Improving the outcome of salvage treatment for non-seminomatous germ cell tumours (NSGCT)

Between 1991–96, 41 patients were treated in this unit for relapsed non-seminomatous germ cell tumours (NSGCT). Twenty-eight patients had raised markers at relapse: 17 required salvage chemotherapy and post-chemotherapy surgery, 11 only chemotherapy. In addition 9 patients received high dose chemotherapy. Overall 16/28 patients (57%) requiring chemotherapy remain alive, 14 (50%) disease free. Of the 17 patients treated with chemotherapy and surgery: 12 remain alive, 10 (59%) with no evaluable disease. Only 4/11 (36%) patients treated with chemotherapy alone remain alive, all in complete remission (CR). For relapse with raised markers, univariate analysis suggests that less than CR to induction therapy, resulting in the presence of residual disease is the most important predictor of poor outcome (P< 0.001). All of the 13 patients relapsing with normal markers remain alive, having been primarily treated surgically. Overall these results indicate an improving outlook for relapsed NSGCT. © 2000 Cancer Research Campaign

Approximately 20-30% of patients with metastatic germ cell tumours will not achieve durable remissions after first line treatment (International Germ Cell Cancer Collaborative Group, 1997). Most suffer relapses or disease progression after a favourable initial response. A variety of salvage chemotherapy regimens have been used, usually involving a combination of cisplatin, ifosfamide and perhaps etoposide or vinblastine (Loehrer et al, 1998). More recently paclitaxel (Motzer et al, 1994) and gemcitabine (Bokemeyer et al, 1999) have been shown to have significant activity. Paclitaxel is therefore currently being investigated in combination with ifosfamide and cisplatin as first line salvage therapy (Motzer et al, 1998). High dose chemotherapy and stem-cell transplantation has also been successfully utilized for these patients (Bhatia et al, 1998).
The essential role of surgery in the treatment of relapsed teratoma is well described (Jansen et al, 1991). Surgery is usually the treatment of choice in relapse with normal markers. Salvage surgery is also vital for post-salvage chemotherapy masses, and can be curative when markers are rising despite salvage chemotherapy (Murphy et al, 1993). If relapse occurs late (several years following initial therapy), in the absence of extensive recurrence or raised markers, salvage surgery may be the primary mode of treatment (Baniel et al, 1995).
Overall the results obtained with salvage therapy remain poor. Reported series indicate that less than 50% of patients will achieve a favourable response to salvage chemotherapy, with 20-25% remaining disease free (Loehrer et al, 1998). While the prognosis of these patients remains unsatisfactory, however, our impression is that improvements are being made. Our aim has been to report our most recent experience, and identify prognostic factors for outcome of relapse.

PATIENTS AND METHODS
This retrospective study comprised all patients treated for NSGCT between 1991 and 1996 at the Beatson Oncology Centre, Glasgow. Patient identification was based on West of Scotland cancer registration. Data were abstracted directly from the patient records. All patients had received platinum-based primary combination chemotherapy. The majority of patients had received BEP (bleomycin, etoposide and cisplatin), or for poorer prognosis patients BOP-VIP (Kaye et al, 1998) or BOP-BEP (Anthoney et al, 1999) in clinical trials. Eighteen patients had received carboplatin-based chemotherapy within a Medical Research Council (MRC) randomized trial for good prognosis patients (Horwich et al, 1997). Patients with residual post-chemotherapy masses had undergone surgery to remove them by a specific specialist surgical team with an interest in this tumour type. If residual malignant disease was identified in the post-chemotherapy resection specimen additional chemotherapy was usually given in the form of either further cisplatin-based combination chemotherapy or oral etoposide.
Patients were defined as having achieved CR if, after chemotherapy, tumour markers and CT scan were normal. Patients who were marker negative after chemotherapy, and who subsequently underwent complete surgical resection of carcinoma or teratoma were classified as no evidence of disease (NED) carcinoma or NED teratoma respectively. Patients with incompletely resected residual masses were described as having residual carcinoma or teratoma.
Recurrence was defined on the basis of rising tumour marker (HCG and/or αFP) and/or the development of new or enlarging metastases on plain radiographs or computerised tomography (CT). The salvage treatment utilised depended on the presentation at relapse and previous treatment. Patients with radiological evidence of a new lesion but normal markers were managed by initial surgical resection and did not receive chemotherapy unless malignant teratoma was seen in the resected specimens. Patients with rising tumour markers were treated with chemotherapy initially. The commonest chemotherapy regimen used was VIP: etoposide, ifosfamide and cisplatin. A number of other regimens were used including VelP; vinblastine, ifosfamide and cisplatin: weekly cisplatin and etoposide; and POMB-ACE: cisplatin, vincristine, methotrexate, bleomycin, actinomycin D, cyclophosphamide and etoposide (Bower et al, 1997). The majority of these patients received between 3 and 6 cycles of chemotherapy (median 4 cycles). Salvage surgery to resect any residual disease was performed, if necessary, after the completion of chemotherapy. Radiotherapy was utilized primarily following resection of cerebral metastases, although it was also successfully used in one patient who relapsed with extensive penile, scrotal and inguinal disease. A further 3 patients with incomplete resection of paraaortic carcinomatous disease were treated with radiotherapy to the tumour bed. This was not standard practice and was only utilized in the absence of other treatment options.
Patients relapsing with rising markers, with evidence of further response to conventional dose regimes, were considered for entry into a non-randomised trial of high dose chemotherapy with peripheral blood stem cell transplantation. High dose chemotherapy comprised carboplatin (total dose was AUC 20 given over 4 days: d1-4), cyclophosphamide (total dose: 120 mg/ kg divided over d1 and d3), and etoposide (total dose 1.2 g/m 2 given over 4 days: d1-4).
After salvage treatment patients were followed up at 2 to 3 monthly intervals in the first year, then 4 to 6 monthly in subsequent years, with clinical, radiological and biochemical examination (αFP, HCG, LDH). Survival times were measured from the date of diagnosis of progression to the date of death or date last seen.
Survival times were calculated from the date of diagnosis of relapse to date of death or date last seen. The Mantel-Haenszel logrank test was used to examine the prognostic value of the variables in Table 5. Multivariate analyses were not performed because of the small number of events involved. Kaplan-Meier estimates were used for calculating the percent alive at 2 years in the same table and for drawing the survival curves.

