Risk of Stopping Trial Early for Treatment Benefit - Story of Veru's Sabizabulin for Treatment of Covid

Today, we learned that FDA declined to approve the emergency use authorization (EUA) for Veru's sabizabulin for the treatment of Covid. The decision is not surprising given that in Nov 9, 2022, the FDA advisory committee voted 8-5 against the sabizabulin treatment aimed at hospitalized patients with moderate to severe infection who are at high risk for acute respiratory distress syndrome.

• "Veru's EUA ambitions crash after FDA declines to authorize its COVID treatment"
• Veru Provides FDA Update on Request for Emergency Use Authorization for Sabizabulin for Hospitalized COVID-9 Patients at High Risk for ARDS
• FDA Adcomm Votes Against Veru's COVID-19 Therapeutics

However, this case is a perfect example demonstrating the risk of stopping a clinical trial early for treatment benefit (or for overwhelming efficacy). 

The pivotal study was an international, multicenter, randomized, double-blind, placebo-controlled, parallel group study of 60 days duration that evaluated the efficacy and safety of VERU-111 (sabizabulin) in hospitalized adult subjects with COVID-19 infection described as being at “high risk for ARDS”. Subjects were included if they met criteria for World Health Organization (WHO) ordinal scale category 5 (non-invasive ventilation or high-flow oxygen) or category 6 (mechanical ventilation); or category 4 (oxygen by mask or nasal prongs) with the following comorbidities: asthma, chronic lung disease, diabetes, hypertension, severe obesity (BMI ≥40), 65 years of age or older, primarily residing in a nursing home or long-term care facility, immunocompromised. The primary efficacy endpoint is all-cause mortality at Day 60. 

According to the study protocol, approximately, 300 subjects were planned to be randomized at a 2:1 ratio into two treatment arms (200 subjects in the VERU-111 treated group and 100 subjects in the Placebo treated group). The protocol also pre-specified a formal interim analysis conducted by an independent data monitoring committee:

An efficacy interim analysis of the primary endpoint, all-cause mortality at Day 60, will be conducted when approximately 67% of the subjects (~200 subjects) have completed day 60, died or withdrawn for other reasons. The interim and final analyses will follow an O'Brien-Fleming group sequential design, with a plan of stopping the trial at the interim analysis if the results are statistically significant in favor of VERU-111. The criterion for efficacy at the interim analysis will be a two-sided 0.0121 p-value (a one-sided p-value <=0.0061 in favor of VERU-111). If the criterion is not met, the trial will continue, and the final analysis will have a criterion for efficacy of a two-sided p-value <=0.0463 (one-sided 0.0232 in favor of VERU111).

The formal interim analysis was conducted after 204 patients were randomized into the study and followed up to 21 days (134 patients in Sabizabulin group and 70 patients in Placebo group). Sabizabulin treatment resulted in a 24.9 percentage point absolute reduction and a 55.2% relative reduction in deaths compared with placebo (odds ratio, 3.23; 95% CI confidence interval, 1.45 to 7.22; P=0.0042). The mortality rate was 20.2% (19 of 94) for sabizabulin versus 45.1% (23 of 51) for placebo.

Given that the p-value 0.0042 is less than the criterion for stopping the trial for efficacy (0.0121), the sponsor (Veru) stopped the study and declared the overwhelming efficacy in the hope of FDA's approval for emergency use authorization. 

• Veru’s Stops COVID-19 Treatment Trial in Face of ‘Overwhelmingly Positive’ Results

The interim analysis results were then published in the top medical journal - New England Journal of Medicine "Oral Sabizabulin for High-Risk Hospitalized Adults with Covid-19: Interim Analysis".

The original sample size of 300 patients was already pretty small in trials for the treatment of COVID-19. The sample size was further decreased due to the early stop for 'overwhelming' efficacy. During the FDA's review, the small sample size was one of the big issues. FDA expressed their concern about the evidence for supporting the benefit outweighing the risk in their briefing document provided for the advisory committee:  

The FDA Review team acknowledges that Study 902 met its prespecified primary endpoint of all-cause mortality at Day 60. We also note that the VERU-111 program is quite small in size compared to other therapeutic programs for patients hospitalized with COVID-19. As detailed in the briefing document, our review has identified a number of uncertainties with the data, which we raise in the context of this small trial in critically ill patients. These include:

• High placebo group mortality rate
• Potential for unblinding events with enteral tube administration
• Baseline imbalances in standard of care therapies
• Differences in hospitalization duration prior to trial enrollment
• Uncertain effects of goals of care decisions on all-cause mortality
• Negative studies with other microtubule disruptors in COVID-19 
• Uncertainty in identification of a clinically relevant patient population 

Based on our review, none of these uncertainties or imbalances alone invalidate the mortality benefit observed in Study 902, but all of these issues together in a small trial which is more vulnerable to imbalances raise questions about the results. We conducted sensitivity analyses to investigate the potential impact of the noted imbalances. However, these analyses cannot eliminate the concern that certain baseline imbalances across treatment groups may have impacted study outcomes, due to the small study size. In addition, these issues raise concern that, even when using an objective endpoint such as mortality, observed results can be subject to biases in a small trial of short duration in critically ill patients. We ask the AC panel to consider these uncertainties together and how they affect the interpretation of the mortality data. 

When considering whether to authorize the emergency use of a product under EUA, the Agency must determine, among other requirements (see section 2.1.3), whether “the known and potential benefits of the product when used to diagnose, prevent, or treat the identified serious or life-threatening disease or condition, outweigh the known and potential risks of the product”. Evaluation of the potential risks in the VERU-111 development program is limited by the atypically small safety database, comprising a total of 149 subjects who received VERU-111 for the proposed use in COVID-19. While the small safety database limits our ability to identify clinically significant safety signals, potential safety signals identified in our review are urinary tract infections (including serious infections), ......

There is no guarantee that Veru's EUA application would be approved had the study not been stopped early. But I think that the chance of EUA approval would be much higher if the study was allowed to be completed according to the protocol and 100 additional patients were randomized into the study. To decide whether a study should be stopped early for efficacy, the stopping criterion is not the only factor to consider. I wished that the sponsor had consulted with the regulatory authorities before the decision to stop the study early for efficacy. 

On a separate note, FDA's "Good Review Practice: Clinical Review of Investigational New Drug Applications" spelled out their concerns about early stopping for efficacy: