Recent advancements in large language models (LLMs) enable real-time web search, improved referencing, and multilingual support, yet ensuring they provide safe health information remains crucial. This perspective evaluates seven publicly accessible LLMs-ChatGPT, Co-Pilot, Gemini, MetaAI, Claude, Grok, Perplexity-on three simple cancer-related queries across eight languages (336 responses: English, French, Chinese, Thai, Hindi, Nepali, Vietnamese, and Arabic). None of the 42 English responses contained clinically meaningful hallucinations, whereas 7 of 294 non-English responses did. 48 % (162/336) of responses included valid references, but 39 % of the English references were.com links reflecting quality concerns. English responses frequently exceeded an eighth-grade level, and many non-English outputs were also complex. These findings reflect substantial progress over the past 2-years but reveal persistent gaps in multilingual accuracy, reliable reference inclusion, referral practices, and readability. Ongoing benchmarking is essential to ensure LLMs safely support global health information dichotomy and meet online information standards.

Colorectal cancer remains the third most common malignancy in Australia with the peritoneum being the second most common metastatic site. Colorectal peritoneal carcinomatosis (CPC) can be treated with cytoreductive surgery and hyperthermic intraperitoneal chemotherapy but this is only limited to a small subset of patients. Those with inoperable disease have a particularly poor prognosis. While the ideal systemic regimen has not been defined, 5‐fluorouracil‐based chemotherapy regimens appear to provide overall and progression free survival benefits. The role of targeted agents such as bevacizumab (vascular endothelial growth factor inhibitor) or cetuximab (epidermal growth factor inhibitor) in the setting of CPC is still evolving. Currently, retrospective analyses have shown promising results for the use of bevacizumab in addition to systemic chemotherapy but similar results have not been seen with cetuximab or panitumumab. However, there is significant heterogeneity in the trial data, lack of prospective randomized controlled trials and demonstrated treatment variability based on age and tumour characteristics. This review summarises the current literature in regard to treatment in the unresectable CPC setting as well as discussing issues with the current data and highlighting the need for further trials.

PURPOSE: Although large language models (LLMs) are increasingly used in clinical practice, formal assessments of their quality, accuracy, and effectiveness in medical oncology remain limited. We aimed to evaluate the ability of ChatGPT, an LLM, to generate physician and patient letters from clinical case notes. METHODS: Six oncologists created 29 (four training, 25 final) synthetic oncology case notes. Structured prompts for ChatGPT were iteratively developed using the four training cases; once finalized, 25 physician-directed and patient-directed letters were generated. These underwent evaluation by expert consumers and oncologists for accuracy, relevance, and readability using Likert scales. The patient letters were also assessed with the Patient Education Materials Assessment Tool for Print (PEMAT-P), Flesch Reading Ease, and Simple Measure of Gobbledygook index. RESULTS: Among physician-to-physician letters, 95% (119/125) of oncologists agreed they were accurate, comprehensive, and relevant, with no safety concerns noted. These letters demonstrated precise documentation of history, investigations, and treatment plans and were logically and concisely structured. Patient-directed letters achieved a mean Flesch Reading Ease score of 73.3 (seventh-grade reading level) and a PEMAT-P score above 80%, indicating high understandability. Consumer reviewers found them clear and appropriate for patient communication. Some omissions of details (eg, side effects), stylistic inconsistencies, and repetitive phrasing were identified, although no clinical safety issues emerged. Seventy-two percent (90/125) of consumers expressed willingness to receive artificial intelligence (AI)-generated patient letters. CONCLUSION: ChatGPT, when guided by structured prompts, can generate high-quality letters that align with clinical and patient communication standards. No clinical safety concerns were identified, although addressing occasional omissions and improving natural language flow could enhance their utility in practice. Further studies comparing AI-generated and human-written letters are recommended.

Overall survival (OS) is the optimal marker of efficacy of treatments in randomized clinical trials (RCTs) but can take a considerable amount of time to mature. Progression free survival (PFS) has served as an early surrogate of OS however is imperfect. Time to deterioration (TTD) in quality of life (QoL) measures could be an alternative surrogate of OS. Phase 3 RCTs in solid malignancies that reported OS, PFS and TTD in QoL or physical function (PF) published between 1^st of January 2010 and 30^th of June 2022 were evaluated. Weighted regression analysis was utilised to assess the relationship between PFS, TTD QoL and TTD PF with OS. The coefficient of determination (R²) was used to quantify surrogacy. 138 phase 3 RCT were included. 47 trials evaluated immune checkpoint inhibitors (ICIs) and 91 investigated non-ICI agents. TTD QoL (137 RCTs) and TTD PF (75 RCTs) performed similarly to PFS as surrogates of OS (R² 0.18 vs 0.19 and R² 0.10 vs 0.09, respectively). For ICI studies, TTD PF had a higher association with OS compared to PFS (R² 0.38 vs 0.19) and PFS and TTD PF were not correlated with each other (R²=0). When used together, the coefficient of determination increased (R² = 0.57). TTD of PF appears to be a OS surrogate measure of particular importance for ICI treatment efficacy. The combination of TTD PF with PFS may enable better prediction of OS treatment benefit in ICI RCTs than either PFS or TTD PF alone.

Background: Treatment response to immune checkpoint inhibitors (ICIs) varies considerably, a phenomenon known as treatment effect heterogeneity (HTE). While HTE is explored via one-variable-at-a-time subgroup analyses in randomised controlled trials (RCTs), this method has limitations, which the risk-modelling approach seeks to address. Methods: Applying the risk-modelling approach, individual patient data from ten RCTs (six supporting FDA atezolizumab label: OAK, IMpower130, IMpower150, IMpower133, IMbrave150, IMspire150; four unlabelled indications: IMpower131, and IMpower132, IMmotion151, IMvigor211) were analysed by an extreme gradient-boosting algorithm to predict pre-treatment prognosis for overall survival. The predicted risk scores were evaluated as efficacy modifiers categorically (high, intermediate, low risk groups) and continuously in Cox models with treatment-by-risk-group interaction terms. Sensitivity and exploratory analyses investigated absolute and meta-analysed treatment effect and compared the results with established prognostic tools and treatment effect predictors. Statistical significance tests are two-sided. Results: Among the ten RCTs (N = 7053), one trial—IMvigor211 - showed significant HTE by pre-treatment prognosis across all evaluations (risk groups, risk scores, sensitivity analyses: p < .001). Amongst other trials, no significant HTE was detected (risk group and risk score analysis interaction test: OAK p = .61, 0.77; IMpower130 p = .13, 0.52; IMpower131: p = .21, 0.02; IMpower150: p = .14, 0.36; IMpower133: p = .38, 0.12; IMbrave150: p = .15, 0.08; IMspire150: p = .24, 0.6; IMpower132: p = .15, 0.81; IMmotion151: p = .48, 0.21). Conclusion: The risk-modelling approach showed no clear link between pre-treatment prognosis and ICI efficacy in most RCTs, particularly those supporting atezolizumab’s FDA label. In IMvigor211, patients with better pre-treatment prognosis were more likely to benefit from atezolizumab treatment for platinum-refractory metastatic urothelial carcinoma.