Advertisement

Hormones

, Volume 17, Issue 2, pp 255–259 | Cite as

HbA1c presents low sensitivity as a post-pregnancy screening test for both diabetes and prediabetes in Greek women with history of gestational diabetes mellitus

  • Michael Apostolakis
  • Stavroula A. Paschou
  • Evangelia Zapanti
  • Vasiliki Sarantopoulou
  • Vasiliki Vasileiou
  • Eleni Anastasiou
Original Article
  • 21 Downloads

Abstract

Background and aim

Women with a history of gestational diabetes mellitus (GDM) are at increased risk for type 2 diabetes (T2D). It is thus recommended that an oral glucose tolerance test (OGTT) be performed after delivery. Recently, the use of glycated haemoglobin A1c (HbA1c) has been proposed as a simpler and faster method to diagnose glucose disorders. The aim of this study was to investigate whether HbA1c measurement can replace OGTT in the detection of prediabetes and T2D in women with a history of GDM.

Patients and methods

We studied 1336 women (35.3 ± 5.8 years old) with a history of GDM 16.6 ± 28.2 months after delivery. All women were evaluated through an OGTT and a simultaneous HbA1c measurement. American Diabetes Association (ADA) criteria were used for the assessment of glucose disorders. Sensitivity and specificity of HbA1c were measured for the prediction of T2D and prediabetes, while Cohen’s coefficient of agreement (k) was calculated. ROC analysis was performed to evaluate the sensitivity and specificity of HbA1c.

Results

Based on OGTT, 725 women (54.3%) were normal, 406 (30.4%) presented impaired fasting glucose (IFG), 48 (3.6%) impaired glucose tolerance (IGT), 74 (5.5%) combined IFG+IGT, and 83 presented with T2D (6.2%). By contrast, using HbA1c as a screening test, 1150 women (94.1%) were normal, while 49 (4.0%) had prediabetes and 23 (1.9%) T2D. Sensitivity of HbA1c for the diagnosis of prediabetes was 5.3% in comparison to OGTT, specificity was 99.2%, while for the diagnosis of T2D, the percentages were 29.6 and 100%, respectively. The consistency in classifying impaired glucose tolerance between HbA1c and OGTT was 59.7%. Cohen’s coefficient of agreement was k = 0.116, indicating slight agreement. Performing a ROC curve, the optimal value of distinctive ability of HbA1c was 4.6% in the case of prediabetes, while for diabetes, it was 5.5%.

Conclusion

This study provided evidence that HbA1c can identify fewer women with prediabetes and T2D than OGTT, indicating that HbA1c cannot be recommended as an alternative post-pregnancy screening method.

Keywords

Gestational Diabetes Prediabetes HbA1c OGTT 

Introduction

Women with a history of gestational diabetes mellitus (GDM) have an increased risk for type 2 diabetes (T2D). Specifically, this risk is six to seven times higher compared to women without history of GDM [1, 2]. Thus, it is recommended that all women with GDM undergo an oral glucose tolerance test (OGTT) after delivery. As the presence of either prediabetes or T2D occurs quite early after pregnancy, the first screening with OGTT takes place at 4–12 weeks after delivery according to the American Diabetes Association (ADA) guidelines [3]. The non-pregnancy diagnostic criteria are used and if the test is negative, women with a history of GDM should have lifelong screening at least every 3 years. Furthermore, women with a history of GDM who are found to have prediabetes should receive intensive lifestyle interventions and/or metformin to prevent T2D [3].

Glycated haemoglobin A1c (HbA1c) correlates with the average blood glucose concentrations over the past 2 to 3 months and is now recognised internationally as the simplest and fastest method for diagnosis and follow-up of patients with diabetes [3, 4]. Because of these advantages, it appears ideal also for the post-partum screening of women with GDM, although the ADA does not recommend this test [3]. Data from a few cohorts of women in the post-partum period suggest that the two tests measure rather different aspects of glycaemic dysregulation [5, 6, 7, 8, 9]. Discordance between HbA1c and OGTT in 6–12 weeks [5], 1 year post-partum [6, 7], 36 months post-partum [8], and even up to 5 years post-partum [9] was the common outcome.

However, the number of studies already performed is small; some of them have shown differences in the ability of the two methods to detect prediabetes or T2D, while the possibility of adaptation of such a simple and useful tool as HbA1c to post-partum screening should be further investigated. The aim of this study was to assess whether HbA1c measurement can replace OGTT in the detection of prediabetes and T2D in women with a history of GDM.

Patients and methods

Study participants

We included in our study 1336 women of Greek origin (35.3 ± 5.8 years old, BMI 26.7 ± 5.4 kg/m2) with a history of GDM who attended the diabetes centre of “Alexandra” Hospital in Athens, Greece, between 2000 and 2015, 16.6 ± 28.2 months after delivery. All included women had been examined with an OGGT-100g during the third trimester of their pregnancy in our centre, with the GDM diagnosis being based on the Carpenter and Coustan criteria. Pregnant women with pre-existing diabetes were excluded. The research was approved by the Institutional Review Board of the “Alexandra” Hospital, Athens, Greece. Informed consent was obtained from all participants.

All women were evaluated through an OGTT (with 75 g of glucose) and a simultaneous HbA1c measurement during the post-partum period, 16.6 ± 28.2 months after delivery. The ADA criteria were used for the assessment of dysglycaemia [3], specifically, HbA1c ≥ 6.5% (48 mmol/mol), fasting glucose ≥ 126 mg/dl (7.0 mmol/l), or 2 h glucose (OGTT) ≥ 200 mg/dl (11.1 mmol/l) for the diagnosis of T2D, and HbA1c 5.7–6.4% (39–46 mmol/mol), fasting glucose 100–125 mg/dl (5.6–6.9 mmol/l), or 2 h glucose (OGTT) 140–199 mg/dl (7.8–11.0 mmol/l) for the diagnosis of prediabetes. More specifically, for prediabetes, when fasting glucose is 100–125 mg/dl (5.6–6.9 mmol/l), the condition is defined as impaired fasting glucose, while when 2 h glucose after OGTT is 140–199 mg/dl (7.8–11.0 mmol/l), the condition is defined as impaired glucose tolerance. Some patients can present both.

Assays

Venous samples were collected in BD Vacutainer® spray-coated K2EDTA tubes. Plasma glucose levels were determined by the glucose oxidase method (Integra 400 plus, Roche). HbA1c was determined using high-pressure liquid chromatography (HPLC-ΗΑ8160 Menarini Arkay). The inter- and intra-assay coefficients of variation for HbA1c were less than 2%.

Statistical analysis

The demographics and measured parameters are described with their mean and standard deviations. Pearson’s correlation coefficients reported the association between glucose measurements and HbA1c. The classification of women based on HbA1c and the OGTT criteria is shown with the relevant frequency distributions. The difference of the proportion of women with abnormal glucose homeostasis and the proportion of women with abnormal glucose tolerance was checked for statistical significance. The diagnosis of prediabetes and diabetes based on the two criteria was cross-tabulated into 2 × 2 contingency tables, reporting in each case the resulting parameters (sensitivity, specificity, positive and negative predictive values, and likelihood ratios) with their 95% confidence intervals (CI). Cohen’s coefficient of agreement (k) between OGTT and HbA1c was calculated. Finally, receiver operating characteristic (ROC) curves were plotted to compute the discrimination ability of HbA1c on prediabetes and diabetes. Based on these curves, the best cutoff values that yielded the best tradeoff between sensitivity and specificity for the diagnosis of prediabetes and diabetes were computed and, once more, the parameters of the resulting 2 × 2 contingency tables are reported.

Results

Table 1 displays the descriptive statistics of the demographic data and measured parameters and Table 2 reports the Pearson’s correlation coefficients between the measured parameters. The high correlations of HbA1c with the glucose parameters raise expectations that it can act as a diagnostic tool for prediabetes and diabetes.
Table 1

Descriptive statistics of demographic data and measured parameters

Parameter

Mean

Standard deviation

Assessment time after delivery (months)

16.6

28.2

Age at assessment (years)

35.3

5.8

BMI at assessment (kg/m2)

26.7

5.4

Fasting glucose (mg/dl)

100.3

20.7

2 h glucose after OGTT (mg/dl)

111.2

46.4

HbA1c (%)

4.9

0.7

Table 2

Pearson’s correlation coefficients between the measured parameters

Parameters

Pearson’s correlation (r)

Significance (p)

Fasting glucose with 2 h glucose

0.686

< 0.01

Fasting glucose with HbA1c

0.645

< 0.01

2 h glucose with HbA1c

0.494

< 0.01

Based on OGTT, 725 women (54.3%) were found to be normal, 406 (30.4%) presented impaired fasting glucose (IFG), 48 (3.6%) impaired glucose tolerance (IGT), 74 (5.5%) combined IFG + IGT, and 83 T2D (6.2%). In contrast, using HbA1c as a screening test, 1150 women (94.1%) were normal, 49 (4.0%) presented prediabetes, and 23 (1.9%) T2D. The proportion of women observed with abnormal glucose homeostasis (5.5%) was found to be significantly lower than the proportion of women found with abnormal glucose tolerance (45.7%, p < 0.01). The consistency in classifying impaired glucose tolerance between HbA1c and OGTT was 59.7%. Cohen’s coefficient of agreement was k = 0.116, indicating slight agreement.

The above findings are further corroborated by the contingency cross-tabulation of the diagnosis of prediabetes and diabetes based the HbA1c and OGTT criteria (Table 3). While in both cases the specificity is perfect, the sensitivity is unacceptable. An increase in sensitivity (which, however, will be accompanied by a collateral decrease in specificity) can be achieved if the cutoff values of 5.7% for prediabetes and 6.4% for diabetes used for HbA1c were lowered. This can be achieved in the following manner. The first step is to prove that HbA1c has significant discrimination ability for prediabetes and diabetes. This is accomplished with ROC analysis. As depicted in Fig. 1, the area under the curve (AUC) of HbA1c for prediabetes (0.644) and diabetes (0.816) is in both cases significantly greater than 0.5. Furthermore, the discrimination ability of HbA1c for diabetes is significantly greater than for prediabetes.
Table 3

Diagnostic ability of HbA1c based on established criteria

Estimated parameter

Prediabetes

Diabetes

Value

95% CI

Value

95% CI

Sensitivity

5.3%

3.5–7.8%

29.6%

19.6–41.8%

Specificity

99.2%

98.1–99.7%

100%

99.6–100%

PPV

83.3%

64.5–93.7%

100%

80.8–100%

NPV

59.6%

56.6–62.5%

95.8%

94.4–96.8%

Likelihood ratio +

7.0

2.7–18.2

ΝΑ

ΝΑ

Likelihood ratio −

0.95

0.93–0.97

0.704

0.606–0.819

PPV positive predictive value, NPV negative predictive value, NA cannot be computed because of the presence of a zero in the 2 × 2 contingency table

Fig. 1

ROC curves for the discrimination ability of HbA1c on prediabetes and diabetes. AUC—area under the curve

The second step is to establish independently for prediabetes and diabetes which cutoff value of HbA1c yields the best tradeoff between sensitivity and specificity. In the case of prediabetes, it was found that this value is 4.6%, while for diabetes, it is 5.5%. Based on these cutoff values, once again, 2 × 2 contingency tables were built and the resultant parameters are shown in Table 4.
Table 4

Diagnostic ability of HbA1c based on computed cutoff values

Estimated parameter

Prediabetes (cutoff value 4.6%)

Diabetes (cutoff value 5.5%)

Value

95% CI

Value

95% CI

Sensitivity

69.5%

65.15–73.1%

58.1%

46.1–69.3%

Specificity

50.9%

47.0–54.8%

91.9%

90.2–93.4%

PPV

50.1%

46.3–54.0%

31.4%

22.9–40.0%

NPV

70.1%

65.8–74.1%

97.2%

96.0–98.1%

Likelihood ratio +

1.42

1.28–1.56

7.21

5.49–9.49

Likelihood ratio −

0.60

0.52–0.69

0.46

0.35–0.60

Discussion

This study provided evidence that HbA1c measurement has a very low sensitivity in detecting both prediabetes and T2D in Greek women with a history of GDM compared to OGTT at the same time post-partum, specifically 16.6 ± 28.2 months after delivery. HbA1c identified fewer women with prediabetes and T2D than OGTT, indicating that HbA1c cannot be recommended as an alternative method for screening.

This is the first study in a Greek cohort of women with a history of GDM. Our results regarding the lower sensitivity of HbA1c compared to OGTT are in accordance with the results of most previous similar studies performed at various time points after delivery. Indeed, discordance between HbA1c and OGTT has been demonstrated in 6–12 weeks [5], 1 year post-partum [6, 7], 36 months post-partum [8], and even up to 5 years post-partum [9]. Therefore, it seems that this discordance exists early after delivery and does not show any convergence with the passage of time. In the immediate post-partum period, the lower accuracy of HbA1c could be explained more easily. First of all, women with GDM present better glycaemic control as a result of diet alone or together with hypoglycaemic therapy, and this can be reflected in the HbA1c levels for the next 2–3 months. Furthermore, haematological factors that are related to pregnancy, such as anaemia from iron and follicacid deficiency or haemorrhage, could affect the HbA1c measurements. The time of our study is late enough after delivery for these influential factors to be excluded.

This study showed lower sensitivity for HbA1c measurement regarding both T2D and prediabetes. From the results, it is evident that although HbA1c and glucose are fairly well correlated, the established HbA1c and OGTT criteria target different pathophysiologies, which may not necessarily be tightly connected. The stringent criteria set for HbA1c are not adapted for an adequate positive prediction of prediabetes and diabetes. A few previous studies have reported that HbA1c identified more women with prediabetes compared to OGTT [5, 10], this also indicating that the two tests may measure different aspects of glycaemic dysregulation. One study was performed in 6–12 weeks [5] and the other during the first year post-delivery [10]. In general, abnormal results in OGTT reflect an inability of the woman to process the glucose load, whereas abnormal HbA1c indicates a more general status of dysglycaemia, including impaired fasting levels of glucose. Discordance between these different aspects of glycaemic dysregulation is seen more often in individuals with modestly elevated glucose concentrations [11], and this is the case for most women with previously diagnosed GDM.

The strengths of our study include the large number of participants, the homogeneous cohort of women of Greek origin only, and the simultaneous measurement of HbA1c with the performance of OGTT. In ethnically diverse cohorts, the variability in the presence of haemoglobinopathies or variations in haemoglobin concentrations or racial differences in HbA1c levels should be taken into consideration. A limitation of the study could be that “Alexandra” Hospital in Athens, Greece, is a referral centre for high-risk pregnant women. Therefore, the cases of GDM included were selected among women with greater severity of GDM and this may not reflect the general population of women with GDM. On the other hand, a high-risk population of women with GDM would be expected to have increased levels of dysglycaemia and therefore increased agreement between HbA1c and OGTT results. This aspect confers even greater importance to the interpretation of our findings.

In conclusion, this study provided evidence that HbA1c measurement presents low sensitivity in detecting both prediabetes and T2D in Greek women with a history of GDM after delivery. Therefore, this test cannot be recommended as an alternative method for screening of glycaemic disorders in the post-partum period.

Notes

Compliance with ethical standards

The research was approved by the Institutional Review Board of the “Alexandra” Hospital, Athens, Greece. Informed consent was obtained from all participants.

Conflict of interest

The authors declare that they have no conflict of interest.

References

  1. 1.
    Cheung NW, Byth K (2003) The population health significance of gestational diabetes. Diabetes Care 26(7):2005–2009CrossRefPubMedGoogle Scholar
  2. 2.
    Bellamy L, Casas J, Hingorani A, Williams D (2009) Type 2 diabetes mellitus after gestational diabetes: a systematic review and meta-analysis. Lancet 373(9677):1773–1779CrossRefPubMedGoogle Scholar
  3. 3.
    American Diabetes Association (2017) Standards of medical care in diabetes—2017. Diabetes Care 40(Suppl. 1):S114–S119CrossRefGoogle Scholar
  4. 4.
    Inzucchi SE, Bergenstal RM, Buse JB, Diamant M, Ferrannini E, Nauck M, Peters AL, Tsapas A, Wender R, Matthews DR (2015) Management of hyperglycemia in type 2 diabetes, 2015: a patient-centered approach: update to a position statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care 38:140–149CrossRefPubMedGoogle Scholar
  5. 5.
    Duke A, Yap C, Bradbury R, Hng TM, Kim C, Wansbrough A, Cheung NW (2015) The discordance between HbA1c and glucose tolerance testing for the postpartum exclusion of diabetes following gestational diabetes. Diabetes Res ClinPract 108(1):72–77CrossRefGoogle Scholar
  6. 6.
    Megia A, Näf S, Herranz L, Serrat N, Yañez RE, Simón I, Vendrell J (2012) The usefulness of HbA1c in postpartum reclassification of gestational diabetes. BJOG 119(7):891–894CrossRefPubMedGoogle Scholar
  7. 7.
    Kim C, Herman WH, Cheung NW, Gunderson E, Richardson C (2011) Comparison of hemoglobin A1c with fasting plasma glucose and 2-h postchallenge glucose for risk stratification among women with recent gestational diabetes mellitus. Diabetes Care 34(9):1949–1951CrossRefPubMedPubMedCentralGoogle Scholar
  8. 8.
    Picon J, Murri M, Munoz A, Fernandez-Garcia J, Gomez-Huelgas R, Tinahones F (2012) Hemoglobin A1c versus oral glucose tolerance test in postpartum diabetes screening. Diabetes Care 35(8):1648–1653CrossRefPubMedPubMedCentralGoogle Scholar
  9. 9.
    Noctor E, Crowe C, Carmody LA, Avalos GM, Kirwan B, Infanti JJ, O'Dea A, Gillespie P, Newell J, McGuire B, O'Neill C, O'Shea PM, Dunne FP, investigators ATLANTICDIP (2013) ATLANTIC DIP: simplifying the follow-up of women with previous gestational diabetes. Eur J Endocrinol 169(5):681–687CrossRefPubMedGoogle Scholar
  10. 10.
    Benaiges D, Chillaron JJ, Pedro-Botet J, Mas A, Puig de Dou J, Sagarra E, Carrera MJ, Goday A, Flores-Le Roux JA (2013) Role of A1c in the postpartum screening of women with gestational diabetes. GynecolEndocrinol 29(7):687–690Google Scholar
  11. 11.
    Van’t RE, Alssema M, Rijkelijkhuizen JM, Kostense PJ, Nijpels G, Dekker JM (2010) Relationship between HbA1C and glucose levels in the general Dutch population: the new Hoorn study. Diabetes Care 33(1):61–66CrossRefGoogle Scholar

Copyright information

© Hellenic Endocrine Society 2018

Authors and Affiliations

  • Michael Apostolakis
    • 1
  • Stavroula A. Paschou
    • 2
  • Evangelia Zapanti
    • 1
  • Vasiliki Sarantopoulou
    • 1
  • Vasiliki Vasileiou
    • 1
  • Eleni Anastasiou
    • 1
  1. 1.Department of Endocrinology and Diabetes“Alexandra” HospitalAthensGreece
  2. 2.Division of Endocrinology and Diabetes, “Aghia Sophia” Hospital, Medical SchoolNational and Kapodistrian University of AthensAthensGreece

Personalised recommendations