Abstract
Background: Little is known about the causes of thyroid cancer, but insulin-like growth factor-I (IGF-I) might play an important role in its development due to its mitogenic and antiapoptotic properties.
Methods: This study prospectively investigated the association between serum IGF-I concentrations and risk of differentiated thyroid carcinoma in a case–control study nested within the European Prospective Investigation into Cancer and Nutrition. The 345 incident cases of differentiated thyroid carcinoma were individually matched to 735 controls by study center, sex and age, date, time, and fasting status at blood collection, follow-up duration, and for women menopausal status, use of exogenous hormones, and phase of menstrual cycle at blood collection. Serum IGF-I concentrations were measured by immunoassay, and risk of differentiated thyroid cancer in relation to IGF-I concentration was estimated using conditional logistic regression.
Results: There was a positive association between IGF-I concentrations and risk of differentiated thyroid carcinoma: the OR for a doubling in IGF-I concentration was 1.48 (95% confidence interval, 1.06–2.08; Ptrend = 0.02). The positive association with IGF-I was stable over time between blood collection and cancer diagnosis.
Conclusion: These findings suggest that IGF-I concentrations may be positively associated with risk of differentiated thyroid carcinoma.
Impact: This study provides the first prospective evidence of a potential association between circulating IGF-I concentrations and risk of differentiated thyroid carcinoma and may prompt the further investigations needed to confirm the association. Cancer Epidemiol Biomarkers Prev; 23(6); 976–85. ©2014 AACR.
Introduction
Worldwide, the age-standardized incidence rate of thyroid cancer is 4.0 per 100,000 persons per year, with relatively high rates seen in Europe and North America (1). Thyroid cancer is three times more common in women than in men (1–4), and is now the third most common cancer in women under the age of 45 in highly developed countries (1). Papillary and follicular carcinomas, together called differentiated thyroid carcinoma, are the most common types of thyroid cancer, accounting for approximately 75% and 10% of all thyroid cancer, respectively (3, 5). There has been an increase in thyroid cancer incidence since the 1970s (3), with up to a 2-fold increase in some of the European countries (2, 3) and up to a 3-fold increase in the United States (3, 4), thought to be largely but not solely due to increased medical surveillance (4). Little is known about the causes of thyroid cancer with the only established risk factor being exposure to ionizing radiation, especially during childhood (6).
Insulin-like growth factor-I (IGF-I) is an endogenous hormone with mitogenic and antiapoptotic properties (7), and high circulating concentrations have been shown to be associated with an increased risk of developing cancer at several sites, such as breast, prostate, and perhaps colorectal cancer (8–11). In contrast, there are limited epidemiologic data on the relationship between IGF-I concentrations and thyroid cancer. A case–control study did not find an association between the two (12). It has been suggested that high concentrations of IGF-I may be associated with a higher risk of goiter (13), which has been linked to thyroid cancer (14–16). Furthermore, IGF-I is a determinant of height that is associated with several cancers (17), including differentiated thyroid carcinoma (18). Some small studies of patients with acromegaly, a disorder characterized by high concentrations of IGF-I, have found higher rates of differentiated thyroid cancer in this patient group than in the general population (19–21). However, it is not known whether high prediagnostic circulating concentrations of IGF-I are directly associated with thyroid cancer risk.
We aimed to prospectively investigate the association between serum IGF-I concentrations and risk of differentiated thyroid carcinoma in the European Prospective Investigation into Cancer and Nutrition (EPIC).
Materials and Methods
Study population and blood collection
A detailed description of the recruitment, blood collection, and follow-up procedure in EPIC has been published previously (22). In brief, from 1992 to 2000, approximately 520,000 participants, mostly ages 35 to 70 years, were recruited in 23 centers located in 10 European countries (Denmark, France, Germany, Greece, Italy, Netherlands, Norway, Spain, Sweden, and United Kingdom). Participants provided detailed information on dietary habits, lifestyle, anthropometry, and previous disease.
At recruitment, 385,747 of the participants (74%) gave a blood sample. In most centers, the blood collection and processing were undertaken according to a standardized protocol; whole blood was transported to a local laboratory, processed within 24 hours, and stored in liquid nitrogen (−196°C) at the International Agency for Research on Cancer (IARC). Exceptions to the standardized protocol occurred; in Oxford, United Kingdom, samples were sent at ambient temperature to a laboratory in Norfolk, United Kingdom, with a mean transit time of 1.5 days; in Denmark and Sweden, all blood samples were stored locally in nitrogen vapor (−150°C) and in electric freezers (−70°C), respectively; and in Norway, serum was not collected.
Follow-up for cancer incidence and vital status
Follow-up for cancer incidence and vital status was conducted via record linkage with regional and national cancer and mortality registers, except in France, Germany, and Greece. In these countries, information on cancer incidence was collected via active follow-up through participants and their next-of-kin and verified through health insurance records, and cancer and pathology registries or reports. Information on vital status was collected via active follow-up in Germany and Greece.
Selection of cases and controls
Cases were selected from cohort participants who had donated a blood sample, were free of cancer at baseline (except nonmelanoma skin cancer), were diagnosed with incident thyroid cancer (WHO International Classification of Disease 10th revision code C73) before the end of follow-up, and had not been diagnosed with other tumors (except nonmelanoma skin cancer), before the thyroid cancer diagnosis. The end of follow-up was defined as the last date of complete follow-up for both vital status and cancer incidence, which ranged from December 2006 to December 2009. This analysis focused on differentiated thyroid carcinoma, i.e., papillary (morphologic codes: 8050, 8260, 8340–8344, 8350, and 8450–8460) and follicular carcinomas (8290, 8330–8335; ref. 23), as this is the most common type of thyroid cancer. Participants diagnosed with rare histologic types (28 medullary, 6 anaplastic, 3 other morphologies, and 1 lymphoma) were excluded, and the present study includes 345 cases. Information on tumor–node–metastasis (TNM) stage was used to group cancers into localized (T1, N0, and M0) and more advanced (≥T2, ≥N1, or M1) cancers.
Participants were eligible as controls if they had provided a blood sample at baseline and were alive and free of cancer (other than nonmelanoma skin cancer) at time of the diagnosis of the index case. Randomly chosen controls (three controls for men and two controls for women) were matched to each case according to an incidence density sampling scheme (24). The matching criteria from a parallel study of thyroid hormones and differentiated thyroid carcinoma in our cohort (25) were used; these were: study center, sex, duration of follow-up (duration for the control must be greater than for the index case), and age (± 1 year), date (± 3 months), time of day (± 1 hour), and fasting status (<3; 3–6; >6 hours) at blood collection. Women were also matched by menopausal status (premenopausal; perimenopausal or unknown; postmenopausal; refs. 26, 27), use of exogenous hormones at blood collection (yes; no), and phase of menstrual cycle at blood collection (follicular; periovulatory; luteal; ref. 26) in premenopausal women.
