- Email: [email protected]
Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes
G Model
DIABET-902; No. of Pages 6 Diabetes & Metabolism xxx (2017) xxx–xxx
Available online at
ScienceDirect www.sciencedirect.com
Original article
Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes S. Tanaka a, R. Kawasaki b,*, S. Tanaka-Mizuno c, S. Iimuro d, S. Matsunaga e, T. Moriya f, S. Ishibashi g, S. Katayama h, Y. Ohashi i, Y. Akanuma j, H. Sone e, H. Yamashita k, for the Japan Diabetes Complications Study Group a
Department of pharmacoepidemiology, graduate longitudinal school of medicine and public health, Kyoto university, Yoshida-Konoe-cho, Sakyo-ku, Kyoto, Japan b Department of public health, Yamagata university graduate school of medical science, 2-2-2 Iida-Nishi, Yamagata, Japan c Division of medical statistics, Shiga university of medical science, Tsukinowa Seta-cho, Ohtsu, Shiga, Japan d Teikyo academic research centre, Teikyo university, Kaga, Itabashi-ku, Tokyo, Japan e Department of haematology, endocrinology and metabolism, faculty of medicine, Niigata university, 1-757 Asahi-machi, Chuo-ku, Niigata, Japan f Health care centre, Kitasato university, 1-15-1 Kitasato, Minami-ku, Sagamihara-shi, Kanagawa, Japan g Division of endocrinology and metabolism, school of medicine, Jichi medical university, 3311-1 Yakushiji, Shimotsuke, Tochigi, Japan h Kawagoe clinic, Saitama medical university, 21-7 Wakitahonchou, Kawagoe, Saitama, Japan i Department of integrated science and engineering for sustainable society, Chuo university, 1-13-27, Kasuga, Bunkyo-ku, Tokyo, Japan j The institute for adult diseases Asahi life foundation, 1-6-1 Marunouchi Chiyoda-ku, Tokyo, Japan k Department of ophthalmology and visual sciences, school of medicine, Yamagata university, 2-2-2 Iida-Nishi, Yamagata, Japan
A R T I C L E I N F O
A B S T R A C T
Article history: Received 5 March 2017 Received in revised form 14 June 2017 Accepted 14 June 2017 Available online xxx
Aim. – Hypoglycaemia is a common complication in diabetes patients. However, its relationship with retinopathy has not been well documented in patients with type 2 diabetes (T2D). This study aimed to investigate the associations between hypoglycaemia and the incidence and progression of diabetic retinopathy (DR). Methods. – In this longitudinal cohort study, which was part of the Japan Diabetes Complications Study (JDCS), adult patients with T2D were recruited at 59 diabetes clinics across Japan. Their history of hypoglycaemia was assessed by standardized self-reported questionnaires. Severe hypoglycaemia was defined as having at least one episode with coma requiring an outpatients visit or hospitalization. Adjusted hazard ratios (HRs) for incidence and progression of DR over 8 years of follow-up were determined. Results. – Of 1221 patients without DR, 127 (10.4%) had experienced non-severe hypoglycaemia within the previous year, whereas 10 (0.8%) reported severe hypoglycaemia episodes. During the 8-year followup involving 8492 person-years, 329 patients developed DR. In 410 patients with prevalent DR, the adjusted HRs for incident DR were 4.35 (95% CI: 1.98–9.56; P < 0.01) and, for progression of DR, 2.29 (95% CI: 0.45–11.78; P = 0.32) with severe hypoglycaemia. Conclusion. – Having a history of severe hypoglycaemia was one of the strongest predictors of incident DR in patients with T2D, with a fourfold increased risk. Identifying patients with greater risks of DR based on their history of hypoglycaemia may help to personalize risk evaluation in patients with diabetes.
C 2017 Elsevier Masson SAS. All rights reserved.
Keywords: Diabetic retinopathy Hypoglycaemia Microvascular complications Risk factors Type 2 diabetes
Introduction Hypoglycaemia is the most common complication of intensive glucose-lowering therapy in diabetes patients. In the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial, which was prematurely stopped owing to increased mortality possibly related
* Corresponding author. Fax: +81 23 628 5261. E-mail address: [email protected] (R. Kawasaki).
to hypoglycaemia, intensive glucose control in patients with type 2 diabetes (T2D) resulted in a threefold increase in hypoglycaemias requiring medical assistance [1]. Meta-analysis confirmed the adverse effects of intensive glucose control on hypoglycaemia [2], and similar associations were reported in a meta-analysis of the addition of dipeptidyl peptidase-4 inhibitors to sulphonylureas [3]. The prevalence of hypoglycaemia was estimated to be 45% for mild or moderate hypoglycaemia and 6% for severe hypoglycaemia in the entire study T2D population, whereas the prevalence was
http://dx.doi.org/10.1016/j.diabet.2017.06.002 C 2017 Elsevier Masson SAS. All rights reserved. 1262-3636/
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002
G Model
DIABET-902; No. of Pages 6 2
S. Tanaka et al. / Diabetes & Metabolism xxx (2017) xxx–xxx
50% for mild or moderate hypoglycaemia and 21% for severe hypoglycaemia in patients taking insulin [4]. It is well known that hypoglycaemia is a major cause of emergency hospitalizations in the elderly [5] and is life-threatening [6]. Moreover, a recent metaanalysis of six observational studies revealed that hypoglycaemia increases the risk of macrovascular complications threefold [7]. In earlier studies, rapid normalization of glucose status with intensive glucose management was suggested to have a potential role in aggravating the progression of diabetic retinopathy (DR), known as ‘early worsening’ [8]. The entire mechanism of this early worsening has not been fully elucidated, but hypoglycaemia is known to cause haemorheological changes, white cell activation, vasoconstriction, and the release of inflammatory mediators and cytokines [9]. Furthermore, fear of hypoglycaemia may have a negative impact by lowering adherence to diabetes management and glucose control regimens [10]. Data on microvascular complications related to hypoglycaemia in T2D are sparse. The Action in Diabetes and Vascular Disease: Preterax and Diamicron MR Controlled Evaluation (ADVANCE) study reported that hypoglycaemia increases major microvascular events [11]. Zhao et al. [12] reported that the risk of microvascular complications was increased in those who had hypoglycaemia within a year. However, it remains unknown whether hypoglycaemia specifically increases the incidence of DR, as the ADVANCE and Zhao et al. studies examined a composite outcome of new or worsening nephropathy or retinopathy. Also, a combined analysis of the ADVANCE and ACCORD studies suggested that the inconsistencies in the risk associations for microvascular complications between studies aiming at near-normalization of glucose might have been due to the balance between the beneficial effects of intensive glucose management and harmful effects of hypoglycaemia [13]. These lines of evidence have prompted our investigation into the associations between mild, moderate and severe hypoglycaemia and the incidence and progression of DR in a cohort of Japanese patients with T2D followed for 8 years. Methods Study cohort The present study was part of the Japan Diabetes Complications Study (JDCS), an open-labelled randomized trial originally designed to evaluate the efficacy of a long-term therapeutic intervention that was mainly focused on lifestyle education [14,15]. Incidence rates of DR did not significantly differ between the two randomized groups in that trial; therefore, for our study, data from these two groups were combined [16]. Those eligible for the study were subjects previously diagnosed with T2D and aged 40–70 years, whose HbA1c levels were 6.5% according to values established by the Japan Diabetes Society (JDS) and assays standardized by the Laboratory Test Committee of the JDS. From the outpatient clinics of 59 university and general hospitals nationwide specializing in diabetes care, 2205 patients were initially registered from January 1995 to March 1996. After excluding those with major ocular disease (such as glaucoma, dense cataract, history of cataract surgery), the population ultimately analyzed consisted of 1221 patients without DR (to assess incident DR) and 410 patients with DR (to assess progression of DR) [14,15]. Also analyzed were the follow-up data collected up to March 2003. The protocol for the study, which was in accord with the Declaration of Helsinki and Ethical Guidelines for Clinical/ Epidemiological Studies of the Japanese Ministry of Health Labour and Welfare, received ethics approval from the institutional review boards of all participating institutions. In addition, written informed consent was obtained from all enrolled patients.
