European Heart Journal (2018) 39, 1181–1190 CLINICAL RESEARCH

doi:10.1093/eurheartj/ehx665 Lipids

U-shaped relationship of HDL and risk

of infectious disease: two prospective
population-based cohort studies
Christian M. Madsen1,2,3, Anette Varbo1,2,3, Anne Tybjærg-Hansen2,3,4,5,

Downloaded from https://academic.oup.com/eurheartj/article/39/14/1181/4710060 by guest on 19 August 2020

Ruth Frikke-Schmidt2,3,5, and Børge G. Nordestgaard1,2,3,4*
1
Department of Clinical Biochemistry, Herlev and Gentofte Hospital, Copenhagen University Hospital, Herlev Ringvej 75, 2730 Herlev, Denmark; 2The Copenhagen General
Population Study, Herlev and Gentofte Hospital, Copenhagen University Hospital, Herlev Ringvej 75, 2730 Herlev, Denmark; 3Faculty of Health and Medical Sciences, University
of Copenhagen, Blegdamsvej 3B, 2200 Copenhagen, Denmark; 4The Copenhagen City Heart Study, Frederiksberg Hospital, Copenhagen University Hospital, Nordre Fasanvej 57,
2000 Frederiksberg, Denmark; and 5Department of Clinical Biochemistry, Rigshospitalet, Copenhagen University Hospital, Blegdamsvej 9, 2100 Copenhagen, Denmark

Received 28 April 2017; revised 24 August 2017; editorial decision 27 October 2017; accepted 28 October 2017; online publish-ahead-of-print 8 December 2017

See page 1191 for the editorial comment on this article (doi: 10.1093/eurheartj/ehx734)

Aims Preclinical evidence has indicated that HDL may play an important role in the immune system; however, very little
is known about the role of HDL in the immune system in humans. We tested the hypothesis that low and high
concentrations of HDL cholesterol are associated with risk of infectious disease in the general population.
...................................................................................................................................................................................................
Methods We included 97 166 individuals from the Copenhagen General Population Study and 9387 from the Copenhagen
and results City Heart Study with measurements of HDL cholesterol at baseline. The primary endpoint was any infectious disease
requiring hospital admission, ascertained in the Danish health registries from baseline in 2003–13 or 1991–94 through
2014; 9% and 31% of individuals in the two studies experienced one or more infectious disease events. Using
restricted cubic splines, there was a U-shaped association between concentrations of HDL cholesterol and risk of any
infection. Following multifactorial adjustment, individuals with HDL cholesterol below 0.8 mmol/L (31 mg/dL) and
above 2.6 mmol/L (100 mg/dL) had hazard ratios for any infection of 1.75 (95% confidence interval 1.31–2.34) and
1.43 (1.16–1.76), compared to those with HDL cholesterol of 2.2–2.3 mmol/L (85–95 mg/dL). In the Copenhagen City
Heart Study, corresponding hazard ratios for any infection were 2.00 (1.16–3.43) and 1.13 (0.80–1.60).
...................................................................................................................................................................................................
Conclusion Low and high HDL cholesterol concentrations found in 21% and 8% of individuals were associated with higher risk
of infectious disease in the general population. These findings do not necessarily indicate causality.
䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏 䊏

Keywords Lipids • Lipoproteins • HDL • Infectious disease • General population • Epidemiology

..
Introduction .. HDL may have a role in immunity, as HDL cholesterol concentra-
.. tions decrease rapidly at the onset of sepsis and as low HDL is associ-
..
In recent years, a protective role of HDL in cardiovascular disease .. ated with poor outcome in these patients.4,5 A possible explanation
.. for such phenomena could be the ability of HDL to neutralize and
has been challenged. First, genetically low concentrations of HDL ..
cholesterol were not causally associated with high risk of cardiovas- .. clear pro-inflammatory endotoxins.6,7 HDL may also be involved in
..
cular disease.1,2 Second, randomized clinical trials with HDL choles- .. protection against certain parasitic infections, as apolipoprotein L-I
terol-elevating drugs were unable to reduce vascular disease risk.3 .. found in HDL particles confers human resistance towards infection
..
However, HDL is conserved and present in many species, indicating .. with Trypanosoma brucei brucei by lysing the parasite.8 Mechanistically,
an important yet unknown biological role from an evolutionary .. preclinical studies suggest that HDL can regulate innate and adaptive
..
standpoint, although this has not been clearly determined. . immunity,9 as HDL takes part in the regulation of the proliferation of

* Corresponding author. Tel: þ45 38683297, Fax: þ45 38683311, Email: boerge.nordestgaard@regionh.dk
Published on behalf of the European Society of Cardiology. All rights reserved. V
C The Author 2017. For permissions, please email: journals.permissions@oup.com.
1182 C.M. Madsen et al.

..
haematopoietic stem cells from the bone marrow and is also able to ..
modulate the maturation and function of immune cells through an ..
..
effect on cell surface receptors.10–12 ..
Despite the evidence indicating that HDL is an important modula- ..
..
tor of the immune response, little is known about the relationship of ..
HDL with the immune system in humans, and specifically whether
..
..
HDL cholesterol concentrations associate with infectious disease
risk. Indeed, the association between HDL concentrations and risk of
infectious disease has only been investigated in five studies with 204
to 1719 hospitalized patients showing that low HDL cholesterol at
admission was associated with high risk of developing infectious com-
plications in relation to hospitalization13–17. However, we recently
Statistical analyses
Statistical analyses were performed using Stata 13.1, and P-values were
two sided. Only for adjustment in the few individuals with missing infor-
mation on certain covariates (see Supplementary material online, Tables
S2 and S3), we used multiple imputations with chained equations to fill
out the missing values.
Associations between risk of infectious disease and HDL cholesterol,
..
.. apolipoprotein A1, triglycerides, and LDL cholesterol on a continuous scale
.. were examined using restricted cubic splines, with knots at equally spaced
..
.. percentiles, which were used in multiple-event Cox proportional hazards
.. regression. These analyses were used to help define meaningful reference
.. groups using categories of HDL cholesterol, with the category having the
..
.. lowest risk as the reference group. Individuals were also stratified into four

