Locoregional treatment for colorectal liver metastases aiming for precision medicine

Abstract In patients with colorectal liver metastases (CLM), surgery is potentially curative. The use of novel surgical techniques and complementary percutaneous ablation allows for curative‐intent treatment even in marginally resectable cases. Resection is used as part of a multidisciplinary approach, which for nearly all patients will include perioperative chemotherapy. Small CLM can be treated with parenchymal‐sparing hepatectomy (PSH) and/or ablation. For small CLM, PSH results in better survival and higher rates of resectability of recurrent CLM than non‐PSH. For patients with extensive bilateral distribution of CLM, two‐stage hepatectomy or fast‐track two‐stage hepatectomy is effective. Our increasing knowledge of genetic alterations allows us to use them as prognostic factors alongside traditional risk factors (e.g. tumor diameter and tumor number) to select patients with CLM for resection and guide surveillance after resection. Alteration in RAS family genes (hereafter referred to as “RAS alteration”) is an important negative prognostic factor, as are alterations in the TP53, SMAD4, FBXW7, and BRAF genes. However, APC alteration appears to improve prognosis. RAS alteration, increased number and diameter of CLM, and primary lymph node metastasis are well‐known risk factors for recurrence after CLM resection. In patients free of recurrence 2 y after CLM resection, only RAS alteration is associated with recurrence. Thus, surveillance intensity can be stratified by RAS alteration status after 2 y. Novel diagnostic instruments and tools, such as circulating tumor DNA, may lead to further evolution of patient selection, prognostication, and treatment algorithms for CLM.

multidisciplinary approach, which for nearly all patients will include perioperative chemotherapy. Although optimal response to chemotherapy is significantly related to better survival, surgery for patients with suboptimal response is not a contraindication.
At the University of Texas MD Anderson Cancer Center, FOLFOX (leucovorin calcium, fluorouracil, and oxaliplatin) with bevacizumab for 8-12 weeks and surgery 5-6 weeks later is the standard preoperative treatment for the purpose of minimizing postoperative complications and maximizing treatment response according to the results of previous studies and the fact that bevacizumab is effective in all patients with CLM irrespective of RAS mutation status. [3][4][5] In fact, preoperative chemotherapy was used in more than 80% of patients in 2007 and more than 90% of patients in 2020. In the majority of patients, the number of preoperative chemotherapy cycles was limited to six ( Figure 1A). Since 2004, oxaliplatin-based regimens have become the most frequently used regimens, given to almost 60% of patients who receive preoperative chemotherapy ( Figure 1B). The vascular endothelial growth factor inhibitor bevacizumab is the first-choice biological agent and has been administered to more than 60% of patients since 2011 ( Figure 1C).
Patients with CLM are a heterogeneous cohort, and multidisciplinary treatment must be tailored for individual patients. The classic prognostic model used to select patients for CLM resection was developed in the late 1990s and included only clinicopathologic factors, 6,7 but somatic gene alterations have become increasingly used in the surgical decision-making for patients with CLM. 1,8 In this review article, we discuss four topics that are fundamental in optimizing locoregional treatment of CLM: parenchymal-sparing hepatectomy and/or ablation; two-stage hepatectomy or fast-track two-stage hepatectomy; somatic gene alterations and canonical pathways associated with prognosis; and posttreatment surveillance.

| PAREN CHYMAL-S PARING HEPATEC TOMY AND/OR AB L ATION FOR S MALL CLM
Considering recurrences after resection for initial CLM occur in ~70% of patients, 9 it is important to develop a treatment strategy that anticipates repeat locoregional treatment for recurrent CLM. Therefore, the need for parenchymal-sparing hepatectomy (PSH) and/or ablation is increasing, especially for small CLMs distant from the hilar plate.

| Parenchymal-sparing hepatectomy
PSH is recommended rather than non-PSH because PSH has been shown to result in better survival and higher rates of resectability of recurrent CLM. 10 Some studies have reported improved outcomes of non-PSH versus PSH 11,12 ; however, these studies were not performed using cohorts appropriately matched by number or size of CLM.
Mise et al 13 compared PSH with non-PSH in patients with solitary CLM measuring less than 30 mm and found that patients who underwent PSH had better survival from the date of operation and from the date of recurrence compared to patients who underwent non-PSH ( Figure 2). The reason was that repeat hepatectomy for recurrent CLM was more frequently performed in the PSH group than in the non-PSH group (in 68% vs 24% of patients with recurrent CLM, P < 0.01). That is, PSH maintained "salvageability." In patients with small CLM, preoperative chemotherapy can render the CLM invisible on cross-sectional imaging and make it difficult to identify the CLM intraoperatively. Thus, at MD Anderson Cancer Center, for CLM less than 2 cm in diameter that are not located on the liver surface, placement of fiducial markers is recommended before preoperative chemotherapy. Passot et al 14  Nishioka et al reported that in patients with CLM who underwent R0-intent resection, the rate of local recurrence (recurrence at the resection margin) was not related to either surgical margin width or somatic gene mutation status, and that OS was not related to surgical margin status. 15 The study implied that surgical margin width should not be tailored according to tumor biology to reduce the rate of local recurrence.

| Ablation
At MD Anderson Cancer Center, ablation is guided by cross-sectional imaging because it allows for optimal identification of tumor extent and close monitoring of ablation margins. In 2016, Shady et al 16  Therefore, as local tumor PFS is conditioned by margins and RAS mutational status, 5-mm margins after ablation will be suitable for patients with wildtype RAS, but patients with mutant RAS or unknown RAS status will need margins of at least 10 mm to reduce the risk of local recurrence. In a subsequent retrospective study at MD Anderson that analyzed outcomes when ablation margins were monitored using state-of-the-art 3D imaging, RAS alteration was no longer an independent risk factor for local recurrence, 18 suggesting that the ablation margin may have been appropriately performed using state of the art 3D reconstruction.

