Key Questions

A Technical Expert Panel provided comments on the scope of the review. The following Key Questions and inclusion criteria reflect suggestions received and are in the final protocol. The final protocol was posted on the Effective Health Care website on June 23, 2022 ( https://effectivehealthcare.ahrq.gov/products/radiation-therapy-bone-metastases/protocol) and registered on PROSPERO (CRD42022340073).

Key Question 1. For symptomatic adults with bone metastases who will receive initial radiation for palliation, what is the comparative effectiveness and what are the comparative harms of dose-fractionation schemes and techniques for delivery (e.g., three-dimensional conformal radiation therapy, stereotactic body radiation)?

Key Question 2. For symptomatic adults with bone metastases who will receive re-irradiation for palliation, what is the comparative effectiveness and what are the comparative harms of dose-fractionation schemes and techniques for delivery (e.g., three-dimensional conformal radiation therapy, stereotactic body radiation)?

Key Question 3. What is the effectiveness and what are the harms of EBRT in the palliative treatment of bone metastases in symptomatic adults for the following:

KQ 3a. EBRT compared with another single MBD treatment modality (e.g., surgery, radionuclide therapy, bisphosphonate therapy, ablation kyphoplasty/vertebroplasty)

KQ 3b. EBRT combined with another treatment modality (e.g., surgery, radionuclide therapy, bisphosphonate therapy, ablation kyphoplasty/vertebroplasty) compared with EBRT alone?

KQ 3b. EBRT combined with another treatment modality (e.g., surgery, radionuclide therapy, bisphosphonate therapy, ablation kyphoplasty/ vertebroplasty) compared with the other (same) treatment modality alone?

Contextual Questions

Contextual Question 1. What are common barriers and facilitators to implementing guidance in radiation oncology, specifically related to palliative radiation for MBD?

Contextual Question 2. What strategies could be used to promote the use and implementation of guidance in radiation oncology, specifically related to palliative radiation for MBD?

Contextual Question 3. In symptomatic patients considered for palliative radiation therapy for MBD, to what extent does patient financial distress/hardship differ between EBRT dose/fraction schemes or technique?

Findings in Relation to the Decisional Dilemmas

Planning for palliative radiation of symptomatic metastatic bone disease (MBD) is complex and presents numerous decisional dilemmas. Using a best evidence approach, our synthesis focuses on the best quality evidence directly comparing dose/fractionation schemes and techniques for initial radiation and re-irradiation for palliation of MBD to inform decision-making around these dilemmas.

The key findings and strength of evidence (SOE) for Key Question 1 are summarized in Tables 5-7 of the full report, for Key Question 2 in Table 8 and for Key Question 3, Tables 9-12. All focus on primary outcomes and harms. SOE is further detailed in Appendix G. In addition to the Key Questions, three contextual questions are addressed in the previous section, with additional information found in Appendix C.

Comparison With Other Systematic Reviews

Our review is consistent with other systematic reviews of RCTs directly comparing EBRT single and multiple fraction schemes regarding both effectiveness and harms,174-176 namely that there are no differences between SF EBRT and MF EBRT in overall pain response (based on longest followup), pathological fracture or new MSCC and higher risk of re-irradiation with SF EBRT. Our review includes recently published RCTs, provides information based on length of followup, and includes additional detail on harms. In contrast to our findings, a recent Bayesian network meta-analysis177 reports that single 8 Gy EBRT was associated with better pain control, less risk of pathologic fracture and cord compression and reduced need for re-irradiation compared with multiple fractions of 20 Gy or 30 Gy; however, many estimates lacked precision. Differences in methodological approach to data synthesis in our review at least partially account for the contrast in findings. Our review uses direct head-to-head evidence within RCTs whereas network meta-analyses indirectly compare treatment options across RCTs when there is no direct evidence. Differences in included studies and categorization of pain response also likely contribute to the disparity in findings.