Downloaded from https://academic.oup.com/eurheartj/article/39/14/1181/4710060 by guest on 19 August 2020

found that both low and high HDL cholesterol is associated with high
.. clinically relevant HDL cholesterol groups with 0.5 mmol/L (19 mg/dL)
..
all-cause mortality.18 .. intervals and into 11 groups with 0.2 mmol/L (8 mg/dL) intervals.
.. To allow for repeated infectious disease events for each individual,
We tested the hypothesis that low and high concentrations of ..
HDL cholesterol are associated with risk of infectious disease in the .. multiple-event Cox proportional hazards regression using the approach
.. described by Andersen–Gill with robust standard errors was used as the
general population. .. primary analysis. Age was used as timescale, and analyses were conducted
..
.. with delayed entry (left truncation) and censoring at death or emigration
.. (0.4%). Conventional single-event Cox proportional hazards regressions
Methods ..
.. were carried out as sensitivity analyses and were used for analyses of
.. infectious disease-related death. Infectious disease events were also
A detailed description of the cohorts, endpoints, laboratory analyses, .. analysed prospectively using negative binomial regression, Fine and Gray
covariates, and statistical analyses is given in the Supplementary material ..
.. competing risk regression, and logistic regression.
online, Methods. .. For genetic analyses, genotypes were combined into an unweighted
..
.. allele score of HDL cholesterol increasing alleles. As genotypes are present
The Copenhagen General Population Study .. from birth, we included infectious disease events from birth or 1977 (regis-
The primary study cohort was comprised of 97 166 individuals from the ..
.. try start), whichever came last. Multiple-event Cox proportional hazards
prospective Copenhagen General Population Study with baseline meas- .. regression adjusted for age and gender was used for the genetic analyses.
urements of HDL cholesterol. It was initiated in 2003 and recruitment is ..
ongoing.
.. Risk estimates and confidence intervals (CIs) were corrected for
.. regression dilution bias using a non-parametric method. P-values are for
.. linear trend, and we tested for interactions by incorporating two-factor
The Copenhagen City Heart Study ..
.. interaction terms between HDL cholesterol and included covariates in
A total of 9387 individuals from the 1991 to 1994 examination of the .. categories, or as continuous variables, as indicated. P-values for interac-
Copenhagen City Heart Study with baseline measurements of HDL cho-
..
.. tion were obtained using the Wald test.
lesterol were included as an independent cohort for confirmation of the ..
results obtained in the Copenhagen General Population Study. ..
..
..
..
Results
Endpoints ..
Infectious disease endpoints were based on data from the national .. Individuals included from the Copenhagen General Population Study
Danish Patient Registry, which records all hospital contacts in Denmark. .. were followed for a median of 6 years (range 0–11 years), and 8282 (9%)
..
We only included infectious events listed as the primary cause of admis- .. individuals experienced one or more infectious disease events during
sion until the end of follow-up on 8 November 2014. Infectious diseases .. follow-up. Correspondingly, individuals from the Copenhagen City
..
were grouped based on the International Classification of Diseases codes .. Heart Study were followed up for 20 years (0–23 years), and 2904 (31%)
as shown in Supplementary material online, Table S1. For the endpoint .. developed an infectious disease event. Baseline characteristics for both
infectious disease-related death, data were retrieved from the national
..
.. cohorts are shown in Supplementary material online, Tables S4 to S12.
Danish Register of Causes of Death. .. HDL cholesterol at baseline was associated with parameters of the
..
.. immune system as plasma C-reactive protein (CRP) decreased 29%
Laboratory analyses, genotyping, and .. (95% CI 27–30%; R2 = 0.02) and blood leucocytes decreased
..
covariates .. 0.51  109/L (0.49–0.53  109/L; R2 = 0.02) per 1 mmol/L (39 mg/dL)
HDL cholesterol, LDL cholesterol, triglycerides, and apolipoprotein A1 .. higher HDL cholesterol (see Supplementary material online, Figure S1).
..
were all measured non fasting using standard hospital assays.19 ..
Genotyping was done using TaqMan-based assays. Covariates for statisti- .. HDL cholesterol, apolipoprotein A1,
..
cal adjustment were chosen a priori based on known association with .. triglycerides, and LDL cholesterol
HDL cholesterol and risk of infectious disease. Multifactorial adjustment ..
.. on a continuous scale and risk of
was for age, gender, body mass index, smoking, alcohol intake, physical ..
activity in leisure time, diabetes mellitus, ischaemic heart disease, lipid- .. infectious disease
lowering therapy, estimated glomerular filtration rate (eGFR), LDL cho-
.. Using restricted cubic splines, both low and high concentrations
..
lesterol, and triglycerides. . (U-shaped) of HDL cholesterol were associated with high risk of any
U-shaped relationship of HDL and risk of infectious disease 1183

infection (Figure 1). The same was observed for concentrations of

apolipoprotein A1, whereas no clear association was seen for con-

Fraction of population (Density)

6 1
centrations of triglycerides and risk of any infection. Low concentra-

any infectious disease

5
tions of LDL cholesterol were also associated with high risk of any

Hazard ratio for

infection. 4

3 0.5
HDL cholesterol and risk of infectious
2
disease
When HDL cholesterol concentrations were divided into 0.2 mmol/L 1

(8 mg/dL) intervals, the risk of any infectious disease showed a 0 0

U-shaped pattern with both low concentrations and very high con- mmol/L: 0 1 2 3
mg/dL: 0 39 77 116
centrations being associated with high risk, compared with the refer- HDL cholesterol

Downloaded from https://academic.oup.com/eurheartj/article/39/14/1181/4710060 by guest on 19 August 2020

ence group with HDL cholesterol concentrations of 2.2–2.3 mmol/L

Fraction of population (Density)

0.02
(85–93 mg/dL) (Figure 2, lower part). Individuals with HDL cholesterol 6

any infectious disease

below 0.8 mmol/L (31 mg/dL) had a multifactorially adjusted hazard 5

Hazard ratio for

ratio for any infection of 1.75 (95% confidence interval 1.31–2.34) as
4
compared to those with HDL cholesterol of 2.2–2.3 mmol/L (85–
3 0.01
93 mg/dL). Individuals with HDL cholesterol >_2.6 mmol/L (100 mg/
dL) also showed high risk of infection with a corresponding hazard 2
ratio of 1.43 (1.16–1.76). 1
In age- and gender-adjusted models, there was an inverse associa- 0 0
tion between concentrations of HDL cholesterol and risk of any
mg/dL: 0 100 200 300
infectious disease when HDL cholesterol concentrations were Apolipoprotein A1

Fraction of population (Density)

divided into 0.5 mmol/L (19 mg/dL) intervals (P for trend < 0.0001; 6 1
Figure 2, upper part). This association was attenuated, but persisted,
any infectious disease

5
following multifactorial adjustment with a hazard ratio for any infec-
Hazard ratio for

4
tion of 1.24 (1.05–1.46) for individuals with HDL cholesterol below
1.0 mmol/L (39 mg/dL) compared to individuals with HDL choles- 3 0.5
terol above 2.0 mmol/L (77 mg/dL). 2
The population-attributable risks of infectious disease in individuals
1
with low (<1.2 mmol/L (<46 mg/dL)) and high (>_2.4 mmol/L
(>_93 mg/dL)) HDL cholesterol were 3.2% and 1.9%, respectively. 0 0
mmol/L: 0 1 2 3 4 5
mg/dL: 0 89 177 366 354 443
HDL cholesterol and risk of infectious Triglycerides

