Some prospective epidemiologic studies have suggested that a low plasma cholesterol level may be associated with increased risk of cancer. Certain sequence variants in the proprotein convertase subtilisin/kexin type 9 serine protease gene (PCSK9) are associated with lifelong low total and LDL cholesterol. We therefore analyzed the association of PCSK9 variation with incidence of cancer between 1987 and 2000 in a prospective study (n = 13,250). The frequency of the PCSK9 variants studied was 2.4% in blacks and 3.2% in whites. Neither was associated with increased cancer incidence: age- and sex-adjusted hazard ratios were 0.66 [95% confidence interval (95% CI), 0.31-1.39] in blacks and 0.77 (95% CI, 0.54-1.09) in whites. Low baseline total or LDL cholesterol levels in 1987 to 1989 were also not statistically significantly associated with incident cancer: multivariable-adjusted hazard ratios for the lowest compared with the highest quartiles of LDL cholesterol were 1.05 (95% CI, 0.78-1.40) in blacks and 1.16 (95% CI, 0.99-1.36) in whites. These data suggest that a lifelong low cholesterol concentration, as reflected by these PCSK9 variants, does not increase risk of cancer. (Cancer Epidemiol Biomarkers Prev 2007;16(11):2455–8)

By the early 1990s, several prospective epidemiologic studies had reported a modest association between low plasma cholesterol concentrations and increased risk of cancer, particularly colon and lung cancer (1, 2). Reasons for this observation were unclear, but general consensus was that an association between low cholesterol and cancer was likely not a cause and effect relation. Instead, the epidemiologic association was felt to likely suffer from confounding or from “reverse causality” (i.e., that subclinical cancer could reduce cholesterol levels; refs. 1, 3).

In that era, before widespread 3-hydroxy-3-methylglutaryl CoA reductase inhibitor (statin) trials, there also was concern from rodent studies and human trials that lowering plasma cholesterol pharmacologically might modestly increase cancer risk (4, 5). However, cholesterol lowering with statins does not seem to increase cancer risk (6-8), so concern about low cholesterol leading to cancer has lessened. The National Cholesterol Education Program's Adult Treatment Panel III in 2001 (9) concluded, “There is no evidence that currently used cholesterol-lowering drugs promote the development of cancer or induce subtle neurologic diseases. Moreover, clinical experience with these drugs over periods of 30 years for fibrates and bile acid sequestrants and 15 years for statins has uncovered no long-term side effects. Nonetheless, the possibility of long-term side effects, albeit remote, should be one factor to consider when recommending lifetime therapy with a cholesterol-lowering drug.”

One way to further test whether long-term low plasma cholesterol levels may increase cancer risk is to compare cancer risks of people with or without genetic variants that lead to lifelong low cholesterol levels using the approach called “Mendelian randomization” (10). Cohen et al. (11) recently showed that plasma LDL cholesterol concentrations are substantially reduced in individuals with certain sequence variations in the proprotein convertase subtilisin/kexin type 9 serine protease gene (PCSK9). In blacks, two nonsense mutations (142X and 679X) were associated with a 28% reduction in LDL cholesterol and an 88% reduction of coronary heart disease risk in the Atherosclerosis Risk in Communities (ARIC) Study. In whites, a single nucleotide polymorphism sequence variant (46L) was associated with a 15% reduction of LDL cholesterol and a 47% reduction of coronary heart disease risk.

Whether these PCSK9 variants linked to lifetime low LDL cholesterol concentrations are associated with cancer risk is unknown. We therefore examined this question in the prospective, population-based cohort of the ARIC Study.

Population

The ARIC Study is a cohort study of cardiovascular disease in whites and African-Americans in four U.S. communities. Between 1987 and 1989, 7,082 men and 8,710 women ages 45 to 64 years were recruited from Forsyth County, North Carolina; Jackson, Mississippi (African-Americans only); suburban Minneapolis, Minnesota; and Washington County, Maryland. There was a 46% response in the Jackson cohort and a 65% to 67% response in the other three cohorts. The ARIC Study protocol was approved by the institutional review board of each participating university. After written informed consent was obtained, participants underwent a baseline clinical examination (visit 1). Follow-up examinations of the cohort occurred thrice at intervals of ∼3 years. The response rates for visits 2 (1990-1992), 3 (1993-1995), and 4 (1996-1998) were 93%, 86%, and 80%, respectively. Participants completed annual telephone interviews between visits and following visit 4.