This study was approved by the IARC Ethics Committee and local ethics committees in the participating countries. All participants gave written informed consent.
Laboratory measurements
Laboratory analyses were performed on serum samples, except for those from Norway and Umeå, Sweden, which provided citrated plasma and EDTA plasma samples, respectively.
Concentrations of IGF-I were measured in the Cancer Epidemiology Unit laboratory (Oxford, United Kingdom), using the automated IDS-iSYS immunoassay system from Immunodiagnostic Systems (IDS) Ltd. The intrabatch coefficient of variation was 3.7% and the overall coefficient of variation was 4.7% at a mean concentration of 13.4 nmol/L. Two samples included insufficient sample, resulting in measurements available for 345 cases and 735 controls.
Measurements of circulating concentrations of thyroid-stimulating hormone (TSH), thyroglobulin (Tg), anti-Tg antibodies (TgAb), and thyroid hormones, i.e., triiodo-thyronine (T3) and thyroxine (T4), were conducted at IARC for a previous analysis in this study population (25). Total and free T3 and T4 were measured using direct radioimmunoassay, TSH using direct immunoradiometric assays (Beckmann Coulter), and Tg and TgAb using immunoradiometric assays (DiaSorin). Individuals with TgAb > 100 UI/mL were categorized as TgAb-positive.
Statistical analysis
Participant characteristics were compared between cases and controls within matched sets, for men and women separately, using the weighted paired-sample t test and the likelihood ratio χ2 test where appropriate. No P value was calculated for matching variables. For cases only, years between blood collection and diagnosis, age at diagnosis, TNM stage (localized; more advanced), and histologic type (papillary; follicular) were additionally investigated. History of thyroid disease was not systematically available in EPIC.
Concentrations of IGF-I were transformed logarithmically to approximate the normal distribution in all analyses. Geometric means of IGF-I concentrations by participant characteristics adjusted for age at blood collection (as a continuous variable), body mass index (BMI; in fifths based on the distribution in controls), and sex were investigated using ANOVA, in controls only. Analyses further adjusted for country were also conducted. Tests for linear trend across categories were performed by scoring categories with consecutive whole numbers.
Correlations between concentrations of IGF-I and TSH, Tg, and thyroid hormones were examined in controls by use of partial correlation adjusted for age at blood collection (as a continuous variable), BMI (in fifths based on the distribution in controls), and sex. Concentrations of all analytes, except total T3, were logarithmically transformed to approximate the normal distribution. As presence of TgAb can affect the measurement of Tg concentrations, participants positive (n = 142) or unknown (n = 11) for TgAb were excluded from all analysis of Tg.
ORs and 95% confidence intervals (CI) for differentiated thyroid carcinoma by thirds of IGF-I concentrations (based on the distributions in controls) were estimated by use of conditional logistic regression, conditioned on the matching variables and adjusted for age at blood collection (as a continuous variable) to eliminate any residual confounding by age after the matching. Using sex-specific tertile cutoff points did not materially change the results; therefore, the combined tertile cutoff points were used for all analyses.
The linear trend of IGF-I concentration with risk of differentiated thyroid carcinoma was computed using the logarithm of IGF-I concentration as a continuous variable in the conditional logistic model. As the logarithm to base 2 was used in this analysis, the ORs and 95% CIs correspond to the odds of differentiated thyroid carcinoma associated with a doubling of IGF-I concentration. Because of the incidence density sampling of controls, all ORs estimate the incidence rate ratio for the IGF-I concentrations in relation to differentiated thyroid carcinoma in the study cohort (28).
To investigate the role of potential confounders, further multivariate conditional logistic regression models were fitted. First, a multivariable-adjusted model, which included variables strongly related to case status or IGF-I concentrations in this dataset, i.e., age at blood collection, height, and BMI. Second, a further adjusted model additionally including participant characteristics previously shown to be related to IGF-I concentrations and/or risk of thyroid cancer, i.e., the waist to hip ratio, alcohol, smoking, physical activity (assessed by use of the Cambridge Physical Activity Index; ref. 29), education, and in women only, age at first period, number of full-term pregnancies and age at first pregnancy, and for postmenopausal women only, age at menopause. Finally, concentrations of TSH, Tg, TgAb, and thyroid hormones (in thirds based on the distribution in controls; unknown; except for TgAb: positive; negative; unknown) were added separately to the age-adjusted model.
The ORs in thirds and for a doubling of IGF-I concentrations in subgroups, and the heterogeneity of the ORs between subgroups, i.e., sex, age at blood collection, education, smoking, BMI, TSH, and the blood-processing protocol were examined. The test for heterogeneity was done by use of the likelihood ratio χ2 test based on models with and without an interaction term between the logarithm of IGF-I concentration and the variable of interest. The age-adjusted model was used for subgroup analyses, as multivariate adjustment made little difference to the overall risk estimates.
Furthermore, the association between IGF-I and differentiated thyroid carcinoma was investigated by tumor characteristics, i.e., histologic type, TNM stage, and years between blood collection and diagnosis. Controls were assigned the value of their matched case and the analysis was conducted as described for the subgroups.
All analyses were performed in Stata Statistical Software versions 12.1 and 13.1 (StataCorp. LP). Two-sided P values <0.05 were considered statistically significant in all analyses.
Results
This analysis included 345 cases, of whom 289 (84%) were women and 56 (16%) were men. Among the cases, the mean duration from blood collection to diagnosis was 6.4 years, and they were on average 57.8 years old at diagnosis. Of the 345 cases, 101 (29%) were diagnosed with localized disease and 79 (23%) with more advanced disease; stage was unknown for 165 cases (48%). Papillary carcinoma was diagnosed in 253 cases (73%), follicular carcinoma in 56 cases (16%), and 36 cases (11%) were diagnosed with not otherwise specified differentiated thyroid carcinoma.
Compared with controls, female cases were on average taller, heavier, and had a higher waist to hip ratio (Table 1). No statistically significant differences were found between cases and controls in men, although cases were, on average, taller than controls.
. | Cases . | Controls . | Pa . | ||
---|---|---|---|---|---|
Women | |||||
n | 289 | 571 | |||
Age at blood collection (y): mean, SD | 51.3 | 8.3 | 51.3 | 8.3 | −b |
Anthropometry: mean, SD | |||||
Height (cm) | 161.6 | 6.1 | 160.3 | 6.5 | 0.002 |
Weight (kg) | 67.8 | 12.0 | 65.6 | 11.2 | 0.007 |
BMI (kg/m2) | 26.0 | 4.5 | 25.6 | 4.6 | 0.20 |
Waist to hip ratioc | 0.80 | 0.07 | 0.79 | 0.07 | 0.04 |
Alcohol (g/d): n, % | |||||
0 | 64 | 22.2 | 102 | 17.9 | |
>0–9 | 161 | 55.7 | 323 | 56.6 | |
10–19 | 39 | 13.5 | 93 | 16.3 | |
≥20 | 25 | 8.7 | 53 | 9.3 | 0.29 |
Smoking: n, %c | |||||
Never | 161 | 56.3 | 340 | 60.5 | |
Former | 65 | 22.7 | 119 | 21.1 | |
Current | 60 | 21.0 | 103 | 18.3 | 0.46 |
Physical activity: n, %c | |||||
Inactive | 93 | 33.2 | 168 | 30.5 | |
Moderately inactive | 96 | 34.3 | 209 | 37.9 | |
Moderately active or active | 91 | 32.5 | 174 | 31.6 | 0.59 |
Education: n, %c | |||||
Less than secondary school | 118 | 41.7 | 228 | 40.4 | |
Secondary school | 109 | 38.5 | 221 | 37.4 | |
University degree | 56 | 19.8 | 125 | 22.2 | 0.70 |
Menopausal status: n, % | |||||
Premenopausal | 106 | 44.6 | 212 | 37.1 | |
Perimenopausal or unknown | 54 | 18.7 | 101 | 17.7 | |
Postmenopausal | 129 | 44.6 | 258 | 45.2 | −b |
Hours since last meal at blood collection: n, %c | |||||
<3 | 123 | 44.6 | 242 | 44.3 | |
3–6 | 50 | 18.1 | 95 | 17.4 | |
>6 | 103 | 37.3 | 209 | 38.3 | −b |
Men | |||||
n | 56 | 164 | |||
Age at blood collection (y): mean, SD | 51.5 | 9.3 | 51.7 | 9.4 | −b |
Anthropometry: mean, SD | |||||
Height (cm) | 176.1 | 6.5 | 174.5 | 5.8 | 0.14 |
Weight (kg) | 81.1 | 11.7 | 81.0 | 10.9 | 0.99 |
BMI (kg/m2) | 26.1 | 2.9 | 26.6 | 3.2 | 0.21 |
Waist to hip ratioc | 0.94 | 0.07 | 0.94 | 0.06 | 0.98 |
Alcohol (g/d): n, % | |||||
0 | 5 | 8.9 | 7 | 4.3 | |
>0–9 | 20 | 35.7 | 58 | 35.4 | |
10–19 | 9 | 16.1 | 40 | 24.4 | |
≥20 | 22 | 39.3 | 59 | 36.0 | 0.40 |
Smoking: n, %c | |||||
Never | 18 | 33.3 | 52 | 31.7 | |
Former | 21 | 38.9 | 65 | 39.7 | |
Current | 15 | 27.8 | 47 | 28.7 | 0.99 |
Physical activity: n, %c | |||||
Inactive | 11 | 20.0 | 27 | 16.5 | |
Moderately inactive | 21 | 38.2 | 70 | 42.7 | |
Moderately active or active | 23 | 42.8 | 67 | 40.9 | 0.75 |
Education: n, %c | |||||
Less than secondary school | 17 | 32.1 | 48 | 29.5 | |
Secondary school | 22 | 41.5 | 64 | 39.3 | |
University degree | 14 | 26.4 | 51 | 31.2 | 0.83 |
Hours since last meal at blood collection: n, %c | |||||
<3 | 20 | 37.0 | 68 | 43.0 | |
3–6 | 12 | 22.2 | 22 | 13.9 | |
>6 | 22 | 40.7 | 68 | 43.0 | −b |
. | Cases . | Controls . | Pa . | ||
---|---|---|---|---|---|
Women | |||||
n | 289 | 571 | |||
Age at blood collection (y): mean, SD | 51.3 | 8.3 | 51.3 | 8.3 | −b |
Anthropometry: mean, SD | |||||
Height (cm) | 161.6 | 6.1 | 160.3 | 6.5 | 0.002 |
Weight (kg) | 67.8 | 12.0 | 65.6 | 11.2 | 0.007 |
BMI (kg/m2) | 26.0 | 4.5 | 25.6 | 4.6 | 0.20 |
Waist to hip ratioc | 0.80 | 0.07 | 0.79 | 0.07 | 0.04 |
Alcohol (g/d): n, % | |||||
0 | 64 | 22.2 | 102 | 17.9 | |
>0–9 | 161 | 55.7 | 323 | 56.6 | |
10–19 | 39 | 13.5 | 93 | 16.3 | |
≥20 | 25 | 8.7 | 53 | 9.3 | 0.29 |
Smoking: n, %c | |||||
Never | 161 | 56.3 | 340 | 60.5 | |
Former | 65 | 22.7 | 119 | 21.1 | |
Current | 60 | 21.0 | 103 | 18.3 | 0.46 |
Physical activity: n, %c | |||||
Inactive | 93 | 33.2 | 168 | 30.5 | |
Moderately inactive | 96 | 34.3 | 209 | 37.9 | |
Moderately active or active | 91 | 32.5 | 174 | 31.6 | 0.59 |
Education: n, %c | |||||
Less than secondary school | 118 | 41.7 | 228 | 40.4 | |
Secondary school | 109 | 38.5 | 221 | 37.4 | |
University degree | 56 | 19.8 | 125 | 22.2 | 0.70 |
Menopausal status: n, % | |||||
Premenopausal | 106 | 44.6 | 212 | 37.1 | |
Perimenopausal or unknown | 54 | 18.7 | 101 | 17.7 | |
Postmenopausal | 129 | 44.6 | 258 | 45.2 | −b |
Hours since last meal at blood collection: n, %c | |||||
<3 | 123 | 44.6 | 242 | 44.3 | |
3–6 | 50 | 18.1 | 95 | 17.4 | |
>6 | 103 | 37.3 | 209 | 38.3 | −b |
Men | |||||
n | 56 | 164 | |||
Age at blood collection (y): mean, SD | 51.5 | 9.3 | 51.7 | 9.4 | −b |
Anthropometry: mean, SD | |||||
Height (cm) | 176.1 | 6.5 | 174.5 | 5.8 | 0.14 |
Weight (kg) | 81.1 | 11.7 | 81.0 | 10.9 | 0.99 |
BMI (kg/m2) | 26.1 | 2.9 | 26.6 | 3.2 | 0.21 |
Waist to hip ratioc | 0.94 | 0.07 | 0.94 | 0.06 | 0.98 |
Alcohol (g/d): n, % | |||||
0 | 5 | 8.9 | 7 | 4.3 | |
>0–9 | 20 | 35.7 | 58 | 35.4 | |
10–19 | 9 | 16.1 | 40 | 24.4 | |
≥20 | 22 | 39.3 | 59 | 36.0 | 0.40 |
Smoking: n, %c | |||||
Never | 18 | 33.3 | 52 | 31.7 | |
Former | 21 | 38.9 | 65 | 39.7 | |
Current | 15 | 27.8 | 47 | 28.7 | 0.99 |
Physical activity: n, %c | |||||
Inactive | 11 | 20.0 | 27 | 16.5 | |
Moderately inactive | 21 | 38.2 | 70 | 42.7 | |
Moderately active or active | 23 | 42.8 | 67 | 40.9 | 0.75 |
Education: n, %c | |||||
Less than secondary school | 17 | 32.1 | 48 | 29.5 | |
Secondary school | 22 | 41.5 | 64 | 39.3 | |
University degree | 14 | 26.4 | 51 | 31.2 | 0.83 |
Hours since last meal at blood collection: n, %c | |||||
<3 | 20 | 37.0 | 68 | 43.0 | |
3–6 | 12 | 22.2 | 22 | 13.9 | |
>6 | 22 | 40.7 | 68 | 43.0 | −b |
aThe weighted paired-sample t test and the likelihood ratio χ2 test from conditional logistic regression were used for continuous and categorical variables, respectively.
