Inverse Relationship Between Serum Bilirubin and Atherosclerosis in Men: A Meta-Analysis of Published Studies

Abstract

Bilirubin, a major intravascular product of heme catabolism, is a potent antioxidant compound. Numerous studies have been published showing the relationship between serum bilirubin levels and atherosclerosis. In the present investigation all the epidemiological studies available on the effect of serum bilirubin levels and atherosclerotic disease were analyzed. Studies on the epidemiology of atherosclerotic diseases in relation to serum bilirubin levels were searched in the MEDLINE database. Selected studies were subdivided according to serum bilirubin levels and severity of atherosclerotic disease. Because of the limited number of females involved in the studies, only males were included into meta-analysis. Associations for ordered categorical variables (bilirubin and natural history of graded atherosclerosis) were assessed to find correlation and linear trend between analyzed variables. A stratified analysis was conducted to compare risks of clinical outcomes. Eleven relevant studies were used for analysis. A close negative relationship was found between serum bilirubin levels and severity of atherosclerosis (Spearman rank coefficient r = −0.31,P < 0.0001). The linear trend was confirmed in analysis of proportions with x² values for both disease conditions to be very significant (P < 0.0001). Unambiguous inverse relationship between serum bilirubin levels and atherosclerosis was demonstrated in this preliminary meta-analytic study. These results indicate the importance of hem oxygenase-related products in the prevention of oxidative stress-mediated diseases.

Bilirubin, a major intravascular catabolic product of heme catabolism, is a compound with potent antioxidant capacity. Several clinical epidemiological studies have been published showing the relationship between serum bilirubin levels and oxidative stress-mediated diseases, including in particular atherosclerotic disease (1–11). Although conclusions of the majority of the studies published were rather convincing and suggested important cardiovascular protection from elevated serum bilirubin levels (1, 2, 4–9, 11), this was not the case in all studies (3, 10). Another important limitation was that the designs of the studies were different from each other and also populations studied were stratified according to different serum bilirubin concentrations. In some studies patients with isolated hyperbilirubinemia were not separated from those suffering from apparent liver disease which might have important misleading effect.

In the present investigation we have surveyed all the observational epidemiological studies available dealing with the effect of serum bilirubin levels and both coronary heart and peripheral artery disease. The aim of this study was to bring convincing evidence about the effects of serum bilirubin concentrations on cardiovascular health.

Methods

Studies on the epidemiology of atherosclerotic diseases in relation to the serum bilirubin levels were searched in MEDLINE database. Eleven relevant studies (1–11) were used for analysis (Table I). Some data required for the analysis were calculated from original graphs and papers. Due to limited numbers of females involved in the studies only males were included into meta-analysis. Three analytical approaches were used: Analysis A of the association of bilirubin levels categorized using Schwertner et al.’s scheme (1) and distributions shifting according to health status classification from true healthy across subclinical to clinical disease was assessed by Spearman rank correlation. In an analysis B, the effects of serum bilirubin levels (i.e., bilirubin specific OR/mean study OR) on atherosclerotic outcome were plotted against serum bilirubin that resulted in the regression line (Fig. 1). Analysis C was performed to estimate common effects of contrasted dichotomous bilirubin levels on cardiovascular disease (Table II ). Effects were tested for homogeneity in their orientation and magnitude.

Results

Analysis A.

A close negative relationship was found between ordered serum bilirubin levels and severity of atherosclerosis (Spearman correlation coefficient, r = −0.31, t = −39.94, P < 0.0001) indicating shift of distributions from higher serum bilirubin and better health to lower serum bilirubin and atherosclerotic condition.

Analysis B.

Data in Figure 1 clearly demonstrate strong negative linear association between estimated effects on atherosclerosis and logarithmic scale of serum bilirubin levels supporting results from the study A. As a whole, 34% variability of the effects is explained by bilirubin levels as noted in the result of regression (Fig. 1 ). Serum bilirubin level of 10 μmol/l is an apparent cut-point for discrimination of cardiovascular risk.

Analysis C.

The seven studies included provided data for stratified analysis (Table II ) that further confirmed significant association between bilirubin levels and atherosclerotic outcomes. Because Analysis A and B were focused on categorized although whole bilirubin concentration range, Analysis C is restricted on only two utmost bilirubin level exposure. Common rate ratio (RR) for the prospective studies was 0.79 (95% CI 0.67, 0.90) and for the studies with case-control design common odds ratio (OR) of 0.40 (95% CI 0.29, 0.54) was determined. Both analyzed groups did not show heterogeneity in estimated effects, but heterogeneity existed between follow-up and case-control studies because of different effect of the study designs. In addition, this difference is in magnitude and not in qualitative orientation of the parameters.

