Annals of Oncology最新文献

Background: We provided updated cancer mortality estimates for 2026 in the European Union (EU) and its five most populous countries, with a focus on lung cancer.

Materials and methods: Cancer death certifications and population data were obtained from the World Health Organization and United Nations databases. For the EU, France, Germany, Italy, Poland, Spain, and the UK, we derived data for all cancers combined and major cancer sites since 1970. Linear regression models, based on the most recent age-specific trends identified by Poisson joinpoint regression, were used to estimate deaths in 2026. The number of averted deaths between 1989 and 2026 was computed by applying the 1988 peak rate to subsequent populations.

Results: For 2026, we estimated ∼1 230 000 EU cancer deaths, corresponding to age-standardised rates of 114.1/100 000 males (-7.8% versus 2020-2022) and 74.7/100 000 females (-5.9%). In the EU countries and the whole EU, favourable trends are predicted for most major cancers, except female pancreatic cancer. In the UK, predicted rates are also favourable, except female colorectal cancer. Lung cancer mortality continues to decrease markedly among males, while we predicted a levelling off of rates, around 12.5/100 000, among females in all considered countries and the whole of EU, except for Spain (+2.4%). Among females, lung cancer mortality declines are confined to those aged <65 years, while unfavourable trends continued in older age groups. Around 7.3 (5.0 in males, 2.3 in females) million total cancer deaths have been avoided in the EU since the peak observed in 1988. The corresponding figure for lung cancer is 1.8 million among males, while no averted deaths were recorded among females.

Conclusion: Lung cancer mortality predictions for 2026 indicate a levelling off among EU females, with age- and country-specific differences. Mortality trends in ASRs for most cancers remain favourable in the EU and the UK, though the absolute number of cancer deaths is not declining due to population ageing.

Background: Advanced/metastatic soft tissue sarcomas (STSs) remain an unmet clinical need. We previously reported the feasibility and preliminary activity of trabectedin-olaparib combination in patients with advanced STS progressing after anthracycline-based regimens.

Patients and methods: In this investigator-initiated, open-label, phase II randomized trial, adult patients with advanced STS progressing after one or more prior lines of therapy, including at least one anthracycline-based regimen, were randomly assigned 1 : 1 to trabectedin 1.1 mg/m² every 21 days (q21d) intravenous (i.v.) plus olaparib tablets 150 mg twice a day, or trabectedin 1.5 mg/m² q21d i.v. Randomization stratified patients by histology (L-sarcoma, i.e. leiomyosarcoma and liposarcoma versus non-L-sarcoma) and number of prior therapies (one versus two or more). The primary endpoint was progression-free survival (PFS) rate at 6 months (PFS6m) per RECIST1.1. Secondary endpoints included PFS, overall survival (OS), RECIST1.1 overall response rate (ORR), and safety. Exploratory endpoints encompassed biomarker/molecular analyses.

Results: Between 25 May 2020 and 2 November 2022, 130 patients were enrolled at 13 Italian Sarcoma Group centers (81 female; 67 L-sarcoma; 93 one prior line). With a median follow-up of 37.4 months, PFS6m and median PFS were 32% [95% confidence interval (CI) 22% to 46%] and 3.9 months (95% CI 2.7-5.2 months) with trabectedin-olaparib versus 28% (95% CI 19% to 42%) and 2.9 months (95% CI 2.2-3.6 months) with trabectedin [hazard ratio (HR) 0.722, 95% CI 0.501-1.041, P = 0.081]. Among 126 assessable patients, ORR was 12.7% (95% CI 6.1% to 22.7%) versus 7.9% (95% CI 3.0% to 16.7%), respectively (odds ratio 1.60, 95% CI 0.50-5.16, P = 0.43). In the uterine leiomyosarcoma subgroup, 12-month PFS was 42.9% with trabectedin-olaparib versus 0% with trabectedin. PARP1 expression significantly correlated with improved PFS with trabectedin-olaparib (PFS6m and median PFS were 41.5% and 4.3 months versus 27.8% and 2.5 months; HR 0.537, 95% CI 0.337-0.855, P = 0.009). Grade ≥3 hematological toxicities were significantly more frequent with trabectedin-olaparib.

Conclusions: Although trabectedin-olaparib combination reached the prespecified threshold for statistical significance for PFS (P < 0.10), the benefit was marginal in the all-comers STS population. Nonetheless, patients affected by PARP1-expressing STS and uterine leiomyosarcoma derived substantial benefit from the combination, supporting further histology- and biomarker-driven investigation in these settings.

Background: Manual abstraction of real-world data (RWD) from unstructured health records (HRs) remains resource intensive, error prone, and highly variable across institutions. Large language models (LLMs) offer a scalable alternative, but their performance in multicenter oncology settings is not fully validated.

