Association of Metformin Use and Survival Outcome in Women With Cervical Cancer

Abstract

Objective

Although preclinical studies suggest possible antitumor effects of metformin against cervical cancer, there is currently a lack of clinical data examining the association of metformin use and survival in women with cervical cancer. The aim of this study was to examine survival of women with cervical cancer who were receiving metformin.

Methods

This is a retrospective study examining consecutive cases of stages I to IV cervical cancer between 2000 and 2014. Patient demographics, medication use, tumor characteristics, treatment patterns, and survival outcomes were correlated to metformin use.

Results

There were 70 (8.9%; 95% confidence interval [CI], 6.9–10.9) metformin users and 715 nonusers identified for the analysis. Median follow-up time was 22.6 months. Recurrence/progression of disease and death due to cervical cancer were observed in 236 and 163 cases, respectively. Metformin users were more likely to be older, hypertensive, diabetic, and dyslipidemic compared with nonusers (all, P < 0.05). On univariate analysis, metformin users and nonusers had similar progression-free survival (PFS) (5-year rates; 57.3% vs 61.8%; P = 0.82) and cervical cancer–specific overall survival (71.7% vs 70.7%; P = 0.86). After adjusting for patient demographics and tumor characteristics, metformin use was not associated with PFS (adjusted hazards ratio, 1.11; 95% CI, 0.70–1.74; P = 0.67) or cervical cancer–specific overall survival (adjusted hazards ratio, 0.91; 95% CI, 0.52–1.60; P = 0.75). Among 478 women who received whole pelvic radiotherapy, metformin use was not associated with PFS (P = 0.93) or cervical cancer–specific overall survival (P = 0.32).

Conclusions

In this study population, metformin use was not associated with survival of women with cervical cancer.

Cervical cancer is the third most common cancer diagnosis and cause of death among gynecologic cancers in the United States.¹ Persistent infection with oncogenic human papillomavirus (HPV) is known to be the cause of cervical cancer development.² Surgery is commonly performed as the upfront treatment for early-stage disease, whereas radiotherapy is an integral part of the management of more advanced disease, either as postoperative adjuvant therapy, primary definitive treatment, or palliative intent. Although cure is highly achievable in early-stage disease, advanced-stage and recurrent diseases have a generally poor prognosis, and the management of such conditions is always challenging.³ Therefore, identifying any effective approach that can improve the prognosis of women with cervical cancer is crucial.

Metformin, an insulin-sensitizing biguanide, is an old-class generation and widely accepted first-line agent for type II diabetes mellitus.⁴ The antihyperglycemic effect is a consequence of reduced hepatic gluconeogenesis and increased insulin-stimulated glucose uptake in skeletal muscle and adipose tissues.⁴ In addition to the glycemic control effects, metformin provides various additional effects to correct several physiologic abnormalities.^5,6 Over the past 2 decades, protective effects of metformin use on cancer development risk^7,8 and cancer-related mortality risk of various cancer types have been reported.^9,10 A recent meta-analysis suggested that metformin use may be a useful agent to improve survival outcome of patients with colorectal and prostate cancers, particularly among those who are receiving metformin during pelvic radiotherapy.⁹

Several preclinical studies have suggested metformin use may be effective in antitumor treatment of women with cervical cancer.^11–14 However, whether metformin can improve the survival of women with cervical cancer remains understudied, and currently available clinical data are limited to a select population of diabetic women with cervical cancer.¹⁰ Moreover, there is no previous study examining the association of metformin use during pelvic irradiation and survival of women with cervical cancer. The objective of the study was to examine the survival outcome of women with cervical cancer who were taking metformin.

