- Email: [email protected]
The role of surgery in abdominal non-Hodgkin's lymphoma: Experience from the childrens cancer study group
The Role of Surgery in Abdominal Non-Hodgkin’s Lymphoma: Experience From the Childrens Cancer Study Group ByMichael
P. LaQuaglia,
Charles
J.H. Stolar,
Mark Krailo,
and Denman
Philip Exelby,
Stuart
Siegel,
Anna
Meadows,
Hammond
Arcadia, California l To determine the appropriate role of surgical intervention in non-Hodgkin’s lymphoma primary to the abdomen, we analyzed the effect of multiple prognostic determinants on event-free survival for patients entered into the CCG-551 study. Eighty-four patients were identified with abdominal lymphoma and of these adequate data for analysis was available on 66 (61%). Variables of interest included: extent of disease at diagnosis, completeness of resection, use of bowel resection, radiation to the primary site, and sex/age/ race. The median age on study was 6 years; 79% of patients were white and 85% were male. Thirty-three patients (49%) presented with localized disease. Laparotomy was performed in 67 children (gg%) with complete gross resection in 28 (42%). of the 10 reported surgical complications, 8 occurred in those with extensive disease who were incompletely resected. Radiation to the primary site was given in 60% of patients with median dose of 2,000 cGy. Analysis was performed both for the overall group and for the subgroup receiving optimal chemotherapy for histopathology. Variables with significant predictive effect on event-free survival in univariate analysis included extent of disease (P I .OOl), complete resection (P < .002), and use of bowel resection (P I .004). However, in multivariate analysis, only extent of disease was an independent predictor of outcome. The data support a role for complete operative excision of localized lymphomas especially when accomplished with bowel resection. Aggressive attempts at debulking extensive retroperitoneal or mesenteric lymphomas are contraindicated. Copyright o 1992 by W.B. Saunders Company INDEX WORDS:
Non-Hodgkin’s
lymphoma,
abdominal
sur-
gery.
T
HE PROGNOSIS for childhood non-Hodgkin’s lymphoma improved significantly with the realization that virtually all cases were disseminated at diagnosis and systemic therapy was necessary for long-term disease-free survival. In 1975, Wollner et al’ established the lo-drug LSA,-L, protocol while Meadows et al successfully used the 4-drug COMP regimen to treat childhood non-Hodgkin’s lymphomas.2*3The CCG-551 study, which compared these two regimens, showed that LSA,-L, was more effec-
From the Lymphoma Strategy Group of the Childrens Cancer Study Group, Arcadia, CA. Presented at the 22nd Annual Meeting of the American Pediatric SurgicalAssociation, Lake Buena Vi&a, Florida, May 15-18, 1991. Address reprint requests to Michael P. LaQuaglia, MD, Department of Surgery (Pediatric Surgery), Memorial Sloan-Kettering Cancer Center, 1275 YorkAve, New York, NYlOO21. Copyright o 1992 by EB. Saunders Company 0022-3468/92/2702-0018$03.00/0
230
tive for disseminated lymphoblastic tumors, whereas COMP had a therapeutic advantage in treatment of undifferentiated lesions.4 Survival increased from 10% to 15% to 75% to 80% during this period. In patients with primary presentation in the abdomen the role of surgery remained controversial. Magrath et al supported aggressive operative debulking, defined as >90% tumor removal, prior to chemotherapy after analysis of a large single institutional series of patients with Burkitt’s lymphoma from Uganda.5,6 Kemeny et al7 also suggested a survival advantage in patients undergoing complete resection but noted that extent of disease was not evaluated. Others cautioned against extensive initial surgery, which might delay institution of systemic therapy.8.9 The question as to whether tumor bulk determined resectability and ultimate prognosis, or surgical resection alone had a primary effect on outcome remained unanswered. In this study we analyzed data on the subgroup of patients with abdominal non-Hodgkin’s lymphoma in the CCG-551 study to ascertain variables, including extent of disease and surgical resectability, which significantly affect disease-free survival. Multivariate analysis was used to determine independent outcome predictors. The results of this analysis were used to develop a rational surgical approach to non-Hodgkin’s lymphomas presenting in the abdomen. MATERIALS Patient
AND
METHODS
Selection
Data on the CCG-551 study were obtained from the CCSG operations office. All untreated patients less than 18 years of age and with biopsy-confirmed non-Hodgkin’s lymphoma of any pathological type but with primary anatomic presentation in the abdomen were eligible to enter the study if they had less than 25% blasts in a bone marrow aspirate and none in the peripheral blood. Five months after initiation (30 patients enrolled) the eligibility criteria were expanded to include all patients with “undifferentiated” lymphomas regardless of extent of bone marrow involvement or blasts in the peripheral blood.
Clinical and Pathological Extent mined by aspiration, white-cell gram, and
Evaluation
of disease, based on protocol specifications, was deterhistory, physical examination, hemogram, bone marrow a spinal fluid cell count and cytocentrifugation for morphology, chest x-ray, bone survey, intravenous uroradionuclide or computed axial tomography scans of the
Journalof Pediatric Surgery, Vol27, No 2 (February), 1992: pp 230-235
SURGERY OF NON-HODGKIN’S
231
LYMPHOMA
liver, spleen, and bone. Routine lymphangiography and staging laparotomy were not performed. Data concerning surgical variables were entered onto a surgical check-list completed by the institutional operating surgeon. Of particular interest was the extent of intraabdominal disease, anatomic location of the primary tumor, resectability (complete gross removal versus incomplete resection), and the incidence and type of surgical complications. Localized disease was defined as isolated mural involvement of the bowel with or without positive mesenteric nodes in adjacent echelons, localized involvement of mesenteric nodes but without fixation to retroperitoneal structures or major mesenteric arteries, or localized solid organ involvement. Tissue suitable for histological examination was obtained from all patients and reviewed by the study pathologist who confirmed the diagnosis of non-Hodgkin’s lymphoma. Histopathologic subtype was determined according to the Rappaport system’” by institutional pathologists and the study pathologist.
Patients with localized abdominal disease underwent exploratory laparotomy and complete gross excision if feasible. Children with widespread abdominal disease also underwent exploratory laparotomy with tissue biopsy. In some patients with extensive intraabdominal disease a “debulking” procedure was also performed.
