Efficacy and effect on lipid profiles of switching to ainuovirine-based regimen versus continuing efavirenz-based regimen in people with HIV-1: 24-week results from a real-world, retrospective, multi-center cohort study

ABSTRACT Ainuovirine (ANV), a novel non-nucleoside reverse-transcriptase inhibitor (NNRTI), was approved in China in 2021. In a previous randomized phase 3 trial, ANV demonstrated non-inferior efficacy relative to efavirenz (EFV) and was associated with lower rates of dyslipidemia. In this study, we aimed to explore lipid changes in treatment-experienced people with human immunodeficiency virus (HIV)-1 (PWH) switching to ANV from EFV in real world. At week 24, 96.65% of patients in the ANV group and 93.25% in the EFV group had HIV-1 RNA levels below the limit of quantification (LOQ). Median changes from baseline in CD4 +T cell counts (37.0 vs 36.0 cells/µL, P = 0.886) and CD4+/CD8 +ratio (0.03 vs 0.10, P = 0.360) were similar between the two groups. The ANV group was superior to the EFV group in mean changes in total cholesterol (TC, −0.06 vs 0.26 mmol/L, P = 0.006), triglyceride (TG, −0.6 vs 0.14 mmol/L, P < 0.001), high-density lipoprotein cholesterol (HDL-C, 0.09 vs 0.08 mmol/L, P = 0.006), and low-density lipoprotein cholesterol (LDL-C, −0.18 vs 0.29 mmol/L, P < 0.001) at week 24. We also observed that a higher proportion of patients demonstrated improved TC (13.55% vs 4.45%, P = 0.015) or LDL-C (12.93% vs 6.89%, P = 0.017), and a lower proportion of patients showed worsened LDL-C (5.57% vs 13.52%, P = 0.017) with ANV than with EFV at week 24. In conclusion, we observed good efficacy and favorable changes in lipids in switching to ANV from EFV in treatment-experienced PWH in real world, indicating a promising switching option for PWH who may be more prone to metabolic or cardiovascular diseases.

profiles (10).Furthermore, dyslipidemia is a major long-term adverse effect of current antiretroviral therapy (ART) regimens.Several studies have shown that ART is associated with dyslipidemia (11)(12)(13), and distinct ART regimens appear to promote different modifications in lipid metabolism (14).In people with HIV, ART regimens showed a 63.8% prevalence of dyslipidemia (15).In short, the prevalence of dyslipide mia, whether genetically determined or influenced by ART regimen or HIV infection, is consistently higher in HIV-infected groups (16).Dyslipidemia is characterized by elevated levels of total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), triglycerides (TG), or decreased high-density lipoprotein cholesterol (HDL-C) levels (17).LDL-C is an important causal risk factor for atherosclerotic CVD (18).Moreover, dyslipidemia is usually associated with hypertension, type 2 diabetes, and metabolic syndrome (19,20).It is important to monitor, evaluate, and manage dyslipidemia in people with HIV.
Efavirenz (EFV), a non-nucleoside reverse-transcriptase inhibitor (NNRTI), is one of the most commonly prescribed antiretroviral medications worldwide (21) and is included in the World Health Organization's preferred first-line ART regimens for people with HIV (22).EFV is also widely used in China since its inclusion as a nationally approved free anti-AIDS drug (23).A Chinese retrospective study including 7,623 people with HIV showed that EFV-based regimens were the most commonly selected first-line ART, accounting for 93.1% of the study population (24).Ainuovirine (ANV), a new generation of NNRTI independently developed in China, was approved in 2021.A randomized phase 3 trial indicated that the antiviral efficacy of ANV was non-inferior to EFV when combined with two nucleoside reverse transcriptase inhibitors (NRTIs), and treatment-emergent adverse events (TEAEs) in ANV-treated participants were less frequent regarding liver toxicity, dyslipidemia, neuropsychiatric symptoms, and rash compared with the EFV group (25).Given the strict inclusion criteria in the phase 3 trial, it was important to further understand the antiviral efficacy and effects on the lipid profile of ANV in real-world patients.

