Association of beta-2-microglobulin with cardiovascular and all-cause mortality in the general and non-CKD population

Abstract

β-2 microglobulin, a light chain in the major histocompatibility complex Class 1 molecule, is associated with mortality in dialysis or uremic patients. Current evidence on the relationship between beta-2-microglobulin (B2M) and mortality in the general and non-chronic kidney disease (CKD) population are limited and controversial. Data from the nutrition and health examination survey database and the nutrition and health examination survey linked mortality file were used. In total, 10,388 adults who had complete data for B2M were included. Weighted multivariable Cox proportional hazards regression models and regression splines were employed to evaluate the relationship between B2M with mortality. Moreover, subgroup and sensitivity analyses were performed. During a median follow up of 17.9 years (interquartile range 15.2–18.7), 2780 people died, 902 (32%) from cardiovascular disease. Restricted cubic splines showed that B2M is J-shaped nonlinear positively associated with all-cause mortality and cardiovascular disease mortality in the non-CKD and general population. Based on the multivariable adjustment model, the adjusted hazard ratios comparing the highest versus lowest quartile of the distribution of B2M were 2.50 (95% confidence interval: 1.90, 3.28) for all-cause mortality in the general population, 2.58 (95% confidence interval: 1.52, 4.37) for cardiovascular disease mortality in the general population, 2.58 (1.91, 3.49) for all-cause mortality in the non-CKD population and 2.62 (1.52, 4.53) for cardiovascular disease mortality in the non-CKD population. The positive associations between B2M and outcomes remained broadly significant across subgroups and sensitivity analyses. Higher B2M levels were associated with cardiovascular and all-cause mortality in the general and non-CKD population.

1. Introduction

Beta-2-microglobulin (B2M) is an essential component of the primary histocompatibility complex class I molecule that is expressed on the surface of almost all nucleated cells.^[1] B2M is constantly secreted into circulation from cell surfaces or intracellular release and is almost exclusively eliminated from the kidneys.^[2,3] The serum B2M concentrations are highly inversely associated with the glomerular filtration rate. It is elevated in patients with impaired kidney function, particularly those on maintenance dialysis with anuria or low residual renal function. Moreover, It has been reported that serum B2M levels are associated with all-cause mortality in hemodialysis and peritoneal dialysis patients.^[4,5]

B2M also plays a vital role in the immune system given the complex relationship between B2M and CD8 + lymphocytes^[6] as well as nonclassical histocompatibility complex class I like molecules such as CD1, MR1, HLA-E,-F, -G and neonatal Fc receptor^[7–9] that are involved in mucosal immunity, tumor surveillance, immunoglobulin, and albumin homeostasis.^[3] Therefore, the serum β-2 microglobulin concentrations are elevated in many nonrenal conditions, such as autoimmune diseases,^[10] immunodeficiency,^[11] and hematological cancers.^[12,13] Many studies have also reported positive associations between β-2 microglobulin and fibrosis,^[14] peripheral artery conditions.^[15]

Cardiovascular diseases like coronary heart disease and stroke are chronic inflammatory processes of atherosclerosis. ^[16,17] Many studies have reported the correlation between cardiovascular diseases and serum B2M levels. Amighi et al^[18] have demonstrated that serum B2M was independently associated with adverse cardiovascular outcome in patients with prevalent asymptomatic carotid atherosclerosis. In dialysis patients, It has been suggested that the formation of amyloid fibrils in the vessel wall which impairs vessels is associated with B2M.^[19]

However, the precise mechanisms linking B2M with all-cause and cardiovascular disease (CVD) mortality have yet to be fully understood. On the 1 hand, it was suggested that B2M functioned as an initiator of inflammatory responses^[20] or a marker of middle-molecule uremic toxins,^[21] which means B2M could directly affect CVD and all-cause mortality. On the other hand, B2M may function as a surrogate of residual renal clearance, indirectly affecting all-cause and CVD mortality. Several randomized controlled trials on high flux dialysis showed that enhanced dialytic clearance of B2M has not convincingly been translated into improved survival.^[22] Moreover, some studies conducted in patients with renal disease showed inconsistent associations between B2M and CVD mortality.^[23,24] Therefore, it is necessary to evaluate the association of B2M with CVD and all-cause mortality in the general and non-chronic kidney disease (CKD) population.

This study aimed to quantify the relative contribution of B2M to all-cause mortality and cardiovascular disease mortality using data from the national health and nutrition examination survey (NHANES) 1999 to 2004, a prospective, representative cohort of the US population enrolled from 1999 to 2004 and followed up to December 31, 2019.

