WHO ARE THESE PATIENTS?

Section:

ChooseTop of pageWHO ARE THESE PATIENTS? <<WHEN IS A PHASE II TRIAL ...SHOULD TRIALS OF ANTICOAG...ReferencesCITING ARTICLES

Because of concerns about enrolling patients in whom activated protein C was already approved (i.e., severe sepsis with APACHE II score > 25) and those in whom the drug is contraindicated, the trial enrolled fewer than 5% of screened patients. The majority of the enrolled patients had either pneumonia or aspiration as the etiology of ALI. Although these patients were ill enough to require mechanical ventilation, they could not have had severe sepsis with APACHE II scores greater than 25, as such patients would then have met standard clinical parameters to receive activated protein C (4). Given these strict criteria, it is not surprising that enrollable patients were relatively rare and had lower mortality than previously reported in studies of ALI (5). With lung-protective ventilation, the 60-day mortality in the present study was only 13.5%. This mortality rate is not inconsequential; however, it would require a clinical trial with at least 3,600 patients to evaluate a 25% relative reduction in mortality in this subset of patients with ALI. Given the limited potential market in such patients and the time it would take to find and enroll them in a trial configured in this manner, one wonders if such a clinical study would be performed regardless of the phase II results.

WHEN IS A PHASE II TRIAL TRULY NEGATIVE?

Section:

ChooseTop of pageWHO ARE THESE PATIENTS?WHEN IS A PHASE II TRIAL ... <<SHOULD TRIALS OF ANTICOAG...ReferencesCITING ARTICLES

Liu and colleagues proposed a novel phase II endpoint: change in pulmonary dead space. Given the hypothesized mechanism of activated protein C, the authors reasoned that a proximal effect of this drug in ALI would be to reduce the dead-space fraction by preventing or treating microvascular thrombi. Change in pulmonary dead space is not a well-accepted phase II endpoint in ALI and the association between pulmonary dead space and mortality may not be entirely mediated by vascular phenomena (6). In their negotiations with the U.S. Food and Drug Administration (FDA), the investigators were steered away from pulmonary dead space toward ventilator-free days as a “more clinically relevant [primary] endpoint” and the study was powered to detect a difference of 6.5 ventilator-free days. This is an enormous effect, which far exceeds the observed benefits of lung-protective ventilation (2 ventilator-free days) or conservative fluid strategy (2.5 ventilator-free days) in the landmark trials by the ARDS (Acute Respiratory Distress Syndrome) Network (7, 8). Obviously, the present trial was severely underpowered to detect any realistic effects on clinically relevant outcomes.

How much confidence should readers place in the failure to see even a trend toward benefit from activated protein C in a clinical trial of 75 patients? Readers need look no further than the confidence intervals around the clinical outcomes to see that large benefits and harms have not been excluded by this trial. Assume that activated protein C really does reduce mortality by 25% in this patient population. Now, imagine that we could perform 100 randomized controlled trials just like the one reported by Liu and colleagues looking for a trend (P < 0.2) in mortality. It turns out that only 12% of these small trials would show a beneficial trend in mortality, even if studying an effective therapy. The rest would be incorrectly interpreted as negative. We don't need to rely on hypothetical examples. The pilot study for the ARDS Network lung-protective ventilation trial enrolled 52 patients and tested a protocol similar, but not identical, to the one that ultimately proved to reduce mortality (9). This pilot trial found absolutely no difference or trend in a difference in mortality between the groups. Had lung-protective ventilation been abandoned on the basis of this “negative” trial, the phase III ARDS Network trial never would have been performed.

Phase II trials are a pivotal decision node in drug development. On the basis of the results of data from as few as 30 patients, investigators and regulators must make decisions that have profound clinical and financial implications. Moving beyond phase II involves investing millions of dollars and enrolling hundreds, if not thousands, of subjects in a multicenter trial. The wrong decision can overlook a life-saving treatment or expose a large number of patients to a drug that is ultimately found to be ineffective or harmful.

The FDA provides guidance on the design of phase II studies for evaluating treatments for a variety of diseases (10). Drugs that do not shrink tumors, lower blood sugar, reduce HIV load, or lower cholesterol in phase II studies will probably not move ahead to phase III trials in cancer, diabetes, HIV infection, or hyperlipidemia. There is limited guidance for investigators performing phase II trials in ALI or sepsis. The reason for this is that we have so few examples of treatments that have been studied from phase I through successful phase III trials and such a limited understanding of the heterogeneous mechanisms of critical illness that it has been difficult to identify reliable surrogate outcomes (11). Experience with inhaled nitric oxide and prone mechanical ventilation (which both improve oxygenation but have not been shown to affect mortality) and lung-protective ventilation (which does not improve oxygenation but reduces mortality) suggests that defining physiologic surrogate outcomes in ALI may not be straightforward. Although it is possible that demonstrating reductions in bronchoalveolar lavage and serum inflammatory markers may be reliable surrogates for improved clinical outcomes from ALI, such parameters have not been well validated and have only been explored in the setting of lung-protective ventilation (12). Therefore, the challenge for phase II trials in critical care is not statistical, but scientific. Underpowered studies of clinically relevant outcomes such as mortality or ventilator-free days looking for trends are not a substitute for mechanistically compelling surrogate markers of therapeutic success.

SHOULD TRIALS OF ANTICOAGULANTS AND FIBRINOLYTICS IN ALI BE ABANDONED?

Section:

ChooseTop of pageWHO ARE THESE PATIENTS?WHEN IS A PHASE II TRIAL ...SHOULD TRIALS OF ANTICOAG... <<ReferencesCITING ARTICLES

Although the original study of activated protein C in sepsis was stopped early for efficacy, such positive results have not been replicated in subsequent clinical trials involving activated protein C or other agents that have potent effects on the coagulation system. Activated protein C was not found to be effective in children with sepsis or in septic adults with low risk of mortality (13, 14). Tissue factor pathway inhibitor (TFPI) and antithrombin did not demonstrate efficacy in reducing death rates in patients with severe sepsis (15, 16). In the studies with activated protein C, clinical efficacy, if it exists, appears to be present only in patients with high risk of death. It is important to note that the morality rate in the study by Liu and colleagues was only 13.5%, which is substantially less than that in the lowest APACHE quartile of the Recombinant Human Activated Protein C Worldwide Evaluation in Severe Sepsis (PROWESS) trial, in which no benefit was found with activated protein C.

Given the accumulated data from clinical trials exploring manipulation of the coagulation and fibrinolytic systems in sepsis and ALI, it is difficult to be enthusiastic about this approach in most patient groups, and especially in those with low risk of death. Even in patients with the highest predicted mortality, such as those with multiple organ dysfunction and hypotension requiring vasopressor therapy, efficacy has only been shown in a single study and the validity of these results has been questioned. A confirmatory study of activated protein C in patients with severe sepsis and vasopressor-dependent hypotension is underway and should help to define the utility of this agent in critically ill patients. However, at the present time, there seems to be no reason to assume that activated protein C or other anticoagulant approaches have benefit for nonseptic ALI or sepsis-induced ALI, unless accompanied by severe extrapulmonary organ dysfunction, including fluid-unresponsive hypotension.