Impact of Molecular Status on Metastasectomy of Colorectal Cancer Liver Metastases

Abstract

Although surgical resection could provide better survival for patients with colorectal cancer liver metastases (CRLM), the recurrence rate after resection of CRLM remains high. The progress of genome sequencing technologies has greatly improved the molecular understanding of colorectal cancer. In the era of genomics and targeted therapy, genetic mutation analysis is of great significance to guide systemic treatment and identify patients who can benefit from resection of CRLM. RAS and BRAF mutations and microsatellite instability/deficient deoxyribonucleic acid (DNA) mismatch repair status have been incorporated into current clinical practice. Other promising molecular biomarkers such as coexisting gene mutations and circulating tumor DNA are under active investigation. This study aimed to review the prognostic significance of molecular biomarkers in patients with CRLM undergoing metastasectomy based on the current evidence.

Colorectal cancer (CRC) is the third most common malignancy and the second leading cause of cancer-related mortality worldwide. ^{1 2} CRC liver metastases (CRLM) develop in approximately 50% of CRC patients, ³ and liver metastasis is the main cause of death in patients with CRC. ⁴ Surgical resection of CRLM provides patients with the opportunity to cure, which is the best choice for patients with CRLM to prolong survival. The 5-year overall survival (OS) rates for patients with CRLM undergoing hepatectomy was reported ranging from 20 to 45%, compared with between 0 and 5% for those who did not receive hepatectomy. ^{5 6} Unfortunately, more than 80% of patients with CRLM are initially unresectable. ⁷ Advances in systemic therapy and surgical techniques have allowed more patients with CRLM to have access to hepatectomy. However, the recurrence rate after resection of CRLM remains high, with more than 50% of patients experiencing recurrence within 2 years after surgery. ⁸

The progress of genome sequencing technologies has greatly improved the molecular understanding of CRC, revealing the genomic variants that contribute to CRC carcinogenesis and metastases. ^{9 10} In the era of genomics and targeted therapy, genetic mutation analysis is of great significance to guide systemic treatment and identify patients who can benefit from resection of CRLM. ^{5 11} RAS and BRAF mutations have been widely recognized to be associated with worse survival after resection of CRLM. Additionally, other molecular biomarkers such as TP53, SMAD4, APC, and PIK3CA were also reported to be prognostic factors for CRLM. ^{12 13} This study aimed to review the prognostic significance of molecular biomarkers in patients with CRLM undergoing hepatectomy.

Molecular Status and Metastasectomy of CRLM

A study from Memorial Sloan Kettering Cancer Center defined the genomic landscape of metastatic CRC by sequencing of 1,134 CRCs. ¹⁴ The most frequently mutated gene detected in their patients was APC (79%), followed by TP53 (78%), KRAS (44%), PIK3CA (18%), and SMAD4 (16%). ¹⁴ Few genomic differences were observed between their primary lesions and metastases. ¹⁴ The authors' center delineated the genomic landscape of Chinese patients with CRLM based on 396 metastatic liver samples, with the most frequent mutations being TP53 (82%), APC (76%), KRAS (42%), SMAD4 (14%), FLG (13%), and FBXW7 (11%). ¹⁵

RAS

RAS family proto-oncogenes include KRAS, NRAS, and HRAS genes. Oncogenic mutations in KRAS and NRAS are common in CRC, with mutation frequencies of approximately 40 and 5%, respectively. ¹⁶ RAS mutations can lead to activation of MAPK and PI3K pathways, promoting the development and progression of CRC. ¹⁷ Mutations in RAS are also predictive biomarkers of resistance to therapy of epidermal growth factor receptor (EGFR) monoclonal antibodies in metastatic CRCs. ¹⁸ Furthermore, RAS mutations have significant prognosis predictive value in patients with CRLM. A meta-analysis of Brudvik et al found that KRAS mutations were negatively correlated with OS (hazard ratio [HR] 2.24, 95% confidence interval [CI]: 1.76–2.85) in patients after resection of CRLM, and the recurrence-free survival (RFS) of KRAS-mutated patients was also lower than the wild-type KRAS patients (HR 1.89, 95% CI: 1.54–2.32). ¹⁹ Patients with NRAS mutations were also reported to be associated with worse outcomes than wild-type patients. ^{20 21} The negative prognostic effect of RAS mutation was also observed in our center and integrated into a nomogram to predict survival of patients undergoing metastasectomy of CRLM. ²² Of note, the prognostic effects of RAS mutation status are related to the surgical procedures, primary tumor location, and specific codons mutations in RAS.

