Increased Erythropoiesis Elevated Erythropoietin in ... · cause for this increase in erythropoiesis in IDM(4). Becausethe humoralcontrol oferythropoiesis is de-pendentonerythropoietin

Increased Erythropoiesis and Elevated

Erythropoietin in Infants Born to Diabetic Mothers

and in Hyperinsulinemic Rhesus Fetuses

JOHNA. WIDNESS, JoHN B. SUSA, JOSEPHF. GARCIA, DONB. SINGER, PRABHATSEHGAL,WILLIAM OH, ROBERTSCHWARTZ,and HERBERTC. SCHWARTZ,Sections ofReproductive and Developmental Medicine and of Pathology, Brown UniversityProgram in Medicine, Departments of Pediatrics at Rhode Island Hospital and ofPediatrics and Pathology, Womenand Infants Hospital of Rhode Island,Providence, Rhode Island 02902; Lawrence Berkeley Laboratory, University ofCalifornia, Berkeley, California 94720; NewEngland Regional Primate ResearchCenter Southboro, Massachusetts 01772; Department of Pediatrics, StanfordUniversity Medical Center, Stanford, California 94305

A B S T R AC T The pathogenesis of the increasederythrocytosis and extramedullary erythropoiesis ob-served in infants of diabetic mothers (IDM) has beenobscure. In the present studies, IDM were found tohave elevated umbilical plasma erythropoietin (Ep)concentrations by radioimmunoassay. 22 of 61 IDM(36%) had levels above the range of 28 nonasphyxiated,appropriately grown normal infants. In 16 controls and20 IDM, plasma Ep correlated directly with plasma in-sulin (P < 0.001, r = 0.73). To investigate this relation-ship further, a chronic rhesus model was studied withcontinuous fetal hyperinsulinemia for 21 d in utero inthe last third of pregnancy. In five experimental fetuses,plasma insulin levels averaged 4,210 ,u.U/ml at delivery,whereas plasma Ep was above the range of six controls.In addition, the experimental fetuses had elevatedreticulocyte counts in umbilical cord blood. Themechanism for the increased plasma Ep associated withhyperinsulinemia in the fetus is unexplained but maybe mediated by fetal hypoxia.

INTRODUCTION

Erythrocytosis, increased normoblastemia, and extra-medullary erythropoiesis have been observed duringthe first days after birth in infants of diabetic mothers

Address reprint requests to Dr. Widness at Rhode IslandHospital.

Received for publication 8 September 1980 and in revisedform 22 October 1980.

(IDM)' (1-3). Fetal hypoxia, which may be secondaryto placental insufficiency, has been suggested as thecause for this increase in erythropoiesis in IDM (4).

Because the humoral control of erythropoiesis is de-pendent on erythropoietin (Ep) production in the fetusas well as the adult, we studied Ep levels in umbilicalplasma from infants of normal and diabetic mothers. Inaddition, because fetal hyperinsulinemia is the majorfactor responsible for fetal macrosomia and cellular pro-liferation in IDM (5, 6), the relationship of fetal insu-lin and Ep was also investigated in a hyperinsulinemicfetal rhesus monkey (Macaca mulatta) model, in whichall the mothers were nondiabetic (7).

METHODS

Human studies. Written informed consent was obtainedfrom all study mothers. Two groups of subjects were studied,control infants and IDM. Controls were those singleton birthswhose birthweights were appropriate for gestational age(>10th percentile and <90th percentile) (8), whose APGARscores were .7 at 5 min of life, and whose mothers were notknown to have diabetes. The IDM group was likewise selectedfor being singleton and nonasphyxiated at birth, but not forappropriateness of birthweight. The mothers of these babieswere subgrouped according to the earlier modified classifica-tion of White (9). This classification was based on both severityand duration of diabetes. The mildest group, class A, includedthose womenwith glucose intolerance but without insulin re-quirements, whereas those in classes D and F/R included

'Abbreviations used in this paper: Ep, erythropoietin; HbA,, glycohemoglobin; IDM, infants of diabetic mothers;NRBC, nucleated erythrocyte count; RBC, erythrocyte count.

J. Clin. Invest. X The American Society for Clinical Investigation, Inc. * 0021-9738/81/03/0637/06 $1.00Volume 67 March 1981 637-642

637

those insulin-dependent womenwho manifested vascular dis-ease. Womenfrom classes B and C required insulin but had noevidence of vascular disease.

