It is well-known that allele frequencies of functional variants often differ substantially among groups that have different geographic ancestries . Therefore, our results represent the association from only European Americans in our cohort. In order to generalize our findings, further studies with data collected from other major ethnic population groups are needed. The ethnicity of the subjects was self-determined instead of using ancestry informative markers. It is generally acceptable that use of self-identified ethnicity based on the geographic ancestry is an appropriate surrogate , although less reliable than using explicit genetic data . Allelic frequencies of the genotyped SNPs showed consistent results with NCBI data, which suggest the generalizability of our sample to a European American population. Among the 5 ethnic clusters (sub-Saharan Africans; Europeans and Asians west of the Himalayas; East Asians; inhabitants of New Guinea and Melanesia; and Native Americans) based on 377 genetic markers , sub-Saharan Africans and East Asians are other major ethnic groups that need to be investigated.
The mis-sense SNP 11 (rs4680) of COMT markedly reduces enzyme activity to about 20–40% of wild-type levels [18–20]. This amino acid change may regulate the amounts of active dopamine and norepinephrine in various parts of the brain and therefore may be associated with mood and other mental processes which are closely related to pain perception [21–23] as well as the pathophysiology of neuropsychiatric disease [20, 24]. However, it did not affect clinically induced acute pain responses in our sample. This is not consistent with previous findings reporting higher responses of met/met homozygotes at 158 amino acid position . SNP 11 met/met homozygotes at codon 158, claiming to be associated with higher pain sensitivity, showed no significant association individually with clinically induced acute pain responses in our sample. An opposite trend was suggested as the met/met subjects showed lower pain responses than val/met and val/val subjects though it was not statistically significant (data not shown). Other studies also failed to replicate the higher pain responses in met/met homozygotes [10, 25]. This discrepancy may be explained by the small sample size of Zubieta et al. which compared a val/val homozygous group (n = 3) to a heterozygous group (n = 11), and the met/met homozygous group (n = 4) to a heterozygous subgroup (n = 9 of the 11). The authors did not provide ethnic information of the subjects; even a small amount of population admixture can undermine an association study and lead to false positive results . Differences in the pain induced experimental stimuli (saline injection into the masseter muscle) verses surgical extraction of impacted teeth and subsequent acute inflammation may also produce different phenotypes that activate different pathways.
Inconsistency was also found in the analysis of haploblocks. In contrast to previously reported lower sensory ratings of pain in G_C_G_G haplotypes , our haplotype data (not shown) demonstrated that G_C_G_G homozygotes are actually included in the higher pain sensitive populations, though the association is not very strong. Diatchenko et al. reported association between COMT variations and experimental pain and chronic pain conditions based on a sample composed of 85% European Americans and 15% from other ethnic populations. The experimental pain phenotype combined pressure pain thresholds, thermal pain thresholds and tolerance, temporal summation of thermal pain, ischemic pain threshold and tolerance. These different stimulus modalities likely cause pain via different neural mechanisms. Thermal and cold pain sensitivity, for example, are genetically dissociable  and there are at least 5 fundamentally different types of nociception , which may be inappropriate to mix for genetic analyses. The association between COMT and risk of developing temporomandibular disorders in Diatchenko et al. is based on 15 TMD patients out of 170 subjects and may be confounded by the ethnic admixture. Analyses for the individual SNP8 (rs6269) and SNP10 (rs4818) also did not show any significant association in our patient sample while Diatchenko et al. reported significant associations (p < 0.01). Thus, our failure to replicate previous studies related to COMT and pain [8, 9] may be due to the small sample sizes, population stratification and the composite pain phenotype used in the latter publications.
Instead of SNP 11, SNP 5 (rs740603) in intron 1 showed moderate but significant association with maximum post-operative pain rating. The observed relationship between SNP5 of COMT and maximum post-operative pain rating suggests that another mechanism including different, unidentified but significant genetic factors may play a role in acute clinical pain perception. Even if a polymorphism is not in a coding sequence, it can still affect gene function by altering the stability, splicing, localization of mRNA or generating a small RNA. Emerging evidence suggests that the non-coding portions (i.e., introns) of the protein coding gene transcripts play an important role in regulatory pathways. Regulatory elements for gene expression in introns [28, 29] or intron-derived microRNAs  are good examples. DNA sequences might be involved in the three-dimensional positioning of chromosomes in the nucleus enabling chromosome-chromosome interactions. Any portion of genomic DNA can be meaningful for the phenotype because of the complicated dynamics of DNA structure and gene expression . Gene expression is generally regulated by DNA regions outside of gene regions that actually encode proteins, and there is much that is not yet understood about this process . The information in the genome sequence must be considered in the large context of the chromosome  as well as in relationship with encoding proteins.
