Determinants of Response to Music Therapy in Autism: Insights from the Time-a Study

Background:

Music therapy (MT) may facilitate skills in areas affected by autism spectrum disorder (ASD), such as social interaction and communication. The TIME-A study (Bieleninik et al. JAMA 2017) examined the effect of improvisational MT on generalized social communication skills of children with ASD and represents the largest randomized controlled trial of MT for ASD. In this assessor-blinded trial of 364 children aged 4-7 years across nine countries worldwide, those assigned improvisational MT, where trained music therapists sang or played music with each child, attuned and adapted to the child’s focus of attention, to help children develop affect sharing and joint attention. The proportion of responders to MT, defined as having improved ADOS social affect score after 5 months, was 52%.

Objectives:

The availability of this large data set gives the possibility to assess the determinants of the response to MT by analyzing the baseline profile of 165 subjects randomized to improvisational MT. By using advanced machine learning systems, we are able to develop a semantic connectivity map of the factors associated with response to the therapy with the aim to derive possible new insights in interpretation of trial results.

Methods:

Data of the TIME-A study were re-analyzed with a special kind of unsupervised artificial neural network (Auto-CM). Auto-CM produces a semantic connectivity map in which the matrix of connections, visualized through a minimum spanning tree filter, takes into account nonlinear associations among variables and captures connection schemes among clusters. In this way, the patient state can be viewed as an hyperpoint in a “multimorbidity space” in which each dimension corresponds to a quantitative phenotype.

Results:

In the semantic map(fig.1), male gender was directly linked to response; age over 5 years and presence of mental retardation were linked with response through male gender followed by ADOS score above 14 and verbal ability. Concomitant treatments such as sensory-motor therapy, speech and language therapy, behavioral and educational interventions, as well as therapeutic leisure activities all pointed to the non-response node. A further reanalysis performed in the trial data set confirmed a completely different trend in response rate between the non-MT arm and MT arm. In the first subgroup the total number of hours in concomitant treatment was not correlated with difference in response rate (response rate in < 40 hours/month and > 80 hours/ month of concomitant treatment equal to 50% and to 54% respectively) while in the second group a marked difference in response rate was evident (response rate in < 40 hours/month and > 80 hours/ month of concomitant treatment equal to 66% and to 40% respectively).

Conclusions:

Response to MT depends on a complex interaction of variables. The prototypical responder to MT, as appears from TIME-A study data set, is a male child able to speak, with ADOS-SA over 14, age over 5, and mental retardation. Excess of concomitant treatments seems to interfere with response to improvisational MT.