Characteristics and prognosis of colorectal signet ring cell carcinoma in patients aged ≥ 65 years

Objective: To evaluate the prognostic value of age on colorectal signet ring cell carcinoma (SRCC) and identify the prognosis factors of patients aged ≥ 65 years. Methods: Patients were recruited from the Surveillance, Epidemiology, and End Results (SEER) database. We ensured that all characteristics were well balanced through adopting the propensity score matching (PSM) method. In addition, the prognostic factors were evaluated by the Cox proportional hazardous model. Results: Altogether 3305 cases were enrolled, including 1564 in elderly group (aged ³65 years). As a result, more patients were female, unmarried, white, earlier AJCC stage, tumor size ≤ 5cm, grade III/IV and located in cecum–transverse colon in the elderly group; whereas signicant fewer patients received chemotherapy and radiotherapy. After 1:1 PSM analysis, each group contained 1032 cases. As discovered by Kaplan-Meier analysis, the elderly group was associated with remarkably dismal survival compared with the younger group after PSM analysis (P=0.0091).Moreover, grade, primary site, AJCC stage, tumor size, surgery and chemotherapy were identied as the independent prognostic factors for the elderly patients. Conclusion: The prognosis of colorectal SRCC aged ≥ 65 years was worse than the younger patients. We may need strengthen the chemotherapy of the eldly colorectal SRCC.


Introduction
In the USA, colorectal cancer (CRC) ranks the second place among the common causes of cancerassociated mortality, and it has became severe health threat in the world [1]. Of the diverse CRC subtypes, colorectal signet ring cell carcinoma (SRCC) arouses wide interests recently. Originally depicted by Laufman and Saphirin [2],SRCC is suggested to be mainly derived from undifferentiated colorectal mucosal stem cells; as a result, SRCC usually shows low differentiation, fast proliferation, high metastasis and diffuse in ltration [3,4].
Many diverse clinicopathological features and prognostic outcomes have been reported for the elderly CRC cases in existing studies. It has been suggested that, the younger CRC cases showed unfavorable pathological characteristics and advanced stage relative to the elderly patients [5,6]. However, other studies discover no difference in the pathological characteristics and tumor stage between the younger and elderly populations [7,8]. But no speci c data on the elderly patients with colorectal SRCC are available at present due to the rarity of this disease.
The aim of this study was to compare the clinicopathologic characteristics and survival of colorectal SRCC in two groups of patients (< 65 years and ≥65 years). We also aimed to identify the prognosis factors of patients aged ≥ 65 years.

Ethical statement
The Surveillance, Epidemiology, and End Results (SEER) database formulated by the National Cancer Institute, which is started in 1973 and updated every year, utilizes the population-based data for the development of all-sided sources [9].This databases covers about 30% US population among different geographic areas [10]. To access the SEER database, we signed the SEER Research Data Agreement (reference number,18753-Nov2016). Related data were obtained according to the approved guidelines.
The data analysis was considered by the O ce for Human Research Protection to be non-human subjects who were researched by the United States Department of Health and Human Services, as they were publicly available and de-identi ed. Thus, it did not require approval by the institutional review board.

Study population
The eligible cases were screened by adopting the SEER*State v8. 3 those diagnosed based on autopsy or the death certi cate; (3)those with no available information on AJCC stage; and (4) those with no data on prognosis. All the rest participants were included into the initial SEER cohort.

