KEY WORDS Testis cancer Germ cell cancer Tandem autotransplant

Transcription

1 Biology of Blood and Marrow Transplantation 13: (2007) 2007 American Society for Blood and Marrow Transplantation /07/ $32.00/0 doi: /j.bbmt Utility of Single versus Tandem Autotransplants for Advanced Testes/Germ Cell Cancer: A Center for International Blood and Marrow Transplant Research (CIBMTR) Analysis Hillard M. Lazarus, 1 Patrick J. Stiff, 2 Jeanette Carreras, 3 Brent R. Logan, 3 Luke Akard, 4 Brian J. Bolwell, 5 Richard W. Childs, 6 Robert Peter Gale, 7 John P. Klein, 3 Michael C. Lill, 8 Waleska S. Pérez, 3 Edward A. Stadtmauer, 9 J. Douglas Rizzo 3 1 University Hospitals of Cleveland, Cleveland, Ohio; 2 Loyola University Medical Center, Maywood, Illinois; 3 The Solid Tumor Working Committee of the Center for International Blood and Marrow Transplant Research (CIBMTR), Health Policy Institute, Medical College of Wisconsin, Milwaukee, Wisconsin; 4 Indiana Blood & Marrow Transplantation, Beach Grove, Indiana; 5 Cleveland Clinic Foundation, Cleveland, Ohio; 6 National Institutes of Hematology, Bethesda, Maryland; 7 Center for Advanced Studies in Leukemia, and 8 Cedars Sinai Medical Center, Los Angeles, California; 9 University of Pennsylvania Hospital, Philadelphia, Pennsylvania Correspondence and reprint requests: Hillard M. Lazarus, MD, Department of Medicine, University Hospitals of Cleveland, Euclid Avenue, Cleveland, OH ( hillard.lazarus@case.edu). Received January 25, 2007; accepted February ABSTRACT Tandem autotransplants are used to treat advanced testis cancer patients but their value compared to a single autotransplant is unknown. To evaluate the results of autotransplant in relapsed testicular/germ cell cancer, data from 300 patients undergoing autotransplants were reported to the Center for International Blood and Marrow Transplant Research. We compared results for those patients intended to undergo tandem autotransplant procedures (N 102) versus patients in whom a second autotransplant was not planned (N 198). Five-year survival probability was 35% (95% confidence interval 25%-46%) in the planned tandem transplant cohort compared to 42% (35%-49%) in the group not planned to have a second transplant (P.29). Probability of progression-free survival at 5 years for these cohorts was 34% (25%-44%) and 38% (31%-45%), respectively (P.50). The planned tandem autotransplant cohort had significantly more advanced disease at diagnosis and greater likelihood of cisplatin resistance. Patients intended to receive tandem transplants had a lower treatment-related mortality at 1 year (3% versus 10%, P.02). Using propensity score analysis the planned tandem autotransplant cohort had significantly lower treatment-related mortality (P.044) but no different risk of relapse (P.541) compared to the planned single transplant cohort. Tandem autotransplants for testicular cancer are associated with less treatment-related mortality than a planned single transplant, with no differences in disease-related outcomes or overall survival at 3 years. Patient selection bias for either transplant approach, however, may affect the results of this observational study; a randomized trial is needed to determine which approach, if either, is better American Society for Blood and Marrow Transplantation KEY WORDS Testis cancer Germ cell cancer Tandem autotransplant INTRODUCTION Conventional combination chemotherapy with or without surgical resection is curative in nearly 80% of testis cancer patients, and about 1/4 of relapsed patients are cured using cisplatin and ifosfamide-containing regimens [1-3]. Poor-prognosis germ cell cancer patients can be identified for whom the 5-year survival rates are considerably less than 50% [4,5]; several approaches have been used to improve outcome in these high-risk patients, including increasing dose intensity and alternating chemotherapy regimens [6-8]. Autologous hematopoietic stem cell transplantation (HSCT) has been utilized as therapy for pa- 778

2 Tandem Autotransplants in Testis Cancer 779 tients with refractory and relapsed testis and germ cell cancer and as first-line therapy for poor-risk patients [9-12]. The optimal timing of transplantation and factors predicting survival are controversial. In 1996, Beyer et al. [13] reported prognostic factors for patients with testicular cancer receiving single autologous HSCT. Cisplatin resistance, primary mediastinal cancers, progressive disease before autologous HSCT, and serum beta human chorionic gonadotrophin ( -HCG) 1000 mu/ml before transplantation were adverse prognostic factors [13]. Toimprove results of autologous HSCT, many centers began using 2 consecutive or tandem autotransplant procedures rather than single autotransplants [14-16]. Despite the lack of efficacy data, it has been estimated that 30% of such patients undergo tandem autotransplants. Using data reported to the Center for International Blood and Marrow Transplant Research (CIBMTR) we studied outcomes after autologous HSCT for testicular cancer between January 1, 1989, and December 31, 2001, to determine the value of planned tandem (n 102) versus planned single (n 198) autotransplants. PATIENTS AND METHODS Data Sources The CIBMTR is a research affiliation of the International Bone Marrow Transplant Registry (IBMTR), Autologous Blood and Marrow Transplant Registry (ABMTR), and the National Marrow Donor Program (NMDP) that comprises a voluntary working group of more than 450 transplantation centers worldwide. These centers contribute detailed data on consecutive allogeneic and autologous HSCT to the Statistical Center at the Medical College of Wisconsin in Milwaukee or the NMDP Coordinating Center in Minneapolis. Participating centers are required to report all