Fax: (901) 521-9005
The impact of early resection of primary neuroblastoma on the survival of children older than 1 year of age with stage 4 disease
The St. Jude Children's Research Hospital experience
Article first published online: 15 NOV 2005
Copyright © 2005 American Cancer Society
Volume 104, Issue 12, pages 2837–2846, 15 December 2005
How to Cite
McGregor, L. M., Rao, B. N., Davidoff, A. M., Billups, C. A., Hongeng, S., Santana, V. M., Hill, D. A., Fuller, C. and Furman, W. L. (2005), The impact of early resection of primary neuroblastoma on the survival of children older than 1 year of age with stage 4 disease. Cancer, 104: 2837–2846. doi: 10.1002/cncr.21566
- Issue published online: 8 DEC 2005
- Article first published online: 15 NOV 2005
- Manuscript Accepted: 27 JUL 2005
- Manuscript Revised: 26 JUL 2005
- Manuscript Received: 9 MAR 2005
- Childhood Cancer Solid Tumor Program Project. Grant Number: CA-23099
- Cancer Center Support (CORE)
- National Cancer Institute. Grant Number: CA-21765
- American Lebanese Syrian Associated Charities (ALSAC)
- surgical procedures;
- disease-free survival;
- postoperative complications
It remains unclear whether primary tumor resection benefits patients with metastatic neuroblastoma. The authors assessed the impact of extent and timing of resection on outcome in these patients.
The authors reviewed the records of 124 patients > 1 year of age at diagnosis of International Neuroblastoma Staging System Stage 4 neuroblastoma. The survival estimates of those who did and did not have a gross total resection (GTR) and of those who had initial versus delayed GTR were compared. Surgical complications were reviewed.
The 5-year survival estimates were comparable for the 90 patients who had a GTR and the 17 who underwent surgery but did not have a GTR (29.9% ± 5.1% [standard error] vs. 29.4% ± 10.1%). The 7 patients who underwent GTR at the time of diagnosis had a higher 5-year survival estimate than the 83 patients who had a GTR after induction chemotherapy (83.3% ± 13.9% vs. 25.2% ± 5.0%) (P = 0.001). Five-year event-free survival estimates were similarly higher in the initial-GTR group (57.1% ± 18.7% vs. 14.5% ± 4.2%) (P = 0.002). These two groups did not differ significantly in age at diagnosis (P = 0.118), site of primary tumor (P = 0.34), MYCN amplification status (P = 1), serum lactate dehydrogenase activity at diagnosis (P = 0.34), or treatment protocol (P = 0.22). Twenty-two (21%) patients had a surgical complication.
In this small cohort of patients with metastatic neuroblastoma, GTR at the time of diagnosis offered a survival benefit. Further prospective studies are warranted before this approach can be applied to all patients with metastatic neuroblastoma. Cancer 2005. © 2005 American Cancer Society.
Neuroblastoma is the most common extracranial solid tumor of childhood. Although significant progress has been made in the treatment of infants (younger than 1 yr old),1, 2 most older children have metastatic disease at the time of diagnosis and have a poor prognosis.3 Surgical resection is the treatment of choice for localized neuroblastoma,2–4 but the role and optimal timing of surgery have not been determined for older patients with advanced-stage disease.5–37
We compared the extent and timing of resection with outcome in children older than 1 year of age at the time of diagnosis of International Neuroblastoma Staging System (INSS) Stage 438 neuroblastoma who were treated at St. Jude Children's Research Hospital between 1984 and 2001.
