Table 1. Abbreviations
Closed for comments This consultation ended on at Request commenting lead permission
Evidence summary
Population and studies description
This interventional procedures overview is based on evidence from 1 systematic review and meta-analysis, 2 RCTs and 9 observational studies. The systematic review included a total of 5,447 people and the RCTs included a total of 236 people, 117 of whom had the procedure. The observational studies included a total of 1,609 people, but there was likely substantial overlap between these patient populations. This is a rapid review of the literature, and a flow chart of the complete selection process is shown in figure 1. This overview presents 10 studies as the key evidence in table 2 and table 3, and lists 17 other relevant studies in appendix B, table 5.
All the studies were done in Europe and almost all included at least 1 centre in the UK. Six of the key studies included only a single centre, 5 of which were based in the UK. There were 4 multicentre studies, at least 3 of which included multiple centres from the UK. Median study follow-up duration ranged from 2 to 7 years. There was substantial heterogeneity in the study populations. The median age ranged from 70 to 81 years. One study reported the mean age of its participants (72 and 74 years for the 2 subgroups, respectively). In all the studies, the population was skewed towards males, which represented between 58% and 73% of all people who were included. There was substantial heterogeneity in terms of the stage of the tumours of the study populations. Most studies included all tumour stages, but overall T2 or T3 tumours were the most prevalent.
The key studies included people who had CXB at different points of the care pathway, for example before or after chemoradiotherapy, as a neoadjuvant or adjuvant treatment and in different doses (see the section on procedure technique). They included people with tumours of different radiological stages. Both RCTs investigated CXB as a boost, but one of them compared CXB plus EBRT with EBRT alone, whereas the other had a more complex intervention arm dependent on the tumour size. Table 2 presents study details.
Study no. | First author, date country | Characteristics of people in the study (as reported by the study) | Study design | Inclusion criteria | Intervention | Follow-up |
|---|---|---|---|---|---|---|
1 | Baron 2025 Europe (17 centres) | N=148 [69 had a boost with EBRT at 9 Gy in 5 fractions (Arm A) compared with 72 having a boost with CXB at 90 Gy in 3 fractions (Arm B)] Age Arm A: 69 (61 to 74), Male sex Arm A: 65%, T2 lesions Arm A: 64%, N0 status Arm A: 71%, Distance from anal margin less than 6 cm Arm A: 77%, | RCT (NCT025-05750) | Adults (18 or older) with biopsy proven adenocarcinoma with a cT2, cT3a or T3b tumour up to 10 cm from anal verge, less than 5 cm in diameter, and less than half the rectal circumference. They also had cN0 to cN1 disease (with lymph node less than 8 mm), no metastases and ECOG performance status of 0 or 1 and were fully operable | Boost with CXB at 90 Gy in 3 fractions (before or after chemo-radiotherapy if the tumour was less or more than 3 cm, respectively) the comparator arm was boost with EBRT only | 5 years |
2 | Ortholan 2012 France | N=88 (45 CXB and EBRT compared with 43 EBRT alone) Radiological stage: T2 (n=22), T3 (n=62), unknown (n=4) Nodal stage: NO (n=37), N1 (n=46), unknown (n=1) Differentiation: well (n=39), moderate (n=42), poor (n=1), unknown (n=6) CXB plus EBRT group: median age 69 years (40 to 92 years); 62% (28 out of 45) male EBRT-alone group: median age 67 (29 to 79) years; 67% (29 out of 43) male | RCT | People with histologically confirmed adenocarcinoma of the lower rectum (located within 6 cm of the anal verge), classified as T2 or T3 by endorectal ultrasonography and involving less than two-thirds of the rectal circumference; people with no signs of distant metastases | CXB Total dose of 85 Gy delivered in 3 fractions: 35 Gy, 30 Gy and 20 Gy on days 1, 8 and 21, respectively | 10 years |
3 | Dhadda 2021 UK (2 centres), France (1 centre) | N=194 Median age: 69 (36 to 91) years Male sex: 65% Tumour staging: T1 (n=143), | Retrospective cohort study | People for whom standard therapy was unsuitable or refused; people with no signs of metastatic disease | Adjuvant CXB following local excision | 77 months |
4 | Than 2024a UK (4 centres), Sweden (1 centre) | N=76 Median age: 78 (67 to 84) years Male sex: 65% Tumour staging: T1 (n=27), | Retrospective cohort study | People with stage 1 (T1 or T2-N0-M0) rectal cancer for whom surgery was unsuitable for or declined | CXB Total dose of 90 Gy given in 3 fractions over 4 weeks, with a fourth 20-Gy dose (total 110 Gy) to selected people | 26 months |