Mixed umbilical cord plasma or serum, which had beenrefrigerated as whole blood for <24 h, was separated, frozen(-20°C) and saved for Ep determination. A doubly clampedumbilical cord blood sample was obtained within 15 s of de-livery in a subgroup of 34 IDM and 5 controls. In these indi-viduals, heparinized umbilical venous whole blood wasanalyzed for hemoglobin concentration, volume of packederythrocytes, nucleated erythrocyte count (NRBC), reticulo-cyte count, and total erythrocyte count (RBC) by the usual lab-oratory procedures. In addition, for these IDM delivered bycesarean section, umbilical arterial and venous blood gaseswere analyzed using the Coming 164/2 Blood Gas Analyzer(Coming Glass Works, Coming, N. Y.).

Plasma glucose was measured using a glucose oxidasemethod (YSI 23A Yellow Springs Instrument Co., YellowSprings, Ohio) and insulin by a double antibody radioim-munoassay technique of Hales and Randle (10). Only offspringof White's diabetic classes A and A/B2 were evaluated forvenous plasma insulin levels, since plasma from these infantsdo not have matemal antiinsulin antibodies that cross the pla-centa and interfere with the insulin assay.

Mixed umbilical serum as well as venous plasma wasanalyzed for Ep using a specific double antibody radioim-munoassay technique (11). The purity of the labeled Ep was70,400 U/mg protein (kindly supplied by Dr. Eugene Gold-wasser, Department of Biochemistry, University of Chicago).Confirmation of the three highest Ep values was obtainedwith the polycythemic mouse bioassay for Ep (12).

A retrospective chart review carried out on the study sub-jects and their mothers included: matemal age, parity, dura-tion of pregnancy, route of delivery, smoking history, thirdtrimester glycohemoglobin (Hb Al) level nearest to delivery,and parameters of fetal distress during labor or delivery.Hb A, was determined either by a macrocolumn analyticalcation resin technique or by a high-performance liquidchromatography method as previously reported (13). For theinfants, these details included: birthweight, placental weight,assessment of gestational age, and APGARscores.

Rhesus primate studies. Five fetal rhesus monkeys weremade hyperinsulinemic in utero by an osmotically drivenminipump (Alzet, Alza Corp., Palo Alto, Calif.) delivering 19 Uof porcine insulin per day as previously reported (7). Toachieve this, a hysterotomy was performed during the finalthird of pregnancy in accurately dated rhesus monkeys anda minimpump implanted subcutaneously in the hind limb ofthe fetus. The minipump was left in situ and the preg-nancy allowed to continue. 21 d later (between 134 and 148 dgestation with term gestation being 165 d), the fetus was de-livered by cesarean section. Three of the six control preg-nancies were treated identically, whereas the remainder werenonoperated.

Before delivery of the fetus, free-flowing umbilical venousand arterial samples were obtained and analyzed in 7 of the 11cases for pH, pCO2, and PO2. Arterial and venous plasmasamples were also analyzed for glucose, insulin, glucagon,and Ep. Glucagon was measured by a radioimmunoassay pro-cedure using the Unger antibody (14). Anticoagulated venouswhole blood samples were analyzed for hemoglobin, volumeof packed erythrocytes, NRBC, reticulocyte count, and RBC.

Fetuses were exsanguinated moments before delivery andexamined grossly. At autopsy, total fetal weight was deter-

2 Class A/B refers to chemically diabetic women who re-ceived insuilin only during pregnancy.

mined by adding the blood volume removed to the weight ofthe fetal body. In addition, the placenta and fetal organs, viz.spleen, liver, and kidneys were also weighed. To assess thedegree of hepatic extramedullary erythropoiesis, a histologicpoint counting technique was used in evaluating light micro-scopic sections of the fetal livers with the results expressedas a percentage of the total volume of the liver (15).

Statistical analysis included unpaired t test, chi-square, lin-ear regression analysis, and the Wilcoxon rank-sum test, whereapplicable.

