Nitric oxide involvement in the toxicity of hydroxyguanidine in leukaemia HL60 cells

The free-radical intermediates and the stable products formed on one-electron oxidation of hydroxyguanidine (HOG) were investigated in order to suggest a mechanistic basis for HOG-induced cytotoxicity and cytostasis in leukaemia HL60 cells. The azide radical (generated radiolytically) reacted with HOG to produce a carbon-centred radical which in the absence of oxygen decays by a first-order process (k = 3.2 x 10(3) s-1) to yield nitric oxide (NO) and urea. Although the HOG radical reacts rapidly with oxygen (rate constant for O2 addition, k = 4.2 x 10(8) dm3 mol-1 s-1) this neither prevented the elimination of NO. nor generated alternative nitrogen oxides (e.g. peroxynitrite) capable of contributing to cellular oxidative stress. The detection of NO. in HL60 cells corroborated mechanistic studies that oxidative denitrification of HOG does not require catalysis by nitric oxide synthase. Quantitation of NO. by electron paramagnetic resonance (EPR) spectroscopy (utilising a NO. -selective probe) shows higher amounts of NO. under anoxic conditions, reflecting competition for NO. with molecular oxygen in oxic cells. Inhibition of cytochrome P450 and myeloperoxidase activity decreased NO. production thereby identifying these enzyme systems as capable of oxidizing HOG in vitro. A correlation exists between the intracellular levels of NO. with both the cytotoxic and cytostatic effects of HOG within HL60 cells. A higher toxicity was observed with hypoxic than with oxic cells. The lower levels of NO. associated with aerobic conditions caused a G1 --> S block in the cell cycle which under anoxia potentiated NO. -induced apoptotic cell death.

Oxidative denitrification of N omega-hydroxy-L-arginine by the superoxide radical anion

The superoxide radical anion (O2-.) produced during the catalytic activity of nitric oxide synthase (NOS) and cytochrome P-450 has been implicated in the oxidative denitrification of hydroxyguanidines ( > C = NOH). The reactivity of the radiolytically generated O2-. radical with N omega-hydroxy-L-arginine (NHA) is pH dependent and appears to parallel the prototropic equilibrium of the hydroxyguanidino group ( > C = NOH reversible > C = NO(-)+H+; pK = 8). The N omega-hydroxyguanidino group is more reactive towards O2-. when deprotonated but exhibits negligible reactivity when protonated. Based on a model, the rate constant for the reaction of the O2-. with NHA was estimated as kappa (O2-.+ > C = NO-) approximately 200-500 M-1.s-1, which is probably too low to compete with O2-. reactions with NO- or superoxide dismutase, which occur many orders of magnitude faster. The oxidative elimination of NO from NHA by O2-. was not accompanied by the formation of L-citrulline. Since only 21% of NHA will exist in the deprotonated > C = NO- form at physiological pH, it is unlikely that oxidative denitrification of NHA by cytochrome P-450 or NOS-derived O2-. radicals will prove a major free-radical pathway to NO. and L-citrulline.

Chemical properties which control selectivity and efficacy of aromatic N-oxide bioreductive drugs

Pulse radiolysis was used to generate radicals from one electron reduction of 1,2,4-benzotriazine-1,4-dioxides (derivatives of tirapazamine), and of imidazo [1,2-a]quinoxaline-4-oxides (analogues of RB90740), which have selective toxicity towards hypoxic cells. Radicals from the mono N-oxides (from the latter compounds) react with oxygen approximately 10-40 times faster than does the tirapazamine radical. Radicals from the tirapazamine analogues studied react with oxygen up to approximately 10 times slower than tirapazamine radicals. The quinoxaline N-oxide radicals are involved in prototropic equilibria with pK(a) values (5.5 to 7.4) spanning that reported for tirapazamine (6.0). Generation of radicals radiolytically in the presence of H donors (formate, 2-propanol, deoxyribose) indicate a chain reaction ascribed to H abstraction by the drug radical. The protonated drug radical is much more reactive than the radical anion (H abstraction rate constant approximately equal to 10(2) - 10(3) dm3 mol-1 s-1). Chain termination is ascribed to drug radical-radical reactions, i.e. radical stability in anoxia, with rate constants 2k approximately equal to 1 x 10(7) to 2 x 10(8) dm3 mol-1 s-1 at pH approximately 7.4. Estimates of the reduction potentials of the drug-radical couples in water at pH 7 for two of the mono-N-oxides were in the range-0.7 to 0.8 V vs NHE at pH 7.

