Pilot Study to Improve In Vitro Identification of Culprit Drugs in SJS/TEN Using Granzyme B and Granulysin Secretion Assays

Abstract

Introduction: Stevens–Johnson syndrome (SJS)/toxic epidermal necrolysis (TEN) are severe cutaneous adverse drug reactions (SCARs) with high morbidity and mortality. Accurate identification of the culprit drug is critical and remains challenging, whereas current in vitro tests such as lymphocyte transformation test (LTT) have limited sensitivity. To evaluate measuring granulysin (Gr) and granzyme B (GZB) secretion from drug-stimulated peripheral blood mononuclear cells (PBMCs) for identification of culprit drugs and to compare their diagnostic performance with conventional LTT. Materials and Methods: PBMCs from 12 patients with confirmed SJS/TEN were incubated with suspected drugs. LTT and Gr/GZB detection assays were performed on day 3 and 6. Levels of Gr and GZB were quantified from supernatants, and results were normalized as production ratios. Cut-off values were defined using receiver operating characteristic (ROC) curve analysis with high specificity thresholds. Results: GZB detection on day 3 yielded the highest sensitivity (8/12) compared to LTT (6/12) and Gr detection. Combining GZB and Gr detection improved the positive rate to 10/12. ROC curve analysis demonstrated higher AUCs for cytokine assay than for LTT, particularly for GZB at day 3. Combining GZB and Gr detection with LTT did not significantly increase diagnostic yield. Conclusion: The detection of GZB and Gr production in PBMC cultures provided a more sensitive and specific method than the conventional LTT for identifying culprit drugs in SJS/TEN. In particular, GZB detection at day 3 provided a shorter assay duration, avoided radioactive labeling, and demonstrated superior diagnostic performance. This cytokine-based approach may represent a practical and safer alternative to enhance drug causality assessment in patients with SCARs.