Late relapse of germ cell tumors after cisplatin-based chemotherapy

Annals of Oncology 8: 41-47, 1997. O 1997 Kluwcr Academic Publishers. Printed in the Netherlands.

Original article Late relapse of germ cell tumors after cisplatin-based chemotherapy A. Gerl,1 C. Clemm,2 N. Schmeller,3 M. Hentrich,4 R. Lamerz 5 & W. Wilmanns 1 ' 6 Departments of Internal Medicine lIII, 5II and ^Department of Urology, Klinikum Grofihadem of the University of Munich; "'Department ofInternal Medicine, Clinic of Oncology Bad Trissl; ^Department of Internal Medicine IV, City Hospital Munich Harlaching; 6GSF Forschungszentrum fur Umntlt und Gesundheit, Munich, Germany

Summary Background: Sparse data are available with regard to the incidence, clinical characteristics, therapeutic management and prognosis of male patients with germ cell tumors, who relapse more than two years after completion of cisplatin-based chemotherapy. Patients and methods: A review of 530 patients treated at two institutions from 1978 to April 1994 was conducted. Twenty-five cases of late relapse were identified. Cumulative risk of late relapse was calculated according to the Kaplan-Meier method. Results: 418 of 523 patients (80%) who received their firstline treatment at our institutions were relapse-free at two years. Among these 418 patients 18 cases (4.3%) developed a late relapse. The cumulative risk of late relapse was 1.1% at five years and 4.0% at ten years excluding patients with prior early relapses who carried risks of 9.4% and 29%, respectively (P < 0.0001). No case of late relapse was observed among patients receiving adjuvant chemotherapy. The risk of late relapse

Introduction Cisplatin-based chemotherapy dramatically improved the clinical outlook for male patients with disseminated germ cell tumors [1]. Adjuvant chemotherapy for resected stage II non-seminomatous testicular cancer almost always prevents relapse [2]. Patients with bulky stage II disease, supradiaphragmatic lymph node involvement or visceral metastases attain complete response (CR) rates of approximately 80%, either to chemotherapy alone or to chemotherapy and adjunctive surgery [3]. However, 8% to 16% of aggressively treated patients experience a relapse, the majority of recurrences appearing within the first year after completion of treatment. Fifteen years ago a CR lasting more than two years was considered as equivalent to cure [4]. Late relapses of germ cell tumors are defined as recurrences after a relapse-free interval of more than two years after completion of primary treatment [5]. Some investigators have reported on late relapse after ciplatin-based chemotherapy, but single case reports [6, 7] and small series [8, 9] did not allow for conclusions regarding incidence, risk factors, clinical features, management, and prognosis. Recently, a large single institution experience

was lower in patients with good-risk non-seminomatous germ cell tumors than in poor-risk patients according to Medical Research Council criteria (P < 0.01). Seven further patients were referred from other institutions for treatment of late relapse. At a median follow-up of 38 months (range, 3 to 121) after treatment of late relapse 9 of 25 patients (36%) are continuously disease-free. Six of these nine patients had surgical resection of carcinoma or teratoma as a component of their therapy. Conclusion: The incidence of late relapse after cisplatin-based chemotherapy of germ cell tumors is related to initial tumor burden and is somewhat higher than previously expected. Chemotherapy seems to have only minor curative potential, but localized resectable disease can be cured by surgery. Annual follow-up evaluations allow to detect the majority of late relapses at an asymptomatic stage and should be extended throughout the patient's life. Key words: chemotherapy, cisplatin, germ cell tumor, late relapse, surgery

was published including data on 81 patients with late relapse, 69 of whom had received chemotherapy as part of their primary treatment [10]. We also reviewed our experience and report here on a retrospective study of 25 patients with late relapse after cisplatin-based chemotherapy.

