2024 年 31 巻 1 号 p. 81-89
Aim: A high level of serum lipoprotein(a) [Lp(a)] is associated with kidney disease development in patients with type 2 diabetes (T2DM). Recent studies have suggested that statins may affect serum levels of Lp(a). However, the statin effect is not well-defined in patients with T2DM with kidney dysfunction. This retrospective study aimed to investigate the relevance of kidney dysfunction and statin therapy to Lp(a) in patients with T2DM.
Methods: Japanese patients with T2DM (n=149, 96 men and 53 women) were divided into two groups: statin users (n=79) and non-statin users (n=70). Multiple logistic regression analyses were performed with Lp(a) as the objective variable and estimated glomerular filtration rate (eGFR), hemoglobin A1c, age, gender, and body mass index as the explanatory variables.
Results: Lp(a) serum levels were higher in statin users than in non-statin users (P=0.022). Multivariate regression analysis results showed an inverse correlation of eGFR to log Lp(a) in all patients (P=0.009) and in non-statin users (P=0.025), but not in statin users. In a multiple logistic regression analysis for median Lp(a), there was an inverse association between eGFR and Lp(a) level (odds ratio, 0.965; 95% confidence interval, 0.935–0.997; P=0.030) in non-statin users as well as in all participants, but not in statin users.
Conclusions: The present study suggests that a high Lp(a) level in patients with T2DM, except in statin users, is significantly associated with decreased eGFR, indicating that the increased Lp(a) levels under statin therapy might diminish the relationship between Lp(a) and eGFR.
Cardiovascular disease (CVD) is a leading cause of mortality worldwide. One of the major cardiovascular risk factors is a high level of low-density lipoprotein—cholesterol (LDL-C)1, 2). Statins, first-line drugs for LDL-C lowering, markedly improve cardiovascular outcomes, but nevertheless, cardiovascular events remain in a substantial proportion of patients with dyslipidemia, which is considered a residual cardiovascular risk. Lipoprotein(a) [Lp(a)] is an independent cardiovascular risk factor and is also considered a significant residual cardiovascular risk factor1-4). Lp(a) is composed of apolipoprotein(a) [apo(a)] bound covalently to apolipoprotein B of LDL particles5). Apo(a) consists of repetitive domains, so-called kringles, and carries a number of kringle IV-2 repeats. Apo(a) also varies widely in size due to a polymorphism in kringle IV-2. When the size of apo(a) is small, the concentration of Lp(a) is high3, 5-7). In this way, serum Lp(a) concentration is determined primarily by heredity. In patients with type 2 diabetes (T2DM), a high level of serum Lp(a) increases the CVD risk and is also considered to be involved in the development of diabetic complications, such as retinopathy and kidney disease8-10). Thus, how serum Lp(a) levels should be interpreted and addressed in clinical practice would be presumably important for patients with T2DM.
Statin therapy is established as the cornerstone of lipid lowering for the prevention of CVD. However, previous reports suggest that statins significantly increase serum Lp(a) levels and elevations in Lp(a) after statin therapy warrant attention because of their effects on residual cardiovascular risk2-4). The LDL-C-lowering target is stringent even in primary prevention in patients with T2DM because patients with diabetes with high levels of LDL-C are at high CVD risk11, 12). In patients with T2DM, statins are a first-line drug for lowering LDL-C because statins prevent CVD and improve their life prognosis12-14). Although statins are used in this way, statin-induced elevations of Lp(a) have also been reported15).
Previous studies have reported an association between kidney function and albuminuria with Lp(a) levels in Japanese patients with T2DM10, 16). There may be a correlation between a high Lp(a) level and kidney dysfunction in T2DM, but the statin effect on Lp(a) is not well-defined in T2DM with kidney dysfunction. Therefore, it is not clear how we can evaluate Lp(a) levels in patients with T2DM who have received statin therapy.
The present study investigated the following: (1) whether there is a relationship between Lp(a) serum levels and kidney dysfunction in T2DM and (2) how we can assess the relationship between serum Lp(a) concentrations and kidney function in patients with T2DM on statin therapy. Here we report a retrospective study investigating the relevance of diabetic kidney disease and statins to Lp(a) serum levels in patients with T2DM.
