Interaction of thalidomide, phthalimide analogues of thalidomide and pentoxifylline with the anti-tumour agent 5,6-dimethylxanthenone-4-acetic acid: concomitant reduction of serum tumour necrosis factor-alpha and enhancement of anti-tumour activity.

DMXAA (5,6-dimethylxanthenone-4-acetic acid), a novel anti-tumour agent currently undergoing clinical evaluation, appears to mediate its anti-tumour effects through immune modulation and the production of the cytokine tumour necrosis factor-alpha (TNF). Our previous studies have shown that thalidomide, a potent inhibitor of TNF biosynthesis that has numerous biological effects, including inhibition of tumour angiogenesis, unexpectedly augments the anti-tumour response in mice to DMXAA. We show here that thalidomide (100 mg kg(-1)) has no effect when administered with inactive doses of DMXAA, and that it must be given simultaneously with an active dose of DMXAA to have its maximum potentiating effect on the growth of the murine Colon 38 adenocarcinoma. To address the issue of whether inhibition of serum TNF production is important for potentiation of anti-tumour activity, we have tested three potent analogues of thalidomide. All three analogues, when co-administered with DMXAA to mice at doses lower than those used with thalidomide, inhibited TNF production and were effective in potentiating the anti-tumour activity of DMXAA against transplanted Colon 38 tumours. One of the analogues, N-phenethyltetrafluorophthalimide, was 1000-fold more potent than thalidomide and at a dose of 0.1 mg kg(-1) in combination with DMXAA (30 mg kg(-1)) cured 100% of mice, compared with 67% for the group treated with DMXAA alone. We also tested pentoxifylline and found it to suppress TNF production in response to DMXAA and to potentiate the anti-tumour effect of DMXAA. The results are compatible with the hypothesis that pharmacological reduction of serum TNF levels might benefit the anti-tumour effects of DMXAA and suggest new strategies for therapy using this agent.

drug flax one acetic acid (FAA that are highly actix e against transplantable murine tumours w-ith an established Xasculature (Rex-castle et al. 1989(Rex-castle et al. . 1991. The most potent of these. 5.6-dimethN-lxanthenone4-acetic acid (DMXAA). is now-in phase I clinical trials in New Zealand and the UK. In mice. DNIXAA is 12-fold more potent than FAA and induces a higher percentage of cures acainst the Colon 38 carcinoma (Rexwcastle et al. 1991).
DNIXAA and FAA share a mechanism of action that is different from that of conventional direct cvtotoxic anti-cancer drugs. Both appear to actix-ate. through host and tumour cell components. a complex series of responses involving shutdow n of tumour blood floxw (Z"-i et al. 1989. 1994). stimulation of immune responses Bagulev. 1987: Ching et al. 1991) and elimination of the tumour.
NManv of the biolo2ical actix-ities of DMNXAA and FAA have been attributed to their abilitv to induce cxtokines. in particular TNF and the interferons (Mace et al. 1990: Futami et al. 1992: be mediated by tumour necrosis factor-a (TNT) as antibodies to T-NF ablate FAA-induced tumour x-ascular collapse (Mahadex an et al. 1990). DMXAA induces hiaher lexels of serum TNF than FAA. and the anti-tumour response correlates w ell w-ith TNF production w ithin a series of DMXAA analoaues (Philpott et al. 1995).
As an approach to inx estigating the role of TNF induction in the anti-tumour action of DNMXAA. wxe co-administered thalidomide wxith DMXAA to mice x ith subcutaneous Colon 38 tumours . Thalidomide. best knowxn for its sedatixe and teratogenic effects (Fabro et al. 1967). has receixed attention in recent -ears as a selectixe inhibitor of TNT production Sampaio et al. 1991 ). apparently acting, by increasing the rate of degradation of TNT m_RNA (Moreira et al. 1993). We haxe shoxwn that thalidomide inhibits DMXAA-induced serum TNF lexels. but. unexpectedly. potentiates the anti-tumour response . While these results appear to argue against a role for TNF in the anti-tumour action of D.MXA. they indicate a noxvel approach to augmenting the anti-tumour action of DMXAA.
Recently. sexeral phthalimide-derixved analogues of thalidomide hax-e been described that are more potent than thalidomide in modulating TNF production in cells stimulated with a phorbol ester (Sasaki et al. 1995). The druc pentoxif-lline is also knoxxn to suppress TNF production in response to administration of lipopolx-sacchande (Noel et al. 1990). In this report. w-e have further investioated the interaction of thalidomide and DMXAA and has-e investigated a series of three phthalimide derivatives. as ell as pentoxif lline. to determine whether they also suppress TNF production in response to DMXAA and w hether thev potentiate DMXAA-induced anti-tumour effects.

Measurement of anti-tumour activity
All experiments were carried out in 8to 12-vveek-old C57B1/6 x DBAI2 F (BDF, mice bred in the laboratorx animal facilitv and treated according to institutional auidelines. Fragments of Colon 38 tumour (1 mm¼ ) were implanted subcutaneously in the flank of anaesthetized (sodium pentobarbital. 90mr kg-' i.p.) animals. Experiments were carried out on tumours approximately 4-5 mm in diameter. grenerally 10 days after implantation. Tumour-bearing, mice (at least five per group) were injected w-ith drugs and the tumours measured with callipers two or three times per week thereafter. Tumour volumes were measured as 0.52 a: x b. where a and b are the minor and major axes of the tumour. The arithmetic mean and standard error of the means A-ere calculated for each time point. including animals having zero measured tumour volume. and expressed as a fraction of the pretreatment volume. Growth delav was determined as the difference in the number of days required for the control and treated tumours to reach four times the pretreatment xolume. Statistical tests w-ere carried out using SigmaStat (Jandel Scientific. San Rafael. CA. USA). Mice cured of tumours were kept for at least 3 months to ensure that tumours did not regrow.

Histological examination of tumour sections
Mice w-ith Colon 38 tumours w-ere treated with DMXAA (single dose. 24 mg kg-'). either alone or w-ith thalidomide (100 mg kg-'). Tumours were excised 1-10 dayvs after treatment. fixed overnight in 10% formalin. Fixed tumours were then embedded in paraffin wax and sections were stained w-ith haematoxvlin and eosin. The section across the major diameter of the tumour w-as examined on a grid marked at 0.4-mm intervals and scored for the percentage area of viable tissue as previously described (Bagulev et al. 1989).

Determination of serum TNF
Mice were anaesthetized using halothane and wA-ere bled from the ocular sinus at indicated times after treatment. Blood w as allowed to clot ovemirht on ice and the serum collected by centrifu2ation (2000 g. 30 min) and stored at -20 C until it was assayed for TNF activity. using the standard L929 cvtotoxicitx assay as described (Philpott et al. 1995). L929 cells (3 x 10-) were allowed to adhere ox erniht to the bottom of flat-bottomed 96ell plates. Actinomycin D (final concentration 8 ,gr ml-') was then added to the wells followed by serial dilutions of the serum to be assayed. Killinc of the L929 cells was assessed after 24 h by a colorimetric assay using 3-(4.5-dimethvl-2-thiazoly l)-2.5-dipheny-l-2H-tetrazolium bromide. One unit of TNF A as defined as that reducing cell staining in this assay by 50%. and corresponded to the activitx obtained with 10-" of punrfied murine TNF protein.

RESULTS
Potentiation of anti-tumour activity of DMXAA by thalidomide Thalidomide potentiates the anti-tumour response of DMXAA at its optimal therapeutic dose of 30 mg kg-I when administered simultaneously . To investigate whether thalidomide potentiated suboptimal doses of DMXAA. mice with Colon 38 tumours were treated with thalidomide (100 mg kg-') and a range of doses of DMXAA from 15 mg, kg-I to the maximaltolerated dose (MTD) of 30 mg kg-'. Tumour growth delays and cure rates appeared to be increased by thalidomide after DMXAA doses of 25 and 20 mc kg-' but not after a dose of 15 mg kg-I ( Figure 2 and Table 1). How-ever. the differences between groups were not statistically significant. analysis beina complicated by the percentage of cures in each group. We also examined the effect of changing the timinc of administration. using DMXAA at its MTD.
W'hile ggrowth delays were potentiated by thalidomide given simultaneously wvith DMXAA. in agreement w-ith previous results . they were not significantly increased when thalidomide w-as given 1 day before. or 1. 2 or 3 davs after DMXAA ( Figure 3 and Table 1).
Histology of tumour regrowth after treatment with DMXAA and thalidomide The histology of tumours treated with DMXAA either alone or in combination with thalidomide (100 mr kg-') was examined by measuring the proportion of viable tissue in tumour sections taken at different times after treatment. A suboptimal dose of DMXAA w as used (24 mg kg-' ) to permit a significant amount of viable tissue to be measured. Extensive haemorrhagic necrosis w-as evident in both treatment groups when measured after 24 h. However. pockets of viable tissue that were visible in tumours treated with DMXAA were less evident than those in tumours treated w-ith the combination. and the amount of viable tissue in tumours treated with the combination w-as lower than that in tumours treated with DMXAA alone (   either alone (0) or together with thalidomide (100 mg kg-) (-) no effect was found on the take-rate or growth of the Colon 38 tumour. Rather. the rate of growth was slightly accelerated bv thalidomide treatment. and the tumours became palpable sooner in the treated amnmals ( Figure 5). Thus there wias no evidence for inhibition of neoxascularization. and thus of tumour growth and development. usinc thalidomide alone at doses of 100 mg kg-

