AIDS Treatment News logo      

April 25, 2011

HIV PrEP Explained: Critical Prevention Opportunity

by John S. James, AIDS Treatment News, April 25, 2011


The Good News

In late 2010 the New England Journal of Medicine published an authoritative report of a trial of PrEP (pre-exposure prophylaxis) in over 2,000 HIV-negative but high-risk gay men, showing that one pill a day of the approved HIV medicine Truvada could prevent HIV infection.[1] Science magazine listed this study as one of the top 10 achievements of 2010 [2]; and President Obama issued a statement about this trial.[3] For background, see [4].

The headline result is that Truvada was 44% effective (compared to placebo) in preventing HIV transmission in this population. This sounds disappointingly low. Why then is the study considered so important?

Much of the answer is that 44% is not the whole story. In the group randomly assigned to take a placebo, 64 became infected during the study; in the group randomly assigned to take Truvada, 36 became infected. Because the sizes of the groups were almost exactly equal (1251 participants, vs. 1248), you can compute the 44% by simple arithmetic.

Fortunately the study did blood testing of the 36 people who became infected despite being assigned to take Truvada, to see if they were really taking the drug. About 90% of them (33 of the 36) had no drug in their body when tested, at the last blood draw before they were found to have HIV. (Truvada consists of two drugs, and no detectable level of either drug was found in any of the 33 people.) The drug-level testing methods had been carefully validated, and can detect the active form of at least one of the drugs two weeks after people stop using the pills. Clearly these study participants had not been taking the medicine when tested. When these trial volunteers who were assigned to take the drug (but had not in fact taken it) were not counted, the effectiveness was not 44%, but over 90%.

It gets better. The other 3 participants who became HIV infected during the trial had very low levels of the active forms of the drugs, just above the limit of detection. Clearly they had taken some Truvada, but were far from using it as directed.

So of the 100 participants who became HIV-infected during this trial, 97 had not used Truvada at the time of the last blood draw -- and the other 3 had used it very poorly. Of those who actually took one Truvada pill per day throughout the study, not one became HIV-infected during this trial (vs. 64 of those who had been given the placebo). So for those who actually used the drug as directed, throughout the entire study, it was 100% effective in this trial, which had over 1,000 high-risk gay men assigned to take Truvada.

But the blood draws occurred in this study at weeks 4, 8, 12, 16, 24, and then every 12 weeks. So we cannot rule out the possibility that one or more people who had not taken Truvada for about two weeks or more before one of those visits, then got religion after the visit, and starting taking the pills daily -- yet got infected despite having adequate drug levels. This seems unlikely in view of the overall findings. But if it did happen, then it would mean that the protection was less than 100% in those who were using the drug. It was not possible to have everyone wear a device that that could record drug levels continuously -- and also tell exactly when they got HIV. So there is no way to be sure that nobody had enough drug in their body when they were infected. For this reason, 100% effectiveness is not claimed or reported.[5]

Still, the fact remains that Truvada for HIV prevention worked very well in this trial. Truvada when used properly (once a day, throughout the trial) completely prevented HIV infection, in a study of over 1,000 people -- when a comparable, randomly chosen group of 1,000 other people given a placebo had 64 HIV infections during the same time. Almost no drug for anything works this well.


Using PrEP Now

Since doctors can legally prescribe Truvada for prevention now, the U.S. CDC (Centers for Disease Control and Prevention) in January 2011 published interim guidance for physicians who want to use it to prevent HIV infection in high-risk HIV-negative men who have sex with men.[6,7]. Guidance will change as more information becomes available from other studies currently ongoing.

The main obstacle now is the price. Gilead Sciences, which holds the patent on Truvada, charges about 100 times as much for Truvada in the U.S., as other companies charge profitably for the same generic medicine in countries where Gilead's patent does not apply. And unless the FDA approves a formal "indication" for prevention use, insurance is unlikely to pay. This means that you can get Truvada for prevention in the U.S. today -- if you have over $12,900 per year to spend out of pocket -- or possibly, very good health insurance. (Gilead raised the price early in April, 2011; on April 18 we checked retail prices on Drugstore.com, which offers 90 once-daily pills for $3,180,90.)

