am. j. epidemiol. 2007 holzman 786 94

9
Original Contribution Histologic Chorioamnionitis and Preterm Delivery Claudia Holzman 1 , Ximin Lin 1 , Patricia Senagore 2 , and Hwan Chung 1 1 Department of Epidemiology, Michigan State University, East Lansing, MI. 2 Department of Physiology, Michigan State University, East Lansing, MI. Received for publication October 24, 2006; accepted for publication March 21, 2007. Inconsistent findings linking placental histologic chorioamnionitis (HCA) and preterm delivery may result from variations in HCA definition, population studied, and exclusion criteria. This analysis from the 1998–2004 Preg- nancy Outcomes and Community Health Study (five Michigan communities) includes the first 1,053 subcohort women (239 preterm, 814 term) with completed placental assessments. Multiple HCA definitions were constructed by 1) varying polymorphonuclear leukocytes/high-powered field thresholds and placenta components included and 2) using polymorphonuclear leukocyte characteristics to assign low/high maternal, fetal inflammation stage and grade. In African Americans, HCA was associated with preterm delivery before 35 weeks. The effect size was modest for polymorphonuclear leukocytes/high-powered field thresholds of greater than 10 and greater than 30 (odds ratios (ORs) ¼ 0.8 and 2.0); larger for greater than 100 (OR ¼ 3.2, 95% confidence interval (CI): 1.4, 7.1); strengthened after excluding medically indicated preterm deliveries (OR ¼ 4.9, 95% CI: 2.0, 11.8); and strongest for high maternal/high fetal HCA (OR ¼ 5.6, 95% CI: 1.4, 22.1). These latter HCA criteria also produced the largest effect size in Whites/others (OR ¼ 2.7, 95% CI: 0.3, 26.9). Among preterm deliveries before 35 weeks excluding those medically indicated, 12% of Whites/others and 55% of African Americans had high maternal HCA. The authors conclude that HCA definition, exclusion criteria, and race/ethnicity influence the HCA-preterm delivery association and that HCA contributes to preterm delivery-related ethnic disparity. chorioamnionitis; continental population groups; infection; inflammation; neutrophils; placenta; premature birth; term birth Abbreviations: CI, confidence interval; HCA, histologic chorioamnionitis; OR, odds ratio. Preterm birth is a major cause of infant mortality and morbidity (1–3) and is associated with long-term adverse sequelae (e.g., cerebral palsy and sight, hearing, and learn- ing problems) (4–7). The risk of preterm delivery is in- versely related to indices of social class in most studies (8–11), but the reasons for this remain unclear. In the United States, African Americans have had consistently higher pre- term delivery rates compared with those in Whites (12–14), even in studies attempting to adjust for social class inequal- ities through study design (15–17) or by statistical modeling with relevant covariates (18–20). Undoing these social class and race/ethnic disparities is an important goal and drives many lines of inquiry into preterm delivery. The link between infection and preterm delivery dispar- ities is supported through a series of related observations. Race/ethnic disparity is greatest for earlier preterm delivery (e.g., <35 weeks) (14), and these early deliveries are more often accompanied by premature rupture of membranes and/ or evidence of infection (21–26). Bacterial vaginosis, a poly- microbial overgrowth associated with preterm delivery (27), is more prevalent in disadvantaged (28, 29) and African- American women (30, 31). The placenta is a key tissue for understanding infection and/or inflammation pathways leading to preterm delivery, but the strength of association between histologic chorioam- nionitis (HCA) and preterm delivery has been inconsistent. Correspondence to Dr. Claudia B. Holzman, B601 West Fee Hall, East Lansing, MI 48824 (e-mail: [email protected]). 786 Am J Epidemiol 2007;166:786–794 American Journal of Epidemiology ª The Author 2007. Published by the Johns Hopkins Bloomberg School of Public Health. All rights reserved. For permissions, please e-mail: [email protected]. Vol. 166, No. 7 DOI: 10.1093/aje/kwm168 Advance Access publication July 11, 2007 by guest on November 20, 2014 http://aje.oxfordjournals.org/ Downloaded from

Upload: sri-agustina

Post on 09-Nov-2015

212 views

Category:

Documents


0 download

DESCRIPTION

Epidemiologi

TRANSCRIPT

  • Original Contribution

    Histologic Chorioamnionitis and Preterm Delivery

    Claudia Holzman1, Ximin Lin1, Patricia Senagore2, and Hwan Chung1

    1 Department of Epidemiology, Michigan State University, East Lansing, MI.2 Department of Physiology, Michigan State University, East Lansing, MI.

    Received for publication October 24, 2006; accepted for publication March 21, 2007.

    Inconsistent ndings linking placental histologic chorioamnionitis (HCA) and preterm delivery may result fromvariations in HCA denition, population studied, and exclusion criteria. This analysis from the 19982004 Preg-nancy Outcomes and Community Health Study (ve Michigan communities) includes the rst 1,053 subcohortwomen (239 preterm, 814 term) with completed placental assessments. Multiple HCA denitions were constructedby 1) varying polymorphonuclear leukocytes/high-powered eld thresholds and placenta components included and2) using polymorphonuclear leukocyte characteristics to assign low/high maternal, fetal inammation stage andgrade. In African Americans, HCA was associated with preterm delivery before 35 weeks. The effect size wasmodest for polymorphonuclear leukocytes/high-powered eld thresholds of greater than 10 and greater than 30(odds ratios (ORs) 0.8 and 2.0); larger for greater than 100 (OR 3.2, 95% condence interval (CI): 1.4, 7.1);strengthened after excluding medically indicated preterm deliveries (OR 4.9, 95% CI: 2.0, 11.8); and strongestfor high maternal/high fetal HCA (OR 5.6, 95% CI: 1.4, 22.1). These latter HCA criteria also produced the largesteffect size in Whites/others (OR 2.7, 95% CI: 0.3, 26.9). Among preterm deliveries before 35 weeks excludingthose medically indicated, 12% of Whites/others and 55% of African Americans had high maternal HCA. Theauthors conclude that HCA denition, exclusion criteria, and race/ethnicity inuence the HCA-preterm deliveryassociation and that HCA contributes to preterm delivery-related ethnic disparity.

    chorioamnionitis; continental population groups; infection; inammation; neutrophils; placenta; premature birth;term birth

    Abbreviations: CI, condence interval; HCA, histologic chorioamnionitis; OR, odds ratio.

