Accepted Manuscript

Pharmacokinetics and Placental Transfer of Magnesium Sulfate in Pregnant Women

Kathleen F. Brookfield, MD, PhD, MPH, Felice Su, MD, Mohammed H. Elkomy, PhD,
David R. Drover, MD, Deirdre J. Lyell, MD, Brendan Carvalho, MBBCh, FRCA

PII: S0002-9378(15)02664-2
DOI: 10.1016/j.ajog.2015.12.060
Reference: YMOB 10868

To appear in: American Journal of Obstetrics and Gynecology

Received Date: 28 October 2015

Revised Date: 20 December 2015
Accepted Date: 29 December 2015

Please cite this article as: Brookfield KF, Su F, Elkomy MH, Drover DR, Lyell DJ, Carvalho B,
Pharmacokinetics and Placental Transfer of Magnesium Sulfate in Pregnant Women, American Journal
of Obstetrics and Gynecology (2016), doi: 10.1016/j.ajog.2015.12.060.

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to
our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and all
legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT
1

1 Pharmacokinetics and Placental Transfer of Magnesium Sulfate in Pregnant

2 Women

3 Kathleen F. BROOKFIELD1*, MD, PhD, MPH, Felice SU2, MD, Mohammed H.

4 ELKOMY3, PhD, David R. DROVER4, MD, , Deirdre J. LYELL5, MD, Brendan

PT
5 CARVALHO4, MBBCh, FRCA

RI
6

7 1*. Department of Obstetrics and Gynecology, Stanford University School of Medicine;

SC
8 currently at Department of Obstetrics and Gynecology, Oregon Health and Science

9 University

10
U
2. Department of Pediatric Critical Care, Stanford University School of Medicine
AN
11
12 3. Department of Pharmaceutics and Industrial Pharmacy, Beni Suef University, Egypt
13
14 4. Department of Anesthesia, Stanford University School of Medicine
M

15
16 5. Department of Obstetrics and Gynecology, Stanford University School of Medicine
17
D

18
19 Corresponding author:
TE

20 Kathleen Brookfield, MD, PhD, MPH

21 3181 SW Sam Jackson Park Road
22 Portland, OR 97239
EP

23 Phone: (503) 494-2101

24 Fax: (503) 494-5296
25 Email: brookfie@ohsu.edu
26
C

27 This work was supported by the Arline and Pete Harman Children’s Health Research
28 Institute fund at Lucile Packard Children’s Hospital at Stanford (Dr Carvalho as recipient
AC

29 and principal investigator).

30
31 The authors report no conflict of interest.
32
33 ClinicalTrials.gov Identifier: NCT01709630
34
35 * Figure 3 chosen for publication in print version of manuscript.

36 This study was presented, in part, at the Society for Maternal Fetal Medicine 35th Annual
 ACCEPTED MANUSCRIPT
2

37 Meeting. San Diego, CA. February 2015.

38 Word Count

39 Abstract: 251

40 Main Text: 2709

PT
41
42

RI
43
44
45

SC
46
47
48
49

U
50
51
AN
52
53
54
55
M

56
57
58
D

59
60
TE

61
62
63
EP

64
65
66
67
C

68
69
AC

70
71
72
73
74
75
76
77
78
 ACCEPTED MANUSCRIPT
3

79 Condensation
80
81 Pharmacokinetic modelling indicates that preeclamptic status and maternal weight

82 significantly impact serum magnesium in women administered magnesium sulfate.

83 Short version of title: Pharmacokinetics of magnesium sulfate in pregnancy

PT
84
85

RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT
4

86 Abstract
87
88 Background. Magnesium sulfate is one of the most commonly prescribed intravenous

89 medications in obstetrics. Despite widespread use, there are limited data about

90 magnesium pharmacokinetics, and magnesium is prescribed empirically without dose

PT
91 adjustment for different indications.

RI
92 Objective. The aim of this study was to characterize the pharmacokinetics and placental

93 transfer of magnesium sulfate in pregnant women, and to determine key covariates which

SC
94 impact the pharmacokinetics.

95 Study Design. This is a prospective pharmacokinetic cohort study of pregnant women

96
U
prescribed magnesium sulfate for preeclampsia, preterm labor or extreme prematurity.
AN
97 Women received 4g loading dose and 2g/hr maintenance dose as clinically indicated.

98 Maternal blood samples were obtained prior to and at multiple time points during and
M

99 after magnesium administration. Cord blood was also sampled at delivery. A population
D

100 pharmacokinetic approach using nonlinear mixed–effects modeling was used to

TE

101 characterize magnesium disposition.

102 Results. Pharmacokinetic profiles of 111 pregnant women were analyzed. Magnesium
EP

103 clearance was 3.98 L/h in preeclamptic women and 5.88 L/h non-preeclamptic women.

104 Steady state concentration of magnesium was 7.2 mg/dL in preeclamptic women
C

105 compared to 5.1 mg/dL in non-preeclamptic women. Maternal weight significantly

AC

106 impacted time to steady-state. The ratio of the mean umbilical vein magnesium level to

107 the mean maternal serum magnesium level at the time of delivery was 0.94 + 0.15.

