On October 3rd, in Silver Spring, Maryland, the Federal Food and Drug Administration convened a meeting of its antiviral advisory committee to discuss issues pertaining to the approval of a new HIV drug, tenofovir disoproxil fumarate (TDF), branded "Viread" by its sponsor, Gilead Sciences. Although the opinions of advisory committees are not binding upon the actions of the FDA, and in this case, the regulators did not even ask for a formal vote on tenofovir, historically, the agency has paid careful attention to the views of its expert advisors.
The group's deliberations provide a great deal of insight into the drug approval process, the ways different doctors and scientists view drug safety and efficacy, and the issues raised by this drug in particular.
What follows is an interpreted but faithful report of some of the participants' questions and opinions. For more comprehensive reviews of the data supporting Gilead's application for TDF and detailed descriptions of the clinical trials that supplied the data, you may wish to consult reports prepared by various online information resources (see Web addresses at the end of this article).
Q: One person died while on study 902. Do you know why?
A: The patient had a history of depression and committed suicide.
Q: In study 901, why were average CD4 cell count increases greater at the 300mg doses than at the 600mg doses?
A: The CD4 variability in study 901 is probably due to the small number of patients involved. From the study of high-dose, intravenous tenofovir (study 701), it appeared that the maximum anti-HIV activity was reached at the 300mg dose. So there is no further data on the 600mg dose after the 902 study.
Q: At 24 weeks, 3 percent of patients had the K65R mutation in the viral reverse transcriptase. Has that proportion remained consistent in subsequent weeks and in larger trials?
A: Yes, the rate remains low even after expanded dosing.
Q: How many patients in study 902 had a viral load higher than 50,000 copies/mL?
A: I don't know, but looking at the quartile baseline viral load, in the highest quartile the mean baseline viral load was about 70,000 copies/mL.
Q: Is continuation of 3TC necessary to get the increased activity of TDF against 3TC resistant virus?
A: We don't know.
Q: How durable was viral suppression over time; how often did blips in viral load occur during the year?
A: I don't know.
Q: Do you have any information on the number of previous drugs that the patients had used?
Q: Did you collect adherence data during the trials?
At several points in the meeting the committee turned its attention to specific concerns about an unusual bone formation abnormality associated with lack of phosphorus availability that was seen at high doses of tenofovir in animal studies. Parts of those discussions are collected here.
Q: Were continuous 24-hour urine tests done to look for abnormal phosphorus levels?
A: No, only spot urine collections were performed and these tests showed no statistically significant differences in the median change from baseline phosphorus fractional excretion between people on TDF and those on placebo.
Q: The differences may not have been statistically significant; nevertheless it appears that fractional phosphorus levels were higher for TDF at every time point on the slide.
A: They were not significant differences.
Q: Do you have any information about alcohol abuse or chronic diarrhea in regards to phosphate wasting in the study patients?
A: This was not analyzed.
Gilead hired a bone disease expert to put the toxicity issue into context. Dr. Teitelbaum made his presentation by telephone:
"Some bone basics: This problem is not the decreased mass of mineralized bone known as osteoporosis. In osteomalacia, bone is normally made, but unmineralized bone matrix accumulates because it cannot be mineralized in its environment.
"Bone doctors love to see osteomalacia because they can cure it. Hypophosphatemia (low blood phosphate level) is the most common cause -- the bone is not seeing enough phosphate. If circulating levels of phosphate are normal, osteomalacia will not develop.
"For example, patients on excessive antacid therapy can bind phosphate in the gut and develop osteomalacia; however, they recover when antacids are stopped.
"From the TDF data, this is not about hyperphosphaturia (due to excess renal excretion of phosphorus) because urine phosphate elevations were not seen. Therefore this is not serious in this case.
"The serious osteomalacia seen in the monkeys was reversible. The worst possible scenario for humans is that a patient develops osteomalacia as severe as was seen in the monkeys. They then stop the drug, get supplementation and they are cured."
Back to the Questions
The FDA invited two bone disease experts as independent evaluators of the toxicity data. Dr. Leukert attended by telephone; Dr. Bone (his real name) was present in person:
Q: (Dr. Bone) I would like to see a rate of decline for phosphate. Is there histology from the no-effect dose on monkeys?
A: No, histology is only available from the monkeys with osteomalacia.
Q: On the dog study, vitamin D levels were reduced. Was this seen on any other studies?
A: No other vitamin D levels were taken.
