Since the mid-1990s, a number of new, targeted therapies, including cetuximab, have been introduced to treat metastatic colorectal cancer. Despite being dramatically more expensive than older regimens, the targeted therapies produce relatively modest survival benefits and typically improve median survival by no more than several months ( 1–3 ). Recently, it has been recognized that patients with tumors that bear a wild type KRAS gene are responsive to cetuximab, whereas patients with tumors bearing KRAS mutations are not ( 4 ). Following this recognition of the predictive value of tumor KRAS mutation status, several practice guidelines began recommending KRAS mutation testing to identify patients for whom cetuximab treatment would not be associated with any clinical benefit ( 5 , 6 ). This practice has rapidly become standard. Restricting cetuximab to colorectal cancer patients with wild-type KRAS tumors has been heralded as a step toward the realization of personalized cancer therapy ( 7 ). Such individually tailored treatment strategies can potentially help address the problems of expensive treatment by limiting their administration to the patients who are most likely to benefit, potentially leading to a more favorable cost-effectiveness of treatment.
In this issue of the Journal, Mittmann et al. ( 8 ) reported the cost-effectiveness of cetuximab monotherapy in advanced colorectal cancer patients who participated in the National Cancer Institute of Canada trial CO.17. They found that even when cetuximab therapy is restricted to patients with wild-type KRAS tumors, the incremental cost-effectiveness ratio (ICER) and cost–utility ratio (in 2007 Canadian dollars) were $120 061 per life-year gained and $186 761 per quality-adjusted life-year (QALY) gained, respectively. For the entire study population, the ICER and cost–utility ratio were $199 742 and $299 613, respectively.
As expected, based on differential response to cetuximab by KRAS mutation status, the cost per life-year and cost per QALY were more favorable when the analysis was restricted to the subset of patients with wild-type KRAS tumors. Importantly, these estimates do not include the cost of KRAS mutation testing because patients in both arms of the trial received testing and including it in the costs would not have affected the incremental difference. Extrapolation of these trial-based estimates to the general population would need to consider the cost of KRAS mutation testing and would likely lead to even higher ICERs for cetuximab therapy outside a trial setting. These findings raise important questions about the use of cost-effectiveness analysis in policy making. In a setting with unlimited resources, all effective therapies—even those with modest survival benefits—could be offered to every patient. But, because in most settings, resources are limited, publicly funded health-care programs, health plans, and other policy makers must weigh the costs and benefits of different treatments and make decisions about which treatments will be covered and under what circumstances.
The rationale for any cost-effectiveness analysis is that it offers an explicit and transparent approach to quantify the costs and benefits of a prevention or treatment strategy by using a common denominator (ie, years of life saved, QALYs saved). The resulting cost-effectiveness ratios can then be compared across conditions with each other or with a threshold value, with the goal of identifying the most efficient ways of maximizing health at the population level ( 9 ). This approach has the potential advantage of facilitating a deliberative, systematic, and data-driven decision-making process for the allocation of public resources.
Critical to this process, then, is an understanding of what an ICER of $120 061 per life-year or $186 761 per QALY actually means. For purposes of comparison, the ICERs for colorectal cancer screening compared with no screening range from $10 000 to $25 000 per life-year saved in 2000 US dollars ( 10 ). Colorectal cancer screening in the United States is currently covered by Medicare, most health plans, and other public insurers. Health-care interventions that are perceived as having “high value” typically have ICERs in this range ( 11 ). Although there is no universally accepted cost-effectiveness benchmark in the United States, values of $50 000–$100 000 per QALY are commonly cited thresholds in the medical literature below which therapies are considered cost-effective and above which therapies are considered less cost-effective ( 11 ). However, these cited thresholds are theoretical rather than practical. The cost-effectiveness of treatment is not used as an explicit criterion for coverage or regulatory decisions in the United States. The Food and Drug Administration considers effectiveness, but not cost-effectiveness data, when it makes decisions about drug approval. Medicare, the largest health-care payer in the United States, is legally bound to cover treatments viewed as “reasonable and necessary,” which has not been interpreted to include a consideration of the economic cost of treatment ( 12 ).
In contrast to the United States, Canada, the United Kingdom, and other countries with single-payer health-care systems do consider ICERs when making decisions about what treatments they will cover. Even in the context of such health-care systems in which ICERs can be considered, the systematic use of cost-effectiveness data faces methodological and technical challenges that are related to differences in methods and perspectives across studies and the differential use of comparators and sensitivity analyses. Moreover, the relevant comparators change over time as new standards of care are identified. Although the ICERs for cetuximab treatment in patients with metastatic colorectal cancer and wild-type KRAS reported by Mittmann et al. exceed the commonly accepted $100 000 per QALY threshold that signifies a “good value” health-care intervention, this treatment is currently available and covered by some health plans for colorectal cancer patients in the United States. ICERs for cetuximab vs no cetuximab are similar to those for other cancer therapies that are routinely covered by US health-care payers, such as HER2 testing and trastuzumab treatment for breast cancer patients with HER2 protein overexpression ( 13 ).