Statistical Analysis Variables analyzed for their effect on disease-free survival included: age at diagnosis, sex, race, extent of intraabdominal disease, surgical resectability, use of bowel resection, and use of abdominal radiation therapy. Complete resection was defined as complete gross excision. The product-limit method (KaplanMeier) was used to estimate the distribution of event-free survival and the statistical significance of observed differences was assessed using the log-rank test.“.” Covariates with significant predictive effect on outcome (P I .l) were then considered for entry into a Cox proportional hazards model.”
RESULTS
Patient Characteristics In the CCG-5.51 study 84 patients with nonHodgkin’s lymphoma primary in the abdomen were identified. Sufficient information was available on 68 (81%) for univariate and multivariate analysis. All subsequent findings refer to this group. The median age at diagnosis was 8 years (range, 1 month to 16 years). There were 58 males (85%) and 54 children were white (79%). Non-Hodgkin’s lymphoma was preoperatively diagnosed in 10 (15%). Localized (intraabdominal) disease was encountered in 33 children (49%) and extensive in 35 (51%). Surgery Laparotomy was performed in 67 patients (99%) and one child had an endoscopic transrectal biopsy of a large pelvic tumor. Operation was elective in 31 patients (46%) and urgent in 37 (54%). Preoperative diagnoses are listed in Table 1 and broken down as to whether exploration was elective or urgent. The most
Table 1. Preoperative Diagnoses Preoperative
Urgent
Elective
Diagnosis
29
3
lntussusception
0
17
Appendicitis
0
8
Perforation
0
4
Intestinal obstruction
0
3
Abdominal pain
1
1
Not stated
1
1
31
37
Abdominal mass
Tota I
frequent preoperative diagnosis was an abdominal mass (31), which was associated with tenderness or other symptoms in 5 cases. This was followed by intussusception (17) and then appendicitis (8). Anatomically the main tumor mass involved the bowel in 58 patients (85%). The small bowel was involved in 28 children, the large bowel in 14, and both the small and large bowel in 17. In addition, the liver was involved in 8 patients with simultaneous bowel lymphoma in 5, whereas the kidney was also a primary site in 8 and 7 of these also had tumor infiltrating the bowel. Six patients had ovarian tumors with 5 of these also suffering bowel involvement. All of the 5 patients with splenic lymphoma also had bowel infiltration. The relationship among tumor burden, urgency of surgery, and resectability is listed in Table 2. It can be noted that complete gross resection of abdominal lymphoma was never accomplished when disease was extensive despite 35 operative procedures. In contradistinction, 28 of 33 patients (85%) with localized disease underwent complete resection and the majority (24/28) of these were urgently explored. Of particular note is that 91% of patients undergoing complete resection with localized disease had a bowel resection. There were 10 reported surgical complications in the entire series (15%). Eight of these occurred in incompletely excised patients who also had extensive disease and the remaining 2 developed in children with localized, completely resected disease. Complications are listed in Table 3 and related to disease
Table 2. Extent of Disease Correlated With Resectability, Urgency of Operation, and Bowel Resection Complete
Incomplete
Complete
Incomplete
Resection
Resection
Resection
Resection
4
3
0
24
Emergency
24
2
0
11
Total
28
5
0
35
26 (91%)
3 (60%)
0
8 (23%)
Elective
No. with rasection involving bowel (%)
232
LAQUAGLIA ET AL
Table 3. Complications of Operation Extent of Complication
Disease
urgency
Resection
Infected ascites
Localized
Urgent
Complete
No
Wound infection
Localized
Urgent
Complete
Yes
Persistent ileus
Extensive
Urgent
Incomplete
Yes
Wound infection
Extensive
Elective
Incomplete
Yes
Extensive
Urgent
Incomplete
Yes
Reexploration for bleeding (omenturn)
Extensive
Elective
Incomplete
Yes
Pleural effusion, chest tube needed
Extensive
Elective
Incomplete
Yes
Renal failure
Extensive
Elective
Incomplete
No
Subhepatic abscess (operative drainage re-
Extensive
Urgent
Incomplete
No
Extensive
Urgent
incomplete
No
Blood loss
(> 250
ml)
Alive
quired) Wound dehiscence
extent, resectability, urgency of operation, and survival. Finally, there were 4 reported second-look laparotomies (6%) all done in patients with extensive, incompletely excised disease. Residual tumor that remained unresectable was found in 3 and the status of the fourth could not be determined. Chemotherapy and Radiation
Of the 68 patients in this study 38 (56%) received the LSA,-L, regimen and 30 (44%) received COMP. Five of the patients receiving LSA,-L, had the lymphoblastic subtype and, therefore, received histopathologically appropriate chemotherapy. All 30 patients who got COMP had undifferentiated histology and received appropriate chemotherapy; thus, a total of 35 patients in this study received histologically appropriate chemotherapy. External beam, supervoltage abdominal radiation therapy was given to 41 patients (60%) and median dose was 2,000 cGy (range, 200 to 3,038 cGy). Outcome and Univariate Analysis
Outcome data are summarized in Table 4, which shows that the 46 survivors had a higher rate of complete resection, urgent exploration, and localized disease. Also, the complication rate was lower in this group. The Kaplan-Meier curves for overall and event-free survival (Fig 1) show that actuarial survival in the overall group was approximately 65%. There was no difference between overall and event-free
survival, implying that deaths were related to uncontrolled disease and salvage after relapse was minimal. In univariate analysis age at diagnosis, sex, race, and radiation therapy to the abdomen were not significant predictors of outcome while extent of disease at diagnosis, complete resection of the primary, and excision achieved with bowel resection were significant. Data are summarized in Table 5 and show the results of log-rank analysis performed on the entire group in the study (n = 68) and a parallel analysis on the subgroup who received histologically appropriate chemotherapy (n = 35). Results are basically the same except for the loss of statistical power occasioned by smaller numbers in the subgroup. The survival in patients with extensive versus localized disease is compared in Fig 2. Figure 3 shows the effect of complete gross resection on outcome, whereas Fig 4 demonstrates the impact of bowel resection. Reference to Table 2 again underscores the fact that complete removal usually involved bowel resection. Multivariate Analysis
Variables with significant effect on outcome in univariate analysis were compared and the only independent predictor of event-free survival was extent of abdominal disease at diagnosis (P 2 .OOOOl). A parallel analysis that looked at the subgroup receiving optimal chemotherapy based on histopathologic subtype gave similar results.
