Increased Resilience to the Development of Drug Resistance with Modern Boosted Protease Inhibitor. Part 3
Of the 2350 individuals included in this analysis, at least 1 available pretherapy genotype was available for 1426 (61%). Primary resistance was assessed using a standardized list of mutations suitable for transmitted resistance surveillance. From the earliest available pretherapy genotypes, 131 (9.2 %) showed some evidence of transmitted resistance. A sensitivity analysis was conducted that eliminated those individuals who were known to have transmitted resistance, leaving 1295 patients.
For each individual, the median number of genotypes that could possibly have been included in the analysis for which data was unavailable was 0 (interquartile range [IQR], 0–1) and the median number completed was 1 (IQR, 0–2). A total of 1433 (61.0%) of individuals had all possible genotypes completed.
Outcome measures and predictor variables.The primary outcome in this analysis was the emergence of drug resistance in any of the four resistance categories described previously (yes vs. no). The following baseline predictor variables were investigated: age, sex, CD4 cell count, pVL (log10 transformed), first regimen, AIDS diagnosis, history of injection drug use, year of first therapy, and adherence. Estimates of adherence to ART were based on medications actually dispensed, not prescribed. For this study, we limited our measure of adherence to the first year of therapy, estimated by dividing the number of months of medications dispensed by the number of months of follow‐up. This measure of adherence has been found to be independently associated with HIV suppression and survival among HIV‐infected persons enrolled in the BC‐CfE DTP. Adherence was categorized as 0% to <40%, 40% to <80%, 80% to <95% and 95%. Because pVL was measured over time starting in 1996, our baseline pVLs were obtained on the basis of the standard pVL assay, and our last measurements were obtained on the basis of the ultrasensitive pVL assay; thus, our upper and lower limits of pVLs ranged over time from 500 and to 50 and copies/mL. Therefore, our pVL measurements were recoded to range from 500 to copies/mL in order to standardize the viral load range over time. Statistical analyses.An exploratory logistic regression model was developed for identifying which patient characteristics were most influential in the development of drug resistance during ART. A backward stepwise technique was used in the selection of covariates. The area under the receiver operating characteristic curve was used to measure the model’s ability to discriminate between those in whom resistance developed versus those in whom it did not. Categorical variables were compared using the χ2 or Fisher’s exact test, and continuous variables were compared using the Wilcoxon rank‐sum test. For the purposes of analysis, we followed the intent‐to‐treat principle, with subjects retained in their initial treatment groups irrespective of whether participants subsequently switched to regimens that were available later. This approach provides a conservative estimate of the true treatment effect. All analyses were performed using SAS software (version 9.1.3, service pack 3).