Amniotic Infection Syndrome: Nosology and Reproducibility of Placental Reaction Patterns

1 2 3 RAYMOND W. REDLINE, * ONA FAYE-PETERSEN, DEBRA HELLER, 4 5 6 FAISAL QURESHI, VAN SAVELL, CAROLE VOGLER, AND THE SOCIETY FOR PEDIATRIC PATHOLOGY,PERINATAL SECTION,AMNIOTIC FLUID INFECTION NOSOLOGY COMMITTEE 1Department of Pathology, University Hospitals of Cleveland and Case Western Reserve University, 11100 Euclid Avenue, Cleveland, OH 44106, USA 2Department of Pathology, University of Alabama at Birmingham, Birmingham, AL, USA 3Department of Pathology, University of Medicine and Dentistry of New Jersey—New Jersey Medical School, Newark, NJ, USA 4Pathology Department, DMC University Laboratories, Hutzel Site and Wayne State University, Detroit, MI, USA 5Department of Pathology, Driscoll Children’s Hospital, Corpus Christie, TX, USA 6Department of Pathology, St. Louis University School of Medicine, St. Louis, MO, USA

Received July 1, 2002; accepted May 12, 2003; published online September 11, 2003.

ABSTRACT 0.2–0.6, fair/moderate; Ͼ 0.6, substantial. Kappa values for the 12 lesions evaluated in 20 cases by the six pathol- Clinically responsive placental examination seeks to pro- ogists were: acute chorioamnionitis/maternal inflamma- vide useful information regarding the etiology, progno- tory response (any, 0.93; severe 0.76; advanced stage, sis, and recurrence risk of pregnancy disorders. The 0.49); chronic (subacute) chorioamnionitis (0.25); acute purpose of this study was to assemble and validate a chorioamnionitis/fetal inflammatory response (any, complete set of the placental reaction patterns seen with 0.90; severe, 0.55; advanced stage, 0.52); chorionic vessel amniotic fluid infection in the hope that this might pro- thrombi (0.37); peripheral funisitis (0.84); acute villitis vide a standardized diagnostic framework useful for (0.90); acute intervillositis/intervillous abscesses (0.65), practicing pathologists. Study cases (14 with amniotic and decidual plasma cells (0.30). Adoption of this clearly fluid infection, 6 controls) were reviewed blindly by six defined, clinically relevant, and pathologically reproduc- pathologists after agreement on a standard set of diag- ible terminology could enhance clinicopathologic corre- nostic criteria. After analysis of initial results, criteria lation and provide a framework for future clinical re- were refined and a second, overlapping set of cases were search. reviewed. Majority vote served as the gold standard. Grading and staging of maternal and fetal inflammatory Key words: amniotic infection syndrome, chorioamnio- responses was found to be more reproducible using a nitis, nomenclature, placenta, reproducibility two- versus three-tiered grading system than a three- versus five-tiered staging system (overall agreement INTRODUCTION 81% vs. 71%). Sensitivity, specificity, and efficiency for The general sequence of pathologic changes ac- individual observations ranged from 67–100% (24/30 Ͼ 90%). Reproducibility was measured by unweighted companying amniotic fluid infection has been rec- kappa values and interpreted as follows: Ͻ 0.2, poor; ognized for some time [1, 2]. Other clinically relevant histologic patterns have been added to this *Corresponding author, e-mail: [email protected] basic framework [3–9]. Despite a fairly extensive literature on this topic, there has been little change as possible to that proposed by the College of in the prevailing diagnostic terminology which has American Pathologists consensus group on placen- basically been “acute chorioamnionitis with or tal diagnosis [17]. Tabulated results from the pre- without funisitis.” Recent data has suggested a re- liminary round (round 1) were circulated and dis- lationship between placental inflammation and cussed. Representative photomicrographs of each important clinical outcomes such as neurologic lesion were reviewed at a meeting of the Society for impairment and chronic lung disease [10–12]. This Pediatric Pathology—Perinatal Section. After the has coincided with a more active clinical approach initial review a second revised set of diagnostic to the management of chorioamnionitis and pre- criteria were proposed and agreed upon. Twelve of term labor [13]. These changes have led to the need the original cases were retained, eight new cases for more detailed description, more specific diag- were added (five new cases and three new con- nostic criteria, and greater standardization of di- trols), and the numerical order was changed. This agnostic terminology in this area. second overlapping set of 20 cases was then circu- As in other organ systems, questions have lated for reexamination followed by an analysis of been raised regarding the reliability of placental individual performance and reproducibility (round diagnosis, i.e., the ability of expert placental pa- 2). thologists to agree with each other and with their The revised diagnostic criteria as agreed upon general pathology colleagues [14–16]. With re- by the study participants before the final evalua- newed interest in the clinical significance of pla- tion (round 2) were as follows: cental pathology, the time seems appropriate to introduce more precision and uniformity in pla- Maternal inflammatory responses cental diagnosis. With this in mind, the Perinatal Stage/progression of disease Section of the Society for Pediatric Pathology has Stage 1 (acute subchorionitis/early acute chorion- undertaken an initiative to review, define, and val- itis) ϭ patchy-diffuse accumulations of neutro- idate diagnostic criteria for a number of different phils in the subchorionic plate fibrin (Fig. 1a) placental reaction patterns. The current report pre- and/or membranous chorionic trophoblast layer (a sents the results of the Amniotic Fluid Infection few scattered neutrophils in the lower half of cho- Nosology Committee. rionic plate and/or the membranous chorionic connective tissue are allowed) (Fig. 1b). Stage 2 METHODS (acute chorioamnionitis) ϭ more than a few scat- A study set of 20 cases was assembled from the files tered neutrophils in the chorionic plate or mem- of University Hospitals of Cleveland. Three slides branous chorionic connective tissue and/or the from each case (umbilical cord, placental mem- amnion (Fig. 1c). Stage 3 (necrotizing chorioam- brances, and one full-thickness section of placental nionitis) ϭ degenerating neutrophils (karyorrhe- parenchyma) were selected. Gestational age and xis), thickened eosinophilic amniotic basement placental weight were provided for each case. membrane, and at least focal amnionic epithelial Fourteen placentas were originally diagnosed by degeneration/sloughage (Fig. 1d). one of the pathologists (R.W.R.) with one or more findings consistent with amniotic fluid infection. Grade/intensity The remaining six placentas had other pathologic Grade 1 (mild–moderate) ϭ individual or small diagnoses, but lacked signs of amniotic fluid infec- clusters of maternal neutrophils diffusely infiltrat- tion. Placental reaction patterns relevant to amni- ing amnion, chorionic plate, chorion laevae, and/or otic fluid infection were chosen for evaluation. subchorionic fibrin (Fig. 1c). Grade 2 (severe) ϭ These patterns are described below, summarized three or more chorionic microabscesses (microab- in Table 1, and illustrated in Figures 1–3. All cases scess ϭ confluent neutrophils measuring at least were examined in a blinded fashion by the six 10 ϫ 20 cells in extent) between chorion and members of the study group using a previously decidua in the membranes and/or under the cho- agreed upon set of diagnostic criteria (round 1). rionic plate (Fig. 1e) or a continuous band of Terminology was chosen to correspond as closely confluent chorionic polymorphonuclear leuko-