Clearly PrEP will not be used enough to impact the epidemic, if individuals must pay so much out of pocket. (Just taking a $35 pill before sex is not expected to work. The body must convert both drugs into their active forms, which takes time.)


Adherence Issues

Some of the media discussion on Truvada adherence (taking one pill a day consistently) vs. effectiveness for HIV prevention is confusing, because in this study adherence was measured in different ways. Drug-level testing was clearly the gold standard -- while other methods, like asking people how often they took the pills, or counting pills returned, were often confusing, and much less reliable.

If any drug is so hard to take correctly that people seldom do so, then that drug is much less valuable. This is part of the reason that studies are most conservatively reported as “intent to treat” (counting everybody assigned to take the drug or other treatment as being treated -- even if they never actually take any of that treatment at all), instead of “on treatment” (counting only those who actually did take the drug -- partially a self-selected group, which could make the trial results hard to interpret). The purpose of intent-to-treat reporting is to make the study results both more accurate and also more relevant to social policy -- since doctors can only recommend a treatment, not make sure it gets used. In this case, the difference in result was huge: 44% effective, vs. over 90% -- or even close to 100% effective in this trial (depending on how you count the three participants with extremely low drug levels who were using the medication very poorly).

Why did so many of the study participants not take the Truvada?

No one knows for sure at this time. But something unusual and unexpected happened in this study. This clinical trial took place at 11 sites in 6 countries -- with two of the 11 sites in the United States (Boston and San Francisco -- with 113 U.S. participants on Truvada, 114 on placebo). And adherence at the U.S. sites was much better than at the non-U.S. sites -- 97% among all the U.S. participants, considered extremely good, compared to far less overall. The "44% effective" headline just averages these very different situations.

We know of three theories about this difference: (1) The U.S. participants were about 10 years older on the average (though this would hardly explain the huge adherence differences observed); (2) Many of the non-U.S. participants were living at home and probably not “out” to their families, so they would have needed to conceal their participation in the study, making adherence difficult; or (3) Boston and San Francisco have excellent HIV treatment access, perhaps the best in the U.S., while in many countries most people with HIV are dying with no treatment at all. So trial participants may have given or sold their pills to someone with immediate need.

We think that the latter is most likely.

Whatever the reason, the excellent U.S. adherence shows that people can use the drug properly for prevention. And in the future, patients will know that they are getting the active drug (not a placebo), and that it has been proven to work -- both of which should boost adherence. Researchers need to find out what went wrong at some of the sites, and learn how fix it.

But the bottom line is that we do have proof of principle that Truvada prophylaxis can work in high-risk gay men, and prevent close to 100% of the HIV infections that would otherwise occur.


Who Should Take Truvada?

No one we know thinks that all gay men should take Truvada.

But using this new prevention option in a targeted way should help a lot to control the HIV/AIDS epidemic. Remember that in any epidemic, if each person infected infects fewer than one other person on the average, and this can be maintained, then the epidemic will end.

And it turns out that a disproportionate amount of transmission takes place very early in someone's HIV infection -- when viral load is extremely high, but people do not know and seldom even suspect that they are infected, so they are much less careful about risk to others, than if they did know their status.

Also, in HIV as in most epidemics, a few so-called “superspreaders” account for a very disproportionate amount of transmission.[8] In HIV, they are much like those in the Truvada prevention study -- high risk, meaning that they have many sexual partners, especially with unprotected anal intercourse. (In the iPrEx study, volunteers were interviewed by the researchers, and those considered not to be at high risk were rejected -- so that the trial would be more likely to answer the question of whether or not this PrEP treatment worked.)

So far, the big problem in targeting early infection and/or superspreaders has been getting them diagnosed in time. Many are not aware of being sick at all during their early “primary” HIV infection, when they are so dangerous to their sexual partners. Most do get sick, but with the flu-like symptoms of an ordinary viral infection. They or their doctors have no reason to suspect that they have HIV. And if they do take a standard HIV antibody test, they will test negative even though they have a very high HIV viral load, because the body has not formed antibodies against HIV yet.

Recently there has been much interest in trying to stop the HIV epidemic by a policy of “test and treat” -- find people and get them into treatment right away, because on treatment their HIV viral load will be hundreds of times less than before, and they will be much less likely to transmit the infection to anyone else. One recent study [9] concluded that test-and-treat alone will not be enough in the United States, where only 19% of people with HIV have an undetectable viral load -- a "shocking" figure due largely to poor access to care. And it is very hard to find newly infected people soon enough.


