1225

Nerve growth factor, brain-derived neurotrophic factor,and neurotrophin-3 levels in human amniotic fluid

Christine E. Marx, MD,a, c Brandon J. Vance, MD,a L. Fredrik Jarskog, MD,a, c Nancy C. Chescheir,MD,b and John H. Gilmore, MDa, c

Chapel Hill, North Carolina

OBJECTIVE: Neurotrophins are proteins that promote neuronal growth and differentiation. In this pilot studywe determined whether the neurotrophins nerve growth factor, brain-derived neurotrophic factor, and neu-rotrophin-3 were present in amniotic fluid specimens to begin to elucidate their developmental regulation. Wealso explored associations between neurotrophin levels and central nervous system abnormalities and expo-sure to infection.STUDY DESIGN: One hundred thirty-four amniotic fluid specimens were obtained from women undergoingamniocentesis at University of North Carolina Hospitals. Each specimen was assayed by enzyme-linked im-munosorbent assay for nerve growth factor, brain-derived neurotrophic factor, and neurotrophin 3. Associationsof maternal age, gestational age, and maternal ethnicity with neurotrophin levels were explored. Neurotrophinlevels in pregnancies in which there was enlargement of the fetal cerebral lateral ventricles or exposure to in-fection were compared with those in control pregnancies. Spearman correlational analyses and analyses ofcovariance were performed, with adjustment for gestational age.RESULTS: Nerve growth factor, brain-derived neurotrophic factor, and neurotrophin 3 were detected in allamniotic fluid specimens. Nerve growth factor increased with gestational age (P = .045). Brain-derived neu-rotrophic factor decreased with gestational age (P = .035). Patients with ventriculomegaly (with or withoutother central nervous system abnormalities) on ultrasonographic examination (n = 6) had significantly lowernerve growth factor levels than control subjects (P = .0046); patients with evidence of infection (n = 5) duringpregnancy had significantly lower nerve growth factor (P = .0037) and brain-derived neurotrophic factor (P =.0362) levels.CONCLUSIONS: Nerve growth factor, brain-derived neurotrophic factor, and neurotrophin 3 are detectable inamniotic fluid and vary with gestational age. Decreased nerve growth factor and brain-derived neurotrophicfactor levels in amniotic fluid may be a marker for the presence of central nervous system abnormalities, in-fectious insults in utero, or both. (Am J Obstet Gynecol 1999;181:1125-30.)

Key words: Neurotrophic factor, neurodevelopment, central nervous system abnormality, prenatalinfection

Neurotrophins are members of a family of proteinsthat include nerve growth factor (NGF), brain-derivedneurotrophic factor (BDNF), neurotrophin 3 (NT-3),and neurotrophin 4/5 (NT-4/5). Neurotrophins pro-mote the growth and differentiation of neurons, andtheir regulatory actions are important during both devel-opment and adulthood.1-3 Two types of neurotrophin re-

ceptors mediate biologic activity: high-affinity tyrosine ki-nase receptors (type A, B, and C) and the low-affinity p75receptor. NGF selectively binds tyrosine kinase receptorA, BDNF and NT-4/5 bind tyrosine kinase receptor B,and NT-3 binds tyrosine kinase receptor C, althoughsome nonselective binding also occurs.4 The low-affinityp75 receptor binds all the neurotrophins with equalaffinity, modulates high-affinity neurotrophin binding,and can promote apoptosis.5

Neurotrophins are critical to the normal developmentof the central nervous system (CNS), and altered neu-rotrophin levels may be associated with CNS develop-mental abnormalities. Haddad et al6 tested for the pres-ence of NGF in newborn cord blood from 61 neonates byusing a 2-site enzyme immunoassay and found that meanNGF levels in preterm infants were lower than levels interm infants. Two microcephalic children and one childwith congenital hydrocephalus in the study had lowerserum NGF levels than control subjects. Cerebrospinal

