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

    The content on this page is not current guidance and is only for the purposes of the consultation process.

    Evidence summary

    Population and studies description

    This interventional procedure overview is based on 1,593 people from 5 RCTs (including a pilot RCT), 1 non-randomised comparative study and 1 systematic review and meta-analysis. Of these 1,593 people, 440 people had the RFA procedure after removing duplications. 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 7 studies as the key evidence in table 2 and table 3, 1 Health Technology Assessment in the existing assessment, and lists other relevant studies in table 5.

    Of the 6 primary studies, 3 studies were done in China, 2 in Korea and 1 in Germany. The follow-up duration ranged from 4 months to 32 months in 5 RCTs, and the longest duration was around 7 years in 1 observational study.

    Primary RFA was done in 5 studies (Gao 2021; Albers 2022; Kang 2021a, 2021b; Yang 2018), and both primary and secondary RFAs were carried out in 1 study (Xia 2021). For the systematic review and meta-analysis (Cha 2021), 7 studies involved ERCP-directed RFA and 1 study (n=36) reported percutaneous RFA. Of the 7 studies, 6 studies were for primary RFA and only 1 study (Dutta 2017) reported both primary and secondary RFAs.

    In terms of stent types, plastic stents were initially used in 3 RCTs (Gao 2021; Kang 2021b; Yang 2018), SEMS in 2 RCTs (Albers 2022; Kang 2021a), and both plastic and metal stents in 1 observational study (Xia 2021). It is worth noting that Kang (2021b) changed the plastic stents to uncovered SEMS at 3 months after the index procedure or in case of premature plastic stent malfunction. For the systematic review and meta-analysis (Cha 2021), plastic stents were used in 2 studies and SEMS in 6 studies.

    Across all the studies, biliary strictures or obstructions were caused by cancer that was surgically unresectable or unsuitable for surgery. The most common cancer was CCA (mostly hilar and distal CCA), followed by pancreatic cancer, gallbladder cancer and others. In all but 1 primary study (Yang 2018), other anti-tumour treatments (such as chemotherapy) were allowed concomitantly or after the index procedure. Chemotherapy was identified as a key confounding factor for OS.

    As part of this review, a meta-analysis was done, combining the results from the 6 primary studies which contained relevant and valid data using the same outcome measures. Data analysis and all pooled results including forest plots are available in the appendix.

    Table 2 presents study details.

    Figure 1 Flow chart of study selection

    Table 2 Study details

    No.

    First author, date

    country

    Population characteristics

    Study design

    Inclusion criteria

    Intervention

    Follow up

    1

    Gao (2021) China (3 centres)

    174 (men, n=89; women, n=85)

    • RFA plus plastic stent: n=87 (mean 68 years)

    • Plastic stent only: n=87 (mean 67.9 years)

    RCT (open-label; NCT01844245)

    Age 18 years or over, obstructive jaundice caused by pathology-confirmed extrahepatic CCA or ampullary cancer which was unsuitable for surgery.

    • ERCP-directed RFA (2-session RFA therapy using; Habib EndoHPB catheter and ESG-100 generator) plus plastic stent

    • Plastic stent only

    Median 31.8 months

    2

    Albers (2022)

    Germany (5 centres)

    86 (women, n=47; men, n=39; mean 71.6 years)

    • RFA plus SEMS: n=42 (mean 70.9 years)

    • SEMS only: n=44 (mean 72.3 years)

    RCT (DRKS00018993)

    Age over 18 years, obstructive jaundice caused by unresectable malignancy, and malignant tumour causing consequential biliary stenosis with the need for biliary drainage.

    • ERCP-directed RFA (Habib EndoHBP RFA catheter and VIO 200D generator) plus SEMS (single stent, n=38; more than 1 stent, n=4)

    • SEMS only (single stent, n=38; more than 1 stent, n=6)

    6 months

    3

    Kang (2021a)

    Korea (single centre)

    48 (male, n=30; female, n=18)

    • RFA plus SEMS: n=24 (median 73 years)

    • SEMS only: n=24 (median 67 years)

    RCT (open-label; NCT02646514)

    Age over 19 years, malignancy that was surgically unresectable or medically unfit for surgery, MBS, and life expectancy more than 3 months.

    • RFA (ELRA™ catheter and VIVA ComboTM generator) and uncovered SEMS placement.

    • Uncovered SEMS only.

    Percutaneous RFA was done in 4 people of each group.

