Hemangiopericytoma in Pediatric AgesA Report from the Italian and German Soft Tissue Sarcoma Cooperative Group

Andrea Ferrari, M.D.1

Michela Casanova, M.D.1

Gianni Bisogno, M.D.2

Adrian Mattke, M.D.3

Cristina Meazza, M.D.1

Alessandro Gronchi, M.D.1

Giovanni Cecchetto, M.D.5

Paola Fidani, M.D.4

Denise Kunz, M.D.3

Jorn Treuner, M.D.3

Modesto Carli, M.D.2

1 Pediatric Oncology Unit, Istituto Nazionale Tu-mori, Milano, Italy.

2 Pediatric Hematology Oncology Division, PadovaUniversity, Padova, Italy.

3 Olgahospital Stuttgart, Stuttgart, Germany.

4 Divisione Oncologia Pediatrica, Ospedale BambinGesu, Roma, Italy.

5 Pediatric Surgery Division, Padova University,Padova, Italy.

Address for reprints: Andrea Ferrari, M.D., Pediat-ric Oncology Unit, Istituto Nazionale Tumori, Via G.Venezian, 1 - 20133 Milano MI, Italy; Fax: 139 0226 65 642; E-mail: ferrari@istitutotumori.mi.it

Received June 17, 2001; revision received July 26,2001; accepted August 2, 2001.

BACKGROUND. Hemangiopericytoma (HPC) is very uncommon in childhood and

comprises two different clinical entities, the adult type and the infantile type,

occurring in the first year of age. We report on a series of 27 pediatric patients

treated from 1978 to 1999 by the Italian and German Soft Tissue Sarcoma Coop-

erative Group.

METHODS. Seven patients had infantile HPC; complete resection was achieved in

the tumors of five patients and chemotherapy was given to four patients. Twenty

children had adult type HPC; nine received complete tumor resection (four pa-

tients at diagnosis and five at delayed surgery). Post-operative radiotherapy was

administered to 15 patients, chemotherapy to 19.

RESULTS. Six of seven patients with infantile HPC were alive in first remission; one

patient died of disease. Chemotherapy achieved an objective response in four of

four patients. Among the adult type HPC cases, 5-year event free survival was 64%

(median follow-up 125 months); 12 patients were alive in first remission, eight

patients relapsed and died of disease. Seven of 10 evaluable patients showed good

response to chemotherapy. Statistically significant differences in outcome were

observed in relation to Intergroup Rhabdomyosarcoma Study grouping, size, local

invasiveness, and gender.

CONCLUSIONS. Infantile HPC is a unique entity probably related to infantile myo-

fibroblastic lesions and characterized by a high response to chemotherapy, which

is required in case of unresectable, life-threatening tumors. In children over 1 year

of age, HPC behaves like its adult counterpart; complete surgical resection remains

the mainstay of treatment, but chemotherapy and radiotherapy seem effective and

are recommended in all patients with incomplete tumor resection and/or locally

invasive, large tumors. Cancer 2001;92:2692– 8. © 2001 American Cancer Society.

KEYWORDS: hemangiopericytoma, infants, children, soft tissue sarcoma

Hemangiopericytoma (HPC) is a soft tissue tumor derived frommesenchymal cells with pericytic differentiation.1,2 It is a tumor of

adult life (fifth decade) and is very uncommon in children; the casesreported in children account for less than 10% of all HPC cases.3 Itusually presents as a deep soft tissue mass with insidious growth andcan occur in any part of the body, the lower extremities being themost common location.

Microscopic diagnosis is based primarily on the recognition ofthe characteristic architectural pattern, the so-called pericytoma pat-tern, with tightly packed cells around ramifying thin-walled, endothe-lium-lined vascular channels ranging from small capillary-sized ves-sels to large gaping sinusoidal spaces.1 Tumors display a widespectrum of features as far as cellularity, appearance and type ofvessels, degree of fibrosis, necrosis, and mitotic activity, so it is diffi-

2692

© 2001 American Cancer Society

cult to predict biologic behavior and distinguish low-grade and high-grade lesions on the basis of histologicparameters.1,4

In childhood two distinct clinical entities are de-scribed: the adult-type in children older than 1 year,with a clinical behavior similar to HPC of adults, andthe infantile-type, occurring in the first year of life.Most infantile HPCs are considered congenital andrepresent about one third of pediatric HPC. Their clin-ical features and behavior are peculiar: the tumor typ-ically arises in the subcutis and oral cavity; multifocaland metastatic diseases are reported, as well as goodresponses to chemotherapy and spontaneous regres-sions. The prognosis is usually favorable.4

Since HPC in pediatric ages is so rare, few pub-lished reports and data on its clinical management areavailable.5– 8 Surgical resection is the mainstay oftreatment; the effectiveness of chemotherapy and ra-diotherapy remains to be established, though somereports have suggested a significant role for adjuvanttreatments.9

To contribute further information on the clinicalmanagement of childhood HPC, we report on a seriesof pediatric patients treated by the Soft Tissue Sar-coma Italian Cooperative Group (STS-ICG) and theGerman Cooperative Group (CWS).

MATERIALS AND METHODSBetween January 1978 and December 1999, 27 consec-utive cases of previously-untreated children with adiagnosis of HPC were registered at ICG (16 patients)and CWS (11 patients) centers. This set of patientsrepresented 0.7% of all pediatric soft tissue sarcomasregistered in the study period. Complete informationon clinical data, treatment modalities, and outcomewere available for all patients and were reviewed. Thehistopathologic evaluation was confirmed at the timeof diagnosis by the ICG and CWS board of patholo-gists.

Tumor extent was assessed according to both theclinical Tumor-Nodes-Metastases (TNM) pretreat-ment staging system10 and the Intergroup Rhabdo-myosarcoma Study (IRS) post-surgical grouping sys-tem.11 The TNM definition T1 refers to tumorsconfined to the organ or tissue of origin, while T2lesions invade contiguous structures; T1 and T2groups are further classified as A or B according totumor diameter, # or . 5 cm, respectively. Regionalnode involvement was designated as N1 (no nodeinvolvement - N0), distant metastases at onset as M1(no metastases – M0).10 After initial surgery, patientswere classified according to the IRS system: Group Iincludes completely-excised tumors, Group II indi-cates grossly-resected tumors with microscopic resid-

ual disease, Group III includes patients with grossresidual disease after incomplete resection or biopsy,Group IV comprises patients with metastases at on-set.11

Patients were treated using multimodality thera-peutic approaches including surgery, chemotherapy,and radiotherapy, based on current protocols. Treat-ment strategies did not change substantially over theyears. Primary excision was attempted when completeand non-mutilating resection was considered feasible,otherwise a biopsy was performed. In some cases,repeat surgery (primary re-excision) was recom-mended prior to any other treatment when micro-scopic residual disease was suspected. To avoid re-sorting to mutilating surgery, when the disease wasconsidered not completely resectable at diagnosis,chemotherapy was administered to shrink the tumorand make it more resectable at subsequent surgery.

Thereafter, radiotherapy was given to patientsconsidered at risk of local relapse due to micro-mac-roscopically incomplete resection or large tumor size.External beam irradiation was administered with con-ventional fractionation (200 cGy/day) for a total doseranging from 35 to 70 Gy, median 50 Gy. The radiationtarget volume included the initial mass plus 2-3 cmmargins and the surgical scars as well.

Different chemotherapeutic regimens were adoptedover the years. The VACA regimen consisted of vin-cristine 1.5 mg/m2, Weeks 1, 4, and 7; adriamycin 30mg/m2/day, for 2 days, Weeks 1 and 7; cyclophosph-amide 1200 mg/m2, Weeks 1, 4, and 7; actinomycin-D0.5 mg/m2/day, for 3 days, Week 4, for a total of 37or 46 weeks. Adriamycin was not used in childrenyounger than 1 year (VAC regimen). Ifosfamide re-placed cyclophosphamide in IVA (ifosfamide 3 g/m2/day, for 2 days; vincristine 1.5 mg/m2, actinomycin-D1.5 mg/m2, every 3 weeks for a total of 27 weeks ) andVAIA regimens (vincristine 1.5 mg/m2, Weeks 1-7, ac-tinomycin-D 1.5 mg/m2, Weeks 1 and 7, ifosfamide 3g/m2/day, for 2 days, Weeks 1, 4, and 7, adriamycin 40mg/m2/day, for 2 days, Week 4, for a total of 27 weeks).In two metastatic patients a more intensive regimenincluding carboplatin and etoposide was used: theCEVAIE schedule consisted of carboplatin 500 mg/m2,Week 1; epirubicin 150 mg/m2, Week 1; vincristine 1.5mg/m2, Weeks 1-7; actinomycin-D 1.5 mg/m2, Week 4;ifosfamide 3 g/m2/day, for 3 days, Weeks 4 and 7;etoposide 200 mg/m2/day, for 3 days, Week 7, for atotal of 27 weeks (in all regimens, the maximum ad-ministered dose of vincristine and actinomycin-D was2 mg).

