The impact of software-guided tight glucose control on outcome of critical illness, in relation to nutrition and circulating insulin

Korte inhoud:In the last decades, significant advancements have been made in critical care medicine, allowing most patients nowadays to survive the acute critical condition. Despite this, still a considerable number of patients do not recover swiftly, remain ICU-dependent for a prolonged period of time and will even develop a lifelong legacy hallmarked by persistent physical, cognitive, and psychiatric disorders. Hence, current research increasingly focusses on interventions that may also prevent or minimize those long-term disabilities, apart from reducing acute morbidity and mortality. Two interventions, tight blood glucose control and withholding early parenteral nutrition, both applied in the acute phase of critical illness, have extensively been studied in the last two decades and have demonstrated to fulfill these criteria. Remaining questions and controversies regarding both interventions were addressed in this thesis. Stress-induced hyperglycemia in critically ill patients was traditionally regarded as an adaptive, beneficial response, to provide energy to cells with insulin-independent glucose uptake in times of scarce food intake. Over twenty years, the research group of Prof. G. Van den Berghe proved in 3 consecutive landmark RCTs that hyperglycemia was harmful for critically ill patients, since lowering blood glucose concentrations to the normal age-adjusted fasting range (80-110 mg/dL for adults) with insulin reduced morbidity and mortality, as compared to tolerating severe hyperglycemia (>215 mg/dl). Despite the positive impact of TGC in the Leuven studies, TGC remained highly debated, especially since the largest multicenter RCT (NICE-SUGAR) found an increased mortality risk, subsequently attributed to an increased incidence of hypoglycemia. Whereas the Leuven RCTS were criticized for the early administration of parenteral nutrition, although in line with the then applicable European nutrition guidelines, the NICE-SUGAR RCT showed some methodological weaknesses and together those differences may explain the divergent effect on outcome. In the Leuven RCTs, patients received early parenteral nutrition when the caloric target was not reached with enteral nutrition alone. Subsequently, 2 multicenter RCTs (EPaNIC and PEPaNIC) demonstrated that the administration of early parenteral nutrition in critically ill patients impaired clinical outcome by slowing down recovery and increasing the incidence of new infections. Therefore, recent nutrition guidelines no longer recommend early full feeding. However, the current feeding practice of withholding early PN lowers the severity of hyperglycemia and the insulin need and increases the risk of hypoglycemia while providing TGC. Since withholding early parenteral nutrition was the standard of care in the NICE-SUGAR RCT, this could have contributed to the high incidence of hypoglycemia when providing TGC. Hence, the efficacy and safety of tight glucose control in the absence of early parenteral nutrition remained unclear. Although patients in the NICE-SUGAR RCT did not receive early parenteral nutrition, the RCT has been criticized for the allowance of inaccurate glucose measurement devices for glucose sampling and the use of insulin boluses. These methodological aspects may explain the high risk of (detected and undetected) hypoglycemia in the NICE-SUGAR RCT, which may have driven the harmful impact of TGC in this study. Hence, it had remained unclear whether TGC, in the context of strictly avoiding hypoglycemia with use of accurate monitoring and a performant insulin-titration algorithm, and when omitting early PN is beneficial or not. To address this controversy about TGC, a multicenter RCT that uses accurate blood glucose monitoring and a performant algorithm to guide insulin titration for TGC in patients who do not receive early PN, is needed. In this context, based on the expertise of the ICU nurses in Leuven, the computerized LOGIC-Insulin algorithm has been developed, which provides advice on the insulin dose (or glucose bolus in case of hypoglycemia) and the timing of blood glucose measurements. Using the LOGIC-Insulin-algorithm for providing TGC demonstrated to improve the quality and safety of TGC in Leuven as compared to TGC performed by experienced nurses. However, the algorithm had not been validated in a multicenter context. Moreover, apart from its effect on glucose control, withholding early PN in the EPaNIC RCT accentuated the inflammatory response, as suggested by a higher peak