Q: (Dr. Leukert) When did phosphorus supplementation begin for patients in the studies?
A: It was mixed. Some began supplements at their first low phosphorus reading and 51 patients did not receive supplements. Most had low phosphate levels on no more than two consecutive visits.
Q: Were any specific questions asked to elicit the incidence of bone pain?
Q: (Dr. Bone) Regarding the necropsy studies on dogs and monkeys: Did you find a no-effect dose in that histology?
A: In dogs, the doses were 10-fold higher than in human doses; in rats, doses were 20-fold that of the human dose. The no-effect dose was 100mg/kg (similar to a daily dose of 6,000mg for a person; the recommended adult dose of TDF is 300mg/day). It was only after toxicity showed up in the efficacy studies that bone toxicity studies were designed.
Q: Did you measure magnesium status in humans or animals?
A: No, we only measured total serum magnesium levels.
Q: Your bone mass density (BMD) studies are mostly from lumbar spine and femurs. Were any BMD done on the forearm -- non-load bearing, cortical BMD measurements?
A: No cortical measurements were made.
Q: What is your long-term safety study commitment for TDF?
A: Most safety data will come from study 910 in which patients will be followed until December 2002. The safety data gained from expanded access patients will be reported separately.
Which Side Are You On?
At this point, the FDA representative presented the big question that the agency wanted the committee to consider and a round-table discussion commenced.
In what population has efficacy been shown?
Dr. Yogev: Efficacy has not been proved in a population with high viral loads or in those naïve to antiretroviral therapy. I wonder if the mean reduction of viral load by -0.6 log over placebo will also hold for those with higher viral loads? In other words, should TDF be among the first line of drug choices?
Dr. Hamilton: Whether or not naïve patients have been shown to respond is not as important as the issue of treating experienced patients with higher viral loads. Frankly, most of the people treated in these two studies may not have needed treatment at all, given the new guidelines. (Mean CD4 counts were about 375 and the mean viral load was under 10,000 copies.) I don't feel compelled to drive the virus low when it is futile and unnecessary.
Dr. Pomerantz: Some data is missing but we think we know what that is. We don't have data for naïve patients or for those with high viral loads -- so there are two missing data sets.
Dr. Shapiro: We haven't seen a lot of data. This experienced population may be exposed to certain risks, but we also think that there are benefits that outweigh the risks. However, the risk is greater than the benefit for those who have many other options. So the risk/benefit equation is mixed.
Dr. Tebas: Before using this drug in naïve people, I want them to show me some more data.
Dr. Munk: I have a question for the FDA: In the past, have broad approvals been given to drugs based on data from naïve populations but little data from experienced patients? (Yes.) Then I think it would be a departure from past practices to limit the indication.
Dr. Kumar: There seems to be no significant safety concerns with using this drug in the population where we would be most likely to see side effects. However, if state AIDS drug assistance programs (ADAP) won't pay for it due to a limited indication, that's a problem.
Dr. Stanley: First let me say that in Texas, once a drug is approved, we don't second-guess the physician; our state programs would pay. However, I continue to have concerns about making this broadly recommended at this time. There are unanswered questions: Does TDF need to be used along with a protease inhibitor in a regimen for naïve patients? The study in progress for naïve patients only looks at it combined with a non-nucleoside RT inhibitor (NNRTI). But it is clearly efficacious in salvage therapy and we should go ahead with it.
Dr. Wood: Obviously there is a need for this drug. If it is approved, it will be used. The people who need it and will use it right away will likely have high viral loads. So the sooner we can get efficacy information for them, the better.
Dr. Pomerantz: It's a tough call. HIV treatment is a dynamic field. It's not like it was five years ago. I don't think we need to have two data sets missing and still put it out there.
Dr. Wong: I want to put another face on what we're talking about. We shouldn't ask sponsors to prove that their drugs are the best in all situations. We are considering an approval that is substantially more restrictive than for any other antiretroviral drug. What they've shown is that their drug is safe and effective for adults with HIV infection. And we shouldn't try to split that any further.
Query to the FDA: How long will it take to get data from the trials in naïve patients?
FDA: When the new data is submitted, we will decide if it gets a 6- or a 10-month review. It may take about a year to receive the data, review it and make a decision.
Dr. Johnson: There is excellent data on the treatment-experienced. I think we're splitting hairs. There is no upper limit to salvage use, no matter what the viral load. With regard to the treatment-naïve population, we have no precedent for half a label.
