Artificial Intelligence-Assisted Cancer Status Detection in Radiology Reports.

Cancer research is dependent on accurate and relevant information of patient's medical journey. Data in radiology reports are of extreme value but lack consistent structure for direct use in analytics. At Memorial Sloan Kettering Cancer Center (MSKCC), the radiology reports are curated using gold-standard approach of using human annotators. However, the manual process of curating large volume of retrospective data slows the pace of cancer research. Manual curation process is sensitive to volume of reports, number of data elements and nature of reports and demand appropriate skillset. In this work, we explore state of the art methods in artificial intelligence (AI) and implement end-to-end pipeline for fast and accurate annotation of radiology reports. Language models (LM) are trained using curated data by approaching curation as multiclass or multilabel classification problem. The classification tasks are to predict multiple imaging scan sites, presence of cancer and cancer status from the reports. The trained natural language processing (NLP) model classifiers achieve high weighted F1 score and accuracy. We propose and demonstrate the use of these models to assist in the manual curation process which results in higher accuracy and F1 score with lesser time and cost, thus improving efforts of cancer research. SIGNIFICANCE: Extraction of structured data in radiology for cancer research with manual process is laborious. Using AI for extraction of data elements is achieved using NLP models' assistance is faster and more accurate.

Health Care Resource Use for Modern First-Line Treatments in Metastatic Renal Cell Carcinoma.

Importance: Immuno-oncology agents have changed the treatment paradigm for metastatic renal cell carcinoma (mRCC). Such therapies improve survival but can impose considerable health care resource use (HCRU) and associated costs, necessitating their examination. Objective: To compare HCRU, costs, and clinical outcomes among patients receiving first-line pembrolizumab plus axitinib (P+A) or ipilimumab plus nivolumab (I+N). Design, Setting, and Participants: This retrospective cohort study used data from an administrative claims database on patients with mRCC receiving first-line P+A or I+N that was initiated between January 2018 and May 2020. Data were analyzed from February 2021 to July 2022. Exposure: First-line P+A or I+N. Main Outcome and Measures: HCRU and costs during the first 90 days, full first-line treatment, and full follow-up periods were assessed. Using Kaplan-Meier analysis, time on treatment, overall survival, time to first emergency department (ED) visit, and time to first inpatient stay were compared. Results: Among 507 patients, there were 126 patients receiving P+A (91 male [72.2%]; mean [SD] age, 67.93 [9.66] y) and 381 patients receiving I+N (271 male [71.1%]; mean [SD] age, 66.52 [9.94] years). The median time on treatment was longer for the P+A compared with I+N group (12.4 months [95% CI, 8.40 months to not estimable] vs 4.1 months [95% CI, 3.07 to 5.30 months]; P < .001). The median time to first ED visit was longer for the P+A than I+N group (7.2 months [95% CI 3.9 to 11.1 months ] vs 3.3 months [95% CI, 2.6 to 3.9 months]; P = .005), as was time to first inpatient stay (9.0 months [95% CI 6.5 months to not estimable] vs 5.6 months [95% CI, 3.9 to 7.9 months]; P = .02). During the first 90 days, a lower proportion of the P+A than N+I group had ED visits (43 patients [34.1%] vs 182 patients [47.8%] and inpatient stays (24 patients [19.1%) vs144 patients [37.8%]; P < .001). During full follow-up, mean total adjusted costs were similar for P+A and I+N groups, but adjusted 12-month estimated total costs were higher for P+A than I+N groups ($325 574 vs $ 263 803; P = .03). Conclusions and Relevance: In this study, treatment with P+A was associated with longer time on treatment, time to first ED visit, and inpatient stay, while 12-month estimated costs were higher for the P+A group. This is among the first clinical studies to evaluate economic burden associated with modern treatments for mRCC.

Cabozantinib Plus Nivolumab in Patients with Non-Clear Cell Renal Cell Carcinoma: Updated Results from a Phase 2 Trial.

Treatment options are limited for patients with non-clear cell renal cell carcinoma (nccRCC). Patients with nccRCC experienced a favorable objective response rate (ORR) in a phase 2 trial of cabozantinib plus nivolumab. We now report updated efficacy and safety results at median follow-up of 34 mo for patients with papillary, unclassified, or translocation-associated RCC. Cabozantinib and nivolumab were administered at standard doses to patients with metastatic nccRCC that had progressed on zero or one line of systemic therapy. The primary endpoint was the ORR according to Response Evaluation Criteria in Solid Tumors v1.1. Secondary endpoints included progression-free survival (PFS), overall survival (OS), and adverse events. Forty patients were treated. At median follow-up of 34 mo for survivors, the ORR was 48% (95% confidence interval [CI] 31.5-63.9%). Median PFS was 13 mo (95% CI 7-16); the 12-mo and 24-mo PFS rates were 51% (95% CI 34-65%) and 23% (95% CI 11-37%), respectively. Median OS was 28 mo (95% CI 23-43); the 18-mo and 36-mo OS rates were 70% (95% CI 53-82%) and 44% (95% CI 28-60%), respectively. No new safety signals were seen with cabozantinib and nivolumab. This extended follow-up analysis demonstrates promising efficacy, and highlights the potential for sustained responses with cabozantinib plus nivolumab in patients with metastatic nccRCC.

Clinical Outcomes of Immune Checkpoint Inhibitors in Unique Cohorts Underrepresented in Clinical Trials.

Regulatory approval of immune checkpoint inhibitors (ICIs) was based on results of large, randomized clinical trials, resulting in limited outcomes data in patient cohorts typically underrepresented in such trials. The objective of this study was to evaluate the efficacy and safety of ICIs in these unique patient cohorts. This is a multicenter, retrospective analysis of real-world data at six academic and community clinics in the United States from 1 January 2011 to 1 April 2018. Patients were included if they had received at least one cycle of ICI treatment. Unique patient cohorts included age > 75 years, non-White race, positive smoking history, ECOG performance status (PS) ≥ 2, BMI ≥ 30 kg/m2, autoimmune diseases (AIDs), chronic viral infections (CVI), extensive prior lines of therapy (LOTs), or >three metastatic sites. Immune-related adverse events (irAEs), overall survival (OS), and time to treatment failure were evaluated in the entire cohort and in NSCLC patients treated with PD-(L)1 monotherapy. Outcomes and their association with unique patient cohorts were compared on univariate analysis and multivariate analysis to those without a particular characteristic in the entire NSCLC PD-(L)1 monotherapy cohorts. In total, 1453 patients were included: 56.5%-smokers, 30.4%-non-White, 22.8%-elderly, 20.8%-ECOG PS ≥ 2, 15.7%-history of AIDs, and 4.7%-history of CVI. The common ICIs were nivolumab (37.1%) and pembrolizumab (22.2%). Black patients, compared to White patients, experienced fewer irAEs (OR 0.54, p < 0.001). An ECOG PS of ≥2 (HR = 2.01, p < 0.001) and an increased number of previous LOTs were associated with poor OS (the median OS of 26.2 vs. 16.2 vs. 9.6 months for one vs. two vs. three prior LOTs, p < 0.001). The above results were confirmed in anti-PD-(L)1 monotherapy non-small cell lung cancer patients (n = 384). Overall, ICIs were safe and efficacious in these typically underrepresented patient cohorts. We noted ECOG PS ≥ 2 and an increased prior LOTs were associated with poor ICI efficacy, and Black patients, compared to White patients, experienced fewer irAEs.