PAC

I

(attlohg

Biology of Blood and Marrow Transplantation 11:903-911 (2005)� 2005 American Society for Blood and Marrow Transplantation1083-8791/05/1111-0007$30.00/0doi:10.1016/j.bbmt.2005.07.010

B

aclitaxel-Based High-Dose Chemotherapy withutologous Stem Cell Rescue for Relapsed Germell Cancer

Kim A. Margolin,1 James H. Doroshow,1 Paul Frankel,2 Warren Chow,1 Lucille A. Leong,1 Dean Lim,1

Mark McNamara,1 Robert J. Morgan,1 Stephen Shibata,1 George Somlo,1 Przemyslaw Twardowski,1

Yun Yen,1 Neil Kogut,1 Jeffrey Schriber,1 Joseph Alvarnas,1 Susan Stalter1

1Departments of Medical Oncology and 2Biostatistics, City of Hope National Medical Center, Duarte, California

�

Correspondence and reprint requests: Kim Margolin, MD, Department of Medical Oncology, City of HopeNational Cancer Research Center, 1500 E. Duarte Rd., Duarte, CA 91010 (e-mail: kmargolin@coh.org).

Received January 5, 2005; accepted July 14, 2005

ABSTRACTWe evaluated the antitumor activity of tandem cycles of high-dose chemotherapy with autologous peripheralstem cell transplantation (aPSCT) in relapsed germ cell tumors by using high-dose paclitaxel, carboplatin,etoposide, and ifosfamide. Thirty-three patients were entered, and 31 underwent protocol therapy. Paclitaxel350 mg/m2 (5 patients) or 425 mg/m2 (26 patients) by 24-hour continuous intravenous infusion was followedby 3 daily doses of carboplatin and either etoposide (cycle 1) or ifosfamide/mesna (cycle 2). The carboplatindose had a calculated area under the curve of 7 mg-min/mL, and the daily dose of etoposide was 20 mg/kg(cycle 1). Ifosfamide 3 g/m2/d for 3 days (with mesna uroprotection) was substituted for etoposide in cycle 2.Each cycle was supported by granulocyte colony-stimulating factor–mobilized peripheral blood stem cells.Thirty-one patients were evaluable for response, toxicity, and long-term disease control. Two patients did notundergo aPSCT because of rapid disease progression. Nineteen patients received both cycles of aPSCT, 8progressed after cycle 1, 3 refused the second cycle, and 1 died of fungal infection during cycle 1. Twelvepatients remain relapse free at a median of 67 months from the initiation of therapy. Whereas the InternationalGerm Cell Cancer Collaborative Group category at the time of initial diagnosis did not seem to predictoutcome, the patient’s probability of achieving durable remission was significantly associated with the Beyerprognostic score at the time of protocol entry. Regimens containing the most active agents in relapsednonseminomatous germ cell tumors, including high-dose paclitaxel, are well tolerated and have promisingactivity even in patients with poor-risk features who do not achieve durable remissions with standard therapy.The Beyer prognostic system is a valuable predictor for patients undergoing aPSCT.© 2005 American Society for Blood and Marrow Transplantation

KEY WORDSHigh-dose chemotherapy ● Autologous peripheral stem cell support ● Germ cell tumors

● Paclitaxel

ffitumtom

t

NTRODUCTION

The treatment of advanced germ cell tumorsGCTs) with cisplatin-based combination chemother-py with or without surgery results in a cure for 70%o 80% of unselected patients. Approximately half ofhe patients who do not achieve remission with first-ine therapy have a complete marker response to sec-nd-line therapy at standard doses; however, less thanalf of these patients are cured [1]. Although autolo-

ous peripheral stem cell transplantation (aPSCT) is C

B & M T

requently used in the United States for patients inrst or subsequent relapse [2,3], the precise indica-ions for this modality remain ill defined. Importantnanswered questions include identification of theost active agents, the selection of patients most likely

o benefit from aPSCT, and the potential superiorityf tandem cycles over single cycles of high-dose che-otherapy.

The prognostic system developed by the Interna-ional Germ Cell Cancer Collaborative Group (IGC-

CG) [4] provides important information regarding

903

iaiftphhctrltlcsprti

tcnsraticcip[ourtnmncdWhdaetmw

hcped

aptnipijsia“(

P

P

a�aebripdtcttstnt

nPcma�ntsbpscecthae

K. A. Margolin et al.

9

nitial characteristics that have been widely acceptednd applied in the design of phase II and III trials fornitial therapy of GCT. To date, a comparable systemor patients in relapse or with an incomplete responseo initial chemotherapy has not been routinely ap-lied. A series of phase II studies in the United Statesave led to the widespread use of tandem cycles ofigh-dose carboplatin and etoposide with or withoutyclophosphamide or ifosfamide (see review [3]), buthis approach has not yet been evaluated in a phase IIIandomized trial for patients with refractory or re-apsed GCT. It is common in the United States toreat patients in a “favorable” relapse (low-volume,ow-marker disease after a complete response to first-linehemotherapy) with second-line therapy consisting oftandard doses of cisplatin-based chemotherapy, whereasatients with less favorable characteristics or subsequentelapse are more often referred for aPSCT, by whichime the durable remission rate from tandem aPSCTs only 20% to 40% [3].