    • Median 135.0 days

    • Median 119.5 days

    4

    Kang (2021b)

    South Korea (3 centres)

    30 (male, n=20; female, n=10)

    • Intraductal RFA plus plastic stent: n=15 (mean 76 years)

    • Plastic stent only: n=15 (mean 72 years)

    Pilot RCT (open- label; KCT0003275)

    Age over 19 years, a confirmed MHBO because of CCA or gallbladder cancer which was unresectable or medically inoperable, life expectancy more than 3 months, and ECOG performance status 0 (fully active) to 1 (restricted in strenuous activity)

    • RFA (ELRA™ catheter and VIVA ComboTM generator) and plastic stent insertion.

    • Plastic stent only: bilateral placement.

    For both groups, plastic stent was changed to uncovered SEMS after 3 months or in case of premature plastic stent malfunction.

    12 months

    • Median 178 days

    • Median 134 days

    5

    Yang (2018)

    China (single centre)

    65 (male, n=33; female, n=32)

    • RFA plus plastic stent: n=32 (mean 62 years)

    • Plastic stent only: n=33 (mean 64.5 years)

    RCT (NCT02592538)

    Age 18 to 75 years, pathology-confirmed extrahepatic CCA, initial treatment, and KPS of 50 points or higher

    • ERCP- guided RFA (Habib EndoHPB catheter and RITA 1500X generator) plus plastic stents (mean 4.7 ERCPs per person)

    • Plastic stent only (mean 3.8 ERCPs per person)

    For both groups, stent replaced every 3 months or upon recurrent jaundice or cholangitis symptoms.

    21 months

    6

    Xia (2021)

    China (2 centres)

    883 (male, n=537; female, n=346; mean 65.3 years)

    • RFA plus stent: n=124

    • Stent only: n=759 ([496 in the matched cohort)

    Non-randomised comparative study (retrospective)

    People with inoperable extrahepatic MBS who had ERCP therapies

    • ERCP-guided RFA (Habib catheter and ERBE VIO200D generator) plus stent (plastic or metal stents)

    • Stent only (plastic or metal)

    7 years

    7

    Cha (2021)

    420 (3 RCTs and 5 retrospective observational studies)

    • RFA plus stent: n=190

    • Stent only: n=230

    Age, not reported

    Systematic review and meta-analysis

    People with unresectable extrahepatic CCA or MBO; studies comparing treatment outcomes between endobiliary RFA with stent and stent only; RCTs and case-control studies with adjustment for confounding variables; and studies reporting the relative risks or odds ratios of survival rate and stent patency duration for both treatments.

    ERCP-guided RFA in 7 studies and percutaneous RFA in 1 observational study (n=18); plastic stents used in 2 RCTs and SEMS in 1 RCT and 5 observational studies.

    Not reported

    The risk of bias (RoB) was assessed using the Cochrane RoB Tool for RCTs, ROBINS-I (RoB in Nonrandomized Studies [NRSs] of Interventions) for NRSs, and ROBIS for systematic reviews. The RoB for RCTs was graded as "low" for low risk, "high" for high risk, and "some concerns". For NRSs, each domain of the RoB was rated as "low", "moderate", "serious", or "critical" for RoB. For systematic review, each domain of the RoB was judged as ''low,'' ''high,'' or ''some concerns.''. The overall RoB was determined according to the judgment for each domain.

    Among the RCTs, 3 studies (Gao 2021; Albers 2022; Yang 2018) were rated as low for all RoB domains, whereas for the other 2 studies (Kang 2021a, 2021b), concerns were raised for 1 domain (deviation from intended intervention as PTC was also allowed [4 people in Kang 2021a and 3 people in Kang 2021b] and no subgroup analysis of people with ERCP-guided RFA only) and they were rated as low for all other domains. So, the overall risk of bias was judged as low for 3 studies (Gao 2021; Albers 2022; Yang 2018) and some concerns for 2 studies (Kang 2021a, 2021b).

    For the retrospective, observational study (Xia 2021), it was rated as low for 4 domains (classification of intervention, deviation from intended intervention, missing data and measurement of outcomes) and as moderate for 3 domains (selection bias, (unmeasured) confounding, and reporting). So, the overall risk of bias was judged as moderate.

    In terms of the systematic review and meta-analysis (Cha 2021), based on ROBIS, there were some concerns regarding the synthesis (outcomes for ERCP-guided RFA not reported separately and no sensitivity or subgroup analysis), but the levels of concerns for all other domains were judged as low.

    Table 3 Study outcomes

    First author, date

    Efficacy outcomes

    Safety outcomes

    Gao (2021)

    Total sample, n=174 (RFA plus plastic stent, n=87; plastic stent only, n=87)

    Mortality due to tumour progression during follow up: RFA plus plastic stent, n=67; plastic stent only, n=81, p=0.940

    OS: RFA plus plastic stent versus plastic stent only

    • Median cumulative survival (Kaplan-Meier analysis): 14.3 months (95% CI, 11.9 to 16.7) versus 9.2 months (95% CI, 7.1 to 11.2); HR, 0.488; 95% CI 0.351 to 0.678; p<0.001.

    • Median OS for people with CCA: 13.3 months (95% CI, 10.2 to 16.3) versus 9.2 months (95% CI 7.2 to 11.1); HR, 0.546; 95% CI, 0.386 to 0.771; p<0.001

    • Statistically significant differences in OS were noted in all subgroups, including hilar CCA (HR, 0.414; 95% CI, 0.025 to 0.762; p=0.004), distal CCA (HR, 0.642; 95% CI, 0.421 to 0.981; p=0.038), and ampullary cancer (HR, 0.301; 95% CI, 0.100 to 0.900; p=0.032)

    RFA treatment (HR, 0.498; 95% CI, 0.352 to 0.703; p<0.001) was an independent predictor of increased OS.

    Within the RFA group, ampullary cancer (HR, 0.274; 95% CI, 0.092 to 0.817; p=0.020; compared with hilar CCA) was a statistically significant predictor of increased survival, whereas total bilirubin (HR, 1.804; 95% CI, 1.010 to 3.223; p=0.046) was the predictor of decreased survival after RFA.

    Clinical success for jaundice control: 92.0% (80/87) versus 90.8% (79/87), p>0.99

    Stent occlusions before the second scheduled ERCP (mean 90 days after the first procedure): 22% (19/87) versus 18% (16/87), p=0.570

    After the second procedure the cumulative stent patency duration: 3.7 months (95% CI, 2.8 to 4.5) versus 4.1 months (95% CI, 3.7 to 4.5), p=0.674; HR, 1.069; 95% CI, 0.782 to 0.460

    KPS scores: the median KPS scores at baseline were both 80 in the 2 group. After the procedure, KPS scores were statistically significantly higher in the RFA plus plastic stent group than in the plastic stent only group at 1, 3, 6 and 9 months (all p<0.05).

    Early adverse events (less than 30 days) - RAF plus plastic stent versus plastic stent only: 27.6% (n=4) versus 19.5% (n=17), p=0.211

    • Post-ERCP pancreatitis (mild): 4.6% (n=4) versus 5.7% (n=5), p>0.99

    • Bleeding: 1.1% (n=1) versus 3.4% (n=3), p=0.621

    • Acute cholangitis: 11.5% (n=10) versus 10.3% (n=9), p=0.808

    • Acute cholecystitis, 10.3% (n=9) versus 0; p=0.003; 7 of the 9 people had hilar CCA, 1 had ampullary cancer, and 1 had distal CCA coexisting with gallstones.

    • Perforation: 0 in both groups

    Late adverse events: 39.1% (n=34) versus 36.8% (n=32); p>0.99

    • Delayed cholangitis (treated with systemic antibiotics): 37.9% (n=33) versus 36.8% (n=32), p=0.875

    • Liver abscess: 1.1% (n=1) versus 0, p=0.755

    No severe procedure-related adverse events or procedure-related mortality were noted.

    Albers (2022)

    Total sample: n=86 (RFA plus SEMS, n=42; SEMS only, n=44)

    Technical success of the RFA procedure (defined by passing the malignant stenosis with a guidewire, subsequently with the RFA electrode, and applying the total ablation energy): 100%

    Technical success for SEMS insertion in the correct position: 98.8% (85/86)

    Patency rate (RFA plus SEMS versus SEMS only):

    • After 3 months: 73.1% versus 81.8%; p=1.0

    • After 6 months: 33.3% versus 52.4%, p=0.6

    The addition of RFA did not impact OS (HR, 0.72; p=0.389 for RFA plus SEMS).

