Viable Germ Cell Tumor at Postchemotherapy Retroperitoneal Lymph Node Dissection. Can We Predict Patients at Risk of Disease Progression?

Transcription

1 2700 Viable Germ Cell Tumor at Postchemotherapy Retroperitoneal Lymph Node Dissection Can We Predict Patients at Risk of Disease Progression? Philippe E. Spiess, MD 1 Nizar M. Tannir, MD 2 Shi-Ming Tu, MD 2 Gordon A. Brown, MD 1 Ping Liu, MD 3 Ashish M. Kamat, MD 1 Christopher G. Wood, MD 1 James G. Evans, MD 1 Louis L. Pisters, MD 1 1 Department of Urologic Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas. 2 Department of Genitourinary Medical Oncology, The University of Texas M. D. Anderson Cancer Center, Houston, Texas. 3 Department of Biostatistics and Applied Mathematics, The University of Texas M. D. Anderson Cancer Center, Houston, Texas. BACKGROUND. Patients with viable tumor at time of postchemotherapy retroperitoneal lymph node dissection (PC-RPLND) are at an increased risk of disease progression. The objective of the current study was to determine the clinical variables that predict this adverse outcome. METHODS. Between 1980 and 2003, 236 patients with testicular cancer underwent PC-RPLND, 41 of whom (17%) were found to have viable tumor. The authors retrospectively reviewed the patients medical records for pertinent clinical and treatment-related outcomes. At a median follow-up of 3.9 years, 18 patients (44%) had developed disease recurrence and 12 patients (29%) had died of disease. RESULTS. The group of patients who developed postoperative disease recurrence had a larger median dimension of the retroperitoneal mass (7.0 cm and 3.5 cm, respectively; P 5.03). The use of adjuvant chemotherapy after PC-RPLND was less common in those patients developing postoperative disease recurrence (P 5.06). On multivariate analysis, patients classified as being at intermediate or poor risk according to the International Germ Cell Consensus Classification (IGCCC) had a poorer recurrence-free survival (P and P 5.07, respectively). On multivariate analysis, predictors of disease-specific survival (DSS) included an elevated a fetoprotein (AFP) level before PC-RPLND (P 5.003) and postoperative disease recurrence (P 5.02). A serum AFP level >5.3 ng/ml before PC-RPLND was found to be predictive of a poorer DSS (P ). CONCLUSIONS. Patients with viable tumor at the time of PC-RPLND are at an increased risk of disease progression. Clinical variables including classification as intermediate or poor IGCCC risk, a preoperative serum AFP level >5.3 ng/ml, and postoperative disease recurrence help to better define those patients who are at risk of future adverse outcomes. Cancer 2007;110: Ó 2007 American Cancer Society. KEYWORDS: disease recurrence, viable tumor, postchemotherapy retroperitoneal lymph node dissection, testicular cancer. See editorial on pages , this issue. Address for reprints: Louis L. Pisters, MD, Department of Urologic Oncology, The University of Texas M. D. Anderson Cancer Center, 1515 Holcombe Boulevard, Unit 1373, Houston, TX 77030; Fax: (713) ; Lpisters@mdanderson.org Received March 14, 2007; revision received June 20, 2007; accepted June 21, Testicular cancer patients with metastatic nonseminomatous germ cell tumors (NSGCT) traditionally have been treated with systemic chemotherapy followed by retroperitoneal lymph node dissection (PC-RPLND). 1 3 The presence of viable GCT elements in the PC-RPLND specimen places patients at an increased risk of disease progression and death. 4,5 Nevertheless, a small proportion of patients with viable tumor in the surgical specimen fail to develop disease progression regardless of whether they receive additional therapy. 6 The ability to better define the clinical features predictive of disease progression in these patients would allow clinicians to ª 2007 American Cancer Society DOI /cncr Published online 19 Octoberz 2007 in Wiley InterScience (

2 Disease Recurrence With Viable Tumor at RPLND/Spiess et al tailor adjuvant chemotherapy strategies and surveillance protocols appropriately. The aim of the current study was to review our disease-related outcomes in patients found to have viable GCT at the time of PC- RPLND and determine whether clinical parameters allow us to better define those patients who are at an increased risk of subsequent disease progression and death. MATERIALS AND METHODS Patient Selection A retrospective chart review protocol was approved by our Institutional Review Board before the initiation of this study. From our tumor registry, we identified 236 patients who underwent a PC-RPLND between January 1980 and July 2003 for NSGCT. We excluded patients with fibrosis (n 5 97 patients) or teratoma (n 5 98 patients) in the PC-RPLND specimen, with our study population comprised of the 41 patients found to have viable GCT in the surgical specimen. At the time of diagnosis of testicular cancer, all patients underwent a complete medical evaluation, including serum tumor markers and radiologic imaging comprised of chest and abdominal/pelvic computed tomography (CT) scans. Clinical and pathologic staging was determined using the 1997 TNM staging system. Furthermore, patients were categorized into 3 prognostic groups (good-risk, intermediate-risk, and poor-risk) according to the 1997 International Germ Cell Consensus Classification (IGCCC) guidelines. 