Durability of Response With Selpercatinib in Patients With RET-Activated Thyroid Cancer: Long-Term Safety and Efficacy From LIBRETTO-001.

Clinical trials frequently include multiple end points that mature at different times. The initial report, typically based on the primary end point, may be published when key planned co-primary or secondary analyses are not yet available. Clinical Trial Updates provide an opportunity to disseminate additional results from studies, published in JCO or elsewhere, for which the primary end point has already been reported.LIBRETTO-001 is a registrational phase I/II, single-arm, open-label study of selpercatinib in patients with RET (REarranged during Transfection)-activated cancers (ClinicalTrials.gov identifier: NCT03157128). We present long-term safety and efficacy from LIBRETTO-001 in patients with RET-mutant medullary thyroid cancer (MTC; n = 324) and RET fusion-positive thyroid cancer encompassing different histological subtypes (TC; n = 66). At the data cutoff of January 2023, the objective response rate was 82.5% among patients with cabozantinib/vandetanib-naïve MTC and 95.8% among patients with treatment-naïve TC. At a median follow-up time of 42.4 and 44.0 months in patients with cabozantinib/vandetanib-naïve and pretreated MTC, the median progression-free survival (PFS) was not reached and 41.4 months, respectively. At a median follow-up time of 24.9 and 30.4 months in patients with treatment-naïve and pretreated TC, the median PFS was not reached and 27.4 months, respectively. Three-year PFS rates were 75.2% and 87.3% among patients with cabozantinib/vandetanib-naïve MTC and treatment-naïve TC, respectively. Median PFS was similar to median duration of response for each patient group. The safety profile of selpercatinib was consistent with previous reports. With an additional follow-up of 37 months and 228 more patients from the last disclosure, selpercatinib continued to provide durable and robust responses in treatment-naïve and previously treated patients with RET-mutant MTC and RET fusion-positive TC.

Therapeutic application of extracellular vesicular EGFR isoform D as a co-drug to target squamous cell cancers with tyrosine kinase inhibitors.

Targeting wild-type epidermal growth factor receptor (EGFR) using tyrosine kinase inhibitors (TKIs) never achieved its purported success in cancers such as head and neck squamous cell carcinoma, which are largely EGFR-dependent. We had previously shown that exceptional responders to TKIs have a genetic aberration that results in overexpression of an EGFR splice variant, isoform D (IsoD). IsoD lacks an integral transmembrane and kinase domain and is secreted in extracellular vesicles (EVs) in TKI-sensitive patient-derived cultures. Remarkably, the exquisite sensitivity to TKIs could be transferred to TKI-resistant tumor cells, and IsoD protein in the EV is necessary and sufficient to transfer the phenotype in vitro and in vivo across multiple models and drugs. This drug response requires an intact endocytic mechanism, binding to full-length EGFR, and signaling through Src-phosphorylation within the endosomal compartment. We propose a therapeutic strategy using EVs containing EGFR IsoD as a co-drug to expand the use of TKI therapy to EGFR-driven cancers.

Tepotinib plus osimertinib in patients with EGFR-mutated non-small-cell lung cancer with MET amplification following progression on first-line osimertinib (INSIGHT 2): a multicentre, open-label, phase 2 trial.

