Monday, February 05, 2007

The label came off

I feel ambivalent about pharmaceutical companies at the present time. There is no doubt that they have contributed greatly in the fight against disease; without them we would not have a fraction of the drugs that we do. However, we are not allies in the fight against CLL, we are co-belligerents. We are certainly fighting the same enemy, but I would not want to be associated with some of the methods they are using.

We have a different regulatory framework in the UK to that in the US. The FDA examines a drug to see if it is efficacious and safe. It is a balancing act. Efficacious drugs are not always safe. Even an aspirin can kill you. The FDA has to balance the good that they might do with the harm that they might do. They may put restrictions on their use. In the UK a body known as the MHRA does the same thing.

In making this assessment the FDA relies on the pharmaceutical company releasing to it all the information concerning efficacy and toxicity. As we have learned in recent times Big Pharma sometimes withholds information about toxicity, or seeks to belittle it. It is also possible to talk up the efficacy of an agent. A recent analysis of a new drug for CLL demonstrated an effect in a subset of patients. This was not an effect that the trial had been designed to detect, but one that emerged from data dredging. Subset analysis is a valid methodology, but the statistical tests that are applied to it have to be more stringent. We cannot accept that P<0.05 is significant. After all if you dredge through 20 subsets of data you are bound to find something with a 5% chance of being significant. For example, you have 20 schoolboys. Your hypothesis is that Peter is taller than the others. You have a 1 in 20 chance of being right. However if your hypothesis is one boy is taller than the others, you are almost bound to be right. In the recent British CLL trial subset analysis suggests that overall survival for patients with mutated IgVH genes is better if they are treated with chlorambucil than if they are treated with fludarabine. But that comparison is invalid, because it wasn't one of the things that the trial was designed to detect. It would have needed many more patients to be randomized. It was just this sort of analysis that the manufacturer (not of chlorambucil, I hasten to add) attempted to swing on the FDA. To their credit, the FDA spotted what was going on and refused to license the drug.

Another trick is to use the comparator drug at a lower than optimal dose. For years I have been using chlorambucil at a dose of 10mg a day for two weeks in every four. That's 140 mg per month. The average person is 1.72 sq meters in surface area, so on average I was giving 81 mg/sq m/month. Why am I not surprised when I turn to the pivotal trial for fludarabine and find that patients in the chlorambucil arm received only 40 mg/sq m/month. When I examine the Campath trial which has just shown Campath to be better than chlorambucil as first line therapy, I find that the same rather low dose of chlorambucil has been used. It's not a wrong dose of chlorambucil, but it sets the winning post a little easier to reach.

In the UK we have an organization called NICE that I have referred to before. Its job is to inform the government which drugs are cost-effective and should therefore be paid for by the NHS. It is not perfect, but it tries hard. Not every drug is examined by NICE, but when drugs are not examined, the local health purchaser has to make a decision, and it is likely to be a deal more arbitrary than the one NICE makes. In the US once a drug is licensed it can be prescribed by any doctor as long as the patient can find someone to pay for it. Very often the third party funder will look at the decision that NICE has made in the UK.

Of course the licensing does not cover every possible use a particular drug. Doctors will experiment. For example, Daunorubicin, which has been the mainstay of treatment for acute myeloid leukemia since the 1960s, but for a long period it did not have a license for this use in the UK. In India, a cardiologist who knew a bit about physiology reasoned that sildenafil would be a very useful drug for the treatment of congenital pulmonary hypertension. You can imagine the outrage at a doctor giving Viagra off-license to new-born babies. But he was right and the lives of many babies were saved by his courage.

However, pharmaceutical companies are not allowed to promote their drug for unlicensed conditions. You can imagine why the FDA was outraged to discover that 70% of prescriptions for one popular monoclonal antibody were for conditions outside the licensed indications.

Which brings me to a warning letter sent to a certain pharmaceutical company by the FDA. This is an old letter dating from 7 years ago so I shan’t name the company who no doubt have put their house in order by now. The point of the warning letter was to suggest that the company had violated the terms of the license: “statements or implications by you that this product may indeed be safe and efficacious in the treatment of diseases or patient populations beyond that approved in your application may be considered a violation of the promotional provisions of the Act.” In particular the company was accused of promoting the drug for indications beyond its license. As it happens we now know that that particular drug was useful for that particular indication, and it has become the standard treatment. Nonetheless, it was a transgression.

