Tackling the Pricing Challenges for Advanced Therapies for Rare Diseases

In 2022, three gene therapies for rare conditions won approval in the United States. As these and other advanced therapies make it to market, drug companies and payers need to wrestle with pricing issues, particularly for one-and-done therapies that are potentially curative. We spoke to Alice Valder Curran, partner with Hogan Lovells, about the challenges of value-based pricing for gene therapies, some of the pricing approaches gene therapy developers are employing, and how the existing policy landscape complicates matters.

Daniel Levine: Alice, thanks for joining us.

Alice Valder Curran: My pleasure.

Daniel Levine: We’re going to talk about advanced therapies, pricing, and new payment schemes designed to address the challenges of payers, both absorbing the cost and being willing to pay for such therapies when their benefits and durability may be viewed as uncertain. The issues are by no means limited to one and done gene therapies. But for the purpose of this discussion, let’s focus on those from a pricing point of view. How do developers arrive at a price? What are the parameters and considerations in setting a price for these therapies?

Alice Valder Curran: A lot of different things, and while the particular drivers of a price, I think, are almost certainly going to differ by therapy, common factors that developers look to include things like the investment and cost in developing the product itself, the cost of alternative treatments not just in terms of other therapies, but also services. If you are going to take a product, a therapy, and it’s going to limit the need for provider care, hospital care, something like that, those alternative costs are also considered. You, of course, look at what competitors, if you have any in the market, and they might not be head to head, but they could be in the same therapeutic class or maybe treating symptoms where you want to cure the underlying disorder. And then, of course, a huge focus on the benefits to the patient—both productivity, risk reduction, disease severity, the benefits of adherence, the benefits to their own family and other care providers in terms of the burdens that might be alleviated with that care. Those are some of the types of factors that I think developers look to; but again, I can’t emphasize enough how each product is different and so the valuation process is going to be different for them and it’s really got to focus on what that particular therapy brings to the patient population.

Daniel Levine: If you think about gene therapies for rare diseases, these therapies, when approval with relatively limited data and experience, how concerned are payers about the long-term durability of these therapies, and how does that complicate recognition of their value or the willingness of payers to support their use?

Alice Valder Curran: Well, payers, of course, are worried about the long-term durability of these therapies, just like any purchaser is going to be for any product or service that has a material cost associated with it. And it is a challenge because as you note oftentimes these products do come to the market without a really extended longitudinal amount of data to tell you how durable and efficacious a product is beyond perhaps the two-year threshold. That said, one of the things to keep in mind with a lot of payers, or most payers, is that patients often cycle in and out of a given payer in the course of a year. And so most payers—I think certainly in my experience—payers recognize that if they invest in paying for this product for a particular patient, that patient may not be with that payer in 12 months or 24 months because the patient population is pretty mobile. And so, while they care about the durability, because it goes to the foundational principle of whether the price is worth it I, sometimes they can’t really count on banking those savings, not because of any problem with the therapy, but the practical reality that that patient may not be with them as insured by that plan by the time those savings would accrue.

Daniel Levine: Given that there is a move for high price therapies to spread payment over time, does that make payers more willing to accept such a payment scheme, or does that make them less willing to do so?

Alice Valder Curran: You know, it’s interesting. There has been a lot of discussion around payment over time. And in my experience, it’s something that’s out there, but that very few payers currently are focusing on, for the simple reason that [for] the gene therapies and cell therapies that are out there with very high price tags, the volume of patients that they treat is still relatively low. So as you mentioned, most of these gene therapies are for rare diseases or conditions, so very small populations. So any given payer may only have one or two or three patients in the course of a given year. And so even if you’re looking at the highest price tag for these products, you’re looking at an annual expenditure of under $10 million and that might not be worth the complexity of trying to spread out that cost with payments over time. So, it’s definitely a topic that people have explored, and there’s been some guidance, even from CMS and the Medicaid program about how to accommodate payment over time arrangements. But so far, in my experience, there really hasn’t been any uptake on them simply because the bullets of payments that any particular plan might have to cover in a particular plan year is not so huge that they have an incentive or a need to resort to really a new payment model that is still being trialed.

Daniel Levine: As companies are winning approval or moving towards approval, there are two basic approaches we’re seeing implemented. The first of these is to use outcomes-based or performance based-measures. Bluebird is doing that with Zynteglo, its beta thalassemia gene therapy. How do these arrangements generally work?

