Metformin Suppresses Colorectal Aberrant Crypt Foci in a Short-term Clinical Trial

Colorectal cancer (CRC) is a major neoplasm worldwide (1), and its prevalence and mortality have been increasing (2). A paradigm shift from surveillance for early detection of cancer or adenomas (polypectomy) to new preventive strategies, including chemoprevention, is needed to lower the burden of this disease. Several large epidemiologic and/or clinical studies have evaluated the possible effects of more than 200 agents, including fiber, calcium, and nonsteroidal anti-inflammatory drugs including aspirin and selective cyclooxygenase-2 (COX-2) inhibitors, in protecting against CRC development (3). Nonsteroidal anti-inflammatory drugs, especially COX-2 inhibitors alone or in combination, have shown the most promise for CRC risk reduction (4), but reports have revealed an increased risk of serious cardiovascular events associated with COX-2 inhibitor use (5, 6). In light of the adverse cardiovascular effects of otherwise effective COX-2 inhibitors and ineffective clinical results with other agents that had shown promise in this setting, novel drugs that would be both safe and effective are needed for CRC prevention. CRC is associated with lifestyle-related diseases such as diabetes and obesity (7–10), and these conditions might represent new targets for CRC chemoprevention.

Metformin (1,1-dimethylbiguanide hydrochloride) is a biguanide derivative that has been long and widely used for treating diabetes mellitus (11). It decreases basal glucose output by suppressing gluconeogenesis and glycogenolysis in the liver and by increasing glucose uptake in muscle tissue. Because metformin does not directly stimulate insulin secretion, the risk of hypoglycemia associated with its use is lower than that associated with the use of other oral antidiabetes drugs (12). The molecular mechanism of metformin involves liver kinase B1–dependent activation of AMP-activated protein kinase (AMPK; ref. 13). Patients with type 2 diabetes taking metformin might be at a lower risk of cancer (including CRC) compared with those patients who do not take metformin (14, 15). This evidence suggests that metformin might be a candidate agent for the chemoprevention of CRC in diabetic patients.

Previously, we showed the chemopreventive effect of metformin in two rodent models of colorectal carcinogenesis: one a genetic cancer model, the other a chemically induced cancer model. We showed that metformin suppressed the development of intestinal polyps in adenomatous polyposis coli (APC^Min/+) mice, a murine model of familial adenomatous polyposis (16). Furthermore, we showed that metformin suppressed the azoxymethane-induced formation of colorectal aberrant crypt foci (ACF) by activating AMPK (17). Both studies were performed in nondiabetic mice, suggesting the chemopreventive potential of metformin in nondiabetic patients. Little is known, however, about the effect of metformin on human colorectal carcinogenesis.

The end points of CRC chemoprevention trials are, in general, the incidence of polyps or of the cancer itself. These end points usually require several years of evaluation, and therefore such studies can be fraught with compliance issues and a high dropout rate. To overcome these problems, shorter-term surrogate markers for cancer could be used to advantage. ACF are tiny lesions that develop in the earliest stage of colorectal carcinogenesis and consist of large, thick crypts identified by dense methylene blue staining (18–21). ACF have been shown to be precursor lesions of the adenoma-carcinoma sequence in humans (22) and have been suggested as a suitable surrogate end point for CRC chemoprevention trials. Therefore, we conducted a prospective pilot clinical trial to evaluate the chemopreventive effects and safety of metformin against ACF formation in nondiabetic patients. This is the first report of a clinical trial of metformin for inhibiting colorectal carcinogenesis in humans.

Eligible patients had ACF determined by clinical and histologic criteria, did not have diabetes, and had a medical history and physical examination indicating that they were in acceptable health and had no severe systemic diseases that could affect the results of a colonoscopic examination. Major exclusion criteria included the following conditions: liver or renal dysfunction, pregnancy, history of drug allergy, lactic acidosis or colorectal cancer, and current nonsteroidal anti-inflammatory drug use. Eligible participants were randomized to metformin (250 mg/d) or control (untreated) groups. Each participant was treated (metformin) or followed (control/untreated) for 1 month by protocol, and the number of ACF per patient in the two groups was examined by magnifying colonoscopy before and after 1 month of study. The study was conducted in accordance with the Declaration of Helsinki (revised 1989) and with the approval of the Ethics Committee of Yokohama City University School of Medicine. Each participant provided written informed consent for study participation and the use of their data for research purposes.

Bowel preparation for colonoscopy was carried out using polyethylene glycol solution. A Fujinon EC-590ZW/M-H colonoscope was used to perform the magnifying colonoscopy (Fujinon Toshiba ES Systems, Co., Ltd.). Total colonoscopy was performed before imaging of the rectal ACF. Subsequently, 0.25% methylene blue was applied to the mucosa with a spray catheter, as described previously (22). Methylene blue was applied to the lower rectal region extending from the middle Houston valve to the dentate line. Lesions consisting of large, thick crypts in the methylene blue–stained regions of the colon were defined as ACF (19–21). Dysplastic ACF were defined as crypts in which each lumen was compressed or not distinct, with an epithelial lining that was much thicker than that of the normal surrounding crypts (21). Similarly, ACF were classified according to the number of crypts per focus (small, 1 to 9 crypts; medium, 10 to 19 crypts; and large, 20 crypts or more per focus; ref. 22; Fig. 1). All ACF were recorded photographically and evaluated by two independent observers (H. Takahashi and K. Hosono), who were blinded to the clinical history and study group of the patients.

