Metformin may offer no protective effect in men undergoing external beam radiation therapy for prostate cancer

Introduction

Prostate cancer (PCa) is the second most common cancer in men, with an estimated 1.1 million cases diagnosed worldwide in 2012 1. External beam radiation therapy (EBRT) with androgen deprivation therapy (ADT) is one of the standard methods of curative treatment for localized PCa 2, 3.

Metformin is a biguanide anti-hyperglycaemic agent used in type 2 diabetes because of its significant benefits in reducing cardiovascular disease 4. The use of metformin in PCa has received much recent attention because of its potential role in PCa protection and especially in ameliorating the metabolic effects encountered during the use of ADT 4, 5. Furthermore, in several studies, metformin was shown to have protective effects in PCa, with reduced castrate resistance, early stabilization of PSA in castrate-resistant disease and improved overall survival (OS) 6-9; however, the evidence is conflicting 5, 10.

Metformin is thought to have anti-tumorigenic activity not only via direct action on the tumour, but also indirectly by lowering systemic insulin levels. Both pathways activate AMP-activated protein kinase (AMPK), which in turn inhibits mammalian target of rapamycin (mTOR) and causes reduced cell proliferation 11, 12. Hypoxia-inducible factor 1-α (HIF1α) is a transcription factor that promotes cancer growth, cell proliferation, angiogenesis, metastases 13 and survival of PCa cells 14. In addition, HIF1α has been shown to increase resistance to radiotherapy 15, and targeting the mTOR/HIF1α pathway improves radio-sensitivity in endometrial cancer 16. As mTOR is an upstream regulator of HIF1α 17, we hypothesized that metformin should reverse the effects of HIF1α, thereby reducing radio-resistance and increasing survival in men undergoing EBRT for PCa, and increasing the sensitivity of PCa cells to H₂O₂ toxicity.

The aim of the present study was to assess the effects of metformin on PCa outcomes in men treated with EBRT and to determine its effect on HIF1α and cell survival in vitro.

Patients and Methods

All patients with histological evidence of adenocarcinoma of the prostate and clinical stage T1–3b Nx/0 M0, who were treated with curative intent with EBRT or combined EBRT+ high-dose-rate boost at the Peter McCallum Cancer Center from 2000 to 2007 inclusive, and who had a minimum potential follow-up of 5 years, were included.

The following data were collected: pretreatment PSA value 18 (at least one), Gleason score on biopsy, clinical T stage, family history of PCa, previous cancer history, comorbidity information (27-item Adult Comorbidity Evaluation [ACE-27]) and the use of metformin and relevant associated medications (aspirin, statins), as well as the duration of adjuvant ADT (short-term or long-term). ACE-27 is a validated comorbidity index, which takes into account all systems, including the cardiovascular system, and is widely used in cancer studies including those on PCa 19. Therefore, this was used to assess and score cardiovascular disease and comorbidities, rather than cardiac disease, individually. The following outcome measures of interest were analysed: all follow-up PSA values; use of salvage ADT; time to metastases; and time to and cause of death.

For all cases with data missing for >18 months, or aberrant data (such as a falling PSA in the absence of recorded therapy), the patient medical record was reviewed and, where no data were available for a period of >2.5 years, the patient was deemed to be lost to follow-up at the start of that period.

Endpoints

The primary endpoint was based on biochemical failure (BF), which was defined by the Phoenix criteria 20, whereby failure is defined when PSA reaches a level of ≥2 ng/mL above the lowest PSA recorded previously. Time to BF was taken as the interval from completing radiotherapy to the date of BF. Secondary endpoints were time to commencing salvage ADT, time to distant relapse (metastasis), and time to death from PCa or any cause.

In Vitro Studies

Results

A total of 2055 eligible cases were identified (Table 1). Of these, 113 patients were on metformin and had a median potential follow-up of 95.7 months. A total of 544 patients experienced BF and 463 deaths were recorded, of which 153 were from PCa. Approximately 30% of patients had received short-term ADT (given as neoadjuvant concurrently with radiotherapy), and 25% had received long-term ADT (neoadjuvant and adjuvant). T-stage, PCa risk and ADT usage were similar in those who were on metformin and those not on metformin. While those on metformin were of similar ages to those who were not, the overall comorbidity burden by ACE-27 score (reflected in the number of prescription medications) was higher in the metformin group (Table 1).

Metformin does not Reduce HIF1α Expression or PC3 Cell Survival after Exposure to Oxidative Stress

