GENERALTHORACIC

Evaluation of Outcome and Prognostic Factorsin Thoracic Primitive Neuroectodermal Tumor:A Study of 84 CasesBivas Biswas, MD, Sandeep Agarwala, MD, Nutan Kumar Shukla, MD,Suryanarayan Deo, MD, Dayanand Sharma, MD, Sanjay Thulkar, MD,Sreenivas Vishnubhatla, MD, and Sameer Bakhshi, MDDepartments of Medical Oncology, Surgical Oncology, Radiotherapy, and Radiology, Dr BR Ambedkar Institute Rotary CancerHospital; and Departments of Pediatric Surgery and Biostatistics, All India Institute of Medical Sciences, New Delhi, India

Background. Data on thoracic primitive neuro-ectodermal tumor (PNET) treated with a uniformchemotherapy protocol are minimal in the literature. Weanalyzed patients with thoracic PNET for outcome andprognostic factors.

Methods. This is a single-institutional data review ofpatients treated between June 2003 and November 2011with uniform neoadjuvant chemotherapy, surgical inter-vention, or radiotherapy (RT), or a combination ofthese treatments as local therapy followed by adjuvantchemotherapy.

Results. Thoracic PNET was found in 84 of 374 (22%)patients with PNET with a median age of 15 years (range,3–40 years); 27 (32%) of these patients had metastases.Thirty patients underwent surgical resection; 27 patientsreceived radical RT after neoadjuvant chemotherapy. Theradical RT group did not have adverse tumor character-istics or poor response to neoadjuvant chemotherapy. Atmedian follow-up of 20.8 months (range, 2–104.6 months),5-year event-free survival (EFS), overall survival (OS),and local control rate (LCR) were 24.4% ± 5.9%, 47.9% ±8.4%, and 59.3% ± 9%, respectively, for the entire cohort,

Accepted for publication June 7, 2013.

Address correspondence to Dr Bakhshi, Department of MedicalOncology, Dr BR Ambedkar Institute Rotary Cancer Hospital, All IndiaInstitute of Medical Sciences, New Delhi-110029, India; e-mail: sambakh@hotmail.com.

� 2013 by The Society of Thoracic SurgeonsPublished by Elsevier Inc

and 31% ± 7.7%, 59% ± 10.4%, and 67% ± 9.7%, respec-tively, for the group with localized tumors. In multivar-iate analysis, symptom duration longer than 4 months(p [ 0.03), primary tumor of skeletal origin (p [ 0.03),and radical RT (p [ 0.006) predicted inferior EFS in theentire cohort and those with localized disease; metastaticdisease (p [ 0.002) predicted inferior OS. Radical RTpredicted inferior LCR in the entire cohort and the groupwith localized tumor; tumor diameter larger than 8 cm(p [ 0.02) and symptom duration longer than 4 months(p [ 0.02) predicted inferior LCR in the group withlocalized tumor.Conclusions. This is a single-institutional experience of

84 patients with thoracic PNETs who underwent a uni-form chemotherapy protocol. Novel prognostic factorswere identified for thoracic PNET. All efforts should bemade to resect primary tumor after neoadjuvant chemo-therapy because radical RT results in inferior EFS andLCRdespite good response to neoadjuvant chemotherapy.

(Ann Thorac Surg 2013;96:2006–14)� 2013 by The Society of Thoracic Surgeons

horacic primitive neuroectodermal tumor (PNET)

Tprobably originates from embryonic cells of theneural crest [1] and is also known as Askin-Rosai tumor[2]. It is the most common pediatric chest wall malignancyand composes 6% to 11% of all PNETs [3–5]. It usuallygrows inside the thoracic cavity and generally presentswith tumors larger than PNET of the extremities. Most ofthe reported studies are part of clinical trials [4–9] orpatients treated over a long period with heterogeneoustreatment protocols [10–12]. Data are not comparablebetween studies, and there is lack of evaluation of prog-nostic factors because of the small number of patients[10–12]. Because chest wall PNET is a rare tumor with a

high rate of local relapse and frequent metastasis atpresentation, we analyzed patients with thoracic PNETwho were treated at our institute over the past 8 years foroutcome and prognostic factors.

Patients and Methods

PatientsThis study involves data review of all patients with aproven diagnosis of PNET who were treated in ourdepartment from June 2003 to November 2011. Specif-ically, PNET of the thorax was selected for this analysis.

Appendix Tables 1 to 4 can be viewed in the onlineversion of this article [http://dx.doi.org/10.1016/j.athoracsur.2013.06.062] on http://www.annalsthoracicsurgery.org.