Patients and methods All patients who presented to Klinikum Grosshadern or City Hospital Munich Harlaching with relapse of a testicular or extragonadal germ cell tumor beyond two years from complete or marker negative partial response were included in the study. We reviewed the charts of 530 patients with metastatic germ cell tumors who underwent chemotherapy at our institutions between 1978 and April 1994. Stage Ila/b (retroperitoneal lymph nodes < 5 cm) was primarily treated surgically followed by two to four cycles of adjuvant chemotherapy in the majority of patients [11]. Patients with bulky stage II disease ( > 5 cm), supradiaphragmatic lymph node involvement (stage III) or visceral metastases (stage IV) underwent primary chemotherapy. Until 1984, all patients with non-seminomatous germ cell tumors received their chemotherapy according to the PVB protocol consisting of cisplatin 20 mg/m2 on days 1-5, vinblastine 0.15-0.20 mg/kg on days 1, 2 and bleomycin 30 mg on days 2, 9, 16 [12]. Since 1984, patients with a large tumor burden have been predominantly treated according to the

42 ECBC schedule consisting of etoposide 120 mg/m2 on days 1-4, cisplatin 30 mg/m 2 on days 1-4, bleomycin 15 mg on day 1 (bolus) and 12 mg/m 2 on days 1^1 (24-hour infusion), and cyclophosphamide 300 mg/m 2 on days 1-4 [13]. In 1987 we began to treat patients with low-volume metastatic disease according to the PEB regimen consisting of cisplatin 20 mg/m 2 on days 1-5, etoposide 100 mg/m 2 on days 1-5 and bleomycin 30 mg on days 2, 9, 16 [14]. Patients with residual masses after chemotherapy were referred to adjunctive surgery [15, 16]. Patients with advanced seminomas (lymph nodes > 5 cm or visceral metastases) received cisplatin-ifosfamide-based chemotherapy with either vinblastine or etoposide [17]. Responses to chemotherapy were graded using the Memorial SloanKettering Cancer Center criteria [18]. A complete response (CR) to chemotherapy alone was defined as the disappearance of all clinical, radiographic, and biochemical evidence of disease for a minimum of one month. Patients who achieved no evidence of disease (NED) after resection of mature teratoma were coded as NED-ter. A CR to chemotherapy plus surgery was defined as the excision of all masses, any one or more of which contained viable cancer (NED-ca). A partial response (PR) was defined as a residual radiographic abnormality that remained stable with normal values of serum tumor markers. All other responses were classified as incomplete responses (IR). Follow-up examinations included history, physical examination, biochemical examinations including serum tumor markers human chorionic gonadotropin (HCG) and alpha-fetoprotein (AFP), chest X-ray and abdominal CT scan. Further examinations were performed as indicated by clinical symptoms. During the first two years after commencement of treatment patients were seen every 3 months, at 6 month intervals during the third year, and thereafter annually. The majority of patients was monitored at our institutions. The status of the remaining patients was verified by personal interview of the patients and their physicians. Survival and cumulative risk were calculated using the KaplanMeier method [19]. Survival was measured from date of diagnosis of late relapse to the date of last follow-up or death.

Results Incidence Twenty-five patients were identified who relapsed from a CR (n = 21) or PR (n - 4) more than two years after discontinuation of primary treatment. One of these patients had already experienced a previous late relapse from stage I disease. A second testicular tumor was excluded by ultrasound in all cases. Disease distribution did not argue in favor of a second extragonadal germ cell tumor in any case. Eighteen patients received their primary treatment at our institutions, while the remainder were initially managed elsewhere. Four hundred eighteen of 523 patients (80%) who received first-line treatment at our institutions were relapse-free at two years. Thus, the current incidence of late relapse is 4.3% (18 of 418 patients). Time to late relapse ranged from 25 to 132 months, with a median of 65 months.

receiving adjuvant chemotherapy for resected stage II disease developed a late relapse. Patients with good-risk non-seminomatous germ cell tumors had a lower risk of late relapse than patients with poor-risk disease according to Medical Research Council criteria (P < 0.01) (Tables 1 and 2) [20]. Six of 87 patients (6.9%) with only retroperitoneal disease (stage II), managed by primary chemotherapy, developed a late relapse in comparison with only one of 56 patients (1.8%) with only lung metastases. Patient characteristics at initial diagnosis The median patient age at initial diagnosis was 24 years (range, 17 to 43). Patient characteristics pertaining to the status immediately before primary therapy are summarized in Table 3. Serum tumor markers were available in 24 of the 25 patients. None of the 24 patients presented with normal values for both markers. HCG was elevated in 19 of 24 patients (79%) with a median level of 1240 IU/1 (range, 12 to 1,300,000). AFP was elevated in 21 of 24 patients (88%) with a median concentration of 750 IU/ml (range, 26 to 27,300).

