Real-world usage and adverse events (AE) of immune checkpoint inhibitors (ICI): A large-scale, automated, GDPR-compliant analysis of hospital records

BackgroundICI form a backbone of curative and non-curative treatment across cancer types, yielding survival benefits as well as costs and AEs. It is challenging to derive these combined aspects from trial data and classical real-world data (claims, registries), because of the heterogeneity of patient (pt) populations and recorded outcomes. The Observational Medical Outcomes Partnership Common Data Model (OMOP CDM) is designed for analysis of cross-domain observational health data.MethodsIn 3 Belgian hospitals, we created OMOP data warehouses (DWH) of pts treated with 9 reimbursed ICIs, using both structured and unstructured data sources (natural language processing). Features include demographics, comorbidities, vitals, lab, diagnoses, drug use, care consumption and AEs of special interest (conditions/symptoms that can be ICI-related). We characterized the cohort with a focus on treatment patterns and potential ICI-related AEs.ResultsA total of 1574 pts received 18 584 ICI-administrations for 20 cancer types (March 2017 - August 2022). Lung cancer accounted for 44% of ICI administrations. Median real-world time on treatment ranged from 84 (95% CI: 70 - 124) days for head and neck cancer to 971 (95% CI: 70 - not reached) days for colorectal cancer. In 48% of pts, an AE of special interest occurred in the DWH before start of ICI and was therefore considered a co-morbidity (Table). In 42% of pts, such an AE first occurred during ICI-treatment (from start until 91 days after last dose; pt groups are not mutually exclusive). In-hospital administration of non-corticosteroid immunosuppressants during ICI-treatment was detected in 3%. Table: 1056P Patients, n (%) Total 1574 Female 537 (34) Median age 67y Performance status 0 255 (26) 1 484 (50) 2-4 229 (24) Unknown 606 Primary cancer, n (%) Lung NSCLC 607 (39) SCLC 72 (5) Unspecified 51 (3) Melanoma 229 (15) Head and neck 139 (9) Urothelial 137 (9) Renal cell 133 (9) Mesothelioma 59 (4) Other 147 (9) Detection of AE, n (%) Detection of non-CS immunosuppressant, n Before ICI On ICI Total 760 (48)23 659 (42)34 Acute interstitial nephritis 13 (1)2 22 (1)0 Adrenal insufficiency 17 (1)1 39 (3)2 Arthritis 87 (6)10 35 (2)2 Colitis 82 (5)4 106 (7)13 Dermatitis 87 (6)8 95 (6)3 Diabetes mellitus 337 (21)10 62 (4)3 Hepatitis 43 (3)2 61 (4)5 Hypophysitis 43 (3)2 21 (1)5 Other (predefined) 415 (26)14 490 (31)19 ConclusionsWe built granular DWH across hospitals on >1500 ICI pts, with a large potential for real-world studies. Lung carcinoma was the most frequent indication. Among AEs that can be ICI-related, diabetes mellitus was the main AE detected before start (21% of pts), whereas AEs detected on ICI-treatment varied.Legal entity responsible for the studyLynxcare.FundingLynxcare.DisclosureA.T.L. Verbiest: Financial Interests, Institutional, Invited Speaker: Novartis, Ipsen; Financial Interests, Institutional, Advisory Board: Ipsen; Non-Financial Interests, Principal Investigator, PI of an investigator-initiated multicenter study on real-world outcomes in immuno-oncology, by buiding research-grade clinical data warehouses in collaboration with Lynxcare. P.R. Debruyne: Financial Interests, Institutional, Research Grant: Pfizer; Financial Interests, Personal, Advisory Board: Astellas, BMS, Ipsen, Merck, Pfizer. D.F.E. Hens: Financial Interests, Personal, Ownership Interest: Lynxcare. H. Prenen: Financial Interests, Institutional, Advisory Board: Amgen, Roche, AstraZeneca; Financial Interests, Institutional, Invited Speaker: Bayer, Ipsen, Sanofi. C. Vulsteke: Financial Interests, Personal, Advisory Board: MSD, Janssens-Cilag, GSK, Astellas Pharma, BMS, Leo Pharma, Bayer, AstraZeneca, Pfizer, Merck; Financial Interests, Institutional, Research Grant, Funding for research project on immune related toxicities: MSD. All other authors have declared no conflicts of interest.