2. Methods

2.1. Study population

The NHANES is a multistage stratified survey designed to provide a detailed examination of the health and nutritional status of a nationally representative sample of noninstitutionalized individuals in the USA.^[25] The present study used data from 3 NHANES cycles with 31,126 samples included from 1999 to 2004. Participants aged < 20 years (n = 15,794) or with missing information on B2M (n = 2942), other covariates (n = 1983), and deaths (n = 19) were excluded from the analysis. Finally, As shown in (Figure S1, Supplemental Digital Content, https://links.lww.com/MD/I620), 10,388 participants with complete data were enrolled as the general population in the analysis. CKD was defined as an estimated glomerular filtration rate (eGFR) < 60 mL/minutes/1.73 m2 (using the chronic kidney disease epidemiology collaboration equation) and/or urinary albumin: creatinine ratio > 30 mg/g.^[26,27] The protocols for NHANES were approved by the NCHS of the centers for disease control and prevention institutional review board. All participants gave informed consent. This study followed the strengthening the reporting of observational studies in epidemiology statement.

2.2. Assessments of B2M

B2M was measured in serum using a B2M immunoassay (siemens healthcare diagnostics) on an automated multi-channel analyzer, siemens dimension vista 1500 (siemens healthcare diagnostics). The lower and upper detection limits are 0.72 mg/L and 23.0 mg/L, respectively. For study participants with concentrations of B2M in blood below the level of detection, we imputed an amount of 0·509 mg/L, which is the level of detection divided by the square root of 2. We assessed concentrations of B2M in serum categorically with quartiles.

2.3. Outcome assessment

NHANES Public-use linked mortality data were provided for the participants up to 31 December 2019 (https://www.cdc.gov/nchs/data/datalinkage/2019NDI-Linkage-Methods-and-Analytic-Considerations-508.pdf). Referring to the definition of the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) (https://apps.who.int/iris/handle/10665/46642), the primary outcomes in our analysis were death from all causes, CVD (codes I00–I09, I11, I13, I20–I51, and I60–I69). The follow-up period was calculated based on the date when the physical examination (at the baseline) was received and the date of death from the causes mentioned above or the end of the follow-up period (i.e., 31 December 2019). Those who died due to an event other than the mentioned causes were regarded as censored at the date of death.

2.4. Covariates assessment

Demographic and lifestyle habits covariates in the present study included gender (male/female), age (years), race (Mexican American/non-Hispanic White/non-Hispanic Black/other races), household income (<US$20,000 or ≥$20,000 per year), alcohol consumption status(never/former/current), smoking status (never/former/current) and physical activity (sit during the day, stand/walk a lot, light load/climb stairs often, and heavy work/load). The Healthy Eating Index, derived from food frequency questionnaires and scored on a scale from 1 to 100, was categorized as tertiles.^[28]

Blood specimen collections and measurements of blood pressure, body weight, and height were conducted during mobile physical examinations. Body mass index (BMI, kg/m2) was calculated as weight in kilograms divided by height in meters squared and categorized as normal (<25·0 kg/m²), overweight (25·0–29·9 kg/m²), or obese (≥30·0 kg/m²). The laboratory covariates include urine albumin (mg/L), urine creatinine(mg/dL), serum creatinine (mg/dL), total cholesterol (mmol/L), high-density lipoprotein cholesterol (HDL-C, mmol/L), serum globulin (g/dL), C-reactive protein (mg/dL). Serum Cr-based eGFR was estimated using the CKD Epidemiology Collaboration equation.^[26,27] High cholesterol levels were defined as the total-to-HDL cholesterol ratio was more than 5.9.^[29]

Diabetes (yes, no) was defined as ever been told by a doctor that have a diagnosis of diabetes. Hypertension (yes, no) was defined as taking antihypertensive medications, ever been told by a doctor that have a diagnosis of hypertension or having 3 consecutive systolic blood pressure measurements ≥ 140 mm Hg or diastolic blood pressure ≥ 90 mm Hg.^[29] Histories of cardiovascular disease (yes, no) were recorded as ever being told they had any of the following conditions by a health care provider: congestive heart failure, coronary heart disease, angina pectoris, heart attack, or stroke.^[30] Histories of cancer (yes, no) were classified as answered yes to the following question “Have you ever been informed by a health professional or a doctor that you had cancer or malignancy?.”

