As a second-generation STR, B/F/TAF is recommended as a preferable ART regimen for PLWH [22]. B/F/TAF, unlike E/C/F/TAF, seems to have little effect on blood lipids [23]. However, longitudinal studies comparing the effects of B/F/TAF and E/C/F/TAF on blood lipids are scarce. In this prospective study, we investigate changes in the conventional blood lipid and lipidomic profiles of PLWH treated with E/C/F/TAF and B/F/TAF. For conventional blood lipids analysis, the levels of TG and VLDL at week 24 were better in the B/F/TAF group, while the levels of TC, LDL-C, and HDL-C, as well as the TC/HDL-C ratio at week 24 were not significantly different between the two groups. Additionally, FRS decreased after switching to the B/F/TAF regimen.

Our non-targeted lipidomic approach revealed that E/C/F/TAF exerted a particular effect on glycerolipids and glycerophospholipids, particularly the upregulation of DGs, TAGs, and LPCs, which are associated with FRS. In the B/F/TAF group, significant increases were observed for sphingolipids and glycerophospholipids, which were found to be inversely related to CVD risk. Finally, our fully adjusted analyses identified seven lipid species, i.e., DG (16:0/18:2), DG (18:2/22:6), DG (18:3/18:2), DG (20:5/18:2), TAG (18:3/18:2/21:5), TAG (20:5/18:2/22:6), and LPC (22:6), that were significantly associated with future CVD risk, independent of clinical factors.

Previous studies have demonstrated that E/C/F/TAF is associated with increased TG, TC, LDL-C, and HDL-C [24] and that switching from E/C/F/TAF to B/F/TAF is strongly related to blood lipid improvement [12]. However, this blood lipid improvement was only observed in a cohort of women living with HIV. It was well known that blood lipid levels vary significantly between genders and ethnicities. To the best of our knowledge, this study is the first prospective cohort to observe changes in blood lipid levels and lipidomic profiles after switching from E/C/F/TAF to B/F/TAF in an Asian population.

In this study, it was observed that switching to B/F/TAF is associated with lower TG and VLDL-C levels, which is in line with previous reports [12]. It is important to note, however, that LDL-C, which is considered a conventional cardiovascular risk factor, was decreased in the E/C/F/TAF group in our study, which is different from the results of a previous study [10]. This could be explained by the short follow-up duration (24 weeks), low median age (32.00, IQR 27.00–37.00), and small sample size (44 patients). Furthermore, it has been demonstrated that conventional blood lipid analysis including that of LDL-C cannot encompass perturbations in overall lipid profiles and may not adequately assess the heterogeneity in interindividual atherogenic vulnerability [25]. In some cases, lower LDL-C may not be associated with improved cardiovascular outcomes [26]. Consistent with previous work, it was observed in our study that despite the reduction of LDL-C levels in the E/C/F/TAF group, individual lipid species associated with CVD risk are persistently elevated during the follow-up.

Prospective data from previous studies show that the mortality of PLWH from liver disease was second only to AIDS-related mortality, and NAFLD could soon emerge as the most common liver disease in PLWH [27, 28]. Liver biopsy is the gold standard for diagnosis of NAFLD, but it is difficult to perform in clinical practice. Hence, in the present study, the HSI scoring system, based on BMI, ALT/AST ratio, gender, and history of diabetes, was used to evaluate and identify NAFLD in the participants [19]. In our study, incidence of NAFLD was elevated in the E/C/F/TAF group, although this difference was not statistically significant. Previous studies have demonstrated that the development of NAFLD is a long-term and multiple-step process [29]; hence, it is necessary to be alert to the adverse effects of E/C/F/TAF on the development of NAFLD. Notably, although effective for predicting hepatic steatosis in the general populations, HSI has not been extensively validated in other populations, such as PLWH. Therefore, our findings are exploratory, and further studies with larger sample sizes and more sensitive and accurate NAFLD diagnosis methods are required. In addition, in this study, FRS was applied to predict 10-year cardiovascular risk. Switching to B/F/TAF was found to be associated with decreased FRS, whereas FRS was found to increase mildly in the E/C/F/TAF group, indicating that B/F/TAF instead of E/C/F/TAF could be related to cardiovascular risk reduction. In a prospective study involving 2283 PLWH, FRS was found to accurately reflect risk of CVD events [30]. Thus, we have reason to believe that switching to B/F/TAF reduces cardiovascular risk in PLWH.

To investigate the long-term effects of different ART regimen on lipid metabolism, detailed lipidomic analysis was used to identify the specific lipid species that are significantly altered over 24 weeks treatment. In the longitudinal analysis, treatment with E/C/F/TAF was accompanied by a persistent elevation of several DGs and LPCs, while switching to B/F/TAF appeared to induce a substantially greater change in plasma lipid species, with increased levels of Cers, LPCs, PCs, PGs SMs, and Sos. We then made a horizontal comparison at week 24 to assess the lipid metabolism characteristics of the different ART regimens. It was observed that in comparison with E/C/F/TAF, B/F/TAF seems to have less effect on DGs (FC 0.004–0.389) and TAGs (FC 0.070–0483). Additionally, we identified significant associations of multiple TAGs and DGs with future CVD risk in this prospective cohort, which is in accordance with previous studies [31,32,33]. Notably, however, several DGs and TAGs such as DG (16:0/18:2), DG (18:2/22:6), DG (18:3/18:2), DG (20:5/18:2), TAG (18:3/18:2/21:5), and TAG (20:5/18:2/22:6) were newly identified to be associated with CVD risk after fully adjusted analyses. LPCs have been previously proposed as possible proinflammatory and proatherosclerotic lipids [34]. Further lipidomic research has revealed that the fatty acid composition of LPC particles plays a crucial role in their function. LPCs containing saturated fatty acids (SaFAs) are proinflammatory, while LPCs containing polyunsaturated fatty acids (PUFAs) are antiinflammatory [34]. Elevated circulating SaFAs are associated with a greater risk of CVD, whereas increased PUFA-containing LPCs appear to be associated with a lower risk of CVD and mortality in PLWH [35, 36]. In addition, Chai et al. found that lipid species associated with the risk of carotid artery plaque in LPCs have lower carbon numbers and double-bond contents, which is in line with our results [37]. In this study, LPC (34:2), LPC (34:3), and LPC (38:5), increased in the B/F/TAF group, were negatively associated with FRS. However, LPC (22:6), a PUFA-containing LPC with six double bonds, was found to be positively correlated with FRS, and the association remains robust when adjusting for age, sex, BMI, TC, and HDL-C. Further work is required to assess the relationship between LPC levels and CVD risk.

Sphingolipids and especially Cers are important bioactive lipids involved in atherosclerosis, and their predictive value for cardiovascular risk is superior to that of LDL-C [38, 39]. In the present study, Cer (d18:1/24:1) was found to increase with B/F/TAF treatment and to be positively associated with FRS, which is consistent with previous results in PLWH [40]. Distinct from previous related studies, Cer (d30:0), Cer (d32:0), Cer (d33:0), Cer (d36:0), and Cer (d38:0) were inversely correlated with FRS. This difference may be explained by the distinct fatty acid compositions of Cer particles and requires further study. Sphingoshines, another sphingolipid increased in the B/F/TAF group, was observed to be upregulated in PLWH despite receiving ART treatment [41]. Sphingoshines act as first and second messengers, playing a crucial role in regulating pathobiological processes, such as cancer, inflammation, and infectious diseases [42]. However, the relationship between sphingoshines and clinical cardiovascular disease remains poorly defined and warrants further investigation.

Together, our data show significant differences in lipid metabolism between the two groups, but the reason for this discrepancy remains obscure. Although TAF-containing regimens were previously thought to be related to a worsening of lipid profile, TAF exposure is comparable for the E/C/F/TAF and B/F/TAF groups due to the inhibitory effect of cobicistat on p-glycoprotein (P-gp) and breast cancer resistance protein (BCRP) transporters [43]. Cobicistat, a pharmacological booster, shows a lower potential for worsening lipid metabolism compared with ritonavir, but further investigation into the effects of cobicistat on lipid dysregulation is warranted [44].

To our knowledge, this is the first prospective lipidomic study that has identified lipid species that may be related to a better metabolism profile and lower CVD risk with the use of B/F/TAF rather than E/C/F/TAF. However, our study does have some limitations. First, the study was limited by a small sample size and residual confounding biases. Nevertheless, we selected stable PLWH with no other comorbidities or medications that might have an impact on the metabolic profile and to eliminate heterogeneity within the sample. Additionally, there were no significant demographic, clinical, or laboratory differences at baseline between the two groups.

Second, the study was limited by a short duration of follow-up and a lack of incidence of CVD events. Previous studies have indicated that significant alterations of lipidomic can happen in a short time frame [45], and in this study, STEM analysis was used to identified continuously changing metabolites. Although the FRS, a well-recognized CVD risk-assessment tool, was used to estimate the future risk of CVD for PLWH, we have to admit that direct measurement of CVD events is the gold standard for assessing cardiovascular risk. Therefore, longer-term studies to evaluate how alterations in lipidomic profiles may influence the incidence of actual CVD events are required.

Third, the study population was mainly middle-aged Chinese men, preventing extrapolation of our findings to other populations. We acknowledge that this gender distribution may introduce a bias in interpreting the impact of ART on lipidomic profile changes and its subsequent cardiovascular risk implications. Men and women exhibit significant physiological and metabolic differences that can influence the impact of ART on lipid profiles. For instance, differences in sex hormone levels can affect lipid metabolism, with women typically having higher levels of HDL-C and a lower risk of cardiovascular diseases [46]. In addition, gender may influence individuals’ responses to ART, including drug metabolism rates, the incidence of side effects, and treatment efficacy [47]. In future research endeavors, it is necessary to increase the sample sizes and ensure a more balanced representation of gender ratios within study populations so as to enhance the applicability and generalizability of our findings.

Finally, even though over a thousand individual lipids were identified in this study, the non-targeted lipidomic approach was restricted to lipid compounds known in the database, possibly overlooking some interesting but uncharacterized compounds. Furthermore, the levels of lipid species were semi-quantified without absolute values, and thus a targeted lipidomic on specific interesting lipid species is required to better understand the underlying biological mechanisms.