No Evidence for GADL1 Variation as a Bipolar Disorder Susceptibility Factor in a Caucasian Lithium-Responsive Cohort

To the Editor: Bipolar disorder has a complex mode of inheritance and high heritability (1, 2), although only a fraction is explained by linkage and genome-wide association study findings. This has prompted attempts to narrow bipolar disorder heterogeneity by studying subphenotypes such as treatment response. Our group has led efforts to characterize responders to lithium monotherapy and developed the Alda lithium response scale (1). We showed bipolar disorder family history association with long-term treatment response and familiality of lithium response itself (1). Approximately 30% of bipolar patients show full lithium response, and family history of lithium-responsive bipolar disorder among first-degree relatives increases the likelihood of excellent response (odds ratio=3.7) (2). Other predictors of lithium response are episodic course of illness, low rates of comorbid conditions, and absence of rapid cycling (2).

Recently, GADL1 (glutamate decarboxylase-like 1) was postulated as a bipolar disorder risk gene in lithium-responsive Han Chinese individuals (3). Given the lithium-response familiality, we investigated our multigenerational family cohort collected and characterized for lithium response using the Alda scale (1). We screened six Caucasian families with transmission of bipolar disorder and positive lithium response across generations. DNA from affected individuals, informative family control subjects, and unrelated control subjects were whole genome sequenced. We did not detect the single-nucleotide polymorphisms rs17026651 and rs17026688 or the IVS8+48delG splicing variant found by Chen et al. (3). No other putatively damaging segregating variants emerged, and only two potentially damaging exonic variants were identified in the entire data set, neither segregating with affected status in the family (Table 1).

We validated these findings by scanning our extended cohort of bipolar families for variants in the coding and adjacent intronic sequence of GADL1 using whole exome sequencing in 41 families (lithium responsive, N=20; lithium nonresponsive, N=21). Only three exonic GADL1 variants were identified in both responders and nonresponders, none of which segregated with affected status in any family. Furthermore, we found the IVS8+48delG intronic splicing variant, suggested to be causal in the Chen et al. study, in a single bipolar individual and 32 nonpsychiatric control subjects (Table 1 [also see the data supplement accompanying the online version of this letter]). Secondly, we scanned 45 excellent lithium responders subjected to whole exome sequencing using the same parameters, for which it was not possible to collect additional family members. None of these individuals were found to carry the IVS8+48delG variant, nor any other potentially damaging variant.

Consequently, we cannot conclude that GADL1 is a susceptibility gene for lithium-responsive bipolar disorder in our cohort. The few exonic variants identified here do not segregate with affected status, which is in contradiction with the genetic model of bipolar disorder. Perhaps we did not achieve the result achieved by Chen et al. in our cohort because the variant has a lower population frequency in Caucasians (2). However, given the extremely strong association reported with lithium-responsive bipolar disorder and the fact that we used a comparable clinical characterization in the Alda scale, one would expect at least some evidence for GADL1 in our combined data set of 275 bipolar individuals, more than one-half of whom are excellent lithium responders with Alda scores >7. Our sample of 45 unrelated lithium responders alone had a better than 95% chance to detect the IVS8+48delG splicing variant even if it had a frequency of only 3.3%, far below the rate reported by Chen et al. (55.8%). Our finding is not solitary, since several recent reports did not show support for GADL1 (4, 5), notably for two similarly powered and ethnically matched cohorts with Han Chinese or Japanese ancestry (N=218 and N=154, respectively).