We have previously shown that HIF1α is overexpressed in PC3 under normoxic conditions 21. Pre-treatment of PC3 cells with the mTOR inhibitor rapamycin was shown to inhibit the accumulation of HIF1α 17. Further studies have shown that metformin inhibits mTOR via both AMPK-dependent and -independent pathways 22; therefore, we investigated whether metformin might reduce the expression of HIF1α in PC3 cells via inhibition of mTOR. As shown in Fig. 1A, treatment of PC3 cells with increasing doses of metformin to a maximum concentration of 100 μM had no effect on HIF1α. In contrast, exogenous 10 μM ferric ions (Fe²⁺) resulted in the degradation of HIF1α to 35 ± 7% in PC3 cells compared to untreated control cells, as demonstrated previously 14. Exposure of cells to ionizing radiation during radiotherapy generates oxidative stress via reactive oxygen species including H₂O₂. Radiotherapy and H₂O₂ have both been shown to induce cell death in PC3 cells 23, and we have previously demonstrated that overexpression of HIF1α in PC3 under normoxic conditions contributes to resistance to oxidative stress induced by exogenous H₂O₂ 21. As seen in Fig. 1B, the failure of metformin to increase the sensitivity of PC3 cells exposed to oxidative stress (50 μM H₂O₂) is in line with the fact that metformin did not change the expression of HIF1α. Further, to test if these observations are applicable to androgen-dependent PCa, we measured the expression of HIF1α in LNCaP cells under normoxic conditions. The data shown in Fig. 1C confirm our previous observation of a very low level of basal HIF1α in LNCaP cells 14. Importantly, metformin did not affect the expression of HIF1α, nor did it increase the sensitivity of LNCaP cells exposed to oxidative stress (25 μM H₂O₂), as shown in Fig. 1C.

Discussion

Metformin has been shown to have protective effects in PCa with reduced castrate resistance and improved OS 6-8. A recent meta-analysis demonstrated a reduced risk of biochemical recurrence in PCa 24, a finding not seen with other anti-glycaemic medications 25. Furthermore, metformin in conjunction with lifestyle changes has the added benefit of countering some of the side effects of ADT, such as reducing cardiovascular risks and metabolic syndrome (characterized by a cluster of cardiovascular risk factors) 26. However, the present study failed to show an association between metformin and BF, OS or metastases. This is consistent with other studies which have also failed to demonstrate any benefit from metformin in PCa 10, 27-30.

A number of reasons have been suggested for these differences. One major drawback of the studies showing a clear benefit with metformin is that they were subject to immortal time bias, whereby misclassification occurred between the time of cohort entry and the exposure to metformin in a time-dependent fashion, potentially enhancing its protective effects 30. For example, Spratt et al. 6 (2901 patients) included patients who were on metformin at the time of the diagnosis of PCa or any time after radiotherapy, and 18% of this population initiated metformin therapy after radiation therapy had completed. Furthermore, other factors, such as latency (time of exposure), reverse causality, and dose–response are also recognized biases in these studies 30. To overcome these biases, we examined the use of metformin only at the time of starting radiotherapy to assess the effects of metformin on treatment outcomes. However, factors such as duration of metformin use, non-compliance and undocumented variation in doses may have influenced our longer-term results.

One of the unexpected findings in the present study was that treatment with higher doses of metformin was associated with a higher risk of PCa-specific death in patients with diabetes. Although this result needs to be interpreted with caution because of the small number of PCa-specific deaths in the present study, another study also reported similar findings, demonstrating worse survival with metformin in patients with diabetes and PCa 30. In that study, which corrected for the above-mentioned biases associated with metformin, Bensimon et al. 30 suggested that the poorer survival might be explained by the fact that men with clinically advanced PCa are likely to be kept on oral anti-glycaemic agents such as metformin, rather than converted to insulin injections as a palliative measure. As diabetes has been shown to increase the rate of progression of PCa 27, poor glycaemic control requiring higher levels of metformin could be an additional factor. While the small number of PCa-specific mortalities 31 amongst those with diabetes may explain these findings, a larger group size may be required to demonstrate statistical significance for a small effect and thus the findings should be interpreted with caution.

Radiotherapy and H₂O₂ have both been shown to induce apoptosis of PC3 cells through the production of reactive oxygen species 23. Metformin inhibits AMPK, thereby impacting tumour cell growth via both direct and indirect mechanisms 12. In both pathways, mTOR plays a key role in reducing tumour cell proliferation 12. HIF1α is a key transcription factor, regulated by mTOR, which promotes cancer growth, cell proliferation, angiogenesis, metastases and tumour cell survival 13, 14, and is also thought to increase resistance to radiotherapy 15. However, in the present in vitro study, metformin failed to reduce expression of HIF1α and correspondingly failed to further sensitize PC3 cells against cytotoxic H₂O₂. These results are in agreement with the clinical aspect of the study, where metformin failed to show any benefit following radiotherapy. Although some previous studies have shown inhibitory effects of metformin on tumour cell proliferation or HIF1α expression, dose–response experiments indicated that the inhibitory effect of metformin starts at 1 mM and reaches a maximum at 10 mM 32, 33. The small but significant adverse effect seen in patients taking high doses of metformin calls into question the clinical translation of these in vitro observations, although a correlation between in vitro doses of metformin with in vivo or clinical dosage has not yet been established.

In conclusion, no association was found between the use of metformin and time to metastases, time to BF or OS in men undergoing ADT with or without radiation therapy for PCa; however, in men with diabetes on high doses of metformin there may be an increased risk of PCa-specific mortality. In vitro, low therapeutic concentrations of metformin had no effect on HIF1α, and this observation could explain the conflicting evidence for metformin in men undergoing EBRT for PCa. Higher, more toxic doses of metformin may be required to inhibit the mTOR-HIF1α pathway in this patient group. Currently available data fail to provide a rationale for the use of metformin as an intervention to improve PCa treatment outcomes.

Conflict of Interest

None declared.