In this study, the case–control and family-based association analyses of the GRIN2A rs8049651 and GRIN2B rs2284411 polymorphisms found no significant association of these two polymorphisms with ADHD in Korean subjects. Previously, Turic et al.  reported that the GRIN2A rs8049651 polymorphism was associated with ADHD in a family study, but Adams et al.  failed to replicate this positive result. Similarly, Dorval et al.  reported that the GRIN2B rs2284411 polymorphism was associated with ADHD symptom severity as well as ADHD diagnosis.
The negative results in our study of these loci that had been previously reported as associated with ADHD could be interpreted in several ways. First, if the GRIN2A rs8049651 and GRIN2B rs2284411 polymorphisms are uncommon disease loci or ones of small effect, then our power to detect a gene would be reduced, and the failure to demonstrate a significant association in our case–control analysis might reflect false negative results stemming from the limited sample size and the resulting lower statistical power. Because susceptibility genes for ADHD are likely to have a small effect size , more independent case–control studies with larger sample sizes will be required to ensure sufficient study power to detect small gene effects. Second, there is a possibility that the GRIN2A rs8049651 and GRIN2B rs2284411 polymorphisms might have no involvement in the susceptibility to ADHD, at least in the Korean population. Indeed, ethnic differences may have affected the results. In particular, the allele frequencies for the G allele and for the A allele of the GRIN2A rs8049651 in the ADHD probands of this study were 0.94 and 0.06, respectively, which were very different from those reported in previous studies (0.69, 0.31 for the Turic sample, and 0.72, 0.28 for the Adams sample), suggesting that findings from Western populations may not be generalizable to the Korean population.
With regard to the GRM7 rs37952452 polymorphism, this is the first study to report an association between the GRM7 polymorphism and ADHD using case–control and family-based analyses. The biased transmission of the G allele in the TDT analysis of our ADHD trios suggests a possible role of this allele in the development of ADHD. However, we found no association between the GRM7 rs37952452 polymorphism and ADHD using the case–control approach. As reviewed by Bobb et al. , a relatively high percentage (46%) of the 26 genetic studies using the family-based and case–control approaches on the same population and polymorphism found divergent results, indicating that both methods should be used to limit the possibility of type II error. Our positive result in the family-based analysis alongside a negative result in the case–control analysis with regard to the GRM7 rs37952452 polymorphism may be partially explained by the small sample size. The number of patients having a minor allele was only 29 in the ADHD group and 19 in the control group, a disparity which might have negatively impacted statistical precision. Therefore, further case–control association studies with larger sample sizes are necessary. Another potential explanation is that differences in population substructures can affect bias estimates in case–control study designs. However, the Korean population is notably characterized by a relatively high genetic homogeneity. Therefore, stratification bias is unlikely to have affected the result of case–control analyses in our sample. It is also possible that the GRM7 rs37952452 polymorphism might have no involvement in the susceptibility to ADHD, and the positive result in our family-based analysis could be a false positive finding.
In this study, the GRM7 rs37952452 polymorphism was associated with omission errors on the CPT. Numerous studies using neuropsychological tests in individuals with ADHD have found deficits in selective attention , and continuous performance tests have been frequently used to evaluate attention deficit in ADHD children . In particular, the omission errors are an indicator of deficits in sustained attention in response to target stimuli. Thus, the results of this study suggest that the GRM7 rs37952452 polymorphism could be more closely linked to the intermediate phenotype of selective attention deficits than to the behavioral phenotype or the disease itself. The intermediate phenotype is used to describe a categorical trait or dimension that usually occurs in an “intermediate” position in the causal chain between genes and disease . Because an intermediate phenotype mediates the relationship between genotype and DSM diagnosis and is more closely connected to genetic functionality, larger magnitudes for genetic effects might be evident and more easily detectable in smaller samples . Furthermore, a relationship between omission error scores and the GRM7 rs37952452 genotype was evident in the total sample (including controls) as well as for ADHD probands, suggesting that this polymorphism might be associated with sub-threshold attention-deficit in the general population.
Although the frequency of a comorbid anxiety disorder did not significantly differ according to the GRM7 rs37952452 genotype, it is notable that the GG genotype of the GRM7 rs37952452 polymorphism was associated with higher state and trait anxiety scores even after excluding the subjects with comorbid anxiety disorders. An association between GRM7 and anxiety has been previously reported. Animal studies, for example, have shown that mGluR7 ablation can cause deficits in fear response , and mGluR7 facilitates extinction of aversive memories and controls amygdala plasticity . Elia et al.  observed an ADHD proband with a copy number variation in GRM7 gene who also presented with anxiety and suggested that this copy number variation may potentially explain certain comorbidities frequently associated with ADHD. Our results also support the idea that GRM7 rs3795245 polymorphism represents certain phenotypes such as anxious ADHD as well as the ADHD diagnosis itself.
Several limitations of this study should be noted. First, the patient group included all subtypes of ADHD, which might have contributed to clinical heterogeneity. Second, although the genotype distribution was not different according to ADHD subtypes, when we compared the CPT, CDI, and STAI-C results according to the genotypes, the subtypes might have acted as potential confounders. Third, because of the relatively small sample size, the samples of this study would have limited power to detect genes of small effect. Finally, we genotyped only one specific SNP of GRIN2A, GRIN2B, or GRM7, and this may have been insufficient to fully address the association between these glutamate genes and ADHD.