Twelve years in coming

On February 27th the FDA's Antiviral Drugs Advisory Committee met to consider an application for approval of Hoffmann-La Roche's valganciclovir for the treatment of CMV retinitis in people with AIDS. Michael Marco prepared this position paper on behalf of TAG. A decision on the application is expected later this month.

Cytomegalovirus (CMV) retinitis, a viral disease affecting the eyes and causing loss of vision, was once a rare disease occurring only among individuals with primary immunodeficiency syndromes or autoimmune disorders, organ transplant recipients and immunosuppressed cancer chemotherapy patients. In people with AIDS, CMV retinitis is by far the most common manifestation, accounting for between 77 and 90% of all CMV end-organ disease. In bone marrow transplant recipients, however, there is a much greater incidence of CMV pneumonitis than retinitis.

From the 1980s to 1996, the incidence of CMV end-organ disease among people with AIDS was estimated to range between 10 and 40%. In the new era of highly active antiretroviral therapy (HAART), the incidence rate of CMV—for those who have access to HAART—is down well-below 5%.

CMV end-organ disease occurs late in the course of AIDS and is associated with extremely low CD4 counts. The average CD4 count in persons with newly diagnosed CMV retinitis is below 30 cells/mm3. The main symptoms of CMV retinitis include "floaters," blurred vision, missing portions of vision and flashing light/sparks. Even subtle changes, such as a minor loss of peripheral vision, can indicate the development of CMV retinitis.

Treatments for CMV retinitis are palliative rather than curative. Resistance to the FDA approved antivirals for CMV retinitis, intravenous ganciclovir, foscarnet and cidofovir, is common and no oral drug indicated for CMV induction therapy exists.

In heavily immunosuppressed individuals with CMV retinitis, lifelong maintenance therapy is recommended in order to the hold lesions at a quiescent state. Recently published USPHS/IDSA guidelines indicate that "discontinuation of [secondary CMV] prophylaxis may be considered in patients with a sustained (e.g., greater than 3-6 month) increase in CD4+ T-lymphocyte count to greater than 100-150 cells/mm3 on HAART."

Valganciclovir's pivotal CMV study
Valganciclovir is the valine ester of intravenous ganciclovir. Were the year 1997, the FDA would approve valganciclovir solely on pharmacokinetic data demonstrating that a 900 mg daily (oral) dose of valganciclovir produced comparable drug exposure (area under the curve/AUC) to that of the standard daily intravenous ganciclovir dose of 5mg/kg. Instead, the agency required the sponsor to conduct a randomized controlled clinical trial demonstrating equivalence between valganciclovir and IV ganciclovir for induction therapy of CMV retinitis.

Roche WV 15376 was the sponsor's registrational study which evenly randomized 160 individuals with CMV retinitis to receive valganciclovir (900 mg, orally, twice daily for 3 weeks followed by 900 mg daily for 1 week) or IV ganciclovir (5 mg/kg twice daily for 3 weeks followed by 5 mg/kg daily for 1 week). After the four-week induction phase comparison, all study volunteers received maintenance therapy with open-label valganciclovir (900 mg daily). The primary endpoint was CMV retinitis progression within 4 weeks of initiating treatment using fundus photographs. Progression was defined as movement >750 µm (along a >750 µm front) or new retinitis lesions >750 µm diameter. Statistically, the study was not a traditional head-to-head comparison but a non-inferiority study powered to prove that valganciclovir was not 10% worse than IV ganciclovir. Both arms were evenly balanced for baseline demographics and disease status: ~90% were men; ~70% were on HAART; median CD4 cell count was ~23 cells; median HIV RNA copies/mL was ~4,000; 24% had zone 1 retinitis; and 25% had bilateral retinitis.

Of 146 individuals who completed the 4-week induction phase, 7 of 73 (~10%) individuals in the valganciclovir arm progressed compared to 7 of 73 (~10%) in the IV ganciclovir arm (95% CI, -0.097, 0.100). Sixty-four and 63 individuals had no documented photographic progression in the valganciclovir and IV ganciclovir arms, respectively.

Safety
In Roche WV 15376, safety data were available on 158 individuals. Adverse events were similar for valganciclovir versus IV ganciclovir diarrhea (16% vs. 10%); pyrexia (13% vs. 11%); nausea (8% vs. 14%); vomiting (11% vs. 6%). As expected, the IV ganciclovir had significantly more catheter-related infection, 11% vs. 2%. There were no significant differences in hematological abnormalities between the valganciclovir and IV ganciclovir arms: absolute neutrophil count <750 cells/mL (21% vs. 19%); hemoglobin 6.5 to <8.0 g/dL (5% vs. 3%); platelets 25,000 to <50,000 (0% vs. 1%).