108
 ACCEPTED MANUSCRIPT
5

109 Conclusions. The study accurately characterizes the pharmacokinetics of magnesium

110 administered to pregnant women. Preeclamptic status and maternal weight significantly

111 impact serum magnesium levels. This pharmacokinetic model could be applied to larger

112 cohorts to help tailor magnesium treatment and account for these covariates.

PT
113

RI
114
115 Keywords: magnesium sulfate, pregnancy, neuroprotection, pharmacokinetics,

SC
116 NONMEM

117

U
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT
6

118 Introduction

119 Magnesium sulfate is one of the most commonly prescribed intravenous medications in

120 obstetrics, and is used for seizure prophylaxis in preeclampsia, tocolysis in threatened

121 preterm labor, and neuroprotection of the preterm fetus with anticipated delivery prior to

PT
122 32 weeks gestation.1,2 It is estimated that more than 700,000 newborns in the U.S. are

RI
123 born each year exposed to magnesium sulfate following maternal administration.

124 Although many dosing protocols are designed to prevent an untoward outcome,

SC
125 undesirable effects for both mother and neonate have been observed with varying

126 magnesium sulfate dosing regimens during pregnancy.1,3-16

U
AN
127 Current knowledge of magnesium disposition in pregnant women is limited to

128 studies of women with preeclampsia using minimal blood sampling with limited
M

129 covariate data available.17-19 Standard magnesium sulfate treatment protocols (without

130 pharmacokinetic modeling consideration) were proposed by Prichard in 1955 to provide

D

131 estimated serum magnesium levels necessary to treat eclampsia.20 Recent evidence
TE

132 suggests that many magnesium sulfate treatment protocols do not achieve serum levels

133 considered therapeutic.16 These standardized protocols are administered to all patients,
EP

134 without adjustment for maternal or fetal factors that may affect serum magnesium levels

135 in the mother. Optimal serum magnesium levels for treatment of non-preeclamptic
C

indications are also unknown.21

AC

136

137 The aim of this study was to develop an accurate pharmacokinetic model,

138 describe the placental transfer of magnesium sulfate in pregnant women, and evaluate the

139 impact of potentially influential covariates (gestational age, presence of preeclampsia,

140 maternal weight, and maternal creatinine) on magnesium pharmacokinetics.

 ACCEPTED MANUSCRIPT
7

141

142 Materials and Methods

143 Patients, magnesium sulfate dosing and sampling

PT
144 Following Stanford University Institutional Review Board approval, women and their

145 neonates with the potential for magnesium sulfate exposure were enrolled in this

RI
146 prospective study conducted at Lucile Packard Children’s Hospital Stanford. Informed

147 consent was obtained in Labor and Delivery at the time of hospital admission. The

SC
148 mother, or mother and father (if possible), provided written informed consent for the

U
149 neonates. The study was conducted from October 2012 to May 2014. The study was
AN
150 registered at ClinicalTrials.gov prior to starting patient enrollment. (NCT01709630).

151 Pregnant women, aged 18-45, who were admitted to the hospital with
M

152 preeclampsia, preterm labor or extreme prematurity, and prescribed magnesium sulfate

153 were included. We excluded pregnant women who were on dialysis. Women received the
D

154 standardized dosing protocol for magnesium heptahydrate (MgSO4
7H20) 20g/500ml
TE

155 (4%) utilized at our institution, which is an intravenous 4 g loading dose (over 20 min)

156 and a 2 g/hr maintenance infusion as clinically indicated. Maternal blood samples were
EP

157 obtained for magnesium at baseline prior to magnesium sulfate administration; at 30 min,

158 1 hr, 2 hr, 4 hr, and every 6 hr during magnesium sulfate administration; and at 1 hr, 3 hr,
C
AC

159 6 hr, 9 hr, and 12 hr after magnesium sulfate was discontinued. Whenever possible, a

160 magnesium level was also obtained from umbilical cord venous blood at delivery.

161 Stanford’s Hospital Laboratory measured the magnesium levels using a Dimension®

162 RxL Max® Integrated Chemistry System.

163
 ACCEPTED MANUSCRIPT
8

164 Pharmacostatistical Analysis

165 Detailed pharmacokinetic model development and covariate analysis is described

166 in supplemental materials (Appendix A). Briefly, population PK modeling seeks to

167 measure the physiologic sources of variability within a population that affects drug

PT
168 disposition while maintaining and accounting for the individuality of each patient.

RI
169 NONMEM 7.2 (ICON Development Solutions, Ellicott City, MD) was used for

170 population pharmacokinetic analysis and RStudio Version 0.97.320 (RSTUDIO, Inc,

SC
171 Boston, MA) was used for goodness-of-fit diagnostics. One and two-compartment

172 models were evaluated. Selection between models was based on successful NONMEM

173
U
minimization with at least three significant digits in each parameter estimate, decrease in
AN
174 objective function (OBJ) of >3.84 (p<0.05), visual inspection of diagnostic scatter plots

175 (observed vs. individual and population predicted concentrations, residual/conditional

M

176 weighted residual vs. predicted concentration or time), the precision of the parameter
D

177 estimates measured by the percent standard error of the mean, and changes in the inter-
TE

178 individual and residual variability. The population pharmacokinetic analysis allowed for

179 investigation of the following covariates in the model: gestational age, presence of
EP

180 preeclampsia, maternal weight at the time of magnesium sulfate administration,

181 antepartum or postpartum status, and maternal creatinine.

