Body composition is associated with disease aetiology and prognosis in patients undergoing resection of intrahepatic cholangiocarcinoma

Abstract Background Body composition alterations are frequent in patients with cancer or chronic liver disease, but their prognostic value remains unclear in many cancer entities. Objective We investigated the impact of disease aetiology and body composition after surgery for intrahepatic cholangiocarcinoma (iCCA), a rare and understudied cancer entity in European and North American cohorts. Methods Computer tomography‐based assessment of body composition at the level of the third lumbar vertebra was performed in 173 patients undergoing curative‐intent liver resection for iCCA at the Department of Surgery, Charité – Universitätsmedizin Berlin. Muscle mass and ‐composition as well as subcutaneous and visceral adipose tissue quantity were determined semi‐automatically. (Secondary) sarcopenia, sarcopenic obesity, myosteatosis, visceral and subcutaneous obesity were correlated to clinicopathological data. Results Sarcopenia was associated with post‐operative morbidity (intraoperative transfusions [p = 0.027], Clavien–Dindo ≥ IIIb complications [p = 0.030], post‐operative comprehensive complication index, CCI [p < 0.001]). Inferior overall survival was noted in patients with myosteatosis (33 vs. 23 months, p = 0.020). Fifty‐eight patients (34%) had metabolic (dysfunction)‐associated fatty liver disease (MAFLD) and had a significantly higher incidence of sarcopenic (p = 0.006), visceral (p < 0.001) and subcutaneous obesity (p < 0.001). Patients with MAFLD had longer time‐to‐recurrence (median: 38 vs. 12 months, p = 0.025, log‐rank test). Multivariable cox regression analysis confirmed only clinical, and not body, composition parameters (age > 65, fresh frozen plasma transfusions) as independently prognostic for overall survival. Conclusion This study evidenced a high prevalence of MAFLD in iCCA, suggesting its potential contribution to disease aetiology. Alterations of muscle mass and adipose tissue were more frequent in patients with MAFLD.

Intrahepatic cholangiocarcinoma (iCCA) is the second most common primary liver cancer after hepatocellular carcinoma (HCC). 1 The incidence and mortality of iCCA is rising in western countries, despite an overall declining cancer mortality over the last three decades. 2While many cases of iCCA in Europe arise sporadically, known risk factors include chronic inflammation, cholelithiasis and cirrhosis, while liver fluke infections strongly contribute to iCCA incidence in Asian regions. 3Generally, the aetiology and disease course of iCCA are understudied in western cohorts.
Recently, obesity and non-alcoholic fatty liver disease (NAFLD), or, according to recent pathophysiological classifications, metabolic (dysfunction)-associated fatty liver disease (MAFLD), have been recognized as an iCCA aetiology and suggested as an adverse prognostic factor for overall survival. 4,5MAFLD is estimated to affect 25% of the global population and constitutes an important aetiology of end-stage and malignant liver disease. 6A rising incidence and disease severity is projected for the population worldwide, including the USA, China and Europe. 7atients with fatty liver disease furthermore carry a high overall risk of adverse cardiovascular events, other oncological diseases and musculoskeletal disorders.
Body composition-the quantity of skeletal muscle and the quantity and distribution of adipose tissue-of patients with obesity fulfilling non-alcoholic steatohepatitis (NASH) criteria was characterized by various groups, suggesting that patients typically have a concomitant increase of both adipose tissue and skeletal muscle mass. 8t is currently unclear whether these observations can be translated into oncological settings, where the typical phenotype of cachexia and sarcopenia is the most frequently described alteration.Our group recently reported that sarcopenic obesity may hold value for overall survival in iCCA in a different patient cohort from a German tertiary centre. 9This publication, despite a lower sample size, showed a high prevalence of body composition pathologies in patients with iCCA and perihilar CCA. 9 In gastrointestinal malignancies, an association of body composition with metabolic function, systemic inflammatory processes and overall prognosis has been suggested previously. 10,11Consequently, nutritional and rehabilitative strategies to maintain and increase muscle mass have been incorporated into guidelines for these patients. 12It is currently unclear whether patients suffering from MAFLD-iCCA show alterations in their body composition, and whether these changes affect disease prognosis.We therefore aimed to delineate prognostic roles of disease aetiology and body composition in iCCA.Furthermore, we examined the relationship between underlying liver disease and pathologies in the computed tomography (CT)-based composition and quantity of muscle and adipose tissue distribution in a European cohort of surgically treated patients with iCCA.