RAS Mutation Status and Surgical Margin

RAS mutations were found to be associated with a higher incidence of positive resection margins and narrower resection margins in patients undergoing resection of CRLM. ²³ Zhang et al ²⁴ reported that CRLM patients with mutated KRAS had a higher incidence of micrometastases and a greater distance from hepatic micrometastases to the border of gross tumors than those with wild-type KRAS, which may lead to the narrower margin width and higher rate of positive resection margins observed in KRAS-mutated CRLM patients. In a study by Margonis et al ²⁵ from the Johns Hopkins University School of Medicine, the authors found that R0 resection only improved the OS among CRLM patients who had KRAS wild-type tumors, while surgical margin (R0 vs. R1) was not associated with OS among patients with KRAS-mutated tumors. Margonis et al further divided R0 resection into three groups (1–4, 5–9, and ≥ 10 mm) in their subsequent study. And their results showed that a surgical margin width of 1 to 4 mm was sufficient for KRAS wild-type patients undergoing resection of CRLM, as patients failed to gain more survival benefits with the increase of surgical margin. ²⁶ However, even R0 resection with surgical margins of more than 1 cm did not confer survival benefit in patients with KRAS-mutated CRLM. ²⁶ Consistently, Hatta et al ²⁷ also found that increasing resection margins were not associated with improved RFS and OS among patients with KRAS-mutated CRLM. These findings suggested that the aggressive biologic nature of RAS mutations could not be balanced by extensive resection. Using artificial intelligence-based techniques, a recent multicenter study based on 1,843 patients from Bertsimas et al suggested that the optimal resection margin width in hepatectomy for KRAS-mutated CRLM was 7 mm. ²⁸

RAS Mutation Status and Anatomical Resection

Although macroscopic invasion of intrahepatic vessels in CRLM has proven to be uncommon, microscopic intrahepatic microvascular invasion can be detected at as high as 37.8%. ²⁹ The presence of RAS mutations was found to be linked with increased risk of vascular invasion and hematogenous metastases. ³⁰ Systematic removal of potential “tumor-containing” intrahepatic vessels by anatomical resection is theoretically possible to reduce intrahepatic recurrence or metastasis. Based on this hypothesis, a study from Johns Hopkins University School of Medicine compared anatomical resection and nonanatomical resection for CRLM based on KRAS mutational status. ³¹ They found that patients with KRAS mutations undergoing anatomical resection for CRLM had a significantly longer median disease-free survival (DFS) of 33.8 months, compared with 10.5 months in those treated by nonanatomical resection. However, anatomical resection did not improve the DFS in patients with KRAS wild-type CRLM. A propensity score matching study from MD Anderson Cancer Center reported different results. ³² Anatomical resection did not prolong the RFS and OS in both patients with RAS wild-type and mutated CRLM. ³² Similarly, Choi et al ³³ also found that there was no difference in DFS between CRLM patients treated by anatomical and nonanatomical resection, regardless of KRAS mutation status. As the three above studies were retrospective, the selection bias of liver surgery methods limited the reliability of their results. The current evidence is insufficient to prove that anatomical resection is associated with favorable outcomes in patients with RAS-mutated CRLM. Compared with anatomical resection, nonanatomical resection could preserve liver parenchyma and provide more opportunities for salvage resection following intrahepatic recurrence. ³⁴ It is hoped that the ongoing multicenter randomized controlled trial, ARMANI trial (NCT04678583), could clarify whether anatomical resection could bring survival benefit for patients with RAS-mutated CRLM.