2.5. Statistical analysis

The characteristics of participants grouped by B2M quartiles are summarized in Table 1. Continuous variates are indicated as mean ± standard deviation or median (range) according to data distribution, while categorical variates presented as frequencies and percentages. The Kruskal–Wallis H test (skewed distribution), 1-way ANOVA test (normal distribution), and χ2 (categorical variables) were employed to evaluate variance among different groups. Survival curves were plotted with the Kaplan–Meier method. Restricted cubic splines with 3 knots located at the 10th, 50th and 90th percentiles of the B2M concentration were used to evaluate linear and nonlinear associations. Multivariable Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for the associations between serum B2M concentration and risk of all-cause and CVD mortality in the general, CKD and non-CKD population. Four models were finally fitted. In Model 1, no covariates were adjusted. Model 2 was adjusted for gender, age, race and household income. Model 3 was additionally adjusted for BMI, Alcohol intake, Smoke, C-reactive protein, Serum globulin, Physical activity, Total-to-HDL cholesterol ratio, eGFR, ACR, and Healthy eating index. The covariates presented in Table 1 were fully adjusted in model 4. Stratified and interaction analyses were conducted according to age (<60 vs ≥ 60 years), sex, self-reported race (non-Hispanic white vs other groups), BMI (<25 vs ≥ 25 kg/ m2), Cholesterol ratio (<5.9 vs ≥5.9), Smoking status (Yes vs No), Drinking status (Yes vs No), history of hypertension, diabetes, CVD, or cancer. Finally, several sensitivity analyses were conducted. First, we excluded participants with histories of CVD and cancer. Second, we repeated the main analysis by changing the categorical variates into continuous variates. A 2-sided P < .05 was considered statistically significant. All analyses were performed using Survey package in R software. (version 4.1.3; R Foundation for Statistical Computing, Vienna, Austria).

Table 1 - Baseline characteristics according to quartile of B2M.

	Total	B2M				P value
		Quartile 1 < 1.63 mg/L	Quartile 2 1.63–1.90 mg/L	Quartile 3 1.90–2.3 mg/L	Quartile 4 ≥ 2.3 mg/L
Age, %						<.001
<60	78.6	97.2	90	75.3	40.8
≥60	21.4	2.8	10.0	24.7	59.2
Gender, %
Male	47.9	44.8	50.4	50.2	45.7	.001
Female	52.1	55.2	49.6	49.8	54.3
Race, %						<.001
Mexican American	7.1	10.1	8.0	5.9	3.4
Non-Hispanic White	72.6	60.3	72.9	78.2	81.9
Non-Hispanic Black	10.3	16.5	8.9	6.9	8.3
Other Race	9.9	13.1	10.2	9.0	6.3
Income, %						<.001
≤US$20 000	18.1	14.0	15.4	17.1	29.1
>US$20 000	81.9	86.0	84.6	82.9	70.9
BMI, %						<.001
Normal weight (<25·0 kg/m²)	33.8	44.3	34.3	27.3	26.9
Overweight (25·0–29·9 kg/m²)	34.8	34.3	35.6	35.3	33.5
Obese (≥30·0 kg/m²)	31.5	21.5	30.1	37.4	39.7
Alcohol intake, %						<.001
Never	11.8	9.8	9.9	11.7	17.7
Ever	14.7	12.5	12.9	14.6	20.5
Current	73.4	77.6	77.2	73.7	61.8
Smoke, %						<.001
Never	50.2	55.2	51.1	49.1	43.6
Ever	25.4	18.8	23.8	27.1	34.4
Current	24.4	26.0	25.1	23.9	22.0
ACR, %						<.001
>30mg/g	9.0	4.8	5.9	8.0	20.4
Total-to-HDL cholesterol ratio%						<.001
≥5.9	11.9	7.4	12.5	14.2	14.2
History of Diabetes, %	6.5	2.9	4.2	6.5	14.8	<.001
History of Hypertension, %	38.5	22.3	31.5	41.5	67.0	<.001
History of CVD, %	8.4	2.2	3.9	8.5	23.4	<.001
History of Cancer, %	8.0	3.7	5.2	9.5	16.0	<.001
Healthy eating index, %						.154
First quartile	33.3	35.8	37.7	36.6	33.3
Second quartile	33.3	33.5	30.9	31.6	32.0
Third quartile	33.3	30.7	31.3	31.8	34.7
Physical activity, %						<.001
Sit during the day	24.6	21.7	22.5	23.7	32.9
Stand/walk a lot	50.9	50.8	50.5	52.3	50.1
Light load/climb stairs often	17.1	19.6	19.0	15.8	12.8
Heavy work/load	7.3	7.9	8.0	8.2	4.2
CRP mg/dL, mean, SD	0.43 (0.80)	0.26 (0.43)	0.35 (0.64)	0.47 (0.70)	0.71 (1.31)	<.001
Serum globulin g/dL, mean, SD	2.99 (0.42)	2.95 (0.38)	2.96 (0.39)	2.99 (0.41)	3.11 (0.51)	<.001
eGFR, mean, SD	98.43 (22.51)	112.96 (16.88)	102.53 (17.11)	95.17 (18.03)	76.64 (23.558)	<.001

Values are weighted mean ± S.D for continuous variables or weighted% for categorical variables.

ACR = albumincreatinine ratio, B2M = beta-2-microglobulin, BMI = body mass index, CRP = C-reactive protein, CVD = cardiovascular disease, eGFR = estimate glomerular filtration rate, HDL = high-density lipoprotein, SD = standard deviation.

