Bisphosphonates Zometa and Fosamax Synergize with Metformin to Prevent AOM-Induced Colon Cancer in F344 Rat Model

Colorectal cancer is one of the most frequently diagnosed cancer and is among the leading causes of cancer-related mortalities worldwide. In the United States, colorectal cancer ranks third in annual cancer incidence cases and second for cancer-associated deaths. An estimated 145,600 new cases of colorectal cancer and 51,020 deaths are expected to occur in 2019 (1). When colorectal cancer is detected early, patients with colorectal cancer have better treatment options with an increased 5-year survival (90%); however, close to two-thirds (61%) of colorectal cancers are at advanced stages when diagnosed and, despite advanced treatment options, patients diagnosed at these stages have a very low survival rate. Early diagnosis and prevention of colorectal cancer are crucial to minimize cancer deaths.

Bisphosphonates (BP; also called diphosphonates) are the analogues of pyrophosphate where oxygen is replaced by a carbon atom with various side chains. BPs have high affinity for calcium hence they accumulate in bone resulting in inhibition of bone resorption. Clinically, BPs are used to treat osteoporosis and osteolytic tumor bone metastases (2). BPs block macrophages to osteoclasts differentiation, activity of mature osteoclasts, and induce osteoclast apoptosis (3–4). While the exact molecular mechanisms behind BPs effects are still not clear, it is noteworthy that polarized macrophages are known to drive the progression of colorectal cancer and it is imperative to understand the role of BPs in colorectal cancer prevention. Mechanistically, the nitrogen-containing BPs such as zometa (zoledronate) and fosamax (alendronate) are known to inhibit the mevalonate pathways while non-nitrogen BPs inhibit ATP-dependent enzymes (3). Recent preclinical data showed that zometa can inhibit cancer cells (5–6) by different mechanisms, and highlighted the potential of BPs as anticancer agents, individually and in combination with other agents (2, 7). Accumulating evidence support that BPs suppress bone metastasis due to antiangiogenic activities and by modulating adhesion and migration of cultured endothelial cells (6).

Metformin is a clinically approved hypoglycemic agent used for treatment of type II diabetes and is well tolerated in patients (8). Epidemiologic studies suggest that patients with diabetes can be benefited by cancer risk reduction effects of metformin (9–10). Metformin exerts its antitumor potential through inhibition of tumor proliferation via AMPK pathway activation, and by targeting cancer stem cells selectively (11–12). Metformin use was found to improve response to chemotherapy (13), with possible improved outcomes in patients with type II diabetic colorectal cancer (12). There is a strong need to identify effective chemopreventive agents for colorectal cancer that can be translated into clinical applications. The above mentioned agents have been tested in various cancers; however, they have not been studied in colorectal cancer. In view of the clinically approved advantages and increasing evidence in favor of the antitumor activity of BPs, we determined the preventive efficacy of the BPs (zometa and fosamax) and metformin, individually or in combination, to prevent azoxymethane-induced colon cancer in a rat model.

All animal studies were conducted in accordance with, and with the approval of Institutional Animal Care and Use Committee. Male F344 rat weanlings (4–6 weeks old) were obtained from Harlan Breeding Laboratories and randomized into control and experimental groups (30 rats/group) based on their weight and provided unrestricted access to food and water. Powdered diet with and without experimental agents was prepared using semipurified ingredients purchased from Bio-Serv as described previously (14). Rats were fed experimental diets through food cups, which were replenished twice a week with fresh diet. For experimental diet preparation, zometa, fosamax, and metformin (Fig. 1A) were premixed with a small amount of diet initially and then blended into bulk diet using a Hobart mixer for uniform distribution.