    Preterm birth is a major cause of infant mortality andmorbidity (13) and is associated with long-term adversesequelae (e.g., cerebral palsy and sight, hearing, and learn-ing problems) (47). The risk of preterm delivery is in-versely related to indices of social class in most studies(811), but the reasons for this remain unclear. In the UnitedStates, African Americans have had consistently higher pre-term delivery rates compared with those in Whites (1214),even in studies attempting to adjust for social class inequal-ities through study design (1517) or by statistical modelingwith relevant covariates (1820). Undoing these social classand race/ethnic disparities is an important goal and drivesmany lines of inquiry into preterm delivery.

    The link between infection and preterm delivery dispar-ities is supported through a series of related observations.Race/ethnic disparity is greatest for earlier preterm delivery(e.g.,

  • Some studies have examined only preterm placentas (26, 32,33), and others have included preterm and term placentasassessed for clinical indications (34, 35) with the potentialfor selection bias. In studies comparing term and pretermplacentas, odds ratios for HCA have ranged from 0.8 to 5.9(23, 34, 3643). This wide range may be due, in part, todifferences in maternal characteristics, excluded subgroups(e.g., hypertensives, medically indicated deliveries), pretermcategory (e.g., early vs. all preterm), and HCA definition.

    Throughout the preterm delivery literature, HCA casedefinitions have differed markedly. Most studies can begrouped into one of two categories: those that have usedinflammatory cell thresholds (i.e., polymorphonuclear leu-kocytes/high-powered field) (24, 38, 4348) and those thathave incorporated a complex HCA grading and/or stagingapproach (4951). Within the first category, HCA criteriahave varied with respect to 1) the tissue components in-cluded, for example, placental disc, extraplacental mem-branes, and cord (33, 42, 47, 52), disc only (24, 25, 38, 40,53), or extraplacental membranes only (41, 43, 44, 48); 2)threshold, for example, any inflammation (25, 33, 34, 42,52, 53), greater than five polymorphonuclear leukocytes/high-powered field (24, 38, 47), and greater than 10 poly-morphonuclear leukocytes/high-powered field (43, 45, 46,48); and 3) number of high-powered fields that must exceedthe threshold (43, 46, 48). In the second category, the de-tailed systems of staging/grading have also varied. Typicallythese protocols document polymorphonuclear leukocyte lo-cation, density, and degeneration to estimate intensity andprogression (i.e., grade and stage) (4951) and, in some in-stances, to distinguish fetal from maternal inflammatory re-sponse (50). Differences across studies in HCA definition,population, and exclusion criteria make it difficult to disen-tangle the influences of specific variations on study results.

    In the community-based, prospective Pregnancy Out-comes and Community Health Study, delivered placentasfrom a subset of women (subcohort) were assessed by a studypathologist using a detailed descriptive rather than diagnos-tic approach. This protocol permitted construction of severalcommonly used HCA definitions within a single sample todetermine how these definitions affect associations betweenHCA and preterm delivery. In addition, it provided anopportunity to examine modification of the HCA-pretermdelivery association by factors such as preterm subgroup,exclusion criteria, and race/ethnicity.

    MATERIALS AND METHODS

    Study population

    The Pregnancy Outcomes and Community Health (orPOUCH) Study recruited pregnant women from August1998 to June 2004 from 52 clinics in five Michigan com-munities. Women were enrolled in gestational weeks 15through 27 (87 percent before week 25). Inclusion criteriawere singleton pregnancy with no known congenital anom-aly, maternal age of 15 or more years, maternal serum alpha-fetoprotein screen in gestational weeks 1522, no prepreg-nancy diabetes mellitus, and proficiency in English. Thestudy received approval from institutional review boards at

    Michigan State University, Michigan Department of Com-munity Health, and nine community hospitals. Women wereinvited to participate at the time of prenatal screening. Thestudy included all interested women with unexplained ma-ternal serum alpha-fetoprotein levels greater than two multi-ples of the median (7 percent of cohort) and a stratifiedsample (ethnic-specific strata) of women with normal ma-ternal serum alpha-fetoprotein levels. Of the 3,038 womenenrolled, 19 were lost to follow-up, leaving a cohort of3,019. At enrollment, cohort women were interviewed andhad biologic samples collected and stored.

    In a subcohort (n 1,371), assays were performed onstored biologic samples, prenatal and labor and delivery re-cords were abstracted, and delivered placentas were exam-ined by a study placental pathologist. The subcohortincluded all women who delivered preterm (2 multiples of the median) and term deliveries, and a sam-ple of women with normal maternal serum alpha-fetoproteinlevels and term deliveries (i.e., 72 percent of African-Americanand 23 percent of White/other women in this category). Thesampling scheme was designed to optimize available resour-ces and maximize statistical power for studying at-risk sub-groups (i.e., African Americans and women with highmaternal serum alpha-fetoprotein). Placentas were retrievedfor 1,213 (88 percent) subcohort women, and this analysisincluded the first 1,053 (239 preterm, 814 term) with com-pleted placenta assessments.

