Combining bioreductive drugs and radiation for the treatment of solid tumors☆☆☆

Early concepts in the development of bioreductive drugs 

The idea of bioreductive drug activation and the selective killing of hypoxic cells developed along 2 complimentary pathways. The first of these arose from observations made by Sutherland8 who noted that the hypoxic cell radiosensitizer, metronidazole, appeared to selectively kill the central hypoxic cells in multicellular spheroids. The second approach came from the recognition by Kennedy et al9 and Sartorelli10 that the presence of hypoxia in tumors created an environment conducive to reductive processes. They showed that mitomycin-C (MMC) could be reductively activated by a number of oxidoreductases, resulting in preferential activation of MMC under hypoxic conditions, leading to enhanced killing of hypoxic as apposed to aerobic cells. From these initial observations, it became clear that reduction was an absolute requirement for drug activation, and for this to occur, cells need to possess the appropriate complement of reductase enzymes to carry out the activation process.

There are 4 main classes of bioreductive drugs currently in various stages of clinical and preclinical evaluation. They are quinones, based on the indolequinone nucleus (eg, MMC, EO9); the nitroheterocyclics, such as the hypoxic bioreductive marker drugs, pimonidazole and SR 4554, and the cytotoxins CB1954 and SN23862; the aromatic N-oxides, typified by tirapazamine (TPZ) and aliphatic N-oxides, such as AQ4N. For the activation of these drugs under hypoxic conditions, the initial reduction step involves addition of an electron. In the absence of specific substrate requirements, the more electron-affinic compounds are likely to be reduced more easily and hence are likely to be more toxic. This is shown in Figure 1 in which the toxicity of a range of nitroheterocyclics toward hypoxic V79 cells is correlated with the 1-electron reduction potential (E17) of the different compounds.

View Large Image Download to PowerPoint

Fig. 1. Correlation between the hypoxic toxicity of various nitroheterocycles toward Chinese hamster V79 cells and the 1-electron reduction potential of the compounds. The more electron-affinic compounds (ie, less negative reduction potential) showing the greater toxicity.  ̂, 2-nitroimidazoles; ●, 5-nitroimidazles; ▵, 5-nitrofurons.

Walton and Workman17 showed the importance of the 1-electron reductase cytochrome P450 reductase (P450R) in the hypoxic activation of nitroheterocyclic compounds. This was subsequently confirmed by Patterson et al18 who showed that in a breast cancer cell line transfected with the gene encoding P450R, the toxicity of RSU1069 was increased as a function of the activity of the transfected enzyme. Similarly, the importance of P450R for the activation and metabolism of TPZ was shown.19, 20 This dependence on P450R for drug activation and toxicity has led to the evaluation of the use of P450R in gene-directed enzyme prodrug therapy (GDEPT) (gene directed enzyme prodrug therapy) strategies (see later). In these studies,21 it was also shown that the indolequinone EO9 showed dependence on P450R for activation and toxicity under hypoxic conditions. However, it was apparent for quinones that other reductases could play an important part in the activation process. In particular, DT-diaphorase was shown to be critical for activation of EO9 and other quinones in air (DT-diaphorase is an O2-independent 2-electron reductase). DT-diaphorase or NQO1, its closely related isozyme NQO2, and other O2-independent nitroreductases are also being developed for (antibody-directed enzyme prodrug therapy) and GDEPT approaches used together with agents such as CB1954.22 The aliphatic N-oxide, AQ4N, in contrast to the other bioreductive drugs, is predominantly metabolised by enzymes of the P450 family. Because expression of P450 is down regulated in cell lines in vitro the activity of this agent was only unambiguously revealed when tumor cells were treated in vivo.

Thus, for the optimal use of bioreductive drugs, it will be important to know the enzyme profile of the tumours being treated. Ideally, if this can be combined with a measure of hypoxia in those tumor (which is becoming a realistic possibility with the use of intrinsic molecular markers of hypoxia),23, 24 then bioreductive drugs may be applied in the most efficient manner.

Bioreductive drugs used with radiation 

The observations that hypoxic cell radiosensitizers could also function as hypoxic cell cytotoxins led to a more detailed evaluation of what contribution cytotoxicity could make to the outcome of radiosensitization experiments carried out in vivo. Conventionally, radiosensitizers such as metronidazole, misonidazole, etanidazole, and nimorazole are given to experimental mice before radiation because for sensitisation to occur drugs only need to be present in the hypoxic tumour cells at the instant of irradiation. Whereas for toxicity to occur, it would be expected that exposure of hypoxic cells to the drugs would be most important (ie, concentration × time). Such exposure could equally occur if the drug were given before or after radiation treatment.

