Condition

CRC is the fourth most frequently diagnosed cancer and the second leading cause of cancer death in the United States.¹ It is a cancer that forms in the tissues of the colon and the rectum. Most colorectal cancers are adenocarcinomas, meaning that they are a cancer of the epithelium originating from glandular tissue. Adenocarcinomas develop from adenomas, which are noncancerous tumors in the epithelial tissue. Over time, adenomas can become cancerous. This progression from adenoma to adenocarcinoma occurs through a sequential process of accumulating genetic changes.² Although the most common type of CRC is adenocarcinoma, squamous carcinoma and adenosquamous carcinoma have been reported infrequently.³

An elevated risk of CRC has been associated with obesity, low physical activity, high dietary intake of refined sugars, low dietary intake of fiber, consumption of meat, and consumption of more than two alcoholic drinks per day.⁴ A reduction in risk has been linked to the intake of dietary calcium and diets high in fiber and potassium.^{5, 6}

Diagnosis and Treatment of Colorectal Cancer

The diagnosis of CRC requires pathologic review to characterize and stage the tumor.⁷ Approximately 39 percent of new cases are diagnosed in the localized state, (i.e., no metastases or spread to regional lymph nodes); 36 percent present with regional spread to lymph nodes; 20 percent present with distant, metastatic cancer; and 5 percent present with unstaged disease.⁸ The 5-year survival rate estimated by the National Cancer Institute Surveillance Epidemiology and End Results program (SEER) data analysis was found to be 74.1 percent for stage I, 64.5 percent for stage IIA, 51.6 percent for stage IIB, 32.3 percent for stage IIC, 74 percent for IIIA, 45 percent for IIIB, 33.4 percent for IIIC, and 6 percent for stage IV.⁹ Survival declines with increasing depth of tumor penetration, increasing tumor stage, and patient age. For the 20 percent of patients who are initially diagnosed with distant (i.e., metastatic) disease, the 5-year survival rate is 10 percent or less with treatment. Patients with untreated liver metastases have a 5-year survival rate of less than 3 percent.¹⁰ Survival differs by the extent of liver metastases.

Treatment Strategies

Several local hepatic therapies have been developed to treat patients with hepatic metastases of CRC. In the continuum of care, use of a local hepatic therapy may occur before or after the use of systemic chemotherapy, but it is administered most often in conjunction with systemic chemotherapy. Local hepatic therapies are divided into three groups: (1) ablation (destruction of tissue through procedures involving heating or cooling); (2) embolization (the selective blockage of blood vessels, often with agents that carry a drug to the occluded site); and (3) radiotherapy (directed radiation to destroy abnormal cells). Table A describes the local hepatic therapies included in this review.

Table A

Local nonsurgical therapies for CRC liver metastases reviewed in this report.

Guidelines from the National Comprehensive Cancer Network for metastatic CRC state that ablative therapy for the metastases can be considered when all measurable metastatic disease can in fact be treated.³⁰ However, the group provides no guidance on which ablative therapy is optimal or on the comparative benefits and harms of the various palliative treatments.³⁰ A perception of clinical equipoise and limited randomized controlled trial (RCT) data comparing local hepatic therapies^31,32 contribute to uncertainty regarding which techniques, either alone or in combination, may be preferable for certain patient groups.

Topic Refinement and Review Protocol

The topic for this report was nominated in a public process. With input from Key Informants, the Evidence-based Practice Center (EPC) drafted the initial KQs and, after approval from AHRQ, posted them to a public Web site for 4 weeks for comment. We modified the KQs and the PICOTS based on these comments and discussion with the Technical Expert Panel (TEP). The initial KQs and interventions were stratified by intent of treatment (palliative or curative). This stratification seemed clinically inappropriate and potentially confusing because some interventions could be applied to palliate symptoms and to eliminate (i.e., cure) the liver metastases. The final KQs are distinguished by the population receiving local hepatic therapy (i.e., liver-directed). To be consistent with clinical practice, we modified KQs 1 and 2 to include patients with minimal rather than no extrahepatic disease. In addition, we categorized the 12 interventions to apply to all KQs, we removed some interventions, and we added SBRT. Finally, we expanded the list of harms to be considered.

