Angiogenic and inflammatory biomarkers in midpregnancy and small-for-gestational-age outcomes in Tanzania

Research www.AJOG.org

OBSTETRICS

Angiogenic and inflammatory biomarkers in midpregnancyand small-for-gestational-age outcomes in TanzaniaAnne Marie Darling, MSc; Chloe R. McDonald, MSc; Andrea L. Conroy, PhD; Kyla T. Hayford, PhD;W. Conrad Liles, PhD; Molin Wang, PhD; Said Aboud, PhD; Willy S. Urassa, PhD; Kevin C. Kain, MD;Wafaie W. Fawzi, DrPH

OBJECTIVE: We sought to investigate the relationship between a RESULTS: A total of 60 participants (13.9%) gave birth to SGA infants.
panel of angiogenic and inflammatory biomarkers measured in mid-pregnancy and small-for-gestational-age (SGA) outcomes in sub-Saharan Africa.
STUDY DESIGN: Concentrations of 18 angiogenic and inflammatorybiomarkers were determined in 432 pregnant women in Dar esSalaam, Tanzania, who participated in a trial examining the effect ofmultivitamins on pregnancy outcomes. Infants falling below the 10thpercentile of birthweight for gestational age relative to the appliedgrowth standards were considered SGA. Multivariate binomialregression models with the log link function were used to determinethe relative risk of SGA associated with increasing quartiles of eachbiomarker. Restricted cubic splines were used to test for nonlinearity ofthese associations.

From the Departments of Global Health and Population (Ms Darling and Dr FaSchool of Public Health, Boston, MA; SAR Laboratories, Sandra Rotman Cen(Ms McDonald and Drs Conroy, Hayford, and Kain), Institute of Medical ScieInfectious Diseases, Department of Medicine (Dr Kain), University of TorontoWashington School of Medicine, Seattle, WA (Dr Liles); and Department of MSciences, Dar es Salaam, Tanzania (Drs Aboud and Urassa).

Received Feb. 4, 2014; revised April 18, 2014; accepted May 24, 2014.

This work was supported in part by the Global Alliance to Prevent Prematurityin Global Health: Preventing Preterm Birth Initiative grant number 12003 (K.C115160 and 13721 (K.C.K.); Canada Research Chair in Molecular Parasitolognumber 492028 (C.R.M.); and Eunice Kennedy Shriver National Institute of C

The authors report no conflict of interest.

Reprints: AnneMarie Darling, MSc, Department of Global Health and [email protected].

0002-9378/$36.00 � ª 2014 Mosby, Inc. All rights reserved. � http://dx.doi.org/10.1

Compared to those in the first quartile, the risk of SGA was reducedamong those in the fourth quartiles of vascular endothelial growthfactor-A (adjusted risk ratio [RR], 0.38; 95% confidence interval [CI],0.19e0.74), placental growth factor (adjusted RR, 0.28; 95% CI,0.12e0.61), soluble fms-like tyrosine kinase-1 (adjusted RR, 0.48;95% CI, 0.23e1.01), monocyte chemoattractant protein-1 (adjustedRR, 0.48; 95% CI, 0.25e0.92), and leptin (adjusted RR, 0.46; 95% CI,0.22e0.96).

CONCLUSION: Our findings provide evidence of altered angiogenic andinflammatory mediators, at midpregnancy, in women who went on todeliver SGA infants.

Key words: angiogenesis, inflammation, small for gestational age

Cite this article as: Darling AM, McDonald CR, Conroy AL, et al. Angiogenic and inflammatory biomarkers in midpregnancy and small-for-gestational-age outcomes inTanzania. Am J Obstet Gynecol 2014;211:x.ex-x.ex.

ntrauterine growth restriction
I (IUGR) is a major public healthproblem that affects 4-8% of pregnanciesin developed countries1 and an estimated27% of pregnancies in low- and middle-income countries.2 The consequences ofinadequate fetal growth can be lifelong.Growth-restricted infants have an in-creased risk of morbidity and mortalityduring the neonatal and postneonatal
period; an increased risk of developmentaldelay, short stature, neurodevelopmentalimpairment, and cerebral palsy duringchildhood; as well as an increased risk ofmyriad cardiometabolic disorders duringadulthood.1,3

A small number of interventionshave had modest success in reducing theoccurrence of small for gestational age(SGA),4 a commonly used surrogate

wzi), Epidemiology (DrsWtre for Global Health, Univnce (Ms McDonald and D, Toronto, Ontario, Canadicrobiology and Immunol

and Stillbirth and the Bill a.K.); Canadian Institutes oy grant number 77753134hild Health and Human D

on, Harvard School of Pub

016/j.ajog.2014.05.032

MONTH 2014 Am

outcome for IUGR. Nonetheless, con-siderable demand still exists for effectivepreventive measures. Development ofthese measures requires an improvedunderstanding of the pathogenesis ofIUGR and the ability to identify preg-nancies at risk in early and midgestation.The examination of biomarkers associ-ated with reduced fetal growth can assistin both endeavors.

ang and Fawzi), and Nutrition (Dr Fawzi), Harvardersity Health Network-Toronto General Hospitalr Kain), and Tropical Disease Unit, Division ofa; Department of Medicine, University ofogy, Muhimbili University of Health and Allied

nd Melinda Gates Foundation Grand Challengesf Health Research (CIHR) grant numbers MOP-5 (K.C.K.); CIHRDoctoral Research Award grantevelopment grant number R01 37701 (W.W.F.).

lic Health, 1635 Tremont St., Boston,MA 02120.