Alice Valder Curran: Well, I’ll start just with what Bluebird announced publicly for its product, which is that it is promising to refund up to 80 percent of the product’s cost, and its announced WAC [wholesale acquisition cost] is $2.8 million. If that patient doesn’t achieve and maintain transfusion independence for up to two years post infusion, what that effectively would look like probably is a rebate of some sort of up to 80 percent of WAC or something near WAC if that performance doesn’t occur. Payers and developers out there are used to rebate agreements with healthcare plans and PBMs. So, those arrangements are probably going to have a lot of those same attributes except that the trigger for that rebate, in effect that refund, is not going to be formulary performance or some other requirement like that, but rather tracking a start date for when the infusion occurs, and then monitoring claims or other data over the course of that succeeding two-year period to see if transfusion independence is achieved and maintained. So, there is a requirement for these sorts of outcomes- and performance-based arrangements to be able to define very specifically what performance or the good or bad outcome is to be able to define that very clearly and to be able to identify the data that the plan and the developer can look to, to measure whether that performance has occurred.

Daniel Levine: The other thing we’ve seen is the introduction of what these companies call a warranty, which BioMarin says it’s planning to do for its Roctavian hemophilia A gene therapy. This is not yet approved in the United States, and the company hasn’t detailed it’s plan, but this is intended to be a risk-sharing approach. Is there any difference between a warranty and an outcomes based approach?

Alice Valder Curran: Yes, there is a difference. They are similar in that they both look to provide some sort of price adjustment where the product does not perform as expected or as hoped. So you sort of define what a positive outcome is, and if that outcome is not achieved, there is some sort of price reduction that’s effectuated with the model I just talked about. The typical model is effectively a rebate where outcome doesn’t occur. What’s different about a warranty is that instead of the developer issuing that rebate or refund, it’s a third party. Effectively what happens in the case of a warranty is the developer has taken out an insurance policy with a third party insurer that says, we want to buy insurance that our product is going to work. Again, you have the same sort of definitions about what performance looks like, what failure to perform looks like, what the outcome is, and the data are that you’re going to look at. But in the case of a warranty, when that failure to perform or that negative outcome occurs, instead of a rebate being issued, this third party insurer issues some form of refund or payment to the payer. And so, it’s basically a third party payment, a type of insurance payment to the to the customer, to the payer, rather than a rebate from the developer to the payer—similar but it involves a third party in the mix in the transactions around how that performance or lack of performance is used to adjust the price.

Daniel Levine: As you’ve noted, any of these arrangements will require some form of data collection and monitoring and some agreed upon measure to determine sustained efficacy. How will this work? Who will be responsible for the data monitoring, and will the measures be negotiated or set by the drug developer?

Alice Valder Curran: It’s going to vary by arrangement, but all of these are going to be subject to an agreement. So, my expectation is that some level of negotiation is going to be involved and typically the payer is going to be the first source of whatever the data are for demonstrating efficacy or durability, but the manufacturer may have access to data as well. It’s hard to tell, but the agreement is going to define the performance parameters. The agreement would define the data that you’re going to look to. It would define the frequency of checking on performance, the submission of data, the ability of the developer to validate those data. And there certainly could be negotiation around what satisfactory performance does and does not look like and the data that you could use. But my expectation is that any developer that’s going to a payer with this sort of proposal is going to have a clear idea of what they think performance looks like, the data that they think are needed in order to demonstrate performance or absence of performance, and then it’s up to working with the payer to make sure that’s workable because there are very clear operational issues with making these work, right? The payer has to be able to access the data, the data has to be at sufficient granularity to be indicative of performance as to an individual payer for an individual plan. And so, even when all parties can agree on what performance looks like and the data you need, there can be very nuts and bolts conversations around data access, transmission validation, and the like to determine whether performance has occurred.

Daniel Levine: And a patient has to be available and willing to be subjected to those measures.

Alice Valder Curran: Yes, of course that’s true, but in my experience, all the parties just want to be able to use data that’s readily accessible, which is why claims data are the easiest to use. You can look at a claims history about whether follow-on services were needed that wouldn’t have been needed if the product was used, for example. Or if a patient had to be on a particular maintenance therapy before the gene therapy, the gene therapy should have ceased the need to have that. If then you see the patient going back on that original maintenance therapy, that could be an indication. So, I do think that there’s a desire to not have to have the patient give particular data, but wherever possible use data that’s already digitized and accessible for all the parties.

Daniel Levine: In the United States, there’s a complex and fragmented payment landscape made up of private insurers, corporate self-insurers, and Medicaid among others. How does this complicate the creation of workable payment approaches?

Alice Valder Curran: It certainly does complicate precisely because of the challenge I mentioned earlier, which is patient movement between payers. So, particularly if you want to track how a patient has been doing post-treatment, if that patient leaves that health plan at the end of the calendar year, it can be a challenge to figure out whether that product has continued to be efficacious in the months 13 to 24, for example. If that patient’s moved to a different payer, it’s unclear how you’re going to find those data. So, I think one of the biggest challenges is patient movement across plans so that you can actually track the durability and efficacy that is the organizing principle of these sorts of value based arrangements. Another challenge that I’ve seen involves, particularly with smaller plans and with self-insured employer plans, when they encounter one of these therapies, if they’re self-insured and they’re relatively small insured, it can be a real challenge to cover the cost of even one of those because that really can be a material impact on the spend for that employer in a given year. So, we are starting to see increasing reliance on reinsurance models to address that. But even as I understand it, even the reinsurance market is becoming challenged or certainly may be challenged as we see more and more of these therapies come to market over the next decade. So, both these small plans and the movement of patients across plans are certainly a couple of the different challenges that the nature of our insurance market here presents for these therapies.