RESULTS

Human studies. In Table I are presented selectedclinical features of the control and IDM subjects.Statistical comparisons are indicated in the table forcomparisons of controls with all other groupings. TheIDM were heavier than the controls (P < 0.001 un-paired t test), although there was no difference in ges-tational age. Of the 61 IDM, 18 were large for gesta-tional age (>90th percentile) (8), three were small forgestational age (<10th percentile), and the remainderwere appropriate for gestational age. Maternal third tri-mester Hb Al, levels nearest to the time of deliverywere determined for 14 controls and for 56 of the dia-betic mothers. When individually compared with con-trols, all diabetic classes had elevated Hb Al. values, asdid the diabetic group as a whole.

Umbilical cord Ep values determined by radioimmu-noassay for the two study groups are portrayed in Fig. 1.Because of the extreme range of values (5-30,000mU/ml), the vertical axis for Ep is presented on a logscale for graphic purposes. Statistical comparison ofthe two groups using the Wilcoxon rank-sum test fornonparametric distributions is significant at P < 0.01.22 of the 61 IDM (36%) fell above the range for thecontrols (5-69 mU/ml). The three highest Ep values inthe IDM group, 30,000, 17,000, and 6,200 mU/ml, werecorroborated using the polycythemic mouse bioassay inwhich the Ep values were 20,000, 13,100, and 5,000mU/ml, respectively. These three infants were born ofmothers from White's classes B (two) and D and, there-fore, were not only from the groups with the mostsevere diabetes. Although significant differences in Epvalues were noted when controls were compared withIDM of White's classes A, A/B, C, and D (P < 0.02)(classes B and F/R were too small in number to reachsignificance), no differences of Ep levels were evidentbetween pairings of diabetic subgroups.

The two subsets of control and IDM subjects (classA and A/B only) who were studied with glucose, insu-lin, and Ep measurements at the time of delivery areportrayed in Table II. Although umbilical vein plasmaglucose concentrations were no different between thetwo groups, plasma insulin values were significantlyhigher (P < 0.01, rank-sum test). Similarly, these 20IDM individuals manifested higher Ep values com-pared with the 16 control infants (P < 0.01). Compari-

638 Widness, Susa, Garcia, Singer, Sehgal, Oh, Schwartz, and Schwartz

TABLE IClinical Features of Human Study Groups at Delivery

Subject group No. Gestational age Birth weight Maternal Hb A,

wk g %total Hb

Controls 28 37.0+3.0* 2,750+670 4.6±0.7 (14)IDM 61 37.1±2.1 3,360±720t 6.7± 1.4 (56)White's class

A 14 38.4± 1.4 3,430±5205 5.8± 1.1§ (13)A/Bl" 9 38.4±1.9 3,850±590t 6.1±0.9t (7)B 5 37.0±0.0 3,300+580 6.6±1.1tC 16 36.0±1.5 3,110+720 7.0±1.5t (15)D 12 36.5±2.2 3,400+860 7.3±1.2t (11)F/R 5 36.4±3.8 3,050+930 7.5 ±1.5§

* Mean±SD.t P < 0.001 relative to control group.5 P < 0.01 relative to control group.

1I Refers to chemically diabetic womenwho received insulin therapy only duringpregnancy.( ), number of subjects included, if different from that already indicated.

son of plasma Ep and insulin by linear regression analy-sis in the combined control and IDM groups resultedin a positive correlation (P < 0.001, r = 0.73).

No correlations were found between plasma Ep leveland maternal Hb A,, when analyzed by linear regres-sion. Similarly, the 28 control infants did not manifestany relationship of Ep level with gestational age. Thiswas also true for the IDM alone as well as for the com-bined groups. In the subset of 31 IDM who had um-bilical blood gases measured at delivery, positive cor-relations were observed for Ep concentration and APo2(PuvO2 - PuaO2) (P < 0.012, r = 0.42) as well as APco2(PuaCO2 - PuvCo2) (P < 0.003, r = 0.49). However,

Log mU/ml5r

4

3Ep

2

arithmetic meani and rangeo )control range

If

36%

64%

01 CONTROLS IDM

FIGURE 1 Human umbilical plasma or serum levels oferythropoietin in controls (n = 28) and IDM (n = 61). Arith-metic mean and range for both groups are indicated alongwith 22 individual IDM values that exceeded the controlrange. (Note log scale.) P < 0.01.

no relationship between umbilical arterial Po2 and Epwas observed. There was no correlation betweenplasma Ep and either RBC, NRBC, or reticulocytecounts.

