Abstract
Autologous stem cell transplantation (ASCT) has been a mainstay in myeloma treatment for over three decades, but patient prognosis post-ASCT varies significantly. In a retrospective study of 5259 patients with multiple myeloma (MM) at the University of Arkansas for Medical Sciences undergoing ASCT with a median 57-month follow-up, we divided the dataset into training (70%) and validation (30%) subsets. Employing univariable and multivariable Cox analyses, we systematically assessed 29 clinical variables, identifying crucial adverse prognostic factors, such as extended duration between MM diagnosis and ASCT, elevated serum ferritin, and reduced transferrin levels. These factors could enhance existing prognostic models. Additionally, we pinpointed significant poor prognosis markers like high serum calcium and low platelet counts, though they are applicable to a smaller patient population. Utilizing seven easily accessible high-risk variables, we devised a four-stage system (ATM4S) with primary stage borders determined through K-adaptive partitioning. This staging system underwent validation in both the training dataset and an independent cohort of 514 ASCT-treated MM patients from the University of Iowa. We also explored cytogenetic risk factors within this staging system, emphasizing its potential clinical utility for refining prognostic assessments and guiding personalized treatment approaches.