HEPZATO can be delivered only at REMS-certified sites. A patient reaches one by moving through four sets of hands: an ophthalmologist, an ocular oncologist, a medical oncologist, and the treating center. No single party sees that pathway end to end. This brief maps it from the public record, and outlines a referral-visibility layer for Delcath's field organization.
Delcath is a public company. Its filings, FDA documents, and the clinical literature on uveal melanoma are all open. The replay below is the research trail behind this brief, condensed.
The device is approved and selling. The next stage of growth is a distribution problem: getting appropriate patients to the small, fixed number of centers certified to treat them.
Read together, these numbers point at one lever. Revenue is now a function of procedures per certified center, and procedures are a function of appropriate patients arriving. Each center sits at the end of a referral catchment: the practices that route patients to it. That catchment is where the next procedure comes from, and it is the part of the business no filing describes.
Uveal melanoma is rare, and the route from diagnosis to a certified center passes through several specialties, over years. Each one owns a different segment of the patient.
Two facts sit underneath the pathway. Surveillance is split: the ocular oncologist watches the eye, the medical oncologist watches the liver, and the certified center sees the patient only once a referral is made. And the systemic alternative does not cover everyone: tebentafusp is limited to HLA-A*02:01-positive patients, roughly half of them. source For the other half, and for liver-dominant disease specifically, the certified-center pathway is the access route.
The eligible population is small and well characterized. The open question is operational: of the patients who should reach a certified center, how many actually do, and where do the rest stop? No public source answers this, because no one measures the pathway.
No public figure quantifies how many eligible patients never reach a certified center. That absence is the finding. The loss is invisible because the pathway is not instrumented.
The seams are specific. A surveillance interval is missed. A liver metastasis is found but no referral is made. A referral is made but the patient never arrives. Or the patient arrives outside the treating window, once liver involvement passes the labeled threshold or disease spreads beyond it. Each of these is a different intervention, and today they all read the same from the field: a center that simply is not treating this quarter.
Phase 1 does not track patients. It instruments the pathway around each certified center at the account level, so the field team can see where referrals stall and act on it.
For each of the nine territories, the layer maps the referral catchment of every certified center: which practices and specialties feed it, how recently, and where the flow thins. It turns field anecdotes into a structured, comparable picture. The unit of analysis is the referring practice and the center, never the patient. It is an intelligence and workflow layer, not a CRM, a REMS platform, a patient hub, or a reimbursement system.
A live referral map for every certified center: which practices feed it, where the flow is thinning, and which stalled feeders a territory should work next. Built on public and Delcath-owned account data, no patient records.
The discipline is deliberate. Phase 1 uses non-PHI account data, keeps a person accountable for every action, and stays inside a small set of measurable outcomes: is each center's catchment visible, is the principal stall identified, is the next action owned. Everything else is logged for later, not built now.
Patient-level referral tracking is where "no patient falls through the cracks" becomes literal. It is named here as the expansion, not the starting point, because it carries privacy, consent, and security obligations that Phase 1 does not.
An individual patient followed from liver-metastasis identification to treatment, so a stalled hand-off raises a flag instead of disappearing. Requires a defined consent model, PHI-grade security, and clinical and regulatory review before any build.
The order matters. Prove that a non-PHI view of the pathway changes field behavior and moves procedures first. Only then take on the consent architecture and the data obligations that patient-level tracking requires. The expansion is earned by the Phase 1 result, not assumed.
This brief is a demonstration, not a product. It exists to show that the referral pathway can be mapped from evidence, and that the mapping is worth instrumenting.
The full build is a paid engagement with Oddmind. This demonstration runs on hand-verified public data and carries no patient information.