PROGNOSTIC FACTORS AND SURVIVAL ANALYSIS IN ESOPHAGEAL CARCINOMA

ABSTRACT Background: Despite recent advances in diagnosis and treatment, esophageal cancer still has high mortality. Prognostic factors associated with patient and with disease itself are multiple and poorly explored. Aim: Assess prognostic variables in esophageal cancer patients. Methods: Retrospective review of all patients with esophageal cancer in an oncology referral center. They were divided according to histological diagnosis (444 squamous cell carcinoma patients and 105 adenocarcinoma), and their demographic, pathological and clinical characteristics were analyzed and compared to clinical stage and overall survival. Results: No difference was noted between squamous cell carcinoma and esophageal adenocarcinoma overall survival curves. Squamous cell carcinoma presented 22.8% survival after five years against 20.2% for adenocarcinoma. When considering only patients treated with curative intent resection, after five years squamous cell carcinoma survival rate was 56.6 and adenocarcinoma, 58%. In patients with squamous cell carcinoma, poor differentiation histology and tumor size were associated with worse oncology stage, but this was not evidenced in adenocarcinoma. Conclusion: Weight loss (kg), BMI variation (kg/m²) and percentage of weight loss are factors that predict worse stage at diagnosis in the squamous cell carcinoma. In adenocarcinoma, these findings were not statistically significant.


INTRODUCTION
D espite recent advances in diagnosis and treatment, esophageal cancer still has high mortality. Mean survival for squamous cell carcinoma (SCC) is 13.95±SD 11.2 months and for esophageal adenocarcinoma (EA) is 13.22±SD 10.23 months 4,11,13 .
Prognostic factors associated with patient and with disease itself are multiple and poorly explored. Knowing these parameters can allow a better stratification of high-risk groups 2,3 .
This study aims to assess demographic, clinical and pathological factors in esophageal cancer patients that impact in overall survival and prognostic.

METHODS
This study retrospectively reviewed esophageal cancer patients that were admitted at an oncology referral center between 2009 and 2012.
The analyzed variables were age, sex, performance status, past oncologic history, family oncologic history, tumor size, weight loss and body mass index, tumor location, grade of cellular differentiation, oncologic stage, lymphatic dissection, and curative intent resection.
The studied population was composed of 565 individuals (n=565), of which 444 were SCC and 105 EA. The remaining was composed of other less frequent tumors, such as neuroendocrine tumors.
Demographic, pathological and clinical characteristics were analyzed and compared to clinical stage and overall survival at 60 months. Average follow-up was 19.8 months.

Statistical Analysis
Regarding statistical analysis, to compare group means, ANOVA test was used; to analyze Kaplan-Meier curves, Log-Rank and Wilcoxon tests were used. Influence of prognostic variables was assessed by Cox regression. Significance level admitted was 0.05.

RESULTS
No difference was noted between SCC and EA overall survival curves. After five years, SCC presented 22.81% survival rate against 20.19% for EA ( Figure 1). There is no statistical difference between the curves (Log-Rank p-value=0.473; Wilcoxon p-value 0.098) Of all of the EA patients, 30.4% were eligible for curative intent surgery. This proportion was 20% in SCC patients (p-value for Log-Rank 0.114; for Wilcoxon 0.042). After five years, survival for EA was 58% and for SCC 56.6%. By univariate analysis, curative intention resection was clearly associated to a better survival rate (p-value < 0.001). Figure 2 and 3 present overall survival curves according to oncologic stages at diagnosis.
Longitudinal neoplasm extension at diagnosis was compared to clinical oncologic stage. By Chi-square analysis, it was noted that neoplasm size relate to poor prognosis in SCC (p-value 0.00), but not in EA (p-value 0.173). By univariate Cox regression, only in SCC tumor size was related to survival (p-value 0.001).
Weight loss (kg), BMI variation (kg/m²) and percentage of weight loss from initial symptoms to the diagnosis of esophageal carcinoma are factors that predict worse oncologic stage at diagnosis in the SCC. In EA, this finding was not statistically significant ( Figure 5). By logistic regression, BMI lower than 20 kg∕m² was a predictor of poor survival rate.
Considering only patients submitted to curative intent surgery, more than 23 node resection could not reach a statistically significant improvement in survival rate by univariate analysis (p=0.678 in EA and p=0.493 in SCC). By univariate and multivariate analysis (Tables 1 and  2), variables associated to poor survival rate in EA was weight loss, performance status at the moment of diagnosis and distal location tumors; for SCC, male sex, weight loss, performance status, past history of other malignances and delay in initiating treatment. For both carcinoma types, curative intention resection was more often associated to better prognosis.

DISCUSSION
Several factors have been related to prognosis in esophageal carcinoma 1,6,12 .
The present study analyzed prognostic factors associated to patients (age, gender, performance status, past oncologic history, family oncologic history, weight loss and body mass index); factors associated to neoplasm (tumor size, tumor location, grade of cellular differentiation, stage of cancer); and factors associated to treatment (quality of lymphadenopathy, curative intent resection).

Factors associated to patients
Age had association to bad prognosis only in SCC patients. Eloubeidi et al. also attributed elderly to poor survival rate 5 . Previous paper demonstrated also that family history of esophageal cancer can predict bad prognosis 12 . This finding is not in agreement with the present study.

Factors associated to neoplasm
High tumor size and high oncologic stage was associated with high mortality in SCC. This is in accordance with previous papers 5, 10 . This study evinces SCC with poor cellular differentiation leads to a poor oncologic stage at the moment of diagnosis. Tachibana et al. 10 also demonstrated an association of differentiation grade and prognosis.

Factors associated to surgery
Although it could not be demonstrated the relationship between survival and number of dissected lymphnodes, other studies showed a great importance of this variable.
The number of positive lymph nodes (more vs. less than 5 dissected nodes) is related to an increasing risk of mortality (hazard ratio [HR], 1.29; 95% confidence interval [95%CI], 1.06 -1.56) according to Eloubeidi et al. 5 Rizk et al. 9 showed that patients with more than four involved lymph nodes have survival similar to that of patients with M1 disease. Consequently, the number of lymph nodes removed would be an independent factor for prognosis. For Peyre et al 7 , a minimum of 23 regional lymph nodes should be removed.
In this study, survival improvement after curative intent surgery must be carefully analyzed, once selection for surgery (only not advanced stages) may be a bias.