Update from the International Workshop on HIV Drug Resistance, Treatment Strategies and Eradication
AIDS INFORMATION NEWSLETTER
AIDS Information Center
VA Medical Center, San Francisco
Protease Inhibitor Resistance and Drug Failure
Virological outcomes in ACTG 320, a randomized, placebo controlled
trial of indinavir in combination with two nucleosides
ACTG 320 is a recently completed randomized, double blind,
placebo controlled study of zidovudine (AZT) and lamivudine (3TC)
vs. the combination of AZT, 3TC and indinavir (Crixivan). The study
enrolled 1156 subjects, all with a CD4+ T cell count less than 200
cells/cc. Subjects were 3TC and protease inhibitor naÐve at study
entry. After a median follow-up of 38 weeks, a significant clinical
benefit was observed in the indinavir arm (hazard ratio 0.50, CI
0.33 to 0.76).
In a virology substudy, 374 subjects were analyzed in using
the Amplicor HIV-1 RNA assay (lower limit of detection 500
copies/ml). The baseline HIV-1 RNA was approximately 5.0 log
(100,000) copies/ml in both treatment groups.
After 24 weeks of therapy, 3% of patients on AZT and 3TC had
undetectable levels of HIV-1 RNA, compared to 60% in the AZT, 3TC
and indinavir arm. In the cohort who initiated three drug therapy
with a baseline CD4+ T cell count less than 50 cells/cc, only 40%
maintained undetectable levels of HIV-1 RNA through week 24. These
trends persist through week 40. Finally, a high viral load or low
CD4+ T cell count at baseline was highly predictive of drug failure
[Hammer, abstract 67].
The ACTG 320 experience suggests that about one half of
patients who initiate AZT, 3TC and indinavir late in the natural
history of the disease process can expect to obtained long-term,
complete viral suppression. This is a much lower number than
previously reported (Merck 035). HIV therapy should therefore be
Nelfinavir resistance: significance of pre-existing natural
Amy Patick from Agouron reviewed her data from 170 patients
treated with nelfinavir (Viracept). As reported before, nelfinavir
failure is strongly associated with the D30N substitution. D30N
rarely evolves alone, and is often seen with mutations at positions
10, 35, 36, 62, 63, 71, 77, 88 (among others). Many of these
mutations exist as natural polymorphisms (i.e., they are present
at low frequencies in untreated patients). The significance of
these polymorphisms remains unclear. In this study, the presence
of these polymorphisms at baseline did not predict subsequent
failure to nelfinavir.
Notably, 5% of patients failing nelfinavir developed L90M
(which is associated with high-level saquinavir resistance) [Patick
et al, abstract 18].
Comment: The HIV-1 protease gene is highly variable, even in the
absence of protease inhibitors. Many "mutations" seen with therapy
are actually natural polymorphisms. This study suggests that the
presence of these polymorphisms does not predict failure to a drug.
Resistance after long-term saquinavir therapy
Researchers at Stanford University performed genotypic and
phenotypic assays on viral isolates from 40 patients who "failed"
long-term therapy with saquinavir (Invirase). As expected, most
patients (58%) developed G48V or L90M. The G48V occurred more
frequently in patients treated with high dose saquinavir. A
significant percentage of patients (38%) also had mutations
typically associated with other protease inhibitors, including
M36I, M46I, V82A, and I84A.
All patients (6 of 6) who developed G48V eventually developed
V82A (a mutation associated with resistance to indinavir or
ritonavir), either while on continued saquinavir or after switching
to nelfinavir or indinavir. V82A did not typically develop in the
presence of L90M [see also Eastman, abstract 30; Deeks, abstract
Comment: The development of complex mutation patterns was common
in patients after prolonged (> 1 year) treatment with saquinavir.
In such patients, cross-resistance to other protease inhibitors
would be expected. Preliminary data from this presentation, and
others, suggests that the resistance pathway taken by the virus may
be influenced by the initial presence of L90M vs. G48V.
Indinavir resistance following saquinavir therapy
Sequential protease inhibitor therapy was a common theme at
this year's conference. In this study, 54 patients were treated
with saquinavir (Invirase) and followed prospectively. Twenty-two
of these patients eventually failed saquinavir and were switched
to indinavir (Crixivan). Ten of the 22 patients had a durable
response to indinavir, while 12 eventually failed. Failure was
defined as having a viral load greater than 3.5 log copies/mL after
a mean 4 month period of treatment with indinavir.
Genotypic analysis was performed on the 12 patients who failed
indinavir. Prior to the introduction of indinavir, 5 of the 12
patients had a mutation at position L90M. All 5 patients maintained
L90M in the presence of indinavir; none developed the classic
indinavir associated mutation V82A. A previously uncharacterized
mutation, G73S, was observed in all post-indinavir specimens.
In the other 6 patients who failed indinavir after prolonged
saquinavir thearapy, saquinavir related resistance patterns were
not present at the time of the switch. Surprisingly, after
switching to indinavir, all patients failed with typical saquinavir
related mutations (L90M or G48V). Typical indinavir related
mutations, such as V82A, were rare [Dulioust et al, abstract 16].
Comment: These results suggest that saquinavir related genotypic
resistance confers cross-resistance to indinavir in vivo, despite
the lack of significant cross-resistance in vitro. Furthermore,
patients may have a normal viral genotype (i.e., no mutations
evident on genotypic analysis) after prolonged therapy with
saquinavir, yet still rapidly select for saquinavir related
mutations in the presence of indinavir. This suggests that these
mutants pre-existed below the levels detectable with current
Acquisition of genotypic resistance associated with reduced
susceptibility to saquinavir (hard gel capsules)
Acquisition of genotypic resistance associated with saquinavir (soft gel capsule)
Eastman and colleagues performed genotypic assays on 7
patients who initially responded to and eventually failed
saquinavir (hard gel capsule). Viral rebound was associated with
the development of an L90M mutation alone (n=4), G48V alone (n=2)
or both (n=1). All 3 subjects with the G48V mutation developed
subsequent additional mutations, including the V82A mutation
typically seen with indinavir or ritonavir resistance. Two subjects
acquired an L90M mutation that was subsequently lost in the
presence of the V82A mutation [Eastman et al, abstract 30].
In a companion presentation, 13 patients added an experimental
formulation of saquinavir (saquinavir-soft gel capsule) to a stable
nucleoside analogue regimen. After 8 weeks of therapy, a potent 1.7
log decrease in plasma HIV-1 RNA was observed. Six patients
subsequently failed (defined as a viral rebound of > 1 log). Viral
rebound was associated with L90M (n=1), G48V (n=1) or both (n=2).
One patient rebounded with V77E only (results from the sixth
patient were not available). Interestingly, one patient with prior
ritonavir failure switched to saquinavir soft-gel capsule. At the
time of the switch, this patient had three mutations associated
with high-level resistance to indinavir and ritonavir (mixed
V82V/A, M46M/L and I84I/V). As expected, SQV-SGC selected for L90M,
but V82, M46 and I84 all reverted to wild-type. This patient
subsequently switched to indinavir, and has had a durable 7 month
response, despite the prior presence of mutations highly correlated
with indinavir resistance [Deeks et al, abstract 69].
Comment: These two studies confirm that G48V and L90M are the two
most important mutations associated with saquinavir therapy. They
also suggest that L90M may not be compatible with V82A,
particularly when other so-called "compensatory mutations" are not
present [see also Winters, abstract 17].
Nelfinavir and indinavir therapy following failure of saquinavir
In a companion study to Winters et al (abstract 17), Jody
Lawrence from Stanford reported on 16 patients who switched to
nelfinavir (Viracept) after failing saquinavir. When possible,
nucleoside analogues were modified. Patients switched to nelfinavir
after a mean duration on saquinavir of 11 months.
After 4 weeks of therapy with nelfinavir, the mean decrease
in HIV-1 RNA was 0.56 log. Only 2 of 16 subjects had a > 0.5 log
decline through week 12 of nelfinavir therapy. At week 12, 2
patients failing nelfinavir had evidence of D30N, the classic
mutation associated with nelfinavir resistance. Most patients who
failed nelfinavir had mutations typically seen with other protease
Considering the poor response to nelfinavir, the study was
modified. Ten patients who failed nelfinavir were switched to
indinavir (Crixivan) plus nevirapine (Viramune). After 4 weeks of
therapy, the mean reduction in viral load was 1.8 log. The
durability of this potent response is unknown [Lawrence et al,
Comment: In this study, saquinavir resistance conferred cross-
resistance to nelfinavir. Therefore, the utility of nelfinavir in
patients with prolonged prior saquinavir may be limited. On the
other hand, the combination of indinavir plus nevirapine appeared
to be very active over 4 weeks of follow-up, despite the fact that
these patients had failed two protease inhibitors. The long-term
efficacy of indinavir plus nevirapine as salvage therapy is
unknown. Finally, resistance patterns to a particular protease
inhibitor can be very different when the drug is used in patients
who have already failed a first-line protease inhibitor.
Viral resistance to the combination of ritonavir and saquinavir
The combination of ritonavir and saquinavir is extremely
effective in suppressing HIV-1 replication. In an ongoing study of
this combination, approximately 80 to 90% of subjects have had a
durable virologic response. In this study, genotypic analysis was
performed on viral isolates from fivepatients who exhibited a
rebound in viral load. Rebound was associated with the emergenceof
V82A and I54V in all five patients. I84V and M36I were also seen.
L90M and G48V, both typically seen with saquinavir failure, were
not observed [Molla et al, abstract 83].
Comment: Patients who fail the combination of ritonavir and
saquinavir appear to select primarily for ritonavir related
mutations. Since ritonavir resistance is known to confer cross
resistance to saquinavir, this observation is perhaps not
surprising. More resistance data on this combination is needed.
Drug resistance genotypes from plasma virus of HIV-infected
patients failing combination drug therapy
Genotypic analysis of HIV-1 protease from patients failing highly
active anti-retroviral therapy
Several groups reported cross-sectional results from patients
failing combination therapies.
One group, studying viral isolates from the military health
care system, reported that 21% of patients failing a combination
regimen had no evidence of genotypic resistance. Surprisingly,
over 50% of patients failing a protease inhibitor-containing
regimen had no genotypic changes in the protease gene [Mayers et
al, abstract 80].
In a study of patients in the Denver area, 33 patients failing
a protease inhibitor containing regimen were analyzed (most
patients had received indinavir). Compared to the clade B consensus
sequence, patients failing combination therapy had a mean of 4
mutations in their protease gene. Four of the 33 subjects had no
protease mutations. Despite the infrequent use of saquinavir, L90M
was very common (15 of 37 isolates) [Young et al, abstract 65].
Comment: The clinical experience with protease inhibitors in the
primary care setting is likely to be very different from the
experience seen in controlled clinical trials. These studies
illustrate this point. The observation that many patients fail
without genotypic evidence of resistance may be due to several
factors: (1) variable drug absorption, (2) enhanced drug
metabolism, (3) poor adherence to the drug regimen and/or (4)
limited sensitivity of the genotypic assays.
Back | Next
Table of Contents