PROTON PUMP INHIBITORS AND METHOTREXATE

STUDIJA 1

Proton Pump Inhibitors Inhibit Methotrexate Transport by Renal Basolateral Organic Anion Transporter
hOAT3

Rym Chioukh, Marie-Sophie Noel-Hudson, Sandy Ribes, Natalie Fournier, Laurent Becquemont, and
Celine Verstuyft

EA 4123 Barrières Physiologiques et Réponses Thérapeutiques (R.C., M.-S.N.-H., S.R., L.B., C.V.) and EA
4529 Lipides Membranaires et Régulation Fonctionnelle du Cœur et des Vaisseaux (N.F.), Université
Paris-Sud, Faculté de Pharmacie, Châtenay-Malabry, France; Centre de Recherche Clinique Assistance
Publique Hôpitaux de Paris, Hôpital Bicêtre, Le Kremlin Bicêtre, France (L.B.); and Service de Génétique
Moléculaire, Pharmacogénétique et Hormonologie, Assistance Publique Hôpitaux de Paris, Hôpital
Bicêtre, Le Kremlin Bicêtre, France (C.V.)

Address correspondence to: Céline Verstuyft, EA 4123, Univ. Paris Sud, Service de Génétique
Moléculaire, Pharmacogénétique et Hormonologie, Assistance Publique Hôpitaux de Paris, Hôpital
Bicêtre, 78 rue du General Leclerc, 94275 Le Kremlin Bicêtre, France. E-mail:
celine.verstuyft@bct.aphp.fr

Abstract

The coadministration of methotrexate (MTX) and proton pump inhibitors (PPIs) can result in a
pharmacokinetic interaction that delays MTX elimination and subsequently increases the MTX blood
concentrations. Human organic anion transporters (hOATs) are responsible for the renal tubular
secretion of MTX and are thought to be involved in this drug interaction. The aim of this study was to
evaluate the inhibitory potencies of PPIs on hOAT1 and hOAT3, which are the two isoforms of OATs
predominantly expressed in kidney proximal tubules. Using stably transfected cell systems that express
the uptake transporters human embryonic kidney (HEK)-hOAT1 and HEK-hOAT3, we analyzed the
inhibitory potencies of omeprazole, lansoprazole, and pantoprazole on OAT-mediated [3H]estrone
sulfate (ES), [3H]p-aminohippuric acid (PAH), and [3H]MTX uptake in vitro. hOAT3 is a high affinity
transporter for MTX (Km = 21.17 ± 5.65 µM). Omeprazole, lansoprazole, and pantoprazole inhibited
[3H]MTX uptake in HEK-hOAT3 cells with an IC50 of 6.8 ± 1.16, 1.14 ± 0.26, and 4.45 ± 1.62 µM,
respectively, and inhibited the [3H]ES uptake in HEK-hOAT3 cells with an IC50 of 20.59 ± 4.07, 3.96 ±
0.96, and 7.89 ± 2.31 µM, respectively. Furthermore, omeprazole, lansoprazole, and pantoprazole
exhibited inhibited PAH uptake on hOAT1 in a concentration-dependent manner (IC50 = 4.32 ± 1.26,
7.58 ± 1.06, and 63.21 ± 4.74 µM, respectively). These in vitro results suggest that PPIs inhibit [3H]MTX
transport via hOAT3 inhibition, which most likely explains the drug-drug interactions between MTX and
PPIs and should be considered for other OATs substrates.

Introduction
Methotrexate (MTX), an antifolate drug, is used in a wide range of doses for the treatment of certain
neoplastic diseases, severe psoriasis, and rheumatoid arthritis (Jolivet et al., 1983; Tugwell et al., 1987).
High-dose MTX is widely accepted as the first line treatment of lymphoid malignancy, osteogenic
sarcoma, and acute leukemia, with intravenous doses ranging from 300 mg/m2 to 12 g/m2. MTX is a
highly toxic drug with a low therapeutic index. The therapeutic drug monitoring of MTX is essential to
prevent toxicity from high plasma MTX concentrations, because delayed elimination can result in serious
and potentially life-threatening toxicities.

Renal excretion is the primary route of MTX elimination. In humans 80 to 90% of the intravenous
administered dose is excreted unchanged in the urine within 24 hours (Shen and Azarnoff, 1978). Renal
excretion occurs via glomerular filtration and active tubular secretion mediated in proximal tubular cells
uptake, followed by active efflux in tubular lumen. Organic anion transporters are responsible for the
passage from the blood to proximal tubules (uptake). Many transporters of organic anionic drugs have
been identified on the apical side of the human kidney epithelium, including multidrug-resistance–
related protein (MRP2, ABCC2; MRP4, ABCC4) and breast cancer resistance protein (BCRP, ABCG2),
which are responsible for the secretion into the urine (Takeda et al., 2002a; Burckhardt and Burckhardt,
2003; Launay-Vacher et al., 2006; Nozaki et al., 2007; VanWert and Sweet, 2008).

Among human organic anion transporters (hOATs), hOAT1 and hOAT3 localize to the basolateral
membrane of proximal tubular epithelial cells and have been shown to transport MTX (Uwai et al., 1998;
Nozaki et al., 2007; Rizwan and Burckhardt, 2007). Members of the OAT family transport a variety of
endogenous substances and drugs, including antineoplastic agents, antiviral agents, β-lactam–
antibiotics, diuretics, and angiotensin-converting enzyme inhibitors (Takeda et al., 2002b; Uwai et al.,
2007; Vallon et al., 2008; Vanwert et al., 2008).

Several drugs, including nonsteroidal anti-inflammatory drugs (Nozaki et al., 2007; Uwai et al., 2004;
Maeda et al., 2008), penicillin G (Takeda et al., 2002b), and probenecid (Aherne et al., 1978), are known
to inhibit the elimination of MTX. The molecular mechanism underlying these interactions partially relies
on the blockade of the renal secretion of antifolate via the basal uptake transporters hOAT3 and hOAT1
(Giacomini et al., 2010).

Over the past few years, several case reports in oncology (Reid et al., 1993; Beorlegui et al., 2000; Troger
et al., 2002; Bauters et al., 2008) and two retrospective cohort studies (Suzuki et al., 2009; Santucci et
al., 2010) have suggested that the coadministration of proton pump inhibitors (PPIs), including
omeprazole, pantoprazole, lansoprazole, and rabeprazole, decreased the renal clearance of MTX. The
elimination of MTX was significantly delayed during cycles with one PPI but normalized during
subsequent cycles after PPI discontinuation or substitution with ranitidine.

Because PPIs are frequently used among patients treated with MTX for cancer or autoimmune diseases,
we aimed to investigate the drug-drug interaction of MTX with PPIs.

To elucidate the PPI-MTX drug interaction, we used cell systems that stably express the human uptake
transporters OAT1 and OAT3 and investigated the effect of the three more commonly prescribed PPIs
(omeprazole, lansoprazole, and pantoprazole) on the uptake of OAT substrates.
Materials and Methods

Radiolabeled.

[3H]estrone sulfate ([3H]ES; 250 μCi; 9.25 MBq; 54.26 Ci/mmol, 2.00762 TBq/mmol); 99.5% purity;
[3H]p-aminohippurate acid ([3H]PAH; 1 mCi/ml; 37 mBq; 4.53 Ci/mmol, 167.61 GBq/mmol); 99% purity,
were purchased from Perkin Elmer (Waltham, MA). [3H]methotrexate ([3H]MTX; 250 μCi; 9.25 MBq;
32.3 Ci/mmol); >99% purity, was purchased from Moravek Biochemicals (Brea, CA).

Unlabeled.

p-Aminohippuric acid and estrone sulfate uptake, which are both well-established substrates of hOAT1
and hOAT3, respectively (Burckhardt, 2012), were purchased from Sigma-Aldrich (St. Louis, MO).
Methotrexate ((2S)-2-[(4-[[(2,4-diaminopteridin-6-yl)methyl]
(methyl)amino]phenyl)formamido]pentanedioic acid) was purchased from Interchim (Montluçon,
France).

The inhibitory chemicals, probenicid [4-(dipropylsulfamoyl)benzoic acid], ibuprofen (2-[4-(2-

methylpropyl)phenyl]propanoic acid), omeprazole (6-methoxy-2-[[(4-methoxy-3,5-dimethylpyridin-2-
yl)methane]sulfinyl]-1H-1,3-benzodiazole), lansoprazole (2-([[3-methyl-4-(2,2,2-trifluoroethoxy)pyridin-
2-yl]methane]sulfinyl)-1H-1,3-benzodiazole), and pantoprazole (6-(difluoromethoxy)-2-[[(3,4-
dimethoxypyridin-2-yl)methane]sulfinyl]-1H-1,3-benzodiazole), were purchased from Sigma-Aldrich. All
unlabeled solid compounds were dissolved in dimethylsulfoxide organic solvent (DMSO). The
concentration of DMSO in the final study medium was limited to 1% in presence or absence of
inhibitors.

Scintillation fluid, Ultima-Gold, was from Perkin Elmer Life Science (Boston, MA).

Triton X-100, DMSO, and bicinchoninic acid (BCA) assay kits were obtained from Sigma-Aldrich.
Dulbecco’s modified Eagles medium (DMEM), phosphate-buffered saline, penicillin, streptomycin,
zeocin, and hygromycin B were purchased from Gibco Invitrogen (Cergy-Pontoise, France). Fetal bovine
serum was purchased from PAA Laboratories (Vélizy Villacoublay, France). FuGENE 6 Transfection
Reagent was from Roche Applied Science (Basel, Switzerland).

Cell Culture and Transfection.

Stably transfected human embryonic kidney (HEK) cell lines were established by using the Flp-In
expression system (Invitrogen, Carlsbad, CA) according to the manufacturer's protocol. HEK-293 cells
were routinely grown in DMEM containing 10% fetal calf serum and 1% streptomycin/ampicillin in a
humidified incubator at 37°C and 5% CO2. Briefly, in separate reactions, the cDNAs, including the open
reading frames hOAT1 or hOAT3, were subcloned into the Flp-In expression vector pcDNA5/FRT, which
contained a FRT site linked to a hygromycin resistance gene. The constructs pcDNA5/FRT-hOAT1 and
pcDNA5/FRT-hOAT3 constructs were then cotransfected with the Flp recombinase expression vector
pOG44 into Flp-In HEK-293 cells. Cells stably expressing the transporters were selected in hygromycin
(100 μg/ml) according to the manufacturer's protocol. The cells were grown in flasks cultured in DMEM
supplemented with 10% fetal bovine serum and hygromycin (100 μg/ml). Cultures were maintained in a
humidified atmosphere containing 5% CO2 at 37°C. Cells were split in a 1:5 ratio every 3 to 4 days.

The function of hOAT1 and hOAT3 was evaluated using HEK293 cells stably transfected with
pcDNA5/FRT vector containing hOAT1 and hOAT3 cDNA or empty vector, named HEK-hOAT1, HEK-
hOAT3, and HEK mock, respectively.

Transport Uptake Experiments.

Cells were seeded on poly-D-lysine-coated 12-well plates BD Biocoat from Becton, Dickinson Company
(Franklin Lakes, NJ) at a density of 4 × 105 cells/well and grown for 2 days (37°C/5% CO2) in the absence
of antibiotics. Before the initiation of transport experiments, the culture medium was removed and the
cells were washed twice and preincubated with Krebs-Henseleit buffer at 37°C for 15 minutes. The
Krebs-Henseleit buffer consisted of 118 mM NaCl, 1.2 mM MgSO4, 1.2 mM KH2PO4, 4.7 mM KCl, 26 mM
NaHCO3, 2.5 mM CaCl2, 5 mM glucose, and 12.5 mM HEPES adjusted to pH 7.4.

The equilibration medium was removed before a final application of 400 µl of incubation solution per
sample (buffer containing the radiolabeled compounds) in the presence or absence of prototypic OAT
inhibitors used as positive control probenicid and ibuprofen. [3H]ES or [3H]PAH uptake was measured
for 10 minutes in HEK cells expressing hOAT3 or 2 minutes in HEK-hOAT1 within the linear uptake phase.
We validated the cells systems using [3H]ES (10 nM) uptake in hOAT3 cells in the absence (no inhibitor)
or presence of probenicid and ibuprofen and [3H]PAH (50 nM) uptake in hOAT1 cells in the absence or
presence of probenicid and ibuprofen. For concentration-dependent inhibition studies, PPIs were used
in the following concentrations: 1, 2.5, 5, 10, 25, 50, and 100 µM. After incubation at 37°C for the
specified times, the uptake solutions were removed and the cells were rapidly rinsed three times with
750 µl of ice-cold phosphate-buffered saline. The cells were dissolved in 500 µl of 1 M NaOH and
neutralized after 1 hour with 500 µl of 1 M HCl, and the radioactivity of the aliquots was determined in 4
ml of Ultima-Gold, a scintillation fluid, using a scintillation counter (liquid scintillation counter, Tri-carb
2900TR, Perkin Elmer). The cellular protein content was determined using the BCA-protein
quantification system. Uptake was then normalized to the protein content in the lysates.

Transformations for kinetic calculations were performed using GraphPad Prism software version 4
(GraphPad Software, San Diego, CA), and the Km and Vmax values were calculated from the x and y
intercepts of the Lineweaver-Burk plot, respectively. The Ki values were calculated assuming competitive
inhibition.

Western Blotting Analysis.

Total proteins were extracted from the pellets containing HEK293 cells by homogenizing the pellets in
TENTS [10 mM Tris-HCl at pH 7.4, 5 mM EDTA at pH 8, 126 mM NaCl, 1% (v/v) Triton X-100, and 0.1%
(v/v) SDS] supplemented with leupeptin, aprotinin, pepstatin, and phenyl methane sulfonyl fluoride
(Sigma-Aldrich). The suspensions were gently agitated for 1 hour at 4°C and then centrifuged at 12,000g
at 4°C for 20 minutes. The protein content of the supernatant was determined using the BCA assay.
Next, 25 µg of protein was separated by electrophoresis using the NuPage Novex Bis Tris MiniGels
(Invitrogen) according to the manufacturer's protocol and transferred electrophoretically onto
nitrocellulose membranes. Free binding sites on the membranes were blocked by incubation with Tris-
buffered saline containing 0.1% of Tween-20 (TTBS) and 10% nonfat dried milk for 1 hour at 20–25°C.
The membranes were washed with TTBS and incubated with primary antibodies [anti-hOAT3 rabbit
OAT3 (P-13), Santa Cruz Biotechnology, Inc. Santa Cruz, CA) and anti-hOAT1 rabbit, Sigma-Aldrich, Inc.)]
overnight at 4°C. The primary antibodies were diluted 1:500. The membranes were then washed with
TTBS (5 times for 10 minutes each) and then incubated with secondary antibodies diluted at 1:1000 for 1
hour at 20–25°C. The secondary antibodies were purchased from Dako (Glostrup, Denmark).

The membranes were washed again (5 times for 10 minutes) with TTBS and probed with the Western
Lightning Chemiluminescence Reagent (Perkin Elmer).

Quantitative Real-Time Polymerase Chain Reaction.

Quantitative real-time polymerase chain reaction (PCR) was performed for hOAT1 and hOAT3. RNA
prepared from HEK mock, HEK-hOAT1, and HEK-hOAT3 was purified on RNeasy columns (Qiagen,
Valencia, CA) and then reverse transcribed using Transcriptor First Strand cDNA Synthesis Kit (Roche
Applied Science, Nutley, NJ) using oligo-dT as a primer. Each cDNA sample was subjected to duplicate
real-time PCR reactions using a CFX96 (Bio-Rad, Hercules, CA) thermal cycler with the following
conditions: initial denaturation (95°C for 30 seconds) followed by 44 cycles of denaturation (95°C for 2
seconds), hybridization- extension (60°C for 5 seconds).

