Differences in Protease Inhibitor (PI) Phenotypic Susceptibility after Failure of
the First PI-Containing Regimen
R. Haubrich1, C. Kemper2, M. Witt3, P. Keiser4, M. Dube5, D. Forthal6, J. Currier7,
J. Hwang1, D. Richman1, N. Hellmann8, G. Heilek8, Y. Lie8, J.A. McCutchan1
1Univ. of California, San Diego: San Diego, CA; 2Santa Clara Valley Med. Ctr.: San
Jose, CA; 3Harbor- UCLA: Los Angeles, CA; 4Univ. of Texas: Dallas, TX; 5Univ. of
Southern California: Los Angeles, CA; 6Univ. of California, Irvine: Irvine, CA; 7Univ. of
California, Los Angeles: Los Angeles, CA; 8ViroLogic: South San Francisco, CA
Objective: To evaluate differences in protease inhibitor susceptibility after failure of the
first PI containing regimen. Design: Baseline PI susceptibility was determined in an
ongoing strategy trial of susceptibility testing (ViroLogic PhenoSenseTM assay) and
expressed as the fold change (FC) in IC50 from control. A > 2.5 FC was defined as
resistant for this analysis. Results: Of 82 patients (median CD4 = 281, median HIV
RNA = 4.1 log10, median months NRTI = 34, median months PI = 15), 13 were PI
naïve and had no reduction in susceptibility to PI. Forty-four had prior NFV, 16 IDV, 5
SQV, 3 RTV and 1 APV. In PI experienced patients, the mean baseline protease
inhibitor FC was significantly different between IDV, NFV, RTV, APV and SQV (4.9,
20, 7.0, 2.1, and 3.1 respectively; p < 0.001 by repeated measures ANOVA). The
percent of patients with drug susceptible virus (< 2.5 FC) was 83% for APV, 84% for
SQV, 71% for RTV, 68% for IDV and 26% for NFV. Compared to patients with other
prior PI, prior therapy with NFV was associated with infrequent cross resistance: APV
(7% vs 36%, p = 0.006), IDV (16% vs 60%, p < 0.001) and RTV (14% vs 56%, p <
0.001). Multiple LR models were constructed to predict individual PI resistance based
on baseline values (CD4, HIV RNA, prior therapy, and AIDS diagnosis). For the best
model (NFV, p = 0.0001), correct prediction of resistance was obtained for 81% of
cases. 11% (2/18) of patients predicted to be NFV-sensitive were actually resistant
while 22% (14/64) predicted to be NFV-resistant were actually sensitive. Conclusion:
In this population treated predominately with NFV, viral isolates remained susceptible to
APV and SQV in the majority of cases. NFV, used as the first PI, was associated with
less cross resistance than other prior PIs. LR modeling (using prior history) did not
accurately predict PI resistance. |
