AU2014202609A1 - Personalized therapeutic treatment process - Google Patents

Abstract

Abstract There is disclosed a process for simultaneous determination of the delivery of two or more chemical (therapeutic) agents to diseased tissue. The procedure generates a patient-specific report that is used to select the best chemical agent and dosage on a personalized basis.

Description

Personalized therapeutic Treatment Process The present application is a divisional application from Australian patent application number 2007323782, the entire disclosure of which is incorporated herein by reference. Technical Field There is disclosed a process for simultaneous determination of the delivery of two or more chemical (therapeutic) agents to diseased tissue and pathogens. The procedure generates a patient specific report that is used to select the best chemical agent and dosage on a personalized basis. Background Chemotherapy is a powerful tool available to clinicians for cancer treatment. Selection of drugs, combinations, dosages and schedules is frequently based on group statistics data. Dosage of chemotherapy agents can be difficult. For example, if the dose is too low, therapy will be ineffective against the tumor, while at excessive doses the toxicity (side-effects) will be intolerable to the patient. This has led to the formation of detailed dosing schemes in most treatment settings, which give guidance on the correct dose and adjustment in case of toxicity. In immunotherapy, drugs are in principle used in smaller dosages than in the treatment of malign diseases. In most cases, the dose is adjusted for the patient's body surface area and a composite measure of weight and height that mathematically approximates the body volume. Therefore, it is desirable to select drugs based on analyses of individual patient tumor factors. This should result in enhanced efficacy without increased toxicity. Although dosage of chemotherapy is normally based on body surface area, little data actually support the use of this method (Cancer: Principles and Practice of Oncology, 6 h edition. 335-344, 2001). The major disadvantage of this method is that body surface area poorly correlates with liver function. This is a problem since quite a few cytostatic types of cancer chemotherapeutic drugs are metabolized via the liver. In addition, the present dosage strategies result in inter-patient variants in the area under the curve (AUC) and drug clearance by a factor of several multiples (J. Clin. Oncol. 14:2590-2611, 1996). Due to these shortcomings, a more predictive tool based on individualized tumor characteristics is needed for drug and dosage selection. Treating breast cancer is challenging for patient and clinician because many therapies fail to work as planned. This is often the result of the tumor being non-responsive to the chosen drug. There is currently no way to consistently predict responsiveness prior to initiating treatment, so the trial-and error method is used. Selection of breast cancer chemotherapy agents is currently based on several general parameters such as tumor histology, clinical stage, receptor and antigenic status. These predictors lead to some patients receiving toxic drugs from which they fail to benefit. Moreover, time expended to identify an individualized regimen that is beneficial further delays responsive treatment. A wide range of tumors can be treated effectively when detected early and the patient is placed on a suitable therapeutic regimen. Thus, better information on predictors of tumor response to specific chemotherapy compounds in individual patients is needed. 1 The effectiveness of chemotherapy regimens for breast cancer is highly variable from patient to patient. As a result, a substantial proportion of patients receive toxic drugs from which they fail to benefit, and some patients are delayed in getting the regimen from which they would benefit most. Accurate predictors of early response to chemotherapy compounds 5 may prove valuable in determining optimal clinical management of patients with breast cancer. There have been some ideas tried to try to address the foregoing problem. One was described in Silverman et al., Mol. Imaging Biol. 8(1):36-42, 2006, using 3'-[F-18]fluoro-3' deoxythymidine with positron emission tomography (PET) to try to predict breast cancer response to therapy. Specifically, the authors concluded that a PET scan using the fluorinated 10 deoxythymidine marker acquired two weeks after the end of the first course of chemotherapy was "useful" for predicting longer-term efficacy of chemotherapy regimens for women with breast cancer. This "usefulness" is hardly a direct answer to the choice of therapeutic or an appropriate dosage for the patient. Instead, it is simply an efficacy-monitoring tool to measure tumor shrinkage. 15 In Delgrorio et al., a single compound was administered to animal and cell lines using a therapeutic-level dose and used AMS used to measure total 1 4 C. There was no metabolite quantification attempted by HPLC. U.S. Patent 4,037,100 describes an apparatus which can be used for the detection of electronegative particles and provide data as to their elemental composition. The apparatus 20 includes an accelerator mass spectrometer (AMS) that can be used for making mass and elemental analyses. Accelerator mass spectrometry (AMS) was developed as a sensitive method for counting long-lived but rare cosmogenic isotopes, typically those having half-lives between 10 3 and 2 X 10 7 years. Isotopes with this range of half-lives are too long-lived to detect by conventional decay counting techniques but are too short-lived on geological 25 timescales to be present in appreciable concentrations in the biosphere or lithosphere. Assay of cosmogenic isotopes (such as, 10 Be, 1 4 C, 26Al, 4 Ca, 36 C1, and 1291) by AMS has become a fundamental tool in archaeology, oceanography, and the geosciences, but has not been widely applied to problems of a biological or clinical nature. White and Brown (Trends Pharmacol. Sci. 25:442-447, 2004) is a review providing 30 examples of AMS used in pharmacology and toxicology. 14C-urea was given to children in Sweden and used to detect Helicobacterpylori by measuring 14C in the breath or urine by Accelerator Mass Spectrometry (AMS). 14C-benzene production was followed in mice to liver and bone marrow by AMS. In addition, the absence of a drug getting to DNA was determined by similar micro-dosing studies. One cited study showed that a carcinogen, MeIQx, formed 35 from cooked meats was able to reach normal and tumor tissue in human patients with colon cancer (Mauthe et al., Int. J. Cancer 80:539-545, 1999). Similar DNA damage was seen in human tumor and normal tissue. At a high dose, 1 4 C-tamoxifen had been administered to women who were about to undergo hysterectomies to detect (by AMS) damage to the DNA in the uterus. The study concluded that some DNA damage was seen by AMS but not enough to 2 cause neoplasia. Other isotopes (especially calcium and aluminum) have been followed in humans with AMS. However, despite such uses of AMS, no mention of differential effects of two or more drugs on cancer or normal cells has been disclosed. Brown et al. (Mass Spectrometry Rev. 25:127-145, 2006), looks at AMS for detecting 5 changes in proteins and DNA in test subjects administered ' 4 C-carcinogens. Further, 3 H, "Be, 2 6 Al, 36 C1, 4 1 Ca, and 129I isotopes have been used for biological research with AMS. Wu et al. (Int. J. Gynecol. Cancer 12:409-423, 2002) is a review article on proteomics in cancer research using two-dimensional gels, protein chips, and other methods. Mass spectrometry (but not AMS) was used to resolve and measure small organic molecules from 10 biological samples. A technique called laser capture micro-dissection (LCM) was suggested to separate specific cells under a microscope for subsequent analysis. LCM has also been combined with mass spectrometry to capture and remove cells from surrounding tissue. Lee and Macgregor, Modern Drug Discovery (July):45-49, 2004 looked at drug resistance in cancer cells due to changes in drug uptake, absorption, metabolism, elimination, 15 and other mechanisms. DNA microarray data may be used to detect biomarkers of chemotherapy response. For example, marker CA125 rises in 70% of ovarian cancer patients with relapse to chemotherapy treatment. In vitro testing of patient cancer cells for drug resistance and sensitivity was done by a number of commercial organizations, such as Rational Therapeutics, Oncotech, and Genzyme. 20 The value of these tests is unsubstantiated; and many insurance companies do not cover in vitro testing of tumor cells. Oncotech, for example, claims that its EDR Assay, which exposes primary cancer cells to five days of drugs in soft agar, can predict drug resistance to chemotherapy in patients ("over 99% accuracy for identifying ineffective agents," from Oncotech website). However, Oncotech requires at least two grams of viable tumor tissue; and 25 patients must not be on chemotherapy or radiation therapy within three weeks of specimen collection. Similarly, Genzyme claims 99% accuracy in predicting tumor resistance to chemotherapy. None of these companies claims accuracy for predicting which drug(s) will be efficacious, particularly in combinations of two or more drugs administered at the same time. Lastly, Sharma et al. (Cancer Cell International 5:26-45, 2005) showed sodium MRI 30 imaging was used to monitor taxotere treatment response in rat breast cancer. Elevated intracellular sodium was found in benign and malignant breast tumors. Anticancer agents increased the intracellular sodium. However, apoptosis caused by such drugs leads to disruptions in the distribution of sodium. Therefore, the authors conclude that sodium MRI may be used as an in vivo drug monitoring method to evaluate taxotere chemosensitivity 35 response of tumors. However, the authors further opine that the method with in vivo tissues "raises doubt of accuracy in predicting tumor features." Summary There is disclosed a process for simultaneous determination of the amounts of delivery of two or more agents to diseased tissue, on a cellular basis. The process generates a patient 3 specific report that is used to select the best agent and dosage on an individualized basis for delivery to a site of activity. Specifically, there is disclosed a process for determining personalized therapy, comprising: (a) administering a test cocktail to a test subject, wherein the test cocktail 5 comprises two or more different therapeutic agents at a dosage at least two times lower than an expected therapeutic dose; (b) obtaining a sample biopsy of the relevant diseased tissue for study; and (c) analyzing the sample biopsy for each of the administered therapeutic agents and their metabolites. 10 Preferably, the dosage of the test cocktail is a tracer dose, wherein a tracer dose is less than 10% of a therapeutic dose. Preferably, the analyzing step is performed with an Accelerator Mass Spectrometry (AMS) instrument. Preferably, the process further comprises pausing from about 10 minutes to about two hours after administering a test cocktail to allow for tissue distribution of the test cocktail. Preferably, the sample biopsy is a piece of tissue 15 selected from the group consisting of excised tumor tissue, blood, fractionated blood, isolated pathogen-infected tissue, and combinations thereof. Preferably, the process further comprises fractionating the sample biopsy by sorting the sample into component cell types. Most preferably, when the sample biopsy is blood, the blood sample is fractionated into each type of white and red blood cell. 20 Brief Description of the Figures Figure 1 is a flow chart describing a general workflow for the personalized treatment of cancer. Figure 2 shows separation of cyclophosphamide and paclitaxel by high performance liquid chromatography. 