Associations of Plasma C-Peptide and IGFBP-1 Levels with Risk of Colorectal Adenoma in a Multiethnic Population

Abstract

Background

Circulating levels of insulin and IGF hormones have been associated with colorectal cancer risk, but few studies have examined their associations with colorectal adenoma.

Methods

We measured plasma C-peptide, a marker of insulin secretion, and IGF hormones in a case-control study of 554 pathologically-confirmed, first-time, adenoma cases and 786 controls with normal endoscopy among Caucasians, Japanese and Native Hawaiians in Hawaii.

Results

High plasma levels of C-peptide were statistically significantly associated with risk of colorectal adenoma [multivariate odds ratio (95% confidence interval) for increasing quartiles: 1.0, 0.91 (0.65–1.27), 1.21 (0.86–1.71) and 1.79 (1.23–2.60); p_trend: 0.0002]. We also observed a statistically significant inverse association between plasma IGFBP-1 levels and adenoma risk [1.0, 0.97 (0.70–1.34), 0.82 (0.58–1.15) and 0.47 (0.32–0.70), p_trend: <0.0001]. These associations remain significant after adjusting for each other and were not confounded by known risk factors. IGF-I, IGFBP-3, BMI and waist or hip circumference were not independently associated with adenoma risk.

Conclusion

These results provide evidence for an association of insulin and IGFBP-1 levels with colorectal adenoma.

Impact

This study suggests that hyperinsulinemia and IGF hormones may act as etiological factors in colorectal carcinogenesis, as early as during adenoma formation.

INTRODUCTION

High caloric diets, lack of physical activity and excess body weight have consistently been associated with colorectal cancer risk. These associations have been proposed to result from the proliferative and anti-apoptotic effects of the increased circulating insulin levels that are observed among overweight and obese individuals (1). Indeed, current epidemiologic evidence suggests that high prediagnostic insulin levels are associated with a 35% elevated risk of colorectal cancer (2,3). In contrast, the evidence linking hyperinsulinemia to the risk of colorectal adenoma, a known colorectal cancer precursor, is more limited and has remained inconsistent (4–6). IGF hormones are also suspected to play a role in colon carcinogenesis through mechanisms similar to insulin but their associations with colorectal cancer have been more inconsistent (7).

We report here on the largest study to date examining the association of plasma C-peptide (a marker of insulin secretion) and IGF hormones with colorectal adenoma risk.

METHODS

Subjects

Study design and data collection for this colorectal adenoma study has been described in detail elsewhere (8,9). Briefly, two flexible-sigmoidoscopy screening clinics were first used to recruit participants on Oahu, Hawaii. Adenoma cases were identified either as part of the baseline screening exam at the Hawaii site of the Prostate Lung Colorectal and Ovarian (PLCO) screening trial between July 1996 and February 2000 or at the Kaiser Permanente Hawaii (KPH)’s Gastroenterology Screening Clinic between January 1995 and June 2006. In addition, starting in June 2002, we also attempted to recruit all eligible patients who underwent a colonoscopy in the KPH Gastroenterology Department. Cases were patients with histologically confirmed first-time adenoma(s) of the colorectum and were of Japanese, Caucasian or Hawaiian race/ethnicity. Controls were selected among individuals found to have a normal colon and rectum at endoscopy, and were individually matched to the cases (with a one-to-one ratio) on age, sex, race/ethnicity, screening date (±3 months), clinic and type of examination (colonoscopy or flexible sigmoidoscopy). The study participation rate was 68% for cases and 69% for controls. Blood was provided by 87.5% of cases and 86.3% of controls who were interviewed. The present analyses were based on 554 cases and 786 controls who were interviewed up to February 2007 and gave a blood sample.

