Serum insulin-like, growth factor binding protein-related protein 1 (IGFBP-rP1) and endometrial cancer risk in Chinese women
Hyperinsulinemia and the metabolic syndrome confer increased risks of endometrial carcinoma. The roles of insulin, and, insulin-like growth factor-binding proteins (IGFBPs) in the etiology of endometrial carcinoma, remain unclear. We recruited 206 patients with endometrial carcinoma and 350 healthy women to a case–control study of fasting insulin and IGFBP-related protein 1 (IGFBP-rP1) in a Chinese tertiary centre. Patients with endometrial carcinoma had higher insulin concentrations (14.8 ± 16.7 vs. 8.1 ± 9.4 μU/mL; p < 0.001) and lower IGFBP-rP1 levels (17.5 ± 17.2 vs. 22.4 ± 22.8 μg/L; p = 0.018) than controls. High insulin and IGFBP-rP1 levels were both positively and negatively associated with endometrial cancer (odds ratio for the highest tertile versus the lowest tertile: insulin: 4.11; 95% CI = 2.61–6.47; IGFBP-rP1: 0.38; 95% CI = 0.24–0.60). Logistic regression analysis confirmed the associations between endometrial carcinoma and fasting insulin or IGFBP-rP1 after adjustments for age, BMI, serum glucose, cholesterol, triglycerides and high-density lipoprotein cholesterol (odds ratio for the highest tertile versus the lowest tertile: insulin: 2.13; 95% CI = 1.30–3.49; IGFBP-rP1: 0.57; 95% CI = 0.34–0.94). Hyperinsulinemia and high IGFBP-rP1 levels confer altered risks for endometrial carcinoma.
Endometrial carcinoma is the second most common cancer in the female genital tract. Its incidence in China has increased abruptly in the past two decades.1 There is a close relationship between the metabolic syndrome (obesity, diabetes mellitus, low glucose tolerance, dyslipidemia and hypertension) and endometrial carcinoma.2–4 In recent times, this link has been largely explained by raised serum estrogen levels in obese women,5, 6 though other endocrine factors may play a role in endometrial cancer. Insulin resistance (hyperinsulinemia), the hallmark of the metabolic syndrome, is associated with colorectal, breast and pancreatic cancers.7–10 Few studies11–15 have investigated the link between hyperinsulinemia and endometrial cancer.
Insulin-like, growth factor binding proteins (IGFBPs) are a family of homogenous proteins that decrease levels (bioavailability) of free insulin and insulin-like growth factors (IGFs) by binding with insulin and IGFs.16 IGFBPs divide into two categories17; those having a higher affinity for IGFs (IGFBP1–6), and those for insulin (IGFBP7–15, that have been reclassified as IGF binding protein-related proteins, IGFBP-rPs). IGF binding protein-related protein 1 (IGFBP-rP1) or IGFBP7 is the most important IGFBP-rPs. It has a 500-fold higher affinity for insulin than IGFBP1–6 and can substantially lower the serum level of free insulin. A previous study showed high circulating IGFBP-rP1 to be associated with altered insulin sensitivity, and, circulating IGFBP-rP1 was higher in previously undiagnosed type 2 diabetics patients compared with both known type 2 diabetics patients receiving pharmacological therapy, and nondiabetic men.18 IGFBP-rP1 may be a useful marker for insulin resistance. However, most studies12–14, 19, 20 have focused on the association between IGFBP1–3 and endometrial cancer risk; associations between serum IGFBP-rP1 and endometrial cancer remain unknown. This report describes a case–control study investigating the relationships among serum IGFBP-rP1, fasting insulin, various metabolic factors and the risk of endometrial carcinoma in a Chinese tertiary centre.
Material and Methods
We investigated 206 patients (mean age: 53.2 years, range: 26–81 years; mean parity: 2.3 children, range: 0–6 children) with endometrial cancer, confirmed histopathologically between August 2008 and December 2010, from the Affiliated Women's Hospital, School of Medicine, Zhejiang University, China. Women (including patients and controls) with a history of hysterectomy, bilateral oophorectomy or Lynch syndrome were excluded from the study. Patients with malignant mixed Müllerian tumor were also excluded. The majority of patients (193/206) underwent a staging surgical procedure (FIGO Stage I 179/193, Stage II 12/193 and Stage III 2/193). Two, experienced histopathologists evaluated archived slides stained with hematoxylin–eosin. The histopathological diagnosis (n = 206) included 191 endometrioid-type, 5 mucinous, 5 serous, 3 clear cell and 2 mixed type tumors.