MAO, another catabolizing enzyme of monoamine neurotransmitters, is present in two isoforms (MAO A and MAO B), which share 70% amino acid sequence identity. These two forms differ with regard to several biochemical properties including their substrate specificity, cellular localization and regulation by pharmacologic agents. They are encoded by two closely linked genes with 15 exons on the X chromosome organized in opposite directions, tail to tail, 24 kb apart. However, MAO A and MAO B genes (MAOA and MAOB) have different tissue specific expression and functions which may be caused by differences in gene regulation including TATA box and Sp1 site . While compulsive/aggressive behavior was observed in MAOA knockout mice , MAO activity has not been investigated for its possible association with pain phenotypes. The association studies between genetic polymorphisms of MAOA, MAOB and other human behaviors show inconsistent results [36–38].
We analyzed females and males separately for MAOA and MAOB because they are located on the X chromosome. Generally, males tend to report pain stimuli lower than females and they only have one allele of homozygotes. Therefore, heterozygous patients who can only be females, may show false higher pain ratings than homozygotes who are mixed females and males. It is not surprising that most SNPs from MAOA showed similar significant associations in females because all SNPs in the MAOA region show high linkage disequilibrium with each other (see Additional file 1). Common combinations of tag SNPs of MAOA were analyzed with the maximum post-operative pain ratings and showed similar tendency with the individual SNPs, as expected. However, this association is non-significant, even without the multiple testing corrections. Though MAOA and MAOB are closely located on the X chromosome, they have many different characteristics including the expression sites and substrates. In pain sensitivity, they also have different effects as MAOB SNPs do not show any significant association.
Since the re-uptake of monoamine neurotransmitters through specific transporters is the major method of elimination from the neural synapse, it is not surprising that the genetic variations in norepinephrine transporter gene (SLC6A2) and the serotonin transporter gene (SLC6A4) affect monoamine neurotransmitter mediated human behaviors including responses to painful stimuli. Unlike catabolizing enzyme encoding genes such as COMT and MAO, SNPs of transporter genes showed association with time dependant pain responses though it is not clear how these genetic variations influence pain or analgesic onset. Given the significant findings between genetic variations of monoamine neurotransmitter systems and sensitivity to clinical pain, additional high density genotyping of these genes regions may be informative. More than 30 SNPs with heterozygosity greater than 0.25 are reported and 6 nonsynonymous SNPs exist in COMT alone. Some of these SNPs induce nonsynonymous amino acid change though their heterozygosity is low or not known. Future studies including higher density genotyping around the candidate regions are needed.
Among the dopaminergic receptors, D2 receptor (DRD2) appears to be the major autoreceptor for dopaminergic neurons . It was reported that DRD2 influences baseline nociception in the mouse, although this effect is weak and submodality selective. DRD2 receptors may contribute to attenuation of referred hypersensitivity caused by sustained nociception . Despite the important role of DRD2 in the dopamine system, no significant association between genetic variations in DRD2 and clinically induced pain was found in our sample. Due to complicated interactions not only within monoamine neurotransmitter systems but also with related other enzymes and other pain modulating systems, it is not clear whether genetic polymorphisms in DRD2 affect pain sensitivity. Many association studies for common diseases suggest that many different genes distributed throughout the human genome contribute to the total genetic variability of a particular complex trait, with any single gene accounting for no more than a few percent of the overall variability of the trait .
Even though we found some significant association between genetic variations of monoamine neurotransmitter system and clinical pain sensitivity, for modest genetic effects and for identification of genotypic subgroups, much larger sample sizes are required. The SNPs showing association with pain sensitivity are in introns or untranslated regions. More complicated unidentified mechanism rather than the simple amino acid change of coding sequence may play a major role. Regardless of the significant associations, the risk of chance finding should be considered carefully. When we performed correction of the multiple testing with Bonferroni's correction, all of our significant associations were statistically non-significant. It should be emphasized that our data set is larger and more homogeneous than the studies reporting positive findings of COMT and pain [8, 9]. Therefore, it is possible that the reported genetic associations between monoamine neurotransmitter system including metabolizing enzymes and transporters and pain are not justified by the available evidence.
A sample drawn from only one gender group may increase the ability to study phenotype-genotype association. It is reported that a gender difference exists in the genetic variation effect on pain responses [12, 13, 22, 42, 43]. However, it is still debatable that the independent analysis of each gender is necessary. It was reported that there is no evidence for a sex-specific genetic influence in the liability of heritability of clinical neck pain, especially in the clinical condition . Analyzing females and males together in our study can be meaningful because overwhelming clinical pain and allelic frequencies are not different between genders as well as it enables us to compare the results of COMT with previously studies of Zubieta et al and Diatchenko et al. It is also necessary to take account of substructure of examined population even in a relatively homogenous genetic group such as the Icelanders. Even Icelanders cannot be considered to be a single, randomly interbreeding population . This will probably be more important in larger populations like European Americans or African Americans. Descriptors such as ethnicity capture only some of the ancestral information about the biological and environmental factors that influence phenotypic characteristics . Other potential interacting factors such as psychological profiles should also be considered in the future studies.