Table 1 Cumulative risk of late relapse. Risk group

% Cumulative risk at 5 years (95% CI)

% Cumulative risk at 10 years (95% CI)

All patients (n = 418)

1.6(0.3-3.0)

6.1 (3.0-9.3)

No early relapse (n = 394) Early relapse (n = 24)

1.1 (0-2.3)

4.0(1.4-6.6)

Non-seminoma (n = 362) Seminoma (n = 56) Curative chemotherapy (n = 306) Adjuvant chemotherapy {n = 112) Good risk NSGCT (« = 169) Poor risk NSGCT 1 (n = 83)

9.4 (0-22) 1.2(0-2.4)

29(6.6-51)

P

< 0.0001

6.6(3.2-10) >0.5

2.0 (0-6.0)

2.0 (0-6.0)

2.2(0.4-4.0)

8.2(4.1-12.4)

0

0

0.6(0-1.9)

4.4(0.5-8.4)

5.7 (0.1-11.2)

20 (7.9-32)

<0.01

Abbreviations: CI - confidence interval; NSGCT - non-seminomatous germ cell tumor. * According to Medical Research Council criteria [20] (Table 2).

Riskfactors

Table 2. Risk groups of non-seminomatous germ cell tumors according to Medical Research Council Criteria [20].

The cumulative risk of late relapse for patients receiving first line treatment at our institutions was 1.1% at 5 years and 4.0% at 10 years, if patients with previous early relapses were excluded who had substantially higher risks of 9.4% and 29%, respectively (Table 1). No patient

Poor risk

Good risk

Liver, brain or bone metastases Mediastinal mass greater than 5 cm 20 or more lung metastases Serum HCG > 10,000 IU/1 and/or AFP > 1000 IU/ml None of the above adverse features

43 Table 3. Patient characteristics at the beginning of primary therapy.

Table 4. Primary treatment and response.

No. of patients Year of primary treatment 1978-1983 1984-1988 1989-1993 Site of primary tumor Testis Retroperitoneum Mediastinum Histology of primary tumor Embryonal carcinoma Teratocarcinoma Mature teratoma (elevated markers) Pure seminoma Seminoma elements (mixed non-seminoma/ seminoma) Stage* II III IV Number of metastatic sites One site or marker only Two sites Three or more sites Sites of disease Retroperitoneum Retroperitoneal mass > 5 cm Lungs Mediastinum Cervical nodes Bone Marker only Tumor markers'5 HCG elevation AFP elevation Medical Research Council risk groupsc Good risk Poor risk

15 6 4 22 2 1 11 11 2 1 9 10 7 8 12 9 4 21 16 8 7 5 1 1 19 21 11 14

* Stage according to Royal Marsden Classification: II = retroperitoneal lymphadenopathy, III = supradiaphragmatic lymph node involvement, IV = visceral metastases. b Data on tumor markers available for 24 patients. c See Table 2.

Primary treatment Primary treatment and response are summarized in Table 4. TWo patients, who began their initial chemotherapy before cisplatin was available in Germany, had been pretreated with non-cisplatin-containing regimens but both patients underwent cisplatin-based chemotherapy at a later stage of their primary therapy. Three patients underwent radiotherapy at some stage during their primary treatment. One patient with unresected low-volume stage II disease received an inadequate primary therapy, since a severe trauma owing to a car accident led to discontinuation of chemotherapy after two cycles; nevertheless, primary therapy resulted in a CR. Seven patients underwent additional chemotherapy for an early relapse which occurred within two years from discontinuation of primary therapy. Two further patients

No. RPLND before chemotherapy Chemotherapy Cisplatin, vinblastine, bleomycin Cisplatin, etoposide, bleomycin Vinblastine, ifosfamide, cisplatin Etoposide, cisplatin, bleomycin, cyclophosphamide Additional chemotherapy for early relapse" Total number of cycles Median/range Postchemotherapy surgery RPLND Thoracotomy Neck dissection Response to primary therapy CR NED-ter NED-ca PR

10 17 6 1 1 7 5/2-10 13 5 1 14 5 2 4

Abbreviations: RPLND - retroperitoneal lymph node dissection; CR complete response; NED-ter - no evidence of disease after resection of teratoma; NED-ca - no evidence of disease after resection of cancer; PR - partial response. a Early relapse defined as < 2 years after discontinuation of primary therapy.