Gene expression values were normalized to that of GAPDH in the corresponding cDNA samples.

Kinetic Analyses.

Transformations for kinetic calculations were performed using GraphPad Prism software version 4
(GraphPad Software), and the Km and Vmax values were calculated from the x and y intercepts of the
Lineweaver-Burk plot, respectively. The IC50 values were calculated assuming competitive inhibition.
The kinetic parameters were obtained using the following Michaelis-Menten equation: one saturable
component,

Formula

Where v is the uptake velocity of the substrate (pmoles per milligram of protein per minute). S is the
substrate concentration of the medium (micromolar). Km is the Michaelis constant (micromolar). Vmax
is the maximal uptake velocity (picomoles per milligram of protein per minute).

Statistics.

The uptake experiments were performed in triplicate, where the values are expressed as the mean of
these replicates with error bars represent the standard error. All experiments were performed at least
three times over three independent experiments in triplicate. Statistical significance was calculated by
using unpaired Student’s test. Differences were considered statistically significant if P <0.05.
Results

Characterization of hOAT1- and hOAT3-Expressing HEK Cells.

To test the inhibitory potencies of PPIs in vitro, we stably transfected HEK cells with cDNAs encoding
human OAT1, the SLC22A6 gene, or human OAT3, the SLC22A8 gene. We validated these models by
examining the presence of the proteins and the function of HEK-hOAT1– or HEK-hOAT3–transfected
cells. The gene and protein expression levels of hOAT1 and hOAT3 were evaluated with quantitative
real-time PCR and Western blot analysis. The respective recombinant OAT proteins were detected in the
membrane fractions from OAT-expressing HEK cells but not in the HEK-mock control cells, at molecular
masses of 60 kDa in the membrane fractions from hOAT1-expressing HEK cells, and 62 kDa in the
membrane fractions obtained from hOAT3-expressing HEK cells (Supplemental Data). The quantitative
real-time PCR analysis demonstrated SLC22A6 and SLC22A8 mRNA expression in HEK hOAT1 and HEK
hOAT3 clones, respectively, which were not detected in the vector-transfected HEK-mock cells.

Both transfected HEK cell lines expressed functionally active organic anion transporters, as
demonstrated by the time-dependent PAH and ES uptake, which are both well-established substrates of
hOAT1 and hOAT3, respectively (Supplemental Data). The stably transfected hOAT1 and hOAT3
expressing cell lines also accumulated significantly more standard substrates {[3H]p-aminohippurate
acid ([3H]PAH) for HEK-hOAT1 and [3H]estrone sulfate ([3H]ES) for HEK-hOAT3} than the control cells.
The estimated Km values of PAH uptake by hOAT1 and uptake of ES by hOAT3 were 15.18 ± 1.93 and
28.32 ± 7.11 µM, respectively (Fig. 1A). Similar to previous in vitro studies, probenecid and ibuprofen
inhibited all mediated transport. Probenicid significantly inhibited the uptake of [3H]PAH by HEK-hOAT1
and uptake of [3H]ES by HEK-hOAT3, with IC50 values of 9.02 ± 2.28 and 0.76 ± 0.28 µM, respectively
(Fig. 1B). In addition, ibuprofen inhibited these uptakes with IC50 values of 1.45 ± 1.54 µM for hOAT3
and IC50 = 15.74 ± 4.35 µM for hOAT1 (Supplemental Data).

Characterization of HEK cells stably transfected with cDNAs encoding human OAT1 or OAT3. (A)
Intracellular uptake of PAH in HEK-hOAT1 and ES in HEK-hOAT3: Net intracellular PAH accumulation in
HEK-hOAT1 cells after 2-minute incubation with increasing PAH concentrations. Intracellular ES uptake
in HEK-hOAT3 cells after 10-minute incubation with increasing ES concentrations. The uptake was
obtained by subtracting the uptake in vector-transfected cells (HEK mock) from that in HEK-hOAT1– or
HEK-hOAT3–expressing cells. Km and Vmax values were calculated by fitting the data to a one-site
binding curve. Data are means ± S.E.M. of 3 determinations. (B) Probenicid inhibition in OAT-expressing
cells. [3H]PAH (50 nM) or [3H]ES (10 nM) uptake was measured for 2 minutes in HEK-hOAT1 or 10
minutes in HEK-hOAT3, respectively, in the absence or in presence of increasing inhibitor concentration.
The uptake amounts of [3H]PAH or [3H]ES in HEK-hOAT1 or in HEK-hOAT3, respectively, were
determined and shown as a percentage. IC50 values were calculated by fitting the data to a sigmoidal
dose-response regression curve. Data points are the means ± S.E.M. of three independent experiments.

MTX Uptake.

To evaluate the uptake of MTX in HEK-hOAT1 and HEK-hOAT3, the cells were incubated in a solution
containing 0.5 µM MTX (for HEK-hOAT1) (Fig. 2A) or 25 nM MTX (for HEK-hOAT3; Fig. 2B). The affinity of
MTX for hOAT3 was higher than that for hOAT1 (Fig. 2A). The Km values of MTX uptake by hOAT3 was
21.17 ± 5.65 µM (Fig. 2C). The higher concentration tested on HEK-hOAT1 was 0.5 µM, with an
accumulation of MTX in HEK-hOAT1, which was approximately twofold higher than that in the control
cells. The Km was not determined for HEK-hOAT1, because the difference in the accumulation of MTX
was too low (data not shown). The transporter-mediated uptake of [3H]MTX over time in HEK-hOAT3 is
presented in Fig. 2B and was linear up to 10 minutes.

Intracellular [3H]MTX uptake. (A) Intracellular [3H]MTX uptake in HEK-hOAT1 and HEK mock cells after
5-minute incubation with 0.5 µM [3H]MTX. Data are means ± S.E.M. of three independent experiments.
Error bars in control cells are within the borders of the bars. *P < 0.05 significantly different from the
control values. (B) [3H]MTX uptake in HEK-hOAT3 and HEK mock cell, after 10-minute incubation with 25
nM [3H]MTX. Data are means ± S.E.M. of three independent experiments. Error bars in control cells are
within the borders of the bars. *P < 0.05 significantly different from the control values. (C) Intracellular
MTX uptake in HEK-hOAT3 cells after 10-minute incubation with increasing MTX concentration. The
uptake was obtained by subtracting the uptake in HEK mock from that in HEK-hOAT3. Km and Vmax
values were calculated by fitting the data to a one-site binding curve. Data are means ± S.E.M. of three
determinations. (D) Net transporter-mediated [3H]MTX uptake by HEH-hOAT3 cells over time,
incubation with 25 nM [3H]MTX. Data are means ± S.E.M. of three independent experiments.

Inhibition of hOAT1- and hOAT3-Mediated Transport by PPIs.

The inhibition of hOAT1 and hOAT3 uptake of their specific substrate by PPIs was measured within the
linear uptake phase.

Regarding the inhibition of [3H]PAH uptake by hOAT1, omeprazole, lansoprazole, and pantoprazole
inhibited the transport of PAH in HEK-hOAT1 in a concentration-dependent manner, with IC50 values of
IC50 = 4.32 ± 1.26, 7.58 ± 1.06, and 63.21 ± 4.74 µM, respectively (Fig. 3).

Inhibition of hOAT1-mediated [3H]PAH uptake by PPIs. Inhibitory effect of omeprazole, lansoprazole,

and pantoprazole on hOAT1-mediated [3H]PAH uptake after 2-minute incubation. IC50 values were
calculated by fitting the data to a sigmoidal dose-response regression curve. Data points are the means ±
S.E.M. of three independent experiments.

Each tested PPI significantly inhibited hOAT3-mediated [3H]ES transport in a concentration-dependent

manner (Fig. 4). The calculated half-maximal inhibitory concentration values were in the micromolar
range. We obtained an IC50 of 20.59 ± 4.07 µM for omeprazole, an IC50 of 3.96 ± 0.96 µM for
lansoprazole, and an IC50 of 7.89 ± 2.31 µM for pantoprazole. Likewise, omeprazole, lansoprazole, and
pantoprazole inhibited the transport of [3H]MTX in HEK-hOAT3 cells, with IC50 values of 6.8 ± 1.16, 1.14
± 0.26, and 4.45 ± 1.62 µM, respectively (Fig. 4).

Inhibition of hOAT3-mediated [3H]ES and [3H]MTX uptake by PPIs. Inhibitory effect of omeprazole,
lansoprazole, and pantoprazole on hOAT3-mediated [3H]ES or [3H]MTX uptake after 10-minute
incubation. IC50 values were calculated by fitting the data to a sigmoidal dose-response regression
curve. Data points are the means ± S.E.M. of three independent experiments.
Discussion

MTX is currently used in wide range of doses, and high-dose MTX schedules are associated with an
incidence of nephrotoxicity of 1.8% and a fatality rate of almost 0.1%, despite therapeutic drug
monitoring and supportive therapy (Widemann and Adamson, 2006) Although drug interactions
between MTX and PPIs have been described in the clinic, the specific mechanism for this drug-drug
interaction remains unknown.

Our major finding indicates that hOAT3, an uptake transporter expressed at the basolateral side of renal
proximal tubular cells, selectively mediates the uptake of MTX, and this transporter is dramatically
inhibited in the presence of PPIs. Different studies have suggested the involvement of multiple drug
transporters in the elimination of MTX (Breedveld et al., 2004; Suzuki et al., 2009), but the uptake
transporters have been well established to be the first limiting step of MTX elimination (VanWert and
Sweet, 2008). Among the OATs, OAT1 and OAT3 localize to the basolateral membrane of proximal
tubular cells and have been shown to play a central role in the renal uptake of anionic drugs, namely
MTX.

Our study confirmed that hOAT3 is a high-affinity type transporter of MTX. In our study, the estimated
Km value for hOAT3 was 21.17 ± 5.65 µM, which was consistent with the Km values of MTX uptake (10.9
and 21.1 µM) previously described by Cha et al. (2001) and Takeda et al. (2002a), respectively. Because
this Km value determined in human kidney sections was similar to that observed for hOAT3 in this study
rather than that observed for hOAT1 (553.8 ± 43.2 µM) by Takeda et al. (2002a) in transfected S2 cells,
OAT3 likely more significantly contributes to the net uptake process involved in MTX elimination. We
failed to detect MTX transport in HEK-hOAT1 below a concentration of 50 nM; because the uptake
experiment required the use of 0.5 µM of MTX according to a study described by El-Sheikh et al. (2013),
we could observe an uptake transport by incubating HEK-OAT1 with only 0.5 µM MTX. Unfortunately the
difference from the mock cells was not sufficient to evaluate the drug-drug interaction. Moreover, we
believe that a MTX concentration above 100 µM is not clinically relevant for therapeutic drug
monitoring, because a slow elimination of MTX was defined as plasma concentrations exceeding 15 µM
at 24 hours (Santucci et al., 2010). These concentrations are much higher than the human plasma
concentrations of MTX and seem unlikely clinical practice. Our current results were also consistent with
the findings of Lu et al. (1999), who cloned hPAHT (p-aminohippurate transporter, the first name of
hOAT1), which exhibited no significant MTX uptake activity. Uwai et al. (2004) determined the Km value
for hOAT1-mediated MTX uptake using a Xenopus laevis oocytes expression system to be 724 µM. In
fact, this higher value of Km for hOAT1 supported our result, i.e., this concentration was not clinically
relevant (Uwai et al., 2004). More recently Kurata et al. (2014) confirmed the same result with HEK-
hOAT1. Nozaki et al., (2007) also examined MTX uptake using human tissue sections and estimated Km
values within the same range (48.9 ± 17.3 µM) we observed for hOAT3 (Nozaki et al., 2007). As
mentioned previously by various authors, the discrepancy may be due to species differences in the
transport activity between rat and human OAT1 or differences in the expression system (Takeda et al.,
2002a; Uwai et al., 2004; Uwai and Iwamoto, 2010).
The most striking result of our study was the potent inhibition of MTX uptake transport by all 3 PPIs in
HEK-hOAT3 cells. The observed PPI IC50 values for MTX uptake were in the micromolar range (6.80,
1.14, and 4.45 µM for omeprazole, lansoprazole, and pantoprazole, respectively). Interestingly, the IC50
values for the three PPIs of the MTX uptake transport of by hOAT3 were higher. The observed PPI IC50
values were higher for MTX than ES but were within the same concentration range as the plasma
circulating concentrations. Moreover, the IC50 values observed for each PPI were compared with the
plasma concentrations of PPIs according to the CYP2C19 genotype (Ishizaki and Horai, 1999) (see Table
1). Indeed, PPIs are mainly metabolized by CYP2C19, and because the impact of CYP2C19 polymorphism
on drug concentrations has been well established, different concentrations should be considered
(Goldstein, 2001; Simon et al., 2011). A previous group described the maximum concentration of carriers
of a loss of function allele in the plasma for omeprazole (3.1 µM), lansoprazole (4.9 µM), and
pantoprazole (11.5 µM) according to the CYP2C19 “poor metabolizer” phenotype (Regardh et al., 1990;
Pue et al., 1993; Yasuda et al., 1995; Ieiri et al., 2001; Freston et al., 2003). The plasma concentrations
were lower in carriers of the normal allele with an “extensive metabolizer” phenotype, 1.6, 2.4, and 5.4
µM for omeprazole, lansoprazole, and pantoprazole, respectively.

PPI dosage and maximal total PPI concentration in the systemic circulation (Cmax) were obtained from
the indicated references. [I] = (Cmax total) * (% plasma unbound fraction)/100].

Until recently, most studies investigated the effects of PPIs on different in vivo or in vitro models and
suggested some effect of PPIs on efflux transporters. The effect of PPIs on the uptake transporter was
poorly understood. The present finding also confirms that PPIs potently interact with different uptake
transporters (hOAT1 and hOAT3) and their well-established substrates. Among the three PPIs tested for
the PAH uptake by HEK-hOAT1, two elicited a strong inhibitory effect (omeprazole IC50 = 4.32 ± 1.26 µM
and lansoprazole 7.58 ± 1.06 µM). In agreement with our results, Nies et al. (2011) recently published
that PPIs inhibited hOCT-mediated metformin uptake in vitro. All five tested PPIs (omepazole,
pantoprazole, lansoprazole, rabeprazole, tenatoprazole) significantly inhibited metformin uptake by
HEK-hOCT1, -hOCT2, and -hOCT3 in a concentration dependent manner. Consistent with our result, the
IC50 values of these PPIs were in the low micromolar range (3–36 µM) (Nies et al., 2011). In addition,
the IC50 values of potent OAT drug inhibitors, such as ibuprofen, ketoprofen, piroxicam, indomethacin,
and probenicid, described for adefovir uptake transport by hOAT1 were 8.0, 1.3, 20.5, 3.0, and 7.4 µM,
respectively (Takeda et al., 2002a), which agrees with our results.