Risk Factor Measurements

Risk factors examined in these analyses were ascertained at visit 1 as described in detail in the ARIC Study manuals of operation (12). Participants were asked to fast before the clinical examination. Blood was drawn from an antecubital vein of seated participants into vacuum tubes containing EDTA (for measurement of lipids and DNA extraction) or a serum separator gel (glucose). Serum and plasma aliquots were stored at −70°C and shipped to central laboratories for analyses. Total cholesterol and triglycerides were measured by enzymatic methods, and high-density lipoprotein cholesterol was measured after dextran-magnesium precipitation. LDL cholesterol was calculated (13). Serum glucose was assayed by a hexokinase/glucose-6-phosphate dehydrogenase method. Prevalent diabetes mellitus was defined as a fasting glucose ≥126 mg/dL (14) or a self-reported history of or treatment for diabetes.

Anthropometrics were assessed with the subject wearing a scrub suit and no shoes. Body mass index was calculated (weight in kilograms/height in meters squared). Questionnaires assessed race (self-identified), education, smoking status, number of cigarettes smoked per day and duration of smoking (pack-years computed), and usual consumption of wine, beer, and hard liquor (grams per day computed). Level of sports physical activity was assessed by the Baecke questionnaire (15).

Genotyping

Using stored DNA from all ARIC participants, fluorogenic 5′-nucleotidase assays for the PCSK9 alleles encoding Y142X, C679X, and R46L were done with the use of the Taqman system (Applied Biosystems). The assays were carried out on a 7900HT Fast Real-time PCR instrument with probes and reagents purchased from Applied Biosystems. The assay failure rates for the R46L variant, the Y142X variant, and the C679X variant ranged from 2.7% to 2.9%. The ARIC genotyping laboratory uses a 5% blind replicate quality assurance program for genotype determinations; the agreement for the variants described here was 100%.

Cancer Ascertainment

During each clinical examination, participants were asked whether they had ever been diagnosed with cancer. At each annual telephone interview, participants reported all hospitalizations. Among those not reporting cancer at the baseline visit, incident cancers were identified between January 1, 1987 and December 31, 2000 via linkage to state cancer registries and supplemented by the hospital records. This method and the high completeness of ARIC cancer ascertainment were previously described (16, 17). For this analysis, we focused primarily on total cancer (excluding nonmelanoma skin cancer) but conducted subanalyses for common site-specific cancers (i.e., colorectal, lung, female breast, and prostate).

Data Analysis and Statistical Methods

From the original ARIC cohort (n = 15,792), we successively excluded participants who did not want to participate in cancer research (n = 187), who denied permission for DNA testing (n = 79), who were in very small minority groups (n = 96), who did not provide sufficient data to determine baseline cancer status or who had a history of cancer (n = 877), who had missing DNA or PCSK9 genotypes (n = 795), or who had not fasted 8 h (n = 508). This left 13,250 in the cohort at risk.

Statistical analysis was done by using Statistical Analysis System software (version 9.1; SAS Institute, Inc.). All analyses were conducted race specific because the PCSK9 variants associated with low LDL cholesterol overlapped little between blacks and whites (11). In analyses of total cancer, person-years at risk were calculated from the time of baseline clinical examination until the date of first cancer diagnosis, death, loss to follow-up, or December 31, 2000, whichever occurred first. In analyses of site-specific cancers, people with multiple cancers of interest were counted more than once. To explore possible confounding factors, means or prevalences of various risk factors were computed by PCSK9 genotype. Crude cancer incidence rates (per 1,000 person-years) for PCSK9 genotypes were calculated. Adjusted hazard ratios (HR) for the associations of the PCSK9 variants and of total and LDL cholesterol with cancer incidence were calculated by using Cox proportional hazards regression. The proportional hazards assumption of the Cox model was found not to be violated by testing an interaction between PCSK9 variants and time. The median values for quartiles of cholesterol were used for the χ2 test for trend in HRs.

Because the PCSK9 variants are uncommon, we estimated the detectable race-specific HRs in association with total cancer in this cohort at α = 0.05 and power = 0.8. The detectable HRs for total cancer, assuming the variants increased risk, were 1.51 for blacks and 1.26 for whites.