436 R.W. REDLINE ET AL. Table 1. Placenta reaction patterns related to amniotic fluid infection: nomenclature and definitions Diagnostic categories Suggested diagnostic terminology Definitions Maternal inflammatory response Stage 1—Early Acute subchorionitis or chorionitis PMN in subchorionic fibrin and/or membrane trophoblast (Fig. 1a,b) 2—Intermediate Acute chorioamnionitis Diffuse-patchy PMN in fibrous chorion and/or amnion (Fig. 1c) 3—Advanced Necrotizing chorioamnionitis PMN karyorrhexis, amniocyte necrosis, and/or amnion basement membrane thickening/hypereosinophilia (Fig. 1d) Grade 1—Mild–moderate No special terminology required Not severe as defined below 2—Severe Severe acute chorioamnionitis or with subchorionic Confluent PMN (Ն 10 ϫ 20 cells in extent) between chorion and microabscesses decidua; Ն 3 isolated foci or continuous band (Fig. 1e) Other Chronic (or subacute) chorioamnionitis Subamnionic mononuclear cell infiltrate with occasional PMN (meconium and hemosiderin-laden macrophages excluded) (Fig. 1f) Fetal inflammatory response Stage 1—Early With chorionic vasculitis or umbilical phlebitis Intramural PMN-chorionic vessels and/or umbilical vein (Fig. 2a,b) 2—Intermediate With umbilical vasculitis (one or two arteries Ϯ Intramural PMN-umbilical artery or arteries (Ϯ umbilical vein) (Fig. vein) or umbilical panvasculitis (all vessels) 2c) 3—Advanced With (subacute) necrotizing funisitis or with PMN Ϯ associated debris in concentric bands-rings-halos around concentric umbilical perivasculitis one or more umbilical vessels (Fig. 2d) A MNIOTIC Grade 1—Mild–moderate No special terminology required Not severe as defined below

