J Clin Oncol 23:6149-6156. 2005 by American Society of Clinical Oncology INTRODUCTION

VOLUME 23 NUMBER 25 SEPTEMBER 1 2005 JOURNAL OF CLINICAL ONCOLOGY O R I G I N A L R E P O R T Outcome Analysis for Patients With Elevated Serum Tumor Markers at Postchemotherapy Retroperitoneal Lymph Node Dissection Stephen D.W. Beck, Richard S. Foster, Richard Bihrle, Lawrence H. Einhorn, and John P. Donohue From the Indiana University School of Medicine, Indianapolis, IN. Submitted January 12, 2005; accepted April 25, 2005. Authors disclosures of potential conflicts of interest are found at the end of this article. Address reprint requests to Stephen D.W. Beck, MD, Department of Urology, Indiana Cancer Pavilion, 535 N Barnhill Dr, Ste 420, Indianapolis, IN 46202; e-mail: stdbeck@iupui.edu. 2005 by American Society of Clinical Oncology 0732-183X/05/2325-6149/$20.00 DOI: 10.1200/JCO.2005.11.684 A B S T R A C T Purpose To evaluate the therapeutic benefit of postchemotherapy retroperitoneal lymph node dissection (PCRPLND) in patients with persistently elevated serum tumor markers. Patients and Methods One hundred fourteen patients with metastatic germ cell cancer with elevated serum tumor markers after first-line (50 patients) or second-line chemotherapy (64 patients) who underwent PCRPLND between 1977 and 2000 with a minimum follow-up of 2-years were included in this retrospective study. Results The 5-year overall survival was 53.9%. Sixty-one patients (53.5%) are alive with a medium follow-up of 72 months. Fifty-three patients died of disease, with a medium time to death of 8.0 months. Mean preoperative serum alpha-fetoprotein (AFP) and beta-human chorionic gonadotropin ( HCG) levels were 483 ng/ml and 555 mu/ml, respectively, with no difference in 5-year survival (P.2). Retroperitoneal pathology revealed germ cell cancer in 53.5% of patients, teratoma in 34.2% of patients, and fibrosis in 12.2% of patients, with 5-year survival rates of 31.4%, 77.5%, and 85.7%, respectively (P.0001). Predictors of retroperitoneal pathology included an increasing serum AFP or HCG, HCG more than 100 ng/ml, redo retroperitoneal lymph node dissection (RPLND), and second-line chemotherapy. Poor prognostic variables by multivariable analysis included HCG status, serum AFP level, redo RPLND, and germ cell cancer in the resected specimen. Conclusion A subset of patients with elevated serum tumor markers after chemotherapy is curable with surgery. The prognostic factors predictive of outcome in this analysis include an increasing HCG, serum AFP level, redo RPLND, and germ cell cancer in the resected specimen. These factors, along with clinical and surgical experience, should aid in determining the appropriate integration of surgery and chemotherapy in this population. J Clin Oncol 23:6149-6156. 2005 by American Society of Clinical Oncology INTRODUCTION Postchemotherapy retroperitoneal lymph node dissection (PCRPLND) has traditionally been reserved for patients with normalized serum tumor markers and residual radiographic retroperitoneal disease. Persistently elevated serum tumor markers after cisplatin-based chemotherapy have usually been considered a contraindication to surgery because of presumed presence of persistent, active, germ cell elements. Surgery alone in this setting was felt to have a low likelihood of cure, and thus, second- or third-line chemotherapy was often recommended after increasing markers. At Indiana University, a select population of patients with elevated serum tumor markers has undergone PCRPLND with curative intent. The purpose of this article is 6149

Beck et al to determine the therapeutic benefit of surgery in this clinical setting, identify clinical and pathologic predictors of outcome, and test the hypothesis that surgery is curative in a subset of patients with elevated serum tumor markers. PATIENTS AND METHODS A retrospective review of the testis cancer database at Indiana University Medical Center from 1977 to 2000 was performed to identify all patients who have undergone PCRPLND. One thousand two hundred eighty patients were identified. From this population, we selected those patients with elevated serum tumor markers after first- or second-line chemotherapy at time of PCRPLND. Additionally, at least 2 years of follow-up was required. One hundred twenty-three patients were identified. Nine patients with incomplete medical records were excluded from the data set. One hundred fourteen patients were included in the study cohort. All patients received cisplatin-based chemotherapy, which was induction chemotherapy in 50 patients and second-line chemotherapy in 64 patients (salvage therapy group). Second-line chemotherapy was either vinblastine, ifosfamide, and cisplatin or high-dose chemotherapy. Adjuvant chemotherapy after PCRLND was defined as systemic therapy administered postoperatively to patients with germ cell cancer in the operative specimen. Patients pre- and postoperative characteristics are listed in Table 1. Normal serum tumor marker levels were defined as betahuman chorionic gonadotropin ( HCG) less than 1.56 ng/ml and alpha-fetoprotein (AFP) less than 24 mu/ml ( 15 mu/ml before 1990). At diagnosis, before chemotherapy, the serum tumor marker data were missing in 12 patients. The 50 patients receiving first-line chemotherapy only were subdivided on the basis of serum marker response as follows: (1) serum tumor marker decline with induction chemotherapy that failed to normalize at time of surgery; (2) marker normalization with subsequent increase before surgery; and (3) increasing markers during or within 4 weeks of receiving chemotherapy. Statistical analysis comparing variables in the induction and salvage therapy groups included univariate 2 analysis and Mantel-Haenszel 2 test for ordinal data. The t test was used to compare continuous variables. Survival analysis was performed using the log-rank test and Cox regression analysis. The end point of interest was survival time, which was defined as the time from postchemotherapy surgery to the date of death or date of last follow-up. The date of death of one patient was unknown. RESULTS The presurgical and postsurgical characteristics for the 114 patients are listed in Table 1. For the entire cohort, 51.7% and 44.7% of patients had an elevated preoperative serum AFP and HCG, respectively. Four patients had an elevation of both serum tumor markers. The mean serum tumor marker levels at first diagnosis and before surgery were similar in the first-line and salvage therapy groups. More patients in the salvage therapy group had an increasing AFP (P.0001) and increasing HCG (P.0001) at PCRPLND compared with the first-line group. Retroperitoneal pathology for the 114 patients revealed germ cell cancer in 53.5% of patients, teratoma in 34.2% of patients, and fibrosis in 12.3% of patients. Preoperative clinical features predictive of cancer in the surgical specimen included an increasing serum AFP or HCG, HCG more than 100 ng/ml, redo retroperitoneal lymph node dissection (RPLND), and second-line chemotherapy (Table 2). The incidence of germ cell cancer in the first-line chemotherapy group was 28% compared with 75.8% in the patients receiving second-line chemotherapy (P.0001). Germ cell cancer was detected in 59% and 45% of patients with an elevated AFP and HCG, respectively. Survival Distribution The 5-year overall survival rate for the 114 patients was 53.9% (Fig 1A). Sixty-one patients (53.5%) are alive, and the median follow-up time for survivors was 72 months (range, 24 to 168 