There were 451 incident cancers in 39,236 person-years of follow-up in blacks and 1,269 cancers in 109,617 person-years in whites. The crude incidence rates per 1,000 person-years were 11.5 in blacks and 11.6 in whites.

A low LDL cholesterol level was not a statistically significant risk factor for cancer in this cohort (Table 1). In whites, the adjusted HR for the lowest versus highest quartile of total cholesterol was 1.09 (95% CI, 0.93-1.29) and for LDL cholesterol was 1.16 (95% CI, 0.99-1.36).

Table 1.

Adjusted HRs (95% CI) of cancer by LDL and total cholesterol quartiles in blacks and whites, ARIC, 1987-2000

Quartile 4Quartile 3Quartile 2Quartile 1P trend
Blacks      
    LDL cholesterol (n = 834) (n = 835) (n = 835) (n = 834)  
        Range (mg/dL) >164 135-164 108-135 <108  
        Cases 108 113 105 92  
        Model 1* 1.0 1.11 (0.86-1.45) 1.06 (0.81-1.39) 1.01 (0.77-1.34) 0.86 
        Model 2 1.0 1.16 (0.89-1.54) 1.06 (0.80-1.40) 1.05 (0.78-1.40) 0.73 
    Total cholesterol (n = 841) (n = 833) (n = 844) (n = 842)  
        Range (mg/dL) ≥242 212-241 184-211 ≤183  
        Cases 103 111 104 104  
        Model 1* 1.0 1.13 (0.87-1.48) 1.10 (0.83-1.44) 1.13 (0.86-1.49) 0.79 
        Model 2 1.0 1.16 (0.88-1.54) 1.19 (0.89-1.58) 1.21 (0.91-1.61) 0.56 
Whites      
    LDL cholesterol (n = 2,400) (n = 2,401) (n = 2,401) (n = 2,400)  
        Range (mg/dL) >161 135-161 112-135 <112  
        Cases 333 321 297 314  
        Model 1* 1.0 1.01 (0.87-1.18) 0.97 (0.83-1.14) 1.12 (0.96-1.31) 0.30 
        Model 2 1.0 1.01 (0.86-1.18) 1.00 (0.85-1.17) 1.16 (0.99-1.36) 0.20 
    Total cholesterol (n = 2,430) (n = 2,415) (n = 2,459) (n = 2,456)  
        Range (mg/dL) ≥239 213-238 188-212 ≤187  
        Cases 335 316 329 303  
        Model 1* 1.0 1.00 (0.86-1.17) 1.06 (0.91-1.23) 1.06 (0.91-1.24) 0.80 
        Model 2 1.0 1.03 (0.88-1.20) 1.06 (0.91-1.25) 1.09 (0.93-1.29) 0.74 
Quartile 4Quartile 3Quartile 2Quartile 1P trend
Blacks      
    LDL cholesterol (n = 834) (n = 835) (n = 835) (n = 834)  
        Range (mg/dL) >164 135-164 108-135 <108  
        Cases 108 113 105 92  
        Model 1* 1.0 1.11 (0.86-1.45) 1.06 (0.81-1.39) 1.01 (0.77-1.34) 0.86 
        Model 2 1.0 1.16 (0.89-1.54) 1.06 (0.80-1.40) 1.05 (0.78-1.40) 0.73 
    Total cholesterol (n = 841) (n = 833) (n = 844) (n = 842)  
        Range (mg/dL) ≥242 212-241 184-211 ≤183  
        Cases 103 111 104 104  
        Model 1* 1.0 1.13 (0.87-1.48) 1.10 (0.83-1.44) 1.13 (0.86-1.49) 0.79 
        Model 2 1.0 1.16 (0.88-1.54) 1.19 (0.89-1.58) 1.21 (0.91-1.61) 0.56 
Whites      
    LDL cholesterol (n = 2,400) (n = 2,401) (n = 2,401) (n = 2,400)  
        Range (mg/dL) >161 135-161 112-135 <112  
        Cases 333 321 297 314  
        Model 1* 1.0 1.01 (0.87-1.18) 0.97 (0.83-1.14) 1.12 (0.96-1.31) 0.30 
        Model 2 1.0 1.01 (0.86-1.18) 1.00 (0.85-1.17) 1.16 (0.99-1.36) 0.20 
    Total cholesterol (n = 2,430) (n = 2,415) (n = 2,459) (n = 2,456)  
        Range (mg/dL) ≥239 213-238 188-212 ≤187  
        Cases 335 316 329 303  
        Model 1* 1.0 1.00 (0.86-1.17) 1.06 (0.91-1.23) 1.06 (0.91-1.24) 0.80 
        Model 2 1.0 1.03 (0.88-1.20) 1.06 (0.91-1.25) 1.09 (0.93-1.29) 0.74 
*

