Chemotherapeutic Agents: A Case for Diagnostics to Achieve Therapeutic Dose Management During Clinical Development

It is the objective of oncologists treating patients with chemotherapy drugs to achieve maximum tolerated dose (MTD), the highest dose of a drug that will produce the desired effect without unacceptable toxicity. The amount of chemotherapy administered to achieve MTD has historically been, and still is, typically based on a patient’s body surface area (BSA) measurements. However, the limitations of BSA are now well documented, and technology advancements provide alternative dosing schemes that offer the potential of improving patient care. Therapeutic dose management (TDM) by simple blood tests is a diagnostic tool that is the standard of care in other fields of medicine and can serve as a valuable resource in optimizing cancer drug dosing to individual patient needs.

BSA originates from an outdated formula developed in 1916 and lacks accuracy. The data collected in this nearly century-old study was garnered from eight subjects and was intended to adjust for the basal metabolic rates to estimate the starting dose in humans from the drug doses used in animals. BSA takes only two variables into account (height and weight), when in fact there are numerous factors that can influence the amount of chemotherapeutics an individual needs to achieve MTD, including sex, age, obesity, renal and/or hepatic function, and concomitant medications, among others. Furthermore, these variables most likely, in turn, affect drug biodistribution, metabolism, and clearance. Numerous studies have shown that using BSA to determine proper dose of a drug is highly variable, both from an inter- and intrapatient perspective.

A review of the literature by Gurney in 1996 illustrated that numerous studies have shown a lack of correlation between BSA and toxicity of chemotherapeutics, whereas a positive correlation was observed between PK parameters and toxicity.¹ More recently, Baker et al examined interpatient drug clearance variability in a retrospective analysis of 33 cancer drugs that underwent clinical development between 1991-2001.² A total of 1650 patients received one of 33 agents undergoing clinical testing. Only 5 of these agents exhibited a positive correlation between BSA and clearance. The relative improvement of variability of these five drugs was between 15% and 35%, suggesting that BSA can improve variability by one-third at best. Baker concluded that alternate dosing strategies based on drug exposure should be evaluated when determining the proper dose regimen of chemotherapeutics.

Pharmacokinetically (PK) guided dose adjustment is such an alternative dosing strategy. Although others had reported that individually adjusting the dose of one of the most commonly prescribed chemotherapeutics, 5-fluorouracil (5-FU) could reduce the incidence of side-effects in patients with head and neck cancer, Fety et al were the first to show in a randomized designed trial that personalizing the dose of 5-FU resulted in a significant reduction in drug induced adverse events.^3,4 In their study, 122 head and neck cancer patients were randomized to standard dose or PK-adjusted doses of 5-FU. Of these patients, 17.5% in standard group had grade 3–4 neutropenia and thrombocytopenia toxicities versus 7.6% in the PK arm; these toxicities were associated with higher systemic exposure to the drug. Furthermore, 5.1% in standard arm had grade 3-4 mucositis toxicity versus 0% in the PK arm. Importantly, pharmaco-economic analysis demonstrated a 14.6% reduction in medical costs for the PK group.

Gamelin et al followed this with a prospective Phase III study in 186 patients with metastatic colorectal cancer; all patients were treated with 5-FU.⁵ One arm received the approved Body Surface Area (BSA) dose of 1500 mg/m2 during a continuous 8-hour infusion once weekly, whereas in the second arm doses were adjusted weekly based on 5-FU plasma concentrations. Gamelin found that only 15% of the patients were receiving a dose of the drug in the therapeutic range; 68% of the patients were under dosed and 17% overdosed. The clinical outcomes were equally compelling. Overall response rates for the BSA and dose-adjusted groups were 18.3% and 33.6%, respectively. For these two groups, survival was 16 months for the BSA group and 22 months for the dose-adjusted group.

This issue is not unique to 5-FU. Kantarjian et al reported that interpatient PK variability may explain the variable response observed in patients with chronic myeloid leukemia (CML) after treatment with the tyrosine kinase inhibitor imatinib (Gleevec).^6,7 Follow-up studies by Picard et al and Larson et al in CML patients, and Demetri et al in patients with gastro-intestinal stromal tumors (GISTs) have demonstrated that imatinib was more effective in patients with higher trough blood levels of the drug versus those patients in the lowest quartile of trough blood levels.^8,9,10 In the Picard et al and Larson et al papers, patients in the lowest quartile were more likely to be discontinued, primarily because of unsatisfactory therapeutic effect. In CML patients, the efficient plasma trough levels for imatinib should be above ~1000 ng/ml for positive clinical outcome, whereas in GIST patients, the best clinical response was observed at plasma trough levels ranging between 2041–4182 ng/ml.