In review of the data in mothers' and infants' charts,few differences could be ascertained in comparisons ofthe IDM with the controls. In addition, those 22 IDMwhose Ep values were above the control range (>69mU/ml) were compared with those 39 IDM within thenormal range. There were no differences for the ma-ternal features of age, parity, or smoking history.Cesarean section was, however, more common in theIDM group (P < 0.02 by chi-square). For the infants, nodifferences could be found for the placental weight or5-min APGARscores. However, the IDM with ele-vated Ep values also had significantly higher birth-weight to placental weight ratios compared with IDMwith normal Ep levels (P < 0.02). Furthermore, the

TABLE IIMean Umbilical Venous Plasma Glucose, Insulin, and

Ep Levels for Human Study Groups

Subjectgroup No. Glucose Insulin Ep

mg/dl 'UU/ml mU/ml

Controls 16 108 24 32(48-258) (10-77) (7-64)

IDM* 20 119 73t 961(39-224) (5-330) (14-520)

), range.* Class A and A/B only.t P < 0.01 by Wilcoxon rank-sum test.

Erythropoietin in Infants of Diabetic Mothers 639

three infants with the extraordinarily elevated Eplevels could not be segregated from any of the othersubjects on the basis of severity of maternal diabetes,diabetic control during pregnancy, or possible acute orchronic hypoxia. However, because of possible inter-ference of maternal insulin antibody in the umbilicalplasma, insulin determination was not possible forthese three individuals.

Fetal rhesus studies. Fig. 2 shows the results of thefetal rhesus monkey experiments with respect to theglucose, insulin, and Ep values measured in umbilicalarterial plasma at the time of delivery. Both groups offetuses were born of nondiabetic normal mothers. Thefetal artterial plasma insulin levels achieved in the fivehyperinsulinemic fetuses were two logs higher(2,900-5,300 ,uU/ml) than the controls (29-64 ,uU/ml)with no overlapping of values. Arterial plasma glucoselevels were lower (P < 0.03) in the hyperinsulinemicgroup, but whether this finding per se had any adversephysiologic effect is questionable, since simul-taneously obtained umbilical venous glucose valueswere not different between the groups (7). Similarly,the hyperinsulinemic fetuses manifest plasma Eplevels that were one order of magnitude higher than thecontrols and without overlap. The level for the six con-trol fetuses (8-45 mU/ml) fell within the range of thoseobtained from the control human subjects (8-69mU/ml). The plasma glucagon levels in the hyperinsu-linemic group were lower than the controls, also withno overlap of values.

Parameters of fetal erythropoiesis for both groups ofmonkeys are shown in Fig. 3. There was a tendencytowards increased hepatic extramedullary erythro-poiesis and absolute number of NRBCin the hyper-insulinemic group, but this did not reach the 5% level

mg/dl,. 6

80

60

to

20-

GI

Log,U/ml4.0.

3.0

o 2.0o

e- 1.0,AA

* 0;LUCOSE INSULIN

Log mU/ml3Or

2.0

1.0

*-A-

A

A

aao.

a0a

Ep

4

pr,/ml8'Or

600o

400[

8.0

6.0

4.0

A

A

No.xlCOm3rA

12.0[

8.0aaa

A

A20_

HEPATICEME

4.0[

60A

40

2C

KCORD

NUCLEATEDRBC

Nax19'3'

A10 A

Aa

)Oas

0 CORDARETICULO-CYTES

15

4

4

A

No xl0,Ymm3A

8 A

1.o _ _A

*

3.5

3.C0; a

° CORDRBC

NSa Control A Hyperinsulinemic

FIGURE 3 Rhesus fetus parameters of erythropoiesis: hepaticextramedullary erythropoiesis (EME) (P < 0.06), umbilicalvein NRBC(P < 0.08), reticulocytes (P < 0.007) and RBC.(See legend for Fig. 2.)

of significance in these small groups. Reticulocyteswere, however, increased in the experimental group,while RBCwas no different between the two groups.

As shown in Fig. 4, the body, placenta, and organweights of the hyperinsulinemic fetuses were heavierthan their control counterparts with the exception of thekidneys. The ratio of body weight to placenta weightwas no different between the two groups. There wereno differences in the free-flowing umbilical arterial orvenous blood gases between the groups with respect toPo2, Pco2, or pH.