25 Figure 3 shows a model of the personalized therapeutic approach demonstrating quantitation of the amount of cyclophosphamide and paclitaxel in breast tumor biopsy after administration of test cocktail in three patients A, B and C. Figure 4 shows a scheme utilizing high performance liquid chromatography to generate individual fractions at one minute intervals followed by quantitation of the radiolabel in each 30 fraction using accelerator mass spectrometry (AMS) to produce cyclophosphamide and paclitaxel distribution data in breast tumor biopsy after administration of test cocktail in three patients A, B and C. Figure 5 shows the amount of CVT 337 distributed into tissues 5 minutes after oral and iv administration in mice. Samples from three (3) mice in each treatment group were pooled 35 and measured for total 1 4 C activity by accelerator mass spectrometry (AMS). Figure 6 shows the amount of CVT 337 in tissues over a 24 hour period following iv administration in mice. Samples from three (3) mice in each treatment group were pooled and measured for total 14C activity by accelerator mass spectrometry (AMS). 4 Figure 7 shows red blood cell 14C-folate concentration for the first 8 days (top) and 200 days (bottom) post-dose in a healthy human subject. The three-day delay before appearance of 1 4 C-folate represented the time required for 4 C-folate incorporation into cells in the marrow during maturation. Error bars represent ± 1 standard deviation of triplicate determinations of 5 4C by AMS. Figure 8 shows an HPLC-AMS chromatogram of plasma sample collected one hour after oral administration of 1 4 C-folic acid to a healthy human subject. Absorption was monitored at 292 nm (solid line) and 1 4 C concentration was measured by AMS and expressed as modems (dashed line). The large peak from 2-4 minutes is from ascorbic acid added to the 10 sample to protect against oxidation. Folic acid (FA) and 5-methyltetrahydrofolate (5MTFA) reference standards were added to the plasma before extraction to check for recovery and to mark FA and 5MTFA retention times. Figure 9 shows a simple model depicting the uptake and loss of tracer. Figure 10 shows a multi-compartment model of folate distribution in humans. This 15 schematic depicts major pools involved in folate metabolism. The amount of tracer was measured in all pools shown with a solid line. The tissue pool, shown with a dotted line, was the only pool whose ' 4 C-folate concentration was not measured. Compartmental modeling allowed the distribution of the tracer into the tissue pool to be determined by difference. Figure 11 shows a compartmental model of folate kinetics in a human volunteer. 20 Compartments are numbered 1 through 11 and transfer coefficients shown as k (recipient, donor). Figure 12 shows a recovery of spiked 5FU and PAC from tissue homogenates using different extraction solvents. This experiment is described in Example 3 herein. Figure 13 shows a chromatogram showing separation of 5FU and PAC in mouse tissue 25 homogenate extracts. The 5FU and PAC peaks in the UV trace represent non-labeled reference standards spiked into experimental samples prior to separation to mark the retention time of radiolabeled 5FU and PAC which do not produce a UV signal at the dosed concentrations. One-minute fractions were collected and fractions corresponding to 5FU and PAC were further analyzed by AMS for 14C quantitation. 30 Figure 14 shows total radiolabel signal in plasma, tumor xenografts of HT-29 human colon cancer cell lines and normal lung tissue in mice two hours after receiving 5FU, PAC or a cocktail of both 5FU and PAC. Figure 15 shows the amount of 5FU and PAC in tumor, normal lung tissue and plasma after mice received either 5FU, PAC or both as a cocktail two hours after iv administration. 35 Figure 16 shows chromatographic separation and quantitation of 5FU and PAC in tumor and lung tissue extracts after treatment with either 5FU or PAC. The tissue extracts were chromatographed with and without spiking with less than 1.0 DPM of 5FU. The chart is normalized to PAC. 5 Detailed Description Definitions The following defined terms are used herein: A Test Subject is a patient; study volunteer; or an animal model. 5 A Test Cocktail is a dose of two or more chemical agents administered to a Test Subject. The chemical agents may be mixed together and co-administered by oral, intravenous or other routes of administration. Alternatively, each chemical agent is administered by a different route of administration. Alternatively, each chemical agent is administered at a different time. 10 A Tracer Dose is a sub-therapeutic dose administered to Test Subject at trace levels to reduce chemical exposure risk. A Target Tissue is an organ, tissue or cell mass whose uptake of drugs is to be studied. Distribution Time is the time between administration of Test Cocktail and collection of Target Tissue. 15 Detection System is the analytical method and instrumentation for assessing the levels of each Tracer Dose in Target Tissue. Test Report is a summary of test procedure results. Comparison of Methods The present disclosure provides a test cocktail comprising two or more different 20 therapeutic agents at a dosage at least two times lower than an expected therapeutic dose. This low dose is called a "tracer dose." Further the amount of radioactivity is generally no larger than about 100 nCi. The present disclosure also uses several different detection techniques, including AMS, mass spectroscopy (MS) and a combined liquid chromatography (LC) for separations combined with MS detection. 25 The following table lists prior processes in comparison with the present disclosure. Status Protocol Purpose Regulatory Chemical 1 4 C Detection Dose Size Isotope Method Prior Microdose Research Exploratory <100 ug 100 nCi AMS Art Investigational chemical New Drug (E- dose IND) Prior Therapeutic Research Investigational 40-200 mg 100 nCi AMS Art dose New Drug (IND) chemical dose Current Tracer Dose -Research -Exploratory Typically, 100 nCi AMS Claims of two or - IND <10 mg total LC/MS 6 more agents Diagnostics -Premarket chemical MS Approval for in dose of PET vitro diagnostic each agent use The following procedure employs Tracer Doses of two therapeutic agents mixed in a Test Cocktail and administered to a Test Subject to quantify the delivery of each therapeutic agent in tumor or normal tissue. The minimally-invasive procedure, combined with highly sensitive analytical methods and instrumentation, provides individualized information for 5 selecting an optimal therapeutic agent and optimal dose for personalized therapy. Specifically, this example provides a procedure for personalized treatment of breast cancer. The delivery of two chemotherapy agents, cyclophosphamide and paclitaxel, to malignant breast tissue is assessed. Cyclophosphamide is in a class of drugs known as alkylating agents. It slows or stops the growth of cancer cells in the body. The length of treatment depends on how well the 10 body responds to drug, and the type of cancer. Paclitaxel is widely used as a chemotherapy drug in the treatment of various malignancies, including breast, ovarian, and lung cancers. This procedure is conducted when a Test Subject is scheduled for a needle biopsy procedure. Needle biopsy procedures are routinely performed to obtain tissue samples for pathological determination of benign and malignant conditions. Approximately 80% of 15 subjects that undergo a needle biopsy procedure are determined to have a benign condition. This procedure establishes a high margin of safety, especially for 80% of Test Subjects who are later found to have a benign condition. Chemotherapeutic agents may be toxic to all Test Subjects at therapeutic doses. To minimize this risk to all Test Subjects, a Test Cocktail consisting essentially of Tracer Doses of cyclophosphamide and paclitaxel was formulated. In 20 this example, the amount of each agent in the Test Cocktail was 1000 times less than the corresponding therapeutic dose. Detection of Tracer Doses of cyclophosphamide and paclitaxel in needle biopsies requires highly sensitive analytical methods and instrumentation. Accelerator Mass Spectrometry (AMS) is an extremely sensitive method for detecting Trace Doses of 25 compounds labeled with a tracer such as radiocarbon isotopes (1 4 C). Extremely small amounts of 1 4 C isotope tracer were used in the labeling of cyclophosphamide and paclitaxel due to the extreme sensitivity of AMS. As a result, the radiation risk of the Test Cocktail to Test Subjects is comparable to the radiation risk already present in the environment at natural levels over a one-year cumulative period. Moreover, the extremely low amount of 14 C isotope tracer in the 30 Test Cocktail and Target Tissue precluded special handling procedures or precautions, making this procedure compatible with standard ethical, environmental, medical and laboratory practices. AMS is therefore an ideal detection system for this example. The following table summarizes the foregoing protocol design. Test Subject 60 kg, 5 foot 6 inches female 7 Tracer Dose* 2.4 mg Cyclophosphamide 0.292 mg Paclitaxel Target Tissue Breast tumor biopsy Test Cocktail Single intravenous administration; 10 minute infusion Distribution Time 3 hours Detection System Accelerator Mass Spectrometry (AMS) 1C Isotope Tracer 50 nCi cyclophosphamide and 50 nCi paclitaxel or 100 nCi total Test Report % of administered dose or metabolites in Target Tissue Dose or metabolite quantity/mg protein in Target Tissue * therapeutic dose is 2,400 mg for cyclophosphamide and 292.25 mg for paclitaxel. Cyclophosphamide Cyclophosphamide is in a class of drugs known as alkylating agents; it slows or stops the growth of cancer cells in the body. The length of treatment depends on how well the body 5 responds to drug, and the type of cancer. The drug can be taken by mouth in tablet form or be given by injection into a vein. In order to work, cyclophosphamide is first converted by the liver into two chemicals, acrolein and phosphoramide. Acrolein and phosphoramide are the active compounds, and they slow the growth of cancer cells by interfering with the actions of deoxyribonucleic acid (DNA) within the cancerous cells. It is, therefore, referred to as a 10 cytotoxic drug. Unfortunately, normal cells also are affected, and this results in serious side effects. Cyclophosphamide also suppresses the immune system and is also referred to as immunosuppressive. The usual initial dose for treatment of adults and children is 40-50 mg/kg administered intravenously over 3-5 days in divided doses. The usual oral dose is 1-5 mg/kg daily. 15 Subsequent maintenance doses are adjusted based on the response of the tumor to treatment and the side effects. An intravenous dose preparation for a 60 kg Test Subject contains 40 mg/kg x 60 kg = 2,400 mg cyclophosphamide. Cyclophosphamide is biotransformed principally in the liver to active alkylating metabolites by a mixed function microsomal oxidase system. These metabolites interfere with 20 the growth of susceptible rapidly proliferating malignant cells. Cyclophosphamide is well absorbed after oral administration with a bioavailability greater than 75%. The unchanged drug has an elimination half-life of 3 to 12 hours. It is eliminated primarily in the form of metabolites, but from 5% to 25% of the dose is excreted in urine as unchanged drug. Several cytotoxic and noncytotoxic metabolites have been identified in urine and in plasma. 25 Concentrations of metabolites reach a maximum in plasma 2 to 3 hours after an intravenous dose. Plasma protein binding of unchanged drug is low but some metabolites are bound to an extent greater than 60%. It has not been demonstrated that any single metabolite is responsible for either the therapeutic or toxic effects of cyclophosphamide. Although elevated levels of 8 metabolites of cyclophosphamide have been observed in patients with renal failure, increased clinical toxicity in such patients has not been demonstrated. Cyclophosphamide Tracer Dose Calculations Formula C 7

