Journal of Perinatology (2008) 28, 97–101 r 2008 Nature Publishing Group All rights reserved. 0743-8346/08 $30 www.nature.com/jp ORIGINAL ARTICLE in preterm PROM: effects on amnion and cord histology

A Locatelli1, M Andreani1, A Ghidini1, M Verderio1, A Pizzardi1, P Vergani1 and CM Salafia2 1Department of and Gynecology, University of Milano-Bicocca, Monza, Italy and 2Department of Epidemiology, Mailman School of Public Health, Columbia University, New York, NY, USA

and lower incidence of pulmonary hypoplasia compared with Objective: To investigate the effects of transabdominal amnioinfusion expectant management.1–6 A recent randomized study7 confirmed (TA) on the histology of amnion (A) and umbilical cord (UC). a significant prolongation of in cases managed with Study Design: From a cohort of 56 singleton with amnioinfusion. premature rupture of membranes (PROM) at p24.6 weeks, we excluded However, the infusion of normal saline into the amniotic fluid those who did not develop severe (n ¼ 12). Of the 44 is not fully a physiologic replacement of amniotic fluid, with its patients, 29 consented to TA and 15 declined the procedure. TA was wealth of growth factors and nutritive components. It is thus performed with normal saline solution. conceivable that exposure of the membranes to normal saline could have some deleterious effects on the integrity or histologic Result: Women who consented to TA underwent a median of three features of the amnion and umbilical epithelia. Because the effects procedures had a at PROM of 18.7 weeks and at delivery of of amnioinfusion on placental histology have not been addressed 26.1 weeks, with a latency of 50 days. Compared with subjects who so far, we have investigated whether amnioinfusion in women with declined the procedure, TA had a beneficial effect on clinical variables, second trimester PROM and severe oligohydramnios has an effect mediated primarily by a longer latency. TA was not associated with on the histopathologic integrity of the epithelial surfaces of the identifiable effects on the histologic features of A or UC. amniochorionic membranes and umbilical cord and the Conclusion: A and UC histology is not significantly affected by exposure underlying connective tissue. Since acute inflammation and to saline solution even for prolonged periods of time. uteroplacental ischemia, two factors implicated in the causation of Journal of Perinatology (2008) 28, 97–101; doi:10.1038/sj.jp.7211876; PROM, may also affect the integrity of the epithelial surfaces, in published online 29 November 2007 the analysis we have taken into account the presence and severity of these two processes at histology. Keywords: amnioinfusion; placental histology; preterm premature rupture of membranes Methods Introduction We have accessed a database of consecutive singleton pregnancies When premature rupture of membranes (PROM) occurs in the with preterm PROM at <25.0 weeks lasting >4 days admitted at early second trimester and is associated with severe our institution between January 1998 and December 2004. oligohydramnios, standard approaches have traditionally included Diagnosis of rupture of the membranes was made by observation of termination of pregnancy or expectant management. However, persistent vaginal pooling on sterile speculum examination and alternative strategies have been explored for treatment and repair of serial ultrasonographic observations. Gestational age (GA) was membrane rupture, such as use of serial amnioinfusions, established by menstrual history and confirmed by intracervical tissue sealants, gelatin sponge embolization and ultrasonographic examination before 20 weeks. amniopatch.1–3 Amnioinfusion is the only approach that has been Included in the present study were those with (1) evaluated by independent investigators in nonrandomized series oligohydramnios, defined as deepest pocket of amniotic fluid and it has been shown to lead to higher perinatal survival rates p2 cm and persisting for 4 days or longer, (2) no major fetal anomalies and (3) placental slides available for histopathological Correspondence: Dr A Locatelli, Clinica Ostetrica e Ginecologica, University of Milano- examination. Bicocca, San Gerardo Hospital, Monza (MI), Italy. All patients were given hospital bed rest during the first week E-mail: [email protected] Received 17 April 2007; revised 13 September 2007; accepted 9 October 2007; published online and received an initial 7-day course of antibiotic prophylaxis 29 November 2007 (sulbactam–ampicillin 3 g every 8 h or amoxicillin–clavulanic Amnioinfusion in preterm PROM A Locatelli et al 98 acid 2 g every 8 h i.v.). Digital examination was avoided until of connective tissue. In the umbilical cord, we analyzed epithelial active labor. After the initial 7-day hospital stay, the patient was or Wharton’s jelly necrosis, PMN number, distance and integrity in discharged home on bed rest if GA was <25 weeks; vaginal cultures the cord vessels and in the Wharton’s jelly. For the purpose of the were obtained every 2 weeks, and positive results for potentially analysis, epithelial integrity was graded in a semiquantitative scale pathogenic bacteria were treated. After 25 weeks’ gestation, women from 1 to 4 based on the distribution of the nuclei, which normally were admitted to the hospital until delivery and they received at are distributed at regular intervals