Measurements Extensive surveys using standardized self-reported questionnaires were conducted at baseline and at 5 years after study registration. Patients’ history of hypoglycaemia was assessed by three items in the questionnaires: frequency of hypoglycaemia episodes within the previous year; frequency of coma due to hypoglycaemia within the previous year; and frequency of hypoglycaemia episodes requiring an outpatients visit or hospitalization within the previous year. For our primary analysis, ‘severe hypoglycaemia’ was defined as the occurrence of at least one episode of hypoglycaemia with either coma or the need for an outpatients visit or hospitalization, based on the baseline questionnaires. Compliance with oral medications or insulin was assessed by the following four choices: takes (injects) every time; does not take (inject) occasionally; does not take (inject) frequently; and almost never takes (injects). Other measurements included treatments, physical examinations, blood pressure measurements, neurological/ophthalmological examinations and laboratory tests, including those for HbA1c, fasting plasma glucose/ insulin/C-peptide, serum lipids/creatinine/urea nitrogen and urinalysis. Plasma insulin was not measured in those treated with insulin analogues. HbA1c assays were standardized by the JDS Laboratory Test Committee using JDS values. The US National Glycohemoglobin Standardization Program (NSGP) value for HbA1c was calculated as follows: 0.25 + 1.02 JDS value. All other laboratory measurements were done at each participating institution. Serum low-density lipoprotein (LDL) cholesterol was calculated using Friedewald’s equation except when triglycerides were > 400 mg/dL, in which case the LDL cholesterol data were treated as missing. Outcome measures The primary outcome of the present study was the incidence of DR in at least one eye. DR was evaluated annually by qualified ophthalmologists at each study institution using the International Clinical Disease Severity Scale for DR and for diabetic macular oedema disease with minor modifications [17]. Severity of DR was categorized into five stages: Stage 0, no retinopathy; Stage 1, haemorrhage and hard exudates; Stage 2, soft exudates; Stage 3, intraretinal microvascular abnormalities and venous changes, including beading, loops and duplication; and Stage 4, new vessels, vitreous haemorrhage, fibrous proliferation and retinal detachment. The incidence of DR was defined as having no signs of DR in either eye at baseline, but having mild-to-severe non-proliferative DR or proliferative DR (Stage 1 to Stage 4) confirmed at two consecutive follow-up years. This severity categorization is mostly comparable to the International Clinical Disease Severity Scale for DR Stages 1 and 2: mild-to-moderate non-proliferative diabetic retinopathy (NPDR); Stage 3: severe NDR; and Stage 4: proliferative diabetic retinopathy (PDR) [14]. To validate that the grading across study sites was consistent, fundus images were examined and evaluated by agreement of grading between local ophthalmologists and retinal specialists (R.K. and H.Y.). The estimated kappa statistic for agreement was 0.56 [95% confidence interval (CI): 0.52–0.59] and was above moderate. The secondary outcome was time from registration to progression to Stage 3 or 4 DR in at least one eye. Another outcome was also examined in the sensitivity analysis: time from registration to a major microvascular event, defined as DR or overt nephropathy, whichever happened first. The incidence of overt nephropathy was defined by spot urinary albumin excretion > 300 mg/g of creatinine in two consecutive samples.
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002
G Model
DIABET-902; No. of Pages 6 S. Tanaka et al. / Diabetes & Metabolism xxx (2017) xxx–xxx
Statistical analysis Hazard ratios (HRs) and 95% CIs for DR in relation to severity of hypoglycaemia were estimated by Cox regression models that included treatment with insulin, age, HbA1c, duration of diabetes, body mass index (BMI), systolic blood pressure, LDL cholesterol, high-density lipoprotein (HDL) cholesterol and log-triglycerides as covariates. Missing data were handled by the use of multiple imputations. Crude Cox regression without covariates and three sensitivity analyses were also performed: Cox regression analysis excluding patients not treated with insulin, which accounted for residual confounding related to diabetes treatment; Cox regression analysis with different exposure definitions (a non-severe hypoglycaemia group including patients who reported non-severe hypoglycaemia at baseline and no severe hypoglycaemia episodes at 5 years, and a severe hypoglycaemia group including patients who reported severe hypoglycaemia both at baseline and at 5 years); and Cox regression analysis for associations between hypoglycaemia and major microvascular events (incidence of DR or overt nephropathy). All P-values were two-sided, and the significance level was 0.05. All statistical analyses and data management were conducted at a central data centre using SAS version 9.4 software (SAS Institute Inc., Cary, NC, USA). Results Of the 1221 patients without DR at baseline, 127 (10.4%) had experienced non-severe hypoglycaemia within the previous year, while 10 (0.8%) reported hypoglycaemia episodes with coma or requiring treatment (Table 1). The severity of hypoglycaemia was significantly associated with gender, duration of diabetes, use of hypoglycaemic or lipid-lowering agents, C-peptide reactivity, BMI, and LDL and HDL cholesterol. Table S1 (see supplementary materials associated with this article online) describes the use of hypoglycaemic agents, the frequency of hypoglycaemia, and
3
compliance with medications at 4 to 5 years of follow-up. In fact, the frequency of severe hypoglycaemia during follow-up was rather low, and compliance with oral medications and insulin did not significantly differ among these three groups. The 8-year follow-up rate was 76.8%. There were no notable differences in baseline characteristics between patients who completed the 8-year follow-up and those who did not [10]. Longitudinal trends in mean HbA1c were unstable in the severe hypoglycaemia group, possibly due to our limited sample size (Fig. 1A). Levels of C-peptide reactivity in the non-severe and severe hypoglycaemia groups were lower at baseline than in the no hypoglycaemia group and decreased during the 8-year follow-up (Fig. 1B). Also, during the follow-up, 8492 person-years were studied and 329 incidents of DR were observed (283 with no hypoglycaemia, 38 with non-severe hypoglycaemia and eight with severe hypoglycaemia). Fig. 2 shows Kaplan–Meier curves for incident DR, and Table 2 compares the incidence of DR and Cox regression analysis of crude HRs with those after adjusting for potential confounders. Adjusted HRs were 0.90 (95% CI: 0.61–1.34; P = 0.61) for non-severe hypoglycaemia within the previous year and 4.35 (95% CI: 1.98–9.56; P < 0.01) for severe hypoglycaemia within the previous year. Table 3 shows the results of our three sensitivity analyses. The estimates of HRs ranged within the CIs of the analysis restricted to insulin users (Model 1) and the Cox regression analysis using different exposure definitions (Model 2). After excluding 141 patients who had albumin excretion rates 300 mg/g creatinine at baseline, or other clinical findings of renal disease, 33 major microvascular events (incidence of DR or overt nephropathy) were seen in 1080 patients (288 with no hypoglycaemia, 38 with mild hypoglycaemia and seven with severe hypoglycaemia). The adjusted HRs for major microvascular events were similar to those for DR (Model 3). In another cohort of 410 patients with prevalent DR, 73 (17.8%) had experienced non-severe hypoglycaemia within the previous year, while seven patients (1.7%) reported hypoglycaemia episodes with coma or requiring treatment (Table S2; see supplementary materials associated with this article online). In this cohort, the adjusted HRs for progression of DR were 1.08 (95% CI: 0.48–2.42;
Table 1 Background characteristics of 1221 patients with type 2 diabetes without diabetic retinopathy by history of hypoglycaemia during the previous year.
Any hypoglycaemia (frequency/week) Hypoglycaemia with coma (frequency/week) Hypoglycaemia requiring treatment (frequency/week) Age (years) Women (%) HbA1c (%, NGSP value) Duration of diabetes (years) Treated with oral hypoglycaemic agents without insulin (%) Treated with sulphonylureas without insulin (%) Treated with insulin (%) Treated with antihypertensive agents (%) Treated with lipid-lowering agents (%) C-peptide reactivity (ng/mL) Body mass index (kg/m2) Systolic blood pressure (mmHg) LDL cholesterol (mg/dL) HDL cholesterol (mg/dL) Triglyceridesa (mg/dL)
No hypoglycaemia (n = 1084)
Non-severe hypoglycaemia (n = 127)
Mean
Mean
0.00 0.00 0.00 58.3 43.7 8.2 9.5 64.9 58.1 7.8 25.3 24.8 2.2 23.2 131.3 124.3 53.5 107.0
SD 0.00 0.00 0.00 6.8 1.3 6.6
2.3 3.0 15.9 32.3 16.4 79.0
1.08 0.00 0.00 57.5 55.9 8.4 11.8 32.3 30.7 62.2 22.0 18.1 1.6 22.0 127.9 117.2 59.6 81.0
Severe hypoglycaemiab (n = 10) SD 1.65 0.00 0.00 7.5 1.3 8.2
2.2 3.2 17.3 30.5 16.9 54.0
Mean 1.08 0.90 0.80 58.7 50.0 8.5 12.3 20.0 20.0 80.0 20.0 60.0 1.6 22.5 128.7 102.5 61.2 126.0
P
SD 1.18 0.57 0.63 6.1 1.8 6.0
1.1 2.7 19.3 25.7 19.1 80.0
< 0.01 < 0.01 < 0.01 0.43 0.03 0.21 < 0.01 < 0.01 < 0.01 < 0.01 0.67 0.01 0.03 < 0.01 0.07 < 0.01 < 0.01 0.07
NGSP: National Glycohemoglobin Standardization Program; LDL/HDL: low-density/high-density lipoprotein. a Data are medians and interquartile range. b At least one episode with coma or requiring an outpatients visit or hospitalization.
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002
G Model
DIABET-902; No. of Pages 6 S. Tanaka et al. / Diabetes & Metabolism xxx (2017) xxx–xxx
4
Fig. 1. Longitudinal trends in mean HbA1c (left) and C-reactive protein (right) over 8 years of follow-up according to patients’ history of hypoglycaemia at baseline.
Fig. 2. Kaplan–Meier curves of incident diabetic retinopathy according to patients’ history of hypoglycaemia at baseline.