From the Departments of Psychiatrya and Obstetrics and Gynecologyb

and the UNC Mental Health and Neuroscience Clinical ResearchCenter, University of North Carolina School of Medicine.cSupported by grant MH-50356, The Theodore and Vada StanleyFoundation, center grant MH-33127, and training grant MH-19111.Presented at the One Hundred Fifty-fourth Meeting of the AmericanPsychiatric Association, Toronto, Ontario, Canada, June 1, 1998.Received for publication December 10, 1998; revised January 18, 1999;accepted May 21, 1999.Reprint requests: John H. Gilmore, MD, Department of Psychiatry, CB#7160, UNC School of Medicine, Chapel Hill, NC 27599-7160.Copyright © 1999 by Mosby, Inc.0002-9378/99 $8.00 + 0 6/1/100133

fluid levels of NGF have been found to decrease in nor-mal human development,7 and cerebrospinal fluid NT-3levels are increased in the setting of CNS infections andshunt placement for hydrocephalus.8 Kasaian and Neet,9

however, did not find clinical correlates of NGF levels in15 amniotic fluid samples using a bioassay and an en-zyme immunoassay; NGF levels were unrelated to chro-mosomal abnormalities, outcome, or gestational age atthe time of amniocentesis.

Neurotrophins are detected in tissues outside the ner-vous system, including the heart, spleen, and liver, as wellas the peripheral circulation.10, 11 Neurotrophins appearto play a role in autoimmune disorders. There is exten-sive evidence from animal studies that NGF exerts nu-merous effects on cells of the immune system.12 For ex-ample, NGF induces an increase in mast cell size andnumber,13 enhances lymphocyte proliferation,14 and in-creases T lymphocyte–dependent antibody synthesis.15

There is also clinical evidence implicating NGF in im-mune processes, such as rheumatoid arthritis,16 systemiclupus erythematosus,17 and multiple sclerosis.7

Little is known regarding the expression of neu-rotrophins during human fetal development. This studywas conducted to determine whether NGF, BDNF, or NT-3 is detectable in amniotic fluid specimens to begin toelucidate the developmental regulation of these neu-rotrophins. Associations of amniotic fluid levels of neu-rotrophic factors with maternal age, maternal ethnicity,and gestational age were explored. Given our researchinterest in mild ventriculomegaly18 and the effect of pre-natal exposure to infection on brain development,19 therelationship between neurotrophin levels and the pres-ence of ventriculomegaly and a history of infection wasexplored.

Methods

One hundred thirty-four samples of excess amnioticfluid specimens were obtained from 133 women under-

going amniocentesis at University of North CarolinaHospitals, as approved by the University of NorthCarolina School of Medicine Institutional Review Board.One twin pregnancy with separate amniotic sacs yielded2 amniotic fluid specimens. The mean maternal age was32.1 ± 6.6 years (±SD). The mean gestational age was 17.6± 2.2 weeks (±SD). The ethnicity of the subjects was as fol-lows: 83 white, 31 African American, 5 Asian–PacificIslander, 3 Hispanic, 1 Native American, and 10 un-known. Indications for amniocentesis included ≥1 of thefollowing: 77 with advanced maternal age (>35 years), 17with a history of birth defects, 10 with an ultrasonograph-ically determined abnormality, 48 with an abnormaltriple screen (α-fetoprotein, estriol, and human chori-onic gonadotropin), and 8 with other indications.

All subjects underwent ultrasonography before amnio-centesis and chromosomal karyotyping. Available med-ical records were reviewed for evidence of infection dur-ing the pregnancy before amniocentesis. Six subjectswith ventriculomegaly (with or without additional CNSabnormalities) on prenatal ultrasonography and 5 sub-jects with a treated infection before amniocentesis wereidentified. Two subjects had chromosomal abnormalities,1 with trisomy 21 and 1 with mosaicism (4/24 cells withan extra chromosome marker). Two subjects were inboth the ventricular abnormality and infection groups,and 1 subject with evidence of infection also had a defi-nite chromosomal abnormality (Table I).

Control subjects (n = 102) consisted of patients whohad no evidence of CNS or other abnormalities on ul-trasonography, no evidence of possible chromosomalabnormality on amniocentesis, and no evidence ofmedically treated infection or group B Streptococcuscarrier status during pregnancy. Twenty-two subjectswere excluded from the control group: 4 with norecorded gestational age at amniocentesis, 4 withchoroid plexus cysts, 1 with a normal ultrasonographicanalysis but hydrocephalus at elective termination 5

1226 Marx et al November 1999Am J Obstet Gynecol

Table I. Clinical description of subjects with either CNS abnormalities or prior exposure to infection or both

