DIABETES TECHNOLOGY & THERAPEUTICS Volume 17, Number 9, 2015 ª Mary Ann Liebert, Inc. DOI: 10.1089/dia.2014.0255

ORIGINAL ARTICLE

Continuous Glucose Monitoring in Type 1 Diabetes Shows that Fetal Heart Rate Correlates with Maternal Glycemia

Katarzyna Cypryk, MD, PhD,1 Lukasz Bartyzel, MD,1 Monika Zurawska-Klis, MD, PhD,1 Wojciech Mlynarski, MD, PhD,2 Agnieszka Szadkowska, MD, PhD,2 Jan Wilczynski, MD, PhD,3 Dorota Nowakowska, MD, PhD,3 Lucyna A. Wozniak, PhD,4 and Wojciech Fendler, MD, PhD2

Abstract Background: Much evidence has shown that in women with preexisting diabetes are affected by an increased risk of maternal and fetal adverse outcomes, probably linked to poor glycemic control. Despite great progress in medical care, the rate of stillbirths remains much higher in diabetes patients than in the general population. Recent technological advances in the field of glucose monitoring and noninvasive fetal heart rate monitoring made it possible to observe the fetal–maternal dependencies in a continuous manner. Subjects and Methods: Fourteen type 1 diabetes patients were involved into the study and fitted with a blinded continuous glucose monitoring (CGM) recorder. Fetal electrocardiogram data were recorded using the Monica AN24 device (Monica Healthcare Ltd., Nottingham, United Kingdom), the recordings of which were mat- ched with CGM data. Statistical analysis was performed using a generalized mixed-effect logistic regression to account for individual factors. Results: The mean number of paired data points per patient was 254 – 106, representing an observation period of 21.2 – 8.8 h. Mean glycemia equaled 5.64 – 0.68 mmol/L, and mean fetal heart rate was 135 – 6 beats/min. Higher glycemia correlated with fetal heart rate (R = 0.32; P < 0.0001) and was associated with higher odds of the developing small accelerations (odds ratio = 1.05; 95% confidence interval, 1.00–1.10; P = 0.04). Conclusions: Elevated maternal glycemia of mothers with diabetes is associated with accelerations of fetal heart rate.

Introduction decisions, contributing to better perinatal outcomes.7 Recent technological advances in the field of glucose monitoring and dequate metabolic control of diabetes in pregnant noninvasive fetal heart rate (FHR) monitoring made it pos- women is a crucial method of preventing neonatal and A sible to observe the fetal–maternal dependencies in a con- obstetric complications. The rate of all perinatal complica- tinuous manner. Using two devices—a continuous glucose tions correlates with the mother’s glycated hemoglobin monitoring (CGM) system and a FHR monitor—we inves- (HbA1c) concentration.1–6 Unfortunately, HbA1c is a retro- tigated association between maternal glycemic fluctuations spective parameter, and thus it is not particularly useful and FHR variability. during pregnancy when self-glucose monitoring is essential. However, achieving and maintaining normoglycemia during Subjects and Methods pregnancy in women with diabetes are major challenges. Adding the continuous monitoring to standard treatment The study group consisted of 14 white pregnant women could improve metabolic control and facilitate therapeutic with diabetes, treated with insulin. All available patients

Departments of 1Diabetology and Metabolic Diseases and 2Pediatrics, Oncology, Hematology and Diabetology, Medical University of Lodz, Poland. 3Feto-Maternal and Gynecology Department, Research Institute, Polish Mother’s Memorial Hospital, Lodz, Poland. 4Department of Structural Biology, Faculty of Biomedical Sciences and Postgraduate Education, Medical University of Lodz, Lodz, Poland.

