Efficacy and Safety of TACE Combined with Regorafenib versus TACE in the Third-Line Treatment of Colorectal Liver Metastases

Background The liver is the most common site of metastasis in colorectal cancer. In patients with unresectable colorectal liver metastases, the 5-year survival rate is less than 5%. Many patients with colorectal liver metastases require effective subsequent therapy after the failure of standard first-line/second-line therapy. The purpose of this study is to investigate the efficacy and safety of TACE combined with Regorafenib versus TACE in the third-line treatment of patients with colorectal liver metastases. Method The clinical data of 132 patients with colorectal liver metastases were collected. There were two groups: TACE + Regorafenib group (N = 63); TACE group (N = 69). TACE uses CalliSpheres® drug-loaded microspheres (loaded with irinotecan). Regorafenib is administered at a dose of 120 mg once daily. If the patient is severely intolerable, the regorafenib dose is adjusted to 80 mg once daily. Primary study endpoints were (1) to evaluate the tumor response, ORR, and DCR and (2) to evaluate OS and PFS in the two groups. Secondary study endpoints were (1) to compare the performance status, CEA, CA19-9 after treatment between the two groups and (2) to compare the incidence of adverse events between the two groups. Results There were significant differences in tumor response, ORR, DCR, OS, and PFS after treatment between the two groups. TACE combined with the Regorafenib group versus the TACE group: ORR (57.1% vs 33.3%), DCR (82.5% vs 68.1%), mOS (18.2 months vs 11.3 months), and mPFS (8.9 months vs 5.3 months). The performance status after treatment was better in the TACE + Regorafenib group than in the TACE group (P < 0.05). The CEA and CA19-9 negative rates after treatment were higher in the TACE + Regorafenib group than in the TACE group (P < 0.05). Conclusion For the third-line treatment of colorectal liver metastases, the combination of TACE + Regorafenib had better tumor response, OS, and PFS than TACE TACE + Regorafenib combination could be considered as salvage therapy for colorectal liver metastases who failed the first- and second-line standard therapy.


