This study investigated the predictors of unexpected POLF after conversion hepatectomies and long-term preoperative chemotherapy in patients with initially unresectable CLMs. The results showed that splenomegaly after preoperative chemotherapy was the only independent predictive factor for unexpected POLF in conversion hepatectomy according to multivariate logistic regression analysis. By contrast, no preoperative patient characteristics were significantly associated with splenomegaly. In the analysis limited to the splenomegaly group, the liver resection area, operative time, and blood loss were significant predictors of POLF. This report clearly demonstrates the important association between splenomegaly after preoperative chemotherapy for CLMs and POLF associated with chemotherapy-related hepatotoxicity.
Objective criteria consisting of the ICG clearance test and FLRV, such as Makuuchi’s criteria, are widely used as predictors of POLF, and their usefulness has been reported20, 21. In this study, conversion surgery was performed only in patients using Makuuchi’s criteria; however, the POLF rate was 14%, which was higher than that in previous reports on liver resection for CLMs22, 23. In cases such as initially unresectable CLMs in which chemotherapy-related hepatotoxicity due to long-term chemotherapy is expected, conventional criteria based on the ICG clearance test and FLRV alone may not adequately predict POLF.
Severe SOS, a form of chemotherapy-related hepatotoxicity, is reportedly associated with postoperative morbidities, including POLF9, 24, 25. SOS is known to be associated with oxaliplatin-based chemotherapy26, and POLF should be considered, especially in cases of conversion surgery for initially unresectable CLM, where the efficacy of regimens containing oxaliplatin has been reported27. However, in clinical settings, predictive factors for POLF that can be identified before surgery are more useful than pathological factors such as SOS. Splenomegaly and hypointensity on EOB-MRI have been previously reported as factors associated with SOS12, 14, 28. Regarding preoperative patient-specific factors, including splenomegaly and reticular hypointensity on EOB-MRI, only splenomegaly was a significant predictor of POLF in this study. One possible reason why reticular hypointensity on EOB-MRI was not a predictive factor is that reticular hypointensity on EOB-MRI could not be quantitatively assessed and had to be judged qualitatively; therefore, it may not have predicted the severe SOS that would have resulted in POLF. Therefore, the establishment of a quantitative evaluation method for hypointensity on EOB-MRI is warranted.
Splenomegaly after long-term chemotherapy for CLM is believed to result from portal hypertension caused by oxaliplatin-induced SOS12. The greater intraoperative blood loss in the splenomegaly group in this study may have been due to portal hypertension29. The significantly higher incidence of POLF in the splenomegaly group supports an association between splenomegaly and SOS. In contrast, no other preoperative factors, including total number of chemotherapy cycles, were associated with splenomegaly. In the present study, in which the oxaliplatin regimen was used in almost all patients, it was not possible to determine the factors that led to the development of splenomegaly. Although measuring splenic size is time-consuming, it may be the only useful method for predicting POLF in conversion surgery after long-term chemotherapy for initially unresectable CLM. In previous reports, spleen size was proportional to the cumulative dose of oxaliplatin and tended to increase to 135% or more when the total dose of oxaliplatin exceeded 800mg/m2 12. Measurement of splenic size may be recommended, especially in patients where the cumulative preoperative dose of oxaliplatin exceeds 800mg/m2.
Although the risk of POLF is significantly higher when the spleen enlarges after long-term chemotherapy for CLMs, liver resection is the only potentially curative treatment for patients with CLMs. In the analysis limited to the splenomegaly group, among the surgery-related factors, operative time, blood loss, and liver resection area were significant predictors of liver failure. Of these, operative time and blood loss have been previously reported as risks for POLF30, which is the result of complex liver resection. Only the liver resection area was calculated preoperatively using a three-dimensional image analysis software (SYNAPSE VINCENT, Fujifilm Medical Co., Ltd., Tokyo, Japan). Importantly, more than 80% of patients with a liver resection area of 100cm2 or more, which also implies complex hepatic resections, develop POLF. Thus, in patients with splenic enlargement of 35% or more and requiring complex hepatic resection with a liver resection area of > 100cm2, careful decisions should be made before hepatectomy, including two-stage hepatectomy31 or postponement of surgery for liver functional recovery from chemotherapy-induced damage.
This study has several limitations. First, it had a retrospective design and included a relatively small number of patients from a single institution. There were only 15 patients with POLF in the entire cohort. There is also concern about selection bias based on our institutional referral pattern, patient population, and tumor board recommendations with respect to the definition of initially unresectable CLM and the decision for conversion surgery. Further analyses of large cohorts, including multicenter studies, are required. Second, the current study did not investigate pathological details. Therefore, it is unclear how severe SOS occurs in patients with splenomegaly or POLF. However, the focus of this study was to determine the association between preoperative factors and POLF, and the examination of pathological factors, including SOS, was beyond the scope of this study. Third, in this study, the FLRV was measured using the SYNAPSE VINCENT software, but ischemic and congested areas were not considered. Because we prefer to perform parenchymal-sparing hepatectomy rather than major anatomical hepatectomy32, 33, ischemic and congested liver areas increase compared with anatomical resection. Although it has been reported that functional FLRV is more closely related to postoperative liver function than simple FLRV34, 35, accurate volume simulations for functional FLRV are difficult because of non-anatomically complicated liver resection.