C

182 Once infused, magnesium sulfate dissociates to Mg++ (magnesium ion). The half-
AC

183 life (t1/2) of magnesium was calculated using the following formula: t1/2 = 0.693*V/CL.

184 Final PK model evaluation occurred using a bootstrap analysis to evaluate parameter

185 uncertainty and estimate 95th percentile confidence intervals. One thousand replicates of
 ACCEPTED MANUSCRIPT
9

186 the dataset were created through repeat sampling with replacement. Model parameters

187 were estimated for each of the 1000 datasets.

188

189 Simulation

PT
190 A simulation study using NONMEM was performed to determine expected magnesium

RI
191 plasma concentrations of the study population. Simulated magnesium sulfate dose was 4

192 g loading dose administered over 20 minutes followed by an infusion of 2 g/hr for > 60

SC
193 hours. A second simulation was performed using the same dosing to determine time to

194 reach a serum magnesium level of 4.8 mg/dL. One thousand Monte Carlo simulation

195
U
replicates were performed, incorporating the final population model parameter estimates
AN
196 of fixed effects, inter-individual and residual random variability.

197 Serum magnesium concentration at steady state was calculated using simulation
M

198 for a 70 kg pregnant woman receiving a 4 g magnesium sulfate bolus administered over
D

199 20 minutes, followed by a 2 g/hr infusion. The steady state concentrations (CSS) of
TE

200 magnesium in preeclamptic and non-preeclamptic women were calculated using the

201 following formulas: 1.) For IV infusion, Css= Infusion rate/CL; and 2.) For IV bolus,
EP

202 Css= Dose/(CL*dosing interval).

203
C

204 Correlation with umbilical cord magnesium levels

AC

205 Using the final model, maternal serum magnesium levels were predicted at the time of

206 delivery for each umbilical cord sample obtained. The ratio of the mean umbilical vein

207 magnesium level at the time of delivery to the mean maternal serum magnesium level at

208 the time of delivery was calculated to determine the transfer ratio.
 ACCEPTED MANUSCRIPT
10

209

210 Results

211 All pregnant women who consented to the study, received magnesium sulfate and had at

212 least one magnesium level recorded while receiving magnesium sulfate were included in

PT
213 the data analysis. (Figure 1) A total of 111 maternal subjects with 687 magnesium levels

RI
214 and 66 umbilical cord blood magnesium levels were available for analysis. Baseline

215 characteristics of subjects are shown in Table 1. Preeclamptic women were significantly

SC
216 heavier, delivered at a later gestational age, and had higher baseline creatinine levels

217 compared to non-preeclamptic women who were administered magnesium sulfate.

218
U
Individual concentration-time profile data are outlined in Appendix B.
AN
219 Magnesium disposition in pregnant women was best fit using a one-compartment model,

220 as magnesium rapidly equilibrates between highly perfused organs and peripheral tissues.
M

221 In our covariate pharmacokinetic analysis, the effect of maternal weight on the volume of
D

222 distribution of magnesium, and the effect of preeclampsia on the clearance of

TE

223 magnesium, were statistically significant (p<0.001). Gestational age at the time of

224 magnesium sulfate administration, antepartum or postpartum status, and maternal

EP

225 creatinine were not significant in the final model.

226 Final parameter estimates and inter-individual variability including standard

C

227 errors of the point estimates are represented in Table 2. Observed vs. individual and
AC

228 population predicted concentrations are shown in Figure 2. Based on our pharmacokinetic

229 model, the mean population parameter estimate for magnesium clearance was 5.88 L/hr

230 in non-preeclamptic women and 3.98 L/hr in preeclamptic women. Volume of

231 distribution of magnesium was 22.5 L. The residual (unexplained) variability that existed
 ACCEPTED MANUSCRIPT
11

232 after the data were fitted to the population model was 18%. The half-life (t ½) of

233 magnesium was 2.7 hours in non-preeclamptic women and 3.9 hours in preeclamptic

234 women. The steady state concentration of magnesium in non-preeclamptic women was

235 5.1 mg/dL compared to 7.2 mg/dL in preeclamptic women.

PT
236 Simulations of 4 g loading dose of magnesium sulfate, followed by 2 g/hr infusion

RI
237 among pregnant women with and without preeclampsia and women of lowest (55 kg),

238 mean (88 kg), and highest (157 kg) body weights in our sample are shown in in Figure 3.

SC
239 Among pregnant women with preeclampsia, serum magnesium levels were significantly

240 higher during magnesium sulfate infusion compared to those of non-preeclamptics

241
U
receiving the same magnesium sulfate dosing. Women with the highest body weight had
AN
242 lower serum magnesium levels following the bolus administration of magnesium sulfate
M

243 relative to women with the lowest body weight. Preeclamptic women weighing more

244 than 100 kg took approximately twice as long to achieve a serum magnesium level of at
D

245 least 4.8 mg/dL compared to preeclamptic women weighing less than 100 kg (Figure 4).
TE

246 Weight-based equations derived from pharmacokinetic simulations for purposes of

247 dosing in non-preeclamptic and preeclamptic women are outlined in Appendix C.

