JCO Precision Oncology

A recent randomized clinical trial in non-small cell lung cancer1 confirms what numerous observational studies have reported time of day (ToD) may dramatically influence treatment outcomes in cancer patients. In this recent trial median overall survival (OS) decreased from 28 months in the early ToD arm to 16.8 months in the late ToD arm. We raise the concern that clinical trial outcomes may be influenced by seemingly minor biases in treatment time across arms. We also suggest that by measuring or randomizing treatment-time in clinical trials, we may identify beneficial ToD dependent treatments that would otherwise be overlooked.

BackgroundIn immune checkpoint inhibitor (ICI) trials, overall survival (OS) benefits are well established, yet improvements in quality of life (QoL) are often inconsistent or absent in conventional analyses. This apparent discordance raises important questions: are QoL outcomes truly unrelated to survival, and how can QoL results be better utilized and interpreted? MethodsA model-based meta-analysis (MBMA) of longitudinal EORTC QLQ-C30 global health status/quality of life data from randomized ICI trials was conducted. Longitudinal QoL trajectories were analyzed using a nonlinear mixed-effects model to estimate treatment-related toxicity and long-term QoL improvement. Associations between QoL trajectory parameters and OS were assessed using spearman rank correlation tests and Cox proportional hazards models. ResultsTwenty-seven studies (8,149 ICI and 5,593 control patients) contributed longitudinal QoL data, and 18 studies provided matched OS data. Raw QoL trajectories showed overlap between treatment arms, while OS consistently favored ICIs. MBMA revealed that ICIs had similar toxicity but significantly faster QoL improvement than control therapies (p < 0.0001). Baseline QoL, toxicity, and QoL improvement rate were all significantly associated with OS (p < 0.001). MBMA-based QoL comparisons were more sensitive in detecting associations with survival than raw QoL data, with the strongest association observed at Week 24 (R = -0.37, p = 0.067). ConclusionsConventional analyses comparing QoL at a single time point may obscure meaningful patient-reported benefits. By capturing longitudinal QoL trajectories across trials, MBMA reveals how patient experience evolves alongside survival outcomes and supports improved interpretation and utilization of QoL data in treatment evaluation.

Allogeneic hematopoietic cell transplantation is the only curative option for many patients with acute myeloid leukemia (AML). In the current study, we designed and implemented a personalized assay, called v96, incorporating up to 96 mutations in 30 AML patients undergoing transplantation. The assay was performed on DNA derived in cells from the bone marrow as well as in cell-free plasma. All 30 (100%) of patients harbored molecular evidence of residual leukemia during remission that was detectable by the v96 assay, while only 6 (20%) had evidence of disease as assessed by conventional clinical assays. Furthermore, cell-free DNA from plasma proved to be more sensitive than DNA from cells of the bone marrow for identifying residual leukemia. The median number of mutants was 352-fold higher in plasma taken prior to transplantation for patients who relapsed compared to those who did not relapse. At two months post-transplantation, 27 of the 30 patients still harbored detectable leukemia as assessed by the v96 assay. Twenty-two of these patients had a subsequent decrease in leukemic burden assessed by the v96 assay, usually only after immunosuppression was discontinued and supporting a graft-versus-leukemia effect. These results document the feasibility of using a relatively large panel of carefully chosen mutations and a highly specific assay as non-invasive markers of therapeutic response in AML patients, minimizing the need for multiple bone marrow biopsies. STATEMENT OF SIGNIFICANCEWe report a blood test that tracks up to 96 patient-specific mutations and applied it to patients with AML who had undergone bone marrow transplantation. Using this test to evaluate cell-free plasma DNA, we found evidence of residual leukemia cells both during remission (prior to transplantation) in all patients, and two months following transplantation in 90% of patients. This test can mitigate the need for invasive bone marrow biopsies to follow patients with leukemia. Moreover, the test appears to be more accurate than standard assays for detecting residual leukemia, and has the potential to guide the timing of transplantation and subsequent therapeutic measures, thereby laying the foundation for future prospective studies.