MATERIALS AND METHODS
This retrospective study was approved by the Institutional Review Board of St. Jude Children's Research Hospital, which also approved all institutional treatment protocols. We reviewed the records of 125 children older than 1 year of age at the time of diagnosis of INSS Stage 4 neuroblastoma who were treated at St. Jude between April 1984 and February 2001. Patients who had an initial diagnosis of resectable, localized disease but who later developed metastatic neuroblastoma were excluded. One 13-month-old patient had limited metastatic disease with favorable histology (Shimada classification) and a DNA index of 1.76 (not high risk) and was excluded. This patient had an initial gross total resection (GTR) and became a long-term survivor. Seventeen of the remaining 124 patients were excluded because they did not undergo surgery; 5 of these patients had complete responses to chemotherapy; 12 had no response to chemotherapy and surgery was not possible. The 107 patients analyzed were enrolled on one of four consecutive institutional treatment protocols, with appropriate written informed consent from each patient or parent. Protocol therapy comprised 6 cycles of induction chemotherapy followed by surgical resection of remaining disease and consolidation with either chemotherapy (NB84 and NB8814) or intensive chemotherapy and autologous hematopoietic stem cell rescue (NB91 and NB97) (Table 1). Radiation therapy was not a planned part of any of these protocols.
|Course||NB 84a||NB 88||NB 91||NB 97|
|4||Cyclophosphamide||Cisplatin||High-dose cisplatin||High-dose cisplatin|
|Stem cell harvest|
|6||Cyclophosphamide||Cyclophosphamide||High-dose cisplatin||High-dose cisplatin|
|Evaluation and surgery||Evaluation and surgery||Evaluation and surgery with bone marrow harvest||Evaluation and surgery|
|Stem cell rescue|
|Stem cell rescue|
|9||Cyclophosphamide||Cyclophosphamide||Interferon-α (16 wks)||Eligible for tumor vaccine study|
|Cumulative dosages, mg/m2|
Demographic, clinical, and tumor cell data, including serum lactate dehydrogenase (LDH) values, ferritin concentration, and MYCN gene amplification, were extracted from patients' records. The primary tumor site and extent of metastatic disease were determined by physical examination, diagnostic imaging, bone marrow studies, and/or surgical exploration. The primary tumor site was categorized as favorable (thoracic39, 40 or pelvic41) or unfavorable (adrenal, subdiaphragmatic nonadrenal, or cervical). Before 1988, each patient was assigned a Pediatric Oncology Group (POG) stage at the time of diagnosis; these patients' INSS stage was determined retrospectively. After publication of the INSS staging system in 1988,38 newly diagnosed patients were assigned an INSS stage. We reviewed diagnostic imaging, operative, and pathology reports to confirm that each patient's stage was consistent with the revised INSS criteria.42 Shimada histologic risk classification43 was determined retrospectively for patients who underwent an initial resection and were treated on the NB 84, NB 8814, or NB 91 protocol. Patients treated on NB 97 who had an initial resection were prospectively assigned a Shimada histologic risk classification.
MYCN amplification was defined as a gene copy number > 3, determined by Southern blot analysis44, 45 in samples obtained before 1990 and by fluorescence in situ hybridization (FISH) in samples obtained after January 1990.44, 46 Serum LDH activity > 1500 U/L was considered elevated, as was a ferritin concentration > 142 ng/mL.
Surgical data obtained comprised the procedure(s), timing in relation to chemotherapy, extent of resection, and complications. Significant surgical complications were defined as blood loss estimated to be > 50% of volume, damage to or loss of an organ, damage to a major blood vessel, evidence of an immediate postoperative complication (e.g., infection, hypotension, oliguria, or hypoxemia), and postoperative small bowel obstruction. Data were grouped on the basis of timing of surgery (initial vs. delayed) and the extent of resection reported by the surgeon. Initial surgery was done at the time of diagnosis, whereas delayed surgery followed several courses of chemotherapy. Extent of resection was defined as follows: gross total resection (GTR), removal of more than 95% of visible tumor from the primary site; subtotal resection (STR), removal of more than 50% but less than 95% of visible tumor from the primary site; less than GTR (< GTR), removal of less than 95% of visible tumor from the primary site; less than STR (< STR), removal of less than 50% of visible tumor from the primary site. All resections included removal of involved regional lymph nodes.
As part of the NB 97 protocol, an attending surgeon prospectively evaluated patients' diagnostic imaging studies to determine the frequency of resectable disease at the time of diagnosis and the point in therapy at which disease that is unresectable at diagnosis becomes resectable.47 Tumors were considered unresectable if there was significant involvement of major vascular structures or contiguous organs or if nephrectomy was required. The imaging studies were conducted at the time of diagnosis, after completion of two courses of topotecan, and after completion of remission induction therapy.
