Tuesday, May 15, 2007

Cost benefit analysis of a vaccine (Gardasil)

Kudos to the New York Times (again). They publish a simple, fantastic piece about the risks and benefits of the new Gardasil vaccine against HPV. The risks?

They may include pain, swelling, itching and redness at the injection site, as well as fever, nausea and dizziness.
Well, this sounds like what you might expect from any vaccine. The pain, swelling at the injection site are basically signs that your immune system is working. And the fever as well. Nausea and dizziness are pretty common side effects of any injection. And:
Another concern involves long-term safety. How do we know this vaccine will not eventually cause other problems like autoimmune or neurological disorders or lose its protective powers or foster the dominance of other HPV variants?

Actually, we don’t. But we do have at least five years of safety data that include no hints of long-term risks or waning effectiveness. But if the vaccine should begin to lose potency over time, that could easily be remedied by a booster shot.
Honest, informative. Some people may not feel comfortable with the idea that we don't know the effects of the vaccine in the long term. But any new vaccine, you won't know what the long term effects are for 20 years (until you've been using it for a long term). And so far, there are very few vaccines that have had any serious long term effects.

Hmmm... So should you have give your daughter the vaccine? What can she get from it? Well, it prevents infection with the subtypes of HPV that cause up to 70% of cervical cancer. And, in a 3 year followup, it prevents pre-cancerous lesions in nearly 100% of vaccinees who got the vaccine prior to exposure to HPV. It even provides some protection to women who have previously been exposed to the virus. Up to 17% protection from pre-cancerous lesions for women who had already been exposed.

This is why it's so important to give this vaccine to children, before they are exposed to HPV through sex. Once they're old enough to choose the vaccine for themselves, it's likely already too late to get much benefit from it.

Granted, we can't say whether or not it prevents cancer. It takes twenty to thirty years for cancer to show up in women who have been exposed to HPV, and this vaccine has only been in production for a few years. But it prevents HPV infection of the most prominent subtypes, it prevents precancerous lesions... it's not a stretch to suggest it will almost certainly prevent cancer. And that sounds pretty good to me.

There are no known risks of Gardasil (though to be fair, there is a small chance that longer term studies will find some). There are some very clear benefits to getting the vaccine. So when you're making up your mind for your child - make the calculation - prevent HPV vs. no known risks. It seems pretty clear to me.

Digg!

8 comments:

mhatrw said...

o summarize this published medical journal article:

1. In the FUTURE I trial, GARDASIL demonstrated no clinical efficacy among the general subject population for overall reduction in the rates of grade 2 and grade 3 cervical intraepithelial neoplasia and adenocarcinoma -- the only recognized precursors to cervical cancer.

2. In the larger FUTURE II trial, GARDASIL demonstrated no clinical efficacy among the general subject population for overall reduction in the rates of grade 3 cervical intraepithelial neoplasia and adenocarcinoma -- the strongest (and many would argue only valid) precursors to cervical cancer.

3. Extrapolating from GARDASIL's very limited clinical "success" (in the FUTURE II study only) against grade 2 cervical dysplasias (40% of which regress spontaneously), 129 women would be have to be vaccinated (at a cost of about $60,000) to prevent a single grade 2 cervical dysplasia.

4. GARDASIL's protection against cancer associated HPV strains 16 and 18 appears to cause a disproportionate increase in of pre-cancerous dysplasias associated with other HPV strains associated with cervical cancer "raising the possibility that other oncogenic HPV types eventually filled the biologic niche left behind after the elimination of HPV types 16 and 18."

5. Even if look only at the FUTURE II results (in which for some reason GARDASIL performed better among the general female population), we are talking about just a 17% decrease in all high grade dysplasias -- many of which would spontaneously regress without treatment. So we would have vaccinate 129 women (at about $500 for the three shot regimen) to avoid a single, eminently treatable dysplasia. That's about $60,000 per dysplasia prevented.

This is all directly from the article linked above.

I myself would add that we currently have only 3 years of follow up to go on in terms of both GARDASIL's safety and efficacy among the 16 to 26 year female population, no data concerning its efficacy among 9 to 12 year old girls and only 18 months of follow up on less than 600 total preteen girls in terms of safety data about GARDASIL within its targeted population.

See also: JAMA

It appears that the vaccinated cohort sees a 20%+ increase in high grade cervical dysplasias caused by cancer-associated HPV strains other than HPV 16 and 18.

One possible explanation is that HPV 6 or HPV 11 infections are antagonistic to more dangerous HPV infections.

The Factician said...

mhatrw,

Thanks for your comment. You seem to be making a relatively common mistake in interpreting the data of clinical effectiveness. The data you are citing are for the population as a whole. Many of the people in that part of the study had been previously infected with HPV (and thus are not responding well - the real surprise is that they are responding at all). This suggests there is some protection afforded by the vaccine even if you've already come in contact with the virus.

Among women who have never shown signs of being exposed to HPV prior to vaccination, the success rate is nearly 100%.