Recent reviews of SBRT for palliation in MBD have focused on spine MBD, relied on NRSI, particularly single arm studies, and provided indirect comparisons of the modalities. Our review provides the most up to date synthesis of RCT evidence comparing SBRT with conventional EBRT and importantly includes one RCT in patients without spine MBD. Our finding that SBRT was associated with slightly higher likelihood of overall pain improvement versus conventional EBRT is consistent with one recent review that included NRSI.178 Our findings of low frequency of Grade 3 and 4 toxicities and serious harms for all radiation therapy modalities is consistent with other systematic reviews, however studies may have been underpowered to detect rare outcomes.

Considerations for Clinical Practice and Health Policy

Our review found that, in patients with uncomplicated MBD, the likelihood of achieving overall pain response for SF EBRT and MF EBRT is probably similar, particularly after 4 weeks, although re-irradiation was more common with SF EBRT than with MF EBRT. There may be no differences in other primary outcomes or in adverse events including pathologic fracture and new spinal cord compression, or serious toxicities between SF and MF EBRT, however many harms were uncommon and evidence for many was insufficient.

Substantial variation in the delivery of palliative radiation therapy for MBD has been noted by many. Guidelines and quality measures have generally discouraged the use of multiple fractions in favor of few fractions or SF EBRT. Reasons cited for variability in use of SF EBRT over MF EBRT have included lack of consensus regarding optimal dose and fractions for either initial radiotherapy or retreatment and lack of clarity regarding which patients may benefit most for various dose/fractionation schemes. Variations in implementing recommendations for SF EBRT may also be due to differences in MBD characteristics as well as patient circumstances, characteristics and prognosis.

Our findings generally support guidelines and initiatives encouraging use of single or a limited number of fractions and lower total dose for palliative radiation in patients with uncomplicated MBD based on similarities in likelihoods for achieving overall pain response, noting that there is an association between SF EBRT and re-irradiation. Our findings provide some confirmation that use of lower single fraction doses is less effective for pain improvement in patients with mixed MBD, but evidence related to spine MBD or more complicated pathology was not identified. Our findings comparing lower total dose with higher total dose multiple fraction schemes suggest that the likelihood of overall pain response is probably similar in mixed MBD and in MSCC for the ranges of doses and fractions studied. This supports consideration of fewer fractions and lower total doses when multiple fractions are used. Of note, however, certain studies incorporated estimation of patient prognosis into inclusion criteria or ultimately enrolled patients with poor overall survival, and thus consideration of prognosis likely must be part of clinical decision-making. Although information on harms, particularly radiation-induced spinal cord pathology and serious adverse events was limited across all comparisons, consideration of potential harms is also important for clinical decision making.

There is variability in the use of more advanced techniques such as SBRT and lack of consensus regarding such use. Our review provides a synthesis across recently published RCTs of SBRT. We found that SBRT (usually 1 or 2 fractions) was associated with a slightly higher likelihood of overall pain response versus conventional, multiple fraction EBRT (3 Gy/10 fractions or 4 Gy/5 fractions). Analysis included one small RCT in patients with nonspine MBD (49% from lung cancer) and two small RCTs in patients with spine MBD. The applicability of these findings across a broader range of patients is unclear. The findings may facilitate initial discussion of how SBRT may benefit patients with MBD, although many gaps in evidence remain. In addition, access to treatments is important to consider. Conventional EBRT (2D or 3D) is likely available within 25 miles or less for most patients in the United States, however smaller communities outside of metropolitan areas and rural communities may not have access to newer technologies such as SBRT. Anecdotally, some have questioned the role that reimbursement may play in use of new technologies.

Evidence comparing dose/fraction schemes for re-irradiation is sparse. Low evidence from one large RCT supports decisions for SF versus MF EBRT for re-irradiation; our review found no differences in overall pain response or harms between SF and MF EBRT. Evidence from a retrospective NRSI also suggests no difference between SF versus MF SBRT for pain improvement, however recommendations should be made cautiously due to limited information and confidence on harms. Similarly, evidence for Key Question 3 (EBRT alone or in combination with other therapies), which was primarily comprised of NRSI, was generally insufficient for most comparators and outcomes, making evidence-based decisions and recommendation formulation challenging.