Some of the first experiments to show that toxicity towards hypoxic cells occurred in vivo were carried out by Brown et al.25 In these experiments, tumor-bearing mice were given 1 mg/g misonidazole; then 24 hours later tumors were excised and the extent of necrosis in tumours measured relative to untreated controls. The drug-treated tumors showed a substantial increase in the areas of necrosis, and one explanation was toxicity toward those chronically hypoxic cells surrounding the areas of necrosis. More compelling evidence for hypoxic toxicity brought about by these radiosensitizer/bioreductive drugs in vivo came from experiments in which misonidazole was given before or after radiation. An early example of this type of experiment is given in Figure 2.26

View Large Image Download to PowerPoint

Fig. 2. Growth delay of the MT fibrosarcoma grown in WHT mice given graded doses of radiographs. X, radiation only;  ̂, mice given 1 mg/g misonidazole 30 minutes before irradiation; ●, ▵, mice given 1 mg/g misonidazole 5 minutes after irradiation. The group of mice illustrated by the open triangle were held at 37°C for 3.5 hours after treatment with Misonidazole. Upward arrows indicate cures in those groups.26.

The radiosentizers/hypoxic cell cytotoxins such as misonidazole showed approximately a 10-fold increase in toxicity toward hypoxic versus aerobic rodent cell lines in vitro.14, 27 This differential toxicity was similar to or somewhat greater than that being reported in parallel experiments with MMC.28 These results stimulated a search for agents that could show higher differences in toxicity between hypoxic and aerobic cells.

The first agent to show a substantive improvement in differential hypoxic cell toxicity was RSU1069. This was a 2-nitroimidazole containing a potentially alkylating aziridine group in the N1 side chain. Thus, the compound was a monofunctional alkylating agent in air, whereas under hypoxic conditions the drug could be converted to a bifunctional agent via reduction of the nitro group to nitroso or hydroxylamine reactive species.29, 30 This resulted in the compound being up to 100 times more toxic to hypoxic relative to aerobic cells. When combined with radiation to treat experimental tumours in vivo, RSU1069 was equally effective when given either before or after radiation, which confirmed the potency of the agent as a hypoxic cytotoxin.31 Furthermore, RSU1069 was also effective when given in a fractionated radiation schedule.32 Table 1 details the growth delay obtained in the RIF-1 tumour treated with RSU1069 given with each of 8 fractions of 2.5 Gy radiation delivered every 12 hours. Clearly, the drug treated group responds substantially better than the group receiving radiation only. Table 2 also shows results obtained when the blood supply to the subcutaneous tumor was occluded, by physical clamping, for 30 minutes after each fraction of radiation.

Table 2. Fractionated radiation with RSU1069 plus 30-minute clamping of RIF-1 tumors immediately after radiation

	Treatment	Time to 4 × Volume (d)
	XR → clamp (× 8)	15.8 ± 1.0
	RSU1069 → clamp (× 8)	18.5 ± 2.5
	RSU1069 → XR → clamp	80.9 ± 8.7 (1/6 cures)

Table 1. Fractionated radiation experiments with TPZ and RSU1069 in the RIF-1 tumor

	Treatment	Time to 4 × Volume (days)
	Untreated	5.8 ± 0.5
	8 × 2.5 Gy (XR)	15.9 ± 1.1
	TPZ → XR	22.6 ± 1.4
	RSU1069 → XR	22.3 ± 2.8

RB6145 is a chemical prodrug of RSU1069. It retains both the radiosensitizing and bioreductive properties of RSU1069 but was considerably less toxic in animal studies.34 RB6145 is metabolized in vivo with 30% conversion to RSU1069.35, 36 In experimental models, when RB6145 was given at 3 times the dosage of RSU1069, it was just as effective as a bioreductive drug when used in combination with radiation but substantially less toxic.34 CI-1010, the R-enantiomer of RB6145, caused marked retinal toxicity in mice, rats, dogs, and monkeys. Because of this serious side effect, RB6145 and other alkylating nitroimidazoles were not developed further as bioreductive drugs. The retinal toxicity seen with CI-1010 was seen to a much lesser extent with TPZ (but not detected at all in clinical trials of TPZ, see later).37 Further, retinal toxicity in experimental models has not been detected at all with other bioreductive drugs.37

