Survival Outcomes After Portal Vein Embolization and Liver Resection Compared With Liver Transplant for Patients With Extensive Colorectal Cancer Liver Metastases

Key Points

Question  Which treatment modality—liver transplant or portal vein embolization followed by liver resection—yields the longest overall survival among patients with colorectal cancer and extensive liver tumor load?

Findings  This comparative effectiveness research study of 50 patients with colorectal cancer and widespread liver metastases who underwent liver transplant and 53 matched patients who underwent portal vein embolization and liver resection found that overall survival among patients with left-sided primary tumor and liver grafts exceeded overall survival among those treated with portal vein embolization and liver resection.

Meaning  Highly selected patients with colorectal cancer and nonresectable liver metastases may be considered for liver transplant in prospective studies.

Importance  Portal vein embolization (PVE) has been implemented in patients with extensive colorectal liver metastases to increase the number of patients able to undergo liver resection. Liver transplant could be an alternative in selected patients with extensive liver-only disease, and we have recently shown promising survival outcomes.

Objective  To compare overall survival (OS) among patients with colorectal cancer and high liver metastasis tumor load who were treated with liver transplant or with PVE and liver resection.

Design, Setting, and Participants  This comparative effectiveness research study assessed 50 patients with colorectal cancer liver metastases who were previously enrolled in liver transplant studies between November 2006 and August 2019 at Oslo University Hospital in Norway. Those patients were compared with a retrospective cohort of 53 patients in the Oslo University Hospital PVE database from March 2006 through November 2015 with similar selection criteria who underwent PVE and liver resection.

Main Outcomes and Measures  The OS among patients with high tumor load after liver transplant was compared with that among patients with high tumor load who underwent PVE and liver resection. High tumor load was defined as 9 or more metastatic tumors or a diameter of 5.5 cm or longer for the largest liver lesion.

Results  In the PVE cohort of 53 patients, the median age was 61.8 years (range, 34.3-71.3 years), and 36 patients (68%) were men. The 5-year OS rate among 38 patients who underwent liver resection after PVE was 44.6%. The 5-year OS rate for patients with high tumor load was 33.4% for those who underwent liver transplant and 6.7% for those who underwent PVE. Among patients with high tumor load and left-sided primary tumors, the 5-year OS rate was 45.3% for those receiving a liver allograft and 12.5% for those treated with PVE and liver resection.

Conclusions and Relevance  Patients with nonresectable disease, an extensive liver tumor load, and left-sided primary tumors had long OS after liver transplant, exceeding the survival outcome for those patients treated with PVE and liver resection.

Colorectal cancer (CRC) is the second most common cause of death due to malignant neoplasm in Western societies.¹ Metastatic spread at time of diagnosis or later will be detected in approximately half of patients, and the liver is the most frequent metastatic site.² Liver resection (LR) is considered the standard of care as a curative treatment for patients with colorectal cancer liver metastasis (CRLM). However, only a few patients with CRLM are candidates for LR. The general definition of resectability is the ability to preserve adequate functional volume of the future liver remnant (FLR) with preserved vascular inflow, outflow, and biliary drainage.³ Patients with more advanced CRLM may require expansion of the FLR prior to LR to avoid postoperative liver failure due to insufficient functional liver volume. Currently, portal vein embolization (PVE) is the preferred method for increasing the volume of an inadequate FLR.^4,5 The 5-year overall survival (OS) rate after LR is reported to be 30% to 50%.^2,6 Several scoring systems have been developed to estimate OS after LR for patients with CRLM.^7,8 In the Fong Clinical Risk Score system,⁷ the size of the largest lesion with a diameter longer than 5 cm is one of the parameters included, and several studies indicate that the number of metastatic lesions is of prognostic significance. Viganò et al⁹ and Allard et al¹⁰ reported a 5-year OS rate after LR of 20% and of 30%, respectively, among patients with 8 or more CRLM lesions and with 10 or more CRLM lesions, respectively. Survival outcomes after LR indicate that there is a strong association of size and number of CRLM lesions resected with postoperative survival.¹⁰ The prognostic importance of the size and number of resected CRLM lesions has also been shown by the Tumor Burden Score (TBS).¹¹