Covariates
The clinicopathological features of patients shown below were examined, including age, sex, race, marital status, insurance status, year of diagnosis, tumor size, primary site, AJCC stage, grade, lymph node dissection (LND), surgery, chemotherapy and radiotherapy.
In this study, the single (never married), widowed (having a domestic partner), separated and divorced cases were classi ed into unmarried category.With regard to primary tumor site, it was divided into cecum-transverse colon (like appendix, cecum, hepatic exure, ascending colon, transverse colon), descending colon-sigmoid (like splenic exure, descending colon, sigmoid colon), rectum, multiple or unknown [11]. As for the year of diagnosis, it was classi ed as 2004-2007, 2008-2011, 2012-2015, in accordance with previous studies [12,13]. Additionally, and tumor size [14,15]were also grouped in line with previous reports. Cancer stage was classi ed by the AJCC classi cation system (6th edition) adapting to SEER-derived patients diagnosed between 2004 and 2015 [16]. In this study, we set cancer-speci c survival (CSS) as the endpoint, which referred to the duration between diagnosis and death because of colorectal SRCC.
Propensity score matching (PSM) Selection bias was inevitable in the observational studies; as a result, confounding factors may be unequally distributed between the two groups. PSM refers to the process to assign conditional probabilities to the speci c treatment based on diverse observed covariates [17]. For reducing the selection bias and avoiding the unequal confounding factor distribution, we adopted PSM in the present study [18]. In addition, the logistic regression model was utilized to predict the propensity score for each case using the SPSS plug-in PSM, including each survival-affecting covariate. Thereafter, PSM at the 1:1 nearest neighbor matching was carried out between the elderly and younger groups for all subgroups, and the caliper was set at 0.01 for the sake of accepting the matched pair.

Statistical analysis
All the categorical were expressed in the manner of number and percentage, while continuous variables were expressed as mean ±SD. Fisher's exact test and Pearson's chi-square test were used to compare clinicopathological features pre-and post-PSM. Meanwhile, the Kaplan-Meier method was utilized to estimate patients' survival. Differences in patients' survival between different groups were evaluated by log-rank test. Cox proportional hazards regression was adopted for assessing the relationships of different variables with survival. Later, signi cant variables obtained from univariate analysis (P < 0.05) or those identi ed to be prognostic factors before were enrolled in multivariate analysis. SPSS23.0 (SPSS Inc., Chicago, USA) was utilized for statistical analysis. A two-sided P< 0.05 was considered as statistically signi cant.

Results
Patient selection procedure and age distribution A total of 3305 patients met the inclusion criteria for analyzing characteristics. Figure 1 presents the patients selection procedure. Median age at diagnosis of all patients was 63 years (12-103 years). The age of patients is almost normal distribution, and the age group of 50-70 accounted for 43.4%. Figure 2 shows the age distribution. Besides, in line with previous [19,20], the enrolled patients were classi ed as two age groups, including the older (³65 years, n=1564) and younger ( 65 years, n=1741) groups.
Patient features between two groups Table 1 presents the patient demographic, tumor feature and treatment data. Differences in sex, race, insurance status, marital status, tumor size, grade, primary site, AJCC stage, LND number, chemotherapy and radiotherapy were statistically signi cant between the two groups. There were more female(P<0.001),unmarried(P<0.001), white (P<0.001), medicaid/insured(P=0.023), earlier AJCC stage, tumor size ≤5cm, grade III/IV(P<0.001) and cecum-transverse colon(P<0.001) cancer patients in the elderly group; whereas a lower proportion of these patients received chemotherapy(P<0.001) and radiotherapy(P<0.001) in comparison with those in the younger group. Meanwhile, difference in received surgery was not signi cant between the two groups (P =0.065).
PSM and CSS post-PSM PSM incorporated factors such as age, sex, race, insurance status, marital status, tumor size, grade, primary site, LND number, AJCC stage, chemotherapy, surgery and radiotherapy. After 1:1 PSM analysis, both groups contained 1032 cases, respectively. None of the variables showed signi cant difference after PSM (Table 2). In addition, according to Kaplan-Meier analysis, the elderly patients had markedly poorer survival compared with the younger patients after PSM( P =0.0091) (Figure 3).

