Clinicopathological Features, Treatment Strategy, and Prognosis of Primary Non-Hodgkin's Lymphoma of the Duodenum: A SEER Database Analysis

Objective Primary duodenum lymphoma (PDL) is extremely rare with limited data available in the literature. In this study, we sought to describe clinical features and identify factors affecting survival in patients with PDL using a large population cohort. Methods The Surveillance, Epidemiology, and End Results (SEER) database was queried from 1998 to 2015. Results A total of 1060 cases of PDLs were identified. Clinicopathological features as well as survival data of PDLs were analyzed and compared with 10573 primary gastric lymphomas (PGLs) and 3239 primary small intestinal lymphomas (PSILs) from the SEER database. PDL patients were younger in age (60.96 ± 15.205), and the proportion of stage I (53.21%) was higher in Ann Arbor staging. The proportion of PDLs treated by surgery (8.68%) is the lowest among PDLs, PGLs, and PSILs. The DSS of PDLs were significantly better than those of PGLs and PSILs, respectively (10-year survival rate: 21.24% vs. 20.40%, P=0.027; 10-year survival rate: 21.24% vs. 16.79%, P=0.001). Age, gender, Ann Arbor staging, and histological type were regarded as independent prognostic factor for the DSS by multivariate analysis (all P < 0.05). Patients with <65 years, female, stage I, and FL were found to be significantly associated with good DSS. The treatment modality (surgery vs. conservative treatment) was not statistically related to DSS. The proportion of PDL patients who received surgical treatment gradually decreased from 15.60% in period 2 to 5.26% in period 4. Conclusions The clinicopathologic features of duodenal lymphoma were significantly different from those of gastric lymphoma and small intestinal lymphoma. The prognosis of PDLs was significantly better than those of the other two groups, and there was no statistical survival benefit from surgery in PDLs.


Introduction
e most predominant extranodal site in non-Hodgkin's lymphoma (NHL) is the gastrointestinal (GI) tract [1], accounting for 5% to 20% of all NHL cases and 30% to 45% of all extranodal cases [2]. e lesion can occur in any part of the digestive tract from the mouth to the anus, of which stomach is the most common pathogenic sites (60%-75%) [3].
As primary duodenum lymphomas (PDLs) are exceedingly rare, the current researches about PDLs are based on anecdotal reports [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23]. In the present study, we retrospectively reviewed the clinical and pathological manifestations of lymphomas of duodenum lymphoma for cases based on the largest sample size so far to identify prognostic factors and to clarify the value of treatment modalities in the management of these malignancies. lymphoma in the coding system of the International Classification of Diseases for Oncology (ICD-O)-3 were 9590-9729. e search was limited to adult patients (≥18 years old) with the type of follow-up equal to "active followup." e exclusion criteria were as follows: (i) patients without definitely histological confirmation; (ii) patients with only autopsy or death certificate records; (iii) patients with incomplete survival data and follow-up information; (iv) patients without Ann Arbor stage record; and (v) patients without the information of surgery. After screening, we got a total of 10321 patients.
Clinical and pathological variables (e.g., age, gender, race, sex, age at diagnosis, marital status, year of diagnosis, histological type, Ann Arbor stage, whole body symptom of lymphoma based on the AJCC (6th edition) staging system, treatment modalities employed and information of "cause of death and follow-up," and "multiple primary field") were extracted from the SEER.
Since the SEER cause-specific death classification variable is defined by taking into account cause of death in conjunction with sequence of tumor occurrence (ie, only one tumor or the first of multiple tumors) and comorbidities (e.g., AIDS and/or site-related diseases), we excluded the patients except that lymphoma was the only one primary cancer or the 1st cancer of 2 or more primaries to avoid the ambiguity of the lymphoma-specific survival [24]. e survival data were available in the measurement unit of months, without precise days. Considering the preconditions that no precise survival days were available and that patients with only autopsy or death certificate records were excluded, a survival time of 0 months was recorded as 0.5 months to include patients who died within 1 month of diagnosis but who did not reach the 1-month threshold [25,26].
Since this study only involves analysis of the publically available database (SEER) and does not contain any identifying patient information, the ethical approval of this study by the institutional review board (IRB) is not required.

Statistical Analyses.
Statistical analyses were performed using the statistical software SPSS 22.0 for Apple (SPSS Inc., Chicago, IL). Numerical variables were expressed as mean ± SD and were analyzed by the t-test. Discrete variables were analyzed using the chi-square test or Fisher's exact test. Risk factors for survival were identified by univariate analysis, and COX regression was employed for multivariate analysis. Disease-specific survival (DSS) was analyzed by the Kaplan-Meier method and differences between the curves were compared using the log-rank test.
All P values were two-sided, and P values < 0.05 were considered statistically significant.   Table 2). e results showed that there were no significant differences in age, gender, marital status, race, and other cancers between the surgery and conservative groups. However, primary site, Ann Arbor staging, symptoms, and histological type were significantly different between the two groups (all P < 0.05); that is, incidence of cancers with I stage or A symptoms was significantly higher in the conservative group compared to that in the surgery group. e results showed that age, gender, symptom, and histological type were significantly different between PDLs and PGLs (all P < 0.05); that is, incidence of tumors with younger patients or more follicular lymphoma was significantly higher in the PDL group compared to that in the PGL group. e PDL group also showed younger patients, earlier Ann Arbor staging, more follicular lymphoma, and more surgery treatment in comparison with those of the PSIL group (all P < 0.05).
Furthermore, univariate and multivariate analyses were performed to evaluate the prognosis of PDLs (Table 3). Age, gender, Ann Arbor staging, and histological type were regarded as independent prognostic factors for the DSS (all P < 0.05). Symptom was regarded as a significant risk factor for the DSS by univariate analysis (P � 0.002), while it is not an independent prognostic factor for DSS by multivariate analysis.

Stratified Analysis.
We showed stratified analysis according to several prognostic variables based on multivariate analyses (Figure 2). Patients with <65 years, female, stage I, and FL were found to be significantly associated with good DSS. However, patients with ≥60 years, male, stage IV, and TCL were found to be significantly associated with poor DSS (all P < 0.05).

Discussion
To the best of our knowledge, the current study represented the largest number of PDLs. In this study, we summarized clinical and pathological features of 1060 cases of PDLs. We further analyzed prognosis of PDLs in comparison with that of PDLs and PSILs. It was found that tumors with younger patients or more follicular lymphoma was significantly higher in PDLs. In addition, PDLs had poorer prognosis compared to PGLs and PSILs. ese observations indicate that surgery treatment may not play a role in improving survival in patients as compared to conservative treatment. Since 2000, the proportion of PDL patients undergoing surgery has declined. We know that follicular lymphoma (FL) is primarily a nodal disease and primary FL of the gastrointestinal (GI) tract is rare [27]. However, the most common histological subtype is FL, followed by DLBCL among of PDLs [22]. Our study showed that the proportion of FL was the highest (44.13%) and significantly higher than that of stomach (2.23%) and small intestine (22.97%). erefore, the predominance of the follicular histology in PDL was interesting. e high proportion of follicular lymphoma in duodenal lymphoma might be an important reason why the prognosis was better than that of the stomach and small intestine.  Our study showed that the mean age (60.96 ± 15.205) of patients with PDL was younger than that of the stomach and small intestine, and that the proportion of stage I was also higher than that of the stomach and small intestine. e duodenal anatomy site is special, the tumor growth space is small, and the patient presents the discomfort symptom earlier than the stomach and small intestinal. At the same time, EUS can not only clarify the lesions on the mucosal surface of the gastrointestinal tract but also understand the changes in the hierarchical structure of the gastrointestinal  Figure 2: e stratified analysis according to (a) age, (b) gender, (c) Ann Arbor staging, and (d) histological type in the PDLs. Patients with <65 years, female, stage I, and FL were found to be significantly associated with good DSS. However, patients with ≥60 years, male, stage IV, and TCL were found to be significantly associated with poor DSS (all P < 0.05).
Canadian Journal of Gastroenterology and Hepatology 5 wall and its relationship with adjacent tissues and organs. It might relate to lower age and tumor staging. Based on the assumption that gastrointestinal tract lymphoma is a localised disease, the surgical treatment was traditionally considered the cornerstone of the therapeutical strategy showing impressive results in terms of long survival. Nowadays, this approach has been extensively revised, and the management of gastrointestinal tract lymphoma is centred on systemic treatments such as chemo-and radiotherapy. From the EER data, the proportion of patients undergoing surgery gradually decreased from 2000 to 2015. From our study, the treatment of PDLs was also in line with the current treatment trend, but interestingly, the proportion of PDLs treated by surgery was lower than that of the stomach (9.64%) and the small intestine (35.13%), among which it was significantly lower than that of the small intestine. e reason may be that the duodenal lesion is mostly found in the descending segment [4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23], which has a complex anatomical structure and a small possibility of local resection, unlike the small intestine which can be directly resected, so conservative treatment is more preferred. Once a larger operation is performed, it is bound to cause complications and affect the quality of life. Meanwhile, multivariate analysis confirmed that the treatment modality was unrelated to DSS; that is, surgical treatment did not bring a survival advantage. e results were similar to previous reports (Table 4) [28][29][30][31] that the survival results of nonsurgical treatment were similar or even better than those of surgical treatment. Surgery, thus, is restricted to the treatment of complications such as occlusion, bleeding or perforation. Preventive surgery is sometimes advocated in bulky tumors, when rapid tumor necrosis secondary to chemo/radiotherapy may be associated with a high risk of life-threatening complications [28]. Surgery is also required for removal of residual disease after medical debulking [32]. Since the SEER database does not list the complications, this paper cannot discuss the complications.
Although there was no statistically significant difference in survival by treatment modalities in the multivariate analysis, there are other multiple factors that contribute to survival. In previous studies, female, low-grade histology and good PS have been reported to be associated with high OS. However, age >60 years, advanced stage, poor performance status (PS), and elevated lactic dehydrogenase (LDH) were associated with poor outcome [3,[32][33][34]. In our study, age, gender, Ann Arbor staging, and histological type retained independent prognostic factors in the multivariate analysis. Patients with <65 years, female, stage I , and FL were found to be significantly associated with good DSS. LDH and PS are not mentioned in the SEER database, so statistical analysis cannot be made in this paper.
Although it is an excellent resource for comparative outcome analysis for all malignancies involving the gastrointestinal tract, SEER has its limitations. Since the database provides passive follow-up for its registered cases, incomplete data reporting remains a problem. First, much information could not be obtained from the SEER database, such as PS and LDH. Second, the SEER database also did not describe postoperative complications and quality of life score, so we were unable to assess the complications and quality of life associated with surgery.