It was not possible to capture the nuances of clinical decision-making related to individual patient circumstances, prognosis, tumor location and features and various clinical factors that might inform the need for a specific dose or number of fractions or need for re-irradiation in a review such as this. While evidence from our report will support decision making and formulation of clinical recommendations, individual patient circumstances and preferences for palliative care must also be considered. As noted in information presented for the contextual questions, while aspects of patient financial distress and burden are important to consider, it is currently unclear how to measure, evaluate or consider these in the context of palliative radiotherapy for MBD for either clinical decision making or policy.

Clinical guidelines are intended to facilitate evidence-based decision making and help decrease practice variability but need to be implemented to be effective. Information from contextual questions in our report suggests that online clinical pathways and education-based interventions, particularly when coupled with use of peer-review or audit, may be most effective in promoting guideline implementation. To the extent that reimbursement may play a role in uptake of clinical guidelines, novel payment care models/incentivized quality metrics provide an intriguing area of research.

Applicability

Patient characteristics, primary tumor histology and treatment regimens represented in included studies are similar to those commonly encountered in clinical situations and therefore, many of our findings are likely applicable to typical clinical practice. Given the range of primary tumors, lesion locations and characteristics represented in included studies, we suspect that the study populations may not differ substantially from those encountered in typical radiation oncology settings. The heterogeneity across these factors is reflected in the proportions of males and females in studies as some primary tumors are specific to men (e.g., prostate) or more common in women (e.g., breast cancer); again, this is likely consistent with typical practice. Race and ethnicity were rarely reported in studies. In the three studies reporting race, most participants were White (76% to 81%); applicability of our finding to other racial or ethnic groups is unclear. In most studies, patients likely used several methods for pain control as well as adjunctive therapies; this, too, is likely consistent with clinical practice. Studies in this review most usually employed 8 Gy for SF EBRT, which is consistent with usual clinical practice. Regimens for MF EBRT varied across studies with the most common being 3 Gy x10 fractions or 4 Gy x5 fractions. This is consistent with usual clinical practice. Of note, older studies did not specify whether EBRT was two-dimensional conformal radiation therapy (2DEBRT) or three-dimensional conformal radiation therapy (3DEBRT), which is now likely most commonly used. In some circumstances, however, 2D is still used clinically.

Criteria for moving forward with re-irradiation were not explicitly described in most studies; this is consistent with usual practice. Unfortunately, evidence comparing dose/fraction approaches is sparse for re-irradiation so applicability of our findings for re-irradiation is unclear.

Clinically, a range of patients including those with complicated MBD and others with uncomplicated MBD is likely. Most included studies were in populations of mixed spine/nonspine lesions with uncomplicated MBD. Four RCTs in patients with spine MBD specifically enrolled patients with MSCC consistent with a definition of complicated MBD. Our findings may not be entirely applicable to patients with complicated MBD as optimal single or multiple fraction regimens for complicated MBD remain unclear.

Given growing evidence for improved oncologic outcomes with metastasis-directed therapy for oligometastatic disease, some SBRT studies included patients with limited numbers of metastatic sites and focused more on local recurrence and overall survival; palliative intent was not always clear. Most SBRT studies were in patients with spine MBD. For the comparison of SBRT versus conventional MF EBRT, we included two studies of SBRT focused on MSCC, one in patients with mixed MBD and one in patients with nonspine MBD.9 In the latter RCT, the most common primary tumor site was lung (49%); adenocarcinoma was the most common histology (63%). The extent to which the findings from this one trial may be applicable to a broader scope of patients with symptomatic nonspine MBD is unknown.

Research Recommendations

Gaps in the existing evidence on radiation for palliation in MBD are many. Most RCTs enrolled patients with uncomplicated MBD, thus high-quality comparative studies, preferably RCTs, in patients with complicated MBD would add to the evidence base. Furthermore, evidence for the use of radiation in the prophylactic setting for asymptomatic MBD to reduce risk of skeletal complications is needed. While RCT evidence comparing SBRT with conventional EBRT continues to emerge, studies specifically focused on palliative treatment of MBD are needed for spine and nonspine applications and in population with complicated and uncomplicated MBD. In addition, rigorous studies directly comparing dose/fraction schemes for SBRT are necessary. Such studies need to be sufficiently powered explicitly to directly evaluate the impact of dose/fraction schemes on harms, particularly pathologic fractures, to go beyond the information available from studies looking at predictive factors for fracture. Additional high-quality evidence comparing dose/fraction schemes for re-irradiation are important to verifying current findings for EBRT and SBRT. There is a need for high-quality studies that evaluate the impact of EBRT and another palliative therapies alone and in combination to better clarify joint benefits and harms. To facilitate comparisons across these studies, standardized definitions for important outcomes such as pain response, pain flare, and others are needed and outlined in the International Consensus on Palliative Radiotherapy Endpoints. Additionally, use of consistent scoring and reporting methods for outcomes related to quality of life is needed. Verification of outcomes such as pathologic fractures and cord compression (e.g., via imaging) may be helpful to assure consistency across studies. Describing criteria or rationale for re-irradiation and documenting patient response to re-irradiation separate from initial radiation may refine understanding regarding the discrepancy in re-irradiation between single and multiple fraction regimens. Many studies, particularly NRSI, provide information based on the number of treated sites or vertebral segments, but do not appropriately adjust for correlated data. Although some NSRI used methods to account for competing risks for outcomes like survival, they did not consistently adjust for potential confounding factors using multivariate analysis. Reporting data based on the number of patients is preferable. Better understanding of possible differential effectiveness or harms based on specific patient characteristics (e.g., age, sex), MBD characteristics (complicated, uncomplicated, type of lesion, number of lesions) or other factors which may help identify which patients may benefit most from particular treatment regimens are important. RCTs that include a priori plans for subgroup analyses, including tests for interaction, that are sufficiently powered are needed to effectively evaluate this. Understanding of the extent to which social determinants of health may impact the delivery, effectiveness and harms of palliative radiation therapy will require modification of study designs and analyses beyond reporting limited to patient characteristics (e.g., age, sex) in studies comparing treatment options. Incorporation and analysis of elements of socioeconomic status such as education, employment status, income, insurance status, family and social support together with information on race and ethnicity in studies may provide initial insights regarding social determinants of health, patient distress and disease burden.

This review does not provide cost information.

Strengths and Limitations

Our review has some notable strengths. Our synthesis focuses on the best quality comparative effectiveness evidence directly comparing dose/fractionation schemes for initial radiation and re-irradiation for palliation of MBD. This review also provides updated information comparing SBRT with conventional EBRT and describes evidence gaps for both effectiveness and comparative effectiveness of radiation therapy for MBD, which may stimulate additional research.

Our review has some limitations. We were only able to conduct limited analyses for publication bias due to small numbers of RCTs for many analyses. Our review of study bibliographies and clinical trial registries did not reveal unpublished studies meeting our inclusion criteria that would suggest missing publications. We did not include non-English language publications, however, title/abstract review of such publications captured from our search and listed in bibliographies suggests that this number would be few, and they would not meet our inclusion criteria; we are unlikely to have missed studies that would have changed our conclusions.

Using a best evidence approach, we focused on RCTs where possible. When sufficient RCT evidence was available, comparative NRSIs that were designed to evaluate harms, and which controlled for confounding, were considered. We excluded nonrandomized studies focused on effectiveness outcomes in these instances. NRSI can be misleading due to the subjective nature of pain and the impact of nonspecific effects of patient expectations regarding treatment and attention received on patient reported outcomes. The potential for selection bias and uncontrolled confounding add to the weaknesses of NRSI. In addition, in populations such as those with MBD, confounding by indication and inclusion of additional therapies with radiation therapy are likely to occur in NRSIs, making specific conclusions regarding radiation therapy effectiveness and adverse events challenging. Information from included NRSIs on harms provided limited additional insight beyond what was available from the RCTs when they were available. For questions where RCT evidence was sparse or not identified, comparative NRSI were considered. Ideally such studies would have explored the need to control for prognostic factors such as age, sex, primary tumor histology, pain duration, baseline pain severity, and prior treatments for outcomes of interest for this review (e.g., overall pain response). Many NRSI provided limited information regarding adjustment methods, and many did not adjust for these prognostic factors and reported adjusted estimates for only selected outcomes. The pool of eligible patients and selection of patients from that pool was general not well described and attrition was frequently unclear, raising concerns regarding confounding by indication and selection bias. While comparative NRSIs provide some information in the absence of RCT evidence, limitations of these studies generally led to determination of insufficient evidence for many outcomes and results should be interpreted cautiously. Inclusion of studies that directly compare interventions is most consistent with best evidence given the comparative intent of this review. Thus, single arm studies, including case series and pre-post studies, were excluded.

