There has been increasing interest in the use of metformin and vitamin D to reduce cancer risk and improve outcomes. Metformin, an oral antidiabetic drug, improves insulin resistance and has been associated with reduced cancer incidence and cancer mortality. Low levels of vitamin D have also been associated with increased cancer risk, mainly in retrospective studies, and it has been suggested that vitamin D supplementation might play a role in cancer prevention. Preclinical data provide a biologic rationale for these associations; however, the human data arise predominantly from observational studies and caution is needed in their translation into clinical practice. This is because of the recognized limitations of observational studies, such as time-related survival biases, selection and referral biases, short follow-up, and the presence of confounding factors that can lead to spurious or inaccurate findings. Combined examination of associations with cancer risk and outcome (occurring when exposure in a population that does not yet have cancer is analyzed in relation to cancer death) may yield results that are difficult to interpret. Finally, associations across all cancers may differ from those in specific cancer types. These shortcomings can be overcome in properly designed and adequately powered prospective randomized trials; however, such trials are both expensive and time consuming. We review the literature examining the associations of metformin and vitamin D with cancer, discussing weakness and strengths and making recommendations for further research and clinical practice.
There is considerable interest in repurposing existing agents as anticancer drugs; the benefits include known safety/toxicity profiles and, in most cases, low cost relative to newly developed drugs. Potential agents include metformin, aspirin, beta-blockers, and a range of vitamins, such as vitamin D. Evidence of benefit often arises initially from observational clinical or epidemiologic studies reporting associations of agent use with lower cancer risk or mortality and/or better cancer outcomes. These observations may be bolstered by findings from preclinical research that provide a plausible mechanistic basis for a beneficial anticancer effect, leading to hype that the agent in question may produce beneficial effects and suggesting that physicians should consider using these agents in the clinical setting.
Unfortunately, this hype often does not translate into clinically relevant benefit for a number of reasons. Observational clinical and epidemiologic studies examine associations, but they do not provide evidence of causation (i.e., proof that the agent produces benefit, typically obtained from randomized trials). This is because observational studies are subject to bias and confounding that cannot be fully controlled in statistical analyses.
Common limitations of observational studies include baseline imbalances in those exposed (vs. not exposed) with respect to factors that may influence cancer risk or outcomes (e.g., smoking status, body size, or concurrent medical conditions that may be associated with cancer). For example, patients with diabetes who are receiving metformin may be less likely to be obese or to smoke than those who do not receive the drug and these pre-existing characteristics, rather than metformin exposure, may result in lower cancer risk. Similarly, healthy individuals may eat a diet containing greater amounts of vitamin D, they may be exposed to more sunlight during outdoor exercise, and they may be more likely to take vitamin supplements; these factors, rather than vitamin D exposure, may result in a lower cancer risk. Other aspects of study design can also introduce bias. Comparing ever users to never users of an agent (instead of incorporating a time-dependent exposure into the analysis) can result in guarantee time/survival biases that occur because those who live longer are more likely to use the agent at some point in their illness trajectory, leading to the false conclusion that it was use of the agent that caused the better outcome rather than the converse.
Supporting preclinical evidence may also prove to be misleading. High concentrations of agents that are not attainable clinically may lead to anticancer effects in the laboratory that are not relevant in patients. For this and other reasons, mechanisms of drug action identified in the laboratory may not prove to be clinically relevant.1 In vivo investigation using clinically relevant concentrations of agents and the use of models that mirror the clinical situation may provide more useful information; however, the relative importance of preclinical findings in the complexity of the clinical situation may remain unclear. Investigation of predictors of benefit in randomized trials and conducting mechanistic studies in the clinical setting (e.g., neoadjuvant window of opportunity studies) may help to identify those mechanisms that are most clinically relevant.
In the sections below, we briefly review available evidence regarding the association of two agents, metformin and vitamin D, with cancer, highlighting the limitations of observational evidence and proposing research that would help to resolve uncertainty regarding their clinical benefit. Because of the large number of observational studies that have been reported, we have chosen to focus on high-quality meta-analyses whenever possible. Some data exist from randomized trials of vitamin D supplementation; these data will also be reviewed. Such trials remain the optimal design to establish benefit; however, because they are costly and time consuming their conduct should be limited to situations where evidence is strongest.
