Lipoprotein(a) levels predict endothelial dysfunction in maintenance hemodialysis patients: evidence from vascular reactivity index assessment

Abstract

Background

Patients with maintenance hemodialysis (MHD) present endothelial dysfunction (ED), which is characterized by impaired vasodilation and a pro-inflammatory state. Lipoprotein(a) (Lp(a)) has pro-inflammatory and pro-atherogenic properties. No study has investigated the association between serum Lp(a) and ED in patients with MHD. This study was conducted to address this issue.

Methods

We collected serum specimens from 123 fasting MHD patients. The endothelial function was measured using the vascular reactivity index (VRI) determined by digital thermal monitoring, and VRI values of ≥ 2.0, 1.0 to <2.0, and < 1.0, indicated good, intermediate, and poor vascular reactivity, respectively. Lp(a) levels were measured by enzyme-linked immunosorbent assay.

Results

Of the 123 MHD patients, 54 (43.9%) had good VRI, 51 (41.5%) had intermediate VRI, and 18 (14.6%) had poor VRI. Serum Lp(a) levels (p < 0.001) were significantly higher in the poor and intermediate VRI groups than in the good VRI group. The log-transformed serum Lp(a) (log-Lp(a); p < 0.001) showed a negative correlation with VRI values. Multivariate logistic regression analyses indicated that serum Lp(a) level was independently associated with vascular reactivity dysfunction (both intermediate and poor VRI; p = 0.001) and poor vascular reactivity (poor VRI; p < 0.001). The areas under the receiver operating characteristic curve were 0.754 and 0.853 for predicting vascular reactivity dysfunction and poor vascular reactivity, respectively, by Lp(a).

Conclusion

The serum Lp(a) level had a negative correlation with the VRI, and it may serve as a potential biomarker for early detection of ED in MHD patients.

Graphical Abstract

Introduction

Maintenance hemodialysis (MHD) is the most common form of renal replacement therapy to treat patients with end-stage renal disease (ESRD), accounting for approximately 89% of all dialysis [1]. In patients with MHD, atherosclerosis is a major comorbidity and contributes to increased cardiovascular mortality [2,3]. Endothelial dysfunction (ED) is characterized by impaired vasodilation, increased oxidative stress, a pro-inflammatory state, and prothrombotic properties [4]. ED is one of the earliest symptoms of atherosclerosis, and it is pivotal in the pathogenesis of atherosclerosis and cardiovascular disease in patients with MHD or ESRD [5–7]. Thus, early detection of ED is clinically important because it allows physicians to provide specific clinical interventions to reduce the risk of atherosclerosis in this patient population [7].

Lipoprotein(a) (Lp(a)) is a low-density lipoprotein-like particle with an apolipoprotein(a) (apo(a)) tail covalently bound to it and is the major lipoprotein carrier of oxidized phospholipids [8–10]. Lp(a) has pro-atherogenic, proinflammatory, and prothrombotic properties, and plays a causal role in the development of atherosclerotic cardiovascular disease (ASCVD) [8–10]. It has been proposed that Lp(a) contributes to ASCVD through several key processes, including ED [8,10]. Several in vitro studies have demonstrated that LP(a) can induce inflammation and oxidative stress in the endothelium, leading to ED [10–15]. Particularly, elevated serum Lp(a) has been reported in patients with MHD or ESRD [16–19]. Several studies have shown that Lp(a) is an independent risk factor for ASCVD or mortality in patients with MHD [20–23]. Of note, patients with MHD or ESRD present ED as a pro-atherogenic event [6,7,24]. However, no study has investigated the association between serum Lp(a) and ED in patients with MHD. Positive results from such a study may prove that Lp(a) is a biomarker for the early detection of ED in these patients.

In this study, we investigated the association between Lp(a) and endothelial function in patients with MHD. Moreover, we explored whether Lp(a) could serve as a biomarker of ED in these individuals. We determined the vascular reactivity index (VRI) by digital thermal monitoring to serve as a measure of endothelial function [25–27].

Materials and methods

Ethics statement

This study was approved by the Research Ethics Committee of the Hualien Tzu Chi Hospital (approval No: IRB108-219-A). The study procedures followed the principles of the Declaration of Helsinki, and written informed consent was obtained from all participants. The present study is part of a larger research project, all components of which were conducted under the same ethics approval number.