| METHODS
All consecutive patients undergoing curative-intent hepatectomy for iCCA between March 2010 and December 2020 at the Department of Surgery, Charité -Universitätsmedizin Berlin were retrospectively evaluated for study inclusion.Inclusion criteria were defined as, (a) pathological and radiological diagnosis of iCCA under the exclusion of perihilar and distal tumour subtypes and mixed iCCA-HCC, (b) available CT scans including the third lumbal vertebra within 3 months prior to operation.Patients with only other available imaging modalities, such as magnetic resonance imaging (MRI)-based abdominal staging, were not included in this study.
This study was approved by the ethics committee of the Charité -Universitätsmedizin Berlin (EA1/105/21) and conducted in accordance with the Declaration of Helsinki and the good clinical practice (ICH-GCP) guidelines.Informed consent was waived in agreement with the ethics committee due to retrospective, pseudonymized study design and analysis of available clinical data.Clinical data were retrieved from patients' records and from a prospectively managed institutional database.Recurrence and survival data were obtained from the Charité outpatient clinic and from local outpatient hepatologists and oncologists.
Post-hepatectomy liver failure (PHLF), as a potential result of a diminished overall energy reserve, was defined as based on post-operative (Day 5) international normalized ratio (INR) together with hyperbilirubinemia as previously described. 13The presence of MAFLD, as opposed to just

Conclusion:
This study evidenced a high prevalence of MAFLD in iCCA, suggesting its potential contribution to disease aetiology.Alterations of muscle mass and adipose tissue were more frequent in patients with MAFLD.

K E Y W O R D S
liver cancer, MAFLD, NAFLD, visceral obesity the presence of steatosis in our exploratory analysis, was defined based on the pathophysiology-centred 2020 consensus statement (any presence of pathology-proven hepatic steatosis plus Type 2 diabetes or a body mass index [BMI] ≥25 kg/m 2 or the presence of more than two metabolic abnormalities). 14Histological steatosis was routinely assessed on Haematoxylin-Eosin staining and reported in the non-tumourous area of the resected specimen by a surgical pathologist.In contrast to previous NAFLD criteria, this pathophysiology-centred approach uses positive inclusion criteria and does not rely on patient-reported alcohol consumption, 14 consistent with the paradigm shift towards (a) relevant alcohol consumption being often reported inaccurately (and not routinely investigated prior to oncological surgery as opposed to liver transplant recipients with unclear consumption status) (b) the copresence and thus the substantial etiological overlap of the metabolic syndrome and relevant alcohol consumption 15 (c) the more accurate reflection on the pathophysiological aspects of the metabolic syndrome as drivers of liver disease. 16

| Image and clinical data analysis
As previously described, 9,17 an axial CT image at the level of the third lumbar vertebra from the most recent CT image before surgery was analyzed semi-automatically with 3D Slicer 18 and the Workstation SlicerCIP extension, body composition module (version 4.10.2).Attenuation values from −29 to 150 Hounsfield units (HU) defined skeletal muscle. 19The spinal muscle area (SMA) included the psoas major, spinal (erector spinae, quadratus lumborum), transversus abdominis, external and internal oblique, and rectus abdominis muscles.Attenuation values from −150 to −50 HU indicated visceral adipose tissue, while −190 to −30 HU defined subcutaneous adipose tissue (subcutaneous fat area [SFA]).Skeletal muscle radiation attenuation (SM-RA) in HU within the muscle area was recorded to assess myosteatosis.Muscle and adipose tissue indices (skeletal muscle index [SMI]; subcutaneous fat index [SFI]) were calculated by normalizing the SMA and SFA to the patients' height (area[cm] 2 /height[m] 2 ).The same trained investigator performed the segmentation analysis while being blinded to patients' outcomes (MDP).
Cut-offs for body composition pathologies were derived from large multicentric oncological cohorts to avoid overfitting to the present dataset. 20Primary sarcopenia is defined as low muscle mass and low muscle strength, 21 while our assessment of sarcopenia relied only on the definition of low muscle mass with the following sexspecific cut-off: SMI <52.4 cm 2 /m 2 for men and <38.5 cm 2 / m 2 for women. 22Hereafter, this CT-based diagnosis will be referred to as 'sarcopenia'.The cut-off for myosteatosis was <41 HU if the BMI was <25 kg/m 2 and <33 HU for patients whose BMI equalled or exceeded 25 kg/m 2 . 20The adaptation to BMI is based on the fact that an accumulation of inter-and intramyocellular fat is significantly dependent on the overall amount of body fat and can only be considered pathological in the context of BMI.A visceral fat area (VFA) exceeding 100 cm 2 indicated visceral obesity, 19 while the SFI was dichotomized at the upper tertile of the cohort (71.89 cm 2 /m 2 ). 23Sarcopenic obesity was defined as sarcopenia plus a simultaneous BMI ≥25 kg/m 2 . 9