H

1 5

N

2 C2O 2 P 5 Mol. weight 261.085 g/mol PN H i 0 Route a single intravenous administration Test Subject 60 kg individual 10 Dose 0.04 mg/kg Tracer Dose 0.04 mg/kg x 60 kg = 2.4 mg 14 C Isotope 50 nCi Specific activity Molar equivalent: (2.4 mg cyclophosphamide) x (1 mmol / 261.085 mg) 15 = 0.009192 mmol = 9.192 umol 14 C Isotope tracer: 50 nCi 50 nCi /9.192 umol = 5.43 nCi /umol Source: Cyclophosphamide, (American Radiolabeled Chemicals, St. Louis, MO) 10 pCi M.W. 20 261.085 Specific Activity 50-100 mCi/mmol is diluted to 0.1 mCi/ml Paclitaxel Paclitaxel is widely used as a chemotherapy drug in the treatment of various malignancies, including breast, ovarian, and lung cancers. Paclitaxel is a naturally occurring lipophilic drug that was originally extracted from the pacific yew tree Taxus brevifolia. The 25 drug interferes with microtubule function and results in primary and postmitotic GI arrest in smooth muscle cells, thereby inhibiting proliferation of these cells without inducing apoptosis, or cell death. For the adjuvant treatment of node-positive breast cancer, the recommended regimen is paclitaxel, at a dose of 175 mg/m2 intravenously over 3 hours every 3 weeks for four courses. 30 An intravenous dose preparation for a 60 kg Test Subject contains 175 mg/m 2 x 1.67 m 2 292.25 mg paclitaxel. Paclitaxel must be diluted prior to infusion. Paclitaxel administration is recommended to be diluted to a concentration of 0.3 to 1.2 mg/mL. The solutions are physically and 9 chemically stable for up to 27 hours at ambient temperature (approximately 25 'C) and room lighting conditions. Paclitaxel Tracer Dose Calculations The Tracer Dose in this protocol is intended for iv administration in a 60 kg, 5 foot 6 5 inches Test Subject. The calculated Body Surface Area for medication dosing using the 2 Mosteller formula is 1.67 m2 Formula

C

47

H

5 1

NO

14 Mol. weight 853.906 g/mol KC 6H0 10 Route a single intravenous administration Test Subject 60 kg individual, 1.67 m 2 Dose 0.175 mg/m 2 Tracer Dose 0.175 mg/m2 x 1.67 m 2 = 0.292 mg 14C Isotope 50 nCi 15 Specific activity Molar equivalent: (0.292 mg paclitaxel) x (1 mmol / 853.906 mg) = 0.000342 mmol = 0.342 umol

'

4 C Isotope tracer: 50 nCi 50 nCi /0.342 umol 20 = 146.1 nCi / umol Source: Taxol (paclitaxel), [2-benzoyl ring-' 4 C(U)] (American Radiolabeled Chemicals, St. Louis, MO) 10 tCi M.W. 853.9 Specific Activity 50-100 mCi/mmol is diluted to 0.1 mCi/ml The Target Tissue in this example was breast tissue collected during preliminary diagnosis of malignancy. Once a breast biopsy is recommended after an abnormal 25 mammogram finding, the Test Subject undergoes a minimally invasive alternative to surgery, known as a needle biopsy. The biopsy procedure takes a few minutes and no stitches are required. A breast biopsy is performed to remove a sample of breast tissue. The tissue is then studied by a pathologist under a microscope to determine the presence or absence of 30 malignancy. Several methods of breast biopsy now exist. The most appropriate method of biopsy for a patient depends upon a variety of factors, including the size, location, appearance and characteristics of the breast abnormality. These methods provide sufficient amount of Target Tissue for further histologic analysis by a pathologist and quantification of Tracer Dose. 10 In this protocol, a core needle biopsy was performed to obtain Target Tissue. A core needle biopsy is a percutaneous procedure that involves removing small samples of breast tissue using a hollow "core" needle. Three to six needle insertions are needed to obtain an adequate sample of tissue. Typically, each insertion removes samples approximately 0.75 5 inches long (approximately 2.0 centimeters) and 0.0625 inches (approximately 0.16 centimeters) in diameter. The volume of the removed sample in each needle insertion is approximately 0.04 cm 3 with a mass approximately 40 mg. This provides material for more than 5 separate analytical measurements using 40 mg of Target Tissue. Sample collected from one insertion is snap frozen in liquid nitrogen and kept frozen until further analysis. The rest 10 of the samples collected are sent to the pathology laboratory to determine if a breast lump is cancerous (malignant) or noncancerous (benign). Single needle insertion = 2.0 cm length x 0.16 cm diameter = 0.04 cm 3 Single needle insertion = approx. 40 mg Target Tissue 40 mg Target Tissue = At least 5 separate analytical measurements 15 The procedure for detection of each Tracer Dose and its active metabolites in Target Tissue is a two step process: (1) Tracer Dose and metabolites are separated into discrete fractions by High Performance Liquid Chromatography (HPLC), (2) the amount of 4 C isotope tracer in each fraction is determined by Accelerator Mass Spectrometry (AMS). Accelerator Mass Spectrometry (AMS) is the platform of choice when extreme sensitivity is required. In 20 this example, AMS quantifies the amount of radiocarbon-labeled Tracer Dose in Target Tissue with attomole (1048 M) sensitivity. AMS traces very low doses of compounds using extremely low radiation (<100 nanoCurie) in Test Subjects. Absorption, metabolism, distribution, binding, and elimination are all quantifiable with high precision after appropriate sample preparation. 25 (1) Fractionation by HPLC The major metabolites of paclitaxel in human plasma are 6-alpha-hydroxypaclitaxel; 3'-p-hydroxypaclitaxel; 6-alpha, 3'-p-dihydroxypaclitaxel; 7-epipaclitaxel. The major metabolites of cyclophosphamide in human plasma are 4-hydroxyphosphamide (OH-CP) and carboxyethylphosphamide (CEPM). There are some different metabolites found in human 30 urine: cyclophosphamide; N-dechlorophosphamide (DCL-CP); 4-keto cyclophosphamide (4KetoCP); and carboxy cyclophosphamide (CarboxyCP). Customized protocols permit separation of parent and/or metabolites of paclitaxel and cyclophosphamide. See example below for separation of parent compounds of paclitaxel and cyclophosphamide (J. Pharm. Biomed. Anal. 1;39(1-2):170-6, 2005). 35 Procedure a) Target Tissue is removed from the freezer and allowed to thaw over ice. A portion weighing 20-30 mg is isolated and the wet weight recorded. A 20% 11 weight/volume (w/v) homogenate is prepared in bovine serum albumin (BSA, 40 g/L) in water. b) Homogenates (0.1 mL) are extracted by ethyl acetate (1 mL). c) A 4.6 mm x 250 mm octadecyl silicone (ODS) column and high performance 5 liquid chromatography (HPLC) system is used to separate the fractions using a gradient protocol consisting of acetonitrile-deionized water (J. Pharm. Biomed. Anal. Sep 1;39(1-2):170-6, 2005). d) Individual fractions are collected and further analyzed by AMS. (2) Detection by Accelerator Mass Spectrometry (AMS) 10 Accelerator Mass Spectrometry (AMS) is a sensitive instrument essential for detecting extremely low levels of radiocarbon (1 4 C) in small very samples. The natural 1 4 C content in 20 ul human plasma or 5 mg tissue is approximately 105 x 10-18 (attomole) 14 C, representing less than 1 decay of 1 4 C per hour. This natural level of 1 4 C is not detectable by other instruments, yet is easily quantified to better than 1% precision in less than a minute by AMS 15 instrumentation. AMS can quantify even lower levels of 14C with exceptional sensitivity and precisions. The AMS limit of quantification (LOQ) for 14C is <200 zeptomole (1021 mol) in 20 pl human plasma or 5 mg tissue (BioTechniques 38:S25-S29). AMS is a type of tandem isotope ratio mass spectrometry in which a low energy (tens of keV) beam of negative atomic and small molecular ions is mass analyzed to 1 AMU 20 resolution (mass 14, for example). These ions are then attracted to a gas or solid foil collision cell that is held at very high positive potential (0.5-10 megaVolts). In passing through the foil or gas, two or more electrons are knocked from the atomic or molecular ion, making them positive in charge. These positive ions then accelerate away from the positive potential to a second mass analyzer where an abundant charge state (4* in the case of a 7MV dissociation) is 25 selected. The loss of 4 or more electrons (to the 3* charge state) in the collision cell destroys all molecules, leaving only nuclear ions at relatively high energies (20-100 MeV) that can be individually and uniquely identified by several properties of their interaction with detectors. The core "tricks" of AMS are molecular dissociation to remove molecular isobars and ion identification to distinguish nuclear isobars. Beyond these two fundamental properties of 30 AMS, special tricks unique to each element have been developed. In this case, 14N is separated from 1 4 C in the ion source because nitrogen does not make a negative ion. The sensitivity and specificity of AMS enables pharmacokinetic and distribution studies of Tracer Doses of 1 4 C-labeled chemical agents to Target Tissue of Test Subjects. Any excess 1 4 C isotope above the well-known and measurable pre-dose levels is attributed to the 35 Tracer Dose and its metabolites. AMS sensitivity drastically reduces the chemical dose exposure to 100-1000x below therapeutic levels and the 1 4 C isotope tracer exposure to less than 100 nanoCurie (nCi) levels. Procedure: 12 a) The volume of each HPLC fraction is placed in individual quartz tubes. Carbon carrier (1 mg) from a stock supply of tributyrin is added for efficient graphitization. Quartz tubes are placed in a centrifuge and their contents dried under vacuum. A standard calculation is used to correct signal introduced by the carbon carrier. All 5 samples are individually combusted to CO 2 and the CO 2 is reduced to graphite over a suitable catalyst such as iron (Anal. Chem. 75,2192-2196, 2003). b) AMS measurements are performed on the graphite (Davis, Nucl. Instrum. Methods B40/41. 705-708, 1989; and Proctor, Nucl. Instrum. Methods B40/41. 727 730, 1989). 10 c) Controls - A plasma sample is collected prior to administration of the Test Cocktail to ensure that the Test Subject does not possess 14C isotope tracer levels above natural levels. ANU sucrose, with an activity 1.508 times the 14C activity of 1950 carbon is used as the analytical standard. Test Report 15 Analytical results from the AMS instrument are analyzed and a report is produced ranking the delivery of each chemical agent in the Test Cocktail according to the amount detected in the Target Tissue. The Test Report may be presented in several different formats. 1 4 C Tracer Isotope Calculations AMS provides a direct measure of the 1 4 C/1 2 C ratio in HPLC fractions. This 1 4