Participants

All MHD patients were screened for enrollment between 1 May 2022 and 31 August 2022. Patients aged >20 years who received standard 4-h hemodialysis using bicarbonate dialysate (FX class dialyzer, Fresenius Medical Care, Bad Homburg, Germany) three times a week and for more than six months were screened for enrollment. Patients who had a limb amputated, acute infection, malignancy, liver cirrhosis, chronic obstructive lung disease, and acute heart failure, or who were bedridden, were excluded. Finally, 123 patients with MHD were enrolled in this study.

Measurements of basic clinical characteristics

Medical records were examined to collect data on the patient’s basic clinical characteristics. These data included age, sex, height, body weight, body mass index, systolic blood pressure, diastolic blood pressure, length of hemodialysis treatment, and medical history. The body mass index was calculated as post-hemodialysis body weight divided by height squared [26]. The medical history included diabetes mellitus, hypertension, and drugs taken, such as angiotensin receptor blockers, β-blockers, calcium channel blockers, α-adrenergic blockers, statins, and fibrates.

Measurements of biochemical variables

Before initiating hemodialysis, a sample of approximately 5 mL of blood was taken from each patient. Approximately 0.5 mL of blood was used to determine the blood cell count by an autoanalyzer (Sysmex SP-1000i, Sysmex America, Mundelein, IL). The remaining blood sample was immediately centrifuged at 3000 × g for 10 min to obtain serum for further biochemical analyses, including albumin, total cholesterol, triglyceride, glucose, alkaline phosphatase (ALP), blood urea nitrogen, creatinine, total calcium, and phosphorus, using an autoanalyzer (Siemens Advia 1800; Siemens Healthcare, Henkestr, Erlangen, Germany) [26,27]. The fractional clearance index for urea (Kt/V) and the urea reduction ratio were obtained to quantify the clearance of dialysis, using a formal single-compartment dialysis urea kinetic model. The serum levels of intact parathyroid hormone (iPTH) (IBL International, Hamburg, Germany) and Lp(a) [28] (Abcam, Cambridge, MA) were quantified using enzyme-linked immunosorbent assay kits according to the manufacturer’s instructions.

Measurements of endothelial function

Before measurements, the patients fasted overnight and abstained from using tobacco, alcohol, caffeine, and vasoactive medications. The VRI, a reliable index of endothelial function, was measured using a digital thermal monitoring device (VENDYS-II; Endothelix, Inc., Houston, TX) with methods that were described in detail in previous studies [25–27,29]. We classified patients as having good, intermediate, or poor VRI if VRI was ≥ 2.0, 1.0 to < 1.9, or 0.0 to < 1.0, respectively, as suggested by previous investigators [25–27,29].

Statistical analysis

All variables used in the analysis were fully available for all included patients, resulting in a dataset with 100% complete data. The Kolmogorov-Smirnov test was used to determine if continuous variables had a normal distribution. Continuous variables are presented as mean ± standard deviation (SD) or median with interquartile ranges (IQR) when data were normally and non-normally distributed, respectively. Comparing continuous variables (hemodialysis duration, triglyceride, glucose, ALP, iPTH, and Lp(a)) across groups (poor, intermediate, and good VRI) was performed using either one-way analysis of variance in the cases of normal data distribution or Kruskal-Wallis analysis in the cases of non-normal data distribution. The chi-square test was used to compare the categorical variables, which are presented as numbers and percentages. To ensure normality for additional analysis, logarithmically transformed values of duration of MHD, triglyceride, glucose, ALP, iPTH, and Lp(a) were obtained. The association between VRI and the clinical or biochemical variables was investigated using simple and multivariate linear stepwise regression analysis. Univariate and multivariate logistic regression analyses were performed to assess the level of Lp(a) for either vascular reactivity dysfunction (intermediate vascular reactivity and poor vascular reactivity) or poor vascular reactivity. After determining the impact of the Lp(a) level on the vascular reactivity dysfunction or poor vascular reactivity, analyses of the receiver operating characteristic (ROC) were performed, and the areas under the ROC curve (AUC) were computed to calculate the predictive power. SPSS version 19.0 (SPSS, Chicago, IL) was utilized for all data analyses. A p value of 0.05 was considered statistically significant.

Results

Table 1 shows the clinical features of the whole group of MHD patients (n = 123) and among groups of MHD patients with different VRI. Of the 123 MHD patients, 54 (43.9%) had good VRI, 51 (41.5%) had intermediate VRI, and 18 (14.6%) had poor VRI. Serum levels of Lp(a) (p < 0.001) and ALP levels (p = 0.002) were significantly higher in the poor and intermediate VRI groups than in the good VRI group. The three study groups had no significant differences in other biochemical variables. Additionally, there were no significant differences among the three study groups in clinical variables, including age, sex, height, body weight, body mass index, hemodialysis duration, blood pressure, presence of hypertension or diabetes, and use of antihypertensive drugs and anti-lipid drugs.

