Projected environmental and public health benefits of extended-interval dosing: an analysis of pembrolizumab use in a US national health system.

BACKGROUND: Health care is a major source of greenhouse gas emissions, leading to climate change and public health harms. Changes are needed to improve the environmental sustainability of health-care practices, but such changes should not sacrifice patient outcomes or financial sustainability. Alternative dosing strategies that reduce the frequency with which specialty drugs are administered, without sacrificing patient outcomes, are an attractive possibility for improving environmental sustainability. We sought to inform environmentally sustainable cancer care by estimating and comparing the environmental and financial effects of alternative, clinically equivalent strategies for pembrolizumab administration. METHODS: We conducted a retrospective analysis using a cohort of patients from the Veterans Health Administration (VHA) in the USA who received one or more pembrolizumab doses between May 1, 2020, and Sept 30, 2022. Using baseline, real-world administration of pembrolizumab, we generated simulated pembrolizumab use data under three near-equivalent counterfactual pembrolizumab administration strategies defined by combinations of weight-based dosing, pharmacy-level vial sharing and dose rounding, and extended-interval dosing (ie, every 6 weeks). For each counterfactual dosing strategy, we estimated greenhouse gas emissions related to pembrolizumab use across the VHA cohort using a deterministic environmental impact model that estimated greenhouse gas emissions due to patient travel, drug manufacture, and medical waste as the primary outcome measure. FINDINGS: We identified 7813 veterans who received at least one dose of pembrolizumab-containing therapy in the VHA during the study period. 59 140 pembrolizumab administrations occurred in the study period, of which 46 255 (78·2%) were dosed at 200 mg every 3 weeks, 12 885 (21·8%) at 400 mg every 6 weeks, and 14 955 (25·3%) were coadministered with infusional chemotherapies. Adoption of weight-based, extended-interval pembrolizumab dosing (4 mg/kg every 6 weeks) and pharmacy-level stewardship strategies (ie, dose rounding and vial sharing) for all pembrolizumab infusions would have resulted in 24·7% fewer administration events than baseline dosing (44 533 events vs 59 140 events) and an estimated 200 metric tons less CO2 emitted per year as a result of pembrolizumab use within the VHA (650 tons vs 850 tons of CO2, a relative reduction of 24%), largely due to reductions in distance travelled by patients to receive treatment. Similar results were observed when weight-based and extended-interval dosing were applied only to pembrolizumab monotherapy and pembrolizumab in combination with oral therapies. INTERPRETATION: Alternative pembrolizumab administration strategies might have environmental advantages over the current dosing and compounding paradigms. Specialty medication dosing can be optimised for health-care spending and environmental sustainability without sacrificing clinical outcomes. FUNDING: None.

Impact of lung cancer screening on stage migration and mortality among the national Veterans Health Administration population with lung cancer.

BACKGROUND: Despite randomized trials demonstrating a mortality benefit to low-dose computed tomography screening to detect lung cancer, uptake of lung cancer screening (LCS) has been slow, and the benefits of screening remain unclear in clinical practice. METHODS: This study aimed to assess the impact of screening among patients in the Veterans Health Administration (VA) health care system diagnosed with lung cancer between 2011 and 2018. Lung cancer stage at diagnosis, lung cancer-specific survival, and overall survival between patients with cancer who did and did not receive screening before diagnosis were evaluated. We used Cox regression modeling and inverse propensity weighting analyses with lead time bias adjustment to correlate LCS exposure with patient outcomes. RESULTS: Of 57,919 individuals diagnosed with lung cancer in the VA system between 2011 and 2018, 2167 (3.9%) underwent screening before diagnosis. Patients with screening had higher rates of stage I diagnoses (52% vs. 27%; p ≤ .0001) compared to those who had no screening. Screened patients had improved 5-year overall survival rates (50.2% vs. 27.9%) and 5-year lung cancer-specific survival (59.0% vs. 29.7%) compared to unscreened patients. Among screening-eligible patients who underwent National Comprehensive Cancer Network guideline-concordant treatment, screening resulted in substantial reductions in all-cause mortality (adjusted hazard ratio [aHR], 0.79; 95% confidence interval [CI], 0.67-0.92; p = .003) and lung-specific mortality (aHR, 0.61; 95% CI, 0.50-0.74; p < .001). CONCLUSIONS: While LCS uptake remains limited, screening was associated with earlier stage diagnoses and improved survival. This large national study corroborates the value of LCS in clinical practice; efforts to widely adopt this vital intervention are needed.

Prostate-Specific Antigen Stratification for Predicting Advanced Prostate Cancer Events in Men Approaching Age Limits for Recommended Screening.

