Interim report of phase II study of new nitrosourea S 10036 in disseminated malignant melanoma.

Thirty-three patients with disseminated malignant melanoma were entered in a phase II study of the new nitrosourea S 10036 using a 100-mg/m2 weekly induction schedule for 3-4 consecutive weeks. Patients who responded to this treatment were followed with a maintenance therapy every 3 weeks. Toxic effects were mainly hematological and consisted of delayed thrombocytopenia and leukopenia. Among 30 patients who could be evaluated, eight partial responses were observed (response rate, 26.67%); among seven patients with cerebral metastasis, two partial responses were observed. This multicentric study is currently being continued to confirm this interim report.

Cellular specificity of NB70K, a putative human ovarian tumor antigen.

We have studied the cellular expression of NB/70K, a glycoprotein proposed to be human ovarian tumor associated described previously. Analysis of cells bearing this antigen in malignant ascites shows expression in tumor cells of gynecological origin (ovarian, endometrial) and to a limited degree in breast cancer cells. Within such tumors, there is a weak inverse correlation between labeling index and antigen expression. Furthermore, evidence is presented to show that the cells bearing this glycoprotein are physically separable from those bearing carcinoembryonic antigen.

Preclinical antitumor activity of a new Vinca alkaloid derivative, S 12363.

S 12363 is a new Vinca alkaloid derivative obtained by appending an optically active alpha-aminophosphonate at the C23 position of O4-deacetyl vinblastine. S 12363 was evaluated for cytotoxic and antitumor activity against a spectrum of murine and human tumors. This compound was, respectively, on average, 72- and 36-fold more cytotoxic than were vincristine and vinblastine, when tested on a panel of 2 murine and 37 human tumor cell lines using the microculture tetrazolium assay. S 12363 exhibited significant antitumor activity against murine transplantable tumors (i.p. and s.c. P388 leukemia, i.p. L1210 leukemia, i.p. and i.v. B16 melanoma, i.p. M5076 sarcoma, and s.c. colon adenocarcinoma 38), while no activity was observed on s.c. Lewis lung carcinoma. S 12363, when administered i.p., showed moderate activity on human NCI-H460 lung and PANC-1 pancreas tumor xenografts in nude mice. However, when it was administered i.v., it exerted a significant activity against human HT-29 colon, NCI-H460 lung, NCI-H125 lung, PANC-1 pancreas, and A-431 vulvar tumor xenografts. S 12363 was also active in vivo against a P388 leukemia subline resistant to vincristine. On the in vivo panel of tumors used in this study, S 12363 was at least as active as reference compounds, while its optimal dosage was 10- to 40-fold lower than that of vinblastine, depending on the models studied. The effects of schedule and route of administration on the antitumor activity of S 12363 were studied in both i.p. inoculated P388 leukemia and B16 melanoma, in which the activity was improved by single and intermittent treatment (Days 1, 8, and 15) and i.p. route. S 12363, which differs only by the configuration of the asymmetric carbon atom of the side chain, was 300-fold less cytotoxic and 1000-fold less potent in vivo than was S 12363. These results suggest that S 12363 could present a therapeutic advantage over its congeners and deserves further pharmacological evaluations.

Phase I clinical study of the new amino acid-linked nitrosourea, S 10036, administered on a weekly schedule.

Diethyl-1-[3-(2-chloroethyl)-3-nitrosoureido]ethylphosphonate (S 10036) is a new nitrosourea that has been evaluated in a clinical trial because of its activity in the National Cancer Institute panel screen and its rational chemical approach. A Phase I study was conducted in 22 evaluable patients with advanced cancers. The drug was given as a slow i.v. infusion over a period of 60 min on days 1, 8, 15, and 22 followed by a 4-week rest period. The dose levels ranged from 25 to 200 mg/m2/week for 4 consecutive weeks using a modified Fibonacci scheme. Thrombocytopenia was the only acute dose-limiting toxicity and started at a dose of 100 mg/m2/week and above. Hematological toxicity was delayed, cumulative, and dose related. Nausea and vomiting were moderate to severe and dose related. Three responses (one complete and two partials) have been noted. Phase II studies of S 10036 are planned at a dose of 100 mg/m2/week for 4 consecutive weeks ("induction therapy") for patients without prior therapy and 100 mg/m2/week for 3 consecutive weeks for those with prior chemotherapy or radiotherapy. Because of the cumulative toxicity, the recommended dose for the second cycle of S 10036 chemotherapy ("maintenance therapy") is 100 mg/m2/week every 3 weeks.