Table 1.

Clinical features of maintenance hemodialysis patients (n = 123) stratified according to measures of vascular reactivity index.

Characteristics	All patients (n = 123)	Good vascular reactivity (n = 54)	Intermediate vascular reactivity (n = 51)	Poor vascular reactivity (n = 18)	p Value
Age (years)	60.49 ± 13.10	58.80 ± 11.82	62.48 ± 14.68	59.94 ± 11.84	0.350
Female, n (%)	53 (43.1)	25 (46.3)	18 (35.3)	10 (55.6)	0.268
Height (cm)	162.29 ± 7.86	161.96 ± 7.82	163.62 ± 8.39	159.50 ± 5.62	0.148
Pre-HD body weight (kg)	66.70 ± 14.84	66.33 ± 14.57	68.47 ± 15.92	62.76 ± 12.12	0.366
Post-HD body weight (kg)	64.10 ± 14.35	63.83 ± 14.12	65.74 ± 15.30	60.23 ± 11.98	0.372
Body mass index (kg/m2)	25.16 ± 4.50	25.14 ± 4.34	25.38 ± 4.68	24.65 ± 4.67	0.843
HD duration (months)	42.66 (25.92–79.71)	36.60 (24.96–90.36)	44.04 (24.12–94.80)	51.90 (37.29–87.57)	0.571
Systolic blood pressure (mmHg)	148.71 ± 28.22	151.17 ± 26.18	149.12 ± 29.10	140.17 ± 31.48	0.358
Diastolic blood pressure (mmHg)	80.97 ± 15.31	82.13 ± 14.14	80.43 ± 15.19	79.00 ± 19.29	0.718
Vascular reactivity index	1.80 ± 0.72	2.41 ± 0.36	1.62 ± 0.25	0.49 ± 0.17	< 0.001*
Hemoglobin (g/dL)	10.23 ± 1.28	10.25 ± 1.20	10.18 ± 1.38	10.30 ± 1.31	0.931
Albumin (g/dL)	4.17 ± 0.45	4.24 ± 0.46	4.12 ± 0.43	4.14 ± 0.48	0.387
Total cholesterol (mg/dL)	156.44 ± 38.91	153.09 ± 38.46	158.29 ± 39.22	161.22 ± 40.75	0.678
Triglyceride (mg/dL)	127.00 (84.00–205.00)	136.00 (87.00–205.00)	127.00 (84.00–209.00)	111.50 (77.25–175.25)	0.854
Glucose (mg/dL)	127.00 (103.00–180.00)	120.00 (106.00–203.00)	134.00 (104.00–193.00)	114.00 (81.75–147.00)	0.213
Alkaline phosphatase (U/L)	75.00 (59.00–104.00)	66.00 (55.00–95.00)	75.00 (59.00–104.00)	105.50 (77.75–133.00)	0.002*
Blood urea nitrogen (mg/dL)	60.46 ± 14.65	59.70 ± 14.24	63.16 ± 15.25	55.11 ± 13.06	0.118
Creatinine (mg/dL)	9.30 ± 2.10	9.74 ± 2.12	9.10 ± 1.95	8.55 ± 2.28	0.079
Total calcium (mg/dL)	9.07 ± 0.75	9.07 ± 0.72	9.06 ± 0.76	9.12 ± 0.83	0.961
Phosphorus (mg/dL)	4.85 ± 1.41	5.01 ± 1.43	4.75 ± 1.38	4.61 ± 1.47	0.475
Urea reduction rate	0.73 ± 0.05	0.72 ± 0.05	0.73 ± 0.05	0.74 ± 0.05	0.659
Kt/V (Gotch)	1.31 ± 0.20	1.30 ± 0.19	1.32 ± 0.20	1.35 ± 0.22	0.587
Intact parathyroid hormone (pg/mL)	222.70 (98.90–476.30)	273.40 (143.10–467.10)	159.90 (69.60–530.10)	288.30 (95.73–662.45)	0.364
Lipoprotein(a) (mg/L)	205.89 (120.02–351.93)	151.86 (80.01–225.96)	233.29 (157.68–336.78)	388.10 (351.96–428.03)	< 0.001*
Diabetes mellitus, n (%)	62 (50.4)	28 (51.9)	28 (54.9)	6 (33.3)	0.279
Hypertension, n (%)	75 (61.0)	32 (59.3)	33 (64.7)	10 (55.6)	0.746
ARB use, n (%)	54 (43.9)	26 (48.1)	23 (45.1)	5 (27.8)	0.313
β-blocker use, n (%)	33 (26.8)	15 (27.8)	13 (25.5)	5 (27.8)	0.961
CCB use, n (%)	54 (43.9)	21 (38.9)	23 (45.1)	10 (55.6)	0.455
α-adrenergic blockers, n (%)	37 (30.1)	16 (29.6)	18 (35.7)	3 (16.7)	0.332
Statin use, n (%)	39 (31.7)	19 (35.2)	16 (37.4)	4 (22.2)	0.591
Fibrate use, n (%)	17 (13.8)	8 (14.8)	7 (13.7)	2 (11.1)	0.925

