We collected data including patients’ demographics, preoperative clinical characteristics, preoperative laboratory test results, pathologic, and surgical results. Patients received the appropriate preoperative treatment, such as endoscopic biliary stenting, percutaneous transhepatic biliary drainage (PTBD), chemotherapy (CTx), and portal vein embolisation (PVE), to augment remnant liver volume. Hepatic resection included segmental resection, left or right lobectomy, and extended hepatectomy. If patients presented severe obstructive jaundice with cholangitis, we evaluated the need for preoperative biliary drainage, including PTBD, endoscopic retrograde biliary stent, and endoscopic naso-biliary drainage, depending on the specific condition. Some patients underwent neoadjuvant treatment with the purpose of downstaging in the case of borderline resectable (suspicious tumor invasion to the contralateral portal vein or hepatic artery, and a tumor located over the U-point or P-point) or locally advanced (suspicious local lymph node metastasis) Klatskin tumor (25). PVE was performed to augment the remnant liver volume, and the decision to resect the Klatskin tumor was made using a multidisciplinary approach considering vessel invasion, possibility of achieving negative margins, and future remnant liver volume (26). We collected data on surgical and oncological outcomes, including postoperative morbidity, mortality, ICU admission, length of hospital stay (LOS-H) and LOS-I, overall survival, and recurrence. Criteria for postoperative ICU admission are for patients who need intensive monitoring, treatment, and ventilator care due to medical or surgical complications such as shock, heart failure, respiratory failure, renal injury, and massive bleeding. Postoperative ICU admission was arranged during the total perioperative period with flexibility including the intraoperative situation and patients’ status in the recovery room. Patients who were not directly admitted to the ICU from the operation room but underwent delayed admission or re-admission within 7 days from the surgery date because of an emergent reason, such as complications or re-operation requiring ICU care, were also included. Therefore, the LOS-I was the sum of the total length of ICU stay if admitted within 7 days after surgery. The grade of postoperative morbidity was assigned according to the Clavien–Dindo classification (CDC) within 90 days of surgery or until the discharge date, and severe complications were defined as complications with a CDC ≥ 3 (27). Postoperative mortality was recorded during the same period. To evaluate the preoperative nutritional status of cancer patients, the prognostic nutritional index (PNI), suggested as a clinical predictor of prognosis, was calculated using collected laboratory data (28). PNI was defined as 10 × serum albumin value (g/dL) + 0.005 × lymphocyte count (/mm³), with a cut-off value for low PNI of less than 1.39 indicating nutritional impairment. Cancer recurrence and survival were determined from the time of surgery to the time of event or the most recent follow-up date.

Preoperative sarcopenia was evaluated according to the obtained preoperative abdominal and pelvic CT images within 90 days before surgery which were routinely obtained to diagnose and plan treatment. The cross-sectional surface (cm²) of both psoas muscle areas (PMA) was automatically quantified at the third lumbar (L3) vertebra level using the Aquarius iNtuition Viewer (ver. 4.4.13, TeraRecon Inc., San Mateo, CA, USA) imaging server platform (29). Using this software, each surface can be automatically quantified using a particular CT attenuation range. Non-contrast-enhanced CT images were used to measure the radiation attenuation density (HU) of the muscle and adipose tissue. Using these values, muscle steatosis was evaluated using intramuscular adipose tissue content (IMAC) by dividing the CT attenuation value of the multifidus muscle (HU) by that of the subcutaneous fat HU (30). A higher IMAC indicates that skeletal muscles contain a greater amount of adipose tissue and may result in poor prognosis in several cancers (31–33). Cut-off values for high IMAC (-0.358 for men and -0.229 for women) were defined in a previous study including healthy donors for liver transplantation (34).

The measured PMA was then normalised by patient height squared (cm²/m²), which was termed the psoas muscle mass index (PMI). Optimal cut-off points for PMI data in the maximal predictive value for postoperative ICU admission were determined using receiver operating characteristic (ROC) curve analysis. Sarcopenia was determined when the patients’ PMI was lower than the sex-specific cut-off. The study population was categorised into two groups: sarcopenia and non-sarcopenia patients, and postoperative short- or long-term outcomes were compared.

Categorical variables are expressed as frequencies and percentages. Continuous variables are summarised as means with standard deviations (SDs). Categorical variables were analysed using the χ2 or Fisher’s exact test, and continuous variables were compared using the independent t-test or Mann–Whitney U test. Spearman’s correlation tests were used to assess the relationships between risk factors and continuous outcomes. Multivariate analysis was performed to investigate statistically significant risk factors affecting ICU admission and LOS-I using variables with p-values < 0.05. Survival estimates were obtained using the Kaplan–Meier survival method, and differences in overall survival (OS) and recurrence-free survival (RFS) between the groups were determined using the log-rank test. All results with p-values < 0.05 were statistically significant. All statistical analyses were performed using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp., Armonk, NY, USA).

A total of 330 patients with pathologically confirmed Klatskin tumors who underwent curative hepatic resection were identified during the study period. Among them, 13 patients were excluded because they underwent hepatic wedge resection (n = 2) or cooperation with other organs (n = 11). Ultimately, 317 patients were included in the analysis. The T stage was unmeasured in three patients owing to post-CTx changes in pathological specimens. The mean age of the study population was 65.6 ± 9.0 years, and 202 men (63.7%) and 115 women (36.3%) were included. The PMI (men: 6.91 vs. women: 3.78, p < 0.001) and IMAC (men: -0.44 vs. women: -0.32, p < 0.001) showed significant differences between sexes. Among them, 196 patients (61.8%) were provided ICU care within seven postoperative days, and most of them (n = 193/196, 98.4%) were directly transported from the operating theatre to the ICU. During the same period, 11 patients (3.5%) were readmitted to the ICU after being discharged from the ICU to the general ward. Among them, 4 (36.4%) developed septic shock, 2 (18.2%) underwent hepatic artery embolization or operation due to bleeding, 2 (18.2%) underwent thrombus aspiration and stent insertion due to portal vein thrombosis, 1 (9.1%) had hepatic encephalopathy, 1 (9.1%) had ventilator care due to pulmonary edema, and 1 (9.1%) experienced liver failure. Severe complications (CDC ≥ 3) were observed in 125 (39.4%) patients within 90 days of surgery, with 82 patients requiring surgical, endoscopic, or radiological intervention due to pleural effusion, pericardial effusion, bile leakage, abdominal abscess, ascites or bleeding (CDC = 3). 21 required intensive care or ventilator due to organ dysfunction like renal failure, hepatic failure, cardiogenic shock, respiratory distress, sepsis or stroke (CDC = 4). 22 patients (6.9%) died owing to postoperative mortality (CDC = 5). Table 1 summarises patient characteristics and postoperative outcomes of the study population.

The mean ± SD for PMI in men was 6.91 ± 1.72 cm²/m², whereas that in women was 3.78 ± 1.04 cm²/m² (p < 0.001). Sex-specific PMI cut-off values for sarcopenia were determined at 6.74 cm²/m² for men [sensitivity = 65.4%, specificity = 69.3%, area under the curve (AUC) = 0.700] and 3.39 cm²/m² for women (sensitivity = 47.8%, specificity = 76.1%, AUC = 0.609) by ROC curve analysis. Using these cut-offs, preoperative sarcopenia was observed in 106 men (52.5%) and 44 women (38.3%). The sarcopenia group had a significant lower BMI (23.0 vs. 23.8 kg/m², p = 0.004) and higher preoperative CTx incidence (12.7% vs. 4.8%, p = 0.015) than the non-sarcopenia group. Patients with sarcopenia underwent longer surgeries (550.7 vs. 506.7 min, p = 0.023) with more intraoperative bleeding (1,375 vs. 976 ml, p = 0.006), packed RBC transfusion events (44.7% vs. 28.1%, p = 0.002), and transfusion volumes (427 vs. 210 ml, p < 0.001) than those without sarcopenia.

Patients with preoperative sarcopenia presented with significantly more frequent ICU admissions within 1 week after surgery (77.3% vs. 47.9%, p < 0.001) and longer total LOS-I (2.45 vs. 0.89 days, p < 0.001) than patients without sarcopenia. Furthermore, the sarcopenia group showed significantly higher postoperative LOS-H (27.8 vs. 21.4 days, p = 0.006), severe complication rates (48% vs. 31.7%, p = 0.003), and in-hospital mortality (11.3% vs. 3.0%, p = 0.004) within postoperative 90 days than the non-sarcopenia group. Postoperative re-admission or delayed admission in ICU showed no difference between the two groups. Detailed comparisons of baseline characteristics and short-term outcomes between the sarcopenia and non-sarcopenia groups are presented in Table 2 .

Multivariate analysis showed that sarcopenia was significantly associated with postoperative ICU admission [adjusted odds ratio (OR): 2.461, p = 0.006] and prolonged LOS-I (B: 0.957, p = 0.008). Other factors associated with postoperative ICU admission in the multivariate analysis were serum carbohydrate antigen (CA) 19-9 levels, operation time, and intraoperative packed RBC transfusion events. Alternatively, BMI, Charlson comorbidity index, T stage 3 or 4, operation time, and intraoperative packed RBC transfusion events were significantly associated with LOS-I in multivariate analysis ( Tables 3 , 4 ).

Non-contrast-enhanced CT images were available for 285 patients, and a high IMAC was observed in 12.0% (n = 38/285) of the patients. A strong linear correlation was noted between PMI and IMAC (correlation coefficient; -0.414, p < 0.001), and a high IMAC was more frequent in sarcopenia patients (17.4% vs. 9.4%, p = 0.046) than in patients without sarcopenia. Preoperative nutritional assessment via laboratory data revealed that patients with sarcopenia had lower serum haemoglobin and albumin levels than patients without sarcopenia. A low PNI was noted in 74 (23.3%) patients and was significantly more frequent in the sarcopenia group (28.7% vs. 18.6%, p = 0.034) than in the non-sarcopenia group. A high IMAC (81.6% vs 58.6%, p = 0.007) and low PNI (73.0% vs 58.4%, p = 0. 024) were significantly associated with an increased postoperative ICU admission rate.

In total, the median follow-up duration among the living patients was 44.3 months, and 181 (57.1%) patients experienced recurrence during the study period. The 1-, 3-, and 5-year OS rates of the patients were 84%, 57%, and 44%, respectively, while RFS rates were 74%, 39%, and 25%, respectively. The median RFS was 27.8 months in patients with sarcopenia and 20.3 months in patients without sarcopenia, but the difference was not statistically significant (p = 0.609). However, sarcopenia patients showed a trend towards a lower OS than non-sarcopenia patients, with borderline significance (median 38.7 vs 57.3 months, p = 0.075). In the subgroup analysis, male patients with sarcopenia had significantly lower OS than those without sarcopenia (median 31.8 vs 57.2 months, p = 0.024, Figure 1 ). Further analysis of patients’ IMAC and PNI did not yield significant impacts on RFS and OS, except that patients with a low PNI were associated with lower OS (median 29.7 vs 57.2 months, p = 0.006, Figure 2 ) than patients with a normal PNI.