Recent genetic linkage and association studies in autism suggesting the involvement of specific susceptibility loci and genes have elicited considerable interest, but attempts to replicate the initial findings have met with limited success [Reference Muhle, Trentacoste and Rapin33]. Because of the clinical, genetic and etiological heterogeneity inherent to autism, the use of additional approaches may facilitate the identification of the susceptibility genes [Reference Leboyer27]. One such strategy is the endophenotype-based approach. Endophenotypes are variables that correlate with a given disease, are usually sub-clinical, involve fewer genes than those involved in the disease syndrome itself and could assist in the identification of genes conferring vulnerability to the disorder [Reference Gottesman and Gould14]. To meet the criteria for an endophenotype, a marker must be present in affected individuals, has to be heritable and co-segregate with the illness, it should be state-independent (i.e., manifests in an individual whether or not the illness is active), and must be found among unaffected relatives of patients at a higher rate than in the general population [14,27].
Results from clinical and molecular genetic studies suggest that autism and obsessive compulsive disorder (OCD) could share a common pattern of heritability. Among a large number of clinical variables evaluated, obsessive-compulsive/repetitive behaviors have been found to be highly correlated among autism probands and their sib-pairs [11,50,51], and among autism probands and their first- and second-degree relatives [4,5,31,57]. Furthermore, Hollander et al. [Reference Hollander, King, Delaney, Smith and Silverman20] showed that autistic children with high scores on narrow restricted interests and rituals were significantly more likely to have one or both parents with OCD, especially fathers, than those with low scores. In addition, a recent genome scan in autism showed that the bulk of evidence for linkage to chromosome 1q24 derived from families most severely affected by obsessive-compulsive behaviors [Reference Buxbaum, Silverman, Keddache, Smith, Hollander and Ramoz6]. These studies suggest that the identification of an endophenotype in autism could be facilitated by the search of characteristics shared by OCD and autism. One such common feature could be executive dysfunctions.
Impairments in executive functions have been previously reported in patients with autism or OCD. The executive function domain includes skills required to generate and drive on complex strategies, including planning, working memory, inhibition and flexibility. Planning, i.e., the ability to achieve a goal through a series of intermediate steps [Reference Ozonoff, Cook, Coon, Dawson, Joseph and Klin37], has been mainly explored with the Tower of London task in probands with autism [Reference Hill18] and in their relatives [21,22,39]. These studies provided converging evidence showing that impairments in planning could be part of the broader autism phenotype, defined as the autism-like traits present in relatives of autistic individuals, milder but qualitatively similar to those seen in probands, and considered a manifestation of the genetic liability to autism [Reference Dawson, Webb, Schellenberg, Dager, Friedman and Aylward9]. Results are less obvious for a deficit of working memory in autism, i.e., the capacity to keep in mind and organize spatial or verbal information, as this process is often embedded in tests of planning and rarely tested alone. Only the study by Hughes et al. [Reference Hughes, Leboyer and Bouvard21] included a specific task of working memory and found impairments in parents of probands when compared to controls. Inhibition abilities that require the capacity to interrupt motor or verbal responses have also been explored in patients with autism, with most of the studies suggesting that these skills are preserved [Reference Ozonoff and Jensen38]. Mental flexibility, which refers to the ability to shift to a different thought or action according to the context of situations, has been tested mainly with the Wisconsin Card Sorting Test (WCST). Individuals with autism are highly preservative in their responses, indicating a lack of flexibility [Reference Hill18]. However, such deficits have not been reported in unaffected first-degree relatives of probands with autism [39,52].
In OCD, previous studies have shown that patients are not impaired in their general cognitive functioning [Reference Galderisi, Mucci, Catapano, D'Amato and Maj12] but more specifically in organizational strategies mostly linked with planning ability [40,46]. For instance, OCD patients take more time to generate an alternative strategy after an error when performing a computerized version of the Tower of London task [54,55]. Other executive function domains such as inhibition or flexibility have also been explored but the results remain contradictory (for review see ref. [Reference Kuelz, Hohagen and Voderholzer26]). Most of the studies were limited in size and confounding variables related to the disorder such as age of onset, medication, comorbidity and severity of the disorder could explain the discrepancies observed. The study of executive functions in the unaffected relatives of OCD probands, which has never been conducted, could help to clarify these divergent results.
