Abstract
Background: The present study aimed to evaluate the effects of metformin on the clinical outcomes of patients receiving radiotherapy for inoperable hepatocellular carcinoma. Patients and Methods: The medical records of 217 patients treated with stereotactic body or hypofractionated radiotherapy for inoperable hepatocellular carcinoma were reviewed. Patients were divided into the metformin group (n=19) and the non-metformin group (n=198), including those with diabetes (n=29), and those without (n=169). We performed a propensity score-matching analysis comparing the two groups. Results: In the propensity score-matched cohort (n=76), the overall survival rate of the metformin group was higher than that of the non-metformin group (2-year, 76% vs. 37%, p=0.022). The adjusted Cox proportional hazards model revealed that metformin usage was a significant factor for mortality (adjusted hazard ratio=0.361; 95% confidence interval=0.139-0.935). Conclusion: The use of metformin in patients with hepatocellular carcinoma receiving radiotherapy was associated with higher overall survival.
Hepatocellular carcinoma (HCC) is the sixth most frequently diagnosed cancer worldwide, and yet it was the second most common cause of cancer death in 2008 (1). Surgical resection, orthotopic liver transplantation and radiofrequency ablation (RFA) form the cornerstone of curative treatment for HCC. However, most patients are not suitable for local therapies owing to large tumor size, poor liver function, co-morbidities, or other technical difficulties (2, 3). In such cases, multi-disciplinary modality treatment, such as transarterial chemoembolization (TACE) or radiotherapy is often indicated for local control. Recently multiple clinical studies, including ours, have reported favorable outcomes with precise techniques of radiation therapy, such as stereotactic body radiotherapy (SBRT) and intensity-modulated radiation therapy (4-8). We previously reported that SBRT resulted in high local control and modest survival benefits in patients with HCC without severe toxicity (5).
Metformin (1,1-dimethylbiguanide hydrochloride), one of the most widely prescribed drugs for type 2 diabetes mellitus, has been clinically associated with potential anti-tumor effects (9, 10). A meta-analysis suggested that metformin might reduce the overall incidence of cancer, including HCC (11, 12). In a number of recent epidemiological and cohort studies, metformin usage was shown to improve overall or disease-free survival in patients with various cancer types (13-16). Furthermore, several clinical studies have suggested that metformin provides a synergistic benefit with chemotherapy or radiotherapy against certain cancer types (17, 18).
Moreover, a few studies suggested the involvement of metformin in the epithelial–mesenchymal transition (EMT), a process associated with metastatic dissemination of epithelial cancer cells and cancer cell progression and invasion into the surrounding environment (19, 20). On the other hand, HCC is generally considered insensitive to systemic chemotherapy (21). Therefore, treatment strategies aiming to reduce intrahepatic or extrahepatic metastasis after local treatment is mandatory to achieve survival benefits. Thus, we hypothesized that the use of metformin might increase overall survival by preventing metastasis and increasing local control. We then conducted a retrospective cohort study to evaluate the effect of metformin on the clinical outcomes of patients with HCC treated with SBRT or hypofractionated radiotherapy (HypoRT).
Patients and Methods
Patients. We retrospectively reviewed medical records of 217 patients with HCC who were treated with SBRT or HypoRT at four Institutions between March 2003 and December 2012. These four institutions have comprehensive cancer care Centers. Treatment plans for patients with HCC were decided by the multi-disciplinary hepatogastrointestinal tumor conference. All patients provided written informed consent after being counseled on the possible benefits and complications of SBRT or HypoRT. This study was approved by the Institutional Review Board of each participating institution (K-1302-002-031). The indications for SBRT or HypoRT were as follows: unsuitable for surgery or local ablation, confined to the liver without extrahepatic metastasis, unsuitable for TACE, or an incomplete response to TACE (n=197, 91%). An incomplete response to TACE was defined as incomplete tumor filling by the lipiodol-doxorubicin mixture demonstrated via computed tomographic (CT) evaluation at 1 month after an adequately performed TACE, or as the continuous elevation of tumor markers levels, such as alpha-fetoprotein (AFP) or protein induced by vitamin K absence or antagonist-II even immediately after TACE. In cases of intrahepatic or extrahepatic disease progression, customized salvage treatment was performed, but no further treatment was indicated when no disease progression was observed.
