As noted earlier we were able to evaluate the cognitive performance of 38 children with autism. These 38 participants were similar to the remaining 63 (those unable to perform the cognitive tests) on CGI-Severity score, ABC Irritability score, and gender. Not surprisingly, the participants who could be tested had higher IQs (about 23 points, on average), and they were older (9.0 vs. 7.9 years). Thus, there were important differences in presumed cognitive ability and maturity, but they were similar in terms of symptom severity and gender. However, of those assigned to risperidone, similar proportions of responders and nonresponders could perform the cognitive tests. These findings have implications for generalizability of the findings, a point that we return to later.
No declines, significant or otherwise, were indicated in the measures of attention (i.e., Cancellation Task, timed math test), hand-eye coordination (Purdue Pegboard), or short term verbal memory (verbal learning task). Conversely, significant improvement occurred with risperidone in two areas of cognitive processing, namely on the Cancellation Task (correct detections) and on the Verbal Learning Task (correct recognitions); improvement occurred equivocally on the Spatial Memory Task (Dot Test, difference score). Whereas this stability of cognitive performance may appear to be at odds with reported tiredness with risperidone (Aman et al., 2005
), the two are not necessarily inconsistent. For example risperidone caused a large reduction in irritable/disruptive behavior (E.S.
1.20) and hyperactivity (E.S.
1.00), which may have enabled the participants to perform as well or better, even if somnolence were present.
) review of atypical antipsychotic studies in adults with schizophrenia suggested enhancement in the areas of verbal fluency and attention, although he cautioned that some investigations used open-label designs. In our study, subjects taking risperidone showed significant improvement compared with controls on verbal learning recognition
performance but not on short-term or delayed recall.
Current literature is mixed regarding the effect of risperidone on spatial memory. Results from Reilly, Harris, Keshaven and Sweeney (2006
), in subjects with schizophrenia, showed worsening of deficits in the maintenance of spatial working memory. In contrast, a comparison of clozapine and risperidone in patients with schizophrenia conducted by McGurk et al. (2005
) found improvements in spatial memory with risperidone and worsening with clozapine. Our study suggests an equivocal improvement in delayed spatial memory for the risperidone group compared to placebo controls. (Loss of significance with the ANCOVA may have been due merely to reduced degrees of freedom, given the smallness of the sample.)The “significant” finding was based on a difference score between no-delay and delay conditions, and only 8 participants could conduct the task. As the finding of improvement with risperidone can result from worsening in the no-delay or improvement in the delay condition with risperidone (or the opposite with placebo), this “significant” outcome needs replication.
The choice of cognitive-motor tasks in children with developmental disabilities has varied greatly, and results have also differed across cognitive constructs. Results of the cancellation task used in this study suggest that the risperidone-treated participants were more task oriented than placebo-treated subjects. This is not consistent with the Günther et al. (2006
) study, which found no risperidone effects on attention in adolescents with ADHD. However, in Troost et al. (2006
) and in the present study, aspects of memory and attention did improve (Troost et al., characterized their divided attention task as an index of working memory.) Findings for the Verbal Learning test differed from the Pandina (2007
) study in that we observed enhanced word recognition with risperidone, but Pandina et al. did not. (However, Pandina et al. did report improvements in uncontrolled one year open-label extension studies.) We are not aware of any other study with children having pervasive developmental disorders in which the Visuospatial Memory Task has been used. This is not surprising in that it proved to be very difficult to convey instructions to these children on how to perform the Dot Test.
Children with autism and accompanying irritable behavior are exceptionally difficult to assess for cognitive change. Troost et al. (2006
) commented that a testable rate of 50% of children with PDDs was “very favorable” and “would have been lower if a larger proportion of the children … had an autistic disorder.” (pg. 571) Indeed, only 2 of 14 participants (14.3%) in the Troost et al. study had a diagnosis of autism, whereas the remainder had PDD—NOS or Asperger's disorder. Our study was conducted exclusively with children having autistic disorder accompanied by serious behavior problems. Furthermore, the large majority of the original sample (83%) had intellectual disability. Collectively, these features made this group of children very challenging to test. As such, our data may provide important information on the selection of cognitive-motor tasks in this population. The Purdue Peg Board task was clearly the easiest for these youngsters to understand. By contrast, the “Dot Test” was remarkably difficult for them to grasp. The MVLT-C, Cancellation Task, and Analogue Classroom Task were of intermediate difficulty. It is easy to see how all of these latter tests may reflect important skills that are relevant to learning in traditional classroom settings. The advantages of being able to sustain attention better over time (as in the Cancellation task) and to be more capable of recognizing previously-heard words would seem apparent both in academic settings and in real life. Likewise, enhanced memory for location (Visuospatial Task) would seem to carry advantages in real-life settings as well. As partial eta squared ranged from .16 to .46, this indicates that the drug condition actually accounted for substantial amounts of variance in these outcome variables. Nevertheless, it is also somewhat reassuring that there were no indications of risperidone-associated deterioration on any of the other variables.
This study has several limitations that caution against over-embracing the statistically-significant (and one equivocal) findings. First, given the exploratory nature of this work (there is only one other study of atypical antipsychotics in children with PDDs), we adopted the .05 level for alpha. Had we corrected for multiple comparisons, none
of the comparisons would have exceeded alpha. Second, only a minority of our participants were able to perform these tasks. This resulted in (a) small sample sizes and (b) the observation that the testable group had a higher IQ and was older than the untestable group. This indicates that the findings may not be fully representative of what would be observed if all the 101 clinical trial participants were testable. However, as reported by others (Troost et al. 2006
), children with autism and irritable behavior are very challenging to assess with such cognitive tests, and the proportion of participants assessed was probably quite respectable under the circumstances. Third, we have already stated our reservations about the “significant” drug finding for the Dot test in that only a small number of participants could perform it.
Despite these obvious limitations, the findings are noteworthy for several reasons. First, autistic disorder is often coupled with substantial cognitive disability. Therefore, it is important that pharmacotherapy not increase any functional handicap that is already present. We did not see evidence of risperidone-induced deficits in performance, and there was some indication of enhancement on some variables. Second, the magnitude of change, assessed by partial eta squared, suggests sizable gains in adaptive skills if upheld by future studies. Third, it is worth noting that the data were gathered across multiple sites under double-blind conditions, which may help to discount any individual examiner effects (i.e., unintended bias). Finally, the mechanism of any improvement is unknown. These were performance tests which tapped what the participant was able to achieve on test days. At this stage, it is unknown whether the participants were simply more compliant (i.e., the changes were a secondary consequence of suppressed irritable behavior) or whether risperidone affected true cognitive ability at a more basic level.
The results reported here are reassuring to clinicians prescribing risperidone for school children with autistic disorder. It does not appear to impair academic ability or cognitive-motor performance. However, at the current state of knowledge, clinicians should not conclude that risperidone can be counted on to improve cognitive-motor performance or even that it is innocuous. The main value of these findings is to reduce one, and only one, of several safety concerns, with the serious possibilities of metabolic and neurological risks remaining. Further research will be needed to determine the robustness and extent of any favorable effects on cognitive performance.