Tamoxifen enhances the cytotoxic effects of the nitrosourea fotemustine. Results on human melanoma cell lines.

Fotemustine (Fote) is a new amino acid-linked chloroethyl nitrosourea which has been shown to be useful in disseminated malignant melanoma. The aim of the present study was to analyse the cytotoxic effects resulting from the combination of anti-oestrogens and Fote on human melanoma cell lines. The anti-oestrogens tested were tamoxifen (TMX, 5 x 10(-7) mol/l and 5 x 10(-6) mol/l) and 4OH TMX (5 x 10(-8) mol/l and 5 x 10(-7) mol/l). As a preliminary step, a series of nine human melanoma cell lines was screened in order to identify and quantify the presence of oestradiol receptors (ER) in these cell lines. This led to the selection of an ER-positive (+) cell line. The drugs alone or in combination were then tested against CAL 1 ER (+) and CAL 7 ER (-) melanoma cell lines. Different sequences of drug combinations were tested using clinically compatible drug concentrations. For CAL 1 cells, there was a growth inhibitory effect induced by the anti-oestrogens given alone. Overall, the presence of the anti-oestrogens resulted in higher cytotoxic effects than when cells were exposed to Fote alone. The lowest IC50 Fote values as compared to Fote alone were generated by the sequences in which the anti-oestrogens were administered before Fote. Significantly, these associations with anti-oestrogens enabled the IC50 values of Fote to be reduced by up to 80%. Globally, TMX and 4OH TMX had similar synergistic effects. TMX and 4OH TMX had a modest influence on Fote cytotoxic effects against CAL 7 ER-negative cells. These data may be useful for optimal planning of future clinical trials for malignant melanoma using anti-oestrogens and nitrosoureas.

Evaluation of the time-schedule dependency for the cytotoxic activity of the new vinca alkaloid derivative, S 12363 (vinfosiltine).

S 12363 is a new vinca alkaloid derivative obtained by appending an optically active alpha-aminophosphonate at the C23 position of 04-deacetyl vinblastine. The present study concerns four different human tumour cell lines, which represent the spectrum of vinca alkaloid clinical activity. The influence of time exposure on S 12363 growth inhibition was studied in vitro. Cells were exposed to the drug during the following exposure times: 5, 15, 30 min and 1, 3, 6, 12, 24, 48, 72, 144 h. The concentrations of S 12363 applied were between 1 x 10(-2) and 1 x 10(3) nmol/l. The cytotoxic effects were assessed by using the methyltetrazolium (MTT) semi-automated test. Considering the IC50 values in terms of concentration (C) x time (T), I (C x T)50, it was shown that for an equal growth inhibitory effect (50% of cell death) the increased exposure times required higher cumulative drug exposures. More precisely, only very long exposure (greater than 24 h) resulted in very high I (C x T)50. The drug exposure ratios which correspond to I (C x T)50 values for 144 h divided by the I (C x T)50 values for 0.25 h ranged between 2.8 and 18.3. If T and C had symmetrical effects on the final growth inhibition, the I (C x T)50 ratios should have been equal to one. For all cell lines investigated there were similar dose-response curves following two types of S12363 exposure: a single day exposure or three successive daily exposures, the total C x T values being the same in both experimental situations. The basic pharmacological information provided by the present study may encourage further clinical trials of this potentially interesting new vinca alkaloid.

Nitrosoureas lomustine, carmustine and fotemustine induced hepatotoxic perturbations in rats: biochemical, morphological and flow cytometry studies.

Chloroethylnitrosoureas are reactive compounds that are highly effective against malignant neoplasms in humans and animals. The most widely used nitrosoureas, lomustine and carmustine, are known to be hepatotoxic and to induce pericholangitis and intrahepatic cholestasis, which in the long term lead to cholangiolysis and biliary cirrhosis. However, the nitrosourea fotemustine has proved to be non-hepatotoxic at 20 mg/kg and 50 mg/kg. We have studied the effect of these three nitrosoureas on the cytotoxicity and cellular kinetics of rat liver cells. Lomustine and carmustine modify the proliferation index of liver cells in vivo: flow cytofluorometry showed that DNA cell distribution is quite similar for lomustine and carmustine, with subsequent accumulation of cells in G2 + M phase. 3 months later regressive morphological and cell cycle perturbations are noted for the lower dose of lomustine and carmustine. The most severe lesions are noted with lomustine (50 mg/kg). Fotemustine is not hepatotoxic and preferentially induces S phase perturbations. The more toxic nitrosoureas, lomustine and carmustine, induce comparable hepatocyte cell cycle alterations which differ from those induced by the less hepatotoxic nitrosourea fotemustine.

