We aimed to investigate a causal effect of functional sarcopenia status, including poor handgrip strength and slow walking pace, on cardiovascular diseases. This study was an observational cohort study including observational analysis and Mendelian randomization. We studied the UK Biobank prospective cohort (n = 324,486) for observational analysis with poor handgrip strength or self-reported slow walking pace as the exposures, investigating risk of myocardial infarction or mortality. Genetic instruments for the exposures were developed in 337,138 individuals of white British ancestries, and coronary artery disease outcome (60,801 cases/123,504 controls) from the independent CARDIoGRAMplustC4D cohort was studied by two-sample Mendelian randomization. The findings were replicated by one-sample analysis by polygenic risk score analysis within the UK Biobank. Both slow walking pace and poor handgrip strength were significantly associated with higher risks of incident myocardial infarction and mortality, particularly from cardiovascular deaths, in the observational investigation. Genetically predicted poor handgrip strength (odds ratio: 1.128 [1.041 to 1.222]) and slow walking pace (odds ratio: 1.171 [1.022 to 1.342]) showed causal effects on the coronary artery disease risks in the independent cohort. The results were again identified by the one-sample Mendelian randomization, as the higher polygenic risk score for poor handgrip strength and slow walking pace was associated with a higher risk of mortality. In conclusion, this study supports the causal effects of slow walking pace and poor handgrip strength on the risks of cardiovascular disease and mortality. The functional sarcopenia status are targetable causative factors for interventions aiming to reduce risks of cardiovascular disease or mortality.
Sarcopenia, a reduced skeletal muscle mass state, has been associated with various comorbidities and even a higher risk of death. Gait speed and handgrip strength are well-known functional markers of sarcopenia that can be relatively easily screened in clinical visits. , Previous studies have demonstrated that poor handgrip strength or slow walking pace is a risk factor for coronary artery disease or related to poor survival, particularly for deaths from cardiovascular diseases. , , , However, the causal effect of slow walking pace or poor handgrip strength on cardiovascular disease or mortality needs further investigation. , Mendelian randomization is a method that can reveal the causal effect of a modifiable exposure, reflected by a set of instrumental genetic variants, on a complex disease phenotype. , In this study, we aimed to demonstrate the association between slow walking pace and poor handgrip strength on survival and risks of coronary artery disease. We hypothesized that both observational and genetically predicted slow walking pace or poor handgrip strength would be associated with a higher risk of coronary artery disease or mortality, suggesting causal effects of functional sarcopenia on risk of adverse cardiac outcomes.
The study was performed in accordance with the Declaration of Helsinki. The study was approved by the institutional review boards of Seoul National University Hospital (No. E-2005-008-1120) and by the UK Biobank consortium (application No. 53799). The requirement for informed consent was waived as we investigated the anonymous public database.
The UK Biobank is a prospective population-based cohort study that recruited >500,000 participants aged 40 to 69 years from 2006 to 2010 from various regions of the United Kingdom. The project included various anthropometric evaluations, surveys by touch-screen questionnaire, and laboratory tests from collected blood and urine. The details of the database have been previously described.
For observational analyses, among the available 502,505 UK Biobank participants, we included those with both 2 handgrip strength measurements (left and right) and self-reported walking pace ( Figure 1 ). To assess risk of incident outcomes, those with prior medical history of myocardial infarction (MI), or self-reported history of heart attack, angina, or stroke, were excluded. Those with missing values in the assessed covariates included in the regression models were additionally excluded to perform complete-case analysis. For genetic analyses, we used the provided data from the UK Biobank to apply sample filters, and we included individuals with white British ancestry who were not outliers for heterozygosity or missing rate and were without sex chromosome aneuploidy or excess kinship. The information for the sample filtering was available in the UK Biobank database.
Handgrip strength was measured isometrically using a calibrated Jamar J00105 hydraulic hand dynamometer (Lafayette Instrument Company, Lafayette, Indiana) at the baseline visit and recorded in the UK Biobank. One measurement per hand was performed, and we implemented the average handgrip strength values to determine poor handgrip strength. We followed the poor handgrip strength definition from the Foundation for the National Institutes of Health, and the cutoff values were <26 kg for men and <16 kg for women. Walking pace was self-reported via a touchscreen-based questionnaire by answering the question “How would you describe your usual walking pace?”; the possible answer choices were slow pace, steady average pace, brisk pace, none of the above, and prefer not to answer.
We tested whether walking pace or handgrip strength was significantly associated with an objective parameter of sarcopenia, that is, appendicular lean mass adjusted for body mass index (appendicular lean mass/body mass index). Appendicular lean mass was measured by a BC418MA body composition analyzer (Tanita, Tokyo, Japan). The first outcome in the observational analysis was MI, and the outcome was algorithmically defined by the UK Biobank. The second outcome was mortality, and we assessed all-cause mortality and also deaths from cardiovascular disease or cancer deaths. The details regarding the ascertainment of the outcomes and covariates or statistical analysis methods for observational investigation are described in Supplementary Methods.
To extend the causal inference of the observational findings, we performed a Mendelian randomization analysis developing genetic instruments for poor handgrip strength and slow walking pace, respectively. A genome-wide association study (GWAS) was performed with the imputed genotype data from the UK Biobank, and the details regarding the genotyping process have been previously published. We filtered out variants with minor allele frequency <0.1%. A p value <5 × 10 −8 was implemented to identify significant single nucleotide polymorphisms (SNPs) associated with the phenotype, and we adjusted for age, sex, and the first 20 principal components. After removing SNPs in linkage disequilibrium ( R 2 < 0.1), a genetic instrument was determined to summarize the genetic disposition of poor handgrip strength or slow walking pace. For the two-sample Mendelian randomization analysis, which has strength in a conservative sense as the outcome is assessed from an independent cohort where genetic instruments are developed, we implemented summary statistics from the independent CARDIoGRAMplusC4D cohort including European ancestry individuals (URL: www.cardiogramplusc4d.org ). The GWAS summary statistics included MI (43,676 cases/128,199 controls) and coronary artery disease (60,801 cases/123,504 controls) outcomes, and the details are described in the cited study. Only the overlapping SNPs from the genetic instrument were included, and the palindromic SNPs with intermediate allele frequencies were disregarded. The main two-sample Mendelian randomization method was the conventional fixed effects inverse variance weighted method. Along with the main analysis by the fixed-effect inverse variance method, an analysis by the maximal likelihood method was performed. The weighted-median method, which provides consistent estimates even when invalid instruments are present, was implemented as a sensitivity analysis. Additional sensitivity analysis was performed by the MR-Egger method with bootstrapping, which results in causal estimates robust to pleiotropy. However, the limitation of the method is low power, particularly when the number of included SNPs in a genetic instrument is low. The MR-Egger regression test for directional pleiotropy was additionally performed to calculate the formal statistics of the presence of directional pleiotropy.
Next, we implemented allele-score-based one-sample Mendelian randomization within the UK Biobank cohort using the individual data to assess the causal associations between the exposures and risk of mortality. We used the genetic instrument to calculate a polygenic risk score (PRS) with the regression effect sizes, betas of the GWAS results, and the gene dosage matrix. Associations between PRS, reflecting 10% odds per unit for phenotypic exposure, and survival were tested by Cox regression analysis according to the survival outcome. For survival outcomes, as inborn genetic information was the exposure, the survival period from birth date was calculated. We used the Ensembl Variant Effect Predictor to search genetic information of implemented SNPs and identify potential horizontal pleiotropy by searching previous reports on implemented SNPs. The multivariable model was adjusted for age, sex, genotype measurement batch, the first 10 principal components of the genetic information, and identified phenotypes related to potential horizontal pleiotropy, including diabetes, body mass index, and waist circumference. All of the above processes were performed with PLINK 2.0 (version alpha 2.3), the R package TwoSampleMR, and R. Two-sided p value <0.05 was used to indicate statistical significance.
In this study, 3,242,486 subjects were included for observational analysis ( Figure 1 ). For genetic analysis, we included 337,138 subjects with white British ancestry passing the sample filter. Those with poor handgrip strength mostly had a slow walking pace ( Table 1 ). Those with a slow walking pace had a higher median age and a larger proportion of obese subjects than the other groups. Current smoking was more common in those with a slow walking pace, and the median frequency of moderate physical activity was also lower than the others. Comorbidities, such as hypertension and diabetes mellitus, were more common in those with slow walking pace than those who reported average or brisk walking pace. Both average handgrip strength and graded walking pace showed a strong correlation with the objective appendicular lean muscle mass adjusted for body mass index (Supplementary Figure 1), and the p value for Spearman’s correlation test was <2.2E −16 . In addition, the adjusted appendicular lean mass showed moderate predictability for slow walking pace (area under curve = 0.624) and poor handgrip strength (area under curve = 0.666).
| Variable | Walking Pace | ||
|---|---|---|---|
| Slow (n = 20,091) | Average (n = 169,977) | Brisk (n = 134,418) | |
| Poor handgrip strength (average<26 for male or <16 kg for female) | 5,666 (23%) | 14,488 (8%) | 6,373 (5%) |
| Age (y) | 60 [53;65] | 59 [51;64] | 56 [49;62] |
| Sex | |||
| Female | 11,777 (59%) | 93,249 (55%) | 72,104 (54%) |
| Male | 8,314 (41%) | 76,728 (45%) | 62,314 (46%) |
| BMI (kg/m 2 ) | 30.2 [26.5;34.7] | 27.3 [24.7;30.4] | 25.4 [23.2;27.9] |
| BMI≥30 kg/m 2 | 10,277 (51%) | 46,845 (28%) | 16,745 (12%) |
| Waist circumference (cm) | 98 [88;108] | 91 [82;100] | 86 [77;94] |
| Waist circumference >102 cm for male, >88 cm for female | 12,345 (61%) | 64,366 (38%) | 27,093 (20%) |
| Frequency of moderate physical activity (d/wk) | 2 [0;5] | 3 [2;5] | 4 [2;6] |
| Smoker | |||
| Nonsmoker | 9,340 (46%) | 92,978 (55%) | 78,509 (58%) |
| Ex- | 7,494 (37%) | 59,155 (35%) | 44,374 (33%) |
| Current | 3,257 (16%) | 17,844 (11%) | 11,535 (9%) |
| Hypertension | 6,848 (34%) | 33,869 (20%) | 16,522 (12%) |
| Systolic BP (mm Hg) | 138.5 [126.5;151.5] | 137.5 [125.5;150.5] | 134 [123;147] |
| Diastolic BP (mm Hg) | 83 [76.5;90] | 83 [76;89.5] | 81 [74.5;88] |
| Diabetes mellitus | 2,424 (12%) | 8,167 (5%) | 3,173 (2%) |
| Hemoglobin A1c (mmol/L) | 36.6 [33.8;40.1] | 35.3 [32.9;38.0] | 34.5 [32.2;36.9] |
| Hemoglobin A1c (mg/dl) | 5.5 [5.2;5.8] | 5.4 [5.2;5.6] | 5.3 [5.1;5.5] |
| Dyslipidemia | 5,085 (25%) | 25,398 (15%) | 12,953 (9.64%) |
| Total cholesterol (mmol/L) | 5.56 [4.75;6.38] | 5.73 [5.00;6.49] | 5.72 [5.03;6.45] |
| Total cholesterol (mg/dl) | 215 [184;247] | 222 [193;251] | 221 [195;249] |
| LDL cholesterol (mmol/L) | 3.49 [2.87;4.13] | 3.59 [3.03;4.18] | 3.54 [3.01;4.12] |
| LDL cholesterol (mg/dl) | 135 [111;160] | 139 [117;162] | 137 [116;159] |
| HDL cholesterol (mmol/L) | 1.30 [1.09;1.55] | 1.39 [1.17;1.66] | 1.47 [1.23;1.75] |
| HDL cholesterol (mg/dl) | 50 [42;60] | 54 [45;64] | 57 [48;68] |
| Estimated GFR (ml/min/1.73 m 2 ) | 92.04 [80.94;99.59] | 92.26 [82.46;99.38] | 93.43 [84.28;100.55] |
During a median of 7.74 years of follow-up for all-cause mortality and 6.98 years of follow-up for MI, 8,814 subjects expired, and 3,958 subjects experienced MI. The risk of MI and mortality was significantly different according to the presence of slow walking pace or poor handgrip strength ( Figure 2 , Table 2 ). Subjects with poor handgrip strength also had higher MI risk than those with reference range handgrip strength, although the effect size of the hazard ratios was smaller than the results by the analysis including slow walking pace exposure. All-cause mortality risk was also significantly higher in those with slow walking pace or poor handgrip strength than in those without such functional sarcopenia. There were 1,552 cardiovascular death events and 5,518 cancer deaths in the study cohort. When the risk for the outcomes was analyzed, overall effect sizes according to the exposure variables were larger with cardiovascular deaths than with cancer deaths. Nevertheless, poor handgrip strength and slow walking pace were significantly associated with risks for both cardiovascular deaths and cancer deaths.
