Maternal erythropoietin in singleton pregnancies: A randomized trial on the effect of oral hernatinic supplementation Desmond P. J. Barton, MD, * Marie-Therese joy, SRN, SCM, ' Terence R. J. Lappin, PhD, ' Morteza Afrasiabi, PhD, 'Jorge G. Morel, PhD, 'Ioan O'Riordan, " John F. Murphy, MD, ' and Colm O'Herlihy, MD* Dublin and Belfast, Ireland, and Tampa, Florida

08JECTIVE: Our purpose was to study the effect of hernatinic supplementation on the maternal erythropoietin response during singleton pregnancy. STUDY DESIGN: In a randomized, double-blind trial 97 patients with a first-trimester hemoglobin level ; a- 14.0 gm/dI received either iron and folic acid (hematinic group, n= 53) or a placebo (n = 44). Serial hemoglobin, hematocrit, and serum erythropoietin were recorded from maternal blood and from cord blood on delivery. Serum ferritin was measured in the first trimester, at 36 weeks' gestation, and in cord blood. RESULTS: In both groups (1) the mean hemoglobin was lower (p < 0.01) at 40 weeks' gestation than when first examined and (2) the mean serum erythropoietin was higher (p < 0.01). The mean serum ferritin was lower (p < 0.001) in both groups at 36 weeks' gestation than at presentation but higher (p = 0.04) in the hernatinic group than in the placebo group. The mean hemoglobin and hematocrit were similar in the two groups until the third trimester but thereafter were higher (p < 0.05) in the hematinic group. The mean maternal serum erythropoietin was higher (p < 0.05) in the placebo group than in the hematinic group after 24 weeks' gestation. The mean cord blood hematologic values were similar in the two groups. CONCLUSION: Maternal serum erythropoletin increased during pregnancy, but this response was reduced In the third trimester in the hernatinic-supplemented group. (Am J OesTET GYNEcoL 1994; 170: 896-901. )

Key words: Maternal erythropoietin, hematinic supplementation, double-blind study

Routine hematinic supplementation throughout pregnancy has been recommended by many authors, but a more selective policy has been proposed by others, especially in patients with high hemoglobin and hematocrit values early in pregnancy. ' ' The physio- logic anemia of pregnancy is attributed to differential increases in plasma volume and red cell mass. Erythro- poietin, the main hormonal regulator of erythropoiesis, is released in the nonpregnant state by the stimuli of anemia and by a reduction in the oxygen available to certain organs. ' The kidney is the major site of eryth-

From the National Maternity Hospital and University College Dublin a the Department of Haematoloo Royal Victoria Hospital, ' the Department of Epidemiology and Biostatistws, College of Public Health, University of South Florida, ' and the Department of Hae- matoloo St. James Hospttal. d Received for publication May 21,1993; revised A ugust 19,1993; accepted September 28,1993. Reprint requests: Desmond P. J. Barton, MA Department of Obstet- rics and Gynecoloo Temple University School of Medicine, Broad and Ontario Streets, Philadelphia, PA 19140. Copyright C 1994 by Mosby--Year Book, Inc. 0002-9378194 $3.00 +0 611151696

ropoietin production in the adult, althouth extrarenal sources exist, most notably the liver. Although the physiologic mechanisms involved in the regulation of erythropoietin release in the nonpregnant state are well understood, in pregnancy these are less well defined but presumed to be similar.

Studies on maternal erythropoietin in pregnancy have been diverse, usually cross-sectional, with small sample sizes and heterogeneous groups of patients. Furthermore, because different assays have been used to measure erythropoietin in serum and in other body fluids, it is difficult to determine the maternal erythro- poietin response. ` Most often it has'not been stated whether patients were receiving hematinic supplemen- tation, and the effect of hernatinic supplementation on the maternal erythropoietin response during preg- nancy has not been reported.

The purpose of this study was to determine the effect of oral hematinic supplementation on the maternal erythropoietin response during pregnancy in patients who were not anemic in the first trimester. The null hypothesis was that there would be no difference in the

896

Volume 170, Number 3 Am j Obstet Gvnecol

maternal erythropoietin response during pregnancy in the hernatinic-supplemented and nonsupplemented groups.

