3.1.1.1. Key Points

• Single-fraction (SF) EBRT is probably associated with a small decrease in the likelihood of achieving overall pain response compared with multiple-fraction (MF) EBRT up to 4 weeks post-treatment (9 RCTs, N=1280, 67.5% vs. 71.9%, risk ratio [RR] 0.93, 95% confidence interval [CI] 0.88 to 0.99, I²=0%) (SOE: moderate), but no clear differences between groups at >4 to 12 weeks (7 RCTs, N=2173, 69.4% vs. 68.3% RR 1.01, 95% CI 0.95 to 1.07, I²=0%) (strength of evidence [SOE]: moderate) or >12 weeks (2 RCTs, N=214, RR 0.87, 95% CI 0.68 to 1.12, I²=0%) (SOE: moderate) were seen. This was consistent across trials in populations with mixed spine/nonspine metastatic bone disease (MBD), however no difference was seen between SF EBRT and MF EBRT in analysis confined to patients with spinal metastases (2 RCTs, N=356, 56% vs. 58%, RR 0.96, 95% CI 0.78 to 1.16, I²=0%). Reported pain response likely included data for initial radiation therapy and re-irradiation. Results slightly favoring MF EBRT over SF EBRT at up to 4 weeks may in part reflect patients having received only initial radiation and/or a proportion of patients whose improvement occurred later than 4 weeks.
• There may be no difference between SF EBRT and MF EBRT in pain scores (0–10 scale) or in quality of life (various measures) across trials at any time frame (SOE: low for both).
• Evidence on overall function from two poor-quality trials was insufficient.
• There was probably no difference between SF EBRT and MF EBRT on maintenance or improvement in ambulation as an indicator of spinal cord compression relief up to 4 weeks post-treatment (SOE: moderate) and may be no difference at >4 to 12 weeks (SOE: low).
• Across RCTs, there may be no differences in pathologic fractures between SF EBRT and MF EBRT in patients with mixed spine/nonspine MBD or in one trial in patients with spine metastasis without cord compression (9 RCTs, N=4,086, 4.1% vs. 3.4%, RR 1.18, 95% CI 0.68 to 2.08, I²=53.1%); similarly, there were no differences between SF EBRT and MF EBRT in the likelihood of developing spinal cord compression (5 RCTs, N=2774, 2.9% vs. 2.0%, RR 1.41, 95% CI 0.87 to 2.30, I²=0%) (SOE: low for both outcomes).
• There may be no differences between SF EBRT and MF EBRT on the risk of developing the following adverse outcomes (SOE: low for all): pain flare, impairment of bladder or bowel function, Grade 3 and Grade 4 toxicities, and withdrawal due to adverse events.
• There was insufficient evidence for the following composite measures: skeletal events (re-irradiation and/or fracture) and skeletal adverse events (hospitalization for uncontrolled pain, symptomatic vertebral fracture, interventional procedure, salvage surgery, new or deteriorated neurologic symptoms, and spinal cord or cauda equina compression).
• Across RCTs, SF EBRT was associated with an over 2-fold higher likelihood of re-irradiation compared with MF EBRT (13 RCTs, N=5040, 19.8% vs. 8.2%, RR 2.44, 95% CI 1.79 to 3.66, I²=68.4%). This remained true across RCTs in populations with mixed spine/nonspine MBD (10 RCTs, N= 5040, RR 2.81, 95% CI 2.19 to 3.94, I²=37.5%), but analyses confined to populations with spine MBD (with or without metastatic spinal cord compression [MSCC]), showed no difference between single and multiple fraction (3 RCTs, N=1,031, 17.7% vs. 13.5%, RR 1.28, 95% CI 0.96 to 2.09, I²=35.4%).

3.1.1.2. Description of Included Studies

Twenty-two RCTs (in 29 publications)^{41,43–47,51,53–55,59–62,66,68,70,71,74,75,79,82,87,93,95,96,102–104} and four NRSIs^{111,118,121,122} compared SF EBRT versus MF EBRT for the palliative treatment of bone metastases (Appendix E, Tables E-1 and E-2).

Across the RCTs, sample sizes ranged from 40 to 1,171 (total N=6,623). The average study mean (13 trials)^{43,46,53,55,59–62,68,70,71,74,75,79,82,87,95,103,104} or median (7 trials)^{41,44,45,47,54,66,93,102} age of participants was 65 years across 20 RCTs (range 52 to 70 years); two RCTs did not report patient age.^51,96 The average proportion of males across 21 RCTs was 54 percent (range 30% to 83%); one trial did not report patient sex.⁸⁷ Few trials reported race or ethnicity, comorbidities, or social determinants of health, with one trial reporting nationality (53% Norwegian, 47% Swedish)⁷⁰ and another trial reporting race (76% White, 17% Black, 5% Hispanic/Latino, ≤1% Asian, or other).⁶² The primary tumor types included breast (range, 8% to 49%), lung (range, 1% to 33.5%), and prostate (5% to 80%). One RCT (in 3 publications) did not report primary tumor type.^62,68,71 In two trials, primary tumors were recorded as favorable (10% to 30%), intermediate (70%, only available in one trial),⁵¹ and/or unfavorable (20% to 70%).^51,74

Bone metastases were present at multiple sites in 13 to 86 percent of patients across seven trials; the proportion with metastases to other nonbone/visceral sites ranged from 20 to 41 percent across three trials. Across three trials, the metastatic bone lesions were lytic in 60 to 88 percent, sclerotic in 8 to 33 percent, and mixed in 3 to 7 percent. The site of bone metastases was mixed (i.e., spine and nonspine) in most RCTs (15 trials; spine, 29% to 89% and nonspine, 11% to 71%);^{41,43–47,51,53,54,59–62,66,68,70,71,74,75,79,82,87,93,95,96,102,104} spinal cord compression and pathologic fracture were exclusion criteria in most of the mixed trials. Five trials included bone metastases to the spine only (4 of MSCC and 1 which did not report spinal cord compression), and no trials included bone metastases to nonspine sites only. Pathological fractures were present at baseline in 0 to 15 percent of patients across the 10 trials that reported this information. In the four trials specifically evaluating MSCC, 45 to 67 percent of participants were ambulatory and 9 to 26 percent of participants reported abnormal bladder function at baseline. Most trials did not describe bone metastases as either complicated or uncomplicated; however, many trials did report spinal cord or cauda equina compression, or pathologic fracture at study entry, the presence or absence of which has been used to define complicated and uncomplicated bone metastases.

The single fraction dose was 8 Gy in all but two trials, which used 10 Gy.^47,60,102 The most common multiple fraction doses were 30 Gy (3 Gy x 10) and 20 Gy (4 Gy x 5) over 1 to 2 weeks across 18 trials;^{41,44,45,47,51,53–55,59,61,62,66,68,70,71,79,82,87,93,95,96,102} one trial used 16 Gy (8 Gy x 2) over 1 week,⁷⁴ one trial used 22.5 Gy (4.5 Gy x 5) over 1 week,⁶⁰ one trial used 24 Gy (4 Gy x 6) over an undisclosed time period,^46,75,104 and two used 40 Gy (2 Gy x 20) over 4 weeks.^43,51 Most trials did not clearly report the specific type of EBRT employed but it was most likely two-dimensional (2D) or three-dimensional conformal radiation therapy (3DCRT) as these are the most commonly available techniques. Common concomitant treatments included analgesics (16% to 70%), opioids (35% to 81.2%, one outlier trial at 2.8%), and steroids (19% to 100%). Previous treatments included systemic therapy (37% to 54%, includes primarily chemotherapy and hormone therapy), surgery (2% to 7%), and analgesics (16% to 70%). Most trials excluded patients who had chemotherapy or recent changes in systemic therapy, prior radiation therapy (RT) to the treatment site, prior vertebral fracture, and poor prognosis (generally life expectancy under 6 weeks). Followup periods ranged from 1 to 64.4 months.

One trial was conducted in the United States,^62,68,71 13 in Europe,^{41,45–47,55,59,60,66,70,74,75,82,87,93,95,102,104} three in Australia and New Zealand,^41,66,93 three in Egypt,^43,51,54 two in India,^44,79 one in Iran,⁵³ and two did not report the countries in which they were conducted.^61,96 Most were single center trials and the most common source of funding across the trials was government, followed by university and undisclosed funding.

Fifteen trials were fair quality^{41,43–47,53–55,59,66,70,74,75,87,93,95,102,104} and seven were poor quality.^{51,60–62,68,71,79,82,96} Common limitations included inability to blind patients and providers, lack of assessor blinding, unclear randomization and allocation concealment methods and high attrition (Appendix F, Table F-1). In many cases, the high attrition was due to high mortality (7.1% to 92.3%, longer followups generally above 20%), which is to be expected in this patient population.

Given the number of RCTs that compared SF EBRT versus MF EBRT, the four eligible NRSIs were included for evaluation of harms only (as specified in the Methods) and are described in Appendix B.

3.1.1.3. Detailed Synthesis

3.1.1.3.1. Pain

Definitions of pain responses varied across trials (Appendix B, Table B-2). We focused on overall response below, which encompasses complete and partial response in most trials (Appendix E, Table E-1).

3.1.1.3.1.1. Overall Pain Response

Fourteen RCTs^{41,43–46,53,54,59,60,62,74,87,93,96} comparing SF EBRT with MF EBRT contributed data to meta-analyses of overall pain response. Three RCTs were in patients with spinal metastasis^44,74,93 and 11 included patients with spine or nonspine MBD.^{41,43,45,46,53,54,59,60,62,87,96}

SF EBRT was associated with a small decrease in the likelihood of achieving overall pain response compared with MF EBRT up to 4 weeks post-treatment (9 RCTs, N=1280, 67.5% vs. 71.9%, RR 0.93, 95% CI 0.88 to 0.99, I²=0%)^{43–45,53,59,60,74,87,96} (Figure 3). Exclusion of two poor-quality trials^60,96 did not change the effect estimates (7 RCTs, N=980, 63.3% vs. 67.8%, RR 0.93, 95% CI 0.86 to 1.01, I²=0%),^{43–45,53,59,74,87} with the proportions of patients with overall response slightly lower in each group. The small overall decrease in pain response at 4 weeks was consistent across trials in populations with mixed spine/nonspine metastases (7 RCTs, N=924, 71.9% vs. 77.3%, RR 0.93, 95% CI 0.87 to 0.99, I²⁼0%),^{43,45,53,59,60,87,96} however no difference was seen between SF EBRT and MF EBRT in analysis confined to patients with spinal metastases across two fair-quality trials (2 RCTs, N=356, 56% vs. 58%, RR 0.96, 95% CI 0.78 to 1.16, I²=0%)^44,74 that included one trial in patients with MSCC.⁷⁴ We found no clear difference in the likelihoods of achieving overall pain response between SF EBRT and MF EBRT from 4 to 12 weeks (7 RCTs, N=2173, 69.4% vs. 68.3%, RR 1.01, 95% CI 0.95 to 1.07, I²=0%)^{43,45,46,54,59,62,96} or at >12 weeks (2 RCTs, N=214, RR 0.87, 95% CI 0.68 to 1.12, I²=0%);^43,45 all trials were in patients with mixed spine/nonspine MBD. No difference was seen across two trials where followup time was not reported or unclear (2 RCTs, N= 953, 71.1% vs. 73.1%, RR 0.98, 95% CI 0.82 to 1.09, I²=40%).^41,93 In one trial, 89 percent of patients had spinal metastasis, while the other trial was conducted in a mixed population (Figure 3). Exclusion of poor-quality trials for >4-week to 12-week results did not influence estimates or heterogeneity at the later time frames.

Across trials, results for pain response at all timeframes likely combined response data for initial radiation therapy and re-irradiation. Results slightly favoring MF EBRT over SF EBRT at up to 4 weeks may in part reflect patients having received only initial radiation and/or a proportion of patients whose improvement occurred later than 4 weeks.

There was no difference in overall pain response between SF EBRT and MF EBRT in analysis based on longest followup across trials (14 RCTs, N=3837, 69.4% vs. 70.0%, RR 0.99, 95% CI 0.94 to 1.03, I²=0%)^{41,43–46,53,54,59,60,62,74,87,93,96} or when trials of mixed spine/nonspine MBD (11 RCTs, N=3209, 72.2% vs. 72.0%, RR 0.99, 95% CI 0.95 to 1.03, I²=0%)^{41,43,45,46,53,54,59,60,62,87,96} were considered separately from those in patients with spine metastases (3 RCTs, N=628, 54.9% vs. 59.5%, RR 0.92, 95% CI 0.80 to 1.06, I²=0%)^44,74,93 (Appendix I, Figures I-1 and I-2). There was no indication of publication or small study bias based on funnel plot analysis and Egger’s test (p= 0.405) (Appendix I, Figure I-3).

Figure 3

Single versus multiple fraction EBRT: Overall pain response by timeframe. BPTWG = Bone Pain Trial Working Group; CI = confidence interval; EBRT = external beam radiation therapy; MBD = metastatic bone disease; MF_ctrl = multiple fraction is the control; (more...)

One fair-quality prospective NRSI (N=968)¹¹¹ found no difference in the probability of achieving overall pain response (adjusted odds ratio [OR] 0.86, 95% CI 0.63 to 1.19) for SF EBRT versus MF EBRT or when analyses were confined to patients with complicated MBD (N=335, adjusted OR 0.90, 95% CI 0.51 to 1.61). When patients were asked if bone pain interfered with their ability to care for themselves, SF EBRT was less likely than MF EBRT to improve this ability (adjusted OR 0.62, 95% CI 0.42 to 0.92); this was also true in analyses confined to patients with complicated MBD (adjusted OR 0.49, 95% CI 0.23 to 0.98).

3.1.1.3.1.2. Complete Pain Response and Pain Scores

Across studies for which complete response was reported or could be inferred, there were no differences between SF EBRT and MF EBRT based on data from last followup (14 RCTs, N= 3821, 34.6% vs. 23.1%, RR 1.01, 95% CI 0.92 to 1.10, I²=0%),^{41,43–46,53,54,59,60,62,74,87,93,96} at any time frame, or by population (mixed spine/nonspine MBD, spine only) (Appendix I, Figures I-4 to I-6).

Six trials^{44,46,47,66,70,79} reported pain based on VAS, numerical rating scale (NRS) or European Organisation for Research and Treatment of Cancer (EORTC) scores which were converted to a 0-10 scale for pooled analysis. There was no difference between SF EBRT and MF EBRT for pain posttreatment up to 4 weeks (4 RCTs, N=1854, mean difference [MD] 0.29, 95% CI −0.03 to 0.65, I²=54%);^44,46,66,70 the estimated difference was below the threshold for a small effect of 0.5 (Appendix I, Figure I-7). Sensitivity analyses using VAS data for one trial⁴⁷ and NRS data for another⁷⁰ reduced the heterogeneity and slightly increased the effect size, but the estimate remained below the threshold for a small effect (4 RCTs, N= 1854, MD 0.35, 95% CI −0.17 to 0.56, I²=0%). No differences between SF EBRT and MF EBRT were seen at >4 to 12 weeks (5 RCTs, N=1837, MD 0.03, 95% CI −0.42 to 0.41, I²=66%) or >12 weeks (3 RCTs, N= 1395, MD −0.07, 95% CI −0.46 to 0.29, I²=0%)^{46,47,66,70,79} (Appendix I, Figure I-7). Sensitivity analyses excluding the one poor-quality trial⁷⁹ from the latter two time frames did not impact effect size or reduce heterogeneity. MF EBRT was associated with small pain improvement in two fair-quality trials in patients with MSCC versus SF EBRT post-treatment up to 4 weeks (2 RCTs, N=390, MD 0.53, 95% CI 0.12 to 1.08, I²=0%, 0-10 scale),^44,66 however there were no differences at longer times in this patient population or in trials of populations with mixed spine/nonspine MBD^46,70,79 (Appendix I, Figure I-8). There was no difference between SF EBRT and MF EBRT in analyses based on longest followup time (Appendix I, Figure I-9).