BACKGROUND: Patients with EGFR-mutated non-small-cell lung cancer (NSCLC) and MET amplification as a mechanism of resistance to first-line osimertinib have few treatment options. Here, we report the primary analysis of the phase 2 INSIGHT 2 study evaluating tepotinib, a highly selective MET inhibitor, combined with osimertinib in this population. METHODS: This open-label, phase 2 study was conducted at 179 academic centres and community clinics in 17 countries. Eligible patients were aged 18 years or older with an Eastern Cooperative Oncology Group performance status of 0 or 1 and advanced or metastatic EGFR-mutated NSCLC of any histology, with MET amplification by tissue biopsy fluorescence in-situ hybridisation (FISH; MET gene copy number of ≥5 or MET-to-CEP7 ratio of ≥2) or liquid biopsy next-generation sequencing (MET plasma gene copy number of ≥2·3), following progression on first-line osimertinib. Patients received oral tepotinib 500 mg plus oral osimertinib 80 mg once daily. The primary endpoint was independently assessed objective response in patients with MET amplification by central FISH treated with tepotinib plus osimertinib with at least 9 months of follow-up. Safety was analysed in patients who received at least one study drug dose. This study is registered with ClinicalTrials.gov, NCT03940703 (enrolment complete). FINDINGS: Between Feb 13, 2020, and Nov 4, 2022, 128 patients (74 [58%] female, 54 [42%] male) were enrolled and initiated tepotinib plus osimertinib. The primary activity analysis population included 98 patients with MET amplification confirmed by central FISH, previous first-line osimertinib and at least 9 months of follow-up (median 12·7 months [IQR 9·9-20·3]). The confirmed objective response rate was 50·0% (95% CI 39·7-60·3; 49 of 98 patients). The most common treatment-related grade 3 or worse adverse events were peripheral oedema (six [5%] of 128 patients), decreased appetite (five [4%]), prolonged electrocardiogram QT interval (five [4%]), and pneumonitis (four [3%]). Serious treatment-related adverse events were reported in 16 (13%) patients. Deaths of four (3%) patients were assessed as potentially related to either trial drug by the investigator due to pneumonitis (two [2%] patients), decreased platelet count (one [1%]), respiratory failure (one [1%]), and dyspnoea (one [1%]); one death was attributed to both pneumonitis and dyspnoea. INTERPRETATION: Tepotinib plus osimertinib showed promising activity and acceptable safety in patients with EGFR-mutated NSCLC and MET amplification as a mechanism of resistance to first-line osimertinib, suggesting a potential chemotherapy-sparing oral targeted therapy option that should be further investigated. FUNDING: Merck (CrossRef Funder ID: 10.13039/100009945).

Synergy of Oxygen Octahedra Distortion and Polar Nanodomains Induced Emergent Electrocaloric Effect in NaNbO3-Based Ceramics.

High-performance electrocaloric materials are essential for the development of solid-state cooling technologies; however, the contradiction of the electrocaloric effect (ECE) and temperature span in ferroelectrics frustrates practical applications. In this work, through modulating oxygen octahedra distortion and short-range polar nanodomains with moderate coupling strength, an EC value of ΔT ∼ 0.30 K with an ultrawide temperature span of 85 K is obtained in the x = 0.04 composition [(0.88 - x)NaNbO3-0.12BaTiO3-xLiSbO3 (x = 0-0.06)]. The LiSbO3 dopant induces a P4bm-to-R3cH phase transition and intensifies the oxygen octahedra distortion degree, accompanied by the ferroelectric domain smashing into polar nanodomains. Also, LiSbO3 addition enhances the relaxation degree with a downshift of Tfd (ferroelectric-to-diffuse phase transition temperature) and TJ (temperature of the maximal current density value), and Tfd is shifted to near room temperature with an absence of TJ in x = 0.04. Local energy barriers induced by oxygen octahedra distortion inhibit the phase transition in conjunction with activation of short-range polar order switching under thermal stimuli, which is the underlying mechanism for an excellent EC performance for x = 0.04. This work not only clarifies that ferroelectrics with oxygen octahedra distortion and short-range polar order are expected to achieve remarkable EC performances but also provides a design strategy to seek emergent EC behaviors in complex oxygen-octahedra-distortion materials.

Capmatinib plus nazartinib in patients with EGFR-mutated non-small cell lung cancer.

