Development and evaluation of a Bayesian pharmacokinetic estimator and optimal, sparse sampling strategies for ceftazidime

Data were gathered during an activity-controlled trial in which seriously ill, elderly patients were randomized to receive intravenous ceftazidime or ciprofloxacin and for which adaptive feedback control of drug concentrations in plasma and activity profiles was prospectively performed. The adaptive feedback control algorithm for ceftazidime used an initial population model, a maximum a posteriori (MAP)-Bayesian pharmacokinetic parameter value estimator, and an optimal, sparse sampling strategy for ceftazidime that had been derived from data in the literature obtained from volunteers. Iterative two-stage population pharmacokinetic analysis was performed to develop an unbiased MAP-Bayesian estimator and updated optimal, sparse sampling strategies. The final median values of the population parameters were follows: the volume of distribution of the central compartment was equal to 0.249 liter/kg, the volume of distribution of the peripheral compartment was equal to 0.173 liter/kg, the distributional clearance between the central and peripheral compartments was equal to 0.2251 liter/h/kg, the slope of the total clearance (CL) versus the creatinine clearance (CLCR) was equal to 0.000736 liter/h/kg of CL/1 ml/min/1.73 m2 of CLCR, and nonrenal clearance was equal to + 0.00527 liter/h/kg. Optimal sampling times were dependent on CLCR; for CLCR of > or = 30 ml/min/1.73 m2, the optimal sampling times were 0.583, 3.0, 7.0, and 16.0 h and, for CLCR of < 30 ml/min/1.73 m2, optimal sampling times were 0.583, 4.15, 11.5, and 24.0 h. The study demonstrates that because pharmacokinetic information from volunteers may often not be reflective of specialty populations such as critically ill elderly individuals, iterative two-stage population pharmacokinetic analysis, MAP-Bayesian parameter estimation, and optimal, sparse sampling strategy can be important tools in characterizing their pharmacokinetics.

Pharmacodynamic analysis of sparse data from concentration- and effect-controlled clinical trials guided by a pilot study. An investigation by simulations

The purpose of this investigation was to explore, by computer simulation, the utility of two different clinical trial designs with sparse sampling (three concentration--effect measurements per subject) for population pharmacodynamic studies when the targeted drug concentration or effect measurements are determined by application of optimal sampling theory based on the results of a preceding, data-intensive pilot study. The two design paradigms were concentration-controlled and pharmacologic effect-controlled randomized clinical trials, respectively. The drug concentration--pharmacologic effect relationship was assumed to be describable by the Hill (sigmoid Emax) equation without hysteresis. Intersubject variability was represented by coefficients of variation of 30, 40, and 30% for Emax, EC50, and gamma, respectively. Random controller imprecision and measurement errors were included. Concentration and effect data for 100 subjects were generated by Monte Carlo simulation (ADAPT II), and pharmacodynamic parameter values were obtained by iterative two-stage analysis. These were then used to predict effect intensities over a range of drug concentrations, and the results were compared with those obtained by use of the true parameter values. Concentration- and effect-controlled trial designs were simulated in two forms: unconstrained and constrained with respect to the highest allowed targeted drug concentration or effect intensity. It was found that both types of unconstrained trials yielded good and comparable parameter estimates whereas the constrained trials (which are clinically more realistic) yielded more biased and imprecise estimates of individual pharmacodynamic parameters. Nevertheless, use of the latter to determine the effect intensities produced by different drug concentrations yielded good estimates but only in the range covered by the targeted concentration or effect measurements. For concentration-controlled trials it appears essential that the individuals in the pilot group and the clinical study group be drawn from the same population. Effect-controlled trials gave good results even when the pilot group was not representative of the population (e.g., for an aberrant subpopulation).

Population pharmacokinetics and renal function-sparing effects of amphotericin B colloidal dispersion in patients receiving bone marrow transplants

