Journal of Adolescent Health xxx (2014) 1e6
www.jahonline.org Original article
Diagnosing Dysglycemia in Adolescents With Polycystic Ovary Syndrome Holly Catherine Gooding, M.D. a, *, Carly Milliren, M.P.H. d, Michelle St. Paul a, M. Joan Mansfield, M.D. a, b, and Amy DiVasta, M.D. a, c a
Division of Adolescent and Young Adult Medicine, Boston Children’s Hospital, Boston, Massachusetts Division of Endocrinology, Boston Children’s Hospital, Boston, Massachusetts Division of Gynecology, Boston Children’s Hospital, Boston, Massachusetts d Clinical Research Center, Boston Children’s Hospital, Boston, Massachusetts b c
Article history: Received July 12, 2013; Accepted December 18, 2013 Keywords: Adolescent; Polycystic ovarian syndrome; Screening; Diabetes
A B S T R A C T
Purpose: Screening for impaired glucose tolerance (IGT) is recommended for adolescents with polycystic ovary syndrome (PCOS) with oral glucose tolerance test (OGTT). Whether glycated hemoglobin (HbA1c) can be used for screening in this patient population is unknown. We sought to determine the utility of HbA1c and 2-hour OGTT for diagnosing dysglycemia in adolescents with PCOS. Methods: This was a retrospective cohort study of 68 adolescents with PCOS seen in the Boston Children’s Hospital Division of Adolescent Medicine between 2008 and 2011 and not known to have diabetes. Prevalence of dysglycemia (impaired fasting glucose, IGT, increased risk for diabetes, or diabetes mellitus as diagnosed by fasting plasma glucose, 2-hour OGTT, and/or HbA1c) and sensitivity and specificity of HbA1c for diagnosing dysglycemia compared with OGTT were assessed. Results: Twenty-four participants had abnormal glucose testing, including one participant (1.5%) who met criteria for diabetes mellitus and 23 participants (34%) who met criteria for impaired fasting glucose/IGT/prediabetes. More patients were identified as having dysglycemia by HbA1c than OGTT. Compared with OGTT, HbA1c had a sensitivity of 60% and a specificity of 69% for diagnosing dysglycemia. Conclusions: In adolescents with PCOS, HbA1c had moderate sensitivity and specificity for detecting dysglycemia compared with OGTT. Clinicians should be aware that both tests have benefits and limitations, and the optimal test for follow-up requires further study. Ó 2014 Society for Adolescent Health and Medicine. All rights reserved.
PCOS affects 5%e10% of reproductive-age women and often presents during adolescence [1]. Many patients with PCOS are insulin resistant even if they are not overweight or obese [2], placing them at increased risk for developing impaired glucose tolerance (IGT) and progression to diabetes mellitus (DM) [3e5].
Conflicts of Interest: The authors have no conflicts of interest to disclose. * Address correspondence to: Holly Catherine Gooding, M.D., Division of Adolescent and Young Adult Medicine, Boston Children’s Hospital, 300 Longwood Avenue, Boston, MA 02115. E-mail address: [email protected] (H.C. Gooding).
IMPLICATIONS AND CONTRIBUTION
Our study demonstrates that both glycated hemoglobin (HbA1c) and oral glucose tolerance test (OGTT) are useful for identifying adolescents with polycystic ovary syndrome (PCOS) with dysglycemia and may identify different populations of girls at risk for long-term cardiometabolic consequences. Future studies are necessary to determine which test best predicts future dysglycemia or progression to diabetes in longitudinal follow-up.
Because IGT can present as early as adolescence in patients with PCOS [6,7], health care providers are tasked with initiating the evaluation for IGT and DM in these young women. The Androgen Excess Society recommends IGT screening for adolescent and adult women with PCOS who are obese, over 40 years of age, or have a personal history of gestational diabetes or a family history of type 2 diabetes [8] with a 2-hour OGTT, as fasting plasma glucose (FPG) alone is less sensitive in this population [3]. The American Diabetes Association (ADA) and American Academy of Pediatrics recommend screening obese youth with two or more risk factors for diabetes (one of which can be PCOS) with FPG [9].
1054-139X/$ e see front matter Ó 2014 Society for Adolescent Health and Medicine. All rights reserved. http://dx.doi.org/10.1016/j.jadohealth.2013.12.020
2
H.C. Gooding et al. / Journal of Adolescent Health xxx (2014) 1e6
Given that 70% of adolescents with PCOS are obese, the majority will qualify for screening [10]. Despite these recommendations, pediatric adolescent and gynecology providers report evaluating only 25%e60% of adolescents diagnosed with PCOS for IGT/DM [11]. In 2010, the ADA endorsed the use of HbA1c for the diagnosis of diabetes in adults [12]. HbA1c is a reflection of the average blood glucose levels over the life of a red blood cell (approximately 90 days) and is closely associated with microvascular disease in diabetes [13]. HbA1c is less subject to intraindividual variation and does not require that the patient be fasting. This latter advantage may be particularly relevant for adolescent patients, for whom there may be barriers to access to care and who may be less likely to return for follow-up [14]. However, the Androgen Excess Society concluded in their 2010 position paper that there were insufficient data on the use of HbA1c testing in PCOS to recommend its use at that time [8]. Several recent studies have compared screening for dysglycemia by HbA1c with either FPG or OGTT in adult [15e21] and pediatric [22,23] groups, including adult women with PCOS [24,25]. Most of these studies have found only fair agreement among HbA1c, FPG, and OGTT. The role of HbA1c as a screening tool in adolescents with PCOS has not been clarified and may differ from its use in adults due to the lower prevalence of dysglycemia in adolescents. We thus sought to examine the prevalence of dysglycemia as diagnosed by HbA1c and OGTT in adolescents with PCOS to help clinicians identify strategies for clinical practice. Methods Participants We used International Classification of Diseases (ICD-9) billing codes to identify patients seen in our Adolescent and Young Adult Medicine Clinic from 2008 to 2011 who were diagnosed with PCOS by their treating clinician and underwent both a 2-hour OGTT and HbA1c testing within 90 days (N ¼ 68, 45 of whom had both tests on the same day), as well as N ¼ 125 individuals with PCOS who did not have both OGTT and HbA1c testing for a comparison sample. If multiple OGTT or HbA1c results were recorded within the study period, those closest to the initial evaluation were used for analysis. Data on demographic characteristics, social and family history, physical examination findings, laboratory studies, pelvic ultrasound results, and treatment were collected through a standardized chart abstraction tool and entered into the RedCap database [26]. All participants met 1990 National Institutes of Health (NIH) Consensus Criteria for PCOS and thus had irregular menses and either clinical or biochemical hyperandrogenism [27]. Participants were excluded if they had a prior diagnosis of type I or type II diabetes or IGT or if they were taking medications known to affect the results of an OGTT or HbA1c (such as metformin). This study was approved by the Boston Children’s Hospital (BCH) Committee on Clinical Investigations. Demographic information Age (in months/years) was defined by subtracting the date of chart extraction from the participant’s date of birth. Race/ ethnicity was obtained from self-reported data from the electronic patient registration system. Family history was recorded by the treating clinician.