3.1.1.3.3. Relief of Spinal Cord Compression/Neurological Outcomes

Four trials^51,66,74,102 which enrolled only patients with MSCC assessed ambulatory status using slightly different measures: a 4-point scale, consistent with the World Health Organization (WHO) performance status, based on the validated Medical Research Council muscle power criteria¹³⁸ (Grade 1 = ambulatory without aids and grade 5 of 5 muscle power in all muscle groups; Grade 2 = ambulatory with aids or grade 4 of 5 muscle power in any muscle group; Grade 3 = unable to walk with no worse than grade 2 of 5 power in all muscle groups or grade 2 of 5 power in any muscle group; and Grade 4 = absence [0/5 muscle power] or flicker [1/5 muscle power] of motor power in any muscle group);⁶⁶ Tomita’s functional motor grading system¹³⁹ (Grade 1= ambulatory without aids; Grade 2 = ambulatory with aids; Grade 3 = inability to walk; Grade 4 = paraplegic);⁷⁴ and an author-modified Tomita 3-point scale for mobility (Grade 1= ambulatory without aids; Grade 2 = ambulatory with aids; and 3 = bed-bound).¹⁰² The fourth trial only reported the outcome as “ambulatory” (authors mention evaluating motor function using the Medical Research Council muscle power criteria, scale 0 [complete paralysis] to 5 [normal power], but how this was applied is unclear).⁵¹

There were no differences between SF EBRT and MF EBRT in the proportion of patients who were ambulatory at any timepoint across all four RCTs (Table 2),^51,66,74,102 both when considering patients who either maintained or improved their ambulation status compared with baseline and only patients who improved their ambulation status.

Table 2

Relief of spinal cord compression: ambulatory status in patients with MSCC.

One trial reported a mobility score using the authors’ own modified Tomita mobility scale (1–3 scale; 1 = unaided, 2 = with walking aid, and 3 = bed-bound) and found no differences between SF EBRT versus MF EBRT in change scores from baseline to 5 weeks: mean change 0.06 (standard deviation [SD] 0.75) versus 0.3 (0.78), adjusted difference in change scores −0.28 (95% CI −0.6 to 0.03).¹⁰²

Sphincter, bladder, and bowel control were reported a variety of ways (i.e., improvement, normal, abnormal) across the four RCTs with no differences between fractionation schemes at any timepoint, with one exception (Results Appendix B, Table B-3): one fair-quality trial⁷⁴ found SF EBRT associated with a large increase in the likelihood of achieving good/normal sphincter control post-RT compared with MF EBRT (N=303, 5.9% vs. 1.3%, RR 4.41, 95% CI 0.97 to 20.1), however the estimate was imprecise. When considering only those patients with poor/abnormal sphincter control at baseline (i.e., likelihood of regaining control only as opposed to maintaining or regaining control), there was no statistical difference between groups though the rate with SF EBRT was higher (34.6% vs. 13.3%, RR 2.60, 95% CI 0.64 to 10.5). One trial reported a bladder score using the authors’ own scale for bladder function (1-3 scale; 1 = continent, 2 = incontinent, and 3 = catheterized) and found no differences between SF EBRT versus MF EBRT in change scores from baseline to 5 weeks: mean change 0.17 (SD 0.71) versus 0.22 (0.96), adjusted difference in change scores −0.05 (95% CI −0.45 to 35).¹⁰² In two trials, patients with good baseline bladder function developed poor function requiring an indwelling catheter: 0.8 percent (n=2/258, 1 poor-quality RCT⁵¹) and 4.6 percent (n=12/262, 1 fair-quality RCT⁷⁴). Results were not reported by treatment group.

One poor-quality trial reported no difference between SF EBRT and MF EBRT in the proportion of patients who had sensory deficit at baseline (19% [18/95] vs. 20.5% [39/190] of the total population, respectively) who recovered after treatment (27.8% [5/18] vs. 30.8% [12/39], RR 0.90, 95% CI 0.37 to 2.18).⁵¹

3.1.1.3.5. Secondary Outcomes

Thirteen RCTs comparing SF EBRT and MF EBRT reported rates of re-irradiation. SF EBRT was associated with an over 2-fold higher likelihood of re-irradiation compared with MF EBRT, consistent with a large effect (13 RCTs, N=5040, 19.8% vs. 8.2%, RR 2.44, 95% CI 1.79 to 3.66, I²=68.4%);^{41,43,45,46,59,61,62,66,70,79,87,93,102} substantial heterogeneity was noted (Figure 4). The effect size and statistical heterogeneity were slightly increased with the exclusion of three poor-quality trials^61,62,79 (10 RCTs, RR 2.55, 95% CI 1.73 to 4.21, I²=75%).^{41,43,45,46,59,66,70,87,93,102} The heterogeneity may have in part been due to variation in criteria for performing re-irradiation across studies. The funnel plot for this analysis showed some visual asymmetry, which raises the possibility of publication bias. It may be due to the substantial heterogeneity for the pooled estimate (I²=68%) given variability in decision making criteria for performing re-irradiation across trials. The Egger’s test was not significant (p=0.221) (Appendix I, Figure I-10).

Figure 4

Re-irradiation across fractionation schemes for EBRT. BPTWG = Bone Pain Trial Working Group; CI = confidence interval; Ctrl = control group; EBRT = external beam radiation therapy; Int = intervention group; MBD = metastatic bone disease; MF = multiple (more...)

Across RCTs in patients with mixed spine/nonspine MBD, SF EBRT was associated with an over 2-fold higher likelihood of re-irradiation compared with MF EBRT, consistent with a large effect (10 RCTs, RR 2.81, 95% CI 2.19 to 3.94, I²=37.5%),^{41,43,45,46,59,61,62,70,79,87} however analyses confined to studies in patients with spine MBD, found no difference between dose/fraction schemes (3 RCT, N=1,031, 17.7% vs. 13.5%, RR 1.28, 95% CI 0.96 to 2.09, I²=35.4%).^66,93,102 Two of the RCTs were in patients with MSCC.^66,102 In the RCT of patients without MSCC,⁹³ there was no difference between dose fractionation schemes on the likelihood of re-irradiation (N=272, 29.2% vs. 24.4%, RR 1.19, 95% CI 0.80 to 1.77). In contrast, one NRSI¹²¹ designed to evaluate clinically relevant adverse spinal events in patients with uncomplicated spinal metastases reported that re-irradiation was more common following SF EBRT versus MF EBRT (N=299, 18.2% vs. 6.0%, p=0.004).

One fair-quality prospective NRSI (N=968) found no difference in re-irradiation between SF EBRT and MF EBRT (17% versus 14%).¹¹¹

Data for other secondary outcomes (local control, medication use, need for additional intervention, and overall survival) can be found in Results Appendix B.

3.1.1.3.6. Harms and Adverse Events

Toxicity, adverse events, and harms were inconsistently reported across included studies.

3.1.1.3.6.1. Pathologic Fracture

There was no difference in pathologic fractures between SF EBRT and MF EBRT (9 RCTs, N=4086, 4.1% vs. 3.4%, RR 1.18, 95% CI 0.68 to 2.08, I²=53.1%)^{41,43,45,46,61,62,70,87,93} (Figure 5). Exclusion of two poor-quality trials^61,62 reduced the effect estimate but heterogeneity was similar (7 RCTs, N=3115, RR 1.02, 95% CI 0.50 to 2.07, I²=59%).^{41,43,45,46,70,87,93} One fair-quality trial in patients with spinal metastases who did not have cord compression at baseline reported new or progressive vertebral fractures;⁹³ no difference between radiation schemes was seen (N=272, 4.4% vs. 3.7%, RR 1.18 95% CI 0.37 to 3.78). All trials enrolled patients with various primary tumors.

Figure 5

Pathologic fractures across fractionation schemes for EBRT. BPTWG = Bone Pain Trial Working Group; CI = confidence interval; Ctrl = control group; EBRT = external beam radiation therapy; Int = intervention group; MBD = metastatic bone disease; MF = multiple (more...)

In contrast to the RCT findings, one fair-quality NRSI (N=299)¹²¹ designed to evaluate clinically relevant adverse spinal events in patients with uncomplicated spinal metastases reported greater odds of fracture with SF EBRT (13.6% vs. 3.0%, OR 3.73, 95% CI 1.61 to 8.63). It is unclear if this is an adjusted estimate. In the propensity-matched cohort, symptomatic vertebral fracture risk was also higher with SF EBRT versus MF EBRT (N=132, 13.6% vs 1.5%).¹²¹

See Results Appendix B, Table B-10 for pathological fracture outcomes.

3.1.1.3.6.2. Spinal Cord Compression

There was no difference in the development of spinal cord compression following treatment between SF EBRT and MF EBRT (5 RCTs, N=2774, 2.9% vs. 2.0%, RR 1.41, 95% CI 0.87 to 2.30, I²=0%).^{41,46,70,87,93} All trials were fair quality. All but one trial was in patients with mixed spine/nonspine MBD and had excluded patients with MSCC at recruitment (Figure 6). One trial was primarily in patients with spine metastases (89%), but only one patient had MCSS prior to radiation therapy⁹³ and found no difference in the likelihood of spinal cord compression by dose/fractionation scheme (N=272, 6.6% vs. 5.9%, RR 1.1, 95% CI 0.44 to 2.79) posttreatment. One additional poor-quality trial reported no spinal cord compression in either group.⁸²

Figure 6

New spinal cord compression across fractionation schemes for EBRT. BPTWG = Bone Pain Trial Working Group; CI = confidence interval; Ctrl = control group; EBRT = external beam radiation therapy; Int = intervention group; MBD = metastatic bone disease; (more...)

One fair-quality propensity score matched cohort NRSI¹²¹ designed to evaluate clinically relevant adverse spinal events in patients with uncomplicated spinal metastases reported that cord or cauda equina compression was more common with SF EBRT than with MF EBRT (N=132, 10.6% vs. 0%, p=0.002).

See Results Appendix B, Table B-10, for spinal cord compression outcomes.

3.1.1.3.6.3. Other Adverse Events

Pain Flare. One fair-quality RCT found no difference between SF EBRT and MF EBRT in the likelihood of experiencing a pain flare in patients with spine MBD (N=233, 10% vs. 4%, RR 2.0, 95% CI 0.91 to 5.81).⁹³ There was no clear difference between dose/fraction schemes across two small NRSIs using different definitions of pain flare. All effect estimates were imprecise. Risk of pain flare was similar with SF EBRT and MF EBRT in one NRSI (N=111, 38.6% vs. 39%, RR, 1.35, 95% CI 0.87 to 2.11). The other NRSI evaluated a subset of patients enrolled in the Canadian Bone Metastasis Trial (N=44)¹²² who agreed to complete a 14-day pain diary (MBD sites not reported). It used two different pain flare definitions. Results using both definitions suggest that pain flare may be more common with SF EBRT compared with MF EBRT, however, estimates were imprecise (Tannock definition, 43.5% vs. 23.8%, RR 1.83, 95% CI 0.75 to 4.47; Chow Definition, 56.5% vs. 23.8%, RR 2.37, 95% CI 1.02 to 5.53).

Skeletal Events and Spinal Adverse Events. There was no difference in skeletal-related events, defined as at least one instance of re-irradiation or pathologic fracture, between SF EBRT and MF EBRT in one poor-quality RCT (N=90, 28.8% vs. 13.3% RR 2.17, 95% CI 0.90 to 5.19).

One fair-quality NRSI¹²¹ designed to evaluate clinically relevant adverse spinal events in patients with uncomplicated spinal metastases reported that SF EBRT was associated with higher likelihood of any spinal adverse event (N=299, 27.3% vs. 14.2%, adjusted hazard ratio (HR) 2.78, 95% CI 1.51 to 5.15). Cumulative incidence of any spinal adverse event was consistently higher with SF EBRT: 4 weeks (6.8% vs. 3.5%), 12 weeks (16.9% vs. 6.4%) and 26 weeks (23.6% vs. 9.2%). SF EBRT was associated with higher rate of first spinal adverse event at 12 weeks (N=132, 22.5% vs. 7.7%, HR 3.2, 95% CI 1.3 to 7.5) based on propensity score matched analysis. Spinal adverse events included hospitalization for uncontrolled pain, symptomatic vertebral fracture, interventional procedure, salvage surgery, new or deteriorated neurologic symptoms, and spinal cord or cauda equina compression.

Various Adverse Events. There were no differences between SF and MF EBRT for new impairment of bladder or bowel function in one fair-quality RCT⁶⁶ in patients with MSCC. No differences were seen in bladder impairment prior to 8 weeks (N=638, 43.7% vs. 34.5%, adjusted OR 1.31, 95% CI 0.87 to 1.97) or at 8 weeks (N=317, 31.1% vs 20.5% [34/166], adjusted OR 1.78, 95% CI, 0.93 to 3.39) or in bowel impairment at any time (N=637, 64.4% vs. 63.4%, OR 1.05, 95% CI 0.76 to 1.45) or at 8 weeks (39.1% [59/151] vs. 36.7% [61/166], OR 1.10, 95% CI 0.70 to 1.74).

One fair-quality RCT (N=303)⁷⁴ in patients with MSCC reported no occurrences of radiation induced myelopathy for either dose/fraction scheme. Another fair-quality RCT reported that one patient who received SF EBRT experienced radiation enteritis due to retreatment and one patient in the MF EBRT group experienced a small bowel ileus. Two RCTs reported no adverse event-related study withdrawals.^44,96 One NRSI¹²¹ in patients with uncomplicated spine MBD reported that new or deteriorated neurologic symptoms were more common following SF EBRT versus MF EBRT in a matched propensity score cohort (N=132, 12.1% vs. 4.5%, p=0.10).

3.1.1.3.6.4. Toxicity

Toxicity type, severity, and frequency were variably reported across studies. Some studies reported toxicities by numbers of sites versus number of patients. We focused on Grades 3 and 4 toxicities here; detail on other grades is found in Results Appendix B, Table B-7.

In patients with mixed spine/nonspine MBD, Grade 4 toxicities were rare, ranging from 0 percent to 3 percent, with no differences between SF EBRT and MF EBRT across three RCTs,^54,60,62 however, this may in part be attributed to small sample size given the rare nature of these events. Similarly, Grade 3 toxicities were similar between SF SBRT and MF SBRT for any toxicity (<1% to 3% vs. <1% to 4%) across one fair- and one poor-quality RCT,^54,62 and for specific toxicities of nausea and vomiting (11% vs. 15%) and tiredness/lassitude (10% vs. 14%) in another poor-quality trial.⁶⁰ Two other trials reported that no grade 3 or 4 toxicities occurred.^59,82 There was no difference between SF and MF EBRT for “quite a bit or very much” nausea (39% vs. 40%) or vomiting (20% vs. 21%) in a subset of patients (N=124) from the Bone Trial Working Group RCT⁴¹ based on pain diaries up to 14 days post-treatment (Results Appendix B, Table B-7).

There was limited data comparing toxicities for SF EBRT versus MF EBRT in patients with spine metastasis. Retrospective analysis⁶⁸ of the Radiation Therapy Oncology Group 97-14 trial⁶² in patients with painful spine metastases reported low frequency of acute or late Grade 4 toxicities and no difference by dose/fraction scheme (N= 135, 0% vs. 1% for both acute and late). Results were similar for Grade 3 toxicities (N=124, acute <1% vs. 3%, late 2% vs. 0%). Sample sizes may have been inadequate to identify differences. One fair-quality RCT (N=686)⁶⁶ reported similar risk of Grade 3 or 4 toxicity (20.5% vs. 20.6%) and of death unrelated to treatment (0.9% vs. 1.5%). Small individual trials report no difference in Grade 3 acute gastrointestinal toxicities (N=59, 0% vs. 6%)⁴⁴ or late upper thigh pain (N=52, 0% vs. 0.5%).¹⁰² One of these RCTs¹⁰² reported somewhat lower risk of any Grade 2 or 3 toxicity with SF SBRT versus MF EBRT (N=100) 11.1% vs. 26.1%, RR 0.43, 95% CI 0.14 to 1.05); the estimate is imprecise.

3.1.2.1. Key Points

• LDSF (4 Gy) may be associated with slightly lower likelihood of overall pain response compared with HDSF (6 to 8 Gy) up to 4 weeks posttreatment (2 RCTs, N= 861, RR 0.80, 95%CI 0.58 to 1.02, I²= 76%), >4 to 12 weeks (2 RCTs, N=743, 74.3% vs. 83.3% RR 0.89, 95%CI 0.72 to 1.0, I²=63.9%) and 12 weeks (1 RCT N=180, 82.3% vs. 93.1%, RR 0.88, 95% CI 0.79 to 0.99) (SOE: low).
• There was insufficient evidence to evaluate the impact of different SF doses on function or quality of life from one cluster RCT (SOE: insufficient).
• No patients had pathologic fractures or spinal cord compression within the first 8 weeks of radiation and there may be no differences between LDSF and HDSF for either of these outcomes at >8 weeks in one trial (SOE: low).
• There may be no differences between LDSF (6 Gy) and HDSF (8 Gy) for skeletal events (pathologic fracture, re-irradiation, cord compression), adverse events (not specified) or adverse reactions (not specified) in a cluster RCT in which all patients received zoledronic acid, calcium, and vitamin D (SOE: low for all).
• Two trials found that re-irradiation was more common with LDSF (4 Gy) versus HDSF (8 Gy), however a third trial found no difference in re-irradiation risk for 4, 6, or 8 Gy single fractions.

3.1.2.2. Description of Included Studies

Four RCTs^64,67,69,72 compared different doses of SF schemes for conventional EBRT for the palliative treatment of bone metastases (Appendix E, Table E-1). We reported the lowest doses as the intervention (LDSF) and higher dose as the control (HDSF).

Across the RCTs, sample sizes ranged from 139 to 655 (total N=1391). The average study median (3 trials)^64,67,69 or mean (1 trial)⁷² age of participants was 61 years (range 57 to 64 years). The average study proportion of males across three RCTs was 48 percent (range 37% to 65%); one trial did not report patient sex.⁶⁴ None of the trials reported race, comorbidities, or social determinants of health. Primary tumor types included breast (range, 21% to 46%), lung (range, 19% to 35%), and prostate (range, 13% to 17%); no trial reported primary tumor histology in terms of favorable or unfavorable. All four RCTs included patients with bone metastases at mixed sites. Across three RCTs, spinal metastases accounted for 37 to 59 percent of lesions, and nonspine metastases accounted for 41 to 63 percent;^64,67,69 one RCT did not provide further details.⁷² Pathological fracture and MSCC were exclusion criteria in two trials,^64,67 and just pathological fracture in one⁶⁹ (the fourth trial⁷² did not report these characteristics). The presence or absence of these characteristics have been used to define complicated and uncomplicated bone metastases. No trial reported whether bone metastases were lytic or sclerotic or if concomitant nonbone/visceral metastases were present. One trial included patients with a single bone metastasis⁶⁴ and another included patients with ≤2 bone metastases;⁷² the remaining two trials did not report the number of metastases treated.

The lower dose in three RCTs was 4 Gy;^64,67,69 6 Gy was used in the fourth RCT.⁷² All patients in the latter trial received six 4 mg doses of zoledronic acid (infusion) in addition to daily doses of 500 mg calcium supplement and 400 IU of oral vitamin D during treatment. The higher dose was 8 Gy in 3 RCTs;^64,67,72 the remaining RCT contained two higher dose arms: 6 Gy and 8 Gy.⁶⁹ For purposes of meta analyses, data from these two higher dose arms were combined to form the high dose group. Most trials did not clearly report the specific type of EBRT employed but it was most likely 2D- or 3DCRT as these are most used.

All four RCTs reported baseline analgesic use with most patients taking opioids/narcotics (range, 46% to 64%); in three trials, 17 to 22 percent of patients were not on any analgesics.^64,67,69 Prior RT to the same site was an exclusion criterion in three trials.^64,69,72 One trial each excluded patients with previous surgery⁶⁹ and pretreatment bisphosphonate use⁷² while another trial⁶⁴ included patients with a history of chemotherapy (35%), hormone therapy (25%), and bisphosphonate use (34%) (unclear if concurrent or past treatments). Followup periods ranged from 12 to 156 weeks.

Three trials were conducted in Europe^67,69,72 and one was multinational without specifying details.⁶⁴ Two were single center trials^67,69 and two were multicenter.^64,72 One of the multicenter trials did not report the number of hospitals but used hospital site as the unit of randomization.⁷² The source of funding was reported in one trial as government⁶⁴ and was not reported in the other trials.

All trials were fair quality.^64,67,69,72 Three trials were unable to blind care providers, patients, or outcome assessors^64,67,69 one trial masked care providers and patients but was less clear in whether it blinded outcome assessors.⁷² Other common limitations included unclear randomization and allocation concealment methods (Appendix F, Table F-1). High levels of attrition were common across trials, particularly as followup time increased, due to high mortality rates in this patient population.

3.1.2.3. Detailed Synthesis

3.1.2.3.1. Pain

Definitions of pain responses varied across the four trials (Results Appendix B, Table B-2), as did measures for reporting pain. We focused on overall response below, which encompasses complete and partial response in most trials (Appendix E, Table E-1).

LDSF was associated with a slightly lower likelihood of achieving overall response compared with HDSF up to 4 weeks posttreatment (2 RCTs, N= 861, 64.2% vs. 76.8%, RR 0.80, 95% CI 0.58 to 1.02, I²= 76%),^64,69 which persisted >4 to 12 weeks (2 RCTs, N=743, 74.3% vs. 83.3%, RR 0.89, 95% CI 0.72 to 1.0, I²=64%)^64,69 and to >12 weeks (1 RCT, N=180, 82.3% vs. 93.1%, RR 0.88, 95% CI 0.79 to 0.99).⁶⁴ LDSF was 4 Gy for all trials. HDSF was 8 Gy in one trial. Two HDSF arms, 6 Gy and 8 Gy, were used in one trial; the arms were combined for this meta-analysis (Figure 7).⁶⁹ Analysis confined to the 8 Gy higher dose for this trial increased heterogeneity, accentuated the differences between LDSF and HDSF and decreased precision posttreatment up to 4 weeks (2 RCTs, N=753, 64.2% vs. 10.2%, RR 0.77, 95% CI 0.52 to 1.07 I2= 83%); use of the 8 Gy dose had little impact on estimates, while heterogeneity remained high >4 weeks to 12 weeks (2 RCTs, N= 635, RR 0.89, 95% CI 0.67 to 1.03, I²=72%). LSDF was also associated with slightly lower likelihood of overall pain response compared with HDSF in analyses based on longest followup time (2 RCTs, N=507, 68.5% vs. 81.2%, RR 0.85, 95% CI 0.71 to 0.97, I²= 33%).^64,69

Figure 7

Comparison of lower and higher dose single fractions of EBRT: Overall pain response by timeframe. CI = confidence interval; EBRT = external beam radiation therapy; Gy = Gray; HDSF = higher total dose single fraction EBRT (control); LDSF = lower total (more...)

In analyses of complete pain response as defined by authors, there was no difference between LDSF and HDSF post-treatment up to 4 weeks (3 RCTs, N= 1055, 27.4% vs. 28.6% RR 0.93, 95% CI 0.67 to 1.19, I²=30%) but LDSF was associated with slightly lower likelihood of complete pain response >4 weeks up to 12 weeks compared with HDSF across the same trials (3 RCTs, N=844, 36.8% vs. 42.3%, RR 0.82, 95% CI 0.68 to 0.98, I² = 0%)^64,67,69 (Appendix I, Figure I-11). There was no difference between SF doses at >12 weeks in one trial (1 RCT, N=180, 68.3% vs. 76.2% RR 0.90, 95% CI 0.74 to 1.08).⁶⁴ Definitions of complete response varied (Results Appendix B, Table B-2).

One trial (N=270) reported lower prevalence of pain improvement by ≥1 category (categories of no pain, mild, moderate, severe) with LDSF (4 Gy) compared to HDSF (8 Gy) at 4 weeks (44% vs. 69%, data not available for effect size calculation).⁶⁷ The authors also reported lower response rate in LDSF recipients versus HDSF recipients at this time frame (53% vs. 76%, p<0.01) based on actuarial analysis. The differences between lower and higher doses was less at 8 and 12 weeks (70% vs. 80% for both times as estimated from graphs). A cluster trial, which randomized treatment by hospital, reported higher mean VAS pain scores (0–10 scale) at 30 weeks for LDSF (6 Gy) compared with HDSF (8 Gy) for pain while supine (3.69 vs. 1.79, p=0.067), while seated (1.67 vs. 0.96, p=0.123) and while standing (2.34 vs. 1.27, p=0.006) suggesting small improvement in pain favoring HDSF; authors did not provide sufficient data to calculate effect sizes with confidence intervals. All patients in this trial received six, 4 mg doses of zoledronic acid (infusion) in addition to daily doses of 500 mg calcium supplement and 400 IU of oral vitamin D during this treatment.

3.1.3.1. Key Points

• There was probably no differences between total LDMF and HDMF schemes in overall pain response post-treatment up to 4 weeks (6 RCTs, N= 788, 64.1% vs. 67.0%, RR 0.96, 95% CI 0.87 to 1.06, I²=0%), from >4 to 12 weeks (3 RCTs, N=275, 79.6% vs. 80.4%, RR 1.02, 95% CI 0.89 to 1.12, I²=0%) (SOE: moderate), and maybe no difference at >12 weeks (2 RCTs, N=114, 78.6% vs. 72.4%, RR 1.10, 95% CI 0.86 to 1.38, I²=0%) (SOE: low), with mixed spine and nonspine MBD or in patients with MSCC.
• There was insufficient evidence on overall function reported in three poor-quality RCTs (SOE: insufficient).
• There were no differences between LDMF versus HDMF schemes for any outcomes related to relief of spinal cord compression in patients with MSCC including improvement on the following: ambulatory status (SOE: moderate), walking capacity (SOE: low), motor function (SOE: low), regain of sphincter control (SOE: low).
• There may be no differences in pathologic fractures (SOE: low) or in Grade 3 toxicities (SOE: low) between LDMF versus HDMF schemes.
• Evidence was considered insufficient to compare multiple fraction schemes on risk of radiation induced myelopathy in patients with MSCC or on risk of new spinal cord compression (SOE: insufficient).
• There was no difference in frequency of re-irradiation by multiple fraction scheme.

3.1.3.2. Description of Included Studies

Ten RCTs (in 12 publications)^{42,51,63,73,76,79,80,82,89–92} and one retrospective NRSI¹³² compared different multiple fraction EBRT schemes (Appendix E, Tables E-1 and E-2). We reported the lower total dose multiple fraction EBRT as the intervention (LDMF), which generally had fewer fractions (dose per fraction generally higher) and higher total dose EBRT fractions as the control (HDMF), which generally represented more lower-dose fractions. Given the large number of RCTs, one NRSI was included for information on harms only.¹³²

Across the 10 RCTs, sample sizes ranged from 60 to 300 (total N=1,615). Mean or median ages range from 53 to 68. The average proportion of males in trials was 55 percent (range, 0% to 87%). No trial reported race or ethnicity. Four trials^51,73,76,91 reported some patients as nonambulatory or severely mobile-impaired (range 27% to 62%). Primary tumor types included breast (range 10% to 100%), lung (range 20% to 33%), and prostate (range 5% to 12%). One trial included only breast cancer⁹² and another reported 89 percent of patients had hepatocellular carcinoma.⁶³ One trial reported the primary tumor histology in terms of favorable (10%) or unfavorable (20%) with the remaining classified as “intermediate”.⁵¹ The site of bone metastases was mixed (i.e., spine and nonspine) in six RCTs (spine 27% to 89% and nonspine 11% to 73%)^{42,63,76,80,82,92} (spinal cord compression and pathologic fracture were exclusion criteria in most of these trials), spine only in one RCT (presence or absence of cord compression not report),⁷⁹ and three RCTs included patients with MSCC only.^51,73,91 The proportion of patients with multiple bone metastases ranged from 58 to 86 percent (3 RCTs)^63,82,91 and the proportion with metastases to nonbone/visceral sites ranged from 20 to 77 percent (3 RCTs).^51,73,91 One trial reported lesions in terms of lytic (88%) and sclerotic (8%).⁸²

The total dose in the LDMF arms ranged from 15 Gy to 40 Gy with 20 Gy (4 Gy in 5 fractions), the most common dose-fractionation scheme,^{76,79,80,82,91} followed by 15 Gy (3 Gy in 5 fractions).^42,92 Across the HDMF arms, total dose ranged from 30 Gy to 60 Gy and the most common dose-fractionation scheme was 30 Gy (3 Gy in 10 fractions)^{42,51,79,82,91,92} followed by 30 Gy (2 Gy in 15 fractions).^76,80 Most trials did not clearly report the specific type of EBRT employed but it was most likely 2D or 3DCRT as these are most commonly used. Prior radiation therapy to the site, surgery, and/or chemotherapy were generally exclusion criteria in the trials. However, small proportions of patients had concomitant surgery in two trials,^76,82 and concurrent chemotherapy was common in two trials.^63,76 Other concomitant treatments included dexamethasone in two trials,^51,73 and antiemetic prophylaxis⁷³ and bisphosphonates⁹¹ in one each. One trial gave all patients zoledronic acid in conjunction with EBRT and excluded those with pretreatment bisphosphonates.⁴² The proportion of patients using narcotics at baseline ranged from 4 to 52 percent across three trials.^79,82,92 Median or mean followup periods ranged from 12 to 156 weeks.

Two trials were conducted in Germany,^76,89–91 two in Turkey,^42,82 and one each in Italy,⁷³ India,⁷⁹ China,⁶³ Egypt,⁵¹ Japan,⁸⁰ and Denmark;⁹² most were either single center or did not report how many centers were involved. None of the trials reported funding sources.

Four trials were fair quality^{42,63,73,89–91} and six were poor quality.^{51,76,79,80,82,92} No trial blinded care providers, patients or outcome assessors. Allocation concealment, high attrition, and lack of intention-to-treat analyses were other common limitations (Appendix F, Table F-1).

Given the number of RCTs that compared different multiple fraction EBRT schemes, the eligible NRSI was included for evaluation of harms only (see Methods Appendix A, Process for Selecting Studies) and is described in Appendix B.

3.1.3.3. Detailed Synthesis

3.1.3.3.1. Pain

Definitions of pain responses varied across trials, particularly with regard to definitions of complete response (Appendix B, Table B-2). We focused on overall response below, which encompasses complete and partial response in most trials (Appendix E, Table E-1).

Eight RCTs^{42,63,73,76,79,80,90,92} comparing LDMF with HDMF contributed data to meta-analyses of overall pain response. Five trials^{42,76,79,80,92} were in patients with mixed spine and nonspine MBD, two trials^73,90 were in patients with spinal cord compression and one trial⁶³ in patients with hepatocellular carcinoma did not report on MBD site.

