Apoptosis occurs predominantly in bystander cells and not in productively infected cells of HIV- and SIV-infected lymph nodes.

Although 13 years have passed since identification of human immunodeficiency virus-1 (HIV-1) as the cause of AIDS, we do not yet know how HIV kills its primary target, the T cell that carries the CD4 antigen. We and others have shown an increase in the percentage of apoptotic cells among circulating CD4+ (and CD8+) T cells of HIV-seropositive individuals and an increase in frequency of apoptosis with disease progression. However, it is not known if this apoptosis occurs in infected or uninfected T cells. We show here, using in situ labelling of lymph nodes from HIV-infected children and SIV-infected macaques, that apoptosis occurs predominantly in bystander cells and not in the productively infected cells themselves. These data have implications for pathogenesis and therapy, namely, arguing that rational drug therapy may involve combination agents targeting viral replication in infected cells and apoptosis of uninfected cells.

The specific interaction of helper T cells and antigen-presenting B cells. IV. Membrane and cytoskeletal reorganizations in the bound T cell as a function of antigen dose.

We have used double-immunofluorescence labeling to determine the surface distributions of LFA-1 and CD4, and the intracellular distributions of the cytoskeletal protein talin and of the microtubule organizing center (MTOC) of cloned Th cells in 1:1 cell couples with antigen (Ag)-specific APC of the B cell type (B-APC). The Th cell was directed to a peptide fragment of the Ag OVA in the context of IAd. The B-APC was the transfected A20 B hybridoma cell A20-HL, bearing on its surface a surface Ig specific for the hapten TNP, and pulsed with different concentrations of DNP-OVA. At sufficiently high doses of DNP-OVA (greater than 100 ng/ml), in essentially all couples, LFA-1, CD4, and talin were each concentrated at the Th cell membrane where it was in contact with the B-APC, and the MTOC inside the Th cell was reoriented to face the contact region. At lower doses of DNP-OVA (between 50 and 10 ng/ml), in all couples, LFA-1 and talin were concentrated at the Th/B-APC contact region, but the extent of CD4 clustering, MTOC reorientation, and Th cell proliferation all decreased with decreasing Ag dose. With no Ag, none of these effects was observed. These and other data indicate that two distinct signals are received by the Th cell that is specifically bound to its B-APC. The first signal, at low Ag doses, stimulates a linkage of LFA-1 and talin in the Th cell, and a specific LFA-1-mediated intercellular adhesion; the second signal, at higher Ag doses, is required to induce Th cell proliferation, with which the Th-MTOC reorientation and CD4 clustering are correlated.

The specific direct interaction of helper T cells and antigen-presenting B cells. II. Reorientation of the microtubule organizing center and reorganization of the membrane-associated cytoskeleton inside the bound helper T cells.

We have produced and investigated cell couples formed between cloned Th cells or T hybridoma cells, and either Ag-presenting B hybridoma or B lymphoma cells. The specific direct interaction between a Th and B-APC is here demonstrated by two rearrangements occurring inside the bound Th cell; the MTOC (and presumably the GA) is oriented to face the cell contact region with the B cell, and a membrane-associated cytoskeletal protein, talin, becomes concentrated under the contacting Th membrane. In the absence of the specific Ag or the correct Ia determinant, nonspecific T-B cell couples form that are morphologically indistinguishable from specific cell couples in the light microscope, but neither the MTOC nor the talin rearrangement occurs inside the bound T cell of such nonspecific couples. Furthermore, Ag processing by the B cell is required to produce the MTOC and talin rearrangements within the T cell in specific T-B couples. In the case of allogeneic Th-B cell couples, similar specific MTOC and talin rearrangements are observed inside the Th. Extracellular Ca2+ is required for the MTOC orientation to occur inside the specifically bound Th cell, but not for the talin rearrangement. It is proposed that the MTOC (and GA) reorientation and the talin rearrangement are involved in the directed secretion of GA-derived lymphokines from the Th cell to the bound B cell.

Small splenic B cells that bind to antigen-specific T helper (Th) cells and face the site of cytokine production in the Th cells selectively proliferate: immunofluorescence microscopic studies of Th-B antigen-presenting cell interactions.

