Histopathology Defines Prognostic Subsets of Ganglioneuroblastoma, Nodular

Transcription

1 1150 Histopathology Defines Prognostic Subsets of Ganglioneuroblastoma, Nodular A Report from the Children s Cancer Group Shunsuke Umehara, M.D. 1 Atsuko Nakagawa, M.D. 1 Katherine K. Matthay, M.D. 2 John N. Lukens, M.D. 3 Robert C. Seeger, M.D. 4 Daniel O. Stram, Ph.D. 5 Robert B. Gerbing, M.A. 5 Hiroyuki Shimada, M.D., Ph.D. 1 1 Department of Pathology and Laboratory Medicine, Childrens Hospital Los Angeles, Los Angeles, California. 2 Department of Pediatrics, University of California-San Francisco School of Medicine, San Francisco, California. 3 Pediatric Hematology/Oncology, Vanderbilt University, Nashville, Tennessee. 4 Division of Hematology/Oncology, Childrens Hospital Los Angeles, Los Angeles, California. 5 Children s Cancer Group, Arcadia, California. Supported in part by grant CA from the Division of Cancer Treatment, National Cancer Institute, National Institutes of Health, Department of Health and Human Services. Address for reprints: Hiroyuki Shimada, M.D., Ph.D., Children s Cancer Group, P.O. Box 60012, Arcadia, CA ; fax: ; E- mail: hshimada@chla.usc.edu. Received October 12, 1999; revision received March 24, 2000; accepted April 27, BACKGROUND. Ganglioneuroblastoma, nodular (GNBn) is a rare subtype of the family of neuroblastic tumors (neuroblastoma, ganglioneuroblastoma, and ganglioneuroma) that are classified in the unfavorable histology group according to the International Neuroblastoma Pathology Classification (Shimada system). Tumors of this subtype have been considered to represent a prototypic example of biologically and clinically nonaggressive (Schwannian stroma-rich and stroma-dominant) components combined with biologically and clinically aggressive nodular (Schwannian stroma-poor) components. However, detailed histopathologic analysis as well as thorough prognostic evaluation of patients with this subtype has not been reported. METHODS. Pathology slides and reports from a total of 70 GNBn patients from the Children s Cancer Group (CCG)-3881 and CCG-3891 studies were reviewed. Sixtyeight tumors were classified in the favorable subset (FS) or the unfavorable subset (US) based on the evaluation of nodular components by applying the same histopathologic criteria (patient age, grade of neuroblastic differentiation, mitosiskaryorrhexis index) that are used for neuroblastomas in the International Neuroblastoma Pathology Classification. Patient prognosis as well as clinical and biologic characteristics within the subsets were analyzed, and the results were compared with those from 654 non-gnbn patients who were enrolled in the same CCG studies during the same period. RESULTS. Patients with GNBn tumors, usually diagnosed at age 1 year, had a significantly lower overall 5-year event free survival (EFS) rate than patients with non-gnbn subtypes (44.7% EFS vs. 65.0% EFS; P ). A significant difference in the outcome of the patients between the FS (22 patients; 86.1% EFS; 95.0% survival rate) and the US (46 patients; 29.0% EFS; 40.7% survival rate) of the GNBn subtype (P ) was shown. When the cohort of patients with GNBn tumors was subdivided into FS and US, the outcomes were similar to those of patients with tumors of favorable histology (397 patients; 90.5% EFS; 97.6% survival rate) and with tumors of unfavorable histology (257 patients; 27.0% EFS; 35.7% survival rate) of the non-gnbn type. The patients with US tumors frequently (63.0%) presented with distant metastasis. CONCLUSIONS. The current study demonstrates that the nodular components in GNBn tumors are not always aggressive. The prognosis of these patients can be determined by the analysis of age-linked histopathologic features. Cancer 2000;89: American Cancer Society. KEYWORDS: neuroblastic tumors, ganglioneuroblastoma, nodular, composite, histopathology, International Classification, Shimada system, favorable subset, unfavorable subset, prognosis American Cancer Society

