Metformin Is Associated with a Favorable Outcome in Diabetic Patients with Cervical Lymph Node Metastasis of Differentiated Thyroid Cancer

Metformin has been reported to have an antineoplastic effect in certain cancers [1,2]. It is the first-line antidiabetic medication in type 2 diabetes; it improves insulin sensitivity in peripheral organs and has a glucose-lowering effect. There are reports that cancer patients treated with metformin have better outcomes than patients not treated with metformin [3,4,5,6]. In breast cancer patients, metformin treatment has been associated with better overall survival [7]. In colorectal cancer, diabetic patients treated with metformin have also had 30% better overall survival than those receiving other diabetic medications [8].

Several experimental studies have reported on the antineoplastic effects of metformin on thyroid cancer cells [9,10]. The authors suggested that metformin activates adenosine monophosphate-activated protein kinase (AMPK) and inhibits the mammalian target of rapamycin pathway, which induces catabolism or apoptosis in cancer cells. In a recent study, it was suggested that metformin can reprogram cellular energy metabolism and cause cells adapted to the new environment to behave less aggressively [2].

Differentiated thyroid cancer (DTC) has a favorable prognosis, but some patients suffer a recurrence during follow-up [11]. Recurrence of DTC is associated with various factors including initial therapy, age, tumor size, gross soft tissue invasion, and distant metastasis [11,12]. A previous study suggested that the use of metformin was an independent favorable prognostic factor because there was a significant association between metformin and remission rate in diabetic patients with thyroid cancer [9].

In this study, we compared clinical outcomes in diabetic patients with DTC and matched control DTC patients. To assess the metformin effect on DTC, we evaluated the clinical outcomes of diabetic patients with DTC according to metformin treatment. Finally, we compared the clinical outcomes of diabetic patients with cervical lymph node (LN) metastasis of DTC according to metformin treatment, and evaluated the effect of metformin on advanced DTC.

We retrospectively reviewed all patients diagnosed with DTC after total thyroidectomy between 1995 and 2005 in Asan Medical Center, Seoul, Korea. Patients aged between 45 and 75 years, with tumors between 1 and 4 cm, were included in the study. We excluded patients with distant metastasis before surgery or patients with type 1 diabetes. Among 943 DTC patients, we found 60 patients (6%) who had type 2 diabetes. The diagnosis of diabetes was confirmed if one or more of the following were present: (1) the patient had been treated with oral hypoglycemic agent or insulin before surgery or (2) there was documentation of hyperglycemia according to the diagnostic criteria for diabetes recommended by the World Health Organization in 2011, which included fasting plasma glucose ≥7.0 mmol/l, random plasma glucose ≥11.1 mmol/l, 2-hour plasma glucose ≥11.1 mmol/l during an oral glucose tolerance test, and glycated hemoglobin (HbA_1c) ≥6.5%. We defined type 1 diabetes based on age at diagnosis, body mass index (BMI), type of treatment, C-peptide level, and presence of autoimmunity such as glutamic acid decarboxylase antibody.

Finally, we randomly selected 210 patients matched for age, sex, BMI, and tumor size from the 883 nondiabetic patients with DTC as a control group. The study protocol was approved by the institutional review board of Asan Medical Center.

Patients who had been treated with antidiabetic medication(s) for more than 6 months within 2 years after the thyroidectomy were defined as patients who had been treated with antidiabetic medication including metformin. We then divided those patients into a ‘metformin group' and a ‘nonmetformin group' according to whether or not they had received metformin treatment for more than 6 months within 2 years after thyroidectomy.

Clinical recurrence was defined as recurrence of structural disease such as the reappearance of pathologically proven malignant tissue and/or the appearance of metastatic lesions on imaging studies including neck ultrasonography, radioiodine whole body scan, computed tomography (CT) scan, and/or ¹⁸F-fluorodeoxyglucose positron emission tomography with CT (PET-CT) at least 2 years after surgery.

HbA_1c data were collected for 2 years after thyroidectomy and the mean HbA_1c over that period was used for analysis. We also collected serum C-peptide data within 2 years after surgery.

All patients were treated with thyroxine to suppress thyroid-stimulating hormone and were regularly followed up with physical examinations, serum thyroglobulin (Tg), anti-Tg antibody measurement, and neck ultrasonography examinations every 6-12 months. A diagnostic whole body scan was performed during the first 6-24 months after initial therapy as previously described [13]. Additional diagnostic imaging studies including CT scan, magnetic resonance scan, and PET-CT were performed in some patients as needed. The median follow-up period was 8.9 years.

