The authors have declared that no competing interests exist.
Achilles tendinopathy or rupture and anterior cruciate ligament (ACL) rupture are substantial injuries affecting athletes, associated with delayed recovery or inability to return to competition. To identify genetic markers that might be used to predict risk for these injuries, we performed genome-wide association screens for these injuries using data from the Genetic Epidemiology Research on Adult Health and Aging (GERA) cohort consisting of 102,979 individuals. We did not find any single nucleotide polymorphisms (SNPs) associated with either of these injuries with a p-value that was genome-wide significant (p<5x10-8). We found, however, four and three polymorphisms with p-values that were borderline significant (p<10−6) for Achilles tendon injury and ACL rupture, respectively. We then tested SNPs previously reported to be associated with either Achilles tendon injury or ACL rupture. None showed an association in our cohort with a false discovery rate of less than 5%. We obtained, however, moderate to weak evidence for replication in one case; specifically, rs4919510 in
Achilles tendinopathy or rupture and anterior cruciate ligament (ACL) ruptures are frequent sources of pain and dysfunction in recreational and elite athletes. Recent studies have tested a few single-nucleotide polymorphisms (SNPs) in a small number of candidate genes for association with Achilles tendon injury or ACL rupture in athletes. For Achilles tendon injury, studies found 19 DNA variations residing in 12 genes that were associated with Achilles tendinopathy at p<0.05 using cohorts containing between 52 and 184 athletes [
The purpose of this study was to perform a screen of the entire genome for polymorphisms associated with either Achilles tendon injury or ACL rupture in a large dataset. We also analyzed specific SNPs previously reported to be associated with either injury.
Genome-wide association screens were performed for either Achilles tendon injury (defined as tendinopathy, rupture or repair) or ACL rupture using data from the Genetic Epidemiology Research on Adult Health and Aging (GERA) cohort. The generation of the data and pipeline for data analysis have been previously described in Jorgenson et al., 2015 and Roos et al., 2016 [
The GERA cohort is comprised of 110,266 adult men and women members of Kaiser Permanente Northern California (KPNC) Medical Care Plan. It is a component of the KPNC Research Program on Genes, Environment and Health (RPGEH). A complete description of the cohort and study design can be found in dbGaP (Study Accession: phs000674.v1.p1). The average age of the participants at the time of sample collection was 62.9 years old (standard deviation = 13.8 years). Length of membership in KPNC averaged 23.5 years. The medical record contains the entire patient history, including injuries that occurred prior to enrollment in KPNC, if reported by the patient and recorded by the physician.
Our analysis cohort (n = 102,979) includes 59,671 females, 43,242 males, and 66 individuals of ambiguous sex. Sex was determined by heterozygosity on the X chromosome [
Participants were genotyped at over 650,000 SNPs on four ancestry group-specific Affymetrix Axiom genome-wide arrays optimized for individuals of European (EUR), African-American (AFR), East Asian (EAS), and Latino (LAT) ancestry group [
Genotypes were pre-phased and then imputed to a cosmopolitan reference panel consisting of all individuals from the 1000 Genomes Project [
Determination of genetic ancestry was performed by principal component analysis (PCA), as previously described [
Achilles tendon injuries were identified in the GERA cohort based on clinical diagnoses and surgical procedures captured in the KPNC Electronic Health Record system. Cases were defined as individuals with at least one International Classification of Disease, Ninth Revision (ICD9) or Common Procedure Terminology, Fourth Edition (CPT4) code: ICD726_71, ICD727_67, and/or CPT27650, describing Achilles bursitis or tendinitis, nontraumatic rupture of Achilles tendon, and primary repair of Achilles tendon rupture, respectively (
Description | Code | N (%) |
---|---|---|
Achilles bursitis or tendinitis | ICD726_71 | 4,949 (96) |
Non-traumatic rupture of Achilles tendon | ICD727_67 | 105 (2) |
Repair, primary, open or percutaneous, ruptured Achilles tendon | CPT27650 | 215 (4) |
Total number individuals | 5,148 (100) |
aNumber of individuals with that code in their electronic medical records, and percent of the total number of individuals in parentheses. Some individuals had more than one code.
For ACL rupture, KPNC patients with any potential ACL injury were identified by search of the electronic medical record for the ICD9 codes 717.83 and 844.2 and the CPT4 codes 29888 and 27407 at any time. The imaging studies and surgical reports from these patients were reviewed and those who had strong evidence for a full or partial ACL rupture on MRI and/or underwent ACL reconstruction were considered to have had an ACL rupture. The charts of the remaining patients were then manually reviewed by one of the study physicians (AA) with the assistance of the study orthopedist (JD). Those patients who had confirmatory evidence of ACL rupture (e.g., unambiguous history in a progress note) were also classified as having had an ACL rupture. Patients with ambiguous histories or ACL injury without rupture were not considered to have had a full or partial ACL rupture.