    Placenta examination protocol

    Placentas were formalin fixed and received gross exami-nation using standard protocols. Nine tissue samples wereembedded in paraffin blocks for microscopic assessment:two extraplacental membrane (membrane roll) samples;two umbilical cord samples (one proximal and one distalto disc insertion); and five full-thickness disc samples, oneat the cord insertion, one in central tissue that appearednormal on gross examination, two from central tissue, andone at the margin, these latter three representative of grosslyvisible abnormalities if present. Microscopic findings wererecorded in a descriptive, computer-based instrument adap-ted from a prototype by Dr. Caroline Salafia. The micro-scopic description included details, such as the highestnumber of cells/high-powered field for each leukocyte typein each placenta tissue compartment (e.g., intervillous space,subchorion, chorion and amnion of plate and extraplacentalmembranes, chorionic vessels, and umbilical cord). Whileperforming microscopic examinations, the study pathologistwas blinded to gestational age at delivery, all clinical data,and gross examination findings.

    Denitions of histologic chorioamnionitis

    HCA definitions were constructed to parallel those foundin the literature on HCA and preterm delivery. One set ofdefinitions incorporated variations in placental tissue com-ponents (i.e., cord, plate, and extraplacental membrane; ex-traplacental membrane only) and polymorphonuclearleukocyte threshold (i.e., at least one high-powered field

    Histologic Chorioamnionitis and Preterm Delivery 787

    Am J Epidemiol 2007;166:786794

    by guest on Novem

    ber 20, 2014http://aje.oxfordjournals.org/

    Dow

    nloaded from

  • with a polymorphonuclear leukocyte inflammatory pattern;>10 polymorphonuclear leukocytes; >30 polymorphonu-clear leukocytes; and >100 polymorphonuclear leuko-cytes). A staging and grading system was adapted froman approach described by Redline et al. (50). Maternalstages 13 were assigned as follows: stage 1at leastone high-powered field with greater than 10 polymorphonu-clear leukocytes in subchorionic fibrin but not in chorion oramnion; stage 2polymorphonuclear leukocyte inflamma-tory pattern in chorionic plate or extraplacental membranechorion plus or minus amnion but no polymorphonuclearleukocyte karyorrhexis or necrotizing inflammation; andstage 3polymorphonuclear leukocyte inflammatory pat-tern in chorionic plate and/or extraplacental membrane cho-rion and amnion, plus karyorrhexis and/or necrotizinginflammation. Fetal stages 13 were assigned as follows:stage 1at least one high-powered field with a polymorpho-nuclear leukocyte inflammatory pattern in fetal chorionicplate vessels but not in umbilical vessels; stage 2poly-morphonuclear leukocyte inflammatory pattern in umbilicalvessels, confined to the vessel wall; and stage 3polymor-phonuclear leukocyte inflammatory pattern in umbilicalvessels extending into Whartons jelly. Maternal grade andfetal grade were based on maximum polymorphonuclearleukocytes/high-powered field, that is, grade 1 (110polymorphonuclear leukocytes), grade 2 (1130 poly-morphonuclear leukocytes), grade 3 (31100 polymorpho-nuclear leukocytes), and grade 4 (>100 polymorphonuclearleukocytes).

    Pregnancy outcome

    Gestational age at delivery was calculated by use of thedate of the last menstrual period or gestational age estimatefrom ultrasound if the last menstrual period and ultrasoundestimate differed by more than 2 weeks. The last menstrualperiod-based estimate was similar (within 2 weeks) to theultrasound-based estimate at less than or equal to 20 weeksin 64 percent of the cohort and to the ultrasound-based esti-mate at 2125 weeks in an additional 12 percent of the co-hort. Ultrasound-based estimates were used for 18 percent ofthe cohort with absent or conflicting last menstrual period-based estimates. In the remaining 6 percent, only the lastmenstrual period-based estimates were available. Two ab-stractors, a physician and a study labor-and-delivery nurse,independently reviewed subcohort prenatal and labor anddelivery records to identify medically indicated preterm de-liveries. Disagreements were resolved through reexamina-tion of medical records. Medically indicated pretermdelivery was defined as delivery before 37 weeks that beginsby induction or cesarean section in the absence of spontane-ous labor (cervix dilated2 cm and regular contractions) orrupture of membrane as an initiating event.

    Analytical approach

    Prevalence of HCA and its association with preterm de-livery was calculated using SAS Survey Freq and SurveyLogistic procedures, respectively (54). Weights were ap-plied to reflect oversampling of high maternal serum alpha-

    fetoprotein into the cohort and the subcohort samplingscheme. Odds of HCA in term deliveries were comparedwith those in preterm deliveries at 3536 weeks and before35 weeks by use of various HCA definitions in race/ethnic-specific models. Analyses were repeated after removing allmedically indicated preterm deliveries, a group deliveredearly most often because of maternal vascular disease andrarely HCA. These models were constructed to examinevariation in effect size for the association between HCAand preterm delivery in relation to factors that are inconsis-tent across studies, that is, HCA definition, race/ethnicity,gestational weeks of the preterm delivery, and inclusion/exclusion of medically indicated preterm delivery.

    For both maternal and fetal inflammatory responses,stages 13 and grades 14 were dichotomized into low(i.e., one or two) and high (i.e., three or more), and a five-level HCA severity variable was created: no HCA (refer-ent); low grade/low stage; low grade/high stage; high grade/low stage; and high grade/high stage. The two five-levelvariables representing maternal and fetal inflammatory re-sponses were each incorporated into separate models to as-sess their association with preterm delivery. In a final model,maternal and fetal inflammatory responses were combinedinto a single HCA severity variable. The cutpoints for di-chotomizing low and high responses were based on associ-ations with preterm delivery in the maternal-specific andfetal-specific models. The maternal high inflammatoryresponse was stage 3 or higher and any grade, and lowwas stage 1 or 2 and any grade. The fetal high inflam-matory response was stage and grade 3 or higher, andlow was stage or grades 12. The new five-level variablefor maternal/fetal inflammation was as follows: no HCA(referent); low maternal/low fetal; low maternal/high fetal;high maternal/low fetal; and high maternal/high fetal.