Studies are ongoing as to the effectiveness in preventing these types of pre-cancerous lesions in young girls who have never had sex (and therefore should never have been exposed).

mhatrw said...

I am not making an mistakes. I am merely reporting the clinical facts. Even among the selected subpopulation that had not been infected with either HPV 16 or HPV 18, the best overall efficacy of GARDASIL against high grade cervical lesions was only 27%. And unless every one planning on getting a GARDASIL shot is also planning on paying for an HPV screening test beforehand, GARDASIL's efficacy in the general population of 16 to 26 year old females is also extremely important.

GARDASIL is a new vaccine from Merck that has now been clinically shown to strongly protect against two dangerous, cancer associated HPV strains -- HPV 16 and HPV 18 -- for at least three years among women previously unexposed to these strains. HPV 16 and HPV 18 are currently associated with 70% of all cervical cancer cases. Medical experts hope that this means that GARDASIL could potentially end up reducing cervical cancer contraction rates by as much as 70%. So far, the limited clinical evidence available tells a very different story.

Because so few women contract cervical cancer and because cervical cancer's average latency period after initial HPV infection is over 20 years, it would take hundreds of thousands of subjects and/or decades to conclusively demonstrate that GARDASIL actually reduces cervical cancer rates in any population. Therefore, the end points used to demonstrate GARDASIL's effectiveness against cervical cancer in GARDASIL's two biggest studies (FUTURE I & II) were chosen to be the number of high grade cervical lesions -- or alternatively, the number of patients with high grade cervical lesions -- prevented over a 3 year period. High grade cervical lesions consist of grade 2 and grade 3 cervical intraepithelial neoplasias and all adenocarcinomas in situ. Among women in the USA, the vast majority of high grade cervical lesions either regress spontaneously or are successfully treated. However, the FDA currently considers high grade lesions to be acceptable surrogate outcomes for cervical cancer.

Here are the most recent and by far most comprehensive clinical results for GARDASIL in terms of reduction of high grade cervical lesions over a three year period.


For the general 16 to 26 year old population

FUTURE I

For the unvaccinated group, there were 194 total high grade lesions diagnosed -- 101 associated with HPV 16 & 18 and 93 associated with other dangerous, cancer causing HPV strains. For the vaccinated group, there were 182 total high grade lesions diagnosed -- 76 associated with HPV 16 & 18 and 106 associated with other dangerous HPV strains. That's a 6% overall decrease high grade HPV lesions but a 14% increase in other dangerous unprotected HPV strains for the vaccinated group.

FUTURE II

For the unvaccinated group, there were 361 total high grade lesions diagnosed -- 207 associated with HPV 16 & 18 and 154 associated with other dangerous, cancer causing HPV strains. For the vaccinated group, there were 281 total high grade lesions diagnosed -- 103 associated with HPV 16 & 18 and 178 associated with other dangerous HPV strains. That's a 22% overall decrease high grade lesions but a 16% increase in high grade lesions associated with dangerous unprotected HPV strains for the vaccinated group.

In terms of individual subjects rather than lesions (some subjects had multiple high grade lesions), vaccinated patients were 17% less likely to develop high grade cervical lesions overall but 15% more likely to develop high grade cervical lesions associated with dangerous unprotected HPV strains.

Combined FUTURE I & II

For the unvaccinated groups, there were 555 total high grade lesions diagnosed -- 308 associated with HPV 16 & 18 and 247 associated with other dangerous, cancer causing HPV strains. For the vaccinated groups, there were 463 total high grade lesions diagnosed -- 179 associated with HPV 16 & 18 and 284 associated with other dangerous HPV strains. That's a 16.6% decrease in overall high grade lesions but a 15.0% increase in high grade lesions associated with dangerous unprotected HPV strains for the vaccinated group.


For the sub-population of subjects who tested negative for both HPV 16 and 18

FUTURE I: Merck did not publish these results.

FUTURE II: Vaccinated patients were 27% less likely to develop high grade cervical lesions overall and at most 8% more likely to develop high grade cervical lesions associated with dangerous HPV strains other than HPV 16 & 18.


For the sub-population of subjects who tested positive for either HPV 16 and 18 or both

FUTURE I: Merck did not publish these results.

FUTURE II: Vaccinated patients were 9% less likely to develop high grade cervical lesions overall and at least 105% more likely to develop high grade cervical lesions associated with dangerous HPV strains other than HPV 16 & 18.


This is not an opinion. These are Merck's very own published clinical results for FUTURE I and FUTURE II.

The Factician said...

mhatrw,

You are indeed using data from their study. And, like much of the media, you are gravely misinterpreting it. (A little data is a dangerous thing).

Look at the first figure in the paper. It describes the 3 populations that they examine in the paper. There are only 2 that are important for my discussion. "Intention-to-treat" is everyone. Every woman who signed up falls into that category. "Per-protocol" is every woman who signed up for the study who showed no signs of exposure to HPV prior to or during the vaccination.

