A Critical Analysis of the iPrEx PrEP Study

A big nothing burger

Before we get into this analysis, I would like to point readers who are new to this Substack to read “Do I think HIV exists?” in order to fully understand my motivations in doing these types of analyses. This piece in no way endorses the HIV AIDS story. Also, this post is rather technical.

Today we are going to be analyzing the results of the iPreX (Pre-exposure Prophylaxis Intitiative) PrEP study from December 30, 2010, published in the New England Journal of Medicine. It is one of the most highly-cited studies on PrEP. Here is the text of the study:

Preexposure Chemoprophylaxis for HIV Prevention in Men Who Have Sex with Men

I will note that there is a kind of update to this study, published in 2014 in The Lancet, but it doesn’t really add any wildly new information to the results of the study under discussion today as it mostly discusses how to promote uptake rather than analyzing efficacy; furthermore, I want to look at the original papers promoting PrEP, so iPrEx it is. You can find the 2014 paper at this link.

The first thing I noticed about this study is there are 35 authors listed, and “et al for the iPrEx Study Team.” That’s a whole lot of authors, although I understand that in cases like this, people like lab assistants are listed as coauthors, so I won’t be too fussed about this. The second thing I noticed was that this paper has the shortest abstract background statement I have ever read; it simply states: “Antiretroviral chemoprophylaxis before exposure is a promising approach for the prevention of human immunodeficiency virus (HIV) acquisition.” I mean, I guess they don’t need to say anything more.

Also, before we dive in, I also want to point out the “Funding and Disclosures” section. I’ll add this as a block quote because it is long, and also will note that all boldface in block quotes is my own with the exception of links:

Supported by the Division of Acquired Immunodeficiency Syndrome (DAIDS), National Institute of Allergy and Infectious Diseases, National Institutes of Health, as a cooperative agreement (UO1 AI64002, to Dr. Grant) and by the Bill and Melinda Gates Foundation. Study drugs were donated by Gilead Sciences. The pharmacological studies were sponsored by a cooperative agreement with DAIDS (UO1 AI84735, to Dr. Anderson). Support for some specimen handling came from a grant from DAIDS (RO1 AI062333, to Dr. Grant) and by the J. David Gladstone Institutes. Some infrastructure support at the University of California at San Francisco was provided by a grant from the National Institutes of Health (UL1 RR024131).

Disclosure forms provided by the authors are available with the full text of this article at NEJM.org.

This article (10.1056/NEJMoa1011205) was published on November 23, 2010, and updated on December 1, 2010, at NEJM.org.

Dr. Grant reports receiving support as an advisor to Siemens (the manufacturer of TruGene HIV-1 Genotyping Kit); Dr. Mayer reports receiving grant support from Gilead, Merck, and Bristol-Myers Squibb; Dr. Kallás reports serving on a data and safety monitoring board for Merck; Dr. Schechter reports receiving consulting fees and grants from Gilead; Drs. Liu and Anderson report receiving donations of study drug from Gilead for various PrEP projects; and Drs. Jaffe and Rooney report being employees of Gilead Sciences and owning stock in the company. No other potential conflict of interest relevant to this article was reported.

Employees of Gilead Sciences who own stock in the company? Hmm. Already, I’m thinking their results might be too good to be true. (Spoiler: the results aren’t too good to be true; they’re actually wildly underwhelming.) I’ll give you the main takeaways from this paper first and then do a somewhat deep dive.

Here is the most important thing to remember about the results of this study: 2,499 seronegative men or transgender women were divided into a placebo group and a FTC-TDF group for 3,324 person years (maximum follow up 2.8 years, median 1.2 years). 36 “became infected” in the FTC-TDF group and 64 in the placebo group, representing a 44% reduction in the incidence of “HIV.”

Hold up—44%? Aren’t we told that PrEP is “>99% effective”? How did they get to that from 44%? Actually, it is quite a bit worse than that, because the 95% confidence interval given for “effectiveness” ranges from 15% to 64%. If you remember intro stats, you will know that a confidence interval that wide is a huge red flag to wildly inconsistent results. It means that 95% of results can be expected within that interval, and 5% outside it. But for a confidence interval to be truly meaningful, it needs both a high level of confidence—at least 95%—and a narrow width. A range of 49 percentage points ought to raise suspicion.

Regardless, in both groups there were seroconversions, although the authors attempt to explain this away by extrapolating what might have happened had their subjects been “fully compliant”—more on that later, because it’s some tortured reasoning. Let’s get into the study design and results.

In the Introduction, the authors state that “We selected emtricitabine (FTC) and tenofovir disoproxil fumarate (TDF) combination therapy in a single tablet (FTC–TDF) for evaluation of preexposure prophylaxis because of several favorable characteristics.” I wonder if one of those characteristics is several of the authors’ being employed by and owning stock in Gilead. Also, we know all about those so-called “favorable characteristics.”

Moving on to Methods, note that the protocol was sponsored by DAIDS, a division of the NIH. Countries involved were Peru, Ecuador, South Africa, Brazil, Thailand, and the United States (but only in Boston). Participants must have been “assigned male at birth” and at least 18 years of age, and HIV negative. They were tested for “HIV” and hepatitis B at screening, and follow up visits occurred every 4 weeks, at which participants were given their medications, had pill counts taken from previous prescriptions, given “adherence counseling” (that isn’t creepy or coercive at all), and, once again, rapid tested for “HIV antibodies.” Furthermore, “High-risk behavior was assessed by interview every 12 weeks, and physical examinations and evaluations for sexually transmitted infections were performed at least every 24 weeks.”

Regarding testing—and this is really important, for reasons I will explain, so I’m putting it in a block quote:

Testing for HIV antibody was performed on whole blood with the use of two different rapid tests at every scheduled visit, and reactive rapid tests were tested with the use of Western blot analysis of serum (Figure S1 in the Supplementary Appendix). Subjects with failed rapid tests were retested during the visit. HIV plasma RNA testing with the use of an assay with a lower limit of quantitation of 40 copies per milliliter was performed if seroconversion was detected within 12 weeks after enrollment. RNA testing was also used to identify the first date of laboratory evidence of infection for the as-treated analysis. Testing for drug-resistance genotyping and phenotyping was performed with the use of clinically validated assays on the basis of the viral load at the seroconversion visit.

Why is this so important? (Note that this is not relevant to the results of the study, so you can skip the next two paragraphs if that’s your interest here, but it IS highly relevant in the context of the all too perfect HIV AIDS story, so bear with me if you wish.) First, recall that to be “undetectable” on a viral load test, the criteria is “fewer than 20 copies per milliliter.” So if I’m reading this right, if seroconversion was detected within 12 weeks of enrollment, the first time a person has more than 40 copies of “HIV genetic material” per milliliter, that indicates when seroconversion occurred? That cannot possibly be the case, per the mainstream model of “HIV infection and progression to AIDS.” How many more than 40? The paper never says.

The reason I find this so concerning dovetails with the fact that I noticed that among the 100 “seroconversions,” no viral load data were given or even indicated for any of them, despite the large number of total tests—over 39,000—these participants were given. As a reminder, the mainstream model claims that within weeks of “infection,” many (but not all, and no one really knows how many) will experience a flu-like illness, at which point—per the mainstream—their viral load is allegedly higher than at any other point during infection. In fact, this phenomenon has been cited as indicating that true viremia does occur in “HIV” patients. The problem is that there is little to no clinical information available to support this assertion; rather, “acute HIV infection” or “seroconversion syndrome” is exclusively defined clinically. I can understand why there are few actual studies on viral load in “acute infection;” by necessity, such studies will be retrospective in nature and not carried out in real time. However, the “seroconversions” in this study presented a missed opportunity to quantify the “viral load” during “acute infection,” and I can’t help but to think that there is a reason this analysis was not performed. Perhaps the possibility of false positive viral load results stayed their hand.

Moving on from that oddity, another important item to note is that a “subgroup analysis of drug levels” was performed. This analysis represents the trickery that was used to somehow magically turn 44% efficacy into 95% efficacy.

A prespecified subgroup analysis was performed to investigate whether drug levels correlated with protective effect. Subjects with HIV infection were matched with two control subjects, one from each study group who were selected from among seronegative subjects, according to study site (Figure S5 in the Supplementary Appendix). Plasma was tested for the presence of FTC and tenofovir (TFV), and peripheral-blood mononuclear cells were tested for FTC triphosphate (FTC-TP) and TFV diphosphate (TFV-DP), which are the active intracellular metabolites of FTC and TFV, respectively, with the use of validated liquid chromatography and tandem mass spectrometry assays.