Discussion

Based on in vitro as well as animal studies bilirubin is generally recognized as an important antioxidant substance. This is supported also by the majority of clinical studies focused on the protective effect of serum bilirubin levels on atherosclerosis (1, 2, 4–9, 11). However, other studies did not either confirm this relationship (10), or found more pronounced effect in the highest bilirubin levels (3, 4, 9, 10). This can be clearly shown from the OR column in Table I. There are several possible reasons for this observation. Most importantly, the presence of concomitant liver disease was not excluded in the hyperbilirubinemic subjects as can be demonstrated in Figure when re-analyzing data of Breimer and colleagues (3). After dichotomization of top Breimeŕs bilirubin category the hyperbilirubinemic group subdivide according to the presence of atherosclerosis to a high prevalent subgroup in case of hepatopathy and to that with a very low occurrence in persons without liver disease. Very important is also a manner of bilirubin level scaling. It seems that SI units (μmol/l) are more informative and better discriminate serum bilirubin levels within physiological range than those expressed in mg/dl. Wrong stratification of the study groups according to the serum bilirubin level can negatively influence the overall results of the epidemiological study (10).

All the three types analyses (nonparametric [A], regression [B], and stratified [C]) reliably demonstrate inverse and dose–response relationship between serum bilirubin levels and atherosclerotic process from subclinical to clinical outcomes in this preliminary meta-analytic study.

We are standing at the beginning of the serious assessment of the bilirubin effects for clinical medicine. Further studies carefully designed in particular to the main confounders such as hepatopathy and states of exaggerated consumption of endogenous antioxidants (e.g., smoking) are certainly needed to bring definite evidence for the importance of heme oxygenase-related products in the prevention of oxidative stress-mediated diseases.

Table I.

Characteristics of Studies Reviewed

	Serum bilirubin (μmol/L)		Atherosclerotic outcome						Other study characteristics
											Suitability for meta-analysis
Authors/year of publication	Category	Mid-point mean (SD)	Yes	None (control)	Total	Person-years	Ordered disease stages	Odds ratios	Type (follow- up years)	Age	A^a	B^b	C^c
^a Study for the association of bilirubin levels and distributions shifting according to health status classification from true healthy across subclinical to clinical disease.
^b Study for regression of effects (i.e., bilirubin specific OR/mean study OR) of bilirubin levels on atherosclerotic outcome.
^c Meta-analysis of effects of contrasted dichotomous bilirubin levels on cardiovascular disease.
Schwertner 1994	3.4–	6.7	58	161	219		Subclinical coronary stenosis >49%	1.0	Prevalence	42	+	+	+
	10.0–	11.0	34	165	199			0.6
	12.0–	14.0	28	201	229			0.4
	16–55.7	35.9	25	205	230			0.3
Levinson 1997		9.8 (3.6)	171				Subclinical coronary stenosis >70%	↑	Prevalence	60	−	−	−
		11.5 (5.5)		83
Ishizaka 2001	<10.0	≈6.5	0.05	0.03			Subclinical carotic plaques; intimal-medial thickness ≥1.3 mm	1.0	Prevalence	?	−	+	−
	10.0–	11.0	0.28	0.24				0.6
	12.0–	14.0	0.37	0.34				0.5
	16.0–	19.0	0.19	0.21				0.5
	20.0≤	≈35.0	0.11	0.18				0.3
Djoussé 2001	0.2–	4.8	69	308	377	7624	Ischemic heart disease and/or sudden cardiac death	1.0	Prospective (22)	38	+	+	+
	9.4–	10.3	25	260	285	6154		0.4
	11.11–	12.8	59	517	576	12373		0.5
	14.53–	16.2	36	344	380	8314		0.5
	17.95–66.7	42.3	42	323	365	7788		0.6
Vítek 2002		9.0 (2.7)	33				Ischemic heart disease in Gilbert syndrome		Prospective observed vs. expected cases (3)	49	+	−	+
		9.1 (2.7)		14
		32.6 (13.5)	0	35		105		↓
			(3.1)
Breimer 1995	1.0–	4.0	90	698	788	8036	Ischemic heart disease (25% prevalence at entry	1.0	Prospective (11.5)	50	+	+	+
	7.0–	7.5	77	638	715	7549		0.9
	8.0–	9.0	114	1257	1371	14615		0.7
	10.0–	11.0	79	826	905	9753		0.7
	12–	14.0	81	765	845	12222		0.8
	17.0–86.0	34.5	29	262	291			0.9
Hopkins 1996	2.0–	4.4	49	18	67		Early ischemic heart disease; Age at dg. <55	1.0	Prospective nested case-control (?)	56	+	+	+
	6.9–	8.7	38	16	54			0.9
	10.5–	12.0	10	17	27			0.2
	13.6–	15.3	5	18	23			0.1
	17–54.5	35.8	18	16	34			0.4
Hunt 1996		8.9 (3.3)	54	1057	1111		Ischemic heart disease in high billirubin phenotype	↑	Prospective	39	+	−	+
		19.4 (8.4)	2	127	129			(2)
Temme 2001	1.71–	3.0	41.5	698.5	740	6917	Cardiovascular death	1.0	Prospective (10)	50	+	+	+
	4.27–	5.1	59.2	1256.8	1316	12596		0.8
	5.98–	6.8	58.8	1275.2	1334	12783		0.8
	7.7–	8.5	34.5	845.5	880	8415		0.7
	9.4–65	37.2	52.4	1137.6	1190	11391		0.8
Breimer 1994		6.9 (2.5)	19	(laboratory population			Periphery artery disease	↑	Prevalence	?	−	−	−
		11.0 (3.0)
Wei 2000		10.9 (?)	?		?	?	Coronary death	↑	Prospective nested case-control (11)	?	−	+	−
		13.3 (?)		?	?	?

