Colorectal cancer risk and dyslipidemia: A case–cohort study nested in an Italian multicentre cohort

Cancer Epidemiology 38 (2014) 144–151

Contents lists available at ScienceDirect

Cancer Epidemiology The International Journal of Cancer Epidemiology, Detection, and Prevention journal homepage: www.cancerepidemiology.net

Colorectal cancer risk and dyslipidemia: A case–cohort study nested in an Italian multicentre cohort Claudia Agnoli a, Sara Grioni a, Sabina Sieri a,*, Carlotta Sacerdote b,c, Paolo Vineis c,d, Rosario Tumino e, Maria Concetta Giurdanella e, Valeria Pala a, Amalia Mattiello f, Paolo Chiodini g, Licia Iacoviello h, Amalia De Curtis h, Leonardo Cattaneo a, Fra¨nzel J.B. van Duijnhoven i,j, Salvatore Panico f,1, Vittorio Krogh a,1 a

Epidemiology and Prevention Unit, Fondazione IRCCS Istituto Nazionale dei Tumori, Via Venezian 1, 20133 Milan, Italy Center for Cancer Prevention (CPO-Piemonte), Via Santena 7, 10126 Turin, Italy c Human Genetics Foundation, Via Nizza 52, 10126 Turin, Italy d Imperial College of London, South Kensigton Campus, London SW7 2AZ, UK e Cancer Registry, Department of Prevention, ASP 7, Via Dante 109, 97100 Ragusa, Italy f Department of Clinical and Experimental Medicine, University of Naples Federico II, Via Pansini 5, 80131 Naples, Italy g Department of Mental and Physical Health and Preventive Medicine, Second University of Naples, Via Armanni 5, 80138 Naples, Italy h Laboratory of Genetic and Environmental Epidemiology, Fondazione di Ricerca e Cura ‘‘Giovanni Paolo II’’, Catholic University, Largo Gemelli 1, 86100 Campobasso, Italy i National Institute for Public Health and the Environment (RIVM), Antonie van Leeuwenhoeklaan 9, 3721 MA Bilthoven, The Netherlands j Division of Human Nutrition, Wageningen University, P.O. Box 8129, 6700 EV Wageningen, The Netherlands b

A R T I C L E I N F O

A B S T R A C T

Article history: Received 11 October 2013 Received in revised form 31 January 2014 Accepted 8 February 2014 Available online 11 March 2014

Background: Dyslipidemia is an established risk factor for many diseases, but its effect on colorectal cancer risk is less clear. We investigated the association of colorectal cancer risk with plasma triglycerides, total, HDL, and LDL cholesterol in four Italian EPIC centers. Methods: We conducted a case–cohort study on participants recruited to four Italian EPIC centers (Turin, Varese, Naples, and Ragusa; 34,148 subjects). A random subcohort of 850 subjects was obtained and 286 colorectal cancer cases were diagnosed. Triglycerides, total and HDL cholesterol were determined in plasma samples obtained at baseline and stored at 196 8C; LDL cholesterol was calculated. Hazard ratios (HR) with 95% confidence intervals (CI), adjusted for potential confounders, were estimated by Cox regression models using the Prentice method. Results: The highest tertiles of total (HR 1.66, 95%CI 1.12–2.45) and LDL cholesterol (HR 1.87, 95%CI 1.27–2.76) were associated with increased colorectal cancer risk compared to lowest tertiles. Risks were greater for men than women, and for postmenopausal than premenopausal women. Highest tertiles of total and LDL cholesterol were also significantly associated with increased risks of colon cancer, distal colon cancer, and rectal cancer, but not proximal colon cancer. Conclusions: Our findings suggest that high levels of total and LDL cholesterol increase colorectal cancer risk, particularly in men and postmenopausal women. However additional studies are needed to clarify the role of plasma lipids in these cancers, particularly in view of the conflicting findings of previous studies. ß 2014 Elsevier Ltd. All rights reserved.

Keywords: Prospective study Triglycerides Cholesterol Colorectal cancer

Abbreviations: BMI, body mass index; CI, confidence interval; CV, coefficient of variation; EPIC, European Prospective Investigation into Cancer and Nutrition; HDL, high density lipoprotein; HR, hazard ratio; LDL, low density lipoprotein. * Corresponding author. Tel.: +39 02 2390 3506; fax: +39 02 23903510. E-mail addresses: [email protected] (C. Agnoli), [email protected] (S. Grioni), [email protected] (S. Sieri), [email protected] (C. Sacerdote), [email protected] (P. Vineis), [email protected] (R. Tumino), [email protected] (M.C. Giurdanella), [email protected] (V. Pala), [email protected] (A. Mattiello), [email protected] (P. Chiodini), [email protected] (L. Iacoviello), [email protected] (A. De Curtis), [email protected] (L. Cattaneo), [email protected] (Fra¨nzel J.B. van Duijnhoven), [email protected] (S. Panico), [email protected] (V. Krogh). 1 Principal investigators. http://dx.doi.org/10.1016/j.canep.2014.02.002 1877-7821/ß 2014 Elsevier Ltd. All rights reserved.

