Background
Low back pain (LBP), a common musculoskeletal disorder, involves the muscle, nerve, and bone tissues of back [
1‐
3]. It has been reported that LBP ranked first in terms of disability and sixth in terms of total burden as part of the Global Burden of Disease 2010 Study [
4]. Intervertebral disk degeneration (IDD), describing the natural destruction of intervertebral disk inside the spine, has been considered as one of the major causes to motor losses and LBP. The etiology and pathogenesis of IDD is so complicated that IDD is thought to be the results of co-effects of ageing and relevant environmental factors such as sporting activities, damage, occupation, and smoking [
5‐
9]. However, there have been many articles establishing a close relationship between heredity and IDD recently [
10‐
12].
For the last several years, many genes have been discovered to be associated with IDD, some of which are collagen genes, such as Collagen I, IX and XI genes [
12‐
14]. Among these genes, the association of collagen type IX alpha 3 chain (
COL9A3) gene polymorphism with IDD risk has been studied much frequently.
COL9A3 gene, located in the chromosome 20q13.3, encodes the α3 (IX) chain of type IX collagen which is part of the interior structure of the disc, nucleus pulposus [
14‐
16]. Mutations in
COL9A3 gene leading to an Arg103Trp substitution in its 3 chain (the Trp3 allele at rs61734651 site), in other words, this change in collagen IX by substitution of glutamine by tryptophan, which is relative rare in collagen, can contribute to a disorder in the collagen properties of intervertebral disc. The increasing proportion of tryptophan in collagen can result in alterations in collagen triple helix, as well as interfering the interaction between collagens IX and II or disturbing the process of lysyl oxidase, which catalyzes cross-link formation, finally leading to disc disease [
17‐
20].
However, recent studies have obtained conflicting results. Some of them, such as Toktas et al. [
11] and Paassita et al. [
17], found that trp3 gene was a risk fact of IDD or the spinal stenosis with spondylolisthesis which is one type of IDD. Others, such as Eskola et al. [
21] and Rathod et al. [
22], did not observe a relationship between trp3 and IDD. Besides, Bagheri et al. [
18], only got an association of trp3 with IDD in males. A few articles reported the association of trp3 with ethnicity [
23]. But no meta-analysis has investigated the association between IDD and
COL9A3 trp3 polymorphism up to now. Therefore, we performed a meta-analysis to evaluate the connection between them. In this study, we aim to identify the association of genetic mutations with IDD, which is likely to be of significant importance and might help identify ‘high-risk’ individuals of IDD or guide the clinical treatment of some specific individuals.
Methods
Strategy for literature search
In order to identify all articles that studied the association of COL9A3 Trp3 polymorphism with IDD, we searched electronic databases including PubMed, Web of Science (WOS), Embase and Cochrane library up to January 01, 2018. The search strategy to screen out all possible articles involved the use of the following terms: (“COL9A3” OR “Collagen 9 alpha-3”) AND (“Gene polymorphism”) AND (“Intervertebral Disk Degeneration” OR “Disk Degeneration, Intervertebral” OR “IDD” OR “Disc Degeneration” OR “disc herniation” OR “low back pain”); (“Trp3” OR “rs61734651” OR “20q13.33” OR “arg103”) AND (“Gene polymorphism”) AND (“Intervertebral Disk Degeneration” OR “Disk Degeneration, Intervertebral” OR “IDD” OR “Disc Degeneration” OR “disc herniation” OR “low back pain”); (“COL9A3” OR “Collagen 9 alpha-3”) AND (“Trp3” OR “rs61734651” OR “20q13.33” OR “arg103”) AND (“Intervertebral Disk Degeneration” OR “Disk Degeneration, Intervertebral” OR “IDD” OR “Disc Degeneration” OR “disc herniation” OR “low back pain”). In order to increase the sensitivity of the searching strategy, both MeSH terms and free words were applied.
Inclusion and exclusion criteria
Studies included in this meta-analysis should satisfy the following inclusion criteria: (1) Evaluation of the association between COL9A3 trp3 polymorphism and the risk of IDD; (2) Human subjects; (3) Case-control study; and (4) Available genotype data were provided to calculate the odds ratios (ORs) and 95% confidence interval (CI).