Fraction of population (Density)

disease subgroups 6 1
any infectious disease

For infectious disease subgroups, except the group other infections, 5

Hazard ratio for

there were inverse associations between HDL cholesterol concen- 4

trations and risk of infectious disease in age- and gender-adjusted
3 0.5
models (Figure 3). This association persisted after multifactorial
adjustment for gastroenteritis and bacterial pneumonia with hazard 2
ratios of 1.68 (95% confidence interval 1.10–2.56) and 1.34 (1.02– 1
1.76) for individuals with HDL cholesterol below 1.0 mmol/L (39 mg/
0 0
dL) compared to individuals with HDL cholesterol above 2.0 mmol/L
mmol/L: 0 1 2 3 4 5 6 7 8
(77 mg/dL). In the remaining subgroups of infectious disease, the mg/dL: 0 39 77 116 154 193 232 270 309
association with HDL cholesterol did not persist after multifactorial LDL cholesterol

adjustment. The relationship between HDL cholesterol and infec-
tious disease subgroups was also examined using 0.2 mmol/L (8 mg/
dL) categories of HDL cholesterol, although with less statistical
power (see Supplementary material online, Figure S2). The high risk
of any infectious disease observed with high concentrations of HDL
cholesterol seemed to be driven by bacterial pneumonia, skin infec-
tion, and urinary tract infection.
Grouping the infectious disease subgroups by likely aetiology into
bacterial and viral showed that both low and high concentrations of
HDL cholesterol were associated with high risk of bacterial infectious
disease, whereas only low concentrations of HDL cholesterol were
Figure 1 HDL cholesterol, apolipoprotein A1, triglycerides, and
LDL cholesterol on a continuous scale and risk of any infectious dis-
ease in 97 166 individuals from the Copenhagen General
Population Study. Analyses were conducted using restricted cubic
splines, with hazard ratios and 95% confidence intervals from multi-
ple-event Cox proportional hazards regression. The values of HDL
cholesterol, apolipoprotein A1, triglycerides, and LDL cholesterol
with the lowest hazard ratio were chosen as reference. The light
green, blue, yellow, and purple areas indicate the distribution of
concentrations of HDL cholesterol, apolipoprotein A1, triglycer-
ides, and LDL cholesterol, respectively.
1184 C.M. Madsen et al.

HDL cholesterol Individuals Events Risk of any infectious disease P for Risk of any infectious disease P for
mmol/L (mg/dL) Age and sex adjusted hazard ratio (95% CI) trend Multifactorially adjusted hazard ratio (95% CI) trend

≥2.0 (≥77) 21,617 2,698 1 (reference) 1 (reference)

1.5-1.9 (58-77) 32,219 3,909 1.07 (0.98-1.18) 1.02 (0.93-1.13)

1.0-1.4 (39-58) 34,229 4,306 1.17 (1.06-1.29) 0.98 (0.88-1.09)

<1.0 (<39) 9,101 1,444 1.84 (1.60-2.11) <0.001 1.25 (1.06-1.47) 0.18

≥2.6 (≥100) 4,169 658 1.42 (1.15-1.74) 1.42 (1.16-1.75)

2.4-2.5 (93-100) 3,748 510 1.34 (1.07-1.68) 1.34 (1.07-1.68)

2.2-2.3 (85-93) 5,443 588 1 (reference) 1 (reference)

Downloaded from https://academic.oup.com/eurheartj/article/39/14/1181/4710060 by guest on 19 August 2020

2.0-2.1 (77-85) 8,257 942 1.08 (0.90-1.31) 1.07 (0.88-1.29)

1.8-1.9 (69-77) 10,180 1,216 1.25 (1.03-1.50) 1.21 (1.00-1.45)

1.6-1.7 (62-69) 14,367 1,787 1.29 (1.08-1.53) 1.20 (1.01-1.43)

1.4-1.5 (54-62) 14,631 1,747 1.26 (1.06-1.50) <0.001 1.14 (0.95-1.36) 0.003

1.2-1.3 (46-54) 15,711 1,904 1.28 (1.08-1.52) 1.09 (0.92-1.30)

1.0-1.1 (39-46) 11,559 1,561 1.60 (1.34-1.91) 1.22 (1.01-1.48)

0.8-0.9 (31-39) 6,601 986 1.91 (1.56-2.34) 1.36 (1.10-1.68)

<0.8 (<31) 2,500 458 2.96 (2.29-3.84) 1.75 (1.31-2.34)

0.75 1.0 1.5 2.0 3.0 4.0 0.75 1.0 1.5 2.0 3.0 4.0
Hazard ratio (95% CI) Hazard ratio (95% CI)

Figure 2 HDL cholesterol and risk of any infectious disease in 97 166 individuals from the Copenhagen General Population Study. Individuals are
in the top part divided into 0.5 mmol/L (19mg/dL) intervals, and in the bottom part into 0.2 mmol/L (8 mg/dL) intervals of HDL cholesterol. Hazard
ratios were from multiple-event Cox proportional hazards regression.

..
associated with high risk of viral infectious disease, albeit the number
of events of viral disease were limited (see Supplementary material
online, Figure S3).
Sensitivity analyses
The association between low HDL cholesterol and high risk of any
infectious disease was robust after stratification for age, gender, body
mass index, smoking, alcohol intake, physical activity, statin use, diabe-
tes, ischaemic heart disease, triglycerides, LDL cholesterol, eGFR,
and birth year (Figure 4). Specifically, results were similar when strati-
fying based on alcohol intake, plasma CRP, and blood leucocytes
(Figure 5). When further adjusting for plasma CRP or blood
leucocytes, the association was slightly attenuated for plasma CRP
adjustment but persisted for the groups with extreme high and low
concentrations of HDL cholesterol (see Supplementary material
online, Figure S4). As women on average have higher HDL cholesterol
than men, we also examined restricted cubic spline models in men
and women separately (see Supplementary material online, Figure
S5), with a similar overall pattern in both gender.
Results from individuals without missing information on covariates,
i.e. a complete case analysis, were similar to those with imputed miss-
ing covariates (compare Supplementary material online, Figure S6
with Figure 2). Furthermore, when examining the association
between HDL cholesterol and risk of infectious disease using (i) sin-
gle-event Cox proportional hazard regression (see Supplementary
material online, Figure S7), (ii) competing risk regression with death as
competing event (see Supplementary material online, Figure S8), (iii)
negative binomial regression (see Supplementary material online,
.. Figure S9), and (iv) cross-sectional logistic regression (see
.. Supplementary material online, Figure S10), results were similar to
..
.. those obtained from multiple-event Cox proportional hazard regres-
.. sion shown in Figure 2, with similar risk estimates.
..
.. Finally, examining at infectious disease-related death, the associa-
.. tion with HDL cholesterol was similar to what was observed for any
..
.. infectious disease, as both low and high concentrations of HDL cho-
.. lesterol were associated with high risk of infectious disease-related
..
.. death (Figure 6).
..
..
.. HDL cholesterol and risk of infectious
..
.. disease: independent confirmation
.. Associations between concentration of HDL cholesterol and risk of
..
.. infectious disease were also investigated in an independent cohort,
..
.. the Copenhagen City Heart Study. Using 0.5 mmol/L (19 mg/dL) cut-
.. points of HDL cholesterol, a similar association was observed as in
... the Copenhagen General Population Study for low concentrations of
..
.. HDL cholesterol (compare Figure 7 with Figure 2). Individuals with
.. HDL cholesterol below 1.0 mmol/L (39 mg/dL) had a multifactorially
..
.. adjusted hazard ratio for any infection of 1.54 (95% CI 1.16–2.04)
.. compared to individuals with HDL cholesterol above 2.0 mmol/L
..
.. (77 mg/dL). Also, a similar trend as in the Copenhagen General
..
.. Population Study was observed in the Copenhagen City Heart Study
.. when examining 0.2 mmol/L (8 mg/dL) intervals of HDL cholesterol,
..
.. as individuals with HDL cholesterol below 0.8 mmol/L (31 mg/dL)
.. had a multifactorially adjusted hazard ratio of 2.00 (95% CI 1.16–
..
.. 3.43) compared to individuals with HDL cholesterol of 2.2–
.. 2.3 mmol/L (95% CI 85–93 mg/dL); individuals with very high
U-shaped relationship of HDL and risk of infectious disease 1185

HDL cholesterol Individuals Events Risk of infectious disease subgroups P for Risk of infectious disease subgroups P for
mmol/L (mg/dL) Age and sex adjusted hazard ratio (95% CI) trend Multifactorially adjusted hazard ratio (95% CI) trend

Gastroenteritis

≥2.0 (≥77) 21,617 229 1 (reference) 1 (reference)

1.5-1.9 (58-77) 32,219 313 1.09 (0.83-1.43) 1.04 (0.78-1.37)

1.0-1.4 (39-58) 34,229 354 1.40 (1.06-1.86) 1.17 (0.86-1.59)

<1.0 (<39) 9,101 124 2.62 (1.82-3.77) <0.001 1.66 (1.08-2.55) 0.05

Bacterial pneumonia

≥2.0 (≥77) 21,617 1,138 1 (reference) 1 (reference)

Downloaded from https://academic.oup.com/eurheartj/article/39/14/1181/4710060 by guest on 19 August 2020

1.5-1.9 (58-77) 32,219 1,545 1.02 (0.88-1.17) 1.05 (0.91-1.21)

1.0-1.4 (39-58) 34,229 1,535 0.99 (0.85-1.14) 0.95 (0.81-1.13)

<1.0 (<39) 9,101 511 1.59 (1.26-2.00) 0.02 1.35 (1.03-1.76) 0.37

Skin infection

≥2.0 (≥77) 21,617 487 1 (reference) 1 (reference)

1.5-1.9 (58-77) 32,219 780 1.12 (0.93-1.35) 0.97 (0.80-1.17)

1.0-1.4 (39-58) 34,229 992 1.40 (1.16-1.68) 0.95 (0.78-1.16)

<1.0 (<39) 9,101 373 2.32 (1.81-2.97) <0.001 1.16 (0.89-1.52) 0.56

Urinary tract infection

≥2.0 (≥77) 2,1617 502 1 (reference) 1 (reference)

1.5-1.9 (58-77) 32,219 703 1.11 (0.91-1.35) 1.02 (0.83-1.25)

1.0-1.4 (39-58) 34,229 762 1.28 (1.05-1.56) 1.03 (0.83-1.28)

<1.0 (<39) 9,101 213 1.71 (1.29-2.25) <0.001 1.13 (0.82-1.54) 0.59

Sepsis

≥2.0 (≥77) 21,617 320 1 (reference) 1 (reference)

1.5-1.9 (58-77) 32,219 465 0.98 (0.77-1.24) 0.92 (0.73-1.17)

1.0-1.4 (39-58) 34,229 591 1.17 (0.92-1.48) 0.94 (0.73-1.22)

<1.0 (<39) 9,101 187 1.60 (1.15-2.23) 0.006 0.97 (0.67-1.40) 0.84

Other Infections

≥2.0 (≥77) 21,617 96 1 (reference) 1 (reference)

1.5-1.9 (58-77) 32,219 192 1.37 (0.83-2.25) 1.43 (0.87-2.33)

1.0-1.4 (39-58) 34,229 182 1.03 (0.64-1.65) 1.11 (0.65-1.91)

<1.0 (<39) 9,101 73 1.72 (0.90-3.27) 0.42 1.98 (0.85-4.60) 0.37

0.75 1.0 1.5 2.0 3.0 0.75 1.0 1.5 2.0 3.0
Hazard ratio (95% CI) Hazard ratio (95% CI)

Figure 3 HDL cholesterol and risk of infectious disease in subgroups of diagnoses in 97 166 individuals from the Copenhagen General Population
Study. Hazard ratios were from multiple-event Cox proportional hazards regression.

..
concentrations of HDL cholesterol, i.e. above 2.6 mmol/L (100 mg/dL), .. trend <0.001; left lower panel Figure 8). The risk of any infectious
had a hazard ratio of 1.13 (95% CI 0.80–1.60) for risk of infectious .. disease was lower with higher numbers of HDL cholesterol
..
disease. .. increasing alleles and hence higher genetically determined HDL
.. cholesterol (left upper panel Figure 8). This seemed to be most
..
Genetically determined HDL cholesterol .. pronounced in those with HDL cholesterol above the median
..
and risk of infectious disease .. (centre and right panel Figure 8). The risk of infectious disease
The concentration of HDL cholesterol increased with increasing .. related to the two variants in LIPC and CETP separately are shown
..
number of HDL cholesterol increasing alleles in an unweighted .. in Supplementary material online, Figure S11. All covariates (not
allele score of two common variants in the genes encoding hepatic
.. including apolipoprotein A1) were equally distributed across the
..
lipase (LIPC) and cholesteryl ester transfer protein (CETP) (P for . different allele scores, except for LDL cholesterol, which
1186 C.M. Madsen et al.

Individuals Events Risk of any infectious disease P for

Multifactorially adjusted hazard ratio (95% CI) interaction
for HDL cholesterol<1.0 mmol/L (<39 mg/dL)
vs. HDL cholesterol≥1.0 mmol/L (≥39 mg/dL)
All 97,166 12,357 1.26 (1.11-1.43)
Age
20-39 years 6,636 340 1.77 (0.87-3.62)
40-59 years 46,736 3,095 1.28 (1.00-1.64)
60-79 years 39,301 6,875 1.18 (1.00-1.40)
≥80 years 4,493 2,047 1.32 (0.97-1.80) 0.21
Sex
Women 53,507 5,767 1.33 (1.05-1.69)

Downloaded from https://academic.oup.com/eurheartj/article/39/14/1181/4710060 by guest on 19 August 2020

Men 43,659 6,590 1.26 (1.08-1.46) 0.23
Body mass index (kg/m2)
<20 3,606 534 2.33 (0.69-7.80)
20-24 38,787 4,043 1.48 (1.03-2.12)
25-29 38,885 5,052 1.21 (1.01-1.44)
≥30 15,733 2,700 1.24 (1.00-1.54) 0.42
Smoking
Never 40,231 3,619 1.21 (0.96-1.51)
Former 39,315 5,648 1.25 (1.04-1.50)
Current 17,280 3,066 1.35 (1.03-1.77) 0.72
Alcohol intake
≤7/14 units 54,345 6,442 1.22 (1.03-1.43)
>7/14 units 38,218 4,912 1.11 (0.89-1.39) 0.31
Physical activity
≤4 hours/week 48,967 4,795 1.23 (0.98-1.53)
>4 hours/week 47,382 7,297 1.31 (1.12-1.54) 0.83
Lipid-lowering therapy
No 85,360 9,866 1.26 (1.08-1.46)
Yes 11,471 2,437 1.28 (0.99-1.64) 0.92
Diabetes
No 92,723 10,945 1.30 (1.13-1.49)
Yes 4,126 1,363 1.05 (0.74-1.48) 0.65
Ischemic heart disease
No 91,555 10,425 1.27 (1.10-1.47)
Yes 5,611 1,932 1.20 (0.90-1.61) 0.58
Triglycerides
≤50th percentile 49,548 5,481 1.62 (1.11-2.36)
>50th percentile 47,606 6,873 1.20 (1.06-1.37) 0.22
LDL-cholesterol
≤50th percentile 50,289 7,040 1.40 (1.17-1.67)
>50th percentile 46,560 5,294
1.11 (0.93-1.33) 0.13
eGFR
≤50th percentile 46,709 8,054 1.22 (1.04-1.42)
>50th percentile 50,277 4,289 1.39 (1.12-1.71) 0.66
Birth year
≤1950 49,862 9,856 1.31 (1.12-1.53)
>1950 47,304 2,501 1.24 (0.99-1.53) 0.62

