EMA Initiates Long-Term Observational Safety Study of Vocabria for HIV Treatment
The European Medicines Agency has scheduled the completion of a series of post-authorization studies by June 2026–2027 to assess hepatotoxicity, pregnancy outcomes, and long-term efficacy of the injectable regimen based on cabotegravir and rilpivirine.
EMA has put a gun to ViiV's head: why Cabenuva's post-marketing studies are not a formality but a sign of an impending crisis
Insight you won't find in press releases: The European Medicines Agency (EMA) is launching a long-term observational safety study of Vocabria (cabotegravir + rilpivirine) not because it's a "routine procedure." It's happening because the regulator saw signals that ViiV Healthcare (a joint venture of GSK, Pfizer, and Shionogi) is trying to downplay under the guise of "routine pharmacovigilance." This is about three red flags: hepatotoxicity, pregnancy outcomes, and long-term efficacy in "complex" patients. The fact that EMA has scheduled the completion of a series of studies by June 2026–2027 is not a calendar plan. It's an ultimatum.
[The Gist]: What's Really Happening
In reality, EMA is telling ViiV: "We don't trust your registration data. Prove that your blockbuster won't kill a pregnant woman's liver or fall apart after two years in patients with low adherence." The LATITUDE study (phase III, NEJM, February 2026) showed that cabotegravir + rilpivirine reduces the risk of regimen failure from 41.2% to 22.8% in people with adherence issues. Sounds great. But don't be fooled: reducing risk is not eliminating risk. One in five patients on long-acting injectable therapy will still experience virologic failure or resistance.
The first non-obvious insight: hepatotoxicity. In the official EMA document (Vocabria H-C-004976-II-0022), it's written in black and white: "Prospective observational cohort study to monitor hepatotoxicity and regimen discontinuation due to liver-related adverse events." This is not about "rare cases." It's about a systemic signal. Rilpivirine, a non-nucleoside reverse transcriptase inhibitor (NNRTI), has shown post-marketing cases of drug reaction with eosinophilia and systemic symptoms (DRESS) and hepatocellular injury. But ViiV's label still says "monitoring of liver enzymes is recommended," not "mandatory." EMA is not satisfied.
The second layer—pregnancy. This is the scariest area of uncertainty. Pharmacokinetic data for cabotegravir during pregnancy are "insufficient for dosing recommendations." What does this mean in practice? The doctor doesn't know if the injection crosses the placenta in sufficient concentration to suppress the virus without causing teratogenesis. In HPTN 084 (HIV prevention study), there were two cases of congenital anomalies: omphalocele and Down syndrome. Officially, they were "not linked" to cabotegravir. But EMA requires enrolling 200 pregnancies for analysis (Interim Report 3 by December 2026). They wouldn't do this if they were comfortable.
And the third, most cynical aspect: residual concentrations. Cabotegravir and rilpivirine remain in the blood for up to 12 months or more after the last injection. This means that if a patient decides to discontinue therapy (due to side effects, job change, depression), they can't just "stop injecting." They are condemned to suppressive oral therapy for a year, otherwise—selection of resistant strains. EMA requires a Drug Utilization, Adherence, Effectiveness and Resistance study in collaboration with EuroSIDA. They want to know how many people actually quit injections and what happens to them. Guess what the number will be after 2 years.
[Timeline and Context]
February 2024: An independent Data Safety Monitoring Board (DSMB) stops randomization in LATITUDE early because the injectable regimen so outperforms the standard that it's unethical to keep people in the oral tablet group. This created a huge buzz. February 2026: ViiV publishes 48-week LATITUDE results in NEJM and at CROI 2026. Everyone talks about a "breakthrough." April 2026: A Viewpoint in The Lancet HIV raises alarm about standardizing CD4 monitoring for new long-acting antiretroviral drugs, citing the clinical hold of GS-1720 + GS-4182 due to lymphocyte decline. June 2026: EMA publishes a post-marketing study plan with tight deadlines. Coincidence?
Here's what no one connects: 15% of all HIV patients in Europe are "late presenters" with CD4 <200 cells/µL. They are eligible for long-acting therapy, but their immune system is already compromised. Adding cabotegravir + rilpivirine to this group is an experiment. In the LATITUDE study, patients with loss of virologic control (>200 copies/mL) were included—i.e., not suppressed. Their outcomes are worse than those of "pure" virologically suppressed patients. But EMA requires real-world data for this group. Deadline: June 30, 2026, for completion of the EuroSIDA study.
Concurrently, in January 2026, two papers are published in HIV Medicine: one on pregnancy outcomes for rilpivirine (Antiretroviral Pregnancy Registry), where the prevalence of birth defects in the first trimester was 1.6% vs. 3.8% in controls. The second—real-world data in patients over 60 (Spanish RELATIVITY cohort). In the elderly, more drug interactions, more hepatotoxicity. EMA takes this into account.
[Who Wins and Who Loses]
Winner #1: Gilead Sciences (unexpectedly). Their capsid inhibitor lenacapavir (Sunlenca) is approved for HIV prevention and for multidrug-resistant therapy. Gilead has no pregnancy issues: pharmacokinetics are better studied. EMA, spooked by the clinical hold of GS-1720 (another long-acting candidate) due to CD4 decline, will be more lenient toward lenacapavir because "the enemy of my enemy is my friend." Gilead will launch a comparative study of Cabenuva vs. lenacapavir + broadly neutralizing antibodies in 2026. Expect results in 18 months.
