In a genetically enriched sample of ADHD affected sibling pairs, the CBCL-PBD profile identified a small (8%) but distinct subset of individuals with severe psychopathology. These results are consistent with an earlier report describing similarly increased rates of oppositional defiant, conduct, and anxiety disorders in clinical samples.2
Also in other reports, the CBCL-PBD profile identified increased ADHD, oppositional defiant disorder, conduct disorder, and suicidality in a population based twin sample,11
and increased delinquency and suicidality in an epidemiological sample.32
The increased rate of substance abuse in the CBCL-PBD sample, while not significant, is notable given that many subjects in our sample had not yet entered the age of risk for substance use initiation, but is also consistent with previous findings.33
Looking across our sample and others, it is clear that children and adolescents identified by the CBCL-PBD profile possess significant psychopathology. There remains an ongoing clinical imperative to recognize, characterize, and develop more specific treatment strategies for these severely ill youth.
Although the potential value of QTL analyses in ADHD genetics studies has been discussed,34
we are not aware of any prior studies that have successfully employed this approach using the quantitative CBCL-PBD phenotype. Our previous investigations of ADHD symptoms counts within ADHD affected sibling pairs revealed no significant evidence for linkage,15
in contrast to when the discrete ADHD phenotype was used.16
However, in the present study using the CBCL-PBD quantitative phenotype, we found suggestive evidence for linkage on chromosome 2 in a region that showed nominal linkage (MLS 1.0) in our first scan of ADHD on the discrete ADHD phenotype and later fell below that nominal level in an extended sample.15,16
Of particular related interest, is the recent finding of significant linkage on chromosome 2q22–q24, the same region underlying our peak linkage signal, with bipolar disorder in a study of 52 families of European descent.35
This region has not, to our knowledge, been previously examined in relationship to ADHD. These emerging results may support further study of this region and use of the CBCL-PBD measure as a susceptibility profile within ADHD to identify putative genes involved in that condition and its clinical heterogeneity.
Although the overall rates of bipolar spectrum illnesses in our sample were lower than in other reports, they still represent up to a two-fold elevation above adult rates usually described in community samples, but are not near the increased rates described with ADHD. 36
It is notable that the CBCL-PBD group did have rates of diagnosed bipolar spectrum illness 3–5 times greater than seen in the CBCL-comparison and CBCL-AP groups, although these differences were not testable given the low overall rates in the sample. To our surprise, the CBCL-PBD profile did not identify subjects or families characterized by increased rates of any DSM-IV mood disorder. This contrasts with numerous prior investigations 1,2,4,6–9
but is consistent with some reports.11,32
It is unclear why such large differences in ADHD comorbidity occur across extant studies. Our sample, which required the participation of both biological parents, might be biased towards healthier, better functioning families, although in previous reports we demonstrated expected rates of comorbid disorders and psychiatric impairment.20,21,37
There are subtle assessment differences between investigations. Some studies relied on the epidemiological version of the KSADS (KSADS-E) which contains fewer probes and allows less clinician discretion than the KSADS-PL used in our investigation.1,36
Other reports used the CBCL itself as the diagnostic phenotype,5,10,32
without consideration of DSM defined syndromes.
Another explanation for the relatively low overall rates of bipolar spectrum illness seen in our sample is the method employed at our site in interpreting the DSM. PBD has been defined by some investigators as phenomenologically dissimilar from the adult presentation, more typically characterized by chronic irritability and a relative lack of positive affective features and distinct episodes.33,38,39–41
Others have argued that this approach in community settings has led to inconsistent application of DSM criteria and a general tendency to assign a diagnosis of bipolar disorder to any child with impulse control deficits and affective instability. 42–44
One attempt to address the challenges of establishing valid symptomatic boundaries for PBD is a proposed differentiation of narrow versus broad bipolar phenotypes.44
The narrow phenotype, which informed our application of DSM criteria in this study, is defined by strict DSM criteria for mania or hypomania, including discrete episodes of elevated, expansive, or irritable mood. In contrast, the broad phenotype is defined by chronic, non-episodic illness that generally lacks the hallmark symptoms of grandiosity and hyperarousal. The predominance of anxiety, oppositional defiant disorder, and conduct disorder in our CBCL-PBD group is more suggestive of the chronic irritability described with the broad PBD phenotype. It is noteworthy that while the CBCL-PBD profile failed to differentiate ADHD probands or parents with higher rates of mood disorders, almost 70% of all families had evidence of lifetime parental depression. The overall high rates of depression in these parents have previously been described and might reflect the psychosocial stress of raising multiple siblings with ADHD, in addition to any underlying biological predisposition.20
In particular view of our linkage findings, the increased rates of psychoactive substance abuse in parents of CBCL-PBD families favors a biological etiology for these disorders that merits further investigation.
There should be no disagreement, however, over the existence of a distinct pediatric syndrome characterized by chronic irritability and affective instability. Our suggestive linkage findings provide supporting evidence of a biological basis for this subgroup of ADHD patients. As we develop future editions of the DSM, the nosological choice to be made is whether we should continue to describe these individuals as bipolar, while acknowledging developmental differences in younger patients, or instead operationalize a new diagnostic category that avoids confusion with the typical adult bipolar presentation. As previously proposed, the CBCL-PBD profile appears useful as a standardized measure in understanding subjects across studies and in potentially identifying biological susceptibility to severe comorbidity.
This study has several methodological limitations. First, in constructing our CBCL categories, we relied on previously described group thresholds. 4,10
Other groups have demonstrated that slightly lower cutoffs points or quantitative interpretations of the CBCL-PBD scale lead to increased sensitivity for detecting comorbid pathology.5,7
However, as our interest was in characterizing subjects who met the originally defined categorical CBCL-PBD cutoff, not in establishing the predictive power of the rating, we choose to use the more conservative threshold. This also reflected our wish to avoid repeated testing at various threshold levels in an attempt to force the data to fit some a priori
expectation. Second, we relied on a model of additive effects in our QTL analysis, based on some prior studies.10,29
However, other work has suggested that ADHD might be inherited under a dominant model.45
While this remains a possibility, in QTL analysis the power to detect a dominant or recessive effect (if present) is generally greater than additive alone across a range of allele frequencies and residual polygenic effects.46
Thus, mis-specification of our inheritance model would, in fact, reduce the power available to detect genetic effects. Finally, the study data set has been used in previous genomewide scans for ADHD and reading disorder.15–18,47
Although linkage peaks identified in this study do not overlap with these previous scans, we have chosen not to emphasize nominal linkage peaks due to the increased probability of Type I error. Our findings need be considered with caution in light of the multiple phenotype testing in the sample. They are offered as preliminary data for comparison with other groups using this quantitative trait. Replication is needed before further investigation, i.e. fine mapping and SNP association studies, is warranted.
This study may be one of the first to identify a suggestive genetic association with such severe behavioral problems in children. It remains unclear if the CBCL-PBD profile serves to identify a more homogenous group of children with ADHD, or if it is also useful in indentifying severe psychopathology in individuals without ADHD. The question is relevant as one considers whether the identified individuals should be viewed as suffering from a severe ADHD subtype or a separate, but commonly comorbid, condition. Consensus on the diagnostic classification of these patients will enhance ongoing research and support directed treatment studies that will provide a basis for improved clinical management.