Table 4. Patient Outcome Patient Group Complete
Urgently
vbnge)
NO.
Resection
Explored
Localized
Survivors 2,650 days (766-3.168)
46
25 (54%)
27 (59%)
29 (63%)
Deaths 174 days (30-491)
22
3 (14%)
10 (45%)
4 (18%)
4 (18%)
Total 2,360 days (30-3,168)
68
28 (41%)
37 (54%)
33 (49%)
10 (15%)
Median Follow-up
Complications
6 (13%)
233
SURGERY OF NON-HODGKIN’S LYMPHOMA
I?
Event-tree survival 0 Overall survival n
.2
(68 Rs., 43 Censored) (68 Pk. 46 Censored)
.2 -
I
I
I
I
528
1056
1584
2112
J
I
2640
3168
I 528
0
I 1056
Fig 1. These Kaplan-Meier curves show the overall survival and disease-free survival for the 66 patients included in this study.
DISCUSSION
The results of this study support several conclusions useful in determining the appropriate role of surgical intervention in abdominal non-Hodgkin’s lymphoma in childhood. The strongest predictor of event-free survival is tumor burden at diagnosis, as measured, in this analysis, by extent of abdominal disease. This is consistent with observations clearly showing that outcome is most dependent on the number of malignant cells present at initiation of therapy and can be predicted by measurement of serum lactic dehydrogenase, interleukin II receptor, and P,-microglobulin levels.‘4-‘” The data from the present analysis indicate that the amount of bulk disease in the abdomen is also a reliable predictor. To determine whether surgical removal of the abdominal tumor alters the clinical course or, conversely, whether the presenting mass primarily determines the feasibility of surgical resection, we examined the effect of complete surgical excision on event-free survival. In univariate analysis complete excision does favorably affect outcome but none of the patients with extensive abdominal disease underwent complete excision. Indeed, most complete resections were performed during emergency laparotomies for symptomatic, localized bowel involvement in pa-
Fig 2.
disease
I 2640
I 3168
All Patients (n = 68)
Optimal Therapy Patients (n = 35)
Negative
P 5 ,001
P 2 ,069
Complete resection
Positive
P 5 ,002
P 2 ,071
Use of bowel resection
Positive
P 5 ,004
P 5 ,070
Survival in the extensive versus localized disease groups.
1
.a
.6
Complete resection
(28 Rs.. 24 Censored)
0 Incomplete resection
(40 Rs., 19 Censored)
n
.2
Predictive Effect on Event-Free Survival
Extensive
I 2112
tients explored for intussusception or to rule out appendicitis (Tables 1 and 2). The fact that bowel resection was associated with better prognosis suggests that intestinal involvement is associated with an increased frequency of abdominal symptoms resulting in earlier laparotomies. Alternatively, a mesenteric or retroperitoneal mass that does not involve bowel wall remains clinically silent until a relatively larger tumor burden is reached as is the case with neuroblastoma or Wilms’ tumor. Complete resection is most readily accomplished in patients who have localized bowel involvement and early development of abdominal symptoms. When tumor burden was compared with resectability in multivariate analysis only the extent of disease remained an independent predictor of event-free survival. In essence, extent of disease is the ultimate
Table 5. Log-Rank Comparisons of Variables With Significant
Effect on Outcome
I
i 584 Time (days)
Time (days)
Variable
(35 Pts., 15 Censored) (33 Pts., 26 Censored)
P _<0.001
P _<0.001 0'
n Extensive disease 0 Localizeddisease
0
528
i 584
1056
2112
2640
3168
Time (days) Fig 3. Survival in completely sected patients.
resected
versus incompletely
re-
LAQUAGLIA ET AL
n Bowel resection 0 No bowel resection
(37 Pk., 29 Censored) (29 Pk., 13 Censored)
P _<0.004 0
I 528
I 1056
I 1584 Time
I 2112
1 2640
I 3168
(days)
Fig 4. Patients in this study undergoing better outcome.
bowel resection had a
determinant of resectability rather than surgical resection affecting outcome independently. Finally, from Tables 3 and 4 it can be concluded that exploration of children with extensive abdominal disease is associated with a higher complication rate. Non-Hodgkin’s lymphomas in general and abdominal undifferentiated tumors in particular have rapid doubling times which can be as short as 12 hours.” Because tumor burden steadily increases with time it is imperative that chemotherapy be instituted promptly. Delay, because patients are recovering from surgical complications, will have significant negative effects on outcome. Reference to Table 3 shows that the 4 patients with grave postoperative complications died from disease. In summary, children with abdominal non-Hodg-
kin’s lymphoma can be divided into two groups: those with localized disease often involving bowel wall, and extensive disease patients with bulky mesenteric and retroperitoneal tumor. In the first group of patients operation is often necessitated by abdominal symptoms and complete removal relatively easily performed with resection and anastomosis of a bowel segment. Complete removal under these circumstances should be associated with minimal complications as was noted in this study and has the added advantage of avoiding bowel perforation, gastrointestinal hemorrhage, or the tumor lysis syndrome after the initiation of chemotherapy.h.‘0.2’ In the second group, complete removal will be extremely difficult because of tumor bulk and difficult anatomic location. These patients should have a minimal operative procedure obtaining enough fresh tissue for diagnosis and specialized studies such as immunophenotyping and cytogenetics.“.*” In neither situation should the surgical procedure delay prompt institution of chemotherapy. We conclude that the most important predictor of event-free survival in abdominal non-Hodgkin’s lymphoma is extent of disease at diagnosis. Surgical resection is associated with improved outlook but the ability to accomplish this will be biologically predetermined. In particular, lymphomas confined to the bowel wall and presenting because of abdominal symptoms may have a high rate of complete resectability and subsequent cure. Alternatively, attempts at surgical debulking of extensive abdominal tumors is associated with a greater complication rate and possible delay in initiation of essential systemic chemotherapy. In the situation of extensive tumors, a minimal surgical procedure that obtains adequate diagnostic tissue is more appropriate.
