Clinicopathological Study of Primary Pediatric Gastrointestinal Lymphomas

Abstract

Background:

Primary gastrointestinal lymphomas (PGIL) are uncommon in children and account for <5% of all pediatric malignancies. The objective of our study was to analyze the prognostic factors of pediatric PGILs with reference to its histological subtypes, stage, and outcomes using immunohistochemistry.

Materials and Methods:

Twenty cases were studied over 11 years. Clinicopathological profiles, histological details, and immunohistochemical (IHC) profiles were analyzed.

Results:

The mean age at the presentation was 6 years. Using IHC stains (CD CD10, CD19, CD20, CD3, terminal deoxynucleotidyl transferase, BCL 2 and 6, PAX 5, and MUM1), diffuse large B-cell lymphoma (DLBCL) was most prevalent (45%), followed by Burkitt lymphoma (35%) and lymphoblastic lymphoma (20%). DLBCLs (9) were classified using the Han's algorithm. Six were activated B-cell and 3 were germinal center B-cell subtypes. The cases of lymphoblastic lymphoma and those in Stage I disease had the best prognosis.

Conclusion:

Pediatric PGILs have variable presentation, outcomes, and diverse treatment modalities depending on the histological subtypes. A panel of IHC stains can be a useful adjunct for the categorization and prognosis of pediatric PGILs.

INTRODUCTION

Primary gastrointestinal lymphomas (PGIL) are infrequent in children and account for <5% of all pediatric malignancies. Lymphomas are one of the most common tumors affecting pediatric population with PGILs being the most common extranodal lymphoma.[1,2,3]

The worldwide incidence of pediatric lymphomas is quite variable. PGILs are less common than secondary gastrointestinal (GI) lymphomas where GI tract is affected in addition to other sites.[4,5] Early diagnosis and management are important in view of its marked chemosensitivity. The accurate categorization is also important so that individualized treatment can be administered and outcomes vary according to the subtypes.[6]

Primary PGILs have usually present with long-standing symptoms of abdominal pain and subacute intestinal obstruction. Often these tend to get treated conservatively for the common conditions such as helminthic infestations or mesenteric lymphadenitis. Therefore, these cases often present at an advanced stage; sometimes, as acute abdomen due to intussusception can are associated with aggressive behavior and poor prognosis. Although there are data available on adult GI lymphomas, there is sparse literature available on pediatric PGILs.[7,8]

The objective of our study was to analyze the clinicopathological spectrum and prognosis of pediatric PGILs with special reference to its anatomical distribution, histological subtypes, and outcomes using a panel of immunohistochemical (IHC) stains.

MATERIALS AND METHODS

This was a retrospective study performed over a period of 11 years, following the guidelines and clearance from the Institutional Ethical Committee. There were 20 cases of pediatric GI lymphomas studied from January 2009 to October 2019. Patients <12 years of age, with a diagnosis of PGIL were included in the study. According to Dawson's criteria, primary lymphoma of GI tract includes the cases without any superficial adenopathy at the time of diagnosis, no evidence of mediastinal lymphadenopathy on chest radiography, normal blood cell counts, without involvement of the liver and spleen and with involvement of only regional lymph nodes at laparotomy.[9]

The presenting symptoms were documented. The extent of the disease was established by physical examination, baseline complete hemogram, serum electrolytes, liver function tests, lactate dehydrogenase (LDH) as tumor bulk indicator, uric acid, abdominal ultrasound, and/or contrast-enhanced computed tomography scan of the abdomen and bone marrow biopsy.

The archived blocks were reviewed. The histopathological analysis of all the 20 cases was done on H and E stains. An algorithmic approach using both morphology and IHC expression was carried out. The morphological parameters we assessed were: Architectural pattern of the lymphoid compartment (diffuse or follicular), population of cells (single or mixed), size of cells (small, intermediate, or large), and nuclear features. The final histological subtype was ascertained according to the World Health Organization classification of tumors of Hematopoietic and Lymphoid system–2017 Revised Edition.[10]

The panel of IHC markers used was: CD 10, CD 19, CD20, CD3, terminal deoxynucleotidyl transferase (TdT), BCL2, BCL6, PAX 5, MUM 1, and KI 67, using Dako, prediluted FLEX RTU antibodies on automated platform, DAKO OMNIS. The following clones were used: Monoclonal mouse anti-human CD10 (Clone-56C6, Code-IS648), monoclonal mouse anti-human CD19 (Clone-LE-CD19, Code-IS656), monoclonal mouse anti-human CD20cy (Clone-L26, Code-IS604), polyclonal rabbit anti-human CD3 (Code-IS503), polyclonal rabbit antiTdT (Code-IS001), monoclonal mouse anti-human BCL2 Oncoprotein (dilution, Clone: 124, Code-IS614), monoclonal mouse anti-human BCL6 Protein (Clone-PG-B6p, Code-IS625), monoclonal mouse anti-human B-cell-specific activator protein (Clone-DAK-Pa × 5 Code-IS650), monoclonal mouse anti-human MUM1 Protein (Clone-MUM1p, Code-IS644), and monoclonal mouse anti-human Ki-67 Antigen (Clone-MIB-1, Code-IS626). Appropriate positive and negative controls were examined simultaneously.

The expression in 30% of tumor cells was considered positive for all IHC markers.[11] Negative staining was defined by less expression regarding either percentage or intensity or both membranous positivity was noted in CD 19 and CD 20, whereas Cd3. TdT, BCL2, and PAX5 revealed nuclear positivity. The Ki-67 labeling index was done by counting the positively stained nuclei, expressed as the mean percentage of total cells counted in 5 high-power fields (400 X). The Hans's algorithm using CD10, BCL6, and MUM1 was applied whenever required.[10] CD10 positivity was strictly membranous, but the rest showed nuclear expression.

The tumors were staged according to the St. Jude's staging system.[6] Follow-up data were available in only 14 cases, and 6 patients were eventually lost to follow-up. The duration of the follow-up period varied from 15 days to 37 months. Comprehensive follow-up data included appropriate biochemical parameters, tumor stage at diagnosis, conferred treatment modality, and the current disease status for any recurrence or metastasis.

Statistical analysis

For quantitative variables, the median and mean values with standard deviation were provided. The correlation between the categorical variables was assessed by the Chi-square test. Kaplan-Meier curves were used for the survival analysis. Statistical analyses were done by using the SPSS software version 20.0 (IBM, Armonk, New York, USA), and P < 0.05 was considered to be statistically significant. All the P values found were two tailed.

RESULTS

A total of 20 cases were evaluated over a period of 11 years. The age at the presentation ranged from 1 year to 11 years, and the mean age at presentation was 6.1 ± 2.8 years. Fifteen out of 20 cases (75%) were boys. The principal presenting symptoms were intestinal obstruction (70%) and abdominal mass (65%), followed by abdominal pain (20%), diarrhea (10%), and bloody stools. B symptoms and elevated LDH were found in 45% cases. Ileocoecal junction (50%) was the most common site of involvement, followed by the small intestine (35%) and large intestine (15%) alone. The clinicopathological profiles of these cases are summarized in Table 1. The first six cases have been published earlier as a part of a smaller series.[7]

Table 1.

Clinicopathological spectrum of primary pediatric gatrointestinal lymphomas

Parameters	n=20	Lymphoblastic lymphoma (n=4)	Burkitt lymphoma (n=7)	DLBCL (n=9)	P
Mean age		4±0.81	7.21±2.58	6.16±3.10	0.18
Male	16	2	7	7	0.05*
Female	4	2	0	2
B symptoms
Present	9	1	6	2	0.02*
Absent	11	3	1	7
LDH
Elevated	9	1	6	2	0.02*
Normal	11	3	1	7
Lesion
Single	20	4	7	9	NS
Multiple	0	0	0	0
Macroscopic type
Superficial	0	0	0	0	0.40
Mass forming	17	4	5	8
Diffuse infiltrating	3	0	2	1
Unclassified	0	0	0	0
Stage
I/IIE	17	4	5	8	0.40
III/IV	3	0	2	1
Treatment
Surgery alone	0	0	0	0	0.74
C+S	18	4	6	8
S±C±R	2	0	1	1
Supportive	0	0	0	0

LDH: Lactate dehydrogenase, DLBCL: Diffuse large B-cell lymphoma, NS: Not significant

On macroscopic examination, majority of the lesions were mass forming (17/20; 85%), and remaining three cases were diffuse infiltrating lesions. Following histological analysis and IHC staining, the most common (9/20) subtype was diffuse large B-cell lymphoma (DLBCL). There were seven cases of Burkitt lymphoma, and remaining four cases were lymphoblastic lymphoma [Figures 1-3]. The DLBCL cases were further classified according to Han's algorithm and 6 out of 9 cases were found to be of activated B-cell subtype, and rest 3 was of germinal center B-cell subtype. The IHC profile of the DLBCL cases showed positivity for CD 20 and PAX 5 in all the cases; variable positivity for CD 19, and BCL 2 was positive in 8 out of 9 cases. CD10 was expressed in three cases; in the remaining 6 cases, it was negative. MUM 1 was positive in all six cases which were CD10 negative. All the DLBCLs were negative for TdT, Bcl 6, and CD 3. The Ki-67 labeling index had a mean value >50% (55.77 ± 6.96%). All the Burkitt lymphomas were positive for CD 10, CD 19, CD 20, BCL 6, and PAX 5 and were negative for TdT, CD 3, and BCL2. The mean Ki-67 labeling index was close to 100% (97.71 ± 2.36). Lymphoblastic lymphomas were positive for TdT and CD 3 were positively expressed in 75% of cases. Variably expression was observed for BCL2 and BCL 6, whereas CD 10, CD 20, and PAX 5 were negative. The mean Ki-67 index was higher than DLBCL but lower than Burkitt lymphoma (62 ± 2.94%) [Table 2].

