Looking Beyond Diagnostic Labels: How Depressed Mood Obscured Our Interpretation of Autism Neuroscience Data
Objectives: To provide an example in which latent heterogeneity (here, on depressed mood) within ASD samples masks subgroup effects that are stronger than the reported summary group effect of “ASD” emotion processing.
Methods: Our current sample of n=130 will increase significantly by May 2019. This abstract reflects findings based on a preliminary subset of n=53 adults aged 18-35 with verbal IQ>80, including adults with ASD (n=21), typically developing adults with current depression (TD-depressed, n=13), and non-ASD, non-depressed controls (TD-controls, n=19). Participants completed diagnostic assessments (including the Autism Diagnostic Observation Scale, second edition, and the Structured Clinical Interview for DSM-5 Disorders), self-report questionnaires (e.g., Beck Depression Inventory), and a passive-viewing task employing emotionally-expressive faces, during which pupil motility was used to index cognitive-emotional load in response to single faces presented for 400 milliseconds, then masked for 8 seconds.
Results: In response to emotional faces, the pupil course of ASD participants, compared to TD-depressed and TD-controls, seemed to exhibit the general pattern noted in previous literature: slower initial but overall heightened response to emotional material (see red hatched line in Figure 1, compared to blue and green lines). However, participants with ASD who ranked in the highest third on depressive symptoms (blue solid line in Figure 2A) had faster, larger, and sustained pupil response to sad faces compared to their lesser-depressive peers with ASD (red and green lines on Figure 2A). This pattern of neural responsivity markedly resembled the TD-depressed pupil course to sad faces (Figure 2B), with no significant practical or statistical difference between the groups. Higher initial and sustained response to dysphoric stimuli apparently represents a similarity in emotion processing across ASD and TD populations that is specific to dysphoric mood.
Conclusions: In these data, the overall slow-but-sustained pattern generally associated with ASD masked depression subgroups that replicated our comparison groups (TD with and without current depression). Specifically, a unique pattern of fast-and-sustained cognitive-affective response was associated with greater depressive symptoms across TD and ASD participants. More broadly, conclusions about emotion processing in the ASD population could be subject to latent effects that are similar to our observed effects here. These findings underscore the necessity of stratifying samples on cognitive and affective variables (e.g., depression, anxiety, rumination) regularly when interpreting neuroscientific data, given the potential for latent heterogeneity within autism. We will discuss analogous findings from independent sources.