The Role of Ablative Radiotherapy to Liver Oligometastases from Colorectal Cancer

This review describes recent data supporting locoregional ablative radiation in the treatment of oligometastatic colorectal cancer liver metastases. Stereotactic body radiotherapy (SBRT) demonstrates high rates of local control in colorectal cancer liver metastases when a biologically equivalent dose of > 100 Gy is delivered. Future innovations to improve the efficacy of SBRT include MRI-guided radiotherapy (MRgRT) to enhance target accuracy, systemic immune activation to treat extrahepatic disease, and genomic customization. Selective internal radiotherapy (SIRT) with y-90 is an intra-arterial therapy that delivers high doses to liver metastases internally which has shown to increase liver disease control in phase 3 trials. Advancements in transarterial radioembolization (TARE) dosimetry could improve local control and decrease toxicity. SBRT and SIRT are both promising options in treating unresectable metastatic colorectal cancer liver metastases. Identification of oligometastatic patients who receive long-term disease control from either therapy is essential. Future advancements focusing on improving radiation design and customization could further improve efficacy and toxicity.


Introduction
An estimated 150,000 people are diagnosed with colorectal cancer (CRC) each year in the USA [1]. Approximately 25-30% of CRC patients present with either synchronous or metachronous CRC liver metastases (CLMs) [2,3]. For those with oligometastatic disease, usually defined as fewer than 5 sites of disease, the treatment paradigm includes definitive local treatment of all disease sites as clinical studies have demonstrated that metastatic control increases progression-free survival (PFS) and overall survival (OS) in various disease sites [4]. If feasible, surgical resection of CLMs can produce clinical outcomes comparable to earlier stages [5,6].
However, up to 90% of newly diagnosed liver metastases are inoperable [7]. While multiagent chemotherapy can also convert some patients into surgical candidates, a significant number of CLMs will remain inoperable, emphasizing the need for nonsurgical focal therapy options [8,9].
Radiation therapy is a non-invasive modality that can provide extended local control (LC) when delivered at high doses [10]. Internal beam radiation and external beam radiation (EBRT) are commonly used to achieve disease control within the liver. Within EBRT, SBRT treats liver lesions with millimeter precision. Meanwhile, selective internal radiotherapy (SIRT) delivers high doses of radioactive yttrium-90 (y-90) to CLMs utilizing TARE.
Eric Ku and John Yeakel contributed equally to this manuscript.

Introduction to SBRT
SBRT refers to submillimeter delivery of ablative radiation to tumors outside of the central nervous system in one to five treatments. It involves integration of high-resolution imaging, non-invasive positioning techniques, and respiratory motion management to design conformal beams targeted around a tumor with an additional goal to reduce the amount of radiation to adjacent normal tissue. Currently, it is an essential part of the treatment paradigm for lung, pancreas, and prostate cancers, as well as the palliation of metastatic lesions from a broad array of primary malignancies.

