GPOH

GPOH

Annals of Oncology 12: 311-319. 2001. © 2001 Khiwer Academic Publishers. Printed in the Netherlands. Original article Clinical impact of histologic s...

840KB Sizes 0 Downloads 16 Views

Annals of Oncology 12: 311-319. 2001. © 2001 Khiwer Academic Publishers. Printed in the Netherlands.

Original article Clinical impact of histologic subtypes in localized non-anaplastic nephroblastoma treated according to the trial and study SIOP-9/GPOH A.Weirich,1 I. Leuschner,2 D. Harms, 2 G.M.Vujanic,3 J.Troger,4 U. Abel,5 N. Graf,6 D. Schmidt,2 R. Ludwig1 & P. A.Voute7 ^Department of Pediatric Hematology and Oncology. University of Heidelberg. Heidelberg] 2 Department ofPediatric Pathology. University of Kiel. Kiel.^ermanytf Department of Pathology, University of Wales College of Medicine. Cardiff, UK: 4 Department of Pediatric Radiology and 5 Institute of Biostatistics, University of Heidelberg, Heidelberg; 6Pediatric Hematology and Oncology. University of Saarland. Homburg/Saar, Germany; 1 Department of Pediatric Hematology and Oncology, AMC, Amsterdam, The Netherlands

Summary JBa Background: Histologic subtypes of standard histology Wilms' tumof"TWT) and the effect of preoperative therapy on their clinical and histologic features, deserve to be analysed in respect to outcome to find an adequate baseline for therapy. Patients and methods: The German Society of Paediatric Oncology & Haematology enrolled patients from January 1989 to March 1994 for therapy according the International Society of Paediatric Oncology trial & study 9. Standardised preoperative therapy with dactinomycin and vincristine for 4-8 weeks was generally applied in patients between 0.5 and 16 years with localized renal tumors and imaging typical for WT. In 99.5% of cases representative material was sent for review to the Kiel Paediatric Tumour Registry. For prospective subtyping of 329 WT (258 after preoperative therapy, 71 with immediate surgery) modified Beckwith & Palmer criteria were used. Reduction in volume measured by imaging prior to chemotherapy and surgery was used to assess response (poor response: reduction <40%; good response: reduction ^40%). Results: There were 39% of patients treated with immediate surgery and 12.4% of patients with preoperative therapy in the

Introduction

The aim of all nephroblastoma trials and studies is to find a treatment concept adapted to the expected risk of the children. Histologic features and extent (stage) of disease have been traditionally regarded as the most important prognostic criteria [1]. Recently some other characteristics such as weight of the specimen and age at diagnosis [2] or complete necrosis as reaction to preoperative chemotherapy [3] have been included to define different risk groups used to stratify patients for modern therapeutic protocols. In all classifications of renal tumors of childhood by far the largest group is nephroblastoma of standard histology [4] according the definition of the International Society of Pediatric Oncolgy (SIOP), which is very similar to favorable histology Wilms' tumor (WT) of the National Wilms' Tumor

age group up to 12 months. The difference in age (P - 0.022) was linked with different amounts of epithelial WT (15.5% vs. 3.1%), median age: 0.58 and 0.93 years. Due to the effect of chemotherapy the amount of other WT changed: stromal 0% to 14%, mixed 45.1% to 29.4%, blastemal 39.4% to 9.3%). After preoperative therapy 37.6% of WT were predominantly regressive, 6.6% completely necrotic. Poor response was frequent in differentiated WT (86% of stromal, 75% of epithelial WT) but none relapsed. In the other WT with viable tumor left after preoperative therapy > 70% had good response, poor response was a risk factor (P = 0.0057). Conclusions: Subtyping according modified Beckwith & Palmer can be used in WT after preoperative therapy to stratify postoperative therapy in future. A milder therapy could be tested in differentiated WT at low stages and an intensified in the others with viable tumor left and poor response, i.e., mainly blastemal WT. i

Key words: clinical response, histologic subtype, nephroblastoma, preoperative chemotherapy, prognostic factor, risk of relapse

study and of the United Kingdom Children's Cancer Study Group Wilms' Tumour study [5, 6]. Thus, different attempts have been made to recognise histologic subtypes within this group in order to find subgroups of prognostic value [7-9]. The SIOP panel of pathologists (chairman: Dr J. Delemarre) tried, during the SIOP 6 and 9 trials & studies, to assess a prognostic significance of different forms of the blastemal component. But although there were valuable differences between the entities these could not be proven to be significant [10]. In contrast, during the SIOP 9 nephroblastoma trial & study the Kiel Paediatric Tumour Registry of the German Society of Paediatric Oncology and Haematology (GPOH) prospectively did subtyping of nephroblastomas according to modified Beckwith and Palmer original semiquantitative criteria [7, 11]. The aim of this study was to analyse different histologic subtypes of nephro-

312 blastoma, their clinical and histologic features especially in respect to the effect of preoperative chemotherapy and their prognostic relevance in order to find a more adequate baseline for postoperative treatment.