REFERENCES 1. Wollner N, Lieberman P, Exelby P, et al: Non-Hodgkin’s lymphoma in children: Results of treatment with LSA?-L, protocol. Br J Cancer 31:337-343,1975 (suppl) 2. Meadows AT, Jenkin RDT, Anderson J, et al: A new therapy schedule for pediatric non-Hodgkin’s lymphoma toxicity and preliminary results. Med Pediatr Oncol8:15-24.1980 3. Ziegler JL: Treatment results of 54 American patients with Burkitt’s lymphoma are similar to the African experience. N Engl J Med 297:75-86,1977 4. Anderson JR, Wilson JF, Jenkin DT, et al: Childhood non-Hodgkin’s lymphoma: The results of a randomized therapeutic trial comparing a 4-drug regimen (COMP) with a lo-drug regimen (LSA&). N Engl J Med 308:559-565,1983 5. Magrath IT, Lwanga S, Carswell W, et al: Surgical reduction of tumour bulk in management of abdominal Burkitt’s lymphoma. Br Med J 2:308-312,1974 6. Janus C, Edwards
BK, Sariban
E, et al: Surgical
resection
and
limited chemotherapy for abdominal Cancer Treat Rep 68:599-605,1984
undifferentiated
lymphomas.
7. Kemeny MM, Magrath IT, Brennan MF: The role of surgery in the management of American Burkitt’s lymphoma and its treatment. Ann Surg 196:82-86,1982 8. Zea JM, Exelby PR, Wollner N: Abdominal non-Hodgkin’s lymphoma in childhood. J Pediatr Surg 11:363-369,1976 9. Kaufman BH, Burgert EO Jr, Banks PM: Abdominal kitt’s lymphoma: Role of early aggressive surgery. J Pediatr 22~671-674.1987
BurSurg
10. Nathwani BN, Kim H, Rappaport H, et al: Non-Hodgkin’s: A clinicopathologic study comparing two classifications. Cancer 41:303-325, 1978 11. Kaplan EL, Meier R: Nonparametric estimation plete observations. J Am Stat Assoc 53:457-581,1958 12. Peto R, Peto J: Asymptotically efficient procedures. J R Stat Sot 135:185-206,1977
from incom-
rank invariant
test
SURGERY OF NON-HODGKIN’S
235
LYMPHOMA
13. Cox DR: Regression models and life tables. J R Stat Sot 34: 187-220. 1972 14. Magrath IT, Lee YJ, Anderson T, et al: Prognostic factors in Burkitt’s lymphoma: Importance of total tumor burden. Cancer 45:1.507-1515.1980 15. Magrath IT, Janus C, Edwards B, et al: An effective therapy for both undifferentiated (including Burkitt’s) lymphomas and lymphoblastic lymphomas in children and young adults. Blood 63:1102-1111. 1984 16. Csako G, Magrath IT, Elin R: Serum total and isoenzyme lactic dehydrogenase activity in American Burkitt’s lymphoma. Am J Clin Pathol78:712-717, 1982 17. Hagberg H, Killander A, Simonnsson B: Serum beta-2 microglobulin in malignant lymphoma. Cancer 51:2220-2225, 1983 18. Wagner DK, Kiwanuka J, Edwards BL. et al: Soluble interleukin II receptor levels in patients with undifferentiated and lymphoblastic lymphomas. J Clin Oncol5:1262-1274,1987
19. Iverson U, Iverson African cases of Burkitt’s Cancer 8:305-310,1972
OH, Ziegler JL, et al: Cell kinetics of lymphoma. A preliminary report. Em J
20. Cohen LF, Balow JE, Magrath IT: Acute tumor lysis syndrome: A review of 37 patients with Burkitt’s lymphoma. Am J Med h&486-491,1980 21. Meyers PA, Potter VP, Wollner during initial treatment for childhood Cancer 56:259-261, 1985
N, et al: Bowel perforation non-Hodgkin’s lymphoma.
22. Smith SD, Morgan R, Link MP, et al: Cytogenetic and immunophenotypic analysis of cell lines established from patients with T cell leukemiailymphoma. Blood 67:650-656, 1986 23. Le Beau MM, Rowley JD: Chromosomal leukemia and lymphoma: Clinical and biological Hum Genet 15:1-54, 1986
abnormalities in significance. Adv
Discussion D.M. Hays (Los Angeles, CA): This report has been almost a decade in development and almost “died” on several occasions. I particularly want to thank Drs LaQuaglia and Stolar for picking up this project in midstudy and carrying it to a successful conclusion. Otherwise these data would have been lost. I think Dr LaQuaglia has very correctly pointed out that patients with non-Hodgkin’s lymphoma with actual or simulated acute abdominal conditions are the ones that receive early therapy and have the best survival. Second, when one considers localization of the tumor, tumor resection, and improved survival, localization is the only independent predictor of survival. First question, Dr LaQuaglia, the P values that you show for the aggregate group, that is all therapies, in comparison with the group receiving specific therapies were different and the P values were more significant in the aggregate group than in the specific group, which I find unexplainable. It may be that the numbers were larger in that group. Third, I would like you to visualize yourself in the operating room at 3 AM with an intussusception that can’t be reduced by hydrostatic pressure and the question comes up as to what it is. Is it an irreducible intussusception, is it really a tumor, how do you make this distinction? Do you do biopsies? Can you tell right away? Do you do a “blind” resection? How do you handle it? M.P. LaQuaglia (response): The reason for doing a parallel analysis on a subgroup of the patients in the study was as you remember the CCG-551 study was
used to compare the LSA2-L2 regimen. The surprising finding from that was that disseminated lymphoma was better for the undifferentiated, so it was Dr Krailo, the statistician, who felt that we should do a parallel analysis of the smaller group of patients who had the more specific therapy. Now it’s controversial that there is a difference between the regimens. The National Cancer Institute studies and studies from the pediatric oncology group don’t necessarily support that distinction but we did the subgroup analysis, which reduced the numbers. The statistical difference therefore, the power, is reduced. However, both our institutional statistician, Mark Heller, and Mark Krailo at CCG agree that the general trend is the same and if we had 100 patients with specific therapy for histopathology, the same conclusions would accrue. And indeed in multivariate analysis the same conclusions would accrue. We put those in for completeness. The second question was what do you do in the operating room and I can’t be definitive about I think that many of these patients are going to present urgently and you just have to rely on surgical judgment. It may be that you will have to do a sleeve resection. The only caveat, you can’t do cytogenetics and immunophenotyping on formalin tissue so that some awareness of the possibility that an appendicitis, intussusception, or bowel perforation may be lymphoma may be helpful in terms of getting some of the tissue stored in the refrigerator so that the more esoteric studies can be done.