Metformin is widely used as an antidiabetes drug that improves insulin resistance. Therefore, we investigated the effect of metformin on insulin resistance and plasma lipid level by comparing these end points at baseline and after 1 month within the metformin group; the analyses were blinded as to the timing of the sample. The plasma levels of fasting glucose, insulin, cholesterol, and triglyceride were measured using an ELISA kit (Wako Pure Chemical). The degree of insulin resistance, as represented by the homeostasis model assessment of insulin resistance (HOMA-IR), was investigated (23); HOMA-IR was calculated using the following formula: HOMA-IR fasting = plasma immunoreactive insulin (μU/mL) × fasting plasma glucose (mg/dL)/405.

To evaluate the effects of metformin on cell proliferative and apoptotic activities in rectal epithelium, we used immunohistochemical techniques, and the analyses, which compared baseline to 1-month results within the metformin group, were blinded as to the timing of the samples. Proliferative activity was evaluated by staining for the proliferative cell nuclear antigen (PCNA), and apoptotic activity was evaluated by the terminal deoxynucleotidyl transferase–mediated dUTP-biotin nick-end labeling (TUNEL) method. Colonic epithelial samples were obtained from the same trial patients by biopsy before and after metformin treatment. Samples embedded in paraffin were sectioned at 4 μm thickness and mounted on slides coated with polylysine. Sections were deparaffinized in xylene and dehydrated through a series of graded alcohol solutions. Inhibition of endogenous peroxidase activity was performed with 3% hydrogen peroxide. The blocking of nonspecific interaction was accomplished by incubating the samples with a serum block for 30 minutes at room temperature. Sections were then incubated overnight at 4°C with the primary antibody, PCNA mouse monoclonal antibody (no. 2586, 1:4,000 dilution; Cell Signaling Technology). The antibody was applied to the sections and incubated with Histofine simple stain max PO kit for 30 minutes (Nichirei Laboratories), in accordance with the instructions of the manufacturer. The signals were counterstained with hematoxylin and visualized with diaminobenzidine (peroxidase substrate kit; Vector Laboratories). Six random microscopic fields per sample of ∼250 cells were counted at a magnification of ×400 using a bright-field microscope. Results were expressed as the percentage of positively stained cells.

Cells were assessed for apoptosis by staining with a TUNEL staining kit, in accordance with the instructions of the manufacturer (Wako Pure Chemical). The apoptotic index (percentage of cells showing positive TUNEL staining relative to the total number of cells) was determined similarly to PCNA results.

Data were expressed as mean ± SD, unless otherwise indicated. All comparative results involve within-group comparisons of baseline with 1 month data in the metformin or control group (for ACF) or within the metformin group (for laboratory end points). The statistical significance was determined using a paired or unpaired t test using the StatView software (SAS Institute, Inc.). P < 0.05 was considered as denoting statistical significance.

Trial recruitment and conduct took place between July 2008 and May 2010 at the Yokohama City University School of Medicine (UMIN clinical trial registry 000001241). During this period, 1342 patients were screened in our colorectal neoplasia screening clinic, and 629 patients had ACF and were assessed for trial eligibility. ACF occurred in 80.9% (430/531) of patients also found to have adenomas and in 24.5% (199/811) of patients without adenomas (P < 0.001), showing an association of ACF with presence of adenomas. Of the 629 ACF patients, 402 refused to participate and 201 were ineligible (Supplementary Fig. S1). Reasons for exclusion were NSAID use (54.7%), liver dysfunction (cirrhosis, hepatitis, elevated enzymes; 30.8%), renal dysfunction (e.g., elevated creatinine; 10.5%), pregnancy (1.5%), CRC (1.5%), and drug allergy (1.0%). Patients with adenomas removed during screening colonoscopy were eligible. Therefore, we prospectively randomized 26 eligible nondiabetic patients with ACF into the trial—12 into the metformin group and 14 into the control group. In the metformin group, one patient stopped treatment early and did not have a follow-up colonoscopy and two refused follow-up colonoscopy, leaving nine metformin group patients who were included in our end point analyses. Baseline patient characteristics were similar between the two study groups (Table 1), including the mean numbers of ACF per patient (8.78 ± 6.45 metformin and 7.23 ± 6.65 control).

In the metformin group, the mean number of ACF per patient decreased significantly from 8.78 ± 6.45 (baseline) to 5.11 ± 4.99 (at 1 month, P = 0.007). In contrast, there was no significant change in the mean number of ACF per patient in the control group (7.23 ± 6.65 baseline versus 7.56 ± 6.75 at 1 month, P = 0.609; Fig. 2). Regarding size and dysplasia distributions of ACF, the mean numbers of small and dysplastic ACF were reduced by half in the metformin group between baseline and the 1-month follow-up, whereas size and dysplasia distribution did not change significantly in the control group (Fig. 3).