Study design
A real-world, retrospective, multi-center controlled cohort study was conducted to evaluate the antiviral efficacy and effect on the lipid profile of switching to ANV+2 NRTIs in people with HIV-1 who were previously treated with EFV+2 NRTIs.The study was conducted at six sites in China namely: Jinan, Wuhan, Changsha, Guangzhou, Guiyang, and Kunming (Table S1).
Eligible participants were diagnosed with HIV-1, aged 18 years or older, had received antiretroviral therapy, and achieved virological suppression, which was defined as an HIV-1 RNA level below the limit of quantification (LOQ) (see Table S2 for LOQ criteria).In addition, the study also included patients with HIV-1 RNA levels above the LOQ who were considered appropriate for switching in the ANV group and without treatment failure in the EFV group at the physician's discretion.Participants with severe metabolic abnormalities, cardiovascular diseases, and major neurological and psychiatric diseases were excluded.Women who were pregnant or breastfeeding and women of childbearing age or whose partners were unable to take effective contraception measures were not included either.
The ANV group was selected through the HIV real-world research platform (i-Study), of which the participants all signed the informed consent form.This group of patients had been receiving the EFV-based regimen, which included lamivudine (3TC) and tenofovir disoproxil fumarate (TDF) combined with EFV.Then, these patients switched to the ANV-based regimen, which was a once daily triple-drug therapy, either included with three drugs of ANV (75 mg/tablet ×2 tablets) +3 TC (300 mg ×1 tablet) +TDF (300 mg ×1 tablet) or a fixed-dose compound tablet of ANV 150 mg/3TC 300 mg/TDF 300 mg (Ainuovirine, Lamivudine and Tenofovir Disoproxil Fumarate Tablets, provided by Aidea Pharma, Jiangsu, China) and continued to take the regimen for at least 24 weeks.Drug switching was due to intolerance to the EFV-based regimen or at the physician's discretion.The EFV group was screened from the National Free Antiretroviral Treatment Program Database in China.This group of patients had been receiving the EFV-based regimen, a once daily triple-drug therapy including EFV 600 mg or 400 mg (×1 tablet) +3 TC 300 mg (×1 tablet) +TDF 300 mg (×1 tablet) and continued to take it for at least 24 weeks from baseline.

Assessment
Patient data were collected from medical records.All participants were required to have a lipid profile (TC, TG, HDL-C, LDL-C) determined at baseline and at 12 and 24 weeks.Other laboratory test results, such as HIV-1 RNA level, CD4 + T-cell count, and CD4 + / CD8 + ratio, were also collected.Fasting lipid profile, CD4 + , and CD8 + T-cell counts were measured by Cobas C702 Chemistry Analyzer (Roche Diagnostics, Mannheim, Germany) at all sites, following the manufacturer's recommended procedures.HIV-1 RNA levels were measured using NEPG Automatic HIV-1 RNA Quantitative Test Kit (Northeast Pharm Group Liaoning Biomedicine Co., LTD, Liaoning, China; LOQ = 100 copies/mL) at Wuhan Jinyintan Hospital, and Xpert HIV-1 Viral Load assay (Cepheid, California, USA; LOQ = 20 copies/mL) was used at other sites.All other clinical measurements were routinely conducted in the Clinical Laboratory Department at each site.An abnormal lipid profile was defined according to the criteria outlined in the 2023 Chinese Guidelines for Lipid Management, with thresholds set as follows: TC ≥5.2 mmol/L, TG ≥1.7 mmol/L, HDL-C <1.0 mmol/L, or LDL-C ≥3.4 mmol/L (26).
The primary endpoints were efficacy on viral suppression (HIV-1 RNA levels below the LOQ, change in the CD4 + T-cell count and CD4 + /CD8 + ratio) and changes in lipid profile [TC, TG, HDL-C, LDL-C, and log (TG/HDL-C)] at week 24.The secondary endpoints were efficacy on viral suppression and changes in lipid profile at week 12.