BackgroundOnce the treatment starts, early prediction of treatment benefit and its correlation with overall survival (OS) remains challenging in metastatic castration-resistant prostate cancer (mCRPC). Existing prognostic models require long-term follow-up, limiting their ability to inform timely treatment decisions. To address this gap, we evaluated tumor growth rate (g-rate)-based survival models across multiple treatment lines to assess their ability to predict OS and support early clinical decision-making. MethodsWe developed GxSurv, a Random Survival Forest (RSF)-based framework that incorporates baseline clinical variables and g-rate calculated from serial on-treatment PSA, to construct line-specific prediction models of OS, a direct measure of treatment outcome. Three variants were developed: G3Surv, using the 3-month g-rate; G6Surv, using the 6-month g-rate; and GfSurv, using the final observed g-rate. Model performance was evaluated using Harrells C-index, Unos C-index, Integrated Brier Score (IBS), time-dependent area under the curve (tAUC). Model interpretability was assessed using permutation importance to quantify predictor contributions within the GxSurv framework. FindingsThe study included 15912 treatment records from 11014 patients with mCPRC across four lines of therapy. We found that incorporation of g-rate consistently improved model performance across all treatment lines, with all GxSurv models outperforming Cox proportional hazards (CoxPH). As the earliest prognostic model, our G3Surv demonstrated strong early predictive performance, with Harrells C-index values ranging from 0{middle dot}700 to 0{middle dot}746 and tAUC values of 0{middle dot}766 to 0{middle dot}822 across all lines, representing 5-8% and 4-5% improvements over CoxPH, respectively. These results indicate that G3Surv accurately predicts individual treatment outcomes at 3 months after treatment initiation. Feature importance analyses consistently identified g-rate as a top predictor, followed by baseline PSA and hemoglobin, with relative variation across treatment lines. InterpretationIntegrating g-rate calculated from on-treatment PSA values enables accurate, line-specific prediction of treatment outcomes in mCRPC, with the 3-month g-rate providing robust early prognostic information to support timely, personalized clinical decision-making. FundingU.S. National Science Foundation, National Institutes of Health, American Cancer Society.

IntroductionBRCA1/2 alterations are increasingly recognized as biologically and clinically relevant features in prostate cancer, yet the prognostic and therapeutic significance of zygosity status remains uncertain. Understanding differences between monoallelic and biallelic inactivation may refine risk stratification and guide therapeutic decision-making. Materials and MethodsA retrospective, desk-based observational analysis was performed using publicly accessible datasets from TCGA-PRAD (primary disease) and SU2C/PCF (metastatic disease). BRCA1/2 status was categorized as wild-type, monoallelic, or biallelic based on mutation, copy-number, and loss-of-heterozygosity profiles. Overall survival was evaluated using Kaplan-Meier estimates and Cox models. Systemic therapy outcomes were assessed by treatment class, incorporating exploratory interaction tests. ResultsIn TCGA-PRAD (n=300), OS did not significantly differ by zygosity (global log-rank p=0.45), with median OS of 80.0 months (wild-type), 78.0 months (monoallelic), and 55.0 months (biallelic). In SU2C/PCF (n=200), zygosity stratified outcomes significantly (global log-rank p=0.04): median OS was 22.0 months (wild-type), 14.0 months (monoallelic), and 16.0 months (biallelic). Treatment analyses showed ARSI exposure improved OS in wild-type disease (HR 0.60; 95% CI 0.38-0.95), while interaction testing suggested potential heterogeneity without statistical confirmation (interaction p=0.092). PARP inhibitor exposure showed directionally favorable HRs in wild-type and monoallelic groups but no significant interaction (interaction p=0.757). No therapy class demonstrated consistent effect modification by zygosity. ConclusionBRCA1/2 zygosity shows prognostic relevance in metastatic prostate cancer but not clearly in primary disease. While zygosity did not consistently modify systemic therapy associations in this dataset, findings support zygosity-aware reporting as a practical tool for molecular stratification and future research design.

PurposePrognosis in metastatic non-seminomatous germ cell tumors (mNSGCT) is currently guided by the IGCCCG classification, which incorporates tumor markers, organs involved with metastatic disease, and primary site but not histologic subtype. We aimed to evaluate whether specific histological components provide additional prognostic information in a large international mNSGCT cohort. Patient and MethodsWe analyzed clinical, pathologic, and outcome data from 662 patients with mNSGCT across multiple international centers. Cox regression and multivariable stepwise models were used to evaluate the impact of age, tumor histology, serum markers, primary site of disease, chemotherapy, IGCCCG, and post-chemotherapy surgery on overall survival. Analyses were performed using both complete-case and imputed datasets to account for missing values. ResultsThe presence of any percentage of embryonal carcinoma (EC) was independently associated with improved overall survival HR 0.603 (95% CI: 0.37-0.98, p=0.040), whereas yolk sac tumor (YST) predicted worse prognosis in complete-case analysis HR 2.27 (95% CI: 1.43 - 3.61 p = 0.001). Choriocarcinoma was also associated with a HR 1.58 (95% CI: 1.08 - 2.32 p= 0.019) adverse outcomes. IGCCCG risk classification remained a strong predictor of mortality HR up to 8.9 for Poor vs Good risk, (95% CI: 4.63 - 17.09 p < 0.001), but histologic components added significant independent prognostic value. Post-chemotherapy retroperitoneal lymph node dissection (RPLND) conferred a substantial survival benefit HR 0.44 (95% CI: 0.258 - 0.754 p=0.003). Interestingly, teratoma was not associated with mortality but was linked to younger age, testicular primaries, and higher likelihood of residual disease requiring surgery. ConclusionsHistological composition, particularly the presence of EC or YST, has a significant and independent impact on survival in mNSGCT, beyond established risk classifications. Integration of histological subtypes may enhance prognostic accuracy and guide individualized treatment strategies in advanced germ cell tumors.

