A total of 100 samples were included in the study. Due to amplification failure in 9 samples and amplification success in only one compartment in 15 samples, we were able to analyze 76 pairs of sequences obtained from both plasma RNA and PBMC DNA compartments. Among the 15 samples successfully amplified only from one compartment, 12 of them were from PBMC DNA and the remaining 3 were from plasma RNA, and their VL ranged from 3,000 to 7,600 copies/ml. Therefore, PBMCs had a higher sequencing rate, 97% (95 CI: 93.5%-100%) than plasma, 87% (95% CI: 80%-94%).
Although PBMCs sequencing was performed in Sweden and plasma RNA sequencing was generated in CDC laboratory in Atlanta, USA, analysis of all the sequence raw data was performed in CDC laboratory. The quality of sequences generated using PBMC DNA were overall comparable to those generated from plasma RNA. Some of the detected mutations did not induce resistance to the regimens in use at testing time. Phylogenetic analyses further showed that plasma-PBMC pair sequences were highly related, which argues against contamination, mislabeling and/or introduction of artifacts that could negatively impact comparison of the sequences. Phylogenetic analysis revealed that all newly obtained sequences were HIV-1 subtype B (Data not shown). All sequences were submitted to GenBank and the accession numbers are JN215213-JN215364.
In all the analyses, we defined concordance as concordant drug resistance profile in any pair of sequences from a patient’s plasma RNA and PBMC DNA since mutations could be slightly different between the two compartments and drug resistance profile would still be the same as not all the mutations identified induce clinically significant drug resistance. Discordance means discordant drug resistance profiles. These analyses were carried out by using the Sensitive, Intermediate, Resistant (SIR) alternative in the Stanford HIValg program.
summarizes the demographic and clinical characteristics of the 76 patients analyzed in the study and stratified according to concordant and discordant drug resistance profiles in paired PBMC and plasma samples.
Demographic and clinical characteristics of the 76 patients analyzed in the study (data were stratified according to concordant and discordant drug resistance profiles in paired peripheral blood mononuclear cell DNA and plasma RNA)
Eighty-eight percent (95% CI: 80.3-94.5) of the samples (N=67) had concordant drug resistance profiles between the two compartments. Among them, 70% (95% CI 59.6-79.8) of the samples (N=53) had either identical drug resistance mutations (N=45) or no drug resistance mutation identified (N=8). Amongst the eight patients without drug resistance mutation, three had poor adherence history (missed three or more drug doses in the last month), while the remaining five had good adherence history (missed less than 3 doses of drug in the last month). Although the remaining 18% samples (N=14) had concordant drug resistance profiles, drug resistance mutations identified were slightly different between the two compartments. For instance, eight samples had one drug resistance mutation (L33FL, K138N, E138EK and A62AV) which was different between the plasma and PBMC samples in their entire drug resistance mutation profiles, while the remaining six specimens contained mutations that changed the predicted level of drug resistance from intermediate to resistant level or vice versa between the two compartments. However, no drug was changed from susceptible to intermediate or resistant level ().
Summary of mutations detected in the 67 specimens defined as concordant drug resistance profiles.
Discordant drug resistance profiles were observed in nine (12%, 95% CI: 6.5-21.3) samples. presents a detailed description of the drug resistance mutations in both compartments for these nine specimens. Five of these samples (HN-12, HN-25, HN-28, HN-39, and HN-72) had mutations identified in plasma, not found in PBMCs while the remaining four had mutations in PBMCs which were not present in plasma. These differences in mutation profiles led to differences in drug resistance profiles between the plasma and PBMC compartments that could be of clinical significance.
Specimens with discordant drug resistance profiles between plasma RNA and peripheral blood mononuclear cell DNAa
Bivariate logistic regression analyses indicated that patients on triple drug therapy were 17.24 times more likely to have concordant drug resistance profiles than those on less than triple drug therapies (OR=17.24, 95% CI: 3.48- 83.33, P=0.0005) while those patients with increasing number of years on ART were less likely to have concordant drug resistance profiles than those with less number of years on ART (OR=0.62, 95% CI: 0.43-0.88, P=0.0081). An association was not identified when other factors were analyzed, including age, sex, CD4 count change in an increment of 100 cells/μL, number of regimens, number of PI, NRTI and NNRTI mutations, transmission route (sexual contact vs. mother to child transmission, MTCT) low, intermediate and high VL levels ().
Bivariate Logistic Regression Models to investigate association with different mutation profiles in PBMCs and RNA
When multivariate logistic regression analysis was used to assess the association of triple drug therapy at first ART with concordant drug resistance profile, the result remained significant (OR=15.36, 95% CI: 1.79-142.86, P=0.013, ).
Multivariate analysis of factors that may be associated with concordant or discordant drug resistance profiles in PMBCs-DNA and plasma RNA