Surgical Resection for Primary Rectal Lymphoma: Support for Local Excision?

Abstract

Background

Primary rectal lymphoma is an uncommon and heterogeneous malignancy. Due to its rarity, few data exist to guide treatment or counsel patients. We present the largest series to date of patients undergoing non-palliative surgery for rectal lymphoma. We hypothesize that there will be no difference in overall survival between patients undergoing local or radical resection.

Materials & Methods

The National Cancer Data Base was queried for all cases of resected primary rectal lymphoma between years 2004 and 2014. Exclusion criteria included patients with Stage IV disease and those operated on for palliation. Patients were categorized by resection approach-local (LR) or radical (RR). Approach, along with demographic, histologic, and hospital level factors were analyzed with a Cox Proportional Hazard analysis.

Results

233 patients were identified. Mean age was 63 (IQR 53–73) and 57% of the population was female. The most common histologic subtypes were Marginal (44%), Diffuse Large B-Cell (DLBCL) (20%) and Follicular Lymphoma (17%). 87% underwent local resection. There was no significant difference in R0 resection (LR:38% vs RR: 58%; p=0.07), adjuvant chemotherapy (LR:18% vs RR: 29%; p=0.22), or adjuvant radiation (LR:21% vs RR: 16%; p=0.63) between the groups. Five-year overall survival was 79%, and there was no significant difference in approach (LR:81% versus RR: 56%, p =0.06). Multivariable analysis did not identify an association between approach and overall survival.

Conclusions

Surgical resection of rectal lymphoma is rare. Our data support consideration of local resection when possible, given the lack of convincing survival benefit of radical surgery or R0 resection.

Introduction

The gastrointestinal tract is the most common site of extra-nodal lymphoma. Primary rectal lymphoma (PRL) is an uncommon malignancy, accounting for <1% of colorectal cancers¹ and <10% of all gastrointestinal lymphomas². In addition to its rarity, PRL exhibits great variety in terms of histology and aggressiveness³. Therefore, selection of the optimal treatment strategy is a multidisciplinary challenge without standard guidelines. Treatment may include chemotherapy, radiation, and surgery.

Approximately 48–61% of patients with colorectal lymphoma undergo surgery^4–6. Surgery may palliate symptoms, alleviate or prevent tumor complications, provide local control, or be elected as definitive therapy. However, surgery does not definitively improve survival, is morbid, and fails to address systemic disease^7–9.

PRL is a particular challenge for surgeons. Specific data to guide surgical approach or counsel patients are lacking. This is due to lack of current studies, agglomeration of multiple anatomic sites into case series, and lumping of metastatic and localized disease. Published literature specific to rectal lymphomas are largely case reports. In the rectum, surgeons have the option of local and radical resection. However, due to the paucity of data on PRL, there is a knowledge gap in the influence of approach on the expected short and long term outcomes for surgical patients.

The National Cancer Database (NCDB) captures the majority of newly diagnosed cancers in the United States and corresponds with population-based cancer registries¹⁰. As such, it is an appropriate resource for the study of “real world” outcomes of patients with rare malignancies. The purpose of this study is to provide a modern, multi-institutional description of the short and long term outcomes of patients undergoing non-palliative local excision or radical resection for primary rectal lymphoma. We hypothesize that radical resection will not provide a survival advantage.

Methods

Data Source

Data for this study was obtained from the Participant User File of the National Cancer Database (NCDB). The NCDB - jointly sponsored by the American College of Surgeons and the American Cancer Society—is a clinical oncology database sourced from hospital registry data that are collected in more than 1,500 Commission on Cancer (CoC)-accredited facilities¹¹. NCDB data are used to analyze and track patients with malignant neoplastic diseases, their treatments, and outcomes. Data represent more than 70 percent of newly diagnosed cancer cases nationwide and more than 34 million historical records¹². Permission to use the NCDB was acquired from both Vanderbilt and the CoC.