Antigen (Ag)-specific T helper (Th) cells regulate the proliferation and differentiation of Ag-specific B cells by secreting cytokines and by expressing activating receptors like gp39. In vitro, the cytokines and the activating receptors function in an Ag-nonspecific manner. It is unclear, therefore, how Ag specificity is imposed on B cell responses in physiological Th-B cell interactions. Here we studied, at the single cell level, the interactions between cloned Th cells and small splenic B cells, which served as Ag-specific antigen-presenting cells (APCs) to the Th cells. Digital confocal immunofluorescence microscopy of Th-B cell conjugates revealed significant variability in the molecular and cellular properties of these interactions, in spite of the fact that all the interactions in this system were expected to be Ag specific. After 30 h of incubation B cells began to divide, and this process was entirely dependent on the presence of both Th cells and Ag. Immunofluorescence microscopic studies showed that essentially all the mitotic B cells were bound to Th cells and faced the microtubule organizing center (MTOC) in the Th cells where interleukin 4 was highly concentrated. Other B cells that were bound to the same Th cells but were not close to the Th-MTOC remained in interphase. These results provide the first direct structural and functional evidence that the site of interaction of B cells with Th cells affects their immune response. We propose that, during Ag-induced Th-B cell interactions, B cells that are bound facing the Th-MTOC proliferate preferentially because they are the recipients of locally secreted cytokines. In addition, these B cells may interact with newly expressed receptors, which may also be locally inserted into the Th membrane. The polarized delivery of activating molecules towards the Th-bound APCs may impose functional specificity on effector molecules that otherwise are not Ag specific.

On the mechanism of unidirectional killing in mixtures of two cytotoxic T lymphocytes. Unidirectional polarization of cytoplasmic organelles and the membrane-associated cytoskeleton in the effector cell.

In mixtures of two CTL of the type a anti-b and b anti-c, only the latter is lysed; i.e., killing is unidirectional. Here, we show that two profound types of changes occur in the effector CTL but not in the target CTL upon formation of couples between them. One is that the microtubule organizing center (and presumably the Golgi apparatus that is invariably colocalized with it) is reoriented inside the effector CTL to face the bound target CTL. This unidirectional reorientation, it is proposed, serves to direct putative cytotoxic secretory components derived from the Golgi apparatus of the effector cell to the site of cell-cell binding. The second unidirectional change is in the membrane-associated cytoskeleton of the effector CTL in the area of target cell binding. The cytoskeletal protein talin, but not any of four other such proteins assayed, is highly concentrated at the contacting membrane of the effector CTL, while it is uniformly distributed over the entire membrane of the bound target CTL. This localized, massive cytoskeletal reorganization may reflect a mechanism to protect the membrane of the effector CTL from the effects of putative cytotoxic components secreted by the effector cell into the intercellular space between it and the target cell.

TCR-induced transmembrane signaling by peptide/MHC class II via associated Ig-alpha/beta dimers.

Previous findings suggest that during cognate T cell-B cell interactions, major histocompatability complex (MHC) class II molecules transduce signals, leading to Src-family kinase activation, Ca2+ mobilization, and proliferation. Here, we show that antigen stimulation of resting B cells induces MHC class II molecules to associate with Immunoglobulin (Ig)-alpha/Ig-beta (CD79a/CD79b) heterodimers, which function as signal transducers upon MHC class II aggregation by the T cell receptor (TCR). The B cell receptor (BCR) and MHC class II/Ig-alpha/Ig-beta are distinct complexes, yet class II-associated Ig-alpha/beta appears to be derived from BCR. Hence, Ig-alpha/beta are used in a sequential fashion for transduction of antigen and cognate T cell help signals.

Stereoselective mephobarbital hydroxylation cosegregates with mephenytoin hydroxylation.

The 8-hour urinary recovery of 4-hydroxy-mephobarbital has been measured after oral administration of racemic mephobarbital (90 mg) in 17 extensive (EM) and six poor (PM) metabolizer phenotypes of mephenytoin. The recovery of this metabolite was measurable in every EM and ranged from 2.5% to 48% (10.9% +/- 1.9% of dose), but was not detected in any PM (less than 1% of dose). In EMs, the 8-hour urine recovery of 4-OH-mephobarbital after mephobarbital was approximately half that of 4-OH-mephenytoin over the same time after mephenytoin administration. One EM received similar doses of R- and S-mephobarbital on separate occasions. Urinary recovery of 4-OH-mephobarbital was 33% and less than 1%, respectively. These results suggest that mephobarbital is stereoselectively hydroxylated by the same drug metabolizing enzyme that is responsible for the stereoselective aromatic hydroxylation of mephenytoin.