Statistical analysis
Propensity score weighting using overlap weights was proven to be an appropriate and feasible method to adjust for confounding factors (27).In this study, baseline information, such as age, sex, antiviral treatment duration, WHO stage, comorbidities (hypertension, diabetes, etc.), CD4 + T-cell count, and body weight (BMI) were used to calculate the propensity scores.For all variables, overlap weights were calculated based on propensity scores for weighted analysis.Continuous variables are presented as the mean (standard deviation), and an independent samples t test was used for intergroup comparisons.Categorical variables are reported as numbers (percentages), and the χ 2 test was used for intergroup comparisons.The mean and 95% confidence intervals (CIs) were used to summarize the lipid profile changes from baseline at weeks 12 and 24, and the analysis of covariance was used to adjust the baseline levels for intergroup comparisons.All analyses were performed using R 4.2.2 and SAS Version 9.4.

Patient baseline characteristics
We retrospectively identified 318 patients eligible for the ANV group and 460 patients for the EFV group.Table 1 summarizes the unweighted and overlap weighted patient baseline characteristics and the standardized mean differences (SMDs) for each variable.After overlap weighting, the baseline characteristics between the two groups were well balanced with SMD <0.1 (Table 1).Both groups had 157 patients, the mean age was 40, 90.9% of them were male, and 8.0% of the study population had comorbidities (i.e., hypertension, diabetes).The average duration of antiviral treatment was 34.8 weeks in both groups.At baseline, the proportion of HIV RNA levels below the LOQ was 88.9% in the ANV group and 94.3% in the EFV group (P = 0.012).The mean CD4 + T-cell count in both groups was 510.8 cells/µL (P = 1.000), and the mean CD4 + /CD8 + ratio was 0.68 in the ANV group and 0.69 in the EFV group (P = 0.655).

Efficacy
At week 24, there were numerically more patients with HIV-1 RNA levels below the limit of quantification (LOQ) in the ANV group (baseline 88.9% to 96.65%) than in the EFV group (baseline 94.3% to 93.25%), although there were no significant differences between groups (Table 2).In primary analysis, similar changes from baseline in CD4 + T-cell count and CD4 + /CD8 + ratio were observed in the two groups at week 24 (Table 2).The median increases from baseline in CD4 + T-cell count were 37.0 cells/µL [interquartile range (IQR), 3.0 to 99.0] in the ANV group and 36.0 cells/µL (IQR, −51.0 to 178.0) in the EFV group, with no significant differences between groups (P = 0.886).The median increases in the CD4 + / CD8 + ratio were 0.03 (IQR, −0.01 to 0.10) with ANV and 0.10 (IQR, −0.10 to 0.20) with EFV, and no significant differences were observed between the two groups (P = 0.360).These data demonstrated that the efficacy of switching to the ANV-based regimen was comparable to that of continuing EFV-based regimen at week 24.

