(163) Colorectal Cancer Cohort Comparison Among Four Real-World Databases

Friday, August 26, 2022

8:00 AM – 6:00 PM CEST

Location: Hall C

Publication Number: 494

Presenting Author(s)

Wei Zhou, MD, PhD, FISPE
– Eli Lilly and Company, West Point, Pennsylvania, United States

Co-Author(s)

Ching-Yu Wang, PhD
– Research associate, Evidera, Gainesville, Florida, United States
Changxia Shao
– Merck and Co., Inc., Kenilworth, New Jersey, USA, United States
Alicia McDonald
– Merck & Co., Inc., Philadelphia, Pennsylvania, United States
Edward A. Bortnichak, Ph.D., M.P.H., MBE
– Executive Director, Epidemiology, Merck and Co., Inc., Kenilworth, New Jersey, USA, West Point, Pennsylvania, United States
Xinyue Liu
– Epidemiologist, Merck and Co., Inc., Kenilworth, New Jersey, USA, Lansdale, PA, United States

Background: There are different types of real-world databases including claims, electronic medical records, clinico-genomic databases for cancer patients. It’s important to understand the characteristics of different databases.

Objectives: To compare demographics, clinical characteristics, treatment patterns, and overall survival using colorectal cancer (CRC) population as an example.

Methods: This study utilized GENIE-BPC v1.1-consortium (American Association for Cancer Research Project Genomics Evidence Neoplasia Information Exchange Biopharma Collaborative, AACR GENIE BPC v1.1-consortium), TCGA (The Cancer Genome Atlas), MarketScan (IBM MarketScan Commercial and Medicare Supplemental) and SEER-Medicare (Surveillance, Epidemiology and End Results (SEER)-Medicare) databases. Newly diagnosed CRC patients aged ≥ 18 (≥ 65 for SEER-Medicare) were compared. Newly diagnosed stage IV CRC patients were further compared in treatment patterns. CRC patients were selected based on ICD-9/10-CM codes in MarketScan databases. Treatments were measured based on HCPCS (Healthcare Common Procedure Coding System) and NDC (National Drug Code) in MarketScan and SEER-Medicare databases. Lines of therapy were defined based on the same prespecified algorithms for all databases except TCGA. Regimen was identified within 30 days of initiation of each line of therapy.

Results: A total of 65,976 CRC patients (1,336 from GENIE-BPC; 552 from TCGA; 32,169 from MarketScan, 31,919 from SEER-Medicare) and a total of 13,985 stage IV CRC patients (647 from GENIE-BPC; 90 from TCGA; 8,114 from SEER-Medicare; 5,134 from MarketScan) were identified. SEER-Medicare had the oldest patient population, mean (SD) age of 77.7 (7.9) years old for CRC and 77.3 (7.8) years old for stage IV CRC. GENIE had the youngest patient population, mean (SD) age of 54.1 (12.6) for CRC and 52.7 (12.1) for stage IV CRC years old. The majority of patients were male (except for SEER-Medicare) and of white race across databases, and about half of the patients in TCGA had no race information. GENIE-BPC had the highest proportion of stage IV CRC patients (48.4% vs. other databases 13.8-25.4%) and patients receiving treatments (95.7% vs. 37.6-59.1%). FOLFOX (+ bevacizumab) is the most common regimen across databases accounting for 47.3-78.5% of patients receiving 1st line of therapy. Median survival was 94, 36, 44 months (CRC) and 23, 23, 15 months (stage IV CRC) for TCGA, GENIE-BPC after left truncation, and SEER-Medicare databases.

Conclusions: Compared to other databases, GENIE-BPC had the youngest CRC patients with the most advanced disease and the largest proportion of patients receiving treatment. Treatment patterns were similar across databases.