Lower serum levels of total cholesterol are associated with higher urinary levels of 8-hydroxydeoxyguanosine
Department of Epidemiology and Prevention, Clinical Research Center, National Center for Global Health and Medicine
Shinjuku-ku, Tokyo 162-8655
Background: Lower serum total (TC), high-density lipoprotein (HDL-C) and low-density lipoprotein cholesterols (LDL-C) have been linked to an increased risk of cancer in various sites, but its underlying mechanism remains unclear. In an attempt to clarify the association between cholesterol levels and oxidative DNA damage, we investigated the relationship between serum cholesterol and urinary 8-hydroxydeoxyguanosine levels in a Japanese working population. Methods: The study subjects were 294 men and 209 women aged 21-66 years in two Japanese municipal offices. Urinary 8-hydroxydeoxyguanosine (8-OHdG) was measured using an automated high-pressure liquid chromatography. Linear regression analysis was used to examine the associations of urinary 8-OHdG with TC, HDL-C and LDL-C levels with adjustment for sex, age, smoking and body mass index. Subgroup analyses were conducted by smoking status in men and age in women. Analysis of covariance was employed to estimate adjusted means of urinary 8-OHdG across TC category. Results: After multivariate adjustment, urinary 8-OHdG levels were inversely associated with serum TC levels ( = 0.0015, p < 0.05) and LDL-C levels ( = 0.0012, p = 0.07). The inverse association with TC was apparent among smoking men ( = 0.0017, p < 0.05) and among women aged less than 48 years ( = 0.0040, p < 0.01). 8-OHdG decreased as TC increased (up to 219 mg/dL); subjects with TC levels of <160 mg/dL had a 17.4% higher adjusted mean of 8-OHdG than did those with TC levels of 200-219 mg/dL. Conclusion: Results suggest that circulating low TC levels are associated with higher oxidative DNA damage.
Lower circulating total (TC) [1-3], high-density lipoprotein
(HDL-C)  and low-density lipoprotein cholesterols
(LDL-C)  have been shown to be associated with an
increase risk of total cancer, as well as cancer of the
lung, prostate, stomach or colon [6,7]. Although these
observed associations have been ascribed to the
presence of preclinical cancer that decreases cholesterol
levels [8,9], the association remained significant in several
studies even after excluding cancer incidence that occurred
during early period of follow up [2,6]. Thus, controversy
continues whether low cholesterol increases cancer risk.
A clue to this question may be obtained from biomarker
study linking low cholesterol to carcinogenesis in a healthy
population. Oxidative DNA stress is thought to play a major
role in carcinogenesis . 8-hydroxydeoxyguanosine
(8-OHdG) is a reliable marker of oxidative DNA damage
and its concentrations in urine have been shown to predict
cancer risk . To our knowledge, only two studies
examined the association between urinary 8-OHdG and serum
cholesterol and their results are conflicting. Sakano et.al.
reported a positive association between LDL-C and urinary
8-OHdG and no association between TC and 8-OHdG .
In contrast, Miyamaoto et al. found no apparent association
between TC/HDL-C and urinary 8-OHdG . These
results, however, may be limited due to the use of ELISA
method in measuring 8-OHdG, which is less accurate than
the high performance liquid chromatography (HPLC)
We hypothesized that oxidative DNA damage would be
increased in persons with low TC/HDL-C/LDL-C levels.
Here, we examined the association between these lipids
and urinary 8-OHdG levels measured by using HPLC in
an apparently healthy working population.
Figure 1 shows flowchart of study protocol. In July and
November, 2006, a health survey was conducted among
municipal employees of two offices in northeastern Kyushu,
Japan, as described elsewhere . Full-time employees
were invited to participate in the survey (n = 601) and 547
responded (age range, 21-67 years; response rate, 91.0%).