Although the clinical consequences are not easily predicted based on in vitro data, detailed advantages
and limitations of various in vitro systems for evaluation of drugs as substrates, as inhibitors, or for their
potential for drug-drug interactions have been delineated (Brouwer et al., 2013; Giacomini and Huang,
2013; Hillgren et al., 2013; Zamek-Gliszczynski et al., 2013). According on the decision trees of the
recommendations of the International Transporter Consortium (Giacomini et al., 2010), the values of the
ration [I]/IC50 of the three PPIs tested on the uptake of MTX on HEK hOAT3 are lower than 0.1 except
for lansoprazole in poor metabolizers, for which the ratio is higher than 0.1, giving thought to a clinical
interaction between lansoprazole and MTX. It would be very interesting to confirm if this in vitro drug-
drug interaction would be relevant in humans in a prospective study. Some factors support such an
assumption, because plasma MTX is predictive of the risk of toxicity; therapeutic drug monitoring is
often used for patients to evaluate the delayed elimination. Recently, the delayed elimination of MTX
associated with serious side effects was described in three retrospective clinical studies of patients
treated with high doses of MTX and PPIs (Joerger et al., 2006; Suzuki et al., 2009; Santucci et al., 2010;
Leveque et al., 2011) and one prospective study of low dose MTX (Vakily et al., 2005). Although
conflicting data were reported in some case reports for either omeprazole or pantoprazole (Whelan et
al., 1999; Beorlegui et al., 2000; Troger et al., 2002; Bauters et al., 2008), recent clinical studies are in
line with our results suggesting that PPIs might decrease MTX renal clearance via OAT3-mediated
inhibition. In the first study, Joerger et al. (2006) described 76 patients who received high-dose MTX, 13
of whom received omeprazole or lansoprazole. Patients that received MTX and PPIs were associated
with a 27% decrease in the clearance of MTX, which resulted in a significantly higher plasma
concentration of MTX (Joerger et al., 2006). The second study is a retrospective noninterventional
cohort study that included 79 French patients with cancer treated with high-doses of MTX. The
coprescription of PPIs (pantoprazole, lansoprazole, omeprazole, or esomeprazole) was found in half of
the cycles with delayed elimination and only in 15% of the cycles without delayed elimination (Santucci
et al., 2010). The third study examined 74 Japanese patients with cancer. MTX was administered
intravenously with a concomitant administration of omeprazole, lansoprazole, and rabeprazole. The
MTX residual concentrations (311 measurements of plasma MTX) were analyzed in 171 cycles of high
dose MTX. They found that PPI coadministration was still a significant risk factor for delayed elimination
after adjustment for six variables (Suzuki et al., 2009). Interestingly, the delayed elimination of plasma
MTX previously mentioned in these studies was not observed in all patients who received PPIs; based on
our results, this finding may be due to higher concentrations of PPIs in carriers of CYP2C19 loss of
function variant alleles (Ieiri et al., 2001; Simon et al., 2011).

Renal tubular secretion involves different uptake transporters. A recent study showed that the
basolateral localization of mouse reduced folate carrier (RFC-1) in the kidney is responsible for the
uptake of MTX (Nozaki et al., 2004). Other uptake transporters that are mainly expressed in the liver
(OATP1B) or intestine (OATP1A2) were found to transport MTX in vitro. These transporters were very
recently found in vivo in transgenic mice that expressed liver-specific human OATP1B1, OATP1B3, and
OATP1A2. Further studies are necessary to confirm the influence of this transporter on the
pharmacokinetic of MTX in humans.

Conversely, some ATP binding cassette transporters, such as breast cancer resistance protein [ABCG2
(Suzuki et al., 2009)], multidrug resistance–associated protein (MRP) 2 and MRP4, which are expressed
on the apical membranes of kidneys, are reportedly also involved in the excretion of MTX (Ito et al.,
2001; Chen et al., 2002). Suzuki et al. (2009) tested the effect of PPIs on the uptake of MTX into BCRP-
expressing membrane vesicles. The observed IC50 for each PPI was considerably higher than the plasma
concentrations of the PPIs. They also concluded that the inhibitory effects of PPIs on BCRP-mediated
MTX transport alone could not explain this drug-drug interaction (Suzuki et al., 2009). Reports on the
drug-drug interaction between PPIs-MTX and MRP transporters are lacking, and the role of these
transporters should be clarified in subsequent studies.

In conclusion, we identified PPIs as an important class of drugs that inhibits OAT transporters and
confirmed that MTX has a greater affinity for OAT3 than OAT1. Taken together our results indicate that
hOAT1 is likely not involved in the interaction between MTX and PPIs. The growing use of PPIs to treat
peptic ulcers and the widespread use of MTX for a variety of diseases, namely cancers, suggest that a
number of patients may be at risk for MTX toxicity, and more intensive therapeutic drug monitoring is
advised. Thus, further studies are required to evaluate the clinical consequences of the pharmacological
interaction between PPIs and other OAT3 substrates, such as antiviral drugs.

Studija 2.

Br J Clin Pharmacol. 2014 Sep; 78(3): 565–571.

Published online 2014 Aug 21. doi: 10.1111/bcp.12384

Retrospective evaluation of methotrexate elimination when co-administered with proton pump

inhibitors

David J Reeves,1,2 Elizabeth S Moore,3 Devon Bascom,1 and Brandon Rensing1

The aim was to assess potential interaction between methotrexate (MTX) and proton pump inhibitors
(PPIs) in patients receiving high-dose MTX.

Methods

Records of 56 adults receiving 201 cycles of MTX were reviewed to determine effects of PPI
administration on MTX elimination. Repeated-measures logistic regressions and Cox regressions were
performed to evaluate the possible drug interaction.

Results

Despite a significant difference between those receiving a PPI and not receiving a PPI in median MTX
levels at 24 (8.0 vs. 3.9 μmol l−1, respectively, P = 0.013) and 72 h after MTX administration (0.08 vs.
0.05 μmol l−1, respectively, P = 0.037), there was no difference between those receiving a PPI and not
receiving a PPI in the proportion of patients experiencing delayed elimination at 24 (19.2% vs. 20.2%,
respectively, P = 1.000) and 72 h (36.2% vs. 33.7%, respectively, P = 0.765). When data were analysed
using Cox regression, controlling for multiple cycles of MTX per patient, PPI use was not a significant
predictor of time to MTX < 0.1 μmol l−1. When the clustering effect of multiple cycles of MTX per patient
was controlled for, co-administration of PPIs was not a significant predictor of MTX level (P = 0.969). A
comparison of patients with delayed elimination at any time point and those without delayed
elimination indicated that PPI use was not a significant predictor of delayed elimination (P = 0.607).

Conclusions

This study does not support previous findings of a significant interaction between PPIs and MTX. Based
on these results, the clinical significance of any potential interaction is likely to be small.

Keywords: drug interaction, methotrexate, proton pump inhibitor

What is Already Known about this Subject

Case reports and retrospective studies have implicated that concomitant administration of
methotrexate (MTX) and proton pump inhibitors (PPIs) may decrease the elimination of MTX.

Many cancer patients receive PPIs as supportive care for symptoms related to either their cancer and its
treatment or a common primary diagnosis of gastro-oesophageal reflux disease.

Prior studies did not control for clustering of multiple cycles per patient and included a relatively low
proportion (≤21%) of patients receiving PPIs concomitantly with MTX.

What this Study Adds

Almost half of the patients (47.3%) received PPIs concomitantly with MTX.

Use of PPIs did not have a significant effect on the proportion of patients who experienced delayed MTX
elimination and, after controlling for the effects of clustering, PPI use was not a significant predictor of
MTX level or delayed elimination.

The clinical significance of any potential interaction between MTX and PPIs is likely to be small.

Introduction

Methotrexate (MTX) is an antifolate chemotherapeutic agent that inhibits dihydrofolate reductase. High
doses of MTX are often used in the treatment regimens for multiple malignancies, including acute
leukaemia, aggressive lymphomas and osteosarcoma. Leucovorin rescue after high-dose MTX and
monitoring of drug levels is essential when administering these high doses in order to detect elevated
plasma drug concentrations and direct routine ancillary measures to prevent toxicity (i.e. leucovorin
dosing, urinary alkalinization, etc.). Additionally, through therapeutic drug monitoring, early recognition
of those patients with renal impairment and at risk for severe toxicity secondary to methotrexate can
allow the early administration of glucarpidase to treat toxic methotrexate levels in those with
significantly delayed elimination.

Methotrexate is primarily eliminated through renal excretion, with 95% recovered in the urine
unchanged within 30 h of administration 1. Numerous factors are known to influence MTX elimination,
including urinary pH, renal function, oedema/fluid third spacing and concomitant administration of
drugs known to interact with methotrexate elimination (sulfamethoxazole/trimethoprim, aspirin,
nonsteroidal anti-inflammatory drugs, probenecid, cephalosporins, dantrolene and penicillins) 1,2.

Several case reports, as well as some retrospective studies, have implicated that concomitant
administration of MTX and proton pump inhibitors (PPIs) may decrease the elimination of MTX, leading
to elevated plasma drug concentrations 3–11. However, not all reports have shown that co-
administration of MTX and PPIs results in delayed MTX elimination 2,12. In addition to reports of a
possible interaction, the prescribing information for MTX and the PPIs in the USA has been updated with
a warning of this possible interaction 13,14. The proposed mechanisms for the interaction include
inhibition of renal H+/K+-ATPase involved in active tubular secretion of MTX or the inhibition of breast
cancer resistance protein, which is also involved in mediating MTX transport 15.

Many patients, especially those with serious illnesses such as a malignancy, receive adjunct medications,
such as PPIs. This class of medications is believed to be relatively safe and free from interactions outside
of decreasing the bioavailability of some orally administered medications that rely on lower gastric pH
levels for absorption. Clinically, a significant drug interaction between MTX and PPIs could have
devastating effects on cancer patients given the narrow therapeutic index of MTX. Delays in elimination
could lead to significant adverse effects, such as mucositis and enhanced myelosuppression. Despite
literature supporting a possible interaction between these medications, effects of delayed MTX
elimination are rarely observed given the high frequency of PPI use in the cancer population. The
purpose of this study was to determine whether an interaction exists between MTX and PPIs that results
in delayed MTX elimination.

Methods

Study design

Patients admitted to our institution between 1 January 2008 and 1 August 2012 who received high-dose
MTX [≥1000 mg m−2 intravenously (IV) over 3–4 h or ≥200 mg m−2 IV bolus followed by ≥800 mg m−2 IV
over 22–24 h] were identified via the pharmacy’s electronic database. If a patient was admitted more
than once during this time period, data from each admission were collected. Therefore, it was possible
for patients with more than one admission to have data collected on multiple cycles of MTX. Patients
were excluded if they were <18 years of age, pregnant or did not have MTX levels monitored. Patient
information and data were extracted from the electronic medical records. Study methods were
reviewed and approved by the local institutional review board.

Data collection

Medical records of included patients were reviewed for the duration of the MTX infusion and the time
until plasma MTX levels were ≤0.1 μmol l−1. Baseline demographics (diagnosis, sex, age, height, weight
and body surface area) were collected for all patients, together with MTX dose/infusion time, baseline
serum creatinine (Scr), aspartate aminotransferase (AST), alanine aminotransferase (ALT), urine pH, use
of urinary alkalinization, presence of pleural effusions or ascites and leucovorin administration. Serum
creatinine together with patient age, height, weight, and gender were used to calculate creatinine
clearance (CrCl) via the Cockroft–Gault formula. The use of drugs known to interact with methotrexate
was also noted, including sulfamethoxazole/trimethoprim, aspirin, nonsteroidal anti-inflammatory
drugs, probenicid, penicillins, cephalosporins and dantrolene. Methotrexate levels and Scr were
collected until MTX levels were nontoxic (MTX level <0.1 umol l−1). The standard of care at the
institution was to check MTX levels at 24, 48 and 72 h after the start of the MTX infusion and every
morning thereafter, until MTX level was ≤0.1 μmol l−1. Lastly, the use of a PPI was noted.

End-points
The primary end-point of the study was to determine whether the use of PPIs in conjunction with MTX
resulted in delayed MTX elimination in comparison to those patients not receiving PPIs. Delayed MTX
elimination was defined as plasma MTX concentrations of >10 μmol l−1 at 24 h after the start of MTX
therapy if receiving bolus infusion (i.e. over 3–4 h) or >20 μmol l−1 at 24 h after the start of MTX therapy
if receiving infusional MTX (i.e. over 22–24 h), >1 μmol l−1 at 48 h and >0.1 μmol l−1 at 72 h. Patients
receiving their dose over 3–4 h were classified as having a short infusion time, while those receiving
their dose over 22–24 h were classified as having a long infusion time. Secondary end-points included a
comparison of the time taken to reach an MTX level of <0.1 μmol l−1 in patients receiving PPIs with
those patients not receiving PPIs and a comparison of PPI use among those with and without delayed
elimination.

Statistical analysis

Differences in patient characteristics, MTX dose, MTX infusion time (short/long) and incidence of
delayed elimination were compared using Fisher’s exact test or the Mann–Whitney U test, as
appropriate. To control for variables affecting MTX elimination and clustering effects of multiple cycles
of MTX per patient, a repeated-measures regression analysis using mixed procedures was conducted on
MTX level, a repeated-measures logistic regression using GLIMMIX procedure was conducted on the
nominal variable delayed elimination, and time to MTX level <0.1 μmol l−1 was analysed using the Cox
regression PHREG procedure. The following variables were included in the repeated measures and Cox
regression analyses: patient demographics (age, gender, height, weight); PPI administration; receipt of
additional interacting medications; MTX level (24, 48 and 72 h postinfusion); MTX dose; MTX infusion
time (short/long); baseline Scr and CrCl; baseline AST/ALT; and urinary pH (<7). Data were analysed
using IBM SPSS Statistics for Windows, version 20.0 (IBM Corp., Armonk, NY, USA) and SAS, version 9.1
(Cary, NC, USA). A P value <0.05 was considered to be statistically significant.

Results

Patient characteristics

A total of 56 patients receiving 201 cycles of high-dose MTX were identified and included in this analysis.
Proton pump inhibitors were co-administered with MTX in 95 (47%) of the cycles. Of the 56 patients, 18
(32.1%) did not receive a PPI with any MTX cycle, 23 (41.1%) received a PPI with every MTX cycle, and 15
(26.8%) received a PPI during at least one MTX cycle and did not receive a PPI during at least one MTX
cycle. Results are reported per MTX cycle and not by patient.

The majority of baseline characteristics were similar between patients who received a PPI compared
with those who did not receive a PPI, including renal function as represented by Scr levels and
methotrexate dosing (Table (Table1).1). There were two differences between the groups, with patients
who received concomitant PPIs being more likely to have a long MTX infusion time compared with those
not receiving a PPI (50.5 vs. 34.0%, P = 0.022) and those receiving PPIs also receiving more medications
known to interact with MTX. The difference in infusion time (short vs. long) between groups remained
significant even when logistic regression using the GLIMMIX procedure was used to control for
clustering (P = 0.017).
Patients in both groups received the same supportive care, with all patients receiving urinary
alkalinization and leucovorin rescue. Urine pH was ≥7 in the majority of patients receiving high-dose
MTX at baseline, 24, 48 and 72 h, with no differences between groups seen in those with urinary pH
results available at those time points (Table (Table2).2). A statistically significant difference was found in
the 24 h serum creatinine between those receiving a PPI and those not receiving a PPI (0.8 vs. 0.9 mg
dl−1, respectively, P = 0.028); however, this difference was not present at 48 and 72 h (Table (Table22).