Daniel Levine: What is the Medicaid best price requirement, and what are the implications of that for the use of value-based pricing? How much of a barrier is that to implementing this?

Alice Valder Curran: So, the Medicaid drug rebate program, which was created in 1991, is a program that a developer has to participate in if they want the Medicaid program to pay for any of products. And that statute was created in a market that’s three decades old, frankly doesn’t look anything like our market is today, but it had this very simple foundational principle, which is that the Medicaid program, because of the volume of patients, effectively ensures nationwide that it should get the benefit of the best price or the lowest price that a developer provides on its therapy to anyone in the commercial market in the United States. Given its volume, it should get the biggest discount that a drug company is willing to give to any other of its best customers. That best price concept intuitively makes sense, but it also is a bit of a blunt instrument because one low price on one unit on one patient in a given quarter can drive the discount level nationwide for the entirety of the Medicaid patient population. The challenge it presents for value-based pricing arrangements is developers want to stand behind the value of their product, and so they want to be able to give a significant refund or rebate if a product doesn’t perform. But let’s say a company wants to give a 100 percent rebate if the product shows a failure of performance within 12 months of being infused. In that case, you’d have a $0 price for that therapy. Now, that $0 price only applies if the therapy doesn’t work, but under the old Medicaid best price rules, the manufacturer would have to report a $0 price as its best price, which would mean that the entirety of the Medicaid program would get a 100 percent discount on every unit that they used, regardless of whether that product worked. So, the manufacturer would only want to make that $0 price available if the product didn’t work, but the best price standard would require that price for all units regardless of whether they would work. So, it’s sort of this windfall problem. And the converse of that would be true as well. So, if Medicaid was covering this product and it always worked with patients in the commercial sphere, so the manufacturer never offered a 100 percent discount to anyone, but it did fail. In the case of a Medicaid patient, the manufacturer normally would want to give the Medicaid program a 100 percent rebate because that’s how they stand behind the value of their product. But there wouldn’t be a mechanism to do that. They would just get whatever their standard statutory rebate is, because there wouldn’t have been a $0 price in the commercial market to set that best price. And so that really worked to chill manufacturer developer interests and ability to do value-based pricing arrangements for these therapies. I’ll add that one of the tools that did exist in the program over time that could have enabled this as a concept is called a bundled sale, which basically says that if you offer value-based pricing on some units, but pay them out on some units but not others, you generate a weighted average discount across all of the utilization, and it’s the weighted average discount that you use when you determine best price. And so, if you have a larger population, you might end up giving big discounts on a handful of patients, but the vast majority of patients there are no discounts. So, the weighted average discount rate might be 10 percent overall and that might be something that the manufacturer could live with. The problem with gene therapies, at least the gene therapies that have been approved so far, is, as you noted at the top, they often treat these rare diseases. So, the plans that cover these products might only have one patient in a year, and so you can’t do a weighted average with an N-of-one because then if you give a 100 percent discount, you have a $0 price. So, these small population products couldn’t take advantage of this bundled sale option that existed under Medicaid because the populations they treat are too fragmented and small for each individual payer. And so, they really did need a solution if they were going to start engaging in value-based pricing arrangements and still be in Medicaid.

Daniel Levine: The Centers for Medicare and Medicaid services sought to address the issue with a rule that went into effect in July. What did that rule do? And does it provide a solution?

Alice Valder Curran: It does provide a solution. It’s an optional approach that developers can choose to do annd it’s called the multiple best prices reporting option. So, I just mentioned how best price is this single lowest price provided in the course of a calendar quarter to any commercial customer in the United States? The multiple best prices option says, when you offer a value-based pricing arrangement, so let’s say the developer says, “I’ll give you a 100 percent rebate if it shows a failure of performance within the first 12 months after infusion, and I’ll give you a 50 percent rebate if the product shows failure to perform in months 13 to 24, and after that you get no rebate.” Well, then the manufacturer would say, we are offering this price, this product at three different prices, we’re offering at a $0 price, we’re offering it at a 50 percent discount price, and we’re offering it at a zero percent discount price, or full price. And it’s tied to those different levels of performance. So, the manufacturer then reports three best prices, that’s why it’s called the multiple best prices option. And they report a $0 price, a 50 percent discount price and a full freight price. And then they define the performance that is linked to each one of those prices. The manufacturer has to offer that arrangement to any state Medicaid program that is interested, and for those state Medicaid programs that opt in, those state Medicaid programs then get access to the three tiers of pricing tied to the performance of the product in each particular patient. So it achieves the overall statutory goal of parity between what a developer offers to the commercial market and making sure that Medicaid programs get the benefit of the best offer, the best price that is out there in the commercial market so that they can opt in. And if a Medicaid patient receives a therapy and it shows a failure to perform within the first 12 months after infusion, the Medicaid program will get a hundred percent rebate just like a commercial customer would, but only where that performance matches the price. So only if it fails to perform in the first 12 months and so forth, across the three different tiers of pricing. And so that enables manufacturers to offer this value-based price in the commercial market and not worry that they have to give the extreme discounts to Medicaid even when the product is fully performing—so it does enable it because in my experience, developers are very willing to offer these value-based arrangements to Medicaid as well as the commercial market if there was a way to make sure that there was a parity and approach, and this multiple best prices reporting option does create that pathway should a developer want to go in that direction.