transplants consecutively; compliance is monitored by on-site audits. Patients are followed prospectively, with yearly follow-up. All CIBMTR teams contribute registration data that include disease type, age, sex, pretransplant disease stage, and chemotherapy responsiveness, date of diagnosis, graft type (bone marrow, blood-derived stem cells, or cord blood), pretransplant conditioning regimen, posttransplant disease progression and survival, development of a new malignancy, and cause of death. Research data are collected on selected subsets of registered patients and include comprehensive preand posttransplant clinical information. We developed and used a testis cancer-specific data collection form that included known prognostic factors for outcome, computerized checks for errors, physician reviews of submitted data, and on-site audits of participating centers ensure the quality of data. Based on data collected in the Centers for Disease Control Hospital Surveys [17,18] and the U.S. Government Accounting Office [19,20] and worldwide surveys of transplant activity, approximately 40% of allogeneic transplants worldwide and more than 50% of autotransplants in North and South America are registered with the CIBMTR. Patients Between January 1, 1989, and December 31, 2001, 321 patients who received autotransplants for testicular cancer were reported to the CIBMTR and had complete research data. Twenty-one patients were excluded from further analyses because of histology (12 pure teratoma and 5 pure yolk sac) and because the primary origin site was mediastinal (4); 300 patients were analyzed. The identified cases came from 76 reporting centers in 8 different countries. The largest transplant center has 18 patients. Participating centers were contacted to confirm attribution of second transplant as: planned tandem or delivered for recurrent disease that was not planned ; and to determine whether there were patients for whom a second transplant was planned but not executed. To assure that the research patients were representative of all registered patients, demographics and survival rates between research and registered patients were compared; no differences were noted in age, gender, number of transplants, graft type, interval from diagnosis to transplant, and overall survival (OS). Median follow-up of survivors after autologous HSCT was 62 (range: 2-163) months. Study Endpoints Primary outcomes studied were treatment-related mortality (TRM), cancer progression or relapse, progression-free survival, and OS. TRM was defined as death in the first 28 days after transplant or death beyond 28 days after transplant in the absence of recurrence or cancer progression. Patients alive and in continuous complete remission were censored at the time of last follow-up. PFS was defined as survival without recurrence or cancer progression, as measured by exam, radiographs, and/or an increase in serum cancer markers. Recurrence or progression of disease and death from any cause were considered events. Those who survived without recurrence or progression were censored at the date of last contact. OS was defined as the interval between transplant and death from any cause. Surviving patients were censored at the date of last contact. Disease-risk (favorable, intermediate, poor risk) was based on disease characteristics [13] as well as blood cancer markers at diagnosis [21].

3 780 H. M. Lazarus et al. Statistical Analysis Comparisons of descriptive factors were performed using chi-square testing for discrete covariates, and the Kruskal-Wallis test for continuous covariates. Univariate probabilities of TRM and relapse/progression were computed using cumulative incidence to accommodate competing risks. Relapse/progression is the competing risk for TRM and TRM is the competing risk for relapse/progression [22]. Univariate probabilities of PFS and OS were computed using the Kaplan-Meier method [21]. Estimates of standard error for the survival function were calculated by Greenwood s formula and 95% confidence intervals (CI), using log-transformed intervals. Statistical Methods A propensity score approach was used to compare tandem and single autologous HSCT while adjusting for the varying patient risk factors [23,24]. First, the likelihood of being assigned to the planned tandem arm was modeled with patient, disease, and treatmentrelated variables using logistic regression. Predicted probabilities of being in the planned tandem arm were computed and used to group patients into 5 strata. Multivariate Cox proportional hazards regression with a main effect of treatment group, stratified on propensity score group, and including additional transplantrelated covariates, were used to adjust for other potentially confounding risk factors. A comparison between tandem and single autologous HSCT was performed as an intention-to-treat (ITT) analysis. The assumption of proportional hazards for each factor in the Cox model was tested using time-dependent covariates. When this indicated differential effects over time (nonproportional hazards), models were constructed breaking the posttransplant time course into 2 periods, using the maximized partial likelihood method to find the most appropriate breakpoint. First-order interactions between main effect and strata or significant covariates were tested before and after stepwise modeling. After modeling time-varying effects and interactions, the final multivariate model was built using a forward stepwise model building approach to develop models for TRM, relapse/progression, PFS, and OS, forcing the main effect into the model and stratifying on propensity score strata. Factors significantly associated with the outcome at a 5% level (2-sided) were kept in the final model. The variables