Association of glycation gap with hypoglycemia CGM indices, diabetic complications, and quality of life in Japanese type 2 diabetes patients Authors: Satomi Sonoda, Yosuke Okada, Tomoya Mita, Keiichi Torimoto, Kenichi Tanaka, Fumiya Sato, Naoto Katakami, Hidenori Yoshii, Keiko Nishida, Shingo Nakayamada, Ryota Ishii, Masahiko Gosho, Iichiro Shimomura, Hirotaka Watada Published: 2026 ABSTRACT ABSTRACTBackgroundThe glycation gap (GGap), defined as the discrepancy between glycated hemoglobin (HbA1c) and the value estimated from actual blood glucose level, is associated with diabetic complications, but its association with hypoglycemia remains unclear.MethodsWe evaluated the association between GGap and continuous glucose monitoring (CGM)‐based hypoglycemic indices in patients with type 2 diabetes mellitus (T2DM). Baseline data from a multicenter cohort of 999 T2DM patients without cardiovascular disease were analyzed. The difference between HbA1c and estimated A1c (eA1c) was defined as the GGap, and various CGM indices were compared among low (≤0.16), medium (<0.16 to ≤0.60), and high (>0.60) GGap tertile groups.ResultsIn the high GGap group, the average blood glucose was lower, while the Time Below Range <3.9 mmol/L (TBR<3.9) and <3.0 mmol/L (TBR<3.0), and low blood glucose index (LBGI) were higher than in the low and middle GGap groups. Patients with minimum blood glucose levels of <3.9, <3.0 mmol/L, and TBR<3.9 ≥ 4%, and TBR<3.0 ≥ 1% had significantly higher GGap values.ConclusionThis is the first study to show the strong association of high GGap with CGM‐based hypoglycemic indices in T2DM. To achieve diabetes treatment that effectively prevents the progression of diabetic complications, it is essential to assess the GGap of the individual patient before intensifying diabetes management.Keywords: continuous glucose monitoring, glycation gap, hypoglycemia INTRODUCTION The level of glycated hemoglobin (HbA1c) reflects the glycation rate of hemoglobin in red blood cells over the past 2 to 4 months and is used as a standard indicator of glycemic control. It is associated with the onset and progression of diabetic microvascular complications and mortality risk 1 , 2 , 3 , 4 , 5 , 6 . However, even with the same HbA1c level, differences in the risk of developing diabetic complications are sometimes observed. This is because HbA1c levels are not a direct measure of blood glucose levels; hence, they may be higher or lower than actual blood glucose levels. The discrepancy between actual blood glucose level and HbA1c is termed the glycation gap (GGap), which has been found to be associated with retinopathy 7 , nephropathy 8 , as well as cardiovascular events and overall mortality 9 . This makes it an important index for the long‐term management and personalized care of patients with diabetes. Hypoglycemia has been suggested as a reason for the association between cardiovascular events, mortality, and GGap 10 . A sub‐analysis of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial reported a higher frequency of severe hypoglycemia in patients with large differences between HbA1c levels estimated from fasting blood glucose levels and actual HbA1c levels 11 . A large GGap indicates that the HbA1c level is higher than actual blood glucose levels, which could increase the risk of hypoglycemia if treatment is intensified based on the HbA1c level; although this remains unclear. Until recently, the GGap was quantified using indices calculated from HbA1c levels and fructosamine 12 , 13 , 14 . Nowadays, with continuous glucose monitoring (CGM) becoming widely used for direct evaluation of blood glucose levels, it is easy to calculate the estimated A1c (eA1c) from the CGM‐derived average blood glucose levels. This has enabled a more accurate assessment of the GGap. In addition, CGM allows detailed evaluation of hypoglycemia. To the best of our knowledge, there have been no reports that evaluated the relationship between GGap and hypoglycemia. Furthermore, it is not clear whether patients with a large GGap experience increased hypoglycemia. The present study was designed to determine the relationship between GGap and CGM hypoglycemic indices in patients with type 2 diabetes mellitus (T2DM). MATERIALS AND METHODS Study design This study is an exploratory sub‐analysis of an ongoing prospective observational study 15 designed to determine the relationship between glucose variability, as assessed by CGM, and the incidence of composite cardiovascular events over a 5‐year follow‐up period in Japanese patients with T2DM. Using baseline data from this prospective observational study, we evaluated the relationship between GGap (assessed by CGM and HbA1c levels) and hypoglycemic indices. This study is registered with the University Hospital Medical Information Network Clinical Trials Registry (UMIN‐CTR), a non‐profit organization in Japan, and meets the requirements of the International Committee of Medical Journal Editors (UMIN000032325). Study population The study population consisted of Japanese patients with T2DM who were seen regularly at the Outpatient clinic of the Departments of Diabetes of 34 medical facilities