Alimentary Tract Malignancies in Children

By Colin A.I. Bethel, Nishith Bhattacharyya, Carol Hutchinson, Frederick Ruymann, and Donald R. Cooney Columbus, Ohio

l Alimentary tract malignancies in children are unusual. From 1952 to 1996,54 patients (55 cases) underwent surgery at The Children’s Hospital, Columbus for intestinal malig- nancy. Their records were reviewed retrospectively. The mean age at diagnosis was 9.3 years (range, 1 to 17 years). There were 35 boys and 19 girls (M:F ratio, 1:8). Mean follow-up was 108 months. Laparotomy was performed in all but one child. The primary tumor sites included the colon (31 cases), small bowel (12 cases), appendix (9 cases), and stomach (3 cases). Seventy-five percent of the tumors were non-Hodgkin’s lymphomas (41 cases), followed by appendi- teal carcinoid (9 cases), colon adenocarcinoma (3 cases), and gastric sarcoma (2 cases). Lymphoma occurred in 28 of 31 nonappendiceal large bowel tumors and was the only tumor type seen in the small intestine. There were 21 Burkitt’s, 11 lymphoblastic, 6 small cell, and 5 large cell lymphomas. There was a statistically significant increase in the frequency of small bowel lymphoma after 1982 (IO of 20) in comparison with that before 1982 (2 of 211, P < .05. This coincided with an increase in Burkitt’s lymphoma from 5 of 21 (pre-1982) to 16 of 20 (post-19821, and a decrease in lymphoblastic lympho- mas from 7 of 21 (pre-1982) to 2 of 20 (post-19821, P < .05. Localized disease and complete resection favored survival in lymphoma, whereas age, sex, and urgency of operation had no influence on survival. This study highlights the predomi- nance of lymphoma as the most common small and large bowel tumor and highlights the emergence of nonendemic Burkitt’s as a major entity in pediatric intestinal malignancy. Copyright o 1997 by W.B. Saunders Company

INDEX WORDS: Alimentary tract, intestine, malignancy, lym- phoma, Burkitt’s lymphoma.

P RIMARY ALIMENTARY tract malignancies in the pediatric age group are rare. In the late 1950s the

percentage of all childhood tumors, benign and malig- nant, which arose from intestine was reported to be well under 5%.Q Today, the true incidence of childhood alimentary malignancy remains unknown, but the paucity of clinical series on the subject speaks to the rarity of the entity. In the tumor registty at The Children’s Hos- pital, Columbus, there were 55 cases of primary intesti- nal malignancy from a total of 4,547 tumor registrants (1.2%) over a 44-year period.

From the Division of Pediatric Surgery, Department of Surgery, The Children k Hospital The Ohio State University, Columbus, OH.

Presented at the I996 Annual Meeting of the Section on Surgery of the American Academy of Pediatrics, Boston, Massachusetts, October 26-30, 1996.

Address reprint requests to Colin A.I. Bethel, MD, Children’s Hospital, ED 378, 700 Children’s DC Columbus, OH 43205-2696.

Copyright 0 1997 by WB. Saunders Company 0022-3468/97/3207-0014$03.00/O

1004

Review of the literature over the last 30 years demon- strates widespread acceptance of lymphoma as the most common small bowel malignancy.2-4 In Pickett’s 1967 review, the majority of worldwide cases of colon malig- nancy were adenocarcinoma, with only a few cases of lymphoma.2 This contrasts the experience in our institu- tion in which lymphoma has been the most frequent cancer in both the small and large bowel. The role that nonendemic (American) Burkitt’s lymphoma has had on the incidence, anatomic distribution, and behavior of alimentary lymphomas has not been described.5,6

We have reviewed retrospectively our cases of alimen- tary malignancy with regard to presenting symptomatol- ogy, preoperative diagnosis, operative findings, treat- ment, and outcome. We hoped to obtain incidence data for each of the histologies seen and to identify any changes that may have occurred in the overall behavior of childhood alimentary malignancy in recent decades.