Aminopyrine demethylation measured by breath analysis in cirrhosis.

The method of measuring the rate of aminopyrine demethylation by breath analysis was assessed in 23 normal subjects and 20 patients with cirrhosis. Carbon 14 aminopyrine specifically labeled at the two N-methyl groups was administered by mouth in a dose of 9 mg/kg, including a total radioactivity of 2 muCi. The decay of the specific activity of 14CO2 in breath (kb) was found to correlate (r = 0.91) with the disappearance of aminopyrine from plasma (KP). In normal volunteers, kb was 22.4%/hr; in patients with alcoholic and nonalcoholic cirrhosis it was depressed to 8.4%/hr (p less than 0.001). The degree of functional impairment found with the breath test was similar to the sulfobromophthalein (BSP) disappearance curve and the galactose elimination capacity. Although many questions relating to the aminopyrine breath test remain open, our data confirm and extend previous studies of 14CO2 breath analysis after 14C-aminopyrine administration. It is concluded that it represents a simple and noninvasive procedure which quantitatively reflects the microsomal function of the cirrhotic liver.

Polymorphic dextromethorphan metabolism: co-segregation of oxidative O-demethylation with debrisoquin hydroxylation.

Dextromethorphan hydrobromide, 25 mg po, was given to 268 unrelated Swiss subjects to study urinary drug and metabolite profiles. Rates of O-demethylation yielding the main metabolite dextrorphan were expressed by the urinary dextromethorphan/dextrorphan metabolic ratio. We found a bimodal distribution of this parameter in our population study, which indicates that there are two phenotypes for dextromethorphan O-demethylation. The antimode at a metabolic ratio of 0.3 separated the poor metabolizer (PM; n = 23; prevalence of 9%) from extensive metabolizer (EM) phenotypes. Urinary output of dextrorphan was less than 6% of the dose in all PMs and was 50% in the 245 EMs. Pedigree analysis of 14 family studies revealed an autosomal-recessive transmission of deficient dextromethorphan O-demethylation. In these families, 37 heterozygous genotypes could be identified; however, through use of the urinary drug and metabolite analysis it was not possible to identify the heterozygous genotypes within the EM phenotype group. Co-segregation of dextromethorphan O-demethylation with debrisoquin 4-hydroxylation was also studied. Complete concordance of the two phenotypic assignments was obtained, with a Spearman rank correlation coefficient of rs = 0.78 (n = 62; P less than 0.0001) for dextromethorphan and debrisoquin metabolic ratios. Presumably the two drug oxidation polymorphisms are under the same genetic control. Thus the innocuousness and ubiquitous availability of dextromethorphan render it attractive for worldwide pharmacogenetic investigations in man.

Mephenytoin hydroxylation deficiency: kinetics after repeated doses.

Deficient aromatic hydroxylation of S-mephenytoin was observed in an index subject during a kinetic study of stereoselective metabolism of mephenytoin. A genetic basis for this defect was suggested by decreased urinary recovery of 3-methyl-5-(4-hydroxyphenyl)-5-ethylhydantoin (4-OH-M) in the 24 hr after oral racemic mephenytoin in two brothers of the propositus. The parents and a third brother had urinary recoveries of 4-OH-M of the same order as in a group of 20 normal subjects. The kinetic implications of this defect were studied in the index subject and compared with four normal subjects after a single oral dose of differentially radiolabeled pseudoracemic mephenytoin (5 microCi of 14C-S-mephenytoin, 45 microCi of H3-R-mephenytoin, and 11.5 mumol/kg of both S- and R-mephenytoin) followed by single oral doses of 1.4 mmol of unlabeled racemic mephenytoin daily the next 4 days. In normal subjects, there was substrate stereoselective metabolism with the S-enantiomer rapidly excreted as 4-OH-M and the R-enantiomer slowly excreted as 5-phenyl-5-ethylhydantoin (PEH). Stereoselective metabolism persisted during repeated dosing. In the hydroxylation-deficient subject, there was no evidence of stereoselective metabolism, recovery of 4-OH-M was low, and both enantiomers were slowly excreted, predominantly as PEH. Plasma PEH concentrations and urinary PEH excretion rates were approximately twice that in normal subjects. Thus a genetic deficiency in ability to hydroxylate S-mephenytoin results in the S-enantiomer metabolization by the alternate route of demethylation to PEH that cumulates, thereby, in comparison to the normal, effectively doubling the dose of total hydantoin.

