The effect of metformin when combined with neoadjuvant chemotherapy in breast cancer patients

Metformin has been used to treat type 2 Diabetes Mellitus since long time. It has two proposed anti-neoplastic mechanisms, direct (insulin-independent) and indirect (insulin-dependent) actions. To assess the effect of Metformin on pathological response when combined with neoadjuvant chemotherapy in breast cancer. A prospective study included stage II, III non-diabetic breast cancer patients who received neoadjuvant chemotherapy in our center during the period from May 2017 to March 2019. 59 patients met our inclusion criteria and completed the study, 27 patients received 850 mg Metformin every 12 h with chemotherapy (group A), and 32 patients received chemotherapy without Metformin (group B). Pathological response was assessed by Chevallier classification and residual cancer burden score (RCB). Both groups were well balanced regarding baseline characteristics. The results of our study showed that the rate of pathological complete response (pCR) was 14.8% in group (A) vs. 6.3% in group (B) with a P value of 0.39. RCB class 3 was 40.7% in group (A) vs. 68.8% in group (B) which was statistically significant with a (P value of 0.031). Patients with triple-positive histology who had RCB class 3 were only (14.3%) in group (A) versus (60%) in group B. Patients with body mass index (BMI) ≥ 25 who had RCB 3 were 40% and 66.7% in group (A) and (B), respectively. Metformin may increase the pCR especially in patients with BMI ≥ 25 and patients with triple-positive histology, a larger phase III study is needed to confirm this finding.


Introduction
Breast cancer can be presented at an early stage or locally advanced breast cancer (LABC) [1]. The incidence of LABC is not common in rich and high-income countries. According to US National Cancer Database and European CONCORD study, LABC represents 9.6% and 4% of breast cancer patients, respectively. However, this percent jumps to 60% in developing and low-income countries [2][3][4][5]. The prognosis of LABC is as in early breast cancer, and it depends on clinical stage, tumor grade, and expression of ER, PR, and HER-2 receptors. But, the most important prognostic tool for long-term survival is the pathological complete response (pCR) after neoadjuvant chemotherapy (NACT). pCR is now accepted to be used in all cases where chemotherapy was indicated at diagnosis before surgery for the aim of improving DFS and OS [5]. Many systems were validated to assess the pCR. In general, the systems define two groups; a group with pathological complete response (pCR) and a group with little or no response. The classes of partial response in different systems range from 1 to 4 [6]. In almost all systems, a pCR requires invasive carcinoma absence in the breast. Likewise, the presence of residual ductal carcinoma in situ (DCIS) is considered also pCR as it does not alter survival [7]. In systems that include lymph nodes, the nodes must also be negative for a pCR [7].
Therefore, many trials were conducted to study the effect of adding different agents to neoadjuvant chemotherapy aiming for increasing the rates of pCR. One of the proposed agents that might have an effect on the pCR rate is Metformin.
Diabetes has been found to be a risk factor for breast cancer in some but not all studies [8]. A large meta-analysis that was done to demonstrate the association between breast cancer and diabetes showed that females with type 2 diabetes mellitus (T2DM) had a 23% higher risk of developing breast cancer in comparison to non-diabetic females [9]. The common factor linking diabetes, obesity, and metabolic syndrome to cancer could be the insulin resistance and the resulting hyperinsulinemia associated with these conditions [10][11][12]. Insulin and insulin-like growth factor (IGF) receptors form a complex network of cell surface receptors and all function to mediate insulin and IGF responses [13]. Most cancer cells have insulin and IGF-1 receptors. The insulin receptors are able to stimulate cancer cell proliferation and metastasis besides its metabolic function [14].
Metformin is a biguanide derivative and has been used for nearly a century to treat T2DM [9,15,16]. The proposed mechanisms of the anti-neoplastic effect of metformin involve both direct (insulin-independent) and indirect (insulin-dependent) actions of the drug as illustrated in Fig. (1) in the "supplementary materials." First, the indirect effects of metformin are associated with reduced circulating insulin levels [17]. As insulin has mitogenic and anti-apoptotic effects, the reduction in its level may be important to the anti-cancer mechanisms of metformin, especially in cancers associated with obesity and hyperinsulinemia, such as those of the breast and colon [17]. Moreover, certain cancers, like breast cancer, express high levels of the insulin receptor (IR), and high insulin level is associated with breast cancer recurrence and death [18][19][20]. Second, the direct effects of metformin include activation of AMP-activated protein kinase (AMPK) via phosphorylation on Thr172 by the tumor suppressor liver kinase B1 (LKB1) and a subsequent reduction in mammalian target of rapamycin (mTOR) signaling, protein synthesis, and cell proliferation [17,21].

