A phase I clinical and pharmacokinetic study of the new topoisomerase inhibitor GI147211 given as a 72-h continuous infusion.

GI147211 is a novel, totally synthetic camptothecin with promising preclinical and early clinical activity. This study was designed to determine the maximum tolerated dose of Gl147211 as a 72-h infusion and to describe its pharmacokinetics and pharmacodynamics on this schedule. In a single-arm, rising-dose study in patients with advanced cancer, eight cohorts of three or more patients received 72-h infusions of Gl147211 at doses ranging from 0.25 to 2.5 mg m(-2) day(-1). Forty-four patients received a total of 124 cycles. All patients had refractory tumours and 40 had received prior chemotherapy and/or radiotherapy. Whole-blood Gl147211 lactone, total blood and total concentrations were measured during and over the 12 h following the infusion. Myelosuppression was observed at all dose levels. Neutropenia was dose limiting at 2.0 mg m(-2) day(-1) in minimally pretreated patients, while both neutropenia and thrombocytopenia were limiting at 1.5 mg m(-2) day(-1) in those more heavily pretreated. Phlebitis occurred with infusions through peripheral veins early in this study, necessitating the use of central venous access. Other toxicities included mild nausea and vomiting, fatigue, headache, central venous catheter infections and alopecia. Three partial and two minor responses lasting 8-34+ weeks were noted in patients with ovarian, colon and breast carcinomas and hepatoma. Mean steady-state concentrations of Gl147211 increased with dose over a range of 0.25-1.24 ng ml(-1). The mean terminal elimination half-life was 7.5 h, and the clearance averaged 1074 ml min(-1) m(-2) over the doses studied. The mean fractional excretion of unchanged drug in urine was 0.114. Gl147211 lactone exposure correlated with haematological toxicity. The recommended phase II doses for this regimen are 1.75 mg m(-2) day(-1) and 1.2 mg m(-2) day(-1) for minimally pretreated and heavily pretreated patients respectively. At these doses, steady-state Gl147211 concentrations within the range of those effective in vitro were achieved. Extensive phase II evaluation of this compound and further phase I trials evaluating more prolonged infusions are ongoing.

Keywords: G1147211: GG211: camptothecin analogues: topoisomerase inhibitor; phase trial; pharmacokinetics: pharmacodynamics: continuous infusion GI 147211 (7-4 methyl piperazinomethilene)-10.1 1 -ethylenedioxy-20-(S camptothecin: Figure 1) is a totally synthetic analogue of camptothecin. a natural product isolated from Camptotheca acuminata (WAall et al. 1966: Emerson et al. 1995. Despite the sicrnificant anti-tumour actixvity of the parent compound camptothecin in preclinical models and in early clinical trials. its further development >-as compromised by severe and unpredictable toxicitv involxino the bone marrow. gastrointestinal tract and urothelium (Gottlieb et al. 1970: Mluggia et al. 1972). In part. the insolubility of camptothecin x-as central to this undesirable adx erse picture. The nuclear enzyvme DNA topoisomerase I. x hich relaxes DNA supercoils arising during, replication and gene transcription. has been identified as the specific targret of camptothecin (Hsianer andLiu. 1988: Wall andWani. 1995). The mechanism of cytotoxicity of camptothecin (and of its analogues) involves the formation of a coxalent complex Awith topoisomerase I and DNA: this cleavable complex stabilizes DNA single strand break-s. xwhich mav be converted to double strand break-s upon encountering a replication fork (Zhang et al. 1990: Slichenmver et al. 1993. Topoisomerase I enzN-me lexels in cancer cell lines and in surmical specimens of a range of human tumours are higher than those of normal tissues (Giovanella et al. 1989: Husain et al. 1994. These findings stimulated the clinical evaluation of less toxic. watersoluble analogues of camptothecin. The semisy nthetic derivatives topotecan and irinotecan (CPT-l1b has-e rood preclinical antitumour activitx. reproducible toxicitv in earlx clinical trials and encouraging activitv in several human tumours (Slichenmver et al. 1993;Potmesil. 1994: Wall andW'ani. 1995).
GI 14721 1 is a whollv synthetic camptothecin analogue in w-hich molecular modifications have been made to enhance w-ater solubilitv and to increase affinity for topoisomerase I. As w-ith all camptothecin analogues. GI14721 1 exists in a pH-dependent equilibrium betwxeen the lactone (the active forr) and the open-ring carboxvlate. In xitro. G1147211 has demonstrated substantial *Present address: -Sen-icio de Oncologia Medica. Hospital 12 de Octubre'. Carretera de Andalucia KM 5 4. Madrid 28(041. Spair -Present address: Department of Oncoloes. Pols arth Building. U nisersits of Aberdeen. Forsterhill. Aberdeen AB9 2ZD. LUK Presented in part at the American Soc.ietr of Clinical Oncoloeg, Meeting. Ma\ 1995* Los Angeles. Califormna actixvity against a broad range of cell lines. In vixo studies usinr, colon (HT29 and SW480). breast (MX 1). oxarian (SKOV3). prostate (PC3) and lung (H460) carcinoma xenografts in nude mice confirmed the anti-tumour actixitv (Emerson et al. 1986(Emerson et al. . 1993(Emerson et al. . 1994(Emerson et al. . 1995. In preclinical models. topoisomerase I inhibitors demonstrate greater in vitro and in vivo anti-tumour activity when administered by repeat doses or by prolonged infusion. To exploit this apparent schedule dependency. we initiated a phase I trial of GI147211 by 72-h infusion.