PURPOSE: This phase 1b/2 trial evaluated the efficacy and safety of capmatinib plus nazartinib in patients with advanced EGFR-mutated non-small cell lung cancer (NSCLC). METHODS: In phase 1b, patients with progression on first-/second-generation EGFR-TKIs received escalating doses of capmatinib 200-400 mg bid plus nazartinib 50-150 mg qd. Once the MTD/RP2D was declared, phase 2 commenced with patient enrollment into groups according to mutation status and prior lines of treatment: group 1 (fasted; EGFR-TKI resistant; 1-3 prior lines; EGFRL858R/ex19del; any T790M/MET); group 2 (fasted; EGFR-TKI naïve; 0-2 prior lines; de novo T790M+; any MET); group 3 (fasted; treatment-naïve; EGFRL858R/ex19del; T790M-; any MET); group 4 (with food; 0-2 prior lines; EGFRL858R/ex19del; any T790M/MET). Primary endpoints in phase 2 were investigator-assessed overall response rate (ORR) per RECIST v1.1 (groups 1-3), safety, and tolerability of the combination with food (group 4). Efficacy was assessed by T790M and MET status for a subgroup of patients. RESULTS: The RP2D was capmatinib 400 mg bid plus nazartinib 100 mg qd. In phase 2 (n = 144), the ORR was 28.8 %, 33.3 %, 61.7 %, and 42.9 % in groups 1 (n = 52), 2 (n = 3), 3 (n = 47), and 4 (n = 42), respectively. In group 1 +phase 1b RP2D, the ORR was 45.8 %, 26.2 %, 37.9 %, and 32.4 % in MET+ (n = 24), MET- (n = 42), T790M+ (n = 29), and T790M- (n = 34) patients. Most common any-grade treatment-related adverse events (≥25 %; n = 144) were peripheral edema (54.9 %), nausea (41.7 %), diarrhea (34.0 %), and maculopapular rash (25.0 %). CONCLUSION: Capmatinib plus nazartinib showed antitumor activity in patients with EGFR-TKI-resistant, EGFR-mutated NSCLC. The overall safety profile was acceptable. CLINICAL TRIAL REGISTRATION: ClinicalTrials.gov NCT02335944.

In-Plane Mesoporous 3D Flower-Like Mo2Ti2C3Clx MXene Anodes for Li-Ion Batteries: From Structure to Performance.

MXenes are known for their exceptional electrical conductivity and surface functionality, gaining interest as promising anode materials for Li-ion batteries. However, conventional 2D multilayered MXenes often exhibit limited electrochemical applicability due to slow ion transport kinetics and low structural stability. Addressing these challenges, this study develops a 3D flower-type double transition metal MXene, Mo2Ti2C3Clx, with precisely engineered in-plane mesoporosity using HF-free Lewis acid-assisted molten salt method, coupled with intercalation and freeze-drying. The molar ratio of Lewis acid to eutectic salts is meticulously controlled to create the mesoporosity, which is preserved through freeze-drying. Molecular dynamics (MD) simulations assess the impact of in-plane pore size on the structure and transport dynamics of electrolyte components. Density functional theory (DFT) shows that chlorine surface functional groups significantly reduce Li-ion diffusion barriers, thereby enhancing ion transport and battery performance. Electrochemical evaluations reveal that small-sized (2-5 nm) mesoporous Mo2Ti2C3Clx achieves a specific capacity of 324 mAh g-1 at 0.2 A g-1 and maintains 97% capacity after 500 cycles at 0.5 A g-1, outperforming larger-pored (10 nm) and non-porous variants. This research highlights a scalable strategy for designing mesoporous materials that optimize ion transport and structural stability, essential for advancing next-generation high-performance energy storage solutions.

Integrated care for children and young people with special health and care needs: a systematic review.

CONTEXT: There is a dearth of high-quality evidence on integrated, coordinated and cost-effective care for children with special health and care needs (CSHCN). OBJECTIVE: To assess the effectiveness of integrated/coordinated care models for CSHCN. DATA SOURCES: Embase, Ovid Medline(R), HMIC Health Management Information Consortium, Maternity & Infant Care Database (MIDIRS), PsycARTICLES, PsycINFO, Social Policy and Practice, Cochrane Central Register of Controlled Trials (CENTRAL), Global Health and PubMed. STUDY SELECTION: Inclusion criteria comprised (1) randomised trials, including cluster randomised trials; (2) an integrated/coordinated care intervention; (3) for children and young people under 25 with special healthcare needs including medical complexity; (4) assessing child-centred outcomes, health-related quality of life among parents and carers, and health or social care use, processes of care and satisfaction with care. DATA EXTRACTION: Data were extracted and assessed by two researchers, and descriptive data were synthesised according to outcome and intervention. RESULTS: 14 randomised controlled studies were included. Seven out of the 14 studies had a dedicated key worker/care coordinator as a vital part of the integrated/coordinated care intervention; however, the certainty of evidence for all outcomes was either 'low' or 'very low'. LIMITATIONS: Included studies were mostly from high-income countries. Variable study outcomes and quality of evidence precluded meta-analysis. CONCLUSIONS: Limited evidence favours integrated care for CSHCN using a dedicated key worker/care coordinator; however, heterogeneity in study outcomes and definitions of CSHCN limit the strength and utility of evidence obtained. Recommendations are made for improving integrated care practice, research and evaluation which are important for evidence-based health services for CSHCN. PROSPERO REGISTRATION NUMBER: CRD42020209320.

