April 22nd, 2013

The FDA, Surrogate Endpoints, and Blood Pressure Drugs

In recent years the FDA has come under increasing fire for approving drugs on the basis of surrogate endpoints without any evidence of greater clinical benefit. The most famous example of this is the diabetes drug rosiglitazone. Despite strong evidence demonstrating that it was effective at lower blood glucose levels — the surrogate endpoint — serious questions emerged about the cardiovascular safety of the drug, eventually leading to its near withdrawal from the market in the U.S. (and full withdrawal in Europe). Questions have also been raised about the long-term health effects of drugs targeting a specific endpoint — including cholesterol and other lipids, blood glucose, weight loss, and blood pressure.

Despite what some believe is a long-term trend against the use of surrogate endpoints, the FDA is actually seeking to make it easier for manufacturers of one therapeutic category, antihypertensives, to claim cardiovascular benefit for their drugs despite the absence of evidence to support most of these drugs. A recent FDA guidance document states:

With few exceptions, current labeling for antihypertensive drugs includes only the information that these drugs are indicated to reduce blood pressure; the labeling does not include information on the clinical benefits related to cardiovascular outcomes expected from such blood pressure reduction. However, blood pressure control is well established as beneficial in preventing serious cardiovascular events, and inadequate treatment of hypertension is acknowledged as a significant public health problem. FDA believes that the appropriate use of these drugs can be encouraged by making the connection between lower blood pressure and improved cardiovascular outcomes more explicit in labeling. The intent of the guidance is to provide common labeling for antihypertensive drugs except where differences are clearly supported by clinical data. The guidance encourages applicants to submit labeling supplements containing the new language.

Although the actual changes proposed by the FDA are small, I thought the issue bore further exploration. I first spoke with Norman Stockbridge, the director of the Division of Cardiovascular and Renal Products (DCaRP) in the Office of Drug Evaluation 1 at the FDA. I then asked Harlan Krumholz and Sanjay Kaul to comment on this topic. I first asked Stockbridge for his thoughts about the FDA document:

I don’t think there’s much crisis of faith with regard to hypertension. For many years we were approving antihypertensive drugs without saying anything at all about why we thought it was a good idea to treat it. Now obviously people thought they knew. Because there was a general understanding in the community between blood pressure and CV risk there wasn’t a strong incentive to rectify this many years earlier.”

Stockbridge explained why he thinks it is impractical to require more outcome trials:

The danger I think is that if you become so worried about re-proving the blood pressure reduction hypothesis that you require outcome trials for every new drug or every new class of drugs, you will never see another antihypertensive drug again. Because there are already a fair number of them and if somebody has to do an outcome claim to get another one on the market they’re just not going to do it. We think there’s a fairly robust demonstration that it doesn’t matter how you lower the blood pressure. That doesn’t mean there aren’t other bad things that drugs can do that will more than compensate for the good that these drugs do. We try to make sure that when we approve them we have enough safety data to have confidence about that aspect of it. But the idea that you’d have to do an outcome trial every time you got a new antihypertensive drug would just kill the market, would just kill drug development.”

I asked Stockbridge to comment on the recent trend by the FDA to require outcomes data in diabetes drugs:

The agency has taken that position, that’s true, and we’ll have to see. And it’s fair to say that there are other places in the agency where similar concerns have been raised. There is more talk now about other clinical areas needing some kind of CV safety study — not generally an expectation that you’re going to show benefit in all those cases — because people are generally risk averse here. There is a risk, however, to this kind of risk aversion — that is, that you are going to stifle drug development somewhat. How aggressive you are about this will determine how much stifling you do, but there’s no question that if you’ve got to do outcomes study that bounder the risk of CV events in a setting where the CV event rate is low you will have a great deal of trouble doing that, because it will take an enormous trial to get those kind of confidence [boundaries] — so it’s a difficult problem and I think there’s plenty of room for public discussion on the topic of how risk averse you want to be and what are the costs of being as risk averse as you are.”