7 IGCCC risk categories could be assigned to 36 patients; the remaining 5 patients could not be assigned to any risk group because of missing information regarding their serum tumor markers at the time of diagnosis. Pathologic specimens, including the testicular primary neoplasm and other surgically resected specimens, were reviewed by several of our genitourinary pathologists. All patients received systemic chemotherapy before PC-RPLND, with the mean and median number of chemotherapy cycles being 6.3 cycles and 5 cycles, respectively. The specific chemotherapeutic regimen selected was decided on an individual basis by the treating genitourinary medical oncologist. Patients were administered or switched from a firstline to a second-line chemotherapeutic regimen if they had either persistent serum tumor marker elevation and/or a minimal radiographic response to systemic chemotherapy. In addition, all patients undergoing desperation PC-RPLND (ie, PC RPLND performed in the context of elevated preoperative serum tumor markers, which has been defined previously 8 as an AFP level >15.0 ng/ml and/or beta human chorionic gonadotropin [BHCG] >2.2 miu/ ml) and who had received second-line or additional salvage chemotherapy before surgery were deemed to have received the maximum tolerable dose of preoperative chemotherapy. There has been a historical change in our chemotherapeutic regimens for patients with testicular cancer over the past 25 years. Typically in the 1980s, we administered alternating cycles of the combination of cyclophosphamide, doxorubicin, and cisplatin (CISCA) and that of vinblastine and bleomycin (VB) and gave 2 cycles beyond normalization of serum tumor markers for all patients with NSGCT regardless of their clinical stage, with a minimum of 4 cycles administered. 9,10 In the 1990s, we continued administering alternating CISCA and VB regimens, with a 20% dose reduction but with equivalent survival rates. 11 In patients with high-volume disease, we administered alternating dose-dense chemotherapy. 12 Before 1997, patients with small-volume disease received the combination of carboplatin, etoposide, and bleomycin (CEB) 13 ; however, in recent years, 3 cycles of the bleomycin, etoposide, and cisplatin (BEP) regimen or 4 cycles of the etoposide and cisplatin (EP) combination were given to good-risk NSGCT patients. After PC-RPLND, patients were followed routinely by history, physical examination, and serum tumor markers, as well as chest and abdominal/pelvic CT scans every 3 months (or earlier if so desired by the treating genitourinary medical or urologic oncologist). Statistical Analysis Forty-one patients were included in the statistical analysis, with the median values reported for age at diagnosis, serum tumor markers prior to orchiectomy and before PC-RPLND, the number of chemotherapy cycles before surgery, the largest dimension of the retroperitoneal mass noted on radiologic CT scan, the largest dimension of the retroperitoneal mass noted on pathology, and the number of lymph nodes removed. Frequency distributions are reported for the clinical stage and histology of the testicular primary tumor, the presence of lymphovascular invasion, and the clinical stage of testicular cancer. Baseline patient characteristics were compared between the group of patients with viable tumor noted in the PC-RPLND specimen with subsequent disease recurrences versus those without disease recurrences using either the Wilcoxon rank sum test, Fisher exact test, or Mantel-Haenszel chi-square test. The methods of Kaplan and Meier 14 were used to estimate the median recurrence-free survival (RFS) and disease-specific survival (DSS) rates. For the