tested in the logistic regression model for the likelihood of receiving a planned second transplant and the variables tested in the Cox regression analysis stratified on propensity score are listed in Table 1.An additional subgroup analysis was performed on the group of patients with residual cancer and elevated serum cancer markers before transplantation who are Table 1. Variables Tested in Logistic Regression Model for Likelihood of Receiving a Planned Second Transplant (Propensity Score) and Propensity Score Analysis (Cox with Propensity Score) Logistic regression model: Patient-related variables: Age at transplant: <30 years* versus >30 years Karnofsky performance status at transplant: <90%* versus >90% Disease-related variables at diagnosis: Histology: pure seminoma* versus nonseminoma versus mixed seminoma nonseminoma Risk: nonseminoma poor risk* versus others versus missing Alpha-fetoprotein: good risk* versus intermediate risk versus poor risk versus missing HCG: good risk* versus intermediate risk versus poor risk versus missing Disease-related variables prior to transplant: Number of conventional chemotherapy regimens: 1* versus 2 versus >3 Number of total chemotherapy cycles: 1-5* versus 6-10 versus >11 versus missing Prior salvage attempts: 0* versus 1 versus 2 versus 3 Sensitivity to last platinum-containing regimen: CR PR* versus others versus missing Sensitivity to last platinum containing regimen: CR* versus PR versus others versus missing Disease status prior to first transplant: no evidence of disease, either surgically or clinically defined* versus serum cancer marker elevation only versus residual cancer with elevated serum cancer markers versus residual cancer with normal markers Time from diagnosis to first transplant: <12 months* versus >12 months Treatment-related: Year of transplant: * versus Propensity score analysis: Main effect variable Intended number of transplants: No planned second transplant* versus planned second transplant (intention to treat) Disease-related variables prior to transplant: Conditioning regimen, first transplant: carboplatin VP16 CY* versus carboplatin VP16 versus carboplatin VP16 ifosfamide versus others Conditioning regimen dose intensity: Carboplatin >2000 mg/m 2 or Etoposide >2200 mg/m 2 (high dose) versus others (low dose) Treatment-related: Source of stem cells: BM* versus PBSC versus BM PBSC

4 Tandem Autotransplants in Testis Cancer 781 at high risk of relapse. Univariate Cox regression models as described above were used to estimate the relative risk of relapse for these high risk patients in the tandem versus single autologous transplant groups. There were no statistically significant center effects [25]. RESULTS Three hundred patients were analyzed; 198 planned to receive single, whereas 102 patients planned to receive tandem autotransplants (Table 2). The cohorts were comparable for median age, proportion with performance status 90% at transplant, testicular versus abdominal origin, histology, risk [16], serum -protein and -HCG [26] at diagnosis, serum -HCG at time of first transplant [13], number of chemotherapy regimens prior to HSCT, and interval from diagnosis to first HSCT. For the planned single versus the planned tandem transplant groups, 20% and 14% of patients had no evidence of cancer when they received the first transplant, and an additional 8% and 4% had only serum cancer marker elevation at transplant, respectively. Seventy-one percent and 82% of patients had residual cancer at time of HSCT, respectively. The planned tandem transplant cohort was significantly more likely to receive blood rather than marrow as the stem cell source; however, using the propensity score adjusted Cox model, source of stem cells was not a factor. Further, there also was no difference in the groups for year of HSCT. Importantly, the tandem cohort had a significantly higher proportion of patients with adverse risk features including more advanced disease stage at diagnosis (testicular and retroperitoneal node involvement) and cisplatin-resistance at time of transplantation. Other comorbidities reported by the teams to the Registry included cardiovascular disease, gastrointestinal disorders, hepatic dysfunction, and pulmonary disease. These conditions were evenly distributed between the patient groups (data not shown). A higher percentage of the planned tandem autologous HSCT patients received a carboplatin and etoposide conditioning regimen rather than the drug regimen used more often for single transplants. We tested the conditioning regimen dose and dose intensity in the propensity score adjusted Cox model but found these factors were not significant. Median follow-up of the groups was 63 (range: 5-163) months for the planned single and 53 (range: 2-132) months for the planned tandem transplants. Cumulative incidence of TRM was 10% (95% CI: 6%-14%) at 1 year for the planned single transplant cohort and 3% (95% CI: 1%-7%) for the planned tandem cohort (intent-to-treat, P.02). Table 3 shows the unadjusted TRM remained higher at 3 and 5 years for the planned single transplant patient cohort (P.03 and 0.02, respectively). Relapse/progression incidence was similar for the 2 cohorts. PFS and OS at 1, 3, and 5 years after autotransplant were similar (Figures 1 and 2, Table 3). The probability of PFS at 5 years for the planned tandem transplant cohort was 34% (95% CI, 25%-44%) compared to 38% (31%-45%) in the planned single transplant cohort (pointwise P.50). The probability of 5-year OS was 35% (25%-46%) versus 42% (35%-49%), respectively, in the 2 transplant groups (P.29). None of the 22 patients (15 single- and 7 tandem- HSCT) who received 3 or more salvage attempts before transplant became long-term survivors (Table 2 and Figure 