METFORMIN AND CANCER
Metformin, an oral drug widely used to treat type 2 diabetes, improves glucose metabolism and reduces insulin levels. In recent years, there has been increasing interest in metformin as an anticancer agent. In the preclinical setting, metformin may exert direct (insulin-independent) and indirect (insulin-dependent) anticancer effects.1 The ability of metformin to reduce glucose and insulin levels through inhibition of hepatic gluconeogenesis is the basis for a potential indirect effect of metformin.1 The direct effect is primarily related to activation of adenosine monophosphate kinase (AMPK) (via inhibition of Complex I of the respiratory chain in the mitochondrium), leading to inhibition of mammalian target of rapamycin and reduced protein synthesis and cell proliferation, although other mechanisms have also been proposed. Preclinical observations of an anticancer effect, coupled with the observational data reviewed below, have created considerable hype that metformin will be an effective anticancer agent.
Recent meta-analyses/systematic reviews have examined the overall incidence of cancer in patients with diabetes taking metformin (vs. not), combining results of more than 200 studies (Table 1). Many of the included studies were retrospective, comparing cancer risk in those who had ever used metformin to that in those who had never used metformin. These studies have suggested that patients with diabetes who are treated with metformin have an approximately one-third reduction in the overall incidence of cancer, with a risk ratio (RR) of 0.67 (95% CI, 0.53 to 0.85).2 Metformin has been associated with reduced incidence of colorectal, pancreatic, hepatocellular, and breast cancer.3-7 Additional studies have reported lower cancer-specific mortality and overall mortality in patients with diabetes who are prescribed metformin compared with those who are not (hazard ratio [HR] 0.66, 95% CI, 0.55 to 0.79; HR 0.62, 95% CI, 0.46 to 0.84, respectively).2
Recent Meta-Analysis/Systematic Review Studies of Metformin and Cancer
Some of the strongest evidence relates to breast cancer. Chlebowski et al. have conducted one of the best-designed studies using the Women's Health Initiative dataset.8 They prospectively examined the incidence of breast cancer in women with diabetes by conducting a time-dependent analysis (avoiding a survival bias) and found a lower breast cancer incidence when metformin was used (vs. not used) to treat diabetes (HR 0.75, 95% CI, 0.57 to 0.99), an association that was independent of estrogen receptor, progesterone receptor, and HER2 status.8 Because their research was prospective and the use of metformin was analyzed in a time-dependent manner, these investigators overcame several of the design flaws inherent in earlier observational studies; however, because the use of metformin was not randomly allocated the risk of selection biases persisted. Considering women already diagnosed with breast cancer, in a retrospective neoadjuvant study patients with diabetes receiving metformin were reported to have a significantly higher pathologic complete response rate (pCR) to neoadjuvant chemotherapy compared with those receiving other antidiabetic drugs or with nondiabetic patients (pCR 24% in the metformin group, 8.0% in the no-metformin group, and 16% in the nondiabetic group; p = 0.02).9 Recent neoadjuvant window of opportunity studies have identified changes in markers of proliferation and/or apoptosis after metformin administration. These findings lend credence to the hypothesis that metformin may improve breast cancer outcomes, which is being tested in a phase III adjuvant trial.
Interestingly, a meta-analysis from Noto et al. suggested that metformin use was associated with a lower likelihood of developing lung cancer in patients with diabetes who received metformin (RR 0.67, 95% CI, 0.45 to 0.99).2 In preclinical data, the use of metformin in a nondiabetic mouse model prevented tobacco carcinogen-induced lung tumorigenesis.10 Recently, Quinn et al. suggested systemic inhibition of circulating growth factors and local receptor tyrosine kinases signaling as AMPK-independent mechanisms of action of metformin in preventing lung tumorigenesis.10 These observations raise important questions regarding the etiology and possible prevention of lung cancer.