Because GCT is typically a disease of young pa-ients with favorable organ function, an importantonsideration is to develop regimens that incorporateewer agents with proven activity that have acceptableafety and toxicity profiles. On the basis of a series ofecent studies documenting the activity of paclitaxel assingle agent [5] in GCT and our own studies and

hose of others regarding dose escalation of paclitaxeln aPSCT [6-8], we designed a regimen of tandemycles of aPSCT that incorporated the 2 most activelasses of agent (paclitaxel and carboplatin) while stillncluding etoposide, ifosfamide, and carboplatin, as wereviously evaluated in patients with advanced GCT9]. Because we proposed to administer tandem cyclesf aPSCT, each containing high-dose paclitaxel, wesed data from our own experience and others’ expe-iences regarding the dose-limiting toxicity of pacli-axel with aPSCT; this toxicity consists of peripheraleuropathy at the maximum tolerated dose of 775g/m2 administered as a 24-hour continuous intrave-

ous (IV) infusion after doxorubicin and high-doseyclophosphamide [7] or in combination with high-ose cisplatin and high-dose cyclophosphamide [8].

hen we combined escalating doses of paclitaxel withigh-dose ifosfamide, carboplatin, and etoposide,ose-limiting stomatitis and diarrhea were observed atpaclitaxel dose of 275 mg/m2 over 24 hours. When

toposide was omitted from the regimen, the pacli-axel could be escalated to a maximum dose of 575g/m2 over 24 hours, and the dose-limiting toxicityas neuropathy [8].

With these considerations about the safety ofigh-dose paclitaxel and our intent to use the drug inombination with carboplatin (both cycles) and eto-oside (1 cycle), we incorporated an abbreviated dosescalation of paclitaxel for both cycles. The starting

ose of paclitaxel for each cycle was 350 mg/m2, and 2 i

04

dditional dose levels, 425 and 500 mg/m2, werelanned, with 5 patients in each cohort to be observedhrough the first cycle before dose escalation in theext cohort of patients. The rationale for administer-

ng tandem cycles of high-dose therapy was a generalrinciple of chemotherapy: a single cycle would benadequate to provide optimal cell kill. The specificustifications for tandem cycles of the regimen wetudied were an effort to avoid the severe neurotoxic-ty related to a single high dose of paclitaxel [10] and

desire to combine the paclitaxel and carboplatinbackbone” with a different third drug in each cycleetoposide in cycle 1 and ifosfamide in cycle 2).

ATIENTS AND METHODS

atients

To be eligible for the trial, patients older than thege of 15 years with a Karnofsky performance status70% were required to have biopsy-proven, measur-

ble (by radiographic study or serum tumor markerlevation), and relapsed or refractory GCT judged toe incurable by standard salvage therapy; patients inelapse who had intermediate- or high-risk disease atnitial diagnosis [4] were included. Because of theirarticularly unfavorable prognosis, patients with me-iastinal primary GCT in first relapse and those pa-ients who had received prior paclitaxel (maximumumulative exposure of �600 mg/m2) as third-linereatment were required to be responsive to salvageherapy. Patients with a history of central nervousystem metastases were required to have completedherapy with surgery, radiation, or both and to beeurologically stable off of corticosteroids before pro-ocol enrollment.

There was no limit on prior cisplatin; peripheraleuropathy was not considered an exclusion criterion.atients were required to have a calculated creatininelearance of �70 mL/min before cycle 1 and �60L/min before cycle 2. Before stem cell collections

nd each cycle of aPSCT, serum bilirubin had to be1.6 mg/dL, and aspartate aminotransferase and ala-

ine aminotransferase had to be �2 times the institu-ional upper limit of normal. Patients with positiveerology for hepatitis B or C were to undergo liveriopsy to rule out chronic active hepatitis or cirrhosis;atients with positive human immunodeficiency viruserology were excluded. Before beginning granulocyteolony-stimulating factor (G-CSF)–mobilized periph-ral stem cell collections and before each treatmentycle, patients were required to have an absolute neu-rophil count �1500/�L, platelets �120 000/�L, andemoglobin �10 g/dL. Patients were required to have

normal radionuclide-determined left ventricularjection fraction and no evidence of arrhythmias or

schemia on electrocardiogram. Requirements for pul-

mculitcct

T

1ticw�

cPdb2pIc3rc7edhte

ogatmdvtlt

owg�thfM

dahhphmwu

stoipppe5Cmasop�etcwtmPtadhittpp

cetawcnsngmcr

Paclitaxel-Based Tandem Autologous Stem Cell Transplantations in Germ Cell Cancer

