Evaluation and management of hepatic injury induced by oxaliplatin-based chemotherapy in patients with hepatic resection for colorectal liver metastasis
Department of Surgery, Institute of Health Biosciences, The University of Tokushima Graduate School, Tokushima, Japan
Correspondence: Dr Yuji Morine, Department of Surgery, Institute of Health Biosciences, The University of Tokushima Graduate School, 3-18-15 Kuramoto-cho, Tokushima 770-8503, Japan. Email: email@example.com
Patients with colorectal liver metastasis (CRLM) can be cured with surgical resection. Recent advances in systemic chemotherapy, including molecular target agents, can be used to introduce “conversion surgery” and achieve R0 resection even in patients with initially unresectable CRLM. Furthermore, neoadjuvant chemotherapy also tries to be applied in patients with resectable CRLM to maximize the remnant liver and reduce the residual micrometastasis before surgery. The development of chemotherapy-induced hepatic injuries is increasingly being recognized, including sinusoidal obstructive syndrome (SOS), steatosis, steatohepatitis and biliary sclerosis. Especially, oxaliplatin (L-OHP)-based chemotherapy in clinical settings appears to be primarily associated with SOS. Various reports have tried to demonstrate the rationale of the correlation between L-OHP-based chemotherapy and SOS for the following hepatic surgery. While we can recognize that this pathophysiological disadvantage leads to hepatic dysfunction and the increasing postoperative morbidity, the essential part of this problem including clinical disadvantage, onset mechanism, evaluation systems, and targeted agents for prevention and treatment of SOS continue to be unclear. In this review, we summarize the current experience with hepatic injury induced by L-OHP-based chemotherapy, focusing on SOS-based on clinical and experimental data, in order to assist in the resolution of these identified factors. Finally, the need for reliable methods to identify the risk of SOS, to evaluate SOS status and to predict the safety of surgical treatment in patients with chemotherapy prior to surgery will be emphasized.
IN PATIENTS WITH colorectal cancer, liver metastasis is one of the most significant prognostic factors. Generally, 15–25% of cases have colorectal liver metastasis (CRLM) at diagnosis.[1, 2] Furthermore, CRLM occurs in 25–50% of cases with the resection of primary colorectal tumor over 3 years.[3-5] Hepatic resection is accepted as the only treatment contributing to the long-term survival and cure of patients with CRLM. However, only 15–20% of patients with CRLM are considered candidates for hepatic resection at the time of presentation.[7-10] The significance of other tumor destruction modalities, such as radiofrequency ablation, remains controversial.
Of those patients who undergo hepatic resection, there are at least two categories of patients with CRLM. The first category is clearly or potentially resectable at the time of presentation. The second category is initially unresectable, but convertible to be resectable after treatment with anticancer agents including molecular targeted agents, which we refer to as “conversion surgery”. The purpose of neoadjuvant chemotherapy for resectable CRLM is to downsize CRLM lesions and maximize the remnant liver as well as to reduce the residual micrometastasis, while less extensive resections can be carried out in keeping with the curative intent. However, until now, the role of neoadjuvant therapy prior to the resection of CRLM is not yet proven and remains controversial. The largest prospective trial consisted of 364 patients with less than five initially resectable CRLM (European Organization for Research and Treatment of Cancer Intergroup 40983 trial) randomized with perioperative chemotherapy (four to six preoperative and six postoperative cycles of FOLFOX4) or surgery alone, and showed a clinical benefit in 3-year progression-free survival (36.2% vs 28.1%) but not