There were no differences between schemes using LDMF versus HDMF in overall pain response posttreatment up to 4 weeks (6 RCTs, N= 788, 64.1% vs. 67.0%, RR 0.96, 95% CI 0.87 to 1.06, I²=0%),^{42,73,76,79,90,92} from >4 to 12 weeks (3 RCTs, N=275, 79.6% vs. 80.4%, RR 1.02, 95% CI 0.89 to 1.12, I²=0%),^42,90,92 at >12 weeks (2 RCTs, N=114, 78.6% vs. 72.4%, RR 1.10, 95% CI 0.86 to 1.38, I²=0%),^90,92 or in studies where time of assessment was unclear or not reported (3 RCTs, N=372, 70.4% vs. 74.4%, RR 0.95, 95% CI 0.87 to 1.08, I²=0%)^63,76,80 (Figure 8). Exclusion of poor-quality trials did not substantially change effect estimates, heterogeneity, or conclusions at any timeframe. There was no difference in overall response between LDMF and HDMF in analyses based on longest followup (8 RCTs, N= 902, 70.1% vs. 72.0%, RR 0.99, 95% CI 0.93 to 1.07, I²=0%)^{42,63,73,76,79,80,90,92} or when patients with mixed MBD were considered separately from those with MSCC (Appendix I, Figures I-12 and I-13).

Similarly, there were no differences in complete response between multiple fraction schemes at any timeframe or for analyses based on longest followup (Appendix I, Figures I-14 to I-16).

Figure 8

Comparison of multiple fraction EBRT schemes: Overall pain response by timeframe. CI = confidence interval; EBRT = external beam radiation therapy; HCC = hepatocellular carcinoma; HDMF = higher total dose multiple fractions (control); LDMF = lower total (more...)

3.1.4.1. Key Points

• Studies meeting inclusion criteria did not report primary outcomes of interest
• There may be no differences in pathologic fractures between SF SBRT and MF SBRT in one RCT and one NRSI, both rated fair quality (SOE: low).
• There may be no differences between SF SBRT and MF SBRT in Grade ≥3 toxicities or in Grade ≥2 toxicities in one RCT and one NRSI, both studies were rated fair quality (SOE: low).
• There was insufficient evidence from NRSI regarding the following adverse events: pain flare, transesophageal fistula and Grade 3 or 4 toxicities.

3.1.4.2. Description of Included Studies

One multicenter RCT¹⁰⁶ and four NRSIs^{113,114,120,136} compared single (SF) versus multiple (MF) dose-fractionation schemes of SBRT for the palliative treatment of MBD. Aside from toxicity and harms, primary outcomes of interest for this review were not reported in any of these studies. The primary focus of each study was local control and overall survival, and palliative intent was generally not clear from study descriptions. Study details can be found in the data abstraction (Appendix E, Tables E-1 and E-2).

In the RCT (N=117), median age was 64 years (32–89 years). Most patients were male (71%) with solitary (80%) bone only lesions (3% had bone plus nodal lesions and 9% had nodal only lesions); patients with >5 metastatic lesions were excluded. Inclusion appears to have been based on imaging findings of spinal metastasis, not symptomatic status. Most lesions involved the spine (62%), but authors do not report spinal cord compression. The most common primary cancer was prostate (47%) followed by lung (9%), colorectal (9%) and renal (7%) cancer. A single fraction dose of 24 Gy was compared with a three fractions of 9 Gy delivered every other day (27 Gy total) MF SBRT scheme. Concurrent systemic or hormonal therapy (not specified) was common (61%). Pretreatment with dexamethasone (4 mg twice daily) was primarily given to the SF EBRT group and was selective in patients receiving the 3-fraction scheme. Posttreatment adjuvant therapies were at the physician’s discretion. Baseline pain was not reported.

Across four NRSIs, samples sizes ranged from 43 to 127 (total N=363). The average study mean age ranged from 45 to 64 years and the proportion of males ranged from 40 to 79 percent. One NRSI enrolled patients with renal cell carcinoma with 56 percent of lesions occurring in the spine, 21 percent in pelvic bone structures, 9 percent in the femur and 13 percent in other bones.¹³⁶ Three other studies were in patients with spine metastasis; MSCC was present in all patients in one study,¹²⁰ one study excluded patients with MSCC,¹¹⁴ and the third did not report spinal cord compression.¹¹³ One NRSI enrolled patients with metastatic lesions (56% were to the spine) from renal cell carcinoma;¹³⁶ another NRSI from the same institution enrolled patients with proven high-grade sarcoma metastases to spine.¹¹³ One additional NRSI enrolled patients with spinal metastases from renal cell caricinoma.¹¹⁴ The most common primary tumors in the other NRSI of spine metastases were from breast (21%) and lung (20%). Most spine segment lesions in this study were radiosensitive (58%) and were primarily carcinoma from breast or prostate. Radioresistant lesions (42% of segments) were primarily carcinomas from colon, renal cell, uterine, or thyroid origin (Appendix E, Table E-2). Single fraction doses of 18 to 24 Gy were used in three NRSIs^113,114,136 with 16 or 18 Gy reported in one NRSI.¹²⁰ Various dose and multiple fraction schemes were reported in NRSIs including 20 to 30 Gy (3–5 fractions),¹³⁶ median 28.5 Gy dose (3–6 fractions),¹¹³ 30 Gy (5 fractions),¹¹⁴ and in one NRSI 21 Gy, 24 Gy, 25.5 Gy or 27 Gy (3 fractions) or 30 Gy (5 fractions).¹²⁰

All studies were conducted in the United States. The RCT was partially funded by government sources.¹⁰⁶ The funding source was not reported for three NRSIs^113,114,136 and one NRSI reported that no funding was received.¹²⁰ The RCT was fair quality due to unclear reporting of attrition or assessor blinding (Appendix F, Table F-1). The trial was stopped early due to slow enrollment. Two NRSIs were fair quality^113,136 and two were poor quality^114,120 (Appendix F, Table F-2). Lack of assessor blinding was noted across studies. Other methodological limitations included concerns about patient selection and unclear reporting of attrition. Data on toxicities and adverse events in most studies were based on numbers of lesions or sites versus number of patients and most analyses did not adjust for correlated data.

3.1.4.3. Detailed Synthesis

The included RCT and NRSIs did not provide information by dose/fraction for the primary outcomes of interest for this review except for toxicities and harms. The primary focus of these studies was to evaluate local control/local failure and/or overall survival, the results for which can be found in Results Appendix B. Risk of re-irradiation was not reported in any included study.

3.1.5.1. Key Points

• There is insufficient evidence on primary outcomes of interest or harms from one poor-quality NRSI to compare multiple SBRT fraction schemes.

3.1.5.2. Description of Included Studies

Two NRSIs comparing different multiple dose-fractionation SBRT schemes were identified and provide insufficient information (Appendix E, Table E-2).^107,115 One study of post-operative SBRT (N=80) in patients with spinal metastases provided limited comparison of SBRT dose-fractionation schemes as part of multivariate analysis evaluating predictors of local control.¹⁰⁷ Primary outcomes were not reported. Mean patient age was 59 years and 55 percent were male. Primary cancer for 44 percent of patients was listed as “other”. Common population characteristics included presence of baseline vertebral compression fracture (55%), presence of paraspinal extension (78%), prior EBRT (75%, mean 20 Gy/5 fractions), and ECOG score of −1 (88.7%). Most patients had surgical decompression alone (36%) or with instrumented stabilization (50%). Three patients (3.7%) received a single fraction 24 Gy SBRT dose, 40 percent received 18 to 26 Gy in two fractions, and 56 percent received 18 to 40 Gy (3–5 fractions). Population characteristics were not provided by dose/fractionation scheme. The study was conducted in Canada; no funding was received. It was rated poor quality based on inadequate information on how patients receiving different dose/fractionation schemes compared at baseline, no reporting of attrition, and unclear assessor blinding (Appendix F, Table F-2).

Another small, prospective NRSI (N=57) in patients with spinal metastases did not control for potential confounding and was of poor quality and is included here for completeness.¹¹⁵ Mean patient age was 64 years, and the majority were male (56%). The most common primary tumors were breast (22%), non-small cell lung cancer (20%), prostate (20%) and other (19%); the majority of patients had oligometastases (56%). Most patients had KPS >70 (80%), fourteen patients (26%) had vertebral compression fractures at enrollment and 30 percent of patients had surgical treatment of spinal lesions prior to SBRT. SBRT of 35 Gy (5 fractions) was compared to 48.5 Gy (10 fractions) SBRT. The study, conducted in Sweden, received partial government funding. It was rated poor quality based on unclear criteria for patient selection, high attrition, unclear comparability between treatment groups on baseline characteristics and failure to control for potential confounding (Appendix F, Table F-2).

3.1.5.3. Detailed Synthesis

3.1.6.1. Key Points

• There may be no differences between IMRT and 3DCRT in overall pain or quality of life outcomes at any timepoint in one small fair-quality RCT of spinal metastases (SOE: low).
• Evidence was insufficient for pathologic fractures, Grade 3 or 4 toxicity or treatment related deaths.

3.1.6.2. Description of Included Studies

One RCT (reported in three publications)^22,24,100 and two NRSIs^125,128 compared image guided intensity modulated radiotherapy (IMRT) to three-dimensional conformal radiotherapy (3DCRT) (Appendix E, Tables E-1 and E-2).

One, small RCT (N=60)^22,24,100 compared IMRT with 3DCRT (both delivered in 3 Gy over 10 fraction) in patients with spine metastases (mostly thoracic and lumbar); authors stated that spinal cord compression was not a specific criterion for exclusion but did not indicate if any patients had cord compression at baseline, though 12 percent had a neurological deficit. Mean patient age was 64 years and 55 percent were male with a mean Karnofsky performance status score of 63 (out of 100). Race, social determinants of health, and comorbidities were not reported. Primary tumor sites were lung (45%), breast (22%), or prostate (11%) primarily; authors did not describe tumor histology in terms of favorable or unfavorable. The number of metastases differed between treatment groups: more patients randomized to IMRT had a single metastasis compared with 3DCRT (57% vs. 33%) and fewer had two metastatic sites (13% vs. 30%); the proportion of patients with three metastases was similar (30% vs. 37%). Distant metastases were present in the viscera (40%), lung (22%), brain (15%) and tissue (15%). The proportion of lytic/sclerotic lesions was not reported, nor was the presence of preexisting fractures. Prior RT was an exclusion criterion, but nearly all other prior and concurrent therapies differ between groups by ≥10%; patients in the IMRT group received more medications across all categories, including opiates (67% vs. 57%) and nonsteroidal anti-inflammatory drugs (NSAIDs) (77% vs. 63%) and received more bisphosphonates (43% vs. 23%). About 30 percent of patients in both groups wore an orthopedic corset. The trial was conducted at one center in Germany and followed patients for a median of 4.3 months (range of 0.5 to 10 months). Authors reported that no funding was received. The trial was rated fair quality due to unclear randomization techniques, lack of blinding, and high attrition rates (Appendix F, Table F-1).

Across the two NRSIs, sample sizes ranged from 179 to 716 (total N=895).^125,128 One study reported median age of 61 years,¹²⁸ while the other split age into <65 (65% versus 48%) and >65 years (35% versus 52%).¹²⁵ Neither NRSI reported race or social determinants of health. Most primary tumor sites were lungs (range, 24% to 37%), breast (range, 7% to 19%), and prostate (range, 8% to 21%). Neither study reported tumor histology in terms of favorable or unfavorable or whether lesions were lytic or sclerotic. One study included patients with MSCC only¹²⁵ (58% were ambulatory before treatment) but excluded patients with pre-existing fractures, while the other study included patients with mixed spine (59% of lesions) and nonspine (29% of lesions) metastases or both (12% of lesions);¹²⁸ patients in the latter trial had a mean ECOG performance status score 1.6. In the study that included MSCC, more patients in the IMRT group had other bone metastases (78% vs. 66%) and visceral metastases (65% vs. 49%) at the time of therapy compared with the 3DCRT group, but further details were not reported.¹²⁵ Fewer patients in the IMRT group in this trial had three or more metastases (35% vs. 60% in 3DCRT group). Few to no patients (0% to 4%) had prior RT, no patient had prior surgery and other concurrent treatments (chemotherapy, palliative care management, dexamethasone, corticosteroids) ranged from 39 percent to 78 percent across studies. Neither study reported opioid use at baseline.

One study (mixed site MBD) assessed conformal radiotherapy using a technique designed to mimic IMRT and compared it to nonconformal RT; patients were given a mean total dose of 19.6 Gy over a mean of 4.4 fractions.¹²⁸ The other study (MSCC) used precision RT dosed at 5 Gy in five fractions and compared it to a historical control group that received conventional RT with 4 Gy in five fractions.¹²⁵

One study was conducted in the United States¹²⁸ and the other in Germany.¹²⁵ Followup was 26 weeks in both studies. One study¹²⁵ was funded by government, the other was unclear. One study was fair quality¹²⁵ and the other poor quality.¹²⁸ Common methodological limitations included imbalances in prognostic factors at baseline and lack of blinding (Appendix F, Table F-2); additional concerns in the poor-quality study included unclear attrition.

3.1.6.3. Detailed Synthesis

3.1.6.3.1. Pain

Overall, there were no differences between IMRT and 3DCRT in pain outcomes including overall and complete pain response (Table 3), VAS pain scores, and neuropathic pain (data unclear or not provide for latter two outcomes) immediately after RT or at 12 and 26 weeks in one RCT.¹⁰⁰ The one exception was VAS pain scores at 12 weeks which were slightly better in patients who received IMRT (p=0.04, data not provided).

One NRSI (N=254) found that more patients who received IMRT showed significant improvement in their pain during treatment compared with 3DCRT (30.5% vs. 15.2%, RR 2.04, 95% CI 1.24 to 3.37); however, there was no difference between groups at 8 weeks (28.2% vs. 32.1%).¹²⁸

Table 3

Pain response in one RCT comparing IMRT with 3DCRT.

3.1.6.3.3. Quality of Life

There were no differences in quality of life based on the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire for Patients with Bone Metastases 22 (EORTC-QLQ-BM-22) questionnaire between IMRT and 3DCRT at any timepoint measured in one RCT²⁴ (Table 4). The NRSIs did not report QOL outcomes.

Table 4

Quality of life outcomes in one RCT comparing IMRT with 3DCRT.

3.1.7.1. Key Points

• No primary outcomes of interest were reported by one fair-quality RCT (N=450) comparing the addition of hemibody irradiation (HBI) to EBRT versus EBRT alone.
• Evidence was insufficient for Grade 3 or 4 toxicities and other serious events.

3.1.7.2. Description of Included Studies

One RCT (N=450)⁸⁶ compared a single 8 Gy dose of HBI in addition to 30 Gy (3 x 10 Gy) of EBRT versus 30 Gy of EBRT alone given over 2 weeks. Most patients were 60 years of age or older (69%) and male (59%) with a KPS score ≥70 (79%). The trial did not report race or ethnicity, comorbidities or social determinants of health. The most common primary tumor types included prostate (33%), breast (27%) and lung (24%). The study did not report primary tumor histology in terms of favorable or unfavorable or bone metastases in terms of complicated or uncomplicated. Most patients had multiple bone metastases (82%); the trial did not report the metastases sites. In the HBI group, the targeted hemibody area for most patients was the lower third (64%) and HBI was given within 1 week of the local EBRT. Patients on hormonal therapy were enrolled if therapy was stable for the 2 months prior to randomization; chemotherapy within 2 weeks of entry into the study was an exclusion criterion. This trial was conducted in the United States and was supported by a grant from the Radiation Therapy Oncology Group. It was rated fair quality due to unclear allocation concealment methods and lack of blinding.

