A Multistate Framework for Analyzing Outcomes in Diffuse Large B-cell Lymphoma With Autologous Hematopoietic Stem Cell Transplant
Author(s)
Deepa Jahagirdar, PhD1, Jonas Haggstrom2, Grammati Sarri, MSc, PhD3, Marco Ghiani, PhD4, Anupama Vasudevan, MPH, PhD2.
1Cytel, Ottawa, ON, Canada, 2Cytel, Waltham, MA, USA, 3Cytel, London, United Kingdom, 4Cytel, Berlin, Germany.
1Cytel, Ottawa, ON, Canada, 2Cytel, Waltham, MA, USA, 3Cytel, London, United Kingdom, 4Cytel, Berlin, Germany.
OBJECTIVES: In Diffuse Large B-cell Lymphoma (DLBCL), hematopoietic cell transplantation (HCT) is a key therapy for patients with high-risk or stable disease. This study applies multistate modeling to explore the dynamic sequence and timing of clinical events beyond traditional survival analyses, generating insights that can inform cost-effectiveness evaluations and policy decisions in health technology assessment (HTA).
METHODS: A multistate model was applied to a cohort of primary refractory, chemo-sensitive DLBCL patients* who underwent HCT. The model included transitions between three states: HCT, progression/relapse, and death. Transition hazards were estimated using Cox proportional hazards models, and time-dependent transition probabilities were derived using the Aalen-Johansen estimator.*Dataset was collected by the Center for International Blood and Marrow Transplant Research (CIBMTR) which is supported primarily by the Public Health Service U24CA076518 from the National Cancer Institute; the National Heart, Lung, and Blood Institute; the National Institute of Allergy and Infectious Diseases;75R60222C00011 from the Health Resources and Services Administration;N00014-23-1-2057 and N00014-24-1-2507 from the Office of Naval Research; NMDP; and the Medical College of Wisconsin.
RESULTS: Among 168 patients*, 81 experienced progression and, of them, 65 died during follow-up. On average, patients stayed progression free for 50 months (95%CI:41.1, 60.7) post-HCT. Those who progressed spent on average 18 months (95%CI:14.0, 22.3) in that state before death or censorship. The four-year probability to remain progression-free was 0.38 (95%CI:0.26, 0.45), while probability of death among those who had progressed was 0.93 (95%CI:0.89, 0.97). Male patients and those older than 50 were at increased risk of both progression and death.
CONCLUSIONS: This analysis used multistate modeling to capture dynamic clinical trajectories, estimating risks of progression and death, as well as state occupancy over time. The results offer actionable insights for post-HCT surveillance and treatment planning, with important implications for health technology assessment, cost-effectiveness evaluations, and resource allocation.
METHODS: A multistate model was applied to a cohort of primary refractory, chemo-sensitive DLBCL patients* who underwent HCT. The model included transitions between three states: HCT, progression/relapse, and death. Transition hazards were estimated using Cox proportional hazards models, and time-dependent transition probabilities were derived using the Aalen-Johansen estimator.*Dataset was collected by the Center for International Blood and Marrow Transplant Research (CIBMTR) which is supported primarily by the Public Health Service U24CA076518 from the National Cancer Institute; the National Heart, Lung, and Blood Institute; the National Institute of Allergy and Infectious Diseases;75R60222C00011 from the Health Resources and Services Administration;N00014-23-1-2057 and N00014-24-1-2507 from the Office of Naval Research; NMDP; and the Medical College of Wisconsin.
RESULTS: Among 168 patients*, 81 experienced progression and, of them, 65 died during follow-up. On average, patients stayed progression free for 50 months (95%CI:41.1, 60.7) post-HCT. Those who progressed spent on average 18 months (95%CI:14.0, 22.3) in that state before death or censorship. The four-year probability to remain progression-free was 0.38 (95%CI:0.26, 0.45), while probability of death among those who had progressed was 0.93 (95%CI:0.89, 0.97). Male patients and those older than 50 were at increased risk of both progression and death.
CONCLUSIONS: This analysis used multistate modeling to capture dynamic clinical trajectories, estimating risks of progression and death, as well as state occupancy over time. The results offer actionable insights for post-HCT surveillance and treatment planning, with important implications for health technology assessment, cost-effectiveness evaluations, and resource allocation.
Conference/Value in Health Info
2025-11, ISPOR Europe 2025, Glasgow, Scotland
Value in Health, Volume 28, Issue S2
Code
MSR6
Topic
Clinical Outcomes, Methodological & Statistical Research, Real World Data & Information Systems
Disease
Oncology