Pharmaceutical companies are naturally keen to get their product to market and they often have very large sums of money tied up in the success of any particular product. In the past when I have asked drug company reps about a particular unlicensed use of their product, they will say straight away, I am not allowed to promote that use. But if I press them for information they will have the documentation favoring that use handy. It’s just a question of pushing the right buttons. If challenged would say that they were just being helpful in my quest for knowledge. All very well if the doctor reads and judges the paper himself, rather than relying on the company’s interpretation of it.

In my previous post I reviewed the clinical trials of treatments for CLL. You may have noticed that there was nothing about Revlimid there. This is because I don’t want to encourage its use outside its license. This is what I have written previously about Revlimid:

Revlimid has not been licensed for use in CLL and neither has thalidomide. However in small phase II trials both have been shown to have some activity in CLL. Do not be persuaded to take either of these drugs unlicensed, except in the context of a clinical trial. Clinical trials have to be assessed by ethical committees. They will regard any trial as unethical if the risk to the patient is greater for taking the drug rather than avoiding it. Therefore there will be strict entry criteria for these trials, and patients entered into the trials will be closely monitored. Ideally these trials should be randomized phase III trials.

This is still true.

At ASH last year there were two abstracts on the use of Revlimid in CLL. In the first, from MD Anderson the abstract was accompanied by a statement that Revlimid had been approved for use in 5q- MDS and myeloma, and that its use in CLL should be considered off-label. Among 22 relapsed and refractory patients 7 had a response and one a CR. The major toxicity was bone marrow suppression and one patient died of a fungal infection. Tumor flare (sudden tender enlargement of lymph nodes, liver or spleen, rash and fever) occurred in 6, fatigue in 13, nausea in 10, itching in 7 and diarrhea in 5. In the other trial which was from Roswell Park, 45 relapsed and refractory patients were treated, the overall response rate was 42% with 9% CR. Similar side effects were seen as in the Houston experience, but in addition tumor lysis syndrome occurred in two.

My conclusion is that this treatment is no picnic, but because it acts in different ways to most anti-CLL treatment, it may have a role in CLL. We need more information on whether it is able to help patients with p53 deletions and whether it might synergise with existing therapies.


Anonymous said...

I suspect the drug the good doctor won't mention is Genasense. I don't have a dog in the fight, but I will note that a lot of 'heavy hitters' in CLL appeared before the FDA to argue for its approval.

Perhaps the 70% usage of a monoclonal antibody is rituximab. If correct, so what? Does it work in CLL even though it is not approved for CLL? Yes!

The drug company probably reprinted some medical papers that showed a benefit from using the drug in CLL. Horrors!!! The fact that the company was correct is immaterial???

Pharmaceuticals may be guilty of hype and outright lies, and if that's the case, sock it to them! However, everyone seems to forget the poor little innocent patient in all of the machinations.

Lives are being saved by rituximab, approved for CLL or not.

That is the bottom line, and should not be forgotten.

(For the life of me, I can't understand pharma bashing. Folks, if there were no drug companies, Dr. Hamblin might offer you sympathy and a kind word. Drug companies even manufacture Dr. Hamblin's favorite drug, chlorambucil.

Drug companies are in business for ONE THING, AND ONE THING ONLY. That is to make money. They are not a charity.

I object to the denial of life-saving drugs because the British taxpayer is too cheap to pay for life-saving drugs. Luckily, I live in the US, where, until Hillary gets her way, we can still have good remissions using (horrors!!!) pharmaceuticals.

And, sad to say, Dr. Hamblin misses the point on statistics. A p value of .05 is commonly accepted as the cut-off point whether a particular result is due to chance or valid. It has nothing to do with a particular sample showing that one out of twenty results will show a favorable result.

It is fair to say, though, that a small sample can give meaningless results if it asked to measure what was not selected for study (i.e. a hypothesis). A study has to be carefully designed, and looking for an odd result here and there is only a suggestion for another study, not a mandate showing something terribly meaningful.

I agree with Dr. Hamblin that a study or two cannot be taken as definitive if it isn't a phase III trial conducted appropriately.

OTOH, if you have nothing else, and will die with no treatment, you may be tempted, might you not?

Terry Hamblin said...

I didn't mention several drugs. I was not trying to put blame on any particular drug company. On the whole I approve of them and I am certainly in favor of private enterprise. I have absolutely nothing against the idea that companies are in business to make money. On the other hand I am in business to save lives, which is why I call them co-belligerents rather than allies.