Figure 1.

(a) Gross photograph of a case of diffuse large B-cell lymphoma arising from the ileocecal junction. (b) Low-power view of the mass which shows a tumor mass composed of sheets of uniform lymphoid cells (H and E, ×100). (c) High power of the same showing large lymphoid cells with prominent nucleoli (H and E, ×400). The tumor cells show immunopositivity for CD10 (d) (×100) and CD20 (e) (×100) (f) Ki67 labeling index of >50% (×100). The tumor cells show positivity for MUM 1 (g) and are negative for BCL 2 (h) and BCL6 (i) (×100)

Figure 3.

(a) Gross photograph of a case of lymphoblastic lymphoma arising from the ascending colon. (b) Low-power view shows monotonous population of lymphoid cells infiltrating the mucosa (H and E, ×100). (c) High-power view of the same noted highlighting the colonic glands (H and E, ×400). The tumor cells show positivity for TdT (d) and CD 3 (e) (×100). (f) The Ki 67 labeling index is of >50% (×400). (g) The tumor cells are positive for BCL 2 (×100). (h) High-power view of BCL 2 (×400). (i) The tumor cells are negative for CD 20 (×400)

Table 2.

Immunohistochemical profiles of pediatric gastrointestinal tract lymphomas

IHC markers	Lymphoblastic lymphoma (n=4)	Burkitt lymphoma (n=7)	DLBCL (n=9)	P
TdT
Positive	4	0	0	<0.0001
Negative	0	7	9
CD 20
Positive	0	7	9	<0.0001
Negative	4	0	0
CD3
Positive	3	0	0	0.009
Negative	1	7	9
CD 10
Positive	0	7	3	<0.0001
Negative	4	0	6
Bcl2
Positive	2	0	8	0.002
Negative	2	7	1
Bcl6
Positive	1	7	0	0.0076
Negative	3	0	9
MUM1
Positive	0	4	6	0.0096
Negative	4	3	3
PAX 5
Positive	0	7	9	<0.0001
Negative	4	0	0
MeanKi67LI%	62±2.94	97.71±2.36	55.77±6.96	<0.0001

DLBCL: Diffuse large B-cell lymphoma, IHC: Immunohistochemical, TDT: Terminal Deoxynucleotidyl Transferase, GIT: Gastrointestinal tract

Figure 2.

(a) Gross photograph of a case of Burkitt lymphoma arising from the terminal illeum. (b) Low-power view of the tumor, illeal epithelium is noted at the top. The tumor cells are composed of sheets of lymphoid cells with high N: C ratio, interrupted by pale macrophages (H and E ×100). (c) High power of the same (H and E, ×400). The cells show intense positivity for CD10 (d), positivity for CD 20 (e) and are negative for CD 3 (f) (×400). (g) Cells show PAX 5 positivity (×100). (h) The tumor cells show a high Ki 67 labeling index of almost >90% (×100). (i) The cells are negative for BCL 6 (×400)

According to St. Jude's classification of childhood lymphomas, majority of the cases presented in Stage I (80%), a single case presenting in Stage II was lost to follow-up, and the remaining 3 cases presented in higher stages. Resection followed by chemotherapy was the principal treatment administered. In addition, two cases received radiotherapy as an adjunct. Among the 20 cases, six cases were lost to follow-up. The remaining cases were followed up for a period of 15 days to 37 months. Kaplan-Meier curves were drawn in relation to the histological subtype and stage. Among the histological subtypes, lymphoblastic lymphoma showed best survival (mean = 27.75 months; range = 15–37 months), followed by DLBCL (mean = 9.8 months; range = 1–18 months) and Burkitt (mean = 4.9 months; range = 0.5–18 months) [Figure 4]. Stage at presentation was found to be a key factor for survival and Stage I patients were found to have the best prognosis [Figure 5].

Figure 4.

Kaplan-Meier curves showing the survival analysis in relation to histological subtypes

Figure 5.

Kaplan-Meier curves showing the survival analysis in relation to stage at presentation

DISCUSSION

Lymphomas constitute the third-common cause of pediatric malignancies.[7] Pediatric PGILs represent a heterogeneous group of diseases with regard to its histological subtypes. Their response to therapy and outcomes are also different.[12,13] Most of the pediatric lymphomas are nodal lymphomas. Among the extranodal sites, GI tract is the most common. Other extranodal sites of lymphomas in children include ovary, bone, mediastinum, and infrequently lungs and the liver.

