Cost Implications of Value-Based Pricing for Companion Diagnostic Tests in Precision Medicine

Abstract

Many interpretations of personalized medicine, also referred to as precision medicine, include discussions of companion diagnostic tests that allow drugs to be targeted to those individuals who are most likely to benefit or that allow treatment to be designed in a way such that individuals who are unlikely to benefit do not receive treatment. Many authors have commented on the clinical and competitive implications of companion diagnostics, but there has been relatively little formal analysis of the cost implications of companion diagnostics, although cost reduction is often cited as a significant benefit of precision medicine. We investigate the potential impact on costs of precision medicine implemented through the use of companion diagnostics. We develop a framework in which the costs of companion diagnostic tests are determined by considerations of profit maximization and cost effectiveness. We analyze four scenarios that are defined by the incremental cost-effectiveness ratio of the new drug in the absence of a companion diagnostic test. We find that, in most scenarios, precision medicine strategies based on companion diagnostics should be expected to lead to increases in costs in the short term and that costs would fall only in a limited number of situations.

Appendix

1.1 Full Details for Scenario 1

Figure 2 shows the northeast quadrant of the CE-plane for the comparison of the new drug versus the status quo. If the drug was free, then there would be total incremental health benefits ${\text{TH}}_{\text{All}}$ and incremental non-drug costs $C_{\text{All}}^{\text{N}}$, represented by point A in the CE-plane. Let $C_{\text{All}}^{\text{D}}$ be the total expected drug cost when the Treat All strategy is used, thus yielding a total incremental cost in the Treat All strategy of ${\text{TC}}_{\text{All}} \;{ = }\;C_{\text{All}}^{\text{N}} \;{ + }\;C_{\text{All}}^{\text{D}}$. For the Treat All strategy to be cost effective, the drug cost must be such that the incremental cost of the Treat All strategy, relative to the Status Quo, is less than the payer’s WTP for the benefits obtained by the Treat All strategy. The dashed line from the origin to point C has a slope equal to the payer’s maximum WTP for benefits. Thus, the cost at point C represents the maximum WTP for benefits of ${\text{TH}}_{\text{All}}$. By the definition of this scenario (Treat All is cost effective), the drug manufacturer has chosen $C_{\text{All}}^{\text{D}}$, such that the total cost ${\text{TC}}_{\text{All}}$ sits at some point B along the vertical line between points A and C. The analysis of Scenario 1 depends on whether the untargeted group would receive positive or negative health benefits, as described in subcases 1-a and 1-b, below.

1.2 Subcase 1-a: Positive Health Benefits Among Patients in the Untargeted Group

In this subcase, the untargeted group experiences positive but low health benefits from the new drug (Fig. 4). We first assume that the companion diagnostic test is free. Since Targeting results in fewer people receiving the new drug, and since members of the untargeted group would have experienced positive incremental health benefits from taking the new drug, the total non-drug costs and health benefits are reduced. Let ${\text{TH}}_{\text{Tar}}$, ${\text{TH}}_{\text{Tar}} \;{ < }\;{\text{TH}}_{\text{All}}$, be the total health benefits achieved in the Targeting strategy, and let $C_{\text{Tar}}^{\text{N}}$ $C_{\text{Tar}}^{\text{N}} \;{ < }\;C_{\text{All}}^{\text{N}}$ be the total non-drug incremental healthcare costs in the Targeting strategy. When the drug cost is not included, the incremental cost and health benefits shift to point D in Fig. 4a. When the drug price is included, the incremental cost increases to $C_{\text{Tar}}^{\text{N}} \;{ + }\;C_{\text{Tar}}^{\text{D}}$ (point E, Fig. 4a). However, this figure does not represent the full incremental cost of the Targeting strategy because the cost of the test itself has not been included.