Changes in lipid profile
At baseline, the mean TC, TG, and LDL-C levels in the ANV group were significantly higher than those in the EFV group, and the mean HDL-C level was significantly lower (Table 1).The proportion of patients with abnormal lipid profile was significantly higher in the ANV group than in the EFV group (Table 1); specifically, 24.4% of patients in the ANV group and 14.7% in the EFV group had abnormal TC (P = 0.002), 53.8% vs 39.3% for TG (P < 0.001), 46.2% vs 33.0% for HDL-C (P = 0.001), and 20.2% vs 12.4% for LDL-C (P = 0.006).This was consistent with some patients who had received the EFV-based regimen then switched to the ANV-based regimen due to dyslipidemia in real world.
Considering the baseline lipid profile imbalances between the treatment groups, we used analysis of covariance to adjust each patient's follow-up lipid levels for his or her baseline level, with the advantage of being unaffected by baseline differences.By using analysis of covariance, the baseline imbalances between groups were eliminated.Table 3 summarizes the mean changes from baseline in the lipid profile at week 24 by treatment and subgroup.In all patients, the differences in mean changes in the four lipid parame ters between the ANV and EFV groups were statistically significant (P < 0.05, Fig. 1).By week 24, the mean (95% CI) changes in TC were −0.06 mmol/L (−0.17 to 0.06) for ANV and 0.26 mmol/L (0.18 to 0.35) for EFV (P = 0.006).TG levels were decreased with ANV (−0.60 mmol/L; 95% CI, −0.79 to −0.40) and increased with EFV (0.14 mmol/L; 95% CI, 0.00 to 0.28; P < 0.001).The increases in HDL-C with ANV (0.09 mmol/L; 95% CI, 0.04 to 0.13) were significantly greater than those with EFV (0.08 mmol/L; 95% CI, 0.02 to 0.14; P = 0.006).The LDL-C decrease was −0.18 mmol/L (−0.27 to −0.09) for ANV, compared with an increase of 0.29 mmol/L (0.19 to 0.40) for EFV (P < 0.001).Log (TG/HDL-C) was decreased from baseline both in the ANV (−0.11; 95% CI, −0.15 to −0.07) and EFV (−0.01; 95% CI, −0.03 to 0.02) groups, with no significant differences between groups (P = 0.063).Figure 2 presents the proportion of patients with worsened, improved, and unchanged lipid profiles from baseline at week 24 among all patients.A higher proportion of patients in the ANV group had improved TC (13.55% vs 4.45%, P = 0.015) or LDL-C (12.93% vs 6.89%, P = 0.017) levels at week 24 than those in the EFV group.A numerically higher proportion of patients had their TG (19.61% vs 12.0%, P = 0.076) improved with ANV than with EFV, although with no significant differences.The proportion of patients with improved HDL-C was numerically lower in the ANV group than in the EFV group (17.69% vs 20.23%), with no significant differences between  c Patients who reported no use of lipid-lowering drugs at baseline, week 12, and week 24 were considered to have never taken lipid-lowering drugs.
groups (P = 0.821).The results were similar at week 12, except with no significant differences in the proportion of patients with improved TC between the ANV and EFV group (13.85% vs 5.83%, P = 0.054) (Fig. S1).Patients were stratified according to whether they had dyslipidemia at baseline or whether they were using lipid-lowering drugs and were divided into three subgroups: dyslipidemia at baseline (subgroup A), never taken lipid-lowering drugs (subgroup B), and dyslipidemia at baseline and never taken lipid-lowering drugs (subgroup C).In subgroup analyses, the mean changes in TC, TG, LDL-C, and log (TG/HDL-C) were more favorable with ANV than with EFV at week 24 for each of the three subgroups, although the mean changes in HDL-C favored EFV (Table 3).In subgroup A of dyslipidemia at baseline, the mean (95% CI) changes in TC were −0.08 mmol/L (−0.22 to 0.06) for ANV and 0.20 mmol/L (0.08 to 0.33) for EFV (P = 0.02).The decrease in TG was −0.82 mmol/L (−1.06 to −0.58) with ANV, compared with an increase of 0.03 mmol/L (−0.18 to 0.25) with EFV (P = 0.001).The increases in HDL-C levels were 0.10 mmol/L (0.06 to 0.15) with ANV and 0.15 mmol/L (0.09 to 0.22) with EFV (P = 0.009).LDL-C levels decreased with ANV (−0.18 mmol/L; 95% CI, −0.30 to −0.07) but increased with EFV (0.26 mmol/L; 95% CI, 0.12 to 0.40; P < 0.001).The decreases in log (TG/HDL-C) were −0.15 (−0.19 to −0.11) with ANV and −0.06 (−0.10 to −0.03) with EFV, with no significant differences (P = 0.162) between the two treatment groups.In subgroups B and C, which excluded patients on lipid-lowering medications to rule out the effects of lipid-lowering drugs on lipids, similar changes in lipid profiles were observed as in subgroup A, except that there were no significant differences (P = 0.274) between treatment groups for the mean changes in TC in subgroup C (Table 3).