erican Journal of Obstetrics & Gynecology 1.e1

mailto:[email protected]

http://dx.doi.org/10.1016/j.ajog.2014.05.032

http://www.AJOG.org

http://www.AJOG.org

http://www.AJOG.org

http://www.AJOG.org

Research Obstetrics www.AJOG.org

Previous studies have linked a numberof biomarkers to IUGR5 but few havebeen conducted in resource-constrainedsettings where the burden of IUGR isgreatest.Much of this research has focusedon biomarkers of angiogenesis, thebranching and nonbranching remodelingof the placental vasculature6 that is acrucial process for adequate perfusion ofoxygen and nutrients to the fetus. Otherevidence suggests that IUGR may involvea proinflammatory cytokine bias.7 Forthis reason, evaluating the associationbetween inflammatory biomarkers andIUGR may be of particular relevance indeveloping countries, where commoninfections such as malaria can induce aproinflammatory microenvironment. Inthis study, we investigate the relationshipbetween a range of angiogenic and in-flammatory biomarkers during mid-pregnancy and IUGR as defined by SGA.

MATERIALS AND METHODS

Study site and participantsWe obtained data and specimens for theseanalyses from a randomized, double-blind, placebo-controlled trial of dailymultivitamin supplementation duringpregnancy. A detailed description of thetrial has been published elsewhere.8 Trialparticipants were human immunodefi-ciency virusenegative, between 12-27weeks of gestation, and planning to stayin Dar es Salaam for at least 1 year afterdelivery. At the time of enrollment, par-ticipants were randomly assigned toreceive a daily oral dose of a multivitamincontaining 20 mg of vitamin B1, 20 mg ofvitamin B2, 25 mg of vitamin B6, 100 mgof niacin, 50 mg of vitamin B12, 500 mg ofvitaminC, 30mg of vitaminE, and 0.8mgof folic acid or a placebo. All trial partic-ipants also received daily doses of 60mg ofelemental iron and 0.25 mg of folic acid.We limited the present analyses to primi-gravid women, since they are at higherrisk for fetal growth restriction thanmultigravid women.9 In addition, welimited these analyses to singleton births.Multiple gestations are associated withboth alterations in angiogenic markers10

and a higher risk of adverse pregnancyoutcomes11 and could therefore confoundthe results of the study. From among thesubset of primigravid trial participants

1.e2 American Journal of Obstetrics & Gynecology

with singleton pregnancies, known birthoutcomes, and stored baseline plasmasamples, we randomly selected 432 par-ticipants for these analyses.

Ethics statementThe institutional review boards at theMuhimbili University of Health and Al-lied Sciences in Dar es Salaam, Tanzania,and the Harvard School of Public Healthin Boston, MA, granted ethical approvalfor the study.

ExposureMaternal peripheral blood samples werecollected in EDTA vacutainer tubes atenrollment, plasma separated, andstored at �80�C prior to testing. For thepresent study, we selected 18 biomarkersassociated with inflammatory and an-giogenic pathways previously studiedin pregnancy, infection, or endothelialactivation pathways. These includedangiopoietin (Ang)-1, Ang-2, Ang-like3, vascular endothelial growth factor(VEGF)-A, soluble fms-like tyrosinekinase (sFlt)-1, soluble tumor necrosisfactor receptor 2, placental growth fac-tor (PGF), macrophage inflammatoryprotein-1b (MIP1b)/chemokine (CCmotif) ligand 4 (CCL4), monocyte che-moattractant protein (MCP)-1/chemo-kine (CCmotif) ligand 2 (CCL2), leptin,interleukin-1 b, interleukin-18 bindingprotein, soluble intercellular adhesionmolecule (sICAM)1, complement factorD, soluble endoglin, C-reactive protein,chitinase-3-like protein-1, and com-plement component C5a. All analysesutilized commercially available enzyme-linked immunosorbent assays (Duosets;R&D Systems, Minneapolis, MN). Toincrease sensitivity, samples were incu-bated for 2 hours at room temperature(18-28�C) for the analyses of C-reactiveprotein, complement component C5a,complement factor D, and VEGF-A,and overnight at 4�C for the analysesof all other biomarkers. Enzyme-linkedimmunosorbent assay analysis wasblinded to infant size for gestational ageat birth.

OutcomesAt enrollment, participants reportedthe date of their last menstrual period.

MONTH 2014

These dates were used to calculate ges-tational ages. Research midwives mea-sured birthweights of newborns to thenearest 10 g following delivery. Wedefined SGA births as those falling belowthe 10th percentile of birthweight forgestational age according to growthstandards from a US population12 as inthe parent study.8

Statistical analysisWe tested the distributions of eachbiomarker for normality using theShapiro-Wilks test. Because the distri-butions of each biomarker deviated fromnormality, we used the Wilcoxon ranksum test to nonparametrically examinewhether the levels of each biomarkerdiffered between SGA and appropriate-for-gestational-age (AGA) infants. Weestimated the relative association be-tween each biomarker and SGA bycategorizing biomarker values intoquartiles and using log binomial regres-sion to determine the risk ratios and 95%confidence intervals (CIs) for SGA forparticipants in each of the upper 3quartiles compared to those in the lowestquartile. In most cases, the log binomialmodels failed to converge and werereplaced with log Poisson models, whichprovide consistent but not fully efficientestimates of the risk ratio and its CI.13

Multivariate models also containedterms for covariates that predicted SGAat an alpha level of<0.2 in univariate logbinomial models. These variablesincluded literacy (yes/no), marital status(yes/no), gestational age at study entry(<20, 20-25,>25 weeks), and district ofrecruitment (Ilala/Temeke/Kinondoni).Although randomization should haveyielded a balanced distribution of allbiomarkers between study arms, werepeated the analysis adjusting for trialregimen assignment, since multivitaminsupplementation reduced the risk ofSGA in the parent trial.8

