Male and female BALB/c and BALB/cnu/nu mice aged 5 to 20 weeks were allowed free access to food and water. All animal experiments were carried out with ethical committee approval. The ethical guidelines that were followed met the standards required by the UKCCCR Guidelines for the Welfare of Animals in Experimental Neoplasia (Workman et al, 1998).

LS174T colon, AGS stomach, MCF-7 breast and OVCAR-3 ovarian human cancer cell lines were obtained from the American Type Culture Collection (Manassas, VA, USA). CL1-5 human lung adenocarcinoma cells (Yang et al, 1992) were kindly provided by Dr PC Yang, Department of Internal Medicine, National Taiwan University Hospital (Taipei, Taiwan). LS174T, CL1-5 and AGS cells were cultured in Dulbecco's modified Eagle's medium (Sigma, St Louis, MO, USA) whereas MCF-7 and OVCAR-3 cells were grown in RPMI-1640 (Sigma, St. Louis, MO, USA) supplemented with 10 μg ml⁻¹ bovine insulin. All media were supplemented with 10% heat-inactivated bovine calf serum, 100 U ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin in a 5% CO₂ humidified atmosphere.

Five thousand (10 000 for MCF-7) cells per well were plated in 96 well microtiter plates overnight. Graded amounts of 9AC, topotecan, 9ACG or 9ACG plus 1 μg E coli-derived βG were added for 24 h in triplicate. The cells were then washed twice with sterile PBS and fresh medium was added for another 24 h. Fresh medium containing 1 μC per well [³H]thymidine was added 16 h before the cells were harvested on glass-fibre filters and the radioactivity was measured on a Topcount scintillation counter. The results are expressed as % inhibition=(c.p.m._S×100)/c.p.m._C where c.p.m._i represents counts per minute of sample (S) or untreated controls (C).

9ACG (10 mg ml⁻¹ in PBS) was adjusted to pH 6.5 with sodium carbonate immediately before administration. More than 95% of 9ACG remained in the lactone form for at least 4 h as determined by HPLC (data not shown). 9AC was prepared as a 1 mg ml⁻¹ suspension in Lipiodol ultra-fluide (Laboratorie Guerbet, Bois Cedex, France). Mice were i.p. injected with 50 mg kg⁻¹ 9ACG, i.v. injected with 25 or 50 mg kg⁻¹ 9ACG or s.c. injected with 2.5 or 5 mg kg⁻¹ 9AC in pilot studies using two mice per group. A maximum transient (<72 h) weight loss of less than 25% was considered acceptable. Six 7–8 week old mice (three males and three females) were also i.v. injected with 50 mg kg⁻¹ 9ACG. Prodrug toxicity was further examined in three groups of six mice (three males and three females) at three different ages (5, 10 and 20 weeks). Mice were injected with 9ACG (50 mg kg⁻¹, i.v.) or 9AC (5 mg kg⁻¹, s.c.) and body weights were followed. Initial body weight was defined as the mean value of three determinations taken 4, 2 and 0 days before drug administration.

Two×10⁶ LS174T cells or 1×10⁷ CL1-5 cells were s.c. injected in the right flank of 16 week old female BALB/cnu/nu mice. Tumour bearing mice were subdivided into groups of five to eight mice. Therapy was initiated 10–11 days after tumour inoculation, when the mean tumour volume was 100–200 mm³. Mice treated with prodrug were i.v. injected with 9ACG at the indicated doses and times. The 9AC group was s.c. injected on the contralateral flank with 3 or 5 mg kg⁻¹ 9AC as a suspension in lipiodol. Irinotecan (10 mg kg⁻¹) and topotecan (1.8 mg kg⁻¹) were i.v. injected on the indicated days. Control mice were i.v. injected with PBS following the corresponding 9ACG schedule. Tumour volumes were calculated as length×width×thickness×0.5 and expressed in mm³. Tumour growth inhibition was calculated as:

[Formula ID: equ1]

Statistical significance of differences between mean values was estimated with Microsoft Excel using the independent t-test for unequal variances.

Table 1 compares the in vitro cytotoxicity of topotecan, 9AC, 9ACG and 9ACG mixed with βG to five human cancer cell lines. Topotecan and 9AC displayed similar potencies with IC₅₀ values ranging from 3 to 30 nM. 9ACG was 25 to 60 times less toxic than 9AC but displayed similar cytotoxicity as 9AC after the glucuronide group was enzymatically cleaved by βG.