Exposure information was collected via an interview-administered questionnaire designed to obtain demographic and lifestyle information, including lifetime histories of tobacco smoking and alcohol drinking, weight at time of examination, usual physical activity, personal medical history, family history of colorectal cancer, and for females, reproductive and hormone use histories. The interview also included a validated food frequency questionnaire with more than two-hundred food items (10,11) and a meat module assessing frequency of consumption and degree of doneness for various meats cooked with high temperature methods (broiling, grilling/barbecuing, pan-frying). Detailed information on vitamin and mineral supplement use was also collected. Each subject was also asked to donate a blood sample that was drawn in the morning after a ten-hour fast. Waist and hip circumferences were measured by trained study personnel. Specimens were processed within 2 hours of collection and stored at −80°C until laboratory analysis. All participants with an available plasma sample were included in the study regardless of whether both members of a matched case-control set had given blood.

Plasma IGF-hormones (IGF-I, IGFBP-1, IGFBP-3) and C-peptide were analyzed at the International Agency for Research on Cancer blinded to the subject’s case-control status. Samples of cases and controls were assayed together in the same analytical batch giving priority to matched sets and, in situations where plasma was not available for all members of a matched set (due to refusal to give blood), to cases and controls of the same sex, ethnicity and similar age. C-peptide concentrations were measured with a radioimmunoassay, IGF-I and IGFBP-3 concentrations by enzyme-linked immunoabsorbent assays and IGFBP-1 by immunoradiometric assay (Diagnostic System Laboratories, Webster, TX). The IGF-I assay included an initial acid-ethanol precipitation step to separate IGF-I from its binding proteins. Based on 52 blind duplicate sample pairs analyzed with the study samples, the mean coefficient of variation for these assays was 4.6%, 5.3%, 6.4% and 6.5% for C-peptide, IGFBP-1, IGF-I and IGFBP-3, respectively.

Statistical Analysis

Pearson’s χ² and Wilcoxon’s rank sum test were used to compare the distribution of demographic characteristics between cases and controls. General linear model (GLM) was used to compare BMI, waist circumference and plasma biomarker levels (C-peptide, IGF-I, IGFBP-1, IGFBP-3, and the molar IGF1/IGFBP3 ratio) across ethnic groups by gender, with or without adjustment for age at blood draw and months of post-menopausal estrogen use (men were assigned zero values). Kruskal-Wallis test was also applied in comparing the biomarkers among ethnic groups because not all biomarkers were normally distributed. Partial Spearman’s correlation coefficients were computed to examine correlations among the blood biomarkers, BMI, waist and hip circumferences, waist-to-hip ratio and energy (METs) expended during an average day.

To estimate the risk of colorectal adenoma conferred by BMI and the various blood biomarkers, we used unconditional logistic regression to obtain odds ratios (ORs) and 95% confidence intervals (CIs). Because all subjects who gave blood (86.8% of interviewed subjects) were used in this analysis and because they differed across the three recruitment sources (PLCO, KPH screening clinic, KPH Gastroenterology Department) with respect to sex, age at blood draw, race/ethnicity and BMI, our analysis was unmatched and adjusted for age at blood draw, race/ethnicity, sex, recruitment site, type of examination and BMI. The regression models were further adjusted for other variables associated with adenoma risk, namely, average daily energy expenditure, , pack-years of smoking, duration of postmenopausal estrogen use, daily intake of alcohol (quartiles), daily intake of total folate (from foods and supplements) (≤400; 400–1,000; ≥1,000 DFE/day) and the logarithm of daily caloric intake, where alcohol and folate intake were expressed as nutrient density. Some variables, such as lifetime use of aspirin, years of schooling, dietary fiber, waist and hip circumferences, waist-to-hip ratio, total calcium and processed meat, were not included in the final model because they did not materially change the OR estimates or increase model-fit. BMI and the biomarker measurements were categorized into 4 levels according to the quartiles based on all study subjects. Linear trends in ORs were tested using the median values for each quartile.

Interactions between plasma biomarkers and sex, race, BMI (≤25; 25 to ≤30; >30 kg/m²), history of diabetes and age at blood draw (≤53; 53 to 65; >65 years) was assessed with the likelihood ratio test, comparing the likelihood of a main effect model with a model including both main effect and interaction terms. Possible interactions (up to the third order) among important biomarkers were also examined separately among subjects with and without a history of diabetes. All statistical tests were performed with a significance level of 0.05 (2-sided) using SAS (version 9.1).