Patients and methods Patients and treatment The GPOH joined the SIOP nephroblastoma trial & study 9 in January 1989, i.e., 11 months after its start, and enrolled patients for treatment according to the protocol until the end of March 1994. The only principal difference concerning the treatment protocol between the GPOH and other SIOP participating centers was the use of adriamycin (ADR) instead of epiadriamycin. Patients diagnosed before July 1993 were additionally registered in the databank of the nephroblastoma trial & study SIOP-9 in Villejuif, France. There were 505 patients registered by the GPOH including all trial and study patients, i.e., also patients with bilateral tumors (stage V), distant metastases at diagnosis (stage IV) and those who underwent immediate surgery [12, 13]. Immediate surgery had been performed in patients younger than 6 months (n = 14) and older 16 years (n = 3) at diagnosis, in cases with a doubt in diagnosis (« = 28), for emergency indications (n = 8). and if surgeons were ignorant of the trial (n = 15) or other reasons (n = 3). Preoperative chemotherapy was recommended for children between 6 months and 16 years of age with characteristic imaging features of nephroblastoma. The duration of the standardised preoperative chemotherapy with dactinomycin (AMD) and vincristine (VCR) in patients with unilateral localised disease (stage I—III) was either four or eight weeks (randomised until October 1991, thereafter recommendation of four weeks). The postoperative treatment depended on stage and histology in stages •—III. As patients with stage IV (69 patients) and stage V (25 patients) received a different preoperative chemotherapy they were excluded from the present study.

Diagnostic imaging All participating centers were asked to send the initial diagnostic imaging to the trial & study center in Heidelberg for confirmation of their diagnosis before the start of treatment. Tumor volumes had to be sonographically measured by using the ellipsoid formula (length x thickness x depth x 0.523) for the three dimensions. The measurement had to be performed before preoperative chemotherapy or immediate nephrectomy, or four and eight weeks after the initiation of preoperative chemotherapy. In addition to the diagnostic images, the measurements had to be documented on the treatment sheets sent to the trial & study center together with the images. For tumors which the centers could not measure adequately by ultrasound scan, the measurement by other multiplanar methods of diagnostic imaging (like computed tomography or magnetic resonance imaging) providing the same method as initially was used for the follow-up tumor measurements, was accepted as equivalent for this study. The tumor measurements of the participating centers were reviewed by the central diagnostic reference group in the study center in Heidelberg, where the original diagnostic images were available. In preoperatively treated cases the measurement just prior to surgery was used to calculate the so-called final tumor volume as the basis for tumor volume reduction in comparison with the initial volume. Concerning the clinical response we defined two groups: 1) poor responders, a) no reduction in tumor volume (increase or reduction <10%), and b) poor reduction (reduction 3:10% and <40%); and 2) good responders, a) good reduction (5= 40% and < 70%), and b) very good reduction ( ^ 70%).

Follow-up Follow-up sheets were sent out for all patients every 6 months until at least 48 months after diagnosis, and the recurrences and tumor-related events had to be documented by the participating centres. In case of recurrence, the follow-up sheets were sent out to obtain information during at least first 48 months after the event.

Statistical methods Histology In all but I out of 411 unilateral localized tumors the material was available for histologic analysis by the institutional pathologists. In 99.5% (408 of 410) of cases representative material and reports of the referring pathologists were sent for central review to the Kiel Paediatric Tumour Registry generally prior to postoperative therapy. In addition, a retrospective review, after the closure of the patients registration, of all cases was done by the SIOP panel of pathologists. In 59 cases a diagnosis of non-Wilms' tumor was made including 24 cases of mesoblastic nephroma, 11 clear-cell sarcomas, 9 rhabdoid tumors, and 15 other tumors, and these were all excluded from the study. Finally, 20 cases of anaplastic WT were excluded, leaving 329 non-anaplastic WT which were treated postoperatively as standard histology. These WT constituted the basic sample of this study and included 258 patients who received pre-operative therapy (229 patients 4, and 29 patients 8 weeks) and 71 who underwent immediate surgery. For histologic subtyping of 329 non-anaplastic WT, the original semiquantitative criteria proposed by Beckwith and Palmer, with addition of a subtype with predominantly chemotherapy-induced regressive changes, were used [7, 11]. The criteria were based on the percentage of different tumor components (blastemal, epithelial, stromal, regressive) and if one of these comprised more than 65% of the tumor sample, the tumor was subtyped accordingly (blastemal, epithelial, stromal, or regressive-change predominant). If there was no predominant component in the viable tumor, it was called mixed. Completely necrotic nephroblastoma were coded separately from the other tumors with regressive changes. In addition to hematoxylin and eosin (H&E) stained histologic sections, immunohistochemistry was performed in doubtful cases in order to discriminate the tumor components. In completely necrotic tumors, additional reticulin and/or periodic acid-Shiff (PAS) stainings were done in order to better visualise necrotic structures.

For comparison of the different groups, standard methods [14-16] were applied using the SAS statistical package including Wilcoxon rank sum test (Table 1, age, tumor volume), the chisquare test (Table I, stage distribution; Table 2, Table 4), Fisher's exact test for comparison of proportions (Table 1, sex, tumor side; Table 2, epithelial and blaslemal WT), and the log-rank test. For the evaluation of the recurrence-free survival the Kaplan-Meier estimates and log rank tests were calculated taking as starting point the first day of treatment (Figures 1 and 2).