Data Sources and Selection

To ensure the applicability of the interventions and outcomes data to current clinical practice, MEDLINE^® and Embase^® were searched for randomized, nonrandomized comparative and observational studies that treated patients between January 1, 2000, and June 27, 2012. Date restrictions were selected to ensure applicability of the interventions. Prior to 2000, some interventions were in their infancy and based on current standards used outdated regimens.^53,54,55 Thermal therapies were not used significantly until the late 1990s, and major changes in proton beam and stereotactic therapy occurred during that same period.⁵⁶ Chemoembolization drugs and embolic mixtures have also changed a great deal in the last 10 years and are more standard now. For these reasons, which the TEP strongly supported, we excluded studies where patient treatment preceded 2000. The searches were also limited to the English language.⁵⁷ It was thought that the exclusion of non–English-language articles from this review would not have an impact on the conclusions. The gray literature was also searched, including in databases with regulatory information, clinical trial registries, abstracts and conference papers, grants, federally funded research, and manufacturing information.

Titles and abstracts were screened in duplicate for studies that looked at overall survival, adverse events, and quality of life among our populations of interest. To be excluded, a study needed to be independently excluded by two team members. In cases where there was disagreement, a second-level abstract screening was completed by two independent reviewers. A third reviewer was consulted when necessary. Full-text review was performed when it was unclear if the abstract met study selection criteria.

Data Extraction and Quality (Risk of Bias) Assessment

Data extraction was performed directly into tables created in DistillerSR, with elements defined in an accompanying data dictionary. All team members extracted a training set of five articles into evidence tables to ensure uniform extraction procedures and test the utility of the table design. All data extractions were performed in duplicate, with discrepancies identified and resolved by consensus. The full research team met regularly during the period of article extraction to discuss any issues related to the extraction process. Extracted data included patient and treatment characteristics, outcomes related to intervention effectiveness, and information on harms. Harms included specific negative effects, including the narrower definition of adverse effects. Data extraction forms used during this review are presented in the main report in Appendix C.

Where applicable, we followed the Methods Guide³⁹ in the assessment of risk of bias in individual studies. Our assessment of risk of bias in the included case-series intervention studies was based on a set of study characteristics proposed by Carey and Boden.⁵⁸ The Carey and Boden assessment tool does not conclude with an overall score of the individual study. We created thresholds for converting the Carey and Boden⁵⁸ risk assessment tool into AHRQ standard quality ratings (good, fair, and poor) to differentiate case-series studies of varied quality. These distinctions were used for differentiation within the group of case-series studies, but not for the overall body of evidence described below. The classification into these categories (i.e., good, fair, poor) is distinct for a specific study design. For a study to be ranked as good quality, each of the Carey and Boden⁵⁸ criteria must have been met. For a fair-quality rank, one criterion was not met, and a rank of poor quality was given to studies with more than one criterion not met. These quality ranking forms can be found in the main report in Appendix D.

Data Synthesis

Evidence tables were completed for all included studies, and data are presented in summary tables. Evidence is also presented in text organized by outcome and intervention. No direct comparisons are made. We considered whether formal data synthesis (e.g., meta-analysis) would be possible from the set of included studies. Because the literature was so heterogeneous in terms of the populations (e.g., prior treatments, reason for unresectability, and number and size of lesions) and interventions (e.g., drugs and dose) studied, we concluded that pooling data would be inappropriate for this review. Thus, all data synthesis is based on qualitative summaries and analyses.

KQs 1 and 2

KQs 1 and 2 focus on the comparative effectiveness (KQ1) and harms (KQ2) of the patient population that was ineligible for systematic therapy and had no or only minimal evidence of extrahepatic disease. The evidence base comprised 23 case series and 931 patients. No comparative study met inclusion criteria for this review.

Key Points

• The evidence is insufficient to draw conclusions about overall survival, quality of life, or adverse events (Table D). Due to the absence of comparative data, we are limited in drawing conclusions regarding the efficacy and effectiveness of these interventions. Risk of bias is a primary concern in observational studies. Intended effects are likely to be biased by preferential prescribing of the intervention based on the patients’ prognosis.
• All studies were case series. Carey and Boden quality rankings were converted into AHRQ “good,” “fair,” and “poor” ratings. Eleven studies were rated as good quality,^{64, 66,67,69,71,73–75,80,88,90} nine studies as fair quality,^{61,63,76,81,82,84,86,87,89} and three studies as poor quality.^65,69,72
• The assessment of applicability of the study findings to clinical practice is limited by the poor characterization of the patient populations (e.g., number and size of metastases, performance status) and variations in the delivery of the interventions (e.g., surgical approach, dose and drugs delivered).

Table D

Strength of evidence for KQ1 and KQ2.