Fig. 4

Targeting strategy when the untargeted group accrues positive health benefits. a Result of switching to the Targeting strategy when the cost of the test is not included. b Result of switching to the Targeting strategy for relatively low test cost. c Result of switching to the Targeting strategy for moderate test cost. d Result of switching to the Targeting strategy for relatively high test cost. WTP willingness to pay, ICER incremental cost-effectiveness ratio, $TC_{All}$ total incremental cost of the Treat All strategy, $TC_{Tar}$ total incremental cost of the Targeting strategy, $C_{Tar}^{N}$ total incremental non-drug costs in the Targeting strategy, $C_{Tar}^{D}$ total cost of the new drug in the Targeting strategy, $TH_{All}$ total incremental health benefits in the Treat All strategy, $TH_{Tar}$ total incremental health benefits in the Treat All strategy

Let $C_{\text{Tar}}^{\text{T}}$ be the expected testing cost per person when the Targeting strategy is used. The cost $C_{\text{Tar}}^{\text{T}}$ includes people in both the targeted and untargeted groups since all members of the population would need to be screened in order to implement the Targeting strategy. When the cost of the test is included, the resulting total incremental cost is ${\text{TC}}_{\text{Tar}} \;{ = }\;C_{\text{Tar}}^{\text{N}} \;{ + }\;C_{\text{Tar}}^{\text{D}} \;{ + }\;C_{\text{Tar}}^{\text{T}}$. There are two important thresholds for total cost, shown by points H and G in Fig. 4a. Below, we discuss three cases defined by the value of ${\text{TC}}_{\text{Tar}}$ relative to these two thresholds.

The first possibility is that the test manufacturer chooses a relatively low test price (Fig. 4b), which results in the Targeting strategy sitting at point F₁, where F₁ is on the line segment between E and H, with ${\text{TC}}_{\text{Tar}} \;{ < }\;{\text{TC}}_{\text{All}}$. Since a line from the origin to F₁ has a lower slope than one from the origin to B, the Targeting strategy is more cost effective than the Treat All. However, Treat All is still a policy option, so we consider the incremental cost effectiveness of switching from Targeting to Treat All. In this instance, the slope of the line segment from F₁ to B is greater than the WTP threshold. Thus, Treat All is not cost effective relative to Targeting. The Targeting strategy results in a reduction in total healthcare costs (i.e. since ${\text{TC}}_{\text{Tar}} \;{ < }\;{\text{TC}}_{\text{All}}$) and a reduction in total health benefits (i.e. caused by the loss of benefits of the new drug that would have been experienced by the untargeted group).

The second possibility is that the test manufacturer chooses a slightly higher price, resulting in the Targeting strategy sitting at point F₂ on the line segment between H and G (Fig. 4c). The total incremental cost ${\text{TC}}_{\text{Tar}}$ is still less than the total cost at point G. The slope from the origin to F₂ is less than the slope from the origin to B, so Targeting would be considered cost effective. However, as in the previous case, Treat All is still a policy option, so we consider the incremental cost effectiveness of switching from Targeting to Treat All. In this instance, the slope of the line segment from F₂ to B is less than the WTP threshold. Thus, Treat All is cost effective relative to Targeting. The total incremental costs remain at ${\text{TC}}_{\text{All}}$ despite the availability of a test to target individuals. In this instance, Targeting does not change total healthcare costs. This subcase arises only if the test manufacturer chooses a relatively high price for the test, thereby pricing itself out of the market. However, since the result of choosing a high price is that the Targeting strategy is not adopted and the test manufacturer does not sell any tests, we would not expect a profit-maximizing test manufacturer to pursue this option.

The third possibility is that the test manufacturer chooses a price such that the total cost of the Targeting strategy exceeds the total cost at point G (point F₃, Fig. 4d). In this case, the Targeting strategy is dominated by Treat All (either strict dominance or extended dominance [36], depending on the value of ${\text{TC}}_{\text{Tar}}$). Thus, the Treat All strategy is adopted, total incremental costs remain at ${\text{TC}}_{\text{All}}$ and total benefits remain unchanged. As in the previous instance, we would not expect a profit-maximizing test manufacturer to choose a price that resulted in this outcome because Treat All would be adopted, and thus, no tests would be used.