Neurocognitive dysfunctions are considered among the most promising candidate endophenotypes in psychiatric disorders [Reference Cornblatt and Malhotra7]. Given that planning deficits are observed both in autism and OCD probands, we hypothesized that executive impairments, specifically in planning, could represent an endophenotype shared by relatives of patients with OCD or autism. To explore this question, we evaluated different components of executive functions in unaffected first-degree relatives of OCD probands and in first-degree relatives of patients with autism, and compared them to a group of sex-, age-, and educational level-matched healthy controls.
2. Subjects and methods
Unaffected parents and siblings of probands with autism or OCD were asked to participate in the study. Relatives were directly interviewed by experienced psychiatrists using the French version of the Diagnostic Interview for Genetic Studies (DIGS) [Reference Nurnberger, Blehar, Kaufmann, York-Cooler, Simpson and Harkavy-Friedman35] or the Schedule for Affective Disorders and Schizophrenia for School Age Children – epidemiological version (KIDDIE-SADS-e) [Reference Orvaschel, Puig-Antich, Chambers, Tabrizi and Johnson36], to confirm the absence of current mood disorder or substance abuse/dependence at time of inclusion and the lack of lifetime schizophrenia or OCD. Also, a non-structured interview was conducted by clinicians with expertise in autism spectrum disorders to exclude relatives suffering from autism spectrum disorders. Finally, to be included, relatives could not have been receiving psychotropic treatment at the time of assessment. The sample was composed of 58 unaffected relatives (18 fathers, 21 mothers, 9 brothers, and 10 sisters) of 23 patients with autism and 64 unaffected relatives (25 fathers, 28 mothers, 5 brothers, and 6 sisters) of 30 OCD patients. Demographic characteristics of the relatives included are presented in Table 1.
Kruskall–Wallis test and Pearson χ2 test indicated no significant differences among groups for age (Kruskal–Wallis, χ2 = 2.04, 2 df, p = 0.36), gender (Kruskal–Wallis, χ2 = 0.84, 2 df, p = 0.65) and education level (Pearson, χ2 = 0.26, 2 df, p = 0.87).
a Mean ± SD.
b Education level was considered as a dichotomous variable: less than 11 years of schooling versus 11 or more years of schooling (equivalent to a high school diploma) and was calculated only for subjects older than 18 years.
The probands of the relatives were recruited at two university hospital psychiatric departments for children (Hôpital Robert Debré, Paris) and adults (Hôpital Albert Chenevier, Créteil). Probands had to fulfill DSM-IV criteria for autistic disorder or OCD (American Psychiatric Association 1994). The diagnosis of autism was confirmed with the Autism Diagnostic Interview-Revised (ADI-R) [Reference Lord, Rutter and Le Couteur29]. Probands with autism were included after a medical workup including karyotype, molecular testing for fragile X syndrome, and metabolic screening. Subjects demonstrated to suffer from medical conditions associated with autism, such as fragile X syndrome or chromosomal abnormalities were excluded from the study. Among the 23 patients with autism whose relatives were assessed, there were 13 males and 10 females, with a mean age of 21.2 ± 6.8 years (range 10–36 years); 18 were non-verbal and had moderate to severe mental retardation and 5 had mild mental retardation, with a relatively preserved language. Probands with OCD were evaluated by a direct clinical interview with either the DIGS for patients over the age of 15 years, or the KIDDIE-SADS-e for patients aged less than 15 years. Among the 30 OCD patients whose relatives were assessed, there were 14 males and 16 females, with a mean age of 23.4 ± 12.5 years (range 10–42 years).