The purpose of this study was to evaluate the pharmacokinetics of amphotericin B colloidal dispersion and its effect on creatinine clearance in bone marrow transplant patients with systemic fungal infections. Seventy-five patients (42 females and 33 males) with a median age of 34.5 years and a median weight of 70.0 kg were enrolled in the study. Patients received 1 of 15 dose levels (range, 0.5 to 8.0 mg/kg of body weight) daily for a mean duration of 28 days and a mean cumulative dose amount of 8 g. Plasma samples for amphotericin B determination (median number, 4; range, 2 to 30) and daily serum creatinine values were obtained for each patient. Iterative two-stage analysis, one of several approaches to population pharmacokinetic and pharmacodynamic modelling, was employed for the pharmacokinetic analysis. The plasma data were available for 51 of 75 patients and were best described by a two-compartment model. Both plasma clearance and volume of distribution increased with escalating doses; the overall average terminal elimination half-life was 29 h. Of the covariates studied, only body weight and dose size were significant. Serum creatinine values over the duration of therapy were available for 59 of 75 patients. Overall, there was no net change in renal function over the duration of therapy; 12 patients had > 30% increases in creatinine clearance, whereas 13 had > 30% decreases. No measure of amphotericin B colloidal dispersion exposure, demographic values, or concomitant treatment with other medications was related to changes in the creatinine clearance.

Effect of aging on atenolol pharmacokinetics and pharmacodynamics

A study was conducted to characterize and compare the pharmacodynamics and pharmacokinetics of atenolol in young and elderly men. Six young (mean +/- SD, 25.0 +/- 3.0 years) and six elderly (63.0 +/- 3.2 years) healthy men took atenolol 100 mg orally once daily for 6 days. Heart rate response to submaximal exercise was measured at selected times for 48 hours, and plasma and urine samples were collected over the same time interval. The Sigmoid Emax model was fit to percent reductions in exercise heart rate and atenolol plasma concentrations. The younger men had significantly lower values for area under the steady-state plasma concentration-time curve and higher values for systemic clearance/F and renal clearance. EC50 values showed a trend toward greater sensitivity to the negative chronotropic effects of atenolol among the elderly men. Model-derived percent reductions in heart rate were greater at all concentrations among the elderly men. These data suggest that group differences in atenolol pharmacokinetics were likely a result of age-related decline in renal function, and that the elderly subjects were at least as sensitive as, and maybe even more sensitive than, the younger subjects to the negative chronotropic effects of atenolol.

Optimal sampling theory: effect of error in a nominal parameter value on bias and precision of parameter estimation

The authors examined the robustness of optimal sampling theory in estimating the parameter values of two different populations of patients receiving a constant rate, half-hour intravenous infusion of theophylline. One population consisted of smokers; the other included nonsmokers. The smoking population was predicted to have a serum clearance approximately 50% greater than the nonsmokers because of an induction of the cytochrome P450 system. After an initial study to provide both patient-specific and population mean parameter values, optimal sampling strategies that were derived from each population (seven sample split designs) and the patient's seven sample and four sample design were determined. A second study was performed with an overall sampling strategy that was superset of all the above strategies. The analysis of all samples served as the reference for the parameter values. Bias and precision of the values determined with each of the optimal sampling sets (seven sample sets based on the "correct" and "wrong" populations, the patient's seven and four sample sets) were determined relative to these reference values. Irrespective of the sample set used for analysis, unbiased and precise parameter estimates, particularly of hybrid parameters were provided. With the patient's four sample set, Vss was significantly biased, but the value of (2.2%) was clinically insignificant. The authors conclude that optimal sampling theory, as implemented in this study, provides robust estimates of important pharmacokinetic parameter values, even when errors of 50% are present in the clearance of the population used to calculate the optimal sampling design.

Ticarcillin-clavulanic acid pharmacokinetics in preterm neonates with presumed sepsis

The objective of the reported study was to characterize the pharmacokinetics of ticarcillin and clavulanic acid in premature low-birth-weight (less than 2,200 g) neonates with presumed sepsis. Eleven infants received 12 courses of ticarcillin-clavulanic acid at 75 mg/kg of body weight intravenously every 12 h. Blood samples were collected at 0.5, 1.5, 4, and 8 h following the infusion of the initial dose. The concentrations of ticarcillin and clavulanic acid were determined by a microbiologic assay. Median (interpatient coefficients of variation) values for the volume of the central compartment, total steady-state volume, distributional clearance, total clearance, and terminal elimination half-life for ticarcillin were 0.030 liter/kg (21%), 0.26 liter/kg (48%), 0.41 liter/h/kg (47%), 0.047 liter/h/kg (47%), and 4.2 h (45%), respectively. For clavulanic acid the parameters were 0.28 liter/kg (32%), 0.36 liter/kg (34%), 11 liters/h/kg (36%), 0.12 liters/h/kg (72%), and 1.95 h (40%), respectively. Our results suggest that the current dosing recommendations of 75 mg/kg every 12 h risk subtherapeutic clavulanic acid concentrations and that 50 mg/kg every 6 h is a more rational dosing strategy.

