As of December 2017, there were 170 HIV-1-positive individuals who were regular attendees at the National Hospital Organization Sendai Medical Center, of whom 125 were being treated with antiretroviral therapy including TAF. Of the 125 individuals, 18 who were treated with TAF from the start and 11 whose previous therapy was antiretroviral therapy including abacavir were excluded from the study. In addition, one non-Japanese person and five patients who had switched to TAF after receiving TDF for less than 48 weeks were excluded. Finally, 90 individuals were included in the study. By 288 weeks, which was set as the study period, 27 individuals had been transferred to other hospitals due to work commitments, one individual died after being diagnosed with HIV encephalopathy, one had his hospital visit interrupted due to disappearance, and one switched to a two-drug therapy involving Dolutegravir (DTG)/Lamivudine (3TC) combination from cobicistat (COBI) boosted Darunavir (bDRV)/Emtricitabine (FTC)/TAF combination about 211 weeks after starting TAF to reduce the size and number of components of the pills. Therefore, 60 individuals who continued antiretroviral therapy including TAF for the entire 288 weeks were included in the analysis. As shown in Table 1, HIV-1-positive individuals were taking a variety of the third class of medications when they started taking TDF, but when they switched from TDF to TAF, 51.7% of the 60 population individuals were taking DTG. As others, 23.3% were taking COBI boosted Elvitegravir (EVG), 16.7% were taking Raltegravir (RAL), and 8.3% were taking bDRV. None of the participants switched the third class of medications to that of another for 96 weeks after switching from TDF to TAF. However, with the launch of the Bictegravir (BIC)/FTC/TAF combination in April 2019, the number of HIV-1-positive individuals who switched from each of the third class of medications to BIC increased after 144 weeks. Finally, at 288 weeks, 66.7% of the 60 population individuals was taking BIC. Otherwise, 11.7% were taking DTG, 3.3% were taking EVG, 8.3% were taking RAL, 6.7% were taking bDRV, and 3.3% were taking Doravirine (DOR).

Table 1 shows the characteristics of the 60 individuals. At the switch to antiretroviral therapy including TAF, most of the analyzed patients had good viral control and improved immunity with a median CD4 count of 453 (IQR: 337–569) cells/μL. Percentage values of the third class of medications used in combination with TDF or TAF are also included in Table 1. The median duration of treatment with TDF was 283 weeks (IQR: 123–465 weeks), which is comparable to the 288 weeks of treatment with TAF in this study. The median SCr and eGFR were 0.90 (IQR: 0.75–0.98) mg/dL and 75.15 (IQR: 64.56–89.18) mL/min/1.73 m2, respectively, at the switch from TDF to TAF. Additionally, the median Uβ2MG, which was not measured at the start of TDF therapy, was 267 (IQR: 114–869) μg/L. Similarly, the median body weight and median BMI were 70.6 (IQR: 61.1–75.2) kg and 23.5 (IQR: 21.8 – 26.2) kg/m2, while the median TG, T-cho, HDL-cho, and LDL-cho levels were 150 (IQR: 106–246), 179 (IQR: 154–205), 47 (IQR: 38–53), and 97 (IQR: 85–119) mg/dL, respectively. Subjects who were receiving medications for hypertension, diabetes, and lipid metabolic disorders at the start of TDF and TAF are shown in Table 1.

The eGFR changes in the 60 patients are shown in Fig. 1A. From TDF0 to TAF0, eGFR showed a significant decrease (mean difference (MD) = 23.20 mL/min/1.73 m2, 95% confidence interval (CI) = 19.05–29.26, p < 0.0001). No significant differences remained in TAF48 and TAF96 levels after switching to antiretroviral including TAF. The eGFR after TAF144 (MD = -5.20 mL/min/1.73m2, 95% CI: -8.69 – -1.72, p = 0.0041) was significantly lower than that of TAF0.