Differences in hepatic drug accumulation and enzyme induction after chronic amiodarone feeding of two rat strains: role of the hydroxylator phenotype?

1. It has previously been shown that the extent of hepatic phospholipidosis induced by chronic amiodarone treatment correlates with the degree of drug accumulation in liver tissue. 2. To investigate a possible influence of pharmacogenetic factors, biochemical and morphological investigations were carried out in two rat strains differing in debrisoquine hydroxylation. 3. Plasma and liver tissue concentrations of amiodarone and its main metabolite, desethyl-amiodarone, were significantly higher in rats with deficient hydroxylation. Microsomal enzyme induction, drug cytochrome P-450 complex formation and typical ultrastructural features of phospholipidosis were only seen in rats with deficient hydroxylation and in a more sensitive species, the guinea-pig. 4. It remains to be seen whether deficient debrisoquine hydroxylation in man is associated with an increased susceptibility to amiodarone side effects.

In-vivo and in-vitro dextromethorphan metabolism in SD and DA rat. An animal model of the debrisoquine-type polymorphic oxidation in man.

The female dark Agouti (DA) rat is well established as an animal model for the debrisoquine poor metabolizer phenotype (PM), whereas the SD rat represents the extensive metabolizer (EM). It is not known, however, if the DA rat also is representative for the dextromethorphan (DEM) PM, a compound recently demonstrated to be subjected to the debrisoquine phenotype in man. Studies were performed, therefore, to evaluate in-vivo and in-vitro metabolism of DEM in DA and SD rats. After oral administration of 50 mg/kg of DEM, the DA rat excreted 25 +/- 6% of the dose in 72-hr urine as O-demethylated product (dextrorphan), whereas the SD excreted 40 +/- 9% (P less than 0.002). Metabolic ratio of O-demethylation was 0.46 +/- 0.11 in DA and 0.02 +/- 0.01 in SD (P less than 0.001). As a compensatory mechanism, N-demethylation was ninefold increased in DA compared to SD (8.0 +/- 3% of the dose excreted in urine of DA as methoxymorphinan vs 0.9 +/- 0.7% in SD) (P less than 0.001). Total plasma clearance of DEM was 95 +/- 20 ml/min/kg in SD and 45 +/- 13 ml/min/kg in DA (P less than 0.001). In vitro, microsomal affinity for DEM O-demethylation was greater than 50 times higher in SD than in DA rats (P less than 0.004), whereas Vmax did not differ statistically. Vmax for N-demethylation was 80% increased in DA (P less than 0.01), whereas corresponding Km values did not differ. It appears that the differences in DEM metabolism between DA and SD rats are qualitatively similar to human EM and PM phenotypes, respectively. Whether this is also true for the underlying mechanism(s) however, remains to be resolved.

Defective bile acid transport in an animal model of defective debrisoquine hydroxylation.

Bile acid transport in female DA (dark Aguti) rats, a model for debrisoquine hydroxylation deficiency in man, was investigated. Compared to hydroxylation competent male DA and Sprague-Dawley rats of either sex, the female DA rat had a significantly lower taurocholate maximal secretory rate in vivo. Studies in the perfused liver showed this to be due to a decreased extraction efficiency during exogenous taurocholate loading. To characterize further the defect, taurocholate uptake velocity into isolated hepatocytes was studied. This showed a decreased maximal uptake velocity in the female DA rat (P less than 0.02). Whether this defect in bile acid uptake is related to the defective debrisoquine hydroxylation, remains to be established.

Spectral binding studies of the polymorphically metabolized drugs debrisoquine, sparteine and phenformin by cytochrome P-450 of normal and hydroxylation deficient rat strains.