MATERIALS AND METHODS

Patients and Data Sources

All cases of alimentary tract malignancies presenting between 1952 and 1996 were reviewed. Data were accumulated in a retrospective fashion from patient charts, operative reports, and pathology reports. At our institution. lymphomas have been typed histopathologically accord- ing to the Working Formulation of the American Joint Committee on Cancer (AJCC). Tumors initially designated as “reticulosarcoma” or “lymphosarcoma” in earlier classifications were reclassified according to the AJCC scheme. Staging was by AJCC criteria.

Patients were grouped on the basis of localized versus extensive disease. Localized disease was defined as tumor confined to the bowel, adjacent mesentary, or adjacent structures amenable to resection along with the main bulk of the tumor itself. In addition, resections were defined as complete gross resections if all gross tumor was excised, or incomplete gross resections if tumor was left in situ. Those patients undergoing incomplete gross resection and those undergoing biopsy alone were compared as a single group with those undergoing complete gross resection.

Statistical Analysis

Analysis using x2 was used to determine statistical significance between mcidence of lymphomas before and after 1982, as well as to determine the significance of tumor burden, surgical procedure and urgency of operation on survival. Product-limit and log-rank analysis were used to evaluate survival differences. Paired t tests were used to evaluate the significance of age and sex on survival.

A median split was used to divide the lymphoma patients into two groups: those whose disease was diagnosed before 1982, and those whose disease was diagnosed after 1982. This was helpful in identifymg changes in the epidemiology of alimentary tract lymphoma over the last several decades.

Journa/ofPediatricSurgery,Vol32, No 7 (July),1997:pp 1004-1009

ALIMENTARY TRACT MALIGNANCIES 1005

RESULTS

Pafient Characteristics

There were 55 cases (in 54 patients) of alimentary tract primaries. The total number of malignancies for this period was 4,547. making the relative incidence of alimentary primaries 1.2%. Thirty-five were boys, and 19 were girls. Of the lymphoma patients, 32 were boys and 9 were girls. The mean and median ages for all patients with alimentary malignancy were 9.3 and 9.5 years, respectively (range, 1 to 17 years). The mean and median ages for lymphoma patients were 7.7 and 7.5 years, respectively (range, 2 to 16 years). No statistically significant correlation between age of presentation and incidence of any given histological subtype could be drawn.

Presenting Symptoms and Preoperative Diagtlosis of Alimentary Malignancies

Symptoms and preoperative diagnoses are listed in Table 1. The most common symptom was abdominal pain, followed by vomiting. The asymptomatic abdomi- nal mass was unusual. The mean duration of symptoms before diagnosis was 5.9 weeks (range, 24 hours to 6 months). The most common preoperative diagnosis was abdominal mass, followed by appendicitis and intussus- ception.

Histology ofAlittletztarq Tract Tumors

Non-Hodgkin’s lymphomas comprised 41 of the 55 (74.%) alimentary tract malignancies (Fig 1). Twenty- eight of these were in colon. 12 were in small intestine, and 1 was gastric. The second most common tumor was carcinoid (nine cases); all these were appendiceal. There were three colon adenocarcinomas and two gastric sarco- mas. Burkitt’s, lymphoblastic, small cell, and large cell lymphomas were observed in decreasing frequency in large bowel. Burkitt’s was also the most common small intestinal tumor (Fig 2).

Table 1. Symptoms and Preoperative Diagnoses

Symptoms ” I%) Preoperative Dlagnws n (%)

Abdominal pain 39 (71) Abdominal mass 18 (33) Vomltlng 26 (47) Appendicitis 15 (27) Weight loss 10 (18) lntussusceptlon 13 (24)

Diarrhea 9 (16) ObstructIon 4 (7) Constipation 8 115) lncldental finding* 1 (2) Anorexia 6 (II) No avarlable rnformatron 4 (7) Fever 5 (91 Rectal bleeding 4 (7) Mass wlthout symptoms 2 (4)

‘Appendiceal carcinoid found at time of gastrectomy.

Small bowel lymphom<2 1.8%) Carcinoid (16.3%)

I I Gastric sarcoma Gastric

Colon lymphoma (3.6%)

lymphoma (50.9%) (1.8%)

Fig 1. Histology of alimentary tract malignancies.

