Secondary Logo

Journal Logo

Clinical Science

Patterns of Resistance Emerging in HIV-1 From Antiretroviral-Experienced Patients Undergoing Intensification Therapy With Tenofovir Disoproxil Fumarate

McColl, Damian J PHD; Margot, Nicolas A MA; Wulfsohn, Michael MD, PHD; Coakley, Dion F PHARMD; Cheng, Andrew K MD, PHD; Miller, Michael D PHD

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: November 1st, 2004 - Volume 37 - Issue 3 - p 1340-1350


Nucleoside reverse transcriptase inhibitors (NRTIs) are important drugs in the treatment of HIV-1 infection and have reduced the mortality associated with this disease.1 When combined with a non-nucleoside reverse transcriptase inhibitor (NNRTI) or a protease inhibitor (PI), NRTIs can suppress HIV-1 replication to undetectable levels. Many patients achieve sustained suppression of viral replication but a significant fraction of patients experience failure of their antiretroviral therapy (ART). Reasons for failure include poor patient adherence, drug-associated toxicities, and the development of resistance mutations due to incomplete suppression of viral replication.2 Cross-resistance of HIV-1 to the agents within a class of antiretroviral drugs may limit therapeutic options for treatment-experienced patients.3 Therefore, an understanding of HIV-1 resistance development has become an integral part of HIV-1 treatment design.

Tenofovir is an acyclic nucleotide analogue that is analogous to the monophosphate form of the other NRTIs. Upon entering cells, tenofovir requires 2 phosphorylation steps to reach the active diphosphate form, whereas 3 phosphorylation steps are required by the other NRTIs.4 Tenofovir disoproxil fumarate (tenofovir DF, Viread; Gilead Sciences, Foster City, CA) is an oral prodrug of tenofovir that is rapidly converted to tenofovir upon absorption. Tenofovir has in vitro activity against HIV-1,5–7 HIV-2,8 and human hepatitis B virus.9–11 Tenofovir selects for the K65R mutation in HIV-1 RT in vitro,12 and this mutation results in approximately 3-fold decreased susceptibility to tenofovir. The K65R mutation has also been observed to be selected in patients on antiviral therapy with tenofovir DF.13 The K65R mutation can also be selected by abacavir both in vitro14 and in vivo,15 by didanosine in vivo,16 by stavudine in vitro,17 and by zalcitabine in vitro and in vivo.18,19 However, analysis of a large panel of clinically derived viral isolates has shown that K65R is infrequent among treatment-experienced patients.20 Recent reports have indicated that the incidence of K65R has increased from approximately 1% in 1998 to 4% in 2003,21–23 which may reflect increased use of abacavir and tenofovir DF.

A phase 2 treatment intensification study in 186 treatment-experienced patients (GS-98-902) demonstrated the potency and favorable safety profile of tenofovir DF 300 mg dosed once daily.24 In addition, there was a low rate of emergence (3%) of the K65R mutation through 96 weeks of tenofovir DF therapy.13,25 A phase 3 intensification trial in 552 treatment-experienced patients (GS-99-907) has confirmed the antiviral efficacy and safety of tenofovir DF 300 mg in this patient population.26 Here we describe the patterns of resistance mutations emerging after 48 weeks of tenofovir DF therapy in GS-99-907 and the results of the phenotypic resistance analyses. Treatment of these patients with partially suppressed disease with tenofovir DF therapy for 48 weeks was associated with a significant and durable antiviral response and reduced the incidence of new resistance mutations associated with all classes of antiretrovirals. Furthermore there was a low incidence of genotypic and phenotypic resistance to tenofovir arising during 48 weeks of tenofovir DF therapy. The K65R mutation emerged in only 3% of tenofovir DF–treated patients and was not associated with viral load rebound. The lack of viral load rebound associated with K65R may be due to the low-level phenotypic changes and reduced viral replication capacity27,28 that result from this mutation. The K65R mutation arose only in patients who had no evidence of TAMs at baseline. Thus, these 2 mutational patterns may be incompatible in HIV-1 from patients undergoing ART.


Study Population and Design of Clinical Trial GS-99-907

GS-99-907 (study 907) was a phase 3, randomized, double-blind, placebo-controlled, multicenter study of the safety and efficacy of tenofovir DF administered orally to HIV-1–infected patients with plasma HIV-1 RNA levels ≥400 copies/mL and ≤10,000 copies/mL.26 Patients on stable ART for ≥8 weeks (n = 552) were randomly assigned in a 2:1 ratio to add either tenofovir DF 300 mg once daily or placebo to their existing regimen in a double-blind manner. At 24 weeks after randomization, patients initially assigned to placebo treatment received open-label tenofovir DF for the remainder of the 48-week study.

Half of the patients (n = 274) were randomly assigned to a blinded virology genotyping substudy at baseline. Approximately half of the patients enrolled in the genotyping substudy (n = 137) were randomly assigned to a phenotyping substudy.

Determination of Patient Plasma HIV-1 RNA Levels

HIV-1 RNA concentrations in patient plasma samples were determined in a central laboratory using the Ultrasensitive HIV-1 Monitor test (Roche Diagnostics Corp.) with a lower limit of detection of plasma HIV-1 RNA of 50 copies/mL. Unless otherwise indicated, all pooled HIV-1 RNA results are expressed as mean values (arithmetic mean) in units of log10 copies/mL.

Genotypic and Resistance Phenotypic Analyses

HIV-1 genotypic analyses included amino acids 1–400 of HIV-1 RT and all of HIV-1 protease (Virco Laboratories, Mechelen, Belgium). All baseline, week 24, week 48, or early termination, plasma HIV-1 samples with >50 copies/mL of HIV-1 RNA were analyzed in a blinded fashion. NAMs were defined according to the definitions of the Resistance Collaborative Group and included residues 41, 62, 65, 67, 69, 70, 74, 75, 77, 115, 116, 151, 184, 210, 215, and 219 of RT.29 The S68G mutation was included in the analysis as it appears to be associated with K65R.19 NNRTI-associated mutations were A98G, L100I, K103N, V106A, V108I, Y181C/I, Y188C/H/L, G190A/E/S, P225H, and P236L. Primary PI-associated mutations were D30N, V32I, G48V, I50V, V82A/C/S/T, I84V, or L90M in protease. HIV-1 susceptibility data for tenofovir and all other approved antiretroviral nucleoside analogues were generated using the Antivirogram recombinant virus assay (Virco Laboratories, Mechelen, Belgium).30 A polymerase chain reaction product could not be obtained from 21 patients at baseline (14 tenofovir DF, 7 placebo), resulting in a total of 253 patients for the genotyping substudy. During the placebo-controlled phase, matching baseline and postbaseline genotypic data (week 24 or early termination) were obtained from 163 of 253 patients in the genotyping substudy. The remaining patients had insufficient HIV-1 RNA to genotype (n = 89) or no postbaseline sample (n = 1). Approximately 50% of patients with insufficient HIV-1 RNA to genotype had <50 copies/mL of plasma HIV-1.26 Proportionally fewer patients in the tenofovir DF treatment arm had week 24 genotypic results due to the greater number of tenofovir DF–treated patients having insufficient HIV-1 RNA for analysis (76 tenofovir DF, 14 placebo).