reflecting the size of each least one course of corticosteroid therapy (betamethasone 12 mg epithelial cell. Scattered nuclear dropout with retained cytoplasm i.m. daily  two doses). In the presence of preterm labor, (considered a sign of cell death with loss of nuclear staining but tocolytic treatment (i.v. ritodrine) was administrated in the absence undissolved cytoplasm) was graded 1. Grade 4 was complete loss of of clinical signs of chorioamnionitis or abruptio placentae. Fetal all epithelium on the amnion basement membrane, or a naked heart rate was monitored daily, and biophysical profile was basement membrane without viable cells (with nuclei and performed twice a week. Amniotic fluid volume was assessed cytoplasm), or dead ‘ghost’ cells (no nuclear staining but preserved sonographically weekly on the outpatients, and every other day cytoplasm/cell shape). from admission until delivery. Chorionic and basal plate were evaluated in particular for Consenting women with persistent (>4 days) oligohydramnios evidence of acute inflammation or uteroplacental ischemia, the (maximum cord-free pocket of amniotic fluid <2 cm) were offered two principal pathologic processes that have been implicated in the serial amnioinfusions. Under sonographic guidance a 20-gauge needle causation of PROM and that may affect histologic features of was inserted transabdominally into the amniotic cavity. Transplacental amnion and umbilical cord. Histologic evidence of acute passage was avoided whenever possible. We targeted either a small inflammation was established by the number of neutrophils, their pocket of residual fluid or loops of cord between the fetal limbs, using integrity and the extent of tissue invasion in amnion, color flow mapping to avoid puncturing the umbilical cord. The choriodecidua and chorionic plate. Evidence of placental damage correct positioning of the needle was checked either by aspiration of a related to uteroplacental vascular pathology included abnormalities little amount of amniotic fluid or by ultrasonographic visualization of of uteroplacental vessels, fibrinoid necrosis and atherosis, a free dispersion of normal saline solution within the amniotic cavity. abruption, villous infarcts, terminal villous fibrosis, increased The volume infused of normal saline solution was aimed at restoring a syncytiotrophoblast knotting, cytotrophoblast (‘X-cell’) proliferation normal amount of amniotic fluid; we noted that infusion of 10 ml per and villous hypovascularity. For the purpose of the analysis, the week of GA was usually sufficient for this purpose. In no case did the two pathologic processes (acute inflammation and uteroplacental procedure last longer than 15 min. If oligohydramnios recurred and ischemia) were evaluated in a stochastic way (as present or absent) persisted for >4 days, the procedure was repeated. Amnioinfusion was as well as in a semiquantitative way, by averaging the scores of deemed successful if the mean deepest pocket of fluid during the acute inflammatory and uteroplacental ischemia lesions. latency period was >2 cm. Delivery was expedited in the presence of clinical Statistical analysis chorioamnionitis; fetal tachycardia with diminished variability, Histologic and clinical characteristics of women who underwent recurrent late or severe variable decelerations, or a biophysical amnioinfusion were compared with those of women with similar profile score <6; abruptio placentae; or GA >34 weeks. severity of oligohydramnios who did not undergo the procedure, Placental pathology sampling and examination were performed using one-way ANOVA and Fisher’s exact test; a P<0.05 was following a standard protocol. For each case, at least two samples considered significant. Linear regression analysis was used to assess of the following were obtained: extra-placental membranes near the independent effect of amnioinfusion on epithelial integrity of the point of rupture and near the placenta; chorionic plate amnion and umbilical cord after controlling for acute including the whole wall, distant from the marginal edge; decidua inflammation and uteroplacental vascular pathology in chorionic basalis; and umbilical cord (5 cm from the chorionic plate and at and basal plate, defined as above. Moreover, to assess whether the proximal fetal end). Additional tissue samples were obtained if persistence of infused normal saline had an effect on epithelial indicated by the presence of gross lesions. The slides (hematoxylin integrity, we selected cases with oligohydramnios undergoing serial and eosin staining) were reviewed by an independent pathologist amnioinfusions and we compared the histologic features between (CMS) who was blinded to the clinical data except GA at delivery those that retained the infusate (maintaining an average pocket of and knowledge that in all cases amnioinfusion had been offered for amniotic fluid >2 cm during latency) and those that did not. persistent oligohydramnios. The histopathologic slides were reviewed with particular attention to epithelial integrity of the amnion and umbilical cord and underlying connective tissues. In Results the amnion, we recorded epithelial integrity, number and integrity From a cohort of 56 consecutive cases of PROM at <25.0 weeks of polymorphonuclear leukocytes (PMNs), cellularity and necrosis with placental histology available, we excluded 12 cases because