EP

248 Maternal serum and umbilical cord magnesium levels were highly correlated with

249 an R2 = 0.793 (p<0.0001). The ratio of the umbilical vein magnesium level at the time of
C

250 delivery to maternal serum magnesium level at the time of delivery was 0.94 + 0.15.
AC

251

252 Comments

253 This study characterizes the pharmacokinetics of magnesium sulfate administered to

254 pregnant women, and demonstrates the pharmacokinetics are significantly impacted by
 ACCEPTED MANUSCRIPT
12

255 maternal weight and the presence of preeclampsia. This pharmacokinetic study is the

256 most sample-rich prospective study to date of magnesium sulfate using multiple sampling

257 blood levels of magnesium drawn for each patient. Knowledge of the impact of these

258 covariates will direct clinicians to more accurately achieve and maintain a target

PT
259 magnesium level. Applied to a much larger cohort of pregnant women, this

RI
260 pharmacokinetic model may allow us to individualize dosing protocols, as well as

261 determine therapeutic serum magnesium levels for non-preeclamptic indications.

SC
262 Magnesium disposition was adequately described by a one-compartment model.

263 This is consistent with two reports used for magnesium sulfate pharmacokinetic analysis

264
U
found in the literature.17,19 During pregnancy the body undergoes extensive physiologic
AN
265 alterations, which may alter pharmacokinetic and pharmacodynamic profiles of drugs.
M

266 Preeclamptic women often have impaired renal function that may impact drug clearance.

267 In our study, the volume of distribution and clearance values for magnesium were in
D

268 agreement with the values reported in the literature for pregnant women with
TE

269 preeclampsia (3.98 L/hr in our study vs. 4.81 L/hr, 4.28 L/hr, and 5.0 L/hr,

270 respectively).17-19 While our model included more complete covariate data than
EP

271 previously reported models, preeclampsia was the only covariate with a significant

272 impact on clearance.17-19 It is possible that the larger overall number of samples obtained
C

273 from preeclamptic women in our study compared to the number of samples from
AC

274 preeclamptic women in other studies accounts for the small differences in our modeling

275 results.

276 Magnesium is not metabolized and elimination of the drug is mainly conducted

277 through renal excretion. Therefore, the difference in clearance during and after
 ACCEPTED MANUSCRIPT
13

278 pregnancy likely reflects modifications in renal clearance (glomerular filtration and active

279 tubular secretion/reabsorption).22,23 Maternal adaptation to pregnancy involves dramatic

280 changes in renal hemodynamics, including a 40–65% increase in glomerular filtration

281 rate during the first half of gestation that is maintained during the later stages of

PT
282 pregnancy, as compared with pre-pregnancy or postpartum values.24 Interestingly, we

RI
283 found that pre-partum clearance of magnesium was less than postpartum clearance,

284 however, this effect was attenuated after controlling for preeclampsia. Clearance of

SC
285 magnesium did not change in the early postpartum period in preeclamptic women, likely

286 because improvements in renal function in the recovering preeclamptic patient are offset

287
U
by the normal postpartum physiological changes of decreased renal clearance.
AN
288 The current estimated maternal volume of distribution (22.5 L) is larger than the
M

289 typical plasma volume (3 L) and the typical extracellular fluid volume (15 L), confirming

290 that magnesium ion is freely distributed. In pregnancy, a substantial percentage of the
D

291 weight gain is attributed to increased total body water, of which approximately 80% is
TE

292 extracellular.25 Therefore, pregnancy may be associated with increased drug dilution,

293 particularly in women with more weight gain. We found that maternal weight at the time
EP

294 of delivery was an important covariate in the pharmacokinetic model. Our results suggest

295 the larger the maternal weight, the longer the time to reach steady state, or any specified
C

296 target level, when compared to women of a lighter weight. These findings are in
AC

297 agreement with what has previously been reported for women with large body mass

298 index.26 Women with larger body mass index have an increased volume of distribution,

299 and frequently have sub-therapeutic serum magnesium levels after receiving standard

300 doses of magnesium sulfate.26,27 Minimum levels of serum magnesium concentrations of

 ACCEPTED MANUSCRIPT
14

301 4.8 mg/dL have been suggested for the prevention and treatment of eclamptic seizures.28-
30
302 According to our pharmacokinetic model, preeclamptic women weighing more than

303 100 kg took approximately twice as long to achieve a serum magnesium level of at least

304 4.8 mg/dL compared to preeclamptic women weighing less than 100 kg.

PT
305 Our study suggests pregnant women administered magnesium sulfate for non-

RI
306 preeclamptic indications (i.e. neuroprotection and tocolysis), have an increased clearance

307 of the drug compared to that of preeclamptic women. The significant difference in

SC
308 magnesium clearance among preeclamptics and non-preeclamptics found in our study

309 suggests alternate infusion dosing may be warranted for different indications. In order to

310
U
achieve the same serum concentration of magnesium as that of preeclamptic women,
AN
311 non-preeclamptic women of the same weight need increased infusion doses of
M

312 magnesium sulfate. To our knowledge, this is the first pharmacokinetic model of

313 magnesium sulfate administration in non-preeclamptic women.

D

314 The BEAM, PREMAG, and ACTO trials were multicenter randomized trials
TE

315 evaluating magnesium sulfate administration for neuroprotection.15,31,32 A meta-analysis

316 of these trials, which enrolled almost 4,000 women, suggests magnesium sulfate in some
EP

317 dosage is beneficial in decreasing rates of cerebral palsy.33 However, there were

318 variations in the magnesium sulfate doses and timing intervals that women and fetuses in
C
AC

319 these studies received, therefore the most beneficial dosing regimen remains unknown.

320 The dosing of magnesium sulfate for neuroprotection used in these studies was based on

321 retrospective data.34 The optimal target serum level of magnesium for cerebral palsy risk

322 reduction is unknown. As we gain knowledge on the therapeutic level of serum

 ACCEPTED MANUSCRIPT
15

323 magnesium for other indications, this pharmacokinetic model could potentially be used to

324 adjust dosing based on maternal weight.

325 In addition to providing a detailed maternal pharmacokinetic model accounting for

326 multiple covariates, we have demonstrated that magnesium readily crosses the placenta.

PT
327 The study demonstrated a strong correlation between maternal serum levels and venous

RI
328 cord blood levels entering the fetus. These findings are similar to those previously

329 described by Hallak et al.35 The almost 1:1 ratio of magnesium in the pregnant mother

SC
330 and umbilical cord has important implications for studying neonatal pharmacokinetic and

331 pharmacodynamic effects.

U
AN
332 The strengths of this study include its prospective design, large sample size, and

333 the robust modeling based on multiple magnesium levels for each patient while
M

334 accounting for multiple covariates. Although the overall model adequately predicted the

335 individual serum magnesium concentrations after magnesium sulfate administration,

D

336 potential explanations for residual variability in our model include sample collection and
TE

337 measurement variability. There may also be important covariates to the pharmacokinetic

338 model which were unmeasured and unaccounted for. Given that enrollment in the study
EP

339 was at the time of hospital admission, we were unable to ascertain whether weight gain

340 during pregnancy was an important covariant in the pharmacokinetic model. We only
C
AC

341 classified women as preeclamptic or non-preeclamptic for the purposes of our

342 pharmacokinetic modeling, and we were not able to account for subtle differences among

343 women with mild forms of preeclampsia compared to those with more severe

344 preeclampsia that may impact the pharmacokinetic model. The severity of preeclampsia

345 may be based on blood pressure, lab values, and/or symptoms reported by the patient.36
 ACCEPTED MANUSCRIPT
16

346 Renal impairment is a key characteristic of many preeclamptic patients. Unlike the

347 Salinger et al study, creatinine levels were not a significant covariate in our

348 pharmacokinetic model.19 This may either be due to study limitations in measuring renal

349 function (most likely), or a true lack of correlation between renal function and

PT
350 magnesium disposition. Creatinine was measured as standard of care, rather than at set

RI
351 time points, which limits our ability to assess dynamic renal function.

352 In summary, we have accurately characterized the pharmacokinetics of

SC
353 magnesium sulfate administered to pregnant women. Preeclamptic status and maternal

354 weight significantly impact serum magnesium levels. When applied to larger cohorts of

355
U
pregnant women, this pharmacokinetic model could be used to tailor magnesium
AN
356 treatment protocols to account for these covariates. Future research using this data

357 includes comparing the proportion of pregnant women with therapeutic serum
M

358 magnesium levels at delivery after receipt of standard dosing versus weight-based dosing
D

359 of magnesium sulfate, as well as prediction models of serum magnesium in women who
TE

360 give birth to neonates with or without cerebral palsy. Ultimately, tailoring magnesium

361 sulfate treatment protocols will help advance the management of preeclampsia seizure
EP

362 prophylaxis and fetal neuroprotection by maximizing the benefits of magnesium sulfate

363 exposure while minimizing unwanted maternal and neonatal side effects.
C

364
AC
 ACCEPTED MANUSCRIPT
17

365 References

366 1. Pryde PG, Mittendorf R. Contemporary usage of obstetric magnesium sulfate:

367 indication, contradiction, and relevance of dose. Obstet Gynecol 2009; 114: 669-73.

368 2. Fox NS, Gelber SE, Kalish RB, Chasen ST. Contemporary practice patterns and beliefs

PT
369 regarding tocolysis among U.S. Maternal-fetal medicine specialists. Obstet Gynecol

RI
370 2008; 112: 42-7.