in Japan. The study design, inclusion criteria, and exclusion criteria have been published previously 15 . The study targeted outpatients aged between 30 and 80 years with stable glycemic control, excluding those with a history of cardiovascular events. Among the screened participants, those who met the eligibility criteria were invited to participate in the study. A total of 1,000 eligible participants were recruited between May 2018 and March 2019; however, one individual withdrew consent, leaving 999 participants in the analysis. All experiments were approved by the ethics committees of the representative institution, Juntendo University Hospital, and all other participating medical facilities in accordance with the Declaration of Helsinki and current Japanese regulations. Written informed consent was obtained from each participant after receiving a full explanation of the study. Definitions of various types of diabetic complications Diabetic neuropathy was defined as meeting two of the following three criteria: the presence of subjective symptoms believed to be due to diabetic polyneuropathy, decreased or absent Achilles tendon reflexes bilaterally, and reduced vibration sensation at both medial malleoli. Diabetic retinopathy (DR) was diagnosed by trained ophthalmologists and classified into four stages: no diabetic retinopathy (NDR), simple diabetic retinopathy (SDR), pre‐proliferative diabetic retinopathy (PPDR), and proliferative diabetic retinopathy (PDR). Diabetic nephropathy was classified according to the level of urinary albumin excretion (UAE) into normal albuminuria (<30 mg/gCr), microalbuminuria (≥30 to <300 mg/gCr), and macroalbuminuria (≥300 mg/gCr). Biochemical tests Fasting blood samples were collected, and renal function tests, lipids, and HbA1c (National Glycohemoglobin Standardization Program) levels were determined using standard methods. UAE was measured using the latex agglutination method with spot urine samples. The estimated glomerular filtration rate (eGFR) was calculated using a previously defined formula 16 . Glucose metrics by CGM and calculation of GGap The FreeStyle Libre Pro (FLP) (Abbott Japan, Tokyo, Japan) CGM device (FLP‐CGM) was used at baseline to measure blood glucose levels every 15 min over a period of 14 days 15 , 17 . Apart from wearing the FLP‐CGM, no restrictions were imposed on the daily life activities of the participants. Previous studies showed that the accuracy of the FLP‐CGM decreases during the first 24 h after attachment (Day 1 to Day 2) and during the last 4 days of the 14‐day monitoring period 18 . Accordingly, data from the intermediate 8‐day period were used for analysis. In this study, hypoglycemia <3.0 mmol/L was used as an indicator of severe hypoglycemia. Episodes were assessed during the 8‐day period used for analysis within the 14‐day CGM monitoring period. The downloaded dataset was analyzed, and the average blood glucose level was estimated using the FLP‐CGM data. Glucose variability was assessed using standard deviation (SD), coefficient of variation (CV) 19 , and mean amplitude of glycemic excursions (MAGE). The CV (%) was calculated by dividing the standard deviation (SD) by the corresponding mean blood glucose level. MAGE was calculated using the arithmetic mean of the differences between consecutive peaks and nadirs when the difference exceeded 1 SD of the mean blood glucose level 20 . Time in Range (TIR) was defined as the percentage of time spent with blood glucose level between 3.9 to 10.0 mmol/L, Time Above Range (TAR) above 10 mmol/L and TAR above 13.9 mmol/L were defined as the percentage of time spent above these corresponding glucose levels, and Time Below Range (TBR) below 3.9 mmol/L (TBR<3.9) and TBR below 3.0 mmol/L (TBR<3.0) were defined as the percentages of time below the corresponding glucose levels, as described previously 19 . The average glucose levels, CV, TIR, TAR, and TBR are major outcomes measurable by CGM 21 . The low blood glucose index (LBGI) and high blood glucose index (HBGI) were calculated by converting blood glucose levels into risk scores 22 . In addition, the mean of daily differences (MODD) 23 and interquartile ranges (IQRs) were calculated to assess day‐to‐day glucose variability. MODD was calculated as the mean of the absolute differences in blood glucose levels measured at the same time on consecutive days, and the IQR was calculated using the values at the same time points during the observation period. The eA1c has been developed to derive an estimated HbA1c level based on the average blood glucose levels 24 . The eA1c was calculated using the following formula: eA1c (%) = 3.38 + 0.02345 × (average glucose level [mg/dL]). GMI was calculated using the formula: GMI (%) = 3.31 + 0.02392 × (average glucose level [mg/dL]) 25 . In this study, GGap was defined as the difference between HbA1c and eA1c levels (GGap = HbA1c – eA1c). Statistical analysis Continuous variables were presented as mean ± SD values or median (interquartile range), and categorical variables as numbers (percentages) of patients. Participants were divided into three groups based on the tertiles of the GGap