KQs 3 and 4

KQs 3 and 4 focus on the comparative effectiveness (KQ3) and harms (KQ4) of the various local hepatic therapies in patients who are received local hepatic therapy as an adjunct to systemic therapy for unresectable CRC metastases to the liver and who had no evidence of extrahepatic disease.

The body of evidence (seven studies) comprises case series with the exception of a single RCT⁸¹ that was included as a single-arm study. Two-hundred ninety-six patients were included from these seven studies. No comparative studies were available that met inclusion criteria.

• No conclusions on overall survival, quality of life, length of stay, time to recurrence, local recurrence, or adverse events can be drawn from the body of evidence comparing local hepatic therapies for unresectable CRC metastases to the liver (Table E).
• The literature base for this review is comprised of case series and one RCT85 that was abstracted as a case-series study due to a nonrelevant comparator. Four studies were ranked as good quality^62,70,78,85 and three were ranked as fair quality.^77,79,83
• The assessment of applicability of the study findings to clinical practice is limited by the poor characterization of the patient populations (e.g., number and size of metastases, performance status) and variations in the delivery of the interventions (e.g., surgical approach, dose and drugs delivered).

Table E

Strength of evidence for KQ3 and KQ4.

Key Findings and Strength of Evidence

No comparative studies met inclusion criteria for any of the four KQs about local hepatic therapy for the treatment of unresectable colorectal cancer (CRC) metastases to the liver. Thirty-one studies met our inclusion criteria and addressed local hepatic therapy for unresectable CRC metastases to the liver.

We assessed the strength of evidence for our primary health outcomes of overall survival and quality of life and for the intermediate outcomes of length of stay, local recurrence, and adverse events for all KQs. In addition, strength of evidence was assessed for the intermediate outcomes of time to progression (KQs 1 and 2) and time to recurrence (KQs 3 and 4). We judged the strength of evidence to be insufficient to draw conclusions for all outcomes. The body of evidence provided no comparative information about differences in effectiveness by type of intervention.

We are not aware of any published systematic reviews of the comparative effectiveness of local hepatic therapies for CRC metastases to the liver, as the literature base does not contain studies comparing one local hepatic therapy with another. Some systematic reviews of single local hepatic therapies have been published. Earlier reviews conforming to a high quality standard interpreted their findings similar to ours in the present review; that is, evidence was insufficient to permit conclusions.^32,91

This review sought evidence on the comparative benefits and harms of local hepatic therapies in two patient groups for CRC metastasis to the liver. Although we did not find this evidence the strength of the present review is in the identification of this important evidence gap. Distinct patient groups exist within the population receiving local hepatic therapies, yet data to analyze these differences are limited.

Applicability

It is challenging to comment on the applicability of findings from our CER because we found that the available evidence was insufficient for us to draw conclusions. The degree to which the data presented in this report are applicable to clinical practice hinges on the degree to which the populations in the included studies represent the patient populations receiving clinical care in diverse settings, as well as the availability of the interventions. We comment below on the relevance of included studies for population, intervention, comparator, outcomes, timing, and setting (PICOTS) elements. The PICOTS format provides a practical and useful structure to review applicability in a systematic manner and is employed in the subsections that follow.⁸⁸

The goal of any local hepatic therapy for unresectable CRC metastases to the liver is to prolong life by eliminating the metastases if possible or to palliate symptoms such as pain. This report has reviewed the literature on local hepatic therapies to achieve these goals. Due to the noncomparative nature of the literature base, both clinical and policymakers are limited in their ability to apply the published literature base to decisions on effectiveness and comparative effectiveness of these interventions. Survival estimates from individual studies of local hepatic therapies suggest that local hepatic therapies may provide some benefit in terms of survival and symptom relief for some patients, but without comparative data, it is not possible to choose the therapy that will produce the best outcomes for specific patients.

Population and Settings

The question of which subgroups of patients with CRC metastases to the liver may benefit from any particular local hepatic therapy compared with another remains unanswered. This uncertainty is reflected in the heterogeneity of the patient populations included in the published literature. Patient characteristics were often poorly characterized and not uniformly reported. Patients with varying degrees of resectability, extrahepatic disease, portal vein tumor thrombosis, and size and number of lesions are often grouped together and reported on as one group, even though it is uncertain whether these factors are likely to affect outcomes. Patient heterogeneity, combined with poor reporting of stratified or patient-level data, limited our ability to compare patient groups in any meaningful way. As a result, we are currently unable to determine which patients should be receiving which local hepatic therapies.