BackgroundGlyphosate-based herbicides are among the most widely used agricultural chemicals globally. Concerns regarding their carcinogenic potential, particularly in relation to non-Hodgkins lymphoma (NHL), persist despite multiple prior systematic reviews and meta-analyses. However, these reviews have demonstrated important methodological limitations and inconsistent analytic decisions, limiting confidence in their conclusions. ObjectiveTo conduct a rigorous, up-to-date systematic review and meta-analysis of observational studies examining the association between glyphosate-based herbicide exposure and risk of NHL and its subtypes, while addressing methodological and analytic shortcomings of prior syntheses. MethodsWe searched MEDLINE (1970-February 26, 2026) and EMBASE (inception-February 26, 2026), supplemented by reference list review. Eligible studies included cohort, case-control, and pooled analyses reporting effect estimates (or sufficient data) for glyphosate exposure and NHL incidence. Two reviewers independently assessed risk of bias using the Newcastle-Ottawa Scale (for primary studies) and structured criteria for pooled analyses. Random- and fixed-effects meta-analyses were conducted using inverse-variance methods. Heterogeneity was evaluated using Cochrans Q and I{superscript 2} statistics. Publication bias was assessed using standard and contour-enhanced funnel plots. Sensitivity analyses addressed overlapping cohorts, hazard ratio inclusion, exposure definitions, and model overfitting (events-per-variable considerations). Certainty of evidence was graded using GRADE. ResultsSeventeen publications were identified, representing 20 unique study populations; after accounting for overlap, 10 primary datasets were included in quantitative synthesis. Five studies were assessed as low risk of bias, four as moderate risk, and one as high risk. For ever exposure, the random-effects model across all eligible datasets yielded an odds ratio (OR) of 1.11 (95% CI: 0.98-1.27), with moderate heterogeneity (I{superscript 2}{approx}53%). In sensitivity analyses excluding hazard ratio-only studies and overlapping cohorts, pooled ORs ranged from 1.19 to 1.23, with estimates approaching or reaching statistical significance depending on modeling assumptions. For the highest exposure categories, the random-effects model demonstrated a statistically significant association (OR{approx}1.38; 95% CI: 1.00-1.90), with moderate heterogeneity (I{superscript 2}{approx}61%). Sensitivity analyses excluding selected pooled cohort estimates strengthened the association (OR{approx}1.47; 95% CI: 1.04-2.06). Analyses incorporating alternative cumulative exposure metrics yielded similar significant associations (OR{approx}1.33-1.45) with low or absent residual heterogeneity. Subtype analyses suggested elevated risks particularly for diffuse large B-cell lymphoma and follicular lymphoma in certain datasets. Publication bias assessments revealed evidence of small-study effects in some models, though contour-enhanced analyses suggested that not all asymmetry was attributable to selective publication. Overall certainty of evidence was graded as moderate for highest exposure analyses and low-to-moderate for ever-exposure analyses due to residual heterogeneity and observational design limitations. ConclusionsThis updated synthesis indicates that while associations with ever exposure to glyphosate are modest and sensitive to analytic decisions, higher levels of exposure are consistently associated with increased odds of NHL. Findings are robust across multiple sensitivity analyses addressing overlapping data, exposure classification, and model overfitting. These results support a dose-related association between glyphosate-based herbicide exposure and NHL risk and underscore the need for continued surveillance, improved exposure characterization, and prospective cohort analyses with minimized loss to follow-up and transparent analytic reporting.

BackgroundAdjuvant abemaciclib+endocrine therapy (ET) improves long-term outcomes in high-risk, hormone receptor-positive (HR+)/HER2-negative early breast cancer (eBC). However, treatment is frequently complicated by diarrhea, affecting adherence and quality of life (QoL). Increasing evidence suggests that abemaciclib-induced gastrointestinal toxicity may involve gut microbiota alterations. We conducted a prospective case-control pilot study evaluating the efficacy of MBR-01, a standardized prebiotic/probiotic formulation, in mitigating abemaciclib-induced diarrhea. MethodsWe enrolled 20 patients with high-risk HR+/HER2-negative eBC considered unfit for adjuvant chemotherapy. Patients received abemaciclib+letrozole (control, n=10) or abemaciclib+letrozole+MBR-01 (experimental, n=10). The primary endpoint was the incidence and severity of diarrhea; secondary endpoints included treatment adherence, QoL assessments and exploratory baseline/week-12 microbiota characterization according to treatment arm. Trial registration number: ISRCTN11948182. ResultsDiarrhea occurred in all patients. In the control group, diarrhea was predominantly grade 1 (50%) or grade 2 (40%), with one grade 3 event (10%). In the MBR-01 group, diarrhea frequency and severity were reduced by [~]70% at the end of week-12; 80% of patients experienced only grade 1 diarrhea or none by week-12, and no grade [&ge;]3 events. Dose modification was only required in one control. Alpha-diversity and depletion of F.prausnitzii were associated with earlier diarrhea onset and longer duration; enrichment in E.coli correlated with higher grade events. MBR-01 supplementation seemed to preserve microbial diversity and limited E.coli expansion. QoL was significantly improved with MBR-01. ConclusionMBR-01 may effectively mitigate abemaciclib-induced diarrhea, likely through the achievement of stabilization of gut microbiota composition. Larger prospective studies are warranted to validate these preliminary findings. HighlightsO_LIMBR-01, a prebiotic/probiotic, was given to reduce abemaciclib-induced diarrhea. C_LIO_LIMBR-01 reduced diarrhea by [~]70%, most patients had G0-1, one G [&ge;]3 at week 12. C_LIO_LIMBR-01 patients keep abemaciclib drug dose; 10% of controls required reduction. C_LIO_LIMBR-01 halved stool frequency and improved quality of life. C_LIO_LIMBR-01 preserved gut diversity, maintaining F. prausnitzii and limiting E. coli. C_LI