Dr. Sun: On safety, this drug seems to be safe. On efficacy, it's a lot easier to extrapolate from experienced to naïve patients. There is biological plausibility. Naïve patients will be getting a greater number of active drugs than the patients on these studies did. History may not be helpful here; five years ago we didn't have protease inhibitor-experienced patients. We typically extrapolate data to multiple uses without specifying them on the label.
Dr. Hamilton: We have two potentially competing responsibilities: First, evaluate the data in a cut and dried way. A second responsibility, by implication, is what we recommend to the public. We should separate in our minds what we say. We probably agree about what the data says but should compromise with a cooperative sponsor about what the details of the approval will be.
Dr. Yogev: I have concerns about patients coinfected with hepatitis B virus using this drug if it results in HBV resistance. The label should warn about the risk of using only one active anti-HBV drug. I would also prefer to restrict the indication to people with lower viral loads.
Dr. Stanley: Our recommendations have weight, as Dr. Hamilton said. The drug is important for salvage, but people will use it as they wish. But what I'm comfortable with is to say what we saw in the data.
Dr. Gulick: Let me summarize: We're unanimous about its value for the experienced salvage population. But there are concerns about using it in patients with high baseline viral load and in treatment-naïve patients. This is a tough call. How much extrapolation are we comfortable with? In the guidelines for accelerated approval, a meaningful benefit over existing treatments must be shown, yet we don't have comparative data.
The members who would support a broad indication are more comfortable extrapolating from this data. They also cite precedence as a guideline.
Dr. Gulick then asked for an informal and nonbinding vote.
Wang: Full approval.
Shapiro: Want to see more PK data.
Kumar: Given everything we've seen it should not be restricted.
Yogev: I'm uncomfortable with full approval.
Stanley: This is a whole new kind of drug. We need to know more about the interactions.
Pomerantz: For naïve patients, I'd like to make the sponsor show us that the drug is effective one more time. There's no hurry to jump the gun with full approval. I would recommend TDF for use in those with prior experience.
Wood: Naïve patients usually have high viral load so I don't think it is reasonable to extrapolate.
Tebas: The biggest impact of our decision will be on marketing. If FDA approves the broader indication, then we will see the drug marketed to naïve patients.
Munk: I'd vote for full approval based on prior practices. I'm also concerned that a partial indication may deter some ADAP programs from adding the drug to their formularies.
Sun: Can we get full approval with caveats about the data that is lacking?
Gulick: I'm concerned about the risk/benefit ratio. I'd vote to restrict its indication while waiting for the pending data.
Drs. Leukert and Bone abstained.
On October 26, the FDA approved tenofovir without restrictions for use in adults with HIV.
For more information on the Web:
National AIDS Treatment Advocacy Project: www.natap.org
Treatment Action Group: www.treatmentactiongroup.org
British National AIDS Manual (aidsmap): www.aidsmap.com
amfAR HIV Treatment Directory: www.amfar.org/td
U.S. Food and Drug Administration: www.fda.gov
Gilead Sciences: www.gilead.com
Study 901: This was a Phase I placebo-controlled study of tenofovir monotherapy in 49 individuals for four weeks. Both treatment experienced and naïve patients participated. Oral doses of 75mg, 150mg, 300mg, and 600mg were evaluated. Based on this study the adult dose of 300mg/day was selected. These results were presented to the FDA.
Study 902: This was a Phase II trial in 189 treatment-experienced individuals. Three doses of TDF were compared to each other and to placebo by intensifying stable HAART regimens. Viral load differences were evaluated after 48 weeks. Individuals who added a daily 300mg dose of tenofovir had a mean viral load that was -0.62 logs lower than those who received placebo. These results were presented to the FDA.
Study 903: This is an ongoing 48 week Phase III trial in 600 treatment-naïve patients. The trial is comparing a regimen of efavirenz plus lamivudine (3TC) with either tenofovir or stavudine (d4T). Results will be available near the middle of 2002.
Study 907: This is an ongoing 48 week Phase III trial in 552 treatment-experienced patients. It is a treatment intensification study similar in design to Study 902. By week 24, the mean reduction of viral load for those receiving tenofovir was about -0.6 logs lower than for those who received placebo. These results were presented to the FDA.
Expanded Access Program: Individuals who have failed prior antiretroviral therapy, regardless of their CD4 cell count or viral load, may be eligible to receive tenofovir under this program. Enrollment is through a physicians' hotline. Call 1-800-GILEAD-5.
Back to the GMHC Treatment Issues October 2001 contents page.