5 | Steinke 2023a UK (4 centres), Sweden (1 centre) | N=258 (N=226 who underwent CXB and short-course radiotherapy as primary treatment, and N=32 undergoing this immediately after local excision) Median age: 81 (49 to 103) years Male sex: 69% Tumour staging: T1 or T2 (n=182), T3 or T4 (n=72) | Retrospective cohort study | People who underwent CXB and short-course radiotherapy | CXB boost (ranging from 30 Gy to 120 Gy) after short-course radiotherapy | 24 months |
6 | Than 2024b UK (1 centre) | N=328 (N=224 at low or intermediate risk and N=104 at high risk) Median age: 73 (62 to 80) years Male sex: 73% Tumour staging: T1 (n=1), | Retrospective cohort study | People who underwent (chemo)radiotherapy and CXB, including short- and long-course radiotherapy | CXB boost Total dose of 90 Gy given in 3 fractions over 4 weeks, with a fourth 20-Gy dose (total 110 Gy) to selected people | 33 months |
7 | Steinke 2023b UK (1 centre) | N=193 Median age: 73 (33 to 103) years Male sex: 73% Tumour staging: T1 (n=27), | Retrospective cohort study | People who had at least 1 fraction of CXB | CXB Total dose up to 110 Gy in up to 4 fractions | 31 months |
8 | Than 2024c UK (1 centre) | N=56 Median age: 76 (45 to 91) years Male sex: 71% Tumour staging: T2 (n=11), T3 or T4, N0 (n=14), T3 or T4, N1 or N2 (n=31) Early-stage regrowth (82%), late-stage regrowth (18%) | Retrospective cohort study | People who had CXB as salvage therapy for local rectal cancer regrowth after a watch-and-wait approach | CXB Total dose up to 110 Gy in 3 to 4 fractions over 4 to 6 weeks | 37 months |
9 | Than 2024d UK (1 centre) | N=251 (N=103 starting with EBRT and N=148 with CXB) Mean age (before adjustment): Male sex (before adjustment): EBRT first: 73% | Retrospective analysis with propensity matching and IPTW | People with rectal adenocarcinoma (cT1 to cT3, N0 to 1, M0, grade 1 to 2, size of 3 cm or less) who had both EBRT and CXB, irrespective of treatment sequence | CXB Total dose up to 110 Gy in 3 to 4 fractions over 4 to 6 weeks | EBRT first: 37 months |
10 | Sun Myint 2017 UK (1 centre) | N=200 Median age: 74 (32 to 94) years Male sex: 67% Tumour staging: cT1 (n=21), cT2 (n=89), cT3 (n=87), cT4 (n=3) | Retrospective cohort study | People with histologically proven, well to moderately well differentiated residual rectal cancer after EBRT for whom surgery was not suitable for or who refused surgery; tumour situated less than 12 cm from the anal verge | CXB boost Most people had no more than a total dose of 90 Gy delivered in 3 fractions every 2 weeks in 4 weeks | 2.7 years |
11 | Powell 2025 Multiple countries | N=5,447 There is likely significant overlap with other studies included in this overview. The following studies reported in this systematic review have also been included separately in this overview: Baron (2025) Than (2024a) Than (2024b) Than (2024d) Steinke (2023a) Steinke (2024b) Sun Myint (2017) | Systematic review and meta-analysis | Randomised and non-randomised studies reporting original data regarding outcomes of people whose rectal cancer was treated with CXB | CXB using 50 kVp with a radiation dose of 80 to 110 Gy, where each session is 2 weeks apart over 4 to 6 weeks | |
12 | Than 2025 UK (1 centre) | N=53 Median age: 71 (64 to 77.5) Male sex: 64% Tumour staging: Stage 1: 16 (30%) Stage 2: 15 (28%) Stage 3: 19 (36%) Tx-Nx-Mx: 3 (6%) | Prospective observational study | All consecutive patients who were referred to the Clatterbridge Cancer Centre, Liverpool, from January to October 2023, for CXB treatment either before or after (chemo)radiation, were eligible to participate. People who had CXB as postoperative adjuvant treatment following local rectal cancer excision were excluded | N/A | 6 and 12 months |
Procedure technique
Nine of the 12 studies detailed the device used, although the device could be inferred with high likelihood in the remaining 3 studies. The majority of the studies used the Papillon-50 machine (Ariane Medical Systems) (Baron 2025, Dhadda 2021, Than 2024a, Than 2024b, Than 2024c, Than 2024d). The Papillon-50 machine was also likely used in Steinke 2023a and Steinke 2023b. Sun Myint (2017) reported using the Papillon-50 machine after 2009 and the Therapax machine (Pantak) between 1993 and 2009. Two studies reported using the RT50 machine (Philips) (Ortholan 2012 and Powell 2025). All of the devices were reported to deliver a beam of 50 kV X-rays (half-value layer 0.64 mm Al, 2.7 mA). Experts informed NICE that a new Papillon Plus machine has recently become available. However, this new device was not used in any of the included studies.