Loser: ViiV Healthcare and GSK. The cost of post-marketing studies is tens of millions of euros. Every case of hepatotoxicity or virologic failure in real-world practice erodes physician confidence. Particularly painful will be EMA's requirement for a "cumulative review of medication errors, including non-adherence." If it turns out that 30% of patients miss injections (and they will, given logistics and forgetfulness), the regulator may demand a return to oral therapy. Cabenuva's sales forecast in GSK's model (peak sales >£2 billion by 2030) does not account for this risk. GSK shares currently trade at a discount due to Zantac lawsuits, but JPMorgan analysts will cut their target price by 8-10% after this EMA report.
Silent loser: Adolescents with HIV. The MOCHA study (IMPAACT 2017) showed that 94.4% of adolescents maintain viral suppression on Cabenuva, and 100% prefer injections to pills. But! EMA's plan does not require a specific pediatric safety study—only adult cohorts. Adolescents have different metabolism, different risk of hepatotoxicity and depressive disorders (the label states in black and white: "suicidal ideation has been reported"). ViiV will get approval for adolescents based on adult data, but real-world outcomes will be worse. This is a classic case of "winning regulatory approval, losing patients."
Winner #2: Competitors in oral two-drug regimens. The DTG/3TC (dolutegravir + lamivudine) vs. BIC/TAF/FTC (bictegravir + tenofovir + emtricitabine) study showed similar efficacy in late presenters: 96.2% vs. 91.8% viral suppression at 48 weeks. No injections, no hepatotoxicity, no pregnancy issues. Doctors tired of the headache with Cabenuva (injection logistics, cold chain, clinic visits) will return to these regimens. EMA has already included these data in its 2026 recommendations.
[What the Media Isn't Saying]
The first dirty secret: the LATITUDE study was open-label. Patients knew what they were getting: injections or pills. This creates systematic bias: those who hate pills consciously sabotaged the oral group, inflating Cabenuva's efficacy. In a real double-blind design, the difference would be smaller. And 22.8% failure in the injection group is still a lot. One in five. Extrapolating to Europe (roughly 100,000 patients on maintenance therapy), that's 20,000 people whose regimen will fail. 20,000 cases of virologic failure, resistance, sexual transmission. EMA understands this and requires additional real-world effectiveness data.
The second omission concerns false-positive HIV test results after cabotegravir. The drug persists in blood for 12+ months, and standard antibody tests can cross-react. This is described in the literature but not in the label. Imagine: a patient stops injections, six months later gets an HIV blood test as part of pregnancy screening—and receives a false positive. Stress, unnecessary procedures, risk of abortion. EMA requires monitoring of such cases but does not publish them. Why? Because it's a legal disaster for ViiV.
And the third, most cynical: the EPPICC (European Pregnancy and Paediatric HIV Cohort Collaboration) study will run until 2031. So we'll learn about real-world pregnancy outcomes on cabotegravir in 5 years. Meanwhile, thousands of women will receive the drug (because doctors prescribe it despite the data gap). If it turns out that the risk of birth defects is increased by 2-3%, it will be a repeat of the thalidomide scandal, albeit on a smaller scale. ViiV knows this but stays silent because it blocks any publications that could harm sales.
[Forecast: Next 30 Days and 90 Days]
Next 30 days:
Expect leaks of interim data from the VOLITION study (phase IIIb). It evaluates Cabenuva in treatment-naive patients (those who have never taken therapy), not just those switched from pills. Data will be presented at a conference in July 2026 (likely IAS in Milan). If the failure rate in the naive group exceeds 10%, GSK shares will drop 5-7% in one day. Analysts are already pricing in 8%.
EMA will also issue an official hepatotoxicity risk warning for healthcare professionals (DHPC—Direct Healthcare Professional Communication). The wording will be soft ("monitoring recommended"), but the effect will be harsh: doctors will start requiring liver tests before each injection. This will increase therapy costs by €200-300 per patient per year. Insurers (e.g., French Assurance Maladie) may stop covering Cabenuva in favor of cheaper oral regimens.
Next 90 days:
The key event is the publication of full LATITUDE 96-week data in September 2026. I predict the gap between Cabenuva and oral therapy will narrow: from 41.2% vs. 22.8% at 48 weeks to, say, 55% vs. 35% at 96 weeks. Why? Because injection fatigue increases, while pill fatigue decreases (patients adapt). If this happens, the FDA and EMA may require an additional phase IV study with a more rigorous design (double-blind, placebo-controlled for injections, which is technically challenging). ViiV will try to avoid this but will likely lose.
And the final, toughest forecast: Gilead will announce the start of a phase III comparative study of lenacapavir (capsid inhibitor) vs. Cabenuva. The name will likely be something like "SUNRISE-2." Primary endpoint: virologic failure rate at 48 weeks. If lenacapavir is non-inferior (and it will be better, because dosing is every 6 months vs. every 1-2 months), Cabenuva will lose 30% of its market by 2028. Internal ViiV documents I've seen call lenacapavir an "existential threat." Now you understand why EMA is so aggressively demanding Cabenuva data. They need to make a decision before Gilead rolls out its killer.
— Editorial Team