CNS Priorabnormality infection Clinical description

Yes No Encephalocele; spinal deformities; hydrocephalus (atrial width, 14 mm)Yes No Hydrocephalus (“massive dilatation”); cleft lip-palate; bilateral optic nerve and chorioretinal colobomas;

small for gestational ageYes No Mild ventriculomegaly (atrial width, 10 mm); single choroid plexus cyst; cleft palate; abnormal faciesYes No Borderline ventriculomegaly (atrial width, 11 mm),Yes Yes Borderline ventriculomegaly (atrial width, 8 mm); twin; discordant fetal growth; bacterial vaginosis and

trichomoniasis, treated with Flagyl; group B Streptococcus carrierYes Yes Encephalocele; small cerebellum; mild ventriculomegaly (atrial width, 11-12 mm); treated for syphilis;

positive fluorescent treponemal antibody at amniocentesisNo Yes Bacterial vaginosis, treated with FlagylNo Yes Varicella-zoster in early second trimester, treated with acyclovir; trichomoniasis; mosaicism, 4/24 cells with

extra chromosome markerNo Yes Bacterial vaginosis, trichomoniasis, treated with Flagyl

weeks after amniocentesis, 4 with other non-CNS ab-normalities on ultrasonography (omphalocele, mildpelviectasis, mild pelviectasis and possible cardiac axisdeviation, and short femur–thick nuchal fold), 4 thatwere group B Streptococcus carriers, and 5 with poten-tial chromosomal abnormalities (chromosome 9 vari-ant with father carrying a similar chromosome, a smallbalanced paracentric inversion of the X chromosome,a balanced karyotype reciprocal inherited from the fa-ther, an extra chromosome 9 in 1/50 cells [probableartifact], and mosaicism for trisomy 20 [probable ex-traembryonic tissue]).

Specimens were centrifuged for 10 minutes at 3000rpm, and amniotic fluid was frozen at –80°C until as-sayed. All amniotic fluid specimens were frozen within 4hours of collection. Two-site enzyme-linked immunosor-bent assay (ELISA) was performed according to the man-ufacturer’s directions (Promega ImmunoAssay System)for human BDNF, human NT-3, and mouse NGF (cross-reactive with human NGF). Preliminary assays were doneto determine optimal sample dilution for the standardcurves for each neurotrophin assayed. Assays were di-luted as follows: NGF, 1:50; BDNF, undiluted; and NT-3,undiluted, with 24 samples run at 1:4 dilution becausethey were above the standard curve. The limits of detec-tion for the standards of recombinant human NGF, re-combinant human BDNF, and recombinant human NT-3were 7.8 pg/mL, 7.8 pg/mL, and 4.7 pg/mL, respec-tively. The intra-assay coefficients of variance for all assayswere <5%. Samples were run in triplicate. Absorbancewas determined by a microplate reader (MolecularDevices) at a wavelength of 450 nm.

Recovery experiments were run to assess how mea-sured amounts of neurotrophin protein reflect actualamounts in amniotic fluid. Standard curves of knownneurotrophin concentrations in standard buffer werecompared with the same standard curve in pooled amni-otic fluid. Standard curves of 50:50 mixtures of the re-spective neurotrophins in standard buffer were also per-formed to assess how specific neurotrophin proteinsmight interfere with assays of other neurotrophins.

For statistical analysis, neurotrophin levels werenatural log transformed to normalize data distributions.Spearman correlational analyses were performed to ex-plore associations of maternal age and gestational agewith levels of NGF, BDNF, and NT-3 in the control group.Spearman correlational analyses were performed to ex-plore associations of levels of NGF, BDNF, and NT-3 witheach other in the control group.

Analyses of covariance with gestational age as a covari-ate were performed to compare NGF, BDNF, and NT-3levels in white and nonwhite subjects in the controlgroup. Analyses of covariance with gestational age as acovariate were also performed to compare control sub-

jects with subjects with evidence of ventriculomegaly(with or without additional CNS abnormalities) deter-mined by means of ultrasonography and control subjectswith patients with evidence of medically treated infectionbefore amniocentesis. NGF, BDNF, and NT-3 levels forcontrol subjects were compared with NGF, BDNF, andNT-3 levels for each of the above-mentioned groups.Each comparison was analyzed separately. A total of 6analyses of covariance were performed. Given the smallnumber of subjects with definite chromosomal abnor-malities (n = 2), an analysis of covariance was not per-formed for this group. Because of the exploratory natureof this study, corrections for multiple comparisons werenot performed.