P = 0.86) for non-severe hypoglycaemia and 2.29 (95% CI: 0.45– 11.78; P = 0.32) for severe hypoglycaemia (Table 4). Discussion Our present study of Japanese patients with T2D has demonstrated a significant increase in the risk of DR among patients who had experienced severe hypoglycaemia, but no notable effects from mild hypoglycaemia episodes on outcomes. The striking difference between our findings and those of the ADVANCE study [11] was that, in our cohort, severe hypoglycaemia was one of the strongest predictors of DR, showing an approximately fourfold increase in incidence rate, which is numerically larger than the previously reported effects of HbA1c, duration of diabetes, systolic blood pressure and BMI [14]. Although the definitions of hypoglycaemia were not fully comparable, it was also notable that non-severe hypoglycaemia seemed to be less frequent in our study than in the ADVANCE study [11]. The proportions of subjects experiencing hypoglycaemia within the previous year in our study were 10.4% for non-severe
and 0.8% for severe hypoglycaemias, equivalent to annual cumulative incidences of 0.104 and 0.008 per person-years, respectively. When compared with the reported incidence rates in the ADVANCE study (1.1 and 0.006 per person-years, respectively) [11], the frequency of non-severe hypoglycaemia was rather low in the present study while the frequency of severe hypoglycaemia was comparable. In previous clinical studies, hypoglycaemia was surveyed through self-reported questionnaires, prospective reports or diaries, emergency-department admission records and claims databases [4]. Unlike our study, which used self-reported questionnaires, the ADVANCE study assessed hypoglycaemia prospectively, and their criteria for severe hypoglycaemia were broader, as it was defined as the inability of patients to treat themselves [11]. Differences between the ADVANCE and our study are therefore partly attributable to the use of self-reported questionnaires and different definitions of hypoglycaemia. It is not plausible, however, that the difference in the size of the effect would be a consequence of misclassification, as non-differential misclassification is expected to weaken associations. Another explanation is that the effects of hypoglycaemia were attenuated in the ADVANCE study because the study population included patients with prevalent microvascular complications, and the study outcome was a composite of new or worsening nephropathy or retinopathy. Indeed, the size of the effect on progression of DR was about half that of the incident DR. Our study also examined associations with a composite of microvascular events as a sensitivity analysis, and the size of the effect was still considerably larger than the findings observed in the ADVANCE study (Table 3). Other speculations to explain these contradictory findings include differences in the severity of diabetes, diabetes treatment and ethnicity. At present, the biological mechanisms behind the potential effects of hypoglycaemia on the microvasculature remain hypothetical, but one possible explanation is oxidative and inflammatory stress, which can be provoked by acute and chronic hypoglycaemia [7]. It is known that, in retinal cells, glucose deprivation leads to upregulation of vascular endothelial growth factor [18]. In fact, epidemiological studies of both type 1 diabetes (T1D) and T2D have revealed that markers of inflammation and endothelial dysfunction correlate with DR [19,20]. Unfortunately, data on inflammation and endothelial dysfunction were not available for our cohort. It is also possible that, instead of a direct causal link, severe hypoglycaemia is associated with the risk of DR through coexisting conditions that remain unmeasured or incompletely quantified. It may also be speculated that residual confounders are related to the severity and treatment of diabetes.
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002
G Model
DIABET-902; No. of Pages 6 S. Tanaka et al. / Diabetes & Metabolism xxx (2017) xxx–xxx
5
Table 2 Associations between history of hypoglycaemia during the previous year and incident diabetic retinopathy in 1221 patients with type 2 diabetes.
History of hypoglycaemia None Non-severe hypoglycaemia Severe hypoglycaemiaa Treated with insulin Age (+ 10 years) Gender (women) HbA1c (%, NGSP value) Duration of diabetes (+ 10 years) Body mass index (kg/m2) C-protein reactivity (ng/mL) Systolic blood pressure (+ 10 mmHg) LDL cholesterol (+ 10 mg/dL) HDL cholesterol (+ 10 mg/dL) Log-triglycerides (+ 1 unit)
Crude HR
95% CI
1 1.16 5.03
Reference 0.83 2.34
P
1.63 10.79
0.38 < 0.01
Adjusted HR
95% CI
Reference 0.89 4.30 1.41 0.94 1.11 1.30 1.61 1.04 1.01 1.10 1.02 1.00 1.04
0.60 1.96 1.01 0.80 0.88 1.22 1.38 1.00 0.96 1.02 0.98 0.93 0.83
P
1.32 9.44 1.97 1.11 1.39 1.39 1.87 1.08 1.06 1.18 1.05 1.07 1.31
0.56 < 0.01 0.047 0.50 0.39 < 0.01 < 0.01 0.04 0.69 0.01 0.30 0.96 0.71
Missing data for 37 patients were multiply imputed; multivariate model adjusted for insulin use, gender, age, HbA1c, duration of diabetes, body mass index, C-peptide reactivity, systolic blood pressure; HR: hazard ratio; CI: confidence interval; NGSP: National Glycohemoglobin Standardization Program; LDL/HDL: low-density/high-density lipoprotein. Multivariate model adjusted for insulin use, sex, age, HbA1c, duration of diabetes, body mass index, C-peptide reactivity, systolic blood pressure, LDL cholesterol, HDL cholesterol and triglycerides. a At least one episode of coma or requiring outpatients visit or hospitalization.
Table 3 Sensitivity analysis for associations between hypoglycaemia and incident diabetic retinopathy (DR) or major microvascular events in 1221 patients with type 2 diabetes.
Baseline modelb Non-severe hypoglycaemia Severe hypoglycaemiaa Model 1c Non-severe hypoglycaemia Severe hypoglycaemia Model 2d Non-severe hypoglycaemia Severe hypoglycaemia Model 3e Non-severe hypoglycaemia Severe hypoglycaemia
Adjusted hazard ratio
95% confidence interval
P
0.89 4.33
0.60 1.97
1.32 9.50
0.58 < 0.01
0.95 6.73
0.56 2.84
1.61 15.93
0.56 < 0.01
1.07 3.62
0.73 1.69
1.57 7.74
0.72 < 0.01
0.91 5.45
0.62 2.43
1.33 12.24
0.63 < 0.01
a
Episodes with coma or requiring an outpatients visit or hospitalization. Multivariate-adjusted models for 1221 patients, not adjusted for body mass index with missing data for 37 patients multiply imputed. c Multivariate-adjusted models for 171 insulin-treated patients, not adjusted for insulin use with missing data for 17 patients multiply imputed. d Multivariate-adjusted models for non-severe hypoglycaemia group were patients reporting non-severe hypoglycaemia at baseline and at 5 years with no severe hypoglycaemia episodes; the severe hypoglycaemia group included patients reporting severe hypoglycaemia at baseline and at 5 years, with missing data for 126 patients multiply imputed. e Multivariate-adjusted models for associations between hypoglycaemia and major microvascular events (diabetic retinopathy: 329 events; overt nephropathy: 73 events); 141 patients with an albumin excretion rate 300 mg/g creatinine at baseline or other clinical findings of renal disease were excluded; missing data for 82 patients were multiply imputed. b
Indeed, hypoglycaemia was frequent among users of sulphonylureas and insulin, and decreasing trends in C-protein reactivity were observed in both non-severe and severe hypoglycaemia groups. On the other hand, no notable differences were observed across groups in either glucose control during follow-up (Fig. 1A) or compliance with oral medications and insulin (Table S1; see supplementary materials associated with this article online). Furthermore, and somewhat surprisingly, the associations between hypoglycaemia and DR remained significant even after restricting the population analysis to insulin users (Table 3). The findings from our present analyses and the ADVANCE study markedly contradict observations regarding T1D. A recent re-analysis of the dataset from the Diabetes Control and Complications Trial
Table 4 Associations between history of hypoglycaemia during the previous year and progression of diabetic retinopathy (DR) in 410 patients with type 2 diabetes and prevalent DR.
Non-severe hypoglycaemia Severe hypoglycaemiaa Treated with insulin Age (+ 10 years) Gender (women) HbA1c (%, NGSP value) Duration of diabetes (+ 10 years) Body mass index (kg/m2) C-protein reactivity (ng/mL) Systolic blood pressure (+ 10 mmHg) LDL cholesterol (+ 10 mg/dL) HDL cholesterol (+ 10 mg/dL) Log-triglycerides (+ 1 unit)
Hazard ratio
95% confidence interval
P
1.03 2.61 0.76 0.72 1.50 1.69 1.10 1.02 1.03 1.12 0.92 0.86 0.80
0.45 0.52 0.39 0.49 0.88 1.41 0.73 0.94 0.97 0.96 0.85 0.71 0.44
0.95 0.25 0.42 0.09 0.14 < 0.01 0.64 0.60 0.39 0.15 0.03 0.10 0.46
2.33 13.21 1.49 1.05 2.58 2.02 1.66 1.12 1.09 1.31 0.99 1.03 1.44
Missing data for 51 patients were multiply imputed. NGSP: National Glycohemoglobin Standardization Program; LDL/HDL: low-density/high-density lipoprotein; NGSP: National Glycohemoglobin Standardization Program; LDL/HDL: low-density/ high-density lipoprotein. a Episodes with coma or requiring an outpatients visit or hospitalization.
(DCCT) [21] found that the frequency of severe hypoglycaemia during the DCCT did not predict the risk of DR over 9 years. Furthermore, same-day glucose variability was not associated with DR risk [22], whereas variability in HbA1c was a significant risk factor [23] in the DCCT. These observations are also not consistent with systematic reviews of glycaemic and HbA1c variability in T2D [24,25]. However, it should be noted that the definitions of exposure status in our study differed from those of the DCCT re-analysis. Although the criteria for severity were similar, our study focused on the effects of hypoglycaemia within the previous year, while hypoglycaemia during follow-up was accounted for only in the sensitivity analysis to avoid reverse causation. Our findings have both clinical and research implications. From the viewpoint of diabetes care, screening for DR is particularly recommended for patients experiencing severe hypoglycaemia. As suggested by some researchers, screening programmes could revise their screening intervals according to risk stratification [26]. The absolute risk of DR for a given patient can be calculated based on age, HbA1c and other risk factors [27], and safe screening intervals may be determined based on this calculated risk.