| Statistical analysis
The primary end point of the present study was oncological and overall survival in patients with iCCA depending on their body composition.The SPSS Statistics 24 software (IBM Corp.) was used for statistical analyses and GraphPad Prism 9 (GraphPad Software) was used for visualization of correlation matrices.Categorical variables were presented as number (frequency, %) and compared with the Mann-Whitney U test, while continuous variables (normally distributed) were displayed as mean ± standard deviation and compared with the Pearson's chi-square test.Nonnormally distributed data were presented as median and range.For comparisons between 3 or more variables, the ANOVA with post hoc Bonferroni correction was used for continuous variables, and the Kruskal-Wallis test for categorical variables.Spearman r correlation was calculated and plotted for the association of MAFLD and body composition parameters.A two-sided p-value of ≤0.05 was regarded as statistically significant, unless corrected with the Bonferroni method.Median time to recurrence (TTR) and overall survival (OS) were presented with 95% confidence intervals (CI).For TTR, the time between operation and recurrence was calculated, and patients were censored if they died or were lost to follow-up.For OS, the time from operation to death (from any cause) was calculated, and patients were censored at the time of their loss to followup.Survival differences between groups were compared using the Kaplan-Meier curves and the log-rank test, hazard ratios (HR) were calculated with univariable and multivariable survival analysis.Statistically significant covariates in univariable analysis that are stable over time, and under exclusion of parameters with suspected collinearity were included in the multivariable analyses.

| RESULTS
Of 236 patients with iCCA who were operated at the Department of Surgery, Charité -Universitätsmedizin Berlin within the study period, 173 patients (73%) met the inclusion criteria.The remaining patients either underwent MRI preoperatively or CT images were unavailable for body composition analyses (Figure S1).The study cohort was composed of 87 men (50%) and 86 women (50%) with a mean age of 64 years.

|
In patients with available data on liver steatosis (n = 147), the MAFLD criteria were subsequently applied because patients without liver disease and without steatosis, but with fibrosis/cirrhosis grouped together in the survival analyses.Subsequently, we divided the cohort of patients with available histopathological data on steatosis into patients with MAFLD (n = 58) and without MAFLD (n = 89).A significant difference in TTR was noted between patients with MAFLD aetiology (median TTR: 38 months, 95% CI: 14.7-61.3)and patients without MAFLD (median TTR: 12 months, 95% CI: 8.0-16.0,p = 0.025, Figure 2).The times of OS were similar between groups.Univariable analysis showed a significantly shorter OS of patients with myosteatosis (33 vs. 23 months, HR 1.5, log-rank p = 0.020, Table 3), while other clinical variables associated with TTR and OS are listed in Table S3.
Multivariable analysis of all significant parameters from univariable analysis showed an independent prognostic value of lymphovascular invasion, vascular invasion and lymph node invasion for TTR, while an age >65 years and fresh frozen plasma transfusions were independently prognostic for OS (Table 4).Accordingly, an independent prognostic role of MAFLD for TTR and subcutaneous obesity and myosteatosis for OS was not confirmed.