Material and methods
We did a prospective randomized study that included 118 patients who were eligible for neoadjuvant chemotherapy during the period from May 2017 to March 2019 as illustrated in Fig. 1. The main inclusion criteria were pathologically proven stage II or III breast cancer females' ≥ 20 years, non-diabetic with adequate hematological, hepatic, and renal functions. Using Echocardiography, the ejection fraction had to be ≥ 50%. The main exclusion criteria were pregnancy, congestive heart failure, including severe hepatic or renal impairment, and patients with known hypersensitivity to metformin.
A written informed consent was taken from all patients after approval from the ethics committee in our University. "The Ethics Committee of the Faculty of Medicine-Alexandria University" is constituted and operates according to ICH GCP guidelines and applicable local and institutional regulations and guidelines which govern EC operation. We used AJCC 8th edition to assess both clinical and pathological stages of the patients.

Treatment plan
All patients received (1) AC regimen (4 cycles every 21 days): Adriamycin 60 mg/m 2 IV push and Cyclophosphamide 600 mg/m 2 IV infusion over 20-60 min or (2) AC-T regimen same AC regimen followed by paclitaxel 80 mg/ m 2 every week for 12 weeks. Group (A) received a chemotherapy regimen either AC or AC-T as mentioned above with Metformin 850 mg starting once daily for 1 week to adapt gastrointestinal symptoms and then twice daily, and Metformin was stopped at least 48 h before anesthesia according to FAD guidelines to avoid the risk of lactic acidosis [22]. Group (B) received chemotherapy only, the same regimens as group (A) but without Metformin.

Tolerability and side effects
We followed the tolerability and recorded the side effects of treatment protocol using Common Terminology Criteria for Adverse Events v5.0 (CTCAE) [23].

Clinical response
Clinical response was evaluated using RECIST criteria [24]. All patients underwent clinical examinations before each cycle with the exclusion of any patients who developed progressive diseases confirmed with radiological imaging. All patients had surgery 3-6 weeks after completing chemotherapy.

Pathological evaluation
All pathological specimens were reviewed blindly to assess the post-operative pathology and pathological response in both groups according to Chevallier classification and residual cancer burden (RCB). Macroscopic and microscopic analyses were performed to establish the Chevallier classification which includes four grades, grade 1 is the disappearance of all tumor on microscopic and macroscopic levels, grade 2 is the presence of DCIS in breast only while lymph nodes are free, grade 3 is the presence of invasive carcinoma with stromal alteration, such as sclerosis or fibrosis, and grade 4 is no or few modifications of the tumoral appearance. The definition of pCR encompasses both class 1 and class 2.
RCB score and class were determined using the MD Anderson Cancer Center's online calculator [25]. This system was developed to calculate RCB in breast in patients who received neoadjuvant treatment to predict distant relapse-free survival [26]. RCB system uses some variables including residual invasive tumor, average cellularity in the tumor bed, number of lymph nodes with metastases, and the size of the largest metastasis in the involved LNs, combined mathematically with a continuous index to define the four classes of RCB (RCB-0 through RCB-III). Although the system requires the use of a mathematical formula, a web-based calculation script is present for free at MA-Anderson website to calculate the scores [27].

Results
The 2 groups were well balanced as illustrated in Table 1. Around 44.4% in group (A) and 46.9% in group (B) had breast cancer stage IIIB at presentation, which reflects the high risk in our treatment cohort.
Luminal subtype was the most prevalent subtype representing 92.6% (25 patients) in group A and 81.3% (26 patients) in group B.
All patients in both groups received four cycles of AC regimen. Eight patients (29.6%) in group (A) and nine patients (28.1%) in group (B) received additional 12 weeks of paclitaxel. Pathological assessment post-neoadjuvant treatment: Pathological assessment of the patients after surgery showed that only four patients in group (A) (14.8%) and two patients (6.3%) in group B developed pCR and difference was not statistically significant P = 0.3. The difference between the two groups regarding other pathological responses in the T and N was not statistically significant as well. Final pathological stage is illustrated in Table 2.
When we applied the Chevallier system (Table 3), our data showed that 4 (14.8%) patients in group (A) were grade 1 and two patients (6.3%) in group (B) were grade 1 (P value 0.39), while 22 (81.5%) and 23 patients (71.9%) were grade 3 and in group (A) and (B), respectively (P value 0.38). In comparison, those with grade 4 which represents no change at all in the tumor were one patient (3.7%) and 7 patients (21.9%) in group (A) and (B), respectively, with a P value of 0.06, approaching significance in favor of the Metformin group "group A." When we used RCB classification (Table 4), we found that four patients (14.8%) and two patients (6.3%) in group (A) and (B) had CPR, respectively, with (P value 0.39).
Eleven patients (40.7%) in group (A) had the worst RCB class "class 3" in comparison to 22 patients (68.8%) in group (B) and the difference is statistically significant in favor of the group A "Metformin group" (P = 0.031) as shown in Fig. 2. The RCB index mean value in group (A) was 2.7, while in group (B) was 3.3 with (P value 0.053), also approaching significance in favor of the Metformin group. Moreover, when the patients were stratified by their molecular subtypes, we did not find a significant correlation between both groups in either luminal or HER-2-enriched subtypes. We did a subgroup analysis for HER-2-negative patients " Table 2 in the supplementary materials" and found that patients who had RCB class 1 and 2 were 56.3% in group (A) and 60% in group (B) which were not statistically significant (P = 1). Patients who had RCB class 3 were 81.8% in group (A) and 45.5% in group (B) (P = 0.067).
There was no significant correlation between both the type of received neoadjuvant chemotherapy and the initial clinical stage with the pathological stage at surgery in our sample.
In the whole sample, 42 patients received AC regimen without taxane, 17 (40.5%) of them had RCB classes 0, 1, and 2, while 25 of them (59.9%) had RCB class 3. Among the 17 patients who received taxane plus AC, nine patients (52.9%) had RCB classes 0, 1, and 2 and eight patients  (47.1%) had RCB class 3 with (P value of 0.38). We also tested the effect of metformin with the type of chemotherapy either AC only or AC + taxanes and we found that RCB class 3 was 42.1% in patients who received AC only versus 37.5% in patients who received AC + taxanes and RCB class 1 and 2 was 57.9 and 62.5% in AC group compared to AC + taxanes group, respectively. The difference was not statistically significant (P = 1). The safety and tolerability of Metformin showed that gastrointestinal adverse events, like diarrhea, nausea, vomiting, and epigastric pain, were slightly higher in group (A), yet the difference was not statistically significant. The adverse events in both groups are demonstrated in Table 1 in the supplementary materials.

Discussion
We did a prospective study to assess the effect of the addition of the oral hypoglycemic drug "Metformin" to conventional chemotherapy in the neoadjuvant settings using pathological response as the primary endpoint. In our study, the overall rate of pathological complete response was (10.16%). In the famous National Surgical Adjuvant Breast and Bowel Project (NSABP B27) trial the rate of pathological complete response was (13%) to those who received 4 cycles AC only and (26%) to those who received 4 cycles AC followed by 4 cycles docetaxel [28]. In a large meta-analysis of 12 trials  The main finding of our study is that the rate of complete pathological response was 14.8% (4 patients) in group (A) "Metformin group" versus 6.3% (2 patients) in group (B) "non-metformin group" with a P value of 0.39 which is comparable to what was published by Arce-Salinas et al. in 2016 [29] where they carried out a randomized doubleblinded trial which included (48) hormonal positive & Her-2-negative non-diabetic breast cancer patients. Twenty-eight patients received neoadjuvant chemotherapy (Anthracyclines and Taxanes) with Metformin 850 mg BID and 20 patients in the control arm received neoadjuvant chemotherapy without Metformin. The rate of PCR was 26.9% in the Metformin group versus 5.9% in the non-Metformin group with a (P value of 0.22) which is still not statistically significant. The rate of complete pathological response in the abovementioned study was numerically higher than that in our study as they used Taxanes in all patients and they did not include HER-2-positive subtype as we did.
The justification for including Her-2-positive patients is that Metformin has been shown to inhibit both the tyrosine kinase activity and the expression of the human epidermal growth factor receptor 2 (HER-2) in vitro models of HER-2-positive breast cancer cells [30][31][32][33]. Metformin treatment cause lowering of circulating levels of insulin and insulin-like growth factor (IGF-I) and cell-autonomous suppression of the "mTOR" pathway. Thus Metformin not only directly targets HER-2 receptor but also targets central mechanisms implicated in refractoriness to HER-2-targeted therapy, such as IGF-1 and mTOR pathway [34]. Data from "Asian Breast Cancer Database" showed that the diabetic patients who were receiving Metformin when breast cancer was diagnosed showed a better prognosis only if they had hormonal receptor-positive with HER-2 receptorpositive tumors [35]. Another important note, in an analysis of ALLTO which is phase III trial of HER-2-directed therapy in HER-2-positive patients where they randomized patients to receive 1 year of Trastuzumab alone, Lapatinib alone, their sequence, or their combination, they found that diabetic patients who were receiving Metformin had a statistically significant beneficial effect when their primary tumor was hormonal positive & HER-2 positive [36]. It is well known that Metformin causes a reduction of (Ki-67) level, especially in (HER-2 + ve) tumors [37]. Moreover, Andrea Decensi et al. found in their randomized pre-surgical trial of the effect of Metformin on breast ductal carcinoma in situ proliferation in comparison to a placebo, Metformin causes a 40% reduction of KI-67 level in the Metformin arm in (HER-2 + ve) DCIS particularly when they co-express estrogen receptor [22]. In our study in the Metformin group patients with HR+ & HER-2 + ve tumor had numerically better pathological response. Six patients (85.7%) had RCB classes 0, 1, and 2 versus only one patient (14.3%) among this group had RCB class 3 with a p-value of 0.18 which was not statistically significant due to the small number of our sample. In group (B) control group patients with the same tumor subtype (HR+ & HER-2+) did not show better pathological response as present group (A) (4 patients (40%) had RCB classes 0, 1, and 2 vs 6 patients (60%) had RCB class3). These results may confirm what was observed in previous studies that Metformin may have a good effect on breast cancer tumors with HR+ & HER-2+ subtype and this observation warranted more prospective trials to confirm it.
When we used the RCB scoring system in the analysis of the pathological response and combined class (RCB 0/1/2) versus RCB class 3, we found that RCB class 3 was 40.7% in group (A) versus 68.8% in group (B) which was statistically significant with a P value of 0.031. We used RCB class 0/1/2 vs 3 based on what was published by Campbell et al. [38] where they used residual cancer burden classes as a predictor of recurrence risk and found that RCB class 3 significantly predicts high recurrence risk after adjusting for age, clinical stage, and hormonal receptor status [38].
The maximum effective dose of Metformin used in patients with T2DM to treat hyperglycemia is 1000 mg twice daily. However, the Metformin dose that achieved the maximal anti-cancerous effect is unclear. Results from two xenograft models preclinical studies reported that the human equivalent of 1500-2250 mg/day was needed to inhibit tumorigenesis [39][40][41]. Kim et al. in phase II randomized trial (METEOR) trial of neoadjuvant Metformin plus Letrazole versus placebo plus Letrazole (still an ongoing trial no results published yet) used the dose of 1000 mg BID [42]. Whereas the ongoing NCIC CTG MA32 phase III clinical trial is testing the effect of adjuvant Metformin, utilizing a dosage of 1700 mg/day. It is the most advanced adjuvant trial investigating the effect of Metformin versus placebo on invasive disease-free survival and other outcomes in 3649 women [43]. Arce-Salinas et al. [44] and Martin-Castello et al. [45] as mentioned earlier used the same dose we utilized (1700 mg/day).
Our study confirms that Metformin is a tolerable and safe addition to chemotherapy, and the dropout rate in Metformin arm was (11.8%). Seven out of 59 patients in the Metformin arm withdrew because of Metforminrelated gastrointestinal upset while in the METTEN study the dropout rate in the Metformin arm was 13% (5 out of 38 patients).
A major limitation in our study is the lack of using anti-HER-2 drugs with neoadjuvant chemotherapy in Her-2-positive patients due to lack of resources.
Another limitation in our study is not using Ki-67 testing before and after chemotherapy as Ki-67 is a good marker to be used to see the effect of investigational new drugs in the neoadjuvant setting. Finally, the lack of correlation of pathological complete response to overall survival.

Conclusion
Metformin can increase the rate of pathological response when combined with chemotherapy. We found that adding Metformin to neoadjuvant chemotherapy resulted in better pathological response using residual cancer burden score. Patients with hormone receptor and Her-2-positive tumors may have the most benefit of adding metformin but a larger trial is needed to confirm this finding.
Metformin is a safe addition to neoadjuvant chemotherapy in breast cancer patients and is well tolerated.
Author contributions SME assisted in conceptualization, methodology, validation, investigation, resources, data curation, and writing and preparation of the original draft. MA involved in reviewing & editing of the manuscript and supervision. DA assisted in investigation and resources. AGG, AME, and Omar Shebl Zahra performed supervision and project administration.
Funding No funding was received to assist with the preparation of this manuscript.

Data availability
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Conflict of interest
The authors have no relevant financial or non-financial interests to disclose.
Ethical approval Informed consent was taken from all patients after approval from the ethics committee in our University. "The Ethics Committee of the Faculty of Medicine-Alexandria University" is constituted and operates according to ICH GCP guidelines and applicable local and institutional regulations and guidelines which govern EC operation.
Informed consent Informed consent was obtained from all individual participants included in the study.