PurposeCastration-resistant prostate cancer (CRPC) is characterized by marked clinical heterogeneity and poor long-term survival, underscoring the need for tools that can rapidly and reliably individualize patient risk. While several prognostic models exist, their complexity has limited routine clinical use. Here, we developed and validated PROGRESS (PROstate cancer Global Risk Evaluation and Stratification Score), a simplified prognostic score, derived through machine learning-guided feature selection, to enhance risk stratification and support individualized, risk-informed clinical decision-making. MethodsPROGRESS was developed using baseline data from 2,035 metastatic CRPC patients enrolled in four different phase III trials. An unsupervised machine-learning approach was applied to identify latent patient subgroups with distinct survival outcomes irrespectively of allocated treatment arm, followed by classical multivariable modelling to derive a simple and straight-forward prognostic score based on routinely available objective laboratory variables. External validation was performed in three independent datasets comprising metastatic CRPC patients treated across different therapeutic settings (n=1,239) and non-metastatic CRPC patients managed with standard care (n=660). Overall survival was assessed using Kaplan-Meier and Cox regression analyses. ResultsUnsupervised modelling identified two patient risk subpopulations with significantly different overall survival rates (median 27.4 vs 17.7 months; hazard ratio [HR] 2.20, 95% CI 1.91-2.54; p<.00001). Feature contribution analysis yielded three independent predictors -PSA, ALP, and AST-used to build PROGRESS. In the training cohort, PROGRESS demonstrated strong discrimination (AUC 0.89). Using a prespecified cut-off, patients classified as increased risk had significantly shorter survival than low-risk patients (median 18.3 vs 25.6 months; HR 1.72, 95% CI 1.50-1.97; p<.0001). PROGRESS prognostic performance was consistent across all validation cohorts, including metastatic and non-metastatic disease, with HRs ranging from 1.74 to 3.46 (all p<.0001). ConclusionsBy integrating machine-learning-based pattern discovery with classical statistical modelling, PROGRESS provides a simple, objective, and clinically accessible approach for individual risk stratification in CRPC. Its reliance on three inexpensive, routinely measured laboratory parameters would facilitate practical implementation in clinical settings, enhancing visibility of underlying disease aggressiveness for individual clinical decision-making. PROGRESS could represent a pragmatic first step toward improving patient selection for clinical trials while identifying regulatory meaningful endpoints achievable in a sizeable patient population; further validation in prospective clinical studies and real-world datasets would allow to confirm its clinical utility and generalizability. PROGRESS can be freely accessed for research use only at the following link: https://dev.ai.topazium.com.

IntroductionThe precise determination of diagnostic cut-off points is essential for the development of multimarker panels in oncology. In previous work on pulmonary nodules, it was observed that the standard two-parameter logistic fit could be insufficient for biomarkers with asymmetric distributions. Furthermore, the calculation of empirical cut-off points based on graphical visualization presented limitations in precision and reproducibility. ObjectiveThis study presents a methodological advancement in the data analysis phase (Stage 1), introducing new Python algorithms for the direct analytical calculation of empirical intersections and robust mathematical modeling using Dual Annealing with both two-parameter and four-parameter logistic functions. This improved methodology feeds into the ThresholdXpert 1.0 software tool for combinatorial optimization of biomarker panels (Stage 2), and is applied here to the diagnostic challenge of hepatocellular carcinoma (HCC). MethodsThe methodology was first validated by re-analyzing a dataset of patients with pulmonary nodules (N=895). It was subsequently applied to an HCC dataset derived from the cohort of Jang et al. (208 HCC, 193 cirrhosis, 401 total), randomly divided into a training set (280) and an independent test set (121). Scripts were developed to compare the previous two-parameter logistic fit with the new two- and four-parameter logistic models. Finally, ThresholdXpert 1.0 was used for multimarker panel optimization. ResultsThe integration of empirical calculation, logistic modeling, and combinatorial optimization through ThresholdXpert 1.0 provides a robust and coherent framework for the development of multimarker diagnostic panels. The four-parameter logistic model provided additional validation without substantially modifying cut-off values for most biomarkers, confirming the stability of the approach while offering greater flexibility for complex distributions. When applied to hepatocellular carcinoma, the framework identified a molecular panel composed of AFP, PIVKA-II, OPN, and DKK-1 with sensitivity of 0.77 and specificity of 0.72, and an optimized panel incorporating inverse MELD that achieved the best overall balance (sensitivity 0.73, specificity 0.75) in independent external validation. These results demonstrate the potential of this approach as a generalizable tool for the optimized design of binary diagnostic systems in oncology. ConclusionThe integration of complementary mathematical modeling enhances the capability of ThresholdXpert 1.0 to identify robust diagnostic panels, as in some cases a single biomarker may outperform biomarker combinations, and vice versa. This approach enabled the integration of molecular biomarkers and clinical variables under a unified mathematical framework. Contactroberto117343@gmail.com

BackgroundThe NHS-Galleri trial reported a substantial reduction in Stage IV cancer diagnoses and a four-fold increase in cancer detection rates, but did not meet its primary endpoint of reducing combined Stage III+IV diagnoses in a prespecified group of 12 cancers. We hypothesize that stage slip-- progression of cancers from Stage I/II to Stage III during diagnostic workup--is the primary mechanism behind this statistical masking. MethodsWe developed a Monte Carlo simulation of 142,000 participants (matching NHS-Galleri enrolment) across 12 cancer types, calibrated to NHS England population stage distributions. The model represents three competing clocks: biological sojourn times, screeninginitiated or symptom-driven pathway initiation, and diagnostic infrastructure delays. The median standard-of-care diagnostic delay (92 days) was constructed from five convergent evidence streams. We estimated the number of intervention-arm cases where a cancer was biologically Stage I/II but recorded as Stage III+IV at diagnosis, and determined how many such cases would need to be recovered for the composite endpoint to reach statistical significance. We validated that the 12 deadly cancers produce a sufficiently large early-stage patient pool and confirmed robustness across a wide range of test sensitivity and diagnostic delay assumptions. FindingsThe calibrated control arm reproduces NHS stage distributions within 1-2 percentage points for all 12 cancers. In the intervention arm, we estimate 84 cases of stage slip (95% CI: 68-104) at published sensitivity values. If only 25 of these cases (approximately one in three) had been diagnosed before crossing the Stage II/III boundary, the composite endpoint would have reached p < 0.05 at reported sensitivity levels. The slip count is robust to sensitivity assumptions, varying from approximately 74 to 85 cases across a 30-100% range of published CCGA 3 values. Across diagnostic delays from 65 to 120 days (at realised sensitivity), the estimate ranges from approximately 55 to 95 cases. The expected pool of early-stage cancers in the 12 deadly types ([~] 260-335 screen-detected Stage I/II cases per arm) is large enough that even a 12% slip rate--corresponding to the most optimistic delay assumption--produces sufficient slippage to exceed the significance threshold. InterpretationStage slip provides a quantitatively sufficient and mechanistically transparent explanation for the primary endpoint miss. The tests biological performance is intact: it detects cancers earlier and reduces Stage IV diagnoses. The composite endpoint was attenuated by systemic diagnostic latency in the NHS, not by a failure of the assay. Future MCED trials should consider endpoints less vulnerable to infrastructure delays or incorporate prespecified adjustments for expected diagnostic pathway timing. FundingNone. RegistrationNot applicable.

Purpose: Tumor genomic testing (TGT) is standard-of-care for most patients with advanced/metastatic cancer. Despite established guidelines, patient education prior to TGT is frequently omitted. The purpose of this study was to evaluate the impact and durability of a concise 3-4 minute video for patient education prior to TGT in community versus academic sites and across cancer types. Patients and Methods: Patients undergoing standard-of-care TGT were enrolled at a tertiary academic institution in three cohorts: Cohort 1-breast cancer; Cohort 2-lung cancer; Cohort 3-other cancers. Cohort 4 consisted of patients with any cancer type similarly undergoing SOC TGT at one of three community cancer centers. Participants completed survey measures prior to video viewing (T1), immediately post-viewing (T2), and after return of TGT results (T3). Outcome measures included: 1) 10-question objective genomic knowledge/understanding (GKU); 2) 10-question video message-specific knowledge (VMSK); 3) 11-question Trust in Physician/Provider (TIPP); 4) perceptions regarding TGT. Results: A total of 203 participants completed all survey timepoints. Higher baseline GKU and VMSK scores were significantly associated with higher income and greater years of education. For the primary objective, there was a significant and sustained improvement in VMSK from T1:T2:T3 (Poverall p<0.0001), with no significant change in GKU (p=0.41) or TIPP (p=0.73). This trend was consistent within each cohort (all p[&le;]0.0001). Results for four VMSK questions significantly improved, including impact on treatment decisions, incidental germline findings, and insurance coverage of testing. Conclusions: A concise, 3-4 minute, broadly applicable educational video administered prior to TGT significantly and sustainably improved video message-specific knowledge in diverse cancer types and in academic and community settings. This resource is publicly available at http://www.tumor-testing.com, with a goal to efficiently educate and empower patients regarding TGT while addressing guidelines within the flow of clinical practice.

Immune-related adverse events (irAEs) affect up to 40% of patients receiving immune checkpoint inhibitors, yet their identification depends on laborious and inconsistent manual chart review. Here we developed and evaluated an agentic large language model system to extract the presence, temporality, severity grade, attribution, and certainty of six irAE types from clinical notes. Retrospectively (263 notes), the system achieved macro-averaged F1 of 0.92 for detection and 0.66 for multi-class severity grading; self-consistency improved F1 by 0.14. The best-performing configuration cost approximately $0.02 per note. In prospective silent deployment over three months (884 notes), detection F1 was 0.72-0.79. In a randomized crossover study of clinical trial staff (17 participants, 316 observations), agentic assistance reduced annotation time by 40% (P < 0.001), increased complete-match accuracy (OR 1.45; 95% CI 1.01-2.09; P = 0.045), and improved inter-annotator agreement (Krippendorffs from 0.22-0.51 to 0.82-0.85). These results demonstrate that agentic AI coupled with human verification could enhance efficiency, performance, and consistency for irAE assessment.

IntroductionDespite advancements in non-small cell lung cancer (NSCLC) management through the use of molecular biomarkers, the recently introduced 9th edition of the TNM staging system remains based exclusively on anatomic descriptors, with no consistently demonstrated improvement in risk stratification for early-stage disease. This study explores the integration of a molecular prognostic classifier into the conventional TNM staging system. MethodsWe analyzed 502 patients with stage I-III lung adenocarcinoma (LUAD) who underwent surgical resection with tumor-based gene expression profiling at the Quebec Heart and Lung Institute. A molecular prognostic classifier was developed and integrated into the 9th edition TNM staging system to generate a novel model (TNMEx). Prognostic performance was compared with the 8th and 9th TNM editions using prognostic discrimination and reclassification metrics. External validation of the molecular classifier was performed in 271 LUAD cases from The Cancer Genome Atlas (TCGA). An independent cohort of 606 resected LUAD patients from the National Cancer Center Hospital (Tokyo) was used to externally compare the prognostic performance of the 8th and 9th TNM staging systems in the absence of molecular data. ResultsThe molecular prognostic classifier was developed based on the expression levels of 26 prognosis-associated genes, weighted by their corresponding coefficients. The classifier was subsequently integrated into the 9th edition TNM staging to generate the TNMEx model. The TNMEx system demonstrated superior prognostic performance, achieving a higher concordance index (C-index = 0.72) compared to the 9th edition TNM (C-index = 0.65, p=0.006). Moreover, TNMEx significantly improved patient risk reclassification compared to both the 8th (net reclassification improvement [NRI] = 0.27, integrated discrimination improvement [IDI] = 0.04) and 9th editions (NRI = 0.40, IDI = 0.05), underscoring its superior ability to stratify outcomes. The 8th and 9th editions showed only limited improvement in overall prognostic accuracy and risk stratification, as reflected by their relatively modest C-index values (0.62 and 0.65, respectively) and minimal reclassification gains (NRI = -0.06, IDI = 0.003). ConclusionsIncorporating a molecular-based prognostic model significantly enhanced the ability to recognize patients at high risk and to predict their survival outcomes more accurately than traditional TNM staging systems.

PurposeCirculating tumor DNA (ctDNA) analyses are informative as an early indicator of immunotherapy response in advanced non-small cell lung cancer (NSCLC); however, the clinical value of ctDNA molecular response requires further validation. Patients and MethodsAs part of a prospective clinical protocol (NCT05995821), we conducted targeted error-correction sequencing of ctDNA (n=328) and matched WBC DNA (n=109) from 109 patients with metastatic NSCLC who received anti-PD-(L)1 either as monotherapy or in combination. Following cellular origin resolution of 2,818 variants, landmark molecular response (mR) was defined as undetectable ctDNA within 3-9 weeks of treatment initiation. ResultsPre-treatment ctDNA burden, but not blood tumor mutation burden, predicted survival. Implementing a tumor-naive WBC DNA-informed approach increased the number of evaluable cases without compromising the overall accuracy of landmark ctDNA molecular responses. A direct comparison of single-timepoint on-therapy ctDNA assessment with ctDNA dynamics from baseline to the 3-9-week interval, along with an analysis of heterogeneity in molecular response within the 3-9-week window, showed that undetectable ctDNA at the landmark timepoint can effectively predict survival outcomes. A significant enrichment in landmark ctDNA mR was noted among patients with progression-free survival (PFS) [&ge;]6 months with immunotherapy (p=2.5e-05) and chemo-immunotherapy (p=0.02). Patients in the landmark mR group had longer progression-free (p=1.6e-06) and overall survival (p=2.5e-05) than those with molecular progression. ConclusionsLandmark ctDNA molecular response provides a real-time, accurate approach for monitoring immunotherapy clinical outcomes. Although not currently validated for regulatory use, these findings demonstrate the potential utility of ctDNA as an early endpoint in clinical trials. Translational RelevanceEmploying circulating tumor DNA (ctDNA) dynamics as an early indicator of immunotherapy response requires a roadmap for the next-generation sequencing approach, definition of molecular response and establishment of its clinical sensitivity. In this study, we introduce the concept of a landmark ctDNA molecular response, determined 3-9 weeks after initiation of immunotherapy, that maximizes the number of evaluable patients without sacrificing the specificity of the approach. Notably, when evaluating heterogeneity in ctDNA detection within the landmark 3-9-week window and assessing the impact of landmark interval dynamics on survival, we found that a single ctDNA assessment performed similarly to multiple ctDNA measurements within the landmark window (most notably, regardless of whether the timepoints were concordant or discordant). Our findings demonstrate that a single assessment of early on-therapy landmark ctDNA molecular response, can identify patients at risk of disease progression and enable future intervention and therapy optimization.

Molecular subtyping of cancer is traditionally defined in transcriptomic space, yet routine clinical deployment is limited by the availability and cost of sequencing. Meanwhile, histopathology captures rich morphological information that is known to correlate with molecular state but lacks a principled, mechanistic bridge to gene-level representations. We propose a graph-constrained learning framework that aligns morphology-derived signals with a fixed, data-driven gene network discovered via hierarchical Monte Carlo screening. We can derive new gene sets for classification using random sampling, and use the coexpression network of that graph to enforce the learning of a pure morphology model without using gene expression. The resulting model performs subtype prediction using morphology alone, while being explicitly forced to operate through a gene-structured latent space. Structural alignment is enforced during training. For Moffitt classification in pancreatic cancer using PANCAN and TCGA datasets, the model has a reported 85% AUC using an alternative gene set network structure, while the alternate gene set itself has an 84% AUC in all patients that were classified with subtyping with pancreatic cancer in the dataset. This demonstrates that virtual transcriptomics can provide biologically grounded molecular insights using only routine histopathology slides, potentially expanding access to precision oncology in resource-limited settings.

Gastrointestinal (GI) polyposis is a major risk factor for colorectal cancer (CRC) and a defining feature of hereditary polyposis syndromes such as familial adenomatous polyposis (FAP). Therapy-associated polyposis (TAP), however, is a rare and incompletely characterized condition that develops decades after treatment for childhood or young adult cancers (CYAC), most often following abdominopelvic radiation or exposure to alkylating agents. As long-term CYAC survival improves, the burden of late GI toxicity, including markedly elevated risks of polyps, CRC, and secondary cancers, continues to rise, yet the molecular features of TAP remain poorly understood. Here, we present the largest clinicopathological and genomic study of TAP to date, comprising 29 patients diagnosed at a median age of 49 years and a median latency of 29 years after primary cancer therapy. Most patients (78%) had received alkylating agents and exhibited high rates of secondary malignancies. Histopathology revealed mixed polyp subtypes with a predominance of adenomas. Given these features and the presence of family history in a subset of patients, we investigated the possibility of Hereditary Mixed Polyposis Syndrome (HMPS). Whole-genome sequencing excluded HMPS by demonstrating absence of the canonical 40-kb GREM1 duplication and lack of consistent GREM1 overexpression. Comparative genomic analysis revealed that TAP adenomas exhibit more extensive genome fragmentation and a higher burden of large structural variants than FAP adenomas. Mutational signature profiling identified strong contributions from age-associated signatures (SBS1, SBS5) and a strong, pervasive contribution of the alkylating-agent signature SBS25, even in samples lacking matched normal tissue, whereas platinum-associated SBS31 was minimal. Patient-derived organoids from TAP adenomas showed impaired differentiation, suggesting persistent therapy-induced stem cell dysfunction. Together, these findings define TAP as a distinct polyposis syndrome marked by heterogeneous histology, long latency, profound structural genomic injury, and chemotherapy-specific mutational scars. This work supports early and tailored GI surveillance for CYAC survivors and provides mechanistic insight into the long-term consequences of cytotoxic therapy on intestinal epithelial homeostasis.

