How One Biopharma Company Overcame Trial Failure
A conversation with Eyal Shoshani, vice president clinical affairs, PolyPid
Clinical trial failure is a reality many researchers face in their careers and many companies navigate in the pursuit of drug approval.
Eyal Shoshani and the team at PolyPid are no different.
When the COVID-19 pandemic's unforeseen impact on surgical site infection rates interrupted their SHIELD I Phase 3 trial for D-PLEX100, they turned the setback into a learning opportunity.
In this Q&A, Shoshani reflects on the turnaround, highlighting their second attempt at a trial, as well as the things they did to revamp the trial design, and muses on the impact of AI in this and future trials.
Clinical Leader: Take us back to the initial failure of the PolyPid’s SHIELD I Phase 3 trial for D-PLEX100 in 2022. What happened, and what factors contributed to the trial’s failure?
Eyal Shoshani: When we initiated SHIELD I in 2020, our goal was to evaluate D-PLEX100 in a global Phase 3 setting. This followed a Phase 2 trial in Israel that showed encouraging safety and efficacy results across a range of incision sizes. Building on that success, we designed SHIELD I as a large multinational study of about 1,000 patients, selecting 90 sites across the U.S., Europe, and Israel, with 60 actively recruiting. We built a strong operational infrastructure supported by regional CROs and scaled our internal clinical team, believing we were well positioned for success.
Unfortunately, an unprecedented factor intervened: the COVID-19 pandemic. The suspension of many procedures, heightened use of PPE, mask mandates, and enhanced sterility protocols profoundly reduced the overall incidence of surgical site infections during the study period. While this was, of course, a positive outcome for patients, it meant the treatment effect of D-PLEX100 was obscured, and the trial was unable to meet its primary endpoint.
It was a humbling reminder that even the most carefully prepared studies can be impacted by forces beyond our control. At the same time, the trial was far from wasted. A prespecified subgroup analysis revealed a statistically significant 54% reduction in infections in subpopulation patients with incisions greater than 20 centimeters. That insight became the foundation for SHIELD II and ultimately put us back on a path to success.
Wanting to try again, PolyPid redesigned the trial. What changes were made?
Following the SHIELD I readout, we conducted a prespecified subgroup analysis that revealed a very compelling signal. Patients with incisions greater than 20 centimeters experienced a statistically significant 54% reduction in surgical site infections compared to standard of care alone (p<0.0032). This insight guided a swift protocol amendment and the resumption of SHIELD II, focusing exclusively on this large incision (greater than 20 centimeters) population.
Targeting patients with large incisions was both scientifically justified and clinically meaningful, as these surgeries inherently carry the greatest risk of infection and represent the area of highest unmet need. To support the redesign, we realigned our operational strategy: tightening site oversight, enhancing training to ensure consistent product application, and reinforcing our partnerships with CROs and investigators to maintain quality across a multinational study footprint.
The result was a trial that was both more focused and more efficient. By narrowing our scope while strengthening operational execution, we generated clear, robust, and clinically meaningful top-line results that validated the promise we had seen in the subgroup analysis.
How did evolving technology, specifically AI-assisted tools, impact your design of the trial?
When we designed and began the SHIELD II trial, AI-assisted tools were not yet mature enough or widely available to be integrated into our processes. The rapid emergence of AI solutions has largely taken place over the last 12 to 18 months, with specialized applications for clinical trials appearing only recently — and many are still in development or pilot phases.
Going forward, we recognize that these tools hold significant promise for improving trial design, patient recruitment, data monitoring, and overall efficiency.
In what ways do you think leveraging this technology (or technologies) will improve the odds of a successful trial?
Clinical trials generate enormous volumes of complex data and, too often, the processes behind them still rely on manual entry and reconciliation. That not only consumes resources but also creates opportunities for error. AI offers a real opportunity to streamline these processes. For example, when it comes to safety data monitoring, adverse event reporting, monitoring visit reports, protocol deviations, IP management, and seamless integration of systems like EDC, ePRO, CTMS, and eTMF, AI can enhance accuracy, efficiency, and consistency across trial operations.
Looking ahead, the more we can integrate validated AI tools into trial operations, the stronger our ability will be to deliver reliable data faster and at lower cost. Of course, regulatory bodies are rightly cautious, and AI should always be deployed with clear boundaries and rigorous oversight. But when applied thoughtfully, these technologies can help optimize trial execution, improve data quality, and ultimately support better outcomes for patients.
What advice do you have for those looking to bounce back from a failed trial?
First, don’t view a failed trial as the end of the road. Every study, even if it doesn’t meet its endpoints, yields critical insights that can shape the path forward. At PolyPid, we learned the value of robust feasibility assessments, early and proactive regulatory engagement, and, above all, adaptability. We also learned the importance of investing in hands-on site training and close oversight, because protocols are only as strong as their execution.
Building strong partnerships with CROs and clinical sites is also essential for maintaining data quality and operational momentum. Most importantly, remain resilient and open to strategic pivots. In our case, the subgroup analysis results maintained our confidence in the effectiveness of our product, and that ability to adapt quickly ensured that SHIELD II was a success.
My advice is to treat every setback as an opportunity to refine, focus, and come back stronger.
Specifically, what do you recommend sponsor companies do in terms of considering, selecting, and choosing a technology to support that subsequent trial?
Beyond the proliferation of use of generative AI in our personal lives on general-purpose platforms, we are now witnessing the emergence of specialized AI solutions for clinical trials. These platforms can provide sponsors with the ability to better plan and execute trials with greater accuracy, speed, and efficiency.
Sponsors who fail to adopt and integrate AI tools into clinical trial development and management — an arena already defined by intense competition, complex regulations, and high costs — risk losing both time and resources. In clinical trials, few assets are more critical or sensitive than time.
My recommendation to every sponsor, especially during the early stages of trial planning, is to carefully evaluate the new AI-based tools entering the market. Key considerations should include their fit for the specific needs of the study, their level of maturity and market validation, and how seamlessly they can be integrated into the organization’s structure and workflows.
I believe many sponsors will be pleasantly surprised to discover a wide range of AI solutions already available that can reduce manual workload, streamline cumbersome processes, and ease the burden on the research team— whether on the sponsor side, the clinical site, or the CRO.
About The Expert:
Eyal Shoshani is the vice president of clinical affairs at PolyPid and a seasoned clinical operations leader with 20 years of experience spanning drug, biologic, cell therapy, and medical device development. He has directed multinational Phase 1-4 trials under FDA, EMA, and IMOH regulations, including a successful FDA BLA approval. Previously, he held senior clinical roles at Gamida Cell, Minovia Therapeutics, Pluristem Therapeutics, and Dexcel Pharma, after starting his career at IQVIA.
With expertise in clinical operations, project management, and regulatory compliance, Eyal has successfully led global development programs from early-stage trials through approval. He holds a BSc, MBA, and a certification in orphan drugs clinical trials.