THROMBOSIS RESEARCH 64; 131-141 1991

0049-3848/91 $3.00 + .OO Printed in the USA.

Copyright (c) 1991 Pergamon Press plc. All rights reserved.

DEVELOPMENT AND VALIDATION OF A SIZE EXCLUSION CHROMATOGRAPHY

METHOD FOR DETERMINATION OF MOLECULAR MASSES AND MOLECULAR MASS
DISTRIBUTION IN LOW MOLECULAR WEIGHT HEPARIN

H.I. Kristensenl*, E.M. Tromborg2. J.R. Nielsen3, J.I. Nielsenl,

K.B. Johansen3*, P.B. 0stergaardl.
Heparin Research Laboratoryl, Novo Nordisk, DK-2820 Gentofte,
Denmark, Chemical Control Laboratory2 and Heparin Research Labo-
ratory3, Leo Pharmaceutical Products, DK-2750 Ballerup, Denmark.

(Received 29.4.1991; accepted in revised form 26.7.1991 by Editor I. Danishefsky)

ABSTRACT
A precise and rugged high performance size exclusion
chromatography (HPSEC) method for determination of
peak maximum molecular mass (q ), weight average
molecular mass (Y,), number average molecular mass
(ry,), and molecular mass distribution (MMD) in Low
Molecular Weight heparin (LMW heparin) has been deve-
loped and validated on two UltrahydrogelR 250 (ID 7,8
mm, length 30 cm) columns in series at three labora-
tories using different equipment. The calibration is
based on four local laboratory heparin standards with
relatively narrow molecular mass distribution and Mp
covering the range from 3,000 to 17,000 Da. The cali-
bration curve describing the logarithm of the mole-
cular mass versus retention time is linear in the
range 3,000 to 17,000 Da. The precision (relative
standard deviation) within-run and between-run is
better than 2%. The between-laboratory variation is
below 5%, 3% and 8% for , & and M,,, respectively.
This method is reproducib Y e, rugged, and fast and is
suited for the determination of average molecular
masses and molecular mass distribution of LMW heparins
on a routine basis.

Key words: LMW heparin, Heparin, molecular mass distribution,

size exclusion chromatography

* Authors to whom correspondence should be addressed.

131
132 MW CALIBRATION FOR LMW HEPARIN Vol. 64, No. 2

INTRODUCTION

Heparin, a widely used anticoagulant drug, is a polydispersed

highly sulphated glucosaminoglycan with a molecular mass
distribution ranging from 2,000 to 40,000 Da (1). During the
past decade new low molecular weight (LMW) heparins with
improved pharmacokinetic properties have been developed
(2,3,4). These newly introduced products, which are derived
from conventional heparin either by depolymerization or
fractionation, differ from one another with respect to mean
molecular mass, molecular mass distribution, and biological
activities in vitro and in viva (5).
Methods for the determination of molecular mass have been
studied in international collaborative studies. These have been
organized by the Heparin Subcommittee of the International
Committees on Thrombosis and Haemostasis (ICTH). The first
study showed that high performance size exclusion chromato-
graphy (HPSEC) is a feasible method with acceptable inter-
laboratory variation when a set of well-defined molecular mass
standards are applied by all laboratories (6). The second study
investigated whether a relatively broad molecular mass LMW
heparin produced by enzymatic depolymerization could be used as
a single calibration standard (7). Several papers on molecular
mass determination on heparin have confirmed that HPSEC is a
suitable and reproducible method (8-12) as long as proper
calibration standards are available. By combining HPSEC with
low angle laser light scattering (LALLS) detection, the need
for reference standards can be overcome (13, 14). However,
LALLS detectors are not wide-spread in analytical laboratories
and not very suitable for LMW heparins as the LALLS signals are
very small for molecular masses lower than about 6,000 Da (14).
The aim of the present study was to develop and validate a
generally applicable, reproducible, rugged, and fast method for
routine analysis of mean molecular masses and molecular mass
distribution of LMW heparins, a method enabling quality
assurance of each analysis by inclusion of a marker. In the
collaborative study of ICTH (6) the participating laboratories
used different column configurations and mobile phases but the
same calibration standards. We have investigated the inter-
laboratory variation of M,,, rY, and Q, when the same method was
used in 3 laboratories.

MATERIALS AND METHOD

Materials: Heparin Sodium was produced at Leo Pharmaceutical

Products. Heparinase NOVO was manufactured at Novo Nordisk. Low
Molecular Weight Heparin Sodium was produced by enzymatic
depolymerisation by Heparinase at both companies. Sodium
acetate trihydrate, sucrose, and sodium azide were of analy-
tical grade from Merck (Darmstadt, F.R. Germany).
Vol. 64, No. 2 MW CALIBRATION FOR LMW HEPARIN 133

Molecular Mass Standards: Four Local Laboratory Heparin Stan-

dards (LSl, LS2, LS3, and LS4) were produced. Heparin standard
LSl was produced by precipitation of Heparin Sodium with an
adjusted amount of ethanol (s - 16,600 Da). Standard LS2 was
produced by depolymerization of Heparin Sodium with HNO,. The
end groups were reduced with NaBH,. The molecular mass fraction
was isolated by fractionated ethanol precipitation (q - 9,600
Da). Standards LS3 and LS4 were produced as described for LS2
followed by gelfiltration ($ - 4,500 Da and 3,000 Da, respec-
tively). The Mp of LSl, LS2, and LS3 were determined by LALLS
(15), whereas the Mp of LS4 was determined by identification of
the oligosaccharide at the peak maximum. This oligosaccharide
was shown to be a decasaccharide, which was assigned a mole-
cular mass of 3,000 Da anticipating an average molecular mass
per disaccharide unit of 600 Da. The heparin standards
described above form Standard Set 1.

A second standard set consisting of four Working Local Labo-

ratory Heparin Standards (WSl, WS2, WS3, and WS4) were also
made. Standards WSl, WS2, and WS3 were produced in the same way
as LSl, LS2, and LS3, whereas WS4 was manufactured by enzymatic
depolymerization of Heparin Sodium with Heparinase. This frac-
tion was purified by ethanol precipitation and gel filtration.
Mp of these four WS (Standard Set 2) was determined using
Standard Set 1. Their Mp values were 16,700, 10,600, 6,700,
and 3,100 Da, respectively. Small amounts of these standards
are available from the authors on request.

Four oligosaccharides were produced in the same way as WS4 and

of these oligosaccharides was assigned a molecular mass of
6 0 Da per disaccharide (2,400, 1,800, 1,200, and 600 Da,
"B
respectively).

Chromatographic conditions: The eluent is 0.5 M sodium acetate

including 0.04% sodium azide as preservative. The eluent is
filtered-and degassed through a membrane filter (e.g. 0.45 urn).
Molecular Mass Standards and samples (10 mg/ml) are dissolved
in eluent containing sucrose (1.5 mg/ml) as a marker.
The columns are two UltrahydrogelR 250 (ID 7.8 mm, length 30
cm) in series (Millipore/Waters part no. 11515). The equipment
and the chromatographic conditions used in the 3 participating
laboratories are shown in Table 1. Overlayed chromatograms of
Standard Set 2 are depicted in Fig. la and a chromatogram of a
LMW Heparin is shown in Fig. lb.

Calculation: For the molecular mass standards the retention

time of % is determined. A calibration curve is established by
plotting the logarithm of $ versus the respective retention
time for the 4 standards. The correlation coefficient should be
better than -0.998. In the samples s is determined either
directly from the calibration curve or it is calculated
together with molecular mass distribution (MMD), K, and K by
the integrator software.
134 MW CALIBRATION FOR LMW HEPARIN Vol. 64, No. 2

TABLE 1

toaraohic Conditipas for Each Eouicment of the .Laboratories_

-----___-------------------------------------------------------------
Equipment Laboratory
1 2 3
---------------------------------------------------------------------
Pump WATERS MERCK-Hitachi MERCK-Hitachi
model 510 L-6200 655A-12