Adjusted for age (continuous) and sex.

Adjusted for age (continuous), body mass index (continuous), triglycerides (continuous), diabetes (yes, no), smoking status (current smoker, nonsmoker), pack-years (continuous), ethanol intake (continuous), sport index (continuous), education (<high school graduate, ≥high school graduate), and sex and current hormone replacement therapy (male and female with no hormone replacement therapy, female with hormone replacement therapy).

PCSK9 alleles associated with low LDL cholesterol were relatively uncommon. Among blacks, 2.4% had either the 142X or 679X variants (one had both). Among whites, 3.2% had the 46L variant. There was no overlap among genotypes across races. Other than the expected association with total and LDL cholesterol, these PCSK9 variants were largely unrelated to the other potential risk factors for cancer examined (Table 2).

Table 2.

Race-specific and age- and sex-adjusted means and percentages of baseline risk factors by presence of PCSK9 variants, ARIC, 1987-89

Blacks
Whites
PCSK9142X or PCSK9679X (n = 85)Neither variant (n = 3,392)P for differencePCSK946L (n = 314)Neither variant (n = 9,459)P for difference
Means       
    LDL cholesterol (mg/dL) 98.8 138.9 <0.0001 117.7 138.3 <0.0001 
    Total cholesterol (mg/dL) 172.2 215.7 <0.0001 195.9 215.4 <0.0001 
    Triglycerides (mg/dL) 95.8 111.4 0.06 140.0 137.1 0.59 
    BMI (kg/m229.3 29.6 0.69 26.9 27.1 0.69 
    Pack-years of smoking 206.5 240.5 0.41 393.6 341.3 0.03 
    Ethanol intake (g/wk) 19.0 33.0 0.18 40.2 46.1 0.26 
    Sports activity index (range, 0-5) 2.08 2.16 0.32 2.47 2.53 0.18 
Prevalences       
    High school graduate (%) 56.9 59.1 0.67 82.3 82.8 0.83 
    Diabetes (%) 12.7 16.5 0.35 8.5 8.6 0.96 
    Current smoking (%) 27.9 29.7 0.72 25.2 24.2 0.68 
    Current HRT use, women (%) 22.9 18.9 0.45 30.1 23.6 0.05 
Blacks
Whites
PCSK9142X or PCSK9679X (n = 85)Neither variant (n = 3,392)P for differencePCSK946L (n = 314)Neither variant (n = 9,459)P for difference
Means       
    LDL cholesterol (mg/dL) 98.8 138.9 <0.0001 117.7 138.3 <0.0001 
    Total cholesterol (mg/dL) 172.2 215.7 <0.0001 195.9 215.4 <0.0001 
    Triglycerides (mg/dL) 95.8 111.4 0.06 140.0 137.1 0.59 
    BMI (kg/m229.3 29.6 0.69 26.9 27.1 0.69 
    Pack-years of smoking 206.5 240.5 0.41 393.6 341.3 0.03 
    Ethanol intake (g/wk) 19.0 33.0 0.18 40.2 46.1 0.26 
    Sports activity index (range, 0-5) 2.08 2.16 0.32 2.47 2.53 0.18 
Prevalences       
    High school graduate (%) 56.9 59.1 0.67 82.3 82.8 0.83 
    Diabetes (%) 12.7 16.5 0.35 8.5 8.6 0.96 
    Current smoking (%) 27.9 29.7 0.72 25.2 24.2 0.68 
    Current HRT use, women (%) 22.9 18.9 0.45 30.1 23.6 0.05 

Abbreviations: BMI, body mass index; HRT, hormone replacement therapy.