CRP, which could have been mediated by lower circulating insulin concentrations. However, it remained unclear whether the higher CRP rise by withholding early PN was mediated by any impact of the intervention on cytokines or by non-cytokine mediated effects of the intervention, and whether the CRP effect related to the outcome benefit of the intervention. The general aim of this PhD project was to investigate whether tight blood glucose control is effective and safe in a context of withholding early parenteral nutrition as compared to tolerating severe hyperglycemia, also in a multicenter setting, when provided with accurate measurement devices and guided by a clinically validated computerized insulin-titration algorithm, the LOGIC-Insulin algorithm. In addition, we investigated the potential drivers of the accentuated CRP rise observed by omitting early parenteral nutrition in the EPaNIC RCT and whether this higher CRP surge is related to the improved clinical outcome by late-PN. In the first part, we investigated in an international, multicenter RCT whether blood glucose control guided by the LOGIC-insulin algorithm could improve the quality of glucose control as compared to nurse-guided blood glucose control in critically ill adults. We demonstrated that glucose control guided by the LOGIC-insulin algorithm, also in this multicenter setting and accordingly provided in less experienced centers than in the Leuven ICUs, improved the quality of glucose control as compared to nurse-guided GC, with an improved time-in-range without increasing the risk of severe hypoglycemia. This confirms the results of the monocenter LOGIC-1 RCT and underlines the generalizability of the findings. In the second part, we investigated whether tight blood glucose control is effective and safe in a context of withholding early parenteral nutrition as compared to tolerating severe hyperglycemia, also in a multicenter setting, when provided with accurate measurement devices and guided by the validated LOGIC-Insulin algorithm. In this large multicenter RCT, in which 9230 adult critically ill patients were enrolled (TGC-Fast RCT), hyperglycemia was less severe than in previous RCTs that included use of early parenteral nutrition. Further lowering blood glucose to the normal range, while avoiding severe hypoglycemia, did not alter ICU dependency or mortality. However, severe acute kidney injury and cholestatic liver dysfunction were less prevalent in the tight-GC group, and the subgroup of patients with a neurologic or neurosurgical admission diagnosis had a possibly lower 90-day mortality when applying TGC. In the third part, we studied the likely drivers of the accentuated CRP increase by withholding early parenteral nutrition in the EPaNIC RCT. We investigated the possible role of systemic inflammation, mediated by cytokines (IL-6, IL-10, TNF-a) as well as the potential role of circulating insulin levels and dosages of macronutrients. In addition, we studied the relation of the CRP effect with the improved clinical outcome by withholding early parenteral nutrition. The higher CRP rise when omitting early parenteral nutrition appeared solely explained by the lower macronutrient doses (proteins and carbohydrates), since the lower insulin dose was rather associated with a lower CRP rise and the investigated cytokines were not affected by the intervention. Moreover, the higher CRP when withholding early PN was associated with a lower likelihood of early live ICU discharge and a higher likelihood of developing a new infection. The CRP effect of withholding early parenteral nutrition was therefore not accountable for the improved clinical outcome of the intervention. To conclude, we demonstrated that using the computerized LOGIC-insulin algorithm improved the quality of glycemic control as compared to nurse-guided GC in a multicenter setting, a necessary condition for using the algorithm in the subsequent TGC-fast RCT. In this large multicenter RCT, LOGIC-Insulin guided TGC did not alter the duration of ICU dependency or mortality in a context of withholding early parenteral nutrition. However, new kidney and liver dysfunction were less prevalent, and in the subgroup of patients with a neurologic or neurosurgical admission diagnosis, TGC may have lowered 90-day mortality. When investigating the potential drivers of the observed higher CRP rise when omitting early parenteral nutrition in the EPaNIC RCT, only the lower macronutrient doses appeared accountable. The observed improved clinical outcome in late PN-patients could not be attributed to this higher CRP rise.

Jaar van publicatie:2024

Toegankelijkheid:Open