Long term follow-up of a large patient population with stage 2 and 3 esophageal cancer

The generally poor outcomes in the management of esophageal carcinoma may be largely attributed to the aggressive nature of the cancer and advanced disease at presentation. Though surgical resection has long been the standard of care for management of early stage disease, fewer than half of all patients presenting for treatment have surgically resectable cancer, and an even smaller proportion can undergo surgery with curative intent [2,3]. Studies and clinical practice have demonstrated the effectiveness of esophagectomy, though its benefi ts are more limited with advanced carcinoma [4]. A 2009 study of patients undergoing esophagectomy for stage I-III esophageal cancer showed fi ve-year survival rates of less than 50% for all but those with stage I disease [5]. Persistently poor patient outcomes have driven the use of chemotherapy and/ or radiation in the adjuvant or neoadjuvant setting. Previous studies suggest that chemoradiation may lead to similar or greater outcomes than surgery alone, but results remain controversial [6].


Introduction
Esophageal carcinoma is the sixth deadliest cancer worldwide, currently affecting more than 450,000 individuals [1]. In the United States, approximately 16,000 deaths occur each year. The incidence of esophageal cancer continues to rise for unclear reasons and cure rates remain low, with fewer than 20% of patients surviving 5 years after diagnosis [2].
The generally poor outcomes in the management of esophageal carcinoma may be largely attributed to the aggressive nature of the cancer and advanced disease at presentation. Though surgical resection has long been the standard of care for management of early stage disease, fewer than half of all patients presenting for treatment have surgically resectable cancer, and an even smaller proportion can undergo surgery with curative intent [2,3]. Studies and clinical practice have demonstrated the effectiveness of esophagectomy, though its benefi ts are more limited with advanced carcinoma [4]. A 2009 study of patients undergoing esophagectomy for stage I-III esophageal cancer showed fi ve-year survival rates of less than 50% for all but those with stage I disease [5]. Persistently poor patient outcomes have driven the use of chemotherapy and/ or radiation in the adjuvant or neoadjuvant setting. Previous studies suggest that chemoradiation may lead to similar or greater outcomes than surgery alone, but results remain controversial [6]. trimodality approach has been shown, with the 2006 CALGB 9781 trial of 56 patients demonstrating an improvement of more than 50% in overall survival rates when compared with surgery alone [9]. Similarly, the CROSS study published in 2012 demonstrated signifi cantly increased survival with neoadjuvant chemoradiation compared with surgery alone (49.4 months vs. 24 months). However, the EORTC clinical trial in 1997 and the TROG trial of 2005 showed no difference in survival with trimodality therapy when compared to surgical resection [7]. Several other studies investigating multimodal treatment have excluded patients with stage 3 disease or been limited by small sample size or a lack of statistical signifi cance [9]. As a result, the benefi ts of trimodality therapy have confl icting evidence, and concerns remain regarding both the potential toxicities posed by chemoradiation and what may represent an increased risk of postoperative morbidity with the addition of neoadjuvant therapy [11,12]. Lastly, in addition to these mixed studies, there is limited existing data on real world practice of effi cacy or in terms of morbidity and mortality.
The existing literature on esophageal cancer delineates a multitude of potential treatment strategies but a lack of consensus regarding optimal management of advanced disease, and there remain questions as to when trimodality therapy should play a role. In reviewing existing literature, the majority of prior studies assessing neoadjuvant or adjuvant chemoradiation have primarily focused on early stage esophageal carcinoma in patients who were able to undergo optimal surgical resection. In addition, these analyses have typically not stratifi ed patients by disease stage [8]. Furthermore, most previous studies have compared the outcomes of trimodality therapy or neoadjuvant chemotherapy or radiation with surgery alone. No previous esophageal cancer study has compared three different treatment regimens with a follow-up up to 21 years.
In this review, we assess the survival and treatmentrelated toxicity of stage II and III esophageal cancer patients, with the hypothesis that trimodality therapy may improve survival when compared with surgery alone, chemotherapy and surgery, or chemoradiation alone. The overall aim was to analyze three different approaches for the management of later stage esophageal cancer for which a standard of care has not been well-established.

Methods & analysis
Patient data collection IRB approval was granted by the University of Washington for this study. Data was gathered from a list of patients who presented to the University of Washington Medical Center (UWMC) or the Seattle Cancer Care Alliance (SCCA) in Seattle, WA for management of esophageal cancer between 1995 and 2016. After review of each patient's medical record to determine eligibility, a sample of 358 individuals was selected. Eligible patients had a diagnosis of stage II or stage III esophageal adenocarcinoma or squamous cell carcinoma and received treatment at UWMC or the SCCA involving surgery, chemotherapy, radiotherapy, or a combination of those therapies. Treatment decisions in most cases involved multidisplinary discussion, such as through tumor board discussion, involving radiation oncologists, general surgeons, and medical oncologists.
Patient information collected from medical records included age, gender, race, cancer histologic subtype, cancer stage at the time of treatment initiation, disease-free date if reached, date and location of cancer recurrence if applicable, treatment modality, pre-existing comorbidities, treatment toxicities, mortality status, and estimated cause of death. The date of death was gathered from either the Social Security Death Index or medical records. Cancer staging for each patient was reviewed in accordance with the 7 th edition of the AJCC Cancer Staging Manual. Patient comorbidities were quantifi ed using the Charlson Comorbidity Index, and toxicities graded from 1-5 in accordance with the Common Terminology Criteria for Adverse Events (CTCAE) version 4.

Classifi cation
After data extraction, eligible patients were stratifi ed by both stage of esophageal carcinoma and type of treatment received. Patients were divided into 3 treatment groups: trimodality therapy (group 1, T), surgery alone (group 2, S), and chemoradiation alone (group 3, CH + R).

Statistical analysis
We were interested in several statistical measures in this study. Patients with stage II and III disease were analyzed separately as two groups, as well as among the three treatment groups. Kaplan Meier survival curves were constructed for each group and stage to estimate median overall survival. A subset of patients who became disease-free after treatment was considered for progression-free survival and comparison through the log-rank test. The time metric for progressionfree survival was defi ned to be the time from disease-free date until time of relapse, death, or loss to follow-up.
Cox Regression analysis was performed by treatment group and stage. Multivariate analysis was used to determine whether patient gender, ethnicity, age at diagnosis, histologic subtype, and treatment toxicity grade had signifi cant effects on survival. Adjustment for medical comorbidities, histologic subtype, and the toxicity of each treatment plan was controlled for in multivariate analysis. Analysis was done on ethnicity as white vs. other.

Outcomes
The primary endpoint of this study was calculation of median overall and disease-free survival for each treatment group. We were also interested in survival differences between stage II and stage III patients. Secondary endpoints included assessment of treatment-related toxicity and the identifi cation of other variables potentially having a signifi cant impact on survival.

Patient characteristics
The analysis included 358 patients encompassing 7 ethnic groups and an age range from 31-91. The sample was 79.   A log-rank test was performed to compare T survival to the other three patient groups. By this calculation, the progressionfree survival difference of T patients versus S (p=0.18) and CH + R patients (p=0.56) was not signifi cant.

Cox regression analysis
Cox regression analysis was performed to determine whether the following variables signifi cantly infl uenced survival: treatment, gender, ethnicity, age at diagnosis, histologic subtype, comorbidity score, and treatment toxicity.
Groups S and CH + R were compared with the T group to determine differences in mortality risk by treatment. The S and CH + R groups had hazard ratios of 2.341 (p <0.001) and 2.739 (p<0.001), respectively, indicating greater mortality risk than the T group.
In addition to the aforementioned infl uence of cancer stage and treatment on survival, this analysis revealed that patient gender (HR=1.612, p=0.024) and treatment toxicity (HR=1.341, p=0.002) also signifi cantly affected patient outcomes. Males achieved lower survival than females, and higher toxicity grades led to increased risk of treatment-related mortality. Age at diagnosis, histologic subtype, and comorbidity score did not signifi cantly affect patient survival (Tables 4-6).

Regimens & toxicities
The specifi c chemotherapy regimens and radiation doses given to patients were similar across groups and stages.
Within the T group, the two most common toxicities were pain (69.7%) and pleural effusion (26.7%). 64.9% of patients in this group experienced a toxicity of 3 or higher. In the S group, the most common adverse events were pain (83.3%) and arrhythmia (23.8%). 69% of S patients experienced a treatment-related toxicity of grade 3-5 (Table 7). 5 patients died due to treatment-related toxicity, which included 2 patients in the T group (0.08%) and 3 S patients (7.1%). Causes of death were gastrointestinal bleed and sepsis in the T group, and ARDS with hemothorax, CVA, and aspiration pneumonia in the 3 surgical patients.
of different therapies on progression-free survival as all differences were not signifi cant when compared to T patients with the log rank test. Future studies may wish to investigate this further with a larger cohort of CH + R patients.
In addition to improved survival, our analysis showed that trimodality therapy appears to be relatively well-tolerated by patients. The incidence and severity of treatment-related toxicity among T patients was similar when compared to the other treatment groups. Our data suggest that CH + R patients may have experienced fewer therapy-related side effects overall, though the frequency and severity of adverse events was similar for all groups.
There were several demographic differences among the four groups that may have affected outcomes. As shown in Table 2, the T group had the lowest average age at diagnosis at 62 years, and the greatest proportion of adenocarcinomas (84.1%). We did fi nd that male gender is associated with a signifi cantly higher mortality risk, and there was a greater proportion of females in the CH + R group at 30.8%. Future studies may wish to investigate these potential confounding effects further.
Overall, our analysis shows that trimodality therapy may lead to signifi cantly improved overall survival without notably different toxicity than three other common treatment regimens. Additionally, we demonstrate that cancer stage expectedly impacts survival, with stage 3 patients achieving worse outcomes (HR=1.428, p=0.023). Our results also suggest that patient gender and toxicity score signifi cantly affect outcome, with males and those with higher grade treatment-  [9,[13][14][15][16]. The question of defi nitive chemoradiation without surgery has been suggested to have similar outcomes to trimodality therapy with decreased surgical morbidity, particularly in poor surgical candidates with squamous cell histology. However, survival was only assessed up to 3 years, making it diffi cult to assess long term disease-free survival [10]. Thus, our study adds support to the existing literature but includes signifi cant follow-up to better assess long-term outcomes. Finally, specifi c differences in treatment did exist within the four groups themselves with respect to chemotherapy regimens and radiation doses. Though the majority of patients received platinum-based combination chemotherapy (83.66%) and a cumulative radiation dose of 50.4 Gy (64.52%), differences in therapy cycles and doses could have affected survival within groups. Future randomized controlled trials could control for these differences and assess treatment groups with a greater degree of standardization.
As the eighth most common cancer globally and the sixth leading cause of cancer mortality, esophageal cancer represents an important public health concern [1]. For patients presenting with more advanced disease for whom surgery is unlikely to be curative, outcomes may be greatly improved with adjuvant and neoadjuvant therapies in a multidisciplinary setting. Our data suggest that trimodality therapy may accord the greatest overall survival benefi ts to stage 2 and 3 esophageal cancer patients without signifi cantly affecting the incidence and severity of treatment-related toxicities. Treatment modality, cancer stage, patient gender, and treatment toxicity appear to signifi cantly impact patient survival.
Although selection of optimal treatment remains subject to clinical decision-making and individual patient characteristics, further objective data in the form of randomized controlled studies and retrospective studies are necessary to continue improving and standardizing the care of more advanced esophageal cancer.