A single s.c. injection of 2.5 mg kg⁻¹ 9AC suspended in lipiodol caused minimal toxicity with 10% weight loss whereas a dose of 5 mg kg⁻¹ produced about 20% weight loss (results not shown). Mice that were i.p. injected with 50 mg kg⁻¹ 9ACG experienced progressive weight loss and died within 6 days (results not shown). In contrast, all mice that were i.v. injected with 25 or 50 mg kg⁻¹ 9ACG experienced dose-dependent weight loss until day 6 followed by rapid and complete recovery within a week. A more detailed study of toxicity performed in 7–8 week old male and female mice with 50 mg kg⁻¹ 9ACG (Figure 1) confirmed the preliminary results. Male mice transiently lost about 25% body mass but female mice experienced only about 5% weight loss.

The effect of age and sex on 9ACG toxicity was further examined by injecting 5, 10 and 20-week old male and female mice with 9AC or 9ACG. An i.v. injection of 50 mg kg⁻¹ 9ACG produced about 25% weight loss in 5-week old male mice (Figure 2A), but only 5% weight loss in 20-week old mice (Figure 2C). 9ACG produced minimal (5%) toxicity in 5-week old female mice (Figure 2D) and virtually no weight loss in 10 or 20-week old mice (Figures 2E,F). Regardless of age, 5 mg kg⁻¹ 9AC was more toxic to male mice, producing about 10% weight loss (Figures 2A,B,C) compared with 5% weight loss in female mice (Figures 2D,E,F).

BALB/c nu/nu mice bearing established LS174T tumours were i.v. injected with 50 mg kg⁻¹ 9ACG in PBS or s.c. injected with 3 mg kg⁻¹ 9AC as a suspension in lipiodol. Both 9ACG and 9AC inhibited LS174T tumour growth by about 80% (Figure 3A). Neither 9ACG nor 9AC produced significant weight loss in this experiment (Figure 3B). To investigate the influence of dose and schedule on the anti-tumour activity of 9ACG, BALB/c nu/nu mice bearing LS174T xenografts were i.v. injected once with 50 mg kg⁻¹ 9ACG or three times with 10 mg kg⁻¹ 9ACG. Tumour-bearing mice were also i.v. injected three times with 10 mg kg⁻¹ irinotecan or s.c. injected once with 5 mg kg⁻¹ 9AC. The fractionated 9ACG schedule, despite a lower total dose, produced similar anti-tumour activity as one bolus injection of 9ACG (Figure 3C). The anti-tumour activity produced by 9ACG and CPT−11 was similar. 9AC produced good anti-tumour activity but was unexpectedly toxic, killing three of six mice by day 8. Although 9ACG and CPT-11 produced similar maximum weight loss of about 20%, one of six mice died after treatment with 50 mg kg⁻¹ 9ACG or three fractionated doses of 10 mg kg⁻¹ 9ACG.

Intravenous administration of 50 mg kg⁻¹ 9ACG on days 10 and 20 resulted in complete regression of six of seven CL1-5 tumours (Figure 4A). The treated mice experienced 14% weight loss after the second prodrug injection, but all the mice quickly recovered and long-term observation (100 days after therapy) did not reveal any signs of delayed toxicity. Figure 4B compares the treatment of CL1-5 tumours by 9ACG, 9AC or irinotecan. 9AC significantly (P⩽0.05) suppressed tumour growth but four of six mice experienced tumour relapse within 100 days. Irinotecan produced long-term tumour regression in two of five mice but tumour size was not significantly delayed as compared to PBS-treated control mice. One i.v. injection of 50 mg kg⁻¹ 9ACG significantly (P⩽0.05) suppressed tumour growth; five of seven mice were apparently cured and tumour growth was suppressed in another mouse for over 100 days. One mouse died after prodrug treatment indicating toxicity although the maximum mean weight loss was only 7.5% after prodrug treatment. One of five control tumours failed to grow in this experiment. Figure 4C compares the treatment of CL1-5 tumours with fractionated doses of 9ACG or topotecan. Treatment with six injections of 8 mg kg⁻¹ 9ACG produced significant (P⩽0.05) delay of tumour growth and tumour regressions in three of seven mice. A previously described optimal schedule for topotecan (Zamboni et al, 1998) displayed minimal anti-tumour activity against CL1-5 xenografts and was highly toxic; two of six mice died and mean mouse weight was significantly (P⩽0.0005) decreased as compared with control mice. The weight of mice treated with 9ACG did not significantly differ from control mice in this experiment.