Patients and methods: We conducted a multicenter study within the French Large & Unified Cancer Cohort (LUCC) consortium to compare the accuracy of artificial intelligence (AI)-based data extraction against manual abstraction by clinical research professionals. A fine-tuned LLM was applied to de-identified unstructured HRs in PDF format to extract 31 variables from lung cancer patients across 10 centers. Ground truth was defined as concordant values across sources, with discrepant cases adjudicated by a blinded expert. The primary endpoint was the extraction error rates. Secondary endpoints included per-variable performance, interinstitutional variability, F1-score for multiple-choice variables, added value of hybrid AI-human workflows, and survival analyses.

Results: Among 10 327 patients with AI-based extraction, 311 were included in the test cohort. Across 8708 datapoints for 28 variables with only one correct answer, the LLM achieved a 7.0% error rate, outperforming manual abstraction (14.2%, P <0.001). The F1-scores of three multiple-choice variables were superior (gene alterations 0.97 versus 0.86, comorbidities 0.86 versus 0.76, metastatic sites 0.71 versus 0.69). Interinstitutional variance was lower with AI (0.12% versus 0.39%). A hybrid approach with targeted human review of 30% of low-confidence AI outputs further decreased error rates to 4.4%. Survival analyses based on AI-extracted data closely matched ground truth, with similar median overall and progression-free survival.

Conclusions: In a multicenter setting, our AI pipeline yielded lower error rates and greater consistency than manual abstraction. These findings support the feasibility of next-generation, AI-enabled multicenter studies to generate high-quality RWD at scale, with potential applicability in prospective clinical trials.

Background: Perioperative immunotherapy, particularly anti-programmed cell death protein 1 therapy, combined with neoadjuvant chemotherapy has emerged as one of the standard-of-care options for resectable non-small-cell lung cancer (NSCLC). We report the final analysis of RATIONALE-315, a randomized, double-blind, phase III trial evaluating the efficacy and safety of perioperative tislelizumab plus neoadjuvant chemotherapy versus neoadjuvant chemotherapy alone in patients with resectable, stage II-IIIA NSCLC.

Patients and methods: Adult patients in China were randomized (1 : 1) to receive either perioperative tislelizumab or placebo in combination with neoadjuvant chemotherapy. The dual primary endpoints were event-free survival (EFS) and major pathological response, assessed by blinded independent central review. The secondary endpoints included pathological complete response, overall survival (OS), disease-free survival, and safety.

Results: In total, 453 patients were randomized (226 to tislelizumab and 227 to placebo). At the final analysis (median study follow-up of 38.5 months), patients in the tislelizumab group experienced statistically significantly improved OS versus those in the placebo group {hazard ratio (HR) 0.65 [95% confidence interval (CI) 0.45-0.93]; P = 0.009}. The median OS was not reached in either group. The 36-month OS rate was 79.3% in the tislelizumab group versus 69.3% in the placebo group. The median EFS was not reached in the tislelizumab group versus 30.6 months in the placebo group [HR 0.58 (95% CI 0.43-0.79)]. Survival benefits were generally consistent across subgroups. The safety profile of tislelizumab plus chemotherapy was tolerable and consistent with known profiles of the individual therapies.

Conclusions: Neoadjuvant tislelizumab plus chemotherapy and adjuvant tislelizumab demonstrated a statistically significant, clinically meaningful OS benefit and sustained, clinically meaningful EFS improvement compared with neoadjuvant chemotherapy, with a tolerable safety profile in patients with resectable stage II-IIIA NSCLC. These results support the use of this regimen in this patient population.

Clinical trial number: NCT04379635.

Background: Thymic epithelial tumors (TETs) are rare malignancies that pose significant diagnostic challenges due to their heterogeneous histological patterns and substantial interobserver variability in classification. Despite standardized World Health Organization (WHO) classification criteria, diagnostic concordance remains suboptimal, particularly in nonexpert settings, where second-opinion reviews lead to diagnostic reclassification in up to 57% of cases. Deep learning may offer a tool to reduce diagnostic variability and improve the consistency of histological classification.

Materials and methods: We trained a deep learning-based model using hematoxylin-eosin (H&E) whole-slide images from The Cancer Genome Atlas as a training dataset. The model incorporated a novel hierarchical loss function designed to reflect clinically relevant tumor groupings based on treatment strategies and patient outcomes. We validated the model on 112 consecutive cases from the University of Chicago, with diagnoses confirmed by an expert thoracic pathologist. Model performances were evaluated using both a three-group hierarchical scheme and the six-class WHO classification.

Results: In the clinically relevant hierarchical three-group classification (As: A + AB; Bs: B1 + B2 + B3; and thymic carcinoma), the model achieved an accuracy of 91.1% with Cohen's κ = 0.859, indicating almost perfect agreement. In the six-class classification (A, AB, B1, B2, B3, and thymic carcinoma), the accuracy was 77.7% with κ = 0.716. The model demonstrated 100% sensitivity and 94.6% accuracy for thymic carcinoma detection. Notably, 60% of misclassifications occurred within the same clinical management group, thereby limiting their impact on therapeutic decision making.