bNo P value was calculated for matching factors.
cUnknown for some participants; mean, SD or n %, and tests exclude unknown values.
In controls, the geometric mean IGF-I concentration was higher among individuals who were younger at blood collection and who had a lower weight (Table 2). IGF-I concentrations differed according to BMI, with the highest geometric mean IGF-I concentration in the second fifth of BMI (22.3–24.4 kg/m2) and the lowest concentration in the highest fifth (>28.6 kg/m2). IGF-I concentration also differed by country, being lowest in Greece (15.4 nmol/L) and highest in the Netherlands (19.4 nmol/L; Pdifference = 0.004 across all countries); further adjustment for country did not influence the other associations, except for attenuation of the association with weight (results not shown).
. | n . | Geometric mean and 95% CIa . | Pdifferenceb . | Ptrendc . |
---|---|---|---|---|
All controls | 735 | 17.1 (16.8–17.5) | − | − |
Sex | ||||
Male | 164 | 17.5 (16.7–18.3) | ||
Female | 571 | 17.0 (16.6–17.4) | 0.33 | − |
Age at blood collection, y | ||||
<48 | 253 | 19.3 (18.6–20.0) | ||
48–55 | 253 | 16.8 (16.2–17.4) | ||
≥56 | 229 | 15.3 (14.7–15.9) | <0.0001 | <0.0001 |
Height (cm) | ||||
<160 | 253 | 17.2 (16.5–17.9) | ||
160–164 | 182 | 17.5 (16.7–18.3) | ||
165–169 | 134 | 17.6 (16.8–18.4) | ||
≥170 | 166 | 16.3 (15.4–17.3) | 0.27 | 0.62 |
Weight (kg) | ||||
<60 | 187 | 17.8 (17.1–18.6) | ||
60–69 | 247 | 17.2 (16.6–17.9) | ||
70–79 | 162 | 17.1 (16.4–17.9) | ||
≥80 | 139 | 16.1 (15.3–16.9) | 0.04 | 0.01 |
BMI (kg/m2) | ||||
<22.3 | 147 | 17.5 (16.7–18.3) | ||
22.3–24.4 | 146 | 18.1 (17.3–19.0) | ||
24.4–26.3 | 147 | 17.4 (16.7–18.3) | ||
26.3–28.6 | 148 | 16.8 (16.0–17.6) | ||
>28.6 | 147 | 15.9 (15.2–16.6) | 0.001 | 0.0003 |
Waist to hip ratiod | ||||
<0.76 | 163 | 17.6 (16.7–18.5) | ||
0.76–0.80 | 160 | 17.5 (16.7–18.4) | ||
0.81–0.88 | 195 | 17.3 (16.6–18.0) | ||
≥0.89 | 177 | 16.5 (15.5–17.6) | 0.58 | 0.27 |
Alcohol (g/d) | ||||
0 | 109 | 16.5 (15.6–17.4) | ||
>0–9 | 381 | 17.2 (16.8–17.7) | ||
10–19 | 133 | 17.0 (16.2–17.9) | ||
≥20 | 112 | 17.5 (16.6–18.5) | 0.42 | 0.26 |
Smokingd | ||||
Never | 392 | 17.2 (16.7–17.7) | ||
Former | 184 | 17.4 (16.6–18.1) | ||
Current | 150 | 16.7 (16.0–17.5) | 0.51 | 0.50 |
Physical activityd | ||||
Inactive | 195 | 16.8 (16.1–17.5) | ||
Moderately inactive | 279 | 17.1 (16.6–17.7) | ||
Moderately active or active | 241 | 17.5 (16.9–18.1) | 0.37 | 0.16 |
Educationd | ||||
Less than secondary | 276 | 16.8 (16.2–17.4) | ||
Secondary school | 275 | 17.6 (17.0–18.2) | ||
University degree | 176 | 17.1 (16.4–17.8) | 0.16 | 0.42 |
Menopausal statuse | ||||
Premenopausal | 212 | 17.2 (16.3–18.2) | ||
Perimenopausal or unknown | 101 | 16.5 (15.6–17.5) | ||
Postmenopausal | 258 | 17.2 (16.4–18.0) | 0.44 | − |
Hormone replacement therapy at blood collectiond,e | ||||
No | 462 | 17.1 (16.7–17.6) | ||
Yes | 95 | 17.2 (16.2–18.2) | 0.94 | − |
Hours since last meal at blood collectiond | ||||
<3 | 310 | 17.0 (16.5–17.6) | ||
3–6 | 117 | 16.9 (16.1–17.8) | ||
>6 | 277 | 17.5 (17.0–18.1) | 0.35 | 0.21 |
. | n . | Geometric mean and 95% CIa . | Pdifferenceb . | Ptrendc . |
---|---|---|---|---|
All controls | 735 | 17.1 (16.8–17.5) | − | − |
Sex | ||||
Male | 164 | 17.5 (16.7–18.3) | ||
Female | 571 | 17.0 (16.6–17.4) | 0.33 | − |
Age at blood collection, y | ||||
<48 | 253 | 19.3 (18.6–20.0) | ||
48–55 | 253 | 16.8 (16.2–17.4) | ||
≥56 | 229 | 15.3 (14.7–15.9) | <0.0001 | <0.0001 |
Height (cm) | ||||
<160 | 253 | 17.2 (16.5–17.9) | ||
160–164 | 182 | 17.5 (16.7–18.3) | ||
165–169 | 134 | 17.6 (16.8–18.4) | ||
≥170 | 166 | 16.3 (15.4–17.3) | 0.27 | 0.62 |
Weight (kg) | ||||
<60 | 187 | 17.8 (17.1–18.6) | ||
