Molecularly guided therapy extends disease control in unfavorable cancer of unknown primary
Patients with unfavorable cancer of unknown primary rarely present with a clear therapeutic path. CUPISCO reframed that uncertainty by using comprehensive genomic profiling after induction chemotherapy, then matching treatment to an actionable molecular alteration whenever possible. In the updated analysis, that strategy continued to separate from additional empiric platinum-based chemotherapy on the end point that matters most in this setting: time without progression.
At data cutoff on December 6, 2024, 436 patients had been randomly assigned after disease control with three cycles of induction chemotherapy. The molecularly guided therapy arm included 326 patients, while 110 patients continued chemotherapy for three more cycles. Median follow-up reached 37.0 months, long enough to see whether the early signal held up under more mature observation. It did.
The updated median progression-free survival was 6.1 months with molecularly guided therapy versus 4.4 months with chemotherapy, with a hazard ratio of 0.75 and a P value of .017. The survival curves did not merely drift apart; they maintained a clinically meaningful separation in a disease where durable control is often elusive and treatment options are usually chosen without a biological map.
Genomic selection changed the treatment logic after induction response
The trial design matters because it selected a group with at least some chemotherapy sensitivity up front. Only patients who reached disease control after induction chemotherapy entered the randomization phase, which means the comparison was not between a targeted approach and chemotherapy in a fully refractory population. It was a test of whether genomic information could refine second-phase therapy after initial stabilization had already been established.
That sequence mirrors how many clinicians encounter CUP in practice. First comes empirical treatment to gain control, then the harder question: whether to keep cycling through the same broad cytotoxic strategy or pivot to a therapy chosen on the basis of tumor biology. CUPISCO supports the second option in this molecularly enriched, treatment-responsive subset.
The biologic rationale is straightforward. Broad chemotherapy exposes tumor cells to nonspecific DNA damage and mitotic stress, but it does not distinguish between tumors driven by a discrete oncogenic alteration and those without a clearly targetable dependency. Comprehensive genomic profiling can identify those dependencies and route patients toward a more rational therapy selection process through a molecular tumor board. In a cancer category defined by diagnostic ambiguity, that kind of precision can improve the odds of controlling disease before resistance compounds the problem.
Overall survival favored molecularly guided therapy, but the comparison remained close
Overall survival moved in the same direction, although the result did not reach conventional statistical significance. Median OS was 15.2 months in the molecularly guided therapy group and 12.8 months in the chemotherapy group, with a hazard ratio of 0.79 and a P value of .0689. The numerical difference is relevant because it suggests the PFS advantage was not an isolated artifact of earlier progression assessment, yet the overlap in confidence intervals leaves room for uncertainty about the magnitude of any survival gain.
This is the kind of result that comes up often in advanced oncology: one endpoint crosses the threshold, another trends in the same direction without fully clearing it. In CUP, where post-progression treatment is variable and later lines can dilute a frontline signal, OS is especially difficult to move cleanly. Patients may receive multiple subsequent therapies after progression, and that downstream treatment can blur differences established earlier in the disease course.
Even so, the direction of effect is consistent with the central hypothesis. If a treatment chosen by genomic features slows progression more effectively than a fixed additional course of chemotherapy, the benefit should appear first in PFS and may or may not translate into a larger OS separation depending on subsequent management, tumor heterogeneity, and access to later therapy.
Safety stayed stable across longer follow-up
No new safety signals emerged with the additional follow-up. That matters in CUP, where many patients are already medically fragile and cumulative toxicity from repeated platinum-based therapy can erode function quickly. A molecularly guided strategy has little practical value if it simply exchanges one burden for another; this updated dataset did not show that kind of trade-off.
The abstract does not present a new toxicity signal, and that restraint is important. Without inventing event rates, the available data support a conservative interpretation: the benefit in disease control was not accompanied by an identified late safety penalty during the longer observation period. For clinicians choosing between another empiric chemotherapy cycle and a genomically directed regimen, that balance is central.
Why these results matter at diagnosis, not after the fact
The most actionable implication sits earlier in the pathway. The updated analysis reinforces the idea that comprehensive genomic profiling should be incorporated at initial diagnosis in unfavorable CUP, not deferred until the disease has already advanced through multiple nonspecific treatments. Once metastatic burden increases and performance status declines, the therapeutic window narrows. Genomic information loses value if it arrives too late to shape a treatment decision that the patient can actually tolerate.
CUPISCO also clarifies the role of the molecular tumor board. The therapy was not assigned by algorithm alone; it was investigator-chosen after discussion in that setting. That preserves clinical judgment while still anchoring the decision to molecular data, which is especially relevant when a tumor profile offers several plausible options or when evidence for a specific alteration is limited.
The trial’s updated readout strengthens the case that the biology of an individual CUP tumor can be more informative than the historical category itself. Patients with unfavorable CUP are not a single disease, and the broader label often masks distinct molecular drivers that may be more therapeutically relevant than the unknown primary site. If a genomic strategy can extend progression-free survival without adding new safety concerns, it offers a more defensible way to use precious first-line and post-induction treatment time.
The remaining question is not whether genomic profiling has a role in this disease setting. The open issue is how best to operationalize it across institutions so patients receive profiling early enough to influence care and so actionable findings translate into treatment access rather than just a report in the chart.