Statistical Analyses

All HIV-1 RNA statistical analyses were performed using SAS statistical software (SAS Institute, Cary, NC). All P values provided for changes in HIV-1 RNA were from 2-sided tests (0.05 was used as the statistical significance level). Patients without baseline genotypic data were excluded from the analysis, creating a virology intent-to-treat population of 253 patients.

The protocol-defined primary efficacy endpoint was the time-weighted average change in log10 HIV-1 RNA copies/mL, from baseline to week 24 (DAVG24). The weights used to calculate DAVG were distributed uniformly over time so that each unit of time contributed equally to the calculation of DAVG. DAVG was calculated for the intention-to-treat population that included all patients who received at least 1 dose of medication. Potential viral load rebound was determined as follows: an HIV-1 RNA nadir was calculated as the lowest of the mean HIV-1 RNA values for each set of 2 consecutive time points within the first 12 weeks of treatment. The baseline HIV-1 RNA value was calculated as the mean of the baseline and the prebaseline HIV-1 RNA values. A patient was classified as a responder if the difference between his or her baseline and nadir HIV-1 RNA values was ≥0.5 log10 copies/mL or as a nonresponder if the difference was <0.5 log10 copies/mL. Among responder patients, if there were 2 consecutive HIV-1 RNA values ≥0.5 log10 copies/mL above the nadir during the 48-week analysis period, the patient’s response was classified as a rebound.


Baseline Patient Characteristics and HIV-1 Genotypes

A total of 552 patients were enrolled of whom 550 received study drug in the trial. The mean CD4 cell count at baseline was 427 ± 213.7 cells/mm3 and the mean HIV-1 RNA viral load at baseline was 3.36 ± 0.51 log10 copies/mL. This patient population had a mean duration of prior ART of 5.4 ± 2.9 years. Baseline HIV-1 genotype data were obtained from 253 of the 274 patients assigned to the genotyping substudy. Reflective of their extensive treatment experience, 94% of patients in the virology substudy had at least 1 NAM at baseline, 58% had a primary PI-associated mutation, and 48% had a primary NNRTI-associated resistance mutation (Fig. 1). Most patients (69%) had HIV-1 with typical TAMs at RT codons 41, 67, 70, 210, 215, or 219 (mean of 2.8 mutations); 68% had HIV-1 with the lamivudine/abacavir -associated M184V/I mutations and 45% had combinations of both TAMs and M184V/I. The prevalence of baseline resistance mutations was similar across both the tenofovir DF and placebo arms (data not shown). Only 2% of patients entered the trial with the K65R RT mutation, consistent with its reported low frequency.

. Baseline HIV-1 genotypic analysis of patients in the virology substudy of study 907 (n = 253). The frequency and type of mutations in HIV-1 present at baseline from patients in the virology substudy are shown. For those mutations present at ≤2% at baseline, only K65R is shown. Additional mutations present at ≤2% at baseline were A62V (2%, n = 6), V75T/I (2%, n = 5), Q151M (1.5%, n = 4), and T69S insertions (1%, n = 2).

At baseline, 99% percent of patients were taking ≥1 NRTI, with lamivudine and stavudine being the most commonly used (68 and 59% of patients, respectively) (Fig. 2). Additionally, 58% of patients were taking at ≥1 PI and 39% of patients were taking an NNRTI. There were no significant differences between the treatment groups in baseline ART usage (data not shown). All patients ultimately had exposure to tenofovir DF during the open-label phase of the trial.

. Antiretroviral drugs utilized by patients in the virology substudy of study 907 (n = 253). Tenofovir DF is shown as 100% to include placebo patients who rolled over onto tenofovir DF therapy at week 24 at the commencement of the open-label phase of the study. The virology substudy included 169 patients from the tenofovir DF arm (48 weeks of exposure to tenofovir DF) and 84 patients from the placebo arm (24 weeks of exposure to tenofovir DF).

Treatment Response to Tenofovir DF

Despite extensive RT resistance mutations at baseline, the 274 patients in the genotyping substudy that added tenofovir DF to their existing regimen demonstrated a statistically significant decline in their mean plasma HIV-1 RNA by week 24 of −0.59 ± 0.61 log10 DAVG24 (vs. −0.03 ± 0.39 log10 DAVG24 in the placebo arm, P < 0.0001). The decline in plasma HIV-1 RNA in the patients receiving tenofovir DF was durable through week 48 (−0.56 ± 0.61 log10 DAVG48). Placebo patients adding tenofovir DF to their existing regimen at week 24 showed a similar response at week 48 (−0.70 ± 0.72 log10 DAVG48). By week 24, significantly more patients (P < 0.0001) achieved either <400 copies/mL or <50 copies/mL of plasma HIV-1 RNA on tenofovir DF therapy vs. placebo.26 An extensive analysis of patient responses by baseline genotype has been performed and is reported elsewhere.31

Development of NRTI-Associated Mutations

During the 24-week placebo-controlled phase, 20 patients in the placebo arm developed at least 1 new NAM in RT, which defined the background rate (20/84 patients = 24%) of NAM development in this study population (Table 1). There was a trend toward reduced development of new NAMs in the tenofovir DF treatment arm as compared with the placebo arm (16 vs. 24%, respectively, P = 0.17), suggesting that treatment with tenofovir DF was delaying the development of new NAMs. Patients who developed new NAMs in the tenofovir DF treatment group still showed statistically significant decreases in their HIV-1 RNA at week 24 (−0.51 ± 0.58 log10 DAVG24, n = 27) compared with placebo patients, who developed NAMs (+0.08 ± 0.36 log10 DAVG24, n = 20, P = 0.0003). Using the secondary endpoint of absolute change in HIV-1 RNA from baseline, tenofovir DF–treated patients who developed new NAMs during the placebo-controlled phase achieved an HIV-1 RNA decrease of −0.41 ± 0.57 log10 copies/mL by week 24 (Table 2) compared with +0.15 ± 0.48 log10 copies/mL in the placebo group. Thus, these patients continued to obtain antiviral benefit from tenofovir DF despite the development of these mutations. During the open-label phase, a further 23% of patients (58/253) developed new NAMs, similar to that observed during the placebo-controlled phase. The HIV-1 RNA response in these patients was −0.56 ± 0.57 log10 copies/mL (DAVG48), similar to the response in patients who did not develop NAMs (−0.61 ± 0.65 log10 DAVG48).

