Inadequate functional liver remnant after hepatectomy is the underlying pathophysiology for PHLF (Qadan et al., 2016). Orthotic liver transplantation is considered the best treatment for PHLF, but the shortage of liver donors and the strict inclusion criteria for transplantation are major limitations of this treatment (Søreide & Deshpande, 2021). Therefore, supportive treatment remains the standard of care for PHLF (Søreide & Deshpande, 2021; Qadan et al., 2016). The main components of supportive treatment are early detection and initiation of general care for critically ill patients with focus on organ support, sepsis control, and optimal environment provision for liver generation (Søreide & Deshpande, 2021). Rigorous preoperative assessment and preoperative optimization of patients undergoing hepatectomy are the keys for avoiding PHLF. Preoperative assessment can be carried out using the following parameters: CT volumetric analysis, Child-Pugh classification, evidence of significant portal hypertension, and ICGR15 (Qadan et al., 2016; Walcott-Sapp & Billingsley, 2018). However, PHLF can still occur despite preoperative preparations because of progressively aggressive treatment approaches such as extended hepatectomy, surgery in elderly patients, and hepatectomy following hepatotoxic neoadjuvant chemotherapy (Qadan et al., 2016). The reported incidence of PHLF in current literatures is 1.2–32% (Søreide & Deshpande, 2021; Qadan et al., 2016). From our study, the incidence of PHLF is 3.1%, which is comparable to previous studies.

There are many previous reports on preoperative factors which affects PHLF (Dasari et al., 2019; Chin et al., 2020; Shen et al., 2019; Ye et al., 2020). However, hepatectomy outcomes are also influenced by intraoperative events (Bagante et al., 2019; Grat et al., 2013). Thus, early postoperative parameters would be more accurate than preoperative factors alone for predicting of PHLF (Grat et al., 2013). In the present study, we analyzed preoperative, intraoperative, and postoperative factors. The results showed that a preoperative factor (ICG R15), two intraoperative factors (extent of hepatectomy, blood loss), and some early postoperative parameters (AST, and INR on POD1) were independently associate with PHLF.

According to the ISGLS consensus and the ’50–50’ criteria, PHLF can only be diagnosed on POD5 because the diagnosis is based on biochemical laboratory data taken on POD5 or later (Søreide & Deshpande, 2021). However, waiting until POD5 to make a diagnosis may delay management of patients with PHLF. Therefore, the ability to promptly predict PHLF and deliver early management is crucial to improve the short-term outcomes following hepatectomy (Grat et al., 2013). Regarding serum transaminase, the present study revealed that the serum transaminase levels peaked on POD1. Consistently, Higaki et al. (Higaki et al., 2018) examined the association between ischemic parenchymal volume of the liver after hepatectomy and serum transaminase elevation, and found that serum transaminase level in their cohort also peaked on POD1. In addition, the present ROC analysis of serum transaminase level from POD0 to POD3 revealed that the transaminase level on POD1 has the highest yield for prediction of PHLF. These findings are consistent with those in a study conducted by Grat et al. (Grat et al., 2013) who analyzed POD1 serum biochemical parameters in patients after major liver resection for colorectal metastases. They found that an AST cut-off point of 798 U/L on POD1 can stratify patients into low-risk and high-risk groups for 90-day mortality. Olthof et al. (Yoshino et al., 2021) also retrospectively studied patients who underwent liver resection and found that peak AST level, which normally occurs within 24 h after hepatectomy, of > 828 U/L associated with increased postoperative morbidity and mortality. The higher cut-off point found by Grat et al. compared with the value of 250 U/L in the present study can possibly be explained by the difference in primary outcomes. Specifically, the primary outcome in present study is PHLF while the primary outcomes in the other study were overall mortality and morbidity.

In contrast to the present findings, Bhogal et al. (Bhogal et al., 2016) reported that serum ALT on POD1 was not predictive of post-hepatectomy morbidity and mortality. However, they did not investigate the ability of elevated AST as a predictor. In the present study, AST was associated with PHLF, but not ALT. Another contradictive work is a study by Boleslawski et al. (Boleslawski et al., 2014), who found that post-hepatectomy serum AST and ALT were not independently associate with morbidity. However, their definition of postoperative morbidity was inclusive of all manner of complications, including pulmonary complications, hemorrhage, wound infections, with only 3% of the reported complications were PHLF.