The setting in which treatment occurs is a major factor in the outcomes of local hepatic therapy. Expertise of both clinicians and centers varies. Based on the available clinical expertise and technology, the choice of a local hepatic therapy may be limited to one option in many centers. Local hepatic therapies, such as radioembolization⁹³ and hepatic arterial infusion,⁹⁴ often require high levels of training and familiarity with the procedure. Lack of experience may not only affect patient outcomes but also result in adverse effects; patients treated by less-experienced clinicians and centers will likely experience poorer outcomes.

Detailed analysis of differences in outcomes by center has important implications for the relevance of the findings in the literature. Unfortunately, these data were unavailable as part of our systematic review of the published literature.

Interventions

Even for a single local hepatic therapy, variations in how the procedure is performed may be substantial. For instance, variations may occur in the approach (open vs. percutaneous), the choice of chemotherapy drugs delivered, and the schedule of delivery of chemotherapy and radiation therapy. Given the lack of comparative data, the present review did not allow for a more rigorous and systematic comparison of the relative performance of local hepatic therapies stratified by these factors. How these factors may alter health outcomes remains unclear.

Additional heterogeneity exists for the context in which the intervention was delivered. Patients often receive more than one local hepatic therapy over time or more than one session of the same therapy. This often results in variations of prior therapy at study enrollment. The complex treatment history of each patient can further limit the conclusions that can be drawn about the benefits attributable to any one component of the treatment plan.

Comparators

All studies in this review are observational (including the arm of one RCT that was extracted as a case series); as such, they report on the experience of a particular center with one or more local hepatic therapies. Although case series can be useful for hypothesis generation, this approach cannot provide the comparative data the field needs for evaluating effectiveness. The applicability of any case series to another study group is very limited.

Outcomes

Little controversy exists regarding the most appropriate direct health outcomes to measure in a study of local hepatic therapies for CRC metastases to the liver. Overall survival is the ultimate outcome; it was reported in all of the studies included in this review. Quality of life is also a very important patient-centered outcome, but is not routinely reported in the literature in this review.

The importance of outcomes such as disease-free survival or local progression-free survival can be debated, but few experts would suggest that these outcomes replace the need for data on overall survival.

Studies of a comparative design are needed to measure accurately the differences in overall survival, quality of life, and harms that may be attributed to a local hepatic therapy.

Timing

The timing of followup assessment was appropriate given the natural history of unresectable CRC liver metastases and the primary outcome of overall survival. Median survival was reached in 21 of 24 studies. We judged this to be an appropriate length of assessment. In addition, two of the studies that did not reach median survival followed patients for up to 3 years to assess overall survival rates.

Gaps

This systematic review attempted to compare outcomes of local hepatic therapies for patients treated for unresectable CRC metastases to the liver. The review focused on two patient populations: those patients whose disease is refractory to systemic chemotherapy and patients who are receiving local hepatic therapy as an adjunct to systemic chemotherapy. Evidence on patient outcomes is limited, and the strength of evidence is insufficient for us to draw conclusions on effectiveness or harms for either patient population. As detailed above under applicability, there are specific evidence gaps that, if addressed, could enhance this literature base.

We identified four broad evidence gaps during this review. We present them organized by PICOTS framework. No gaps were identified for timing and setting.

• Populations: An objective of comparative effectiveness research is to understand the comparative effects for different population subgroups. To achieve this, we must fully delineate the population subgroups of interest. As detailed in the population and setting section above, these data are limited. Future studies must present data by subgroups of interest so that evidence can be interpreted by these variables. Based on published multivariate analyses, examples of patient or tumor characteristics found to be associated with improved overall survival include: ECOG status (0 vs. ≥1 and in another study 0 or 1 vs. ≥2), performance status (0 or 1 vs. ≥ 2), number of extrahepatic metastases sites (0 or 1 vs. ≥2), number of lines of previous chemotherapy (0–1 vs. ≥ 2), performance status (0 or 1 vs. ≥ 2), carcinoembryonic antigen response (Yes, No), and Response Evaluation Criteria in Solid Tumors (RECIST). These variables should be considered when designing future studies. Because there are so many variables being collated, clinical risk scores may be particularly beneficial as a summary measure.⁹⁵
• Intervention: There can be substantial variation in the role of local hepatic therapy in the overall treatment strategy for patient populations with unresectable CRC liver metastases reviewed in this report. A thorough delineation of prior and concurrent treatment is necessary to assess the incremental benefit of local hepatic therapy and the comparative outcomes of these therapies for the reviewed patient populations. All other therapies, systemic and local, should be taken into account when evaluating the effectiveness of the intervention under study, as these therapies may have an effect on patient survival. Previous resections and other local hepatic therapies were often not reported in the studies included in this review.
• Comparator: A major limitation of the current evidence review was that there was no comparative evidence at the time of publication of this report comparing the various liver-directed therapies with one another.
• Outcomes: Outcomes of interest to patients and their physicians include survival, quality of life, and adverse effects such as radiation-induced liver disease, liver failure, and local recurrence (i.e., treatment failure). Evidence comparing these outcomes of local hepatic therapies in the populations of interest for the review are needed. For survival and other time-to-event outcomes, it is essential for authors to report the time point from which the event was measured (e.g., time from liver-directed therapy, time from CRC diagnosis, time from diagnosis of metastases).
  Collection and reporting of quality-of-life data (e.g., pain) using standard measurement tools was inconsistently reported in the literature included in this review. These data are particularly important for the population of patients in which palliation of symptoms, rather than cure, is the intent of therapy.