Mathematical examination of dual individualization principles (II): The rate of bacterial eradication at the same area under the inhibitory curve is more rapid for ciprofloxacin than for cefmenoxime

OBJECTIVE

To compare two antibiotics at equal ranges of area under the inhibitory curve (AUIC) exposure to determine if the rate of bacterial eradication differed between these antibiotics.

DESIGN

Retrospective comparison of two previously collected studies of similar patients with nosocomial pneumonia.

SETTING

Hospitalized patients, most intubated in critical care units with nosocomial pneumonia.

PARTICIPANTS

Patients treated with either i.v. ciprofloxacin (n = 74) or the i.v. third-generation cephalosporin cefmenoxime (n = 43) were compared for their length of treatment required to eradicate bacterial pathogens from their respective infection sites, using serial cultures from the site of infection. All patients were also assessed for clinical outcomes. Serum samples were obtained to evaluate individual patient antibiotic pharmacokinetics, which were used to model pharmacodynamics of response. The HPLC assay used for each antibiotic had interday coefficients of variation < 10 percent. Serum concentration versus time profiles were fit using the computer program ADAPT II to determine pharmacokinetic parameters for each patient. The primary drug exposure measure that related to response was the AUIC, calculated as steady-state AUC0-24/minimum inhibitory concentration.

RESULTS

AUIC values in the patients ranged from 6.0 to more than 7000, yet the AUIC value was highly predictive of time to bacterial eradication (p < 0.001). Although more than 75 percent of patients eventually achieved eradication of pathogens from tracheal aspirate cultures, ciprofloxacin and cefmenoxime differed significantly in the time required to sterilize these cultures. At appropriate AUIC values (> 250) for ciprofloxacin, the median time to eradication was two days, while cefmenoxime (also at AUIC values > 250) required six days to achieve the same result.

CONCLUSIONS

We conclude that the more rapid in vitro bacterial killing, which is characteristic of ciprofloxacin at optimal AUIC values, can manifest in vivo as more rapid clearance of bacteria from the respiratory tract of patients, even when both agents are controlled for initial antibacterial exposure (i.e., same AUIC).

Pharmacokinetics and bioavailability of fluconazole in patients with AIDS

Fluconazole pharmacokinetics were evaluated for 10 volunteers with AIDS who had no clinical evidence of gastroenteritis. Single 100-mg intravenous (i.v.) and oral (p.o.) doses were administered in a randomized, crossover design. i.v. doses were delivered by a constant-rate infusion over 30 min. Serum fluconazole concentrations were measured by gas-liquid chromatography. The i.v. and p.o. studies were modelled simultaneously by iterative two-stage analysis, which provided individual parameter estimates and a population pharmacokinetic model. Median areas under the concentration-time curves for i.v. and p.o. studies did not differ (90.6 and 99.3 micrograms/ml.h, respectively). Consistent with this finding, the median fractional bioavailability was 1.1 (range, 0.45 to 1.3), comparable to those in healthy subjects. Serum pharmacokinetics in these AIDS patients were generally similar to published data for healthy volunteers. However, following p.o. dosing, we observed a slightly delayed and highly variable time to maximum concentration in serum (median, 2 h; range, 15 min to 8 h). Data were well described by a linear, two-compartment pharmacokinetic model with first-order absorption and elimination. Repeated-measures analysis of variance found no significant differences among any of the pharmacokinetic parameters between i.v. and p.o. studies. On the basis of our findings, we suggest no change in dosage of p.o. fluconazole in patients with AIDS who show no clinical signs of enteropathy.

Predictors of trough concentrations of oral ciprofloxacin

Patients enrolled in a fixed-dose clinical trial of oral ciprofloxacin had trough concentrations measured to document absorption and monitor compliance. The objective was to determine whether any demographic characteristics might be important predictors of the concentrations. Stepwise multivariate linear regression revealed no correlation between ciprofloxacin trough concentrations and serum creatinine, estimated creatinine clearance (Clcr), weight, height, body surface area, or gender. However, age exhibited a direct linear relationship with trough concentrations (Y in microgram/ml), Y = 0.020.age--0.541 (p < 0.003). We conclude that for patients with Clcr 30 ml/minute or above, age is a more important predictor of ciprofloxacin trough concentration than renal function. Dosage adjustment should not be arbitrary but should be guided by minimum inhibitory concentration, clinical response, and side effects.