Controls (n = 47) were healthy volunteers recruited among blood donors at the Pitié-Salpêtrière Hospital (Paris) for those aged over 18 years (n = 39) and through advertisement at a local middle school for younger subjects (n = 8). Adult controls were included after being interviewed with the DIGS to confirm the absence of lifetime personal history of psychiatric disorders. Subjects younger than 18 years were interviewed with the French version of the Mini-International Neuropsychiatric Interview (MINI-Kid) [Reference Sheehan, Lecrubier, Sheehan, Amorim, Janavs and Weiller49] to confirm the absence of lifetime psychiatric disorders. In addition, questions adapted from the Family Interview for Genetic Studies (FIGS) [Reference Maxwell30] were asked to the healthy volunteers to confirm the absence of family history of psychiatric disorders in first- and second-degree relatives. Control subjects were receiving no psychotropic treatment at the time of assessment. Demographic characteristics of the control group are presented in Table 1. When compared to the group of relatives, controls did not differ for age (Kruskall–Wallis, χ2 = 2.04, df = 2, p = 0.36), gender (Kruskall–Wallis, χ2 = 0.84, df = 2, p = 0.65) or education level (Pearson, χ2 = 0.26, df = 2, p = 0.87).
All unaffected relatives and controls had to be from European descent and French speaking to be included in the study. The local Research Ethics Board approved the study protocol. Written informed consent was obtained from all participating subjects. If the proband was under 18 years old, the proband's consent and written parental consent were obtained.
2.2. Cognitive tasks
All participants were tested individually by an experienced psychologist. Five neuropsychological tests assessing various components of executive functions were carried out in the same order for all subjects. The first test performed was the Tower of London task [Reference Shallice48], requiring planning and spatial working memory [Reference Keith Berg and Byrd25]. The participant was asked to rearrange the three disks placed on three different pegs, drawn at the bottom of a paper sheet, so that they matched the positions of the goal state shown on top. There was no way of manipulating the material so the subject had to solve the problem mentally. The participant was given a set of two-, three-, four- and five-move problems presented randomly. This task was divided into two levels of difficulty: ‘easy’ problems (two and three moves) and ‘difficult’ problems (four and five moves) [Reference Hughes, Leboyer and Bouvard21]. The subject's performance on this task was measured by the number of correct solutions given for each level of difficulty (six items for each level) and the time taken to solve each problem.
The second test carried out was the Trail Making test [Reference Reitan and Wolfson44]. We used part B of this test, which assesses flexibility and mental set shifting [Reference Lezack28]. The subject had to draw a continuous line to alternately connect the digits 1 to 13 and the letters A to L on a sheet of paper. The time taken to complete the task was recorded.
The third test proposed was the Design Fluency task [Reference Jones-Gotman and Milner23]. The ‘fixed condition’ was used, which requires productivity and flexibility on a spatial material [Reference Demakis and Harrison10]. Subjects were asked to generate as many different figures as possible in three minutes, using two vertical parallel lines, one large white circle and one small black circle. The number of correct figures generated was noted and clusters of similar drawings were counted.
The fourth test was the Verbal Fluency task [Reference Benton3], which assesses productivity and flexibility on a verbal material [Reference Troyer, Moscovitch, Winocur, Alexander and Stuss53]. The participant was asked to generate as many words as possible beginning with a specified letter (f, a, or s) within three minutes. The number of correct words was noted (excluding errors and repetitions) as were ‘phonemic clusters’ (groups of words beginning or ending with the same two letters and homonyms).
The last and fifth test was the Association Fluency task. This test was specifically designed by our research team to measure the capacity of subjects to produce a list of associated words after being presented with an initial word. The specificity of this last test is the absence of cues (i.e., a letter or a name of category), thus requiring the subject to use his imagination. Each subject was given three minutes to produce as many words as possible after the presentation of a word (‘house’, ‘airplane’ and ‘camel’); the test was repeated three times, once for each word.