Study Designs To Address These Gaps

RCTs are the gold standard of clinical evaluation, and there is an absence of randomized controlled clinical trial evidence on the use of local hepatic therapies for the included indications. Because we were unable to find comparative studies to answer any of our KQs, we conducted additional discussions with members of our Technical Expert Panel (TEP) to elicit ideas that could address the gaps in the literature. TEP members identified common barriers to conducting RCTs that would answer our KQs, including limited sources of research funding to support RCTs, reluctance of physicians to randomize patients, and reluctance of patients to be randomized.

In addition to the resistance to randomize, consensus around the most compelling hypothesis for a comparative RCT is lacking. Clinical investigators have competing hypotheses of which treatment is best suited for which patients, and these hypotheses are often based on their own institution’s experience. TEP members agreed that certain broad categories of patients with CRC metastasis to the liver, such as the populations included in this review, may well benefit from local hepatic therapies, but they also recognized that the published literature did not permit analysis of patient subgroups to identify characteristics more favorable to one local hepatic therapy over another. RCTs with well-documented patient and treatment characteristics could address the lack of comparative evidence. Lack of funding sources will continue to be an issue under the current regulatory structure. Under this system, the FDA does not require the same level of evidence for device approval as it does for drug approval. Because device companies can obtain approval without data from RCTs, they have very little incentive to provide funding.⁹²

Regardless of the study design, we suggest that studies aiming to address the effectiveness or comparative effectiveness of local hepatic therapies take care to address potential confounders and effect measure modification that could obscure the results. This is particularly important for patient characteristics such as size and number of metastases and performance status, which could serve as both modifiers of the effectiveness and factors that are considered when choosing the best local hepatic therapy.

Although RCTs may not be possible for all comparisons in all centers, multivariate analyses from existing case series can aid in identifying additional factors that should be documented and potentially controlled for in the comparative analysis of these data. Several factors were identified in multivariate analyses in the literature base of this report that impacted overall survival. The following factors should be collected and considered in future studies: number and size of lesions, number of extrahepatic metastases, previous treatment history (i.e., number of lines of previous chemotherapy), CEA, performance status, and tumor response. These analyses can enhance the design of future RCTs or observational studies.

Patient Registries

In the absence of consensus regarding the most salient comparative research question, observational data could be useful in driving the generation and prioritization of hypotheses for future research. One approach is the use of a registry to systematically collect observational data. According to the Agency for Healthcare Research and Quality publication on registries for evaluating patient outcomes, patient registries are often constructed to study patient outcomes, the natural history of disease, and disease management under various treatment scenarios.⁹⁷ Registries need to be created with a question in mind, which will then guide the identification of the target patient population, the interventions of interest (e.g., a local hepatic therapy), the outcomes of interest, the number of patients (to be adequately powered for future analysis), and the length of followup.

The KQs from this CER could serve as guide for designing one or more registries focused on this clinical area. The aim would be to establish a prospective registry that tracks the outcomes, quality of life, and adverse events in those who receive local nonsurgical treatment for unresectable metastatic CRC to the liver in order to identify the most effective local hepatic therapy strategies. The effectiveness of any one local hepatic therapy is expected to vary by patient subgroup. Provider experience with the local hepatic therapy is also an important factor in patient outcomes. We have identified a core set of variables or core dataset, defined as the information set needed to address the critical questions the registry is developed to answer. This is presented in Table F, organized by PICOTS.

Table F

Core dataset elements for local hepatic therapy registry by PICOTS.