Most studies reported details about the procedure. When reported, the procedure was delivered exclusively in an outpatient setting. The total dose varied but in most studies it was 85 to 90 Gy delivered in 3 to 4 fractions in 4 to 6 weeks. Some people had up to 110 Gy (Than 2024a, Than 2024b, Than 2024c, Than 2024d). In the study by Steinke (2023a) the total CXB doses ranged from 30 Gy to 120 Gy. People who had CXB after local excision of the tumour had a dose of 60 Gy in 2 fractions (Dhadda 2021).
Efficacy
Clinical response
Most studies reported outcomes related to clinical response. cCR was reported in 9 of the studies. Two studies also reported nCR and 2 studies reported a combined cCR and nCR. One study also reported partial response and another study reported clinical incomplete response.
Than (2024a) observed a cCR in 82% of people at a median follow-up of 14 months. The cCR was statistically significantly higher in those with T1 tumours than those with T2 tumours (93% compared with 76%; p=0.05). The cCR rate was also higher among people who were fit but refused surgery than in those who were at high risk for surgery and those with history of prior pelvic radiotherapy (92% compared with 81% and 70%; p=0.17). The study population was 76 people. Steinke (2023a) reported a cCR rate of 74% in a cohort of 258 people after a median follow-up of 2 years. The nCR rate in the same cohort was 7%. In this study, the combined cCR/nCR rate was statistically significantly lower in the primary treatment CXB subgroup (78%) than in the CXB after local excision subgroup (97%; p=0.01). The combined cCR/nCR rate was also statistically significantly higher in the CXB first subgroup (85%) than in the EBRT first subgroup (72%, p=0.03).
After a median follow-up of 33 months, Than (2024b) observed a higher cCR rate among people who were at low or intermediate risk (78%) than among people who were at high-risk (73%), but the results were not statistically significant.
Steinke (2023b) reported cCR rates of 82% among people having CXB with radical treatment intent, 100% among people having CXB after local excision, 40% among people having CXB as treatment of recurrent disease after previous EBRT and 83% among people having CXB as treatment of recurrent disease after no previous EBRT. nCR was achieved in 3%, 0%, 20% and 0% of the subgroups, respectively. Partial response was achieved in 4%, 0%, 10% and 0% of the subgroups. The second, third and fourth subgroup included a small number of people.
Ortholan (2012) reported a higher cCR rate at 5-weeks among people who had CXB along EBRT than among people who had only EBRT (26% versus 3%). The study included people who had T2 or T3 tumours only.
Sun Myint (2017) observed a cCR rate of 72% in a cohort of 200 people having CXB as a boost after a median follow-up of 2.7 years. The clinical incomplete response rate in the same cohort was 28%.
Than (2024c) reported a 6-month cCR rate of 57% among 56 people who had CXB as salvage therapy for local rectal cancer regrowth after a watch-and-wait approach. In the same study the authors observed a combined cCR/nCR rate of 61% among 46 people with early-stage regrowth. But the combined cCR/nCR rate was 40% among 10 people with advanced stage regrowth.
Than (2024d) compared the cCR rate among 103 people who had EBRT first and 148 people who had CXB first. The rate was higher (but not statistically significantly different) in the CXB first group in both the unadjusted analysis (79% compared with 88%; p=0.07), propensity-matched analysis (78% compared with 85%; p=0.39) and IPTW analysis (86% compared with 88%; p=0.57).
