Abstract Book Vol. 2 [International Conference on AIDS (14th: 2002: Barcelona, Spain)]

XIV International AIDS Conference

Abstracts ThOrB1385-ThOrB1388 377

ThOrB1385 The GUESS study: An international evaluation of
the accuracy, precision & consistency of expert
HIV-1 genotype interpretation
A.R. Zolopa1, L.C. Lazzeroni1, A. Rinehart2, D. Kuritzkes3. 1Stanford University,
1101 welch road, suite a3, palo alto, california 94025, United States;
2Tibotec-Virco, Durham,NC, United States; 3Harvard University, Boston, MA,
United States
Introduction: We evaluated genotype interpretations from clinical specimens by
an international panel of experts. We tested the ability to predict assayed phenotype, agreement between experts in predicting phenotype, antiretroviral drug
activity and consistency of treatment recommendations.
Methods: We randomly selected 50 genotypes from the VIRCO database with
matching phenotype. The isolates had to be resistant to at least one drug (IC50
>4-fold). For each genotype, experts were asked to predict the susceptibility to
each of 16 antiretrovirals (ARV's) as a categorical fold-change (FC) in IC50 (<2.5;
2.5-<4; 4-<7; 7-<10; 10-<20; >20). Expected drug activity was rated on a 6 point
scale and experts noted whether or not the ARV would be recommended for use
in treatment.
Results: 12 experts interpreted 50 genotypes. Accuracy in predicting FC varied
within nRTI's from 25% for abacavir to 74% for 3TC; within Pl's it varied from 26%
for NFV to 40% for LPV/r. Experts were able to predict FC for NNRTI's about 2/3's
of the time. Median difference between predicted FC and assayed FC category
was 0 (ie no difference) for all drugs except AZT, D4T, DDI, DDC and ABC where
the median difference was +1 category The correlation (rho) between experts in
predicting phenotype ranged from 0.23 for TDF to 0.84 for 3TC for nucleos/tide
analogs, from 0.89 for EFV to 0.93 for NVP for NNRTI's and 0.68 for APV to 0.79
for IDV and RTV for Pl's. The inter-expert correlation in predicting drug activity
showed a similar range and pattern as predicted phenotype. Consensus between
experts on treatment recommendations ranged from 62% for DDI to 90% for 3TC
with a median of 79%.
Conclusions: This panel was able to predict phenotype FC within a relatively narrow range. However, experts systematically overestimated the FC for AZT, d4T,
ddl, ddC, and ABC. Inter-expert agreement in predicting FC, predictions of drug
activity and treatment recommendations was high but varied by ARV.
Presenting author: andrew zolopa, 1101 welch road, suite a3, palo alto, california 94025, United States, Tel.: [phone], Fax: [phone], E-mail:
[email protected]
ThOrB1386 Genotypic (GT) and virtual phenotypic (vPT)
correlates of virologic response to Abacavir
(ABC)-based therapy in the ZORRO trial (ESS40009)
P. Ruane1, J. Hernandez2, G. Richmond3, A. Williams4, T Fralich5, L. Yau2,
S. Hessenthaler2, E.R. Lanier2. 1Tower ID, LA, United States; 2GlaxoSmithKline,
RTP United States; 3Broward General Medical Center, Ft. Lauderdale, United
States; 4 Graduate Hospital, Philadelphia, United States; 5imperial Point Medical
Center, Ft. Lauderdale, United States
Background: Resistance testing may improve response in ART-experienced subjects. GT and PT testing have some utility in this setting. We tested the ability of
the Virco GT and vPT to predict response to open-label ABC-based therapy in
ART-experienced subjects.
Methods: 833 ART-naive and experienced subjects were treated in the US with
ABC 300mg BID and > 2 other AR drugs, and followed for 16 weeks (wks). GT
and vPT were done on ART-experienced subjects as per investigator discretion.
Subjects with isolates susceptible to ABC and > 2 other AR drugs were enrolled.
Changes in therapy were allowed as clinically indicated. Virologic responses (ITT,
Observed analyses) were assessed at 8 and 16 wks and correlated with BL GT
and vPT characteristics.
Results: 180 subjects had GT and vPT performed. Median BL vRNA was 4.3
logs; median CD4 count was 315 cells/mm3. 98% of isolates had < 4 nucleosideassociated mutations and a vPT to ABC of < 4.5 fold compared to control virus
(VFR). At Wk 8, median change in vRNA from BL was -1.38; median CD4 cell
increase was 69, and the proportion of subjects with vRNA < 400 c/mL or > 1
log drop from BL was 71%. At Wk 16, median change in vRNA from BL was -
1.33, median CD4 cell increase was 47 and the proportion of subjects with vRNA
< 400 c/mL or >_ 1 log drop from BL was 65%. The table shows the proportion
of subjects who achieved vRNA <400 c/mL or _ 1 log drop from BL by BL resistance for ABC (and likely other NRTIs). *Note that all TAM categories are +/M184V. Notably, overall responses to these multi-switch regimens appeared to
be predictable based on GT and vPT results. Additional analysis to include other
variables will be presented.