P. LaQuaglia,
Charles
J.H. Stolar,
Mark Krailo,
and Denman
Philip Exelby,
Stuart
Siegel,
Anna
Meadows,
Hammond
Arcadia, California l To determine the appropriate role of surgical intervention in non-Hodgkin’s lymphoma primary to the abdomen, we analyzed the effect of multiple prognostic determinants on event-free survival for patients entered into the CCG-551 study. Eighty-four patients were identified with abdominal lymphoma and of these adequate data for analysis was available on 66 (61%). Variables of interest included: extent of disease at diagnosis, completeness of resection, use of bowel resection, radiation to the primary site, and sex/age/ race. The median age on study was 6 years; 79% of patients were white and 85% were male. Thirty-three patients (49%) presented with localized disease. Laparotomy was performed in 67 children (gg%) with complete gross resection in 28 (42%). of the 10 reported surgical complications, 8 occurred in those with extensive disease who were incompletely resected. Radiation to the primary site was given in 60% of patients with median dose of 2,000 cGy. Analysis was performed both for the overall group and for the subgroup receiving optimal chemotherapy for histopathology. Variables with significant predictive effect on event-free survival in univariate analysis included extent of disease (P I .OOl), complete resection (P < .002), and use of bowel resection (P I .004). However, in multivariate analysis, only extent of disease was an independent predictor of outcome. The data support a role for complete operative excision of localized lymphomas especially when accomplished with bowel resection. Aggressive attempts at debulking extensive retroperitoneal or mesenteric lymphomas are contraindicated. Copyright o 1992 by W.B. Saunders Company INDEX WORDS:
Non-Hodgkin’s
lymphoma,
abdominal
sur-
gery.
T
HE PROGNOSIS for childhood non-Hodgkin’s lymphoma improved significantly with the realization that virtually all cases were disseminated at diagnosis and systemic therapy was necessary for long-term disease-free survival. In 1975, Wollner et al’ established the lo-drug LSA,-L, protocol while Meadows et al successfully used the 4-drug COMP regimen to treat childhood non-Hodgkin’s lymphomas.2*3The CCG-551 study, which compared these two regimens, showed that LSA,-L, was more effec-
From the Lymphoma Strategy Group of the Childrens Cancer Study Group, Arcadia, CA. Presented at the 22nd Annual Meeting of the American Pediatric SurgicalAssociation, Lake Buena Vi&a, Florida, May 15-18, 1991. Address reprint requests to Michael P. LaQuaglia, MD, Department of Surgery (Pediatric Surgery), Memorial Sloan-Kettering Cancer Center, 1275 YorkAve, New York, NYlOO21. Copyright o 1992 by EB. Saunders Company 0022-3468/92/2702-0018$03.00/0
230
tive for disseminated lymphoblastic tumors, whereas COMP had a therapeutic advantage in treatment of undifferentiated lesions.4 Survival increased from 10% to 15% to 75% to 80% during this period. In patients with primary presentation in the abdomen the role of surgery remained controversial. Magrath et al supported aggressive operative debulking, defined as >90% tumor removal, prior to chemotherapy after analysis of a large single institutional series of patients with Burkitt’s lymphoma from Uganda.5,6 Kemeny et al7 also suggested a survival advantage in patients undergoing complete resection but noted that extent of disease was not evaluated. Others cautioned against extensive initial surgery, which might delay institution of systemic therapy.8.9 The question as to whether tumor bulk determined resectability and ultimate prognosis, or surgical resection alone had a primary effect on outcome remained unanswered. In this study we analyzed data on the subgroup of patients with abdominal non-Hodgkin’s lymphoma in the CCG-551 study to ascertain variables, including extent of disease and surgical resectability, which significantly affect disease-free survival. Multivariate analysis was used to determine independent outcome predictors. The results of this analysis were used to develop a rational surgical approach to non-Hodgkin’s lymphomas presenting in the abdomen. MATERIALS Patient
AND
METHODS
Selection
Data on the CCG-551 study were obtained from the CCSG operations office. All untreated patients less than 18 years of age and with biopsy-confirmed non-Hodgkin’s lymphoma of any pathological type but with primary anatomic presentation in the abdomen were eligible to enter the study if they had less than 25% blasts in a bone marrow aspirate and none in the peripheral blood. Five months after initiation (30 patients enrolled) the eligibility criteria were expanded to include all patients with “undifferentiated” lymphomas regardless of extent of bone marrow involvement or blasts in the peripheral blood.
Clinical and Pathological Extent mined by aspiration, white-cell gram, and
Evaluation
of disease, based on protocol specifications, was deterhistory, physical examination, hemogram, bone marrow a spinal fluid cell count and cytocentrifugation for morphology, chest x-ray, bone survey, intravenous uroradionuclide or computed axial tomography scans of the
Journalof Pediatric Surgery, Vol27, No 2 (February), 1992: pp 230-235
SURGERY OF NON-HODGKIN’S
231
LYMPHOMA
liver, spleen, and bone. Routine lymphangiography and staging laparotomy were not performed. Data concerning surgical variables were entered onto a surgical check-list completed by the institutional operating surgeon. Of particular interest was the extent of intraabdominal disease, anatomic location of the primary tumor, resectability (complete gross removal versus incomplete resection), and the incidence and type of surgical complications. Localized disease was defined as isolated mural involvement of the bowel with or without positive mesenteric nodes in adjacent echelons, localized involvement of mesenteric nodes but without fixation to retroperitoneal structures or major mesenteric arteries, or localized solid organ involvement. Tissue suitable for histological examination was obtained from all patients and reviewed by the study pathologist who confirmed the diagnosis of non-Hodgkin’s lymphoma. Histopathologic subtype was determined according to the Rappaport system’” by institutional pathologists and the study pathologist.
Patients with localized abdominal disease underwent exploratory laparotomy and complete gross excision if feasible. Children with widespread abdominal disease also underwent exploratory laparotomy with tissue biopsy. In some patients with extensive intraabdominal disease a “debulking” procedure was also performed.
