Conditional survival of children, adolescents and young adults (0–24 years) diagnosed with leukaemia during 2000–2014 world-wide: (CONCORD-3)

Abstract

Background: Population-based survival estimates provide valuable insights into cancer care patterns world-wide. Access to optimal treatment leads to better outcomes, however, treatment pathways vary globally. Conditional survival is the probability that patients who have already survived for a given number of years since diagnosis will live for an additional number of years. It is a useful proxy to assess the success of initial treatment or remission of leukaemia. Methods: We analysed data for 164,563 patients aged 0–24 years diagnosed during 2000–2014, from 258 population-based cancer registries in 61 countries. Using the Pohar-Perme estimator, we estimated net survival at five years, conditional on surviving at least one year, and at 10 years conditional on surviving five years. To control for background mortality, we used life tables of all-cause mortality by single year of age, sex, country and calendar year. All-ages survival estimates were standardised to the marginal age distribution. Findings: During 2010–2014, age-standardised five-year conditional net survival ranged from 61.8 % in Mexico to 90 % or more in 20 countries. By 2010–2014, five-year conditional survival in most high-income countries exceeded 90 % for children, but not for older patients, and for acute myeloid leukaemia it was typically 5–10 % lower than for lymphoid leukaemia. Ten-year conditional survival was 90 % or higher in most countries, with less variation world-wide. Interpretation: World-wide variation in survival was less marked for patients who survived the first year(s) after diagnosis. Notable gains occurred in countries with initially lower five-year survival (e.g., China or Mexico), where legislative changes contributed to improved access to treatment for young patients with cancer. None- theless, inequalities persisted between high-income and low- and middle-income countries. Population-based cancer registry data remain essential to monitor further improvements.