SBRT for CLMs
Investigating the optimal SBRT dose to liver metastases, retrospective studies looking at hepatic metastases of various etiologies have noted the importance of achieving a biological effective dose (BED) of at least 100 Gy to achieve durable LC [11,12]. Published reports using SBRT to treat liver metastases have shown actuarial LC rates ranging from 50 to 100% with higher doses associated with better LC. A multiinstitutional phase I/II study of SBRT for liver metastases showed the safety of dose escalation from 36 Gy up to 60 Gy in 3 fractions with 2-year actuarial in-field LC rates of 92% [13]. Lesions smaller than 3 cm had a 100% LC at 2 years.
The few retrospective studies that have focused solely on CLMs treated with SBRT have found a possible benefit of BED greater than 110 Gy [13][14][15][16][17][18][19][20]. A multi-institutional pooled analysis of 102 CLMs in 65 patients with 1 to 4 hepatic sites appreciated significant differences in LC when utilizing a BED cutoff value of 75 Gy [21]. On tumor control probability modeling, they estimated a BED of 117 Gy was needed to achieve 90% LC at 1 year. Joo et al. stratified lesions by BED (< 80 Gy, 100-112 Gy, > 132 Gy) and noted an improvement in the 2-year LC from 83 to 89% with BED > 130 Gy compared to 100-112 Gy, a small but not statistically significant difference [22 •]. BED greater than 132 was predictive of OS and LC on univariate analysis. However, no differences in PFS were noted between the three groups.
A large prospective Dutch-Belgian registry followed 515 patients with 668 liver metastases treated using SBRT with a variety of fractionations ranging from 54 to 60 Gy in 3, 5, 8, and 12 fractions [23]. Eighty percent of patients presented with CLMs and approximately 50% completed prior therapy. Most patients completed SBRT to 1 liver lesion. Overall LC after RT was 87% at 1 year, 75% at 2 years, and 68% at 3 years. No differences in LC were appreciated by SBRT fractionation scheme. OS at 1, 2, and 3 years was 84%, 63%, and 44%. The rate of grade 3 or higher CTCAE gastrointestinal toxicity was 3.9% including one event of grade 5 hepatobiliary toxicity. Analysis of treated lesions by liver segment did not reveal differences in toxicity outcomes but an association between segment 3 lesions and lower LC approached significance (HR 3.72, 0.94-14,63, P = 0.06). Only 14 segments and 3 livers were treated making these results prone to sample bias. Recent retrospective studies note similar findings with 3-year LC rates ranging from 68 to 79% (Table 1).
Few prospective trials have investigated SBRT solely in the setting of oligometastatic CLMs (Table 2). Rusthoven et al. at the University of Colorado first demonstrated 2-year LC rates greater than 90% with SBRT to 1 to 3 liver metastases after 60 Gy in 3 fractions but only included 32% CRC patients [13]. Shortly thereafter, Erasmus University published their prospective experience of SBRT for CLMs looking at patients with limited hepatic involvement unamenable to radiofrequency ablation and treated with 37.5 Gy in 3 fractions [20]. Nine local failures were seen in the 31 CLMs treated, which is higher compared to prior large retrospective studies but may be explained by the lower dose given (BED = 84.4 Gy). A phase 2 Italian trial conducted by Predicting which patients with oligometastatic CLMs achieve durable disease control after SBRT remains unclear. Utilizing clinical risk score (CRS) based on the number of metastases, size of the lesion(s), serum CEA level, disease-free interval, and presence of lymph nodes at diagnosis, Creasy et al. appreciated significantly longer disease-free interval after resection in those with lower CRS [31]. Future SBRT studies implementing a multifactorial stratification tool like CRS could help identify those that would significantly benefit from ablative radiation.
The importance of patient selection is further demonstrated in a recently published phase I trial from UT Southwestern looking at single-fraction SBRT of 35 to 40 Gy [28••]. Eligible patients had good performance status with less than 5 liver metastases that were at least 2 cm away from the portal vein and its bifurcation within the liver due to concern for radiation toxicity. At median follow-up of 28.3 months, only two local failures were noted. Despite 42.5% patients having extrahepatic disease, 4-year OS was 49.7%, and only three grade 2 gastrointestinal toxicities were noted. None of the patients developed radiation-induced liver disease. The high LC is likely explained by the high BED delivered (> 150 Gy) although BED estimates for singlefraction treatment can be inaccurate. The total disease burden was also limited with 87.9% of patients having fewer than 5 sites of disease and 66.6% having 2 or fewer. All patients in this study only received treatment to 1 to 2 liver metastases with a median maximal tumor length of 2 cm (0.5-5 cm). Additionally, only allowing tumors further from major vessels may select for patients with a lower risk of further disease spread. A limitation of this study is low incidence of inoperable patients with limited hepatic involvement reflected by 9-year accrual of only 33 patients. Nonetheless, this trial encourages further investigation into SBRT providing longterm cure for those with oligometastatic CLMs.
In summary, retrospective evidence shows that SBRT provides durable LC of CLMs supported by increased LC with higher BED delivered (> 110-130 Gy). The low rates of PFS may be explained by higher overall disease burden and extrahepatic disease. Future studies may benefit from scoring systems to estimate overall disease burden to select patients most likely to benefit from SBRT.

Proton-Based SBRT
The use of protons with SBRT is potentially attractive given the advantage of lack of exit dose when compared to photonbased techniques. Several treatment planning comparison studies have been reported showing improved normal tissue dosimetry of proton-based radiotherapy compared to photon-based radiotherapy in the treatment of liver, lung, and adrenal lesions [30••, 32, 33]. SBRT with protons have been used to treat hepatocellular carcinoma (HCC) using various hypofractionated regimens with minimal acute toxicity, no reported radiation-induced liver disease, and 3-year PFS of 60% [34][35][36][37]. Phase I data of ablative proton SBRT (BED 180 Gy) shows it to be well tolerated in patients with limited liver metastases with no acute toxicities grade 2 or higher [38].