Study participants were asked to fill out the survey
questionnaire beforehand, which were checked by research
staff for completeness. Also obtained were data that were
routinely collected in the health examination, including job
title, anthropometric measurements, biochemical data,
and information about medical history, smoking, physical
activity and alcohol drinking.
We excluded subjects with a history of cancer (n = 8),
cardiovascular disease (n = 4), cerebrovascular diseases
(n = 6) or diabetes (n = 12), subjects who were currently
under the care of physician due to chronic liver disease
(n = 2) or nephritis (n = 1), subjects who were receiving
medication for hyperlipidemia (n = 3) and pregnant women
(n = 2). We further excluded subjects who had missing
information on any of the variables used in the present
analysis. Some participants had two or more conditions
for exclusion. After exclusion, 503 (294 Japanese men
and 209 women) were eligible for analysis. The protocol of
the health study was approved by the ethics committee of
601 full-time employees in two municipal offices
547 participants in the survey
History of inflammatory diseases (n=23)
Missing data on 8-OHdG levels (n=11)
503 subjects (294 men and 209 women)
Figure 1 Flowchart of study protocol.
the National Center for Global Health and Medicine,
and the written informed consent was obtained from
Measurement of serum cholesterol levels and urinary 8-OHdG
During the health checkup, blood and urine samples were
obtained. Serum TC, HDL-C and triglycerides were
measured by using Quick auto neo T-Cho (Shino-Test Co.,
Tokyo), Cholestest N HDL (Daiichi Kagaku-Co., Tokyo),
Wako L-Type (Wako Co. Tokyo), respectively. Among
participants with fasting blood samples (n = 481), LDL-C
was estimated according to the Friedewald formula:
[LDL-C = TC-HDL-C-triglycerides/5]. Urinary 8-OHdG and
creatinine were determined by HPLC method . The
accuracy of the measurement, estimated from the
recovery of an added 8-OHdG standard, was 90-98%.
When the same urine sample was analyzed three times,
the variation of the data was within 7%.
For continuous variables, normality was checked before
statistical analysis. 8-OHdG levels were adjusted for urinary
creatinine levels and log-transformed for association
analyses. Multiple linear regression analyses were used
to examine the associations of urinary 8-OHdG with
TC, HDL-C and LDL-C levels while adjusting for sex,
age (continuous), smoking (smoker or nonsmoker
including past smoker) and body mass index (BMI) (kg/m2,
continuous). To test the interaction, we added an
interaction term of sex and each TC, HDL-C, LDL-C levels
to the model. Subgroup analyses were conducted by
smoking status in men (smoker and nonsmoker) and
age group in women (<48 and 48 years, the median age
of menopause in Japanese women ). Due to few female
smokers (n = 4), subgroup analysis by smoking status was
not performed in women.
After subjects were divided into seven groups according
to TC concentrations (<160, 160-179, 180-199, 200-219,
220-239, 240-259 or 260 mg/dL), analysis of covariance
was employed to estimate the means of log-transformed
8-OHdG and their standard errors. The means and
their 95% confidence limits in log-scale were then
backtransformed. A trend test was performed by treating TC as
a continuous variable in a multivariable regression analysis.
We repeated the above analyses by adding other covariates,
including physical activity, alcohol consumption,
hypertension status and job-title. All statistical tests were two-tailed
and considered to be statistically significant at the 0.05
level. All analyses were done with Stata, version 9.1.
The 294 subjects were men (58.4%), the mean age was
42.4 years old (range 21-66), and mean (SD) BMI was
22.5 3.5 kg/m2. As shown in Table 1, age and BMI
Table 1 Characteristics of study subjects by serum total cholesterol concentrations
BMI (S.D.), kg/m2
BMI Body mass index.