Patients who received a PPI had significantly higher MTX levels at 24 and 72 h compared with patients
who did not receive a PPI (Table (Table3).3). However, when data were analysed to control for the
clustering effect of multiple cycles of MTX per patient, using repeated-measures mixed procedures, co-
administration of PPIs was not a significant predictor of MTX level (P = 0.969). Significant predictors
identified included height (P = 0.010), CrCl (P = 0.003) and time (P < 0.001). Patients who were taller had
higher MTX levels, whereas those with a higher CrCl had lower MTX levels. Similar to the effect of CrCl,
time also showed a negative correlation with the MTX level.

There was a statistical difference between groups in the time taken to reach a nontoxic MTX level based
on observed drug levels at 24, 48 and 72 h, as represented by the differences in interquartile ranges,
even though the median time to reach an MTX level <0.1 μmol l−1 was 72 h in both groups (Table
(Table3).3). When data were analysed using Cox regression, controlling for multiple cycles of MTX per
patient, PPI use was not a significant predictor of the time taken to reach an MTX level <0.1 μmol l−1.
The only significant predictor was baseline CrCl (P = 0.008). The higher the CrCl value, the less time it
took to achieve an MTX level <0.1 μmol l−1. Additionally, the proportion of patients experiencing
delayed elimination was not different between groups at any time point (Table (Table33).

Delayed methotrexate elimination

When patients with delayed elimination at any time point after the MTX infusion were compared with
those without delayed elimination, the only statistically significant differences between the groups were
patient age and baseline Scr (Table (Table4).4). When data were analysed using repeated-measures
logistic regression with the GLIMMIX procedure to control for possible clustering, there was not an
association between PPI use and delayed elimination of MTX (P = 0.607). Patient age and time were the
only significant predictors of delayed MTX elimination (P = 0.008 and P < 0.001, respectively). Older
patients had an increased risk of delayed elimination while, as one would expect, the risk of delayed
elimination lessened over time.

Discussion

Delayed elimination of MTX after administration of high doses could have devastating effects on
patients, including increased risk for serious adverse effects. It is particularly important for practitioners
to evaluate a patient’s risk for delayed MTX elimination critically and take the necessary precautions
during MTX administration, including altering the home drug regimen to minimize drug–drug
interactions. Proton pump inhibitors have relatively few known pharmacokinetic drug interactions;
however, there have been multiple reports of delayed MTX elimination when co-administered with PPIs.
Results from our study showed that once the data were controlled for confounders and possible
clustering, PPI use was not a significant predictor of MTX level, delayed elimination or the time taken to
reach an MTX level <0.1 μmol l−1, despite the fact that the PPI group had a significantly longer infusion
time and were more likely to receive other interacting medications.

Results of the present study echo those of a recent evaluation of the effect of drug interactions on MTX
elimination and toxicity 2. The retrospective, case–control study identified 73 patients receiving MTX (23
cases with delayed elimination and 50 control patients without delayed elimination). It was determined
that, along with other potential interacting medications, PPIs were not significantly associated with
delayed MTX elimination [odds ratio (95% confidence interval), 1.50 (0.55–4.06), P = 0.454]. Use of PPIs
occurred in 11 case patients (47.8%) and 19 control patients (38.0%). The present study also did not
observe any differences in the use of PPIs between those with and without delayed elimination.
Additionally, when the present study used an alternative method to analyse the data, in that patients
were grouped and compared based on whether or not they received a PPI during their admissions for
MTX administration, the results were similar, in that PPIs were not a significant predictor of MTX levels
or elimination.

Multiple reports have shown an association between PPI co-administration and delayed MTX
elimination. Five case reports describing a possible interaction between PPIs and MTX were identified 3–
7. Four reports described increased levels while receiving PPIs concomitantly with high-dose MTX, while
the fifth described a case of increased toxicity (myalgia) with low-dose weekly intramuscular MTX. An
additional case report that did not find an interaction with omeprazole was also identified 12. This case
described a patient who had identical MTX clearance during two cycles of high-dose MTX for
osteosarcoma despite receiving omeprazole only during the first cycle.

In addition to the multiple case reports described above, a case series, two retrospective clinical studies
and a pharmacokinetic study described an association between delayed MTX elimination and
concomitant PPI use 8–11. The case series included five patients who received glucarpidase for delayed
MTX elimination 8. All patients received a PPI during the course of MTX therapy; however, in previous
and subsequent cycles without PPI co-administration, delayed elimination was not observed. In the two
retrospective clinical studies, PPI use was associated with delayed elimination 10,11. These studies
included 171 and 177 cycles of high-dose MTX, and 18 and 21% of the patients in each study received
PPIs, respectively. This is significantly less than the proportion of patients receiving PPIs in the present
study (47.3%). For the published studies, data were primarily analysed comparing patients with delayed
elimination with those without delayed elimination, which differs from how data were analysed in the
present study. Moreover, previous studies did not control for multiple cycles in the same patient. In the
present study, a statistically significant difference was found in MTX levels and the time to achieve MTX
levels <0.1 μmol l−1 between patients who received and did not receive PPIs when data were compared
without adjusting for potential clustering. However, after controlling for clustering, a statistically
significant difference was not found. It is possible that had earlier published studies controlled for
multiple cycles of MTX, they would not have found the increased use of PPIs in patients with delayed
elimination to be significant. Another consideration is the clinical significance of differences in MTX
elimination observed between groups. Although median MTX levels at 24 and 72 h were higher in the
PPI group prior to adjusting for clustering, this is likely to be clinically insignificant given the fact that the
median levels in both groups were below the generally accepted thresholds for 24 and 72 h
methotrexate levels (24 h, ≤10–20 μmol l−1; and 72 h, ≤1 μmol l−1). Likewise, the difference in the time
to reach a nontoxic MTX level is likely to be of little clinical significance given the fact that the medians
were the same and that this effect disappeared upon controlling for clustering.

The lack of an association of PPI use with MTX elimination in this study differs from previous studies and
may be related to the greater number of patients receiving PPIs in our study. Additionally, methods used
for data analysis differed from previous studies. Similar to the other studies describing a possible
interaction, the present study is limited by its retrospective nature. Another limitation is that our study
did not assess adverse effects associated with MTX. It is possible that patients experience increased
adverse effects when receiving the combination of MTX and PPIs without a noticeable change in
elimination, as described in the case report of myalgia with low-dose MTX 5. However, this is unlikely
given the strong association between MTX drug levels/duration of exposure and adverse effects 1. There
were two differences between groups in baseline characteristics; however, these were unlikely to
influence the results of the study given the fact both would tend to lead to increased MTX levels in the
PPI group (patients receiving PPIs also received more interacting medications and had longer infusion
times). These would benefit the group not receiving PPIs; however, such a benefit was not observed in
the analysis. Additionally, repeated measures, logistic regression and Cox regression did not identify use
of interacting medications and infusion time as having a significant effect on PPI use, MTX levels or MTX
elimination. Another difference between groups was in the 24 h serum creatinine, with those not
receiving a PPI having a slightly higher median serum creatinine (Table (Table2).2). This difference would
favour the PPI group. Creatinine clearance was taken into account in the Cox regression and repeated-
measures testing.

Conclusion

Our study suggests that there is not an association between concomitant use of PPI and MTX level, time
to elimination, or delayed elimination of MTX, despite the fact that patients receiving PPIs had
significantly longer infusion times and received more interacting medications. Although the presence of
an interaction between MTX and PPIs cannot be ruled out by this study alone, the clinical significance of
any potential interaction between MTX and PPIs is likely to be small, especially in those at low risk for
delayed elimination. Future studies should include a large prospective trial to allow for the control of
multiple cycles of MTX per patient and other possible confounders, such as MTX infusion time and
patient characteristics.

STUDIJA 3

Drug Saf. 2014; 37(4): 201–211.

Published online 2014 Feb 19. doi: 10.1007/s40264-014-0144-0

Pharmacokinetic Drug Interaction Profiles of Proton Pump Inhibitors: An Update

Ralph-Steven Wedemeyer corresponding author and Henning Blume

Abstract

Proton pump inhibitors (PPIs) are used extensively for the treatment of gastric acid-related disorders,
often over the long term, which raises the potential for clinically significant drug interactions in patients
receiving concomitant medications. These drug–drug interactions have been previously reviewed.
However, the current knowledge is likely to have advanced, so a thorough review of the literature
published since 2006 was conducted. This identified new studies of drug interactions that are modulated
by gastric pH. These studies showed the effect of a PPI-induced increase in intragastric pH on
mycophenolate mofetil pharmacokinetics, which were characterised by a decrease in the maximum
exposure and availability of mycophenolic acid, at least at early time points. Post-2006 data were also
available outlining the altered pharmacokinetics of protease inhibitors with concomitant PPI exposure.
New data for the more recently marketed dexlansoprazole suggest it has no impact on the
pharmacokinetics of diazepam, phenytoin, theophylline and warfarin. The CYP2C19-mediated
interaction that seems to exist between clopidogrel and omeprazole or esomeprazole has been shown
to be clinically important in research published since the 2006 review; this effect is not seen as a class
effect of PPIs. Finally, data suggest that coadministration of PPIs with methotrexate may affect
methotrexate pharmacokinetics, although the mechanism of interaction is not well understood. As was
shown in the previous review, individual PPIs differ in their propensities to interact with other drugs and
the extent to which their interaction profiles have been defined. The interaction profiles of omeprazole
and pantoprazole sodium (pantoprazole-Na) have been studied most extensively. Several studies have
shown that omeprazole carries a considerable potential for drug interactions because of its high affinity
for CYP2C19 and moderate affinity for CYP3A4. In contrast, pantoprazole-Na appears to have lower
potential for interactions with other medications. Lansoprazole and rabeprazole also seem to have a
weaker potential for interactions than omeprazole, although their interaction profiles, along with those
of esomeprazole and dexlansoprazole, have been less extensively investigated. Only a few drug
interactions involving PPIs are of clinical significance. Nonetheless, the potential for drug interactions
should be considered when choosing a PPI to manage gastric acid-related disorders. This is particularly
relevant for elderly patients taking multiple medications, or for those receiving a concomitant
medication with a narrow therapeutic index.

Introduction

Proton pump inhibitors (PPIs) achieve a greater degree and longer duration of gastric acid suppression,
and better healing rates in various gastric acid-related disorders, than histamine H2 receptor antagonists
[1–3]. They are thus considered essential in the management of gastro-oesophageal reflux disease,
peptic ulcer disease (PUD) and Zollinger–Ellison syndrome. PPIs are also a key part of triple therapy (with
two antibiotics, such as clarithromycin, amoxicillin or metronidazole) for the eradication of H. pylori in
PUD [4], and may be used in the prophylaxis of stress- and NSAID-induced PUD [5, 6]. Many of these
disorders generally require long-term treatment, which increases the potential for clinically significant
drug interactions in patients (such as hospitalised patients and community-dwelling older people [7, 8])
receiving PPIs and other medications [9].
A previous review published in 2006 highlighted the similarities and differences among the PPIs in terms
of the likelihood, relevance and mechanisms of drug–drug interactions [10]. In the review, the authors
discussed how, by elevating pH, PPIs can modify the intragastric release of other drugs from their
dosage forms, and also how PPIs influence drug absorption and metabolism by interacting with
adenosine triphosphate-dependent P-glycoprotein or with the cytochrome P450 (CYP) enzyme system
[10]. At the time of the review, the interaction profiles of omeprazole and pantoprazole sodium
(pantoprazole-Na) had been studied most extensively. The authors concluded that omeprazole carried a
considerable potential for drug interactions because of its high affinity for CYP2C19 and moderate
affinity for CYP3A4, whereas pantoprazole-Na appeared to have a lower potential for interactions than
omeprazole based on extensive evidence. Lansoprazole and rabeprazole also seemed to have a weaker
potential for interactions than omeprazole, but this was based on limited evidence only. Much of the
review remains relevant today; however, several PPI drug interaction papers have been published since
2006. Thus, here we present an update of the 2006 review, which, when read in conjunction with the
original article, provides a comprehensive overview of drug interactions associated with the use of PPIs
[10].

This review is based on literature published from 1 January 2007 to 31 December 2012 identified by
searching (i) MEDLINE using Medical Subject Heading (MESH) terms for ‘drug-interactions’ and ‘proton
pump inhibitors’; and (ii) EMBASE using (Omeprazole/drug interaction) OR (Esomeprazole/drug
interaction) OR (Lansoprazole/drug interaction) OR (Pantoprazole/drug interaction) OR
(Rabeprazole/drug interaction) OR (Proton-Pump-Inhibitor/drug interaction). Searches were limited to
English language and excluded comments, editorials, letters, notes or conference papers or reviews.
PUBMED and EMBASE results were combined and duplicates removed; the remaining results were
divided into articles investigating PPI interactions with clopidogrel (where this term was used in the title,
abstract or as CAS number for MEDLNE or as descriptor for EMBASE) and other drug interaction articles.
Additional articles were also obtained from manual searches of the reference lists of relevant reviews
and papers. In total, 132 articles for interactions with clopidogrel and 174 articles for interactions with
other drugs were obtained. The two authors independently selected additional articles for inclusion
based on appropriate study design for drug-interaction studies, and any discrepancies were discussed
and agreed. Forty new references were identified and used in this updated review.

STUDIJA 4

Oncologist. 2012 Apr; 17(4): 550–554.

Published online 2012 Apr 3. doi: 10.1634/theoncologist.2011-0431

Accumulating Evidence for a Drug–Drug Interaction Between Methotrexate and Proton Pump Inhibitors

Shewit Bezabeh,corresponding authora Ann Corken Mackey,a Paul Kluetz,b Dilara Jappar,b and Joyce
Korvickb

A number of medications are known to interact with methotrexate through various mechanisms. The
aim of this article is to apprise practitioners of a new labeling change based on the accumulating
evidence for a possible drug–drug interaction between methotrexate (primarily at high doses) and
proton pump inhibitors (PPIs).

Methods.

The U.S. Food and Drug Administration (FDA) Adverse Event Reporting System (AERS) database of
spontaneous adverse event reports and the published literature were searched for cases reporting an
interaction between methotrexate and PPIs.

Results.

A search of the AERS database and existing literature found several individual case reports of drug–drug
interactions and three additional supportive studies that suggest potential underlying mechanisms for
the interaction.

Conclusion.

There is evidence to suggest that concomitant use of methotrexate (primarily at high doses) with PPIs
such as omeprazole, esomeprazole, and pantoprazole may decrease methotrexate clearance, leading to
elevated serum levels of methotrexate and/or its metabolite hydroxymethotrexate, possibly leading to
methotrexate toxicities. In several case reports, no methotrexate toxicity was found when a histamine
H2 blocker was substituted for a PPI. Based on the reviewed data, the FDA updated the methotrexate
label to include the possible drug–drug interaction between high-dose methotrexate and PPIs.
Physicians should be alerted to this potential drug–drug interaction in patients receiving concomitant
high-dose methotrexate and PPIs.

Keywords: Methotrexate, Proton pump inhibitor, Drug interaction

Introduction

Methotrexate was first approved by the U.S. Food and Drug Administration (FDA) in December 1953. It
inhibits dihydrofolic acid reductase and thus interferes with DNA synthesis, repair, and cellular
replication [1]. Methotrexate is used in a wide range of doses for the treatment of certain neoplastic
diseases, severe psoriasis, and rheumatoid arthritis [1]. High-dose methotrexate is used for the
treatment of malignancies such as high-grade lymphoma, osteosarcoma, and acute leukemia in doses of
300 mg/m2 to 12 g/m2. At higher dose levels, monitoring of serum methotrexate elimination is
performed because delayed elimination can result in serious and potentially life-threatening toxicities.
The FDA label for methotrexate contains a number of boxed warnings for proper administration and
monitoring, for various adverse events, and for potential drug–drug interactions. Besides methotrexate,
its major metabolite 7-hydroxymethotrexate may also contribute to toxicities [2].