PARP4 interacts with hnRNPM to regulate splicing during lung cancer progression.

BACKGROUND: The identification of cancer driver genes from sequencing data has been crucial in deepening our understanding of tumor biology and expanding targeted therapy options. However, apart from the most commonly altered genes, the mechanisms underlying the contribution of other mutations to cancer acquisition remain understudied. Leveraging on our whole-exome sequencing of the largest Asian lung adenocarcinoma (LUAD) cohort (n = 302), we now functionally assess the mechanistic role of a novel driver, PARP4. METHODS: In vitro and in vivo tumorigenicity assays were used to study the functional effects of PARP4 loss and mutation in multiple lung cancer cell lines. Interactomics analysis by quantitative mass spectrometry was conducted to identify PARP4's interaction partners. Transcriptomic data from cell lines and patient tumors were used to investigate splicing alterations. RESULTS: PARP4 depletion or mutation (I1039T) promotes the tumorigenicity of KRAS- or EGFR-driven lung cancer cells. Disruption of the vault complex, with which PARP4 is commonly associated, did not alter tumorigenicity, indicating that PARP4's tumor suppressive activity is mediated independently. The splicing regulator hnRNPM is a potentially novel PARP4 interaction partner, the loss of which likewise promotes tumor formation. hnRNPM loss results in splicing perturbations, with a propensity for dysregulated intronic splicing that was similarly observed in PARP4 knockdown cells and in LUAD cohort patients with PARP4 copy number loss. CONCLUSIONS: PARP4 is a novel modulator of lung adenocarcinoma, where its tumor suppressive activity is mediated not through the vault complex-unlike conventionally thought, but in association with its novel interaction partner hnRNPM, thus suggesting a role for splicing dysregulation in LUAD tumorigenesis.

Training immunophenotyping deep learning models with the same-section ground truth cell label derivation method improves virtual staining accuracy.

Introduction: Deep learning (DL) models predicting biomarker expression in images of hematoxylin and eosin (H&E)-stained tissues can improve access to multi-marker immunophenotyping, crucial for therapeutic monitoring, biomarker discovery, and personalized treatment development. Conventionally, these models are trained on ground truth cell labels derived from IHC-stained tissue sections adjacent to H&E-stained ones, which might be less accurate than labels from the same section. Although many such DL models have been developed, the impact of ground truth cell label derivation methods on their performance has not been studied. Methodology: In this study, we assess the impact of cell label derivation on H&E model performance, with CD3+ T-cells in lung cancer tissues as a proof-of-concept. We compare two Pix2Pix generative adversarial network (P2P-GAN)-based virtual staining models: one trained with cell labels obtained from the same tissue section as the H&E-stained section (the 'same-section' model) and one trained on cell labels from an adjacent tissue section (the 'serial-section' model). Results: We show that the same-section model exhibited significantly improved prediction performance compared to the 'serial-section' model. Furthermore, the same-section model outperformed the serial-section model in stratifying lung cancer patients within a public lung cancer cohort based on survival outcomes, demonstrating its potential clinical utility. Discussion: Collectively, our findings suggest that employing ground truth cell labels obtained through the same-section approach boosts immunophenotyping DL solutions.

Exploring Clinical Remission in Moderate Asthma - Perspectives from Asia, the Middle East, and South America.

