Synthesis of 1-aldofosfamide-perhydrothiazines.

Aldofosfamide-perhydrothiazine derivatives are a new class of prodrugs which spontaneously, with half-life times of 2 to > 12 h hydrolyse to the corresponding aldophosphamide in aquous solution. Synthesis of 1-aldofosfamide-perhydrothiazine (N,N'-(2-chloroethyl)-phosphorodiamide-2-(2'-[4'-carboxy-1',3'- perhydrothiazinyl])-ethylester) and a derivative, in which one 2-chlorethyl group of the alkylating function is substituted by a mesyl-ethyl-group (N-(2-Chloroethyl)-N'-(methanesulphonylethyl)- phosphorodiamide-2-(2'-[4'-carboxy-1',3'-perhydro-thiazinyl] )-ethylester), is described.

Structure/activity studies with thiazolidinyl- and perhydrothiazinylphosphamide ester.

Structure/activity studies were carried out with thiazolidinyl- and perhydrothiazinylphosphamide ester, which differ in the structure of the phosphamide moiety and in the rate of spontaneous hydrolysis to activated oxazaphosphorines. Antitumour activity in mice with advanced P388 tumours growing subcutaneously was increased 30- to 40-fold when one 2-chloroethyl group in l-aldophosphamide-perhydrothiazine was substituted by a mesylethyl group.

Thiazolidinyl- and perhydrothiazinylphosphamidesters: toxicity and preliminary antitumour evaluation.

Aldophosphamide thiazolidine (NSC 613060) and aldophosphamide perhydrothiazine (NSC 612567), which hydrolyse spontaneously to 4-hydroxycyclophosphamide (4-OH-CP) in aqueous solution, were synthesised. These substances are prototypes of a new class of prodrugs for activated oxazaphosphorines. They were developed according to our hypothesis on the mechanism of action of oxazaphosphorine cytostatics. According to this hypothesis, toxicity and canceroselectivity are the results of phosphoramide mustard (PAM) release from 4-OH-CP catalysed by two classes of phosphodiesterase. 4-OH-CP toxicity results (a) from oxazaphosphorine-specific toxicity due to reactivity of the hemiaminal group with thiol groups of membrane proteins and (b) from PAM release catalysed by ubiquitous phosphodiesterases present in blood and tissues. Specific cytotoxicity suitable for antitumour therapy is based on specific PAM release in the vicinity of the target molecule DNA by the exonuclease subsites of DNA polymerases delta and epsilon. To unfold this specific core, which, we assume, improves efficacy in cancer treatment, low, long-lasting concentrations of OH-CP have to be guaranteed beneath the affinity range of the ubiquitous phosphodiesterase. This goal is facilitated by the rapid transfer of 4-OH-CP released from the perhyrothiazine derivative NSC 612567 to protein SH groups, as shown by protein-binding studies. Half-lives of hydrolysis and dissociation constants of the thiazolidine and perhydrothiazine derivatives, in which the reactivity of the hemiaminal group is inactivated by inclusion into the thiazolidine or perhydrothiazine ring, were determined to be 23 h and 6.0 x 10(-6) mol/l for NSC 613060 and 1.5 h and 1.1 x 10(-4) mol/l for NSC 312567. Accordingly the compounds guarantee low but long-lasting steady-state concentrations of 4-OH-CP. The acute toxicity determined in mice was 2400 mg/kg for NSC 613060 and 1900 mg/kg for NSC 612567. Except for a 30% decrease in leucocytes, daily i.p. injections of 260 mg/kg NSC 612567 (15% of LD50) were tolerated without signs of toxicity over a period of 4 weeks. In contrast, equitoxic doses of cyclophosphamide caused severe signs of toxicity, only five daily applications were tolerated. In mice treated repeatedly with NSC 613060, oxazaphosphorine toxicity was overlapped by thiazolidine toxicity. Scheduled activity tests in mice bearing P815 ascites tumour showed optimal therapeutic response when mice were treated daily. Repeated applications of 4% LD50 of NSC 613060 and 13% LD50 of NSC 612567 prevented tumour growth in mice with advanced, P388 lymphomas, implanted subcutaneously, without signs of overall toxicity to the host.

Low toxicity cancer chemotherapy by suicide inactivation of DNA polymerase alpha holoenzyme: first results with new thiazolidinyl- and perhydrothiazinyl-ethyl-N-mustard-phosphamide esters.

