CKM Glu83Gly Is Associated With Blunted Creatine Kinase Variation, but Not With MyalgiaCLINICAL PERSPECTIVE
Background—To test the association of a recently reported variant in the creatine kinase (CK) muscle gene, CKM Glu83Gly (rs11559024) with constitutive creatine phosphokinase (CK) levels, CK variation, and inducibility. Given the diagnostic importance of CK in determining muscle damage, we tested the association of the variant with myalgia.
Methods and Results—Meta-analysis between longitudinal cohort GoDARTS (Genetics of Diabetes Audit and Research, Tayside Scotland), minor allele frequency (=0.02), and randomized clinical trial (JUPITER [Justification for the Use of Statins in Primary Prevention: An Intervention Trial Evaluating Rosuvastatin], minor allele frequency=0.018) was used to replicate the association with baseline CK measures. GoDARTS was used to study the relationship with CK variability. Myalgia was studied in JUPITER trial participants. Baseline and SDs of CK were on average 18% (P value=6×10−63) and 24% (P value=2×10−5) lower for carriers of the variant, respectively. The variant was not associated with myalgia (odds ratio, 0.84; 95% confidence interval, 0.52–1.38).
Conclusions—This study highlights that a genetic factor known to be associated with constitutive CK levels is also associated with CK variability and inducibility. This is discussed in the context of evidence to suggest that the variant has an impact on inducibility of CK by trauma through a previously reported case of a homozygous carrier. However, the lack of association between the variant and myalgia suggests that it cannot reliably be used as a biomarker for muscle symptoms.
Creatine phosphokinase also known as creatine kinase (CK) is an enzyme, 381 amino acids in length.1 It catalyzes the reversible reaction that uses creatine to produce phosphocreatine and ADP by the dephosphorylation of ATP. This is an exergonic reaction and is important in the maintenance of energy homeostasis in all muscle tissues. CK is a dimer composed of 2 subunits, CK-M (muscle) and CK-B (brain) that combine to form 3 isoforms. The CK-MM isoform is found predominantly in skeletal muscle, CK-MB in the brain and cardiac muscle, and the CK-BB in smooth muscle. CK-MM makes up the majority of measured serum CK.2–4 CKM encodes the CK-M subunit.
See Editorial by de Denus et al
The enzyme creatine phosphokinase (CK) is a routine biochemical test performed in the clinical setting with widespread applicability. CK-M is used as a marker for tissue damage, muscle breakdown, muscular dystrophy,5,6 infection,7 acute kidney failure,8 myocardial infarctions,9–12 rheumatoid arthritis,13,14 and some diseases of the liver.15–17 Notably, it is used as a marker of muscle damage or myopathy in adverse reactions to statins.8,18,19 In addition, CK-M has documented associations with hypothyroidism20 and blood pressure.21,22
In 2014, Dubé et al23 performed an original genome-wide association study in a Canadian population of statin users. They report a single nucleotide polymorphism rs11559024 (Glu83Gly) in CKM as being associated with serum CK levels. They found that heterozygous carriers of the rare allele (Glu83Gly) had a mean CK level of 68.13 U/L (SD=35.57 U/L) compared with 119.32 (SD=84.74) for homozygous carriers of the common allele (Glu83Glu). They also report that statin dose and duration of use did not impact this association.23 The minor allele frequency (MAF) in the Canadian study was 0.010.23 Kristjansson et al4 replicated these findings in a genome-wide association study performed on 63 159 Icelanders with CK measurements. They report the main effect of the CKM Glu83Gly variant as being associated with serum CK (β=−0.446, P value=1.8×10−115). They also report that the variant was not associated with statin-related side effects in a subcohort of ≈8900 statin users. The MAF in the Icelandic cohort was 0.0215.4
Given the robust association and biologically apparent role of the gene in the production of CK, we wanted to gain insight into the relationship between the variant and CK response. Because CK elevations in response to appropriate stressors is the clinically significant feature of the biomarker, we sought to understand the impact of the variant on inducibility of CK. The analytic cohort used was the GoDARTS study (Genetics of Diabetes Audit and Research, Tayside Scotland). GoDARTS is a rich source of data, combining complete electronic medical records, including prescription information, all laboratory results from clinical visits from a cohort of 18 190 individuals in Tayside, Scotland. Genetic information is available for a proportion of these individuals. Because of its longitudinal nature, the GoDARTS database has a median of 9 CK tests per individual. Therefore, it is the ideal template with which to examine intraindividual variability of CK by genotype. Further, because raised levels are used as markers of muscle-based symptoms, we sought to understand the relationship between the variant and the development of myalgia. If the variant confers no protective effect, this might provide a novel mechanistic rationale for a subpopulation of individuals presenting with statin intolerance or myalgia without raised CK levels.18
The findings of this analysis could impact the viability of CK as a reliable biomarker. We provide evidence to suggest that CK test results should be analyzed in the context genotypic data, especially for statin intolerance.