I also have no problem with people spending their own money on whatever they choose. I might think they're crazy, but it's their money. I do object to people spending my money on stupid things, though, and whether it is spent via a taxation-based system or an insurance based system, the bottom line is that someone else is paying for it. Americans tend to forget that taxation funds a large proportion of the healthcare budget via Medicare, Medicaid, County Hospital ER rooms and the VA. Not to mention tax breaks for companies.

Does rituximab work for CLL? I certainly think so, though the evidence is not yet convincing. It certainly has activity, though not so much as in follicular lymphoma, and it lacks the toxicity of cytotoxic drugs. But rituximab has been around a long time now, so where is the evidence from properly designed controlled trials? There are two studies that compare FC versus FCR in either first-line or salvage situations. We will not get information from them until 2008 or even 2009. Why were these trials not started earlier?

The story from several years ago where a company was reprimanded by the FDA for breaking the terms of their license is all about process, not outcome. Sure the drug was useful for the new indication, but without the rule of law anarchy proceeds. Try reading the speech by Sir Thomas More in Robert Bolt's play, 'A Man for all Seasons' about what happens when you remove the rule of law. Drugs can be used off-license as I have illustrated, but they should only be used in that way by people who know what they are doing, who appreciate the possible risks and benefits.

I don't have favorite drugs, not even chlorambucil. It is clear that chlorambucil is very much an unfavorite drug of the pharmaceutical industry, though. That might have something to do with the fact that it is cheap and off-patent. On the whole I would rather patients needed no drugs at all for CLL, and I am rather in favor of rituximab alone for some cases, but every case and every drug has to be considered on its merits for every case.

My attempt at explaining statistics conflated two one in 20 chances, and was misleading. Thank you for pointing it out. p<0.05 simply means that there is less than a 5% chance that such a finding would occur by accident. We generally accept that as a meaningful statement that something is true. We design trials to include sufficient numbers of patients that a difference between two groups being compared is large enough and consistent enough that it can be detected with that degree of certainty. If, after the event, someone dredges through the data to find other differences that the trial wasn't designed to look for the p<0.05 degree of significance just won't do. This is because if there are lots of subsets there is a pretty good chance that you can find something that gives a 19 in 20 chance of being true.

In desperate circumstances doctors will take greater risks. But they don't try just anything. They look for a course that they might choose to follow and are willing to accept less convincing evidence. But most of these treatment decisions, taken on less convincing evidence are not in life or death situations.

Advertising is often misleading; false impressions are given even when no lie is told. Indeed, one of the skills of a copywriter is to mislead without actually lying. Advertising to professionals is obviously a service since we assume (with some but not absolute justification) that professionals are discerning. Advertising of prescription medicines to the general public is not something that is permitted in Britain, but is a fairly recent innovation in America. I'm not entirely convinced it is a good idea.

Anonymous said...

When it comes to prescription medicine, the patient cannot generally (unless he/she is a person who can write prescriptions themselves) medicate themselves. Some foolish/ineffective/dangerous drug therapies are thus negated.

There is, of course, the problem with unscrupulous/unethical doctors or nurse practitioners who will write prescriptions that they should not. It happens, and I've known several doctors who would write prescriptions they had no reason to write, presumably out of a desire to help a patient, money, drugs, or other reason.

You are right, of course, in your unrelenting plead for phase III trials comparing rituximab and combinations such as FCR against 'standard' therapies perhaps such as chlorambucil or fludarabine. It is, as you have noted, problematic to use historical studies as controls since there is a glaring surfeit of variables that cannot be controlled for.

Gosh I know there is a problem with both off-label use and direct to consumer advertising. Without off-label use, though, many patients walking around today might not. It is SOOO expensive to run a complete clinical trial, just so one can advertise. It's like the proverbial mousetrap. If you create something that works, you don't have to really advertise to sell it.

If we had a cure for CLL (or, better yet, baldness) would the company have to spend a dime on advertising? Perhaps for crowd control...

Direct to consumer ads are perhaps a problem, but what would Bob Dole, former Senator and Presidential candidate do if not to advertise a certain 'little blue pill'?

On could argue that it does put pressure on doc to explain why Viagra won't necessarily work for women or for 20 year old men, but that's one responsibility of the doc (though I know they fear patients will just go somewhere else).

What a crazy world in which we live, isn't it?

Manu Manickvel said...

'Drugs can be used off-license as I have illustrated, but they should only be used in that way by people who know what they are doing' - Doc, imho only such people can talk about it...when will the rest learn?! To keep quiet and benefit?