DISCUSSION

Ep is the major hormonal regulator of erythropoiesisin the mammalian fetus and infant (16). In the presentstudy, Ep was found to be elevated in umbilical plasmaat delivery of infants of diabetic mothers. In one-third

50C

0a.06 40C

200[A

0GLUCAGON

30C

oLI

aControl A Hyperinsulinemic

FIGURE 2 Rhesus umbilical arterial glucose (P < 0.03), in-sulin (P < 0.001), Ep (P < 0.001), and glucagon (P < 0.001)levels in control and hyperinsulinemic fetuses. (Note logscale for insulin and Ep values.) Mean values for each groupare indicated by the horizontal line.

A

150 A

A A1.50A& A

125 * OA

A

A 1.00ia 100 O

0

i 0.50 B° 75,° O

BODYOPL ,LE0BODY PLACENTA SPLEEN

20r A

A

0

0A10j-

00<-

LIVER

4.0

zo

A

0 i

0

Ao A00

KIDNEYSNS

O3CONTROL AHYPERINSULINEMIC

FIGURE 4 Rhesus fetus somatic (P < 0.003), placenta(P < 0.001), and organ (spleen [P < 0.001], liver [P< 0.015], and kidneys) weights (in grams) at delivery. (Seelegend for Fig. 2.)

640 Widness, Susa, Garcia, Singer, Sehgal, Oh, Schwartz, and Schwartz

2

r..,

r

2

of the infants studied, plasma Ep at delivery exceededthe range of values observed in normal infants. Threevalues were among the highest previously reported byeither bioassay or the more sensitive radioimmunoas-say used here. There was no relationship observed be-tween plasma Ep and maternal insulin therapy or ma-ternal vascular disease. At least two important ques-tions arise from this observation: (a) what factorsdetermined this abnormality in utero and (b) what isthe biological significance of this elevation of Ep?

The fetus and newborn of the diabetic mother aresubject to the altered maternal substrate milieu that canaffect a variety of developing fetal systems (17). Mostdata have been interpreted to support the concept thatmaternal hyperglycemia results in fetal hyperglycemia,which stimulates fetal insulin production and release(5). Fetal hyperinsulinemia in the presence of adequatesubstrate produces macrosomia and selective organo-megaly (7). Although several of the diverse complica-tions of the infant of the diabetic mother have beenrelated to fetal hyperinsulinemia, the etiology of someproblems including polycythemia and extramedullaryerythropoiesis has remained obscure (17). The findingof elevated plasma Ep in IDM and the correlation ofumbilical plasma Ep with timbilical insulin in controland class A and A/B IDM may be interpreted as evi-dence supporting another role for fetal hyper-insulinemia.

In a small study of eight IDM, Finne could not de-tect any abnormality in umbilical plasma Ep using amouse bioassay system (18). In contrast, the presentstudy included several infants whose Ep levels, whenmeasured by both polycythemic mouse bioassay andradioimmunoassay, were extraordinarily high. It is un-clear why certain infants had such high levels. As yetthere is no evidence to support the possibility of an in-creased Ep production in contrast to a relative decreasein Ep receptors, which would permit the accumulationof exceptionally high circulating levels. The dif-ferences in magnitude of Ep levels observed in theIDM subjects compared with the hyperinsulinemicrhesus fetuses may be related to species differences,the limited number of rhesus fetuses studied, or un-known adverse uteroplacental factors present in the hu-man diabetic subjects.

Unlike the human IDM, the experimental fetal mon-key has been made hyperinsulinemic in utero within ametabolically normal mother. Under these conditions,it is unlikely that fetal hypoxia occurs secondary tomaternal/placental insufficiency. However, the in-creased erythropoiesis and elevated Ep levels ob-served in the hyperinsulinemic fetal monkeys mightresult from fetal hypoxia secondary to hyperinsulin-emia. Carson et al. (19) have recently observed thathyperinsulinemic fetal sheep have a progressive de-crease in arterial oxygen content over several days pos-

sibly due to increased fetal and/or placental oxygenconsumption or to a decrease in umbilical bloodflow (19).

Since the hyperinsulinemic monkey fetus is macro-somic and has hepatosplenomegaly (7), a hypoxic stim-ulus for Ep production may occur secondarily to the in-creased cellular proliferation. Alternatively, Ep pro-duction in the fetus might be directly stimuliated byinsulin. Golde et al. (20) have demonstrated that avariety of nonhematologic hormones, e.g. growth hor-mone, thyroxin, and prolactin may stimulate murineand human bone marrow erythroid progenitors in vitro(20). In addition, insulin alone was shown to have somestimulatory effect in vitro on Friend erythro-leukemiacells. Zanjani et al. (21) have recently demonstrated invivo that the increased erythropoiesis produced bytestosterone and thyroxine in the fetal sheep is Epmediated (21).

In the present studies, the hyperinsulinemic rhesusfetus with increased erythropoiesis and Ep levels hadlow plasma glucagon concentrations. Naets and Gans(22) have suggested from recent studies of rats and micethat increased glucagon may have an inhibitory role inthe control of erythropoiesis (22). Whether decreasedglucagon would result in enhanced erythropoiesis inthe presence of Ep has not yet been established.

Studies involving extirpation of the kidneys and/orliver in the sheep fetus haye localized the site of Epproduction to the liver (21, 23, 24). However, there islittle data from direct organ extraction of Ep in thefetus. In contrast, the adult can produce Ep from extra-renal sites under unusual conditions (25).

In the present study, the ,elevated Ep levels did notcorrelate with the RBCin either the IDM or the hyper-insulinemic fetal monkeys. This could be due to a mild,compensated hemolytic state and/or to ineffectiveerythropoiesis. The increased pulmonary excretion ofcarbon monoxide, an index of bilirubin production, ob-served by Stevenson et al. (26) in IDM may also re-flect ineffective erythropoiesis or hemolysis. Althoughthe hyperinsulinemic fetal monkeys had suggestiveincreases in erythropoiesis correlating well with in-creased Ep levels, measurements of erythrocyte pro-duction with ferrokinetics, and of erythrocyte survivalmay further elucidate this problem.

In summary, the rhesus and human umbilical plasmadata on Ep levels presented here are most consistentwith a secondary effect of fetal hyperinsulinemia onboth Ep levels and erythropoiesis. This effect may berelated to an increase in cellular proliferation andmetabolism resulting in increased oxygen con-sumption. The fetus, being in a relatively hypoxic statecompared to the adult may be more sensitive to changesin its internal environment. There is little direct evi-dence for fetal hypoxia in diabetic pregnancies in theabsence of advanced vascular disease. The present data

Erythropoietin in Infants of Diabetic Mothers 641I

are compatible with a major role for fetal insulin in theincreased erythropoiesis of IDM.

ACKNOWLEDGMENTSThe authors wish to thank Ms. K. S. Petzold for her experttechnical assistance and Ms. D. Perry for preparation of themanuscript. The Alza Corp. kindly donated Alzet 28-dminipumps.

This work was supported in part by grants from the NationalInstitutes of Health, HD-11343, AM-25603, and HL-22469;Division of Research Resources grant RR-00168 and RR-81;the Rhode Island Hospital Research Fund, the Division ofEnvironmental Research for the U. S. Department of Energy,and the Thrasher Foundation.

REFERENCES1. Naeye, R. L. 1965. Infants of diabetic mothers: a quanti-

tative, morphologic study. Pediatrics. 35: 980-988.2. Zetterstrom, R., B. Strindberg, and R. G. Amhold. 1958.

Hyperbilirubinemia and ABOhemolytic disease in new-born infants of diabetic mothers. Acta Paediatr. 47:238-250.

3. Gross, G. P., W. E. Hathaway, and H. R. McGaughey.1973. Hyperviscosity in the neonate. J. Pediatr. 82:1004-1012.

4. MacKay, R. B. 1957. Observations on the oxygenation ofthe foetus in normal and abnormal pregnancy. J. Obstet.Gynaecol. Br. Commonw. 64: 185-197.

5. Pedersen, J. 1977. The Pregnant Diabetic and Her New-born. The Williams & Wilkins Company, Baltimore.123- 134.

6. Hill, D. E. 1979. Effect of insulin on fetal growth. InThe Diabetic Pregnancy. A Perinatal Perspective. I. R.Merkatz and P. A. J. Adam, editors. Grune & Stratton,Inc., New York. 155-165.

