Pharmacogenomic testing: the cost factor - Nature

The Pharmacogenomics Journal (2001) 1, 171–174

 2001 Nature Publishing Group All rights reserved 1470-269X/01 $15.00 www.nature.com/tpj

EELS (Ethical, Economic, Legal & Social) ARTICLE

Pharmacogenomic testing: the cost factor PJ Wedlund1 and J de Leon2 1

College of Pharmacy, University of Kentucky, Lexington, KY, USA; 2College of Medicine, University of Kentucky, Lexington, KY, USA INTRODUCTION A number of reports in the literature have characterized the clinical relevance of genetic variations on therapeutic response and outcomes to drugs.1–7 These and other reports support the general belief that genetic testing could be used to improve patient therapy.8–10 The urgency of moving in this direction has been further underscored by reports that adverse drug reactions add billions of dollars per year to our health care bill, and that over 100 000 patients per year may die from taking prescription medications properly.11,12 This has convinced many researchers that genetic tests could improve how drugs are selected and prescribed to patients, and that such tests should become clinically available for this purpose within the next 10–20 years. Beyond the ethical and legal questions surrounding genetic testing, the clinical use of patient genomic information to improve therapeutic treatment often ignores the most fundamental question about this technology—is it cost effective? The clinical implementation of genomic testing hinges on answering a very simple question: ‘Under what condition(s)/situation(s) are the savings provided by a pharmacogenomic test greater than the cost of genetic testing?’ The answer to this question requires two pieces of information: (1) the cost of performing a genetic test; and (2) the savings that could be realized if genetic testing were employed in place of current clinical methods. When the savings realized are greater than the cost of performing

the test, it makes economic sense to apply pharmacogenomic testing. If pharmacogenomic testing had been proven to save money by now, it would already be a routine clinical tool. It is the lack of adequate data to answer this simple question that has delayed the introduction of pharmacogenomic testing into therapeutic practice. There is a progression of steps that must be followed to both understand when and how to perform a cost analysis of pharmacogenomic testing. This starts with a description of where genetic testing has already been shown to be cost effective and why, and proceeds by defining the elements that are required before one can begin to apply similar efforts to a cost analysis of pharmacogenomic testing. There must also be an appreciation for the size of the savings that pharmacogenomic testing must generate in order to be a cost-effective tool. Equally important is the need for researchers to understand that such studies must be performed not in selected populations, but in the routine clinical environment. Finally, there are many different approaches to cost analysis, but one must distinguish between what is ideal and what is practical. It is hoped the following points and ideas will stimulate discussion in this area and help focus attention on what needs to be done to move pharmacogenomic testing from the bench to the bedside. GENETIC TESTING THAT IS COST EFFECTIVE Genetic testing has made some clinical in-roads, notably in areas where it is

cost effective to implement. For example, delays associated with culturing and the difficulty in culturing certain pathogenic organisms increase clinic and patient care costs to the point where genetic testing provides a cost-efficient means for improving the efficiency and response times of microbiology lab services.13 It has been estimated that genetic testing can more quickly identify Mycobacterium tuberculosis and its resistance at a London hospital and could save the hospital $70 000–$210 000 per year relative to the current conventional methods for M. tuberculosis identification.14 Similarly, detection of herpes simplex virus, Chlamydia and Neisseria gonorrhoeae by polymerase chain reaction (PCR) based genetic tests is considered cost effective relative to other clinical methods.15 Genetic testing has been embraced because it has been shown to provide an obvious monetary and practical benefit not provided by any other techniques. THERAPEUTIC AND ECONOMIC ISSUES SURROUNDING ROUTINE PHARMACOGENOMIC TESTING Most studies of human genetic variation have focused on how a specific genetic variation or set of variations affects therapeutic outcomes.1–7 Genomic variation must have a well defined therapeutic relevance before its impact on health care costs can be addressed (Table 1). There is a natural progression to first define the therapeutic importance of a genomic variation followed by its economic impact. The economic impact requires information on two additional factors: (1) the cost of performing the pharmacogenomic test; and (2) the savings provided by the pharmacogenomic testing. Advances in genetic testing by PCR methodologies have significantly affected genetic testing expenses which reportedly range between $80– $250 per sample depending on the lab, precise techniques and number of alleles and genes being tested.16–18

