- Email: [email protected]
Biological variables in thoracic neuroblastoma: A Pediatric Oncology Group study
Biological Variables in Thoracic Neuroblastoma: A Pediatric Oncology Group Study By Jonathan A. Morris, Stephen J. Shochat, E. Ide Smith, A. Thomas Look, Garrett M. Brodeur, Alan B. Cantor, and Robert P. Castleberry
Stanford, California; Dallas, Texas; Memphis, Tennessee; Philadelphia, Pennsylvania; Tampa, Florida; and Birmingham, Alabama • The prognosis for patients with neuroblastoma is related to the age and stage at time of presentation, as well as to the presence or absence of biological markers such as N-myc amplification and the degree of DNA ploidy. However, previous studies have shown that neuroblastoma in the thoracic site also is a favorable prognostic indicator, in that children with mediastinal neuroblastoma have a better survival rate, regardless of age or stage at time of presentation. This study was designed to evaluate the biological differences between thoracic and nonthoracic neuroblastoma with respect to N-myc amplification, DNA index as a measure of DNA ploidy, serum lactate dehydrogenase levels, and serum ferritin levels. Patients enrolled in the Pediatric Oncology Group study protocols for neuroblastoma were evaluated retrospectively, and log-rank analysis allowed the impact of each biological variable on survival to be determined for each cohort of patients. There were 1,335 neuroblastoma patients in the data base; 227 had thoracic-site neuroblastoma. Through analysis, it was apparent that patients with thoracic neuroblastoma have better survival rates than do their nonthoracic counterparts (P < .0001), and they are less likely to have N-myc amplification (P -- .001), more likely to have an LDH level of less than 1,500 (P < .0001), and usually have a DNA index of greater than 1 (P < .003). Both thoracic and nonthoracic patients have low serum ferritin levels (86% of thoracic versus 83% of nonthoracic patients). Thoracic-site patients tend to be younger than their nonthoracic counterparts (P < .0001), and they present at a lower surgicopathologic stage (P < .0001). Thoracic site was prognostically favorable for survival, when univariately stratifying for DNA index (P < .0001), lactate dehydrogenase (LDH) level (P < .0001), or age (P < .O001l. In addition, thoracic site is associated with a more favorable prognosis for patients with stage C disease (P = .001), and there is a trend toward a favorable effect among stage D patients (P = .06). Patients with a thoracic site and nonamplified N-myc have a better chance of survival than do patients with nonthoracic sites who have Nomyc nonamplification (P = .002); however, there were too few
thoracic patients with N-myc amplification to determine its significance among them (P = .40). Although patients with thoracic neuroblastoma have a favorable biological profile (N.myc nonamplified, DNA index >1, LDH level <1,500, low serum ferritin level) and clinical profile (age <1 year, favorable stage), this may not fully explain their better survival rates.
Copyright© 1995by W.B. SaundersCompany INDEX WORDS: Neuroblastoma, thoracic site, biological variables.
HE LAST DECADE has brought invaluable improvements in the understanding of the biologiT cal activity of neuroblastoma. N-myc amplification
From Stanford University Medical Center, Stanford, CA; The University of Texas SouthwesternMedical School, Dallas, TX; St Jude Children's Research Hospital, Memphis, TN; The Universityof Pennsylvania School of Medicine, Philadelphia, PA; The Pediatric Oncology Statistical Office and Universityof South Florida, Tampa, FL; and The Universityof Alabama School of Medicine, Birmingham,AL. Presented at the 25th Annual Meeting of the American Pediatric SurgicalAssociation, Tucson, Arizona, May 14-17, 1994. Supported in part by Grant No. CA-30969 from the National Cancer Institute of the National Institutes of Health, Bethesda, MD. Address reprint requests to Stephen J. Shochat, MD (8741), c/o Pediatric Oncology Group, 645 N Michigan Ave, Suite 910, Chicago, IL 60611. Copyright © 1995 by W.B. Saunders Company 0022-3468/95/3002-0025503.00/0
and DNA index as a measure of tumor cell ploidy have gained acceptance as predictive biological markers in neuroblastoma. Patients with N-myc amplification have rapid tumor progression and poor clinical outcome regardless of stage at time of presentation, 1,2 whereas a DNA index of greater than 1 (hyperdiploidy) correlates with improved chemotherapeutic response and overall prognosis) ,4 Four different genetic abnormalities can characterize certain types of tumors, including the loss of heterozygosity on the short arm of chromosome 1, amplification of N-myc, DNA ploidy, and defects in expression or function of the nerve growth-factor receptor? Isolated studies have suggested that serum ferritin levels6 and lactate dehydrogenase levels7 also may be predictive of survival. Other possible laboratory predictors include increased levels of trisialogangliosides, 8 neuropeptide Y expression, 9 and chromogranin A expression, x° but their significance is not clear at this time. Historically, the age at time of presentation and the disease stage at time of diagnosis determine the likelihood of survival for patients with neuroblastoma. In addition, many studies have established that thoracic and mediastinal neuroblastoma are associated with a favorable prognosis, u-z° Adams et al2° recently confirmed that children with mediastinal neuroblastoma have a favorable outcome; however, this was not because of earlier clinical presentation or earlier expression of symptoms. Because the children in their series had better survival rates, even after controlling for age and stage of disease, it was suggested that the basic biology of the tumor might be
296
Journalof PediatricSurgery,Vo130,No 2 (February),1995:pp 296-303
BIOLOGICAL VARIABLES IN THORACIC NEUROBLASTOMA
different from that of neuroblastoma at other sites. Shochat et a121 first noted that there were differences in ganglioside composition between tumors of thoracic origin and those with similar histological subtypes located in other sites. The present study was designed to determine the clinical and biological features of thoracic neuroblastoma in a large series of patients treated with modern therapy. The variables examined included age, stage, DNA index (degree of hyperdiploidy), N-myc amplification, serum LDH level, and the serum ferritin level, with outcome measured as years of survival. In addition, we sought to determine whether the more favorable outcome of thoracic primary tumors was attributable to their association with favorable features, or whether thoracic site was an independent prognostic variable. MATERIALS AND METHODS All patients were enrolled in Pediatric Oncology Group (POG) protocols. The patients used in the present analysis were from POG trials 8104, 8741/42, 8441, and 8743. Descriptions of the treatment regimens and differences between the study groups have been discussed elsewhere.4,7,2° POG 8104 enrollment occurred from October 1981 until February 1990, POG 8441 from December 1984 until April 1987, POG 8741/42 from April 1987 until December 199I, and POG 8743 from May 1987 until October 1990. One thousand three hundred thirty-five patients were entered in the above POG protocols. All were under 2l years of age and were found to have histologically proven neuroblastoma or ganglioneuroblastoma. The initial clinical evaluation included a chest roentgenogram, metastatic bone survey, bone marrow aspiration, and radionuclide bone scan. The patients were staged according to the POG surgicopathologic staging system (Table 1). Computed tomography, magnetic resonance imaging, and myelograms were performed as necessary. Flow cytometry on stained tumor cell suspensions wa s performed as previously described,4 and the DNA index was determined using the modal channel number of G0/Gi-phase cells versus normal diploid cells. N-myc amplification was calculated using a Southern blot hybridized probe for the N-rnyc oncogene.~ Lactate dehydrogenase (LDH) levels were determined Table 1. POG Neuroblastoma Surgicopathologic Staging System Stage A
DS
Description Complete surgical excision of the primary tumor, histologically negative or positive margins; nonadherent intracavitary lymph nodes, histologically negative for tumor; liver negative for tumor in abdominal and pelvic primary tumor Incomplete surgical resection of primary; lymph nodes histologically negative for tumor, as in A Complete or incomplete surgical resection of primary; nonadherent intracavitary lymph nodes, histologically positive for tumor; liver histologically negative for tumor Disseminated disease beyond intracavitary nodes (ie, bone marrow, bone, liver skin, or lymph nodes beyond the cavity of the primary) Localized primary tumor with dissemination limited to liver, skin, and/or marrow (no evidence of bone metastasis)
297
Table 2. Log-Rank Tests of Selected Biological and Clinical Variables and Their Effects on Survival Prognostic Variable
No. of Patients
Deaths
Expected
P Value
Thoracic site Yes No Age
227 1,108
39 523
120 442
<.0001
< 1 yr > 1 yr Stage
490 845
76 486
247 315
< .0001
A B C D DS DNAindex
211 118 248 675 83
7 15 61 465 14
119 63 109 230 42
<.0001
228 426
129 120
72 177
<.0001
Nonamplified Amplified LDH
396 96
94 73
147 21
< .0001