had early relapses with only mature teratoma which were managed surgically. Altogether, primary therapy resulted in a CR in 21 patients and in a PR in four patients. Two of these four patients in PR underwent incomplete resection of mature teratoma, while the remaining two refused adjunctive surgery. These four patients presented with viable cancer at diagnosis of late relapse, one at an initially involved site only, one at an initially involved and another site, and two at other sites only. Patient characteristics at late relapse Sites of late relapse are listed in Table 5. At diagnosis of late relapse both tumor markers were normal in seven of Table 5. Patient characteristics at late relapse. No. of patients (%) Time to late relapse 2-5 years > 5 years Number of metastatic sites One site Two sites Sites of disease Retroperitoneum Mediastinum Pelvis Liver Lungs Cervical nodes Suprarenal gland Abdominal wall

11(44) 14(56) 19(76) 6(24) 12(48) 8(32) 3(12) 3(12) 2(8) 1(4) 1(4) 1(4)

44

n = 25

120

150 MONTHS

Figure 1. Survival from diagnosis of late relapse.

24 patients (29%). Only two patients (8%) presented with abnormal HCG levels at 25 and 46 months, respectively. In both cases, HCG elevation was found concurrently with radiologic detection of relapse. An increase of AFT* occurred in 16 patients (67%), and in 12 of these cases a rise of AFP was documented 6 to 44 months (median 27 months) prior to radiological detection of late relapse. The initial doubling time of serum AFP ranged from 56 to 600 (median 155) days indicating slow progression of occult disease. All patients with preceding AFP increase were asymptomatic at the time of radiologic manifestation of late relapse. In five cases diagnosis of late relapse was not made in connection with routine follow-up. Three of these five patients underwent regular annual follow-up examinations. Management of late relapse Three patients with a CR to primary treatment developed a late relapse with only mature teratoma and were managed by surgery alone. Two of these patients are alive with no evidence of disease at 3 and 82 months, whereas the third developed a second late relapse which was associated by an increase of AFP. The latter patient underwent chemotherapy at second late relapse, but he ultimately died of disease. Twenty-two patients had pathologic or serologic evidence of cancer. Two of these patients refused any salvage treatment. Of the remaining 20 patients 16 were primarily managed by chemotherapy and four by surgery. Only one of the four surgically treated patients remained diseasefree, while the other three were referred to chemotherapy. Overall, 20 patients underwent chemotherapy at some stage during their treatment of late relapse. Eighteen of these 20 patients (90%) were treated according to a protocol consisting of etoposide, ifosfamide and cisplatin. Nine patients underwent postchemotherapy adjunctive surgery, seven of whom harbored viable cancer and two necrosis/fibrosis. Sarcomatous components or other nongerm cell malignancies were not identified in any of these patients. Overall, 12 patients attained disease-free status after

Figure 2. Survival from diagnosis of late relapse comparing subgroups: 1 - patients with lymph node involvement only, 2 - patients with visceral metastases (P < 0.01).

treatment for late relapse, three of whom relapsed again. Ten patients had an IR to treatment of late relapse, and one patient achieved tumor marker normalization but radiologic response could not be assessed because of early toxic death (case report no. 2). After failure of initial management of late relapse ten patients received further treatment which led to NED status in only one patient. Four of these ten patients underwent high-dose chemotherapy, none of whom achieved a CR. After management of late relapse nine patients (36%) remain continuously disease-free at a median follow-up of 38 months (range, 3 to 121). Two of these nine patients had resection of only mature teratoma, and a further patient had resection of a large teratoma containing only small foci of viable cancer. Three patients underwent resection of chemorefractory disease, whereas two further heavily pretreated patients achieved a CR to chemotherapy alone, as complete resection of residual masses disclosed only necrosis/fibrosis. In the latter cases time to late relapse was 80 and 97 months, respectively; observation time after management of late relapse is 38 and 18 months, respectively. One patient with a pure seminoma had a PR to chemotherapy and achieved a CR after consolidation with radiotherapy. One patient with recurrent disease after initial management of late relapse attained a CR to further chemotherapy and is alive with NED at 138 months. Two patients (8%) are currently alive with disease, and 13 (52%) died of disease or treatment-related toxicity. Survival after diagnosis of late relapse is shown in Figure 1. Patients with only lymph node involvement at diagnosis of late relapse had a significantly better prognosis than patients presenting with visceral metastases (Figure 2). Neutropenic infections after chemotherapy caused three deaths, while a fourth patient died after a surgical intervention. Three of the four patients with toxic deaths were heavily pretreated and had evidence of active disease.