RAS Mutation Status and Special Surgical Procedures

Two-stage hepatectomy (TSH) offers an opportunity to prolong survival for patients with bilateral CRLM whose lesions cannot be resected at one operation. ³⁵ TSH is a highly demanding treatment strategy, including two consecutive hepatectomies, systemic chemotherapy, and interval portal vein embolization in the majority of patients. It is particularly important to select patients who are expected to complete TSH and can obtain survival benefit. Passot et al reported that although RAS mutation status did not affect the completion of TSH, RAS mutation was identified as an independent factor associated with progression-free survival and OS for patients undergoing TSH for CRLM. ³⁶ Patients with RAS-mutated CRLM undergoing TSH showed very poor survival in their study. All their RAS-mutated patients had recurrence within 18 months after first-stage resection, and only one patient survived beyond 5 years. ³⁶ It seems that RAS mutation has greater prognostic impact on patients after TSH than single-stage hepatectomy. Accordingly, RAS mutation status should be an important consideration when selecting patients for TSH.

Associating liver partition and portal vein ligation for staged hepatectomy (ALPPS) is a TSH variant that could rapidly promote the increase of liver volume and improve the resection rates in patients with unresectable CRLM due to insufficient future liver remnant. ³⁷ A study of Serenari et al from Italy reported that patients with KRAS-mutated CRLM just obtained a median OS of 15.3 months after ALPPS, significantly lower than the 38.3 months in those with KRAS wild-type CRLM. ³⁸

Likewise, a study which enrolled 510 ALPPS patients from 22 centers revealed that the presence of RAS mutation was associated with reduced cancer-specific survival (CSS), with 5-year CSS of 41% for RAS wild-type versus 11% for RAS-mutated patients. ³⁹ In view of the less survival benefit but greater surgical risk, the authors suggested that ALPPS should not be conducted in patients with RAS-mutated CRLM. ³⁹

RAS Mutation Status and Primary Tumor Location

Right-sided and left-sided CRCs have different underlying disease biology, with significant differences in terms of incidence, pathogenesis, presentation, and outcomes. ⁴⁰ Our group's meta-analysis of 12 studies with a total of 6,387 patients showed that patients with primary right-sided CRC after resection of CRLM had a significantly decreased 5-year OS compared with those with primary left-sided CRC, with a pooled HR of 1.354 (95% CI: 1.238–1.482). ⁴¹ The prognostic impact of RAS mutation status on CRLM patients was found to be related to the primary tumor location. ^{42 43} A study from Sasaki et al involving 426 CRLM patients reported that KRAS mutations were detected in half of patients with primary right-sided CRC, higher than the detection rate of one-third of patients with primary left-sided CRC. ⁴² Mutant-type KRAS was associated with worse RFS and OS in their patients with primary left-sided CRC who underwent hepatectomy for CRLM, but did not affect the prognosis in those with primary right-sided CRC. ⁴² Consistently, our study also confirmed that RAS mutations were associated with right-sided CRC and only had negative effects on the prognosis of patients undergoing hepatectomy for CRLM derived from a primary left-sided CRC. ⁴³

Also, the effect of primary tumor site on prognosis differs with the status of RAS mutation. Margonis et al ⁴⁴ reported that right-sided primary tumor was associated with a worse 5-year OS (43.7% vs. 55.5%) in their KRAS wild-type patients who underwent curative-intent resection for CRLM. However, primary tumor site had no impact on 5-year OS in their KRAS-mutated patients. ⁴⁴ In a study of 227 patients undergoing simultaneous curative-intent surgery for synchronous CRLM, right-sided and KRAS-mutated CRLM patients showed the worst OS and DFS, while left-sided and KRAS wild-type CRLM patients had the best OS and DFS. ⁴⁵