Continuous variables with normal distribution are presented as means ± standard deviation and were tested by one-way analysis of variance. Continuous variables with non-normal distribution are presented as medians and interquartile range and were tested by Kruskal-Wallis analysis. Categorical variables are presented as frequency (%) and were analyzed by the Chi-square test. Patients were classified as having good, intermediate, or poor vascular reactivity index (VRI) if VRI was ≥ 2.0, 1.0 to < 2.0, or 0.0 to < 1.0, respectively.

HD: hemodialysis; Kt/V: fractional clearance index for urea; ARB: angiotensin receptor blocker; CCB: calcium channel blocker.

*p < 0.05 was considered statistically significant.

Simple regression analyses revealed that the serum level of logarithmic transformed Lp(a) (log-Lp(a), r = −0.578, p < 0.001) and log-ALP (r = −0.252, p = 0.005) showed a negative correlation with VRI values, while serum creatinine (r = 0.188, p = 0.037) showed a positive correlation with VRI values (Table 2). Further multivariate forward stepwise linear regression analyses indicated that serum log-Lp(a) (β = −0.562, adjusted R² change = 0.328, p < 0.001) and log-ALP (β = −0.212, adjusted R² change = 0.040, p = 0.004) levels were found to be negatively associated with VRI values (Table 2). Figure 1(A,B) show two-dimensional scatter plots of VRI values with serum log-Lp(a) and log-ALP levels, respectively, in MHD patients.

Table 2.

Simple and multivariable regression analyses for the correlation between clinical variables and the level of the vascular reactivity index in patients with maintenance hemodialysis.

Variables	Vascular reactivity index
	Simple regression		Multivariable regression
	r	p Value	Beta	Adjusted R2 change	p Value
Age (years)	–0.085	0.349	–	–	–
Height (cm)	0.062	0.498	–	–	–
Pre-HD body weight (kg)	0.071	0.438	–	–	–
Post-HD body weight (kg)	0.074	0.415	–	–	–
Body mass index (kg/m2)	0.049	0.590	–	–	–
Log-HD duration (months)	–0.008	0.932	–	–	–
Systolic blood pressure (mmHg)	0.135	0.136	–	–	–
Diastolic blood pressure (mmHg)	0.119	0.192	–	–	–
Hemoglobin (g/dL)	0.031	0.735	–	–	–
Albumin (g/dL)	0.115	0.206	–	–	–
Total cholesterol (mg/dL)	–0.108	0.236	–	–	–
Log-Triglyceride (mg/dL)	0.078	0.390	–	–	–
Log-Glucose (mg/dL)	0.082	0.369	–	–	–
Log-ALP (U/L)	–0.252	0.005*	–0.212	0.040	0.004*
Blood urea nitrogen (mg/dL)	0.091	0.316	–	–	–
Creatinine (mg/dL)	0.188	0.037*	–	–	–
Total calcium (mg/dL)	–0.004	0.962	–	–	–
Phosphorus (mg/dL)	0.134	0.140	–	–	–
Log-iPTH (pg/mL)	–0.040	0.662	–	–	–
Log-Lipoprotein(a) (mg/L)	–0.578	< 0.001*	–0.562	0.328	< 0.001*
Urea reduction rate	–0.119	0.191	–	–	–
Kt/V (Gotch)	–0.135	0.138	–	–	–

HD: hemodialysis; ALP: alkaline phosphatase; iPTH: intact parathyroid hormone; Kt/V: fractional clearance index for urea.

Data of HD duration, triglyceride, glucose, ALP, iPTH, and lipoprotein(a) showed skewed distribution and these data were log-transformed before analysis.

Analysis of data was performed using simple linear regression analyses or multivariable stepwise linear regression analysis (adapted factors were log-ALP, creatinine, and log-lipoprotein(a)).

*p < 0.05 was considered statistically significant.

Figure 1.

