2830 Diabetes Care Volume 37, October 2014

Felix O. Dupont,1 Marie-France Hivert,1,2,3 Glycation of Fetal Hemoglobin Catherine Allard,1 Julie Menard,´ 1 Patrice Perron,1,2 Luigi Bouchard,1,4,5 Reflects Hyperglycemia Exposure Julie Robitaille,6 Jean-Charles Pasquier,1,7 Christiane Auray-Blais,1,8 and In Utero Jean-Luc Ardilouze1,2 Diabetes Care 2014;37:2830–2833 | DOI: 10.2337/dc14-0549

OBJECTIVE The lifetime risk of metabolic diseases in offspring of women with gestational diabetes mellitus (GDM) depends, at least in part, on the impact of glycemic fetal programming. To quantify this impact, we have developed and validated a unique mass spectrometry method to measure the percentage of glycated hemoglobin in cord blood.

RESEARCH DESIGN AND METHODS This case-control study includes 37 GDM women and 30 pregnant women with normal glucose tolerance (NGT).

RESULTS Glycation of the a-chain (Gla) was higher in neonates from GDM (2.32 vs. 2.20%, P < 0.01). Gla strongly correlated with maternal A1C measured at delivery in the r P < r P < overall cohort ( =0.67, 0.0001) as well as in each group (GDM: = 0.66, 1 r P Centre de Recherche du Centre Hospitalier Uni- 0.0001; NGT: = 0.50, = 0.01). versitaire de Sherbrooke, Sherbrooke, QC, Canada 2Endocrine Division, Universite´ de Sherbrooke, CONCLUSIONS Sherbrooke, QC, Canada Thus, Gla may reflect hyperglycemic exposure during the last weeks of fetal de- 3Harvard Pilgrim Health Care Institute, Depart- velopment. Future studies will confirm Gla is a predictive biomarker of prenatally ment of Population Medicine, Harvard Medical School, Boston, MA programmed lifetime metabolic health and disease. 4Department of Biochemistry, Universitede´ Sherbrooke, Sherbrooke, QC, Canada The fetal programming theory suggests that maternal hyperglycemia during pregnancy 5ECOGENE-21 and Lipid Clinic, Chicoutimi Hospi- tal, Saguenay, QC, Canada has lifelong consequences for metabolic health in offspring (1,2). However, assessing 6 fl Department of Food Science and Nutrition, fetal impacts poses a challenge. Birth weight is often used as a marker but is in uenced Laval University, Quebec, QC, Canada by multiple determinants (3). Cord blood glucose, C-peptide, and insulin are also 7Department of Gynaecology-Obstetrics, Uni- measured, albeit only once upon delivery (4–6), despite labor-associated stress and versite´ de Sherbrooke, Sherbrooke, QC, Canada exercise affecting fetal glucose metabolism in opposite ways. A marker of long-term 8Department of Pediatrics, UniversitedeSher-´ fetal glucose metabolism is needed. Although fetal A1C was a promising candidate, its brooke, Sherbrooke, QC, Canada NOVEL COMMUNICATIONS IN DIABETES measurement consistently resisted standard methods because of technical difficulties Corresponding author: Jean-Luc Ardilouze, jean-luc. [email protected]. in face of wide biological variations (7). We have recently developed an accurate method to measure glycation of hemo- Received 2 March 2014 and accepted 26 June 2014. globin (Hb) chains in fetal cord blood (8). Here, we compare results from neonates Clinical trial reg. no. NCT01025973, clinicaltrials born to women with gestational diabetes mellitus (GDM) and women with normal .gov. glucose tolerance (NGT). F.O.D. and M.-F.H. contributed equally to this RESEARCH DESIGN AND METHODS study. © 2014 by the American Diabetes Association. Research Design Readers may use this article as long as the work This case-control study includes 37 women diagnosed and treated for GDM (all were is properly cited, the use is educational and not under insulin therapy) and 30 pregnant women classified as NGT. This case-control for profit, and the work is not altered. care.diabetesjournals.org Dupont and Associates 2831

