The Survival Effect of Metformin on Non-Small Cell Lung Cancer Treated with Chemotherapy: A Systematic Review

,


Introduction
Lung cancer remains the type of cancer causing most deaths in the world [1]. Tere were still a signifcant number of patients with the advanced stage at diagnosis, most of which were unresectable. Stage IV NSCLC patients could occupy from 2% to 9% of the total patients at diagnosis, while stage III patients could occupy nearly 20% [2,3]. Although surgery was recommended as the standard of care for most early-stage NSCLC from stage I to IIIB, the role of adjuvant therapy and systemic therapy remained critical in lung cancer management. Postoperative chemotherapy was recommended for early-stage patients undergoing surgery with negative margins but with high-risk characteristics, including poorly diferentiated tumors, vascular invasion, visceral pleural involvement, and unknown lymph node status [4]. Patients undergoing surgery and with positive margin or lymph node invasion (N1 or N2) were recommended to receive postoperative chemotherapy or concurrent chemoradiation therapy [4]. Chemotherapy or concurrent chemoradiotherapy was also recommended for NSCLC patients with unresectable advanced disease who are not eligible for targeted therapy or immunotherapy. Te survival benefts of chemotherapy required further improvements. Te median overall survival (OS) of chemotherapy in advanced NSCLC patients was reported to reach 10 months-26 months [5][6][7]. Te OS of concurrent chemoradiotherapy in unresectable NSCLC patients was from 23% to 32% as reported [8,9]. Studies pursuing improvement in survival benefts of chemotherapy has always been on the way.
Metformin, the most widely used antidiabetes agent in treating type 2 diabetes, has been extensively studied for its antitumor ability and addictive efects in combination of chemotherapy. Preclinical models have shown that metformin could activate adenosine monophosphate-activated kinase and inhibit mammalian target of rapamycin, subsequently suppressing protein synthesis and cancer cell growth and proliferation [10,11]. Tere have been clinical studies on the impact of metformin on the survival of NSCLC patients undergoing chemotherapy, either in retrospective or prospective design. But due to the diferences in study samples and intervention details, consistent results guiding on clinical applications are still lacking.
Herein, the objective was to conduct a systematic review on the impact of metformin on the survival of NSCLC patients undergoing chemotherapy.

Research Strategy.
Electronic literature search was conducted in PubMed, Embase, and Web of Science databases following the preferred reporting items for systematic reviews and meta-analyses guidelines (up to April 2023) [12]. Te search terms are shown in Table 1.

Study Selection.
Te inclusion criteria of the studies were as follows: (a) randomized controlled trials (RCTs) or observational studies; (b) patients with unresectable NSCLC; (c) chemotherapy (or chemoradiotherapy) with or without metformin was applied to treatment; and (d) the efect of metformin on the survival of unresectable NSCLC patients receiving chemotherapy was reported.
Te exclusion criteria were as follows: (a) studies with only abstract published; (b) among studies based on the same population, the study with the most recent or detailed data was selected; and (c) efect estimation of overall survival (OS) or progression-free survival (PFS) was not directly reported or could not be calculated from other supportive data. Two independent reviewers, Q Li and Q Fan, completed the selection process. Te disagreements were dealt through discussion.

Data Extraction and Quality Assessment.
Te data extraction and quality assessment of the included studies were independently conducted by two reviewers. Data of interest were as follows: frst author, publication year, study region, study year, study design, histological subtype of the NSCLC, tumor stage, treatment strategy, sample size, follow-up duration, diabetes status, OS and PFS, and HR of OS or PFS and their corresponding 95% CI. Te OS means the length of time that the patients live after the start of treatment. Te PFS means the length of time that the patients live until any progression of disease after the treatment. We also reviewed characteristics of study population including age, comorbidities, histology, metastasis, and stage at diagnosis. Te quality assessment tools were Cochrane collaboration's tool for RCT and the Newcastle-Ottawa scale (NOS) for observational studies, respectively. Te Newcastle-Ottawa scale is a star system that judges the quality of retrospective studies on three broad perspectives: the selection of the study groups, the comparability of the groups, and the ascertainment of either the exposure or outcome of interest. Two independent reviewers, Q Fan and J Wu, evaluated the quality of the included studies. Any discrepancies would be discussed with a third reviewer Q Li and resolved.

