In this real-world longitudinal study, we examined kidney and metabolic changes in a cohort including mainly old patients (mean age > 70 years) with CHB before and after switching from TDF to TAF or ETV. In this population, we did not observe clinically relevant changes in observed biomarkers following treatment switch; these variations are more likely attributable to the physiological decline of organ function associated with ageing than to drug-specific effects. The imbalance between treatment groups observed in this cohort likely reflects two converging factors: the broad applicability of Italian reimbursement criteria for TAF in this older population (where age ≥ 60 years alone represented a sufficient criterion for eligibility) and the high proportion of patients previously treated with lamivudine (66%), which lowers the resistance barrier to ETV [21, 22] determining a clinical preference for TAF.

Several findings emerged. First, renal function remained largely stable after switching, with no clinically relevant improvement, consistent with the predominant role of age-related decline in this older cohort [23]. Second, phosphate trajectories differed between groups, with partial recovery after switching in the ETV group, likely reflecting pre-existing tubular dysfunction related to selection bias. Third, total cholesterol increased immediately after TDF discontinuation, consistent with withdrawal of TDF’s lipid-lowering effect.

A key feature of our cohort is the advanced age (mean 72 years), which distinguishes this population from most previously published studies. The progressive ageing of CHB cohorts, mainly of those which acquired HBV before HBV vaccine was available, has been widely recognized, with increasing prevalence of chronic kidney disease (CKD), metabolic syndrome, and cardiovascular comorbidities in long-term treated patients [6]. Ageing itself contributes to renal decline, with an estimated reduction in glomerular filtration rate of approximately 8 mL/min per decade after the age of 40 [21]. In our cohort, the observed eGFR decline before and after switching (approx. 0.7–1.1 mL/min/year) closely mirrors the expected physiological age-related trajectory (approx. 0.8–1.0 mL/min/year), suggesting that in this older population with prolonged TDF exposure, age-related renal ageing may be the predominant driver of renal decline, making it difficult to isolate the contribution of TDF nephrotoxicity. In this context, antiviral-related toxicity cannot be interpreted independently of age-related organ vulnerability. Current guidelines and expert reviews explicitly recommend prioritizing TAF or ETV over TDF in older patients or those with renal risk, reflecting the increasing importance of safety profiles in lifelong therapy [2, 5]. However, evidence regarding the magnitude of benefit associated with switching after prolonged TDF exposure in older real-world populations remains limited. It should be acknowledged that the non-significant baseline differences between the two groups may partly reflect Italian reimbursement criteria, which consider age (> 60 years) and kidney dysfunction in treatment eligibility [18]. Our findings extend this concept by showing that, in such patients, a late treatment switching may have limited capacity to reverse established damage: age-related decline in renal function may partially explain the modest and heterogeneous changes observed after switching therapy, as physiological renal aging may attenuate the potential for recovery after removal of TDF exposure. Furthermore, TDF remains widely used globally because of its availability, cost-effectiveness, and established antiviral efficacy. Therefore, understanding the expected renal and metabolic effects of switching after long-term TDF exposure remains highly relevant for the management of ageing HBV populations.

One of the most consistent findings across all models was the effect of baseline values on longitudinal outcomes. Baseline creatinine, eGFR, phosphate, and total cholesterol were all strongly associated with biomarker levels at follow-up timepoints, reflecting the fact that inter-individual differences at baseline substantially outweigh longitudinal changes over time. Notably, demographic variables such as age and sex were not independent predictors in the multivariable models. This finding aligns with broader evidence indicating that renal outcomes during antiviral therapy are strongly influenced by pre-existing renal dysfunction, comorbidities such as diabetes and hypertension, and cumulative drug exposure [15]. In older patients with long-standing TDF exposure, this suggests that biological reserve at the time of switching may be a key determinant of reversibility. In our cohort switching therapy was associated with attenuation or stabilization of biomarker trajectories rather than clear improvements in absolute levels, suggesting mitigation of TDF-related toxicity rather than full reversibility of prior damage. This supports earlier switching strategies in older individuals, before irreversible organ changes occur.