The benztriazine-di-N-oxide TPZ (SR4233) emerged from a National Cancer Institute (United States) radiosensitizer synthesis and screening programme awarded to Brown and colleagues. The agent was found to kill unirradiated hypoxic (relative to aerobic) cells in the radiosensitizer screen and Brown recognized that TPZ could be a potent bioreductive drug, which indeed proved to be the case.38, 39, 40 TPZ is activated by 1-electron reduction to a nitroxide radical intermediate, which is thought to abstract a hydrogen atom from DNA to produce single- and double-strand breaks, leading to chromosomal defects. These aberrations are significantly more difficult to repair than those produced by x-rays.41, 42 TPZ exhibits considerable selective cytotoxicity, with oxic/hypoxic differentials in the order of 30- to 300-fold in most murine and human cell lines.38, 43 When combined with radiation, the drug was equally effective if given just before or after single doses of radiation. TPZ is also effective when given with fractionated radiation treatment.44, 45 An example of the potency of the drug is given in Table 1; it is apparent that TPZ is at least as effective as RSU1069. Brown et al2, 46 further developed the theory of rehypoxiation between fractions of radiation. A consequence of this would be that, when adding a hypoxic cytotoxin to radiation, the potentiation of the radiation effect, or enhancement ratio, should increase with dose fractionation (ie, there should be a greater potentiation with fractionated irradiation than with single doses) Precisely the opposite occurs with hypoxic cell radiosensitizers because of reoxygenation between doses (eg, the enhancement ratio for tumour response when a sensitizer is added to radiation decreases with increasing number of fractions).47 This prediction has been tested and found to hold when TPZ is added to single or multiple doses of radiation given to the SCCVII tumor.48 The fact that hypoxic cytotoxins can take advantage of reoxygenation/rehypoxiation between doses is a major reason for the theoretical superiority of these agents over hypoxic radiosensitizers.

Preclinical studies have shown that TPZ is also extremely effective in potentiating the activity of chemotherapeutic drugs such as cisplatin and cyclophosphamide. As a consequence, TPZ is currently in a variety of phase 2 and 3 clinical trials as an adjunct to radiotherapy alone or cisplatin-based chemotherapy or chemo/radiotherapy. Recent results have shown a dramatic survival advantage for patients with non–small-cell lung cancer treated with TPZ and cisplatin, compared to cisplatin monotherapy.16

MMC can be regarded as the prototype bioreductive drug. It was isolated from Streptomyces caespitosus and has been used in the treatment of cancer for nearly 20 years, gaining a place in the therapy of a number of solid tumors, including lung and breast cancer. MMC shows a small but significant increase in toxicity toward hypoxic cells in vitro,28, 43 and when combined with radiation to treat experimental tumors in vivo, MMC causes a reduction of the tumor hypoxic cell fraction.49 Interestingly, the combination of MMC with radiation appears to be additive,49 whereas in the experimental work with TPZ and RSU1069/RB6145, it appeared (at least within the sensitivity of the in vivo/in vitro excision assay) that the latter agents eliminated the radioresistant hypoxic cell population within tumours.31, 32 Nevertheless, MMC has been used in a randomized study in squamous-cell carcinoma of the head and neck looking at the role of postoperative radiation therapy, with or without MMC.50, 51 The drug was administered soon after the start of 5 weeks of radiotherapy, and some patients received a second dose 6 weeks later. From a cohort of 117 patients, an increase in the actuarial disease-free survival was found in those given the combined treatment, compared with patients receiving radiotherapy alone (75% v 49%, P < .07, median follow-up >5 years). It was concluded that a treatment benefit was obtained without a significant increase in toxicity related to the MMC. The dose-limiting toxicity includes prolonged myelosuppression, pulmonary fibrosis, cardiotoxicity, and nephrotoxicity, and because of these side effects, other bioreductive drugs related to MMC but with greater oxic/hypoxic differentials for toxicity have been sought.

Porfiromycin is an analog of MMC and has a greater aerobic/hypoxic differential cytotoxicity than MMC.52 Compared with MMC, which has an approximately 2-fold selective toxicity toward hypoxic cells, porfiromycin exhibits a broader 10-fold oxic/hypoxic differential cytotoxicity. This superior hypoxia selectivity is predominantly because of a lowering of the aerobic toxicity and has been shown to be related to the lower incidence of DNA crosslinks generated by porfiromycin under aerobic conditions. Porfiromycin is activated by both 1- and 2-electron reductases53 and, like MMC, it is the single-electron reduction pathway that confers the hypoxia-selective properties of this agent.54, 55 Rockwell et al52 showed the activity of porfiromycin when combined with radiation to treat experimental tumours and subsequently in a phase 1 toxicity study, porfiromycin was given concomitantly with radiotherapy to patients with head and neck tumors. This study was later extended into a phase 3 comparison of radiotherapy plus porfiromycin to radiotherapy plus MMC, again in patients with head and neck tumors. The early results suggested that the toxicity profile was acceptable.56