The mechanisms of polymorphic drug hydroxylation of debrisoquine, sparteine and related drugs in vivo have been investigated using Cyt P-450 preparations of inbred rat strains as an in vitro model of the poor and extensive metabolizer phenotypes found in various rat strains and in man. Optical difference spectroscopy with debrisoquine, sparteine, phenformin and three other drugs (selected test compounds with proven or suspected hydroxylation polymorphisms in man) exhibited Type 1 binding in normal Sprague-Dawley, Fischer and Lewis Cyt P-450, whereas no Type I drug binding was found in the hydroxylation deficient DA rat liver Cyt P-450. Cyt P-450 content and Type II drug binding of metiamide was the same in normal and hydroxylation deficient rat liver microsomes. The pronounced Type I drug binding in extensive hydroxylation Cyt P-450 and the defective Type I binding in DA Cyt P-450 in vitro, therefore, closely parallels the polymorphic hydroxylation pattern of these test drugs found in the four rat strains studied in vivo. Consequently, missing binding properties of Cyt P-450 or of its micro-environment might represent the enzymatic defect underlying the genetically determined hydroxylation deficiency of polymorphically metabolized drugs in the poor metabolizer phenotype in the DA rat and, by inference, in man.

Metabolism and pharmacokinetics of oral and intravenous ifosfamide.

The initial metabolism of ifosfamide (IF) consists of two different pathways: enzymatic hydroxylation at carbon-4 forms the cytostatically active metabolite 4-OH-IF ("activated ifosfamide") whereas side-chain oxidation results in the liberation of chloroacetaldehyde, a compound with possible neurotoxic properties. The pharmacokinetics of ifosfamide and its activated form were investigated in 12 cancer patients, who received both oral and i.v. treatment in a randomized sequence on days 1 and 3 at a dose of 1 g/m2 (n = 7) or 1.5 g/m2 (n = 5). In 3 patients the pharmacokinetics of chloroacetaldehyde were also investigated. After oral application, ifosfamide absorption proceeded rapidly, the oral bioavailability was 0.92. Independent of the route of ifosfamide application on day 1, the terminal half-life on day 3 (when the drug was given by the alternative route) was decreased in 6 out of the 12 patients, thus indicating self-induction of hepatic metabolism. 4-OH-IF was already present 20 min after ifosfamide administration. In the individual patient the concentrations of 4-OH-IF were always higher after oral than after i.v. IF application: the mean p.o.:i.v. ratios for cmax and the area under the concentration/time curve were 2.3 and 1.7 respectively (P less than 0.05). In a first series of 3 patients the chloroacetaldehyde concentrations measured after oral ifosfamide application were about twice as high as those when the drug was given intravenously. These results indicate that (in comparison to the i.v. route) orally administered ifosfamide may be more cancerotoxic but also leads to higher levels of chloroacetaldehyde. This would explain the neurotoxic side-effects previously seen after oral administration of comparatively low ifosfamide doses.

Subcutaneous continuous infusion of ifosfamide and cyclophosphamide in ambulatory cancer patients: bioavailability and feasibility.

The oxazaphosphorines ifosfamide (IFO) and cyclophosphamide (CTX) are standard alkylating agents. Both drugs show an increased therapeutic index when given as a fractionated dosage over several days. Maximal fractionation is achieved by continuous infusion. We have studied the feasibility and bioavailability of a subcutaneously (s.c.) administered isotonic and neutral (pH 7) solution of IFO (10 h up to 5 days infusion) and CTX (12-24 h infusion) in patients with advanced cancer. A portable disposable gas-driven infusor syringe was used for ambulatory patients. Our results show 90%-100% bioavailability of s.c. IFO and CTX. The isotonic solution of IFO and CTX (pH 7) showed no significant local toxicity (one local infection in 51 cycles) during or after s.c. administration of 33 cycles with IFO and 18 with CTX. Haematotoxicity of both drugs was equal after s.c. and i.v. application. For IFO-treated patients no uro- or neurotoxicity was observed. We conclude that this novel continuous s.c. oxazaphosphorine infusion over a prolonged period is a rational, well-tolerated and economic way of delivering this drug on an outpatient basis.

Continuous 5-day infusion of ifosfamide with mesna in inoperable pancreatic cancer patients: a phase II study.