B

onary function included a room air arterial oxygenoncentration �70 mm Hg, a forced expiratory vol-me in 1 second of �2.0 L or �75% of the predicted

ower limit of normal, and no history of bleomycin-nduced pulmonary toxicity. All patients providedheir voluntary, written informed consent after proto-ol approval by the institutional review board, in ac-ordance with an assurance filed with and approved byhe Department of Health and Human Services.

reatment Protocol

Stem cell mobilization. All patients received G-CSF0 �g/kg/d subcutaneously for 4 days before the ini-iation of leukapheresis, followed by daily G-CSF dur-ng collections until the total collected product ex-eeded 4 � 106 CD34� cells per kilogram. G-CSFas withheld if the total white blood count was80 000/�L.

Chemotherapy regimens. Cycle 1 chemotherapyonsisted of paclitaxel, etoposide, and carboplatin.remedication for paclitaxel consisted of 2 doses ofexamethasone 20 mg orally given at 12 and 6 hoursefore the initiation of paclitaxel, which was given as a4 hour continuous IV infusion at 350 mg/m2 (first 5atients) or 425 mg/m2 (all other patients) on day �7.mmediately before the infusion, patients also re-eived diphenhydramine 50 mg orally and cimetidine00 mg orally. On days �6, �5, and �4, each patienteceived etoposide 20 mg/kg IV over 2 hours andarboplatin at a calculated area under the curve [11] ofmg-min/mL IV over 30 minutes. Premedication for

toposide consisted of hydrocortisone 50 mg IV andiphenhydramine 50 mg IV, which were repeatedalfway through the etoposide infusion. Premedica-ion for carboplatin included dexamethasone, loraz-pam, and a 5-hydroxytryptamine-3 antiemetic agent.

Patients with increasing serum tumor markers orther clinical or radiographic evidence of tumor pro-ression before cycle 2 were taken off protocol ther-py. Otherwise, all patients who had recovered fromhe acute toxicities associated with cycle 1 were per-itted to begin cycle 2 at 2 to 4 weeks after hospital

ischarge. All of the pretreatment tests detailed pre-iously were required before cycle 2, with the excep-ion of brain magnetic resonance imaging, viral sero-ogic tests, and cardiac and pulmonary functionesting.

Cycle 2 chemotherapy consisted of paclitaxel, if-sfamide, and carboplatin. Paclitaxel and carboplatinere administered exactly as in cycle 1. Ifosfamide 3/m2 was given daily IV over 30 minutes on days �6,5, and �4. Allopurinol 300 mg orally was adminis-

ered daily on day �7 through day 0, and alkalineydration and diuresis were established before ifos-amide and continued throughout its administration.

esna was given as an initial 1 g/m2 IV bolus imme- m

B & M T

iately before the first dose of ifosfamide, followed by10 g/m2 continuous IV infusion over 72 hours (24

ours beyond the final ifosfamide dose). All patientsad a urinalysis daily on days �6 through �3. If theatient developed hematuria (�50 red blood cells perigh-power field), ifosfamide was to be withheld,esna was to be continued, and additional hydrationas to be provided; ifosfamide was to be resumedpon the resolution of hematuria.

Stem cell reinfusion and supportive care. Cell infu-ions and protocol-specific supportive care were iden-ical for the 2 cycles. Because we had previously dem-nstrated earlier granulocyte recovery by splitting thenfusion of autologous peripheral stem cells [12], allatients received 12.5% of the total CD34� stem cellroduct on day �2 of each cycle and 37.5% of theroduct on day 0 of each cycle. Premedication forach stem cell infusion consisted of diphenhydramine0 mg IV and acetaminophen 650 or 1000 mg orally.ontinuous oximetry and frequent vital sign measure-ents were required during and for at least 4 hours

fter each infusion, and symptom management con-isted of meperidine for chills and oxygen for hypoxiar dyspnea in association with the cell infusions. Allatients received G-CSF 5 �g/kg/d IV starting on day2 and continuing until the absolute neutrophil count

xceeded 1000/�L for 3 consecutive days. Plateletransfusion support was provided to keep the plateletount �20 000/�L, and red blood cell transfusionsere provided to keep the hemoglobin �9 g/dL. An-

imicrobial prophylaxis consisted of levofloxacin 500g/d orally or IV and amphotericin B 10 mg/d IV.atients who became febrile were treated with addi-

ional broad-spectrum antibiotics, and antimicrobialgents were adjusted as indicated by the clinical con-ition and culture results. All patients with positiveerpes simplex serology received acyclovir 250 mg/m2

ntravenously every 8 hours during the period of neu-ropenia and until the resolution of stomatitis. Pa-ients were treated in single rooms equipped withositive-pressure air flow during the period of neutro-enia.