2 Prognostic Subsets of Ganglioneuroblastoma, Nodular/Umehara et al Ganglioneuroblastoma, nodular (GNBn) is a rare subtype of neuroblastic tumors (NTs) (including neuroblastoma, ganglioneuroblastoma, and ganglioneuroma). Patients with GNBn have a high potential for distant metastasis and have a poor clinical outcome. Tumors in this subtype are comprised of different clones characterized by macroscopically visible and often hemorrhagic neuroblastomatous (Schwannian stroma-poor) nodule(s) coexisting with nonaggressive component(s) of either ganglioneuroblastoma, intermixed (Schwannian stroma-rich) or ganglioneuroma (Schwannian stroma-dominant), maturing or mature morphology. 1 3 In 1947, Stout 4 first described a tumor containing fully malignant neuroblastomatous nodules coexisting with ganglioneuromatous tissue; he introduced the idea of a composite tumor for this subtype of NTs and suggested its aggressive clinical behavior. Adam and Hochholzer, in 1981, 5 also reported a poor prognosis for patients with this subtype of ganglioneuroblastoma. In 1984, Shimada et al. 1 defined morphologic and clinical characteristics of this subtype, named it stroma-rich (ganglioneuroblastoma), nodular, and separated it from other subtypes of NTs. In 1999, the International Neuroblastoma Pathology Committee (INPC) proposed a classification of NTs 6,7 by adopting the original Shimada system. 1 In their report, this subtype of NTs was renamed ganglioneuroblastoma, nodular (composite, Schwannian stroma-rich/stroma-dominant and stroma-poor) with a similar but more precisely defined morphologic description. 6 Although tumors of this subtype are classified into an unfavorable histology group by the Shimada system, the INPC suggested further histopathologic evaluation and analysis to more fully define the biologic and clinical characteristics of GNBns. This study, by analyzing the largest series of patients with GNBn yet, was designed to characterize clinical and pathologic features and to evaluate the prognostic significance of histopathologic findings of the tumors in this subtype. MATERIALS AND METHODS In total, 911 NT cases were registered in CCG studies 3881 and 3891 from August 1, 1991, to August 1, 1995 (both studies were open during this period, 8 encompassing neuroblastoma patients with all clinical stages). Among these 911 patients, pathology slides from 724 tumors that were biopsied or surgically removed prior to chemotherapy and/or irradiation therapy were reviewed centrally. There were 70 tumors (9.7%) of the GNBn subtype and 654 tumors (90.3%) of other subtypes (non-gnbn: neuroblastoma [Schwannian stromapoor]; ganglioneuroblastoma, intermixed [Schwannian TABLE 1 Ganglioneuroblastoma, Nodular versus Other Subtypes of Neuroblastic Tumors Characteristic Ganglioneuroblastoma, nodular (%) Other subtypes (%) P value Age (yrs) (14.3) 285 (43.6) (65.7) 319 (48.8) 5 14 (20) 50 (7.6) Clinical stage I 6 (8.6) 82 (12.5) II 16 (22.9) 125 (19.1) III 12 (17.1) 149 (22.8) IV 36 (51.4) 244 (37.3) IVs 0 (0) 54 (8.3) Primary site Head and neck 0 (0) 26 (4.0) Chest 20 (28.6) 125 (19.1) Thoracoabdominal 1 (1.4) 23 (3.5) Adrenal 28 (40.0) 264 (40.4) Abdomen 14 (20.0) 111 (17.0) Pelvis 6 (8.6) 25 (3.8) Others 1 (1.4) 24 (3.7) Unknown 0 (0) 56 (8.6) MYCN status Amplified 2 (2.9) 106 (16.2) Nonamplified 54 (77.1) 446 (68.2) Not tested 14 (20.0) 102 (15.6) CCG study (42.9) 409 (62.5) (57.1) 245 (37.5) Total 70 (100) 654 (100) Event free survival (5 yrs) 44.7% 65.0% CCG: Children s Cancer Group. stroma-rich]; and ganglioneuroma [Schwannian stroma-dominant], maturing and mature). Clinical and biologic data (patient age distribution, Evan clinical stage, 9 primary tumor site, MYCN status by Southern blot analysis, study number, and clinical outcome) of the GNBn and non-gnbn patients are summarized in Table 1. Treatment protocols for patients who were included in the CCG studies (CCG-3881 and CCG-3891) depended on risk group, which was defined by disease stage, patient age, and tumor biology. 8,10 12 Low risk patients (Stages I, II, and IVs disease) generally underwent surgery or biopsy alone. Intermediate risk patients (biologically favorable Stage III children and Stage IV infants) received a course of moderate therapy according to the CCG-3881 protocol, and high risk patients (biologically unfavorable Stage III children and Stage IV infants or Stage IV at age 1 year) received intensive chemotherapy and possibly underwent autologous bone marrow transplantation according to the CCG-3891 protocol. Appropriate in-