The present study was based on a retrospective, cross-sectional design. The study participants were patients with T2DM who visited the internal medicine outpatient clinic of the Division of Diabetes, Metabolism, and Endocrinology at the Jikei University Kashiwa Hospital from August 2018 to May 2020. Exclusion criteria were patients with poor glycemic control (glycohemoglobin A1c [HbA1c] levels ＞10.0%), diabetic ketoacidosis and hyperglycemic crisis, type 1 diabetes, including secondary diabetes, endocrine diseases, gastrointestinal disorders, pancreatic diseases, cancer, and malignant neoplasms; patients on medication with steroid, anticancer drugs and molecular target drugs; patients with an estimated glomerular filtration rate (eGFR) under 30 mL/min/1.73 m2 (stages 4 and 5 of chronic kidney disease [CKD]); and premenopausal women17).
This study conformed to the Declaration of Helsinki and was approved by the Ethics Committee of the Jikei University School of Medicine (approval number: 30-010), and a waiver for informed consent was approved with opt-out consent.
Data Collection and Laboratory MeasurementsPatient characteristics (age, gender, body mass index [BMI], blood pressure, and medication status) and biochemical laboratory data (glucose, HbA1c, lipids, uric acid (UA), and liver and kidney function tests of fasting samples) were collected from electronic medical records. Missing blood and urine data of participants were collected on another day visit at the hospital outpatient clinic.
Statin users were defined as the participants who were prescribed statins for at least 3 months until the day their characteristics and biochemical laboratory data were collected.
HbA1c was measured by high-performance liquid chromatography with a glycohemoglobin analyzer HA-8190V (Arkray, Japan), and other data were measured by routine methods. For example, total cholesterol (TC) and triglyceride (TG) were measured by conventional methods using a Determiner L TC and TG (Minaris Medical, Hitachi, Japan). High-density lipoprotein—cholesterol (HDL-C) was measured by a direct method using a Metabolead HDL-C (Minaris Medical). LDL-C was determined by the Friedewald formula (LDL-C=TC − HDL-C − [TG/5])18) or a direct method (Metabolead LDL-C, Minaris Medical). Urinary albumin excretion (urine albumin-to-creatinine ratio [UACR], mg/g) and serum Lp(a) were measured by immunonephelometry using Autokit Micro Albumin (Fujifilm Wako, Japan) and latex agglutination immunoassay (Sekisui Medical, Japan), respectively. In addition, eGFR (mL/min/1.73 m2) was calculated using the following formula19): 194×serum creatinine (mg/dL)−1.094×age (years)−0.287 for men, and this calculated eGFR×0.739 for women. The diagnosis of diabetic kidney disease was given using the Classification of Diabetic Nephropathy 2014 20).
Statistical AnalysisAll statistical analyses were performed using the Statistical Package for Social Sciences (version 25.0; IBM, Tokyo, Japan). The required number of patients to observe a difference in Lp(a) between statin users and non-statin users was determined to be 134. This sample size was calculated by using G*Power version 3.1.9.7 (Franz Faul, Kiel University, Germany) 21). An α level of 0.05, 80% power and 0.5 effect size.
The continuous variables involved in this study were represented by median (interquartile range) and mean±standard deviation (SD). The Mann–Whitney U test and Welch t test were used for group comparison. The Kolmogorov–Smirnov test was used to determine the homogeneity of the variables in this study. The Welch t test was used to compare the variables that conform to normal distributions (age, BMI, systolic blood pressure, diastolic blood pressure, creatinine, eGFR, UA, glucose, HbA1c, TC, HDL-C, LDL-C). The Mann–Whitney U test was used to compare the variables that conform to non-normal distributions (serum albumin, UACR, TG, Lp[a]). Categorical variables (gender, anti-diabetic drugs, insulin, other lipid-lowering drugs, anti-hypertensive drugs, UA-lowering drugs) were represented by frequency and percentage. Fisher’s exact test was used for group comparison.