PrEP's Big Advantage with Test-and-Treat

With effective PrEP, the problem of early HIV diagnosis goes away. If the high-risk individual never gets infected in the first place, a whole branch of the epidemic tree can be cut off. And as we explained above, the effectiveness of Truvada PrEP in gay men has been close to 100% so far, when it has been used properly (one pill a day, taken consistently).

The way to identify high-risk individuals is to let them identify themselves. After a conversation with their doctor (or a specialist physician at an AIDS organization), and with testing to make sure they do not already have HIV (in which case they would need medical management, not PrEP), they should be able to get a one-pill-a-day supply of the drug without obstacles. Periodic testing could be required, to guard against side effects (such as possible kidney damage), and to make sure the patient is still using the drug, and does not have HIV.

Of course there will be some worried well who end up taking Truvada for prevention even though they do not need to. The doctor can explain the small but real risks of doing so. Truvada is one of the least problematic AIDS medicines.

Note that the “superspreaders” are usually the same high-risk individuals who have many sexual partners and do not take needed precautions. They know who they are. And their own interest in protecting themselves closely parallels the public interest in stopping the HIV epidemic.

Notice that PrEP gives people a new, big incentive to get tested -- the chance to avoid HIV entirely if they get the good news, perhaps even without changing their sexual practices -- not just the chance to start treatment earlier in case of bad news.

We need combination HIV prevention; no one size fits all. But we now know that Truvada PrEP can provide very good protection for high-risk men who have sex with men. And by doing so, it will also be a major tool to help control and eventually end the global HIV epidemics.


Note on FEM-PrEP Trial Closing

Shortly before we went to press, a similar Truvada PrEP trial in women failed, and was closed early. It was ended because exactly the same number of women were infected in the group assigned to take Truvada, as in the group assigned to take the placebo. The study's Independent Data Monitoring Committee secretly examined these interim results, and determined that this trial was very unlikely to show that Truvada PrEP worked in the women. Therefore, the trial was stopped, due to "futility."

Why did the same iPrEx study drug work in men but not in women? No one knows. Blood samples were saved, and drug-level testing and other data will need to be analyzed, to see what can be learned.

There are at least two plausible theories as to why this may have happened:

(1) The iPrEx trial focused on anal HIV transmission, FEM-PrEP on vaginal. It is possible that oral Truvada does not block vaginal transmission, probably because it does not reach high enough levels in the vagina.

(2) There is also another possibility. As explained above, the iPrEx trial worked much better at the U.S. sites (Boston and San Francisco) -- than at other sites, where participants would often have strong reason to give their medication to someone who was sick due to HIV, and not tell the researchers. All of the FEM-PrEP sites were in Africa, where access to life-saving HIV treatment is much worse than in Boston or San Francisco. Unlike iPrEx, FEM-PrEP had no U.S., European, or other sites where no one would need to divert the pills to save the life of a family member or friend.

Neither one of these two possibilities would contradict the iPrEx results -- or the CAPRISA 004 results. In either case, both these studies remain as solid as they ever were.

Note: Truvada alone is not considered proper HIV treatment today, since at least one other anti-HIV drug is needed to more fully suppress the virus and prevent it from developing drug resistance. But if no other treatment is available, Truvada alone might restore someone with AIDS to health and save their life for some time. It is probably better than any HIV treatment anywhere before 1996. Of course it would be much better to take Atripla (or a generic equivalent, or a comparable combination); Atripla is Truvada plus a third drug that works well with it, a modern HIV treatment in one pill per day.

We will report new information as it becomes available.


References and Footnotes

[1] Preexposure Chemoprophylaxis for HIV Prevention in Men Who Have Sex with Men. New England Journal of Medicine (full text free); December 30, 2010, published online November 3, 2010, http://www.nejm.org/doi/full/10.1056/NEJMoa1011205.