Results The main clinical and pathologic features of the study group are presented in Table 1. Concerning the age at diagnosis of the two groups, i.e., the immediately operated and the preoperatively treated patients, there was a certain unbalance (P = 0.022). In particular, the proportion of patients diagnosed in their first year of life differed markedly, i.e., 39% (28 of 71) and 12.4% (32 of 258), respectively. This higher proportion of very young patients in the immediately operated group may be responsible for the fact that this group had significantly smaller tumors when compared with the preoperatively treated patients. The distribution of histologic subtypes of WT is presented in Table 2. The differences between the two groups, i.e., the immediately operated and the preoper-

313 Table I. Main clinicopathologic characteristics of 329 patients. Total (%) Immediate Preoperative /"-value chemosurgery therapy (%) (%) Age at diagnosis (years) Mean 3.16 Median 2.51 Sex 180(55) Boys 149(45) Girls Tumor side 170(52) Left" 159(48) Right" Stage distribution I 189(57) 72 (22) II N 68(21) IIN+/III Median initial tumor volume (cm 3 ) 368C Median final tumor volume (cm 3 ) Total

329(100)

0.022 3.27 2.00

3.13 2.58 0.8

42 (59) 29 (41)

138(53) 120(47)

35"(49) 36 b (51)

135(52) 123(48)

31 (44) 12 (17) 28 (39)

158(61) 60 (23) 40(16)

0.9

0.29

224d

397

-

163

71 (100)

< 0.0001

258(100)

" One extrarenal, b four extrarenal. Data available only for 326 patients, the size of 3 large tumors is missing. d Data available only for 68 patients, the size of 3 large tumors is missing.

c

Table 2. Distribution and age at diagnosis of the histologic subtypes of nephroblastoma. Histologic subtypes

Immediate surgery"

Preoperative chemotherapy"

n (%")

Median age (years)

Epithelial predominant Stromal predominant Blastemal predominant Mixed Regressive predominant Completely necrotic WT

11(15.5) 0(-) 28 (39.4) 32(45.1)

0.58 4.10 2.67 -

8(03.1) 36(14.0) 24 (09.3) 76 (29.4) 97 (37.6) 17(06.6)

0.93 1.73 2.84 2.39 3.61 4.07

Total

71 (100)

2.00

258(100)

2.58

1

OH 0(-)

n (%")

Median age (years)

P = 0.001 (chi-square test).

atively treated patients, were highly significant (P = 0.001) in regard to the proportion of subtypes. In particular, this was true for epithelial (P < 0.001) and the blastemal WT(/>< 0.001). The stage and median size distribution of the histologic subtypes is shown in Table 3. The tumor size was available in all but three older patients who presented with very large tumors that had ruptured prior to immediate emergency surgery. The clinical tumor response to preoperative chemotherapy was assessed on the basis of its change in volume. There was a significant difference between the subtypes

in regard to their proportion of poor or good responders as shown in Table 4. The majority (61%) of the stromal predominant WT showed little change in their size during preoperative therapy. As 13 out of 36 even increased up to 10% of the initial volume and 9 out of 36 that showed only a marginal decrease of less than 10% of their initial volume. In the remaining cases no volume reduction of more than 50% was observed. Moreover, half of the tumors showed minimal regression on histologic analysis (less than 25% regression). The other half showed 25%—35% regressive changes of the tumor mass. The clinical response of the epithelial predominant WTwas similar as there was only one with very good reduction, but six of eight patients had only a poor response. In contrast, more than half of the tumors in each of the other WT (mixed, blastemal predominant, regressive-change predominant and completely necrotic) were good responders. Still, 4 tumors with predominant regressive changes even doubled their initial size due to big cystic structures progressing further during pretreatment and even 2 of 17 completely necrotic tumors enlarged markedly during pretreatment. Follow-up information (range 48-126 months) was available in all but two patients who were lost to followup in their first complete remission (CR) at 2.1 and 2.9 years after the diagnosis, respectively. Additionally, five other patients died of other reasons than the primary or recurrent WT including three who died of treatment toxicity during the first, second and fourth month of treatment, one who died of a metachronous disease on the contralateral side that developed 3.3 years after the primary diagnosis, and one who died of a non-tumor related illness while in CR for more than 3.5 years. The vast majority of the patients, whose course was properly followed for at least four years (298 of 322, 92.6%) survived with no recurrence. The histologic subtypes and stages with respect to relapses of preoperatively treated patients are shown in Table 5. No recurrences occurred in the stromal predominant, epithelial predominant and completely necrotic WTeither after immediate surgery or preoperative therapy. The four children who relapsed in the group of immediately operated children died. Two of them died in the progression of the first relapse (one with initial stage I tumor of mixed subtype and another with a blastemal predominant tumor of stage III), the third following his third relapse (initial stage III), and another one with initial stage I tumor of mixed histology died in CR after his first relapse due to severe late side effect of the recurrence treatment. The four years recurrence-free survival of the patients who received preoperative chemotherapy was 95.5% (231 of 258) (Figure 1). Of the 27 patients who relapsed, 16 survived and have remained in CR for at least 57 months (maximum 116 months, median 78 months) since the date of their last event. The death of 10 children was tumor related and 1 further patient died in second CR due to the late side effect of the recurrence treatment. Three of the nine children who relapsed with blastemal

314 Table 3. The stage and median size distribution of the histologic subtypes of nephroblastoma. Histologic subtypes

Stage I n (%)

Epithelial predominant Immediate surgery Preoperative chemotherapy Stromal predominant" Blastemal predominant Immediate surgery Preoperative chemotherapy Mixed Immediate surgery Preoperative chemotherapy Regressive-change predominant 3 Completely necrotic WT"