Ortholan (2012) also reported the rates of clinical response greater and less than 50% at 5 weeks. Clinical response greater than 50% rate was higher in the CXB plus EBRT group (69%) than in the EBRT-alone group: (67%) and the difference was statistically significant (p<0.001). Conversely, the clinical response less than 50% rate was lower in the CXB plus EBRT group (5%) than in the EBRT-alone group (31%) (no p-value stated).
Powell (2025) included 24 studies (N=2,365) reporting cCR. The pooled rate of cCR was 82% (95% CI 76% to 88%). Meta-analysis of two RCTs included in this systematic review showed a significantly higher rate of cCR when CXB has been combined with EBCRT compared to EBCRT with EBRT boost, with an overall odds ratio of 7.89 (95% CI 3.31 to 18.83, n=222, p <0.01). 17 studies (N=668) reported cCR rate according to T-stage. Early-stage disease has a higher rate of cCR compared to more advanced disease, with pooled cCR rates of 91%, 76% and 75%, respectively for T1, T2, and T3 disease. Increased cCR rates were also noted with the addition of concomitant therapies.
Organ preservation
Six studies reported outcomes related to organ preservation. Three of them reported an organ preservation rate. Baron (2025) compared the 5-year organ preservation rate between people randomised to have an EBRT boost or a CXB boost. The organ preservation rate was statistically significantly lower for people who had the EBRT boost (56% compared with 79%; HR 0.4, p=0.003). Than (2024d) found the 3-year organ preservation rate higher in people who had CXB first compared with people who had EBRT first. The organ preservation rates were 69% versus 75% (unadjusted analysis), 70% versus 75% (propensity-matched analysis) and 73% versus 80% (IPTW analysis). But, the difference was not statistically significant in any of the analyses. In another study, Than (2024b) found a 3-year organ preservation rate lower in people at low or intermediate surgical risk compared with those at high risk, but the difference was not statistically significant (80% versus 87%; p=0.25).
Dhadda (2021) reported an organ preservation rate of 95% among 194 people who had adjuvant CXB after local excision of a tumour across 3 centres with a median follow-up of 77 months.
Ortholan (2012) reported 4 outcomes relevant to organ preservation in the small RCT in France. The authors observed more sphincter saving procedures in the CXB plus EBRT group than in the EBRT-alone group (76% compared with 44%). They also reported fewer people needing a colostomy in the CXB plus EBRT group than in the EBRT-alone group (31% compared with 63%) and less people needing an abdominoperineal resection in the CXB plus EBRT group than in the EBRT-alone group (24% compared with 56%). It was not reported if the results were statistically significantly different. The authors observed a lower actuarial colostomy rate in the combined treatment group (29% compared with 63%; p<0.001). Finally, Steinke (2023a) reported a permanent stoma rate of 6% in a cohort of 258 people after a median follow-up of 2 years. There was no statistically significant difference in the rate between people who had CXB and short-course radiotherapy as primary treatment those who had the procedure after local excision of a tumour.
Powell (2025) reported consolidated findings from 15 studies (N=1,133), 80% (95% CI 73% to 87%) of people whose rectal cancer was treated with CXB ultimately did not require resection of the rectum.
Survival
Eleven of the 12 studies reported OS estimates. The longest follow-up was reported by Ortholan (2012) who observed a nearly identical 10-year OS among people who had CXB and EBRT and people who had EBRT only (55% and 56%, respectively) in their small RCT. In the larger and more recent RCT, Baron (2025) observed no statistically significant difference in the 5-year OS between people randomised to have EBRT or CXB (91% versus 92%; p=0.5). Than (2024a) observed OS of 97% (1-year), 75% (3-year) and 58% (5-year) among 56 people with stage I cancer. OS did not statistically significantly differ for people with T1 compared with people with T2 cancer, but was statistically significantly lower in people who were at high risk for surgery (HR 2.54, p=0.02) and those with history of prior pelvic radiotherapy (HR 2.75, p=0.03) than in those who were fit but refused surgery.
Steinke (2023a) reported OS at 2 years, 3 years and 5 years, separately for people who had CXB as primary treatment and people who had CXB after local excision. The OS rates were 67% and 86%, 55% and 64%, 33% and 32%, respectively. The study included 4 centres. In a single centre analysis, Steinke (2024b) reported OS among people who had CXB as treatment with radical intent only. The OS at 2 years, 3 years and 5 years was 81%, 77% and 65%, respectively. Similarly, in a single centre Sun Myint (2017) recorded OS at 2 years, 3 years and 5 years for people having CXB after EBRT. The observed rates were 88%, 82% and 64%, respectively. Than (2024d) investigated whether the OS differed for people in a single centre dependent on the sequence of treatment. OS was higher for people who had CXB first. The magnitude was high in the propensity-matched analysis, but small in the unadjusted and the IPTW analyses. The difference was statistically significant in the multivariable analyses in all 3 cases.