As shown in Table 3, there was no evidence that cholesterol-lowering variants of PCSK9 were associated with increased risk of total cancer in blacks (HR, 0.66; 95% CI, 0.31-1.39) or whites (HR, 0.77; 95% CI, 0.54-1.09). This was true in blacks for both 142X and 679X when analyzed separately. There was no appreciable change in the HRs after adjusting for the other risk factors in the footnotes to Table 1 (HR of 0.78 in blacks and HR of 0.77 in whites). Numbers were limited, precluding conclusions about common site-specific cancers (Table 3).

Table 3.

Crude incidence rate and race-specific and age- and sex-adjusted HRs (95% CI) of cancer by presence of PCSK9 variants, ARIC, 1987-2000

No. developing cancerPerson-yearsIncidence rateHR (95% CI)
All cancer     
    Blacks     
        Neither variant 428 37,445 11.43  
        PCSK9142X or PCSK9679X 937 7.47 0.66 (0.31-1.39) 
    Whites     
        Neither variant 1,253 107,341 11.67  
        PCSK946L 33 3,665 9.00 0.77 (0.54-1.09) 
Colorectal cancer     
    Blacks     
        Neither variant 44 38,629 1.14  
        PCSK9142X or PCSK9679X 954 2.10 1.87 (0.45-7.71) 
    Whites     
        Neither variant 129 111,777 1.15  
        PCSK946L 3,775 1.32 1.16 (0.47-2.83) 
Lung cancer     
    Blacks     
        Neither variant 61 38,720 1.58  
        PCSK9142X or PCSK9679X 959 1.04 0.69 (0.10-4.97) 
    Whites     
        Neither variant 186 112,018 1.66  
        PCSK946L 3,793 0.53 0.32 (0.08-1.28) 
Breast cancer (women)     
    Blacks     
        Neither variant 81 23,956 3.38  
        PCSK9142X or PCSK9679X 645 1.55 0.45 (0.06-3.26) 
    Whites     
        Neither variant 259 58,027 4.46  
        PCSK946L 1,996 3.00 0.67 (0.30-1.50) 
Prostate cancer     
    Blacks     
        Neither variant 120 14,090 8.52  
        PCSK9142X or PCSK9679X 315 3.17 0.35 (0.05-2.50) 
    Whites     
        Neither variant 234 51,879 4.51  
        PCSK946L 1,739 5.18 1.14 (0.59-2.23) 
No. developing cancerPerson-yearsIncidence rateHR (95% CI)
All cancer     
    Blacks     
        Neither variant 428 37,445 11.43  
        PCSK9142X or PCSK9679X 937 7.47 0.66 (0.31-1.39) 
    Whites     
        Neither variant 1,253 107,341 11.67  
        PCSK946L 33 3,665 9.00 0.77 (0.54-1.09) 
Colorectal cancer     
    Blacks     
        Neither variant 44 38,629 1.14  
        PCSK9142X or PCSK9679X 954 2.10 1.87 (0.45-7.71) 
    Whites     
        Neither variant 129 111,777 1.15  
        PCSK946L 3,775 1.32 1.16 (0.47-2.83) 
Lung cancer     
    Blacks     
        Neither variant 61 38,720 1.58  
        PCSK9142X or PCSK9679X 959 1.04 0.69 (0.10-4.97) 
    Whites     
        Neither variant 186 112,018 1.66  
        PCSK946L 3,793 0.53 0.32 (0.08-1.28) 
Breast cancer (women)     
    Blacks     
        Neither variant 81 23,956 3.38  
        PCSK9142X or PCSK9679X 645 1.55 0.45 (0.06-3.26) 
    Whites     
        Neither variant 259 58,027 4.46  
        PCSK946L 1,996 3.00 0.67 (0.30-1.50) 
Prostate cancer     
    Blacks     
        Neither variant 120 14,090 8.52  
        PCSK9142X or PCSK9679X 315 3.17 0.35 (0.05-2.50) 
    Whites     
        Neither variant 234 51,879 4.51  
        PCSK946L 1,739 5.18 1.14 (0.59-2.23) 

In several previous cohorts that were mostly white, a low plasma cholesterol level was reported to be a risk factor for cancer (1, 2). Low total and LDL cholesterol in ARIC was not associated with an increased risk of cancer, although the HR for low LDL cholesterol in whites was slightly elevated (1.16; 95% CI, 0.99-1.36) in comparison with LDL cholesterol in the highest quartile. The novel finding reported here was that the PCSK9 variants associated with presumably lifelong low plasma cholesterol concentrations were also not related to cancer incidence. This offers further reassurance that any association between plasma total or LDL cholesterol and cancer is unlikely to be causal.