Effect of thalidomide alone on tumour development and growth
Thalidomide is known to inhibit anciooenesis (DlAmato et al. 1994). and angiogenesis antagonists can increase the anti-tumour response to radiation and chemotherapy through inhibition of tumour neovascularization (Teicher et al. 1993: D'Amato et al. 1994. To test the hypothesis that thalidomide. administered alone. was affecting the growth of Colon 38 tumours. mice w-ere treated either w-ith a single dose or ,iith multiple doses of thalidomide (100 mg kgper injection). No inhibition of tumour growth was observed in mice treated w-ith a single injection of thalidomide (100 mg kg-'). or ith injections three times A eekly for the duration of the experiment (1100 mg kgper injection). Furthermore. Awhen thalidomide alone A as gixven at the time of tumour implantation and three times A eeklv thereafter (100 mg kg-' per injection).
Ability of analogues of thalidomide to inhibit serum TNF production and potentiate the anti-tumour action of DMXAA A series of simple phthalimide derivatives that are more potent than thalidomide in modulating serum TNF production has been reported (Sasaki et al. 1995). We synthesized three of these more dose-potent derivatives (PP. PEP and PEFP) and compared their abilitx v ith thalidomide to suppress DMLXAA-induced serum TNF production and to potentiate anti-tumour action. PEFP w-as the most toxic of the three compounds. inducing deaths at 3 days at doses above 50 mg kg-'. whereas the other two derivatives were A-ell tolerated at 100 mg kg-1. In comparison. thalidomide was well tolerated up to 250 mg kg-I in mice . Comparisons of inhibition of serum TNF production w-ere carried out using a dose of DMXAA (50 mg kg-') that Awas optimal British Joumal of Cancer (1998)  inhibited TNF lexels to 58. 18 and 8.5%k. respectively. of DMXVAA controls (Figure 6). They xx-ere thus more potent than thalidomide.
"-hich was actix-e at a dose of 0.3 mg kg-. In contrast to thalidomide. none of the phthalimide derivatixves caused sedation (results not sho"-n).
We next compared the ability of the phthalimide analogues at a dose (10 mg kg-') that suppressed serum TNF production (Ficure 6) to potentiate the anti-tumour response of DMXAA (30mgr kg-). Combination of DMXAA A-ith PEFP and PP produced cures against the Colon 38 tumour in 100% of the mice (Ficure 7). A-hile combination with PEP caxve cures in 75%7e of the mice and extended the growth delay to oxer 60 days. At this dose. thalidomide "-as the least actix-e of the four agents. Thus. all three phthalimide derixatixes appeared to be more potent than thalidomide in potentiating the anti-tumour actix itv of DMXAA.
As PEFP "-as the most actix-e in reducing serum TNF lexels Ficure 6). A e tested this compound at low er doses for potentiation of the anti-tumour action. PEFP A. 1 mg kgin combination "-ith DNIX.AA (30 mc kg-1) induced cures in 100%' of the animals (Figrure 8) and thus >-as at least 1000-fold more potent than thalidomide. A hich required a dose of 100 mc ko' in combination wvith DNMXAA for a 100% cure rate (Table 1). PEFP had no antitumour effects on its ow n at its maximal-tolerated dose (Figure 8). and this >-as also the case for PEP and PP (100 mg k-': data not shox-n).
Inhibition of serum TNF production and potentiation of DMXAA anti-tumour action by pentoxifylline The studies w-ith the phthalimide derixatixes extended our prexious demonstration that inhibition of serum DMXAAinduced TNF production xxas concomitant x-ith the potentiation of DMXAA anti-tumour action (Figures 6 and 7). We therefore examined pentoxifylline. a structurallv unrelated inhibitor (Figyure 1). w-hich. like thalidomide. inhibits TNF production in response to lipopolx-saccharide (Noel et al. 1990). We found that co-administration of pentoxifylline at doses of 12.5-100 mg kg-1 reduced DMXAA-induced serum T'NF bv 50-805 (data not shown). Co-administration of pentoxifxlline (50 mg kgYincreased the growth delav induced by DNIXAA (30 mg kg-) from 19 to more than 40 dayvs ( Figure 9A) and increased the cure rate from 64A' to 82%. but these differences were not statistically si2nificant. Co-administration of DNIXAA w-ith pentoxifylline (100 mg kgx-) induced complete tumour regressions in 100% of the mice ( Figure 9B DISCUSSION W'e have previously demonstrated that thalidomide. w hile reducina DNIXAA-induced increases in serum TNF. potentiates the anti-tumour response of DNIXAA  ). We have show n similar effects to thalidomide firstly for three derivatives of phthalimide structurally related to thalidomide (Sasaki et al. 1995) and secondly for pentoxifA-lline. w-hich differs from thalidomide in both structure and mechanism of TNT inhibition (Han et al. 1990: Sampaio et al. 1991. The augmentation of anti-tumour activitv bv these drugs (Figures 7-9) is difficult to assess statisticallv because DMXAA alone induces a percentage of cures. and a large number of animals are therefore required to achiex-e statisticall1 sienificant differences in cure rates. How-ever. taken together. the results are consistent w-ith the conclusion that the potentiation of the antitumour action of DMNXAA is associated w ith the common property of pentoxifylline. thalidomide and the phthalimides to decrease serum TNF production. Co-administration of agents low erincy serum TNF may thus represent an innoxvative strateav for increasing the anti-tumour efficacy of DMIX.AA. a drug that is currently in phase I clinical trial.
Thalidomide has a number of pharmacological actions. includinc the inhibition of angiogenesis (DAmato et al. 1994). raisin2 the question of w hether potentiation of DMXAA actix ity is mediated by its anti-angiogenic properties. Administered alone. thalidomide had no inhibitorx effect on the row-th of the Colon 38 tumour . and repeated dosing did not affect the development and growxth of the Colon 38 tumour (Fioure 5. A single dose. administered at the same time as DMXAA (Fiaure 3).
was required for the synergistic effect. and thalidomide did not affect tumour growth when administered with an ineffective dose of DMXAA (Figure 2). Inhibition of tumour angiogenesis requires repeated or continuous application of the angiogenesis antaconist (D'Amato et al. 1994). and it is likely that higher doses than that used in these experiments are necessarx for inhibition of neox ascularization (RJ d'Amato. personal communication). W'hile we have not ruled out completely that thalidomide may be potentiating the anti-tumour activitv of DMXAA through inhibition of anaioeenesis. it appears more likely that it is acting in some other fashion.
Histoloaical studies indicate that the augmented tumour growth inhibition of the combined therapy stems from an acute effect resulting in a greater reduction in tumour cell survival at 24 h and from a subsequent slower rate of regeneration (Figure 4. These obserx ations suggest that thalidomide minht in some w-ax increase the actions of the induced cvtokines. The mechanism bx which thalidomide. the phthalimide derixatixes and pentoxifx lline inhibit serum TNF production is not vet British Joumal of Cancer (1998)  Thalidomide plus DMXAA (50) Figure 6 Suppression of DMXAA-induced serum TNF actvity by thalidomide and its analogues. Individual mice were treated with DMXAA (50 mg kg-') alone or together with thalidomide, PP, PEP or PEFP at the indicated doses, or with drug aloe. Sera were collected 2 h later and assayed for TNF actty. The control actvity for PP and PEP was 3393 units and that for PEFP and thalidomide was 2440 units understood. In cultured cells. thalidomide can either inhibit or enhance TNF production. depending on the conditions and the cell type. TNF production by HL-60 cells in response to phorbol esters as increased by co-incubation with thalidomide, as well as by the phthalimide analogues used in this study (Nishimura et al. 1994). Thalidomide also enhanced TNF svnthesis (in response to lipopolysaccharide) in THP-1 cells and in human mononuclear cells enriched for adherent cells. but inhibited TNF svnthesis in cultures of unfractionated peripheral blood cells (Shannon and Sandoval. 1996). We do not yet know whether these mechanisms will apply in vivo or whether other mechanisms are insols ed. We are currently investigating whether these drugs affect the in vi--o pharmacokinetics of DMXAA.
The observation that lowering of serum TNF levels is associated with improving the anti-tumour activity of DMXAA appears to contradict previous studies that show that antibodies to TNF inhibit tumour vascular collapse and ameliorate FAA-induced anti-tumour action (Mahadevan et al. 1990: Pratesi et al. 1990). In another tumour model. we have shown that antibodies to TNF partly inhibit the action of DMXAA. consistent with the above observations (WL Browne et al. manuscript in preparation). In recent studies. we have shown that TNF is produced in response to DMXAA in tumour tissue and that co-administration of thalidomide. while reducing serum TNF. does not reduce tumour-associated TNF (WL Browne et  TNF is produced in tumours and to identify the source of the TNT in serum. How-ever. we can conclude from the data obtained so far that the inhibition of serum TNF bv thalidomide and the other drugs investigated here does not contradict the hypothesis that TNF mediates the effect of DMXAA in tumours.
In conclusion. we have shown here that pharmacological modulators of TNT production can be applied to reduce serum levels of TNF while increasing the anti-tumour effects of DMXAA. The phthalimide derivatives are more activ-e and potent pentoxifylline at a dose of 100 mg kgthan thalidomide and lack the sedatory effects. presumably through the elimination of the Olutarimide substituent of the molecule and interaction with the glutamate receptors in the brain. The co-administration of pharmacological TNF modulators such as these may lead to improved strategies to exploit the novel biological properties of DMXAA in a clinical situation.