Measures Body mass index (BMI) was calculated using the formula: weight (kg)/height2 (m). Sexual maturity rating of pubertal development for both breast and pubic hair was recorded by the treating clinician. Blood pressure was obtained by GE Dinamap DPC100X-US, measured in the right arm resting at heart level with both feet on the ground per standard clinic policy. Laboratory assays An OGTT was performed in the morning on a random day of the menstrual cycle in 68 fasting individuals who were instructed to consume a normal amount of carbohydrates for the 3 days preceding the test in either the BCH clinical laboratory or another certified laboratory (n ¼ 4). Blood glucose was measured at baseline (FPG) and at 120 minutes (BG 120) after a 75 g oral glucose load. ADA criteria [12] were used to classify participants as having normal fasting glucose or glucose tolerance (FPG < 100 mg/dl or BG 120 < 140 mg/dl), impaired fasting glucose (IFG) or IGT (FPG, 100e125 mg/dl or BG, 120 140e199 mg/dl), or diabetes (FPG 126 mg/dl or BG 120 200 mg/dl) based on their OGTT results. For HbA1c results (turbidimetric inhibition immunoassay, BCH clinical laboratory), we classified subjects as “normal” if HbA1c < 5.7%, “increased risk for diabetes” if HbA1c 5.7%e6.4%, or “diabetic” if HbA1c 6.5% based on ADA criteria. Additional laboratory values (total cholesterol, low density lipoprotein cholesterol, high density lipoprotein cholesterol, triglycerides, alanine aminotransferase, total and free testosterone levels, dehydroepiandrosterone sulfate, and sex hormone binding globulin) were obtained using standard assays in either the BCH clinical laboratory or another certified laboratory. Elevation in testosterone levels was defined as values greater than the normative range in that laboratory for a given patient’s sexual maturity rating. Statistical analysis Descriptive statistics are presented as mean (SD) or frequency (%). Median (interquartile range) are presented in lieu of mean (SD) for continuous factors with non-normal distributions upon visual inspection. Demographic and clinical factors at initial visit were compared by HbA1c categories. All tests were performed at an alpha level of .05 using SAS software, version 9.3 (SAS Institute Inc., Cary, NC). Independent two-tailed t tests were used to assess differences between participants classified as “normal” and “abnormal” by HbA1c on continuous factors. Wilcoxon rank-sum nonparametric test was used to assess differences between HbA1c categories on continuous factors with a non-normal distribution where indicated in results and tables. Chi-square or Fisher’s exact test (where appropriate) was used to assess differences between the HbA1c categories on categorical factors. Diagnostic characteristics of FPG and HbA1c were compared using OGTT in terms of the predictive ability of these tests to classify participants as having “normal” or “abnormal” glucose findings. Measures of the diagnostic ability calculated were sensitivity, specificity, positive and negative likelihood ratios, and the kappa statistic as a measure of overall agreement between the test and the gold standard. Empirically derived cut points that maximize sensitivity and specificity were also generated from the data. Receiver operating characteristic (ROC) curves were constructed using the binary OGTT result classification of “normal”
H.C. Gooding et al. / Journal of Adolescent Health xxx (2014) 1e6
and “abnormal” as the gold standard to determine the predictive ability of FPG and HbA1c at the range of all possible cut points for these tests. To determine whether the test had good predictive ability to discriminate glucose findings, the areas under the ROC curve (AUC) were compared with the null value of .5 (discrimination due to chance) using a ManneWhitney test to construct 95% confidence intervals (CIs) for the AUC. A test of the contrast between the ROC curves for HbA1c to FPG in reference to OGTT was also computed to discern whether the tests differed from each other in terms of discriminatory ability. Results Demographic and clinical characteristics of the 68 patients who completed both OGTT and HbA1c testing within 90 days, stratified by HbA1c category, are listed in Table 1. The median time between OGTT and HbA1c testing was 0 days (interquartile range 23.5). Findings did not differ when analysis was restricted to the 45 individuals who had OGTT and HbA1c testing on the same day. Compared with 125 patients with PCOS seen during the same period who did not have both OGTT and HbA1c testing, patients who had both tests performed had higher BMI Z-scores (1.83 vs. 1.27, p < .01). However, patients with both tests performed were no more likely to have a family history of DM and did not differ from other PCOS patients with respect to race/ ethnicity, age, or free testosterone (pg/dl). Subjects with an HbA1c 5.7% were more likely to be black (p ¼ .03) and to have higher low-density lipoprotein cholesterol (p ¼ .04) compared with individuals with a normal HbA1c. There were no statistically significant differences by HbA1c category in other cardiometabolic risk factors, including blood pressure and BMI, nor in androgen levels or criteria met for PCOS diagnosis. Subjects with an HbA1c 5.7% had higher mean FPG levels and were more