The association between intra‐amniotic infection and preterm labor and birth has been well established.1 The prevalence of microbial infection of amniotic fluid (AF), as detected by AF culture, has been reported to range from 4.2% to 21.6% in women who subsequently delivered preterm neonates. Intra‐amniotic infection is often subclinical with signs such as fever, uterine tenderness, foul‐smelling vaginal discharge, fetal tachycardia, and maternal leukocytosis, which usually occur late and are present in only a small portion (12.5%) of women with microbiologic evidence of infection.2 Neonates born to mothers with an intra‐amniotic infection also have a higher risk of both infectious and noninfectious complications than infants born to mothers with negative AF cultures.3
The need to promptly treat patients with a subclinical intra‐amniotic infection has led to considerable interest in developing rapid, yet highly sensitive and specific, assays for the identification of microbial invasion of the amniotic cavity. Examples of these tests include the AF Gram stain, white blood cell count, leukocyte esterase, glucose concentration, and interleukin‐6 assay. Of these tests, AF interleukin‐6 is probably the most accurate predictor of a positive AF culture, but it is limited by a low positive predictive value.4 The limited sensitivity and specificity of the other tests also hinder their clinical usefulness.
Locksmith et al reported that measuring matrix metalloproteinase‐9 in AF appeared to be reliable for diagnosing intra‐amniotic infection and portend preterm delivery regardless of microbiologic confirmation.5 Matrix metalloproteinases are an important family of enzymes that use zinc‐dependent catalysis to degrade extracellular matrix components. They are constitutively produced by reproductive tissues, and gene expression of various matalloproteinases has been noted to change during different stages of parturition.6 They are involved in conceptus implantation7 and uterine involution after delivery.8 Matrix metalloproteinase‐9 is believed to be a terminal enzyme in remodeling the extracellular matrix and, in contrast to many metalloproteinases, the production and release of matrix metalloproteinase‐9 are induced under specific conditions.9
Our primary objective was to assess the accuracy of AF matrix metalloproteinase‐9 analysis for detecting intra‐amniotic infection compared with AF interleukin‐6 in women with preterm contractions, preterm labor, preterm premature rupture of membranes (preterm PROM), or suspected subclinical intra‐amniotic infection.
MATERIALS AND METHODS
From July 1, 1997, through May 31, 2000, we performed transabdominal amniocentesis in 84 women between 22 and 35 weeks' gestation who presented to the labor and delivery unit at Yale‐New Haven Hospital with preterm contraction, preterm labor, preterm PROM, or clinical suspicion of intra‐amniotic infection. The institutional Human Investigation Committee approved this study. In addition to giving written informed consent for the procedure, all subjects signed a separate institutional consent form to allow analysis of their AF in this study. Preterm contraction and labor were defined as the presence of regular uterine contractions, at least four per hour. Cervical dilatation of at least 2 cm at presentation was documented in cases with preterm labor. Preterm PROM was documented clinically and with positive nitrazine and fern tests. Suspicion of intra‐amniotic infection was based on maternal fever, fetal tachycardia, or uterine tenderness. We excluded women with antibiotic use within 7 days of presentation, poly‐hydramnios, multiple gestations, fetal anomalies, or maternal diabetes.
Our purpose for performing amniocentesis was to determine if intra‐amniotic infection was present. After collection, AF samples were transported to the laboratory for Gram stain, leukocyte and glucose concentration analyses, and culture for aerobes, anaerobes, and mycoplasmas. From each sample, we saved 3–5 mL of amniotic fluid, centrifuged it at 3000 revolutions per minute for 15 minutes, and stored it in 0.5 mL of aliquots at −80C until measurements of AF matrix metalloproteinase‐9 and interleukin‐6 were performed. Results of AF culture, Gram stain, glucose, and leukocytes were used to make management decisions, but levels of matrix metalloproteinase‐9 and interleukin‐6 were not.
We used a commercial, two‐site enzyme‐linked immunosorbent sandwich assay (R&D Systems, Minneapolis, MN) to quantitate total matrix metalloproteinase‐9 and interleukin‐6 levels in the AF samples. The assay for matrix metalloproteinase‐9 measures concentrations most reliably in the 0.2–20 ng/mL range with interassay and intra‐assay coefficients of variance of 7.5% and 1.9%, respectively. All samples were diluted at 1:10 for the preliminary analysis. Samples that contained greater than 200 ng/mL were diluted at 1:50 for the final analysis. The assay for interleukin‐6 measures reliably in the range of 0.7 pg/mL to 0.3 ng/mL with interassay and intra‐assay coefficients of variance of 4.5% and 2.6%, respectively. Each sample was analyzed in duplicate at a dilution of 1:20. Amniotic fluid matrix metalloproteinase‐9 and interleukin‐6 concentrations were determined by interpolation from constructed curves from known standards of matrix metalloproteinase‐9 and interleukin‐6.