Scavenging of nitrogen dioxide, thiyl, and sulfonyl free radicals by the nutritional antioxidant beta-carotene

Mechanisms of free radical scavenging by the nutritional antioxidant beta-carotene have been investigated by pulse radiolysis. Free radicals, which can initiate the chain of lipid peroxidation, including nitrogen dioxide (NO2.), thiyl (RS.), and sulfonyl (RSO2.) radicals, are rapidly scavenged by beta-carotene. Absolute rate constant k[NO2. + beta-carotene] = (1.1 +/- 0.1) x 10(8) m-1 s-1 and for the glutathione thiyl radical k[GS. + beta-carotene] = (2.2 +/- 0.1) x 10(8) m-1 s-1 have been determined. The mechanisms however are mutually exclusive, the former involving electron transfer to generate the radical-cation [ beta-carotene]+. and the latter by radical-addition to generate an adduct-radical [RS... beta-carotene].. Rate constants for thiyl radical-addition reactions vary from 10(6) to 10(9) m-1 s-1 and correlate with the lipophilicity of the thiyl radical under study. Sulfonyl radicals undergo both electron abstraction, [ beta-carotene]+. and radical-addition, [RSO2... beta-carotene]. in an approximate 3:1 ratio. The beta-carotene radical-cation and adduct-radicals are highly resonance stabilized and undergo slow bimolecular decay to non-radical products. These carotenoid-derived radicals react differently with oxygen, a factor which is expected to influence the antioxidant activity of beta-carotene within tissues of varying oxygen tension in vivo.

Efficacy of vancomycin and teicoplanin alone and in combination with streptomycin in experimental, low-level vancomycin-resistant, VanB-type Enterococcus faecalis endocarditis

The efficacy of vancomycin (VM) and teicoplanin (TE), alone and in combination with streptomycin (SM), against enterococci that express low-level VanB-type VM resistance was investigated in experimental endocarditis using isogenic strains of Enterococcus faecalis susceptible to glycopeptides and aminoglycosides or inducibly resistant to low levels of VM (MIC = 16 micrograms/ml). VM was significantly less active against the resistant strain than against the susceptible strain, establishing that low-level VanB-type VM resistance can influence therapeutic efficacy. By contrast, TE had equally good activity against both strains. VM or TE combined with SM was synergistic and bactericidal against the resistant strain in vitro. While both combinations were efficient in reducing bacterial density in vivo, TE plus SM was significantly superior to VM plus SM if valve sterilization was considered. These data suggest that despite the presence of low-level VanB-type resistance, combination therapy with a glycopeptide and SM (and presumably other aminoglycosides to which there is not high-level resistance) will nevertheless provide effective bactericidal activity.

Inactivation of xanthine oxidase by oxidative radical attack

Steady-state radiolysis, pulse radiolysis and epr studies, combined with enzyme activity measurements, were carried out on the mechanism by which radical attack, through one-electron oxidation, inactivates xanthine oxidase. Electron transfer to both the N3 and Br2- radical species was used to initiate oxidative damage on the enzyme. Inactivation was found to occur to a greater extent at low than at high pH and is associated with the initial formation of a tryptophanyl radical which converts by a known intramolecular pathway to a tyrosyl radical with a rate constant of 5 x 10(3) S-1. The tyrosyl radical in turn slowly loses around half of its absorbance at an intramolecular rate constant of 350S-1 and is consistent with the establishment of a radical equilibrium with cysteine residue(s). The sequence of reactions could be repeated several times on the same irradiated sample implying that restitution of the implied cysteinyl radical occurs leading to other damage in the protein. N3+Trp/N-->Trp/N-->Tyr/O<-->Cys/S-->?. Epr evidence implies that inactivation of the enzyme from the above sequence of reactions arises in part from alternations to Fe/S center I in the enzyme.

Radicals from one-electron reduction of nitro compounds, aromatic N-oxides and quinones: the kinetic basis for hypoxia-selective, bioreductive drugs

Drugs based on nitroarene, aromatic N-oxide or quinone structures are frequently reduced by cellular reductases to toxic products. Reduction often involves free radicals as intermediates which react rapidly with oxygen to form superoxide radicals, inhibiting drug reduction. The elevation of cellular oxidative stress accompanying oxygen inhibition of reduction is generally less damaging than drug reduction to toxic products, so the drugs offer selective toxicity to hypoxic cells. Since such cells are resistant to radiotherapy, these bioreductive drugs offer potential in tumour therapy. The basis for the selectivity of action entails kinetic competition involving the contesting reaction pathways. The reduction potential of the drug, radical pKa and nature of radical/radical decay kinetics all influence drug activity and selectivity, including the range of oxygen tensions over which the drug offers selective toxicity. These properties may be quantified using generation of radicals by pulse radiolysis, presenting a physicochemical basis for rational drug design.

Comparison of the bactericidal activities of ofloxacin and ciprofloxacin alone and in combination with ceftazidime and piperacillin against clinical strains of Pseudomonas aeruginosa

On the basis of MIC data, ciprofloxacin exhibits superior activity against Pseudomonas aeruginosa than the other currently available fluoroquinolones do. Despite the antipseudomonal advantage noted for ciprofloxacin monotherapy, it is unknown whether this advantage is maintained when the fluoroquinolones are used in combination with antipseudomonal beta-lactams such as ceftazidime and piperacillin. Twelve healthy volunteers were enrolled in this open-label, randomized, steady-state, six-way cross-over, comparative trial. All subjects received the following regimens: (i) 400 mg of ofloxacin given intravenously (i.v.) every 12 h (q12h), (ii) 400 mg of ciprofloxacin given i.v. q12h, (iii) 400 mg of ofloxacin given i.v. q12h plus 1 g of ceftazidime given i.v. every 8 h (q8h), (iv) 400 mg of ciprofloxacin given i.v. q12h plus 1 g of ceftazidime given i.v. q8h, (v) 400 mg of ofloxacin given i.v. q12h plus 4 g of piperacillin given i.v. q8h, and (vi) 400 mg of ciprofloxacin given i.v. q12h plus 4 g of piperacillin given i.v. q8h. Serum bactericidal titers with subsequent calculation of the area under the bactericidal curve were determined against three clinical isolates of P. aeruginosa. As monotherapy, ciprofloxacin demonstrated superior antipseudomonal activity than ofloxacin did; however, combination of these agents with ceftazidime yielded remarkably similar and statistically comparable activity profiles. In contrast, ciprofloxacin-piperacillin retained a bactericidal advantage over ofloxacin-piperacillin. Although ciprofloxacin exhibits superior antipseudomonal activity when used as monotherapy, combination of ofloxacin or ciprofloxacin with ceftazidime yielded equivalent activity profiles against susceptible strains of P. aeruginosa.

Clinical pharmacokinetics of newer cephalosporins

Several new cephalosporins have been developed in recent years. These agents include several oral and parenteral agents with extended activity against Gram-negative pathogens. The pharmacokinetic literature for these agents is quite extensive; therefore, we have summarised this information and presented it in tabular form for critical comparison. With a few exceptions, the newer cephalosporins share similar pharmacokinetic properties. Cefixime, cefetamet pivoxil and ceftibuten differ from the others in that they exhibit nonlinear pharmacokinetic properties. The nonlinear nature of these agents is reflected by decreasing maximal concentrations with escalating doses of cefixime and cefetamet pivoxil, decreasing area under the serum concentration-time curve with increasing doses for cefixime, and a reduced bioavailability with large doses of ceftibuten. Attention to such characteristics aid the clinician in selecting appropriate dosage regimens that will optimise drug absorption. The majority of agents are primarily renally eliminated; however, renal elimination accounts for only 20% of cefixime elimination. The pharmacokinetic parameters noted for the newer cephalosporins are not influenced by multiple-dose administration, suggesting lack of drug accumulation over time. The pharmacodynamics of antimicrobials should be considered when extrapolating pharmacokinetic information into the clinical arena. In the case of the beta-lactams, the time which drug concentrations remain above some critical threshold, such as the minimal inhibitory concentration, appears to have the greatest influence on bactericidal activity. Therefore, it is important to select dosage regimens that will optimise the time serum concentrations remain above this threshold. We present an evaluation of these agents with respect to their activity against a variety of pathogens in an effort to demonstrate a pharmacokinetically-based process of antimicrobial selection.