Table 1
Table 1:
Development of NAMs* by Week 48
Table 2
Table 2:
HIV-1 RNA Responses in Patients Developing Mutations

In both the placebo and tenofovir DF arms, the majority of patients developing new mutations during the placebo-controlled phase (31/47) developed new TAMs while taking other NRTIs (zidovudine, stavudine, abacavir, or didanosine) concomitantly. Most of these patients (22/31) added additional TAMs onto a background of preexisting TAMs. There were no significant differences in the development of any of the TAMs between patients in the tenofovir DF and placebo-controlled arms of the study (Table 1). Development of the D67N mutation occurred more frequently in the tenofovir DF arm during the first 24 weeks, but this did not achieve statistical significance (P = 0.28, Fisher’s exact test). Among the 7 patients whose HIV-1 developed a D67N mutation, there was continued viral load suppression (−0.94 ± 0.68 log10 DAVG24). Thus the background therapies probably caused the development of this mutation.

During the open-label phase, a similar percentage of patients (19%) developed ≥1 new TAM. When the placebo-controlled and open-label phases were compared, there were no significant differences between the 2 groups in the frequency of development of any of the individual TAMs. The use of antiretroviral agents known to select TAMs and their similar distribution between the 2 arms is consistent with the background antiretroviral agents being primarily responsible for their development.

During the placebo-controlled phase, 15 patients developed NAMs at codons that were neither TAMs nor K65R (positions 62, 69, 74, 75, 115, 151, or 184 of RT). Thirteen additional patients developed mutations at these positions during the open-label phase. The emergence of these substitutions correlated with the use of NRTIs (didanosine, stavudine, abacavir, zidovudine, and lamivudine) that have been shown to select these mutations. The most frequent change was observed at residue 74, where 7 of 253 patients (3%) had an amino acid change (L74V/I) during the open-label phase. However, this pattern of mutation occurred at a higher rate (6%) in the placebo group during the placebo-controlled phase. For all other codons listed above, a new mutation developed in <1% of tenofovir DF–treated patients, and their distribution between the tenofovir DF and placebo-control arms was similar.

Development of the K65R Mutation

During the placebo-controlled phase, the K65R mutation developed in 5 patients, all of whom were in the tenofovir DF treatment arm (Table 1); a further 3 patients developed this mutation during the open-label phase. Thus, K65R developed in 8/253 patients (3%) during the study and all these patients were on tenofovir DF therapy when the mutation developed. The median time to emergence of K65R was 24 weeks and it often appeared initially as a mixture of mutant and wild-type sequences (Table 3). Four of these 8 patients were also taking abacavir or didanosine, both of which can select the K65R mutation, and these drugs along with tenofovir DF may have contributed to its selection.14–19 The other 4 patients were receiving lamivudine concomitantly along with zidovudine (n = 2) or stavudine (n = 2) (Table 3). Stavudine has been reported to select the K65R mutation in vitro.17

Table 3
Table 3:
Treatment Regimens and Responses Status of Patients Who Developed K65R

The virologic responses to tenofovir DF therapy in the 8 patients who developed K65R were highly variable, with a mean DAVG48 of −0.28 ± 0.75 log10 copies/mL (range −1.15 to +0.86 log10 copies/mL). Three of the 8 patients were classified as “responders,” 3 patients were classified as “non-responders,” and 2 patients were classified as “rebounders” (Table 3). Viral load rebound in these latter patients (patients 3 and 7) could not be attributed specifically to development of K65R alone as concurrent development of NNRTI resistance was also observed in these 2 patients (Table 3). Patient 3 developed the nevirapine-associated mutation V106A while taking nevirapine concomitantly. Patient 7 developed noncanonical mutations at codons associated with NNRTI resistance (V106G and G190V),32 and resistance to the efavirenz in the patient’s regimen was observed.

None of the 8 patients who developed the K65R mutation had detectable TAMs present at baseline. The presence of baseline TAMs was significantly associated with the lack of emergence of the K65R mutation (P < 0.0001). This was striking given that nearly 70% of all patients in the study had ≥1 TAM at baseline (Fig. 1). Furthermore, only 1 of these 8 patients who developed the K65R mutation subsequently developed any TAMs by week 48 (patient 8). This suggests that the K65R mutation and the TAMs form 2 evolutionary pathways that may be mutually exclusive or there is a significant disadvantage to the virus to select both TAMs and K65R in patients undergoing therapy with NRTIs. This observation prompted an analysis of the frequencies of TAM development among patients with and without detectable TAMs at baseline. Through week 48, there were no significant differences in the frequencies of development of new TAMs in the HIV-1 of patients with or without TAMs at baseline (48/175, 27%, vs. 23/78, 29%, respectively). Similarly, when comparing the tenofovir DF arm vs. placebo arm, there were no significant differences in the frequency of emergence of TAMs at week 24 in patients with TAMs at baseline vs. those without TAMs at baseline (data not shown). However, new TAMs emerged approximately 3 times as frequently as K65R (23/78, 29% vs. 8/78, 10%, respectively) in the group of patients whose HIV-1 had no TAMs present at baseline.