3 1152 CANCER September 1, 2000 / Volume 89 / Number 5 FIGURE 1. Nodular formation in ganglioneuroblastoma, nodular subtype. There were 23 single nodular tumors and 33 multiple nodular tumors. Six tumors were considered as variant: Four were comprised predominantly of stroma-poor component (SP) without a macroscopically recognizable stroma-rich/ stroma-dominant component (SR/SD), and two had ganglioneuroma (Schwannian stroma-dominant) histology in the primary site and neuroblastoma (Schwannian stroma-poor) histology in metastatic lesion to the bone. The numbers of nodules were not determined in eight tumors. formed consent procedures were followed, and consent was obtained from parents or guardians. Pathology Review The files of the 70 GNBn tumors in the repository at the Department of Pathology and Laboratory Medicine, Childrens Hospital Los Angeles, Los Angeles, California, had varying numbers of hematoxylin and eosin stained sections (1 24 sections; average, 7 sections per case). Pathology review was conducted blindly by two pathologists (S.U. and H.S.) without knowledge of clinical information (patient age, disease stage, and outcome) of the cases. All cases were accompanied by surgical pathology reports from the contributing institutions. In this study, the following pathology features were recorded for each case: 1) the number (and size, if available) of macroscopically visible nodule(s) of stroma-poor neuroblastomatous component(s) based on both the description of the surgical pathology report from the contributing institutions and careful review of the pathology slides; 2) the histologic findings of the nodular component, including grade of neuroblastic differentiation (undifferentiated, poorly differentiated, or differentiating), and mitosis karyorrhexis-index (MKI; a low MKI of 2% or 100 per 5000 cells, an intermediate MKI of 2 4% or per 5000 cells, or a high MKI of 4% or 200 per 5000 cells); 3) the number of different histologies that made up the nodular component of each tumor; and 4) the histologic appearance of the stroma-rich/stromadominant component. Two different nodular histologies, a favorable nodular histology (FN) and an unfavorable nodular histology (UN), were distinguished in the GNBn tumors by the pathology review. To determine nodular histology, the criteria of age-linked histopathologic evaluation for neuroblastoma (Schwannian stromapoor) tumors used in the International Neuroblastoma Pathology Classification (Shimada system) 7 were applied to the stroma-poor components of the GNBn tumors. Histopathologic features for FN included 1) poorly differentiated or differentiating neuroblastoma morphology with low or intermediate MKI diagnosed at age 1.5 years and 2) differentiating neuroblastoma morphology with low MKI diagnosed from ages years. Histopathologic features for UN included 1) any neuroblastoma morphology with high MKI at any age, 2) any neuroblastoma morphology with intermediate MKI at age 1.5 years, 3) undifferentiated neuroblastoma morphology at any age, 4) poorly differentiated neuroblastoma morphology diagnosed at age 1.5 years, and 5) any neuroblastoma morphology diagnosed at age 5 years. After the determination of nodular histology, the GNBn tumors were classified into two prognostic sub-

4 Prognostic Subsets of Ganglioneuroblastoma, Nodular/Umehara et al sets. The favorable subset (FS) of GNBn included tumors that contained single or multiple FNs. The unfavorable subset (US) included tumors with single unfavorable nodules (UNs) or multiple nodules containing at least one UN. Statistical Analysis Using the two-sided Pearson chi-square test or a twosample t test, analyses for histopathologic, clinical, and biologic characteristics of the tumors in the GNBn subtype and the non-gnbn subtypes were performed. The prognoses of the subsets of patients with GNBn tumors (FS vs. US) were evaluated, and their estimated EFS and survival experience was calculated using the Kaplan Meier method. 13 The EFS and survival curves for these two subsets of patients also were compared with those curves of patients with non-gnbn tumors (favorable histology [FH] vs. unfavorable histology [UH] according to the International Neuroblastoma Pathology Classification 7 ) generated from the data bases at the CCG Operations Center (Arcadia, CA). RESULTS Clinical and Biologic Characteristics Table 1 shows that the tumors of GNBn subtype were diagnosed in children who were significantly older than those who displayed non-gnbn tumor subtypes: 85.7% of the GNBn tumors were diagnosed later than age 1 year (P ), and 20.0% were diagnosed after age 5 years (P ). No significant differences were observed in tumor distribution for primary tumor sites or for clinical stages (nonadvanced [Stages I, II, and IVs] vs. advanced [Stages III and IV]) between the GNBn subtype and the non-gnbn subtypes. Of the patients with GNBn