Metformin has also been associated with lower overall cancer-related mortality (Table 1).2,11 Because most of the studies investigating this association examined mortality in populations who were cancer-free at study entry (thereby combining associations with cancer incidence and mortality), it is not clear whether the lower risk of death reflects lower cancer incidence or better outcomes after cancer diagnosis. Yin et al. directly examined associations between metformin use and outcomes in patients with both diabetes and cancer; they reported significantly lower cancer-specific and all-cause mortality in those receiving metformin (p < 0.001).12 Some, but not all, studies reported lower risk of death from pancreatic, colorectal, breast, lung, and prostate cancer in patients with diabetes receiving metformin. Of interest, Franciosi et al. identified a lower risk of cancer-related death in metformin users when data from observational studies were analyzed (odds ratio [OR] 0.65, 95% CI, 0.53 to 0.80) but not when data from randomized clinical trials (RCTs) were analyzed (RR 0.91, 95% CI, 0.70 to 1.18). 13 The latter was also reported by Stevens.14 It should be noted that follow-up was very short in many of the RCTs (only five had follow-up longer than 2.5 years) and it is possible that associations would have emerged with longer follow-up. Nonetheless, these results introduce a note of caution into the hype surrounding metformin.
Many of these studies of metformin and cancer used retrospective designs. They were often conducted using hospital-based rather than population-based registries, introducing the possibility of selection and referral biases as discussed above. Furthermore, the use of metformin (vs. not) was not randomized and characteristics (e.g., age, smoking status, obesity) of those who received metformin often differed from those who did not, thus it is possible that baseline cancer risk differed in those who received metformin and that the associations identified reflected those differences rather than metformin use. The short follow-up in many studies, the lack of standard dose of metformin, and the possibility of exposure biases and guarantee time/survival biases (discussed above) raise additional concerns. Furthermore, the heterogeneity of the treatments received in the comparator arm (e.g., insulin or sulfonylureas, which may be associated with an increased cancer risk) raises concerns that associations of metformin may have been over- or underestimated.
Although the findings of these observational studies are provocative, caution is needed in their translation into clinical practice. We do not recommend use of metformin as an anticancer agent (in the prevention or treatment setting) until more convincing information becomes available. There are at least three ongoing phase III randomized trials involving metformin that should provide more definitive information regarding anticancer effects of metformin. In the MA.32 trial, 3,649 patients with early breast cancer who were receiving standard therapy were randomized to receive metformin or placebo for 5 years. The primary outcome is invasive disease-free survival; results should be available in the next 3 years. The MAST study is currently recruiting patients with low-risk prostate cancer to test whether metformin prolongs time-to-progression. Finally, a chemoprevention trial of metformin to lower endometrial cancer risk is currently recruiting patients, focusing on those who are overweight and have high insulin levels.
VITAMIN D AND CANCER
The association of vitamin D with cancer is a highly polarized area that has received enormous attention in medical circles as well as in the lay press. Vitamin D (calcitriol) is normally absorbed from the gut but it is also produced in the skin in the presence of sunlight (there is a feedback loop that prevents overproduction). Initial interest in vitamin D and cancer came from ecologic studies that reported higher cancer risk in countries distant from the equator, where there is less sun exposure.15 Low sun exposure (ultraviolet-B radiation), low vitamin D intake, and low blood levels of 25-hydroxyvitamin D (25-OHD) have been linked to a higher incidence of cancer in some, but not all, ecologic and epidemiological studies.16 Preclinical studies support a link between vitamin D and cancer. Vitamin D binds to vitamin D receptor elements to alter expression of a broad range of genes, many of which have been implicated in cancer prevention and outcome by modulating cellular growth and development.17 The potential presence of an association is based largely on observational research; fortunately, data are now beginning to emerge from RCTs.
The Institute of Medicine has recently reviewed the literature on vitamin D and cancer and concluded that there is insufficient evidence to confirm an association between vitamin D and cancer.18 Here, we focus on research into the association of blood levels of vitamin D with cancer risk and prognosis and on RCTs examining the potential contribution of vitamin D supplementation to cancer risk and prognosis, rather than on less direct research linking sun exposure or geography to cancer.