I 2—Severe With a severe fetal inflammatory response or with Near confluent intramural PMN-chorionic and/or umbilical vessels NFECTION intense chorionic (umbilical) vasculitis with attenuation/degeneration of VSMC (Fig. 2e,f) Other With associated fetal vessel thrombi Recent thrombosis associated with intramural PMN (Fig. 2f) Other specific features Peripheral funisitis Focal aggregates of PMN at the umbilical cord surface (Fig. 3a) S

YNDROME Acute villitis PMN in villous stroma (or between trophoblast and stroma) (Fig. 3b) Acute intervillositis with intervillous abscesses Patchy-diffuse PMN in intervillous space (Fig. 3c) Decidual plasma cells Unequivocal plasma cells in decidua basalis or capsularis (Fig. 3d)

437 PMN, polymorphonuclear leukocyte; VSMC, vascular smooth muscle cell. Figure 1. Histologic features of reaction patterns related Necrotizing (maternal stage 3) chorioamnionitis: degen- to the maternal inflammatory response to amniotic fluid erating subamnionic neutrophils, thickened amnionic infection. a. Acute subchorionitis (maternal stage 1): basement membrane, and desquamation of amnionic epi- neutrophils emanating from the intervillous space are thelial cells are seen in placental membranes. e. Severe (ma- diffusely scattered in the subchorionic fibrin. b. Acute ternal grade 2) acute chorioamnionitis: (sub)chorionic chorionitis of chorion laevae (maternal stage 1): a band microabscesses are depicted comprised of confluent ag- of neutrophils emanating from decidual small venules gregates of maternal neutrophils, at least 10 ϫ 20 cells in are gathered at the chorion laevae trophoblast layer extent. f. Chronic (subacute) chorioamnionitis: patchy without spread to the overlying amnion. c. Acute chorio- maternally derived histiocytic infiltrate of the chorionic amnionitis, not otherwise specified (maternal stage 2, plate concentrated immediately below the amnionic sur- grade 1): neutrophils are seen in both chorionic and am- face (left). Occasional neutrophils are also present. nionic connective tissue in the placental membranes. d. cyte (PMN) more than 10 cells in width occupy- casional maternally derived mononuclear cells ing more than half of the subchorionic fibrin or (usually macrophages) in the chorionic plate one revolution of the membrane role. Chronic (most frequently below the amnion) in cases (or subacute) chorioamnionitis ϭ more than oc- without exogenous pigment deposition (i.e.,

438 R.W. REDLINE ET AL. Figure 2. Histologic features of reaction patterns related ical artery. d. Subacute necrotizing funisitis (fetal stage to the fetal maternal inflammatory response to amniotic 3): an arc of degenerating neutrophils and mineralized fluid infection. a. Acute chorionic vasculitis (fetal stage 1, cell debris in Wharton’s jelly surrounds an umbilical ves- grade 1): neutrophils are scattered throughout the up- sel. e. Intense chorionic vasculitis (fetal grade 2): near per amnionic aspect of a major chorionic plate vessel. b. confluent fetal neutrophils infiltrate the outer (am- Umbilical phlebitis (fetal stage 1): neutrophils are iden- nionic) wall of a chorionic vessel. f. Chorionic vasculitis tified between the widely separated smooth muscle fas- with adherent fetal vessel thrombus: a recent nonocclu- cicles of the umbilical vein. c. Umbilical arteritis (fetal sive thrombus adheres to the outer (amnionic) wall of an stage 2): neutrophils infiltrate the subendothelial cush- acutely inflamed chorionic vessel. ion and tightly packed smooth muscle cells of the umbil-

meconium, hemosiderin). Neutrophils may be Fetal inﬂammatory responses rare or abundant, but a coexistent acute chorio- Stage/progression of disease amnionitis should be present in at least one sec- Stage 1 (chorionic vasculitis/umbilical phlebitis) ϭ tion (Fig. 1f). neutrophils in the wall of any chorionic plate vessel