In addition to K65R, other mutations were observed to develop in these 8 patients (Table 3). Patient 2 developed S68G and patient 3 had this mutation at baseline. S68G has been reported in association with K65R in virologic failure patients from a study investigating an abacavir, didanosine, and stavudine triple-nucleoside regimen.33 A site-directed mutant virus carrying both mutations has demonstrated no significant changes in susceptibility to any of the NRTIs tested, including tenofovir.34 S68G may partially compensate for the reduction in viral replication capacity that results from the K65R mutation.27,28,35 Patient 6, who was taking abacavir along with tenofovir DF, developed the abacavir-associated mutation Y115F,36 but no significant change in phenotypic susceptibility to tenofovir DF was observed (data not shown). Patient 7 developed a D67G mutation that has been observed in viruses containing T69S insertions.37 The phenotypic susceptibility of this patient’s virus to tenofovir was unchanged relative to baseline and was equivalent to wild-type (data not shown). Patient 8 entered the trial with a Q151M multidrug-resistant genotype plus the M184V mutation and NNRTI-resistance mutations (V108I, G190A). This patient subsequently developed the M41L mutation in addition to K65R. Whether these 2 mutations were present on the same virus or there was a mixed viral species present in this patient is unclear. A database of nearly 300 patient-derived viruses containing the K65R mutation has shown that the TAMs very rarely occur in combination with K65R.23

Development of Potential Novel Tenofovir Resistance Mutations

The K65R mutation appeared to be the single NAM that was clearly associated with tenofovir DF therapy. All other RT residues not previously associated with NRTI or NNRTI resistance were analyzed to detect potential novel resistance mutations selected by tenofovir DF. Substitutions that developed in RT in ≥3 patients (>1%) by week 48 were identified (Table 4). These are not established polymorphic residues in RT; however, 7 of these mutations were also present at a similar low frequency in the baseline samples and in the Stanford ART-experienced patient database. With the exception of T139K, all developed at higher frequency in the placebo group and are unlikely to be novel tenofovir resistance mutations. The 3 patients who developed T139K while on tenofovir DF responded well to therapy (−0.82 ± 0.49 log10 mean DAVG48) with no evidence of viral load rebound. Thus, no novel resistance mutations to tenofovir DF were identified in this study.

Table 4
Table 4:
Development of Other RT Mutations by Week 48

Development of NNRTI- and PI-Associated Mutations

Overall, development of primary resistance mutations to NNRTIs and PIs was infrequent during both the placebo-controlled (6% and 4%, respectively) and open-label phases (2.9% and 2.9%, respectively) of this study. The HIV-1 of 17 patients developed a new primary NNRTI-associated resistance mutation during the placebo-controlled phase, 11 of whom had an NNRTI-associated mutation at baseline. Fewer patients in the tenofovir DF arm developed new NNRTI-associated mutations (5% tenofovir DF vs. 9% placebo); however, this did not achieve statistical significance (P = 0.29, Fisher’s exact test). Ten patients developed a new primary PI-associated mutation during the placebo-controlled phase, 7 of whom had a PI-associated mutation at baseline. Again, fewer patients in the tenofovir DF arm developed PI-associated mutations when compared with the placebo arm, and this difference did achieve statistical significance (2% tenofovir DF vs. 8% placebo, P = 0.02, Fisher’s exact test). Similarly, during the open-label phase, most patients who developed new NNRTI- and PI-associated mutations already had resistance mutations to these agents at baseline (5/8 for NNRTIs and 6/8 for PIs). These results suggest that tenofovir DF therapy was retarding the development of NNRTI- and PI-associated mutations, consistent with the significant decreases in viral load observed in these tenofovir DF–treated patients.

HIV-1 RNA Responses in Patients Developing Mutations

The impact of mutations that developed on the virologic response to tenofovir DF was also examined with respect to the secondary endpoint, the absolute change in HIV-1 RNA from baseline (as opposed to DAVG, the time-weighted average change). NAMs that developed in >2% of patients during the placebo-controlled phase were analyzed. Five groups of patients fit these criteria; those whose HIV-1 developed any NAM (n = 27, 10.6%), developed any TAM (n = 19, 7.5%), developed M41L (n = 8, 3.1%), developed D67N (n = 7, 2.7%), or developed K70R/Q/N (n = 6, 2.3%) (Table 2). Among patients from the tenofovir DF arm who were included in these 5 groups, the mean absolute change in HIV-1 RNA from baseline to week 24 ranged from −0.41 ± 0.57 to −0.70 ± 0.56 log10 copies/mL. In comparison, patients in the placebo control group who developed these mutations showed mostly increases in their HIV-1 RNA levels (range −0.03 ± 0.33 to +0.39 log10 copies/mL). When the same patients from the tenofovir DF arm were analyzed at week 48, there was some loss of response between week 24 and week 48. The strongest diminution of antiviral activity was observed among the patients whose HIV-1 developed M41L. Indeed, most of the mutations in these 5 groups are TAMs and would diminish the antiviral response to the background NRTIs in these patients’ regimens, not just the response to tenofovir DF. Nevertheless, analysis of HIV-1 RNA responses in GS-99-907 by baseline genotype has shown decreased response to tenofovir DF among patients whose HIV-1 at baseline had ≥3 TAMs including the M41L and L210W mutations.31 Thus, the development of M41L would be expected to affect the antiviral efficacy of both the thymidine analogues and tenofovir DF in these patients’ regimens.

Phenotypic Changes in Response to Tenofovir DF Therapy

Baseline, week 24, and week 48 phenotypic analyses were attempted for all patients randomly assigned into the virology phenotyping substudy (n = 137). For patients initially in placebo arm, the last sample prior to initiation of tenofovir DF at week 24 was used as the baseline. Of these 137 patients, a baseline and post-baseline phenotype was obtained from 74 patients, including 35 patients with 48 weeks of tenofovir DF exposure. At baseline there was reduced susceptibility to all NRTIs (Table 5). The greatest reduction in susceptibility was observed for zidovudine and lamivudine, consistent with nearly 70% of patients having TAMs or the M184V/I mutation at baseline (Fig. 1). At baseline, mean phenotypic susceptibility to tenofovir DF was 1.8-fold reduced, relative to wild-type HIV-1 (n = 74, range 0.2–7.5-fold). Eight patients had >4-fold reduced susceptibility to tenofovir at baseline (mean 5.2-fold). These 8 patients also had significantly reduced susceptibility to zidovudine (mean 16.5-fold), stavudine (mean 2.0-fold), abacavir (mean 9.5-fold), and lamivudine (mean >23-fold). Patterns of multiple TAMs were present in the HIV-1 of these 8 patients (mean 3.8 TAMs) and 6 had HIV-1 with ≥3 TAMs including M41L and L210W. Although this pattern of TAMs is associated with reduced susceptibility to tenofovir,31 it does not guarantee high-level resistance to tenofovir. The majority of patients (62%) with this pattern of TAMs had tenofovir susceptibilities within 4-fold of wild-type.

