2013

http://informahealthcare.com/gyeISSN: 0951-3590 (print), 1473-0766 (electronic)

Gynecol Endocrinol, 2013; 29(7): 691–694! 2013 Informa UK Ltd. DOI: 10.3109/09513590.2013.797398

GESTATIONAL DIABETES

Homeostatic indices of insulin resistance among gestational diabetics inanticipating pregnancy complications

Nor Azlin Mohamed Ismail1, Maslinda Mohd Kasim2, Azimatun Noor Aizuddin3, and Nor Aini Umar4

1Department of Obstetrics & Gynaecology, Faculty of Medicine, 2Department of Pathology, 3Department of Community Medicine and4Department of Pathology, Universiti Kebangsaan Malaysia, Kuala Lumpur, Malaysia

Abstract

Objective: This was to determine HOMA-IR score as well as to assess its association in fetal andmaternal outcomes among pregnant women with diabetes risks.Methods: A prospective cohort study of pregnant women with diabetes risks was done. GDMwas diagnosed using modified glucose tolerance test. Serum insulin was taken and measuredby an electrochemiluminescence immunoassay method. Plasma glucose was measured byenzymatic reference method with hexokinase. HOMA-IR score was calculated for each patient.Maternal and fetal outcomes were analyzed.Results: From 279 women recruited, 22.6% had GDM with higher HOMA-IR score (4.07� 2.44versus 2.08� 1.12; p¼ 0.001) and fasting insulin (16.76� 8.63 mIU/L versus 10.15� 5.07 mIU/L;p¼ 0.001). Area under ROC curve for HOMA-IR score was 0.79 (95% confidence interval, 0.74–0.84) with optimum cut-off value of 2.92 (sensitivity¼ 63.5%; specificity¼ 89.8%), higher thanrecommended by IDF (2.38). This point showed significant association with neonatalhypoglycemia (p¼ 0.02) and Cesarean section (p¼ 0.04) in GDM mothers.Conclusions: HOMA-IR score and insulin resistance levels were higher in GDM women in ourpopulation. With the cut-off HOMA-IR value of 2.92, neonatal hypoglycemia and Cesareansection were significant complications in GDM mothers. This can be used in anticipation ofmaternal and fetal morbidities.

Keywords

Diabetes, insulin resistance, pregnancy

History

Received 9 January 2013Revised 1 March 2013Accepted 26 March 2013Published online 4 June 2013

Introduction

Pregnancy results in insulin resistance and hyperinsulinemia,render pregnant women prone to diabetes. This happens whenthere is insufficient insulin secretion to meet the demandespecially in gestational diabetes mellitus (GDM) which isdefined as glucose intolerance with onset or first recognitionduring pregnancy [1]. GDM is known to associate with maternaland fetal morbidity and mortality involving stillbirth, birthtrauma, cesarean section, pre-eclampsia, hypoglycemia, respira-tory distress, polycythemia, hyperbilirubinemia, hypocalcemiaand increase neonatal intensive care unit admissions [2]. Withincreasing prevalence of diabetes mellitus globally, a higherincidence of GDM is especially seen among the Asian ethnicgroup [3]. Different ethnicity shows different ability to cope withthe increased insulin resistance during pregnancy. As pregnancyprogresses, the levels of placental hormones also increase whichthen peaked between 26th to 33rd weeks of gestation leading to afurther increase in insulin resistance. Recent studies [4,5] hadreported pregnant women from East Asia and South Asia withmore susceptibility to GDM as compared to the Caucasians due tothis matter.

Clinically insulin resistance is defined as impaired ability ofinsulin (either endogenous or exogenous) to lower blood glucoselevels. Women with GDM have a greater insulin resistance

compared to those seen in normal pregnancies. The quantitativeassessment of insulin resistance is not routinely used as part of adiagnostic purpose in biochemical investigations. However, withthe emerging importance of insulin resistance, this has led to amore extensive research application. The hyperinsulinemiceuglycemic glucose clamp is the gold standard for determinationof insulin sensitivity; however, it requires tedious and laboriousprocedure which clinically is not practicable. A few compatibleindices have therefore, been employed to simplify the testincluding homeostatic model assessment of insulin resistance(HOMA-IR). This model correlates with estimation using theeuglycemic clamp method (r¼ 0.88) and was proven to be arobust clinical epidemiological tool in describing pathophysi-ology of diabetes [6]. This current study was done to determineHOMA-IR score as well as to assess its utility in predicting fetaland maternal complications among our pregnant women with thediabetes risks.