AMNIOTIC INFECTION SYNDROME 439 Figure 3. Histologic features of other specific features of trophils are also present in the intervillous fibrin. c. Acute amniotic fluid infection. a. Peripheral funisitis: a triangu- intervillositis/intervillous abscesses: near confluent neu- lar neutrophilic microabscess is seen just below the um- trophils are noted in the intervillous space with only fo- bilical surface epithelium. b. Acute villitis: fetal neutro- cal involvement of adjacent villous stroma. d. Decidual phils are present in the villous stroma and plasma cells: small aggregates of plasma cells are present subtrophoblastic space of terminal villi. Occasional neu- in the decidua basalis.

(Fig. 2a) or the umbilical vein (Fig. 2b). Extension 2a). Grade 2 (severe) ϭ chorionic plate (or umbil- of neutrophils into Wharton’s jelly is allowed if not ical) vessels with near confluent neutrophils plus aggregated in a concentric band, ring, or halo attenuation and/or degeneration of vascular around the umbilical vein. Stage 2 (umbilical vas- smooth muscle cells on the side facing the amni- culitis) ϭ neutrophils in one or both umbilical otic cavity (Fig. 2e). Associated fetal vessel thrombi arteries Ϯ vein (Fig. 2c). Extension of neutrophils ϭ recent thrombosis of chorionic plate (or umbil- into Wharton’s jelly is allowed if not aggregated in ical) vessel(s) in the context of a fetal inflammatory a concentric band, ring, or halo around the umbil- response (Fig. 2f). ical vessel(s). Stage 3 (necrotizing funisitis or concentric umbilical perivasculitis) ϭ neutrophils, cel- Other specific features lular debris, eosinophilic precipitate, and/or Peripheral funisitis ϭ small punctate microab- mineralization arranged in a concentric band, scesses with neutrophils on the outer surface of ring, or halo around one or more umbilical vessels cord (Fig. 3a). Acute villitis ϭ neutrophils in the (generally most severe on the side nearest the pe- villous stroma and/or the adjacent subtrophoblas- riphery of the cord) (Fig. 2d). tic space (Fig. 3b). Acute intervillositis/intervillous Grade/intensity abscesses ϭ neutrophils within perivillous fibrin or Grade 1 (mild–moderate) ϭ scattered neutrophilic the intervillous space (Fig. 3c). Decidual plasma infiltrate in the subendothelial or intramural por- cells ϭ unequivocal plasma cells in either the de- tions of any chorionic (or umbilical) vessel (Fig. cidua basalis or capsularis (Fig. 3d).

440 R.W. REDLINE ET AL. Data analysis Grading and staging The gold standard for the presence or absence of a All possible histologic gradations of intensity diagnosis in each case was the group consensus. (grade) and progression of disease (stage) for ma- The diagnosis from the original institution was to ternal and fetal inflammatory responses are poten- be used as the “tie breaker” in split cases, but was tially relevant to clinical outcome. However, exces- not required for any diagnosis in any case. Individ- sive subclassification in the absence of ual variations from consensus grades and stages documented clinical significance is one cause of were tabulated for rounds 1 and 2 and the percent unacceptable interobserver variability amongst pa- agreement was determined. The final analysis thologists. Table 2 compares the group perfor- (round 2) evaluated the following 12 reaction pat- mance using two different classification schemes, terns: any maternal of fetal inflammatory response and shows the results from round 1 using a slight (grade and stage not ϭ 0), severe maternal or fetal modification of a previously described classifica- inflammatory response (grade 2), advanced stage tion system used in research studies correlating maternal or fetal inflammatory response (stage 3), placental diagnosis with clinical outcome [19]. and the six other features listed above. Collective This scheme utilized three-tiered grading and five- performance of the group was assessed by combin- tiered staging of both maternal and fetal inflammatory responses. Overall agreement using this ing the individual diagnoses to derive specificity system was modest (71%) and particular problems (true negatives/true negatives plus false positives), existed in separating intermediate and low grades sensitivity (true positives/true positives plus false of inflammation (Table 2). These distinctions are negatives), and efficiency (true positives plus true of no documented clinical significance. Most of negatives/total observations) relative to the gold disagreement was the result of minor disagree- standard. Interobserver reliability was assessed by ments (Ϯ 1) suggesting that a simplified scheme unweighted kappa analysis [18]. Kappa analysis is might yield better reproducibility without sacrific- a measurement of agreement between observers ing the clinical significance associated with high on a case to case basis. Kappa values can vary grade and stage. For round 2, the number of grades from Ϫ1toϩ1, with Ϫ1 indicating perfect in- was reduced to two, retaining the high grade cate- verse correlation; 0, no correlation; and ϩ1, per- gories because of their established clinical signifi- fect positive correlation. Interpretation of kappa cance (see Discussion). The number of stages was values can vary. We used the following guide- reduced to three, distinguishing only those cases Ͻ lines: 0.2, poor agreement; 0.2–0.6, fair/moder- with findings suggestive of early and advanced Ͼ ate agreement; 0.6 substantial agreement [18]. stage infection from those without these character- istics. The results shown in Table 2 showed a sub- RESULTS stantial improvement in overall agreement (81%). Preliminary analysis Final group survey results (round 2) The initial study set of 20 cases, 14 with features A complete analysis of the group’s performance is associated with amniotic fluid infection and 6 con- shown in Table 3. Twelve reaction patterns were trols, was circulated to group members for blinded evaluated: presence of any maternal or fetal in- review and coding on standardized forms (round flammation, presence of severe (grade 2) maternal 1). Analysis of the first round results revealed un- or fetal inflammation, presence of advanced (stage acceptable interobserver variation. A revised and 3) maternal or fetal inflammation, chronic (sub- simplified set of definitions was circulated and acute) chorioamnionitis, fetal vessel thrombi, pe- agreed upon and a second overlapping set of study ripheral funisitis, acute villitis, acute intervillositis/ cases was circulated for review (round 2). The le- intervillous abscesses, and decidual plasma cells. sions, terminology, and definitions used in the final The gold standard for the presence or absence of round are described in the Methods section, sum- each condition (prevalence) in the 20 study cases marized in Table 1, and illustrated in Figures 1, 2, was group consensus. Group consensus was unan- 3. imous for 43/60 (72%) diagnoses and near unani-

AMNIOTIC INFECTION SYNDROME 441 Table 2. Stage and grade of infection, preliminary and final rounds, comparing expanded versus contracted scales Stage/grade No. of cases Individual deviations from consensusa consensus (% of observers) positive Initial evaluation (round 1): three-tiered -2 -1 0 ϩ1 ϩ2 grading/five-tiered staging Maternal inflammatory response Grade 12— 867817 2 2 085042— 3 10 3 17 80 —— Combined 14 2 14 74 8 2 Stage 11— 17 83 0 0 2 4 0 256780 3 5 0 7 70176 4 3 0 176716— 5 1 16 17 67 —— Combined 14 1 16 68 13 2 Fetal inflammatory response Grade 14— 8 541721 2 4 0 135433— 3 5 0397—— Combined 14 0 8 68 17 7 Stage 14— 88840 2 1 0 505000 3 2 0 0 83170 4 4 3 11833 — 5 1 0 0 100 —— Combined 14 1 10 83 6 0 Overall performance (total observations) (324) 1 14 71 11 3 Maternal inflammatory response Grade 15— 38413— 2901189—— Combined 14 0 8 87 5 — Stage 12— 8 59258 2 8 06868— 3401783—— Combined 14 0 10 81 8 1 Fetal inflammatory response Grade 18— 76726— (continued)

442 R.W. REDLINE ET AL. Table 2. (Continued) Stage/grade No. of cases Individual deviations from consensusa consensus (% of observers) positive 2601486—— Combined 14 0 10 76 14 — Stage 14— 17 83 0 0 2 7 0 117910— 3302872—— Combined 14 0 16 79 5 0 Overall performance (total observations) (336) 0 11 81 8 Ͻ 1 aPositive and negative numbers specify the number of stages or grades over or under the consensus score for each individual observer.

