Keegan, Theresa H M

PURPOSE Survivors of adolescent and young adult (AYA)-onset cancers require comprehensive cancer surveillance care. Guidelines recommend 1-4 clinician visits annually for 5 years depending on stage and disease. The goal of this study was to identify factors associated with patterns of post-treatment primary and oncology care in a large cohort of survivors of AYA-onset cancers diagnosed within an integrated health care system. METHODS Patients diagnosed with cancer between 2006 and 2020 age 15-39 years were included. Surveillance visits were identified from electronic medical records 2-5 years after diagnosis. Multivariable logistic regression was used to assess associations with zero oncology specialty visits. RESULTS Of 7,925 survivors, 46% were Hispanic, 6% non-Hispanic Black, 11% non-Hispanic Asian, 35% non-Hispanic White, and 65% female. One-quarter had no oncology specialty visits in the first surveillance year rising to 38% in year 5; 31% had 3+ visits in the first year, declining to 13% in year 5. Over the surveillance period, 17% did not have any oncology specialty visits and 6% had no primary care or oncology visits. Those who were male (odds ratio [OR], 1.21 [95% CI, 1.02 to 1.45]), 20-24 years at diagnosis (OR, 1.58, 95% CI, 1.27 to 1.9; v 35 to 39), or non-Hispanic Black (OR, 1.38, 95% CI, 1.05 to 1.82; v non-Hispanic White) or had high-deductible commercial or Medicaid insurance (OR, 1.35, 95% CI, 1.15 to 1.59; OR, 1.42, 95% CI, 1.11 to 1.82, respectively; v no or low deductible commercial) were more likely to be in the 0-visit group for oncology specialty care. CONCLUSION Although the majority of this cohort received either primary or oncology specialty care, our study identifies those who may need tailored outreach for follow-up care. Continued research into development and testing of interventions to improve uptake of appropriate health care for survivors of AYA-onset cancers is critical.

PURPOSE Chemotherapy-induced peripheral neuropathy is a potentially debilitating adverse effect of cancer therapy that can lead to delay, reduction, or discontinuation of cancer treatment. Population-based data on peripheral neuropathy incidence among young cancer survivors are lacking. METHODS Using a linkage between Medicaid, the California Cancer Registry, and hospitalization and emergency department data, we identified 6,028 adolescents and young adults (15-39 years) with Hodgkin lymphoma (HL), non-Hodgkin lymphoma (NHL), breast cancer, colorectal cancer, or testicular cancer and 418 children (<15 years) with HL or NHL during 2005-2017. We determined the cumulative incidence of peripheral neuropathy and its association with neurotoxic chemotherapy identified from Medicaid claims, using multivariable Cox proportional hazards regression models. RESULTS Of 6,446 patients, 1,007 were diagnosed with peripheral neuropathy. Across each cancer type, incidence was higher among patients receiving neurotoxic drugs. For example, compared with non-neurotoxic agents, 5-year cumulative incidence was higher with oxaliplatin for colorectal cancer (24.0% v 6.2%) and paclitaxel for breast cancer (22.6% v 5.1%). In multivariable analysis, the agents most strongly associated with peripheral neuropathy were brentuximab (±other neurotoxic drugs) for HL (hazard ratio [HR], 9.53 [95% CI, 5.95 to 15.26]); brentuximab (±vinca alkaloids; HR, 7.00 [95% CI, 4.13 to 11.87]) for NHL, paclitaxel for breast cancer (HR, 4.03 [95% CI, 3.05 to 5.31]); oxaliplatin for colorectal cancer (HR, 3.46 [95% CI, 2.23 to 5.36]); and cisplatin and etoposide for testicular cancer (HR, 2.06 [95% CI, 1.37 to 3.11]). CONCLUSION The high incidence of peripheral neuropathy highlights the need for frequent monitoring, new supportive care approaches, and development of novel therapeutic agents to minimize toxicity while maintaining treatment efficacy.

PURPOSE Patient-centered outcomes (PCOs) are pivotal in cancer treatment, as they directly reflect patients' quality of life. Although multiple studies suggest that factors affecting breast cancer–related morbidity and survival are influenced by treatment side effects and adherence to long-term treatment, such data are generally only available on a smaller scale or from a single center. The primary challenge with collecting these data is that the outcomes are captured as free text in clinical narratives written by clinicians. MATERIALS AND METHODS Given the complexity of PCO documentation in these narratives, computerized methods are necessary to unlock the wealth of information buried in unstructured text notes that often document PCOs. Inspired by the success of large language models (LLMs), we examined the adaptability of three LLMs, GPT-2, BioGPT, and PMC-LLaMA, on PCO tasks across three institutions, Mayo Clinic, Emory University Hospital, and Stanford University. We developed an open-source framework for fine-tuning LLM that can directly extract the five different categories of PCO from the clinic notes. RESULTS We found that these LLMs without fine-tuning (zero-shot) struggle with challenging PCO extraction tasks, displaying almost random performance, even with some task-specific examples (few-shot learning). The performance of our fine-tuned, task-specific models is notably superior compared with their non–fine-tuned LLM models. Moreover, the fine-tuned GPT-2 model has demonstrated a significantly better performance than the other two larger LLMs. CONCLUSION Our discovery indicates that although LLMs serve as effective general-purpose models for tasks across various domains, they require fine-tuning when applied to the clinician domain. Our proposed approach has the potential to lead more efficient, adaptable models for PCO information extraction, reducing reliance on extensive computational resources while still delivering superior performance for specific tasks.