Effect of Long Term Prediagnostic Aspirin Intake on the Prognosis of Esophageal Squamous Cell Carcinoma Receiving Radical Surgery


 Background:The effect of long term prediagnostic aspirin intake on the prognosis of esophageal squamous cell carcinoma (ESCC) is unclear. We aimed to reveal the effect of long term prediagnostic aspirin intake on survival of ESCC patients receiving radical surgery. Methods:147 eligible ESCC patients who received radical surgery for primary treatment were enrolled in this study. Patients who had used aspirin regularly for more than 3 months before diagnose were classified as aspirin group and patients who had never used aspirin before diagnose and surgery were served as non-aspirin group. The recurrence rate, disease-free survival (DFS) and overall survival (OS) were compared between the two groups to verify the effect of aspirin. Results:Patients were clarified into aspirin group (n=57) and non-aspirin group (n=90). The DFS and OS were both significantly shorter in aspirin group than non-aspirin group (DFS: 23.1±18.0 months vs. 30.9±19.8 months, P=0.018; OS: 29.8±17.4 months vs. 35.2±18.2 months, P=0.082). Survival analysis revealed that OS decreased in aspirin group than in non-aspirin group, however, it did not reach significance (P=0.074). DFS decreased significantly in aspirin group than non-aspirin group in both univariate (P=0.007) and multivariate (P=0.002) survival analysis. Subgroup analysis revealed that in pTNM stage 2, OS and DFS were reduced in non-aspirin group compared with aspirin group (P=0.048 and P=0.003, respectively), while no difference was found in stage 3.Conclusions:Long term prediagnostic aspirin intake may cause poor DFS in ESCC patients receiving radical surgery, especially for those in pTNM stage 2.


Background
Esophageal cancer is a major health problem all over the world. The estimated new cases and deaths of esophageal cancer are 572,034 and 508,585 respectively in the worldwide in 2018 [1]. Esophageal cancer is the top 10 most frequently cause of cancer death in both male and female [1]. In China, esophageal cancer turned out to be the top 10 most commonly diagnosed cancer in both men and women (3rd in men and 6th in women) [2]. Esophageal squamous cell carcinoma (ESCC) is the major type and accounts for more than 90% of esophageal cancer in China [3]. The effective preventive methods and prognostic indicators are very necessary for ESCC.

Page 3/11
Aspirin is an analgesic and antipyretic agent which is widely used. Studies have indicated aspirin may reduce the incidence and mortality of certain cancers [4,5]. It was observed that the anti-tumor mechanisms of aspirin include COX dependent and COX independent pathways through inhibition of angiogenesis [6], induction of autophagy and apoptosis [6][7][8][9], anti-proliferative activity [10], and inhibition of metastasis [11][12][13]. U.S. Preventive Services Task Force has recommended aspirin as a primary preventive agent of cardiovascular disease and colorectal cancer in adults in 2016 [14]. However, increasing studies indicated the possible risk of clinical aspirin use which make the role of aspirin uncertain. A study including 6694 endometrial cancer patients with a maximum follow-up of 13 years did not found indication that pre-or post-diagnostic low-dose aspirin use reduced mortality for endometrial cancer [15]. Previous studies suggest that aspirin may reduce prostate cancer risk, while recent studies found aspirin was not associate with reduced prostate cancer risk [16,17]. A large Asian cohort study did not nd impact of aspirin on pancreatic cancer development [18]. Besides, there was no evidence that low-dose aspirin use before or after diagnosis was associated with a reduced risk of adverse outcomes overall in breast cancer [19,20].
It was reported aspirin could be used in esophageal cancer patients as an adjuvant chemotherapy following the standard surgery [21,22]. However, two large cohort studies in England contained 4654 esophageal cancer patients and 3833 gastric cancer patients showed that the cancer mortality proportions of participants surviving 1 year were similar in aspirin users versus non-users after diagnosis with cancer [23]. Another cohort of esophageal cancer patients indicated pre-diagnosis aspirin use was not associated with all-cause or cancer-speci c mortality but could increase the risk of interval metastatic disease [24]. Considering the con icting results of aspirin on esophageal cancer, this study aimed to investigate the effect of long term pre-diagnostic aspirin intake on the prognosis of ESCC.

Patients
We collected all newly diagnosed ESCC patients who received potential radical surgery for primary treatment in Qilu Hospital of Shandong University from 1 January 2010 to 31 December 2014. We selected all the patients who had regularly used aspirin for a long-term (≥ 3 months) before diagnose and surgery as the aspirin group and randomly chose another 90 ESCC patients who had never used aspirin or other COX-2 selective NSAIDs before diagnose and surgery as the control group. The reasons of aspirin intake were treatments for cardiovascular and cerebrovascular diseases, which were not the death reasons for the aspirin group. The American Joint Committee on Cancer TNM staging system was used in this study [25]. Undergoing radiotherapy and/or chemotherapy or not after the operation is not an exclusion criterion in this study. The exclusion criteria: lost to follow up, coexistence of other malignancies, resection not for curative intent, stage 0 disease, distant metastasis, non-cancer death. Ultimately, 147 patients were included in this study (57 in aspirin group, 90 in control group). The characteristics of patient, tumor, treatment and the nal outcome were retrieved from the Medical Records Room or by phone call follow-up. During the rst 2 years after surgery, follow-up visits were performed every 3 months. After that follow-up visits were performed every 6 months up to death or the end of the study (29 November 2016).