C. Agnoli et al. / Cancer Epidemiology 38 (2014) 144–151

1. Introduction Dyslipidemia is an established risk factor for many diseases, particularly cardiovascular diseases [1], but it is less clear whether it is associated with greater risk of cancer. Epidemiological studies have suggested links between lipid or lipoprotein levels in blood and gynecological [2–6], breast [5,7,8], and prostate [2,9] cancer; however inverse as well as direct associations were reported. A 1974 study [10] which pooled cohort cases from six prospective cardiovascular studies on men, found that the 90 colon cancer cases had lower mean serum cholesterol levels than the overall cohort mean. A 1983 review also found an inverse relationship between blood cholesterol and colon cancer [11] as did the Framingham [12] and Honolulu Heart [13] studies. By contrast, however, a 1991 review of prospective studies [14] found no longterm association of colorectal cancer with low cholesterol. More recently some prospective studies have found that people with metabolic syndrome are at increased risk of developing colorectal cancer [15,16]. Metabolic syndrome is characterized by three or more components among abdominal obesity, high blood pressure, hyperglycemia, hypertriglyceridemia, and low HDL cholesterol [17]. Several mechanisms, mainly involving abdominal obesity and insulin resistance (typically present in metabolic syndrome) have been proposed to link the syndrome with colorectal cancer [18]. In particular, the dyslipidemia component of metabolic syndrome is linked to chronic low grade inflammation [19], oxidative stress [20] and insulin resistance [21] – all of which may enhance carcinogenesis. However, prospective studies exploring the relation of alterations in individual components of blood lipids with colorectal cancer have yielded conflicting results [2,5,6,15,16,22–24]. The aim of this case–cohort study was to investigate the association of colorectal cancer risk with plasma concentrations of triglycerides, total cholesterol, HDL cholesterol, and LDL cholesterol in participants recruited to four Italian EPIC centers [25]. 2. Materials and methods 2.1. Study population and data collection We performed the present case–cohort study, nested in participants recruited to four EPIC Italy centers, to investigate the association of colorectal cancer risk with dyslipidemia. The study formed part of a wider case–cohort design study to investigate three other outcomes (myocardial infarction, stroke, and breast cancer). The study protocol was approved by the ethics committee of the Human Genetics Foundation (Turin, Italy). The cohort consisted of 34,148 participants recruited prospectively in 1993–1998 to the four Italian EPIC Italy centers of Varese and Turin (Northern Italy), Naples (Southern Italy, women only), and Ragusa (Southern Italy). At baseline, after participants had given written informed consent, detailed information was collected on lifestyle using a standardized lifestyle questionnaire, and on diet over the previous year using food frequency questionnaires specifically developed to capture local Italian dietary habits [26]. Also at baseline, weight, height, and blood pressure were measured using standardized procedures [27] and a 30 ml fasting blood sample was collected according to a standardized protocol. The blood samples were divided into 0.5 ml aliquots of plasma, serum, red blood cells, and buffy coat, on the day of collection, and stored in liquid nitrogen at 196 8C [25]. 2.2. Case–cohort study, identification of colorectal cancer cases, and follow-up procedures A center-stratified random sample of 850 participants (284 men, 566 women) was obtained from the 34,148 participants to

145

serve as subcohort for the present study. This number was decided based on expected number of events and costs of laboratory analyses. The subcohort included two women who developed colorectal cancer during follow-up. In Varese, Turin, and Ragusa, incident colorectal cancer cases were identified by linkage to the databases of regional cancer registries. In Naples, incident cases were identified through linkage to the regional archive of hospital discharges, and by direct telephone contact where necessary. Cancers were primary incident cases, identified by the following International Classification of Diseases (10th Revision) codes: C18.0–18.5 (proximal colon); C18.6–C18.7 (distal colon); C18.8 (overlapping colon sites); C18.9 (unspecified colon); C19 (rectosigmoid junction); and C20 (rectum). The corresponding ICD-9 codes are: 153.0, 153.1, 153.4, 153.5, 153.6 and 153.7 for proximal colon; 153.2 and 153.3 for distal colon; 153.9 for unspecified colon; 154.0 for rectosigmoid junction; and 154.1 for rectal cancer. Anal cancers were excluded. End of follow-up varied with center: December 31, 2006 for Varese and Naples; December 31, 2008 for Turin and Ragusa. A total of 286 colorectal cancer cases was identified. The sample on which we performed most analyses therefore consisted of 1134 participants: 850 in the randomly selected cohort and 286 cases (two in the subcohort). The sample for the analysis of LDL cholesterol consisted of only 1129 participants (848 subcohort members, 283 cases, including two in the subcohort) since 5 participants had triglycerides above 400 mg/dl and for these LDL cholesterol could not be estimated. 2.3. Analysis of plasma samples Total cholesterol, HDL cholesterol and triglycerides were measured in plasma samples using a colorimetric enzyme kit (Instrumentation Laboratory, Milan, Italy) and an automatic analyzer (IL 350, Instrumentation Laboratory). LDL cholesterol was calculated using the Friedewald formula [28]. Quality control was assured using commercial (high and low) laboratory standards and an in-house plasma pool. Coefficients of variation (CV) for the high level external standards were 5.5% for cholesterol, 5.0% for triglycerides, and 6.1% for HDL cholesterol. CVs for the low level external standards were 5.8% for cholesterol, 7.9% for triglycerides, and 7.0% for HDL cholesterol. CVs for the in-house plasma pool were 2.6% for cholesterol, 3.5% for triglycerides and 5.3% for HDL cholesterol. To render the results in plasma samples comparable with literature data on serum samples, the following conversion factors were applied: 1.338 for triglycerides, 1.335 for total cholesterol, 1.344 for HDL cholesterol. These factors were determined in the laboratory by comparison of analyte concentrations in plasma and serum samples from 222 persons. For all analyses, laboratory staff were blind to the case–control status of the samples. 2.4. Statistical analysis Baseline characteristics of the subcohort members, according to tertiles of total cholesterol, were summarized as means and standard deviations (continuous variables) or as frequencies (categorical variables). Plasma triglycerides, total cholesterol, HDL cholesterol, and LDL cholesterol were grouped into tertiles (based on distributions in the subcohort). Prentice-weighted Cox proportional hazards modeling was used to produce HRs with 95% CI with the lowest tertile as reference, as estimates of the association between plasma lipid concentrations and colorectal cancer risk, with age as the underlying time scale [29]. Using this approach, a case generated outside the subcohort contributes only when it becomes a case; weightings are assigned as follows: 0 for

C. Agnoli et al. / Cancer Epidemiology 38 (2014) 144–151

146

cases generated outside the subcohort before failure; 1 for cases generated outside the subcohort at failure, cases in the subcohort before and at failure, and non-cases in the subcohort. We ran minimally adjusted models, with age, and sex as covariates; and fully adjusted models, with the following additional covariates: body mass index (BMI) (continuous variable), smoking status (never, former, current), total physical activity (inactive, moderately inactive, moderately active, and active; continuous), education (8 years, >8 years), alcohol consumption (0.1 g/d, >0.1 to 12 g/d, >12 to 24 g/d, >24 g/d), red meat consumption (beef, pork, processed meat; continuous), dietary fiber consumption (continuous), and dietary calcium (continuous). All models were stratified by center. The significance of linear trends across tertiles was tested by treating each tertile as a continuous variable in the Cox models. HRs were also calculated for 1 standard deviation increments of plasma lipids as continuous variables. For triglycerides, the HR for a 1 standard deviation increment was calculated on the log transformed variable because of its skewed distribution. We ran models for the whole cohort and the following subcategories: men, women (fully-adjusted model further adjusted for menopausal status), all colon, proximal colon, distal colon, and rectal cancer. We also ran separate models for premenopausal and postmenopausal women. Ps for interaction of plasma lipids with sex and menopausal status were calculated adding to the models a product term of plasma lipids and, respectively, sex and menopausal status. All analyses were performed with Stata version 11.2 (College Station, TX, USA). 3. Results Table 1 shows baseline characteristics of the cohort by tertiles of total cholesterol. Age increased with increasing total cholesterol tertiles. Cohort members in the highest tertile consumed more alcohol, smoked less, and were less educated. Postmenopausal women had higher total cholesterol than premenopausal women. Cohort members from Turin had highest mean total cholesterol. Table 2 shows HRs for developing colorectal cancer according to tertiles of plasma lipids in all study subjects, and in men and women separately. When all study subjects were considered together, the highest tertiles of total cholesterol (HR 1.66, 95%CI 1.12–2.45) and LDL cholesterol (HR 1.87, 95%CI 1.27–2.76) were associated with significantly greater colorectal cancer risk than the lowest respective tertiles. Risks also increased significantly with 1 standard deviation increments of total and LDL cholesterol as continuous variables. Plasma triglycerides and HDL cholesterol were not significantly associated with colorectal cancer risk. For

Table 1 Baseline characteristics of subcohort members by tertiles of total cholesterol. Tertile I (n = 286)

Characteristic

Tertile II (n = 282)

Tertile III (n = 282)

50.3  7.6 25.6  3.7 81.3  45.4 24.1  11.3 1013  402

52.0  7.5 26.7  3.8 79.5  47.1 24.5  11.4 1030  426

Mean  SD Age, years Body mass index, kg/m2 Red meat intake, g/d Dietary fiber, g/d Dietary calcium, mg/d

47.8  8.1 26.2  4.5 78.6  46.4 26.6  11.3 1057  470 N (%)

Total physical activity Inactive Moderately inactive Moderately active Active Center Turin Varese Naples Ragusa

98 113 35 40

(34.2) (39.5) (12.2) (14.0)

82 95 60 45

(29.0) (33.7) (21.3) (16.0)

80 128 37 37

(28.4) (45.4) (13.1) (13.1)

68 59 87 72

(23.8) (20.6) (30.4) (25.2)

95 52 105 30

(33.7) (18.4) (37.2) (10.7)

143 41 69 29

(50.7) (14.5) (24.5) (10.3)

Gender Male Female

92 (32.2) 194 (67.8)

111 (39.4) 171 (60.6)

81 (28.7) 201 (71.2)

Education 8 yrs >8 yrs

130 (45.6) 156 (54.4)

128 (45.5) 154 (54.5)

161 (57.1) 121 (42.9)

Smoking Current smoker Ex smoker Never smoker

81 (28.3) 77 (26.9) 128 (44.8)

72 (25.5) 79 (28.0) 131 (46.5)

71 (25.2) 69 (24.5) 142 (50.3)

Alcohol consumption 0.1 g/d >0.1–12 g/d >12–24 g/d >24 g/d

52 138 44 52

50 141 43 48

39 139 50 54

(18.2) (48.2) (15.4) (18.2)

Menopausal status (women only) Postmenopausal 58 (29.9) Premenopausal 118 (60.8) Perimenopausal 18 (9.3)

(17.7) (50.0) (15.3) (17.0)

76 (44.4) 82 (48.0) 13 (7.6)

(13.8) (49.3) (17.7) (19.2)

122 (60.7) 62 (30.8) 17 (8.5)

men, the increased risks of colorectal cancer associated with the highest tertiles of total and LDL cholesterol were greater than those for all study subjects (HR 2.54, 95%CI 1.35–4.79 and HR 2.90, 95%CI 1.51–5.56, respectively). For women, by contrast, high total and LDL cholesterol were significantly associated with increased colorectal cancer risk only for 1 standard deviation increments

Table 2 HRs for developing colorectal cancer according to tertiles of plasma lipid concentrations. Tertile I

Tertile II

Tertile III

P trend

Continuous (for each 1 SD increase)

All Triglycerides Range, mg/dl Cases/controls HRa HRb

32.04–89.44 63/291 1 1

90.78–137.50 103/276 1.45 (1.00–2.11) 1.42 (0.97–2.07)

138.84–1124.07 120/283 1.39 (0.96–2.02) 1.32 (0.89–1.95)

0.110 0.226

1.11 (0.96–1.28)d 1.08 (0.92–1.27)d

Total cholesterol Range, mg/dl Cases/controls HRa HRb

72.25–204.71 66/286 1 1

206.05–244.85 91/282 1.10 (0.74–1.73) 1.11 (0.74–1.67)

246.19–417.46 129/282 1.67 (1.13–2.46) 1.66 (1.12–2.45)

0.007 0.009

1.33 (1.13–1.56) 1.32 (1.12–1.56)

HDL cholesterol Range, mg/dl Cases/controls HRa

24.19–55.10 126/324 1

56.45–67.20 83/261 0.88 (0.62–1.26)

68.54–129.02 77/265 0.81 (0.5–1.18)

0.275

0.90 (0.77–1.06)

C. Agnoli et al. / Cancer Epidemiology 38 (2014) 144–151

147

Table 2 (Continued ) Tertile I

Tertile II

Tertile III

P trend

Continuous (for each 1 SD increase)

HR

1

0.89 (0.62–1.28)

0.85 (0.56–1.28)

0.416

0.92 (0.77–1.09)

LDL cholesterolc Range, mg/dl Cases/controls HRa HRb

24.10–121.53 59/283 1 1

121.58–155.15 87/283 1.17 (0.78–1.76) 1.12 (0.74–1.70)

155.17–290.27 137/282 1.91 (1.29–2.81) 1.87 (1.27–2.76)

0.001 0.001

1.39 (1.19–1.63) 1.38 (1.18–1.63)

Men Triglycerides Range, mg/dl Cases/controls HRa HRb

32.04–89.44 17/62 1 1

90.78–137.50 42/96 1.57 (0.78–3.16) 1.64 (0.79–3.40)

138.84–1124.07 71/126 1.97 (1.02–3.79) 1.77 (0.88–3.55)

0.045 0.161

1.25 (1.00–1.58)d 1.16 (0.91–1.49)d

Total cholesterol Range, mg/dl Cases/controls HRa HRb

72.25–204.71 32/92 1 1

206.05–244.85 39/111 0.96 (0.53–1.76) 1.04 (0.55–1.98)

