Halving Enrollment Time in a Phase III Solid Tumor Trial

Background

A biotech sponsor running a Phase III randomized controlled trial for an investigational PD-1 inhibitor in advanced solid tumor indications faced a familiar problem: 14 months into the planned 18-month enrollment window, they had reached only 38% of target enrollment. Screen failure rates across their 14 sites averaged 61%. The sponsor's CRO had already recommended adding 6 new sites — at an estimated cost of $720,000 — to recover the timeline.

The Challenge

The protocol had a complex set of inclusion criteria including specific prior treatment requirements, ECOG performance status thresholds, and biomarker eligibility. Sites were relying on manual chart review to identify candidates — a process that took approximately 8–12 coordinator hours per eligible patient identified, with most candidates failing at formal pre-screening due to undocumented prior treatment history.

What TrialVyx Did

TrialVyx deployed its phenotype engine across all 14 existing sites, connecting to their Epic FHIR R4 endpoints. The protocol's inclusion/exclusion criteria were mapped to computable phenotype dimensions within two weeks of engagement. Within three weeks of activation, the first cohort of pre-screened referrals began flowing to site coordinators.

Key capabilities used in this deployment:

• Prior treatment history extraction from structured medication records and clinical notes (NLP)
• ECOG performance status inference from recent clinical note language and functional assessment data
• Hard exclusion flagging for patients with documented contraindications to immunotherapy
• Real-time match score updates when patients' clinical status changed (e.g., new lab results ruling in or out)

Results

At 11 months post-activation (and 14 months into the original enrollment window), the trial reached full enrollment. The sponsor cancelled the planned site expansion. Screen failure rate dropped from 61% to 29% — a 52% reduction — reflecting the improvement in candidate quality entering formal pre-screening.

Coordinator time spent on patient identification decreased by approximately 70%. The freed coordinator capacity was redirected to patient engagement, consent discussions, and data collection — reducing data query rates in parallel.