PATIENTS AND METHODS Patient selection
This phase I study w as conducted as an international collaboratix e trial betuxeen Fox Chase Cancer Center and the Cancer Research Campaign (CRC) Department of Medical Oncology at the Universitv of Glasgow. Elicible patients had histologically documented solid tumours which were considered to be refractory to consentional therapv. They were over 18 years of age. with an Eastern Cooperative Oncologv Group (ECOG) performance status of 0-2. They had adequate bone marrow (WBC count >4000 l1j. g5ranulocyte >2000 gl-' and platelet count >100 000 l1il ) lixer [bilirubin level <1.5 mg dl-I and aspartate aminotransferase (AST) .4x upper normal value] and kidney (serum creatinine concentration <1.5 ma dl-I or creatinine clearance .60 ml min-' ) function.
Patients were required to have recovered from all toxicities of prior treatment and to have received no cytotoxic chemotherapy w ithin the previous 3 weeks (6 w eeks in the case of nitrosoureas or mitomvcin C). All patients rasve written informed consent. During the course of this study. placement of a central vein catheter A-as made a requirement.
Before therapy a medical history. physical examination. complete blood count. biochemical profile. urinalysis. electrocardioggraphy and chest radiography were performed. Patients were monitored w ith weekly blood counts and biochemical profiles. and clinical examinations were performed on exerv course. Doses were not escalated wxithin patients. Results are reported using the Common Toxicits Criteria (Cancer Therapy Ev-aluation Program. National Cancer Institute. Bethesda. MD. 1988). Patients with measurable disease were evaluated (usually by radiographic scan or X-rav) every other course: those with stable disease or better x ere continued on therapy. (Miller et al. 1981

Drug administration
Initial stability studies rexealed that GI 14721 1 xA as stable either as a concentrate or as a more dilute solution in 5%7 dextrose (axvera-e pH 5). A uniform preparation of 5%7 dextrose USP was proxided as the sole acceptable diluent for this trial. After dilution in 96 ml. the drucg was administered as a continuous infusion using an ambulatory infusion pump. Medication bags and extension tubingr were changed evers day during therapy and x ere protected from light at all times. Aliquots of the administered solution were obtained to confirm the stabilitv of GI 14721 1.
Significant interspecies differences in toxicity prompted selection of a starting dose of G1147211 of 0.25 mc m-day-. which xxas less than 1/30 the munrne LD,I,. GI 14721 1 was administered as a 72-h infusion: courses xxere repeated every 4 wxeeks. provided  Analytical procedure A GLP-v-alidated HPLC assav A-as used for all GI114721 1 measurements (Stafford and St Clairfe. 1995). 61147211 is extracted from cold blood by a mixture of 1:4 (vIv) acetonitrile and butylchloride.
Samples are kept in an ice slurry bath before extraction as the conversion of lactone to carboxyvlate is slowsed down dramatically by the low temperatures of the ice slumn. The orgai phase is collected and evaporated dowsn under a stream of nitrogen. The reidue is reconstituted vN ith a 1:4 (v/v) solution of acetronitrile and sodium phosphate buffer (pH 4). This solution is inj'ected onto an HPLC system equipped wsith a BDS Hyperfil C~column (250 mm x 4.6 mm) and fluorescence detector. The range of the assay is 0.15-100 ng, ml' with sufficient precision and accuracv (coeffecient of variation < 10%7). The mobile phase consists of 25%7 acetonitrile and 10% ammonium acetate buffer pumped at 1.7 ml rmm'n. The internal standard used in this assav is 6.7dimethoxy-4-methylcoumnarin (Stafford and St Claire. 1995).