Thiazolidinyl- and perhydrothiazinyl-ethyl-N-mustard-phosphamide esters were designed to act as highly specific suicide inactivators of DNA polymerase alpha holoenzymes. Acute and subacute toxicity of these drugs in mice was very small. By daily i.p. injection, on day 0-4 mice were cured of P 388 lymphatic leukaemia with no depression of blood leucocytes. The findings suggest that suicide inactivators of DNA polymerase alpha holoenzyme may be promising drugs for low toxicity cancer chemotherapy.

The enzymatic basis of cyclophosphamide specificity.

Metabolic activation of cyclophosphamide (CP) by microsomal mixed function hydroxylases yields 4-hydroxycyclophosphamide and aldophosphamide defined as activated CP. Activated CP shows relatively high cancerotoxic selectivity in vivo and cytotoxic specificity in vitro and can be trapped rapidly by reversible reaction of hemiaminal group of the oxazaphosphorine ring with protein thiols to form protein bound activated CP (protein-S-CP). Protein-S-CP stores activated CP in a highly stable form. From pharmacokinetics of activated CP in mice after the injection of cyclophosphamide, it was calculated that about 17% of the CP dose given was stored in a pool of protein bound activated CP lasting for several days. From therapy studies with 4-(S-ethanol)-sulfido-CP in combination with excess of cysteine, it was concluded that the protein-S-CP pool may be that form of activated CP which is mainly responsible for the specific cytotoxic effects in the tumor cells. On the other hand free unbound 4-OH-CP was shown to contribute mainly to overall toxicity. No spontaneous toxicogenation of activated CP was noted under in vivo conditions. 3'-5' Exonucleases were found to hydrolyze 4-OH-CP, yielding phosphoramide mustard and acrolein as split products. Because of the low affinity of 4-OH-CP for plain 3'-5' exonucleases, it seems however unlikely that these enzymes play a major role in the antitumor effect of CP in vivo. 3'-5' Exonucleases associated to DNA polymerase like in DNA polymerase delta from rabbit bone marrow or in DNA polymerase I from E. coli are more likely candidates for 4-OH-CP toxicogenation because of the much higher specific activity with 4-OH-CP as substrate. In experiments with DNA polymerase I from E. coli, 4-OH-CP was shown to inhibit DNA polymerase activity after toxicogenation by the 3'-5' exonuclease subsite of the enzyme. This suggests an enzyme mechanism based suicide inactivation of the DNA polymerase. Because of the close spatial cooperation of the DNA polymerase and 3'-5' exonuclease subsites with primer/template a site-specific alkylation of DNA is also postulated. Thus we raised the hypothesis that cytotoxic specificity of activated CP is based on the interaction of protein-S-CP (protein bound activated CP) with DNA polymerase/3'-5' exonuclease as the target. In a P 815 mouse mast-cell tumor we determined by means of 5' AMP agarose affinity chromatography two/third of total DNA polymerase to be associated with 3'-5' exonuclease.

Pharmacokinetics of "activated" cyclophosphamide and therapeutic efficacies.

Estimation of "activated" cyclophosphamide (4-OH-CP) in blood of cancer patients and laboratory animals has revealed significant differences between pharmacokinetics of cyclophosphamide (CP) in man and laboratory animals after CP treatment. Whereas in blood of mice and rats relatively high concentrations of 4-OH-CP were found to exist for a relatively short time, in blood of humans only low, but longer-lasting, blood levels were detected after administration of comparable CP doses. In order to examine whether these different pharmacokinetic behaviors might account at least in part for the known differences of antitumor activity and toxicity of CP between humans and laboratory animals, the authors studied the influence of pharmacokinetics of activated CP on therapeutic efficacy and toxicity after injection of 4-(S-ethanol)-sulfido-cyclophosphamide (P1), a pro drug of activated CP, into nude mice bearing heterotransplanted human bladder sarcoma. With P1, which hydrolyzes quickly in blood to yield 4-OH-CP, different blood level shapes of 4-OH-CP could be established either by single bolus injection of P1 or by repetitive injection of a loading dose followed by several maintenance doses which caused nearly constant levels of activated CP for a longer time period. With these models it was found that 4-OH-CP showed more therapeutic efficacy when present in blood at relatively low levels for longer times than after bolus injection of the same dose resulting in a sharp peak level of activated CP. So after single intraperitoneal (IP) injection of 300 mg/kg P1 which caused a bioavailability of 36 mumol/ml-1/minute a 67% inhibition of tumor growth was achieved, whereas a tumor growth reduction of 83% was obtained after injection of the same dose in 6 fractions resulting in constant blood levels with a bioavailability of only 17 mumol/ml-1/minute. In contrast to the significant influence on antitumor efficacy of activated CP, practically no effect of pharmacokinetics on toxicity of 4-OH-CP could be observed. Therefore, the bioavailability of activated CP, which killed 50% of the animals, was determined to be approximately 89 mumol/ml-1/minute after adjustment of pharmacokinetics to yield constant levels and approximately 79 mumol/ml-1/minute after single bolus injection. The experiments presented show that by adjustment of pharmacokinetics the therapeutic index of P1, defined as bioavailability causing 50% of animals to die, referred to bioavailability causing 90% tumor growth inhibition, could be more than doubled.

