Clinical Cases in Pharmacology Clinical Cases  ·  Medical Oncology Vol. III  ·  Skin Cancer, Sarcomas, and Unknown Primary Site  ·  Molecular Profiling Against Empiric Chemotherapy in CUP
Medical Oncology Vol. III, Case 0012 — Skin Cancer, Sarcomas, and Unknown Primary Site

Cancer of Unknown Primary with a HER2 Signal: Trusting a Biomarker Without a Matching Trial

A single patient whose cancer never told the team where it started, and a genomic report that offers an answer of a different kind. The disagreement is about how much confidence a biomarker found on a broad panel deserves when the specific population it would treat has never actually been studied.

Abbreviations, terms, and other agents mentioned in this case Cancer of unknown primary (CUP) — a metastatic cancer for which, despite thorough workup, the original site of tumor origin cannot be identified  ·  Favorable-subset CUP — a small group of CUP presentations whose overall clinicopathologic picture closely matches a specific known cancer type, allowing site-specific therapy despite the unidentified primary  ·  Comprehensive genomic profiling (NGS panel) — broad next-generation sequencing of a tumor to identify actionable mutations or amplifications, independent of the tumor's tissue of origin  ·  ECOG — Eastern Cooperative Oncology Group performance-status scale  ·  Antibody-drug conjugate — an antibody carrying a cytotoxic payload directly to cells bearing its target antigen — trastuzumab deruxtecan is the HER2-directed example here
Presentation

R.T., a 71-year-old man, spent thirty-nine years as an accountant before retiring, work he says trained him to want a clear number before making any decision -- an instinct that has made the past three weeks genuinely difficult. He presented with a firm left supraclavicular lymph node, and biopsy showed a poorly differentiated adenocarcinoma; an exhaustive search for the primary site -- CT of the chest, abdomen, and pelvis, upper and lower endoscopy, PSA, and a whole-body PET scan -- found nothing else, leaving him with a diagnosis of cancer of unknown primary rather than a named disease with its own established treatment pathway.

Comprehensive genomic profiling, sent as part of his workup, identified a HER2 gene amplification in the tumor -- a finding that matters because CUPISCO, the trial testing molecularly-matched targeted therapy against standard empiric platinum-based chemotherapy in CUP patients outside a defined favorable clinical subset, found a real progression-free survival advantage for the molecularly-guided approach. What CUPISCO's result doesn't yet settle is whether that PFS advantage has translated into longer overall survival, and it doesn't erase the fact that HER2-directed antibody-drug conjugates like trastuzumab deruxtecan have never been tested in their own pivotal trials on a population of patients whose primary tumor site is unknown -- every prior approval rests on a defined cancer type his diagnosis, by definition, doesn't have.

The one place genuinely biomarker-driven CUP therapy already has a long track record is different from his situation in a way worth naming plainly: certain CUP presentations -- axillary-node-only disease in a woman, for instance, treated as though it were breast cancer -- are matched to a specific known cancer type by their whole clinicopathologic picture, not by a single mutation on a broad panel, and that favorable-subset paradigm has mature outcome data behind it that a HER2 amplification found in isolation does not yet have.

He has asked, in nearly the same words each visit, for someone to simply tell him which number is the right one to trust -- a request the team has had to answer honestly has no single correct figure behind it yet, only two real bodies of evidence that measure different things and haven't been reconciled by a trial built to compare them directly.

R.T. · 71 No primary site identified after full workup
History
Left supraclavicular adenopathy, biopsy-proven poorly differentiated adenocarcinoma
Workup
CT chest/abdomen/pelvis, upper/lower endoscopy, PSA, whole-body PET — no primary site identified
Genomic profiling
HER2 gene amplification identified on comprehensive NGS panel
ECOG performance status
1
Comorbidities
Well-controlled type 2 diabetes

How much a biomarker can stand in for a known cancer type

Medical Oncologist Opening

CUPISCO is the one randomized trial actually built to test this question, and it found a real progression-free survival advantage for molecularly-matched therapy over empiric platinum-based chemotherapy in CUP patients outside a favorable subset. He has an actionable HER2 amplification. I don't think we should default to empiric treatment when a trial specifically designed for this scenario points the other way.

Second Medical Oncologist Response

I'd want to be precise about what CUPISCO has and hasn't shown. Its progression-free survival benefit hasn't yet translated into a confirmed overall survival advantage. And no HER2-directed antibody-drug conjugate has ever been tested in a pivotal trial on a population defined by unknown primary site -- every existing approval rests on a named cancer type he doesn't have. Empiric platinum-based chemotherapy is the actual validated standard for CUP as its own disease category.

I take CUPISCO's result seriously as real evidence, not nothing. I just don't think a progression-free survival signal, without a matured survival benefit and without any trial data specific to HER2-directed therapy in an unknown-primary population, is strong enough yet to displace the treatment that's actually been validated for his exact diagnosis.

Pathologist Final

I'd add a distinction that I think gets lost when we call this 'biomarker-driven CUP therapy' as though it's all one thing. The form of that approach with genuinely mature outcome data -- axillary-node-only disease in a woman treated as breast cancer, for instance -- matches a patient's whole clinicopathologic picture to a known cancer type, not one mutation on a broad panel in isolation. His HER2 amplification is real information. It's not the same category of evidence as that whole-picture match.

Given that, and given the genuine uncertainty on both sides here, I'd start empiric carboplatin-paclitaxel now rather than treat the HER2 finding as though it already carries favorable-subset-level confidence -- and hold trastuzumab deruxtecan explicitly in reserve for next-line therapy if he progresses, at which point a real trial of molecularly-matched treatment, with much less to lose by that point, becomes a clearly reasonable next step.

Regimen selected
Carboplatin + Paclitaxel
Platinum Agent + Taxane · Empiric first-line
Chosen as the actual validated standard of care for CUP outside a defined favorable clinicopathologic subset, given genuine uncertainty about extrapolating HER2-directed therapy to an untested population.
Trastuzumab Deruxtecan — Held in Reserve
HER2-Directed Antibody-Drug Conjugate · Contingent on progression
Retained as a next-line option given his real HER2 amplification, to be used if empiric chemotherapy fails rather than as an unproven first-line substitute for it.
Trastuzumab Deruxtecan, Immediate — Not Adopted
Considered, not adopted as the first step
A real, biologically plausible option given CUPISCO's PFS finding, but judged premature given the absence of any pivotal trial data for HER2-directed therapy in a population defined by unknown primary site.
Where this was left

Agreed: start empiric carboplatin-paclitaxel, with trastuzumab deruxtecan held explicitly in reserve as a next-line option given his HER2 amplification, contingent on progression rather than used as an immediate substitute for standard treatment.

Educational content only — a composite teaching case, not a real patient encounter or a substitute for clinical guidance. About These Cases →