values: low group (GGap ≤0.16%, GGap≤0.16), middle group (GGap of >0.16 to ≤0.60%, GGap0.16–0.60); and high group (GGap >0.60%, GGap>0.60). Continuous data were compared using analysis of variance, Kruskal‐Wallis test, or Student's t‐test as appropriate, and categorical data were compared using the chi‐squared test. The relationship between various parameters measured by the FLP‐CGM and GGap, as well as patient background factors, was compared among the three GGap groups. In addition, participants were classified according to the stage of diabetic nephropathy and DR, and also divided into groups that met or did not meet the following criteria: minimum blood glucose level <3.9 mmol/L, minimum blood glucose level <3.0 mmol/L, TBR <4%, and TBR <1%, and the GGap was compared among these groups. Finally, receiver operating characteristic (ROC) curve analysis was performed to determine the cutoff value of the GGap for predicting the achievement of TBR of <4% and TBR of <1%. The cutoff value was determined based on the Youden index (sensitivity + specificity − 1). All statistical tests were two‐sided with a significance level of 5%. To evaluate the association between GGap and CGM‐defined hypoglycemia outcomes, multivariable logistic regression analyses were performed. The outcomes included the presence of minimum glucose level <3.9 mmol/L, minimum glucose level <3.0 mmol/L, TBR <3.9 mmol/L ≥4%, and TBR <3.0 mmol/L ≥1%. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated. The models were adjusted for age, sex, duration of diabetes, HbA1c, insulin use and dose, estimated glomerular filtration rate, hemoglobin level, and the use of sulfonylureas and other glucose‐lowering agents. Similar analyses were performed using GMI‐derived GGap as a sensitivity analysis. All analyses were performed using SAS software version 9.4 (SAS Institute, Cary, NC). RESULTS Baseline clinical characteristics The study included 999 patients (608 men and 391 women), and the clinical background of all patients has been reported previously 15 . The mean age was 64.6 ± 9.6 years, body mass index (BMI) 24.6 ± 3.9 kg/m2, HbA1c 7.1 ± 0.8%, and eGFR 73.4 ± 20.6 mL/min/1.73m2, with a median duration of T2DM of 11.0 (6.0–18.0) years. The overall eA1c was 6.7 ± 0.8%, and the GGap was 0.38 ± 0.57%. Diabetes medications were used by 89.5% of the patients, with dipeptidyl peptidase 4 inhibitors in 57.8%, metformin in 54.4%, sodium‐glucose transport protein 2 (SGLT2) inhibitors in 23.1%, sulfonylureas (SUs) in 12.7%, and insulin in 15.8%. DR was detected in 22.2% of the patients, with the following stage distribution in the right/left eyes: SDR 12.6/12.3%, PPDR 4.9/4.8%, and PDR 3.7/3.6%. Furthermore, 27.0% of the patients had diabetic nephropathy, with the following stage distribution: normoalbuminuria 68.0%, microalbuminuria 24.7%, and macroalbuminuria 7.3%. Comparison of clinical characteristics by GGap tertiles The participants were divided into three groups based on their GGap values (Table 1). The GGap>0.60 group included more women (43.9%, compared with 33.4% for GGap≤0.16 and 39.9% for GGap0.16–0.60, P = 0.02), patients with longer duration of diabetes (13.9 (8.0–20.0) years, compared with 10.2 (5.0–16.0) and 11.0 (6.0–17.0), respectively, P < 0.001), higher BMI (25.0 ± 4.0 kg/m2, compared with 24.2 ± 3.8 and 24.6 ± 3.8, respectively, P = 0.023), and higher HbA1c levels (7.5 ± 0.9%, compared with 6.8 ± 0.7 and 6.9 ± 0.6, respectively, P < 0.001), than the other two GGap groups. In addition, the GGap>0.60 group included a larger proportion of patients with DR and more advanced stages of retinopathy. Furthermore, patients of the GGap>0.60 group showed higher use of SGLT2 inhibitor (P = 0.040), metformin (P = 0.013), glucagon‐like peptide‐1 (GLP‐1) receptor agonist (P < 0.001), and insulin (P < 0.001) than the other groups. Insulin therapy was used in 25.2% of the GGap>0.60 group, compared with only 10.6% of the GGap≤0.16 and 11.4% of the GGap0.16–0.60 groups. Furthermore, the daily insulin dose was also higher in the GGap>0.60 group (18.0 (11.0–26.0) units/day, compared with 15.0 (9.0–27.0) and 9.5 (6.0–20.0), respectively, P = 0.016). Comparable findings were observed when the comparison was performed using GMI‐derived GGap (Table S1). Table 1Patient demographic and background characteristics by glycation gap tertileLow (≤0.16)Moderate (>0.16 to ≤0.60)High (>0.60) P value n = 329 n = 333 n = 337Age (years)65.1 ± 9.864.8 ± 9.163.9 ± 10.00.253Female gender (%)110 (33.4)133 (39.9)148 (43.9)0.020Estimated duration of diabetes (years)11.7 (1.0–44.0)12.3 (1.0–46.0)14.5 (1.0–58.0)<0.001Body mass index (kg/m2)24.2 ± 3.824.6 ± 3.825.0 ± 4.00.023Systolic blood pressure (mmHg)130 ± 16132 ± 15132 ± 140.074Diastolic blood pressure (mmHg)75 ± 1176 ± 1176 ± 110.379Diabetic complications (%)Diabetic neuropathy96 (29.2)86 (25.8)104 (30.9)0.341Diabetic retinopathy64 (19.5)67 (20.1)91 (27.0)0.034Rt. NDR/SDR/PPDR/PDR267 (81)/42 (13)/11 (3)/9 (3)272 (82)/41 (12)/14 (4)/6 (2)248 (74)/43 (13)/24 (7)/22 (6)0.005Lt. NDR/SDR/PPDR/PDR269 (82)/41 (12)/11 (3)/8 (2)271 (81)/42 (13)/14 (4)/6 (2)252 (75)/40 (12)/23 (7)/22 (6)0.006Diabetic nephropathy93 (28)73 (21.9)104 (30.9)0.028Normo‐/micro‐/macro‐albuminuria (%)217 (67)/77 (24)/30 (9)233 (71)/81 (24)/16 (5)220 (66)/86 (26)/26 (8)0.250HbA1c (%)6.8 ± 0.76.9 ± 0.67.5 ± 0.9<0.001HbA1c (mmol/mol)50.6 ± 8.151.8 ± 6.358.6 ± 9.6<0.001Estimated glomerular filtration rate (mL/min/1.73 m2)74.9 ± 21.473.0 ± 17.872.4 ± 22.40.262Urinary albumin excretion (mg/g Creatine)105.3 (1.3–3110.0)54.5 (0.2–1620.0)131.6 (0.9–7398.0)0.077Use of oral glucose‐lowering agents (%)Sodium‐glucose cotransporter 2 inhibitors (%)64 (19.5)74 (22.2)93 (27.6)0.040Dipeptidyl peptidase‐4 inhibitors (%)183 (55.6)193 (58.0)201 (59.6)0.574Metformin (%)158 (48.0)186 (55.9)199 (59.1)0.013Sulfonylureas (%)39 (11.9)46 (13.8)42 (12.5)0.740Glinides (%)20 (6.1)19 (5.7)29 (8.6)0.269α‐glucosidase inhibitors (%)47 (14.3)59 (17.7)66 (19.6)0.186Thiazolidinediones (%)41 (12.5)46 (13.8)56 (16.6)0.294Glucagon‐like peptide‐1 receptor agonists (%)16 (4.9)17 (5.1)41 (12.2)<0.001Use of insulin (%)35 (10.6)38 (11.4)85 (25.2)<0.001Dose of insulin (unit/day)17.9 (5.0–61.0)14.8 (3.0–48.0)21.1 (4.0–101)0.016Hemoglobin (g/dL)14.2 ± 1.514.2 ± 1.414.1 ± 1.50.234Data are mean ± SD or number (%) of patients. Kruskal‐Wallis test was used for continuous variables. The chi‐squared test was used for categorical data.NDR, no diabetic retinopathy; PPDR, pre‐proliferative diabetic retinopathy; PDR, proliferative diabetic retinopathy; SDR, simple diabetic retinopathy. Patient demographic and background characteristics by glycation gap tertile Data are mean ± SD or number (%) of patients. Kruskal‐Wallis test was used for continuous variables. The chi‐squared test was used for categorical data. NDR, no diabetic retinopathy; PPDR, pre‐proliferative diabetic retinopathy; PDR, proliferative diabetic retinopathy; SDR, simple diabetic retinopathy. Comparison of FLP‐CGM–derived metrics by GGap tertiles Table 2 shows the differences in CGM metrics among the three GGap groups. In the GGap>0.60 group, blood glucose level was lower (7.47 ± 1.74 mmol/L, compared with 8.52 ± 2.03 and 7.42 ± 1.34, respectively, P < 0.001), and the CV, a marker of glucose variability, was higher (26.81 ± 0.40%, compared with 25.78 ± 5.70 and 26.02 ± 5.15, respectively, P = 0.054). Additionally, the hypoglycemic indicators, both the TBR<3.9 and TBR<3.0 were higher, and the LBGI was significantly greater in the GGap>0.60 group than in the other two groups. The same results were obtained when GMI‐derived GGap data were used in the analysis (Table S2). Table 2FLP‐CGM–derived metrics by glycation gap tertileLow (≤0.16)Moderate (>0.16 to ≤0.60)High (>0.60) P value n = 329 n = 333 n = 337Mean glucose (mmol/L)8.52 ± 2.037.42 ± 1.347.47 ± 1.74<0.001CV (%)25.78 ± 5.7026.02 ± 5.1526.81 ± 6.400.054MAGE (mmol/L)5.70 ± 2.105.25 ± 1.805.43 ± 2.080.013SD (mmol/L)2.19 ± 0.691.93 ± 0.531.99 ± 0.62<0.001TIR (%)72.96 ± 21.7183.59 ± 13.7679.98 ± 17.94<0.001TAR>13.9 mmol/L (%)6.44 ± 12.121.91 ± 5.263.23 ± 8.73<0.001TAR>10 mmol/L (%)25.99 ± 22.0414.59 ± 14.3316.44 ± 18.43<0.001TBR<3.9 mmol/L (%)1.06 ± 3.751.81 ± 3.623.58 ± 5.99<0.001TBR<3.0 mmol/L (%)0.23 ± 1.380.20 ± 0.780.55 ± 2.09<0.001HBGI7.05 ± 5.714.58 ± 3.055.14 ± 4.42<0.001LBGI1.04 ± 1.491.51 ± 1.302.11 ± 1.96<0.001MODD (mmol/L)1.84 ± 0.711.61 ± 0.471.76 ± 0.69<0.001IQR (mmol/L)2.27 ± 0.951.97 ± 0.582.19 ± 0.85<0.001eA1c (%)6.98 ± 0.866.51 ± 0.566.54 ± 0.73<0.001GMI (%)6.98 ± 0.876.51 ± 0.586.53 ± 0.75<0.001Data are mean ± SD or number (%) of patients. Kruskal‐Wallis test was used for continuous variables.FLP‐CGM, FreeStyle Libre Pro continuous glucose monitoring; CV, coefficient of variation; MAGE, mean amplitude of glycemic excursion; SD, standard deviation; TIR, time in range; TAR, time above range; TBR, time below range; HBGI, high blood glucose index; LBGI, low blood glucose index; MODD, mean of daily differences, IQR, interquartile range; eA1c, estimates A1c; GMI, glucose management indicator. FLP‐CGM–derived metrics by glycation gap tertile Data are mean ± SD or number (%) of patients. Kruskal‐Wallis test was used for continuous variables. FLP‐CGM, FreeStyle Libre Pro continuous glucose monitoring; CV, coefficient of variation; MAGE, mean amplitude of glycemic excursion; SD, standard deviation; TIR, time in range; TAR, time above range; TBR, time below range; HBGI, high blood glucose index; LBGI, low blood glucose index; MODD, mean of daily differences, IQR, interquartile range; eA1c, estimates A1c; GMI, glucose management indicator. Comparison of GGap with FLP‐CGM–derived hypoglycemic metrics For this analysis, we divided the participants into two groups in order to assess the effect of the severity of hypoglycemia. First, patients were divided into those with a minimum blood glucose level of <3.9 and ≥3.9 mmol/L. Second, patients were divided into those with a minimum blood glucose of <3.0 and ≥3.0 mmol/L. Third, patients were divided into those with TBR<3.9 of ≥4 and <4%, and TBR<3.0 of ≥1% and <1 (Figure 1A–D). Comparable findings were observed when the comparison was performed using GMI‐derived GGap (Figure S1). In all these groups, the GGap was significantly higher in the group with severe hypoglycemia. Our analysis also showed that the GGap cutoff