Conclusion: This deep learning model demonstrates strong potential as a diagnostic tool for TETs classification, particularly in settings with limited thoracic pathology expertise. The high sensitivity for thymic carcinoma detection and robust performance across different tissue processing conditions suggest its clinical applicability for improving diagnostic consistency and supporting pathological decision making in both specialized and nonspecialized settings.

Background: Intravenous (IV) paclitaxel requires prolonged infusion and is associated with hypersensitivity reactions and peripheral neuropathy. This study evaluated the efficacy and safety of DHP107, a novel oral formulation of paclitaxel, compared to IV paclitaxel in patients with HER2-negative, recurrent, or metastatic breast cancer.

Patients and methods: This multinational, multicenter, open-label, randomized phase III trial evaluated the non-inferiority of DHP107 compared to IV paclitaxel, with a predefined non-inferiority margin of 1.33. Eligible patients had recurrent or metastatic breast cancer and had received no prior chemotherapy in the metastatic setting. Patients were randomized to receive either DHP107 (200 mg/m² orally twice daily on days 1, 8, and 15) or paclitaxel (80 mg/m² intravenously on days 1, 8, and 15) in a 28-day cycle. The primary endpoint was investigator-assessed progression-free survival (PFS) in the per-protocol set (PPS). Secondary endpoints included independent central review-assessed PFS, overall survival (OS), tumor response, quality of life (QoL), and safety.

Results: Between January 2018 and December 2023, 549 patients were randomized to receive either DHP107 (n=277) or paclitaxel (n=272); 519 patients were included in the PPS. DHP107 demonstrated non-inferiority in PFS with a median PFS of 10.0 months compared to 8.5 months for paclitaxel (hazard ratio [HR], 0.869; 95% confidence interval [CI], 0.707-1.068). The median OS was 32.6 months for DHP107 and 31.8 months for paclitaxel (HR 0.967; 95% CI, 0.762-1.227). DHP107 was associated with higher rates of neutropenia, febrile neutropenia, nausea, diarrhea, and vomiting, whereas peripheral neuropathy and hypersensitivity reactions were more common with paclitaxel. No treatment related-death occurred in the DHP107 group, and in one patient (0.4%) in the paclitaxel group. Quality of life was comparable between the two groups.

Conclusions: DHP107 demonstrated non-inferior efficacy compared to IV paclitaxel, with a manageable safety profile, supporting its use as an effective and convenient alternative in HER2-negative breast cancer.

Background: Homozygous loss of MTAP (methylthioadenosine phosphorylase) occurs in approximately 15% of cancers and leads to partial PRMT5 inhibition, creating a selective vulnerability to PRMT5 inhibitors. The frequency and therapeutic relevance of MTAP loss in oncogene-driven non-small cell lung cancers (NSCLCs) remain underexplored.

Methods: MTAP status was assessed by next-generation sequencing (NGS) or immunohistochemistry (IHC) in >13,000 NSCLC samples in four cohorts. Prevalence was assessed across oncogenic drivers and temporal dynamics in pre- and post-treatment biopsies. Clinical outcomes were analyzed in EGFR- and ALK-rearranged NSCLC treated with first-line osimertinib and alectinib, respectively. The MTA-cooperative PRMT5 inhibitor BMS-986504 was tested alone and in combination with targeted therapies (TT) in MTAP-deleted models in vitro, ex vivo and in vivo.

Results: MTAP loss was frequent in ALK-rearranged (27% and 33% by NGS; 36% and 45% IHC), RET-rearranged (18.5% and 26% by NGS; 35% by IHC), and EGFR-mutant NSCLC (17% and 24% by NGS; 24% and 29% by IHC), with CDKN2A co-deletion in 98% of cases. MTAP loss was typically present prior to TT. MTAP loss did not significantly impact response or overall survival with first-line osimertinib or alectinib in EGFR mutant and ALK-rearranged NSCLC, respectively. In preclinical studies, BMS-986504 showed nanomolar activity in 11/18 MTAP-deleted models, including 5/8 EGFR- and 5/5 ALK-driven models, regardless of TT sensitivity. Synergistic or additive effects with TT were observed in 11/18 models. In ex vivo ALK-rearranged spheroids resistant to crizotinib, the combination of BMS-986504 and alectinib improved antitumor activity over monotherapy. In an osimertinib-resistant, EGFR mutant PDX model, BMS-986504 with or without osimertinib controlled tumor growth, without weight loss.

Conclusions: MTAP loss is frequent in oncogene-driven NSCLC, particularly in ALK-, RET-, and EGFR-altered subtypes. MTA-cooperative PRMT5 inhibition demonstrates broad activity in MTAP-deleted, oncogene-driven models and, and may enhance targeted therapy efficacy in selected settings.