Auto WATERS Wisp MERCK-Hitachi L

sampler model 712 655 A-40

Integrator WATERS Maxima 820 MERCK-Hitachi Perkin-Elmer

GPC-software D-2520 GPC Sigma 15
version 3.02 Data Station

Refractive WATERS Shodex RI Knauer

index detector model 410 SE-61

Column WATERS WATERS TCM i

heater (43'C) (35°C)

Flow rate 0.5 ml/min 0.7 ml/min 0.7 ml/min

Injection vol. 50 p1 50 p1 100 p1

15 20 25 15 20
;- L
25
I 8.

Retention
time(min) Retention time (min)

Fig. la Fig. lb
Overlayed chromatograms. Chromatogram of LMW heparin
Standard Set 2. 20.14 min - q
25.90 min - Sucrose Retention Time
26.73 min - Salt Peak Retention Time
Vol. 64, No. 2 MW CALIBRATION FOR LMW HEPARIN 135

RESULTS

Recovery: Recovery of LMW heparin has been investigated by

examinations of Standard Set 2 demonstrating linearity between
the amount of heparin injected (mg/injection) and the heparin
peak area (volt set). The recovery of WS3 is depicted in Fig.
??

The correlation coefficients of Standard Set 2 are r,,, =

$999, rws2 = 1.000, r,,, = 1.000, and r,,, = 0.999, respecti-
vely. Furthermore, the intercepts with the y-axis were in all
cases less than 2% of the response obtained by injection of a
test substance in the normal working concentration. In conc-
lusion, the recovery is complete at least when the molecular
mass distribution is in the range of the Molecular Mass
Standards and when the amount injected is between 0.125 and
1.250 mg.
Arer(voltr*rsc)
361
,’

30r

26- /
20- /'
/*

lo-

16- /
6 _ /'

/"
OL* / 1 / I / 1
0 0.2 1.2 1.4
'.'AmounoP~mg/i~j:ction)'

Fig. 2. Recovery of heparin (WS3) on UltrahydrogelR 250

columns (2 x 30 cm). Peak area as a function of different
amount of heparin.

Column configuration and calibration curves: Two Ultrahydrogel

250 columns (2 x 30 cm) in series have been shown to have a
suitable hydrodynamic range for analysis of LMW heparins but
too poor a resolution at molecular masses above 20,000 Da to
be ideal for conventional heparin (15).
Standard Set 2 gives a linear calibration curve in the range
between 3,100 Da and 16,700 Da with a correlation coefficient
of -1.000 when the logarithm of q is plotted against retention
time (Fig. 3). When Standard Set 2 is supplemented with the 4
oligosaccharides a cubic curve (r=-1.000) coinciding with the
linear curve in the range 3,100 to 16,700 Da is formed (Fig.
3).
For routine analysis a linear calibration curve based on
Standard Set 2 is found adequate.
136 MW CALIBRATION FOR LMW HEPARIN Vol. 64. No. 2

log. Mp
-/
5.00 -

4.00 -

3.00 -

2,oo J , , , , , , , , ,

20 30 40

Retention Time (min.)

Fig. 3. Graphs of the linear calibration curve obtained from

Standard Set 2 (r=-1.000) and of the cubic curve obtained when
the standards are supplemented with 4 oligosaccharides
(r = -1.000).

Stability of the HPSEC-system:

1) Influence of the pump flow: Table 2 shows the effect of a
change of the pump flow rate on s, q, and M,, for WS3. WS3 was
run at flow rates of 0.700 ml/min, (0.700 f 1%) ml/min, and
(0.700 + 2%) ml/min. At each run the results were calculated
using a calibration curve obtained from Standard Set 2 at a
flow rate of 0.700 ml/min.

TABLE 2
The influence of the pump flow on%, bh, and % of WS3.
a) The numerical deviation in per cent from 0.7 ml/min.
b) The numerical deviation in per cent from the value
determined at 0.7 ml/min.