Pharmacogenomic testing: the cost factor PJ Wedlund and J de Leon

172

Table 1 Factors affecting the therapeutic and economic relevance of pharmacogenomic variability A genomic variation will have therapeutic relevance when ALL the following criteria are met: (1) Genomic variation affects the expression, level and/or function of the final gene product. (2) There is an established relationship between the genomic variation and therapeutic response and/or outcome. (3) The genomic variation affecting therapeutic response and/or outcomes are of clinical importance and their effects can not be more easily assessed by some direct clinical or paraclinical measurement. A genomic variation may have economic relevance if one or more of the following are met: (1) Genomic variation affects length of in-patient stay or likelihood of hospital admission when patients prescribed certain drug(s). (2) Genomic variation affects number of clinician visits when patients prescribed certain drug(s). (3) Genomic variation affects clinical lab services when patients prescribed certain drug(s). (4) Genomic variation affects use of ancillary health care services when patients prescribed certain drug(s). (5) Genomic variation affects the cost of drug therapy by its effect on the use or effectiveness of drug therapy and/or treatment outcomes.

Unquestionably, lower per sample costs are possible with higher throughput and more automated testing systems,15,19 but the cost of labor, overheads, equipment depreciation, supplies and profit must still be factored into this equation. Realistically, genomic testing by most clinical labs is not likely to fall much below about $50 per test in the immediate future, and is more likely to remain near $100 or so per sample in many clinical labs. The current trend is to test for multiple variations in a given gene and multiple genes that can affect therapeutic response,7,8,16 and this trend adds to the cost of genetic testing even as testing methods improve. Thus, the actual clinical cost of genomic testing will probably remain within a range of between $50 000–$250 000 per 1000 samples tested, with the actual cost hovering close to $100 000–$150 000 per 1000 samples tested over the short term in spite of improved efficiencies in genetic testing methodologies. The other piece of information needed is whether the savings provided by pharmacogenomic testing services will justify this expense. To answer this question, one must first ask: ‘What is the cost of a genetic variation?’ There have been only limited efforts to address this outside the realm of monogenetic diseases and diseases associated with known environmental agents producing disease.17,20–23 Disease has a much better defined cost range than therapeutic complications, and because most diseases are not cured, one can extrapolate expenses to get life-time cost estimates for the dis-

The Pharmacogenomics Journal

ease. One cannot do this for therapeutic complications. However, it was recently reported that psychiatric inpatients deficient in CYP2D6 enzyme expression cost on average $4000– $6000 more per year to treat than patients who express this enzyme when prescribed drugs that are metabolized by the CYP2D6 enzyme.24 This translates to an estimated $112 000– $168 000 per year in added health care expenses at this psychiatric hospital directly related to this single gene polymorphism. Unfortunately, this estimate was based on a population size too small to prove that pharmacogenomic testing would save money. ADDRESSING THE SAVINGS FROM PHARMACOGENOMIC TESTING There are many factors that must be taken into account to characterize the cost analysis of genomic testing,25 and the precise approach must consider the specific patient population and genomic variation being evaluated. Ultimately, the average savings (S) generated per affected patient by a pharmacogenomic test must be greater than the unit cost (C) of performing the test and the frequency (F) of the genomic variation being tested in the patient population based on equation 1: S ⬎ C/F

(1)

This implies that a genomic variation that is infrequent in the population must either be associated with a low cost of detection, or a very large saving in order to justify its routine determi-

nation in clinical practice. This condition is what has limited the broader and more routine testing for genetic variations responsible for monogenetic diseases.20–22 The direct annual health care savings associated with pharmacogenomic testing is affected by four major factors (Table 2). The total savings can never be accurately determined for most pharmacogenomic tests since these will accrue over the lifetime of a patient, and there is no simple way to accurately define lifetime exposure of a patient to various therapeutic agents. There is also no way to accurately assess the actual annual savings generated by pharmacogenomic testing without performing a study for this purpose in a patient population. It may be possible to obtain estimates for the type and size of the patient population affected. It may even be possible with some genomic variations to Table 2 Major factors affecting the annual savings associated with pharmacogenomic testing (1) Type and annual number of patients therapeutically compromised by genomic variation. (2) Severity of therapeutic complication and/or outcome produced by the genomic variation. (3) Predictability of therapeutic complication and/or outcome based on information about the genomic variation. (4) Ability to apply the pharmacogenomic information to prevent/avoid the inappropriate therapeutic complication and/or outcome.