Response to treatment was evaluated after 9weeks of chemotherapy based on the reduction in thesum of the products of the perpendicular diameters of

Childhood Hemangiopericytoma/Ferrari et al. 2693

all measurable lesions, defined as follows: completeresponse (CR) 5 complete disappearance of disease;partial response (PR) 5 tumor reduction over 50%;minor response (MR) 5 a reduction over 25%. Stabledisease or a reduction under 25% was recorded as noresponse, while an increase in tumor size or the de-tection of new lesions was considered as progressionof disease.

Event free survival (EFS) and overall survival(OS) were estimated according to the Kaplan-Meiermethod.12 Patients were evaluated from the date ofdiagnosis to disease progression, relapse, or deathfrom any cause (whichever occurred first) for EFS, andto death for OS. The time scale extended up to thelatest follow-up if no event was observed. The log-ranktest was used to compare the survival curves of thedifferent subgroups of patients to define the potentialvalue of prognostic factors.13 Patient follow-up, as ofApril 2001, ranged from 24 to 276 months (median 125months).

RESULTSThe diagnosis was performed in the first year of life inseven patients, with 20 children were over 1 year ofage. Data from patients with infantile and with adult-type HPC were analyzed separately.

Infantile-HPCTable 1The diagnosis of HPC was obtained in all cases at birthor during the first two months of life. Six of the sevenpatients were male. The site of the primary tumor wasthe head and neck region in three patients, with trunk,pelvis, extremities, and liver sites in one patient each.

According to the TNM system, three patients wereclassified as T1A, one as T1B, one as T2A, two as T2B;one child had nodal involvement at diagnosis (N1),none had distant metastases or multifocal disease.The maximum tumor diameter was smaller than 3 cmin two patients, between 3 and 5 cm in two patients,between 5 and 10 cm in three patients. According tothe IRS system, after initial surgery two children werestaged as Group I (complete excision) and five asGroup III (four incomplete resections, one biopsy).

The two patients with complete resection receivedno adjuvant therapy. Four of the five Group III chil-dren received chemotherapy (two VAC, two VAIA reg-imen): one achieved CR, while three had PR and un-derwent subsequent complete resection of theirresidual disease. In these patients, delayed re-excisionwas scheduled after 10-12 weeks of chemotherapy,continued thereafter to complete the required 27-37weeks of treatment. In one patient with a macroscopicincomplete resection of a large paravertebral cervicaltumor, chemotherapy was initially refused by parents.A conspicuous local progression was observed 2months after surgery. From then on, different chemo-therapeutic regimens obtained only partial and tem-porary control of the disease, and the patient died 2years after diagnosis. Overall, six out of seven patientswith infantile-HPC were alive in first CR 26-160months (median 104) after diagnosis, with a 5-year OSand EFS of 80% and 85.7%, respectively.