likely to have an abnormal 2-hour glucose (Table 2). There were no statistically significant differences by OGTT category in demographic variables, androgen concentrations, homeostasis model assessment of insulin resistance, or cardiometabolic risk factors (data not shown). One subject was classified as having diabetes; this subject met criteria for diabetes by both HbA1c and OGTT. Three patients were classified as having IFG on the basis of an abnormal FPG, while nine subjects were classified as having IGT based upon BG 120. Twentythree patients were classified “at increased risk for diabetes” by HbA1c. Due to the low number of patients with DM in our sample, DM and IFG/IGT/“increased risk for diabetes” were considered together as dysglycemia for all remaining analyses. The majority of patients (N ¼ 22, 84%) with abnormal glucose testing of any type were obese; HbA1c was no more or less likely than OGTT to identify normal or overweight individuals as having dysglycemia. Figure 1 shows the number of patients diagnosed as having dysglycemia by each of the three tests. HbA1c failed to classify four individuals identified as abnormal by BG 120 and one individual identified as abnormal by FPG, but identified 17 individuals as abnormal who were not identified by either component of the OGTT. Only one individual was identified as abnormal by all three tests. Compared with OGTT, HbA1c had a sensitivity of 60.0% and a specificity of 69.0% for diagnosing dysglycemia (Table 3); these values did not differ for the empirically derived cut point of 5.6%. HbA1c was more sensitive but less specific for making the diagnosis of dysglycemia than FPG in this cohort when compared with OGTT. The AUC for continuous HbA1c was .68 (95% CI .49e.87), and the AUC for continuous FPG was .67 (95% CI .48e.86). After
3
Table 1 Characteristics of study population at first test by HbA1c categories (N ¼ 68) Participant characteristic
Age (years) Race White Black or African-American Hispanic or Latino Other/Multiracial Family history of diabetes Family history of PCOS BMI z-score BMI categoryc Normal Overweight Obese Systolic BP (mm Hg) Diastolic BP (mm Hg) Total cholesterol (mg/dl) LDL cholesterol (mg/dl) HDL cholesterol (mg/dl)* Triglycerides (mg/dl)* ALT (IU/L)* Total testosterone (ng/dl) Free testosterone (pg/dl)* DHEAS (mg/dl)* SHBG (nmol/L)* PCOS diagnosisd Clinical hyperandrogenism (n ¼ 61) Biochemical hyperandrogenism (n ¼ 55)
Normal (99 mg/dl) Empirically determined cut point (>87 mg/dl)
Sensitivity (%)
Specificity (%)
þLR
LR
Kappa
c Statistic
60.0 60.0
69.0 69.0
1.93 1.93
.58 .58
.18 .18
.645 .645
10.0 80.0
97.0 62.0
3.33 2.11
.93 .32
.09 .23
.533 .710
FPG ¼ fasting plasma glucose; HbA1c ¼ hemoglobin A1c; þLR ¼ positive likelihood ratio; LR ¼ negative likelihood ratio; OGTT ¼ oral glucose tolerance test; PCOS ¼ polycystic ovary syndrome.
An alternative approach involves measurement of both HbA1c and OGTT at the initial evaluation with repeated measurement of the more abnormal test on a separate day. If IGT or increased risk for diabetes is confirmed on either test, this test could be followed at regular intervals to measure change with lifestyle modification or medication therapy. Either of these strategies has the potential to increase both financial [29] and psychological costs due to falsepositive results. A prospective study comparing these strategies over time would allow assessment of outcomes and cost. Adolescent girls with PCOS are at high risk for dysglycemia. Based on our results, it appears that both HbA1c and OGTT are useful for identifying adolescents with PCOS with dysglycemia and may actually identify different populations of girls at risk for long-term cardiometabolic consequences. Future studies are necessary to determine whether OGTT or HbA1c results best predict future dysglycemia or progression to diabetes in longitudinal follow-up. In the meantime, clinicians need to be aware of the characteristics and limitations of each of the glucose screening tests available and continue to counsel all adolescents with PCOS on the benefits of lifestyle modification for the prevention of diabetes. It is our opinion that the best screening strategy is the one most likely to be completed by clinicians and their patients in order to identify, counsel, and treat those patients at highest risk for diabetes. Acknowledgments The authors thank our patients, their families, and our clinical providers as well as Dr. S. Jean Emans and Dr. Henry Feldman for thoughtful comments on study design and analysis. Dr. Gooding drafted the manuscript; none of the authors were paid an honorarium for this work. Funding Sources Personnel support for data collection was funded in part by the Leadership Education in Adolescent Health Training Program MCH/HSRA T71MC00009 and Dr. DiVasta is supported by NICHD K23HD060066; neither funder directly sponsored the study nor influenced data collection or interpretation. References [1] Orsino A, Van Eyk N, Hamilton J. Clinical features, investigations and management of adolescents with polycystic ovary syndrome. Paediatr Child Health 2005 Dec;10:602e8. [2] Dunaif A, Segal KR, Futterweit W, et al. Profound peripheral insulin resistance, independent of obesity, in polycystic ovary syndrome. Diabetes 1989 Sep;38:1165e74.