A number of medications are known to interact with methotrexate therapy through various
mechanisms, and the drug is labeled for these drug–drug interactions [1]. Nonsteroidal anti-
inflammatory drugs (NSAIDs) have been reported to elevate and prolong serum methotrexate levels by
reducing tubular secretion. Other drugs, such as salicylates, phenylbutazone, phenytoin, and
sulfonamides, may increase methotrexate toxicity by displacing albumin-bound methotrexate.
Probenecid inhibits renal tubular transport, which can result in higher serum concentrations of
methotrexate. Oral antibiotics such as tetracycline, chloramphenicol, and nonabsorabable broad
spectrum antibiotics may decrease intestinal absorption of methotrexate or interfere with the
enterohepatic circulation by inhibiting bowel flora and suppressing metabolism of the drug by bacteria.
Finally, penicillin may reduce the renal clearance of methotrexate.

Proton pump inhibitors (PPIs) belong to the benzimidazole chemical family. Omeprazole (Prilosec®,
AstraZeneca, Wilmington, DE) was approved by the FDA in 1989 as the first PPI; other approved PPIs
include lansoprazole (Prevacid®, Takeda Pharmaceuticals U.S.A. Inc., Deerfield, IL), pantoprazole
(Protonix®, Wyeth Pharmaceuticals Inc., Philadelphia, PA), rabeprazole (Aciphex®, Eisai Inc., Woodcliff
Lake, NJ), esomeprazole (Nexium®, AstraZeneca, Wilmington, DE), omeprazole/bicarbonate (Zegerid®,
Santarus Inc., San Diego, CA), esomeprazole/naproxen (Vimovo®, AstraZeneca, Wilmington, DE), and
dexlansoprazole (Dexilant®, Takeda Pharmaceuticals U.S.A. Inc., Deerfield, IL). These agents inhibit
H+/K+-ATPase and decrease the secretion of gastric acid [3]. Some of the approved indications for PPIs
include the treatment of symptoms of gastroesophageal reflux disease, the treatment of erosive
esophagitis, the treatment of gastric ulcer, Helicobacter pylori eradication, and the reduction of NSAID-
associated gastric ulcers.

In this paper, we report the results of a literature and Adverse Events Reporting System (AERS) database
review of a potential drug–drug interaction between methotrexate and PPIs. These potential
interactions were noted for a number of different PPIs, including omeprazole, esomeprazole, and
pantoprazole.

Methods

The FDA maintains a database of adverse event reports associated with drugs and biologic products
used by humans. Drug manufacturers are required by law to submit such reports that come to their
attention. Health care professionals and consumers can voluntarily report via FDA's MedWatch program
[4]. The adverse drug event reports are collected, maintained, and retrieved through the AERS database
[5].

The AERS database was searched using the drug name methotrexate and the individual PPIs
(omeprazole, lansoprazole, pantoprazole, rabeprazole, esomeprazole, dexlansoprazole) for drug–drug
interactions from the initiation of marketing to July 31, 2011. The identified reports were then reviewed
for suspected drug–drug interactions between methotrexate and PPIs. In addition, the PubMed
database was searched on July 31, 2011 for articles on methotrexate and PPI-associated drug–drug
interactions and for articles on possible drug–drug interaction–related toxicity.

Results

AERS Database
A search of the AERS database identified several cases in which the reporters suspected an interaction
between methotrexate and a number of PPIs leading to methotrexate toxicity. The identified cases
include omeprazole (n = 7), esomeprazole (n = 5), and pantoprazole (n = 2). No cases of suspected
methotrexate interaction were identified for lansoprazole, rabeprazole, or dexlansoprazole in the AERS
database. When reported, clinical and laboratory outcomes included renal toxicity, hematologic events,
and myopathy. Two reports suggested a mechanism of PPI interference with methotrexate elimination.
However, most reporters did not provide blood levels or dechallenge/rechallenge information. Of the 14
cases, 13 cases were reported from foreign sources. Eleven patients were receiving methotrexate as
part of cancer chemotherapy; the other indications for use included psoriatic arthropathy (n = 1),
pulmonary fibrosis (n = 1), and rheumatoid arthritis (n = 1). Two cases of a positive dechallenge were
reported in 2009–2010 from foreign sources, and are described below.

A pharmacist reported that a 47-year-old male patient being treated for Burkitt's lymphoma
experienced decreased methotrexate clearance (drug levels were recorded but not provided in the
report) when esomeprazole was added to the patient's chemotherapy regimen. The patient was
hospitalized (the adverse clinical event associated with the decreased renal clearance was not reported).
His methotrexate clearance normalized after esomeprazole was omitted during his second
chemotherapy course.

A physician reported that a 15-year-old male patient being treated for large cell lymphoma experienced
decreased elimination of methotrexate while on concomitant omeprazole. The report stated that
methotrexate elimination was normal during the first two cycles, but elimination lasted 1 week instead
of 2 days (blood levels were provided) after omeprazole was added to the third cycle. The patient
experienced no manifestations of toxicity associated with the delayed methotrexate elimination. During
the subsequent two methotrexate cycles, omeprazole was discontinued and methotrexate elimination
was reportedly normal.

Literature

A search of the existing literature identified four case reports describing a possible interaction between
methotrexate and PPIs.

The first case of a possible interaction between omeprazole and high-dose methotrexate was reported
by Reid et al. [6] in 1993 in a patient with osteosarcoma. The patient received high-dose methotrexate
with leucovorin rescue while receiving omeprazole and other drugs. During the first course of therapy,
his serum methotrexate levels remained elevated for several days, prompting prolonged administration
of hydration, urine alkalinization, and leucovorin therapy for 8 days. After omeprazole was discontinued,
the patient's serum methotrexate levels rapidly declined, with normal kinetics for the next three cycles
of methotrexate therapy.

Beorlegui et al. [7] reported a case of delayed elimination of high-dose methotrexate associated with
concomitant omeprazole administration. An 11-year-old male patient experienced delayed
methotrexate elimination after receiving methotrexate to treat osteoblastic osteosarcoma with
concomitant omeprazole (40 mg a day), megestrol, and sucralfate. Delayed elimination was determined
via methotrexate plasma concentrations. The clinicians suspected an interaction with omeprazole
leading to altered methotrexate levels and substituted ranitidine. Methotrexate elimination was normal
in subsequent cycles administered with concomitant ranitidine in place of omeprazole.

A case was reported by Tröger et al [8] involving a 59-year-old male patient who developed symptoms of
myalgia with methotrexate toxicity. The patient had underlying folliculotropic cutaneous T-cell
lymphoma and was receiving low-dose pulse methotrexate therapy (15 mg i.m., once a week) along with
pantoprazole to treat Barrett's esophagus. This case described positive dechallenge and rechallenge
results. The patient experienced myalgia that lasted several days with symptoms recurring over the
following four methotrexate cycles. When pantoprazole was discontinued and ranitidine was given in its
place, the symptoms of myalgia subsided and eventually disappeared. The patient was rechallenged
with pantoprazole 8 weeks later with the patient's consent and the symptoms recurred. Serum
concentrations of methotrexate and its metabolite 7-hydroxymethotrexate were monitored with and
without pantoprazole. The investigator found that the concentration–time curves were identical for
methotrexate in both periods but differed considerably for 7-hydroxymethotrexate. The area under the
curve between 0 and 144 hours was ∼70% higher for 7-hydroxymethotrexate when methotrexate was
given with pantoprazole than without pantoprazole, and the half-life was doubled when pantoprazole
was given with methotrexate.

Bauters et al. [9] described a 15-year-old male patient with underlying acute lymphocytic leukemia who
experienced severe mucositis after taking methotrexate (5 g/m2) and omeprazole concomitantly. In the
first cycle, he received high-dose methotrexate and ranitidine and experienced no problems. In the
second cycle, omeprazole was substituted for ranitidine. During that cycle, his methotrexate clearance
was delayed and his methotrexate levels remained elevated for several days (blood levels provided). He
experienced severe mucositis, but his creatinine and liver function tests were normal. Omeprazole was
discontinued and the mucositis resolved. He experienced normal methotrexate clearance with the third
cycle of therapy, given without omeprazole.

The literature search also identified several articles, including some that assessed the potential
underlying mechanism for the interaction.

Breast cancer resistant protein (BCRP) is present in the kidney [10]. It is thought that BCRP is responsible
for methotrexate secretion in the kidneys during the process of renal excretion of methotrexate.
Breedveld et al. [11] reported on their investigation of the mechanism of interaction in vitro in
membrane vesicles from cells infected with a baculovirus containing human BCRP. They found that
benzimidazoles differentially affect transport of methotrexate mediated by BCRP and multidrug
resistance associated protein, and they concluded that competition for BCRP may explain the interaction
between methotrexate and benzimidazoles.

An article by Suzuki et al. [12] reported retrospective data on plasma methotrexate concentrations for
171 cycles of high-dose methotrexate therapy in 74 patients. They conducted a multiple logistic
regression analysis that supported coadministration of PPIs (i.e., omeprazole, lansoprazole, rabeprazole,
pantoprazole) as a risk factor for delayed elimination as well as renal and liver dysfunction. Although all
four PPIs inhibited BCRP-mediated transport of methotrexate, the half-maximal inhibitory
concentrations were higher than the plasma concentrations of PPIs. These authors concluded that the
drug–drug interaction is likely not solely a result of the PPI effects on BCRP-mediated methotrexate
transport.

Joerger et al. [13] reported on a 24- and 48-hour blood sample analysis from 76 patients receiving high-
dose methotrexate in which the concentration–time data were subjected to population pharmacokinetic
and covariate analyses. The study concluded that, in patients receiving high-dose methotrexate,
concurrent administration of a benzimidazole was associated with a significant decrease in the clearance
of methotrexate and its metabolite hydroxymethotrexate, resulting in higher plasma concentrations.

Santucci et al. [14] retrospectively analyzed the causes of delayed methotrexate elimination in six
patients who had received glucarpidase rescue. Of those who had received a PPI (three of six),
rechallenge in the absence of the PPI resulted in normal elimination. The same authors then conducted
a larger retrospective cohort study in patients receiving methotrexate at doses >1 g/m2 [15]. The
retrospective study revealed that, of the 79 patients who were treated at their institution with a total of
197 cycles, 32 cycles (16%) displayed delayed elimination (>15 micromole/L at 24 hours, >1.5
micromole/L at 48 hours, or >0.15 micromole/L at 72 hours). Of these delayed cycles, coadministration
of a PPI occurred in 17 of 32 cycles (53%) whereas co administration of a PPI occurred in only 24 of 165
or 15% of the cycles in which elimination was not delayed.

As opposed to the above described case reports and series that suggest an interaction between
methotrexate and PPIs, Whelan et al. [16] reported a case of no drug interaction between methotrexate
and omeprazole. The case involved a 24-year-old male patient with underlying chondroblastic
osteosarcoma who was receiving methotrexate and omeprazole. The patient was receiving omeprazole
for 2 months for dyspeptic symptoms. Methotrexate blood levels were elevated at 67 mmol/L at 24
hours, prompting an adjustment of the folinic acid dose. Continued monitoring of the methotrexate
level demonstrated delayed elimination, with the level falling to <0.1 mmol/L by 140 hours. Omeprazole
was omitted from the second cycle of methotrexate (using the same methotrexate dose, hydration, and
other medications) and his serum levels measured at 24 hours revealed delayed methotrexate clearance
identical to what was seen in the first course. However, there were no reported levels of the metabolite
7-hydroxymethotrexate. Additionally, a prospective evaluation of low-dose methotrexate in 28
rheumatoid arthritis patients receiving lansoprazole and naproxen for 7 days did not reveal any change
in methotrexate pharmacokinetics [17].

Discussion

Methotrexate is eliminated primarily via renal excretion. Methotrexate clearance rates vary widely and
are generally lower at higher doses [18]. This review identified cases of decreased methotrexate
clearance with some symptoms of renal toxicity, hematologic events, mucositis, and myalgia possibly
resulting from drug–drug interactions with various PPIs. The frequency of the interaction is difficult to
ascertain. The AERS database cannot be used to estimate the true incidence rates of events because
spontaneous reporting systems such as the AERS are subject to underreporting. In fact, it has been
suggested that <10% of adverse events are reported to the FDA [19].

Two mechanisms for PPI-induced interference with methotrexate elimination have been proposed in
the literature. First, there is evidence that H+/K+-ATPase is present in renal epithelium as well as gastric
parietal cells [20]. It was proposed that PPIs inhibit renal H+/K+-ATPase, which supports the active
tubular secretion of methotrexate, resulting in an increased half-life of methotrexate [6]. However, it
was also found that the administration of omeprazole does not change the pH of urine, suggesting the
absence of a significant effect of PPIs on renal H+/K+-ATPase [21]. Therefore, further research is needed
to evaluate the validity of this hypothesis for this first proposed mechanism. A second proposed
mechanism involves possible PPI inhibition of ATP-dependent efflux of methotrexate by BCRP in human
kidney proximal tubules. Although it was demonstrated in vitro that PPIs inhibit BCRP-mediated
transport of methotrexate, the clinical relevance of this interaction is uncertain because the half-
maximal inhibitory concentrations of PPIs are significantly higher than the therapeutic unbound plasma
concentration [10].

The majority of the reported cases occurred with the administration of high-dose methotrexate.
Although the product label describes high-dose therapy as 12 g/m2 as a 4-hour infusion [1], doses ≥500
mg/m2 are frequently cited as “high dose” in the literature, and the majority of the above case reports
and series were reported from patients receiving doses of 300 mg/m2 to 12 g/m2. It appears the risk is
greatest in the high-dose setting; however, there were also cases of patients taking a PPI and
experiencing toxicity at doses as low as 15 mg of methotrexate per week.

Based on reports in the literature and evidence from the AERS, it appears that PPIs may interfere with
the elimination of methotrexate (primarily at high doses), possibly leading to accumulation of
methotrexate and its metabolite hydroxymethotrexate. Decreased clearance of methotrexate and its
metabolite may result in an increased risk for known methotrexate toxicities, some of which can be
severe and life threatening. This review cites one report of a patient who did not experience delayed
elimination after receiving methotrexate and omeprazole, but the report did not mention any
measurement of the metabolite hydroxymethotrexate.

Conclusion

There is accumulating evidence to suggest that concomitant use of methotrexate (primarily at high
doses) and PPIs such as omeprazole, esomeprazole, and pantoprazole may decrease methotrexate
clearance. Decreased clearance may result in elevated and prolonged serum levels of methotrexate
and/or its metabolite hydroxymethotrexate, possibly leading to methotrexate toxicities. In addition,
there may be a class effect because this interaction has been described for multiple PPIs.

PPIs are used extensively, and health care practitioners should be aware of this potential interaction. In
several case reports, methotrexate elimination normalized and no methotrexate toxicity was found
when a histamine H2 blocker was substituted for a PPI, although no formal studies have been
conducted. This possible drug–drug interaction has been added to the labels for i.v. methotrexate and
PPIs. Clinicians should consider substituting H2 blockers for PPIs when acid suppression is clinically
indicated during methotrexate therapy. The FDA encourages reporting of suspected cases of drug–drug
interactions to their MedWatch program (http://www.fda.gov/medwatch).