The sponsor also carried out a 200 person open-label randomized safety study, Roche WV 15705. No significant differences in adverse events or hematological toxicities were noted between the valganciclovir and IV ganciclovir arms.

Discussion
Pharmacokinetically, valganciclovir has shown to provide systemic exposure (AUC) comparable to IV ganciclovir. The sponsor fulfilled the FDA Antiviral Drugs Division's approval requirements by carrying out a randomized controlled Phase III study of valganciclovir which documented that valganciclovir was non-inferior to ganciclovir for induction therapy of AIDS-related CMV retinitis. The safety profile of valganciclovir is almost identical to well-characterized IV ganciclovir. Its most serious toxicity is neutropenia in which ~20% experience a grade 3 or 4 event by week 4.

The sponsor should be commended for studying valganciclovir against the gold standard IV ganciclovir. Other approved CMV antivirals were merely approved using the quick and easy (some say lazy) immediate vs. deferred trial design. Valganciclovir is the first drug that has been tested against IV ganciclovir for induction since SOCA tested foscarnet vs. IV ganciclovir in 1990.

It has been 12 years since IV ganciclovir was approved by the FDA in 1989. It has taken far too long for industry to bring an orally formulated anti-CMV therapy to market. We needed one years ago. The incidence of AIDS-related CMV retinitis has dramatically declined to less than 5%; nevertheless, valganciclovir will certainly change the clinical management of CMV therapy. With easier administration, adherence to valganciclovir should be better than IV ganciclovir and result in less resistance.

Valganciclovir may hold great promise as prophylaxis for CMV retinitis. Hoffmann-La Roche should continue its positive, collegial relationship with the Adult AIDS Clinical Trials Group (AACTG) and support it in accruing and completing AACTG a3050, "Valganciclovir Pre-emptive [Prophylactic] Therapy for Cytomegalovirus." If international sites in Spain, Australia, or Canada are needed to help with accrual, Hoffmann-La Roche must help in funding these units.

Lastly, Hoffmann-La Roche should cease with its erroneous patient-directed ad campaign warning individuals not to discontinue their CMV maintenance therapy with oral ganciclovir even though they are on HAART and fit the discontinuation guidelines outlined by the USPHS/IDSA. ¤

Bemoaning virtual monotherapy

In January the Antiviral Drugs Advisory committee of the FDA met to discuss the trial design for salvage therapy in HIV infection. TAG's Yvette Delph was there and prepared this report.

The meeting, chaired by Roy Gulick of Cornell University, was convened to focus on clinical trial design issues for HIV-infected individuals who have limited antiretroviral therapeutic options. Heidi Jolson, the FDA's outgoing Director of the Division of Antiviral Drug Products, noted that the FDA's interest was in registrational rather than treatment strategy trials and that for the purpose of the meeting, the "highly treatment experienced (HTE) population" would refer to individuals who lacked or had lost response to two or more HAART regimens and who had experience with one or more members of each class of drugs.

Martin Schechter of the Canadian Clinical Trials Network gave an overview of trial design options in adults. He noted that as individuals moved from treatment naïve to salvage, population heterogeneity increased, but that heterogeneity only matters if the variables are strongly prognostic. He questioned whether blinding in salvage trials introduced more bias than it prevented, as it could wipe out the adherence advantage of an arm with lower pill burden. He rued the fact that factorial designs are woefully underutilized in the medical arena.

Carlton Hogan of the Coalition for Salvage Therapy (CST) gave the patients' perspective. The CST was uncomfortable with intensification designs that simply add one new agent to an already failing regimen ("virtual monotherapy") and he stressed that it was vitally important that the FDA proactively offer clear, unambiguous guidance related to the use of more than one investigational agent. The CST advocated strongly that background therapy be optimized individually based on genotypic and phenotypic resistance testing (optimized background regimen (OBR)) and recommended a modified factorial design on top of OBR. With three new agents (A, B, C) this would be:

• A+B+OBR
• A+C+OBR
• B+C+OBR
• A+B+C+OBR

Such a design would improve chances of virologic control in each arm, reduce development of resistance to new agents as well as OBR drugs, allow the effects of individual agents to be discerned, and may be attractive to potential participants. However, there is no true control arm; it may be difficult to ascribe adverse events to one particular agent with certainty; and, if certain combinations fail to achieve virologic control, study participants would be exposed to toxicity without commensurate benefit.

Dr. Jolson pointed out that in 1999 the FDA had written to sponsors indicating that more than one investigational agent could be used and that the Agency was extremely supportive of factorial designs. The FDA had concerns about intensification trials in salvage populations.