mous (5/6) for another 12%. Overall agreement (stage 3) maternal inflammatory response (Fig. between group prevalence and a priori prevalence 1d), severe (grade 2) fetal inflammatory response (the original diagnosis of the submitting patholo- (Fig. 2e) advanced (stage 3) fetal inflammatory re- gist, R.W.R.) for the 10 lesions in the 20 cases was sponse (Fig. 2d), chronic (subacute) chorioamnio- 87%. The range of individual prevalences for each nitis (Fig. 1f), fetal vessel thrombi (Fig. 2f), and lesion among the 20 cases was generally narrow. decidual plasma cells (Fig. 3d)—was fair to mod- Group performance was evaluated in two ways. erate. First, each individual response was classified as true or false positive, or true or false negative, DISCUSSION relative to the group consensus and overall sensi- Premature delivery is the major cause of perinatal tivity, specificity, and diagnostic efficiency. Overall morbidity and mortality in the United States ac- diagnostic efficiency was 84% or greater for all counting for 70% of deaths, nearly half of cerebral lesions studied. Sensitivity and specificity were palsy, and a substantial proportion of other disor- also high for the majority of diagnoses. Problem ders including chronic lung disease, mental retar- areas included relatively low sensitivity (Ͻ 85%) dation, and sensorineural impairment [20]. Infec- for advanced (stage 3) fetal inflammatory re- tion of the amniotic fluid is either the initiating sponse, fetal vessel thrombi, decidual plasma cells, event or the final common pathway leading to de- and chronic chorioamnionitis, and relatively low livery in more than a third of these patients [21, specificity (Ͻ 85%) for decidual plasma cells. The 22]. The term amniotic infection syndrome was second measure of individual performance was to originally coined to encompass the circumstances evaluate overall agreement using the unweighted in which microorganisms enter the normally ster- kappa (k) statistic. For this analysis, ␬-values were ile amniotic sac, the sequence of maternal and fetal interpreted as follows: 0.00–0.20, poor; 0.21–0.60, inflammatory reactions these organisms elicit, and fair/moderate; 0.61–1.00, substantial/near perfect the adverse consequences of infection and inflam- [18]. By these criteria, none of the diagnoses fell mation for the mother and fetus [1]. Pathologists into the poor category. Reproducibility for pres- can provide important information related to all ence of any maternal inflammatory response, pres- three components. In terms of predisposing condi- ence of any fetal inflammatory response, severe tions, pathologists contribute by recognizing bac- (grade 2) maternal inflammatory response (Fig. terial vaginosis on pap smear, detecting cervico- 1e), peripheral funisitis (Fig. 3a), acute villitis (Fig. vaginal group B streptococci by culture, and by the 3b), and acute intervillositis/intervillous abscesses diagnosis of subacute or chronic deciduitis (endo- (Fig. 3c) was substantial to near perfect. Reproduc- metritis) in the delivered placenta [20, 23, 24]. ibility for the remaining diagnoses—advanced With respect to the sequence of maternal and fetal