0003-4975/$36.00http://dx.doi.org/10.1016/j.athoracsur.2013.06.062

Abbreviations and Acronyms

CR = complete responseEFS = event-free survivalLCR = local control rateLFU = lost to follow-upN = number of patientsORR = overall response rateOS = overall survivalPD = progressive diseasePNET = primitive neuroectodermal tumorPR = partial responseSD = stable disease

2007Ann Thorac Surg BISWAS ET AL2013;96:2006–14 PRIMITIVE NEUROECTODERMAL TUMOR OF THORAX

GENERALTHORACIC

Thoracic PNET included PNET arising from rib, sternum,scapula, clavicle, and soft tissue of chest wall and pleura.PNET arising from thoracic vertebrae were excluded fromthis analysis. Patient baseline clinicopathologic features,metastatic workup, treatment modality, treatment-relatedgrade 3/4 toxicities (according to National Cancer Insti-tute Common Terminology Criteria for Adverse Events,version 3.0) [13], and outcome data were collected. Ethicalclearance was approved by the institutional ethics andreview committee. Informed consent was obtained frompatients or guardians (in the case of minors) in patientsenrolled prospectively.

Diagnostic WorkupBiopsy samples (true cut or incisional) of the primarylesion were obtained from all patients and underwentimmunohistochemical analysis. Diagnosis of PNETrequired the presence of small blue round cell tumor andpositivity for CD99 (MIC-2) or synaptophysin or chro-mogranin. Other round cell tumors were ruled out byperforming immunohistochemical analysis with leuko-cyte common antigen, desmin, and myogenin. Evaluationfor translocation [t(11;22)(q24;q11.2-12)] was not per-formed in this cohort. To observe the extent of the locallesion, all patients underwent computed tomography ormagnetic resonance imaging of the thorax. All patientshad a metastatic workup, including a bone scan and abone marrow biopsy.

Treatment and Response EvaluationTreatment protocol consisted of 3 phases: neoadjuvantchemotherapy for 9 to 12 weeks, local therapy, and thenadjuvant chemotherapy. Chemotherapy was given usingvincristine, doxorubicin, actinomycin, cyclophosphamide,ifosfamide, and etoposide [14]; 4 patients received thesedrugs according to the institutional protocol of St. JudeChildren’s Research Hospital [15]. Local therapy includedRT or surgical intervention with or without postoperativeRT. The choice of local therapy was made on an individualbasis, depending on the primary tumor site and resect-ability of the tumor after neoadjuvant chemotherapy, withcare being taken to avoid long-term morbidity anddisfigurement. Operative procedures were preferredwhenever wide local excision was possible with clearsurgical margins and without a mutilating operation

(decision for surgical resection was made by the treatingsurgeon). In general, patients with pathologic completeremission (CR) were not given adjuvant RT, although de-cisions for adjuvant RT were made on an individual case-by-case basis after discussion in the multidisciplinaryclinic. After local therapy, all patients were to receiveadjuvant chemotherapy up to a planned 48 weeks.Disease response was determined by radiologic pro-

cedures after neoadjuvant chemotherapy and completionof local therapy. CR, partial response (PR), stable disease(SD), and progressive disease (PD) were defined accordingto Response Evaluation Criteria in Solid Tumors [16]. Allpatients without metastases underwent local therapy ifthey had CR/PR/SD after neoadjuvant chemotherapy,whereas all patientswithmetastases received local therapyif they had CR/PR at both primary and metastatic sites.

Statistical AnalysisThe c2 test or Fischer’s exact test was used to detect anassociation between qualitative variables. Student’s t testor Wilcoxon’s rank sum test was applied to compare be-tween categorical and continuous variables. Survival wasestimated by the Kaplan-Meier method and comparedusing the log-rank test. Data were censored on November30, 2012. A univariate Cox proportional hazard modelfollowed by stepwise multivariate Cox regression analysiswas done to identify the predictors of outcome. Factorswith significance (p� 0.1) in univariate analysis were takeninto multivariate analysis. Event-free survival (EFS) wascalculated from the date of diagnosis to the date of diseaserelapse or progression or death from any cause. Overallsurvival (OS) was calculated from the date of diagnosis tothe date of death from any cause. Local and distant sitefailures were combined as local failure for analysis of localcontrol rate (LCR). Patients who were lost to follow-up orabandoned treatment were included in EFS and OS ana-lyses, and outcome in these patients was confirmed bytelephone contact. Abandonment of treatment wasincluded for survival analysis in the present study becauseit has been proposed that noncompliant patients should beincluded in survival analyses for studies from developingnations to provide a true picture of outcome from thesecountries [17]. STATA/SE 9.0 (StataCorp LP, College Sta-tion, TX) was used for statistical analysis.

Results

Clinicopathologic ProfileThe clinical findings are summarized in Table 1. Duringthe study, a total of 403 patients with PNET were regis-tered at our center and according to our intention-to-treatanalysis, 374 patients were evaluable for survival analysis.Of the total 403 registered patients, 93 (23%) had thoracicPNET. Nine patients did not receive therapy, so weanalyzed the baseline characteristics and survival of 84 of374 (22%) patients.