371 3. Duley L, Gulmezoglu AM, Henderson-Smart DJ, Chou D. Magnesium sulphate and

SC
372 other anticonvulsants for women with pre-eclampsia. Cochrane Database Syst Rev 2010:

373 CD000025.

374
U
4. Mittendorf R, Pryde PG, Roizen N: Second overview of relationships between
AN
375 antenatal pharmacologic magnesium sulfate and neurologic outcomes in children. J

376 Perinat Med 2004; 32: 201-10.

M

377 5. Mittendorf R, Pryde PG: A review of the role for magnesium sulphate in preterm
D

378 labour. BJOG 2005; 112 Suppl 1: 84-8.

TE

379 6. Mittendorf R, Dammann O, Lee KS: Brain lesions in newborns exposed to high-dose

380 magnesium sulfate during preterm labor. J Perinatol 2006; 26: 57-63.
EP

381 7. Greenberg MB, Penn AA, Thomas LJ, El-Sayed YY, Caughey AB, Lyell DJ: Neonatal

382 medical admission in a term and late-preterm cohort exposed to magnesium sulfate. Am J
C

383 Obstet Gynecol 2011; 204:515.e1-7.

AC

384 8. Riaz M, Porat R, Brodsky NL, Hurt H: The effects of maternal magnesium sulfate

385 treatment on newborns: a prospective controlled study. J Perinatol 1998; 18: 449-54.

386 9. Rasch DK, Huber PA, Richardson CJ, L'Hommedieu CS, Nelson TE, Reddi R:

387 Neurobehavioral effects of neonatal hypermagnesemia. J Pediatr 1982; 100: 272-6.

 ACCEPTED MANUSCRIPT
18

388 10. Green KW, Key TC, Coen R, Resnik R: The effects of maternally administered

389 magnesium sulfate on the neonate. Am J Obstet Gynecol 1983; 146: 29-33.

390 11. del moral T, Gonzalez-Quintero VH, Claure N, Vanbuskirk S, Bancalari E: Antenatal

391 exposure to magnesium sulfate and the incidence of patent ductus arteriosus in extremely

PT
392 low birth weight infants. J Perinatol 2007; 27: 154-7.

RI
393 12. Shokry M, Elsedfy GO, Bassiouny MM, Anmin M, Abozid H: Effects of antenatal

394 magnesium sulfate therapy on cerebral and systemic hemodynamics in preterm

SC
395 newborns. Acta Obstet Gynecol Scand 2010; 89: 801- 6.

396 13. Yokoyama K, Takahashi N, Yada Y, Koike Y, Kawamata R, Uehara R, Kono Y,

397
U
Honma Y, Momoi MY: Prolonged maternal magnesium administration and bone
AN
398 metabolism in neonates. Early Hum Dev 2010; 86: 187-91.

399 14. Lyell DJ, Pullen K, Campbell L, Ching S, Druzin ML, Chitkara U, Burrs D, Caughey
M

400 AB, El-Sayed YY: Magnesium sulfate compared with nifedipine for acute tocolysis of
D

401 preterm labor: a randomized controlled trial. Obstet Gynecol 2007; 110: 61-7.
TE

402 15. Rouse DJ, Hirtz, DG, Thom E et al. A randomized, controlled trial of magnesium

403 sulfate for the prevention of cerebral palsy. N Engl J Med 2008; 359:895-905.
EP

404 16. Okusanya BO, Oladapo OT, Long Q et al. Clinical pharmacokinetic properties of

405 magnesium sulphate in women with preeclampsia and eclampsia: a systematic review.
C

406 BJOG 2015; DOI: 10.1111/1471-0528.13753.

AC

407 17. Chuan FS, Charles BG, Boyle RK et al. Population pharmacokinetics of magnesium

408 in preeclampsia. Am J Obstet Gynecol 2001;185:593-9.

409 18. Lu J, Pfister M, Ferrari P et al. Pharmacokinetic-pharmacodynamic modeling of

410 magnesium plasma concentration and blood pressure in preeclamptic women. Clin
 ACCEPTED MANUSCRIPT
19

411 Pharmacokinet 2002; 41: 1105-1113.

412 19. Salinger D, Mundle S, .Regi A, Bracken H, Winikoff B, Vicini P, Easterling T.

413 Magnesium sulphate for prevention of eclampsia: are ntramuscular and intravenous

414 regimens equivalent? A population pharmacokinetic study. BJOG 2013;120:894–900.

PT
415 20. Pritchard JA. The use of magnesium sulfate in preeclampsia-eclampsia. J Reprod

RI
416 Med 1979; 23: 107-14.

417 21. McPherson JA, Rouse DJ, Grobman WA, Palatnik A, Stamilio DM. Association of

SC
418 duration of neuroprotective magnesium sulfate infusion with neonatal and maternal

419 outcomes. Obstet Gynecol 2014; 124: 749-55.

420
U
22. Kokko J, Tannen RL, Kersey R. Magnesium disorders: fluids and electrolytes. 3rd ed.
AN
421 Philadelphia (PA): WB Sanders Company; 1996. p. 421.

422 23. Quamme GA, Dirks JH. Magnesium metabolism. In: Narins RG, ed. Clinical
M

423 disorders of fluid and electrolyte metabolism. 5th ed. New York (NY): McGraw Hill;
D

424 1994. p. 373.

TE

425 24. Jeyabalan A, Conrad KP. Renal function during normal pregnancy and preeclampsia.