173

anticipate the severity of the therapeutic complications and/or outcomes. However, it will never be possible to accurately define the resultant costs or the ability to apply pharmacogenomic testing in clinical practice without performing these studies in a routine clinical environment.

PATIENT SELECTION FOR STUDIES OF SAVINGS ATTRIBUTABLE TO PHARMACOGENOMIC TESTING Attempting to extrapolate highly controlled clinical studies to the routine patient population is dangerous and misleading. For example, the absence of a metabolic enzyme may not be associated with any major complication in a patient who only receives a single drug. However, this same drug may produce toxicity when that patient is exposed to two or more drugs concurrently,26 a common practice in routine clinical settings. Alternatively, single dose studies may not predict the real therapeutic relevance of genomic variation when patients are treated chronically with drugs due to non-linear kinetics, tolerance and/or compensatory homeostatic mechanisms. The goal in performing a cost analysis of pharmacogenomic testing is not to optimize the detection of a clinical difference between genetically distinct groups, it is to define the true clinical relevance of the genomic variation. To accomplish this objective, clinical studies must be performed in the actual clinical environment(s) in which the pharmacogenomic test is likely to be used. Unfortunately, this point is not well appreciated or understood by very many researchers. Most studies that attempt to describe the therapeutic importance of pharmacogenomic variations are based on research that has little or no relevance to routine clinical practice. Often patient selection and/or the study design are not reflective of the real clinical environment, but are designed primarily to optimize the potential for detecting a significant difference between two genetically distinct groups. If pharmacogenomic testing is to move from the bench to the

bedside, it will require addressing its therapeutic relevance at the bedside. METHODS FOR COST ANALYSIS OF PHARMACOGENOMIC TESTS There are a number of different methods that have been used to assess the value of clinical procedures.25 Standardized methods have been promulgated by the Panel on CostEffectiveness in Health and Medicine.27–29 In brief, to demonstrate a clinical procedure is cost-effective requires the procedure be evaluated in the patient population in which it will be used, and that its cost, relative to the length or quality of life, is reasonable (typically less than $50 000 per year of life gained or quality adjusted life year).25,30 Cost-effective analysis measures the outcomes relative to a group receiving a standard treatment. Cost-benefit analysis defines the dollar benefit of a treatment, and divides that benefit by the cost to provide it. When the benefit/cost ⱖ 1 it is considered cost-effective to provide the treatment. The criticism of cost-benefit analysis is that it ignores clinical procedures that may improve the quality or length of life, but do not provide any direct measurable savings.25 However, a costbenefit analysis is important to those who pay the bills. The cost of performing cost analysis research is a major factor hindering the execution of these studies. For this research, there must be two randomized groups—patients who receive standard therapy, and patients where the treatment is based on the additional clinical information provided by a pharmacogenomic test. Some objective or subjective measurement must be used to quantify the outcome of applying or not applying the results from a pharmacogenomic test. When the outcome measured is in dollars, the analysis is a cost-benefit analysis. When the outcome is measured by more subjective measures it is a cost-utility or cost-effectiveness study. These types of studies are somewhat more difficult for healthcare providers to evaluate since they do not relate the intervention with the direct healthcare dollar saved. For many genomic variations that

are expected to impact only a few percent of the population, thousands of patients may have to be recruited to establish (with sufficient power and significance) if pharmacogenomic testing is cost-effective. The need for large population sizes in pharmacogenomic research on cost-effectiveness or costbenefit analysis makes this clinical research extremely expensive, and time consuming to perform. Patient clinical evaluations (the most time consuming and costly parts) increase these expenses significantly. Randomized studies with large patient groups may not even be possible in the current clinical environment controlled by managed care companies. Thus, there is a need to incorporate practicality with what is the ideal study design. An easier and less expensive approach to determining if pharmacogenomic testing has the potential for providing value to therapeutics is by initially defining the cost of care for patients with and without a particular genomic variation. Cost comparison studies provide an estimate of the maximum potential savings that may be realized by pharmacogenomic testing. The true savings will depend on the extent to which the pharmacogenomic test can be applied in clinical practice to avoid the therapeutic complications and costs attributed to the genomic variation. The advantage of cost-comparison studies is the much lower expense associated with performing them. Unlike other cost evaluations that require comparing costs and outcomes in a control and treatment group, a cost-comparison only requires evaluating the difference in costs between the different genomic groups. Randomization is not necessary provided the population is of adequate size and is selected from all patients receiving standard therapy. CONCLUSIONS Assessing differences in the cost of treating typical patients with different genotypes provides a simple answer to the question: ‘What is the cost of genomic variation?’ If that cost is greater than the cost associated with pharmacogenomic testing, it makes economic