Introduction
Globally, colorectal cancer is a malignant tumor with the second highest incidence and the third highest mortality [1], which seriously threatens human life and health. Te liver is the most common site of metastasis in colorectal cancer [2,3]. Colorectal liver metastases (CRLM) include synchronous liver metastases and metachronous liver metastases. Liver metastasis occurs at the same time or within 6 months after the diagnosis of colorectal cancer, which belongs to synchronous liver metastasis; liver metastasis occurs more than 6 months after the diagnosis of colorectal cancer, which belongs to metachronous liver metastasis [4]. Studies have reported that about 50% of colorectal cancer patients may have liver metastases during the course of the disease. Of these, 15-25% have synchronous liver metastases [5]. Even after achieving curative resection of the primary tumor of colorectal cancer, the chance of developing metachronous liver metastases is as high as 30% [3]. Liver metastasis from colorectal cancer is the leading cause of death [6]. Te survival time of patients with untreated colorectal liver metastases was 6.9 months. Approximately 35% of patients with colorectal cancer have metastases confned to the liver alone and become the ultimate cause of death in approximately 50% of these patients [7,8]. Studies have reported that when colorectal cancer patients develop liver metastases, only 10-20% of patients can be treated with radical surgery [9], and their 5-year survival rate is around 35-55% [10]. However, in patients with unresectable CRLMs, the 5-year survival rate is less than 5%.
For patients with unresectable colorectal liver metastases, a two-drug or three-drug chemotherapy regimen of 5-FU (or capecitabine) combined with oxaliplatin and/or irinotecan, combined with targeted medicine, is a standard frst-and second-line treatment regimen and can signifcantly improve the prognosis of patients [10]. However, many patients with CRLMs experience intrahepatic disease progression after frst-line/second-line therapy and require continued treatment [11]. Clinical studies have found that colorectal liver metastases can be localized in the liver for a long time without extrahepatic metastases [12], so the treatment strategy for colorectal liver metastases should be to actively implement efective local treatment under the premise of multidisciplinary treatment, such as ablation procedures [13], regional hepatic intra-arterial chemotherapy [14], chemoembolization [15], radioembolization [16], and radiation therapy [17], including stereotactic radiation therapy.
Transcatheter arterial chemoembolization (TACE) is an efective treatment for primary liver cancer and liver metastases [18]. Its main principle is to inject chemotherapeutic drugs followed by embolic agents into blood vessels supplying nutrients to the tumor via a catheter. TACE treats liver cancer through the following mechanisms. First, the cytotoxicity of pharmacological drugs can induce the apoptosis of tumor cells and inhibit the proliferation of tumor cells. Second, after the embolization of tumor vessels, it leads to ischemia, hypoxia, and necrosis of tumor tissues [19]. According to the diferent embolic agents used in TACE, TACE consists of conventional TACE (C-TACE) and TACE with drug-eluting beads (DEB-TACE). Karanicolas et al. [20] reported that DEB-TACE appears to ofer a survival beneft in the secondline treatment of the unresectable CRLMs. CalliSpheres ® drug-loaded microspheres are a new type of embolization material. It has the following advantages: (1) Its component is PVA, which is a permanent embolic material. It can efectively embolize tumor blood vessels, leading to tumor ischemia, hypoxia, and necrosis. (2) When the microspheres reach the target vessel, they can slowly release the chemotherapy drugs they have been preloaded. It can maintain a high concentration of chemotherapy drugs in the tumor area for up to one month; meanwhile, the concentration of chemotherapy drugs in the peripheral blood is low, reducing the systemic toxicity of chemotherapy drugs [21]. However, hypoxia after TACE induces an increase in VEGF levels, leading to tumor recurrence and metastasis [22]. Regorafenib is a multitarget tyrosine kinase inhibitor. Regorafenib can inhibit a variety of kinases involved in angiogenesis (including lymphangiogenesis), cell proliferation, and tumor growth as well as the tumor microenvironment, mainly including VEGFR1∼3 (angiogenesis), KIT, RET (tumor formation), PDGFR-β, and FGFR (tumor microenvironment), thereby controlling tumor growth and delaying disease progression [23]. Regorafenib is recommended as a third-line treatment for patients with unresectable CRLM and beyond [24]. Mechanistically, the combination of TACE and Regorafenib has complementary efects and can compensate for the problems posed by the rise in VEGF levels after TACE [25]. Although some studies have reported that TACE combined with Regorafenib was more efective in patients with CRLM than Regorafenib alone [26], there has been no comparative study between TACE + Regorafenib and TACE alone in the treatment of CRLM. Te aim of this study was to investigate the efcacy and safety of TACE combined with Regorafenib versus TACE in the third-line treatment of patients with colorectal liver metastases.

General Information.
Tis was a single-center retrospective study. Te clinical data of 132 patients with colorectal liver metastases admitted to the Department of Intervention Terapy, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, from January 2016 to December 2019 were collected. Inclusion criteria were (1) the primary tumor was confrmed as colorectal cancer by pathology; (2) the lesion was diagnosed as colorectal liver metastasis by pathology or imaging examination; (3) there was no metastasis to other organs than the liver; (4) the lesion progressed after standard frst-and second-line treatment; (5) liver function: Child-Pugh class A or B, performance score (ECOG) 0-2; (6) intrahepatic lesions had not been treated with TACE before; and (7) estimated survival time ≥3 months; and (8) age 18-70 years old. Exclusion criteria were (1) liver function: Child-Pugh class C, physical score (ECOG) > 2; (2) lesions more than 60% of the liver volume; (3) severe coagulopathy and cannot be corrected; (4) combined heart, lung, brain, kidney, and other serious dysfunction; (5) pregnant or lactating patients; (6) allergic to a contrast agent or Regorafenib; (7) the patient was not followed up on time according to the doctor's advice, or the follow-up examination data were incomplete. Te patients were divided into two groups: TACE + regorafenib group (N � 63); TACE group (N � 69) ( Figure 1). Te baseline data of patients including gender, age, location of the primary tumor, whether the primary tumor was surgically resected, Child-Pugh classifcation of liver function, ECOG score, ALT, AST, total bilirubin, CEA, CA19-9, number and the maximum diameter of lesions were collected.

