Background:  Phenotypic resistance to ATV, and other antiretrovirals (ARV), was assessed in 40 patients undergoing screening for protocol PACTG 1020A. 

Methods:  Phenotypic resistance scores (PRS) were calculated as the log₁₀ of the ratio of patient vs wild type IC₅₀s. 36/40 patients had genotypic testing.  For analyses dealing with cross-resistance, these genotypic data were summarized into genotypic resistance scores (GRS) for each drug, using an algorithm. An alpha level of 0.01 was used as the criterion for statistical significance.

Results:  The geometric mean of PRS to ATV was 5.6, with 52.5% of the sample showing scores >5.  Correlations between the PRS of ATV and both the PRS and GRS of the other ARV were as follows:

                                      Medication:

                                  3TC   ABC   AMP  D4T  DDI  EFV  IDV  NFV  NVP  RTV  SQV  ZDV

Phenotypic(n=40):     .03     .36      .82*   .41*  .24    .14    .87*   .82*   .25     .88*   .88*   .37

Genotypic(n=36):     -.05     .56*    .77*   .47*  .40    .12    .80*   .78*   .24     .80*   .77*   .56*

    (*=p<0.01)

A multiple regression analysis (backwards selection), with PRS to ATV as the outcome variable and the PRS of the other ARVs with which it was significantly correlated as the predictors, showed that phenotypic resistance to APV, NFV, RTV, and SQV all made unique and significant contributions towards predicting the extent of phenotypic resistance to ATV (r²=0.94).  A similar regression, with predictors consisting of the GRSs which were significantly correlated with ATV, indicated that genotypic resistance to NFV and IDV made unique and significant contributions towards predicting phenotypic resistance to ATV (r²=0.70).

Exploratory analyses testing whether ATV phenotypic resistance varied as a function of the presence or absence of mutations identified 13 significant positions (p<0.01).  A regression model with the ATV PRS as the outcome revealed that mutations at PI-54, PI-84 and RT-215 (r² =0.76) made unique and significant contributions towards predicting phenotypic resistance to ATV.

Conclusions:  These results indicate that heavily treated pediatric patients exhibited relatively high levels of resistance to ATV.  Cross-resistance with other protease inhibitors (PI) was such that 94% of the variance in ATV PRS could be predicted on the basis of cross-resistance with other PI.  These findings should be interpreted with caution, since they come from exploratory analyses involving multiple comparisons, where some results may have been due to chance.