Simple correlations between Lp(a), an objective variable, and other explanatory continuous variables were evaluated by Spearman correlation analysis. Next, multiple regression analyses were performed for Lp(a) as an objective variable with the explanatory variables eGFR and HbA1c and the universal confounding factors age, gender, and BMI.
In the whole study group, the median value of serum Lp(a) was 7 mg/dL. The study patients were classified into Group 1 (high Lp(a) ≥ 8 mg/dL; n=69) or Group 2 (low Lp(a) ＜8 mg/dL; n=80). For subgroup analyses according to median Lp(a) values, statin users were classified as Group 3 (high Lp(a) ≥ 10 mg/dL; n=39) or Group 4 (low Lp(a) ＜10 mg/dL; n=40), and non-statin users were classified as Group 5 (high Lp(a) ≥ 5 mg/dL; n=34) or Group 6 (low Lp(a) ＜5 mg/dL; n=36).
In addition, univariate and multivariate logistic regression analyses were performed to analyze the relationship between nominal objective variables (groups by Lp(a) median values) and continuous explanatory variables (one SD increase in age, BMI, HbA1c, and eGFR) or nominal explanatory variables (gender, presence or absence of statins, and BMI [high group ≧25 (obese, n=69) or low group ＜25 kg/m2 (non-obese, n=80)] instead of the continuous variable BMI). The results were expressed as odds ratios (ORs) and 95% confidence intervals (CIs), and the predictive values of explanatory factors for high levels of Lp(a) were investigated. A P value ＜0.05 was considered statistically significant.
The log-transformation was used for Lp(a) in the multiple regression analyses and UACR in the univariate and multivariate logistic regression analyses because these two kinds of variables had non-normal distributions.
Statin users (n=79) were using five different statins (rosuvastatin, 49; pravastatin, 11; pitavastatin, 14; atorvastatin, 4; simvastatin, 1). Characteristics, background, and biochemical data are shown in Table 1. There were no differences in age, BMI, gender distribution, blood pressure, albumin, creatinine, eGFR, UA, UACR, plasma glucose, HbA1c, TG, or medications between the 79 statin users and the 70 non-statin users. In the lipid data, TC and LDL-C were lower, but Lp(a) was higher in statin users than in non-statin users. Because Lp(a) levels were significantly different between statin users and non-statin users, simple correlation analyses for continuous variables with serum Lp(a) as an objective value were performed in the statin user and non-statin user groups, respectively. In statin users, Lp(a) levels were positively correlated with HDL-C and inversely correlated with BMI (Table 2). In non-statin users, age and plasma glucose were positively correlated with Lp(a), whereas eGFR was inversely correlated with Lp(a) (Table 2).
| Total (n = 149) | Statin users (n = 79) | Statin non-users (n = 70) |
Statin users vs non-users P value |
|
|---|---|---|---|---|
| Age (years) | 68±11 | 66±11 | 69±12 | 0.139 |
| BMI (kg/m2) | 25.8±4.2 | 26.0±3.8 | 25.7±4.7 | 0.747 |
| Gender, male/female (%) | 96 (64)/53 (35) | 48 (60)/31 (39) | 48 (68)/22 (31) | 0.320 |
| Systolic blood pressure (mmHg) | 130.7±15.9 | 129.3±17.0 | 132.2±14.5 | 0.273 |
| Diastolic blood pressure (mmHg) | 73.8±12.0 | 73.3±12.6 | 74.3±11.4 | 0.628 |
| Albumin (g/dL) | 4.1 (3.8‐4.3) | 4.2 (3.9‐4.3) | 4.05 (3.8‐4.3) | 0.149 |
| Creatinine (mg/dL) | 1.0±0.3 | 1.0±0.4 | 0.9±0.3 | 0.120 |
| eGFR (ml/min/1.73 m2) | 61.4±19.4 | 58.9±19.7 | 64.2±18.9 | 0.093 |
| Uric acid (mg/dL) | 5.6±1.4 | 5.5±1.4 | 5.8±1.4 | 0.282 |
| UACR (mg/g Cr) | 53.1 (13.3‐234) | 51.7 (14.0‐635) | 57.5 (12.1‐194) | 0.287 |
| Glucose (mg/dL) | 147.5±49.1 | 143.3±49.9 | 152.2±48.1 | 0.272 |
| HbA1c (%) | 7.5±1.0 | 7.5±1.0 | 7.6±1.0 | 0.864 |
| Total cholesterol (mg/dL) | 187.0±37.4 | 180.0±37.8 | 195.0±35.6 | 0.014 |
| Triglyceride (mg/dL) | 128 (88‐181) | 136 (93.5‐187) | 119 (85.8‐165) | 0.221 |
| HDL-C (mg/dL) | 57.2±14.8 | 56.3±15.4 | 58.1±14.1 | 0.461 |
| LDL-C (mg/dL) | 99.8±30.8 | 91.8±31.4 | 108.8±27.7 | 0.001 |
| Lipoprotein (a) (mg/dL) | 7 (2-20) | 9 (3-24) | 4 (2-12) | 0.022 |
| Medication | ||||
| Anti-diabetic drugs (n, %) | 138 (93.6) | 75 (94.9) | 63 (90.0) | 0.403 |
| Insulin (n, %) | 75 (50.3) | 41 (51.9) | 34 (48.6) | 0.685 |
| Other lipid-lowering drugs (n, %) | 18 (12.1) | 12 (15.2) | 6 (8.6) | 0.216 |
| Anti-hypertensive drugs (n, %) | 98 (65.8) | 55 (69.2) | 43 (61.4) | 0.293 |
| Uric acid-lowering drugs (n, %) | 28 (18.8) | 15 (19.0) | 13 (18.6) | 0.948 |
Continuous variables are expressed as mean±SD or median (inter quartile range, IQR), and categorical variables were expressed as number (percentage). Abbreviations: N, number; BMI, body mass index; eGFR, estimated glomerular filtration rate; HbA1c, glycohemoglobin; UACR, urine albumin to creatinine ratio; HDL-C, high-density lipoprotein-cholesterol; LDL-C, low-density lipoprotein-cholesterol; SD standard deviation.
| statin users | statin non-users | |||
|---|---|---|---|---|
| rs | P value | rs | P value | |
| Age | 0.156 | 0.169 | 0.346 | 0.003 |
| BMI | -0.248 | 0.027 | -0.021 | 0.862 |
| Gender | 0.206 | 0.068 | -0.051 | 0.678 |
| eGFR | -0.192 | 0.090 | -0.386 | 0.001 |
| Uric acid | -0.012 | 0.916 | 0.181 | 0.133 |
| UACR | 0.019 | 0.867 | 0.144 | 0.235 |
| Glucose | -0.045 | 0.693 | 0.244 | 0.042 |
| HbA1c | -0.074 | 0.516 | 0.049 | 0.689 |
| Total cholesterol | 0.185 | 0.102 | 0.120 | 0.323 |
| Triglyceride | -0.009 | 0.935 | -0.162 | 0.179 |
| HDL-C | 0.248 | 0.028 | 0.126 | 0.297 |
| LDL-C | 0.173 | 0.128 | 0.128 | 0.289 |
Abbreviations: BMI, body mass index; UACR, urine albumin to creatinine ratio; eGFR, estimated glomerular filtration rate; HbA1c, glycohemoglobin A1c; HDL-C, high-density lipoprotein-cholesterol; LDL-C, low-density lipoprotein-cholesterol; rs, Spearman’s rank correlation coefficient
Then, multivariate regression analysis results showed an inverse correlation of eGFR to log Lp(a) in all participants (Table 3). An inverse correlation of eGFR to log Lp(a) was also found in non-statin users, but not in statin users. In addition, the significant correlation between BMI and log Lp(a) was not found in the multiple regression analysis (Table 3).