BMI changes
At week 24, there were no significant differences in mean (SD) BMI compared to baseline in either the ANV (P = 0.891) or EFV group (P = 0.698), and there were no significant differences in BMI distribution for either group (Table S3).The same results were observed at week 12 (Table S3).These data showed that the ANV-based regimen had no significant effect on body weight (BMI).

DISCUSSION
To the best of our knowledge, this is the first real-world study to compare the efficacy and safety of switching to ANV from EFV with continuing EFV in treatment-experienced people with HIV-1.Our findings indicate that switching to ANV-based regimen resulted in considerable efficacy and favorable changes in lipid profile.
Throughout the 24-week study period, participants who switched to ANV-based regimen showed maintenance of virologic suppression, with no significant differences compared with the EFV group.Immune restoration, as measured by median changes from baseline in CD4 + T-cell counts and CD4 + /CD8 + ratio, was similar between the ANV and EFV groups.The efficacy of ANV was comparable to that of EFV, which was consistent with the results of a previous randomized phase 3 trial (25).
Many antiretroviral regimens, including EFV, are related to adverse effects in serum lipids.This often necessitates lipid-lowering treatment (28,29).Multiple studies have shown that EFV demonstrate more negative changes in TC, TG, and LDL-C than elvitegravir (EVG) (30)(31)(32), raltegravir (RAL) (33,34), dolutegravir (DTG) (35), nevirapine (NVP) (36), and rilpivirine (RPV) (37) when used in people with HIV-1.According to the DRIVE-AHEAD trial, the EFV group exhibited higher mean increases in fasting LDL-C and non-HDL-C at week 96 compared with week 48 (38,39), suggesting that the chance of developing dyslipidemia increases with prolonged use of EFV.In line with the previous studies, our data showed that participants who continued the EFV-based regimen for 24 weeks exhibited significant increases in TC, TG, and LDL-C (Fig. 1B).
The association of adverse impact on serum lipids with certain antiretroviral therapies has prompted a strategy of switching the potentially offending agent for an alternative regimen.According to the European AIDS Clinical Society (EACS) guideline 2022, in the event of adverse effects, it is recommended to switch within the same drug class if the alternative has equal potency and there is no evidence of resistance (40).Recent studies indicated that newer NNRTIs, such as doravirine (DOR), have more favorable impact on lipids (39,41).Other studies on switching therapy have found that switching to DOR demonstrates favorable changes in lipid profiles.In the DRIVE-SHIFT Trial, participants who switched to DOR-based regimen from ritonavir-boosted protease inhibitor had significantly greater reductions in fasting LDL-C and non-HDL-C at week 24 (42).A retrospective multi-center Italian study demonstrated a significant reduction in both cholesterol and triglyceride levels 24 weeks after switching to DOR-containing/-based regimens (43).In the present study, patients who switched to ANV, another novel NNRTI, showed decreases in mean TC, TG, and LDL-C both at weeks 12 and 24.In contrast, patients continuing with EFV experienced increases in these lipid parameters (Fig. 1).Subgroup analyses confirmed that after ruling out the confounding effects of baseline lipid levels and the use of lipid-lowering drugs, ANV was still superior to EFV for changes in TC, TG, and LDL-C (Table 3).The atherogenic index of plasma (AIP), one of the predictors for coronary atherosclerosis risk, is calculated as Log (TG/HDL-C) (44).Log (TG/HDL-C) at baseline was higher in the ANV group than in the EFV group (Table 1) and decreased numerically more with ANV than with EFV at week 24 (Table 3).Similar results were observed in all three subgroups.Moreover, a significantly higher proportion of patients had improved TC and LDL-C in the ANV group than in the EFV group at week 24 (Fig. 2).Since higher LDL-C levels are associated with an increased risk of CVD and "lower is better for longer" (45) the favorable changes in LDL-C in participants who switched to ANV from EFV could have a beneficial effect on CVD prevention.
Weight gain after ART initiation is common in people with HIV (46).A pool analysis of 8 randomized controlled clinical trials, including 5,680 people with HIV-initiating ART between 2003 and 2015, reported that 48.6%, 36.6%, and 17.3% of participants had a weight gain of 3%, 5%, and 10% from baseline, respectively (47).And the weight gain was greater among those taking integrase strand transfer inhibitors (INSTIs) than NNRTIs and PIs (47).This study found that ANV had little effect on body weight (BMI) and would not cause weight gain.
Several limitations to our study should be noted.The first is the limitation inher ent to retrospective analysis in real clinical setting, including the risk of confounding bias and selection bias, which were inevitable.Since patients were not randomly assigned, decisions of switching from EFV to ANV were made upon the attending physician's discretion and potentially had a risk of confounding variables.Furthermore, the follow-up period was only 24 weeks, which could be inadequate to determine the effectiveness of switching to ANV, as improvement in lipid profile and CVD risk need long-term evaluation.We will retain the cohort for long-term follow-up, and the results of 48 weeks or even longer will be presented later.Third, there may be individ ual disparities in atherosclerotic cardiovascular disease (ASCVD) risk score at baseline between the two groups, which could influence the outcomes but were not stratified in the current analysis.However, propensity score weighting using overlap weights was used to overcome this limitation by balancing the baseline characteristics between the two groups.Nevertheless, these findings should be interpreted with caution.Further randomized controlled trials with CVD risk stratification and longer follow-up to verify the lipid-friendly properties of ANV are needed, and the mechanisms of newer NNRTIs on lipid metabolism require further exploration (48).

