Rationalizing Treatment and Coverage Decisions With Predictive Biomarkers

Beyond providing prognostic information, molecular biomarkers can help optimize therapy decisions. Predictive biomarkers are beginning to help physicians assign value to associated therapies in terms of likelihood of benefit and benefit-to-risk profiles for individual patients. A predictive factor is a baseline patient or tumor characteristic that identifies a specific qualitative outcome (eg, response or survival) driven by a treatment.

Healthcare providers and payers hope to be able to use predictive biomarkers to rationally apportion therapy. “We want to enrich the treated population with those who are likely to benefit from therapy,” explained Mark Green, MD, chief medical officer, Xcenda, who led the roundtable discussion Prognostic vs Predictive Biomarkers and the Role of Companion Diagnostic Testing in Payer Decision Making during the 2012 Academy of Managed Care Pharmacy Annual Meeting. “Individuals likely to benefit should be willing to accept the risks of receiving that therapy— namely, costs for that particular treatment, loss of opportunity to try something else at that point in time, and treatment-related toxicities,” said Green.

Implications for Payers

On the other hand, he said, “Individuals whose tumors fail to express a biomarker directly predictive of a reasonable likelihood of response to a certain therapy would be much better served by exploring other therapies and not exposing themselves to an extremely low likelihood of benefit with no reduction in risk.” The loss of opportunity for trying an alternative therapy that might offer a better chance for response is a significant risk to this group.When a biomarker test is capable of predicting response to a particular therapy, payers should have the opportunity to incorporate that test into their drug benefit management strategy. It is Green’s opinion that by encouraging and covering testing, payers can facilitate rational treatment and coverage decisions. He is expecting that electronic medical records will help bolster awareness, buy-in, and integration of biomarker testing.

According to Green, the field is still finding its way in actualizing the use of biomarkers, not only for therapy selection but for coverage decision making. “It is a new world for providers, payers, patients, and the entire consortium of healthcare stakeholders, but we have to make rational use of them,” he commented.

Applying biomarkers in this way, he cautions, will require discipline and biomarker assays of excellent quality that are associated with high reproducibility, specificity, sensitivity, and accuracy. It will also require stakeholders to agree upon test criteria, thresholds, and decisions related to companion diagnostics. Once that happens, according to Green, plans and providers should be in a position to address the optimum strategy for care. “It is evidence-based, it is science-driven, it is an opportunity cost for optimized patient care and it is a responsible behavior. That is what we are striving for and that is where we need to be going,” he remarked.

Table. Locally Advanced or Metastatic ALK-Positive NSCLC Efficacy Results^a,b

Efficacy Parameters

Study A

(N = 136)

Study B

(N = 119)

Objective response rate,

CR + PR (95% CI)

50%

(42%-59%)

61%

(52%-70%)

Number of responders

68

71

Median duration of response (weeks)

41.9

48.1

CR indicates complete response; NSCLC, non—small cell lung cancer; PR, partial response.

^a Investigated in 2 multicenter, single-arm studies (studies A and B).

^b Investigator-assessed response.

Source: Xalkori [package insert]. New York, NY: Pfizer Labs; 2011. pfizer.com. Accessed May 23, 2012.

Why Bother Testing?

An exciting development in the field of predictive diagnostics is the FDA approval of crizotinib (Xalkori), along with a companion diagnostic test, for patients with ALK-positive locally advanced or metastatic non—small cell lung carcinoma (NSCLC). Although crizotinib costs more than $9000 per month, the response rates in patients with ALK-positive tumors are dramatic. The majority of eligible patients responded to therapy (Table), demonstrating the robust enrichment in likelihood of response achievable with predictive biomarkers.Giving these targeted agents a try in everyone is not an advantageous approach for a number of reasons. These drugs are extremely expensive and there is the issue previously raised about giving someone a therapy from which they are not likely to benefit while exposing them to all the associated costs. The overall rate of response to EGFR inhibitors among patients with advanced NSCLC positive for an EGFR mutation is around 70%, while in patients without an EGFR mutation, it is essentially zero. Use of the EGFR inhibitor gefitinib as initial therapy for patients with EGFR mutation—negative tumors suggests targeted therapy can actually worsen progression-free survival (Figures on Page 2). In addition, some trials show that if you do not test and instead simply randomly assign patients to either erlotinib or chemotherapy as first-line therapy, overall survival may be significantly worse in those individuals who are assigned to receive targeted therapy rather than chemotherapy as first-line management. Green elaborated, “So you have to have the test results if you want to use this targeted therapy in the first-line setting and if the patient carries a mutation and you use gefitinib or erlotinib first-line, the likelihood of response is extremely high.”

Figure. Kaplan-Meier Curves for Primary End Point of Progression-Free Survival in Patients Receiving Either Carboplatin-Paclitaxel or Gefitinib

Green reiterated the goal of predictive therapy as, “You want to be able to dichotomize a population into those who have a high likelihood of benefit and those who do not.” The key, he says, is to avoid missing that person who has a chance to receive a therapy that may provide a 70% to 80% chance of response and has been shown to remain active at 12 months or longer. “We are looking to be able to rationalize, not in the sense of justify, but in the sense of being able to think clearly of how to assign therapeutic recommendations.”

CI indicates confidence interval; EGFR, epidermal growth factor receptor.

A, Overall population of patients; B, patients who were positive for the EGFR mutation; C, patients who were negative for the EGFR mutation; and D, patients with unknown EGFR mutation status.

Analyses were performed on an intention-to-treat population.

Source: Reprinted with permission from Mok TS, Wu Y-L, Thongprasert S, et al. Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med. 2009;361(10):947-957.

Funding Source: None.

Author Disclosure: The author reports no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design; drafting of the manuscript; and critical revision of the manuscript for important intellectual content.