Another single-centre study observed no statistically significant difference between the 5-year OS of people at low or intermediate surgical risk and people at high surgical risk (67% compared with 64%; p=0.88) (Than 2024b).
Than (2024c) observed OS of 100% (1 year), 82% (3 years) and 65% (5 years) among people who having CXB as salvage therapy for local rectal cancer regrowth. Six-year OS among people having adjuvant CXB following local excision was 81% according to Dhadda (2021).
Eight of the studies also reported disease-free survival. The longest follow-up was again reported by Ortholan (2012) who observed no statistically significant difference for people in the CXB plus EBRT and EBRT-only group (10-year disease-free survival of 53% compared with 54%; p=0.99). Than (2024a) observed disease-free survival of 80% (1-year), 70% (3-year) and 66% (5-year) among 56 people with stage I cancer; it was not statistically significantly different for people with T1 compared with T2 cancer or based on the reason for not having surgery. Disease-free survival was similar in the primary treatment CXB subgroup observed by Steinke (2023a): 75%, 63% and 60% at 2, 3 and 5 years, respectively. But survival was higher in the CXB after local excision subgroup: 93%, 93% and 93% for the same timepoints. The number of people in the latter subgroup was notably smaller. In another study Steinke (2023b) reported the disease-free survival for people having CXB with radical intent only; the observed survival probabilities were 90% at 2 years, 85% at 3 years and 75% at 5 years. The authors also reported relapse-free survival probabilities which were slightly lower: 85% at 2 years and 79% and 3 years. Sun Myint (2017) observed slightly lower disease-free survival at the same intervals: 72% (2 years), 65% (3 years) and 53% (5 years). The authors also reported very similar local PFS: 74% (2 years), 66% (3 years) and 52% (5 years).
Than (2024d) found disease-free survival to be higher for people who had CXB first rather than EBRT first in both the unadjusted, propensity-matched and IPTW analyses, but the results were not statistically significant in either the univariable or multivariable analyses. In another study, the author observed 5-year disease free survival of 62% for people at low or intermediate surgical risk and 64% for people at high surgical risk: 64% (Than 2024c). The difference was not statistically significant (p=0.46).
In their small retrospective study Than (2024c) recorded lower disease-free survival among people having CXB as salvage therapy: 69% (1 year), 51% (3 years) and 51% (5 years).
Instead of disease-free, the RCT by Baron (2025) reported DSS. DSS at 5 years was found to be statistically significantly higher among people randomised to the CXB group (73%) than that of people in the EBRT group (46%; p=0.003).
The reporting of survival outcomes within the included studies in Powell (2025) was inconsistent, primarily because of differences in follow-up durations. The most commonly cited survival outcomes were 3- and 5- year DFS and OS. A pooled analysis from 6 studies including 580 people revealed a 3-year DFS rate of 81% (95% CI 73% to 89%) and a 5-year rate of 69% (95% CI 60–78%) from 8 studies involving 781 people. Regarding OS, the data presented include a 3-year rate of 85% (95% CI 61% to 99%) from 3 studies with 175 people, and a 5-year rate of 73% (95% CI 65% to 82%) from 11 studies with 963 people.
Disease regrowth
Studies included data on local regrowth. Baron (2025) provided evidence for statistically significantly lower local regrowth at 5 years in people randomised to CXB compared to people having EBRT (16% versus 33%; p=0.02). Local regrowth after initial cCR was observed in 11% of people by Sun Myint (2017) after a median follow-up of 2.7 years. Than 2024a reported local regrowth rates at a median follow-up of 14 months. They were 18% for all people, 16% for people with T1 cancers only and 19% for people with T2 cancers only. The difference between the latter subgroups was not statistically significant (p=0.95). In a single-centre study the same author observed local regrowth rates of 17% and 22% for people at low/intermediate risk and high risk, respectively (but the difference was not statistically significant; p=0.41). The local regrowth rates were not statistically significantly different for people who had EBRT or CXB first in either the unadjusted, propensity-matched or IPTW analyses by Than (2024d).