7. Susa, J. B., K. L. McCormick, J. A. Widness, D. B. Singer,W. Oh, K. Adamsons, and R. Schwartz. 1979. Chronichyperinsulinemia in the fetal rhesus monkey. Diabetes.28: 1058-1063.

8. Babson, S. G., R. E. Behrman, and R. Lessel. 1970. Fetalgrowth. Liveborn birth weights for gestational age ofwhite middle class infants. Pediatrics. 45: 937-944.

9. White, P. 1974. Diabetes mellitus in pregnancy. Clin.Perinatal. 1: 331-347.

10. Hales, C. N., and P. J. Randle. 1963. Immunoassay of in-sulin with insulin antibody precipitate. Lancet. I: 200.

11. Garcia, J. F., J. Sherwood, and E. Goldwasser. 1979.Radioimmunoassay of erythropoietin. Blood Cells. 5:405-419.

12. Garcia, J. F., and J. C. Schooley. 1971. Dissociation oferythropoietin from erythropoietin-antierythropoietincomplexes. Proc. Soc. Exp. Biol. Med. 138: 213-215.

13. Widness, J. A., T. L. Rogler-Brown, K. L. McCormick,K. S. Petzold, J. B. Susa, H. C. Schwartz, and W. Schwartz.1980. Rapid fluctuations in glycohemoglobin (hemo-globin Ai,) related to acute changes in glucose. J. Lab.Clin. Med. 95: 386-394.

14. Faloona, G., and R. Unger. 1974. Glucagon. In Methodsof Hormone Radioimmunoassay. B. Jaffe and H. Behrman,editors. Academic Press, Inc., New York. 317-327.

15. Chalkley, H. W. 1943. Method for the quantitative mor-phologic analysis oftissues.J. Natl. Cancer Inst. 4:47-53.

16. Finne, P. H., and S. Halvorsen. 1972. Regulation oferythropoiesis in the fetus and newborn.Arch. Dis. Child.47: 683-687.

17. Cornblath, M., and R. Schwartz. 1976. Disorders of Car-bohydrate Metabolism in Infancy. In Major Problems inClinical Pediatrics. A. J. Schaffer, editor. W. B. SaundersCompany, Philadelphia. 115-154.

18. Finne, P. H. 1966. Erythropoietin levels in cord blood asan indicator of intrauterine hypox ia. Acta Paedia tr. Sca nd.55: 478-488.

19. Carson, B. S., A. F. Philipps, M. A. Simmons, F. C. Bat-taglia, and G. Meschia. 1980. Effects of a sustained insu-lin infusion upon glucose uptake and oxygenation of theovine fetus. Pediatr. Res. 14: 147-152.

20. Golde, D. W., N. Bersche, and M. J. Cline. 1979. Hor-monal effects on erythroid stem cells. In Proceedings ofthe Conference on Cellular and Molecular Regulation ofHemoglobin Switching. G. Stamatoyannopoulos, editor.Grune & Stratton, Inc., New York. 305-321.

21. Zanjani, E. D., and M. Banisadre. 1979. Hormonal stimu-lation of erythropoietin production and erythropoiesis inanephric sheep fetuses. J. Clin. Invest. 64: 1181-1187.

22. Naets, J. P., and M. Gans. 1980. Inhibitory effect of gluca-gon on erythropoiesis. Blood. 55: 997-1002.

23. Zanjani, E. D., J. Poster, H. Burlington, L. I. Mann, andL. R. Wasserman. 1977. Liver as the primary site oferythropoietin formation in the fetus. J. Lab. Clin. Med.89: 641-644.

24. Schooley, J. C., and L. J. Mahlmann. 1974. Extrarenalerythropoietin production by the liver in the weanling rat.Proc. Soc. Exp. Biol. Med. 145: 1081-1083.

25. Erslev, A. J., J. Caro, E. Kansu, and R. Silver. 1980.Renal and extrarenal erythropoetin production in anemicrats. Br. J.Haematol. 45: 65-72.

26. Stevenson, D. K., A. L. Bartoletti, C. R. Ostrander, and J.D. Johnson. 1979. Pulmonary excretion of carbonmonoxide in the human infant as an index of bilirubinproduction. II. Infants of diabetic mothers.J. Pediatr. 94:956-960.

642 Widness, Susa, Garcia, Singer, Sehgal, Oh, Schwartz, and Schwartz