www.nature.com/tpj


174

sense to consider applying the genomic testing to therapeutics. Without any information about the difference in cost of treating patients with specific genomic variations, it is impossible to know if pharmacogenomic testing is just a useful research method or a practical clinical tool. Scientific and lay journals have proposed that new genetic technologies will be used to individualize therapy in 10–20 years. If pharmacogenomics is to become a great therapeutic tool for individualization of patient care in the 21st century, a key element in its rise to prominence must be a demonstration for health providers of its economic value. DUALITY OF INTEREST None declared. Correspondence should be sent to PJ Wedlund, College of Pharmacy, Division of Pharmaceutical Sciences, University of Kentucky, Lexington, KY 40536–0082, USA Tel: ⫹1 859 257 5788 Fax: ⫹1 859 257 7564 E-mail: pjwedl1얀pop.uky.edu

REFERENCES 1 Drazen IE, Liggett SB, Boushey HA, Cherniack RM, Chinchilli VM, Cooper DM et al. The effect of polymorphisms of the beta(2)adrenergic receptor on the response to regular use of albuterol in asthma. Amer J Respir Crit Care Med 2000; 162: 75–80. 2 Benhattar J, Cerottini J-P, Saraga E, Metthez G, Givel J-C. p53 mutations as a possible predictor of response to chemotherapy in metastatic colorectal carcinomas. Int J Cancer 1996; 69: 190–192. 3 Richard F, Helbecque N, Neuman E, Guez D, Levy R, Amouyel P. APOE genotyping and response to drug treatment in Alzheimer’s disease. Lancet 1997; 349: 539. 4 Persson K, Sjo¨ stro¨ m S, Sigurdardottir I, Molna´ r V, Hammarlund-Udenaes M, Rane A. Patient-controlled analgesia (PCA) with codeine for postoperative pain relief in ten extensive metabolizers and one poor metabolizer of dextromethorphan. Br J Clin Pharmacol 1995; 39: 182–186. 5 Schu¨ tz E, Gummert J, Mohr F, Oellerich M. Azathioprine-induced myelosuppression in thiopurine methyltransferase deficient heart transplant recipient. Lancet 1993; 341: 436.