[3] Legro RS, Kunselman AR, Dodson WC, et al. Prevalence and predictors of risk for type 2 diabetes mellitus and impaired glucose tolerance in polycystic ovary syndrome: A prospective, controlled study in 254 affected women. J Clin Endocrinol Metab 1999 Jan;84:165e9. [4] Ehrmann DA, Barnes RB, Rosenfield RL, et al. Prevalence of impaired glucose tolerance and diabetes in women with polycystic ovary syndrome. Diabetes Care 1999 Jan;22:141e6. [5] Legro RS, Gnatuk CL, Kunselman AR, et al. Changes in glucose tolerance over time in women with polycystic ovary syndrome: A controlled study. J Clin Endocrinol Metab 2005 Jun;90:3236e42. [6] Palmert MR, Gordon CM, Kartashov AI, et al. Screening for abnormal glucose tolerance in adolescents with polycystic ovary syndrome. J Clin Endocrinol Metab 2002 Mar;87:1017e23. [7] Nur MM, Newman IM, Siqueira LM. Glucose metabolism in overweight Hispanic adolescents with and without polycystic ovary syndrome. Pediatrics 2009 Sep;124:e496e502. [8] Wild RA, Carmina E, Diamanti-Kandarakis E, et al. Assessment of cardiovascular risk and prevention of cardiovascular disease in women with the polycystic ovary syndrome: A consensus statement by the Androgen Excess and Polycystic Ovary Syndrome (AE-PCOS) Society. J Clin Endocrinol Metab 2010 May;95:2038e49. [9] Type 2 diabetes in children and adolescents. American Diabetes Association. Pediatrics 2000 Mar;105(3 Pt 1):671e80. [10] Bekx MT, Connor EC, Allen DB. Characteristics of adolescents presenting to a multidisciplinary clinic for polycystic ovarian syndrome. J Pediatr Adolesc Gynecol 2010 Feb;23:7e10. [11] Bonny AE, Appelbaum H, Connor EL, et al. Clinical variability in approaches to polycystic ovary syndrome. J Pediatr Adolesc Gynecol 2012 Aug;25: 259e61. [12] American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care 2010 Jan;33(Suppl 1):S62e9. [13] Sacks DB. A1C versus glucose testing: A comparison. Diabetes Care 2011 Feb;34:518e23. [14] Peters A, Laffel L. Diabetes care for emerging adults: Recommendations for transition from pediatric to adult diabetes care systems: A position statement of the American diabetes Association, with representation by the American College of Osteopathic Family Physicians, the American Academy of Pediatrics, the American Association of Clinical Endocrinologists, the American Osteopathic Association, the Centers for Disease Control and Prevention, Children with Diabetes, The Endocrine Society, the International Society for Pediatric and Adolescent Diabetes, Juvenile Diabetes Research Foundation International, the National Diabetes Education Program, and the Pediatric Endocrine Society (formerly Lawson Wilkins Pediatric Endocrine Society). Diabetes Care 2011 Nov;34:2477e85. [15] Heianza Y, Hara S, Arase Y, et al. HbA1c 5.7-6.4% and impaired fasting plasma glucose for diagnosis of prediabetes and risk of progression to diabetes in Japan (TOPICS 3): A longitudinal cohort study. Lancet 2011 Jul 9;378:147e55. [16] Jorgensen ME, Bjerregaard P, Borch-Johnsen K, et al. New diagnostic criteria for diabetes: Is the change from glucose to HbA1c possible in all populations? J Clin Endocrinol Metab 2010 Nov;95:E333e6. [17] Carson AP, Reynolds K, Fonseca VA, et al. Comparison of A1C and fasting glucose criteria to diagnose diabetes among U.S. adults. Diabetes Care 2010 Jan;33:95e7. [18] Lipska KJ, De Rekeneire N, Van Ness PH, et al. Identifying dysglycemic states in older adults: Implications of the emerging use of hemoglobin A1c. J Clin Endocrinol Metab 2010 Dec;95:5289e95. [19] Olson DE, Rhee MK, Herrick K, et al. Screening for diabetes and prediabetes with proposed A1C-based diagnostic criteria. Diabetes Care 2010 Oct;33:2184e9. [20] James C, Bullard KM, Rolka DB, et al. Implications of alternative definitions of prediabetes for prevalence in U.S. adults. Diabetes Care 2011 Feb;34: 387e91.
6
H.C. Gooding et al. / Journal of Adolescent Health xxx (2014) 1e6
[21] Mann DM, Carson AP, Shimbo D, et al. Impact of A1C screening criterion on the diagnosis of pre-diabetes among U.S. adults. Diabetes Care 2011 Oct; 33:2190e5. [22] Nowicka P, Santoro N, Liu H, et al. Utility of hemoglobin A(1c) for diagnosing prediabetes and diabetes in obese children and adolescents. Diabetes Care 2011 Jun;34:1306e11. [23] Lee JM, Wu EL, Tarini B, et al. Diagnosis of diabetes using hemoglobin A1c: Should recommendations in adults be extrapolated to adolescents? J Pediatr 2011 Jun;158:947e52. e941-943. [24] Hurd WW, Abdel-Rahman MY, Ismail SA, et al. Comparison of diabetes mellitus and insulin resistance screening methods for women with polycystic ovary syndrome. Fertil Steril 2011 Oct;96:1043e7. [25] Velling Magnussen L, Mumm H, Andersen M, et al. Hemoglobin A1c as a tool for the diagnosis of type 2 diabetes in 208 premenopausal women with polycystic ovary syndrome. Fertil Steril 2011 Nov;96:1275e80. [26] Harris PA, Taylor R, Thielke R, et al. Research electronic data capture (REDCap)ea metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009 Apr;42:377e81. [27] Zawadski J, Dunaif A. Diagnostic criteria for polycystic ovary syndrome: Towards a rational approach. In: Dunaif A, Givens JR, Haseltine FP, et al., eds.
[28]
[29]
[30]
[31] [32]
[33]
[34]
Polycystic Ovary Syndrome. Boston, MA: Blackwell Scientific Publications; 1992:377e84. Kester LM, Hey H, Hannon TS. Using hemoglobin A1c for prediabetes and diabetes diagnosis in adolescents: Can adult recommendations be upheld for pediatric use? J Adolesc Health 2012 Apr;50:321e3. Wu EL, Kazzi NG, Lee JM. Cost-effectiveness of screening strategies for identifying pediatric diabetes mellitus and dysglycemia. JAMA Pediatr 2013;167:32e9. Libman IM, Barinas-Mitchell E, Bartucci A, et al. Reproducibility of the oral glucose tolerance test in overweight children. J Clin Endocrinol Metab 2008 Nov;93:4231e7. Inzucchi SE. Clinical practice. Diagnosis of diabetes. N Engl J Med 2012 Aug 9;367:542e50. Forbes LE, Storey KE, Fraser SN, et al. Dietary patterns associated with glycemic index and glycemic load among Alberta adolescents. Appl Physiol Nutr Metab 2009 Aug;34:648e58. Herman WH, Cohen RM. Racial and ethnic differences in the relationship between HbA1c and blood glucose: Implications for the diagnosis of diabetes. J Clin Endocrinol Metab 2012 Apr;97:1067e72. Lee JM, Eason A, Nelson C, et al. Screening practices for identifying type 2 diabetes in adolescents. J Adolesc Health 2013 Aug 19.