426 Frontiers in bioscience: a journal and virtual library. 2007; 12: 2425-37.
EP

427 25. Krauer B, Krauer F. Drug kinetics in pregnancy. Clin Pharmacokinet 1977; 2: 167-81.

428 26. Tudela CM, McIntire DD, Alexander JM. Effect of maternal body mass index on
C

429 serum magnesium levels given for seizure prophylaxis. Obstet Gynecol 2013;121: 314–
AC

430 20.

431 27. Dayicioglu V, Sahinoglu Z, Kol E, Kucukbas M. The use of standard dose of

432 magnesium sulfate in prophylaxis of eclamptic seizures: do body mass index alterations

433 have any effect on success? Hypertens Pregnancy 2003;22: 257–65.

 ACCEPTED MANUSCRIPT
20

434 28. Sibai BM. Hypertension. In: Gabbe SG, Niebyl JR, Simpson JL editors. Obstetrics

435 Normal and Problem Pregnancies, 4th edn. New York: Churchill Livingston; 2002. pp.

436 969.

437 29. Redman C. Hypertension in pregnancy. In: de Swiet SM editor. Medical Disorders in

PT
438 Obstetric Practice, 2nd edn. Oxford: Blackwell Scientific Publications; 1989. pp. 278.

RI
439 30. Dildy GA, Phelan JP, Cotton DB. Complications of Pregnancy- Induced

440 Hypertension. In: Clark SL, Cotton DB, Hankins GDV, Phelan JP editors. Critical Care

SC
441 Obstetrics, 2nd edn. Oxford: Blackwell Scientific Publications; 1991. pp. 256.

442 31. Marret S, Marpeau L, Zupan-Simunek V, Eurin D, Le ́veˆque C, Hellot M, Be ́nichou

443
U
J on behalf of the PREMAG trial group. Magnesium sulphate given before very-preterm
AN
444 birth to protect infant brain: the randomised controlled PREMAG trial. BJOG

445 2007;114:310–318.
M

446 32. Crowther CA, Hiller JE, Doyle LW, Haslam RR. Effect of magnesium sulfate given
D

447 for neuroprotection before preterm birth: a randomized controlled trial. JAMA
TE

448 2003;290:2669-2676.

449 33. Conde-Agudelo A, Romero R. Antenatal magnesium sulfate for the prevention of
EP

450 cerebral palsy in preterm infants less than 34 weeks’ gestation: a systematic review and

451 metaanalysis. Am J Obstet Gynecol 2009; 200: 595-609.

C

452 34. Nelson KB, Grether JK. Can magnesium sulfate reduce the risk of cerebral palsy in
AC

453 very low birthweight infants? Pediatrics 1995;95:263-9.

454 35. Hallak M, Berry SM, Madincea F et al. Fetal serum and amniotic fluid magnesium

455 concentrations with maternal treatment. Obstet Gynecol 1993;81:185-8.

456 36. American College of Obstetricians and Gynecologists: Task Force on Hypertension in
 ACCEPTED MANUSCRIPT
21

457 Pregnancy. Hypertension in pregnancy. Report of the American College of Obstetricians

458 and Gynecologists’ Task Force on Hypertension in Pregnancy. Obstet Gynecol 2013;

459 122: 1122-31.

460

PT
RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT
22

Table 1. Baseline characteristics for study cohort by indication.

Mean maternal age 29.3 ± 1.3

(years)
Mean maternal weight 87.6 ± 3.8
(kg)

PT
Mean gestational age 33.7 ± 0.9
(weeks)
Mean baseline 1.96 ± 0.04

RI
magnesium (mg/dL)
Mean baseline creatinine 0.80 ± 0.06
(mg/dL)

SC
Preeclamptic 92 (83%)

* Values are presented as mean + standard deviation, or number (%)

U
AN
M
D

Pharmacokinetics of magnesium sulfate in pregnancy

TE
C EP
AC
 ACCEPTED MANUSCRIPT
23

Table 2. Final parameter estimates for the population PK model. IIV is the inter-

individual variability, RSE is the relative standard error.

PT
Bootstrap
Parameter Estimate %IIV %RSE
95% CI
CL (L/hr) – without preeclampsia 5.88 3.7 5.1 (5.35, 6.55)

RI
CL (L/hr) – with preeclampsia 3.98 41 4.5 (3.65, 4.36)
V (L/70 kg) 22.5 23.6 4.1 (20.7, 24.4)
Residual Error 0.73 18 (0.6, 0.86)

SC
*Model parameterized by clearance (CL, L/hr) and volume of central compartment (V, L)

U
AN
M

Pharmacokinetics of magnesium sulfate in pregnancy

D
TE
C EP
AC
 ACCEPTED MANUSCRIPT
24

Figure Legends:

Figure 1. Flow diagram of subjects included in analysis. Forty patients were excluded

from the analysis because magnesium sulfate was not administered, or a serum

magnesium level was not obtained after the initiation of magnesium sulfate.

PT
RI
Figure 2. Observed versus population-predicted and observed versus individual-

predicted concentrations.

