Impact of Overdiagnosis on Long-Term Survival of Breast Cancer

Background: While overdiagnosis becomes the main concern in breast cancer screening with mammography its influence on long-term survival of breast cancer is poorly understood and intractably quantified because of indistinguishability between overdiagnosis and curation. We aimed to assess respective independent contributions of overdiagnosis and curation to long-term survival of breast cancer. Methods: We elucidated this thorny issue by using a Swedish Dalarna cohort with long-term follow-up of 1346 invasive breast cancers from 1996 onwards together with a zero (cured or overdiagnosis)-inflated model design and analysis. The zero part represents both types of non-progressive cancer without potential of dying from BC, the cured due to effective treatment and the over-diagnosed due to mammography screening. These two types can be distinguished by the provision of information on detection modes (screen-detected cases and interval cancer plus cancers from non-participants). The count part represents the progressive breast cancer with potential of dying from BC depending on prognostic factors during follow-up. Findings: The probability for non-progressive BC (the zero part) was 56·14% including 44·34% due to the curation after early detection and initial treatment and 11·80% due to overdiagnosis resulting from mammography screening program (8·94%) and high awareness (2·86%). Among 43·86% progressive BC (the count part), 32·11% undergoing subsequent adjuvant therapies still remained alive after 15-years of follow-up when adjusting for significant prognostic factors. The 15-year prognosis-adjusted cumulative survival of BC was dropped from 88·25% to 74·80% after correcting for the zero-inflated part of overdiagnosis. Interpretation. The proposed zero-inflated model design and analysis together with information on detection mode with 15 years of follow-up revealed 76% survivors of BC due to the curation resulting from mammography screening and accompanying effective treatment and therapy and 12% due to overdiagnosis. Funding: HHC, AMFY, SLSC, and CYH are supported by the Ministry of Science and Technology grant (grant number MOST 106-2118-M-002-006-MY2; MOST 106-2118-M-038-002-MY2; MOST 106-2811-M-002-075; MOST 107-3017-F-002-003). HHC is supported by The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan (NTU-107L9003). Declaration of Interest: The authors declare no conflict of interest. Ethical Approval: This study was approved by the Joint Institutional Review Board of Taipei Medical University (TMU-JIRB, approval numbers N201607008).