3.1.7.3. Detailed Synthesis

3.1.8.1. Key Points

• SBRT was associated with a small increase in the likelihood of experiencing overall pain response compared with conventional EBRT posttreatment up to 4 weeks (2 RCTs [excluding poor quality], N=325, 60% vs. 48%, RR 1.24, 95% CI 0.98 to 1.57, I²=0%) (SOE: low), at 12 weeks (4 RCTs, N=408, 59% vs. 44%, RR 1.31, 95% CI 1.05 to 1.61, I²=0%) (SOE: moderate) and up to 26 weeks (3 RCTs, N=324, 50% vs. 37%, RR 1.32, 95% CI 1.01 to 1.92, I²=24.3%) (SOE: low). At 36 weeks a small RCT in patients with nonspine MBD found moderate increase in the likelihood of overall response (SOE: low) with SBRT versus EBRT.
• SBRT was also associated with large improvement in VAS pain score (0-10 scale) >12 weeks (SOE: low); evidence was insufficient at other time frames.
• SBRT was associated with improved stability based on the Spinal Instability in Neoplasia Score (SINS) (0-18 scale) at 12 weeks; there was no difference at 26 weeks compared with conventional EBRT (SOE: low).
• There were no differences between SBRT and EBRT on any quality-of-life measures at 12, 26 (SOE: low), or 52 weeks (SOE: insufficient).
• There were no differences between SBRT and EBRT on spinal cord compression by 26 weeks (SOE: low), pathologic fracture at 12 weeks (SOE: low), or pain flare within 2 days of treatment or at 26 weeks (SOE: low).

3.1.8.2. Description of Included Studies

Four RCTs (in 6 publication)^{9,23,84,94,99,101} and four NRSIs^{108,116,130,133} compared SBRT versus conventional EBRT. Another population based comparative NRSI compared advanced techniques (IMRT, 3DCRT, SBRT) with simple conventional EBRT¹²⁴ (Appendix E, Tables E-1 and E-2).

Across the RCTs, sample sizes ranged from 60 to 229 (total N=559). The average age of participants was a mean 62.3 years (range 62 to 63 years) in two trials^9,23,99,101 and a median 64 years in two trials.^84,94 The average proportion of males across trials was 56.2% (range 51% to 62%). Few trials reported comorbidities, social determinants of health or race or ethnicity, with one exception regarding race (79% White, 6% Black, 7% Hispanic/Latino, 3% Asian, and 4% other).⁹ The most common primary tumor types across all RCTs included lung (range 26% to 49%) and breast (range 9% to 31%), as well as prostate in two trials (14% to 51%).^9,84 None of the trials reported the primary tumor histology in terms of favorable or unfavorable. Bone metastases were present at multiple sites in 16 percent to 22 percent of patients across two trials;^9,23,99,101 in one of these trials most patients had metastases to other nonbone sites (47% visceral, 27% lung, 18% brain, and 16% tissue).^23,99,101 In one trial, the metastatic bone lesions were lytic in 41 percent, sclerotic in 28 percent, and mixed in 30 percent of participants. The site of bone metastases was limited to the spine in two RCTs,^23,94 mixed spine (55%) and nonspine (45%) in one RCT,⁸⁴ and nonspine sites only (pelvis primarily, 59%) in one RCT.⁹ Spinal cord compression was an exclusion criterion in all trials including spine metastases. In the two RCTs that included spinal metastases only, 29 percent of patients in one trial had a preexisting pathological fracture²³ and 27 percent in the other had <50 percent vertebral body collapse (2% had ≥50% collapse) aselyne.⁹⁴ No trials described bone metastases as either complicated or uncomplicated.

The SBRT dose varied among RCTs (12 to 24 Gy in one fraction, 24 Gy total in two fractions, 30 Gy total in three fractions, 35 Gy in five fractions).^{9,23,84,94,99,101} The most common EBRT dose was 30 Gy (3 Gy x 10); one trial primarily used single fraction EBRT (8 Gy) and two trials also used 20 Gy (4 Gy x 5).^84,94 Most trials did not clearly report the specific type of EBRT employed but it was most likely 2D or 3DCRT. Concomitant treatments included analgesics and systemic therapies in 41 and 46 percent of participants. Previous treatments included systemic therapy and targeted therapies. Most trials excluded patients who had chemotherapy, prior RT to the treatment site, spinal cord compression, compression fracture, and surgery. The proportion of patients who used opioids at baseline ranged from 38 to 51 percent in two trials that reported this information.^23,84,99,101 Followup periods ranged from 12 to 104 weeks.

One RCT was conducted in the United States,⁹ two in Europe,^23,84,99,101 and one in Canada and Australia⁹⁴ and most were single center trials. The most common source of funding across the trials was government, followed by industry, private and unclear funding.

Three RCTs were fair quality^{9,23,94,99,101} and one was poor quality.⁸⁴ Common limitations included unclear randomization and allocation concealment methods, lack of blinding, and high attrition (Appendix F, Table F-1). In many cases, the high attrition was due to high mortality (range 15.7% to 56.9%) which is to be expected in this patient population.

Across the NRSIs of SBRT versus EBRT, sample sizes ranged from 44 to 131 (total N=277). The average study mean age of participants was 64 years (range 59 to 66 years) in two studies^130,133 and the median age was 51 years in two studies (range, 46 to 57).^108,116 The average proportion of males in trials was 61 percent (range 25% to 93%). No studies reported on race or ethnicity, comorbidities or social determinants of health. The primary tumor types reported included breast (range, 24% to 50%), lung (range, 22% to 36%), and prostate (30% in one study¹³³); one trial enrolled only patients with hepatocellular carcinoma. None of the NRSIs reported the primary tumor histology in terms of favorable or unfavorable. Bone metastases were present at multiple sites in 53 to 65 percent of patients across two studies.^130,133 The site of bone metastases was mixed (i.e., spine and nonspine) in two NRSIs (spine, 35.8% to 45% and nonspine, 64.2% to 65%).^108,133 Two NRSIs included bone metastases to the spine only and no trials included bone metastases to nonspine sites only. No studies described bone metastases as either complicated or uncomplicated. The NRSI (N=1,712)¹²⁴ of advanced techniques versus EBRT (categorized as simple, parallel opposed pair RT) consisted of mostly males (64%), 50 to 70 years old (50%). The most common primary tumors were prostate, breast, and lung. The most common MBD sites were spine (55%) and pelvis (21%).

Total doses and fractionation schedules varied across the SBRT and EBRT arms across all studies (Appendix E, Table E-2). Most studies did not clearly report the specific type of EBRT employed but it was most likely 2D or 3DCRT; one study reported using 3DCRT.¹³³ Concomitant treatments reported were surgery, radioisotope injection (samarium), and analgesics in two studies.^116,133 Previous treatments included systemic therapy in one study.¹⁰⁸ Most studies did not report exclusion criteria, but prior treatment and surgery were exclusion criteria in one trial.¹¹⁶ Followup periods ranged from 4 to 22 weeks. The NRSI of advanced techniques versus EBRT did not report specific doses and fractions; the most used advanced technique was IMRT (67%), followed by 3DCRT (26%) and SBRT (8%).¹²⁴

Two studies were conducted in the United States,^108,116 one in Europe,¹³³ one in South Korea,¹³⁰ and one in Canada¹²⁴ and most were single center studies. The most common source of funding across the trials was government, followed by unclear funding. All five NRSIs were fair quality (Appendix F, Table F-2).^{108,116,124,130,133} Common limitations included imbalances in prognostic factors between groups at baseline and unclear attrition.

3.1.8.3. Detailed Synthesis

3.1.8.3.1. Pain

All four RCTs contributed data to meta-analyses of overall pain response. There was no difference between SBRT and conventional EBRT in overall response at 4 weeks post-RT (3 RCTs, N=394, 57.4% vs. 50.0%, RR 1.15, 95% CI 0.83 to 1.43, I²=40%) (Figure 9).^9,84,94 However, exclusion of the one poor-quality trial resulted in a small increase in the likelihood of achieving overall pain response with SBRT at this timepoint and eliminated heterogeneity (2 RCTs, N=325, 60.2% vs. 48.4%, RR 1.24, 95% CI 0.98 to 1.57, I²=0%).^9,94 SBRT was associated with a small increase in the likelihood of achieving overall pain response compared with EBRT at 12 weeks (4 RCTs, N=408, 59.4% vs. 44.4%, RR 1.31, 95% CI 1.05 to 1.61, I²=0%)^9,23,84,94 and at 26 weeks (3 RCTs, N=324, 49.7% vs. 36.8%, RR 1.32, 95% CI 1.01 to 1.92, I²=24.3%),^9,23,94 and a moderate increase at 36 weeks in one trial in patients with nonspine metastases (N=48, 77.3% vs. 46.2%, RR 1.67, 95% CI 1.04 to 2.69) (Figure 9).⁹ Exclusion of the poor-quality trial at 12 weeks resulted in a slightly larger but similar estimate (3 RCTs, N=354, 59.7% vs. 42.2%, RR 1.41, 95% CI 1.13 to 1.77, I²=0%).^9,23,94 Similarly, SBRT was associated with a small increase in the likelihood of achieving overall pain response compared with conventional EBRT in analysis based on longest followup (12 to 36 weeks) across trials (4 RCTs, N= 370, 51.6% vs. 37.5%, RR 1.35, 95% CI 1.02 to 1.95, I²=39.2%) (Appendix I, Figure I-17);^9,23,84,94 exclusion of the one poor-quality trial resulted in a moderate increase in the likelihood of achieving overall pain response and eliminated heterogeneity (3 RCTs, N=316, 50.3% vs. 34.2%, RR 1.52, 95% CI 1.16 to 2.21, I²=0%).^9,23,94

When RCTs were analyzed separately at longest followup based on the site of MBD, SBRT was associated with a small increase in the likelihood of achieving overall pain response compared with EBRT in populations with spine metastases (2 RCTs, N=268, 45.9% vs. 31.9%, RR 1.46, 95% CI 0.97 to 2.71, I²=34.9%)^23,94 and a moderate increase in patients with nonspine metastases (1 RCT, N=48, 77.3% vs. 46.2%, RR 1.67, 95% CI 1.04 to 2.69)⁹ (Appendix I, Figure I-17); in both populations, these associations were first seen at the 12-week followup and persisted through final followup (Appendix I, Figure I-18). There were no differences between treatment groups at any timepoint in the poor-quality trial that included a population with MBD at mixed (i.e., spine and nonspine) sites.⁸⁴

Figure 9

SBRT versus conventional EBRT: Overall pain response by timeframe. 3DCRT = three-dimensional conformal radiation therapy; CI = confidence interval; C-EBRT = conventional external beam radiation therapy; MBD = metastatic bone disease; PL = profile likelihood; (more...)

There was no difference between SBRT and conventional EBRT in complete pain response at 4 weeks (2 RCTs, N=298, 23.3% vs. 16.9%, RR 1.40, 95% CI 0.65 to 2.44, I²=0%)^84,94 or 12 weeks (3 RCTs, N=329, 32.1% vs. 15.5%, RR 2.09, 95% CI 0.66 to 4.08, I²=58.8%).^23,84,94 However, after exclusion of the poor-quality, outlier trial at 12 weeks SBRT was associated with a large increase in the likelihood of achieving complete pain response compared with EBRT across the two trials in patients with spinal metastases (N=275, 36.5% vs. 14.5%, RR 2.52, 95% CI 1.42 to 4.46, I²=0%);^23,94 this effect persisted at 26 weeks (2 RCTs, N=268, 35.3% vs. 14.8%, RR 2.31, 95% CI 1.25 to 7.15, I²=35.2%)^23,94 (Appendix I, Figures I-19 to I-21). There was no difference between SBRT and EBRT in complete pain response at any timepoint (4 or 12 weeks) in the poor-quality trial in a population of mixed spine and nonspine bone metastases.

SBRT was associated with a large improvement in pain intensity (on a 0-10 scale) compared with EBRT at 26 weeks in one RCT in patients with spinal metastases (N=39, MD −2.13, 95% CI −3.59 to −0.67);²³ there were no differences between treatment groups at earlier timepoints (up to 4 weeks: 2 RCTs, N=143, MD 0.84, 95% CI −0.45 to 2.31, I²=0%; and >4 to 12 weeks: 2 RCTs, N=135, MD −0.90, −2.34 to 0.76, I²=0%) across both RCTs^23,84 (Appendix I, Figure I-22). One of these trials in patients with spine metastases reported no difference between groups in neuropathic pain at any timepoint up 26 weeks, but it is unclear how this outcome was measured.²³

Consistent with results from the RCTs, there were no differences between SBRT and EBRT in pain outcomes at 4 weeks in matched-pairs analyses across two NRSIs in patients with spinal metastases.^116,130 In one study in patients with primary hepatocellular carcinoma, complete pain relief (adjusted for pain medication) was reported by 21.4% (6/28) of SBRT versus 10.7% (3/28) of EBRT patients (p=0.83) and the mean change in VAS pain scores was −3.7 (SD 2.7) vs. −2.8 (SD 2.4), respectively, (p=0.13)¹³⁰ The second study did not provide data but stated that pain relief (excellent/good: complete relief with or without pain medication) did not differ between treatments (p=0.11).¹¹⁶ A third NRSI in patients with mixed spine and nonspine MBD from renal cell carcinoma reported significantly better symptom control (i.e., stable disease, partial pain response or complete pain response) with SBRT through 2 years (74.9% vs. 35.7%, p=0.020).¹⁰⁸

A fourth NRSI in patients with mixed spine and nonspine MBD also found no differences between advanced techniques and simple EBRT based on estimates (adjusted for age, primary histology, sex and treatment region) for partial pain response or complete response. Compared with simple EBRT (referent) adjusted estimates for partial pain response were: 3DCRT (OR 1.08, 95% CI 0.46 to 2.56), IMRT (OR 1.02, 95% CI 0.49 to 2.09) and SBRT (OR 0.64, 95% CI 0.10 to 4.12); adjusted estimates for complete pain response were: 3DCRT (OR 0.74, 95% CI 0.21 to 2.58), IMRT (OR 1.29, 95% CI 0.52 to 3.23), and SBRT (OR 1.16, 95% CI 0.13 to 10.36).¹²⁴

3.2.2.1. Description of Included Studies

Two RCTs (N=173 and 850)^58,103 compared re-irradiation SF EBRT with MF EBRT for the palliative treatment of bone metastases in populations with mixed spine/nonspine metastases (Appendix E, Table E-1). One trial¹⁰³ was a subsequent publication of the Dutch Bone Metastasis Study and reported on the subgroup of patients who underwent re-irradiation within 1 year of followup. The average study mean age of participants was 65 years for both trials and most were male (59% and 61%). Neither trial reported race or ethnicity, comorbidities or social determinants of health. The primary tumor types included breast (34% and 39%), lung (22% and 28%), and prostate (22% and 23%). Neither trial reported the primary tumor histology in terms of favorable or unfavorable. The proportion of patients with bone metastases at multiple sites was 51 percent in one trial¹⁰³ and not reported in the other. The locations of bone metastases were pelvis (36% and 40%); spine (22% and 28%); humerus (7% one trial) or upper limbs (10% one trial); and femur (8%, one trial¹⁰³). The proportion of metastases to nonbone/visceral was not reported in either trial, nor was the proportion of metastatic bone lesions that were lytic or sclerotic or complicated or uncomplicated. Spinal cord compression and pathologic fracture were exclusion criteria for both trials.