Journal of Perinatology Amnioinfusion in preterm PROM A Locatelli et al 99 they never developed oligohydramnios. In the remaining 44 cases, in 39/44 cases (89%); amnioinfusion was not associated with any which constituted the study population, the mean±s.d. of the significant change in the histologic features analyzed, with the deepest pocket of amniotic fluid was 1.4±0.8 cm. Fifteen out of 44 exception of lower grade of epithelial necrosis in the umbilical cord women (34%) declined amnioinfusion and 29 (66%) consented, of those who underwent amnioinfusions. with a median of three procedures (range 1 to 7). Median GA at Histologic evidence of acute placental inflammation was present PROM was 18.7±3.2 weeks, and GA at delivery was 26.1±5.0 in 22/29 or 76% of the cases in the amnioinfusion group vs 15/15 weeks, with a latency period of 50±38 days. or 100% of the non-amnioinfused cases (P ¼ 0.07), with an Table 1 displays the clinical characteristics of the study average acute inflammation score of 0.5±0.45 and 0.9±1.3, population in relation to whether amnioinfusion was performed or respectively (P ¼ 0.025). Our study had adequate statistical power not. On average, women who consented to the procedure had lower to detect a difference of 26% of acute placental inflammation GA at membrane rupture, but significantly longer latency period, between the two groups; in other words, 51 total women would be resulting in similar GA at delivery as women who declined the required for the observed difference to achieve statistical procedure. significance (b ¼ 0.20, a ¼ 0.05). Histologic evidence of Table 2 displays the histologic characteristics of amnion, uteroplacental ischemia was present in 16/29 or 54% of the cases umbilical cord, and underlying connective tissue in relation to in the former group vs 3/15 or 20% in the latter group amnioinfusion (Figures 1 and 2). Amnion histology was available (P ¼ 0.052), with an average score of 0.3±0.7 and 0.06±0.3, respectively (P ¼ 0.005). Linear regression analysis demonstrated that amnioinfusion was not significantly associated with scores of epithelial integrity after Table 1 Characteristics of women with oligohydramnios due to pPROM in taking into consideration GA at PROM, mean amniotic fluid pocket relation to amnioinfusion during latency, duration of latency interval, histologic evidence of Amnioinfusion, Amnioinfusion, P-value yes (n ¼ 29) no (n ¼ 15)

Nulliparity 18 (62%) 9 (60%) 0.84 Steroid administration 29 (100%) 15 (100%) 0.92 Gestational age at membrane 17.6±2.6 21.0±3.0 <0.001 rupture (weeks) Latency period (days) 62.8±38.3 27.5±25.0 0.002 Gestational age at delivery (weeks) 26.7±5.1 25.0±5.0 0.29 Median amniotic fluid pocket (mm) 17.1±7.8 9.5±7.4 0.003 Birth weight (g) 978±631 898±616 0.69 Neonatal survival 14 (48%) 8 (53%) 0.76

Number (%) or mean±s.d. Figure 1 Hematoxylin and eosin stained amnion showing health amnion epithelium and scattered necrotic amnion epithelial cells with subjacent normal dense amnion connective tissue (40 Â ). Table 2 Histologic characteristics of amnion and cord among cases with oligohydramnios in relation to amnioinfusion

Lesion score (0–4) Amnioinfusion, Amnioinfusion, P-value yes (n ¼ 29) no (n ¼ 15)