In 2006, a pilot study (Secondary Cancer I [SECA-I])¹² investigating liver transplants (LTs ) in patients with CRC and nonresectable liver-only metastases found an estimated 5-year OS rate of 60%. Pretransplant maximal tumor diameter exceeding 5.5 cm, level of carcinoembryonic antigen prior to LT higher than 80 ng/mL (to convert to micrograms per liter, multiply by 1.0), failure of chemotherapy, and a time interval from resection of the primary tumor to LT of less than 2 years were all factors associated with decreased survival. By assigning 1 point for each of those factors, a scoring system termed the Oslo Score was developed and may be used to stratify posttransplant OS among patients with CRLM.¹² Similarly, the Fong Clinical Risk Score and metabolic tumor volume assessed by fluorodeoxyglucose positron emission tomography–computed tomography (FDG-PET-CT) prior to LT are both closely associated with OS.^7,13,14 Based on these 3 different scoring systems for selection of patients with nonresectable CRLM, 5-year OS rates of approximately 70% to 100% may be obtained.¹⁴ We have previously shown a marked difference in OS after LT among patients with the primary tumor in the ascending colon compared with the primary tumor located in the remainder of the colon or rectum (termed herein left-sided tumors).^14,15 In the present report, we compared OS among patients with resectable CRLM receiving PVE prior to resection and among patients with nonresectable CRLM treated with LT in the same institution.

In total, 50 patients with CRLM underwent LT between November 2006 and August 2019 at Oslo University Hospital, Norway, as part of approved prospective studies ([SECA-I]¹² and NCT01479608 [SECA-II]¹⁵). All protocols were approved by the regional ethics committee and institutional review board of Oslo University Hospital. All patients provided written informed consent before inclusion, which was obtained in a manner that was consistent with the Declaration of Helsinki.¹⁶ No one received compensation or was offered any incentive for participating in this study.

The inclusion and exclusion criteria, as well as the immunosuppression therapy used in the different studies, have been previously reported.^12,15 To exclude extrahepatic metastases, FDG-PET-CT was performed for all patients prior to LT.

All patients included in the LT studies were considered to have nonresectable CRLM by a multidisciplinary tumor board. Patients with present or prior extrahepatic disease, other malignant neoplasm, or primary tumor in situ were excluded except 1 patient with prior resected pulmonary metastasis who was accepted in the SECA-II group D. No patient received adjuvant chemotherapy after LT. Patients had regular outpatient follow-up once a month the first year, every 3 months the second year, and every 6 months thereafter. Treatment at time of relapse was at the discretion of the responsible physician.

All patients underwent review at a multidisciplinary tumor board meeting to determine resectability and resection strategy. Portal vein embolization was used to induce hypertrophy of the FLR to enable LR in patients with an insufficient FLR, defined as an FLR less than 30% of the total estimated liver volume, calculated based on body surface area. The standardized FLR was calculated by dividing the CT-measured FLR by the total estimated liver volume. The PVE procedure was performed by percutaneous transhepatic glue embolization of the ipsilateral portal vein.¹⁷ Embolization of segment 4 portal vein branches was performed in selected cases requiring an extended right side hemihepatectomy. A total of 109 patients who underwent PVE were included in the prospective PVE database at Oslo University Hospital from March 2006 through November 2015. From the PVE database, 53 patients with CRLM but without extrahepatic disease, primary tumor in situ, or other prior malignant neoplasm were selected. In this comparative effectiveness study, all patients in the LT and PVE groups had an Eastern Cooperative Oncology Group score of 0 to 1, were younger than 72 years of age, and had the primary tumor resected before LT or PVE. All patients received standard-of-care chemotherapy. More than 50% of patients with LT had received second or later lines of chemotherapy at the time of transplant.