< 1,500 > 1,500 Ferritin Low High
894 255
299 195
425 69
<.0001
209 42
116 35
130 21
=.0006
1 >1
N-myc
in the standard fashion and are expressed in units per milliliter. Serum ferritin levels were calculated in the standard fashion and are expressed as "low" or "high," depending on the child's age. 22,23 For 0 to 2 month olds, high levels were more than 500 ng/mL; for 2 to 4 month olds, they were more than 400 ng/mL; and for children over 4 months of age, they were more than 150 ng/mL. Survival measurements date from diagnosis until the time of death or the time the patient was last seen for evaluation. The quantitative variables (DNA index, N-rnyc copy number, serum [LDH] and ferritin levels) and age were dichotomized at values previously reported to be prognostic.l-4,6,7,22,23 The association of each variable with site was evaluated using SAS PROC FREQ, with the usual contingency table X2 P value. 24 The association of each variable with survival time was evaluated using the log-rank test,2s and a stratified version of this test was used to assess the prognostic importance of thoracic site after adjusting for the other variables. Survival curves were constructed by the method of Kaplan-Meier,;6 with standard errors of Peto et al. 27
RESULTS
One thousand three hundred thirty-five patients were enrolled in the POG protocols, and all had documentation of primary site of occurrence, age at time of presentation, and stage at time of presentation. One thousand one hundred forty-nine patients had serum LDH levels recorded, and 654 had DNA index recorded. N-mycamplification was recorded for 452, and serum ferritin levels were documented for 251. Because of the large number of missing covariants, it was not deemed valid to conduct a multivariate Cox analysis. Table 2 shows results of unstratified log-rank tests,
298
MORRIS ET AL
in which prognostic variables were correlated with survival. The total number of patients who had data collected for the particular biological variable is listed under "no. of patients." Two hundred twenty-seven patients (17%) were found to have thoracic neuroblastoma, as defined by location of the tumor in the thorax, mediastinum, or diaphragm. Thoracic site is associated with a significantly lower mortality rate; as 39 of 227 patients died as compared to 523 of the 1,108 nonthoracic neuroblastoma patients (P < .0001). Figure 1 shows the survival curves for thoracic and nonthoracic patients. The median survival and follow-up period for all patients was 2.8 years (2.4 years for nonthoracic Patients, 4.9 years for thoracic patients). For all patients with neuroblastoma, the stage and age at tim e of presentation also were shown t0 be highly significant factors with respect to survival (P < .0001). Patients w i t h a DNA index Of greater than 1 (426 patients, 120 of whom died) survived longer than did those with an index of 1 (228 patients, 129 of whom died) (P < .0001). In addition, N-myc amplification is associated with a poorer outcome. Seventy-three of the 96 patients with N-myc amplification died, as did only 94 of the 396 without N-myc amplification (P --- .003). Low levels of serum markers (LDH and ferritin) also were associated with a better prognosis. Two hundred ninety-nine of the 894 patients with low LDH levels died, compared with 195 of the 255 with high LDH levels (P < .0001). One hundred sixteen of the 209 patients with low serum ferritin levels died, compared with 35 of less than the 42 with high serum ferritin levels (P = .0006). The relationship between the frequency of each clinical and biological variable and thoracic site is noted in Table 3. Thoracic patients were younger (.51% were < I year of age) than nonthoracic patients (34% were < 1 year of age) (P < .0001). Thoracic patients also presented at a lower stage (59% stage A or B) than their nonthoracic counterparts (18% stage A or B) (P < .0001); 631 of the 1,108 nonthoracic patients (57%) presented with stage D disease, versus only 44 of the 227 of the thoracic patients (19%).
l°°~sg)
. . . . .
•
0
1
2
_ . . . . . .
......................
C3a=
3
4
5
6
7
8
YeatsFollowed
9
10
11
12
13
14
Fig 1. Survival curves for thoracic (n = 227) and nonth.oracic patients (n = 1,108) (P < ,0001, Numbers above the curves represent cases with follow-up until or beyond the specified time-points.
Table 3. Frequency of Selected Biological and Clinical Variables Among Thoracic and Nonthoracic Patients Variable Age <1 yr > 1 yr Stage A B C D DS DNA index =1 >1
N-myc Nonamplified Amplified LDH < 1,500 > 1,500 Ferritin
Low High
Thoracic
Nonthoracic
227 115 (51%) 112 (49%) 227 80 (36%) 55 (24%) 43 (19%) 44 (19%) 5 (2%) 114 26 (23%) 88 (77%) 84 79 (94%) 5 (6%)
1,108 378 (34%) 730 (66%) 1,108 131 (12%) 63 (6%) 205 (19%) 631 (56%) 78 (7%) 540 202 (37%) 338 (63%) 409 318 (78%) 91 (22%)
191 182 (95%) 9 (5%) 22 i9 (86%) 3 (14%)
958 712 (74%) 246 (26%) 229 190 (83%) 39 (17%)
P Value < .0001
< ,0001
=.003
=.001
< .0001
=.68
Seventy-seven percent of patients with a thoracic site had a DNA index of greater than 1 (P = .003). The majority of thoracic patients (78 of 83) did not have N-myc amplification (P = .001), and 95% of the thoracic patients (182 of 191) had LDH levels below 1,500 (P < .0001). Only five thoracic patients had N-rnyc amplification, and 0nly nine thoracic patients had LDH levels above 1,500. The majority of both thoracic and nonthoracic patients had low serum ferritin levels (86% v 83% respectively; P = .68). Although thoracic patients tend to be younger than thei r nonthoracic counterparts; they have better survival rates regardless of age at time of presentation (Table 4). Among all 112 thoracic neuroblastoma patients under 1 year of age, there were only seven deaths; the thoracic patients over 1 year of age also had a better survival rate (P < .0001) than their older nonthoracic counterparts. Patients who presented with Stage A, B, or DS, both thoracic and nonthoracic, have an equally good survival rates (P = .46, P = 196, and P -- .89, respectively). Only 36 of the 412 combined thoracic and nonthoracic patients in these three staging groups died of disease during the follow-up period. However, patients with stage D disease did poorly, regardless of tumor site (465 of 675 died). The import of thoracic site for survival among this group is marginal (P = .06). The one stage in which thoracic site is associated with a better chance of survival is stage C. Only two of the 43 stage C thoracic patients died of disease, versus 59 of the 205 stage C nonthoracic patients (P = .001).
BIOLOGICAL VARIABLES IN THORACIC NEUROBLASTOMA
299
Table 4. Clinical Variables and the Effect of Thoracic Site on Survival Clinical Variable
Thoracic
Nonthoracic
n
Deaths
Expected
112 115
7 32
19 93
227
39
A
80
B C
55 43
D DS Total
n
Deaths
Expected
P Value
378 730
69 454
57 393
=.002 < .0001
111
1,108
523
451
<.0001
2
3
131
5
4
=.46
7 2
7 12
63 205
8 59
8 49
=.96 =.001
44 5
27 1
38 1
631 78
438 13
427 13
=.58 =.89
227
39
61
1,108
523
501
=.003
Age < 1 yr > 1 yr Total Stage
Table 5 shows the results of log-rank tests in which thoracic site was compared with each biological variable to determine which are independent determinants of survival for patients with neuroblastoma. Thoracic site was associated with a significantly greater chance of survival, regardless of DNA index. Thoracic patients with a DNA index of greater than i had the best long-term survival rate (Fig 2), whereas the nonthoracic patients with a DNA index of 1 had the lowest survival rate. The stratified log-rank test showed that thoracic site is independent of DNA index (P < .0001). Thoracic patients without N-myc amplification (Fig 3) have a better outcome than nonthoracic patients without amplification (P = .002). However, there is no significant difference between thoracic and nonthoracic patients when N-rnycampli-
fication is present (P = .40). The overall results of the stratified log-rank test show the independence of these two variables (P = .003). Thoracic patients with an LDH level below 1,500 have a significantly better survival rate than that of their nonthoracic counterparts (P < .0001). For patients with LDH levels above 1,500, site is not significantly associated with a better survival rate, although the number of thoracic patients in this group is small. The stratified log-rank test shows that, overall, thoracic site and LDH levels are independent factors with respect to survival among neuroblastoma patients (P < .0001). Thoracic patients with a low serum ferritin level had a better survival rate than their nonthoracic counterparts (P = .03), but there was no significant difference with respect to high ferritin levels (P = .68). This must be
Table 5. Biological Variables and the Effect of Thoracic Site on Survival Biological Variable
Nonthoracic
Thoracic n
Deaths
Expected
n
Deaths
Expected
P Value
26 88
9 12
20 25
202 338
120 108
110 91
=.009 <.0001
114
21
48
540
228
201
<.0001
N-myc Nonamplified Amplified
79 5
10 2
23 4
318 91
84 71
71 70
=.002 =.40
Total
84
12
26
409
155
141
=.003
LDH < 1,500
182
30
73
> 1,500
9
4
8
712 246
269 191
227 187
<.0001 =.125
191
34
81
958
460
413
<.0001
Ferritin Low High
19 3
6 3
14 2
190 39
107 32
99 32
=.03 =.68
Total
22
9
16
229
139
132
=.067
DNA index =1 >1 Total
Total
300
100
MORRIS ET AL
""
80
(73) ...... ~'"-,..,_ 46) (24) ....................... 7) .......................................................