In the United Kingdom, the National Institute for Clinical Excellence uses cost-effectiveness data to inform coverage decisions for the National Health Service. ICER values below the range of £20 000–£30 000 per QALY (approximately $33 000–$49 000 per QALY in US dollars) are thresholds that have been used by National Institute for Clinical Excellence in coverage decisions ( 14 ). Although cetuximab is not currently covered by the National Health Service ( 15 ), it is available to patients in the United Kingdom who choose to purchase it privately. National Institute for Clinical Excellence's deliberations about coverage of drugs used to treat patients who are at the end of life includes some flexibility and consideration of additional criteria, including consideration of the size of the patient population at risk, the degree to which the treatment is targeted to patients with short life expectancy, and the availability of alternative treatments ( 16 ). In Canada, deliberative groups similar to National Institute for Clinical Excellence incorporate cost-effectiveness data into formulary recommendations for publicly funded health plans. Each province has some autonomy and discretion about whether treatments will be covered. Cetuximab is currently available and covered in some provinces in Canada. ICERs are also increasingly being used in European countries to inform their decision making about coverage ( 14 , 16 ).
How could cost-effectiveness information about oncology treatments inform health policy in the United States? In practice, some therapies with ICERs that are well in excess of the noted thresholds are currently covered, and it is anticipated that many therapies under development will also have ICERs well above these thresholds. The sustainability of this current approach to decision making about coverage is unclear, particularly in light of escalating health-care costs. However, there remains a profound reluctance by publicly funded health-care programs, health plans, and other policy makers to consider economic information in making coverage decisions.
ICERs can be a useful metric for informing coverage decisions because they can be used to systematically compare the value of health interventions across conditions. Yet, the context of the health intervention is also increasingly being recognized as an important part of the decision-making process. For example, even if the ICERs for two health interventions are formally and quantitatively equivalent, one intervention might offer the potential for extending life for patients with a terminal illness and the other might represent the less tangible benefit of preventing illness across a large population. Other considerations for coverage decision making include the budgetary impact of the intervention and availability of treatment across patient populations ( 11 ). Importantly, coverage decisions need not be a binary yes or no. Alternative approaches to coverage could incorporate value-based pricing principles, in which prevention or treatment strategies with very low ICERs (eg, colorectal cancer screening) might be completely covered without a patient copayment, and other treatments with slightly higher and much higher ICERs could be partially covered but have slightly higher and much higher patient copayments, respectively. This value-based pricing strategy may exacerbate health disparities, however, particularly for low-income cancer patients who may not have the means to pay for treatment out of pocket. Other coverage models merit scrutiny. For example, in the United Kingdom, a payment-for-results strategy is currently being used, wherein the National Health Service reimburses for the drug only when it has a therapeutic effect.
How does cost-effectiveness information affect oncology practice in the United States? Researchers and professional societies have developed guidance for oncologists about how to talk to patients about the cost of treatment ( 17 , 18 ), suggestions that are based, in part, on the growing concerns about health-care costs and cost burden for the patient. Yet, according to recent surveys of oncologists in the United States, most reported that patients should have access to effective drugs regardless of the cost ( 19 , 20 ). Most oncologists also reported that although costs influenced their treatment recommendations, they did not feel prepared to interpret cost-effectiveness data ( 20 ), and many do not talk to patients about drug costs ( 20 , 21 ). Thus, it is unclear how cost-effectiveness information could inform clinical practice in the absence of policy changes.
Although “personalizing” therapy by identifying the subset of metastatic colorectal cancer patients most likely to benefit makes the ICER for cetuximab therapy reported by Mittmann et al. more favorable, it is still higher than thresholds commonly used to identify “good value.” In the United States and other countries that explicitly consider ICERs in decision making, publicly funded health-care programs, health plans, and policy makers face many challenges related to the escalating cost of cancer care. Cancer patients and their physicians also face challenges related to the affordability of care. Health-care spending as a percentage of gross domestic product has grown steadily in the United States, and is expected to continue growing. Systematic and transparent approaches for assessing value for money are needed to ensure that increases in the costs of cancer care are accompanied by increases in high-quality effective cancer care.