value to achieve a TBR <3.9 mmol/L at <4% was 0.36%, and the cutoff value for achieving a TBR <3.0 mmol/L at <1% was 0.39% (Figure 2). The same results were obtained when GMI‐derived GGap data were used in the analysis (Figure S2). GGap by FLP‐CGM–derived metrics. (a) Comparison of GGap of patients with minimum blood glucose level of <3.9 and ≥3.9 mmol/L. (b) Comparison of GGap of patients with minimum blood glucose level of <3.0 and ≥3.9 mmol/L. (c) Comparison of GGap of patients with TBR<3.9 mmol/L of <4 and ≥4%. (d) Comparison of GGap of patients with TBR<3.0 mmol/L of <1 and ≥1%. Data are mean ± SD. *P < 0.001, by the Student's t‐test. GGap, glycation gap; FLP‐CGM, FreeStyle Libre Pro continuous glucose monitoring; TBR, time below range. TBR<3.9 mmol/L (%) and TBR<3.0 mmol/L (%) predicted GGap cutoff values based on ROC curve analysis. (a) The cutoff value of GGap for achieving TBR<3.9 mmol/L at <4% was 0.36% (area under the ROC curve: 0.70; 95% CI: 0.66–0.74). (b) The cutoff value of GGap for achieving TBR<3.0 mmol/L at <1% was 0.39% (area under the ROC curve: 0.65; 95% CI: 0.59–0.89). GGap, glycation gap; %CV, % coefficient of variation; AG, average glucose; CI, confidence interval; ROC, receiver operating characteristic; TBR, time below range. Higher GGap values were independently associated with CGM‐derived hypoglycemic outcomes after adjustment for potential confounders, including age, sex, duration of diabetes, HbA1c, insulin use and dose, and the use of sulfonylureas and other glucose‐lowering agents (Table 3). Importantly, these associations were replicated when GGap was calculated using GMI (Table S3), confirming the robustness of the findings. Table 3Odds ratio for CGM‐derived hypoglycemic outcomes according to glycation gap (GGap) tertilesVariablesGGap TertilesLow (≤0.16)Moderate (>0.16 to ≤0.60)High (>0.60) n = 329 n = 333 n = 337Minimal glucose level <3.9 mmol/LUnadjusted OR (95% CIs)1.002.47 (1.81, 3.38)4.67 (3.37, 6.49)Adjusted OR (95% CIs)1.002.56 (1.85, 3.55)4.95 (3.50, 6.99)Minimal glucose level <3.0 mmol/LUnadjusted OR (95% CIs)1.001.69 (1.10, 2.59)2.83 (1.89, 4.26)Adjusted OR (95% CIs)1.001.74 (1.11, 2.71)2.52 (1.64, 3.88)TBR<3.9 mmol/L ≧4%Unadjusted OR (95% CIs)1.002.47 (1.46, 4.18)5.40 (3.30, 8.85)Adjusted OR (95% CIs)1.002.73 (1.58, 4.71)5.61 (3.32, 9.49)TBR<3.0 mmol/L ≧1%Unadjusted OR (95% CIs)1.001.55 (0.76, 3.18)3.09 (1.61, 5.91)Adjusted OR (95% CIs)1.001.57 (0.75, 3.30)2.27 (1.13, 4.57)Adjusted for age, sex, BMI, duration of diabetes, eGFR, Hb, use of insulin therapy, use of sulfonylureas, and use of other glucose‐lowering agents.OR, Odds ratio; CI, confidence interval. Odds ratio for CGM‐derived hypoglycemic outcomes according to glycation gap (GGap) tertiles Adjusted for age, sex, BMI, duration of diabetes, eGFR, Hb, use of insulin therapy, use of sulfonylureas, and use of other glucose‐lowering agents. OR, Odds ratio; CI, confidence interval. In addition to tertile‐based analyses, we evaluated GGap as a continuous variable. Higher GGap was significantly associated with increased odds of hypoglycemia outcomes per 0.1% increase (Table 4). Similar results were observed for GMI‐derived GGap (Table S4), indicating a dose–response relationship between GGap and hypoglycemia risk. Table 4Association between GGap and CGM‐derived hypoglycemic outcomesGGap (per 0.1% increase)Odds ratio (95% CI) P valueMinimal glucose level <3.9 mmol/LUnadjusted OR (95% CIs)1.12 (1.10, 1.15)<0.001Adjusted OR (95% CIs)1.13 (1.10, 1.16)<0.001Minimal glucose level <3.0 mmol/LUnadjusted OR (95% CIs)1.08 (1.05, 1.11)<0.001Adjusted OR (95% CIs)1.06 (1.03, 1.10)<0.001TBR <3.9 mmol/L ≧4%Unadjusted OR (95% CIs)1.13 (1.09, 1.16)<0.001Adjusted OR (95% CIs)1.13 (1.09, 1.17)<0.001TBR <3.0 mmol/L ≧1%Unadjusted OR (95% CIs)1.10 (1.06, 1.15)<0.001Adjusted OR (95% CIs)1.06 (1.02, 1.11)0.008Adjusted for age, sex, BMI, duration of diabetes, eGFR, Hb, use of insulin therapy, use of sulfonylureas, and use of other glucose‐lowering agents.OR, Odds ratio; CI, confidence interval. Association between GGap and CGM‐derived hypoglycemic outcomes Adjusted for age, sex, BMI, duration of diabetes, eGFR, Hb, use of insulin therapy, use of sulfonylureas, and use of other glucose‐lowering agents. OR, Odds ratio; CI, confidence interval. In a descriptive analysis stratified by tertiles of mean glucose, GGap tended to be higher in participants with hypoglycemia outcomes across all strata (Table S5). In sensitivity analyses using ROC‐derived cutoffs, participants with higher GGap showed consistently greater hypoglycemia burden across multiple metrics (Tables S6 and S7). Analysis of GGap in patients with DR and diabetic nephropathy The relationship between DR and GGap was analyzed by comparing the GGap among patients free of retinopathy (n = 777), with SDR (n = 133), PPDR (n = 50), and PDR (n = 39). The GGap values for the above respective groups were 0.36 ± 0.51, 0.41 ± 0.67, 0.57 ± 0.75, and 0.61 ± 0.79, with a significantly larger GGap observed in patients with more advanced diabetic complications (P = 0.005) (Figure 3a). GGap by diabetic retinopathy, nephropathy, and neuropathy. (a) Relationship between severity of