The TBS was calculated as previously described.¹¹ Prior studies have shown that patients with more than 8 or 10 CRLM lesions have reduced OS after LR.^9,10 Furthermore, having a maximal tumor diameter exceeding 5 cm is associated with decreased postresection survival,⁷ and patients with lesions 5.5 cm or longer in diameter have decreased OS after LT.¹² High tumor load (HTL) is defined herein as 9 or more metastatic tumors or a maximal tumor diameter is equal to or longer than 5.5 cm, whereas patients below these limits were categorized as having low tumor load (LTL).

Survival analyses were estimated using the Kaplan-Meier method. Overall survival was calculated from the date of LT to the end of follow-up on August 1, 2019, in the LT studies and from the date of PVE plus LR after PVE to August 1, 2019, in the PVE cohort. The log-rank test was used to compare outcomes between the 2 groups. The difference in the median values between the groups was calculated using the Mann-Whitney test. A 2-sided P < .05 was considered statistically significant. The analyses were performed using IBM SPSS, version 25.0 (IBM SPSS).

All patients (n = 50) included in the LT studies of SECA-I (23 patients), SECA-II (17 patients), and SECA-II group D (10 patients) were included in this report. A matched cohort of 53 patients with resectable CRLM lesions and insufficient FLR (<30%) received PVE to expand the FLR prior to LR but otherwise displayed similar selection criteria as patients treated with LT. The median age of this PVE cohort was 61.8 years (range, 34.3-71.3 years), and 36 patients (68%) were men. The size and number of CRLM lesions before and after chemotherapy, age, carcinoembryonic antigen level, sex, and the T category and N category of the primary tumor in the PVE cohort are presented in Table 1. One patient had more than 20 CRLM lesions before chemotherapy, and no patient had 20 lesions or more after chemotherapy at the time of the PVE procedure. Patients with CRLM lesions considered resectable at the multidisciplinary tumor board meeting and who received PVE had a median OS of 32.7 months (95% CI, 17.7-47.8 months) and a 5-year OS rate of 32.0% from the time of the PVE procedure. Fifteen patients (28%) did not proceed to LR as planned owing to either insufficient regeneration of the FLR or tumor progression. The median OS in this subgroup was 10.9 months (95% CI, 6.6-15.2 months), with no patient surviving for 5 years (Figure 1A). The remaining 38 patients undergoing LR after PVE had a median OS from PVE of 43.9 months (95% CI, 17.7-70.1 months) with a 5-year OS rate of 44.6% (Figure 1A). Of the 38 patients with LR, 13 did not receive adjuvant chemotherapy, and 18 patients started adjuvant chemotherapy, with 11 patients treated with oxaliplatin-containing regimens. The adjuvant treatment status was unknown for 7 patients. The decision to initiate adjuvant chemotherapy was at the discretion of the oncologist at the local hospital. The aforementioned difference in OS between the patients receiving LR and those with no resection after PVE was significant (P < .001).

The LT and PVE groups were subclassified into 2 subgroups according to tumor load. In both the PVE and LT groups, patients in the LTL groups had significantly better OS compared with patients in the HTL groups. There were 8 patients without available CT scans after chemotherapy, and we therefore did not have the number or size of the largest lesions for these patients at the time of the PVE procedure. These patients could therefore not be classified as having either LTL or HTL. Patients in the PVE group with LTL (n = 30) had a median OS of 68.9 months (95% CI, 26.1-111.7 months) and a 5-year OS rate of 53.1% (Figure 1B). However, 7 patients with LTL receiving only PVE and no LR had a median OS of 10.5 months (95% CI, 9.7-11.3 months), with the longest OS of 46.4 months. The 5-year OS rate of 23 patients with LTL receiving LR after PVE was 69.3%. There were 15 patients with HTL within the PVE cohort, and their median OS from the PVE procedure was 19.2 months (95% CI, 0-39.5 months), with 1 patient (6.7%) who survived for 5 years (HTL-PVE vs LTL-PVE; P = .001) (Figure 1B). Seven patients with HTL receiving PVE only had a median OS of only 10.9 months (95% CI, 7.6-14.2 months) from the procedure, and 8 patients with HTL receiving LR after PVE had a median OS of 29.8 months (95% CI, 11.1-48.5 months) from the PVE procedure (P = .04) (Figure 2).