3. Results

3.1. Baseline characteristics

Overall, 10,388 participants aged 20 years or older from 3 NHANES cycles were included in this study. (Weighted population, 172,006,053; weighted mean age, 45.8 [16.5] years; 5387 (weighted proportion, 51.9% female). The overall serum B2M concentrations ranged from 0.509 mg/L to 23 mg/L which were equally divided into 4 groups according to the median B2M concentration 1.900 (1.630–2.300) mg/L. Participants who had the highest concentrations of B2M were older, poorer, and more likely to be female, to ever consume alcohol, to never smoke cigarettes, to be overweight and to have less physical activity. Elevated amounts of cholesterol in serum and history of hypertension and diabetes were more prevalent among adults of higher B2M. Besides, samples with higher B2M levels are more prone to have higher values in urinary albumin to creatinine ratio (Table 1). Participants excluded from the current study owing to missing data on covariates were older, of high B2M, and more likely to be women, non-Hispanic black, as shown in (Table S1, Supplemental Digital Content, https://links.lww.com/MD/I621).

3.2. Primary outcomes

Analysis of restricted cubic splines indicated that adjusted HRs were steeper at higher concentrations of B2M in serum than at lower concentrations. Specifically, in the non-CKD population, we observed a J-shaped exposure-response relationship between B2M and all-cause and CVD mortality risk (Fig. 1). Kaplan–Meier curves for survival showed that participants with the highest B2M had a worse prognosis than individuals with the lowest B2M. The survival curve becomes steeper with a higher B2M quartile, as shown in (Figure S2, Supplemental Digital Content, https://links.lww.com/MD/I622). Table 2 shows the association of the quartiles of the serum B2M levels relative to quartile 1 with mortality among the general population both in unadjusted and adjusted models. After adjusting for sociodemographic characteristics and other covariates, including age, gender, race, income, BMI, alcohol intake, smoke, C-reactive protein, serum globulin, physical activity, total-to-HDL cholesterol ratio, eGFR, ACR, Healthy eating index, and history of comorbidities, the hazards rations when adults of high B2M were compared with adults of low B2M were 2.50 (95% CI: 1.90–3.28) for all-cause mortality and 2.58 (95% CI: 1.52–4.37) for CVD mortality in the general population. The hazard ratios for all-cause mortality increased with a higher B2M quartile. The hazard ratios without adjustment were larger. In the non-CKD population, when full adjustments except for eGFR were made for Model 3, the adjusted hazards were 2.58 (95% CI: 1.91–3.49) for all-cause mortality and 2.62 (95% CI: 1.52–4.53) for CVD mortality at the highest B2M concentration status (Table 3). When we further adjusted for eGFR in Model 3, the adjusted hazards were 2.21 (95% CI: 1.59–3.08) for all-cause mortality and 2.25 (95% CI: 1.23–4.09) for CVD mortality at the highest B2M concentration status. In contrast, HRs for all-cause mortality and CVD mortality at the highest B2M concentration status among the CKD population were 3.61 (95% CI: 1.79–7.27) and 3.93 (95% CI: 1.42–10.9) before eGFR was adjusted. When eGFR was additionally adjusted, HRs for all-cause mortality and CVD mortality at the highest B2M concentration status among the CKD population were 3.08 (95% CI: 1.54–6.16) and 2.93 (95% CI: 1.03–8.27).

Table 2 - The association of B2M with CVD and all-cause mortality in the general population.

	Deaths, No. (%)	HR (95% CI), P value
		Model 1	Model 2	Model 3	Model 4
All-cause mortality
Q1	166 (6.15)	ref	ref	ref	ref
Q2	328 (12.5)	1.82 (1.40, 2.37)	1.37 (1.05, 1.79)	1.21 (0.92, 1.59)	1.20 (0.93, 1.57)
Q3	663 (26.2)	3.81 (2.90, 5.00)	1.96 (1.50, 2.58)	1.61 (1.20, 2.14)	1.57 (1.19, 2.08)
Q4	1623 (63.8)	14.54 (11.42,18.53)	4.35 (3.38, 5.61)	2.66 (2.02, 3.51)	2.50 (1.90, 3.28)
Cardiovascular disease mortality
Q1	41 (1.52%)	ref	ref	ref	ref
Q2	99 (3.78%)	2.62(1.54,4.46)	1.84 (1.07, 3.17)	1.58 (0.90, 2.77)	1.56 (1.56 (0.91, 2.69)
Q3	183 (7.24%)	4.01 (2.38, 6.73)	1.81 (1.08, 3.04)	1.39 (0.79, 2.43)	1.36 (1.36 (0.79, 2.35)
Q4	579 (22.8%)	21.72 (13.67,34.50)	5.42 (3.43, 8.57)	2.83 (1.66, 4.84)	2.58 (2.58 (1.52, 4.37)

Model 1: no adjustment. Model 2: Model 1 + Age, Gender, Race, Income. Model 3: Model 2 + BMI, Alcohol intake, Smoke, CRP, Serum globulin, Physical activity, Total-to-HDL cholesterol ratio, eGFR, ACR, Healthy eating index. Model 4: Model 3 + History of Diabetes, History of Hypertension, History of CVD, History of Cancer.