Pharmacokinetics of cefotaxime in healthy volunteers and patients

Cefotaxime is a third-generation cephalosporin that has maintained good susceptibility pattern despite its extensive use. It is available for intravenous and intramuscular administration. Its pharmacokinetic property includes a small volume of distribution with low protein binding. Cefotaxime's half-life is approximately 1.1 h, and it is primarily eliminated by the kidney. It has an active metabolite desacetyl-cefotaxime that displays pharmacokinetic properties similar to cefotaxime. Desacetyl-cefotaxime has a half-life of 1.5 h and also is eliminated by the kidneys by both glomerular filtration and active secretion. The half-life of cefotaxime and its metabolite is altered in patients with severe renal dysfunction requiring dosage adjustment. Despite its relatively short half-life, cefotaxime may be dosed every 12 h based on its pharmacokinetic and pharmacodynamic properties.

Cephalosporin-metronidazole combinations in the management of intra-abdominal infections

With our current understanding of antimicrobial pharmacokinetics and pharmacodynamics, optimal antimicrobial dosing strategies can be developed for a variety infectious processes. Herein, we discuss the clinical utility of a combination containing a third-generation cephalosporin plus metronidazole as compared to conventional single agents (cefoxitin and ampicillin-sulbactam) for the management intra-abdominal infections. At present, several studies have been performed that compare the bactericidal activity of such combinations to that of single agents for organisms commonly isolated from these intra-abdominal process. From these studies it appears that the use of a third-generation cephalosporin with strong activity against common aerobic organisms associated with intra-abdominal infections in combination with a potent anaerobic drug such as metronidazole provides improved antibacterial activity and optimizes the pharmacodynamic profile of the agents over the dosing interval compared to conventional single agents. As a result of the pharmacokinetic and pharmaco-dynamic superiority of the combination regimen, considerable pharmacoeconomic advantages may be realized with the clinical implementation of a third-generation cephalosporin plus metronidazole regimen. This approach should result in maximal clinical efficacy and is important not only for individual patient therapy, but also for formulary management decisions.

Acute promyelocytic leukaemia and Hodgkin's disease in the same patient

Acute non-lymphocytic leukaemia is a well recognised and increasingly reported complication of treated Hodgkin's disease. The prognosis is generally poor with a disappointing response to chemotherapy. A patient in whom myelodysplastic features appeared after treatment for Hodgkin's disease, to be followed shortly afterwards by acute promyelocytic leukaemia, is reported. Complete remission was achieved and sustained until Hodgkin's disease reappeared three years later when the patient was autografted with a marrow harvested four years earlier. The patient remains in good health with platelet support at the time of writing.

Free hydroxyl radicals are formed on reaction between the neutrophil-derived species superoxide anion and hypochlorous acid

Superoxide anion reacts with hypochlorous acid to yield free hydroxyl radicals, as shown by the hydroxylation of benzoate. This reaction is analogous to the Haber-Weiss reaction but in the absence of metal ions is at least six orders of magnitude faster.

Oxidative metabolism of amsacrine by the neutrophil enzyme myeloperoxidase

Oxidative metabolism of the anti-cancer drug amsacrine 4'-(9-acridinylamino) methane-sulphan-m-anisidide has been suggested to account for its cytotoxicity. However, enzymes capable of oxidizing it in non-hepatic tissue have yet to be identified. A potential candidate, that may be relevant to the metabolism of amsacrine in blood and its action in myeloid leukaemias and myelosuppression, is the haem enzyme myeloperoxidase. We have found that the purified human enzyme oxidizes amsacrine to its quinone diimine, either directly or through the production of hypochlorous acid. In comparison, the 4-methyl-5-methylcarboxamide derivative of amsacrine, CI-921 9-[[2-methoxy-4[(methylsulphonyl)-amino]phenyl]amino)-N, 5-dimethyl-4-acridine carboxamide, reacted poorly with myeloperoxidase, although it was oxidized by hypochlorous acid. Detailed studies of the mechanism by which myeloperoxidase oxidizes amsacrine revealed that the semiquinone imine free radical is a likely intermediate in this reaction. Oxidation of amsacrine analogues indicated that factors other than their reduction potential determine how readily they are metabolized by myeloperoxidase. Both amsacrine and CI-921 inhibited production of hypochlorous acid by myeloperoxidase. CI-921 acted by trapping the enzyme as the inactive redox intermediate compound II. Amsacrine inhibited by a different mechanism that may involve conversion of myeloperoxidase to compound III, which is also unable to oxidize Cl-. The susceptibility of amsacrine to oxidation by myeloperoxidase indicates that this reaction may contribute to the cytotoxicity of amsacrine toward neutrophils, monocytes and their precursors.

Absorption spectra of radical forms of 2,4-dihydroxybenzoic acid, a substrate for p-hydroxybenzoate hydroxylase

Combined optical and conductimetric measurements in aqueous solution indicate that at high pH (greater than or equal to 10).OH radicals react with the phenoxide form of 2,4-dihydroxybenzoic acid to form transiently phenoxyl radicals and a small amount of hydroxyeyclohexadienyl (HCHD) radicals by 150 ns. The respective yields of 88 and 12% of the total.OH radical yield were deduced from conductance and optical changes as well as from studies using a low potential reductant. The HCHD radical possesses a pKa of 8.0 +/- 0.1 and the constructed spectrum of the deprotonated forms of HCHD has a lambda max at 420 nm with a minimum extinction coefficient of approximately 7250 M-1 cm-1. The red shift in lambda max and increase in extinction coefficient compared to the revised spectral properties of the protonated form of the HCHD radical (lambda max at 390 nm with extinction coefficient of approximately 4500 M-1 cm-1), together with the pKa of the HCHD radical, provide an explanation for the pH-dependent spectral changes of the so-called highly absorbing intermediate II species, observed in the functioning of the enzyme p-hydroxybenzoate hydroxylase. These results add further to the evidence in support of the proposal that intermediate II is composed of species which absorb similarly to the flavin 4(a)-hydroxide and a form of the substrate/product such as the HCHD radical (Anderson, R. F., Patel, K. B., and Stratford, M. R. L. (1987) J. Biol. Chem. 262, 17475-17479).

Absorption spectra of the hydroxycyclohexadienyl radicals of substrates for phenol hydroxylase

The absorption spectra of the hydroxycyclohexadienyl radicals formed upon the addition of OH radicals to six substrates for phenol hydroxylase have been determined using pulse radiolysis. Combining the radical spectra of thiophenol (lambda max, 390 nm; epsilon, 10,500 M-1 cm-1) and resorcinol (lambda max, 340 nm; epsilon, 4,100 M-1 cm-1) with their respective published spectra of enzyme-bound reduced flavin that is substituted in the C(4a) position of the dihydroflavin ring gave composite spectra that closely match the spectra formed concomitantly with the introduction of an oxygen atom into the substrates, the so-called Intermediate II species. A similar procedure for the substrates hydroquinone, 3-aminophenol, 3-chlorophenol, and 3-methylphenol yielded spectra that are also consistent with the known characteristics of their Intermediate II species. These spectral results give further support to the proposed biradical mechanism (Anderson, R.F., Patel, K. B., and Stratford, M. R. L. (1987) J. Biol. Chem. 262, 17475-17479) for the functioning of this class of flavoprotein hydroxylases.

Effect of iron deficiency on renal function

Renal function was assessed by determining three hour creatinine clearance (THCC) values in 20 patients (13 males, seven females; age 16-55 years) of nutritional iron deficiency anaemia. Mean transferrin saturation was 4.6% (SD 2.3). Haemoglobin and THCC were determined twice at the interval of three days before therapy. All patients received total dose iron-dextran intravenously. Three days after therapy, haemoglobin and THCC were determined again. Paired 't' test was used to determine the significance of the difference. There was no significant difference between the two pretherapy mean haemoglobin values (6.1 +/- 3.5 g/dl and 5.6 +/- 4.5 g/dl; p greater than 0.2), and the two pretherapy mean THCC values (67.2 +/- 36.9 ml/min and 70.3 +/- 22.8 ml/min; p greater than 0.5). There was no significant difference (p greater than 0.5) between pre-and post-therapy mean haemoglobin levels (6.6 +/- 2.2 g/dl). The difference between the pre-therapy and post-therapy THCC (95.3 +/- 34.0 ml/min) was statistically significant (p less than 0.01). It is concluded on the basis of these results that renal function as measured by THCC is impaired in iron deficiency anaemia, and it improves significantly within three days of total dose intravenous iron-dextran therapy when there is no significant increase in haemoglobin value. This is likely to represent the effect of iron at the tissue level independent of the anaemia.