in 5-year overall survival. The FOLFOX regimen may reduce the risk of events in terms of progression-free survival but not necessarily improve long-term survival compared with surgery alone in eligible and initially resectable patients. On the other hand, regarding initially unresectable CRLM, the “conversion surgery” strategy has been widely used and accepted. Actually, 5-fluorouracil (5-FU)/leucovorin (LV) plus oxaliplatin (L-OHP); FOLFOX or irinotecan (Iri); FOLFIRI or combination of both; FOLFOXIRI with or without molecular-targeted agents as preoperative strategy have recently achieved higher conversion rates and better clinical outcomes.[13-20] Particularly in L-OHP-based chemotherapy, the conversion rate ranged 7–51% in patients with unresectable CRLM. The effectiveness of triple combination chemotherapy, FOLFOXIRI, for patients with initially unresectable CRLM has been reported to have an improved response rate (60% vs 34%) and higher R0 resection rate among patients with CRLM only compared with the FOLFIRI regimen (36% vs 12%). Regarding the addition of molecular-targeted agents such as bevacizumab, the First Beat and the randomized phase-III NO16966 trial demonstrated that 225 out of 1914 patients (11%) with unresectable liver metastasis could undergo curative intent surgery, and R0 resection was achieved in 173 out of 225 patients (76.9%).[19, 20]
Hence, as a regimen using more powerful chemotherapy is developed in one of the multidisciplinary treatments for CRLM, hepatic toxicity is likely to be exacerbated, such as sinusoidal obstructive syndrome (SOS), steatosis (non-alcoholic fatty liver disease), steatohepatitis (non-alcoholic steatohepatitis) and biliary sclerosis (Table 1). Especially, L-OHP-based chemotherapy with molecular targeting agents, such as bevacizumab, cetuximab or panitumumab, plays a central role of initial chemotherapy for unresectable colorectal cancer in Japanese Society for Cancer of the Colon and Rectum guidelines and it was well known that L-OHP-based chemotherapy appears to be primarily associated with SOS. In this review, we attempt to summarize the current experience with hepatic injury induced by L-OHP-based chemotherapy focusing on SOS.
Table 1. Type of hepatic injury induced by anti-colon cancer agents
Pattern of liver damage
Sinusoidal obstructive syndrome
5-FU + folinic acid
SOS Induced by L-OHP-Based Chemotherapy
Mechanism and characteristics
VENO-OCCLUSIVE DISEASE INDUCED by a lethal poisoning of pyrrolizine alkaloids in humans was first reported in 1920, and the abnormalities of the central vein and the centrilobular localization of the damage were recognized. In 1999, De Leve et al. established the rat hepatic veno-occlusive disease model induced by monocrotaline. In this article, congestion and dilatation of the hepatic sinusoids, discontinuity in the sinusoidal membrane and collagen deposits in the perisinusoidal spaces were proven as histopathological features. This pathophysiology, which has an impressive macroscopic character “blue liver”, has been well known in SOS (Fig. 1). Recently, the induction of L-OHP-based chemotherapy for advanced colorectal cancer has developed the frequent onset of SOS. SOS is defined as a disruption of the sinusoidal membrane, collagenization of the perisinusoidal space and sinusoidal dilatation. A part of the molecular pathophysiology of SOS involves the depolymerization of F-actin in sinusoidal endothelial cells, which leads to the increased expression of matrix metalloproteinase (MMP)-9 and MMP-2 by sinusoidal endothelial cells. As morphological change, it was microscopically revealed that red blood cells penetrated under the sinusoidal endothelial cell barrier and dissected the endothelium off the extracellular matrix in the Disse space. At the same time, anticancer agents (L-OHP or Taxan with 5-FU) made it possible to induce oxidative stress.[25, 26] These SOS can be associated with fibrosis and consequent portal hypertension and liver dysfunction.