PURPOSE: Our goal was to quantify the ability of various PSA values in predicting the likelihood of developing metastatic or fatal prostate cancer in older men. MATERIALS AND METHODS: We used a random sample of patients in the US Veterans Health Administration to identify 80,706 men who had received PSA testing between ages 70 to 75. Our primary end point was time to development of either metastatic prostate cancer or death from prostate cancer. We used cumulative/dynamic modeling to account for competing events (death from non-prostate cancer causes) in studying both the discriminative ability of PSA as well as for positive predictive value and negative predictive value at 3 time points. RESULTS: PSA demonstrated time-dependent predictive discrimination, with receiver operating characteristic AUC at 5, 10, and 14 years decreasing from 0.83 to 0.77 to 0.73, respectively, but without statistically significant difference when stratified by race. At PSA thresholds between 1 and 8 ng/mL, the positive predictive value of developing advanced prostate cancer was significantly greater in Black than White patients. For instance, at a PSA > 3, at 5, 10, and 14 years, White patients had 2.4%, 2.9%, and 3.7% risk of an event, whereas Black patients had 4.3%, 6.5%, and 8.3% risk. CONCLUSIONS: In men aged 70 to 75 deciding whether to cease PSA testing with borderline-elevated PSA values, the risk of developing metastatic or fatal prostate cancer is quantifiable and relatively low. Risk assessment in this setting must account for the higher incidence of prostate cancer in Black men.

An NLP Framework for the Extraction of Concept Measurements from Radiology and Pathology Notes.

Natural language processing (NLP) tools can automate the identification of cancer patients eligible for specific pathways. We developed and validated a cancer agnostic, rules-based NLP framework to extract the dimensions and measurements of several concepts from pathology and radiology reports. This framework was then efficiently and cost-effectively deployed to identify patients eligible for breast, lung, and prostate cancers clinical pathways.

Lung Cancer Survival Trends in the Veterans Health Administration.

INTRODUCTION: Lung cancer survival is improving in the United States. We investigated whether there was a similar trend within the Veterans Health Administration (VHA), the largest integrated healthcare system in the United States. MATERIALS AND METHODS: Data from the Veterans Affairs Central Cancer Registry were analyzed for temporal survival trends using Kaplan-Meier estimates and linear regression. RESULTS: A total number of 54,922 Veterans were identified with lung cancer diagnosed from 2010 to 2017. Histologies were classified as non-small-cell lung cancer (NSCLC) (64.2%), small cell lung cancer (SCLC) (12.9%), and 'other' (22.9%). The proportion with stage I increased from 18.1% to 30.4%, while stage IV decreased from 38.9% to 34.6% (both P < .001). The 3-year overall survival (OS) improved for stage I (58.6% to 68.4%, P < .001), stage II (35.5% to 48.4%, P < .001), stage III (18.7% to 29.4%, P < .001), and stage IV (3.4% to 7.8%, P < .001). For NSCLC, the median OS increased from 12 to 21 months (P < .001), and the 3-year OS increased from 24.1% to 38.3% (P < .001). For SCLC, the median OS remained unchanged (8 to 9 months, P = .10), while the 3-year OS increased from 9.1% to 12.3% (P = .014). Compared to White Veterans, Black Veterans with NSCLC had similar OS (P = .81), and those with SCLC had higher OS (P = .003). CONCLUSION: Lung cancer survival is improving within the VHA. Compared to White Veterans, Black Veterans had similar or higher survival rates. The observed racial equity in outcomes within a geographically and socioeconomically diverse population warrants further investigation to better understand and replicate this achievement in other healthcare systems.

Quantifying the Risk of Technology-Driven Health Disparities in Radiation Oncology.

PURPOSE: New technologies are continuously emerging in radiation oncology. Inherent technological limitations can result in health care disparities in vulnerable patient populations. These limitations must be considered for existing and new technologies in the clinic to provide equitable care. MATERIALS AND METHODS: We created a health disparity risk assessment metric inspired by failure mode and effects analysis. We provide sample patient populations and their potential associated disparities, guidelines for clinics and vendors, and example applications of the methodology. RESULTS: A disparity risk priority number can be calculated from the product of 3 quantifiable metrics: the percentage of patients impacted, the severity of the impact of dosimetric uncertainty or quality of the radiation plan, and the clinical dependence on the evaluated technology. The disparity risk priority number can be used to rank the risk of suboptimal care due to technical limitations when comparing technologies and to plan interventions when technology is shown to have inequitable performance in the patient population of a clinic. CONCLUSIONS: The proposed methodology may simplify the evaluation of how new technology impacts vulnerable populations, help clinics quantify the limitations of their technological resources, and plan appropriate interventions to improve equity in radiation treatments.

