Effect of metformin on outcomes of patients treated with immune checkpoint inhibitors: a retrospective cohort study

Abstract

Background

Immune checkpoint inhibitors have transformed the treatment landscape of cancer treatment, but only a fraction of patients responds to treatment, leading to an increasing effort to repurpose clinically approved medications to augment ICI therapy. Metformin has been associated with improved survival outcomes in patients undergoing conventional chemotherapy. However, whether metformin provides survival benefits in patients receiving immune checkpoint inhibitors (ICIs) is unknown.

Methods

We performed a retrospective cohort study at two tertiary referral centers in Taiwan. All adult diabetes mellitus patients who were treated with ICIs between January 2015 and December 2021 were included. The primary and secondary outcomes were overall survival (OS) and progression-free survival (PFS), respectively.

Results

In total, 878 patients were enrolled in our study, of which 86 patients used metformin and 78 patients used non-metformin diabetes medications. Compared with non-users, metformin users had a longer median OS (15.4 [IQR 5.6–not reached] vs. 6.1 [IQR, 0.8–21.0] months, P = 0.003) and PFS (5.1 [IQR 2.0—14.3] vs. 1.9 [IQR 0.7—8.6] months, P = 0.041). In a univariate Cox proportional hazard analysis, the use of metformin was associated with a reduction in the risk of mortality (HR: 0.53 [95% confidence interval: 0.35—0.81], P = 0.004) and disease progression (HR: 0.69 [95% CI 0.49—0.99], P = 0.042). The use of metformin remained associated with a lower risk of mortality after adjusting for baseline variables such as age, cancer stage, and underlying comorbidities (OS, HR: 0.55 [95% CI 0.34–0.87], P = 0.011). Similarly, the use of metformin was associated with a lower risk of disease progression. Importantly, the use of metformin before ICI initiation was not associated with a reduction in mortality (HR: 0.61 [95% CI 0.27—1.42], P = 0.25) or disease progression (HR: 0.69 [95% CI 0.33—1.43], P = 0.32).

Conclusion

The use of metformin is associated with survival benefits in patients undergoing immunotherapy. Prospective clinical trials are warranted to define the role of metformin in augmenting immunotherapy.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00262-022-03363-6.

Introduction

Immune checkpoint inhibitors (ICIs) are antibodies that target programmed cell death protein-1 (PD-1) or its ligand (PD-L1), and cytotoxic T-lymphocyte-associated-4 (CLTA-4), thereby activating the immune response against tumor cells [1, 2]. The development of ICI has substantially improved the outcomes of cancer patients and brought about a paradigm shift in the landscape of cancer therapy [3, 4]. However, most of the patients do not benefit from ICI therapy, representing an unmet need to improve patient response to treatment [5, 6].

One potential adjuvant drug theorized to improve patients’ response to ICI therapy is metformin, a biguanide class of antidiabetic drugs commonly used in the treatment of type 2 diabetes mellitus [7]. Metformin decreases serum glucose levels by increasing insulin sensitivity, which results in increased glucose uptake and decreased gluconeogenesis [7]. Other than its antidiabetic effects, metformin also affects cell growth and proliferation by lowering insulin levels in the body [8]. In both cell lines and mouse models, metformin has been demonstrated to inhibit cancer cell proliferation and delay tumor progression [8]. In clinical studies, metformin is associated with improved patient outcomes in patients undergoing conventional chemotherapy across different tumor types [9–11]. These observations have led to efforts to investigate whether metformin may be repurposed to augment and improve response to immunotherapy.

Preclinical studies have shown that metformin enhances the response to immunotherapy, potentially by decreasing the levels of PD-L1 expression and increasing tumor-infiltrating lymphocyte levels [12–14]. However, clinical observation studies have not been consistent on whether metformin is associated with survival benefits in patients treated with ICIs [15–19]. Of note, many of these cohort studies used patients without type 2 diabetes as control patients. This would represent a selection bias as patients who received metformin are expected to have type 2 diabetes mellitus. The presence of diabetes is a poor prognostic factor for cancer outcomes and is likely to have reduced the survival benefit of metformin in previous analyses [20]. Thus, whether metformin may improve the survival outcomes of patients undergoing immunotherapy remains inconclusive. In this study, we aimed to investigate the impact of metformin in patients receiving ICI therapy on clinical outcomes.