subtype tumors, 68.6% presented at advanced stages, and 60.1% of the patients with non-gnbn subtype tumors presented at advanced stages. However, in those with GNBn subtype tumors, 51.4% (36 of 70) patients were presented with Stage IV disease, and no tumors were diagnosed in patients with Stage IVs disease. MYCN amplification was seen in only 2 of 56 patients (3.6%) with the GNBn subtype of tumor, whereas 106 of 552 patients (19.2%) with tumors of the non-gnbn subtypes had amplified MYCN (P 0.005). Among the patients with GNBn tumors, 30 patients were in the CCG-3881 study, and 40 patients were in the CCG-3891 study. Among the patients with non-gnbn subtype tumors, 409 patients were in the CCG-3881 study, and 245 patients were in the CCG-3891 study. The expected 5-year EFS rate for the patients with GNBn subtype tumors (44.7%) was significantly lower than that for patients with non- GNBn subtype tumors (65.0%; P ). Pathology Review of the Tumors of the GNBn Subtype Macroscopic appearance By macroscopic examination (Figs. 1, 2), it was relatively easy to distinguish the stroma-rich/stromadominant component (solid, tan-yellow) from the stroma-poor nodule(s) (usually hemorrhagic and often necrotic with or without calcification). Macroscopically visible nodular formation(s) were confirmed by the institutional pathologists and after the central review in 64 (91.4%) cases. Six cases (8.6%) were considered as variant forms of GNBn. Four (5.7%) of those tumors were comprised predominantly of a huge nodule or nodules with a stroma-poor component and without any stroma-rich/stroma-dominant portion that was recognizable clearly by macroscopic examination. The diagnoses of GNBn subtype for those four variant cases were made after microscopic examination by identifying thin stroma-rich/stroma-dominant component(s) either in the peripheral portion or in the septal area of the primary tumor tissue. In the remaining two (2.9%) variant cases, nodular formation was not detected in pathology samples from primary tumor sites (one in the adrenal and the other in a retroperitoneal location): They both showed histology of the ganglioneuroma (Schwannian stroma-dominant), maturing subtype and had metastatic lesions of stroma-poor neuroblastomatous component to the bone. Among the 64 patients with macroscopically identifiable nodules in primary tumors, 23 had a single nodule (single nodular tumors), and 33 had two or more nodules (multiple nodular tumors). The number of nodules was not determined in eight tumors. In 15 tumors, precise measurement of the individual nodules (varying from 0.3 cm to 10.5 cm in the greatest dimension) was described in the surgical pathology reports from the contributing institutions. In other cases, the nodules were described as large or discrete in the reports. Histologic Evaluation Individual nodules of the 23 single nodular tumors were comprised of either a single histology (22 patients) (Fig. 3a c) or two different histologies (1 patient) (Fig. 3d) of the stroma-poor component. The nodular lesions in the 33 multiple nodular tumors were also comprised of either a single histology (27 patients) (Fig. 3e) or two different histologies (6 patients) (Fig. 3f). Nodular lesions in 12 of the patients in whom the number of nodules was not determined (including four tumors that were comprised predominantly of a stroma-poor component without a macroscopically recognizable stroma-rich/stroma-domi-

5 1154 CANCER September 1, 2000 / Volume 89 / Number 5 FIGURE 2. Ganglioneuroblastoma, nodular subtype. Note discrete nodule(s) in a single nodular tumor (a) (original magnification, 9) and a multiple nodular tumor (b) (original magnification, 6). Some nodules show infiltrative growth of neuroblastic cells into the Schwannian stroma-rich/stroma-dominant portion: An example is shown in c (original magnification, 40). nant component) showed a single histology (11 patients) or double histology (1 patient). For the two patients in whom a stroma-poor component was found only at metastatic sites (bone), further histologic features were not determined because of limited amounts of tissue for evaluation. These two patients, therefore, were excluded from further analysis. The histologic features (grade of neuroblastic differentiation and MKI) of the stroma-poor components for the 68 fully analyzed cases are summarized in Table 2. With regard to the Schwannian stroma-rich/stromadominant component, 46 tumors (67.6%) showed ganglioneuroblastoma, intermixed morphology, whereas 22 tumors (32.4%) displayed ganglioneuroma, maturing (19 patients) or mature (3 patients) morphology (see Table 3). Prognostic Subsets in GNBn Sixty-eight of 70 GNBn tumors were classified into FS or US based on the age-linked histologic evaluation of their stroma-poor component(s). The nodules that were made up of a single histology were classified into either FS (21 of 60 patients; 35.0%) or US (39 of 60