Specific KRAS Codon Mutations and Prognosis

KRAS mutations were most common in codons 12 and 13, whereas mutations were less common in codons 61 and 146. ⁴⁶ Margonis et al found that patients with KRAS codon 12 mutations undergoing resection of CRLM had worse OS compared with wild-type KRAS patients, while KRAS codon 13 mutation had no correlation with prognosis. ⁴⁷ This finding was validated in a study from China, although this study focused on patients with unresectable CRLM. ⁴⁸ A stratified analysis according to primary disease site showed that both codon 12 and codon 13 KRAS mutations were related to poor OS in patients with primary left-sided colon cancer after resection of CRLM. ⁴⁹ However, neither codon 12 nor codon 13 KRAS mutations had correlation with OS in those with liver metastases derived from right-sided colon cancer and rectal cancer. ⁴⁹ A subsequent study by the Margonis group further investigated the impact of codon-specific KRAS mutations on recurrence patterns in patients with CRLM, and their results showed that codon 13 KRAS mutations patients were linked with the higher risk of extrahepatic recurrence and lung-specific recurrence, while KRAS 12 codon mutations did not. ⁵⁰

BRAF

BRAF is a protein kinase that plays a key role in the MAPK signaling pathway, and the aberrant activation of this pathway is associated with the development of CRC. ⁵¹ Mutations in the BRAF gene lead to a sustained downstream activation of the MEK/ERK/MAPK pathways, affecting the differentiation, proliferation, and migration of tumor cells. ⁵² The frequency of BRAF mutation was reported ranging from 4.7 to 20% in CRC. ⁵² However, BRAF mutation was present in only 1 to 6.1% of resectable CRLM patients, which may be due to the rapid tumor growth and metastasis caused by mutations in BRAF gene. ⁵² Patients with BRAF-mutated metastatic CRC are more likely to develop peritoneal metastases, rarely present disease confined to the liver. ⁵³ BRAF mutations are associated with poor prognosis in patients with metastatic CRC, with mortality rates nearly three times that of BRAF wild-type patients. ⁵²

Schirripa et al reported that the median RFS and OS after resection of CRLM in BRAF-mutated patients were only 5.7 and 22.6 months, far lower than the figures of 14.4 and 63.3 months in wild-type patients. ⁵⁴ A meta-analysis of six studies incorporating 1,857 patients revealed that patients with BRAF-mutated CRLM undergoing hepatectomy had significantly worse OS (HR 2.80, 95% CI: 2.09–3.77) and DFS (HR 2.29, 95% CI: 2.41–3.71) compared with BRAF wild-type patients. ⁵⁵ Valine substitution for glutamate in codon 600 (V600E) is the most common BRAF mutation in metastatic CRC, and patients with V600E and non-V600E BRAF mutations have significantly different prognosis. ⁵⁶ A multicenter study of 9,643 patients with metastatic CRC reported that non-V600E BRAF mutations occurred in 208 (2.2%), accounting for 22% of all RAF mutations. ⁵⁷ The median OS in patients with non-V600E BRAF mutations was 60.7 months, significantly longer than both those with V600E mutation (11.4 months) and wild-type BRAF (43.0 months). ⁵⁷ Another study of 849 CRLM patients undergoing curative-intent resection from Johns Hopkins University revealed that V600E BRAF mutation was associated with significantly worse OS (HR 2.76, 95% CI: 1.74–4.37) and DFS (HR 2.04, 95% CI: 1.30–3.20), while patients with non-V600E BRAF mutations had similar survival to those with wild-type BRAF genotypes. ⁵⁸

Given the poor prognosis for patients with V600E BRAF mutation, several scholars do not recommend surgical treatment for such patients. ⁵² A study from Mayo Clinic found that patients with BRAF V600E metastatic CRC undergoing metastasectomy had significantly longer median OS (29.1 vs. 22.7 months) and progression-free survival (13.6 vs. 6.2 months) compared with the non-metastasectomy patients. ⁵⁹ A recent multicenter study from Japan reported a good median OS of 31.1 months in their 33 patients with BRAF V600E-mutated CRLM after hepatectomy. ⁶⁰ Similarly, in a study from France, 49 BRAF V600E-mutated CRLM patients also obtained a good median OS of 34 months after hepatectomy. ⁶¹ Accordingly, V600E BRAF mutation should not be considered as a contraindication for metastasectomy.