Adrenal-Permissive Germline HSD3B1 Allele and Prostate Cancer Outcomes.

Importance: The adrenal androgen-metabolizing 3ß-hydroxysteroid dehydrogenase-1 enzyme, encoded by the HSD3B1 gene, catalyzes the rate-limiting step necessary for synthesizing nontesticular testosterone and dihydrotestosterone production. The common adrenal-permissive HSD3B1(1245C) allele is responsible for encoding the 3ß-HSD1 protein with decreased susceptibility to degradation resulting in higher extragonadal androgen synthesis. Retrospective studies have suggested an association of the HSD3B1 adrenal-permissive homozygous genotype with androgen deprivation therapy resistance in prostate cancer. Objective: To evaluate differences in mortality outcomes by HSD3B1 genetic status among men with prostate cancer. Design, Setting, and Participants: This cohort study of patients with prostate cancer who were enrolled in the Million Veteran Program within the Veterans Health Administration (VHA) system between 2011 and 2023 collected genotyping and phenotyping information. Exposure: HSD3B1 genotype status was categorized as AA (homozygous adrenal-restrictive), AC (heterozygous adrenal-restrictive), or CC (homozygous adrenal-permissive). Main Outcomes and Measures: The primary outcome of this study was prostate cancer-specific mortality (PCSM), defined as the time from diagnosis to death from prostate cancer, censored at the date of last VHA follow-up. Secondary outcomes included incidence of metastases and PCSM in predefined subgroups. Results: Of the 5287 participants (median [IQR] age, 69 [64-74] years), 402 (7.6%) had the CC genotype, 1970 (37.3%) had the AC genotype, and 2915 (55.1%) had the AA genotype. Overall, the primary cause of death for 91 patients (1.7%) was prostate cancer. Cumulative incidence of PCSM at 5 years after prostate cancer diagnosis was higher among men with the CC genotype (4.0%; 95% CI, 1.7%-6.2%) compared with the AC genotype (2.1%; 95% CI, 1.3%-2.8%) and AA genotype (1.9%; 95% CI, 1.3%-2.4%) (P = .02). In the 619 patients who developed metastatic disease at any time, the cumulative incidence of PCSM at 5 years was higher among patients with the CC genotype (36.0%; 95% CI, 16.7%-50.8%) compared with the AC genotype (17.9%; 95% CI, 10.5%-24.7%) and AA genotype (18.5%; 95% CI, 12.0%-24.6%) (P = .01). Conclusions and Relevance: In this cohort study of US veterans undergoing treatment for prostate cancer at the VHA, the HSD3B1 CC genotype was associated with inferior outcomes. The HSD3B1 biomarker may help identify patients who may benefit from therapeutic targeting of 3ß-hydroxysteroid dehydrogenase-1 and the androgen-signaling axis.

Artificial intelligence to unlock real-world evidence in clinical oncology: A primer on recent advances.

PURPOSE: Real world evidence is crucial to understanding the diffusion of new oncologic therapies, monitoring cancer outcomes, and detecting unexpected toxicities. In practice, real world evidence is challenging to collect rapidly and comprehensively, often requiring expensive and time-consuming manual case-finding and annotation of clinical text. In this Review, we summarise recent developments in the use of artificial intelligence to collect and analyze real world evidence in oncology. METHODS: We performed a narrative review of the major current trends and recent literature in artificial intelligence applications in oncology. RESULTS: Artificial intelligence (AI) approaches are increasingly used to efficiently phenotype patients and tumors at large scale. These tools also may provide novel biological insights and improve risk prediction through multimodal integration of radiographic, pathological, and genomic datasets. Custom language processing pipelines and large language models hold great promise for clinical prediction and phenotyping. CONCLUSIONS: Despite rapid advances, continued progress in computation, generalizability, interpretability, and reliability as well as prospective validation are needed to integrate AI approaches into routine clinical care and real-time monitoring of novel therapies.

Genetic risk and likelihood of prostate cancer detection on first biopsy by ancestry.

Despite differences in prostate cancer risk across ancestry groups, relative performance of prostate cancer genetic risks scores (GRS) for positive biopsy prediction in different ancestry groups is unknown. This cross-sectional retrospective analysis examines the association between a polygenic hazard score (PHS290) and risk of prostate cancer diagnosis upon first biopsy in male veterans using 2-sided tests. Our analysis included 36 717 veterans (10 297 of African ancestry). Unadjusted rates of positive first prostate biopsy increased with higher genetic risk (low risk: 34%, high risk: 58%; P < .001). Among men of African ancestry, higher genetic risk was associated with increased prostate cancer detection on first biopsy (odds ratio = 2.18, 95% confidence interval = 1.93 to 2.47), but the effect was stronger among men of European descent (odds ratio = 3.89, 95% confidence interval = 3.62 to 4.18). These findings suggest that incorporating genetic risk into prediction models could better personalize biopsy decisions, although further study is needed to achieve equitable genetic risk stratification among ancestry groups.

Intricacies of Human-AI Interaction in Dynamic Decision-Making for Precision Oncology: A Case Study in Response-Adaptive Radiotherapy.

Background: Adaptive treatment strategies that can dynamically react to individual cancer progression can provide effective personalized care. Longitudinal multi-omics information, paired with an artificially intelligent clinical decision support system (AI-CDSS) can assist clinicians in determining optimal therapeutic options and treatment adaptations. However, AI-CDSS is not perfectly accurate, as such, clinicians' over/under reliance on AI may lead to unintended consequences, ultimately failing to develop optimal strategies. To investigate such collaborative decision-making process, we conducted a Human-AI interaction case study on response-adaptive radiotherapy (RT). Methods: We designed and conducted a two-phase study for two disease sites and two treatment modalities-adaptive RT for non-small cell lung cancer (NSCLC) and adaptive stereotactic body RT for hepatocellular carcinoma (HCC)-in which clinicians were asked to consider mid-treatment modification of the dose per fraction for a number of retrospective cancer patients without AI-support (Unassisted Phase) and with AI-assistance (AI-assisted Phase). The AI-CDSS graphically presented trade-offs in tumor control and the likelihood of toxicity to organs at risk, provided an optimal recommendation, and associated model uncertainties. In addition, we asked for clinicians' decision confidence level and trust level in individual AI recommendations and encouraged them to provide written remarks. We enrolled 13 evaluators (radiation oncology physicians and residents) from two medical institutions located in two different states, out of which, 4 evaluators volunteered in both NSCLC and HCC studies, resulting in a total of 17 completed evaluations (9 NSCLC, and 8 HCC). To limit the evaluation time to under an hour, we selected 8 treated patients for NSCLC and 9 for HCC, resulting in a total of 144 sets of evaluations (72 from NSCLC and 72 from HCC). Evaluation for each patient consisted of 8 required inputs and 2 optional remarks, resulting in up to a total of 1440 data points. Results: AI-assistance did not homogeneously influence all experts and clinical decisions. From NSCLC cohort, 41 (57%) decisions and from HCC cohort, 34 (47%) decisions were adjusted after AI assistance. Two evaluations (12%) from the NSCLC cohort had zero decision adjustments, while the remaining 15 (88%) evaluations resulted in at least two decision adjustments. Decision adjustment level positively correlated with dissimilarity in decision-making with AI [NSCLC: ρ = 0.53 ( p < 0.001); HCC: ρ = 0.60 ( p < 0.001)] indicating that evaluators adjusted their decision closer towards AI recommendation. Agreement with AI-recommendation positively correlated with AI Trust Level [NSCLC: ρ = 0.59 ( p < 0.001); HCC: ρ = 0.7 ( p < 0.001)] indicating that evaluators followed AI's recommendation if they agreed with that recommendation. The correlation between decision confidence changes and decision adjustment level showed an opposite trend [NSCLC: ρ = -0.24 ( p = 0.045), HCC: ρ = 0.28 ( p = 0.017)] reflecting the difference in behavior due to underlying differences in disease type and treatment modality. Decision confidence positively correlated with the closeness of decisions to the standard of care (NSCLC: 2 Gy/fx; HCC: 10 Gy/fx) indicating that evaluators were generally more confident in prescribing dose fractionations more similar to those used in standard clinical practice. Inter-evaluator agreement increased with AI-assistance indicating that AI-assistance can decrease inter-physician variability. The majority of decisions were adjusted to achieve higher tumor control in NSCLC and lower normal tissue complications in HCC. Analysis of evaluators' remarks indicated concerns for organs at risk and RT outcome estimates as important decision-making factors. Conclusions: Human-AI interaction depends on the complex interrelationship between expert's prior knowledge and preferences, patient's state, disease site, treatment modality, model transparency, and AI's learned behavior and biases. The collaborative decision-making process can be summarized as follows: (i) some clinicians may not believe in an AI system, completely disregarding its recommendation, (ii) some clinicians may believe in the AI system but will critically analyze its recommendations on a case-by-case basis; (iii) when a clinician finds that the AI recommendation indicates the possibility for better outcomes they will adjust their decisions accordingly; and (iv) When a clinician finds that the AI recommendation indicate a worse possible outcome they will disregard it and seek their own alternative approach.