The Lancet August 10, 2024

Molecularly guided therapy versus chemotherapy after disease control in unfavourable cancer of unknown primary (CUPISCO): an open-label, randomised, phase 2 study

Alwin Krämer, Tilmann Bochtler, Chantal Pauli, et al.

Bottom Line

In patients with unfavorable cancer of unknown primary who achieved disease control after induction chemotherapy, comprehensive genomic profiling followed by molecularly guided therapy significantly improved progression-free survival compared to continued platinum-based chemotherapy.

Key Findings

1. The primary endpoint of progression-free survival (PFS) was significantly prolonged in the molecularly guided therapy (MGT) arm, with a median PFS of 6.1 months compared to 4.4 months in the standard chemotherapy arm (HR 0.72; 95% CI, 0.56–0.92; P=0.0079).
2. Overall survival (OS) data were immature at the primary data cutoff, demonstrating a median OS of 14.7 months for MGT versus 11.0 months for chemotherapy, which did not reach statistical significance (HR 0.82; 95% CI, 0.62–1.09; P=0.1779).
3. The objective response rate (ORR) was significantly improved in the MGT group, showing an absolute increase of 9.6% (95% CI, 2.4–16.8; P=0.0141) over continued chemotherapy.
4. The PFS benefit was particularly pronounced in patients with an actionable molecular profile who received matched targeted therapies or atezolizumab, achieving a median PFS of 8.1 months compared to 4.7 months on chemotherapy (HR 0.65; 95% CI, 0.42–0.99).

Study Design

Design
Phase II RCT
Open-Label
Sample
436
Patients
Duration
24.1 mo
Median
Setting
Multinational
Population Patients with newly diagnosed, unfavorable, non-squamous cancer of unknown primary (CUP) who achieved disease control after 3 cycles of induction platinum-based chemotherapy
Intervention Molecularly guided therapy (MGT), selected from multiple targeted or immunotherapy regimens based on comprehensive genomic profiling and molecular tumor board advice
Comparator Continuation of standard platinum-based induction chemotherapy for at least 3 additional cycles
Outcome Investigator-assessed progression-free survival (PFS)

Study Limitations

The open-label design introduces a risk of investigator and patient bias, particularly concerning subjective assessments like progression-free survival and response rates.
The trial restricted randomization to patients who had already achieved disease control following three cycles of induction chemotherapy, excluding early progressors who constitute a significant and extremely poor-prognosis subset of CUP patients.
The broad inclusion of multiple distinct targeted therapies and immunotherapies as a single 'molecularly guided therapy' arm makes it difficult to ascertain which specific regimens drove the overall efficacy.
At the primary data cutoff, overall survival data were relatively immature, limiting definitive conclusions regarding the long-term survival impact of the intervention.

Clinical Significance

The CUPISCO trial provides the first prospective, randomized evidence that a biomarker-driven precision oncology framework can alter the natural history of unfavorable cancer of unknown primary (CUP). By demonstrating that comprehensive genomic profiling (CGP) successfully matches patients to targeted therapies that outperform empiric chemotherapy, it establishes early CGP and subsequent molecularly guided therapy as a new standard of care for this historically intractable disease.

Historical Context

Cancer of unknown primary (CUP) has historically accounted for 2-5% of all newly diagnosed malignancies and carries a notoriously dismal prognosis, especially in the 'unfavorable' subset. For decades, standard treatment relied on empiric, broad-spectrum platinum-based chemotherapy. Earlier attempts to improve outcomes by using tissue-of-origin gene expression profiling to direct site-specific therapies (such as in the GEFCAPI 04 trial) failed to demonstrate superiority over standard chemotherapy. CUPISCO represents a paradigm shift, proving that instead of attempting to predict the tissue of origin, directly matching comprehensive genomic alterations to targeted treatments yields superior clinical outcomes.

Guided Discussion

High-yield insights from every perspective

Med Student
Medical Student

How does the biological concept of cancer of unknown primary (CUP) differ from traditional metastasis, and why does comprehensive genomic profiling (CGP) provide a more rational therapeutic target than site-specific histology in unfavorable CUP?

Key Response

CUP represents a heterogeneous group of metastatic tumors where the primary site cannot be identified. Traditional oncology targets the tissue of origin. Since unfavorable CUP lacks a known origin and has a poor prognosis with empiric chemotherapy, CGP shifts the paradigm to treating the specific molecular driver (e.g., BRAF, ERBB2 mutations) rather than the anatomical origin, illustrating the shift to agnostic precision oncology.

Resident
Resident

The CUPISCO trial required patients to achieve disease control with 3 cycles of induction chemotherapy before randomization to molecularly guided therapy. Clinically, why is initial empiric platinum-based chemotherapy still standard before transitioning to targeted therapy in these patients?

Key Response

In unfavorable CUP, patients often present with rapidly progressing, symptomatic disease. Waiting for comprehensive genomic profiling results can take weeks, risking rapid clinical deterioration. Initial empiric platinum-based chemotherapy serves to gain rapid disease control, buying time for molecular profiling and ensuring patients are fit enough for targeted therapy once results return.

