Amniotic Fluid Glucose Concentration: A Marker for Infection in Preterm Labor and Preterm Premature Rupture of Membranes

Amniotic fluid Gram stain and culture have been utilized as laboratory tests of microbial invasion of the amniotic cavity. The Gram stain of amniotic fluid has a low sensitivity in the detection of clinical infection or microbial invasion of the amniotic cavity, and amniotic fluid culture results are not immediately available for management decisions. Glucose concentration is used to diagnose infection in other sites such as cerebrospinal fluid. Objective: The purpose of this study was to evaluate the usefulness of amniotic fluid glucose concentration in detecting microbial invasion of the amniotic cavity associated with preterm labor and preterm premature rupture of membranes. Methods: Amniocentesis was performed in 60 women with preterm labor and/or preterm premature rupture of membranes. Gram stain and culture for Mycoplasma hominis, Ureaplasma urealyticum, aerobic, and anaerobic bacteria were performed. Subjects were studied prospectively for the development of positive amniotic fluid cultures and the development of clinical chorioamnionitis. Results: The diagnosis of clinical chorioamnionitis was made in 25% (15/60) of women entered into the study. Low amniotic fluid glucose concentration Was considered < 15 mg/dl. The sensitivity, specificity, and positive predictive value of low amniotic, fluid glucose concentration to predict clinical chorioamnionitis were 73.3%, 88.1%, and 68.8% respectively, while positive amniotic fluid culture, hada sensitivity of 43.8%, specificity of 79.5%, and positive predictive value of 43.8%. Conclusions: Amniotic fluid glucose concentration was more sensitive in predicting chorioamnionitis than either Gram stain or culture. Amniotic fluid glucose concentration was better in predicting clinical chorioamnionitis than predicting positive amniotic fluid culture results. Gestational age-dependent normal ranges and pathologic conditions that may alter amniotic fluid glucose concentrations should be considered when interpreting amniotic fluid glucose values to diagnose microbial invasion of the amniotic cavity.

KEY WORDS Chorioamnionitis, glucose, intra-amniotic infection nfection may be a significant etiologic factor in the development of preterm labor (PTL) and preterm premature rupture of" membranes (PPROM). Amniocentesis is sometimes employed to obtain amniotic fluid for analysis from patients presenting with PTL or PPROM in order to ex-clude infection as an underlying cause. [1][2][3][4] Amniotic fluid culture has been utilized for the diagnosis of microbial invasion of the amniotic cavity; however, the results of bacteriologic cultures may take in excess of 2 days. In addition, biologic properties of amniotic fluid such as bacteriolytic/static activity may potentially result in negative culture results, s The Gram stain is a rapid test that provides useful information for clinical management. However, at least 105 organisms/ml must be present to detect microbial invasion of the amniotic cavity by Gram stain, thus limiting the sensitivity of amniotic fluid Gram stain to 45%. 6 Low glucose concentration has been used as a marker for detecting infection in cerebrospinal fluid (CSF).  Recently the use of amniotic fluid glucose concentration has been reported in the detection of microbial invasion of the amniotic cavity in patients with PTL and PPROM.  The purpose of this study was to compare the value of Gram stain of amniotic fluid and amniotic fluid glucose in the identification of microbial invasion of the amniotic cavity and the development of clinical chorioamnionitis in patients presenting with PTL or PPROM. The relationship between presenting clinical diagnoses (PTL, PPROM, PTL with PPROM) and incidence of positive diagnostic tests was also evaluated. PTL was defined as regular painful uterine contractions with a frequency of at least three contractions within 10 minutes or documented cervical change. PPROM was confirmed by speculum examination with nitrazine and fern testing. The clinical diagnosis of chorioamnionitis was made according to the criteria described by Gibbs and coworkers s as a syndrome of fever, with either maternal or fetal taehycardia, uterine tenderness, foul-smelling amniotic fluid, or peripheral blood leukocytosis.

MATERIALS AND METHODS
Patients with antecedent antimicrobial therapy within month of presentation were excluded from the study. Patients were enrolled in the study regardless of previous tocolytic therapy, which included terbutaline or magnesium sulfate. Patients with known glucose intolerance were excluded from the study.
Amniocentesis was performed under local anesthesia and with ultrasound guidance. Samples of amniotic fluid were submitted to the hospital laboratory for clinically indicated tests such as Gram stain, culture and sensitivity, and pulmonary maturity studies.
The remainder of the amniotic fluid specimen was used for this study. Amniotic fluid was placed in a BBL Port-a-cul vial (Becton Dickinson and Company, Cockeysville, MD) and in separate sterile polypropylene tubes (Becton Dickinson Labware, Lincoln Park, NJ) that were immediately refrigerated and transported for storage in a 70C freezer until glucose assay. Gram stain was performed with reagents (BBL Microbiology Systems, Cockeysville, MD) under standard conditions. Quantitation of both microorganisms and white blood cells (WBCs) were as follows: 0, none present; < -I-, to occasional/oil immersion field (OIF); +, 1/OIF; 2 +, 2-5/OIF; 3 +, 5-10/OIF; and 4+, > 10/OIF. Culture media were obtained from Regional Media Laboratories, Inc. (REMEL, Lenexa, KS). Samples were placed on blood agar (Trypticase Soy Agar (TSA) with 5% sheep blood), chocolate agar, Thayer Martin agar (improved), and phenylethyl alcohol (PEA) agar with 5 % sheep blood at 35C in a CO 2 incubator. Amniotic fluid was plated on MacConkey's agar in a 35C non-CO 2 incubator. Anaerobic cultures were plated on anaerobic KV blood agar [Centers for Disease Control (CDC) formulation] and anaerobic PEA blood agar (CDC formulation) and grown in an anaerobic chamber at 35C. A-7 medium was utilized for the isolation of Mycoplasma and Ureaplasma (Scott Lab) and incubated at 35C in CO 2. Quantitative amniotic fluid cultures were performed with a Il/ml loop. Amniotic fluid cultures were considered positive if Mycoplasma or Ureaplasma were isolated.
Amniotic fluid glucose concentrations were determined in singlicate using the glucose oxidase method. Amniotic fluid glucose measurements were not used in the clinical management of patients. Subjects with positive amniotic fluid Gram stains were placed on antibiotics and then were delivered.   Of the 60 study subjects, 44 had normal amniotic fluid glucose concentration and 16 (Table 3). Thus, at least one-half of patients with positive cultures would be expected to have negative Gram stains for organisms. There were nine false negative and no false positive Gram stains for microorganisms. Nine of the 16   Among the 16 patients with positive amniotic fluid cultures, 7 had PPROM, 6 had PTL, and 3 had PPROM with PTL. There was no significant difference in amniotic fluid glucose concentration between patients with PPROM and PTL (19.7 vs. 25.3 mg/dl; P 0.58).
Amniotic fluid WBC count was lower in association with PPROM than with PTL; however, this did not reach statistical significance (0.57 + vs. 2.08 + ;P 0.077). The relationships between primary clinical diagnoses (PTL, PPROM, or PTL with PPROM) and clinical tests or outcomes are shown in Figure 2.
Among the 16 patients with low amniotic fluid glucose (<15 mg/dl), 6 had negative cultures and did not develop chorioamnionitis. One patient with Rh alloimmunization prematurely delivered 18 hours following amniocentesis. One patient delivered in 6 hours with abruptio placentae, which was believed to be the underlying cause of PTL. Another patient with PTL delivered in 91 hours with fetal distress and suspected pregnancy-induced hypertension. A patient with PPROM and oligohydramnios delivered after 232 hours with fetal distress; the neonate was found to have congenital syphilis. Two patients with low amniotic fluid glucose and negative cultures had prolonged amniocentesis-to-delivery times of 36 and 44 days.

DISCUSSION
The effects of leukocytes and bacteria on the glucose concentration of CSF in bacterial meningitis have been well established. 7 Both activated leukocytes and bacteria are required to lower CSF glucose concentration. 8-10 Altered glucose transport across inflamed membranes has also been a proposed mechanism for hypoglycorrhachia in bacterial meningitis. 16 This understanding makes it plausible that the presence of bacteria andactivated leukocytes in amniotic fluid may result in decreased concentrations of glucose. Weiss  pregnancies complicated by growth retardation or fetal malformation. [17][18][19] All papers published to date regarding amniotic fluid glucose concentration in detecting microbial invasion of the amniotic cavity have reported different thresholds of low amniotic fluid glucose. [11][12][13][14] Romero and colleagues 11 studied 168 women with PTL and intact membranes. They found that a cutoff of <14 mg/dl resulted in a reasonable compromise between the true positive and false positive rate. With this threshold, the sensitivity, specificity, and PPV were 86.9%, 91.7%, and 62.5%, respectively. Kirshon and associates lz studied 39 patients with PTL and/or PPROM. All nine patients with microbial invasion of the amniotic cavity were noted to have amniotic fluid glucose concentrations of < 10 mg/dl, and none with > 10 mg/dl developed chorioamnionitis. Kiltz and coworkers 13 evaluated the amniotic fluid glucose concentrations of 84 patients who were either in PTL, near term but not in labor, or at term in labor. They found that an amniotic fluid glucose concentration < 5 mg/dl had a PPV of 90% for positive culture and at > 20 mg/dl had a NPV of 98 %. The PPV at < 15 mg/dl was 46%, and the NPV at > 15 mg/dl was 97%. The PPV for amniotic fluid glucose concentration in our study was 43.8% at < 15 mg/dl and 47.1% at < 16 mg/dl, similar to the findings of Kiltz and co-workers.
In the clinical setting of PTL or PPROM, the ramifications of a false positive test include induction of labor and premature delivery, which are highly undesirable. Likewise, the results of clinical action based upon a false negative test may include unwarranted tocolysis, delayed delivery and withholding of antibiotic therapy, which may lead to adverse maternal and neonatal outcomes. We chose to use 15 mg/dl as a cutoff for normal amniotic fluid glucose concentration based upon the ability to predict positive cultures and the development of clinical chorioamnionitis. Analysis of our data demonstrates that this threshold for amniotic fluid glucose concentration is comparable or superior to the other laboratory tests evaluated for detection of clinical chorioamnionitis, but not for positive cultures. It should be recognized that immediate delivery for positive amniotic fluid Gram stain or culture may have a significant effect in reducing the number of subjects who would later develop clinical chorioamnionitis.
It appears that a cutoff for amniotic fluid glucose concentration at 15 mg/dl would be appropriate for general clinical use at this time. The use of a lower cutoff point would reduce the false positive rate at the expense of sensitivity, especially at earlier gestational ages. Studies examining the factor of gestational age have not yet been conducted. Gestational age may play an important role in interpretation of these data, since an amniotic fluid glucose concentration of 14 mg/dl may actually be normal at 3 5 weeks but considered low at 3 2 weeks of gestation. This is demonstrated by our data in Table 3: four patients (numbers 14,38,41, and 45)were found to have normal (> 15 mg/dl) amniotic fluid glucose concentrations in the presence of both positive Gram stain for microorganisms and positive amniotic fluid cultures. If gestational age were taken into consideration, these four patients would have been considered to have low amniotic fluid glucose. Further studies examining the effects of gestational age appear warranted. Due to limitations in sensitivity and specificity, we recommend that amniotic fluid glucose concentration should not be used alone in making clinical decisions, but should be factored in with clinical signs, symptoms, and other laboratory tests.