Here is Krumholzs response:

There is confusion about this issue. It is not about the blood pressure reduction hypothesis — but about acknowledging that drugs that lower blood pressure have many effects on the body and the only way to know the net effect on patients is to study the drugs in outcomes trials. Even within a class there may be different net effects on patients. How many times do we need to observe that favorably modifying a risk factor does not always equal favorably modifying risk — even if the risk factor is a strong predictor of outcomes and is part of the mechanism of adverse outcomes — it is not about the risk factor, but about the breadth of effects of the drugs and the need to know the net effect. That being said, there may be circumstances where the uncertainty about the net effect is unavoidable — but it should be a purposeful decision.”

Finally, here is Kauls response:

The guidance calls for a limited general statement regarding stroke, myocardial infarction, and cardiovascular mortality outcome benefit associated with blood-pressure lowering regardless of how it is lowered, followed by a more detailed drug and/or class-specific outcome data or lack thereof. Of the 15 classes of drugs approved for chronic treatment of hypertension, 13 are supported by outcome data. Plus, the link between blood-pressure lowering and CV outcome benefit is more robust and consistent than the relationship between other surrogate endpoints (such as blood-glucose lowering) and CV outcome benefit.

The guidance was developed based on the recommendations of two Cardiorenal advisory committee meetings convened in 2005 and 2006 to discuss class labeling for cardiovascular outcome claims for drugs that are indicated to treat hypertension. It was acknowledged during the panel deliberations that the failure to include information about cardiovascular outcomes in the drug label was contributing to inadequate treatment of hypertension, a significant public health problem. It was the consensus of the committee members that inclusion of CV outcomes claim in the label would serve as a teaching moment for both physicians and patients thereby promoting more adequate treatment of hypertension. Furthermore, the potential to incentivize drug development was viewed as an additional benefit. It is important to keep in mind that over the last two decades the FDA has approved only one new class of antihypertensive drug — aliskiren, a direct renin inhibitor approved in 2007. In contrast, seven new classes of drugs to treat diabetes have been approved by the FDA during this time.

Thus, although this guidance sets a regulatory precedent for extending an outcome claim across a set of pharmacologically distinct drug classes, I think it is quite reasonable and justified.

As to Stockbridge’s statement about the potential of diabetes drug guidance to stifle drug development, it should be noted that the initiation of over a dozen randomized trials enrolling more than 100,000 subjects after the issuance of the 2008 FDA diabetes guidance which calls for outcome data ruling out unacceptable CV risk can hardly be viewed as ‘stifling’ diabetes drug development.”


One Response to “The FDA, Surrogate Endpoints, and Blood Pressure Drugs”

  1. Franz Messerli and Sripal Bangalore sent the following comment:

    We take issue with Dr Stockbridge’s commentary “there’s a fairly robust demonstration that it doesn’t matter how you lower the blood pressure”. LIFE, ALLHAT, and ACCOMPLISH showed clear inferiority for CV outcomes reduction of atenolol, doxazosyn and HCTZ to their respective comparators despite a similar fall in BP. In the US alone, millions of patients are still exposed to the adverse effects, inconvenience and cost of atenolol without reaping any benefits. No question atenolol lowers BP, but it has never been shown to reduce heart attacks, stroke or death in uncomplicated hypertension. According to the FDA’s principle “we have enough safety data to have confidence about that aspect” other beta blockers such as carvedilol and nebivolol have been approved without having to show efficacy. It has been falsely assumed that there is evidence for beta blockers for hypertension for nearly 4 decades and only in the past few years has this bubble been busted and beta blockers have been relegated to a 4th line agent by major national and international hypertension guidelines. Although, carvedilol and nebivolol look good on paper based on surrogate endpoint and properties, they have been widely used by physicians based on this false assumption of CV risk reduction. When a drug sells regardless of presence or absence of outcomes data, there is no incentive for pharmaceutical companies to conduct an outcomes trial. Similarly, for numerous other antihypertensive drugs and drug classes (antiadrenergics, i. e. clonidine, arteriolar dilaltors, i.e. hydralazine) outcome data are completely lacking. Clearly, treating high blood pressure is not a question of having “enough safety data” but rather of demonstrating outcome benefits at least of the drug class if not the individual drug. The FDA advisory will therefore simply encourage the pharmaceutical company to aim for the minimum required-i.e., demonstrate reduction in surrogate endpoint of blood pressure without attempting to attempt to show CV risk reduction. It wouldn’t be a surprise if in a few years we will have to face a situation akin to medications used for diabetes mellitus. After all, antihypertensives lower blood pressure, but so do leeches!