Survival and prognostic factors for the elderly group patients
For the 1564 elderly cases, the survival time ranged from 0 to 155 (median, 13.0) months. The CSS rates at 1, 3 and 5 years were 57.25%, 37.66% and 33.64%. As revealed by univariate analysis, sex, tumor size, primary site, AJCC stage, grade, LND number, surgery and chemotherapy were identi ed as the predicting factors for CSS (all P 0.05). According to multivariate analysis, tumor size ≤5cm, primary site at cecumtransverse colon, earlier AJCC stage and grade I/II were identi ed to be the favorable prognostic factors. Moreover, patients who received surgery or chemotherapy were associated with superior prognosis compared to those who did not [(HR:0.674, 95% CI: 0.498-0.912, P =0.010) and (HR:0.488, 95% CI: 0.424-0.562, P<0.001)] (Table 3).

Discussion
The present population-based, retrospective study was carried out to illustrate age distribution in colorectal SRCC and evaluate the signi cance of age in prognosis prediction. As far as we know, this study is the rst large scale population-based study that utilizes PSM for assessing the signi cance of age in predicting colorectal SRCC prognosis. As a result, the age of colorectal SRCCs patients was almost distributed normally. Difference in survival was signi cant between patients of the two age groups. Moreover, tumor size, primary site, AJCC stage, grade, chemotherapy and surgery were identi ed to be the independent prognostic factors for the elderly patients.
Colorectal SRCC, an uncommon CRC histological subtype, is classi ed by the WHO classi cation of tumors to be the colorectal adenocarcinoma variant, where over 50% cancer cells have obvious intracytoplasmic mucin [21]. Such cancer is featured by different clinical presentations and prognostic outcomes [22]. SRCC shows dismal prognostic outcome and high malignancy grade relative to additional CRC subtypes [23,24].
Due to the low prevalence of colorectal SRCC, little information is available regarding the role of age in predicting the colorectal SRCC prognosis.As discovered by Ping He, the elderly (³ 65 years old) showed dismal prognosis compared with the younger patients( 65 years old) [19], which is conformed to our ndings.
In addition, our study suggested that chemotherapy and surgery were the factors for the favorable prognosis among the elderly patients, indicating that the aggressive and effective treatment markedly decreased the colorectal SRCC-associated mortality among the elderly patients. Surgery exerts an important part in treating locoregional tumors [25]. The value of chemotherapy has also been a rmed by many studies [14,19,26]. Our results suggested that the proportions of the patients who received surgery were similar between the two age groups, while that of patients who received chemotherapy signi cantly decreased in the elderly group (39.0% vs. 74.8%).Such nding is commonly found among the elderly patients who receive anticancer treatment in other tumors [27][28][29]. Therefore, we should strengthen the treatment of elderly patients.
So far, the present study is the rst large-scale retrospective analysis on the role of age in predicting colorectal SRCC prognosis. Our study assists the clinical workers in further understanding the clinicopathological characteristics and treatment for the elderly colorectal SRCC patients (³65 years old). Nonetheless, certain limitations should be noted in the present study. First of all, selection bias was inevitable in the observational study. The application of PSM analysis could reduced the bias resulting from the unequal distribution of covariates measured, but bias induced by the unmeasured variables were inevitable. Secondly, we enrolled altogether 14 variables, but some variables still could not enrolled in the SEER database, like vascular invasion, status of surgical margin or the molecular-genetic pro les.
Thirdly, detailed information about chemotherapy was not available. Therefore, we can not further explore which chemotherapy regimen is safe and effective for elderly patients. Our future work will focus on solving the above limitations.

Conclusion
In conclusion, the clinicopathological features were signi cantly different between younger and older group, and the prognosis of colorectal SRCC in aged ³65 years patients was signi cantly worse than that in younger patients. Primary site, grade, tumor size, AJCC stage, surgery and chemotherapy were independent risk factors for the prognosis of elderly patients. Nevertheless, more studies are warranted to determine whether more intense treatment is necessary for colorectal SRCC patients if appropriate.

Declarations Ethics approval
Approval from the ethical board for this study was not required because of the public nature of all the data.

Funding
No funding.

Con ict of interest
All authors declare that they have no con icts of interest.