Clinical and data monitoring. From the initiation ofhemotherapy until recovery of blood counts and ad-quate oral intake, all patients had daily laboratoryesting consisting of a complete blood cell count (with

differential white blood cell count when the totalhite blood cell count was �500/�L) and a serum

hemistry panel that included electrolytes, blood ureaitrogen, creatinine, and glucose. A comprehensiveerum chemistry panel including hepatic transami-ases, bilirubin, alkaline phosphatase, lactate dehydro-enase, uric acid, albumin, calcium, phosphorus, andagnesium was performed at least 3 times weekly, and

oagulation studies were performed twice weekly. Se-um tumor markers (alfa fetoprotein [AFP] and �-hu-

an chorionic gonadotropin [HCG]) were measured

905

wcrmw

Ccweafidtywuo

S

bitfdaf(

R

P

mwglc3(ssbptlavmtAt(ia

rT�ifdmnttBc

trirppedNtaaa(

T

T

D

D

N

R

L

K. A. Margolin et al.

9

eekly, starting at the time of protocol enrollment. Ahest radiograph was performed at least weekly. Afterecovery from cycle 2, patients whose serum tumorarkers had normalized and who had residual massesere referred for surgical resection.

The National Cancer Institute Common Toxicityriteria, version 2.0 (http://ctep.info.nih.gov/reporting/

tc-3test.html), were used to assess toxicities. Patientsere evaluated as often as necessary to confirm recov-ry from acute toxicities and then underwent clinicalnd laboratory evaluation at monthly intervals for therst year. Evaluations were performed every 2 monthsuring the second year, every 4 months during thehird year, twice during the fourth year, and thenearly. Radiographic evaluations for tumor assessmentere required with every second clinical evaluationntil 5 years after treatment and then at the discretionf the physician.

tatistical Methods

Overall survival and progression-free survival,ased on intention to treat, were calculated from the

nitiation of protocol procedures (G-CSF mobiliza-ion before stem cell collection) to the time of deathrom any cause or tumor marker progression. Stan-ard Kaplan-Meier methods were used for survivalnalysis by using S-Plus software (S-Plus 6.0; Insight-ul, Seattle, WA). All significance testing was 2 sidedlog-rank test).

ESULTS

atient Characteristics and Treatment Details

Thirty-three patients were initially enrolled andet the eligibility criteria before protocol proceduresere initiated. Two of these patients had rapidly pro-ressive disease, and 1 also had poor stem cell mobi-ization that precluded protocol therapy. The patientharacteristics are listed in Table 1. Patients included2 men and 1 woman, with a median age of 30 yearsrange, 17-49 years). Twenty-four patients were clas-ified as having had poor-risk GCT at initial diagno-is, on the basis of 1 or more of the criteria establishedy the IGCCCG [4]. The criteria by which theseatients fell into the poor-risk category included lac-ate dehydrogenase �10 times the institutional upperimit of normal (1 patient; initial values were notvailable in 6 patients), metastatic disease in 1 or moreisceral (brain, bone, or liver) sites (all 24 patients),ediastinal-primary nonseminomatous GCT (3 pa-

ients), serum �-HCG �50 000 U/L (8 patients), andFP �10 000 ng/mL (7 patients). The median dura-

ion of response to first-line therapy was 4 monthsrange, 0-128 months). Twelve patients had normal-zed serum tumor markers in response to salvage ther-

py, and 21 had increasing markers. Two patients had p

06

esponded to paclitaxel-containing salvage therapy.he tumor histologic results were choriocarcinoma (n8), embryonal carcinoma/teratocarcinoma (includ-

ng yolk sac and endodermal sinus and including the 1emale patient; n � 18), mediastinal germ cell syn-rome with poorly differentiated carcinoma (n � 1),ixed GCT (n � 1), and pure seminoma (n � 5). The

umbers of prior treatment regimens were 1 (3 pa-ients), 2 (27 patients), or 3 (3 patients). Eleven pa-ients had prognostic scores of �2 by the system ofeyer et al., and 22 patients had scores �2 by theseriteria [13].