Our HOMA-IR analysis found no significant differences in blood glucose levels or degree of insulin resistance in the metformin group between baseline and 1 month of treatment (Table 2). There were also no significant differences in the metformin group in plasma cholesterol or triglyceride levels between baseline and 1 month of treatment.

To gain insight into the mechanism of suppressive metformin effects on ACF, we investigated the PCNA labeling index of proliferative activity in the rectal epithelium. The PCNA index decreased significantly in the metformin group between baseline and the 1 month follow-up (Fig. 4A and B).

To investigate whether metformin induced apoptosis, we performed TUNEL staining of the rectal epithelium. Metformin did not significantly alter the percentage of apoptotic cells in the metformin group between baseline and at 1 month of treatment (Fig. 4C and D).

Metformin at a low dose of 250 mg/d did not produce any side effects, including lactic acidosis, hypoglycemia, or diarrhea in this 1-month study.

The present study clearly shows that metformin suppressed the formation of human colorectal ACF. Furthermore, metformin (at this study's dose for 1 mo), did not decrease blood glucose or affect insulin resistance, plasma cholesterol, or plasma triglyceride levels of nondiabetic ACF patients, indicating that the ACF suppression effect was direct rather than due to the attenuation of insulin resistance or hyperlipidemia.

We examined the potential direct effects of metformin on ACF via PCNA immunostaining for colorectal cell proliferative activity and TUNEL for apoptosis. The PCNA index decreased significantly following metformin treatment, although the apoptotic index did not change significantly. These data suggest that metformin effects in suppressing cell proliferation mediate suppressive effects on ACF formation.

Metformin activates AMPK, which inhibits the mammalian target of rapamycin (mTOR) pathway (13). The mTOR pathway plays an important role in protein translational machinery and cell proliferation (24). Inhibition of the mTOR pathway might be a direct mechanism of ACF suppression. The best-characterized downstream effector of mTOR is S6 kinase, which regulates the initiation and elongation phases of translation (25). Activation of the mTOR pathway accelerates cell cycle progression from G₁ to S in CRC DLD-1 cells (26). Therefore, AMPK activation may lead to the inhibition of cell growth and proliferation as a result of suppressed protein synthesis and thus might have a potent antiproliferative effect. Recent evidence indicates that metformin has a suppressive effect on tumorigenesis and cancer cell growth (27–29). Metformin activated AMPK and consequently decreased cellular proliferative activity and produced a general decrease in protein synthesis in vitro in human breast carcinoma cells (27). Metformin also inhibited the proliferation of human prostate cancer cells (29). Our group has shown that mTOR inhibition with rapamycin reduced ACF formation in adiponectin-deficient mice under high-fat diet conditions (30).

We previously showed that metformin suppressed intestinal polyp growth in APC^Min/+ mice (16) and colorectal ACF formation in a chemical carcinogen–induced murine model (17). We clarified that the molecular mechanism of ACF suppression was the inhibition of the mTOR pathway by the metformin-mediated activation of AMPK, which consequently led to the suppression of cellular proliferation (17). The present study extended these murine model studies to the human clinical setting, in which metformin also suppressed colorectal ACF formation. The relevance of ACF to the adenoma-carcinoma sequence in humans (22) and thus to CRC prevention is strengthened by our patient data showing that ACF were significantly associated with adenomas in the overall population of screened patients.

Metformin was re-evaluated as an antidiabetes drug with a cardioprotective effect based on evidence showing that the incidence of cardiovascular events was significantly lower in diabetics taking metformin than those taking other diabetes drugs (31). Our results similarly suggest that metformin should be re-evaluated as an antidiabetes drug with a chemopreventive effect against CRC. Our pilot trial, the first reported clinical study of the inhibitory effect of metformin on colorectal carcinogenesis, has shown that metformin, at 250 mg/d for 1 month, safely and directly suppressed both colorectal epithelial proliferation and ACF formation, supporting further, larger-scale studies of metformin in small doses for CRC chemoprevention in diabetic or nondiabetic patients. Determining some important issues, such as the potential optimal dose, schedule, and duration, will be necessary prior to conducting a long-term clinical trial of metformin for preventing colorectal polyps or cancer.

No potential conflicts of interest were disclosed.

We thank Machiko Hiraga for her technical assistance.

Grant Support: A Grant-in-Aid for research on the Third-Term Comprehensive Control Research for Cancer from the Ministry of Health, Labour and Welfare, Japan (A. Nakajima), a grant from the National Institute of Biomedical Innovation (NBIO; A. Nakajima), a grant from the Ministry of Education, Culture, Sports, Science and Technology, Japan (KIBAN-B; A. Nakajima), a grant from the Yokohama Foundation for Advancement of Medical Science (K. Hosono), and the grant program “Collaborative Development of Innovative Seed” from the Japan Science and Technology Agency.

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.