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002
G Model
DIABET-902; No. of Pages 6 S. Tanaka et al. / Diabetes & Metabolism xxx (2017) xxx–xxx
6
Limitations of our present study include the accuracy of hypoglycaemia assessment, which was self-reported and may have been subject to recall bias. The frequency of severe hypoglycaemia is rather high in this analysis. Our definition of ‘severe hypoglycaemia’ was at least one episode with coma or requiring an outpatients visit or hospitalization, thereby including a range of episodes from an outpatients visit to coma. In addition, only 10 patients without DR and seven patients with DR at baseline had experienced severe hypoglycaemia, limiting our statistical analysis in terms of details within this subpopulation and its effects on progression. Given these limitations, the generalizability of the present findings need to be confirmed by other cohorts with larger sample sizes and/or longer follow-ups, with detailed assessments of both hypoglycaemic events and DR. Moreover, data on the dosages of antidiabetic agents were not available, and the association between insulin doses and hypoglycaemia was unfortunately unknown. A further limitation was the diagnostic accuracy of DR based on clinical diagnosis compared with grading based on seven-field stereo fundus photography. Finally, our findings in Japanese patients may not be generally applicable to populations with different ethnic and genetic factors. In conclusion, severe hypoglycaemia is one of the strongest predictors of incident DR in patients with long-standing T2D and has an approximately fourfold greater risk. Also, tracking the number of episodes of hypoglycaemia may be helpful for personalizing diabetes care and for devising risk-based screening intervals for DR. Funding Funding/Support: This work was financially supported by the Ministry of Health, Labor and Welfare, Japan. Financial disclosures: No financial disclosures. Disclosure of interest The authors declare that they have no competing interest. Acknowledgements We sincerely thank the late Professor Nobuhiro Yamada, former director of the JDCS, who always provided warm spiritual support to us all. We thank all the patients and diabetologists at the 59 participating institutions for their long-standing collaboration in the JDCS. Thanks are extended to Ms Mami Haga and Ms Natsuko Tada, and Niigata University, for their excellent secretarial assistance. A medical editor in NAI Inc. reviewed the writing of the manuscript. The sponsor had no role in the design and conduct of the study.
Appendix A. Supplementary data Supplementary materials (Tables S1 and S2) associated with this article can be found at http://www.scincedirect.com at http:// dx.doi.org/10.1016/j.diabet.2017.06.002.
References [1] Action to Control Cardiovascular Risk in Diabetes Study Group. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med 2008;358:2545–59. [2] Boussageon R, Bejan-Angoulvant T, Saadatian-Elahi M, et al. Effect of intensive glucose lowering treatment on all-cause mortality, cardiovascular death, and microvascular events in type 2 diabetes: meta-analysis of randomised controlled trials. BMJ 2011;343:d4169.
[3] Salvo F, Moore N, Arnaud M, Robinson P, Raschi E, De Ponti F, et al. Addition of dipeptidyl peptidase-4 inhibitors to sulphonylureas and risk of hypoglycaemia: systematic review and meta-analysis. BMJ 2016;353:i2231. [4] Edridge CL, Dunkley AJ, Bodicoat DH, Rose TC, Gray LJ, Davies MJ, et al. Prevalence and incidence of hypoglycaemia in 532,542 people with type 2 diabetes on oral therapies and insulin: a systematic review and metaanalysis of population based studies. PLoS One 2015;10(6):e0126427. [5] Budnitz DS, Lovegrove MC, Shehab N, Richards CL. Emergency hospitalizations for adverse drug events in older Americans. N Engl J Med 2011;365:2002–12. [6] Bonds DE, Miller ME, Bergenstal RM, Buse JB, Byington RP, Cutler JA, et al. The association between symptomatic, severe hypoglycaemia and mortality in type 2 diabetes: retrospective epidemiological analysis of the ACCORD study. BMJ 2010;340:b4909. [7] Goto A, Arah OA, Goto M, Terauchi Y, Noda M. Severe hypoglycaemia and cardiovascular disease: systematic review and meta-analysis with bias analysis. BMJ 2013;347:f4533. [8] The Kroc Collaborative Study Group. Diabetic retinopathy after two years of intensified insulin treatment. Follow-up of the Kroc Collaborative Study. JAMA 1988;260:37–41. [9] Wright RJ, Frier BM. Vascular disease and diabetes: is hypoglycaemia an aggravating factor? Diabetes Metab Res Rev 2008;24:353–63. [10] Wild D, von Maltzahn R, Brohan E, Christensen T, Clauson P, Gonder-Frederick L. A critical review of the literature on fear of hypoglycaemia in diabetes: implications for diabetes management and patient education. Patient Educ Couns 2007;68:10–5. [11] Zoungas S, Patel A, Chalmers J, de Galan BE, Li Q, Billot L, et al. Severe hypoglycaemia and risks of vascular events and death. N Engl J Med 2010;363:1410–8. [12] Zhao Y, Campbell CR, Fonseca V, Shi L. Impact of hypoglycaemia associated with antihyperglycemic medications on vascular risks in veterans with type 2 diabetes. Diabetes Care 2012;35:1126–32. [13] O’Connor PJ, Ismail-Beigi F. Near-Normalization of glucose and microvascular diabetes complications: data from ACCORD and ADVANCE. Ther Adv Endocrinol Metab 2011;2:17–26. [14] Kawasaki R, Tanaka S, Tanaka S, Yamamoto T, Sone H, Ohashi Y, et al. Incidence and progression of diabetic retinopathy in Japanese adults with type 2 diabetes: 8-year follow-up study of the Japan Diabetes Complications Study (JDCS). Diabetologia 2011;54:2288–94. [15] Tanaka S, Tanaka S, Iimuro S, Yamashita H, Katayama S, Ohashi Y, et al. Cohort profile: the Japan diabetes complications study: a long-term follow-up of a randomised lifestyle intervention study of type 2 diabetes. Int J Epidemiol 2014;43:1054–62. [16] Sone H, Tanaka S, Iimuro S, Tanaka S, Oida K, Yamasaki Y, et al. Long-term lifestyle intervention lowers the incidence of stroke in Japanese patients with type 2 diabetes: a nationwide multicentre randomised controlled trial (the Japan Diabetes Complications Study). Diabetologia 2010;53:419–28. [17] Wilkinson CP, Ferris 3rd FL, Klein RE, Lee PP, Agardh CD, Davis M, et al. Proposed international clinical diabetic retinopathy and diabetic macular oedema disease severity scales. Ophthalmology 2003;110:1677–82. [18] Sone H, Kawakami Y, Okuda Y, Kondo S, Hanatani M, Suzuki H, et al. Vascular endothelial growth factor is induced by long-term high glucose concentration and up-regulated by acute glucose deprivation in cultured bovine retinal pigmented epithelial cells. Biochem Biophys Res Commun 1996;221:193–8. [19] Schram MT, Chaturvedi N, Schalkwijk CG, Fuller JH, Stehouwer CD, EURODIAB Prospective Complications Study Group. Markers of inflammation are crosssectionally associated with microvascular complications and cardiovascular disease in type 1 diabetes — the EURODIAB Prospective Complications Study. Diabetologia 2005;48:370–8. [20] van Hecke MV, Dekker JM, Nijpels G, Moll AC, Heine RJ, Bouter LM, et al. Inflammation and endothelial dysfunction are associated with retinopathy: the Hoorn Study. Diabetologia 2005;48:1300–6. [21] Kilpatrick ES, Rigby AS, Atkin SL, Frier BM. Does severe hypoglycaemia influence microvascular complications in type 1 diabetes? An analysis of the diabetes control and complications trial database. Diabet Med 2012;29:1195–8. [22] Kilpatrick ES, Rigby AS, Atkin SL. Effect of glucose variability on the long-term risk of microvascular complications in type 1 diabetes. Diabetes Care 2009;32:1901–3. [23] Kilpatrick ES, Rigby AS, Atkin SL. A1C variability and the risk of microvascular complications in type 1 diabetes: data from the Diabetes Control and Complications Trial. Diabetes Care 2008;31:2198–202. [24] Smith-Palmer J, Bra¨ndle M, Trevisan R, Orsini Federici M, Liabat S, Valentine W. Assessment of the association between glycaemic variability and diabetesrelated complications in type 1 and type 2 diabetes. Diabetes Res Clin Pract 2014;105:273–84. [25] Gorst C, Kwok CS, Aslam S, Buchan I, Kontopantelis E, Myint PK, et al. Longterm glycaemic variability and risk of adverse outcomes: a systematic review and meta-analysis. Diabetes Care 2015;38:2354–69. [26] Thomas RL, Dunstan F, Luzio SD, Roy Chowdury S, Hale SL, North RV, et al. Incidence of diabetic retinopathy in people with type 2 diabetes mellitus attending the Diabetic Retinopathy Screening Service for Wales: retrospective analysis. BMJ 2012;344:e874. [27] Tanaka S, Tanaka S, Iimuro S, Yamashita H, Katayama S, Akanuma Y, et al. Predicting macro- and microvascular complications in type 2 diabetes: the Japan Diabetes Complications Study/the Japanese Elderly Diabetes Intervention Trial risk engine. Diabetes Care 2013;36:1193–9.