Materials and methods

This prospective cohort study was conducted in a tertiary teachinghospital for the duration of 12 months from Jun 2010 until July2011. All pregnant women with diabetes risks (diabetes mellitusin first degree relatives, previous pregnancy with gestationaldiabetes mellitus, previous macrosomic baby, previous unex-plained stillbirth, previous baby with congenital abnormally,recurrent miscarriages, age 35 years and above, obese or pre-pregnancy weight more than 80 kg) and indicated for selectivescreening of modified glucose tolerance test (MGTT) wererecruited into the study. Patients who were having multiple

Address for correspondence: Nor Azlin Mohamed Ismail, Department ofObstetrics & Gynaecology, Faculty of Medicine, Universiti KebangsaanMalaysia, Jalan Yaacob Latiff, Cheras, Kuala Lumpur 56000, Malaysia.E-mail: azlinm@ppukm.ukm.edu.my

Gyn

ecol

End

ocri

nol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 11

/23/

14Fo

r pe

rson

al u

se o

nly.

pregnancies were excluded from the study. The study wasapproved by the Institutional Research and Ethics ReviewBoard. The patients were explained about the study andconsented. They were requested to fast overnight before fastingblood glucose level was taken in the morning. At the same time,they had a total of 6 mL of fasting blood (4 mL in plain tube and2 mL in sodium fluoride tube) for measurement of serum insulinand plasma glucose. A 75 g glucose drink was given and anotherblood sample was taken two hours later.

GDM was diagnosed based on MGTT result (either one resultor both fasting plasma glucose was �6.0 mmol/L or two-hourplasma glucose was �7.8 mmol/L). Serum insulin was measuredby the electrochemiluminescence immunoassay ‘‘ECLIA’’method on Roche Diagnostics Elecsys� (Roche Diagnostics,Germany). The analyzer automatically calculated the analyteconcentration of each sample (either in micro unit per milliliter(mU/mL) or picomol per liter (pmol/L). The Insulin assaymeasured concentrations up to 300 mU/mL with the lowerdetection limit of 2 mU/mL. The intra-assay coefficient ofvariation (CV) ranged from 7.1 to 8.0 and 5.9 to 7.0 for lowand high Insulin levels, respectively. On the other hand, plasmaglucose was measured by enzymatic reference method withhexokinase on Roche Diagnostics Cobas

�with the intra-assay CV

of 3.2%. HOMA-IR was calculated for each pair of fasting seruminsulin and plasma glucose level using the formula:

Fasting serum glucose ðmmol=LÞ � Fasting insulin ðmIU=LÞ22:5

Further medical managements followed the usual hospitalprotocol. Maternal demographic data including body mass index(BMI), ethnicity, glycemic control, diabetes risks, progress ofpregnancies and pregnancy outcomes were recorded andanalyzed.

All clinical and laboratory data were stored and analyzed usingthe Statistical Package for Social Sciences (SPSS) version 19.0software (Chicago, IL). Assessment for normal distribution of thedata was performed with chi-square and student t-test. Mann–Whitney test was used for unpaired nonparametric data, with the95% confidence intervals calculated for each mean or medianvalue. The correlations between two continuous variables weredetermined by Pearson test. Receiver Operator Characteristic(ROC) curve was constructed to evaluate the level of HOMA-IRin determining GDM group and non GDM group. The diagnosticvalue of HOMA-IR in GDM was established by its sensitivity,specificity and optimal diagnostic cut-off from analysis of theROC curve. In all statistical analyses, p50.05 (95% confidenceinterval) was considered to be significant.

Results

A total of 279 pregnant women with risk of GDM were recruitedfor this study. Of these, 63 (22.58%) had abnormal MGTT basedon WHO criteria and were diagnosed with GDM. Maternaldemographic features were described in Table 1. The mean pre-pregnancy body mass index was significantly higher in diabeticwomen compared to those non GDM women (25.98� 5.24 kg/cm2 versus 24.09� 4.75 kg/cm2; p¼ 0.01).

The study had shown that the mean values for HOMA-IR,fasting insulin, fasting glucose and 2 h glucose were higher inGDM patients than non GDM patients. There was a statisticalsignificant different in these values of the parameters. The meanvalues for HbA1C and fructosamine for glycemic control in GDMpatients were within the acceptable range (Table 1). There wereno statistically significant results when risk factors were beingcompared in these women (Table 2).

ROC analysis was performed to evaluate the diagnosticperformance of the HOMA-IR against MGTT to diagnose GDMpatients, as well as to determine the appropriate cut-off value forour population in this study. The area under the ROC curve(AUC) for HOMA-IR to diagnose GDM was 0.789 [95%confidence interval (CI), 0.737–0.836]. The cut-off point ofHOMA-IR level that maximized sensitivity (63.49%) and speci-ficity (89.81%) was 2.92.

AUC for fasting glucose was 0.757 [95% CI, 0.703–0.806]. Thecut-off point for fasting glucose was 4.8 mmol/L, which maximizedboth sensitivity (63.49%) and specificity (80.09%). However, thisvalue was lower than the hospital protocol cut-off value. For fastinginsulin, the AUC was 0.694 [95% CI, 0.694–0.799).

Using the cut-off level of 2.92 in those with GDM, had shownstatistically significant results among neonatal complication in thedevelopment of hypoglycemia. Similarly, Cesarean section wasalso a significant maternal complication (Table 3).

Discussion

All over the world including the Asian countries, the prevalenceof Diabetes Mellitus (DM) has increased tremendously [7] thusgiving rise into a critical medical problem to be addressedurgently. GDM is a renowned ‘window period’ before a full-blown feature of DM sets in at a later period of life time [8,9].Thus, recognition of early insulin resistance among GDM womenin anticipation of possible complications must escalate as GDMinherits its trans-generational potential. The insulin resistancefeature also contributes to basic mechanism of the disease processthat must be tackled at forefront. Although insulin resistance haslong been the topic of research in this area, there is no universalcut off value and this keeps changing between different parts ofthe world with different ethnicity [10].

In this current study, among the cohort of relatively homo-genous women, 22.58% of the women developed GDM which wasin agreement to previous local studies [11,12]. Multiparouswomen were noted to be the majority of them with similarly to bethe majority for non-GDM and GDM. The homogeneity wasproven when the risk factors (Tables 1 and 2) among the non-GDM and GDM women were scrutinized, revealing no factorswere statistically significant different except for pre pregnancyBMI (25.98� 5.24 kg/cm2 in GDM versus 24.09� 4.75 kg/cm2 innon-GDM group; p¼ 0.01). The results of fasting insulin andHOMA-IR were significantly higher in the GDM group (Table 1),suggesting of a higher insulin resistance in GDM patients whichwas also in agreement with previous studies [13,14] despitehaving had a good glycemic control. Other studies [15,16] hadalso proven that GDM patients had a higher insulin resistancecompared to those non-GDM and the resistance was worsening asthe pregnancy advances. This somehow could also be attributed tothe women obesity which was seen in our GDM patients andsimilarly seen in other studies [16].

As different ethnicity showed different preponderance forGDM [4] which may also be reflected by genetic inheritance [17],HOMA-IR cut off value in our patients showed a higher level(2.92) compared to the cut-off recommended by IDF, which was2.38. This study value was also among the highest insulinresistance compared to previous studies [13,14] in Turkey exceptfor those pregnant Arab women in Scandinavian study [18]. Basedon the ROC curve, a comparison was made between HOMA-IR,fasting glucose and fasting insulin and HOMA-IR was shown tobe more specific than both fasting glucose and fasting insulinseparately.

In this current study, HOMA-IR has the best sensitivity(63.49%) and specificity (89.81%) with the cut-off value of 2.92,in comparison to the cut-off value of 2.24 taken from Gokcel et al.

692 N. A. M. Ismail et al. Gynecol Endocrinol, 2013; 29(7): 691–694

Gyn

ecol

End

ocri

nol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 11

/23/

14Fo

r pe

rson

al u

se o

nly.

study [19] (the sensitivity and specificity were 76.19% and64.81%, respectively). By applying Ozcimen et al. [14] cut-offvalue (2.60) on the other hand, the HOMA-IR sensitivity andspecificity would be 66.67% and 73.61%, respectively.

The pregnancy outcomes were analyzed using the cut-off valueof 2.92, and it was noted to be significantly associated with thedevelopment of neonatal hypoglycemia (Table 3). This perhapscould be explained by a higher level of insulin in the diabeticmothers causing hyperplasia of fetal pancreatic b-cells andsubsequent fetal hyperglycemia. The neonate develops hypogly-cemia due to its own insufficient glucose after birth. Cesareansection showed the significant maternal complication using thiscut off value, with majority (83%) had HOMA-IR value�2.92. The other 17% with the HOMA-IR value52.92 also hadCesarean section, which may be contributed by other factorsbesides GDM.

As the sample size of women with GDM was limited in thisstudy and not representing the true population, the HOMA-IR cut-off value obtained from this study, is still insufficient to be used asa recommended tool for diagnosis of GDM. However, it is hopedthat this study will be a platform for researchers to conduct a

larger scale research as to come upon a consensus for values ofHOMA-IR in different ethnic groups.

In conclusion, HOMA-IR score and insulin resistance levelswere higher in GDM women in our population. With the cut-offHOMA-IR value of 2.92, neonatal hypoglycemia and Cesareansection were significant complications in GDM mothers. This canbe used in anticipation of maternal and fetal morbiditiesespecially in the Asian population.

Acknowledgements

The authors wish to thank all patients who participated in this study andstaff of the Specialised Endocrine Laboratory for their technicalassistance.

Declaration of interest

This study was financially supported by UKMMC Fundamental ResearchGrant (FF-178-2010).

The authors report no conflicts of interest. The authors alone areresponsible for the content and writing of the paper.

Table 1. Maternal demographic and biochemical data.

Diagnosis

GDM n (%) Non-GDM n (%) Total n (%) p

Race Malay 43 (68.3) 170 (78.7) 213 (76.3) 0.10Chinese 19 (30.2) 39 (18.1) 58 (20.8)Indian 1 (1.6) 7 (3.2) 8 (2.9)

Parity Primigravide 13 (20.6) 71 (32.9) 84 (30.1) 0.02**Pseudoprimip 6 (9.5) 8 (3.7) 14 (5.0)Multiparous (2-5) 44 (69.8) 127 (58.8) 171 (61.3)Grandmultip (45) 0 (0) 10 (4.6) 10 (3.6)

#Age (years) 31.62 (4.52) 30.65 (4.83) 0.16#Weight (kg) 63.03 (13.18) 58.64 (12.26) 0.01**#HOMA-IR 4.07 (2.44) 2.08 (1.12) 0.001**#Fasting Insulin (mIU/L) 16.76 (8.63) 10.15 (5.07) 0.001**#Fasting Glucose (mmol/L) 5.33 (1.22) 4.55 (0.42) 0.001**#2 Hours Glucose (mmol/L) 9.08 (1.40) 6.00 (1.04) 0.001**#HbA1C (%) 5.67 (0.61)#Fructosamine (mmol/L) 195.38 (20.29)

#Data presented in mean (SD).**Significant p50.05.

Table 2. Risk factors among women with diabetic risks.

Glucose tolerance

Abnormal NormalRisk factors n (%) n (%) �2 p

I. Family history of diabetes No 25 (39.7) 83 (38.4) 0.03 0.86II. Past pregnancy history Yes 38 (60.3) 133 (61.6)GDM No 57 (90.5) 208 (96.3) 3.47 0.06

Yes 6 (9.5) 8 (3.7)Birth of big baby (44 kg) No 61 (96.8) 208 (96.3) 0.04 0.84

Yes 2 (3.2) 8 (3.7)Polyhydramnios No 62 (98.4) 214 (99.1) 0.20 0.65

Yes 1 (1.6) 2 (0.9)Recurrent abortion No 61 (96.8) 212 (98.1) 0.41 0.52

Yes 2 (3.2) 4 (1.9)Intrauterine death No 62 (98.4) 212 (98.1) 0.02 0.89III. Current Pregnancy Yes 1 (1.6) 4 (1.9)Glycosuria No 56 (88.9) 197 (91.2) 0.31 0.58

Yes 7 (11.1) 19 (8.8)Maternal age �35 No 55 (87.3) 179 (82.9) 0.71 0.40

Yes 8 (12.7) 37 (17.1)Maternal weight480 kg No 35 (55.6) 143 (66.2) 2.39 0.12

Yes 28 (44.4) 73 (33.8)

Table 3. Maternal and neonatal complications in GDM women.

HOMA-IR

52.92 �2.92 �2 p

Neonatal Yes 1 (4.3) 2 (5.0) 0.01 0.91Macrosomic (44 kg) No 22 (95.7) 38 (95.0)Stillbirth No 23 (100) 40 (100) – –Birth Trauma Yes 0 1 (2.5) 0.58 0.45

No 23 (100) 39 (97.5)Hypoglycemia Yes 0 8 (20.0) 5.27 0.02*

No 23 (100) 32 (80.0)Hyperbilirubinemia Yes 5 (21.7) 7 (17.5) 0.17 0.68

No 18 (78.3) 33 (82.5)Polycythemia Yes 1 (4.3) 1 (2.5) 0.16 0.69

No 22 (95.7) 39 (97.5)Maternal Yes 3 (17.6) 15 (32.6) 4.28 0.04*Cesarean section No 20 (82.4) 25 (67.4)Assisted delivery Yes 2 (11.8) 2 (4.3) 0.34 0.56

No 21 (88.2) 38 (95.7)

*Significant p50.05.Data presented: n (%).

DOI: 10.3109/09513590.2013.797398 GDM using HOMA-IR score 693

Gyn

ecol

End

ocri

nol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 11

/23/

14Fo

r pe

rson

al u

se o

nly.

References

1. Buchanan TA, Xiang AH, Kjos SL, Watanabe R. What is gestationaldiabetes? Diabetes Care 2007;30:S105–11.

2. Kjos SL, Buchanan TA. Gestational diabetes mellitus. N Engl J Med1999;341:1749–56.

3. Getahun D, Nath C, Ananth CV, et al. Gestational diabetes in theUnited States; Temporal trends 1989 through 2004. Am J ObstetGynecol 2008;198:525.e1–5.

4. Mørkrid K, Jenum AK, Sletner L, et al. Failure to increase insulinsecretory capacity during pregnancy-induced insulin resistance isassociated with ethnicity and gestational diabetes. Eur J Endocrinol2012;167:579–88.

5. Retnakaran R, Hanley AJ, Connelly PW, et al. Low serumlevels of high-molecular weight adiponectin in Indo-Asianwomen during pregnancy: evidence of ethnic variation inadiponectin isoform distribution. Diabetes Care 2006;29:1377–9.

6. Wallace TM, Levy JC, Matthews DR. Use and abuse of HOMAmodeling. Diabetes Care 2004;27:1487–95.

7. Letchuman GR, Wan Nazaimoon WM, Wan Mohamad WB, et al.Prevalence of diabetes in the Malaysian National Health MorbiditySurvey III 2006. Med J Malaysia 2010;65:180–6.

8. Ben-Haroush A, Yogev Y, Hod M. Epidemiology of gestationaldiabetes mellitus and its association with Type 2 diabetes. DiabetMed 2004;21:103-13.

9. Pappa KI, Gazouli M, Economou K, et al. Gestational diabetesmellitus shares polymorphisms of genes associated with insulinresistance and type 2 diabetes in the Greek population. GynecolEndocrinol 2011;27:267-72.

10. Esteghamati A, Ashraf H, Khalilzadeh O, et al. Optimal cut-off ofhomeostasis model assessment of insulin resistance (HOMA-IR) forthe diagnosis of metabolic syndrome: third national surveillance of

risk factors of non-communicable diseases in Iran (SuRFNCD-2007). Nutr Metabol 2010;7:1-8.

11. Mohamed Ismail NA, Mohd Aris N, Mahdy ZA, et al. Gestationaldiabetes mellitus in primigravidae: a mild disease. Acta Medica2011;54:21-4.

12. Shamsuddin K, Mahdy ZA, Siti Rafiaah I, et al. Risk factorscreening for abnormal glucose tolerance in pregnancy. Int JGynecol Obstet 2001;75:27-32.

13. Yilmaz O, Kucuk M, Ilgin A, Dagdelen M. Assessment of insulinsensitivity/resistance and their relations with leptin concentrationsand anthropometric measures in a pregnant population with andwithout gestational diabetes mellitus. J Diabet Complicat 2010;24:109–14.

14. Ozcimen EE, Uckuyu A, Ciftci FC, et al. Diagnosis of gestationaldiabetes mellitus by use of the homeostasis model assessment-insulin resistance index in the first trimester. Gynaecol Endocrinol2008;24:224–9.

15. Georgiou HM, Lappas M, Georgiou GM, et al. Screening forbiomarkers predictive of gestational diabetes mellitus. Acta Diabetol2008;45:157-65.

16. Endo S, Maeda K, Suto M, et al. Differences in insulin sensitivity inpregnant women with overweight and gestational diabetes mellitus.Gynecol Endocrinol 2006;22:343�9.

17. Nor Khatijah MA, Nor Azlin MI, Zaleha AM, et al. An analysis oftargeted single nucleotide polymorphisms for the risk prediction ofgestational diabetes mellitus in a cohort of Malaysian patients. Asia-Pacific J Mol Med 2011;1:1-8.

18. Shaat N, Ekelundi M, Lernmark A, et al. Genotypic and phenotypicdifferences between Arabian and Scandinavian women with gesta-tional diabetes mellitus. Diabetologia 2004;47:878-84.

19. Gokcel A, Baltali M, Tarim E, et al. Detection of insulin resistancein Turkish adults: a hospital-based study. Diabetes Obes Metab2003;5:126-30.

694 N. A. M. Ismail et al. Gynecol Endocrinol, 2013; 29(7): 691–694

Gyn

ecol

End

ocri

nol D

ownl

oade

d fr

om in

form

ahea

lthca

re.c

om b

y U

nive

rsity

of

Mel

bour

ne o

n 11

/23/

14Fo

r pe

rson

al u

se o

nly.