DATE: September 6, 2001

TO: Advisory Committee Members and Guests

FROM: Tenofovir Review Team

THROUGH: Debra Birnkrant, M.D.

Acting Director

Division of Antiviral Drug Products

FDA’s analysis of the safety and efficacy data submitted in the NDA support the applicant’s results and conclusions. This NDA provides clear evidence of tenofovir’s antiviral activity (over 24 weeks) when added to a stable background regimen. In two pivotal studies statistically significant reductions in HIV RNA were observed in antiretroviral-experienced patients. These results are noteworthy given that the majority of approvals have been based primarily on studies in treatment-naïve patients. For many drugs, there is a need for more data characterizing their activity in treatment-experienced patients with prior exposure to the same drug class.

Although tenofovir’s risk benefit ratio was favorable among those patients included in the principal studies, the Division is convening this meeting to solicit the committee’s comments on four issues: 1) the breadth of the treatment indication, 2) the non-clinical and clinical assessment of bone toxicities, 3) the clinical implications and limitations of the resistance data and 4) the design of a second confirmatory study for traditional approval. Given the diversity of these issues, we have invited several committee guests, some with expertise in bone metabolism and others with expertise in HIV resistance. The text and appendices of this background document provide detailed information related to these issues.

The applicant is proposing a broad indication for the treatment of HIV infection based on the results of two principal trials. However, in both of these studies the antiviral activity of tenofovir when added to stable background therapy was assessed in antiretroviral experienced patients with relatively low plasma HIV RNA levels and high CD4 cell counts (mean baseline values were 3.4 log₁₀ copies/mL and 410 cells, respectively). In both studies treatment with tenofovir produced sustained mean decreases in HIV RNA levels approximating 0.5 log₁₀. As discussed in section III, CD4 changes in these studies were modest. However, small changes in CD4 cell counts may be expected in treatment experienced patients with relatively stable CD4 cell counts at baseline. Given that the study population in this NDA was quite select, we would like to hear the committee’s discussion regarding the most appropriate indication for tenofovir. Please refer to section III for a summary of the clinical efficacy data.

The second issue relates to the potential for tenofovir-induced bone toxicity in humans as observed in the animal toxicology data. Reductions in bone mineral density (BMD) were observed in three animal species following tenofovir treatment. Bone mineral loss occurred in juvenile rhesus monkeys following ten months of subcutaneous administration of tenofovir 30 mg/kg/day. Decreased bone mineral content and density was observed in rats at oral doses of 300 and 1000 mg/kg and in dogs at doses of 30 mg/kg given for 13 or 42 weeks. Mechanisms for changes in bone mineral density have not been defined but may have been a consequence of renal tubular reabsorption defects and/or a direct drug-related decrease in intestinal absorption of phosphate in some species. Review of the clinical trial data suggests no clinically significant changes in phosphate, calcium, PTH or BMD over time; however, PTH and BMD were only available for a small subset of patients. In addition, no obvious increases in fracture rates were observed over time in studies 902 and 907. Please refer to Section VB for further details on bone abnormalities.

After reviewing the entire animal toxicity data, along with human data, it is not likely that tenofovir related clinical fractures will occur over 48 weeks assuming the mechanism for bone abnormalities is mediated by renal phosphate wasting or decreases in intestinal absorption of phosphate. However, the exact mechanism of bone toxicity is unknown. Controlled safety data in more patients receiving tenofovir for longer periods of time are needed to address definitively the potential for bone toxicity in the clinical setting. This may be accomplished in an ongoing phase 3 study (#903) to be used for traditional approval (See section VI).

During the meeting we will be seeking your assessment of both nonclinical and clinical data with regard to bone abnormalities. We would like your general impression of the clinical implications of these data and your recommendations for additional preclinical and/or clinical studies to address the potential for bone toxicity. Additionally, we would like to hear your recommendations for long term monitoring for this potential toxicity

The third issue relates to the clinical virology data. Please refer to section IV for further details. This NDA included more clinical resistance data than any previous NDA for an antiretroviral drug. Prospective and exploratory analyses were conducted to evaluate HIV RNA response according to the presence of baseline mutations and baseline susceptibility. The genotypic and phenotypic data suggest potential for some cross resistance between tenofovir and abacavir, didanosine, stavudine, zalcitabine and zidovudine. Cross resistance was not observed between lamivudine and tenofovir. For some primary NRTI mutations or multi-drug resistance NRTI mutations, there were too few patients expressing these mutations at baseline to determine the clinical significance. It appears that the number and type of thymidine analogue mutations (TAMs, defined as M41L, D67N, K70R, L210W, T215Y/F and K219Q) present at baseline affect tenofovir efficacy.