Coexisting Gene Mutations

TP53 mutations are present in the majority of CRC patients, ^{62 63} which have been detected in 82% of CRLM patients in the authors' center. ¹⁵ The effect of TP53 mutations on the prognosis of patients after resection of CRLM has been reported inconsistently in the literature, but most studies support that TP53 mutations have no prognostic effect on these patients. ^{12 62 63 64 65} A recent study from Memorial Sloan Kettering Cancer Center revealed that co-alteration of RAS/BRAF and TP53 was associated with extremes of survivorship among metastatic CRC patients following complete metastasectomy. ⁶⁶ Compared with patients with RAS/BRAF or TP53 mutations alone, patients with co-altered RAS/BRAF and TP53 had the worst OS. ⁶⁶ Likewise, Chun et al reported that double mutation of RAS and TP53, present in 31.4% of CRLM patients, was correlated with reduced survival after resection of CRLM. ⁶² APC and PIK3CA mutations alone also showed no effect on the prognosis of patients with CRLM after surgical resection. ^{13 64 67} However, cooccurring mutation of APC and PIK3CA was found to be associated with worse RFS and OS after resection of CRLM. ¹³ This may be explained by the cooperativity between coexisting gene mutations in tumor invasion and malignant transformation. ⁶⁸

A study of Kawaguchi et al from Japan reported that mutations in RAS, TP53, and SMAD4 were independently associated with worse OS and RFS among patients with CRLM after hepatectomy. ¹² Their further analyses revealed that patients with coexisting mutations in RAS, TP53, and SMAD4 had significantly decreased OS and RFS compared with those who harbored any two and one or none of these gene mutations. ¹²

ERBB2 Amplification

The protein ERBB2 (also known as HER2) is a member the EGFR family of receptor tyrosine kinases, which also includes EGFR (ERBB1), ERBB3, and ERBB4. ⁶⁹ The overexpression of ERBB2 induced by ERBB2 gene amplification is associated with aberrant signaling pathways involved in cell proliferation and apoptosis. ⁶⁹ ERBB2 amplification is present in 3% of patients with metastatic CRC, and has been found to be related to resistance to treatment with anti-EGFR in these patients. ^{70 71} Amplification of ERBB2 is predominantly found in CRCs located on the left colon or rectum. ⁷² Compared with ERBB2 wild-type CRC patients, ERBB2-amplified CRC patients have more aggressive disease and worse prognosis. ^{73 74}

Han et al first evaluated the prognostic impact of ERBB2 amplification on their 208 patients with CRLM undergoing hepatectomy. ⁷⁵ In their study, ERBB2 amplification was detected in 13 (6.25%) patients and associated with left-sided RAS/RAF wild-type status. ⁷⁵ Among their left-sided RAS/RAF wild-type patients, ERBB2 amplification was significantly associated with decreased OS (30.2 vs. 50.9 months) and RFS (5.77 vs. 19.97 months). ⁷⁵ Due to the small sample size, these findings need to be further confirmed.

Microsatellite Instability and Deficient Deoxyribonucleic Acid Mismatch Repair

Deficient deoxyribonucleic acid (DNA) mismatch repair (dMMR) system can cause errors during DNA replication and thus results in microsatellite instability (MSI). The judgment of dMMR is based on immunohistochemistry staining of four MMR proteins including MLH1, PMS2, MSH2, and MSH6, while the MSI status is determined by genetic sequencing. According to the extent of MSI, CRCs are classified as MSI-high (MSI-H), MSI-low, and microsatellite stable (MSS). ⁷⁶ Approximately 15% of patients with CRC harbor dMMR, with 80% due to methylation of the MLH1 gene promoter. ⁷⁷ A meta-analysis of 136 studies by Toh et al revealed that MSI-H was associated with better DFS and OS in stage II and stage III CRC, and MSI-H CRC patients had lower rate of lymph node and distant metastases. ⁷⁸

However, dMMR was only present in 5% of metastatic CRCs. ⁷⁹ A pooled analysis of 3,063 patients from four phase III studies in first-line treatment of metastatic CRC revealed that patients with dMMR had significantly worse median PFS (HR 1.33, 95% CI: 1.12–1.57) and OS (HR 1.35, 95% CI: 1.13–1.61) compared with proficient MMR patients. ⁷⁹ This study also found that dMMR was related to V600E BRAF mutation, which was detected in 34.6% of these dMMR patients. ⁷⁹ MSI-H/dMMR metastatic CRCs are less responsive to conventional chemotherapy, but respond well to immune checkpoint inhibitor therapy. In the KEYNOTE-177 trial, pembrolizumab (a programmed death 1 blockade) brought double the PFS to patients with MSI-H/dMMR metastatic CRCs compared with traditional chemotherapy (16.5 vs. 8.2 months). ⁷⁷

Sanhueza and colleagues identified 75 MSI-H patients from 1,268 patients with metastatic CRC and the results showed that patients receiving metastasectomy of metastases had significantly longer median OS than those without metastasectomy (82.0 vs. 13.9 months). ⁸⁰ A national analysis of 2,743 patients with CRLM undergoing metastasectomy from U.S. found that MSI was independently associated with worse OS (HR 1.21, 95% CI: 1.01–1.46). ⁸¹ In another study by Matteo et al from Italy, patients with MSI status also had a high rate of intrahepatic recurrence compared with those with MSS status (54% vs. 21%), although no difference was observed between the two groups of patients in terms of DFS and OS. ⁸²

Circulating Tumor DNA

Circulating tumor DNA (ctDNA) is mainly derived from the DNA fragments released into the circulation by the apoptotic or necrotic tumor cells, which has emerged as a noninvasive and highly sensitive biomarker for CRC patients. ⁸³ An increasing number of studies suggest that ctDNA could predict outcomes in patients with CRLM undergoing metastasectomy. ^{84 85}

In a recent prospective study from MD Anderson Cancer Center, perioperative ctDNA analyses of blood samples were conducted in 48 patients with CRLM undergoing hepatectomy. ⁸⁶ Patients with positive ctDNA before and after surgery (ctDNA+/+ ) just had a median RFS of 6.0 months, significantly lower than the 33.0 months of patients with negative ctDNA before and after surgery (ctDNA −/− ). ⁸⁶ Another prospective study from Denmark reported that positive ctDNA after resection of CRLM was associated with decreased RFS (HR 4.5, 95% CI: 2.1–9.5) and ctDNA could be detected before radiological recurrence (up to 10.6 months in advance, with a median time of 3.1 months). ⁸⁷

Tie et al also found that patients with detectable ctDNA after CRLM resection had both worse RFS (HR 6.3, 95% CI: 2.58–15.2) and OS (HR 4.2, 95% CI: 1.5–11.8) compared with those without detection of postoperative ctDNA. ⁸⁸ In addition, among their 11 patients with ctDNA detected after surgery, ctDNA clearance occurred in 3 patients during adjuvant chemotherapy, of which 2 patients still did not relapse. All the 8 patients with ctDNA detected continuously after postoperative adjuvant chemotherapy developed recurrence. However, ctDNA clearance during neoadjuvant chemotherapy was not related to better RFS. These findings suggest that end-of-treatment ctDNA detection is of greater significance.

Summary

Tumor mutation analysis can provide valuable prognostic information and guide perioperative systemic therapy in patients with CRLM. The genomic landscape of patients with CRLM has been well delineated. RAS mutation status is the core molecular biomarker for CRLM patients, and its prognostic effects are related to the surgical procedures, primary tumor location, and specific codons mutations. Although V600E BRAF mutation is associated with poor prognosis, metastasectomy could also provide survival benefits for well-selected patients. Coexisting gene mutations showed more powerful prognostic effects compared with single gene mutation. Furthermore, ctDNA has been emerged as a promising biomarker for patients with CRLM. Postoperative ctDNA analysis could not only predict recurrence but also detect recurrence earlier than the current standard surveillance tools.

Funding Statement

Funding None.