Later analysis of the trial, not ready for the above publication, was included at the 18th Conference on Retroviruses, especially in the March 1, 2011, 10:00 a.m. session, "Advances in PrEP," at http://www.retroconference.org/2011/data/files/webcast_2011.htm (you need to select 'Tuesday'). Besides the Scientific Overview, also note “Adherence Indicators and Interim Guidance: Preexposure Prophylaxis for the Prevention of HIV Infection in Men Who Have Sex with Men," and "PrEP Drug Levels in the iPrEx Study,” later in the same session.

[2] Science Magazine Names USAID-Funded HIV Research As Top 10 Breakthrough of the Year. [Both the CAPRISA trial (1% tenofovir vaginal gel for women) and the iPrEx trial (Truvada, a tenofovir + FTC pill for men) made the top-10 list.] Medical News Today, December 28, 2010, http://www.medicalnewstoday.com/articles/212499.php

[3] White House press release, November 23, 2010, http://www.whitehouse.gov/the-press-office/2010/11/23/president-obama-welcomes-new-hiv-prevention-research-results

[4] Background note: This international clinical trial is called iPrEx, which is an abbreviation for the Spanish name of the study; many participants were in Spanish-speaking countries. It used Truvada, a common HIV treatment marketed by Gilead Sciences; Truvada consists of two drugs, tenofovir and emtricitabine, combined in a single pill taken once a day. A related HIV prevention study in women, using a vaginal gel containing tenofovir, was also successful (CAPRISA 004, also recognized in the Science top-10 list. But a prevention trial using Truvada in women recently failed; see the discussion of FEM-PrEP in the text above.

[5] A major reason the "44%" effective (instead of "close to 100%") is widely headlined, instead of is that researchers are expected to highlight the study analysis that they planned in advance to do -- not other analyses that they may choose after looking at the study results. In this case, the planned analysis was the overall HIV-prevention result, with all the study sites considered together. The huge difference between the U.S. and some of the other sites was not expected.

The U.S. sites showed proof of principle that taking one Truvada pill a day can work very well to prevent HIV infection. The sites where people did not take the pills showed that these sites had other problems, which need to be investigated and addressed.

Those interested in statistical methods will note that headlining the pre-planned analysis turned out to be misleading in this case, because of the emergence of a "black swan" -- something totally unexpected which changes the entire picture. As seen also in the financial world (http://www.guardian.co.uk/business/2011/apr/05/world-economy-faces-black-swans-imf-strauss-kahn), modern analysis does not handle black swans well. In clinical research, a major purpose for favoring hypotheses stated in advance is to prevent researchers from falsely enhancing the value of their study by building hypotheses around the resulting data -- including its random variations.

We believe that an analysis revolution is possible here, since there must be better ways to handle these problems that what is done today. As a stopgap, we suggest letting medical researchers claim a black-swan exemption, when justified by an unexpected result that changes the whole picture -- giving them more freedom to highlight interpretations developed after they had seen the results, while also calling in more scrutiny from other professionals, to check that they were using the exemption in ways that were not misleading, and that made practical sense overall.

[6] Interim CDC Guidance on Pre-Exposure Prophylaxis for HIV Prevention in MSM (short overview with highlights for physicians, January 28, 2011), http://aids-clinical-care.jwatch.org/cgi/content/full/2011/214/1

[7] Interim Guidance: Preexposure Prophylaxis for the Prevention of HIV Infection in Men Who Have Sex with Men (the official document), http://www.cdc.gov/mmwr/preview/mmwrhtml/mm6003a1.htm

[8] On the importance and the difficulty of reaching superspreaders in HIV, see: Autonomous targeting of infectious superspreaders using engineered transmissible therapies. PLoS Computational Biology; March 17, 2011, http://www.ploscompbiol.org/article/info%3Adoi%2F10.1371%2Fjournal.pcbi.1002015.

[9] Test-and-treat not enough to control HIV epidemic in the US, http://www.aidsmap.com/Test-and-treat-not-enough-to-control-HIV-epidemic-in-the-US/page/1724334/

Note: This post is Part I. We are researching Part II, on obstacles and opposition to using Truvada PrEP for individual protection, or to help end the HIV/AIDS epidemic. Part II will be controversial.

10 comments:

incidence0 said...