Another flaw in the current MTD dosing scheme is that it is determined in a very limited number of subjects. In general, the Fibonacci dose escalation “3+3 schema” is employed. In this schema, three patients are enrolled in each proposed dose level. Once a dose limiting adverse event is observed in even one subject, three more subjects are added to that particular dosage group; if two or more of these subjects experience a drug limiting toxicity, then the preceding dose is considered the MTD. Thus, data from as few as six subjects might be employed to determine the MTD that will be used during the drug’s entire clinical development plan. Taken together, the primary means of establishing the proper dose of a chemotherapeutic during clinical development is determined on data from a very limited number of subjects, and independent of the exposure of each subject to the drug.

These findings beg the question why all patients with cancer are not dosed individually based on systemic exposure as a meaningful improvement over BSA. It is critical that a more effective algorithm be adopted to ensure that the patient receives the right dose of the right drug at the right time. Optimizing drug dose on a patient-by-patient basis should be viewed favorably by regulatory agencies and payers. Indeed, regulatory agencies are placing an increased emphasis on tailoring medical treatment based on the specific biology or pathology of each patient (i.e., personalizing the drug regimen). A recent publication in The New England Journal of Medicine by Margaret Hamburg and Francis Collins, the heads of the FDA and NIH, respectively, stated that scientists are developing and using diagnostic tests to predict better patient responses to therapy.¹² Hamburg and Collins share the vision that the FDA and NIH need to support personalized medicine, and the agencies plan to move forward on several fronts. One such effort is the recently published guidance on the co-development of a therapeutic with a companion diagnostic test, where the companion diagnostic might be a genetic test, an assay for a protein biomarker or assay for TDM.¹³ Clearly, they believe it is imperative that the proper diagnostic tools are available to steer the patient to the right dose of the right drug. To date, only Herceptin, Ontak, Tykerb, Vecitibix, and Erbitux are FDA-approved drugs with mandatory companion diagnostic tests to identify the subset of patients most likely to have a favorable clinical outcome; 11 other oncology drugs have mention of biomarkers in the label.¹⁴

As already mentioned, having an assay that can measure levels of the drug in a patient’s blood is considered a companion diagnostic. Whereas a genetic test or protein assay will (in theory) ensure the right patient receives the right drug by providing information on the state of the disease, the right dose of the drug will still need to be optimized. De Jonge et al published a comprehensive review article in 2005 on the importance on TDM.¹⁵ When doses of chemotherapeutic agents were administered based on BSA, plasma blood levels varied as much as 100-fold from patient to patient, as in the case of 5-FU, presumably because of the high variability of PK observed with chemotherapeutic agents as already mentioned.^4,16 Given these data, one can argue that the reasons for conducting TDM during the development of a therapeutic is just as important as genetic testing or protein assays. The study of Gamelin et al summarized above is a landmark study highlighting the importance of TDM in patients with cancer.⁵

As regulatory agencies push the use of companion diagnostics in personalized medicine, it appears the pharmaceutical and biotechnology industries are getting the message. As of 2008, only Roche and Novartis had invested in internal companion diagnostic research. Other companies decided to partner for diagnostic expertise and as of that time, there were seven pharmaceutical-diagnostic partnerships. This number more than doubled by 2010, and in a report from the Tufts Center for the Study of Drug Development in that same year, 94% of companies surveyed stated they were investing in personalized medicine research and that 12%-50% of their pipelines are personalized medicines.¹⁷ What is the reason for this relatively late entry by the industry into personalized medicine and the use of companion diagnostics during drug development? Some companies undoubtedly waited because they did not want to stratify the patient population and thus reduce market size. Others may not have wanted to complicate clinical development by codeveloping a companion diagnostic with their therapeutic, and some may just not have appreciated the significance of using a companion diagnostic as part of the development plan. Finally, some companies may not have wanted to invest in internal expertise or to form partnerships, both of which take time and consume internal resources. Regardless, FDA’s refusal to approve ChemGenex’s Omapro for chronic myeloid leukemia (CML) without a validated companion diagnostic should be a wake-up call to those companies not pursuing a specific companion diagnostic test during clinical development, when appropriate. ¹⁸