Table II.

Effects of “higher” bilirubin levels (lower bilirubin as reference) on ischemic heart disease

Studies			Effect^a (95% CI)^b		Test of homogeneity^c
Follow-up	Bilirubin	Cases/PT^d	Stratum rate ratio	Common rate ratio	Score χ²
Breimer 1995	<8	167/15585	0.80
	10+	189/21975	(0.65, 0.99)
Djoussé 2001	<11.1	94/13778	0.71
	14.5+	78/16102	(0.52, 0.96)	0.79	heterogeneity NO
Temme 2001	<6.0	101/19512	0.84	(0.67, 0.90)	3.143 (3 DF)
	7.7+	87/19806	(0.63, 1.22)		P = 0.370
Vítek 2002	≈32.6	0/3.1^e	0.87
	(Gilbert sy)		(0, 0.97)

Case-control		Cases/controls	Stratum odds ratio	Common odds ratio
Hopkins 1996	<10.5	87/34	0.27
	14+	23/34	(0.14, 0.51)
Hunt 1996	≈8.9	54/1057	0.33	0.40	heterogeneity NO
	≈19.4	2/127	(0.05, 1.08)	(0.29, 0.54)	2.202 (2 DF)
Schwertner 1994	<12	92/326	0.46		P = 0.3326
	12+	43/406	(0.32, 0.67)

ALL studies		Common relative risk
^a Effects are calculated as median unbiased estimate (ref. 12).
^b 95% confidence intervals (calculated as Mid-P exact limits).
^c Score chi-square is calculated using Mantel-Haenszel method for testing of homogeneity of effects across studies (strata), (ref. 12).
^d Population-Time (or person-years).
^e Observed vs expected ratio (3 years of follow-up).
Cases/PT	716/106758	0.69	heterogeneity YES
Cases/Controls	311/1984	(0.69, 0.78)	17.27 (2 DF)
			P = 0.00018

Figure 1.

Regression of effects of bilirubin levels on atherosclerotic outcome. *Effects in top Breimer bilirubin category when liver status is taken into account.

Footnotes

This work was supported by the Czech Ministry of Education Grants J 13/98:111100002 and LN 00B107.

1

To whom requests for reprints should be addressed at 4th Dept. Int. Med., 1st Medical Faculty, Charles University, Prague U Nemocnice 2 Praha 2, 128 08 Czech Republic. E-mail:

References

Schwertner HA, Jackson WG, Tolan G. Association of low serum concentration of bilirubin with increased risk of coronary artery disease. Clin Chem 40 : 18–23, 1994.

Levinson SS. Relationship between bilirubin, apolipoprotein B, and coronary artery disease. Ann Clin Lab Sci 27 : 185–192, 1997.

Breimer LH, Wannamethee G, Ebrahim S, Shaper AG. Serum bilirubin and risk of ischemic heart disease in middle-aged British men. Clin Chem 41 : 1504–1508, 1995.

Hopkins PN, Wu LL, Hunt SC, James BC, Vincent GM, Williams RR. Higher serum bilirubin is associated with decreased risk for early familial coronary artery disease. Arterioscler Thromb Vasc Biol 16 : 250–255, 1996.

Hunt SC, Wu LL, Hopkins PN, Williams RR. Evidence for a major gene elevating serum bilirubin concentration in Utah pedigrees. Arterioscler Thromb Vasc Biol 16 : 912–917, 1996.

Breimer LH, Spyropolous KA, Winder AF, Mikhailidis DP, Hamilton G. Is bilirubin protective against coronary artery disease? Clin Chem 40 : 1987–1988, 1994.

Ishizaka N, Ishizaka Y, Takahashi E, Yamakado M, Hashimoto H. High serum bilirubin level is inversely associated with the presence of carotid plaque. Stroke 32 : 580–583, 2001.

Vítek L, Jirsa M, Brodanová M, Kaláb M, Mareček Z, Danzig V, Novotný L, Kotal P. Gilbert syndrome and ischemic heart disease: A protective effect of elevated bilirubin levels. Atherosclerosis 160 : 449–456, 2002.

Djousse L, Levy D, Cupples LA, Evans JC, D’Agostino RB, Ellison RC. Total serum bilirubin and risk of cardiovascular disease in the Framingham Offspring Study. Am J Cardiol 87 : 1196–1200, 2001.

10.

Temme EHM, Zhang JJ, Schouten EG, Kesteloot H. Serum bilirubin and 10-year mortality risk in a Belgian population. Canc Caus Cont 12 : 887–894, 2001.

11.

Wei M, Schwertner HA, Zeng Q, Blair SN. Fasting serum bilirubin concentrations and the risk of subsequent coronary heart disease death in men (abstract). Hepatology 32 : 314A, 2000.

12.

Martin DO, Austin H. An exact method for meta-analysis of case-control and follow-up studies. Epidemiology 11 : 255–260, 2000.