We tested for SNP associations with a logistic regression model using allele in an additive genetic model for each of the five ancestry groups separately, as in Roos et al., 2016 [
For Achilles tendinopathy or rupture, the principal components used were: AFR (6 PCs), EAS (6 PCs), EUR (10 PCs), LAT (6 PCs) and SAS (6 PCs). For ACL rupture, there were only 7 cases in the SAS and AFR populations, and so these populations were not analyzed further. For the remaining populations for ACL rupture, the principal components used were: EAS (2 PCs), EUR (4 PCs) and LAT (3 PCs). 10,551,193 SNPs were processed in the meta-analysis for Achilles tendinopathy or rupture and 8,303,052 SNPs were processed for ACL rupture meta-analysis.
To assess for inflation due to population stratification, the genomic control parameter (λ) was calculated [
The fixed-effects and random-effects p-values for all of the SNPs tested in this study are available from NIH GRASP:
We searched our meta-analysis results for previously-published genetic associations with either Achilles injury or ACL rupture [
Based on our knowledge of biological functions of the Achilles tendon and ACL, we generated a list of 90 candidate genes potentially associated with either Achilles or ACL injury. These genes code for structural components or the development of ligaments and/or tendons. A set of all known SNPs found within 2kb of the start and end of these genes was generated using SCAN (
Gene-based testing was performed using the VErsatile Gene-bASed test VEGAS2 [
This study analyzed stored data from RPGEH. The health and genotype data for the subjects were de-identified. All study procedures were approved by the Institutional Review Board of the Kaiser Foundation Research Institute.
We obtained access to injury information and genotype data from the Research Program on Genes, Environment and Health cohort (Materials and Methods). This program includes genotype and medical data from 102,979 patients in the Kaiser-Permanente system in Northern California. For Achilles tendon injury, we interrogated the electronic medical records of these individuals for those that had incurred tendinopathy or a rupture (
Cases (%) | Controls (%) | Total | |
Overall | 5,148 (5.0%) | 97,831 (95.0%) | 102,979 |
European | 4,258 (5.1%) | 79,006 (94.9%) | 83,264 |
Latin-American | 413 (4.8%) | 8,147 (95.2%) | 8,560 |
East Asian | 268 (3.6%) | 7,250 (96.4%) | 7,518 |
African | 192 (6.1%) | 2,969 (93.9%) | 3,161 |
South-East Asian | 17 (3.6%) | 459 (96.4%) | 476 |
Female | 2,934 (4.9%) | 56,737 (95.1%) | 59,671 |
Male | 2,211 (5.1%) | 41,031 (94.9%) | 43,242 |
Unknown | 3 (4.5%) | 63 (95.5%) | 66 |
Age |
62.3 (62.2–62.4) |
62.7 (62.6–62.8) | 62.6 (62.5–62.7) |
Cases (%) | Controls (%) | Total | |
Overall | 598 (.61%) | 98,744 (99.39%) | 99,342 |
European | 495 (.59%) | 82,769 (99.61%) | 83,264 |
Latin-American | 54 (.63%) | 8,506 (99.37%) | 8,560 |
East Asian | 49 (.65%) | 7,469 (99.35%) | 7,518 |
Female | 349 (.61%) | 57,257 (99.39%) | 57,606 |
Male | 249 (.59%) | 41,421 (99.41%) | 41,670 |
Unknown | 0 (0%) | 66 (100%) | 66 |
Age |
52.0 (50.8–53.1) |
62.7 (62.6–62.8) | 62.6 (62.5–62.7) |
a Mean age (years) at enrollment (95% CI).
bThere is a small difference in ages between cases and controls (p = 0.01).
c The difference in ages between the cases and controls is highly significant (p<10−100).