A) Change in eGFR (mean ± SD) over 288 weeks in HIV-positive individuals who switched from TDF to antiretroviral therapy including TAF and continued taking it. B Change in eGFR (mean ± SD) over 288 weeks in each group classified by eGFR level at the time of switching from TDF to TAF (week 0, TAF0). In both (A) and (B), paired t-tested was performed at 48 weeks (TAF48), 96 weeks (TAF96), 144 weeks (TAF144), 192 weeks (TAF192), 240 weeks (TAF240), and 288 weeks (TAF288), using TAF0 as the reference, with a significance level of p < 0.05. The sample size of B) at each survey from TAF0 to TAF288 was 13 in group G1, 37 in group G2, and 10 in groups G3a, b. At TDF0, there were 8 in group G1, 32 in group G2, and 9 in groups G3a, b

Figure 1B shows the changes in eGFR classified into three groups based on GFR categories according to the eGFR values at TAF0. There were no G4 or G5 patients in this study. TDF durations (mean ± SD) for groups G1 (291 ± 158 weeks), G2 (296 ± 212 weeks), and G3a, b (311 ± 188 weeks) showed no significant difference.

After switching from TDF to TAF, the eGFR in group G1 continuously decreased for up to 288 weeks, with a particularly significant decrease after week 144 (MD = -16.96 mL/min/1.73 m2, 95% CI: -29.65 – -4.27, p = 0.0130). In group G2, the decline in eGFR was suppressed until 96 weeks, finally showing a significant decrease (MD = -4.11 mL/min/1.73 m2, 95% CI: -7.69 – -0.53, p = 0.0256) at 288 weeks. In contrast, in groups G3a and b, eGFR significantly increased at 48 weeks (MD = 6.06 mL/min/1.73 m2, 95% CI: 1.87–10.26, p = 0.0097) compared to that at TAF0. After 96 weeks, the inhibition of eGFR decline continued until week 288, although this difference was not statistically significant.

Changes in Uβ2MG are shown in Fig. 2A as an indicator of renal tubular damage. Uβ2MG significantly decreased at 48 weeks (MD = -2753.5 μg/L, 95% CI: -6471.9 –—964.8, p < 0.0001) compared to TAF0 and remained significantly lower thereafter until TAF288 (MD = -2700.2 μg/L, 95% CI: -6273.9 –—873.4, p = 0.0013). The changes in Uβ2MG in groups G1, G2, and G3a, b based on GFR classification are shown in Fig. 2B. After switching from TDF to TAF, Uβ2MG in the G3a and b groups decreased significantly at TAF48 (MD = -14,881.0 μg/L, 95% CI: -44,606.0 –—14,843.5, p = 0.0156), and the significant decrease was maintained thereafter until TAF192 (MD = -15,335.0 μg/L, 95% CI: -44,914.0 –—14,243.5, p = 0.0156). However, there was no significant difference at TAF240 and TAF288. In the group G2, Uβ2MG significantly decreased at 48 weeks (MD = -866.8 μg/L, 95% CI: -2024.4 –—290.8, p = 0.0125) compared to TAF0, and the decrease was maintained thereafter until TAF288 (MD = -917.4 μg/L, 95% CI: -2050.9 –—216.2, p = 0.0146). In the group G1, Uβ2MG was originally low at TAF0 after the switch from TDF to TAF, and there was consistently no significant difference up to TAF288. On testing the association between eGFR and Uβ2MG at TAF0 and TAF288, Spearman’s rank correlation coefficient was -0.3859 (p = 0.0052) at TAF0 and 0.0587 (p = 0.6643) at TAF288. The results of the single regression analysis were p = 0.0320 at TAF0 and p = 0.8914 at TAF288. Thus, a negative correlation between eGFR and Uβ2MG was suggested when taking antiretroviral therapy, including TDF; however, this association disappeared on switching to TAF (Fig. 3A and B).

