Survival and causes of death in primary mitochondrial diseases

Background: Primary mitochondrial diseases affect 1 in 4300 livebirths and cause early death or lifelong disability but our understanding of their impact on survival and causes of death is limited. There is conflicting evidence about whether they cause an increased incidence of cancer. We aimed to determine survival, causes of death and risk of developing cancer in people with mitochondrial disease by linking genetic results with routinely collected healthcare data at a national scale. Methods: Individuals with mitochondrial diseases were identified in England (population 60.2 million) by National Health Service diagnostic laboratories and Genomics England. A curated cohort was assembled. Individual consent was not required because the National Disease Registration Service has legal permission to collect patient data. We used national datasets to find out the vital status, cause of death and any cancer diagnoses for each individual. Kaplan-Meier survival, Cox regression model and standardised cancer incidence ratio were analysed. Results: 1266 unique individuals were identified, with 1164 having a known region in England. Following clinical review, 1107 were included (527 males, 580 females, median age 48 years, age range from infancy to 90 years). 74% had a mitochondrial DNA diagnosis and 26% had nuclear genetic diagnosis. 170 were deceased. 159 were diagnosed during childhood (14.3%). Most childhood deaths were due to the underlying mitochondrial disorder. Common causes of death in adults included cardiac, cancers, and pneumonia. Survival after diagnosis varied by genetic diagnosis and by age at diagnosis with poorer survival for people diagnosed during childhood (hazard ratio 2.11, p < 0.001) or older age (2.83, p < 0.001) and those with recessive POLG disease (hazard ratio 3.06, p < 0.001). For the whole cohort, survival at 10 years after genetic diagnosis was 80.3% (95% confidence interval 76.9% to 83.7%) and survival at 20 years after diagnosis was 58.9% (95% confidence interval 49.6% to 70.0%). Standardised cancer incidence ratios did not show a difference in cancer incidence compared to the general population. Conclusions: Survival was much longer than anticipated, with differences due to the precise genetic diagnosis and age at onset. Despite prior contradictory evidence, there was no increased cancer risk in this large cohort.