Regarding renal outcomes, our results differ in part from several clinical trials and observational studies reporting improvement after switching from TDF to TAF. In randomized and real-world settings, TAF has demonstrated improved renal safety compared with TDF, with stabilization or modest improvement in eGFR and tubular markers [9, 10]. However, in our cohort, we did not observe a clear improvement in creatinine or eGFR after switching. Instead, renal function remained largely stable or continued to decline modestly, without significant differences between TAF and ETV. This apparent discrepancy may reflect differences in patient characteristics. Prior studies often included younger populations or patients with shorter TDF exposure, in whom removal of nephrotoxicity translates into measurable recovery. By contrast, our cohort had advanced age and mildly reduced baseline eGFR which may limit the extent of observable improvement despite potential stabilization of renal function. This interpretation is consistent with recent evidence indicating that the comparative renal safety of TAF and ETV may be similar in patients with preserved renal function, while differences become more evident in selected high-risk populations [12]. Taken together, our data suggest that in older patients, the main benefit of switching may be prevention of further decline rather than restoration of renal function.

Phosphate dynamics provided additional insight into treatment-specific renal tubular effects. In the ETV group, phosphate declined significantly during TDF therapy and showed meaningful recovery after switching, consistent with partial reversal of TDF-associated proximal tubular dysfunction. By contrast, the TAF group showed stable phosphate levels throughout, suggesting that patients allocated to TAF had less severe tubular involvement at baseline—a finding consistent with the selection bias already discussed. These results reinforce the concept that the renal tubular benefit of switching from TDF is most evident in patients with pre-existing tubular dysfunction, and highlight the importance of accounting for pre-switch trajectories when interpreting post-switch changes [10, 24].

Another finding is the lack of significant differences between TAF and ETV after switching. Across all models and timepoints, between-group contrasts were consistently non-significant. This aligns with large comparative studies and meta-analyses indicating broadly similar efficacy and, in many settings, comparable safety profiles between ETV and tenofovir-based therapies [12, 25, 26]. From a clinical perspective, this supports an individualized approach to treatment selection in older patients, in which drug choice is guided less by expected differences in short-term outcomes and more by patient-specific factors such as prior treatment exposure, renal risk, and comorbidity burden. The metabolic findings of our study deserve particular attention. We observed a significant increase in total cholesterol following TDF discontinuation followed by stabilization over time. This pattern is highly consistent with existing literature. This was further supported by the presence of a significant immediate increase in cholesterol levels at the time of treatment switch in the fixed step-change analysis. TDF has been repeatedly shown to exert lipid-lowering effects, whereas TAF is considered metabolically neutral or associated with modest lipid increases [16]. The magnitude of the immediate cholesterol increase (approx. 18 mg/dL) was similar in both treatment groups, further supporting the interpretation that this reflects TDF withdrawal rather than a TAF-specific effect. Importantly, recent studies have demonstrated that increases in cholesterol after switching to TAF occur predominantly in patients previously treated with TDF, supporting the hypothesis that this reflects withdrawal of TDF-induced lipid suppression rather than a direct adverse effect of TAF [26]. These findings are further supported by multiple cohort studies and meta-analyses showing increases in total cholesterol, LDL, and triglycerides after switching from TDF to TAF, along with a higher incidence of dyslipidemia [14, 15]. However, the clinical implications of these changes remain uncertain. While some studies suggest a modest increase in estimated cardiovascular risk or atherosclerotic cardiovascular disease scores after switching [16, 17], others have not demonstrated a clear increase in cardiovascular events, at least in the short to medium term [27]. Therefore, the interpretation of lipid changes should be cautious: these alterations likely reflect a shift toward a “physiological” lipid profile rather than a true drug-induced metabolic toxicity.

From a geriatric perspective, this issue is particularly relevant. Older patients with CHB already have an elevated baseline cardiovascular risk due to age and comorbidities. In this context, the increase in cholesterol after TDF discontinuation may warrant closer metabolic monitoring and optimization of cardiovascular risk factors, rather than avoidance of TAF or ETV. Statin use and global risk assessment may be more important than the choice of antiviral agent per se in determining long-term outcomes.

This study has several limitations, mainly related to its retrospective design. First, treatment allocation was not randomized, introducing selection bias; although this was addressed through multivariable adjustment, residual confounding cannot be excluded. Second, the relatively small size of the ETV group may have reduced statistical power for between-group comparisons. Third, comorbidity data were not systematically available, and were therefore not included in the analysis. However, the real-world design, the longitudinal modelling approach, and the focus on an older population represent important strengths, addressing a clinically relevant gap in the current clinical practice.