    RESULTS

    The maternal characteristics of this subcohort sample arepresented in table 1 without sampling weights; therefore, theoverall prevalences and differences between term and pre-term in this table do not reflect those in the entire cohort. Inthis subcohort sample, 54 percent were insured by Medicaid,38 percent were African Americans, and a small percentagewas other ethnic minorities. In the following analyses, sam-pling weights were used, and the other ethnic group wascombined with Whites because HCA prevalence in theother group more closely paralleled that in Whites thanthat in African Americans.

    HCA prevalence in term deliveries ranged from 85 percentdown to 7 percent and in preterm delivery from 63 percentdown to 4 percent, depending on the polymorphonuclearleukocytes/high-powered field threshold used to defineHCA (figure 1). Overall, the prevalence dramatically de-creased as the threshold increased, and for most thresholds,the HCA prevalence was lower when based on examinationof extraplacental membranes only versus examination ofall placental tissue compartments (figure 1). In term and pre-term deliveries, the HCA prevalence was higher in AfricanAmericans compared with that in Whites/others, and theserace/ethnic differences increased as the polymorphonuclear

    788 Holzman et al.

    Am J Epidemiol 2007;166:786794

    by guest on Novem

    ber 20, 2014http://aje.oxfordjournals.org/

    Dow

    nloaded from

  • leukocytes/high-powered field threshold increased (figure 1).When cord, plate, and extraplacental membrane were exam-ined and a polymorphonuclear leukocytes/high-poweredfield threshold of greater than 30 was used to define HCA,the prevalence of HCA in African Americans versus Whites/others was 37 percent versus 25 percent in term deliveries, 23percent versus 14 percent in preterm deliveries at 3536weeks, and 46 percent versus 6 percent in preterm deliveriesbefore 35 weeks.

    In race/ethnic-specific logistic regression models, HCAdefined by any of the polymorphonuclear leukocytes/high-powered field thresholds was not associated with pretermdelivery at 3536 weeks (table 2). HCA was significantlyrelated to preterm delivery before 35 weeks but only in

    African Americans, and only when the polymorphonuclearleukocytes/high-powered field threshold for HCA wasgreater than 100 (odds ratio (OR) 3.2, 95 percent confi-dence interval (CI): 1.4, 7.1). After removal of the medicallyindicated preterm deliveries, this association was furtherstrengthened (OR 4.9, 95 percent CI: 2.0, 11.8), and evenHCA defined by lower thresholds (e.g., >30 polymorpho-nuclear leukocytes/high-powered field) appeared related topreterm delivery before 35 weeks in African Americans(table 2).

    By use of a staging and grading system for maternal andfetal inflammatory response (HCA) and excluding medi-cally indicated preterm deliveries, the association betweenhigh maternal/high fetal inflammatory response and pretermdelivery before 37 weeks was weak and statistically non-significant among Whites/others (OR 1.9, 95 percent CI:0.4, 8.1) and, again, more pronounced in African Americans(OR 3.2, 95 percent CI: 1.2, 8.3) (table 3). The sameanalytical models were repeated but this time comparingHCA in term with that in preterm deliveries before 35 weeks(table 4). The odds ratio estimate for high maternal/highfetal inflammatory response increased slightly in Whites/others (OR 2.7, 95 percent CI: 0.3, 26.9), but the confi-dence interval was wider because of the smaller sample sizeafter excluding 35- to 36-week preterm deliveries. In Afri-can Americans, preterm delivery before 35 weeks was as-sociated with high maternal/low fetal response (OR 3.6,95 percent CI: 1.0, 13.4) and with high maternal/high fetalresponse (OR 5.6, 95 percent CI: 1.4, 22.1).

    DISCUSSION

    In this study, the HCA-preterm delivery relation wasstrongly influenced by HCA definition, inclusion/exclusionof preterm delivery at 3536 weeks and medically indicatedpreterm delivery, and the race/ethnic group studied. Surpris-ingly, there has been little discussion regarding variations inHCA definition across studies. Within studies, kappas forinterrater reliability have ranged from 0.15 to 0.83, with lessagreement for mild HCA or severity of HCA and greateragreement for the presence of any HCA (5557). However,in most reliability studies, the HCA definition and protocolwere standardized. Across studies, two approaches for as-sessing HCA predominate, one using polymorphonuclearleukocytes/high-powered field thresholds and the other us-ing a complex staging and/or grading scheme, with eachapproach having its own set of varied definitions. Blanc(49) first proposed three stages based on maternal polymor-phonuclear leukocyte movement from subchorion to chorionto amnion overlying the placental plate. Later classificationschemes, for example, those reported by Salafia et al. (51)and Redline et al. (50), incorporated location, density, andcharacteristics of polymorphonuclear leukocyte infiltrates.