Material and methods Patient eligibility and randomization. All patients

who were examined during the first trimester at the National Maternity Hospital, Dublin, with a singleton pregnancy and with hemoglobin ýý: 14.0 gm/dl were eligible for study. In the hospital population of single- ton pregnancies a first-trimester hemoglobin of 14.0 gm/dl represented the 90th percentile. Patients were excluded if they had had a recent blood transfusion, chronic respiratory disease, chronic hypertension, renal disease, diabetes mellitus, a history of a hematologic disorder (including rhesus isoimmunization), and alco- hol dependence ....... Each eligible patient was inter- viewed by two of the authors (D. P. J. B. and M. -T. J. ) after the initial visit, and informed consent was obtained. The study was approved by the Hospital's Ethics Com- mittee. Patients could withdraw from the study at any stage during the pregnancy and were withdrawn if anemia (hemoglobin <I 0 gm/dl on two consecutive occasions 2 weeks apart) developed.

By means of computer-generated numbers patients were randomized, double-blind, to the placebo or he- matinic groups. Patients in the hernatinic group re- ceived iron and folic acid tablets, one tablet to be taken by mouth twice daily (each tablet contained 0.5 mg of folic acid and 60 mg of elemental iron). The placebo tablets, also to be taken by mouth twice daily, were identical in size, shape, and color to the iron and folic acid tablets and contained the same excipients (Glaxo, Dublin). The patients began taking the placebo tablets or hematinics at the end of the first trimester.

Antenatal care. Patients had hospital visits for the study at 24,28,32,36, and 40 weeks' gestation, when blood was drawn for hemoglobin, hematocrit, and erythropoietin determinations and for serum ferritin values on entry into the study and at 36 weeks' gesta- tion. Cord venous blood was drawn on delivery, and the same hematologic index values were determined. He- moglobin and hematocrit were determined on an automated Coulter counter (Coulter Electronics, Harpenden, Herts, U. K. ). Blood drawn for erythropoi- etin was allowed to clot, and serum was separated within 4 hours and stored at - 70* C until analyzed by radio- immunoassay, as previously described. " The interassay coefficient of variation was 13%, and the intraassay variation was 6%. Serum ferritin (micrograms per liter) was measured by radioimmunoassay (Travenol, Cam- bridge, Mass. ). Patients were seen at each hospital visit and after delivery by one of the authors (M. -Tj. ). During the study the 10th percentile for birth weight at term was approximately 2700 gm.

Barton et al. 897

Table 1. Demographic data

Placebo) I (n = 44

Hematin' (n = 53)

Nulliparous 20 25 Cigarette smoking 14 25 Hypertensive disorder 4 4 Antepartum hemorrhage 2 S Birth weight < 2.7 kg 7 5 Perinatal death 0 1 Cesarean delivery 4 4

Statistical analysis. Published data were reviewed to determine sample size. We used a variance in the maternal serum erythropoietin of 225 (SD of 15 mU/mJ) at each gestation of the study; a difference between the mean serum crythropoietin in the two groups of > 10 mU/ml was considered significant. With " probability of 0.05 of making a type I error, to detect " difference between the mean serum crythropoietin of at least 10 mU/ml with a power of 0.8,35 patients would be required in each group. Because it was antic- ipated that some patients would request to be with- drawn from the study, default from antenatal visits, or be delivered before term, our intention was to have 45 patients in each group.

The data were analyzed with the generalized linear model procedure of the SAS software. Analysis of covariance was used to determine, at each different time point (for hemoglobin, hematocrit, and erythropoictin), whether adjusted means (least-square means) of the placebo and hematinic groups were the same. The hematinic values (hemoglobin, hematocrit, and cryth- ropoietin) at the initial visit (that is, before that patients received hernatinic or placebo) were the covariates. To assess the possible effect of cigarette smoking on the outcome variables, analysis of covaTiance was used to compute adjusted means (least-square means). For the cord blood values a two-tailed t test was performed. The paired t test was used to compare hematologic values on entry into the study with those at 40 weeks' gestation (or 36 weeks' gestation in the case of serum ferritin). Results are expressed as mean ± SE.

