In the recent pilot trial reported by Strandenes et al., the authors studied the use of the cold storage of platelets in the setting of complex cardiac surgery and “focused on feasibility and safety to provide preliminary data to evaluate hemostatic potential of cold stored platelets.”  Although their study highlights an important area of transfusion medicine, a number of study design and registration deficiencies add considerable uncertainty to its interpretation. Although feasibility (of performing a subsequent larger trial) and safety were its stated goals and are consistent with the principles of pilot study design  the design and reporting of the study appear to interchangeably mix pilot trial elements with those of an definitive trial primarily assessing the effect of an intervention (i.e., cold storage). For example, their primary hypothesis that “relative to transfusion with room temperature-stored platelets, cold-stored platelets reduced postoperative bleeding with sufficient platelet aggregation in patients who developed indications for platelet transfusion during complex cardiothoracic surgery,” clearly indicates a primary trial objective addressing an intervention, not the feasibility and safety that defines a pilot trial.

Irrespective of what arguably can be nuanced differences between these study types, there are other aspects of this study that also challenge its interpretation. For example, the initial trial was registered as a noninferiority trial, although the design was subsequently changed (and updated on clinicaltrials.gov; NCT02495506) to reflect a superiority trial design. Thus, the trial appears never to have been intended to be designed (or registered) as a feasibility study. The originally registered primary outcome was platelet aggregation, although this too was subsequently changed (and registered after patient enrollment) to 24-h chest tube output. However, even that 24-h chest tube output primary outcome was subsequently changed (but not updated in the trial registry) to chest tube output by 8:00 am the morning after surgery (i.e., less than the originally registered 24 h). There were also a number of other unexplained changes from the trial registration. For example, the number of patients enrolled in the trial was initially registered as 30 per group but was changed (with the registration updated) to 20 per group; however, their article eventually reported on 25 per group. Part way through the trial, and after an interim analysis (which itself is somewhat problematic because of the lack of statistical adjustment for this early look at the data), a second observational arm was added, in a previously unplanned phase II of the study that extended the cold storage of platelets to 14 days. This phase II was registered to include 10 patients, although 15 patients were eventually included in the study report.

Another important and well-recognized goal of a pilot trial is to examine safety.  Although one could argue that, in a trial such as Strandenes et al. that includes only 65 patients, it is difficult to confidently understand the point estimates of safety; at the very least, the safety endpoints should be clearly defined a priori. Indeed, the authors defined the safety endpoints in their methods section as venous and arterial thromboembolism, transfusion reactions, length of intensive care unit stay, and mortality at 28 days, but they subsequently also reported time to extubation as a safety parameter, seemingly because this was statistically different between the groups. Furthermore, in the trial’s registration, the safety endpoints of intensive care unit length of stay and 28-day mortality were not included. Last, other potentially important platelet-related safety endpoints were not reported, such as acute kidney injury or neurologic dysfunctio both of which could be adverse reactions plausibly related to platelets. Because bacterial contamination of platelets is a well-known risk factor of storage, including an infectious safety endpoint would also have been optimal. Interestingly, they did perform bacterial testing in the platelet concentrates at day 1, but they did not repeat this bacterial testing at the time of transfusion (up to 14 days later).

Their report’s conclusions state that their “pilot trial supports the feasibility of platelets stored cold for up to 14 days and provides guidance for future pivotal trials.” We argue that it would be difficult to confidently ascertain the feasibility of performing a subsequent definitive trial when patient enrollment (n = 65) took 41 months and the study had a multitude of both reported and unreported trial design features that may have significantly impacted the results. That is, as typical feasibility endpoints (e.g., patient enrollment rates) were not explicitly stated in either the trial registry or in the article itself, whether they actually met a threshold for feasibility at all cannot be determined. Without doubt, the efficacy of prolonged platelet storage would be a valuable goal, both to preserve the platelet donor pool and optimize the effectiveness of transfusion. However, a trial that states feasibility as an objective but only reports the intervention’s effect (in a likely underpowered manner) seems unlikely to confidently inform a future definitive trial.