The small retrospective study of people who had CXB as salvage therapy for local rectal cancer regrowth after a watch-and-wait approach observed further early-stage local regrowth after a median follow-up of 39 months in 21% of people (Than 2024c). Further advanced-stage local regrowth was observed in 2 out of 4 people (50%).
Data on local regrowth was available for 28 studies (N=2,963) included in Powell (2025). The pooled local regrowth rate was 20% (95% CI 15% to 25%). The results of this paper appear to suggest that EBCRT, when combined with CXB, appears to have a protective effect against the risk of local regrowth with a pooled estimate of 13% (95% CI 10% to 17%, n=494) for studies using CXB and EBCRT, compared to 24% (95% CI 16% to 32%, n=1,424) for studies using CXB and EBRT, and 22% (95% CI 12% to 33%, n=1,045) for CXB alone.
Disease recurrence
Local recurrence was reported in 3 studies. The longest follow-up was available from the small RCT by Ortholan (2012). The local recurrence rate (Kaplan–Meier estimate) at 10 years was 10 in the CXB plus EBRT group and 15% in the EBRT-alone group. The difference was not statistically significant (p=0.69). Steinke (2023a) only reported local relapse in people who had achieved cCR. The respective rates were 16% in the CXB as a primary treatment subgroup and 3% in the CXB after local excision subgroup. The difference was highly statistically significant (p<0.001). The multi-centre study by Dhadda (2021) reported a crude local relapse at a median follow-up of 77 months of 7.7%. The local relapse rate (Kaplan–Meier estimate) was 9% at 6 years for people having adjuvant CXB following local excision. The local recurrence rate was lower for people who had CXB prior to EBRT (4.5%) than for those having EBRT prior to CXB (15%). The difference was statistically significant (p=0.037), but based on a small number of events.
In addition to the local relapse rate, Steinke (2023a) also reported that the median time to local relapse was 14 months. It was not statistically significant difference between subgroups (people who had CXB as a primary treatment and people who had CXB after local excision).
Three studies also estimated regional relapse. With a median follow-up of 2 years, Than (2024a) observed regional relapse in 2 out of 76 people with stage I cancer, which represented 3% of the study population. In a larger single-centre study the authors noted regional relapse in 1% of people who were at a low or intermediate surgical risk and 3% of people who were at high risk. The median follow-up was 33 months and the difference was not statistically significant (p=0.33).
One study reported nodal relapse, whereas another reported combined regional/nodal relapse. Nodal relapse was statistically significantly lower among people at low/intermediate risk compared with people at high risk (2% versus 6%; p=0.05) in the single centre, retrospective cohort study by Than (2024b). In another single centre, retrospective study by Than (2024d), nodal/regional relapse was higher among people who had EBRT first in both the unadjusted analysis (4% versus 3%; p=0.60) and propensity-matched analysis 3% versus 1%; p=1.00). The rates were the same in the IPTW analysis (3% versus 3%; p=1.0).
Seven studies included data on distant recurrence. In the recent multicentre RCT, Baron (2025) observed similar rates of distant metastases among people randomised to EBRT (14%) and people randomised to CXB (13%) at 5 years. Similarly, another small RCT found very similar rates of distant recurrence at 10 years in people who had CXB with EBRT (27%) and people who had EBRT only (26%; Ortholan 2012). This trend was also observed by Than (2024d). Distant relapse was lower in the CXB first group and higher in the EBRT first group, but the difference was not statistically significant in the unadjusted, propensity-matched or IPTW analyses.
The single-centre retrospective study by Sun Myint (2017) reported distant metastases in 8% of all cases at the end of study period, with a median follow-up of 2.7 years. In the small retrospective study by Than 2024a distant relapse occurred in 2 people (3%) only. The study population was only people with stage 1 (T1 or T2-N0-M0) rectal cancer and the median follow-up was slightly above 2 years (26 months).
Steinke (2023a) reported a rate of distant metastasis of 3% across 5 centres after a median follow-up of 2 years. The authors found no statistically significant difference between the 2 subgroups (people having CXB as primary treatment and people having CXB after local excision). But, in a large cohort from a single centre, distant relapse occurred statistically significantly less often in people at low or intermediate surgical risk than people at high surgical risk (11% compared with 21%; p=0.01).
The single-centre study by Dhadda 2021 reported distant metastases in 9% of people who had adjuvant CXB following local excision after a median follow-up of 77 months.