Results "ere were 53 patients in the hematinic group and

44 in the placebo group, which represented about one fifth of all eligible patients during the course of the study. The demographic data are shown in Table 1. The mean gestation of entry into the study was 12 weeks, and no patient in the study was first seen after the first trimester. There was no significant difference in the numbers of smokers in the two groups (-X' 2.36, P>0.05).

The serial mean hematocrit and hemoglobin data are

898 Barton et aL March 1994 Am j Obstet Gynecol

Table 11. Serial maternal hemoglobin levels (grams per deciliter) during pregnancy

I Placebo Hematintc

Gestational age (wk) I

No. Mean t SE No. Mean -2: SE Significance

Presentation 44 14.42 ± 0.05 (14.43 ± 0.05) 53 14.31 ± 0.04 (14.31 0.05) p=0.08 (p = 0.08) 24 40 12.43 ± 0.57 (12.68 ± 0.53) 45 13.44 ± 0.54(13.33 0.48) p=0.21 (p = 0.37) 28 41 12.29 t 0.13 (12.36 -t 0.13) 50 12.55 0.12 (12.53 -t 0.12) p=0.144 (p = 0.33) 32 42 12.29 ± 0.24 (12.30 ± 0.25) 49 12.87 0.23 (12.85 --!: 0.23) p=0.09 (P = 0.11) 36 40 12.57 ± 0.32 (12.71 t 0.38) 47 13.46 ± 0.29 (13.30 ± 0.34) P=0.043 (p = 0.25) 40 18 11.97 t 0.35 (12.01 :t0.35) 30 13.68 ± 0.27 (13.66 :t0.27) p<0.001 (P < 0.001)

Analysis of covariance with results at presentation as the covariate. Results in parentheses adjusted for cigarette smoking.

Table III. Serial maternal hematocrit during pregnancy Placebo Hematinic

Gestational age (wk) No. Mean ± SE No. Mean ± SE Significance

Presentation 44 0.429 ± 0.002 (0.428 :t0.002) 52 0.425 :t0.002 (0.425 :t0.002) P=0.27 (p = 0.30) 24 38 0.374 ± 0.004 (0.374 ± 0.004) 44 0.393 ± 0.004 (0.392 ± 0.004) p=0.003 (p = 0.004) 28 20 0.371 ± 0.007 (0.373 ± 0.007) 42 0.377 ± 0.005 (0,377 ± 0.005) p=0.419 (p = 0.65) 32 41 0.370 ± 0.004 (0.372 ± 0.005) 47 0.390 ± 0.004 (0,390 ± 0.004) P<0.001 (p = 0.004) 36 39 0.375 ± 0.004 (0.377 ± 0.004) 47 0.399 ± 0.004 (0.399 ± 0.004) p<0.001 (p < 0.001) 40 17 0.366 ± 0.009 (0.367 ± 0.009) 29 0.410 ± 0.007 (0.410 ± 0.007) p<0.001 (p < 0.001)

Analysis of covariance with results at presentation at the covariate. Results in parentheses adjusted for cigarette smoking

Table IV. Serial maternal serum erythropoietin (milliunits per milliliter) during pregnancy Placebo Htmaitnic

Gestational age (wk) No. Mean ± SE No. Mean ± SE Significance

Presentation 44 21.57 t 1.11 (21.18 :t1.07) 50 22.86 ± 0.92(22.82 ± 0.98) p=0.37 (p = 0.26) 24 39 34.17 ± 1.84(31.81 ± 2.33) 45 31.78 ± 1.73 (32.82 ± 2.09) p=0.348 (p = 0.75) 28 38 39.08 ± 1.81 (37.18 ± 2.13) 48 32.81 ± 1.70(33.56 ± 1.92) p=0.014 (p = 0.21) 32 41 48.90 ± 3.16(46.91 ± 3.61) 49 37.87 ± 2.99(38.35 ± 3.20) p=0.008 (p = 0.08) 36 36 54.39 ± 4.30(51.43 4.96) 48 42.67 ± 3.81 (43.11 4.23) P=0.045 (p = 0.20) 40 16 60.49 ±: 4.85(60.27 5.47) 29 37.33 ± 3.67 (36.47 4.13) P=0.001 (p = 0.001)