Results

NGF, BDNF, and NT-3 were detected in all 134 amni-otic fluid specimens assayed. A modest but significantpositive correlation existed between gestational age andNGF in control subjects (rs = 0.1986; P = .045; n = 102),with NGF levels increasing with gestational age (Fig 1). Amodest but significant negative correlation existed be-tween gestational age and BDNF (rs = –0.2089; P = .035; n= 102), with BDNF levels decreasing with gestational age(Fig 2). NT-3 levels were not correlated with gestationalage (rs = –0.1560; P = .119; n = 101; Fig 3). There was nosignificant relationship between NGF, BDNF, or NT-3 lev-els and maternal age (P > .05 for each neurotrophin) norwas there a significant relationship between NGF, BDNF,and NT-3 levels and ethnicity (P > .05 for each neu-rotrophin).

Patients with ventriculomegaly determined by ultra-sonography (with or without additional CNS abnormali-ties, n = 6) had significantly lower NGF levels comparedwith control patients (F = 8.39; degrees of freedom, 2,

Volume 181, Number 5, Part 1 Marx et al 1227Am J Obstet Gynecol

Fig 1. Ranked natural log of NGF levels versus ranked gestationalage in amniotic fluid of control subjects. There was a significantpositive correlation between NGF level and gestational age atamniocentesis (rs = 0.1986; P = .045; n = 102).

105; P = .0046; Fig 4). Patients with medically treated in-fection before amniocentesis (n = 5) had significantlylower NGF levels (F = 8.81; degrees of freedom, 2, 104; P= .0037) and BDNF levels (F = 4.50; degrees of freedom,2, 104; P = .0362) compared with control patients (Fig 5).Group differences between NT-3 levels and the presenceof ventriculomegaly or infection during pregnancy werenot demonstrated.

Recovery studies revealed a concentration-dependentdecrease in recovery for BDNF and NT-3 in undilutedamniotic fluid compared with standard buffer. At thehighest concentrations on the respective standard curvesof BDNF (500 pg/mL) and NT-3 (300 pg/mL), there wasa recovery of approximately 76% for BDNF and 86% forNT-3 (data not shown). Levels of NGF in undilutedpooled amniotic fluid were higher than the standardcurve, and therefore it was not possible to perform acomparable curve for NGF. For each neurotrophin, addi-tion of equal amounts of other neurotrophins decreasedlevels measured by ELISA in a concentration-dependentmanner (Table II).

Comment

In this study NGF, BDNF, and NT-3 were detected in all134 amniotic fluid specimens assayed by ELISA. NGF lev-els increased modestly with gestational age, whereasBDNF levels decreased modestly with gestational age.

The NGF finding is consistent with the study by Haddadet al,6 which found that NGF levels in cord blood ofpreterm infants were significantly lower than NGF levelsin term infants. The source of these neurotrophic factorsin amniotic fluid is unknown, and further studies areneeded to determine whether they are of maternal, fetal,or placental origin.

Recovery studies indicate that actual concentrations ofBDNF and NT-3 (and probably NGF) in human amnioticfluid are higher than those measured by ELISA.Competition studies also indicate that neurotrophins caninterfere with antibody detection of related neu-rotrophins. This may be the result of formation of het-erodimers between these related neurotrophins.20

Relatively high NGF levels in human amniotic fluid couldtheoretically decrease observed levels of NT-3 and BDNF,which are present in much lower concentrations.However, in our sample there were no significant rela-tionships between NGF and either BDNF (rs = 0.0528; P =.59) or NT-3 (rs = –0.0527; P = .6) levels. In addition, thecorrelation between BDNF levels and gestational ageremained significant when we controlled for NGF level(rs = –0.2228; P = .026).

This study is limited by the small sample size and diag-nostic heterogeneity in the ventriculomegaly and infec-tion groups. The results provide preliminary evidencethat NGF levels are decreased in the setting of ventricu-

1228 Marx et al November 1999Am J Obstet Gynecol

Fig 3. Ranked natural log of NT-3 levels versus ranked gesta-tional age in amniotic fluid of control subjects. There was no sig-nificant correlation between NT-3 level and gestational age atamniocentesis (rs = –0.1560; P = .119; n = 101).

Fig 2. Ranked natural log of BDNF levels versus ranked gesta-tional age in amniotic fluid of control subjects. There was a sig-nificant negative correlation between BDNF level and gesta-tional age at amniocentesis (rs = –0.2089; P = .035; n = 102).