Table 5
Table 5:
Baseline and Week 48 Phenotypes

The mean fold changes in phenotypic susceptibility by week 48 (defined as the mean of the individual fold changes of all week 48 samples relative to their matching baseline sample) for all patients (n = 74) were relatively low for all drugs. The highest changes were observed for didanosine (2.1-fold) and zidovudine (1.9-fold). Among those patients whose HIV-1 developed TAMs, the highest mean change was also observed for didanosine (2.3-fold) and zidovudine (2.1-fold). Consistent with the broad cross-resistance mediated by TAMs, some smaller effects on the mean fold change for all drugs were observed among patients who developed these mutations.

Baseline and week 48 phenotypes were available for 5 of the 8 patients who developed K65R. For these 5 patients, the strongest effects on susceptibility were observed for lamivudine and zalcitabine (mean changes from baseline of 2.8-fold and 3.2-fold, respectively). In contrast, the mean fold changes from baseline for zidovudine, stavudine, didanosine, and abacavir were within 1.3-fold of the baseline susceptibility, suggesting that even if K65R does emerge, phenotypically viable future treatment options are available. For tenofovir, addition of K65R resulted in a mean fold change of 1.7 relative to baseline (range 0.5–4.9). Similar phenotypic changes to tenofovir mediated by K65R have been observed using the ViroLogic Phenosense Assay (ViroLogic, South San Francisco, CA) to analyze a database of nearly 300 patient viruses with K65R.23 For 42 viruses with K65R alone, the median phenotypic change to tenofovir is 1.7-fold over wild-type, considerably less than the 4-fold cutoff that corresponds to “no clinically significant response” to tenofovir DF.23 Among the 5 patients from study 907 who developed K65R, the largest individual fold change resulted from 1 patient whose HIV-1 had a Q151M multidrug-resistant genotype at baseline and after addition of K65R had a 4.9-fold reduction in tenofovir susceptibility. HIV-1 containing the Q151M genotype alone retains susceptibility to tenofovir38 but the combination of K65R and Q151M is often associated with >4-fold reduced susceptibility to tenofovir.23

Phenotypic data were available on 4 patients whose HIV-1 developed NAMs that were neither TAMs nor K65R (T69N, L74V/I, F77L, Y115F). No consistent pattern of mutations was present in these 4 patients and there was no notable mean fold change in tenofovir susceptibility (1.3-fold). However, significant mean changes in susceptibility to zidovudine (7.6-fold), stavudine (2.7-fold), didanosine (10.8-fold), and abacavir (4.9-fold) were noted. The mean changes to the latter 3 drugs were driven primarily by the HIV-1 from 1 patient who had apparent phenotypic hypersusceptibility to these drugs at baseline (0.1, 0.2, and 0.5-fold for didanosine, stavudine, and abacavir, respectively). By week 48, this patient’s phenotypic susceptibilities to didanosine, stavudine, and abacavir were 4.0-, 1.8-, and 5.3-fold reduced relative to wild-type, respectively (mean fold changes of 40.0, 9.0, and 10.6, respectively). Similarly, the mean change observed for zidovudine in this group was also due to 1 patient whose HIV-1 was hypersusceptible at baseline (0.3-fold) but had a 4.1-fold reduction in zidovudine susceptibility at week 48 (mean fold change 13.7). As expected, for patients with no additional mutations by week 48 (n = 41), no significant changes in susceptibility relative to baseline were observed for any of the NRTIs. Overall, addition of tenofovir DF therapy for 48 weeks was associated with mostly low-level changes in phenotypic susceptibility to tenofovir DF and other NRTIs.


The therapy options for the heavily treatment-experienced patients entering study 907 had been greatly reduced due to substantial resistance development. A new drug introduced as an intensifying agent into the failing regimens of such patients must be effective against resistant virus, have limited potential for further resistance development, and be tolerable. In study 907, addition of once-daily tenofovir DF therapy to the background regimens of treatment-experienced patients resulted in a statistically significant reduction of HIV-1 RNA of approximately −0.6 log10 copies/mL that was maintained through 48 weeks. In contrast, studies of short-term tenofovir DF monotherapy in treatment-naive patients showed a larger reduction in viral load of −1.6 log10 copies/mL.39 Virologic responses in study 907 were variable in patients who developed the K65R mutation, with this group of patients overall showing a reduced response to tenofovir DF by week 48. Patients who developed TAMs showed a reduction in response to tenofovir DF (mean absolute change of −0.28 log10 copies/mL, Table 2), with the strongest diminution of response being associated with the development of M41L (mean absolute change of −0.19 log10 copies/mL). A pattern of ≥3 TAMs including M41L and L210W has been associated with reduced clinical response to tenofovir DF therapy, based on analyses of baseline resistance mutations in study 907.31 Patients in study 907 were also taking additional NRTIs other than tenofovir DF; thus the development of TAMs would also be expected to affect the response to these components of the patients’ regimens.

Intensification with tenofovir DF reduced the rate of development of new mutations to all classes of HIV-1 drugs in this study population, with reduction in new PI-resistance mutations achieving statistical significance. This is a desired result of intensification, the aims of which are both to suppress the viral load and to prevent the further emergence of resistance to the background regimen. Of the patients who developed new mutations, most developed new TAMs, and despite the potential of these mutations to mediate cross-resistance, tenofovir DF therapy had a continued antiviral benefit in these patients. With the exception of the K65R mutation, the mutations that developed and their frequency were very similar between the tenofovir DF and placebo-controlled arms, suggesting that the background therapies taken concomitantly with tenofovir DF were driving the development of these new mutations in these partially suppressed patients.

The K65R mutation is selected by tenofovir12 and can also be selected by other NRTIs.14–19 Nevertheless, its prevalence in treatment-experienced patients is low (2–4%),20–23 and consistent with its low prevalence, only 8/253 patients (3%) in the virology substudy of study 907 developed K65R. Four of these patients were taking abacavir or didanosine, and these NRTIs may have contributed to its selection. The low rate of emergence of K65R after 48 weeks of tenofovir DF therapy in study 907 (3%) is similar to the rate observed after 48 weeks in treatment-naive patients (2.3%) in study 903.40 Despite the very different patient populations in these 2 trials, the K65R mutation developed at similar low frequencies in both trials.