Minimum epithelial integrity 1.79±1.76 2.20±1.699 0.48 Maximum epithelial integrity 2.93±1.377 2.87±1.727 0.91 PMN number (minimum) 0.25±0.897 0.07±.258 0.45 PMN number (maximum) 0.85±1.312 1.33±1.676 0.32 PMN integrity (0–4) 1.16±1.668 1.73±1.944 0.34 Connective tissue cellularity 1.83±1.274 2.43±1.015 0.17 Connective tissue necrosis 1.65±1.770 1.33±1.234 0.54 Cord PMN number 0.39±0.84 0.35±0.67 0.87 Epithelial cord necrosis 0.4±0.86 1.17±1.43 0.04 Figure 2 Hematoxylin and eosin stained amnion showing complete loss of Abbreviation: PMN, polymorphonuclear leukocytes. amnion epithelium with subjacent marked loss of normal connective tissue Mean±s.d. composition with reduced collagen content and edema (40 Â ).

Journal of Perinatology Amnioinfusion in preterm PROM A Locatelli et al 100

Table 3 Characteristics of the study population of cases with oligohydramnios lower connective tissue necrosis and lower PMN integrity. The who underwent amnioinfusion in relation to whether the procedure was successful association between retained infusate and PMN integrity (P ¼ 0.8) or not or connective tissue necrosis (P ¼ 0.26) lost significance after Amnioinfused Amnioinfusion P-value controlling for latency period, histologic evidence of acute solution retained solution lost inflammation and uteroplacental ischemia. (n ¼ 10) (n ¼ 19) Comment Nulliparity 9 (90%) 9 (47%) 0.04 We have found that amnioinfusion does not exert an adverse effect Number of procedures 3 (1–7) 1 (1–7) 0.21 (median, range) on the histopatologic features of the structures in contact with Steroid administration 7 (70%) 7 (37%) 0.12 the amniotic cavity. In particular, epithelial integrity of amnion Gestational age at membrane 17.6±2.9 17.6±2.6 0.97 and umbilical cord, which may be more sensitive to changes rupture (weeks) in the composition of the fluid in the amniotic cavity, was not Latency period (days) 88.4±38.5 47.2±29.4 0.003 significantly different between cases that did or did not undergo Gestational age at delivery 30.4±4.6 24.5±4.0 0.001 amnionfusion after taking into consideration potential (weeks) confounders. Moreover, we have observed a lack of effect of Birth weight (g) 1513±603 651±379 0.0001 prolonged retention of the infused solution on epithelial integrity. Pulmonary hypoplasia 2 (20%) 11 (58%) 0.11 Although the degree of confidence of our findings is tempered by the Neonatal survival 8 (80%) 6 (32%) 0.02 limited statistical power due to the inevitably small number of cases, Number (%) or mean±s.d. the findings provide reassuring evidence that amnioinfusion per se,or prolonged exposure of the epithelium of the amnion to the infused solution (normal saline), does not damage epithelial cells. Table 4 Histologic characteristics of cases with oligohydramnios who underwent Our study was underpowered to show a significant difference in amnioinfusion in relation to whether infused solution was retained (successful rates of acute placental inflammation between the two groups. amnioinfusion) or not Moreover, the observed 0.4 point difference in placental acute Lesion score (0–4) Amnioinfused Amnioinfused P-value inflammatory score between the two groups, although statistically solution retained solution significant, is probably clinically negligible given that the score (n ¼ 10) lost (n ¼ 19) ranges from 0 to 4. Although a difference was present in uteroplacental ischemia between the group undergoing Epithelial integrity (minimum) 0.7±1.1 2.0±1.8 0.09 amnioinfusion and the one that did not, the interpretation is Epithelial integrity (maximum) 2.2±1.4 3.0±1.5 0.24 PMN number (minimum) 0±0 0.3±0.9 0.43 hampered by GA at delivery, an important confounder that was PMN number (maximum) 0.2±0.4 1.0±1.4 0.14 significantly different between the two groups. PMN integrity 0.1±0.2 1.6±1.8 0.04 We were unable to provide insights into the mechanisms and Connective tissue cellularity 2.6±1.5 1.3±2.7 0.23 processes responsible for rapid loss of infused solution leading to Connective tissue necrosis 0.3±0.5 2.0±1.8 0.019 failure of serial amnioinfusions to maintain an adequate pocket of Cord PMN number 0 0.56±0.97 0.17 amniotic fluid pocket during latency. Intrauterine inflammation Epithelial cord necrosis 0.66±0.81 1.33±1.58 0.34 has been invoked to play a causal role in the failure of retention of Abbreviation: PMN, polymorphonuclear leukocytes. the amnioinfused solution, as it may alter the dynamics of 8 Mean±s.d. amniotic fluid production and reabsorption by the amnion, thus reducing the probability of success of the procedure. Moreover, intrauterine infection may cause an increase in the resting tone of acute inflammation or uteroplacental ischemia (P ¼ 0.76 and the myometrium, leading to an accelerated loss of the infused P ¼ 0.12, respectively). solution. Alternatively, characteristics of membrane rupture (for Table 3 displays the characteristics of the study population example, size or location of the defect in relation to the internal among women with oligohydramnios who underwent serial cervical os) may be responsible for a more prompt loss of the amnioinfusions, in relation to whether the infusate was retained infused solution.9 (maintaining an average pocket of amniotic fluid >2 cm during We confirm previous reports that amnioinfusion is latency) or not. Retained fluid was associated with longer latency, associated with prolonged latency period and thus with improved greater GA at delivery, birth weight and survival. Table 4 shows the perinatal outcome. In light of the negligible effects of association between retained fluid and histologic features of amnioinfusion on histologic placental characteristics, the amnion, cord and underlying connective tissues. Retention of the procedure should be considered for cases of extremely preterm infused solution did not affect epithelial integrity, but it resulted in PROM with severe oligohydramnios. Our study was limited