Methods

Study design

This retrospective cohort study was conducted at two tertiary referral centers in Taiwan. The ethics committee of both hospitals approved this study (Chung Shan Medical University Hospital CS2-21,183, Taipei Tzu Chi Hospital 11-X-035). We included all patients who were administered at least 2 cycles of immunotherapy between January 2015 and September 2021. We excluded patients with incomplete data, less than 20 years old, who received only 1 cycle of immunotherapy, and who did not have diabetes mellitus. We further excluded patients who did not receive metformin or other diabetes medications within 30 days of receiving ICI therapy. We collected data on patient characteristics and clinicopathological features, including age, sex, Eastern Cooperative Oncology Group (ECOG) Performance Status (ECOG-PS), cancer stage, cancer pathology, ICI type and treatment cycle, and underlying comorbidities. We also collected data on the type of diabetes medications prescribed (Supplemental Table 1).

In the primary analysis, patients who received metformin were compared to patients who received non-metformin diabetes medications. In the secondary analysis, patients who received metformin were compared to patients who did not receive metformin, regardless of the presence of diabetes. The non-metformin diabetes medications included the following: insulin, dipeptidyl-peptidase 4 (DPP-4) inhibitors, sulfonylureas, alpha-glucosidase inhibitors, and sodium-glucose co-transporter 2 (SGLT2) inhibitors (Supplemental Table 1).

Outcome definitions

The primary endpoint was overall survival (OS), and the secondary endpoint was progression-free survival (PFS). We adjudicated the tumor-based endpoints using a clinician and radiological-based validated approach because the Response Evaluation Criteria in Solid Tumors (RECIST) criteria were not feasible in a retrospective study based on electronic health record-derived data [21].

Statistical analysis

We evaluated the OS and PFS between metformin and non-metformin users using Kaplan–Meier analysis and Cox proportional hazard model analysis. The variables in the multivariate Cox model were selected a priori and included the following variables: age, gender, performance status, cancer stage, cancer type, class of ICI, hypertension, hyperlipidemia, chronic kidney disease, chronic obstructive pulmonary disease, ischemic heart disease, stroke, heart failure, surgery, and HbA1c. Missing data for HbA1c were addressed using multiple imputations [22]. We carried out two sensitivity analyses. First, we evaluated the relationship between the use of metformin before ICI initiation and oncological outcomes. Second, we investigated if the use of metformin is associated with improved survival benefits when compared with non-diabetes patients. A P value less than 0.05 for a two-sided test indicates statistical significance. All analyses were conducted using Stata version 16.0 (StataCorp LLC, College Station, TX).

Results

Patient characteristics

We identified 878 patients who met the inclusive criteria. We excluded patients with missing data, patients under the age of 20, and those who received only one cycle of ICI. We further excluded patients if they had no medical history of diabetes mellitus or if they did not receive diabetes medications within 30 days of ICI therapy (Fig. 1). A total of 685 patients (86 metformin users, 78 non-metformin users, and 521 non-diabetes patients) were subsequently enrolled in this study. The median ages of metformin users, non-metformin users, and non-diabetes patients were 66 (Interquartile range (IQR), 59–70), 66 (Interquartile range (IQR), 59–75), and 62 (Interquartile range (IQR), 54–69), respectively. The median HbA1c for the metformin and non-metformin users was 7 (6–8) and 7 (6–10), respectively. The most common cancer type was lung cancer (46%) and hepatobiliary cancer (18%). The baseline characteristics of all included patients are summarized in Supplemental Table 2.

Fig. 1.