Fellow
Fellow

Given that comprehensive genomic profiling in CUPISCO was likely performed on tissue obtained before induction chemotherapy, how might clonal evolution and tumor heterogeneity induced by platinum-based chemotherapy impact the efficacy of the subsequent molecularly guided therapy?

Key Response

Platinum chemotherapy induces DNA damage and selective pressure, potentially leading to clonal evolution or resistance mutations. If targeted therapy is chosen based on pre-chemotherapy biopsy results, the dominant molecular driver may have shifted. Fellows must consider whether repeat biopsies or circulating tumor DNA (ctDNA) analysis at randomization might better reflect the current actionable genomic landscape.

Attending
Attending

While CUPISCO demonstrates a PFS benefit for MGT in unfavorable CUP after induction chemotherapy, how should we balance the high cost and access barriers of comprehensive genomic profiling and off-label targeted therapies with the modest survival gains in real-world practice?

Key Response

The transition from trial to practice requires pragmatism. Attendings must weigh the statistical significance of PFS against absolute clinical benefit, financial toxicity, and logistical hurdles of insurance approval. A key teaching point is that actionable mutations do not guarantee dramatic responses, requiring shared decision-making regarding quality of life versus pursuing every molecular target.

Scholarly Review

Critical appraisal through the lens of expert reviewers and guideline development

PhD
PhD

The CUPISCO trial utilized an open-label, randomized phase 2 design with progression-free survival (PFS) as the primary endpoint. From a methodological standpoint, how does the open-label nature introduce bias in assessing PFS, and what statistical methods could robustly validate PFS as a surrogate for overall survival in this highly heterogeneous population?

Key Response

Open-label designs are prone to assessment bias, as knowledge of the treatment arm can influence scan interpretation or timing of progression, making blinded independent central review critical. Furthermore, validating PFS as a surrogate for OS in CUP is statistically challenging due to molecular heterogeneity and post-progression crossover, requiring trial-level meta-regression or copula-based models to establish true correlation.

Journal Editor
Journal Editor

The MGT arm in CUPISCO is essentially a basket of different targeted therapies tailored to individual genomic alterations. As an editor, how do you critically appraise a trial where the intervention arm is a heterogeneous strategy, and what specific supplementary data must be demanded?

Key Response

A major threat to validity in strategy trials is that a statistically significant overall benefit might be driven by a small subset of patients (e.g., those with BRAF or NTRK alterations) who have exceptional responses. An editor must demand detailed forest plots of PFS by specific molecular pathway and agent to determine if the strategy is universally beneficial or merely masking a few highly effective outlier treatments.

Guideline Committee
Guideline Committee

Current ESMO and NCCN guidelines recommend empiric platinum-based chemotherapy for unfavorable CUP and suggest comprehensive genomic profiling primarily for trial eligibility. Based on the CUPISCO phase 2 results, is there sufficient evidence to upgrade the recommendation for CGP and subsequent molecularly guided therapy to a standard-of-care Level I recommendation?

Key Response

While CUPISCO provides strong prospective Phase 2 data (Level II evidence) that MGT improves PFS over maintenance chemotherapy, guideline committees typically require Phase 3 data with an Overall Survival benefit for a Level I recommendation. The committee would likely upgrade CGP to a strongly recommended tier (e.g., NCCN Category 2A) to guide maintenance but may hesitate to make it an absolute standard without definitive OS data and cost-effectiveness analyses.

Clinical Landscape

Noteworthy Related Trials

2015

SHIVA Trial

n = 195 · Lancet Oncol

Tested

Molecularly targeted therapy based on tumor profiling

Population

Patients with advanced cancer refractory to standard care

Comparator

Treatment at physician's choice

Endpoint

Progression-free survival

Key result: The use of molecularly targeted therapy outside approved indications did not improve progression-free survival compared to physician's choice of treatment.
2019

WJTOG1201 Trial

n = 130 · J Clin Oncol

Tested

Site-specific therapy dictated by gene expression profiling

Population

Patients with untreated unfavorable carcinoma of unknown primary

Comparator

Empirical chemotherapy (Paclitaxel and Carboplatin)

Endpoint

1-year survival rate

Key result: Site-specific treatment based on gene expression profiling failed to significantly improve 1-year survival compared to standard empirical chemotherapy.
2020

NCI-MATCH Trial

n = 6,000 · J Clin Oncol

Tested

Targeted therapies matched to specific actionable genetic mutations

Population

Patients with relapsed or refractory solid tumors and lymphomas

Comparator

Historical controls (Phase 2 basket trial design)

Endpoint

Objective response rate

Key result: Matching targeted therapies to specific molecular alterations across various tumor types yielded meaningful clinical responses in molecularly defined patient subsets.

Tailored to your role

Want this tailored to you?

Add your specialty or training stage to get role-specific takeaways and more questions.

Personalize this analysis