The single fraction dose was 8 Gy in both trials. The multiple fraction dose was 24 Gy (4 Gy x 6)¹⁰³ and 20 Gy (4 Gy x 5, with some exceptions based on target field and previous radiation therapy).⁵⁸ One RCT allowed both 2D- and 3D-EBRT,⁵⁸ while the type of EBRT was not reported in the other trial. Concomitant treatments included nonspecified bone-modifying agents and systemic therapy (no proportions provided) at the discretion of the treating physician in one RCT⁵⁸ and narcotic and nonnarcotic analgesics in both trials. The proportion of patients who used opioids at baseline was 32 percent at a mean daily dose of morphine equivalence of 44 mg in the trial that reported baseline analgesic use.⁵⁸ Previous treatments as reported by one trial included nonspecified systemic therapy in 51 percent of the population which differed according to primary cancer type (79% with breast cancer, 81% with prostate cancer, and 12% of with lung cancer).¹⁰³ One trial excluded patients with metastases of renal cell carcinoma, melanoma, and cervical spine metastasis.¹⁰³ Patients with treatment areas associated with previous palliative surgery or who were receiving systematic radiotherapy or half-body irradiation within 30 days of randomization were excluded in the other trial.⁵⁸ Median followup was 12.2 months⁵⁸ and a maximum of 2 years (mean followup duration not reported).¹⁰³

One trial was conducted at 17 sites in The Netherlands and did not report funding source.¹⁰³ The other trial was conducted in Canada, Australia, New Zealand, United States, Israel, Switzerland, United Kingdom, The Netherlands, and France and funded primarily by national cancer institutes.⁵⁸

One trial was rated fair quality⁵⁸ and the other poor quality.¹⁰³ Methodological limitations included inability to blind care providers or patients, unclear randomization and allocation concealment methods, unclear if randomized groups were similar at baseline, and high attrition, only partly due to high mortality (Appendix F, Table F-1).

Two NRSI compared SF versus MF EBRT for re-irradiation (Appendix E, Table E-2). Sample sizes were 60 and 62 (total N=122).^127,129 The median age of participants was 55 years in one study and 63 years at primary radiation therapy in the other and just over half were male (52% and 56%). Neither study reported race or ethnicity, comorbidities or social determinants of health. The primary tumor types included breast (27% and 37%), lung (8% and 10%), prostate (3% and 35%), and multiple myeloma (13%)¹²⁹ or myeloma/lymphoma (7%).¹²⁷ Neither study reported the primary tumor histology in terms of favorable or unfavorable or reported the proportion of patient with bone metastases at multiple sites or to nonbone/visceral sites, nor were bone metastases described as complicated or uncomplicated. One NRSI excluded patients with lytic lesions >3 cm or >50% cortical erosion of bone diameter. The site of bone metastases was mixed in one study (55% spine, 45% nonspine)¹²⁹ while all patients in the second study had MSCC (36% lumbar spine alone, 34% thoracic spine alone, 24% cervical and thoracic spine, and 6% thoracic and lumbar spine).¹²⁷ Spinal cord compression was an exclusion criterion in the mixed spine/nonspine metastases study along with any high-risk lesions for pathological fracture. In the NRSI of patients with spinal cord compression, neurologic outcomes such as pain and incontinence were not studied, although all had motor deficits of the lower limbs with baseline motor function scores of Grade 1 (ambulatory without aid, 10%), Grade 2 (ambulatory with aid, 65%), Grade 3 (not ambulatory, 24%), and Grade 4 (paraplegia, 1.6%).

The single fraction dose was 8 Gy in both NRSIs; the multiple fraction dose was 20 Gy (4 Gy x 5) over 5 days or 8 days (if spine/whole pelvis involved)¹²⁹ and 20 Gy (4 Gy x 5) or 15 Gy (3 Gy x5) depending on initial radiation treatment (treatment duration not reported).¹²⁷ Neither study reported the specific EBRT used. All patients received concomitant chemotherapy and/or hormonal systemic therapy and bisphosphonates in one study,¹²⁹ while concomitant treatment was not described in the other study.¹²⁷ Previous treatments such as surgery or chemotherapy for the targeted spinal area were not allowed in one study¹²⁷ and not reported in the other.¹²⁹ The proportion of patients who used opioids at baseline was 90 percent¹²⁹ or not reported.¹²⁷ Median followup times were 7 months¹²⁹ and 12 months.¹²⁷

The NRSIs were conducted in Europe at one or more sites¹²⁷ and at a single site in Egypt.¹²⁹ Neither study reported funding source. Both NRSIs were rated poor quality due to unclear blinding of outcome assessors, unclear if comparison groups similar at baseline, and lack of adjustment for prognostic confounding variables (Appendix F, Table F-2).

3.2.2.2. Detailed Synthesis

3.2.3.1. Description of Included Studies

One retrospective NRSI (N=228; 348 lesions)¹¹⁷ compared re-irradiation with SF SBRT (mean 16.3 Gy) versus MF SBRT (mean 20.6 Gy in 3 fractions, 23.8 Gy in 4 fractions, and 25.4 Gy in 5 fractions) in patients with primarily (75%) thoracic and lumbar spinal metastases (Appendix E, Table E-2). Individuals with frank spinal cord compression or spinal instability were excluded. Most patients (56%) had previous EBRT and received SBRT due to pain (97%). Primary tumor types included breast (25%), lung (18%), renal cell (15%), and thyroid (8%) cancer. The mean patient age was 59 years and 48 percent were male. Other patient (race/ethnicity, comorbidities, social determinants of health, primary tumor histology in terms of favorable or unfavorable) and bone (number of metastases, metastases to nonbone/visceral sites, lytic or sclerotic lesions, complicated versus uncomplicated lesions) characteristics were not reported. Median followup was 360 days. This study was conducted in the United States and partially funded by industry. It was rated fair quality due to baseline differences between groups, unclear blinding of outcome assessors and lack of adjustment for all prognostic variables, although adjustment was made for initial tumor volume (Appendix F, Table F-2).

3.2.3.2. Detailed Synthesis

3.3.2.1. Description of Included Studies

Two RCTs conducted in patients with metastatic prostate cancer compared EBRT with strontium-89 chloride (a radioisotope that delivers radiation to cancerous areas)^81,88 (Appendix E, Table E-1). Sample sizes were 111 and 203, median ages were 67 and 71, and all patients were male. Neither trial reported race or ethnicity, comorbidities or social determinants of health. One trial reported the median number of hot spots on bone scans was 11 with about 35 percent of patients reporting one painful metastasis and about 44 percent reporting two painful metastases.⁸¹ Neither trial reported the primary tumor histology in terms of favorable or unfavorable or bone metastases in terms of complicated or uncomplicated. One trial required all patients to have sclerotic bone metastases for study entry.⁸⁸ Risk of spinal cord compression or pathological fracture were exclusion criteria in one trial.⁸⁸ Both trials required patients to have hormone-resistant prostate cancer.

The dose of EBRT was 20 Gy (4 Gy x 5 or a single dose of 8 Gy) for local radiotherapy in one trial⁸⁸ or usual radiotherapy regimen per the treatment center (median of 4 Gy x 5) in the other.⁸¹ The type of EBRT (two- or three-dimensional) was not reported in the trials. The dose of Strontium-89 Chloride was 150 MBq in one trial and 200 MBq in the other. Concomitant therapies were not reported. One trial reported about 49 percent of patients were receiving at least 20 mg of morphine equivalents daily at study entry, while the other trial reported 35 percent of patients were receiving level-4 narcotics. One trial followed patients until death,⁸¹ whereas followup was 12 weeks in the other trial.⁸⁸ Of note, in this latter trial, patients who failed to achieve pain relief with the treatment assigned at randomization (either EBRT [4 Gy x 5 or 8 Gy x 1] or strontium-89 200 MBq) were offered the other treatment at 8 weeks.

One trial was conducted in the United Kingdom; it is unclear if the other trial was conducted solely in The Netherlands or was multinational. One trial received support from Amersham Laboratories (strontium), and the Scottish Urological Oncology Group; funding was not reported in one trial. One trial was rated fair quality and other rated poor quality⁸⁸ due to lack of blinding, reporting of findings for some patients at 12 weeks and for other patients at 8 weeks (those who crossed over to other treatment), and high attrition (Appendix F, Table F-1).

3.3.2.2. Detailed Synthesis

3.3.3.1. Description of Included Studies

One NRSI compared EBRT with cryoablation (use of extreme cold to kill cancer cells) (Appendix E, Table E-2).¹¹² Of 175 participants enrolled, 150 were treated with EBRT (n=125) or cryoablation (n=25). (The remaining 25 patients were treated with a combination of EBRT and cryoablation and are described under Key Questions 3b and 3c.) Patients were matched via propensity score analysis. Mean patient age was 68 years and 49 percent were male. Comorbidities, social determinants of health and race/ethnicity were not reported. Mean Karnofsky score ranged from 70 to 89 in exactly half of the patients and 91 to 100 in the other half. Primary tumors included prostate (33%), lung (29%), breast (23%), kidney (9%), and colorectal (7%). The study did not report primary tumor histology in terms of favorable or unfavorable or bone metastases in terms of complicated or uncomplicated. Tumor metastatic locations included pelvis (40%), sacrum (24%), vertebrae (17%), humerus (7%), and femur (3%). All patients were taking narcotic analgesics at enrolment, 63 percent were receiving chemotherapy, 29 percent bisphosphonates, 28 percent hormonal therapy, and 9 percent immunotherapy. Patients with evidence of spinal cord or cauda-equina compression or pathological fracture were excluded. Treatment with 3D-conformal beams (5 x 20 Gy) or cryoablation (2, 15-minute freezes with a 10-minute thaw in between) were compared. Outcomes were reported at 12 weeks. This study was conducted Italy and rated moderate quality (Appendix F, Table F-2); funding was not reported.

3.3.3.2. Detailed Synthesis

3.3.4.1. Description of Included Studies

One RCT⁶⁵ compared a single dose of 8 Gy EBRT with ibandronate 6 mg in patients with metastatic prostate cancer (Appendix E, Table E-1). Patients could cross over to the other treatment after 4 weeks whether or not they experienced improvement in pain. Twenty-seven percent of patients (128/470) crossed over to the other treatment (23.8% initially treated with EBRT crossed over to ibandronate and 30.6% initially treated with ibandronate crossed over to EBRT, RR 0.78, 95% CI 0.58 to 1.05). The sample size in the trial was 470; median age was 73 years; all were male. Race/ethnicity, comorbidities, or social determinants of health were not reported. Areas of metastases were not reported, although the primary sites of pain were in the abdomen (78%) and thorax (15%). This trial did not report the primary tumor histology in terms of favorable or unfavorable or complicated or uncomplicated. Ninety percent of patients were receiving or had recently received hormone therapy and 3 percent were receiving chemotherapy. Median followup was 11.7 months. This trial was conducted in the United Kingdom and sponsored by the University College London; funding from Cancer Research UK; Roche Products Limited provided ibandronate; the trial was rated fair quality (Appendix F, Table F-1).

3.3.4.2. Detailed Synthesis

3.3.5.1. Description of Included Studies

One NSRI compared EBRT with androgen deprivation therapy in patients with oligometastatic prostate cancer (Appendix E, Table E-2); only patients with bone metastases are included here.¹¹⁰ This study reported only secondary outcomes (no harms) and is summarized in Results Appendix B.

3.4.2.1. Description of Included Studies

One RCT⁸³ and four NRSIs^{109,123,126,137} compared either conventional EBRT or SBRT with surgery to EBRT or SBRT alone for the palliative treatment of bone metastases (Appendix E, Tables E-1 and E-2). Results for EBRT and SBRT are presented separately below. In all five studies, spinal metastases were an inclusion criterion. The RCT and two NRSIs,^109,126 including the SBRT study,¹⁰⁹ also required patients to have MSCC.

One RCT⁸³ included 101 participants with MSCC and a median age of 60 years, 69 percent of whom were male. Race and other patient characteristics were not reported. Lung cancer was the most common primary histology (26%), followed by prostate (19%) and breast cancer (13%). Metastases to nonspine sites were not reported. Many patients were nonambulatory (32%), incontinent (39%), or had spinal instability (38%) at baseline. Surgery for all patients included circumferential decompression, with stabilization including use of cement, metal rods, or bone grafts for those with spinal instability. Patients received EBRT within two weeks after surgery. Total EBRT dose in both treatment groups was 30 Gy (3 Gy x 10). All patients also received dexamethasone. Baseline opioid use was not reported, but post-treatment use was a study outcome. Median followup for surgical patients was 15 weeks, 13 weeks for control patients. The trial was conducted at seven U.S. academic centers, with government funding, and was rated fair quality due to unclear reporting of the following: allocation concealment methods, assessor blinding and attrition (Appendix F, Table F-1).

Three NRSIs of EBRT and surgery (N=534)^123,126,137 in patients with spinal metastases included 67 percent male patients, with mean age 52 years across two studies (not reported for the third). One study included only patients with cancer of unknown origin¹²³ and one only those with primary lung cancer;¹³⁷ most patients (54%) in the third study¹²⁶ also had lung cancer, and 19 percent had cancer of unknown origin. Across two studies, 45 percent of patients had nonspine bone metastases, and 42 percent had visceral metastases (not reported in the third). In one study 36 percent of patients were nonambulatory at baseline, 30 percent in another (Frankel grade C or less); the third did not report baseline motor function. Surgical procedures included decompression and/or stabilization. Total EBRT dose was 30 to 45 Gy in 10 to 20 fractions across two studies, not described in the third; none reported treatment duration. Additional reported treatments included chemotherapy (2 studies), bisphosphonates (1 study), erlotinib (1 study), and dexamethasone (1 study). Followup (mean or median) ranged from 22 to 38 weeks. Two studies were conducted in one of two academic centers in China, the third at multiple centers in the U.S., Europe, and Saudi Arabia. Funding sources were unclear. All three studies were rated fair quality, with methodologic limitations including differences in baseline characteristics (2 studies), unclear methods for patient selection and ascertainment of exposures and outcomes, no blinding reported, and no analysis to control for confounding (1 study) (Appendix F, Table F-2).

One NRSI of SBRT and surgery¹⁰⁹ enrolled 57 patients with 69 metastatic lesions and spinal cord compression. Most results were reported by lesion rather than by patient. Mean age was 60 years, and 43 percent of lesions occurred in males. Race and other patient characteristics were not reported. Primary tumors varied widely, including renal cell (26%), breast (25%), lung cancer (16%), and 14 other histologies. Non-spine metastases and baseline function were not reported. Patients with high-grade SCC underwent surgery with decompression and stabilization. Total SBRT dose was 16 to 30 Gy in 1 to 5 fractions, with duration not reported, and some lesions (43%) had previously been treated with EBRT. Median followup was 43 weeks. The setting was a single U.S. academic center, with funding not reported. The study was rated poor quality (Appendix F, Table F-2): confounding by indication was a concern, as patients with higher-grade MSCC, fracture, or instability were treated with surgery, others with SBRT alone, and the study did not control for confounding. Prior treatment differed across treatment groups, methods for ascertaining exposures and outcomes were unclear, and blinding was not reported.

3.4.2.2. Detailed Synthesis

3.4.3.1. Description of Included Studies

Two RCTs^57,98 compared EBRT with dexamethasone to EBRT alone for the palliative treatment of bone metastases (Appendix E, Table E-1). One RCT⁹⁸ enrolled patients with MSCC and gave dexamethasone to improve motor deficits. The other trial⁵⁷ excluded patients with MSCC and gave dexamethasone to mitigate an adverse effect of EBRT (pain flare).