Search Results.
Te fow chart of the search process is shown in Figure 1. A total of 3399 records were identifed through initial database search, and 1930 records were removed due to duplication. Te titles and abstracts of the remaining 1469 records were reviewed for eligibility for our study, and 28 potential studies were fltered out. Ten, the full text of the 28 studies was carefully read, and 20 studies were excluded according to exclusion criteria. Finally, 11 studies were included in the systematic review [12][13][14][15][16][17][18][19][20][21][22].

Characteristics of the Included Studies.
Te characteristics of the 11 included studies are demonstrated in Table 2. All of the studies were published in or after 2010, and the study years were mainly in the past two decades. A total of 4606 patients were included into these studies, of which 1977 (42.92%) patients were treated with chemotherapy (or chemoradiotherapy) plus metformin and the others were treated with chemotherapy (or chemoradiotherapy) alone. Regarding the study design of the included studies, fve were RCTs and six were retrospective case-control studies. Patients of four studies were treated with chemoradiotherapy, while the others were treated with chemotherapy.
Characteristics of study populations are given in Table 3. For the ages of patients in the included studies, most had mean or median of age over 55 years. Five of the studies only included patients with diabetes mellitus, four studies only included patients without diabetes mellitus, and the other two studies had diabetic status mixed. Two studies declared to exclude patients with coexisting malignancies other than lung cancer [16,22]. Te comorbidities that were reported or excluded, the metastasis, and the stage at diagnosis are also shown in Table 3.  Table 1). Te six retrospective studies were of high quality with NOS scores of 7-8 (Supplementary Table 2).

Efect of Metformin
Addition on the OS. Ten of the eleven included studies reported the OS of patients. Of these, only Lin et al. and Chuang et al. reported signifcant diference between the metformin group and the nonmetformin group [14,22], in which the metformin group showed better OS. Te HR for OS of each included study is shown in Figure 2

Summary of Evidence.
Te impact of addition of metformin on the survival of advanced NSCLC patients treated with chemotherapy was inconclusive. While most of the studies included in our systematic review showed

Te Reason for the Addition of Metformin.
As reported in the former studies, the median overall survival (OS) of chemotherapy in advanced NSCLC patients reached 10 months-26 months [5][6][7]. Te explorations of new antineoplastic medications never stopped. Metformin, as a common antidiabetes medication, has been shown in the laboratory to have antineoplastic efects. Tere have been explanations for the mechanism of its antineoplastic efect; the activated AMPK and the suppressed mTOR pathway by metformin have been proposed to explain its inhibition in tumor growth and proliferation [10,11,23,24]. Te cytotoxic efects of metformin have also been demonstrated in preclinical studies, both in vivo and in vitro [25,26]. As a result, metformin was soon put into clinical application in the feld of cancer treatment. In addition to its antineoplastic efects on cancer patients, studies have shown metformin's ability to reduce cancer incidence [27][28][29]. Studies were subsequently performed on the synergistic efect on the survival of metformin with chemotherapy, radiation therapy, targeted therapy with tyrosine kinase inhibitors, and immune-checkpoint inhibitors.

Characteristics of Patients Receiving Metformin.
Chemotherapy was recommended to NSCLC patients who undergo surgery at an early stage but have high-risk features or with positive margins and lymph node invasion. Patients with an advanced stage of NSCLC but not eligible for targeted therapy were also recommended for chemotherapy [4]. Two of our enrolled studies involved patients at an early stage. Wen-Xiu et al. [16] studied NSCLC patients in stage I-IV (28 patients in I-II) for the addition of metformin in platinum-based chemotherapy. Ahmed et al. [13] studied patients in stage I-IV (25 in I-II) treating with platinumbased chemotherapy and radiation therapy. Both studies found no signifcant efect of metformin on survival, which might be explained by the limited sample size [16]. We were not able to perform subgroup analysis in terms of the stage since no data specifc to the stage were provided. Metformin was originally a type of antidiabetic drug. In our enrolled studies, NSCLC patients with diabetes treated with metformin made up a signifcant part of the study population. Our subgroup showed that metformin had signifcant benefts on the survival in patients with diabetes but no signifcant efect in patients without diabetes. It drove the puzzle of whether metformin's impact on survival was due to its efect on diabetes outcomes. However, studies have reported that the cause of death in patients was primarily cancer progression rather than competing risks due to diabetes      [14,19]. Meanwhile, all studies on NSCLC patients without diabetes did not conclude that metformin had a benefcial efect on survival. We look forward to further studies to solve the abovementioned puzzle on the metformin efect.

Timing and Dose of Metformin Use in Enrolled Studies.
Lin et al.'s and Chuang et al.'s studies [14,22] were the only two studies to identify a signifcant beneft of metformin on survival. In both studies, the study population was already on metformin at the time of lung cancer diagnosis. Study populations in the included retrospective studies were mostly taking metformin prior to lung cancer diagnosis. In those studies, metformin was more likely to be used as a baseline treatment than as add-on chemotherapy. It was unclear whether the efects of metformin had accumulated before chemotherapy or were synergistic with chemotherapy. Te included randomized controlled trials started metformin medication after randomization, though still weeks before the initiation of cytotoxic therapy [12,17,18].
Wang et al.'s study evaluated the survival beneft of diferent metformin doses during the frst 3 months after lung cancer diagnosis [19]. Tey found that a daily dose of at least 1500 mg and a cumulative defned daily dose of 21 or higher would bring improved survival outcomes [19]. Among the other three studies that also reported metformin doses, all three planned the full-day doses of 2000 mg [12,13,18]. Tere was less concern about the association between the dose and the survival beneft. We look forward to more concentration on the efect of the metformin dose on survival outcomes.

Impact on the Survival of Metformin Addition.
Two retrospective studies identifed signifcant benefts of metformin addition to survival of patients receiving chemotherapy [14,22]. But it was surprised to see that one of the included RCTs reported a signifcantly worse survival outcome of metformin addition. Tsakiridis et al. [12] included 26 patients in the metformin addition group while 28 in the control group. Although the sample size was somewhat limited compared to those two retrospective studies which reported a positive impact of metformin addition, Tsakiridis et al. still identifed signifcant diference between the two groups regarding both OS and PFS [12]. Except for the study design, one of the most prominent diferences was that Tsakiridis et al. included patients without diabetes, while studies of Lin et al. [14] and Chuang et al. [22] included patients with diabetes. Tere might be a hypothesis that metformin could bring a protective efect on survival to those patients with diabetes, which awaited further evidence.
4.6. Implications on Future Direction. Te exploration of the role of metformin in cancer treatment remained insufcient. In addition to lung cancer, metformin might have the potential for protective efects on other types of cancer. One study on female patients taking tamoxifen showed that addition of metformin in treatment reduced the tamoxifeninduced endometrial hyperplasia [30]. A randomized trial showed that addition of metformin to colorectal cancer patients following polypectomy could reduce the prevalence of metachronous colorectal cancer [31]. Further investigation into the protective efects of metformin on other cancer is warranted.

Limitations and the Risk of Bias.
Our results should be interpreted with some limitations. First, the fve enrolled RCTs were identifed as a high risk of bias. Tis was due to their open-label design since the medication of metformin was hard for blind design. Second, the age limitation, comorbidities, stage and the metastasis status at diagnosis, and the presence of other cancers were not consistent across all included studies, which could be the potential sources of heterogeneity. We reviewed those variables in a table to help comprehensively understand the population characteristics of each study. Tird, the diferent chemotherapy regimen among the enrolled studies was another great source of heterogeneity.

Conclusion
Our systematic review showed that the survival impact of metformin in combination with chemotherapy on unresectable NSCLC patients remains inconclusive. Te survival beneft of metformin addition might be more prominent in patients with diabetes. More prospective studies are needed to evaluate this efect.

Data Availability
Te data underlying this article will be shared on reasonable request to the corresponding author.