Annals of Oncology最新文献

Background: Global inequities in access to cancer diagnostics and treatment contribute to wide variation in cancer mortality-to-incidence ratios (MIRs), a proxy for survival. We aimed to develop an interpretable machine learning framework to quantify country-specific health system contributors to MIR and inform policy prioritization.

Materials and methods: We assembled national MIRs from GLOBOCAN 2022 for 185 countries and health system indicators from multilateral sources, including gross domestic product (GDP) per capita, universal health coverage (UHC) index, radiotherapy centers per population, health spending (%GDP), out-of-pocket expenditure, work force densities (physicians; nurses/midwives; surgical work force), pathology availability, Human Development Index, and gender inequality index. A CatBoost gradient-boosting model was trained with repeated leave-one-country-out cross-validation (10 repeats; 1850 predictions). Nested hyperparameter optimization and strict leakage control were used. Model interpretability employed SHapley Additive exPlanations (SHAP; TreeExplainer) to generate global and country-level feature attributions. SHAP values, model-derived metrics quantifying each factor's contribution to cancer outcomes, were generated. Performance metrics included R², root mean squared error (RMSE), mean absolute error, and Pearson correlation; uncertainty was estimated by bootstrap resampling.

Results: The model showed strong out-of-sample performance [R² = 0.852, 95% confidence interval (CI) 0.801-0.891; RMSE 0.057, 95% CI 0.050-0.064]; correlation between predicted and observed MIRs was r = 0.923 (P = 8.30 × 10^-78). Global SHAP contributions ranked GDP per capita (22.5%), radiotherapy centers per population (15.4%), and UHC index (12.9%) as the leading determinants. Country-specific SHAP profiles revealed substantial heterogeneity in dominant drivers across settings, enabling tailored policy levers (e.g. infrastructure, coverage expansion, or financial protection). An accompanying web interface provides country-level SHAP summaries for decision support.

Conclusions: An explainable machine learning approach accurately predicts national MIRs and decomposes predictions into country-specific health system attributions. While ecological and noncausal by design, the SHAP profiles translate population-level associations into actionable hypotheses for prioritizing investments-highlighting, across many contexts, radiotherapy capacity and UHC expansion as recurrent levers, and underscoring that higher total health spending alone may be insufficient without strategic allocation. Prospective, country-specific evaluations are warranted to test whether targeting model-identified drivers improve cancer outcomes.

Background: Standard treatment of localized muscle-invasive bladder cancer (MIBC) is neoadjuvant chemotherapy (NAC) followed by radical cystectomy (RC), but ∼50% of patients relapse within 2 years. Adjuvant immunotherapy is currently administered based on pathology and clinical assessment in high-risk patients only, potentially resulting in under- and overtreatment.

Patients and methods: TOMBOLA is a Danish, multicenter, open-label, single-arm phase II trial evaluating serial circulating tumor DNA (ctDNA) testing to guide postoperative immunotherapy. Low- and high-risk patients with MIBC (cT2-4aN0-1M0) treated with NAC and RC were monitored postoperatively with tumor-informed ctDNA assays. ctDNA-positive patients initiated atezolizumab for up to 1 year, irrespective of imaging; ctDNA-negative patients received immunotherapy only upon radiographic detection of metastases. The primary endpoint was molecular and radiographic complete response. Key secondary endpoints included recurrence-free survival and overall survival.

Results: In total 192 patients were enrolled, and among 178 assessable patients in the intention-to-treat population (median follow-up 34 months), 104 (58%) were ctDNA-positive within 2 years after RC, 63% within 4 months. The median lead time from ctDNA detection to imaging-confirmed recurrence was 90 days (range -61 to 961 days). Of the ctDNA-positive patients, 84 completed atezolizumab and had scanning and ctDNA analyses available for primary endpoint assessment. Some 60% (50/84) of patients achieved the primary endpoint of complete response. One-year recurrence-free survival was 97% in ctDNA-negative patients and 76% in ctDNA-positive patients. Prespecified biomarker analyses showed that ctDNA status and levels, risk stratification, and immune-related gene expression signatures were associated with both recurrence risk and response to immunotherapy. Treatment was well tolerated with no new safety concerns.