Results of meta-analyses and systematic reviews have suggested that serum 25-OHD levels are inversely associated with cancer risk. To a large extent, investigators have focused on individual types of cancer. For example, Chung et al. reported that each 10 nmol/L increase in serum 25-OHD levels was associated with a 6% lower risk of colorectal cancer.19 Liu et al. reported no association between 25-OHD levels and risk of pancreatic cancer in a nine-study meta-analysis.20 For breast cancer, a meta-analysis of prospective studies reported that a 5 ng/mL increase in 25-OHD was linked to lower risk of postmenopausal (but not premenopausal) breast cancer with a RR of 0.88 per 5 ng/mL (95% CI, 0.79 to 0.97).21 However, Amir et al. did not find a significant association of 25-OHD levels with breast cancer risk, (OR for upper versus lower quartile 0.77, 95% CI, 0.55 to 1.06, p = 0.11).22
Stronger evidence is available from four randomized trials that have evaluated the effect of vitamin D supplementation on risk of bone fracture and mortality (Table 2); they examined risk of cancer as a secondary outcome.23-26 The United Kingdom trial compared vitamin D supplementation to placebo; no cancer risk reduction was observed.23 In the Nebraska trial, patients were randomly assigned to receive calcium and vitamin D, calcium alone, or placebo; there was no direct comparison of vitamin D alone versus placebo.24 There were few cancers overall; however, those randomized to calcium/vitamin D supplementation (vs. placebo) had a lower overall cancer incidence.24 The larger Women's Health Initiative trial did not show a reduction in risk of colorectal, breast, or any cancer in those randomized to vitamin D.25 Finally, in the Record trial, vitamin D versus placebo administration was not associated with a reduction in cancer risk.26 These largely negative results may reflect inadequate vitamin D dosing (Table 2), small sample sizes, or the true absence of an association. The promising findings of the Nebraska study may reflect the combined use of calcium and vitamin D, rather than vitamin D alone.
Randomized Trials of Vitamin D Supplementation and Cancer Incidence
Recently, interest has focused on the potential role of vitamin D receptor (VDR) gene polymorphisms and risk of cancer, including prostate, colorectal, and breast cancer. In colorectal cancer, for example, Bsml VDR polymorphism was linked to a higher risk of cancer.27 However, on the whole, the association of VDR gene polymorphisms with cancer risk is inconsistent and further research is necessary.
A recent meta-analysis of 42 RCTs reported an inverse correlation between vitamin D supplementation and all-cause mortality; cancer specific mortality was not examined.28 For established breast cancer, our group reported a recent meta-analysis of eight studies and found that low vitamin D levels were linked to higher risk of recurrence and death (HR 2.13, 95% CI, 1.64 to 2.78 and HR 1.74, 95% CI, 1.35 to 2.30, respectively); however, we cautioned that bias and confounding factors inherent in the included studies could have accounted for the observed association.29
Three of the RCTs discussed above (United Kingdom, WHI, and Record) also evaluated cancer outcomes (Table 2). No significant difference in cancer death between the vitamin D versus non vitamin D supplementation arms was identified in these three studies (HR 0.86, 95% CI, 0.61 to 1.20; HR 0.90, 95% CI, 0.77 to 1.05; and HR 0.85, 95% CI, 0.68 to 1.06, respectively).23,25,26
The role of vitamin D in cancer is unproven. The results of observational studies are subject to bias and confounding. Furthermore, in trials of supplementation, the dose and duration of vitamin D administration and the effect of supplement administration on blood levels may have been inadequate to identify clinically relevant effects. We do not believe that the available evidence is sufficiently robust to recommend vitamin D supplementation as a means of lowering cancer risk or improving cancer outcomes. Ongoing randomized trials in melanoma, breast, and colorectal cancer should contribute important information regarding the contribution of vitamin D to cancer prevention and treatment.