The influence of the protector thiol L-cystein on the toxic and therapeutic responses of stabilized "activated" cyclophosphamide (4-(S-ethanol)-sulfido-cyclophosphamide).

The influence of L-cystein on the toxic and therapeutic responses of 4-(S-ethanol)-sulfido-cyclophosphamide (P1), a stabilized "activated" cyclophosphamide, was investigated. Stabilized "activated" cyclophosphamides hydrolyze under physiological conditions to 4-hydroxycyclophosphamide (4-OH-CP). The antitumor activity of P1 was investigated on a heterotransplanted human bladder sarcoma in nude mice and in perfusion experiments carried out on the isolated tumor bearing limb in rats. Due to its rapid hydrolysis to 4-OH-CP, P1 exhibits severe local toxicity which is decreased by the protector thiol L-cystein. Simultaneous application of double molar amounts of L-cystein reduces toxicity in nude mice to approximately one-third. Therapeutic activity is not affected by this ratio of L-cystein so that the protector thiol increases the therapeutic efficacy of P1. Higher amounts of L-cystein reduce both the acute toxicity in nude mice and the therapeutic efficacy against the human xenograft. The perfusion experiments demonstrate that a P1 concentration necessary to cure rats with tumor bearing limb is only tolerated in combination with L-cystein.

Activated cyclophosphamide: an enzyme-mechanism-based suicide inactivator of DNA polymerase/3'-5' exonuclease.

DNA polymerase I from E. coli can toxify activated cyclophosphamide (CP) by means of the 3'-5' exonuclease activity associated with the enzyme. Acrolein and an alkylating moiety are released in the process. Preincubation of DNA polymerase I with activated CP for 15-60 min leads to an increasing inhibition of DNA polymerase activity, which can be prevented when preincubation of DNA polymerase I with activated CP is carried out in the presence of 5' AMP, a competitive inhibitor of the 3'-5' exonuclease subsite of the enzyme. This demonstrates that toxification of activated CP by the 3'-5' exonuclease subsite of DNA polymerase is a prerequisite for the inhibition of DNA polymerase activity. The kinetics and the degree of DNA polymerase inhibition suggest that the alkylating moiety rather than acrolein released from activated CP during toxification is responsible for the inhibition of DNA polymerase. DNA polymerase with associated 3'-5' exonuclease activity has also been isolated from eukaryotic cells, and toxification of activated CP by such an enzyme (DNA polymerase delta from rabbit bone marrow) has been shown previously. Thus we suggest that toxification of activated CP by DNA polymerases/3'-5' exonucleases present mainly in proliferating cells might lead to the specific alkylation of macromolecules involved in the cell proliferation process, such as the DNA polymerase subsite of these enzymes and probably also the DNA bound to the enzymes. The relatively high cancerotoxic selectivity and cytotoxic specificity of activated CP could be based on this specific enzyme-mediated alkylation.

[Intracavitary chemotherapy of S 180 ascites sarcoma in mice with 4-(S-ethanol)-sulfido-cyclophosphamide in combination with protector thiols]. / Intrakavitäre Chemotherapie des S 180-Ascites-Sarkoms der Maus mit 4-(S-Ethanol)-sulfido-cyclophosphamid in Kombination mit Protektorthiolen.