Conclusions: Tumor-informed ctDNA testing after NAC and RC predicts recurrence risk and enables personalized postoperative management in MIBC. TOMBOLA demonstrates that ctDNA-positive, low-risk patients may benefit from early immunotherapy, while ctDNA-negative high-risk patients may safely avoid adjuvant treatment without compromising outcomes.

Background: The optimal duration of immune checkpoint blockade (ICB) in lung cancer remains undefined. Indefinite treatment in long-term responders increases healthcare burden, exposes patients to avoidable toxicities, and is not supported by any clinical and biological rationale or translational data. Prospective strategies to determine the optimal duration of immunotherapy in lung cancer are urgently needed.

Patients and methods: In this retrospective cohort study, 455 patients from 21 nNGM centers with ≥2 years of disease control on first-line ICB-based therapy were grouped into PET/CT-guided discontinuation (cohort A, n=126) or continued ICB without PET/CT (cohort B, n=329), and assessed for overall survival (OS). Matched pre- and post-ICB tumor samples from cohort A patients with persistent or progressive disease were analyzed by comprehensive genomic profiling, histological TIL quantification, and spatial transcriptomics to explore mechanisms of late resistance.

Results: After a median follow-up of 55 months, cohort A showed significantly longer OS (median not reached vs. 82 months; HR 0.35 [0.18-0.67], p = 0.002), despite substantially shorter treatment duration (27 vs. 45 months; p < 0.001). Discontinuation was either PET-driven (A) or resulted from immune-related toxicity, progression, or patients' choice (B). Systematic re-biopsies in cohort A revealed a high incidence of second primary lung cancers (SPLC, 28%). All progression events were managed exclusively with local (ablative) treatments in 53% (A) vs. 17% (B). Post-treatment tumors exhibited features of acquired resistance, whereas SPLC displayed characteristics of primary resistance, including low PD-L1 expression, low TMB, and immunologically cold tumor microenvironments.

Conclusions: A structured discontinuation strategy appears to provide a safe approach for long-term ICB responders, enabling earlier detection of resistance before generalized progression. A confirmatory prospective non-inferiority randomized trial within the nNGM is underway.

Background: The TROPION-Breast01 study (NCT05104866) demonstrated statistically significant and clinically meaningful improvement in progression-free survival (PFS) by blinded independent central review (BICR) with the trophoblast cell-surface antigen 2-directed antibody-drug conjugate (ADC) datopotamab deruxtecan (Dato-DXd) versus investigator's choice of chemotherapy (ICC) in patients with previously treated, inoperable/metastatic hormone receptor (HR)-positive, human epidermal growth factor receptor 2 (HER2)-negative breast cancer. In this article, we report results from the final overall survival (OS) analysis.

Patients and methods: Patients with inoperable/metastatic HR-positive/HER2-negative breast cancer, who had disease progression on endocrine therapy and for whom endocrine therapy was unsuitable, and had received one to two prior lines of chemotherapy in the inoperable/metastatic setting were randomly assigned in a 1 : 1 ratio to Dato-DXd (6 mg/kg every 3 weeks) or ICC (eribulin/capecitabine/vinorelbine/gemcitabine). Dual primary endpoints were PFS by BICR and OS.

Results: At data cut-off, median follow-up was 22.8 months. OS for the Dato-DXd versus ICC arm did not reach statistical significance (hazard ratio 1.01, 95% confidence interval 0.83-1.22, P = 0.9445). Use of ADCs (trastuzumab deruxtecan and sacituzumab govitecan) as subsequent therapy was imbalanced: 12.3% in the Dato-DXd arm versus 24.0% in the ICC arm. Secondary efficacy endpoints (PFS by investigator assessment, objective response rate, duration of response, disease control rate at 12 weeks, time to first and second subsequent therapy or death, and time to second progression or death) continued to favor Dato-DXd at this final analysis. The overall safety profile of Dato-DXd remained favorable compared with ICC, and no new safety signals were observed with longer follow-up.

Conclusions: TROPION-Breast01 met its dual primary endpoint of PFS by BICR. While there was no statistically significant improvement in the dual primary endpoint of OS with Dato-DXd versus ICC, subsequent ADC treatment may have affected OS results. The totality of efficacy and safety data supports Dato-DXd as a new treatment option for patients with previously treated, inoperable/metastatic HR-positive/HER2-negative breast cancer.