3 2702 CANCER December 15, 2007 / Volume 110 / Number 12 analysis of RFS, disease recurrence was counted as an event and for the analysis of DSS, disease-specific death was the only event noted. Patients who survived were censored at the date of their last followup. Patients defined as dead of other causes had no evidence of recurrence of the testicular cancer at the time of their last follow-up visit and were censored at their date of death. The number of patients at risk and survival rates at 0, 5, and 10 years were assessed for both DSS and RFS. Using the Cox proportional hazards regression model, 15 we tested the statistical significance of several potential prognostic factors for predicting RFS and DSS. This modeling was performed in a univariate fashion. From this model, we estimated the hazards ratio (HR) for each potential prognostic factor with a 95% confidence interval. All potential prognostic factors with a P value <.10 from the univariate analysis were then included in a saturated model, and backward elimination was used to remove variables from the model based on the likelihood ratio test in the multiple regression analysis. The type of chemotherapeutic regimen used was not included as a variable in our univariate and multivariate Cox proportional hazards regression models of DSS and RFS because >10 different chemotherapeutic regimens were used at our institution over the 23 year period of this study, making it difficult to compare the treatment-related outcomes of these various treatment regimens. A serum AFP value of 5.3 ng/ml before PC-RPLND was chosen as a specific cutoff point because this numeric value of AFP represents the 75% quartile of AFP values and was found to be most optimal in differentiating DSS among our study population. The mean and median length of follow-up for our study population from the time of PC-RPLND was 5.2 years and 3.9 years, respectively (range, years). All statistical analyses were performed using SAS software (version 9.1; SAS Institute Inc, Cary, NC), setting the statistical significance level at P <.05. RESULTS Patient Characteristics The clinical characteristics of our study population are shown in Table 1. The median age of patients before orchiectomy was 27.0 years (range, years) and the median serum tumor values were an AFP level of ng/ml, BHCG level of 21.6 miu/ ml, and lactate dehydrogenase (LDH) level of 452 IU/L. The pathologic stage of the primary tumor was pt1 in the majority of cases (73.2%), with lymphovascular invasion reported in 17.1% of patients. The most common primary tumor histology was mixed NSGCT (85.3%). The clinical stage of testicular cancer was stage IIA in 1 patient (2.4%), stage IIB in 12 patients (29.3%), stage IIC in 20 patients (48.8%), and stage III in 8 patients (19.5%). The median of the largest dimension of the retroperitoneal masses as measured on preoperative radiologic imaging was 3.0 cm (range, 1 17 cm). Patients received a median of 5 cycles of preoperative chemotherapy before PC- RPLND, with the median serum tumor marker values before PC-RPLND being 3.3 ng/ml for AFP, 1.0 miu/ ml for BHCG, and IU/L for LDH. PC-RPLND was performed in 25 patients after first-line chemotherapy and in 8 patients after second-line chemotherapy; the remaining 8 patients underwent desperation PC-RPLND. All patients underwent a full-bilateral template RPLND, with a median of 18 lymph nodes removed per case. All patients were deemed to have undergone a complete surgical resection at the time of PC-RPLND, with 3 patients undergoing additional procedures at the time of surgery (nephrectomies in 2 patients and an aortic graft in 1 patient). Our overall postoperative complication rate (occurring <1 month from the date of PC- RPLND) was 32%, with the major complications being chylous ascites (n 5 4 patients), ileus (n 5 3 patients), and atelectasis (n 5 3 patients). In addition, there was 1 postoperative death that occurred secondary to sepsis. All the patients in the current study had viable GCT elements, with the histology consisting of mixed NSGCT in 36 patients (87.8%), embryonal carcinoma in 3 patients (7.3%), and yolk sac tumor in 2 patients (4.9%). Thirty-six of the 41 patients received adjuvant chemotherapy after PC-RPLND. Of all patients (N 5 236) treated with PC-RPLND at our center over the study period, the incidence of viable cancer at time of PC-RPLND has dramatically declined (P 5.02) from 34.6% in the 1980s to 18.0% in more recent years ( ). Survival Analysis All 41 patients with viable GCT elements in the surgical specimen were included in our survival analysis, with 25 patients still alive, 12 dead of disease, and 4 dead of other causes at the time of analysis. Eighteen of the 41 patients in the current study experienced a postoperative disease recurrence, with the median RFS being 3.0 years. The 5-year DSS and RFS rates were 71% and 50%, respectively, with the 10-year DSS and RFS rates being 66% and 50%, respectively. The Kaplan-Meier DSS and RFS analyses are shown in Figure 1.

