Frailty, as measured by the Liver Frailty Index (LFI), is increasingly incorporated into cirrhosis decision-making as a measure of physiological reserve beyond traditional biochemical parameters 1, 2. In this timely study, Kabelitz et al. addressed the practical question of whether baseline LFI meaningfully stratifies risk in patients with cirrhosis undergoing TIPS 3. In this retrospective review of a prospective cohort of 123 patients undergoing elective TIPS, baseline LFI was examined as a predictor for overt hepatic encephalopathy (OHE) and transplant-free survival (TFS). Despite 40 participants experiencing post-TIPS OHEs and 30 deaths/transplants during follow-up, baseline LFI was not independently associated with OHE or TFS even after stratification by established categories and quartiles. However, patients in the lowest LFI quartile had significantly better TFS. Additionally, frailty was dynamic, with 24.5% experiencing an improvement in LFI whereas 11.5% experienced a decline. These observations highlight a clinically relevant divergence between functional frailty and sarcopenia. Sarcopenia has been repeatedly associated with post-TIPS OHE, plausibly through reduced skeletal muscle–mediated ammonia disposal and broader metabolic vulnerability 4. Conversely, in the present cohort, LFI did not predict TFS or OHE. This discordance suggests that, among elective TIPS candidates, physical performance (LFI) and muscle mass (sarcopenia) represent partially distinct pathophysiologic domains: diminished physiological reserve versus liver disease-related muscle dysfunction. Therefore, a practical implication of this study is that pre-TIPS assessment may benefit from pairing functional testing with LFI with objective morphometric measures (e.g., CT-based skeletal muscle index) when the clinical question concerns encephalopathy risk 5. Other limitations of the study merit consideration. The absence of the control group introduces selection bias, especially when considering frail participants who underwent TIPS. Second, the major aetiology of liver disease was due to alcohol (50% ALD, 7% MetALD), where significant clinical improvement may also occur with sobriety 6. Third, despite high rates of frailty, MELD scores were below 16 in most patients, and the established interaction between MELD and LFI was not examined 1. In addition, while only 26% had a history of OHE, 71% were on lactulose and/or 54% were on rifaximin at the time of TIPS. There is growing evidence that prophylactic therapy for OHE post-TIPS reduces the risk of new OHE; however, it is unclear whether the centres follow this practice 7. Finally, the longitudinal improvement signal should be interpreted cautiously, as it is susceptible to survivorship bias, attrition bias, intercurrent illness and learning effects that may influence performance of LFI components. Future multicentre prospective studies should pre-specify missing-data handling, integrate parallel measures of muscle quantity and function, and include patient-centred endpoints such as falls, activities of daily living and quality of life, consistent with current cirrhosis frailty guidance 8. Overall, Kabelitz et al. provide an essential corrective to overly deterministic use of frailty metrics: elevated LFI alone should not preclude elective TIPS. Rather than serving as an exclusionary gatekeeper, LFI may identify a subset with a particularly favourable prognosis, encouraging clinicians to consider TIPS early in advanced cirrhosis and to pursue targeted optimization. Sai Kiran Kuchana: writing – original draft, writing – review and editing. Raj Vuppalanchi: conceptualization, investigation, writing – review and editing, supervision. Archita P. Desai: conceptualization, investigation, writing – review and editing, supervision. The authors have nothing to report. The authors declare no conflicts of interest. This article is linked to Kabelitz papers. To view this article, visit papers https://doi.org/10.1111/apt.70315 and https://doi.org/10.1111/apt.70554. Data sharing not applicable to this article as no datasets were generated or analysed during the current study.
Kuchana et al. (Mon,) studied this question.