Activity and unexpected lung toxicity of the sequential administration of two alkylating agents--dacarbazine and fotemustine--in patients with melanoma.

We report the results and discuss the toxicity of clinical trials based on a single concept: the decrease in O6alkyl DNA alkyltransferase (O6AT) resistance mechanism when a chloroethylating agent is used sequentially after a methylating agent. This decrease in O6AT being dose dependent, several increasing doses of dacarbazine (DTIC) have been tested (400 mg/m2 to 1000 mg/m2 every 4 weeks, 3-4 h before fotemustine (100 mg/m2 intravenously every 4 weeks). These results (mean overall response rate 27%) compared with reference regimes, demonstrate that DTIC is able to increase the alkylating power of fotemustine: same range of response rate with only half of the two drug doses compared to an alternated combination, high activity rate especially in lung metastases (10/42 complete responses + 13/42 partial responses), different pattern for haematotoxicity, and occurrence of a new side-effect: acute lung toxicity as adult respiratory distress syndrome (ARDS). This lung toxicity was totally unexpected since several hundreds of patients had been so far treated with fotemustine as single agent or in other combinations with DTIC without any case of acute or delayed lung toxicity. Prophylactic administration of corticoids was not effective and monitoring of the respiratory function was of no predictive value. Due to the additional depleting effects of DTIC on at least two main defence mechanisms--the O6AT system and cytosolic and/or nuclear glutathione--we suppose that the sequence is able to increase the alkylating power of fotemustine to an excessive extent and/or that the detoxication capacity of the cell against DTIC and/or fotemustine metabolites is overwhelmed. Other depletors of the O6AT activity which do not generate metabolites that compete for the same detoxication pathway as the chloroethylnitrosourea (CENU) metabolites should be tested.

Overview of docetaxel (Taxotere)/cisplatin combination in non-small cell lung cancer.

Cisplatin-based chemotherapy is effective in non-small cell lung cancer (NSCLC), although it prolongs survival only modestly. Single-agent docetaxel (Taxotere; Rhône-Poulenc Rorer, Antony, France) is highly active against NSCLC. The activity and tolerability of two docetaxel/ cisplatin regimens were therefore investigated in two multicenter phase II studies, one in Australia and one in France. Chemotherapy-naive patients with inoperable NSCLC received either docetaxel 75 mg/m2 on day 1 plus cisplatin 75 mg/m2 3 weekly (n = 47; Australian study) or docetaxel 75 mg/m2 on day 1 plus cisplatin 100 mg/m2 every 3 weeks for three cycles then every 6 weeks (n = 51; French study). The majority of the population (74%) had metastatic disease. Seventy-eight patients were evaluable for efficacy. Overall response rates were 36% (95% confidence interval, 25 to 47) in all evaluable patients and 34% in patients with metastases. Median duration of response was 6 months, with a 4-month median time to progression. Median survival time was 9 months, with a 1-year survival rate of 34%. A median of four (range, one to nine) treatment cycles were administered. Febrile neutropenia occurred in 14% of patients. Severe infection, which occurred in less than 7% of patients, led to two toxic deaths. Other severe toxicities were rare, with severe stomatitis and severe neurosensory side effects reported in 2% and 1%, respectively, of treated patients. No severe fluid retention occurred. Docetaxel/cisplatin, administered as two different schedules, is well tolerated and exhibits efficacy in the range of the most established combinations in the treatment of advanced NSCLC.

Effects of a new triazinoaminopiperidine derivative on adriamycin accumulation and retention in cells displaying P-glycoprotein-mediated multidrug resistance.

A new triazinoaminopiperidine derivative, Servier 9788 (S9788), was investigated for its ability to increase Adriamycin (ADR) accumulation and retention in two rodent (P388/ADR and DC-3F/AD) and three human (KB-A1, K562/R and COLO 320DM) cell lines displaying the P-glycoprotein (P-gp)-mediated multidrug resistance (MDR) phenotype. Depending on the cell line S9788 was shown to be two to five times more active and five to 15 times more potent than Verapamil (VRP) in increasing ADR accumulation in resistant cells. ADR retention in KB-A1 cells maintained in a concentration of 10 microM S9788 was twice that in VRP-treated cells, and similar to that measured in the untreated sensitive KB-3-1 cells. Although 5 microM S9788 and 50 microM VRP gave the same values of ADR uptake in KB-A1 cells, S9788 was shown to induce a greater ADR retention following cell wash and post-incubation in resistance modifier- and ADR-free medium. Taking into account that S9788 had no effects on ADR accumulation and retention in sensitive KB-3-1 cells, it can be suggested that S9788 inhibits specifically the P-gp dependent ADR efflux, and in a manner less reversible than that observed with VRP. Moreover, [3H]azidopine photolabeling of P-gp, in P388/ADR plasma membranes, was completely inhibited by 100 microM S9788. Although S9788, as VRP, had no effect on the cell cycle of P388 cells, 5 microM S9788 increased 700-fold the efficacy of ADR to block P388/ADR cells in the G2+M phase of the cell cycle. Together, these results show that the sensitization, by S9788, of cell lines resistant to ADR is mainly due to an increase in ADR accumulation and retention, leading to an increase in the number of resistant cells blocked in the G2+M phase.

Sequence-dependent cytotoxic effects of the combination of a new nitrosourea, fotemustine, with 5-fluorouracil plus folinic acid.

The present study was designed to analyse the cytotoxic effects of the combination of fotemustine with 5-fluorouracil (5-FU) plus folinic acid (FA). Two human tumor cell lines were used; one line was derived from colon cancer (WIDR) and the other, from a non-small-cell lung cancer (CAL 12). Cytotoxic effects were assessed using the MTT (tetrazolium bromide) semi-automated test in 96-well incubation plates. The effects of various drug combinations were evaluated by the isobologram method. The drug combinations tested included fotemustine concentrations of 20, 30, 40, 50 and 70 micrograms/ml, 5-FU concentrations of 5, 15 and 30 micrograms/ml, and a constant FA concentration of 10(-5) M. A total of 180 different experimental conditions were tested. When cells were exposed to fotemustine prior to treatment with 5-FU, the final cytotoxic effects on both cell lines were additive or synergistic in the majority of cases (P less than 0.001). The 5-FU concentration was a determinant factor that modified the effects of the drug combination from antagonism (at low 5-FU concentrations) to synergism (high 5-FU concentrations; P less than 0.001). The addition of FA (10(-5) M) resulted in a significant shift towards synergistic associations in both cell lines. Administration of 5-FU prior to treatment with fotemustine caused marked antagonism, which 10(-5) M FA could not significantly shift towards simple additivity.

In vitro activity of S 9788 on a multidrug-resistant leukemic cell line and on normal hematopoietic cells-reversal of multidrug resistance by sera from phase I-treated patients.

The triazinoaminopiperidine derivative S 9788 is a new multidrug-resistance modulator that is currently being evaluated in phase I clinical trials. In this study, the reversal effect of S 9788 in comparison with verapamil was shown in vitro in human T-leukemic CCRF-CEM/VLB cells expressing the multidrug-resistance (MDR) phenotype. S 9788 increased in a dose-dependent manner the cytotoxic activity of doxorubicin or vinblastine, with complete reversal of resistance occurring at 2 microM for a concomitant continuous exposure (96 h) to the cytotoxic drugs. At respective concentrations equivalent to the IC10 value (the concentration inhibiting 10% of cell growth), S 9788 was 44 times more potent than verapamil in CCRF-CEM/VLB cells. S 9788 at 2 microM did not enhance the in vitro toxicity of doxorubicin or vinblastine in the human normal bone-marrow erythroid (BFU-E) and myeloid (CFU-GM) progenitors. The effect of exposure duration and concentrations on the synergistic action of modulator and cytotoxic agent closely depended on the cytotoxic agent studied. Post-incubations with S 9788 alone after a 1-h coadministration with vinblastine and S 9788 dramatically increased the reversal effect (4-41 times) in proportion to both the duration of postincubation and the concentration of S 9788. In contrast, for doxorubicin resistance, post-incubation with S 9788 alone induced a maximal 2-fold increase in the reversal effect that was not proportional to the post-incubation duration. In patients treated with S 9788 as a 30-min intravenous infusion during phase I trials, a good correlation was found between the serum levels of S 9788 and the ability to reverse MDR in CCRF-CEM/VLB cells. The reversal effect was dose-dependent and was effective beginning at a plasma concentration of 0.25 microM. These data form a basis for the design of phase II trials using a combination of a loading dose of S 9788 given before vinblastine or doxorubicin administration followed by a maintenance infusion of S 9788 alone for a period of 2-24 h.