| DISCUSSION
Postoperative recurrence of iCCA with impaired longterm survival are frequent and severe occurrences after curative-intent surgery, despite surgical and perioperative advances in the field. 26In this context, due to the comparative rarity of iCCA and the challenges of curativeintent liver resections, prognostic biomarkers are severely lacking.Recently, next-generation sequencing revealed molecular subtypes of iCCA that carry prognostic and predictive potential, 27 while the pathological criteria of lymph node and lymphovascular invasion are well-characterized prognostic factors across European and Asian patient collectives. 26,28At the same time, patient-or host-centred determinants of prognosis remain relatively unexplored in this rare tumour entity. 29Accordingly, we investigated tumour aetiology together with body composition in a large, homogenous European cohort of patients undergoing curative-intent liver resection for iCCA.
In this study, we were able to identify that a categorical body composition parameter -sarcopenia-was associated with perioperative outcome, but none of the body composition categories held independent prognostic value in multivariable TTR and survival analysis.As such, this study pointed towards an association of sarcopenia with perioperative morbidity, such as intraoperative transfusions, Clavien-Dindo ≥IIIb complications and an elevated post-operative CCI.While this observation is novel to iCCA, it is shared with curative-intent surgery for other gastrointestinal malignancies, such as HCC and pancreatic adenocarcinoma, 30,31 and is a result of malnutrition, systemic inflammation and overall catabolic processes. 32Furthermore, we noted inferior OS in univariable analysis in patients with myosteatosis, without observing differences when splitting the analysis by gender.Similarly, a study in a smaller palliative cohort across different CCA subtypes recently suggested a prognostic role of both myosteatosis and sarcopenia for overall survival.In the present study, a third of our cohort fulfilled the 2020 consensus MAFLD criteria. 14This subgroup had a higher incidence of body composition alterations, namely, sarcopenic, visceral and subcutaneous obesity, than patients without liver disease or with predominant fibrotic alterations.Recently, an association of NASH and elevated BMI as well as bioelectrical impedance analysis-assessed fat mass and skeletal muscle mass was published in patients undergoing bariatric surgery, 8 but data from oncological settings have not yet been reported.Because our cohort was composed of curatively treated patients without signs of systemic disease and with adequate preoperative performance status, severe cachexia, as observed in advanced gastrointestinal cancers, was less prevalent than it would be expected in palliative cohorts.
Patients with MAFLD had significantly longer TTR than patients without or with other underlying liver disease, while their OS was similar to the overall cohort.The aetiology of iCCA is relatively little studied in European/western cohorts and large meta-analyses are oftentimes skewed towards South Asian/Pacific populations with a predominant aetiological role of Opisthorchis viverrini. 3,34Across various CCA studies, the presence of cirrhosis, as well as hepatitis B and C, was not associated with shorter disease-free survival, 34 and the results on OS in patients with cirrhosis or chronical hepatitis B infection are disparate, with an apparent trend towards shorter OS in these patients. 35,36To date, the vast majority of iCCA studies investigating clinical prognostic factors do not report on either body composition, steatosis, or MAFLD as an iCCA aetiology or as a prognostic factor. 34,37Recently, an Italian study reported that patients fulfilling NASH criteria had an inferior OS after iCCA surgery but did not observe a difference in TTR between aetiologies.In fact, it remained unclear in this setting whether the accelerated mortality in the NASH group derived from cancer-related deaths or other causes, such as increased cardiovascular mortality. 5he role of MAFLD as a tumour aetiology is more apparent and better characterized in HCC, for which the estimated annual incidence in patients with NASH cirrhosis is from 0.5% to 2.6%. 38Recent years have brought a deeper understanding of NASH-HCC mechanisms, delineating impaired immune surveillance, 39 autoreactive immune cells, 40 systemic inflammation, 41 oxidative stress due to dysregulated lipid metabolism and dysbiosis. 42In contrast, fatty liver disease as a contributor to iCCA has only been characterized superficially.In this context, our study contributes to a growing body of evidence that patients with MAFLD-iCCA may constitute a distinct population for whom further metabolic, functional and immunological characteristics remain to be clarified.
The present study has a considerable sample size for European single-centre iCCA studies, with a homogenous surgically-treated patient collective.Nevertheless, the following shortcomings limit our conclusions: besides the exploratory single-centre design with a retrospective evaluation of body composition and MAFLD, the patient collective had, in part, post-operative adjuvant treatment based on negative pathological outcome factors, which was the recommended protocol before the universal recommendation for adjuvant chemotherapy from the BILCAP trial. 25Furthermore, we did not examine potential prognostic differences in tumour biology, such as gene variants, genetic mutations or molecular subtypes with a documented prognostic role in CCA and did not clarify their distribution across tumour aetiologies. 27,43n conclusion, we aimed to link the complex pathophysiology of disease aetiology and body composition by investigating liver disease of iCCA patients together with CT-derived fat and muscle parameters in the largest of these patient cohort to date.Herein, we found that MAFLD is frequent in iCCA patients, may hold potential prognostic value for time to recurrence and is significantly associated with alterations of body composition.In conjunction with the systemic changes observed in both MAFLD and in body composition pathologies, these findings illustrate the necessity of exploring systemic metabolic, performance and immunological changes in MAFLD-iCCA patients in future studies.and editing (equal).Dominik Geisel: Resources (supporting); writing -review and editing (equal).Timo Alexander Auer: Resources (supporting); writing -review and editing (equal).Uwe Pelzer: Resources (supporting); writing -review and editing (equal).Dominik Paul Modest: Writing -review and editing (equal).Nathanael Raschzok: Writing -review and editing (equal).Igor Maximilian Sauer: Resources (supporting); writing -review and editing (equal).Wenzel Schöning: Resources (supporting); writing -review and editing (equal).Frank Tacke: Funding acquisition (supporting); resources (supporting); writing -review and editing (equal).Johann Pratschke: Resources (supporting); writing -review and editing (equal).Georg Lurje: Conceptualization (equal); funding acquisition (lead); project administration (equal); resources (lead); supervision (lead); writing -review and editing (equal).