Pharmacodynamics of intravenous ciprofloxacin in seriously ill patients

Seventy-four acutely ill patients were treated with intravenous ciprofloxacin at dosages ranging between 200 mg every 12 h and 400 mg every 8 h. A population pharmacokinetic-pharmacodynamic analysis relating drug exposure (and other factors) to infectious outcome was performed. Plasma samples were obtained and assayed for ciprofloxacin by high-performance liquid chromatography. Samples from patients were frequently cultured so that the day of bacterial eradication could be determined. The pharmacokinetic data were fitted by iterative two-stage analysis, assuming a linear two-compartment model. Logistic regression was used to model ciprofloxacin exposure (and other potential covariates) versus the probabilities of achieving clinical and microbiologic cures. The same variables were also modelled versus the time to bacterial eradication by proportional hazards regression. The independent variables considered were dose, site of infection, infecting organism and the MIC for it, percent time above the MIC, peak, peak/MIC ratio, trough, trough/MIC ratio, 24-h area under the concentration-time curve (AUC), AUC/MIC ratio (AUIC), presence of other active antibacterial agents, and patient characteristics. The most important predictor for all three measures of ciprofloxacin pharmacodynamics was the AUIC. A 24-h AUIC of 125 SIT-1.h (inverse serum inhibitory titer integrated over time) was found to be a significant breakpoint for probabilities of both clinical and microbiologic cures. At an AUIC below 125 (19 patients), the percent probabilities of clinical and microbiologic cures were 42 and 26%, respectively. At an AUIC above 125 (45 patients), the probabilities were 80% (P < 0.005) and 82% (P < 0.001), respectively. There were two significant breakpoints in the time-to-bacterial-eradication data. At an AUIC below 125 (21 patients), the median time to eradication exceeded 32 days; at an AUIC of 125 to 250 (15 patients), time to eradication was 6.6 days: and at AUIC above 250 (28 patients), the median time to eradication was 1.9 days (groups differed; P < 0.005). These findings, when combined with pharmacokinetic data reported in the companion article, provide the rationale and tools needed for targeting the dosage of intravenous ciprofloxacin to individual patients' pharmacokinetics and their bacterial pathogens' susceptibilities. An a priori dosing algorithm (based on MIC, patient creatine clearance and weight, and the clinician-specified AUIC target) was developed. This approach was shown, retrospectively, to be more precise than current guidelines, and it can be used to achieve more rapid bacteriologic and clinical responses to ciprofloxacin, as a consequence of targeting the AUIC.

Development of a population pharmacokinetic model and optimal sampling strategies for intravenous ciprofloxacin

Data obtained from 74 acutely ill patients treated in two clinical efficacy trials were used to develop a population model of the pharmacokinetics of intravenous (i.v.) ciprofloxacin. Dosage regimens ranged between 200 mg every 12 h and 400 mg every 8 h. Plasma samples (2 to 19 per patient; mean +/- standard deviation = 7 +/- 5) were obtained and assayed (by high-performance liquid chromatography) for ciprofloxacin. These data and patient covariates were modelled by iterative two-stage analysis, an approach which generates pharmacokinetic parameter values for both the population and each individual patient. The final model was used to implement a maximum a posteriori-Bayesian pharmacokinetic parameter value estimator. Optimal sampling theory was used to determine the best (maximally informative) two-, three-, four-, five-, and six-sample study designs (e.g., optimal sampling strategy 2 [OSS2] was the two-sample strategy) for identifying a patient's pharmacokinetic parameter values. These OSSs and the population model were evaluated by selecting the relatively rich data sets, those with 7 to 10 samples obtained in a single dose interval (n = 29), and comparing the parameter estimates (obtained by the maximum a posteriori-Bayesian estimator) based on each of the OSSs with those obtained by fitting all of the available data from each patient. Distributional clearance and apparent volumes were significantly related to body size (e.g., weight in kilograms or body surface area in meters squared); plasma clearance (CLT in liters per hour) was related to body size and renal function (creatinine clearance [CLCR] in milliliters per minute per 1.73 m2) by the equation CLT = (0.00145.CLCR + 0.167).weight. However, only 30% of the variance in CLT was explained by this relationship, and no other patient covariates were significant. Compared with previously published data, this target population had smaller distribution volumes (by 30%; P < 0.01) and CLT (by 44%; P < 0.001) than weight- and CLCR- matched stable volunteers. OSSs provided parameter estimates that showed good to excellent estimates of CLT (or area under the concentrations-time curve [AUC]) were unbiased and precise (e.g., r2 for AUC for all data versus AUC for OSS2 was > 0.99) and concentration-time profiles were accurately reconstructed. These results will be used to model the pharmacodynamic relationships between ciprofloxacin exposure and response and to aid in developing algorithms for individual optimization of ciprofloxacin dosage regimens.