6 Prognostic Subsets of Ganglioneuroblastoma, Nodular/Umehara et al FIGURE 2. (continued) patients; 65.0%) because of the presence of FN or UN, respectively. Histologically different nodular components that were found in the same primary tumor tissue were evaluated separately: A tumor with FN/FN (one patient) was classified as FS, and tumors with FN/UN (5 patients) and UN/UN (2 patients) were classified as US (Fig. 4). There was a total of 22 tumors (32.4%) in the FS group and 46 tumors (67.7%) in the US group (see Table 2). The characteristics of these subsets are summarized in Table 3. It was noted that children with US tumors were significantly older than those with FS tumors (97.8% vs. 59.1% diagnosed after age 1 year; P ; 30.4% vs. 0% diagnosed after age 5 years; P ). FS tumors were distributed evenly in various clinical stages, whereas patients with US tumors frequently (63.0%; 29 of 46 tumors) presented with distant metastases (Stage IV) (P ). No significant difference was found in primary distribution sites between the subsets. Figure 5 shows the EFS and survival curves for the patients in the FS and US groups with the GNBn subtype of tumor. The 5-year EFS and survival rates differed significantly between these two prognostic subsets of patients: 86.1% and 95% for patients in the FS group and 29.0% and 40.7% for patients in the US group, respectively (P ). Also shown in Figure 5 are EFS and survival curves for the patients in the FH group (397 patients; 90.5% 5-year EFS rate and 97.6% 5-year survival rate) and patients in the UH group (257 patients; 27.0% 5-year EFS and survival rate and 35.7% 5-year survival rate) with tumors of the non-gnbn subtypes for comparison. The EFS curves for patients in the FS and FH groups were almost identical, as were those for patients in the US and UH groups. Further analyses showed that the number of nodules (single vs. multiple) and the nodular histology (single vs. double) of stroma-poor component had no significant prognostic effects. In this series, there were 30 patients (18 patients in the FS group and 12 patients in the US group) in the CCG-3881 study and 38 patients (4 patients in the FS and 34 patients in the US) in the CCG-3891 study. To further analyze the prognostic effects of the FS and US in the patients with GNBn, two clinical classes, i.e., a high risk class and a low and intermediate risk class, were distinguished based on the data of disease stage, patient age, and MYCN status (56 patients were available for the analysis, and 12 patients were excluded due to missing MYCN data). The high risk class included the patients with Stage IV disease who were age 1 year at diagnosis. The rest of the patients were classified into the low and intermediate risk class. The high risk class was comprised predominantly (25 of 27 patients; 92.6%) of the patients in the US, and these patients generally had very low 5-year EFS (13.3%) and 5-year survival (22.7%) rates. Only two patients from the FS were found in the high risk class: One patient died 2 years after the diagnosis, and the other patient is alive 3 years after the diagnosis. Among the patients in the low and intermediate risk class, there were significant differences in both the 5-year EFS rate (92.9% vs. 57.1%; P ) and the 5-year survival rate (100% vs. 62.9%; P ) be-

7 1156 CANCER September 1, 2000 / Volume 89 / Number 5 FIGURE 3. Various morphologies of the stroma-poor component in ganglioneuroblastoma, nodular subtype: Single nodular component tumors usually are comprised of a single histology, such as undifferentiated neuroblastoma with low mitosis-karyorrhexis index [MKI] (a) (original magnification, 200), poorly differentiated neuroblastoma with low MKI (b) (original magnification 100), or differentiating neuroblastoma with low MKI (c) (original magnification, 100). Also seen are tumors with a single nodular component that is comprised of two different histologies, such as differentiating neuroblastoma with low MKI (D) and poorly differentiated neuroblastoma