To summarize, cost savings (or cost neutrality) should be expected when the drug is cost effective without Targeting, and the untargeted group would experience positive health benefits. This is because a profit-maximizing test manufacturer has an incentive to choose the highest price such that Targeting is cost effective, but it is not cost effective to switch from Targeting to Treat All. At any higher price, the Treat All strategy would be adopted and the test manufacturer would not gain any revenue. In this situation, Targeting results in removing a technology that is effective but not cost effective among some members of the population. That is, Targeting reduces costs not by removing an ineffective or harmful drug but by removing from treatment those individuals who would experience low benefits.

1.3 Case 1-b: Negative Health Benefits Among Patients in the Untargeted Group

Next, we consider the case where the untargeted group experiences negative health consequences by using the new drug. This scenario could occur if the Status Quo treatment was clinically superior for a portion of patients or if the test helped to identify a portion of patients who would experience serious side effects. In Fig. 5, as in Fig. 4, point A represents the total incremental cost if the drug were free, and point B represents the total incremental cost, including the drug cost (both relative to the Status Quo). In this instance, by Targeting, total incremental health increases because the untargeted group can continue using the Status Quo treatment, which is superior to the new drug for that group of patients. Compared with the Treat All strategy, this strategy results in lower incremental healthcare costs (excluding the drug and test costs) and in higher incremental health benefits (${\text{C}}_{\text{Tar}}^{\text{N}} \;{ < }\;C_{\text{All}}^{\text{N}}$ and ${\text{TH}}_{\text{Tar}} {\text{ > TH}}_{\text{All}}$; point I in Fig. 5). Thus, the incremental cost, including the drug cost, would fall somewhere on the line between points I and J.

Fig. 5

Targeting strategy when the untargeted group accrues negative health benefits from the new drug. WTP willingness to pay, ICER incremental cost-effectiveness ratio, $TC_{All}$ total incremental cost of the Treat All strategy, $C_{Tar}^{N}$ total incremental non-drug costs in the Targeting strategy, $TH_{All}$ total incremental health benefits in the Treat All strategy, $TH_{Tar}$ total incremental health benefits in the Treat All strategy

When the test manufacturer chooses the cost of the test, four possibilities arise, depending on the value of ${\text{TC}}_{\text{Tar}}$ relative to points J, K and L: (1) If the test manufacturer chooses a test price such that ${\text{TC}}_{\text{Tar}}$ is less than the cost at point J, then the Treat All strategy is dominated by Targeting, and total healthcare costs are reduced (i.e. ${\text{TC}}_{\text{Tar}} {\text{ < TC}}_{\text{All}}$). (2) If the test manufacturer chooses a test price such that ${\text{TC}}_{\text{Tar}}$ is greater than the cost at point J but less than the cost at point K, then Treat All would be removed from consideration due to extended dominance. The test would be adopted, and total healthcare costs would increase. (3) If the test manufacturer chooses a test price such that ${\text{TC}}_{\text{Tar}}$ is greater than the cost at point K but less than the cost at point L, then the Treat All strategy is no longer dominated. In this case, Treat All would be cost effective relative to the Status Quo, but it would also be cost effective to switch from the Treat All strategy to Targeting. Since the cost is greater at point K than at point B, total healthcare costs would increase (i.e. ${\text{TC}}_{\text{Tar}} \;{ > }\;{\text{TC}}_{\text{All}}$). The reason for the cost increase is that a portion of the population avoids negative health consequences, and the test manufacturer has set a price to capture the payer’s WTP for this benefit (similar to the value-based pricing strategy described elsewhere [37, 38]). (4) If the test manufacturer chooses a price such that ${\text{TC}}_{\text{Tar}}$ is greater than the total cost at point L, then it would not be cost effective to switch from the Treat All strategy to Targeting. The Treat All strategy would be adopted, and there would be no change in strategy or costs as a result of having the test.

To summarize, total healthcare costs decrease in only one of the four cases described above. This case requires the test manufacturer to choose a relatively low price for the test. However, a profit-maximizing manufacturer would always be expected to choose a test price that causes overall costs to rise, relative to Treat All. A test manufacturer that was confident in the payer’s WTP would be expected to choose as high a price as possible, so it would be cost effective to switch from Treat All to Targeting.