Statistical Analysis Variables analyzed for their effect on disease-free survival included: age at diagnosis, sex, race, extent of intraabdominal disease, surgical resectability, use of bowel resection, and use of abdominal radiation therapy. Complete resection was defined as complete gross excision. The product-limit method (KaplanMeier) was used to estimate the distribution of event-free survival and the statistical significance of observed differences was assessed using the log-rank test.“.” Covariates with significant predictive effect on outcome (P I .l) were then considered for entry into a Cox proportional hazards model.”
RESULTS
Patient Characteristics In the CCG-5.51 study 84 patients with nonHodgkin’s lymphoma primary in the abdomen were identified. Sufficient information was available on 68 (81%) for univariate and multivariate analysis. All subsequent findings refer to this group. The median age at diagnosis was 8 years (range, 1 month to 16 years). There were 58 males (85%) and 54 children were white (79%). Non-Hodgkin’s lymphoma was preoperatively diagnosed in 10 (15%). Localized (intraabdominal) disease was encountered in 33 children (49%) and extensive in 35 (51%). Surgery Laparotomy was performed in 67 patients (99%) and one child had an endoscopic transrectal biopsy of a large pelvic tumor. Operation was elective in 31 patients (46%) and urgent in 37 (54%). Preoperative diagnoses are listed in Table 1 and broken down as to whether exploration was elective or urgent. The most
Table 1. Preoperative Diagnoses Preoperative
Urgent
Elective
Diagnosis
29
3
lntussusception
0
17
Appendicitis
0
8
Perforation
0
4
Intestinal obstruction
0
3
Abdominal pain
1
1
Not stated
1
1
31
37
Abdominal mass
Tota I
frequent preoperative diagnosis was an abdominal mass (31), which was associated with tenderness or other symptoms in 5 cases. This was followed by intussusception (17) and then appendicitis (8). Anatomically the main tumor mass involved the bowel in 58 patients (85%). The small bowel was involved in 28 children, the large bowel in 14, and both the small and large bowel in 17. In addition, the liver was involved in 8 patients with simultaneous bowel lymphoma in 5, whereas the kidney was also a primary site in 8 and 7 of these also had tumor infiltrating the bowel. Six patients had ovarian tumors with 5 of these also suffering bowel involvement. All of the 5 patients with splenic lymphoma also had bowel infiltration. The relationship among tumor burden, urgency of surgery, and resectability is listed in Table 2. It can be noted that complete gross resection of abdominal lymphoma was never accomplished when disease was extensive despite 35 operative procedures. In contradistinction, 28 of 33 patients (85%) with localized disease underwent complete resection and the majority (24/28) of these were urgently explored. Of particular note is that 91% of patients undergoing complete resection with localized disease had a bowel resection. There were 10 reported surgical complications in the entire series (15%). Eight of these occurred in incompletely excised patients who also had extensive disease and the remaining 2 developed in children with localized, completely resected disease. Complications are listed in Table 3 and related to disease
Table 2. Extent of Disease Correlated With Resectability, Urgency of Operation, and Bowel Resection Complete
Incomplete
Complete
Incomplete
Resection
Resection
Resection
Resection
4
3
0
24
Emergency
24
2
0
11
Total
28
5
0
35
26 (91%)
3 (60%)
0
8 (23%)
Elective
No. with rasection involving bowel (%)
232
LAQUAGLIA ET AL
Table 3. Complications of Operation Extent of Complication
Disease
urgency
Resection
Infected ascites
Localized
Urgent
Complete
No
Wound infection
Localized
Urgent
Complete
Yes
Persistent ileus
Extensive
Urgent
Incomplete
Yes
Wound infection
Extensive
Elective
Incomplete
Yes
Extensive
Urgent
Incomplete
Yes
Reexploration for bleeding (omenturn)
Extensive
Elective
Incomplete
Yes
Pleural effusion, chest tube needed
Extensive
Elective
Incomplete
Yes
Renal failure
Extensive
Elective
Incomplete
No
Subhepatic abscess (operative drainage re-
Extensive
Urgent
Incomplete
No
Extensive
Urgent
incomplete
No
Blood loss
(> 250
ml)
Alive
quired) Wound dehiscence
extent, resectability, urgency of operation, and survival. Finally, there were 4 reported second-look laparotomies (6%) all done in patients with extensive, incompletely excised disease. Residual tumor that remained unresectable was found in 3 and the status of the fourth could not be determined. Chemotherapy and Radiation
Of the 68 patients in this study 38 (56%) received the LSA,-L, regimen and 30 (44%) received COMP. Five of the patients receiving LSA,-L, had the lymphoblastic subtype and, therefore, received histopathologically appropriate chemotherapy. All 30 patients who got COMP had undifferentiated histology and received appropriate chemotherapy; thus, a total of 35 patients in this study received histologically appropriate chemotherapy. External beam, supervoltage abdominal radiation therapy was given to 41 patients (60%) and median dose was 2,000 cGy (range, 200 to 3,038 cGy). Outcome and Univariate Analysis
Outcome data are summarized in Table 4, which shows that the 46 survivors had a higher rate of complete resection, urgent exploration, and localized disease. Also, the complication rate was lower in this group. The Kaplan-Meier curves for overall and event-free survival (Fig 1) show that actuarial survival in the overall group was approximately 65%. There was no difference between overall and event-free
survival, implying that deaths were related to uncontrolled disease and salvage after relapse was minimal. In univariate analysis age at diagnosis, sex, race, and radiation therapy to the abdomen were not significant predictors of outcome while extent of disease at diagnosis, complete resection of the primary, and excision achieved with bowel resection were significant. Data are summarized in Table 5 and show the results of log-rank analysis performed on the entire group in the study (n = 68) and a parallel analysis on the subgroup who received histologically appropriate chemotherapy (n = 35). Results are basically the same except for the loss of statistical power occasioned by smaller numbers in the subgroup. The survival in patients with extensive versus localized disease is compared in Fig 2. Figure 3 shows the effect of complete gross resection on outcome, whereas Fig 4 demonstrates the impact of bowel resection. Reference to Table 2 again underscores the fact that complete removal usually involved bowel resection. Multivariate Analysis
Variables with significant effect on outcome in univariate analysis were compared and the only independent predictor of event-free survival was extent of abdominal disease at diagnosis (P 2 .OOOOl). A parallel analysis that looked at the subgroup receiving optimal chemotherapy based on histopathologic subtype gave similar results.
Table 4. Patient Outcome Patient Group Complete
Urgently
vbnge)
NO.