with intermediate MKI (P) (d) (original magnification, 100). There are tumors with multiple nodules that are comprised of either a single histology, an example of which is shown in e (poorly differentiated neuroblastoma with low MKI) (original magnification, 100), or double histologies, an example of which is shown in f (differentiating neuroblastoma with low MKI [D] and poorly differentiated neuroblastoma with intermediate MKI [P]) (original magnification, 200). tween patients in the FS (15 patients) and the US (14 patients). With regard to the stroma-rich/stroma-dominant component, tumors with ganglioneuroma (Schwannian stroma-dominant) morphology were found in significantly older children (age years) than those with ganglioneuroblastoma, intermixed (Schwannian stroma-rich) morphology (age years) (P 0.005; two-sample t test). It is noteworthy that tumors with ganglioneuroma morphology almost always had the UN component (20 of 22 tumors; 90.9%), whereas slightly more than half of the tumors with ganglioneuroblastoma, intermixed (56.5%; 26 of 46 tumors) had the UN component (P ). DISCUSSION The GNBn subtype of neuroblastic disease, which is found rarely in infants and usually is diagnosed in older children age 1 year, causes a significantly lower overall EFS rate for the patients than non-gnbn subtypes of NTs. The literature suggests that this subtype represents a prototypic example of a composite tumor made up of biologically and clinically nonaggressive clone(s) i.e., ganglioneuroblastoma, inter-

8 Prognostic Subsets of Ganglioneuroblastoma, Nodular/Umehara et al FIGURE 3. (continued) mixed (Schwannian stroma-rich) or ganglioneuroma (Schwannian stroma-dominant) component(s) and biologically and clinically aggressive clone(s) i.e., neuroblastomatous (Schwannian stroma-poor) component(s). 1 However, the current study clearly demonstrates that the neuroblastomatous components in this subtype are not always aggressive and that the prognosis of patients can be determined by precise evaluation of histology. The same age-linked criteria used for neuroblastoma (Schwannian stroma-poor) tumors are applicable to the nodular component for distinguishing FN and UN. Two distinct prognostic subsets, FS and US, have been identified in the GNBn subtype. Tumors in the FS are composed of 1) a Schwannian stroma-rich/stroma-dominant component and 2) the FN component. These tumors are distributed evenly in all stages and show a significantly better EFS and survival rate than the US that are comparable to those of the FH group in non-gnbn subtypes. Conversely, tumors in the US containing UN frequently present with distant metastasis and have a signifi-

9 1158 CANCER September 1, 2000 / Volume 89 / Number 5 FIGURE 3. (continued) cantly lower EFS and survival rate that are comparable to those of the UH group in non-gnbn subtypes. Notably, greater than 2 of 3 of the tumors in the GNBn subtype fall into the US. It is possible that the FN component, although it is morphologically less mature than the stroma-rich/ stroma-dominant component in the same tumor tissue, may have the potential for differentiation and, thus, may belong in the age-matched maturational sequence of biologically favorable NTs defined by the Shimada system. 1,7 These FNs may proceed along their course of maturation and take on a stroma-rich/ stroma-dominant histology, eventually losing the characteristic feature of macroscopically visible, nodular formation of composite tumors in older patients. According to the Shimada system, histologic changes from stroma-poor to stroma-rich/stroma-dominant morphology of these FNs are expected to take place in patients before age 5 years. 1,7 By contrast, UN in the GNBn subtype would remain histopathologically unfavorable, regardless of patient age. This may be one of the reasons why more FS tumors than US tumors

10 Prognostic Subsets of Ganglioneuroblastoma, Nodular/Umehara et al TABLE 2 Histologic Features of the Nodular Lesion in Ganglioneuroblastoma, Nodular Tumor type Single nodular tumors (n 23) and predominantly neuroblastomatous tumors (N 4) Single histology (n 25) Poorly Diff. and low MKI (FS, 3; US, 10) 13 Poorly Diff. and intermediate MKI (FS, 1; US, 2) 3 Poorly Diff. and high MKI (US, 2) 2 Diff. and low MKI (FS, 7) 7 Double histologies (n 2) Poorly Diff. and low MKI (FS); Diff. and low MKI (FS) 1 Poorly Diff. and intermediate MKI (US); diff. and low MKI (FS) 1 Multiple nodular tumors (n 33) Single histology (n 27) Undiff. and low MKI (US, 1) 1 Poorly Diff. and low MKI (FS, 3; US, 12) 15 Poorly Diff. and intermediate MKI (US, 3) 3 Poorly Diff. and high MKI (US, 2) 2 Diff. and low MKI (FS, 6) 6 Double histologies (n 6) Poorly Diff. and low MKI (FS); Poorly Diff. and high MKI (US) 1 Poorly Diff. and low MKI (US); Poorly Diff. and high MKI (US) 1 Diff. and low MKI (FS); Poorly Diff. and low MKI (US) 2 Diff. and low MKI (FS); Poorly Diff. and intermediate MKI (US) 1 Diff. and low MKI (US); Poorly Diff. and low MKI (US) 1 No. of nodules, unknown (n 8) Single histology (n 8) Undiff. and low MKI (US, 1) 1 Poorly Diff. and low MKI (US, 4) 4 Poorly Diff. and intermediate MKI (US, 2) 2 Diff. and low MKI (FS, 1) 1 No. of cases Undiff.: undifferentiated; Poorly Diff.: poorly differentiated; Diff.: differentiating; MKI: mitosis-karyorrhexis index; FS: favorable subset; US: unfavorable subset (see text). TABLE 3 Prognostic Subsets of Ganglioneuroblastoma, Nodular a Feature Favorable subset (%) Unfavorable subset (%) P value b CCG study (81.8) 12 (26.1) (18.2) 34 (73.9) Survival 95.0% 40.7% Event free survival (5 yrs) 86.1% 29.0% Age (yrs) (40.9) 1 (2.2) (59.1) 31 (67.4) 5 0 (0) 14 (30.4) Clinical stage I 4 (18.2) 2 (4.3) II 7 (31.8) 9 (19.6) III 6 (27.3) 6 (13.0) IV 5 (22.7) 29 (63.0) Primary site Chest 9 (40.9) 11 (23.9) Thoracoabdominal 0 (0) 1 (2.2) Adrenal 7 (31.8) 20 (43.5) Abdomen 4 (18.2) 9 (19.6) Pelvis 2 (9.1) 4 (8.7) Others 0 (0) 1 (2.2) Schwannian stroma-rich/ stroma-dominant histology Ganglioneuroblastoma, intermixed 20 (90.9) 26 (56.5) Ganglioneuroma 2 (9.1) 20 (43.5) Total 22 (100) 46 (100) CCG: Children s Cancer Group. a Two variant cases (no macroscopically visible nodular formation in the primary site) were excluded from the analysis because of a limited amount of stroma-poor component for histopathologic evaluation in the metastatic bone lesion. b For a detailed explanation of the significant difference for each item, see text. are diagnosed in younger children. It also may explain the fact that the vast majority of the tumors in older children that had a ganglioneuroma (Schwannian stroma-dominant) morphology did not have FN. While handling surgically removed or biopsy samples of the NT, pathologists may encounter some difficulty in diagnosing this subtype of GNBn. Because it is a composite tumor comprised of different clones, 1,3,6,7 careful examination and tissue sampling is critical for histologic as well as molecular evaluation. A detailed description and discussion of this issue can be found in the article published by the INPC. 6 In the current report, we simply mention the fact that, in our series, we found six variant patients in whom the diagnosis of GNBn was not possible by macroscopic examination of the tumors. Four of these patients had a huge stroma-poor component with only microscopically identifiable stroma-rich/stromadominant component at the rim or septal portion in the primary tumor tissue. In another two patients, no macroscopically visible nodule(s) of stroma-poor component were found in (or sampled from) the primary ganglioneuroma tumor, but metastatic neuroblastomatous tissue was found in the bone. Because the nodule(s) could be small in size, careful macroscopic examination and tissue sampling from any suspicious areas of the primary tumor tissue are required for diagnosing GNBn. It should be added here that the very low incidence of MYCN-amplified tumors we observed in this subtype may or may not have been the result of sampling error. Pathologists also should be aware that the nodular histology of this subtype can vary from one nodule to another in the multiple nodular tumor, or even in the different microscopic fields of the same nodule of either single- or multiple-nodular tumors. Histologi-

11 1160 CANCER September 1, 2000 / Volume 89 / Number 5 FIGURE 4. Prognostic subsets in ganglioneuroblastoma, nodular subtype. The favorable subset contains only favorable histology nodules (FN), and the unfavorable subset contains at least one unfavorable histology nodule (UN). FIGURE 5. Event free survival and survival curves for patients with ganglioneuroblastoma, nodular subtype and other subtypes of neuroblastic tumors. FS: favorable subset of ganglioneuroblastoma, nodular subtype; US: unfavorable subset of ganglioneuroblastoma, nodular subtype; FH: favorable histology group of other subtypes of neuroblastic tumors; UH: unfavorable histology group of other subtypes of neuroblastic tumors.