"~.
E
(99)
(43)
........
T~:~. 0 •
% ............
o~.r. ul>I
"\
d'4i.............................................. (24)
0
1
2
3
4
(19)
5
i.....oa~ o1=1
6
7
8
9
10
11
12
13
Years Followed
Fig 2. Survival curves for thoracic (114) and nonthoracic (540) patients for whom the DNA index was recorded (stratified log-rank test; P < .0001). Numbers above the curves represent cases with fonow-up until or beyond the specified time-points,
interpreted with caution because there were only three thoracic patients with elevated ferritin levels, all of whom died. DISCUSSION
Our series is similar to most other populations of neuroblastoma patients in that 17% of our patients had a thoracic primary site, and 37% were under f year of age at the time of presentation. Sixty-nine percent of our patients (923) presented with stage C or D disease. For all patients, primary site as well as stage and age at time of presentation had been recorded. However, not all patients had determinations of DNA index (measured in 654), LDH levels (1149), N-myc amplification (492), or serum ferritin levels (251). It was not until 1984 and 1985 that Brodeur and Seeger first showed that N-myc gene amplification predicted rapid disease progression and poor outcome, and only shortly beforehand was DNA index begun to be recorded on a regular basis. 1-4 Because our patient population was drawn from all modern treatment protocols (1981 to 1991), biological data for the early patients was erratic and incomplete. The outcome for patients with thoracic neuroblastoma is better than that of their nonthoracic counterparts. Some have suggested that this is because thoracic and mediastinal tumors become symptom100
,
(41)
(20)
~
~
a¢ N , on
80
'~
.-..,.J~!~................................. ~32)....................................................................
>
~4o
23
]
0
1
~(;~ )
2
(1)
3
4
Th0~ ~p
5
6
7
8
9
10
Years Followed
Fig 3. Survival curves for thoracic (84) and nonthoracic (409) patients for whom N-myc amplification was recorded (stratified log-rank test; P = ,003). Numbers above the curves represent cases with follow-up until or beyond the specified time-points.
atic earlier in the disease process, 5,92 but this has recently been shown by Adams et al and others to not be true. 7,13,2° Although Adams et al showed that age at time presentation and the stage distribution were different for thoracic neuroblastoma, they also showed that thoracic site was an independent favorable risk factor, even when corrected for age and stage at time of presentation, and suggested that the basic biology of the tumors at different sites is different. Complete clinical data were available for each patient in the present study. The age at time of presentation for thoracic patients was significantly lower than that for nonthoracic patients; 51% of thoracic patients were under 1 year of age, compared with 34% of nonthoracic patients. These findings are similar to our previously published results for a smaller cohort of patients. 2° Stage at time of presentation is known to be an important prognostic variable for children with neuroblastoma. In this series, 135 of the 227 thoracic patients (59%) had stage A or B, whereas 194 of the 1,108 nonthoracic patients (18%) had stage A or B. The fact that so many of thoracic patients presented at a lower stage is significant, and likely plays a large role in their better survival rate. All children with stage A, B, or DS disease have a better prognosis than do patients with stage D disease, regardless of site. The only stage in which thoracic site is associated with a significantly higher chance of survival is stage C; this is because thoracic stage C cases behave like thoracic stage A and B cases. Despite the advances in our understanding of neuroblastoma, stage D patients (57% of all nonthoracic patients in our study) continue to have a poor survival rate, regardless of site of presentation. We have shown that the biology of thoracic neuroblastoma is indeed different from that of neuroblastoma located in nonthoracic sites, but these factors might not fully explain the favorable prognosis associated with thoracic site. A DNA index of greater than 1 is known to be associated with improved survival rates, and this was true for our total patient population. However, thoracic site was an independent predictor of survival, irrespective of DNA index. Six hundred fifty-four patients had data for this variable, and statistically significant differences were noted. N-myc amplification is known to portend a poor prognosis for patients with neuroblastoma. In our series, patients with N-myc amplification had a significantly lower survival rate than did those without amplification. The majority (78%) of nonthoracic patients did not have amplification, and almost all (78 of 83; 94%) the thoracic patients did not. Only five thoracic patients had N-myc amplification, two of
BIOLOGICAL VARIABLES IN THORACIC NEUROBLASTOMA
301
whom died in the follow-up period. Data regarding this group must be interpreted cautiously because there were so few thoracic patients with N-myc amplification. For patients with LDH levels of less than 1,500, thoracic site is associated with a greater chance of survival. However, only nine thoracic patients had LDH levels above 1,500, making survival comparisons difficult among this small group. Despite the fact that only a few patients had high values, there were enough patients overall (in the stratified log-rank test) to show that LDH and thoracic site are independent predictors of survival. Some investigators have suggested that serum ferritin level is an important predictor of outcome. 23 Both thoracic and nonthoracic patients had low serum ferritin levels, as determined by age and level stratification. Elevated serum ferritin portended a poor prognosis, regardless of site (83% overall mortality rate). However, only three thoracic patients had elevated serum ferritin levels (all died), making any statistical comparisons difficult between the elevated thoracic and nonthoracic patients. This study confirms our previous observation, from a smaller cohort of patients, that thoracic site neuroblastoma has a better prognosis than nonthoracic neuroblastoma. Previously we showed that the improved survival rate is not attributable to the earlier presentation or expression of symptoms, and that it is independent of age and stage at time of presentation. We now suggest that although thoracic patients tend to have a favorable biological profile (LDH level <1,500, low serum ferritin level, DNA index = 1, and N-myc nonamplification), the studied variables do not fully explain the improved survival rate for our patients. However, many patients in this series had only one or two biological variables recorded. It is also clear that some trends that emerge make an equal distribution of variables between
thoracic and nonthoracic patients unlikely. Very rarely did thoracic patients have N-rnyc amplification (5 patients), serum LDH levels above 1,500 (9 patients), or high serum ferritin levels (3 patients). There is growing evidence that in neuroblastoma, genetic and molecular differences exist and that survival likely will be related to inherent qualities of the tumor. Although N-myc amplification and DNA index were accepted in the last decade as being useful biological markers that helped predict survival, there are new genetic markers that may be more sensitive and useful in the diagnosis and treatment of neuroblastoma. Tanaka et a128and Chan et a129have studied the expression of the H-ras oncogene product p21, and Nakagawara et aP ° have suggested that high levels of expression of the proto-oncogene TRK is strongly predictive of a favorable outcome in neuroblastoma. Cooper et al have presented evidence 3l that neuroblastoma cells represent arrest at various levels of adrenal medullary cell differentiation and that a process of differentiation/dedifferentiation may be responsible for the seemingly random switch in some tumors from malignant to benign phenotype. The expression of TRK, the function of the nerve growth-factor receptor, and the expression of the H-ras oncogene product p21 are but three examples of new markers that might prove useful in the future for identifying high-risk neuroblastoma patients. We have clearly shown that in a large group of patients for whom survival tends to be favorable (thoracic site) and for whom biological variable data are incomplete, the currently measured variables do not fully explain the better survival rate among these patients. Future studies of neuroblastoma should include a thorough and complete assessment of the effect of biological variables on survival and also should be directed at evaluating the molecular genetics of this group, in an attempt to identify the predictive and therapeutic variables that portend an improved survival.