45 Case reports Two cases of late relapse are described in the following sections. Case report no. 1 A 30-year-old man underwent right orchiectomy for embryonal carcinoma with seminoma elements in 1984. CT scans revealed a retroperitoneal mass of 10 cm and slightly enlarged mediastinal lymph nodes. Serum HCG and AFP were elevated at 1400 IU/1 and 650 IU/ml, respectively. The patient underwent five cycles of chemotherapy consisting of cisplatin, vinblastine and bleomycin. As postchemotherapy RPLND disclosed viable cancer, two additional cycles of chemotherapy consisting of etoposide, ifosfamide and cisplatin were administered. After discontinuation of treatment the patient remained in CR for 18 months. Thereafter HCG rapidly rose to 329 IU/1. As CT scans and testicular ultrasound were normal, an early relapse with marker elevation only was diagnosed. Three cycles of etoposide, ifosfamide and cisplatin were given, and the patient attained a second CR. Fiftyfour months later an AFP level of 41 IU/ml was found at an annual follow-up visit. The serum concentration of AFP remained slightly elevated for two years, and repeat testicular ultrasound and CT scans did not reveal any lesion. Thereafter AFP rapidly rose to 224 IU/ml, and an abdominal CT scan disclosed an ipsilateral right-sided pelvic mass of 8 cm. Chemotherapy was instituted with etoposide, ifosfamide, and cisplatin, and after three cycles AFP returned to normal. Resection of the pelvic mass revealed only necrosis/fibrosis. The patient is alive with NED status 38 months after completion of treatment for late relapse. Comment. This pattern of late relapse was typical for one half of the patients. A relatively long phase of slowly rising AFP preceded a short phase of rapid increase, in which late relapse was confirmed radiologically. Of note, this patient had a CR to chemotherapy despite extensive prior treatment. Case report no. 2 A 21-year-old man underwent right orchiectomy for a teratocarcinoma. Serum tumor markers returned to normal, and RPLND did not reveal any metastatic lymph node involvement. Two months later a solitary pulmonary metastasis and a mediastinal lymph node infiltrating the sternum were identified. Serum AFP and HCG were elevated at 382 IU/ml and 15 IU/1, respectively. Four cycles of chemotherapy consisting of cisplatin, vinblastine, and bleomycin were administered, and the patient achieved a CR. He underwent regular follow-up evaluations for more than eight years, with normal results. Forty months after the last follow-up visit the patient presented with dyspnoea. CT scan disclosed a mass at the initially involved mediastinal region, and AFP was elevated at 11,400 IU/ml. Four cycles of chemotherapy consisting of etoposide, ifosfamide, and cisplatin were administered.

At the beginning of the fourth cycle AFP returned to normal. After the fourth cycle the patient developed pancytopenia and died of sepsis with renal and pulmonary failure. Post mortem examination was not performed. Comment. This case demonstrates that annual follow-up evaluations should not cease at 5 to 10 years after discontinuation of treatment.

Discussion A recent report by Baniel et al. [10] demonstrated that late relapses of germ cell tumors constitute a distinct entity. At initial presentation germ cell tumors show an aggressive biologic behavior and are highly responsive to cisplatin-based chemotherapy. In contrast, late relapses often show a slow tumor growth and usually respond to chemotherapy poorly. Only two patients of the above mentioned report were continuously disease-free after chemotherapy alone, and it was emphasized that both patients had not received prior chemotherapy [10]. Our data support the contention that late relapses frequently show a poor response to chemotherapy, as only 2 of 20 patients attained a continuous disease-free status. However, it is of note that these two patients were heavily pretreated with 10 and 7 cycles of prior chemotherapy, respectively. A further patient is a long-term disease-free survivor after third-line chemotherapy. Although conclusions have to be drawn with caution, these three cases suggest that chemotherapy may have curative potential in a small subset of chemotherapy-pretreated patients with late relapse. Patients who present with negative tumor markers should always undergo surgery as first mode of treatment to recognize disease reactivation with only mature teratoma which is adequately managed surgically [21]. Moreover, a second non-germ cell neoplasm is an important differential diagnosis [22]. Taking the poor results with primary chemotherapy into account, patients with a small lesion that is deemed completely resectable should be referred to surgery as first mode of treatment. If surgical margins are clear and tumor markers normalize postoperatively, patients probably will not benefit from adjunctive chemotherapy. Reported data [10] and our own experience with primary surgical management of late relapse are limited to a relatively small number of patients and therefore do not allow for definite conclusions, but it seems that cure rates are relatively low in patients with viable cancer. We therefore currently prefer primary chemotherapy, if there is any doubt with regard to the feasibility of complete resection. Patients undergoing primary chemotherapy for unresectable disease and achieving normalization of tumor markers should be always considered for adjunctive surgery, as the chance of harboring viable cancer is high. If tumor markers do not normalize after four cycles of chemotherapy, resection of localized disease should be also reconsidered. Surgery has