Surgeu?, atzd Adjuvarzt Therupq

All but one patient, who had rapid clinical deteriora- tion. underwent laparotomy. Surgery was urgent in 32 (22 bowel resections. 8 appendectomies, 2 biopsies); and elective in 22 (12 bowel resections, 10 biopsies). There were three second-look procedures in patients who had lymphoma (a cecal lymphoblastic lymphoma with omen- tal implants, an unresectable cecal Burkitt’s, and an extensive ileal Burkitt’s). Chemotherapy and radiation protocols varied during the study period. Thirty-six of the lymphoma patients received chemotherapy. and 11 under- went radiation.

Lymphoma

Figure 3 shows that the frequency of alimentary lymphomas in small intestine increased from 2 of 20 (10%) before 1982 to 10 of 20 (50%) after 1982 (x” = 7.62, P < .Ol). Similarly, the incidence of Bur-

11 11 10 1

g9 9

.s? 8

z 8 a7

'"06 7

25 6

I

5 E4 23 4

3 2 2 1 1

Large bowel Small bowel

w BUrkitt‘s Smali cd IsB Carcmoid

0 Lymphoblastic 0 Large cell pig Adenocarcinoma

Fig 2. Distribution of large and small bowel malignancies.

1006 BETHEL ET AL

post-l 982 pre-1982

/----A

Fig 3. Distribution of large and small bowel lymphomas pre- and post-1982. The number of small bowel lymphomas after 1982 was statistically higher than that before 1982 (P < .Ol). There was a statistically significant increase in the percentage of Burkitt’s after 1982 (P < .Ol). Conversely, the percentage of lymphoblastic lympho- mas decreased significantly from the pre-1982 to post-1982 periods (P < ,011.

kitt’s lymphomas increased from 5 of 20 (25%) before 1982 to 16 of 20 (80%) after 1982 (x2 = 10.03, P < .Ol). Ninety percent of the small bowel lymphomas presenting after 1982 were of Burkitt’s type. Survival by stage and histological subtype are depicted in Table 2. The overall survival for all alimentary lymphomas was 6 l%, and that for the Burkitt’s sub group was 71% (Fig 4). Overall lymphoma survival was 43% before 1982, and was 75% after 1982. As depicted in Table 3, there were 24 survivors and 17 nonsurvivors in the lymphoma group. Survival was higher in those with localized versus extensive disease, P < .Ol. The performance of urgent versus elective laparotomy did not impact on survival. Survival rates after complete gross resection were higher that those after incomplete gross resection or biopsy alone, P < .Ol.

Carcinoid

Eight of nine patients (seven girls; two boys) who had a carcinoid had symptoms of appendicitis ranging from 18 hours to 2 days in duration. A single carcinoid was

Fig 4. Actuarial survival of Burkitt’s and non-Burkitt’s lymphomas. Survival in Burkitt’s lymphoma for the entire period of study was 71%, and survival in all other lymphomas was 61%. A statistically significant difference in survival between these two groups could not be demonstrated.

discovered incidentally in a patient with a synchronous gastrointestinal stromal tumor of the stomach. Eight patients had localized appendiceal tumors of less than 2 cm in diameter and were cured by simple appendectomy. Meso-appendiceal extension was present in one requiring right hemicolectomy. In six cases there were polymorpho- nuclear cells in the appendix with lumenal obstruction by the tumor. There were no cases of metastasis or carcinoid syndrome and no adjuvant treatment was given.

Colonic Adenocavcinoma

Both patients with mutinous adenocarcinoma died. Pain and fever had been present for 3 weeks in a 16-year-old patient. A left colon mass and diffuse carcino- matosis were found. Biopsy, omentectomy and colos- tomy were performed. The patient died after 2 months without adjuvant treatment. The second patient was a 15-year-old who had 3 months of abdominal pain and weight loss. Multiple peritoneal implants were found; right hemicolectomy was performed. The patient died after a 3-week course of 5-flourouracil, having refused further chemotherapy. The third patient was an ll-year- old who had adenocarcinoma in a sigmoid polyp diag- nosed after rectal bleeding. Polypectomy was curative. No patients had familial polyposis or inflammatory bowel disease.

