Factors Affecting Survival in Neuroendocrine Tumors: A 15-Year Single Center Experience

Background: Neuroendocrine tumors are a heterogeneous group of tumors that can originate from all of the
neuroendocrine cells in the body, mostly from the gastrointestinal tract. In addition to early diagnosis, streaming
patients into appropriate prognostic groups is an important component of treatment. In this study, we examined the
factors that affect survival in patients we followed in our center between 2000-2016. Methods: The demographic data,
clinical and pathological features of patients were obtained from their medical files. TNM staging and tumor grading
were performed according to AJCC and WHO 2010 classification. SPSS 15.0 for Windows programme was used for
statistical analysis. Results: 85 patients (32 male, 53 female) were included into the study. The median age of the patients
was 55,7 (27-83) years. Eighty percent of the tumors were of gastroenteropancreatic system, most commonly stomach
(27.1%) origin. Nineteen patients (22.4%) died during follow-up. In univariate analysis; age (p<0,001), stage (p=0.002),
primary tumor localization (p=0.005), grade (p<0.001), Ki-67 value (p<0.001), number of metastases (p=0.001) and
type of surgery (p<0.001) were found to be factors affecting survival. Age (p=0.024) and Ki67 (p <0.001) were the
independent prognostic factors for survival in multivariate analysis. For the cut-off value of 6%, Ki-67 had a sensitivity
of 83.3% and specifity of 71.4% for survival determination. Conclusion: Ki-67 ratio and age were the most important
factors affecting survival in neuroendocrine tumors in our study. Ki-67 ratio has a high sensitivity and specificity for
predicting survival, a cut-off value of 6% may be used to predict survival.


Introduction
Neuroendocrine tumors (NET) are a heterogeneous group of tumors that can originate from all of the neuroendocrine cells in the body, mostly from the lung and gastrointestinal tract including stomach, pancreas, small and large intestine, rectum. They can occur at any age, although it is often seen over 50 years. The incidence of NET is higher in men than in women. Even though they usually exhibit indolent clinical course, they may become very aggressive and rapidly become metastatic. Since most of NET are not functional, they often cause no signs and symptoms, which makes early diagnosis difficult and decreases survival by reducing the chance of curative treatment (Yao et al., 2008). In addition to early diagnosis, streaming patients into appropriate prognostic groups is an important component of treatment. However, the absence of frequently accepted classifications limits its benefit on survival (Bilimoria et al., 2007).
There is insufficient information about the incidence of carcinoid syndrome, stage, location of the primary, the location and number of metastases, type of surgery and treatment applied were obtained from their medical files. Patients with incomplete data, missing data, or multiple primers were excluded from the study. A total of 85 patients (32 males and 53 females) were included in the study. The TNM staging of patients and grading (G) of the tumor were performed according to AJCC and 2010 WHO classification, respectively (Bosman, 2010); World Health Organization; International Agency for Research on Cancer. WHO Classification of Tumours of the Digestive System. 4th ed. Lyon: International Agency for Research on Cancer). The study protocol was approved by the Corporate Ethics Committee and found to comply with ethical principles for epidemiological investigations.
SPSS 15.0 for Windows program was used for statistical analysis. Descriptive statistics were given as mean, standard deviation, minimum, maximum for numerical variables, number and percentage for categorical variables. The numerical variables in the independent two groups were analyzed by Student t test and Mann Whitney U test if normal distribution condition was provided and not met, respectively. The comparisons of ratios between groups were made with Chi Square Analysis. Monte Carlo simulation was applied when conditions were not met. The survival analyzes were performed with Kaplan Meier Analysis. Determinants for survival were examined by Cox Regression Analysis. In univariate analysis, forward stepwise model was used for values with p<0.100. The cut-off values were determined by using Roc Curve Analysis. The statistical significance level of alpha was accepted as p <0.05.

Results
A total of 85 patients, 32 (37.6%) male and 53 (62.4%) female, were included in the study. The median age was 55.7 (27-83) years. Eighty percent of the tumors were of gastroenteropancreatic system, most commonly stomach (27.1%) origin.
In addition, the rates of G3 (p<0.001) and stage IV disease (0.016) were significantly higher, whereas curative surgery rate was significantly lower (p=0.007) in patients who died (Table 3). In univariate analysis; age (p<0.001), stage (p=0.002), primary tumor localization (p=0.005), grade (p <0.001), Ki-67 ratio (p<0.001), the number of metastasis (p=0.001) and the type of surgery (p <0.001) were found to be the factors affecting survival (Table 4). When factors affecting the OS were evaluated, age (p=0.024) and Ki-67 ratio (p<0.001) were found to be the most significant factors according to Forward Stepwise analysis based on model consisted of variables of which p values were determined as <0.100 in univariate analysis (age, smoking, primary tumor localization, grade, stage, metastasis, type of surgery, Ki-67 ratio) (Table 5).
There was a statistically significant difference in survival rates in the Ki-67 ratio groups (p<0.001). Patients with a Ki-67 ratio of >20% had a statistically significant lower survival rate than those with ≤2% (p<0.001) and 3-20% (p=0.002). On the orher hand, no significant difference in survival rates was detected between patients with Ki-67 value of ≤2% and 3-20% (p=0.094) ( Figure  1)  According to the ROC analysis for the determination of mortality, the sensitivity and specificity of the Ki-67 ratio were found as 83.3% and 71.4% for the cut-off value of >6% (AUC:0.813 (%95 CI: 0.664-0.963) (Figure 2). The cumulative survival rate of patients with a Ki67 ratio of ≥6% was found to be statistically significantly lower than those with a Ki67 ratio of <6% (p<0.001) (Figure 1). The median survival was not reached in the group with a Ki67 ratio of <6% whereas it was 86±41.9 (95% CI:3.8-168.1) months in the group with a Ki67 ratio of ≥6%.

Discussion
The naming and classification of NET have been changed several times, making it difficult to collect epidemiological information and compare studies published in the literature. The actual incidence of NETs is not known due to the lack of sufficient multicentric and epidemiological studies. This may explain the difference in incidence of NET between gender, race, country and continent (Hauso et al., 2008).
Over the last decade, attempts have been made to develop existing classification systems. There is limited data on long-term follow-up and survival in patients with NET. Because of infrequency and the differences in the diagnosis of NET, it is difficult to identify high risk factors. There are only a few studies that define prognostic factors, thus, factors affecting survival of patients with NET is lacking in many countries (Faggiano et al., 2012).
The median age of the patients at our study was 55.7 years, similar to other studies (Niederle et al., 2010;Araujo et al., 2013;Lewkowicz et al., 2015;Nikou et al., 2016). Five percent of the cases were asymptomatic. The incidence of carcinoid syndrome was 8.2%. Similar to other studies, the most common symptom was abdominal pain (Araujo et al., 2013;Lewkowicz et al., 2015). The most common disease grade seen in our study was G1. The most common localizations of the G1 disease were of the rectum and appendix in other studies, whereas it was of pancreas and appendix in our study (Niederle et al., 2010;Lewkowicz et al., 2015).
The pancreas and lung were the most common primary localizations in the study by Nikou et al., (2016). In another study, the most common primary localizations were alined as rectum, duodenum, pancreas and stomach while the most frequent stage, grade and metastatic site were stage 1, grade 1 and the liver, respectively (Lim et al., 2017). In our study, unlike other studies, the most common localizations were stomach, pancreas and small bowel (Garcia-Carbonero et al., 2010;Niederle et al., 2010;Lim et al., 2011;Lewkowicz et al., 2015). The most common distant metastasis site was liver.
The only curative treatment method in NET is the surgical resection. Surgery should be considered in patients with early stage, locoregional and resectable metastatic disease (Bilimoria et al., 2007). In our study, curative surgery was applied to 63.5% of patients and lymph node metastasis was detected in 30.5% of patients who underwent surgical treatment. Of our patients, 44.8% were metastatic at the time of diagnosis.
In a study that evaluated prognostic factors after resection of pancreatic NET, the presence of tumor necrosis, lymph node and liver metastasis was found to be associated with disease-free survival whereas age, tumor grade and the presence of distant metastasis were detected as the most significant determinants of survival (Bilimoria et al., 2008). Various studies have reported different survival rates according to tumor localization (Garcia-Carbonero et al., 2010;Lim et al., 2011;Lewkowicz et al., 2015). In the study by Lewkowicz et al., (2015) advanced stage, G2 and presence of metastasis at diagnosis were determined to be associated with poor prognosis in the univariate analysis, while presence of advanced stage and metastasis was found as the independent risk factors for poor outcome in the multivariate analysis. In another study, grade and stage were found as the independent risk factors for survival (Garcia-Carbonero et al., 2010). In the study by Ma et al., (2017) it was determined that 5-year survival rate of patients with advanced age, tumor localized in stomach, duodenum and colon, a tumor size of ≥4 cm and G3 disease was lower in univariate analysis. In multivariate analysis, age, stage, lymph node and distant metastasis were found to be independent risk factors affecting the prognosis of patients. In our study, age, Ki-67 and mitosis rate, stage, gastric localization, presence of distant metastases at the time of diagnosis, number of metastases, presence of lymph node metastasis and CT use were found to be factors affecting survival in univariate analysis. In multivariate analysis, age and Ki-67 ratio were found to be the most significant factors. The lower survival rates in gastric localization may be due to higher values of Ki-67 ratios of tumors in gastric localization. Furthermore, the use of CT in the treatment of symptomatic patients with high tumor burden may explain the lower survival rates in this group.
As a conclusion, in our study, age, Ki-67 ratio, number of mitosis, number of metastases, gastric localization of the primary tumor, presence of distant metastases, presence of lymph node metastases, G3 and stage IV disease and CT utilization rates were statistically significant higher in patients with exitus compared with those alive. The Ki-67 ratio and age were determined as the most important factors affecting survival. Ki-67 ratio has high sensitivity and specificity in predicting survival. We think that the Ki-67 ratio of ≥6% might be used to estimate survival.

Funding Source
None.