246.19–417.46 59/81 2.54 (1.40–4.59) 2.54 (1.35–4.79)

0.002 0.004

1.64 (1.22–2.20) 1.61 (1.18–2.20)

HDL cholesterol Range, mg/dl Cases/controls HRa HRb

24.19–55.10 77/164 1 1

56.45–67.20 40/74 1.08 (0.64–1.81) 1.21 (0.70–2.10)

68.54–129.02 13/46 0.59 (0.28–1.25) 0.69 (0.30–1.59)

0.296 0.664

0.83 (0.62–1.11) 0.87 (0.63–1.22)

LDL cholesterolc Range, mg/dl Cases/controls HRa HRb

24.10–121.53 25/88 1 1

121.58–155.15 39/103 1.29 (0.68–2.46) 1.24 (0.62–2.47)

155.17–290.27 65/92 2.70 (1.45–5.02) 2.90 (1.51–5.56)

0.001 0.001

1.68 (1.25–2.26) 1.72 (1.25–3.37)

Women Triglycerides Range, mg/dl Cases/controls HRa HRb HRe

32.04–89.44 46/229 1 1 1

90.78–137.50 61/180 1.55 (1.00–2.40) 1.60 (1.02–2.51) 1.63 (1.04–2.57)

138.84–1124.07 49/157 1.13 (0.70–1.82) 1.11 (0.66–1.86) 1.12 (0.66–1.89)

0.587 0.623 0.602

1.04 (0.86–1.26)d 1.02 (0.82–1.26)d 1.02 (0.82–1.27)d

Total cholesterol Range, mg/dl Cases/controls HRa HRb HRe

72.25–204.71 34/194 1 1 1

206.05–244.85 52/171 1.34 (0.80–2.26) 1.35 (0.79–2.33) 1.38 (0.80–2.38)

246.19–417.46 70/201 1.33 (0.80–2.23) 1.39 (0.83–2.33) 1.41 (0.83–2.39)

0.321 0.242 0.238

1.26 (1.04–1.53) 1.30 (1.06–1.60) 1.32 (1.07–1.63)

HDL cholesterol Range, mg/dl Cases/controls HRa HRb HRe

24.19–55.10 49/160 1 1 1

56.45–67.20 43/187 0.77 (0.48–1.24) 0.72 (0.44–1.18) 0.72 (0.44–1.20)

68.54–129.02 64/219 0.84 (0.53–1.34) 0.87 (0.53–1.44) 0.87 (0.52–1.47)

0.534 0.660 0.692

0.94 (0.77–1.13) 0.94 (0.77–1.16) 0.95 (0.77–1.17)

LDL cholesterol Range, mg/dl Cases/controls HRa HRb HRe

24.10–121.53 34/195 1 1 1

121.58–155.15 48/180 1.20 (0.71–2.03) 1.16 (0.67–1.99) 1.18 (0.68–2.04)

155.17–290.27 727,190 1.58 (0.95–2.61) 1.62 (0.97–2.70) 1.65 (0.97–2.78)

0.066 0.053 0.051

1.32 (1.10–1.60) 1.38 (1.13–1.70) 1.40 (1.14–1.72)

b

a b c d e

Adjusted for age and gender; stratified by center. Adjusted for age, gender, BMI, smoking, total physical activity, alcohol consumption, dietary red meat, dietary fiber, and dietary calcium; stratified by center. Analysis based on 1129 subjects because 5 had missing LDL-cholesterol. HR calculated on log-transformed variable. Further adjusted for menopausal status (1 woman with missing datum excluded).

of the continuous variables. No interaction was found between any plasma lipid and gender. The analyses for postmenopausal and premenopausal women (Table 3) indicated increased colorectal cancer risk for postmenopausal women with high total and LDL cholesterol, but the risk increases were significant only for 1 standard deviation increments of total and LDL cholesterol as continuous variables (HR 1.44, 95%CI: 1.11–1.88 and HR 1.60, 95%CI: 1.23–2.09, respectively). Plasma triglycerides and HDL cholesterol were not significantly associated with colorectal cancer risk in premenopausal women.

Nevertheless, no interaction was found between any plasma lipid and menopausal status. The results (all study subjects) for each colorectal sub-site are shown in Table 4. For all colon, distal colon, and rectal sites, the results were similar to those found for colorectal cancer in general, with significantly increased cancer risks in those with the highest tertiles of total and LDL cholesterol; however for rectal cancer, risks were significantly higher only for 1 standard deviation increments of total and LDL cholesterol. The risk of developing proximal colon cancer was unrelated to levels of any plasma lipid investigated.

C. Agnoli et al. / Cancer Epidemiology 38 (2014) 144–151

148

Table 3 HRs for developing colorectal cancer in women according to tertiles of plasma lipid concentrations, stratified by menopausal status. Tertiles I

P trend (median)

Continuous (for each 1 SD increase)

II

III

Postmenopausal women Triglycerides Range, mg/dl 32.04–82.77 24/67 Cases/controls HRa 1

84.10–126.82 41/86 1.23 (0.66–2.28)

128.16–1124.07 43/103 0.93 (0.49–1.76)

0.742

1.02 (0.77–1.33)b

Total cholesterol Range, mg/dl Cases/controls HRa

90.98–203.38 21/58 1

204.71–248.87 30/84 0.93 (0.44–1.97)

250.21–417.46 57/114 1.46 (0.73–2.90)

0.194

1.44 (1.11–1.88)

HDL cholesterol Range, mg/dl Cases/controls HRa

24.19–57.79 41/84 1

59.14–69.89 31/91 0.66 (0.36–1.19)

71.23–122.30 36/81 0.75 (0.40–1.40)

0.362

0.88 (0.66–1.18)

LDL cholesterol Range, mg/dl Cases/controls HRa

42.82–120.18 18/54 1

120.19–155.79 28/86 0.93 (0.43–2.01)

156.08–290.27 60/115 1.76 (0.86–3.60)

0.063

1.60 (1.23–2.09)

Premenopausal women Triglycerides Range, mg/dl 32.04–82.77 Cases/controls 15/103 HRa 1

84.10–126.82 17/87 1.50 (0.64–3.55)

128.16–1124.07 11/72 1.09 (0.42–2.80)

0.761

1.10 (0.74–1.64)b

Total cholesterol Range, mg/dl Cases/controls HRa

90.98–203.38 10/117 1

204.71–248.87 25/89 2.53 (1.09–5.88)

250.21–417.46 8/56 1.15 (0.35–3.81)

0.610

1.09 (0.71–1.68)

HDL cholesterol Range, mg/dl Cases/controls HRa

24.19–57.79 12/102 1

59.14–69.89 14/69 1.65 (0.70–3.88)

71.23–122.30 17/91 1.35 (0.57–3.23)

0.500

0.95 (0.70–1.28)

LDL cholesterol Range, mg/dl Cases/controls HRa

42.82–120.18 12/116 1

120.19–155.79 20/88 1.66 (0.72–3.87)

156.08–290.27 11/58 1.50 (0.51–4.41)

0.387

1.11 (0.74–1.68)

a b

Adjusted for age, gender, BMI, smoking, total physical activity, alcohol consumption, dietary red meat, dietary fiber, and dietary calcium; stratified by center. HR calculated on log-transformed variable.

