Daniel Matthias Aebersold

Checkpoint kinase 2 (CHK2) participates in the DNA damage response (DDR) by regulating the cell cycle or inducing apoptosis, and its abnormal activity can lead to cancer development. CHK2 loss-of-function mutations correlate with the onset of many types of tumors and with therapy resistance. As the exact molecular mechanisms underlying these phenomena are not fully elucidated, we aimed to understand how CHK2 deficiency in cancer cells affects their DDR functionality and their DNA repair ability post-treatment. Despite compromised RAD51 foci formation, we found that CHK2 -deficient HCT116 cancer cells exhibit higher homologous recombination repair (HRR) frequency than CHK2 -proficient cells. Transcriptomics analysis revealed a significantly decreased expression of DNA-PKcs and increased NBS1 mRNA levels in CHK2 knockout (KO) cells compared to their isogenic CHK2 WT line. These changes were also confirmed on a protein level. Consequently, the repression of the NHEJ pathway by the DNA-PKcs small molecule inhibitor peposertib further decreased the viability of HCT116 CHK2 -proficient cells compared to ionizing radiation alone, while the blockage of HRR by the RAD51 inhibitor CAM833 did not radiosensitize cancer cell lines with distinct CHK2 alterations. In conclusion, our results indicate that CHK2 KO influences DNA repair pathways engagement, and the targeting of these pathways in combination with irradiation deserves further studies to uncover the significance of CHK2 alterations in shaping radiation response.

Background: Large language models (LLMs) can accurately classify biomedical documents, but strong benchmark performance does not establish that predictions are grounded in the supplied text. In biomedical literature tasks, titles, abstracts, digital object identifiers (DOIs), journal metadata, and trial identifiers may have been seen during pretraining and can trigger parametric knowledge or learned associations. Objective: This study aimed to test whether oncology randomized trial success classification is driven by abstract evidence or by identifier-conditioned prior knowledge, and assess whether models follow counterfactual outcome evidence when it conflicts with original trial identifiers. Methods: We evaluated 250 two-arm oncology randomized controlled trials from 7 major journals published between 2005 and 2023, each with a single primary endpoint and previously adjudicated positive or negative ground-truth label. The corpus included 58.4% (146/250) positive and 41.6% (104/250) negative trials. GPT-5.2, Gemini 3 Flash, and Claude Opus 4.5 were queried via vendor APIs under default settings using a single-token output instruction. For each trial, we created 5 deterministic input conditions: title+abstract, title only, DOI only, counterfactual title+abstract in which the primary endpoint outcome statement was minimally flipped, and the same counterfactual input paired with the original DOI to create an identifier-text conflict. Performance was assessed using valid format rate, accuracy, sensitivity, specificity, and F ₁ -score. Results: The models showed high format adherence, with valid prediction rates of 97.2% to 100%. In the title+abstract condition, all models achieved high and balanced performance (accuracy and F ₁ -score=0.96-0.97; sensitivity=0.96-0.97; specificity=0.96-0.98). Removing evidence reduced performance stepwise: title-only accuracy and F ₁ -score fell to 0.79 to 0.88, and DOI-only performance fell to 0.63-0.67, exceeding the 58.4% majority class baseline but indicating limited identifier-driven signal. Counterfactual edits were concentrated in outcome-bearing text, with the Results and Conclusions sections modified for all trials, whereas the titles and Methods sections required edits in only 5.2% (13/250) and 1.6% (4/250) of trials. Against inverted labels, models followed counterfactual evidence with near-ceiling performance (accuracy and F ₁ -score=0.96-0.99). Reintroducing the original DOI caused little change for GPT-5.2 (accuracy and F ₁ -score=0.99) but modestly reduced F ₁ -scores for Gemini (0.97) and Claude (0.95), mainly through lower sensitivity. Conclusions: The evaluated LLMs robustly followed explicit end point statements in abstracts, including when those statements contradicted original trial outcomes. However, above-chance title-only and DOI-only performance, together with small decrements under counterfactual DOI conflicts, showed that identifiers can carry predictive signal and occasionally compete with textual evidence. Progressive content removal combined with counterfactual identifier-text conflicts offers a practical, reproducible audit for grounding in biomedical LLM evaluations.