The phenotypic data showed reduced response rates at > 4 fold reduced tenofovir susceptibility. Also decreased susceptibility to zidovudine at baseline affected virologic response. It does not appear that treatment with tenofovir altered the susceptibility of other NRTIs over 24 weeks compared to placebo.

We would like the committee’s comments on the clinical resistance analyses conducted during the development of tenofovir. We would also like the committee to comment on the type of resistance data warranting inclusion into package inserts, including tenofovir and other drugs in the pipeline.

Finally, we will be asking the committee to comment on the applicant’s proposed second confirmatory study for traditional approval and suggestions for alternative study designs or patient populations that should be studied in phase 4. A summary of the proposed confirmatory study can be found in section VI.

This NDA contains clinical data from four trials conducted with tenofovir tablets, including 2 pivotal studies (GS-98-902 and GS-99-907) and two supportive studies. Both pivotal studies were similar in design and evaluated the safety and efficacy of tenofovir versus placebo when added to a stable antiretroviral regimen in patients with prior nucleoside analogue experience. In addition, activity against nucleoside-resistant HIV was assessed in prospectively defined resistance subgroups.

Two supportive studies were also submitted. Study 901 investigated 5 doses of tenofovir vs. placebo in treatment naïve and experienced patients for 35 days. Study 908 was a compassionate use safety study in patients with limited therapeutic options. Summaries of these trials are provided in Table IIA.

The choice of the 300 mg dose was based on results from studies 901 and 902. In studies 901 and 902, decreases in HIV RNA were greater in the 300 mg dose group compared to 75 mg and 150 mg dose groups. No further HIV RNA reductions were observed for the 600 mg dose in study 901. Based on pharmacokinetic, safety and activity data from studies 901 and 902, Gilead chose the 300 mg dose for phase 3 studies and marketing.

Studies 902 and 907 were similar in design and enrolled patients with similar baseline characteristics. As shown in Table IIIA, both studies predominately enrolled Caucasian men with a mean age of 41 years and approximately 4 years of prior antiretroviral treatment. Study 902 enrolled patients with baseline HIV RNA levels between 400 and 100,000 copies/mL, whereas study 907 restricted baseline HIV RNA levels to values between 400 and 10,000 copies/mL. Consequently, the mean baseline HIV RNA level in study 902 was slightly higher than that of 907, whereas the mean baseline CD4 count was slightly higher in study 907.

The primary efficacy endpoint in the two pivotal studies was the time-weighted change in log₁₀ HIV RNA over 24 weeks (DAVG₂₄). Such analyses are useful for assessing antiviral activity in patients in whom plasma HIV-RNA reductions below the assay limit may not be frequently achieved. Therefore DAVG₂₄ was deemed an acceptable endpoint for evaluating virologic responses in treatment-experienced patients, such as those enrolled in studies 902 and 907. Secondary endpoints in these studies included the proportion of patients achieving HIV RNA levels below 400 and 50 copies/mL.

Table IIIB summarizes the efficacy results for studies 902 and 907. In both studies statistically significant differences of approximately 0.5-0.6 log₁₀ were observed for the primary efficacy endpoint (DAVG₂₄) favoring the tenofovir 300 mg over placebo. In study 902, patients receiving 300 mg had sustained HIV RNA reductions over 48 weeks that were larger than those receiving either 150 mg or 75 mg.

In study 902 there were no numerical differences between the 4 treatment groups with respect to the proportion of patients achieving HIV RNA levels < 400 or < 50 copies/mL. However, for patients with extensive prior antiretroviral therapy and/or higher baseline viral loads (> 5,000), the addition of one drug to a background regimen may not be expected to decrease HIV RNA to levels below assay limits. Of note, study 902 may not have been of sufficient size to detect treatment differences for the secondary endpoint of proportion below an assay limit.

In study 907, a much larger study than 902, a greater proportion of patients in the tenofovir group achieved HIV RNA < 400 copies/mL compared to placebo. In addition, baseline HIV RNA levels in this study were lower than that of study 902, making it more feasible to achieve HIV RNA levels below an assay limit.