ACR = albumin: creatinine ratio, B2M = beta-2-microglobulin, BMI = body mass index, CI = confidence interval, CVD = cardiovascular disease, CRP = C-reactive protein, eGFR = estimate glomerular filtration rate, HR = hazard ratios, HDL = high-density lipoprotein.

Table 3 - The association of B2M with CVD and all-cause mortality in the non-CKD population (non-CKD, n = 8590).

	HR (95% CI), P value
	Deaths, No. (%)	Model 1	Model 2	Model 3
All-cause mortality
Q1	148 (5.79)	ref	ref	ref
Q2	277 (11.4)	1.79 (1.34, 2.38)	1.35 (1.01, 1.79)	1.24 (0.95, 1.63)
Q3	515 (23.5)	3.59 (2.71, 4.76)	1.88 (1.41, 2.51)	1.65 (1.25, 2.18)
Q4	719 (51.0)	10.36 (7.97, 13.46)	3.53 (2.58, 4.83)	2.58 (1.91, 3.49)
CVD mortality
Q1	33 (1.29)	ref	ref	ref
Q2	83 (3.42)	2.63 (1.45, 4.76)	1.77 (0.97, 3.24)	1.61 (0.90, 2.87)
Q3	133 (6.06)	3.65 (2.09, 6.40)	1.56 (0.91, 2.68)	1.33 (0.77, 2.29)
Q4	232 (16.4)	14.64 (8.85, 24.20)	3.82 (2.30, 6.33)	2.62 (1.52, 4.53)

Model 1: no adjustment. Model 2: Model 1 + Age, Gender, Race, Income. Model 3: Model 2 + BMI, Alcohol intake, Smoke, CRP, Serum globulin, Physical activity, Total-to-HDL cholesterol ratio, Healthy eating index, History of Cancer, History of Diabetes, History of Hypertension, History of CVD.

BMI = body mass index, B2M = beta-2-microglobulin, CI = confidence interval, CKD = chronic kidney disease, CVD = cardiovascular disease, CRP = C-reactive protein, eGFR = estimate glomerular filtration rate, HR = hazard ratios, HDL = high-density lipoprotein, UACR = urine albumin: creatinine ratio.

Figure 1.: Dose-response curves for B2M in blood and mortality. B2M and all-cause mortality in the general population (A) and non-CKD population, (B) B2M and cardiovascular mortality in the general population, (C) and non-CKD population, and (D) B2M = beta-2-microglobulin, CKD = chronic kidney disease.

We further investigated the stratified, interaction, and sensitivity analyses between B2M and all-cause mortality as well as CVD mortality of model 4. Stratified and interaction analyses suggest that the associations of B2M and all-cause mortality differ by age, chronic comorbidities and the associations of B2M and CVD mortality differ by alcohol consumption and history of hypertension. Relatively stronger associations between B2M and all-cause mortality were observed among younger adults, participants with non-hypertension and non-heart disease (Table 4). The increased risk for CVD mortality was more pronounced in drinker and people with no-hypertension (Table 5). For sensitivity analysis, results were not substantially changed when we further adjusted for substituting the categorical variates by continuous variates, as shown in (Table S2, Supplemental Digital Content, https://links.lww.com/MD/I623). We repeated the Cox proportional hazards models excluding data of participants with histories of CVD and cancer, as shown in (Table S3, Supplemental Digital Content, https://links.lww.com/MD/I624). The results were consistent with those of the main analysis.

Table 4 - Subgroup analyses of the associations between B2M and all-cause mortality.