Two patients did not proceed to aPSCT becausehey had rapidly progressive disease. Twelve patientseceived only the first cycle of aPSCT, for the follow-ng reasons: 8 had progressive GCT after cycle 1, 3efused a second cycle (2 patients developed severeeripheral neuropathy, and 1, with a history of su-raventricular tachycardia, developed a symptomaticpisode during the first cycle), and 1, a heavy smoker,ied during cycle 1 as a result of fungal pneumonia.ineteen patients completed both cycles of protocol

herapy. The median interval between day 1 of cycle 1nd day 1 of cycle 2 was 58 days (range, 41-155 days),nd the median interval between hospital dischargefter cycle 1 and admission for cycle 2 was 32 daysrange, 12-134 days).

oxicities of aPSCT

The grade 3 and 4 toxicities of therapy for the 31

able 1. Patient Characteristics

Variable Data

iseasePoor risk at diagnosis, in first relapse 24

�-HCG >50 000 U/L 8AFP >10 000 ng/mL 7LDH >10 times upper limit of normal 1 (unknown in 6)Mediastinal primary tumor 3Visceral metastasis 22

Histologic characteristicsChoriocarcinoma 8Embryonal/yolk sac 17 male, 1 femaleMixed nonseminometous GCT 1Pure seminoma 5Mediastinal germ cell syndrome with

poorly differentiated carcinoma 1emographicsMale/female 32/1Age range, y (median) 17–49 (30)o. of prior regimens1 32 273 3

esponse status immediately beforeprotocol therapy

Responsive disease (markers normal) 12Progressive disease (markers increasing) 21

DH indicates lactate dehydrogenase.

atients who began the aPSCT regimen are listed in

Tt5aoccmdsddmOc

pieppeofpri

T

rorntrGsregrmtttpsOw

TT

SNH

DDR

A

*

†‡

Paclitaxel-Based Tandem Autologous Stem Cell Transplantations in Germ Cell Cancer

B

able 2. Severe stomatitis was predominantly limitedo cycle 1 at the paclitaxel dose of 425 mg/m2. One of

patients at the lower paclitaxel dose of 350 mg/m2

nd 5 of 19 patients who underwent the second cyclef aPSCT experienced grade 3 or 4 stomatitis. Be-ause dose-limiting mucositis occurred in the firstycle at the second dose level of paclitaxel (425 mg/

2), we did not further escalate the paclitaxel doseuring the first cycle. Although the mucositis in theecond cycle of therapy was only mild and was notose limiting, we elected not to escalate the paclitaxelose in that cycle because of the concern that patientsight develop irreversible cumulative neurotoxicity.ther gastrointestinal toxicities were similar in both

ycles. All patients developed a subacute sensory neuro-

able 2. Grade 3/4 Acute Regimen-Related Toxicities of Protocolherapy*

VariableCycle 1

(n � 31)†Cycle 2

(n � 19)‡

tomatitis 21 6ausea/vomiting 9 3epaticBilirubin alone 3 2Bilirubin and transaminase 1 1Transaminase alone 7 6iarrhea 2 3ermatitis 2 0egimen-related mortality 1 0

ll patients had grade 2 neuropathy (moderate, not interfering withactivity, and slowly resolving over several weeks to months afterall protocol therapy).

National Cancer Institute Common Toxicity Criteria, version2—Bone Marrow Transplantation scale.

Paclitaxel, etoposide, and carboplatin.Paclitaxel, ifosfamide, and carboplatin.

Figure 1. Progression-free (PFS) and overall (OS) survival for all 33

B & M T

athy characterized by generally painless hypesthesiasn a stocking-glove distribution that began within sev-ral days to 1 week after the cycle 1 dose of paclitaxel,eaked after the second cycle, and subsided in mostatients over several weeks to months after the lastxposure to paclitaxel. No patient experienced grade 3r 4 nephrotoxicity in either cycle. The characteristicsor stem cell mobilization and the results of hemato-oietic reconstitution were similar to those that weeported previously in our study of tandem cycles offosfamide, carboplatin, and etoposide [9].

herapeutic Outcomes

Twelve patients (36% of all patients who wereegistered) were alive and progression free at a medianf 67 months (Figure 1). Two of these patients did noteceive cycle 2 because of refusal or toxicities that didot meet off-protocol criteria. Of the other 21 pa-ients, 1 died of therapy-related infection, and theemaining patients relapsed and died as a result ofCT. Eight patients who achieved normalization of

erum tumor markers underwent surgical excision ofesidual masses at a median of 6 months after studyntry. One patient had progressive disease before sur-ery, and the other patients had surgery for residualadiographic masses and normalized serum tumorarkers. Of the 5 who had no pathologic evidence of

umor (4 with necrosis/fibrosis and 1 with matureeratoma), 4 (including the patient with mature tera-oma) remain disease free after protocol therapy; all 3atients with viable GCT (all with embryonal or yolkac histologic results) have died of progressive disease.f the 24 patients whose initial prognostic categoryas poor risk by the IGCCCG criteria, the outcome

patients entered onto the protocol (Kaplan-Meier analysis).

907

ae

aalrgpbeaspus

sF

D

tcmtaama

at diag

K. A. Margolin et al.

9

fter protocol therapy was similar to those of thentire patient population (Figure 2).