R version 3.0 and the R libraries prodlim, car, Cairo, and survival were used for analyzing data and drawing graphs (R Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org). Continuous variables between two groups were compared using Student's t test. Categorical variables were compared using the χ² test or Fisher's exact test. A Cox proportional hazard model was used to evaluate the risks of recurrence. The multivariate analysis included age, sex, BMI, duration of diabetes, extrathyroidal extension, presence of tumor at resection margin, and serum Tg level at ablation. Disease-free survival (DFS) curves were constructed by the Kaplan-Meier method, and log-rank tests were used to evaluate differences of DFS between patient groups. All p values were two-sided, with p < 0.05 considered statistically significant.

Mean age of the patients was 58.3 ± 6.1 years, and 194 patients (72%) were female. Mean BMI at thyroid surgery was 25.2 ± 2.7. Of the 270 patients, 255 patients (94%) were diagnosed with the classical variant of papillary thyroid carcinoma and 7 patients (3%) were diagnosed with the follicular variant of papillary thyroid carcinoma. Five patients (2%) had follicular thyroid carcinoma and 3 patients (1%) had Hürthle cell carcinoma. There were no significant differences in histology, T stage, N stage, extrathyroidal extension, lymphovascular invasion, tumor at resection margin, or radioiodine remnant ablation between the diabetic patients and the control group (table 1).

Mean duration between diagnoses of diabetes and thyroid surgery was 4.7 years (range: 0.5-21 years). Of the 60 diabetic patients, 35 patients (58%) were treated with metformin. Mean duration of metformin treatment was 7.4 ± 4.8 years, and the mean dose was 979 mg (range: 250-2,000). Thirty patients received 1,000 mg/day or less of metformin and 5 patients received over 1,000 mg/day of metformin. There was no difference in antidiabetic medication other than metformin between the metformin and nonmetformin groups. Mean HbA_1c was 7.3 ± 1.1%, and the mean C-peptide level was 3.5 ± 2.3 ng/ml. Cervical LN metastasis was more prevalent in the metformin group than in the nonmetformin group (OR 3.52, 95% CI: 1.08-12.37, p = 0.035; table 2). There were no significant differences between the groups in age, sex, BMI, preoperative thyroid-stimulating hormone values, the surgeons, extent of surgery, tumor size, HbA_1c, C-peptide, histology, T stage, extrathyroidal extension, lymphovascular invasion, tumor at resection margin, or radioiodine remnant ablation (table 2).

Six of the 60 diabetic patients (10%), and 33 of the 210 controls (16%) suffered recurrence (table 1). In 6 diabetic patients with recurrence, 4 patients underwent additional surgery for recurrence, and 1 of these patients received radioactive iodine (RAI) treatment after surgery. Among 33 patients with recurrence in the control group, 26 patients underwent additional surgery for recurrence and 11 of these patients received RAI treatment after surgery. One patient from the control group received RAI treatment for recurred lesion without additional surgery. There was no significant difference in DFS between the two groups (p = 0.28; fig. 1a), and their rates of recurrence at distant organs were similar (p = 0.31).

Three of the 35 patients (9%) in the metformin group, and 3 of the 25 (12%) in the nonmetformin group had recurrence (table 2). There was no difference in DFS between the two groups (p = 0.72; fig. 1b). There was also no difference in recurrence at distant organs (p = 0.99). When we compared the recurrence rate and DFS according to the dose of metformin (1,000 mg/day or less, and over 1,000 mg/day), there were no differences in recurrence rate (p = 0.38) and recurrence-free survival (p = 0.54) between the two groups.

We performed a subgroup analysis of the 30 diabetic patients with cervical LN metastasis of DTC because LN metastasis was more prevalent in the metformin group than in the nonmetformin group. There were no significant differences in age, sex, BMI, HbA_1c, C-peptide, the surgeons, extent of surgery, tumor size, histology, T stage, extrathyroidal extension, radioiodine remnant ablation, or serum Tg at ablation between the metformin and nonmetformin subgroups (online suppl. table 1, see www. karger.com/doi/10.1159/000437365). Two of the 22 patients (9%) in the metformin subgroup and 3 of the 8 patients (38%) in the nonmetformin subgroup had recurrent DTC. The metformin subgroup had significantly longer DFS (17.1 ± 1.0 years) than the nonmetformin subgroup (8.6 ± 1.6 years; HR 0.16, 95% CI: 0.03-0.95, p = 0.021; fig. 1c), and metformin treatment was significantly associated with lower recurrence in multivariate analysis. The HR for metformin treatment was 0.02 (95% CI: 0.0004-0.85, p = 0.041) after adjustment for age, sex, BMI, and duration of diabetes, and 0.03 (95% CI: 0.002-0.47, p = 0.01) after adjustment for age, BMI, duration of diabetes, and extrathyroidal extension. This effect was also significant after adjusting for age, BMI, duration of diabetes, presence of tumor at resection margin, and serum Tg level at ablation (HR = 0.03, 95% CI: 0.001-0.78, p = 0.035; table 3). There was no significant difference in recurrence at distant organs between the metformin and nonmetformin subgroups (p = 0.47).