Subgroups	B2M				P value for interaction
	Q1	Q2	Q3	Q4
Age					.016
<60	ref	1.12 (0.84,1.49)	1.41 (0.99,2.00)	2.20 (1.55,3.13)
≥60	ref	1.01 (0.76,1.37)	1.26 (0.95,1.68)	2.01 (1.51,2.68)
Alcohol drinking status					.105
Abstainer	ref	0.95 (0.52,1.69)	1.38 (0.81,2.37)	1.79 (1.03,3.09)
Drinker	ref	1.30 (0.97,1.74)	1.60 (1.19,2.15)	2.86 (2.12,3.87)
Race					.231
Non-Hispanic white	ref	1.12 (0.76,1.66)	1.47 (0.99,2.20)	2.38 (1.64,3.43)
Others	ref	1.44 (1.10,1.89)	1.92 (1.38,2.68)	2.80 (2.01,3.91)
Smoking					.096
Yes	ref	1.22 (0.85,1.75)	1.70 (1.21,2.37)	2.60 (1.91,3.54)
No	ref	1.08 (0.72,1.63)	1.23 (0.83,1.82)	2.11 (1.33,3.37)
BMI					.052
<25 kg/m2	ref	1.18 (0.77,1.79)	1.58 (1.10,2.26)	2.65 (1.87,3.77)
≧25 kg/m2	ref	1.15 (0.84,1.58)	1.49 (1.05,2.12)	2.37 (1.67,3.34)
Diabetes					.291
Yes	ref	1.42 (0.70,2.88)	1.81 (0.86,3.78)	3.19 (1.48,6.84)
No	ref	1.13 (0.86,1.49)	1.43 (1.05,1.93)	2.26 (1.68,3.03)
Hypertension					<.001
Yes	ref	1.00 (0.70,1.43)	1.28 (0.87,1.88)	2.06 (1.48, 2.88)
No	ref	1.26 (0.88,1.80)	1.61 (1.17,2.21)	2.69 (1.86,3.90)
CVD					.025
Yes	ref	1.04 (0.51,2.11)	1.23 (0.65,2.30)	1.78 (0.92,3.45)
No	ref	1.19 (0.89,1.58)	1.51 (1.14,2.01)	2.56 (1.96,3.36)
Gender					.726
Male	ref	1.18 (0.75,1.85)	1.70 (1.07,2.70)	2.86 (1.88,4.35)
Female	ref	1.22 (0.84,1.77)	1.41 (1.03,1.92)	2.14 (1.44,3.17)
Cancer					.144
Yes	ref	0.90 (0.36,2.22)	1.36 (0.56,3.30)	1.82 (0.68,4.82)
No	ref	1.23 (0.92,1.65)	1.52 (1.14,2.03)	2.57 (1.94,3.42)
Cholesterol ratio					.051
≥5.9	ref	1.0 (0.48,2.08)	1.31 (0.61,2.78)	1.51 (0.70,3.25)
<5.9	ref	1.21 (0.94,1.55)	1.56 (1.22,1.98)	2.64 (2.04,3.43)
CKD					.997
Yes	ref	1.43 (0.74,2.73)	2.07 (1.05,4.06)	3.47 (1.74,6.91)
No	ref	1.23 (0.94,1.62)	1.62 (1.23,2.13)	2.59 (1.93,3.50)

Adjusted covariates: Age, Gender, Race, Income, BMI, Alcohol intake, Smoke, CRP, Serum globulin, Physical activity, Total-to-HDL cholesterol ratio, eGFR, ACR, Healthy eating index, History of Diabetes, History of Hypertension, History of CVD, History of Cancer. The variable used for stratification was not included in the given model. The interaction was tested using a continuous beta-2-microglobulin term and the exposure of interest using a Wald test for dichotomous variables. Statistically significant results are shown in bold.

ACR = albumin: creatinine ratio, BMI = body mass index, B2M = beta-2-microglobulin, CVD = Cardiovascular disease, CKD = chronic kidney disease, CRP = C-reactive protein, eGFR = estimate glomerular filtration rate, HDL = high-density lipoprotein.

Table 5 - Subgroup analyses of the associations between B2M and cardiovascular mortality.

Subgroups	B2M				P value for interaction
	Q1	Q2	Q3	Q4
Age					.994
<60	ref	1.43 (0.78,2.61)	0.87 (0.42,1.81)	1.69 (0.76,3.77)
≥60	ref	1.64 (0.82,3.25)	1.58 (0.85,2.96)	2.92 (1.57,5.44)
Alcohol drinking status					.005
Abstainer	ref	0.86 (0.36, 2.03)	0.95 (0.42,2.15)	1.39 (0.60,3.24)
Drinker	ref	1.96 (1.13,3.40)	1.56 (0.87,2.77)	3.30 (1.87,5.81)
Race					.209
Non-Hispanic white	ref	1.63 (0.73,3.63)	1.29 (0.58,2.86)	2.61 (1.23,5.52)
Others	ref	1.48 (0.90, 2.45)	1.72 (0.97, 3.07)	2.68 (1.50,4.78)
Smoking					.135
Yes	ref	1.28 (0.66,2.50)	1.21 (0.67,2.20)	2.09 (1.13,3.84)
No	ref	2.03 (0.70,5.88)	1.55 (0.56,4.24)	3.42 (1.18,9.86)
BMI					.332
<25 kg/m2	ref	1.96 (0.80,4.77)	1.53 (0.74,3.17)	2.98 (1.48,6.02)
≧25 kg/m2	ref	1.37 (0.75, 2.49)	1.25 (0.69,2.28)	2.44 (1.32,4.53)
Diabetes					.064
Yes	ref	1.21 (0.42,3.51)	1.03 (0.32,3.29)	2.21 (0.74,6.60)
No	ref	1.49 (0.86,2.59)	1.22 (0.67,2.24)	2.29 (1.28,4.10)
Hypertension					.014
Yes	ref	1.02 (0.55,1.89)	0.92 (0.53,1.59)	1.90 (1.11,3.24)
No	ref	2.32 (0.92,5.83)	2.00 (0.78, 5.10)	2.98 (1.20,7.42)
CVD					.090
Yes	ref	1.56 (0.51,4.79)	1.51 (0.56,4.07)	2.05 (0.81,5.20)
No	ref	1.44 (0.85,2.46)	1.09 (0.63,1.89)	2.55 (1.51,4.32)
Gender					.565
Male	ref	1.36 (0.67,2.75)	1.20 (0.57,2.53)	2.50 (1.27,4.93)
Female	ref	1.83 (0.81,4.13)	1.62 (0.73,3.60)	2.71 (1.19, 6.17)
Cancer					.351
Yes	ref	1.15 (0.30,4.44)	1.11 (0.34,3.62)	1.24 (0.37,4.10)
No	ref	1.55 (0.91,2.63)	1.28 (0.74,2.21)	2.73 (1.60,4.65)
Cholesterol ratio					.165
≥5.9	ref	1.50 (0.50,4.48)	1.39 (0.44,4.39)	1.39 (0.36,5.45)
<5.9	ref	1.49 (0.83,2.68)	1.25 (0.70,2.22)	2.62 (1.51,4.55)
CKD					.550
Yes	ref	1.73 (0.61,4.86)	2.09 (0.82,5.34)	3.81 (1.38,10.57)
No	ref	1.56 (0.87,2.79)	1.29 (0.75,2.21)	2.58 (1.50,4.41)

Adjusted covariates: Age, Gender, Race, Income, BMI, Alcohol intake, Smoke, CRP, Serum globulin, Physical activity, Total-to-HDL cholesterol ratio, eGFR, ACR, Healthy eating index, History of Diabetes, History of Hypertension, History of CVD, History of Cancer. The variable used for stratification was not included in the given model. The interaction was tested using a quartile 4 beta-2-microglobulin term and the exposure of interest using a Wald test for dichotomous variables. Statistically significant results are shown in bold.

ACR = albumin: creatinine ratio, BMI = body mass index, B2M = beta-2-microglobulin, CVD = cardiovascular disease, CKD = chronic kidney disease, CRP = C-reactive protein, eGFR = estimate glomerular filtration rate, HDL = high-density lipoprotein.

4. Discussion

In this large retrospective cohort of US adults, we observed an increased risk of all-cause and CVD mortality at the higher B2M levels among both general and non-CKD population. These associations are still significant after adjusting for potential confounders. The subgroup analysis and interaction test showed that this association was similar across different population settings. The results suggest that the associations of B2M with all-cause and CVD mortality were independent and did not differ by renal function. We also observed a J-shaped exposure-response relationship between the B2M and all-cause mortality risk, which partially explains why the associations were more pronounced at the highest B2M quartile.

Elevated serum B2M in mortality has been discussed before. Several studies examined the association of B2M with mortality in peritoneal dialysis,^[24] hemodialysis,^[4]or kidney transplantation patients,^[31] given that B2M is intimately related with renal function. These studies showed that B2M is a predictor of all-cause and infectious disease mortality in people with renal impairment.^[32] The Chronic Renal Insufficiency Cohort Study found that B2M was an independent predictor of all-cause mortality among persons with CKD.^[33] A retrospective observational cohort study with 29 all-cause deaths among 312 patients with advanced CKD displayed that compared to the lowest tertile, the highest tertile group for B2M had strong association with all-cause death after multivariate adjustment.^[34] Previous reports also illustrated the link between B2M and increased risk of various physical disorders, such as diabetes,^[35] cancer,^[36] and chronic obstructive pulmonary disease.^[37] Therefore, the strong mortality risk of B2M could be explained by other nonfiltration factors. Moreover, the serum levels of B2M varied largely between the CKD and non-CKD population. Examining the link between B2M and mortality in the general and non-CKD population is necessary. The Atherosclerosis Risk in Communities Study with 1425 all-cause deaths among 9988 general population demonstrated that a higher level of B2M was associated with a higher risk for all-cause mortality (Q5c: aHRs, 3.01; 95% CI: 2.41–3.75) after multivariate adjustment.^[38] A prospective cohort study showed that B2M had strong associations with all-cause mortality and cardiovascular mortality among 5632 participants with baseline estimated glomerular filtration rate ≥ 60ml/min/1.73m².^[39] In present study, we adjusted for renal function and other non-filtration factors and showed that B2M was independently associated with all-cause mortality in the general population (Q4 vs Q1:HR, 2.50; 95% CI: 1.90–3.28). A similar trend could be found in the non-CKD population when we adjusted for renal function (Q4 vs Q1:HR, 2.58; 95% CI: 1.91–3.49).