2). Variables from the logistic regression model that predict likelihood of being in the planned tandem transplant cohort included disease state at first transplant and disease stage at diagnosis (Table 4). Specifically, patients with residual cancer and elevated serum cancer markers were more likely to be in the planned second transplant group (odds ratio [OR] 2.28, 95% CI , P.029), compared to patients with no evidence of disease. Also, patients with retroperitoneal cancer involvement with or without testis involvement at diagnosis were more likely to be in the planned second transplant cohort (OR 3.84, 95% CI , P.003). In Cox regression analysis stratified on propensity score (Table 4) the planned tandem autotransplant group had significantly lower TRM (P.044). Propensity score analysis indicated that there was no statistically significant difference in risk of relapse in the planned tandem transplant group compared to the planned single transplant population (P.54) (Table 4). Table 4 shows the risk of treatment failure and overall mortality. The effect of tandem transplantation on PFS and OS was time dependent. Patients who received tandem transplantation had significantly better PFS within the first 9 months after transplant and higher OS within the first 3 months after transplant compared to those receiving only 1 transplant. In those who survived beyond these time points, there was no significant difference between the groups. A subgroup analysis of relapse in the high-risk (residual cancer with elevated serum cancer markers at transplant) cohort found a significant time-dependent effect. There was no significant difference in relapse within the first 9 months after transplant (RR 0.72, 95% CI , P.16); however, the relapserisk was higher in the tandem group in those who survived disease free for 9 months after transplant (RR 4.59, 95% CI , P.019). The overwhelming reason for failure was disease recurrence in both cohorts (79% planned single and 94% planned tandem autotransplant), but organ fail-

5 782 H. M. Lazarus et al. Table 2. Characteristics of Patients Receiving Bone Marrow and/or Peripheral Blood Stem Cell Autologous Transplantation for Testicular Cancer Reported to the CIBMTR, from 1989 to 2001, by Intended Number of Transplants No Planned 2nd tx Planned 2nd tx (ITT) Variable N Eval N (%) N Eval N (%) P-value b Number of patients 198 a 102 a Age, median (range), years c (15-59) (17-63).83 Age at transplant, years c (10) 12 (12) (40) 42 (41) (34) 34 (33) (15) 9 (9) >51 2 (1) 5 (5) Karnofsky score pretransplant c (1) 1 (1) 60 2 (1) 2 (2) (9) 9 (9) (24) 17 (17) (65) 71 (71) Karnofsky score pretransplant <90 c (35) (29).27 Origin of primary cancer at diagnosis c Testicular primary 183 (92) 92 (90) Abdominal nodes primary 15 (8) 10 (10) Stage at diagnosis Testis primary Testis only 49 (25) 11 (11) Testis retroperitoneal only 23 (12) 22 (21) Any CNS involvement 9 (4) 4 (4) Widespread involvement 102 (51) 55 (54) Extragonadal primary Retroperitoneal only 2 (1) 2 (2) Widespread involvement 13 (7) 8 (8) Histology at diagnosis c Seminoma any nonseminoma 31 (16) 20 (20) Pure seminona 28 (15) 18 (18) Pure choriocarcinoma 18 (9) 7 (7) Pure embryonal 24 (12) 9 (9) Pure (other) d 4 (2) 1 (1) Mixed nonseminoma (without seminoma) 87 (45) 44 (45) None 1 (1) 0 Risk at diagnosis ce Nonseminoma good risk 40 (21) 13 (13) Nonseminoma intermediate risk 27 (14) 13 (13) Nonseminoma poor risk 68 (36) 35 (34) Nonseminoma not specified 8 (4) 6 (6) Nonseminoma none 1 (1) 1 (1) Seminoma good prognosis 14 (7) 9 (9) Seminoma intermediate prognosis 16 (8) 16 (16) Seminoma not specified 8 (4) 6 (6) None nonseminoma seminoma 9 (5) 2 (2) Serum -fetoprotein at diagnosis, (exclude pure seminoma), median (range), ng/ml cf (0-278,455) (0-58,649).19 Serum -fetoprotein at diagnosis (exclude pure seminoma) cfg Good risk 63 (57) 42 (69) Intermediate risk 34 (31) 12 (20) Poor risk 13 (12) 7 (11) Serum -fetoprotein at diagnosis (exclude pure seminoma) cf <100,000 ng/ml 108 (98) 61 (100) >100,000 ng/ml 2 (2) 0 -HCG at diagnosis, median (range), IU/L ch (0-140,000) (0-113,440).57 -HCG at diagnosis chg Good risk 85 (63) 41 (61) Intermediate risk 28 (21) 12 (18) Poor risk 21 (16) 14 (21) -HCG at diagnosis ch <100,000 IU/L 130 (97) 65 (97) >100,000 IU/L 4 (3) 2 (3)

6 Tandem Autotransplants in Testis Cancer 783 Table 2. (Continued) No Planned 2nd tx Planned 2nd tx (ITT) Variable N Eval N (%) N Eval N (%) P-value b Initial surgery c Orchiectomy only 140 (71) 73 (71) Unilateral retroperitoneal node dissection and orchiectomy only 13 (7) 11 (11) Biospsy other 13 (7) 6 (6) Debulking only 1(<1) 1 (1) Other i 2 (1) 0 Initial surgery, not specified 2 (1) 0 No initial surgery 26 (13) 11 (11) Initial chemotherapy c VAB 1 (1) 0 PVB 18 (9) 5 (5) BEP 118 (60) 66 (66) EP 29 (15) 14 (14) Other etoposide based regimens 2 (1) 1 (1) Other vinblastine based therapy 6 (3) 2 (2) Other chemotherapy j 4 (2) 2 (2) Chemotherapy given, not specified 2 (1) 0 No chemotherapy 16 (8) 10 (10) Number of conventional chemotherapy regimens prior to transplant c (13) 21 (20) (51) 55 (54) 3 55 (28) 18 (18) 4 15 (7) 7 (7) No chemotherapy k 1 (1) 1 (1) Number of conventional chemotherapy cycles c (23) 33 (32) (62) 57 (56) >11 24 (12) 7 (7) Chemotherapy given, cycles unknown 4 (2) 4 (4) No chemotherapy l 1 (1) 1 (1) Prior salvage attempts cm (lines after 1 st relapse or progression) (41) 39 (38) 2 40 (20) 26 (25) 3 27 (14) 8 (8) 4 2 (1) 0 Others 11 (6) 4 (4) No salvage attempt 10 (5) 4 (4) No relapse 25 (13) 19 (19) First relapse missing 1 (<1) 2 (2) Sensitivity to 1 st platinum containing regimen cn CCR 2 (1) 0 CR 71 (37) 25 (26) PR 84 (44) 51 (54) SD 4 (2) 8 (9) NR 9 (5) 3 (3) PD 12 (6) 4 (4) NE 4 (2) 2 (2) No cisplatin or carboplatin 3 (2) 1 (1) No chemotherapy 2 (1) 1 (1) Sensitivity to last platinum containing regimen co CCR 1 (1) 0 CR 35 (19) 18 (18) PR 91 (49) 42 (43) SD 19 (10) 14 (14) NR 9 (5) 6 (6) PD 13 (7) 13 (13) ME 3 (2) 0 NETD 0 1 (1) NE 8 (4) 3 (3) No cisplatin or carboplatin 3 (2) 1 (1) No chemotherapy 2 (1) 1 (1)