A preliminary trial of the gel test for blood group serology

The gel test, first reported in 1988, can be used for most areas of blood group serology, but this report deals specifically with ABO/Rh typing and antibody screening and investigation. The technique is simple and quick to perform and the reading of results is standardised and clean with no false-positive reactions being found in antiglobulin tests. Since gel anti-human globulin tests require no washing and the test cards may be sealed during centrifugation, the system is particularly advantageous for 'loosely bound' antibodies and 'high-risk' samples.

The significance of CD34 and TdT determinations in patients with untreated de novo acute myeloid leukemia

The results of intensive chemotherapy given to 247 adults at the University of Maryland Cancer Center with previously untreated de novo acute myeloid leukemia (AML) were reviewed with respect to expression of terminal deoxynucleotidyl transferase (TdT) and CD34. Of the 228 patients with data for TdT, 32 (14%) had > 5% of the leukemia cells positive by an immunofluorescence assay. The median age of the TdT-positive patients was approximately 10 years less than the TdT-negative patients (50 versus 60 years). Patients with TdT-positive AML had similar median survival (12 versus 10.5 months) and complete remission (CR) rates (53 versus 59%), but a greater frequency of long-term complete responders (60 of complete remitters versus 20%, p = 0.08) than TdT-negative patients. Of 126 patients tested, 59% were CD34-negative (< 20% reactivity with leukemia cells). These 74 patients (median age 60 years) had a greater CR rate (71 versus 48%, p = 0.008) than the 52 CD34-positive patients (median age 60 years), and improved survival (p = 0.013 by Wilcoxon) although there was no difference in the duration of CR between the CD34-positive and negative groups. Of CD34-positive patients 12/52 remain in continuous CR, and 16/74 CD34-negative patients remain in continuous CR. None of eight patients strongly positive for CD34 (> 70% reactivity) remain disease-free. Positivity for TdT or CD34 was associated with less differentiated AML. Of CD34-positive patients, 44% had FAB M0/M1 morphology versus 13% of CD34-negative patients (p = 0.0001); similarly, 47% of TdT-positive patients were FAB M0/ML1 versus 25% of TdT-negative patients (p = 0.01). Of seven patients with FAB M4E0, five were CD34-positive. Of the 12 CD34-positive survivors, four had FAB M4E0. Thus CD34 expression predicts for CR rate and overall survival in adults with AML. TdT expression does not significantly affect overall outcome but may be associated with longer CR durations.

Approaches to optimal dosing of hexamethylene bisacetamide

HMBA is a potent differentiating agent capable of causing > 95% morphological differentiation in cell lines in vitro. The induction of differentiation is dependent on both the concentration of and the duration of exposure to HMBA. However, acute toxicities (neurotoxicity and acidosis) have limited the maximal HMBA css value to < 2 mM, which is at the lower limit of effective in vitro concentrations. When HMBA css values have been maintained at 1-2 mM, thrombocytopenia has limited the duration of HMBA infusion to < or = 10 days. The present studies were performed to determine whether exposure to HMBA could be individualized and maximized without resulting in intolerable toxicity to patients and to determine which factors would predispose a patient to the development of acute toxicity during treatment with HMBA. For these investigations, patients were given HMBA at a target css using an adaptive-feedback-control method rather than at a set dose. Because HMBA administration produces large anion gaps, a simple maneuver such as alkalinization might enable the escalation of plasma HMBA css values to > 2 mM. HMBA was given as a 5-day CI to 14 patients (26 courses) at 2 target HMBA css levels near the maximal tolerated value in the presence or absence of concurrent alkalinization with sodium bicarbonate. Symptomatic acidosis occurred in one patient who did not receive bicarbonate. Neurotoxicity proved to be dose-limiting at the target HMBA css value of 1.5-2.0 mM in the absence of concurrent alkalinization and at a css level of > 2 mM, regardless of alkalinization. No neurotoxicity was seen at target HMBA css values of 1.5-2.0 mM in patients who did receive concurrent alkalinization. Alkalinization was not associated with any detectable changes in plasma HMBA metabolites. With the maximal tolerable 5-day HMBA css having thus been defined at 1.5-2.0 mM, we attempted to maximize exposure to HMBA by defining a tolerable duration of infusion. Individualization of the duration of HMBA infusion to a target nadir PLT was performed in patients who had received an initial 5-day CI of HMBA at a css 1.5-2.0 mM along with concurrent alkalinization. The AUC achieved and the thrombocytopenia produced during this first course were used to predict the duration of infusion that each patient would subsequently tolerate (at an HMBA css of 1-2 mM) to achieve a nadir PLT of 75,000-100,000/microliters.(ABSTRACT TRUNCATED AT 400 WORDS)