We nonparametrically examined thepossibility of nonlinear relationships be-tween continuous biomarker levels andSGA status using restricted cubic splines14

with 4 knots placed at the points corre-sponding to the 20th, 40th, 60th, and80th percentiles. For biomarkers thatdid not depart from linearity in relation

http://www.AJOG.org

TABLE 1Descriptive characteristics of study participantsCharacteristics n (%) or median (IQR)

Age, y 20.6 (18.6e22.6)

Gestational age at enrollment, wk 22 (19.4e24.6)

Body mass index, kg/m2

<18.5 5 (1.3)

18.5-24.9 252 (67.4)

25-29.9 98 (26.2)

�30.0 19 (5.1)

Mid upper arm circumference, cm

�22.5 47 (11.0)

22.6-25.0 172 (40.4)

25.1-29.0 165 (38.7)

>29.0 42 (9.9)

Education, y

0-4 30 (6.9)

5-7 289 (66.9)

8-11 97 (22.5)

�12 16 (3.7)

Literate 389 (90.5)

Marital status

Married 339 (79.2)

Divorced/single/widowed 89 (20.8)

Per day spending for food �TSh 500 87 (20.1)

District of recruitment

Ilala 287 (66.4)

Temeke 45 (10.4)

Kinondoni 100 (23.2)

Received multivitamin regimen 226 (52.3)

IQR, interquartile range; TSh, Tanzanian shilling.

Darling. Biomarkers of SGA in Tanzania. Am J Obstet Gynecol 2014.

www.AJOG.org Obstetrics Research

to SGA status, we then tested for thepresence of linear trends by assigningeach quartile the median value andmodeling this variable as a continuousvariable.

We considered assigned treatment armin the parent trial as a potential mod-ifier of the relationship between eachbiomarker and SGA births. To do so, wedichotomized each biomarker at themedian to create “high” and “low” cate-gories, computed cross-product terms bymultiplying the dichotomous biomarkervariables by the indicator variable fortreatment arm, and assessed the signi-ficance of this cross-product term usinga likelihood ratio test that comparedthe �2 log likelihoods of the modelscontaining the cross-product term tothe models without cross-product terms.All statistical analyses were performedusing software (SAS 9.2; SAS Institute,Cary, NC).

RESULTS

Participant characteristics are presentedin Table 1. At the time of enrollment intothe parent trial, participants had a me-dian age of 20.6 years (interquartilerange, 18.6e22.6) and a median gesta-tional age of 22 weeks (interquartilerange, 19.4e24.6). The majority of par-ticipants (67.4%) were of normal weightfor height, were able to read (90.5%),had completed <8 years of schooling(73.8%), and were married (79.2%).Approximately one fifth of participantsreported that their household spent�500 Tanzanian shillings (US $0.31) perperson per day on food. Of the 432participants, 60 (13.9%) delivered SGAinfants. Table 2 displays the medianvalues of each biomarker among SGAand AGA deliveries. Compared towomen who gave birth to AGA infants,women who gave birth to SGA infantshad notably lower median levels ofVEGF-A, sFlt-1, PGF, MCP-1/CCL2,and leptin, and notably higher levels ofsICAM1.

Since the biomarkers examined haveunique profiles with respect to ges-tational age, we examined biomarkerlevels according to the gestational age atwhich the plasma sample was collected.Median levels of VEGF-A and sFlt-1

were consistently lower in the SGAgroup compared to the AGA group at allgestational ages (Figure 1, A and C).Median levels of PGF, MCP-1/CCL2,and leptin were higher in the SGA groupcompared to the AGA group amongwomen between 12-16 weeks of ges-tation, but were lower among womenat other gestational ages (Figure 1, B,D, and E). Higher median levels ofsICAM1 were observed among the SGAgroup at all weeks of gestation (Figure 1,E and F).

MONTH 2014 Am

Table 3 shows the relative risks for anSGA delivery according to quartiles ofbiomarker values. Women with VEGF-A levels in the third and fourth quar-tiles had a reduced risk of giving birthto an SGA infant compared to womenin the first quartile. Spline analysisconfirmed a significant nonlinear rela-tionship (P ¼ .001) between VEGF-Aand SGA status that appeared L-shaped(Figure 2). Women with plasma PGFin the highest quartile had a reducedrisk of delivering an SGA infant (72%;


http://www.AJOG.org

TABLE 2Median biomarker values according to small-for-gestational-age status

Biomarker na

Appropriate forgestational age

na

Small forgestational age

P valuebMedian [IQR] Median [IQR]