Relationships between vascular reactivity index and (A) log-transformed lipoprotein(a) (log-lipoprotein(a)) and (B) log-transformed alkaline phosphatase (log-ALP) among patients with maintenance hemodialysis.

Compared to MHD patients with good VRI, univariate logistic regression analyses revealed that serum Lp(a) level was associated with vascular reactivity dysfunction (intermediate vascular reactivity and poor vascular reactivity) (odds ratio [OR] = 1.005 confidence interval (CI) = 1.002–1.008; p = 0.001) and poor vascular reactivity (OR = 1.006, 95% CI = 1.003–1.010, p < 0.001) (Table 3). Further multivariate logistic regression analyses indicated that serum Lp(a) level was independently associated with vascular reactivity dysfunction (OR = 1.006; 95% confidence interval (CI) = 1.003–1.009; p = 0.001) and poor vascular reactivity index (OR = 1.010, 95% CI = 1.005–1.016, p < 0.001) after full adjustment for potential confounding factors (Model 3, Table 3). Multivariate logistic regression analyses with partial adjustment for potential confounding factors revealed similar associations between serum Lp(a) level and vascular reactivity dysfunction or poor vascular reactivity index (Models 1 and 2, Table 3).

Table 3.

Multivariate logistic regression analyses for the association between lipoprotein(a) and vascular reactivity dysfunction (intermediate vascular reactivity and poor vascular reactivity) and poor vascular reactivity in patients with maintenance hemodialysis.

Model	Lipoprotein(a) (per 1 mg/L of increase) for vascular reactivity dysfunction		Lipoprotein(a) (per 1 mg/L of increase) for poor vascular reactivity
	OR (95% CI)	p Value	OR (95% CI)	p Value
Crude model	1.005 (1.002–1.008)	0.001*	1.006 (1.003–1.010)	< 0.001*
Model 1	1.006 (1.003–1.009)	< 0.001*	1.007 (1.003–1.011)	< 0.001*
Model 2	1.006 (1.003–1.009)	< 0.001*	1.007 (1.003–1.011)	< 0.001*
Model 3	1.006 (1.003–1.009)	0.001*	1.010 (1.005–1.016)	< 0.001*

Kt/V: fractional clearance index for urea; OR: odds ratio; CI: confidence interval.

Model 1: Adjusted for demographic and clinical covariates (age, sex, body mass index, diabetes mellitus, hypertension). Model 2: Further adjusted for dialysis-related parameters (hemodialysis duration, systolic blood pressure, diastolic blood pressure, urea reduction rate, Kt/V) with model 1. Model 3: Fully adjusted model including laboratory values (hemoglobin, blood urea nitrogen, creatinine, albumin, total cholesterol, triglyceride, glucose, alkaline phosphatase, calcium, phosphorus, intact parathyroid hormone) with model 2.

*p < 0.05 was considered statistically significant.

Analyses of ROC showed that the AUC was 0.754 (95% CI = 0.668–0.827, p < 0.001) for predicting vascular reactivity dysfunction by Lp(a), and the AUC was 0.853 (95% CI = 0.778–0.910, p < 0.001) for predicting poor vascular reactivity by Lp(a) (Figure 2 and Table 4). The cutoff values of Lp(a) for predicting vascular reactivity dysfunction and poor vascular reactivity were 183.97 and 318.98 (mg/L), respectively.

Figure 2.

Receiver operating characteristic (ROC) curves for predicting (A) vascular reactivity dysfunction and (B) poor vascular reactivity using lipoprotein(a) as a predictor in patients with maintenance hemodialysis. AUC: area under the ROC curve.

Table 4.

Diagnostic value of lipoprotein(a) on vascular reactivity dysfunction (intermediate vascular reactivity and poor vascular reactivity) or poor vascular reactivity in patients with maintenance hemodialysis.

Vascular reactivity dysfunction
	AUC (95% CI)	p value	Cutoff	Sen (%)	Spe (%)	PPV (%)	NPV (%)
Lipoprotein(a) (mg/L)	0.754 (0.668–0.827)	< 0.001	183.97	72.5	70.4	75.8	66.7
Poor vascular reactivity
	AUC (95% CI)	p value	Cutoff	Sen (%)	Spe (%)	PPV (%)	NPV (%)
Lipoprotein(a) (mg/L)	0.853 (0.778–0.910)	< 0.001	318.98	88.9	82.9	47.1	97.8

Analyses of the receiver operating characteristic were performed, and the areas under the curve (AUC) were computed to calculate the predictive power. CI, confidence interval; Sen, sensitivity; Spe, specificity; PPV, positive predictive value; NPV, negative predictive value.