4 Disease Recurrence With Viable Tumor at RPLND/Spiess et al TABLE 1 Clinical Characteristics of the Study Population (n 5 41) Variable Mean Median Range Age at diagnosis, y ( ) Tumor markers prior to orchiectomy: AFP (ng/ml) (1.1 65,255) BHCG (miu/ml) (11 93,000) LDH (IU/L) ( ) Tumor markers prior to PC-RPLND: AFP (ng/ml) ( ) BHCG (miu/ml) ( ) LDH (IU/L) (15.1,22014) No. of chemotherapy cycles prior to PC-RPLND (2 16) Largest dimension of retroperitoneal mass (radiology), cm (1 17) Largest dimension of retroperitoneal mass (pathology), cm (0.8 21) No. of lymph nodes removed (3 48) Variable No. (%) Clinical stage of primary testicular tumor T1 30 (73.2%) T2 4 (9.8%) T3 4 (9.8%) Tx 3 (7.3%) Histology of primary testicular tumor Mixed NSGCT 35 (85.3%) Embryonal 3 (7.3%) Teratoma 3 (7.3%) Lymphovascular invasion No 34 (82.9%) Yes 7 (17.1%) Clinical stage IIA 1 (2.4%) IIB 12 (29.3%) IIC 20 (48.8%) III 8 (19.5%) IGCCC risk category Good risk 18 (43.9%) Intermediate risk 5 (12.2%) Poor risk 13 (31.7%) Missing 5 (12.2%) AFP indicates a fetoprotein; BHCG, b-human chorionic gonadotropin; LDH, lactate dehydrogenase; PC-RPLND, postchemotherapy retroperitoneal lymph node dissection; NSGCT, nonseminomatous germ cell tumor; IGCCC, International Germ Cell Consensus Classification. FIGURE 1. Kaplan Meier estimate of disease-specific survival (DSS) and recurrence-free survival (RFS) for all patients with viable tumor in the postchemotherapy retroperitoneal lymph node dissection specimen. Comparison of Patients With and Without Disease Recurrence Of the 41 patients with viable tumor noted in the PC-RPLND specimen, 18 (44%) developed disease recurrence after surgery. The clinical characteristics of patients with or without disease recurrences after PC-RPLND were contrasted as shown in Table 2. The group of patients with viable tumor in the PC- RPLND specimen who developed a disease recurrence had a larger median dimension of the retroperitoneal mass (7.0 cm and 3.5 cm, respectively; P 5.03). Furthermore, all patients developing postoperative disease recurrence had received adjuvant chemotherapy, and the use of adjuvant chemotherapy after PC-RPLND was more common than in those patients who did not develop postoperative disease

5 2704 CANCER December 15, 2007 / Volume 110 / Number 12 TABLE 2 Comparison of Clinical Characteristics of Patients With Viable Tumor With or Without Disease Recurrence Baseline variable Patients without recurrence (n 5 23) Patients with recurrence (n 5 18) P* Median age at diagnosis, y Median tumor markers prior to orchiectomy: AFP (ng/ml) BHCG (miu/ml) LDH (IU/L) Clinical stage of testicular primary tumor T T2 1 3 T3 2 2 Tx 1 2 Lymphovascular invasion No Yes 3 4 Histology of testis primary tumor Mixed NSGCT Embryonal 1 2 Teratoma 1 2 Median tumor markers prior to PC-RPLND: AFP (ng/ml) BHCG (miu/ml) LDH (IU/L) Median no. of chemotherapy cycles prior to PC-RPLND Clinical stage IIA IIB 8 4 IIC 12 8 III 2 6 Histology of PC-RPLND specimen Mixed NSGCT Pure embryonal 3 0 Yolk sac 1 1 Median of the largest dimension of retroperitoneal mass (pathology),cm Use of adjuvant chemotherapy No Yes Median length of follow-up (range),y 4.4 ( ) 2.0 ( ).78 AFP indicates a fetoprotein; BHCG, b-human chorionic gonadotropin; LDH, lactate dehydrogenase; NSGCT, non-seminomatous germ cell tumor; PC-RPLND, postchemotherapy retroperitoneal lymph node dissection. * P values were derived from either the Wilcoxon rank sum, Fisher exact, chi square, or Fisher Freeman Halton test. recurrence (P 5.06). All 5 of the patients who did not develop postoperative disease recurrence despite not receiving postoperative adjuvant chemotherapy had normalized preoperative and postoperative serum