The experimental and pharmacokinetic basis for the local chemotherapy of body cavities with 4-(S-ethanol)-sulfido-cyclophosphamide (P1), a stable derivative of activated cyclophosphamide (CP), was evaluated on the S 180 ascites sarcoma in mice. The severe local toxicity of P1 observed after intraperitoneal administration was markedly reduced by increasing the injection volume (belly bath) without significant loss of cytotoxic activity on the S 180 tumor. Simultaneous application of L-cysteine as a "protector thiol" resulted in further reduction of toxicity without significantly decreasing the cytotoxic effect on tumor cells. Thus the therapeutic index was increased (2.5 fold) by the combination of belly bath and protection by L-cysteine, contrary to 2-mercaptoethane sulfonic acid sodium salt (Mesna) as protector thiol which reduced both the acute toxicity and the curative effectiveness of P1. Pharmacokinetic parameters were determined by measuring the concentrations of P1 and 4-hydroxycyclophosphamide (4-OH-CP), carboxyphosphamide and 4-ketocyclophosphamide in blood and peritoneal fluid. As a result of these measurements the reduction of toxicity of P1 after high volume i.p. administration is due to increased enzymatic detoxification of 4-OH-CP to 4-ketocyclophosphamide and carboxyphosphamide. The effect of L-cysteine on the toxicity of P1 is mainly the consequence of transmercaptalisation of P1 to 4-(S-cysteine)-sulfido-CP. By this reaction formation of the toxic 4-OH-CP in the peritoneal cavity is diminished, and the peritoneal clearance of "activated" CP reduced.

Enzymatic toxicogenation of "activated" cyclophosphamide by 3'-5' exonucleases.

3'-5' Exonucleases from various sources were found to toxicogenate 4-hydroxycyclophosphamide ("activated" cyclophosphamide) by splitting the oxazaphosphorinane ring and releasing an alkylating moiety and acrolein. Neither cyclophosphamide (CP) nor the deactivated metabolites of CP, 4-keto-CP and carboxyphosphamide nor 4-(S-ethanol)-sulfido-CP were attacked by 3'-5' exonucleases. DNA polymerases with proofreading activity, such as DNA polymerase I from E. coli or DNA polymerase delta from rabbit bone marrow, exhibited a tenfold higher specific activity with "activated" CP than "plain" 3'-5' phosphodiesterases such as snake venom phosphodiesterase or 3',5'cyclic AMP phosphodiesterase from bovine heart tissue. High levels of toxicogenating activity were estimated in peripheric human lymphocytes and tissues of lymphatic origin, suggesting that enzymatic toxicogenation plays a key role with respect to the cytotoxic specificity of "activated" CP.

Comparative study on human pharmacokinetics of activated ifosfamide and cyclophosphamide by a modified fluorometric test.

The activated metabolites of ifosfamide and cyclophosphamide (4-hydroxy-ifosfamide and 4-hydroxy-cyclophosphamide) were analysed fluorometrically by condensation of liberated acrolein with m-aminophenol yielding 7-hydroxyguinoline. Interfering fluorescence of blood and urine was eliminated by extraction with dichlormethane and determination of blanks in which the liberated acrolein reacted with hydrazine to non-fluorescent pyrazoline. The modified test proved effective in identifying low levels of activated metabolites in man. After i.v. injection of 20 mg/kg cyclophosphamide or ifosfamide peak levels of activated cyclophosphamide (4.7 nmol/ml) and the area under the curve (c.t = 16.7 nmol.ml/h) showed mean values three times higher than those found for activated ifosfamide. One per cent of the applied dosis of cyclophosphamide vs. 0.3% of ifosfamide was excreted as activated metabolites. Due to the high stability of activated cyclophosphamide and ifosfamide in urine a low liberation rate of acrolein was found, the concentration of which in urine was below 0.5 nmol/ml. Acrolein, which was directly liberated from activated cyclophosphamide or ifosfamide, does not seem to play an important role in the urotoxicity of these cytostatics.

[Effect of damaged liver parenchyma, renal insufficiency and hemodialysis on the pharmacokinetics of cyclophosphamide and its activated metabolites]. / Einfluss von Leberparenchymschäden, Niereninsuffizienz und Hämodialyse auf die Pharmakokinetik von Cyclophosphamid und seinen aktivierten Metaboliten.