Table 2. Lymphoma Survival by AJCC Stage

I II III IV

Alwe Dead Alive Dead AlWe Dead AllW Dead

Burkitt’s 6 6 3 2 1 3 Lymphoblastic 2 5 1 - - 3

Small cell - 1 2 - - 3 Large cell 1 - - - - 2

NOTE. AJCC Staging is defined as follows: I, single lymph node (LN) region or single extralymphatic (EL) organ/site; II, two or more LN regions on same side of draphragm or localized involvement of single EL organ/Me; III, LN regrons on both sides of the diaphragm with or wrthout localized involvement of EL organ/site, spleen, or both, IV, disseminated involvement of one or more EL organs with or without associated LN

involvement or associated EL organ involvement with distal nodal involvement (A) asymptomatrc, (B) greater than 10% weight loss or temperatures greater than 38” or night sweats.

ALIMENTARY TRACT MALIGNANCIES

Table 3. Outcome in Alimentary Lymphoma

SllWlVOE NOW3AWlVOKS

(n = 24) (n = 171 x2 P Value

Localized

Extensive Urgent

Elective CGR

IGR Biopsy

Chemotherapy

Radiation

21

3 16

8 20

3

20

4

3 16.15 co1

14 7 0.81 NS

9 2 15.66 <.Ol

9 5

16

Abbreviations: CGR, complete gross resection; IGR, incomplete

gross resectlon.

Gastric Sarcornatous Tumors

The first of these was an epithelioid leiomyosarcoma with liver metastasis in a 15-year-old boy. After 1 year of vomiting, a gastric mass was found on upper endoscopy findings. Subtotal gastrectomy was performed. After chemotherapy, the liver nodules were resected and found to represent areas of necrosis with no active tumor, At 30 months the patient was alive, and had undergone cryosur- gery for recurrent liver metastases. The second patient was a 17-year-old girl who had gastrointestinal stromal tumor who presented with abdominal pain. A gastric mass was seen on contrast radiographs. Subtotal gastrectomy and resection of a liver nodule, without chemotherapy, were curative.

DISCUSSION

In our institution, alimentary tract primary malignan- cies accounted for 1.2% of all cases of pediatric cancer. This has not changed in the last 40 years. In 1960, the percentage of childhood malignancies with bowel prima- ries was 1% .’ The low incidence in children contrasts that in adults in whom 22% of cancers are attributable to the alimentary tract.*

Despite the presence of symptoms long before surgery in the majority of patients, the diagnosis of malignancy was most often made at urgent exploration. This suggests that a delay in diagnosis until intussusception, right lower quadrant pain, and/or obstruction require urgent interven- tion. Despite these delays, a higher percentage of patients explored urgently underwent complete tumor resection in comparison with those explored electively. In addition, there was no demonstrable difference in survival between the two groups (Table 3).

Lymphoma is by far the most common bowel primary malignancy in children. It has long been recognized that lymphomas are the most common small bowel malig- nancy, but the clear predominance of lymphoma in the large bowel, as demonstrated in this study, has not been reported.4 It is important to recognize. therefore, that colonic lymphoma was by far the most common malig-

nancy observed in our series, and before 1982, almost all the lymphomas of the gastrointestinal tract were colonic (Fig 3).

The shift in anatomic distribution of lymphomas from a primarily colonic location before 1982 to an even distribution between small and large bowel after 1982 coincided with the increase in the incidence of Burkitt’s lymphoma in our hospital and in the United States. Kaufman et all9 reported more small than large bowel primary malignancies in 11 patients with Burl&t’s lym- phoma. Whether Burkitt’s lymphoma occurs more often in small bowel than in colon remains to be demonstrated in larger series. The role that American Burkitt’s lym- phoma has played in the observed shift from large to small bowel primary malignancies in the United States remains unclear. Similarly, the mechanism behind this shift remains unknown.