All 8 patients whose HIV-1 developed K65R in study 907 were derived from the approximately 30% of patients who did not have detectable TAMs at baseline, and with only 1 exception, these patients did not develop any TAMs. This observation suggests that the K65R and TAM pathways may represent 2 distinct paths in viral evolution. A strong negative association of K65R with TAMs has been reported in several studies.21,23 The virologic and host factors that determine whether HIV-1 will select the K65R mutation vs. TAMs are largely unknown, particularly in patients taking both tenofovir DF and thymidine analogues as occurred in study 907. It is unclear whether K65R and the TAMs are functionally incompatible substitutions in the viral RT. The complex mutational interplay resulting from competing NRTIs in a regimen may also have a role in the apparent mutual exclusion of these 2 resistance pathways. It has been observed that the presence of K65R increases susceptibility to zidovudine in vitro41 and this has also been observed in the phenotypes of patients who developed K65R in study 903.34,40

A recent retrospective pooled analysis of abacavir monotherapy and combination therapy has suggested that the presence of zidovudine in a regimen can strongly reduce the emergence of K65R but, instead, results in greater TAM emergence.42 Five of the 8 patients who developed K65R in study 907 were taking thymidine analogues (3 zidovudine, 2 stavudine) concomitantly, demonstrating that the presence of a thymidine analogue in a regimen is not sufficient to completely prevent the emergence of K65R. Recently, K65R has been shown to be selected under stavudine drug pressure and can cause reduced susceptibility to stavudine in enzymatic assays.17 Interestingly, 2 virologic failure patients in the stavudine-containing arm of study 903 (stavudine + lamivudine + efavirenz) also developed K65R by week 48.34 These data demonstrate that there is potential to select K65R even in the presence of a thymidine analogue. It is important to distinguish between the presence of preexisting TAMs, which may strongly antagonize the selection of K65R, as was observed in study 907, and the potential to influence selection of K65R in the presence of a thymidine analogue. Influencing the NRTI resistance pathway must be balanced with the cost, complexity, and safety considerations of adding another NRTI and an understanding of the consequences of the alternate resistance pathway.

Given that K65R is a single-point mutation, what factors result in its low rate of emergence, even in treatment-experienced patients with ongoing viral replication? The much lower frequency of development of K65R compared with TAMs in study 907 may reflect the observation that viruses carrying K65R have reduced replication capacity (approximately 50% of wild-type).27,28,34 Similar to the M184V mutation,43–45 the K65R mutation reduces replication capacity in viruses with a variety of genetic backgrounds when both site-directed mutant viruses and clinically derived viruses are examined.27,28,34 The combination of the K65R and M184V mutations results in a virus with an additive reduction in replication capacity.27,28 In contrast, the TAMs do not exert as strong an effect on replication capacity (K65R, RC = 53–56%; M184V, RC = 57%; K65R + M184V, RC = 24–29%; TAMs, RC = 73–82%).27,28 Six of the 8 patients who developed K65R had M184V at baseline, and by week 48, 5 of the 8 patients had plasma HIV-1 with both mutations (Table 3). All 5 patients were on regimens that included either lamivudine or abacavir, NRTIs that would have maintained selection pressure to retain this mutation. Patients whose HIV-1 developed K65R while on tenofovir DF therapy lose this mutation rapidly upon removal of tenofovir, suggesting that the K65R mutation is detrimental to the virus (unpublished observations). Additional mutations that can associate with K65R, such as S68G, may partially compensate for the replication capacity defect that results from K65R.23,27,28,35

Another factor that may influence development of K65R is the fold-resistance that this mutation imparts to the virus. The median phenotypic change (ViroLogic Phenosense Assay) to tenofovir in patient viruses with K65R alone is 1.7-fold over wild-type, considerably less than the 4-fold cutoff that corresponds to “no clinically significant response” to tenofovir DF.23,31 In contrast, selection of the M184V mutation in the presence of lamivudine leads to high-level lamivudine resistance (>100-fold)44 due to the steric hindrance of lamivudine triphosphate binding.46 The low level of resistance against tenofovir mediated by the K65R mutation may result in less selective advantage to the virus than that mediated by the M184V mutation in the presence of lamivudine.

Other than K65R, no novel mutations were detected that could be attributed specifically to tenofovir DF therapy in study 907. Thus, after 2 trials in treatment-experienced patients (studies 902 and 907)13 and 2 years of tenofovir DF therapy in treatment-naive patients (study 903),34,40 the only mutation that can currently be attributed to tenofovir DF therapy is K65R and this occurred at a low frequency. Compared with these studies, higher frequencies of this mutation have been observed in recent studies that have investigated once-daily, nucleoside-only regimens in treatment-naive patients. High rates of virologic failure in the trials of these triple-nucleoside-based regimens (abacavir/lamivudine/tenofovir DF once daily and didanosine/lamivudine/tenofovir DF once daily) have resulted in early closure of these trials.47,48 In all 3 studies, most virologic failure patients developed resistance mutations, with nearly 100% developing M184V and approximately 50% developing both K65R and M184V. The use of once-daily triple-nucleoside regimens may have resulted in suboptimal potency, thereby facilitating resistance selection. Other triple-nucleoside regimens that have been studied, including stavudine plus didanosine plus abacavir, have also shown poor virologic efficacy and resistance development.33,49 As the reasons for the reduced potency of triple-nucleoside regimens remain unclear, the results of these trials highlight the importance of combining NRTIs with antiretrovirals of other classes, as is currently recommended as the standard of care in the International AIDS Society HIV treatment guidelines.50 In study 907, patients intensified their partially suppressive regimens with the addition of tenofovir DF, achieving both a sustained virologic response and reduction in the emergence of new resistance mutations. In treatment-experienced patients, tenofovir DF combined with other antiretroviral agents was effective and had a low rate of resistance development in the form of the K65R mutation, thereby preserving future treatment options.


The authors thank all study site personnel and patients who participated in study 907, Mick Hitchcock, Craig Gibbs, Alvan Fisher, and Kirsten White for their careful critique of this manuscript, and Justin Hendrix and Susan Edl for assistance in the preparation of this manuscript.