diabetic retinopathy and GGap. The GGap for each stage of DR was as follows: NDR (n = 777): 0.363 ± 0.515, SDR (n = 133): 0.407 ± 0.665, PPDR (n = 50): 0.571 ± 0.750, and PDR (n = 39): 0.608 ± 0.789, with significant differences among the four groups (P = 0.006). (b) Relationship between severity of diabetic nephropathy and GGap. The GGap for each stage of diabetic nephropathy was as follows: normoalbuminuria (n = 670): 0.388 ± 0.526, microalbuminuria (n = 244): 0.411 ± 0.603, macroalbuminuria (n = 72): 0.324 ± 0.770, with no significant differences among the three groups. (c) Relationship between severity of diabetic neuropathy and GGap. The GGap was as follows: without neuropathy (−) (n = 729): 0.382 ± 0.537, with neuropathy (+) (n = 270): 0.408 ± 0.638, with no significant differences between the two groups. Data are mean ± SD. *P < 0.001, by analysis of variance. GGap, glycation gap; NDR, no diabetic retinopathy; SDR, simple diabetic retinopathy; PPDR, preproliferative diabetic retinopathy; PDR, proliferative diabetic retinopathy”. In contrast, analysis of the relationship between DR and GGap based on the severity of renal complications [which included patients with normoalbuminuria (n = 670), microalbuminuria (n = 244), and macroalbuminuria (n = 72)], showed no significant differences in GGap values among the three groups (normoalbuminuria: 0.39 ± 0.53, microalbuminuria: 0.41 ± 0.60, macroalbuminuria: 0.32 ± 0.77) (Figure 3b). The same results were obtained when GMI‐derived GGap data were used in the analysis (Figure S3). Relationship of GGap with DTR‐QOL and lifestyle factors With regard to the relationship between GGap and lifestyle factors, 11.6% of subjects of the GGap>0.60 group reported experiencing hypoglycemic symptoms in the past month, which was significantly higher than that in the GGap≤0.16 (4.3%) and GGap0.16–0.60 groups (4.2%) (P < 0.001). Additionally, for the DTR‐QOL scores, both the average scores for all items and the scores for each individual item were significantly lower in the GGap>0.60 group than the other two groups (Table S8). The distribution of GGap The distribution of GGap is shown in Figure 4 and was approximately normal. The distributions of GGap stratified by hypoglycemia status are shown in Figure 5. GGap values tended to be higher in groups with higher TBR. Similar patterns were observed for GMI‐derived GGap (Figures S4 and S5). The distribution of GGap. Histogram showing the distribution of GGap (mean ± SD, 0.39 ± 0.57). Distribution of GGap according to hypoglycemia status. (a) Distribution of GGap according to TBR <3.9 mmol/L. Histogram and kernel density plots of GGap stratified by TBR <3.9 mmol/L (<4% vs. ≥4%). The distribution shifted toward higher values in the ≥4% group. (<4%: mean ± SD, 0.33 ± 0.56; ≥4%: mean ± SD, 0.70 ± 0.51). (b) Distribution of GGap according to TBR <3.0 mmol/L. Histogram and kernel density plots of GGap stratified by TBR <3.0 mmol/L (<1% vs. ≥1%). The distribution shifted toward higher values in the ≥1% group. (<1%: mean ± SD, 0.37 ± 0.56; ≥1%: mean ± SD, 0.69 ± 0.61). DISCUSSION This is the first study to describe the association of high GGap with CGM‐derived indicators of hypoglycemia in patients with T2DM. The study also determined the GGap cutoff value associated with a higher risk of hypoglycemia based on CGM‐derived indicators of glucose management and demonstrated the presence of a relationship between GGap increase and progression of retinopathy. Based on these findings, we believe it is essential to fully understand the GGap in each individual case before intensifying diabetes treatment in order to avoid hypoglycemia and achieve treatment that prevents the progression of diabetic complications. In this study, we demonstrated the association of high GGap with CGM‐derived hypoglycemic indicators. Previous studies reported that GGap is associated with cardiovascular events and mortality, with hypoglycemia being suggested as a contributing factor 10 . A sub‐analysis of the ACCORD trial reported a higher frequency of severe hypoglycemia in patients with a large discrepancy between fasting glucose‐estimated HbA1c and actual HbA1c levels 11 . Our study is the first to accurately assess the GGap using CGM and identify the association of GGap with CGM markers of hypoglycemia. The rate of hemoglobin glycation varies among individuals, leading to the occurrence of GGap 14 , 26 , 27 . This gap differs between individuals. Therefore, the reliance of clinicians on HbA1c only in the management of T2DM may increase the incidence of hypoglycemia. In fact, in this study, the high GGap group included a significantly higher number of users of SGLT2 inhibitors, metformin, GLP‐1 receptor agonists, and insulin, suggesting clinicians' attempts to intensify pharmacotherapy in this group. Particularly in the GGap>0.60 group, both the proportion of insulin users and the daily insulin dosage were higher, suggesting that intensified treatment might contribute to an increase in hypoglycemia. Previous studies using the hemoglobin glycation index (HGI), which represents the difference between HbA1c and fasting glucose‐predicted HbA1c levels, reported