Angiotensin II Receptor Blockers Improve Survival in Patients With Locally Advanced Oral Squamous Cell Carcinoma- A Propensity-Score-Matched Study

Objectives: Angiotensin II receptor blockers (ARBs) improve the survival rates of patients with various cancers. However, it remains unclear whether ARBs confer a survival benet on patients with oral squamous cell carcinoma (OSCC). Here, we assessed the associations between ARB use and survival in patients with OSCC of different stages. Materials and Methods: This was a 10-year retrospective cohort study of OSCC patients. We enrolled 7,763 patients diagnosed with oral cancer between January 2007 and December 2017 whose details had been entered into the Chang Gung Research Database. A total of 930 patients receiving surgery were recruited from the Chang Gung Research Database after performing 1:4 propensity score-matching between ARB users and non-users. Cox’s regression models with adjusted covariates were employed to detect factors inuencing the survival rates of patients with OSCC. Results: Kaplan-Meier analysis revealed that the overall survival(OS) and disease-specic survival (DSS) rates of 180-day ARB users increased. Cox’s regression models indicated that ARB use, early-stage OSCC, and only surgical intervention were independently prognostic of improved OS and DSS. An increased survival rate was also observed in 180-day ARB users in the stage III, Iva, and IVb categories. Conclusions: ARB use for more than 180 days is associated with an increased survival rate and is a positive, independent prognostic factor in patients with OSCC. These ndings highlight the clinical usefulness of ARBs in OSCC patients with advanced disease.


Introduction
Oral cancer is one of the most frequently occurring cancers worldwide. Oral squamous cell carcinoma (OSCC) represents the most common type of oral cancer, constituting approximately 90% of all oral cancers 1 . In 2018, more than 355,000 individuals were diagnosed with oral cancer worldwide, and approximately 177,000 oral cancer-related deaths were reported 2 . Despite advances in surgical techniques and chemoradiotherapy, the prognosis of patients with OSCC remains unsatisfactory, especially for those diagnosed with advanced disease. Therefore, the identi cation of novel therapeutic targets in OSCC is of high clinical importance.
The renin-angiotensin system (RAS) is involved in the regulation of blood pressure. Therefore, angiotensin I-converting enzyme inhibitors (ACEIs) and angiotensin II type 1 receptor blockers (ARBs) are among the most widely used anti-hypertensive drugs. A retrospective cohort study conducted by Lever and colleagues showed that the long-term ACEI use protected against cancer 3 , suggesting that the local RAS played roles in tumor development and progression. Additionally, the RAS has been implicated in most human cancers; thus, the use of ACEIs/ARBs has been proposed as a promising anti-tumor strategy, which could potentially suppress tumor progression through inhibition of cancer cell proliferation and neovascularization 4 . Indeed, the combination of ACEIs/ARBs with conventional anti-cancer therapies has been shown to improve clinical outcomes of patients with various types of cancer, including lung, urothelial, breast, and esophageal cancers 5-8 . However, the clinical usefulness of RAS inhibitors in patients with OSCC remains unclear. Also, most previous studies did not separately evaluate the anti-neoplastic effects of ACEIs and ARBs; the drug classes were combined when exploring the clinical outcomes of cancer patients. The effects of ARBs alone were inconsistent 9,10 , suggesting that the ACEIs exerted all of the observed anti-neoplastic effects. Thus, we investigated the e cacy of ARBs in patients with OSCC receiving surgery. We used the Chang Gung Memorial Hospital database to perform a 10-year, retrospective cohort study. Furthermore, we explored the effects of ARBs on patients with advanced-stage OSCC.

Study cohort
We enrolled 7,763 patients diagnosed with oral cancer between January 2007 and December 2017 whose details had been entered into the Chang Gung Research Database [11][12][13] . Figure 1 is a ow chart of the cohort study design for statistical analysis. Among the 7763 oral cancer patients, a total of 1560 patients were not recruited after applying the exclusion criteria. We performed propensity score-matching to balance covariates between ARB users and non-users. Hence, data from 930 patients were analyzed in our study, including 186 patients treated with ARBs and 744 matched patients who did not receive ARBs. This retrospective cohort study was approved by the Institutional Review Board (IRB) of Kaohsiung and Chiayi Chang Gung Memorial Hospital (approval nos. 202001463B0 and 201700253B0C602), and all experiments were performed in accordance with relevant guidelines and regulations. The Chang Gung Medical Foundation IRB (approval nos. 202001463B0) approves the waiver of the participants' consent.

Statistical Analysis
Categorical data (e.g., sex, comorbidities, lifestyle risk factors, cancer sites, and AJCC stage) were analyzed using a two-sided Fisher's exact test or a two-sided Pearson's chi-squared test. Parametric and non-parametric continuous data were analyzed using Student's t-test and the Mann-Whitney U test, respectively. To minimize the confounding effects due to non-randomized allocation, data were analyzed from a 1:4 propensity score-matched cohort (ARBs vs. nil), which had been identi ed by the Greedy method with a 0.25 caliper width using NCSS software, version 10 (NCSS Statistical Software, Kaysville, UT, USA). Propensity scores were calculated using a logistic regression model with sex, age, pathological AJCC stage, and the diagnostic year of OSCC as covariates (Table S1). In ARB users, we calculated the survival time from the day of OSCC diagnosis if ABR was already used or from the day of starting ARB use if the patient had not used it after OSCC diagnosis. Because the diagnostic years were matched in both groups, the calculated survival time in non-users started from the same day as its match to make the comparison between the two groups fair. After adjustment for these confounding factors, the Kaplan-Meier method and the log-rank test were used to evaluate the effects of ARBs on the primary outcome. The correlations with different comorbidities were evaluated by Pearson's correlation coe cient to prevent multicollinearity before building a regression model ( Figure S1). A Cox proportional hazards model tested the dependence of primary factors on other prognostic factors in multivariate survival modeling. A strati ed analysis was performed and adjusted to analyze the e cacy of ARBs in patients at different pathological stages. All statistical analyses were performed using SPSS Statistics for Windows, version 22 (IBM Corp., Armonk, NY, USA). P-values < 0.05 were considered statistically signi cant.

Results
Baseline clinicopathological characteristics of the study cohort are summarized in Table 1. Of the 930 patients with OSCC, 92.7% (n = 862) were men and 7.3% (n = 68) were women. The median age at diagnosis was 58 years, and the mean follow-up time was 62.4 months. Tongue (34.8%) and buccal mucosa (34.7%) were the most common tumor sites. Five hundred and eighty-two patients (62.6%) had early-stage cancer (stage I or II), while the remaining 348 patients (37.4%) had advanced stage tumors (stage III, IVA, or IVB). Six hundred and eight patients (65.4%) underwent surgery alone, and 322 (34.6%) underwent surgery plus adjuvant radiotherapy or concurrent chemoradiotherapy. At the end of the study period, 256 (27.5%) patients had died; 133 of these (14.3%) had died of primary head and neck cancer.  (Table 2). In this 10-year cohort study, patients receiving ARBs for more than 180 days exhibited a signi cantly higher OS rate, compared to those who did not receive ARBs (Fig. 2). Similarly, the DSS rate was signi cantly higher in patients receiving ARBs than in those who did not receive ARBs (Fig. 3).  Tables 3 and 4). We also evaluated the effect of ARBs on the survival rates of advanced OSCC patients in the T and N categories. The ARB treatment outcomes for OSCC patients in advanced T (T3 and 4) or either N category in terms of OS rate were signi cantly improved (Table S2). Overall, these strati ed analyses suggested that patients with late-stage OSCC were the most likely to bene t from ARB use for more than 180 days after OSCC diagnosis.

Discussion
To the best of our knowledge, this was the rst study to investigate the potential clinical bene t of ARBs in patients with OSCC receiving surgery. In this 10-year retrospective cohort study, patients who received ARBs for at least 180 days had improved OS and DSS, compared to patients who did not receive ARBs. In addition, patients with locally advanced OSCC experienced the greatest bene t from ARBs.
The RAS consists of several enzymatic and non-enzymatic protein components and is essential for maintenance of vascular homeostasis. Angiotensinogen is produced in the liver and cleaved by the aspartyl protease renin to angiotensin I. Angiotensin I is subsequently cleaved by the angiotensin Iconverting enzyme to produce angiotensin II (Ang II). Ang II is a key component of the RAS, which exerts its actions by binding to two G protein-coupled receptors: angiotensin receptor 1 (AT1R) and the lesser known angiotensin receptor 2 14 . It is increasingly evident that, in addition to systemic effects on blood pressure and uid homeostasis, AT1R and Ang II have important roles at the local tissue level. AT1R overexpression has been reported in numerous cancers, including ovarian, breast, and bladder cancer 15,16 . Consistent with these ndings, ARBs and ACEIs have been reported to reduce tumor growth and vascularization in a wide range of cancers, suggesting a role for Ang II in cancer development and Several clinical studies have suggested a role for the RAS in terms of the effectiveness of anti-cancer therapies [17][18][19] ; however, the clinical utilities of ARBs alone have not been reported. Although a nationwide population-based study showed that long-term ARB use was associated with a lower incidence of cancer 9 , a meta-analysis of randomized controlled trials found that ARBs were associated with a modest increase in cancer risk 10 . The ARB Trialists Collaboration study and a recent metaanalysis concluded that ARB use had no effect on cancer incidence 20,21 . Brie y, the varying extent of concurrent ACEI use among the studies may explain the discrepancies. We evaluated patients who had received ARBs only (Table S3).
Head and neck cancer is the sixth most common cancer worldwide 22 . The critical role of RAS in head and neck cancer has been shown in various tissues, including the oral mucosa 23 . Additionally, the ACEI perindopril has been shown to reduce the growth of head and neck squamous cell carcinoma (HNSCC) in vivo 24 , suggesting a role for Ang II in HNSCC. Ang II also has been found to promote HNSCC cell migration and invasion 25 . The effects of Ang II on autocrine and paracrine signaling pathways are mediated by AT1R, suggesting that ARBs might provide a clinical bene t in patients with HNSCC.
However, no large study has yet shown that ARBs improve the clinical outcomes of patients with headand-neck cancer.
We found that ARB use for at least 180 days improved the 10-year OS and DSS rates of patients with OSCC ( Figs. 2 and 3); the effects were most pronounced for patients with late-stage resectable OSCC (Tables 3 and 4). This was also true of subgroups of patients de ned via TN staging; ARB use was associated with increased survival of patients with advanced T (T3 and 4) category (Table S2). As T refers to local tumor containment, the data suggest that ARBs are bene cial for patients with locally advanced OSCC. It remains unclear how ARBs affect OSCC progression and improve survival rates. An prior in vitro study might explain this effect: co-injection of cancer cells with stromal cells increased tumor size and brosis; ARB treatment attenuated these effects 26 .
Our study had a few limitations. First, medical records were incomplete for some patients. Therefore, some critical clinicopathological characteristics (e.g., surgical margin, extranodal extension, and depth of tumor invasion) could not be analyzed in our study. However, the large cohort size supported the validity of conclusions drawn for the assessed clinicopathological characteristics. Second, although our study included only data from patients treated with ARBs, a considerable proportion of the patients were receiving concurrent treatment with other agents to control hypertension, which might have in uenced our ndings. Several patients received amlodipine and hydrochlorothiazide (Table S3); however, there is no evidence that these agents suppress cancer development or progression 27 28 . Therefore, we presume that treatment with these agents had a minimal in uence on our ndings. Third, various ARBs were included; therefore, the standardized effective dosage was di cult to calculate and the dose effect could not be measured. On the other hand, the use of propensity score matching (PSM) and the large database were highlights of our study. PSM attempts to reduce the bias due to confounding variables that can arise if the effects of treatment are assessed simply by comparing outcomes between patients that received the treatment versus those that did not. We created a 1:4 propensity-score-matched study group to minimize any confounding effect of non-randomized allocation when comparing the groups.  Figure 2 Kaplan-Meier survival curve of OS rates between angiotensin II receptor blockers users (≥ 180 days) and non-users. The estimated 5-and 10-year OS rates of ARB non-users (None) were 62.7% and 55.9%, respectively. The estimated 5-and 10-year OS rates of ARB users (≥180 days) were 79.5% and 61.0%, respectively. ARBs, angiotensin II receptor blocker; OS, overall survival Figure 3 Kaplan-Meier survival curve of DSS rates between angiotensin II receptor blockers users (≥ 180 days) and non-users. The estimated 5-and 10-year DSS rates of ARB non-users (None) were 74.6% and 70.2%, respectively. The estimated 5-and 10-year DSS rates of ARB users (≥180 days) were 85.4% and 85.4%, respectively. ARBs, angiotensin II receptor blockers; DSS, disease-speci c survival