Pharmacokinetic analysis
Plasma G11472 11 concentration vs time curves wxere evaluated using model-independent methods (Gibaldi. 1984). The averagte steady -state concentration (C A was determined by taking h average concentration after steadyv-state wxas achieved. Syvstermic blood clearance (CL) wxas estimated using, the equation:

CL=K/ C_
A-here K, is the infusion rate. The terrminal rate constant Xz xxas determined by linear regression of log, transformation of the blood concentration vs time curse followsing the end of the infusion. The  ..  Haematological toxicity Eight dose lev-els ranging from 0.25 to 2.5 m m-dayvwere studied. Myvelosuppression w as obser-ed at all dose levels and w-as dose dependent and non-cumulative (Table 2). Only grade I or II haematological toxicity was noted at doses of 1.0 mg m-' day-'. with the exception of a single patient who had previously experienced severe myelotoxicitv followina treatment with carboplatin and taxotere. At doses > 1.5 mg, m-' dav-l. 'grade IV myelosuppression was observed and criteria for dose-limiting toxicitv fulfilled.
The severity of mvelosuppression seemed to be related to the extent of prior therapy: accordingly. a representative sample of both heavilyv and minimally pretreated patients was accrued as stipulated previously.
Patients who were heavily pretreated experienced more myelosuppression at a gixen dose level ( Table 2). Four of ten patients Relationship between dose and lactone clearance. An increase in clearance is shown at doses > 1 mg m-2 day-, neutropenia episodes (median duration 5 days. range 1-14) developed fever and two had microbiologically documented infections.

Non-haematological toxicity
Twenty-six patients experienced grade I or II nausea and 16 had arade I or II vomiting (Table 3). Three patients had grade III vomiting. Fifteen of these patients had nausea and/or vomiting before starting, the infusion of GI 147211. Emesis was not dose dependent. usually started on day 1 or 2 and pursued an intermittent course during the infusion. It was easily controlled with standard antimetics. Mild diarrhoea occurred in seven patients and did not appear to be temporally related to treatment. Half of the patients had mild anorexia and/or fatigue which lasted for approximately 2-3 days after the infusion. Alopecia. observed in 14 patients. occurred most frequently in patients who received three or more courses at doses > 1.5 mg m-' day-l. Twenty per cent of patients complained of headache while receiving the infusion. In one patient. the infusion was discontinued on the third course as she developed sensitivity to the drug (cutaneous rash, respiratory difficulty. eosinophilia). These symptoms first appeared about 10 days after the initial infusion and became progressively more severe with subsequent cycles. The patient s metastatic colorectal cancer was stable through this period. Two patients presented with a new diagnosis of a bleeding duodenal ulcer and oesophageal varices. shortly after entrv into the trial. Five patients treated at doses < 1.0 mg m-' day-1 had the GI147211 infusions administered through peripheral veins. All of these patients developed phlebitis at the site of venepuncture. frequently requiring a change of the i.v. cannula at 24-48 h. Phlebitis recovered in 7-24 days without apparent effect of the applied local treatment. One patient had extravasation of drug in the left forearm: this did not produce necrosis and was successfully managed with local steroids and ice packs. The other 39 patients were treated through central venous catheters. Five developed associated infections requiring iv. antibiotics. The only toxic  Table 4. There were no significant differences in drug elimination between the two sites. Steady-state G1147211 blood concentrations were reached in most patients between 14 and 26 h into the 72-h infusion. Linear regression analysis revealed that the GI147211 lactone Css increased with dose over the range 0.25-1.24 ng ml-' (r = 0.58. P < 0.05) (Figure 2). After the end of infusion. the plasma lactone elimination was monoexponential with a mean half-life of 7.5 h. Mean clearance values tended to increase with dose (r'=0.31: P<0.05) and ranged from 776 to 1554 ml min-1 m-' (average 1074ml min-' m-2) over the dose range studied (Figure 3). Total GI14721 1 blood concentrations (lactone plus carboxylate) were approximately four times greater than lactone concentrations. The variability of total drug was greater than that of the lactone (the active moiety). Tenninal half-life of total GI14721 1 was longer than that of the lactone: the harmonic mean value was 20. subjects. The mean fraction of drug excreted unchanged in the urine (fe) was 0.114 ± 0.041 and the renal clearance averaged 164 ± 61 ml minacross the doses measured.

Pharnacodynamics
The CSS of the GI147211 lactone correlated with the decrease in neutrophil and platelet counts using the sigmoid E model ( Figure 4A and B). While variability was high. a steep concentration-response curve was observed between 0.5 and 1.5 ng ml-l. Based on this model. neutrophils were more sensitive than platelets to G1147211 toxicity. The half-maximal concentration was 0.75 ng ml-1 for the fonner and 1.4 ng ml-1 for the latter.
Anti-tumour activity Three partial responses lasting 8. 30 and 34+ weeks were observed. One patient with breast cancer. who had received two previous chemotherapy regimens (CMF and epirubicin) as well as two prior regimens of hormonal therapy. experienced shrinkage of cutaneous and lymph node disease. A patient with ovarian cancer who had been treated with two platinum-based regimens experienced a substantial decrease in CA125 and had control of ascites.
A patient with colorectal cancer metastatic to the liver, previously treated with 5-fluorouracil and leucovorin, also had a partial response. Two additional patients, one with colorectal cancer and the other with hepatoma, had decreases in hepatic lesions of 44% and 36% respectively. All of these patients were treated at doses > 1.5 mg m-2 day-'. with the exception of the patient with ovarian carcinoma. who received 0.5 mg m-'day-'.

DISCUSSION
Treatment with GI 14721 1 administered as a 72-h continuous infusion in adults with malignant solid tumours was well tolerated on an outpatient basis. As in studies of G1147211 given daily for 5 days, myelosuppression was the dose-limiting toxicity (Eckardt et al. 1995). In the current trial, bone marrow toxicity appeared at all dose levels and involved all haematopoietic lineages in a dosedependent manner. The degree of platelet toxicity was somewhat more pronounced than that observed on a similar schedule with topotecan (Burris et al. 1994). though more protracted infusions of that drug result in comparable effects on both lineages (Hochster et al. 1994). The extent of prior therapy has also been a determinant of bone marrow toxicity in the 5-day study of GI147211 and in phase I trials of other topoisomerase I inhibitors (Grochow et al. 1992: Saltz et al. 1993Haas et al. 1994: Eckardt et al, 1995: Rubin et al. 1995. Both neutropenia and thromobocytopenia were dose limiting in this study and both appeared to be related to the extent of previous treatment. although the patient numbers in each population did not allow for rigorous statistical analysis. In heavily pretreated patients. grade HI or IV toxicity was observed in all patients treated at or above 1.5 mg m-2 day-'. Thus. the recommended dose for phase H trials for previously treated patients is 1.2 mg m-2 day-' Among untreated and minimally pretreated patients. the MTD was 2.0 mg m-day-'. At this dose. three of ten patients developed grade IV neutropenia and four grade IH thrombocytopenia. All of these episodes were uncomplicated and were of short duration (less than 5 days). apart from one case of neutropenic fever. Therefore. an appropriate phase II dose would be 1.75 mg m-day-'. While the ability of the bone marrow to recover from GI147211 is an important factor. it may be observed from Table 2 that prior treatment alone is insufficient to account for the wide variability in toxicity at a particular dose level (see, for example. the 2 mg m-' day-' dose level). Nor do differences in drug exposure explain the variability: as may be observed in Figure 4. a broad range of toxicity outlines the characteristic sigmoidal curve. The steep concentration-response curve is characteristic of this class of drug (Grochow et al. 1992: Haas et al. 1994) and emphasizes the need to understand the pharmacodynamic basis of drug effect. In vitro studies of cell lines with varying topoisomerase I content suggest that the expression of topoisomerase I may be a detenrninant of camptothecin effect: the higher the topoisomerase I expression the more sensitive the cell line to camptothecin-induced cytotoxicity (Pommier et al. 1994). However. Pommier et al (1994) have shown that topoisomerase I activity alone does not explain varying sensitivity. In peripheral mononuclear cells from patients undergoing treatment with the topoisomerase I inhibitor topotecan. the opposite relationship has been found: those expressing lower topoisomerase I levels had more myelotoxicity (Khater et al. 1995). Clearly, a simple explanation for the variable toxicity is not yet evident and additional pharmacodynamic studies are needed.
The modest incidence of non-haematological toxicity makes GI14721 1 an excellent candidate for combination with other drugs or radiation therapy. Mild and easily preventable emesis and shortlasting fatigue were the main complaints. In contrast to irinotecan. G1147211 does not induce diarrhoea (Slichenmyer et al. 1993;Potmesil. 1994: Abigerges et al. 1995. This may relate to the fact that G1147211 is the active compound. while irinotecan is a prodrug which is converted to SN-38. the active metabolite. An important site of this conversion appears to be the intestinal mucosa. in which high levels of SN-38 may be formed locally. with attendant mucosal damage (Gupta et al. 1994). The lack of gastrointestinal toxicity of G1147211 doubtless accounted for the infrequent finding of sepsis. even in the face of profound neutropenia. Also remarkable is the absence of mucositis in the Phase I study of Gl147211 as 72-h infusion 1335 present study. Only one episode of grade IH stomatitis was seen. and this occurred at 2.5 mg m-2 day-'. concomitantly with neutropenic fever. By contrast. mucositis was dose limiting in a 5-day infusion phase I study of topotecan in leukaemic patients (Kantarjian et al. 1993). The absence of urothelial toxicity is similar to topotecan and is probably due to improved water solubility (Emerson et al. 1995). The propensity of GI147211 to induce phlebitis at venepuncture sites with infusion times of less than 24 h makes the use of central vein catheters a prerequisite for protracted administration schedules.
Anti-tumour activity was documented in five patients with colon. ovary and breast cancers and hepatoma. These three partial and two minor responses are particularly noteworthy as they occurred in patients who had been previously treated with conventional chemotherapy. This. coupled with the significant antiproliferative activity of GI 147211 in preclinical models in breast. colon and ovarian tumours (Emerson et al. 1995). and the anti-tumour activity seen with the daily x 5-day regimen. justifies further phase II evaluation GI14721 1. Except for the partial response in the patient with ovarian cancer observed at 0.5 mg m-2 day-'. all other tumour regressions appeared at higher dose levels (1.5-2.0 mg inday-'). suggesting a dose-response relationship. This high rate of response was also identified in another GI14721 1 phase I trial and these data suggest that the drug may have useful activity when used in a population with less advanced disease.
Pharmacokinetic studies showed that Css was related to dose but there was a twoto threefold variation in Css with dose level. Css at doses 2 1.2 mg m-2 day-' were above 1 ng mF' which are potentially cytotoxic (e.g. 50% growth inhibition concentration for the melanoma cell line Lox was 0.592 ng ml-') (Emerson et al. 1995).
The Css of topotecan lactone at MTD of 1.6 mg m-2 day-' administered as a continuous infusion for 3 days every 3 weeks was 5.5 ng ml-l (Burris et al. 1994). These differences in drug levels at doses that produce similar degrees of myelosuppression may be accounted for by topotecan being measured from plasma and G1147211 from whole blood. and the observation that the latter compound is 2-5 times more active in in vitro models (Slichenmyer et al. 1993: Emerson et al. 1995. It remains to be seen if this enhanced potency of GI 14721 1 will translate to a selective advantage in clinical trials. The terminal half-life (7.5 h) in the present study was higher than that observed after 30 min infusions. possibly reflecting the absence of tissue distribution which has already occurred during the 72-h infusion period. Better characterization of the elimination of the carboxylate would be achieved with later sampling points. From these data. the slower clearance of the carboxylate form (mean terminal half-life 20 h) may also provide a source of continuing formation of lactone during the elimination phase. Although greater interpatient variability was observed, the average clearance increased with increasing dose. particularly at doses > 1.0 mg m-' day-'. These data do not support a definitive conclusion. but ratelimited elimination or binding processes may be operating at higher doses. As the major route of metabolism of the lactone is its conversion to the open ring acid form. it is unlikely that the compound induces its own metabolism. In this study. approximately 11% of total GI 147211 was recovered unaltered in urine.
confirming animal data that biliary and/or intestinal excretion are the main routes of elimination for this drug.
In summary. myelosuppression is the DLT of GI147211 as a 72-h infusion every 3 weeks. Haematological toxicity was doserelated. non-cumulative. reversible and dependent on the extent of prior chemotherapy. Central venous catheters are required as prolonged infusions induce phlebitis. At the recommended doses for phase II trials (1.75 mg m-2 day-' and 1.2 mg m-2 day-' for untreated/minimally treated and heavily pretreated patients respectively). the schedule is well tolerated in an outpatient basis with minimal non-haematological toxicity and promising evidence of anti-tumour activity. The anti-tumour activity documented in this trial supports a broad phase II evaluation of this drug. which is now in progress using the five daily dose schedule. In attempting to optimize the therapeutic efficacy of GI1472 11. the duration of tumour exposure to the drug may be increased by prolonging the infusion time or by oral administration. For this reason. phase I trials testing 14-and 21-day infusions every 4-5 weeks and a bioavailability study of an oral formulation of the compound have been started.