Materials and Methods
Tayside Medical Ethics Committee approved the GoDARTS study, and informed consent was obtained for all participants. The data set contains complete electronic medical records, prescription information, and laboratory results from 18 306 Scottish white individuals. In GoDARTS, genotype data for the CKM Glu83Gly variant were available for 6271 individuals, of whom 4578 were genotyped using the Human Exome-12 VI_A_chip and 1693 using TAQMAN. The MAF of CKM Glu83Gly in the GoDARTS cohort was 0.02. The JUPITER trial (Justification for the Use of Statins in Primary Prevention: An Intervention Trial Evaluating Rosuvastatin) protocol was approved by the local institutional review board at each participating center. Genotyping for 8749 JUPITER trial participants was performed on the Omni1-Quad platform (Illumina, San Diego). The CKM Glu83Gly variant, rs11559024, was directly typed, and MAF was 0.018.24 The variant was in Hardy–Weinberg equilibrium in both cohorts.
CK measures were gathered from the GoDARTS population where available. Test results from wards, such as Accidents and Emergency, Cardiac Care, stroke, surgical wards, and high dependency units, were excluded for the analysis of baseline or uninduced CK levels with the variants. Individuals who had a history of thyroid disease, and those who had suffered a myocardial infarction, kidney disorder, or had been involved in a serious accident in the 6 months preceding the test date were excluded from the analysis of baseline or uninduced CK with the variant.
For the analysis of CK variability, all measures were considered irrespective of referring center, and intraindividual SDs were computed for individuals with ≥3 CK tests. The upper limit of the normal range for women over 19 years of age is 120 IU/L and 180 IU/L for men of the same age as specified by the metadata documents provided by the Health Informatics Centre (Farr Institute). All clinical and biochemical testing in the GoDARTS cohort is performed at centralized laboratories in Ninewells Hospital, Dundee, United Kingdom. The CK test results used for replication in the JUPITER trial were tested at baseline, when the population was treatment naive. The end products measured by the CK enzyme assay are provided in Figure I in the Data Supplement.
Myalgia in JUPITER
The analysis in JUPITER focused on 4381 study participants randomized to receive statin treatment and 4368 who received the placebo.24,25 Because the trial focused on the role of low-grade underlying inflammation (evidenced by high C-reactive protein levels), patients with inflammatory conditions, such as severe rheumatoid arthritis, lupus, or inflammatory bowel disease, were excluded, as were patients using immunosuppressant agents.25 Myalgia was ascertained by physicians blinded to trial allocation arm. Only treatment-emergent adverse events were reported.26 Myalgia was observed in 837 trial participants and was reported by Hsia et al26 to be independent of their assigned therapy.
All statistical analyses for GoDARTS were performed in SAS 9.3 (SAS Institute, Cary, NC). Plots were generated in R studio.27 Statistical analyses in the JUPITER trial were performed using R.28 Fixed-effects meta-analyses were performed using the metafor package in the R studio environment.27–29 The Cochrane Q test was used to determine heterogeneity. Logarithmic transformations were applied to all CK levels to normalize their distribution. The first recorded CK measurement taken in an ambulatory setting for each individual was used to study the association between the variant and CK levels using linear regression. Intraindividual SDs were calculated for those with ≥3 CK measures. The association between intraindividual SD and genotypes was tested using linear regression. The β, SE, and R2 are reported. Binary logistic regression models and a survival analyses were used to test associations of the variant with myalgia in the JUPITER trial using R.28
Baseline Characteristics of the GoDARTS Study Population
The average age was 66 years at the time of recruitment into the study (range 27–93); CK measures were available retrospectively and prospectively from recruitment date. Females comprised 40% of the study population. Certain factors that might be associated with CK levels, such as age, sex, body mass index, type 2 diabetes mellitus status, and creatinine levels, were tested for the association with the CKM Glu83Gly genotype, and none were found to be significant (Table 1). Effect estimates of the genotype in analyses were therefore unaffected by the inclusion of these variables. Baseline features of the JUPITER population are described elsewhere.24–26
Association With Baseline CK: GoDARTS and JUPITER
CK levels were statistically significantly associated with the CKM Glu83Gly variant (n=4598, P value=2×10−16) in the GoDARTS population. Carriers of the CKM variant (Glu83Gly:T/C) had mean CK of 86 (±68) compared with 126 (±82) for homozygotes of the common allele as seen in Table 2. A boxplot of the difference in CK levels by genotype is presented in Figure II in the Data Supplement.