Controls (n = 350; mean age: 53.5 years, range: 28–83 years; mean parity: 2.1, range: 0–5 children) were randomly selected from lists of healthy women undergoing routine physical examinations during the same period.
We reviewed the hospital medical records according to an institutionally approved protocol. Body weight and height, diastolic and systolic blood pressure, menopause, reproductive history, previous hormone replacement therapy use, personal and family cancer history, smoking, alcohol consumption, primary hypertension and diabetes mellitus were recorded. Body mass index (BMI) was calculated as mass divided by height squared (kg/m2). BMI categories were defined as normal (18.5–24.9 kg/m2), overweight (25.0–29.9 kg/m2) and obese (≥30.0 kg/m2).21 We grouped overweight and obesity together because of the relatively small number of obese women in our population.
Preoperative fasting total serum cholesterol (TC), triglycerides (TG), high-density lipoprotein cholesterol (HDL-C) and glucose (GLU) were also obtained from the medical records. The upper normal reference values for fasting blood glucose, cholesterol and triglycerides are 6.11 mmol/L, 5.55 mmol/L and 1.71 mmol/L, respectively, while the lower normal reference value for HDL-C is 1.10 mmol/L.22 Patients with values outside the reference range for TC, TG, GLU and HDL-C were recorded as abnormal. Patients with diabetes mellitus were automatically categorized as abnormal. The missing data included body weight or height in eight cases, previously hormone replacement therapy use in one case, TC, TG and HDL in two cases and six controls, and GLU in one case and three controls.
Blood samples were drawn from patients and controls before surgery or any cancer therapy. Women fasted overnight (at least for 8 hr). Serum was obtained and centrifuged at 4°C in the blood collection tube containing a specific separation gel for biochemical tests (BD Biosciences, San Jose, CA). The serum was kept in ∼80°C freezer until use.
Assays of serum insulin and IGFBP-rP1 were performed blind to the case–control status of the samples. Each sample was examined in duplicate with the average value as the final result.
Serum IGFBP-rP1 concentrations were analyzed by commercial, double-antibody, sandwich ELISA kit (R&D Systems, Minneapolis, MN) as per the manufacturer's procedures. Concentrations were calculated according to the standard concentration and the corresponding optical density value at 450 nm. The assay range for IGFBP-rP1 was 3–90 μg/L. The intra- and interassay coefficients of variation were 5.2 and 10.8%, respectively.
Serum insulin was measured by a chemiluminescent, microparticle immunoassay (the ARCHITECT Insulin Assay, Abbott Laboratory, IL). The assay was automatically accomplished by the ARCHITECT i system. The calibration range is 0–300 μU/mL. The intra- and interassay coefficients of variation were 4.7 and 8.9%, respectively.
SPSS 13.0 (SPSS, Chicago, IL) was used for statistical analysis. We used the Spearman's correlations to analyze correlations between different measures and one-way ANOVA to detect the descriptive results between cases and controls. Subjects were divided into two groups as normal/abnormal or presence/absence according to various parameters. Serum levels of insulin and IGFBP-rP1 were further categorized into tertiles based on the values in cases and controls for univariate and multivariate analysis. Missing data were removed in the statistical analysis. Odds ratio (OR) and its 95% confidence interval (CI) were calculated to estimate the relative risks for endometrial cancer. The relative cancer risks of serologic factors were measured as women in tertiles 2 or 3 (T2, T3) when the first tertile 1 (T1) was taken as the baseline. The threshold was determined at 0.05 (two-tailed). Subgroup analysis stratified by menopausal status, reproductive history (as continuous variables) or hormone replacement therapy use was conducted to determine whether these factors influenced cancer risks of insulin and IGFBP-rP1. An unconditional logistic regression model was used to estimate the ORs of serum insulin and IGFBP-rP1, adjusted by age at 5-year intervals, and different metabolic variables including BMI, GLU, TC, TG, HDL-C, insulin (for IGFBP-rP1) and IGFBP-rP1 (for insulin). The variables would be removed from the final models if they did not change risk estimates with more than 10% (removal probability = 0.10).