study included all neonates previously in coefficients of variation (CVs) for Gla delivered earlier (38.5 6 0.8 vs. 39.8 6 our method development (8). Addi- levels were 2.10/3.72%. The linearity 0.9 weeks, P , 0.001). GDM neonates tional participants were included in (r2)ofthea-chain glycation response, were normoglycemic (4.1 6 0.9 vs. this project aiming at investigating the measured from 1.97 to 3.11%, was 4.3 6 0.8 mmol/L, P = NS) but displayed pathophysiology relevance of measur- 0.9990. In the current samples, the min- hyperinsulinemia (89 6 104 vs. 41 6 32 ing fetal A1C, but always with the imum value that we measured was 1.9% pmol/L, P = 0.02) and impaired insulin same inclusion and exclusion criteria as and the maximum value was 2.8%; the sensitivity as estimated by the glucose- when recruiting for method develop- interquartile range was 2.12–2.40%. to-insulin ratio (0.09 6 0.06 vs. 0.14 6 ment. According to Canadian recom- 0.06, P =0.01). mendations (9), GDM was defined in Biomarker Measurements The mean level of Gla was higher in Maternal A1C (Bio-Rad VARIANT, Hercu- this study as two out of three abnormal GDM neonates (2.32 vs. 2.20%, P =0.01) les, CA) analysis was performed by plasma glucose values during the 75-g and was correlated with maternal A1C the Centre Hospitalier Universitaire de oral glucose tolerance test (fasting $5.3, levels in the overall group (r =0.67,P , Sherbrooke (CHUS) Clinical Biochemis- 1h$10.6, and 2 h $8.9 mmol/L). 0.0001) and in each group (GDM: r = try laboratory. Cord blood samples for Exclusion criteria were as follows: 0.66, P , 0.0001; NGT: r =0.50,P = future measurements were centrifuged multiple pregnancies, mother’s first tri- 0.01). Gla was not correlated with birth at 2,500g during 10 min at 48C. Aliquots mester BMI $30 kg/m2, gestational age weight (GDM: r =0.16,P = 0.34; NGT: r = of plasma were stored at 2808C until ,37 weeks, and neonate birth weight 0.03, P = 0.88) or with gestational age measurement. Glucose was measured ,2,500 g. Gestational age was deter- (GDM: r =0.01,P =0.96;NGT:r =0.09, using a colorimetric method (Wako mined based on ultrasound dating, a P =0.64). Diagnostics, Mountain View, CA); the standardized clinical procedure at our intra-assay and interassay CVs were institution. CONCLUSIONS both ,5%; minimum detectable con- Maternal weight and height at first Gla was higher in neonates exposed to centration was 0.28 mmol/L. For insulin, trimester and data related to delivery GDM. Our sophisticated MS method al- C-peptide, and proinsulin concentration and neonatal outcomes were retrieved lows for precise Gla measurement, in- measurements (radioimmunoassay; from medical records. Blood for mater- dependently of the HbF-to-HbA ratio Millipore Corp., Billerica, MA), the nal A1C measurements was sampled (8). Standard methods for measuring fe- intra-assay and interassay CVs were upon arrival at the Obstetrics Depart- tal Hb glycation have consistently failed, all ,10% for the three analytes; mini- ment, before delivery. Cord blood sam- due to inherent inaccuracy (7,10) and mum detectable concentrations of insu- ples for glycation measurements were the wide range of HbF (65–90% in term lin, C-peptide, and proinsulin were, collectedwithin15minafterdelivery, neonates) and HbA variations attribut- respectively, 16.29 pmol/L, 0.033 ng/mL, in BD Vacutainer tubes with sodium able to individual characteristics and fe- and 3.055 pmol/L. fluoride/Na2-EDTA as an additive (8). tal gestational age (11). An alternative Insulin sensitivity was assessed by the Statistical Analyses method using electrospray ionization glucose-to-insulin ratio in cord blood (6). Continuous variables were assessed for MS has been proposed before but re- distribution and log transformed to mained with limitations (12); as demon- Principles achieve normal distribution as neces- strated in our methodology paper (8), Hb in cord blood is a composite of HbF sary. Variables were compared between we were able to clearly establish the (a2g2 chains) and HbA (a2b2). Accord- groups (GDM vs. NGT) using Student linearity of the response, selectivity, ing