UPDATES FOR GLUCOSE TARGETS

 Observational data consistently report that the association between death and hyperglycemia is markedly affected by preexisting glucose control . Moreover, the risk of treatment-induced hypoglycemia, a strong predictor of increased mortality, is greatest in those with preexisting hyperglycemia .

 A recent single-center retrospective cohort study of 3,625 patients with at least four blood glucose measurements and a HbA1c measurement reported that for patients with a HbA1c  8.0% mortality was lower if the proportion of time spent with a blood glucose between 180 and 250 mg/dL (10 and 14 mmol/L) was greater . Such data support the hypothesis that glucose concentrations that are regarded as safe and desirable in those without diabetes (108–180 mg/dL) might actually be undesirable and harmful in patients with preexisting glucose intolerance, which substantiates the concept that glucose targets should be ‘personalized’ based on preexisting glycemia .

 The prospective studies evaluating a more liberal approach to glucose targets in patients with diabetes are limited to single-center studies that each have substantial methodological weaknesses . Nonetheless, the signal from these studies is that allowing modest hyperglycemia before commencing insulin and targeting blood glucose concentrations between 180 and 250 mg/dL (10 and 14 mmol/ L) appears well-tolerated with a tendency to reduce episodes of hypoglycemia . Recently an important multicenter trial, the CONTROLLING trial, attempted to address the question: Does a ‘personalized’ approach to glycemic targets, which is based on a glycated hemoglobin on admission, reduce 90-day all-cause mortality? Within a parallel group design, critically ill adult patients were randomized (1:1) to a ‘personalized’ or ‘conventional’ approach to blood glucose targets . 

Critically ill patients who were expected to remain in the ICU for 48 h and were not able to ingest food had their HbA1c measured within 96 h of ICU admission. For patients allocated to personalized glucose control the HbA1c determined the blood glucose target and for patients allocated to conventional control the blood glucose target was 180 mg/dL (10 mmol/L) or less. Unlike the studies mentioned above that only include patients with preexisting diabetes, two-third of patients in this study did not have diabetes. This resulted in a large proportion of patients within the CONTROLLING trial having blood glucose targets that were similar to the ‘tight glucose control’ (81–108 mg/dL or 4.5– 6.0 mmol/L) that was demonstrated to be harmful in NICE-SUGAR [34]. The independent data safety monitoring board (DSMB) recommended that the CONTROLLING trial be ceased at the first interim analysis due to the high incidence of hypoglycemia in participants and the low likelihood of the alternative hypothesis being proven true. 

PAPER LINK

It should be noted that there was no preexisting DSMB charter or stopping rules. The trial was ceased after 2,075 patients with data included for 1,917 participants. Death occurred in 308 of 938 (32.8%) patients in the personalized control and 295 of 968 (30.5%) patients in the convention control group . Moderate hypoglycemia (<72 mg/dL or <4.0 mmol/L) was more frequent with personalized control (294 of 942 (31%) vs. 154 of 975 (16%); P < 0.01) but there was no statistically significant difference for incidence of severe hypoglycaemia (<40 mg/dL or <2.2 mmol/L; 37 of 942 (3.9%) vs. 24 of 975 (2.5%); P ¼ 0.09). A post hoc analysis of patients without diabetes revealed a signal toward increased 90-day mortality with personalized glucose targets (Hazard Ratio: 1.3, 95% CI 1.05–1.59). The CONTROLLING trial results complement the NICE-SUGAR trial results to provide greater certainty that ‘tight blood glucose’ control, or targeting a blood glucose  110 mg/dL (6.1 mmol/L), is harmful regardless of whether the patient previously had diabetes, and should be avoided. However, whether a strategy to ‘personalize’ targets and allow moderate hyperglycemia (> 180 mg/dL or 10 mmol/L) to reduce absolute and/or relative hypoglycemia in those patients with preexisting diabetes was not addressed in the CONTROLLING trial. Future trials, such as LUCID, will help to inform decision-making in this cohort .