Abstract
Background: Over the last decades, it has been repeatedly suggested that psychosocial factors such as depression and anxiety increase the risk of breast cancer, through mechanisms such as mutation, DNA repair, neuroendocrine processes, immunological processes, or unhealthy behaviours. With individual participant data meta-analyses, we aimed to test whether depression, anxiety, recent loss event, and perceived social support increase the risk for breast cancer. We also explored the effects of neuroticism, general distress, and relationship status.
Materials and methods: IPD meta-analyses were performed with up to twenty-two studies in the PSY-CA consortium (up to: N = 220,258, person years = 2,502,822, breast cancer incidences = 5724). At stage 1, Cox regression models were fitted in each cohort for each psychosocial factor (outlined above) and breast cancer outcome. Two models were tested: a minimally-adjusted model (correcting for sociodemographic covariates) and a maximally-adjusted model (additionally correcting for several health behaviors and other potential confounders such as parity). At stage 2, hazard ratios (from stage 1) were pooled using random-effects meta-analyses.
Results: Most psychosocial factors were not related to breast cancer incidence, with the exception of anxiety symptoms which showed a protective effect (HR = 0.95 [0.91, 0.998], p = 0.04) in the minimally adjusted model. When adjusting for further potential confounders, this effect was no longer statistically significant (HR = 0.96 [0.90, 1.02], p = 0.14).
Conclusions: We found no consistent evidence for an association between psychosocial factors and the incidence of breast cancer, with the possible exception of anxiety symptoms showing a small, protective effect. Further research is needed to test whether health-related behaviours, such as unhealthy behaviours or menopausal status, moderate the association between psychosocial factors and breast cancer.
Materials and methods: IPD meta-analyses were performed with up to twenty-two studies in the PSY-CA consortium (up to: N = 220,258, person years = 2,502,822, breast cancer incidences = 5724). At stage 1, Cox regression models were fitted in each cohort for each psychosocial factor (outlined above) and breast cancer outcome. Two models were tested: a minimally-adjusted model (correcting for sociodemographic covariates) and a maximally-adjusted model (additionally correcting for several health behaviors and other potential confounders such as parity). At stage 2, hazard ratios (from stage 1) were pooled using random-effects meta-analyses.
Results: Most psychosocial factors were not related to breast cancer incidence, with the exception of anxiety symptoms which showed a protective effect (HR = 0.95 [0.91, 0.998], p = 0.04) in the minimally adjusted model. When adjusting for further potential confounders, this effect was no longer statistically significant (HR = 0.96 [0.90, 1.02], p = 0.14).
Conclusions: We found no consistent evidence for an association between psychosocial factors and the incidence of breast cancer, with the possible exception of anxiety symptoms showing a small, protective effect. Further research is needed to test whether health-related behaviours, such as unhealthy behaviours or menopausal status, moderate the association between psychosocial factors and breast cancer.
Original language | English |
---|---|
Pages (from-to) | S6-S7 |
Number of pages | 2 |
Journal | European Journal of Cancer |
Volume | 175 |
Issue number | Supplement 1 |
Early online date | 17 Nov 2022 |
DOIs | |
Publication status | Published - Nov 2022 |