Patients were classified by the system of Beyer etl. according to their prognostic factors just beforePSCT. This system assigns a score of 2 for �-HCGevels �1000 U/L, 2 for tumor that is absolutelyefractory to cisplatin-based therapy for relapse (pro-ressing on therapy), and 1 each for a mediastinalrimary tumor, a tumor that is refractory to cisplatin-ased therapy (progressing within 4 weeks of the lastxposure), and disease progression immediately beforePSCT [13]. Of 22 patients with favorable prognosticcores (�2), 10 (45%) achieved remission and remainrogression free, whereas only 2 (18%) of 11 withnfavorable prognostic scores (�2) remain progres-

Figure 2. A, Progression-free survival by IGCCCG risk group

ion free after aPSCT (P � .05; Figure 3A). Overall d

08

urvivals paralleled disease-free survivals (P � .03;igures 1, 2B, and 3B).

ISCUSSION

The last 3 decades have witnessed the evolution ofherapy (from palliative to curative) for advanced germell cancer in the vast majority of patients. Further-ore, most of the acute toxicities of standard chemo-

herapy regimens for GCT are now readily manage-ble with appropriate supportive care interventions,nd the long-term toxicities of curative therapy areodest, consisting mainly of reversible neuropathy

nd restrictive lung disease, possible accelerated car-

nosis. B, Overall survival by IGCCCG risk group at diagnosis.

iovascular events, sexual dysfunction and diminished

fnoaytwtnabtt

aa

cdpect

infcmfbp

Fo

Paclitaxel-Based Tandem Autologous Stem Cell Transplantations in Germ Cell Cancer

B

ertility, and a slightly increased risk of second malig-ancies (see review [14]). Thus, although the numberf patients for whom major improvements in therapyre still needed is modest, they are predominantlyoung men for whom the individual effect of a cura-ive therapy is enormous. It is unknown at presenthether improvements in aPSCT regimens will lead

o a higher fraction of cured patients or whether thiseed can be met only with the discovery of improvedgents for first-line therapy that will be based on aetter understanding of the molecular genetics ofhese malignancies and the individuals who develophem.

When our study was developed, there was generalgreement that a substantial fraction of patients with

igure 3. A, Progression-free survival by Beyer risk group at the tif protocol entry.

dvanced disease whose initial therapy failed could be a

B & M T

ured with aPSCT. In the United States, this was ad-ressed with the use of tandem cycles containing carbo-latin and etoposide with or without a third agent, gen-rally an alkylating agent such as ifosfamide oryclophosphamide. Data in support of the use of pacli-axel originated from the report of Motzer et al [5].

Because there was, at the time of this study’snception in 1995, no widely used accepted prog-ostic system for classifying patients in relapse oror predicting their response to therapy, we ac-epted a heterogeneous group of patients for treat-ent in this study. Because the IGCCCG system

or classifying patients at the time of diagnosis haseen widely incorporated in recent protocols foratients receiving first-line therapy, we initially an-

protocol entry. B, Overall survival by Beyer risk group at the time

me of

lyzed the entire study population and the subsets of

909

pCsmb(hicsttlfeb

aiwifrHpanftahlcslbanfitgdtmhtalaraoGfimco

pmctp

tspUttt(paamtpopr4idccsssisweaawraa(ccpshbag

cecaI

K. A. Margolin et al.

9

atients whose initial characteristics in the IGC-CG fell into the poor-risk category on the basis of

erum lactate dehydrogenase, AFP, �-HCG, sites ofetastasis (visceral [including bone, liver, and

rain] versus nonvisceral), and site of primary tumortestis or retroperitoneal versus mediastinal). In oureterogeneous study population, the initial present-

ng characteristics did not seem to predict the out-ome of protocol therapy. This observation is noturprising, because many other factors, especiallyhe rate of decline of serum tumor markers duringherapy, probably influence the outcome of first-ine therapy [15-18]. The use of such predictiveactors to tailor therapy is currently undergoingvaluation in a large US Intergroup study and haseen incorporated into smaller studies as well.

The prognostic system for patients in relapse waslso derived from an international database of patientsn relapse from initial therapy who were then treatedith aPSCT [13]. In multivariate analysis, the most

mportant prognostic factors for long-term relapse-ree survival were progressive disease before aPSCT,efractory or absolutely refractory disease, and anCG level �1000 U/L. After this study was com-

leted, we assigned a prognostic score to each patientccording to these characteristics (mediastinal versusonmediastinal primary tumor, progression just be-ore aPSCT, degree of refractoriness to prior cispla-in-based therapy, and level of �-HCG). Although theccuracy of this system as applied to our patients mayave been limited by missing marker data between the

