Variability in HbA1c and the risk of major clinical outcomes in type 2 diabetes with chronic kidney disease: Post hoc analysis from the CREDENCE trial

Glycated hemoglobin (HbA1c) levels, a measure of average glucose levels over approximately 3 months, are widely accepted clinical indicators of glycaemic control and are recommended for ongoing monitoring in accordance with established guidelines. HbA1c variability refers to the fluctuations in a person's HbA1c levels over longer periods of time and reflects changes in glycaemic control over longer periods, for example, months to years. Recent studies have highlighted the importance of HbA1c variability as an independent predictor of microvascular and macrovascular complications in Type 1 and Type 2 diabetes (T1D and T2D). For instance, a meta-analysis by Shen et al. demonstrated a significant association between HbA1c variability and increased cardiovascular risk in T2D patients.1 A study following 1240 Type 1 diabetics for over 20 years linked higher HbA1c variability to increased microvascular complications.2 Studies have suggested that higher HbA1c variability is associated with increased microvascular complications, such as retinopathy, microalbuminuria,3, 4 cardiovascular events and mortality,1, 5 in people with T1D and with a specifically higher mortality risks6 in people with Type 2 diabetes (T2D). Furthermore, recent guidelines from the American Diabetes Association (ADA) emphasize the need to evaluate HbA1c variability in high-risk populations, such as those with chronic kidney disease (CKD).7 However, the prospective association between HbA1c variability and significant clinical outcomes, such as cardiovascular and kidney events, in patients with T2D and CKD remains underexplored. This study aims to quantify HbA1c variability in T2D patients with CKD and examine its association with major clinical outcomes, leveraging data from the CREDENCE trial. Understanding HbA1c variability dynamics could enhance patient management protocols, identify patients at high risk and mitigate the progression of diabetic complications. This study aims to quantify HbA1C variability in T2D patients with CKD and examine the association between HbA1c variability and significant clinical outcomes. Detailed methods and main findings from the study have been previously published.8 In brief, CREDENCE (NCT02065791) was a double-blind, randomized trial evaluating the effects of the Sodium-Glucose Cotransporter 2 (SGLT2) inhibitor canagliflozin on renal outcomes in 4401 patients aged ≥30 years with T2D, with a glycated hemoglobin level between 6.5 % to 12.0% (6.5% to 10.5% in Germany), and albuminuric CKD conducted across 34 countries with all patients providing written consent. CKD was defined as an estimated glomerular filtration rate (eGFR, using the CKD Epidemiology Collaboration formula) of 30 to <90 mL per minute per 1.73 m2 of body-surface area and albuminuria (urinary albumin-to-creatinine ratio (UACR), >300 to 5000 mg/g), measured centrally. Variability in HbA1c was assessed using the coefficient of variation (CVHbA1c). Participants were grouped by thirds of CVHbA1c (presented as a percentage) defined as low variability (<6.27%; reference group), moderate variability (≥6.27% to <10.00%) and high variability (≥10.00%). This analysis examined major cardiovascular events (myocardial infarction, heart failure hospitalization, unstable angina, cardiovascular death) and kidney events (serum creatinine doubling to >200 μmol/L, end-stage kidney disease (continuous eGFR <15 mL/min/1.73 m2, maintenance dialysis) and kidney death), stroke and all-cause mortality. Visit-to-visit HbA1c across the first 3, 6 and 12 months of the study was the exposure period (Figure S1). Participants with <3 HbA1c measurements, study outcomes in year one, the initiation of dialysis/transplantation and the occurrence of diabetic ketoacidosis, anemia or hemoglobinopathy were excluded (Figure S2). Log-linear trends across CVHbA1c categories at baseline were tested by linear regression or logistic regression analysis, as appropriate. The proportion of study outcomes across the three groups was assessed using a chi-square test. Cox regression models were used to estimate hazard ratios (HRs) and corresponding 95% confidence intervals (CIs) for HbA1c variability, adjusting for baseline participant demographic and medical characteristics (Table S1). Subgroups were defined by HbA1c <7.5% vs. ≥7.5%, adjusting for covariates using Cox regression models. In sensitivity analysis, we used patient baseline characteristics 12 months after randomization, that is, at the start of this study. All analyses were performed with R version 4.4.1. The authors declare that all supporting data are available within the article (and its online supplementary files). A two-sided p-value less than 0.05 was considered statistically significant. Of the 4401 CREDENCE participants, 3080 (69.98%) were eligible for inclusion in the current study (Figure S2). The mean participant age was 63.3 years; 1016 (33.0%) were female (Table 1). Across increasing tertiles of HbA1c variability, participants were more likely to be younger, have a shorter duration of diabetes and have higher levels of albuminuria. Other characteristics were broadly similar across the three groups (Table 1). During a median follow-up of 2.62 years (interquartile range 25th–75th percentile: 2.0 to 3.2), 182 (5.9%) participants experienced a major cardiovascular event, 112 (3.6%) experienced kidney events, 35 (1.1%) experienced a stroke and 102 (3.3%) died of any cause. Increasing variability was associated with higher numbers of kidney events (Table S2). In unadjusted analysis, greater HbA1c variability was independently associated with a higher risk of kidney events (HR for moderate/high vs. low variability: 1.58 95% CI: 0.88–2.82 and 3.51, 95% CI:2.08–5.94, respectively) (Figure 1A), with no statistically significant associations between greater HbA1c variability and the risk of major macrovascular events, stroke or all-cause mortality. In adjusted models, greater HbA1c variability was independently associated with a higher risk of major macrovascular events (HR for moderate/high vs. low variability: 1.15 95% CI: 0.79–1.67 and 1.47, 95% CI:1.02–2.13, respectively) and kidney events (HR for moderate/high vs. low variability: 1.53, 95% CI: 0.84–2.79 and 2.75, 95% CI:1.57–4.83, respectively) (Figure 1B). In subgroup analyses, using adjusted cox-regression models, the group with greater mean HbA1c within the high variability group had a lower risk of kidney events (HR for HbA1c ≥ 7.5% compared with HbA1c < 7.5%: 0.43 95% CI: 0.24–0.76) with no statistically significant association between greater HbA1c and the risk of major macrovascular events, stroke or all-cause mortality (Table S3). Sensitivity analyses confirmed the primary findings (Table S4). In this analysis of 3080 patients with Type 2 Diabetes and chronic kidney disease, higher HbA1c variability significantly predicted increased major cardiovascular and kidney events. These findings align with previous studies that have linked HbA1c variability to adverse outcomes in diabetic patients. For example, a recent meta-analysis by Xu et al. demonstrated that long-term HbA1c variability is associated with an increased risk of kidney-related outcomes in patients with diabetes.9 Our results further corroborate these findings, suggesting that HbA1c variability is a key prognostic marker in T2D patients with CKD. Contrary to expectations from prior cohorts where older age was associated with greater HbA1c variability,10 our study observed younger age in the high variability group. This discrepancy may reflect differences in study populations, as CREDENCE enrolled participants with advanced CKD, who may exhibit distinct glycemic patterns due to comorbidities or therapeutic interventions (e.g., limited use of SGLT2 inhibitors). Additionally, younger individuals with shorter diabetes duration might have less stable glycemic control due to evolving treatment regimens or lifestyle factors. These findings highlight the need for age-stratified analyses in future studies to clarify the interplay between age, glycemic variability and CKD progression. We observed modest differences in diastolic blood pressure (DBP) across HbA1c variability tertiles. While DBP could theoretically confound the association between glycemic variability and outcomes, our adjusted models accounted for baseline blood pressure, minimizing residual confounding.11 Nevertheless, hemodynamic fluctuations linked to glycemic variability might indirectly influence renal and cardiovascular risk, warranting further mechanistic investigations. However, our study extends these findings by highlighting the role of HbA1c variability as an independent risk factor. A recent meta-analysis by Li et al. further supports that HbA1c variability independently predicts cardiovascular events even after adjusting for mean HbA1c.12 Unlike previous studies that focused primarily on mean HbA1c levels,10 our analysis suggests that HbA1c variability may provide additional prognostic information, particularly in patients with CKD. The clinical implications of our findings are significant. HbA1c variability could be a useful marker for identifying high-risk patients who may benefit from more intensive glycemic monitoring and management. Current KDIGO guidelines recommend targeting HbA1c <7.0% in CKD patients but do not address variability.13 Our results suggest incorporating HbA1c variability into clinical algorithms could enhance risk stratification. Furthermore, interventions aimed at reducing HbA1c variability, such as using SGLT2 inhibitors or GLP-1 receptor agonists, may improve outcomes in T2D patients with CKD. For instance, the EMPA-REG OUTCOME trial demonstrated that empagliflozin reduces glycemic variability and cardiovascular risk, aligning with our findings.14 Several limitations of our study should be acknowledged. First, as a post-hoc analysis, our findings may be subject to selection bias. Second, the median follow-up period of 2.62 years may be insufficient to assess long-term outcomes. Third, variations in HbA1c measurement methods across different centers (e.g., differences in detection kits) could introduce measurement bias. Finally, our study population was derived from a clinical trial setting, which may limit generalizability to routine clinical practice. Higher HbA1c variability predicts worse cardiovascular and renal outcomes in T2D patients with CKD, independent of mean HbA1c levels. These findings underscore the importance of monitoring glycemic variability as a complementary strategy to traditional HbA1c targets. Future studies should explore interventions targeting HbA1c variability and validate these findings in real-world populations. YZ, SK, MJ, RF, CA and BN contributed to the concept and rationale for the study and interpretation of the results. YZ conducted a statistical analysis and drafted the manuscript with advice from SK. All authors contributed to the discussion and reviewed and edited the manuscript. YZ and SK are the guarantors of this work and, as such, had full access to all the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis. Open access publishing facilitated by University of New South Wales, as part of the Wiley - University of New South Wales agreement via the Council of Australian University Librarians. The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/dom.16363. Access to these data, used with CREDENCE steering committee approval, is restricted and they are not publicly available. Data S1. Supporting Figures. Data S2. Supporting Tables. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.