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002
DIABET-902; No. of Pages 6 Diabetes & Metabolism xxx (2017) xxx–xxx
Available online at
ScienceDirect www.sciencedirect.com
Original article
Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes S. Tanaka a, R. Kawasaki b,*, S. Tanaka-Mizuno c, S. Iimuro d, S. Matsunaga e, T. Moriya f, S. Ishibashi g, S. Katayama h, Y. Ohashi i, Y. Akanuma j, H. Sone e, H. Yamashita k, for the Japan Diabetes Complications Study Group a
Department of pharmacoepidemiology, graduate longitudinal school of medicine and public health, Kyoto university, Yoshida-Konoe-cho, Sakyo-ku, Kyoto, Japan b Department of public health, Yamagata university graduate school of medical science, 2-2-2 Iida-Nishi, Yamagata, Japan c Division of medical statistics, Shiga university of medical science, Tsukinowa Seta-cho, Ohtsu, Shiga, Japan d Teikyo academic research centre, Teikyo university, Kaga, Itabashi-ku, Tokyo, Japan e Department of haematology, endocrinology and metabolism, faculty of medicine, Niigata university, 1-757 Asahi-machi, Chuo-ku, Niigata, Japan f Health care centre, Kitasato university, 1-15-1 Kitasato, Minami-ku, Sagamihara-shi, Kanagawa, Japan g Division of endocrinology and metabolism, school of medicine, Jichi medical university, 3311-1 Yakushiji, Shimotsuke, Tochigi, Japan h Kawagoe clinic, Saitama medical university, 21-7 Wakitahonchou, Kawagoe, Saitama, Japan i Department of integrated science and engineering for sustainable society, Chuo university, 1-13-27, Kasuga, Bunkyo-ku, Tokyo, Japan j The institute for adult diseases Asahi life foundation, 1-6-1 Marunouchi Chiyoda-ku, Tokyo, Japan k Department of ophthalmology and visual sciences, school of medicine, Yamagata university, 2-2-2 Iida-Nishi, Yamagata, Japan
A R T I C L E I N F O
A B S T R A C T
Article history: Received 5 March 2017 Received in revised form 14 June 2017 Accepted 14 June 2017 Available online xxx
Aim. – Hypoglycaemia is a common complication in diabetes patients. However, its relationship with retinopathy has not been well documented in patients with type 2 diabetes (T2D). This study aimed to investigate the associations between hypoglycaemia and the incidence and progression of diabetic retinopathy (DR). Methods. – In this longitudinal cohort study, which was part of the Japan Diabetes Complications Study (JDCS), adult patients with T2D were recruited at 59 diabetes clinics across Japan. Their history of hypoglycaemia was assessed by standardized self-reported questionnaires. Severe hypoglycaemia was defined as having at least one episode with coma requiring an outpatients visit or hospitalization. Adjusted hazard ratios (HRs) for incidence and progression of DR over 8 years of follow-up were determined. Results. – Of 1221 patients without DR, 127 (10.4%) had experienced non-severe hypoglycaemia within the previous year, whereas 10 (0.8%) reported severe hypoglycaemia episodes. During the 8-year followup involving 8492 person-years, 329 patients developed DR. In 410 patients with prevalent DR, the adjusted HRs for incident DR were 4.35 (95% CI: 1.98–9.56; P < 0.01) and, for progression of DR, 2.29 (95% CI: 0.45–11.78; P = 0.32) with severe hypoglycaemia. Conclusion. – Having a history of severe hypoglycaemia was one of the strongest predictors of incident DR in patients with T2D, with a fourfold increased risk. Identifying patients with greater risks of DR based on their history of hypoglycaemia may help to personalize risk evaluation in patients with diabetes.
C 2017 Elsevier Masson SAS. All rights reserved.
Keywords: Diabetic retinopathy Hypoglycaemia Microvascular complications Risk factors Type 2 diabetes
Introduction Hypoglycaemia is the most common complication of intensive glucose-lowering therapy in diabetes patients. In the Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial, which was prematurely stopped owing to increased mortality possibly related
* Corresponding author. Fax: +81 23 628 5261. E-mail address: [email protected] (R. Kawasaki).
to hypoglycaemia, intensive glucose control in patients with type 2 diabetes (T2D) resulted in a threefold increase in hypoglycaemias requiring medical assistance [1]. Meta-analysis confirmed the adverse effects of intensive glucose control on hypoglycaemia [2], and similar associations were reported in a meta-analysis of the addition of dipeptidyl peptidase-4 inhibitors to sulphonylureas [3]. The prevalence of hypoglycaemia was estimated to be 45% for mild or moderate hypoglycaemia and 6% for severe hypoglycaemia in the entire study T2D population, whereas the prevalence was
http://dx.doi.org/10.1016/j.diabet.2017.06.002 C 2017 Elsevier Masson SAS. All rights reserved. 1262-3636/
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002
G Model
DIABET-902; No. of Pages 6 2
S. Tanaka et al. / Diabetes & Metabolism xxx (2017) xxx–xxx
50% for mild or moderate hypoglycaemia and 21% for severe hypoglycaemia in patients taking insulin [4]. It is well known that hypoglycaemia is a major cause of emergency hospitalizations in the elderly [5] and is life-threatening [6]. Moreover, a recent metaanalysis of six observational studies revealed that hypoglycaemia increases the risk of macrovascular complications threefold [7]. In earlier studies, rapid normalization of glucose status with intensive glucose management was suggested to have a potential role in aggravating the progression of diabetic retinopathy (DR), known as ‘early worsening’ [8]. The entire mechanism of this early worsening has not been fully elucidated, but hypoglycaemia is known to cause haemorheological changes, white cell activation, vasoconstriction, and the release of inflammatory mediators and cytokines [9]. Furthermore, fear of hypoglycaemia may have a negative impact by lowering adherence to diabetes management and glucose control regimens [10]. Data on microvascular complications related to hypoglycaemia in T2D are sparse. The Action in Diabetes and Vascular Disease: Preterax and Diamicron MR Controlled Evaluation (ADVANCE) study reported that hypoglycaemia increases major microvascular events [11]. Zhao et al. [12] reported that the risk of microvascular complications was increased in those who had hypoglycaemia within a year. However, it remains unknown whether hypoglycaemia specifically increases the incidence of DR, as the ADVANCE and Zhao et al. studies examined a composite outcome of new or worsening nephropathy or retinopathy. Also, a combined analysis of the ADVANCE and ACCORD studies suggested that the inconsistencies in the risk associations for microvascular complications between studies aiming at near-normalization of glucose might have been due to the balance between the beneficial effects of intensive glucose management and harmful effects of hypoglycaemia [13]. These lines of evidence have prompted our investigation into the associations between mild, moderate and severe hypoglycaemia and the incidence and progression of DR in a cohort of Japanese patients with T2D followed for 8 years. Methods Study cohort The present study was part of the Japan Diabetes Complications Study (JDCS), an open-labelled randomized trial originally designed to evaluate the efficacy of a long-term therapeutic intervention that was mainly focused on lifestyle education [14,15]. Incidence rates of DR did not significantly differ between the two randomized groups in that trial; therefore, for our study, data from these two groups were combined [16]. Those eligible for the study were subjects previously diagnosed with T2D and aged 40–70 years, whose HbA1c levels were 6.5% according to values established by the Japan Diabetes Society (JDS) and assays standardized by the Laboratory Test Committee of the JDS. From the outpatient clinics of 59 university and general hospitals nationwide specializing in diabetes care, 2205 patients were initially registered from January 1995 to March 1996. After excluding those with major ocular disease (such as glaucoma, dense cataract, history of cataract surgery), the population ultimately analyzed consisted of 1221 patients without DR (to assess incident DR) and 410 patients with DR (to assess progression of DR) [14,15]. Also analyzed were the follow-up data collected up to March 2003. The protocol for the study, which was in accord with the Declaration of Helsinki and Ethical Guidelines for Clinical/ Epidemiological Studies of the Japanese Ministry of Health Labour and Welfare, received ethics approval from the institutional review boards of all participating institutions. In addition, written informed consent was obtained from all enrolled patients.