Adult-type HPCTable 2This group included 11 boys and 9 girls. Age rangedbetween 3 and 21 years (median 13); 13 patients were

TABLE 1Infantile Hemangiopericytoma

Pt GenderAge at dx(wks) Site TNM

Size(cm) IRS Surgery

Chemotherapy(CT) Radiotherapy Outcome

1 F 3 trunk T2AN0M0 3 - 5 I complete resection no no alive in 1st CR at 160 mos2 M at birth head-neck T2BN0M0 5 - 10 III incomplete resection refused no local PRO at 2 mos; DOD at 26 mos3 M at birth head-neck T2BN0M0 5 - 10 III incomplete resection;

delayed completesurgery

VAIA no PR to CT; alive in first CR at 158 mos

4 M 2 extremity T1AN0M0 , 3 I complete resection no no alive in first CR at 132 mos5 M 4 pelvis T1BN0M0 5 - 10 III biopsy; delayed complete

surgeryVAIA no PR to CT; alive in first CR at 76 mos

6 M at birth head-neck T1AN1M0 , 3 III incomplete resection VAC no CR to CT; alive in first CR at 28 mos7 M 6 liver T1AN0M0 3 - 5 III incomplete resection;

delayed completesurgery

VAC no PR to CT; alive in first CR at 26 mos

Dx: diagnosis; TNM: Tumor Nodes Metastases; IRS: Intergroup Rhabdomyosarcoma Study; CR: complete remission; PRO: progression; DOD: dead of disease; PR: partial remission; VAIA: Vincristine, Actinomycin-D,

Ifosfamide, Adriamycin; VAC: Vincristine, Actinomycin-D, Cyclophosphamide.

2694 CANCER November 15, 2001 / Volume 92 / Number 10

older than 10 years. The site of origin of the tumor wasthe extremities in nine patients (six lower), the headand neck in five patients, the trunk in three patients,and the abdomen in three patients. According to theTNM system, the tumor was staged as T1A in 4 pa-tients, T1B in 4 patients, and T2B in 12 patients; 1patient had nodal involvement (N1) and 3 had lungmetastases (M1) at diagnosis. Tumor size was smallerthan 5 cm in 4 patients, between 5 and 10 cm in 6patients, and larger than 10 cm in 10 patients.

At diagnosis, complete tumor excision with histo-logically-free margins (IRS Group I) was achieved infour patients, through primary re-excision in one andgross resection with suspected microscopical residualdisease (Group II) in five. Eleven patients received amacroscopically-incomplete resection (eight patientsin Group III, three patients in Group IV); nine patientshad a tumor size larger than 10 cm. Delayed resec-tion was performed in six patients after 10-12 weeksof chemotherapy: five patients had a complete ex-

TABLE 2Adult Type Hemangiopericytoma

Pt SexAge at dx(years) Site TNM

Size(cm) IRS Surgery

Chemotherapy(CT) Radiotherapy Outcome

1 M 21 extremity T1BN0M0 5 - 10 I complete resection VAIA 54 Gy alive in first CR, 46 mosfrom dx

2 M 13 head-neck T1AN0M0 3 - 5 I complete resection VAIA no alive in first CR at 100 mos3 M 9 pelvis T2BN0M0 5 - 10 I complete resection IVA no alive in first CR at 124 mos4 M 14 extremity T1BN0M0 5 - 10 I complete resection

(primary re-excision)VACA 64 Gy local relapse at 62 mos;

DOD at 159 mos5 M 14 extremity T2BN0M0 5 - 10 II incomplete resection VAIA 45 Gy alive in first CR at 38 mos6 F 9 extremity T1AN0M0 3 - 5 II incomplete resection VAIA 35 Gy alive in first CR at 128 mos7 F 20 head-neck T1AN0M0 3 - 5 II incomplete resection

(primary re-excision)VACA 50 Gy alive in first CR at 176 mos

8 M 3 abdomen T2BN0M0 . 10 II incomplete resection,delayed completesurgery

VACA no alive in first CR at 199 mos

9 F 4 head-neck T1BN0M0 5 - 10 II incomplete resection VACA 45 Gy alive in first CR at 276 mos10 F 13 extremity T2BN0M0 . 10 III biopsy no 50 Gy local PRO at 1 mo; DOD at