ast chemotherapeutic regimen and the time of proto-ol entry, we found a strong association between thiscore and the outcome of protocol therapy, particu-arly for patients with a high score. These patients maye better served by participating in studies of novelgents than by undergoing aPSCT. The Beyer prog-ostic score should also be considered as a prestrati-cation criterion in future phase III studies. For pa-ients with favorable scores in this system, it may be ofreater value to explore the effect of variables such asrug, dose, single (most common in European cen-ers) versus tandem (used in most US centers) versusultiple cycles of high-dose therapy supported by

ematopoietic stem cells. An example of the latter ishe TICE regimen, consisting of 2 cycles of paclitaxelnd ifosfamide for cytoreduction and stem cell mobi-ization, followed by 3 cycles of high-dose carboplatinnd etoposide, each supported by stem cells. Theecent report of this regimen, piloted by Motzer [19]nd updated by Kondagunta [20] showed that �75%f patients with poor-prognosis nonseminomatousCT (incomplete serum tumor marker response to

rst-line chemotherapy, relapsed extragonadal pri-ary tumor, or second relapse) achieved a durable

omplete remission with chemotherapy with or with-

ut surgery. This novel regimen consists of 2 cycles of s

10

aclitaxel and ifosfamide to cytoreduce the tumor andobilize autologous peripheral blood hematopoietic

ells, followed by 3 cycles of cell-supported chemo-herapy consisting of high-dose carboplatin and eto-oside [19,20].

Among the most important questions that remaino be answered in this field is the true power of aPSCTtrategies to overcome drug resistance and, thus, cureatients in relapse after 1 or more prior regimens.ntil recently, the literature was composed of uncon-

rolled trials of small to moderate size featuring pa-ients with heterogeneous characteristics who werereated with aPSCT regimens containing 2 or 3 drugscarboplatin and etoposide with or without cyclophos-hamide or ifosfamide) [3]. The preliminary results ofn important European multicenter trial to comparePSCT with standard-dose second-line therapy regi-ens were recently published [21]. Patients in this

rial had advanced GCT in first relapse but were notlatinum refractory (their disease had not progressedr relapsed within 1 month of prior exposure to cis-latin-containing chemotherapy). Those who wereandomized to standard-dose chemotherapy receivedcycles of VIP or VeIP (etoposide or vinblastine with

fosfamide and cisplatin), and those who were ran-omized to aPSCT received 3 cycles of standard-dosehemotherapy followed by a single cycle of high-dosearboplatin, etoposide, and cyclophosphamide withtem cell support. The results of this trial demon-trated no statistically significant difference in overallurvival between the 2 groups. The treatment mortal-ty in the aPSCT arm was higher than that in thetandard-dose arm, and all other outcome parametersere the same in the 2 groups (response rate, 1-yearvent-free survival, and overall survival). Although theuthors reported that patients in the standard-therapyrm did not often cross over to receive aPSCT, thereere other potential explanations for this negative

esult. Two different reinduction regimens were avail-ble, and only a single cycle of non–taxane-basedPSCT was administered. The dose of carboplatin1-2.2 g/m2) is similar to that which is used in eachycle in the ongoing US Intergroup and previous andurrent Memorial Sloan-Kettering trials and is com-arable to the range of carboplatin doses used in thistudy. Furthermore, the patients enrolled in that trialad an unusually poor prognosis (65% of patients inoth arms had an incomplete response to initial ther-py) but were not stratified by IGCCCG or Beyer riskroup.

We have shown that aPSCT using tandem cyclesontaining high-dose paclitaxel in combination withtoposide plus carboplatin and with ifosfamide plusarboplatin can be given safely and possesses encour-ging activity in patients with poor-prognosis GCT.n view of the excellent activity of paclitaxel and its

ynergy with the other agents used in this regimen, we

buupdouceauinG

A

p

R

1

1

1

1

1

1

1

1

1

1

2

2

Paclitaxel-Based Tandem Autologous Stem Cell Transplantations in Germ Cell Cancer

B

elieve that it should undergo further testing, partic-larly in comparison with other regimens commonlysed in this setting, such as tandem cycles of etoposidelus carboplatin. Many questions remain to be ad-ressed as we attempt to identify the ideal treatmentf patients with recurrent GCT. These include these of other agents directed at molecular targets spe-ific to GCT and the design of disease-specific strat-gies that do not depend on the modest dose responsechievable with currently available drugs. A betternderstanding of the biology and unique susceptibil-

ties of these tumors will then allow investigators toarrow the gap between cure and death in advancedCT.

CKNOWLEDGMENTSSupported by City of Hope Cancer Center Sup-

ort Grant CA 33572.

EFERENCES

1. Einhorn LH. Treatment of testicular cancer: a new and im-proved model. J Clin Oncol. 1990;8:1777-1781.

2. Motzer RJ, Bosl GJ. High-dose chemotherapy for resistantgerm cell tumors: recent advances and future directions [re-view]. J Natl Cancer Inst. 1992;84:1703-1709.