RESULTS
Forty-one patients were treated for relapsed teratoma between 1991 and 1996. Twenty-eight of 41 patients had raised markers at relapse. The features of these 28 patients at primary diagnosis and at relapse are presented in Tables 1 and 2. Their median age at relapse was 29 years, with a median time to relapse of 13 months (range: 7 to 121 months). Twelve of these 28 patients have died following treatment for subsequent relapse (Figure 1), 14 (50%) remain alive with NED. For the 16 living patients the median follow-up is 72 months (range 26 to 121 months). Table 3 describes the treatment of relapsing disease: all 28 patients were treated with chemotherapy, with 17 patients having postchemotherapy resection of operable residual disease (Table 3). Of these 17 patients treated with chemotherapy and surgery, 10 remain alive with NED (59%), 2 remain alive with residual disease, and 5 have died. The other 11 patients were treated with chemotherapy alone: only 4 (36%) of these patients remain alive in CR, the rest having died of progressing disease.
An additional 13 patients developed recurrent disease with normal markers. They were primarily treated surgically (Table 4). These 13 patients all remain alive with NED, and are not included in the prognostic factor univariate analyses.

Prognostic factors associated with relapse
A recent study by Fossa et al (1999) identified 3 main factors predicting a poor prognosis at relapse. These included progression-free interval (<2 years vs >2 years), response to induction therapy (CR vs not in CR) and level of serum HCG and αFP at relapse (<100 or >100). We performed a univariate analysis of these 3 potential prognostic factors, the results of which are shown in Table 5. This study confirms that an incomplete response to primary treatment, resulting in residual unresected disease, is the most important prognostic parameter. Progression free interval and level of tumour markers at relapse were not found to be independent prognostic factors in our study. When these three prognostic factors are used together, however, patients with 0-1 of these risk factors were found to have a significantly better outcome than those with 2-3 risk factors (P = 0.0027).

DISCUSSION
A widely accepted prognostic factor based staging system to guide the treatment of NSGCT patients progressing after induction cisplatin-based chemotherapy has not yet been defined. A recent study analysing 164 relapsing patients described an overall survival rate of 30% . This study identified a number of poor prognostic factors and indicated that treatment in a large specialist centre could improve outcome. Indeed our findings do suggest that the outcome of relapsed NSGCT is improving, and confirm the prognostic significance of the factors identified by Fossa et al (1999). There are a number of potential reasons for this improvement in outcome. Firstly, 57% (94 of 164 patients) of the relapse patients described by Fossa et al were treated for relapse prior to 1986. Secondly, management of NSGCT has been shown to have a better outcome in a larger specialist treatment centre (Harding et al, 1993;Collette et al, 1999). Thirdly, since surgery is a key component of salvage therapy, access to a specialist surgical team with an interest in NSGCT is crucial. These results therefore have implications for future decisions regarding patterns of health care. We propose that the management of relapsed NSGCT should be carried out with curative intent, in a specialist unit, with access to a dedicated specialist surgical team. Two patients had cerebral metastasis and received post-resection radiotherapy; one had a small focus of malignant disease, in largely mature teratoma, which was completely resected: he received post-resection oral VP-16 for 12 weeks.