Daniel Levine: Does it leave other issues that need to be resolved?

Alice Valder Curran: Yes, the one issue that still needs to be resolved is how to treat these arrangements in the average sales price calculation, or ASP, which is the price that has to be reported for use in setting payment amounts under Medicare Part B for physician administered products. And these gene therapies to the extent they treat a Medicare population, would be the type of therapies that are payable under Part B. The reason it’s a challenge is the average sales price for a product is defined by reference to sales that are counted in best price. So, it’s what you count in ASP is defined directly by reference to what you have to count in best price. Well, with the multiple best prices option, a developer reports two sets of best price data. They report the three price tiers I referred to in my previous example for those cases where a state opts into the multiple best prices option. But it also reports a non-value-based price, best price for those situations where the state Medicaid program does not choose to engage in the value-based pricing arrangement. So, there are two sets of best price data for any manufacturer that opts into the multiple best prices recording option. The question is, which of those data sets do we use in ASP? Do we use the value-based pricing data or do we use the non-value-based pricing best price data? This is unresolved. It’s not acute in terms of urgency at the moment because first of all, the multiple best prices rule, as you noted, just went into effect a few months ago. But also, the gene therapies that are on the market now that are the products that are most likely to be interested in, or be subject to ASP reporting, typically don’t treat Medicare populations. They’re usually treating pediatric populations. But as we have more and more of these therapies come to market as there is the potential for Medicare beneficiaries to potentially be treated by these therapies, the need to address ASP is going to become more urgent. And I’ll make a plug for the fact that I think CMS can readily interpret the statutory definition of ASP to only look at the non-value-based pricing best price data and not look at the multiple best prices value-based pricing data in determining ASP and that certainly would clear the path and remove any disincentives to manufacturers for entering into these arrangements if they have ASP exposure. We just need the Medicare program to issue guidance on this, one way or another.

Daniel Levine: There have been limited examples of value-based pricing, but these agreements have been around for 10 years or so outside of gene therapies. What have we learned from the experience with them and is that translatable to one-time gene therapies?

Alice Valder Curran: Danny, you’re absolutely right. These have been around for a while. I think the difference is that the value-based pricing arrangements that have existed over the last decade have targeted or involved therapies that involved much larger patient populations and because they involve much larger patient populations, in my experience, although I don’t know if this is what those developers did in my experience, it certainly would’ve been possible to manage the best price risk for those arrangements through this bundled sale approach where the developer could say, if the product fails and you the payer incur additional costs for that reason, we’ll rebate you data, but we’re going to allocate those rebate dollars across the totality of utilization you have for that particular product. And so, when you sort of allocate or smooth those discount dollars over the value of all the utilization in a quarter or a year, the weighted average discount rate is manageable from a Medicaid rebate or best price perspective. And that’s what I assume has happened. But to be clear, that’s just an assumption. I don’t know, because we’re not talking about any specific examples and I don’t know if that’s what was done, but I think that’s likely how it was managed. And it’s precisely because with these gene therapies that bundled sale tool wasn’t a solution because they treat these very small patient populations where any given plan might have one, two, at max three patients a year, there was too much risk in assuming that if you did have to pay out a large rebate to account for performance, you couldn’t be sure that the weighted average discount rate would be low enough that you could manage it from a rebate perspective. And that was one of the driving forces, I believe, behind the need to create this alternative pathway and this multiple best prices model that CMS was willing to do. I want to give CMS credit for standing behind its public statements about the desire to enable and support value-based pricing arrangements. Finalizing this regulation creating this pathway was really putting that promise into action. Now it’s up to developers to find ways to use it and to be able to bring those value-based pricing arrangements to the market.

Daniel Levine: Alice Valder Curran, partner with Hogan Lovells. Alice, thanks so much for your time today.

Alice Valder Curran: Thank you, Danny. It’s been my pleasure.