Stage II n (%)

Initial size (cm3)

Final size (cm3)

3 (08.4)

132 151 372

126 350

Stage IIN+/III n (%)

10(90.9) 7(87.5) 25 (69.4)

0(-) 1(12.5) 8(22.2)

1 (09.1)

10(35.7) 11 (45.8)

6(21.4) 8 (33.4)

12(42.9) 5 (20.8)

369b 473

219

11 (34.4) 54(71.1) 52(53.6) 9 (53.0)

6(18.7) 16(21.1) 22 (22.7) 5 (29.4)

15(46.9) 6 (07.9) 23 (23.7) 3(17.6)

228C 376 418 449

129 146 115

OH

" No cases in patients with immediate surgery. b The tumor size of one large tumor is missing as not measurable, i.e., diagnosed after tumor rupture. c The tumor sizes of two large tumors are missing as not measurable, i.e., diagnosed after tumor rupture.

Table 4. Tumor volume reduction from the initial to the final measurement. Histologic subtypes

Epithelial predominant Stromal predominant Blastemal predominant Mixed Regressive-change predominant Completely necrotic WT

I Poor responder"

II Good responder a

Reduction in volume < 4 0 % n (%)

Reduction in volume > 4 0 % " (%)

6(75.0) 31 (86.1) 11 (45.8) 21 (27.7)

2 (25.0) 5(13.9) 13(54.2) 55 (72.3)

28 (29.4) 4(23.4)

69 (70.6) 13(76.6)

" P = 0.001 (chi-square test).

Table 5. Relapses and stages in nephroblastoma after preoperative chemotherapy. Histologic subtypes

Stage I

Stage II n (%)

Stage IIN+/I1I n (%)

Epithelial predominant Stromal predominant Blastemul predominant Mixed Regressive-change predominant Completely necrotic WT

0/7 (0) 0/25(0) 5/11(45.4) 5/54(09.3)

0/1 (0) 0/8 (0) 2/8 (25.0) 3/16(18.8)

- (0) 0/8 (0) 0/3 (0) 0/36(0) 2/5 (40.0) 9/24(37.5) 0/6 (0) 8/76(10.5)

4/52(07.7) 1/22(04.5) 5/23(21.7) 0/9 (0) 0/5 (0) 0/3 (0)

Total «(%)

10/97(10.3) 0/17(0)

predominant WT had a tumor related death including two with initial stage 1 disease who died during the first relapse and one with an initial stage III disease who died of his second relapse. On the other hand, seven of eight patients with relapses of a mixed WT achieved a second CR. One of the seven who initially presented with stage I died later in second CR due to side effects of the pulmonary irradiation which he received for lung metastases. Another child with initial stage II N - tumor died later in a second local relapse after a successful treatment

of the first relapse. Eight out of ten recurrences in tumors with predominant regressive changes achieved a second CR. Two of them later developed a second relapse which could not be cured. Finally, 6 of 10 children were succesfully cured of the recurrent disease. The recurrence-free survival of stromal predominant, epithelial predominant, and completely necrotic subtype (n - 61) was significantly better than that of the other subtypes, i.e., blastemal predominant, mixed and regressive predominant nephroblastoma (Figure 2). Clinical poor response was a significant risk factor (P = 0.0057, log-rank test, no table given) associated with a poor prognosis (14 relapses out of 60 patients) when compared to good clinical response (13 relapses out of 137 patients) for certain subtypes, i.e., blastemal predominant, mixed and predominant regressive tumors. Within these 14 poor responders developing later relapses, there were 8 patients with stage I, blastemal predominant (3 patients), mixed (3 patients) and regressive predominant (2 patients) tumors. This clinical poor response was also a significant risk factor (P - 0.0099) if evaluated in the same subtypes after 4 weeks pretreatment only (13 relapses out of 57 cases with poor response vs. 11 relapses out of 118 cases with good response).

DisCUSSion

The data of the SIOP-9/GPOH trial & study include 95% of the patients diagnosed in that 1989-1994 period and registered in the German Childhood Cancer Registry in Mainz [12] and, therefore, represent a realistic distribution of WT. The previous GPOH study had mainly followed the principles of the NWTS trial and studies in which initial surgery is recommended as the first step in treatment of unilateral localised WT [17] and preoperative treatment is restricted to selected cases. Therefore, introduction of the SIOP protocol which included preoperative chemotherapy as the initial therapeutic step for patients between 6 months and 16 years was a

315 1.0H_L" Jll_

U.

-

0.8

0.8

0.4

0.2

0.0-

-1

1

1

1

1

!

,

1

,

1—1

1

1

1

1

1

1

1

0

1

1

.-—i

r—T

1

i

i

2

1

1

1

1

,

3

r

4

S

years









predom. -epithelial, -stromal WT mixed triphasic WT predominant blastemal WT at least 2/3 regressive WT totally regressive WT

-

N 44 76 24 97 17

# recurrences 0 8 9 10 0

4-years survival 1.00 0.89 0.61 0.90 1.00

SIOP 9/GPOH October 2000

95%-Q: 1.00 - 1.00 0.83 - 0.88 0.41 -- 0.81 0.84 -- 0.96 1.00 -- 1.00 ZMBT HEIDELBERG

Figure I. Standard Wilms stage 1—III after preoperative chemotherapy (n = 258). Recurrence-free survival, recurrences di = 27).