60–69 | 247 | 17.2 (16.6–17.9) | ||
70–79 | 162 | 17.1 (16.4–17.9) | ||
≥80 | 139 | 16.1 (15.3–16.9) | 0.04 | 0.01 |
BMI (kg/m2) | ||||
<22.3 | 147 | 17.5 (16.7–18.3) | ||
22.3–24.4 | 146 | 18.1 (17.3–19.0) | ||
24.4–26.3 | 147 | 17.4 (16.7–18.3) | ||
26.3–28.6 | 148 | 16.8 (16.0–17.6) | ||
>28.6 | 147 | 15.9 (15.2–16.6) | 0.001 | 0.0003 |
Waist to hip ratiod | ||||
<0.76 | 163 | 17.6 (16.7–18.5) | ||
0.76–0.80 | 160 | 17.5 (16.7–18.4) | ||
0.81–0.88 | 195 | 17.3 (16.6–18.0) | ||
≥0.89 | 177 | 16.5 (15.5–17.6) | 0.58 | 0.27 |
Alcohol (g/d) | ||||
0 | 109 | 16.5 (15.6–17.4) | ||
>0–9 | 381 | 17.2 (16.8–17.7) | ||
10–19 | 133 | 17.0 (16.2–17.9) | ||
≥20 | 112 | 17.5 (16.6–18.5) | 0.42 | 0.26 |
Smokingd | ||||
Never | 392 | 17.2 (16.7–17.7) | ||
Former | 184 | 17.4 (16.6–18.1) | ||
Current | 150 | 16.7 (16.0–17.5) | 0.51 | 0.50 |
Physical activityd | ||||
Inactive | 195 | 16.8 (16.1–17.5) | ||
Moderately inactive | 279 | 17.1 (16.6–17.7) | ||
Moderately active or active | 241 | 17.5 (16.9–18.1) | 0.37 | 0.16 |
Educationd | ||||
Less than secondary | 276 | 16.8 (16.2–17.4) | ||
Secondary school | 275 | 17.6 (17.0–18.2) | ||
University degree | 176 | 17.1 (16.4–17.8) | 0.16 | 0.42 |
Menopausal statuse | ||||
Premenopausal | 212 | 17.2 (16.3–18.2) | ||
Perimenopausal or unknown | 101 | 16.5 (15.6–17.5) | ||
Postmenopausal | 258 | 17.2 (16.4–18.0) | 0.44 | − |
Hormone replacement therapy at blood collectiond,e | ||||
No | 462 | 17.1 (16.7–17.6) | ||
Yes | 95 | 17.2 (16.2–18.2) | 0.94 | − |
Hours since last meal at blood collectiond | ||||
<3 | 310 | 17.0 (16.5–17.6) | ||
3–6 | 117 | 16.9 (16.1–17.8) | ||
>6 | 277 | 17.5 (17.0–18.1) | 0.35 | 0.21 |
aAdjusted for age at blood collection (as a continuous variable), BMI (in fifths based on the distribution in controls), and sex, in which the adjustment variable was not the variable of interest. Analysis of weight was not adjusted for BMI.
bP values refer to tests of difference between the logarithm of IGF-I concentration in the separate categories (excluding unknowns) calculated by ANOVA.
cP values refer to tests of linear trend calculated by scoring categories with consecutive whole numbers (excluding unknowns).
dUnknown for some participants.
eIn women only.
The partial correlations between concentrations of IGF-I and TSH, Tg, and thyroid hormones in controls, adjusted for age at blood collection, BMI, and sex, were all low (correlation coefficients ranged from −0.05 to 0.03).
Cases had a higher geometric mean IGF-I concentration than controls (17.8 nmol/L; 95% CI, 17.3–18.4 vs. 17.1; 95% CI, 16.8–17.5; Pdifference = 0.03), and IGF-I concentration was positively associated with risk of differentiated thyroid carcinoma; the OR for a doubling in IGF-I concentration was 1.48 (95% CI, 1.06–2.08; Ptrend = 0.02) after adjustment for age at blood collection (Table 3). Compared with those in the lowest third of IGF-I concentration, those in the highest third had a 37% increased risk (OR, 1.37; 95% CI, 0.99–1.91) after adjustment for age at blood collection, whereas risk was not elevated in the middle third (OR, 0.99; 95% CI, 0.71–1.39). Additional adjustment for height and BMI did not materially change the results, whereas further adjustment for lifestyle and reproductive factors attenuated the results slightly, resulting in a borderline significant result (Table 3). Adding the concentrations of TSH, Tg, TgAb, and the thyroid hormones separately to the age-adjusted model did not change the results markedly; the ORs for a doubling of IGF-I concentration varied between 1.45 (95% CI, 1.02–2.04; Ptrend = 0.03) when adjusting for TSH and 1.61 (95% CI, 1.14–2.27; Ptrend = 0.01) when adjusting for free T4, and all models stayed statistically significant.
