Radioembolization for Treatment of Salvage Patients with Colorectal Cancer Liver Metastases: A Systematic Review

DISCUSSION

From the studies included in this systematic review, survival proportions of approximately 50% at 12 mo after treatment were found. Therefore, in this group of salvage CRCLM patients, who otherwise have no regular treatment options and a life expectancy of less than 6 mo, ⁹⁰Y radioembolization seems to be a hopeful treatment option (3,4).

⁹⁰Y radioembolization is an emerging treatment for CRCLM and is performed as monotherapy or with concomitant chemotherapy. Over the past few decades, multiple phase 1 and 2 trials, randomized clinical trials, cohort studies, and several reviews have been published on ⁹⁰Y radioembolization. However, the reviews focused mainly on technical details concerning the angiographic procedures, did not focus primarily on liver metastases of colorectal cancer, or assessed only a selection of the available evidence (6,8,37,38). In the present systematic review, we aimed to provide a comprehensive overview of all currently available data on tumor response and survival after ⁹⁰Y radioembolization for salvage patients with CRCLM.

To evaluate treatment outcome based on multiple studies, comparability of studies and patient populations is needed. International guidelines have stated widely accepted inclusion and exclusion criteria for treatment with ⁹⁰Y radioembolization (39). These guidelines do, however, leave substantial room for subjectivity due to several poorly defined or relative contraindications (such as the amount of extrahepatic disease). Any relative contraindications must be evaluated per individual patient, leading to large inter- and intrainstitute variability in patient selection. Recently, Salem et al. published a paper focusing on research-reporting standards in the field of radioembolization (40). This paper encourages researchers to use standardized terminology in order to facilitate comparability between studies and correspondence between authors. During the selection process and data extraction for this systematic review, large heterogeneity in inclusion criteria and in patient populations and response assessment methods became apparent. This problem is also reflected in the critical appraisal (Supplemental Table 3).

Heterogeneity in patient populations arises in several aspects. First, heterogeneity in the definition of tolerable amount of extrahepatic disease for treatment exists. A widely accepted selection criterion for ⁹⁰Y radioembolization is the presence of unresectable, liver-dominant tumors. However, no consensus on the definition of liver dominance and acceptable tumor load outside the liver is currently available from the literature, even though the presence of extrahepatic disease has been shown to influence treatment outcome and survival (10,12,30,32,36). Nevertheless, sometimes details on the criterion for liver-dominant disease cannot be extracted from articles or only the rather general comment is made that “patients with extrahepatic disease not deemed clinically relevant were accepted for treatment.” Yet in 3 articles on ⁹⁰Y radioembolization as monotherapy and 4 on combination treatment with chemotherapy, the applied definition of liver-dominant disease was specified, either in terms of number (10,15), amount (size or percentage) (31), or localization of acceptable extrahepatic lesions (18,27,32). Hendlisz et al. did not accept any extrahepatic disease (i.e., treated disease limited to the liver only) (25), and Seidensticker et al. accepted patients with nonprogressive extrahepatic deposits for treatment (9).

Second, acceptable intrahepatic tumor burden is generally advised to be no more than 70% of the liver. Previous studies have confirmed the prognostic value of liver involvement for treatment outcome (12,16). Of the articles in this review, some have included only patients with a tumor burden of less than 50% (9,15), whereas others have accepted larger liver involvements. For example, one quarter of patients described in the study by Cianni et al. had more than 50% liver involvement (20).

Third, patient performance status has also been demonstrated to have prognostic value for the outcome of ⁹⁰Y radioembolization (12,30). Only half the articles that qualified for inclusion in this review reported patient performance scores. Most studies included mainly patients with a World Health Organization performance score of 0. However, in the study by Stubbs et al., for example, 50% of patients had a performance score greater than 0 (36).

Last, the prognostic value of previous systemic treatments has not been the focus of much research until now, although some studies indicate influence of prior therapies and response to prior therapies on current treatment outcome (18). However, knowledge of previous systemic treatments of included patients is important. On one hand, heavily pretreated patients generally have more advanced disease and will possibly benefit less from ⁹⁰Y radioembolization treatment. On the other hand, patients who have not yet received all standard available systemic treatment regimes may do so after ⁹⁰Y radioembolization, thereby obscuring the survival benefit of ⁹⁰Y radioembolization. To be able to compare studies, information on specific chemotherapeutic agents and number of treatments is needed. However, definitions for “a line” of chemotherapy may vary across institutions. Critical appraisal of reported previous systemic treatments (Supplemental Table 3) showed that approximately half the studies did not report, or only incompletely reported, previous systemic treatments.

Most studies on ⁹⁰Y radioembolization as monotherapy in this review have treated patients who had received a median of 3 or more previous systemic treatments. Mulcahy et al. treated a relatively large number of patients who received two or fewer lines of previous systemic therapy (12). For the studies on ⁹⁰Y radioembolization in combination with chemotherapy, previous treatments of patients were obviously related to the study inclusion criterion with respect to the concomitant treatment (Supplemental Table 2).