tumor markers and had focal, viable GCT elements in the PC-RPLND specimen (2 positive lymph nodes). We did not note a significant difference between these 2 groups in terms of age at diagnosis, serum tumor markers before orchiectomy or before PC-RPLND, stage of the primary testicular tumor, histology of the testicular neoplasm, the presence of lymphovascular invasion in the primary tumor, the amount of preoperative chemotherapy administered, clinical stage of the disease, and pathology of the PC-RPLND specimen. There was no statistically significant difference (P 5.78) noted in the median durations of follow-up between the group of patients with and those without disease recurrence. Predictors of RFS, DSS, and Overall Survival Univariate and multivariate Cox proportional hazards analyses of potential predictors of RFS are shown in Table 3. On multivariate analysis, patients classified as being at intermediate or poor risk according to IGCCC criteria were found to have a worse RFS (P and P 5.07, respectively). Univariate and multivariate Cox proportional hazards analyses of potential predictors of DSS are shown in Table 4. On multivariate analysis, an elevated serum AFP before PC-RPLND (P 5.003) and the presence of a post-

6 Disease Recurrence With Viable Tumor at RPLND/Spiess et al TABLE 3 Factors Associated With Recurrence Free Survival Variable No. of Patients (recurrence) Median survival,years Univariate P value HR 95% CIfor HR Multivariate P value HR 95% CI for HR Age at diagnosis, y 41 (18) Presentation of testicular cancer Local or no symptoms 31 (12) Systemic or both symptoms 10 (6) Clinical stage of primary tumor T1 30 (11) T2 or higher 11 (7) Tumor markers prior to orchiectomy AFP 31 (12) BHCG 32 (14) LDH 21 (7) Presence of lymphovascular invasion No 34 (14) Yes 7 (4) Clinical stage IIA/IIB 13 (4) 1.00 IIC/III 28 (14) Tumor markers prior to PC-RPLND AFP 38 (15) B HCG 38 (15) LDH 39 (16) IGCCC risk category Good risk 18 (3) Intermediate risk 5 (5) Poor risk 13 (8) Missing 5 (2) Context of PC-RPLND First-line chemotherapy 25 (8) 1.00 Second-line chemotherapy 8 (6) Chemotherapy cycles prior to PC-RPLND 41 (18) Largest dimension of retroperitoneal mass (radiology), cm 25 (10) Largest dimension of retroperitoneal mass (pathology), cm 36 (15) Other procedure at time of PC-RPLND No 26 (8) 1.00 Yes 15 (10) Postoperative complications No 32 (13) Yes 9 (5) No. of positive lymph nodes 38 (17) No. of lymph nodes removed 41 (18) Systemic adjuvant therapy No 5 (0) 1.00 Yes 36 (18) HR indicates hazards ratio; 95% CI, 95% confidence interval; AFP, a fetoprotein; BHCG, b-human chorionic gonadotropin; LDH, lactate dehydrogenase; PC-RPLND, postchemotherapy retroperitoneal lymph node dissection; IGCCC, International Germ Cell Consensus Classification. operative disease recurrence (P 5.02) were found to be predictive of poorer DSS. The presence of a serum AFP level >5.3 ng/ml before PC-RPLND was predictive of poorer DSS, as shown in Figure 2. There were 9 patients with a preoperative serum AFP level >5.3 ng/ml, with all of these patients having significantly higher serum AFP levels (median of 6279 ng/ml) before orchiectomy. A Kaplan Meier survival analysis contrasting the DSS of patients with or without postoperative recurrences demonstrated the prognostic impact of this variable (P 5.02) (Fig. 3). On multivariate analysis of predictors of overall survival, advanced clinical stage of the primary tumor (T2 or higher; P 5.01), an elevated serum AFP before PC-RPLND (P 5.05), an increased amount of preoperative chemotherapy (> 5cycles;P 5.002), and the development of a postoperative complication (P 5.03) were all found to be associated with a poorer overall survival.