Study design
A hospital-based retrospective cohort study was performed. Patient-related characteristics (age, gender, smoking, drinking, prediagnostic aspirin intake) and tumor-related characteristics (tumor location, tumor length, differentiation grade, pathological tumor-node-metastasis (pTNM) classi cation, treatment method) were investigated. We used disease-free survival (DFS) and over-all survival (OS) to evaluate the prognosis. The de nition of DFS in this paper was the period of time from the radical operation to the identi able time for rst relapse or death or last follow-up, whichever rstly occurred. OS was from the radical operation date to the death date from tumor cause or last follow-up.

Statistical analysis
In data analysis of patient-related characteristics and tumor-related characteristics, Chi-square test and student's t-test were applied for categorical variables and continuous variables, respectively. Survival curves were constructed with the Kaplan Meier method to estimate the distribution of the DFS time and OS time in the two different groups. Cox proportional hazards modeling was used to investigate multiple risk factors that have been shown to in uence ESCC prognosis [95% con dence intervals (CIs)]. P values less than 0.05 were considered to be statistically signi cant. SPSS version 21.0 (SPSS Inc., Chicago, IL, USA) was used for all statistical analyses.

Results
Patients' characteristics 147 patients were recruited in this study, including 123 men and 24 women (49 males and 8 females in aspirin group, 74 males and 16 females in non-aspirin group, P = 0.550). Twenty patients were less than 65 years old and 37 patients were over 65 years old in aspirin group, while 47 less than 65 years old and 43 over 65 years old in non-aspirin group (P = 0.042). All other patient-related characteristics of aspirin group and non-aspirin group were not signi cantly different (all P > 0.05). Patients' characteristics are shown in Table 1.

Univariate and multivariate survival analysis results
In the univariate survival analysis, we used Kaplan-Meier curves to identify the effect of long term prediagnostic aspirin intake on OS and DFS. OS decreased in aspirin group than in non-aspirin group, however, it did not reach signi cance (P = 0.074). DFS decreased signi cantly in aspirin group than in non-aspirin group (P = 0.007). Signi cant factors of univariate survival analysis were included in the Cox proportional hazards model ( Table 2). The results showed that only pTNM stage was signi cant predictor for OS (2.496(1.742-3.575); P < 0.001). For DFS prediction, aspirin intake (HR, 0.502(0.325-0.777); P = 0.002) was signi cant. Besides, tumor location, differentiation grade, pTNM stage and treatment methods were signi cant as prediction biomarkers. Subgroup analysis of aspirin intake were further conducted according to pTNM stage. In pTNM stage 2, OS and DFS were reduced in non-aspirin group compared with aspirin group (P = 0.048 and P = 0.003, respectively), while no difference was found in stage 3 (P = 0.315 and P = 0.387, respectively). Subgroup analysis were not conducted in stage 1 and 4 for the patient number was less than 10 ( Fig. 1).

Discussion
Published evidence showed that aspirin may have protective effects for several kinds of cancers, such as lung cancer, prostate cancer, breast cancer, colon cancer, pancreatic cancer, esophageal cancer [26]. Especially in colorectal cancer, aspirin has been approved as a primary preventive agent in adults by U.S. Preventive Services Task Force in 2016 [14]. Contrary to our expected result, we observed that long term prediagnostic aspirin intake may cause poor outcome of ESCC patients who received radical surgery for the rst treatment. The recurrence rate and tumor-speci c mortality rate of aspirin group were both higher than non-aspirin group. The DFS and OS were both signi cantly shorter in aspirin group than non-aspirin group. It was observed that DFS of ESCC decreased signi cantly in aspirin group than non-aspirin group in both univariate and multivariate survival analysis. Long term prediagnostic aspirin intake decreased OS in univariate survival analysis while it was not signi cant. Subgroup analysis revealed that in pTNM stage 2, OS and DFS were signi cantly reduced in non-aspirin group compared with aspirin group, while no difference was found in stage 3 group. Previous studies conclusions were not consistent on the role of aspirin in esophageal cancer. The current points of aspirin effect on tumor were different. Some studies suggested to use aspirin as an adjuvant chemotherapy following the standard surgery [21,22], while other studies did not reveal the positive role of aspirin use [23]. Another cohort of esophageal cancer patients indicated pre-diagnosis aspirin use could increase the risk of interval metastatic disease. Multicenter prospective studies are necessary to reveal the role of aspirin plays in tumor.
Previous studies tried to investigate the potential mechanism of aspirin in cancer. It was observed that the anti-tumor mechanisms of aspirin include COX dependent and COX independent pathways through inhibition of angiogenesis [6], induction of autophagy and apoptosis [6-9], anti-proliferative activity [10], and inhibition of metastasis [11][12][13]. It was reported that post-diagnosis aspirin therapy improved overall survival of colorectal cancer patients, especially for patients with positive PTGS2 (COX-2) expression and mutated PIK3CA tumors [27]. Besides, aspirin suppresses growth in PI3K-Mutant breast cancer by activating AMPK and inhibiting mTORC1 signaling [28]. One study revealed the cytotoxicity and synergistic potential of aspirin and aspirin analogues towards esophageal and colorectal cancer [29]. However, we did not nd the clinical positive role of aspirin in esophageal cancer. In vivo and in vitro experiments are necessary to reveal mechanisms of aspirin in esophageal cancer.
The current work has several limitations. Firstly, the work was a retrospective cohort study; some of the results were based on the phone-call follow-up, which may in uence the accuracy. Secondly, the sample size was small which may bring bias on the results. We called on large sample, prospective, randomized controlled studies on this eld. Thirdly, due to the small sample size, we did not carry out a subgroup analysis according to the dose and frequency of aspirin intake.

Conclusions
In conclusion, our work found long term prediagnostic aspirin intake caused poor outcome of ESCC and reminded that aspirin cannot bene t all cancer patients.

Declarations
Competing interests: The authors declare that they have no competing interests.