INTRODUCTION: Clinical remission is a relatively new concept in asthma but recent research initiatives suggest it could be an ambitious and achievable therapeutic target for patients with asthma. METHODS: In this modified Delphi study (comprising two online surveys, completed either side of a virtual scientific workshop), the opinions of a panel of respiratory physicians were evaluated to summarize perspective statements on key therapeutic outcomes and criteria for on-treatment clinical remission in patients with moderate asthma. An agreement threshold was pre-defined as agreement by ≥ 75% of participants. RESULTS: Surveys 1 and 2 were completed by 20 and 18 participants, respectively. Most participants (95%) agreed with the concept of clinical remission in moderate asthma and that this should be a desirable treatment goal (90%). Based on a composite measure of 4-6 desirable therapeutic outcomes, current understanding of clinical remission was considered as 12 months with no exacerbations, no oral corticosteroids, no daytime or night-time asthma symptoms (Asthma Control Test score ≥ 20 or Asthma Control Questionnaire score ≤ 0.75), stable lung function, and no treatment-related adverse events. No agreement was reached on the role of relievers in defining therapeutic outcomes or on the wider use of biomarkers and airway hyperresponsiveness for defining asthma remission in clinical practice. CONCLUSIONS: In line with recent consensus statements from the United States and Europe, there was a high level of agreement on the elements of clinical remission among a panel of respiratory physicians from Asia, the Middle East, and South America. Extension of the concept of clinical remission to patients with moderate asthma was considered aligned with the potential of clinical remission as a goal of therapy.

Real-world outcomes of pemetrexed-platinum chemotherapy plus osimertinib after progression on first-line osimertinib in advanced EGFR-mutated NSCLC.

OBJECTIVES: First-line pemetrexed-platinum chemotherapy + osimertinib(Pem-Plat-Osi) improves progression-free survival as compared to osimertinib alone in advanced epidermal growth factor (EGFR)-mutated non-small cell lung cancer (NSCLC). However, many patients are hesitant to commence chemotherapy upfront. We describe outcomes to Pem-Plat-Osi after first-line osimertinib failure. MATERIALS AND METHODS: Patients with advanced EGFR-mutated (ex19del/L858R) NSCLC who had Pem-Plat-Osi between 1/7/2018-30/9/2023 after progression on first-line osimertinib at National Cancer Centre Singapore, Prince of Wales Hospital and Chinese University of Hong Kong were identified. Key endpoints were time to treatment failure (TTF) and overall survival (OS). RESULTS: A total of 60 patients were included. Median age at diagnosis was 62, 53.3 % (32/60) were male and 76.7 % (46/60) were never smokers. Ex19del comprised 56.7 % (34/60) and L858R 43.3 % (26/60). Baseline central nervous system (CNS) metastases were present in 66.7 % (40/60). Median TTF on osimertinib (TTF1) was 14.4 months(m) and median time to initiation of Pem-Plat-Osi was 41 days(d) (range 0-652) after progression on osimertinib. Partial response (PR) or stable disease to Pem-Plat-Osi was achieved in 81.7 %(49/60). Intracranial disease control was achieved in 90.6 % (29/32) of patients with measurable CNS metastases, including those who did not undergo brain radiotherapy. At median follow up of 31.2 m, median TTF on Pem-Plat-Osi (TTF2) was 6.6 m. Median TTF1 + TTF2 was 23.4 m and median OS was 34.2 m. Survival outcomes were similar comparing ex19del and L858R (median TTF1 + TTF2 21.8 m vs 23.5 m, p = 0.90; median OS 34.2 m vs 36.8 m, p = 0.37) and in patients without/with baseline CNS metastases (median TTF1 + TTF2 21.8 m vs 23.4 m, p = 0.44; median OS 36.2 m vs 31.9 m, p = 0.65). TTF1 duration was not significantly associated with TTF2 (p = 0.76). Patients who started Pem-Plat-Osi within 20d of progression on osimertinib had significantly longer TTF2 as compared to patients who started after 20d (median 8.4 m versus 6.0 months, p = 0.03), which remained statistically significant on multivariable analysis. CONCLUSIONS: Our real-world data supports the efficacy of Pem-Plat-Osi after progression on first-line osimertinib, including L858R and baseline CNS metastases. Chemotherapy initiation within 20d of Osi progression was predictive of superior TTF2.

Gut-liver axis: Potential mechanisms of action of food-derived extracellular vesicles.