REFERENCES 1. Brodeur GM, Seeger RC, Schwab M, et al: Amplification of
N-myc in untreated human neuroblastomas correlates with advanced stage of disease. Science 224:1121-1124, 1984 2. Seeger RC, Brodeur GM, Sather H, et al: Association of multiple copies of the N-myc oncogene with rapid progression of neuroblastomas. N Engl J Med 313:1111-1116, 1985 3. Look AT, Hayes FA, Nitschke R, et al: Cellular DNA content as predictor of response to chemotherapy in infants with unresectable neuroblastoma. N Engl J Med 311:231-235, 1984 4. Look AT, Hayes A, Shuster J J, et al: Clinical relevance of tumor cell ploidy and N-myc gene amplification in childhood neuroblastoma: A Pediatric Oncology Group Study. J Clin Oncol 9:581-591, 1991 5. Brodeur GM, Azar C, Brother M, et al: Neuroblastoma:
Effect of genetic factors on prognosis and treatment. Cancer 70:1685-I694, 1992 6. Silber Evans AE, Fridman M: Models to predict outcome from childhood neuroblastoma: The rote of serum ferritin and tumor histology. Cancer Res 51:1426-1433, 1991 7. Shuster JJ, McWilliams NB, Castleberry R, et al: Serum lactate dehydrogenase in childhood neuroblastoma: A Pediatric Oncology Group Recursive Partitioning Study. J Clin Oncol 15:295-303, 1992 8. Shochat SJ, Abt AB, Schengrund CL: VCN-releasable sialic acid and gangliosides in human neuroblastomas. J Pediatr Surg 12:413-418, t977 9. Cohen PS, Cooper M J, Helman LJ, et al: Neuropeptide Y
302
MORRIS ET AL
expression in the developing adrenal gland and in childhood neuroblastoma tumors. Cancer Res 50:6055-6061, 1990 (suppl) 10. Hsiao RJ, Seeger RC, Yu AL, et al: Chromogranin A in children with neuroblastoma: Serum concentration parallels disease stage and predicts survival. J Clin Invest 85:1555-1559, 1990 11. Catalono PW, Newton WA, Williams TE, et al: Reasonable surgery for thoracic neuroblastomas in infants and children. J Thorac Cardiovasc Surg 76:459-464, 1978 12. Jaffe N: Neuroblastoma: Review of the literature and examination of the factors contributing to its enigmatic character. Cancer Treat Rev 3:61-82, 1976 13. Filler RM, Traggis DG, Jaffe N: Favorable outlook for children with mediastinal neuroblastoma. J Pediatr Surg 7:136-143, 1972 14. Kajanti M: Neuroblastoma in 88 children: Clinical features, prognostic factors, results, and late effects of chemotherapy. Ann Clin Res 15:1-68, 1983 (supp139) 15. Pelton JJ, Rattner 1A: Neuroblastoma of the thoracic inlet. J Pediatr Surg 25:547-549, 1990 16. Koop CE: The role of surgery in resectable, nonresectable, and metastatic neuroblastomas. JAMA 205:157-158, 1968 17. deLorimier AA, Bragg KU, Linden G: Neuroblastoma in childhood. Am J Dis Child 118:144-150, 1969 18. Sandstedt B, Jereb B, Eklund G: Prognostic factors in neuroblastoma. Acta Pathol Microbiol Immunol Scand 91:365-371, 1983 19. Swank RL, Fenerman GH, Sieber WK, et al: Prognostic factors in neuroblastoma. Ann Surg 174:428-435, 1971 20. Adams GA, Shochat SJ, Smith EI, et al: Thoracic neuroblastoma: A Pediatric Oncology Group Study. J Pediatr Surg 28:372378, 1993 21. Shochat SJ, Corbelletta NL, Repman MA, et al: A biochemi-
cal analysis of thoracic neuroblastomas: A Pediatric Oncology Group Study. J Pediatr Surg 22:660-664, 1987 22. Saarinen UM, Slimes MA: Serum ferritin in assessment of iron nutrition in healthy infants. Acta Paediatr Scand 67:745-751, 1978 23. Hann HWL, Levy HM, Evans AE: Serum ferritin as a guide to therapy in neuroblastoma. Cancer Res 40:1411-1413, 1980 24. SAS Procedure Guide, Version 6 (ed 3). Cary, NC, SAS Institute Inc, 1990, pp 325-363 (chap 20) 25. Peto R, Peto J: Asymmetricallyefficient rank invariant test procedures. J R Stat Soc 135:185-198, 1972 26. Kaplan EL, Meier R: Non-parametric estimation from incomplete observation. J Am Stat Assoc 53:457-481, 1958 27. Peto R, Pike MC, Armitage P, et al: Design and analysis of randomized clinical trials requiring prolonged observation of each patient: lI analysis and examples. Br J Cancer 35:1-39, 1977 28. Tanaka T, Slamon DJ, Shimoda H, et al: Expression of H-ras oncogene products in human neuroblastomas and the significant correlation with a patients prognosis. Cancer Res 48:1030-1034, 1988 29. Chan HSL, Haddad G, Thorner PS, et al: P-glycoprotein expression as a predictor of the outcome of therapy for neuroblastoma. N Engl J Med 325:1608-14, 1991 30. Nakagawara A, Arima-Nakagawara M, Scavarda NJ, et al: Association between high levels of expression of the TRK gene and favorable outcome in human neuroblastoma. N Engl J Med 328:847-854, 1993 31. Cooper MJ, Steinberg SM, Chatten J, et al: Plasticity of neuroblastoma tumor cells to differentiate along a fetal adrenal ganglionic lineage predicts for improved patient survival. J Clin Invest 90:2402-2408, 1992
Discussion J. Ternberg (St Louis, MO): I t h i n k we all acknowledge that site m a k e s a difference in various tumors. W e have seen the effect of site in r h a b d o m y o s a r c o m a . T h e q u e s t i o n in my m i n d , though, is what m a k e s the difference? It is n o t just the site. T h e r e has to be s o m e t h i n g different a b o u t the t u m o r in that area that m a k e s the difference, a n d I t h i n k this is what is b e i n g e x a m i n e d in this report. It is the stage C p a t i e n t s w h e r e the thoracic site really seems to m a k e a difference with respect to survival. Did they look to see which o n e s had h a d c o m p l e t e resection. D i d that m a k e any difference? T h e o t h e r p o i n t that I did n o t notice in the r e p o r t was some r e f e r e n c e to the t u m o r histology. W e know histology m a k e s a difference in W i l m s ' t u m o r prognosis. W e are b e g i n n i n g to talk a b o u t it now in regard to r h a b d o m y o s a r c o m a . T h e r e is the S h i m a d a classification for n e u r o b l a s t o m a ; have the thoracic cases b e e n e x a m i n e d from this standpoint? I t h i n k that in the future, a n a t o m i c staging n o t be as i m p o r t a n t as it is today. W e are going to be able to p u t all this i n f o r m a t i o n , i n c l u d i n g the m o l e c u l a r biol-
ogy i n f o r m a t i o n , into the c o m p u t e r a n d come out with b e t t e r staging as a basis for t r e a t m e n t stratification, a n d hopefully s o m e t i m e in the f u t u r e we will b e able to refine our t r e a t m e n t . M. Asch (Los Angeles, CA): A l o n g the same lines as D r T e r n b e r g ' s question, was t h e r e any difference b e t w e e n the survival rates for p a t i e n t s whose t u m o r s were able to b e resected d u r i n g the initial o p e r a t i o n a n d those whose t u m o r s r e q u i r e d a second p r o c e d u r e for c o m p l e t e resection ( b e c a u s e they were u n r e s e c t able the first t i m e ) ? J.M. Laberge (Montreal, Quebec): I am n o t very good in statistics, b u t have you d o n e logistic regression analysis to look at all the factors to see w h e t h e r thoracic site is an i n d e p e n d e n t v a r i a b l e ? G.M. Haase (Denver, CO): It a p p e a r s that the biological factors were at least as i m p o r t a n t as the thoracic site, b e c a u s e the u n a m p l i f i e d t u m o r s were associated with a good survival rate, a n d the amplified t u m o r s were i n t e r m e d i a t e with respect to survival. Was there an e v a l u a t i o n within biological groups as o p p o s e d to thoracic groups for resectability?
BIOLOGICAL VARIABLES IN THORACIC NEUROBLASTOMA
Just an observation: There are about 1,000 metastatic neuroblastoma patients in this decade of experience from the Pediatric Oncology Group. There are another 1,000 patients with metastatic disease in the Children's Cancer Group studies. With this combined experience, it seems to me that the two groups could answer the question regarding the importance of resectability for metastatic neuroblastoma. It is something we should get together and accomplish in the next few years. J.A. Morris (response): With respect to Dr Ternberg's first question, yes, there is a difference in stage C patients. Stage C thoracic patients have a significantly higher survival rate than do nonthoracic patients, whether they have complete resection or not. With respect to histology, we were not looking specifically at histology in this study. We believe that a combination of anatomic and clinical presentation, along with identifying the important biological characteristics that can be used to predict survival, will be the key to future staging systems.
303
With respect to Dr Asch's question regarding resection during the initial operation, in a smaller cohort of these patients, previously presented, we showed that regardless of complete or incomplete resection in the thoracic site, survival data were unchanged. Both Dr Laberge and Dr Haase asked about logistic regression analysis. The log-rank test is a form of partition analysis that Dr Cantor, our statistician, uses. It allows us to obtain specific information about smaller groups, and then "collapse" those groups to present data regarding specific variables. The results of a multivariate analysis would be difficult to interpret because not all our patients had all the variables recorded. Within each biological group, Dr Haase, we looked previously at resectability, and the thoracic site did not appear to make a difference. We did not look specifically at the abdominal primary tumors in the other 1,108 nonthoracic patients or at their ability to be resected.