Flowchart of patient enrolment

Survival outcomes

Compared with non-metformin users, metformin users had a longer median OS (15.4 [IQR, 5.6—not reached] vs. 6.1 [IQR, 0.8–21.0] months; P = 0.003) and PFS (5.1 [IQR, 2.0–14.3] vs. 1.9 [IQR, 0.7–8.6] months; P = 0.041) (Fig. 2 and Supplemental Table 3). In Cox proportional hazard model analysis, the use of metformin was associated with a 47% lower risk of all-cause mortality (univariate HR, 0.53 [95% CI 0.35–0.81, P = 0.004) and 31% lower risk of disease progression or death (univariate HR, 0.69 [95% CI 0.49–0.99, P = 0.042) (Table 1). These trends were also observed in multivariate Cox proportional hazard analyses adjusting for underlying characteristics and comorbidities (All-cause mortality, multivariate HR, 0.55 [95% CI 0.34–0.87, P = 0.011; disease progression or death, multivariate HR, 0.66 [95% CI 0.44–0.99, P = 0.045). Similarly, the use of metformin was associated with a lower risk of death and disease progression after adjusting for HbA1c levels (Table 1).

Fig. 2.

Survival outcomes of patients treated with and without metformin. A Overall survival, B Progression-free survival

Table 1.

Cox proportional hazards regression of all-cause mortality or disease progression

Analysis	Outcome	Univariate model HR (95% CI)	P value	Multivariate model 1 HR (95% CI)	P value	Multivariate model 2 HR (95% CI)	P value
Metformin versus non-metformin user	All-cause mortality	0.53 (0.35–0.81)	0.004	0.55 (0.34–0.87)	0.011	0.46 (0.30–0.69)	< 0.001
	Disease progression or mortality	0.69 (0.49–0.99)	0.042	0.66 (0.44–0.99)	0.045	0.57 (0.40–0.82)	0.002
Metformin before ICI initiation versus non-metformin before ICI initiation	All-cause mortality	0.61 (0.27–1.42)	0.25	0.49 (0.19–1.26)	0.14	0.39 (0.15–1.03)	0.057
	Disease progression or mortality	0.69 (0.33–1.43)	0.32	0.61 (0.27–1.40)	0.24	0.57 (0.24–1.32)	0.19
Metformin versus non-diabetes	All-cause mortality	1.06 (0.76–1.49)	0.72	1.01 (0.69–1.46)	0.98	–	–
	Disease progression or mortality	1.06 (0.76–1.49)	0.96	0.96 (0.71–1.29)	0.77	–	–

HR hazard ratio; ICI immune checkpoint inhibitor

In multivariate model 1, the following variables were used: age, gender, cancer stage, cancer type, Eastern Cooperative Oncology Group Performance Status, class of immune checkpoint inhibitors, surgery, underlying comorbidities that include hypertension, hyperlipidemia, chronic obstructive pulmonary disease, chronic kidney disease, ischemic heart disease, stroke, and heart failure

In multivariate model 2, the following variables were used: age, gender, cancer stage, cancer type, Eastern Cooperative Oncology Group Performance Status, class of immune checkpoint inhibitors, surgery, underlying comorbidities that include hypertension, hyperlipidemia, chronic obstructive pulmonary disease, chronic kidney disease, ischemic heart disease, stroke, and heart failure, and HbA1c. Missing values for HbA1c were addressed using multiple imputation

Sensitivity analysis

In a sensitivity analysis, patients who used metformin before ICI initiation had a similar median OS (3.9 [IQR, 1.7–not reached] vs. 1.8 [IQR, 0.3–19.7] months; P = 0.25) and PFS (3.0 [IQR, 1.3–7.1] versus 1.1 [IQR, 0.2–7.4] months; P = 0.31) compared with those who did not use metformin before ICI initiation (Supplemental Fig. 1 and Supplemental Table 3). The use of metformin before ICI initiation was not associated with a decrease in the risk of all-cause mortality and disease progression or death in both univariate and multivariate Cox regression analyses (Table 1).

In another sensitivity analysis, diabetes patients who were treated with metformin had a similar median OS (16.6 [IQR, 7.5–not reached] vs. 25.6 [IQR, 6.1–58.4] months; P = 0.72) and PFS (6.4 [IQR, 3.5–15.2] vs. 6.5 [IQR, 2.3–24.0] months; P = 0.96) compared with patients who did not have diabetes mellitus (Supplemental Fig. 2 and Supplemental Table 3). Compared with non-diabetes patients, the use of metformin was not associated with better survival outcomes (Table 1).

Discussion

In this cohort study, we report two important findings. First, patients who received metformin experienced a better survival outcome than patients who received other types of diabetes medications. Second, the use of metformin prior to the initiation of ICI was not associated with a better clinical benefit. Overall, our results suggest that metformin may potentially be used to augment and improve response to immunotherapy.

Most of the previous clinical studies have not found an association between the use of metformin and improved clinical outcomes among patients treated with ICIs [15–19]. Only one study conducted by Afzal et al. found a trend toward improved clinical benefits in patients receiving metformin [15]. An explanation for this is that the controls selected for comparison against metformin users were mostly patients without diabetes [15–19]. Patients who do not have diabetes mellitus generally have a better survival outcome than those who have diabetes mellitus [23]. Thus, the potential survival benefits associated with metformin might have been confounded or masked by the presence of diabetes in the metformin group. In line with this theory, our analysis found that metformin users had a similar OS and PFS when compared with non-diabetes patients, but a longer OS and PFS when compared with diabetes patients treated with non-metformin diabetes medications. We also adjusted for the baseline differences in HbA1c levels before the initiation of ICI. This was important because metformin is commonly used as first-line therapy for diabetes and patients who received metformin would tend to have better glycemic control than those who received other diabetes medications [24, 25]. Thus, our study provides robust data to support metformin’s role in augmenting and improving the response to immunotherapy.

There are several potential mechanisms as to why metformin might lead to improved survival over other diabetes medications. Metformin has been found to possess antitumor effects, possibly through activating the immune system to fight against tumor cells [8, 13]. In mice, metformin was found to increase the number of CD8 + tumor-infiltrating lymphocytes and protect these cells from apoptosis and exhaustion in the tumor microenvironment [8, 13]. Based on this mechanism, metformin might work synergistically with immune checkpoint inhibitors, which would activate and increase the antitumor activity of these T lymphocytes. Metformin also has cationic amphiphilic properties. Medications with these properties can permeabilize the lysosomal membrane and trigger cascades of events, resulting in plasma membrane rupture and cell death [26]. Alternatively, destabilizing the lysosomal membrane may also replenish PD-L1 expression, leading to an enhanced response to immune checkpoint blockade [27]. Consistent with this hypothesis, one previous study showed that patients treated with antihistamines possessing cationic amphiphilic properties experienced survival benefits over patients who were not treated with these antihistamines [27].

This study has several limitations. Because this was a retrospective study, we were unable to utilize the RECIST criteria to adjudicate the tumor-based endpoints. Nevertheless, we did use a validated approach that is appropriate for real-world data [21]. There might be residual confounders that were not included in the multivariate cox models used to adjust for the baseline differences between metformin users and users of other diabetes medications. However, we included most of the important confounders, including HbA1c, the key indicator of glycemic control in our models. The survival benefits associated with metformin were observed for patients with type 2 diabetes mellitus, and further studies would be required to investigate if these benefits are also seen in patients without diabetes mellitus. There was a substantial loss to follow-up in this study, which was inevitable given that this was a cancer population. Finally, this study was derived from an Asian cohort and therefore these observations might not be directly applicable to other populations.

Conclusion

In conclusion, the use of metformin is associated with improved patient survival and response to ICI therapy. Prospective clinical trials are needed to further clarify the role of metformin in augmenting immunotherapy.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

CSMUH

Chung Shan Medical University Hospital

ICI

Immune checkpoint inhibitor

TTCH

Taipei Tzu Chi Hospital

Author contributions

CMP and CHC contributed to study concept and design. CHC, YJC, CHC, CYC, YCC, SSW, XYS, CSH, CYP, and YPH contributed to acquisition of data. CHC, YJC, and CHC contributed to analysis of data. CHC, YJC, CHC, CHC, and CMP contributed to drafting of the manuscript. CHC, YJC, CHC, CYC, YCC, SSW, XYS, CSH, CYP, YPH, CMP, and CHC contributed to critical revision of the manuscript for important intellectual content. CMP and CHC contributed to study supervision.

Funding

No funding.

Data availability

The data that support the findings of this study are available on request from the corresponding author, CHu Chiang. The data are not publicly available as their containing information could compromise the privacy of research participants.

Declarations

Conflict of interests

We have no conflicts of interest to declare.