Child-Pugh Classes.
Patients were scored for hepatic encephalopathy, ascites, total bilirubin, albumin, and prolonged prothrombin time. Grade A is 5 ̶ 6 points, grade B is 7-9 points, and grade C is ≥ 10 points. Te patient was placed in a supine position, and the inguinal region was disinfected and draped. Local anesthesia was performed after the injection of 1% lidocaine at the femoral artery puncture site. Te femoral artery was accessed using the Seldinger technique, and a 5F vascular sheath was placed. Te 5 F Yashino catheter was inserted into the celiac trunk and superior mesenteric artery through the vascular sheath, and arteriography was performed to identify the feeding artery of the tumor. A 2.7 F microcatheter was then superselectively cannulated into the tumor-feeding artery and injected with CalliSpheres ® drug-loaded microspheres, and the embolization endpoint was that the blood fow in the tumor-feeding artery was close to the stagnant state. Patients received a reexamination of their enhanced CT or MRI, CEA, and CA19-9 every 3 ̶ 4 weeks. According to reexamination conditions, it was decided whether TACE should be performed again. CalliSpheres ® drug-loaded microspheres with 100-300 µm particle sizes were selected. CalliSpheres ® drugloaded microspheres were frst aspirated and then allowed to stand, and the supernatant was discharged after the microspheres sank; 120 mg irinotecan was dissolved using 5 ml of the vehicle and loaded into a bottle of CalliSpheres ® drugloaded microspheres for 30 min [27]. Nonionic contrast medium was mixed 1 : 1 with CalliSpheres ® Solution before injection, and then slowly injected through the catheter at an injection rate of 1 mL/min to maintain forward blood fow during the injection. CalliSpheres ® were administered according to intrahepatic lesions. Embolization endpoints were near the stasis of forward fow in the target vessel. Once the embolic endpoint is reached, the injection is stopped, whether or not the embolic agent is completely injected. If the embolic endpoint is still not reached after injection of a vial of CalliSpheres ® , blank microsphere (8Spheres ® ) may be supplemented. If it is necessary to supplement blank microspheres 8Spheres ® , select 300-500 µm particle size.
Regorafenib is administered at a dose of 120 mg once daily for three weeks, followed by a one-week break. If the patient is severely intolerable due to adverse drug reactions, the Regorafenib dose is adjusted to 80 mg once daily. In the TACE combined with the Regorafenib group, Regorafenib was temporarily discontinued 3 days after each TACE treatment.

Primary Study Endpoints
(1) Patients in both groups were evaluated for tumor response after 3 months of treatment using mRE-CIST criteria, including complete response (CR), partial response (PR), stable disease (SD), and progressive disease (PD) (2) To evaluate the objective response rate (ORR), and disease control rate (DCR) of the two groups (3) To evaluate OS and PFS in the two groups

Secondary Study Endpoints
(1) Changes in liver function and performance status (ECOG score) before and after treatment in the two groups  (2) Te changes in tumor markers (CEA, CA19-9) before and after treatment in the two groups (3) Te incidence of adverse events after treatment in the two groups Te patients were compared before treatment and at the time of reexamination after 3 months of treatment.
ORR: Proportion of patients achieving CR + PR in total patients after 3 months of treatment.
DCR: Proportion of patients achieving CR + PR + SD in total patients after 3 months of treatment.
PFS: It refers to the time from the initiation of treatment to the occurrence of tumor progression or death.

Statistical Methods.
Statistical analysis was performed using the SPSS 24.0 software. Enumeration data were expressed as the number of cases (percentage), and a chisquare test was used for diferences, including Pearson chisquare and Fisher's exact test. Measurement data were expressed as mean ± standard deviation, and two independent samples t-test was used. OS and PFS were shown by Kaplan ̶ Meier curves, and the log rank test was used to compare OS and PFS between the two groups. P < 0.05 was considered to indicate a statistically signifcant diference.

Basic Information
(1) Comparison of pretreatment enumeration data between the TACE + Regorafenib group and the TACE group (Table 1). Tere were no statistically signifcant diferences in gender, liver function grade, ECOG score, CEA, CA19-9, location of the primary tumor, whether the primary tumor was surgically resected, and size/ number of liver metastases between the two groups. Chi-square test was used for comparison between two groups with aP value > 0.05 and no statistical diference.
(2) Comparison of pretreatment measurement data between the TACE + Regorafenib group and the TACE group (Table 1).
Tere was no statistically signifcant diference in age, ALT, AST, and total bilirubin before treatment between the two groups.
Comparisons between two groups were performed using the t-test with P value > 0.05 and no statistical diference.

Changes in Liver Function before and after Treatment in Both Groups
(1) Comparison of liver function after treatment between the two groups ( A paired sample t-test was used, and a P value < 0.05 was considered to indicate a statistical diference.
Comparisons between two groups were performed using the chi-square test, and a P value < 0.05 was considered to indicate a statistical diference.

Changes in Physical Status and Tumor Marker Levels before and after Treatment in Both Groups
(1) Comparison of physical status and tumor marker levels after treatment between the two groups ( Table 5); Te performance status after treatment in the TACE + Regorafenib group was better than that in the TACE group, with a statistically signifcant difference (P < 0.05). Te CEA and CA19-9 negative rates after treatment in the TACE + Regorafenib group were higher than those in the TACE group, 4 Journal of Oncology with a statistically signifcant diference (P < 0.05). TACE + Regorafenib had higher negative rates of CEA and CA19-9, indicating better efcacy. (2) Compare the changes in physical status and tumor marker levels before and after treatment in each group (Tables 6 and 7).

TACE + Regorafenib group.
Tere was no signifcant statistical diference in the performance status of patients before and after treatment (P > 0.05). Te negative rates of CEA and CA19-9 after treatment were higher than those before treatment, with a statistically signifcant diference (P < 0.05).

TACE Group.
Te performance status of patients after treatment was worse than that before treatment, with a statistical diference (P < 0.05). Te negative rates of CEA and CA19-9 after treatment were higher than those before treatment, with a statistically signifcant diference (P < 0.05).
Te chi-square test was used for comparison between the two groups, and P value < 0.05 was statistically diferent.

Response Evaluation of Tumors after Treatment in the Two Groups
(1) Comparison of tumor response between the two groups (   Chi-square test was used for comparison between the two groups, and a P value < 0.05 was statistically diferent. (3) Comparison of OS and PFS between the two groups ( Table 9, Figures 2 and 3).
Te log-rank test was used for comparison between the two groups, and a P value < 0.05 was statistically signifcant.