C/'

2 C 20 ratio is known as the Fraction Modem (FM). The FM value is used to calculate the amount of 4C isotope tracer in each HPLC fraction. A series of calculations is shown below for quantitation of Tracer Doses in the protocol. Step 1 - Conversion of AMS results to 1 4 C isotope amount (attomole) attomole 1 4 CHPLC fraction 25 (FM) x (97.9 attomol l4!C x (mg Ccarrier + mg CHPLC fraction) -Ccarrier) mg Ctotal Step 2 - HPLC fractions are essentially carbon free since the HPLC mobile phase consists of volatile components that are all removed during vacuum centrifugation. Therefore, the mg CHPLC fraction is considered to be zero. The above equation is reduced to: 30 attomole 14CHPLC fraction (FM) x (97.9 attomol 14C) x (mg Ccarrier) - (1 4 Ccarier) mg Carrier Step 3 - A common carbon carrier is tributyrin (80 uL, 20 mg/mL). The total carbon mass added as carrier is 0.9536 mg C. The Fraction Modems is 0.1014. 35 attomole 1 4 Ccarrier= (0.1014 FM) x (97.9 attomol 14C) x (0.9536 mg Ccarrier) mg Ccarrier attomole 1 4 Ccarrier = 9.466 attomole 14C in tributyrin carrier Step 4 - In this example, an HPLC fraction with carbon carrier added has a Fraction Modem of 0.2345, then: 13 attomole 14 CHPLC fraction (0.2345) x (97.9 attomol 1 4 C) x (0.9536 mg Ccarier) - (9.466 attomol 14 Ccarrier) mg Ccarrier attomole 14 CHPLC fraction = 12.426 attomole 1 4 C in HPLC fraction 5 Step 5 - The amount of 1 4 C in attomoles is converted to fCi 1 4 C. fCi 14 CHPLC fraction = (12.426 attomole 1C) x (6.11 fCi) = (97.9 attomole) fCi 14 CHPLCfraction= 0.7755 fCi 1 4 C 10 Step 6(a) - If this HPLC fraction represents cyclophosphamide, then the amount of 14C cyclophosphamide in the HPLC fraction is calculated using the specific activity of the administered 1 4 C-cyclophosphamide Tracer Dose. pmol 14C-cyclophosphamide = (0.7755 fCi 4 C) x (1 nCi) x (1 umol) x (106 pmol) 15 (106 fCi) (5.43 nCi) (1 umol) pmol 14C-cyclophosphamide = 0.1428 pmol 14C-cyclophosphamide % administered dose in HPLC fraction = 0.1428 pmol 1 4 C-cyclophosphamide in HPLC fraction 9.192 umol 14C-cyclophosphamide in administered dose 20 % administered dose in HPLC fraction = 0.00000155 % Step 6(b) - If this HPLC fraction represents paclitaxel, then the amount of 14C-paclitaxel in the HPLC fraction is quantified using the specific activity of the administered 14C-paclitaxel Tracer Dose. pmol 1 4 C-paclitaxel = 25 (0.7755 fCi 14C) x (1 nCi) x (1 umol) x (10' fmol) (106 fCi) (146.1 nCi) (1 umol) pmol 14C-paclitaxel = 5.308 fmol 1 4 C-paclitaxel % administered dose in HPLC fraction = 5.308 fmol 14C-paclitaxel in HPLC fraction 30 0.342 umol 14C-paclitaxel in administered dose % administered dose in HPLC = 0.00000155 % The calculations above are applied to all fractions collected by HPLC. The specific activity of the parent molecule is used to calculate the amount of metabolites present in each fraction. For simplicity, the term 'equivalent' is used to signify that specific activity of the 35 parent molecule is used for calculating the amount of metabolites. paclitaxel cyclophosphamide equivalent 6-alpha-hydroxypaclitaxel equivalent 4-hydroxyphosphamide equivalent 3'-p-hydroxypaclitaxel equivalent carboxyethylphosphamide Substantial inter-individual variability exists in the pharmacokinetic and metabolism of 40 chemotherapeutic agents. Individuals may obtain different paclitaxel plasma concentrations 14 after fixed doses of paclitaxel. A 4-5-fold difference in paclitaxel area under the concentration-time curve (AUC) was reported after a fixed-dose administration (J. Clin. Oncol. 15:317-29, 1997). Target Tissue guided selection of chemotherapeutic agents and optimization of dosing on an individualized basis can assist in attaining desired treatment 5 outcome. The process described permits individualized determination of the delivery of therapeutic compounds. Data are presented, for example, for two chemotherapy compounds, cyclophosphamide and paclitaxel, to Target Tissue. Quantitative distribution of Tracer Doses of cyclophosphamide and paclitaxel co-administered in a Test Cocktail can serve to forecast 10 delivery of these usually toxic agents doses into breast tumor. Such predictions are difficult to make due the large amount of variability amongst individuals. However, distribution measurements improve the accuracy with which the effectiveness of each chemotherapy agent is predicted. Alterations in drug metabolism are particularly important in cancer patients, who are 15 typically undergoing multi-drug therapy and/or may suffer liver disease due to drug toxicity or liver metastasis. Several metabolic enzymes may be either up or down-regulated under these conditions. In addition, several factors can influence drug metabolism outcomes, including, for example, genetic factors, disease state, age, diet, and physiological status. Example 1 20 This example illustrates the quantification of ' 4 C-CPT 377 distribution into multiple tissues in mice by accelerator mass spectrometry (AMS). Diagnoses of Type II diabetes and complications associated with diabetes have risen to epidemic proportions in the last decade on a global scale. It is currently estimated that more than 150 million people suffer from Type II diabetes, with the characteristic hallmarks of insulin resistance in peripheral tissues, 25 hyperglycemia, and pancreatic B-cell dysfunction. Although several marketed therapies are available, many have significant side effects or become ineffective for patients who receive these treatments. In addition, as disease progression occurs, the need for additional therapeutic intervention and supplemental insulin treatment is necessary. As such, considerable effort towards the development of therapeutic agents with novel mechanisms of action continues, 30 with the hope to reduce the occurrence, progression and complications of this debilitating disease. One such approach is to obtain effective insulin sensitizing agents by targeting the insulin signaling pathway directly, and more specifically through Protein Tyrosine Phosphatase lB (PTP1B). PTP1B is a ubiquitously expressed, non-receptor enzyme that negatively 35 regulates insulin and leptin signaling in vivo. PTPlB knockout (PTPlB-/-) mice are healthy and show increased insulin sensitivity, improved glucose tolerance and resistance to weight gain when fed a high fat diet. In addition, studies from tissue specific attenuation of PTP1B suggest a role for this enzyme in certain insulin sensitive tissues. Therefore, a small molecule 15 reversible, competitive inhibitor of this well-validated target would provide a positive therapeutic benefit in treating Type II diabetes. An early stage highly potent lead molecule from a chemical series designed to specifically inhibit PTP1B, was found to be efficacious in a mouse model of diabetes and 5 obesity (C57Bl/6J ob/ob). Reductions in daily plasma glucose levels and positive effects in a standard glucose tolerance test were seen after 3-5 days of a once daily oral treatment with this compound (CPT 377). Traditional pharmacokinetic analysis revealed the oral bioavailability of the molecule to be low (<6%) with a serum half-life of 1.5 hours. Based on the reported role of PTP1B in peripheral tissues of insulin sensitivity, an AMS study was designed to 10 delineate if the positive efficacy results of this compound were due to residence time of the compound in specific tissues. This information would help establish the potential correlation between tissue exposure and effect. An experiment was set up with a Test Article being an oral and iv formulation of a lightly-labeled 1 4 C-CPT 377 (small molecule). The specific activity was 0.12 mCi/mmol, 15 storage conditions were 2-8*C, the oral dose was 20mg/kg, and iv dose was 4mg/kg in 38 male mice strain CD-1 (albino Swiss equivalent, Charles River Laboratories, Hollister, CA) weighing about 20 g each. The animals were housed individually at room temperature and 50+20% relative humidity, in rooms with at least ten room air changes per hour and fed Laboratory Rodent Diet with water provided ad libitum. The photoperiod was diurnal; 12 20 hours light, 12 hours dark and the animals were first acclimated for four days. Approximately 1 mL whole blood was collected by cardiac puncture. Plasma and erythrocytes were separated at the time of collection by centrifugation at approximately 2800 RPM for 15 minutes at 4 *C. Plasma was removed and stored in Fisher brand glass threaded vials. 25 Tissue samples from various organs were obtained by surgical removal and the wet weight recorded. A 20% weight/volume (w/v) homogenate was prepared in 20 mM potassium phosphate dibasic (KH 2

PO

4 ), pH 7.4. The total carbon concentration in each sample (75-100 RL) was measured by freezing the sample over liquid nitrogen in individual Costech tin capsules (Ventura, CA) followed by 30 overnight lyophilization. Each capsule was then placed inside a second tin capsule, rolled into a ball and analyzed for total carbon concentration using a Carlo-Erba carbon analyzer (Pella, Am. Lab. 22:116-25, 1990). Calculations were based on sample weights measured to four decimal places. Mass % Tissue Sample (ug) Carbon Plasma 1 2.1 4656 3.75 Plasma 2 2.2 3679 4.10 16 Plasma 3 2.3 4121 3.31 Plasma 4 8.1 3313 4.80 Plasma 5 8.2 2724 5.72 Plasma 6 8.3 3169 6.60 Plasma 7 11.1 4667 4.23 Plasma 8 11.2 2610 5.21 Plasma 9 11.3 4196 4.79 Plasma 10 14.1 4664 4.42 Plasma 11 14.3 5617 3.87 Plasma 12 17.1 5013 3.32 Plasma 13 17.2 5403 4.02 Plasma 14 17.3 3886 4.27 Plasma average (% Carbon) 4.46 Brain 1 42.1 2972 5.13 Brain 2 42.3 6975 3.11 Brain average (% Carbon) 4.12 Liver 1 48.1 5293 3.05 Liver 2 48.2 7029 2.74 Liver average (% Carbon) 2.90 Fat 1 80.1 4654 0.76 Fat 2 80.2 3324 0.98 Fat average (% Carbon) 0.87 Muscle 1 89.1 5106 2.32 Muscle 2 89.2 4821 2.88 Muscle 3 89.3 6366 2.74 Muscle average (% Carbon) 1 1 2.65 Approximately 20 uL tissue homogenate was placed in individual quartz tubes and graphitized (Anal. Chem. 75:2192-2196, 2003). In this process, all samples were combusted to