The first RCT⁹⁸ included 57 patients with MSCC, 63 percent of whom were ambulatory at baseline. Median age was 62 years, and 32 percent of patients were male. Most patients (60%) had primary breast tumors, with gastrointestinal (11%) and prostate (9%) tumors next most common. Non-spine metastases were not reported. Total EBRT dose was 28 Gy, given in seven daily 4-Gy fractions; dexamethasone dose started at 96 mg/day and was tapered off over 1.4 weeks. Previous treatment for epidural metastasis was an exclusion criterion. Followup was 104 weeks or until death (actual followup not reported). The trial was conducted at a single center in Denmark, with nonprofit funding, and was rated poor quality for unclear randomization and allocation concealment methods, differences between groups in sex at baseline, lack of blinding and failure to report attrition (Appendix F, Table F-1).

The second trial⁵⁷ enrolled 298 patients with painful bone metastases but without MSCC or pathologic fracture. Median age was 69 years, and 57 percent of patients were male. Primary tumor histology was lung (28%), prostate (25%), breast (22%) and other solid tumors (25%). Site of metastasis was mixed, with 35 percent spine and 65 percent nonspine; 78 percent of patients had solitary metastases, and 22 percent had multiple metastases. At baseline most patients (55%) had Karnofsky score of 70 to 80, and worst pain score of 7 to 10 (51%). EBRT was given in a single 8-Gy fraction, and dexamethasone dose was 8 mg/day for 5 days (one dose before EBRT and 4 doses after). Patients had narcotics prescribed at baseline, but prior radiation therapy, recent or concurrent systemic steroids, NSAIDs, and planned chemotherapy were exclusion criteria. Median followup was 6 weeks. The trial was conducted at 23 centers in Canada, received government and university funding, and was rated good quality (Appendix F, Table F-1).

3.4.3.2. Detailed Synthesis

3.4.4.1. Description of Included Studies

One RCT¹⁰⁵ and two NRSIs^119,135 (all considered poor quality) compared EBRT with bisphosphonates to EBRT alone for palliation of bone metastases (Appendix E, Tables E-1 and E-2). None of the three studies reported spinal cord compression at baseline.

The RCT enrolled 40 patients, all with bladder cancer.¹⁰⁵ Median age was 54 years (mean not reported), and 78 percent of participants were male. The most common site of metastases was the pelvis (73%), followed by the spine (55%) and the femur, humerus, and ribs (25% for each site). Thirty percent of patients had a single metastasis, 32.5 percent had two or three, and 37.5 percent had four or more. Patients with visceral metastases were excluded. Total EBRT dose was 13 Gy (6.5 Gy x 2 over 24 hours) in 65 percent of patients, and 20 Gy (4 Gy x 5 over 4 days) in 35 percent. Patients were randomized to receive zoledronate (4 mg intravenous) or placebo (i.e., EBRT alone group) given monthly over 6 months. Two patients in the control group received chemotherapy during the study; analgesia was available to all patients, but opioid use was not specified. Eighty percent of patients had prior radical cystectomy. Median followup was 24 weeks. The trial took place at a single academic center in Egypt, and source of funding was unclear. The trial was assessed as poor-quality due to unclear randomization and allocation concealment methods, inadequate data to compare treatment groups at baseline, lack of blinding and failure to report attrition (Appendix F, Table F-1).

Median age was 65 years in one NRSI,¹¹⁹ and not reported in the other.¹³⁵ Across the two studies, 56 percent of participants were male, 41 percent had spine metastases, and 47 percent had visceral metastases. One study included only patients with renal cell carcinoma, who were classified into favorable- (3%), intermediate- (77%), and high (19%) -risk groups.¹¹⁹ In the other NRSI, breast (32%), colorectal (19%) and lung cancer (15%) were most common primary tumor types; 75 percent of bone metastases were osteolytic, and 25 percent osteoblastic.¹³⁵ All patients in the latter study had complete or impending pathologic fractures and were treated with radiation therapy after stabilizing orthopedic surgery. Median total dose was 39 Gy in one study, given in 5 fractions over 5 weeks, and 30 Gy in the other, with 10 fractions in 35 percent and 2 fractions in 19 percent of patients (duration not reported). Zoledronate was given every 3 to 4 weeks at a dose of 4 mg in the study with mixed primary tumor types; the dose was reduced based on renal function in the study of patients with renal cell carcinoma (actual doses not reported). In this study 77 percent of patients had undergone nephrectomy, and 55 percent had systemic treatment in addition to study drugs, including sunitinib, interferon, interleukin-2, sorafenib, everolimus, and temsirolimus. For patients receiving zoledronate, the effect of treatment with sunitinib was also reported. Median followup was 87 weeks in this study, and 39 weeks in the study of patients undergoing orthopedic surgery. Both were single-center studies, one set in Japan, the other in Germany, and both were rated poor quality because of unclear methods for patient selection and for ascertainment of exposures and outcomes, baseline characteristics that differed across groups or were not clearly reported, retrospective design, blinding not reported, and no control for confounding (Appendix F, Table F-2).

3.4.4.2. Detailed Synthesis

3.4.5.1. Description of Included Studies

Three RCTs (reported in 4 publications)^77,78,85,97 compared EBRT plus a radioisotope versus EBRT alone for the palliative treatment of bone metastases (Appendix E, Table E-1). Across the RCTs, sample sizes ranged from 64 to 126 (total N=221). The average study mean age of participants was 70 years (range 67 to 73 years). Two trials^77,78,85 enrolled only males; the proportion of males was not reported in the remaining trial.⁹⁷ None of the trials reported on race or ethnicity, comorbidities, or social determinants of health. In two of the trials, the primary tumor was in the prostate,^77,78,85 and in the remaining trial the primary tumor types were prostate (69%), breast (20%) and other locations (11%).⁹⁷ None of the trials reported the primary tumor histology in terms of favorable or unfavorable. None of the trials reported the location or characteristics of bone metastases, though the inclusion criteria of one of the trials⁸⁵ required multiple bone metastases at enrollment.

EBRT dosage varied among the trials. One trial⁷⁸ utilized a dose-fractionation scheme that included either 8 Gy single fraction or multiple fraction doses of 20 Gy (4 Gy in 5 fractions) over 1 week or 30 Gy (3 Gy in 10 fractions) over 2 weeks. Dosage in the second trial⁸⁵was 30 Gy (3 Gy in 10 fractions) for 14 days or 20 Gy (5 Gy in 4 fractions) for 7 days. In the third trial,⁹⁷ the planned dosage was 30 Gy (3 Gy in 10 fractions), although the study reported that some patients received 8 Gy single fraction; the timing of EBRT delivery was not reported in this trial. None of the trials clearly reported the specific type of EBRT employed but it was most likely 2D or 3DCRT. Radioisotopes used in the studies were radium-223⁷⁸ or strontium-89.^85,97 Concomitant use of NSAIDs or other pain medication was permitted in all three trials, though the proportion of patients using specific treatments was not reported apart from baseline opioid used which was 50 percent in one trial.⁸⁵ One trial⁷⁸ excluded patients with previous radiotherapy and prior therapy was unclear in one trial.⁸⁵ In the third trial, the proportion of previously treated patients was 28 percent for radiotherapy, 12 percent for chemotherapy and 17 percent for hormone therapy.⁹⁷ Duration of followup was 6 months in two trials^85,97 and 2 years in the remaining trial.^77,78

Two trials were conducted in Europe,^77,78,97 and one in Canada.⁸⁵ Two^77,78,85 were multicenter studies and one⁹⁷ was a single center study. Funding (by a private source) was only reported in one study.^77,78

One trial was good quality^77,78 and two were fair quality^85,97 (Appendix F, Table F-1). Limitations of the fair quality studies included unclear randomization and allocation concealment, between-group differences at baseline, and high rates of attrition, which is to be expected in this patient population due to high mortality (range 60% to 78%).

3.4.5.2. Detailed Synthesis

3.4.6.1. Description of Included Studies

One NRSI¹¹² compared EBRT plus cryoablation (use of extreme cold to kill cancer cells) with EBRT alone (Appendix E, Table E-2). Of 175 participants enrolled, 150 were treated with EBRT plus cryoablation (n=25) or EBRT alone (n=125). (The remaining 25 patients were treated with cryoablation alone and are described under Key Questions 3a and 3c.) Patients were matched via propensity score analysis. The mean patient age was 68 years and 49 percent were male. Comorbidities, social determinants of health and race/ethnicity were not reported. Mean Karnofsky score ranged from 70 to 89 in about half of the participants (49%) and 91 to 100 in the other half (51%). Primary tumor types included prostate (33%), lung (29%), breast (24%), colorectal (7%), and renal (6%). The study did not report primary tumor histology in terms of favorable or unfavorable or bone metastases in terms of complicated or uncomplicated. Enrollment criteria required the presence of a single, painful bone metastases, which was present primarily in the pelvis (41%), sacrum (24%), or vertebrae (17%). Spinal cord compression and treatment site fracture were exclusion criteria.

The EBRT dose was 20 Gy (4 Gy x 5) over one week and was delivered 15 days after cryoablation. The specific type of EBRT employed was not reported but was most likely 2D or 3DCRT. Percutaneous cryoablation (−100° C) was delivered in two, 15-minute sessions separated by 10 minutes. Concomitant treatments included narcotic analgesics in 100 percent of participants. Previous treatments were not reported. The duration of followup was 12 weeks.

The study was conducted in Italy, in a single center; no funding source was reported. The study was rated fair quality. Limitations included unclear overall and differential attrition and lack of blinding (Appendix F, Table F-2).

3.4.6.2. Detailed Synthesis

3.4.7.1. Description of Included Studies

One RCT (N=57)⁵⁶ compared EBRT plus hyperthermia (n=29) with EBRT alone (n=28). (Appendix E, Table E-1). The mean age was 58 years and 56 percent were male. Race/ethnicity, social determinants of health, and comorbidities were not reported, although trial inclusion criteria required an ECOG score between 0 and 3 at baseline. The primary tumor type was breast or prostate in 19 percent of the population; tumor type was not reported for the remaining 81 percent of the population. A single bone metastasis was present in 44 percent of the population, and 56 percent had multiple bone metastases. Fifty percent of metastases were in the spine, 28 percent were in pelvic bones, and 21 percent were in the sternum, ribs or extremities. Patients with previous radiotherapy at the metastatic site or with a pathologic fracture requiring surgery were excluded from the trial; presence of spinal cord compression was not reported.

The EBRT dose was 30 Gy (3 Gy x 10 fractions) over 2 weeks. The specific type of EBRT employed was 3DCRT. Hyperthermia was delivered over 2 weeks for a total of four sessions of at least 40 minutes per session using targeted heating (median 559 watts) through the Thermatron R-8 device. Concomitant analgesic use was permitted per inclusion criteria, but specific rate of use was reported. The duration of followup was 12 weeks.

The trial was conducted in Taiwan, in a single academic (university) center; funding was through the university. The RCT was rated poor quality due to numerous limitations that included unclear randomization, allocation concealment, reporting of attrition and loss to followup (Appendix F, Table F-1). The trial initially had a planned 3-year duration and an enrollment of at least 152 patients but was stopped early at 3 months due enrollment difficulties and due to the observed complete response in the EBRT plus hyperthermia group. As a result, the trial was underpowered to detect differences between treatment groups.

3.4.7.2. Detailed Synthesis

3.4.8.1. Description of Included Studies

One RCT⁵² (N=84) compared EBRT plus capecitabine (n=42) with EBRT alone (n=42). (Appendix E, Table E-1). The mean age was 47 years. Sex was not reported, but the trial enrolled only patients with breast cancer so presumably all or most patients were female. Race/ethnicity and social determinants of health were not reported. A single bone metastasis was present in 31 percent of the population, and 69 percent had multiple bone metastases; specific sites were not reported. In addition to bone metastases, 10 percent of patients had lung metastases and 12 percent had liver metastases. Spinal cord compression and treatment site fracture were exclusion criteria.

The EBRT dose was 30 Gy (3 Gy x 10 fractions) for 5 days. The specific type of EBRT employed was not reported. Oral capecitabine (825 mg/m²) was given twice a day at the time of EBRT. The study group randomized to EBRT alone did not receive a corresponding oral placebo. The duration of followup was 12 weeks, and outcomes were assessed at 1, 2, 4, 8, and 12 weeks.

The trial was conducted in Egypt, in a single hospital radiology center; study funding was not reported. The RCT was rated fair quality due to unclear allocation concealment and blinding of outcome assessors and patients (Appendix F, Table F-1).

3.4.8.2. Detailed Synthesis

3.5.2.1. Description of Included Studies

One NRSI¹¹² compared EBRT plus cryoablation (use of extreme cold to kill cancer cells) with cryoablation alone (Appendix E, Table E-2). Of the 175 participants enrolled, 50 were treated with EBRT plus cryoablation (n=25) or cryoablation alone (n=25). (The remaining 125 patients were treated with EBRT alone and are described under Key Questions 3a and 3b.) Patients were matched via propensity score analysis. The mean patient age was 68 years and 50 percent were male. Comorbidities, social determinants of health and race/ethnicity were not reported. Mean Karnofsky score ranged from 70 to 89 in just over half (54%) of the patients and 91 to 100 in the other 46 percent. Primary tumor types included prostate (32%), lung (24%), breast (24%), renal (12%) and colorectal (8%). The study did not report primary tumor histology in terms of favorable or unfavorable or bone metastases in terms of complicated or uncomplicated. Enrollment criteria required the presence of a single, painful bone metastases, which was present in the pelvis (34%), sacrum (26%), vertebrae (17%), rib, humerus, or femur (8% each). Spinal cord compression and treatment site fracture were exclusion criteria.

The EBRT dose was 20 Gy (4 Gy x 5) over 1 week and was delivered 15 days after cryoablation. The specific type of EBRT employed was not reported but was most likely 2D or 3DCRT. Percutaneous cryoablation (−100° C) was delivered in two, 15-minute sessions separated by 10 minutes. Concomitant treatments included narcotic analgesics in 100 percent of participants. Previous treatments were not reported. The duration of followup was 12 weeks.

The study was conducted in Italy, in a single center; no funding source was reported. The study was rated fair quality. Limitations included unclear overall and differential attrition and lack of blinding (Appendix F, Table F-2).

3.5.2.2. Detailed Synthesis

3.5.3.1. Description of Included Studies

One retrospective NRSI¹³⁴ compared strontium-89 plus EBRT (n=53) with strontium-89 alone (n=53) for palliation of multiple bone metastases from primarily lung (36%), breast (27%), prostate (17%), and epipharynx (14%) cancer (Appendix E, Table E-2). The sites of the bone metastases and other characteristics were not reported. Mean patient age was 57 years and 64 percent were male. Strontium-89 148 MBq was given once every 3 to 6 months. One month after strontium injection, patients in the combination group received EBRT 30 to 60 Gy (delivery technique and fractionation scheme not reported) over 2 to 4 weeks. The study was conducted in China at a single center and was rated poor quality because of unclear methods for patient selection and for ascertainment of exposures and outcomes, baseline characteristics were not robustly reported, blinding not reported, and no control for confounding, though primary tumor types were balanced across treatment groups (Appendix F, Table F-2).

3.5.3.2. Detailed Synthesis

3.5.4.1. Description of Included Studies

One fair-quality NRSI (N=60)¹³¹ compared EBRT plus surgical stabilization versus surgical stabilization alone for palliation of pathological (61%) or impending (39%) fractures due to bone metastases (Appendix E, Table E-2). Patients were selected using inpatient International Classification of Diseases (ICD)-9 and surgical codes. Previous RT to the fracture site was an exclusion criterion. Fracture sites included femoral trochanter (33%), femoral shaft (30%), femoral head/neck (28%), humerus (5%), and other (5%). Primary cancer types were breast (33%), lung (23%) and prostate (13%), primarily. Surgical procedures were classified as either fracture fixation (75%) or an arthroplasty reconstruction (25%). Patients in the combined group received EBRT (median dose 30 Gy, fractions not reported) within a median of 14 days postoperatively. An equal number of patients who received the combine treatment had pathologic (51%) or impending (49%) fracture whereas those treated with surgery alone had primarily pathologic fractures (72%). Several other baseline characteristics differed between the two groups: patients in the combined EBRT plus surgery group were younger (58 vs. 64 years) with a greater proportion of females (69% vs. 55%) as well as with lung cancer as the primary tumor type (31% vs. 14%). Patients in the combined group also had better extremity functional status before fracture (Grade 1 [normal, pain free], 88% vs. 58%). Authors performed multivariate logistic regression analyses to control for these imbalances. The study was conducted in the United States at two sites and was rated fair quality because of concerns around the use of ICD-9 and procedures codes to accurately select patients and lack of blinding (Appendix F, Table F-2).