www.jahonline.org Original article
Diagnosing Dysglycemia in Adolescents With Polycystic Ovary Syndrome Holly Catherine Gooding, M.D. a, *, Carly Milliren, M.P.H. d, Michelle St. Paul a, M. Joan Mansfield, M.D. a, b, and Amy DiVasta, M.D. a, c a
Division of Adolescent and Young Adult Medicine, Boston Children’s Hospital, Boston, Massachusetts Division of Endocrinology, Boston Children’s Hospital, Boston, Massachusetts Division of Gynecology, Boston Children’s Hospital, Boston, Massachusetts d Clinical Research Center, Boston Children’s Hospital, Boston, Massachusetts b c
Article history: Received July 12, 2013; Accepted December 18, 2013 Keywords: Adolescent; Polycystic ovarian syndrome; Screening; Diabetes
A B S T R A C T
Purpose: Screening for impaired glucose tolerance (IGT) is recommended for adolescents with polycystic ovary syndrome (PCOS) with oral glucose tolerance test (OGTT). Whether glycated hemoglobin (HbA1c) can be used for screening in this patient population is unknown. We sought to determine the utility of HbA1c and 2-hour OGTT for diagnosing dysglycemia in adolescents with PCOS. Methods: This was a retrospective cohort study of 68 adolescents with PCOS seen in the Boston Children’s Hospital Division of Adolescent Medicine between 2008 and 2011 and not known to have diabetes. Prevalence of dysglycemia (impaired fasting glucose, IGT, increased risk for diabetes, or diabetes mellitus as diagnosed by fasting plasma glucose, 2-hour OGTT, and/or HbA1c) and sensitivity and specificity of HbA1c for diagnosing dysglycemia compared with OGTT were assessed. Results: Twenty-four participants had abnormal glucose testing, including one participant (1.5%) who met criteria for diabetes mellitus and 23 participants (34%) who met criteria for impaired fasting glucose/IGT/prediabetes. More patients were identified as having dysglycemia by HbA1c than OGTT. Compared with OGTT, HbA1c had a sensitivity of 60% and a specificity of 69% for diagnosing dysglycemia. Conclusions: In adolescents with PCOS, HbA1c had moderate sensitivity and specificity for detecting dysglycemia compared with OGTT. Clinicians should be aware that both tests have benefits and limitations, and the optimal test for follow-up requires further study. Ó 2014 Society for Adolescent Health and Medicine. All rights reserved.
PCOS affects 5%e10% of reproductive-age women and often presents during adolescence [1]. Many patients with PCOS are insulin resistant even if they are not overweight or obese [2], placing them at increased risk for developing impaired glucose tolerance (IGT) and progression to diabetes mellitus (DM) [3e5].
Conflicts of Interest: The authors have no conflicts of interest to disclose. * Address correspondence to: Holly Catherine Gooding, M.D., Division of Adolescent and Young Adult Medicine, Boston Children’s Hospital, 300 Longwood Avenue, Boston, MA 02115. E-mail address: [email protected] (H.C. Gooding).
IMPLICATIONS AND CONTRIBUTION
Our study demonstrates that both glycated hemoglobin (HbA1c) and oral glucose tolerance test (OGTT) are useful for identifying adolescents with polycystic ovary syndrome (PCOS) with dysglycemia and may identify different populations of girls at risk for long-term cardiometabolic consequences. Future studies are necessary to determine which test best predicts future dysglycemia or progression to diabetes in longitudinal follow-up.
Because IGT can present as early as adolescence in patients with PCOS [6,7], health care providers are tasked with initiating the evaluation for IGT and DM in these young women. The Androgen Excess Society recommends IGT screening for adolescent and adult women with PCOS who are obese, over 40 years of age, or have a personal history of gestational diabetes or a family history of type 2 diabetes [8] with a 2-hour OGTT, as fasting plasma glucose (FPG) alone is less sensitive in this population [3]. The American Diabetes Association (ADA) and American Academy of Pediatrics recommend screening obese youth with two or more risk factors for diabetes (one of which can be PCOS) with FPG [9].
1054-139X/$ e see front matter Ó 2014 Society for Adolescent Health and Medicine. All rights reserved. http://dx.doi.org/10.1016/j.jadohealth.2013.12.020
2
H.C. Gooding et al. / Journal of Adolescent Health xxx (2014) 1e6
Given that 70% of adolescents with PCOS are obese, the majority will qualify for screening [10]. Despite these recommendations, pediatric adolescent and gynecology providers report evaluating only 25%e60% of adolescents diagnosed with PCOS for IGT/DM [11]. In 2010, the ADA endorsed the use of HbA1c for the diagnosis of diabetes in adults [12]. HbA1c is a reflection of the average blood glucose levels over the life of a red blood cell (approximately 90 days) and is closely associated with microvascular disease in diabetes [13]. HbA1c is less subject to intraindividual variation and does not require that the patient be fasting. This latter advantage may be particularly relevant for adolescent patients, for whom there may be barriers to access to care and who may be less likely to return for follow-up [14]. However, the Androgen Excess Society concluded in their 2010 position paper that there were insufficient data on the use of HbA1c testing in PCOS to recommend its use at that time [8]. Several recent studies have compared screening for dysglycemia by HbA1c with either FPG or OGTT in adult [15e21] and pediatric [22,23] groups, including adult women with PCOS [24,25]. Most of these studies have found only fair agreement among HbA1c, FPG, and OGTT. The role of HbA1c as a screening tool in adolescents with PCOS has not been clarified and may differ from its use in adults due to the lower prevalence of dysglycemia in adolescents. We thus sought to examine the prevalence of dysglycemia as diagnosed by HbA1c and OGTT in adolescents with PCOS to help clinicians identify strategies for clinical practice. Methods Participants We used International Classification of Diseases (ICD-9) billing codes to identify patients seen in our Adolescent and Young Adult Medicine Clinic from 2008 to 2011 who were diagnosed with PCOS by their treating clinician and underwent both a 2-hour OGTT and HbA1c testing within 90 days (N ¼ 68, 45 of whom had both tests on the same day), as well as N ¼ 125 individuals with PCOS who did not have both OGTT and HbA1c testing for a comparison sample. If multiple OGTT or HbA1c results were recorded within the study period, those closest to the initial evaluation were used for analysis. Data on demographic characteristics, social and family history, physical examination findings, laboratory studies, pelvic ultrasound results, and treatment were collected through a standardized chart abstraction tool and entered into the RedCap database [26]. All participants met 1990 National Institutes of Health (NIH) Consensus Criteria for PCOS and thus had irregular menses and either clinical or biochemical hyperandrogenism [27]. Participants were excluded if they had a prior diagnosis of type I or type II diabetes or IGT or if they were taking medications known to affect the results of an OGTT or HbA1c (such as metformin). This study was approved by the Boston Children’s Hospital (BCH) Committee on Clinical Investigations. Demographic information Age (in months/years) was defined by subtracting the date of chart extraction from the participant’s date of birth. Race/ ethnicity was obtained from self-reported data from the electronic patient registration system. Family history was recorded by the treating clinician.
Measures Body mass index (BMI) was calculated using the formula: weight (kg)/height2 (m). Sexual maturity rating of pubertal development for both breast and pubic hair was recorded by the treating clinician. Blood pressure was obtained by GE Dinamap DPC100X-US, measured in the right arm resting at heart level with both feet on the ground per standard clinic policy. Laboratory assays An OGTT was performed in the morning on a random day of the menstrual cycle in 68 fasting individuals who were instructed to consume a normal amount of carbohydrates for the 3 days preceding the test in either the BCH clinical laboratory or another certified laboratory (n ¼ 4). Blood glucose was measured at baseline (FPG) and at 120 minutes (BG 120) after a 75 g oral glucose load. ADA criteria [12] were used to classify participants as having normal fasting glucose or glucose tolerance (FPG < 100 mg/dl or BG 120 < 140 mg/dl), impaired fasting glucose (IFG) or IGT (FPG, 100e125 mg/dl or BG, 120 140e199 mg/dl), or diabetes (FPG 126 mg/dl or BG 120 200 mg/dl) based on their OGTT results. For HbA1c results (turbidimetric inhibition immunoassay, BCH clinical laboratory), we classified subjects as “normal” if HbA1c < 5.7%, “increased risk for diabetes” if HbA1c 5.7%e6.4%, or “diabetic” if HbA1c 6.5% based on ADA criteria. Additional laboratory values (total cholesterol, low density lipoprotein cholesterol, high density lipoprotein cholesterol, triglycerides, alanine aminotransferase, total and free testosterone levels, dehydroepiandrosterone sulfate, and sex hormone binding globulin) were obtained using standard assays in either the BCH clinical laboratory or another certified laboratory. Elevation in testosterone levels was defined as values greater than the normative range in that laboratory for a given patient’s sexual maturity rating. Statistical analysis Descriptive statistics are presented as mean (SD) or frequency (%). Median (interquartile range) are presented in lieu of mean (SD) for continuous factors with non-normal distributions upon visual inspection. Demographic and clinical factors at initial visit were compared by HbA1c categories. All tests were performed at an alpha level of .05 using SAS software, version 9.3 (SAS Institute Inc., Cary, NC). Independent two-tailed t tests were used to assess differences between participants classified as “normal” and “abnormal” by HbA1c on continuous factors. Wilcoxon rank-sum nonparametric test was used to assess differences between HbA1c categories on continuous factors with a non-normal distribution where indicated in results and tables. Chi-square or Fisher’s exact test (where appropriate) was used to assess differences between the HbA1c categories on categorical factors. Diagnostic characteristics of FPG and HbA1c were compared using OGTT in terms of the predictive ability of these tests to classify participants as having “normal” or “abnormal” glucose findings. Measures of the diagnostic ability calculated were sensitivity, specificity, positive and negative likelihood ratios, and the kappa statistic as a measure of overall agreement between the test and the gold standard. Empirically derived cut points that maximize sensitivity and specificity were also generated from the data. Receiver operating characteristic (ROC) curves were constructed using the binary OGTT result classification of “normal”
H.C. Gooding et al. / Journal of Adolescent Health xxx (2014) 1e6
and “abnormal” as the gold standard to determine the predictive ability of FPG and HbA1c at the range of all possible cut points for these tests. To determine whether the test had good predictive ability to discriminate glucose findings, the areas under the ROC curve (AUC) were compared with the null value of .5 (discrimination due to chance) using a ManneWhitney test to construct 95% confidence intervals (CIs) for the AUC. A test of the contrast between the ROC curves for HbA1c to FPG in reference to OGTT was also computed to discern whether the tests differed from each other in terms of discriminatory ability. Results Demographic and clinical characteristics of the 68 patients who completed both OGTT and HbA1c testing within 90 days, stratified by HbA1c category, are listed in Table 1. The median time between OGTT and HbA1c testing was 0 days (interquartile range 23.5). Findings did not differ when analysis was restricted to the 45 individuals who had OGTT and HbA1c testing on the same day. Compared with 125 patients with PCOS seen during the same period who did not have both OGTT and HbA1c testing, patients who had both tests performed had higher BMI Z-scores (1.83 vs. 1.27, p < .01). However, patients with both tests performed were no more likely to have a family history of DM and did not differ from other PCOS patients with respect to race/ ethnicity, age, or free testosterone (pg/dl). Subjects with an HbA1c 5.7% were more likely to be black (p ¼ .03) and to have higher low-density lipoprotein cholesterol (p ¼ .04) compared with individuals with a normal HbA1c. There were no statistically significant differences by HbA1c category in other cardiometabolic risk factors, including blood pressure and BMI, nor in androgen levels or criteria met for PCOS diagnosis. Subjects with an HbA1c 5.7% had higher mean FPG levels and were more likely to have an abnormal 2-hour glucose (Table 2). There were no statistically significant differences by OGTT category in demographic variables, androgen concentrations, homeostasis model assessment of insulin resistance, or cardiometabolic risk factors (data not shown). One subject was classified as having diabetes; this subject met criteria for diabetes by both HbA1c and OGTT. Three patients were classified as having IFG on the basis of an abnormal FPG, while nine subjects were classified as having IGT based upon BG 120. Twentythree patients were classified “at increased risk for diabetes” by HbA1c. Due to the low number of patients with DM in our sample, DM and IFG/IGT/“increased risk for diabetes” were considered together as dysglycemia for all remaining analyses. The majority of patients (N ¼ 22, 84%) with abnormal glucose testing of any type were obese; HbA1c was no more or less likely than OGTT to identify normal or overweight individuals as having dysglycemia. Figure 1 shows the number of patients diagnosed as having dysglycemia by each of the three tests. HbA1c failed to classify four individuals identified as abnormal by BG 120 and one individual identified as abnormal by FPG, but identified 17 individuals as abnormal who were not identified by either component of the OGTT. Only one individual was identified as abnormal by all three tests. Compared with OGTT, HbA1c had a sensitivity of 60.0% and a specificity of 69.0% for diagnosing dysglycemia (Table 3); these values did not differ for the empirically derived cut point of 5.6%. HbA1c was more sensitive but less specific for making the diagnosis of dysglycemia than FPG in this cohort when compared with OGTT. The AUC for continuous HbA1c was .68 (95% CI .49e.87), and the AUC for continuous FPG was .67 (95% CI .48e.86). After
3
Table 1 Characteristics of study population at first test by HbA1c categories (N ¼ 68) Participant characteristic
Age (years) Race White Black or African-American Hispanic or Latino Other/Multiracial Family history of diabetes Family history of PCOS BMI z-score BMI categoryc Normal Overweight Obese Systolic BP (mm Hg) Diastolic BP (mm Hg) Total cholesterol (mg/dl) LDL cholesterol (mg/dl) HDL cholesterol (mg/dl)* Triglycerides (mg/dl)* ALT (IU/L)* Total testosterone (ng/dl) Free testosterone (pg/dl)* DHEAS (mg/dl)* SHBG (nmol/L)* PCOS diagnosisd Clinical hyperandrogenism (n ¼ 61) Biochemical hyperandrogenism (n ¼ 55)
Normal (99 mg/dl) Empirically determined cut point (>87 mg/dl)
Sensitivity (%)
Specificity (%)
þLR
LR
Kappa
c Statistic
60.0 60.0
69.0 69.0
1.93 1.93
.58 .58
.18 .18
.645 .645
10.0 80.0
97.0 62.0
3.33 2.11
.93 .32
.09 .23
.533 .710
FPG ¼ fasting plasma glucose; HbA1c ¼ hemoglobin A1c; þLR ¼ positive likelihood ratio; LR ¼ negative likelihood ratio; OGTT ¼ oral glucose tolerance test; PCOS ¼ polycystic ovary syndrome.
An alternative approach involves measurement of both HbA1c and OGTT at the initial evaluation with repeated measurement of the more abnormal test on a separate day. If IGT or increased risk for diabetes is confirmed on either test, this test could be followed at regular intervals to measure change with lifestyle modification or medication therapy. Either of these strategies has the potential to increase both financial [29] and psychological costs due to falsepositive results. A prospective study comparing these strategies over time would allow assessment of outcomes and cost. Adolescent girls with PCOS are at high risk for dysglycemia. Based on our results, it appears that both HbA1c and OGTT are useful for identifying adolescents with PCOS with dysglycemia and may actually identify different populations of girls at risk for long-term cardiometabolic consequences. Future studies are necessary to determine whether OGTT or HbA1c results best predict future dysglycemia or progression to diabetes in longitudinal follow-up. In the meantime, clinicians need to be aware of the characteristics and limitations of each of the glucose screening tests available and continue to counsel all adolescents with PCOS on the benefits of lifestyle modification for the prevention of diabetes. It is our opinion that the best screening strategy is the one most likely to be completed by clinicians and their patients in order to identify, counsel, and treat those patients at highest risk for diabetes. Acknowledgments The authors thank our patients, their families, and our clinical providers as well as Dr. S. Jean Emans and Dr. Henry Feldman for thoughtful comments on study design and analysis. Dr. Gooding drafted the manuscript; none of the authors were paid an honorarium for this work. Funding Sources Personnel support for data collection was funded in part by the Leadership Education in Adolescent Health Training Program MCH/HSRA T71MC00009 and Dr. DiVasta is supported by NICHD K23HD060066; neither funder directly sponsored the study nor influenced data collection or interpretation. References [1] Orsino A, Van Eyk N, Hamilton J. Clinical features, investigations and management of adolescents with polycystic ovary syndrome. Paediatr Child Health 2005 Dec;10:602e8. [2] Dunaif A, Segal KR, Futterweit W, et al. Profound peripheral insulin resistance, independent of obesity, in polycystic ovary syndrome. Diabetes 1989 Sep;38:1165e74.