MATERIALS AND METHODS

Eligibility Criteria

After institutional review board approval was obtained at the University of Southern California, a retrospective study was conducted to examine consecutive cases of stages I to IV invasive cervical cancer diagnosed and managed at the LAC + USC Medical Center between January 1, 2000 and December 31, 2014. A previously established database for cervical cancer was used to identify those eligible cases.¹⁵ Cases with preinvasive cervical dysplasia, sarcoma, and metastatic tumors to the uterine cervix were excluded from the analysis. Among eligible cases, patient demographics, laboratory test results, tumor characteristics, treatment patterns, and survival outcomes were collected from the archived medical records. Strengthening the Reporting of Observational studies in Epidemiology guidelines were consulted to outline the study description for this retrospective cohort study.¹⁶

Clinical Information

Patient demographics at cervical cancer diagnosis included age, ethnicity, body mass index (BMI, kg/m²), medical comorbidities (hypertension, diabetes mellitus, and hypercholesterolemia), medication types, and cigarette use. Medication history was verified through multiple information sources. The dose of metformin per day at diagnosis was calculated by multiplying the dose and quantity of tablets dispensed per prescription. Tumor characteristics included histologic subtypes and cancer stage. Patterns for the initial treatment after cervical cancer diagnosis included use of primary hysterectomy, systemic chemotherapy, and whole pelvic radiotherapy (WPRT).

Definition

Obesity was defined as a BMI of greater than or equal to 30 kg/m².¹⁷ Daily metformin use was grouped per the median dose (≤1000 vs >1000 mg/d) in this study because of absence of previous data to support the survival significance. Cancer stage was based on the International Federation of Gynecology and Obstetrics classification.¹⁸ In this study, cancer stage was further grouped into the following: early-stage (stages IA1–IB1), locally advanced stage (stages IB2–IVA), and distant metastasis (stage IVB) per a previous study.¹⁵ Medication types were grouped as antihypertensive, antiglycemic, anticholesterol, analgesia, ant-acid, psychiatric, and other class agents. Progression-free survival (PFS) was defined as the time interval between the initial cervical cancer diagnosis and the date of the first disease recurrence/progression or the last date of-follow-up if censored. Overall survival (OS) was defined as time interval between the initial cervical cancer diagnosis and the date of death due to cervical cancer or the last date of follow-up if patient was alive or died of other causes.

Statistical Analysis

The primary interest of analysis was to determine the effects of metformin use on survival outcome of women with cervical cancer. The secondary interest of analysis was to examine the effects of metformin use on survival outcome of women with cervical cancer who received WPRT. Continuous variables were assessed for normality with Kolmogorov-Smirnov test, expressed with mean (SD) or median (range) as appropriate. Student t test, Mann-Whitney U test, or Kruskal-Wallis test were used to assess statistical significance for continuous variables as appropriate. Categorical variables were evaluated with Fisher exact test or χ² test as appropriate.

Log-rank test for univariate analysis and a Cox proportional hazards regression model for multivariate analysis were used for survival analysis. The association of metformin use and survival outcome was adjusted for a priori clinically known prognostic factors^10,19 including age (<60 vs ≥60 years), histologic subtype (squamous cell, adenocarcinoma, adenosquamous, and others), and stage (early, locally advanced, and distant metastasis). Magnitude of statistical significance was expressed with hazards ratio (HR) and 95% confidence interval (CI). Kaplan-Meier method was used to construct survival curves. A P value of less than 0.05 was considered statistically significant (all, 2-tailed). The Statistical Package for Social Science software (SPSS version 24.0; Ill (IBM Corp, Armonk, NY) was used for all analyses.

RESULTS

Patient’s Characteristics

There were 802 women with cervical cancer identified during the study period. Of those, 17 patients (2.1%) had no record for the initial information and were excluded. The remaining 785 patients were available for analysis of metformin use at the time of cervical cancer diagnosis and included 70 (8.9%; 95% CI 6.9–10.9) metformin users and 715 (91.1%) nonusers.