Patients with impaired liver function have a reduced biotransformation rate of the cytostatic agent cyclophosphamide. With pathologically reduced serum cholinesterase activity the half-life of the drug increases from normally 4.3 h to 6.7 h. These patients show significantly lower peak levels of activated cyclophosphamide (4-hydroxy-cyclophosphamide + aldophosphamide). Because of the low renal clearance of cyclophosphamide (16 ml/min) and equally low renal excretion of activated cyclophosphamide amounting to only 1% of the applied dose more than 80% of the drug is still metabolized and the area under the curve of activated cyclophosphamide (cXt) remains relatively constant. No change in the pharmacokinetics of cyclophosphamide and its activated metabolite is observed in an anuric patient. However, an accumulation of toxic, directly alkylating metabolites with a fourfold alkylation rate of plasma proteins is found in this case. Hemodialysis sufficiently eliminated the toxic alkylating metabolites without a measurable influence on the pharmacokinetics of activated cyclophosphamide.

Does acrolein contribute to the cytotoxicity of cyclophosphamide?

To determine whether the release of acrolein from oxazaphosphorinane-cytostatics contributes to their cytotoxic action, the effect of 4-hydroperoxycyclophosphamide, 4-hydroperoxy-semi-cyclophosphamide, 4-hydroperoxy-dechloro-cyclophosphamide, and acrolein on murine L 1210 leukemia cells in vitro was compared by measuring the median survival time (MST) after transplantation of the tumor cells in DBA2/Han mice. We found that only 4-hydroperoxycyclophosphamide, which is able to release both acrolein and the alkylating metabolite phosphoramide-mustard, decreased the transplantability of L 1210 cells, while the structurally analogous 4-hydroperoxy-dechloro-cyclophosphamide and 4-hydroperoxy-semi-cyclophosphamide, which under physiological conditions only release acrolein but no alkylating split products showed no cytotoxicity. Acrolein itself showed only a marginal effect, when administered in concentrations equivalent to the release of acrolein from the oxazaphosphorinane-derivatives in test. In this case, however, significant lysis of the L 1210 cells was observed by estimating dye exclusion, while acrolein released intracellularly from 4-hydroperoxy-oxazaphosphorinane-compounds did not. This points to a different mechanism of the cytotoxic action of extracellular acrolein and acrolein released intracellularly from activated oxazaphosphorinane-compounds. The results suggest that the cytotoxic effect of activated cyclophosphamide is based on the alkylating moiety of the molecule. Neither the 4-hydroperoxy-group nor the activated oxazaphosphorinane-ring itself, nor acrolein released intracellularly during toxification of activated cyclophosphamide exert a direct cytotoxic effect. Thus, the release of acrolein from activated CP apparently does not contribute to the cytotoxicity of CP in vivo.

[Blood level and urinary excretion of activated cyclophosphamide and its deactivation products in man (author's transl)]. / Uber Blutspiegel und Urin-Ausscheidung von aktiviertem Cyclophosphamid und seinen Deaktivierungsprodukten beim Menschen.

Blood levels and urinary excretion of cyclophosphamide and its metabolites were determined in cancer patients receiving cyclophosphamide. Activated cyclophosphamide (4-hydroxycyclophosphamide aldophosphamide) was assayed by TLC after derivatisation to stable 4-(S-benzyl)-sulfido-cyclophosphamide. Twenty minutes after injection of 10(20) mg/kg cyclophosphamide mean peak levels of activated cyclophosphamide were found to be 1.4(2.6) nmol/ml. The rate constant for biotransformation (=activation) of cyclophosphamide in man (km = 0.132 h-1) was only 1/50 of the value found in the mouse whereas the elimination rate constant of activated cyclophosphamide (ke[M] approximately 6.78 h-1) was much higher equalling that of laboratory animals. 4-ketocyclophosphamide, carboxyphosphamide, and phosphoramidemustard reached their peak levels between 4 and 6 h after cyclophosphamide injection. Increasing quantities of cyclophosphamide metabolites were bound to plasma proteins reaching a constant level after 24 h lasted for several days. Fifty per cent of those metabolites were reversibly bound to plasma proteins. Within 24 h, the cumulative excretion of cyclophosphamide and its metabolites amounted to 50% of the dose applied. The main metabolites excreted were phosphoramide-mustard and carboxyphosphamide whereas only 2% consisted of activated cyclophosphamide. The significance of the different pharmacokinetics of cyclophosphamide in laboratory animals and man for the therapeutic index is discussed.

[Fluorometric determination of "activated" cyclophosphamide and ifosfamide in blood (author's transl)]. / Fluorometrische Bestimmung von "aktiviertem" Cyclophosphamid und Ifosfamid im Blut.

