Why do pharmaceutical companies keep screwing up?

Perhaps it’s because they are trying too hard. Or lying to hard. But probably it’s because it is really difficult to bring new compounds to market, costs a lot of shareholder money and is very high risk for failure. It must be similar to designing a new commercial aircraft, or even bringing out a new car model that might or might not be market successful. The difference between a pharmaceutical and a vehicle, however, is not only one of size, it is of efficacy and safety. The auto industry knows before hand that they can design a safe product that actually works. Their only concern is whether there are enough buyers. For the pharmaceutical industry, there are no guarantees at the design stage that the product will work, nor that it will be safe.

The statins are a wonderful example of this difference and of the financial pressures on big pharmaceutical companies. Thus Merck embarked in a joint venture with Schering-Plough on a clinical trial to demonstrate that when their two products are combined into a single tablet called Vytorin, that treatment would be significantly superior to simvastatin alone, Merck’s main cholesterol lowering agent which was about to come off patent and thus be reproduced by generic drug manufacturers at a tiny fraction of the price Merck was able to get when it had the patent-monopoly on the product.

The idea to combine the two compounds made biological sense in that they acted on different sites in the cholesterol metabolic pathways. The Vytorin Enhance trial began in June 2002, aimed to recruit 720 subjects with primary hypercholesterolemia, and after 24 months of treatment to show that the cholesterol plaques in their carotids had statistically significantly shrunk. The results were eagerly anticipated by physicians, patients and shareholders. Although the study end-date was April 2006, by November 2007, the results had not been made public. Merck/Schering-Plough issued a statement to explain the delay:

“The independent panel recommended focusing the primary endpoint to the common carotid artery to expedite the reporting of the study findings.  Merck/Schering-Plough now anticipates that these results of the ENHANCE study will be presented at the American College of Cardiology meeting in March 2008.”

This is a curious statement because the study had only one primary endpoint: carotid artery intima-media thickness per subject over 24 months, comparing baseline reading with the endpoint reading (trial registration, www.clinicaltrials.gov)
Why did the head investigator, Dr. Enrico P. Vetri, a Schering-Plough employee, need to “re-focus”? There was only one focal point. It also does not escape notice that as long as the study was not published, revenue from sales of Vytorin would remain unaffected.

The answer, as it appears in a statement by the two companies on January 14, 2008, (now coming on 2 years after the projected end date of the trial) is that Vytorin doesn’t work. It won’t fly. Not only that, but the data strongly suggests that the combination product is harmful. Subjects in the trial who took simvastatin alone had cleaner carotid arteries than those who took Vytorin.

Now, in fairness to the companies, the result must have been a surprise and they must have wondered if somewhere in the analysis of the carotid artery results a data-entry problem might have occurred, perhaps that the patient data files got mixed up. (Nonetheless, in their November 17 statement, they should have been clear that this was not a question of re-focusing on end points, but something far more serious. Patients taking both drugs did worse. And this is not just that they felt worse, their carotid arteries carrying blood to their brains got more clogged up!

Pharmaceutical company corporate behaviour, we are assured, is changing. Merck claims to "put patient safety first". Yet, this kind of press statement along with the fact that it appeared 16 months after the trial end date belies that claim. The Orwellian language of the press releases must have been a deliberate choice of words by the marketing department, pressured as it always is by Finance. Obfuscation, truthiness - if not lies.

The companies have taken pains to point out that the results show only a tendency towards clogging of the carotid arteries by their product, there is no proof - i.e. statistical significant proof. Sure, the result might have arisen by chance. But the difference in intimal carotid artery plaque was 0.111 mm in the Vytorin patients and only 0.0053 in the simvasatin group. Thus plaques grew almost twice as big in the Vytorin group (91% bigger).

The Vytorin trial also demonstrates that although huge amounts of money are spent annually by patients and governments on statins, there is precious little evidence that they do anything more than lower blood cholesterol. Most folks would not give a damn about some molecule or other in their blood were it not for the fact that they believe that the blood level of their molecule is highly correlated with their atherosclerosis and of their chances of having a heart attack or a stroke. What’s measured in trials however, is 2 steps away from what is of benefit to patients - cholesterol, plaques, strokes. In fact there isn’t much evidence that lowering cholesterol reduces rates of heart attacks and strokes and this study suggests that the the combination product Vytorin may be harmful.

The same is true for most pharmaceuticals. We evaluate whether they work or not, not on the basis of a test drive or flight - will this thing actually fly - but on a set of theoretical arguments that it ought to fly. This is equivalent to Boeing saying “Hey, the engine started it ought to go up!”

In a way it is surprising that Merck and Schering-Plough agreed to go ahead with this trial. They could have proposed a trial that had the soft intermediate end points like levels of cholesterol. Indeed, of the 18 ongoing trials of Vytorin, all but one have intermediate metabolic endpoints - usually levels of LDL-c. Only one large trial - to be completed in 2011, will look at what really matters to patients and doctors - rates of heart attacks and strokes. I wonder now if it is ethical to continue further studies of Vytorin? At a minimum one would have to warn study subjects that it looks like Vytorin doesn’t prevent the clogging of their carotid (and presumably coronary) arteries and there’s a pretty good chance that it may make them worse.

Vytorin is not the only example of a drug trial going wrong. I believe that Merck got into trouble with Vioxx because they made a judgement error in terminating the collection of data on drug adverse events before the termination of the collection period for data on good outcomes. They looked for good outcomes for a longer time than they looked for bad outcomes. The extra deaths that occurred in the termination intervals did not materially effect the estimates of risk for Vioxx, but it sure looked bad for the company - that they had deliberately made this decision once they had looked at the data, or at least had seen it coming. It turns out, really, that Vioxx is no worse than the other drugs in this class - they all carry a risk of cardiovascular disease (and their benefit is no greater than less toxic drugs).

Why not just come clean? Pharmaceutical companies know that not all the drugs they design will fly. Hanging on to losers and trying to bend clinical trial results to hoodwink the FDA and physicians is not good enough. They need to set their own internal standards higher than is required and much higher than they are now. This will undoubtedly increase the cost of bringing new compounds to market, but it will preserve their reputations and long run viability.

To come clean will mean that the research divisions must be separated from the finance and marketing divisions not by a Chinese wall, but by a cement one. For Merck and Schering-Plough to choose an employee as the Principal Investigator is a mistake of Titanic proportions. Not that Dr. Vetri is not capable: It is that he could never be seen as independent. But picking a principal investigator is almost beside the point. The entire research enterprise must be in an intellectual vault. Researchers should be rewarded with attractive positions, but those positions and salaries must receive no monetary incentives related to the launch of a successful product. And they must have complete independence in choosing designing the research protocol.

The Finance departments must re-evaluate upward their risk assessments for new compounds and new studies that are proposed. With a truly independent research division it is likely that more studies will fail to show enough efficacy and safety to permit a test flight of the product on the open market. This is a cost of doing business in this high-risk environment, but it is a risk that must be factored into company bottom lines. Such a policy ought to re-assure shareholders and decrease the downside risk of endless collective action lawsuits.