Introduction

Osteoarthritis (OA), also considered as chronic arthritis, refers to a degenerative process in the articular cartilage of the joints, subchondral bone reactive hyperplasia, inflammation and osteophyte formation based on joint degeneration or aging, and characterized by joint swelling, pain or dysfunction [1]. OA is a chronic bone and joint disease caused by cartilage degeneration and bone hyperplasia of the joint, also known as proliferative arthritis, degenerative arthritis and osteoarthritis. This disease mostly occurs in the elderly, but also in young people [2]. OA is a major disease that causes joint pain and limited activity in the elderly, and Middle-aged and elderly people suffer from serious health problems as a result of it. Several studies have found that age, joint trauma, obesity and genetic susceptibility are risk factors for OA [3, 4]. Growth differentiation factor 5 (GDF5) is also known as CDMP-1 and BMP-14. It is the growth differentiation factor that regulates tissue growth, and it is bone morphogenetic protein and a member of the transforming growth factor β family, which plays a crucial role in the progression, protection and rehabilitate of bone and cartilage [5]. It has been reported that GDF5 gene mutations can directly lead to some bone-related diseases. The content of GDF5 increases in gradient in the area where cartilage precursor cells gather, the cartilage core where long bones develop, and the joint formation areas, thereby exerting its special biological functions, such as regulating limb bud cell development and maintaining cell dynamic balance [6]. Furthermore, GDF5 also regulates the proliferation and differentiation of limb bud cells in the embryonic stage. In view of GDF5’s vital function, OA is considered to be related to it [7,8,9]. Some previous studies considered the correlation between GDF5 and OA, but some studies still believe that there is no significant correlation.

A number of shortcomings have been found in previous meta-analyses, including incorrect data extraction and insufficient analysis of population subgroups. Additionally, some literature has been updated. The main purpose of this article is to make a comprehensive analysis of the association between GDF5 rs143383 SNP and OA in different types of osteoarthritis, such as knee osteoarthritis, hip osteoarthritis and hand osteoarthritis. At the same time, from the analysis of gender and different nationalities, we can get the most accurate conclusion at present, which can be used to guide clinical work and carry out related drug research.

Materials and methods

Search strategy

This study conformed to PRISMA guidelines. The databases searched included PubMed, Cochrane Library, the Web of Science, EMBASE, China National Knowledge Infrastructure (CNKI), Wanfang Data Knowledge Service Platform and other databases to investigate the correlation between GDF5 rs143383 SNP and OA. In terms of search strategy, we used (“growth and differentiation factor 5” or “GDF5” or “rs143383”) and (“SNP” or “polymorphism”) and (“OA” or “osteoarthritis”). The language is limited to English and Chinese, and the time limit for retrieval is from the institution of the database to June 2022.

Selection criteria

Selection criteria for the review included the following:

  1. (1)

    Case–control study.

  2. (2)

    The case group meets the diagnostic criteria of OA (the diagnosis of OA was based on the American College of Rheumatology criteria), and the control group was healthy.

  3. (3)

    The full text of the original literature is available, involving GDF5 rs143383, and there are specific data on sample size, genotype and gene frequency in the case group and the control group.

The following criteria were used to exclude studies:

  1. (1)

    Other observational study designs include pedigree correlation studies, case reports, clinical trials, reviews and comments.

  2. (2)

    The case group was not in accordance with the OA diagnosis, and the control group was not healthy.

  3. (3)

    GDF5 rs143383 are not the SNP of interest, or OA is not the phenotype of interest.

Data extraction

The literature has been read by both authors in its entirety and according to the selection criteria. They searched independently the above-mentioned databases and extracted the information included in the literature. We did not include literature that could not be retrieved from the database or that was undergoing review and for which we were unable to retrieve it. The authors of this article were blinded to the authors and institutions of the studies undergoing review. Finally, data were thoroughly analyzed and extracted from all relevant studies, including a comprehensive search and a comprehensive information extraction process. When collecting data to initially screen the literature, a large number of articles are included, and there may be inconsistencies in the screening results between the two authors. In this case, our two authors will screen the literature again in strict accordance with the inclusion and exclusion criteria. Or read the full text of the article carefully with the other authors, and accurately screen out the literature that meets the inclusion and exclusion criteria listed in this article, in order to ensure the accuracy of the article.