Stanford, California; Dallas, Texas; Memphis, Tennessee; Philadelphia, Pennsylvania; Tampa, Florida; and Birmingham, Alabama • The prognosis for patients with neuroblastoma is related to the age and stage at time of presentation, as well as to the presence or absence of biological markers such as N-myc amplification and the degree of DNA ploidy. However, previous studies have shown that neuroblastoma in the thoracic site also is a favorable prognostic indicator, in that children with mediastinal neuroblastoma have a better survival rate, regardless of age or stage at time of presentation. This study was designed to evaluate the biological differences between thoracic and nonthoracic neuroblastoma with respect to N-myc amplification, DNA index as a measure of DNA ploidy, serum lactate dehydrogenase levels, and serum ferritin levels. Patients enrolled in the Pediatric Oncology Group study protocols for neuroblastoma were evaluated retrospectively, and log-rank analysis allowed the impact of each biological variable on survival to be determined for each cohort of patients. There were 1,335 neuroblastoma patients in the data base; 227 had thoracic-site neuroblastoma. Through analysis, it was apparent that patients with thoracic neuroblastoma have better survival rates than do their nonthoracic counterparts (P < .0001), and they are less likely to have N-myc amplification (P -- .001), more likely to have an LDH level of less than 1,500 (P < .0001), and usually have a DNA index of greater than 1 (P < .003). Both thoracic and nonthoracic patients have low serum ferritin levels (86% of thoracic versus 83% of nonthoracic patients). Thoracic-site patients tend to be younger than their nonthoracic counterparts (P < .0001), and they present at a lower surgicopathologic stage (P < .0001). Thoracic site was prognostically favorable for survival, when univariately stratifying for DNA index (P < .0001), lactate dehydrogenase (LDH) level (P < .0001), or age (P < .O001l. In addition, thoracic site is associated with a more favorable prognosis for patients with stage C disease (P = .001), and there is a trend toward a favorable effect among stage D patients (P = .06). Patients with a thoracic site and nonamplified N-myc have a better chance of survival than do patients with nonthoracic sites who have Nomyc nonamplification (P = .002); however, there were too few
thoracic patients with N-myc amplification to determine its significance among them (P = .40). Although patients with thoracic neuroblastoma have a favorable biological profile (N.myc nonamplified, DNA index >1, LDH level <1,500, low serum ferritin level) and clinical profile (age <1 year, favorable stage), this may not fully explain their better survival rates.
Copyright© 1995by W.B. SaundersCompany INDEX WORDS: Neuroblastoma, thoracic site, biological variables.
HE LAST DECADE has brought invaluable improvements in the understanding of the biologiT cal activity of neuroblastoma. N-myc amplification
From Stanford University Medical Center, Stanford, CA; The University of Texas SouthwesternMedical School, Dallas, TX; St Jude Children's Research Hospital, Memphis, TN; The Universityof Pennsylvania School of Medicine, Philadelphia, PA; The Pediatric Oncology Statistical Office and Universityof South Florida, Tampa, FL; and The Universityof Alabama School of Medicine, Birmingham,AL. Presented at the 25th Annual Meeting of the American Pediatric SurgicalAssociation, Tucson, Arizona, May 14-17, 1994. Supported in part by Grant No. CA-30969 from the National Cancer Institute of the National Institutes of Health, Bethesda, MD. Address reprint requests to Stephen J. Shochat, MD (8741), c/o Pediatric Oncology Group, 645 N Michigan Ave, Suite 910, Chicago, IL 60611. Copyright © 1995 by W.B. Saunders Company 0022-3468/95/3002-0025503.00/0
and DNA index as a measure of tumor cell ploidy have gained acceptance as predictive biological markers in neuroblastoma. Patients with N-myc amplification have rapid tumor progression and poor clinical outcome regardless of stage at time of presentation, 1,2 whereas a DNA index of greater than 1 (hyperdiploidy) correlates with improved chemotherapeutic response and overall prognosis) ,4 Four different genetic abnormalities can characterize certain types of tumors, including the loss of heterozygosity on the short arm of chromosome 1, amplification of N-myc, DNA ploidy, and defects in expression or function of the nerve growth-factor receptor? Isolated studies have suggested that serum ferritin levels6 and lactate dehydrogenase levels7 also may be predictive of survival. Other possible laboratory predictors include increased levels of trisialogangliosides, 8 neuropeptide Y expression, 9 and chromogranin A expression, x° but their significance is not clear at this time. Historically, the age at time of presentation and the disease stage at time of diagnosis determine the likelihood of survival for patients with neuroblastoma. In addition, many studies have established that thoracic and mediastinal neuroblastoma are associated with a favorable prognosis, u-z° Adams et al2° recently confirmed that children with mediastinal neuroblastoma have a favorable outcome; however, this was not because of earlier clinical presentation or earlier expression of symptoms. Because the children in their series had better survival rates, even after controlling for age and stage of disease, it was suggested that the basic biology of the tumor might be
296
Journalof PediatricSurgery,Vo130,No 2 (February),1995:pp 296-303
BIOLOGICAL VARIABLES IN THORACIC NEUROBLASTOMA
different from that of neuroblastoma at other sites. Shochat et a121 first noted that there were differences in ganglioside composition between tumors of thoracic origin and those with similar histological subtypes located in other sites. The present study was designed to determine the clinical and biological features of thoracic neuroblastoma in a large series of patients treated with modern therapy. The variables examined included age, stage, DNA index (degree of hyperdiploidy), N-myc amplification, serum LDH level, and the serum ferritin level, with outcome measured as years of survival. In addition, we sought to determine whether the more favorable outcome of thoracic primary tumors was attributable to their association with favorable features, or whether thoracic site was an independent prognostic variable. MATERIALS AND METHODS All patients were enrolled in Pediatric Oncology Group (POG) protocols. The patients used in the present analysis were from POG trials 8104, 8741/42, 8441, and 8743. Descriptions of the treatment regimens and differences between the study groups have been discussed elsewhere.4,7,2° POG 8104 enrollment occurred from October 1981 until February 1990, POG 8441 from December 1984 until April 1987, POG 8741/42 from April 1987 until December 199I, and POG 8743 from May 1987 until October 1990. One thousand three hundred thirty-five patients were entered in the above POG protocols. All were under 2l years of age and were found to have histologically proven neuroblastoma or ganglioneuroblastoma. The initial clinical evaluation included a chest roentgenogram, metastatic bone survey, bone marrow aspiration, and radionuclide bone scan. The patients were staged according to the POG surgicopathologic staging system (Table 1). Computed tomography, magnetic resonance imaging, and myelograms were performed as necessary. Flow cytometry on stained tumor cell suspensions wa s performed as previously described,4 and the DNA index was determined using the modal channel number of G0/Gi-phase cells versus normal diploid cells. N-myc amplification was calculated using a Southern blot hybridized probe for the N-rnyc oncogene.~ Lactate dehydrogenase (LDH) levels were determined Table 1. POG Neuroblastoma Surgicopathologic Staging System Stage A
DS
Description Complete surgical excision of the primary tumor, histologically negative or positive margins; nonadherent intracavitary lymph nodes, histologically negative for tumor; liver negative for tumor in abdominal and pelvic primary tumor Incomplete surgical resection of primary; lymph nodes histologically negative for tumor, as in A Complete or incomplete surgical resection of primary; nonadherent intracavitary lymph nodes, histologically positive for tumor; liver histologically negative for tumor Disseminated disease beyond intracavitary nodes (ie, bone marrow, bone, liver skin, or lymph nodes beyond the cavity of the primary) Localized primary tumor with dissemination limited to liver, skin, and/or marrow (no evidence of bone metastasis)
297
Table 2. Log-Rank Tests of Selected Biological and Clinical Variables and Their Effects on Survival Prognostic Variable
No. of Patients
Deaths
Expected
P Value
Thoracic site Yes No Age
227 1,108
39 523
120 442
<.0001
< 1 yr > 1 yr Stage
490 845
76 486
247 315
< .0001
A B C D DS DNAindex
211 118 248 675 83
7 15 61 465 14
119 63 109 230 42
<.0001
228 426
129 120
72 177
<.0001
Nonamplified Amplified LDH
396 96
94 73
147 21
< .0001
< 1,500 > 1,500 Ferritin Low High
894 255
299 195
425 69
<.0001
209 42
116 35
130 21
=.0006
1 >1
N-myc