In study 902 the protocol permitted changes to background therapy after week four. Changes to background regimens were discouraged in study 907; however, some patients did change their regimen. FDA conducted exploratory analyses to evaluate the impact of the addition of a new agent on the overall study results. Overall, treatment changes did not appear to affect the HIV RNA response. FDA concluded that the decline in HIV RNA in the tenofovir 300 mg groups observed in both studies was not attributed to the addition of another drug.

Subgroup analyses are presented below for study 907. The applicant conducted several subgroup analyses based on age, sex, and race. Statistically significant differences in HIV RNA decreases favoring tenofovir over placebo were observed for each of the age, race and sex subgroups.

Randomization for study 907 was stratified according to baseline HIV RNA (<5000, > 5000 copies/mL), baseline CD4 (<200, > 200 cells) and prior antiretroviral therapy (< 4 agents, > 4 agents). Treatment differences between the 300 mg group and placebo were significant favoring tenofovir for all randomization strata.

No significant interaction was seen for the subgroups based on baseline CD4 cell counts. The net benefit in reduction of viral load due to tenofovir was similar in patients who had either < 200 or > 200 cell/mm³ at baseline.

CD4 cell count changes are presented in table IIIC. In study 902, the mean DAVG₂₄ for CD4 cell counts was –10.5 cells/mm³ for the 300 mg group compared to –3.6 cells/mm³ for the placebo group. The difference between the 300 mg tenofovir dose compared to placebo was not statistically significant.

For study 907, there was a statistically significant difference in the DAVG₂₄ for CD4 favoring tenofovir. The mean DAVG₂₄ was +12 cells/mm³ in the tenofovir group compared to –10 cells/mm³ in the placebo group, resulting in a net treatment difference of approximately 20 cells/mm³.

Table IIIC: FDA Analyses: CD4 Cell Count Response

To further investigate the modest changes in CD4 cell counts observed in these studies, FDA conducted subgroup analyses to evaluate CD4 responses by baseline CD4 cell count based on pooled data. Patients with baseline CD4 cell counts < 200 had similar CD4 responses compared to patients with baseline CD4 counts > 200 (See Table IID). This finding is encouraging because CD4 increases in patients with lower cell counts are important for minimizing the risk of opportunistic infections.

It is also important to note that the study populations in studies 902 and 907 may not have been optimal for observing large increases in CD4 cell counts, given the fact that only one new drug was added to a stable background regimen. Similar results were reported for the abacavir CNA2003 study. CNA2003 was similar in design to tenofovir studies 902 and 907. The activity of abacavir versus placebo when added to stable antiretroviral regimens was evaluated in treatment experienced patients. No statistically significant differences for change in CD4 from baseline was noted between abacavir and placebo; however there was a numerical difference favoring the abacavir group (+30 vs +1 cells/mm³). Results from studies CNA2003, 902 and 907 suggest that small changes in CD4 cell counts may be expected in treatment experienced patients with relatively stable CD4 cell counts at baseline. For these patients, the addition of one new agent to a stable background antiretroviral regimen did not produce substantial increases in CD4 cell counts over time. Further evaluations of CD4 cell counts in studies with different designs are needed.

Studies 902 and 907 included analysis plans to prospectively evaluate HIV RNA response according to the presence of baseline mutations. Primary analysis plans specified the evaluation of thymidine analogue mutations (TAMs) and the M184V mutation. The six mutations commonly defined as TAMs are M41L, D67N, K70R, L210W, T215Y/F and K219Q. Secondary analyses evaluated virologic response by other primary NRTI, NNRTI and PI associated mutations.

Table IVA summarizes FDA analyses of virologic response for the protocol specified baseline mutation groups. These analyses incorporated pooled data from studies 902 and 907. As shown in Table IVA, decreases in HIV RNA were observed in tenofovir treated patients with baseline TAMs or M184V. The data suggest that tenofovir’s activity was not affected by the M184V mutation, indicating absence of cross resistance between tenofovir and lamivudine

The applicant also performed analyses to determine if the presence of the M184V mutation at baseline resulted in better responses to tenofovir compared to those without this mutation. These results did not show "enhanced" activity when the M184V mutation was present despite in vitro evidence of increased tenofovir susceptibility in the presence of this mutation. Tenofovir treated patients with M184V and TAMs had slightly greater HIV RNA decreases compared to patients with TAMS and no M184V (DAVG₂₄ –0.52 vs –0.45 log₁₀). However, when considering the effect of placebo, the net treatment effects in these groups were similar (-0.60 vs –0.58 log₁₀).

Table IVA RNA Responses by Baseline Resistance Mutations (ITT)^A