Dowling RJ, Niraula S, Stambolic V, et al. Metformin in cancer: translational challenges. J Mol Endocrinol
Noto H, Tsujimoto T, Sasazuki T, et al. Significantly increased risk of cancer in patients with diabetes mellitus: a systematic review and meta-analysis. Endocr Pract
Thakkar B, Aronis KN, Vamvini MT, et al. Metformin and sulfonylureas in relation to cancer risk in type II diabetes patients: a meta-analysis using primary data of published studies. Metabolism
Col NF, Ochs L, Springmann V, et al. Metformin and breast cancer risk: a meta-analysis and critical literature review. Breast Cancer Res Treat
Singh S, Singh H, Singh PP, et al. Antidiabetic medications and the risk of colorectal cancer in patients with diabetes mellitus: a systematic review and meta-analysis. Cancer Epidemiol Biomarkers Prev
Singh S, Singh PP, Singh AG, et al. Anti-diabetic medications and risk of pancreatic cancer in patients with diabetes mellitus: a systematic review and meta-analysis. Am J Gastroenterol
Zhang H, Gao C, Fang L, et al. Metformin and reduced risk of hepatocellular carcinoma in diabetic patients: a meta-analysis. Scand J Gastroenterol
Chlebowski RT, McTiernan A, Wactawski-Wende J, et al. Diabetes, metformin, and breast cancer in postmenopausal women. J Clin Oncol
Jiralerspong S, Palla SL, Giordano SH, et al. Metformin and pathologic complete responses to neoadjuvant chemotherapy in diabetic patients with breast cancer. J Clin Oncol
Quinn BJ, Dallos M, Kitagawa H, et al. Inhibition of lung tumorigenesis by metformin is associated with decreased plasma IGF-I and diminished receptor tyrosine kinase signaling. Cancer Prev Res
Zhang P, Li H, Tan X, et al. Association of metformin use with cancer incidence and mortality: a meta-analysis. Cancer Epidemiol
Yin M, Zhou J, Gorak EJ, et al. Metformin is associated with survival benefit in cancer patients with concurrent type 2 diabetes: a systematic review and meta-analysis. Oncologist
Franciosi M, Lucisano G, Lapice E, et al. Metformin therapy and risk of cancer in patients with type 2 diabetes: systemic review. PLoS ONE
Stevens RJ, Ali R, Bankhead CR, et al. Cancer outcomes and all-cause mortality in adults allocated to metformin: systematic review and collaborative meta-analysis of randomised clinical trials. Diabetologia
Porojnicu AC, Lagunova Z, Robsahm TE, et al. Changes in risk of death from breast cancer with season and latitude: sun exposure and breast cancer survival in Norway. Breast Cancer Res Treat
van der Rhee H, Coebergh JW, de Vries E. Sunlight, vitamin D and the prevention of cancer: a systematic review of epidemiological studies. Eur J Cancer Prev
Xu H, Posner GH, Stevenson M, et al. Apc(MIN) modulation of vitamin D secosteroid growth control. Carcinogenesis
Rosen CJ, Abrams SA, Aloia JF, et al. IOM committee members respond to Endocrine Society vitamin D guideline. J Clin Endocrinol Metab
Chung M, Lee J, Terasawa T, et al. Vitamin D with or without calcium supplementation for prevention of cancer and fractures: an updated meta-analysis for the U.S. Preventive Services Task Force. Ann Intern Med
20. Liu SL, Zhao YP, Dai MH, et al. Vitamin D status and the risk of pancreatic cancer: a meta-analysis. Chin Med J (Engl). 2013;126:3356-3359.
Bauer SR, Hankinson SE, Bertone-Johnson ER, et al. Plasma vitamin D levels, menopause, and risk of breast cancer: dose-response meta-analysis of prospective studies. Medicine (Baltimore)
Amir E, Cecchini RS, Ganz PA, et al. 25-hydroxy vitamin-D, obesity, and associated variables as predictors of breast cancer risk and tamoxifen benefit in NSABP-P1. Breast Cancer Res Treat
Trivedi DP, Doll R, Khaw KT. Effect of four monthly oral vitamin D3 (cholecalciferol) supplementation on fractures and mortality in men and women living in the community: randomised double blind controlled trial. BMJ
Lappe JM, Travers-Gustafson D, Davies KM, et al. Vitamin D and calcium supplementation reduces cancer risk: results of a randomized trial. Am J Clin Nutr
Bolland MJ, Grey A, Gamble GD, et al. Calcium and vitamin D supplements and health outcomes: a reanalysis of the Women's Health Initiative (WHI) limited-access data set. Am J Clin Nutr
Avenell A, MacLennan GS, Jenkinson DJ, et al. Long-term follow-up for mortality and cancer in a randomized placebo-controlled trial of vitamin D(3) and/or calcium (RECORD trial). J Clin Endocrinol Metab
Touvier M, Chan DS, Lau R, et al. Meta-analyses of vitamin D intake, 25-hydroxyvitamin D status, vitamin D receptor polymorphisms, and colorectal cancer risk. Cancer Epidemiol Biomarkers Prev
Zheng Y, Zhu J, Zhou M, et al. Meta-analysis of long-term vitamin d supplementation on overall mortality. PLoS One
Rose AA, Elser C, Ennis M, et al. Blood levels of vitamin D and early stage breast cancer prognosis: a systematic review and meta-analysis. Breast Cancer Res Treat