In 2004, Rubbia-Brandt et al. published the first clinical series of SOS in non-tumorous liver induced by L-OHP administration as preoperative chemotherapy. In this article, they defined a semiquantitative classification of SOS as follows: 0, absent; 1, mild (centrilobular involvement limited to one-third of the lobular surface); 2, moderate (centrilobular involvement extending in two-thirds of the lobular surface); and 3, severe (complete lobular involvement). In their series, 34 of 43 patients (79.1%) with L-OHP developed SOS. Up to now, several investigators have reported the correlation between SOS and preoperative administration of L-OHP (Table 2).[27-37] These studies showed that 8.3–51.6% of patients who received the administration of L-OHP with 5-FU-based chemotherapy developed grade 2–3 SOS. Regarding the correlation between the number of cycles of L-OHP treatment and the onset of SOS, Kishi et al. showed that sinusoidal injury was recognized in 46 of 79 patients (46%) and 22 of 38 patients (58%) after short (1–8 cycles) and long (≥9 cycles) duration preoperative FOLFOX, respectively. In most studies, there was no correlation between the cumulative dose of L-OHP and the presence or severity of SOS. Nakano et al. only asserted that at least six cycles of L-OHP administration significantly correlated with the onset of SOS. Meanwhile, the defined discontinuous duration of treatment with L-OHP for regression of SOS has remained unclear. Though reports regarding this problem are few, Nakano et al. also reported that the interval between the cessation of L-OHP-based chemotherapy and hepatic resection was significantly longer in patients without SOS (6.5 ± 1.2 months) than in those with SOS (3.6 ± 0.8 months). In contrast, the persistence or progression of SOS 32 months after the cessation of L-OHP administration has been reported. Though continuous duration of SOS induced by L-OHP is not still proven, most investigators have adopted 1–3 months as the time interval between cessation of chemotherapy and hepatic resection.[29, 30, 34, 37]
Table 2. Summary of clinical evaluation between oxaliplatin based chemotherapy and hepatic injury (high grade 2–3 SOS)
Influence of intra- and post-operative complication
Schiffer et al. demonstrated that the presence of SOS the induced impairment of liver regeneration, obstruction of hepatic microcirculation, increased portal pressure and decreased bile flow associated with a decreased bile excretion of 153gadobenate dimeglumine, after partial hepatectomy in this rat model of monoclotarine-induced SOS, suggesting that L-OHP may augment hepatic regeneration following major hepatic resection with increased perioperative complications. There is only one clinical report regarding liver regeneration in patients who underwent portal vein embolization after L-OHP-based chemotherapy. SOS inhibited the future remnant liver hypertrophy after portal vein embolization and induced postoperative liver failure.
In clinical settings of patients with hepatic resection, some investigators evaluate the implication of perioperative complication in patients with L-OHP treatment (Table 2). Though there were more than a few reports indicating that morbidity risks did not increase after surgery, they were not investigating SOS cases in particular, but in all patients including those without SOS after preoperative chemotherapy. Aloia et al. assessed the relationships between preoperative chemotherapy, and surgical outcomes by comparing patients treated with either FU and LV or FU, LV, and L-OHP and non-treated patients. According to their findings, patients with L-OHP tended to have a higher incidence of morbidity compared to patients without any chemotherapy. Furthermore, they demonstrated that the mean transfusion rate for packed red blood cells was four-fold higher in the patients with L-OHP compared to patients without any chemotherapy. Mehta et al. also noted a similar assertion that intra-operative blood transfusion requirement was higher in patients with L-OHP-based chemotherapy (34.2%) than in patients without chemotherapy (18.6%). Nakano et al. further investigated perioperative liver dysfunction including surgical outcomes according to the presence or absence of L-OHP-induced SOS. In their series of 90 patients, preoperative indocyanine green retention rate at 15 min (ICG-R15) (9.7 ± 0.7% vs 7.6 ± 0.8%; P = 0.026) and postoperative maximum total bilirubin levels (33.2 ± 4.5 vs 22.0 ± 1.7 µM/L; P = 0.023) were significantly higher, and hospital stay was significantly longer in patients with SOS. Particularly in patients with a major hepatectomy, SOS was significantly associated with higher morbidity (40.0% vs 6.3%; P = 0.026) including 10% liver insufficiency and longer hospital stay (17.0 ± 1.8 vs 10.9 ± 0.9 days; P = 0.006). Considering these findings, we must pay attention to perioperative complication particularly in major hepatic resection for patients with severe SOS induced by treatment with L-OHP-based chemotherapy.
The recent strong chemo-regimen FOLFOXIRI containing 5-FU, L-OHP and Iri has greater efficacy in down-staging unresectable colorectal liver metastasis. Masi et al. reported that this regimen had a 70% response rate and allowed an R0 surgery in 19% of unselected patients with initially unresectable metastatic colorectal cancer. Among these patients undergoing hepatic resection, the incidence of postoperative complication was 27% without mortality. In addition, they further indicated that all patients developed SOS, but no grade 3 SOS was found (grade 2; 48%).
Prediction and evaluation
Some investigators explored several parameters to evaluate and predict the SOS state of the liver after chemotherapy (Table 3).[32, 40-48] As a potential consequence of SOS, sinusoidal injury associated with L-OHP increased resistance to blood flow between the portal and hepatic venous systems. Then, portal hypertension developed splenomegaly, persistent thrombocytopenia, and bleeding of esophageal and hemorrhoidal varices. Overman et al. evaluated the relationship between L-OHP-induced hepatic sinusoidal injury, increased volume of spleen and the subsequent development of thrombocytopenia. In their study, increased volume of spleen correlated with cumulative L-OHP dose and higher rates of thrombocytopenia. They suggested that 50% increase in spleen volume was a predictor of higher histological grades of sinusoidal injury. Miura et al. also evaluated the correlation between splenic volume, aspartate aminotransferase to platelet ratio (APR) and sinusoidal injury, and demonstrated that 30% increases in spleen volume due to six cycles of FOLFOX could predict grade 2–3 SOS, if the APR before FOLFOX was 0.17 or higher.
Table 3. Summary of evaluation and risk of SOS induced chemotherapy
a0, none; 1, fine reticulations visible on a minority of sections; 2, diffuse reticulations or localized, coalescent areas of high signal; and 3, diffuse reticulations visible on all sections or densely coalescent areas of high signal visible on multiple sections.
bReticular hypointensity: 1, definitely not present; 2, probably not present; 3, equivocal; 4, probably present; 5, definitely present.
APR, aspartate aminotransferase to platelet ratio; AST, aspartate aminotransferase; Cx, chemotherapy; EOB-MRI, ethoxybenzyl multiple resonance imaging; Hx, hepatectomy; ICG-R15, indocyanine green retention rate at 15 min; SOS, sinusoidal obstructive syndrome; SPIO, superparamagnetic iron oxide.
In other imaging modalities for detecting the liver injury, Oki et al. introduced the usefulness of elastography (FibroScan; Echosens, Paris, France), and demonstrated that the stiffness of the liver was increased after chemotherapy within 48 h and that the hepatic stiffness gradually increased after repeated FOLFOX4 in some cases with liver injury. The observed stiffness of the liver may cause portal hypertension and splenomegaly after repeated chemotherapy. Regarding magnetic resonance imaging (MRI), there were two interesting reports using contrast radiography. Ward et al. suggested that superparamagnetic iron oxide-enhanced T2-weighted gradient echo imaging before hepatectomy was useful to evaluate subclinical SOS with L-OHP treatment. They graded the presence and severity of abnormal areas of reticular hyperintensity on a 4-point ordinal scale (0, none; 1, fine reticulations visible on a minority of sections; 2, diffuse reticulations or localized, coalescent areas of high signal; and 3, diffuse reticulations visible on all sections or densely coalescent areas of high signal visible on multiple sections), and defined that a severity score of 2 or 3 was considered positive for SOS.
Recently, gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid has been well known as a newly available hepatobiliary contrast agent, which had both dynamic and hepatobiliary phase imaging ability for the qualitative diagnosis of liver tumors. Shin et al. also graded the presence of reticular hypointensity on hepatobiliary phase images using a 5-point original scale (1, definitely not present; 2, probably not present; 3, equivocal; 4, probably present; 5, definitely present). They defined that their confidence score of 4 or 5 was considered positive diagnosis for SOS. Ethoxybenzyl MRI may be established as the most useful imaging modality to evaluate both tumor detection or diagnosis and severity of liver injury induced by preoperative chemotherapy as a “one-stop shop”.