Discussion

Limited research has been conducted on the association between serum Lp(a) and ED in patients with MHD. In this study, our analyses indicated that the VRI determined using digital thermal monitoring was negatively associated with Lp(a) serum levels. Additionally, higher serum levels of Lp(a) were independently and significantly associated with vascular reactivity dysfunction or poor vascular reactivity in MHD patients after controlling for potential confounding variables. Furthermore, serum levels of Lp(a) had good or excellent AUC values for predicting vascular reactivity dysfunction and poor vascular reactivity. Collectively, these findings led us to hypothesize that Lp(a) may have a detrimental impact on endothelial function in MHD patients, and that it may serve as a potential biomarker for ED in these patients.

ED is a common comorbid condition in patients with MHD or ESRD [5–7]. ED is characterized by reduced nitric oxide bioavailability with or without an imbalance between endothelium-derived contracting and relaxing factors associated with a pro-inflammatory and prothrombotic status [4]. Several pathogenetic mechanisms, including inflammation and oxidative stress, are involved in producing low nitric oxide bioavailability in the endothelium of patients with MHD or ESRD [4–7]. To this end, similar to other lipoproteins, Lp(a) is also susceptible to oxidative modifications, leading to extensive formation of proinflammatory oxidized phospholipids, oxysterols, oxidized lipid-protein adducts in Lp(a) particles [9,10]. As a result, the endothelium becomes a preferential target of inflammation and oxidative stress caused by circulating Lp(a). Several in vitro studies have demonstrated that exposure of endothelial cells to Lp(a) or its oxidized phospholipid content produced inflammation [12,13] and oxidative stress [13,14], leading to ED. These observations confer a theoretical explanation for the relationship between serum Lp(a) and ED observed in this study.

The pro-atherogenic property of Lp(a) has been well recognized [8–10]. Atherosclerosis is a major comorbidity and contributes to increased cardiovascular mortality in patients with MHD [2,3]. As such, several studies have shown that Lp(a) is an independent risk factor for ASCVD or mortality in patients with MHD. Liang et al. [20] conducted a retrospective cohort study and demonstrated an independent and positive association between serum Lp(a) levels and the risk of ASCVD in MHD patients. Cressman et al. [21] performed a prospective study and showed that serum Lp(a) is an independent risk factor for clinical events attributed to ASCVD in MHD patients. Koda et al. [22] conducted a prospective follow-up study and reported that serum Lp(a) is an independent risk factor for atherosclerotic cardiovascular death in MHD patients. ED is one of the earliest symptoms of atherosclerosis, and it plays an important role in the pathogenesis of ASCVD in patients with MHD or ESRD [5–7]. Thus, our findings regarding the association between serum Lp(a) and ED provide additional evidence to support the notion that serum Lp(a) is an independent risk factor for ASCVD in this patient population.

MHD patients have been shown to have increased Lp(a) levels compared with healthy controls [17–19]. For example, Webb et al. [17] reported that their MHD patients and control subjects had medium plasma Lp(a) levels of 174 and 94 mg/L, respectively. Hirata et al. [18] showed that their MHD patients and control subjects had mean plasma Lp(a) levels of 264 and 149 mg/L, respectively. This study’s findings regarding the medium serum Lp(a) levels in the intermediate and poor VRI groups (233 and 388 mg/L) generally agree with these observations. Recent expert recommendations [30] suggest that clinicians should consider using risk thresholds with ‘grey’ zones (e.g. 300–500 mg/L) to either rule-in (≥500 mg/L) or rule-out (<300 mg/L) the risk of ASCVD in a non-specified population. Clearly, this recommendation can also be applied to the finding regarding the assessment of the ED risk by serum Lp(a). In our cohort of MHD patients, we identified serum Lp(a) thresholds of 183.97 mg/L and 318.98 mg/L for predicting vascular reactivity dysfunction and poor vascular reactivity, respectively. These values align with, but are slightly lower than, the internationally suggested “grey zone” thresholds for cardiovascular risk, which typically range from 300 to 500 mg/L, with ≥500 mg/L considered high risk in the general population [30]. The slightly lower cutoffs observed in our study may reflect the heightened vascular vulnerability in MHD patients, whose endothelial function is already compromised due to uremia, oxidative stress, and chronic inflammation. Importantly, ED represents an earlier stage of arterial damage compared to overt cardiovascular events, which may explain why Lp(a) levels associated with ED in this study are lower than those associated with clinical cardiovascular outcomes in broader populations. These findings suggest that even moderate elevations in Lp(a) could be clinically relevant for early vascular injury detection in the MHD population. Incorporating Lp(a) levels into multimodal cardiovascular risk assessment frameworks, especially when combined with endothelial function measurements such as VRI, may allow earlier identification of high-risk individuals before irreversible atherosclerotic damage occurs. Future studies are needed to determine whether risk thresholds for Lp(a) should be lowered specifically for ESRD populations and to explore whether Lp(a)-lowering strategies improve endothelial or cardiovascular outcomes in these patients. The circulating Lp(a) levels in an individual are primarily determined by genetics but are subject to modification by several non-genetic factors [8,10,11]. It has been shown [31,32] that increases in Lp(a) levels start early during the course of chronic kidney disease and become more pronounced with increased severity of the disease. Studies [32,33] have demonstrated a rapid decrease of Lp(a) levels after renal transplantation, which provides support for a metabolic role of the kidney in Lp(a) catabolism.