The patient demographics are shown in Table 1. For the entire cohort, mean age was 49.2 years, and most were Hispanic (71.8%). The most common medical comorbidity was obesity (39.3%) followed by hypertension (25.0%), diabetes mellitus (14.0%), and hypercholesterolemia (8.0%). Cigarette users were seen in 14.7%. The tumor characteristics and treatment patterns are shown in Table 2. The most common histology was squamous cell carcinoma (75.7%), and locally advanced disease was the most common cancer stage in this study population (56.8%). The most common primary treatment modality was WPRT (60.9%) followed by primary hysterectomy (27.6%) and systemic chemotherapy (9.0%). Across the drug classes examined in the study (Table 3), the most common type of medication was angiotensin converting enzyme (ACE) inhibitors (8.2%) followed by β-blockers (7.3%), diuretics (6.9%), and other nonsteroidal anti-inflammatory drugs (6.0%).

TABLE 1.

Demographics for cervical cancer patients

	All	Metformin (+)	Metformin (−)
Characteristics	785 (100%)	70 (8.9%)	715 (91.1%)	P
Age, y	49.2 (15.6–88.7)	55.9 (25.9–85.0)	47.8 (15.6–88.7)	<0.001
<60	414 (52.7%)	20 (28.6%)	394 (55.1%)
≥60	371 (47.3%)	50 (71.4%)	321 (44.9%)
Ethnicity				0.08
White	65 (8.3%)	0	65 (9.1%)
African	50 (6.4%)	4 (5.7%)	46 (6.4%)
Hispanic	564 (71.8%)	55 (78.6%)	509 (71.2%)
Asian	102 (13.0%)	10 (14.3%)	92 (12.9%)
Others	4 (0.5%)	1 (1.4%)	3 (0.4%)
BMI, kg/m2	28.0 (16.5–58.3)	28.1 (16.5–58.3)	27.6 (18.1–47.0)	0.59
<30	455 (60.7%)	41 (60.3%)	414 (60.8%)	0.94
≥30	294 (39.3%)	27 (39.7%)	267 (39.2%)
Hypertension				<0.001
No	589 (75.0%)	27 (38.6%)	562 (78.6%)
Yes	196 (25.0%)	43 (61.4%)	153 (21.4%)
Diabetes				<0.001
No	675 (86.0%)	2 (2.9%)	673 (94.1%)
Yes	110 (14.0%)	68 (97.1%)	42 (5.9%)
Hypercholesterolemia				<0.001
No	722 (92.0%)	48 (68.6%)	674 (94.3%)
Yes	63 (8.0%)	22 (31.4%)	41 (5.7%)
Smoker				0.59
No	663 (85.3%)	61 (88.4%)	602 (85.0%)
Yes	114 (14.7%)	8 (11.6%)	106 (11.6%)

Number (%) or median (range) is shown. The χ², Fisher exact, or Mann-Whitney U tests were used for P values. Significant P values are emboldened.

TABLE 2.

Tumor characteristics and treatment patterns for cervical cancer patients

	All	Metformin (+)	Metformin (−)
Characteristics	785 (100%)	70 (8.9%)	715 (91.1%)	P
Histology				0.46
Squamous cell	594 (75.7%)	51 (72.9%)	543 (75.9%)
Adenocarcinoma	137 (17.5%)	16 (22.9%)	121 (16.9%)
Adenosquamous	31 (3.9%)	1 (1.4%)	30 (4.2%)
Other	23 (2.9%)	2 (2.9%)	21 (2.9%)
Stage				0.13
Early stage	281 (36.0%)	25 (36.2%)	256 (36.0%)
Locally advanced stage	443 (56.8%)	35 (50.7%)	408 (57.4%)
Distant metastasis	56 (7.2%)	9 (13.0%)	47 (6.6%)
Primary hysterectomy				0.31
No	568 (72.4%)	47 (67.1%)	521 (72.9%)
Yes	217 (27.6%)	23 (32.9%)	194 (27.1%)
Radiotherapy (WPRT)				0.68
No	307 (39.1%)	29 (41.4%)	278 (38.9%)
Yes	478 (60.9%)	41 (58.6%)	437 (61.1%)
Chemotherapy				0.83
No	714 (91.0%)	63 (90.0%)	651 (91.0%)
Yes	71 (9.0%)	7 (10.0%)	64 (9.0%)

Number (%) is shown. The χ² or Fisher exact tests were used for P-values. Significant P values are emboldened.