Regarding predictive factors of sinusoidal injury induced by preoperative chemotherapy, Nakano et al. reported that using ICG-R15 and the aspartate aminotransferase level before hepatectomy they were able to predict the occurrence of SOS. Soubrane et al. indicated high preoperative APR score as the most reliable indicator. In their report, the mean interval between chemotherapy and surgery was 7.2 weeks. Brouquet et al. demonstrated that serum γ-glutamyltransferase level before chemotherapy was an independent high-risk factor of SOS. Interestingly, they also suggested that aspirin intake was an independent factor associated with a reduced risk for SOS.
Prevention and treatment
Recently, bevacizumab, an antibody against vascular endothelial growth factor (VEGF), which is an angiogenesis inhibitor, approved by the US Food and Drug Administration for the treatment of metastatic colorectal cancer in 2004, has been used with L-OHP-based chemotherapy to obtain further improvement of outcome for patients with advanced colorectal cancer.[13-20] Conveniently, some authors have asserted that anti-VEGF had a protective effect based on the result that VEGF is one of the causative cytokines for SOS.[24, 49] On the other hand, another report has asserted that the inhibition of VEGF receptors had an adverse effect on liver regeneration in the murine experimental model. Clinically, Aussilhou et al. demonstrated that bevacizumab impaired hypertrophy of the future remnant liver after portal vein embolization, particularly in patients who received six cycles or more of bevacizumab treatment, and warned that major liver resection should be considered with caution in patients who have received bevacizumab. Most recently, encouraging data has been published based on hepatic volumetric analysis of bevacizumab-treated patients who underwent hepatectomy. This study included 41 patients who underwent major hepatectomy (≥3 segments) with more than four cycles of neoadjuvant chemotherapy including less than 3 months of bevacizumab treatment and compared the matched 41 patients administrated the equivalent systemic chemotherapy without bevacizumab. In preoperative characteristics, patients with bevacizumab received a median of six cycles of chemotherapy that was discontinued for a median of 52 days before hepatic resection (median cessation interval of bevacizumab, 65 days). As a result, postoperative liver regeneration was not influenced by the type of hepatic resection, the number of courses of chemotherapy (≥6 cycles or ≥10 cycles) or age factor (>65 years old), and no intergroup differences in overall morbidity or postoperative liver failure were observed. Our experimental study with rats also revealed that preoperative bevacizumab (7 days before hepatectomy) administration significantly increased liver regeneration, and induced heat shock protein 70 mRNA, which had protective effects for organ injury, just before hepatectomy (Fig. 2).
Actually, several reports are beginning to emerge on the safety and efficacy of bevacizumab in patients treated with preoperative chemotherapy and surgical treatment (Table 4).[30, 35, 37, 52-57] Most investigators have indicated that bevacizumab reduced the severity of SOS as well as decreased the incidence of SOS induced by preoperative chemotherapy, and did not increase the risk of morbidity and mortality. Ribero et al. first reported the protective effect of bevacizumab for grade 2–3 SOS in patients receiving bevacizumab preoperatively in patients receiving chemotherapy with 5-FU and L-OHP. Kesmodel et al. also reported a surgical series analyzing the safety of preoperative chemotherapy including bevacizumab. In their series, they confirmed the effect of bevacizumab to reduce the severity of SOS, as well as the fact that there was no difference in morbidity or mortality after hepatic resection between the presence and absence of bevacizumab treatment. Reddy et al. further demonstrated that preoperative bevacizumab treatment was associated with less blood loss (median, 425 vs 600 mL) and lower red blood cell transfusion rates (43.9% vs 23.1%) after hepatic resection and also that the addition of bevacizumab to preoperative L-OHP/Iri did not increase morbidity after hepatic resection, if bevacizumab administration was discontinued at least 8 weeks before hepatic resection.