Table 1 shows the clinicopathological variables and serum levels of insulin and IGFBP-rP1 between cases and controls. Patients with endometrial cancers had significantly higher frequencies of a history of diabetes mellitus, nulliparity, higher BMI, higher levels of GLU, TG, TCH and serum insulin, and lower serum IGFBP-rP1 than controls (p < 0.05). The age at menopause, pre- or postmenopausal, hormone replacement therapy and history of primary hypertension did not differ significantly between cases and controls (p > 0.05). Correlations between the measured serological factors and BMI in healthy controls are presented in Table 2. Briefly, fasting insulin level was positively correlated with BMI, serum GLU, TG and TCH. IGFBP-rP1 was not significantly associated with insulin and other metabolic variables.
Table 1. Clinicopathological features and distribution of serum insulin and IGFBP-rP1 in cases and controls1
Table 2. Spearman's correlation coefficients of the associations between BMI (kg/m2) and measured serologic factors in healthy controls
There was an increased cancer risk in women with diabetes versus women without diabetes (OR = 2.30, 95% CI = 1.12–4.73, p = 0.018), in women with obesity or overweight versus those with normal BMI (OR = 2.31, 95% CI = 1.59–3.35, p < 0.001), and in women with nulliparity versus multiparity (OR = 4.45, 95% CI = 2.15–9.22, p < 0.001). Women aged >50 years, or with primary hypertension or a history of hormone replace therapy did not show significantly heterogeneous cancer risks compared with baseline (p > 0.05). We found that increased cancer risk was found in women with high levels of GLU, TG and TCH and decreased cancer risks in women with high serum HDL-C (data not shown).
There was a dose-dependent response between insulin or IGFBP-rP1 concentration and endometrial cancer risk (Table 3). We detected no influence on these results when stratified by menopause, reproductive history or hormone replacement therapy use. To eliminate the influence of potential confounders, we further applied a multiple logistic regression model to calculate relative risks for endometrial cancer of serum insulin and IGFBP-rP1 in T2 or T3 versus T1. The logistic regression model-derived ORs and 95% CIs are given in Table 3. Serum insulin or IGFBP-rP1 seemed to be an independent risk or protective factor for endometrial cancer, respectively.
Table 3. Unadjusted and multiple logistic regression model-derived ORs of serum insulin and IGFBP-rP1 for endometrial carcinoma1
In this study, we found that women with increased IGFBP-rP1 levels had decreased risks of endometrial cancer, and those with hyperinsulinemia had an increased risk of endometrial cancer. These associations persisted after adjustments for weight, hyperglycemia, dyslipidemia and other related variables. Our results suggest that dysfunction of the insulin-IGFBPs system may influence the development of endometrial cancer.
Contemporary epidemiological and experimental data focus on the well-established, “unopposed estrogen” theory. Excess estrogen is associated with aromatization of androstenedione to estrogen in adipose tissue, and increased free estrogen with decreased serum levels of sex hormone-binding globulin.6, 23 The increased endometrial cancer risk in women with type 2 diabetes, obesity or polycystic ovarian syndrome3, 24, 25 suggests a possible association between hyperinsulinemia and endometrial cancer. Our observations support this association. Furthermore, circulating insulin seems to be an independent risk factor for endometrial cancer. We found five (three prospective and two case–control) studies that studied the association between serum insulin and endometrial carcinoma.11–15 Our study is consistent with recent prospective studies that support the hyperinsulinemia-endometrial cancer hypothesis.11–13 Insulin has mitogenic and antiapoptotic functions in cancer cells.26, 27 Insulin can also increase free estradiol by decreasing serum sex hormone-binding globulin synthesis.28 Both roles may contribute to the relationship between hyperinsulinemia and endometrial carcinoma.
In keeping with a previous study,13 we also found a dose-dependent relationship between insulin and endometrial cancer that remained significant after adjustment for BMI. This observation suggests that the positive association between serum insulin levels and endometrial cancer may only be partially explained by the metabolic syndrome. Alternatively, the current metabolic parameters including BMI and fasting glucose, may be less reliable. For example, waist–height ratio, which reflects visceral adiposity, might be better than BMI in the assessment of insulin resistance in Chinese adults.29 As a marker of energy balance, insulin may be more helpful than other anthropometric measures.
