OUP user menu

Meta-analysis urges the development of new strategies to treat depression in order to improve cardiac prognosis: reply

Amanda Nicholson, Hannah Kuper, Mika Kivimaki, Harry Hemingway
DOI: http://dx.doi.org/10.1093/eurheartj/ehm128 1661-1662 First published online: 5 May 2007

de Jonge and Van Melle's model, suggesting that the somatic symptoms of depression may impair prognosis in CHD independently of severity of disease, offers a useful framework for future work.1 The first step will be to confirm that such an independent effect exists. Their findings need to be confirmed in other studies which control effectively for the severity of cardiac disease and examine different dimensions of depression. A meta-analysis using individual patient data would allow systematic adjustment for confounders and severity of disease. If an independent effect of depression is confirmed by such work, some doubts about residual reverse causality in the observed effect may persist. Results of studies based on depression reports will remain open to the possibility that the somatic items on depression symptom scales are acting as a better measure of physical health than the items being used to control for severity of disease.

Failure to demonstrate reversibility in the association between depression and impaired prognosis in heart disease remains a strong argument against a causal effect. There is some evidence of an effect of selective serotonin reuptake inhibitors (SSRIs) in improving prognosis in those suffering depression after an MI,2,3 although this requires confirmation in further studies. We wonder whether pharmacogenetic studies might add extra support to a causal link if an effect of SSRIs on prognosis of CHD is confirmed. Polymorphisms in genes influencing brain monoamines, for example, 5-HTTLPR and NET, may influence the impact of SSRI or NRI antidepressants on depressive symptoms.4,5 Assuming an independent effect of depression on CHD prognosis, any improvement in CHD prognosis due to treatment should be less in subjects with the genotype associated with a poorer response of depression to medication.

Considering the rapid expansion in genomic knowledge, future research may identify genetic variants suitable for Mendelian randomization studies, to mimic exposure to depression, in addition to gene-environment interaction studies related to treatment. The underlying assumption of Mendelian randomization studies is that, because alleles are randomly allocated from parents to offspring at conception, those individuals with genetic variants that relate to a higher risk of depression will have in effect been randomly allocated to somewhat higher levels of depression across their life course than individuals with genotypes related to lower risk of depression.6 These differences in depression should therefore translate into differences in CHD risk if depression were causally related to CHD. An obvious strength of this approach, compared with conventional observational data, is that the genetic variants affecting risk of depression represent proxies for depression that are largely free from reverse causation bias and confounding by behavioural factors. Although this approach is limited by the likely modest associations between genotype and depression, it may offer new avenues for enquiry.


View Abstract