Table 4 HRs for developing all colon, proximal colon, distal colon and rectal cancer according to tertiles of plasma lipid concentrations. Tertiles

P trend (median)

Continuous (for each 1 SD increase)

138.84–1124.07 88/283 1.31 (0.87–1.98) 1.23 (0.79–1.92)

0.251 0.447

1.10 (0.94–1.29)d 1.06 (0.89–1.27)d

206.05–244.85 71/282 1.17 (0.76–1.81) 1.18 (0.75–1.85)

246.19–417.46 96/282 1.67 (1.08–2.56) 1.66 (1.08–2.56)

0.017 0.019

1.32 (1.10–1.59) 1.33 (1.10–1.61)

24.19–55.10 92/324 1 1

56.45–67.2 64/261 0.92 (0.62–1.35) 0.92 (0.61–1.37)

68.54–129.02 58/265 0.79 (0.52–1.19) 0.84 (0.53–1.32)

0.262 0.444

0.89 (0.74–1.05) 0.91 (0.75–1.10)

24.10–121.53 46/283 1 1

121.58–155.15 60/283 1.02 (0.65–1.60) 0.98 (0.62–1.57)

155.17–290.27 106/282 1.86 (1.22–2.83) 1.85 (1.21–2.82)

0.002 0.002

1.39 (1.16–1.66) 1.40 (1.16–1.68)

I

II

III

All colon cancers Triglycerides Range, mg/dl Cases/controls HRa HRb

32.4–89.44 48/291 1 1

90.78–137.50 78/276 1.42 (0.94–2.14) 1.41 (0.92–2.15)

Total cholesterol Range, mg/dl Cases/controls HRa HRb

72.25–204.71 47/286 1 1

HDL cholesterol Range, mg/dl Cases/controls HRa HRb LDL cholesterolc Range, mg/dl Cases/controls HRa HRb Proximal colon cancer Triglycerides

C. Agnoli et al. / Cancer Epidemiology 38 (2014) 144–151

149

Table 4 (Continued ) Tertiles

P trend (median)

Continuous (for each 1 SD increase)

I

II

III

Range, mg/dl Cases/controls HRa HRb

32.04–89.44 21/291 1 1

90.78–137.50 31/276 1.29 (0.72–2.33) 1.26 (0.68–2.33)

138.84–1124.07 34/283 1.15 (0.64–2.10) 0.99 (0.51–1.92)

0.696 0.908

1.08 (0.85–1.38)d 1.01 (0.76–1.35)d

Total cholesterol Range, mg/dl Cases/controls HRa HRb

72.25–204.71 20/286 1 1

206.05–244.85 29/282 1.07 (0.58–2.00) 1.00 (0.51–1.95)

246.19– 417.46 37/282 1.36 (0.71–2.61) 1.38 (0.71–2.68)

0.335 0.315

1.16 (0.89–1.51) 1.17 (0.87–1.56)

HDL cholesterol Range, mg/dl Cases/controls HRa HRb

24.19–55.10 37/324 1 1

56.45–67.20 27/261 0.88 (0.51–1.52) 0.84 (0.47–1.51)

68.54–129.02 22/265 0.62 (0.34–1.14) 0.64 (0.32–1.31)

0.125 0.226

0.75 (0.58–0.97) 0.76 (0.56–1.03)

LDL cholesterolc Range, mg/dl Cases/controls HRa HRb

36.84–121.53 17/283 1 1

121.58–155.15 27/283 1.19 (0.62–2.28) 1.07 (0.53–2.16)

155.17–290.27 41/282 1.81 (0.95–3.46) 1.86 (0.96–3.59)

0.060 0.055

1.23 (0.96–1.58) 1.24 (0.95–1.63)

Distal colon cancer Triglycerides Range, mg/dl Cases/controls HRa HRb

32.4–89.44 22/291 1 1

90.78–137.50 40/276 1.59 (0.91–2.78) 1.66 (0.94–2.93)

138.84–1124.07 42/283 1.36 (0.78–2.36) 1.34 (0.73–2.44)

0.349 0.434

1.08 (0.88–1.31)d 1.06 (0.85–1.31)d

Total cholesterol Range, mg/dl Cases/controls HRa HRb

72.25–204.71 20/286 1 1

206.05–244.85 33/282 1.34 (0.74–2.43) 1.42 (0.76–2.65)

246.19– 417.46 51/282 2.15 (1.24–3.72) 2.20 (1.26–3.86)

0.006 0.005

1.50 (1.20–1.88) 1.55 (1.22–1.97)

HDL cholesterol Range, mg/dl Cases/controls HRa HRb

24.19–55.10 42/324 1 1

56.45–67.20 30/261 1.01 (0.60–1.69) 1.04 (0.60–1.80)

68.54–129.02 32/265 1.09 (0.65–1.84) 1.20 (0.68–2.13)

0.750 0.540

1.05 (0.85–1.30) 1.09 (0.87–1.37)

LDL cholesterolc Range, mg/dl Cases/controls HRa HRb

36.84–121.53 22/283 1 1

121.58–155.15 25/283 0.92 (0.49–1.72) 0.91 (0.48–1.74)

155.17–290.27 56/282 2.12 (1.24–3.61) 2.11 (1.23–3.59)

0.003 0.003

1.57 (1.26–1.95) 1.63 (1.29–2.06)

Rectal cancer Triglycerides Range, mg/dl Cases/controls HRa HRb

32.4–89.44 15/291 1 1

90.78–137.50 25/276 1.55 (0.79–3.06) 1.44 (0.74–2.83)

138.84–1124.07 32/283 1.66 (0.87–3.16) 1.53 (0.79–2.95)