The following information was extracted from studies included in reviewing:

  1. (1)

    First author’s last name;

  2. (2)

    The year of publication;

  3. (3)

    A description of the study’s country of source;

  4. (4)

    Ethnicity;

  5. (5)

    Sample size, genotypes and alleles of the OA group and the control group.

Quality assessment

We applied the modified Newcastle–Ottawa scale (NOS) to review the literature on the relevance between GDF5 rs143383 SNP and OA included in the study and assess its quality. The modified NOS has a total of 9 stars and includes three aspects: selection, comparability and outcome. When ≥ 5 stars, it can be regarded as good quality, and it may be necessary to adjust the relative threshold depending on the technology used.

Credibility analysis

R4.1.3 software is used to analyze the extracted data by Meta. Two-classified variables are expressed by the odds ratio (OR) and 95% confidence interval (CI). We calculate, respectively, the OR and 95% CI of the GDF5 rs143383 allele model (T vs. C), codominant homozygote model (TT vs. CC), codominant heterozygote model (TC vs. CC), dominant model (TT + TC vs. CC) and recessive model (TT vs. TC + CC), and statistics were considered significant when P values were less than 0.05. It is necessary to determine whether gene frequency in the literature is consistent with Hardy–Weinberg equilibrium (HWE). If P < 0. 05, the gene frequency distribution of the control group does not accord with HWE, and if P > 0. 05, the gene frequency distribution of the control group accords with HWE. According to ethnicity, the included population was divided into Asian and Caucasian subgroups for analysis. Heterogeneity is evaluated by I2: I2 < 50%, the heterogeneity is small, and fixed effect model is used; I2 ≥ 50%, heterogeneity is large, and random effect model is used.

Results

Literature retrieval results

According to the above retrieval strategy, a total of 308 associated articles were searched. Based on the abstracts and titles of the papers, repetition and articles unrelated to the study’s objectives were excluded. We screened 26 articles related to the topic, and inclusion and exclusion criteria were strictly followed throughout the entire text. Finally, 17 foreign articles and 1 Chinese article were included, and the 18 articles included 12,060 patients with OA (case group) and 18,401 controls (control group). As shown in Fig. 1, the literature screening procedure and outcome as well as the basic characteristics included in literature research are listed in Table 1.

Fig. 1
figure 1

Flow diagram of the literature search

Table 1 The basic characteristics of the study included

Quality evaluation result of included literature

The quality of the included literature was assessed by the modified NOS. Research quality score of 17 articles was all above 5 stars (Table 1), and the overall quality was higher. All the inclusion researches were case–control studies, and there was no exclusion or inclusion criteria that they did not meet.

Meta-analysis results

Most of meta-analysis results are P < 0.05. However, the high heterogeneity may lead to uncertainty in the results. We only reliably analyze the data with I2 < 50% to draw accurate conclusions.

OA of knee, hip and hand

All eighteen studies [10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27] presented the OA of knee, hip and hand data. In the studies of the correlation between the GDF5 rs143383 SNP and OA of knee, hip and hand, the meta-analysis showed that GDF5 rs143383 SNP codominant heterozygote model (OR = 1.17, 95% CI 1.07–1.27) is associated with the susceptibility to OA of knee, hip and hand. Conclusions of subgroup analysis proved that sensibility to OA of knee, hip and hand is associated with codominant heterozygote model (OR = 1.31, 95% CI 1.12–1.53) in the Asian population, while in Caucasian population codominant homozygote model (OR = 1.28, 95% CI 1.14–1.43), codominant heterozygote model (OR = 1.12, 95% CI 1.01–1.23) and dominant model (OR = 1.19, 95% CI 1.09–1.31) are susceptibility factors to OA of knee, hip and hand (Fig. 2). The specific results are shown in Table 4.