Study Design

We performed a retrospective, cohort analysis of a prospectively maintained national database. All adult patients in the NCBD from 2004–2014 with a diagnosis of primary rectal lymphoma (International Classification of Diseases for Oncology, 3rd Edition SEER Topography Codes C20.9 or C21.8 with Histology Code contained in Appendix 1). Exclusion criteria included surgery performed outside the reporting facility, metastatic disease, and operations purely for palliative intent. Cases were stratified by the primary exposure of surgical approach- either local or radical resection as defined by NCDB coding.

Local resection is defined by the NCDB as either “Local tumor excision, NOS”, “Polypectomy”, or “Excisional biopsy”. Radical resection of the rectum includes low anterior resection and abdominoperineal excision. This study was reviewed and deemed exempt by the Vanderbilt University Medical Center Institutional Review Board (Protocol 161707).

Outcomes

Overall survival was the primary outcome of this study. R0 resection, 30 & 90-day mortality, and readmission were reviewed as secondary outcomes. The NCDB does not contain information on recurrence or disease-free survival.

Variables

For cases identified by the above criteria, patient level variables included age, gender, race, and education and income as markers of socio-economic status were reviewed. Comorbidities were captured by the NCDB’s modification of the Charlson score, which excludes the primary cancer diagnosis condition and then reports the Charlson score as 0 (no comorbidities), 1 (Total Charlson score of 1) or 2 (Total Charlson Score of 2 or more). Hospital level data assessed included the cancer designation of the hospital where the surgery was performed (community cancer program, comprehensive community cancer program, academic/research program). Type of lymphoma was assessed using the International Classification of Diseases for Oncology, 3rd Edition SEER Histology code. (Appendix 1). Stage is recorded in the NCDB as pathological stage when available, or may be recorded as clinical stage when pathological stage is not documented. For the survival analysis, lymphoma types were collapsed to marginal/follicular lymphoma, diffuse large B-cell lymphoma and other.

Statistical Analysis

We present continuous and categorical data as mean with standard deviation and proportions respectively. The exception being continuous variables with non-normal distributions that are presented as the median with the interquartile range. Continuous and categorical variables were analyzed with the Student’s T-test and Chi-square test respectively. For survival analysis, bivariate comparison was determined using the log-rank test. For the adjusted analysis, a multivariable Cox proportional hazard model was built using all factors found to be significant for overall survival in a bivariate analysis as well as any variables determined to be clinically relevant. Data management and analysis was performed using SAS (version 9.3; SAS Institutes Inc., Cary NC, USA) was used for all analyses. All tests were two sided with an alpha level of 0.05.

Results

Demographics

After applying exclusion criteria, we identified 233 patients with PRL who underwent non-palliative surgery from 2004–2014 in the NCDB (Table 1). The majority of patients (86.7%) underwent a local excision. The median age was 63 (inter-quartile range 53–73) and 132 (56.6%) were female. Age and gender was not different between groups. Groups did not differ in terms of race, Charlson comorbidity score, household income, or facility type (Table 1). Histologic sub-types were diverse in both surgical cohorts and varied significantly between groups (Table 1). The most common histologic subtype was diffuse large B-cell lymphoma (DLBCL) in the radical resection cohort (48.4%) and marginal zone lymphoma in the local excision cohort (43.8%). The majority of patients (77.4%) were Stage I. Stage varied between groups (Table 2), with significantly more patients having Stage I disease in the local excision group than in the radical resection group (81.2% vs. 54.8%; p = 0.001).

Table 1:

Patient characteristics grouped by treatment with local resection (n=202) versus radical resection (n = 31).