The 5-year OS rate following LT for 21 patients with LTL was 72.4%. Among 29 patients with HTL who received LT, the median OS was 40.5 months (95% CI, 26.4-54.5 months), with a 5-year OS rate of 33.4%. This difference between the LT subgroups with LTL vs HTL was significant (P = .002).

The 5-year OS rate among patients in the LTL-LT group was 72.4%, and among patients in the LTL-PVE group, the 5-year OS was 53.1% (P = .08). The median number and size of the largest liver metastatic tumors in the LTL-PVE-LR group (n = 23) was 4 lesions (95% CI, 3-5 lesions) and 22 mm (95% CI, 13-34 mm) compared with 5 lesions (95% CI, 4-7 lesions) and 28 mm (95% CI, 11-34 mm) in the LTL-LT group. None of these differences were significant. However, the LTL-LT group had a significantly higher median TBS value (7.1; interquartile range, [IQR], 5.8-8.5) compared with the LTL-PVE-LR group (median, 4.5; IQR, 3.3-5.5) (P < .001).

In the HTL groups, there was a significant increase (P = .007) in OS among patients receiving LT compared with patients receiving PVE, with a median OS of 40.5 months (95% CI, 26.3-54.7 months) in the LT group and 19.2 months (95% CI, 0.0-39.5 months) in the PVE group (Figure 3). The median number of CRLM tumors was 15 (range, 1-53) in the in HTL-LT group and 10 (range, 2-15) in the HTL-PVE group (P = .06), with 14 of 29 patients in the HTL-LT group having 16 or more CRLM tumors. The median size of the largest lesions was 54 mm (range, 10-130 mm) in patients with HTL and 30 mm (range, 6-160 mm) in patients receiving LT and PVE (P > .99). The TBS in the HTL-LT group was 17.7 (range, 6.7-53.1) compared with 11.0 (range, 5.9-16.9) in the HTL-PVE group (P = .06). Tumor Burden Score zone 3 (worse prognosis), defined as a score of 9 of higher,¹¹ was observed in 28 of 29 patients in the HTL-LT group and in 11 of 15 patients in the HTL-PVE group.

Patients with HTL-LT and primary left-sided tumors had a median OS of 59.9 months (95% CI, 26.2-93.6 months), with a 5-year OS rate of 45.3%, and patients with HTL-LT and the primary tumor in the ascending colon had a median OS of 12.2 months (95% CI, 0.0-27.7), with a 5-year OS rate of 0% (P < .001) (Figure 2). One patient with an ascending colon primary tumor was alive with new metastatic disease and receiving palliative chemotherapy 27.4 months after LT. There were no significant differences in TBS, number of lesions and size of largest lesion, median carcinoembryonic antigen values, sex, age, and T category of the primary tumor between patients with the primary tumor in the ascending colon and patients with a left-sided primary tumor (Table 2). However, there was a significantly higher number of patients having pN+ primary disease (P = .04) among patients with the primary tumor located in the ascending colon (median, 2.0; range, 0.0-2.0; IQR, 1.2-2.0; mean, 1.6; SD, 0.6) vs a left-sided primary tumor (median, 1.0; range, 0.0-2.0; IQR, 0.0-2.0; mean, 0.9; SD, 0.7). Furthermore, 21 patients in the HTL-LT group with a left-sided primary tumor had significantly longer OS compared with 8 patients in the HTL-PVE group who received LR after the PVE procedure (P = .04), with a median OS of 59.9 months (95% CI, 26.2-93.6 months) in the LT group compared with a median OS of 29.8 months (95% CI, 11.1-48.5 months) in the PVE-LR group. The 5-year OS was 45.3% among the patients in the HTL-LT group with a left-sided primary tumor and 12.5% among the patients in the HTL-PVE-LR group (Figure 2).