Fig. 2
figure 2

Forest plot of the correlation between GDF5 gene polymorphism and OA of knee, hip and hand risk. Codominant model (TC vs. CC)

Knee OA

Among the 17 articles [10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26], 18 items of data studied the connection between GDF5 rs143383 SNP and KOA (Table 2). The total meta-analysis explained that the susceptiveness to KOA is associated with GDF5 rs143383 SNP allele model, codominant homozygote model, codominant heterozygote model and dominant model. Analyzing subgroups revealed the following that codominant homozygote model (OR = 1.58, 95% CI 1.33–1.88), codominant heterozygote model (OR = 1.23, 95% CI 1.03–1.46) and dominant model (OR = 1.40, 95% CI 1.18–1.65) are associated with the susceptibility to KOA in the Asian population. We also found that among the Caucasian population allele model (OR = 1.19, 95% CI 1.12–1.27), codominant homozygote model (OR = 1.39, 95% CI 1.20–1.60) and dominant model (OR = 1.24, 95% CI 1.09–1.42) are associated with the susceptibility to KOA, while there is no statistical significance in the codominant heterozygote genes (OR = 1.14, 95% CI 0.99–1.30) (Figs. 3 and 4). The specific results are shown in Table 4.

Table 2 Characteristics of the included studies for osteoarthritis of the knee, hip and hand
Fig. 3
figure 3

Forest plot of the correlation between GDF5 gene polymorphism and knee OA risk. Codominant model (TT vs. CC)

Fig. 4
figure 4

Forest plot of the correlation between GDF5 gene polymorphism and knee OA risk. Dominant model (TT + TC vs. CC)

Hip OA

Five articles [10, 11, 14,15,16] (6 items of data) studied the relativity between the GDF5 rs143383 SNP and hip OA (Table 2), and the overall heterogeneity is high and it is impossible to draw an accurate conclusion. Through the heterogeneity analysis of the subgroup, it was found that the heterogeneity came from the Asian population group, so only the Caucasian population was analyzed. The outcome of subgroup analysis explicated that in Caucasian population allele model (OR = 1.08, 95% CI 1.01–1.16) and recessive model (OR = 1.12, 95% CI 1.02–1.24) are related to the susceptibility to hip OA, while there is no statistical significance in the codominant homozygote model (OR = 1.15, 95% CI 0.99–1.34) and dominant model (OR = 1.09, 95% CI 0.95–1.26). The specific results are shown in Table 4.

Hand OA

In the studies [11, 14, 15] of the relevance between the GDF5 rs143383 SNP and hand OA (Table 2), the meta-analysis showed that GDF5 rs143383 SNP codominant homozygote model (OR = 1.28, 95% CI 1.05–1.55) and dominant model (OR = 1.09, 95% CI 0.95–1.26) are associated with the susceptibility to hand OA, while there is no statistical significance in the codominant heterozygote model (OR = 1.15, 95% CI 0.95–1.40) (Fig. 5). The specific results are shown in Table 4.

Fig. 5
figure 5

Forest plot of the correlation between GDF5 gene polymorphism and hand OA risk. Codominant model (TT vs. CC)

Male OA of knee, hip and hand

Eight articles studied the relativity between the GDF5 rs143383 SNP and male OA (Table 3). The overall meta-analysis showed that GDF5 rs143383 SNP allele model, codominant homozygote model, codominant heterozygote model and dominant model are associated with the susceptibility to male OA, while there is no statistical significance in the recessive model. Analyses of subgroups revealed the following: codominant homozygote model (OR = 2.08, 95% CI 1.17–3.71) is related to the susceptibility to OA in the Asian males, while it is no statistical significance in allele model (OR = 1.25, 95% CI 0.99–1.59) and dominant model (OR = 1.55, 95% CI 0.94–2.55). The codominant homozygote model (OR = 1.24, 95% CI 1.02–1.51), codominant heterozygote model (OR = 1.24, 95% CI 1.02–1.50) and dominant model (OR = 1.24, 95% CI 1.03–1.49) are related to the susceptibility to OA in the Caucasian males, while it is no statistical significance in allele model (OR = 1.09, 95% CI 0.99–1.19) and recessive model (OR = 1.05, 95% CI 0.93–1.20) (Figs. 6 and 7). The specific results are shown in Table 4.

Table 3 Characteristics of the included studies for osteoarthritis of sex
Fig. 6
figure 6

Forest plot of the correlation between GDF5 gene polymorphism and male OA of knee, hip and hand risk. Codominant model (TT vs. CC)

Fig. 7
figure 7

Forest plot of the correlation between GDF5 gene polymorphism and male OA of knee, hip and hand risk. Dominant model (TT + TC vs. CC)

Table 4 The analysis results of genetic models on the association of GDF5 rs143383 polymorphism with OA

Female OA of knee, hip and hand

Eight articles studied the relativity between the GDF5 rs143383 SNP and female OA (Table 3), and the overall meta-analysis revealed that GDF5 rs143383 SNP allele model, codominant homozygote model and dominant model are related to the susceptibility to female OA, while there is no statistical significance in GDF5 rs143383 codominant heterozygote model. According to the results of subgroup analysis, allele model (OR = 1.41, 95% CI 1.16–1.72) and codominant homozygote model (OR = 1.96, 95% CI 1.28–2.98) are related to the susceptibility to OA in the Asian females, while it is no statistical significance in the codominant heterozygote model (OR = 1.14, 95% CI 0.76–1.72) and dominant model (OR = 1.46, 95% CI 0.99–2.14). The GDF5 rs143383 allele model OR = 1.17, 95% CI 1.09–1.25),codominant homozygote model (OR = 1.35, 95% CI 1.17–1.56), dominant model (OR = 1.23, 95% CI 1.07–1.40) and recessive model (OR = 1.22, 95% CI 1.11–1.35) are related to the susceptibility to OA in the Caucasian females, while there is no statistical significance in the codominant heterozygote model (OR = 1.14, 95% CI 0.99–1.31) (Figs. 8 and 9). The specific results are shown in Table 4.

Fig. 8
figure 8

Forest plot of the correlation between GDF5 gene polymorphism and female OA of knee, hip and hand risk. Allele model (T vs. C)

Fig. 9
figure 9

Forest plot of the correlation between GDF5 gene polymorphism and female OA of knee, hip and hand risk. Codominant model (TT vs. CC)

Heterogeneity and sensitivity analyses

Meta-analysis revealed heterogeneity, which may be explained by factors associated with HWE. However, HWE test (P > 0.05) was used to determine if subjects were selected reliably in each study, which showed a high level of reliability in each one. By using subgroup and the analysis of heterogeneity to determine the heterogeneity’s source, we found that Zhang et al. [26] were responsible for heterogeneity. We carefully analyzed this study and used the data provided in this paper, we calculated the OR and 95% CI, and however, we found that they were not consistent with the final results. We believe that data errors may be the main cause of heterogeneity, and the heterogeneity decreased after excluding the literature of Zhang et al. The sensitivity analysis of the included literature is carried out by using the method of excluding each study one by one, and the OR values of other studies are combined. It can be seen that the results are stable, indicating that the outcomes of Meta-analysis are believable. An analysis of subgroups was conducted, which no significant heterogeneity was detected in several studies of the Asian subgroup, indicating a good consistency. At the same time, we carried out a sensitivity analysis, and we finally found that the result is stable. We did not notice any significant changes in genotypes when we limited the number of high quality and HWE studies. The detailed results are shown in Table 4.

Publication bias

We performed funnel plots and Begg’s test to assess the publication bias of the literature. According to the funnel plots, publication bias was not evident (Fig. 10). For statistical evidence, Begg’s tests were conducted and indicate that publication bias is not apparent.