CO

2 and the CO 2 was reduced to graphite over a suitable catalyst, such as iron. 97.9 amol radiocarbon radiocarbonsampe = fraction modem x x carbonsample mg carbon 5 AMS measurements were performed as previously described (Davis, Nucl. Instrum. Methods. B40/41. 705-708, 1989 and; Proctor, Nucl. Instrum. Methods. B40/41. 727-730, 1989). Predose plasma and urine was collected from one mouse to ensure the animals did not 17 possess 14C concentrations above natural levels. An additional six samples are available to use for 14 C control measurements. ANU sucrose, with an activity 1.508 times the 14C activity of 1950 carbon was used as the analytical standard. The time of dose administration was referred to as To. All subsequent time points, 5 given in minutes, were referred to as 'minutes since dose'. The 1 4 C calculations were "modems" to four decimal places at 3% precision. Modems can be thought of as a measure of 1 4 C/1 2 C ratio. The 1 4 C concentration was calculated from the modem values by using a reference number for total carbon (1C) in various sample types. Once 1 4 C concentration was determined, the 14C-dose concentration was calculated using the specific activity of the 10 administered dose. Plasma, urine, bile fmol 14 C-dose/uL Tissue fmol 14 C-dose/g tissue 13.56 dpm _ 6.11 fCi 97.9 amol radiocarbon 1 modern = = =____ _________ g carbon mg carbon mg carbon 1 mol 14C = 14 g 1 4 C fCi 1 4 C-compound = 1 fmol 14C-compound 15 g 14C = 1 g 14C-compound The compound (iv administered) was cleared rapidly from plasma within 25 minutes and over 95% cleared within 3 hours of dosing. With oral administration, Cmax was reached at approximately 1.5-2 hours. There was an approximately 1 00x difference between oral and iv concentration at Cmax. There was very little or no distribution of iv or oral dose in brain. 20 With an iv dose in heart tissue, there was a rapid drop in concentration by 25 minutes post dose and a return close to baseline by approximately 6 hours. The oral dose saw a very rapid drop in concentration and return to baseline within 25 minutes. The difference in maximum concentration between iv and oral routes is less marked in heart. In the liver/kidneys/Testes/Bone, (iv dose) the maximum signal was observed at 5 minutes with rapid 25 drop in compound concentration within 25 minutes and return to baseline by 6 hours post dose. With the oral dose, there was a rapid drop in concentration and return close to baseline within 6 hours. The iv dose in muscle showed the maximum signal was observed at 5 minutes with slower drop in dose concentration than other tissues. It returned to baseline by 6 hours post 30 dose. The oral dose showed a slower drop in concentration than most other tissues. It returned to baseline by 6 hours post dose. It should be noted that there was little or no difference in concentration between iv and oral routes of administration. In fat tissue, the iv dose maximum signal was observed at 5 minutes with rapid drop in compound concentration within 25 minutes and return to baseline by 6 hours post dose. 35 Similarly, with the oral dose a rapid drop in concentration was observed and a return close to baseline within 6 hours. 18 Compound CPT 377 was an early lead compound in a series of potent, well characterized inhibitors of PTP 1 B. It showed consistent oral efficacy in an ob/ob mouse model of diabetes and obesity. Results from a traditional rat pharmacokinetics study using CPT 377, revealed the compound to have a less than optimal pharmacokinetics profile with a low 5 bioavailability (<6%). In order to further understand the disconnect between the efficacy and bioavailability of 377 and to determine if the compound had enough residence time in insulin sensitive tissues to provide an effect, an ultra-sensitive method for analyzing the distribution profile of low levels of compound was necessary. CPT 377 was easily labeled with "C in house at a significant cost and time savings and was dosed to mice by oral gavage or 10 intraveneous, and samples collected at various time points through 24 hours. The tissue distribution profile and bioavailability determination from the AMS study provided the information needed to focus on alternate chemical modifications in order to benefit from the potency, avoid rapid elimination, and focus on specific target tissues of interest. The study demonstrated the utility of AMS as an ultra-sensitive detection platform for 15 quantifying drug kinetics and distribution in small animals. AMS analytical methods require no additional method development, including chromatographic or mass spectrometric optimization for specific chemical structures. This makes AMS even more suitable for early phase drug development where analytical resources or methods may not be readily available, or where early preclinical characterization may help select the most suitable candidate within a 20 series of similar compounds. As such, AMS-based protocols permit assessment of pharmacokinetic and ADME characteristics of new drugs earlier in development, with minimum bioanalytical contribution and with tremendous sensitivity. Example 2 This example shows a quantification of "C-folic acid distribution in red blood cells and 25 more particularly a quantification and compartmental modeling of "'C-folic acid distribution in red blood cells after a single oral administration in a human subject. Recent advances in mass spectrometry and the availability of stable isotope-labeled compounds have made kinetic tracing of nutrients a powerful tool for understanding nutrient metabolism in humans. The quality of the data generated by such studies was dictated to a 30 large degree by limitations due to sample preparation and analysis. Accelerator mass spectrometry (AMS) provided an alternative approach to study tracer kinetics by measuring "C-labeled in human samples at attomole concentrations. Folate plays an important role in the etiology of many diseases. This relationship encompasses pathology of multiple organs occurring at different stages during development 35 (Wadsworth, Canadian Journal Of Public Health-Revue Canadienne De Sante Publique 88:304-304, 1997). This increases the challenge faced by researchers aiming to identify the mechanisms underlying many of these pathologies. Many biological processes are involved in the coordination of folate homeostasis and metabolism. Folate balance is also influenced to a great degree by environmental factors such as dietary intake. The interplay between biological 19 regulation determined at the genetic level (genotype), and environmental factors, produce the observed state of the organism (phenotype). Kinetic modeling can identify the metabolic phenotype of the organism. Genetic and/or environmental factors that contribute to the observed phenotype can then be determined. This will help to identify the etiology of folate 5 related diseases and to develop rational therapies. No reliable method has ever been developed to model folate kinetics in humans under physiological conditions. In this study, the extreme sensitivity of accelerator mass spectrometry permitted the modeling of folate kinetics in a healthy adult male for the first time. The instrument will be useful for scientists as a biomedical research tool and for clinicians as a diagnostic tool. 10 Although accelerator mass spectrometry (AMS) has been used in the past for environmental, geologic and archeological research, only recently has it been applied to the biomedical sciences. Animal and human toxicological research has yielded novel insights into interactions between macromolecules and toxicants (Creek, Carcinogenesis 18:2421-2427, 1997; Kautiainen, Chemico-Biological Interactions 106:109-121, 1997; Turteltaub, Mutation 15 Research-Fundamental and Molecular Mechanisms OfMutagenesis 3 76:243-252, 1997; White, Chemico-Biological Interactions 106, 149-160, 1997)). A recent investigation revealed femtomole quantities of 1 4 C0 2 in the expired breath of humans after ingestion of 40 nCi 14C triolein (Stenstrom, Appl. Radiat. Isot. Apr.; 47(4):417-22, 1996). We are not aware of biomedical AMS research that has been attempted with the sampling density and quantitative 20 precision required by this study. These experimental protocols were based on current understanding of folate kinetics in humans. Important parameters, such as dosing level and sampling schedule, were planned to aid the mathematical modeling of the data. The materials used in this study included L-Glutamic acid [14C (U)] (250 mCi/mmol) (Moravek Biochemicals), folic acid, 5-methyltetrahydrofolic acid and folinic acid standards 25 (Sigma), acetonitrile and water (OPTIMA grade from Fisher). Pteroyl-[14C(U)]-glutamic acid (folic acid) was synthesized according to the method of Plante (Plante, Methods in Enzymology 66, 533-5, 1980) with some modifications (Clifford, Adv. Exp. Med. Biol.445:239-51, 1998). The concentration was measured by UV-VIS spectrometry after separation by reverse-phase HPLC. Radioactivity was measured by scintillation counting and specific activity calculated at 30 1.24 mCi/mmol. The specific activity of the final product was lower than the specific activity of the 14C-glutamic acid because of dilution with non-labeled glutamic acid. 14C-Folic acid was administered to an healthy informed male volunteer weighing 85 kg consumed 50 mL water orally containing 35 tg "'C-folic acid (80 nmol, specific activity 1.24 mCi/mmol) in the morning followed by a light breakfast. Residual dose in the container was 35 rinsed with approximately 100 mL water and ingested. All procedures were approved by Institutional Review Boards at the University of California, Davis and Lawrence Livermore National Laboratory. A small volume of sample required for AMS measurement allowed frequent sampling of blood in the first 24 hours of the study. The early dynamic stage of folate absorption and distribution was determined by collection of-8 mL blood at 10 minute 20 intervals postdose. Sampling frequency was reduced to 20-minute intervals after one hour and to 30-minute intervals after 3 hours. A total of 24 samples were collected in the first 24 hours. Sampling of - 24 mL blood continued for the next 200 days at weekly to monthly intervals. Red blood cells were isolated by collecting whole blood prior to administration of the 5 oral dose and post-dose through a catheter for the first week and by venupuncture thereafter. Red blood cells were separated from whole blood within an hour after collection by centrifugation at 3,500 RPM for 5 minutes. Plasma was removed, the buffy-coat discarded and red blood cells washed four times with an approximately equal volume of isotonic buffer (150 mM sodium chloride, 10 mM potassium phosphate, pH 7.2, 0.05 mM EDTA, 2 % ascorbate). 10 Plasma and red blood cells were stored at -20'C for AMS, folate and carbon measurements. The total carbon concentration in each sample (75-100 pLL) was measured by freezing the sample over liquid nitrogen in individual Costech tin capsules (Ventura, CA) followed by overnight lyophilization. Each capsule was then placed inside a second tin capsule, rolled into a ball and analyzed for total carbon concentration using a Carlo-Erba carbon analyzer (Pella, 15 1990). Calculations were based on sample weights measured to four decimal places. Evidence that the subject was in steady-state was provided by the carbon concentration in the blood and carbon losses in urine and feces. The carbon concentration of each sample was measured using a modified protocol that simplified pipetting and packaging samples prior to analysis. The results, shown in Table 3 below, provided evidence that carbon homeostasis 20 was maintained over the 200-day study period. Table 3. Carbon in samples collected over a 200-day period Sample Mean ± Standard deviation Plasma (n=59) 4.42 ± 0.14 g/dL Red blood cells (n=59) 0.60 ± 0.04 g/g hemoglobin Urine (n=63) 9.68 ± 1.9 g/day Feces (n=39) 9.09 ± 1.5 g/day AMS measurements were performed as previously described (Davis, Nucl. Instrum. Methods B40/41:705-708, 1989 and; Proctor, Nucl. Instrum. Methods B40/41:727-730, 1989). Briefly, a beam of C- ions was produced by bombarding the cool, cesiated surface of a graphite sample with about 5 keV Cs+ ions. The C- beam produced by the sputtering of the sample by 25 the Cs* beam was accelerated, focused, and mass analyzed into mass 14, and 13 amu beams. These beams were then accelerated to high energy in sequence by successively changing their energy as they passed through the mass analyzer so that they were on the correct trajectory for transmission into a 1.5SDH-1 Pelletron accelerator. The energy changing sequencer was adjusted about 10 times a second so that about 1 part in 103 of the mass 13 30 beam, and 99.9% of the mass 14 beam passed into the accelerator keeping average accelerated and beam loading currents very low and X-rays produced directly or indirectly by high energy ions also very low. The beam of negative ions was about 500 keV in energy when it reached a 21 region of relatively high argon gas pressure, the stripper canal, located in the high voltage terminal of the 1.5SDH- 1 Pelletron. The fast moving negative ions lose electrons and become predominantly C+ ions when passing through the stripper canal. Also critical to the AMS process, negative molecular ions, such as CH- and CH- 2 , are broken into C*" and H+ ions by the 5 argon gas. This eliminates interferences that might be caused by molecular ions when counting 1 4 C+ ions later in the system. The charge 1 positive ions are accelerated from the high voltage terminal to ground gaining an additional 0.5 MeV in energy for a total of about 1 MeV. The ions are magnetically deflected and focused at 900 by the analyzing magnet so that the pulses of 3 C+ are separated 10 from the 1 4 C+ and measured in a Faraday cage. The 14C+ ions and a small number of 1 2 C+ or 1 3 C* ions from the molecular breakup in the terminal that have changed charge state at exactly the right places in the accelerating tube so that their energy is enough greater than the 1 4 C* ions. to be transmitted around the 90* magnet are then allowed to pass into a 900 electrostatic spherical analyzer (ESA) which deflects the faster 12 C and 1 4 C* ions away from the 1 4 C* ion 15 beam path. The ESA also provides a final focusing so that the 1 4 C+ ions are transmitted to a solid-state detector where they are counted. By recording the 1 3 C current and 1 4 C counts as known and unknown samples are sputtered, the amount of 1 4 C present in a sample is determined to high accuracy. The dose of 14C-folic acid had a specific activity of 1.24 mCi/mmol. Exactly 441.4 pL 20 of the synthesized preparation was used for the dose with an activity of 222,000 DPM (5.028 DPM/tL). This corresponded to 100 nCi of activity and 80.6 nmol of folic acid. This amount (35.6 pg) was approximately equivalent to 1 /6h the current RDA for folic acid. The dose was mixed with ~ 150 mL water and ingested at 7:12 AM. The first blood sample was taken after 10 minutes the time of subsequent samples recorded in minutes after 25 the dose and expressed as days (after ingestion of dose). A 24-hour urine sample was collected before ingestion of the dose followed by 6-hour collections for the first day and 24-hour collections thereafter. All values were expressed at the end-point of each collection. A pre dose fecal sample was collected 24 hours before ingestion of the dose. The time of each collection was recorded and values for that sample are expressed at that time point. 30 The modem value for each sample was determined three times by accelerator mass spectrometry and the mean value used to calculate 1 4 C-folic acid concentration. The modem value, carbon concentration of each sample and specific activity of the dose were used to calculate the concentration of 1 4 C-folate in each sample. Since the identity of the folate molecule was not determined, this value was expressed as a molar quantity to eliminate 35 ambiguities between different folate molecules due to differences in molecular weight. The same approach was taken in expressing 1 4 C-folate concentration in urine and feces. These samples contained catabolites of folate but concentrations were expressed as moles of I 4