Food-derived extracellular vesicles (FEVs) are nanoscale membrane vesicles obtained from dietary materials such as breast milk, plants and probiotics. Distinct from other EVs, FEVs can survive the harsh degrading conditions in the gastrointestinal tract and reach the intestines. This unique feature allows FEVs to be promising prebiotics in health and oral nanomedicine for gut disorders, such as inflammatory bowel disease. Interestingly, therapeutic effects of FEVs have recently also been observed in non-gastrointestinal diseases. However, the mechanisms remain unclear or even mysterious. It is speculated that orally administered FEVs could enter the bloodstream, reach remote organs, and thus exert therapeutic effects therein. However, emerging evidence suggests that the amount of FEVs reaching organs beyond the gastrointestinal tract is marginal and may be insufficient to account for the significant therapeutic effects achieved regarding diseases involving remote organs such as the liver. Thus, we herein propose that FEVs primarily act locally in the intestine by modulating intestinal microenvironments such as barrier integrity and microbiota, thereby eliciting therapeutic impact remotely on the liver in non-gastrointestinal diseases via the gut-liver axis. Likewise, drugs delivered to the gastrointestinal system through FEVs may act via the gut-liver axis. As the liver is the main metabolic hub, the intestinal microenvironment may be implicated in other metabolic diseases. In fact, many patients with non-alcoholic fatty liver disease, obesity, diabetes and cardiovascular disease suffer from a leaky gut and dysbiosis. In this review, we provide an overview of the recent progress in FEVs and discuss their biomedical applications as therapeutic agents and drug delivery systems, highlighting the pivotal role of the gut-liver axis in the mechanisms of action of FEVs for the treatment of gut disorders and metabolic diseases.

Capmatinib plus nivolumab in pretreated patients with EGFR wild-type advanced non-small cell lung cancer.

INTRODUCTION: Dysregulated MET is an established oncogenic driver in non-small cell lung cancer (NSCLC). MET signaling may also suppress anticancer immune responses. Concomitant MET inhibition with capmatinib (a MET inhibitor) synergistically enhanced the efficacy of immunotherapies in murine cancer models, regardless of tumor dependency to MET signaling. Here, we report results of a multicenter, open-label, phase 2 study of capmatinib plus nivolumab (a PD-1 inhibitor) in patients with EGFR wild-type advanced NSCLC, previously treated with platinum-based chemotherapy. METHODS: Patients were allocated into high-MET or low-MET groups according to MET expression determined by immunohistochemistry, MET gene copy number as assessed by fluorescence in-situ hybridization, and presence of MET exon 14 skipping mutation, then received capmatinib 400 mg, oral, twice daily in combination with nivolumab 3 mg/kg intravenously every 2 weeks. The primary endpoint was investigator-assessed 6-month progression-free survival (PFS) rate per RECIST v1.1. RESULTS: The primary endpoint was met in both the high-MET (N = 16) and low-MET (N = 30) groups. In the high-MET and low-MET groups, respectively, the estimated mean 6-month PFS rate (95 % credible interval) by Bayesian analysis was 68.9 % (48.5-85.7) and 50.9 % (35.6-66.4). The Kaplan-Meier median PFS (95 % CI) was 6.2 months (3.5-19.2) and 4.2 months (1.8-7.4). The overall response rate (95 % CI) was 25.0 % (7.3-52.4) and 16.7 % (5.6-34.7). Most frequent treatment-related adverse events (≥30 % any grade, N = 46) were nausea (52.2 %), peripheral edema (34.8 %), and increased blood creatinine (30.4 %). CONCLUSIONS: Capmatinib plus nivolumab showed clinical activity and manageable safety in pretreated patients with advanced EGFR wild-type NSCLC, independent of MET status. TRIAL REGISTRATION: ClinicalTrials.gov NCT02323126.

Transcriptome Deconvolution Reveals Absence of Cancer Cell Expression Signature in Immune Checkpoint Blockade Response.

Immune checkpoint therapy (ICB) has conferred significant and durable clinical benefit to some patients with cancer. However, most patients do not respond to ICB, and reliable biomarkers of ICB response are needed to improve patient stratification. Here, we performed a transcriptome-wide meta-analysis across 1,486 tumors from ICB-treated patients and tumors with expected ICB outcomes based on microsatellite status. Using a robust transcriptome deconvolution approach, we inferred cancer- and stroma-specific gene expression differences and identified cell-type specific features of ICB response across cancer types. Consistent with current knowledge, stromal expression of CXCL9, CXCL13, and IFNG were the top determinants of favorable ICB response. In addition, we identified a group of potential immune-suppressive genes, including FCER1A, associated with poor response to ICB. Strikingly, PD-L1 expression in stromal cells, but not cancer cells, is correlated with ICB response across cancer types. Furthermore, the unbiased transcriptome-wide analysis failed to identify cancer-cell intrinsic expression signatures of ICB response conserved across tumor types, suggesting that cancer cells lack tissue-agnostic transcriptomic features of ICB response. SIGNIFICANCE: Our results challenge the prevailing dogma that cancer cells present tissue-agnostic molecular markers that modulate immune activity and ICB response, which has implications on the development of improved ICB diagnostics and treatments.

Use of Natural Language Processing to Infer Sites of Metastatic Disease From Radiology Reports at Scale.

PURPOSE: To evaluate natural language processing (NLP) methods to infer metastatic sites from radiology reports. METHODS: A set of 4,522 computed tomography (CT) reports of 550 patients with 14 types of cancer was used to fine-tune four clinical large language models (LLMs) for multilabel classification of metastatic sites. We also developed an NLP information extraction (IE) system (on the basis of named entity recognition, assertion status detection, and relation extraction) for comparison. Model performances were measured by F1 scores on test and three external validation sets. The best model was used to facilitate analysis of metastatic frequencies in a cohort study of 6,555 patients with 53,838 CT reports. RESULTS: The RadBERT, BioBERT, GatorTron-base, and GatorTron-medium LLMs achieved F1 scores of 0.84, 0.87, 0.89, and 0.91, respectively, on the test set. The IE system performed best, achieving an F1 score of 0.93. F1 scores of the IE system by individual cancer type ranged from 0.89 to 0.96. The IE system attained F1 scores of 0.89, 0.83, and 0.81, respectively, on external validation sets including additional cancer types, positron emission tomography-CT ,and magnetic resonance imaging scans, respectively. In our cohort study, we found that for colorectal cancer, liver-only metastases were higher in de novo stage IV versus recurrent patients (29.7% v 12.2%; P < .001). Conversely, lung-only metastases were more frequent in recurrent versus de novo stage IV patients (17.2% v 7.3%; P < .001). CONCLUSION: We developed an IE system that accurately infers metastatic sites in multiple primary cancers from radiology reports. It has explainable methods and performs better than some clinical LLMs. The inferred metastatic phenotypes could enhance cancer research databases and clinical trial matching, and identify potential patients for oligometastatic interventions.

Towards proactive palliative care in oncology: developing an explainable EHR-based machine learning model for mortality risk prediction.

BACKGROUND: Ex-ante identification of the last year in life facilitates a proactive palliative approach. Machine learning models trained on electronic health records (EHR) demonstrate promising performance in cancer prognostication. However, gaps in literature include incomplete reporting of model performance, inadequate alignment of model formulation with implementation use-case, and insufficient explainability hindering trust and adoption in clinical settings. Hence, we aim to develop an explainable machine learning EHR-based model that prompts palliative care processes by predicting for 365-day mortality risk among patients with advanced cancer within an outpatient setting. METHODS: Our cohort consisted of 5,926 adults diagnosed with Stage 3 or 4 solid organ cancer between July 1, 2017, and June 30, 2020 and receiving ambulatory cancer care within a tertiary center. The classification problem was modelled using Extreme Gradient Boosting (XGBoost) and aligned to our envisioned use-case: "Given a prediction point that corresponds to an outpatient cancer encounter, predict for mortality within 365-days from prediction point, using EHR data up to 365-days prior." The model was trained with 75% of the dataset (n = 39,416 outpatient encounters) and validated on a 25% hold-out dataset (n = 13,122 outpatient encounters). To explain model outputs, we used Shapley Additive Explanations (SHAP) values. Clinical characteristics, laboratory tests and treatment data were used to train the model. Performance was evaluated using area under the receiver operating characteristic curve (AUROC) and area under the precision-recall curve (AUPRC), while model calibration was assessed using the Brier score. RESULTS: In total, 17,149 of the 52,538 prediction points (32.6%) had a mortality event within the 365-day prediction window. The model demonstrated an AUROC of 0.861 (95% CI 0.856-0.867) and AUPRC of 0.771. The Brier score was 0.147, indicating slight overestimations of mortality risk. Explanatory diagrams utilizing SHAP values allowed visualization of feature impacts on predictions at both the global and individual levels. CONCLUSION: Our machine learning model demonstrated good discrimination and precision-recall in predicting 365-day mortality risk among individuals with advanced cancer. It has the potential to provide personalized mortality predictions and facilitate earlier integration of palliative care.