Local control
Two small retrospective studies reported rates of local control. The actuarial local control rate in 76 people with stage 1 rectal cancer was 84% (90% for people with T1 cancer and 80% for people with T2 cancer; Than 2024a). The local control rate did not statistically significantly differ based on the reason for not having surgery. Than (2024c) only reported the local control rates per subgroup. The rates were 79% among 28 people with early-stage tumour regrowth and 50% among 2 people with advanced stage tumour regrowth.
Bowel function
Only 1 study assessed bowel function (Baron 2025). The authors reported a slightly higher LARS score in people who were randomised to have an EBRT boost (24%) than in people who had a CXB boost (17%), but the difference was not statistically significant.
Safety
Mortality
Four studies contained data on mortality. Baron (2025) observed 7 deaths in the EBRT group and 5 in the CXB group at 5 years. Nine of those were cancer related. The small French RCT comparing CXB and EBRT with EBRT alone recorded 10-year mortality rates of 24% in the CXB plus EBRT group and 28% in the EBRT-only group (Ortholan 2012).
Steinke (2023a) reported that nearly half of 258 people who had CXB across 5 centres were dead at the latest follow-up timepoint (this was not reported; the median follow-up was 2 years). The number was extracted from a table with results, but the authors did not interpret it in the study.
Sun Myint (2017) reported only CXB-related deaths in a retrospective analysis of a single-centre cohort of 200 people. No CXB-related deaths were observed.
Rectal bleeding
Five studies recorded rectal bleeding. This outcome occurred in 17% of people randomised to EBRT and 64% of people randomised to CXB in the OPERA trial (Baron 2025). But the bleeding presented as CTCAE grade 1 or 2 only. Similarly, Than (2024d) observed higher rates of rectal bleeding in the CXB first group (32%) than in the EBRT first group (18%). The difference was statistically significant (p=0.01). Sun Myint (2017) observed grade 1 bleeding due to telangiectasia in 56 out of 200 people (28%) in a single centre. Than (2024b) recorded rectal bleeding in 18% of people. Most of the cases were grade 1 or 2, but grade 3 bleeding, which required blood transfusion, occurred in 3 people (1%). Slightly higher rates were observed by Than (2024a) in the small multicentre study of people with stage 1 cancer. Grade 1 or 2 rectal bleeding occurred in 26% of all people and grade 3 bleeding in 4%.
Fourteen studies (N=1,119) in Powell (2025) reported rectal bleeding. Overall, the rate was moderate, at 24% (95% CI 13% to 35%). Most rectal bleeding was mild and self-limiting. 7% of people required blood transfusion, and 6% underwent argon beam therapy to control rectal bleeding.
Proctitis
Proctitis was recorded by Baron (2025), Than (2024a) and Than (2024c). The outcome occurred in 6% of people randomised to the EBRT group and 13% of people randomised to the CXB group. A similar rate (9%) was observed by Than (2024a) among 76 people with stage 1 cancer.
Than (2024c) noted that proctitis symptoms (erratic bowel habits) as acute reactions and late rectal bleeding occurred in 10 out of 56 people (18%) who had CXB as salvage therapy for local rectal cancer regrowth after a watch-and-wait approach.
In Powell (2025) the overall occurrence of proctitis was moderate at 17% (95% CI 12% to 22%). But most cases were mild, with severe grade 3 or higher proctitis occurring in only 7% of people.
Anecdotal and theoretical adverse events
Expert advice was sought from consultants working in the field. They were asked if they knew of any other adverse events for this procedure that they had heard about (anecdotal), which were not reported in the literature. They were also asked if they thought there were other adverse events that might possibly occur, even if they had never happened (theoretical).
They listed the following anecdotal adverse events:
rectal telangiectasia.
They listed the following theoretical adverse events:
rectal stenosis/fistula
rectal perforation
infection
diarrhoea
tenesmus.
Thirteen professional expert questionnaires for this procedure were submitted. Find full details of what the professional experts said about the procedure in the specialist advice questionnaires for this procedure.
Quality of life and patient reported outcomes
In Than (2025), QoL was assessed in 53 people who had CXB for various clinical indications, with 51, 47, and 42 remaining at the end of treatment, 6-month and 12-month follow-ups, respectively. Overall, symptom and functional scores from EORTC-QLQ-CR29 remained stable throughout the follow-up period. Significant improvements were observed in abdominal pain, flatulence, urinary frequency and body weight at 12 months. HADS and EQ-5D-3L scores remained stable, while EQ-VAS scores showed improvement, indicating a good overall quality of life following CXB treatment.