2 mos11 F 11 extremity T2BN0M0 . 10 III biopsy VACA no no response to CT;

metastatic PRO at 4mos; DOD at 14 mos

12 F 9 head-neck T2BN0M0 . 10 III biopsy; delayed completesurgery

VACA 50 Gy PR to CT; local relapse at16 mos; DOD at 26 mos

13 M 14 trunk T2BN0M0 5 - 10 III incomplete resection;delayed completesurgery

VAIA 40 Gy PR to CT; local relapse at22 mos; DOD at 35 mos

14 F 6 trunk T2BN0M0 . 10 III incomplete resection VACA 42 Gy PR to CT; metastaticrelapse at 29 mos; DODat 78 mos

15 M 14 retro-peritoneum

T2BN0M0 . 10 III biopsy; delayed completesurgery

VACA 54 Gy PR to CT; alive in first CRat 108 mos

16 M 15 trunk T2BN0M0 . 10 III biopsy; delayedincomplete surgery

VACA 52 Gy PR to CT; alive in first CRat 164 mos

17 M 8 head-neck T1AN1M0 3 - 5 III incomplete resection VACA 70 Gy PR to CT; alive in first CRat 234 mos

18 M 18 extremity T1BN0M1 . 10 IV biopsy; delayed completesurgery on T and M

CEVAIE lung 20 Gy no response to CT; alive in1st CR at 24 mos

19 F 20 extremity T2BN0M1 . 10 IV incomplete resection VAIA 54 Gy no response to CT;metastatic PRO at 5mos; DOD at 23 mos

20 F 16 extremity T2BN0M1 . 10 IV incomplete resection CEVAIE 45 Gy 1 lung20 Gy

CR to CT; metastaticrelapse at 22 mos; DODat 29 mos

Dx: diagnosis; TNM: Tumor Nodes Metastases; IRS: Intergroup Rhabdomyosarcoma Study; CR: complte remission; PRO: progression; DOD: dead of disease; PR: partial remission; VAIA: Vincristine, Actinomycin-D,

Ifosfamide, Adriamycin; IVA: Ifosfamide, Vincristine, Actinomycin-D; VACA: Vincristine, Actinomycin-D, Cyclophosphamide, Adriamycin; CEVAIE: Carboplatin, Epiadriamycin, Vincristine, Actinomycin-D,

Ifosfamide, Etoposide.

Childhood Hemangiopericytoma/Ferrari et al. 2695

cision of residual tumor, and one an incompleteresection.

Post-operative local radiotherapy was adminis-tered to 15 patients: 10 children received radiotherapydue to previous incomplete tumor excision and fivedue to large tumor size, although the tumor had beencompletely resected (two at diagnosis, three at delayedre-operation). In two metastatic patients lung irradia-tion was performed.

All but one patient received chemotherapy: VACAregimen in 10 patients, VAIA in 6 patients, CEVAIE in2 patients, IVA in 1 patient. Chemotherapy was notadministered to one girl with a huge invasive tumor ofthe thigh and gluteus, with uncontrollable progressionof disease during radiotherapy; the patient died 2months after diagnosis.

Response to chemotherapy was evaluable in 10patients with measurable disease and was as follow: 1CR, 6 PR, 3 no response. In four patients with tumorsconsidered not completely resectable at diagnosis,chemotherapy induced tumor shrinkage and madesubsequent surgery feasible.

With a follow-up of 24-276 months (median 125),OS and EFS were 68.8% and 64% at 5 years, and 62.6%and 57.6% at 10 years. Twelve patients were alive infirst CR: three of four patients in IRS Group I, five offive patients in Group II, three of eight patients inGroup III, and one patient of three in Group IV.Among the patients with metastases at onset, one wasstill in CR 24 months after the diagnosis. This childreceived chemotherapy without any objective re-sponse, followed by complete resection of the primarytumor in the thigh, then bilateral pulmonary metasta-sectomy and lung irradiation.

Eight patients had tumor progression or relapse,local in four patients, with lung metastases in fourpatients, 1-62 months (median 22) after diagnosis.They all died of disease despite further treatment2-159 months (median 27) after diagnosis. Among theeight relapsing patients, one was T1B and seven wereT2B, six were females, six were over 10 years old, andfive had a tumor of the extremities. None of the pa-tients with a tumor diameter smaller than 5 cm hasrelapsed through the time of writing.

Patients with adult-type HPC were stratified ac-cording to different characteristics and the estimatedOS and EFS were compared in univariate analysis(Table 3). Despite the limited number of patients, astatistically significant influence on prognosis wasseen for IRS group (Group I-II versus Group III), tumorsize, local invasiveness, and gender; p was not signif-icant for age, site, or achievement of complete resec-tion (in primary or re-surgery).