1. Squires KE. An introduction to nucleoside and nucleotide analogues. Antivir Ther. 2001;6:1–14.
2. Volberding PA. HIV therapy in 2003: consensus and controversy. AIDS. 2003;17:S4–S11.
3. Whitcomb JM, Parkin NT, Chappey C, et al. Broad nucleoside reverse-transcriptase inhibitor cross-resistance in human immunodeficiency virus type 1 clinical isolates. J Infect Dis. 2003;188:992–1000.
4. Naesens L, Snoeck R, Andrei G, et al. HPMPC (cidofovir), PMEA (adefovir) and related acyclic nucleoside phosphonate analogues: a review of their pharmacology and clinical potential in the treatment of viral infections. Antivir Chem Chemother. 1997;8:1–23.
5. Mulato AS, Cherrington JM. Anti-HIV activity of adefovir (PMEA) and PMPA in combination with antiretroviral compounds: in vitro analyses. Antiviral Res. 1997;36:91–97.
6. Balzarini J, Aquaro S, Perno CF, et al. Activity of the (R)-enantiomers of 9-(2-phosphonylmethoxypropyl)-adenine and 9-(2-phosphonylmethoxypropyl)-2,6-diaminopurine against human immunodeficiency virus in different human cell systems. Biochem Biophys Res Commun. 1996;219:337–341.
7. Srinivas RV, Robbins BL, Connelly MC, et al. Metabolism and in vitro antiretroviral activities of bis(pivaloyloxymethyl) prodrugs of acyclic nucleoside phosphonates. Antimicrob Agents Chemother. 1993;37:2247–2250.
8. Balzarini J, Holý A, Jindrich J, et al. Differential antiherpesvirus and antiretrovirus effects of the (S) and (R) enantiomers of acyclic nucleoside phosphonates: potent and selective in vitro and in vivo antiretrovirus activities of (R)-9-(2-phosphonomethoxypropyl)-2, 6-diaminopurine. Antimicrob Agents Chemother. 1993;37:332–338.
9. Ying C, De Clercq E, Nicholson W, et al. Inhibition of the replication of the DNA polymerase M550V mutation variant of human hepatitis B virus by adefovir, tenofovir, L-FMAU, DAPD, penciclovir and lobucavir. J Viral Hepat. 2000;7:161–165.
10. Nelson M, Portsmouth S, Stebbing J, et al. An open-label study of tenofovir in HIV-1 and Hepatitis B virus co-infected individuals. AIDS. 2003;17:F7–F10.
11. Ristig MB, Crippin J, Aberg JA, et al. Tenofovir disoproxil fumarate therapy for chronic hepatitis B in human immunodeficiency virus/hepatitis B virus-coinfected individuals for whom interferon-alpha and lamivudine therapy have failed. J Infect Dis. 2002;186:1844–1847.
12. Wainberg MA, Miller MD, Quan Y, et al. In vitro selection and characterization of HIV-1 with reduced susceptibility to PMPA. Antivir Ther. 1999;4:87–94.
13. Margot NA, Isaacson E, McGowan I, et al. Extended treatment with tenofovir disoproxil fumarate in treatment-experienced HIV-1-infected patients: genotypic, phenotypic, and rebound analyses. J Acquir Immune Defic Syndr Hum Retrovirol. 2003;33:15–21.
14. Tisdale M, Alnadaf T, Cousens D. Combination of mutations in human immunodeficiency virus type 1 reverse transcriptase required for resistance to the carbocyclic nucleoside 1592U89. Antimicrob Agents Chemother. 1997;41:1094–1098.
15. Harrigan PR, Stone C, Griffin P, et al. Resistance profile of the human immunodeficiency virus type 1 reverse transcriptase inhibitor abacavir (1592U89) after monotherapy and combination therapy. J Infect Dis. 2000;181:912–920.
16. Winters MA, Shafer RW, Jellinger RA, et al. Human immunodeficiency virus type 1 reverse transcriptase genotype and drug susceptibility changes in infected individuals receiving dideoxyinosine monotherapy for 1 to 2 years. Antimicrob Agents Chemother. 1997;41:757–762.
17. Garcia-Lerma JG, MacInnes H, Bennett D, et al. A novel genetic pathway of human immunodeficiency virus type 1 resistance to stavudine mediated by the K65R mutation. J Virol. 2003;77:5685–5693.
18. Gu Z, Fletcher RS, Arts EJ, et al. The K65R mutant reverse transcriptase of HIV-1 cross-resistant to 2′,3′-dideoxycytidine, 2′,3′-dideoxy-3′-thiacytidine, and 2′,3′- dideoxyinosine shows reduced sensitivity to specific dideoxynucleoside triphosphate inhibitors in vitro. J Biol Chem. 1994;269:28118–28122.
19. Zhang D, Caliendo AM, Eron JJ, et al. Resistance to 2′,3′-dideoxycytidine conferred by a mutation in codon 65 of the human immunodeficiency virus type 1 reverse transcriptase. Antimicrob Agents Chemother. 1994;38:282–287.
20. Bloor S, Kemp SD, Hertogs K, et al. Patterns of HIV drug resistance in routine clinical practice: a survey of almost 12000 samples from the USA in 1999. Antivir Ther. 2000;5:132.
21. Parikh U, Koontz D, Hammond J, et al. K65R: a multi-nucleoside resistance mutation of low but increasing frequency. Antivir Ther. 2003;8:S152.
22. Lanier ER, Scott J, Ait-Khaled M, et al. Prevalence of mutations associated with resistance to antiretroviral therapy from 1999–2002. Paper presented at: 10th Conference on Retroviruses and Opportunistic Infections; February 10–14, 2003; Boston.
23. Miller MD, McColl DJ, White KL, et al. Genotypic and phenotypic characterization of patient-derived HIV-1 isolates containing the K65R mutation in reverse transcriptase (RT). Paper presented at: 43rd Annual Interscience Conference on Antimicrobial Agents and Chemotherapy; September 14–17, 2003; Chicago.
24. Schooley RT, Ruane P, Myers RA, et al. Tenofovir DF in antiretroviral-experienced patients: results from a 48-week, randomized, double-blind study. AIDS. 2002;16:1257–1263.
25. Margot NA, Isaacson E, McGowan I, et al. Genotypic and phenotypic analyses of HIV-1 in antiretroviral-experienced patients treated with tenofovir DF. AIDS. 2002;16:1227–1235.
26. Squires K, Pozniak AL, Pierone G Jr, et al. Tenofovir disoproxil fumarate in nucleoside-resistant HIV-1 infection. Ann Intern Med. 2003;139:313–321.
27. White KL, Margot NA, Wrin T, et al. Molecular mechanisms of resistance to human immunodeficiency virus type 1 with reverse transcriptase mutations K65R and K65R + M184V and their effects on enzyme function and viral replication capacity. Antimicrob Agents Chemother. 2002;46:3437–3446.