TreatmentAll 84 patients received neoadjuvant chemotherapy witha median of 6 cycles (range, 4–7 cycles). Eleven patients

Table 1. Baseline Clinical and Tumor Characteristics of Whole Cohort (N ¼ 84)

Patients Characteristics Number Percentage

Age (median) 15 y (range, 3–40 y)�15 y 44 52>15 y 40 48

SexMale 59 70Female 25 30

SymptomsSwelling 65 77Pain 65 77Cough, dyspnea 14 17

Systemic symptoms 30 36Symptom duration (median) 4 mo (range, 0.4–24 mo)

�4 mo 47 56>4 mo 37 44

Tumor diameter (median), n ¼ 82 9 cm (range, 1.6–20 cm)�8 cm 36 44>8 cm 46 56

Tumor siteRibs 44 54Scapula 21 25Clavicle 6 7Other sitea 13 14

Tumor originSkeletal 71 85Soft tissue 13 15

Pleural effusionNo 58 69Yes 26 31

Hemoglobin (median), n ¼ 77 11.2 g/dL (range, 6.2–15.1 g/dL)�10 g/dL 20 26>10 g/dL 57 74

White blood cell count (median), n ¼ 76 8,500/mL (range, 1,500–26,200/ mL)�11,000/mL 58 76>11,000/mL 18 24

Lactate dehydrogenase (median), n ¼ 66 522 U/L (range, 60–2,450 U/L)�500 U/L (2 � N) 31 47>500 U/L (2 � N) 35 53

Serum albumin (median), n ¼ 73 4.3 g/dL (range, 2.6–5.2 g/dL)�3.5 g/dL 8 11>3.5 g/dL 65 89

Serum alkaline phosphatase (median), n ¼ 75 311 IU/L (range, 63–1,432 IU/L)�480 IU/L 63 84>480 IU/L 12 16

MetastasesNo 57 68Yes 27 32

Local treatmentSurgical resection � RT 30 53RT only 27 47

Response to neoadjuvant chemotherapyCR/PR 66 79SD/PD 18 21

(Continued)

2008 BISWAS ET AL Ann Thorac SurgPRIMITIVE NEUROECTODERMAL TUMOR OF THORAX 2013;96:2006–14

GENERALTHORACIC

Table 1. Continued

Patients Characteristics Number Percentage

RT dose (median) (Gy) 45 (range, 30–60 Gy)�45 28 60>45 19 40

a Chest wall soft tissue ¼ 9 patients; pleural soft tissue ¼ 4 patients.

CR ¼ complete response; PD ¼ progressive disease; PR ¼ partial response; RT ¼ radiotherapy; SD ¼ stable disease.

2009Ann Thorac Surg BISWAS ET AL2013;96:2006–14 PRIMITIVE NEUROECTODERMAL TUMOR OF THORAX

GENERALTHORACIC

abandoned treatment during neoadjuvant chemotherapy(PR, n ¼ 5; SD, n ¼ 1; and not evaluated, n ¼ 5),14 patients experienced PD while receiving neoadjuvantchemotherapy, and 2 patients died of neutropenic sepsis.Thus, 57 patients (68%) received local therapy: Surgicalresection was performed in 30 patients (53%), 20 of whomreceived adjuvant RT, whereas radical RT alone wasadministered in 27 (47%) patients. Of the 30 patients whounderwent surgical resection, R0 resection was achievedin 24 patients. RT was given at a median dose of 55 Gy(range, 50–60 Gy) in 1.8- to 2-Gy/d doses, 5 days a weekover 5 to 6 weeks in the group with localized tumor; amedian dose of RT of 45 Gy was administered to thegroup with metastases. While receiving local treatment, 5patients experienced PD, 1 patient in CR died of toxicity,and 2 patients were lost to follow-up (1 with PR andanother with CR). Thus, the remaining 49 patientsreceived adjuvant chemotherapy (Fig 1).

Treatment ToxicityTwenty-five patients (30%) had treatment-related toxicityas follows: sepsis/febrile neutropenia (n ¼ 22), hemor-rhagic cystitis (n ¼ 1), ifosfamide-induced encephalopa-thy (n ¼ 1), and cardiomyopathy from doxorubicintreatment (n ¼ 1). Three patients died of febrile

neutropenia. There was no recordable grade 3 or 4toxicity with RT. Six patients received a reduced dose ofchemotherapy after toxicity.

Response to Neoadjuvant Chemotherapy and LocalTherapyOf 84 patients, reevaluation after neoadjuvant chemo-therapy was performed in 77 patients (5 abandonedtreatment and 2 died of sepsis) and response was as fol-lows: CR in 13 patients, PR in 48 patients, SD in 2 patients,and PD in 14 patients, with an overall response rate(ORR) of 79% (n ¼ 61 of 77). Of the 77 patients who un-derwent response evaluation after neoadjuvant chemo-therapy, 57 received local treatment; 20 patients did notreceive local treatment (14 patients had PD and 6 patientswere lost to follow-up [5 had PR and 1 had SD]. Of these57 patients, response after local therapy was CR in35 patients, PR in 17 patients, and PD in 5 patients, withan ORR of 91% (n ¼ 52 of 57 patients).

