Why This Study Matters
Newly diagnosed, unfavorable, nonsquamous cancer of unknown primary (CUP) remains a challenge to treat. CUPISCO was designed to test whether a molecularly guided therapy (MGT) strategy, informed by comprehensive genomic profiling (CGP) and a molecular tumor board, could improve on platinum-based chemotherapy. In the authors' words, it was the first randomized study of MGT, to their knowledge, to report positive results in this tumor entity. This paper reports the updated and final efficacy and safety analyses for category 1 patients.
Study Design
All patients first received three cycles of platinum-based induction chemotherapy (investigator's choice). Category 1 (CAT1) patients who reached disease control (complete or partial response, or stable disease) were randomly assigned 3:1 to MGT (experimental arm) or to at least three additional cycles of chemotherapy (control arm). In the experimental arm, patients with an actionable molecular profile received targeted therapy or atezolizumab monotherapy; patients without one received atezolizumab plus chemotherapy. Category 2 patients, who had progressive disease after induction, were reported separately.
The primary end point was investigator-assessed progression-free survival (PFS, RECIST v1.1). Overall survival (OS) and safety were secondary end points. Stratification factors were gender and chemotherapy response. Subgroup analyses were post hoc, and safety data are descriptive.
Patient Population
1,505 patients screened; 636 enrolled; 436 randomly assigned (326 MGT; 110 chemotherapy) at a 3:1 ratio; median follow-up 37.0 months (range, 0.0–67.8).
Of the 636 enrolled, 573 (90%) completed induction and 438 reached disease control, of whom 436 were randomly assigned. Baseline characteristics were reported as well balanced. The data cutoff was December 6, 2024 (the primary analysis cutoff was February 14, 2023, with 24.1 months median follow-up). Number of sites, countries and dosing details: Not specified in source.
Primary Endpoint Results
| Analysis | MGT | Chemotherapy | HR (95% CI) | P |
|---|---|---|---|---|
| Updated median PFS | 6.1 mo (4.7–6.5) | 4.4 mo (4.2–6.4) | 0.75 (0.59–0.95) | .0177 |
| Primary analysis PFS | 6.1 mo (4.7–6.5) | 4.4 mo (4.1–5.6) | 0.72 (0.56–0.92) | .0079 |
| Updated median OS | 15.2 mo (13.9–18.4) | 12.8 mo (9.9–15.4) | 0.79 (0.61–1.02) | .0689 |
| Primary analysis OS | 14.7 mo (13.3–17.3) | 11.0 mo (9.7–15.4) | Not specified in source | Not specified in source |
The authors state that the updated results were consistent with the primary analysis and show a stable PFS benefit. OS trends favored MGT, but CUPISCO was not designed or powered to test an OS effect. Detecting a 30% reduction in the hazard with 80% power would need more than 330 OS events; 304 had occurred at the final cutoff.
Subgroup Analyses
Subgroups were defined by whether the patient had an actionable molecular profile. These analyses were post hoc, and the authors state that statistical significance was not evaluated in them. Both comparisons below use the safety-evaluable population for CAT1 patients (per Figures 2C and 3C).
| Subgroup | Endpoint | Experimental arm | Control arm | HR (95% CI) |
|---|---|---|---|---|
| Actionable profile (MGT vs chemotherapy) | Median PFS | 8.2 mo (5.0–9.3) | 5.5 mo (4.0–6.9) | 0.62 (0.41–0.93) |
| No actionable profile (atezolizumab + chemotherapy vs chemotherapy) | Median PFS | 5.5 mo (4.5–6.4) | 4.5 mo (4.2–6.3) | 0.83 (0.60–1.15) |
| Actionable profile | Median OS | 17.8 mo (13.6–23.4) | 14.5 mo (10.5–20.5) | 0.75 (0.48–1.17) |
| No actionable profile | Median OS | 14.5 mo (11.9–17.4) | 11.0 mo (9.7–17.8) | 0.87 (0.61–1.23) |
The authors note CUPISCO was not designed to assess the benefit of adding atezolizumab to chemotherapy in patients without an actionable target, so firm conclusions cannot be drawn in that subgroup.
Safety Profile
| Safety-evaluable population | MGT (n = 312) | Chemotherapy (n = 102) |
|---|---|---|
| ≥1 AE | 297 (95.2%) | 90 (88.2%) |
| ≥1 AE with fatal outcome | 15 (4.8%) | 0 |
| ≥1 treatment-related AE | 269 (86.2%) | 85 (83.3%) |
| ≥1 grade 3–5 AE | 205 (65.7%) | 44 (43.1%) |
| ≥1 treatment-related grade 3–5 AE | 153 (49.0%) | 37 (36.3%) |
| ≥1 serious AE | 119 (38.1%) | 15 (14.7%) |
| ≥1 treatment-related serious AE | 55 (17.6%) | 7 (6.9%) |
| ≥1 AE leading to withdrawal from treatment | 84 (26.9%) | 17 (16.7%) |
No major safety signals were raised, and the profile of each therapy remained unchanged from the primary analysis.
— Krämer et al., J Clin Oncol 2026, Discussion
The authors state that longer exposure in the MGT group led to higher rates of related serious AEs, grade 3–5 AEs, and AEs leading to treatment withdrawal.
Interpretation and Broader Context
The authors conclude that additional follow-up was consistent with the primary analysis and that CGP with subsequent MGT improved outcomes over standard platinum-based chemotherapy. They highlight CGP in the initial diagnostic workup to guide MGT or immunotherapy for unfavorable nonsquamous CUP, and suggest that chemotherapy plus immunotherapy combinations warrant further study. They also advise that novel therapies in a clinical trial context should be considered for all patients without an actionable molecular target.
Limitations
The OS difference was not statistically significant, and the trial was not designed or powered for OS; detecting a 30% reduction in the hazard with 80% power would need more than 330 events, and 304 had occurred. The subgroup analyses were post hoc, and the authors state that statistical significance was not evaluated in them. CUPISCO was not designed to assess the benefit of adding atezolizumab to chemotherapy in patients without an actionable target, so firm conclusions cannot be drawn in that subgroup. OS may have been affected by later-line therapies, including targeted treatments in the control arm (information not available for all patients). The primary end point was investigator-assessed, safety data are descriptive, and the experimental arm's longer treatment exposure was associated with higher rates of related serious and grade 3–5 adverse events.