28. Miller MD, White KL, Petropoulos CJ, et al. Decreased replication capacity of HIV-1 clinical isolates containing K65R or M184V RT mutations. Paper presented at: 10th Conference on Retroviruses and Opportunistic Infections; February 10–14, 2003; Boston.
29. DeGruttola V, Dix L, D’Aquila R, et al. The relation between baseline HIV drug resistance and response to antiretroviral therapy: re-analysis of retrospective and prospective studies using a standardized data analysis plan. Antivir Ther. 2000;5:41–48.
30. Hertogs K, de Bethune MP, Miller V, et al. A rapid method for simultaneous detection of phenotypic resistance to inhibitors of protease and reverse transcriptase in recombinant human immunodeficiency virus type 1 isolates from patients treated with antiretroviral drugs. Antimicrob Agents Chemother. 1998;42:269–276.
31. Miller MD, Margot N, Lu B, et al. Genotypic and phenotypic predictors of the magnitude of treatment response to tenofovir disoproxil fumarate in antiretroviral-experienced patients. J Infect Dis. 2004;189:837–846.
32. Bacheler LT, Anton ED, Kudish P, et al. Human immunodeficiency virus type 1 mutations selected in patients failing efavirenz combination therapy. Antimicrob Agents Chemother. 2000;44:2475–2484.
33. Roge BT, Katzenstein TL, Obel N, et al. K65R with and without S68: a new resistance profile in vivo detected in most patients failing abacavir, didanosine and stavudine. Antivir Ther. 2003;8:173–182.
34. Miller MD, Margot NA, McColl DJ, et al. Characterization of resistance mutation patterns emerging over 2 years during first-line antiretroviral treatment with tenofovir DF or stavudine in combination with lamivudine and efavirenz. Antivir Ther. 2003;8:S151.
35. Garcia-Lerma JG, Gerrish PJ, Wright AC, et al. Evidence of a role for the Q151L mutation and the viral background in development of multiple dideoxynucleoside- resistant human immunodeficiency virus type 1. J Virol. 2000;74:9339–9346.
36. Miller V, Ait-Khaled M, Stone C, et al. HIV-1 reverse transcriptase (RT) genotype and susceptibility to RT inhibitors during abacavir monotherapy and combination therapy. AIDS. 2000;14:163–171.
37. Larder BA, Bloor S, Kemp SD, et al. A family of insertion mutations between codons 67 and 70 of human immunodeficiency virus type 1 reverse transcriptase confer multinucleoside analog resistance. Antimicrob Agents Chemother. 1999;43:1961–1967.
38. Wolf K, Walter H, Schnell T, et al. The drug resistance profile of tenofovir: a story of resistance and resensitization. Antivir Ther. 2002;7:S22.
39. Louie M, Hogan C, Hurley A, et al. Determining the antiviral activity of tenofovir disoproxil fumarate in treatment-naive chronically HIV-1-infected individuals. AIDS. 2003;17:1151–1156.
40. Miller MD, Margot NA, McColl DJ, et al. Genotypic and phenotypic characterization of virologic failure through 48 weeks among treatment-naive patients taking tenofovir DF or stavudine in combination with lamivudine and efavirenz. Paper presented at: VI International Congress on Drug Therapy in HIV Infection; November 17–21, 2002; Glasgow.
41. Bazmi HZ, Hammond JL, Cavalcanti SCH, et al. In vitro selection of mutations in the human immunodeficiency virus type 1 reverse transcriptase that decrease susceptibility to (−)-beta-D-dioxolane-guanosine and suppress resistance to 3′-azido-3′-deoxythymidine. Antimicrob Agents Chemother. 2000;44:1783–1788.
42. Ait-Khaled M, Lanier R, Richards N, et al. Zidovudine appears to prevent selection of K65R and L74V mutations normally selected by abacavir mono- or combination therapies not containing zidovudine. Antivir Ther. 2002;7:S141.
43. Diallo K, Marchand B, Wei X, et al. Diminished RNA primer usage associated with the L74V and M184V mutations in the reverse transcriptase of human immunodeficiency virus type 1 provides a possible mechanism for diminished viral replication capacity. J Virol. 2003;77:8621–8632.
44. Naeger LK, Margot NA, Miller MD. Increased drug susceptibility of HIV-1 reverse transcriptase mutants containing M184V and zidovudine-associated mutations: analysis of enzyme processivity, chain-terminator removal and viral replication. Antivir Ther. 2001;6:115–126.
45. Wei X, Liang C, Gotte M, et al. Negative effect of the M184V mutation in HIV-1 reverse transcriptase on initiation of viral DNA synthesis. Virology. 2003;311:202–212.
46. Gao H-Q, Boyer PL, Sarafianos SG, et al. The role of steric hindrance in 3TC resistance of human immunodeficiency virus type-1 reverse transcriptase. J Mol Biol. 2000;300:403–418.
47. Farthing C, Khanlou H, Yeh V. Early virologic failure in a pilot study evaluating the efficacy of abacavir, lamivudine and tenofovir in the treatment naive HIV-infected patients. Antivir Ther. 2003;8:S195.
48. Gallant JE, Rodriguez A, Weinberg W, et al. Early non-response to tenofovir DF (TDF) + abacavir (ABC) and lamivudine (3TC) in a randomized trial compared to Efavirenz (EFV) + ABC and 3TC: ESS30009 unplanned interim analysis. Paper presented at: 43rd Interscience Conference on Antimocrobial Agents and Chemotherapy; September 14–17, 2003; Chicago.
49. Gerstoft J, Kirk O, Obel N, et al. Low efficacy and high frequency of adverse events in a randomized trial of the triple nucleoside regimen abacavir, stavudine and didanosine. AIDS. 2003;17:2045–2052.
50. Yeni PG, Hammer SM, Carpenter CCJ, et al. Antiretroviral treatment for adult HIV infection in 2002: updated recommendations of the International AIDS Society-USA Panel. JAMA. 2002;288:222–235.

HIV-1; drug resistance; tenofovir; mutation; reverse transcriptase (RT); genotype; phenotype

© 2004 Lippincott Williams & Wilkins, Inc.