Besides heterogeneity in patient populations, heterogeneity in response assessment also exists between selected studies for this review. Imaging response to treatment is almost always scored with RECIST (Response Evaluation Criteria in Solid Tumors) or World Health Organization criteria. However, not always can the precise use of these criteria (with respect to application to the liver only, target lesions only, or whole body) be deduced from the article, even though this determination is of pivotal importance for the interpretation of treatment results. Some authors do state explicitly that, for example, criteria are used only for hepatic lesions and do not take extrahepatic disease into account (7,16,25).

Second, for accurate interpretation of reported tumor responses, information on other parameters of response assessment is crucial. These include interval between follow-up scans, total duration of follow-up, and interval for reported response. An often-used interval for determining response is at 3 mo after treatment (16,22,23,29,31,36), although earlier outcome assessments have also been described (15,20,30). Some authors report patient’s best response (12,25–27,32), and especially in these cases, total follow-up duration is of interest. Without information on sampling interval and follow-up duration, comparison of responses between studies becomes complicated. In addition, number of patients lost to follow-up is important, including a statement on the handling of this parameter (e.g., calculating worst-case scenario by considering patients lost to follow-up as having progressive disease) (17).

Because of the heterogeneity in response assessment, we did not find it helpful to calculate a pooled estimate for the response rate. There may even be a need for new response criteria specifically designed for locoregional therapies, in which some tumors either are not treated simultaneously or are not treated at all, when considering extrahepatic disease localizations.

Given the heterogeneities in patient populations and reporting of results, pooling of data was not appropriate and only descriptive comparisons can be made. With respect to tumor responses for ⁹⁰Y radioembolization as monotherapy, the lowest response rate was reported by Fahmueller et al. (disease control rate of 29%) (17), and the highest response rate was reported by Kennedy et al. (disease control rate of 90%) (23). Response was measured with PET/CT in the former study and with CT and RECIST criteria in the latter, potentially explaining in part the differences in response. For ⁹⁰Y radioembolization in combination with chemotherapy, the lowest response rate was reported by Lim et al. (disease control rate of 59%) (30), and the highest response rates were reported by Van Hazel et al. and Sharma et al. (26,29) (both had a disease control rate of 100%). However, these last 2 studies are relatively small (11 and 20 patients, respectively). The largest study is by Stubbs et al. (36), with 100 included patients and measuring disease control rate at 3 mo. However, a rather large proportion of patients was lost to follow-up at that moment (20/100 patients). Thus, disease control was confirmed at the 3-mo follow-up in only 75 of 100 patients. Furthermore, response and progression in this study were respectively defined as decrease and increase of lesion size, without the specific percentages as are used in RECIST.

When survival data were compared for ⁹⁰Y radioembolization as monotherapy, the shortest overall survival was reported by Seidensticker et al. (8.3 mo) (9) and the longest overall survival was reported by Sato et al. (15.2 mo) (7). Comparison of these 2 studies is difficult. One used resin-based spheres and the other glass-based spheres of which the median administered activity was not reported. Seidensticker et al. included a relatively large percentage of patients with extrahepatic disease. Unfortunately, details on this cannot be extracted from the other paper. For ⁹⁰Y radioembolization in combination with chemotherapy, the shortest overall survival was reported by Hendlisz et al. (10.0 mo) (25) and the longest overall survival was reported by Van Hazel et al. and Kosmider et al. (both 29.4 mo) (26,32). Reasons for these differences are not readily apparent, other than the fact that these were all relatively small studies (21, 11, and 19 patients, respectively). The largest cohort in this group of articles is by Stubbs et al., with 100 patients, reporting an overall survival of 11 mo (36).

To enable comparison of the survival statistics of the individual studies, survival proportions at 12 mo after treatment were determined, either through measurements from a Kaplan–Meier plot or through calculations presuming exponential survival distributions. In our opinion, survival at 12 mo is clinically valuable information in the described patient populations. Because most median survivals were close to 12 mo, we also feel that any errors due to the assumption of exponential decay are small.

At present, ⁹⁰Y radioembolization is provided mainly to patients with progressive disease after first-line (fluoropyrimidine-based) and second-line (irinotecan-based) chemotherapy. Possible third-line systemic treatment consists of monoclonal antibodies (i.e., panitumumab or cetuximab). Van Cutsem et al. conducted a randomized controlled trial for panitumumab versus best supportive care and reported an overall survival for both groups of approximately 6 mo (3). In a randomized controlled trial on cetuximab versus best supportive care, Jonker et al. demonstrated that cetuximab significantly improved overall survival compared with best supportive care alone (6.1 vs. 4.6 mo, respectively) (4). Included patients in these studies were similar to those deemed eligible for ⁹⁰Y radioembolization, withstanding their potential extrahepatic disease load, since treatment with monoclonal antibodies is systemic. With that in mind, the results of this review show that, in the appropriately selected patient (with liver-dominant disease, acceptable disease burden, and preserved hepatic function and performance status), longer overall survival can be expected after ⁹⁰Y radioembolization than with current third-line systemic treatment options.