Validation of the advanced‐stage Hodgkin lymphoma international prognostic index (A‐HIPI) in the GHSG HD21 trial

The International Prognostic Score (IPS) has served as the standard risk stratification tool in advanced-stage classical Hodgkin Lymphoma (cHL) for over two decades. 1 The IPS-7 incorporates seven dichotomized clinical factors-age, sex, serum albumin, hemoglobin, Ann Arbor stage, leukocytosis, and lymphocytopenia-to assign patients into risk categories for disease progression at 5 years.The score's ability for prognostication has diminished considerably as modern regimens have led to substantially improved outcomes.2,3 The advanced-stage Hodgkin lymphoma international prognostic index (A-HIPI) was developed with contemporary statistical methods as a prognostic model using individual patient data pooled from eight clinical trials and was validated in prominent registries.4 A-HIPI models employ, with slight variation, age, lymphocyte count, hemoglobin, and albumin levels as continuous variables besides sex, Ann Arbor stage, and bulky disease to predict 5-year (5y) probabilities of progression-free survival (PFS) and overall survival (OS).A-HIPI demonstrated improved calibration for PFS and OS and enhanced discrimination for OS compared to IPS-7.The original model was developed and validated primarily with ABVD-treated patients, while additional validations have been published.[5][6][7] There is a need to validate prediction models for patients treated with novel agents such as brentuximab vedotin (Bv) 8 or nivolumab, 9,10 which are increasingly used in the first-line treatment.11 The randomized, Phase III GHSG HD21 trial (NCT02661503) compared BrECADD (brentuximab vedotin, etoposide, cyclophosphamide, doxorubicin, dacarbazine, and dexamethasone) with eBEACOPP (escalated bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, and prednisone) in advancedstage cHL.12 The trial demonstrated superior efficacy of BrECADD compared to eBEACOPP.The aim of this analysis was to assess the performance of the A-HIPI in the HD21 trial population and to compare it with the IPS-7.This analysis included 1446 patients from HD21 (BrECADD: 729; eBEACOPP: 717).The cohort was restricted to intention-to-treat (ITT) patients with available A-HIPI variables and laboratory values within the predefined ranges (S1).The full model equations of the A-HIPI have been reported elsewhere.4 To assess the model's performance, discrimination and HemaSphere.