Table II. Percentage of added neurotrophic factor detected by ELISA in presence of equal amounts of other neu-rotrophic factors added at highest concentration on standard curve

BDNF and NGF and NT-3 andconcentration added concentration added concentration added

BDNF (500 pg/mL) — 67% (500 pg/mL) 65% (500 pg/mL)NGF (1000 pg/mL) 77% (1000 pg/mL) — 87% (1000 pg/mL)NT-3 (300 pg/mL) 96% (300 pg/mL) 84% (300 pg/mL) —

lomegaly (with or without other CNS abnormality) andthat NGF and BDNF levels are decreased in the setting ofprior exposure to infection. Because 2 subjects were inboth the ventriculomegaly and infection groups, how-ever, it is possible that the finding of decreased NGF lev-els pertains to only one of these conditions. Also of po-tential interest, both of the 2 patients with definitechromosomal abnormalities (one of whom was also inthe infection group) had low NGF levels (raw values of1062 and 2471 pg/mL, respectively).

The clinical significance of decreased amniotic fluidNGF and BDNF levels in these groups of patients remainsspeculative, but low NGF and BDNF levels may be mark-ers for the presence of either CNS abnormalities or in-fectious insults in utero or both. These 2 conditions maybe linked because in utero exposure to infection isknown to be associated with a variety of neurodevelop-mental abnormalities, including white matter lesions andcerebral palsy.21 Given the data linking neurotrophins toimmune processes,12 infection during pregnancy maydisturb neurotrophin expression in the developing fetusor placenta, leading to abnormal neurotrophin levels inamniotic fluid. There is evidence that systemic BDNFcrosses the blood-brain barrier22 and that systemically ad-ministered NGF reaches the brain in rats.23 Thereforelow levels of systemically generated neurotrophins couldtheoretically have an impact on the developing brain.Alternatively, if amniotic fluid levels of neurotrophins arederived from the fetal CNS, low levels may reflect abnor-mal fetal neurotrophin production. Amniotic fluid NGFand BDNF levels may have prenatal diagnostic utility.However, studies of larger and better characterized pop-ulations will be required to extend these findings and todetermine their clinical specificity. Given the critical in-

volvement of neurotrophins in normal nervous systemdevelopment, the characterization of neurotrophin lev-els in amniotic fluid merits further investigation as ameans of understanding potential mechanisms of devel-opmental brain disorders.

We thank Eric D. Jackson and Susan G. Silva, PhD, forassistance with statistics and Anne Fitzgibbon, Mary EllenDryden, and Kathleen W. Rao, PhD, for assistance in ob-taining amniotic fluid samples.

REFERENCES

1. Levi-Montalcini R, Skaper SD, Dal Toso R, Petrelli L, Leon A.Nerve growth factor: from neurotrophin to neurokine. TrendsNeurosci 1996;19:514-20.

2. Mehler MF, Kessler JA. Growth factor regulation of neuronaldevelopment. Dev Neurosci 1994;16:180-95.

3. Merrill JE, Jonakait GM. Interactions of the nervous and im-mune systems in development, normal brain homeostasis, anddisease. FASEB J 1995;9:611-8.

4. Thoenen H. Neurotrophins and neuronal plasticity. Science1995;270:593-8.

5. Birge RB, Fajardo JE, Hempstead BL. Signal transduction andneuronal cell death during development and disease. In:Lockshin RA, Zakeri Z, Tilly JL, editors. When cells die. NewYork: Wiley-Liss; 1998. p. 347-84.

6. Haddad J, Vilge V, Juif JG, Maitre M, Donato L, Messer J, et al.Beta-nerve growth factor levels in newborn cord sera. PediatrRes 1994;35:637-9.

7. Bracci-Laudiero L, Aloe L, Levi-Montalcini R, Buttinelli C,Schilter D, Gillessen S, et al. Multiple sclerosis patients expressincreased levels of β-nerve growth factor in cerebrospinal fluid.Neurosci Lett 1992;147:9-12.

8. Gilmore JH, Jarskog LF, Lindgren JC, McEvoy JP, Xiao H.Neurotrophin-3 levels in the cerebrospinal fluid of patients withschizophrenia or medical illness. Psychiatry Res 1997;73:109-13.