AstraZeneca’s Iressa is another example where failure to codevelop a companion diagnostic resulted in either restricted labeling or nonapproval by regulatory agencies. Originally approved by FDA in 2003 as third-line monotherapy in patients with non small-cell lung cancer (NSCLC), FDA placed very restrictive labeling on the drug when the Iressa Survival Evaluation in Lung Cancer (ISEL7) study failed to show an overall survival benefit.¹⁹ Subanalysis of the data demonstrated that patients of Asian descent and nonsmokers benefitted from the drug. Subsequent research led to the discovery that only patients with a mutation of EGFR tyrosine kinase (EGFRTK) responded to the drug. AstraZeneca partnered with DxS Ltd (now Qiagen) to make widely available the assay for EGFRTK mutation and in parallel reapplied for regulatory approval. Sales of Iressa turned around when it gained EMA approval in 2009 for advanced metastatic NSCLC in patients expressing the EGFRTK mutation. In fact, worldwide Iressa sales in 2010 were $393 million, a 28% increase over 2009.²⁰ Clearly, in the cases of Omapro and Iressa, not developing a companion diagnostic in parallel during clinical development was detrimental to the drugs’ regulatory approval and marketing success.

The issues with Omapro and Iressa can be applied to TDM. Cyclosporin and warfarin are two examples where TDM is required by regulatory agencies. Unfortunately, most oncology practices do not have the pharmacy capabilities, expertise, or even the tools needed to collect and process blood samples for TDM. All of this is complicated by the need to collect and process the samples in a relatively short period of time. Still, given the high cost of drug development and the risk of failure if the wrong dose is given during clinical development, it seems intuitive that companies should invest the time and money needed during clinical development to have the best chance of observing a positive clinical outcome, regardless if the effort is internal or via a partnership. Indeed, given DiMasi’s 2003 findings that capitalized clinical costs per approved new drug averaged $467 million and average development time was 90.³ months, each month saved in development time will save the sponsor on average $5 million, not to mention the huge advantage of getting to the market sooner.²¹ Having available an assay with the proper characteristics to measure drug exposure early in clinical development will increase the chances of getting to an early-go or no-go decision. If a no-go decision is made, the sponsor can quickly reallocate resources to another project(s). If results from clinical development are positive, the sponsor can move very quickly to efficacy trials. It is reasonable to assume that fewer patients will be needed to show efficacy if every patient receives the correct therapeutic dose and are neither overdosed nor underdosed. It is also reasonable to assume that the need to study fewer patients will lead to corresponding decreases in development costs and time, as well as extending the marketing life-span of the drug, all of which are near and dear to the heart of the pharmaceutical and biotech industries.

It should be standard research practice to develop PK-PD (pharmacodynamics) relationships early in the drug development process for chemotherapeutics when the drug is given as a single agent. Phase II may be the ideal place to establish concentration-effect relationships since more patients are studied (versus Phase I). It is also critical to consider early in development the type of assay to be employed in the critical Phase III trials and, ultimately, to be employed in clinical practice. Whereas HPLC or liquid chromatography mass spectroscopy (LC/MS) are more than sufficient to measure systemic concentrations of drug levels, both are too time consuming and labor intensive for large scale trials, and too expensive to be used in the clinic. Immunoassays have been used routinely in several indication areas in medicine since the 1970s, and they may be ideal for TDM because such an assay is highly sensitive and specific, high throughput, robust, can be run on analyzers used in routine clinical laboratory research, and relatively inexpensive. Proper project planning is important if an immunoassay is going to be employed in Phase III. It may take 12–15 months to generate the appropriate antibody and validate a Research Use Only (RUO) assay. Another 15–18 months may be needed to validate a commercial assay and gain regulatory approval. Considering these timelines, initiation of an immunoassay needs to begin 2.5–3 years before market launch.