The association with the CKM Glu83Gly variant was replicated in the JUPITER trial (n=8745, P value<2×10−45). A meta-analysis with the GoDARTS cohort showed a highly robust association (β=−0.18, P value=1×10−63), and the test of heterogeneity was nonsignificant (P value=1). A forest plot of the association is presented in Figure 1.
From the meta-analyses of the GoDARTS and JUPITER populations, we conclude that carriers have 18% lower (log 10 transformed) CK on average, which translates to a 1.5 IU/L lower baseline CK level.
Association With CK Variability
To assess if CK levels respond to stress differently by Glu83Gly genotype in CKM, we undertook an analysis of the intraindividual variation (represented by SDs) for individuals with ≥3 CK measures in the GoDARTS population (n=3246). The SD of an individual’s CK test results were stratified by genotype as seen in Figure 2. We observe that Glu83Gly variant exerts a strong effect on the variability of CK measures in an individual (β=−0.24%, P value=2×10−5). Carriers of the variant have a reduced range of variability in their measures. There is on average 24% reduction in the (log 10 transformed) SD for carriers, indicating that their CK levels do not always respond in a manner similar to noncarriers.
Association With Myalgia in JUPITER
Given the use of CK as a marker of muscle damage, especially for statin users, we examined if this variant that is associated with blunted CK response was associated with the development of myalgia in a clinical trial. We observed no association between the variant and the development of myalgia in the JUPITER population (odds ratio, 0.84; 95% confidence interval, 0.52–1.38; P value=0.5; Kaplan–Meier plot presented in Figure 3). This analysis was conducted using the genotype as a dominant trait because there were only 3 homozygous carriers of the variant. The analysis had 90% power to detect an effect of the size of the point estimate.30 There was no association in analyses stratified by treatment allocation arms of the trial: rosuvastatin arm (odds ratio, 0.64; 95% confidence interval, 0.31–1.29) or placebo arm (odds ratio, 1.12; 95% confidence interval, 0.57–2.17).
We report the replication of the association between the CKM Glu83Gly variant and constitutive CK levels using data obtained in a prospective cohort study (GoDARTS) and in the treatment-naive population of a clinical trial (JUPITER). Further, we show that the variant is also associated with variability of CK levels. The use of SDs to calculate the variability gives us a conservative estimate of the effect, as contrasting the highest and lowest CK for each individual could also be used to assess the impact of the variant. The conclusions about the impact of the Glu83Gly variant in CKM on CK inducibility is further strengthened by the single individual in the GoDARTS cohort who was a homozygous carrier of the variant (Gly83Gly:C/C), presented as a case report by Wallace et al.31 Details of the medical history are presented in Figure III in the Data Supplement. During hospitalization for necrotizing fasciitis (event 1), a condition during which there is aggressive infection of the tissue and where CK levels could rise to >600 IU/L,32 the patient’s CK levels did not exceed 10 IU/L. In response to the subsequent development of gangrene (event 2), the patient underwent a debridement procedure; postoperatively, CK levels were at a maximum of 28 IU/L. Later, the patient underwent a hemicolectomy for bowel cancer (event 3). The patient’s pre- and postoperative CK levels remained relatively unchanged and, in fact, seemed lower (34 and 25 IU/L, respectively). Notably, the patient was a statin user who had undergone over 4 switches in statin therapy between 3 types of statins, namely pravastatin, atorvastatin, and simvastatin. This switching is attributed to complaints of intolerance; however, her CK measurements had been deemed normal by clinicians looking for evidence of statin-induced myopathy.31
In the case of the homozygous carrier of the variant, we see the lack of CK response in conditions that would normally cause extremely high CK levels, such as severe tissue infection and surgical trauma.31 On the basis of the CK assay that measures enzyme activity, we conclude that carriers of this CKM Glu83Gly variant are less likely to produce large quantities of measurable and, therefore, functioning CK in response to tissue damage.