months). Fifty-three patients died of disease, with a median time to death of 8.0 months (range, 1 to 84 months). Survival distribution based on serum marker elevation and retroperitoneal pathology are shown in Figures 1B and 1C, respectively. The 5-year overall survival rates based on pathology were 31.4%, 77.5%, and 85.7% for germ cell cancer, teratoma, and fibrosis, respectively (P.0001). The 5-year overall survival rate for the primary chemotherapy group was 81.1%. The median survival time for the 64 patients in the salvage therapy group was 15 months (95% CI, 9 to 21 months), with a 5-year overall survival rate of 33.3% (P.0001). Univariate Survival Analysis Eighteen variables were subjected to univariate analysis for prognostic significance (Table 3). The following factors were associated with an adverse prognosis for the entire cohort of 114 patients: presurgery serum HCG and AFP levels as both continuous and categoric variables, serum tumor marker status (increasing v declining) before surgery, history of marker normalization, first-line versus salvage chemotherapy, redo RPLND, retroperitoneal pathology, and unresectable disease. Number of sites of disease and adjuvant chemotherapy did not influence survival on univariate analysis. The 5-year overall survival rates for patients with germ cell cancer receiving (18 patients) and not receiving (42 patients) adjuvant chemotherapy were 44.4% and 25.7%, respectively (P.067; Fig 2). Multivariable Survival Analysis Eight factors that were significant on univariate analysis were entered into the multivariate regression analysis using the Cox proportional hazards model. Serum AFP more than 1,000 mu/ml (five patients), HCG more than 1,000 ng/ml (four patients), and unresectable disease (three patients) were excluded from multivariate analysis because of small numbers. The following four variables were determined to have independent prognostic significance for 6150 JOURNAL OF CLINICAL ONCOLOGY

Desperation Surgery for Germ Cell Tumors Table 1. Patient Characteristics Entire Cohort (N 114) First-Line Chemotherapy (n 50) Secondary/Salvage Chemotherapy (n 64) Characteristic P Pathology.0001 Cancer 61 53.5 14 28 47 75.8 Teratoma 39 34.2 26 52 13 20.9 Fibrosis 14 12.3 10 20 4 6.3 Tumor marker.3775 AFP 59 51.7 23 39 36 61 Cancer 35 59.3 9 39.1 26 72.2 Teratoma 16 27.1 8 34.8 8 22.2 Fibrosis 8 13.6 6 26.1 2 5.5 HCG 51 46 26 53 25 40 Cancer 23 45 5 19 18 45 Teratoma 22 43 17 65 5 20 Fibrosis 6 12 4 16 2 8 AFP/HCG 4 1 3 Cancer 3 0 3 Teratoma 1 1 0 Fibrosis 0 0 0 Tumor marker AFP, ng/ml Initial Mean 13,126 15,754 11,550.9765 Range 20-130,000 166-54,000 20-130,000 Preoperative Mean 483 220.9 645.0839 Range 15-9,437 30-1,086 15-9,437 HCG, mu/ml Initial Mean 147,561 184,299 105,312.0716 Range 109-704,000 109-704,000 500-545,000 Preoperative Mean 555 35.2 1,057.0562 Range 3.0-11,095 5-224 3-11,095 Period of marker normalization during.0001 systemic therapy, No./total No. AFP 35/59 8/23 27/36 HCG 25/51 3/26 22/25 AFP/HCG 2/4 0/1 2/3 Preoperative marker status.0001 HCG Increasing 22 3 11 19 68 Plateau 18 11 41 7 25 Declining 15 13 48 2 7 AFP Increasing 44 11 46 33 85 Plateau 12 7 30 5 13 Declining 7 6 25 1 2 AFP, ng/ml Initial level 1,000 19 3 16 1,000-10,000 14 7 7 10,000 23 11 12 Preoperative level 100 30 13 17 100-1,000 28 10 18 1,000 5 1 4 (continued on following page) www.jco.org 6151