Detection of MYCN Amplification by FISH
Several patients' tumor MYCN status was not determined before therapy but was determined from archived tumor samples by FISH. Briefly, dual-color FISH assays were performed on formalin-fixed, paraffin-embedded tumor sections as previously reported.48 Bacterial artificial chromosome (BAC)-derived probes targeting MYCN (2p24.1; RP11-355H10, Invitrogen, Carlsbad, CA) and (2q35; contig of RP11-296A19 and 384O8, Invitrogen) were used. DAPI (0.5 μL/mL) (Insitus Biotechnologies, Albuquerque, NM) was used as a nuclear counterstain. Tumors whose nuclei contained numerous MYCN signals and a MYCN:2q35 ratio > 5 were considered to have MYCN amplification.
The duration of survival was defined as the interval between diagnosis and death or last follow-up visit. The duration of event-free survival (EFS) was defined as the interval between diagnosis and relapse, disease progression, second malignancy, death, or last follow-up visit. Survival and EFS distributions were estimated by the method of Kaplan and Meier,49 and the associated standard errors were calculated by the method of Peto and Pike.50 The exact log-rank test was used to compare survival and EFS distributions. Fisher exact test, the Wilcoxon exact rank-sum test, and the exact chi-square test were used to compare patient characteristics and biologic data according to the extent and timing of resection. Fisher exact test was used to compare the rate of complications among surgical groups.
The characteristics of the 107 patients are shown in Table 2. At the time of analysis, 25 (23%) patients were alive, with a median follow-up of 7.4 years (range, 2.1–17.8 yrs). Twenty-two of the 25 (88%) survivors had undergone follow-up within the last year, and all had been seen or contacted within the past 2.7 years.
|Age at diagnosis|
|< 2 yrs||26 (24.3)|
|≥ 2 yrs||81 (75.7)|
|Not amplified||64 (59.8)|
|Not available||13 (12.2)|
|≥ 1500||18 (16.8)|
|Not available||1 (0.9)|
|≥ 142||54 (50.5)|
|Not available||29 (27.1)|
Extent of Resection
Figure 1 shows the surgical treatment of the 124 patients reviewed. Ten patients underwent primary tumor resection before chemotherapy (7 GTRs; 3 STRs) and before referral to our institution. The indication for surgery was either presumed diagnosis of Wilms tumor or initial diagnosis of locoregional neuroblastoma. Most patients (n = 97) underwent primary tumor resection after 6 cycles of multiagent chemotherapy (Table 1). Eighty-eight of these 97 patients underwent delayed definitive surgery (83 GTRs; 5 STRs). The remaining nine patients underwent other delayed surgical procedures (< STR). Seventeen patients did not undergo a surgical procedure and were excluded from further analysis.
Impact of Extent of Resection (GTR vs. < GTR)
Characteristics of the 90 patients who underwent GTR were compared with those of the 17 patients who had < GTR (Table 3). No statistically significant difference was observed in MYCN status (P = 1), treatment protocol (P = 0.50), age at diagnosis (P = 0.96), or site of primary tumor (P = 1). Serum LDH activity at diagnosis was lower in patients who had < GTR (P = 0.071). Although ferritin concentration was unavailable for approximately 30% of patients, we found no significant association between extent of resection and normal versus elevated ferritin concentration.