Analysis of covariance with results at presentation as the covariate. Results in parentheses adjusted for cigarette smoking

shown in Tables 11 and 111. In the placebo group the hemoglobin and hematocrit values fell gradually in the second half of pregnancy, although there was a small increase in both at 36 weeks' gestation. In contrast, in the hematinic group, after a fall in hemoglobin and hematocrit, these values then increased throughout the third trimester. When paired data only were analyzed, in the placebo group the mean hemoglobin was lower (p < 0.000 1) at 40 weeks' gestation (12.01 ± 0.32) than at the initial examination (14.44 - 0.1). Similarly, in the hematinic group the mean hemoglobin was lower (p = 0.002) at 40 weeks' gestation (13.47 ± 0.25) than at presentation (14.28 ± 0.05). No patients were with- drawn from the study because of anemia. The mean hemoglobin levels between the groups were signifi- cantly different at 36 and 40 weeks' gestation, a result unaffected when the data were adjusted for maternal smoking. There were, however, fewer data in the pla-

cebo group at 40 weeks' gestation because of delivery before then or to blood sampling errors. The mean hematocrit values differed in the third trimester (p < 0.05), results unaffected when corrected for ma- ternal cigarette smoking.

Although the two groups had high first-trimester hemoglobin levels, the mean maternal serum erythro- poietin increased significantly from the first trimester during pregnancy (Table IV). Wien. paired data only were analyzed, in the placebo group the mean maternal serum erythropoictin level (U/ml) at 40 weeks' gestation (58.88 t 7.61) was higher (p < 0.0001) than that at presentation (22.56 t 2.04). Similarly, in the hematinic

group the mean maternal serum erythropoietin level

was higher (p < 0.0001) at 40 weeks' gestation (37.67 ± 2.85) than at presentation (22.44 t 0.83). In the first two trimesters the mean erythropoietin levels of the two groups were similar. They differed throughout

Volume 170, Number 3 Am j Obstet Gynecol

Table V. Maternal serum ferritin (micrograms per liter) during pregnancy

Barton et al. 899

Placebo Hematinic

Gestational age (wk) No. Mean :t SE No. Mean ± SE Significance

Presentation 38 43.93 :t4.34 (43.10 t 6.20) 36 34 12.78 t 2.81 (10.76 t 7.25)

46 47.53 t 6.48 (47,53 ±: 5.50) p=0.66 (p = 0.59) 43 32.60 t 8.21 (32.45 ýt 6.32) p=0.04 (p = 0.03)

Results in parentheses adjusted for cigarette smoking.

Table VI. Mean cord blood values

Placebo Hematinic

Hematologic parameter No. Mean t SE No. Mtan t SE Significance

Hemoglobin (gm/dl) 34 16.71 ± 0.22 (16.83 t 0.62) 47 17.30 t 0.64 (17.31 ± 0.50) p=0.45 (p = 0.54) Hematocrit (%) 33 0.52 ± 0.01 (0.52 ±: 0.01) 45 0.52 :t0.01 (0.52 ± 0.01) p=0.91 (P = 0.81) Serum ferritin (ýLg/L) 32 156.34 t 13.27 (144.27 ± 15.63) 43 165.23 ± 14.24 (162.42 ± 12-79) p=0.66 (p = 0.36) Erythropoietin (mU/ml) 32 46.19 ± 6.05 (47.47 i: 5.71) 44 52.77 ± 4.13 (52.88 ± 4.61) p=0.35 (p = 0.46)

Results in parentheses adjusted for cigarette smoking.

the third trimester, although the difference was just

significant at 36 weeks' gestation (Fable IV). In the hematinic group the mean maternal serum erythropoi- etin changed little after 28 weeks' gestation.