Resection
Explored
Localized
Survivors 2,650 days (766-3.168)
46
25 (54%)
27 (59%)
29 (63%)
Deaths 174 days (30-491)
22
3 (14%)
10 (45%)
4 (18%)
4 (18%)
Total 2,360 days (30-3,168)
68
28 (41%)
37 (54%)
33 (49%)
10 (15%)
Median Follow-up
Complications
6 (13%)
233
SURGERY OF NON-HODGKIN’S LYMPHOMA
I?
Event-tree survival 0 Overall survival n
.2
(68 Rs., 43 Censored) (68 Pk. 46 Censored)
.2 -
I
I
I
I
528
1056
1584
2112
J
I
2640
3168
I 528
0
I 1056
Fig 1. These Kaplan-Meier curves show the overall survival and disease-free survival for the 66 patients included in this study.
DISCUSSION
The results of this study support several conclusions useful in determining the appropriate role of surgical intervention in abdominal non-Hodgkin’s lymphoma in childhood. The strongest predictor of event-free survival is tumor burden at diagnosis, as measured, in this analysis, by extent of abdominal disease. This is consistent with observations clearly showing that outcome is most dependent on the number of malignant cells present at initiation of therapy and can be predicted by measurement of serum lactic dehydrogenase, interleukin II receptor, and P,-microglobulin levels.‘4-‘” The data from the present analysis indicate that the amount of bulk disease in the abdomen is also a reliable predictor. To determine whether surgical removal of the abdominal tumor alters the clinical course or, conversely, whether the presenting mass primarily determines the feasibility of surgical resection, we examined the effect of complete surgical excision on event-free survival. In univariate analysis complete excision does favorably affect outcome but none of the patients with extensive abdominal disease underwent complete excision. Indeed, most complete resections were performed during emergency laparotomies for symptomatic, localized bowel involvement in pa-
Fig 2.
disease
I 2640
I 3168
All Patients (n = 68)
Optimal Therapy Patients (n = 35)
Negative
P 5 ,001
P 2 ,069
Complete resection
Positive
P 5 ,002
P 2 ,071
Use of bowel resection
Positive
P 5 ,004
P 5 ,070
Survival in the extensive versus localized disease groups.
1
.a
.6
Complete resection
(28 Rs.. 24 Censored)
0 Incomplete resection
(40 Rs., 19 Censored)
n
.2
Predictive Effect on Event-Free Survival
Extensive
I 2112
tients explored for intussusception or to rule out appendicitis (Tables 1 and 2). The fact that bowel resection was associated with better prognosis suggests that intestinal involvement is associated with an increased frequency of abdominal symptoms resulting in earlier laparotomies. Alternatively, a mesenteric or retroperitoneal mass that does not involve bowel wall remains clinically silent until a relatively larger tumor burden is reached as is the case with neuroblastoma or Wilms’ tumor. Complete resection is most readily accomplished in patients who have localized bowel involvement and early development of abdominal symptoms. When tumor burden was compared with resectability in multivariate analysis only the extent of disease remained an independent predictor of event-free survival. In essence, extent of disease is the ultimate
Table 5. Log-Rank Comparisons of Variables With Significant
Effect on Outcome
I
i 584 Time (days)
Time (days)
Variable
(35 Pts., 15 Censored) (33 Pts., 26 Censored)
P _<0.001
P _<0.001 0'
n Extensive disease 0 Localizeddisease
0
528
i 584
1056
2112
2640
3168
Time (days) Fig 3. Survival in completely sected patients.
resected
versus incompletely
re-
LAQUAGLIA ET AL
n Bowel resection 0 No bowel resection
(37 Pk., 29 Censored) (29 Pk., 13 Censored)
P _<0.004 0
I 528
I 1056
I 1584 Time
I 2112
1 2640
I 3168
(days)
Fig 4. Patients in this study undergoing better outcome.
bowel resection had a
determinant of resectability rather than surgical resection affecting outcome independently. Finally, from Tables 3 and 4 it can be concluded that exploration of children with extensive abdominal disease is associated with a higher complication rate. Non-Hodgkin’s lymphomas in general and abdominal undifferentiated tumors in particular have rapid doubling times which can be as short as 12 hours.” Because tumor burden steadily increases with time it is imperative that chemotherapy be instituted promptly. Delay, because patients are recovering from surgical complications, will have significant negative effects on outcome. Reference to Table 3 shows that the 4 patients with grave postoperative complications died from disease. In summary, children with abdominal non-Hodg-
kin’s lymphoma can be divided into two groups: those with localized disease often involving bowel wall, and extensive disease patients with bulky mesenteric and retroperitoneal tumor. In the first group of patients operation is often necessitated by abdominal symptoms and complete removal relatively easily performed with resection and anastomosis of a bowel segment. Complete removal under these circumstances should be associated with minimal complications as was noted in this study and has the added advantage of avoiding bowel perforation, gastrointestinal hemorrhage, or the tumor lysis syndrome after the initiation of chemotherapy.h.‘0.2’ In the second group, complete removal will be extremely difficult because of tumor bulk and difficult anatomic location. These patients should have a minimal operative procedure obtaining enough fresh tissue for diagnosis and specialized studies such as immunophenotyping and cytogenetics.“.*” In neither situation should the surgical procedure delay prompt institution of chemotherapy. We conclude that the most important predictor of event-free survival in abdominal non-Hodgkin’s lymphoma is extent of disease at diagnosis. Surgical resection is associated with improved outlook but the ability to accomplish this will be biologically predetermined. In particular, lymphomas confined to the bowel wall and presenting because of abdominal symptoms may have a high rate of complete resectability and subsequent cure. Alternatively, attempts at surgical debulking of extensive abdominal tumors is associated with a greater complication rate and possible delay in initiation of essential systemic chemotherapy. In the situation of extensive tumors, a minimal surgical procedure that obtains adequate diagnostic tissue is more appropriate.
