Dopaminergic brain systems have been implicated to play a major role in drug reward [20
], thus making genes involved in these circuits plausible candidates for influencing susceptibility to substance use disorders. In fact, several genes coding for the dopaminergic system have been investigated in cocaine dependence including genes for the dopamine receptor D2 (DRD2) [16
] the dopamine receptor D3 (DRD3) [10
], the dopamine receptor D4 (DRD4) [3
], the dopamine transporter (DAT) [17
] and the catechol-O-methy-transferase (COMT) gene [27
]. The results of all these studies have been conflicting with some positive reports and some negative findings, possibly due to small sample sizes and the complex genetic nature of cocaine dependence. In this study we investigated whether polymorphisms in the CDNF
gene, another gene involved in dopaminergic neurotransmission, confer risk to cocaine dependence. Neither single marker analysis nor haplotype analysis produced evidence in support of an association between the CDNF
gene and cocaine dependence among individuals of African descent. While these results suggest that polymorphisms in CDNF
do not play a major role in cocaine dependence, several limitations in our study must be considered carefully before excluding CDNF
as a susceptibility gene for cocaine dependence.
One of the conceptual strengths, and at the same time a weakness of genetic association studies, lies in their design. The key concept of DNA population association studies is the use highly informative genetic markers as surrogates for the block structure of the human genome. Investigating markers across a gene that are in strong LD, and thus likely to be inherited together as a block, reduces the amount of tests needed for association analysis. While this approach has the advantage of detecting major susceptibility factors in LD using only a few selected markers, it fails to detect rare variants that may be causative and does not take into account the complex LD patterns that may exist in the tested population. We selected four Hapmap tagging SNPs in order to provide broad coverage of the CDNF
gene. We observed moderate LD values between markers that spanned the gene with exception, however. The 3′ region of CDNF
between SNP3 (rs7900873) and SNP4 (rs2278871) is approximately 2.5kb. The observed LD between these two markers was low (D′=0.51) (), indicating that our coverage of this region was weak. Weak LD may be explained by a frequent occurrence of conversions [1
] and homologous recombination in that region of the gene, a phenomenon termed as a “recombination hotspot”. Recombination hotspots are DNA sites at which recombination occurs at increased rates due to discrete recombination signaling sequences and interacting proteins [49
]. They occur on average every 200 kilobases across the human genome [29
]. Given this weak LD, there may be polymorphisms between SNP3 and SNP4 that are associated with cocaine dependence which our study failed to detect. Additional studies are needed to genotype more SNPs in that region or conduct large scale sequencing of cases and controls to comprehensively cover the gene. Such gene coverage is necessary before conclusively excluding CDNF
as a genetic risk factor of cocaine dependence.
Another limitation of this study was the sample size, which had limited statistical power to detect risk alleles that contribute small effects to the overall disorder. This limited power could explain our negative results. Additional studies with larger sample sizes should be conducted to test for polymorphisms in the CDNF
that may contribute small effects to susceptibility to cocaine dependence. However, it must also be considered that larger sample sizes may increase genetic heterogeneity and contribute to undetected population stratification, which would negatively impact the interpretation of association analysis [28
]. Unaccounted differences in population structure can create associations in and of themselves and lead to inaccurate interpretation of results [40
]. The possibility of unaccounted differences in population structure is especially relevant to analyses involving individuals of African descent since there is substantial genetic heterogeneity among African Americans [35
]. Genomic controls and/or the utilization of family-based association studies may control for these stratification issues [2
]. Future studies investigating associations between CDNF
and cocaine dependence with larger samples sizes should utilize genomic controls or family-based association paradigms to control for stratification issues.
When conducting genetic association studies, the clinical phenotype might add additional heterogeneity which could obscure an association or lead to a false positive finding. Patients with co-morbid alcohol dependence/abuse and nicotine dependence were not excluded from our cocaine dependent population. While all patients were diagnosed with cocaine dependence in accordance with the DSM-IV criteria, co-morbid alcohol and nicotine use might have differed between patients. It has been shown that genetic factors play a role in nicotine dependence [53
] and alcoholism [19
], thus perhaps shared genetic factors contribute to all substance use disorders. Hence, unaccounted clinical heterogeneity within a population may have exacerbated genetic heterogeneity among the cocaine cases and may have led to false negative or positive findings. Although all patients were diagnosed according to DSM-IV criteria and the diagnosis of cocaine dependence was supported by urine drug screen data, the control subjects were assessed using semi-structured interviews but did not undergo urine drug testing. While drug testing is useful in establishing a diagnosis, it might not be useful for assessment of controls since it does not rule out past exposure or substance use. Unreported or minimized substance abuse in the control population is thus an important limitation that needs to be considered; however, even under the assumption that the control group had 1% of undetected cocaine dependence cases, assuming the general prevalence rate of cocaine dependence in the control population, this factor might have only minor impact when comparing cocaine cases to the group of controls.
In summary, our results do not support an association between polymorphism in the CDNF gene and cocaine dependence in individuals of African descent. However, additional studies using larger sample sizes, comprehensive SNP coverage, and independent, homogenous populations are necessary before excluding CDNF as a contributing genetic risk factor for cocaine dependence.