Background
The kidney is a vital organ for volume homeostasis, uremic toxin clearance, maintenance of body electrolyte balance, and various biological functions. A state of impaired kidney function, chronic kidney disease (CKD), is an emerging comorbidity with a high prevalence and large socioeconomic burden, thus, assessment of kidney function with the estimated glomerular function rate (eGFR) is commonly performed in diverse clinical conditions [
1].
The close linkage between myocardial infarction (MI) and eGFR has been noted previously [
2,
3]. CKD is one of the most widely acknowledged risk factors for MI, and reduced eGFR is associated with a poor prognosis in MI patients. In addition, recent observational studies reported that a state of supranormal eGFR, kidney hyperfiltration, was associated with a higher risk of cardiovascular diseases [
4‐
7]. To further confirm that the observational findings were from the causal effects on eGFR on MI risks, Mendelian randomization (MR) analysis, an analytical tool widely used in the recent medical literature, can be helpful. MR analysis has strengths in demonstrating causal estimates minimally affected by reverse causation or confounding effects, as the method implements inborn-fixed genetic instrument variables. However, previous MR studies reported null causal effects of kidney function parameters on MI [
8,
9], a finding that was contradictory to previous observational findings. Nevertheless, considering that a causal effect may be nonlinear, conventional summary-level MR analysis would not capture the complex exposure-outcome relation because it assumes linearity, particularly for a suspected U-shaped relationship between eGFR and cardiovascular risk [
7,
10,
11]. Therefore, further nonlinear MR analysis is warranted to investigate the shape of the causal estimates on MI according to eGFR values.
In this study, we hypothesized that a causal effect of eGFR on MI risk would be present with a nonlinear exposure-outcome relationship. We performed a nonlinear MR analysis utilizing the largest individual-level genetic database that includes MI phenotyping and eGFR measurements, the UK Biobank.
Discussion
In this MR study, we identified that genetically predicted eGFR is significantly associated with MI risk with a quadratic shape. Our results indicated that a reduction in eGFR may be a causal factor for higher MI risk in individuals with an eGFR in the low range. In addition, the results suggested that supranormal eGFR values, commonly known as kidney hyperfiltration, may causally elevate the risk of MI.
Kidney function impairment is one of the most widely recognized risk factors for cardiovascular diseases. Such a close linkage raised suspicion that kidney function impairment may causally increase the risk of MI. However, as there are shared risk factors such as hypertension and diabetes for CKD and MI and the possibility of reverse causation remains, the causal effects of kidney function on MI risk have been difficult to be confirmed by conventional observational studies. To solve this issue, recent studies have implemented MR analysis [
8,
9]. In MR, causal estimates can be yielded as genetically predicted exposure is determined before birth; thus, the instrumental variable approach is minimally affected by confounding effects or reverse causation. However, previous MR results indicated null causal effects from cystatin C- or creatinine-related parameters and did not support that clinical interventions targeting kidney function impairment would also be helpful for reducing the risk of MI [
8,
9]. However, previous MR analyses were based on the linearity assumption and tested the causal estimates throughout the entire range of kidney function exposure. As supranormal eGFR values were reported to be associated with all-cause mortality or atherosclerotic cardiovascular diseases, kidney function could have a parabolic causal effect on MI risk [
4‐
6,
33,
34]. We implemented nonlinear MR analysis methods to investigate the issue and identified that genetically predicted eGFR was significantly associated with MI risk with a quadratic shape of the exposure-outcome relation. Therefore, this MR study supports that kidney function impairment would be a causal factor for a higher MI risk, and the linkage would not be from external confounding or reverse causal effects.
A reduction in eGFR, usually below 60 mL/min/1.73 m
2 where stage 3 chronic kidney disease is defined [
35], has been reported to be an independent risk factor for coronary artery disease. The observational associations remained significant even after adjustment for known traditional risk factors such as hypertension or diabetes [
2]. Recent findings suggested the clinical significance of fibroblast-growth factor 23-mediated pathways or calcium-phosphate metabolism in regard to the risk of CKD progression and MI [
36,
37]. A platelet-related mechanism has also been suggested to mediate the linkage between CKD and MI, as a reduction in eGFR was associated with higher thrombotic activity and poor responses to antiplatelet agents [
38‐
40]. There have been other pathophysiologic mechanisms, such as the induction of inflammation, vascular calcification, or endothelial dysfunction, that may explain the close association between CKD and MI [
41]. With the current MR findings, a decrease in eGFR below the reference range < 60 mL/min/1.73 m
2, may be considered a “causal” factor that elevates the risk of MI. Further, the identified causal effects imply that clinical interventions targeting kidney function impairment may also be beneficial for preventing MI.
Kidney hyperfiltration has been reported to be associated with the risk of cardiovascular diseases [
7,
11], even in a report where direct measurements of GFR were performed [
42]. Specifically, a previous systematic meta-analysis including 24 observational cohorts of 637,315 individuals showed that eGFR ≥ 105 mL/min/1.73 m
2 was significantly associated with higher risks of adverse cardiovascular risks [
7]. Our results suggested that MI risk was higher in higher ranges of genetically predicted eGFR values, similar to previous observational findings, independent of major comorbidities, suggesting that kidney hyperfiltration may be another “causal” factor for MI similarly as the state of reduced eGFR below 60 mL/min/1.73 m
2. This interpretation should be made carefully because eGFR is an estimated value and even cystatin C may be affected by nonkidney factors [
43,
44]. However, as kidney hyperfiltration is considered another state of impaired kidney function associated with future rapid eGFR decline [
45], it may be acceptable that early kidney function impairment represented as supranormal eGFR may affect MI risk. Upregulation of the renin-angiotensin-aldosterone system and increased proximal tubular sodium-glucose reabsorption are reasons for glomerular hyperfiltration as they affect tubuloglomerular feedback [
46]. Considering that renin-angiotensin aldosterone system blockade or sodium-glucose cotransporter inhibitors 2 reduce both glomerular hyperfiltration and cardiovascular risk [
47‐
50], the linkage may be explained by the mediating mechanism. A future study is warranted to validate our findings and confirm the mechanism of supranormal eGFR in regard to the risk of MI. In addition, a study may test the potential benefits of identifying the cause of kidney hyperfiltration or clinical interventions to reverse supranormal eGFR.
On the other hand, the parabolic shape of the association between genetically predicted eGFR and MI risk explained the null causal estimates reported by previous MR studies [
8,
9] and our summary-level MR analysis. This study emphasizes that nonlinear MR analysis should be considered when a U-shaped causal estimate according to the exposure variable is suspected, as conventional summary-level MR analysis relies on the linearity assumption and can be attenuated for such quadrative relations. In addition, the overall effect size of the localized averaged causal estimates was relatively small compared to the findings in observational studies [
7]. This finding may imply that the previously reported observational association between eGFR and MI risks might have been overestimated due to residual confounding effects.
There are some limitations of this study. First, MR analysis cannot prove the clinical utility of modifying an exposure to affect an outcome [
51]. Although this study suggests the causal linkage between kidney function impairment and MI risk, a result based on a clinical trial is necessary to suggest the clinical implications of our findings. In addition, as it is difficult to provide interpretable effect sizes of the causal estimates in non-linear MR analysis, the degree of the suggested causal effects could not be determined herein. Second, MR analysis cannot provide a direct mechanistic explanation for the identified causal effects. In particular, as eGFR values are as an estimated value and cystatin C levels could also be affected by external factors, the mechanism of supranormal kidney hyperfiltration and its clinical significance should be validated in future studies. Third, the possibility of selection bias remains. Although the UK Biobank dataset is the largest genetic dataset where nonlinear MR analysis was possible, the UK Biobank cohort has a healthy volunteer bias [
52]. Additional studies may be necessary to retest the causal estimates in a population with characteristics that are closer to those of the general population. Last, nonlinear MR studies rely on allele score-based analysis. Although we adjusted clinical covariates to support the attainment of the independence assumption, potential unmeasured pleiotropic effects should be considered.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.