U SC
Figure 3. Simulation of 4g loading dose of magnesium sulfate, followed by 2g/hr
AN
infusion among pregnant women with and without preeclampsia and women of lowest

(55 kg), mean (88kg), and highest (157kg) body weights in our study.
M
D

Figure 4. Simulation to predict time to serum magnesium level > 4.8 mg/dL for our study
TE

population following the administration of 4 g magnesium sulfate bolus over 20 minutes,

followed by a 2 g/hr infusion.

C EP
AC

Pharmacokinetics of magnesium sulfate in pregnancy

 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

PT
RI
U SC
AN
M
D
TE
EP
C
AC
 ACCEPTED MANUSCRIPT

Appendix A

Pharmacokinetic Model Development

All models were run using the first order conditional estimation with

interaction (FOCE-I) method. RStudio Version 0.97.320 (RStudio, Inc., Boston, MA)

PT
was used for goodness-of-fit diagnostics. Selection between models was based on

RI
successful NONMEM minimization with at least three significant digits in each

parameter estimate, decrease in objective function (OBJ) of >3.84 (p<0.05), visual

SC
inspection of diagnostic scatter plots (observed vs. individual and population

predicted concentrations, residual/conditional weighted residual vs. predicted

U
AN
concentration or time), the precision of the parameter estimates measured by the

percent standard error of the mean, and changes in the inter-individual and residual
M

variability.

One and two-compartment models were evaluated. The inter-individual

D

variability (IIV) of parameters was described using an exponential variance model:

TE

Pi = θkexp(ηki), where Pi is the estimated parameter value for individual i, θk is the

typical population value of parameter k, and ηki is the inter-individual random effect
EP

for individual i and parameter k.

C

Additive, proportional and combined (additive and proportional) error

AC

functions were evaluated to model the residual variability. An additive residual

error function was incorporated into the final model: Cobs,ij = Cpred,ij+εij where Cobs,ij is

the observed concentration j in individual i, Cpred,ij is the individual predicted

concentration, εij is the residual random error.

 ACCEPTED MANUSCRIPT

Covariate Analysis

Once the structural pharmacokinetic model was established, biologically or

clinically plausible covariates were evaluated for their influence on pharmacokinetic

parameters. Covariate analysis was conducted using a standard forward addition

PT
and backward elimination procedure. An increase in OBJ >7.88 (p <0.005) during

RI
backward elimination was considered significant and warranted inclusion of the

covariate in the final model. Covariates of weight, estimated gestational age,

SC
creatinine, presence of preeclampsia and time of delivery were explored.

A linear weight model standardized to an adult body weight of 70kg was

U
AN
used to evaluate the influence of weight on clearance and volume of distribution:

Pi = Pref * (WTi /70 kg)

M

where Pi is the typical parameter value for individual i with weight WTi, Pref is the

parameter value for an individual with weight equal to the reference weight of 70kg.
D

The effects of estimated gestational age and creatinine on clearance were

TE

explored using power models

Pi = θk*COVθcov
EP

where Pi is the parameter value for individual i, θk is the population estimate of

C

parameter k, COV is the value of the covariate (estimated gestational age or serum
AC

creatinine) and θcov is the estimated power parameter.

Finally, the effects of the presence of preeclampsia and time of delivery on

clearance were evaluated as dichotomous variables.

Pi = θk*COVC
 ACCEPTED MANUSCRIPT

where Pi is the parameter value for individual i, θk is the population estimate of

parameter k, COV is a factor that describes the magnitude of effect of preeclampsia

or time of delivery, and C is an indicator value assigned to 1 when the patient

experienced preeclampsia or was post-partum.

PT
RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Appendix B: Individual concentration-time profile data

The dots represent actual observations for each individual. The solid line

represents the individual predicted serum magnesium levels by the PK model. The

dotted line represents the population predicted serum magnesium levels by the PK

PT
model.

RI
U SC
AN
M
D
TE
C EP
AC
 ACCEPTED MANUSCRIPT

Appendix C: Weight-based equations derived from pharmacokinetic modeling

simulations for purposes of dosing in non-preeclamptic and preeclamptic women:

1. A 20-minute loading dose that produces an area under the curve (AUC0-20min)

PT
equal to that in a 70 kg woman administered a 4g IV bolus of magnesium sulfate

RI
over 20 minutes.

Eq(C.1): In non-preeclamptic women:

SC
0.32

 
() =
1.94 − 0.324 × () × 1 −
⁄( ) "

Eq(C.2): In preeclamptic women:

U
AN
0.145

 
() =
1.31 − 0.324 × () × 1 −
$⁄( ) "
M

2. An infusion rate that produces an AUC0-24hr equal to that in a 70 kg woman

D

administered a 2g/hr IV infusion of magnesium sulfate.

TE

Eq(C.3) In non-preeclamptic women:

EP

236.7
%&'  (
(⁄ℎ) =
140.9 − 0.324 × () × 1 −
$,$.$⁄( ) "
C

Eq(C.4) In preeclamptic women:

AC

107.7
%&'  (
(⁄ℎ) =
95.0 − 0.324 × () × 1 −
-.,./⁄( ) "