C

folate since one mole of catabolite was derived from exactly one mole of 1 4 C-folic acid. The equations for calculating 1 4 C-folate concentrations are shown below. 22 1 mol 4 C = 14 g 1 4 C 13.56 dpm 6.11 fCi 97.9 amol radiocarbon 1 modern === g carbon mg carbon mg carbon 1.24 fCi 1 4 C-folic acid = 1 fmol 1 4 C-folic acid 6.2741 x 10~4 g 1 4 C = 1 g 1 4 C-folic acid Calculation Example: Plasma (P1) 5 (1.4180 modems) - (1.0986 modems) = 0.3194 modems (adjusted) (0.3194 modems) x (97.9 amol 1 4 C/mg C) = 31.2692 amol 1 4 C/mg C (31.2692 amol 1 4 C/mg C) x (44.38 mg C/mL plasma) = 1,387.7 amol 14C/mL plasma (387.7 amol 1 4 C/mLplasma) x (14 ag 1 4 C/amol 14C) = 19,428.2 ag 14C/mL plasma (19,428.2 ag 1 4 C/mL) x (1 ag 14C-folic acid/6.2741 x 10~4 ag 14 C) 10 = 3.0965 x 10' ag 1 4 C-folic acid /mL (3.0965 x 107 ag 14C-folic acid /mL) x (1 amol 14C-folic acid/441.4 ag 1 4 C-folic acid) = 70,151 amol 1 4 C-folic acid/mL plasma = 70.1 fmol 1 4 C-folic acid/mL plasma Red blood cell 1 4 C-folate concentration was measured by AMS and the data expressed as 15 fmol 1 4 C-folate/gram hemoglobin. This approach eliminated differences in cell dilution after washing in buffer by normalizing folate values against the hemoglobin concentration. The data for the first eight days and for the entire study period are shown in the top and bottom of Figure 7, respectively. Samples collected in the first 24 hours possessed a small amount of 1 4 C above the background. However, this returned to background by 36 hours after the dose and 20 remained low until day 3. Samples collected after day 4 showed a rapid rise in red blood cell 1 4 C-folate concentration and reached a maximum of 1650 fmol 14C-folate/gram hemoglobin by day 19. Using the subject's hemoglobin concentration and reference values for total blood volume, the total amount of hemoglobin in circulation was estimated to be 6.6 g (13.3 g 25 hemoglobin/dL, 5 L blood). The total amount of 14 C-folate in red blood cells by day 19 was approximately 10.9 pmol 1 4 C-folate or about 0.014 % of the administered bolus. This value corresponded to 2.18 fmol 1 4 C-folate/mL blood, well above the detection limit of AMS. Figure 7 shows red blood cell 14 C-folate concentration for the first 8 days (top) and 200 days (bottom) postdose. The three-day delay before appearance of 14C-folate represented the 30 time required for 1 4 C-folate incorporation into cells in the marrow during maturation. Error bars represent ± 1 standard deviation of triplicate determinations of 14C by AMS. Folate was extracted from plasma using C 18 solid phase extraction cartridges and folate binding protein-affinity chromatography. The extracted folate molecules were separated by reverse-phase HPLC with detection at 292 nm. Folate from 100 pIL plasma did not produce a 35 detectable signal because the endogenous folate concentration was below the limit of detection. Folate standards were added to the plasma as internal standards before extraction to check for 23 recovery and to mark the retention of folate under the HPLC conditions. Folic acid (FA) and 5-methyltetrahydorfolate (5MTFA) standards were added since plasma folate was in the form of 5MTFA and the administered dose was in the form of FA. Folate standards, extraction buffers and HPLC solvents were screened by AMS to ensure there would be no 1c 5 contamination. Fractions were collected every minute, lyophilized and carbon carrier added prior to AMS measurement. Figure 8 shows an HPLC-AMS chromatogram of plasma sample collected one hour after ingestion of 1 4 C-folic acid. Absorption was monitored at 292 nm (solid line) and 14C concentration was measured by AMS and expressed as modems (dashed line). The large early 10 peak was due to ascorbic acid added to the sample to protect against oxidation. Folic acid (FA) and 5-methyltetrahydrofolate (5MTFA) standards were added to the plasma before extraction to check for recovery and to mark folate retention times. A compartmental model determines the transfer rates between body compartments by dividing the body into discrete pools. These pools may or may not represent actual organs 15 with biological correlates. The model was built on the simple schematic depicted in Figure 9. The simplest model consisted of a single body pool with input for the tracer and an output for all excretions. It was essential to know the amount of tracer administered. The bioavailability of the tracer, however, complicated quantification of the amount of tracer that actually entered the body pool. Figure 9 shows a simple model depicting the uptake and loss of tracer. 24 This problem has been traditionally resolved by intra-venous administration of a second tracer to determine the bioavailability and make the appropriate adjustments. The bioavailability of the tracer in this study was measured directly and no such adjustments were necessary. Once the tracer was absorbed it was distributed to various tissues and eventually 5 excreted from the body in the urine and feces. If the tracer could be lost through the lungs, expired air should also be collected and included as a route of excretion. Skin could also represent a route of excretion for some compounds. In this study, urine and feces were considered to be the major routes of folate excretion from the body (Krumdieck, American Journal of Clinical Nutrition 31:88-93, 1978). 10 Blood is an easily accessible pool in the body. Whole blood was sampled frequently and separated into plasma and red blood cells, each representing a discrete pool. Red blood cells represented a tissue pool that could be easily sampled. The main components of the model are shown in Figure 10 for folate distribution in humans. This schematic depicts major pools involved in folate metabolism. The amount of tracer was measured in all pools shown 15 with a solid line. The tissue pool, shown with a dotted line, was the only pool whose ' 4 C-folate concentration was not measured. Compartmental modeling allowed the distribution of the tracer into this pool to be determined by simple difference. Compartments that were measured are shown with solid lines, and the inaccessible tissue compartment is shown with a dotted line. 20 Once the main components of the model were identified, biologically relevant components were added. These components were based on current knowledge of folate metabolism in humans and animals. For example, 1 4 C-folic acid in the gut had to pass through intestinal cells before entering the plasma. An intestinal pool was added to account for this process. It was known that maturing red blood cells in the marrow incorporated folate before 25 being released into circulation (Bills, Blood 79:2273-80, 1992). A compartment representing the marrow was added as well as a delay element to provide the necessary delay observed experimentally (Strumia, Medical Times 96:1113-24, 1968). Since the red blood cell folate dynamics was governed by the cell's lifespan in circulation, a delay element was added to provide for this biological process. The 24-hour transit time of material through the 30 gastrointestinal tract also necessitated a delay element to be incorporated into this portion of the model. Compartmental modeling made certain assumptions regarding fluxes of the tracer between compartments. Initially, the model assumed that all fluxes were governed by first order processes, mathematically represented as flux(2,1)=k(2,1)*ql, where 1 was the donating 35 pool, 2 was the receiving pool, k(2,1) was the transfer coefficient between the pools, and qI was the concentration of the tracer. The flux, therefore, varied as the concentration in pool 1 changed. In most cases, the assumption that a first-order process governed flux was valid, however, the equation could be altered to meet the system's needs. 25 Another assumption made in the modeling was that the tracer paralleled the behavior of the endogenous compound being studied. Isotopically labeling folate permitted discrimination of the tracer from endogenous sources. The assumption was made that biological processes, such as absorption, protein binding and enzymatic conversion to metabolites, were not affected 5 by isotope labeling. In this study, 14C-labelling of folic acid ensured minimal or no isotope effects. In previous models of nutrient kinetics the ratio of labeled and nonlabeled nutrient was used to derive the specific activity of the tracer. This value was then used to model the nutrient dynamics between various compartments since the specific activity was a measure of tracer enrichment in that compartment. In this study, 14C-folate concentration alone was used to 10 build the compartmental model based upon the assumption that 1 4 C-folate dynamics represented folate dynamics. Since modeling treated body compartments as pools, the concentration data in plasma and red blood cells were converted to total amounts of 1 4 C-folate in each pool using reference values for plasma volume and total grams of hemoglobin in circulation. These amounts, as 15 well as cumulative 14C-folate excreted in the urine and feces, were associated with their respective pools in the model. All values, in femtomoles 14C-folate, were assigned a fractional standard deviation of 0.025 to represent the uncertainty in the AMS measurements. The time of collection for each sample was converted to hours after ingestion of the bolus. The gut compartment received a bolus of 8.0 x 10 7 fmol 1 4 C-folic acid at time zero. 20 The colon compartment received the experimentally measured portion of the bolus that was not absorbed equaling 9.39 x 106 fmol 1 4 C-folate. The colon to feces transfer contained a delay element of 24 hours. The marrow to red blood cells transfer contained a delay element of 100 hours while the degradation delay was one hour. Compartments were added to represent tissue folate distribution consisting of fast and slow turnover pools. The fast turnover tissue 25 contained a delay element of 20 hours. The model was generated using the SAAM II software package (SAAM Institute, University of Washington). The model, shown in Figure 11, was an expansion of the basic model shown earlier. The entire 80 nmol 1 4 C-folate bolus entered the gut compartment at time zero. Exactly 9.39 nmol 14C-folate of this was lost to the colon compartment and represented the portion of the 30 bolus that was not absorbed, as determined experimentally. The remainder of the bolus entered the intestine compartment. This pool represented the intestinal cells that absorbed folic acid and reduced and methylated it before passing it on to plasma (Whitehead, British Journal of Haematology 13:679-86, 1972). The plasma compartment distributed folate to urine and colon, for excretion, and to 35 tissues for storage. Two compartments represented tissue storage, a fast-turnover tissue pool and a slow-turnover tissue pool. The red blood cell pool received the tracer through a marrow compartment which was added to represent folate incorporation into maturing leukocytes before their release into circulation (Bills, Blood 79:2273-80, 1992). 26 Initial coefficients were assigned for each parameter based on a best guess. The differential equations were solved simultaneously using algorithms in the software package. The algorithm adjusted these coefficients with every iteration in order to minimize the sum squares of errors between predicted and experimental values in the four compartments where 5 data was available. Transfer coefficients were finally derived that satisfactorily predicted experimentally determined values. The earlier assumption that fluxes in the compartmental model were driven by first order processes was not applicable to transfer of the tracer to the red blood cell and urine pools. For simplification, it was assumed that 1 4 C-folate was incorporated into the red blood cell pool 10 in a discrete pulse. This assumption was incorporated into the model by forcing the transfer coefficient from plasma to marrow, k(9,3) to 0 after 24 hours. This allowed 1 4 C-folate to enter the marrow pool by a first-order process for the first 24 hours. The 0.030 hr- transfer coefficient within that period permitted 1 4 C-folate uptake into the marrow pool that reflected the red blood cell pool's influx of 1 4 C-folate determined experimentally. The model was 15 unable to predict the red blood cell 14C-folate data when this provision was removed. The red blood cell pool presented a special case in compartmental modeling since flux out of the red blood cell pool was governed by the lifespan of these cells in circulation. Once folate entered this pool, it became unavailable for removal due to polyglutamation, until the cells themselves were removed from circulation (Rothenberg, Blood 43:437-43, 1974; Ward, J. 20 Nutr. 120:476-84, 1990; Brown, Present knowledge in nutrition, 6th edition, 1990). This was evident in the approximately 100-day presence of 14C-folate in the red blood cell pool, closely matching the known lifespan these cells. This process was incorporated into the model by adjusting the transfer coefficient of the flux out of the red blood cell pool. This transfer coefficient was changed from 4 x 10-5 hr-1 to 0.00072 hr-' at 2300 hours. 25 Aged red blood cells were removed from circulation by a macrophage-mediated process that engulfed the cells and returned them to the spleen and liver for degradation (Rosse, Journal of Clinical Investigation 45:749-57, 1966). The degradation pool, q10, represented this process. Although the 1 4 C-folate in these cells was capable of being recycled, their removal to a dead-end degradation pool did not affect the model since the total amount of 30 14C-folate in the red blood cell pool represented just 0.0 14 % of the bolus. Urine output of 1 4 C-folate was not governed by first-order processes. Experimental data showed that there was a constant output of the tracer into the urine over the 42-day period. Since plasma 1 4 C-folate was highly dynamic in the first 24 hours, flux to the urine pool could not depend on plasma concentration. Output of 1 4 C-folate, as metabolites or catabolites, was 35 driven by a selective process from renal glomerulii. These included active transport mechanisms that concentrated 1 4 C-folate against a gradient (Das, Br. J. Haematol. 