Compounds in the EO series, typified by EO9, are synthetic indolequinones, some of which are substantially more active than MMC under hypoxic conditions.57, 58 For EO9 and EO8, hypoxic/oxic differentials of 30 and 5000 respectively have been observed.59, 60 Initial studies focused on EO9 because DT diaphorase had been identified as an important enzyme for its bioreduction, and various in vitro studies correlated the potency of EO9 with the intracellular expression of the enzyme.57, 58 EO9 has been shown to strongly potentiate the action of radiation in vivo at drug doses in which E09 alone showed little cytotoxic effect.59 It was concluded that EO9 was functioning as an efficient hypoxic cell cytotoxin in vivo. In support of these observations, EO9 was shown to result in a greater than additive effect when given postradiotherapy to 4 different rat tumors.6 EO9 completed phase 1 and phase 2 clinical evaluations in non–small-cell lung cancer,62, 63 but these trials failed to show any significant antitumor activity. The reason for this may be partly because of the design of these studies since patients in both trials were treated with EO9 alone. It is perhaps inappropriate to evaluate any hypoxic cytotoxin as a single agent because only the hypoxic subpopulation of a tumor will be affected. Better results might have been achieved by combining the treatment with radiotherapy, in a similar approach to the studies with MMC, porfiromycin, and TPZ.

AQ4N is an aliphatic di-N-oxide analogue of mitoxantrone, which is metabolized by a concerted 2-electron reduction to AQ4M and then to AQ4. Despite the absence of a detectable 1-electron intermediate, metabolic reduction is readily inhibited by oxygen. The mechanism of inhibition may involve competition between oxygen and AQ4N for reducing species.64 This active metabolite, AQ4, is a DNA-affinic topoisomerase II inhibitor. AQ4N has a relatively weak antitumour activity per se. When combined with radiation in vivo, a substantial enhancement of antitumor efficacy is observed with a radiation-dose modification factor of about 2. This enhancement is found irrespective of whether AQ4N is administered before or after the radiation exposure, providing strong evidence that this agent does act as a bioreductive cytotoxin.65, 66, 67 AQ4N is also effective when given in a fractionated radiation schedule.65, 66 The greatest antitumor effects were obtained when AQ4N was given daily with each fraction; however, substantial benefit was also achieved when AQ4N was given on a once or twice weekly basis. The agent is now in phase 1 clinical trial in combination with radiation in the treatment of esophageal cancer.

Future directions 

Denny and colleagues3 first articulated the trigger/effecter concept in bioreductive drug development. Examples are given in Figures 3 and 4 in which the bioreductive prodrugs are putatively inactive until subjected to reductive metabolism under hypoxic conditions.

View Large Image Download to PowerPoint

Fig. 3. Mechanism of the reductive elimination of aspirin from a 2-nitroimidazole prodrug.68.

View Large Image Download to PowerPoint

Fig. 4. Mechanism of the reductive elimination of an alkylating mustard from an indolequinone prodrug.69.

Table 3. Toxicity of the prodrug MUP98176 and the putative toxic product of bioreduction (nitrogen mustard) toward hypoxic and aerobic MDA468 breast carcinoma cells (values of IC50 in μmol/L after 3 hours drug exposure)

		Air	N2	Differential
	MUP98176	34	0.07	485
	Active mustard	0.17	0.11	1.5

An important feature of MUP98176 is that the evicted mustard is capable of diffusion, resulting in a bystander cell kill. The bystander effect can be defined as an amplification of the effects of a prodrug beyond the cell in which it was activated. Diffusion of an activated metabolite establishes a differential K-value for metabolism and cytotoxicity, a property that can be exploited to avoid normal tissue toxicity while retaining complementary cell killing with ionising radiation. A reason for this is that the radical anions of quinones react orders of magnitude faster with oxygen than do nitroarene radicals.70 This is reflected in nitro radical anions being half maximally back oxidized at pO2 values around 0.5%, whereas semiquinone radicals are much more sensitive to O2 quenching, with this occurring around 0.02% O2. The marked sensitivity of semiquinones to back oxidation is reflected in their K-curves or oxygen-dependence for cell killing, where the half-value (K-value) is of the order of 0.01% O2 or less for MMC.71 In contrast, for most nitroimidazoles, the K-value is around 0.1% to 0.5% O2,13, 14, 15, 72 and for TPZ, it is even higher at 1% O2.15 Because the K-value for radiation sensitisation by oxygen is ~0.5% O2,73 the use of radiotherapy in combination with bioreductive drugs, such as the indolequinones that require stringent hypoxia for maximal efficiency, will leave cells at intermediate oxygen tensions resistant to radiation and unable to activate the bioreductive drug.74 However, this exquisite hypoxia dependence may be an advantage for indolequinones because the oxygen tensions required do not exist in pathophysiologically normal tissues. In contrast, intermediate oxygen concentrations occur in some normal tissues, and this may be associated with some of the side effects, such as retinopathy, observed in preclinical models treated with TPZ and the alkylating nitroimidazole RB6145.37 MUP98176, which consists of a diffusible cytotoxin linked to an indolequinone nucleus, will remain latent until the compound is reduced at extremely low oxygen concentrations, resulting in release of the oxygen-insensitive cytotoxic entity. Utilization of a cytotoxic leaving group with both a desirable pKa value and an inherent latency as a consequence of electron withdrawal (such as the aliphatic mustard in MUP98176) will ensure that futile redox cycling of the indolequinone will not result in the inappropriate release of the cytotoxin. The very rapid kinetics of back oxidation of semiquinones ensures the hypoxia specificity of this process.70, 75

The bystander effect is also important when applied in a GDEPT context (see later), in which the gene vector is unlikely to target greater than 10% to 20% of tumor cells in vivo, it is advantageous that the prodrug kills not only cells expressing the enzyme but also surrounding nontransfected cells. Thus, less than 100% infection of tumor cells can still achieve total cell kill by exploiting active metabolite transfer to induce a bystander effect.

There are a variety of GDEPT approaches using bioreductive drugs, but most of these use O2-independent nitroreductases so that hypoxia selectivity is lost.22 However, there are a number of examples in which hypoxic directed gene therapy has been combined with bioreductive drugs.76, 77 In one of these studies, the gene encoding P450R (which efficiently activates many bioreductive drugs under hypoxic conditions) has been placed under the control of transcriptional enhancer sequences called hypoxic responsive elements in the context of a minimal promotor sequence.77 This construct has been introduced into HT1080 cells and a stably transfected clone (R9) isolated. Table 4 shows the sensitivity of the transfected clone exposed to RSU1069 for 18 hours in vitro under aerobic or hypoxic conditions compared to empty vector transfected wild type cells.

Table 4. Toxicity of the bioreductive drug RSU 1069 toward parental HT1080 and R-9 (HRE/P450 reductase transfected) cells under oxic and hypoxic conditions77

		IC50 (μmol/L)
	Cells	Air	N2	HCR
	HT1080	54	1.61	33
	R-9	50	0.05	1,000

Abbreviations: HCR, hypoxia cytotoxicity ratio.

View Large Image Download to PowerPoint

Fig. 5. Tumor-free survival data for HT1080 (EGFP-5 and R-9) xenografts treated with radiographs plus or minus 250 mg/kg RB6145.77.

Conclusions 

The development of bioreductive drugs might be regarded as a mature science waiting for definitive, repeatable, clinical evidence that their use in combination with radiotherapy alone or chemo/radiotherapy will be of therapeutic benefit. However, the indicators are promising. There is firstly the positive interaction of TPZ combined with cisplatin in the treatment of non–small-cell lung cancer. Secondly, there are the positive clinical trials combining radiotherapy with MMC in the treatment of head and neck cancer. There is also the benefit seen in head and neck cancer when nimorazole is used as a radiosensitizer,78 which is now routinely used in some European countries. These latter results taken together with the measurements of pO2 in head and neck cancer using the Eppendorf oxygen electrode, which show a low O2 status (greater level of hypoxia) to be an adverse prognostic indicator, would argue strongly that bioreductive drugs should have a role in the treatment of this disease.

It is also known that hypoxia is an adverse prognostic indicator (as measured by the Eppendorf electrode or by the presence of intrinsic molecular markers23, 24) in many other cancers. Thus, bioreductive drugs are likely to feature in future treatments. It may or may not be those currently under clinical evaluation, but with the platform technology detailed earlier and the development of other bioreductive drugs of diverse structure,79, 80, 81, 82, 83 there is every prospect of the right drug being designed for the appropriate tumor treatment.