Out of the 217 patients with HCC, 48 (22%) had type 2 diabetes mellitus and 19 (9%) received metformin during radiotherapy. These 19 patients received metformin for at least one year (range=16-243 months). The median daily dose of metformin was 1,000 mg (range=500-2000 mg). Other patients received sulfonylurea (n=16), thiazolidinedione (n=3), meglitinide (n=1), alpha-glucosidase inhibitor (n=1), or insulin (n=8) for the treatment of type 2 diabetes mellitus. Hemoglobin A1c (HbA1c) levels within 6 months before and after radiotherapy were recorded for 83% of diabetic patients. Patients were divided into the metformin (n=19, 9%) and non-metformin (n=198, 91%) groups. Those in the non-metformin group were further divided into the diabetic non-metformin (n=29, 13%) and non-diabetic (n=169, 78%) subgroups. This information was identified from the review of patient and pharmacy records.
Stereotactic body radiotherapy/hypofractionated radiotherapy. In the present study, SBRT and HypoRT were delivered using a RapidArc (Varian Medical Systems, Palo Alto, CA, USA), Cyberknife (Accuray Inc., Sunnyvale, CA, USA), or tomotherapy (Accuray Inc.) machine. Breathing-related tumor motion was controlled by abdominal compression, gating, or tracking techniques. Daily imaging guidance by orthogonal radiography or on-board CT was used to relocalize the target lesion before treatment delivery. Doses of 25 to 60 Gy in three to six fractions were delivered and the median biologically effective dose (BED) was 137.7 Gy10 (range=50.2-180.0 Gy10).
Study outcomes. The study outcomes included overall survival (OS), progression-free survival (PFS), local control, (LC) intrahepatic metastasis-free survival (IHFS), and extrahepatic metastasis-free survival (EHFS). Treatment response was assessed using the modified Response Evaluation Criteria in Solid Tumors (22). LC was defined as the absence of progressive disease within the treated lesion, whereas intrahepatic metastasis was defined as disease recurrence in hepatic areas other than the treated lesion, and extrahepatic metastasis was defined as disease recurrence in areas of the body outside the liver. PFS was defined as living status without any progression of the treated lesion and without intrahepatic or extrahepatic metastasis. Data were analyzed with a closeout (study censor) date of August 20, 2013. Follow-up time was calculated from the date of definitive treatment initiation to that of the last contact or death for OS and the last imaging assessment or death for other outcomes, respectively. Time to failure was calculated from the date of definitive treatment initiation to that of the relevant event.
Statistical analysis. Propensity score matching was performed to minimize the differences in baseline characteristics between the metformin and non-metformin groups. Propensity scores, defined as the conditional probability of being treated given the covariates, were used to adjust for possible bias due to nonrandom assignments of patients with different conditions (23). This propensity model consisted of the following 11 baseline covariates: age, gender, Eastern Cooperative Oncology Group (ECOG) performance status, etiology, the number of TACE session, portal vein tumor thrombus (PVTT), BED, tumor size, Child-Pugh class, AFP level, and multiple tumor lesions. Patients were matched based on the logit of propensity score using a caliper of width equal to 0.2 of the standard deviation of the estimated propensity score logit. Such a caliper width has been found to result in the optimal estimation of risk differences in a variety of settings (24). Accordingly, we performed a 1-to-3 matched analysis comparing the metformin and non-metformin groups. Before and after propensity score matching, a balance test was then performed using standardized differences to assess the success of propensity score matching. We aimed to achieve an absolute standardized difference (ASD) of 0.1 or less for all covariates (25).
Survival rates were calculated using the Kaplan–Meier method. Using the matched study sample, we computed the quartiles of the estimated propensity score. Matched patients were stratified into four approximately equal-size groups using such quartiles of the estimated propensity score (25). Intergroup comparisons were performed using the stratified log-rank test. All factors with p≤0.1 in univariate analysis, propensity score, and diabetic status were subjected to multivariate analysis using a Cox proportional hazards model with a forward stepwise procedure. All statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS) (version 14.0; SPSS, Inc., Chicago, IL, USA), and p≤0.05 were considered statistically significant.
Results
Baseline characteristics. Out of the 217 patients with HCC included in this study cohort, 19 (9%) received metformin during radiotherapy. Baseline characteristics for the entire cohort and the propensity score-matched cohort are listed in Table I. Before matching, the proportion of patients with BED over 100 Gy10 tended to be higher in the metformin group than in the non-metformin group (n=13, 68% vs. n=100, 51%; ASD=0.415). The average tumor size of the metformin group was likely to be smaller than that of the non-metformin group (5.03 vs. 6.48 cm, ASD=0.345). However, after propensity score matching, patient characteristics were mostly similar between the two groups, except for diabetic status. The absolute standardized differences of all variables were 0.1 or less (Table I). There was no significant difference in HbA1c level (average 7.4 vs. 7.3%, ASD=0.098) between the metformin and diabetic non-metformin group, and no imbalance in the existence of co-morbidities between two groups.
Unmatched total cohort. After a median follow-up period of 15 months (range=1-83 months), five (26%) patients in the metformin group, 13 (45%) in the diabetic non-metformin group, and 89 (53%) in the non-diabetic group had died [metformin vs. non-metformin, hazard ratio (HR)=0.385; 95% confidence interval (CI)=0.157-0.947]. The 2-year OS rates for the metformin, diabetic non-metformin, and non-diabetic groups were 76%, 45%, and 46%, respectively (p=0.035 by log-rank test, Figure 1). Survival analysis revealed that OS rates was worse for the diabetic non-metformin group (p=0.097 by log-rank test) and significantly worse for the non-diabetic group (p=0.027 by log-rank test) than for the metformin group. Univariate analysis revealed ECOG performance status, PVTT, Barcelona Clinic Liver Cancer stage, BED, tumor size, and AFP level as significant prognostic factors for OS. Multivariate analysis identified that ECOG performance status, BED, AFP level were significant prognostic factors for OS.
Propensity score-matched cohort. In the propensity score-matched cohort (n=76), the median follow-up period was 16 months (range=1-76 months). The use of metformin (HR=0.361; 95% CI=0.139-0.935) was a significant factor for mortality in the multivariate analysis adjusted on propensity score and diabetic status. The Kaplan–Meier survival curve indicated that the OS rate of the metformin group was higher than that of the non-metformin group (2-year, 76% vs. 37%, p=0.022 by stratified log-rank test, Figure 2A). Similarly, the PFS rate for the metformin group was also higher than that of the non-metformin group (2-year, 46% vs. 16%, p=0.045 by stratified log-rank test, Figure 2B). LC and IHFS rates were not statistically significantly different between the two groups (2-year, 87% vs. 77%, p=0.664 by stratified log-rank test and 52% vs. 29%, p=0.262 by stratified log-rank test, respectively). On the other hand, the EHFS rate for the metformin group was higher than that of the non-metformin group, with borderline statistical significance (2-year, 85% vs. 64%, p=0.090 by stratified log-rank test).
Discussion
In the present study, we found that the use of metformin significantly increased the OS and PFS of patients with HCC treated with radiotherapy. The use of metformin was an independent predictive factor of OS in patients with HCC after adjustment for other prognostic factors. In addition, such use tended to increase the EHFS in patients with HCC.
Recently, many studies involving high doses of radiation, such as SBRT and HypoRT, have reported favorable outcomes in patients with HCC, including an LC rate of more than 90% (4-8). However, as seen with other local therapies, intrahepatic or extrahepatic metastasis outside the radiotherapy-treated lesions remain a problem, providing the rationale for combining regional or systemic treatments with SBRT (5, 6). A combination of radiotherapy and concurrent chemotherapy or TACE has been performed at several institutions (5, 27), and a phase III randomized trial of sorafenib-alone versus SBRT followed by sorafenib for locally advanced HCC is ongoing (RTOG, Radiation Therapy Oncology Group 1112). Similarly, a combination of radiotherapy as the local modality and metformin as the systemic treatment may be a potential therapeutic option.
Furthermore, many recent studies have suggested that metformin use is significantly associated with longer survival in patients with non-metastatic cancer (Table II). Metformin reportedly reduced overall or cancer-specific mortality by 32–76% in various cancer types. It might also provide a synergistic benefit with chemotherapy or radiotherapy against certain types of cancer (13-17, 28-33). Metformin in patients with rectal cancer receiving chemoradiotherapy was associated not only with significantly higher pathological complete response rates, but also improved survival (17). Furthermore, metformin was associated with a dose-dependent increased response to chemoradiotherapy in patients with esophageal cancer and might be a sensitizer to such therapy (30). It was also associated with an independent and significant decrease in early biochemical relapse rates in patients with localized prostate cancer, with improved tumor oxygenation and response to radiotherapy (33). In addition, metformin might also reduce distant metastasis of cancer. In patients with localized prostate cance treated with external-beam radiation therapy, both OS and distant metastasis-free survival were significantly lower in the diabetic non-metformin group (31). In patients with triple-negative breast cancer, compared to the metformin group, the diabetic non-metformin and non-diabetic groups tended to have a higher risk of distant metastases (32). Several clinical trials are currently underway to investigate the efficacy of metformin in breast, prostate, endometrial and pancreatic cancer, and other solid malignancies. The largest to date is a phase III randomized multi-center trial (NCT01101438) that aims to recruit 3,582 participants with early-stage nonmetastatic breast cancer (34).