Mp 8 b’ M, % b’ % % b’
Da Da Da

0.686 -2 5,496 19.8 5,035 17.1 6,595 17.4

0.693 -1 6,112 10.8 6,051 0.4 7,963 0.3

0.700 0 6,852 0 6,076 0 7,985 0

0.707 +l 7,641 11.5 6,903 13.6 9,126 14.3

0.714 +2 8,496 24.0 8,469 39.4 11,145 39.6

Vol. 64, No. 2 MW CALIBRATION FOR LMW HEPARIN 137

2) The use of sucrose as a marker: The molecular mass standards

and the samples are dissolved in mobile phase with sucrose
added as a marker. Table 3 shows the stability of the HPSEC-
system expressed as the retention time of sucrose (RTsuc).
RTsuc =’
independent of the concentration between 0.125 and
1.250 mg heparin per injection and molecular mass distribution
of the heparin within the range of the standards.

BY using a marker the same calibration curve may be used as

long as the retention time of the marker stays within the
limits defined from the results obtained when the calibration
curve was established. We have been able to use the same
calibration curve for months and have experienced that any
technical problem with t,he. analytical system, e.g. leaking
seals in the pump or injection system, air bubbles, column
deterioration and so on will be detected immediately by a
change in the retention time of the marker.

TABLE 3

Influence of the molecular mass and

the amount of heparin injected on RT,"..

10 different amounts in the range 0.125-1.25 mg/injection was

tested and the mean and RSD calculated.

RT.SUC(min)
Standards Mean % RSD

WSl a) 16,700 Da 25.80 0.04%

WS2 a) 10,600 Da 25.80 0.06%
WS3 b) 6,700 Da 25.66 0.05%
WS4 b) 3,100 Da 25.62 0.07%

a. Calibration curve: In M, = -0.5714x + 20.0596

b. Calibration curve: In M, = -0.5890x + 20.3378

3) Concentration dependence of Rr: In Fig. 4 the concentration

dependence of R, is depicted. The figure shows that R, is
influenced by the amount of heparin injected especially in the
high molecular mass end (>lO,OOO Da). Consequently, it is
important that the amount (mg/injection) of standards are com-
parable to the amount of the corresponding molecular mass frac-
tion in the sample.
138 MW CALIBRATION FOR LMW HEPARIN Vol. 64, No. 2

0 0.2 0.4 0.6hg~~&o~~' 1.2 1.4

hnount

---a4 -+- wsa 'Icws2 -s-W81

Fig. 4: Graph of retention time in minutes against the amount

of heparin standard injected.

Reproducibility within-run and between-run: Table 4 shows the

variation of Mp, 4, and 4, when a given LMW heparin sodium
(LMWH) is analysed six times within one run. Each LMWH is
calculated both against a cubic (C) and a straight line (R)
calibration curve (Fig. 3). There is no significant difference
between the two sets of results. In addition the molecular mass
distribution of LMW heparin is stated as per cent molecules
above 10,000 Da, between 10,000 and 1,500 Da and below 1,500
Da. There is an acceptable agreement between the results from
the cubic and the straight line calibration curve. The same two
LMWH are examined in six different runs and these data are
shown in Table 5. The precision of q, M,,, and K within-run
and between-run are the same (< 1.3%).

TABLE 4
Within-run reproducibility. Each sample is repeated six times
and calculated against a cubic (C) and a linear (R) calibration
curve (Fig. 3). RSD is stated in the brackets.

3atch Curve %MMD

;MWH >10,000 Cl,500 1,500-10,000

A C 5,400 4,000 6,800 20.0 75.0

(0.9%) (0.3%) (1.1%) (1.0%) ;1098%) (0.2%)

A R 5,400 4,200 6,800 20.2 77.5

(1.0%) (0.2%) (0.9%) (0.9%) ?i45%) (0.3%)

B C 4,800 3,800 6,500 17.8 76.8

(1.0%) (0.7%) (1.3%) (1.5%) Yl%) (0.5%)

B R 4,800 4,100 6,500 17.9 79.5

(1.0%) (0.6%) (1.0%) (1.5%) K%) '(0.4%)
Vol. 64, No. 2 MW CALIBRATION FOR LMVV HEPARIN 139

TABLE 5
Between-run reproducibility. Each sample is examined once in
six independent days and calculated against a cubic (C) and a
linear (R) calibration curve (Fig. 3). The relative standard
deviation is stated in the brackets.