to our previous analyses (8), only t test or Mann-Whitney U test depending and data deconvolution processing in a glycation of the a-chain (Gla)wasspe- on the normality of the distribution. Fre- range that is appropriate for measure- cifically assessed in this study because of quencies, presented as percentages, ment of glycation of a-andg-chains. the following: 1)thea-chain is common were compared using x2 tests. Correla- Here, we reported associations with to HbF and HbA and represents the sum tions between maternal/neonatal char- clinically relevant maternal and fetal of both forms, which are produced by acteristics and the level of glycation of glycemic traits. the fetus; 2)thea-chain is not a function the a-chain were assessed by Pearson Because fetal Hb life span is 60–80 of the HbF-to-HbA ratio, an important coefficients. P values ,0.05 were con- days (11), we assumed that Gla reflects consideration since this ratio is not 1:1 sidered significant. Statistical analyses mean fetal glucose exposure over the and varies substantially at birth; and 3) were performed using SAS version 9 last 4–6 weeks of pregnancy. Most im- the number of g-chain variants and (SAS Institute, Cary, NC). portantly, the strong correlations with b-chain methodological interferences maternal A1C levels suggest that Gla ac- make the measurement of these chains RESULTS curately reflects the degree of hypergly- unsuitably long. GDM women (Table 1) were heavier at cemia exposure. One could argue that Glycation of Hb chains was measured first trimester (BMI = 27.6 6 5.7 vs. maternal A1C may also serve as marker using a quadrupole orthogonal acceler- 24.6 6 5.9 kg/m2, P = 0.04) and had a of glycemic exposure; yet given the im- ation time-of-flight mass spectrometer higher A1C at delivery time (5.8 6 0.4 perfect, albeit strong, correlation, we (SYNAPT MS; Waters Corporation, Mil- vs. 5.4 6 0.3%, P =0.001).Neonatesex believe that direct measurement of ford, MA) with an electrospray ioniza- distribution and birth weight (3,358 6 Gla captures additional information tion source working in positive mode 394 vs. 3,457 6 468 g, P =NS)were and offers a more accurate in situ glycemic and W-Optics (8). The intra/interday similar in both groups. GDM women fetal exposure. Whether Gla is a stronger 2832 Glycation of a-Chain Diabetes Care Volume 37, October 2014

Table 1—Characteristics of mothers and neonates of fetal glucose metabolism. Because NGT (n = 30) GDM (n = 37) P value even slight maternal hyperglycemia is associated with higher Gla levels in Mother clinical characteristics cord blood, we propose Gla as an accu- Age (years) 28.9 6 5.2 29.6 6 5.1 NS First trimester BMI (kg/m2)* 24.6 6 5.9 27.6 6 5.7 0.04 rate biomarker of maternal glycemic Gestational age at delivery (weeks) 39.8 6 0.9 38.5 6 0.8 ,0.001 exposure. Maternal A1C level at delivery (%) 5.4 6 0.3 5.8 6 0.4 0.001 Delivery (% of caesarian section) 1 (3.3%) 9 (24.3%) 0.02 Neonate clinical characteristics Acknowledgments. The authors acknowledge Sex (% female) 11 (36.7%) 15 (40.5%) NS the Blood Sampling in Pregnancy clinic at CHUS Weight (g) 3,457 6 468 3,358 6 394 NS and Sun Life Financial, which supports research Cord blood activities integrated with the clinic, the assis- tance of clinical research nurses (M. Gerard´ and HbF (%) 77.0 6 4.7 79.9 6 4.3 0.01 M.-J. Gosselin), research assistants (C. Rousseau HbA (%) 16.1 6 5.0 13.0 6 4.6 0.01 and P. Brassard) for recruiting mothers and ob- Glucose (mmol/L) 4.3 6 0.8 4.1 6 0.9 NS taining consent for the study, and the labora- 6 6 Insulin (pmol/L) 41 32 89 104 0.02 tory technician (Lucie Bouffard) for cord blood 6 6 C-peptide (ng/mL) 0.26 0.22 0.40 0.47 NS endocrine measurements. 6 6 Proinsulin (pmol/L) 31.7 20.3 35.2 25.2 NS Funding and Duality of Interest. The study Glucose-to-insulin ratio 0.14 6 0.06 0.09 6 0.06 0.01 was supported by Fondation des Etoiles´ (oper- Glycated a-chain (%) 2.20 6 0.15 2.32 6 0.21 0.01 ating grant to J.