Therapy-Emergent Drug Resistance to Integrase Strand Transfer Inhibitors in HIV-1 Patients: A Subgroup Meta-Analysis of Clinical Trials

Abstract

Background: Integrase strand transfer inhibitors (INSTIs) are a novel class of anti-HIV agents that show high activity in inhibiting HIV-1 replication. Currently, licensed INSTIs include raltegravir (RAL), elvitegravir (EVG) and dolutegravir (DTG); these drugs have played a critical role in AIDS therapy, serving as additional weapons in the arsenal for treating patients infected with HIV-1. To date, long-term data regarding clinical experience with INSTI use and the emergence of resistance remain scarce. However, the literature is likely now sufficiently comprehensive to warrant a meta-analysis of resistance to INSTIs.

Methods: Our team implemented a manuscript retrieval protocol using Medical Subject Headings (MeSH) via the Web of Science, MEDLINE, EMBASE, and Cochrane Central Register of Controlled Trials databases. We screened the literature based on inclusion and exclusion criteria and then performed a quality analysis and evaluation using RevMan software, Stata software, and the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE). We also performed a subgroup analysis. Finally, we calculated resistance rates and risk ratios (RRs) for the three types of drugs.

Results: We identified 26 references via the database search. A meta-analysis of the RAL data revealed that the resistance rate was 3.9% (95% CI = 2.9%-4.9%) for the selected randomized controlled trials (RCTs). However, the RAL resistance rate reached 40.9% (95% CI = 8.8%-72.9%) for the selected observational studies (OBSs). The rates of resistance to RAL that were associated with HIV subtypes A, B, and C as well as with more complex subtypes were 0.1% (95% CI = -0.7%-0.9%), 2.5% (95% CI = 0.5%-4.5%), 4.6% (95% CI = 2.7%-6.6%) and 2.2% (95% CI = 0.7%-3.7%), respectively. The rates of resistance to EVG and DTG were 1.2% (95% CI = 0.2%-2.2%) and 0.1% (95% CI = -0.2%-0.5%), respectively. Furthermore, we found that the RRs for antiviral resistance were 0.414 (95% CI = 0.210-0.816) between DTG and RAL and 0.499 (95% CI = 0.255-0.977) between EVG and RAL. When RAL was separately co-administered with nuclear nucleoside reverse transcriptase inhibitors (NRTIs) or protease inhibitors (PIs), the rates of resistance to RAL were 0.2% (95% CI = -0.1%-0.5%) and 0.2% (95% CI = -0.2%-0.6%), respectively. The ten major integrase mutations (including N155H, Y143C/R, Q148H/R, Y143Y/H, L74L/M, E92Q, E138E/A, Y143C, Q148Q and Y143S) can reduce the sensitivity of RAL and EVG. The resistance of DTG is mainly shown in 13 integrase mutations (including T97T/A, E138E/D, V151V/I, N155H, Q148, Y143C/H/R, T66A and E92Q).

Conclusions: Our results reveal that the DTG resistance rate was lower than the RAL resistance rate in a head-to-head comparison. Moreover, we confirmed that the EVG resistance rate was lower than the RAL resistance rate. In addition, our results revealed that the resistance rate of RAL was lower than that of efavirenz. The rates of resistance to RAL, EVG and DTG were specifically 3.9%, 1.2% and 0.1%, respectively. Compared with other types of antiviral drugs, the rates of resistance to INSTIs are generally lower. Unfortunately, the EVG and DTG resistance rates could not be compared because of a lack of data.

Fig 1. PRISMA diagram of the literature search.

The PRISMA diagram illustrates the process through which literature was filtered, according to the designated inclusion and exclusion criteria. At each step, the reason for exclusion is indicated, where “n” represents the number of papers.

Fig 2. Quality evaluation of RCTs using RevMan software.

In each dimension, the area of different colors represents the proportion of different publication biases derived from the included literatures. When no clear answer could be obtained for a dimension, it was classified as presenting a high risk of bias.

Fig 3. Histogram of quality evaluation for observational studies using STROBE.

Different colors represent different dimensions in STROBE. The different regions on the x-axis illustrate different authors and publication dates. Each dimension was independently scored. “Sources” indicates the total points obtained for the quality evaluation process.

Fig 4. Funnel plot of the publication bias associated with the OBSs, as generated using Stata.

Fig 5. Forest plot for the rate of resistance to RAL, as determined using Stata.

RCTs and OBSs formed the basis of this classification. “OBS” indicates the observational nature of the study. ES denotes the effect value (i.e., resistance rate). The important indicator I² was used to evaluate the heterogeneity of the data. A hollow diamond represents the result of the meta-analysis. “n” indicates the different trial numbers for a given piece of literature. A black diamond represents the resistance rate for each trial. The width of the horizontal line passing through the black diamond denotes the 95% CI. The meta-analysis was completed using a random-effects model.

Fig 6. Forest plot for the rate of resistance to EVG, as determined using Stata.

Fig 7. Forest plot for the rate of resistance to DTG, as determined using Stata.

Fig 8. Forest plot for the RR between DTG and RAL, as determined using Stata.

The RR was obtained using the following formula: DTG resistance rate divided by the RAL resistance rate. The data calculation was performed according the Mantel-Haenszel method.

Fig 9. Forest plot for the RR between EVG and RAL, as determined using Stata.

Fig 10. Forest plot for the RR between RAL and efavirenz, as determined using Stata.

Fig 11. Forest plot for the rate of resistance to RAL, based on therapeutic time subgroup analysis.

Fig 12. Forest plot for the rate of resistance to RAL, based on HIV-1 subtypes subgroup analysis.

The complex subtype contains the all subtypes except for the A, B and C subtypes. When the same data contained different subtypes, it was assigned to different subtypes according to weight.

Fig 13. Forest plot for the rate of resistance to RAL based on cross-resistance subgroup analysis.