|Characteristic||GTR N = 90||< GTR N = 17||P value|
|Male||53 (59)||10 (59)|
|Female||37 (41)||7 (41)|
|White||69 (77)||13 (76)|
|Black||15 (17)||2 (12)|
|Hispanic||2 (2)||1 (6)|
|Other||4 (4)||1 (6)|
|Age at diagnosis|
|1–2 yrs||19 (21)||7 (41)||0.119|
|> 2 yrs||71 (79)||10 (59)|
|Adrenal||68 (76)||6 (35)|
|Nonadrenal||16 (18)||11 (65)|
|Thoracic||5 (6)||0 (0)|
|Cervical||1 (1)||0 (0)|
|Amplified||26 (29)||4 (24)|
|Not amplified||54 (60)||10 (59)|
|Not available||10 (11)||3 (18)|
|0–1499||71 (79)||17 (100)|
|≥ 1500||18 (20)||0 (0)|
|Not available||1 (1)||0 (0)|
|0–141||19 (21)||5 (29)|
|≥ 142||48 (53)||6 (35)|
|Not available||23 (26)||6 (35)|
|NB84||35 (39)||10 (59)|
|NB88||17 (19)||2 (12)|
|NB91||21 (23)||3 (18)|
|NB97||17 (19)||2 (12)|
There was no evidence of association between survival or EFS and the extent of surgery (P = 1 for both tests; Fig. 2). Five-year survival and EFS estimates were 29.9% ± 5.1% and 17.8% ± 4.5% for patients with GTR and 29.4% ± 10.1% and 17.6% ± 8.0% for patients with < GTR, respectively.
Impact of Initial versus Delayed GTR
Most patients underwent GTR after chemotherapy (n = 83). All of the seven patients who had initial GTR had unfavorable Shimada histologic features and six of the seven had bone or bone marrow metastases (Table 4). Despite the small number of patients who had initial GTR, we analyzed patient characteristics and outcome according to the timing of GTR. We found no significant difference between the initial and delayed GTR groups in treatment protocol, serum LDH elevation, primary tumor site (favorable vs. unfavorable), or MYCN amplification (Table 5), nor did the groups differ by gender (P = 0.44) or race (White vs. Other) (P = 1). Patients who had initial GTR were slightly older at diagnosis (3.9 yrs vs. 3.0 yrs) (P = 0.118). Ferritin concentration was unavailable for approximately 25% of patients, thus preventing statistical analysis.
|Patient||Age in yrs||Primary site||Shimada histologic risk assessment||MYCN amplification||Metastatic sites|
|Lymph node(s)||Bone||Bone marrow||Other|
|1||3.7||Right adrenal||Unfavorable||Not Amp||NE||Yes||No||No|
|4||3.9||Left adrenal||Unfavorable||Not Amp||No||No||Yes||No|
|5||4.3||Left adrenal||Unfavorable||Amp||Yes||No||Yes||Liver, pleural effusion|
|6||16.8||Thoracic||Unfavorable||Not Amp||No||Yes||No||Pleural effusion|
|Initial GTR n (%)||Delayed GTR n (%)||P value|
|Age at diagnosis|
|1–2 yrs||1 (14)||18 (22)||1.0|
|< 2 yrs||6 (86)||65 (78)|
|Adrenal||5 (71)||63 (76)|
|Nonadrenal||1 (14)||15 (18)|
|Thoracic||1 (14)||4 (5)|
|Pelvic||0 (0)||0 (0)|
|Cervical||0 (0)||1 (1)|
|Amplified||2 (29)||24 (29)|
|Not amplified||4 (57)||50 (60)|
|Not available||1 (14)||9 (11)|
|0–1499||7 (100)||64 (77)|
|≥ 1500||0 (0)||18 (22)|
|Not available||0 (0)||1 (1)|
|0–141||4 (57)||15 (18)|
|≥ 142||0 (0)||48 (58)|
|Not available||3 (43)||20 (24)|
|NB84||1 (14)||34 (41)|
|NB88||2 (29)||15 (18)|
|NB91||1 (14)||20 (24)|
|NB97||3 (43)||14 (17)|
Initial GTR was associated with a better outcome. Five-year EFS estimates were 57.1% ± 18.7% for patients who underwent initial GTR versus 14.5% ± 4.2% for patients who underwent delayed GTR (P = 0.002; Fig. 3A). Five-year survival estimates were 83.3% ± 13.9% for patients who underwent initial GTR and 25.2% ± 5.0% for patients who underwent delayed GTR (P = 0.001; Fig. 3B).