Measurements Extensive surveys using standardized self-reported questionnaires were conducted at baseline and at 5 years after study registration. Patients’ history of hypoglycaemia was assessed by three items in the questionnaires: frequency of hypoglycaemia episodes within the previous year; frequency of coma due to hypoglycaemia within the previous year; and frequency of hypoglycaemia episodes requiring an outpatients visit or hospitalization within the previous year. For our primary analysis, ‘severe hypoglycaemia’ was defined as the occurrence of at least one episode of hypoglycaemia with either coma or the need for an outpatients visit or hospitalization, based on the baseline questionnaires. Compliance with oral medications or insulin was assessed by the following four choices: takes (injects) every time; does not take (inject) occasionally; does not take (inject) frequently; and almost never takes (injects). Other measurements included treatments, physical examinations, blood pressure measurements, neurological/ophthalmological examinations and laboratory tests, including those for HbA1c, fasting plasma glucose/ insulin/C-peptide, serum lipids/creatinine/urea nitrogen and urinalysis. Plasma insulin was not measured in those treated with insulin analogues. HbA1c assays were standardized by the JDS Laboratory Test Committee using JDS values. The US National Glycohemoglobin Standardization Program (NSGP) value for HbA1c was calculated as follows: 0.25 + 1.02 JDS value. All other laboratory measurements were done at each participating institution. Serum low-density lipoprotein (LDL) cholesterol was calculated using Friedewald’s equation except when triglycerides were > 400 mg/dL, in which case the LDL cholesterol data were treated as missing. Outcome measures The primary outcome of the present study was the incidence of DR in at least one eye. DR was evaluated annually by qualified ophthalmologists at each study institution using the International Clinical Disease Severity Scale for DR and for diabetic macular oedema disease with minor modifications [17]. Severity of DR was categorized into five stages: Stage 0, no retinopathy; Stage 1, haemorrhage and hard exudates; Stage 2, soft exudates; Stage 3, intraretinal microvascular abnormalities and venous changes, including beading, loops and duplication; and Stage 4, new vessels, vitreous haemorrhage, fibrous proliferation and retinal detachment. The incidence of DR was defined as having no signs of DR in either eye at baseline, but having mild-to-severe non-proliferative DR or proliferative DR (Stage 1 to Stage 4) confirmed at two consecutive follow-up years. This severity categorization is mostly comparable to the International Clinical Disease Severity Scale for DR Stages 1 and 2: mild-to-moderate non-proliferative diabetic retinopathy (NPDR); Stage 3: severe NDR; and Stage 4: proliferative diabetic retinopathy (PDR) [14]. To validate that the grading across study sites was consistent, fundus images were examined and evaluated by agreement of grading between local ophthalmologists and retinal specialists (R.K. and H.Y.). The estimated kappa statistic for agreement was 0.56 [95% confidence interval (CI): 0.52–0.59] and was above moderate. The secondary outcome was time from registration to progression to Stage 3 or 4 DR in at least one eye. Another outcome was also examined in the sensitivity analysis: time from registration to a major microvascular event, defined as DR or overt nephropathy, whichever happened first. The incidence of overt nephropathy was defined by spot urinary albumin excretion > 300 mg/g of creatinine in two consecutive samples.
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002
G Model
DIABET-902; No. of Pages 6 S. Tanaka et al. / Diabetes & Metabolism xxx (2017) xxx–xxx
Statistical analysis Hazard ratios (HRs) and 95% CIs for DR in relation to severity of hypoglycaemia were estimated by Cox regression models that included treatment with insulin, age, HbA1c, duration of diabetes, body mass index (BMI), systolic blood pressure, LDL cholesterol, high-density lipoprotein (HDL) cholesterol and log-triglycerides as covariates. Missing data were handled by the use of multiple imputations. Crude Cox regression without covariates and three sensitivity analyses were also performed: Cox regression analysis excluding patients not treated with insulin, which accounted for residual confounding related to diabetes treatment; Cox regression analysis with different exposure definitions (a non-severe hypoglycaemia group including patients who reported non-severe hypoglycaemia at baseline and no severe hypoglycaemia episodes at 5 years, and a severe hypoglycaemia group including patients who reported severe hypoglycaemia both at baseline and at 5 years); and Cox regression analysis for associations between hypoglycaemia and major microvascular events (incidence of DR or overt nephropathy). All P-values were two-sided, and the significance level was 0.05. All statistical analyses and data management were conducted at a central data centre using SAS version 9.4 software (SAS Institute Inc., Cary, NC, USA). Results Of the 1221 patients without DR at baseline, 127 (10.4%) had experienced non-severe hypoglycaemia within the previous year, while 10 (0.8%) reported hypoglycaemia episodes with coma or requiring treatment (Table 1). The severity of hypoglycaemia was significantly associated with gender, duration of diabetes, use of hypoglycaemic or lipid-lowering agents, C-peptide reactivity, BMI, and LDL and HDL cholesterol. Table S1 (see supplementary materials associated with this article online) describes the use of hypoglycaemic agents, the frequency of hypoglycaemia, and
3
compliance with medications at 4 to 5 years of follow-up. In fact, the frequency of severe hypoglycaemia during follow-up was rather low, and compliance with oral medications and insulin did not significantly differ among these three groups. The 8-year follow-up rate was 76.8%. There were no notable differences in baseline characteristics between patients who completed the 8-year follow-up and those who did not [10]. Longitudinal trends in mean HbA1c were unstable in the severe hypoglycaemia group, possibly due to our limited sample size (Fig. 1A). Levels of C-peptide reactivity in the non-severe and severe hypoglycaemia groups were lower at baseline than in the no hypoglycaemia group and decreased during the 8-year follow-up (Fig. 1B). Also, during the follow-up, 8492 person-years were studied and 329 incidents of DR were observed (283 with no hypoglycaemia, 38 with non-severe hypoglycaemia and eight with severe hypoglycaemia). Fig. 2 shows Kaplan–Meier curves for incident DR, and Table 2 compares the incidence of DR and Cox regression analysis of crude HRs with those after adjusting for potential confounders. Adjusted HRs were 0.90 (95% CI: 0.61–1.34; P = 0.61) for non-severe hypoglycaemia within the previous year and 4.35 (95% CI: 1.98–9.56; P < 0.01) for severe hypoglycaemia within the previous year. Table 3 shows the results of our three sensitivity analyses. The estimates of HRs ranged within the CIs of the analysis restricted to insulin users (Model 1) and the Cox regression analysis using different exposure definitions (Model 2). After excluding 141 patients who had albumin excretion rates 300 mg/g creatinine at baseline, or other clinical findings of renal disease, 33 major microvascular events (incidence of DR or overt nephropathy) were seen in 1080 patients (288 with no hypoglycaemia, 38 with mild hypoglycaemia and seven with severe hypoglycaemia). The adjusted HRs for major microvascular events were similar to those for DR (Model 3). In another cohort of 410 patients with prevalent DR, 73 (17.8%) had experienced non-severe hypoglycaemia within the previous year, while seven patients (1.7%) reported hypoglycaemia episodes with coma or requiring treatment (Table S2; see supplementary materials associated with this article online). In this cohort, the adjusted HRs for progression of DR were 1.08 (95% CI: 0.48–2.42;
Table 1 Background characteristics of 1221 patients with type 2 diabetes without diabetic retinopathy by history of hypoglycaemia during the previous year.
Any hypoglycaemia (frequency/week) Hypoglycaemia with coma (frequency/week) Hypoglycaemia requiring treatment (frequency/week) Age (years) Women (%) HbA1c (%, NGSP value) Duration of diabetes (years) Treated with oral hypoglycaemic agents without insulin (%) Treated with sulphonylureas without insulin (%) Treated with insulin (%) Treated with antihypertensive agents (%) Treated with lipid-lowering agents (%) C-peptide reactivity (ng/mL) Body mass index (kg/m2) Systolic blood pressure (mmHg) LDL cholesterol (mg/dL) HDL cholesterol (mg/dL) Triglyceridesa (mg/dL)
No hypoglycaemia (n = 1084)
Non-severe hypoglycaemia (n = 127)
Mean
Mean
0.00 0.00 0.00 58.3 43.7 8.2 9.5 64.9 58.1 7.8 25.3 24.8 2.2 23.2 131.3 124.3 53.5 107.0
SD 0.00 0.00 0.00 6.8 1.3 6.6
2.3 3.0 15.9 32.3 16.4 79.0
1.08 0.00 0.00 57.5 55.9 8.4 11.8 32.3 30.7 62.2 22.0 18.1 1.6 22.0 127.9 117.2 59.6 81.0
Severe hypoglycaemiab (n = 10) SD 1.65 0.00 0.00 7.5 1.3 8.2
2.2 3.2 17.3 30.5 16.9 54.0
Mean 1.08 0.90 0.80 58.7 50.0 8.5 12.3 20.0 20.0 80.0 20.0 60.0 1.6 22.5 128.7 102.5 61.2 126.0
P
SD 1.18 0.57 0.63 6.1 1.8 6.0
1.1 2.7 19.3 25.7 19.1 80.0
< 0.01 < 0.01 < 0.01 0.43 0.03 0.21 < 0.01 < 0.01 < 0.01 < 0.01 0.67 0.01 0.03 < 0.01 0.07 < 0.01 < 0.01 0.07
NGSP: National Glycohemoglobin Standardization Program; LDL/HDL: low-density/high-density lipoprotein. a Data are medians and interquartile range. b At least one episode with coma or requiring an outpatients visit or hospitalization.
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002
G Model
DIABET-902; No. of Pages 6 S. Tanaka et al. / Diabetes & Metabolism xxx (2017) xxx–xxx
4
Fig. 1. Longitudinal trends in mean HbA1c (left) and C-reactive protein (right) over 8 years of follow-up according to patients’ history of hypoglycaemia at baseline.
Fig. 2. Kaplan–Meier curves of incident diabetic retinopathy according to patients’ history of hypoglycaemia at baseline.