0.75 1.0 1.5 3.0

Hazard ratio (95%CI)

Figure 4 Risk of any infectious disease in different strata, for HDL cholesterol above vs. below 1 mmol/L in 97 166 individuals from the
Copenhagen General Population Study. Hazard ratios were from multiple-event Cox proportional hazards regression. P for interaction was from
Wald test. The total number of individuals varies slightly with the availability of the variables used for stratification.
U-shaped relationship of HDL and risk of infectious disease 1187

Alcohol intake Plasma C-reactive protein Blood leukocytes

≤7/14 units/week (women/men) ≤50th percentile ≤50th percentile
Individuals: 54,345 1 Individuals: 50,440 1 Individuals: 51,730 1

Fraction of population (Density)

6

6
Events: 6,442 Events: 4,263 Events: 5,491
infectious disease (95% CI)

5
5
Hazard ratio for any

4
4

0.5 0.5 0.5

3
3

2
2

1
1

1
0 0 0

0
0

Downloaded from https://academic.oup.com/eurheartj/article/39/14/1181/4710060 by guest on 19 August 2020

>7/14 units/week (women/men) >50th percentile >50th percentile
Individuals: 38,218 Individuals: 46,628 Individuals: 45,296 1

Fraction of population (Density)

1 1

6
6

Events: 4,912 Events: 8,086 Events: 6,847

infectious disease (95% CI)

5
5
Hazard ratio for any

4
4

0.5 0.5 0.5

3
3

3
2
2

2
1
1

1
0 0 0
0
0

0
mmol/L: 0 1 2 3 0 1 2 3 0 1 2 3
mg/dL: 0 39 77 116 0 39 77 116 0 39 77 116
HDL cholesterol HDL cholesterol HDL cholesterol

P for intercation: 0.07 0.23 0.24

Figure 5 Risk of any infectious disease stratified by alcohol intake, plasma C-reactive protein, and blood leucocytes for HDL cholesterol on a con-
tinuous scale in 97 166 individuals from the Copenhagen General Population Study. Analyses were conducted using restricted cubic splines, with haz-
ard ratios and 95% confidence intervals from multiple-event Cox proportional hazards regression. The value of HDL cholesterol with the lowest
hazard ratio was chosen as reference. P for interaction was from Wald test. The light green areas indicate the distribution of concentrations of HDL
cholesterol. The total number of individuals varies slightly with the availability of the variables used for stratification.

.. of infectious disease for LDL cholesterol did not attenuate the

..
.. association rather it was strengthened (data not shown).
Fraction of population (Density)

Events: 1335 ..
disease related death (95% CI)

6 1
..
Hazard ratio for infectious

5 ..
.. Discussion
4 ..
..
3 0.5 .. This study showed that low and high concentrations of HDL choles-
.. terol found in 21% and 8% of the individuals in the general population
2 ..
.. were associated with high risk of infectious disease (Take home figure).
1 ..
.. The highest risk estimates were seen for gastroenteritis and bacterial
0 0 .. pneumonia. These findings are novel and were confirmed for low
mmol/L: 0 1 2 3
..
mg/dL: 0 39 77 116
.. HDL cholesterol in an independent cohort. Importantly, these obser-
HDL cholesterol
.. vational findings do not necessarily indicate causality.
..
..
Figure 6 HDL cholesterol on a continuous scale and risk of infec- .. Possible mechanisms and comparison
tious disease-related death in 97 166 individuals from the ..
Copenhagen General Population Study. Analyses were conducted
.. with existing evidence
..
using restricted cubic splines, with hazard ratios and 95% confidence .. It is likely that the explanation for high infectious disease risk differ in
intervals from multiple-event Cox proportional hazards regression. .. individuals with low and high HDL cholesterol. Relevant for low HDL
..
The value of HDL cholesterol with the lowest hazard ratio was .. cholesterol, preclinical studies have provided several possible mecha-
chosen as reference. The light green area indicates the distribution .. nisms by which HDL might contribute to normal functioning of the
..
of concentrations of HDL cholesterol. .. immune system.9 Studies in mice have shown that HDL helps to con-
.. trol the proliferation of bone marrow progenitor cells.10,11
..
.. Specifically, HDL inhibits the proliferation of haematopoietic stem
decreased with increasing number of HDL cholesterol increasing .. cells in the bone marrow, thereby controlling the development of
..
alleles (see Supplementary material online, Table S12). However, .. immune cells and avoiding inappropriate leucopoiesis. In later phases
further adjusting the association between the allele score and risk
.. of the immune response, HDL acts as a modulator of immune cell
1188 C.M. Madsen et al.

HDL cholesterol Individuals Events Risk of any infectious disease P for Risk of any infectious disease P for
mmol/L (mg/dL) Age and sex adjusted hazard ratio (95% CI) trend Multifactorially adjusted hazard ratio (95% CI) trend

≥2.0 (≥77) 1,886 985 1 (reference) 1 (reference)

1.5-1.9 (58-77) 3,316 1,696 1.05 (0.90-1.23) 1.01 (0.86-1.18)

1.0-1.4 (39-58) 3,523 1,930 1.18 (1.01-1.39) 1.05 (0.87-1.25)

<1.0 (<39) 662 527 2.05 (1.59-2.65) <0.001 1.54 (1.16-2.03) 0.04

≥2.6 (≥100) 404 229 1.14 (0.81-1.61) 1.13 (0.80-1.61)

2.4-2.5 (93-100) 281 135 0.99 (0.66-1.48) 0.99 (0.66-1.48)

2.2-2.3 (85-93) 461 243 1 (reference) 1 (reference)

2.0-2.1 (77-85) 740 378 0.99 (0.72-1.36) 0.97 (0.70-1.34)

Downloaded from https://academic.oup.com/eurheartj/article/39/14/1181/4710060 by guest on 19 August 2020

1.8-1.9 (69-77) 1,080 562 1.03 (0.77-1.38) 1.02 (0.76-1.37)

1.6-1.7 (62-69) 1,431 725 1.10 (0.82-1.48) 1.04 (0.77-1.40)

1.4-1.5 (54-62) 1,652 837 1.07 (0.80-1.44) <0.001 0.99 (0.74-1.34) 0.01

1.2-1.3 (46-54) 1,545 844 1.21 (0.89-1.63) 1.09 (0.80-1.49)

