The use of surrogate endpoints in clinical research is becoming increasingly common as a way to accelerate getting drugs to market. But, some experts are critical of aspects of this practice for a number of reasons.
Unlike an outcome that represents a direct clinical benefit, such as survival, decreased pain, or the absence of disease, a surrogate endpoint is merely a prediction of clinical benefit.
The use of surrogates has become particularly prevalent in oncology, where surrogate endpoints like progression-free survival are overtaking cancer’s gold-standard clinical endpoint, overall survival (OS).
Not only do surrogates not necessarily provide real proof on whether a drug works, in some cases, critics argue, that they are poorly understood by both patients and the clinicians aiming to treat them.
London School of Economics associate professor of health policy Dr Huseyin Naci has been speaking out about the use of surrogate endpoints for some time and in a recent paper published in the BMJ argued for more selective use of surrogate endpoints when evaluating new drugs, restricting their use to chronic diseases.
Clinical Trials Arena spoke to Naci to discuss the problems of surrogate endpoints, their limitations and how they can negatively affect patients, clinicians and decision-makers.
Kezia Parkins: How did aducanumab’s approval shine a light on the problem with surrogate endpoints?
Huseyin Naci: I think to those of us closely watching and following FDA drug approvals, it came as no surprise that the FDA would approve a drug on the basis of a surrogate because the vast majority of clinical trials that support new drug approvals are now based on surrogate endpoints alone. I guess the surprising aspect of the aducanumab decision was the fact that there has been quite a large body of literature that looks at the strength of the association between the surrogate for that drug and the clinical endpoint, and not finding any evidence in support of that association. So the FDA actually approving it despite that evidence was quite surprising. Then of course the fact that the FDA panel wasn’t advised about the possibility of approving the drug on the basis of the accelerated approval pathway and then the panel members resigning… definitely created the perfect storm and drew a lot of attention to the issue of surrogates – which lots of people have been writing about and have been concerned about for quite some time.
KP: In what kinds of trials are surrogate endpoints most prevalent?
HN: They’re quite prevalent, especially in cancer drug trials. We see the majority, about four fifths of clinical trials supporting cancer drugs using surrogate endpoints like progression-free survival or response rates. They are also common in other areas like cardiovascular disease where you see LDL cholesterol-lowering used as a surrogate quite frequently. In diabetes trials, HBA1C lowering is used frequently as a surrogate. It’s across the board, but cancer does stand out as being the area that’s using surrogates quite a lot.
KP: What are your concerns about using surrogate endpoints in cancer trials in particular?
HN: This is not unique to cancer; we have the same issues in other therapeutic areas. If the relationship between the intermediate outcome for the surrogate and what really matters for people, which is the clinical outcome, has not been evaluated and the relationship is not strong, then we have no way of knowing if the drug actually will have the intended effect.
So what is quite clear is that the drug will have the side effects already observed but it may not have any of the benefits that we hope to observe. Therefore the benefit-risk balance is really not well established when we’re using surrogate endpoints, and this applies in cancer therapeutic areas.
KP: What are some of the main problems with surrogate endpoints?
HN: They may provide false hope and optimism to patients as well as clinicians. We know from previous surveys that people often misinterpret or misunderstand what FDA drug approvals actually mean and how strong the evidence is when a new drug enters the market. There may not be a lot of awareness or familiarity with this concept of using surrogates to approve drugs so patients and clinicians may incorrectly assume that the drug is beneficial and will extend or improve their lives. That may not actually be established so it certainly sends the wrong signal to patients and clinicians. Using surrogates just doesn’t provide enough information to inform shared decision making in clinical practice. If anything, it actually complicates it.
Another issue for policymakers is that when the FDA, health technology assessment organisations like NICE or any other regulatory agency approves a drug on the basis of a surrogate endpoint then the evidence available on that drug when we want to look at its cost-effectiveness, has to rely on those surrogate endpoints to make projections about how well the drug will work and whether the additional costs of the drug are worth its benefits. Using a surrogate makes it very uncertain how to establish that cost-effectiveness which also poses challenges for the budget, the affordability of the healthcare system, and how we allocate resources. Their use has huge downstream effects for patients, clinicians as well as policymakers.
