Transition versus Nontransition to Dialysis and Glycemic Status and Mortality

Because most previous trials of glycemic control excluded patients with advanced kidney disease, the optimal glycemic target in patients with chronic kidney disease (CKD) remains unclear. Earlier studies have shown microvascular and macrovascular benefits of intensive glycemic control in patients with both type 1 and type 2 diabetes with minimal to no kidney damage. However, more recent trials have shown lack of benefit and increased mortality in populations with long-term type 2 diabetes and cardiovascular risk, a population similar to patients with advanced CKD, challenging the safety of lower glycemic targets.

Kidney Disease Improving Global Outcomes (KDIGO) Quality Initiative and KDIGO guidelines suggest a target hemoglobin A1c (HbA1c) of 7% to prevent or reduce progression of microvascular complications. The guidelines also advise HbA1c levels >7% in patients with comorbidities, limited life expectancy, and increased risk of hypoglycemia.

Connie M. Rhee, MD, MSc, and colleagues recently demonstrated that among US veterans with diabetes and advanced CKD progressing to end-stage renal disease (ESRD), there were associations between higher averaged random glucose and HbA1c levels measured in the pre-ESRD period and higher morality risk post-ESRD. Noting that those findings may not be generalizable to patients with CKD who do not transition to dialysis therapy, the researchers reexamined the association of pre-ESRD glycemic status, defined by averaged random glucose and HbA1c levels, with post-ESRD mortality in a national cohort of veterans with CKD and diabetes who were transitioning to dialysis. They then compared  the inter-relationships between glycemic status and survival among a matched cohort of patients with CKD who did not transition to dialysis. Results were reported in the Journal of Renal Nutrition [2019;29(2):82-90].

The study utilized longitudinal data from the Transition of Care in CKD study, a retrospective study that examined transition to dialysis in a cohort of US veterans with ESRD. The source population consisted of 52,172 patients from the national Veterans Affairs database who transitioned to dialysis from October 1, 2007, to September 30, 2011 (transition cohort). Parallel analyses were conducted among a matched sample of adult patients with CKD from the RCAV (Racial and Cardiovascular Risk Anomalies in CKD) cohort of US veterans who received care within the VA healthcare system from 2004 to 2013 (nontransition cohort).

The primary outcome of interest in the transition cohort was 1-year post-ESRD all-cause mortality. Patients were censored for kidney transplantation, loss to follow-up, end of the study period (1-year after commencement of at-risk time), or last date of available follow-up data, whichever occurred first. In parallel analyses of the nontransition cohort, the researchers examined associations of the 1-year averaged random glucose and HbA1c with 1-year all-cause mortality, with at-risk time beginning the day after the 1-year averaged random glucose and HbA1c exposure periods.

Of the 17,121 patients in the transition cohort eligible for the primary analyses of averaged random glucose, mean age was 69 years, 28% were black, and 8% were Hispanic. Most patients had moderate to advanced CKD: 0.9% were categorized as stage 1, 4% as stage 2, 19% as stage 3, and 77% as stage 4+5 CKD. They were matched to 8711 patients in the nontransition cohort who had a similar balance of baseline characteristics.

In the transition cohort, those in the lowest averaged random glucose category (<80 mg/dL) tended to be younger, were more likely to have diabetes, less likely to have hypertension as the underlying cause of ESRD, more likely to be living in the northeast and less likely to be living in the west, had higher Charlson Comorbidity Index scores, and higher 1-year average eGFR and lower serum albumin levels.

In secondary analyses of HbA1c, 17,819 participants in the transition cohort met eligibility criteria. Mean age was 68 years, 28% were black, and 7% were Hispanic. Most had mild to moderate CKD: 1% were categorized as stage 1, 4% as stage 2, 19% as stage 3, and 76% as stage 4+5 CKD. They were matched with 10,848 patients in the nontransition cohort who had a similar balance in baseline characteristics.

Patients in the transition cohort contributed a total of 13,207 patient-years of follow-up. There were 3850 deaths during follow-up. Results of expanded case-mix analyses revealed a J-shaped association between averaged random glucose and mortality. The highest category (≥200 mg/dL) was associated with higher death risk (reference:100-<120 mg/dL): adjusted hazard ratio (aHR), 1.26; 95% confidence interval (CI), 1.13-1.40.

Among the matched participants in the nontransition cohort who underwent ≥1 random glucose measurement, there were 1092 deaths over 6699 patient-years of follow-up. There was a U-shaped association between averaged random glucose and mortality: lower levels (<100 mg/dL) and higher levels (≥160 mg/dL) were associated with higher risk of death: aHRs, 1.70 (95% CI, 1.18-2.44), 1.34 (95% CI, 1.07-1.69), 1.15 (95% CI, 0.94-1.41), 1.12 (95% CI, 0.90-1.41), 1.55 (95% CI, 1.10-1.83), 1.42 (95% CI, 1.10-1.83), and 1.34 (95% CI, 1.08-.165) for glucose categories <80, 80 to <100, 120 to <140, 140 to <160, 160 to <180, 180 to <200, and ≥200 mg/dl, respectively.

In the secondary analyses, patients in the transition accounted for 15,119 patient-years of follow-up; during that time, there were 4374 deaths. Results of expanded case-mix analyses found a J-shaped association between HbA1c and mortality: higher categories of 8 to <10 and ≥10% were each associated with higher risk of death (reference: 6 to <8%): aHR, 1.21 (95% CI, 1.11-1.33) and aHR, 1.43 (95% CI, 1.21-1.69), respectively.

In the matched sample of patients from the nontransition cohort who underwent >1 HbA1c measurement, there were 1130 deaths over 5402 patient-years. There was no significant J-shaped association between HbA1c levels and mortality. In expanded case-mix adjusted splines, there was a trend toward HbA1c levels <5% and higher mortality, and toward HbA1c levels above ~8% and greater survival; however, the prevalence of patients in those outlier HbA1c ranges was low.

The researchers cited some limitations to the study, including defining glycemic status by random glucose levels measured in the clinical setting, possible confounding of HbA1c measurements, and the inability to prove a causal association between glycemic status and mortality risk due to the observational design of the study.

The researchers said, “Our study is the first to show a differential relationship between glycemic status and survival among diabetic CKD patients who did versus did not transition to ESRD after 1 year of follow-up. Our observations of a J-shaped association between glycemic status and post-ESRD mortality suggest liberal glycemic status is associated with long-term mortality risk, whereas the U-shaped glucose–mortality relationship among patients who did not transition indicate intensive glycemic status is associated with short-term risk. Further studies are needed to define individualized optimal glycemic targets among diabetic CKD patients according to their underlying longevity and health status.”

Takeaway Points

1. Researchers conducted an analysis of the association of glycemic status prior to end-stage renal disease (ESRD) with post-ESRD mortality in patients who transitioned to dialysis. They also compared the relationships between glycemic status and survival in a matched cohort of patients with CKD who did not transition to dialysis.
2. In the transition cohort, there was an association between averaged random glucose ≥200 mg/dl and higher mortality risk. There was also an association between HbA1c 8-<10% and 10% and higher mortality.
3. In the nontransition cohort, averaged random glucose <100 mg/dL and ≥160 mg/dL were associated with higher risk of death. There was no association between HbA1c and mortality.