1.0 - j - 1 ^

' V

^

0.8

0.6:

0.4

0.2

0.0-

years mixed, predom. blastemal, > 2/3 regressive 197 predom.-epithelial,-stromal, 100% regressive 61 SIOP 9/GPOH October 2000

# recurrences 27 0

LOG RANK TEST P = 0.0032

4 - years survival 0.86 1.00

95%-Q 0.81 - 0.91 1.00 - 1.00 ZMBT HEIDELBERG

Figure 2. Standard Wilms stage I—III after preoperative chemotherapy (n = 258). Recurrence-free survival, recurrences (ft = 27).

significant change for the GPOH centers. This resulted in a cautious approach to the new strategy and, consequently, a higher percentage of initially operated patients in the GPOH compared to the other SIOP-9 centers [18]. There has been only one larger study on the effect of

preoperative therapy in different subtypes of WT [19]. However, it dealt with 119 highly selected unilateral cases from the NWTS-3 trial which were deemed inoperable because of large size or extent of disease. Pathologic material from a definite resection was available in 83

316 cases, 38 of which had been accompanied by pretherapy biopsy. Only pretherapy biopsy slides had been available from the remaining 36 patients. Finally, follow-up data for the subtypes of the so-called favorable histology WT were given for 68 patients including 19 of stage IV. Preoperative chemotherapy was not standardized, and although the majority of patients received chemotherapy as for high risk WT including ADR, its duration varied considerably (3-27 weeks) and included in some patients also abdominal irradiation. Other studies have included smaller series [20] or analysed one selected subtype only [3]. In contrast, our study dealt with a group of unilateral localized WT which were treated with a standardised preoperative chemotherapy, and a small not randomized but representative group of immediately operated WT. This gave us an opportunity to assess clinicopathologic features of the subtypes in both groups and draw conclusions about the effect of preoperative chemotherapy on their histologic features and their prognostic significance. Although the group of immediately operated patients is not too small for itself, its comparison with the much larger group of patients treated with preoperative therapy is statistically unbalanced. Therefore, the results of previous studies which included cases with immediate surgery and where the same histologic subtyping was performed [17, 21] had to be combined in order to reach better founded conclusions. A noticable but not significant difference between the cases treated with immediate surgery and preoperative therapy was in stage distribution, where a percentage of lower stages (stage I-II) increased from 61% to 84% (P = 0.36) of cases, respectively (Table 1). This was mainly due to the well-known fact that preoperative therapy results in downstaging of tumors [4, 22], but also due to some emergency indications (e.g., preoperative tumor ruptures or massive tumor bleeding) which prompted immediate surgery and were associated with high local stages [23]. Another difference between the two groups was in distribution of histologic subtypes (Table 2), most noticable being the absence of stromal predominant WT in the immediately operated patients. However, the results from the previous GPOH study, which followed more or less the principles of the NWTS trial concerning the initial therapeutic approach, showed that only 4 of 373 (1.0%) patients had stromal predominant WT [17]. A low incidence of this subtype in immediately operated patients has also been shown in the NWTS-4 trial where 35 (1.6%) such cases were seen amongst 2077 favorable histology WT of stage I-IV [24]. In contrast, there were 14% of stromal predominant WT in the group that received preoperative therapy (Table 2). Interestingly, these tumors showed only poor clinical response (Tables 3 and 4). This has been noted before [25], and in a recent paper on a small series of fetal rhabdomyomatous WT it has been regarded as a proof of a tumor's resistance to therapy [26]. However, in our study proportion of predominant stromal WT increased up to 14% suggesting that only few of them

were stromal predominant before chemotherapy. It is highly likely that chemotherapy induced further differentiation and maturation of other subtypes as shown in other tumors like rhabdomyosarcoma [27, 28]. Since the percentage of mixed type tumors decreased from 45.1% in the immediately operated patients to 29.4% in the preoperatively treated, one could speculate that many of these tumors became stromal predominant after chemotherapy [30]. In Zuppan et al. series the percentage of tumors that contained skeletal muscle increased from 4% to 31% after chemotherapy, and the percentage of cases with the stromal pattern increased to 7.4% (5 of 68) of the follow-up cases (stage I-IV) [19]. The diseasefree survival of patients with stromal predominant WT was 100% in our and the NWTS-3 series [19]. Still, there have been rare case reports of this tumor type with relapses at higher stages, especially when the tumors were not resectable [29, 30]. Recently, it has been shown that stromal predominant WT were not exclusively but typically associated with WT1 mutations [31] and that the loss of the WT1 function leads to ectopic myogenesis in WT [32]. Compared to the stromal predominant WT, the number of the epithelial predominant subtype is rather small in our series and permits only limited conclusions. Our data showed that there was about a 10% difference in the proportion between the two groups (15.5% vs. 3.1%). But bearing in mind the fact that there was a big difference in the percentage of children diagnosed in the first year of life when epithelial predominant WT are far more common (Table 2), we can assume that the true difference between the two groups is only about 3%-5%. However, the NWTS-4 results [24] showed 9.3% of epithelial predominant WT, the vast majority of which were at lower stages, as also found in our series. An excellent outcome for patients with epithelial predominant WT in our series confirms previous reports which have also shown its favorable prognosis especially in lower stages [24]. However, in inoperable cases in higher stages the prognosis for these tumours can be much worse, with a disease free survival of only 57% in Zuppan et al. NWTS-3 series [19] and 42.9% in the NWTS-4 series for stage III and IV [24]. There is up to now no evidence that preoperative therapy influences the prognosis of the epithelial predominant subtype since even old reports from the pre-preoperative chemotherapy era had shown that prognosis was excellent for young patients with tumors of lower stages [33, 34]. However, Zuppan observed [19] that in preoperatively treated cases it is a prognostically favorable finding if there are more than 75% of differentiated elements in the residual tumor. An earlier analysis of more than 2,000 favorable WT outlined also that the differentiated WT, i.e., the stromal predominant and the epithelial predominant, have these above characteristics at one time - the low aggressivity, as typically represented at low stages with an excellent follow-up, and the high resistancy to therapy, when very rarely seen at higher stages [24].