1.0-1.1 (39-46) 1,131 658 1.36 (1.00-1.85) 1.10 (0.80-1.53)

0.8-0.9 (31-39) 536 425 1.93 (1.35-2.75) 1.51 (1.04-2.20)

<0.8 (<31) 126 102 3.37 (2.02-5.64) 1.99 (1.16-3.42)

0.75 1.0 1.5 2.0 3.0 4.0 0.75 1.0 1.5 2.0 3.0 4.0
Hazard ratio (95% CI) Hazard ratio (95% CI)

Figure 7 HDL cholesterol and risk of any infectious disease in 9387 individuals from the Copenhagen City Heart Study. Individuals are in the top
part divided into 0.5 mmol/L (19 mg/dL) intervals and in the bottom part into 0.2 mmol/L (8 mg/dL) intervals of HDL cholesterol. Hazard ratios were
from multiple-event Cox proportional hazards regression.

function and activity, which in part is believed to be based on the abil-

..
..
ity of HDL to change the cholesterol content in lipid rafts in the cell ..
membrane, thereby affecting the activity of different cell surface
..
..
receptors and the function of immune cells.9,12 Additionally, HDL ..
..
carries lipopolysaccharide-binding protein (LBP), which enables neu- ..
tralization and clearance of endotoxins.20 In future studies, it would ..
..
be interesting to determine the association between HDL concentra- ..
tions and concentrations of LBP in the HDL particles and how this ..
..
affects infectious disease risk. Of note, potential defensive activities of ..
HDL might also happen at mucosal barriers where HDL is secreted ..
..
rather than in the blood stream.
Relevant for high HDL cholesterol, the very high concentrations of
HDL cholesterol could be due to genetic variants that might have
detrimental effects affecting disease susceptibility such is the case for
certain mutations in, e.g. LIPC and SCARB1, which are associated with
high risk of coronary heart disease.21,22 Alternatively, the functional
property of the HDL particles may be compromised in individuals
with very high HDL cholesterol leading HDL to no longer function
properly. HDL is an extremely complex particle that varies greatly in
size and composition, and it has been attributed to several different
functions, including being anti-inflammatory.23 Hence, it is possible
that some people with extreme high concentrations of HDL choles-
terol may have HDL particles that are dysfunctional.
Reports on the association between HDL concentrations and
infectious disease risk in humans are limited to five studies with 204
to 1719 hospitalized patients.13–17 In accordance with our results,
these studies have shown an association between low concentrations
of HDL cholesterol at admission and high risk of infectious disease
during hospital admission or shortly following. To the best of our
knowledge, this study is the first one to examine the relationship
between concentrations of HDL cholesterol and risk of infectious
disease in two large prospective studies of the general population.
In addition to the high risk of infection seen at low concentrations
of HDL cholesterol, we also observed high risk of infectious disease
among those with high concentrations of HDL cholesterol in the
Copenhagen General Population Study but not in the Copenhagen
City Heart Study. An explanation could be that high concentrations
of HDL cholesterol can be due to genetic variations that could have
detrimental effects, as shown for cardiovascular disease.22,24
.. Interestingly, in the ILLUMINATE trial with the CETP inhibitor torce-
.. trapib, the torcetrapib group, with elevated HDL cholesterol, had
..
.. more infectious disease-related deaths than the placebo group with
.. lower HDL cholesterol (9 vs. 0 events).25 This might reflect a qualita-
..
.. tive difference in HDL raised in response to CETP inhibition different
.. from that of ambient HDL measured in the general population.
..
.. Concentrations of HDL cholesterol show an inverse correlation
.. with concentrations of triglycerides2; however, adjusting for triglycer-
..
.. ides in the multifactorially adjusted analysis gave similar risk estimates
..
.. and plasma concentrations of triglycerides were not associated with
.. the risk of infectious disease. Hence, it is unlikely that the correlation
..
.. between HDL cholesterol and triglycerides explained the findings in
.. this study. Furthermore, low concentrations of apolipoprotein A1
..
.. were also associated with high risk of infectious disease similar to
.. HDL cholesterol. In light of the association with HDL cholesterol,
..
.. the absence of any relationship between risk of infection disease and
.. plasma triglyceride concentrations is surprising. This could be
..
.. explained by the fact that HDL cholesterol or triglycerides are not
.. perfectly correlated and that it really is HDL per se that is important
U-shaped relationship of HDL and risk of infectious disease 1189

All individuals HDL cholesterol ≤50th percentile HDL cholesterol >50th percentile

infectious disease (95% CI) 1.05 P for trend 0.045 P for trend 0.47 P for trend 0.03
Hazard ratio for any

1.00

0.95

0.90

0.85

2.4 +17%
+12%
+5%
2.0 0%
HDL cholesterol, mmol/L

+17%

Downloaded from https://academic.oup.com/eurheartj/article/39/14/1181/4710060 by guest on 19 August 2020

+12%
+6%
1.6 0%
+19%
+6% +12%
1.2 0%

0.8

0.4

0
0 1 2 3-4 0 1 2 3-4 0 1 2 3-4
No. of HDL cholesterol increasing alleles No. of HDL cholesterol increasing alleles No. of HDL cholesterol increasing alleles
No. of individuals: 19,374 41,198 30,089 9,454 9,640 20,643 15,091 4,717 9,734 20,555 14,998 4,737
No. of events: 6,413 13,109 9,641 2,906 3,360 6,904 5,276 1,527 3,053 6,205 4,365 1,379

Figure 8 Genetically determined HDL cholesterol and risk of any infectious disease in 100 115 individuals from the Copenhagen General
Population Study and Copenhagen City Heart Study combined. Lower panel shows the mean (±SEM) concentration of HDL cholesterol for individu-
als with the different allele scores with the percent change in HDL cholesterol compared to individuals in the group with zero HDL cholesterol
increasing alleles. Upper panel shows the risk of any infectious disease with hazard ratios from multiple-event Cox proportional hazards regression
adjusted for age and gender. For this analysis, infectious disease events were included from 1977 (start of the registry) or from birth, whichever came
last. The left column includes all individuals, whereas the centre and right columns include individuals with HDL cholesterol below or above the
median. Individuals were divided by the median of HDL cholesterol after the effect of the genotypes on the concentration of HDL cholesterol was
removed.

..
for the risk of infectious disease. In addition, there are situations .. association between HDL cholesterol and the risk of infectious dis-
where a very high level of HDL cholesterol is associated with a high .. ease after multifactorial adjustment, indicating that part of the associ-
..
rather than a low concentration of plasma triglycerides, as is seen in .. ation could be explained by possible confounders. HDL cholesterol
people who consume excessive amounts of alcohol. However, strati-
.. is reduced in many chronic conditions which themselves are associ-
..
fying for alcohol intake provided similar results in those with low and .. ated with high risk of infectious disease such as diabetes.28 However,
high alcohol intake.
.. we did not observe an association between high concentrations of
..
Although recent results from genetic studies and randomized clini- .. triglycerides, which are often present in individuals with poor health
..
cal trials have failed to show a beneficial effect of raising HDL choles- .. status, and risk of infectious disease, and also we adjusted and strati-
terol for the prevention of cardiovascular disease, there is still a .. fied our analyses for diabetes with similar results. Furthermore, there
..
strong focus on HDL cholesterol in this context in the current .. seemed to be reduced risk of infection in those with genetically
European Society of Cardiology (ESC)/European Atherosclerosis .. higher HDL cholesterol, which could indicate a causal relationship, as
..
Society (EAS) Guidelines for the Management of Dyslipidaemias.26 .. genetic variants are not subject to reverse causation and are less
This focus should in the future perhaps be directed more at .. likely to be confounded. This is because genetic variants are present
..
triglyceride-rich remnant lipoproteins when it comes to the preven- .. at birth and because the random assortment of genetic variants dur-
tion of cardiovascular disease,27 while other possible roles of HDL
.. ing gamete formation leads to equal distribution of confounders in
..
outside of cardiovascular disease should receive additional attention, .. individuals with the different genotypes. However, the results from
such as increased all-cause mortality18 or the presently documented
.. the genetic analyses has to be interpreted with caution and needs to
..
U-formed relationship with infectious disease. .. be confirmed in additional studies, as the association between HDL
..
.. cholesterol and risk of infectious disease does not appear to be linear
Strengths and limitations .. and as the effect of the allele score on HDL cholesterol concentra-
..
Due to the observational nature of our study, we are unable to con- .. tions was only modest. Another limitation relates to the use of the
clude whether low HDL cholesterol is a causal risk factor for infec- .. national registries for diagnoses of infectious diseases. These only
..
tious disease. Thus, we cannot exclude that the association could be .. capture infections severe enough to lead to hospital contact and not
due to residual confounding or reverse causation; especially, this .. all are confirmed by microbiological examinations. In addition, there
..
could be the case at the extremes of the HDL cholesterol distribu- .. might be a referral bias for individuals who are generally less well
tion. Further to this, we also observed a strong attenuation of the
.. who are more often referred to hospital and treated for infections.
1190 C.M. Madsen et al.

Risk of infectious disease

Fraction of population (Density)

5 HDL cholesterol Triglycerides 1
Hazard ratio (95% CI)

3
0.5
2

Downloaded from https://academic.oup.com/eurheartj/article/39/14/1181/4710060 by guest on 19 August 2020

0 0
mmol/L: 0 1 2 3 0 1 2 3 4 5
mg/dL: 0 39 77 116 0 89 177 366 354 443

Take home figure HDL cholesterol and triglycerides on a continuous scale and risk of any infectious disease in 97 166 individuals from the
Copenhagen General Population Study. Analyses were conducted using restricted cubic splines, with hazard ratios and 95% confidence intervals
from multiple-event Cox proportional hazards regression. The values of HDL cholesterol and triglycerides with the lowest hazard ratio were chosen
as the reference. The light green and yellow areas indicate the distribution of concentrations of HDL cholesterol and triglycerides respectively.

That said, such a referral bias will not operate for the endpoint infec- ... Conflict of interest: none declared.
..
tious disease-related death, and results using this endpoint were simi- ..
lar to the overall results. Also, the use of data from the national .. References
.. 1. Frikke-Schmidt R, Nordestgaard BG, Stene MC, Sethi AA, Remaley AT, Schnohr
Danish registries on infectious disease has previously been suggested ..
to provide valid results.29 Furthermore, generalizability of the results .. P, Grande P, Tybjaerg-Hansen A. Association of loss-of-function mutations in the
.. ABCA1 gene with high-density lipoprotein cholesterol levels and risk of ischemic
may be reduced by the fact that we only investigated White individu- .. heart disease. JAMA 2008;299:2524–2532.
als of Danish descent, but we are not aware of data to indicate that
.. 2. Varbo A, Benn M, Tybjærg-Hansen A, Jørgensen AB, Frikke-Schmidt R,
.. Nordestgaard BG. Remnant cholesterol as a causal risk factor for ischemic heart
the results of this study may be different in other ethnic groups. ..
Strengths include the large number of prospectively recruited indi-
.. disease. J Am Coll Cardiol 2013;61:427–436.
.. 3. Keene D, Price C, Shun-Shin MJ, Francis DP. Effect on cardiovascular risk of high
viduals from the general population, no losses to follow-up, and .. density lipoprotein targeted drug treatments niacin, fibrates, and CETP inhibitors:
detailed information on covariates. In addition, we confirmed our
.. meta-analysis of randomised controlled trials including 117,411 patients. BMJ
..
findings in an independent cohort including individuals examined in .. 4. 2014;349:g4379. Lekkou A, Mouzaki A, Siagris D, Ravani I, Gogos CA. Serum lipid profile, cytokine
another time period.
..
.. production, and clinical outcome in patients with severe sepsis. J Crit Care 2014;
.. 29:723–727.
.. 5. Chien JY, Jerng JS, Yu CJ, Yang PC. Low serum level of high-density lipoprotein
.. cholesterol is a poor prognostic factor for severe sepsis. Crit Care Med 2005;33:
Conclusions ..
.. 6. 1688–1693.
In conclusion, we found that low and high concentrations of HDL cho-
.. Guo L, Ai J, Zheng Z, Howatt DA, Daugherty A, Huang B, Li XA. High density
.. lipoprotein protects against polymicrobe-induced sepsis in mice. J Biol Chem
lesterol found in 21% and 8% of all individuals in the general population .. 2013;288:17947–17953.
.. 7. Birjmohun RS, van Leuven SI, Levels JH, van ’T Veer C, Kuivenhoven JA, Meijers
were associated with high risk of infectious.This indicates that HDL .. JC, Levi M, Kastelein JJ, van der Poll T, Stroes ES. High-density lipoprotein attenu-
may play a role in normal function of the human immune system. .. ates inflammation and coagulation response on endotoxin challenge in humans.
..
.. Arterioscler Thromb Vasc Biol 2007;27:1153–1158.
.. 8. Vanhamme L, Paturiaux-Hanocq F, Poelvoorde P, Nolan DP, Lins L, Van Den
Supplementary material .. Abbeele J, Pays A, Tebabi P, Van Xong H, Jacquet A, Moguilevsky N, Dieu M,
.. Kane JP, De Baetselier P, Brasseur R, Pays E. Apolipoprotein L-I is the trypano-
..
Supplementary material is available at European Heart Journal online. .. 9. some lytic factor of human serum. Nature 2003;422:83–87.
Catapano AL, Pirillo A, Bonacina F, Norata GD. HDL in innate and adaptive
..
.. immunity. Cardiovasc Res 2014;103:372–383.
Acknowledgements .. 10. Blaho VA, Galvani S, Engelbrecht E, Liu C, Swendeman SL, Kono M, Proia RL,
We would like to thank the staff and participants from the .. Steinman L, Han MH, Hla T. HDL-bound sphingosine-1-phosphate restrains lym-
..
Copenhagen General Population Study and the Copenhagen City .. 11. phopoiesis and neuroinflammation. Nature 2015;523:342–346.
Yvan-Charvet L, Pagler T, Gautier EL, Avagyan S, Siry RL, Han S, Welch CL, Wang
Heart Study for their valuable contributions.
.. N, Randolph GJ, Snoeck HW, Tall AR. ATP-binding cassette transporters and HDL
..
.. suppress hematopoietic stem cell proliferation. Science 2010;328:1689–1693.
Funding .. 12. Wang SH, Yuan SG, Peng DQ, Zhao SP. HDL and ApoA-I inhibit antigen
.. presentation-mediated T cell activation by disrupting lipid rafts in antigen pre-
Herlev and Gentofte Hospital, Copenhagen University Hospital, ..
Denmark, Chief Physician Johan Boserup and Lise Boserup’s Fund, .. 13. senting cells. Atherosclerosis 2012;225:105–114.
Shor R, Wainstein J, Oz D, Boaz M, Matas Z, Fux A, Halabe A. Low HDL levels
Denmark, and the Novo Nordisk Foundation, Denmark.
.. and the risk of death, sepsis and malignancy. Clin Res Cardiol 2008;97:227–233.
U-shaped relationship of HDL and risk of infectious disease 1190a

14. Delgado-Rodrı́guez M, Medina-Cuadros M, Martı́nez-Gallego G, Sillero-Arenas

..
..
M. Total cholesterol, HDL-cholesterol, and risk of nosocomial infection: a pro- ..
spective study in surgical patients. Infect. Control Hosp Epidemiol 1997;18:9–18. ..
15. Rodriguez-Sanz A, Fuentes B, Martinez-Sanchez P, Prefasi D, Martinez-Martinez ..
M, Correas E, Diez-Tejedor E. High-density lipoprotein: a novel marker for risk
of in-hospital infection in acute ischemic stroke patients? Cerebrovasc Dis 2013;
35:291–297.
16. Grion CM, Cardoso LT, Perazolo TF, Garcia AS, Barbosa DS, Morimoto HK,
Matsuo T, Carrilho AJ. Lipoproteins and CETP levels as risk factors for severe
sepsis in hospitalized patients. Eur J Clin Invest 2010;40:330–338.
17. Canturk NZ, Canturk Z, Okay E, Yirmibesoglu O, Eraldemir B. Risk of nosoco-
mial infections and effects of total cholesterol, HDL cholesterol in surgical
patients. Clin Nutr 2002;21:431–436.
18. Madsen CM, Varbo A, Nordestgaard BG. Extreme high high-density lipoprotein
cholesterol is paradoxically associated with high mortality in men and women:
two prospective cohort studies. Eur Heart J 2017;38:2478–2486.
22. Zanoni P, Khetarpal SA, Larach DB, Hancock-Cerutti WF, Millar JS, Cuchel M,
DerOhannessian S, Kontush A, Surendran P, Saleheen D, Trompet S, Jukema JW,
De Craen A, Deloukas P, Sattar N, Ford I, Packard C, Majumder A, Alam DS, Di
Angelantonio E, Abecasis G, Chowdhury R, Erdmann J, Nordestgaard BG,
.. Nielsen SF, Tybjaerg-Hansen A, Schmidt RF, Kuulasmaa K, Liu DJ, Perola M,
.. Blankenberg S, Salomaa V, Mannisto S, Amouyel P, Arveiler D, Ferrieres J,
.. Muller-Nurasyid M, Ferrario M, Kee F, Willer CJ, Samani N, Schunkert H,
.. Butterworth AS, Howson JM, Peloso GM, Stitziel NO, Danesh J, Kathiresan S,
.. Rader DJ; Consortium CHDE, Consortium CAE, Global Lipids Genetics
..
.. Consortium. Rare variant in scavenger receptor BI raises HDL cholesterol and
.. increases risk of coronary heart disease. Science 2016;351:1166–1171.
.. 23. Barter PJ, Rye KA. HDL cholesterol concentration or HDL function: which mat-
.. ters? Eur Heart J 2017;38:2487–2489.
.. 24. Agerholm-Larsen B, Nordestgaard BG, Steffensen R, Jensen G, Tybjaerg-Hansen
.. A. Elevated HDL cholesterol is a risk factor for ischemic heart disease in white
.. women when caused by a common mutation in the cholesteryl ester transfer
..

Downloaded from https://academic.oup.com/eurheartj/article/39/14/1181/4710060 by guest on 19 August 2020

19. Nordestgaard BG, Langsted A, Mora S, Kolovou G, Baum H, Bruckert E, Watts ..
GF, Sypniewska G, Wiklund O, Boren J, Chapman MJ, Cobbaert C, Descamps ..
OS, von Eckardstein A, Kamstrup PR, Pulkki K, Kronenberg F, Remaley AT, Rifai
..
..
N, Ros E, Langlois M; European Atherosclerosis Society (EAS) and the European ..
Federation of Clinical Chemistry and Laboratory Medicine (EFLM) joint consen- ..
sus initiative. Fasting is not routinely required for determination of a lipid profile: ..
clinical and laboratory implications including flagging at desirable concentration ..
cut-points-a joint consensus statement from the European Atherosclerosis
..
..
Society and European Federation of Clinical Chemistry and Laboratory Medicine.
Eur Heart J 2016;37:1944–1958.
20. Wurfel MM, Kunitake ST, Lichenstein H, Kane JP, Wright SD. Lipopolysaccharide
(LPS)-binding protein is carried on lipoproteins and acts as a cofactor in the neu-
tralization of LPS. J Exp Med 1994;180:1025–1035.
21. Andersen RV, Wittrup HH, Tybjaerg-Hansen A, Steffensen R, Schnohr P,
Nordestgaard BG. Hepatic lipase mutations, elevated high-density lipoprotein
cholesterol, and increased risk of ischemic heart disease: the Copenhagen City
Heart Study. J Am Coll Cardiol 2003;41:1972–1982.
protein gene. Circulation 2000;101:1907–1912.
25. Barter PJ, Caulfield M, Eriksson M, Grundy SM, Kastelein JJ, Komajda M, Lopez-
Sendon J, Mosca L, Tardif JC, Waters DD, Shear CL, Revkin JH, Buhr KA, Fisher
MR, Tall AR, Brewer B; ILLUMINATE Investigators. Effects of torcetrapib in
patients at high risk for coronary events. N Engl J Med 2007;357:2109–2122.
26. Catapano AL, Graham I, De Backer G, Wiklund O, Chapman MJ, Drexel H,
Hoes AW, Jennings CS, Landmesser U, Pedersen TR, Reiner Z, Riccardi G,
Taskinen MR, Tokgozoglu L, Verschuren WM, Vlachopoulos C, Wood DA,
Zamorano JL; Authors/Task Force M, Additional C. 2016 ESC/EAS Guidelines
.. for the management of dyslipidaemias. Eur Heart J 2016;37:2999–3058.
.. 27. Nordestgaard BG. Triglyceride-rich lipoproteins and atherosclerotic cardiovascu-
.. lar disease: new insights from epidemiology, genetics, and biology. Circ Res 2016;
.. 118:547–563.
.. 28. Haase CL, Tybjærg-Hansen A, Nordestgaard BG, Frikke-Schmidt R. HDL choles-
.. terol and risk of type 2 diabetes: a Mendelian randomization study. Diabetes
..
.. 2015;64:3328–3333.
.. 29. Benfield T, Jensen JS, Nordestgaard BG. Influence of diabetes and hyperglycaemia
. on infectious disease hospitalisation and outcome. Diabetologia 2007;50:549–554.