DISCUSSIONOur study represents the largest reported series onHPC in children and adolescents. The rarity of thistype of soft tissue sarcoma in pediatric age accountsfor the few published reports and explains why littleinformation is available on clinical features and man-agement.6 – 8

The most important series on pediatric HPC pub-lished to date has come from St. Jude Children’s Re-search Hospital on 12 children (3 infants) treated overa 35-year period.6 In agreement with that report, ourseries confirms that HPC in pediatric ages comprisestwo different clinical entities. Although it shares cer-tain histologic features with the adult-type, infantile-HPC represents a distinct entity, considered by someto be closely related to infantile myofibroblastic le-sions.14,15 Histologic findings that usually indicate apoor prognosis in HPC of adults – such as increasedmitotic activity and focal necrosis – are often de-scribed,1 but the clinical behavior is typically favor-able. Our data confirm known features of HPC ininfants, such as a male predilection and superficiallocation (subcutaneous tissue of head and neck), aswell as a good response to chemotherapy and favor-able outcome in the majority of patients.16,17 In ourseries, all patients had a good response to chemother-apy and all but one were alive in first CR at the time ofthe report. Unlike other reports, no multifocal lesionor spontaneous regression were observed. In the re-view from St. Jude Children’s Research Hospital, theauthors emphasized the ability of infantile-HPC todifferentiate into more mature tissue; in one of theirpatients, infantile-HPC matured to hemangioma afterchemotherapy.6 In three patients in the current series,surgery was delayed until after chemotherapy, butnone revealed any maturation into benign vasculartumor at histologic evaluation.

Given the high chemoresponsiveness of infantile-HPC lesions and their frequently aggressive and life-

TABLE 3Adult Type Hemangiopericytoma: Univariate Analysis

5-yr EFS (%)

Overall 64Male (n. 11) vs female (n. 9) 90 vs 33.3 P 5 0.0201IRS Group I–II (n. 9) vs Group III (n. 8) 100 vs 37.5 P 5 0.0206Complete surgery (n. 9) vs incomplete surgery

(n. 11) 75 vs 56.3 P 5 0.6506T1 (n. 8) vs T2 (n. 12) 100 vs 41.7 P 5 0.0433, 10 cm (n. 10) vs . 10 cm (n. 10) 90 vs 36 P 5 0.0320, 10 yrs (n. 7) vs . 10 years (n. 13) 71.4 vs 59.8 P 5 0.3509Extremities (n. 9) vs other sites (n. 11) 53.3 vs 72.7 P 5 0.0937

EFS: event free survival; IRS: Intergroup Rhabdomyosarcoma Study.

2696 CANCER November 15, 2001 / Volume 92 / Number 10

endangering presentation, a primary chemotherapyshould be preferred to mutilating surgery or a “waitand see” approach.18 Whenever feasible, surgical ex-cision remains the mainstay of treatment; in patientswho underwent a primary complete resection, chemo-therapy might be avoided. In the current study, thetwo patients with complete excision at diagnosis re-ceived no adjuvant therapy and were alive in first CR.Considering the high chemoresponsiveness of thesetumors, as well as the early age of patients, less toxicchemotherapeutic regimens that limit antracyclinesand/or alkylating-agents might be investigated.

The clinical data from our subset of patients olderthan 1 year confirm that the behavior of adult-typeHPC is similar to cases developing in adulthood. How-ever, the favorable response to chemotherapy in ourcohort was particularly significant and higher thanmost reported adult series: among the 10 childrenwith measurable disease, 70% (six PR and one CR)responded to chemotherapy. In four children withhuge unresectable lesions at diagnosis, tumor shrink-age induced by chemotherapy permitted subsequentsurgery to be performed. Previous reports have docu-mented a significant response rate to chemotherapeu-tic regimens in use for soft tissue sarcomas, includingcombinations of vincristine, actinomycin-D, cyclo-phosphamide, ifosfamide, adriamycin, and dacarba-zine.19 –22 Even if the role of chemotherapy is not en-tirely clear yet, most investigators consider HPC arelatively chemosensitive neoplasm. The rarity of thistumor hinders the accrual of a sufficient number ofpatients for randomized trials to establish the effec-tiveness of chemotherapy. Nevertheless, in the currentseries chemotherapy was used for large unresectablelesions to enable subsequent excision, in the case ofmetastases, or after incomplete tumor excision. Sinceearly microdissemination was suggested by the evi-dence of distant metastases with no local recurrencein a subset of relapsing patients, chemotherapy wasalso given to completely resected patients to controloccult metastatic foci.