317 In contrast to these chemoresistent WT, the percentage of the blastemal predominant subtype decreased significantly from 39.4% to 9.3% (Table 2). It has been noted earlier that blastema is the most responsive tumor component to chemotherapy. In Zuppan et al.'s series the percentage of tumors showing blastema decreased from 99% in pretherapeutic biopsies to 65% in the specimen after chemotherapy. Also, the percentage of cases with the blastemal predominant pattern decreased markedly to only 13.2% (9 of 68), and most of them (7 of 9) were stage IV patients [19]. However, the comparison of our results with that study is difficult due to the fact that their patients were treated with a more aggressive chemotherapy consisting of three drugs, which is likely to have caused more extensive destruction of the blastema and the whole tumor. The SIOP-9 study [3] has shown that a number of completely necrotic nephroblastoma is markedly higher in stage IV patients whose preoperative chemotherapy always included an anthracycline compared to those with localised disease treated with AMD and VCR only. In contrast, the duration of chemotherapy of four or eight weeks had no influence on the number of completely necrotic nephroblastoma [3]. Also, the number of stage I cases in both arms differed not significantly [18] although there was a certain further decrease of the tumor size observed within the second four weeks if randomized to eight weeks. Although the proportion of the blastemal predominant WT decreased after preoperative therapy, this subtype still representing after preoperative chemotherapy as blastema resistent to chemotherapy was the most likely one to develop relapses irrespectively of stage (Table 5). However, the chance of these patients to be cured from recurrence was comparable to that of the other subtypes. The NWTS-3 study showed similar results as disease-free survival of children with the blastemal pattern resistant to chemotherapy was only 56% (5 of 9) [19]. Still, as the number of our cases of blastemal predominant WT after preoperative therapy was so small and the confidence interval concerning the recurrences very broad (Figure 2), the results have to be interpreted carefully. But according to our data at least blastemal predominant WT after preoperative chemotherapy which had besides this pathologic sign of low regression a clinical poor response require a more intensive postoperative strategy. The prognosis of the mixed and predominant regressive WT was much better (Figure 2) and they showed in most cases a good clinical response (Table 4). When completely necrotic and the predominantly regressive WT are grouped and analysed together, they showed only 10 relapses among 114 patients, with a 92% fouryear recurrence-free survival. Zuppan et al. were the first who noted that 'massive necrosis' was associated with an excellent prognosis as disease-free survival of their series was 94% (15 of 16) [19]. In our previous papers we described typical imaging features of WT [35] and their response to preoperative therapy [36]. In this study, for the first time, we showed

in a representative series that WT could be evaluated concerning its clinical response to preoperative therapy in relation to its histologic subtypes. Clinically good response was in most cases associated with a histologically good response in the sense of a large amount of histologic regressive changes. On the other hand, a histologically good response is not necessarily linked with a good clinical response as 2 of 17 completely necrotic tumors increased in size and 4 of 97 predominantly regressive WTeven doubled their size due to the further increase of big cystic structures, as rarely noted before [37]. Finally, a histologic tumor response in form of further maturation or differentiation of its components, as it is characteristic for stromal predominant WT which typically keep their initial size and solid structure in most cases, cannot be visualised on imaging studies. Therefore, changes of tumor volume or structure on radiologic imaging alone can never be regarded as an absolute criterion for the tumors response to chemotherapy but a detailed histologic assessment of the tumor must be done. In general, the aggressiveness of postoperative treatment is adapted to the expected risk of failure, which is based on its histologic analysis concerning stage and type. According to the SIOP-9 and the ongoing SIOP93-01 protocol, stage I tumors of low risk [38] receive no postoperative treatment, whereas the other stages receive a postoperative treatment like standard histology or intermediate risk WT, i.e., it always includes an anthracycline. As none of the differentiated WT (stromal and epithelial predominant) in our sample which represented mainly lower stages developed any relapse, it seems reasonable to test in future a reduced or milder postoperative therapy for the low stages, i.e., when tumors are completely resected (stage I and II N - ) . Other WT (blastemal predominant, mixed and predominantly regressive subtype), were more frequently associated with relapses. Relapses were especially common within the group of WT which still contained predominantly blastemal tissue in the viable tumor after preoperative therapy suggesting its limited responsiveness to the drugs given (Table 4). Furthermore, within the 'higher' risk tumors only-poor clinical response was outlined as a significant risk factor for a later relapse. Therefore, they should be regarded as 'higher' risk tumors and it could be reasonable to test an intensification of their postoperative treatment. This would mean to adapt a strategy which is already successfully used in the treatment of stage IV where the aggressiveness of postoperative treatment depends on local stage and histology and clinical response of the metastases to preoperative chemotherapy and their resectability thereafter [39, 40].