7 784 H. M. Lazarus et al. Table 2. (Continued) No Planned 2nd tx Planned 2nd tx (ITT) Variable N Eval N (%) N Eval N (%) P-value b Sensitivity to any platinum-containing chemotherapeutic agent (as reported by team) c <.001 Sensitive 138 (77) 47 (50) Resistant 31 (17) 42 (44) Untreated 9 (5) 1 (1) Refractory 2 (1) 5 (5) -HCG prior to first transplant c <1,000 IU/L 120 (85) 63 (78) >1000 IU/L 22 (15) 18 (22) Disease status prior to first transplant c No evidence of disease surgically defined 9 (5) 2 (2) No evidence of disease clinically defined 28 (15) 12 (12) Serum cancer marker elevation only 15 (8) 4 (4) Residual cancer with elevated markers 71 (37) 59 (60) Residual cancer with normal markers 65 (34) 22 (22) Not evaluable 3 (2) 0 Disease status prior to second transplant p NA 74 No evidence of disease clinically defined 20 (27) Serum cancer marker elevation only 2 (3) Residual cancer with elevated markers 28 (38) Residual cancer with normal markers 20 (27) Not evaluable 4 (5) Time from diagnosis to first transplant, median (range), months c (<1-274) (2-257).06 Time from first to second transplant, median (range), months p 2 10 (1-19) 85 1 (1-3) Mononucleated cells infused, first transplant, median (range), 10 8 /kg c 82 6 (<1-34) 42 4 (<1-39).38 CD34 cells infused, first transplant, median (range), 10 6 /kg cq 90 4 (<1-104) 68 3 (<1-71).21 Conditioning regimen, first transplant r <.001 Carboplatin VP16 34 (17) 45 (44) Carboplatin VP16 Ifosfamide 38 (19) 9 (9) Carboplatin VP16 CY 91 (46) 32 (31) Others s 35 (18) 16 (16) Number of conditioning drugs, first transplant < (<1) 2 (2) 2 43 (22) 46 (45) (74) 54 (53) 4 6 (3) (1) 0 Number of conditioning drugs, second transplant NA (3) 2 39 (49) 3 38 (47) 4 1 (1) Graft type c BM 59 (30) 14 (14) PBSC 120 (61) 76 (74) BM PBSC 19 (9) 12 (12) Planned 2 nd transplant per protocol planned, 1 delivered 196 (99) 1 planned, 2 delivered 2 (1) 2 nd planned, 1 delivered 17 (17) 2 nd planned, 2 delivered 85 (83) Number of transplants < (99) 17 (17) 2 2 (1) 84 (82) (1) Year of first transplant c (2) (3) 3 (3) (4) 3 (3) (5) 7 (7) (9) 1 (1) (4) 5 (4) (9) 7 (7) (14) 16 (16)

8 Tandem Autotransplants in Testis Cancer 785 Table 2. (Continued) No Planned 2nd tx Planned 2nd tx (ITT) Variable N Eval N (%) N Eval N (%) P-value b (18) 19 (18) (14) 18 (18) (9) 9 (9) (7) 7 (7) (2) 7 (7) Median follow-up of survivors, months (5-163) (2-132) tx indicates transplant; CCR, continued complete response; CR, complete response; PR, partial response; SD, stable disease; NR, no response; PD, progressive disease; ME, markers elevated; NETD, not evaluable, toxic death; NE, not evaluable; VAB, cisplatin carboplatin bleomycin cyclophosphamide vinblastine; PVB, cisplatin carboplatin bleomycin vinblastine; BEP, bleomycin etoposide (VP16) cisplatin carboplatin; EP, etoposide platinum; VP16, etoposide; CY, cyclophosphamide; BM, bone marrow; PBSC, peripheral blood stem cell; CNS, central nervous system; NA, not applicable; ITT, intention to treat. a FU completeness index 90% (no planned 2 nd transplant 90%; planned 2 nd transplant (ITT) 89%). b The chi-square test was used for discrete covariates; the Kruskal-Wallis test was used for continuous covariates. c First transplant. d Other histology includes: Mixed malignant germ cell (n 1). Mixed germ cell cancer (n 2). Rhabdomyosarcoma alveolar (n 1). Unknown (n 1). e Cancers that had a mixture of seminoma and nonseminoma components were classified as nonseminoma. Classification for good, intermediate, and poor risk: Nonseminoma: Good: all of the following: AFP 1000 ng/ml, HCG 5000 IU/L, and LDH 1.5 upper limit of normal; nonmediastinal primary; no nonpulmonary visceral metastasis. Intermediate: all of the following: AFP ,000 ng/ml, HCG ,000 IU/L, or LDH upper limit of normal; nonmediastinal primary site; no nonpulmonary visceral metastasis. Poor: any of the following: AFP 10,000 ng/ml, HCG 50,000 IU/L, or LDH 10 upper limit of normal; mediastinal primary site; nonpulmonary visceral metastasis present. Seminoma: Good: no nonpulmonary visceral metastasis. Intermediate: nonpulmonary visceral metastasis present. f There were 63 nonseminoma cases with unknown information. g Mead GM: International consensus prognostic classification for metastatic germ cell tumours treated with cisplatinum-based chemotherapy: final report of the International Germ Cell Cancer Collaborative Group (IGCCCG). Proc Am Soc Clin Oncol 1995;14:235. Classification for AFP (ng/ml) and HCG (ng/ml): Good: 1000 ng/ml. Intermediate: ,000 ng/ml. Poor: 10,000 ng/ml. h There were 64 nonseminoma cases with unknown information. i Other initial surgery were: B/L RP node (n 1). Craniotomy/orchiectomy (n 1). j Other chemotherapy were: Carboplatin adriamycin (n 1). Cisplatin adriamycin (n 1). Cisplatin thiotepa (n 1). Cisplatin alone (n 2). Cyclophosphamide gemcitabine (n 1). k Patients received: Initial surgery only (n 1). Surgery only (n 1). l Patients received: Initial surgery only (n 1). Surgery only (n 1). m Median number of salvage attempts: 2. n First line (n 44); second line (n 147); third line (n 73); fourth line (n 21). o Patients who did not received chemotherapy (n 3), cisplatin regimen (n 4), or missing (n 17) were excluded in the multivariate analysis phase. p Second transplant. q There were 38 cases with unknown CD34 information. r 10 of 87 patients had a different conditioning regimen for the 2 nd transplant. s Other conditioning regimen includes: Carboplatin other (n 28). Others (n 23).