The Pharmacogenomics Journal

6 Kuivenhoven JA, Jukema JW, Zwinderman AH, de Knijff P, McPherson R, Bruschke AVG et al. The role of a common variant of the cholesteryl ester transfer protein gene in the progression of coronary atherosclerosis. N Engl J Med 1998; 338: 86–93. 7 Arranz MJ, Munro J, Birkett J, Bolonna A, Mancama D, Sodhi N et al. Pharmacogenetic prediction of clozapine response. Lancet 2000; 355: 1615–1616. 8 Evans WE, Relling MV. Pharmacogenomics: translating functional genomics into rational therapeutics. Science 1999; 286: 487–491. 9 Ingelman-Sundberg M, Oscarson M, McLellan RA. Polymorphic human cytochrome P450 enzymes: an opportunity for individualized drug treatment. Trends Pharmacol Sci 1999; 20: 342–349. 10 March R. Pharmacogenomics: the genomics of drug response. Yeast 2000; 17: 16– 21. 11 Classen DC, Pestotnik SL, Evans RS, Lloyd JF, Burke JP. Adverse drug events in hospitalized patients. Excess length of stay, extra costs, and attributable mortality. J Amer Med Assoc 1997; 277: 301–306. 12 Lazarou J, Pomeranz BH, Corey PN. Incidence of adverse drug reactions in hospitalized patients. A meta-analysis of prospective studies. J Amer Med Assoc 1998; 279: 1200–1205. 13 Louie M, Louie L, Simor AE. The role of DNA amplification technology in the diagnosis of infectious diseases. Can Med Assoc J 2000; 163: 301–309. 14 Drobniewski FA, Watterson SA, Wilson SM, Harris GS. A clinical, microbiological and economic analysis of a national service for the rapid molecular diagnosis of tuberculosis and rifampicin resistance in Mycobacterium tuberculosis. J Med Microbiol 2000; 49: 271–278. 15 Fedricks DN, Relman DA. Application of polymerase chain reaction to the diagnosis of infectious diseases. Clin Infect Dis 1999; 29: 475–488. 16 Chen SQ, Chou WH, Blouin RA, Mao ZP, Humphries LL, Meek QC et al. The cytochrome P450 2D6 (CYP2D6) enzyme polymorphism: screening costs and influence on clinical outcomes in psychiatry. Clin Pharmacol Ther 1996; 60: 522–534. 17 Bassett ML, Leggett BA, Halliday JW, Webb S, Powell LW. Analysis of the cost of population screening for haemochromatosis using biochemical and genetic markers. J Hepatol 1997; 27: 517–524. 18 Bapat B, Noorani H, Cohen Z, Berk T, Mitri A, Gallie B et al. Cost comparison of predictive genetic testing versus conventional

19

20

21

22

23

24

25

26

27

28

29

30

clinical screening for familial adenomatous polyposis. Gut 1999; 44: 698–703. Beutler E, Gelbart T. Large-scale screening for HFE mutations: methodology and cost. Genet Test 2000; 4: 131–142. Rosenberg T, Jacobs HK, Thompson R, Home JM. Cost-effectiveness of neonatal screening for Duchenne muscular dystrophy—how does this compare to existing neonatal screening for metabolic disorders? Soc Sci Med 1993; 37: 5541–5547. Van der Riet AAPM, van Hout BA, Rutten FFH. Cost effectiveness of DNA diagnosis for four monogenetic diseases. J Med Genet 1997; 34: 741–745. Vintzileos AM, Ananth CV, Smulian JC, Fisher AJ, Day-Salvatore D, Beazoglou T. A cost-effectiveness analysis of prenatal carrier screening for cystic fibrosis. Obstet Gynecol 1998; 91: 529–534. Bartell SM, Ponce RA, Takaro TK, Zerbe RO, Omenn GS, Faustman EM. Risk estimation and value-of-information analysis for three proposed genetic screening programs for chronic beryllium disease prevention. Risk Anal 2000; 20: 87–99. Chou WH, Yan FX, de Leon J, Barnhill J, Rogers T, Cronin M et al. Extension of a pilot study: impact from the cytochrome P450 2D6 polymorphism on outcome and costs associated with severe mental illness. J Clin Psychopharmacol 2000; 20: 246–251. Veenstra DL, Higashi MK, Phillips KA. Assessing the cost-effectiveness of pharmacogenomics. AAPS Pharmsci 2000; 2: Article 29. (http://www.pharmsci.org/) Horsemans Y, Lannes D, Pessayre D, Larrey D. Possible association between poor metabolism of mephenytoin and hepatotoxicity caused by Atrium, a fixed combination preparation containing phenobarbital, febarbamate and difebarbamate. J Hepatol 1994; 21: 1075–1079. Russell LB, Gold MR, Siegel JE, Daniels N, Weinstein MC. The role of cost-effectiveness analysis in health and medicine. J Amer Med Assoc 1996; 276: 1172–1177. Siegel JE, Weinstein MC, Russell LB, Gold MR. Guidelines for pharmacoeconomic studies. Recommendations from the panel on cost effectiveness in health and medicine. Panel on cost effectiveness in Health and Medicine. Pharmacoeconomics 1997; 11: 159–168. Weinstein MC, Siegel JE, Gold MR, Kamlet MS, Russell LB. Rcommendations of the panel on cost-effectiveness in health and medicine. J Amer Med Assoc 1996; 276: 1253–1258. Garber AM, Phelps CE. Economic foundations of cost-effectiveness analysis. J Health Econ 1997; 16: 1–31.