The role of adjuvant radiotherapy after incom-plete removal of the tumor in adult-type HPC seemsestablished.9 Different data suggested a benefit of ra-diotherapy in local control both for gross residualtumor and for microscopic incomplete excision.23–25

In the current series, the effectiveness of radiotherapyin controlling microscopic disease is supported by theabsence of relapses in IRS Group II patients. In con-trast, local recurrence within radiation fields was ob-served in three children irradiated after a completetumor resection (one at primary re-excision, two atdelayed surgery).

In our experience, local invasiveness, tumor size,

and gross resectability (IRS Group I-II versus GroupIII) appear to be the most important predictors ofsurvival. The differences in outcome were statisticallysignificant despite the small number of patients. Wealso observed a worse outcome in female patients: theprognostic role of gender observed in our analysis hasnot been reported elsewhere and may be related to thelarge proportion of girls with T2B tumors consideredunresectable at diagnosis.

In our series, pathologic re-evaluation of histo-logic specimens or slides to assess mitotic index, de-gree of cellularity, and necrosis was not possible inmost cases, due to the number of centers in differentcountries involved in the study. However, several au-thors have emphasized the difficulty in predicting thebehavior of this tumor in children; no distinct histo-logic criteria capable of defining the grade of malig-nancy have been identified to date.4,6,7

Local and distant relapses after a prolonged dis-ease free interval have been reported.26 This suggestsa mandatory long-term follow-up is, which must betaken into account when considering survival curvesin the current series. The median follow-up in thecurrent group was 125 months, and the latest relapseoccurred 62 months after diagnosis. Some patientswho are tumor free at time of writing may eventuallyrelapse (three patients had a follow-up of less than 60months).

In conclusion, the current report attempts to con-tribute toward a better definition of the clinical man-agement of this rare pediatric tumor. Childhood HPCincludes two distinct clinical entities in childrenyounger or older than 1 year of age. In infants, HPChas a more favorable behavior, sharing clinical fea-tures with infantile myofibroblastic lesions: it re-sponds well to chemotherapy, which is required incases of unresectable and aggressive life-threateninglesions. Less toxic therapeutic regimens might be in-vestigated in further studies. The clinical behavior ofHPC in children older than 1 year of age does notappear to differ from its adult counterpart. Completesurgical resection remains the mainstay of treatment,but - unlike most other soft tissue sarcomas apartfrom rhabdomyosarcoma – chemotherapy and radio-therapy seem effective and are recommended in allpatients failing to achieve complete resection and/orwith tumor with locally invasive, large tumors.

REFERENCES1. Enzinger FM, Weiss SW. Perivascular tumor. In: Enzinger

FM, Weiss SW, editors. Soft Tissue Tumors. 3rd ed. St Louis:CV Mosby, 1995: 701-33.

2. Stout AP, Murray MR. Hemangiopericytoma: a vascular tu-mor featuring Zimmermann’s pericytes. Ann Surg 1942;116:26-33.

Childhood Hemangiopericytoma/Ferrari et al. 2697

3. Enzinger FM, Smith BH. Hemangiopericytoma: an analysisof 106 cases. Hum Pathol 1976;7:61-82.

4. Coffin CM. Vascular tumors. In: Coffin CM, Dehner LP,O’Shea PA, editors. Pediatric Soft Tissue Tumors. A clinical,pathological, and therapeutic approach. 1st ed. Baltimore:Williams and Wilkins, 1997:40-79.

5. Kauffman SL, Stout AP. Hemangiopericytoma in children.Cancer 1960; 13:695-710.

6. Rodriguez-Galindo C, Ramsey K, Jenkins JJ, Poquette CA,Kaste SC, Merchant TE, et al. Hemangiopericytoma in chil-dren and infants. Cancer 2000; 88:198-204.