All of the data that you are discussing comes from the "Intention-to-treat" section of the paper. These women include women who had exposure to HPV during or prior to vaccination. You're right that it seems to work at a rather low rate in this population. The surprise is that it works at all. One wouldn't expect that this vaccine would protect women who had already been exposed to the virus (this suggests we don't understand completely how HPV induces cancers). This is good news! The vaccine works (albeit at a low rate) in women who already have the virus.

Within the "per-protocol" population (that is, the population who showed no signs of prior exposure to HPV) the rate of success is near 100% in every category.

That said, you're absolutely right on one point, but only one point. These are only surrogate outcomes. But they are reasonable surrogate outcomes. We won't know for 20 years whether or not the Gardasil vaccine truly prevents cancer. But every sign so far suggests very strongly that it will.

mhatrw said...

What signs point to GARDASIL's efficacy against high grade lesions caused by ALL dangerous HPV strains? GARDASIL's highest demonstrated three year clinical efficacy in any subpopulation of either major FUTURE study was 27%.

Yes, GARDASIL was very effective against HPV 16 and 18. But what does this matter if other cancer causing HPV strains fill the gap?

In the FUTURE II study, GARDASIL was 27% effective in reducing the number of subjects with high grade cervical lesions caused by all dangerous, cancer associated HPV strains for the sub-population who tested negative for both HPV 16 and 18 when the study began, 17% effective in reducing the number of subjects with high grade cervical lesions caused by all dangerous, cancer associated HPV strains for the entire population and 9% effective in reducing the number of subjects with high grade cervical lesions caused by all dangerous, cancer associated HPV strains for the sub-population who tested positive for either HPV 16 or 18 when the study began.

Those are Merck's own numbers for the major study and GARDASIL preformed far better in FUTURE I than in FUTURE II.

You and Merck would like us believe -- based on nothing but blind faith -- that GARDASIL's actual efficacy among a virginal subpopulation would be much higher. But where are the clinical data to support this rank assumption? Merck could have easily released the overall efficacy percentages for just the subjects who had never been infected with any HPV strains associated with high grade cervical lesions or even just the subjects who had never been exposed to HPV period. However, Merck chose not to publish these results. In fact, in its FUTURE I study, Merck released no subpopulation analyses whatsover in terms of GARDASIL's efficacy against all HPV strains. And in Merck's FUTURE II study, Merck published just a single subpopulation efficacy result against all dangerous HPV strains -- the 27% efficacy among subjects who tested negative for both HPV 16 and 18 when they started the experiment.

Is it up to me to disprove efficacy or up to Merck to demonstrate efficacy? Personally, I would love to believe that GARDASIL will be more than 27% against all HPV strains when injected into a (largely) virginal sub-population. But where are the clinical data that support this assumption?

The Factician said...

mhatrw,

I'm sorry. I hope you don't think I'm being rude, but almost everything you say is dead wrong. It's not a different interpretation of the data. It's completely, and totally wrong.

I've addressed most of the points you raise in earlier responses. I suggest you re-read them. Otherwise, I'm not sure what I can do to help you.

Personally, I would love to believe that GARDASIL will be more than 27% against all HPV strains when injected into a (largely) virginal sub-population.

Hmmm.... Well, I'm not sure what else to say to you. I've tried to spell it out clearly for you, but you are still misinterpreting the paper. (gravely). I would suggest that if you really would like to think that the success rates are higher than your belief that you re-read my earlier responses.

Your belief has very little to do with the matter. Belief (that is to say, thoughts without proof) is irrelevant. The fact of the matter is that they've shown astounding success rates (see above, or re-read the paper) using surrogate outcomes.

mhatrw said...

OK, let me quote directly from Merck's own FUTURE II study (linked via my name):

Among women who tested negative for both HPV-16 and HPV-18 at enrollment (4693 in the vaccine group and 4703 in the placebo group), high-grade cervical disease developed in 95 subjects in the vaccine group and in 130 in the placebo group, a reduction of nearly 27% in the vaccine group.

So what exactly am I misinterpreting when I say that GARDASIL demonstrated 27% efficacy against high grade cervical lesions caused by all HPV strains associated with cervical cancer?

If I am misinterpreting something, please find me a published efficacy percentage against high grade cervical lesions caused by all HPV strains associated with cervical cancer that is higher than 27% for any sub-population.

This should be good.

The Factician said...

mhatrw,

Please re-read the section I wrote above describing the populations involved in the study. Many of the intention-to-treat population tested negative at start of the study, and yet showed signs of infection during the study. Keep in mind, that the ages of women included in the study were likely sexually active, and amongst the highest risk population for being infected. These women likely were infected prior to the final dose.

The only important data for consideration of effectiveness are in the per-protocol population. These are women who received all 3 doses of vaccine and showed no signs of infection during that time. These women show effectiveness in the 98% and higher range.

Given that we have been repeating the same things over and over, I won't be responding any more in this thread. Best of luck.