Effects of Intensive Glycemic Control on Lower Extremity Amputations
Christopher J Clark, Matthew P. Goldman, Timothy E. Craven, Gabriela Velazquez, Ross P. Davis, Randolph L. Geary, Justin Hurie, Timothy K. Williams, Matthew S. Edwards
Wake Forest University School of Medicine, Winston Salem, NC
Background: Diabetes mellitus is a major risk factor for peripheral arterial disease and subsequent progression to critical limb ischemia and lower extremity amputation. In this study we evaluated the effects of intensive blood glucose control on risk of lower extremity amputation (LEA) in type 2 diabetics during a randomized-controlled multicenter trial and extended post-trial follow-up.
Methods: The Action to Control Cardiovascular Risk in Diabetes (ACCORD) trial randomized 10251 type 2 diabetics to intensive glycemic control (IGC; HbA1c target <6.0%) versus standard glycemic control (SGC; HbA1c target 7.0 - 7.9%). Of those randomized, 9985 participants (4984 IGC, 5001 SGC) had at least one follow-up with surveillance for LEA. The primary outcome of the ACCORD trial was a composite of non-fatal MI, non-fatal stroke, or cardiovascular death. The glycemic management trial was discontinued by the sponsor after 3.7 years due to an observed increased mortality risk in the IGC group; however, all living participants were invited to participate in a long-term post-trial follow-up that included surveillance for diabetes-related complications. Using both an intention-to-treat analysis of randomized treatment assignment and an analysis of post-randomization on-treatment mean HbA1c, we examined relationships between glycemic control and incident/recurrent LEA during the clinical trial and subsequent follow-up.
Results: At time of randomization, mean age was 62±7 years, mean HbA1c was 8.3±1.1%, and average duration of diabetes was 10.8±7.7 years. A total of 2.6% of participants had a prior history of LEA and 3.8% had a prior history of lower extremity revascularization. Mean post-randomization HbA1c over the course of the trial and post- trial follow-up was 7.3±0.9% (6.8±0.8% in the IGC arm, 7.7±0.7% in the SGC arm). Baseline demographics and traditional risk factors were examined by treatment assignment; small but statistically significant differences were observed for number of blood pressure medications and patient reported history of any neuropathy. Patients were followed for a mean of 7.9±3.1 (median 8.9) years after randomization. A total of 124 participants had at least one LEA over the course of the trial and post-trial follow- up, 73 randomized to the SGC arm and 51 to the IGC arm (p = 0.049). A Cox proportional-hazard analysis found that randomization to IGC was associated with decreased rate of LEA (HR 0.69, 95% CI 0.483 - 0.987, p = 0.042). This association remained significant after controlling for death as a competing risk with a 31% reduction in risk of LEA. Figure 1 shows Kaplan-Meier estimates of risk of LEA by treatment arm assignment. In a Cox model that included both post-randomization mean HbA1c and glycemic control arm, mean HbA1c was a powerful predictor of LEA (HR 1.84 per 1% increase in mean HbA1c, 95% CI 1.51-2.23, p < 0.0001) while treatment arm assignment was no longer associated with the outcome. Model-predicted risk of LEA for mean post-randomization HbA1c of 8.5%, 7.5% and 6.5% are shown in Figure 2. Post- randomization mean HbA1c remained a strong predictor of LEA after controlling for other important covariates (diabetes duration, gender, smoking status, and histories of prior amputation or revascularization) and competing risk of death (HR 1.78 per 1% increase in HbA1c, 95% CI 1.40-2.27, p < 0.0001).
Figure 1. Kaplan-Meier estimates of follow-up LEA risk by treatment arm assignment. Numbers at risk are shown for each time point.
Figure 2. Predicted risk of LEA from Cox model for patients with mean post- randomization HbA1c values of 8.5%, 7.5% and 6.5%. Total numbers at risk are shown for each time point.
Conclusions: In older patients with type 2 diabetes mellitus, intensive glycemic control was associated with a significant reduction in the long-term risk for lower extremity amputation. Our analysis shows that after only 3.7 years of intensive glycemic control there was an enduring protective effect against subsequent amputation. Furthermore, examination of mean post-randomization HbA1c levels showed that improved glycemic control, even outside the range of IGC, was a strong predictor of decreased risk for subsequent LE amputation. This study suggests that tight glycemic control, even over a relatively short time period, has potential to profoundly reduce risk of limb loss.
Back to 2018 Abstracts