After randomization, 7 weeks old male F344 rats were fed the modified AIN-76A diet for a week, after which colon carcinogen azoxymethane was administered by subcutaneous injections at a dose of 15 mg/kg body weight once weekly for 2 weeks. Saline was injected into the vehicle group rats. Four weeks after the last azoxymethane injection, animals were fed experimental diets with zometa or fosamax at 0, 20, or 100 ppm; metformin 1,000 ppm; or combinations (zometa/fosamax 20 ppm, + metformin 1,000 ppm) for 40 weeks (Fig. 1B). At this stage, carcinogen-treated rats develop preneoplastic aberrant crypt foci, representing high-risk individual of colorectal cancer. Later, rats were killed by CO₂ asphyxiation and the colons were removed, rinsed in PBS, opened longitudinally, and flattened for colon tumor evaluation as described previously (14). The location (proximal and distal), number, and size of all colon tumors were recorded (Fig. 1C). Colonic tumors were snap-frozen in liquid nitrogen for molecular analysis or saved in 10% buffered formalin for 24 hours, followed by transferring to 80% ethanol for histologic processing.

Whole blood was collected terminally from the control and experimental group rats by cardiac puncture. Serum was prepared from clot blood by centrifugation (10 minutes at 6,000 rpm). The serum was then analyzed using an IDEXX Catalyst instrument to evaluate the drugs' effects on liver function using alanine aminotransferase (ALT) and alkaline phosphatase (ALKP); kidney function using blood urea nitrogen (BUN) and creatinine; and other serum parameters, such as cholesterol, glucose, amylase, lipase, total protein, and albumin, following the manufacturer's instructions.

Colon tumor tissues saved in formalin/ethanol were processed, paraffin embedded, and cut into 4-μm thick sections. For histopathology, the tissue sections were deparaffinized in xylene, rehydrated in alcohol gradient, and stained with hematoxylin and eosin. Stained sections masked for treatment information were analyzed for tumor grade by a pathologist as described previously (14). Formalin-fixed, paraffin-embedded tumor tissue sections were also used for biomarker evaluation using IHC, as described in our previous paper (15). Primary antibodies used were PCNA (ab29; 1:3,000), caspase-3 (cs9662; 1:250), and p21 (sc397; 1:50). Detection was conducted using a IHC Detection Kit (Histostain-Plus, Life Technologies). Primary antibody was replaced with TBS or the respective antibody serum for negative control slide. Digital images of the staining were recorded with an Olympus DP70 camera attached to Olympus microscope IX71. These images were analyzed using IHC profiler following semiautomated analysis protocol (16).

For marker analysis, total cell lysates were prepared from colon tumors from different treatment groups as described previously (14). Proteins were resolved on SDS-PAGE, transferred to a polyvinylidene difluoride membrane, and probed with specific antibodies [PCNA (ab-29; 1:1,000); Cyclin D1 (ab-134175; 1:2,000); p21 (cs-2947; 1:1,000); caspase-9 (ab-52298; 1:1,000); caspase-3 (cs-9662; 1:1,000); PARP (cs-9542; 1:1,000); Actin (cs-4970; 1:1,000)] overnight at 4°C. The membranes were washed with TBS (pH 7.4) and were incubated with anti-rabbit horseradish peroxidase–conjugated secondary antibody (1:5,000, 2.5% skimmed milk in TTBS) and visualized using chemiluminescence reagent, followed by autoradiography as described previously (15). β-actin was used as the loading control.

Colon tumor multiplicity (average number of tumors per rat), was analyzed by unpaired Student t test with Welch Correction. Fisher exact two-tailed test was used to analyze tumor incidence (percent of rats with colon tumors). Differences between body weights were analyzed by ANOVA. The data are presented as means ± SEM. Differences between groups with P < 0.05 were considered significant. GraphPad Prism Software 8.0 (GraphPad Software, Inc.) was used for statistical analysis.

Rats fed experimental diets containing zometa, fosamax, or metformin individually or in combination showed no significant difference in the body weight gain between these groups throughout the study (Supplementary Fig. S1A). The average weights of the organs, such as kidneys, liver, and spleen, were also found to be similar among the treatment groups (Supplementary Fig. S1B–S1D), except for high-dose zometa, for which significant differences were observed in the liver and spleen weights compared with those of controls. At the point of necropsy, no effects of the agents were noted on the gross observations, such as size, shape, color, and texture. Serum from animals (n = 4) was analyzed for liver and kidney function parameters ALKP, ALT, creatinine, and BUN (Supplementary Fig. S2) and various other parameters (Supplementary Table S1). The experimental drugs had no significant effect on these parameters compared with control. Taken together, the gross, histologic, and biochemical data suggested that the chemopreventive agents, alone or in combination, did not cause any observable toxicity.

Administration of azoxymethane led to formation of colon tumors in all control group rats, while saline-treated rats were free from colon tumors (Fig. 1C). To determine the chemopreventive effect of the test agents, colon tumor multiplicity was compared with the control group animals. Rats of the control group had colon adenocarcinoma multiplicity of 4.62 ± 0.52 (Fig. 2A). Rats fed zometa at 20 and 100 ppm had colon tumor multiplicity of 3.76 ± 0.44 (P > 0.05) and 3.70 ± 0.41 (P > 0.05), respectively. Similarly, rats administered fosamax at both doses had colon tumor multiplicity of 3.72 ± 0.43 (P > 0.05) and 3.83 ± 0.44 (P > 0.05), respectively (Fig. 2A). Both BPs showed moderate tumor inhibition (zometa 19%–20% inhibition and fosamax 17%–19% inhibition); with no statistical significance. There was no difference in the inhibitor effects of the two agents and their two tested doses. Metformin alone had a very poor chemopreventive effect, with average colon adenocarcinoma multiplicity of 4.61 ± 0.57 (P > 0.05) in this treatment group (Fig. 2A). Treatment of the azoxymethane-injected rats with zometa, fosamax, or metformin individually had no notable effect on colon tumor incidence, although 3%–7% of the BP-treated rats were free from colon tumors; this effect was not significant (P > 0.05; Fig. 2B).

Although the agents showed marginal tumor inhibition when administered individually, low dose (20 ppm) of either of the BPs (zometa or fosamax) when combined with metformin (1,000 ppm) showed a synergistic chemopreventive effect leading to a significant inhibition of colon tumor multiplicity and incidence. The zometa plus metformin combination showed 48% inhibition of colon adenocarcinoma (2.4 ± 0.28; P < 0.0002), while the fosamax plus metformin combination showed 46% inhibition (2.49 ± 0.32; P < 0.0005) of colon adenocarcinoma when compared with the control group (Fig. 2A). Interestingly, a significant number of rats given a combination of metformin and zometa/fosamax were observed to be free from azoxymethane-induced colon adenocarcinoma, suggesting that the combination could successfully prevent tumor formation. About 23% (P < 0.001) of the rats treated with zometa plus metformin, and 27% (P < 0.005) of the rats treated with fosamax plus metformin were free from colon tumors compared with 100% incidence in control group, suggesting that the two agents synergize, leading to prevention of colon tumors (Fig. 2B).

To further evaluate the effect of zometa, fosamax, and metformin on colon tumor progression, the histopathology grade of the colon tumors from all treatment groups was determined. Total colon tumors of the control diet fed animals 4.62 ± 0.52 (Fig. 2A) were classified into noninvasive (1.53 ± 0.24; 33%) and invasive adenocarcinoma (3.1 ± 0.38; 66%) based on histologic criteria (Fig. 3 and 4). Zometa showed inhibition of the noninvasive adenocarcinoma at both dose (0.96 ± 0.16 at 20 ppm; P < 0.026 and 1.04 ± 0.18 at 100 ppm; P = 0.05; Fig. 3A). The invasive adenocarcinomas were also slightly inhibited at both doses of zometa (2.80 ± 0.36 and 2.66 ± 0.32, respectively; P > 0.05). However, the effect was not significant when compared with controls (Fig. 3B). Fosamax had a stronger chemopreventive effect than did zometa, and there was a dose-dependent inhibitory effect of the noninvasive adenocarcinoma (0.86 ± 0.13, P < 0.009 and 0.60 ± 0.10, P < 0.0004; Fig. 3C). However, similar to zometa, fosamax had a slight and nonsignificant effect on invasive adenocarcinoma multiplicity (2.86 ± 0.33 and 3.23 ± 0.38, respectively; P > 0.05; Fig. 3D). Metformin had no effect on tumor progression when used alone. The combination of zometa and metformin inhibited both noninvasive (0.80 ± 0.15; P < 0.006) and invasive adenocarcinoma (1.60 ± 0.21; P < 0.0005; Fig. 3A and B). Similarly, the fosamax and metformin combination was effective in preventing both types of colon tumor types (0.56 ± 0.10, P < 0.0002 and 1.93 ± 0.24, P < 0.006; Fig. 3C and D). Thus, the results suggest that the BPs are able to inhibit noninvasive adenocarcinoma. However, their combination with metformin may also have a strong inhibitory effect on invasive adenocarcinoma, due to their synergistic effect.

We also analyzed the incidence of both histologic types of colon tumors among the control and treated animals. Except for 100 ppm fosamax, none of the individual treatments of BPs or metformin had a significant inhibitory effect on the incidence of noninvasive adenocarcinoma (20%–27% inhibition) and invasive colon adenocarcinoma (10%–24% inhibition; Fig. 4A–D). Fosamax alone at the 100 ppm dose showed significant inhibition (46% less; P < 0.0005) of only noninvasive adenocarcinoma incidence (Fig. 4C). The combination of metformin with either of the BPs inhibited both types of adenocarcinoma; the effect was more profound with the fosamax combination. The combination of zometa with metformin showed 44% (P < 0.05) and 37% (P < 0.005) inhibition of the noninvasive and invasive adenocarcinoma, respectively (Fig. 4A and B). On the other hand, the fosamax and metformin combination had a much stronger inhibition of these two tumor types, noninvasive adenocarcinoma 60% inhibition (P < 0.0001) and invasive adenocarcinoma 40% inhibition (P < 0.005). Although both BPs were able to inhibit both types of adenocarcinoma, the effect seemed to be more pronounced with fosamax.

Tissue sections from the untreated and treated colon tumors were analyzed to study the expression of some of the proteins that affect tumor growth. There was a correlation between the tumor outcome and biomarker expression. IHC and Western blot analysis showed that the control tumors had a strong expression of proliferation markers, PCNA (Fig. 5 and 6) and Cyclin D1 (Fig. 6). Individual treatment with fosamax, zometa, or metformin did not show any clear inhibitory effect on the expression of these proteins. However, suppression of PCNA and Cyclin D1 expression was seen in the combination treatment tumors with an increased expression of p21 (Cdk1A; Fig. 5 and 6). Proapoptotic markers caspase 3, caspase 9, and Parp-1 were markedly increased in the treatment groups, and particularly in the combination treatment group, while the control group animal tumors showed low expression of these proteins, suggesting an increase in apoptosis upon treatment (Fig. 6).

BPs and metformin are two medications that have been widely used for the past several decades. While metformin is commonly prescribed for type II diabetes, BPs are used extensively for various indications, such as the osteoporosis treatment, its prevention in the high population, and treating patients with bone metastasis of gastrointestinal and various other cancers. It is however unclear whether these drugs will play a role in the primary and secondary prevention of colorectal cancers.

Developing novel agents for cancer chemoprevention takes several decades, and side effects are a significant concern. Hence, the repurposing of approved agents with demonstrated clinical safety is being pursued as an alternative approach to develop agents for cancer chemoprevention (17). In this context, there is a growing interest in both metformin and BPs as chemopreventives, as recent studies indicated that their use is associated with reduced risk for some cancers (9, 18–19). These agents are also the most commonly prescribed agents for the rapidly increasing elderly population, who are also at higher risk for many cancers. Therefore, it is important to use appropriate models to study the effects of these two agents on one of the most commonly diagnosed cancers. Here, we evaluated the chemopreventive effect of BPs (zometa and fosamax), antidiabetic drug metformin, and their combination in a preclinical rat model of colorectal cancer.

In vitro studies showed that BPs reduce proliferation and induce apoptosis of colon cancer cells (20), while in vivo BPs were also found to inhibit colorectal carcinogenesis in an ulcerative colitis experimental model (21–22). In this study, we found a substantial colon tumor inhibition, particularly the noninvasive colon tumors, using BPs alone (Fig. 3). While both BPs showed significant inhibition of the noninvasive adenocarcinoma, they failed to prevent invasive adenocarcinoma (Fig. 3). This finding may suggest that BPs may be more effective in the early stages of tumor development, in which the cells are much simpler in terms of molecular complexity and aggressive nature. Hence, BPs make a good candidates for chemoprevention, where the objective is to treat the at-risk population who may harbor these types of early-stage polyps or tumors.

Although BPs are used to treat late-stage cancers to prevent cancer metastasis to the bone, we did not see any strong effect on the advanced invasive adenocarcinoma. Unfortunately, metformin alone failed to show any colon tumor inhibition at the tested dose, and a higher dose may be required to see the chemopreventive effect. These findings are in agreement with earlier studies showing an inverse correlation with BPs use and colon cancer (18–19). Several meta-analysis studies suggest that the use of oral BPs is associated with reduced risk of colorectal cancer and that this association is directly dependent on the number of prescriptions and duration of use (18–19, 23). In older populations, particularly postmenopausal women, BP intake was associated with a substantial and significant reduction (40%) in the risk of overall colon cancer–related deaths, as well as incidence of the colon cancer (24). Moreover, BPs use has been found to be associated with risk reduction of breast and endometrial cancers in women (25–27).

In this study, we found that although BPs show a moderate colon tumor inhibitory effect, there was a significant inhibition of tumor multiplicity and incidence of the noninvasive and invasive adenocarcinoma when BPs were combined with metformin (Figs. 2 and 3). Thus, our data clearly indicate that BPs given in combination with other agents may synergize to enhance the chemopreventive effects. These findings are consistent with earlier studies that showed similar synergistic/additive effects of the BPs with other targeted agents (28–30).

The antitumor effects of BPs may be attributed to their pharmacologic and diverse molecular effects (31). When administered orally, BPs are known to have poor absorption through the gastrointestinal tract (bioavailability ranges from 0.6% to 1%). Thus, a substantial amount is delivered directly to the colon (32–33). The high concentration of BPs in the colon may have cancer inhibitory effect on the colon cancer cells. Many in vitro studies attribute the antitumor effects of BPs to their ability to inhibit protein prenylation through the inhibition of the mevalonate pathway, which affects cancer cell growth and metastasis. There is substantial evidence that BPs stimulate adaptive and innate immunity (34); inhibit tumor angiogenesis, invasion, and adhesion of tumor cells; and impede overall tumor progression. Although precise molecular mechanism of action for metformin effects on cancer cells is not fully elucidated, studies suggest that it can activate AMPK pathways, resulting in energy metabolism aberration thereby inhibiting cell growth (35). The findings from our work, along with strong epidemiologic and preclinical data, suggest that BPs may serve as potential chemopreventive agents for colon cancers, particularly in the high-risk older population, and warrant further investigation.

No potential conflicts of interest were disclosed.

Conception and design: V. Madka, A. Mohammed, V.E. Steele, C.V. Rao

Development of methodology: V. Madka, A. Mohammed, C.V. Rao

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): V. Madka, G. Kumar, G. Pathuri, Y. Zhang, A. Mohammed, C.V. Rao

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): V. Madka, G. Kumar, G. Pathuri, S. Lightfoot, A. Mohammed, V.E. Steele, C.V. Rao

Writing, review, and/or revision of the manuscript: V. Madka, G. Kumar, A. Mohammed, V.E. Steele, C.V. Rao

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): V. Madka, G. Kumar, S. Lightfoot, V.E. Steele, C.V. Rao

Study supervision: V. Madka, A.S. Asch, A. Mohammed, C.V. Rao

Other (discussion/clinical relevance): A.S. Asch

This study was supported by funding (NCI-N01-CN53300 and NCI R01 CA213987 to C.V. Rao and NCI CCSG P30CA225520) from the NIH. Authors thank Rodent barrier facility for providing assistance with animal studies and to Ms. Kathy Kyler for editorial help.

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.