RESULTS

Table 1 displays the main characteristics of the participants by case-control status. Compared to cases, controls were more educated, smoked less, had lower BMI and smaller waist and hip circumferences, were more likely to have had a previous colorectal endoscopy, and consumed less alcohol and processed meat, and more fiber from vegetables, total folate and total calcium. The distribution of other variables, including aspirin use and daily energy expenditure, was similar between cases and controls.

Table 1.

Main Characteristics^* of Study Participants

	Cases (n = 554)	Controls (n=786)	p‡
Male (%)	64.1	64.9	0.76
Race (%)			0.77
Japanese	34.8	34.5
Caucasian	45.1	46.8
Hawaiians	20.1	18.7
Site and Type of Examination (%)			<0.0001
Kaiser (Colonoscopy)	30.9	22.0
Kaiser (Flexible Sigmoidoscopy)	51.8	51.2
PLCO (Flexible Sigmoidoscopy)	17.3	26.8
Age at blood draw (years)	63 (57–69)	63 (58–69)	0.23
Previous History of Colorectal Endoscopy (%)	32.7	39.2	0.01
Ever Use of Aspirin# (%)	32.1	30.1	0.42
Past History of Diabetes (%)	13.2	10.2	0.09
Education (years)	14 (12–17)	16 (12–17)	0.01
Smoking Status (%)			0.0003
Never smoker	40.7	48.0
Past smoker	45.2	44.2
Current smokers	14.1	7.8
Pack-years†	24 (10–43)	19 (5–37)	0.06
BMI (kg/m2)	26.6 (24.0–30.2)	25.9 (23.4–29.2)	0.002
Waist Circumference (cm)	95.0 (87.1, 104.2)	93.0 (84.5, 102.0)	0.005
Hip Circumference (cm)	104.0 (98.5, 112.9)	103.1 (97.0, 110.3)	0.01
Waist-to-Hip Ratio	0.91 (0.84, 0.96)	0.91 (0.84, 0.95)	0.11
Total Calories (kcal/day)	2,061 (1587–2821)	2,071 (1620–2779)	0.89
Alcohol Consumption (g/day)	2.0 (0.3–17.2)	1.3 (0.2–11.1)	0.02
Daily Energy Expenditure (METs)	57 (49, 66)	57 (50, 66)	0.67
Dietary Fiber from Vegetables (g/day)	15.9 (12.5–19.5)	16.5 (13.2–20.7)	0.004
Total Calcium§ (mg/day)	828 (622–1257)	918 (679–1320)	0.01
Total Folate§ (DFE/day)	769 (499–1194)	933 (567–1291)	<0.0001
Total Processed Meat (g/day)	22.2 (10.9–42.0)	20.1 (8.5–39.2)	0.02

^*Data are medians (interquartile range), except as indicated.

^#Ever use: twice a week for 3 consecutive months or more

^†Pack-years: pack-years of cigarette/cigar/pipe smoking among ever smokers

^‡P-values are from Pearson’s χ² test for percentages and Wilcoxon rank sum test for continuous traits comparing cases and controls.

^§From foods and supplements (adjusted for total calories). DFE=Dietary Folate Equivalents

Table 2 presents the median values of the anthropometric measurements and the blood biomarkers in control subjects only, by sex and race. BMI, waist circumference and plasma C-peptide levels were significantly different across ethnic groups for both men and women (p < 0.01 with or without adjustment for age at blood draw, and postmenopausal estrogen use in women). Unadjusted test also suggested that the distributions of IGFBP-3 levels in men and IGFBP-1 levels in women were different across ethnic groups (both p’s = 0.01). However, these associations were weakened after controlling for age at blood draw in both sexes and further for menopausal estrogen use in women (both p’s = 0.06). Kruskal-Wallis test gave similar p-values to the GLM test in the unadjusted analysis.

Table 2.

Median BMI, Waist Circumference and Biomarker Levels in Control Subjects.

	Men (n=510)				Women (n=276)
	JPN	CAU	NH	p*	JPN	CAU	NH	p*
BMI (kg/m2)	25.8	26.5	28.5	<10−4	23.8	24.9	28.1	<10−4
Waist Circumference (cm)	92	99	101	<10−4	80	86	93	<10−4
C-Peptide (ng/ml)	3.0	3.4	3.5	0.008	2.8	3.0	4.2	0.002
IGF-I (ng/ml)	190	201	189	0.23	147	162	164	0.54
IGFBP-1 (ng/ml)	28.4	20.7	22.5	0.53	37.9	43.7	25.5	0.01#
IGFBP-3 (ng/ml)	3616	4031	3603	0.01#	4213	4420	4281	0.73
IGF-I/IGFBP-3	0.19	0.18	0.20	0.15	0.14	0.14	0.15	0.13

^*p-values for crude differences in biomarker distribution across ethnic/racial groups within each sex.

^#These p-values became non-significant (>0.05) after adjustment for age at blood draw, and post-menopausal estrogen use (men were assigned zero). The significance of other p-values remained unchanged after adjustment.

JPN: Japanese; CAU: Caucasian; NH: Native Hawaiian.

The partial Spearman’s correlation coefficients are shown in Table 3 for the same variables, after adjustment for age at blood draw, sex, ethnicity, case-control status and duration of post-menopausal estrogen use. Plasma C-peptide levels were highly negatively correlated with IGFBP-1 levels (r=−0.58, p<0.0001). BMI was also correlated positively to C-peptide levels (r=0.50, p<0.0001) and negatively to IGFBP-1 levels (r=−0.44, p<0.0001). Plasma IGF-I was also correlated with levels of its binding proteins [IGFBP-3 (r=0.58, p<0.0001) and IGFBP-1 (r=−0.20, p<0.0001)]. The correlation between BMI and IGF-I or IGFBP-3 was not statistically significant. BMI was related positively to waist and hip circumferences, and waist-to-hip ratio and negatively to daily energy expenditure (correlation coefficients 0.85, 0.84, 0.45 and −0.14, respectively) (all p’s<0.0001)

Table 3.

Partial Spearman Rank Correlation Coefficients^* (p-values) (N=1,318)

	BMI	Waist Circumf.	C-Peptide	IGF-I	IGFBP-1
Waist Circumf.	0.85 (<0.0001)
C-Peptide	0.50 (<0.0001)	0.53 (<0.0001)
IGF-I	−0.02 (0.45)	−0.02 (0.59)	0.06 (0.03)
IGFBP-1	−0.44 (<.0001)	−0.44 (<0.0001)	−0.58 (<0.0001)	−0.20 (<0.0001)
IGFBP-3	0.01 (0.78)	0.04 (0.14)	0.12 (<0.0001)	0.58 (<0.0001)	−0.19 (<0.0001)

^*Adjusted for age at blood draw, sex, ethnicity, case-control status, and duration of post-menopausal estrogen use (men were assigned 0).

The ORs of adenoma risk associated with quartiles of BMI and the biomarkers are presented in Table 4, after adjustment for the matching variables and important risk factors (see Methods). Increased plasma levels of C-peptide were associated with a higher risk of adenoma (p _trend = 0.0002), after controlling for BMI and other risk factors. The OR for the highest compared to the lowest quartile was 1.79 (95% CI: 1.23–2.60) and 1.61 (95% CI: 1.07–2.45), before and after adjusting for plasma IGF-I, IGFBP-1 and IGFBP-3 levels, respectively. Higher plasma IGFBP-1 levels were associated with a decreased risk (p_trend < 0.0001). The OR for the 4^th compared to the 1^st quartile was 0.47 (95% CI: 0.32–0.70) and 0.55 (95% CI: 0.36 – 0.85), before and after adjustment for C-peptide, IGF-I and IGFBP-3 levels, respectively. BMI, IGF-I, IGFBP-3 or the ratio IGF-I/IGFBP-3 did not show any significant association with adenoma. No association was observed for waist and hip circumferences and waist-to-hip ratio after adjustment for BMI and other factors.