3.5.4.2. Detailed Synthesis

3.6.1.1. Key Points

• Despite the existence of Clinical Practice Guidelines (CPGs) there is great heterogeneity in application of radiation therapy for MBD with relative underutilization of single fraction RT
• Barriers to implementation of CPGs include healthcare professional factors, guideline-related factors and external factors, including healthcare organizations, communities of practice and patients
• Facilitators include CPGs that are accessible and easy to use, commitment to resources needed to support implementation, accessibility to the multidisciplinary care model, incentives toward education and practice improvement for all care team members
• Palliative medicine professionals value an individualized approach to management of serious illness; they did not view the use of CPGs to be inconsistent with high-quality palliative care
• Patients appreciate the opportunity to participate in creation of their radiation treatment plan.

3.6.1.2. Detailed Synthesis

Clinical Practice Guidelines are created with the intent of maximizing quality and consistency of patient care. The benefit of CPGs to clinicians and patients is contingent on effective implementation. Despite the existence of CPGs for palliative radiation therapy of MBD and general consensus that single fraction radiation treatment (SFRT) offers patients equal equivalent pain-relief with increased convenience and decreased financial burden, there is great heterogeneity in application of guidelines with relative underutilization of single fraction radiotherapy.

We identified two reviews^145,146 that provide a broad framework for understanding the barriers and facilitators to implementing CPGs across healthcare. Two other reviews explore barriers and facilitators specific to cancer care and endorse similar themes.^147,148

Barriers are organized into the following categories: healthcare professional factors, guideline-related factors and external factors. Amongst healthcare professional factors, the most significant contributors are knowledge deficits around CPGs and attitude toward practice change, including confidence and motivation. Factors related directly to CPGs include strength of evidence, feasibility of implementation and applicability to real-world patients, particularly complex patients. External factors included those associated with healthcare organizations, communities of practice and patients themselves. Clinicians benefit from strong leadership, a culture of continuous learning, support for interprofessional/multidisciplinary practice and availability of resources. Patient factors were broadly identified by the review articles as lack of knowledge of diagnosis and guidelines as well as lack of trust in or involvement with care team around decisions.

One review¹⁴⁹ sought to understand the attitude toward acceptance of CPGs amongst over a thousand palliative care professionals in Germany. This multidisciplinary cohort included physicians (55.5%) and nurses (30.3%) as well as mental and spiritual health professionals, with 15.2% of physician participants practicing cancer care. This study revealed significant skepticism around the quality of CPGs and concern that they offer a “one-size-fits-all” approach rather than the holistic/patient-centered approach prioritized by palliative care clinicians. Another review¹⁴⁷ similarly identified this concern for “cookbook medicine” amongst cancer care clinicians. Nonetheless, the providers in the first study felt the existence of CPGs for palliative patients was not inherently inconsistent with high-quality palliative care nor palliative care values.¹⁴⁹

Regarding palliative radiation for MBD, two studies looked at patient preference for single versus multi-fraction treatment regimens. One offered education based on the Dutch Bone Metastasis Study to a cohort of patients at a university hospital system in Singapore.¹⁵⁰ They found that 85 percent of these patients chose multi-fraction. Patients attributed this preference to lower rates of re-treatment and decreased frequency of fractures. For the patients that did elect the single fraction regimen, they credited convenience and cost. The other study evaluated a cohort of patients at a Canadian cancer center and, with aide of a decision tool, 76 percent of this group favored a single fraction regimen.¹⁵¹ Though outcome differed, the same main factors were most influential: convenience as the reason for SFRT and risk for fracture supporting MF. Regardless of treatment regimen preference, both studies found that patients strongly appreciated the opportunity to participate actively with choosing their own treatment plan. Some of the variability in these studies may relate to difference between Asian and Canadian populations. We did not find any similar studies conducted in American healthcare setting. In addition to the barriers described above, Technical Expert Panel members described additional potential barriers to implementing clinical guidelines on palliative radiation for patients with MBD, including challenges in the referral process, referring provider unawareness of treatment options, general requirement for treatment at local facilities, and financial incentives.

Facilitators for implementation of CPGs are largely intuitive based on the barriers noted above. Clinicians and patients benefit from CPGs that are accessible and easy to use and benefit from the translation of CPGs into practical decision guides and patient communication tools. From institutions and leaders, facilitative factors include commitment to technology and staff needed to support implementation, accessibility to multidisciplinary care model, as well as incentives toward education and practice improvement for all care team members.

Choosing Wisely is an initiative from the American Board of Internal Medicine Foundation aimed at streamlining access to guidelines and improving conversations between physicians and patients to promote evidence-based and truly necessary beneficial care. There is some discrepancy between recommendations by participating body: for instance, the American Society of Radiation Oncology includes “Don’t routinely use extended fractionation schemes (>10 fractions) for palliation of bone metastases.”¹⁵² In contrast, the Canadian Medical Association and its cancer societies explicitly include: “Don’t recommend more than a single fraction of palliative radiation for an uncomplicated painful bone metastasis,” and the American Academy of Hospice and Palliative Medicine¹⁵³ suggest “Don’t recommend more than a single fraction of palliative radiation for an uncomplicated painful bone metastasis.” One study¹⁹ found for the 196 patients treated with palliative radiation to MBD across the state of Michigan, 7.7 percent received SFRT and of the 70 patients with simple painful bone metastases, 12.9 percent received SFRT. Another publication from Canada¹⁶ found 50.2 percent utilization of SFRT, varying from 60 percent in British Columbia to 31 percent in Saskatchewan. While multiple structural factors likely underlie the variation in recommendations between societies, the variation may reflect opportunity for Choosing Wisely recommendations to facilitate changes in practice.

Among the tools found useful in promoting CPG implementation are the creation of algorithms or clinical pathways as explored by two studies^154,155 Toward that end, one of these studies¹⁵⁴ offers guidance for the creation and implementation of a radiation oncology treatment algorithm based on CPGs as well as patient and disease factors. The other study¹⁵⁵ investigates the impact of an electronic clinical pathway for maximizing guideline-concordant care and significantly finds an increase in appropriate SFRT rates from 18 percent pre-pathway to 48 percent post-pathway. Additional strategies are further discussed in Contextual Question 2.

3.6.2.1. Key Points

• The most effective strategies to promote guideline implementation appear to be the use of online clinical pathways and education-based interventions, particularly when use of peer review/audits were included.
• The least effective appear to be electronic medical record-based interventions; guideline dissemination appears to improve utilization and adherence, but impact may be transient.
• Novel payment care models/incentivized quality metrics provide an intriguing area of research though may benefit from targeting of radiation oncology providers who decide treatment decisions.

3.6.2.2. Detailed Synthesis

We identified 18 studies evaluating strategies intended to improve guideline uptake and adherence (Appendix C, Table C-1). Five evaluated some type of online platform with or without peer review or electronic medical record alert intended to increase use of single-fraction/short-course radiation regimens and reduce use of extended course treatment,^155–159 six focused on provider education followed by peer-review or practice audit,^160–165 two focused on guideline dissemination alone,^26,166 and two described the impact of a payment model or otherwise incentivized implementation of quality metrics.^167,168 One additional publication described clinical algorithm creation, presenting an example case,¹⁵⁴ and another analyzed the influence of physician peer-based groups alone.¹⁶⁹ One additional publication described the implementation of a dedicated palliative radiation oncology service line.¹⁷⁰ Ten studies were conducted in the United States,^{155–159,165,167–170} five in Canada,^{26,162–164,166} and three were conducted in Europe (United Kingdom, Italy, Switzerland).^154,160,161

Briefly, with regard to the five studies evaluating online platforms or electronic medical record alerts, the interventions generally consisted of online care pathways designed to assist providers in deciding between fractionation regimens for palliation of bone metastases.^155–159 Clinical pathways were designed based on a variety of different inputs, including national guidelines, expert input, and literature review as well as prognostication scoring systems. Implementation either did or did not include a component of peer review prior to treatment.

The six studies evaluating education-based interventions utilized educational sessions based on literature review and national guidelines describing indications for single-fraction radiation treatment, often including a component of department-level or provider-level audit of single-fraction radiation therapy rates.^160–165 The two studies focusing on guideline dissemination evaluated use of single-fraction radiation after release or electronic dissemination of clinical practice guidelines.^26,166

Overall, the most effective interventions appear to be online clinical pathways or targeted educational interventions, which generally demonstrated increase in single-fraction or shorter-course radiation regimens/decrease in extended courses of radiation therapy after implementation.^{155,157,158,160,162,163,165} These interventions varied widely in the included components as well as how educational intervention/clinical pathways were constructed, with variation in the degree of impact; degree of change may be influenced by the targeted group/practice environment though with some suggestion of improvement across both academic and community settings. Notably, a key component across effective interventions appears to be a component of peer review or provider auditing. A single study evaluating the impact of prospective peer review alone without pathway/education intervention was effective in changing practice patterns.¹⁵⁸ Furthermore, a single study evaluating electronic medical record-based alerts without peer review found no difference in compliance rates with national quality recommendations.¹⁵⁹

The importance of leveraging peer review may also be reflected in the studies evaluating guideline dissemination alone. Generally, these studies demonstrated that guideline dissemination alone may not provide long-term changes in utilization. One study reports increased use of single-fraction radiation therapy immediately post-guideline dissemination but a return to pre-guideline levels within a few years, with minimal impact to inter-center variations in fractionation use.²⁶ The other study indicated no impact of guideline dissemination alone, with use of single-fraction radiation therapy varying widely between individual providers.¹⁶⁶

Novel payment schemes designed to alter provider incentives in making treatment decisions remain an intriguing area of research, particularly in the United States where alternative payment models are being explored. One study evaluated the impact of the Oncology Care Model, an alternative payment model designed to improve the quality and value of care in oncology practices but found no effect on fractionation patterns;¹⁶⁸ however, this may reflect the design of the model, which focuses on episodes of chemotherapy administration and not directly on radiation oncology providers. A study that provided incentives to facilities meeting quality metrics regarding use of extended courses of radiation therapy for bone metastases found low levels of use followed the intervention.¹⁶⁷ Of note, the development of alternative payment models for radiation oncology providers has been approached in the United States and supported by the American Society for Radiation Oncology, though given debate regarding the components of the model an alternative payment model has not been implemented.

A final study evaluating the impact of a dedicated radiation oncology service line did demonstrate increases in the use of shorter-course radiation therapy regimens.¹⁷⁰ While such programs are gaining in popularity in the United States, implementation may be more restricted to academic settings that promote provider specialization, though community-based programs have been explored.

While these interventions overall suggest clinical pathways may increase uptake of single-fraction or shorter-course radiation regimens in concordance with national guidelines, challenges remain in the ability to handle nuanced individual patient care situations/refinement of necessary components to influence care without increasing unnecessary workflow disruptions. Complete uptake of such regimens may not be clinically appropriate for all patients though this is reflected in national guidelines, which note certain populations that were excluded or less reflected in trial data.

3.6.3.1. Key Points

• We did not identify any study that focused on, measured, or described patient financial distress or toxicity related to EBRT for palliative treatment of MBD.
• Very limited information across three studies suggests that fewer radiotherapy sessions may be less burdensome on patients. None of these studies were conducted in the United States. Given differences in health systems, insurance and care delivery, the applicability of findings from these studies is unclear.

3.6.3.2. Detailed Synthesis

While studies formally evaluating the costs or cost-effectiveness of radiotherapy techniques or fractionation schemes for palliative treatment of MBD were identified, they did so from a health system or payer perspective and did not assess the direct financial impact on patients.

We identified three studies which provided limited information describing patient financial burden by dose/fraction schemes for palliative radiotherapy for MBD. Three are based on RCTs included in this review.^79,94,171 One trial from Canada,⁹⁴ directly asked patients regarding perceptions of financial difficulty. In another study, a subset of patients from the Dutch Bone Metastasis Study answered a questionnaire regarding costs of nonradiotherapy and nonmedical costs as part of a cost-utility analysis.¹⁷¹ An RCT from India, reported on costs of patient travel.⁷⁹ Study and patient information are found in Appendix C, Table C-2.

Only one study directly asked patients about financial difficulty. This fair-quality trial included in this review compared 24 Gy/2 fraction SBRT with 20 Gy/5 fraction conventional EBRT in 299 patients with confirmed spinal MBD who did not have neurological deficit.⁹⁴ Patient perception of financial difficulty was assessed based on the question “Has your physical condition or medical treatment caused you financial difficulties?” included in the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire (EORTC-QLQ-C30, 0-100 scale, higher scores worse difficulty). SBRT was associated with improved perception of financial strain compared with conventional EBRT based on mean change scores at 1 month, (−5.9 vs. 1.5 p=0.03) but the difference was no longer statistically significant at 3 or 6 months and large standard deviations suggest lack of precision (Appendix C, Table C2). More SBRT recipients reported improved perception of financial distress compared with conventional EBRT from baseline to 6 months (35% vs. 22%) and fewer reported worsening of financial distress (15% vs. 29%) with similar proportions of patients in each group reporting stability (50% vs. 48%). Authors conclude that SBRT was associated with improved perception of financial strain versus conventional EBRT and suggest that the finding could reflect more general financial strain in terminally ill patients as well as a differential effect of fewer sessions.

Another study evaluating a subset of patients (N=166) from the Dutch Bone Metastasis Study who completed cost questionnaires suggests that 8 Gy single fraction EBRT may be somewhat less burdensome for patients than 5 fractions of 4 Gy.¹⁷¹ Authors do not clearly delineate which costs are paid by patients or describe financial distress. Estimated costs (in 2002 USD) for radiotherapy (time travel, and out-of-pocket expenses) and for nonmedical costs (time/travel, out-of-pocket, domestic help, paid an unpaid labor) are assumed to be patient costs. SFRT was associated with lower estimated radiotherapy costs for time, travel and out-of-pocket expenses compared with multiple fraction radiation treatment (MFRT) ($134, 95% CI $87-$181 vs. $704, 95% CI $396 to $1012, p<0.001). There was no association between nonmedical costs overall or the individual components with fraction scheme however (Appendix C, Table C-2). Authors speculate that although retreatment was more common with SFRT versus MFRT (25% vs. 7%), most patients may find an extra SFRT less burdensome. Similarly, another included poor-quality RCT from India compared 8 Gy/1 fraction, 20 Gy/5 fractions and 30 Gy/10 fractions (N=60).⁷⁹ Compared with the SFRT, average travel distance and cost per patient were greater in the MFRT schemes. Formal statistical comparison across fractionation schemes was not reported for distance or cost. Authors concluded that the 20Gy/5 fraction may be more economically feasible than the 30Gy/10 fraction scheme and more favorable than the SFRT given lower frequency of re-irradiation (20% for SFRT vs. 5%). The only study conducted in the US retrospectively evaluated the National Cancer Data Base¹⁷² and reported that in both univariate and multivariate analysis, greater distance to treatment was associated with increased odds of SFRT compared with MFRT use but provides no financial information; conclusions regarding patient financial impact are not possible.