3atch Curve %MMD

MWH ~10,000 Cl,500 1,500-10,000

A C 5,500 4,000 6,900 20.5 74.9

(1.3%) (1.0%) (1.3%) (1.7%) :i68%) (0.5%)

R 5,400 4,300 6,800 77.1

(1.3%) (0.8%) (1.1%) (0.5%)

C 4,800 3,900 6,500 18.0 76.9

(1.3%) (0.8%) (1.3%) (1.3%) :410%, (0.4%)

R 4,800 4,100 6,500 79.5

(1.3%) (0.6%) (1.0%) (0.4%)

Reproducibility between laboratories: Standard Set 2 was cali-

brated against Standard Set 1 at the three independent labora-
tories. The mean of M,,, M,,, K including the relative standard
deviation and the polydispersity Q/M,, is shown in Table 6
(n=18). To show conformity between the two Set of Standards one
batch of LMW heparin has been run ten times against both Stan-
dard Set 1 and Standard Set 2. No statistical difference was
found (student's t-test).

TABLE 6
Reproducibility between laboratories. Each standard is deter-
mined six times at each laboratory. The results are mean of
eighteen determinations. The relative standard deviation is
stated in the brackets.

Mp Y-4 Y Polydispersity
Da (Mw/Mn)

16,700 13,400 17,200 1.3

(1.9%) (4.0%) (1.8%)

10,600 7,800 11,100 1.4

(4.0%) (5.4%) (1.1%)

6,700 5,900 8,000 1.4

(1.8%) (5.6%) (1.4%)

3,100 2,800 3,500 1.3

(1.6%) (7.9%) (2.6%)
140 MW CALIBRATION FOR LMW HEPARIN Vol. 64, No. 2

DISCUSSION

In this study we have developed and validated a high perfor-

mance size exclusion chromatography method for the determi-
nation of average molecular masses and molecular mass distri-
bution of low molecular weight heparin at three independent
laboratories using different chromatographic systems. The
chromatographic conditions and the column configuration of the
method applied well for all three HPLC systems, and the use of
a marker in each chromatographic run made it possible to
evaluate the validity of each analysis.
The within-run, between-run, and between-laboratory variations
are very satisfactory showing that the method is precise and
rugged. The accuracy is impossible to assess as no "true
values" of molecular weight characteristics for any heparin
product exist. Our method gives reproducible result not only of
average molecular masses but also of molecular mass distri-
bution.

When this method is used for quality control analysis the use
of a marker is important since even small differences in pump
flow result in considerable variation in the results (Table 2).

The major draw-back of HPSEC is its dependence on calibration

standards which are not easily available. As mentioned in the
introduction HPSEC with low angle laser light scattering
(LALLS) detection is independent of calibration standards. Such
detectors are, however, not common in analytical laboratories
and not very suitable for low molecular mass polymers as LMW
heparin (14).

In our study we have used relatively narrow molecular mass

heparin calibration standards. Our method should, however, be
equally well suited in connection with a broad molecular mass
calibration standard as the one suggested for an international
standard (9). In this study we have shown that the retention
time especially at relatively high molecular masses does depend
on the amount of substance loaded to the columns indicating
that overloading is a potential source of error.

In summary, we have developed a fast and reproducible method

for the routine determination of average molecular masses and
molecular mass distribution of LMW heparin. The method is
dependent on the availability of calibration standards, but we
have shown that 4 standards covering the molecular mass range
of LMW heparins are sufficient and that a linear calibration
curve is adequate. Small amounts of our 4 working standards are
available from the authors on reguest.