7 2706 CANCER December 15, 2007 / Volume 110 / Number 12 TABLE 4 Factors Associated With Disease-Specific Survival Variable No. of patients (DSD) Median survival, years Univariate P value HR 95% CI for HR Multivariate P value HR 95% CI for HR Age at diagnosis, y 41 (12) Presentation of testicular cancer Local or no symptoms 31 (7) Systemic or both symptoms 10 (5) Clinical stage of primary tumor T1 30 (6) 1.00 T2 or higher 11 (6) Tumor markers prior to orchiectomy AFP 31 (7) BHCG 32 (8) LDH 21 (5) Presence of lymphovascular invasion No 34 (9) Yes 7 (3) Clinical stage IIA, IIB 13 (3) 1.00 IIC, III 28 (9) Tumor markers prior to PC-RPLND AFP 38 (10) B HCG 38 (10) LDH 39 (10) IGCCC risk category Good risk 18 (2) 1.00 Intermediate risk 5 (2) Poor risk 13 (7) Missing 5 (1) Context of PC-RPLND First-line chemotherapy 25 (4) 1.00 Second-line chemotherapy 8 (4) Chemotherapy cycles prior to PC-RPLND 41 (12) Largest dimension of retroperitoneal mass (radiology), cm 25 (6) Largest dimension of retroperitoneal mass (pathology), cm 36 (10) Other procedure at time of PC-RPLND No 26 (4) 1.00 Yes 15 (8) Postoperative complications No 32 (8) Yes 9 (4) No. of positive lymph nodes 38 (12) No. of lymph nodes removed 41 (12) Postoperative recurrence No 23 (1) Yes 18 (11) Systemic adjuvant therapy No 5 (1) 1.00 Yes 36 (11) DSD indicates disease specific death; HR, hazards ratio; 95% CI, 95% confidence interval; AFP, a fetoprotein; BHCG, b-human chorionic gonadotropin; LDH, lactate dehydrogenase; PC RPLND, postchemotherapy retroperitoneal lymph node dissection; IGCCC, International Germ Cell Consensus Classification. DISCUSSION Patients with viable tumor in the PC-RPLND specimen are at high-risk of disease progression despite the majority of these patients receiving adjuvant systemic chemotherapy. 4,6 In the current study, we report a disease recurrence rate of 44% and a disease-specific mortality rate of 29% at a median follow-up of 3.9 years. Nevertheless, a select subgroup of patients with viable GCT in the surgical specimen did not develop disease progression. Thus, the ability

8 Disease Recurrence With Viable Tumor at RPLND/Spiess et al FIGURE 2. Kaplan Meier disease-specific survival curve of patients based on serum a fetoprotein (AFP) level before postchemotherapy retroperitoneal lymph node dissection. FIGURE 3. Kaplan Meier disease-specific survival curve of patients based on the development of postoperative disease recurrence. to predict those patients at risk of disease progression would allow us to tailor surveillance protocols and adjuvant systemic chemotherapy regimens appropriately. In the current study, patients diagnosed with intermediate risk or poor-risk disease according to the IGCCC criteria were found to have a poorer RFS. In addition, a serum AFP level >5.3 ng/ ml before PC-RPLND and the development of a postoperative disease recurrence are predictive of a poorer DSS in this patient population. The prognostic significance of preoperative serum tumor markers has been shown previously 3,8 and cannot be overemphasized. Patients should be offered adequate preoperative systemic chemotherapy until the normalization of serum tumor markers is achieved. Although there is a subset of patients in whom preoperative serum tumor marker elevation may result from other causes (eg, hypogonadism, hepatotoxicity, and illicit drug use), for the most part preoperative serum tumor marker elevation should be assumed to result from viable GCT elements and such patients should be offered additional systemic chemotherapy until normalization of tumor markers occurs or until maximum tolerable amounts of chemotherapy are administered. A multidisciplinary approach to such patients may be beneficial because insight from both medical and surgical oncologists is provided. The prognostic factors identified in patients with viable GCT at the time of