Table 4.

Odds ratios (95% CIs) for Adenoma Risk According to Quartiles of the Biomarkers and BMI

	Quartiles
	1st	2nd	3rd	4th	p trend
C-Peptide
Ca/Co	117/218	112/224	144/191	181/153
Median (ng/ml)	1.9	2.8	4.1	6.7
OR (95% CI)*	1.0	0.91 (0.65–1.27)	1.21 (0.86–1.71)	1.79 (1.23–2.60)	0.0002
OR (95% CI)#	1.0	0.89 (0.62–1.26)	1.09 (0.75–1.58)	1.61 (1.07–2.45)	0.004
IGF-I
Ca/Co	134/201	151/185	149/186	120/214
Median (ng/ml)	110	162	200	263
OR (95% CI)	1.0	1.22 (0.88–1.69)	1.23 (0.88–1.72)	0.85 (0.60–1.20)	0.29
OR (95% CI)2	1.0	1.21 (0.85–1.72)	1.15 (0.79–1.69)	0.83 (0.54–1.27)	0.26
IGFBP-1
Ca/Co	167/168	152/182	141/193	92/242
Median (ng/ml)	6.6	16.3	31.7	62.6
OR (95% CI)	1.0	0.97 (0.70–1.34)	0.82 (0.58–1.15)	0.47 (0.32–0.70)	<0.0001
OR (95% CI)3	1.0	1.08 (0.77–1.51)	0.96 (0.66–1.38)	0.55 (0.36–0.85)	0.001
IGFBP-3
Ca/Co	142/193	128/205	146/189	136/197
Median (ng/ml)	2637	3641	4419	5309
OR (95% CI)	1.0	0.87 (0.63–1.21)	1.00 (0.71–1.39)	0.79 (0.56–1.12)	0.29
OR (95% CI)4	1.0	0.83 (0.58–1.19)	0.94 (0.65–1.37)	0.78 (0.51–1.19)	0.37
IGF-I/IGFBP-3
Ca/Co	139/194	137/197	147/189	129/204
Median	0.12	0.15	0.19	0.24
OR (95% CI)	1.0	1.11 (0.79–1.54)	1.28 (0.91–1.81)	1.16 (0.81–1.68)	0.39
OR (95% CI)5	1.0	1.09 (0.78–1.54)	1.19 (0.83–1.69)	1.09 (0.75–1.59)	0.65
BMI
Ca/Co	120/217	139/195	137/196	158/178
Median (kg/m2)	22.1	25.0	27.7	32.6
OR (95% CI)	1.0	1.23 (0.86–1.76)	1.11 (0.72–1.71)	1.18 (0.61–2.27)	0.74
OR (95% CI)6	1.0	1.13 (0.78–1.65)	1.03 (0.65–1.63)	1.07 (0.54–2.11)	0.93

^*Adjusted for sex, race, age at blood draw, recruitment clinic, examination type, daily energy expenditure, BMI (where appropriate), pack-years of smoking, duration of post-menopausal estrogen use, and alcohol and total folate intakes, and log of energy intake.

^#Further adjusted for the other biomarkers (IGF-I, IGFBP-1, IGFBP-3, or C-peptide)

Race-specific ORs for plasma biomarkers are shown in Supplementary Table 1. Similar inverse associations were observed for plasma IGFBP-1 with adenoma risk in the three ethnic groups. The direct association with C-peptide and adenoma risk was observed for Japanese and whites but not for Native Hawaiians, possibly due to the relatively smaller sample size in this group. However, the test for interaction with race for C-peptide did not reach statistical significance (p: 0.12).

We also examined interactions between C-peptide and the IGF hormones with sex, race, age at blood draw and history of diabetes on the association with adenoma. No notable interaction was observed in all subjects or after stratification on history of diabetes. Supplementary Table 2 shows the adenoma ORs for the biomarkers after exclusion of subjects with a history of diabetes. The direct association with C-peptide and inverse association with IGFBP-1 remained statistically significant.

DISCUSSION

In this case-control study, we observed a direct association of plasma levels of C-peptide, a marker of insulin secretion, and an inverse association of plasma IGFBP-1 with risk of colorectal adenoma. These associations were independent from each other and seemingly not explained by known risk factors. IGF-I, IGFBP-3, BMI and waist or hip circumferences were not associated with adenoma risk in our data.

Although lifestyle factors associated with elevated insulin secretion have been linked to risks of adenoma and colorectal cancer, and although direct measures of circulating insulin have been associated with colorectal cancer risk (1,3), few studies have assessed the association of hyperinsulinemia and colorectal adenoma. Schoen et al. (4) found a direct association between serum insulin and IGF-I levels and presence of adenoma detected by flexible sigmoidoscopy in the Pittsburgh site of the PLCO trial (202 cases, 256 controls). An association was also observed between C-peptide, but not IGFBP-1, and colorectal adenoma in a case-control study nested in the Nurses’ Health Study (380 case-control pairs) (5). In contrast, no association was found for C-peptide or IGFBP-1 concentrations with adenoma in a case-control study nested in the CLUE II cohort (132 cases, 260 controls) (6).

The associations that we observed with plasma C-peptide and IGFBP-1 were stronger and independent from those with body size and physical activity, suggesting that these biomarkers are not solely surrogate markers for lifestyle and anthropometric characteristics. C-peptide is considered a better marker of insulin secretion than insulin itself because it has a longer half-life and its levels do not fluctuate as much. IGFBP-1 is known to be modulated by insulin levels and can regulate the bioactivity of IGF-I (7). IGF-I circulating levels are mainly dependent on growth hormone secretion and are weakly affected by insulin in well-nourished populations. Thus, this study suggests that insulin secretion and increased biologically available IGF-I may be etiologic factors in the formation of adenoma, as it has been proposed for colorectal cancer. Possible mechanisms include the proliferative and mitogenic/anti-apoptotic effects of insulin and IGF-I (1,7).

Of note in our data is the suggestion of a lack of association between plasma C-peptide and adenoma risk among Native Hawaiians. The native Polynesian populations of Hawaii and New Zealand are known to experience high rates of obesity and Type 2 diabetes but only low-to-moderate rates of colorectal cancer. This is consistent with the recent observation of a lack of association between diabetes history and colorectal cancer risk among Native Hawaiian participants in the Multiethnic Cohort Study^*. In contrast, diabetes was associated with an increased colorectal cancer risk among the other racial/ethnic groups (Whites, Japanese Americans, Latinos and African Americans) in this prospective study. Native Hawaiians tended to have higher C-peptide levels in our study (Table 2). Whether our results are explained by a ceiling effect, above which insulin would not affect colorectal neoplasia, and/or by a more rapid progression to insulin deficiency in Native Hawaiians, or by other mechanisms, will require further study.

A potential limitation of this study is that biomarkers levels were measured after diagnosis of adenoma, and only once. So it is not known whether the levels measured reflect the participants’ past habitual levels. Adenoma patients are not typically asked to change their lifestyle as the result of their diagnosis, suggesting that post-polypectomy levels may reflect past levels. It is also unlikely that colorectal adenomas would affect production or circulating levels of C-peptide and IGF-hormones. Two independent studies measured within-person variation in IGF-I by collecting two separate samples from the same individuals six and eight weeks apart, with correlations of 0.94 and 0.65, respectively (12,13). The coefficient of variation for C-peptide measured for 7 consecutive days and two weeks apart was reported to be 9.3% and 9.7%, respectively (14,15). Thus, a single measurement should be reasonably representative of usual levels. Finally, our study was not adequately powered to allow for meaningful interaction or sub-group (e.g., by anatomical sub-site or size of adenoma) analyses.

In summary, this study provides evidence for an association of insulin and IGFBP-1 with colorectal adenoma and suggests that hyperinsulinemia and IGF hormones may act as etiological factors in colorectal carcinogenesis, as early as during adenoma formation.