AACE 2011: Endocrinology, Critical Illness, and Clinical Trials: Perils and Pitfalls
APRIL 14, 2011
Jamie Reiter, PhD
Poorly designed and executed trials evaluating hormonal control in critically ill patients may lead to erroneous conclusions with serious consequences for patients.
Patients suffering from critical illness (CI) are extremely challenging to treat, explained Greet Van den Berghe, MD, PhD, at the 20th Annual Meeting and Clinical Congress of the American Association of Clinical Endocrinologists. She heads the Department of Intensive Care Medicine, University Hospital Gasthuisberg, is affiliated with the Catholic University of Leuven as professor of medicine, and chairs the Division of Acute Medical Sciences there. Van den Berghe explained that the challenge in treating these patients is not only related to their sensitive condition. She said that there are several problems with randomized controlled trials (RCTs) evaluating hormonal control in CI patients, which may lead to inaccurate or misleading conclusions.
The median stay in the ICU for critically ill patients is 31 days, and 2 out of 10 patients don’t survive. Typical problems seen in this population are weakness (wasting of lean tissue, preservation of adipose tissue), non-recovering organ failure, and increased risk of infections. Tissue wasting is characterized by a breakdown of protein, which is metabolized by hormonal controls such as growth hormone, testosterone, insulin, thyroid hormone, and glucosteroids. “Clearly, there are opportunities to use hormone treatments to improve the outcomes in critical illness,” said Van den Berghe. She cautioned, however, that lack of insight into pathophysiology and inadequate study design can result in adverse consequences.
She reviewed several studies as examples of these two main errors in research. In one RCT, high-dose growth hormone (GH) treatment was evaluated in prolonged CI. After four weeks and 532 patients, results showed increased mortality. The fatal flaw was the researchers extrapolated results from acute CI to prolonged CI, two types of CI with different characteristics. For example, T3 and T4 are thyroid hormones often targeted for treatment, as low levels are often observed in the ICU. However, reasons for these low levels differ between the two hormones. Studies in rabbits suggest a drop in T3 may be associated with a lack of food. In the ICU, particularly in the first few days, patients are provided with very limited food, if at all. Therefore the low T3 may be due to the lack of food in acute CI, whereas low T4 may be attributed to the hypothalamus in prolonged CI.
Other issues with RCTs include sample sizes too small to address mortality, doses that are too high, studies starting too early in the disease, and use of the wrong hormone. As another example of flawed design, Van den Berghe described a study examining glucose control. Because hyperglycemia contributes to adverse outcomes, researchers have hypothesized that strict control of blood glucose will save lives. In the NICE-SUGAR study, patients were put on intermediate versus strict control of glucose. Surprisingly, mortality was higher in the strict control group. Van den Berghe pointed out that the study utilized a range of meters for monitoring glucose, sometimes even using more than one type with the same patient. There were also issues with precision and sensitivity of the meters. Therefore the “NICE-SUGAR study should be repeated with more attention paid to the accuracy of the glucose measuring device,” she said.
Van den Berghe emphasized the point repeatedly that we may be rejecting “another excellent hypothesis because we used the wrong test.” There are opportunities for treating patients, she explained, “but before we can get there, we have to do this right.”