There was general support among the advisory panel for a modified factorial design in HTE individuals. It was felt that it would be unethical to enroll "deep salvage" individuals in trials—they should be granted expanded or compassionate access to new agents.

Michael Marco presented TAG's position.

Daniel Vittecoq of the European Agency for the Evaluation of Medicinal Products (EMEA) summarized the new approach adopted by the EMEA to the registration of new antiretroviral drugs with particularly promising resistance or pharmacokinetic profiles for use in HTE individuals.

Phase II studies would establish the drug dosage in treatment naïve persons or HIV-negative volunteers. Phase III trials in HTE individuals should be superiority designs—either substitution or intensification trials. There would be a 2-4 week assessment of efficacy and a 12-16 week assessment of durability and safety. Primary endpoints would be the proportion of participants achieving "undetectable" viral loads and = 0.5 log10 difference between arms.

Jim Rooney of the Intercompany Collaboration (ICC) gave the agency's perspective. He noted some of the difficulties of trials involving more than one agent: isolating the benefit of a single agent, the limited availability of new agents, and unexpected interactions and toxicities. He suggested that placebos should not be used often in the salvage setting as they would greatly increase the pill burden, (thereby decreasing adherence), and advocated using change in viral load as an endpoint. He suggested a trial design with short-term monotherapy followed by longer-term combination therapy:

Stage 1: New agent "A" vs. placebo or no treatment for 1-4 weeks

Stage 2 (for both arms): A + optimized background regimen (which could include other experimental agents) for 12-16 weeks

Stage 1 would provide evidence of antiviral activity, allow for resistance monitoring and provide short-term safety and tolerability data.

In the discussion that followed, members of the panel suggested that trial designs include an early escape mechanism for non-responders. Endpoints suggested include decrease in viral load, increase in CD4 count, proportion with plasma viral load below detectability and the slope of plasma viral load decline during the first few weeks on monotherapy or initial intensification of a failing regimen. Community members stressed the need for a system for monitoring the safety and toxicity of agents post-registration.

Victor DeGruttola of the Harvard School of Public Health further addressed the issue of the modified factorial design. Although it does not have the full efficiency of the true factorial design, it allows comparison of several agents (A vs. B, B vs. C and A vs. C). Some of the ability to discern all of the interactions among agents is lost, but it provides the best power to look at interactions between drugs.

Michael Saag of the University of Alabama at Birmingham suggested that study participants be stratified prior to randomization according to the number of treatment options available based on resistance testing. Those with >3 options would be randomized to receive:

• OBR+placebo
• OBR+A
• OBR+B
• OBR+C

Those with <3 options would be randomized to receive:

• OBR+A+B
• OBR+B+C
• OBR+A+C

Dr. DeGruttola presented on the choice of endpoints for salvage studies. He questioned whether study withdrawal should be counted as failure or as censored, noting that each analysis is likely biased. He recommended doing both analyses as well as more sophisticated analyses.

Dr. Gulick presented data on the low response rate in HTE individuals. He noted that predictors of response include adherence, baseline plasma viral load, baseline CD4 count, resistance profile, number of active drugs, and drug plasma levels.

In the discussion that followed, it was suggested that baseline plasma viral load should be the plasma viral load off therapy. Endpoints should include: change in plasma viral load (0.4-0.5 log10 reduction in plasma viral load over 8 weeks); resistance; viral fitness; toxicity; adherence and quality of life. The proportion of study participants with undetectable plasma viral load or plasma viral load reduction of 0.4-0.5 log10 over 8 weeks should be correlated with clinical outcome. Predictors of success on therapy should be studied.

Finally, the role of shorter-term trials (e.g., 16 weeks) in assessing the safety and efficacy of new antiretrovirals in HTE individuals was discussed. Some clinicians suggested that "deep salvage" individuals need urgent access to new agents and so shorter-term registrational trials in HTE individuals may be needed. The community members present were unanimous in expressing concern about approving agents with fewer than 24-weeks of data for conditional approval and fewer than 48-weeks for full approval. They noted that there is great need for adequate safety data, especially Phase IV post-registration, and that industry had not demonstrated a willingness to do systematic long-term safety monitoring. Before the FDA grants approval based on shorter-term trials, sponsors should demonstrate that they are committed to looking at long-term toxicities—and systems (cohort studies, databases, etc.) should be set up to monitor drug adverse events. Because FDA approval permits off-label use of drugs, earlier approval may be dangerous. Compassionate or expanded use programs should be set up as soon as the investigational agents go into larger Phase III trials so that "deep salvage" individuals who need these agents to construct a viable regimen may have access to them.

In concluding, Dr. Jolson said that safety data beyond 8 weeks is needed and encouraged sponsors to follow up study participants from early trials for as long as possible. ¤