| Extracted explanatory variables | Total (n = 149) | Statin users (n = 79) | Statin non-users (n = 70) | |||
|---|---|---|---|---|---|---|
| β | P value | β | P value | β | P value | |
| eGFR | -0.244 | 0.009 | -0.172 | 0.223 | -0.290 | 0.025 |
| BMI | -0.148 | 0.075 | -0.230 | 0.055 | 0.020 | 0.875 |
| HbA1c | -0.056 | 0.492 | -0.084 | 0.488 | 0.057 | 0.621 |
| log UACR | 0.023 | 0.796 | 0.001 | 0.995 | 0.092 | 0.424 |
| Age | 0.065 | 0.466 | 0.046 | 0.694 | 0.195 | 0.157 |
| R2 | 0.107 | 0.006 | 0.097 | 0.182 | 0.196 | 0.014 |
Abbreviations: eGFR, estimated glomerular filtration rate; BMI, body mass index; HbA1c, glycohemoglobin A1c; UACR, urine albumin to creatinine ratio; R2, coefficient of determination; β, standard partial regression coefficient.
Subsequently, a multivariate logistic regression analysis was performed using one SD increases in age, BMI, HbA1c, eGFR, and UACR as predictor variables for Lp(a) levels in all participants, the statin user group, and the non-statin user group (Table 4). In all participants (OR, 0.975; 95% CI, 0.955–0.996; P=0.018) and the non-statin user group (OR, 0.965; 95% CI, 0.935–0.997; P=0.030), eGFR was independently associated with high Lp(a) (≥ 8 mg/dL), but not in the statin user group. However, BMI was independently associated with high Lp(a) in the statin user group. On the contrary, when analyzed using the nominal variable BMI, the association between BMI and Lp(a) was not found (Table 5).
| All participants (n = 149) | |||
| β | OR | P value | |
| eGFR | -0.025 | 0.975 (0.955-0.996) | 0.018 |
| BMI | -0.047 | 0.954 (0.875-1.040) | 0.283 |
| HbA1c | -0.243 | 0.784 (0.545-1.128) | 0.190 |
| log UACR | -0.118 | 0.889 (0.573-1.380) | 0.600 |
| Age | 0.017 | 1.017 (0.983-1.052) | 0.323 |
| Statin users (n = 79) | |||
| β | OR (95% CI) | P value | |
| eGFR | -0.015 | 0.985 (0.955-1.015) | 0.326 |
| BMI | -0.141 | 0.868 (0.755-0.999) | 0.048 |
| HbA1c | -0.237 | 0.789 (0.463-1.343) | 0.382 |
| Log UACR | -0.069 | 0.933 (0.494-1.761) | 0.830 |
| Age | 0.023 | 1.023 (0.977-1.072) | 0.332 |
| Statin non-users (n = 70) | |||
| β | OR (95% CI) | P value | |
| eGFR | -0.036 | 0.965 (0.935-0.997) | 0.030 |
| BMI | 0.042 | 1.043 (0.918-1.185) | 0.515 |
| HbA1c | 0.072 | 1.074 (0.598-1.930) | 0.811 |
| log UACR | 0.428 | 1.535 (0.762-3.093) | 0.231 |
| Age | 0.037 | 1.038 (0.982-1.097) | 0.192 |
Data are expressed as nominal variables [lipoprotein(a) (high groups ≧the median values, low group ＜the median values)] or continuous variables (eGFR, BMI, HbA1c, log UACR, Age). Abbreviations: β, partial regression coefficient; OR, odds ratio; CI, confidence interval; eGFR, estimated glomerular filtration rate; BMI, body mass index; HbA1c, glycohemoglobin A1c; UACR, urine albumin to creatinine ratio.
| All participants (n = 149) | |||
| β | OR | P value | |
| eGFR | -0.025 | 0.975 (0.955-0.996) | 0.020 |
| BMI | |||
| obese (n = 69) or non-obese (n = 80) | -0.285 | 0.752 (0.371-1.525) | 0.430 |
| HbA1c | -0.242 | 0.785 (0.545-1.129) | 0.191 |
| log UACR | -0.119 | 0.887 (0.572-1.377) | 0.594 |
| Age | 0.019 | 1.019 (0.986-1.054) | 0.267 |
| Statin users (n = 79) | |||
| β | OR (95% CI) | P value | |
| eGFR | -0.011 | 0.989 (0.960-1.019) | 0.460 |
| BMI | |||
| obese (n = 39) or non-obese (n = 40) | -0.711 | 0.491 (0.184-1.312) | 0.156 |
| HbA1c | -0.294 | 0.745 (0.442-1.257) | 0.270 |
| Log UACR | -0.048 | 0.953 (0.512-1.775) | 0.880 |
| Age | 0.025 | 1.026 (0.979-1.074) | 0.285 |
| Statin non-users (n = 70) | |||
| β | OR (95% CI) | P value | |
| eGFR | -0.037 | 0.963 (0.933-0.995) | 0.023 |
| BMI | |||
| obese (n = 34) or non-obese (n = 36) | 0.358 | 1.430 (0.464-4.409) | 0.534 |
| HbA1c | 0.074 | 1.076 (0.614-1.887) | 0.797 |
| log UACR | 0.000 | 1.000 (0.999-1.001) | 0.958 |
| Age | 0.034 | 1.035 (0.980-1.093) | 0.216 |
Data are expressed as nominal variables {lipoprotein(a) [high groups ≧the median values, low group ＜the median values] and BMI [high group ≧ 25 (obese), or low group ＜25 kg/m2 (non-obese)]}or continuous variables (eGFR, HbA1c, log UACR, Age). Abbreviations: β, partial regression coefficient; OR, odds ratio; CI, confidence interval; eGFR, estimated glomerular filtration rate; BMI, body mass index; HbA1c, glycohemoglobin A1c; UACR, urine albumin to creatinine ratio.
The present study shows that a decrease in eGFR, as an indicator of renal function, is significantly associated with a high Lp(a) level in statin-free patients with T2DM, but not in statin users. As reported previously, serum Lp(a) levels were significantly higher in statin users than in non-statin users15). Therapeutic drugs for diabetes, hypertension, and dyslipidemia, except for statins, were not different between statin users and non-statin users in our study (Table 1).
Effects of statin therapy on serum Lp(a) levels have been reported previously, including a dose-dependent increase in Lp(a) levels by lovastatin therapy22). A recent meta-analysis of six previously reported studies revealed that statins (atorvastatin, rosuvastatin, and pravastatin) increase Lp(a) levels, but one of these studies did not show an increase15, 23). This different result might be due to the study’s sample size effects and/or ethnic differences15, 23). In addition, the effect of statins on Lp(a) is still unclear, and it is not possible to accurately determine whether Lp(a) is elevated by the administration of statins or whether statin users just had higher Lp(a) levels than non-statin users in the present study. Although the underlying mechanisms by which statins affect Lp(a) levels are not fully defined, it has been suggested that statin-induced increases in the production of apo(a) and proprotein convertase subtilisin/kexin type 9 protein-related potentiation of Lp(a) may lead to increased serum levels of Lp(a)15, 24). Hence, the increased Lp(a) serum levels in statin users in the present study are presumably consistent with the statin effects on Lp(a) serum levels reported in the previous studies.
The univariate and multivariate correlation analysis data showed that eGFR was inversely correlated with Lp(a) in all participants and in non-statin users, but not in statin users. These results indicate that low levels of eGFR would be a potent marker for high Lp(a). Then, a multivariate logistic regression analysis was used to evaluate the predictive values of eGFR for Lp(a) levels. A one SD increase in eGFR was a significant inverse predictor for high Lp(a) in all participants (OR, 0.975) and in non-statin users (OR, 0.965), but not in statin users. Therefore, these results in the present cross-sectional study indicate an independent association between low eGFR and high Lp(a) in non-statin users with diabetes. A recent Mendelian randomization study demonstrated the relevance of CKD to high Lp(a)25). The increased Lp(a) found in patients with CKD may be attributed to enhanced production of Lp(a) and impaired catabolism of Lp(a) in patients with proteinuria and lowered eGFR, respectively26). In addition, Lp(a) lowering might prevent the progression of CKD25). Conversely, a high Lp(a) level might be considered causal for a decrease in eGFR. Lp(a) has high homology for plasminogen, and it has been demonstrated in vitro that Lp(a) competitively inhibits the fibrin-dependent activation of plasminogen to plasmin, which plays a crucial role in the catabolism of extracellular matrix proteins27). An experimental study showed that abnormalities in lipoprotein metabolism cause glomerular and tubulointerstitial damage by causing macrophage activation and infiltration in the kidney28). Namely, a high level of serum Lp(a) could be not only attributed to the lowered eGFR but also might provide a cause for the decrease in eGFR. A longitudinal, observational cohort study demonstrated that the Lp(a) level was an independent prognostic factor for the future development of CKD in Korean patients with T2DM29). Nevertheless, our study results indicate that low levels of eGFR would be a potent marker for high Lp(a).