an increase in hypoglycemic events in insulin‐treated patients with T2DM who had high HGI values 28 . In addition, previous studies have reported that, among patients with type 2 diabetes with a TIR >70%, a high HGI was associated with increased hypoglycemia 29 , and that, among patients with type 2 diabetes and acute coronary syndrome, hypoglycemia occurred significantly more frequently in the high‐HGI group 30 . In patients with a large GGap, measured HbA1c is relatively higher than the actual mean glucose level assessed by CGM. Therefore, when treatment decisions are based solely on HbA1c, glycemic control may be judged to be insufficient even when the actual mean glucose level is not markedly elevated. As a result, glucose‐lowering therapy may be intensified, leading to an excessive reduction in actual glucose levels and thereby increasing the risk of hypoglycemia. In this study, we identified the GGap cutoff value required to achieve hypoglycemia management goals using CGM. While there is not yet sufficient evidence regarding the standard for the GGap, Gu et al. 10 examined the relationship between GGap calculated from GA and mortality, and reported that patients with GGap exceeding 0.38% were at high risk of all‐cause mortality and cardiovascular death. Patients who experience hypoglycemic episodes are reported to have a two‐ to four‐fold higher risk of cardiovascular events and mortality 31 , 32 , 33 , suggesting that this relationship may be due to hypoglycemia. However, Gu et al. 10 did not examine thoroughly the relationship of GGap with hypoglycemia thoroughly. Our study is the first to demonstrate that a GGap of 0.36–0.38% or higher is associated with increased risk of hypoglycemia. The GGap thresholds derived from ROC analysis in this study should be interpreted as tools for risk stratification within this study population, rather than as established diagnostic cutoffs for hypoglycemia. Because both GGap and hypoglycemia outcomes are derived from CGM data, part of the observed association may reflect a structural relationship driven by mean glucose levels. To address this issue, we performed a descriptive analysis stratified by tertiles of mean glucose, which showed a consistent tendency for higher GGap values in participants with hypoglycemia across strata. These findings suggest that differences in mean glucose alone may not fully explain the observed association; however, the results should be interpreted with caution. We also demonstrated the association of the progression of DR and GGap. Significantly higher proportions of DR and advanced‐stage retinopathy were noted in subjects of the GGap>0.60 group. The extent of the GGap remains consistent over long periods of time 34 , and it seems to be associated with the progression of DR 7 , 35 , nephropathy 7 , 13 , 36 , and the prevalence of macroalbuminuria 8 . When the GGap is high—meaning that the actual HbA1c level is higher—cells in various organs and tissues, not just red blood cells, which are susceptible to diabetic complications, are exposed to similar glycation. Therefore, individuals with high GGap values are likely to experience more cellular damage than those with a lower GGap, even at the same level of glucose exposure. Consequently, the risk of various complications is expected to increase in patients with high GGap. In this study, no significant association was observed between the GGap and diabetic nephropathy. Although both diabetic retinopathy and nephropathy are microvascular complications and share several common risk factors, certain factors such as hypertriglyceridemia 37 , 38 , genetic background 39 , and smoking 40 , 41 have been specifically implicated in the development of nephropathy, whereas their association with retinopathy remains less clear. These factors may have influenced the manifestation of nephropathy in our cohort. Moreover, the widespread use of nephroprotective agents, such as SGLT2 inhibitors and GLP‐1 receptor agonists 42 , 43 , 44 in our study population may have contributed to the attenuation of progression of nephropathy, potentially obscuring any association with the GGap. One of the strengths of this study is its large‐scale, multicenter collaborative design. To the best of our knowledge, this study of 999 patients with T2DM is the largest to date examining the utility of CGM‐derived GGap. However, this study also has a few limitations. First, it included only Japanese patients with T2DM. Since it has been reported that the average blood glucose and HbA1c levels differ among ethnic groups 45 , 46 , the GGap cutoff values may vary between populations. Second, although this study found the association of CGM‐derived GGap with HbA1c values and indices of hypoglycemia, it was a cross‐sectional study; therefore, a causal relationship remains unclear. To address this, we are currently conducting a long‐term follow‐up study in the same cohort 15 . Third, the CGM measurements in this study were obtained from a maximum of 8 days of CGM data. This period may be insufficient for full evaluation of the overall glycemic control. On the other hand, the formula for eA1c 24 was developed