Association Between Time in Range and HbA1c in Japanese Patients With Type 2 Diabetes Mellitus

There are no large-scale studies on the association between time in range (TIR) and hemoglobin A1c (HbA1c) in Japanese patients with type 2 diabetes mellitus (T2DM) only. The aim of this study was to dene the relationship between TIR and HbA1c levels in Japanese patients with T2DM. The glycemic prole of 999 patients was analyzed with FreeStyle Libre Pro Continuous Glucose Monitoring (FLP-CGM) while they continued their prescribed glucose-lowering medications. FLP-CGM data recorded over 8 consecutive days were analyzed. The regression model for HbA1c on TIR was HbA1c = 9.4966 − 0.0309×TIR. The predicted HbA1c level for TIR of 70% was 7.33%, and is higher than recent reports subjecting mostly T1DM. The TIR corresponding to HbA1c 7.0% was 80.64%. HbA1c level correlated signicantly with many FLP-CGM-derived metrics. The patients with low TIR tended to have long history of diabetes, on higher daily insulin dose and had high body mass index, HbA1c, liver dysfunction and triglyceride. Furthermore, relatively higher percentages of patients of this group used sulfonylureas, glinides, glucagon like peptide-1 receptor agonists and insulin. Our data showed that the predicted HbA1c corresponding to TIR is largely depends on the study population, thus is not uniform. Our results provide new insights on the management of T2DM.


Introduction
The UK Prospective Diabetes Study (UKPDS) 35 on patients with type 2 diabetes mellitus (T2DM) reported a linear relationship, without threshold values, between hemoglobin A1c (HbA1c) level and various T2DM-related complications, such as cardiovascular events 1 . Furthermore, the Kumamoto study reported that patients with HbA1c < 6.9% were less likely to develop microangiopathy 2 . These and other studies suggest that HbA1c can be used as an indicator of glycemic control in patients with T2DM. On the other hand, many large-scale clinical studies involving T2DM patients reported that reductions in HbA1c levels were more pronounced following intensive therapy compared with conventional therapy. While intensive therapy is reported to reduce the incidence of microangiopathy, it does not affect that of cardiovascular disease (CVD) events 3,4 . In fact, one study reported a signi cant increase in the overall death rate with intensive therapy 5 . In addition, Currie et al. 6 analyzed patients who were switched from monotherapy to combination oral glucose-lowering therapy or in whom insulin therapy was started. They reported that there was a U-shaped rather than a linear relationship between cardiovascular events and HbA1c levels, suggesting that there is a problem in relying on HbA1c only as an indicator of glycemic control in the treatment of diabetes.
In 2019, the international panel of the Advanced Technologies and Treatments for Diabetes (ATTD) Congress provided recommendations on glycemic control guidelines by Continuous Glucose Monitoring (CGM) 7 . According to the international guideline, the target range of blood glucose level is 70-180 mg/dL (except in pregnant women). Time in range (TIR) is de ned as the time spent within the target range, and the time above range (TAR) and time below range (TBR) are de ned as the time spent above and below the target range, respectively. Lu et al. 8 investigated the association between TIR and intima-media thickness ≥ 1.0 mm using the CGM data of 2215 patients. They reported signi cantly higher frequency of intima-media thickness ≥ 1.0 mm among patients with lower TIR. Furthermore, Beck et al. 9 reanalyzed the 7-point self-monitored blood glucose data from the Diabetes Control and Complications Trial (DCCT) in order to determine the association between TIR and microangiopathy. They found higher frequencies of retinopathy and microalbuminuria among patients with low TIR.
While the association between TIR and diabetes-related complications has been assessed, as described above, to our knowledge, there are no large-scale studies on the association between TIR and HbA1c levels in patients with T2DM only. The aim of the present large multi-institutional study was to determine the relationship between TIR and HbA1c levels in T2DM.

Methods
Patinets. The study population was Japanese patients with T2DM who regularly attended the Outpatient Diabetes Clinics of 34 institutions across Japan (listed in Supplementary Table 1). The study design, inclusion criteria and exclusion criteria had been published previously 10 . Patients who met the eligibility criteria were asked to participate in the present study. A total of 1,000 outpatients with T2DM, stable control and no history of apparent CVD were recruited between May 2018 and March 2019. One patient withdrew consent after enrollment. The study protocol was approved by the institutional review board of each participating institution in compliance with the Declaration of Helsinki and current legal regulations in Japan. Written informed consent was obtained from all participants after a full explanation of the study.
Study design. This study is an exploratory sub-analysis of an ongoing, observational, prospective cohort study designed to investigate the relationships between glucose variability evaluated with CGM and the incidence of composite cardiovascular events over a 5-year follow-up period, as described in detail previously 10 . The present study used baseline data from the cohort study and is registered in the University Hospital Medical Information Network Clinical Trials Registry (UMIN-CTR), which is a non-pro t organization in Japan that meets the requirements of the International Committee of Medical Journal Editors (UMIN000032325). All participants wore the FreeStyle Libre Pro (FLP) CGM (FLP-CGM) (FreeStyle Librepro®, Abbott Japan, Tokyo, Japan) for 14 days and the data used for analysis were obtained from days 4 to 11 of the FLP-CGM. No changes were made to diet or exercise therapy or glucose-lowering agents/insulin dose. The primary outcome was HbA1c predicted by TIR. The secondary outcomes were the correlation between HbA1c and FLP-CGM-derived metrics other than TIR, and the comparison of patient background for each TIR group.
Clinical laboratory tests. Fasting blood samples were obtained at outpatient visits and analyzed for renal function status, lipid levels, and HbA1c (National Glycohemoglobin Standardization Program) using standard techniques. Urinary albumin excretion (UAE) was measured using a latex agglutination assay on a spot urine sample. The estimated glomerular ltration rate (eGFR) was calculated using a standard formula 11 . CGM using FreeStyle Libre Pro Device. The FLP-CGM device was used in this study as reported previously 10 . Apart from wearing the FLP-CGM, no restrictions were imposed on the participants with respect to daily activities/routines. After the end of the 14-day recording, the CGM stored data were downloaded and analyzed. Glucose variability was assessed using the mean amplitude of glucose excursion (MAGE), SD, and coe cient of variation (CV). MAGE was calculated as the arithmetic mean of the differences between consecutive peaks and nadirs, provided that the differences were greater than one SD of the mean glucose value. CV was calculated by dividing SD by the mean of the corresponding glucose readings. The original statistical analysis plan (SAP) for this study was reported in the initial study protocol 9 . After the publication of the SAP, the ATTD Congress proposed some CGM-derived metrics as useful clinical targets that complement HbA1c 7 . Thus, we updated the SAP by adding certain CGMderived metrics in this study prior to database lock. Mean glucose was measured from the data collected during FLP-CGM. TIR was de ned as percentage of time spent within the target range of 70-180 mg/dl (time in range: TIR 70-180mg/dl), time above the target glucose range (TAR > 180 mg/dl, TAR > 250 mg/dl), and time below the target glucose range (TBR < 70 mg/dl, TBR < 54 mg/dl). Low blood glucose index (LBGI) and high blood glucose index (HBGI) formulae were implemented by converting glucose values into risk scores 12 . In addition, the mean of daily differences (MODD) 13 in glucose levels and interquartile range (IQR) were calculated to assess inter-day glucose variability. MODD was calculated as the mean of the absolute difference between glucose levels measured at the same time on two consecutive days. IQR was calculated using values from the same time of day during the monitoring period. Since a previous study demonstrated that FLP-CGM was less accurate during the rst 24 hours (from the rst day to the second day) after insertion and during the last four days of its 14-day lifetime 14 , we analyzed FLP-CGM data over the middle 8-day period, as stated above.
Statistical analysis. All variables are summarized as mean, standard deviation (SD), minimum and maximum for continuous variables and number (proportion) of patients for categorical variables. The Pearson correlation coe cient and linear regression model were used to assess the relationship between two variables. Continuous data were compared using analysis of variance and Kruskal-Wallis test, and categorical data were compared using χ2 test and Fisher exact test. All statistical tests were two-sided with 5% signi cance level. All analyses were performed using SAS software version 9.4 (SAS Institute, Cary, NC).

Results
Characteristics of study population. As reported in our previous study 15  Correlation between HbA1c and FLP-CGM-derived metrics and comparison of patient background for each TIR group. HbA1c level correlated signi cantly with many FLP-CGM-derived metrics, with the exception of CV and TBR<54 ( Table 2, Fig. 1). Patients with low TIR tended to have a longer history of diabetes, be on higher daily insulin dose, and have higher BMI, HbA1c, alanine aminotransferase, gammaglutamyl transpeptidase and triglyceride, as shown in Table 3. In addition, the complication rates of all microangiopathies were relatively higher in these patients, with more severe diabetic retinopathy and nephropathy. Furthermore, the incidence of use of sulfonylureas, glinides, glucagon like peptide-1 (GLP-1) receptor agonists, and insulin was likely to be higher in these patients (Table 4).   22 . Patients with low TIR had a longer history of diabetes, higher HbA1c and TG levels, and a larger percentage used insulin, than the other TIR groups. These results con rmed the early ndings of Lu et al. 23 However, the use of glucoselowering medications was not analyzed in the previous studies. In the present study, higher percentages of patients with low TIR used sulfonylureas, GLP-1 receptor agonists, and insulin, compared to the other TIR groups. According to the basic data collected in 2019 by the Japan Diabetes Clinical Data Management (JDDM) Study group, many T2DM patients treated with insulin or GLP-1 receptor agonists exhibited poor control of HbA1c (≥8%), indicating that the present ndings are consistent with the realworld data. In the treatment of diabetes in Japan, sulfonylureas, insulin, and GLP-1 receptor agonists are considered to re ect the fact that they are often administered to T2DM patients with poor glycemic control. Including these drugs, we believe that active therapeutic intervention from an early stage is important to achieve the target glycemic control.
The strength of this study is its relatively large-sample, multicenter, study design. However, our study has certain limitations. First, FLP-CGM derived metrics were evaluated by FLP-CGM measurements during a limited time. Thus, FLP-CGM derived metrics may not represent overall glycemic control of subjects. In order to attain the best of measurements of glucose uctuation by FLP-CGM at baseline, we employed a blind CGM system that enable subjects not to alter their lifestyle behaviors based on the results of glucose readings. Second, this study used FLP-CGM, which does not require calibration, and the mean glucose level may have been lower due to the difference in the sensor used. Third, this study was conducted on Japanese patients with T2DM, and racial differences should be considered. In the present study, the predicted HbA1c corresponding to TIR of 70% and the mean TIR were higher and mean values of TAR and TBR were lower than those reported in previous studies. We only recruited Japanese patients with T2DM, and the low proportion of patients using insulin. These constraints may limit the generalizability of our results.

Conclusion
There are no large-scale studies that have evaluated the association between TIR and HbA1c levels in patients with T2DM only. To our knowledge, this multi-institutional study of 999 patients with T2DM only is the largest and the results provide new insights on the management of T2DM.