TABLE 3.

Medication types and metformin in cervical cancer

	All	Metformin (+)	Metformin (−)
Characteristics	785 (100%)	70 (8.9%)	715 (91.1%)	P
Antihypertensive
ACE inhibitor	64 (8.2%)	24 (34.3%)	40 (5.6%)	<0.001
β-blocker	57 (7.3%)	11 (15.7%)	46 (6.4%)	0.012
Diuretics	54 (6.9%)	4 (5.7%)	50 (7.0%)	0.69
Ca-blocker	35 (4.5%)	5 (7.1%)	30 (4.2%)	0.23
ARB	13 (1.7%)	3 (4.3%)	10 (1.4%)	0.10
Antiglycemic
Sulfonylurea	36 (4.6%)	25 (35.7%)	11 (1.5%)	<0.001
Insulin	20 (2.5%)	6 (8.6%)	14 (2.0%)	0.006
Anticholesterol
Statin	43 (5.5%)	20 (28.6%)	23 (3.2%)	<0.001
Analgesia
ASA	44 (5.6%)	19 (27.1%)	25 (3.5%)	<0.001
Other NSAIDs	47 (6.0%)	1 (1.4%)	46 (6.4%)	0.11
Acetaminophen	40 (5.1%)	2 (2.9%)	38 (5.3%)	0.57
Opioid	40 (5.1%)	5 (7.1%)	35 (4.9%)	0.39
Antiacid
PPI	18 (2.3%)	6 (8.6%)	12 (1.7%)	0.003
H2 blocker	9 (1.1%)	2 (2.9%)	7 (1.0%)	0.19
Psychiatric
SSRI/SNRI	8 (1.0%)	1 (1.4%)	7 (1.0%)	0.53
Benzodiazepine	4 (0.5%)	0	4 (0.6%)	0.99
Other classes
Synthroid	22 (2.8%)	4 (5.7%)	18 (2.5%)	0.12
Steroid	13 (1.7%)	1 (1.4%)	12 (1.7%)	0.99
Antihistamine	5 (0.6%)	1 (1.4%)	4 (0.6%)	0.37
Warfarin	4 (0.5%)	1 (1.4%)	3 (0.4%)	0.31

Number (%) is shown. The χ² test was used for P values. Significant P values are emboldened. ACE indicates angiotensin converting enzyme; Ca-blocker, calcium channel blocker; ARB, angiotensin receptor blocker; ASA, aspirin; NSAID, nonsteroidal anti-inflammatory drug; PPI, proton pump inhibitor; SSRI, selective serotonin reuptake inhibitor; and SNRI, serotonin/norepinephrine reuptake inhibitor.

Metformin users were compared with the nonusers (Tables 1–3). Metformin users were more likely to be older, hypertensive, diabetic, and dyslipidemic compared with the nonusers (all P < 0.001; Table 1). Frequencies of obesity and cigarette use were similar between the metformin users and the nonusers (Table 1). Tumor characteristics and treatment patterns were also similar between the metformin users and nonusers (all P > 0.05; Table 2). Correlation between metformin use and other medication use was examined (Table 3). Metformin users were more likely to take ACE inhibitors, β-blockers, other antiglycemic agents (sulfonylurea and insulin), statins, aspirins, and proton pump inhibitors (all P < 0.05).

Survival Outcome of Metformin Users

Survival analyses based on metformin use were performed (Table 4). Median follow-up time for the entire cohort was 22.6 months. There were 236 cases (30.1%) of cervical cancer recurrence/progression and 163 women (20.8%) who died of the disease. Median follow-up time for the metformin users was similar compared with the nonusers (19.9 vs 22.4 months; P = 0.96). On univariate analysis, metformin users and nonusers had similar PFS (5-year rates, 57.3% vs 61.8%; P = 0.82; Fig. 1A) and OS (71.7% vs 70.7%; P = 0.86; Fig. 1B). Medical comorbidities, other medication types, and cigarette use were not associated with survival outcomes (all P > 0.05). On multivariable analysis, metformin use was not associated with PFS (adjusted HR, 1.11; 95% CI 0.70–1.74; P = 0.67; Table 4) or OS (adjusted HR, 0.91; 95% CI, 0.52–1.60; P = 0.75; Table 4) after adjusting for patient demographics and tumor characteristics. Among 478 patients who received WPRT as the primary therapy (metformin users, 41; nonusers, 437), metformin use was not associated with PFS (5-year rates, 55.1% vs 53.3%; P = 0.93; Fig. 1C) or OS (5-year rates, 65.8% vs 72.5%; P = 0.32; Fig. 1D). Among recurrent/progressed cases, metformin use was not associated with survival time after recurrence/progression (2-year rates, 58.1% vs 50.6%; P = 0.23).

TABLE 4.

Multivariate survival analysis for women with cervical cancer

		PFS Multivariate			OS Multivariate
Characteristics	No.	5-y (%)	HR (95% CI)	P	5-y (%)	HR (95% CI)	P
Age, y
<60	414	68.9	1		75	1
≥60	371	52.7	1.11 (0.85–1.46)	0.43	65.8	0.93 (0.67–1.29)	0.66
Histology
Squamous cell	594	60.7	1		69.5	1
Adenocarcinoma	137	66.3	1.04 (0.73–1.50)	0.82	79.1	0.81 (0.51–1.29)	0.38
Adenosquamous cell	31	60.3	1.42 (0.73–2.79)	0.31	75	1.17 (0.48–2.88)	0.73
Others	23	49.1	1.21 (0.66–2.19)	0.54	45.9	1.28 (0.68–2.42)	0.44
Stage
Early stage	281	90.1	1		95.1	1
Locally advanced stage	443	51.6	6.86 (4.29–10.9)	<0.001	63.3	9.58 (4.84–19.0)	<0.001
Distant metastasis	56	4.5	39.7 (22.9–68.6)	<0.001	8.3	67.6 (32.0–143.0)	<0.001
Metformin
No	715	61.8	1		70.7	1
Yes	70	57.3	1.11 (0.70–1.74)	0.67	71.7	0.91 (0.52–1.60)	0.75

A Cox proportional hazards regression model for multivariate analysis. All the listed covariates were entered in the final model. Significant P values are emboldened.

FIGURE 1.

Survival curves of cervical cancer based on metformin use. Log-rank test for P-values. Kaplan-Meier methods to construct survival curves were used for (A) PFS for all cases, (B) OS for all cases, (C) PFS for WPRT cases, and (D) OS for WPRT cases. No. indicates number; met, metformin.

Subanalysis was performed by stratifying patients based on the following diabetic status: (1) diabetic metformin users, 68; (2) diabetic metformin nonusers, 42; and (3) nondiabetic patients, 673 (2 patients who were metformin users but were not diabetic were excluded). Diabetic metformin users and diabetic metformin nonusers were more likely to be older, hypertensive, and dyslipidemic compared with the nondiabetic women (all P < 0.01; Supplementary Table 1, http://links.lww.com/IGC/A503). Obesity and cigarette use rates were similar across the 3 groups (Supplementary Table 1, http://links.lww.com/IGC/A503). Tumor characteristics and treatment patterns except for hysterectomy use were also similar among the 2 groups (Supplementary Table 2, http://links.lww.com/IGC/A503). Nondiabetic patients were less likely to take ACE inhibitors, β-blockers, statins, aspirins, and proton pump inhibitors compared with patients in the 2 diabetic groups (all P < 0.05; Supplementary Table 3, http://links.lww.com/IGC/A503). The 5-year PFS rates were similar across the groups: diabetic metformin use, 56.6%; diabetic metformin nonuse, 57.2%; and nondiabetic, 62.0% (P = 0.94). Similar findings were also observed for the 5-year OS rates: diabetic metformin use, 71.2%; diabetic metformin nonuse, 78.5%; and nondiabetic, 70.2% (P = 0.79).

Metformin users were further divided into the 2 groups based on the daily dose at diagnosis (≤1000 mg/d, n = 36; >1000 mg/day, n = 29). The patient demographics, tumor characteristics, and treatment patterns were also similar between the 2 groups (all P > 0.05; Supplementary Tables 4, 5 http://links.lww.com/IGC/A503 ). Daily metformin dose was not associated with PFS (P = 0.66) or OS (P = 0.58).

DISCUSSION

This study demonstrated that use of metformin was not associated with clinical outcomes in women with cervical cancer. Similarly, among women who received pelvic irradiation, metformin use was also not associated with survival. Hypothetical plausibility for this nonsignificant association of metformin use and cervical cancer deserves further discussion.

A recent study has reported that metformin use may significantly reduce the risk of developing cervical cancer compared with other antiglycemic agents among type II diabetic patients, especially when the cumulative duration of metformin use is longer than 2 years.⁸ Another study suggested that there is decreased cervical cancer–specific and overall mortality among diabetic women aged 66 years or older with prolonged cumulative duration of metformin use after cervical cancer diagnosis.¹⁰ However, interpretations of these studies are limited as they only include elder or diabetic patients, whereas the typical cervical cancer population is younger and nondiabetic.

Previous in vitro studies have been highlighting both direct and indirect antitumor effects of metformin.^20,21 A proposed schema is shown in Figure 2. The direct effects of metformin on cervical cancer cells are divided into 2 pathways: adenosine monophosphate-activated protein kinase (AMPK)-dependent pathway^11,13,14 and AMPK-independent pathway.¹² Indirectly, metformin may suppress cancer cell proliferation, which is stimulated by insulin and insulin-like growth factors through activation of phosphoinositide3-kinase (PI3K)-AKT signaling.^21,22 Metformin also indirectly inhibits inflammation through improving metabolic disturbances or through inhibiting the proinflammatory cancer-promoting nuclear factor κB and signal transducer and activator of transcription 3 pathways, which play a critical role in inflammation-mediated oncogenesis.²³ To date, there is no in vivo study reporting the results of metformin use in a cervical cancer model.

FIGURE 2.

Proposed schematic figure for the mechanism of metformin effects in cervical cancer. Systemic chronic inflammation caused by obesity leads to insulinemia and glycemia, which activate PI3K/MAPK signaling pathway implicated in oncogenesis and tumor progression. The AMPK/mTOR pathway plays a pivotal role in maintaining and promoting cancer cells. In obesity-related cancer, metformin suppresses the proliferation of cancer throughout inhibition of the indirect pathway (PI3K/MAPK signaling pathway) and the direct inhibition pathway (AMPK/mTOR pathway). However, in HPV-related cancer, metformin cannot sufficiently suppress the initiation and progression of cancer, although it may partially have inhibitory function throughout the indirect and direct inhibition pathway. The HPV oncoproteins, E6 and E7, largely overcome negative growth regulation by the host cell proteins, p53 and Rb, and induce genomic instability. The involvement of metformin on this pathway has not been well understood. IR indicates insulin receptor; IFG1-R, insulin-like growth factor-1 receptor; MAPK, mitogen-activated protein-kinase; AMPK, 5′ adenosine monophosphate–activated protein kinase; Rb, retinoblastoma.

The indirect tumor-inhibiting pathway of metformin may be more effective for obesity-related cancer such as colon,²⁴ breast,²⁵ and endometrium cancers (Fig. 2).²⁶ Increased body habitus is well correlated to increased mortality rates for all these 3 cancers,²⁷ and chronic inflammation associated with obesity has been known to be a major factor contributing to progression of these 3 cancers.^28,29 Adipose tissue is a complex endocrine organ that secretes a variety of both anti-inflammatory and proinflammatory cytokines, which cause a state of chronic systemic inflammation.²⁸ In this theory, obesity stimulates inflammatory pathways that promote tumor development, mainly by releasing proinflammatory cytokines, enhancing angiogenesis, inducing cell proliferation, suppressing immune system, and generating reactive oxygen species for DNA damage.^30,31