Conclusions The histologic response of a standard histology WT to preoperative therapy can be resistance, further differentiation and maturation, or regression. Preoperative

318 chemotherapy alters histologic and clinical features in most cases so markedly that a large group of localised unilateral WT treated according to the SIOP-9 protocol for standard histology can be further subdivided in subtypes of 'lower' risk (epithelial predominant and stromal predominant) and subtypes of 'higher' risk (predominant blastemal, mixed and those with predominant regressive changes). It could be tested in future studies if postoperative treatment can be adapted to these different risks of failure. As preoperative chemotherapy can be regarded as an 'in vivo test' of the cytostatics used, tumor's histologic and clinical no or poor response could be seen as an indicator of ineffectiveness of these drugs and a different postoperative chemotherapy protocol could be used.

Acknowledgements Grant sponsor: 'Deutsche Krebshilfe'; grant number: M61/89/Lul. We thank all the 78 participating centers, whose physicians and other health professionals, made this study possible by their care of the children. For her most accurate data management we thank C. Pallus-Barutzki of the Institute of Biostatistics, University of Heidelberg. We are grateful to Dr U. Janig for her review of tumors sent to the Kiel Paediatric Tumour Registry. We thank Prof. B. Beckwith for his help in cases sent for a second opinion and we are grateful to Dr B. Sandtstedt, for reviewing and supporting the manuscript as chairman of the SIOP panel of pathologists. We thank the radiologists Prof. Dr K.. Rieden and Dr W. Rohrschneider of the University of Heidelberg for their review of the diagnostic imaging sent to the study center. Prof. B. Royer-Pokora and Dr V. Schumacher of the Institute of Human Genetics, University of Diisseldorf, supported us with their helpful discussions. We thank Prof. Walter Niitzenadel and Dr B. Selle of the Children's Hospital, University of Heidelberg, for their most appreciated personal support.

References 1. Charles A, Vujanic GM, Berry PJ et al. Renal tumours of childhood. Histopathology 1998; 32: 293-309. 2. Green DM, Beckwith J B, Weeks DA et al. The relationship between microsubstaging variables. Age at diagnosis and tumor weight of children with stage I/favorable histology Wilms' tumor. A report from the National Wilms' tumor study. Cancer 1994; 74: 1817-20. 3. Boccon-Gibod L, Rey A, Sandstedt B et al. Complete necrosis induced by preoperative chemotherapy in Wilms' tumor as an indicator of low risk. Med Pediatr Oncol 2000; 34, 183-90. 4. Tournade MF, Com-Nougue C, Voiite PA et al. Results of the Sixth International Society of Pediatric Oncology Wilms' Tumor Trial and Study: A risk-adepted therapeutic approach in Wilms' tumor. J Clin Oncol 1993; 11: 1014-23. 5. D'Angio GJ, Evans A, Beckwith JB et al. Treatment of Wilms' tumor. Results of the Third National Wilms'tumor study. Cancer 1989; 64: 349-60.

6. Pritchard J, Imeson J, Barnes J et al. Results of the United Kingdom Children's Cancer Study Group First Wilms' Tumor Study. J Clin Oncol 1995; 13: 124-33. 7. Beckwith JB, Palmer NF. Histopathology and prognosis of Wilms' tumor. Results from the First National Wilms' Tumor Study. Cancer 1978; 41: 1937-48. 8. Marsden HB, LawlerW, CarrTet al. A scoring system for Wilms' tumour. Pathological study of Second Medical Research Council (MRC) Trial. Int J Cancer 1984;33: 365-8. 9. Delemarre JFM, Sandstedt B, Marchant G et al. SIOP nephroblastoma trials and studies, morphological aspects. Proceedings of the XHIth Annual Meeting of the International Society of Pediatric Oncology, Marseille, September 1981; 19-24. 10. Vujanic G, Delemarre JFM, Harms D et al. Histopathological aspects of the SIOP Nephroblastoma Trial and Study 6. Proceedings of the XVIIIth Annual Meeting of the International Society of Pediatric Oncology, September 1986, 249-50. 11. Schmidt D, Harms D, Leuschner 1. Malignant renal tumors of childhood. Path Res Pract 1992; 188: 1-15. 12. Ludwig R, Weirich A, Burger D et al. Application of a new therapeutic strategy for nephroblastoma in the area of the Society of Paediatric Oncology and Haematology SIOP 9/GPOH. Monatsschr Kinderheilk 1997; 145: 128-35. 13. Ludwig R, Weirich A, Potter R et al. Priioperative Chemolherapie des Nephroblastoms. Vorlaufige Ergebnisse der Therapiestudie SIOP-9/GPOH. Klin Padiatr 1992; 204: 204-13. 14. Fisher RA (ed). Statistical Methods for Research Workers. Edinburgh: Oliver & Boyd 1972. 15. Peto R, Pike MC, Armitage P et al. Design and analysis of randomized clinical trial requiring prolonged observation of each patient. II Analysis and examples. Br J Cancer 1977; 35: 1-39. 16. Marabini E,Vaisecchi MG. Competing Risks. Analyzing Survival Data from Clinical Trials and Observational Studies. New York: John Wiley & Son 1995. 17. Gutjahr P, Kaatsch P, Spaar HJ et al. Klinik, Therapie und Prognose bei 373 Kindern mit Wilms-Tumoren. Ergebnisse der bundesweiten Studie 1980-1988. Aktuelle Urol 1989; 21: 132-41. 18. Tournade MF, Com-Nougue' C, de Kraker J et al. Optimal duration of preoperative therapy in unilateral and non metastatic Wilms' tumor in children over six months of age. Results of the 9th SIOP Wilms' tumor trial and study. J Clin Oncol (submitted). 19. Zuppan CW, Beckwith JB, Luckey DW et al. Effect of preoperative therapy on Wilms' tumor histology: Analysis of cases from the third National Wilms'tumor study. Cancer 1991; 68: 385-94. 20. Guarda LA, Ayala AG, Jaffe N et al. Chemotherapy-induced histologic changes in Wilms' tumor. Pediatric Pathology 1984; 2: 197-206. 21. Breslow N, Sharpies K, Beckwith JB et al. Prognostic factors in nonmetastatic favorable histology Wilms' tumor. Results of the Third National Wilms' Tumor Study. Cancer 1991; 68: 2345-53. 22. Lemerle J, Voute PA, Tournade MF et al. Effectiveness of preoperative chemotherapy in Wilms' tumor: Results of an International Society of Paediatric Oncology (SIOP) clinical trial. J Clin Oncol 1983; 1: 604-9. 23. Godzinski J, Weirich A, Tournade MF et al. Primary nephrectomy for emergency in the International Society of Paediatric Oncology Nephroblastoma Trial and Study No. 9. Eur J Pediatr Surg (in press). 24. Beckwith JB, Zuppan CE, Browning NG et al. Histological analysis of aggressiveness and responsiveness in Wilms' tumor. Med Pediatr Oncol 1996; 27: 422-8. 25. Harms D, Gutjahr P, Hohenfellner R et al. Fetal rhabdomyomatous nephroblastoma. Pathologic histology and special clinical biologic features. Eur J Pediatr 1980; 133: 167-72. 26. Maes P, Delemarre J, de Kraker J et al. Fetal rhabdomyomatous nephroblastoma: A tumour of good prognosis but resistant to chemotherapy. Eur J Cancer 1999; 35: 1356-60. 27. Schmidt D, Reimann O, Treuner J et al. Cellular differentiation and prognosis in embryonal rhabdomyosarcoma. A report from