REFERENCES 1. Wollner N, Lieberman P, Exelby P, et al: Non-Hodgkin’s lymphoma in children: Results of treatment with LSA?-L, protocol. Br J Cancer 31:337-343,1975 (suppl) 2. Meadows AT, Jenkin RDT, Anderson J, et al: A new therapy schedule for pediatric non-Hodgkin’s lymphoma toxicity and preliminary results. Med Pediatr Oncol8:15-24.1980 3. Ziegler JL: Treatment results of 54 American patients with Burkitt’s lymphoma are similar to the African experience. N Engl J Med 297:75-86,1977 4. Anderson JR, Wilson JF, Jenkin DT, et al: Childhood non-Hodgkin’s lymphoma: The results of a randomized therapeutic trial comparing a 4-drug regimen (COMP) with a lo-drug regimen (LSA&). N Engl J Med 308:559-565,1983 5. Magrath IT, Lwanga S, Carswell W, et al: Surgical reduction of tumour bulk in management of abdominal Burkitt’s lymphoma. Br Med J 2:308-312,1974 6. Janus C, Edwards
BK, Sariban
E, et al: Surgical
resection
and
limited chemotherapy for abdominal Cancer Treat Rep 68:599-605,1984
undifferentiated
lymphomas.
7. Kemeny MM, Magrath IT, Brennan MF: The role of surgery in the management of American Burkitt’s lymphoma and its treatment. Ann Surg 196:82-86,1982 8. Zea JM, Exelby PR, Wollner N: Abdominal non-Hodgkin’s lymphoma in childhood. J Pediatr Surg 11:363-369,1976 9. Kaufman BH, Burgert EO Jr, Banks PM: Abdominal kitt’s lymphoma: Role of early aggressive surgery. J Pediatr 22~671-674.1987
BurSurg
10. Nathwani BN, Kim H, Rappaport H, et al: Non-Hodgkin’s: A clinicopathologic study comparing two classifications. Cancer 41:303-325, 1978 11. Kaplan EL, Meier R: Nonparametric estimation plete observations. J Am Stat Assoc 53:457-581,1958 12. Peto R, Peto J: Asymptotically efficient procedures. J R Stat Sot 135:185-206,1977
from incom-
rank invariant
test
SURGERY OF NON-HODGKIN’S
235
LYMPHOMA
13. Cox DR: Regression models and life tables. J R Stat Sot 34: 187-220. 1972 14. Magrath IT, Lee YJ, Anderson T, et al: Prognostic factors in Burkitt’s lymphoma: Importance of total tumor burden. Cancer 45:1.507-1515.1980 15. Magrath IT, Janus C, Edwards B, et al: An effective therapy for both undifferentiated (including Burkitt’s) lymphomas and lymphoblastic lymphomas in children and young adults. Blood 63:1102-1111. 1984 16. Csako G, Magrath IT, Elin R: Serum total and isoenzyme lactic dehydrogenase activity in American Burkitt’s lymphoma. Am J Clin Pathol78:712-717, 1982 17. Hagberg H, Killander A, Simonnsson B: Serum beta-2 microglobulin in malignant lymphoma. Cancer 51:2220-2225, 1983 18. Wagner DK, Kiwanuka J, Edwards BL. et al: Soluble interleukin II receptor levels in patients with undifferentiated and lymphoblastic lymphomas. J Clin Oncol5:1262-1274,1987
19. Iverson U, Iverson African cases of Burkitt’s Cancer 8:305-310,1972
OH, Ziegler JL, et al: Cell kinetics of lymphoma. A preliminary report. Em J
20. Cohen LF, Balow JE, Magrath IT: Acute tumor lysis syndrome: A review of 37 patients with Burkitt’s lymphoma. Am J Med h&486-491,1980 21. Meyers PA, Potter VP, Wollner during initial treatment for childhood Cancer 56:259-261, 1985
N, et al: Bowel perforation non-Hodgkin’s lymphoma.
22. Smith SD, Morgan R, Link MP, et al: Cytogenetic and immunophenotypic analysis of cell lines established from patients with T cell leukemiailymphoma. Blood 67:650-656, 1986 23. Le Beau MM, Rowley JD: Chromosomal leukemia and lymphoma: Clinical and biological Hum Genet 15:1-54, 1986
abnormalities in significance. Adv
Discussion D.M. Hays (Los Angeles, CA): This report has been almost a decade in development and almost “died” on several occasions. I particularly want to thank Drs LaQuaglia and Stolar for picking up this project in midstudy and carrying it to a successful conclusion. Otherwise these data would have been lost. I think Dr LaQuaglia has very correctly pointed out that patients with non-Hodgkin’s lymphoma with actual or simulated acute abdominal conditions are the ones that receive early therapy and have the best survival. Second, when one considers localization of the tumor, tumor resection, and improved survival, localization is the only independent predictor of survival. First question, Dr LaQuaglia, the P values that you show for the aggregate group, that is all therapies, in comparison with the group receiving specific therapies were different and the P values were more significant in the aggregate group than in the specific group, which I find unexplainable. It may be that the numbers were larger in that group. Third, I would like you to visualize yourself in the operating room at 3 AM with an intussusception that can’t be reduced by hydrostatic pressure and the question comes up as to what it is. Is it an irreducible intussusception, is it really a tumor, how do you make this distinction? Do you do biopsies? Can you tell right away? Do you do a “blind” resection? How do you handle it? M.P. LaQuaglia (response): The reason for doing a parallel analysis on a subgroup of the patients in the study was as you remember the CCG-551 study was
used to compare the LSA2-L2 regimen. The surprising finding from that was that disseminated lymphoma was better for the undifferentiated, so it was Dr Krailo, the statistician, who felt that we should do a parallel analysis of the smaller group of patients who had the more specific therapy. Now it’s controversial that there is a difference between the regimens. The National Cancer Institute studies and studies from the pediatric oncology group don’t necessarily support that distinction but we did the subgroup analysis, which reduced the numbers. The statistical difference therefore, the power, is reduced. However, both our institutional statistician, Mark Heller, and Mark Krailo at CCG agree that the general trend is the same and if we had 100 patients with specific therapy for histopathology, the same conclusions would accrue. And indeed in multivariate analysis the same conclusions would accrue. We put those in for completeness. The second question was what do you do in the operating room and I can’t be definitive about I think that many of these patients are going to present urgently and you just have to rely on surgical judgment. It may be that you will have to do a sleeve resection. The only caveat, you can’t do cytogenetics and immunophenotyping on formalin tissue so that some awareness of the possibility that an appendicitis, intussusception, or bowel perforation may be lymphoma may be helpful in terms of getting some of the tissue stored in the refrigerator so that the more esoteric studies can be done.