"Activated" N-(2-Chloroethyl)amido-oxazaphosphorines like 4-hydroxycyclophosphamide, 4-hydroperoxycyclophosphamide, 4-hydroxyifosfamide, and 4-hydroperoxyifosfamide can be determined fluorometrically by condensation of liberated acrolein with m-aminophenol yielding 7-hydroxychinolin. The method permits determination of 10(-10) mol and is specific for "activated" N-(2-Chloroethyl)amido-oxazaphosphorine metabolites which liberate acrolein under conditions of the test. Neither cyclophosphamide nor ifosfamide or other metabolites of this cytostatics interfere with the test. Blood levels of free 4-hydroxycyclophosphamide and 4-hydroxyifosfamide were determined after injection of cyclophosphamide and ifosfamide into mice.

[Permeability of N,N-bis(2-chloroethyl)-diamido-phosphoric-acid into tumor cells (author's transl)]. / Zur Frage der Permeabilität von N,N-Bis(2-chloräthyl)-phosphorsäurediamid in Tumorzellen.

The uptake of tritiated N,N-bis(2-chloroethyl)-diamido-phosphoric-acid into Ehrlich-Ascites-Tumor cells of mice was studied by means of the siliconoil-filtration technique. At 10 mM concentration no permeation of the metabolite into the tumor cells could be found within 5 min at 1 degrees C, while its congenors cyclophosphamide and 4-hydroperoxycyclophosphamide (1 mM) were shown to permeate into the cells very easily reaching saturation values. Thus lack of permeation into tumor cells of N,N-bis(2-chloroethyl)-diamidophosphoric-acid seems to be the reason for the poor cytotoxic activity of this metabolite of cyclophosphamide.

Synthesis and preliminary antitumor evaluation of 4-(SR)-sulfido-cyclophosphamides.

Crystalline 4-(SR)-sulfidocyclophosphamides, sulfido derivatives of activated cyclophosphamide (4-hydroxycyclophosphamide), were synthesized by ozonation of cyclophosphamide and reaction of the intermediate 4-hydroxycyclophosphamide with various thiols (HSR). The products were characterized by elemental analysis, 1H NMR and IR spectroscopy, and mass spectrometry. 1H NMR and polarimetric analysis demonstrated that they consist of racemic cis-isomers that are stable in the crystalline state at room temperature. In aqueous solution these derivatives are hydrolyzed to 4-hydroxycyclophosphamide and the corresponding thiol, with half-lives ranging between 4 and 17 min at 37 degrees C and pH 7. The cytotoxicity of 4-(S-ethyl)- and 4-(S-ethanol)-sulfidocyclophosphamide against Yoshida sarcoma ascites cells and the toxicity in rats were found to be practically identical with those of activated cyclophosphamide. A preliminary evaluation of the curative effect after a single IV injection of 4-(S-ethane)- and 4-(S-ethanol)-sulfidocyclophosphamide in rats bearing Yoshida ascites sarcoma or of 4-(S-ethanol)-sulfidocyclophosphamide in nu/nu mice bearing human breast carcinoma xenografts suggested that these sulfido derivatives possess the same oncostatic efficacy as activated cyclophosphamide itself.

Mass spectrometric characterization of activated N-(2-chloroethyl)amino oxazaphosphorine derivative.

The hydroperoxy and several alkylthio derivatives of the antitumor agents cyclophosphamide (2-bis(2-chloroethyl)amino tetrahydro-2H-1,3,2-oxazaphosphorine 2-oxide), ifosfamide (3-(2-chloroethyl)-2-(2-chloroethylamino)tetrahydro-2H-1,3,1-oxazaphosphorine 2-oxide) and trofosfamide (3-(2-chloroethyl)-2-(bis(2-chloroethyl)amino)tetrahydro-2H-1,3,2-oxazaphosphorine 2-oxide) were characterized by electron impact and field desorption mass spectrometry. The compounds, which are stabilized derivatives of the activated hydroxylated intermediates of cyclophosphamide (ifosfamide, trofosfamide), could be identified as 4-hydroperoxy and 4-alkylthio oxazaphosphorines. The existence of diastereomers of these products was demonstrated by thin-layer chromatography and f.d. mass spectra. Derivatization with benzylmercaptan was found to be an appropriate method for the quantitative isolation and mass spectral identification of the activated metabolic intermediates of cyclophosphamide from biological material. Using this reaction, 4-hydroxycyclophosphamide and its acyclic tautomer, aldophosphamide, which are too unstable for direct identification, were detected in urine and serum of patients treated with 3H-cyclophosphamide.