9 786 H. M. Lazarus et al. Table 3. Univariate Probabilities of Transplant Outcomes among Patients Who Underwent Bone Marrow and/or Peripheral Blood Stem Cell Autologous Transplantation for Testicular Cancer Reported to the CIBMTR, from 1989 to 2002, by Intended Number of Transplants No Planned 2nd Transplant Planned 2nd Transplant (ITT) Outcome Event N Prob (95% CI) N Prob (95% CI) P-value* TRM year 10 (6-14) 3 3 years 11 (7-16) 4 5 years 11 (7-16) 4 (1-9).02 Progression/relapse year 42 (35-49) 53 3 years 49 (42-56) 61 5 years 51 (43-58) 62 (52-72).07 PFS year 48 (41-55) 44 3 years 40 (33-47) 35 5 years 38 (31-45) 34 (25-44).50 Overall survival year 64 (57-70) 67 3 years 47 (40-54) 44 5 years 42 (35-49) 35 (25-46).29 TRM indicates treatment-related mortality; PFS, progression-free survival; PROB, probability; CI, confidence interval; ITT, intention to treat. *Probabilities of treatment-related mortality and progression/relapse were calculated using the cumulative incidence estimate. Progression-free survival and overall survival were calculated using the Kaplan-Meier product limit estimate. Time to events measured from date of first transplant. ure (4%), infection (4%), and hemorrhage (2%) accounted for many deaths in the planned single autotransplant population. DISCUSSION Several early transplant series indicate that a single course of high-dose chemotherapy and autologous HSCT in the management of relapsed or refractory germ cell cancers could induce durable complete remissions (CR) in 23%-34% of patients at 2 years or beyond with TRM ranging from 3%- 22% [11,16,27,28]. Sophisticated supportive care technologies were not routinely available during that era; tandem transplants were not considered feasible because of severe mucositis, infection, bleeding, and renal injury. Nichols et al. [29] reported in 1989 a phase I/II high-dose carboplatin (dose escalation) and etoposide (fixed dose) autologous HSCT study in 33 testis cancer patients who had received at least 2 prior cisplatin-based regimens or were considered cisplatinrefractory. Seven patients (21%) died from treatment, and only 20 patients received tandem autotransplants, yet 4 were long-term disease-free survivors. The Eastern Cooperative Oncology Group conducted a phase II study using tandem autologous HSCT (carboplatin and etoposide) in 40 patients (22 of 38 evaluable received a second transplant); CCR was seen in 5 patients [30]. More recently, the group at Indiana University reported a case series of 65 relapsed germ cell cancer patients (excluding mediastinal primaries) who received tandem autologous HSCT (carboplatin 2100 mg/m 2 and VP mg/m 2 ) separated by a period of 4-5 weeks [12]. No patient in this relatively large series experienced trans- Figure 1. Adjusted probability of progression-free survival after autotransplants for testicular cancer, by intended number of transplants. Figure 2. Adjusted probability of overall survival after autotransplants for testicular cancer, by intended number of transplants.

10 Tandem Autotransplants in Testis Cancer 787 Table 4. Logistic Regression Model for Likelihood of Receiving a Planned Second Transplant and Propensity Score Analyses Factor Level N OR (95% CI) P-value Logistic regression model: Disease status prior to 1st transplant.016 No evidence of disease, either surgically or clinically defined Serum cancer marker elevation only ( ).79 Residual cancer with elevated markers ( ).029 Residual cancer with normal markers ( ).82 Missing ( ).99 Stage.012 Testis primary testis only Retroperitoneal / testis ( ).003 Widespread involvement or CNS involvement ( ).051 Variables N RR of Death (95% CI) P-value Propensity score analyses: TRM* Intended number of transplants No planned 2nd transplant Planned 2nd transplant ( ).044 Progression/relapse* Intended number of transplants No planned 2nd transplant 193 Planned 2nd transplant ( ).54 Progression-free survival* Intended number of transplants.032** Within first 9 months after transplant No planned 2nd transplant 193 Planned 2nd transplant ( ).026 Beyond first 9 months after transplant No planned 2nd transplant 193 Planned 2nd transplant ( ).16 Overall survival* Intended number of transplants.038** Within first 3 months after transplant No planned 2nd transplant 196 Planned 2nd transplant ( ).01 Beyond first 3 months after transplant No planned 2nd transplant 196 Planned 2nd transplant ( ).91 RR indicates relative risk; CI, confidence interval; OR, odds ratio; CNS, central nervous system. *Model stratified on propensity score strata (5 groups). **Two degree of freedom test of no difference in risk between planned and no planned second transplant both early and late. plant-related death, and with more than 4 years of follow-up, 54% (35 of 65) of patients remain in CCR. Finally, a randomized trial in relapsed advanced germ cell tumors of 3 cycles of vinblastine, ifosfamide, and cisplatin salvage chemotherapy (VeIP) plus 1 cycle of high-dose chemotherapy and autologous HSCT compared with 4 cycles of VeIP showed no survival benefit, suggesting that multiple autotransplants are necessary to achieve a favorable outcome [31,32]. Results of these and other series have been considered promising enough by some transplanters to justify tandem autologous transplantation. In the United States, tandem autologous HSCT using peripheral blood stem cells (PBSC) appears to be a common approach. In our series, a significant percent of patients undergoing planned tandem HSCT had poorer risk features including more advanced disease at diagnosis and greater likelihood of exhibiting cisplatin resistance when compared to subjects where 2 autotransplants were not planned. Despite these characteristics, patients intended to receive tandem transplants had a significantly lower TRM, 3% at 1 year compared to 10% in the planned single transplant group (P.02) (Table 4). This low TRM is similar to the Indiana University published experience. The intensity of the transplant preparative regimens may partly explain this finding. Only 53% of the planned tandem transplant group received a regimen containing 3 or more chemotherapeutic agents in contrast to 78% in the planned single transplant group. Patients who were given a tandem autotransplant had a relapse/progression incidence that did not differ significantly from the planned single transplant group