Correspondingly, the exclusion criteria were defined as: (1) Comments, reviews or animal studies; (2) Duplicate reports with previous publications; (3) the study only described data of case population; (4) Studies without available genotype frequencies.
All retrieved articles were evaluated and discussed to achieve accordance by two junior investigators depending on the inclusion and exclusion strategies independently. If a conflict (among the basic information, data, and the quality of articles separately extracted by two investigators) still existed, a senior author was invited to extract the specific data independently using blind method. Then comparing the results with the two junior investigators to solve the problem and finally come to a consistency.
The following characteristics of each study were collected: (1) name of the first author; (2) year of publication; (3) country of enrollment; (4) ethnicity of the study population; (5) age and gender of individuals included; (6) diagnostic criteria for IDD cases; (7) genotyping methods; (8) source of controls; (9) matching items; (10) number of subjects under IDD cases and controls; (11) Relation with IDD; Data were extracted carefully from all eligible publications independently by two investigators. For conflict resolution, an agreement was reached by discussion.
Methodological quality assessment
The two investigators assessed the qualities of all the included studies separately using the Clark scores system, which contains 10 items [
24,
25]. Scores below 5 indicate low quality, while 5–7 scores represent moderate quality and 8–10 scores denote high quality [
24,
25].
Statistical analysis
The PRISMA checklists and their guidelines were cautiously followed during the whole process of the study [
26]. The association strength between
COL9A3 trp3 polymorphism and IDD risk was assessed by combining ORs with 95%CI. The estimations of pooled ORs were determined by the weighted average OR from each study. Significance was identified by a
P-value less than 0.05 in Z-test. The pooled ORs and 95%CI were calculated for trp3 positive (the mutation type) versus trp3 negative (the wild type). Because seven studies [
11,
14,
18,
23,
27‐
29] included in this meta-analysis only exhibited data in “Trp3 positive versus Trp3 negative” form. In other words, these studies did not have enough data to calculate ORs and 95%CI for five comparison models. In addition, although the other four studies [
17,
21,
22,
30] included this meta-analysis showed separate data in homozygous type (TGG/TGG), heterozygous type (TGG/CGG) and wild type (the others which not include TGG at this site, such as CGG/CGG) for trp3, the number of subjects for homozygous type (TGG/TGG) was too small with no more than two subjects observed in each study. So we combined homozygous and heterozygous type together as trp3 positive (TGG/TGG, TGG/CGG) and the wild type was defined as trp3 negative (the others which not include TGG at this site). In other words, the trp3 positive was defined as the presence of at least 1 Trp3 allele and the trp3 negative were the types without Trp3 allele. The statistical heterogeneity was verified by
I2 statistics. Fixed-effect model was used to estimate the ORs and 95%CI when heterogeneity was low (
I2 < 50%), while the random effects was adopted when heterogeneity was high (
I2 > 50%) [
31]. Sensitivity analyses were also performed to evaluate the function of an individual study on the pooled ORs by removing each study in turn. All analyses were performed using STATA 14 (Stata, College Station, TX). Subgroup analyses were performed to find whether sex or ethnicity of studies was linked to the value of the pooled ORs and 95%CI as well. Because only few studies included separate data of degree of IDD, we do not conduct a subgroup analysis stratified by disease degree. All
P-values were two-sided. Publication bias was checked using the Begg funnel plot [
32] and the Egger’s test [
33] (
P < 0.05 was considered statistically significant).
Discussion
IDD, a common musculoskeletal disease, is widely considered as multifactorial diseases enforcing economic and medical burdens to society. Genetic factors have been considered as one of the leading causes of IDD [
7,
11,
34].
COL9A3, an extracellular matrix molecule present in the nucleus pulposus of the intervertebral disc and cartilage, codes for Collagen IX [
35]. Collagen IX is vital for the normal cartilage development or maintenance. Mutations in
COL9A3 could cause chondrodysplasias in humans as well as articular cartilage and intervertebral discs degeneration in mice [
36].
COL9A3 gene was observed to be a key genetic influencer in the process of IDD [
23]. Previous studies have reported the association between the
COL9A3 trp3 polymorphism and IDD, but with conflicting results. With the studies with larger sample sizes of predisposing gene polymorphism, it would be much more reliable to discover the connection between candidate genes and specific type of diseases. In order to solve the inconsistence, meta-analysis was performed to examine the association of
COL9A3 trp3 polymorphism with IDD risk by critically reviewing 11 studies. Its strength came from the accumulation of various published data, offering more information to explore significant differences.
The pooled ORs (trp3 positive versus trp3 negative) and 95%CI did not show a significant association of
COL9A3 trp3 polymorphism with IDD risk in the overall populations. Different
COL9A3 trp3 frequencies have been reported in two genders of subjects (male and female) [
18]. So we performed a stratified analysis by gender to determine whether there was an association between trp3 and IDD differed by gender. We also found no association of
COL9A3 trp3 polymorphism with IDD risk both in male or female subgroup. To our limited knowledge, ethnicity may contribute to different genetic characteristics of IDD. Hence, we also performed a subgroup analysis by ethnicity and the outcomes indicated no association of
COL9A3 trp3 polymorphism with IDD risk in any of ethnicity subgroup. Based on the analyses above, we speculated that
COL9A3 trp3 might be a minor factor in genetic etiology of IDD risk due to the small amount of
COL9A3 inside the intervertebral discs [
37]. All of the results above do not eliminate the possibility of a clinically vital association that remains to be explored more carefully in convincing studies of larger sample sizes.
No significant heterogeneity was observed in subgroup analysis of gender, whereas there existed heterogeneities in the overall comparisons and subgroup analysis of continent for trp3 and IDD risk. To search the source of heterogeneity, we observed that
I2 values had significantly decreased after excluding Paassita et al. [
17] or Kelempisioti et al. [
30] in overall analysis. We also found that
I2 values had significantly decreased after excluding a study of Paassita et al. [
17] in subgroup analysis of Europe continent. The results indicated that the major source of the heterogeneity might result from these studies. However, heterogeneity did not seem to influence the results, because the lack of association between trp3 and IDD was not altered after excluding either of study mentioned above.
Moreover, no significant change of results was identified by sensitivity analyses, which indicating the reliability of results. These suggested the reliability of the results. Publication bias was also tested in this study. On the basis that a meta-analysis collects various data from numerous studies, the effect of publication bias among the articles included in the study can influence the meta-analytic results. Neither the Egger’s test nor the Begg funnel plot showed significant publication bias for this analysis. Although the results are reliable, more studies are required to be conducted in order to confirm the outcome of this meta-analysis.
Our meta-analysis has several strengths. Firstly, to our best knowledge, this is the first meta-analysis focusing on the connection between COL9A3 trp3 polymorphism and the risk of IDD. We suggest that such a method of incorporating the outcomes of related studies may help us to understand the effect of polymorphism on disease development better. Secondly, we also have taken the gender and ethnicity of subjects into account. This study included researches of Asia (Iran, China and India), Europe (Finland, Greece and Turkey) and America (USA), containing different kinds of ethnicities and enrolling both male and female. So the results are much more comprehensive. Moreover, several strategies and strict principle were applied to evaluate the methodological quality of the studies and most of the studies included in this meta-analysis possessed moderate or high qualities.
The present meta-analysis also has a few limitations that should be taken into account. Firstly, the number of filtered studies for COL9A3 trp3 polymorphism is a little bit small. Secondly, the heterogeneity was a little bit high when overall and sub group of ethnicity analyses were conducted, contributing to a cautious acceptance of the results. What’s more, some studies were removed from our research for lacking detailed data, which may contribute to selection bias. Limited to data, we only analyze trp3 positive versus trp3 negative to estimate the ORs and 95%CI rather than five models (allele, homozygote, recessive, dominant and heterozygote models). This may also influence the reliability of outcomes. Finally, although most of articles in this meta-analysis made a good match of age, gender or other items which might influence the results, some articles did not take certain items into account or even did not mention the match points. These confounding factors might affect the results.