Editorial: Moving From Innovation to Validation With Independent Sample Testing

Abstract

It is well established that family history is the most important known risk factor for the development of a mood disorder.1 Approximately 25% of at-risk offspring will develop a mood disorder1,2; yet, this is only slightly higher than the lifetime rate of a mood disorder overall of 21.4%.3 These statistics suggest heterogeneity within mood disorders generally and within people at familial risk. More information, particularly in the form of trait-level biomarkers, is sorely needed to accurately predict mood disorder risk and development. Trait-level biomarker identification may also help to decrease the well-known delay in accurately diagnosing bipolar disorders specifically,4 which could ultimately improve treatment and outcomes. Nimarko etal.5 aimed to identify trait-level biological markers of social reward-related neural vulnerability that distinguish youths genetically at risk for mood disorder. Nimarko and colleagues reported a well-designed study consisting of extremely well-matched youths who, by virtue of parentage, were at risk of bipolar disorder, were at risk of major depressive disorder, or were healthy (control group).5 The sample sizes reflected the challenges associated with doing this type of research with children, but the importance of the relationships was clear and strong. Specifically, the results showed reduced bilateral putamen activity (left: z= 4.39, p= .003; right: z= 4.56, p< .001) and reduced left putamen connectivity with right posterior cingulate cortex (z= 4.04, p< .01) and left anterior cingulate cortex (z > 3.64, p< .001) during an emotional face social reward task as potential trait markers that distinguished at-risk samples.5 In another study, similarly reduced reward-related connectivity in genetically at-risk youths for bipolar disorder relative to at-risk and healthy groups was shown to monetary reward and loss.6 Importantly, in both of these studies, reward-related connectivity group differences were observed before threshold symptoms developed, suggesting that early identification and intervention may be possible. In addition, this reduced reward-related connectivity was not related to subthreshold symptoms in either study. Taken together, these findings suggest that social reward and monetary reward and loss may be interconnected at a neural level in youths genetically at risk for bipolar disorder before symptom development. Exploratory analyses in Nimarko etal.5 also showed that reduced reward-related connectivity may be a trait-level biomarker of risk of future conversion to an Axis I disorder over time (bipolar disorder risk: hazard ratio= 8.28, p< .01); major depressive disorder risk: hazard ratio= 2.31, p= .02). Although Acuff etal.6 did not address the longitudinal question of conversion directly, the authors speculated that reduced reward-related connectivity was related to increased risk of future bipolar disorder. This serendipitous independent replication of reward-related connectivity is a promising lead, but repeated testing in independent samples is needed. However, direct independent sample testing is uncommon in psychiatry and in psychiatric neuroscience.7,8 This is partially due to factors such as time and money, but also to the value placed on innovation, which is prized by funding agencies, journal editors, and society. Independent sample testing is not, by definition, innovative, and while innovation is an important cornerstone of science and should be encouraged, we also need to test previously identified findings using independent samples.