The HICKORY trial data landed in March 2026 with a finding that stopped AstraZeneca’s hypophosphatasia program cold. After 25 weeks of treatment, patients receiving efzimfotase alfa walked farther on the Six-Minute Walk Test than placebo patients did. The drug moved the needle. The trial still failed to hit statistical significance on its primary endpoint. AstraZeneca’s explanation: the trial composition was wrong. The patients enrolled were not the patients who would show the effect. That is a design failure, not an efficacy failure, and the distinction carries a specific operational consequence for everyone who built, activated, and monitored the HICKORY sites.
Clinical operations teams tend to treat trial design as someone else’s document. The protocol comes down from clinical development, the feasibility questionnaire goes out to sites, and the CRO begins site activation. What gets lost in that handoff is a structured interrogation of whether the enrolled population will actually produce a readable signal at the primary endpoint. In a rare metabolic bone disease like hypophosphatasia, where patient heterogeneity is extreme and ambulatory function varies enormously across age and disease severity, the Six-Minute Walk Test can produce wildly different baseline values across sites. If the enrolled mix skews toward patients whose baseline performance is already high or low enough that 25 weeks of treatment cannot move them far enough to cross a significance threshold, the trial is mathematically compromised before the first site initiation visit.
That is the operational problem HICKORY surfaces. Not misconduct. Not GCP failures. A design signal that should have triggered an enrichment review before first-patient-in, and did not.
Where the Feasibility Process Breaks Down
Sites filling out feasibility questionnaires are not asked to validate endpoint sensitivity. They are asked how many eligible patients they can identify, whether they have the equipment and staff, and roughly what their screen failure rate will look like. Those questions tell the sponsor whether activation is possible. They do not tell the sponsor whether the enrolled population, aggregated across 20 or 30 sites with varying referral patterns and diagnostic practices, will produce a trial-level distribution tight enough for the primary endpoint to detect a treatment difference. That analysis lives in the protocol’s statistical section, and clinical operations almost never reads it during site activation.
The Tufts Center for the Study of Drug Development put the median direct cost of implementing a substantial protocol amendment in a Phase III trial at $535,000 USD, not counting delay costs or lost opportunity. In a rare disease program with a small eligible population, the indirect costs dwarf that figure: sites that screened and enrolled patients under one set of criteria now have to rescreen, coordinators have to relearn eligibility, IRB amendments consume weeks, and the patient community gets a message that the trial they signed up for has changed. Mid-trial redesigns in rare disease programs do not just cost money. They cost the trust of the advocacy networks that referred patients in the first place.
The question HICKORY forces onto every clinical operations lead running a rare disease Phase 3 right now is whether their current feasibility process would have caught this. Across the sites I work with at CliniBiz, the honest answer at most of them is no, because the feasibility process is calibrated to confirm site readiness, not to pressure-test endpoint sensitivity against the likely enrolled distribution.
A 2016 analysis by Hwang et al. covering 640 novel therapeutics that entered pivotal trials between 1998 and 2008 found that 344, or 54%, failed in clinical development. The reasons for failure spanned multiple categories, including efficacy and safety. That number has been cited in strategy decks for a decade without changing the feasibility workflow that sits upstream of it. HICKORY is one more data point in a pattern that the industry keeps acknowledging and not operationalizing.
The Enrichment Review No One Runs
Patient enrichment strategy in rare disease trials deserves its own operational checkpoint, separate from standard feasibility. The protocol defines eligibility criteria. Enrichment strategy asks a different question: within the eligible population, which patients are positioned to show the endpoint effect within the trial window? For a functional endpoint like the Six-Minute Walk Test, that means identifying the baseline performance band where treatment-driven improvement is both physiologically plausible and statistically detectable. Sites cannot do this analysis. The CRO’s feasibility team rarely does it either, because it requires clinical development and biostatistics to participate in a conversation that happens, if at all, in a protocol design review that clinical operations does not attend.
The fix is procedural. Before site qualification visits begin, the clinical operations lead should require a written population characterization document from the clinical team: what is the expected baseline range on the primary endpoint across sites, what is the within-group variability assumption embedded in the power calculation, and which site referral patterns are likely to produce patients outside that range. This document should be part of the site selection package. Sites with patient populations that skew outside the sensitivity band of the power calculation should be flagged before they are activated, not after enrollment is underway.
Sponsor oversight obligations under established GCP frameworks place explicit responsibility on sponsors to ensure that protocol design supports reliable data collection and that sites have the capability to enroll the intended population. Under longstanding sponsor oversight obligations, a design that mismatches the enrolled population with the endpoint’s sensitivity range represents an oversight failure, even if every site ran a perfect SIV.
What Changes on Monday
For sponsors activating rare disease Phase 3 programs now, the operational directive from HICKORY is concrete. Add a pre-activation enrichment review to your site qualification checklist. It does not need to be elaborate: a one-page summary from the biostatistics team that translates the power calculation assumptions into plain baseline eligibility guidance, specifying which patient profiles at the site level are and are not within the sensitivity window of the primary endpoint. Give that document to your CRAs. Train them to ask site coordinators the right screening questions before the first patient is consented.
For sites, the ask is simpler but requires a posture shift. When a sponsor’s feasibility questionnaire arrives for a rare disease indication, ask for the statistical sensitivity parameters before committing to an enrollment target. A site that has 12 HPP patients in its registry is not necessarily a site that has 12 patients who will move the endpoint. Understanding that distinction before the SIV protects the site from the second-worst outcome in clinical research: enrolling patients into a trial that cannot answer its own question.
AstraZeneca says it is evaluating next steps for the efzimfotase alfa program. Whatever redesign follows HICKORY will have to resolve the enrichment question the original protocol left unanswered. The sites that activated under the original design, the coordinators who screened and enrolled, the patients who completed 25 weeks of treatment: all of them did their jobs. The failure happened upstream, in a feasibility process that confirmed execution without validating the science it was executing against. The $535,000 median cost of a protocol amendment is recoverable. The enrollment window in a rare disease with a small global patient pool often is not.
References
- Endpoints News, “AstraZeneca blames trial design for Phase 3 failure of bone disease drug”
- Hwang et al. (2016), PubMed, “Failure of Investigational Drugs in Late-Stage Clinical Development and Publication of Trial Results”
- Tufts Center for the Study of Drug Development, PubMed, “Cost and impact of protocol amendments in Phase III clinical trials”