	All (n=233)		Local (n=202)	Radical (n= 31)	p-value
Age, median, y (IQR)	63 (53–73)		63 (53–73)	69 (52–75)	0.29
Female gender	132 (56.6%)		114 (56.4%)	18 (58.0%)	0.86
Race					0.17
White	194 (83.2%)		168 (83.2%)	26 (83.9%)
Black	16 (6.9%)		12 (5.9%)	4 (12.9%)
Asian/Other	23 (9.9%)		22 (10.9%)	1 (3.2%)
Insurance Type					0.46
Private	105 (45.1%)		94 (46.5%)	11 (35.5%)
Medicare	109 (46.8%)		91 (45.0%)	18 (58.1%)
Medicaid	12 (5.1%)		11 (5.4%)	1 (3.2%)
Other	4 (1.7%)		4 (1.9%)	0 (0%)
Uninsured	3 (1.3%)		2 (1.0%)	1 (3.2%)
Median Household Income by ZipCode					0.63
< $38,000	18 (8%)		21 (10.5%)	5 (16.6%)
$38,000-$47,999	33 (14.7%)		34 (17.1%)	6 (20.0%)
$48,000–62,999	52 (23.1%)		64 (32.2%)	10 (33.3%)
> $63,000	122 (54.2%)		80 (40.2%)	9 (30.3%)
Low Education (> 21% of adults not graduating from high school)	30 (12.8%)		25 (12.4%)	5 (16.1%)	0.56
Urban-Rural Continuum					0.65
Metro	203 (87.1%)		177 (87.6%)	26 (83.9%)
Urban	28 (12.0%)		23 (11.4%)	5 (16.1%)
Rural	2 (0.8%)		2 (1.0%)	0 (0%)
Distance Traveled, median, miles (IQR)	7.6 (4.2–16.9)		7.4 (4–16.5)	10.7 (5–25.6)	0.27
Charlson Score					0.41
0	196 (84.1%)		172 (85.1%)	24 (77.4%)
1	25 (10.7%)		21 (10.4%)	4 (12.9%)
>1	12 (5.1%)		9 (4.4%)	3 (9.7%)
Facility					0.33
Community Cancer Program	63 (27.0%)		56 (27.7%)	7 (22.6%)
Comprehensive Community Cancer Program	85 (36.4%)		70 (34.6%)	15 (48.4%)
Academic/Research Program	85 (36.4%)		76 (37.6%)	9 (29.0%)
Lymphoma Subtype					0.0001
DLBCL	47 (20.2%)		32 (15.8%)	15 (48.4%)
Burkitt	5 (2.1%)		4 (1.9%)	1 (3.2%)
Follicular	40 (17.2%)		36 (17.8%)	4 (12.9%)
Mantle Cell	8 (3.4%)		7 (3.5%)	1 (3.2%)
Marginal	102 (43.8%)		98 (48.5%)	4 (12.9%)
NHL NOS	14 (6.0%)		13 (6.4%)	1 (3.2%)
T-Cell	4 (1.7%)		3 (1.5%)	1 (3.2%)
Other	13 (5.6%)		9 (1.5%)	4 (12.9%)

DLBCL: diffuse large B-cell lymphoma, NHL: non-Hodgkin lymphoma, NOS: not otherwise specified. IQR: inter-quartile range.

Table 2:

Treatment characteristics, short term outcomes, and overall survival of patients treated with local resection (n=202) versus radical resection (n = 31).

	All (n=233)		Local (n=202)	Radical (n= 31)	p-value
Preoperative Chemotherapy	5 (2.1%)		3 (1.5%)	2 (6.4%)	0.07
Preoperative Radiation	2 (0.8%)		1 (0.5%)	1 (3.2%)	0.24
R0 resection	94 (40.3%)		77 (38.1%)	17 (57.8%)	0.07
Stage					0.001
I	181 (77.7%)		164 (81.2%)	17 (54.8%)
II	24 (10.3%)		15 (7.4%)	9 (29.0%)
III	5 (2.1%)		4 (1.9%)	1 (3.2%)
Unknown	23 (9.9%)		19 (9.4%)	4 (12.9%)
Length of Stay, d median (IQR)	0 (0–0)		0 (0–0)	6.5 (4–11)	<0.001
Post Operative Chemotherapy	46 (19.7%)		37 (18.3%)	9 (29.0%)	0.22
Post Operative Radiation	48 (20.6%)		43 (21.3%)	5 (16.1%)	0.63
30-day Readmission	6 (2.6%)		3 (1.5%)	3 (9.7%)	0.007
30-day Mortality	2 (0.8%)		1 (0.5%)	1 (3.2%)	0.12
90-Day Mortality	5 (2.1%)		3 (1.5%)	2 (6.4%)	0.07
5 Year Overall Survival	79.1%		81.4%	56.8%	0.06

IQR = inter-quartile range.