In this study, we evaluated clinical outcomes in 60 diabetic patients and 210 nondiabetic patients matched for age, sex, BMI, and tumor size. There were no differences in the clinicopathological features of DTC in the two groups. We evaluated clinical outcomes in the diabetic patients according to metformin treatment and found that metformin was associated with a favorable outcome only in the diabetic patients with cervical LN metastasis of DTC: the metformin group had a significantly longer DFS than the nonmetformin group.

Previous studies have suggested that metformin has antineoplastic effects including induction of catabolism, downregulation of cell proliferation, cell cycle arrest, and apoptosis in the absence of tumor-specific compensatory mechanisms [1,14]. Cancer cells replace the respiration of oxygen occurring in normal body cells by fermentation and produce large amounts of lactate by glycolysis even when oxygen supply is sufficient, so they require higher levels of mitochondrial oxidative phosphorylation. This fermentation phenomenon is called the ‘Warburg effect' [15]. Previous work has shown that metformin inhibits oxidative phosphorylation and reduces ATP production [2,16]. The decline in ATP level triggers activation of AMPK, which inhibits the mammalian target of rapamycin pathway and downregulates growth factor signaling [1,2]. Nevertheless, cancer cells which are defective in AMPK signaling and downstream effectors are paradoxically hypersensitive to metformin. They cannot reduce their energy consumption by activating AMPK in conditions of energy stress, and therefore suffer an energy crisis and may undergo cell death [17,18]. Metformin also leads to suppression of fatty acid synthase gene expression and inactivation of acetyl-CoA carboxylase. This causes reduction in lipogenesis and synthesis of the acetyl-CoA carboxylase product malonyl-CoA resulting in increased fatty acid oxidation. Moreover, AMPK activation may result in p53 activation, which is a tumor suppressor that is often mutated in other type of cancer [19,20]. There have been several in vitro and in vivo studies on papillary, medullary, and anaplastic thyroid cancer cell lines evaluating the effects of metformin on the suppression of clonal growth, self-renewal, and proliferation of cancer stem cells [9,10,21,22].

Several clinical studies have suggested that metformin is beneficial in diabetic patients in preventing the progression of certain cancers [1,3,4,5,7,8,23]. Cancer patients treated with metformin before cancer diagnosis had significantly increased survival than nondiabetic patients or diabetics taking other antidiabetic medications [3,7,8,24]. Some studies have suggested that increased serum insulin was associated with cancer proliferation, and both insulin levels and cancer cell proliferation were blunted by metformin [7,25]. Although Klubo-Gwiezdzinska et al. [9] reported that metformin was an independent factor increasing the remission rate and progression-free survival of DTC patients with diabetes, there have been few studies of the association between metformin and the prognosis of thyroid cancer.

It is difficult to establish an effect of metformin on the prognosis of DTC because many clinical factors such as BMI, insulin resistance, cancer subtype, and drug level can influence its effects [26,27,28]. It has been reported that metformin caused a marked reduction of Ki-67 in a subgroup of breast cancer patients with high BMI, high insulin resistance, or high C-reactive protein levels [26]. In our study, metformin treatment was associated with a favorable outcome only in diabetic patients with cervical LN metastasis of DTC even though cervical LN metastasis was more prevalent in the metformin group than the nonmetformin group. We found that metformin was associated with prolonged DFS in these patients at high risk for DTC recurrence.

This study had the limitation that it was retrospective and there was only a small number of diabetic patients with DTC. We matched for age, sex, BMI, and tumor size to avoid the selection bias. However, we did not control the pathological characteristics of DTC except the tumor size because only tumor size could be preoperatively estimated. It was an incidental finding that cervical LN metastasis was more prevalent in the metformin group than control group. In this study, thyroid surgery was done by four surgeons. The recurrence rate could be affected by the extent of surgery and the skills of the surgeons. However, there were no significant differences in the surgeons and extent of surgery between the metformin and nonmetformin groups. Nevertheless, this is the first study to suggest that the metformin effect is only significant in the advanced stage of DTC. A further larger study is needed to establish a clear protective effect of metformin on DTC recurrence.

In conclusion, metformin treatment is associated with longer DFS in diabetic patients with cervical LN metastasis of DTC. Our findings suggest that metformin has an antineoplastic effect at a certain stage of DTC.

This study was supported by the National Research Foundation of Korea (DIRAMS) grant funded by the Korea government (MSIP, No. 50600-2014).

The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.