Preoperative AST was found to be better than preoperative ALT for predicting outcomes following hepatectomy in previous reports (Ye et al., 2020; Liu et al., 2020; Saadat et al., 2021; Shi et al., 2021). Liu et al. (Liu et al., 2020) and Saadat et al. (Saadat et al., 2021) conducted large-population studies on the preoperative factors of PHLF, and found that preoperative AST > 40 U/L was associated with PHLF. Postoperative AST had also reported to be more strongly associated with postoperative outcomes than postoperative ALT (Olthof et al., 2016; Grat et al., 2013). Olthof et al. (Olthof et al., 2016) reported that peak postoperative AST, but not ALT, was associated with the overall postoperative morbidity and mortality. Consistently, Grat et al. (Grat et al., 2013) reported that postoperative AST, INR, and bilirubin were associated with 90-day mortality. Meanwhile, AST alone was not significantly associated with overall hepatic complications, including PHLF, delayed recovery of liver function, bile leakage, and subphrenic abscess. Yu et al. (Yu et al., 2018) reported that among PHLF patients, sustained ALT elevation beyond POD1 was associated with increased mortality. However, the observation was only made within a PHLF cohort with no comparison against a non-PHLF cohort. To the best of our knowledge, there are no previous reports on the association of peak postoperative serum transaminase levels with PHLF as the primary outcome.

From the present result with PHLF is the primary outcome, only serum AST > 250 U/L on POD1 had significant association with PHLF, while ALT did not. An explanation for the association of AST with PHLF may be hepatocellular injury, which can be caused by multiple factors (Søreide & Deshpande, 2021; Qadan et al., 2016; Murtha-Lemekhova et al., 2021). Aminotransferases (also known as transferases) are enzymes involved in the transfer of amino groups from aspartates to ketoglutaric acid and are markers of hepatocellular injury (Robles-Diaz et al., 2015). Elevated levels of AST and/or ALT, ALP, and bilirubin can suggest the occurrence of hepatocellular injury and are associated with increased liver-related mortality in the general population, as well as in post-hepatectomy patients (Kwo et al., 2017). In addition, plasma transaminase levels are measured after liver surgery as markers of hepatocellular injury and have been used as endpoints in numerous previous clinical trials (Murtha-Lemekhova et al., 2021; Beck-Schimmer et al., 2012; Nguyen et al., 2019). However, the mechanism of post-hepatectomy hepatocellular injury dependent on multi-factors. Some of these factors were (1) preexisting conditions (such as hepatic steatosis, fibrosis, or cirrhosis) that can reduce liver regeneration capacity, (2) vascular inflow occlusion during the operation, and (3) ongoing injury due to hyperperfusion in a proinflammatory environment (Murtha-Lemekhova et al., 2021). Thus, further large-population prospective studies on the relationship between serum AST or ALT and PHLF should be conducted to confirm the findings of the present study.

The present results also showed that INR on POD1 was independently associated with PHLF. INR was shown to temporarily increase after hepatectomy in some patients, with a peak on POD1 or 2 (Balzan et al., 2005). Usually, INR gradually normalize on POD3 or 4 (Martin 2nd et al., 2003). However, those with PHLF had sustained elevation after POD5, leading to the currently accepted criteria for diagnosis of PHLF (Rahbari et al., 2011). Although non-PHLF patients can have elevated INR level on POD1, high INR on POD1 was demonstrated by Roberts et al. (Roberts et al., 2013) to signify increasing severity of PHLF. These findings are similar to the present findings and suggest that INR on POD1 can be an early warning for physician that PHLF may be underway.

There are a few limitations to the present study. First, because of its retrospective nature, some selection bias may have been present. Second, the characteristics of patients undergoing hepatectomy can be heterogenous, and there was a lack of data on the degree of background liver disease, postoperative serum glucose, and serum lactate level in the study. Third, the small population of the study, which there were relatively small number of PHLF patient which would affecting the power of the analysis. In addition, there were only four patients who died from PHLF, and thus multiple logistic regression analyses could not be performed to evaluate the use of the identified parameters for prediction of mortality.