    Applying the polymorphonuclear leukocytes/high-powered field approach with low thresholds (i.e., >0 or>10 polymorphonuclear leukocytes/high-powered field) inthe cord, extraplacental membranes, plate, or subchorion, weobserved that 5476 percent of White/other term placentasand 6485 percent of African-American term placentas met

    TABLE 1. Maternal characteristics of the subcohort with

    completed placenta assessments (n 1,053), PregnancyOutcomes and Community Health Study,* 19982004

    Gestational week at delivery

  • these HCA criteria. Other studies examining term placentashave reported HCA prevalences ranging from 4 percent to79 percent (23, 34, 3643, 51). These findings underscorethe need to compare inflammatory cells/patterns in pretermplacentas with those in term placentas to arrive at meaning-ful relations between HCA and preterm delivery. Previousstudies have often focused solely on HCA in preterm pla-centas (21, 26, 33, 45, 52), and many studies have used a lowpolymorphonuclear leukocytes/high-powered field thresh-old to define HCA (24, 38, 47), which would have an atten-uating effect on the HCA-preterm delivery association. Thehigh prevalence of low intensity HCA in term placentas maymark the beginning of a maternal response to microbialascent during labor. In addition, infection may precipitatedelivery at any point along the gestational continuum, anda less fulminating inflammation may adequately triggerdelivery in late gestational tissues already primed. We ob-served high stage/high grade fetal inflammation in 3 percentof White/other and 10 percent of African-American termplacentas, evidence that this problem is not confined topreterm delivery and that ethnic disparities persist at term.

    For each polymorphonuclear leukocytes/high-poweredfield threshold in our analyses, assessment of cord, plate,

    and extraplacental membranes tended to produce a higherHCA prevalence than that found in extraplacental mem-branes alone. One explanation could be empirical; the moretissue locations sampled, the greater the likelihood of findinga locus with a high number of polymorphonuclear leuko-cytes/high-powered field. However, these results highlightthe challenges of comparing HCA results across studies thatinclude only extraplacental membranes (23, 41, 43, 48), onlydisc (24, 25, 38, 40, 53), or both (26, 34, 36, 46).

    In our HCA stage/grade analyses that excluded medicallyindicated preterm deliveries, all but one placenta deliveredbefore 35 weeks with high fetal response also had highmaternal response. However, among term placentas withhigh fetal response, a substantial proportion (10 of 16White/others, 14 of 33 African American) had a low mater-nal response. This may reflect the ability of a more maturefetus to mount a robust immune response earlier in the pro-cess of exposure to infection and possibly discordancein maternal and fetal immune responsiveness.

    We also observed a higher prevalence of HCA in AfricanAmericans compared with that in Whites/others withinterm, preterm delivery at 3536 weeks, and preterm deliverybefore 35 weeks. In deliveries before 35 weeks that were not

    FIGURE 1. Prevalence of histologic chorioamnionitis (HCA) by race/ethnicity and gestational week at delivery, Pregnancy Outcomes andCommunity Health (POUCH) Study, 19982004. AD compare different criteria used to dene HCA: A and C, polymorphonuclear leukocytesevaluated in plate, cord, and extraplacental membranes in Whites/others and African Americans, respectively; B and D, polymorphonuclearleukocytes evaluated in extraplacental membranes only in Whites/others and African Americans, respectively. Prevalences are weighted toaccount for the subcohort sampling design. HCA was considered present if the highest number of polymorphonuclear leukocytes/high-poweredeld (PMNL/hpf) observed in any tissue component was equal to or greater than the dened cutoff (threshold). The lowest threshold was greaterthan zero PMNL/hpf and an inammatory pattern.

    790 Holzman et al.

    Am J Epidemiol 2007;166:786794

    by guest on Novem

    ber 20, 2014http://aje.oxfordjournals.org/

    Dow

    nloaded from

  • medically indicated, 12 percent of White/other placentasand 55 percent of African-American placentas showed evi-dence of a high maternal inflammatory response. The largestHCA odds ratio for very preterm delivery was 5.6 in AfricanAmericans but only 2.7 and not statistically significant inWhites/others. These results are in agreement with many(21, 5860), but not all (40), previous studies showing thatHCA is a key element in race/ethnic disparity in early pre-term delivery. The findings also suggest that the importanceof an HCA pathway to preterm delivery could vary consid-erably with the population studied. Often reports on HCAand preterm delivery come from inner-city teaching hos-pitals with large proportions of minority populations andurban poor (38, 59, 60).

    Among the African-American deliveries in our study,three factors increased the effect size of the associationbetween HCA and preterm delivery: 1) higher polymorpho-nuclear leukocytes/high-powered field threshold or moreadvanced maternal and fetal inflammatory response; 2) re-stricting the outcome to preterm delivery before 35 weeks;and 3) exclusion of medically indicated preterm delivery.Thus, even studies with similar populations may be ex-pected to produce different results if these factors are notconsistent. Our findings are in agreement with previous re-search showing that inflammation/infection plays a largerrole in the earliest preterm delivery (2126) and in pretermdelivery that begins with spontaneous labor or rupture ofmembranes (21, 32, 33) as compared with medically indi-cated preterm delivery.

    One study limitation was sample size. We had 26 White/other and 22 African-American deliveries before 35 weeksafter excluding medically indicated preterm deliveries.Based on the low prevalence of high maternal/fetal inflam-matory response in the White/other placentas, one wouldneed at least 324 nonmedically indicated preterm deliveriesbefore 35 weeks in Whites/others to have 80 percent powerto detect an odds ratio of 2.0. We also had too few extremepreterm deliveries, that is, before 33 weeks, to separatelyexamine this subgroup. The importance of HCA may bemore similar across race/ethnic groups in the before-33-week deliveries compared with the marked race/ethnic dis-parities that we observed in the before-35-week deliveries.Another limitation is one inherent in much of the literatureon HCA. We used the location of polymorphonuclear leu-kocytes to infer an inflammatory infiltrate and direction ofchemotaxis, and we had no additional immunohistologicstaining or RNA microarray evidence that these polymor-phonuclear leukocytes were recruited as part of an inflam-matory process.