319

28.

29.

30.

31.

32.

33. 34.

35.

36.

the Cooperative Soft Tissue Sarcoma Study (CWS). Virchows Arch 1986; 409: 183-94. Lollini PL, de Giovanni C, Del Re B et al. Myogenic differentiation of human rhabdomyosarcoma cells induced in vitro by antineoplastic drugs. Cancer Res 1989; 49: 3631-6. Van der Kwast TH, ten Kate FJW, Vuzevski VD et al. Hepatic rhabdomyomatous tumor: Late sequel of a fetal rhabdomyomatous nephroblastoma. Pediatr Pathol 1991; 12: 449-56. Ceccamea A, Dominici C, Clerico A et al. Bilateral rhabdomyomatous tumor relapsed as typical triphasic Wilms' tumor. Tumori 1987; 73: 85-9. Schumacher V, Schneider S, Figge A et al. Correlation of germline and two-hit inactivation of the WTl gene with Wilms tumors of stromal-predominant histology. Proc Natl Acad Sci USA; 1997; 94: 3972-7. Miyagawa K, Kent J, Moore A et al. Loss of WTl function leads to ectopic myogenesis in Wilms' tumour. Nature Genet 1998; 18: 15-7. Chatten J. Epithelial differentiation in Wilms' tumor: A clinicopathological appraisal. Perspect Pediatr Pathol 1976; 3: 225-54. Chambers CH, Camitta BM, Garg U et al. Nephroblastoma (Wilms' tumor): Tubule density and prognosis. Med Pediatr Oncol 1978; 5: 127-35. Rieden K, Weirich A, Troger J et al. Accuracy of diagnostic imaging in nephroblastoma before preoperative chemotherapy. EurRadiol 1993; 3: 115-22. Rieden K, Weirich A, Troger J et al. Tumor response to pre-

37.

38.

39.

40.

operative chemotherapy in nephroblastoma. In Breit A (ed): Advanced Radiation Therapy, Tumor Response and Treatment Planning. Berlin: Springer-Verlag 1992; 249-53. Shimizu H, Jaffe N, Eftekhari F. Massive Wilms' tumor: Sonographic demonstration of therapeutic response without alteration in size. Pediatr Radiol 1987; 17: 493-4. Vujanic GM, Delemarre JFM, Sandstedt B et al. The new SIOP (Stockholm) Working Classification of renal tumours of childhood. Med Pediatr Oncol 1996; 26: 145-6. De Kraker J, Lemerle J, Voute PA et al. Wilms tumor with pulmonary metastases at diagnosis: The significance of primary chemotherapy. J Clin Oncol 1990; 8: 1187-90. De Kraker J, Tournade MF, Weirich A et al. Wilms tumour stage IV. A report from the SIOP-9 study. Med Pediatr Oncol 1997; 29: 370.

Received 26 June 2000; accepted 8 November 2000.

Correspondence to: A. Weirich, MD Abteilung Hamatologie und Onkologie Universitatskinderklinik Im Neuenheimer Feld 150 69120 Heidelberg Germany E-mail/[email protected]

/