Several studies have investigated the relationship between IGFBPs and endometrial carcinoma. To date, all the studies focused on the association of the first IGFBP catalogue (IGFBP1–6) and cancer, with variable and inconsistent results.11–14, 19, 20 One explanation of these results is that these IGFBPs primarily regulate the bioavailability of serum IGF-I. IGF-I is not a reliable marker for insulin resistance, and its relationship with endometrial cancer risk remains inconclusive. IGFBPs in the second catalogue (i.e., IGFBP-rPs) show much higher binding affinity for insulin.17, 18 For example, IGFBP-rP1 has a 500-fold higher affinity with insulin than that with IGF1. Increased serum concentrations of IGFBP-rP1 are associated with insulin resistance by decreasing levels of free insulin.18 Theoretically, IGFBP-rPs should perform better than IGFBPs in the evaluation of the relationship between hyperinsulinemia and cancer. Our study supports this hypothesis since we observed a dose-dependent, inverse relationship between serum IGFBP-rP1 level and endometrial carcinoma. To our knowledge, this is the first study to demonstrate the relationships among serum IGFBP-rP1, insulin resistance and endometrial cancer risk.
We found no correlation between serum IGFBP-rP1 and insulin. Only one study previously investigated the correlation between fasting IGFBP-rP1 and insulin.18 That study also showed no direct correlation between serum IGFBP-rP1 and fasting insulin in the nondiabetic men. This paradoxical observation may indicate that total serum insulin does not precisely reflect free (functional) insulin. Alternatively, circulating IGFBP-rP1 is a serum insulin-independent, protective factor against endometrial cancer. In our opinion, there are two possible interpretations. First, IGFBP-rP1 is a protein secreted by smooth muscle cells, endothelial cells and fibroblasts. It is a potential tumor suppressor gene that inhibits cell proliferation, induces cell apoptosis and differentiation and promotes cell senescence.30–32 It inhibits BRAF-MEK-ERK signaling and regulates p53 response instead of insulin and IGF-1 activity. Second, endometrial glands and stromal cells in the late secretory phase express IGFBP-rP1 150-fold higher than in the proliferative phase.33 IGFBP-rP1 is essential for decidualization and implantation. This expression pattern implies a potential relationship between IGFBP-rP1 and progesterone. IGFBP-rP1 can also directly suppress estrogen synthesis in granulosa cells.34
Our study has two important advantages. First, we used fasting blood samples in (preoperative) cases and controls, and direct measurement of fasting insulin. Both strategies reflect the basic levels of serum insulin more precisely than C-peptide measurement in nonfasting plasma as in previous studies.13 Second, we obtained relevant clinical data for each sample including BMI, reproductive history, histopathological diagnosis, use of hormone replacement therapy, careful clinical screening of controls, serum levels of GLU, TG, etc. Assessment of these factors permits a detailed analysis of the co-effects of metabolic factors on potential associations between endometrial cancer, and insulin or IGFBP-rP1. However, our study also has several limitations. First, it is a retrospective study with a relatively small sample size, and the potential for bias. Second, we did not assess whether the intervention on metabolic abnormalities may influence the serum insulin or IGFBP-rP1 and related cancer risks. Our previous study indicated that effective control of type 2 diabetes mellitus may decrease the incidence of endometrial cancer.3 Finally, some potentially sensitive and reliable anthropometric factors such as waist–hip ratio, serum free insulin levels and IGFBP-rP1 may enable more precise estimation of risk.
In summary, we found that hyperinsulinemia was a risk factor for endometrial cancer while elevated IGFBP-rP1 was associated with decreased endometrial cancer risk. These results suggest that a disordered insulin/IGFBP system is involved in the pathogenesis of endometrial carcinoma. Further studies to elucidate the pathophysiological mechanisms of insulin and IGFBP-rP1 in the development of endometrial cancer will be helpful. Studies to determine whether insulin and IGFBP-rP1 might be useful in the prediction and monitoring of women at high risk of endometrial cancer, may also be helpful.
We thank Dr. Xiangyang Lou (Department of Biostatistics, University of Alabama at Birmingham, USA) for his kind help with the English language and statistics for the manuscript.