Wendy Carter chose her words with care on Tuesday afternoon. As acting director of the FDA’s Office of Infectious Diseases, she announced the approval of GSK’s Hibsago (bepirovirsen) with a sentence that promised exactly as much as it could defend: the drug “provides a new treatment approach that gives some patients the possibility of maintaining control of the virus after all chronic HBV treatment has ended.” The FDA approval announcement never said which patients she meant. GSK’s own trial counts say it more bluntly than anything the agency put on the page.
Among trial patients who started with 1,000 IU/mL or less of hepatitis B surface antigen (HBsAg, the blood marker of persistent infection), 200 of 768 reached functional cure, or 26 percent, according to the GSK B-Well results release. Take those counts out of the overall totals and you are left with 452 treated patients who started between 1,000 and 3,000 IU/mL, of whom 33 were cured, about 7 percent, or roughly one in 14. The label the FDA signed this week covers both groups.
| Group | Value (percent) |
|---|---|
| 1,000 IU/mL or below | 26 |
| Above 1,000 to 3,000 | 7 |
Hibsago is the first antisense oligonucleotide approved for chronic hepatitis B, a short synthetic strand that breaks down viral RNA, slows replication and, by its developers’ account, primes the immune system to finish the job. The disease it targets is one medicine has long managed without ending. GSK, citing WHO figures, counts about 1.7 million Americans and more than 240 million people worldwide living with chronic infection. Most who are treated take a nucleoside or nucleotide analogue pill every day, often for life, which keeps the virus quiet and almost never clears it; the company puts functional cure on that regimen at less than 1 percent, citing a 2020 meta-analysis.
The zero in the placebo column
GSK ran two identical phase 3 trials, B-Well 1 and B-Well 2, across 29 countries, with results published in May in the New England Journal of Medicine (NEJM phase 3 paper). Every participant was already on long-term antiviral pills, had no cirrhosis, and carried surface antigen no higher than 3,000 IU/mL. In the efficacy counts, 1,220 got a weekly 300 mg injection of bepirovirsen for 24 weeks and 614 got placebo, all while staying on their pills. Patients whose antigen and viral DNA had dropped out of measurable range qualified to stop the pills at week 48, and at week 72, after 24 weeks off all treatment for those who stopped, investigators counted functional cures: surface antigen undetectable, viral DNA below the limit of quantification.
| Measure | Share |
|---|---|
| of treated patients met the criteria to stop their antivirals | 24 percent |
Pooled, 233 of the 1,220 treated patients got there, 19 percent. None of the 614 on placebo did. That zero reads like a revelation and is mostly a description, because the placebo patients kept taking pills that rarely clear the antigen, and the antigen stayed. Measured against the less than 1 percent baseline, 19 percent is a large step, and GSK’s headline in May called it “unprecedented.” Tony Wood, the company’s chief scientific officer, phrased it more carefully in the body of the release, saying the drug “offers the possibility of significantly better functional cure rates” than standard care.
The pooled figure blends two very different populations. Trial by trial, the CIDRAP trial report on the NEJM data shows functional cure at 25 percent in B-Well 1 and 28 percent in B-Well 2 for patients who started at 1,000 or below, and at 10 percent and 5 percent for those who started higher. Anna Lok of the University of Michigan, whose editorial ran alongside the paper, called the upper group’s rate “substantially lower than that among those with the lower HBsAg level.” The American College of Gastroenterology’s evidence review put it in one line: “Baseline quantitative HBsAg stratification dramatically changed response rates.”
The FDA set eligibility at 3,000 IU/mL or less, measured by a companion test authorized alongside the drug, Roche’s Elecsys HBsAg II quant II RxDx assay, for adults without cirrhosis who have been on antivirals at least six months with suppressed viral levels. Those 26 and 7 percent figures are group averages, not forecasts for any one patient, but they are the best estimate a prescriber has, and the gap between them is wide.
Six months of needles, and then the wait
A one-in-14 chance might be worth taking if it cost nothing. Hibsago carries a boxed warning, the FDA’s most serious label caution, for serious liver reactions and for low platelet counts that can raise the risk of potentially life-threatening bleeding, with further warnings for worsening kidney function, serious inflammatory reactions such as vasculitis, and harm from improper injection. In the pooled trials, 91 percent of bepirovirsen patients had an adverse event against 73 percent on placebo; serious adverse events ran 7 percent against 4 percent, and events of grade 3 or worse ran 16 percent against 3 percent. The most common of those severe events was a rise in the liver enzyme ALT, in 6 percent of treated patients. Counted at a lower bar, ALT at least three times the upper limit of normal, the ACG review found elevations in 24 percent of patients during active treatment.
The ACG reviewer reads those flares as immune-mediated clearance of infected cells rather than drug-induced liver injury. Clearance and injury are not mutually exclusive, and the FDA, looking at the same data, put serious liver reactions in a box at the top of the label. Platelets and kidney function also dipped during treatment and recovered afterward, a pattern familiar from the antisense drug class.
The trials monitored blood every one to two weeks, and the ACG reviewer expects practice to need the same. So a working adult on Hibsago spends six months on weekly injections and frequent lab draws, then keeps taking the daily pill for at least 24 more weeks before learning whether they qualify to stop. According to the FDA, 24 percent of treated patients met that bar, which means 76 percent finished the course still on the pill the drug was meant to replace. Neither the agency’s announcement nor Reuters’ approval-day report mentioned a price.
What the trials measured is functional cure 24 weeks after treatment ended. Lok called the results “impressive but cannot be generalized to patient groups that were not included in the trials,” and said longer follow-up is needed to confirm the antigen loss lasts; the ACG review lists cirrhosis and HIV, hepatitis C and hepatitis D coinfection among the untested groups and calls multiyear durability unestablished. GSK’s release adds that antigen loss is associated with an 89 percent lower risk of liver cancer and a 62 percent lower risk of death from any cause. Those numbers come from a retrospective cohort study of US health records, in people who lost the antigen by any route. Nobody in them received bepirovirsen, and the trials were neither long enough nor designed to test either claim.
GSK funded both trials, its scientists sat in the international study group, and it licensed the molecule from Ionis, which developed it with GSK. The FDA gave the application Fast Track status, Breakthrough Therapy designation and Priority Review.
For a noncirrhotic patient whose antigen sits under 1,000, the drug offers something close to one-in-four odds of getting off the daily pill and staying off it through the 24 weeks the trials watched, and the ACG reviewer wrote that it “will change my practice” for those patients. The agency drew its line at 3,000, the enrollment ceiling of GSK’s own trials, and gave the public the pooled 19 percent. The 7 percent did not make the announcement.