The inverse association between Lp(a) and eGFR found in patients without statin therapy in the present study is consistent with the results of previous studies, but this association was not found in patients with statin therapy. The existing relationship between Lp(a) and eGFR might be diminished in part by the increased Lp(a) levels under statin therapy, and as such, the absence or presence of statin therapy should be noted when evaluating Lp(a) for the risk of developing complications in patients with T2DM and kidney disease.
Furthermore, the univariate correlation and multivariate logistic analysis data showed that BMI was inversely associated with Lp(a) in statin users, but not in all participants or in non-statin users. Thus far, there is no report that indicates a direct correlation between BMI and Lp(a). However, some studies indicated a negative correlation between metabolic syndrome and serum Lp(a), and several prospective cohort studies reported that individuals with metabolic syndrome have significantly lower Lp(a) levels than those without metabolic syndrome30-32). Insulin resistance can modulate the genetically determined Lp(a) production in the liver33). Previous reports demonstrated an inverse association of Lp(a) with markers of insulin resistance in participants with dyslipidemia, but another study showed no evidence for a causal role of Lp(a) and insulin33, 34). Unfortunately, in our study, data on serum insulin concentration and insulin resistance markers were not investigated. However, BMI, which is frequently high in metabolic syndrome, could be speculated to be inversely correlated with Lp(a), which appears to be consistent with the inverse relationship between BMI and Lp(a) found in statin users in our study. In contrast, it is inconsistent with the lack of significant correlations between them in all participants and non-statin users. Meanwhile, when analyzed by using the nominal variable BMI (high group ≧25 [obese] or low group ＜25 kg/m2 [non-obese]) instead of the continuous variable BMI, the association between BMI and Lp(a) was also not found. Nevertheless, this discrepancy and issues related to the relevance of BMI and metabolic syndrome to Lp(a) remain to be resolved.
There are several limitations to our study. First, this study is a retrospective and cross-sectional study. Hence, the causal relationship between increased Lp(a) and decreased kidney function remains to be clarified, although it might be considered that decreased kidney function increases Lp(a). Second, the relevance of kidney function and statins to Lp(a) serum levels in patients under lipid-lowering therapy with proprotein convertase subtilisin/kexin type 9 inhibitors or apolipoprotein B antisense oligonucleotides, which markedly affect Lp(a) serum levels, should be investigated in future studies35). Third, plasma Lp(a) concentrations are primarily determined by the apo(a) gene, LPA 5), but genetic testing data were not acquired in our study. Finally, this study was conducted only at a single institution and was relatively small. Therefore, a prospective, large-scale study to investigate the causal relationship should be performed in the future.
The present study shows that an inverse association between Lp(a) and eGFR was found in patients free of statin therapy. This study also suggests that decreased eGFR is significantly associated with high Lp(a) in patients with T2DM. Nevertheless, the influence of statin therapy on the relationship between eGFR and Lp(a) should be presumably considered because this relationship might be diminished by the increased Lp(a) levels under statin therapy. However, further investigations will be needed to clarify the causal relationship between increasing Lp(a) and decreasing kidney function in patients with T2DM.
We appreciate helpful supports from members of Department of Laboratory medicine, the Jikei University Kashiwa Hospital.
This study was supported by Grant-in-Aid for Scientific Research (number 17K09560) from Japan Ministry of Education, Culture, Sports, Science, Research Grant from research support program of Roche Diagnostics and Technology, and the Jikei University Research Fund from the Jikei University School of Medicine (H. Yoshida).
Prof. Hiroshi Yoshida has received honoraria for speaking activities from Denka and Kowa. Other authors declare that they have no competing interests.