Phase II study of fotemustine as second-line treatment after failure of immunotherapy in metastatic renal cell carcinoma.

Sixteen patients who after prior systemic immunotherapy had progressing disease received fotemustine (100 mg/m2) i.v. on days 1, 8, and 15 followed by a 5-week rest period. In responding or stabilized patients, maintenance therapy consisted of 100 mg/m2 fotemustine given once every 3 weeks until progression on toxicity occurred. No objective response was observed. Four patients showed stable disease (median duration: 4 months; range: 3-19). The main toxicities were neutropenia (WHO grade 3 and 4: 27%) and thrombocytopenia (WHO grade 3 and 4: 27%). Fotemustine was administered on an outpatient basis and was generally well tolerated, but in our series of patients it had no antitumour activity in metastatic renal cell carcinoma after failure of immunotherapy.

Differential in vitro action of S-12363, a new vinblastine derivative, and of its epimer on microtubule proteins.

The action of two epimers of a new vinblastine derivative that differ in their in vivo antitumor activity and their cytotoxicity was studied in vitro in brain microtubule proteins. These two compounds, called S-12363 and S-12362, could not be distinguished from one another or from other active vinca alkaloids by their ability to prevent microtubule assembly. However, they differed strongly both from one another and from vincristine and vinblastine in their ability to induce the formation of tubulin paracrystals and in the stability of the paracrystals following temperature shifts from 0 degree to 37 degrees C and vice versa. The most potent drug, S-12363, induced considerable tubulin aggregation, which was even more pronounced than that observed in the presence of vincristine. Previous results have shown that S-12363, in contrast to vincristine, induces no neurotoxic effects. This observation is in disagreement with a direct relationship between tubulin aggregation and neurotoxicity.

In vitro chemosensitivity testing of Fotemustine (S 10036), a new antitumor nitrosourea.

Fotemustine (S 10036) is a new anti-tumor nitrosourea characterized by a phosphonoalanine carrier group coupled to the nitrosourea moiety, which potentially increases the cellular penetration of the drug. Using human tumor cell lines, the activity of S 10036 was compared with that of the more established nitrosoureas BCNU and CCNU. Growth-inhibiting effects were evaluated by the [3H]-thymidine incorporation test. In a panel of 12 human cancer cell lines [melanoma (4), ovary (2), head and neck (3), lung (1), bladder (1), breast (1)], the dose-response curves of S 10036 (0-100 microM) were similar to those obtained with equimolar concentrations of BCNU and CCNU; they indicated a moderately more marked effect for two and an equal effect for six melanoma cell lines with S 10036 as compared with BCNU. Moderate but significant synergistic combinations were obtained when S 10036 (0-80 microM) and CDDP (0-100 microM) or DTIC (250-6,500 microM) were combined in melanoma cell lines. In conclusion, the new nitrosourea S 10036 shows promising activity, particularly against human melanoma cell lines.

Renal effects of S10036 in man.

The renal hemodynamic and tubular effects of S10036 (fotemustine) were evaluated in seven patients with advanced malignancy. Initial evaluation carried out prior to treatment and repeated 1 day after the first fotemustine infusion and 7 days after the second included clinical, haematological parameters, liver-function tests, and determination of the glomerular filtration rate, renal blood flow and enzymuria. The glomerular filtration rate was 108 +/- 3.7 ml/min before treatment and remained stable after the first (117 +/- 5 ml/min) and second (124 +/- 6 ml/min) fotemustine infusions. Renal blood flow and urinary beta 2-microglobulin and N'-acetylglucosaminidase excretion were also not modified by fotemustine administration. We conclude that fotemustine does not acutely alter renal haemodynamics, nor does it have direct tubular toxicity.

In vivo evidence of complete circumvention of vincristine resistance by a new triazinoaminopiperidine derivative S 9788 in P388/VCR leukemia model.

S 9788, a new triazinoaminopiperidine derivative, was found to be a potent reversant of vincristine resistance in the in vivo murine leukemic P388/VCR model. In two treatment regimens (Q4D days 1, 5 and 9 and QD days 1-9), S 9788 enhanced the antitumor activity of vincristine in a dose-dependent manner, resulting in a complete circumvention of drug resistance for well-tolerated doses of S 9788. S 9788 was also effective in enhancing therapeutic effects of vincristine in the treatment of sensitive P388-bearing mice. These results strongly suggest that S 9788 may be a potential candidate for circumvention of multidrug resistance (MDR) in clinical practice.

Fotemustine: an overview of its clinical activity in disseminated malignant melanoma.

Fotemustine is a new chloronitrosourea which is active against disseminated malignant melanoma (DMM), and especially against cerebral metastases (CM). This efficacy has been widely demonstrated through many phase II studies. A multicentre trial of monotherapy was undertaken in 153 evaluable French patients. A response rate (RR) of 24.2% (25.0% RR in CM; 31.8% in non-visceral metastases (NVM) was obtained. Three other phase II studies confirmed these results with respective objective RR of 16.7%, 20.0% and 47.0% and RR in CM of 8.3%, 14.3% and 60.0%. Fotemustine has also been used in combination with dacarbazine (DTIC), in patients with DMM. The RR among 103 patients was 27.2% (26.3% in CM, 37.5% in NVM), confirming the activity of fotemustine. The two drugs have also been administered sequentially, in order to exploit their synergism in interfering with the O6 alkyltransferase. Impressive RR have been achieved, especially in patients with visceral metastases (VM) but at the expense of a pulmonary toxicity that does not arise with other treatment schedules and which precludes its use outside of strictly conducted clinical trials.

Fotemustine, dacarbazine, vindesine combination chemotherapy in advanced malignant melanoma: a phase II study of 43 patients.

Fotemustine and dacarbazine constitute the most active single chemotherapeutic agents in the treatment of melanoma. In this phase II study we evaluated the activity and toxicity of a combination of fotemustine, dacarbazine and vindesine as a means of increasing response rate and survival time. Between September 1989 and November 1993, 43 patients with advanced melanoma were treated with a combination of 100 mg/m2 fotemustine on days 1 and 8, 250 mg/m2 dacarbazine on days 15 and 16 and 2 mg/m2 vindesine on days 15 and 16 as induction treatment. After a 5-week rest period, the patients exhibiting a response or stable disease received the same drugs administered once every 28 days as maintenance therapy until either progression or toxicity was observed. Among 41 evaluable patients, there were six complete responses and eight partial responses. The overall response rate was 32% (95% confidence interval: 18-46%), with 8 months median duration of response. Median survival time was 10 months. This regimen was well tolerated. From this large phase II study, we conclude that such a combination is active against advanced malignant melanoma and seems to be more effective than fotemustine or dacarbazine used alone, especially on visceral metastatic sites.

Combined treatment with dacarbazine, cisplatin, fotemustine and tamoxifen in metastatic malignant melanoma.

The combination of dacarbazine (DTIC), cisplatin (DDP), carmustine and tamoxifen (TAM) has been reported to yield a high rate of response in patients with metastatic melanoma, but responders often experience intracranial recurrences. As fotemustine (FOT) has demonstrated activity on cerebral metastases, the rationale of this study was to replace carmustine by FOT in this four-drug regimen. Twenty patients with metastatic melanoma received FOT (100 mg/m2) on days 1 and 8, DTIC (220 mg/m2 per day) and DDP (25 mg/m2 per day) from day 1 to day 3 and from day 28 to day 30, and continuous daily treatment with TAM (20 mg/day). If stabilization or response was observed at the end of the 8th week, patients received maintenance courses of FOT on day 1, and DTIC (220 mg/m2 per day) and DDP (25 mg/m2 per day) on days 1 to 3. Nineteen patients were evaluable. Of these, six had brain metastases. The overall response rate was 10.5% (two out of 19); both of the responders had only partial responses. The best responding site was lung. No response was obtained in the four patients with evaluable brain metastases, but no patient had therapy failure due to new brain metastases. The median overall survival was 5 months (range 1-45 months). Toxicity was mainly haematological. The use of this combination is not recommended.