Conversion of xanthine dehydrogenase to xanthine oxidase as a possible marker for hypoxia in tumours and normal tissues

The enzyme activities of endogenous xanthine dehydrogenase (XDH) and xanthine oxidase (XO) have been measured in 10 different types of mouse tumour and seven normal tissues. The conversion of XDH to XO has been observed in two tumour types upon the prolonged clamping off of the blood supply to the tumours. It is proposed that a similar conversion might also occur naturally in chronically hypoxic cells and that the ratio of the XO activity to the combined XO + XDH activities (%XO activity) could well serve as a marker for tissue hypoxia. A qualitative relationship exists between the %XO activity and literature values of the hypoxic fraction for some tumours measured by radiobiological assays. The influence of tumour size (about 0.2-1.8 g) on %XO activity is presented for all 10 tumours as well as %XO activity determinations for four of the normal tissues.

Absorption spectra of radicals of substrates for p-hydroxybenzoate hydroxylase following electrophilic attack of the .OH radical in the 3 position

The spectra of radicals formed upon the addition of .OH radicals to the 3 position of substrates for p-hydroxybenzoate hydroxylase (4-hydroxybenzoic acid, 2,4-dihydroxybenzoic acid, and 4-aminobenzoic acid) have been determined using pulse radiolysis combined with the results from high performance liquid chromatography measurements. The 3-hydroxy radical forms of the substrates absorb maximally in the 365-410 nm region with extinction coefficients in the range 3700-5250 M-1 cm-1. Upon combining these radical spectra with the known spectrum of enzyme-bound reduced flavin that is substituted in the C(4a) position of the isoalloxazine ring, spectra are found which closely resemble the species long thought to be formed concomitantly with the introduction of an oxygen atom into substrates. On the basis of these spectral results a new radical mechanism is proposed for the functioning of this class of flavoprotein hydroxylases.

Changes in the survival curve shape of E. coli cells following irradiation in the presence of uncouplers of oxidative phosphorylation

Four uncouplers of oxidative phosphorylation (UOP) (carbonyl cyanide m-chlorophenylhydrazone, 2,4-dinitrophenol, 4-hydroxybenzylidenemalonitrile and N-phenylanthranilic acid) have been found to alter the shape of the radiation survival curves of several cell lines of E. coli when present during irradiation in oxia. Incubation of cells with high concentrations of UOP for 30 min before irradiation induced an increase in extrapolation number (n) in cell lines AB 1157 (wild-type), AB 1886(uvrA-) and KMBL(polA-) but not GR 501(lig-)ts, AB 2463(recA-) and AB 2480(uvrA-recA-). In addition the UOP all effect a decrease in mean lethal dose (D0) even when tested at low concentrations or short contact times. Studies with wild-type cells correlate the increase in n with measured increased levels of ATP (above oxic control cells) produced upon incubation with UOP. The increased levels of ATP most likely arise from the UOP overstimulating glycolysis. The decrease in D0 cannot be associated with any of the repair pathways investigated and it is concluded that the highly lipophilic UOP directly or indirectly potentiate other target(s) to radiation damage as well as DNA under oxic conditions. Treatment of the cells with UOP did not result in the deleterious depletion of energy substrates, loss of non-protein thiols or the production of cytotoxins upon irradiation.

The influence of thiols on the pre-irradiation incubation effect of nitroimidazoles in E. coli cells

The increase in the degree of radiosensitization of Escherichia coli cells following prolonged pre-irradiation incubation with nitroimidazoles is not correlated with the loss of intracellular non-protein thiols (NPSH) alone. The rates of reduction of the nitro compounds and the NPSH removal do not show strong dependencies on the lipophilicities of the nitroimidazoles whereas the highly lipophilic compound RGW-609 effects an increase in radiosensitization in a much shorter incubation time than the other nitroimidazoles. Exogenous dithiothreitol (DTT) increased the rate of reduction of misonidazole in the cells but did not alter the fraction converted to the amine. Added DTT (0.15 mmol dm-3) completely protected against the pre-irradiation incubation effect of misonidazole (2.5 mmol dm-3) when added at the start of the incubation but only partially protected when added before irradiation. It is suggested that NPSH can intercept metabolite(s) (or their precursors) of nitroimidazoles which can potentiate cell killing by radiation.