12 Prognostic Subsets of Ganglioneuroblastoma, Nodular/Umehara et al cally different nodules or areas in the same tumor need to be evaluated separately and individually for determination of UN or FN status. Those tumors that contain any UN component should be classified into the US. With the results of this study, we can support the INPC proposal and would like to urge a precise morphologic description for both components i.e., Schwannian stroma-rich/stroma-dominant and stromapoor of the GNBn subtype in surgical pathology reports. The Schwannian stroma-rich/stroma-dominant component is classified into either ganglioneuroblastoma, intermixed or ganglioneuroma, maturing or mature. For the nodular Schwannian stroma-poor component, the grade of neuroblastic differentiation and MKI need to be documented. Next, the nodular histology (FN or UN) should be determined based on the Shimada system of age-linked histopathologic evaluation, 1,7, so that each tumor can be classified as FS or US. We are planning an additional study to test the reproducibility of identifying these two subsets of patients with GNBn and their prognostic effects by analyzing currently accumulated cases from the nationwide intergroup COG studies. After successful completion of the study, these two prognostic subsets should be distinguished for assigning appropriate patient treatment protocols. REFERENCES 1. Shimada H, Chatten J, Newton WA, Sachs N, Hamoudi AB, Chiba T, et al. Histopathologic prognostic factors in neuroblastic tumors; definition of subtypes of ganglioneuroblastoma and an age-linked classification of neuroblastomas. J Natl Cancer Inst 1984;73: Aoyama C, Qualman SJ, Regan M, Shimada H. Histopathologic features of composite ganglioneuroblastoma: immunohistochemical distinction of the stromal component is related to prognosis. Cancer 1990;65: Schmidt ML, Salwen HR, Chagnovich D, Bauer KD, Crawford SE, Cohn SL. Evidence for molecular heterogeneity in human ganglioneuroblastoma. Pediatr Pathol 1993;13: Stout AP. Ganglioneuroma of the sympathetic nervous system. Surg Gynecol Obstet 1947;84: Adam A, Hochholzer L. Ganglioneuroblastoma of the posterior mediastinum: a clinicopathologic review of 80 cases. Cancer 1981;47: Shimada H, Ambros IM, Dehner LP, Hata J, Joshi VV, Roald B. Terminology and morphologic criteria of neuroblastic tumors: recommendations by the International Neuroblastoma Pathology Committee.Cancer 1999;86: Shimada H, Ambros IM, Dehner LP, Hata J, Joshi VV, Roald B, et al. The International Neuroblastoma Pathology Classification (Shimada system). Cancer 1999;86: Matthay KK, Perez C, Seeger RC, Broduer GM, Shimada H, Atkinson JB, et al. Successful treatment for Stage III neuroblastoma based on prospective biologic staging: a Children s Cancer Group Study. J Clin Oncol 1998;16: Evans AC, D Angio GJ, Randolph J. A proposed staging for children with neuroblastoma: Children s Cancer Study Group A. Cancer 1971;27: Matthay KK, Villablanca JG, Seeger RC, Stram DO, Harris RE, Ramsay NK, et al. Treatment of high-risk neuroblastoma with intensive chemotherapy, radiotherapy, autologous bone marrow transplantation, and 13-cis-retinoic acid. N Engl J Med 1999;341(16): Nickerson HJ, Matthay KK, Seeger RC, Brodeur GM, Shimada H, Perez C, et al. Favorable biology and outcome of Stage IV-S neuroblastoma with supportive care or minimal therapy: a Children s Cancer Group study. J Clin Oncol 2000;18: Perez CA, Matthay KK, Atkinson JB, Seeger RC, Shimada H, Haase GM, et al. Biologic variables in the outcome of Stages I and II neuroblastoma treated with surgery as primary therapy: a Children s Cancer Group study. J Clin Oncol 2000;18: Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958;53: