Survivors of first adult-onset cancers are at risk of developing second primary cancers (SPCs) and are at risk of death from their subsequent cancers and other competing causes. Here we investigated patterns of incident SPC risk and cause-specific mortality in survivors of adult-onset cancer during the past three decades.
Data were extracted from the population-based Tasmanian Cancer Registry in Australia. Patients diagnosed with a first primary cancer between 1980 and 2009 were followed for incident SPCs to December 31, 2013 and for deaths to December 31, 2014. SPC risks were quantified by using standardised incidence ratios (SIRs). Trends in SPC risk over time were assessed in multivariable Poisson models. The cumulative incidence and subdistribution hazard ratios (SHR) of cause-specific deaths were estimated using competing risk models.
5,339 SPCs were observed from 51,802 cancer survivors. The SIRs for any SPC increased from 0.98 with a first cancer diagnosis in 1980-1984 to 1.12 in 2005-2009. The increase in SIRs was significant in multivariable Poisson models (Ptrend< 0.001). Deaths were identified in 39,976 (69.8%) of 57,288 patients. The 5-year cumulative incidence of death due to first primary cancer gradually decreased from 57.2% for a first cancer diagnosis in 1980-1984 to 30.7% in 2005-2009. However, the 5-year cumulative incidence of deaths due to subsequent cancers varied across periods of first cancer diagnosis, with an increase from 1.0% in 1980-1984 to 1.7% in 1995-1999, and a decrease to 1.4% in 2005-2009. The SHR of deaths due to first primary gradually decreased over time in multivariable competing risk models, but varied over time for deaths due to subsequent cancers: the SHR increased from 1.00 (reference) in 1980-1984 to 1.19 (95%CI 1.03-1.36) in 1995-1999, then decreased to 0.80 (95%CI 0.69-0.94) in 2005-2009.
The risk of SPC has increased in Tasmania over the last three decades. While the risk of death due to first primary cancer decreased over time, the risk of death due to subsequent cancers did not. The increased risk of deaths from subsequent cancers might be an outcome of overdiagnosis of first primary cancer in the 1990s.
Clinical trial identification
Legal entity responsible for the study
Menzies Institute for Medical Research, University of Tasmania
All authors have declared no conflicts of interest.