Discussion
Colorectal cancer is one of the most common malignant tumors, and the incidence is increasing year by year. Te liver is the most important target organ of hematogenous metastases of colorectal cancer, and CRLM is one of the main causes of death. Currently, CRLM is commonly classifed as resectable, potentially resectable, and unresectable. Although surgical resection is the mainstay of treatment for CRLM, only 15-20% of patients have the opportunity to undergo surgical treatment due to conditions such as lesion size, disease extent, and inadequate liver functional reserve. While for potentially resectable and unresectable patients, the mainstay of treatment remains chemotherapy [28]. Commonly used chemotherapy regimens are FOLFOX, FOLFIRI, FOLFOXIRI, CAPEOX, and so on. Te advent of molecularly targeted agents, represented by bevacizumab and cetuximab, has further prolonged the survival of patients with CRLM [29]. Unfortunately, these tumors that initially respond to treatment will eventually progress. Terefore, it is signifcant to combine efective local therapy in patients with CRLM under the premise of multidisciplinary treatment. Commonly used local treatments for CRLM include ablation therapy [30,31], I125 seed implantation, transcatheter arterial chemoembolization (TACE) [32], hepatic arteria infusion chemotherapy (HAIC) [33], and TARE [34]. TACE was frst reported in the 1970s, and the technique of TACE has been continuously improved with the advent of new devices, drugs, and embolization materials [15]. As one of the commonly used means to treat primary liver cancer, TACE can signifcantly prolong the OS and PFS of patients [35]. Although the metastatic route of the CRLM is the portal vein, its blood supply comes from the hepatic artery [36]. Terefore, TACE is also a commonly used local treatment for CRLM. C-TACE is performed with lipiodol mixed with chemotherapeutic drugs for embolization, and its efcacy is related to whether the tumor is hypervascular [37]. However, CRLM is hypovascular usually, so the efcacy of C-TACE is sometimes unsatisfactory [37,38]. In 2006, DEB-TACE was frst reported to treat liver metastases [39]. C. Aliberti [39] reported that DEB-TACE (loading irinotecan) was safe and efective in treating patients with CRLMs. Peter Huppert [40] found that TACE (loaded with irinotecan) using SAP microspheres can lead to necrosis of CRLMs, and it is safe and efective. It is reported by An Ngo [41] that ffty-three patients of CRLMs received 125 treatments with DEBIRI (irinotecan-loaded beads), their mOS was 14.5 months, and their mPFS was 5 months. Drugeluting microspheres used in DEB-TACE are novel embolic agents that slowly and continuously release their preloaded chemotherapeutic agents when injected into the feeding arteries of tumors. Tus, chemotherapeutic drugs can maintain high drug concentrations for a long time in the tumor area. Drug-eluting microspheres can efectively embolize tumor vessels and slowly release chemotherapeutic drugs, so it is very suitable for the chemoembolization of hypovascular tumors. Eichler [42] reported that DEBIRI (irinotecan-loaded DC beads) is a safe treatment for CRLMs with good pharmacokinetics and a high technical success rate. Martin et al. [43] found that, for patients with CRLMs who failed standard treatment, the use of DEBIRI had good efcacy and high safety. Teir mPFS was 247 days and their mOS was 343 days from the frst treatment. CalliSpheres ® drug-loaded microspheres are a new type of embolic agent, which is composed of polyvinyl alcohol, 2-acrylamide, and 2-methylpropanesulfonic acid polymerization and is an ionexchange embolic microsphere. CalliSpheres ® drug-loaded   [44,45]. Camillo Aliberti [46] suggested that DEBIRI-TACE could be used as a palliative treatment for unresectable CRLMs who failed chemotherapy. Giammaria Fiorentini [47] reported that     Journal of Oncology DEBIRI-TACE was well tolerated in pretreated patients with CRLMs, and the median length of hospital stay was 3 days. Huang et al. [48] found that for initially unresectable KRAS wild-type CRLMs, chemotherapy combined with TACE improved the negative rate of re-resection in patients and had a better survival beneft compared with chemotherapy alone or chemotherapy plus cetuximab. It is reported by Zacharias et al. [49] that for unresectable CRLMs who failed at least frst-line therapy, TACE had a higher OS than others (TACE: 21.3 months vs. HAIC: 13.2 months vs. TARE: 10.7 months). Iezzi et al. [50] suggested that for CRLMs who failed chemotherapy, DEBIRI with capecitabine had a well technical success rate, high safety, and good patient tolerance (mPFS 4 months, mOS 7.3 months). Govindarajan Narayanan [51] found that DEBIRI was well tolerated and safe as a palliative treatment for CRLMs (mOS 13.3, 95% CI: 6.8-19.8 months). However, the tumor microenvironment will change after TACE. Because hypoxia after embolization induces an increase in VEGF, which can lead to the proliferation of new blood vessels and promote the implantation and growth of circulating tumor cells, resulting in tumor recurrence and metastasis. In order to make up for the lack of TACE, some scholars have tried TACE combined with antiangiogenic drugs in the treatment of CRLMs. Fiorentini et al. [52] found that in the treatment of unresectable CRLMs, the CR rate was 6% and 31%, and the PR rate was 13% and 46% after TACE and TACE-bevacizumab at 3 months; the study showed that TACE-bevacizumab was efective and tolerable. Regorafenib is a drug with indications for the treatment of metastatic colorectal cancer, so TACE combined with regorafenib was chosen in our study. Regorafenib is a multitargeted tyrosine kinase inhibitor [53]. Regorafenib can inhibit a variety of kinases involved in angiogenesis, cell proliferation, and tumor growth as well as the tumor microenvironment, mainly including VEGFR1∼3, KIT, RET, PDGFR-β, and FGFR. It can play a role in both the targeted inhibition of tumor cell proliferation and antiangiogenesis [54]. Mechanistically, the combination of TACE and Regorafenib can play a complementary role. TACE combined with targeted drugs has been widely used in the treatment of primary liver cancer, and has achieved very good efcacy. Tomas Walter [11] reported that in the third-line treatment of CRLMs, Regorafenib TAS-102 and SIRT were more effective than the best supportive care. Yamaguchi et al. [55] found that Regorafenib improved the overall survival (mOS was 6.9 months) of patients with CRLM who progress after standard therapies in a real-life setting. Dhillon [56] suggested that Regorafenib was an efective treatment for refractory CRLMs, with controllable adverse events (AEs), which were mostly mild to moderate. Petrioli et al. [57] found that Regorafenib used a modifed 2/1 schedule for refractory CRLMs older than 75 years and with poor performance status, which was well tolerated by patients and achieved a good therapeutic efect (mPFS: 4.8 months, mOS: 8.9 months). In a multicenter study (REBECCA) [58], Regorafenib was safe and efective in patients with CRLMs who failed standard treatments. Regorafenibrelated AEs occurred in 524 patients (80%), and the most common AEs were poor appetite, asthenia, diarrhea, hand-foot skin reactions, hypertension, and mucositis. Te phase-3 trial (CONCUR) [59] showed that for patients with refractory CRLM, the Regorafenib group had a better survival beneft compared with the placebo (mOS, Regorafenib group vs. placebo group: 8.8 months vs. 6.3 months), which is an important option for patients after progression to standard treatment. Tanaka et al. [60] reported that 16 patients with CRLM who failed standard treatment received Regorafenib, with a DCR of 75.0% (95% CI: 50.0-93.8), PFS of 9.0 weeks (8.5-9.5 weeks), and OS of 26.6 weeks (5.0-48.1 weeks).
However, TACE combined with targeted drugs (Regorafenib) is still in the exploratory stage in CRLM. Tis study found that there was a statistically signifcant diference in the efcacy evaluation of tumors after TACE + Regorafenib in the TACE group (P < 0.05). Te ORR and DCR in the TACE + Regorafenib group were signifcantly better than those in the TACE group. Although the negative rates of serological markers (CEA, CA19-9) after treatment in the TACE + Regorafenib group were increased compared with those before treatment, the CEA negative rate in the TACE + Regorafenib group was significantly higher than that in the TACE group (P < 0.05). Te OS in the TACE + Regorafenib group (mOS 18.2 months) was higher than that in the TACE group (mOS 11.3 months) (P < 0.05). Meanwhile, the PFS in the TACE + Regorafenib group (mPFS 8.9 months) was longer than that in the TACE group (mPFS 5.3 months) (P < 0.05). Te reason may be that the combination of TACE and Regorafenib plays a synergistic and complementary efect. VEGF levels increased after TACE, and the antiangiogenic efect of Regorafenib compensated for the lack, prolonged the duration of chemoembolization, and inhibited tumor recurrence and metastasis; meanwhile, Regorafenib could inhibit tumor cell proliferation through a variety of pathways. Terefore, the TACE + Regorafenib group was superior to the TACE alone group in both the efcacy evaluation of the tumor and the survival beneft of the patients after treatment. Cao et al. [26] reported that Regorafenib combined with DEB-TACE could achieve better tumor response, PFS, and OS in CRLMs who failed standard treatment compared with Regorafenib monotherapy. Our fndings were similar but not identical to them, they compared DEB-TACE + regorafenib with regorafenib alone, and our study compared DEB-TACE + Regorafenib with DEB-TACE alone. Te combination of Regorafenib with DEB-TACE was safe and the post-embolization syndrome was tolerable in patients. Kennedy et al. [61] found that the treatment of CRLM using Regorafenib followed by TARE was tolerable (mPFS was 3.7 months and mOS was 12.1 months).  [63] found that the common AEs after TACE included hypertension, abdominal pain, fever, nausea, and vomiting. Common adverse reactions of Regorafenib include fatigue, hand-foot reactions, hypertension, proteinuria, diarrhea, and so on. It is reported by Ducreux et al. [64] that the AEs of Regorafenib in the treatment of CRLMs in the real world was the same as those reported in phase III trials, and the most AEs (grade III-IV) included hypertension (6%), fatigue (9%), and hand-foot reactions (7%). Van Cutsem et al. [65] reported that for CRLMs who failed standard therapy (including systemic chemotherapy and monoclonal antibodies targeting VEGF or EGFR), Regorafenib was safe and efective. Kakizawa et al. [66] found that, under the premise of adequate management of patients' AEs, patients with a morphologic response may predict that patients can beneft from Regorafenib treatment. In CONCUR, higher than grade III Regorafenib-related AEs were hand-foot skin reaction (16%), hypertension (11%), hyperbilirubinaemi (7%), hypophosphataemia (7%), ALT increased (7%), AST increased (6%), lipase increased (4%), and maculopapular rash (4%). Whether the combination of TACE and Regorafenib will increase the adverse reactions of patients is a concern for many scholars. Tis study found that ALT, AST, and total bilirubin in the TACE + Regorafenib group increased after treatment compared with those before treatment, but there was no statistical diference in liver function indicators after treatment between the two groups. However, TACE + Regorafenib had more adverse reactions than the TACE group to fatigue, hand-foot reaction, hypertension, proteinuria, and diarrhea (P < 0.05). However, there was a signifcant diference in ECOG score between the TACE + Regorafenib group and the TACE group. Te performance status after treatment was better in the TACE + Regorafenib group than in the TACE group. Te reason may be that both groups of patients underwent TACE "as needed" rather than "as timed." Because the tumor control of the TACE + Regorafenib group is better, the patients in this group receive TACE less frequently and the interval of TACE is longer. While TACE is an invasive procedure, it has an impact on the physical, psychological, and performance status of patients. Compared with the efect of adverse reactions of Regorafenib on the physical status of patients, TACE has a greater impact on patients. Terefore, patients may have better performance status if the times of TACE is reduced. Te incidence of abdominal pain after treatment was higher in the TACE + Regorafenib group than in the TACE group (P < 0.05). Te reason maybe that due to the antiangiogenic efect of Regorafenib, the artery is thinner in the TACE + Regorafenib group, and the ischemic symptoms of patients are more obvious after TACE, so the incidence of abdominal pain is higher.