F I G U R E 1
Typical physiological body composition versus typical metabolic dysfunction-associated fatty liver disease (MAFLD)-associated body composition.Compared to physiological body composition (A), the typical MAFLDassociated changes of body composition depicted here (B) are an increased subcutaneous fat index (subcutaneous fat in light green), increased visceral fat area (visceral fat area, dark green/petrol), normal spinal muscle index (skeletal muscle index, red, no sarcopenia) and lower Hounsfield units-values (grey intramuscular areas), indicative of intramuscular fat accumulation.(C) Spearman r correlation matrix of MAFLD and body composition parameters (n = 173 patients included for all columns except MAFLD [n = 147]) (D) Distribution of body composition pathologies between non-MAFLD and MALFD patients with intrahepatic cholangiocarcinoma; chisquare test with only significant values shown.

Time to recurrence and overall survival
Preoperative patient characteristics stratified by the main clinicopathological aetiologies.
Abbreviations: BMI, body mass index; CCI, comprehensive complication index; G, Grade; iCCA, intrahepatic cholangiocarcinoma; ICU, intensive care unit; MAFLD, metabolic dysfunction-associated fatty liver disease; N, node; R, rest; T, tumour; UICC, Union Internationale Contre le Cancer.Note: Data presented as median and range if not noted otherwise.T A B L E 1 (Continued)T A B L E 2 (n = 75, median 22 months, 95% CI: 0.1-43.9months), compared to patients with fibrosis or no liver disease in the resected specimen (fibrosis: n = 44, 15 months, 95% CI: 10.4-19.6 months and no liver disease: n = 44, 12 months, 95% CI: 2.3-21.7 months, respectively, Figure 33 9,24 presented as mean and standard deviation if not noted otherwise.Body composition features were defined as follows and as partly described previously9,24: BMI -weight[kg]/height 2 [m 2 ], sarcopenia-SMI < 52.4 cm 2 /m 2 .for men and <38.5 cm 2 /m 2 for women, myosteatosis-<41 HU if BMI <25 kg/ m 2 and <33 HU if BMI ≥25 kg/m 2 , visceral obesity if VFA ≧100 cm 2 , subcutaneous obesity derived from the subcutaneous fat index, dichotomized at the upper tertile of the cohort (71.89 cm 2 /m 2 ), sarcopenic obesity defined as BMI >25 kg/m 2 and SMI ≦38.5 cm 2 /m 2 in women and ≦52.4 cm 2 /m 2 in men.Bold values indicate p values regarded as statistically significant.Based on ANOVA with post hoc Bonferroni correction, with p values given for comparisons to the no Liver Disease column.Bonferroni correction for three groups resulted in a level of significance α = 0.016.; **Based on chi-square test, with p values given for comparisons to the no Liver Disease column. *