Conclusions
This retrospective study suggested that in this scenario, ANV-based regimen showed comparable virologic/immunologic effect compared with EFV-based regimen.ANV exhibited favorable changes in lipid profiles in participants who switched from EFV.In addition, ANV did not cause weight gain.Overall, ANV is a promising switching option for patients who may have other risk factors for metabolic syndrome or cardiovascular diseases.

FIG 1
FIG1 Mean changes from baseline in the lipid profile at week 12 (A) and week 24 (B) in all patients.The results were obtained through weighted analysis of propensity scores.Analysis of covariance was used for intergroup comparisons.At week 12, the P values for intragroup comparison of changes in lipids in the ANV and EFV groups were 0.057 and 0.001 for TC, P < 0.001 and P < 0.001 for TG, P < 0.001 and P = 0.574 for HDL-C, and P < 0.001 and P < 0.001 for LDL-C, respectively.At week 24, the P values for intragroup comparison of changes in lipids in the ANV and EFV groups were 0.341 and <0.001 for TC, P < 0.001 and P = 0.047 for TG, P < 0.001 and P = 0.008 for HDL-C, and P < 0.001 and P < 0.001 for LDL-C, respectively.ANV, ainuovirine; EFV, efavirenz; HDL-C, high density lipoprotein cholesterol; LDL-C, low density lipoprotein cholesterol; TC, total cholesterol; TG, triglyceride.

FIG 2
FIG 2Proportion of patients with worsened, improved, and unchanged lipid profiles at week 24 among all patients.The results were obtained through weighted analysis of propensity scores.In the analyses, worsening was defined as the lipid level changing from normal at baseline to abnormal; improvement was defined as the lipid level changing from abnormal at baseline to normal; and the lipid level remaining normal or abnormal was defined as unchanged.ANV, ainuovirine; EFV, efavirenz; HDL-C, high density lipoprotein cholesterol; LDL-C, low density lipoprotein cholesterol; TC, total cholesterol; TG, triglyceride.

TABLE 2
Efficacy outcomes at week 24 a a ANV, ainuovirine; EFV, efavirenz; IQR, interquartile range; LOQ, limit of quantification; SD, standard deviation.The number of patients and missing data were unweighted, and the remaining results were obtained through weighted analysis on propensity score.Paired t test (week 24 vs baseline) was used for intragroup comparison, and analysis of covariance was used for intergroup comparison of changes from baseline.b P value is not valid for intergroup comparison.c P < 0.05 for intragroup comparison.

TABLE 3
Changes from baseline in the lipid profile at week 24 by treatment and subgroup a