Serum total cholesterol concentrations (mg/dL)
180-199 200-219 220-239
106 122 85
were positively correlated with TC levels. Mean (SD)
TC was 206 33 mg/dL (men: 209 33 mg/dL; women:
201 34 mg/dL), HDL-C was 61.5 16 mg/dL (men:
55.7 14 mg/dL; women: 69.8 15 mg/dL) and LDL-C
was 122 30 mg/dL (men: 125 30 mg/dL; women:
117 30 mg/dL). Median (interquartile range) 8-OHdG
concentration was 3.01 (2.37, 4.03) g/g creatinine (men:
3.09 [2.43, 3.97]; women; 2.99 [2.24, 4.21]), ranging from
0.8 to 10.0 g/g creatinine.
Table 2 shows the results of multiple linear regression
analysis. For overall subjects, urinary 8-OHdG levels were
significantly, inversely associated with serum TC levels
( = 0.0015, p < 0.01) and marginally, inversely associated
with LDL-C levels ( = 0.0012, p = 0.07) with adjustment
for age, sex, smoking status and body mass index. Urinary
8-OHdG levels were not associated with HDL-C levels
( = 0.0010, p = 0.46). There was no significant interaction
of sex on the association between urinary 8-OHdG and
Subgroup analysis showed that the inverse association
with TC was statistically significant among smoking men
( = 0.0017, p < 0.05) and among women aged less than
48 years ( = 0.0040, p < 0.01), but not among
nonsmoking men ( = 0.0015, p = 0.15) and among women
aged 48 years or older ( = 0.0014, p = 0.47). In women,
urinary 8-OHdG levels were significantly, inversely
associated with LDL-C ( = 0.0030, p < 0.05). Moreover, a
marginally significantly, inversely association between with
HDL-C and urinary 8-OHdG levels was observed among
women aged less than 48 years ( = 0.0052, p = 0.08).
Figure 2 shows mean urinary 8-OHdG by TC levels with
adjustment for age, sex, BMI and smoking status. There
was a decreasing trend of 8-OHdG with increasing TC
(up to 219 mg/dL). The multivariate adjusted means
(standard errors) of urinary 8-OHdG levels from the lowest
to the highest group of serum TC levels were 3.36 (0.50),
3.19 (0.34), 3.08 (0.26), 2.86 (0.22), 3.02 (0.29), 2.82 (0.37),
2.94 (0.46) (p for trend < 0.05). Subjects with TC levels
of <160 mg/dL had a 17.4% higher adjusted mean of
8-OHdG than did those with TC levels of 200-219 mg/dL.
Results were virtually unchanged after adding other
covariates, including alcohol drinking, physical activity, job-title
and hypertension status.
We hypothesized that oxidative DNA damage is
increased in persons with low cholesterol levels. In the
present study of a healthy working population, we found
a statistically significant inverse association between
serum TC and urinary 8-OHdG concentrations even
Table 2 Regression coefficients in multiple linear regression analyses between serum cholesterols and urinary 8-OHdG
8-OHdG 8-hydroxydeoxyguanosine, S.E standard error, HDL high-density lipoprotein, LDL low-density lipoprotein.
All analyses were adjusted for sex, age, body mass index and smoking status.
LDL-cholesterol was estimated by applying the Friedewalds formula (LDL cholesterol = total cholesterol - HDL cholesterol - trigriceride/5) only among subjects
with fasting blood samples (n = 481).
Serum total cholesterol (mg/dL)
Figure 2 Mean urinary 8-hydroxydeoxyguanosine (8-OHdG) levels by serum total cholesterol concentrations. Bar indicates geometric
means levels. Error bar indicates upper limit of 95% confidence interval. *Adjusted for age, sex, body mass index and smoking status.
after adjustment of confounding factors. Subgroup
analysis showed that the inverse association was more
pronounced in smoking men and younger women. This
is one of few epidemiological studies which examined
the association between TC levels and oxidative DNA
Available data on associations between circulating
cholesterol levels and markers of oxidative DNA damage
are limited and conflicting. In a study of 677 middle-aged
city officers, urinary 8-OHdG concentrations were
statistically, positively associated with LDL-C, but not with
TC . In another study among 90 non-smokers who
participated in a health program, a weak non-significant
inverse association was observed between TC and urinary
8-OHdG levels (r = 0.155) . In measuring urinary
8-OHdG, these studies used ELISA method, which is
less accurate than HPLC method .
In the present study, a significant inverse association
between TC and urinary 8-OHdG levels was found
among smokers ( = 0.0017, p < 0.01) but not among
non-smokers ( = 0.0015, p = 0.15) in men. However,
the estimated slope () of smokers was similar to that of
nonsmokers. In women, the association was more apparent
among women aged less than 48 years ( = 0.0040,
p < 0.01) than those aged 48 years or older ( = 0.0014,
p = 0.47). This result may suggest that menopausal status
may modify the association between circulating TC and
oxidative stress. Such differential association has also been
observed in epidemiological studies on cancer. For instance,
low serum TC was associated with an increased risk of
breast cancer among pre-menopausal women, but not
among post-menopausal women .
Biological mechanism underlying the association between
lower serum TC and higher oxidative DNA damage
remains unclear. Cholesterol is an important structural
lipid for maintaining cell functions. In addition, cholesterol
plays an important role in the absorption of lipid-soluble
vitamins including vitamin E, the major membrane-bound
antioxidant, and controls the flow of these vitamins in and
out of cell membranes [19,20]. Persons with low cholesterol
levels may thus have a decreased antioxidant capacity,
leading to increased oxidative DNA damage.
In women aged less than 48 years, urinary 8-OHdG
levels were significantly, inversely associated with LDL-C
( = 0.0041, p < 0.05), and marginally significantly
associated with HDL-C levels ( = 0.0052, p = 0.08). Plausible
mechanisms underlying these associations are unclear, but
the following has been suggested. HDL-C can prevent
oxidation of LDL-C . Thus, LDL-C derived oxidized stress
might be increased among subjects with lower levels of
HDL-C. Meanwhile, LDL-C level itself is not associated
with oxidized LDL-C level . Although LDL-C might be
partially oxidized, most of the remaining unoxidized LDL-C
might play a protective role against oxidative stress.
Major strengths of the present study include high
response rate (91%), control of known background and
lifestyle factors associated with oxidative DNA damage,
and use of reliable technique (HPLC) in measuring
biomarker of oxidative DNA damage . Limitations
of the study also warrant mention. The cross-sectional
design of this study does not allow us to infer causality.
Because study participants were employees of municipal
offices in Japan, the present findings may not be applicable
to populations with different background. Finally, the
sample size of this study may not be sufficient to detect
a modest association.
In conclusion, serum TC levels were inversely
associated with urinary 8-OHdG concentrations in a healthy
working population. This finding suggests that
oxidative DNA damage is increased in persons with low
cholesterol levels, and thus may support a link of low
TC to carcinogenesis.
8-OHdG: 8-hydroxydeoxyguanosine; HPLC: The high performance liquid
chromatography; BMI: Body mass index; TC: Total cholesterol; LDL-C:
Low-density lipoprotein cholesterol; HDL-C: High-density lipoprotein
The authors declare that they have no competing interest.
TM, HKA, MS and KK developed study design. TM, AN and AH collected data.
HKI, AN and TM performed statistical analysis and prepared draft version of
the manuscript. All authors critically revised the manuscript. All authors read
and approved the final manuscript.
This work was supported by a Grant-in-Aid for the Third Term
Comprehensive 10-Year Strategy for Cancer Control from the Ministry
of Health, Labour and Welfare of Japan; and Grant-in-Aids for Scientific
Research (C) (no. 18590601) and (B) (no. 21390213) from the Japan Society
for the Promotion of Science (To Dr Mizoue). We thank Tamani Hatano,
Yasumi Kimura, Akihiro Tanaka and Yuko Ejima (Kyushu University) Mio
Ozawa (Fukuoka Womens University): and Akiko Hayashi and Kie Nagao
(International Medical Center of Japan) for their help in data collection.