In conclusion, personalized medicine will only continue to become a more important component of drug development, and TDM will be as much a part of it as genetic testing or using a companion diagnostic to identify a disease specific biomarker because managing the right dose is just as important as selecting the right drug. The importance of dose management for all therapeutics where exposure is variable is unquestioned, but it is especially important for oncology products, which also often have a narrow therapeutic window and dose limiting toxicity. It is also unquestioned that regulatory agencies will increasingly demand that the various components of personalized medicine are incorporated as part of the clinical development plan and that, when appropriate, approval of a therapeutic will be linked to approval of a companion diagnostic.
Companies not embracing this development strategy will be left behind. Regardless of the type of companion diagnostic, the assay must be robust, stable, inexpensive and available for routine use.

References

1.Gurney. Dose Calculation of Anticancer Drugs: A review of the current practice and introduction of an alternative. 1996. J. Clin. Oncology. 14(9): 2590-2611.

2.Baker et al. Role of body surface area in dosing of investigational anticancer agents in adults 1991-2001. J. Nat. Cancer Inst. 94, (24): 1883-1886.

3.Santini et al. 5FU therapeutic monitoring with dose adjustment leads to an improved therapeutic index in head and neck cancer. 1989; Br. J. Cancer 59: 287-290.

4.Fety et al. Clinical Impact of Pharmacokinetially-guided Dose Adaptatio of 5-Fluorouracil: results from a multicenter Randomized trial in Patients with Locally Advanced Head and Neck Carcinomas. 1998. Clinical Cancer Research 4: 2039-2045.

5.Gamelin et al. Individual 5-fluorouracil dose adjustment based on pharmacokinetic follow-up compared with conventional dosage: results of a multicenter randomized trial in patients with metastatic colorectal cancer. J. Clin Oncol. 2008; 26(13):2099-2105.

6.Kantarjian et al. High-dose imatinib mesylate therapy in newly diagnosed Philadelphia chromosome-positive chronic phase chronic myeloid leukemia. Blood. 2004; 103:2873-2878.

7.Kantarjian et al. Dose escalation of imatinib mesylate can overcome resistance to standard-dose therapy in patients with chronic myelogenous leukemia. Blood. 2003;101:473-475.

8.Picard et al. Trough imatinib plasma levels are associated with both cytogenetic and molecular responses to standard-dose imatinib in chronic myeloid leukemia. Blood. 2007;109:3496-3499.

9.Larson et al. Imatinib pharmacokinetics and its correlation with resonse and safety in chronic-phase chronic myeloid leukemia: a subanalysis of the IRIS study. Blood. 2008; 111(8): 4022-4028.

10.Demetri et al. Imatinib plasma levels are correlated with clinical benefit in patients with unresectable/metastatic gastrointestinal stromal tumors. J. Clin Oncol 2009; 27(19): 3141-3147.

11.DuBois and DuBois. A formula that approximates body surface area if height and weight are known. Arch. Intern. Med. 1916; 17:863-871.

12.Hamburg and Collins. The path to personalized medicine. N. Eng. J. Med. 363(4) 301-304.

13.FDA Guidance: http://www.fda.gov/MedicalDevices/DeviceRegulationandGuidance/ GuidanceDocuments/ucm262292.htm

14.FDA, Pharmacogenomic Biomarkers in Drug Labels, www.fda.gov

15.de Jonge et al, Individualized cancer chemotherapy: strategies and performance of prospective studies on therapeutic drug monitoring with dose adaptation. Clin. Pharmacokinet 2005: 44(2): 147-173.

16.Undevia et al. Pharmacokinetic variability of anticancer agents. Nat Rev Cancer. 2005; 5(6):447-458.

17.Personalized Medicine is Playing a Growing Role in Biopharmaceutial Development Pipelines. www.ageofpersonalizedmedicine.org/center/publications/report-2010-Tufts.asp

18.ChemGenex press release, July 2010. http://www.chemgenex.com/2010/07/us-fda-agree-on-potential-regulatory-pathwayo/

19.Case Study: Personalized cancer therapy. DATAMONITOR March, 2011.

20.AstraZeneca 2010 year-end financial report http://www.astrazeneca-annualreports.com/outputPDF.aspx

21.DiMasi et al, The price of innovation: new estimates of drug development costs. J. Health Economics 22 (2003):151-185.

Salvatore J. Salamone is the founder, senior vice presidet, and chief scientific officer at Saladax Biomedical. Jack M. DeForrest is the president of JDF PharmaConsulting.