PC- RPLND are clinically useful to the treating urologist and genitourinary oncologist. In our comparison of the clinical variables of the patients with and those without disease recurrence, we observed that the median size of the retroperitoneal masses after surgical resection was significantly larger in the group of patients with viable GCT who developed disease recurrence. Furthermore, the use of adjuvant systemic chemotherapy was higher in the group of patients who developed a disease recurrence, illustrating that as physicians we may have a sense of who may be at risk of disease progression among those patients with viable tumor in the surgical specimen. This point is clearly illustrated by the fact 5 patients were not offered adjuvant chemotherapy and did not develop a postoperative disease recurrence due to the fact that they had a low volume of viable disease in the PC-RPLND specimen (2 positive lymph nodes) and they had normalized preoperative and postoperative serum tumor markers. However, the heterogeneity of our study population is an alternate explanation for why we noted fewer disease recurrences in patients not receiving adjuvant chemotherapy. In addition, our analysis is limited by the small number of patients in this subset and we hope future studies may help to further clarify the prognosis and necessity of adjuvant chemotherapy in patients with a low volume of viable disease at the time of PC-RPLND. In our patient population, we noted that the incidence of viable GCT at the time of PC-RPLND has declined nearly 2-fold (from 34.6% to 18%) from the 1980s until more recently ( ), which we speculate may be the result of more effective preoperative chemotherapeutic regimens. A comparison of the baseline characteristics (eg, serum tumor markers before PC-RPLND, IGCCC risk categories, and clinical tumor stage) of the patient population in the current study was made with those of patients in our testicular cancer database who were found to have fibrosis alone (n 5 97 patients) in the surgical specimen (data not presented). We found no statistically significant differences in terms of any of their baseline patient characteristics.

9 2708 CANCER December 15, 2007 / Volume 110 / Number 12 In our multivariate analysis of predictors of DSS, an elevated serum AFP before PC-RPLND was associated with a poorer DSS, which is consistent with prior studies. 4 We were able to identify a specific value for the serum AFP level of >5.3 ng/ml before PC-RPLND as being associated with a poorer DSS. It is somewhat surprising that an elevated serum AFP level was predictive of DSS and not RFS, for which we do not have a clear explanation. In addition, patients with viable tumor at the time of PC-RPLND in whom a postoperative disease recurrence developed were found to have a nearly 15-fold higher risk of dying of disease than patients without disease recurrence, which strongly supports the need for an aggressive treatment approach to postoperative recurrences. On multivariate analysis, patients classified as being at intermediate or poor risk according to the IGCCC criteria have a poorer RFS, further validating the important prognostic value of this classification system. However, we were quite surprised that patients with intermediate-risk disease had a higher HR of disease recurrence than those with poor-risk disease (HRs of 9.4 and 3.5, respectively), which we attributed to the small number of patients included in this analysis (particularly in the intermediate-risk group). Nevertheless, the clinical parameters identified in this study may help guide the need for adjuvant therapy and the stringency of surveillance. The limitations of the current study must be addressed. First, our study is retrospective and chemotherapeutic regimens have evolved at our center over the past 20 years, which may have affected our treatment