Because the variant inhibits measurable CK levels from rising, potentially obfuscating the correct diagnosis, it might be essential to factor in the genotype of the individual before determining the validity or normalcy of the result. This is especially crucial because the genotype shows no association with myalgia, an outcome most commonly associated with elevated CK levels.
The diminished quantity of functional CK in the serum (Figure I in the Data Supplement) would have to be compensated for by other mechanisms in the body to maintain energy homeostasis. One potential hypothesis revolves around the switch between aerobic and anaerobic metabolisms. An oxygen debt created when there is insufficient phosphocreatine to make enough ATP needed during periods of high exertion. The process of energy generation then shifts from muscles to the liver (where ATP is generated anaerobically via glycolysis and the Cori cycle, which delays the oxygen consumption process). Therefore, those with low CK activity (CKM carriers) might make this switch sooner, to maintain homeostasis.
The MAF for the CKM Glu83Gly variant in the GoDARTS population was 0.02; in JUPITER, it was 0.018, similar to the European population in the 1000 Genome project (0.022). The Canadian (0.01) and the Icelandic (0.0215) cohorts also had a similar MAFs. This indicates that the populations under study were not experiencing participant selection bias. The frequency of the CKM variant in the Kenyan Masai population is 0.223.33 This significant difference in the allele frequencies in the Masai is striking and warrants further investigation in other populations and for association with features of muscle or athletic performance. Undertaking whole-exome sequencing of homozygous carriers would enable us to determine if there are other rare variants in CKM that might be causative.
Statins are the most popular class of lipid-lowering medications. Intolerance to statins manifests as muscle symptoms, including myalgia, and is observed in 10% of statin users. Intolerance results in poor compliance to medication, which in turn increases the risk of adverse cardiovascular events, such as myocardial infarctions and strokes.34–36 Definitions of statin intolerance in population-based studies hinge on the elevation of CK and, in some instances, on the resolution of muscle-associated CK elevations on the discontinuation of statin therapy.19,37,38 The most established threshold for a clinically relevant CK elevation is ≥4× the upper limit of normal (4 × upper limit of the normal) and 10 × upper limit of the normal.37 However, as this study suggests, these outcomes are less likely to occur among carriers of the CKM variant. Indeed, we note the underrepresentation of variant carriers in 231 clinically adjudicated cases of statin-induced myopathy of white ancestry in the PREDICTION-ADR study,37 who met the CK elevation criteria and had been exome sequenced. However, because of the size of the study and the MAF of the variant, we are underpowered to determine if the difference is significant. We posit that the employment of traditional classifications would lead to the artificial enrichment of noncarriers of the CKM variant being classified as having statin-induced myalgia or myopathy. Indeed, this artifact is noted in the occurrence of CK above the upper limit of the normal in GoDARTS (Table I in the Data Supplement).
As highlighted by Stroes et al,18 in the European Atherosclerosis Society consensus panel statement on the impact of statin-associated muscle symptoms on statin therapy, often cases of statin-associated muscle symptoms are not accompanied by marked CK elevations. Phillips et al39 have reported cases of statin-associated myopathy, that were resolved on discontinuation and return on rechallenge, that showed no elevations in CK. The journal for American Family Physician also highlights the potential for this to occur, concluding that pathological evidence of myopathy might be present among patients with adverse muscle symptoms with normal CK levels.40 This is congruent with the lack of association between this CK-associated variant and nonspecific myalgia in this study.
It is of great clinical and scientific interest to understand the compensatory mechanism that maintains energy homeostasis for carriers of the variant. This might be accomplished by metabolomic studies in populations where the variant is rare (whites) with those in which it is more frequent, such as the Kenyan Masai population (MAF=0.22).33 It would also be important to see if there is a commensurate rise in CK-B levels for carriers of the CKM variant, and if the variants encoding them are related.
Because the variant confers no protective effect against the development of myalgia, this might provide a novel mechanistic rationale for a subpopulation of individuals presenting with statin intolerance or myalgia without raised CK levels.18 Given the MAF of 0.02 of the Glu83Gly variant in CKM, ≈4 in every 100 individuals is likely to be a carrier. With the rising number of statin users worldwide, CK monitoring for those complaining of adverse reactions is also likely to increase. The CKM Glu83Gly variant is therefore a key tool to be used while determining the normalcy of CK test results.
We are grateful to all the participants in the GoDARTS study (Genetics of Diabetes Audit and Research, Tayside Scotland), the general practitioners, the Scottish School of Primary Care for their help in recruiting the participants, and to the whole team, which includes interviewers, computer and laboratory technicians, clerical workers, research scientists, volunteers, managers, receptionists, and nurses. The study complies with the Declaration of Helsinki. We acknowledge the support of the Health Informatics Centre, University of Dundee, for managing and supplying the anonymized data and National Health Service Tayside, the original data owner.
Sources of Funding
This work was funded by the European Community’s Seventh Framework Programme (FP7/2007–2013) Under Grant Agreement no. 602108 through the PREDICTION-ADR (Adverse Drug Reactions) project. GoDARTS (Genetics of Diabetes Audit and Research, Tayside Scotland) was funded by The Wellcome Trust (072960/Z/03/Z). Genotyping was funded as part of WTCCC2 (084726/Z/08/Z, 084727/Z/08/Z, 085475/Z/08/Z, and 085475/B/08/Z), UK Medical Research Council (Award G0601261) and as part of the EU IMI-SUMMIT program. Genetic analysis in the JUPITER trial (Justification for the Use of Statins in Primary Prevention: An Intervention Trial Evaluating Rosuvastatin) was supported by a research grant jointly to Drs Chasman and Ridker.
The Data Supplement is available at http://circgenetics.ahajournals.org/lookup/suppl/doi:10.1161/CIRCGENETICS.117.001737/-/DC1.
- Received February 16, 2017.
- Accepted June 23, 2017.
- © 2017 American Heart Association, Inc.
- 1.↵National Center of Biotechnology Information. Creatine Kinase [Homo sapiens] [Internet]. http://www.ncbi.nlm.nih.gov/protein/AAA96609.1. Accessed January 28, 2016.
- Wyss M,
- Kaddurah-Daouk R
- Cabaniss C.
- Kristjansson RP,
- Oddsson A,
- Helgason H,
- Sveinbjornsson G,
- Arnadottir GA,
- Jensson BO,
- et al
- Moat SJ,
- Korpimäki T,
- Furu P,
- Hakala H,
- Polari H,
- Meriö L,
- et al
- D’Amico A,
- Catteruccia M,
- Baranello G,
- Politano L,
- Govoni A,
- Previtali SC,
- et al
- Crum-Cianflone NF
- Savonitto S,
- Granger CB,
- Ardissino D,
- Gardner L,
- Cavallini C,
- Galvani M,
- et al
- Alpert JS,
- Thygesen K,
- Antman E,
- Bassand JP
- Apple FS,
- Quist HE,
- Doyle PJ,
- Otto AP,
- Murakami MM
- Stroes ES,
- Thompson PD,
- Corsini A,
- Vladutiu GD,
- Raal FJ,
- Ray KK,
- et al
- Brewster LM,
- Mairuhu G,
- Bindraban NR,
- Koopmans RP,
- Clark JF,
- van Montfrans GA
- Dubé MP,
- Zetler R,
- Barhdadi A,
- Brown AM,
- Mongrain I,
- Normand V,
- et al
- Chasman DI,
- Giulianini F,
- MacFadyen J,
- Barratt BJ,
- Nyberg F,
- Ridker PM
- Hsia J,
- MacFadyen JG,
- Monyak J,
- Ridker PM
- 27.↵RStudio Team. RStudio: Integrated Development Environment for R. 2015. http://www.rstudio.com/. Accessed June 7, 2016.
- 28.↵R Development Core Team. R: A Language and Environment for Statistical Computing. 2008. http://www.r-project.org. Accessed June 7, 2016.
- Rannanheimo PK,
- Tiittanen P,
- Hartikainen J,
- Helin-Salmivaara A,
- Huupponen R,
- Vahtera J,
- et al
The findings of this research suggest that a single variant in the CKM gene (Glu83Gly) is associated with lower levels of the enzyme creatine kinase, as well as lower variation in enzyme levels. The findings point to a genetic subpopulation of individuals who will show a blunted response to stressors that usually result in increased enzyme levels, with a potential for clinical misclassification. The impact of altered creatine kinase response will be observed especially while attempting to identify muscle-based adverse reactions to statins, which are the most widely used medication for cholesterol control. At present, creatine kinase is the principal biomarker for statin-association muscle symptoms. We suggest the use of genotype data to determine the normalcy of creatine kinase test results.