A) Change in Uβ2MG (median; IQR) over 288 weeks in HIV-positive individuals who switched from TDF to antiretroviral therapy including TAF and continued taking it. B) Change in Uβ2MG (median; IQR) over 288 weeks in each group classified by eGFR level at the time of switching from TDF to TAF (week 0, TAF0). In both (A) and (B), Wilcoxon signed rank test was performed at 48 weeks (TAF48), 96 weeks (TAF96), 144 weeks (TAF144), 192 weeks (TAF192), 240 weeks (TAF240), and 288 weeks (TAF288), using TAF0 as the reference, with a significance level of p < 0.05. The sample size of B) at each survey from TAF0 to TAF288 was 11 in group G1, 33 in group G2, and 7 in groups G3a, b

Scatter plots of eGFR and Uβ2MG at (A) week 0 (TAF0) and (B) week 288 (TAF288) for 60 subjects

Figure 4A shows BMI as an indicator of weight. BMI showed a significant increase at each time point after 48 weeks (MD = 0.56 kg/m2, 95% CI: 0.26–0.87, p = 0.0005) compared to that at TAF0. Thereafter, it remained near the upper reference limit of 25.0 kg/m2 [24] until TAF288 (MD = 2.50 kg/m2, 95% CI: 1.63–3.37, p < 0.0001). Additionally, a comparison between TAF48 and TAF288 showed no significant difference.

Changes in (A) BMI (mean ± SD), (B) TG (mean ± SD), (C) T-cho (mean ± SD), (D) HDL-cho (mean ± SD), and (E) LDL-cho (mean ± SD) over 288 weeks of individuals who switched from TDF to antiretroviral therapy including TAF and continued taking it. In A), B), C), D), and E), paired t-test was performed at 48 weeks (TAF48), 96 weeks (TAF96), 144 weeks (TAF144), 192 weeks (TAF192), 240 weeks (TAF240) and 288 weeks (TAF288), with TAF0 as the reference, and the significance level was set at p < 0.05

Figures 4B, C, D and E show changes in TG, T-cho, HDL-cho, and LDL-cho as parameters of lipid metabolism. The TG levels were already above the reference level (50–149 mg/dL) at the time of switching from TDF to TAF. The TG levels at each point were compared with those at TAF0, all of which continued to show no significant differences. T-cho showed a significant increase at 48 weeks (MD = 22.1 mg/dL, 95% CI: 13.6–30.7, p < 0.0001) when compared to TAF0 at each point and then remained within the reference levels (150–219 mg/dL) until TAF288 (MD = 23.8 mg/dL, 95% CI: 13.2–34.4, p < 0.0001). HDL-cho showed a significant increase at 48 weeks also (MD = 4.5 mg/dL, 95% CI: 2.4–6.5, p < 0.0001) when compared to TAF0 at each time point and then remained within reference levels (men: 40–86 mg/dL, women: 40–96 mg/dL) until TAF288 (MD = 9.4 mg/dL, 95% CI: 4.0–14.7, p = 0.0009). Moreover, LDL-cho showed a significant increase at 48 weeks (MD = 14.7 mg/dL, 95% CI: 8.5–20.8, p < 0.0001) when compared to TAF0 at each point and remained within reference levels (70—139 mg/dL) until TAF288 (MD = 11.7 mg/dL, 95% CI: 4.3–19.2, p = 0.0027). A comparison between TAF48 and TAF288 showed no significant differences in any of the aforementioned parameters.

The prediction results of factors affecting the change in eGFR at 288 weeks after switching from TDF to TAF are shown in Table 2. The outcome was whether the decline in eGFR between TAF0 and TAF288 was ≥ 2.0 mL/min/1.73 m2, and the associated factors that might reduce eGFR above the mean annual rate of decline in the Japanese population were identified. Factors affecting the decrease in eGFR were: history of AIDS (odds ratio (OR) = 0.1361, 95% CI: 0.0205–0.9015, p = 0.0232) and Uβ2MG (OR = 0.9996, 95% CI: 0.9992–0.9999, p = 0.0087). However, when these ORs are considered, switching from TDF to TAF affected the suppression of eGFR decline in the group with a history of AIDS. Switching from TDF to TAF also reduced Uβ2MG and had little effect on eGFR decline.