7. Kumar R, Corbally M. Childhood hemangiopericytoma. MedPediatr Oncol 1998; 30:294-6.

8. Atkinson JB, Mahour GH, Isaacs H Jr., Ortega JA. Heman-giopericytoma in infants and children. A report of six pa-tients. Am J Surg 1984; 148:372-4.

9. Miser JS, Triche TJ, Kinsella TJ, Pritchard DJ. Other softtissue sarcomas of childhood. In: Pizzo PA, Poplack DC,editors. Principles and Practice of Pediatric Oncology. 3rded. Philadelphia: Lippincott, 1997:865-88.

10. Harmer MH. TNM classification of pediatric tumors. Ge-neva: International Union Against Cancer, 1982:23— 8.

11. Maurer HM, Beltangady M, Gehan EA, Crist W, HammondD, Hays DM, et al. The Intergroup RhabdomyosarcomaStudy I: A final report. Cancer 1988;61:209-20.

12. Kaplan EL, Meier P. Non-parametric estimation from in-complete observations. J Am Stat Assoc 1958; 53:457-481.

13. Conover WJ. Practical nonparametric statistics. New York:Wiley, 1980:153-69.

14. Mentzel T, Calonje E, Nascimento AJ, Fletcher CDM. Infan-tile hemangiopericytoma versus infantile myofibromatosis.Study of a series suggesting a continuous spectrum of in-fantile myofibroblastic lesions. Am J Surg Pathol 1994;18:922-30.

15. Variend S, Bax NMA, van Gorp J. Are infantile myofibromato-sis, congenital fibrosarcoma and congenital haemangioperi-cytoma histogenetically related? Histopathology 1995;26:57-62.

16. del Rosario ML, Saleh A. Preoperative chemotherapy forcongenital hemangiopericytoma and review of the litera-ture. J Pediatr Hematol Oncol 1997;19:247-50.

17. Bailey PV, Weber TR, Tracy TF, O’Connor DM, Sotelo-AvilaC. Congenital hemangiopericytoma: an unusual vascularneoplasm of infancy. Surgery 1993;114:936-4.

18. Toren A, Perlman M, Polak-Charcon S, Avigad I, Katz M,Kuint Y, et al. Congenital hemangiopericytoma / infantilemyofibromatosis: radical surgery versus a conservative “waitand see” approach. Pediatr Hematol Oncol 1997;14:387-93.

19. Ortega JA, Finklestein JZ, Isaacs H Jr., Hittle R, Hastings N.Chemotherapy of malignant hemangiopericytoma of child-hood. Report of a case and review of the literature. Cancer1971;27:730-5.

20. Wong PP, Yagoda A. Chemotherapy of malignant heman-giopericytoma. Cancer 1978;41:1256-60.

21. Beadle GF, Hillcoat BL. Treatment of advanced malignanthemangiopericytoma with combination adriamycin andDTIC: a report of four cases. J Surg Oncol 1983;22:167-70.

22. Celik I, Bascil N, Yalcin S, Gullu IH, Kars A, Barista I, et al.Ifosfamide-based chemotherapy for recurrent or metastatichemangiopericytoma. Acta Oncol 1997;36:348.

23. Staples JJ, Robinson RA, Wen BC, Hussey DH. Hemangio-pericytoma – the role of radiotherapy. Int J Radiat Oncol BiolPhys 1990;19:445-51.

24. Jha N, McNeese M, Barkley HT Jr., Kong J. Does radiother-apy have a role in hemangiopericytoma management? Re-port of 14 new cases and a review of the literature. Int JRadiat Oncol Biol Phys 1987;13:1399-402.

25. Mira JG, Chu FCH, Fortner JG. The role of radiotherapy inthe management of malignant hemangiopericytoma. Reportof eleven new cases and review of the literature. Cancer1997;39:1254-9.

26. Spitz FR, Bouvet M, Pisters PW, Pollock RE, Feig BW. He-mangiopericytoma: a 20-year single-institution experience.Ann Surg Oncol 1998;5:350-5.

2698 CANCER November 15, 2001 / Volume 92 / Number 10