Unlike these well-known obesity-related cancers, obesity is not a risk factor for developing cervical cancer. Rather, persistent infection with high-risk oncogenic HPV is the major driving factor for developing cervical cancer.² The HPV oncoproteins E5, E6, and E7 are the primary viral factors responsible for the initiation and progression of cervical cancer, largely through overcoming negative growth regulation by the host cell proteins and by inducing genomic instability.³² On the other hand, chronic inflammation caused by obesity has not been considered as the major factor for the initiation trigger of cervical cancer. A recent retrospective study has shown that metformin use did not demonstrate beneficial effects on cause-specific survival in patients with head and neck cancer, another HPV-related cancer, compared with nonmetformin users and nondiabetics.^33,34 In addition, although obesity affects cancer mortality in both cervical and endometrial cancer,²⁷ the survival benefit of metformin was seen only in endometrial cancer (obesity-related cancer).²⁶ Collectively, these findings suggest that metformin may not be effective for antitumor effects against HPV-related cancer, and these previous findings in another HPV-related cancer partly support our results in that metformin use was not associated with survival of women with cervical cancer (Fig. 2).

Efficacy of metformin during radiotherapy has been tested in several cancers.^35–37 In vitro studies for lung, colon, liver, and prostate cancer have shown that metformin sensitizes cancer cells to ionizing radiation through activation of AMPK, enhanced apoptosis, decreased DNA repair, and improved tumor oxygenation.^35–37 On the contrary, the activation of AMPK by metformin may conversely contribute to radiation resistance in colon cancer cells.³⁸ Although metformin use seems to be beneficial for survival in patients with colorectal cancer and prostate cancer,⁹ our study of cervical cancer found that metformin use was not associated with survival among the subset of women who received WPRT. Although the exact reason is unknown, this lack of radio-sensitization effect of metformin in cervical cancer may be because of a difference in tumor biology compared with non-HPV related cancers.³⁹ A randomized, multicenter phase II study of standard chemoradiation in combination with metformin versus standard chemoradiation alone in women with locally advanced cervical cancer has been currently undertaken for patient accrual.⁴⁰ This study will provide more information about the effectiveness of metformin during radiotherapy for women with cervical cancer. In our results, among the subset of patients receiving WPRT, there was a nonsignificant trend for improved outcomes after 1 year associated with metformin, compared with the nonmetformin group (Fig. 1C).

A strength of this study is that it is one of the few studies that examined the survival effects of metformin use in cervical cancer. In addition, the sample size is relatively large among previous studies, and meticulous efforts were made by analyzing a large number of covariates. We are also aware of a number of weaknesses and limitations in the study. First, this is a retrospective observational study that may have missed possible confounding factors. For example, we were not able to abstract information regarding the duration of metformin use before the initial diagnosis of cervical cancer. In addition, there is a possibility that other drugs ameliorated the antineoplastic effects of metformin. Some antiglycemic drugs such as insulin, incretin-based therapy, and thiazolidinedione may have been involved with tumor progression,^41–43 although there was no significant association between use of their antidiabetic drugs and cervical cancer–specific mortality as shown in our study.¹⁰ There is a possible selection bias in this study; metformin is typically given to only diabetic patients, and it remains unknown if metformin use is associated with survival outcomes in a nondiabetic population. The prevalence of metformin use was less than 10% in this study. Because women with cervical cancer are generally young and relatively healthy compared with women with endometrial cancer, this is likely the cause of the observed lower prevalence of metformin use in the cervical cancer population. In addition, most of the patients in our study were Hispanic, and our results may not reflect a generalized population.

In summary, the direct oncogenic pathway (HPV-related signaling and AMPK signaling) and indirect obesity-related pathway (insulin-like growth factors signaling) may be the determinant factors to predict the outcome of metformin treatment across the cancer types. Further studies are warranted to elucidate this association of the direct/indirect pathways and metformin response in cancer patients.