2.3. Statistical analysis
Because the observed distribution of continuous variables was not normal in the groups of unaffected relatives and controls, inter-group comparisons were made with a non-parametric test, the Kruskal–Wallis test. For non-continuous variables, the Pearson χ2 test was used. To test diagnostic specificity, pair-wise comparisons were performed with the non-parametric Mann–Whitney U-test. All relatives and controls completed all the tests, thus the number of subjects included in all the comparisons was the same. All p values are nominal and not corrected for multiple testing.
Mean performances on executive function tests for both groups of unaffected relatives and for controls are summarized in Table 2. Among the five tests assessing executive functions, inter-group comparisons with the Kruskall–Wallis test revealed significant differences between unaffected relatives of probands with OCD, unaffected relatives of probands with autism, and controls only for the Tower of London task. Differences were observed for the total score, as well as the scores for the first and the second level of difficulty. Significant differences among the three groups were also observed for the total response times and the response times of the first and second level of difficulty of the test. By contrast, no significant differences were observed among the three groups in terms of reaction time in the Trail Making test, number of drawings and clusters in the Design Fluency test, number of words and clusters during the Verbal Fluency test, and number of words produced during the Association Fluency test.
Data represent mean ± SD. Inter-group comparisons were made with the Kruskal–Wallis test; *p < 0.05, **p < 0.01, ***p < 0.001. Pair-wise comparisons were made with the Mann–Whitney test; autism vs. controls, #p < 0.05, ##p < 0.01, ###p < 0.001; OCD vs. controls, §p < 0.05, §§p < 0.01, §§§p < 0.001.
Pair-wise comparisons of the number of responses on the Tower of London task revealed that relatives of probands with autism gave fewer correct answers than controls during the first and the second levels of difficulty of the task, so the total number of correct solutions was also significantly lower. Furthermore, the relatives of individuals with autism took significantly more time than controls during the first and second levels of difficulty; the total response time was also significantly higher in the autism relatives than in controls.
Similar abnormalities were observed when comparing the number of responses in the unaffected relatives of subjects with OCD to controls. The OCD relatives gave fewer correct answers for both levels of difficulty of the Tower of London task when compared to controls, and the total score was also lower. In addition, the relatives of OCD probands took more time than controls on the first level of the test, but not on the second level or total response times.
Finally, the pair-wise comparison of the two groups of relatives for the first level of the test (for which both groups took more times to solve problems) did not reveal any significant differences (U = 1671, p = 0.34). Furthermore, no difference was observed between the two groups concerning the number of correct answers given, whatever the level of the task performed (first level, U = 1788, p = 0.70; second level, U = 1812, p = 0.82; entire task, U = 1802, p = 0.79).
Neurocognitive dysfunctions, specifically executive impairments, were tested as putative shared abnormalities in unaffected relatives of patients with OCD and autism. Our study revealed that both groups of relatives performed significantly worse than controls on the Tower of London task. To our knowledge, this is the first study to report executive dysfunctions in unaffected relatives of OCD patients. We suggest that these dysfunctions could be a characteristic shared by the unaffected relatives of patients with OCD and autism. Thus, specific executive processes, such as planning and spatial working memory, could aggregate in these families and might be a heritable component of OCD and autism. These specific executive traits might be used as endophenotypes to dissect the genetic complexity of both disorders. Furthermore, our findings contribute to a growing body of evidence suggesting a relationship between OCD and autism.
4.1. Executive dysfunctions in autism and OCD
Impairments in executive functions have been commonly reported in autism [Reference Pennington and Ozonoff41]. Although a few studies have failed to observe deficits in autistic subjects (e.g., refs. [8,16]), this could be a function of the tests used. Probands of all ability ranges have been found to be disturbed, especially in planning and working memory [Reference Hill18]. The same dysfunctions have been reported in their unaffected relatives, with impairments commonly observed in the Tower of London task [21,22,39]. The very significant differences in terms of performances and response times we obtained on this test between first-degree relatives of individuals with autism and controls, regardless of the level of difficulty, further strengthen that these executive processes belong to the broader autism phenotype [18,37]. The heritability of deficits in planning, which is suggested by the familial aggregation, could represent one of the most important heritable components of autism [Reference Dawson, Webb, Schellenberg, Dager, Friedman and Aylward9].