Conclusion
For the third-line treatment of CRLM, the combination of TACE + Regorafenib had better tumor response, OS, and PFS than TACE alone. Although the TACE + Regorafenib combination group had slightly more adverse reactions, the adverse reactions were mild and tolerable. Patients in the TACE + Regorafenib group had better performance status and quality of life. Terefore, TACE + Regorafenib combination could be considered as salvage therapy for CRLM who failed the frst-and second-line standard therapy.
Tis is work on a small subset of one study. Te shortcomings of this study are that the data are derived from a single center, and it is a retrospective study with limited sample size. Multicenter, large-sample, prospective studies can be carried out at a later stage to provide more help for clinical work.

Data Availability
Te datasets used and analysed during the current study are available from the corresponding author on reasonable request.

Ethical Approval
Te medical ethics committee of Union Hospital, Tongji Medical College, Huazhong University of science and technology, Wuhan, Hubei Province, approved the retrospective study and gave up the written informed consent. Although the ethics committee gave up the written consent, for the safety and rationality of the research, the authors still signed the informed consent for all the participants. During the follow-up, the authors informed the patients about the study, and they agreed to use their data. Te authors confrm that all methods were performed in accordance with the relevant guidelines and the Declaration of Helsinki.

Consent
All participants' information, images and inspection results are agreed to be published and have obtained written consent. Te details and images will be freely available on the Internet and may be seen by the general public. Te consent form is available if requested. Tese patients have not been reported in any other submission by anyone.

Conflicts of Interest
Te authors declare that they have no fnancial and personal relationships with other people or organizations that can inappropriately infuence our work and there is no professional or other personal interest of any nature or kind in any product, service and company that could be construed as infuencing the position presented in this manuscript.

Authors' Contributions
HH L has made substantial contributions to the conception and design of the work, and the acquisition, analysis of data, as well as manuscript writing. CS Z has made contributions to the design of the work. L. F. has made contributions to the acquisition and analysis of data. B X have made contributions to the analysis, interpretation of data, and manuscript writing. All authors have agreed both to be personally accountable for the author's own contributions and to ensure that questions related to the accuracy or integrity of any part of the work. All authors have read and approved the manuscript. HH L is the frst author. CS Z is the corresponding author and is responsible for ensuring that all listed authors have approved the manuscript before submission.