0.131 0.225

1.12 (0.89–1.41)d 1.11 (0.86–1.44)d

Total cholesterol Range, mg/dl Cases/controls HRa HRb

72.25–204.71 19/286 1 1

206.05–244.85 20/282 0.93 (0.47–1.83) 0.94 (0.46–1.92)

246.19– 417.46 33/282 1.71 (0.87–3.36) 1.70 (0.84–3.45)

0.111 0.136

1.36 (1.06–1.73) 1.33 (1.03–1.70)

HDL cholesterol Range, mg/dl Cases/controls HRa HRb

24.19–55.10 34/324 1 1

56.45–67.20 19/261 0.79 (0.44–1.43) 0.84 (0.45–1.55)

68.54–129.02 19/265 0.91 (0.49–1.71) 0.96 (0.46–1.97)

0.706 0.847

0.97 (0.73–1.28) 0.98 (0.71–1.36)

LDL cholesterolc Range, mg/dl Cases/controls HRa HRb

36.84–121.53 13/283 1 1

121.58–155.15 27/283 1.70 (0.83–3.49) 1.61 (0.77–3.39)

155.17–290.27 31/282 2.05 (0.98–4.27) 1.92 (0.91–4.09)

0.054 0.089

1.39 (1.08–1.80) 1.35 (1.04–1.75)

a b c d

Adjusted for age and gender; stratified by center. Adjusted for age, gender, BMI, smoking, total physical activity, alcohol consumption, dietary red meat, dietary fiber, and dietary calcium; stratified by center. Analysis based on 1129 subjects because 5 subjects had missing LDL cholesterol. HR calculated on log-transformed variable.

150

C. Agnoli et al. / Cancer Epidemiology 38 (2014) 144–151

4. Discussion In this case–cohort study, we have found that high plasma levels of total and LDL cholesterol were associated with increased risk of colorectal cancer. The risk increases were greater for men than women; and among women, risk increases were mainly evident for those in postmenopause. As regards individual subsites, the risk increases were significant for whole colon, distal colon and, less evidently, rectum, but not for proximal colon. Plasma triglycerides and HDL cholesterol were unrelated to risk of colorectal cancer. Numerous studies have investigated the relation between blood lipids and colorectal cancer but have not provided a clear picture. Early epidemiological studies found an inverse relationship between cholesterol and colon cancer risk [10,12,13] and the 1988 National Health and Nutrition Examination Survey I Epidemiologic Follow-up Study found an inverse relationship between total cholesterol and colorectal cancer [30]. Two recent prospective studies found high blood triglycerides were associated with increased colorectal [16] or colon [2] cancer risk, and a 1986 prospective study found increased colorectal cancer risk for high blood total cholesterol among men [31]. However other fairly recent prospective studies found no association of blood triglycerides with colorectal cancer risk [23,32,33]. A case–control study nested in the EPIC-Europe cohort [24] found an inverse relationship between HDL cholesterol and colon cancer, but no association with rectal cancer. Our findings on total cholesterol are in line with those of the recently published Me-Can cohort paper [34], in which a positive though not statistically significant association between total serum cholesterol and colon and rectal cancer was found, in both men and women. However, the effects of cholesterol were much stronger in our study than that of Strohmaier et al. [34] A possible explanation for the larger effect size found by us is that there were competing risks for cardiovascular disease in the MeCan study that were not present in our case–cohort study. A ‘‘small-study effect’’ [35] is also possible-a stronger risk statistically related to small study size. A possible explanation for the finding that blood cholesterol was inversely related to colorectal cancer, could be reverse causation: The 1991 review on prospective studies conducted by Law & Thompson found, in fact, that the inverse association was present only in cancers diagnosed shortly after the cholesterol measurement but was not present in cases that developed long after the measurement [14], so metachronous cholesterol levels were unlikely to have influenced the development of the disease, or the cancer itself could have modified cholesterol levels. The inverse association might also be due to differences between the etiologies of distal and proximal colon cancers [36]. Thus, while distal colon cancer seems related to environmental factors, proximal colon cancer has been suggested as related to genetically determined low blood cholesterol [11]. However this explanation would be reasonable only if, in the studies with an inverse cholesterol–colorectal cancer association, proximal cancers formed the majority of cases; unfortunately these studies did not differentiate between the distal and proximal sites. It is noteworthy that cholesterol was unrelated to risk of proximal colon cancer in the present study. It is also noteworthy that our findings differ from those of the EPIC-Europe nested case–control study [24], which included cohorts from 23 centers in 10 European countries, and included all five EPIC Italy centers. EPIC-Europe found that LDL cholesterol had no influence on risk and that high HDL cholesterol was protective. When the EPIC-Europe data pertaining to the four centers included in the present study were re-analyzed, no significant association between total and LDL cholesterol and colorectal cancer was found (van Duijnhoven, personal communication), but risks were in the