This study suggests that LT may have a role in the treatment of selected patients with CRLM and resectable disease with HTL. Liver resection has been considered the only curative treatment for patients with CRLM. However, the majority of patients with LR will have recurrence within 3 years. The high recurrence rate is a reality, even in a cohort with a median of only 1 lesion.⁶ Several LR studies have shown that survival outcomes after LR for CRLM is associated with the number of resected CRLM lesions as well as the size of the largest resected CRLM lesion.^9-11 On the basis of the number and size of the largest CRLM tumor, Sasaki et al¹¹ developed the TBS and showed a close association with OS after LR, with patients having TBS zone 3 showing significantly shorter OS compared with patients having TBS zones 1 or 2. Patients with resection in TBS zone 3 had a 5-year OS rate of 25.5%.¹¹ The OS after LT in the LT-HTL group in TBS zone 3 was higher in our study than that reported in studies on LR with the same tumor load assessed as TBS. The median number of lesions in most publications of LR in patients with CRC is 1 to 3 lesions, and the prevalence of patients eligible for resection having 8 or more liver lesions is in most reports less than 10%, often down to 2% to 4% of patients.^7,18-20

Deceased donor grafts are a scarce resource in all countries. Hence, it is important to select patients who would benefit the most to avoid the futile use of liver grafts. As previously reported, well-selected patients with nonresectable CRLM lesions and low Fong Clinical Risk Score, Oslo Score, or PET-derived metabolic tumor volume values may attain acceptable 5-year OS rates.¹⁴ In the present report, we obtained similar results for patients with LTL receiving LT, with a 5-year OS rate of 72%. Long OS among patients with CRC after LT was observed despite the fact that most of the patients who received LT had a relapse. The extended OS from time of relapse has been reported to be due to the majority of recurrent disease being small pulmonary lesions with a slow growth pattern, and many of these pulmonary lesions were resected.²¹ By contrast, a right-sided CRC primary tumor has been associated with a short OS from the time of LT,^14,15 and LT should be implemented with caution in this cohort of patients with CRLM. Other studies have reported 5-year OS rates after LR of 20% and 30% among patients with 8 or more CRLM lesions and 10 or more CRLM lesions, respectively.^9,10 We are not aware of any publication with OS similar to that observed in the present HTL-LT cohort among patients with CRC and a median of 15 CRLM lesions receiving LR or any other medical treatment. Liver transplantation among selected patients with left-sided primary tumors and numerous liver metastases may, despite heavy tumor load, still yield sufficiently long OS to be considered for LT (Figure 2). Furthermore, some patients who have received extensive neoadjuvant or adjuvant chemotherapy combined with LR may eventually develop postchemotherapy liver failure. Patients in this category who are recurrence free after long follow-up may also be good candidates for LT in the future. In our data set, no patient had postchemotherapy liver failure.

Portal vein embolization is an established technique to expand the FLR volume so that a metastasectomy can be performed with reduced risk of postresection liver failure. Similar to previous results, our results showed that patients receiving surgery after PVE had much better OS compared with patients not undergoing resection for CRLM after PVE (Figure 1A). The outcome was highly dependent on tumor load, with patients in the LTL-PVE group having significantly better OS compared with patients in the HTL-PVE group (Figure 1B). By contrast, patients not undergoing resection after PVE irrespective of tumor load displayed OS comparable to patients receiving second-line chemotherapy.²² Furthermore, only 8 of the 15 patients in the HTL-PVE group received the planned LR after PVE, which may suggest that alternative treatment options, such as LT, may be relevant to consider in this cohort. Patients with left-sided primary tumors and HTL who underwent LT displayed significantly better outcomes compared with patients receiving PVE followed by LR (Figure 2). Taken together, these results may suggest that patients with extensive liver-only disease, with the exclusion of patients with the primary tumor in the ascending colon, may benefit from LT compared with standard-of-care PVE and liver resection (Figure 2). This topic is clearly a highly controversial issue that can be addressed further only by properly designed prospective trials adjusting for the selection bias in the present study. Furthermore, development in both PVE and resection techniques may in the future increase resectability rates and, in some instances, the OS after this procedure. The fundamental problem in HTL-CRLM, however, regardless of resection technique used, is the prospect of frequent microscopic residual disease after resection in these patients, ultimately leading to high intrahepatic recurrence rates. A possible future strategy may be to resect the liver in patients with resectable HTL and perform LT in patients who later develop liver-only recurrent disease.

During the last few years, increased use of FLR expansion techniques, such as associating liver partition and portal vein ligation (ALPPS), for staged hepatectomy has been established as a treatment option in extensive metastatic disease and small FLR. Compared with PVE treatment, the ALPPS procedure has been reported to result in increased resection rates.²³ Two-year OS rates among patients with CRLM who underwent the ALPPS procedure have been reported to be 62% in Olthof et al²⁴ and 66% in Wanis et al,²⁵ with a 5-year OS rate of 27% in another study recently reported.²⁶ A randomized clinical trial investigating ALPPS vs PVE for patients with CRLM (the LIGRO trial)²⁷ reported an estimated median survival in the ALPPS group of 46 months compared with 26 months in PVE group. We have previously reported in the SECA-I and SECA-II LT studies 2-year OS of 90% and 100%, respectively.^12,15 The LTL-PVE group had long OS from the time of the PVE procedure, and patients with LTL who are able to receive PVE followed by LR have a good prognosis.

The present comparison has several limitations. Although the results might suggest that better outcomes may be obtained with LT than with resection for patients with HTL, the most important interpretation of this study is the hypothesis-generating properties in guiding possible future prospective studies to assess this possibility in a scientifically stringent manner. There were some major differences between the LT and PVE groups. The use of FDG-PET-CT was required for the LT group and not for the PVE group, but all patients with extrahepatic metastatic disease as assessed using either CT or magnetic resonance imaging or during surgery in the PVE group were excluded from the present comparative study. The LT group was heavily pretreated at the time of LT, with 81% of the patients having received a second line or later lines of chemotherapy at the time of LT. The maximum number of CRLM lesions in the PVE group was 15, whereas 14 of the 29 patients in the HTL-LT group had 16 or more CRLM lesions at the time they underwent LT. In both cohorts, patients having progressive disease while undergoing chemotherapy were included. Furthermore, all patients with LT were considered to have nonresectable disease by the multidisciplinary tumor board, whereas patients receiving PVE were all considered to have resectable disease if they obtained an acceptable FLR after the PVE procedure.

In conclusion, the present results suggest that LT is a treatment that may yield long OS among selected patients with advanced CRLM with a left-sided primary tumor, whereas a right-sided primary tumor is a distinct negative prognostic factor. Patients with fewer than 9 liver metastatic tumors, all smaller than 5.5 cm in diameter, may obtain long OS after LR following the PVE procedure. Patients with a high liver tumor load may benefit from LT even if they are considered to have resectable disease. Patients who do not respond to PVE may, if they fulfill all transplant criteria, be evaluated for LT. Liver transplant among patients with CRLM should still be considered a work in progress, and patients offered LT should therefore be included in prospective trails.

Accepted for Publication: January 10, 2021.

Published Online: March 31, 2021. doi:10.1001/jamasurg.2021.0267

Corresponding Author: Svein Dueland, MD, PhD, Section for Transplantation Surgery, Department of Transplantation Medicine, Oslo University Hospital, Postbox 4950 Nydalen, N-0424 Oslo, Norway (svedue@ous-hf.no).

Author Contributions: Drs Dueland and Yaqub had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Dueland, Hagness, Brudvik, Line.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Dueland, Carling, Line.

Critical revision of the manuscript for important intellectual content: All authors.

Statistical analysis: Dueland.

Obtained funding: Dueland, Line.

Administrative, technical, or material support: Dueland, Yaqub, Syversveen, Brudvik, Line.

Supervision: Dueland, Line.

Conflict of Interest Disclosures: None reported.

Funding/Support: The study was supported by Oslo University Hospital, the Norwegian Cancer Society, and South-Eastern Norway Regional Health Authority.

Role of the Funder/Sponsor: The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.