The exact mechanism of action of metformin in patients with HCC treated with radiotherapy is elusive. We postulated that metformin improved OS by both enhancing the radiosensitizing effect and reducing metastasis. In vitro, metformin activated 5’ adenosine monophosphate-activated protein kinase (AMPK), inhibited the mammalian target of rapamycin and down-regulated Livin protein expression. Such activities suggest that metformin is effective at initiating apoptosis and inhibiting key survival signaling pathways in HCC cells (35). The combination of low metformin doses and ionizing radiation resulted in a much stronger cytotoxicity than the treatment with metformin or ionizing radiation alone, leading to an ~80% decrease in cell viability and ~35% increase in cell-cycle arrest in two HCC cell lines (21). Metformin also activated AMPK and modified ionizing radiation responses of human cancer cells in a liver kinase B1-independent manner, resulting in the induction of p21WAF/CIP and regulation of the cell cycle and survival (36). Another newly-proposed mechanism of action involved the improvement of tumor oxygenation by metformin (37). Tumor hypoxia is a negative prognostic factor in cancer and HypoRT can result in a significant decrease in tumor cell killing compared to standard fractionation as a result of tumor hypoxia (37). Metformin treatment significantly improved tumor oxygenation in two xenograft models. It also led to improved radiotherapy responses when administered immediately before irradiation in mice (33). However, we were unable to demonstrate the synergistic effect of metformin with radiotherapy. In our study, all patients received SBRT or HypoRT in two weeks, which could be relatively short for achieving a radiosensitizing effect. The LC rates were considerably high, with a small difference in both groups (2-year, 87% in metformin and 77% in non-metformin group). These may make it difficult to find a statistically significant difference.
Another proposed mechanism of action by metformin is via reducing distant metastasis of cancer cells. Patients with breast cancer with high expression of cluster of differentiation (CD)-24 experienced limited distant metastasis-free survival, and metformin suppressed the metastasis-associated protein CD24 in MDA-MB-468 triple-negative breast cancer cells (38). In a xenograft model of ovarian cancer, metformin not only reduced tumor size but also the number and size of metastatic lung nodules, along with AMPK activation (39). Metformin might also suppress EMT in cancer cells (18). EMT is a key developmental process is often activated during cancer invasion and metastasis. Specifically, most patients with HCC have underlying chronic liver diseases, particularly liver cirrhosis. Cirrhotic liver-derived hepatocytes have been shown to exhibit increased vimentin and type 1 collagen expression, as well as morphological features consistent with EMT (40). Our data, that indicated that metformin use tended to increase EHFS of patients with HCC, providing clinical support for such a mechanism of action by reducing distant metastasis of cancer cells.
The strength of the present study was the homogenous subject population, including only those with inoperable HCC, no distant metastases, and curative radiotherapy. We also performed propensity score matching to minimize the differences in baseline characteristics between the metformin and non-metformin groups. Therefore, important confounders of cancer mortality, such as differences in tumor characteristics and treatment diversity, were eliminated. Furthermore, to the best of our knowledge, the present study is the first to demonstrate the positive effect of metformin on the OS of patients with HCC with adjustment for other factors through propensity score matching. However, there were several limitations. Firstly, we inevitably encountered the limitations inherently associated with a retrospective study of a limited sample size. Further studies such as a large-scale prospective trial are, therefore, required. Secondly, our study did not consider the impact of metformin dose and duration. Thirdly, the matched non-metformin group included both diabetic and non-diabetic patients. Diabetes is an important risk factor for the development of HCC and is also associated with poorer survival in HCC (41). Comparing diabetic (with poor prognosis) with non-diabetic group (with better prognosis) may attenuate the impact of metformin use. Nevertheless, we found that the use of metformin significantly increased the OS and PFS of patients with HCC treated with radiotherapy. These results were similar after adjustment for other prognostic factors including diabetic status.
In conclusion, our results indicated that patients with HCC who received metformin had higher OS and PFS than those who did not. To our knowledge, the present study is the first to demonstrate such survival benefits by metformin in patients with HCC receiving radiotherapy. Therefore, metformin might be useful in the combination treatment for inoperable HCC. Further studies to evaluate the potential of metformin as an antitumor agent are warranted, and our data provide the theoretical support for such trials.
Acknowledgements
The Authors declare that they have no competing interests. This study was supported by the National Nuclear R&D Program of the Ministry of Science, ICT and Future Planning, Republic of Korea, and a grant from the KRP (KCCH Research Competence Promotion) (Grant No. 50244-2014) and RTR (Radiological Translational Research program) (Grant No. 50476-2015), Korea Institute of Radiological & Medical Sciences. The study was approved by the Korean Radiation Oncology Group (KROG 14-02). This study was presented as a poster discussion at the 5th Asia-Pacific Primary Liver Cancer Expert Meeting (2014APPLE) and selected for the Best Poster Award, July 11–13 2014, Taipei, Taiwan.
- Received April 30, 2015.
- Revision received June 6, 2015.
- Accepted June 8, 2015.
- Copyright© 2015 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved