The journal Science is warning readers that an article it published on a much-hyped potential HIV remission approach didn’t include all the facts.
In a so-called “editorial expression of concern” published today, Science Editor in Chief Jeremy Berg, PhD, explained that the journal discovered that the strain of simian immunodeficiency virus (SIV) used in a non-human primate HIV remission trial is different than what the authors stated in their 2016 paper. And this undisclosed SIV variant “could affect the results presented in this Research Article.”
“There’s certainly nothing to indicate an intention to deceive,” Berg told Medscape Medical News. But he still felt it was important to tell researchers interested in pursuing this avenue that the literature was incomplete. This step will likely be followed by some further action, Berg said. That could include a correction, a revised and annotated journal article, or a full retraction.
“It’s more a question of imprecise use of language or an assumption that people would understand what it meant, even though it was not explicitly stated in the paper,” Berg said. He thinks Science may have enough information to make their decision in a few months. Whatever the decision, it will be published in the journal.
A Question of Precision
Science‘s statement of concern is significant because the 2016 Science study produced long-term remissions in macaques. As reported by Medscape Medical News, two of the seven monkeys that received the antibody that binds the alpha4beta7 integrin on the surface of CD4 cells maintained viral suppression for more than 9 months. The others did, too, after initial spikes in viral load. In contrast, the monkeys in the control arm all progressed and died of the virus.
Those findings prompted Anthony Fauci, MD, director of the National Institute of Allergy and Infectious Diseases (NIAID), to call the findings “striking” at the HIV Research for Prevention conference in October 2016.
And it led to a new monkey study to try to replicate these findings, also performed by NIAID. In addition, Fauci’s group launched a small study in humans, targeting the cell receptor alpha4beta7 integrin using vedolizumab, a drug already approved by the Food and Drug Administration to treat Crohn’s disease and ulcerative colitis.
But those studies didn’t get anything close to the results of the previous study, Fauci told Medscape Medical News this week.
“We used the same virus [in the new monkey study] and we did not get that striking result,” Fauci said. “That’s why I brought it to Jeremy [Berg’s] attention.”
Apparently he wasn’t the only one. Berg told Medscape Medical News that he received tips from a variety of directions, starting at the beginning of the year, that the virus used in the study that started all of this wasn’t exactly what the published paper said it was.
A Mutated Nef
One person who knew exactly what virus the 2016 study used, as well as the follow-up nonhuman primate study that wasn’t successful, was Francois Villinger, DVM, of the University of Louisiana at Lafayette. He’s the one who provided the virus for both animal studies.
And it was, Villinger told Medscape Medical News, the same virus. Both were a strain of the engineered SIV virus called SIVmac239 with a stop codon on the nef gene — that is to say, it’s a strain of SIVmac239 with a nef “stop” instead of “go.”
The problem, according to Berg and Fauci, is that the paper only identifies the strain as SIVmac239, not SIVmac239 with a stop codon on the nef gene.
This mutated virus “is somewhat weakened when the monkeys are first infected,” Berg said. “Then over time, it basically gets corrected in the monkeys.”
The conflict, then, isn’t over whether the mutated version of SIVmac239 was used and not specifically stated in the paper. (Berg, Fauci, Villinger, and Aftab Ansari, PhD, of Emory University, who was the lead author on the Science paper, all now agree that that was the strain that was used.)
The questions are who knew about the mutated strain when, why wasn’t it written in the paper, and whether it matters for the results.
The first question is somewhat still up in the air. Villinger said that it was obvious to him that if they were going to use SIVmac239 in a study that would last a year, they couldn’t use an unaltered version of it.
SIVmac239 is so virulent, he said, that in previous studies, monkeys died within 6 months of infection. For a study set to run for 12 months, that obviously wouldn’t do.
“If you have to place animals on [antiretroviral treatment] before 2 weeks because the virus is too virulent (ie, a study I am currently involved in), to ensure animals have an immune response worth salvaging therapeutically, the value of such a model is, in my mind, even more questionable,” he told Medscape Medical News. The understanding that a nef-mutated strain of SIVmac239 would be useful in studying HIV vaccines goes back to 1991, he said (Cell. 1991; 65(4):P651-662).
“It just appears that many investigators are either unaware or forgot about this fact.”
“I Wish the Hell I Knew”
But who else knew is unclear. Ansari said he didn’t learn that the virus he’d used in his study was a variant until December 2016, months after the study was published in Science.
“I had no idea it was a nef stop,” he said. “I wish the hell I knew.”
He said he regrets “not being clear in the Science paper how the virus was put together.”
Likewise, Fauci, who is a named author on the Science paper and whose team provided details used in one of the figures in the paper, said he didn’t learn that the virus was a nef stop until after publication.
“These are the subtle types of things that could explain the reasons we didn’t replicate the study,” Fauci said, “even though we used the same virus.”
What It Means
But whether the viral mutation is the reason others haven’t been able to replicate the Science study is unclear. Fauci thinks it may be, but Ansari told Medscape Medical News that that seems unlikely.
H. Clifford Lane, MD, NIAID’s deputy director for clinical research and special projects, was involved in the study to replicate the Science paper. He told Medscape Medical News that “to the best of our knowledge” the virus they used, which was a “gift” from Villinger, was the same virus used in the Science study.
Yet “we did not have any animals that totally suppressed SIV in the absence of antiretroviral drug,” he said. In a presentation of that study at the 2018 International AIDS Conference, the researchers make clear that they used the nef-stop version of SIVmac239.
Yet both Ansari and Villinger said that they don’t believe that the virus type they used invalidates their results, no matter when they discovered what it was.
After Ansari discovered that they’d used a variant in the study, he said he contacted Villinger and said, “We need to address this.”
They came to an agreement: Villinger would sequence the nef gene from all the monkeys to see whether the stop codon was still present in the monkeys before they began antiretroviral therapy and then the anti-alpha4beta7 treatment. If the virus had repaired itself and the stop codon was gone, Ansari said, perhaps there’s some mechanism, some dynamic between the monkeys’ own immune systems and the anti-alpha4beta7 treatment that could still offer hope as an avenue of investigation.
Villinger did do that sequencing, he said. And “by week 2, all animals had virus with fully coded nef replicating in vivo.” In other words, they may have administered a virus with a stop codon on nef, but the virus in the bodies of the monkeys by the time they administered the treatment didn’t have a nef mutation after all.
That data hasn’t been published yet, Villinger said.
Dismissing the Science paper’s results because of the mutation, Ansari said, “is like throwing the baby out with the bath water.”
“How did the animals control the viremia?” he asked. “I’m the first to admit that something funny is going on. But I’m trying to point out the fact that [NIAID] got the virus from Villinger and got different results.”
Why is that? He said he’s seen the raw data from NIAID that hasn’t been published yet, and the two groups analyzed the results in different ways, which could explain some of the discrepancy.
But Fauci isn’t so sure.
“I’m reluctant to say any field is dead; alpha4beta7 is an important pathway in HIV infection,” he said. “But whether an antibody to alpha4beta7 can lead to consistent, long-term suppression of the virus — that is clearly now questionable.”
And in the middle of this debate is Science and journals that present other researchers with the scientific disagreements and data produced using potentially different methods and different approaches to data analysis.
“This was a very interesting observation that got a lot of attention when it was published,” Berg said of the Science paper. “If it had a surprising effect and it was right [in its methods], that’s really important. If they got a surprising effect and it was wrong, we need to understand why.”
Science. Published online March 21, 2019. Full text
The Science study was funded by the NIAID. Fauci, Lane, Villinger, and Ansari have disclosed no relevant financial relationships.
By Heather Boerner