Citation: Caspi, A., et al., Influence of life stress on depression: moderation by a polymorphism in the 5-HTT gene. Science, 2003. 301 (5631): p. 386-9.
Design: Birth cohort followed from age 3 to 26. Setting: New Zealand Patients: 1037 children (91% of births between April 1972 to March 1973) . 96% intact at age 26. 7% were excluded as they were Maori. N=847 . Depression before 21 excluded. Follow up: At age 3,5,7,9,11,13,15,18,21 and 26.
Risk factor and exposure: Serotonin transporter gene variation – 5HTTLPR , stressful life events between 21-25 –Life history calendar ( prior 5 years) and childhood maltreatment between ages 3 and 11 through observation, parental reports and retrospective reports by participants.
Outcome: Depression at age 26 (within the last year) with Diagnostic interview schedule.
Analysis: Model for depression as an outcome included sex as a covariate and genotype, stress and stress*genotype
- s/s homzygotes= 17%; s/l heterozygotes= 51%; l/l homozyotes= 31%
- Stressful life events (SLE) occurring after age 21 predicted the occurrence of depression at age 26 in those with the s/s (b=0.79, p= 0.002)and among s/l( b=0.41, p=0.001) but not among l/l carriers.
- SLE predicted suicidal ideation among s allele carriers but not in l/l homozygotes (p=0.05)
- The above results continued to be significant after taking informant reports into account (p<0.01)
- To prove that this was not a pure genetic effect, G*E (21-26) association with depression at age 26 positive G*E (21-26) association with depression at age 18 and 21 negative
- Childhood maltreatment predicted adult depression in s allele carriers only
- 5 HTT gene association with depression and stress was not a function of the MAO-A gene.
Validity criteria (Users guide and JAMA evidence)
- Was the sample of patients representative? Large community sample from New Zealand. Not entirely representative of patients seen in our practice as they are ethnicaly different.
- Were the patients sufficiently homogenous with respect to prognosis? Yes.
- Was the phenotype properly defined and recorded by someone blind to the genotype? Yes
- Do the phenotype proportions observe Hardy-Weinberg equilibrium? Yes
- Was follow up sufficiently complete? Yes
- Are the results consistent with other studies? Partially. Recent metaanalysis does not support this finding.
- How large and precise are the associations? Associations met statisitcal significance.
- How do I apply this in my clinical setting? At this time, there is no direct clinical application.
- Will the genetic associations improve predictive power beyond easily measured clinical variable? No.
- Will the results change my management? No
- Will my patient be better off knowing the result of the test? No