46 definite curative potential in chemorefractory disease as shown in three of our patients and in several cases of the aforementioned report [10]. Successful management of late relapse is critically dependent on early recognition. 20 of 23 patients undergoing regular follow-up evaluations had asymptomatic disease at diagnosis of late relapse. It must be emphasized that patients with symptomatic disease carried a poor prognosis; all four patients with viable cancer and symptomatic disease died within one year from diagnosis of late relapse. Furthermore, in concordance with a report by Borge et al. [23] patients presenting with visceral metastases at diagnosis of late relapse carried a poor prognosis. As in the report by Baniel et al. [10] approximately three quarters of our patients presented with elevated tumor markers, whereby AFP was the leading tumor marker of late relapse. Of note, only two patients presented with an increase of HCG, although HCG was elevated at primary diagnosis in a similar proportion as AFP. This finding may be explained by the study of Carl et al. [24] who desribed that HCG producing cells tended to grow faster than AFP producing cells, and were 3-5fold more sensitive to chemotherapy than AFP producing cells. In one half of our patients a rise of AFP preceded radiologic detection by 6 to 44 months. Treatment was not started before radiologic confirmation of late relapse, since AFP elevation may be also caused by a second testicular tumor [25] or by liver dysfunction [26]. Patients with bulky retroperitoneal disease appear to be at an increased risk of developing a late relapse. As tumor markers did not rise in approximately one quarter of late relapses, this patient group should undergo CT scans at annual follow-up evaluations. In the remaining patients history, physical examination, tumor markers and chest X-ray may allow to detect the majority of late relapses at an asymptomatic stage. Annual follow-up should not cease at ten years after treatment, as late recurrences were observed up to 32 years [10]. It is of note that in our series none of 112 patients who underwent primary RPLND followed by adjuvant chemotherapy for stage Ila/b disease developed a late relapse. Baniel et al. compared primary RPLND and primary chemotherapy in the management of stage Ila/b disease and found that patients receiving RPLND had an advantage in terms of late relapse [27]. In a series of 122 patients treated by primary chemotherapy three late relapses were observed at a median follow-up of 5.5 years [28]. In summary, the cumulative risk of late relapse in patients appearing relapse-free at two years after first line chemotherapy is 4% at ten years. The risk of late relapse is correlated to tumor burden at start of primary therapy. Annual follow-up evaluation allows to detect the majority of late relapses at an asymptomatic stage, whereby the tumor marker AFP plays a predominant role. However, follow-up examinations cannot circumvent few cases of symptomatic late relapse due to aggressive tumor biology, and patients should be therefore informed about typical symptoms such as persistent back pain,

