Purpose: To estimate diagnostic yield and genotype-phenotype correlations in a cohort of 811 patients with lissencephaly or subcortical band heterotopia. Methods: We collected DNA from 756 children with lissencephaly over 30 years. Many were tested for deletion 17p13.3 and mutations of LIS1, DCX, and ARX, but few other genes. Among those tested, 216 remained unsolved and were tested by a targeted panel of 17 genes (ACTB, ACTG1, ARX, CRADD, DCX, LIS1, TUBA1A, TUBA8, TUBB2B, TUBB, TUBB3, TUBG1, KIF2A, KIF5C, DYNC1H1, RELN, and VLDLR) or by whole-exome sequencing. Fifty-five patients studied at another institution were added as a validation cohort. Results: The overall mutation frequency in the entire cohort was 81%. LIS1 accounted for 40% of patients, followed by DCX (23%), TUBA1A (5%), and DYNC1H1 (3%). Other genes accounted for 1% or less of patients. Nineteen percent remained unsolved, which suggests that several additional genes remain to be discovered. The majority of unsolved patients had posterior pachygyria, subcortical band heterotopia, or mild frontal pachygyria. Conclusion: The brain-imaging pattern correlates with mutations in single lissencephaly-associated genes, as well as in biological pathways. We propose the first LIS classification system based on the underlying molecular mechanisms.