P = 0.86) for non-severe hypoglycaemia and 2.29 (95% CI: 0.45– 11.78; P = 0.32) for severe hypoglycaemia (Table 4). Discussion Our present study of Japanese patients with T2D has demonstrated a significant increase in the risk of DR among patients who had experienced severe hypoglycaemia, but no notable effects from mild hypoglycaemia episodes on outcomes. The striking difference between our findings and those of the ADVANCE study [11] was that, in our cohort, severe hypoglycaemia was one of the strongest predictors of DR, showing an approximately fourfold increase in incidence rate, which is numerically larger than the previously reported effects of HbA1c, duration of diabetes, systolic blood pressure and BMI [14]. Although the definitions of hypoglycaemia were not fully comparable, it was also notable that non-severe hypoglycaemia seemed to be less frequent in our study than in the ADVANCE study [11]. The proportions of subjects experiencing hypoglycaemia within the previous year in our study were 10.4% for non-severe
and 0.8% for severe hypoglycaemias, equivalent to annual cumulative incidences of 0.104 and 0.008 per person-years, respectively. When compared with the reported incidence rates in the ADVANCE study (1.1 and 0.006 per person-years, respectively) [11], the frequency of non-severe hypoglycaemia was rather low in the present study while the frequency of severe hypoglycaemia was comparable. In previous clinical studies, hypoglycaemia was surveyed through self-reported questionnaires, prospective reports or diaries, emergency-department admission records and claims databases [4]. Unlike our study, which used self-reported questionnaires, the ADVANCE study assessed hypoglycaemia prospectively, and their criteria for severe hypoglycaemia were broader, as it was defined as the inability of patients to treat themselves [11]. Differences between the ADVANCE and our study are therefore partly attributable to the use of self-reported questionnaires and different definitions of hypoglycaemia. It is not plausible, however, that the difference in the size of the effect would be a consequence of misclassification, as non-differential misclassification is expected to weaken associations. Another explanation is that the effects of hypoglycaemia were attenuated in the ADVANCE study because the study population included patients with prevalent microvascular complications, and the study outcome was a composite of new or worsening nephropathy or retinopathy. Indeed, the size of the effect on progression of DR was about half that of the incident DR. Our study also examined associations with a composite of microvascular events as a sensitivity analysis, and the size of the effect was still considerably larger than the findings observed in the ADVANCE study (Table 3). Other speculations to explain these contradictory findings include differences in the severity of diabetes, diabetes treatment and ethnicity. At present, the biological mechanisms behind the potential effects of hypoglycaemia on the microvasculature remain hypothetical, but one possible explanation is oxidative and inflammatory stress, which can be provoked by acute and chronic hypoglycaemia [7]. It is known that, in retinal cells, glucose deprivation leads to upregulation of vascular endothelial growth factor [18]. In fact, epidemiological studies of both type 1 diabetes (T1D) and T2D have revealed that markers of inflammation and endothelial dysfunction correlate with DR [19,20]. Unfortunately, data on inflammation and endothelial dysfunction were not available for our cohort. It is also possible that, instead of a direct causal link, severe hypoglycaemia is associated with the risk of DR through coexisting conditions that remain unmeasured or incompletely quantified. It may also be speculated that residual confounders are related to the severity and treatment of diabetes.
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002
G Model
DIABET-902; No. of Pages 6 S. Tanaka et al. / Diabetes & Metabolism xxx (2017) xxx–xxx
5
Table 2 Associations between history of hypoglycaemia during the previous year and incident diabetic retinopathy in 1221 patients with type 2 diabetes.
History of hypoglycaemia None Non-severe hypoglycaemia Severe hypoglycaemiaa Treated with insulin Age (+ 10 years) Gender (women) HbA1c (%, NGSP value) Duration of diabetes (+ 10 years) Body mass index (kg/m2) C-protein reactivity (ng/mL) Systolic blood pressure (+ 10 mmHg) LDL cholesterol (+ 10 mg/dL) HDL cholesterol (+ 10 mg/dL) Log-triglycerides (+ 1 unit)
Crude HR
95% CI
1 1.16 5.03
Reference 0.83 2.34
P
1.63 10.79
0.38 < 0.01
Adjusted HR
95% CI
Reference 0.89 4.30 1.41 0.94 1.11 1.30 1.61 1.04 1.01 1.10 1.02 1.00 1.04
0.60 1.96 1.01 0.80 0.88 1.22 1.38 1.00 0.96 1.02 0.98 0.93 0.83
P
1.32 9.44 1.97 1.11 1.39 1.39 1.87 1.08 1.06 1.18 1.05 1.07 1.31
0.56 < 0.01 0.047 0.50 0.39 < 0.01 < 0.01 0.04 0.69 0.01 0.30 0.96 0.71
Missing data for 37 patients were multiply imputed; multivariate model adjusted for insulin use, gender, age, HbA1c, duration of diabetes, body mass index, C-peptide reactivity, systolic blood pressure; HR: hazard ratio; CI: confidence interval; NGSP: National Glycohemoglobin Standardization Program; LDL/HDL: low-density/high-density lipoprotein. Multivariate model adjusted for insulin use, sex, age, HbA1c, duration of diabetes, body mass index, C-peptide reactivity, systolic blood pressure, LDL cholesterol, HDL cholesterol and triglycerides. a At least one episode of coma or requiring outpatients visit or hospitalization.
Table 3 Sensitivity analysis for associations between hypoglycaemia and incident diabetic retinopathy (DR) or major microvascular events in 1221 patients with type 2 diabetes.
Baseline modelb Non-severe hypoglycaemia Severe hypoglycaemiaa Model 1c Non-severe hypoglycaemia Severe hypoglycaemia Model 2d Non-severe hypoglycaemia Severe hypoglycaemia Model 3e Non-severe hypoglycaemia Severe hypoglycaemia
Adjusted hazard ratio
95% confidence interval
P
0.89 4.33
0.60 1.97
1.32 9.50
0.58 < 0.01
0.95 6.73
0.56 2.84
1.61 15.93
0.56 < 0.01
1.07 3.62
0.73 1.69
1.57 7.74
0.72 < 0.01
0.91 5.45
0.62 2.43
1.33 12.24
0.63 < 0.01
a
Episodes with coma or requiring an outpatients visit or hospitalization. Multivariate-adjusted models for 1221 patients, not adjusted for body mass index with missing data for 37 patients multiply imputed. c Multivariate-adjusted models for 171 insulin-treated patients, not adjusted for insulin use with missing data for 17 patients multiply imputed. d Multivariate-adjusted models for non-severe hypoglycaemia group were patients reporting non-severe hypoglycaemia at baseline and at 5 years with no severe hypoglycaemia episodes; the severe hypoglycaemia group included patients reporting severe hypoglycaemia at baseline and at 5 years, with missing data for 126 patients multiply imputed. e Multivariate-adjusted models for associations between hypoglycaemia and major microvascular events (diabetic retinopathy: 329 events; overt nephropathy: 73 events); 141 patients with an albumin excretion rate 300 mg/g creatinine at baseline or other clinical findings of renal disease were excluded; missing data for 82 patients were multiply imputed. b
Indeed, hypoglycaemia was frequent among users of sulphonylureas and insulin, and decreasing trends in C-protein reactivity were observed in both non-severe and severe hypoglycaemia groups. On the other hand, no notable differences were observed across groups in either glucose control during follow-up (Fig. 1A) or compliance with oral medications and insulin (Table S1; see supplementary materials associated with this article online). Furthermore, and somewhat surprisingly, the associations between hypoglycaemia and DR remained significant even after restricting the population analysis to insulin users (Table 3). The findings from our present analyses and the ADVANCE study markedly contradict observations regarding T1D. A recent re-analysis of the dataset from the Diabetes Control and Complications Trial
Table 4 Associations between history of hypoglycaemia during the previous year and progression of diabetic retinopathy (DR) in 410 patients with type 2 diabetes and prevalent DR.
Non-severe hypoglycaemia Severe hypoglycaemiaa Treated with insulin Age (+ 10 years) Gender (women) HbA1c (%, NGSP value) Duration of diabetes (+ 10 years) Body mass index (kg/m2) C-protein reactivity (ng/mL) Systolic blood pressure (+ 10 mmHg) LDL cholesterol (+ 10 mg/dL) HDL cholesterol (+ 10 mg/dL) Log-triglycerides (+ 1 unit)
Hazard ratio
95% confidence interval
P
1.03 2.61 0.76 0.72 1.50 1.69 1.10 1.02 1.03 1.12 0.92 0.86 0.80
0.45 0.52 0.39 0.49 0.88 1.41 0.73 0.94 0.97 0.96 0.85 0.71 0.44
0.95 0.25 0.42 0.09 0.14 < 0.01 0.64 0.60 0.39 0.15 0.03 0.10 0.46
2.33 13.21 1.49 1.05 2.58 2.02 1.66 1.12 1.09 1.31 0.99 1.03 1.44
Missing data for 51 patients were multiply imputed. NGSP: National Glycohemoglobin Standardization Program; LDL/HDL: low-density/high-density lipoprotein; NGSP: National Glycohemoglobin Standardization Program; LDL/HDL: low-density/ high-density lipoprotein. a Episodes with coma or requiring an outpatients visit or hospitalization.
(DCCT) [21] found that the frequency of severe hypoglycaemia during the DCCT did not predict the risk of DR over 9 years. Furthermore, same-day glucose variability was not associated with DR risk [22], whereas variability in HbA1c was a significant risk factor [23] in the DCCT. These observations are also not consistent with systematic reviews of glycaemic and HbA1c variability in T2D [24,25]. However, it should be noted that the definitions of exposure status in our study differed from those of the DCCT re-analysis. Although the criteria for severity were similar, our study focused on the effects of hypoglycaemia within the previous year, while hypoglycaemia during follow-up was accounted for only in the sensitivity analysis to avoid reverse causation. Our findings have both clinical and research implications. From the viewpoint of diabetes care, screening for DR is particularly recommended for patients experiencing severe hypoglycaemia. As suggested by some researchers, screening programmes could revise their screening intervals according to risk stratification [26]. The absolute risk of DR for a given patient can be calculated based on age, HbA1c and other risk factors [27], and safe screening intervals may be determined based on this calculated risk.
