that used the most stringent inclusion criteria82,86 failed to support the GxE association.91 Some have also noted there is an inverse relationship between the power of the replication studies and support for the 5-HTTLPR association, precisely the opposite of what one would expect if the association is valid.91 Moreover, the most direct replication attempt of the Caspi findings, which was not included in any prior meta-analysis, found no evidence in support of the GxE effect on depression. This was a longitudinal birth cohort study following a similar population (New Zealand residents), for a similar length of time (30 years), and using comparable phenotypic measures.92 The authors observed no interaction between stressful life events and 5HTTLPR genotype, even after conducting 104 different regression models.92