It is innacurate to say that "Of those who actually took one Truvada pill per day throughout the study, not one became HIV-infected during this trial", because the researcher do not know who took their pills everyday and have no means to know that informantion (they were not there to check what each and every participants were doing). This is why the figure of 44% is used and not that of 100% because in the real world that's what the efficacy is.
The drug study was done on a subgroup wich is not representative of the overall trial. It is a matched cohort of less than 200 ppts and it is not appropriate to draw to much form such a small substudy.

John S James said...

We DO know that everyone in the entire study (more than 1,000 participants in the Truvada arm) who were infected DID NOT take their pills as directed, at least part of the time.

Of these 36 people, 33 had no detectable level of either drug in their bodies, when the drugs should have been there. And the other three had very poor levels, just above the limit of detection, consistent with very poor adherence.

Everything we know suggests that these 36 could not have been taking one pill a day consistently in the weeks before their blood was tested for the drugs. Whether they took the pills at other times in the trial we don't know.

Everyone else in the Truvada arm was not infected during this trial. This compares with 64 who were infected in the placebo arm.

True, we don't know how many in the Truvada arm were protected by the drugs, and how many were just lucky.

But we do know that NOBODY who did use the drugs once a day throughout the whole study, as directed, got HIV. Unless the researchers could not find the drugs when they looked -- and they were very careful in validating these drug-level tests.

While the total number in the drug-level study was under 200, that represents ALL of the Truvada "failures" (36 persons), with two matched controls selected for each. There was no real need to test for drug levels in the people who did not get HIV. So this "small" drug-level study actually represents the whole study, with over 1,000 people in each of the two arms (Truvada and placebo arms).

incidence0 said...

I agree with the firt part or your comment which does not say the same thing as your main article.

However when you say that "But we do know that NOBODY who did use the drugs once a day throughout the whole study, as directed, got HIV." this does not mean that,s because they were protected, as you note yourself, they may have been lucky, or simply they may have not been in contact with the virus. And again we don't know who used the drug everyday in the whole study. You are making an inference here:

Your main post suggest that Truvada work 100% when taken as prescribed (daily) but this is an inference as the trial was not comparing daily dosage versus non-daily dosage. This misundertsanding has been repeated many times in the news with the 73% protection figure when the drug was taken 90% of the case, because it is not what the trial was investigating. The subtudy used matched controle whose drug level was detected in 51% of the case (you can check this by watching Bob Grant slides on the CROI website). This clearly suggest that despite drug nit being detected they remained HIV-Neg.

Robert Grant never claimed, and would never claimed that it work 100% of the time in this study or in the real world.

Please note that it does not mean that it does not work. I am just trying to introduce some cautionary note in your otherwise good article!

incidence0 said...

PS:

Grant's presentation at CROI:

http://bit.ly/grjVUg

Also whatch Rivet Amico about adherence and what it is possible to say and not to say.

In short you can only say that those who became infected did not take their pills regularly.

John S James said...
This comment has been removed by the author.
John S James said...

Let's start with where we agree. To quote Dr. Grant, the principle investigator of the entire iPrEx study, "No one in iPrEx acquired HIV infection with a drug level that would have been expected with daily dosing" (at the Retroviruses conference, in the video you linked to in a comment above). 

I believe you DISAGREE with the second sentence in the following, from my article: 

********** 

"Of those who actually took one Truvada pill per day throughout the study, not one became HIV-infected during this trial (vs. 64 of those who had been given the placebo). So for those who actually used the drug as directed, throughout the entire study, it was 100% effective in this trial, which had over 1,000 high-risk gay men assigned to take Truvada. 

"But the blood draws occurred in this study at weeks 4, 8, 12, 16, 24, and then every 12 weeks. So we cannot rule out the possibility that one or more people who had not taken Truvada for about two weeks or more before one of those visits, then got religion after the visit, and starting taking the pills daily -- yet got infected despite having adequate drug levels. This seems unlikely in view of the overall findings. But if it did happen, then it would mean that the protection was less than 100% in those who were using the drug. It was not possible to have everyone wear a device that that could record drug levels continuously -- and also tell exactly when they got HIV. So there is no way to be sure that nobody had enough drug in their body when they were infected. For this reason, 100% effectiveness is not claimed or reported.[5] 

"Still, the fact remains that Truvada for HIV prevention worked very well in this trial. Truvada when used properly (once a day, throughout the trial) completely prevented HIV infection, in a study of over 1,000 people -- when a comparable, randomly chosen group of 1,000 other people given a placebo had 64 HIV infections during the same time. Almost no drug for anything works this well." 

********** 

I think this puts the "100% effective" in an accurate, limited context -- though reasonable people can differ. It's the observed on-treatment result from this trial -- not from the next trial, not from using Truvada PrEP out in the world. Of course the drug will not work perfectly, so the real protection will be somewhat less than 100%. But in this trial, NO ONE who used Truvada at all properly got HIV -- compared to 64 who did not use Truvada at all, in the randomized placebo control group that was designed to be similar to the Truvada group. This looks like 100% to me, as the readout from this particular trial. 

Notice that the arithmetic steps to get the 100% are exactly the same steps that the iPrEx team used to get the 44%. The difference is that the iPrEx team used intent-to-treat, meaning that it counted 36 people who were assigned to take Truvada as doing so, even though WE KNOW that they used Truvada grossly improperly, if they used it at all, at least part of the time during the study. NONE of them had drug levels expected to be protective (90% of them had no detectable drug at all) -- while 97% of the U.S. participants did have adherence as seen by drug levels, showing that it is possible for a great majority of patients to protect themselves. 

Here's the math: 
(64-36)/64 = 43.75%, approximately 44% 
(64-0)/64 = 100% 

The reason for the 44% intent-to-treat result is that something went wrong at some of the trial sites. No one yet knows what went wrong. Our guess is that many participants gave or sold their pills to people who had no other HIV treatment access and needed them to stay alive -- and of course concealed this from the researchers, so that their supply would not be cut off. The two U.S. sites were in Boston and San Francisco -- probably the best cities in the U.S. for access to HIV care. Probably none of the U.S. participants knew anybody who wanted their pills.

Anonymous said...

Drug levels are a reflection of adherence. Adherence is one aspect of behavior. How do you know that those who adhered to the drug regimen did not also adhere to condom use? Very difficult to isolate a single behavior with any certainty. Intent to treat analysis avoids that trap. Be cautious about claims greater than 44% reduction in risk of infection.

J. Jeff McConnell said...

Dear John,

Thank you so much for this. I regret that we did not do more as investigators to help the media and the public understand the results of iPrEx. It was quite apparent to us that the 44% figure far from revealed the high level of effectiveness that Truvada reached in MSM. I find no flaws in any of your reasoning or math here and appreciate your candor. The media, at least initially never really got the study results wrong, I am quite thankful for that, but they rarely got the interpretation quite right either. Many reported 51% and the 72% efficacy numbers too claiming that they represented efficacy when people took their pills as directed. I find these numbers the least informative because they are based on adherence self-reports that we knew were very poor indicators of actual adherence, they simply suggested more pills is more protection. And we simply do not know who, if anybody, actually took the pill daily as directed.

We are still working to better understand what happened in iPrEx and doing what we can to help figure out why Truvada failed in FemPrEp, a crushing outcome. For me, in regards to MSM right now the most pressing question is how many men and transgendered women will take how many pills now that they know that Truvada works at least as much as reducing infections by over 90% in people who take it enough to keep drug levels high, and as you have reasonably pointed out quite possibly as much as 100%. Once again I am so pleased that our trial was so international (even though that can be a rough haul some days). We now know that the answer to that question may be profoundly different for individuals in different parts of the world despite being unified by the risk of HIV infection. Our sites are getting ready to start the open label extension (iPrExOLE) where we invite all the men and women of iPrEx back for 18 months and offer everybody the real pill and tell them what we learned with their help from the first trial. We assume that we will get more than 50% adherence under these conditions but we do not know, and if so we do not know how high it might go.

John, I trust you will be around to explain our results. I look forward to it.

Best regards,
Jeff
Investigator, Global iPRrEx

Anonymous said...

Hi John, I have authored an article that will appear in the next issue of Current HIV Research. You may be interested in reading this article. The abstract is e-published ahead of print. you can email me at the address in the article and I will forward the full text to you. Sincerely, Marc C. E. Wagner

Pearly Home Remedy said...

The drug study was done on a subgroup wich is not representative of the overall trial. It is a matched cohort of less than 200 ppts and it is not appropriate to draw to much form such a small substudy.