Let’s move on to their results. This won’t take long because I’ve told you everything you really need to know at the outset. The rates of self-reported pill use were 89% among the FTC-TDF group versus 92% in the placebo group initially, but after 8 weeks were similar between the two groups at 95%. This is pretty high uptake, per self reporting, although as we will see, the study team clearly didn’t trust self reporting.

There is also discussion of adverse events, focusing mainly on serum creatinine levels, an indication of kidney function. A total of 10 elevations of serum creatinine led to drug discontinuation. Also, “Serum creatinine levels were elevated at more than one consecutive test in 5 subjects in the FTC–TDF group (<1%) and in none of the subjects in the placebo group. All elevations in the serum creatinine level resolved after the discontinuation of a study drug, within 4 weeks in 3 subjects, within 12 weeks in 1 subject, and within 20 weeks in 1 subject. Four of the subjects resumed taking FTC–TDF without recurrence of the elevation.” Given that the median length of observation for this study was 1.2 years, the fact that this side effect showed up so soon is alarming, although as the Truvada disaster indicates, perhaps it shouldn’t be too shocking.

Here is the money quote; the important part is bolded by me:

HIV rapid testing was performed at 39,613 visits, during which there were false reactive tests for 3 subjects at 7 visits; each subject had multiple negative tests afterward. HIV seroconversion was observed in 110 persons, of whom 10 had plasma HIV RNA subsequently detected in specimens obtained at the enrollment visit. A finding of fewer than 40 copies per milliliter of plasma HIV RNA was documented for the other 100 HIV-infected subjects before seroconversion. Among the 100 subjects with emergent HIV infection, 36 occurred in the FTC–TDF group, and 64 occurred in the placebo group, representing a relative reduction of 44% in incidence in the modified intention-to-treat population (95% confidence interval [CI], 15 to 63; P=0.005)

After adjustment for the difference in age between the two groups, the efficacy was 43% (95% CI, 14 to 62). The rate of pill use on 50% or more of days was recorded on the basis of pill counts, self-report, and dispensation records at 81% of visits on which efficacy was 50% (95% CI, 18 to 70; P=0.006). This rate did not differ significantly (P=0.48) from the efficacy at visits with less than 50% pill use of 32% (95% CI, −41 to 67%) (Figure 3). Efficacy of less than 30% could not be ruled out in the modified intention-to-treat analysis (P=0.15) or in the prespecified as-treated analysis at 50% pill use (P=0.09). There was no evidence of a change in HIV efficacy with longer follow-up (P=0.44).

43% efficacy, with a confidence interval of 14% to 62%. That upper limit is not anywhere near 99% or even 90%. Also, returning to the “acute infection,” we have the following, which I have bolded in its entirety for a reason:

Among the 10 subjects in whom plasma HIV RNA was subsequently detected in specimens obtained at enrollment, 5 had symptoms of an acute viral syndrome at enrollment, 2 had symptoms 1 week later (prompting an interim study visit), 1 had an anal sore, and 2 had leukopenia at enrollment. In these subjects, the clinicians did not suspect acute HIV infection, because the symptoms were attributed to an upper respiratory tract infection, sinusitis, or other non-HIV cause.

Why was a viral load test not performed? Why did the clinicians assume that their symptoms were caused by “an upper respiratory tract infection, sinusitis, or other non-HIV cause” when they could have, presumably, per the mainstream narrative, simply tested their viral load and found it to be sky high? This would not only have the effect of “confirming acute infection,” but also of providing actual clinical evidence that “acute HIV infection” is not just a figment of the imagination. Moving on. We’re almost done.

The last subsection of interest is called “Drug-level detection and prophylactic effect.” Considering their wildly unremarkable results, the study authors came up with the neat trick of attempting to detect traces of FTC-TDF in dried blood spots obtained from participants, and correlating the levels detected with the probability of seroconversion. This is statistical trickery, pure and simple. I don’t have to point out the inherent lack of rigor in such a methodology. Here’s the last block quote:

In the FTC–TDF group, among subjects with a detectable study-drug level, as compared with those without a detectable level, the odds of HIV infection were lower by a factor of 12.9 (95% CI, 1.7 to 99.3; P<0.001), corresponding to a relative reduction in HIV risk of 92% (95% CI, 40 to 99; P<0.001). After adjustment for reported unprotected receptive anal intercourse, the relative risk reduction was 95% (95% CI, 70 to 99; P<0.001).

There are a few glaring problems with this. The first is the utility of this methodology at all, but would you look at the confidence interval? We are given two; one is from 1.7% to 99.3% efficacy, which is completely useless as it basically encompasses every possibly level of efficacy or non-efficacy; the second is from 40% to 99%, which is only marginally better. Ignore the p-values; they have limited utility (that’s a topic for another day). The fact is that if you need such a wide range of values in which to fit 95% of your results, that doesn’t tell you very much at all. In fact, what it tells you is that the results demonstrate so much variability as to render them nearly useless.

We’re moving into the Discussion now, so we’re almost done. Thank goodness. Once again, here is their conclusion, the only takeaway you really need: “Once-daily oral FTC–TDF provided 44% additional protection from HIV among men or transgender women who have sex with men who also received a comprehensive package of prevention services. The protective effect of FTC–TDF was significant but not as high as originally hypothesized during the design of the study.” The authors can’t seem to let go of the idea of measuring drug concentrations to ensure “compliance”—after all, the results of this paper without the drug concentration analysis represent nothing new, so no wonder they love this idea. But it veers into creepy territory with the following statement: “Drug level may have a role in monitoring trials, programs, and individual users. Methods for inexpensively measuring long-term drug exposure, such as that afforded by analysis of hair, would be helpful once such a method is fully validated.” All I can say to that is YIKES.

Regarding side effects, this study also provides some ominous foreshadowing. “Side effects may have contributed to low pill use among some subjects. As with treatment of HIV infection and the use of FTC–TDF in postexposure prophylaxis,²⁰the initiation of FTC–TDF preexposure prophylaxis was associated with self-limited start-up symptoms in a few subjects. The trial design involving a placebo may also have contributed to lower-than-expected pill use.” Of course, we are all aware that placebo controls have been abandoned in trials for “anti-HIV drugs,” but what possible justification can there be to even suggest that placebo may have been a problem in a study of HIV-negative people? That this statement was even included is alarming. They also mention that “TDF treatment is known to cause decreases in renal function.” No kidding.

Finally, one more note on “acute infection”: “Ways to increase recognition of acute HIV infection would include routine measurement of body temperature and testing for HIV antibodies to evaluate viral syndromes, regardless of whether the presentation suggests HIV infection or another cause. Testing for HIV RNA at the time of the initiation of preexposure prophylaxis should be considered where available.” Why not just do it every time? We know from Covid that PCR testing can easily be done on a massive scale, so why not make this routine if “acute infection” is suspected? This would be the easiest way to shut down a lot of critics, but do we see it happening?

We have finally come to the end of this paper, and here is their final conclusion:

The optimal regimen for preexposure prophylaxis has not been established, and data from the subjects in our study cannot be applied to other populations. Alternative regimens in different populations are being studied.

In academia, this can be translated to mean “Our study didn’t really show anything.” Period, full stop. This entire paper is a giant nothing burger and establishes very little, yet it is one of the most highly cited papers on PrEP efficacy. That should tell you everything you need to know. We are pushing PrEP on entire communities based on this kind of evidence. It’s total insanity. This paper reminds me a little of the 1997 Padian paper on heterosexual transmission that did not show what it intended to show, which was condom efficacy. Instead, they followed over 400 discordant couples for ten years and observed no seroconversions despite inconsistent, and sometimes no, condom use.

I would like to note as a kind of addendum that the web version of this paper allowed comments, and some of the comments were rather interesting. There were comments to the effect that in real life, the level of follow up adhered to in this study would not be feasible and so we could expect results to be even worse than indicated in the paper. (How could they be worse? A confidence interval that goes from 1% to 99% is about as bad as you can get.) But there was one interesting comment, by an individual named Eugene Semon, whose name sounds familiar—is he an AIDS critic? Regardless, an excerpt of his comment reads: “With all due respect to the hard work this study represents: its results appear to offer nothing new for the practicing clinician - which I’m not.” And with that, I will conclude because I really couldn’t say it better myself.

As always, let me know what you think in the comments.

To support my work on Substack, please purchase my book for yourself or for a friend, and leave a review on Amazon. You can learn about efforts to ban my book here. You can also buy my new book Almost Cancelled.

If you’re a new reader and would like some background as to my views on HIV AIDS, including the “existence” question, please refer to this post and the links contained therein.