A required sample size of 45 was estimated, assuming that a valuable test predicts the presence or absence of infection for 80% of measurements, compared with 50% accuracy for chance alone. The confidence coefficient was 95%, and the desired power level was 80%.5 We constructed receiver operating characteristic (ROC) curves for each of AF matrix metalloproteinase‐9 and interleukin‐6 levels. The ROC curves for matrix metal‐loproteinase‐9 were tested for significance against those of null hypothesis and interleukin‐6 by comparing the areas under the curves.10 The Mann‐Whitney U test was used to compare AF matrix metalloproteinase‐9 and interleukin‐6 concentrations in infected versus noninfected subjects. Spearman rank correlation was used for correlation analysis between AF matrix metalloproteinase‐9, interleukin‐6, leukocytes, and glucose concentration. Sensitivity, specificity, and positive and negative predictive values of AF matrix metalloproteinase‐9 and interleukin‐6 for diagnosing intra‐amniotic infection were calculated with 95% confidence intervals using culture as the criterion standard. Results were expressed as medians with ranges.
Table 1 presents the demographic characteristics of AF culture‐positive and culture‐negative women. There were no statistical differences between the two groups in maternal age, gestational age, parity, race, and cervical dilatation at the time of the amniocentesis. Women with preterm PROM were more likely to have a positive AF culture compared with women with intact membranes. The median level and ranges of AF matrix metalloproteinase‐9 in women with culture‐confirmed intra‐amniotic infection were significantly higher than those of women with negative AF cultures, 134.5 ng/mL (0.0–541.9) compared with 0.0 ng/mL (0.0–13.6), P < .001 (Figure 1). Amniotic fluid interleukin‐6 levels were also significantly higher in women with intra‐amniotic infection than in women without infection, 81.1 ng/mL (0.0–91.7) versus 1.4 ng/mL (0.0–28.9), P < .001 (Figure 2). Amniotic fluid levels of matrix metalloproteinase‐9 and interleukin‐6 were significantly correlated (r = 0.813, P < .001). These levels correlated directly with AF leukocytes (r = 0.818, P < .001 and r = 0.712, P < .001) and correlated inversely (r = −0.598, P < .001 and r = −0.510, P < .001) with AF glucose concentration, respectively.
Figure 3 shows the ROC curves for AF matrix metalloproteinase‐9 and interleukin‐6. Using a cutoff level of 13.6 ng/mL for matrix metalloproteinase‐9 and 11.4 ng/mL for interleukin‐6 seemed to be the most appropriate for predicting the presence or absence of infection. Amniotic fluid matrix metalloproteinase‐9 had better sensitivity, 77% compared with 73%, and specificity, 100% compared with 79%. The positive predictive values were 100% versus 61%. The negative predictive values were 90% versus 86%. Combining AF matrix metalloproteinase‐9 and interleukin‐6 slightly improved the sensitivity (77% to 80%) and negative predictive values (90% to 92%) in diagnosing intra‐amniotic infection (Figure 4).
In this study, AF concentrations of matrix metalloproteinase‐9 and interleukin‐6 were significantly elevated in women with intra‐amniotic infection. The distribution of AF matrix metalloproteinase‐9 and interleukin‐6 values was remarkable. In 58 women with negative AF cultures, matrix metalloproteinase‐9 was undetectable in 42, whereas only two women had undetectable interleukin‐6. However, in 26 women with positive AF cultures, 22 demonstrated matrix metalloproteinase‐9 concentrations ranging from 30.1 to 541.9 ng/mL; all of the women demonstrated an interleukin‐6 range from 0.7 to 91.7 ng/mL. The performance statistic of AF interleukin‐6 analysis was similar to that of matrix metalloproteinase‐9 concentration and was consistent with previous studies on AF interleukin‐6.4 However, we found higher specificity and positive predictive values for AF matrix metalloproteinase‐9 than for AF interleukin‐6. Matrix metalloproteinase‐9 was virtually undetectable in culture‐negative AF samples.