Treatment Failure and OutcomeTreatment failure was observed in 42 patients (50%): 17patients had recurrences after achieving CR and 25 pa-tients experienced PD during or after therapy (14 patientswith PD after neoadjuvant chemotherapy, 5 patients with

Fig 1. Flow diagram showing treatment andoutcome of study patients. (CR ¼ completeresponse; LFU ¼ lost to follow-up;PD ¼ progressive disease; PNET ¼ primitiveneuroectodermal tumor; PR ¼ partialresponse; SD ¼ stable disease.)

Table 2. Univariate Analysis for EFS and OS in the Entire Cohort (N ¼ 84)

Variable Category

EFS OS

5-Year EFSEstimate (%) HR p Value

5-Year OSEstimate (%) HR p Value

Age (y)a �15 (n ¼ 44) 29.9 0.78 44.8>15 (n ¼ 40) 29.9 1.08 68.6 0.67 0.3

Sex Male (n ¼ 59) 17.7 0.07 37.9Female (n ¼ 25) 43.2 0.54 64.9 0.54 0.19

Systemic symptoms No (n ¼ 54) 25.7 0.97 58.1Yes (n ¼ 30) 21.3 0.99 34.7 1.29 0.51

Symptom duration (mo) �4 (n ¼ 47) 38.6 0.1 56>4 (n ¼ 37) 9.4 1.6 50 1.28 0.51

Tumor site Rib/scapula (n ¼ 65) 15.3 0.01 36.8Other site (n ¼ 19) 63.5 0.31 88.2 0.25 0.04

Tumor origin Skeletal (n ¼ 71) 19.1 0.02 40.9Soft tissue (n ¼ 13) 62.3 0.25 91.7 0.17 0.09

Tumor diameter (cm) �8 (n ¼ 36) 35.9 65.7>8 (n ¼ 46) 17.4 1.53 0.15 33.4 1.98 0.09

Pleural effusion No (n ¼ 58) 22.4 44.8Yes (n ¼ 26) 25.8 1.21 0.52 52.1 0.86 0.72

Hemoglobin (g/dL) �10 (n ¼ 20) 19.3 0.26 39>10 (n ¼ 57) 29.8 0.69 47.5 0.66 0.33

White blood cellcount (/mL)

�11,000 (n ¼ 58) 25.4 0.76 39.7>11,000 (n ¼ 18) 33.3 0.9 61.2 0.75 0.57

LDH (U/L) �500 (n ¼ 31) 27.9 0.31 40.9>500 (n ¼ 35) 20.9 1.38 45.4 1.33 0.48

Metastases No (n ¼ 57) 31 0.004 59.3Yes (n ¼ 27) 7.5 2.34 15 4.08 <0.001

Local treatmentb Surgical resection � RT (n ¼ 31) 49.5 0.005 82RT only (n ¼ 27) 11.5 2.76 34.6 4.25 0.02

a Three-year estimated value; b Four-year estimated value.

EFS ¼ event-free survival; HR ¼ hazard ratio; LDH ¼ lactate dehydrogenase; OS ¼ overall survival; RT ¼ radiotherapy.

2010 BISWAS ET AL Ann Thorac SurgPRIMITIVE NEUROECTODERMAL TUMOR OF THORAX 2013;96:2006–14

GENERALTHORACIC

PD after local therapy, and 6 patients with PD during orafter adjuvant chemotherapy). Site of failure was localalone in 15 patients, distant metastases in 18 patients, andcombined failure in 9 patients. Median time to local fail-ure was 15 months (range, 6.2–40.2 months).

At the time of analysis, 25 patients (30%) were alive(CR, n ¼ 22; PR, n ¼ 1; PD, n ¼ 2). Although EFS and LCRcould be assessed in the majority of patients, some pa-tients did not opt for salvage therapy after PD, and thusfinal survival status of 31 (37%) patients could not beassessed for OS (abandoned treatment, n ¼12; PD,n ¼ 19). A total of 28 (33%) patients died during or aftertreatment (PD, n ¼ 20; death from toxicity, n ¼ 3; death athome [cause could not be ascertained], n ¼ 5 ).

After median follow-up of 20.8 months (range, 2–104.6months), 5-year EFS, OS, and LCR were 24.4% � 5.9%,47.9% � 8.4%, and 59.3% � 9%, respectively, for the entirecohort. In those with localized disease (n ¼ 57), 5-yearEFS, OS, and LCR were 31% � 7.7%, 59% � 10.4%, and67% � 9.7%, respectively. The 3-year LCR in the surgicalresection group, the surgical resection plus RT group, andthe radical RT group was 76.2% � 14.8%, 89.4% � 7.1%,and 33.5% � 14.6%, respectively.