Combination SBRT and Immunotherapy for Extrahepatic Control
Despite high LC rates with SBRT, median OS remains less than 3 years. This is attributed to the high rates of out-offield disease progression within the first 2 years emphasizing the importance of systemic control [39]. Examining changes in PD-L1 expression and CD8 + T-cell infiltration before and after chemoradiation in rectal cancer patients, Lim et al. found that persistently high tumor surface PD-L1 and low CD8 + T-cells after treatment correlated with an increased risk of distant disease relapse, suggesting anti-PD-1 therapy could improve outcomes in rectal cancer [40]. SBRT has been shown to promote tumor antigen release and increase T-cell receptor diversity [41] and the combination of checkpoint inhibitors and SBRT has been extensively investigated in preclinical and early phase trials [42] (Fig. 1A). However, the dose and fractionation of radiation to prime the immune system against colorectal cancer remain unknown. A recent multi-institutional phase 2 trial examined 8 Gy per day for 3 days versus 2 Gy twice a day for 2 days for those with 1 to 2 microsatellite stable CLMs given in combination with anti-CTLA-4 and anti-PD-1 therapy [43•].
Despite noting significant changes in immune cell populations within treated tumor specimens, tumor regression outside of the radiation field did not occur in either arm. Further trials examining checkpoint blockade with SBRT for CLMs are underway and will provide more information (NCT02837263, NCT03927898).

Integrating Genomics to Improve SBRT for CLM
Previous retrospective studies have noted a BED > 110 Gy needed to achieve durable LC, suggesting that CLMs may be more radioresistant than other primary tumor types. Ahmed et al. developed a radiation sensitivity index (RSI), with higher scores indicating more radioresistance, based  These studies raise the question of whether radiation dose should be customized based on genomic alterations of CLMs (Fig. 1B). They also open further exploration to improve outcomes after SBRT of CLMs including combination locoregional therapy to overcome radioresistance (NCT03963726) and the addition of radiosensitizers to augment SBRT (NCT01569984, NCT03223779).

MRI-Guided Radiotherapy for CLMs
Most EBRT machines utilize x-ray imaging during SBRT to guide treatment. Due to poor soft tissue delineation of liver targets on x-rays, fiducials are often placed within metastatic liver lesions as surrogate markers for targeting and assessing motion [45]. However, fiducial placement carries risk of pain, pneumothorax, bleeding, and fiducial migration [46].
The integration of magnetic resonance imaging (MRI) into radiation treatment machines allows for real-time MRI acquisition during SBRT, resulting in improved soft tissue visualization, commonly referred to as MRgRT [47]. Cine-MRI captures images every 4 to 8 s to allow for real-time treatment monitoring of tumor motion without the need for fiducials [48]. Stereotactic MR-guided online-adaptive radiotherapy uses daily MRI to modify dose and treatment plan in response to changes in tumor response, target motion, and organs at risk motion.
Studies exploring high-dose focal RT using MRgRT for CLMs are in the early stages. In a multi-institutional retrospective review of 26 patients with both primary and secondary liver malignancies treated to 50 Gy in 10 fractions, Rosenberg et al. showed freedom from local progression of CLMs at 75%. Only 2 patients experienced grade 3 toxicities and none experienced grade 4 or higher toxicity [49]. van Dams et al. recently published their phase I results exploring the safety and feasibility of MR-guided SBRT in 20 patients treated with 25 liver tumors (8 primary, 12 secondary) [50•]. This trial examined a 3-dose tier for single and multiple lesion plans while prioritizing limiting the dose to normal liver. They appreciated a LC benefit when liver tumors were treated with a BED > 100 Gy consistent with previous literature [11,12,51]. No acute grade 3 toxicities were noted and only 1 patient had late grade 3 duodenal toxicity. With the ability to deliver ablative doses of radiotherapy with higher precision, MRgRT is a promising tool with the potential to overcome CLM radioresistance while minimizing toxicity due to real-time target and organs at risk visualization (Fig. 1C). Prospective trials comparing CT-based SBRT and MRgRT outcomes and toxicity are eagerly awaited.

Transarterial Radioembolization
Transarterial radioembolization (TARE) with y-90 has been shown to be an effective treatment for CRC metastases to the liver. The beta radiation's mean tissue penetration of 2.5 mm allows for high doses to be administered with doses > 100 Gy frequently delivered.
The ideal dose to allow for disease control without significant toxicity to the normal liver or lung is still unclear. In the setting of HCC, several studies have shown that a higher absorbed tumor dose is a superior prognostic marker of survival with one study even suggesting 205 Gy as an appropriate target dose [52,53]. Dosage data in CRC is less robust, but there are also studies that suggest improved benefit with higher doses. Alsultan et al. looked at dose-response in patients with CLMs treated with y-90. They reported median absorbed dose by tumor response of 196 Gy for complete response, 177 Gy for partial response, 72 Gy for stable disease, and 95 Gy for progressive disease. Based on their data, they concluded that a tumor absorbed dose of > 189 Gy was recommended as a target to achieve the best response [54]. Further study is needed in this arena, particularly in the setting of CRC metastases.

Resin vs Glass TARE
Two devices are available for y-90 delivery, glass spherebased y-90 (TheraSphere™, Boston Scientific Corporation, Marlborough, MA) and resin sphere-based y-90 (SIR-Spheres®, Sirtex Medical, Sydney, New South Wales, Australia). The main difference between the two devices lies in the activity per sphere with glass spheres having an activity on the order of 50 times that of resin spheres. This translates into a significantly larger amount of resin beads required to deliver the same amount of radioactivity which, in turn, can affect the biodistribution of the spheres and confers a potentially higher embolic effect to resin spheres, potentially precluding delivery of the entire prescribed dose if stasis is met first. The dosimetry of the two devices has historically been calculated differently, as well. The details are beyond the scope of this review, but briefly, resin-based therapy dosimetry has largely been based on a body surface area (BSA) calculation and glass-based therapy dosimetry on a medical internal radiation dose (MIRD) calculation.

SIRT for CLMs
While resin-based therapy is specifically FDA approved for CRC metastases, and glass-based therapy is only FDA approved for HCC, several studies have shown the safety and efficacy of both of these devices in the chemorefractory or salvage setting of CRC. A single-center study of 302 patients treated with resin-based y-90 radioembolization demonstrated a median survival of 10.5 months after radioembolization performed in the salvage setting with relatively limited clinical toxicity. Hepatic tumor involvement was found to be a statistically significant factor in OS with 0-25%, 26-50%, and 51-75% having 11.6, 9.1, and 5.6 months of median survival [55]. Another study looking at 531 patients at multiple centers who were treated with glass-based y-90 radioembolization similarly showed a median OS of 10.6 months with low toxicity. This study also showed better survival outcomes for hepatic tumor burden no more than 25% on univariate analysis [56]. These studies show the benefit of y-90 in this setting and suggest improved outcomes with better patient selection, such as those with lesser disease burden.
The favorable results of these studies and earlier smaller studies prompted investigation into the use of TARE in the first-line setting in an attempt to improve patient outcomes. Three multicenter, randomized, worldwide phase 3 trials (FOXFIRE, SIRFLOX, and FOXFIRE-Global) with 1103 total patients were performed, with a combined analysis, comparing y-90 radioembolization plus chemotherapy to chemotherapy alone in treatment-naïve patients. While the study did show better liver control in the radioembolization group with a lower cumulative incidence of the first progression in the liver (31% vs. 49%) and higher objective response (72% vs. 63%), there was no statistically significant difference in the primary endpoint of OS (22.6 months vs. 23.3 months) [57••].
However, post hoc analysis of some of this data yielded interesting results for potential use of TARE in the firstline setting. The data from the SIRFLOX and FOXFIRE-Global trials was looked at in the setting of patients with right-sided metastatic CRC, based on right-sided primary being a poor prognostic factor. Looking at these 2 studies alone, there was again no statistically significant difference in OS when looking at all 739 patients (24.3 months vs. 24.6 months), but in the 179 patients with right-sided primary, an improved OS was seen in those receiving y-90 radioembolization in addition to chemotherapy (22.0 months vs. 17.1 months) [58]. Another post hoc analysis of 472 patients from SIRFLOX looked at secondary technical resectability of the two arms. While there was no difference in resectability at baseline (11.9% vs. 11.0%), both arms demonstrated an increase in resectability compared to baseline, with more patients resectable in the radioembolization group compared to the control (38.1% vs. 28.9%) [59]. While these findings are somewhat limited due to the post hoc nature of these studies, their findings suggest that TARE may still have a role in the first-line setting in select patients. Table 3 summarizes recent studies examining SIRT for CLMs.