PURPOSE: The growing volume of biomedical literature, especially in oncology, necessitates automated tools for extracting clinically relevant information. Large language models (LLMs) offer promising capabilities for data extraction. However, their potential to extract clinically relevant information from case reports detailing rare treatment interactions remains underexplored. METHODS: We systematically searched PubMed for case reports on interactions between radiotherapy (RT) and pembrolizumab, cetuximab, or cisplatin. A random sample of 100 report abstracts for each therapy was manually classified by two independent medical experts using 23 Boolean questions about patient demographics, treatment, cancer type, and outcomes with mutually exclusive answers, forming a ground truth. An LLM-based system with the open-source Generative Pretrained Transformer (GPT) models (GPT-OSS-120B and GPT-OSS-20B) was applied to classify these reports and the remaining data set entries using the defined question structure. Performance of the approach was evaluated using the standard classification metrics accuracy, precision, recall, and F1-scores. RESULTS: The searches yielded 320 (pembrolizumab), 147 (cetuximab), and 2055 (cisplatin) publications. Inter-rater agreement for manual classification was high (Cohen's kappa = 0.85), though lower for specific outcome and cancer type questions. The LLM-based classification (GPT-OSS-120B model) achieved high overall performance with an F1-score of 93.64% (95.19% accuracy, 93.23% precision, 94.05% recall). Performance was consistent across systemic therapies (STs), with the GPT-OSS-20B model showing similar results (F1-score 93.22%). Analysis of the entire data sets revealed that 56.14% of publications described patients who received both RT and ST. Proportions of positive and negative outcomes varied by therapy and sequencing. CONCLUSION: LLM-based classification systems demonstrate high performance for curating scientific case reports on RT and ST interactions. These findings support their potential for high-throughput hypothesis generation and knowledge base construction, particularly for underutilized case reports, with even smaller open-source models proving to be effective.

Background: Randomized controlled trials (RCTs) are the gold standard for evaluating interventions in oncology, but reporting can be subject to “spin”—presenting results in ways that mislead readers about true efficacy. Objective: This study aimed to investigate whether large language models (LLMs) could provide a standardized approach to detect spin, particularly in the conclusions, where it most commonly occurs. Methods: We randomly sampled 250 two-arm, single–primary end point oncology RCTs from 7 major medical journals published between 2005 and 2023. Two authors independently annotated trials as positive or negative based on whether they met their primary end point. Three commercial LLMs (GPT-3.5 Turbo, GPT-4o, and GPT-o1) were tasked with classifying trials as positive or negative when provided with (1) conclusions only; (2) methods and conclusions; (3) methods, results, and conclusions; or (4) title and full LLM performance was evaluated against human annotations. Afterward, trials incorrectly classified as positive when the model was provided only with the conclusions but correctly classified as negative when provided with the whole abstract were analyzed for patterns that may indicate the presence of spin. Model performance was assessed using accuracy, precision, recall, and F ₁ -score calculated from confusion matrices. Results: Of the 250 trials, 146 (58.4%) were positive, and 104 (41.6%) were negative. The GPT-o1 model demonstrated the highest performance across all conditions, with F ₁ -scores of 0.932 (conclusions only; 95% CI 0.90-0.96), 0.96 (methods and conclusions; 95% CI 0.93-0.98), 0.98 (methods, results, and conclusions; 95% CI 0.96-0.99), and 0.97 (title and abstract; 95% CI 0.95-0.99). Analysis of trials incorrectly classified as positive when the model was provided only with the conclusions revealed shared patterns, including absence of primary end point results, emphasis on subgroup improvements, or unclear distinction between primary and secondary end points. These patterns were almost never found in trials correctly classified as negative. Conclusions: LLMs can effectively detect potential spin in oncology RCT reporting by identifying discrepancies between how trials are presented in the conclusions vs the full abstracts. This approach could serve as a supplementary tool for improving transparency in scientific reporting, although further development is needed to address more complex trial designs beyond those examined in this feasibility study.