In OCD, fewer reports have explored cognitive functions in probands, but executive dysfunctions have been observed [Reference Kuelz, Hohagen and Voderholzer26]. The difficulties to solve problems in the Tower of London task that we observed in the unaffected relatives of individuals with OCD reinforce several studies in probands suggesting planning and working memory dysfunctions [15,42,43,54,55]. A recent study using a computerized version of the Tower of London task to compare OCD patients with controls found results similar to those we obtained in relatives of subjects with OCD: a lengthened time to answer the conditions requiring one or two moves and decreased performance scores whatever the level of difficulty of the task [Reference Van den Heuvel, Veltman, Groenewegen, Cath, van Balkom and van Hartskamp54]. Interestingly, this same study showed that independently from the severity of the illness, OCD patients show decreased frontal-striatal responsiveness during planning, mainly in the dorsolateral prefrontal cortex and caudate nucleus [Reference Van den Heuvel, Veltman, Groenewegen, Cath, van Balkom and van Hartskamp54]. The state-independent nature of these abnormalities, their neuro-anatomical correlates, and their familial aggregation, reinforce the hypothesis that planning deficit could be an endophenotype in OCD.
4.2. Executive dysfunctions as a shared characteristic of OCD and autism
Despite results from clinical and molecular genetic studies suggesting that autism and OCD could share genetic factors, it is surprising that symptomatically so different disorders could both be linked to the same cognitive and/or neural dysfunctions. One possibility is that even if both groups of relatives performed worse than healthy controls on the Tower of London task, they carried out the task in different ways. For example, given that OCD is an anxiety disorder, it is conceivable that the relatives of OCD probands made more errors or took longer time during the task because they were more worried about their performance. By contrast, as autism involves a preference for rule systems, it is also conceivable that parents of individuals with autism may have spent more time wanting to run through different permutations in their mind to see how this would lead to different outcomes, and thus took more time to fulfill the task. Thus, the same pattern of performance in both groups of relatives could be underlined by different cognitive ‘styles’.
Another explanation could be that the executive dysfunction observed in OCD and autism could depend on a specific clinical feature shared by both disorders. Specifically, repetitive behaviors constitute a core symptom observed in OCD as well as in autism. A criterion for autism diagnosis is the presence of rigid and repetitive behaviors, restricted interests and marked resistance to change. A preliminary study reported that children with autism who had a higher total score on the repetitive domain of the ADI-R had significantly more often one or both parents with OC traits or disorder, compared with those who had a low score [Reference Hollander, King, Delaney, Smith and Silverman20]. Moreover, a recent study reported similar frequencies of OC symptoms in OCD and Asperger syndrome [Reference Russell, Mataix-Cols, Anson and Murphy45]. In addition, Hollander et al. [Reference Hollander, Anagnostou, Chaplin, Esposito, Haznedar and Licalzi19] reported that high ADI repetitive behavior scores are strongly correlated with increased volume of the right caudate in individuals with autism. Since large caudate volumes have also been described in adults with OCD [Reference Scarone, Colombo, Livian, Abbruzzese, Ronchi and Locatelli47], possible common pathways associated with repetitive behaviors in OCD and autism could exist [Reference Hollander, Anagnostou, Chaplin, Esposito, Haznedar and Licalzi19].
4.3. Executive dysfunction as a specific component of OCD and autism?
As in OCD and autism, executive dysfunctions have been found to aggregate in families of individuals with schizophrenia or bipolar disorder [17,24]. Therefore, the specificity of the cognitive dysfunctions observed in this study remains to be clarified. The five different tests that were proposed in this study tried to explore further the specificity of executive processes related to OCD and/or autism. Only the Tower of London task showed highly significant differences between relatives of both proband groups and controls. One explanation could be the task complexity, the Tower of London task being considered as the most difficult task of our study. However, our results showed that both groups of relatives performed less well than controls even in the easiest level of the task. A better explanation could be that none of the other tests implicates the simultaneous participation of planning and spatial working memory processes (i.e., organization and manipulation of visual mental images). Finally, as recently suggested by Goldberg & Weinberger [Reference Goldberg and Weinberger13], behavioral measures used alone might be insufficiently discriminative to further explore the specificity and the complexity of cognitive processes implicated in psychiatric phenotypes. Thus, to improve the exploration of the cognitive processes shared by autism and OCD and their neuro-anatomic correlates, it would be useful to replicate our findings associating cognitive tasks and functional neuro-imaging.
There are several limitations to our study that need to be considered when interpreting our current findings. As in previous family studies that explored executive functions in autism [Reference Hughes, Leboyer and Bouvard21] or schizophrenia [Reference Kathmann, Hochrein, Uwer and Bondy24], both groups of relatives and controls were matched on the education level and not on a general measure of intelligence (IQ). Although some studies analyzing the association between executive functions and IQ found few or no correlations between WISC [1,2] or WAIS subtests [Reference Welsh, Labbe and Delayney56] and executive tests, it is usually admitted that IQ is correlated with performance on executive functions. Thus, it would be necessary to further control the potential impact of intellectual functioning between groups by the use of a test measuring IQ in additional works. However, due to the restricted size of the groups in the present study, it seems improbable that, whatever the strategy used, we can rule out a confounding effect of IQ on our results.
A second limitation concerns how we determined that relatives of autism and OCD patients were affected or not. It is well known that phenotypes are more dimensional than categorical in autism [Reference Micali, Chakrabarti and Fombonne31] as well as in OCD [Reference Miguel, Leckman, Rauch, do Rosario-Campos, Hounie and Mercadante32]. Although numerous studies have tried to better define the broader phenotype of autism [Reference Dawson, Webb, Schellenberg, Dager, Friedman and Aylward9] or OCD [Reference Nestadt, Addington, Samuels, Liang, Bienvenu and Riddle34], the candidate traits to describe these broad phenotypes remain uncertain. Thus, the exclusion criteria for the relatives in our study were based on a narrow phenotype as described in the DSM-IV, i.e., presence of full criteria for OCD or for one of the autism spectrum disorders.
A third limitation is that the sample of probands with autism was not representative of the sex ratio characteristic of this disorder (1 female:4 males), as almost half of them were female. Even if no relation has ever been demonstrated between the gender of the probands and the cognitive performances in autism relatives, this characteristic of our data set could restrict the generalization of our findings.
Finally, the use of “super-normal” controls (without personal or family history of psychiatric disorders) could generate a bias if we hypothesize that they could carry protective genetic factors for psychiatric disorders. If this were true, then part of our results could be due to these protective factors in the controls and not to the vulnerability factors for executive dysfunctions present in relatives. However, this possibility seems unlikely, given that previous studies have found results similar to ours when studying the usual control populations.
On the basis of the present results, we conclude that deficits in planning and spatial working memory processes aggregate in families of probands with OCD and autism and might constitute a shared intermediate cognitive phenotype. Further work is required to confirm our findings in relatives of OCD patients. If such studies confirm that specific executive dysfunctions aggregate in both types of families, the assessment of this cognitive domain could become an important quantitative trait in alternative phenotypic strategies for genetic studies of autism and OCD. Such assessments could also improve our understanding of the developmental mechanisms underlying executive dysfunctions in affected patients and unaffected relatives.
We thank the patients and the families who participated in this study. This work was supported by grants from Assistance Publique-Hôpitaux de Paris (CRC and PHRC AOM95076), Fondation pour la Recherche Médicale, INSERM, Ministère de la Recherche (contrat Cognitique) and Fondation de France.