REAL-WORLD TREATMENT PATTERNS AND MACHINE LEARNING PREDICTORS OF LATER-LINE THERAPY IN METASTATIC COLORECTAL CANCER
Author(s)
Vikash K. Verma, MBA, PharmD1, Louis Brooks Jr, MS2, Marissa Seligman, PharmD3, Abhimanyu Roy, MBA4, Abhinav Nayyar, MBA, MBBS5, Ankitkumar Arora, MPharm6, Ram K. Mishra, PhD7, Rashmi Tulsi, Jr., Other8, Anuj Gupta, MSc9, Pooja Sharma, BA10, Vishan Khatavkar, MBA11, Khushboo -12;
1Optum Lifesciences, Boston, MA, USA, 2Optum, Bloomsbury, NJ, USA, 3Optum, Winchester, MA, USA, 4Optum, Gurgaon, India, 5Optum Life Sciences, Gurugram, India, 6Optum Global Solutions, Gurgaon, India, 7Optum Global Solution, Gurugram, India, 8Optum Global Solution, NOIDA, India, 9Optum Lifesciences, Noida, India, 10Optum Lifesciences, Delhi, India, 11Optum Lifesciences, Gurugram, India, 12Gurugram, India
1Optum Lifesciences, Boston, MA, USA, 2Optum, Bloomsbury, NJ, USA, 3Optum, Winchester, MA, USA, 4Optum, Gurgaon, India, 5Optum Life Sciences, Gurugram, India, 6Optum Global Solutions, Gurgaon, India, 7Optum Global Solution, Gurugram, India, 8Optum Global Solution, NOIDA, India, 9Optum Lifesciences, Noida, India, 10Optum Lifesciences, Delhi, India, 11Optum Lifesciences, Gurugram, India, 12Gurugram, India
OBJECTIVES: Metastatic colorectal cancer (mCRC) remains a major cause of cancer‑related mortality, and outcomes decline substantially for patients advancing to third‑ or fourth‑line (3L/4L) therapy. This study examined real‑world epidemiology, treatment patterns, and machine‑learning-derived predictors of progression to later‑line therapy using a large US claims‑linked clinical dataset.
METHODS: A retrospective observational cohort study was performed using Optum® Market Clarity data (January 1, 2019-June 30, 2025). Adult patients (≥18 years) with a confirmed diagnosis of mCRC were included. The index date was the first mCRC diagnosis, and eligibility required ≥2 post‑diagnosis visits or ≥12 months of continuous enrollment. Lines of therapy (LOTs) were derived using a rule‑based LOT algorithm. Patients initiating 3L or 4L therapy formed the analytic cohort. Outcomes included demographics, clinical characteristics, treatment regimens, and predictors of LOT progression using logistic regression and XGBoost models.
RESULTS: Among 30,904 patients with mCRC, 55.40% were female, 63.87% were non‑Hispanic White, and 33.08% resided in the US South region. Mean±SD age was 65.25±12.38 years. Approximately 13% had documented metastatic sites, with the liver/intrahepatic ducts most common (44.8%), followed by the retroperitoneum/peritoneum (27.5%) and other sites (14.7%). Progression through LOTs declined steeply: 42.45% advanced to second‑line therapy, 21.98% to 3L, and only 12.42% to 4L. Later‑line regimens demonstrated marked heterogeneity, including anti‑HER2 combinations, EGFR/BRAF‑directed baskets, VEGF‑based chemotherapy regimens, and TAS‑102 plus bevacizumab. Machine‑learning analyses identified several strong predictors of progression to later‑line therapy. Significant associations included metastasis site (especially liver involvement), indicators of disease progression or deterioration, older age, ECOG performance status, and additional clinical factors. Both logistic regression and XGBoost models consistently highlighted these predictors, with XGBoost showing superior variable discrimination.
CONCLUSIONS: In later LOTs, real-world treatment patterns in mCRC are highly heterogeneous, reflecting the use of diverse chemotherapy and targeted therapy options. Further research should assess effectiveness and tolerability.
METHODS: A retrospective observational cohort study was performed using Optum® Market Clarity data (January 1, 2019-June 30, 2025). Adult patients (≥18 years) with a confirmed diagnosis of mCRC were included. The index date was the first mCRC diagnosis, and eligibility required ≥2 post‑diagnosis visits or ≥12 months of continuous enrollment. Lines of therapy (LOTs) were derived using a rule‑based LOT algorithm. Patients initiating 3L or 4L therapy formed the analytic cohort. Outcomes included demographics, clinical characteristics, treatment regimens, and predictors of LOT progression using logistic regression and XGBoost models.
RESULTS: Among 30,904 patients with mCRC, 55.40% were female, 63.87% were non‑Hispanic White, and 33.08% resided in the US South region. Mean±SD age was 65.25±12.38 years. Approximately 13% had documented metastatic sites, with the liver/intrahepatic ducts most common (44.8%), followed by the retroperitoneum/peritoneum (27.5%) and other sites (14.7%). Progression through LOTs declined steeply: 42.45% advanced to second‑line therapy, 21.98% to 3L, and only 12.42% to 4L. Later‑line regimens demonstrated marked heterogeneity, including anti‑HER2 combinations, EGFR/BRAF‑directed baskets, VEGF‑based chemotherapy regimens, and TAS‑102 plus bevacizumab. Machine‑learning analyses identified several strong predictors of progression to later‑line therapy. Significant associations included metastasis site (especially liver involvement), indicators of disease progression or deterioration, older age, ECOG performance status, and additional clinical factors. Both logistic regression and XGBoost models consistently highlighted these predictors, with XGBoost showing superior variable discrimination.
CONCLUSIONS: In later LOTs, real-world treatment patterns in mCRC are highly heterogeneous, reflecting the use of diverse chemotherapy and targeted therapy options. Further research should assess effectiveness and tolerability.
Conference/Value in Health Info
2026-05, ISPOR 2026, Philadelphia, PA, USA
Value in Health, Volume 29, Issue S6
Code
MSR112
Topic
Methodological & Statistical Research
Disease
No Additional Disease & Conditions/Specialized Treatment Areas, SDC: Oncology