-L.A. and C.A.-B.), Sun Life Fi- Variables are given as n (%) or mean 6 SD. P value ,0.05 was considered statistically significant. nancial (grant to J.-L.A.), and Faculty of *First trimester medical record: n =29inNGTandn = 37 in GDM. Medicine and Health Sciences Start-up Fund (to J.-L.A.). M.-F.H., J.-C.P., C.A.-B., and J.-L.A. are members of the Fonds de la Recherche du predictor of short- and long-term compli- all were intensively monitored and effi- Quebec-Sant´ e(FRQS)´ –funded Centre de Re- cations compared with maternal A1C at ciently treated throughout pregnancy, cherche of the CHUS. F.O.D. received a Fonda- tion des Etoiles´ scholarship. M.-F.H. is an FRQS delivery will necessitate larger studies as shown by their predelivery A1C, which, research scholar and was awarded a Clinical and longitudinal follow-up. We cannot ex- albeit slightly higher than in control sub- Scientist Award by the Canadian Diabetes Asso- clude that contamination of maternal jects, was well within the normal range ciation and the Maud Menten Award from blood in the cord blood samples might for pregnant women at delivery (15). the Institute of Genetics–Canadian Institutes ’ influence our results, but this is usually Based on these results, we are confident of Health Research (CIHR). C.A. is a master s student in statistics. She was awarded the limited; blood from a heel prick could be that our method is capable of detecting CIHR–Frederick Banting and Charles Best Canada considered as a source of neonatal blood subtle differences in glycemic exposure, Graduate Scholarships. L.B., J.R., J.-C.P., and J.-L.A. for future studies. even within the “controlled” range. On are FRQS-Junior 2 research scholars. No other High birth weight is considered the other hand, we admit that the strict potential conflicts of interest relevant to this a classical marker of fetal glucose selection of our participants limits the article were reported. Author Contributions. F.O.D. conceived the exposure (13,14), although it is a func- generalizability of our results; future MS study design, performed clinical data col- tion of multiple factors and is an admit- studies should include women of a lection and MS data analysis and interpretation, tedly imperfect reflection of maternal larger range of BMI, different levels of and wrote the manuscript. M.-F.H. provided hyperglycemia (3). Interestingly, we glycemic control, and other ethnicities. assistance with the statistical analysis, actively found no correlation between Gl Further investigation into Gl may participated in data interpretation, and wrote a a the manuscript. C.A. performed acquisition of and birth weight, arguing again that open up two entirely new fields of re- data and data analysis and interpretation and birth weight is a relatively crude search. First, Gla may help refine the revised and edited the manuscript. J.M. partic- measurement of exposure to maternal glycemic targets of treatment of diabe- ipated in data collection and data analysis and hyperglycemia. tesinpregnancy.Thisisanimportant interpretation and revised and edited the fi manuscript. P.P., L.B., J.R., and J.-C.P. partici- In contrast to the long-term signi - issue for care that should translate into pated in data analysis and interpretation and cance of Gla, other common surrogates normal Gla levels at birth. However, cur- revised and edited the manuscript. C.A.-B. of in utero hyperglycemic exposure, such rent glycemic treatment targets for conceived the MS study design, performed as cord blood glucose, insulin, or C-peptide GDM are admittedly (16) far from glu- data analysis and interpretation, and revised – and edited the manuscript. Measuring glycation (4 6), have drawbacks limiting reliabil- cose levels observed in normal preg- ’ fi of Hb chains is J.-L.A. s brain child. J.-L.A. also ity; they are measured only once, i.e., nancy and should be rede ned. Our conceived the study design, received funding, at birth, despite labor-associated fetal titer of higher Gla values in neonates participated in data analysis and interpretation, stress, exercise, and hypoxia, which affect born from ideally treated GDM pregnan- and wrote the manuscript. M.-F.H., C.A.-B., and fetal glycemia in opposite ways. ciesclearlyshowsthisisnowwithin J.-L.A. are the guarantors of this work and, as The method proposed here is both grasp. Second, Gla may help screen such, had full access to all the data in the study and take responsibility for the integrity of the accurate and enables detection of fetal newborns and identify those at risk for data and the accuracy of the data analysis. in situ glycemic exposure. Importantly, future metabolic diseases, thus provid- our study was conducted under strin- ing reliable quantification of the impact References gent conditions, i.e., in neonates devel- of fetal programming. Such investiga- 1. Wright CS, Rifas-Shiman SL, Rich-Edwards JW, Taveras EM, Gillman MW, Oken E. Intrauterine oping in the healthiest of maternal tions will require longitudinal studies. exposure to gestational diabetes, child adiposity, environments. Indeed, none of the Current technology now enables ac- and blood pressure. Am J Hypertens 2009;22: mothers were obese at conception and curate measurement of Gla and thus 215–220 care.diabetesjournals.org Dupont and Associates 2833

2. Nolan CJ, Damm P, Prentki M. Type 2 diabe- umbilical cord: evaluation of the glycemic con- 11. Lurie S, Mamet Y. Red blood cell survival tes across generations: from pathophysiology to trol indicators in neonates. J Perinatol 2011;31: and kinetics during pregnancy. Eur J Obstet prevention and management. Lancet 2011;378: 430–433 Gynecol Reprod Biol 2000;93:185–192 169–181 8. Dupont FO, Gagnon R, Ardilouze JL, Auray- 12.DavisonAS,GreenBN,RobertsNB.Fetal 3. Catalano PM, Hauguel-De Mouzon S. Is it Blais C. Determination of glycated and acety- hemoglobin: assessment of glycation and acet- time to revisit the Pedersen hypothesis in the lated hemoglobins in cord blood by time-of-flight ylation status by electrospray ionization mass face of the obesity epidemic? Am J Obstet Gy- mass spectrometry. Anal Chem 2011;83:5245– spectrometry. Clin Chem Lab Med 2008;46: necol 2011;204:479–487 5252 1230–1238 4. Schwartz R, Gruppuso PA, Petzold K, Brambilla 9. Canadian Diabetes Association Clinical 13. Pedersen J, Osler M. Hyperglycemia as the D, Hiilesmaa V, Teramo KA. Hyperinsulinemia and Practice Guidelines Expert Committee. Cana- cause of characteristic features of the foetus macrosomia in the fetus of the diabetic mother. dian Diabetes Association 2008 clinical prac- and newborn of diabetic mothers. Dan Med Diabetes Care 1994;17:640–648 tice guidelines for the prevention and Bull 1961;8:78–83 5. Catalano PM, Presley L, Minium J, Hauguel-de management of diabetes in Canada. Canadian 14. Freinkel N. Banting Lecture 1980. Of preg- Mouzon S. Fetuses of obese mothers develop Jounal of Diabetes. 2008;32(Suppl. 1):S1– nancy and progeny. Diabetes 1980;29:1023–1035 insulin resistance in utero. Diabetes Care 2009; S201 15. Mosca A, Paleari R, Dalfra` MG, et al. Refer- 32:1076–1080 10. Dupont FO, Gagnon R, Menard´ J, Auray- ence intervals for hemoglobin A1c in pregnant 6. Luo ZC, Delvin E, Fraser WD, et al. Maternal Blais C, Ardilouze JL. Evaluation of the glyce- women: data from an Italian multicenter study. glucose tolerance in pregnancy affects fetal insu- mic control in neonates: a novel technical Clin Chem 2006;52:1138–1143 lin sensitivity. Diabetes Care 2010;33:2055–2061 approach for measuring fetal-glycated hemo- 16. Combs CA, Moses RG. Aiming at new tar- 7. Koga M, Murai J, Saito H, et al. Measurement globin. J Perinatol 2011;31:807; author reply gets to achieve normoglycemia during preg- of glycated hemoglobin and glycated albumin in 808 nancy. Diabetes Care 2011;34:2331–2332