Validity and generalisability
Only 2 of the studies were designed to include random assignment. The remaining studies were retrospective cohort analyses. But some of those compared subgroups; 1 of the studies (Than 2024d) performed statistical adjustment of those subgroups to improve comparability.
There is a lack of long-term follow-up, with the longest available being 10 years, as reported in Ortholan (2012). But this was somewhat different from the other studies (see below). The next longest was 77 months, and the rest of the studies with follow-up details are between 1 and 5 years' follow-up.
11 out of 12 studies were fully or primarily from the UK. 1 small RCT was conducted in France (Ortholan 2012).
Ortholan (2012) produced results that differed somewhat from those observed in the other studies included in this review. This may be because of the use of older data, a different device, or a higher proportion of stage T2 or T3 people in the study population.
There may be overlap in the populations of some studies, in particular the retrospective cohort studies and the systematic review. The potential for selection bias was also noted by expert advisers.
The evidence is highly heterogeneous in terms of the position of CXB in the treatment pathway, and therefore in the type of care received before and along with CXB. Several of the studies have investigated CXB as immediate neoadjuvant treatment, with or without EBRT. A minority of studies included people having CXB as an adjuvant treatment.
There was notable heterogeneity in terms of the stage of the tumours of the study populations. Most studies included all tumour stages, but overall T2 or T3 tumours were the most prevalent. Some studies based inclusion or exclusion criteria on tumour stages. Experts highlighted that this is a key consideration as there is variation in how tumour stages are defined. They agreed that cT1 or cT2 tumours can be classified as early stage. But, they disagreed with the previous definition of T3 or T4 tumours as locally advanced in NICE guidance IPG659. One expert adviser said that most clinicians will consider a T3 tumour as locally advanced only when the circumferential resection margin is involved. A T4 with N2 disease would also be considered locally advanced. In other words, T3 cancers without circumferential resection margin involvement would be considered early stage and immediate treatment with CXB would be appropriate. This definition of early-stage rectal cancer is consistent with the one used in the OPERA trial (Baron 2025).
The size of the tumour is an important factor, which may affect treatment choice. Expert advisers gave different upper limits for the maximal tumour size for which CXB is safe and effective. Those upper limits ranged from 3.5 cm to 6 cm. But, they advised that the procedure should only be considered for tumours less than 3 cm if CXB is used as a boost (after there is significant shrinkage of the tumour following EBRT).
Expert advisers also highlighted that there is a learning curve associated with the procedure. But the majority of the key evidence is from centres with a long expertise in CXB, so the learning curve is unlikely to be a confounding factor.
There was no evidence on important efficacy outcomes, including rectal perforation and rectal necrosis. The evidence was limited for other outcomes such as radiation toxicity and rectal ulceration.
There is a paucity of evidence available to validate the QoL and patient reported outcomes outlined in this overview. Larger prospective studies with homogeneous study populations and extended follow-up periods could be used to fully assess functional outcomes following CXB treatment.
Powell (2025) included multiple older studies from as early as 1974. This may limit generalisability to the modern clinical context. New diagnostic approaches, surveillance and administration of adjuvant therapies introduces significant heterogeneity and may impact outcomes.
Only 2 of the studies included in Powell (2025) are of a randomised nature.
Ongoing trials
NCT02505750: A multicentre, open-label, phase 3 RCT to evaluate a boost of EBRT with a boost with CXB in adults with cT2, cT3a or cT3b adenocarcinoma of low-mid rectum. The study is known as the OPERA trial; 5-year results are included in the key evidence. Sponsor:Centre Antoine Lacassagne, France. Estimated completion: 2030.
NCT05772923: A prospective phase 2 feasibility study to evaluate CXB compared with extension of the waiting interval with or without local excision in people with rectal cancer with a good clinical response after neoadjuvant (chemo)radiation. Sponsor: The Netherlands Cancer Institute. Estimated completion: 2029.
NCT06402864: A feasibility phase 3 trial to evaluate total neoadjuvant treatment and a CXB boost in people with T2 or T3 tumours. Sponsor: The Gustave Roussy, Cancer Campus, Grand Paris. Estimated completion: 2030.
How are you taking part in this consultation?
You will not be able to change how you comment later.
You must be signed in to answer questions