KP: Can using surrogate endpoints actually draw out processes for drug makers, as they may have to conduct more trials post-approval?
HN: Absolutely right. The key motivation or the rationale for using surrogates is to look at the drug effectiveness at an earlier time point without waiting for that ultimate clinical endpoint to materialise and approve the drug and get it on the market as quickly as possible so that people can benefit from it. But actually, the time saved by stopping the trial early and evaluating a surrogate may not be that long. It may just be a matter of months, in some cases up to a year. So that’s something to keep in mind.
I would agree that the use of surrogates actually makes it harder for companies to do those post-marketing trials. That’s because once you approve the product to go into the marketplace, people will assume that it already works. So it will be more difficult to identify patients or participants to be recruited on those clinical trials to evaluate its effectiveness and measure those outcomes that ultimately matter to people.
KP: How do you think the industry is doing at communicating data around surrogates?
HN: We have a huge problem with how we communicate the data that we see on surrogate endpoints to patients. A surrogate endpoint like progression-free survival – ‘survival’ in the name may imply that they will live longer, even though there’s no evidence in many therapeutic areas or many types of cancers that extending progression-free survival will extend overall survival.
The way we explain and communicate what survival endpoints mean can have huge implications in people’s understanding and how enthusiastic they would be to start therapy. I don’t think we’re doing a very good job at this and regulators are certainly not doing a good job of communicating what a surrogate endpoint means when they approve new drugs.
There has been some research done in the US, which showed that a label that clearly conveyed the benefits as well as the harms and the uncertainties of drugs when they are approved by the FDA and reporting this in an accessible way to patients can really improve their understanding and influence their decisions. So I think regulators have a responsibility to improve how they communicate, report and summarise the benefits and harms.
KP: How do you think surrogates should fit into clinical reporting?
HN: I think surrogate endpoints have a role to play in early-phase trials as a signal about the potential effectiveness of investigational products, but they should rarely be used as a basis for approving new drugs. At the moment, we’re seeing that regulators are not particularly selective in their use of surrogate endpoints. I think there’s a lot of room for improvements there in terms of really narrowing down the scope of using them and when they are used to evaluate them and to talk about the uncertainties that remain when we make decisions based on them.
KP: With the increasing use of technology and biomarkers to accelerate drug development, are we moving away from that very important aspect – patient benefit?
HN: I think we have very much embraced this idea that faster is better in terms of accessing new investigational products and surrogates are just a symptom of that underlying trend. Ultimately, we do need to question what is the right balance we need to strike between having sufficient evidence and having timely access to medications and at the moment, I think the balance is a little off and we are really prioritising speed of access rather than evidence.
That evidence is rarely evaluated and generated after the products enter the markets. You can approve products very early on in their lifecycle if we can do those post-marketing trials in a robust way and as comprehensively as possible, but we now have decades of evidence suggesting that this is not the case. Actually, once products are approved, there’s very little incentive for companies to do those post-marketing trials, even when they are required or mandatory by regulators. So I think we need to rethink our approach to getting products onto the marketplace as fast as possible and really think about how best to generate the meaningful evidence needed to make decisions.
KP: How can we better involve patients and patient organizations to ensure the conditions of surrogate endpoints are benefiting them?
HN: To me, that is one of the most important questions. The entire enterprise is working under the assumption that this is what patients want, but there is no single patient interest. Patient interests vary. There may be some who want very fast access on the basis of surrogate endpoints, but other patients with the same condition may be a lot more cautious and may actually be harmed by our lowering of evidence standards. Future patients are certainly harmed when we don’t have the evidence to support new drug approvals or have robust evidence to support new drug approvals in the future.
We need to think about how to get patient input in a representative way and not talk to just a few patients that support our way of approaching that. More involvement from patients and surveys of patients in earlier stages of trials would be very helpful I think.
KP: Do surrogate endpoints work against trial diversity?
HN: One thing that comes to mind is when we are able to use surrogate endpoints in clinical trials, this means that we can do smaller trials because the effect is easier to discern. But the smaller the population, the less diverse the population by definition. So by doing larger trials, when we’re aiming to collect data on clinical endpoints, we may actually be able to achieve diversity as well.