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002
G Model
DIABET-902; No. of Pages 6 S. Tanaka et al. / Diabetes & Metabolism xxx (2017) xxx–xxx
6
Limitations of our present study include the accuracy of hypoglycaemia assessment, which was self-reported and may have been subject to recall bias. The frequency of severe hypoglycaemia is rather high in this analysis. Our definition of ‘severe hypoglycaemia’ was at least one episode with coma or requiring an outpatients visit or hospitalization, thereby including a range of episodes from an outpatients visit to coma. In addition, only 10 patients without DR and seven patients with DR at baseline had experienced severe hypoglycaemia, limiting our statistical analysis in terms of details within this subpopulation and its effects on progression. Given these limitations, the generalizability of the present findings need to be confirmed by other cohorts with larger sample sizes and/or longer follow-ups, with detailed assessments of both hypoglycaemic events and DR. Moreover, data on the dosages of antidiabetic agents were not available, and the association between insulin doses and hypoglycaemia was unfortunately unknown. A further limitation was the diagnostic accuracy of DR based on clinical diagnosis compared with grading based on seven-field stereo fundus photography. Finally, our findings in Japanese patients may not be generally applicable to populations with different ethnic and genetic factors. In conclusion, severe hypoglycaemia is one of the strongest predictors of incident DR in patients with long-standing T2D and has an approximately fourfold greater risk. Also, tracking the number of episodes of hypoglycaemia may be helpful for personalizing diabetes care and for devising risk-based screening intervals for DR. Funding Funding/Support: This work was financially supported by the Ministry of Health, Labor and Welfare, Japan. Financial disclosures: No financial disclosures. Disclosure of interest The authors declare that they have no competing interest. Acknowledgements We sincerely thank the late Professor Nobuhiro Yamada, former director of the JDCS, who always provided warm spiritual support to us all. We thank all the patients and diabetologists at the 59 participating institutions for their long-standing collaboration in the JDCS. Thanks are extended to Ms Mami Haga and Ms Natsuko Tada, and Niigata University, for their excellent secretarial assistance. A medical editor in NAI Inc. reviewed the writing of the manuscript. The sponsor had no role in the design and conduct of the study.
Appendix A. Supplementary data Supplementary materials (Tables S1 and S2) associated with this article can be found at http://www.scincedirect.com at http:// dx.doi.org/10.1016/j.diabet.2017.06.002.
References [1] Action to Control Cardiovascular Risk in Diabetes Study Group. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med 2008;358:2545–59. [2] Boussageon R, Bejan-Angoulvant T, Saadatian-Elahi M, et al. Effect of intensive glucose lowering treatment on all-cause mortality, cardiovascular death, and microvascular events in type 2 diabetes: meta-analysis of randomised controlled trials. BMJ 2011;343:d4169.
[3] Salvo F, Moore N, Arnaud M, Robinson P, Raschi E, De Ponti F, et al. Addition of dipeptidyl peptidase-4 inhibitors to sulphonylureas and risk of hypoglycaemia: systematic review and meta-analysis. BMJ 2016;353:i2231. [4] Edridge CL, Dunkley AJ, Bodicoat DH, Rose TC, Gray LJ, Davies MJ, et al. Prevalence and incidence of hypoglycaemia in 532,542 people with type 2 diabetes on oral therapies and insulin: a systematic review and metaanalysis of population based studies. PLoS One 2015;10(6):e0126427. [5] Budnitz DS, Lovegrove MC, Shehab N, Richards CL. Emergency hospitalizations for adverse drug events in older Americans. N Engl J Med 2011;365:2002–12. [6] Bonds DE, Miller ME, Bergenstal RM, Buse JB, Byington RP, Cutler JA, et al. The association between symptomatic, severe hypoglycaemia and mortality in type 2 diabetes: retrospective epidemiological analysis of the ACCORD study. BMJ 2010;340:b4909. [7] Goto A, Arah OA, Goto M, Terauchi Y, Noda M. Severe hypoglycaemia and cardiovascular disease: systematic review and meta-analysis with bias analysis. BMJ 2013;347:f4533. [8] The Kroc Collaborative Study Group. Diabetic retinopathy after two years of intensified insulin treatment. Follow-up of the Kroc Collaborative Study. JAMA 1988;260:37–41. [9] Wright RJ, Frier BM. Vascular disease and diabetes: is hypoglycaemia an aggravating factor? Diabetes Metab Res Rev 2008;24:353–63. [10] Wild D, von Maltzahn R, Brohan E, Christensen T, Clauson P, Gonder-Frederick L. A critical review of the literature on fear of hypoglycaemia in diabetes: implications for diabetes management and patient education. Patient Educ Couns 2007;68:10–5. [11] Zoungas S, Patel A, Chalmers J, de Galan BE, Li Q, Billot L, et al. Severe hypoglycaemia and risks of vascular events and death. N Engl J Med 2010;363:1410–8. [12] Zhao Y, Campbell CR, Fonseca V, Shi L. Impact of hypoglycaemia associated with antihyperglycemic medications on vascular risks in veterans with type 2 diabetes. Diabetes Care 2012;35:1126–32. [13] O’Connor PJ, Ismail-Beigi F. Near-Normalization of glucose and microvascular diabetes complications: data from ACCORD and ADVANCE. Ther Adv Endocrinol Metab 2011;2:17–26. [14] Kawasaki R, Tanaka S, Tanaka S, Yamamoto T, Sone H, Ohashi Y, et al. Incidence and progression of diabetic retinopathy in Japanese adults with type 2 diabetes: 8-year follow-up study of the Japan Diabetes Complications Study (JDCS). Diabetologia 2011;54:2288–94. [15] Tanaka S, Tanaka S, Iimuro S, Yamashita H, Katayama S, Ohashi Y, et al. Cohort profile: the Japan diabetes complications study: a long-term follow-up of a randomised lifestyle intervention study of type 2 diabetes. Int J Epidemiol 2014;43:1054–62. [16] Sone H, Tanaka S, Iimuro S, Tanaka S, Oida K, Yamasaki Y, et al. Long-term lifestyle intervention lowers the incidence of stroke in Japanese patients with type 2 diabetes: a nationwide multicentre randomised controlled trial (the Japan Diabetes Complications Study). Diabetologia 2010;53:419–28. [17] Wilkinson CP, Ferris 3rd FL, Klein RE, Lee PP, Agardh CD, Davis M, et al. Proposed international clinical diabetic retinopathy and diabetic macular oedema disease severity scales. Ophthalmology 2003;110:1677–82. [18] Sone H, Kawakami Y, Okuda Y, Kondo S, Hanatani M, Suzuki H, et al. Vascular endothelial growth factor is induced by long-term high glucose concentration and up-regulated by acute glucose deprivation in cultured bovine retinal pigmented epithelial cells. Biochem Biophys Res Commun 1996;221:193–8. [19] Schram MT, Chaturvedi N, Schalkwijk CG, Fuller JH, Stehouwer CD, EURODIAB Prospective Complications Study Group. Markers of inflammation are crosssectionally associated with microvascular complications and cardiovascular disease in type 1 diabetes — the EURODIAB Prospective Complications Study. Diabetologia 2005;48:370–8. [20] van Hecke MV, Dekker JM, Nijpels G, Moll AC, Heine RJ, Bouter LM, et al. Inflammation and endothelial dysfunction are associated with retinopathy: the Hoorn Study. Diabetologia 2005;48:1300–6. [21] Kilpatrick ES, Rigby AS, Atkin SL, Frier BM. Does severe hypoglycaemia influence microvascular complications in type 1 diabetes? An analysis of the diabetes control and complications trial database. Diabet Med 2012;29:1195–8. [22] Kilpatrick ES, Rigby AS, Atkin SL. Effect of glucose variability on the long-term risk of microvascular complications in type 1 diabetes. Diabetes Care 2009;32:1901–3. [23] Kilpatrick ES, Rigby AS, Atkin SL. A1C variability and the risk of microvascular complications in type 1 diabetes: data from the Diabetes Control and Complications Trial. Diabetes Care 2008;31:2198–202. [24] Smith-Palmer J, Bra¨ndle M, Trevisan R, Orsini Federici M, Liabat S, Valentine W. Assessment of the association between glycaemic variability and diabetesrelated complications in type 1 and type 2 diabetes. Diabetes Res Clin Pract 2014;105:273–84. [25] Gorst C, Kwok CS, Aslam S, Buchan I, Kontopantelis E, Myint PK, et al. Longterm glycaemic variability and risk of adverse outcomes: a systematic review and meta-analysis. Diabetes Care 2015;38:2354–69. [26] Thomas RL, Dunstan F, Luzio SD, Roy Chowdury S, Hale SL, North RV, et al. Incidence of diabetic retinopathy in people with type 2 diabetes mellitus attending the Diabetic Retinopathy Screening Service for Wales: retrospective analysis. BMJ 2012;344:e874. [27] Tanaka S, Tanaka S, Iimuro S, Yamashita H, Katayama S, Akanuma Y, et al. Predicting macro- and microvascular complications in type 2 diabetes: the Japan Diabetes Complications Study/the Japanese Elderly Diabetes Intervention Trial risk engine. Diabetes Care 2013;36:1193–9.
Please cite this article in press as: Tanaka S, et al. Severe hypoglycaemia is a major predictor of incident diabetic retinopathy in Japanese patients with type 2 diabetes. Diabetes Metab (2017), http://dx.doi.org/10.1016/j.diabet.2017.06.002