Lymphoma survival was greatest for low-stage disease and correlated favorably with localized disease and complete gross resection (Tables 2 and 3), findings consistent with those from other series.‘O-l3 The overall survival in Burkitt’s lymphoma appeared to be higher (though without statistical significance, probably related to sample size), than survival in the non-Burl&t’s lympho- mas (Fig 4). It is difficult to determine whether this represents favorable biology of Burkitt’s lymphomas, or whether this is a manifestation of improved survival in the latter half of the study period as a result of better adjuvant therapy.

Appendiceal carcinoids in children most commonly present with symptoms of appendicitis. In this series, histological evidence of appendicitis was present in the majority of cases. The “appendicitis” is presumably the result of lumenal obstruction by the tumor. However, two of the patients had right lower quadrant symptoms without appendiceal inflammation, demonstrating that appendiceal carcinoid can cause localized symptoms in the absence of appendicitis. This series supports the principle that the majority of pediatric carcinoids are small, nonfunctioning, and do not metastasize.i4-i6 The female preponderance in this series has been observed in other studies.r6-I8

Colorectal malignancies in children are rare: 50% of these are mucin-secreting adenocarcinomas.4,19 The prog- nosis for this variant is poor, with survival reported as low as 2.5%.20 None of the colon adenocarcinoma patients in this series had ulcerative colitis or polyposis syndromes. In these not-at-risk patients, early diagnosis is unusual because of a low level of suspicion for this entity. The two cases of gastric sarcomatous tumors in this series demonstrate their low incidence in children. A 1988 meta-analysis of intestinal leiomyosarcomas re-

1008 BETHELETAL

ported only 22 world-wide cases21 Although almost all the pediatric tumors were resected for cure, the 5-year survival rate in adults was only 40% to 63%.22

The alimentary tract remains an unusual location for pediatric malignancy. We have demonstrated that lym- phoma is by far the most common histology observed in both small and large bowel primary malignancies. There have been significant changes in the epidemiology of alimentary lymphoma over the last four decades: an increase in the incidence of nonendemic Burkitt’s, and an increase in the percentage of lymphomas in the small

bowel. The contribution that American Burkitt’s lym- phoma has made toward these changes remains to be defined, as does the role it may play in the overall behavior of pediatric alimentary malignancies in the future.

ACKNOWLEDGMENT

The authors acknowledge the contributions of John Hayes, PhD. of the Wexner Institute for Pediatric Research for invaluable assistance with statistical analysis, and Donna Caniano and Gail Besner for reviewing the manuscript.

REFERENCES

1. Longmo LA, Martin LW: Abdommal mass m the newborn infant. Pediatrics 21:596-604, 1958

2. Pickett LK, Briggs HC: Cancer of the gastrointestinal tract m childhood. Pediatr Clin North Am 14:223-234, 1967

3. Altman AJ, Schwartz AD: Tumors of the gastromtestinal tract, in Altman AJ, Schwartz AD (eds): Malignant Diseases of Infancy. Childhood, and Adolescence, (ed 2). Philadelphia. PA, Saunders, 1983. pp’510-523

4. Goldthorn JF, Canizaro PC: Gastrointestinal malignancies in infancy, childhood and adolescence. Surg Clin North Am 66:845-861, 1986

5. Burkitt D: A sarcoma involving the jaws in Afrtcan children. Br .I Surg 46:218-223, 1958

6. Arseneau JC, Canellos GP, Banks PM, et al: American Burkitt’s lymphoma: A clinicopathologic study of 30 cases. Am J Med 58:314- 321.1975

7. Peller S: Cancer in Childhood and Youth. Bristol. England, John Wright & Sons. Ltd, 1960. p 17

8. Epidemiology of Cancer, from National Vital Statistics Division and Bureau of the Census, United States, 1968

9. Kaufman BH, Burgert EO. Banks PM: Abdominal Burkitt’s lymphoma: Role of early aggressive surgery. J Pediatr Surg 22:671-674. 1987

10. LaQuaglia MP, Stolar CJ, Krailo M, et al: The role of surgery in abdommal non-Hodgkin’s lymphoma: Experience from the Children’s Cancer Study Group. J Pediatr Surg 27:230-235. 1992