Information about surgical complications was available for 81 of the 83 patients who had delayed GTR; 19 (23%) of these patients had complications, compared with 3 of 7 (43%) patients who had initial GTR. However, no statistically significant difference was observed (P = 0.36). The most common complication in the initial-GTR group was nephrectomy. This finding reflects the presumed diagnosis of Wilms tumor in most of these patients.
Outcome According to Prospective Resectability
The above findings suggest that patients with metastatic disease who undergo initial GTR have a better outcome. However, it is difficult to rule out the possibility that patients with resectable disease at diagnosis have a better prognosis regardless of the timing of surgery. The imaging studies of 22 patients treated on NB97 who did not undergo initial resection were analyzed to determine at what point the tumors appeared resectable.47 Seven patients were judged to have resectable tumors at the time of diagnosis but did not receive an initial GTR. We found no statistically significant difference in survival (P = 0.113) or EFS (P = 0.204) between these 7 patients and the 15 patients considered to have unresectable tumors at diagnosis (Fig. 4).
In this group of 107 patients older than 1 year of age at diagnosis of metastatic neuroblastoma who were treated on well defined protocols, initial GTR was associated with a significantly improved likelihood of survival, which is otherwise poor in this group. Most available studies have not reported similar findings.
Numerous studies have assessed the impact of the extent of resection on survival in patients with metastatic neuroblastoma, with contradictory results. Studies in the 1980s both did19, 28 and did not30, 34 demonstrate improved survival of patients with metastatic disease when the primary tumor was resected. CCSG study 321P221 demonstrated that total resection at some point during therapy markedly improved the survival of patients with metastatic neuroblastoma. The Study Group of Japan35 reported that total resection improved the 2-year survival of patients with metastatic neuroblastoma. Subsequent studies have reported no improvement in overall survival23, 29, 33 and a survival benefit6, 27, 36 for patients with metastatic disease whose primary tumor is totally resected. Recent studies in Germany,16 Spain,9 and Japan,51 at Memorial Sloan-Kettering,17 and by the CCSG52 have tried to assess the importance of GTR in the management of metastatic neuroblastoma. The German study16 found that total resection improved outcome only for patients whose tumors had MYCN amplification. In the Spanish study,9 outcome was influenced more by metastatic relapse than by the extent of primary tumor resection. Both U.S. studies17, 52 concluded that complete resection may improve the survival of patients with metastatic neuroblastoma.
Clearly, there is no consensus as to whether total resection of the primary tumor improves the survival of patients with metastatic neuroblastoma. All studies to date, including ours, have been retrospective and are, thus, plagued by biases inherent in such studies. In addition, in most of these studies including ours, the majority of children received chemotherapy before primary tumor resection. With this therapeutic strategy, chemoresponsive tumors are much more likely to undergo resection than relatively chemoresistant tumors. Therefore, patients with chemoresponsive disease, who have an increased likelihood of survival, may be overrepresented in the delayed-GTR group. Our study showed that even with that potential bias toward improved outcome in the delayed-resection group, there was no evidence that total resection after induction chemotherapy was associated with improved survival.
Our second question was whether the timing of total resection affected survival in this set of patients. The current approach is to delay definitive surgical resection of high-risk neuroblastoma until after several courses of chemotherapy to minimize surgery-related morbidity, particularly in older children with bulky, locally invasive (INSS 3) or metastatic (INSS 4) neuroblastoma. Pretreatment reportedly results in smaller, firmer tumors,5 less significant intraoperative bleeding, and easier dissection of tumor from vital structures.9, 18, 32 The three reports that include the impact of the timing of surgery on survival in CCSG studies between 1975 and 199021, 30, 34 indicated no advantage to early resection except a trend toward improved survival (P = 0.093) on CCSG studies 371P and 371.30 Shamberger32 also reported a trend toward improved survival of patients > 1 year of age at diagnosis of metastatic neuroblastoma who underwent initial primary tumor resection and were treated on institutional protocols. Total resection at the time of diagnosis improved local control in a study reported by Matthay et al.53 In an analysis by La Quaglia,27 GTR at diagnosis did not improve the survival of 70 patients > 1 year of age with metastatic neuroblastoma treated at a single institution between November 1979 and August 1992. Several other studies have found no survival advantage for patients with metastatic disease who undergo initial resection of the primary tumor.16, 33, 35, 52
Unlike most reports, ours shows a significant survival advantage for patients with metastatic disease who undergo initial GTR of the primary tumor. One possible explanation is a selection bias. Patients who had an initial GTR in our study may have had less tumor burden at diagnosis (most had bone or bone marrow disease but not both) than patients in the delayed-GTR group. Another possibility is that the patients who underwent initial resection had a more favorable prognosis in any case; however, the initial-GTR and delayed-GTR groups did not differ significantly in any of the known prognostic indicators we examined.