We used Linkage Disequilibrium Score Regression to analyze the GWAS data for: 1) heritability, 2) fraction of the genetic association signal derived from polygenic associations, and 3) overlap with GWAS data for other traits [
We plotted the p-value for every SNP on Manhattan plots (
Achilles tendon injury | ||||||
SNP | EA |
MAF |
P (FE) |
OR (FE)(95% CI) |
P (RE) | OR (RE) (95% CI) |
rs1937810 | C | 0.17 | 1.5x10-7 | 1.16 (1.11–1.21) | 1.5x10-7 | 1.16 (1.11–1.21) |
rs57104447 | C | 0.04 | 3.0x10-7 | 1.28 (1.19–1.37) | .01 | 1.25 (1.16–1.34) |
rs57224706 | G | 0.03 | 6.0x10-7 | 1.36 (1.24–1.48) | 6.0x10-7 | 1.36 (1.24–1.48) |
rs60713544 | C | 0.02 | 7.2x10-7 | 1.43 (1.29–1.57) | 7.2x10-7 | 1.43 (1.29–1.57) |
ACL rupture | ||||||
SNP | EA |
MAF |
P (FE) |
OR (FE) (95% CI) |
P (RE) | OR (RE) (95% CI) |
rs4067493 | G | 0.04 | 2.4x10-7 | 1.94(1.69–2.19) | 3.1x10-5 | 2.06(1.81–2.31) |
rs113435565 | C | 0.03 | 4.0x10-7 | 1.91 (1.66–2.16) | 4.0x10-7 | 1.91 (1.66–2.16) |
rs11960097 | G | 0.08 | 5.9x10-7 | 1.54 (1.37–1.71) | 5.9x10-7 | 1.54 (1.37–1.71) |
aEffect allele.
bMinor Allele Frequency in the control population.
cP-value from fixed effects (FE) or random effects (RE) meta-analysis.
dAllelic odds ratio (95% confidence interval) for the effect allele. FE, fixed effect; RE, random effect.
Previous studies have reported candidate SNPs that have shown an association with either Achilles tendon injury or ACL rupture using p<0.05 as a cutoff. Specifically, 19 SNPs in 12 candidate genes have been reported to show an association with Achilles tendinopathy (using p<0.05 as a cutoff)[
For Achilles tendinopathy or rupture, 14 of these SNPs were contained in our dataset whereas for ACL rupture, 6 SNPs were present. We attempted to replicate the previous results for these candidate SNPs using our dataset. Because we are testing only a small number of SNPs, the threshold for statistical significance can be much lower than the genome-wide threshold that we used for the genome-wide study above (p<5x10-8). It is still important, however, to adjust the p-value threshold to compensate for multiple testing. We used the Benjamini-Hochberg method to set the false discovery rate to q = 0.05. In addition, the association should show the same direction of effect (i.e. same risk allele) as in the previous publication.
For Achilles tendinopathy or rupture, none of the 14 SNPs were found to be significant using the Benjamini-Hochberg threshold (
rs1045485 |
CASP8 | C | 2.6E-03 | 1.10 (1.04–1.16) | [ |
rs4747096 | ADAMTS14 | G | 0.55 | 0.98 (0.93–1.04) | [ |
rs2761884 | BMP4 | T | 0.22 | 1.03 (0.98–1.07) | [ |
rs1134170 | COL5A1 | A | 0.45 | 0.97 (0.88–1.05) | [ |
rs12722 | COL5A1 | T | 0.62 | 0.97 (0.88–1.06) | [ |
rs3196378 | COL5A1 | A | 0.75 | 1.01 (0.93–1.10) | [ |
rs1559186 | COL5A3 | C | 0.75 | 1.01 (0.97–1.05) | [ |
rs331079 | FBN2 | C | 0.32 | 1.03 (0.93–1.10) | [ |
rs4919510 | MIR608 | G | 0.35 | 0.98 (0.92–1.03) | [ |
rs591058 | MMP3 | T | 0.34 | 1.06 (1.02–1.10) | [ |
rs679620 | MMP3 | T | 0.35 | 1.06 (1.02–1.10) | [ |
rs4789932 |
TIMP2 | G | 8.7E-03 | 1.08 (1.02–1.13) | [ |
rs1330363 | TNC | C | 0.26 | 1.02 (.98–1.07) | [ |
rs2104772 | TNC | A | 0.29 | 0.98 (.94–1.02) | [ |
rs1516797 | ACAN | T | 0.99 | 1.00 (0.77–1.43) | [ |
rs516115 | DCN | A | 0.29 | 1.07 (0.94–1.18) | [ |
rs970547 | COL12A1 | T | 0.11 | 1.11 (0.98–1.25) | [ |
rs2276109 | MMP12 | T | 0.44 | 1.07 (0.90-.1.24) | [ |
rs1800255 | COL3A1 | A | 0.17 | 1.09 (1.00–1.31) | [ |
rs331079 | FBN2 | G | 0.45 | 1.07 (.91–1.25) | [ |
a Effect Allele.
b P-value from this study.
c Reference showing original association of candidate SNP with Achilles tendinopathy.