Beck et al Characteristic Table 1. Patient Characteristics (continued) Entire Cohort (N 114) First-Line Chemotherapy (n 50) Secondary/Salvage Chemotherapy (n 64) HCG, mu/ml Initial level 5,000 9 3 6 5,000-50,000 10 5 5 50,000 24 15 9 Preoperative level 10 16 11 5 10-100 26 14 12 100 13 2 11 Redo RPLND 23 20.1 2 4.1 21 30.6.0004 sites of disease.0815 1 75 66 29 57.5 46 72 2 31 27 16 32.5 15 23 3 8 7.0 5 1.0 3 5.0 Pulmonary metastases 24 21 12 24.5 12 19.3838 Nonpulmonary visceral metastases Liver 5 4.5 4 8.2 1 1.6.1679 Bone 0 0 0 0 0 0 Brain 0 0 0 0 0 0 Additional procedures Nephrectomy 32 29 13 26.5 19 30.6.6346 Vena cava resection 10 9.0 2 4.1 8 13.1807 Aortic replacement 6 5.4 2 4.1 4 6.5.6923 Adjuvant chemotherapy 18 16.2 7 14.3 11 9.9.6237 Abbreviations: AFP, alpha-fetoprotein; HCG, human chorionic gonadotropin; RPLND, retroperitoneal lymph node dissection. P survival in the Cox model: HCG status (increasing v declining or plateau), serum AFP level (continuous variable), redo RPLND (Fig 1D), and germ cell cancer in the surgical specimen (Table 4). In subset analysis, retroperitoneal pathology (with 5-year overall survival rates of 36.9%, 96.5%, and 100% for germ cell cancer, teratoma, and fibrosis, respectively; P.00001) was the single predictor of outcome on multivariable analysis for the 50 patients in the induction group. Significant variables predictive of outcome for the 64 patients in the salvage therapy group are listed in Table 4. Pathology was not predictive of outcome in the salvage therapy group, with no statistical difference in 5-year survival rates between the pathologies (29.6%, 40.4%, and 50% for germ cell cancer, teratoma, and fibrosis, respectively; P.20). DISCUSSION In this retrospective analysis, we have demonstrated a therapeutic benefit of postchemotherapy surgery in patients with elevated serum tumor markers (desperation RPLND). The 5-year overall survival rate in select patients undergoing surgery was 55%. Despite elevated serum tumor markers, only 50% of patients had germ cell cancer, of whom one third had a long-term survival with no observed benefit from adjuvant chemotherapy. Risk factors negatively associated with survival included an increasing preoperative serum HCG, an elevated serum AFP (continuous variable), redo RPLND, and germ cell cancer in the surgical specimen. Previous published studies evaluating the therapeutic benefit of desperation RPLND have reported survival rates ranging from 33% to 75%. 1-6 Most of these studies have reported improved survival in patients with an elevated serum AFP (as opposed to an elevated HCG) and stable serum tumor markers at time of surgery. 2,4,6 In the current series, no absolute level could be detected for either serum tumor marker above which an adverse outcome could be predicted, although an increasing serum HCG at surgery and the greater the absolute serum AFP value preoperatively influenced outcome. Germ cell cancer in this series was present in 61 (53.5%) of 114 patients undergoing desperation surgery. Active germ cell cancer in the retroperitoneum increased the risk of death by 2.5-fold; however, surgery provided a 6152 JOURNAL OF CLINICAL ONCOLOGY

Desperation Surgery for Germ Cell Tumors Factor Table 2. Univariate Analysis to Evaluate Clinical Features Predictive of Retroperitoneal Pathology Patients Cancer Teratoma/Fibrosis Tumor marker.224 AFP 59 35 59.3 24 40.7 HCG 51 23 45.1 28 54.9 Both 4 3 75.0 1 25.0 HCG marker status.025 Plateau/declining 33 11 33.3 22 66.7 Increasing 22 15 68.2 7 31.8 AFP marker status.025 Plateau/declining 19 7 36.8 12 63.2 Increasing 44 31 70.4 13 29.6 HCG 10 mu/ml.864 No 16 6 37.5 10 62.5 Yes 39 20 51.3 19 48.7 HCG 100 mu/ml.025 No 42 16 38.1 26 61.9 Yes 13 10 76.9 3 23.1 AFP 100 ng/ml.20 No 30 15 50.0 15 50.0 Yes 33 23 69.7 10 30.3 Sites of disease.1 1 75 42 56.0 33 44.0 2 31 14 48.4 16 51.6 3 8 4 50 4 50 Redo RPLND.01 No 91 42 46.2 49 53.8 Yes 23 19 82.6 4 17.4 Chemotherapy.0001 First line 50 14 28.0 36 72.0 Second line 64 47 73.4 17 26.6 Abbreviations: AFP, alpha-fetoprotein; HCG, human chorionic gonadotropin; RPLND, retroperitoneal lymph node dissection. P cure for one third of these patients at 5 years, with no