Background: The growing volume of biomedical literature, especially in oncology, necessitates automated tools for extracting clinically relevant information. Large Language Models (LLMs) offer promising capabilities for data extraction in this domain. However, their potential to extract clinically relevant information from case reports detailing rare treatment interactions, remains underexplored. Methods: We systematically searched PubMed for case reports on interactions between radiotherapy (RT) and Pembrolizumab, Cetuximab, or Cisplatin. A random sample of 100 report abstracts for each therapy was manually classified by two independent medical experts using 17 Boolean questions about patient demographics, treatment, cancer type and outcome with mutually exclusive answers, forming a ground truth. An LLM-based system with the open-source GPT models (GPT-OSS-120B and GPT-OSS-20B) was applied to classify these reports and the remaining dataset entries using the defined question structure. Performance of the LLM-based information extraction was evaluated using the standard classification metrics accuracy, precision, recall, and F1-scores. Results: The systematic searches yielded 320 (Pembrolizumab), 147 (Cetuximab), and 2055 (Cisplatin) publications. Inter-rater agreement for manual classification was high (Cohen’s kappa = 0.87), though lower (0.60-0.80) for specific outcome and cancer type questions. The LLM-based classification (GPT-OSS-120B model) achieved high overall performance with an F1-score of 94.33% (95.83% accuracy, 93.69% precision, 94.98% recall). Performance was consistent across systemic therapies, with the smaller GPT-OSS-20B model showing similar results (F1-score 94.06%). Analysis of the entire datasets revealed that 56.02% of publications described patients who received both RT and systemic therapy. Proportions of positive and negative outcomes varied by therapy and sequencing. Conclusions: LLM-based classification systems demonstrate high accuracy and reliability for curating scientific case reports on RT and systemic therapy interactions. These findings support their potential for high-throughput hypothesis generation and knowledge base construction in oncology, particularly for underutilized case reports, with even smaller open-source models proving effective for such tasks.

Background: Large language models (LLMs) hold promise for supporting clinical tasks, particularly in data-driven and technical disciplines such as radiation oncology. While prior evaluation studies have focused on examination-style settings for evaluating LLMs, their performance in real-life clinical scenarios remains unclear. In the future, LLMs might be used as general AI assistants to answer questions arising in clinical practice. It is unclear how well a modern LLM, locally executed within the infrastructure of a hospital, would answer such questions compared with clinical experts. Objective: This study aimed to assess the performance of a locally deployed, state-of-the-art medical LLM in answering real-world clinical questions in radiation oncology compared with clinical experts. The aim was to evaluate the overall quality of answers, as well as the potential harmfulness of the answers if used for clinical decision-making. Methods: Physicians from 10 departments of European hospitals collected questions arising in the clinical practice of radiation oncology. Fifty of these questions were answered by 3 senior radiation oncology experts with at least 10 years of work experience, as well as the LLM Llama3-OpenBioLLM-70B (Ankit Pal and Malaikannan Sankarasubbu). In a blinded review, physicians rated the overall answer quality on a 5-point Likert scale (quality), assessed whether an answer might be potentially harmful if used for clinical decision-making (harmfulness), and determined if responses were from an expert or the LLM (recognizability). Comparisons between clinical experts and LLMs were then made for quality, harmfulness, and recognizability. Results: There were no significant differences between the quality of the answers between LLM and clinical experts (mean scores of 3.38 vs 3.63; median 4.00, IQR 3.00-4.00 vs median 3.67, IQR 3.33-4.00; P=.26; Wilcoxon signed rank test). The answers were deemed potentially harmful in 13% of cases for the clinical experts compared with 16% of cases for the LLM (P=.63; Fisher exact test). Physicians correctly identified whether an answer was given by a clinical expert or an LLM in 78% and 72% of cases, respectively. Conclusions: A state-of-the-art medical LLM can answer real-life questions from the clinical practice of radiation oncology similarly well as clinical experts regarding overall quality and potential harmfulness. Such LLMs can already be deployed within the local hospital environment at an affordable cost. While LLMs may not yet be ready for clinical implementation as general AI assistants, the technology continues to improve at a rapid pace. Evaluation studies based on real-life situations are important to better understand the weaknesses and limitations of LLMs in clinical practice. Such studies are also crucial to define when the technology is ready for clinical implementation. Furthermore, education for health care professionals on generative AI is needed to ensure responsible clinical implementation of this transforming technology.