11 788 H. M. Lazarus et al. (Table 4). Further, PFS and OS did not differ between the 2 transplant groups (Figures 1 and 2) after 9 months from transplant. The probability of survival at 5 years was approximately 40% for both groups. Relapses also appeared to occur infrequently beyond 2 years after autotransplant. At 1 year after transplant, 94 and 41 patients are at risk of relapse in the single and tandem autotransplant arms, and at 2 years there are 79 and 33 patients, respectively. Nonetheless, the major cause of treatment failure, even in the tandem autotransplant group, was disease recurrence; 94% of the deaths in the planned tandem transplant group and 79% in the planned single transplant group were attributed to cancer recurrence. None of the 22 patients (15 single and 7 tandem) who received 3 or more salvage attempts before transplant became longterm survivors. Although these numbers are small, such data suggest that this patient subgroup might best be treated with non-hsct approaches. This study has several limitations including the accrual of patients during a 12 year period at multiple institutions. Although we used statistical techniques to adjust for differences in patient and disease characteristics associated with whether a single or tandem autologous HSCT was performed, residual patient selection bias as determined by the transplant centers may affect our results. Additionally, prognostic data on serum cancer markers at presentation was available for only 60% of patients. Our finding that high-risk patients have a greater likelihood of relapse with tandem transplantation after 9 months following HSCT should be interpreted with caution, as small numbers of patients were at risk in that time period. We could not definitively determine the role of tandem versus single autologous transplantation because of these limitations. Our observational data suggest that planned tandem autologous HSCT for testicular cancer have lower TRM than planned single autotransplants but similar long-term outcomes, even though the patients selected for tandem autotransplant may be at higher risk for cancer recurrence. Factors considered important by transplant centers when selecting patients for either approach, however, may not be adequately captured in an observational database. Our findings support the conduct of a randomized trial to compare these 2 approaches. Consideration of the costs of both transplant approaches was beyond the scope of this study. There have been no published studies comparing costs of different transplant approaches, nor studies that consider costs of transplantation compared to alternative salvage treatments. This limitation may further justify conduct of a well-designed clinical trial that includes cost analysis. Other strategies to improve long-term cancer control also include sequential conventional chemotherapy and subsequent autotransplant. One group recently reported that repetitive cycles of etoposide, ifosfamide, and cisplatin chemotherapy followed immediately by high-dose chemotherapy and autotransplant as initial therapy for advanced germ cell cancer provided a 73% 5-year OS rate, 76% for gonadal and retroperitoneal versus 67% for mediastinal primaries [28]. Earlier intervention using autotransplant or instituting new regimens that demonstrate antitumor activity with acceptable toxicity, such as gemcitabine and oxaliplatin [33], in heavily pretreated patients, possibly even in the posttransplant setting, may prevent recurrences. These and other phase I/II data need confirmation, although a number of regimens that showed promise in single-arm studies failed to show superior efficacy in phase III trials [32]. ACKNOWLEDGMENTS The authors thank Lawrence H. Einhorn, MD, and Rafat Abonour, MD, for careful manuscript review and thoughtful suggestions. The CIBMTR is supported by Public Health Service Grant U24- CA76518 from the National Cancer Institute, the National Institute of Allergy and Infectious Diseases, and the National Heart, Lung and Blood Institute; Office of Naval Research; Health Services Research Administration (DHHS); and grants from AABB, Abbott Laboratories; Aetna; AIG Medical Excess; American Red Cross; Amgen, Inc.; Anonymous donation to the Medical College of Wisconsin; AnorMED, Inc.; Astellas Pharma US, Inc.; Berlex Laboratories, Inc.; Biogen IDEC, Inc.; Blue Cross and Blue Shield Association; BRT Laboratories, Inc.; Celgene Corp.; Cell Therapeutics, Inc.; CelMed Biosciences; Cubist Pharmaceuticals; Dynal Biotech, LLC; Edwards Lifesciences RMI; Endo Pharmaceuticals, Inc.; Enzon Pharmaceuticals, Inc.; ESP Pharma; Gambro BCT, Inc.; Genzyme Corporation; GlaxoSmithKline, Inc.; Histogenetics, Inc.; Human Genome Sciences; International Waldenstrom Macroglobulinemia Foundation; Kirin Brewery Company; Ligand Pharmaceuticals, Inc.; Merck & Company; Millennium Pharmaceuticals; Miller Pharmacal Group; Milliman USA, Inc.; Miltenyi Biotec; National Center for Biotechnology Information; National Leukemia Research Association; National Marrow Donor Program; Nektar Therapeutics; NeoRx Corporation; Novartis Pharmaceuticals, Inc.; Novo Nordisk Pharmaceuticals; Ortho Biotech, Inc.; Osiris Therapeutics, Inc.; Pall Medical; Pfizer, Inc.; Pharmion Corp.; Protein Design Labs, Inc; QOL Medical; Roche Laboratories; StemCyte, Inc.; Stemco Biomedical; StemSoft Software, Inc.; SuperGen, Inc.; Sysmex; The Marrow Foundation; THERAKOS, a Johnson & Johnson Co.; University of Colorado Cord Blood Bank; Valeant Pharmaceuticals; ViaCell, Inc.; ViraCor Laboratories; WB Saunders Mosby Churchill; Wellpoint, Inc.; and Zelos Therapeutics, Inc.

