Validation of an online application to identify potential immune-related adverse events associated with immune checkpoint inhibitors based on the patient’s symptoms

by Takahiro Osawa, Takashige Abe, Hiroshi Kikuchi, Ryuji Matsumoto, Sachiyo Murai, Takafumi Nakao, Shinji Tanaka, Ayu Watanabe, Nobuo Shinohara

Immune checkpoint inhibitors (ICIs) are increasingly being used to treat malignancies. Some patients experience immune-related adverse events (irAEs), which may affect any organ/tissue. IrAEs are occasionally fatal and usually have nonspecific symptoms. We developed a three-step application (https://irae-search.com/) to provide healthcare professionals with information on the diagnosis, treatment options, and published reports for 38 categories of irAEs encountered in clinical practice.

IrAEs reported in ≥5 cases were identified from articles published between October 2018 and August 2020 by searching Japanese (SELIMIC, JAPIC-Q Service, and JMED Plus) and international (MEDLINE, EMBASE, Derwent Drug File) databases. The cases’ symptoms were entered into the application to identify irAEs, which were verified using the reported diagnosis, to evaluate the application’s sensitivity and specificity.

Overall, 1209 cases (1067 reports) were analyzed. The three most common categories of irAEs were pituitary or adrenal disorders (14% of cases), skin disorders (13%), and diabetes mellitus (10%). The top three primary diseases were lung cancer (364 cases), melanoma (286 cases), and renal cell carcinoma (218 cases). The average sensitivity was 90.8% (range 44.4%–100.0%) initially, and improved to 94.8% (range 83.3%–100.0%) after incorporating the symptoms reported in published cases into the application’s logic for two irAE categories. The average specificity was 79.3% (range 59.1% [thyroid disorders]–98.2% [arthritis]).

irAE Search is an easy-to-use application designed to help healthcare professionals identify potential irAEs in ICI-treated patients in a timely manner to facilitate prompt management/treatment. The application showed high sensitivity and moderate-to-high specificity for detecting irAEs.