observed benefit from adjuvant chemotherapy. Even in the presence of poor prognostic clinical and pathologic features, including elevated serum tumor markers and germ cell cancer in the surgical specimen, RPLND remains a treatment options in this population because of its therapeutic benefit. Despite elevated serum tumor markers and presumed active disease, surgical pathology revealed teratoma in 34.2% of patients and fibrosis in 12.3% of patients, with 5-year overall survival rates of 77.5% and 85.7%, respectively. Other series have also reported an incidence of teratoma or fibrosis ranging from 20% to 40%, despite elevated serum tumor markers at time of PCRPLND. 2,3,6 In our series, clinical parameters predictive of teratoma or fibrosis included declining serum tumor markers, low serum HCG ( 100 ng/ml), and first-line chemotherapy only. Patients who received first-line chemotherapy and who had declining serum tumor markers at surgery had the highest incidence of teratoma/necrosis in the resected specimen ( 75%), which further supports the argument for surgery in lieu of chemotherapy in this population. Possible explanations for nonviable germ cell elements (teratoma/fibrosis) and not finding active germ cell cancer at RPLND in light of elevated serum tumor markers include residual germ cell cancer not resected, nondiagnosed distant disease, and unrecognized germ cell elements in the pathologic specimen. Cerebral metastases was identified early postoperatively in three patients (all dead as a result of disease) who were felt only to harbor fibrosis, underscoring the importance of extensive metastatic evaluation. Alternatively, the elevated presurgical tumor makers may not be related to active germ cell cancer, with false-positive HCG related to marijuana use or cross reactivity with leuteinizing hormone and false-positive AFP elevation as a result of hepatic dysfunction. Likewise, the subset of patients with declining serum tumor markers at PCRLND may have normalized over time, and as such, the surgery would not be classified as desperation with its high anticipation of active germ cell cancer. Van der Gaast et al 7 hypothesized a slowleak phenomenon from postchemotherapy cystic teratoma as an explanation for presurgical serum AFP elevation. www.jco.org 6153

Beck et al Fig 1. (A) Overall survival for 114 patients. (B) Survival based on serum tumor marker elevation. (C) Survival based on pathology. (D) Survival for redo postchemotherapy retroperitoneal lymph node dissection (PCRPLND). AFP, alpha-fetoprotein; HCG, human chorionic gonadotropin. Although the explanation remains unclear, teratoma or fibrosis is identified in the surgical specimen in nearly half the patients at desperation surgery and carries an improved overall prognosis. In the management of residual retroperitoneal disease after chemotherapy, incomplete initial surgical resection necessitating redo surgery portends a poor prognosis. Donohue et al 8 reported a 63% survival rate (118 of 188 patients) for patients undergoing redo PCRPLND versus 86% (529 of 613 patients) for patients undergoing initial PCRPLND. In the current series, 23 patients underwent a redo PCRPLND, with a 5-year overall survival rate of 16% compared with 64% for the 90 patients undergoing initial postchemotherapy surgery (P.0001). Redo RPLND, which is probably the only prognostic variable not absolutely dictated by the biologic aggressiveness of the disease, largely reflects inadequate retroperitoneal technique, underscoring the importance of complete surgical resection at initial RPLND. Postoperative chemotherapy seems to improve diseasefree recurrence in patients with germ cell cancer after first-line chemotherapy but not when administered after second-line chemotherapy. 9 This lack of benefit in the salvage setting is thought to be secondary to the development of tumor chemotherapy resistance. Persistently elevated serum tumor markers after systemic therapy implies a degree of chemotherapy resistance, and in the current series, adjuvant chemotherapy did not improve patient survival. Because of the heterogeneity of the chemotherapy used in the postoperative setting and lack of randomization, no absolute conclusion can be reached; however, adjuvant chemotherapy was not significant in either subgroup and would not be anticipated to improve survival in the salvage therapy population. Administration of adjuvant chemotherapy to patients in the first-line subset with declining markers (resembling patients undergoing standard RPLND), although not demonstrated in this series because of small numbers, may positively influence outcome and must be individualized. An initial tenet in the management of testicular cancer was the exclusion of surgery for patients with elevated serum tumor markers, with chemotherapy being the preferred 6154 JOURNAL OF CLINICAL ONCOLOGY

Desperation Surgery for Germ Cell Tumors Continuous Variables Patients Table 3. Results of Univariate Analysis Events 5-Year Survival (%) Survival (months) Initial HCG 43 12.110 Initial AFP 53 23.700 Presurgery HCG 55 26.002 Presurgery AFP 59 26.018 Categoric variables Presurgery HCG, mu/ml 100 42 16 64.0.003 100 13 10 23.0 7 2.0 to 12.0 1,000 51 22 58.6.01 1,000 4 4 0.0 3 0.0 to 20.0 Presurgery AFP 100 30 10 65.0.044 100 32 19 40.0 22 0.0 to 46.0 1,000 57 24 56.6.0019 1,000 5 5 0.0 11 9.0 to 13.0 Presurgery HCG status.0001 Declining 15 1 93.3 Plateau 18 8 60.0 84 Increasing 22 17 22.7 6 0.0 to 12.0 Presurgery AFP status.0103 Declining 7 1 85.7 Plateau 11 2 81.8 Increasing 44 26 38.7 23 0.0 to 55.0 Serum tumor marker.272 AFP 58 26 53.7 HCG 51 23 56.6 AFP/HCG 4 3 25 3 0.0 to 20.0 Sites of disease.999 1 74 34 54.8 2 31 14 53.0 3 8 4 50.0 22 Redo RPLND.0001 No 90 33 63.9 Yes 23 19 16.3 9 7.0 to 11.0 Pathology.0001 Cancer 60 41 31.4 10 1.0 to 19.0 Teratoma 39 9 77.0 Fibrosis 14 2 85.0 Adjuvant chemotherapy.067 No 42 31 25.7 8.0 4.0 to 12.0 Yes 18 10 44.4 30 0.0 to 69.0 Marker normalization.0001 No 51 11 80.0 Yes 62 41 31.5 17 6.0 to 28.0 Chemotherapy.0001 Primary 49 10 81.1 Secondary 64 42 33.3 15 9.0 to 21.0 Unresectable.0014 Yes 110 49 55.4 No 3 3 0.0 6 1.0 to 11.0 Median Abbreviations: HCG, human chorionic gonadotropin; AFP, alpha-fetoprotein; RPLND, retroperitoneal lymph node dissection. 95% CI P treatment. In this report, however, we have demonstrated a clear therapeutic benefit of surgery in this population; this proves that, even when disease is chemotherapy refractory and despite elevated serum tumor markers, not all disease is systemic and, moreover, that the disease remains curative with aggressive surgery. This being stated, it is strict selection (identifying disease that is likely chemotherapy refractory and likely not systemic) that enables surgery (desperation RPLND) www.jco.org 6155

Beck et al Table 4. Results of Multivariable Analysis Variable Relative Risk 95% CI P Fig 2. Survival stratified according to adjuvant chemotherapy. Entire cohort AFP, continuous 1.0 1.0 to 1.0.048 HCG, continuous 1.0 1.0 to 1.0.546 AFP status 0.49 0.89 to 2.7.413 HCG status 5.06 1.36 to 18.8.015 Preoperative HCG 100 mu/ml 1.077 0.43 to 2.67.872 Redo RPLND 2.86 1.49 to 5.47.001 Cancer v teratoma/fibrosis 2.65 1.25 to 5.58.011 Induction v salvage chemotherapy 0.722 0.31 to 1.66.445 History of marker normalization 1.74 0.68 to 4.46.247 Induction group AFP, continuous 1.0 0.998 to 1.003.852 History of marker normalization 1.62 0.294 to 8.96.579 Adjuvant chemotherapy 0.42 0.93 to 1.88.258 Cancer v teratoma/fibrosis 23.1 3.38 to 157.9.001 Salvage group HCG, continuous 