Perioperative Outcomes

Preoperative administration of chemotherapy (2.1%) or radiotherapy (0.8%) was rare, suggesting perhaps that the identified patients are not failures of chemotherapy, but represent a “surgery first” approach. Pre-operative therapy did not differ between local versus radical resection groups. The overall R0 resection rate was low (40%) and did not differ between groups (local excision 38.1% vs. radical resection 57.8%; p = 0.07). The most common approach to local excision was polypectomy (either by endoscopic or transanal approach (51.5%) followed by “local excision not otherwise specified” (26%). Adjuvant chemotherapy and radiation was only given in the minority of patients and did not differ between groups (Table 2). Overall, 19.7% of patients received chemotherapy post-operatively (local excision 18.3% vs. radical resection 29.0%; p =0.22). Among all patients, 20.6% received radiation (local excision 21.3% vs. radical resection 16.1%; p = 0.63). Not surprisingly, length of stay and readmission rate were significantly higher in the radical resection group, but rates of 30- and 90-day mortality were low and did not differ between groups (Table 2).

Overall Survival

228 patients (97.8%) survived 90 days and were included in the analysis of overall survival. Over a median follow up of 60.5 months, the 5-year overall survival was 79.1% in the entire cohort. Surgical approach was not associated with a significant difference in survival, but there was a clear trend toward decreased overall 5-year survival in the radical resection group (local excision 81.4% vs. radical resection 56.8%; p = 0.07), (Figure 1).

Figure 1-.

Kaplan Meier Curve for Overall Survival Stratified by Approach.

On multivariate analysis using a Cox Proportional Hazard model, local excision was not associated with overall survival (HR = 1.31; 95% CI: 0.63–2.72; p =0.47). Factors that remained significantly associated with 5-year overall survival included age > 65 years, Charlson comorbidity score > 0, and DLBCL or Other histologic subtype (Table 3). An analysis of survival by common histologic subtypes demonstrated follicular lymphoma to be associated with the best survival and DLBCL with worst (Figure 2).

Table 3:

Results of bivariate and multivariable Cox proportional hazards model analysis of patient, treatment, and tumor factors associated with overall survival.

Overall Survival	Bivariate			Multivariable
Variable	HR (95% CI)	P		HR (95% CI)	P
Local Resection	1.95 (0.97–3.94	0.06		1.40 (0.65–3.03)	0.39
Age > 65	0.29 (0.15–0.57)	0.0003		0.31 (0.15–0.61)	0.0007
R0 resection	1.28 (0.72–2.28)	0.40
Female Gender	1.74 (0.97–3.11)	0.06
Caucasian	0.55 (0.22–1.39)	0.55
Charlson > 0	0.29 (0.15–0.56)	0.29		0.44 (0.22–0.87)	0.01
Facility
Community	REF
Comprehensive	0.90 (0.88–1.78)	0.77
Academic	2.01 (0.88–4.54)	0.09
Lymphoma Subtype
Follicular/Marginal	REF			REF
DLBCL	0.24 (0.12–0.48)	<0.001		0.27 (0.14–0.55)	0.003
Other	0.33 (0.16–0.69)	0.003		0.37 (0.17–0.79)	0.01
Stage
I	REF			REF
II	1.29 (0.50–3.34)	0.58		1.33 (0.50–3.57)	0.55
III	0.17 (0.05–0.55)	0.003		0.16 (0.05–0.56)	0.004
Unknown	0.62 (0.27–1.41)	0.25		0.78 (0.31–1.92)	0.58

Figure 2-.

Kaplan Meier Curve for Overall Survival Stratified by Lymphoma Sub-type.

Discussion

We present a current, multi-institutional cohort of 233 patients undergoing surgical resection for primary rectal lymphoma. Our study reveals that most PRL patient undergo local excision and that local excision patients have equivalent rates of neo-adjuvant and adjuvant therapy administration, equivalent rates of R0 resection, and have at least equivalent overall 5-year survival. Overall survival was influenced by age, disease stage, comorbidity, and histologic subtype.

In contrast to other large published series of gastrointestinal lymphoma^3–6,9, our study is unique in that it focuses on a particular anatomic site and sub-analyzes surgical approach for non-metastatic, non-palliative patients. The largest study of primary colorectal lymphoma included 425 rectal lymphoma patients collected from the Surveillance, Epidemiology, and End Results (SEER) database from 1990–2014, including Stage IV disease and non-surgical patients⁹. Five-year survival for PRL patients was 61.4%, considerably lower than the 79.4% in our study, likely reflecting inclusion of older data and Stage IV patients. In the SEER study, multivariate analysis found that surgery did not influence overall survival compared to non-surgical treatment in colorectal lymphoma patients. These results are similar to a prior SEER analysis indicating that surgery only provided a survival benefit among early-stage patients with right-sided lesions⁶. Similarly, improved survival with surgery was only seen in patients with B-cell histology in a relatively large Korean series of 581 patients with intestinal lymphoma⁴. Lymphoma histologic subtype was associated with survival in this study as well. As lymphoma can arise from any stage of lymphocyte maturation, histologic variety is enormous. Early consultation with a multidisciplinary oncology team and expert pathologist may improve treatment selection in these rare patients.

Compared to the colon, the rectum uniquely allows full thickness local excision of neoplastic lesions. Standard transanal excision^13–14, transanal endoscopic microsurgery¹⁵, and endoscopic resection¹⁶ have all been described for lymphoma in small case series. In rectal adenocarcinoma, these surgical approaches have an accepted role in the treatment of early, low-risk cancers with an increased rate of local recurrence, but equivalent overall survival and obvious benefits in organ preservation and surgical morbidity¹⁷. For PRL, there are no prospective comparisons of local versus radical resection for PRL, and no large database study prior to the current work sub-analyzes survival by surgical approach. Given the rarity of PRL it is unlikely that high quality prospective data are possible in the near future. Our data clearly show equivalent overall survival in patients undergoing local excision versus radical resection and trend toward worse survival for radical resection patients. Considering the unclear survival benefit of surgery in colorectal lymphoma in general, and the equivalent outcomes of local excision patients in our study, we propose that surgeons should pursue local resection for PRL patients whenever technically possible. In terms of therapy beyond surgery, in this study the rate of adjuvant chemo or radiotherapy administration was low and was not different between patients with and without an R0 resection. Reasons for the low rate of adjuvant therapy administration could include patient preference, fitness, provider choice, and perception of therapeutic benefit.

Our study has several limitations due to the retrospective and de-identified nature of the data. As we do not have reliable data surrounding surgical decision-making, there is the potential for selection bias in the survival analysis. Lack of data on tumor size and exact location is particularly important as both may influence selection of surgical approach. Additionally, while we were able to exclude palliative operations, lack of coding makes it impossible to determine whether surgery was undertaken for urgent indications, such as bleeding or perforation, or for oncologic purposes. Finally, only overall survival is captured by the NCDB. An increase in local recurrence, as has been observed in local excision in rectal adenocarcinoma, would not be captured by the study.

Conclusion

Surgical patients with PRL who undergo a local excision have similar rates of R0 resection, chemotherapy and/or radiation administration, and at least equivalent overall survival as compared to patients with radical resection. Prediction of surgical patient outcome may be improved by these results: overall survival is not influences by surgical approach, but depends on age, comorbidity, disease stage, and histologic sub-type. Although we recognize the limitations of this study, these data may assist physicians in surgical planning, patient counselling, and designing optimal treatment algorithms for this rare malignancy.