In the present study we found significant associations between the COMT polymorphisms rs4633 and rs4680 (Val158Met) and pain reduction at long term follow-up among fused and non-fused patients with chronic LBP. The polymorphism rs4633 was in complete linkage disequilibrium (LD) with rs4680 and hence showed similar results.
Dai et al were the first to test the association between COMT
variants and surgical outcome with regard to pain and disability in 69 patients with lumbar fusion. They reported an association between improvement in ODI score and both a single marker (rs4633) and a COMT
]. Our results are in line with Dai et al. with regard to the associations observed between success of treatment and rs4633 but do not replicate their findings, as our associations were with pain reduction and not with ODI. In addition we reported association for rs4680 as well, which was not found in the study by Dai et al.
Although the pain genes could only explain 3.0% of the variance in pain change, association remained significant after adjustment for covariates. The most important covariates were the reported use of pain medication and anxiety and depression at follow-up, as the explained variance increased from about 11% for the pain gene with age, gender and treatment to about 31% with addition of pain medication and anxiety and depression. Twice as many fused patients (51%) used analgesics daily at follow up, compared with the cognitive intervention and exercises (24%) and although the co linearity and correlation statistics was acceptable for performing multiple regression analysis, fusion was both negatively (with pain medication) and positively associated with change in pain (Table ). The observed associations suggest that it seems unlikely that treatment responses can be predicted solely by the analysis of COMT gene polymorphisms and it may be recommended to assess the influence of candidate genes for both disc degeneration and pain in a larger cohort.
In the present study the largest improvement was observed for A/G
heterozygous patients using the recessive genetic model. These results are contrary to the dose effect of the SNP and along with lack of association observed in the additive model, suggest that the effect is not increasing. Studies have shown that the three genotypes of the rs4680 influence the human experience of pain differently, with Met/Met
homozygous patients being more pain sensitive compared to the Val/Val
and those heterozygous possessed an intermediate pain sensitivity [24
]. Increased enzyme activity is inversely related with the pain sensitivity and it has been reported that Val/Val
homozygotes produce an effective enzyme and vice versa, while the heterozygotes express an intermediate COMT activity [14
]. Lotsch et al. reported that carriers of Val/Val
alleles were more sensitive to pain compared to the non carriers [43
]. The present findings of a larger pain reduction in heterozygous individuals after stratification according to genotypes, does not fit with the already reported results. Frequency of Val/Met
heterozygotes was much higher in our sample compared to Val/Val
homozygotes. Due to a small sample size the findings could be false and by chance as we have not observed pain improvement in homozygous individuals which should have been in line with assumed functional effect of the enzyme. A case control study involving 61 Turkish fibromyalgia patients has previously reported a higher frequency of Val/Met
] and also an association between individuals heterozygous for COMT
gene polymorphisms rs464312 and rs6269 and pain sensitivity has previously been reported [10
The association observed in the present study was with change in pain scores at long term follow-up, while Dai et al. have assessed the change at 1
year and in contrast to our findings Dai et al. reported a greater improvement with rs4633 homozygotes and an intermediate improvement with heterozygotes. In the present study change in ODI and VAS LBP were moderately correlated (r2
0.6) and therefore it is not unlikely to detect an association with one variable and not with the other. Contrary to Dai et al. who reported a significant association between a haplotype involving rs4633, rs4680, rs6269 and rs4818 and greater improvement in ODI score, our associations for similar haplotypes did not reach statistical significance neither for ODI nor for pain change. This is despite the fact that we have found association for two individual SNPs i.e. rs4680 and rs4633, which were also a part of this haplotype. Diatchenko L et al have also reported that as a single SNP, rs4680 was not associated with pain sensitivity despite possessing the same amino acid sequence as part of different haplotypes possessing different COMT activity and pain sensitivities [8
]. These minor discrepancies could be due to population differences in LD and haplotype structures. The contrasting results make it difficult to establish the true genotypic effect. Nevertheless both these studies support the polymorphism as being associated. Genetic studies ideally require a large patient number that can tolerate the correction for multiple testing in order to avoid false positive results. Patient sample of the present study is larger as compared to Dai et al. and included fused and non-fused patients.
The mechanism of pain modulation is complex and many studies have either failed to show any associations between COMT
variation and pain sensitivity [45
], while other studies have found an association with pain sensitivity and variation in dosage requirement of morphine in cancer pain patients [25
Inclusion of covariates in the analysis and homogeneity with regard to genetic make up and ethnicity are the strengths of this study. The distribution of different genotypes of rs4680 (A/A
) was homogenous among both surgical and cognitive-exercise group (p
0.6), and hence there was no evidence for selection bias on the basis of genetic make up among patients treated by the two different modules.
The conservative estimate for smallest difference for VAS LBP change was set at 17.4 in this study. Despite having limitations of sample size and being underpowered, this study gives suggestion towards the direction in which future research should head in examining the different clinical phenotypes related to patients with lumbar disc degeneration.
LBP is a major cause of disability among the population of the industrialized world, in turn leading to a socio-economic deterioration [47
]. The pathogenesis and etiology of lumbar disc degeneration is complex regarding both degenerative process and patient’s symptomatology with regard to pain and disability. Radiologically assessed changes in the discs may or may not cause pain [48
] and there is a visible variation in the patient outcomes after treatment. Thus, there is an indication and need for spine researchers to expand their knowledge at the molecular level, with regard to genetics of the degeneration of the lumbar discs and an inter-individual difference in pain sensitivity following the treatments.
Co-morbidity in patients with LBP has also been reported. Hagen et al, reported that pain was more often located to the whole spine (not just the lower back), legs and head, in patients with LBP as compared to the normal healthy controls [49
]. Another study reported an increased prevalence of subjective health complaints in patients suffering from sciatica due to a herniated disc, compared to the normal healthy population [50
]. These findings suggest a role of underlying genetic factors related to the pain sensitivity and perception that can predispose the patient to chronic pain, not just in the lower back but in other parts of the body as well. Previous studies have focussed on disc degeneration and genetics to reveal spine specific risk factors, while the focus on the role of pain genetics in the clinical outcome of LBP patients has been lacking.