Postpartum testing for antecedent gestational diabetes

Postpartum testing for antecedent gestational diabetes Marshall W. Carpenter, MD, Donald R. Coustan, MD, John A. Widness, MD, Philip A. Gruppuso, MD, Marianne Malone, RN, and Lee M. Rotondo, RN Providence, Rhode Island Gestational diabetes is a predictor of glucose intolerance in subsequent pregnancies and in the nongravid state. Many pregnant women are not tested for gestational diabetes, although they or their offspring may show signs suggestive of antecedent hyperglycemia. We examined the diagnostic utility of a postpartum (within 48 hours), 100 gm, oral glucose tolerance test and cord plasma glucose, cord plasma C-peptide, and 2-hour neonatal plasma glucose tests to detect antecedent gestational diabetes in women with documented gestational diabetes (n = 37) or with normal glucose tolerance test results late in the third trimester (n = 28). The 1-hour, 2-hour, and incremental 1-hour + 2-hour ([1-hour - fasting] + [2-hour - fasting)) glucose values of the postpartum glucose toelrance test showed significant differences between study participants with and without gestational diabetes (164 ± 30 versus 115 ± 22, 145 ± 31 versus 101 ± 21, and 153 ± 51 versus 67 ± 33 mg/di, respectively, p < 0.025). Maternal fasting and 3-hour postpartum glucose tolerance test glucose, cord plasma glucose, cord plasma C-peptide, and 2-hour neonatal plasma glucose values showed no significant between-group differences. Receiver operating characteristic curve analyses for these tests indicated that the incremental 1-hour + 2-hour postpartum glucose tolerance test glucose values best sustain test specificity at the low test threshold values necessary for high test sensitivity. A threshold of 110 mg/di for this test yielded a predicted specificity of 90% and sensitivity of 80% with regard to antecedent gestational diabetes. (AM J OssTET GYNECOL 1988;159:1128-31.)

Key words: Gestational diabetes, postpartum glucose tolerance test, metabolic control

Gestational diabetes complicates 1.4% to 2.5% of all pregnancies. 1-3 Gravid women with gestational diabetes are more likely to have carbohydrate intolerance during future pregnancies and later in life. The Second International Workshop Conference on Gestational Diabetes 4 has recommended that all gravid women be screened for gestational diabetes. However, in its technical bulletin the American College of Obstetricians and Gynecologists does not recommend screening for women younger than 30 years of age. 5 A survey with regard to clinical practice patterns indicates that 59% of obstetricians do not perform universal screening for this disorder. 6 Maternal hyperglycemia or glucosuria during the puerperium and neonatal macrosomia, hypoglycemia, polycythemia, and hypocalcemia are nonspecific signs of possible antecedent maternal glucose intolerance during pregnancy. A more specific test, which could be

From the Departments of Obstetrics and Gynecology and Pediatrics, Brown University Program in Medicine, and Women and Infants and Rhode Island Hospitals. Supported in part by the Rhode Island Department of Health and the Centers for Disease Control. Presented at the Forty-seventh Annual Meeting of the American Diabetes Association, Indianapolis, Indiana, June 6-9, 1987. Received for publication March 17, 1988; revised May 17, 1988; accepted June 4, 1988. Reprint requests: Marshall W. Carpenter, MD, Women and Infants Hospital, 101 Dudley St., Providence, RI 02905.

1128

performed during the puerperium to detect antecedent gestational diabetes, would be useful when these signs are present in cases that lack antenatal testing. Surveillance of the patient with a history of gestational diabetes may thereby be undertaken during future pregnancies or later in life. We compared three neonatal variables (cord plasma glucose, cord plasma C-peptide, and 2-hour capillary plasma glucose values) with results of a maternal postpartum, 100 gm, 3-hour oral glucose tolerance test (GTT) to determine which was the most useful in the diagnosis of antecedent gestational diabetes.

Methods Participants were selected in a random fashion from known gestational diabetic and nondiabetic gravid women who were enrolled in a population-based study of gestational diabetes in the Providence, R.I., metropolitan area. All participants had a 50 gm, oral, I-hour screening test at 24 to 28 weeks. Control participants (n = 28) had values <130 mg/di and normal 100 gm, 3-hour, oral GTT results at 35 to 38 weeks. These tests were administered after a 3-day preparation with a high-carbohydrate diet. Test results were normal if at least three of four values were <95, 180, 155, and 140 mg/di limits. 7 Participants with gestational diabetes (n = 37) had a screening test value ""130 mg/di at 24 to 28 weeks. They were given a 3-hour GTT early in

Volume 159 Number 5

Postpartum glucose tolerance test

1129

.

1.0 .-------------- ---+6-1'>.fl

~

~~

~f.~.

0.8 -

:f

""t. . .

0.6 I

•

:I::

rn

31

t.

0.4 -

0.2 -

•

•• ~~=

•

-.6,

a~

I

.. I

•

•

•

••• • • t.

:f

1

0.4

t;.

2 hr glucose

•

1 hr glucose

1 + 2 hr Incremental glucose

cord c-peptide cord glucose 2 hr neonatal glucose

0.0 --t--~---i,r--r--,,----.--.,--.---.--.---l 0.0 0.2 0.4 0.6 0.8 1.0 1 - specificity

0.2

0.0

0.2

0.4

0.6

0.8

1.0

1 - specificity

Fig. 1. ROC curves of neonatal blood tests for gestational diabetes.

Fig. 2. ROC curves of postpartum GTT values for gestational diabetes.

the third trimester by the same protocol, with at least two of four values equal to or in excess of the preceding criteria. All but one (36 of 37) of the participants with gestational diabetes had weekly fasting, 2-hour postbreakfast, and 3 PM glucose surveillance. These 36 women had uniformly good metabolic control, with mean whole blood glucose values of 89.2 ± 7.3 mg/di (mean ± SD), as determined by a nurse using a calibrated reflectance meter. Insulin therapy was required by 13 of the 37 participants with gestational diabetes because of hyperglycemia. Lactated Ringer's solution or a normal saline solution was infused during labor into all but two participants, who received glucose solutions. Neonatal data for these two women were not included in data analysis. Mixed arterial and venous cord plasma samples were obtained for glucose and C-peptide analyses. Capillary blood was obtained from neonates 2 hours after birth. Plasma glucose was measured by the hexokinase method. We chose C-peptide as a marker for fetal insulin release to avoid assay interference by transplacentally acquired insulin antibodies in fetal blood. 8 Plasma C-peptide levels were measured by means of a double-antibody radioimmunoassay that used guinea pig antihuman Cpeptide antibody (Linco Research, Inc., Eureka, Ill.), 25 ' I-tyrosylated human C-peptide, and unlabeled recombinant human C-peptide (provided by Bruce Frank, Eli Lilly and Co., Indianapolis, Ind.). Interassay and intraassay coefficients of variation for the Cpeptide assay were 6% and 4%, respectively. Intraassay and interassay coefficients of variation for the glucose assay were I% to 2% a!ld 3%, respectively, depending on the analyzer used. Thirty-one of the 37 participants with gestational di-

abetes experienced labor, and two of them had cesarean sections. Six of these women were delivered of infants abdominally without labor. Of the 28 control participants, two had cesarean sections after labor and one had a cesarean section without preceding labor. Within 48 hours of parturition all participants had a 100 gm 3-hour oral GTT. All participants fasted ""'12 hours, although no other attempt was made to influence diet before the test. All testing began between 7 and 8:30 AM. Data analysis. Postpartum GTT values that were evaluated included the individual fasting, I-hour, 2-hour, and 3-hour concentrations as well as the sum of incremental I-hour and 2-hour values (incremental I-hour + 2-hour values = [I-hour - fasting] + [2-hour - fasting] values). Comparisons of test values between groups were examined by means of the Student unpaired two-tailed t test. The diagnostic utility of each test to detect gestational diabetes was evaluated by means of receiver operating characteristic (ROC) curve analysis. 9 By this method, the effect of test threshold selection on the reciprocal relationship of test sensitivity and specificity is displayed throughout the range of test values and allows comparisons of function among competing tests to be made. Results Significant differences in I-hour, 2-hour, and incremental I-hour + 2-hour glucose values were observed between patients with gestational diabetes and control participants (Table I). In contrast, none of the fasting or 3-hour postpartum GTT glucose values or cord plasma glucose, cord plasma C-peptide, or 2-hour neo-

1130 Carpenter et al.

November 1988 Am J Obstet Gynecol

Table I. Postpartum tests for antecedent gestational diabetes Control participants Test

Postpartum GTT (mg/di) I-hour glucose 2-hour glucose Incremental I-hour + 2-hour glucose Cord glucose (mg/di) Cord C-peptide (pmol/ml) 2-hour neonatal glucose

No

I

Mean

I

Gestational diabetes study SD

No

I

Mean

I

SD

Significance

27 27 27

115 IOI 67

22 21 33

37 36 36

164 115 153

30 31 51

p < 0.025 p < 0.025

24 18 22

87 0.29 58

25 O.Q7

33 31 36

93 0.40 60

31 0.27 23

NS NS NS

natal capillary plasma glucose values was significantly different between groups. The diagnostic utility of each of these variables in the identification of earlier gestational diabetes was examined by means of ROC analysis. The ROC curves for each of the three neonatal variables followed a nearly diagonal path from coordinates 0,0 to I, I, which indicates that test sensitivity for these variables cannot be obtained without loss of test specificity (Fig. I). In contrast, the ROC curves of the I-hour, 2-hour, and Ihour + 2-hour incremental postpartum GTT values indicate the test threshold may be reduced to achieve test sensitivity '30.6 before test specificity is lost (Fig. 2). Of these three variables, the I-hour + 2-hour incremental postpartum GTT glucose values achieved a sensitivity of 0.8 at a test threshold of I I 0 mg I di before specificity fell to <0.9. Twenty-eight of the 37 participants (76%) in the gestational diabetes group met National Diabetes Data Group criteria 10 for gestational diabetes. Comparison of tests between these 28 women and the 28 control participants showed the same differences seen with all 37 cases compared with the control group. In addition, the ROC analyses that used this diagnostic criterion revealed qualitatively similar results. Seven of the 37 infants of women with gestational diabetes and seven of the 28 infants from the control group had birth weights >4000 gm. Among participants delivered of infants >4000 gm, the I-hour and 2-hour incremental postpartum GTT values were significantly higher in those with gestational diabetes (195 ± 4I versus 76 ± 26 mg/di, p < O.OOOI). Two infants of gestational diabetic participants and three infants of the control group women had birth weight percentiles >90." Their mean I-hour and 2-hour incremental postpartum GTT values were I94 and 63 mg/ di, respectively. Comment

Diagnosis of gestational diabetes serves to identify gravid women at risk for perinatal morbidity, including fetal macrosomia, birth trauma, and operative delivery,

15

p<

0.025

as well as neonatal hypoglycemia, hypocalcemia, polycythemia, and hyperbilirubinemia. Criteria for the application of insulin treatment in these women vary. They include exceeding of thresholds of fasting and postmeal glucose values obtained weekly, daily use of prophylaxis in all women with abnormal tolerance test results, and elevated amniotic fluid insulin values in the third trimester. 12 The application of all criteria depends on the antepartum identification of abnormal glucose tolerance. O'Sullivan et al.' have demonstrated a detection rate of 53% when historical risk factors alone are used to determine the need for a glucose tolerance test. Current recommendations of the American College of Obstetricians and Gynecologists' and recent surveys with regard to obstetric practice suggest that universal screening for gestational diabetes occurs in only a minority of practices. In one study6 one third of respondent practices screened patients at 24 to 28 weeks' gestation and less than one fourth used the recommended• 50 gm oral glucose challenge test. In this context clinical conditions at parturition will often raise the question of antecedent gestational diabetes. Several investigators have pointed out altered but persistent pregnancy-related changes in glucose homeostasis during the puerperium. Burt et al., 13 by means of a cross-sectional design, demonstrated lower insulin and glucose concentrations in response to an oral, 50 gm, glucose challenge performed 4 days postpartum, compared with those evoked by the same challenge during the third trimester. Reductions in both fasting and 2.5-hour glucose concentrations were also evident after the same glucose challenge at 2 to 3 days postpartum, compared with antenatal values in a cohort study. 14 Lind and Harris" performed 50 gm oral glucose challenge tests at 36 to 38 weeks' gestation, the second and fifth postpartum days, and 6 weeks after parturition. Insulin responses returned to levels similar to the 6-week test by the fifth postpartum day. However, plasma glucose levels, in response to the oral challenge, remained elevated through the fifth postpartum day. Weiss et al. 16 have established norms for umbilical cord blood insulin concentrations m normal-

Postpartum glucose tolerance test

Volume 159 Number 5

birthweight neonates. Cord insulin assays and puerperal maternal oral glucose tolerance tests were performed in 22I instances when babies weighed >4 kg. A threefold increase (I8.6% versus 6.2%) in abnormal glucose tolerance was identified in cases with cord insulin values >20 µU/ml (mean+ 2.45 SD). The foregoing studies suggested that umbilical cord glucose and C-peptide tests and the 2-hour neonatal glucose test might function as well as a postpartum oral glucose tolerance test in the identification of antecedent gestational diabetes. However, ROC analysis indicated that the neonatal variables studied had no utility in the irlentification of antecedent gestational diabetes. On the other hand, variables derived from the postpartum GTT appeared to offer useful information about previous glucose tolerance when performed within the first 48 hours post partum. The incremental I-hour + 2hour glucose values of the postpartum GTT function best in this regard. At at test threshold of I IO mg/ di, 90.8% of the nondiabetic population and 80.0% of those with gestational diabetes will be correctly identified on the basis of these data. Participants for this investigation were selected from a populational study of gravid women in the Providence, R.I., metropolitan area where the incidence of gestational diabetes is 3.3%, according to 62I4 screening glucose challenge tests and the aforementioned diagnostic criteria. On the assumption of this incidence, the positive predictive accuracy (true positive/[true positive + false positive]) is 22.9% and the negative predictive accuracy (true negative/[true negative + false negative]) is 99.3% on the basis of an incremental I-hour + 2-hour glucose threshold of I IO mg/di. We assume that the incidence of this disorder would be higher in instances in which clinical events suggest antecedent gestational diabetes. This would substantially increase the positive predictive accuracy with negligible effect on the negative predictive accuracy. Among the few instances of neonatal macrosomia in this series, the incremental I-hour + 2-hour glucose values continued to differentiate between nondiabetic and gestational diabetic gravid women. Consequently, we recommend that if antecedent gestational diabetes is suspected at parturition but is undocumented, a 2-hour, I 00 gm, oral glucose tolerance test may be administered. An abnormal result, on the

1131

basis of incremental I-hour + 2-hour glucose concentration values (~I IO mg/di), may justify either further testing in future pregnancies or diagnostic surveillance later in the patient's life. We acknowledge the able assistance of Theresa C. Bienieki. REFERENCES 1. O'Sullivan JB, Mahan CM, Charles D, Dandrow R. Screening criteria for high-risk gestational diabetic patients. AM J 0BSTET GYNECOL 1973; 116:895. 2. Lavin JP, Barden TP, Miodovnik M. Clinical experience with a screening program for gestational diabetes. AM J 0BSTET GYNECOL 1981; 141 :491. 3. Marquette GP, Klein VR, Repke JT, Niebyl JR. Costeffective criteria for glucose screening. Obstet Gynecol 1985;66: 181. 4. Summary and recommendations of the second international workshop-conference on gestational diabetes mellitus. Diabetes 1985;34:123. 5. American College of Obstetricians and Gynecologists. Management of diabetes mellitus in pregnancy. Washington: American College of Obstetricians and Gynecologists, 1986; ACOG Technical Bulletin no. 92. 6. Furman GJ, Steinberg MC. Diabetes screening during pregnancy [Abstract]. Diabetes 1987;36(suppl 1):90A. 7. Carpenter MW, Coustan DR. Criteria for screening tests for gestational diabetes. AM J OBSTET GYNECOL 1982;144:768. 8. DiMario U, Fallucca F, Gargiulo P, et al. Insulin-antiinsulin complexes in diabetic women and their neonates. Diabetologia 1984;27:83. 9. Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1978;8:283. 10. National Diabetes Data Group. Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 1979;28:1039. 11. Thomson AM, Billewicz WZ, Hytten FE. The assessment of fetal growth. J Obstet Gynaecol Br Commonw 1968; 75:903. 12. Weiss PAM, Hofman H, Winter R, Purstner P, Lichtenegger W. Gestational diabetes and screening during pregnancy. Obstet Gynecol l 984;63:776. 13. Burt RL, Leake NH, Rhyne L. Glucose tolerance during pregnancy and the puerperium-a modification with observations on serum immunoreactive insulin. Obstet Gynecol 1969;33:634. 14. MacDonald HN, Good W, Schwarz K, Stone J. Serial observations of glucose tolerance in pregnancy and the early puerperium. J Obstet Gynaecol Br Commonw 1971;78:489. 15. Lind T, Harris VG. Changes in the oral glucose tolerance test during the puerperium. Br J Obstet Gynaecol l 976;83:460. 16. Weiss PAM, Hofmann H, Purstner P, Winter R, Lichtenegger W. Fetal insulin balance: gestational diabetes and postpartal screening. Obstet Gynecol 1984;64:65.