The mean serum ferritin was similar in both groups at presentation but had fallen in the two groups by 36

weeks' gestation (Table V). When paired results were analyzed, the mean serum ferritin (ýLg/L) had decreased in the placebo group (47.47 t 5.45 to 12-86 t 3.10, p<0.0001) and in the hernatinic group (42.69 ±: 4.23

to 23.83 t 5.78, p=0.000 1). The mean serum ferritin

was higher in the hematinic group at 36 weeks' gesta- tion (Table V).

The mean cord blood values in the two groups were similar (Table VI). Wien the cord blood results were corrected for cigarette smoking during pregnancy no significant change was noted. The cord blood results were not significantly affected by the mode of delivery (vaginal vs cesarean), although there were only eight cesarean deliveries (three elective, five in labor) in the study. The condition of the newborn, as determined by

the I- and 5-minute Apgar score, the need for neonatal resuscitation, and admission to the neonatal intensive

care unit, did not have a significant impact on the cord blood results.

The hematologic profiles of those cases associated with anteparturn hemorrhage (n = 5), hypertensive dis-

orders (n = 8), or low birth weight (<I Oth percentile) (n = 12) were similar to those of the remaining pa- tients. There was one unexplained perinatal death at 37

weeks' gestation in the hematinic group of an appro- priately grown, normal fetus (as confirmed on post- morten examination) to a multiparous woman who smoked cigarettes in pregnancy.

Comment During the first two trimesters the hematologic index

values of the two groups were similar. In the third trimester significant differences were observed in the mean hematocrit, hemoglobin, and mean serum eryth- ropoietin. In both groups the changes in hemoglobin

and hematocrit were mirrored by reciprocal changes in

erythropoietin, a relationship that has been clearly documented in the nonpregnant state. Because of the effect of cigarette smoking on the serum erythropoietin, we analyzed our results to determine whether this was a contributing factor to any observed differences between the two groups. It appeared that cigarette smoking did

not significantly affect the results, but it is difficult to quantify cigarette smoking in pregnancy, especially be-

cause patients enrolled in the study may have been

motivated to stop or substantially reduce their habit during pregnancy.

Until recently most studies on cord blood erythropoi- etin have been on samples taken at delivery. The King's College group'"' has reported that fetal erythropoi- etin, measured from cordocentesis samples, was in-

creased in pregnancies complicated by red blood cell isoimmunization, diabetes mellitus, and intrauterine

growth retardation. Labor has also been reported to be

asscwiated with increased cord erythropoietin levels,

whereas Halmesmaki et al. ' reported that the mode of delivery did not affect the cord erythropoietin levels. In

contrast to our findings, Fenton et al. " and Romsto et al. " noted an increased cord ferritin level in patients receiving hematinic supplementation, but their study groups were not similar to ours. We found that the cord blood results were similar in both groups, suggesting that the maternal and fetal hematologic system function

900 Barton et al.

independently, at least in patients with a high hemo- globin level in the first trimester of pregnancy.

In our nonpregnant healthy population the normal range of serum erythropoietin is 5 to 30 mU/ml. " The erythropoietin values in early pregnancy in both groups were within this range in 91.8% (n = 89) of patients. A similar finding for early pregnancy has been reported in other studies, including 40 of 41 patients in the study by Howells et al. ' and Beguin et al. "' During the second and third trimesters there is expansion of blood volume and red cell mass. ' Before the end of the first trimester there is an increase in the blood volume and, in spite of this dilutional effect, the serum erythropoietin concen- tration in the first trimester was not reduced compared with nonpregnant levels. There is dispute, however,

about whether there is an expansion or a reduction in the red blood cell mass in the first trimester! The latter response would be difficult to explain and would not be in keeping with the other maternal physiologic adapta- tions during pregnancy!