19:203-21, 1970; Henderson, J. Membr. Biol. 101:247-58, 1988; and Pristoupilova, Folia Haematol. Int. Mag. Klin. Morphol. Blutforsch 113:759-65, 1986). It was therefore unjustifiable to use first order flux equations to describe transfer of 1 4 C-folate from plasma to urine. Instead, the linear 27 regression model that described the cumulative excretion of 14C-folate in urine, was used to mathematically model the flux of 1 4 C-folate to the urine pool. Transfer coefficients, shown in Table 4, were derived that successfully predicted the amounts of 14C-folate in the various pools. 5 Figure 11 shows a compartmental model of folate kinetics in a human volunteer. Compartments are numbered 1 through 11 and transfer coefficients shown as k (recipient, donor). Table 4. Transfer coefficients from donor to recipient pools Parameter From To Transfer Coefficient, hr 1 k(8,1) Gut Colon 0.0002 k(2,1) Gut Intestine 3.800 k(3,2) Intestine Plasma 0.094 k(4,3) Plasma Fast tissue 2.300 k(3,4)' Fast tissue Plasma 0.059 k(5,3) Plasma Slow tissue 1.900 k(3,5) Slow tissue Plasma 0.0022 k(9,3) Plasma Marrow 0.0303 k(l0,9)1 Marrow RBC 0.2104 k(l l,0) 1 RBC Degrade 0.000045 k(6,3) Plasma Urine 0.0266 k(8,3) Plasma Colon 0.020 k(7,8) 1 Colon Feces 1.300 Transfer to recipient pool via a delay element 2 Transfer determined by the bioavailability of the bolus 3 Forced to 0 after 24 hours 4 Forced from 0 to 0.21 after 58 hours 5 Forced to 0.00072 after 2300 hours 6 Forcing function used to describe transfer to the urine pool (see text) Example 3 This example shows fluorouracil (5FU), a drug that is used in the treatment of cancer. It 10 belongs to the family of drugs called antimetabolites. It is a pyrimidine analog. 5FU, which has been in use against cancer for about 40 years, acts in several ways, but principally as a 28 thymidylate synthase inhibitor, interrupting the action of an enzyme which is a critical factor in the synthesis of thymine. Some of the principal uses of 5FU are in the treatment of colorectal cancer and pancreatic cancer, where it has served as the established form of chemotherapy for decades. As a pyrimidine analogue, 5FU is transformed inside the cell into different cytotoxic 5 metabolites which are then incorporated into DNA and RNA, finally inducing cell cycle arrest and apoptosis by inhibiting the cell's ability to synthesize DNA. Capecitabine is a prodrug that is converted into 5FU in the tissues. It can be administered orally. Paclitaxel (PAC) is a mitotic inhibitor used in cancer chemotherapy. PAC is now used to treat patients with lung, ovarian, breast cancer, head and neck cancer, and advanced forms of 10 Kaposi's sarcoma. PAC works by interfering with normal microtubule growth during cell division and destroying the cell's ability to use its cytoskeleton in a flexible manner. Specifically, PAC binds to the p subunit of tubulin. Non-cancerous cells are also affected adversely, but since cancer cells divide much faster than non-cancerous cells, they are far more susceptible to PAC treatment. 15 Several in vitro studies on human solid tumor cell lines have demonstrated the positive and schedule-dependent interaction of PAC and 5FU (Kano et al., Br. J. Cancer 1996, 74:704 710; Smorenburg et al., Eur. J. Cancer 2001, 37:2310-2323; Johnson et al., Anticancer Research 2002, 22:3197-3204). A synergistic effect was obtained only when tumor cells were exposed to PAC followed by antimetabolites. Conversely, simultaneous exposure to the two 20 drugs or pretreatment with 5FU reduced overall cell killing compared to PAC alone. The large molecular weight and bulky chemical structure of PAC delay peritoneal clearance, increase exposure in the peritoneal cavity, and can thus be exploited in the treatment of gastric cancers. Furthermore, PAC exerts its cytotoxic effects through a mechanism different from that of 5FU, and thus shows no cross-resistance with 5FU. In tumor cell lines, the 25 combination of PAC and 5FU has demonstrated additive cytotoxicity, especially with sequential exposure. A number of new agents have been introduced recently, either alone or in combination, for the treatment of solid tumors. PAC is one newly developed anticancer drug which has appeared promising for the treatment of gastric cancer, especially for patients with advanced 30 and refractory peritoneal dissemination. Although a number of clinical trials have examined the effects of PAC alone, response rates were approximately 25% and no survival advantages were shown in most of these reports. Therefore, some groups have started clinical trials to examine several new combination regimens of PAC with other chemotherapeutic agents. Cascinu et al. reported a phase I study of weekly 5FU plus PAC every 3 weeks in patients with 35 advanced gastric cancer that was refractory to the existing regimen (5FU, leucovorin, cisplatin, epidoxorubicin). Bokemeyer et al. performed a phase II study of weekly 5FU/leucovorin plus PAC every 3 weeks and showed a 32% response rate for advanced gastric cancer. Despite the promising evidence provided by previous studies, there remains a need for more phase I studies to investigate other combinations of chemotherapeutic agents with weekly PAC. 29 The pharmacological data demonstrated that weekly PAC doses between 60 and 90 mg/m2 produce plasma PAC levels that remain above 0.01 mmol/l for at least 24 hours after the administration over 1 hour, and an AUC of 90 mg/m2 is similar to that observed with a dose of 105 mg/m2 delivered over 3 hours. Prolonged exposure to a low concentration of PAC, 5 on the order of 0.01 mmol/l, has been shown to induce apoptosis in several different cell lines. The results demonstrate that PAC doses can be safely escalated to 90 mg/m2/week, with a fixed dose of 5 days continuous 5FU infusion, with hematological and liver toxicity limiting further dose escalation. Overall, this regimen was adequately tolerated for up to 8 weeks and was associated with moderate toxic effects. Although the MTD of PAC combined with 5FU on 10 this schedule was 90 mg/m2 weekly for three out of every 4 weeks, the recommended dose for a future phase II trial was one level lower. Female athymic nude mice at 6-7 weeks of age were obtained from Taconic Laboratories (Germantown, NJ). The mice were housed in microisolator housing, with food and water provided ad libitum, and quarantined for 4 days prior to the initiation of the study. 15 The HT-29 human colon cancer cell line was used in this study (American Type Culture Collection). HT-29 cells were maintained in DMEM and McCoy's5A medium supplemented with 10% fetal bovine serum respectively. All cells were cultured at 37 degrees Centigrade in an atmosphere of 95% air/5% CO 2 and 100% humidity. Cells were fed every third day and passaged weekly. When cells reached 80% confluence, they were harvested using 0.25% 20 trypsin/EDTA solution. The harvested cells were washed once with phosphate buffered saline (PBS) and re-suspended in PBS at a density of 1 x 107 cells/100 pLl. To each mouse, 100 p1l of the cell suspension was subcutaneously injected to the right flank. Once animals were implanted with cancer cells, they were observed daily for tumor development. When tumors reach approximately 100 mm 3 , the mice were divided equally into 25 three groups and dosed intravenously with 1 4 C-PAC, 1 4 C-5-fluoruracil, or both. PAC was dissolved in a 50% CremophorC EL, 50% dehydrated ethanol solution and diluted with 5% dextrose to prepare the intravenous dose. Fluorouracil was dissolved in water and diluted with 5% dextrose to prepare the intravenous dose. Animal receiving a single agent were dosed with 5 nCi of the radiolabel and the cocktail group received a total of 10 nCi. The administered 30 dose was up to 5 mg/kg body weight in a volume of 20 microliters for the single agent group and 40 microliters for the cocktail group. Mice were sacrificed at 0.5, 2, or 4 hours after dosing and plasma and organ samples were taken immediately and stored frozen until transfer to analytical laboratory. Tissue was placed in individual disposable tissue grinders (Fisher), water was added (2:1 v/w) and 35 homogenized for two minutes to form a homogeneous slurry. Methanol was added to the homogenate (3:1 methanol/homogenate, v/v) and vortexed for 30 seconds. The tubes were then centrifuged at 2,000 g for 10 minutes and the supernatant removed for further analysis. The pellet was retained for further analysis of unextracted radiolabel signal. 30 Tissue extracts (300 uL) were dried using vacuum centrifugation for approximately 2 hours and resuspended in mobile phase A (25 mM ammonium phosphate, pH 6.08). Unlabeled reference standards for PAC and 5FU were spiked into this solution to mark the retention time of PAC and 5FU in the UV trace. Between 50-100 uL if the mixture was injected on a Luna 5 phenyl-hexyl CI 8 column (Phenomenex), 5 um particle size, 4.60mm x 250mm. The chromatography system consisted of a Shimadzu Prominence HPLC with auto sampler, UV detector and fraction collector. A gradient system was used consisting of mobile phase A (25 mM ammonium phosphate, pH 6.08) and mobile phase B (100% acetonitrile). The unit was programmed to deliver 0% mobile phase B from 0-3 minutes, ramp to 90% mobile phase B by 10 19 minutes and hold at 90% mobile phase B until the end of the chromatogram with a flow rate of 1.0 mL/minute. Individual fractions were collected at one minute intervals throughout the chromatogram. Accelerator mass spectrometry (AMS) was used to quantify the radiolabel signal in the fractions. Figure 12 shows a recovery of spiked 5FU and PAC from tissue homogenates using 15 different extraction solvents. Complete recovery of free 5FU (unincorporated) and PAC was obtained using 100% methanol at a ratio of three volumes solvent to one volume tissue homogenate. Figure 13 shows a chromatogram of the separation of 5FU and PAC in mouse tissue homogenate extracts. The 5FU and PAC peaks represent non-labeled reference standards spiked into experimental samples prior to separation to mark the retention time of 20 radiolabeled 5FU and PAC which do not produce a UV signal at the dosed concentrations. One-minute fractions around the 5FU and PAC were collected and further analyzed by AMS for quantitation of radiolabel. The peaks around 5FU represent mouse endogenous compounds which are not radiolabeled and do not interfere with downstream AMS quantitation of 5FU by AMS. 25 Figure 14 shows total radiolabel signal in plasma, tumor xenografts of HT-29 human colon cancer cell lines and normal lung tissue in mice two hours after receiving 5FU, PAC or a cocktail of both 5FU and PAC. The total plasma signal (DPM/mL plasma) for mice that received 5FU is comparable to the group that received PAC. This is expected since each group received 5 nCi of the dose and the established plasma half-life of PAC is 0.34 hours and 0.25 30 0.30 hours for 5FU. When 5FU and PAC were co-administered in a cocktail (10 nCi total), there was a nearly 20-fold increase in the total plasma signal strongly suggesting a synergistic effect between these two agents. Other researchers have also seen this additive effect in certain cell lines (Kano, British Journal of Cancer 74(5):704-10, 1996) and employed the combination therapy in clinical trials, demonstrating improved pharmacological profile of PAC (Kondo, 35 Japanese Journal of Clinical Oncology, 35(6)332-337, 2005). Mouse tumor and normal lung tissue (total DPM/mL extract) in this example also demonstrates this additive effect. In this case, there was an approximately 10-fold increase in total signal when 5FU and PAC were administered as a cocktail. Notably, the methanol extraction procedure is designed to recover PAC and the free unincorporated pool of 5FU. 5FU incorporated into RNA and DNA is 31 recovered in the pellet and measured separately. The results in these charts consistently show low 5FU signal demonstrating near complete incorporation of 5FU into the RNA and DNA fraction. To further delineate the source of signal in each sample type, high performance liquid 5 chromatography was employed to separate 5FU from PAC followed by quantitation by accelerator mass spectrometry. Figure 15 shows the amount of 5FU and PAC in tumor, normal lung tissue and plasma after mice received either 5FU, PAC or both as a cocktail two hours after iv administration. Again, the methanol extract of tissue homogenates was used in the chromatography capturing data for the free unincorporated pool of 5FU. The limit of 10 quantitation is shown as a dotted line and was calculated based on background signal of the carrier carbon added to each sample during processing. The PAC treated group shows signal corresponding to PAC but not 5FU in tumor, lung and plasma. The 5FU treated group shows signal corresponding to 5FU but not PAC in plasma. No free 5FU was detected in the methanol extracts of tissue homogenates demonstrating near complete incorporation of free 15 5FU into RNA and DNA which were recovered in the pellet after centrifugation of the methanol extract. Quantitation of total radiolabel in the pellet fraction yielded a signal of approximately 1.103 DPM and 0.914 DPM for tumor and lung tissue, respectively, in 5FU treated animals. This compared to 0.468 DPM and 0.927 DPM of PAC for tumor and lung tissue, respectively, in the methanol extract fraction of PAC treated animals. Furthermore, in 20 plasma, where 5FU is expected to exist only in the free form, 5FU is clearly detected in the methanol extract of mice that received 5FU or the cocktail. Fractionation of tissue homogenates into a solvent-based supernatant and a pellet allowed tracing of not only 5FU distribution into tissue but also provided the ability to distinguish the free 5FU pool from the DNA and RNA incorporated pool. 25 Figure 16 shows chromatographic separation and quantitation of 5FU and PAC in tumor and lung tissue extracts after treatment with either 5FU or PAC. The tissue extracts were chromatographed as described earlier and again after spiking with less than 1.0 DPM of 5FU. The chart is normalized to PAC and demonstrates the expected increase in the 5FU signal after spiking with 5FU, further reinforcing the robust chromatographic method 30 developed in this example. 32