using Dexcom sensors with at least 10 days, preferably 14 or more days of CGM data. Accordingly, the application of this formula to short‐term data from the FreeStyle Libre Pro may have resulted in either under‐ or over‐estimation of eA1c. In addition, whereas HbA1c reflects the average blood glucose levels over the past 1–2 months, the eA1c in our study was calculated from only 8 days of CGM data. This discrepancy in assessment periods between the two indices may have introduced a temporal mismatch, potentially leading to inaccuracies in the GGap estimation. Furthermore, while we used a blinded CGM system to prevent participants from altering their behavior based on glucose measurements, it is impossible to completely eliminate the placebo effect. Fourth, it cannot be ruled out that there may be discrepancies between actual blood glucose levels and those measured by the FLP‐CGM. Studies on the mean absolute relative difference (MARD) in T2DM have reported that, within the blood glucose range of 70–250 mg/dL, the MARD ranges from 11.4% to 16.7%, with slightly lower accuracy in the hypoglycemic range 47 . Fifth, insulin secretory capacity is known to correlate with glycemic variability assessed by CGM 48 . Therefore, in patients with impaired insulin secretion, such as those receiving insulin therapy, glycemic variability that is less readily detected by HbA1c may be present, which could result in elevated GGap values. In our study, we also examined patients' self‐reported hypoglycemic symptoms, and the results showed that the GGap>0.60 group experienced a higher frequency of hypoglycemic symptoms. Sixth, as this was a cross‐sectional study, the temporal relationship between GGap and hypoglycemia cannot be established. While a higher GGap may be associated with increased hypoglycemia risk, reverse causation is also possible; frequent hypoglycemia may lower mean glucose levels and consequently widen the GGap. Seventh, although the proposed GGap cutoff may be useful for risk stratification, the AUC was moderate, indicating that its discriminative ability as a single marker is limited. CONCLUSION There is substantial evidence in support of the use of HbA1c as a treatment target for glycemic control and the ultimate goal of prevention of diabetic complications. However, this study suggests that in patients with large GGap values, relying solely on HbA1c values during treatment may increase the risk of hypoglycemia. The increase in hypoglycemia among patients with high GGap values may lead to further progression of glycation in tissues and, consequently, increased T2DM‐related complications. To achieve diabetes treatment that effectively prevents the progression of complications, it is important to understand the value of GGap of each individual before intensifying diabetes management. Future large‐scale prospective studies of other ethnic groups may be required to validate our findings. FUNDING This study was funded by the Japan Agency for Medical Research and Development (AMED) under Grant Number JP20ek0210105 (to H.W.) and by the Manpei Suzuki Diabetes Foundation (to H.W.). DISCLOSURE H.W. has received research funds from Abbott Japan and is a member of the Advisory Board of Abbott Japan. N. K. received lecture fees from Abbott Japan Co., Ltd. The other authors declare no conflicts of interest. Approval of the research protocol: N/A. Informed consent: N/A. Registry and the registration no. of the study/trial: N/A. Animal studies: N/A. AUTHOR CONTRIBUTIONS All authors contributed to the article's validation, writing, review, and editing. They read and approved the final article. S.S., Y.O., K.T., T.M, K.T., F.S., N.K., H.Y., K.N., and R.I. collected the data. M.G. analyzed the data. H.W. received funding for the study. H.W. investigated the data. Y.O., T.M., and H.W. were responsible for project administration. S.N., I.S., and H.W. supervised the article. S.S. wrote the original draft. Supporting information Figure S1. GMI‐derived GGap by FLP‐CGM–derived metrics. Figure S2. TBR <3.9 mmol/L (%) and TBR <3.0 mmol/L (%) predicted GMI‐derived GGap cutoff values based on ROC curve analysis. Figure S3. GMI‐derived GGap by diabetic retinopathy, nephropathy and neuropathy. Figure S4. The distribution of GMI‐derived GGap. Figure S5. Distribution of GMI‐derived GGap according to hypoglycemia status. Table S1. Patient demographic and background characteristics by GMI‐derived glycation gap tertile. Table S2. FLP‐CGM–derived metrics by GMI‐derived glycation gap tertile. Table S3. Odds ratio for CGM‐derived hypoglycemic outcomes according to glycation gap (GMI‐derived GGap) tertiles. Table S4. Association between GMI‐derived GGap and CGM‐derived hypoglycemic outcomes. Table S5. Distribution of GGap according to hypoglycemia outcomes across mean glucose tertiles. Table S6. Comparison of hypoglycemia metrics according to the cutoff value of GGap for achieving TBR<3.9 mmol/L at <4%. Table S7. Comparison of hypoglycemia metrics according to the cutoff value of GGap for achieving TBR<3.0 mmol/L at <1%. Table S8. DTR‐QOL and lifestyle factors by glycation gap tertile. Table S9. List of sites and investigators.