1.0 1.0 to 1.0.436 HCG v AFP 2.88 1.2 to 6.76.015 AFP status 2.04 0.537 to 7.74.295 HCG status 5.49 1.46 to 20.5.011 Redo RPLND 3.63 1.77 to 7.43.001 Cancer v teratoma/fibrosis 1.84 0.824 to 4.12.137 Abbreviations: AFP, alpha-fetoprotein; HCG, human chorionic gonadotropin; RPLND, retroperitoneal lymph node dissection. Marker status reflects statistics for increasing markers v plateau/declining. to be potentially curative in this uncommon clinical setting. General guidelines for selection to proceed with surgery include declining or plateau serum tumor markers after chemotherapy, slowly increasing markers after an initial complete response to either first- or second-line chemotherapy, radiographic resectable residual disease in one or two sites, and increasing markers with resectable disease after exhausting all chemotherapeutic agents. In conclusion, we report a 54% 5-year overall survival rate for patients undergoing postchemotherapy surgery with elevated serum tumor markers. Half of the patients were found to have residual, viable, nonteratomatous germ cell tumor, with 31% alive at 5 years with no observed benefit from adjuvant chemotherapy. This study demonstrates that a subset of patients with elevated serum tumor markers after chemotherapy is curable with surgery. The decision to proceed with surgery in lieu of second- or thirdline chemotherapy includes identifying both patients who are unlikely to obtain a complete response with systemic therapy (and thus require surgery) and patients with resectable tumors that are potentially curable with surgery. Prognostic variables predictive of outcome at desperation surgery include an increasing preoperative HCG, elevated AFP (continuous variable), redo RPLND, and germ cell cancer in the surgical specimen. These factors, along with clinical and surgical experience, should aid in determining the appropriate integration of surgery and chemotherapy in this population. Authors Disclosures of Potential Conflicts of Interest The authors indicated no potential conflicts of interest. REFERENCES 1. Eastham JA, Wilson TG, Russell C, et al: Surgical resection in patients with nonseminomatous germ cell tumor who fail to normalize serum tumor markers after chemotherapy. Urology 43:74-80, 1994 2. Coogan CL, Foster RS, Rowland RG, et al: Postchemotherapy retroperitoneal lymph node dissection is effective therapy in selected patients with elevated tumor markers after primary chemotherapy alone. Urology 50:957-962, 1997 3. Albers P, Ganz A, Hannig E, et al: Salvage surgery of chemorefractory germ cell tumors with elevated tumor markers. J Urol 164:381-384, 2000 4. Murphy BR, Breeden ES, Donohue JP, et al: Surgical salvage of chemorefractory germ cell tumors. J Clin Oncol 11:324-329, 1993 5. Ravi R, Ong J, Oliver RT, et al: Surgery as salvage therapy in chemotherapy-resistant nonseminomatous germ cell tumours. Br J Urol 81:884-888, 1998 6. Wood DP Jr, Herr HW, Motzer RJ, et al: Surgical resection of solitary metastases after chemotherapy in patients with nonseminomatous germ cell tumors and elevated serum tumor markers. Cancer 70:2354-2357, 1992 7. van der Gaast A, Hoekstra JW, Croles JJ, et al: Elevated serum tumor markers in patients with testicular cancer after induction chemotherapy due to a reservoir of markers in cystic differentiated mature teratoma. J Urol 145:829-831, 1991 8. Donohue JP, Leviovitch I, Foster RS, et al: Integration of surgery and systemic therapy: Results and principles of integration. Semin Urol Oncol 16:65-71, 1998 9. Fox EP, Weathers TD, Williams SD, et al: Outcome analysis for patients with persistent nonteratomatous germ cell tumor in postchemotherapy retroperitoneal lymph node dissections. J Clin Oncol 11:1294-1299, 1993 6156 JOURNAL OF CLINICAL ONCOLOGY