Fig. 10
figure 10

Funnel plot for publication bias among selected studies. A OA of knee, hip and hand codominant model (TC vs. CC). B Knee OA codominant model (TT vs. CC). C Knee OA dominant model (TT + TC vs. CC)

Discussion

The latest epidemiological study found that 8.1% of people over 45 years old had symptoms of OA, and the highest incidence rate was between 60 and 69 years old. And with the aging of the population becoming more and more serious, it is estimated that over 50-year-olds will account for 49% of the total population by 2050. If we do not conduct in-depth research in this direction, the proportion of elderly people suffering from OA will inevitably increase significantly [28]. Up to now, there is no effective drug to prevent or improve the progress of osteoarthritis in the whole world, and it is still focused on reducing pain and improving joint function [29, 30]. Severe OA seriously affects patients’ actions and pain, resulting in an extremely low happiness index. At present, operation is the main treatment method to eliminate pain and recover joint range of motion [31, 32]. At this stage, the cost of joint replacement surgery is relatively high. With the increasingly serious aging of the population, the social and economic burden of OA treatment will inevitably increase in the future. Therefore, the study on the pathogenesis of OA is urgent and has positive practical significance.

GDF5 participates in tissue differentiation and is related to the various bone-related diseases, which have been confirmed at the genetic level and through molecular biological mechanisms. It can provide help for early diagnosis of bone-related diseases and potential targets for treatment, and further understanding the mechanism of disease development, which can bring many new ideas to medical researchers [6, 33]. GDF5 regulates chondrogenesis through the canonical Wnt signaling pathway. Developmental dysplasia of the hip (DDH) is a multifactorial disease, which occurs under environmental and genetic influence. The genes displaying the most statistically significant co-expression link to GDF5. A specific polymorphism in GDF5 has been linked to DDH, and DDH patients more frequently carry the T allele [6, 34, 35]. This may be closely related to the occurrence and development of OA, possibly through the same mechanism. Studies in UK and Netherland have confirmed that the SNP of GDF5 rs143383 is closely correlated with the incidence rate of OA [11, 15, 16]. However, no correlation was found between OA and GDF5 gene polymorphisms in South Korea and Japanese populations [17, 18], which may be related to the genetic differences between eastern and western ancestors. This conclusion still lacks the correct guiding conclusion and has become the focus of the debate.

The previous meta-analysis data have not been updated [36], and some analyses have recorded identical data [37], which does not fully explain the relationship between GDF5 and OA. The Meta has carried out the most comprehensive analysis at present, analyzing different types of arthritis, gender and race. Although individual data are highly heterogeneous, which affects accurate conclusions, we have analyzed data with low heterogeneity and can still make effective summary because of the combination of numbers, so as to draw relatively more accurate conclusions.

The low heterogeneity gene model was selected for analysis throughout this meta. Based on the results, there was a significant correlation between GDF5 and OA in the total OA codominant heterozygote gene model in all populations with OA of the knee, hip, and hand, Asian populations and Caucasian populations, especially in Caucasian populations codominant homozygote gene model and dominant gene model.

As osteoarthritis differs in type, the correlation of knee osteoarthritis is the most obvious. The association between KOA and GDF5 can be evaluated based on an overall analysis of the four genotypes. In the overall allele gene, codominant homozygote gene, codominant heterozygote gene and dominant gene model, it is shown that there is a relationship between GDF5 gene and KOA. And it suggests the same results in both among Asians and Caucasians. In hip OA, it is generally explained that there is no significant relevance between GDF5 SNP and hip OA, and because of the high heterogeneity, this conclusion needs to be further verified, and the current study of hip OA is mainly Caucasian population and needs more data research and other ethnic groups. In hand OA, the overall analysis of the correlation between GDF5 and hand OA can be evaluated using these two genotypes. Codominant homozygote gene and dominant gene model suggest that GDF5 gene is associated with hand OA. However, the current study is focused on the Caucasian population, and the outcomes should be proved by an abundant case in other ethnic groups.

OA can arise naturally, with the aging population witnessing an increase in diagnoses of this pathology, but the root causes of OA have yet to be identified, and increasing interest is arising toward investigating biological sex as a risk factor. Clinical studies show increased prevalence and worse clinical outcomes for female patients [38, 39]. In the analysis of OA in different genders, we found that GDF5 was associated with OA in both males and females’ osteoarthritis. In males with OA, it is possible to reckon the association between GDF5 gene and OA using four genotypes. In the overall allele gene, codominant homozygote gene, codominant heterozygote gene and dominant gene model, the GDF5 gene is associated with OA, and the same results are suggested in the Caucasian population. In females with OA, three genotypes can be used to evaluate the relevance between GDF5 gene and OA. The overall allele gene, codominant homozygote gene and dominant gene model suggest that GDF5 gene is associated with OA, and the same results are suggested in Asian and Caucasian populations.

At present, it is generally believed that the pathogenesis of OA is due to the imbalance of cartilage tissue synthesis and catabolism, and the destruction of articular cartilage is difficult to repair, and the imbalance of chondrocytes synthesis and catabolism [40, 41]. Many factors are interrelated with the occurrence and progression of OA, such as joint composition and environmental factors, genetic predispositions, endocrine and metabolic diseases and mechanical injuries. Genetic factor is a significant pathogenic risk factor, and the genetic polymorphisms of multiple gene loci have been proved to be interrelated with the occurrence of OA in the region [42,43,44,45]. GDF5 belongs to the family of bone morphogenetic proteins (BMP), which is involved in bone growth and repair, such as the proliferation, differentiation, angiogenesis, and bone and cartilage formation. Research shows that GDF5 genetic polymorphisms are closely related to OA [46, 47]. However, it is not completely clear what role GDF5 plays in the occurrence and development of OA, and how it affects other signal pathways. The transmembrane serine/threonine kinases I and II can initiate GDF5’s signal cascade just like other BMPs. As a result of GDF5 binding, receptors are phosphorylated, the downstream Smad pathway is activated, and Smad is transferred to the nucleus to regulate gene transcription [48, 49]. Additionally, type I receptors bind both GDF5 and BMP2, and their complexes can recruit type II receptors, activating MAPK in the process [50]. Kan et al. [51] found that Sox11 (SRY-related HMG box11) transcription factor regulates the expression of GDF5, and Sox11 overexpressed in vitro and microsphere cell culture can directly activate the increase in GDF5 gene expression in chicken limb bud cells. The binding site of the Sox family is in the 5 ′—UTR region of GDF5 gene, indicating that Sox11 can specifically bind to this site, and the Sox11 can be used as a potential regulatory site of GDF5. However, the specific mechanism remains to be further studied and verified.

The main advantages of this study include: (1) Most of the studies contained in the meta-analysis are high quality case–control studies; (2) based on relatively large samples, we extract and analyze specific gene results. However, the current meta-analysis also has many limitations: (1) Although the meta-analysis contains a relatively large sample volume, it could still lead to overestimation and does not explain all the results; (2) the subjects of the study only include East Asian and Caucasian races and cannot reflect the overall situation. In subgroup analysis, the sample volume of every subgroup is smaller, which will also cause the analysis results to deviate from the actual situation; (3) publication bias and language bias caused by the unpublished results of some negative studies would affect the results of meta-analysis; (4) when patients’ informed consent is required in clinical research, and it comes to medical ethical issues, it is likely to result in low literature quality evaluations and inevitable biases, which will affect the reliability of meta-analysis conclusions; and (5) this paper only includes OA of knee joint, hip joint and hand joint, which is not all OA, and further data are needed to improve it. Therefore, the conclusion demands to be further confirmed by larger sample randomized controlled trials.

Conclusion

Collectively, we concluded that the GDF5 rs143383 SNP has a significant relationship with the occurrence of OA in the whole population with OA of the knee, hip, and hand. From the analysis of each group, we got the same conclusion in KOA and hand OA, but which need further verification in hip OA. Considering gender, we found a close relationship between GDF5 rs143383 SNP and OA of the knee, hip and hand, both for men and women. This conclusion is more obvious in Caucasian people.