Of the known metalloproteinases, matrix metalloproteinase‐9 is an enzyme that can be induced under specific conditions. Because it is an inducible enzyme, matrix metalloproteinase‐9 has drawn interest as a marker for the prediction of preterm delivery11 and the diagnosis of subclinical intrauterine infection.5 Fortunato et al found matrix metalloproteinase‐9 in the fetal membranes of women with intra‐amniotic infection, and they also showed that matrix metalloproteinase‐9 gene expression could be induced in cultured membranes by exposure to lipopolysaccharide or peptidoglycan polysaccharide.12 Others have found elevated matrix metalloproteinase‐9 concentration in AF13 and plasma11 in women with preterm PROM. In this study, women with preterm PROM were more likely to have a positive AF culture (59%) than women with intact membranes (24%). However, we noted elevated AF matrix metalloproteinase‐9 levels in cases of preterm PROM only when intra‐amniotic infection was confirmed. Levels of AF matrix metalloproteinase‐9 were virtually undetectable in the absence of intra‐amniotic infection whether the membranes were ruptured or not. This finding may explain the importance of microbial invasion of the amniotic cavity for inducing the production and release of matrix metallo‐proteinase‐9, and it was consistent with other in vivo and in vitro studies.12,14
Interleukin‐6 has been regarded as an important mediator in the host response to intra‐amniotic infection. Our data showed significant correlation between AF levels of matrix metalloproteinase‐9 and interleukin‐6. There was also a significant positive correlation between AF matrix metalloproteinase‐9 and interleukin‐6 concentrations and AF leukocyte count and an inverse correlation with AF glucose concentration. This correlation suggests a common underlying process for the elevation of AF matrix metalloproteinase‐9 and interleukin‐6, which occurs coincidental with the invasion of the amniotic cavity by microbes. However, we found a higher specificity and positive predictive value for AF matrix metalloproteinase‐9 compared with AF interleukin‐6 despite the similarity in their ROC curves. The reason for that difference is not clear, but this can be explained by the nonspecific release of interleukin‐6 from stimulated immune cells present in amnion and chorion/decidua and AF in response to other and noninfectious inflammatory causes.15
We did not analyze preterm delivery as an outcome because many of the women who delivered before term had labor induced or had cesarean delivery based on the clinical impression that maternal or fetal benefits outweighed those of pregnancy continuation. We used the cutoff values of 13.6 ng/mL for AF matrix metalloproteinase‐9 and 11.4 ng/mL for AF interleukin‐6. These values were determined from the constructed ROC curves. The cutoff value we used for AF interleukin‐6 was consistent with previous reports by Romero et al.4 In this study, the two groups were similar except for the membrane status. However, we do not believe that preterm PROM was a confounding factor because AF matrix metalloproteinase‐9 was undetectable in women with preterm PROM with negative AF culture. Accordingly, reasonable estimates of the accuracy of AF matrix metalloproteinase‐9 for prediction of intra‐amniotic infection can be made.
Our findings suggest that analysis of AF matrix met‐alloproteinase‐9 concentration has good sensitivity, specificity, and positive and negative predictive values in diagnosing intra‐amniotic infection. We suggest that developing a colorimetric test to determine the status of matrix metalloproteinase‐9 in AF will be a useful and rapid test in predicting intra‐amniotic infection with high specificity and predictive value.
1. Romero R, Mazor M. Infection and preterm labor. Clin Obstet Gynecol 1988;31:553–84.
2. Romero R, Sirtori M, Oyarzun E, Avila C, Mazor M, Callahan R, et al. Infection and labor. V. Prevalence, microbiology, and clinical significance of intra-amniotic infection in women with preterm labor and intact membranes. Am J Obstet Gynecol 1989;161:817–24.
3. Watts DH, Krohn MA, Hillier SL, Eschenbach DA. The association of occult amniotic fluid infection with gestational age and neonatal outcome among women in preterm labor. Obstet Gynecol 1992;79:351–7.
4. Romero R, Yoon BH, Mazor M, Gomez R, Diamond MP, Kenney JS, et al. The diagnostic and prognostic value of amniotic fluid blood cell count, glucose, interleukin-6, and Gram stain in patients with preterm labor and intact membranes. Am J Obstet Gynecol 1993;169:805–16.
5. Locksmith GJ, Clark P, Duff P, Schultz GS. Amniotic fluid matrix metalloproteinase-9 levels in women with preterm labor and suspected intra-amniotic infection. Obstet Gynecol 1999;94:1–6.
6. Bryant-Greenwood GD, Yamamoto SY. Control of peripheral collagenolysis in the human chorion-decidua. Am J Obstet Gynecol 1995;172:63–70.
7. Graham CH, Lala PK. Mechanism of control of trophoblast invasion in situ. J Cell Physiol 1991;148:228–34.
8. Sellers A, Woessner JF. The extraction of a neutral metal-loproteinase from the involuting rat uterus, and its action on cartilage proteoglycan. Biochem J 1980;189:521–31.
9. Opdenakker G. On the roles of extracellular matrix remodeling by gelatinase B. Verh K Acad Geneeskd Belg 1997;59:489–514.
10. Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating curves derived from the same cases. Radiology 1983;148:839–43.
11. Tu FF, Goldenberg RL, Tamura T, Drews M, Zucker SJ, Voss HF. Prenatal plasma matrix metalloproteinase-9 levels to predict spontaneous preterm birth. Obstet Gynecol 1998;179:1248–53.
12. Fortunato SJ, Menon R, Lombardi SJ. Collagenolytic enzymes (gelatinases) and their inhibitors in human amniochorionic membrane. Am J Obstet Gynecol 1997; 177:731–41.
13. Vadillo-Ortega F, Hernandez A, Gonzalez-Avila G, Bermejo L, Iwata K, Strauss JF III. Increased matrix metallo-proteinase activity and reduced tissue inhibitor of metallo-proteinases levels in amniotic fluids from pregnancies complicated by premature rupture of membranes. Am J Obstet Gynecol 1996;174:1371–6.
14. Buhimschi IA, Kramer WB, Buhimschi CS, Thompson LP, Weiner CP. Reduction-oxidation (redox) state regulation of matrix metalloproteinase activity in human fetal membranes. Am J Obstet Gynecol 2000;182:458–64.
© 2002 by The American College of Obstetricians and Gynecologists.
15. Witkin S, McGregor JA. Infection-induced activation of cell-mediated immunity: Possible mechanism for preterm birth. Clin Obstet Gynecol 1991;34:112–21.