PGIL is uncommon in infants, toddlers, and adolescents. In our clinicopathological analysis of 20 cases of childhood PGILs, the mean age at the presentation was 6.1 ± 2.8 years, and a male preponderance was noted. Most of the other case series also report similar age/sex profiles, and a comparative data are summarized in Table 3.[8,14,15,16,17]

Table 3.

A comparative analysis of prior published series of pediatric primary gastrointestinal lymphomas

Parameters	Present study	Indian study Qadri et al.[8]	Asian study Khurshed et al.[14]	Egyptian study Morsi et al.[15]	USA study Naeem et al.[17]	Study among adult population Raina et al.[19]
Total number of cases	20	11	60	43	452	77
Duration of study (years)	11	12	2	7	41	15
Age range (years)	1-11	1-14	0-19	0.4-17	0-19	9-80
Study design	Retrospective	Combined retrospective and prospective, descriptive	Cross-sectional	NCI Cairo University data based	NCI SEER data based	Retrospective
Mean age (years)	6	6.6	11.6	6.6	11	32
Male: female	4:1	4.5:1	12:1	2.3:1	3.2:1	2.2:1
Commonest presenting symptoms	Intestinal obstruction	Abdominal pain	Abdominal mass	Abdominal pain	-	Abdominal pain
Commonest site	Ileocecal junction	Terminal ileum	Small bowel	Ileocecal junction	Ileum	Stomach
Commonest histology	DLBCL	DLBCL	BURKITT	BURKITT	BURKITT	DLBCL
Commonest stage	Stage I/II	Stage II	Stage III	Stage II	Stage I/II	Stage II

DLBCL: Diffuse large B-cell lymphoma, NCI: National cancer institute, BURKITT: Burkitt, SEER: Surveillance, epidemiology and end results

In our study, the common presenting symptoms were intestinal obstruction and abdominal mass. Pediatric PGILs most commonly involve the small and the large gut, in contrast to adults where the stomach is the most common site of involvement.[18,19] Most of the papers have shown that ileocoecal junction to be the most common site of involvement; followed by the small intestine.[7,15] Low-grade lymphomas such as MALToma and IPSID have not been reported in the pediatric age group. In our study too, ileocoecal junction was the most common site of involvement (50%), followed by the small intestine (35%) and large intestine (15%).

Most of the published series report Burkitt lymphoma to be the most common histological subtype in children.[15,17,20,21,22] However, data among the Asian population have reported DLBCL to be the most common histological subtype of pediatric PGILs.[7,8,14] In our study, DLBCL was the most common histological subtype (45%), followed by Burkitt lymphoma (35%) and lymphoblastic lymphoma (20%). It is difficult to explain this variation in distribution; however, our study has been based on immunohistochemistry alone in the absence of flow-cytometric analysis and molecular confirmation. Regional differences in the frequency of EB virus-induced neoplasias could be an incriminatory factor. However, the confirmatory translocation t (8;14) for Burkitt lymphoma was not done. Similarly, FISH analysis for CMYC and 11q for DLBCL was not performed. Applying the recently described Han's algorithm, mainly used for the nodal lymphomas, majority of the cases of DLBCL were activated B-cell type, which explains the poor outcome. This subclassification is extremely important in view of the specific chemotherapeutic regimens, predictive response, and outcome. In a resource-constraint setting, immunohistochemistry is often the only tool available, before starting therapy.

Pediatric PGILs are highly aggressive tumors with early dissemination and lymph node involvement.[16,22] In our study, 85% of the cases presented in Stage I/II and 15% in Stage III/IV. The stages I and II (limited stage) have shown definite survival advantage.[17,20,22,23]

The management protocol for pediatric PGILs remains controversial. Recent studies showed chemotherapy to be an effective mode of treatment in all stages of the disease.[17] The stage at the presentation of the disease and the extent of involvement still remains the most important criteria for determining survival advantage.[7,15]

In our current study, Stage III patients were treated with limited resection followed by chemotherapy and were the most common mode of treatment offered. A case of Burkitt lymphoma and another case of DLBCL were treated with chemoradiation. In our study, lymphoblastic lymphoma had the best outcome, followed by DLBCL, whereas Burkitt showed the worst outcome.

CONCLUSION

DLBCL was the most common subtype of pediatric GI lymphoma, followed by Burkitt lymphoma and lymphoblastic lymphoma. The best outcome was associated with lymphoblastic lymphoma followed by DLBCL, whereas Burkitt lymphoma had the worst outcome. Although molecular testing is desirable, in a resource-constraint setting, a panel of IHC markers is a useful adjunct for the categorization and prognosis of pediatric PGILs.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.