AMNIOTIC INFECTION SYNDROME 443 444 ..R R.W. DIEE AL ET EDLINE

Table 3. Final group survey results, round 2 Lesion prevalence for the 20 study Group Collective individual performance Interobserver . cases consensus (20 cases scored for 12 lesions) reproducibility Placental lesion reaction pattern A priori Group Individual range 6/6 5/6 4/65 Sensitivity Specificity Efficiency Kappa Maternal inflammatory response Any 0.70 0.70 (0.65–0.70) 14 0 0 100 94 98 0.93 Grade 2 (severe) 0.20 0.45 (0.25–0.55) 7 1 1 88 95 92 0.76 Stage 3 (advanced) 0.15 0.15 (0.15–0.30) 2 1 0 93 87 88 0.49 Chronic (subacute) 0.10 0.10 (0–0.20) 0 0 2 75 94 91 0.25 chorioamnionitis Fetal inflammatory response Any 0.65 0.70 (0.65–0.70) 12 1 1 96 100 98 0.90 Grade 2 (severe) 0.30 0.30 (0.25–0.55) 4 1 1 92 85 87 0.55 Stage 3 (advanced) 0.15 0.10 (0–0.20) 0 2 0 83 94 93 0.52 Fetal vessel thrombi 0.15 0.20 (0.15–0.35) 0 2 2 75 87 84 0.37 Other specific features Peripheral funisitis 0.10 0.10 (0.10–0.20) 2 0 0 100 98 98 0.84 Acute villitis 0.10 0.10 (0.05–0.10) 1 1 0 92 100 98 0.90 Acute intervillositis/abscesses 0.05 0.05 (0.05–0.15) 1 0 0 100 97 97 0.65 Decidual plasma cells 0.05 0.05 (0.05–0.15) 0 0 1 67 79 91 0.30 inflammation, histologic chorioamnionitis has re- ence of neutrophils in the walls of large fetal ves- peatedly been shown to be both sensitive and spe- sels in the chorionic plate and umbilical cord [1, cific for infection and is considered the gold stan- 29]. The presence of umbilical cord inflammation dard against which other clinical predictors of has been shown to be a risk factor for clinical infection should be measured [25–28]. Finally, as manifestations of the fetal inflammatory response described below, certain reaction patterns such as syndrome including intraventricular hemorrhage severity of inflammation, associated fetal vascular and CNS echolucencies in preterm infants [6, 31]. thrombosis, patterns suggestive of a specific caus- A greater association with complications in pre- ative organism, and estimation of the risk for neo- term infants may relate in part to an increased natal sepsis may identify infants at increased risk threshold for mounting fetal inflammatory re- for specific perinatal complications. sponses in the less mature fetus [32]. It has gener- Most attempts to subclassify the inflamma- ally been observed that chorionic plate vessels tory responses to amniotic fluid infection have and/or the umbilical vein are involved prior to the built upon the observations of Blanc who de- umbilical arteries. Support for this sequence has scribed what he believed to be the anatomic se- recently been published by two recent studies dem- quence of events [29]. In this scheme, stage 1 was onstrating increased fetal cytokine levels and in- defined by neutrophils in the subchorionic fibrin, creased morbidity in patients with inflammation in stage 2 by neutrophils in the chorionic plate, and one or both umbilical arteries compared to those stage 3 by neutrophils in the amnion. Van Hoeven with involvement of the vein alone [33, 34]. and colleagues later found that amnionic necrosis Keenan et al. also demonstrated the additional (necrotizing chorioamnionitis), a late complica- clinical significance of arterial inflammation by tion of inflammation in the amnion, was associ- showing that umbilical panvasculitis (veins plus ated with an increased risk of perinatal death and arteries) was an independent risk factor for neona- preterm delivery [9]. For this reason, they argued tal sepsis [4]. In view of this data, it seems impor- that this feature should be specified in the final tant to differentiate between involvement of the pathologic diagnosis. Mueller-Heubach and co- umbilical vein alone (phlebitis) and umbilical vas- workers also described associations of necrotizing culitis with arterial involvement. Use of the generic chorioamnionitis with preterm labor, premature term “funisitis” is not sufficiently specific and rupture of membranes, and decreased gestata- should be discouraged. tional age in very low birthweight infants [7]. Nararro and Blanc were also the first to de- Severity of the maternal inflammatory re- scribe a later stage of fetal inflammation, subacute sponse has been estimated in a couple of ways. keenan and colleagues in 1977 showed that sever- necrotizing funisitis [3]. In this lesion, fetal neu- ity, as determined by the formation of subchori- trophils migrate into the umbilical stroma (Whar- onic microabscesses, was associated with an in- ton’s jelly), arranging themselves in a circular are creased risk of presumed or documented fetal around the vessel. This pattern has been attributed sepsis [4]. Muller-Heubach and colleagues pro- to the precipitation of immune complexes formed posed an alternative numerical system in which Ͼ by microbial antigens diffusing in from the amni- 30 neutrophils in the upper half of the chorionic otic fluid and maternal antibodies diffusing out plate was considered severe [7]. However, severity from the umbilical vessels [5]. Such immune com- defined in the latter fashion has not been predictive plexes are chemotactic for fetal neutrophils. Mat- of clinically significant complications [10, 11, 30]. suda and coworkers later showed that infants with This numerical definition was also poorly repro- this lesion are at significantly increased risk for ducible in our study (round 1). Better reproducibil- chronic lung disease [35]. These observations de- ity and the clinical significance described by fine three temporal stages of fetal involvement: Keenan et al. [4] lead us to recommend this as a stage 1, neutrophils in chorionic and/or umbilical more useful criterion for severity. vein, stage 2, neutrophils in an umbilical artery, In Blanc’s original formulation, a fetal in- and stage 3, concentric arcs of degenerating neu- flammatory response was indicated by the pres- trophils in Wharton’s jelly.