dFor rs1045485 and rs4789932, the direction of the effect was opposite to the published results; i.e. the minor C allele of rs1045485 and the G allele of rs4789932 are associated with increased risk in this work but decreased risk in the prior work [
In our gene association study, we used sex, age and ancestry group as covariates whereas previous studies did not use these covariates. To control for this difference, we re-tested the candidate SNPs without using covariates in order to align our analysis with those that had been done previously (
For ACL rupture, when we repeated the analysis without using covariates, we found that rs1800255 in
To demonstrate how these genome-wide data can be applied, we used the data to evaluate a new set of candidate genes for association with Achilles tendinopathy or rupture and with ACL rupture. We created a list of 90 genes encoding proteins involved in the formation of ligaments or tendons, including the set of candidate genes from previous publications (
Besides evaluating each SNP individually, SNPs were analyzed as part of genes using gene-based analysis [
For ACL rupture,
Achilles tendon and ACL injury are common in recreational and elite athletes, and even in non-athletes [
Here, we performed a study to find genetic variants associated with Achilles tendon or ACL injury by obtaining access to large-scale genotype and phenotype data from the Research Program on Genes, Environment and Health. The data contains information from 102,979 individuals of whom 5,418 had Achilles tendinopathy or rupture (from all five ancestry groups) and 598 had an ACL rupture (from the EAS, EUR and LAT ancestry groups). To date, this is the largest gene association study for either Achilles tendon or ACL injury reported in terms of number of SNPs that were genotyped and number of injury cases.
We were unable to find any SNPs associated with either injury type at a genome-wide significance level. For a hypothetical common SNP with a minor allele frequency of 5% and a genotype relative risk of 1.3, power calculations indicate that this SNP would have about a 78% chance of being detected in our Achilles study and a 59% chance for detection in our ACL study. Thus, it is unlikely that there are many common SNPs with a medium-to-strong association with these injuries (i.e. genotype relative risk > 1.3).
We found four independent polymorphisms, however, associated with Achilles tendinopathy or rupture with a borderline significant p-value between 1x10-6 and 5x10-8 (
Previous candidate gene studies have reported some SNPs that show a weak association with Achilles tendon or ACL injury [
Although we re-tested candidate SNPs that were previously reported to show an association with either Achilles or ACL injuries, we did not find strong evidence for replication of the candidate gene results. We performed the replication analysis using age, sex and ancestry group as covariates (as was done in the genome-wide analysis) as well as without these covariates (as was done in the candidate gene studies). For Achilles tendinopathy or rupture, we found moderate evidence for replication of one of the 14 tested SNPs; specifically, rs4919510 in
For ACL rupture, none of the other SNPs showed evidence of replication either with or without using covariates in the analysis. Previous work has shown that the association of rs12722 in
One possible explanation for the failure to replicate previous candidate gene studies is that our cohort consists of patients in the Kaiser-Permanente medical system in Northern California regardless of activity level, whereas previous studies evaluated cohorts of competitive athletes. Varying levels of physical activity may affect risk for sustaining an Achilles tendon injury or ACL rupture.
For Achilles tendon injury, another possible explanation for the failure to replicate previous results is that cases were identified based on diagnosis and procedure codes in the electronic medical record. Patients in large, administrative data sets may have been misdiagnosed, introducing information bias. Non-differential misclassification of Achilles tendon injury would tend to reduce the strength of associations. A second possible explanation is that we evaluated patients with Achilles tendinopathy, bursitis or rupture as a single injury group. Achilles tendon rupture may represent an increased injury severity versus tendinopathy with a correspondingly stronger genetic effect, or conversely may be associated with acute trauma with limited genetic effect. A third possibility is that the genetic risk factors for Achilles bursitis may be different from those for intrinsic Achilles tendon pathology. Identifying cases of ACL rupture was more definitive, as cases had either shown an ACL rupture by MRI or had undergone an ACL reconstruction procedure.
We performed a candidate-gene study for Achilles tendon injury and ACL rupture using 2855 SNPs in 90 candidate genes. In the first analysis, SNPs were tested individually but none showed a significant association with either type of injury. In the second analysis, gene-based analysis was used to aggregate SNPs into genes and then test each of the genes for association with these injuries. None of the genes showed an association with Achilles tendon injury but
The summary statistics from the GWAS for Achilles tendon and ACL injury are available to the public at NIH GRASP:
Tested SNPs within
(TIF)
(DOCX)
A. Achilles tendon injury. B. ACL rupture.
(XLS)
(DOCX)
(DOCX)
(XLSX)
(XLS)
(DOCX)
The authors thank the Kaiser Permanente Northern California RPGEH team for access to data and assistance in data management. Data used in this study were provided by the Kaiser Permanente Research Program on Genes, Environment, and Health (RPGEH): Genetic Epidemiology Research on Adult Health and Aging (GERA), funded by the National Institutes of Health [RC2 AG036607 (Schaefer and Risch)], the Robert Wood Johnson Foundation, the Wayne and Gladys Valley Foundation, The Ellison Medical Foundation, and the Kaiser Permanente Community Benefits Program. Access to RPGEH data used in this study may be obtained by application to the Kaiser Permanente Research Bank (KPRB) via