An additional possibility is that resectability at diagnosis is an independent prognostic factor. A few studies have investigated whether resectability at diagnosis is associated with more favorable biologic characteristics. Haase et al.22 studied 58 patients with Evans Stage III disease and concluded that certain favorable prognostic features, such as age and histopathology, do not necessarily predict tumor resectability. A follow-up report of 52 children with Stage IV disease and 9 with Stage III disease found no difference in resectability on the basis of MYCN amplification.21 In the St. Jude NB 97 study, each patient's diagnostic imaging was assessed prospectively to determine when the tumor became resectable. Tumors that were resectable at diagnosis were not found to differ from others in primary site or MYCN status.47 In addition, patients who had a resectable tumor at diagnosis but did not undergo initial resection did not have a better outcome than those judged to have an unresectable primary tumor at diagnosis. Therefore, resectability at diagnosis does not appear to be associated with favorable biologic characteristics (such as absence of MYCN amplification) and does not, in itself, appear to be a favorable prognostic factor if the tumor is not resected at that time.
If initial resection improves the outcome of patients with metastatic neuroblastoma, then a biologically plausible explanation must be considered. Our findings support the hypothesis that neuroblastoma is curable below a critical threshold of tumor burden. According to the Goldie and Coldman hypothesis,54, 55 an important cause of treatment failure is the development of resistant clones of tumor cells because of spontaneous somatic mutations. In that case, surgical debulking before the initiation of chemotherapy would reduce the number of tumor cells and, thereby, reduce the likelihood of spontaneous mutation.
This hypothesis is not without precedent. In one study, patients with metastatic renal cell carcinoma who were candidates for nephrectomy were randomly assigned to undergo nephrectomy followed by therapy with interferon α-2b versus therapy with interferon α-2b alone. The median survival was 11.1 months for the 120 patients who underwent nephrectomy versus 8.1 months for the 121 who received only interferon α-2b (P = 0.05).56 The biologic mechanism by which removal of the primary tumor improves the survival of patients with metastatic disease is unclear. These patients with renal cell carcinoma did not receive cytotoxic therapy, and, therefore, the Goldie and Coldman hypothesis54 should not apply. It is possible that the primary tumor secretes factors that influence the growth of metastatic disease.
Given the limited evidence from retrospective studies, clinicians must weigh the risks and benefits of initial GTR for patients with high-risk neuroblastoma. The frequency of complications is high. Several groups have reported infection, diarrhea, obstructive adhesions, hemorrhage caused by vascular injury or laceration of internal organs, and urologic complications due to nephrectomy.18, 22–24, 29 We observed many of these complications as well. However, we did not find a statistically significant increase in the frequency of surgical complications in patients who underwent initial GTR. This finding suggests that initial surgical resection does not pose a significantly greater risk than delayed surgical resection.
Our study suggests that initial surgical resection improves the overall outcome of patients with metastatic high-risk neuroblastoma. However, the nonprospective nature of this study and the small number of patients may have introduced bias. At this time, aggressive surgical resection of the primary tumor is not indicated for all patients with metastatic neuroblastoma. However, pediatric oncologists and surgeons should address the questions of the optimal timing of surgical resection and the appropriate candidates for early resection through appropriately controlled and monitored prospective studies.
The authors thank Sharon Naron for editorial assistance and Mickey Cain, Valerie McPherson, Joanna Hwang, and Loraine Avery for data management.