    Despite these limitations, this study also had multiplestrengths. The sample was socioeconomically diverse, andparticipants were recruited from a large number of prenatalclinics that serve urban, suburban, and rural women withlow- and high-risk pregnancies, thereby improving the gen-eralizability of findings. During gross and histologic exami-nations, the pathologist was unaware of the gestational weekat delivery, pregnancy complications, delivery circums-tances, and race/ethnicity associated with each placenta,

    TABLE 2. Variation in denition of histologic chorioamnionitis and its association with preterm delivery at 3536 weeks and at less

    than 35 weeks, Pregnancy Outcomes and Community Health Study,* 19982004

    Thresholdpolymorphonuclear

    leukocytes/high-poweredeld for histologicchorioamnionitisy

    Preterm delivery at 3536 weeksz Preterm delivery at 10 0.4 0.3, 0.6 0.5 0.3, 0.8 0.2 0.1, 0.3 0.3 0.1, 0.7 0.3 0.1, 0.6 0.4 0.1, 1.2

    >30 0.5 0.3, 0.9 0.6 0.3, 1.2 0.2 0.1, 0.6 0.2 0.1, 1.0 0.3 0.1, 1.1 0.5 0.1, 2.0

    >100 0.7 0.3, 1.7 0.5 0.1, 1.6 0.5 0.1, 2.3 0.8 0.2, 3.5 1.1 0.2, 4.8 1.6 0.4, 7.1

    African Americans

    >0 0.3 0.1, 0.6 0.4 0.2, 0.9 0.3 0.1, 0.6 1.0 0.4, 2.1 0.6 0.2, 1.7 2.0 0.7, 5.2

    >10 0.5 0.2, 0.9 0.6 0.3, 1.3 0.8 0.3, 1.7 2.0 0.9, 4.3 2.5 0.6, 4.0 4.0 1.5, 10.2

    >30 0.5 0.2, 1.1 0.7 0.3, 1.6 1.5 0.7, 3.3 1.3 0.6, 3.0 2.5 1.0, 6.0 2.0 0.8, 4.6

    >100 0.9 0.3, 2.2 1.3 0.5, 3.2 3.2 1.4, 7.1 2.9 1.2, 7.0 4.9 2.0, 11.8 4.1 1.6, 10.4

    * Referred to as the POUCH Study.

    yHistologic chorioamnionitis was considered present if the highest number of polymorphonuclear leukocytes/high-powered eld observed inany tissue component was equal to or greater than the dened cutoff (threshold).

    z Term deliveries are the referent group, and all odds ratios and 95% condence intervals incorporate cohort/subcohort sampling weights asapplicable; the column subtitles denote tissue components used to assess evidence of histologic chorioamnionitis.

    Greater than zero polymorphonuclear leukocytes/high-powered eld and an inammatory pattern.

    Histologic Chorioamnionitis and Preterm Delivery 791

    Am J Epidemiol 2007;166:786794

    by guest on Novem

    ber 20, 2014http://aje.oxfordjournals.org/

    Dow

    nloaded from

  • TABLE 3. Associations between maternal and fetal inammatory responses and preterm delivery (
  • information that could influence pathologists impressions.Placental assessment followed a strict protocol and incorpo-rated considerable detail to permit construction of multipleHCA definitions; this combination of rigor and flexibilitywas a unique feature of this study. We were able to assessthe effects of HCA definition, race/ethnicity, and pretermdelivery subtypes within a single study, whereas compari-sons across studies are often confounded by other salientfactors such as variations in study samples and methodolo-gies. In summary, after noting large inconsistencies in resultsfrom studies linking HCA and preterm delivery, we hypoth-esized that factors such as HCA definition, exclusion ofmedically indicated preterm deliveries, gestational weeksof the preterm deliveries, and race/ethnicity might explainsome of the heterogeneity. Our results supported thesehypotheses and offered further evidence that HCA maybe a powerful contributor to ethnic differences in pretermdelivery risk.

    ACKNOWLEDGMENTS

    This study was supported by funding from the following:National Institute of Child Health and Human Developmentand the National Institute of Nursing Research (grant R01HD34543), March of Dimes Foundation (grants 20FY01-38and 20-FY04-37), Thrasher Research Foundation (grant02816-7), and Centers for Disease Control and Prevention(grant U01 DP000143-01).

    The authors would also like to thank the PrematurityStudy Group, project director Dr. Bertha Bullen, andphysician abstractors Dr. Joseph Marshall, Dr. Lynn Reuss,and Dr. Judith Suess.

    Conflict of interest: none declared.

    REFERENCES

    1. MacDorman MF, Martin JA, Mathews TJ, et al. Explaining the20012002 infant mortality increase in the United States: datafrom the linked birth/infant death data set. Int J Health Serv2005;35:41542.

    2. Kramer MS, Demissie K, Yang H, et al. The contribution ofmild and moderate preterm birth to infant mortality. Fetal andInfant Health Study Group of the Canadian Perinatal Surveil-lance System. JAMA 2000;284:8439.

    3. Stevenson DK, Wright LL, Lemons JA, et al. Very low birthweight outcomes of the National Institute of Child Health andHuman Development Neonatal Research Network, January1993 through December 1994. Am J Obstet Gynecol 1998;179:16329.

    4. Anderson P, Doyle LW. Neurobehavioral outcomes of school-age children born extremely low birth weight or very pretermin the 1990s. JAMA 2003;289:326472.

    5. Bhutta AT, Cleves MA, Casey PH, et al. Cognitive and be-havioral outcomes of school-aged children who were bornpreterm: a meta-analysis. JAMA 2002;288:72837.