Table 4. Summary of safety and efficacy of bevacizumab on preoperative chemotherapy for hepatic resection
In our experience of patients with L-OHP-based chemotherapy for unresectable colorectal liver metastasis, the incidence and severity of SOS was significantly lower in patients with bevacizumab (n = 9) than in patients without bevacizumab (n = 7) (grade 2–3, 22.2% vs 71.4%; P < 0.05). Furthermore, the change in spleen volume and serum hyaluronic acid, which was used to assess the damage of sinusoidal endothelial cells, were significantly lower in patients with bevacizumab compared with patients without bevacizumab (changes in spleen volume, 110.3 ± 27.5% vs 146.3 ± 34.2%, P < 0.05; serum hyaluronic acid levels, 33.6 ± 21.2 vs 124.5 ± 34.0 ng/mL, P < 0.05) (Fig. 3).
Regarding the relationship between the cumulative dose of bevacizumab and postoperative complications, Anne et al. reported that the addition of bevacizumab with L-OHP-based chemotherapy may protect against sinusoidal injury without increasing the risk of morbidity, and neither duration of chemotherapy (1–5 vs ≥6 cycles) nor the interval between cessation (5 weeks) of chemotherapy and hepatic resection were associated with postoperative complications despite the bevacizumab treatment. Meanwhile, in the series published by Kishi et al. patients with short (1–8 cycles) or long (≥9 cycles) preoperative chemotherapy with FOLFOX with or without bevacizumab were analyzed. In this article, they revealed that nine cycles or more of FOLFOX was the only independent prognostic factor for postoperative liver insufficiency, and concluded that the addition of bevacizumab may significantly reduce the incidence of SOS, but did not impact on the rate of postoperative liver insufficiency in patients with extended duration of chemotherapy.
Other possible treatments in experimental model
In experimental studies for reduction of SOS, several investigators have tried some possible agents except bevacizumab for the monoclotarine-induced SOS model (Table 5).[58-62] Narita et al. demonstrated that preoperative upregulation of heme oxygenase-1 by a phosphodiesterase-III inhibitor was effective for maintenance of the sinusoidal lining in sinusoidal endothelial cells and blockage of monoclotarine-induced SOS, and resulted in a significant improvement in survival rate after 70% hepatectomy. In addition, they also revealed that Japanese kampo medicine “Dai-kenchu-to” attenuates monoclotarine-induced liver injury through the preventing neutrophil-induced liver injury through blockage of upregulation of cytokine-induced neutrophil chemoattractant and intracellular adhesion molecule-1 mRNA level. Additionally, Ezzat et al. reported that flavonoid (monoHER2) prevented portal hypertension and hepatic injury including MMP-9 suppression. Nakamura et al. reported that sorafenib, which was a multiple tyrosine kinase receptor inhibitor targeting Ras/Raf kinase that also inhibits certain tyrosine kinases, reduced the severity of monoclotarine-induced SOS in rats through suppression of MMP-9 and c-Jun N-terminal kinase (JNK) activity. Also, it was reported that sesame oil attenuated SOS by decreasing the recruitment of inflammatory cells including Kupffer cells and neutrophils, downregulating MMP-9 and upregulating tissue inhibitor of matrix metalloproteinase-1 expression. All of these agents may be considered for possible clinical application in the near future.
Table 5. Summary of inhibitors for SOS in experimental models
IN THIS REVIEW, the current recognition of hepatic injury induced by L-OHP-based chemotherapy was summarized, particularly focusing on SOS. Even today, the pathophysiological mechanism of L-OHP-induced SOS remains unclear. Therefore, clinical disadvantage, evaluation system and targeted agents for preventing and reduction of SOS are yet to be fully elucidated. At the present stage, the algorithm to deepen understanding of the current status of SOS is shown in Figure 4. In future, further investigation should be conducted based on the molecular biology and pathology combined with drug targeting systems, which can provide some new ideas for the treatment of SOS.