in the standard fashion and are expressed in units per milliliter. Serum ferritin levels were calculated in the standard fashion and are expressed as "low" or "high," depending on the child's age. 22,23 For 0 to 2 month olds, high levels were more than 500 ng/mL; for 2 to 4 month olds, they were more than 400 ng/mL; and for children over 4 months of age, they were more than 150 ng/mL. Survival measurements date from diagnosis until the time of death or the time the patient was last seen for evaluation. The quantitative variables (DNA index, N-rnyc copy number, serum [LDH] and ferritin levels) and age were dichotomized at values previously reported to be prognostic.l-4,6,7,22,23 The association of each variable with site was evaluated using SAS PROC FREQ, with the usual contingency table X2 P value. 24 The association of each variable with survival time was evaluated using the log-rank test,2s and a stratified version of this test was used to assess the prognostic importance of thoracic site after adjusting for the other variables. Survival curves were constructed by the method of Kaplan-Meier,;6 with standard errors of Peto et al. 27
RESULTS
One thousand three hundred thirty-five patients were enrolled in the POG protocols, and all had documentation of primary site of occurrence, age at time of presentation, and stage at time of presentation. One thousand one hundred forty-nine patients had serum LDH levels recorded, and 654 had DNA index recorded. N-mycamplification was recorded for 452, and serum ferritin levels were documented for 251. Because of the large number of missing covariants, it was not deemed valid to conduct a multivariate Cox analysis. Table 2 shows results of unstratified log-rank tests,
298
MORRIS ET AL
in which prognostic variables were correlated with survival. The total number of patients who had data collected for the particular biological variable is listed under "no. of patients." Two hundred twenty-seven patients (17%) were found to have thoracic neuroblastoma, as defined by location of the tumor in the thorax, mediastinum, or diaphragm. Thoracic site is associated with a significantly lower mortality rate; as 39 of 227 patients died as compared to 523 of the 1,108 nonthoracic neuroblastoma patients (P < .0001). Figure 1 shows the survival curves for thoracic and nonthoracic patients. The median survival and follow-up period for all patients was 2.8 years (2.4 years for nonthoracic Patients, 4.9 years for thoracic patients). For all patients with neuroblastoma, the stage and age at tim e of presentation also were shown t0 be highly significant factors with respect to survival (P < .0001). Patients w i t h a DNA index Of greater than 1 (426 patients, 120 of whom died) survived longer than did those with an index of 1 (228 patients, 129 of whom died) (P < .0001). In addition, N-myc amplification is associated with a poorer outcome. Seventy-three of the 96 patients with N-myc amplification died, as did only 94 of the 396 without N-myc amplification (P --- .003). Low levels of serum markers (LDH and ferritin) also were associated with a better prognosis. Two hundred ninety-nine of the 894 patients with low LDH levels died, compared with 195 of the 255 with high LDH levels (P < .0001). One hundred sixteen of the 209 patients with low serum ferritin levels died, compared with 35 of less than the 42 with high serum ferritin levels (P = .0006). The relationship between the frequency of each clinical and biological variable and thoracic site is noted in Table 3. Thoracic patients were younger (.51% were < I year of age) than nonthoracic patients (34% were < 1 year of age) (P < .0001). Thoracic patients also presented at a lower stage (59% stage A or B) than their nonthoracic counterparts (18% stage A or B) (P < .0001); 631 of the 1,108 nonthoracic patients (57%) presented with stage D disease, versus only 44 of the 227 of the thoracic patients (19%).
l°°~sg)
. . . . .
•
0
1
2
_ . . . . . .
......................
C3a=
3
4
5
6
7
8
YeatsFollowed
9
10
11
12
13
14
Fig 1. Survival curves for thoracic (n = 227) and nonth.oracic patients (n = 1,108) (P < ,0001, Numbers above the curves represent cases with follow-up until or beyond the specified time-points.
Table 3. Frequency of Selected Biological and Clinical Variables Among Thoracic and Nonthoracic Patients Variable Age <1 yr > 1 yr Stage A B C D DS DNA index =1 >1
N-myc Nonamplified Amplified LDH < 1,500 > 1,500 Ferritin
Low High
Thoracic
Nonthoracic
227 115 (51%) 112 (49%) 227 80 (36%) 55 (24%) 43 (19%) 44 (19%) 5 (2%) 114 26 (23%) 88 (77%) 84 79 (94%) 5 (6%)
1,108 378 (34%) 730 (66%) 1,108 131 (12%) 63 (6%) 205 (19%) 631 (56%) 78 (7%) 540 202 (37%) 338 (63%) 409 318 (78%) 91 (22%)
191 182 (95%) 9 (5%) 22 i9 (86%) 3 (14%)
958 712 (74%) 246 (26%) 229 190 (83%) 39 (17%)
P Value < .0001
< ,0001
=.003
=.001
< .0001
=.68
Seventy-seven percent of patients with a thoracic site had a DNA index of greater than 1 (P = .003). The majority of thoracic patients (78 of 83) did not have N-myc amplification (P = .001), and 95% of the thoracic patients (182 of 191) had LDH levels below 1,500 (P < .0001). Only five thoracic patients had N-rnyc amplification, and 0nly nine thoracic patients had LDH levels above 1,500. The majority of both thoracic and nonthoracic patients had low serum ferritin levels (86% v 83% respectively; P = .68). Although thoracic patients tend to be younger than thei r nonthoracic counterparts; they have better survival rates regardless of age at time of presentation (Table 4). Among all 112 thoracic neuroblastoma patients under 1 year of age, there were only seven deaths; the thoracic patients over 1 year of age also had a better survival rate (P < .0001) than their older nonthoracic counterparts. Patients who presented with Stage A, B, or DS, both thoracic and nonthoracic, have an equally good survival rates (P = .46, P = 196, and P -- .89, respectively). Only 36 of the 412 combined thoracic and nonthoracic patients in these three staging groups died of disease during the follow-up period. However, patients with stage D disease did poorly, regardless of tumor site (465 of 675 died). The import of thoracic site for survival among this group is marginal (P = .06). The one stage in which thoracic site is associated with a better chance of survival is stage C. Only two of the 43 stage C thoracic patients died of disease, versus 59 of the 205 stage C nonthoracic patients (P = .001).
BIOLOGICAL VARIABLES IN THORACIC NEUROBLASTOMA
299
Table 4. Clinical Variables and the Effect of Thoracic Site on Survival Clinical Variable
Thoracic
Nonthoracic
n
Deaths
Expected
112 115
7 32
19 93
227
39
A
80
B C
55 43
D DS Total
n
Deaths
Expected
P Value
378 730
69 454
57 393
=.002 < .0001
111
1,108
523
451
<.0001
2
3
131
5
4
=.46
7 2
7 12
63 205
8 59
8 49
=.96 =.001
44 5
27 1
38 1
631 78
438 13
427 13
=.58 =.89
227
39
61
1,108
523
501
=.003
Age < 1 yr > 1 yr Total Stage
Table 5 shows the results of log-rank tests in which thoracic site was compared with each biological variable to determine which are independent determinants of survival for patients with neuroblastoma. Thoracic site was associated with a significantly greater chance of survival, regardless of DNA index. Thoracic patients with a DNA index of greater than i had the best long-term survival rate (Fig 2), whereas the nonthoracic patients with a DNA index of 1 had the lowest survival rate. The stratified log-rank test showed that thoracic site is independent of DNA index (P < .0001). Thoracic patients without N-myc amplification (Fig 3) have a better outcome than nonthoracic patients without amplification (P = .002). However, there is no significant difference between thoracic and nonthoracic patients when N-rnycampli-
fication is present (P = .40). The overall results of the stratified log-rank test show the independence of these two variables (P = .003). Thoracic patients with an LDH level below 1,500 have a significantly better survival rate than that of their nonthoracic counterparts (P < .0001). For patients with LDH levels above 1,500, site is not significantly associated with a better survival rate, although the number of thoracic patients in this group is small. The stratified log-rank test shows that, overall, thoracic site and LDH levels are independent factors with respect to survival among neuroblastoma patients (P < .0001). Thoracic patients with a low serum ferritin level had a better survival rate than their nonthoracic counterparts (P = .03), but there was no significant difference with respect to high ferritin levels (P = .68). This must be
Table 5. Biological Variables and the Effect of Thoracic Site on Survival Biological Variable
Nonthoracic
Thoracic n
Deaths
Expected
n
Deaths
Expected
P Value
26 88
9 12
20 25
202 338
120 108
110 91
=.009 <.0001
114
21
48
540
228
201
<.0001
N-myc Nonamplified Amplified
79 5
10 2
23 4
318 91
84 71
71 70
=.002 =.40
Total
84
12
26
409
155
141
=.003
LDH < 1,500
182
30
73
> 1,500
9
4
8
712 246
269 191
227 187
<.0001 =.125
191
34
81
958
460
413
<.0001
Ferritin Low High
19 3
6 3
14 2
190 39
107 32
99 32
=.03 =.68
Total
22
9
16
229
139
132
=.067
DNA index =1 >1 Total
Total
300
100
MORRIS ET AL
""
80
(73) ...... ~'"-,..,_ 46) (24) ....................... 7) .......................................................
"~.
E
(99)
(43)
........
T~:~. 0 •
% ............
o~.r. ul>I
"\
d'4i.............................................. (24)
0
1
2
3
4
(19)
5
i.....oa~ o1=1
6
7
8
9
10
11
12
13
Years Followed
Fig 2. Survival curves for thoracic (114) and nonthoracic (540) patients for whom the DNA index was recorded (stratified log-rank test; P < .0001). Numbers above the curves represent cases with fonow-up until or beyond the specified time-points,
interpreted with caution because there were only three thoracic patients with elevated ferritin levels, all of whom died. DISCUSSION
Our series is similar to most other populations of neuroblastoma patients in that 17% of our patients had a thoracic primary site, and 37% were under f year of age at the time of presentation. Sixty-nine percent of our patients (923) presented with stage C or D disease. For all patients, primary site as well as stage and age at time of presentation had been recorded. However, not all patients had determinations of DNA index (measured in 654), LDH levels (1149), N-myc amplification (492), or serum ferritin levels (251). It was not until 1984 and 1985 that Brodeur and Seeger first showed that N-myc gene amplification predicted rapid disease progression and poor outcome, and only shortly beforehand was DNA index begun to be recorded on a regular basis. 1-4 Because our patient population was drawn from all modern treatment protocols (1981 to 1991), biological data for the early patients was erratic and incomplete. The outcome for patients with thoracic neuroblastoma is better than that of their nonthoracic counterparts. Some have suggested that this is because thoracic and mediastinal tumors become symptom100
,
(41)
(20)
~
~
a¢ N , on
80
'~
.-..,.J~!~................................. ~32)....................................................................
>
~4o
23
]
0
1
~(;~ )
2
(1)
3
4
Th0~ ~p
5
6
7
8
9
10
Years Followed
Fig 3. Survival curves for thoracic (84) and nonthoracic (409) patients for whom N-myc amplification was recorded (stratified log-rank test; P = ,003). Numbers above the curves represent cases with follow-up until or beyond the specified time-points.
atic earlier in the disease process, 5,92 but this has recently been shown by Adams et al and others to not be true. 7,13,2° Although Adams et al showed that age at time presentation and the stage distribution were different for thoracic neuroblastoma, they also showed that thoracic site was an independent favorable risk factor, even when corrected for age and stage at time of presentation, and suggested that the basic biology of the tumors at different sites is different. Complete clinical data were available for each patient in the present study. The age at time of presentation for thoracic patients was significantly lower than that for nonthoracic patients; 51% of thoracic patients were under 1 year of age, compared with 34% of nonthoracic patients. These findings are similar to our previously published results for a smaller cohort of patients. 2° Stage at time of presentation is known to be an important prognostic variable for children with neuroblastoma. In this series, 135 of the 227 thoracic patients (59%) had stage A or B, whereas 194 of the 1,108 nonthoracic patients (18%) had stage A or B. The fact that so many of thoracic patients presented at a lower stage is significant, and likely plays a large role in their better survival rate. All children with stage A, B, or DS disease have a better prognosis than do patients with stage D disease, regardless of site. The only stage in which thoracic site is associated with a significantly higher chance of survival is stage C; this is because thoracic stage C cases behave like thoracic stage A and B cases. Despite the advances in our understanding of neuroblastoma, stage D patients (57% of all nonthoracic patients in our study) continue to have a poor survival rate, regardless of site of presentation. We have shown that the biology of thoracic neuroblastoma is indeed different from that of neuroblastoma located in nonthoracic sites, but these factors might not fully explain the favorable prognosis associated with thoracic site. A DNA index of greater than 1 is known to be associated with improved survival rates, and this was true for our total patient population. However, thoracic site was an independent predictor of survival, irrespective of DNA index. Six hundred fifty-four patients had data for this variable, and statistically significant differences were noted. N-myc amplification is known to portend a poor prognosis for patients with neuroblastoma. In our series, patients with N-myc amplification had a significantly lower survival rate than did those without amplification. The majority (78%) of nonthoracic patients did not have amplification, and almost all (78 of 83; 94%) the thoracic patients did not. Only five thoracic patients had N-myc amplification, two of
BIOLOGICAL VARIABLES IN THORACIC NEUROBLASTOMA
301
whom died in the follow-up period. Data regarding this group must be interpreted cautiously because there were so few thoracic patients with N-myc amplification. For patients with LDH levels of less than 1,500, thoracic site is associated with a greater chance of survival. However, only nine thoracic patients had LDH levels above 1,500, making survival comparisons difficult among this small group. Despite the fact that only a few patients had high values, there were enough patients overall (in the stratified log-rank test) to show that LDH and thoracic site are independent predictors of survival. Some investigators have suggested that serum ferritin level is an important predictor of outcome. 23 Both thoracic and nonthoracic patients had low serum ferritin levels, as determined by age and level stratification. Elevated serum ferritin portended a poor prognosis, regardless of site (83% overall mortality rate). However, only three thoracic patients had elevated serum ferritin levels (all died), making any statistical comparisons difficult between the elevated thoracic and nonthoracic patients. This study confirms our previous observation, from a smaller cohort of patients, that thoracic site neuroblastoma has a better prognosis than nonthoracic neuroblastoma. Previously we showed that the improved survival rate is not attributable to the earlier presentation or expression of symptoms, and that it is independent of age and stage at time of presentation. We now suggest that although thoracic patients tend to have a favorable biological profile (LDH level <1,500, low serum ferritin level, DNA index = 1, and N-myc nonamplification), the studied variables do not fully explain the improved survival rate for our patients. However, many patients in this series had only one or two biological variables recorded. It is also clear that some trends that emerge make an equal distribution of variables between
thoracic and nonthoracic patients unlikely. Very rarely did thoracic patients have N-rnyc amplification (5 patients), serum LDH levels above 1,500 (9 patients), or high serum ferritin levels (3 patients). There is growing evidence that in neuroblastoma, genetic and molecular differences exist and that survival likely will be related to inherent qualities of the tumor. Although N-myc amplification and DNA index were accepted in the last decade as being useful biological markers that helped predict survival, there are new genetic markers that may be more sensitive and useful in the diagnosis and treatment of neuroblastoma. Tanaka et a128and Chan et a129have studied the expression of the H-ras oncogene product p21, and Nakagawara et aP ° have suggested that high levels of expression of the proto-oncogene TRK is strongly predictive of a favorable outcome in neuroblastoma. Cooper et al have presented evidence 3l that neuroblastoma cells represent arrest at various levels of adrenal medullary cell differentiation and that a process of differentiation/dedifferentiation may be responsible for the seemingly random switch in some tumors from malignant to benign phenotype. The expression of TRK, the function of the nerve growth-factor receptor, and the expression of the H-ras oncogene product p21 are but three examples of new markers that might prove useful in the future for identifying high-risk neuroblastoma patients. We have clearly shown that in a large group of patients for whom survival tends to be favorable (thoracic site) and for whom biological variable data are incomplete, the currently measured variables do not fully explain the better survival rate among these patients. Future studies of neuroblastoma should include a thorough and complete assessment of the effect of biological variables on survival and also should be directed at evaluating the molecular genetics of this group, in an attempt to identify the predictive and therapeutic variables that portend an improved survival.