    6. Marlow N, Wolke D, Bracewell MA, et al. Neurologic anddevelopmental disability at six years of age after extremelypreterm birth. N Engl J Med 2005;352:919.

    7. Hack M, Taylor HG, Drotar D, et al. Chronic conditions,functional limitations, and special health care needs of school-aged children born with extremely low-birth-weight in the1990s. JAMA 2005;294:31825.

    8. Ancel PY, Saurel-Cubizolles MJ, Di Renzo GC, et al. Socialdifferences of very preterm birth in Europe: interaction withobstetric history. Am J Epidemiol 1999;149:90815.

    9. Luo ZC, Kierans WJ, Wilkins R, et al. Disparities in birth out-comes by neighborhood income: temporal trends in rural andurban areas, British Columbia. Epidemiology 2004;15:67986.

    10. Pickering RM, Deeks JJ. Risks of delivery during the 20th tothe 36th week of gestation. Int J Epidemiol 1991;20:45666.

    11. Thompson JM, Irgens LM, Rasmussen S, et al. Secular trendsin socio-economic status and the implications for pretermbirth. Paediatr Perinat Epidemiol 2006;20:1827.

    12. Branum AM, Schoendorf KC. Changing patterns of lowbirthweight and preterm birth in the United States, 198198.Paediatr Perinat Epidemiol 2002;16:815.

    13. Singh GK, Yu SM. Adverse pregnancy outcomes: differencesbetween US- and foreign-born women in major US racial andethnic groups. Am J Public Health 1996;86:83743.

    14. Joyce T, Martin JA, Hamilton BE, et al. Births: final datafor 2003. National vital statistics reports. Hyattsville, MD:National Center for Health Statistics, 2005.

    15. Adams MM, Read JA, Rawlings JS, et al. Preterm deliveryamong black and white enlisted women in the United StatesArmy. Obstet Gynecol 1993;81:6571.

    16. McGrady GA, Sung JF, Rowley DL, et al. Preterm deliveryand low birth weight among first-born infants of Black andWhite college graduates. Am J Epidemiol 1992;136:26676.

    17. Schoendorf KC, Hogue CJ, Kleinman JC, et al. Mortalityamong infants of black as compared with white college-educated parents. N Engl J Med 1992;326:15226.

    18. Zhang J, Savitz DA. Preterm birth subtypes among blacks andwhites. Epidemiology 1992;3:42833.

    19. Collins JW Jr, Hammond NA. Relation of maternal race to therisk of preterm, non-low birth weight infants: a populationstudy. Am J Epidemiol 1996;143:3337.

    20. Migone A, Emanuel I, Mueller B, et al. Gestational durationand birthweight in white, black and mixed-race babies. Pae-diatr Perinat Epidemiol 1991;5:37891.

    21. Ogunyemi D, Murillo M, Jackson U, et al. The relationshipbetween placental histopathology findings and perinatal out-come in preterm infants. J Matern Fetal Neonatal Med 2003;13:1029.

    22. Salafia CM, Vogel CA, Vintzileos AM, et al. Placental path-ologic findings in preterm birth. Am J Obstet Gynecol 1991;165:9348.

    23. Hillier SL, Martius J, Krohn M, et al. A case-control study ofchorioamnionic infection and histologic chorioamnionitis inprematurity. N Engl J Med 1988;319:9728.

    24. Naeye RL, Peters EC. Causes and consequences of prematurerupture of fetal membranes. Lancet 1980;1:1924.

    25. Zlatnik FJ, Gellhaus TM, Benda JA, et al. Histologic cho-rioamnionitis, microbial infection, and prematurity. ObstetGynecol 1990;76:3559.

    26. Lahra MM, Jeffery HE. A fetal response to chorioamnionitis isassociated with early survival after preterm birth. Am J ObstetGynecol 2004;190:14751.

    27. Leitich H, Bodner-Adler B, Brunbauer M, et al. Bacterialvaginosis as a risk factor for preterm delivery: a meta-analysis.Am J Obstet Gynecol 2003;189:13947.

    28. Holzman C, Leventhal JM, Qiu H, et al. Factors linked tobacterial vaginosis in nonpregnant women. Am J PublicHealth 2001;91:166470.

    Histologic Chorioamnionitis and Preterm Delivery 793

    Am J Epidemiol 2007;166:786794

    by guest on Novem

    ber 20, 2014http://aje.oxfordjournals.org/

    Dow

    nloaded from

  • 29. Bradshaw CS, Morton AN, Garland SM, et al. Higher-riskbehavioral practices associated with bacterial vaginosis com-pared with vaginal candidiasis. Obstet Gynecol 2005;106:10514.

    30. Royce RA, Jackson TP, Thorp JM Jr, et al. Race/ethnicity,vaginal flora patterns, and pH during pregnancy. Sex TransmDis 1999;26:96102.

    31. Koumans EH, Sternberg MR, McQuillan G, et al. Prevalenceof bacterial vaginosis in the United States, 20012002. Pre-sented at the 2006 National STD Prevention Conference ofCDC, Jacksonville, Florida, May 811, 2006.

    32. Andrews WW, Goldenberg RL, Faye-Petersen O, et al. TheAlabama Preterm Birth Study: polymorphonuclear andmononuclear cell placental infiltrations, other markers of in-flammation, and outcomes in 23- to 32-week preterm newborninfants. Am J Obstet Gynecol 2006;195:8038.

    33. Lamont RF, Taylor-Robinson D, Newman M, et al. Sponta-neous early preterm labour associated with abnormal geni-tal bacterial colonization. Br J Obstet Gynaecol 1986;93:80410.

    34. Beebe LA, Cowan LD, Altshuler G. The epidemiology ofplacental features: associations with gestational age and neo-natal outcome. Obstet Gynecol 1996;87:7718.

    35. Perkins RP, Zhou SM, Butler C, et al. Histologic chorioam-nionitis in pregnancies of various gestational ages: impli-cations in preterm rupture of membranes. Obstet Gynecol1987;70:85660.

    36. Arias F, Rodriquez L, Rayne SC, et al. Maternal placentalvasculopathy and infection: two distinct subgroups amongpatients with preterm labor and preterm ruptured membranes.Am J Obstet Gynecol 1993;168:58591.

    37. Dollner H, Vatten L, Halgunset J, et al. Histologicchorioamnionitis and umbilical serum levels of pro-inflam-matory cytokines and cytokine inhibitors. BJOG 2002;109:5349.

    38. Guzick DS, Winn K. The association of chorioamnionitis withpreterm delivery. Obstet Gynecol 1985;65:1116.

    39. Hillier SL, Krohn MA, Kiviat NB, et al. Microbiologic causesand neonatal outcomes associated with chorioamnion infec-tion. Am J Obstet Gynecol 1991;165:95561.

    40. Mueller-Heubach E, Rubinstein DN, Schwarz SS. Histologicchorioamnionitis and preterm delivery in different patientpopulations. Obstet Gynecol 1990;75:6226.

    41. Naeye RL. Factors that predispose to premature rupture of thefetal membranes. Obstet Gynecol 1982;60:938.

    42. Potkul RK, Moawad AH, Ponto KL. The association of sub-clinical infection with preterm labor: the role of C-reactiveprotein. Am J Obstet Gynecol 1985;153:6425.

    43. Sherman DJ, Tovbin J, Lazarovich T, et al. Chorioamnionitiscaused by gram-negative bacteria as an etiologic factor inpreterm birth. Eur J Clin Microbiol Infect Dis 1997;16:41723.

    44. Amory JH, Adams KM, Lin MT, et al. Adverse outcomes afterpreterm labor are associated with tumor necrosis factor-alphapolymorphism -863, but not -308, in mother-infant pairs. Am JObstet Gynecol 2004;191:13627.

    45. De Dooy J, Colpaert C, Schuerwegh A, et al. Relationshipbetween histologic chorioamnionitis and early inflammatoryvariables in blood, tracheal aspirates, and endotrachealcolonization in preterm infants. Pediatr Res 2003;54:11319.

    46. De Felice C, Toti P, Laurini RN, et al. Early neonatal brain in-jury in histologic chorioamnionitis. J Pediatr 2001;138:1014.

    47. Elimian A, Figueroa R, Spitzer AR, et al. Antenatal cortico-steroids: are incomplete courses beneficial? Obstet Gynecol2003;102:3525.

    48. Hillier SL, Witkin SS, Krohn MA, et al. The relationship ofamniotic fluid cytokines and preterm delivery, amniotic fluidinfection, histologic chorioamnionitis, and chorioamnion in-fection. Obstet Gynecol 1993;81:9418.

    49. Blanc W. Placental inflammation. In: Naeye RL, KissaneJM, Kaufman N, eds. Perinatal diseases. Baltimore, MD:Williams & Wilkins, 1981:6799.

    50. Redline RW, Faye-Petersen O, Heller D, et al. Amniotic in-fection syndrome: nosology and reproducibility of placentalreaction patterns. Pediatr Dev Pathol 2003;6:43548.

    51. Salafia CM, Weigl C, Silberman L. The prevalence and dis-tribution of acute placental inflammation in uncomplicatedterm pregnancies. Obstet Gynecol 1989;73:3839.

    52. Tauscher MK, Berg D, Brockmann M, et al. Association ofhistologic chorioamnionitis, increased levels of cord bloodcytokines, and intracerebral hemorrhage in preterm neonates.Biol Neonate 2003;83:16670.

    53. Cherouny PH, Pankuch GA, Botti JJ, et al. The presence ofamniotic fluid leukoattractants accurately identifies histologicchorioamnionitis and predicts tocolytic efficacy in patientswith idiopathic preterm labor. Am J Obstet Gynecol 1992;167:6838.

    54. SAS Institute, Inc. Base SAS 9.1.3 procedures guide. Cary,NC: SAS Publishing, 2006.

    55. Grether JK, Eaton A, Redline R, et al. Reliability of placentalhistology using archived specimens. Paediatr Perinat Epide-miol 1999;13:48995.

    56. Simmonds M, Jeffery H, Watson G, et al. Intraobserver andinterobserver variability for the histologic diagnosis of cho-rioamnionitis. Am J Obstet Gynecol 2004;190:1525.

    57. Kramer MS, Chen MF, Roy I, et al. Intra- and interobserveragreement and statistical clustering of placental histopatho-logic features relevant to preterm birth. Am J Obstet Gynecol2006;195:16749.

    58. Dammann O, Leviton A, Allred EN. What explains away theincreased risk of histological chorioamnionitis in African-American mothers of very-low-birthweight infants? Develop-mental Epidemiology Network Investigators. Paediatr PerinatEpidemiol 2000;14:209.

    59. Greig PC, Ernest JM, Teot L, et al. Amniotic fluid interleukin-6 levels correlate with histologic chorioamnionitis and amni-otic fluid cultures in patients in premature labor with intactmembranes. Am J Obstet Gynecol 1993;169:103544.

    60. Holcroft CJ, Askin FB, Patra A, et al. Are histopathologicchorioamnionitis and funisitis associated with metabolic aci-dosis in the preterm fetus? Am J Obstet Gynecol 2004;191:201015.

    794 Holzman et al.

    Am J Epidemiol 2007;166:786794

    by guest on Novem

    ber 20, 2014http://aje.oxfordjournals.org/

    Dow

    nloaded from