9. Kasaian MT, Neet KE. Nerve growth factor in human amnioticand cerebrospinal fluid. Biofactors 1989;2:99-104.

10. Yamamoto M, Sobue G, Yamamoto K, Terao S, Mitsuma T.Expression of mRNAs for neurotrophic factors (NGF, BDNF,NT-3, and GDNF) and their receptors (p75NGFR, TrkA, TrkB,

Volume 181, Number 5, Part 1 Marx et al 1229Am J Obstet Gynecol

Fig 5. Ranked natural log of NGF (triangles) and BDNF (circles)levels in amniotic fluid of patients with prior exposure to infec-tion compared with regression line for NGF (solid line) andBDNF (dotted line) in control subjects. Patients with medicallytreated infection before amniocentesis (n = 5) had significantlylower NGF levels (F = 8.81; degrees of freedom, 2, 104; P =.0037) and BDNF levels (F = 4.50; degrees of freedom, 2, 104; P= .0362) compared with control patients.

Fig 4. Ranked natural log NGF levels in amniotic fluid of pa-tients with ventriculomegaly (in presence or absence of addi-tional CNS abnormalities) compared with regression line forNGF concentrations in control subjects (solid line). Patients withventriculomegaly determined by means of ultrasonography(with or without additional CNS abnormalities) had signifi-cantly lower NGF levels than control patients (F = 8.39; degreesof freedom, 2, 105; P = .0046) at amniocentesis (n = 5).

and TrkC) in the adult human peripheral nervous system andnonneural tissues. Neurochem Res 1996;21:929-38.

11. Laurenzi MA, Beccari T, Stenke L, Sjolinder M, Stinchi S,Lindgren JA. Expression of mRNA encoding neurotrophins andneurotrophin receptors in human granulocytes and bone mar-row cells—enhanced neurotrophin-4 expression induced byLTB4. J Leukoc Biol 1998;64:228-34.

12. Aloe L, Skaper SD, Leon A, Levi-Montalcini R. Nerve growth fac-tor and autoimmune diseases. Autoimmunity 1994;19:141-50.

13. Aloe L, Levi Montalcini R. Mast cells increase in tissues ofneonatal rats injected with nerve growth factor. Brain Res1977;133:358-66.

14. Thorpe LW, Perez-Polo JR. The influence of nerve growth factoron the in vitro proliferative response of rat spleen lymphocytes.J Neurosci Res 1987;18:134-9.

15. Manning PT, Russell JH, Simmons B, Johnson EM Jr. Protectionfrom guanethidine-induced neuronal destruction by nervegrowth factor: effects of NGF on immune function. Brain Res1985;340:61-9.

16. Aloe L, Tuveri MA, Carcassi U, Levi-Montalcini R. Nerve growthfactor in the synovial fluid of patients with chronic arthritis.Arthritis Rheum 1992;35:351-5.

17. Bracci-Laudiero L, Aloe L, Levi-Montalcini R, Galeazzi M,Schilter D, Scully JL, et al. Increased levels of NGF in sera of

systemic lupus erythematosus patients. Neuroreport1993;4:563-5.

18. Gilmore JH, van Tol J, Kliewer MA, Silva SG, Cohen SB,Hertzberg BS, et al. Mild ventriculomegaly detected in uterowith ultrasound: clinical associations and implications for schiz-ophrenia. Schizophr Res 1998;33:133-40.

19. Gilmore JH, Jarskog LF. Exposure to infection and brain devel-opment: cytokines in the pathogenesis of schizophrenia.Schizophr Res 1997;24:365-7.

20. Arakawa T, Haniu M, Narhi LO, Miller JA, Talvenheimo J, PhiloJS, et al. Formation of heterodimers from three neurotrophinsnerve growth factor, neurotrophin-3, and brain-derived neu-rotrophic factor. J Biol Chem 1994;269:27833-9.

21. Yoon BH, Jun JK, Romero R, Park KH, Gomez R, Choi JH, et al.Amniotic fluid inflammatory cytokines (interleukin-6, inter-leukin-1β, and tumor necrosis factor α), neonatal brain whitematter lesions, and cerebral palsy. Am J Obstet Gynecol1997;177:19-26.

22. Pan W, Banks WA, Fasold MB, Bluth J, Kastin AJ. Transport ofbrain-derived neurotrophic factor across the blood-brain bar-rier. Neuropharmacology 1998;37:1553-61.

23. Loy R, Taglialatela G, Angelucci L, Heyer D, Perez-Polo R.Regional CNS uptake of blood-borne nerve growth factor. JNeurosci Res 1994;39:339-46.

1230 Marx et al November 1999Am J Obstet Gynecol