Radiation Segmentectomy for Oligometastatic Disease
Another potential use for y-90 in the treatment of CRC is radiation segmentectomy (RS). RS uses higher, ablative doses of radiation, often > 200 Gy, injected into a smaller territory, to treat tumors primarily when they are in only 1 or 2 segments of the liver. This technique is best suited for patients with limited disease, but who are not candidates for resection or ablation. The higher doses to a smaller area, segmental or subsegmental delivery, result in improved necrosis to the tumor and surrounding liver. This technique has been studied extensively in the treatment of HCC where it has been shown to be safe, efficacious, and even potentially curative [61][62][63]. However, its role in CRC metastases remains unclear given their hypovascularity, frequent multifocality, and potential for micrometastases not visible on imaging. Small studies have shown excellent results in the safety and efficacy of this technique in CRC and other metastases. One study looking at initial efficacy treated 10 patients with this technique with 5 showing complete response, 1 with partial response, and 4 with stable disease with a mean dose of 261 Gy [64]. Another study treating 14 tumors in 10 patients showed 2-and 3-year local tumor PFS of 83% and 69% and OS of 41.5 months with a mean delivered dose of 293 Gy [60]. These results are promising for the use of this technique in treatment of patients with oligometastatic disease.

Improving TARE Dosimetry
Advances in dosimetry may further improve the efficacy and safety of TARE. Much of the early data acquired on TARE is in the setting of previously mentioned dosimetry methods which have significant limitations. More centers are now using the partition model which is more work-intensive, but theoretically more accurate and patient-specific taking into consideration tumor volume, normal liver volume, tumor to normal ratio, and lung shunt. Currently, technetium-99 m macroaggregated albumin (Tc-99 m MAA) is injected into the hepatic artery during a mapping angiogram primarily to evaluate the lung shunt. While previously calculated with planar imaging, this is now performed via SPECT/CT imaging for more accurate calculations. These Tc-99 m MAA SPECT/CT images can also be used to improve dosimetry with more accurate tumor to normal absorbed dose ratios. This is, unfortunately, somewhat limited by the differential distribution patterns of Tc-99 m MAA and y-90 microspheres. There is ongoing research into other radiolabeled particles which may better approximate the y-90 distribution pattern. Technological advancements in y-90 PET/CT imaging have allowed for a more accurate assessment of posttreatment y-90 distribution and, as a result, a more accurate calculation of tumor absorbed dose [65•] (Fig. 1D). This information can be used to determine undertreated areas of tumor which may require additional therapy, such as a more selective TARE, SBRT, or other locoregional therapies. A recent study combining EBRT and SIRT utilizing y-90based dosimetry in 10 HCC patients appeared safe [66]. Further studies are needed regarding the safety and efficacy of TARE using these more advanced dosimetry methodologies. Larger population studies such as metaanalyses are needed to assess the effectiveness of SIRT and SBRT depending on various CLM characteristics (i.e., segment location, proximity to vessels, tumor volume, number of lesions genomic profile).

Conclusions
SBRT and SIRT are ablative locoregional therapies that provide durable LC with minimal toxicity, but their role in oligometastatic CLMs requires further investigation. Several promising areas of SBRT research for CLMs are underway including combination with immunotherapy to combat extrahepatic disease, genomic sequencing to reveal customized approaches to overcoming radioresistance, and enhanced precision targeting with MgRT. TARE in combination with chemotherapy has been shown in multicenter phase 3 trials to improve liver disease control compared to chemotherapy alone. Subgroup analysis reveals a possible OS benefit for those with reduced liver involvement. Radiation segmentectomy delivers even higher ablative doses (> 200 Gy) and has been utilized in treatment of HCC. Y-90 TARE dosimetry with PET could reveal opportunities to combine SIRT and SBRT together. Further refinement of those with minimal metastatic burden could result in a prolonged disease-free interval after SBRT and/or SIRT of CLMs.

Conflict of Interest
The authors declare no competing interests.

Human and Animal Rights and Informed Consent
This article does not contain any studies with human or animal subjects performed by any of the authors.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.