AMNIOTIC INFECTION SYNDROME 445 Severity of fetal inflammation was found by tervillositis and intervillous abscess formation Spong and coworkers to be a risk factor for severe (Fig. 3c). Acute villitis in these cases is less exten- variable decelerations and increased circulating sive than the intervillositis and is presumed to rep- NRBC in very low birth weight infants [36]. Other resent secondary spread to villi. This pattern is studies by Redline et al. have shown that severe most commonly seen in infections caused by Lis- fetal inflammation is a risk factor for neurologic teria monocytogenes [38]. Campylobacter fetus and impairment in both very low birth weight and term a variety of rare infections, such as coccidiomyco- infants [11, 19]. While severity is more easily as- sis, psittacosis, tularemia, and brucellosis, may sessed and of better documented significance in also manifest some or all of these findings [39, 40]. chorionic plate vessels, umbilical vessels can also In the present study, the shared feature of acute be assessed in cases where the chorionic plate has villitis was easily recognized and agreed upon by not been sampled. Much of the effect of severe fetal virtually all observers (kappa ϭ 0.90). The addi- inflammation in the second group of studies cited tional finding of acute intervillositis/intervillous above was attributable to coexisting chorionic ves- abscesses was recognized by all observers, but was sel thrombosis. Reproducibility for the diagnosis occasionally overdiagnosed in cases of acute villitis of fetal vessel thrombi was somewhat disappoint- with occasional perivillous neutrophils leading to a ing in the present study, largely owing to ambigu- lower kappa value (0.65). ity in cases with small foci of endothelial fibrin The final two patterns, decidual plasma cells deposition. For diagnostic purposes, we recom- and chronic (subacute) chorioamnionitis, were mend close adherence to the classic criteria for the less reproducible and often missed in consensus- identification of true thrombi (glassy red-blue positive cases. Plasma cells in decidua, like those in staining, lamination, and adhesion to the vessel nonpregnant endometrium, are notoriously diffi- wall) when making this diagnosis (Fig. 2f). cult to detect. Overdiagnosis of cells lacking one or Peripheral funisitis, the formation of superfi- more of the classic morphologic features can also cial umbilical cord abscesses, was a highly repro- be a problem. Chronic (subacute) chorioamnioni- ducible finding (kappa ϭ 0.84). This pattern, when tis requires the observer to recognize an underly- diffusely distributed over the cord, is virtually pa- ing histiocytic infiltrate in membranes already suf- thognomonic for candida infection [5, 37]. When fused with neutrophils. It is possible that special the peripheral inflammatory response is focal and stains such as syndecan-1 for plasma cells and located near the placental cord insertion, the asso- CD68 for histiocytes could improve reproducibil- ciation with candida is less specific. Our study set ity. Plasma cells in particular, and B-lymphocytes contained one case of each type and both were in general, are always abnormal in the endometrial recognized by 100% of observers. Because of this cavity [41]. Their presence is an indicator of inap- overlap, a final diagnosis of candidal chorioamnio- propriate antigenic exposure. One source of anti- nitis requires demonstration of fungal hyphae in gen is bacterial subacute endometritis, which may tissue sections. lead to recurrent acute chorioamnionitis in subse- The presence of neutrophils in the villous quent pregnancies [20]. Uterine plasma cells in the stroma may be found in two distinct patterns. The absence of acute inflammation are also associated first pattern, acute villitis with a minimal intervil- with infertility, spontaneous abortion, and chronic lous component, is believed to reflect overwhelm- villitis [42–44]. We believe that the presence of ing fetal sepsis, often caused by gram negative plasma cells should always be noted in the final bacilli such as E. coli [17]. The histologic findings diagnosis [45]. Chronic chorioamnionitis, on the are acute capillaritis with emigration of fetal neu- other hand, is a heterogeneous entity that can ac- trophils to villous stroma. Clusters of neutrophils company either acute chorioamnionitis or chronic in acute villitis often accumulate between villous villitis [46, 47]. This lesion overlaps with a very stroma and the villous trophoblast basement mem- recently characterized lesion known as subacute brane (Fig. 3b). Bacteria are usually evident by chorioamnionitis [48]. In the latter report, pub- routine hematoxylin and eosin (H&E) stain. The lished after completion of our study, subacute cho- second pattern is acute villitis accompanied by in- rioamnionitis when combined with subacute ne-

446 R.W. REDLINE ET AL. croti