cough and dyspnoea, or a mass in the supraclavicular fossa. Whereas chemotherapy has only minor curative potential in the treatment of late relapse, patients with localized resectable disease can be cured by surgery. References 1. Einhorn LH. Treatment of testicular cancer: A new and improved model. J Clin Oncol 1990; 11:1777-81. 2. Motzer RJ. Adjuvant chemotherapy for patients with stage II nonseminomatous testis cancer. Semin Oncol 1995; 22: 641-6. 3. Gerl A, Clemm C, Schmeller N et al. Advances in the management of metastatic non-seminomatous germ cell tumours: A single institution experience. Br J Cancer 1996; 74: 1280-5. 4. Einhorn LH. Testicular cancer as a model for a curable neoplasm: The Richard and Hinda Rosenthal Foundation Award Lecture. Cancer Res 1981; 41: 3275-80. 5. Terebelo HR, Taylor HG, Brown A et al. Late relapse of testicular cancer. J Clin Oncol 1983; 1: 566-71. 6. Embil JM, Weinerman BH, Pascoe EA. Late recurrence of testicular cancer. Can J Surg 1994; 37:165-7. 7. Bokemeyer C, Kuczyk MA, Fichna B et aL Late relapse of testicular cancer presenting as differentiated teratoma: Report of a case and discussion of the clinical implications. Oncol Rep 1995; 2: 615-8. 8. Deleo MJ, Greco FA, Hainsworth JD et aL Late recurrences in long-term survivors of germ cell neoplasms. Cancer 1988; 62: 985-8. 9. Lianes P, Paz-Ares L, Rivera F et al. Late recurrences (LR) in malignant germ cell tumors (MGCT). Ann Oncol 1992; 3 (Suppl 5): 165 (Abstr). 10. Baniel J, Foster RS, Gonin R et al. Late relapse of testicular cancer. J Clin Oncol 1995; 13:1170-6. 11. Gerl A, Clemm C, Kohl P et al. Adjuvant chemotherapy of stage II non-seminomatous testicular cancer. Oncol Rep 1994; 1: 209-12. 12. Einhorn LH, Donohue J. Cis-diamminedichloroplatinum, vinblastine, and bleomycin combination chemotherapy in disseminated testicular cancer. Ann Intern Med 1977; 87: 293-8. 13. Gerl A, Clemm C, Hentrich M et al. Etoposide, cisplatin, bleomycin, and cyclophosphamide (ECBQ as first-line chemotherapy for poor-risk non-seminomatous germ cell tumors. Acta Oncol 1993; 32: 541-6. 14. Williams SD, Birch R, Einhorn LH et al. Treatment of disseminated germ-cell tumors with cisplatin, bleomycin, and either vinblastine or etoposide. N Engl J Med 1987; 316:1435-40. 15 Gerl A, Clemm C, Schmeller N et al. Sequential resection of residual abdominal and thoracic masses after chemotherapy for metastatic non-seminomatous germ cell tumours. Br J Cancer 1994; 70: 960-5. 16. Gerl A, Clemm C, Schmeller N et aL Outcome analysis after postchemotherapy surgery in patients with non-seminomatous germ cell tumours. Ann Oncol 1995; 6:483-8. 17. Clemm C, Hartenstein R, Willich N et al. Vinblastine-ifosfamidecisplatin treatment of bulky seminoma. Cancer 1986; 58: 2203-7. 18. Mencel PJ, Motzer RJ, Mazumdar M et al. Advanced seminoma: Treatment results, survival, and prognostic factors in 142 patients. J Clin Oncol 1994; 12:120-6. 19. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53:457-81. 20. Mead GM, Stenning SP, Parkinson MC et al. The second Medical Research Council study of prognostic factors in non-seminomatous germ cell tumors. J Clin Oncol 1992; 10: 85-94. 21. Gelderman WAH, Oosterhuis JW, Schraffordt Koops H et aL Late recurrence of mature teratoma in non-seminomatous testicular tumors after PVB chemotherapy and surgery. Urology 1989; 33: 10-4. 22. Gerl A, Clemm C, Schmeller N et al. Prognosis after salvage treatment for unselected male patients with germ cell tumours. Br J Cancer 1995; 72: 1026-3Z

47 23. Borge N, Fossa SD, Ous S et al. Late recurrence of testicular cancer. J Clin Oncol 1988; 6:1248-53. 24. Carl J, Christensen TB, von der Maase H. Cisplatinum dose dependent response in germ cell cancer evaluated by tumour marker modelling. Acta Oncol 1992; 31: 749-53. 25. Gerl A, Clemm C, Kohl P et al. Testicular tumor after cisplatinbased chemotherapy for germ cell malignancy. Eur Urol 1994; 25: 216-9. 26. Germa JR, Llanos M, Tabernero JM et al. False elevations of alpha-fetoprotein associated with liver dysfunction in germ cell rumors. Cancer 1993; 72: 2491-4. 27. Baniel J, Roth BJ, Foster RS et al. Cost and risk benefit in the management of clinical stage II non-seminomatous testicular tumors. Cancer 1995; 75: 2897-903. 28. 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: 72-8. Received 24 July 1996; accepted 3 December 1996.

Correspondence to: Dr. Arthur Gerl Medizinische Klinik III Klinikum GroBhadern Ludwig-Maximilians-Universitat Munchen Marchioninistrafie 15 81377 Munchen Germany