619 620 CYPRYK ET AL. treated during the 2012–2013 period were assessed for with the CGM recording up to the nearest minute. Small FHR eligibility and asked to participate in the study. The in- accelerations and decelerations were defined, respectively, as clusion criteria were set as following: singleton pregnancy increases in the FHR from the baseline greater than 10 beats/ > 30 weeks of gestation and established type 1 diabetes. We min (bpm) and lasting for at least 15 s and as decreases of excluded patients with any comorbidities necessitating ‡ 10 bpm for ‡ 10 s. A large acceleration event was defined as pharmacological treatment, patients after in vitro fertil- an increase greater than 15 bpm and lasting for more than ization, pregnancies at risk of premature termination, ab- 15 s. A large deceleration event was defined as a fall in FHR normal findings in prenatal ultrasonography, or triple test. from the baseline, where the area below the baseline was The study was approved by the Bioethics Committee of the greater than 20 beats. Imputation of missing data was not Medical University of Lodz, Lodz, Poland. Eligible patients attempted. invited to the study, after their informed consent was ob- The paired CGM/ECG data points were used in the analysis. tained, were fitted with a CGM recorder (iPro2; Medtronic, Frequency of hypoglycemia ( < 70 mg/dL [3.9 mmol/L] and Minneapolis, MN) for a standard, at least 48-h-long, re- < 54 mg/dL [ < 3 mmol/L]) and hyperglycemia ( > 126 mg/dL cording. Throughout the study the patients had the option of [7 mmol/L]) were presented as percentage of the total CGM staying in the hospital or remaining in their own home. time. Mean glucose values and three parameters of glycemic The sensor was placed in the subcutaneous tissue using a variability (SD, M100, and J index) were calculated using an dedicated insertion device. Calibration of the CGM device online CGM variability calculator designed by the authors was done using the ACCU-CHEK glucometer (Roche, (GlyCulator) and verified using Microsoft (Redmond, WA) Basel, Switzerland), and measurements were performed at Excel. These indices of glycemic variability were selected as least four times per day. The team was blinded to their methodology allows for noncontinuous data, which was CGM recordings during the study, and its results did not the case in our study.8–11 influence their clinical management during the study. Statistical analysis was performed using a generalized The endocrinological team was blinded to fetal electro- mixed-effect logistic regression model to estimate the cardiography data but could adjust insulin treatment de- glucose-dependent risk of FHR disturbances while account- pending on current glucose levels when necessary (blood ing for individual effects. The lme4 package for R was used glucose concentration less than 70 mg/dL [3.9 mmol/L] and for this analysis. more than 126 mg/dL [7 mmol/L]). Fetal electrocardiogram (ECG) data were recorded using the Monica AN24 device Results (Monica Healthcare Ltd., Nottingham, United Kingdom). All patients were fitted with this device between 1 p.m. and 5 Mean age of patients was 30.4 – 4.2 years (range, 25–37

p.m. after a run-in period of at least 3 h of CGM. After a years). Mean duration of diabetes was 14.6 – 7.6 years. Two calibration period of the ECG, the recording was started and patients had a body mass index of > 30 kg/m2, with the mean continued for at least 20 h. value in the whole group being 28.1 – 3.8 kg/m2.Gestational After removal of both devices, the ECG recording was week at evaluation was 33.5 – 1.0 (range, 32–36) (Table 1). No analyzed for a 5-min time frame using the manufacturer’s adverse events during the CGM/ECG recordings were noted in software (Monica DK; Monica Healthcare Ltd.), matched any of the studied patients. The mean number of paired data

Table 1. Study Group Characteristics Duration BMI Third- Gestational Neonatal Maternal of (kg/m2) trimester age Place Gestational birth Patient age Type of diabetes Insulin before HbA1c (weeks) at of age (weeks) weight ID (years) diabetes (years) therapy pregnancy (%) evaluation observation at delivery (g) 1 36 T1DM 26 MDI 28.6 6.20 33 Hospital 38 3,660 2 29 T1DM 8 CSII 26.5 5.86 33 Hospital 36 3,200 3 26 T1DM 17 CSII 27.9 6.45 33 Hospital 38 3,460 4 31 T1DM 23 CSII 34.1 5.30 33 Home 37 3,800 5 30 T1DM 18 CSII 29.4 6.03 33 Home 38 4,160 6 36 T1DM 8 CSII 24.3 5.55 33 Home 33 3,140 7 27 T1DM 2 MDI 25.1 5.60 33 Hospital 39 3,390 8 29 T1DM 25 CSII 28.8 5.68 33 Home 37 4,080 9 28 T1DM 22 CSII 37.6 6.12 34 Hospital 35 2,950 10 31 T1DM 7 CSII 24.7 6.10 35 Hospital 39 3,900 11 35 T1DM 15 MDI 25.5 6.30 34 Hospital 40 3,770 12 25 T1DM 12 CSII 28.6 6.00 36 Hospital 38 3,680 13 25 T1DM 15 CSII 23.9 6.97 32 Home 33 2,690 14 37 T1DM 7 CSII 27.8 5.16 34 Hospital 39 4,160 Mean – SD 30.4 – 4.2 — 14.6 – 7.6 — 28.1 – 3.8 5.95 – 0.48 33.5 – 1.0 — 37.1 – 2.17 3,574 – 456

Individual data of all 14 participants are presented in the respective rows. Means with SDs are presented in the last row. BMI, body mass index; CSII, continuous subcutaneous insulin infusion; HbA1c, glycated hemoglobin; MDI, multiple daily injections; T1DM, type 1 diabetes mellitus.

Table 2. Individual Data Recording Characteristics Patient ID Mean – SD or 1 2 3 4 5 6 7 8 9 10 11 12 13 14 median (25–75%) Glucose concentration (mmol/L) Mean 5.34 4.71 6.52 5.52 6.90 5.06 5.00 5.84 5.85 5.06 5.59 5.03 6.67 5.92 5.64 – 0.68 SD 2.00 1.50 2.07 1.76 2.14 1.10 1.29 0.90 2.06 1.37 2.15 1.49 2.11 1.68 1.69 – 0.42 M100 9.55 6.47 4.66 4.12 5.22 2.03 2.90 0.96 5.01 2.54 5.66 3.37 5.27 3.06 4.34 – 2.15 J index 17.45 12.46 23.87 17.2 26.5 12.29 12.86 17.87 20.28 13.39 19.41 13.77 24.96 18.7 17.93 – 4.74 Glucose concentration (mmol/L) Minimum 2.83 2.44 3 2.33 3.22 2.22 3.83 4.56 2.44 2.78 2.83 2.67 4.67 3.17 3.07 – 0.78 Maximum 9.89 10 10.78 8.11 9.78 8.22 7.61 8.11 9.89 8.11 10.61 10.22 10.78 9.44 9.40 – 1.28 621 % > 7 mmol/L 23.4 10.0 15.9 3.3 53.6 4.4 9.9 28.4 23.4 4.4 21.3 14.7 54.1 17.4 16.65 (9.93–23.40) % < 3.9 mmol/L 8.1 46.0 11.7 15.7 5.3 13.7 1.7 0.0 7.7 12.4 28.7 14.9 0.0 5.5 9.90 (5.35–14.60) % < 3.0 mmol/L 1.4 11.2 2.2 6.2 0.0 3.4 0.0 0.0 2.4 1.7 2.4 2.2 0.0 0.0 1.95 (0.00–2.40) FHR (bpm) Mean 128 138 131 123 135 133 138 137 144 136 132 133 145 134 135 – 6 SD 6.03 7.19 7.49 9.06 6.05 8.24 4.57 8.49 8.26 7.84 6.23 9.52 5.35 8.11 7.32 – 1.46 FHR (bpm) Minimum 117 116 101 105 115 112 122 113 125 115 117 112 133 114 116 – 8 Maximum 142 165 156 147 153 164 150 163 179 166 157 159 167 157 159 – 9 Percentage small Acceleration 0.43 0.41 0.33 0.37 0.38 0.39 0.49 0.5 0.4 0.43 0.23 0.46 0.44 0.47 0.41 – 0.07 Deceleration 0.18 0.24 0.31 0.31 0.18 0.09 0.18 0.13 0.18 0.18 0.25 0.29 0.19 0.32 0.22 – 0.07 R for FHR/glycemia correlation 0.29a 0.25a 0.21b 0.09 0.36a - 0.02 0.11 0.21a 0.38a 0.11b 0.35a 0.40a 0.10 0.34a NA

aP < 0.0001, bP < 0.05. bpm, beats/min; FHR, fetal heart rate; NA, not applicable; R, Pearson’s correlation coefficient. 622 CYPRYK ET AL.

FIG. 1. Correlation between the time-point-matched maternal glycemia and fetal heart rate for pooled data points of all available patients/recordings. points per patient was 254 – 106, representing an observation In our study we did not observe any large, long-lasting period of 21.2 – 8.8 h. The mean glucose level equaled deceleration, the most dangerous heart disturbance, probably 5.64 – 0.68 mmol/L, and the mean FHR was 135 – 6bpm. because diabetes control achieved in the studied patients was Glycemic variability data and FHR characteristics are excellent. The mean HbA1c was 5.95 – 0.48%, and glycemia summarized in Table 2. In nine patients FHR showed sig- during the observation ranged from 2.22 to 10.78 mmol/L. nificant, positive correlation with the time-point-matched maternal glycemia, which consequently held true for an Discussion aggregate correlation analysis (R = 0.32; P < 0.0001) (Fig. 1). In five patients no such correlations were noted, and in one Adequate metabolic control of diabetes in pregnant case a significant negative correlation was noted. In mixed- women is a crucial method of preventing neonatal and ob- effect logistic regression analysis, higher glucose levels stetric complications. Hyperglycemia in pregnant women can were associated with higher odds of the fetus developing cause fetal malformations, growth disturbances, development small accelerations (odds ratio = 1.05; 95% confidence in- delay of the central nervous system, chronic hypoxemia, and, terval, 1.00–1.10; P = 0.04). None of the analyzed individual- finally, spontaneous abortion, stillbirth, prematurity, and level factors, like patient’s age (P = 0.38), body mass index many other fetal complications, including hypoglycemia, (P = 0.66), duration of diabetes (P = 0.32), gestational week macrosomy, etc. Those are well documented in the in vitro (P = 0.93), or nighttime measurements (P = 0.07), signifi- and in human studies.1–5 cantly affected the association between maternal glucose A Swedish study proved that stillbirth rate in women with and FHR accelerations. The risk of small decelerations did type 1 diabetes was more than three times higher than in the not, however, depend significantly on current maternal glu- background population.6 (fetal ECG) is an cose levels (odds ratio = 0.97; 95% confidence interval, established method to monitor fetal well-being and is es- 0.92–1.03; P = 0.32), with nighttime measurements signifi- sential to avoid intrauterine death by early detection of fetal cantly decreasing the odds of deceleration events (odds ra- compromise. Bjo¨rklund et al.12 revealed that maternal hy- tio = 0.49; 95% confidence interval, 0.41–0.58; P < 0.0001). poglycemia induced during hyperinsulinemic/hypoglycemic Similarly, with acceleration events, none of the evaluated clamp with induction and maintenance of an arterial blood individual factors—patient’s age (P = 0.51), body mass in- glucose level of about 2.2 mmol/L was associated with an dex (P = 0.56), duration of diabetes (P = 0.83), or gestational increase in frequency and amplitude of FHR accelerations week (P = 0.35)—showed any associations with the odds of and a slight decrease in the pulsatility index of the umbilical deceleration events. artery and with a rise in the maternal catecholamine levels. FETAL HEART RATE AND MATERNAL GLYCEMIA 623

However, Reece et al.13 did not observe an influence of of the mother and fetal well-being in various levels of gly- maternal hypoglycemia in pregnant women with diabetes on cemic control. the mean number of fetal limb, body, and breathing move- ments or a heart rate, although maternal epinephrine and Acknowledgments growth hormone levels were significantly increased. This study was supported financially by funds of the In a study by Serra-Serra et al.14 the impact of maternal Medical University of Lodz (project number 502-03/0-160- blood glucose concentration on cardiotocographic results in 01/502-04-014). pre- and postmeal stage was measured. The authors con- cluded that FHR parameters are unaffected by prandial gly- cemic changes over a wide range (4.2–14.8 mmol/L) of Author Disclosure Statement maternal glucose levels in any of the groups of women with No competing financial interests exist. gestational diabetes, pregestational diabetes, and healthy pregnant volunteers without diabetes.14 On the other hand, Buscicchio et al.15 showed that gesta- References tional diabetes did impact FHR. The alteration was slight but 1. Vargas R, Repke JT, Ural SH: Type 1 diabetes mellitus and 15 evident, and it correlated with neonatal reactivity. Un- pregnancy. Rev Obstet Gynecol 2010;3:92–100. fortunately, in this study the blood glucose level was mea- 2. Inkster ME, Fahey TP, Donnan PT, Leese GP, Mires GJ, sured only once, at delivery. Murphy DJ: Poor glycated haemoglobin control and ad- Wiener et al.16 revealed significantly reduced FHR varia- verse pregnancy outcomes in type 1 and type 2 diabetes tion as well as frequency of accelerations in of women mellitus: systematic review of observational studies. BMC with well-controlled diabetes compared with fetuses of Pregnancy 2006;30:6–30. mothers without diabetes. In this study, published in 1996, 3. Jensen DM, Korsholm L, Ovesen P, Beck-Nielsen H, HbA1c was only estimated every 3 months during pregnancy, Moelsted-Pedersen L, Westergaard JG, Moeller M, Damm self-monitoring of glucose was performed, and there were no P: Peri-conceptional A1C and risk of serious adverse data about blood glucose concentration during the test. pregnancy outcome in 933 women with type 1 diabetes. Costa et al.17 also found a positive correlation between Diabetes Care 2009;32:1046–1048. basal FHR and mean glycemia. A significant negative cor- 4. Gizzo S, Patrelli TS, Rossanese M, Noventa M, Berretta R, relation was observed in this study between short-term var- Di Gangi S, Bertin M, Gangemi M, Nardelli GB: An update iation and mean glycemia. on diabetic women obstetrical outcomes linked to precon- All these studies did not combine directly maternal blood ception and pregnancy glycemic profile: a systematic lit- erature review. Sci World J 2013;6;254901. glucose concentration with fetal ECG data. Our study seems 5. Jensen DM, Damm P, Moelsted-Pedersen L, Ovesen P, to confirm direct relationship between maternal glycemia and Westergaard JG, Moeller M, Beck-Nielsen H: Outcomes in fetal well-being as it demonstrates that elevated maternal type 1 diabetic pregnancies: a nationwide, population-based glycemia of mothers with diabetes is associated with accel- study. Diabetes Care 2004;27:2819–2823. erations of FHR, even in very well-controlled diabetes. We 6.PerssonM,NormanM,HansonU:Obstetricandperi- are aware that the study does have its limitations due to a natal outcomes in type 1 diabetic pregnancies: a large, single time point of CGM/ECG examination, a lack of a population-based study. Diabetes Care 2009;32:2005– healthy control group, and limited sample size. It would be, 2009. however, extremely difficult to justify such an experiment on 7. Murphy HR, Rayman G, Lewis K, Kelly S, Johal B, Duf- healthy pregnant women without any a priori reason to field K, Fowler D, Campbell PJ, Temple RC: Effectiveness conduct a cumbersome CGM examination. Moreover, the of continuous glucose monitoring in pregnant women with dynamic range of glycemic changes in healthy women is diabetes: randomised clinical trial. BMJ 2008;25;337: incomparably narrower than that observed in women with a1680. diabetes. This ultimately convinced us that if any meaningful 8. Dobbe JGG, Lunshof S, Boer K, Wolf H, Grimbergen CA: associations between GCM and ECG variability are to be The technique and algorithms for computerized analysis of discovered, it should be done first in a group in which fluc- long-term fetal heart rate recordings. Prenatal Neonatal tuations of both heart rate and glycemia may have direct Med 2001;6:280–289. consequences. Finally, having established that FHR does 9. Schlichtkrull J, Munck O, Jersild M: The M-value, an index correlate with maternal glycemia, we are able to focus on the of blood-sugar control in diabetics. Acta Med Scand 1965; impact of further inquiries on fetal outcomes of either pa- 177:95–102. rameter fluctuations. 10. Wojcicki JM: ‘‘J’’-index. A new proposition of the as- sessment of current glucose control in diabetic patients. This is the first study conducted in real life that shows this Horm Metab Res 1995;27:41–42. correlation in a continuous manner. We believe that inves- 11. Czerwoniuk D, Fendler W, Walenciak L, Mlynarski W: tigation using a combined fetal ECG/maternal CGM is fea- GlyCulator: a glycemic variability calculation tool for sible and provides insight into impact of maternal diabetes on continuous glucose monitoring data. J Diabetes Sci Technol fetal well-being. 2011;5:447–451. 12. Bjo¨rklund AO, Adamson UK, Almstro¨m NH, Enocksson Conclusions EA, Gennser GM, Lins PE, Westgren LM: Effects of hy- poglycaemia on fetal heart activity and umbilical artery Elevated maternal glycemia of mothers with diabetes is Doppler velocity waveforms in pregnant women with insulin- associated with accelerations of FHR. Further studies are dependent diabetes mellitus. Br J Obstet Gynaecol 1996; needed to explore the association between glucose variability 103:413–420. 624 CYPRYK ET AL.

13. Reece EA, Hagay Z, Roberts AB, DeGennaro N, Homko plicated by maternal diabetes. Eur J Obstet Gynecol Reprod CJ, Connolly-Diamond M, Sherwin R, Tamborlane WV, Biol 1996;27;70:111–115. Diamond MP: Fetal Doppler and behavioral responses 17. Costa VN, Nomura RM, Reynolds KS, Miyadahira S, Zu- during hypoglycemia induced with the insulin clamp gaib M: Effects of maternal glycemia on fetal heart rate in technique in pregnant diabetic women. Am J Obstet Gy- pregnancies complicated by pregestational diabetes melli- necol 1995;172:151–155. tus. Eur J Obstet Gynecol Reprod Biol 2009;143:14–17. 14. Serra-Serra V, Camara R, Sarrio´n P, Jaren˜o M, Cervera J, Bellver J, Perales A: Effects of prandial glycemic changes on objective fetal heart rate parameters. Acta Obstet Gy- Address correspondence to: necol Scand 2000;79:953–957. Katarzyna Cypryk, MD, PhD 15. Buscicchio G, Gentilucci L, Giannubilo SR, Tranquilli AL: Department of Diabetology and Metabolic Diseases Computerized analysis of fetal heart rate in pregnancies Medical University of Lodz complicated by gestational diabetes mellitus. Gynecol En- ul. Pomorska 251 docrinol 2010;26:270–274. 92-216 Lodz, Poland 16. Weiner Z, Thaler I, Farmakides G, Barnhard Y, Maulik D, Divon MY: Fetal heart rate patterns in pregnancies com- E-mail: [email protected]