STUDIJA 5

Br J Clin Pharmacol. 2009 Jan; 67(1): 44–49.

doi: 10.1111/j.1365-2125.2008.03303.x

Co-administration of proton pump inhibitors delays elimination of plasma methotrexate in high-dose

methotrexate therapy

Kunihiro Suzuki,1,2 Kosuke Doki,2 Masato Homma,1,2 Hirofumi Tamaki,3 Satoko Hori,3 Hisakazu
Ohtani,3 Yasufumi Sawada,3 and Yukinao Kohda1,2

Abstract

To assess whether or not co-administration of proton pump inhibitors (PPIs) is a risk factor for delayed
elimination of plasma methotrexate (MTX) in high-dose MTX (HDMTX) therapy for malignant diseases.

METHODS

To assess the effects of PPI co-administration on elimination of plasma MTX, we examined plasma MTX
concentration data on 171 cycles of HDMTX therapy performed in 74 patients. We performed multiple
logistic regression analysis to evaluate PPI co-administration as a risk factor. Inhibitory potencies of
omeprazole, lansoprazole, rabeprazole and pantoprazole on MTX transport via breast cancer resistance
protein (BCRP, ABCG2) were also investigated in an in vitro study using membrane vesicles expressing
human BCRP.

RESULTS

We identified co-administration of PPIs as a risk factor for delayed elimination (odds ratio 2.65, 95%
confidence interval 1.03, 6.82) as well as renal and liver dysfunction. All four PPIs inhibited BCRP-
mediated transport of MTX, with half-maximal inhibitory concentrations of 5.5–17.6 µM – considerably
higher than the unbound plasma concentrations of the PPIs.

CONCLUSIONS

Our results support previous findings suggesting that PPI co-administration is associated with delayed
elimination of plasma MTX in patients with HDMTX therapy. This drug interaction, however, cannot be
explained solely by the inhibitory effects of PPIs on BCRP-mediated MTX transport.

Keywords: ABCG2, breast cancer resistance protein, delayed elimination, methotrexate, multiple logistic
regression analysis, proton pump inhibitors

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

Co-administration of proton pump inhibitors (PPIs) increases plasma methotrexate (MTX) concentration
in cancer patients receiving high-dose MTX (HDMTX) therapy.

There is controversy as to whether or not co-administration of PPIs affects plasma MTX elimination in
HDMTX therapy.

Inhibitory activity of PPIs on breast cancer resistance protein (BCRP) is a possible mechanism for the
drug interaction between MTX and PPIs.

WHAT THIS STUDY ADDS

Co-administration of a PPI (omeprazole, lansoprazole, or rabeprazole) was more frequently observed in

the delayed MTX elimination group than in the normal MTX elimination group.

Multiple logistic regression analysis with adjustment for significant covariates revealed that PPI co-
administration was a significant risk factor for delayed plasma MTX elimination.

The half-maximal inhibitory concentration of each PPI in inhibiting BCRP function was much higher than
the therapeutic unbound concentration in the plasma.

Introduction

High doses of methotrexate (MTX), an antifolate drug, are an accepted treatment for lymphoid
malignancy, osteogenic sarcoma and acute leukaemia [1]. Therapeutic drug monitoring of MTX is
essential to prevent toxicity from high plasma MTX concentrations [2], a problem that commonly
requires rescue administration of calcium folinate. Large inter- and intraindividual variations in plasma
MTX elimination are associated with co-administration of nonsteroidal anti-inflammatory drugs (NSAIDs)
[3, 4] and vancomycin [5], as well as renal dysfunction, third space fluid accumulations such as pleural
effusion and ascites, and insufficient hydration [1]. A recent report has suggested that co-administration
of proton pump inhibitors (PPIs), including omeprazole and lansoprazole, decreased MTX clearance,
resulting in delayed plasma MTX elimination [6]. In contrast, one case report showed that omeprazole
did not alter MTX clearance [7]. It is therefore controversial as to whether or not co-administration of
PPIs affects elimination of plasma MTX in high-dose MTX (HDMTX) therapy. Because membrane
transport of MTX is known to be mediated by breast cancer resistance protein (BCRP, ABCG2) and
multidrug-resistance-related protein, the inhibitory effects of PPIs on these transporters may be
involved in the mechanism of the PPI–MTX interaction [8].

We therefore used multiple logistic regression analysis to examine the impact of PPI co-administration
on plasma MTX elimination in patients undergoing HDMTX therapy. We also assessed the inhibitory
effects of four PPIs (omeprazole, lansoprazole, rabeprazole and pantoprazole) on BCRP-mediated
transport of MTX in an in vitro study of membrane vesicles expressing human BCRP.

Methods

Patients
Plasma MTX data on 171 cycles of HDMTX therapy were evaluated in 74 patients (45 male and 29
female) treated at Tsukuba University Hospital (Ibaraki, Japan). Twenty-five HDMTX treatment protocols
with a median MTX dose of 3500 mg m−2 (range 1000–5000 mg m−2) for malignant lymphoma,
leukaemia, chronic active Epstein–Barr virus infection or plasmacytoma were included. Median HDMTX
infusion duration was 6 h (range 2.9–29.0 h). The patients were well hydrated to keep the urine pH > 7
during treatment. Intravenous or oral calcium folinate rescue was conducted when the plasma MTX
concentration was high, according to each protocol guideline. Any suspected presence of pleural
effusion and ascites was confirmed by radiography or computed tomography. Laboratory data such as
serum aspartate aminotransferase (AST) and alanine aminotransferase (ALT), serum creatinine (SCr) and
blood urea nitrogen levels for the assessment of liver and kidney function were determined at baseline.
Cut-off values for judging liver and kidney dysfunction were estimated in accordance with the
corresponding normal laboratory data ranges used at our hospital.

The study was approved by the ethics committee of the University of Tsukuba. Informed consent was
obtained from the patients.

Determination of MTX concentration for group categorization

Plasma MTX concentrations were determined 24, 48 and/or 72 h after the start of HDMTX therapy. A
TDx Abbott analyser (Abbott Japan, Tokyo, Japan) with a detection limit of 0.02 µmol l−1 was used for
MTX assay. The 171 cycles of HDMTX therapy were categorized into delayed (n = 41) and normal (n =
130) elimination groups according to the following definitions of toxic MTX concentrations at each
measurement time. Patients with plasma MTX concentrations ≥10 µmol l−1 at 24 h after the start of
HDMTX therapy, ≥1 µmol l−1 at 48 h or ≥0.1 µmol l−1 at 72 h were classed in the delayed elimination
group [9, 10]. In those patients in whom the duration of infusion exceeded 24 h we did not use the
plasma MTX concentration 24 h after the start of HDMTX in our assessment of delayed elimination.

Statistical analysis

Differences in patient characteristics and MTX dosing regimen were compared between the delayed and
normal elimination groups by the χ2 test, Fisher's exact probability test or Mann–Whitney U-test.
Multiple logistic regression analysis was performed to identify possible risk factors for delayed
elimination of plasma MTX. Six variables for which there were significant differences between the
normal and delayed elimination groups (Table 1) were subjected to stepwise logistic analysis to establish
the best predictive model. The odds ratio (OR) and its 95% confidence interval (CI) were calculated by
StatView 5.0 (SAS Institute, Cary, NC, USA). A P-value < 0.05 was considered to be statistically significant.

Inhibitory effects of PPIs on BCRP

BCRP-expressing membrane vesicles, which express human BCRP on their membranes in an ‘inside-out’
manner, were purchased from SOLVO Biotechnology (Budapest, Hungary). 3H-MTX [3′,5′,7′-3H(N)] was
purchased from Moravek Biochemicals, Inc. (Brea, CA, USA). Uptake of 3H-MTX into the membrane
vesicles was measured by a rapid filtration technique. The membrane vesicles were incubated for 2 min
at 37°C in uptake buffer (pH 7.0) containing 100 µmol l−1 MTX and PPI (5, 20, or 100 µmol l−1).
Adenosine triphosphate (ATP)-dependent uptake of MTX was determined by subtracting the uptake in
the absence from that in the presence of ATP.

Results

We evaluated 311 measurements of plasma MTX in 171 cycles of HDMTX therapy. Plasma MTX
concentrations in the delayed elimination group were significantly higher than those in the normal
elimination group (median 11.50 vs. 1.10 µmol l−1 at 24 h after the start of HDMTX therapy; 0.87 vs.
0.11 µmol l−1 at 48 h; 0.23 vs. 0.05 µmol l−1 at 72 h, P < 0.05, Table 1), resulting in the need for frequent
calcium folinate rescue (median 24 vs. 12 rescues, P < 0.05).

Patient characteristics and MTX dosing regimen were compared between the delayed elimination group
and the normal elimination group (Table 1). Gender, age, weight and MTX dose did not differ between
the groups. Abnormally high SCr and serum AST and ALT concentrations were significantly more
frequent in the delayed elimination group than in the normal elimination group (19.5% vs. 5.4%, 31.7%
vs. 8.5%, and 41.5% vs. 25.4%, respectively, P < 0.05). Infusion time for HDMTX was significantly longer
in the delayed elimination group than in the normal elimination group (median 7.3 vs. 6 h, P < 0.05).
Pleural effusion and ascites were significantly more frequent in the delayed elimination group than in
the normal elimination group (29.3% vs. 10.8%, P< 0.05). Co-administration of PPIs, including
omeprazole, lansoprazole and rabeprazole, was also significantly more frequent in the delayed
elimination group than in the normal elimination group (31.7% vs. 13.8%, P < 0.05). No difference was
found in NSAID or vancomycin co-administration between the groups.

Multiple logistic regression analysis was performed to identify the significant risk factors for delayed
plasma MTX elimination (Figure 1). After adjustment for six variables – PPI co-administration, pleural
effusion and ascites, MTX infusion time, and SCr, AST and ALT levels – PPI co-administration was still a
significant risk factor for delayed elimination (adjusted OR 2.65; 95% CI 1.03, 6.82), as well as SCr
elevation (adjusted OR 4.60; 95% CI 1.31, 16.16) and AST elevation (adjusted OR 4.12; 95% CI 1.19,
14.28).

Adjusted odds ratios for delayed methotrexate elimination

To confirm the significant effects of PPI on MTX elimination, the incidence of delayed plasma MTX
elimination was also compared between HDMTX therapy with (n = 31) and without (n = 140) PPI co-
administration. Delayed elimination of plasma MTX was more frequent in HDMTX therapy with PPI co-
administration than without PPI co-administration (41.9% vs. 20.0%, P < 0.05).

All four PPIs inhibited the ATP-dependent uptake of MTX into BCRP-expressing membrane vesicles in a
concentration-dependent manner, with half-maximal inhibitory concentrations (IC50 values) of 17.6,
14.4, 8.5 and 5.5 µmol l−1 for omeprazole, lansoprazole, rabeprazole and pantoprazole, respectively
(Figure 2). Inhibitory potencies, as shown by the IC50 values and pharmacokinetic parameters, were
almost the same among omeprazole, lansoprazole and rabeprazole, which are the three PPIs used in
Japan (Table 2).
Discussion

Multiple logistic regression analysis revealed that PPI co-administration was a risk factor for delayed
elimination of MTX in HDMTX therapy and for kidney and liver dysfunction. PPI co-administration
increased the risk of delayed MTX elimination by 2.65 times (Figure 1). These results support the
previous finding that concurrent administration of PPIs causes higher plasma MTX concentration in
HDMTX therapy [6].

Although other variables, such as co-administration of NSAIDs or vancomycin and pleural effusion and
ascites, have been associated with delayed MTX elimination in HDMTX therapy, their contribution to
MTX elimination was not statistically significant here. NSAIDs are considered to suppress renal MTX
excretion by inhibiting renal prostaglandin synthesis, and vancomycin acts by inducing nephrotoxicity [3,
5]. However, unlike PPI co-administration, these effects were not factors strongly affecting MTX
elimination. Joerger et al. reported that PPI co-administration decreased MTX clearance by 27% –
greater than that induced by NSAID co-administration (16%) [6]. Compatible with these observations,
our results also show that the contribution of PPI co-administration to delayed plasma MTX elimination
was greater than that of NSAIDs.

HDMTX therapy is commonly avoided in patients with third space effusions, pleural effusions and ascites
[1]. However, 26 cycles of HDMTX were given to 19 patients with third space effusions because they had
no other therapeutic options to control their malignancies. The presence of third space effusions did not
significantly affect the elimination of MTX, a result similar to those of a previous report [6].

BCRP is expressed on the apical membranes of renal epithelial cells and is associated with ATP-driven
efflux of MTX [8]. All four PPIs showed in vitro inhibition of MTX transport via BCRP. However, their IC50
values were 50–200 times their therapeutic unbound plasma concentrations in human subjects (Table
2), suggesting that the interaction between MTX and PPIs cannot be explained solely by the inhibitory
effects of PPIs on renal BCRP. It may be possible that other transporters responsible for the renal
excretion of MTX, such as organic anion transporter 3, multidrug resistance-associated protein (MRP) 2
and MRP4 [16], are involved in this drug interaction, although the inhibitory effects of PPIs on these
transporters remain to be investigated.

It is considered that BCRP is responsible for MTX secretion by the kidneys during the process of renal
excretion of MTX. We estimated the renal clearance for MTX secretion to be 64.5 ml min−1 by
subtracting the glomerular filtration clearance of MTX (59.5 ml min−1), which is calculated from the
glomerular filtration rate (120 ml min−1) and the plasma unbound fraction (fu; 49.6%) [17], from the
renal clearance of MTX (124 ml min−1) [6, 18]. Assuming that renal reabsorption of MTX is negligible,
the maximum contribution of MTX secretion to MTX renal clearance would be 52.0%. However, because
transporters other than BCRP may also be involved in the renal secretion of MTX, the contribution of
BCRP itself may be <52.0%. In any case, renal clearance of MTX may not be totally impaired even if renal
BCRP function is abolished by PPI co-administration.
Delayed elimination of plasma MTX was not observed in every patient receiving PPIs; the reason is
unclear. Polymorphism of CYP2C19, a principal enzyme involved in the metabolism of PPIs [19], can be
considered a potential explanation if the drug interaction is plasma PPI concentration-dependent.
Subjects with CYP2C19*2 and *3 mutated alleles, both of which provide higher plasma concentrations of
PPIs [20], may be susceptible to the drug interaction. Another genetic risk factor for this drug interaction
may involve polymorphism of BCRP [21].

In conclusion, by using multiple logistic regression analysis we have confirmed that PPI co-administration
was a possible risk factor for delayed plasma MTX elimination. The mechanisms of this delayed
elimination, including the genetic factors and the contribution of other MTX transporters, remain to be
clarified in further investigations.

STUDIJA 7

BMJ. 2002 Jun 22; 324(7352): 1497.

