Representation of classical and rare head and neck cancers in basket clinical trials: A pooled analysis.

Abstract

e18137 Background: Basket clinical trials (BCTs) aim to expand access to novel therapies across tumor types, including rare cancers. Head and neck cancers (HNC) comprise biologically heterogeneous entities with marked differences in incidence, etiology, and therapeutic sensitivity. How these factors influence specific representation of classical squamous cell carcinoma (HNSCC) and rare HN histotypes in BCTs remains poorly characterized. Methods: We conducted a systematic review of BCTs published between April 2018 and December 2024. BCTs were classified by biomarker (BM) selection: none, single, or multiple. Cancer-specific crude incidence rates (IR per 100,000 persons; 2022) were derived from the Global Cancer Observatory and harmonized across the EU and US (both sexes, age >20 years). Associations between IR and BCT enrollment were assessed using semi-log and log-log linear models. HNC representation was evaluated for HNSCC, nasopharyngeal carcinoma (NPC), salivary gland carcinoma (SGC). Adenoid cystic carcinoma (AdCC) was analyzed descriptively using literature-derived IR. Observed-to-expected (O/E) enrollment ratios (ER) were calculated comparing observed BCT accrual with expected numbers based on subtype-specific incidence-weighted counts. Drug class (immunotherapy [IO], targeted therapy [TT], antibody–drug conjugates [ADC]) and geographic distribution were annotated. Results: Among 137 screened publications, 26 BCTs enrolling 2,488 patients across 372 centers (57% US, 29.3% EU), met the inclusion criteria: 8 no-BM, 12 single-BM and 6 multi-BM. Across 25 cancer types, IR was associated with enrollment in both semi-log (β ≈ 69.7, 95%CI 26.8–112.5, p=0.003, R² ≈ 0.34) and log–log models (β ≈ 0.39, 95%CI 0.08–0.69, p=0.015, R² ≈ 0.24), indicating a robust but not-proportional relationship. Overall, 134 HNC patients (5.3%) were included: 68 HNSCC, 34 NPC, 19 SGC, 13 AdCC. HNSCC patients were included in 11 BCTs (6 TT-, 3 IO-, 2 ADC-based, O/E ER 0.6), 69% requiring a single BM. Most NPC patients (88%) were enrolled in 2 no-BM IO trials (O/E ER 8.8), while 89.5% of SGC accrual occurred in TT- or ADC-based single-BM trials (O/E ER 2.2). AdCC patients were enrolled in 6 no-BM, 3 single-BM, 4 multi-BM trials, showing the highest O/E ER (25.9). Neither NPC nor SGC appeared in multi-BM trials. When contextualized by IR in comparison with other cancers, HNSCC was under-represented in the semi-log but well-represented in the log-log analyses. Conclusions: Representation of HNC in contemporary BCT is highly heterogeneous. Patients with rare entities such as NPC and SGC achieve proportional enrollment, and AdCC is well represented, whereas patients with HNSCC experience limited absolute accrual. These findings suggest that other factors - e.g. tumor biology, therapeutic mechanism, clinical referral to specialized facilities - rather than incidence alone, influence the access of HNC patients to BCTs.