Claims (7)

1. A process for determining personalized therapy, comprising: (a) administering a test cocktail to a test subject, wherein the test cocktail comprises two or more different therapeutic agents at a dosage at least two times lower than an 5 expected therapeutic dose; (b) obtaining a sample biopsy of the relevant diseased tissue for study; and (c) analyzing the sample biopsy for each of the administered therapeutic agents and their metabolites.

2. The process for determining personalized therapy of claim 1 wherein the dosage 10 of the test cocktail is a tracer dose, wherein a tracer does is less than 10% of a therapeutic dose.

3. The process for determining personalized therapy of claim 1 wherein the analyzing step is performed with an Accelerator Mass Spectrometry (AMS) instrument.

4. The process for determining personalized therapy of claim 1 wherein the process further comprises pausing from about 10 minutes to about two hours after 15 administering a test cocktail to allow for tissue distribution of the test cocktail.

5. The process for determining personalized therapy of claim 1 wherein the sample biopsy is a piece of tissue selected from the group consisting of excised tumor tissue, blood, fractionated blood, isolated pathogen-infected tissue, and combinations thereof.

6. The process for determining personalized therapy of claim 5 wherein when the 20 sample biopsy is blood, the blood sample is fractionated into each type of white and red blood cell.

7. The process for determining personalized therapy of claim 1 wherein the process further comprises fractionating the sample biopsy by sorting the sample into component cell types. 25 33