same direction. Although the cohorts included in both studies were the same, there were fewer cases and follow-up was shorter in the EPIC-Europe study because it was conducted earlier. The shorter follow-up and hence fewer cases are likely to explain the lack of significant association between cancer risk and blood lipids in the EPIC-Europe study, as also suggested by the wide 95%CIs. Various mechanisms may explain the association between cholesterol and colorectal cancer risk. Several epidemiological studies suggest involvement of bile acids. Feces from people in Western countries-where colorectal cancer incidence is highcontain more steroids and have higher proportions of anaerobes that deconjugate primary bile acids, than feces from people where colorectal cancer incidence in lower [37]; furthermore secondary fecal bile acid levels are higher in colorectal cancer patients than healthy volunteers [38]. In addition, cholecystectomy – a condition characterized by higher fecal loss of bile acids and an increased ratio of secondary to primary bile acids – is a risk factor for proximal colorectal cancer [39]. Secondary bile acids may promote tumor growth by stimulating proliferation and invasion of colorectal cancer cells via activation of protein kinase C and COX-2, as reviewed in Debruyne et al. [40]. Cholesterol itself could also be involved in colorectal carcinogenesis via its effect on inflammation which may promote cellular proliferation and inhibit apoptosis [41,42]. Interestingly, single nucleotide polymorphisms in CYP7A1, the rate-limiting enzyme in the pathway from cholesterol to primary bile acids, have been linked to colorectal cancer risk [43]. Hypercholesterolemia is also related to oxidative stress [44] which in turn may play a role in cancer development [45], perhaps by altering gene expression. High circulating LDL cholesterol may occur in metabolic syndrome and insulin resistance [21], both of which have been suggested as risk factors for colorectal cancer [46,47]. A study limitation is that we assessed lipids in a single plasma sample, and cannot provide indications of long-term variation in lipid levels since baseline. However any such changes are likely to weaken the association between lipids and cancer. There is also the problem of possible short-term intra-individual variation in analyte levels. However data suggest that, with the exception of triglycerides, variation of plasma lipids is contained [48]. Another limitation is that samples were collected between 1993 and 1998, stored at 196 8C, and assayed up to 17 years later and it is possible that lipids decay during long-term storage. However data indicate that concentrations of triglycerides and cholesterol in serum samples stored at 70 8C are relatively stable for at least five years [49]. Furthermore if decay half-life is independent of initial concentration, then the rank of concentrations in samples will not change to bias the risk evaluation. There are also limitations arising from the case–cohort design. We chose this design because we needed to investigate different outcomes, so a nested case–control design was unsuitable. In comparison with the nested case–control design, the case–cohort design is less able to account for batch differences and long-term storage because there is no matching. Strengths of our study are its prospective design, relatively large sample size, and use of detailed information on lifestyle, dietary, and anthropometric variables, allowing control for confounding effects. To conclude, the findings of this prospective study suggest that elevated plasma total and LDL cholesterol – but not triglycerides or HDL cholesterol – may be risk factors for colorectal cancer, especially all colon and distal colon cancer. Risks were stronger for men and, among women, for those who were postmenopausal. It is possible that cholesterol is a risk factor overlying unknown but probably complex biological mechanisms and further studies are required to elucidate how they may give rise to colorectal cancer.

C. Agnoli et al. / Cancer Epidemiology 38 (2014) 144–151

Conflict of interest statement All authors declare no conflict of interest. Acknowledgements We thank all participants the EPIC Italy study, A. Evangelista and D. Del Sette for technical support, Don Ward for help with the English, and the Italian Association for Cancer Research (AIRC) and Compagnia di San Paolo for financial support.

[23]

[24]

[25]

[26]

References [27] [1] Expert Panel on 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). J Am Med Assoc 2001;285(19):2486–97. [2] Borena W, Stocks T, Jonsson H, Strohmaier S, Nagel G, Bjorge T, et al. Serum triglycerides and cancer risk in the metabolic syndrome and cancer (Me-Can) collaborative study. Cancer Causes Control 2011;22(2):291–9. [3] Cust AE, Kaaks R, Friedenreich C, Bonnet F, Laville M, Tjonneland A, et al. Metabolic syndrome, plasma lipid, lipoprotein and glucose levels, and endometrial cancer risk in the European Prospective Investigation into Cancer and Nutrition (EPIC). Endocr Relat Cancer 2007;14(3):755–67. [4] Swanson CA, Potischman N, Barrett RJ, Berman ML, Mortel R, Twiggs LB, et al. Endometrial cancer risk in relation to serum lipids and lipoprotein levels. Cancer Epidemiol Biomarkers Prev 1994;3(7):575–81. [5] Tulinius H, Sigfusson N, Sigvaldason H, Bjarnadottir K, Tryggvadottir L. Risk factors for malignant diseases: a cohort study on a population of 22,946 Icelanders. Cancer Epidemiol Biomarkers Prev 1997;6(11):863–73. [6] Ulmer H, Borena W, Rapp K, Klenk J, Strasak A, Diem G, et al. Serum triglyceride concentrations and cancer risk in a large cohort study in Austria. Br J Cancer 2009;101(7):1202–6. [7] Hoyer AP, Engholm G. Serum lipids and breast cancer risk: a cohort study of 5,207 Danish women. Cancer Causes Control 1992;3(5):403–8. [8] Melvin JC, Seth D, Holmberg L, Garmo H, Hammar N, Jungner I, et al. Lipid profiles and risk of breast and ovarian cancer in the Swedish AMORIS study. Cancer Epidemiol Biomarkers Prev 2012;21(8):1381–4. [9] Kok DE, van Roermund JG, Aben KK, den Heijer M, Swinkels DW, Kampman E, et al. Blood lipid levels and prostate cancer risk; a cohort study. Prostate Cancer Prostatic Dis 2011;14(4):340–5. [10] Rose G, Blackburn H, Keys A, Taylor HL, Kannel WB, Paul O, et al. Colon cancer and blood-cholesterol. Lancet 1974;1(7850):181–3. [11] Sidney S, Farquhar JW. Cholesterol, cancer, and public health policy. Am J Med 1983;75(3):494–508. [12] Williams RR, Sorlie PD, Feinleib M, McNamara PM, Kannel WB, Dawber TR. Cancer incidence by levels of cholesterol. J Am Med Assoc 1981;245(3):247–52. [13] Kagan A, McGee DL, Yano K, Rhoads GG, Nomura A. Serum cholesterol and mortality in a Japanese-American population: the Honolulu Heart program. Am J Epidemiol 1981;114(1):11–20. [14] Law MR, Thompson SG. Low serum cholesterol and the risk of cancer: an analysis of the published prospective studies. Cancer Causes Control 1991;2(4):253–61. [15] Aleksandrova K, Boeing H, Jenab M, Bas Bueno-de-Mesquita H, Jansen E, van Duijnhoven FJ, et al. Metabolic syndrome and risks of colon and rectal cancer: the European prospective investigation into cancer and nutrition study. Cancer Prev Res (Phila) 2011;4(11):1873–83. [16] Stocks T, Lukanova A, Bjorge T, Ulmer H, Manjer J, Almquist M, et al. Metabolic factors and the risk of colorectal cancer in 580,000 men and women in the metabolic syndrome and cancer project (Me-Can). Cancer 2011;117(11):2398–407. [17] Alberti KG, Zimmet P, Shaw J. Metabolic syndrome – a new world-wide definition. A Consensus Statement from the International Diabetes Federation. Diabet Med 2006;23(5):469–80. [18] Pais R, Silaghi H, Silaghi AC, Rusu ML, Dumitrascu DL. Metabolic syndrome and risk of subsequent colorectal cancer. World J Gastroenterol 2009;15(41):5141–8. [19] Esteve E, Ricart W, Fernandez-Real JM. Dyslipidemia and inflammation: an evolutionary conserved mechanism. Clin Nutr 2005;24(1):16–31. [20] Hutcheson R, Rocic P. The metabolic syndrome, oxidative stress, environment, and cardiovascular disease: the great exploration. Exp Diabetes Res 2012;2012:271028. [21] Avramoglu RK, Basciano H, Adeli K. Lipid and lipoprotein dysregulation in insulin resistant states. Clin Chim Acta 2006;368(1–2):1–19. [22] Chung YW, Han DS, Park YK, Son BK, Paik CH, Lee HL, et al. Association of obesity, serum glucose and lipids with the risk of advanced colorectal