Multi-cancer early detection (MCED) tests can detect several cancer types and stages. We previously developed a methylation and protein (MP V1) MCED classifier. In this study, we present a refined MP V2 classifier, developed by evaluating model architectures that improved performance in prospectively enrolled case-control cohorts under standard testing conditions. The newly developed MP V2 classifier was trained to be more generalizable and achieve increased early-stage sensitivity at a target specificity of [&ge;]97.0%. MP V1 and MP V2 classifier performances were compared using a previously described test set, and MP V2 performance was also evaluated in a new independent clinical validation set. Compared to MP V1, the MP V2 classifier demonstrated a 7.3% increase in overall sensitivity, with sensitivity increases of 7.6%, 9.2%, and 8.3% for stages I, II, and stages I/II, respectively, in the intended use (breast and prostate cancers excluded) test set. In an independent validation intended use set, the MP V2 classifier showed an overall sensitivity of 55.6%, with sensitivities of 26.8%, 42.9%, and 34.8% for stages I, II, and stages I/II, respectively. In a case-control setting, the MP V2 classifier offered improved sensitivity for early-stage cancers at a lower specificity target.

Introduction: Manual data extraction from unstructured clinical notes is labor-intensive and impractical for large-scale clinical and research operations. Existing automated approaches typically require large language models, dedicated computational infrastructure, and/or task-specific fine-tuning that depends on curated data. The objective of this study is to enable accurate extraction with smaller locally deployed models using a disease-site specific pipeline and prompt configuration that are optimized and reusable. Materials/Methods: We developed OncoRAG, a four-phase pipeline that (1) generates feature-specific search terms via ontology enrichment, (2) constructs a clinical knowledge graph from notes using biomedical named entity recognition, (3) retrieves relevant context using graph-diffusion reranking, and (4) extracts features via structured prompts. We ran OncoRAG using Microsoft Phi-3-medium-instruct (14B parameters), a midsize language model deployed locally via Ollama. The pipeline was applied to three cohorts: triple-negative breast cancer (TNBC; npatients=104, nfeatures=42; primary development), recurrent high-grade glioma (RiCi; npatients=191, nfeatures=19; cross-lingual validation in German), and MIMIC-IV (npatients=100, nfeatures=10; external testing). Downstream task utility was assessed by comparing survival models for 3-year progression-free survival built from automatically extracted versus manually curated features. Results: The pipeline achieved mean F1 scores of 0.80 +/- 0.07 (TNBC; npatients=44, nfeatures=42), 0.79 +/- 0.12 (RiCi; npatients=61, nfeatures=19), and 0.84 +/- 0.06 (MIMIC-IV; npatients=100, nfeatures=10) on test sets under the automatic configuration. Compared to direct LLM prompting and naive RAG baselines, OncoRAG improved the mean F1-score by 0.19 to 0.22 and 0.17 to 0.19, respectively. Manual configuration refinement further improved the F1-score to 0.83 (TNBC) and 0.81 (RiCi), with no change in MIMIC-IV. Extraction time averaged 1.7-1.9 seconds per feature with the 14B model. Substituting a smaller 3.8B model reduced extraction time by 57%, with a decrease in F1-score (0.03-0.10). For TNBC, the extraction time was reduced from approximately two weeks of manual abstraction to under 2.5 hours. In an exploratory survival analysis, models using automatically extracted features showed a comparable C-index to those with manual curation (0.77 vs 0.76; 12 events). Conclusions: OncoRAG, deployed locally using a mid-size language model, achieved accurate feature extraction from multilingual oncology notes without fine-tuning. It was validated against manual extraction for both retrieval accuracy and survival model development. This locally deployable approach, which requires no external data sharing, addresses a critical bottleneck in scalable oncology research.

Medical oncology education faces a dual crisis: knowledge velocity that outpaces static curricula and large language model (LLM) risks--hallucination and automation bias--that threaten the fidelity of AI-assisted learning. We present Onco-Shikshak V7, an AI-native adaptive learning platform that addresses both challenges through a unified cognitive architecture grounded in learning science. The system replaces isolated educational modules with four authentic clinical workflows--Morning Report, Tumor Board, Clinic Day, and AI Textbook--each scaffolded by a nine-module pedagogy engine that integrates ACT-R activation dynamics (illness scripts), Item Response Theory (adaptive difficulty), the Free Spaced Repetition Scheduler (FSRS v4), Zone of Proximal Development (scaffolding), and metacognitive calibration training (Brier score). Six specialist AI agents--medical oncology, radiation oncology, surgical oncology, pathology, radiology, and oncology navigation--engage in multi-disciplinary deliberation with per-specialty retrieval-augmented generation (RAG) grounding across nine authoritative guideline sources including NCCN, ESMO, and ASTRO. The platform provides 18 clinical cases with decision trees across six cancer types, maps every interaction to 13 ACGME Hematology-Oncology milestones, and implements four closed-loop feedback mechanisms that connect session errors to targeted flashcards, weak domains to suggested cases, and all interactions to a persistent learner profile. Technical validation confirms algorithmic correctness across eight subsystems. To our knowledge, this is the first system to unify ACT-R, IRT, FSRS, ZPD, and metacognitive calibration in a single medical education platform. Formal learner evaluation via randomized controlled trial is planned.