Metastatic DiseaseTwenty-seven (32%) patients had metastatic disease atbaseline. Notably, no patient with PNET of the clavicle hadmetastasis at presentation. All patients received neo-adjuvant chemotherapy and achieved an ORR of 70% (PR,19 patients), whereas only 13 patients received local ther-apy (RT only in 9 patients, surgical intervention with orwithout RT in 4 patients). This group had a higher inci-dence of systemic symptoms (52% versus 28%; p ¼ 0.03),larger tumor diameter (10.4 cm versus 8.4 cm; p ¼ 0.02),higher lactate dehydrogenase levels (762U/L versus 513U/L; p ¼ 0.01), and greater use of radical RT as local therapy(69% versus 41%; p ¼ 0.07) compared with patients withlocalized disease. A total of 12 patients were alive at thetime of analysis with 3-year EFS,OS, and LCR rates of 7.5%� 6.9%, 18.4% � 14%, and 20.2% � 17.8%, respectively.

Univariate Analysis for Prognostic FactorsENTIRE COHORT. On univariate analysis, a rib/scapula pri-mary tumor (p ¼ 0.01), a tumor of skeletal origin (p ¼0.02), metastatic disease (p ¼ 0.004), and radical RT aslocal treatment predicted adverse EFS. Radical RT (p ¼0.02), a rib/scapula primary tumor (p ¼ 0.04), and

2011Ann Thorac Surg BISWAS ET AL2013;96:2006–14 PRIMITIVE NEUROECTODERMAL TUMOR OF THORAX

GENERALTHORACIC

metastatic disease (p < 0.001) were found to predict anadverse outcome for OS, whereas skeletal origin (p ¼0.09), and tumor diameter larger than 8 cm (p ¼ 0.09)showed a trend toward inferior OS (Table 2). Metastaticdisease (p ¼ 0.05) and radical RT (p ¼ 0.02) adverselyaffected LCR (Appendix Table 1).LOCALIZED DISEASE. On univariate analysis, symptomduration longer than 4 months (p ¼ 0.012), a rib/scapulaprimary site (p ¼ 0.02), and radiation alone as localtherapy (p ¼ 0.007) predicted adverse EFS. No factorsignificantly predicted OS (Appendix Table 2). Tumordiameter larger than 8 cm (p ¼ 0.023) and radical RT aslocal treatment (p ¼ 0.03) predicted an adverse LCR,whereas symptom duration longer than 4 months showeda trend toward inferior LCR (p ¼ 0.053) (AppendixTable 1).

Multivariate AnalysisENTIRE COHORT. On multivariate analysis, symptom dura-tion longer than 4 months (p ¼ 0.03), skeletal origin(p ¼ 0.03), and radical RT (p ¼ 0.006) emerged as inde-pendent prognostic factors and predicted inferior EFS(Figs 2A–2C). For OS, metastatic disease (p ¼ 0.002)independently predicted inferior outcome (Fig 2D).Radical RT (p ¼ 0.02) independently predicted lower LCRon multivariate analysis (Table 3) (Fig 2E).LOCALIZED GROUP. On multivariate analysis, symptomduration longer than 4 months (p ¼ 0.001), skeletal origin(p ¼ 0.013), and radical RT (p ¼ 0.002) emerged as

Fig 2. (A) Three-year event-free survival (EFS) of total cohort (N ¼ 84) wversus 11.5% � 7.5% for radical RT group. (B) Five-year EFS was 38.6%9.4% � 6% for symptom duration longer than 4 months in whole cohort (N ¼skeletal origin versus 62.3% � 19.8% for patients with tumor of soft tissue o16.5% � 7.2% for surgical resection � RT group versus 66.5% � 14.6% fosurvival (OS) was 66.4% � 8.6% for group with localized tumor versus 18.year OS was 57% � 9.8% for group with localized tumor versus 13.2% � 1status could not be assessed) (n ¼ 53).

independent prognostic factors and predicted inferiorEFS (Table 3) (Fig 3A–3C). Tumor diameter larger than 8cm (p ¼ 0.02), radical RT as local treatment (p ¼ 0.04), andsymptom duration longer than 4 months (p ¼ 0.02)emerged as independent prognostic factors for LCR withinferior outcome (Table 3) (Figs 3D–3F).

Analysis After Exclusion of Patients Whose FinalSurvival Status Could Not Be Assessed for OSAfter excluding the patients whose final survival statuscould not be assessed (n ¼ 31), metastatic disease atpresentation (p ¼ 0.01) was found to be an independentprognostic factor on multivariate analysis, with worseoutcomes for OS for the entire cohort (Table 3) (Fig 2F).When we excluded those patients from the group withlocalized tumor (n ¼ 23), none of the factors were found toinfluence OS.