3. Margolin KA. High-dose chemotherapy and stem cell supportin the treatment of germ cell cancer [review]. J Urol. 2003;169:1229-1233.

4. International Germ Cell Cancer Collaborative Group. Interna-tional germ cell consensus classification: a prognostic factor-based staging system for metastatic germ cell cancers. J ClinOncol. 1997;15:594-603.

5. Motzer RJ, Bajorin DF, Schwartz LH, et al. Phase II trial ofpaclitaxel shows antitumor activity in patients with previouslytreated germ cell tumors. J Clin Oncol. 1994;12:2277-2283.

6. Stemmer SM, Cagnoni PJ, Shpall EJ, et al. High-dose pacli-taxel, cyclophosphamide, and cisplatin with autologous hema-topoietic progenitor-cell support: a phase I trial. J Clin Oncol.1996;14:1463-1473.

7. Somlo G, Doroshow JH, Synold T, et al. High-dose paclitaxelin combination with doxorubicin, cyclophosphamide and pe-ripheral blood progenitor cell rescue in patients with high-riskprimary and responding metastatic breast carcinoma: toxicityprofile, relationship to paclitaxel pharmacokinetics and short-term outcome. Br J Cancer. 2001;84:1591-1598.

8. Leong LA, Raschko J, Somlo G, et al. Phase I tandem cyclehigh-dose chemotherapy with cisplatin, etoposide, and cyclo-

phosphamide followed by ifosfamide, carboplatin, escalating

B & M T

doses of paclitaxel and stem cell rescue [abstract 479]. Proc AmSoc Clin Oncol. 2001;20:121a.

9. Margolin KA, Doroshow JH, Ahn C, et al. Treatment of germcell cancer with two cycles of high-dose ifosfamide, carboplatin,and etoposide with autologous stem-cell support. J Clin Oncol.1996;14:2631-2637.

0. Iniguez C, Larrode P, Mayordomo JI, et al. Reversible periph-eral neuropathy induced by a single administration of high-dosepaclitaxel. Neurology. 1998;51:868-870.

1. Calvert A, Newell D, Gumbrell L, et al. Carboplatin dosage:prospective evaluation of a simple formula based on renal func-tion. J Clin Oncol. 1989;7:1748-1756.

2. Somlo G, Sniecinski I, Odom-Maryon B, et al. Effect ofCD34� selection and various schedules of stem cell reinfusionand granulocyte colony-stimulating factor priming on hemato-poietic recovery after high-dose chemotherapy for breast can-cer. Blood. 1997;89:1521-1528.

3. Beyer J, Kramar A, Mandanas R, et al. High-dose chemother-apy as salvage treatment in germ cell tumors: a multivariateanalysis of prognostic variables. J Clin Oncol. 1996;14:2638-2645.

4. Raghavan D. Testicular cancer: maintaining the high cure rate.Oncology. 2003;17:218-230.

5. Toner GC, Geller NL, Tan C, et al. Serum tumor markerhalf-life during chemotherapy allows early prediction of com-plete response and survival in nonseminomatous germ celltumors. Cancer Res. 1990;50:5904-5910.

6. Mazumdar M, Bajorin DF, Bacik J, et al. Predicting outcome tochemotherapy in patients with germ cell tumors: the value ofthe rate of decline of human chorionic gonadotrophin andalpha-fetoprotein during therapy. J Clin Oncol. 2001;19:2534-2541.

7. Motzer RJ, Mazumdar M, Gulati SC, et al. Phase II trial ofhigh-dose carboplatin and etoposide with autologous bonemarrow transplantation in first-line therapy for patients withpoor-risk germ cell tumors. J Natl Cancer Inst. 1993;85:1828-1835.

8. Motzer RJ, Mazumdar M, Bajorin DF, et al. High-dose carbo-platin, etoposide, and cyclophosphamide with autologous bonemarrow transplantation in first-line therapy for patients withpoor-risk germ cell tumors. J Clin Oncol. 1997;15:2546-2552.

9. Motzer RJ, Mazumdar M, Sheinfeld J, et al. Sequential dose-intensive paclitaxel, ifosfamide, carboplatin, and etoposide sal-vage therapy for germ cell tumor patients. J Clin Oncol. 2000;18:1173-1180.

0. Kondagunta GV, Bacik J, Bajorin DF, et al. Sequential dose-intensive paclitaxel plus ifosfamide and carboplatin plus etopo-side with autologous stem cell transplantation for germ celltumor (GCT) patients (pts) [abstract]. Proc Am Soc Clin Oncol.2005;23:378.

1. Pico JL, Rosti G, Kramar A, et al. A randomized trial ofhigh-dose chemotherapy in the salvage treatment of patientsfailing first-line platinum chemotherapy for advanced germ cell

tumors. Ann Oncol. 2005;16:1152-1159.

911