Although our study highlights serum Lp(a) as a potential biomarker for ED in MHD patients, it also raises the question of how clinicians should respond when elevated Lp(a) levels are detected in this population. Currently, there are no specific guideline-based interventions targeting Lp(a) in dialysis patients. However, Lp(a)-selective apheresis has been used in high-risk individuals with progressive cardiovascular disease and extremely elevated Lp(a) levels, though its use in MHD patients remains limited due to resource intensity and access issues. Emerging therapies, such as pelacarsen (an antisense oligonucleotide targeting apo(a) mRNA), have shown significant Lp(a)-lowering effects in early-phase trials [34,35] and may hold promise for MHD patients in the future. While these therapies are not yet approved or tested in the dialysis population, their development underscores the importance of identifying high-risk patients, such as those with elevated Lp(a), to prioritize early intervention and monitoring. Future clinical trials are warranted to determine the efficacy and safety of such Lp(a)-lowering strategies in patients with ESRD on dialysis.

Based on our findings, we propose a clinical flowchart to guide Lp(a) testing and vascular risk stratification in MHD patients (Supplemental Figure 1). Clinicians should consider Lp(a) testing in MHD patients with prior ASCVD, prolonged dialysis, family history of premature ASCVD, or signs of ED such as low VRI. Our data suggest that Lp(a) >184 mg/L may warrant intensified vascular surveillance, while levels ≥319 mg/L are associated with poor vascular reactivity and may prompt early cardiovascular referral and discussion of emerging therapies, such as antisense oligonucleotides like pelacarsen, pending future availability. This personalized approach may support earlier identification and intervention in high-risk dialysis patients. It is important to note that while our proposed algorithm may help guide vascular risk assessment in MHD patients, there is currently no formal guideline, consensus, or evidence-based recommendation specific to Lp(a) management in the hemodialysis population. The existing cardiovascular risk thresholds for Lp(a) (e.g. ≥300–500 mg/L) are derived primarily from general population studies and may not directly apply to patients with MHD. Furthermore, no clinical trial to date has demonstrated that lowering Lp(a) improves cardiovascular outcomes in MHD patients. As such, the thresholds proposed in our study (184 mg/L and 319 mg/L) should be interpreted as exploratory cutoffs, reflecting associations with endothelial dysfunction rather than validated clinical outcomes. Therefore, while our findings suggest a role for Lp(a) in early vascular monitoring, they should not be used as a basis for treatment decisions at this time. Future prospective studies and randomized controlled trials are needed to determine whether interventions targeting Lp(a), such as apheresis or antisense oligonucleotides like pelacarsen, are safe, feasible, and beneficial in dialysis populations. Until then, Lp(a) testing in MHD patients should be considered hypothesis-generating and adjunctive to comprehensive cardiovascular risk assessment.

The Lp(a) thresholds identified in our study (183.97 mg/L and 318.98 mg/L) correspond closely to the “grey zone” of 300–500 mg/L proposed by Kronenberg et al. [30] for cardiovascular risk in the general population. Notably, our lower threshold for predicting vascular dysfunction (184 mg/L) suggests that MHD patients may experience endothelial injury at relatively lower Lp(a) concentrations, potentially due to their heightened pro-inflammatory and pro-oxidative milieu. This highlights the need for dialysis-specific Lp(a) thresholds, which may differ from those established in non-CKD populations. Future studies with a longitudinal design and clinical endpoints are needed to define clinically actionable cutoffs for this group. A comparative table (Supplemental Table 1) summarizes existing serum Lp(a) thresholds used in the general population [30], in patients with CKD not yet on dialysis [36], and in patients undergoing MHD, including the cutoffs derived from our study. Our findings may bridge nephrology and cardiology, underscoring the potential for shared management strategies in high-risk populations such as patients with ESRD on dialysis.

This study also found that serum levels of ALP, a marker of osteoblastic activity, were elevated in the poor and intermediate VRI groups and negatively correlated with VRI values in MHD patients, a finding that has been reported in our previous studies [27]. It has been reported that elevated ALP is associated with increased all-cause or cardiovascular mortality among MHD patients [37–39]. Why ALP levels were negatively correlated with VRI values in our patients remains unclear. However, serum ALP negatively affects endothelium-dependent vasodilation in naïve hypertensive patients [40]. Elevated serum ALP is associated with the risk of atherosclerosis in patients with coronary artery disease [41]. Serum ALP is independently associated with arterial stiffness in participants without cardiovascular disease [42]. The role of serum ALP as an inducer of endothelial dysfunction and vascular calcification in patients with chronic kidney disease has also been suggested [43]. All these observations support the notion that serum ALP can be another biomarker for ED in MHD patients.

In this study, we assessed whether the sample size was adequate to detect meaningful associations in the logistic regression models. A post hoc power analysis was conducted using G*Power version 3.1. Assuming a two-sided α = 0.05, an odds ratio (OR) of 1.009 (based on our observed effect size for poor vascular reactivity), a base event rate of 14.6% (prevalence of poor VRI), and a sample size of 123, the calculated power to detect this effect was approximately 80%. For detecting vascular reactivity dysfunction (combining intermediate and poor VRI, ∼56.1%), the power exceeded 90% for an OR of 1.006. These estimates suggest that the study was adequately powered to detect small to moderate associations between serum Lp(a) levels and endothelial dysfunction outcomes in MHD patients.

While this study provides novel insights into the association between serum Lp(a) and ED in MHD patients, there are several limitations that warrant acknowledgment. First, although statistically significant associations were observed, a formal power calculation for logistic regression was not performed prior to the study. A post hoc power analysis, based on the observed effect sizes (e.g. OR = 1.009 for poor vascular reactivity), sample size (n = 123), and event rate (14.6% for poor VRI), suggests that the study had approximately 80% power to detect this effect at α = 0.05. Nonetheless, future studies with larger sample sizes and a priori power calculations are necessary to ensure adequate statistical power, especially for detecting smaller effects. Second, the cross-sectional and single-center nature of the study limits causal inference and generalizability. To validate our findings, multicenter or international cohort studies involving diverse ethnic and clinical populations are needed. Furthermore, longitudinal studies are strongly encouraged to evaluate temporal relationships between rising Lp(a) levels and the progression of endothelial dysfunction or cardiovascular outcomes in dialysis patients. Third, although the VRI thresholds (≥2.0 for good, 1.0–1.9 for intermediate, <1.0 for poor) have been used in previous studies of MHD [25,26], their specific validation in MHD populations remains limited. The physiological changes associated with hemodialysis, such as fluid shifts, oxidative stress, and vascular remodeling, may necessitate dialysis-specific cutoffs for more accurate risk stratification. Future research should aim to validate or recalibrate these thresholds specifically for the dialysis population to improve clinical utility. Fourth, we acknowledge that key factors such as inflammatory markers (e.g. C-reactive protein), lipid parameters (e.g. LDL-C and HDL-C), and fluid status indicators (e.g. BNP or bioimpedance spectroscopy) were not measured in this study. These unmeasured variables may act as potential confounders in the association between Lp(a) and endothelial dysfunction, and their absence limits the comprehensiveness of our multivariate model. Future studies should aim to incorporate these factors to better elucidate the independent role of Lp(a) in vascular health among hemodialysis patients.

Conclusion

To our knowledge, this is the first study directly linking serum Lp(a) levels to ED assessed via VRI in patients undergoing MHD. We found that the serum Lp(a) level had a negative correlation with the VRI, and it may serve as a potential biomarker for early detection of ED in MHD patients. Future research should validate these Lp(a) cutoffs in multiethnic cohorts, explore whether lowering Lp(a) improves endothelial function in MHD patients, and examine the role of Lp(a) isoforms or oxidized Lp(a) in vascular health.

Funding Statement

This study was supported by grants from Hualien Tzu Chi Hospital, Buddhist Tzu Chi Medical Foundation, Hualien, Taiwan, grant numbers: TCRD-112-070 and TCMF-A 109-07.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All data generated or analyzed during this study are fully included in this article. Any further inquiries may be directed to the corresponding author