Estimating Long-Term Survival in a Cohort of Allogeneic Hematopoietic Stem Cell Transplant Patients

Introduction: Allogeneic hematopoietic cell transplantation (HCT) is a common treatment for many hematologic diseases. Most deaths occur in the first 2 years after HCT due to relapse, graft-versus-host disease, infections, malignancies, or other toxicities. Among patients who are alive and recurrence free at 2 years after HCT, survival at 10 years is between 80% and 92%.

Advances in transplantation practices have led to improved outcomes and more long-term HCT survivors. As survival outcomes continue to improve and new treatments emerge, understanding and quantifying the full lifetime benefit of HCT in terms of mean overall survival (OS) is clinically relevant. In this analysis, we estimate the mean OS of a cohort of HCT patients.

Methods: A systematic literature review of all studies reporting OS post-HCT was conducted. Extracted data were incorporated into a long-term survival model using a step-wise approach: short-term survival (up to 2 years post-HCT) using data reported by Uhlin et al (Haematologica. 2014;99:346-52), and longer-term survival (more than 2 years post-HCT) using data reported by Wingard et al (J Clin Oncol. 2011;29:2230-9) and the age-adjusted life tables for the general UK population.

Available published data provided OS estimates up to 15 years post-HCT, and beyond this time, OS estimates are uncertain. To estimate mean OS and address this uncertainty, three different survival scenarios were modeled: best-case, worst-case, and base-case. For the best-case, it was assumed that HCT patients were cured and had the same OS as the age-adjusted general population. For the worst-case, it was assumed that HCT patients carried excess mortality for the rest of their lives. The excess mortality was calculated from Wingard et al 2011, which showed 20% mortality from year 2-15 post-HCT; this is an order of magnitude greater than the general population (2%). For the base-case, a Weibull parametric function was fit to the data published by Wingard et al 2011 to estimate the survival curve from year 2 post-HCT until death. To incorporate the excess mortality, the lowest survival was chosen cycle over cycle between the parametric estimate and the age-adjusted life tables.

Sensitivity analysis for the mean survival estimate was performed by fitting several different parametric functions using exponential, Gompertz, log-logistic, and log-normal. A lifetime analysis was undertaken for a cohort of patients starting at the weighted mean age at the time of HCT, calculated from Wingard et al 2011.

Results: Only two published articles were found that provided OS in patients without complications after HCT (Uhlin et al 2014 and Wingard et al 2011). Data from Uhlin et al 2014 was used to estimate the percent alive at 2 years post-HCT, which was 65%.

For a cohort of HCT patients that received their transplant at age 23.5, the estimated mean OS for the base-case was 25.9 years post-HCT, with parametric models ranging between 25.9 and 29.5 years. The best-case and worst-case estimates were 31.7 and 23.9 years, respectively.

Conclusions: The mean OS for a cohort of HCT patients was estimated to be 25.9 years. This estimate helps to understand and quantify the full lifetime survival benefit to HCT patients, including the tail end of the survival curve, and the potential added benefits of future treatments post-HCT. The sensitivity analysis revealed a narrow range for the estimated mean OS, minimizing the uncertainty of the results. Our estimates are based on data published by Wingard et al 2011, which incorporates excess mortality after 2 years post-HCT. This is consistent with the clinical assumption that HCT patients continue to have excess mortality throughout their lifetime. Since the first 2 years post-HCT has the highest mortality rate, new treatments that can improve survival during this time may change the impact on the lifetime benefit of the therapy.