Background: Addition of pembrolizumab to neoadjuvant chemotherapy (NACT) is a standard-of-care treatment in non-metastatic triple-negative breast cancer (TNBC). Trials employing reduced doses of immune checkpoint inhibitors have yielded encouraging activity in several tumor types. Whether low-dose pembrolizumab enhances pathological complete response when added to NACT in TNBC remains uncertain.

Patients and methods: A phase II, open-label, randomized controlled study was conducted at a tertiary cancer center in New Delhi, India. Eligible participants had previously untreated stage II-III TNBC and lacked access to standard-dose pembrolizumab. Patients were randomly assigned (1 : 1) to receive dose-dense NACT-four cycles of doxorubicin-cyclophosphamide followed by four cycles of paclitaxel with or without a low dose of pembrolizumab (50 mg every 6 weeks for three cycles). The primary endpoint was pathological complete response.

Results: Between February 2024 and February 2025, 157 patients were enrolled. Baseline characteristics were comparable between study arms. Surgery was completed in 152 patients. In the intention-to-treat population, pathological complete response was achieved in 53.8% [90% confidence interval (CI) 43.9% to 63.5%] of patients receiving low-dose pembrolizumab + NACT versus 40.5% (90% CI 31.1% to 50.4%) with NACT alone, corresponding to an absolute difference of 13.3% (90% CI 0.3% to 26.3%, one-sided P = 0.047). Among those who underwent surgery, the respective pathological complete response rates were 56.7% and 41.0% (absolute difference 15.7%, one-sided P = 0.031). Grade ≥3 toxicities occurred in 50% of patients in the low-dose pembrolizumab arm and in 59.5% in the control arm.

Conclusions: Addition of low-dose immunotherapy to NACT led to a statistically significant increase in pathological complete response rates, and the benefit appears to be numerically similar to that seen in the KEYNOTE-522 trial. In settings where access to the standard regimen is restricted, a low-dose approach may offer a feasible and cost-effective treatment option for patients with TNBC.

Background: Anti-epidermal growth factor receptor (EGFR) drug rechallenge could be of therapeutic value in a subgroup of refractory metastatic colorectal cancer (mCRC).

Patients and methods: The randomized phase II CAVE-2 GOIM trial compared two rechallenge regimens in RAS/BRAF wild-type mCRC (cetuximab monotherapy or in combination with avelumab). Patients were selected according to baseline plasma circulating tumor DNA (ctDNA) comprehensive genomic profiling (CGP) by FoundationOne Liquid CDx. Primary endpoint was overall survival (OS).

Results: From August 2022 to December 2024, 156 out of 328 screened patients were randomly assigned in a 2 : 1 ratio to cetuximab plus avelumab (arm A, 104 patients) or cetuximab (arm B, 52 patients). Median progression-free survival (mPFS) was 5.3 months [95% confidence interval (CI) 4.3-6 months] in arm A versus 4.3 months (95% CI 3.5-5.5 months) in arm B [hazard ratio (HR) 0.78, 95% CI 0.55-1.10, P = 0.158]. Median OS (mOS) was 14.8 months (95% CI 12.1-18.3 months) in arm A versus 12.9 months (95% CI 11 months-not evaluable) in arm B (HR 1.00, 95% CI 0.65-1.52, P = 0.983). A pre-planned exploratory analysis evaluated the impact of genomic pathogenic variants on therapeutic efficacy in the EGFR pathway. In the whole study population, 124/156 patients had tumors without any genomic alteration in KRAS, NRAS, BRAF, EGFR extracellular domain, PIK3CA exon 20, MAP2K1, AKT1, MET, PTEN, and ERBB2 ('negative hyperselection'). These patients had significantly improved mPFS [5.35 months (95% CI 4.4-5.9 months) versus 3.65 months (95% CI 2.8-4.8 months); HR 0.62, 95% CI 0.42-0.92, P = 0.017] and mOS [15.0 months (95% CI 12.6-19.9 months) versus 11.1 months (95% CI 8.6-15.6 months); HR 0.61, 95% CI 0.39-0.97, P = 0.037] compared with patients having at least one pathogenic variant ('positive hyperselection'). Similarly, improved objective response rate, mPFS, and OS were observed for each treatment arm in patients with 'negative hyperselection'.

Conclusion: Addition of avelumab does not increase efficacy of cetuximab rechallenge. Liquid biopsy CGP identifies patients who benefit from anti-EGFR rechallenge supporting its implementation in the continuum of care of mCRC.