. | . | Bottom third . | Middle third . | Top third . | Doublinga . | ||
---|---|---|---|---|---|---|---|
. | . | <15.3 (13.0) nmol/Lb . | 15.3–19.6 (17.4) nmol/Lb . | >19.6 (22.8) nmol/Lb . | . | . | . |
. | . | OR (95% CI) . | OR (95% CI) . | OR (95% CI) . | OR (95% CI) . | Ptrend . | Pheterogeneityc . |
ncase/ncontrol | 105/245 | 101/245 | 139/245 | ||||
Risk overalld | |||||||
Age-adjusted model | 345/735 | 1 (ref.) | 0.99 (0.71–1.39) | 1.37 (0.99–1.91) | 1.48 (1.06–2.08) | 0.02 | − |
Multivariable-adjusted modele | 345/735 | 1 (ref.) | 1.00 (0.71–1.41) | 1.37 (0.98–1.92) | 1.46 (1.04–2.07) | 0.03 | − |
Further adjusted modelf | 345/735 | 1 (ref.) | 1.02 (0.72–1.46) | 1.33 (0.93–1.88) | 1.41 (0. 99–2.01) | 0.06 | − |
Risk in subgroupsd | |||||||
Sex | |||||||
Male | 56/164 | 1 (ref.) | 1.01 (0.45–2.28) | 1.63 (0.73–3.65) | 1.39 (0.61–3.16) | 0.43 | |
Female | 289/571 | 1 (ref.) | 0.99 (0.68–1.44) | 1.33 (0.92–1.91) | 1.50 (1.04–2.18) | 0.03 | 0.87 |
Age at blood collection, y | |||||||
<50 | 146/310 | 1 (ref.) | 1.55 (0.84–2.84) | 1.59 (0.91–2.78) | 1.54 (0.90–2.66) | 0.12 | |
≥50 | 199/425 | 1 (ref.) | 0.79 (0.52–1.20) | 1.39 (0.91–2.13) | 1.45 (0.94–2.22) | 0.09 | 0.86 |
Educationg | |||||||
Less than secondary school | 135/273 | 1 (ref.) | 1.10 (0.66–1.86) | 1.47 (0.87–2.48) | 1.80 (1.08–3.00) | 0.02 | |
Secondary school or more | 201/436 | 1 (ref.) | 0.84 (0.53–1.31) | 1.20 (0.79–1.83) | 1.27 (0.83–1.93) | 0.27 | 0.27 |
Smoking | |||||||
Never | 179/390 | 1 (ref.) | 1.02 (0.65–1.60) | 1.37 (0.87–2.17) | 1.62 (1.03–2.56) | 0.04 | |
Ever | 161/326 | 1 (ref.) | 0.96 (0.58–1.58) | 1.39 (0.87–2.24) | 1.39 (0.88–2.20) | 0.16 | 0.62 |
BMI (kg/m2) | |||||||
<25 | 158/347 | 1 (ref.) | 0.92 (0.55–1.53) | 1.32 (0.81–2.16) | 1.45 (0.88–2.38) | 0.14 | |
≥25 | 187/388 | 1 (ref.) | 1.05 (0.67–1.64) | 1.44 (0.92–2.26) | 1.55 (1.00–2.42) | 0.05 | 0.84 |
Thyroid stimulation hormone (mIU/L)g | |||||||
<1.27 | 204/365 | 1 (ref.) | 1.03 (0.66–1.61) | 1.37 (0.89–2.11) | 1.46 (0.96–2.22) | 0.08 | |
≥1.27 | 140/366 | 1 (ref.) | 1.00 (0.59–1.67) | 1.35 (0.83–2.22) | 1.52 (0.91–2.52) | 0.11 | 0.90 |
Risk by tumor characteristicsd | |||||||
Histologic typeg | |||||||
Papillary | 253/538 | 1 (ref.) | 1.30 (0.86–1.95) | 1.62 (1.09–2.40) | 1.69 (1.13–2.55) | 0.01 | |
Follicular | 56/122 | 1 (ref.) | 0.60 (0.27–1.38) | 0.92 (0.41–2.08) | 1.13 (0.53–2.37) | 0.76 | 0.35 |
TNM stageg | |||||||
Localized | 101/211 | 1 (ref.) | 0.92 (0.47–1.78) | 1.48 (0.82–2.70) | 1.44 (0.81–2.54) | 0.22 | |
More advanced | 79/172 | 1 (ref.) | 2.01 (1.00–4.06) | 1.53 (0.75–3.13) | 1.91 (0.90–4.08) | 0.09 | 0.55 |
Years between blood collection and diagnosis | |||||||
<5 | 137/293 | 1 (ref.) | 0.79 (0.46–1.35) | 1.33 (0.80–2.21) | 1.48 (0.84–2.61) | 0.18 | |
≥5 | 208/442 | 1 (ref.) | 1.15 (0.74–1.80) | 1.42 (0.92–2.20) | 1.48 (0.97–2.26) | 0.07 | 0.99 |
. | . | Bottom third . | Middle third . | Top third . | Doublinga . | ||
---|---|---|---|---|---|---|---|
. | . | <15.3 (13.0) nmol/Lb . | 15.3–19.6 (17.4) nmol/Lb . | >19.6 (22.8) nmol/Lb . | . | . | . |
. | . | OR (95% CI) . | OR (95% CI) . | OR (95% CI) . | OR (95% CI) . | Ptrend . | Pheterogeneityc . |
ncase/ncontrol | 105/245 | 101/245 | 139/245 | ||||
Risk overalld | |||||||
Age-adjusted model | 345/735 | 1 (ref.) | 0.99 (0.71–1.39) | 1.37 (0.99–1.91) | 1.48 (1.06–2.08) | 0.02 | − |
Multivariable-adjusted modele | 345/735 | 1 (ref.) | 1.00 (0.71–1.41) | 1.37 (0.98–1.92) | 1.46 (1.04–2.07) | 0.03 | − |
Further adjusted modelf | 345/735 | 1 (ref.) | 1.02 (0.72–1.46) | 1.33 (0.93–1.88) | 1.41 (0. 99–2.01) | 0.06 | − |
Risk in subgroupsd | |||||||
Sex | |||||||
Male | 56/164 | 1 (ref.) | 1.01 (0.45–2.28) | 1.63 (0.73–3.65) | 1.39 (0.61–3.16) | 0.43 | |
Female | 289/571 | 1 (ref.) | 0.99 (0.68–1.44) | 1.33 (0.92–1.91) | 1.50 (1.04–2.18) | 0.03 | 0.87 |
Age at blood collection, y | |||||||
<50 | 146/310 | 1 (ref.) | 1.55 (0.84–2.84) | 1.59 (0.91–2.78) | 1.54 (0.90–2.66) | 0.12 | |
≥50 | 199/425 | 1 (ref.) | 0.79 (0.52–1.20) | 1.39 (0.91–2.13) | 1.45 (0.94–2.22) | 0.09 | 0.86 |
Educationg | |||||||
Less than secondary school | 135/273 | 1 (ref.) | 1.10 (0.66–1.86) | 1.47 (0.87–2.48) | 1.80 (1.08–3.00) | 0.02 | |
Secondary school or more | 201/436 | 1 (ref.) | 0.84 (0.53–1.31) | 1.20 (0.79–1.83) | 1.27 (0.83–1.93) | 0.27 | 0.27 |
Smoking | |||||||
Never | 179/390 | 1 (ref.) | 1.02 (0.65–1.60) | 1.37 (0.87–2.17) | 1.62 (1.03–2.56) | 0.04 | |
Ever | 161/326 | 1 (ref.) | 0.96 (0.58–1.58) | 1.39 (0.87–2.24) | 1.39 (0.88–2.20) | 0.16 | 0.62 |
BMI (kg/m2) | |||||||
<25 | 158/347 | 1 (ref.) | 0.92 (0.55–1.53) | 1.32 (0.81–2.16) | 1.45 (0.88–2.38) | 0.14 | |
≥25 | 187/388 | 1 (ref.) | 1.05 (0.67–1.64) | 1.44 (0.92–2.26) | 1.55 (1.00–2.42) | 0.05 | 0.84 |
Thyroid stimulation hormone (mIU/L)g | |||||||
<1.27 | 204/365 | 1 (ref.) | 1.03 (0.66–1.61) | 1.37 (0.89–2.11) | 1.46 (0.96–2.22) | 0.08 | |
≥1.27 | 140/366 | 1 (ref.) | 1.00 (0.59–1.67) | 1.35 (0.83–2.22) | 1.52 (0.91–2.52) | 0.11 | 0.90 |
Risk by tumor characteristicsd | |||||||
Histologic typeg | |||||||
Papillary | 253/538 | 1 (ref.) | 1.30 (0.86–1.95) | 1.62 (1.09–2.40) | 1.69 (1.13–2.55) | 0.01 | |