11. Shamberger RC, Weinstein HJ: The role of surgery m abdominal Burkitt’s lymphoma. J Pediatr Surg 27:236-240. 1992

12. Gahukamble DB. Khamage AS: Limitations of surgery in mtraabdominal Burkitt’s lymphoma in children. J Pediatr Surg 30:5 19. 522,1995

13. Hutchtson RE, Murphy SB, Fairclough DL, et al. Diffuse small noncleaved cell lymphoma m children, Burkitt’s versus non-Burkitt’s types. Cancer 64:23-28. 1989

14. Field JL, Adamson LF: Review of carcinoids in children, functioning carcmoid in a 15-year-old male. Pediatrics 29:953-960, June. 1962

15. Ryden SE, Drake RM. Franciosi RA: Carcinoid tumors of the appendix in children. Cancer 36:1538-1542. 1975

16. Ayulo JA: Carcinoid tumors of the gastrointestinal tract. Am J Gastroenterol51:138-144, 1969

17. Ponka JL: Carcmoid tumors of the appendtx. Am J Surg 126:77-83. 1973

18. Zeitels J, Naunheim K. Kaplan EL, et al: Carcmoid tumors, a 37.year experience. Arch Surg 117:732-737, 1982

19. Pratt CB, Rivera G, Shanks E, et al: Colorectal carcinoma m adolescents, implications regarding etiology. Cancer 40:2464-2472. 1977

20. Andersson A, Bergdahl L: Carcinoma of the colon in children: A report of six new cases and a revrew of the hterature. J Pediatr Surg 11:967-971. 1976

21. McGrath PC. Neifeld JP, Kay S. et al: Prmciples in the management of pediatric mtestinal leiomyosarcomas J Pediatr Surg 23:939-941, 1988

22. Wurlitzer FP, Mares AJ, Isaacs H: J Pediatr Surg 8:421-427, 1973

Discussion

R.C. Shamberger (Boston, MA): Dr Bethel and his colleagues have identified 45 cases over a 44-year interval demonstrating the relative rarity of these lesions. They demonstrated a fascinating shift in the epidemiol- ogy with an increased frequency of Burkitt’s lymphoma in the latter half of this study accounting for an increase in the small bowel lesions. The prognosis of colon carci- noma in this series was dismal, consistent with the recent report from Sloan-Kettering in which a high incidence. of signet cell lesions was identified.

Because of the extended period of this study, caution must be employed in assessing the survival statistics as chemotherapy has made rapid progress during this four- decade period. Current survival figures for Burkitt’s

lymphoma for stage I and II are 90% and for stage III, 85%, and stage IV, 70%. Lymphoblastic lymphoma and large cell lymphoma have an 85% survival for stages I and II and 70% for stages III and IV, respectively, a marked change from four decades ago. The statistics used for the comparison in this study were x2, and with the low numbers, the Fisher’s Exact test is preferable.

It must be remembered that these diseases are sys- temic. Prior studies have suggested that limited, but not extensive resection should be performed to allow earlier systemic therapy. The CCC experience showed on univari- ate analysis that extent of disease, complete resection, and use of bowel resection were all significant predictors of survival. When multivariate analysis was used, only

ALIMENTARY TRACT MALIGNANCIES

the extent of disease was an independent predictor of outcome. They felt extensive attempts at debulking retroperitoneal and mesenteric lymphoma were contrain- dicated. In our review, we also concluded that explor- atory surgery for the purpose of resection of the primary was only justified in patients who had limited ileocecal and mesenteric involvement.

Was a multivariate analysis was done on to assess the role of surgery? I would also ask about the complications which occurred in this series. It would also be helpful to know the number of days that transpired before the onset of chemotherapy.

C.A.I. Bethel (response): A multi-variate analysis was

performed. In the interest of time I did not present all of that data. As you suggested, stage and size were not significant variables in determining outcome.

The other question referred to the number of days before chemotherapy. The span of time for this study made it extremely difficult to get any conclusive data on these 55 patients with regard to the use of chemotherapy. In fact, as one might expect, in the early portion of this series many of the patients did not receive chemotherapy.

I underscore and agree with your comments on colon cancer and their poor prognoses. The statistical analysis was performed with both x2 and Fisher’s test. The data were the same.