It is already known that an elevated concentration of serum B2M acts as a potential risk factor for the development of dialysis-related amyloidosis.^[40] However, it is still unclear whether B2M is an active player in vascular damage or a surrogate of residual renal clearance. In Rist study, serum B2M was not associated with cardiovascular outcomes after further adjustment for eGFR.^[41] On the contrary, Wilson study found that elevated B2M levels correlated with severity of the peripheral arterial disease.^[19] A single-center acute ischemic stroke cohort study found that patients with higher levels of serum B2M had a higher risk of acute ischemic stroke recurrence than patients with lower levels of B2M.^[42] What’ more, B2M was found to be independently associated with adverse cardiovascular outcome in patients with prevalent asymptomatic carotid atherosclerosis.^[18] With adverse effects of B2M on vascular structure, it is reasonable to hypothesize that B2M is directly associated with cardiovascular mortality risk, which had been evaluated by several studies,^[43,44] In our analysis, we found B2M quartile 4 (HR = 2.58, 95% CI: 1.52–4.37 in the general population; HR = 2.62, 95% CI: 1.52–4.53 in the non-CKD population) was significantly associated with CVD mortality after adjusted for renal function and other non-filtration factors.

In stratified and interaction analysis, we found that the HR of the B2M for all-cause mortality was higher in the younger adults, participants with non-hypertension and non-heart disease. On the contrary, the association between B2M and CVD mortality varied by history of hypertension and drinking status.

Evidence has accumulated that the pathogenic pathways of hypertension is complex with a significant diversity and variability of the mechanisms.^[45] Some of these pathways include regulation of sodium excretion by the kidney,^[46] activity of the central and sympathetic nervous systems and contractile processes in the vasculature,^[47] immune and inflammatory response activation.^[48] Previous studies also reported the relationship between B2M and diseases related to immunity and inflammation.^[10,49,50] B2M was also found to trigger an inflammatory process in vivo.^[51] However, the influences of hypertension on mortality are unlikely to be dominated by the same pathways as for B2M. Hence, when we adjust the confounding factors for B2M in the participants with hypertension group, the association between B2M and mortality may be limited. Trends were similar for the aging and cardiovascular disease group. As for the interaction effect between B2M and drinking status, the serum levels of B2M may be affected by drinking, while the direct evidence is lacking.

The present study demonstrated that higher serum B2M level was at increased risk of all-cause and CVD mortality. Possible mechanisms for these links are as follows: The association of B2M with CVD mortality could be related to vascular inflammation, considering inflammation is a significant component of atherosclerotic syndromes;^[52,53] The complex interaction between B2M, systemic inflammation and immunity contribute to all-cause mortality;^[54] B2M may damage vessels by participating in amyloid formation in the vessel wall.^[19]

Our study has several strengths. Mostly notably, compared with previous studies,^[55,56] we included a larger sample size, making our results more convincing and applicable. The large sample size also allowed us to perform joint and stratified analyses with sufficient statistical power. Furthermore, we explore the nonlinearity relationship between B2M and mortality. A J-shaped exposure-response relationship was observed for B2M and mortality risks.

Nevertheless, this study has several limitations. First, the measurement of serum B2M was relied on a single examination at baseline, which may not accurately predict death over decades. Second, even though we adjusted for various covariates, we cannot exclude residual confounding that might result in an overestimation of the effect of concentrations of B2M in serum. Third, those excluded from the analysis because of missing covariates were more likely to be of higher B2M. Finally, Creatinine-based eGFR is affected by an individual’s muscle mass. Muscle wasting due to chronic illness is associated with lower creatinine generation, giving a bias toward higher estimated GFR in such individuals. Further studies with measured GFR are needed.

5. Conclusion

B2M is a promising filtration marker for gauging prognosis and predicting mortality. A higher serum B2M concentration was independently associated with increased all-cause and CVD mortality risks in this prospective cohort study of a nationally representative sample of US adults. However, further studies are still needed to validate our findings.

Acknowledgements

We would like to thank the academic editor and reviewers for their important contributions that improved the quality of this article.

Author contributions

Conceptualization: Hang Fang.

Data curation: Hang Fang.

Formal analysis: Hang Fang.

Funding acquisition: Lie Jin.

Investigation: Hang Fang.

Methodology: Qiankun Zhang, Lie Jin.

Project administration: Qiankun Zhang, Lie Jin.

Software: Hang Fang.

Supervision: Qiankun Zhang, Lie Jin.

Validation: Qiankun Zhang, Lie Jin.

Writing – original draft: Hang Fang.

Writing – review & editing: Qiankun Zhang.