Journal of Perinatology Amnioinfusion in preterm PROM A Locatelli et al 101 by its non-randomized design. Multicenter prospective randomized 4 Hadi HA, Hodson CA, Strickland D. Premature rupture of the membranes between trials are needed to confirm the results of the cohort series 20 and 25 week’s gestation: role of amniotic fluid in perinatal outcome. Am J Obstet published so far. Gynecol 1994; 170: 1139–1144. 5 De Santis M, Scavo M, Noia G, Masini L, Piersigilli F, Romagnoli C et al. Transabdominal amnioinfusion treatment of severe oligohydramnios in preterm premature rupture of membranes at less than 26 gestational weeks. Fetal Diagn Ther 2003; 18:412–417. References 6 Vergani P, Locatelli A, Strobelt N, Mariani S, Cavallone M, Arosio P et al. Amnioinfusion 1 Locatelli A, Ghidini A, Verderio M, Strobelt N, Pezzullo J, Andreani M et al. Predictors for prevention of pulmonary hypoplasia in second-trimester rupture of membranes. of perinatal survival in a cohort of pregnancies with severe oligohydramnios Am J Perinatol 1997; 14: 325–329. due to premature rupture of membranes at less than 26 weeks managed with 7 Tranquilli AL, Giannubilo SR, Bazzeccheri V, Scagnoli C. Transabdominal serial amnioinfusion. Eur J Obstet Gynecol Reprod Biol 2006; 128(1–2): amnioinfusion in preterm premature rupture of membranes: a randomised controlled 97–102. trial. BJOG 2005; 112:1–5. 2 Ogunyemi D, Thompson W. A case controlled study of serial transabdominal amnioinfusions 8 Ross MG, Brace RA, NIH Workshop participants. National Institute of Child Health and in the management of second trimester oligohydramnios due to premature rupture of Development Conference summary: amniotic fluid biologyFbasic and clinical aspects. membranes. Eur J Obstet Gynecol Reprod Biol 2002; 102: 167–172. J Matern Fetal Med 2001; 10: 2–19. 3 Vergani P, Ghidini A, Locatelli A, Cavallone M, Ciarla I, Cappellini A et al. Risk factors 9 Devlieger R, Millar LK, Bryant-Greenwood G, Lewi L, Deprest JA. Fetal membrane healing for pulmonary hypoplasia in second trimester premature rupture of membranes. Am J after spontaneous and iatrogenic membrane rupture: a review of current evidence. Obstet Gynecol 1994; 170: 1359–1364. Am J Obstet Gynecol 2006; 195(6): 1512–1520.

Journal of Perinatology