Ang-1, ng/mL 362 19.63 [11.68e29.11] 60 18.13 [11.90e27.26] .47

Ang-2, ng/mL 362 4.64 [2.04e8.21] 60 4.05 [2.05e9.10] .93

Angptl3, ng/mL 363 124.74 [86.50e167.15] 60 108.05 [74.19e152.41] .16

VEGF-A, pg/mL 362 51.08 [7.81e363.54] 60 9.43 [7.81e85.38] .0006

sFlt-1, ng/mL 363 1.31 [0.55e4.11] 60 0.94 [0.23e2.38] .03

sTNFR2, ng/mL 363 5.68 [3.87e8.20] 60 5.62 [4.30e7.71] .86

PGF, ng/mL 343 1.61 [0.91e2.84] 59 1.18 [0.65e1.98] .01

MIP1b/CCL4,pg/mL

350 163.76 [81.14e346.06] 59 135.31 [63.64e295.72] .30

MCP-1/CCL2,pg/mL

353 48.14 [11.56e247.91] 60 20.91 [7.81e187.09] .07

Leptin, ng/mL 362 7.74 [4.76e12.88] 60 6.36 [4.04e10.42] .10

IL1b, pg/mL 362 19.96 [3.91e76.40] 60 19.99 [3.91e65.39] .40

IL-18 BP, ng/mL 363 13.57 [9.29e18.78] 60 13.33 [8.76e18.97] .83

sICAM1, ng/mL 363 157.68 [109.40e237.01] 60 184.53 [120.93e241.80] .03

Complementfactor D, ng/mL

356 478.00 [322.72e674.42] 57 505.07 [366.77e994.25] .34

sEng, ng/mL 362 22.41 [16.42e29.67] 60 22.90 [15.27e28.24] .71

CRP, mg/mL 350 1.86 [0.78e4.17] 57 2.05 [0.05e4.61] .81

CHI3L1, ng/mL 363 38.69 [22.52e66.53] 60 41.15 [25.87e69.48] .45

C5a, ng/mL 352 32.49 [16.80e64.74] 58 28.92 [16.89e52.38] .48

Ang, angiopoietin; Angptl3, angiopoietin-like 3; BP, binding protein; C5a, complement component C5a; CCL, chemokine (CCmotif) ligand; CHI3L1, chitinase-3-like protein-1; CRP, C-reactive protein; IL, interleukin; IQR, interquartile range; MCP,monocyte chemoattractant protein; MIP, macrophage inflammatory protein; PGF, placental growth factor; sEng, solubleendoglin; sFlt, soluble fms-like tyrosine kinase; sICAM, soluble intercellular adhesion molecule; sTNFR2, soluble tumor ne-crosis factor receptor 2; VEGF, vascular endothelial growth factor.

a Sample sizes were not equal for all biomarkers due to indeterminate assay results; b P values from Wilcoxon rank sum test.



95% CI, 59e88%) compared to those inthe lowest quartile. Similar to the resultsfor VEGF-A, spline analysis showed thatPGF had a significant nonlinear rela-tionship to SGA (P ¼ .008) that showedan L-shape (Figure 3).

We also observed associations be-tween leptin, sFlt-1, MCP-1/CCL2, andsICAM1 and SGA status. Compared tobiomarker levels in the first quartile,levels of plasma leptin, MCP-1, andsFlt-1 in the fourth quartile were asso-ciated with a reduced risk of SGA.Although the increased relative risksfor the upper quartiles of sICAM1 werenot statistically significant, spline anal-ysis suggested a nonlinear relationship


between sICAM1 and SGA (P ¼ .02).Treatment assignment in the parent trialalso did not significantly modify the as-sociations between any of the otherbiomarkers and SGA (all P values fromlikelihood ratio test comparing modelswith and without interaction terms> .05). None of the associations weobserved between biomarker levels andSGA materially changed when weincluded treatment assignment as a co-variate in the multivariate models (datanot shown).

COMMENT

We found that higher plasma levels of theproangiogenic markers VEGF-A, PGF,

MONTH 2014

and the antiangiogenic marker sFlt-1 atmidpregnancy were associated with areduced risk of giving birth to an SGAinfant. The L-shape of these associationsimplies that reaching critical thresholdsof these proteins may be necessary foradequate fetal growth. Based on our data,these thresholds appear to exist around100 pg/mL for VEGF-A and 4 ng/mL forPGF, although we cannot rule out thatthe fit of the spline models may beinfluenced by chance patterns in the data.Higher levels of the inflammatory pro-tein, MCP-1 and the adipocytokine hor-mone, leptin were also associated with areduced risk of delivering an SGA infant.Receipt of multivitamins did not modifythese associations, potentially due tolimited power for this analysis.

VEGF-A, PGF, and sFlt all belong tothe VEGF family of structurally similarproteins that are essential for placentalvascular development. VEGF-A isnecessary for vasculogenesis,6 the for-mation of the blood vessels from a pre-viously avascular area, which occursprimarily in the first trimester. In thesecond and third trimester, both VEGF-A and PGF regulate angiogenesis, theelongation, branching, and vascularremodeling critical to support rapid fetalgrowth in late pregnancy.15 The inverseassociation we observed between PGFand SGA is in agreement with previousstudies.16-23 Of the 2 previous studies ofmaternal VEGF and IUGR we couldidentify, one also found an inverse asso-ciation between VEGF and IUGR duringmidpregnancy,22 but the other found apositive association between VEGF andIUGR at delivery.24 Up-regulation ofVEGF due to placental hypoxia25 mayexplain these discrepant findings.

sFlt-1 is an alternatively splicedvariant of vascular endothelial growthreceptor-1 (VEGFR1) that has an in-hibitory effect on VEGF-A and PGFsignalling.26 During pregnancy, tropho-blast cells secrete large amounts of sol-uble receptors that modulate the levelsof bioavailable VEGF-A and PGF.27 Inthis study higher levels of sFlt-1 wereassociated with a reduced risk of SGA,in contrast to others showing eitherno association,17,28-30 or an inverseassociation between sFlt-1 and fetal

http://www.AJOG.org

FIGURE 1Median levels and interquartile ranges of selected biomarkers according to gestational age at sample collection

A, Vascular endothelial growth factor (VEGF)-A; B, placental growth factor (PGF); C, soluble fms-like tyrosine kinase (sFlt)-1; D, monocyte chemo-

attractant protein (MCP)-1/CCL2; E, leptin; F, soluble intercellular adhesion molecule (sICAM)1.AGA, appropriate for gestational age; SGA, small for gestational age.