Comment

Approximately 17% of all malignant chest wall tumors arePNETs, and as many as 15% of them are localized to thechest wall [18]. The incidence of chest wall PNET in ourcohort was 22%, compared with 11.3% found by Sham-berger and colleagues [5] and 11% found by Schuck andcolleagues [19]. Patients in most of the reported studies ofchest wall PNET were treated with different chemo-therapy protocols and over long time spans [10–12], sothe results are not comparable. We have discussed 84patients with chest wall PNET who were treated with a

as 55.7% � 10.2% for surgical resection � radiotherapy (RT) group� 8.3% for patients with symptom duration 4 months or less versus84). (C) Five-year EFS was 19.1%� 5.7% for patients with tumor ofrigin in whole cohort (N ¼ 84). (D) Three-year local failure rate wasr radical RT group in whole cohort (n ¼ 57). (E) Three-year overall4% � 14% for metastatic group in whole cohort (N ¼ 84). (F) Three-0.6% for metastatic group (after exclusion of patients whose survival

Table 3. Multivariate Analysis

Variable Category N

Entire Cohort

Localized Group(n ¼ 57)

All Patients(N ¼ 84)

After ExcludingPatients With UnknownStatus for OS (n ¼ 53)

HR p Value HR p Value N HR p Value

A: Event-free survivalTumor origin Skeletal 71 . 47

soft tissue 13 0.2 0.03 10 0.07 0.013Symptom duration �4 mo 47 . 33

>4 mo 37 2.25 0.03 24 4.94 0.001Local treatment Surgical

resection � RT30 . 26

Radical RT 27 2.97 0.006 18 4.61 0.002B. Local control rate

Symptom duration �4 mo . . . 25>4 mo 19 5.4 0.02

Tumor diameter (cm) �8 . . . 20>8 . 23 6.52 0.02

Local treatment Surgicalresection � RT

30 . . 26

Radical RT 27 3.7 0.02 18 4.25 0.043C. Overall survival

Metastatic disease No 57 .

Yes 27 5.44 0.002 4 0.01

HR ¼ hazard ratio; RT ¼ radiotherapy.

2012 BISWAS ET AL Ann Thorac SurgPRIMITIVE NEUROECTODERMAL TUMOR OF THORAX 2013;96:2006–14

GENERALTHORACIC

uniform treatment protocol. Notably, our cohort patientswith metastases composed 32% of our cohort, which isrelatively higher than in the studies of Laskar and co-workers [20] and Saenz and associates [11] but wascomparable to other studies [10, 12]. This difference mayresult from referral bias.

Our data showed that patients who received radical RTas local treatment had significantly inferior EFS and LCRcompared with patients who underwent surgical resec-tion, whether data were analyzed with or without patientswith metastases at baseline. The radical RT grouphad a borderline higher incidence of metastatic disease(33% versus 13%; p ¼ 0.07) compared with patientswho underwent surgical resection, which may have beenresponsible for the adverse outcome. There was no dif-ference in clinical characteristics between the 2 groupsother than the tumor site (Appendix Table 3). Further, inthe radical RT group, response after neoadjuvantchemotherapy was not significantly different from thesurgical resection group (p ¼ 0.22). Although this is not arandomized study to assess the role of RT versus surgicalintervention as primary treatment, the analysis suggeststhat those receiving radical RT were not necessarily poorresponders and did not have any adverse baseline tumorcharacteristics, yet the group receiving radical RT hadworse EFS and LCR. Thus, even if there is good responseto neoadjuvant chemotherapy in thoracic PNET, anattempt should be made to perform surgical resection.

We found no difference in 5-year EFS and OS betweenpatients receiving adjuvant RT and those not receiving

adjuvant RT after undergoing surgical resection afterneoadjuvant chemotherapy; EFS was 50% in both groups(p ¼ 0.98) and OS was 78.8% and 84.2%, respectively (p ¼0.66).Among thosewith localizeddisease, of the 26patientswho underwent surgical resection, the group receivingadjuvant RT had a relatively higher number of patientswith residual tumor after neoadjuvant chemotherapyversus the group that did not receive adjuvant RT (13 of 19versus 2 of 7; p¼ 0.07). It is difficult to discern the role of RTin such situations with conviction; however, these limiteddata suggest that RT can perhaps be avoided in patients inwhomno residual tumor is found in the resected specimen.Large tumor size emerged as an independent prog-

nostic factor for LCR in contrast to previous studies inwhich it had no significance [12, 20]. In our study, 15% oftumors of soft tissue origin were in the thorax, which hasnot been evaluated at this site in previous studies. Inter-estingly, soft tissue PNET in this region emerged as afavorable prognostic factor for EFS.In the entire cohort, as well as in the group with

localized tumor, longer symptom duration (>4 months)predicted worse EFS in this site of PNET. Notably, therewas no association of duration of symptoms with tumorsize (p ¼ 0.93) and metastases (p ¼ 0.6).Survival in our cohort was inferior compared with

other large series of chest wall PNET reported in theliterature (Appendix Table 4). This inferior result isexplained by a higher burden of metastases and delayedpresentation in our cohort as well as a resource-challenged setting.