Severe myalgia from an interaction between treatments with pantoprazole and methotrexate

U Tröger Institute of Clinical Pharmacology

B Stötzel Department of Dermatology and Venerology

J Martens-Lobenhoffer Institute of Clinical Pharmacology

H Gollnick Department of Dermatology and Venerology

F P Meyer Institute of Clinical Pharmacology, Hospital of the Otto-von-Guericke-University, D-39120

Magdeburg, Germany

A 59 year old man with a folliculotropic cutaneous T cell lymphoma had been receiving treatment with
interferon alfa-2a. After the treatment was discontinued because his tumour had relapsed, he started
taking low dose pulse methotrexate (15 mg intramuscularly, once a week). Concomitantly he also had a
Barrett oesophagus, which was treated with pantoprazole (20 mg/day, orally), and arterial hypertension,
which was treated with atenolol. After the first injection of methotrexate the patient had severe
generalised myalgia and bone pain. This led to partial immobility that began three to four hours after
the injection and continued—albeit to a lesser degree—for several days. The symptoms recurred over
the following four methotrexate cycles. Multiple clinical and laboratory examinations excluded clinical
causes such as an underlying focal or disseminated infection and the systemic exacerbation of the
cutaneous T cell lymphoma. A drug interaction with methotrexate was suspected because the symptoms
arose when the drug was given. After pantoprazole was replaced by ranitidine, the symptoms subsided
dramatically and finally disappeared.

The patient was rechallenged with pantoprazole eight weeks later, having given his informed consent.
Treatment with pantoprazole was started six days before the weekly dose of methotrexate was given so
that the maximum effect could be attained. Serum concentrations of methotrexate and its metabolite 7-
hydroxymethotrexate were monitored with and without pantoprazole by a high performance liquid
chromatography-fluorescence assay.1 The two study periods were separated by a washout phase of six
weeks. Atenolol, clemastine to treat local itching, and ascorbic acid to treat a temporary deficiency of
vitamin C were given concomitantly.

The symptoms reappeared in response to the challenge but did not reappear without pantoprazole. The
concentration-time curves (figure) are identical for methotrexate in both periods, but they differ
considerably for 7-hydroxymethotrexate. The area under the curve between 0 and 144 h after drug
administration was nearly 70% higher for 7-hydroxymethotrexate plus pantoprazole than without
pantoprazole (1929 ng·h/ml v 1131 ng·h/ml); the half life of 7-hydroxymethotrexate was doubled when
methotrexate was given with pantoprazole (81.4 h v 36.4 h). This indicates an interaction in renal
elimination, rather than a metabolic interaction.

About 30% of all patients discontinue the low dose pulse methotrexate because of adverse effects. Drug
interactions contribute considerably to the number of patients who stop using the drug.2,3 Reactions
after dosing occur in about 10-15% of patients with rheumatoid arthritis receiving this treatment.4,5
This report shows for the first time a link between methotrexate post-dosing reactions and an
interaction of 7-hydroxymethotrexate and concomitantly given pantoprazole. The German Federal
Institute for Drugs and Medical Devices and the Drug Commission of the German Medical Profession
were informed about the case. The manufacturer of methotrexate (Medac GmbH, Wedel, Germany) was
not aware of this drug interaction.

STUDIJA 8

J Transl Med. 2013; 11: 268.

Published online 2013 Oct 24. doi: 10.1186/1479-5876-11-268

Proton pump inhibitor chemosensitization in human osteosarcoma: from the bench to the patients’ bed

Stefano Ferrari,1 Francesca Perut,2 Franca Fagioli,3 Adalberto Brach Del Prever,3 Cristina Meazza,4
Antonina Parafioriti,5 Piero Picci,6 Marco Gambarotti,7 Sofia Avnet,2 Nicola Baldini,2 and Stefano
Faiscorresponding author8

Abstract

Major goals in translational oncology are to reduce systemic toxicity of current anticancer strategies and
improve effectiveness. An extremely efficient cancer cell mechanism to avoid and/or reduce the effects
of highly cytotoxic drugs is the establishment of an acidic microenvironment, an hallmark of all
malignant tumors. The H + −rich milieu that anticancer drugs meet once they get inside the tumor leads
to their protonation and neutralization, therefore hindering their access into tumor cells. We have
previously shown that proton pump inhibitors (PPI) may efficiently counterattack this tumor advantage
leading to a consistent chemosensitization of tumors. In this study, we investigated the effects of PPI in
chemosensitizing osteosarcoma.

Method

MG-63 and Saos-2 cell lines were used as human osteosarcoma models. Cell proliferation after
pretreatment with PPI and subsequent treatment with cisplatin was evaluated by using erythrosin B dye
vital staining. Tumour growth was evaluated in xenograft treated with cisplatin after PPI pretreatment.
Subsequently, a multi-centre historically controlled trial, was performed to evaluate the activity of a pre-
treatment administration of PPIs as chemosensitizers during neoadjuvant chemotherapy based on
methotrexate, cisplatin, and adriamycin.

Results

Preclinical experiments showed that PPI sensitize both human osteosarcoma cell lines and xenografts to
cisplatin. A clinical study subsequently showed that pretreatment with PPI drug esomeprazole leads to
an increase in the local effect of chemotherapy, as expressed by percentage of tumor necrosis. This was
particularly evident in chondroblastic osteosarcoma, an histological subtype that normally shows a poor
histological response. Notably, no significant increase in toxicity was recorded in PPI treated patients.

Conclusion

This study provides the first evidence that PPI may be beneficially added to standard regimens in
combination to conventional chemotherapy.

Introduction

Low pH is a major cause of tumor unresponsiveness to the vast majority of cytotoxic drugs mostly due to
the fact the H + −rich tumor microenvironment leads to protonation of the drug causing both its
neutralization outside the cells and abrogation of drug entry within the target cell [1]. The prime cause
of tumor microenvironment acidification is the Warburg effect, which leads to overproduction of lactic
acid by malignant tumors. However, this condition progressively selects cells that live in the acidic
microenvironment due to overexpression and function of proton pumps that avoid intracellular
acidification [2]. Vacuolar-type H + −ATPases play a key role in the acidification of tumor
microenvironment [2]. Under experimental conditions, pretreatment of drug-resistant tumor cells with
proton pump inhibitors (PPI) increases tumor cells sensitivity to a variety of anticancer drugs [3].
Pretreatment with PPI, followed by the administration of anticancer drug, both in vitro and in vivo,
resulted in the most efficient approach. This effect was correlated with both an inhibition of ATPase
activity and a PPI-induced marked increase in drug retention within tumor cells. PPI-mediated
chemosensitization occured independently of both tumor histology and the type of cytotoxic drug [3].

More recently, it has been shown that the in vivo PPI-mediated effect on human tumors is transient,
particularly in terms of pH gradient at the cellular level [4]. Lastly, PPI were able to increase sensitivity of
cells responding to standard chemotherapy, but also to revert multidrug resistance (MDR) [3]. A PPI-
based approach might therefore be extremely interesting to test both in solid tumors unresponsive to
drugs and in tumors responding to chemotherapy but undergoing MDR and paying also the price of high
level of toxicity of very aggressive chemotherapy. Osteosarcoma is a rare tumor with an overall
incidence of 0.2 new cases/100,000. It is more frequently diagnosed in adolescents and young adults
where it accounts for >10% of all solid cancers [5]. Currenty, strategy is based on a combination of
surgery and chemotherapy. Chemotherapy is delivered before and after surgical removal of the tumor
(neoadjuvant chemotherapy). The most effective drugs used for osteosarcoma are methotrexate (MTX),
cisplatin (CDP), doxorubicin (ADM), and ifosfamide (IFO). In case of patients without evident metastatic
disease at presentation the event-free survival is 60% at 5 years [6]. Late (>5 years) relapse are
uncommon in patients with osteosarcoma (6,7). Inadequate surgical control of the tumor leads to local
recurrence in about 5% of patients. In about 35% of patients tumor resistance towards the four-drug
combination therapy is responsible for the failure of the systemic chemotherapy treatment [6]. The
rationale for neoadjuvant chemotherapy is based on an early use of chemotherapy and on the
possibility to assess chemosensitivity by means of histological evaluation of the chemotherapy-induced
tumor necrosis on the surgical specimen [7]. There is a strong correlation between chemotherapy-
induced tumor necrosis and prognosis in patients with osteosarcoma [8] with a higher probability of
disease-free survival obtained in patients having a good histologic tumor response to neoadjuvant
chemotherapy [6,7]. Hence, the histologic response to neoadjuvant chemotherapy is a very important
predictive factor of survival and a reliable parameter of chemosensitivity. Previous studies have shown
that the preoperative chemotherapy dose-intensification [9] or the preoperative deliver of intra-arterial
cisplatin [10,11] are able to increase the percentage of cases with a good response to neoadjuvant
chemotherapy. Both strategies are however associated with more severe side effects and discomfort for
the patient [9,11]. Preclinical data showing that pretreatment of drug-resistant tumor cells with proton
pump inhibitors (PPI), renders tumor cells more sensitive to a variety of anticancer drugs [3] have
suggested that the administration of PPI as chemosensitizers might be an innovative approach to
increase the sensitivity of osteosarcoma cancer cells to the currently used drugs. Moreover, an in vivo
study performed in companion animals with spontaneous occurring tumors, including osteosarcoma,
has shown an amazing clinical response to high dosage PPI/chemotherapy combination [12].

Based on the hypothesis that microenvironmental acidity may represent a key factor in tumor
homeostasis, mostly involved in resistance to cytotoxic drugs, this study presents preclinical and clinical
data evaluating the effectiveness of PPIs as chemosensitizer against human osteosarcoma. More
specifically, this study was aimed at exploring the potential use of high dose PPI pretreatment in
osteosarcoma patients undergoing neoadjuvant chemotherapy.

Materials and methods

In vitro studies

Cells
Continuous cell lines from human osteosarcoma (Saos-2, MG-63), obtained from the American Type
Culture Collection (ATCC, Rockville, MD), were maintained in Iscove's Modified Dulbecco's Medium
(IMDM, Invitrogen, Carlsbad, CA), supplemented with 10% fetal bovine serum (FBS, Mascia Brunelli,
Milan, Italy), penicillin (100 U/ml) and streptomycin (100 mg/ml) (Invitrogen) at 37°C and 5% CO2. Only
cells in exponential growth phase were used.

Cell proliferation assay

Cells were seeded in duplicate in 12-well plates (3x104 cells/well for Saos-2 and 2x104 cells/well for MG-
63) in unbuffered medium. After 24 h the culture medium was replaced with fresh medium containing
60 μM esomeprazole (ESOM) (Sigma-Aldrich) dissolved in DMSO. As controls, cells were incubated with
medium at the same concentration of DMSO. After 24 h the culture medium was replaced with fresh
medium containing 0.5-5-50 μM cisplatin (Sigma-Aldrich). After 72 h, pH of the medium was tested and
cells were harvested. The number of viable cells was evaluated by the erythrosin B dye vital staining
[13]. Results were expressed as growth inhibition ratio in respect to cells in 0.5 μM cisplatin medium
without ESOM pretreatment. The experiment was repeated three times in duplicate.

Statistical analysis

Statistical analysis was performed using the StatView™ 5.0.1 software for Windows (SAS Institute, Cary,
NC). Differences were analysed using the non-parametric Wilcoxon Signed Rank test (significant level of
p < 0.05).

In vivo study

Female CB.17 SCID/SCID mice aged 4–5 weeks (Harlan, Italy) were kept under specific pathogen-free
conditions and fed ad libitum. Mice were housed in micro-isolator cages, and all food, water, and
bedding were autoclaved prior to use. Each mouse was injected subcutaneously in the right flank with 3
x106 human osteosarcoma cells (Saos-2) that had been resuspended in 0.2 mL of RPMI-1640 containing
10% FCS. Once tumors became evident (at least 0.10 cm, approximately 10 days after the tumor cell
injection), ESOM was administered by i.p. injection [4] at a dose of 25 mg/kg. Cisplatin was administered
weekly by i.p. injection at a dose of 5 mg/kg. Tumor size (mm3) was measured three times per week
with calipers with the formula length x width2. Morbidity was considered as the end-point according to
standard clinical criteria including oversized tumor (>1 cm), weight loss (>20%), rough hair coat, and
general illness. At least 5 mice were used for each treatment group. Data are expressed as the mean
value of tumor weight with 95% confidence intervals. Mice were monitored for the duration of the in
vivo experiments for body weight, hair ruffling, and the presence of diarrhea. Animal care was
conformed to European Council Directive 86/609/EEC and the study was approved by institutional
review board.

Differences between groups were analyzed by Mann-Whitney test, student T test or by ANOVA as
appropriate. Data are expressed as mean 6 SD and p values reported are 2-sided.

Clinical study
The study was a multi-centre historically controlled trial, evaluating the activity of a pre-treatment
administration of PPIs as chemosensitizers in a neoadjuvant chemotherapy based on methotrexate
(MTX), cisplatin (CDP), and adriamycin (ADM).

Patients aged ≤40 years with resectable nonmetastatic osteosarcoma of the extremities, with normal
bone marrow, hepatic, cardiac and renal function, without contraindications to the use of MTX, CDP,
ADM, were included into the study and after given written informed consent. Patients were instructed
about the use of the study drug. In a separate sheet of the discharging letter they had to fill a table
reporting the exact time of administration of the study drug and signs or symptoms related to the study
drug.

Patients received, according to the standard strategy of treatment for osteosarcoma [3] neoadjuvant
chemotherapy that in this study consisted of two blocks of MTX (12 g/m2), CDP (120 mg/m2) and ADM
(75 mg/m2) (Figure 1). MTX (12 mg/m2, top dose 24 g) was given i.v. over 4 hour (T0-T4) infusion. The
minimum hydration required was 2.5 L/m2/day. From T24, folinc acid rescue was started every 6 hours
for 11 administration (up to T84). CDP was given i.v. 120 mg/m2 over 48 hours. The minimum hydration
required was 2 L/m2/24 hours After CDP infusion patients received ADM 75 mg/m2 i.v. over 24 hours.

In vitro studies on PPI effect on osteosarcoma cells. (a) Comparison of CDP activity with or without
ESOM pre-treatment (60 μM) in MG-63 cells. The activity of CDP is significantly higher in pretreated cells
(significant level * for p < 0.05) ...

All patients eligible for the study received the study drug ESOM. This was orally administered in the two
days prior to each cytotoxic agent. Patients had to receive ESOM the two days before chemotherapy
because preclinical studies have clearly shown that this is the only effective approach is the
pretreatment [3]. The daily dose of ESOM was 60 mg/day (before MTX), that is the minimal effective
dose calculated on the basis of the preclinical studies [3,4]. In patients weighting under 30 Kg the daily
dose was reduced at 40 mg/day. Before CDP/ADM cycles, the daily dose of ESOM was 120 mg/day. The
study drug was supplied by the local pharmacy.

Baseline studies included: complete blood count, serum electrolytes, and glomerular filtration rate
estimation; serum alkaline phosphatase and lactate dehydrogenase levels; bilirubin and
aminotransferase levels and echocardiography.

After neoadjuvant chemotherapy, the patients underwent surgical removal of the tumor.

Histological analysis of the tumor map was performed in accordance with a method reported previously
[8]. When the percentage of tumor necrosis was equal or higher than 90%, patients were classified as
good responders (GR); with a lower percentage they were defined as poor responders (PR).

Diagnosis, histological subtype, and histological response were reviewed by an expert panel of
pathologists.
The rarity of osteosarcoma, with an incidence of 0.2/100,000 new cases/year [5], made it difficult to run
a randomized phase 2 study, and for this reason it was decided to perform a prospective phase 2 study
with a comparison with historical control.

The comparator of the present study was the previous Italian Sarcoma Group ISG/OS-1 study, whose
results have already been published [14-17]. In particular, the group of patients treated according to
arm A, that had received the same drugs and schedule adopted in the present study, was used as
comparator.

In ISG/OS-1 the percentage of patients with a good pathologic response was 50%.

A sample size of 85 patients was calculated to detect a difference of 15% or higher, between the study
group and the historical control (study power of 80%, type I error of 5%).

Results

Preclinical studies

In vitro

This set of experiments was aimed at defining the ability of the PPI ESOM to increase the sensitivity of
human osteosarcoma cells to cytotoxic drugs. We first investigated the changes in the
microenvironmental pH induced by the osteosarcoma cells growth. Osteosarcoma cells were able to
acidify the medium by decreasing extracellular pH. In fact, after 24 h of cell culture in unbuffered
medium, the extracellular pH (pHe) of both osteosarcoma cell lines (MG-63 and Saos-2) was respectively
6.67 ± 0.01 and 6.89 ± 0.05. Thus, there was the suitable condition for the best protonation of ESOM in
order to allow the transformation of the pro-drug to the active molecule (tetracyclic sulfenamide).

Then we performed experiments aimed at evaluating the PPI-induced sensitization of human

osteosarcoma cells to CDP. Results showed that 24 h ESOM pre-treatment significantly increased the
activity of CDP in both osteosarcoma cell lines (Figure 1a and b). Particularly, after ESOM pre-treatment,
the activity of CDP, at the lower dosage tested, was double in MG-63 and three times in Saos-2 model (p 
= 0.0277) compared to control. These results strongly supported the use of ESOM in the therapy of
osteosarcoma patients.

In vivo

By analogy with the in vitro studies, we used the Saos-2 cellular model, ESOM as a PPI and CDP as one of
the drugs currently used in treatment of osteosarcoma patients. The in vivo model was set up following
the same procedure of the studies performed for other tumor types [3]. Briefly, CB.17 SCID/SCID mice
were injected s.c. with 3 x106 human osteosarcoma cells (Saos-2) and after 10 days 25 mg/kg ESOM was
administered i.p., 24 hours before the i.p. injection of 5 mg/kg CDP. This treatment was repeated weekly
for up to 4 consecutive weeks. Tumor size was measured three times/week until the end of the
experiments. Three groups of at least 5 mice were treated with CDP alone, ESOM + CDP combination, or
left untreated.

Results showed that, as expected, CDP alone induced a significant inhibition of tumor growth, but also
that ESOM pre-treatment induced a complete inhibition of tumor growth, with no evidence of the
tumor at the end of the experiments (Figure 2). Supporting previous results obtained with human cell
lines of different histotypes [3], ESOM alone, at this dosage and with this schedule of treatment did not
induce significant inhibition of tumor growth (not shown). This set of experiments further supported the
potential use of PPI in improving the effectiveness of chemotherapy in osteosarcoma patients.

In vivo effects of ESOM on tumor growth in CB.17 SCID/SCID mice. Mice were engrafted with Saos-2 cells
via s.c. injection in the right flank. At the time of tumor appearance (approximately 7–10 days after
injection), mice were left untreated ...

Clinical study

From January 2006 to December 2009, 99 patients [median age 15 years (6–40)] were registered into
the study. One of them was excluded because a central revision did not confirm the diagnosis of
osteoblastic osteosarcoma. All the 98 eligible patients were evaluable for histological response. No
protocol deviation was registered

A resection was performed in 93 patients, whereas 5 patients underwent amputation. The surgical
margins evaluation was available in 93 patients. In 9 patients the surgical margins were classified as
marginal (8 patients) or wide but contaminated (3 patients), the remaining patients had wide or radical
surgical margins. On histology, a GR was detected in 56 (57%) patients (Figure 3). According to the
histological histotype the highest rate of GR was seen in patients with the chondroblastic histotype
followed by those with telangiectatic, fibroblastic, and osteoblastic variants (Figure 3). In the comparator
group (ISG/OS-1), the incidence of GR was 47%, close to that observed in the investigational group
(Figure 3). According to the histotype, a similar percentage of good response was observed in
osteoblastic and in the pool of fibroblastic and telangiectatic osteosarcoma whereas in patients with
chondroblastic osteosarcoma the incidence of GR was only 25%, a remarkably lower rate as compared
to the 61% recorded in the group of patients pre-treated with ESOM.

Incidence of GR in eligible patients (PPI-Osteosarcoma) versus the comparator group (ISG/OS-1).

Comparison among the groups by means of Fisher’s exact test: Osteoblastic p = 0.32, Fibroblastic + 
Telangectatic 0.70, ...

Overall, the incidence of leukopenia G4 was 28%, thrombocytopenia G4 30%. The incidence of febrile
neutropenia was 27%. Red blood cells and platelets transfusion were required in 18% and 16% of the
chemotherapy cycles respectively. The incidence of delayed MTX excretion was 3% with transient G1
renal toxicity in 3 patients. Forty-two patients (mainly children and adolescents) experienced G4
transaminitis that in less than 30% of cases required medical treatment with steroids and fluid i.v.
administration. In no cases the liver toxicity was cause of drug discontinuation, but it more frequently
caused delay in the subsequent course of chemotherapy. No toxic deaths were reported. No serious
adverse events were reported related to the pretreatment with ESOM.

Discussion

Microenvironmental acidity is a factor influencing the tumor homeostasis. Preclinical data suggest that a
low pH is involved in resistance to cytotoxic drugs [3] and that tumor alkalinization through proton
pump inhibitors (PPI), renders tumor cells more sensitive to chemotherapy [1,3].

Previous studies have shown as PPI have an anti-tumor effect against different tumor types [4,18],
supporting the use of this family of compounds in the treatment of the vast majority of human tumors.
Moreover, since 2010 the International Society for Proton Dynamics in Cancer (ISPDC), has gathered a
consistent part of scientists that worldwide are studying tumor acidity and the proton dynamics
underlying the low pH of tumors, with the aim to explore new modalities of treatment for cancer [19].

The preclinical experiments have shown that PPI pre-treatment induce a clear increase in the sensitivity
of human osteosarcoma cells to chemotherapeutics. The in vitro experiments showed a significantly
increase in the CDP-mediated cytotoxicity following ESOM pre-treatment against osteosarcoma cells.
This set of experiments also provided the proof of concept that osteosarcoma cells cultured in
unbuffered condition were fully able to acidify the medium, just providing the suitable condition for
transformation of the ESOM pro-drug to the active molecule. In fact, in this condition while CDP alone
was effective against MG-63 and Saos-2 osteosarcoma cells, esomeprazole significantly increased the
CDP-mediated cytotoxicity at the lower dose of CDP. Notably, in MG-63 cells ESOM pre-treatment
lowered ten times the dose of CDP able to induce a 50% inhibition of cell growth (0.5 μΜ vs 5 μΜ). The
in vitro results were strongly supported by the in vivo results obtained in the human osteosarcoma-SCID
mice xenografts. Consistently with the results obtained with the human osteosarcoma cell cultures, pre-
treatment with ESOM significantly increased the effectiveness of CDP alone on the growth of human
osteosarcoma in SCID mice, getting to a total inhibition of the tumor growth at the end of the
experiments.

Thus, we tested the effectiveness of PPI pre-treatment in a clinical trial. The possibility to evaluate the
pathological response to neoadjuvant chemotherapy by mapping the resected specimen and the clear
relation between chemo-induced tumor necrosis and survival makes osteosarcoma an interesting model
to assess a possible role of PPIs as an innovative therapeutic solution. The rarity of osteosarcoma made
difficult to run a randomized phase 2 study, and for this reason we decided to perform a prospective
phase 2 study with a comparison with historical control. On the other hand the expected pathological
response to neoadjuvant chemotherapy in osteosarcoma is well defined, based on several previous
experiences [16,20].

Overall, the addition of PPIs, to MTX, CDP, and ADM allowed a higher rate of GR compared to the
control group, although this did not reach the statistical significance. It is important to notice that by
comparing the histological response according to the different histotypes arises a difference for the
chondroblastic variant, with a 61% of good responses in PPI pretreated patients (25% was the good
response rate in the control group). It is well known that the expected rate of GR is low in the
chondroblastic variant of osteosarcoma [20,21] and the results achieved in the present study suggest a
relation between chondroblastic component, tumor acidity and the use of PPIs.

On the other hand it is important to underline that the results reported were achieved with a multidrug
regimen, and that the drugs used have different pharmacokinetics and pharmacodynamics
characteristic. For this reason the study does not allow any conclusions on the possible interaction
between PPIs and each single drug used in the protocol.

The chemotherapy protocols currently used are characterized by a remarkable toxicity. Previous studies
performed in the clinical centers involved in this project have shown that IV grade neutropenia is
detectable in > 40% of patients despite the prophylactic treatment with G-CSF. The toxicity profile
registered in patients who received pretreatment with ESOM was the same observed in patients having
the same clinical characteristics and who received the same chemotherapy protocol. Preclinical data
showing that PPI pre-treatment increases the sensitivity of human osteosarcoma cells to antineoplastic
agents (particularly to CDP) is an interesting basis for further studies combining PPIs and antineoplastic
agents at a reduced dose in an attempt to ameliorate the chemotherapy-related toxicity.

It is note worth, that a recent report highlighted the interaction between PPIs and MTX toxicity [22]
leading to an increased MTX due to a concomitant use of the two drugs. In our report, however, the
toxicity profile of MTX was not modified by pre-treatment with ESOM.

Conclusions

This is the first translational study reporting preclinical and clinical investigations on the use of PPI in
human osteosarcoma. Pre-clinical data clearly show that PPIs may directly affect the sensitivity of
human osteosarcoma cells to chemotherapeutic drugs, such as CDP, and suggest that CDP could be
delivered at reduced dose in an attempt to achieve a better toxicity profile.

Clinical data, while obtained with a surrogate endpoint, such as the chemotherapy-induced tumor
necrosis, represent the first proof of concept that PPI may be included in the future strategies against
cancer, and in particular in the treatment of osteosarcoma patients. The high pathological response rate
in patients with the chondroblastic variant of osteosarcoma recommends further studies to assess the
relation between PPIs and sarcoma tumors with chondroblastic component.

STUDIJA 9

Oxford JournalsMedicine & Health Rheumatology Volume 45, Issue 3Pp. 362-363.

Disease-modifying anti-rheumatic drugs are only one of a number of potential causes of

myelosuppression: a careful drug history is necessary to elucidate the cause of an adverse event

R. W. Marshall, V. J. Marshall 1 and R. Hull

Rheumatology, Queen Alexandra Hospital, Portsmouth and 1Drug Safety Research Unit, Southampton,
UK
Correspondence to: R. W. Marshall. E-mail: R.W.Marshall@doctors.org.uk

SIR, We read with interest the article by Lim et al. [1] concerning methotrexate (MTX)-induced
pancytopenia, and present here three cases of drug-induced myelosuppression that illustrate that
disease-modifying antirheumatic drugs (DMARDs) are not always to blame.

Case 1 was a 76-yr-old lady who was treated with MTX for her rheumatoid arthritis (RA). A routine blood
test on 1 July 2004 showed that her haemoglobin (Hb) had fallen over 1 month from 11.1 g/dl to 8.4 
g/dl, the white blood count (WBC) from 5.9 × 109/l (neutrophils 3.1 × 109/l) to 2.1 × 109/l (neutrophils
0.5 × 109/l) and platelets from 581 × 109/l to 310 × 109/l. The MTX was thought to be responsible. She
was admitted to hospital, her medications were stopped and she was given folinic acid rescue. Her blood
parameters improved.

At clinic review a drug history revealed that she had been taking MTX, folic acid and azapropazone
without problem for over 10 yr, and ferrous sulphate for the past 6 months. Two months previously she
had received intravenous flucloxacillin and benzylpenicillin, and had started lansoprazole.
Haematological abnormalities are well recognized in these three drugs [2], and on balance it was felt
that the lansoprazole was responsible. With much reservation the patient agreed to restart MTX, and
regular haematological monitoring has been unremarkable since.

The second case was a 65-yr-old lady with longstanding RA, who had previously reacted to MTX (nausea
and headaches), sulphasalazine (diarrhoea and vomiting), gold injections (rash) and penicillamine (flu-
like illness). In March 2003 she commenced etanercept with an excellent response. Later that year she
was prescribed meloxicam.

On 19 April 2004 a routine blood test showed Hb 12.9 g/dl, WBC 3.0 × 109/l (neutrophils 0.9 × 109/l) and
platelets 218 × 109/l, having previously been normal. Etanercept and meloxicam were stopped, but
repeat blood counts over the next 10 days showed no improvement. At clinic review on 4 May, a careful

history revealed that she had commenced aspirin 75 mg daily on 19 February following a transient
ischaemic attack.

Aspirin was stopped and she was monitored closely. She developed oesophageal candidiasis, which was
successfully treated with fluconazole. A bone marrow and trephine biopsy showed maturation arrest of
neutrophil production, with no evidence of dysplasia. By August 2004 her blood count had returned to
normal (WBC 4.4 × 109/l, neutrophils 2.4 × 109/l). Her RA flared so she was commenced on
prednisolone 10 mg daily, prior to the commencement of adalimumab.

Haematological abnormalities are well recognized with aspirin [2], but are rare with etanercept:
neutropenia is listed as occurring in between 1/1000 and 1/10 000 patients. To July 2004, the
Committee on Safety of Medicines (CSM) had received 16 case reports of neutropenia with etanercept.
Aspirin was the likely cause in this patient as the neutropenia occurred within a few weeks of
commencing therapy.
Case 3 was a 65-yr-old lady with longstanding RA who had been taking azathioprine 150 mg daily for 3 
yr, along with coproxamol. A routine blood test in May 1997 showed Hb 11.1 g/dl, WBC 1.6 × 109/l
(neutrophils 1.2 × 109/l) and platelets 30 × 109/l. She had been feeling unwell with anorexia, dry mouth,
sore throat, weight loss and dysphagia, and was pyrexial. She was admitted to hospital, where repeat
blood tests showed Hb 9.0 g/dl, WBC 0.4 × 109/l (neutrophils 0.2 × 109/l), platelets 9 × 109/l,
international normalized ratio 1.8, Activated Partial Thromboplastin ratio (APTR) 2.7, bilirubin 56 µmol/l
(normal range 3–20), albumin 23 g/l (37–50), aspartate transaminase 117 IU/l (12–40) and alkaline
phosphatase 195 IU/l (30–95).

Examination showed established rheumatoid arthritis, but no synovitis; temperature 38°C and a
purpuric rash on the legs but no other localizing signs. Azathioprine was stopped and she was given
supportive treatment and repeated transfusions. Blood cultures were consistently sterile, although she
developed oropharyngeal candidiasis and a coliform urinary infection. Ultrasound of the abdomen
confirmed hepatosplenomegaly.

She improved over 1 month and was discharged. A drug history revealed that prior to admission she had
received a week's triple therapy with clarithromycin, metronidazole and lansoprazole for eradication of
Helicobacter pylori. The lansoprazole was thought to be the cause of her pancytopenia. Four months
later her RA started to flare, and at the patient's insistence azathioprine was restarted. Eight years later
she is still taking the drug without problem. Myelosuppression is a recognized complication of DMARD
therapy, and patients should be counselled and monitored according to BSR guidelines [3]. When an
adverse drug reaction (ADR) occurs it is critical to elucidate which drug is responsible, and this should be
reported to the CSM. It is easy to ascribe causality to the more ‘toxic’ drug, but adverse events occur
more commonly with drugs that have been recently introduced [4, 5]. The diagnostic processes
underlying causality assessments have been well described, and include consideration of the time to
onset of ADR, differential causes, a process of elimination, and the recurrence of the event on
rechallenge [6]. In clinical practice, rechallenge after an ADR is not always possible.