Our data clearly show for the first time that in

patients with a high hemoglobin early in pregnancy the mean maternal serum erythropoietin concentration in- creased from the first trimester and that the response in the third trimester was reduced in patients taking he-

matinic supplementation. In cross-sectional studies Be- guin et al., ̀ '" investigated whether the erythropoietin response in pregnancy was appropriate for the degree

of anemia and reported an impaired response in early pregnancy that had recovered in late pregnancy. Whether patients were receiving hernatinic supplemen- tation was not stated, and part of the comparative analysis was based on erythropoeitin data obtained from nonpregnant normal women and from patients with anemias of various causes. Our data show that in

spite of the substantial increase in plasma volume in

early pregnancy the serum erythropoietin level had increased. Furthermore, no patients were withdrawn from the study because of anemia.

To explain this increase in erythropoietin, a mecha- nism involving maternal tissue hypooxygenation seems unlikely, because the physiologic changes of pregnancy, such as the increased circulating blood volume, the increased blood flow to the kidneys, and the rise in 2,3-diphosphoglycerate, enhance tissue perfusion and tissue oxygenation. Not all authors, however, agree on the changes in 2,3-diphosphoglycerate during preg- nancy. " Factors other than circulating oxygen level, red blood cell mass, and hernatocrit may be important for

erythropoietin release, and recently increased plasma viscosity has been shown to inhibit the erythropoietin response in the nonpregnant state. "' In addition, serum levels of erythropoietin only crudely assess both the response to stimuli and hormone activity. In the non- pregnant state understanding of the interaction be-

March 1994 Am j Obstet Gynecol

tween erythropoietin and its receptor, receptor expres- sion, and signaling pathways and the interaction of erythropoietin with other hormones and with cytokines remains incomplete. " It has yet to be established whether a fetoplacental mechanism affects the maternal erythropoietin response. For example, in earlier animal studies placental lactogen augmented erythropoictin activity. " Widness et al. " reported in 1984 that the changes in serum erythropoietin and human placental lactogen in pregnancy were similar in six patients, but these could be parallel and unrelated responses. Al- though erythropoietin can be transferred across the placenta in mice, it is unlikely to cross the human placenta because its molecular mass is 30 kd. 2- "

It is of interest that the erythropoietin results were significantly different only in the last trimester of preg- nancy. Fetal iron requirements are greatest and mater- nal ferritin levels lowest during this period, when as much as 8001r of the total fetal iron stores are acquired. "' These results suggest that iron replacement in preg- nancy could be delayed until the third trimester, at least in patients with a high first-trimester hemoglobin level. However, further study on maternal iron stores would be necessary before this practice could be recom- mended.

Whether these data on the maternal erythropoietin response could be extrapolated to patients whose first- trimester hemoglobin level was < 14.0 gnVdl remains to be determined. Nevertheless, in future studies on ma- ternal erythropoietin in pregnancy hematinic supple- mentation should be considered a variable in the study design and in the data analysis.

REFERENCES 1. Murphy J F, O'Riordan J, Newcombe RG, Coles EC, Pear-

son J F. Relation of haemoglobin levels in first and second trimesters to outcome of pregnancy. Lancet 1986; 1: 992-5.

2. Taylor Dj, Lind T. Red cell mass during and after preg- nancy. Br J Obstet Gynaecol 1979; 86: 364-70.

3. jelkmann W. Erythropoictin: structure, control of produc- tion, and function. Physiol Rev 1992; 72: 449-89.

4. Zivny J, Kobilkova J, Neuwirt J, Andrasova V. Regulation of erythropoiesis in fetus and mother during normal preg- nancy. Obstet Gynecol 1982; 60: 77-81.

5. Widness JA, Clemons G K, Garcia J F, Schwartz R. Plasma immunoreactive erythropoietin in normal women studied sequentially during and after pregnancy. Am J Oasrrr GYNECOL 1984; 149: 646-50.

6. Howells MR, Jones SE, Napier JAF, Saunders K, Cavill 1. Erythropoiesis in pregnancy. Br J Haematol 1986; 64: 595-9.

7. Teramo KA, Widness JA, Clemons GK, Voutilaien P, McKinlay S, Schwartz R. Amniotic: fluid erythropoietin correlates with umbilical plasma erythropoietin in normal and abnormal pregnancy. Obstet Gynecol 1987; 69: 710-5.

8. Halmesmaki E, Teramo KA, Widness JA, Clemons GK, Ylikorkala 0. Maternal alcohol abuse is associated with elevated fetal erythropoietin levels. Obstet Gynecol 1990; 76: 219-22.

9. Voutilainen PEJ, Widness JA, Clemons GK, Schwartz R, Teramo KA. Amniotic fluid erythropoietin predicts fetal

Volume 170, Number 3 Am j Obstet Gynecol

distress in Rh-immunized pregnancies. Am j OBsTET Gy- NECOL 1989; 160: 429-34.

10. Widness JA, Susa JB, Garcia JF, et al. Increased erythro- poiesis and elevated erythropoietin in infants born to diabetic mothers and in hyperinsulinernic rhesus fetuses. j Clin Invest 1981; 67: 637-42.

11. Lappin TRJ, Elder GE, Taylor T, McMullin MF, Bridges JM. Comparison of the mouse spleen cell assay and a radioimmunoassay for the measurement of serum eryth- ropoietin. Br j Haernatol 1988; 70: 117-20.

12. Thilaganathan B, Salvesen Dr, Abbas A, Ireland RM, Nicolaides KH. Fetal plasma erythropoietin concentration in red blood cell-isoimmunized pregnancies. Am j OwrEi GYNECOL 1992; 167: 1292-7.

13. Salvesen Dr, Brudenell JM, Snijders RJM, Ireland RM, Nicolaides KH. Fetal plasma erythropoietin in pregnan- cies complicated by matemal diabetes mellitus. Am j OB-

sT-rr GYNECOL 1993; 168: 88-94. 14. Snijders RJM, Abbas A, Melby 0, Ireland RM, Nicolaides

KH. Fetal plasma erythrOpoietin concentration in severe growth retardation. Am j ORsTET GYNECOL 1993; 168: 615-9.

15. Fenton V, Cavill I, Fisher J. Iron stores in pregnancy. Br Haematol 1977; 37: 145-9.

16. Romslo 1, Haram K, Sagen N, Augensen K. Iron require- ments in normal pregnancy as assessed by serum ferritin, serum transferrin saturation and erythrocytic protopor- phyrin determinations. Br j Obstet Gynaecol 1983; 90: 101-7.

Barton et al. 901

17. Beguin Y, Lipscei G, Otis R, Thoumsin H, Fillet G. Serum immunoreactive erythropoietin during pregnancy and in the early postpartum. Br J Haematol 1990; 76: 545-9.

18. Beguin Y, Lipscei G, Thoumsin H, Fillet G. Blunted erythropoietin production and decreased erythropoiesis in early pregnancy. Blood 1991; 78: 89-93.

19. MacDonald RG, MacDonald HN. Erythrocyte 2,3-diphos- phoglycerate and associated haematological parameters during the menstrual cycle and pregnancy. Br I Obstet Gyanecol 1977; 84: 427-33.

20. Singh A, Eckardt KU, Zimmerman A, et al. Increased plasma viscosity as a reason for inappropriate erythropoi- etin formation.

.1 Clin Invest 1993; 91: 251-6.

21. Faquin WC, Schneider'rj, Goldberg MA. Effect of inflam- matory cytokines on hypoxia-induced erythropoietin pro- duction. Blood 1992; 79: 1987-94.

22. Jepson J H, Endocrine control of maternal and fetal eryth- ropoiesis. Can Med Assn J 1968; 98: 844-7.

23. Widness JA, Clemons GK, Garcia JF, Oh W Schwartz R. Increased immunoreactive erythropoietin in cord serum after labor. A-m J OBs i Fi GYNECOL 1984; 148: 94-7.

24. Koury NIJ, Bonclurant MC, Graber SE, Sawyer Sr. Eryth- ropojetm messenger RNA levels in developing mice and transfer of 1251 -erythropoietin by the placenta. J Clin Invest 1988; 82: 154-9.

25. Saddi R, Schapira G. Iron-requirements during growth. In: Hallberg L, Hayweth GH, Vamotti A, eds. Iron defi- ciency. New York: Academic Press, 1970: 183-98.