Fig 3. (A) Three-year event-free survival (EFS) of group with localized tumor (n ¼ 57) was 61% � 11% for surgical resection � radiotherapy (RT)group versus 16.5% � 10.3% for radical RT group. (B) Five-year EFS was 53.6% � 10.5% for symptom duration 4 months or less versus 22% �9.3% for symptom duration 4 months or more in group with localized tumor (n ¼ 57). (C) Five-year EFS was 24.6% � 7.8% for patients withtumor of skeletal origin versus 67.5% � 20.7% for patients with tumors of soft tissue origin in group with localized tumor (n ¼ 57). (D) Five-yearlocal failure rate was 17.2% � 9.4% for patients with symptoms 4 months or longer versus 58.75 � 19% for patients with symptoms 4 months orlonger in the group with localized tumor (n¼ 44). (E) Five-year local failure rate was 10.6%� 7% for patients with tumor diameter 8 cm or smallerversus 59.8% � 15.8% for patients with tumor diameter 8 cm or larger in the group with localized tumor (n ¼ 44). (F) Three-year local failure ratewas 8.9% � 6% for surgical resection � radiotherapy (RT) versus 56.3% � 16.5% for radical RT in group with localized tumor (n ¼ 44).

2013Ann Thorac Surg BISWAS ET AL2013;96:2006–14 PRIMITIVE NEUROECTODERMAL TUMOR OF THORAX

GENERALTHORACIC

The strength of our study is that we evaluated all pa-tients with intention-to-treat analysis, and all received auniform chemotherapy protocol. Metastatic disease wasfound to be an independent poor prognostic factor for OSin the entire cohort and continued to maintain its signif-icance after exclusion of patients whose survival status forOS could not be assessed.

The limitations of our study include the lack of study oftranslocation (EWS-FLI) in our patients at diagnosis, eventhough all patients were evaluated with an extensiveimmunohistochemistry panel. Further, approximately14% (n ¼ 12) of our patients abandoned treatment duringthe study, and the final survival status could be notascertained for these patients.

In conclusion, thoracic PNET was found in 22% of theentire cohort of patients with PNET who underwenttherapy, with 32% of patients presenting with metastaticdisease. Patients with tumors at this site who presentedwith symptom duration longer than 4 months and askeletal primary tumor had inferior EFS on multivariateanalysis and may merit more aggressive therapy. Besidessymptom duration, tumor diameter larger than 8 cmpredicted inferior LCR. Further, our data suggest that allpossible efforts should be made to resect the primarytumor after neoadjuvant chemotherapy and that radicalRT alone, despite good response to neoadjuvant chemo-therapy, results in inferior EFS and poor LCR. Perhaps RTis not required after resection if there is no residual tu-mor, although larger studies are required to definitivelyaddress this issue.

References

1. O’Regan S, Diebler MF, Meunier FM, Vyas S. A Ewing’ssarcoma cell line showing some, but not all, of the traits of acholinergic neuron. J Neurochem 1995;64:69–76.

2. Askin FB, Rosai J, Sibley RK, Dehner LP, McAlister WH.Malignant small cell tumor of the thoracopulmonary regionin childhood: a distinctive clinicopathologic entity of uncer-tain histogenesis. Cancer 1979;43:2438–51.

3. Cotterill SJ, Ahrens S, Paulussen M, et al. Prognostic factorsin Ewing’s tumor of bone: analysis of 975 patients from theEuropean Intergroup Cooperative Ewing’s Sarcoma StudyGroup. J Clin Oncol 2000;18:3108–14.

4. Thomas PR, Foulkes MA, Gilula LA, et al. Primary Ewing’ssarcoma of the ribs. A report from the intergroup Ewing’ssarcoma study. Cancer 1983;51:1021–7.

5. Shamberger RC, LaQuaglia MP, Gebhardt MC, et al. Ewingsarcoma/primitive neuroectodermal tumor of the chest wall:impact of initial versus delayed resection on tumor margins,survival, and use of radiation therapy. Ann Surg 2003;238:563–7; discussion 567–8.

6. Shamberger RC, Laquaglia MP, Krailo MD, et al. Ewingsarcoma of the rib: results of an intergroup study withanalysis of outcome by timing of resection. J Thorac Car-diovasc Surg 2000;119:1154–61.

7. Schuck A, Ahrens S, Konarzewska A, et al. Hemithoraxirradiation for Ewing tumors of the chest wall. Int. J RadiatOncol Biol Phys 2002;54:830–8.

8. Schuck A, Hofmann J, R€ube C, et al. Radiotherapy in Ewing’ssarcoma and PNET of the chest wall: results of the trialsCESS 81, CESS 86 and EICESS 92. Int J Radiat Oncol BiolPhys 1998;42:1001–6.

9. Nesbit ME Jr, Gehan EA, Burgert EO Jr, et al. Multimodaltherapy for the management of primary, nonmetastaticEwing’s sarcoma of bone: a long-term follow-up of the FirstIntergroup study. J Clin Oncol 1990;8:1664–74.

2014 BISWAS ET AL Ann Thorac SurgPRIMITIVE NEUROECTODERMAL TUMOR OF THORAX 2013;96:2006–14

GENERALTHORACIC

10. Denbo JW, Shannon OrrW,Wu Y, et al. Timing of surgery andthe role of adjuvant RT in Ewing sarcoma of the chest wall: asingle-institution experience. Ann Surg Oncol 2012;19:3809–15.