outcomes. Second, the surgical pathology data we reported consisted of data derived from our review and/or the interpretation of pathology reports. Consequently, pathology specimens were reported by different genitourinary pathologists at our institution, each of whom may have interpreted the pathologic specimens slightly differently. In conclusion, the current study defines the disease-related outcomes and predictors of disease progression in patients with viable GCT in the PC- RPLND specimen. Patients with viable tumor at the time of surgery are at a high risk of disease recurrence (44%); nevertheless, a subgroup of these patients will not develop disease progression. The clinical parameters that help to better define those patients at risk of such adverse outcomes were a preoperative serum AFP level of >5.3 ng/ml and postoperative disease recurrence. Recognition of these important predictors of outcome may facilitate our management of and follow-up protocols for patients with viable GCT in their PC-RPLND specimens, although we await validation of our findings at other centers. REFERENCES 1. Donohue JP, Thornhill JA, Foster RS, et al. The role of retroperitoneal lymphadenectomy in clinical stage B testis cancer: the Indiana University experience (1965 to 1989). J Urol. 1995;153: Horwich A, Norman A, Fisher C, et al. Primary chemotherapy for stage II non-seminomatous germ cell tumors of the testis. J Urol. 1994;151: Sheinfeld J. The role of adjunctive post-chemotherapy surgery for non-seminomatous germ-cell tumors: current concepts and controversies. Semin Urol Oncol. 2002;20: Hollender A, Stenwig EA, Ous S, et al. Survival of patients with viable malignant non-seminomatous germ cell tumour persistent after cisplatin-based induction chemotherapy. Eur Urol. 1997;31: Spiess PE, Brown GA, Pisters LL, et al. Viable malignant germ cell tumor in the postchemotherapy retroperitoneal lymph node dissection specimen: can it be predicted using clinical parameters? Cancer. 2006;30: Fizazi K, Tjulandin S, Salvioni R, et al. Viable malignant cells after primary chemotherapy for disseminated nonseminomatous germ cell tumors: prognostic factors and role of postsurgery chemotherapy-results from an international study group. J Clin Oncol. 2001;19: International Germ Cell Consensus Classification: a prognostic factor-based staging system for metastatic germ cell cancers. International Germ Cell Cancer Collaborative Group. J Clin Oncol. 1997;15: Stephenson AJ, Bosl GJ, Motzer RJ, et al. Retroperitoneal lymph node dissection for nonseminomatous germ cell testicular cancer: impact of patient selection factors on outcome. J Clin Oncol. 2005;23: Logothetis CJ, Swanson DA, Dexeus F, et al. Primary chemotherapy for clinical stage II nonseminomatous germ cell tumors of the testis: a follow-up of 50 patients. J Clin Oncol. 1987;5: Logothetis CJ, Samuels ML, Selig D, et al. Improved survival with cyclic chemotherapy for nonseminomatous germ cell tumors of the testis. J Clin Oncol. 1985;3: Fizazi K, Do KA, Wang X, et al. A 20% dose reduction of the original CISCA/VB regimen allows better tolerance and similar survival rates in disseminated testicular non-seminomatous germ-cell tumors: Final results of a phase III randomized trial. Ann Oncol. 2002;13: Fizazi K, Prow DM, Do KA, et al. Alternating dose-dense chemotherapy in patients with high volume disseminated non-seminomatous germ cell tumors. Br J Cancer. 2002; 86: Prow DM, Amato RJ, Jones DM, et al. Carboplatin, etoposide, and bleomycin (CEB) combination chemotherapy for the treatment of small volume (SV) metastatic non-seminomatous germ cell tumor of the testis (NSGCT). Proc Am Soc Clin Oncol. 1997;16:340a. 14. Kaplan EL, Meier EP. Nonparametric estimation from incomplete observations. J Am Stat Assoc. 1958;53: Cox DR. Regression models and life tables (with discussion). J R Stat Soc B. 1972;34: