This was a pilot study to determine the safety and potential efficacy of lenalidomide in a small number of patients and was not intended to have adequate power to achieve statistically significant findings. The fact that the majority of the patients with history of regression showed almost no expressive ability at baseline is not atypical for some regressive subtypes seen clinically, but it limited the ability of the study to measure any significant change. This study suggests that the oral TNF-α
inhibiting agent lenalidomide is a potential novel mechanism for treatment of autism with regression in the presence of elevated TNF-α
markers in CSF and serum. Clear clinical lowering of TNF-α
in both serum and CSF was observed in four patients completing the 12-week study. These findings build upon prior animal models of inflammatory cytokine elevation and change in rat offspring behavior with cytokine treatment [17
] and appear to have potential application in human patients based on these pilot study results. In addition to the number of subjects who completed the study, a limitation of this study included logistical limitations in capturing CSF data at exact baseline and followup, which caused variability in the timeline between these measurements. Due to this we cannot conclude that all the observed changes in CSF levels were due to therapy alone.
Pharmacokinetic data was successfully obtained and safety was monitored successfully during the trial. Variability was seen in one patient only in the pharmacokinetic data. Rashes that subsequently recovered were seen on 2 patients. In clinical practice with lenalidomide, rashes are a potential reflection of treatment effectiveness and many will continue treatment through a rash but this protocol specified discontinuation of treatment if a rash occurred. Mild absolute neutrophil count (ANC) decline to <1500 was the cutoff for the FDA safety requirements on this protocol. No hepatic, renal, hematologic or thrombotic changes other than this single ANC event occurred.
The results showed a trend towards improvement in language. Three of the seven patients with initial phrase speech were observed by parents and the investigator to show improvement in receptive language and CGI, reflecting significant levels of change in social and receptive language. These observations were validated through improvements in one-word receptive language testing scores, which showed clinically meaningful changes by 6 weeks of treatment that were measurable yet not statistically significant. There were an insufficient number of cases after 12 weeks of followup to compute statistics for expressive and receptive language scores. CARS improved on average greater than 2 points and showed significance after 6 weeks. These exploratory outcomes suggest some relationship to treatment were seen, but the small sample size in an open-label study limits conclusions of treatment affect.
In addition to the small sample size, this study has several limitations, one of which was the inconsistency with the laboratory processing of TNF-α, as well as the inconsistent timing between CSF samples. There is no consensus on standard laboratory tests that should be analyzed for children with autism. TNF-α and other cytokine measurements are done infrequently so there is no standardization in how these labs are processed. The current lack of commercially standardized TNF values and test protocols somewhat limits exact correlation; however CSF and serum TNF samples were processed in the same manner at baseline and 12 weeks in all patients, so we felt the values were valid for analysis in this study—especially serum levels which reflected true baseline versus end of treatment changes. In addition, serum level values could be correlated to behavioral and language scoring. Consistent testing standards are necessary and timing of samples before true meaningful cytokine data in autism can be assessed in treatment protocols.
Patients' repeated exposure to the tests and developmental maturation are potential limitations to the research findings. Further, some patients completed neuropsychological testing after lab work, the medical appointment with the physician, and additional wait time, which led to behavior noncompliance that likely affected test results. No patients suffered severe adverse reactions and all three patients who dropped out of study recovered fully from ANC levels and rashes.
Despite the limitations, to our knowledge, this open-label study represents the first attempt to treat autism by specifically targeting elevated innate inflammatory cytokine levels. Safety monitoring and pharmacokinetic data were successfully completed during this pilot study and exploratory observations of clinical and cytokine changes suggest a trend towards improvement. Correlating treatment outcomes with cytokine level changes may be a target in future autism spectrum treatment, especially in those with known maternal or postnatal immunological risk factors. Larger blinded and placebo-controlled studies assessing cytokine measurement and cytokine-targeted treatment in autism patients with TNF-α or other inflammatory cytokine elevation are warranted.