F I G U R E 1
Chronological trends from 1998 through 2020 in preoperative chemotherapy for patients with colorectal liver metastases with respect to (A) number of cycles, (B) cytotoxic agents, and (C) molecular targeted therapy. VEGF, vascular endothelial growth factor; EGFR, epidermal growth factor receptor.

| Planned incomplete resection with completion ablation
MD Anderson Cancer Center now uses a sequential treatment strategy for patients with extensive distribution of CLM, consisting of a planned incomplete (R2) resection followed by postoperative image-guided percutaneous completion ablation of the remaining, intentionally untreated lesions. 19 In a retrospective study comparing outcomes between this completion ablation strategy and standard intraoperative ablation, Okuno et al found that the 5-y cumulative incidence of local tumor progression was significantly lower in the completion ablation group than in the standard intraoperative ablation group (31.7% vs 62.4%, P = 0.030), whereas the 5-y OS rate did not differ between groups (53% for completion ablation vs 42% for intraoperative ablation, P = 0.41). The complication rate was also significantly lower for completion ablation (31.7% for completion ablation vs 62.4% for intraoperative ablation, P = 0.03). 20 The study suggested that postoperative ablation can avoid the risk of resection for small CLM that are difficult to approach intraoperatively.

| T WO -S TAG E HEPATEC TOMY FOR B IL ATER AL CLM
Bilaterally distributed CLM pose special challenges. In 2000, Adam et al proposed treating such CLM with "two-stage hepatectomy" (TSH) to achieve R0 resection and avoid postoperative liver failure due to a small future liver remnant. 21,22 In the typical TSH, the first-stage operation involves minor resections of metastatic lesions within the left liver. This is followed by embolization of the right portal vein. After regeneration and adequate hypertrophy of the left side of the liver, a formal right hepatectomy is performed as the second-stage operation.
At MD Anderson Cancer Center, 148 of 1779 patients (8.3%) with preoperative oxaliplatin-and/or irinotecan-based chemotherapy for initial CLM underwent both stages of TSH during 1998-2020 (Table 1)  In a separate study, our group found that in patients who required a major hepatectomy, hepatectomy combined with ablation was associated with a lower 5-y OS rate than TSH (24% vs 35%, P = 0.01), a higher rate of postoperative major morbidity (32% vs 14%, P = 0.003), and a higher incidence of postoperative hepatic insufficiency (28% vs 6%, P < 0.0001). This study shows that the use of simultaneous ablation should be avoided in patients undergoing major hepatectomy. 23 One of the major disadvantages of TSH is dropout prior to the second-stage hepatectomy because of the long interval between stages. To shorten the interval, we set up a hybrid room that combines the capabilities of a standard operating room with those of an interventional radiology suite. 24 This room contains a fluoroscopy  In addition to RAS alteration, TP53 alteration and SMAD4 alteration were reported to be independently associated with worse prognosis. 8 Figure 4A). After 10 cycles of therapy with FOLFOX and bevacizumab, a partial response was obtained ( Figure 4B). Extended left hepatectomy with common bile duct and the caudate lobe resection was performed because disease invaded the hilar plate. Seven mo later, a recurrence was identified in a retroportal lymph node ( Figure 4C). After 12 cycles of therapy with XELOX (capecitabine and oxaliplatin), a partial response was obtained again ( Figure 4D).

| G ENE ALTER ATI ON S A SSO CIATED WITH PROG NOS IS AF TER CLM RE S EC TI ON
Lymphadenectomy was performed because the recurrent lesion was localized. At the most recent follow-up, 7 y after lymphadenectomy, there had been no evidence of recurrence even though the patient did not receive chemotherapy. Gene panel analysis covering 50 genes revealed a CTNNB1 alteration, but no RAS, TP53, or SMAD4 alterations ( Figure 4E). The prognostic impact of CTNNB1 alteration has not been elucidated in CLM. This case suggests that genetic mutation analysis may be useful in determining whether an extensive, highly invasive operation is warranted.

| P OS T TRE ATMENT SURVEILL AN CE
The principal aim of surveillance after resection of CLM is to permit early detection of disease recurrence and thus enable physicians to deliver repeated locoregional therapy in a timely manner.
In patients with recurrence in the liver or lung after CLM resection, repeat hepatic or lung metastasectomy along with chemotherapy for recurrence is associated with better survival than chemotherapy alone. 37,38 However, frequent follow-up tests and imaging are associated with increased medical costs. Therefore, the surveillance protocol should be based on the patient's risk of recurrence.
According to the National Comprehensive Cancer Network's 2020 guidelines, the follow-up interval after resection for stage IV colorectal cancer can be extended starting 2 y after resection because ~70% of recurrences occur within the first 2 y. 9,39,40 For instance, the guidelines recommend that oncologists perform serum carcinoembryonic antigen measurements and axial imaging every 3-6 mo until 2 y after resection and every 6-12 mo starting 2 y after resection. The risk factors associated with recurrence within the initial 2 y after resection of CLM include primary lymph node metastasis, greater number and size of CLM, and RAS alteration.
However, the only risk factor associated with recurrence beyond 2 y after resection is RAS alteration. 41 The data suggests that in patients with RAS alteration, the more frequent surveillance should be main- proved survival in patients who were negative for ctDNA before and after surgery. Patients who were positive for ctDNA before surgery but negative for ctDNA after surgery had survival similar to that of patients who were negative for ctDNA both before and after surgery ( Figure 6A,B). 43 Similarly, Øgaard et al 44   Clinical Trials Support Resource.

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors declare no conflicts of interest for this article.

E TH I C A L A PPROVA L
This study was performed in accordance with the Declaration of