REFERENCES 1. Brodeur GM, Seeger RC, Schwab M, et al: Amplification of
N-myc in untreated human neuroblastomas correlates with advanced stage of disease. Science 224:1121-1124, 1984 2. Seeger RC, Brodeur GM, Sather H, et al: Association of multiple copies of the N-myc oncogene with rapid progression of neuroblastomas. N Engl J Med 313:1111-1116, 1985 3. Look AT, Hayes FA, Nitschke R, et al: Cellular DNA content as predictor of response to chemotherapy in infants with unresectable neuroblastoma. N Engl J Med 311:231-235, 1984 4. Look AT, Hayes A, Shuster J J, et al: Clinical relevance of tumor cell ploidy and N-myc gene amplification in childhood neuroblastoma: A Pediatric Oncology Group Study. J Clin Oncol 9:581-591, 1991 5. Brodeur GM, Azar C, Brother M, et al: Neuroblastoma:
Effect of genetic factors on prognosis and treatment. Cancer 70:1685-I694, 1992 6. Silber Evans AE, Fridman M: Models to predict outcome from childhood neuroblastoma: The rote of serum ferritin and tumor histology. Cancer Res 51:1426-1433, 1991 7. Shuster JJ, McWilliams NB, Castleberry R, et al: Serum lactate dehydrogenase in childhood neuroblastoma: A Pediatric Oncology Group Recursive Partitioning Study. J Clin Oncol 15:295-303, 1992 8. Shochat SJ, Abt AB, Schengrund CL: VCN-releasable sialic acid and gangliosides in human neuroblastomas. J Pediatr Surg 12:413-418, t977 9. Cohen PS, Cooper M J, Helman LJ, et al: Neuropeptide Y
302
MORRIS ET AL
expression in the developing adrenal gland and in childhood neuroblastoma tumors. Cancer Res 50:6055-6061, 1990 (suppl) 10. Hsiao RJ, Seeger RC, Yu AL, et al: Chromogranin A in children with neuroblastoma: Serum concentration parallels disease stage and predicts survival. J Clin Invest 85:1555-1559, 1990 11. Catalono PW, Newton WA, Williams TE, et al: Reasonable surgery for thoracic neuroblastomas in infants and children. J Thorac Cardiovasc Surg 76:459-464, 1978 12. Jaffe N: Neuroblastoma: Review of the literature and examination of the factors contributing to its enigmatic character. Cancer Treat Rev 3:61-82, 1976 13. Filler RM, Traggis DG, Jaffe N: Favorable outlook for children with mediastinal neuroblastoma. J Pediatr Surg 7:136-143, 1972 14. Kajanti M: Neuroblastoma in 88 children: Clinical features, prognostic factors, results, and late effects of chemotherapy. Ann Clin Res 15:1-68, 1983 (supp139) 15. Pelton JJ, Rattner 1A: Neuroblastoma of the thoracic inlet. J Pediatr Surg 25:547-549, 1990 16. Koop CE: The role of surgery in resectable, nonresectable, and metastatic neuroblastomas. JAMA 205:157-158, 1968 17. deLorimier AA, Bragg KU, Linden G: Neuroblastoma in childhood. Am J Dis Child 118:144-150, 1969 18. Sandstedt B, Jereb B, Eklund G: Prognostic factors in neuroblastoma. Acta Pathol Microbiol Immunol Scand 91:365-371, 1983 19. Swank RL, Fenerman GH, Sieber WK, et al: Prognostic factors in neuroblastoma. Ann Surg 174:428-435, 1971 20. Adams GA, Shochat SJ, Smith EI, et al: Thoracic neuroblastoma: A Pediatric Oncology Group Study. J Pediatr Surg 28:372378, 1993 21. Shochat SJ, Corbelletta NL, Repman MA, et al: A biochemi-
cal analysis of thoracic neuroblastomas: A Pediatric Oncology Group Study. J Pediatr Surg 22:660-664, 1987 22. Saarinen UM, Slimes MA: Serum ferritin in assessment of iron nutrition in healthy infants. Acta Paediatr Scand 67:745-751, 1978 23. Hann HWL, Levy HM, Evans AE: Serum ferritin as a guide to therapy in neuroblastoma. Cancer Res 40:1411-1413, 1980 24. SAS Procedure Guide, Version 6 (ed 3). Cary, NC, SAS Institute Inc, 1990, pp 325-363 (chap 20) 25. Peto R, Peto J: Asymmetricallyefficient rank invariant test procedures. J R Stat Soc 135:185-198, 1972 26. Kaplan EL, Meier R: Non-parametric estimation from incomplete observation. J Am Stat Assoc 53:457-481, 1958 27. Peto R, Pike MC, Armitage P, et al: Design and analysis of randomized clinical trials requiring prolonged observation of each patient: lI analysis and examples. Br J Cancer 35:1-39, 1977 28. Tanaka T, Slamon DJ, Shimoda H, et al: Expression of H-ras oncogene products in human neuroblastomas and the significant correlation with a patients prognosis. Cancer Res 48:1030-1034, 1988 29. Chan HSL, Haddad G, Thorner PS, et al: P-glycoprotein expression as a predictor of the outcome of therapy for neuroblastoma. N Engl J Med 325:1608-14, 1991 30. Nakagawara A, Arima-Nakagawara M, Scavarda NJ, et al: Association between high levels of expression of the TRK gene and favorable outcome in human neuroblastoma. N Engl J Med 328:847-854, 1993 31. Cooper MJ, Steinberg SM, Chatten J, et al: Plasticity of neuroblastoma tumor cells to differentiate along a fetal adrenal ganglionic lineage predicts for improved patient survival. J Clin Invest 90:2402-2408, 1992
Discussion J. Ternberg (St Louis, MO): I t h i n k we all acknowledge that site m a k e s a difference in various tumors. W e have seen the effect of site in r h a b d o m y o s a r c o m a . T h e q u e s t i o n in my m i n d , though, is what m a k e s the difference? It is n o t just the site. T h e r e has to be s o m e t h i n g different a b o u t the t u m o r in that area that m a k e s the difference, a n d I t h i n k this is what is b e i n g e x a m i n e d in this report. It is the stage C p a t i e n t s w h e r e the thoracic site really seems to m a k e a difference with respect to survival. Did they look to see which o n e s had h a d c o m p l e t e resection. D i d that m a k e any difference? T h e o t h e r p o i n t that I did n o t notice in the r e p o r t was some r e f e r e n c e to the t u m o r histology. W e know histology m a k e s a difference in W i l m s ' t u m o r prognosis. W e are b e g i n n i n g to talk a b o u t it now in regard to r h a b d o m y o s a r c o m a . T h e r e is the S h i m a d a classification for n e u r o b l a s t o m a ; have the thoracic cases b e e n e x a m i n e d from this standpoint? I t h i n k that in the future, a n a t o m i c staging n o t be as i m p o r t a n t as it is today. W e are going to be able to p u t all this i n f o r m a t i o n , i n c l u d i n g the m o l e c u l a r biol-
ogy i n f o r m a t i o n , into the c o m p u t e r a n d come out with b e t t e r staging as a basis for t r e a t m e n t stratification, a n d hopefully s o m e t i m e in the f u t u r e we will b e able to refine our t r e a t m e n t . M. Asch (Los Angeles, CA): A l o n g the same lines as D r T e r n b e r g ' s question, was t h e r e any difference b e t w e e n the survival rates for p a t i e n t s whose t u m o r s were able to b e resected d u r i n g the initial o p e r a t i o n a n d those whose t u m o r s r e q u i r e d a second p r o c e d u r e for c o m p l e t e resection ( b e c a u s e they were u n r e s e c t able the first t i m e ) ? J.M. Laberge (Montreal, Quebec): I am n o t very good in statistics, b u t have you d o n e logistic regression analysis to look at all the factors to see w h e t h e r thoracic site is an i n d e p e n d e n t v a r i a b l e ? G.M. Haase (Denver, CO): It a p p e a r s that the biological factors were at least as i m p o r t a n t as the thoracic site, b e c a u s e the u n a m p l i f i e d t u m o r s were associated with a good survival rate, a n d the amplified t u m o r s were i n t e r m e d i a t e with respect to survival. Was there an e v a l u a t i o n within biological groups as o p p o s e d to thoracic groups for resectability?
BIOLOGICAL VARIABLES IN THORACIC NEUROBLASTOMA
Just an observation: There are about 1,000 metastatic neuroblastoma patients in this decade of experience from the Pediatric Oncology Group. There are another 1,000 patients with metastatic disease in the Children's Cancer Group studies. With this combined experience, it seems to me that the two groups could answer the question regarding the importance of resectability for metastatic neuroblastoma. It is something we should get together and accomplish in the next few years. J.A. Morris (response): With respect to Dr Ternberg's first question, yes, there is a difference in stage C patients. Stage C thoracic patients have a significantly higher survival rate than do nonthoracic patients, whether they have complete resection or not. With respect to histology, we were not looking specifically at histology in this study. We believe that a combination of anatomic and clinical presentation, along with identifying the important biological characteristics that can be used to predict survival, will be the key to future staging systems.
303
With respect to Dr Asch's question regarding resection during the initial operation, in a smaller cohort of these patients, previously presented, we showed that regardless of complete or incomplete resection in the thoracic site, survival data were unchanged. Both Dr Laberge and Dr Haase asked about logistic regression analysis. The log-rank test is a form of partition analysis that Dr Cantor, our statistician, uses. It allows us to obtain specific information about smaller groups, and then "collapse" those groups to present data regarding specific variables. The results of a multivariate analysis would be difficult to interpret because not all our patients had all the variables recorded. Within each biological group, Dr Haase, we looked previously at resectability, and the thoracic site did not appear to make a difference. We did not look specifically at the abdominal primary tumors in the other 1,108 nonthoracic patients or at their ability to be resected.