[28]

[29] [30]

[31]

[32]

[33]

[34]

[35]

[36]

[37] [38]

[39]

[40] [41] [42]

[43]

[44]

[45]

[46]

[47] [48] [49]

151

adenoma and cancer: a case–control study in Korea. Dig Liver Dis 2006;38(9):668–72. Tsushima M, Nomura AM, Lee J, Stemmermann GN. Prospective study of the association of serum triglyceride and glucose with colorectal cancer. Dig Dis Sci 2005;50(3):499–505. van Duijnhoven FJ, Bueno-de-Mesquita HB, Calligaro M, Jenab M, Pischon T, Jansen EH, et al. Blood lipid and lipoprotein concentrations and colorectal cancer risk in the European Prospective Investigation into Cancer and Nutrition. Gut 2011;60(8):1094–102. Palli D, Berrino F, Vineis P, Tumino R, Panico S, Masala G, et al. A molecular epidemiology project on diet and cancer: the EPIC-Italy Prospective Study. Design and baseline characteristics of participants. Tumori 2003;89(6):586– 93. Pala V, Sieri S, Palli D, Salvini S, Berrino F, Bellegotti M, et al. Diet in the Italian EPIC cohorts: presentation of data and methodological issues. Tumori 2003;89(6):594–607. Rose G. Standardisation of observers in blood pressure measurement. Lancet 1965;1(7387):673–4. 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(6):499–502. Prentice RL. A case–cohort design for epidemiologic cohort studies and disease prevention trials. Biometrika 1986;73(1):1–11. Schatzkin A, Hoover RN, Taylor PR, Ziegler RG, Carter CL, Albanes D, et al. Sitespecific analysis of total serum cholesterol and incident cancer in the National Health and Nutrition Examination Survey I Epidemiologic Follow-up Study. Cancer Res 1988;48(2):452–8. Tornberg SA, Holm LE, Carstensen JM, Eklund GA. Risks of cancer of the colon and rectum in relation to serum cholesterol and beta-lipoprotein. N Engl J Med 1986;315(26):1629–33. Ashbeck EL, Jacobs ET, Martinez ME, Gerner EW, Lance P, Thompson PA. Components of metabolic syndrome and metachronous colorectal neoplasia. Cancer Epidemiol Biomarkers Prev 2009;18(4):1134–43. Saydah SH, Platz EA, Rifai N, Pollak MN, Brancati FL, Helzlsouer KJ. Association of markers of insulin and glucose control with subsequent colorectal cancer risk. Cancer Epidemiol Biomarkers Prev 2003;12(5):412–8. Strohmaier S, Edlinger M, Manjer J, Stocks T, Bjorge T, Borena W, et al. Total serum cholesterol and cancer incidence in the Metabolic syndrome and Cancer Project (Me-Can). PLoS One 2013;8(1):e54242. Sterne JA, Gavaghan D, Egger M. Publication and related bias in meta-analysis: power of statistical tests and prevalence in the literature. J Clin Epidemiol 2000;53(11):1119–29. Glebov OK, Rodriguez LM, Nakahara K, Jenkins J, Cliatt J, Humbyrd CJ, et al. Distinguishing right from left colon by the pattern of gene expression. Cancer Epidemiol Biomarkers Prev 2003;12(8):755–62. Hill MJ, Drasar BS, Hawksworth G, Aries V, Crowther JS, Williams RE. Bacteria and aetiology of cancer of large bowel. Lancet 1971;1(7690):95–100. Kishida T, Taguchi F, Feng L, Tatsuguchi A, Sato J, Fujimori S, et al. Analysis of bile acids in colon residual liquid or fecal material in patients with colorectal neoplasia and control subjects. J Gastroenterol 1997;32(3):306–11. McMichael AJ, Potter JD. Host factors in carcinogenesis: certain bile-acid metabolic profiles that selectively increase the risk of proximal colon cancer. J Natl Cancer Inst 1985;75(2):185–91. Debruyne PR, Bruyneel EA, Li X, Zimber A, Gespach C, Mareel MM. The role of bile acids in carcinogenesis. Mutat Res 2001;480(48135):9–369. Erdman SE, Poutahidis T. Roles for inflammation and regulatory T cells in colon cancer. Toxicol Pathol 2010;38(1):76–87. Kim S, Keku TO, Martin C, Galanko J, Woosley JT, Schroeder JC, et al. Circulating levels of inflammatory cytokines and risk of colorectal adenomas. Cancer Res 2008;68(1):323–8. Wertheim BC, Smith JW, Fang C, Alberts DS, Lance P, Thompson PA. Risk modification of colorectal adenoma by CYP7A1 polymorphisms and the role of bile acid metabolism in carcinogenesis. Cancer Prev Res (Phila) 2012;5(2):197–204. da Silva PR, Tatsch E, Bochi GV, Kober H, Duarte T, Dos Santos Montagner GF, et al. Assessment of oxidative, inflammatory, and fibrinolytic biomarkers and DNA strand breakage in hypercholesterolemia. Inflammation 2013;36(4):869–77. Valko M, Izakovic M, Mazur M, Rhodes CJ, Telser J. Role of oxygen radicals in DNA damage and cancer incidence. Mol Cell Biochem 2004;266(1–2):37– 56. Esposito K, Chiodini P, Colao A, Lenzi A, Giugliano D. Metabolic syndrome and risk of cancer: a systematic review and meta-analysis. Diabetes Care 2012;35(11):2402–11. Giovannucci E. Insulin, insulin-like growth factors and colon cancer: a review of the evidence. J Nutr 2001;131(11 Suppl.):3109S–20S. Shumak SL, Campbell NR. Intraindividual variation in lipid and lipoprotein levels. CMAJ 1993;149(6):843–4. Shih WJ, Bachorik PS, Haga JA, Myers GL, Stein EA. Estimating the long-term effects of storage at 70 8C on cholesterol, triglyceride, and HDL-cholesterol measurements in stored sera. Clin Chem 2000;46(3):351–64.