ThOrB1 387 Multivariate logistic regression analysis of factors
predictive of the virological response in the Narval
trial
M. Vray1, J.L. Meynard2, C. Dalban', L. Morand-Joubert2, F. Brun-Vezinet3,
F. Clavel4, D. Costagliola', P.M. Girard2. 'Inserm SC4, Inserm SC4, 56
Boulevard Vincent Auriol, 75013 Paris, France; 2Hopital Saint Antoine, Paris,
France; 3Hopital Bichat, Paris, France; 41nserm, Paris, France
Objective: To identify factors predictive of W12 virological response.
Methods: The NARVAL trial was conducted to assess the respective benefit of
Phenotype (P),Genotype (G) versus standard of care (SOC) for the choice of
antiretroviral therapy in patients failing a PI-containing regimen. Virological success was defined as viral load (VL) le 200 copies/ml at W12. Baseline variables
including demographic, clinical and biological characteristics, drug resistance mutations, randomization arm, compliance to treatment and drugs prescribed were
analyzed univariately. Significant variables were included in a logistic regression
model.
Results: 541 patients were randomized. Median baseline CD4 counts and VL
were 339 cells/mm3 and 3.4 log cp/ml, respectively. 202 patients reached virological success at W12. In multivariate analysis, the factors significantly associated with virological success were: prescription of EFV in NNRTI-naive patients
(OR=4.37; 95%CI, 2.76-6.90), randomization in G arm (OR=2.13, 1.20-3.80),
prescription of 3TC (OR=1.69, 1.01-2.83) and prescription of abacavir (ABC) in
ABC-naive patients (OR=1.66, 1.02-2.72). Factors significantly associated with
virological failure were: prescription of NFV (OR=0.30, 0.13-0.68), baseline VL
(OR=0.37, 0.28-0.50), presence of ge 5 PI mutations (OR=0.42, 0.26-0.66), presence of ge 3 NAMs (OR=0.61, 0.39-0.97), previous PI exposure ge 30 months
(OR=0.64, 0.41-0.99).
Conclusions: Randomization in G group, prescription of EFV, ABC or 3TC were
associated with an increased virological success rate. In contrast, a high number
of resistance mutations, long prior exposure to PIs, high baseline VL and prescription of NFV were associated with a decreased virological success rate.
Presenting author: Muriel Vray, Inserm SC4, 56 Boulevard Vincent Auriol, 75013
Paris, France, Tel.: [phone], Fax: [phone], E-mail:
[email protected]
ThOrB1388 The clinical relevance of non-nucleoside reverse
transcriptase inhibitor (NNRTI) hypersusceptibility:
a prospective cohort analysis
R. Haubrich1, N. Hellmann2, P. Keiser3, C. Kemper4, M. Witt5, D. Forthal6,
J. LeedoM7, M. Leibowitz8, D. Richman9, J.A. McCutchan', C.C.T. Group10.
SUCSD, San Diego, CA, United States; 2 ViroLogic, South San Francisco, United
States; 3UT Southwestern, Dallas, United States; 4SCVMC, San Jose, United
States; 5Harbor- UCLA, Los Angeles, United States; 6UCI, Irvine, United States;
7USC, Los Angeles, United States; 8 UCLA, Los Angeles, United States; 9UCSD,
SD VA, La Jolla, United States; o10, United States
Objective: To evaluate the prevalence, associated factors and clinical significance of NNRTI hypersusceptibility (NNRTI-HS).
Methods: Patients without prior NNRTI therapy who were failing (HIV RNA >
400) stable antiretroviral therapy had ViroLogic phenotype assay, HIV RNA, and
CD4 cell counts before changing to a new NNRTI containing regimen at baseline. NNRTI HS was defined as a fold change in IC50 of less than 0.4 (patient
compared to reference control).
Results: At baseline, the 177 patients had a mean HIV RNA of 4.1 log10, CD4
cell count of 322 and a mean of 41 months of prior treatment. NNRTI-HS was
common: EFV-HS = 24%, DLV-HS = 17.5%, and NVP = 20%. EFV-HS was associated with duration of previous NRTI use (p < 0.001), number of nucleoside
agents (p = 0.002), use of ZDV (p = 0.04), and reduced susceptibility to ZDV
and ABC (fold change in IC50 > 5.0 compared to control, p< 0.005). The mean
change in HIV RNA (area based measure) 6 months after starting a new NNRTI
based regimen was greater in the 21 patients with NNRTI hypersusceptible virus
compared to the 77 patients without HS (1.2 versus 0.8 reduction in log10 HIV
RNA; p = 0.016). The difference persisted at month 12 (p = 0.023). Multiple linear regression models confirmed that HS to NNRTI was a significant independent
predictor of the magnitude of early HIV RNA reduction (up to month 4), after accounting for the baseline HIV RNA and the number of newly prescribed drugs
to which the patient's virus was susceptible (p < 0.02). CD4 cell increases were
also greater for patients with HS virus. Between months 4 to 12, the average additional CD4 gain in those with NNRTI HS was 28- 60, differences that trended
toward significance at months 4, 6, and 10.
Conclusion NNRTI HS occurred in about 20% of nucleoside experienced patients and, in those who initiated a NNRTI, appeared to result in greater reduction
of HIV RNA and CD4 cell gain for up to one year.
Presenting author: Richard Haubrich, 150 W. Washington, Suite 100, San
Diego, CA, 92103, United States, Tel.: +[phone], Fax: +[phone],
E-mail: [email protected]

ABCvPT<3
(N=152)
Wk 8 74%
Wk 16 70%

WT
(N=67)
74%
75%

M184V only
(N=31)
80%
58%

1-2 TAMs*
(N=43)
74%
79%

3-4 TAMs*
(N=35)
44%
36%

>5 TAMs*
(N=4)
0%
0%

Conclusions: ABC-based therapy, optimized by GT and vPT results, was effective for experienced subjects in clinical practice.
Presenting author: Siegrid Hessenthaler, GlaxoSmithKline, 5 Moore Drive, Research Triangle Park, NC 27709-3398, United States, Tel.: [phone], Fax:
[phone], E-mail: [email protected]