11. Saenz NC, Hass DJ, Meyers P, et al. Pediatric chest wallEwing’s sarcoma. J Pediatr Surg 2000;35:550–5.

12. Indelicato DJ, Keole SR, Lagmay JP, et al. Chest wall Ewingsarcoma family of tumors: long-term outcomes. Int J RadiatOncol Biol Phys 2011;81:158–66.

13. National Cancer Institute, Cancer Therapy Evaluation Program.Common Terminology Criteria for Adverse Events v3.0. Avail-ableat: http://ctep.cancer.gov/protocolDevelopment/electronic_applications/docs/ctcaev3.pdf. Accessed August 30, 2013.

14. Grier HE, Krailo MD, Tarbell NJ, et al. Addition of ifosfamideand etoposide to standard chemotherapy for Ewing’s sar-coma and primitive neuroectodermal tumor of bone. N EnglJ Med 2003;348:694–701.

15. Gururangan S, Marina NM, Luo X, et al. Treatment of chil-dren with peripheral primitive neuroectodermal tumor orextraosseous Ewing’s tumor with Ewing’s-directed therapy.J Pediatr Hematol Oncol 1998;20:55–61.

New Website and Online Jofor The Annals of Thoracic S

The Annals of Thoracic Surgery has a new onlinehome: annalsthoracicsurgery.org. CTSNet no longerhosts journals.Online Journal CME is available on the new site via:

annalsthoracicsurgery.org/cme/home, where 96 CME ac-tivities are available to be taken at any time.If you completed any activities on our previous CME

site on CTSNet, your course history was migrated andmerged onto the new site.

HOW TO REGISTER AND CLAIM YOUR ONLINEACCESS ON THE NEW SITEAll subscribers are encouraged to register and claimtheir online access.REGISTRATION IS REQUIRED AS OF AUGUST 1,

2013.

1. Go to annalsthoracicsurgery.org and click “Register” inthe banner at the top right, and then choose “Registerand Activate Your Subscription.”

2. Enter your e-mail address in the top field. (This will beyour username going forward.) Click the button to theleft of “Register anAccount,” and then click “Continue.”

3. At the Register a New Account page, create a pass-word and fill out the requested profile information.Select that you have read the terms of use and click“Register.”

4. On the confirmation page, click the link near thebottom that says “Claim access to full-text articles.”

5. On the “Claim Your Online Access” page:

(a) if you are a member of The Society of ThoracicSurgeons (STS):

� 2013 by The Society of Thoracic SurgeonsPublished by Elsevier Inc

16. Therasse P, Arbuck SG, Eisenhauer EA, et al. Newguidelines to evaluate the response to treatment in solidtumors. European Organization for Research and Treatmentof Cancer, National Cancer Institute of the United States,National Cancer Institute of Canada. J Natl Cancer Inst2000;92:205–16.

17. Mostert S, Arora RS, Arreola M, et al. Abandonment oftreatment for childhood cancer: position statement of aSIOP PODC Working Group. Lancet Oncol 2011;12:719–20.

18. Burt M. Primary malignant tumors of the chest wall. TheMemorial Sloan-Kettering Cancer Center experience. ChestSurg Clin N Am 1994;4:137–54.

19. Schuck A, Ahrens S, Paulussen M, et al. Local therapy inlocalized Ewing tumors: results of 1058 patients treated in theCESS 81, CESS 86, and EICESS 92 trials. Int J Radiat OncolBiol Phys 2003;55:168–77.

20. Laskar S, Nair C, Mallik S, et al. Prognostic factors andoutcome in Askin-Rosai tumor: a review of 104 patients. Int JRadiat Oncol Biol Phys 2011;79:202–7.

urnal CME Siteurgery

–click “I receive my subscription through a societymembership.”–Select the STS option shown in the dropdownmenu.–Enter your 6-digit STS membership ID number andlast name, then click “Claim.” (If you need assistancewith your STS membership ID number, contactSarah Foreman, Member Services Coordinator, atmembership@sts.org.)

(b) if you are a non-member subscriber:

–click “I have purchased a personal subscription.”–Enter the Account Number from the mailing labelof your print issue, then click “Claim.” (If youneed assistance with your Account Number, seeElsevier Customer Support contact informationbelow.)

6. Click either the journal link or “Finish.” Both willdirect you back to the journal homepage, which willnow show you as logged in.

Still have questions? Subscribers in the US, Canada,and all other non-European countries should contactElsevier Customer Support at:

� JournalsOnlineSupport-usa@elsevier.com� (800) 654-2452 (toll-free in the US & Canada)� (314) 447-8871 (outside US & Canada)

Europe-based subscribers should contact:

� eurosupport@elsevier.com� +44 (0) 1865-843177

Ann Thorac Surg 2013;96:2014 � 0003-4975/$36.00