Phase II studies on ifosfamide and mesna in pancreatic cancer have mostly been inconclusive. In all of these studies ifosfamide was administered as an i.v. bolus or by short infusions. Since dose fractionation of ifosfamide over several days increases its therapeutic index, we chose to maximize the dose fractioning by selecting a continuous-infusion schedule (1.75 g/m2 on days 1-5 every 21-28 days, with mesna 60%-100% of the ifosfamide dose up to 12 h after ifosfamide). Since 1987 29 patients (performance status less than or equal to 2) with advanced inoperable adenocarcinoma of the pancreas were studied (8 women and 21 men; median age 58 years: 36-73 years). A total of 25 patients are evaluable for response (1 ineligible; 3 inevaluable: 2 early deaths due to disseminated intravascular coagulation, 1 refusal). One female patient with a complete response on computed tomography scan (after five cycles) but residual liver metastases on surgical exploration survived for 473 days. Three male patients with partial response survived for 205, 335 and 355 days. Six more patients with minor response (3) or no change (3) but significant decrease of tumour marker CA 19-9 had a median survival of 213 days (106-243). Responders seemed to benefit in terms of pain relief and general well-being. The median overall survival of all patients was 148 days (21-473). Haematotoxicity was rarely dose-limiting [median nadirs: white blood cells = 2.1 x 10(9)/l (0.45-6.4), Hb = 10.7 g/dl (7.5-13), platelets = 137 x 10(9)/l (21-411)]. Nausea and vomiting were mild with prophylactic oral metoclopramide. No central nervous system toxicity or urotoxicity was observed. Alopecia was seen in all patients who had received at least two cycles. Continuous infusion of ifosfamide was generally well tolerated and useful for palliation in 10 of 25 patients. A higher dose intensity is recommended.

Increased urinary losses of carnitine during ifosfamide chemotherapy.

Chloroacetaldehyde and thiodiglycolic acid, two metabolites of ifosfamide, interfere with mitochondrial function and may sequester carnitine. Urinary excretion of carnitine was measured in five patients before and during a continuous infusion of ifosfamide over 5 days at a dose of 2.8-3.2 g/m2 per day. The excretion of free and total carnitine increased from 85+/-53 to 2697+/-1393 micromol/day on the 1st day of chemotherapy and then gradually decreased. The average loss of carnitine during a chemotherapy cycle amounted to 8.5 mmol. The formation and excretion of esters of carnitine and metabolites of ifosfamide and/or a decreased renal tubular reabsorption could account for this marked loss, which might lead to symptomatic carnitine deficiency after several chemotherapy cycles.

Depletion of total cysteine, glutathione, and homocysteine in plasma by ifosfamide/mesna therapy.

The sulfhydryl status of cells, particularly the intracellular concentration of glutathione, is a critical determinant of the response of tumor and normal cells to cytostatic drugs. Recent data indicate that the administration of mercaptoethane sulfonate (mesna), which is often combined with ifosfamide, markedly decreases the circulating concentration of total cysteine and could thereby influence the response of the organism to the cytotoxic effects of chemotherapy. The aim of the present study was to assess the effects of the combination of ifosfamide/mesna on sulfhydryl and disulfide homeostasis in tumor patients. Ifosfamide was infused into 14 patients with advanced sarcoma for 5 days at a dose of 2.4-3.2 g/m2 per day together with mesna. The plasma concentrations of total mesna, cysteine, glutathione, and homocysteine were measured before and on days 1 and 6 of the first course of ifosfamide/mesna therapy and prior to the next course of chemotherapy, and the urinary excretion of cysteine and mesna was monitored daily using a high-performance liquid chromatography (HPLC) method. Ifosfamide/mesna resulted in a marked depletion of circulating total cysteine, i.e., cysteine, cystine, and cysteine mixed disulfides [from 245 +/- 36 to 50 +/- 14 nmol/ml (mean +/- 95% CI) on day 6], total glutathione (from 6.9 +/- 1.1 to 2.5 +/- 1.1 nmol/ml), and total homocysteine (from 12.3 +/- 2.1 to 1.4 +/- 1.1 nmol/ml). The values returned to baseline levels prior to the next course of chemotherapy. The urinary excretion of cysteine increased significantly from 0.28 to 1.82 mmol/day on the 1st day, whereupon it returned toward baseline. An average of 62% +/- 6% of the delivered dose of mesna was recovered in urine. The combination of ifosfamide/mesna results in depletion of circulating total cysteine, glutathione, and homocysteine. This marked derangement of sulfhydryl and disulfide homeostasis could modulate the efficacy and toxicity of ifosfamide/mesna therapy.