ACKNOWLEDGEMENT

The authors wish to thank Mr. W. Batsberg, Rise National Labo-

ratory for the LALLS calibration of the heparin standards.
Helle Bak, Novo Nordisk, Mai-Britt Nielsen and Marianne Hansen,
LEO Pharmaceutical Products are thanked for technical assistan-
ce, and Anne Christensen, Novo Nordisk for typing this manu-
script.
Vol. 64, No. 2 MW CALIBRATION FOR LMW HEPARIN 141

REFERENCES

1. CHOAY, J. Structure and activity of heparin and its frag-

ments: An overview. Semin. Thromb. Hemostas. 15, 359-364,
1989.
2. MATZSCH, T., BERGQVIST, D., HEDNER, U. and 0STERGAARD, P.
Effects of an enzymatically depolymerized heparin as
compared with conventional heparin in healthy volunteers.
Thromb. Haemostas. 57, 97-101, 1987.
3. BARA, L., BILLAUD, E., GRAMOND, G., KHER, A. and SAMAMA, M.
Comparative pharmacokinetics of a low molecular weight
heparin (PK 10169) and unfractionated heparin after
intravenous and subcutaneous administration. Thromb. Res.
3, 631-636, 1985.
4. BRATT, G., TORNEBOHM, E., WIDLUND, L. and LOCKNER, D. Low
molecular weight heparin (Kabi 2165, Fragmin): Pharmaco-
kinetics after intravenous and subcutaneous administration
in human volunteers. Thromb. Res. 42, 613-620, 1986.
5. FAREED, J., WALENGA, J.M., HOPPENSTEADT, D., RACANELLI,
B.A. and COYNE, E. Chemical and biological heterogeneity in
low molecular weight heparins: Implication for clinical use
and standardization. Semin. Thromb. Hemostas. 15, 440-463,
1989.
6. VAN DEDEM, G. Collaborative assay of heparin molecular
weights by high performance size exclusion chromatography.
Report, Diosynth BV, Oss, The Netherlands, 1985.
7. VAN DEDEM, G. Oral presentation at the 36th Annual Meeting
of the Scientific and Standardization Committee for
Heparin. Barcelona, Spain, June 21, 1990.
8. RODRIGUEZ, H.J. Accurate and reproducible determination of
molecular weight distribution of sodium heparin U.S.P. by
HPLC. Anal. Letters. 9, 497-506, 1976.
9. SUGISAKA, N. and PETRASEK, F.J. Rapid molecular size chara-
cterization of heparins by high pressure liquid chromato-
graphy. Fed. Proc. 36, 89-92, 1977.
10. RODRIGUEZ, H.J. and VANDERWIELEN, A.J. Molecular weight
determination of commercial heparin sodium USP and its
sterile solutions. J. Pharm. Sci. 68, 588-591, 1979.
11. HARENBERG, J. and DE VRIES, J.X. Characterization of
heparins by high-performance size exclusion liquid
chromatography. J. Chromatoq. 261, 287-292, 1983.
12. MULLER, D., NDOUME-NZE, M. and JOZEFONVICZ, J. High-
pressure size-exclusion chromatography of anticoagulant
materials. J. Chromatog. 297, 351-358, 1984.
13. MIKLAUTZ, H., RIEMANN, J., and VIDIC, H.J. The molecular
weight distribution of heparin determined with a HPLC-LALLS
coupling technique. J. Liq. Chromatoq. 9, 2073-2093, 1986.
14. HENNINK, W.E., VAN DEN BERG, J.W.A. and FEIJEN, J.
Standardization of heparins by means of high performance
liquid chromatography equipped with a low angle laser light
scattering detector. Thromb. Res. 45, 463-475, 1987.
15. BATSBERG, W., HANDLOS, V.N., HVASS, AA., JOHANSEN, K.B.,
NIELSEN, J.I. and ~STERGAARD, P.B. Determination of the
molecular mass distribution of heparin. Eighth Interna-
tional Symposium on HPLC of Proteins, peptides and Poly-
nucleotides. Copenhagen Denmark. October 31-November 2,
1988.