growth.16,31-34 Åsvold et al23 observedthat low levels of sFlt-1 in early preg-nancy were associated with an increasedrisk of SGA, which they postulate reflectslow angiogenic activity overall duringthis period. We studied sFlt-1 levelssomewhat later in pregnancy, at whichtime decreases in sFlt-1 may reflect acompensatory mechanism preventingsFlt-1 from binding to free VEGF whenits levels are depressed. The dramaticallylower median levels of VEGF-A amongthe SGA group that were visible evenamong women who were only 12-16weeks’ gestation (Figure 1, A) may sup-port this explanation. Lower levels ofsFlt-1 in the SGA group may also resultfrom occult placental malaria infec-tion,35 a contributor to fetal growthrestriction.36

MCP-1/CCL2 is a chemokine thatattracts and activates monocytes andmacrophages to sites of inflammation.MCP-1/CCL2 has also been implicated inmediating angiogenesis in vitro andin vivo in a mouse model37,38 and in theregulation of trophoblast invasion into

the placental bed.39 Our finding thatwomen with the highest MCP-1 levelshad the lowest risk of a SGA birth ispartially in accordance with a previousobservation by Georgiou et al40 thatmothers of SGA infants had significantlylower concentrations of MCP-1 at 7-10weeks of gestation. While we alsoobserved lower median levels of MCP-1overall among mothers of SGA infants,our results contrastingly showed highermedianMCP-1 levels among those whosespecimens were tested between 12-16weeks (Figure 1). Briana et al41 foundlower postnatal concentrations of MCP-1among SGA infants and their mothers,although one cannot infer the temporal-ity of the association from their study.Further research into MCP-1 may help toclarify the role of this chemokine in thepathobiology of fetal growth restriction.The adipocytokine hormone leptin

takes part in the regulation of energybalance, metabolism, the immune res-ponse, and T-cell activation. Duringpregnancy, leptin is synthesized by theplacenta and contributes to placental

MONTH 2014 Am

growth, nutrient transfer, angiogenesis,and trophoblast invasion.42 As reviewedby Briana and Malamitsi-Puchner,42

previous studies have reported contra-dictory findings related to leptin and fetalgrowth restriction. Some studies havefound associations between increasedleptin levels and IUGR. A proposedexplanation for these findings is thatleptin secretion from trophoblast cellsincreases under hypoxic conditions,43

which can be characteristic of IUGR.Conversely, other studies have observedassociations between reduced plasmaleptin levels and IUGR comparable to theresults we report here. These latter studiesmay indicate that under some circum-stances, the placenta fails to increase lep-tin secretion in response to reducedplacental perfusion.42 Lower leptin in theSGA group, like lower sFlt-1, may alsosuggest a contribution of occult placentalmalaria infection35 to growth restrictionin this population.

Strengths of this study include its pro-spective design, comprehensive assess-ment of angiogenic and inflammatory


http://www.AJOG.org

TABLE 3Relative risks for small-for-gestational-age infants according to quartiles of biomarker levelsa

Variable Q1 Q2 Q3 Q4 P trendb

Ang-1, ng/mL 1.00 (Ref) 1.19 (0.63e2.24) 0.90 (0.46e1.77) 0.69 (0.33e1.44) .20

Ang-2, ng/mL 1.00 (Ref) 1.14 (0.60e2.17) 0.72 (0.34e1.53) 1.14 (0.58e2.24) .84

Angptl3, ng/mL 1.00 (Ref) 0.67 (0.37e1.23) 0.59 (0.30e1.17) 0.55 (0.28e1.07) .09

VEGF-A, pg/mL 1.00 (Ref) 0.56 (0.30e1.05) 0.24 (0.11e0.53) 0.38 (0.19e0.74)

sFlt-1, ng/mL 1.00 (Ref) 0.93 (0.52e1.68) 0.65 (0.34e1.23) 0.48 (0.23e1.01) .05

sTNFR2, ng/mL 1.00 (Ref) 1.32 (0.67e2.60) 1.32 (0.66e2.63) 1.09 (0.52e2.25) .98

PGF, ng/mL 1.00 (Ref) 0.72 (0.39e1.34) 0.64 (0.34e1.20) 0.28 (0.12e0.61)

MIP1b/CCL4, pg/mL 1.00 (Ref) 0.83 (0.43e1.60) 0.71 (0.36e1.40) 0.78 (0.41e1.50) .65

MCP-1/CCL2, pg/mL 1.00 (Ref) 0.44 (0.23e0.87) 0.53 (0.29e0.98) 0.48 (0.25e0.92) .30

Leptin, ng/mL 1.00 (Ref) 0.63 (0.33e1.21) 0.80 (0.44e1.46) 0.46 (0.22e0.96) .06

IL1b, pg/mL 1.00 (Ref) 0.42 (0.19e0.93) 0.87 (0.48e1.56) 0.75 (0.41e1.40) .94

IL-18 BP, ng/mL 1.00 (Ref) 0.87 (0.45e1.69) 0.89 (0.45e1.76) 1.08 (0.57e2.09) .69

sICAM1, ng/mL 1.00 (Ref) 1.33 (0.66e2.68) 1.34 (0.67e2.70) 1.52 (0.75e3.11)