12 Tandem Autotransplants in Testis Cancer 789 REFERENCES 1. Einhorn EH. Testicular cancer: an oncological success story. Clin Cancer Res. 1997;3: Sonneveld DJ, Hoekstra HJ, van der Graaf WT, et al. Improved long term survival of patients with metastatic nonseminomatous testicular germ cell carcinoma in relation to prognostic classification systems during the cisplatin era. Cancer. 2001;91: Loehrer PJ Sr, Gonin R, Nichols CR, et al. Vinblastine plus ifosfamide plus cisplatin as initial salvage therapy in recurrent germ cell tumor. J Clin Oncol. 1998;16: International Germ Cell Cancer Collaborative Group. International Germ Cell Consensus Classification: a prognostic factorbased staging system for metastatic germ cell cancers. J Clin Oncol. 1997;15: Hartmann JT, Nichols CR, Droz JP, et al. Prognostic variables for response and outcome in patients with extragonadal germcell tumors. Ann Oncol. 2002;13: Nichols CR, Williams SD, Loehrer PJ, et al. Randomized study of cisplatin dose intensity in poor-risk germ cell tumors: a Southeastern Cancer Study Group and Southwest Oncology Group protocol. J Clin Oncol. 1991;9: de Wit R, Stoter G, Sleijfer DT, et al. Four cycles of BEP versus an alternating regime of PVB and BEP in patients with poor-prognosis metastatic testicular non-seminoma; a randomised study of the EORTC Genitourinary Tract Cancer Cooperative Group. Br J Cancer. 1995;71: Kaye SB, Mead GM, Fossa S, et al. Intensive induction-sequential chemotherapy with BOP/VIP-B compared with treatment with BEP/EP for poor-prognosis metastatic nonseminomatous germ cell tumor: a Randomized Medical Research Council/ European Organization for Research and Treatment of Cancer study. J Clin Oncol. 1998;16: Motzer RJ, Mazumdar M, Bajorin DF, et al. High-dose carboplatin, etoposide, and cyclophosphamide with autologous bone marrow transplantation in first-line therapy for patients with poor-risk germ cell tumors. J Clin Oncol. 1997;15: Decatris MP, Wilkinson PM, Welch RS, et al. High-dose chemotherapy and autologous haematopoietic support in poor risk non-seminomatous germ-cell tumours: an effective first-line therapy with minimal toxicity. Ann Oncol. 2000;11: Broun ER, Nichols CR, Turns M, et al. Early salvage therapy for germ cell cancer using high dose chemotherapy with autologous bone marrow support. Cancer. 1994;73: Bhatia S, Abonour R, Porcu P, et al. High-dose chemotherapy as initial salvage chemotherapy in patients with relapsed testicular cancer. J Clin Oncol. 2000;18: Beyer J, Kramar A, Mandanas R, et al. High-dose chemotherapy as salvage treatment in germ cell tumors: a multivariate analysis of prognostic variables. J Clin Oncol. 1996;14: Broun ER, Nichols CR, Gize G, et al. Tandem high dose chemotherapy with autologous bone marrow transplantation for initial relapse of testicular germ cell cancer. Cancer. 1997; 79: Motzer RJ, Mazumdar M, Bosl GJ, et al. High-dose carboplatin, etoposide, and cyclophosphamide for patients with refractory germ cell tumors: treatment results and prognostic factors for survival and toxicity. J Clin Oncol. 1996;14: Vaena DA, Abonour R, Einhorn LH. Long-term survival after high-dose salvage chemotherapy for germ cell malignancies with adverse prognostic variables. J Clin Oncol. 2003;21: Statistics NCfH. National Hospital Discharge Survey for 1990 and 1991, in Branch HCS. Hyattsville, MD: U. S. Department of Health and Human Services, Public Health Service, Centers for Disease Control. 18. Graves EJ. Detailed diagnoses and procedures, National Hospital Discharge Survey, Vital Health Stat. 1991;13: Curtis RE, Rowlings PA, Deeg HJ, et al. Solid cancers after bone marrow transplantation. N Engl J Med. 1997;336: Silberman G, Crosse MG, Peterson EA, et al. Availability and appropriateness of allogeneic bone marrow transplantation for chronic myeloid leukemia in 10 countries. N Engl J Med. 1994; 331: Kaplan E. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53: Gooley TA, Leisenring W, Crowley J, et al. Estimation of failure probabilities in the presence of competing risks: new representations of old estimators. Stat Med. 1999;18: Rosenbaum P, Rubin D. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70: Rosenbaum P, Rubin D. Reducing bias in observational studies using subclassification on the propensity score. J Am Stat Assoc. 1984;79: Andersen PK, Klein JP, Zhang MJ. Testing for centre effects in multi-centre survival studies: a Monte Carlo comparison of fixed and random effects tests. Stat Med. 1999;18: Mead GM. International consensus prognostic classification of metastatic germ cell tumours treated with cisplatinum based chemotherapy: final report of the International Germ Cell Cancer Collaborative Group (IGCCCG). Proc Am Soc Clin Oncol Droz JP, Pico JL, Ghosn M, et al. Long-term survivors after salvage high dose chemotherapy with bone marrow rescue in refractory germ cell cancer. Eur J Cancer. 1991;27: Schmoll HJ, Kollmannberger C, Metzner B, et al. Long-term results of first-line sequential high-dose etoposide, ifosfamide, and cisplatin chemotherapy plus autologous stem cell support for patients with advanced metastatic germ cell cancer: an extended phase I/II study of the german testicular cancer study group. J Clin Oncol. 2003;21: Nichols CR, Tricot G, Williams SD, et al. Dose-intensive chemotherapy in refractory germ cell cancer a phase I/II trial of high-dose carboplatin and etoposide with autologous bone marrow transplantation. J Clin Oncol. 1989;7: Nichols CR, Andersen J, Lazarus HM, et al. High-dose carboplatin and etoposide with autologous bone marrow transplantation in refractory germ cell cancer: an Eastern Cooperative Oncology Group protocol. J Clin Oncol. 1992; 10: Pico JL, Rosti G, Kramar A, et al. A randomised trial of high-dose chemotherapy in the salvage treatment of patients failing first-line platinum chemotherapy for advanced germ cell tumours. Ann Oncol. 2005;16: Motzer RJ, Sheinfeld S. Poor-risk testicular cancer and highdose chemotherapy. J Clin Oncol. 2003;21: Kollmannsberger C, Beyer J, Liersch R, et al. Combination chemotherapy with gemcitabine plus oxaliplatin in patients with intensively pretreated or refractory germ cell cancer: a study of the German Testicular Cancer Study Group. J Clin Oncol. 2004;22: