Testing the Utility of Positive, Negative, and Cognitive Dimensions for Parsing ASD Heterogeneity

Background: Autism spectrum disorder (ASD) is very heterogeneous in its clinical presentation. While gold-standard assessment tools are effective for capturing the general presentation of the disorder and making diagnostic discriminations, there are currently no assessments that parse potentially diverse presentations within symptom domains; instead, existing assessments focus on the general notion of “abnormality.” For example, many assessments probe for “atypical prosody,” yet, due to broad wording, questions cannot differentiate whether prosody is over-animated and cartoonish versus robotic and monotone. This lack of specificity limits the capacity to characterize phenotypic heterogeneity, which may reflect dissociable underlying biology or relate to different treatment outcomes. In schizophrenia, conceptualizing symptoms along positive, negative, and cognitive dimensions has been fruitful in treatment development (Chen et al. 2013); it has recently been proposed that applying a similar conceptualization to the ASD phenotype could be a meaningful endeavor (Foss-Feig et al. 2015).

Objectives: To test the utility of capturing ASD symptoms along positive (presence of an atypical feature not seen in normative development), negative (decrease or absence of a behavior characteristic of normative development) and cognitive (abnormality in thought processes) dimensions as a way to assess variability within facets that intersect traditional symptom domains and differentially associate with low-level processes.

Methods: A battery of social, language, sensory, and diagnostic assessments was administered to 28 children (age 10-13 years) with ASD. Individual items were selected from the Social Responsiveness Scale (SRS) and Clinical Evaluation of Language Fundamentals Pragmatics Profile (CELF-PP) that best mapped onto the Positive, Negative, and Cognitive dimensions in the social domain (5, 6, 6 items, respectively), as outlined in Foss-Feig et al (2015). Bivariate correlations evaluated whether variability along these new dimensions was dissociable from traditional scales of social dysfunction (ADOS, SCQ), and/or differentially associated with particular subsets of low-level behaviors (sensory hyporesponsiveness, hyperresponsiveness, and seeking).

Results: Neither ADOS Social Affect Score nor SCQ Total Score were significantly correlated with any of the Positive, Negative, and Cognitive dimensions (all p’s>.4), indicating that these scales tap new constructs overlooked in gold-standard scales for quantifying social dysfunction. Positive, Negative, and Cognitive scores were differentially associated with specific sensory responses: more positive symptoms correlated significantly with greater levels of social hyperresponsiveness (r=.610, p=.003), nonsocial hyperresponsiveness (r=.564, p=.006), and nonsocial hyporesponsiveness (r=.500, p=.018), whereas negative symptoms correlated positively and significantly only with nonsocial hyporesponsiveness (r=.512, p=.015), and cognitive symptoms correlated negativelyand significantly with nonsocial hyporesponsiveness (r=-.494, p=.012).

Conclusions: Only 27 of 107 questions across the SRS and CELF-PP uniquely captured symptoms along Positive, Negative, and Cognitive dimensions, indicating that existing scales inadequately measure variability on these dimensions. Despite few suitable questions, our results support the notion that Positive, Negative, and Cognitive symptom scales capture variability in new ways that can begin to deconstruct the heterogeneity of ASD. Future work should move toward development of new measures that capture ASD symptomatology along these dimensions, which may then contribute to new ways of parsing heterogeneity, grouping subsets of ASD, understanding neurobiology of particular phenotypes, and informing targeted treatment.