The PCSK9 variants studied were uncommon, so it would take a very large study to verify that these variants do not increase cancer risk slightly or to study site-specific cancers, which we could not effectively do. However, we had adequate power to detect an important elevation of overall cancer risk if there had been one. HRs of cancer, if anything, tended to be reduced rather than increased in carriers compared with noncarriers of the variants.

In summary, using a Mendelian randomization design, we found no evidence that variants in PCSK9 associated with low LDL cholesterol levels increase risk of total cancer.

Grant support: National Cancer Institute grant R03-CA65473 and National Heart, Lung, and Blood Institute contracts N01-HC-55015, 55016, 55018, 55019, 55020, 55021, and 55022.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank the staff and participants in the ARIC Study for their important contributions and Drs. Jonathan Cohen and Helen Hobbs for helpful discussions about PCSK9.

1
Jacobs D, Blackburn H, Higgins M, et al. Report of the conference on low blood cholesterol: mortality associations.
Circulation
1992
;
86
:
1046
–60.
2
Law MR, Thompson SG. Low serum cholesterol and the risk of cancer: an analysis of the published prospective studies.
Cancer Causes Control
1992
;
2
:
253
–61.
3
Kritchevsky SB, Wilcosky TC, Morris DL, Truong KN, Tyroler HA. Changes in plasma lipid and lipoprotein cholesterol and weight prior to the diagnosis of cancer.
Cancer Res
1991
;
51
:
3198
–203.
4
Newman TB, Hulley SB. Carcinogenicity of lipid-lowering drugs.
JAMA
1996
;
275
:
55
–60.
5
Davey Smith G, Pekkanen J. Should there be a moratorium on the use of cholesterol lowering drugs?
BMJ
1992
;
304
:
431
–4.
6
Friis S, Olsen JH. Statin use and cancer risk: an epidemiologic review.
Cancer Invest
2006
;
24
:
413
–24.
7
Setoguchi S, Glynn RJ, Avorn J, Mogun H, Schneeweiss S. Statins and the risk of lung, breast, and colorectal cancer in the elderly.
Circulation
2007
;
115
:
27
–33.
8
Dale KM, Coleman CI, Henyan NN, Kluger J, White CM. Statins and cancer risk: a meta-analysis.
JAMA
2006
;
295
:
74
–80.
9
Expert Panel on the Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Executive Summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III).
JAMA
2001
;
285
:
2486
–97.
10
Smith GD, Ebrahim S. Mendelian randomization: prospects, potentials, and limitations.
Int J Epidemiol
2004
;
33
:
30
–42.
11
Cohen JC, Boerwinkle E, Mosley TH, Jr., Hobbs HH. Sequence variations in PCSK9, low LDL, and protection against coronary heart disease.
N Engl J Med
2006
;
354
:
1264
–72.
12
ARIC Investigators. ARIC Study Web site [accessed 2007 June 1]. Available from: http://www.cscc.unc.edu/aric/.
13
Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low density lipoprotein cholesterol in plasma without use of the preparative ultracentrifuge.
Clin Chem
1972
;
18
:
499
–502.
14
The Expert Committee on the Diagnosis and Classification of Diabetes Mellitus. Report of the Expert Committee on the Diagnosis and Classification of Diabetes Mellitus.
Diabetes Care
1997
;
20
:
1183
–97.
15
Baecke JA, Burema J, Frijters JE. A short questionnaire for the measurement of habitual physical activity in epidemiological studies.
Am J Clin Nutr
1982
;
36
:
936
–42.
16
Ahmed RL, Schmitz KH, Anderson KE, Rosamond WD, Folsom AR. The metabolic syndrome and risk of incident colorectal cancer.
Cancer
2006
;
107
:
28
–36.
17
Tande AJ, Platz EA, Folsom AR. The metabolic syndrome is associated with reduced risk of prostate cancer.
Am J Epidemiol
2006
;
164
:
1094
–102.