Use of adaptive control with feedback to individualize suramin dosing

Suramin is the first putative growth factor inhibitor in clinical trial that has demonstrated antitumor activity. Administration of suramin is complicated by a narrow therapeutic index and significant interpatient variability of measured pharmacokinetic parameters. Because both antitumor response and dose-limiting toxicities are related to plasma suramin concentration profiles, individualized dose schedules are required for optimal administration of the compound. In this report, the use of optimal sampling theory to derive sparse data monitoring and control strategies for use with suramin is described. A fixed rate continuous infusion schedule was used in seven patients, and the time to peak concentration (280-300 micrograms/ml) ranged from 7.7-21 days (mean, 13.2 days) with a decline to 150 micrograms/ml in 3-22 days (mean, 11 days). An initial population pharmacokinetic model was fit using a maximum likelihood algorithm. The mean volume of the central compartment was 4.5 +/- 6.7 liters/m2, volume of the peripheral compartment 10.6 +/- 1.4 liters/m2, distributional half-life 25 +/- 5.4 h, and elimination half-life 29.7 +/- 6.9 h. The terminal half-life was shorter than previously reported. These parameters were used as the initial population model for an iterative 2-stage analysis. The resulting distributional half-life of 22.3 +/- 2.7 h and elimination half-life of 28.2 +/- 5.0 h were similar, reflecting the intensive sampling. The iterative 2-stage analysis model was then used to determine the optimal sampling times and to simulate 20 data sets for a protocol designed to maintain plasma concentrations in a defined concentration range. This strategy is currently under investigation in phase I clinical trials.

Pharmacokinetic and pharmacodynamic analysis of fluorouracil during 72-hour continuous infusion with and without dipyridamole

During a phase I trial of 3-day simultaneous continuous intravenous infusions of varying doses of fluorouracil (5FUra) and 7.7 mg/kg/d of dipyridamole, we examined the relationships between 5FUra dose and steady-state plasma concentration (Css) and the percentage reduction in WBCs, as well as the percentage frequency of stomatitis. The 5FUra was administered at doses ranging from 185 mg/m2/d times three to 3,600 mg/m2/d times three. In 42 patients, 86 cycles of 5FUra plus dipyridamole and 28 cycles of 5FUra alone were analyzed. The Css of 5FUra varied even within the same dose level. When patients receiving the same 5FUra dose were considered, the interpatient coefficient of variation of 5FUra Css in cycles of 5FUra plus dipyridamole was 23% +/- 4.2%. For courses of 5FUra alone, the coefficient of variation of 5FUra was 15.6% +/- 6.5%. When the occurrence of any degree of stomatitis was related to the Css 5FUra, with patients grouped in cohorts of 2-mumol/L increments, the following equations accurately described the frequency of stomatitis: for 5FUra plus dipyridamole, percentage frequency of stomatitis = 100(1-e-0.114Css), r2 = 0.88; for 5FUra alone, percentage frequency stomatitis = 100(1-e0.122Css), r2 = 0.80. When 5FUra dose was substituted for Css, these relationships were as follows: percentage frequency of stomatitis = 100(1-e-0.00031 [dose]), r2 = 0.85; and percentage frequency of stomatitis = 100(1-e-0.00051 [dose]), r2 = 0.80. When the relationship between the percentage reduction in WBC and Css 5FUra was examined, statistically significant relationships were also apparent: for 5FUra plus dipyridamole, percentage reduction in WBC = 100(1-e-0.085Css), r2 = 0.46; for 5FUra alone, percentage reduction in WBC = 100(1-e-0.060Css), r2 = 0.61. When 5FUra dose was substituted for Css, these relatinships were as follows: percentage reduction in WBC = 100(1-e-0.00023 [dose]), r2 = 0.40; percentage reduction in WBC = 100(1-e-0.00024 [dose]), r2 = 0.65. The relationship between either Css 5FUra or dose 5FUra and either stomatitis or myelosuppression were also well described by the modified Hill equation (J Theor Biol 20:171-201, 1968). These analyses indicate that it should be possible to develop therapeutic regimens wherein 5FUra is administered to achieve a targeted Css determined by the risk and severity of toxicity deemed acceptable.