[3] Legro RS, Kunselman AR, Dodson WC, et al. Prevalence and predictors of risk for type 2 diabetes mellitus and impaired glucose tolerance in polycystic ovary syndrome: A prospective, controlled study in 254 affected women. J Clin Endocrinol Metab 1999 Jan;84:165e9. [4] Ehrmann DA, Barnes RB, Rosenfield RL, et al. Prevalence of impaired glucose tolerance and diabetes in women with polycystic ovary syndrome. Diabetes Care 1999 Jan;22:141e6. [5] Legro RS, Gnatuk CL, Kunselman AR, et al. Changes in glucose tolerance over time in women with polycystic ovary syndrome: A controlled study. J Clin Endocrinol Metab 2005 Jun;90:3236e42. [6] Palmert MR, Gordon CM, Kartashov AI, et al. Screening for abnormal glucose tolerance in adolescents with polycystic ovary syndrome. J Clin Endocrinol Metab 2002 Mar;87:1017e23. [7] Nur MM, Newman IM, Siqueira LM. Glucose metabolism in overweight Hispanic adolescents with and without polycystic ovary syndrome. Pediatrics 2009 Sep;124:e496e502. [8] Wild RA, Carmina E, Diamanti-Kandarakis E, et al. Assessment of cardiovascular risk and prevention of cardiovascular disease in women with the polycystic ovary syndrome: A consensus statement by the Androgen Excess and Polycystic Ovary Syndrome (AE-PCOS) Society. J Clin Endocrinol Metab 2010 May;95:2038e49. [9] Type 2 diabetes in children and adolescents. American Diabetes Association. Pediatrics 2000 Mar;105(3 Pt 1):671e80. [10] Bekx MT, Connor EC, Allen DB. Characteristics of adolescents presenting to a multidisciplinary clinic for polycystic ovarian syndrome. J Pediatr Adolesc Gynecol 2010 Feb;23:7e10. [11] Bonny AE, Appelbaum H, Connor EL, et al. Clinical variability in approaches to polycystic ovary syndrome. J Pediatr Adolesc Gynecol 2012 Aug;25: 259e61. [12] American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care 2010 Jan;33(Suppl 1):S62e9. [13] Sacks DB. A1C versus glucose testing: A comparison. Diabetes Care 2011 Feb;34:518e23. [14] Peters A, Laffel L. Diabetes care for emerging adults: Recommendations for transition from pediatric to adult diabetes care systems: A position statement of the American diabetes Association, with representation by the American College of Osteopathic Family Physicians, the American Academy of Pediatrics, the American Association of Clinical Endocrinologists, the American Osteopathic Association, the Centers for Disease Control and Prevention, Children with Diabetes, The Endocrine Society, the International Society for Pediatric and Adolescent Diabetes, Juvenile Diabetes Research Foundation International, the National Diabetes Education Program, and the Pediatric Endocrine Society (formerly Lawson Wilkins Pediatric Endocrine Society). Diabetes Care 2011 Nov;34:2477e85. [15] Heianza Y, Hara S, Arase Y, et al. HbA1c 5.7-6.4% and impaired fasting plasma glucose for diagnosis of prediabetes and risk of progression to diabetes in Japan (TOPICS 3): A longitudinal cohort study. Lancet 2011 Jul 9;378:147e55. [16] Jorgensen ME, Bjerregaard P, Borch-Johnsen K, et al. New diagnostic criteria for diabetes: Is the change from glucose to HbA1c possible in all populations? J Clin Endocrinol Metab 2010 Nov;95:E333e6. [17] Carson AP, Reynolds K, Fonseca VA, et al. Comparison of A1C and fasting glucose criteria to diagnose diabetes among U.S. adults. Diabetes Care 2010 Jan;33:95e7. [18] Lipska KJ, De Rekeneire N, Van Ness PH, et al. Identifying dysglycemic states in older adults: Implications of the emerging use of hemoglobin A1c. J Clin Endocrinol Metab 2010 Dec;95:5289e95. [19] Olson DE, Rhee MK, Herrick K, et al. Screening for diabetes and prediabetes with proposed A1C-based diagnostic criteria. Diabetes Care 2010 Oct;33:2184e9. [20] James C, Bullard KM, Rolka DB, et al. Implications of alternative definitions of prediabetes for prevalence in U.S. adults. Diabetes Care 2011 Feb;34: 387e91.
6
H.C. Gooding et al. / Journal of Adolescent Health xxx (2014) 1e6
[21] Mann DM, Carson AP, Shimbo D, et al. Impact of A1C screening criterion on the diagnosis of pre-diabetes among U.S. adults. Diabetes Care 2011 Oct; 33:2190e5. [22] Nowicka P, Santoro N, Liu H, et al. Utility of hemoglobin A(1c) for diagnosing prediabetes and diabetes in obese children and adolescents. Diabetes Care 2011 Jun;34:1306e11. [23] Lee JM, Wu EL, Tarini B, et al. Diagnosis of diabetes using hemoglobin A1c: Should recommendations in adults be extrapolated to adolescents? J Pediatr 2011 Jun;158:947e52. e941-943. [24] Hurd WW, Abdel-Rahman MY, Ismail SA, et al. Comparison of diabetes mellitus and insulin resistance screening methods for women with polycystic ovary syndrome. Fertil Steril 2011 Oct;96:1043e7. [25] Velling Magnussen L, Mumm H, Andersen M, et al. Hemoglobin A1c as a tool for the diagnosis of type 2 diabetes in 208 premenopausal women with polycystic ovary syndrome. Fertil Steril 2011 Nov;96:1275e80. [26] Harris PA, Taylor R, Thielke R, et al. Research electronic data capture (REDCap)ea metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform 2009 Apr;42:377e81. [27] Zawadski J, Dunaif A. Diagnostic criteria for polycystic ovary syndrome: Towards a rational approach. In: Dunaif A, Givens JR, Haseltine FP, et al., eds.
[28]
[29]
[30]
[31] [32]
[33]
[34]
Polycystic Ovary Syndrome. Boston, MA: Blackwell Scientific Publications; 1992:377e84. Kester LM, Hey H, Hannon TS. Using hemoglobin A1c for prediabetes and diabetes diagnosis in adolescents: Can adult recommendations be upheld for pediatric use? J Adolesc Health 2012 Apr;50:321e3. Wu EL, Kazzi NG, Lee JM. Cost-effectiveness of screening strategies for identifying pediatric diabetes mellitus and dysglycemia. JAMA Pediatr 2013;167:32e9. Libman IM, Barinas-Mitchell E, Bartucci A, et al. Reproducibility of the oral glucose tolerance test in overweight children. J Clin Endocrinol Metab 2008 Nov;93:4231e7. Inzucchi SE. Clinical practice. Diagnosis of diabetes. N Engl J Med 2012 Aug 9;367:542e50. Forbes LE, Storey KE, Fraser SN, et al. Dietary patterns associated with glycemic index and glycemic load among Alberta adolescents. Appl Physiol Nutr Metab 2009 Aug;34:648e58. Herman WH, Cohen RM. Racial and ethnic differences in the relationship between HbA1c and blood glucose: Implications for the diagnosis of diabetes. J Clin Endocrinol Metab 2012 Apr;97:1067e72. Lee JM, Eason A, Nelson C, et al. Screening practices for identifying type 2 diabetes in adolescents. J Adolesc Health 2013 Aug 19.