Limitations in the evidence base are reflected in the limitations to the review. There is substantial heterogeneity in enrolled populations across studies related to primary tumor types as well as number of sites or lesions. Studies that enrolled patients with spine and nonspine MBD did not report results by MBD sites. Few studies characterized MBD lesions regarding type (e.g., osteolytic, osteoblastic, favorable, unfavorable). RCTs comparing SF EBRT and MF EBRT of populations with mixed spine and nonspine MBD enrolled patients with uncomplicated MBD, based on the suggested definition (no pathological fractures or spinal cord or cauda equina compression); thus, the effectiveness and harms of SF EBRT versus MF EBRT in patients with complicated MBD requires further research. Similarly, no high-quality comparative evidence was identified in patients with complicated MBD for comparisons of single fractions, multiple fractions or in populations having re-irradiation were identified.

Few studies reported pain response based on the visual analogue scale (VAS), numeric rating scale (NRS), or similar method to measure pain improvement. Most studies used various definitions of pain response (Results Appendix B, Table B-2). Typically pain response was a composite measure (e.g., use of a pain measure combined with frequency of analgesic use). Some studies used specific thresholds for pain to categorize response (e.g., 20% pain response), and some employed the International Consensus on Palliative Radiotherapy Endpoints. Heterogeneity in definitions has been noted in other systematic reviews. We focused on overall pain response; this was most consistently reported and encompassed complete and partial responses based on authors' descriptions. We may have misclassified pain response in some instances. We did not report composite outcomes that combined aspects of pain response with imaging, given our focus on alleviating pain for this review. Similarly, definitions of pain flare varied across studies.

Some outcomes may not be routinely or systematically assessed either in clinical studies or clinical practice. For example, fractures and cord compression may be asymptomatic and studies may not have routinely confirmed these via imaging. Similarly, outcomes such as pain flare and local recurrence were variably defined across studies and may or may not be symptomatic.

Evidence from methodologically rigorous studies comparing SBRT with conventional EBRT and comparing various SBRT dose/fraction schemes for palliation remains sparse, particularly in patients with MBD not involving the spine. Some included SBRT studies primarily included patients with oligometastases and appeared to focus on aspects of survival and provided limited information on primary outcomes of interest to this review (e.g., pain response). Studies of SBRT compared single or two fraction SBRT with multiple EBRT fractions. Studies comparing dose/fraction schemes for re-irradiation were limited even with the inclusion of comparative NRSI. This is also true for comparisons of EBRT alone or in combination with other palliative treatments for MBD.

There was insufficient information from included trials on differential effectiveness or harms based on patient characteristics, tumor characteristics or other factors for all comparisons and interventions. Studies were underpowered to effectively evaluate this. There was also insufficient information from included studies to assess factors that may provide insight into how social determinants of health may impact delivery, effectiveness, and harms of palliative radiation therapy for MBD. Included studies did not provide information on patient sociodemographic characteristics beyond age and sex. Evaluation of the impact of social determinants of health requires a different approach to study design and analysis than is usually used in clinical intervention studies. Understanding of a broader literature that examines factors beyond what have been traditionally described as socioeconomic status is needed. Consideration of socioeconomic status together with factors related to social determinants are important to understanding distress and hardship experienced by patients receiving palliative radiation therapy; however, there was no evidence specific to the populations and comparisons of interest identified for this review.

A range of followup times was reported across studies. Unfortunately, candidates for palliative radiation therapy for MBD have a limited remaining life span. Substantial loss due to death at longer followup times occurred across studies, creating challenges for drawing conclusions across various comparisons due to diminishing sample size.