Complement factor D, ng/mL 1.00 (Ref) 1.03 (0.51e2.09) 1.03 (0.51e2.10) 1.36 (0.70e2.66) .35

sEng, ng/mL 1.00 (Ref) 0.68 (0.32e1.44) 1.27 (0.70e2.30) 0.74 (0.37e1.50) .66

CRP, mg/mL 1.00 (Ref) 1.01 (0.51e2.03) 1.18 (0.59e2.37) 1.25 (0.62e2.52) .68

CHI3L1, ng/mL 1.00 (Ref) 1.46 (0.71e2.98) 1.39 (0.67e3.87) 1.79 (0.89e3.62) .52

C5a, ng/mL 1.00 (Ref) 1.06 (0.55e2.06) 1.17 (0.61e2.24) 0.72 (0.34e1.54) .32

Ang, angiopoietin; Angptl3, angiopoietin-like 3; BP, binding protein; C5a, complement component C5a; CCL, chemokine (CC motif) ligand; CHI3L1, chitinase-3-like protein-1; CRP, C-reactiveprotein; IL, interleukin;MCP, monocyte chemoattractant protein; MIP, macrophage inflammatory protein; PGF, placental growth factor; sEng, soluble endoglin; sFlt, soluble fms-like tyrosine kinase;sICAM, soluble intercellular adhesion molecule; sTNFR2, soluble tumor necrosis factor receptor 2; VEGF, vascular endothelial growth factor.

a Relative risks and 95% confidence intervals were estimated using binomial regression with log link function. When log-binomial model failed to converge, log Poisson models were used.Multivariate models adjusted for literacy (yes/no), marital status (married vs divorced/single/widowed), gestational age at study entry (<20, 20-25, >25 wk), and district of recruitment (Ilala,Temeke, Kinondoni); b P value for test for linear trend test calculated with median biomarker in each quartile modeled as continuous variable. P values for trend tests not reported for biomarkerswith observed nonlinear relations with small-for-gestational-age status (VEGF-A, P nonlinearity ¼ .001; PGF, P nonlinearity ¼ .008; sICAM1, P nonlinearity ¼ .02).



biomarkers, and ability to control for awide range of covariates as potentialconfounders. Although the examinationof numerous biomarkers increases thepotential for type I error throughmultipletesting, we maintained an alpha level of0.05 for each test. Applying a Bonferronicorrection to these results would havereduced the alpha level to 0.003 (0.05/18).The use of such a conservative alpha levelwould likely inflate the probability of typeII error.44 As the goal of this discoveryanalysis was to identify biomarkers forfurther investigation in relation to SGA,we did not wish to dismiss any associa-tions of interest. Furthermore,we selectedbiomarkers for inclusion in the studybased on a priori hypotheses about theirrole in the pathogenesis of fetal growthrestriction.


Our study may be limited by the use ofSGA as the outcome, which is definedas being below the 10th percentile ofbirthweight for gestational age, since it isnot entirely analogous to IUGR. Theformal definition of IUGR refers to thefailure of an infant to reach its geneticgrowth potential45 and IUGR is typicallyascertained through ultrasound technol-ogy unavailable in this setting. Therefore,we may have misclassified some consti-tutionally small but healthy infants asSGA. We expect any misclassification ofthis sort would have occurred non-differentially and led to conservativerelative risk estimates. Limiting the defi-nition of SGA to birthweights below the3rd percentile conversely attenuatedmostof our results, most likely due to lowpower. Some misclassification of last

MONTH 2014

menstrual periodebased gestational agesmay have occurred due to recall error, butany such errors would have also occurredindependently of exposure status andlikewise attenuated the results.

In summary, lower maternal levelsof VEGF-A, PGF, MCP-1, and leptinappear to precede SGA in this studycohort. Given the importance of eachof these factors in placental vasculardevelopment, our findings support thehypothesis that alterations in levels ofcritical mediators of angiogenesis atmidpregnancy contribute to the de-velopment of placental vascular insuf-ficiency, whereby the placenta cannotmeet the metabolic demands of thegrowing fetus resulting in SGA out-comes. Although these findings areexploratory, they compel further

http://www.AJOG.org

FIGURE 2Nonlinear relation (P [ .001)between midpregnancy VEGF-Alevels (pg/mL) and SGA

Adjusted for literacy (yes/no), marital status

(yes/no), gestational age at study entry (<20,

20-25, >25 weeks), and district of recruitment

(Ilala/Temeke/Kinondoni). Because VEGF-A

levels were nonnormally distributed, only those

between 5-95th percentile are displayed. Solid

line shows estimated odds ratio (OR) for mor-

tality for each increasing pg/mL of VEGF

compared to reference value of 74.07. Hori-

zontal line represents null OR of 1.0. Dashed

lines signify upper and lower bounds of 95%

confidence interval for OR.

SGA, small for gestational age; VEGF-A, vascular endothelialgrowth factor.

Darling. Biomarkers of SGA in Tanzania. Am J ObstetGynecol 2014.

FIGURE 3Nonlinear relation (P [ .008)between midpregnancy PGFlevels (ng/mL) and SGA

Adjusted for literacy (yes/no), marital status

(yes/no), gestational age at study entry (<20,

20-25, >25 weeks), and district of recruit-

ment (Ilala/Temeke/Kinondoni). Because placental

growth factor (PGF) levels were nonnormally

distributed, only those between 5-95th percentile

are displayed. Solid line shows estimated odds

ratio (OR) for mortality for each increasing pg/mL

of VEGF compared to reference value of 0.03.

Horizontal line represents null OR of 1.0. Dashed

lines signify upper and lower bounds of 95%

confidence interval for OR.

PGF, placental growth factor; SGA, small for gestational age;VEGF, vascular endothelial growth factor.

Darling. Biomarkers of SGA in Tanzania. Am J ObstetGynecol 2014.


investigation into the role of alteredangiogenesis in the pathobiology ofSGA as well as into the use of thesemarkers as potential early diagnostictools or targets for interventions to re-duce SGA. -

ACKNOWLEDGMENTS

We thank the mothers and children; the fieldteams, including nurses, midwives, supervisors,and laboratory staff; and the administrative staffwho made the study possible.

REFERENCES

1. Suhag A, Berghella V. Intrauterine growthrestriction (IUGR): etiology and diagnosis. CurrObstet Gynecol Rep 2013;2:102-11.2. Lee AC, Katz J, Blencowe H, et al. Nationaland regional estimates of term and preterm ba-bies born small for gestational age in 138 low-income and middle-income countries in 2010.Lancet Global Health 2013;1:e26-36.3. Longo S, Bollani L, Decembrino L, DiComite A, Angelini M, Stronati M. Short-termand long-term sequelae in intrauterine growthretardation (IUGR). J Matern Fetal Neonatal Med2013;26:222-5.

MONTH 2014 Am

4. Morris RK, Oliver EA, Malin G, Khan KS,Meads C. Effectiveness of interventions for theprevention of small-for-gestational age fetusesand perinatal mortality: a review of systematicreviews. Acta Obstet Gynecol Scand 2013;92:143-51.5. Conde-Agudelo A, Papageorghiou AT,Kennedy SH, Villar J. Novel biomarkers for pre-dicting intrauterine growth restriction: a sys-tematic review and meta-analysis. BJOG2013;120:681-94.6. Kaufmann P, Mayhew TM, Charnock-Jones DS. Aspects of human fetoplacentalvasculogenesis and angiogenesis, II: changesduring normal pregnancy. Placenta 2004;25:114-26.7. Raghupathy R, Al-Azemi M, Azizieh F. Intra-uterine growth restriction: cytokine profiles oftrophoblast antigen-stimulated maternal lym-phocytes. Clin Dev Immunol 2012;2012:734865.8. Fawzi WW, Msamanga GI, Urassa W, et al.Vitamins and perinatal outcomes among HIV-negative women in Tanzania. N Engl J Med2007;356:1423-31.9. Watson-JonesD,WeissHA,ChangaluchaJM,et al. Adverse birth outcomes in United Republicof Tanzaniaeimpact and prevention of maternalrisk factors. Bull World Health Organ 2007;85:9-18.10. Maynard SE, Moore Simas TA, Solitro MJ,et al. Circulating angiogenic factors in singletonvs multiple-gestation pregnancies. Am J ObstetGynecol 2008;198:200.e1-7.11. Berkowitz RL. Multiple gestations. In:Gabbe SG, Niebyl JR, Simpson JL, eds. Ob-stetrics: normal and problem pregnancies. NewYork: Churchill Livingstone; 1986:739-67.12. Brenner WE, Edelman DA, Hendricks CH.A standard of fetal growth for the United Statesof America. Am J Obstet Gynecol 1976;126:555-64.13. Spiegelman D, Hertzmark E. Easy SAS cal-culations for risk or prevalence ratios and dif-ferences. Am J Epidemiol 2005;162:199-200.14. Durrleman S, Simon R. Flexible regressionmodels with cubic splines. Stat Med 1989;8:551-61.15. Geva E, Ginzinger DG, Zaloudek CJ,Moore DH, Byrne A, Jaffe RB. Human placentalvascular development: vasculogenic andangiogenic (branching and nonbranching)transformation is regulated by vascular endo-thelial growth factor-A, angiopoietin-1, andangiopoietin-2. J Clin Endocrinol Metab2002;87:4213-24.16. Crispi F, Dominguez C, Llurba E, Martin-Gallan P, Cabero L, Gratacos E. Placentalangiogenic growth factors and uterine arteryDoppler findings for characterization of diffe-rent subsets in preeclampsia and in isolated in-trauterine growth restriction. Am J ObstetGynecol 2006;195:201-7.17. Poon LC, Zaragoza E, Akolekar R,Anagnostopoulos E, Nicolaides KH. Maternalserum placental growth factor (PlGF) in small forgestational age pregnancy at 11(þ0) to 13(þ6)


http://refhub.elsevier.com/S0002-9378(14)00500-6/sref15


















































































http://www.AJOG.org


weeks of gestation. Prenat Diagn 2008;28:1110-5.18. Taylor RN, Grimwood J, Taylor RS,McMaster MT, Fisher SJ, North RA. Longitudinalserum concentrations of placental growth fac-tor: evidence for abnormal placental angiogen-esis in pathologic pregnancies. Am J ObstetGynecol 2003;188:177-82.19. Romero R, Nien JK, Espinoza J, et al.A longitudinal study of angiogenic (placentalgrowth factor) and anti-angiogenic (solubleendoglin and soluble vascular endothelial growthfactor receptor-1) factors in normal pregnancyand patients destined to develop preeclampsiaand deliver a small for gestational age neonate.J Matern Fetal Neonatal Med 2008;21:9-23.20. Cowans NJ, Stamatopoulou A,Matwejew E, von Kaisenberg CS, Spencer K.First-trimester placental growth factor as amarker for hypertensive disorders and SGA.Prenat Diagn 2010;30:565-70.21. Benton SJ, Hu Y, Xie F, et al. Can placentalgrowth factor in maternal circulation identify fe-tuses with placental intrauterine growth restric-tion? Am J Obstet Gynecol 2012;206:163.e1-7.22. Bersinger NA, Ødegard RA. Serum levels ofmacrophage colony stimulating, vascularendothelial, and placenta growth factor in rela-tion to later clinical onset of pre-eclampsia and asmall-for gestational age birth. Am J ReprodImmunol 2005;54:77-83.23. Åsvold BO, Vatten LJ, Romundstad PR,Jenum PA, Karumanchi SA, Eskild A. Angio-genic factors in maternal circulation and the riskof severe fetal growth restriction. Am J Epidemiol2011;173:630-9.24. Borras D, Perales-Puchalt A, RuizSacedon N, Perales A. Angiogenic growth fac-tors in maternal and fetal serum in pregnanciescomplicated with intrauterine growth restriction.J Obstet Gynaecol 2014;34:218-20.25. Trollman R, Amann K, Schoof E, et al.Hypoxia activates the human placental vascularendothelial growth factor system in vitro andin vivo: up-regulation of vascular endothelialgrowth factor in clinically relevant hypoxicischemia in birth asphyxia. Am J Obstet Gynecol2003;188:517-23.26. Kendall RL, Wang G, Thomas KA. Identifi-cation of a natural soluble form of the vascular


endothelial growth factor receptor, FLT-1, andits heterodimerization with KDR. Biochem Bio-phys Res Commun 1996;226:324-8.27. BurtonGJ, Charnock-Jones DS, Jauniaux E.Regulation of vascular growth and function in thehuman placenta. Reproduction 2009;138:895-902.28. Shibata E, Rajakumar A, Powers RW, et al.Soluble fms-like tyrosine kinase 1 is increased inpreeclampsia but not in normotensive preg-nancies with small-for-gestational-age neo-nates: relationship to circulating placentalgrowth factor. J Clin Endocrinol Metab 2005;90:4895-903.29. Wathén KA, Tuutti E, Stenman UH, et al.Maternal serum-soluble vascular endothelialgrowth factor receptor-1 in early pregnancyending in preeclampsia or intrauterine growthretardation. J Clin Endocrinol Metab 2006;91:180-4.30. Rajakumar A, Jeyabalan A, Markovic N,Ness R, Gilmour C, Conrad KP. Placental HIF-1alpha, HIF-2 alpha, membrane and solubleVEGF receptor-1 proteins are not increased innormotensive pregnancies complicated by late-onset intrauterine growth restriction. Am JPhysiol Regul Integr Comp Physiol 2007;293:R766-74.31. Chaiworapongsa T, Espinoza J, Gotsch F,et al. The maternal plasma soluble vascularendothelial growth factor receptor-1 concen-tration is elevated in SGA and the magnitude ofthe increase relates to Doppler abnormalities inthe maternal and fetal circulation. J Matern FetalNeonatal Med 2008;21:25-40.32. Stepan H, Krämer T, Faber R. Maternalplasma concentrations of soluble endoglin inpregnancies with intrauterine growth restriction.J Clin Endocrinol Metab 2007;92:2831-4.33. Boutsikou T, Malamitsi-Puchner A,Economou E, Boutsikou M, Puchner KP,Hassiakos D. Soluble vascular endothelialgrowth factor receptor-1 in intrauterine growthrestricted fetuses and neonates. Early Hum Dev2006;82:235-9.34. Wallner W, Sengenberger R, Strick R, et al.Angiogenic growth factors in maternal and fetalserum in pregnancies complicated by intrauter-ine growth restriction. Clin Sci (Lond) 2007;112:51-7.

MONTH 2014

35. Conroy AL, Liles WC, Molyneux ME,Rogerson SJ, Kain KC. Performance charac-teristics of combinations of host biomarkers toidentify women with occult placental malaria: acase-control study from Malawi. PLoS One2011;6:e28540.36. Umbers AJ, Aitken EH, Rogerson SJ. Ma-laria in pregnancy: small babies, big problem.Trends Parasitol 2011;27:168-75.37. Kim MY, Byeon CW, Hong KH, Han KH,Jeong S. Inhibition of the angiogenesis by theMCP-1 (monocyte chemoattractant protein-1)binding peptide. FEBS Lett 2005;579:1597-601.38. Salcedo R, Ponce ML, Young HA, et al.Human endothelial cells express CCR2 andrespond to MCP-1: direct role of MCP-1 inangiogenesis and tumor progression. Blood2000;96:34-40.39. Naruse K, Innes BA, Bulmer JN,Robson SC, Searle RF, Lash GE. Secretionof cytokines by villous cytotrophoblast andextravillous trophoblast in the first trimesterof human pregnancy. J Reprod Immunol2010;86:148-50.40. Georgiou HM, Thio YS, Russell C, et al.Association between maternal serum cytokineprofiles at 7-10 weeks’ gestation and birth-weight in small for gestational age infants. Am JObstet Gynecol 2011;204:415.e1-12.41. Briana DD, Boutsikou M, Baka S, et al.Perinatal plasma monocyte chemotacticprotein-1 concentrations in intrauterine growthrestriction. Mediators Inflamm 2007;2007:65032.42. Briana DD, Malamitsi-Puchner A. Reviews:adipocytokines in normal and complicatedpregnancies. Reprod Sci 2009;16:921-37.43. Mise H, Sagawa N, Matsumoto T, et al.Augmented placental production of leptin inpreeclampsia: possible involvement of placentalhypoxia. J Clin Endocrinol Metab 1998;83:3225-9.44. Rothman KJ. No adjustments are neededfor multiple comparisons. Epidemiology 1990;1:43-6.45. Goldenberg RL, Cliver SP. Small for gesta-tional age and intrauterine growth restriction:definitions and standards. Clin Obstet Gynecol1997;40:704-14.























































































































































http://www.AJOG.org

Angiogenic and inflammatory biomarkers in midpregnancy and small-for-gestational-age outcomes in Tanzania

Documents