Follicular | 56/122 | 1 (ref.) | 0.60 (0.27–1.38) | 0.92 (0.41–2.08) | 1.13 (0.53–2.37) | 0.76 | 0.35 |
TNM stageg | |||||||
Localized | 101/211 | 1 (ref.) | 0.92 (0.47–1.78) | 1.48 (0.82–2.70) | 1.44 (0.81–2.54) | 0.22 | |
More advanced | 79/172 | 1 (ref.) | 2.01 (1.00–4.06) | 1.53 (0.75–3.13) | 1.91 (0.90–4.08) | 0.09 | 0.55 |
Years between blood collection and diagnosis | |||||||
<5 | 137/293 | 1 (ref.) | 0.79 (0.46–1.35) | 1.33 (0.80–2.21) | 1.48 (0.84–2.61) | 0.18 | |
≥5 | 208/442 | 1 (ref.) | 1.15 (0.74–1.80) | 1.42 (0.92–2.20) | 1.48 (0.97–2.26) | 0.07 | 0.99 |
aOR, 95% CI, and Ptrend were based on the logarithm (base2) of IGF-I concentrations (as a continuous variable) and correspond to the risk associated with a doubling in IGF-I concentration.
bTertile cutoff points were based on the distribution controls only. The numbers shown are the range (median) of IGF-I concentrations in each third.
cP values refer to the likelihood ratio χ2 test between models of the linear trend in logarithm (base 2) of IGF-I concentrations with and without an interaction term with the variable of interest.
dAll analyses were logistic regression models conditioned on the matching variables and adjusted for age at blood collection (as a continuous variable).
eThe multivariable-adjusted model was adjusted for age at blood collection (as a continuous variable), height (as a continuous variable), and BMI (in fifths based on the distribution in controls).
fThe further adjusted model was in addition to age at blood collection, height, and BMI adjusted for the waist to hip ratio (below and above the median based on the distribution in controls; unknown), alcohol (0; >0–9; 10–19; ≥20 g/d), smoking (never; former; current; and unknown), physical activity (inactive; moderately inactive; moderately active or active; and unknown), education (less than secondary school; secondary school; university degree; and unknown), and in women only, age at first period (<12; 12–14; >14 years; unknown), number of full-term pregnancies and age at first pregnancy (nulliparous, 1–2 children and <25 years; 1–2 children and ≥25 years; ≥3 children and <25 years; ≥3 children and ≥25 years; and unknown), and for postmenopausal women only, age at menopause (<50; ≥50 years; unknown).
gUnknown for some participants.
There was no significant heterogeneity in ORs between subgroups, including the blood-processing protocol (results not shown), or by tumor characteristics (Table 3). The ORs for a doubling in IGF-I concentrations were 1.69 (95% CI, 1.13–2.55) for papillary carcinoma and 1.13 (95% CI, 0.53–2.37) for follicular carcinoma.
Discussion
This prospective multicenter study suggests that IGF-I concentrations are positively related to risk of differentiated thyroid carcinoma. The prospective nature of the study design and the finding of no heterogeneity by time from blood collection to diagnosis suggest that the apparent association of IGF-I and differentiated thyroid carcinoma is not explained by any effect of preclinical tumors on IGF-I concentrations.
Little is known about the association between IGF-I concentrations and risk of thyroid cancer. Some relevant evidence is provided by studies of anthropometry and thyroid cancer. For example, IGF-I concentration early in life is a determinant of height, and adult height has been found to be positively associated with risk of differentiated thyroid carcinoma (18), although IGF-I in middle age seems to be only weakly associated with height (30). It is possible that IGF-I concentrations measured earlier in life might be more strongly related to risk of thyroid cancer than concentrations measured in middle age. Previous studies that have investigated IGF-I concentrations in relation to thyroid cancer have not found an association (12, 19, 20, 31). However, three of these studies were performed in patients with acromegaly, with a small numbers of cases (n = 3–7), and were retrospective in design. The fourth study, a case–control study did not find an association of IGF-I with thyroid cancer overall nor with its subtypes (12).
The detailed information on participant characteristics, TSH, Tg, TgAb, and thyroid hormones available in this study allowed a thorough investigation of the influence of potential confounding. Including anthropometry and lifestyle in the models did not materially change the results, although slight attenuation was seen after adjustment for reproductive factors. However, information about benign thyroid disease, which is positively associated with thyroid cancer risk (14–16) and IGF-I concentrations (13), was not available for the majority of participants and thus, we are not able to exclude the possibility of confounding by preexisting related conditions. Family history of thyroid cancer could also be a potential confounding factor in this study. However, the familial form of thyroid carcinoma only accounts for a small proportion of papillary and follicular thyroid carcinomas (32). Furthermore, no information was available on iodine deficiency, which may be positively associated with follicular carcinoma (33, 34). However, severe iodine deficiency is rare in Europe (35). We also had no information on radiation exposure.
Our findings that serum IGF-I concentration was not correlated with concentration of TSH or thyroid hormones are consistent with other reports finding no or weak associations (36, 37), and suggest that these hormones are not likely to confound the association between IGF-I concentration and differentiated thyroid carcinoma. Furthermore, additional adjustment for TSH, Tg, TgAb, and thyroid hormones did not materially change the risk estimates, despite TSH, Tg, and TgAb being associated with differentiated thyroid carcinoma in this study population (25). The opposing direction of the relationship between differentiated thyroid carcinoma and IGF-I (a positive association) and TSH (an inverse association) is intriguing. Although TSH is the strongest growth factor for thyroid follicular cells, it has been suggested that low TSH concentration may lead to predisposition to malignant transformation via reduced differentiation of thyroid follicular cells (38). There is some evidence from cell studies of a joint effect of IGF-I and TSH, as IGF-I might stimulate the transcription of the thyroid-stimulating hormone receptor gene in thyroid cells (39), but there was no evidence of heterogeneity between the association between IGF-I and differentiated thyroid carcinoma at low and high concentrations of TSH in the present study.
Our findings were based on a single IGF-I measurement. Moderate to high reproducibility of IGF-I over 4 months to 15 years has been reported (intraclass or Spearman correlation coefficients ranged between 0.4 and 0.9; refs. 40–45). Thus, some attenuation of the association between IGF-I concentration and differentiated thyroid carcinoma is likely due to our results being based on a single IGF-I measurement, and the true association might be stronger than reported in this article.
A limitation of this study is the small numbers in each group in the subgroup analyses, affecting the strength of our conclusions. Thus, further larger studies should be conducted to confirm the results of the present study. Future studies might also investigate other IGFs and IGF-binding proteins in relation to risk of differentiated thyroid carcinoma.
In conclusion, the findings of the present study suggest that IGF-I concentrations may be positively associated with risk of differentiated thyroid carcinoma.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Authors' Contributions
Conception and design: J.A. Schmidt, N.E. Allen, K.K. Tsilidis, E. Weiderpass, K. Overvad, A. Tjønneland, H. Boeing, V. Krogh, R. Tumino, H.B. Bueno-de-Mesquita, P.H. Peeters, E. Ardanaz, K.-T. Khaw, T.J. Key, R.C. Travis
Development of methodology: S.J. Tipper, K.K. Tsilidis, E. Weiderpass, R. Tumino, H.B. Bueno-de-Mesquita, R.C. Travis
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Almquist, S.J. Tipper, E. Weiderpass, K. Overvad, A. Tjønneland, M.-C. Boutron-Ruault, R. Kaaks, A. Lukanova, H. Boeing, P. Lagiou, D. Trichopoulos, A. Trichopoulou, D. Palli, V. Krogh, S. Panico, R. Tumino, R. Zanetti, H.B. Bueno-de-Mesquita, P.H. Peeters, E. Lund, A. Agudo, M.-J. Sánchez, M.-D. Chirlaque, E. Ardanaz, N. Larrañaga, J. Hennings, M. Sandström, K.-T. Khaw, N. Wareham, T.J. Key
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): J.A. Schmidt, E. Weiderpass, P.H. Peeters, A. Agudo, E. Ardanaz, M.J. Gunter, R.C. Travis
Writing, review, and/or revision of the manuscript: J.A. Schmidt, N.E. Allen, M. Almquist, S. Franceschi, S. Rinaldi, S.J. Tipper, K.K. Tsilidis, E. Weiderpass, K. Overvad, A. Tjønneland, M.-C. Boutron-Ruault, L. Dossus, S. Mesrine, R. Kaaks, A. Lukanova, H. Boeing, P. Lagiou, D. Trichopoulos, A. Trichopoulou, D. Palli, V. Krogh, S. Panico, R. Zanetti, H.B. Bueno-de-Mesquita, P.H. Peeters, E. Lund, A. Agudo, M.-J. Sánchez, M.-D. Chirlaque, E. Ardanaz, N. Larrañaga, J. Hennings, M. Sandström, K.-T. Khaw, N. Wareham, I. Romieu, M.J. Gunter, E. Riboli, T.J. Key, R.C. Travis
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): S.J. Tipper, E. Weiderpass, K. Overvad, R. Tumino, H.B. Bueno-de-Mesquita, P.H. Peeters, V. Menéndez, M.-J. Sánchez, M.-D. Chirlaque, E. Ardanaz
Study supervision: E. Weiderpass, R. Tumino, H.B. Bueno-de-Mesquita, E. Ardanaz, R.C. Travis
Acknowledgments
The authors thank all participants in the EPIC cohort for their invaluable contribution to the study. The authors also thank Paul Appleby, Cancer Epidemiology Unit, University of Oxford, for comments on the article.
Grant Support
This work was supported by Cancer Research UK (A1 1692; co-investigator, T.J. Key), which funded the analysis of IGF-I for this study. The analyses of TSH, Tg, TgAb, and thyroid hormones were funded by World Cancer Research Fund (2009/92; principal investigator, S. Rinaldi; co-investigator, S. Franceschi). The EPIC (all authors) is financially supported by the European Commission; the International Agency for Research on Cancer; Danish Cancer Society (Denmark); Ligue contre le Cancer, Institut Gustave Roussy, Mutuelle Générale de l'Education Nationale, and Institut National de la Santé et de la Recherche Médicale (INSERM; France); Deutsche Krebshilfe, Deutsches Krebsforschungszentrum, and Federal Ministry of Education and Research (Germany); the Hellenic Health Foundation (Greece); Associazione Italiana per la Ricerca sul Cancro (AIRC; Italy); Dutch Ministry of Public Health, Welfare and Sports (VWS), the Netherlands Cancer Registry (NKR), LK Research Funds, Dutch Prevention Funds, Dutch ZON (Zorg Onderzoek Nederland), and World Cancer Research Fund (WCRF; the Netherlands); European Research Council, Norwegian Cancer Society, and Norwegian Research Council (Norway); the Health Research Fund (FIS) of the Spanish Ministry of Health (exp. P10710130), Regional Governments of Andalucía, Asturias, Basque Country, Murcia (no. 6236), Navarra and the Catalan Institute of Oncology, La Caixa (BM 06–130), RTICC-RD06/10091 (Spain); Swedish Cancer Society, Swedish Scientific Council, and Regional Government of Skåne and Västerbotten (Sweden); and Cancer Research UK and Medical Research Council (United Kingdom).
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.