    The patency rate in the subgroup with extrahepatic strictures indicated no statistically significant difference after 3 and 6 months between groups (SEMS only [n=37], 77.2% and 53.3%, respectively [p=0.51]; RFA plus SEMS [n=37], 75% and 36.4%, respectively [p=0.45]). Accordingly, patency rates in the subgroup with pancreatic cancer showed no statistical difference after 3 and 6 months (SEMS only [n=18], 87.5% and 40%, respectively [p=0.25]; RFA plus SEMS [n=16], 65% and 25%, respectively, [p=0.11])

    Overall mortality after 3 and 6 months:

    • RFA plus SEMS: 24.3% and 41.9%, p=0.56

    • SEMS only: 26.8% and 50%, p=0.99

    Similar results were found in the cohort with extrahepatic strictures (SEMS only [n=37], 22.6% and 53.3%, respectively [p=0.77]; RFA plus SEMS [n=37], 23.5% and 39.3%, respectively [p=0.38]) and in the pancreatic cancer subgroup (SEMS only [n=18], 37.9% and 62.1%, respectively [p=0.23]; RFA plus SEMS [n=16], 20% and 38.9%, respectively [p=0.48])

    Subgroup analysis of 74 people with extrahepatic versus 12 with intrahepatic strictures and of 54 people with pancreatic cancer versus 32 with other entities: no statistically significant difference in the patency rate and overall mortality after 3 and 6 months.

    Adverse events within the first 30 days after the intervention (RFA plus SEMS versus SEMS alone): 4 (10.5%) versus 1 (2.3%), p=0.18

    • Minor events:

      • Bleeding: 1 (2.6%) versus 1 (2.3%)

      • Pancreatitis: 1 (2.6%) versus 0

      • Cholangitis: 1 (2.6%) versus 0

    • Major event:

      • Cholecystitis with gallbladder perforation: 1 (2.6%) versus 0

    Kang (2021a)

    Total sample, n=48 (RFA plus SEMS, n=24; SEMS only, n=24)

    Comparison of outcomes: RFA plus SEMS versus SEMS only

    • Technical success: 100% versus 100%, p=1.000

    • Clinical success: 87.5% (n=21) versus 83.3% (n=20), p=1.000

    • 90-day stent patency: 58.3% (n=14) versus 45.8% (n=11), p=0.386

    • Duration of stent patency (median): 132.0 days (95% CI, 99.6 to 164.4) versus 116.0 days (95% CI, 52.4 to 179.6), p=0.440

    • OS (median): 244.0 days (95% CI, 117.8 to 370.0) versus 180.0 days (95% CI, 27.8 to 332.2), p=0.281

    Subgroup analysis of people with MBS at the CBD only: RFA plus SEMS versus SEMS only

    • 90-day stent patency: 70.6% (12/17) versus 42.9% (9/27), p=0.087

    • The median duration of stent patency and the median OS were not statistically significantly different in this subgroup analysis (exact data not reported).

    Analysis with stratification of people according to the primary site of malignancy did not find any statistically significant differences in outcomes.

    Comparison of complications within 30 days: RFA plus SEMS versus SEMS only

    • Early complications within 7 days: 1 (4.2%) versus 3 (12.5%), p=0.609

    • Pancreatitis (all mild): 0 versus 3 (12.5%), p=0.234

    • Cholangitis: 1 (4.2%) versus 0, p=1.000

    A cholangitis resulting in septic shock and death was reported as unlikely related severe adverse event.

    • Postprocedure pain (24 hours), VAS: median 6.0 (IQR 2.0 to 6.0) versus 6.0 (IQR 3.3 to 7.0), p=0.163

    • Postprocedure pain (24 hours), VAS 7 or more: 3 (12.5%) versus 8 (33.3%), p=0.086

    All procedure-related complications were managed conservatively, and the people recovered uneventfully, except the death mentioned above.

    No perforation, bleeding and late complications (more than 7 days) in both groups.

    Kang (2021b)

    Total sample, n=30 (RFA plus plastic stent, n=15; plastic stent only, n=15; per-protocol analysis, n=22 [12 versus 10])

    Total event-free stent patency:

    • ITT analysis: 178 days (95% CI, 96.2 to 259.8) versus 122 days (95% CI, 111.2 to 132.8), p=0.154

    • Per-protocol analysis: 175 days (95% CI, 144.4 to 205.6) versus 122 days (95% CI, 111.2 to 132.8), p=0.623

    OS:

    • ITT analysis: 230 days (95% CI, 77.0 to 383.0) versus 144 days (95% CI, 0 to 323.1), p=0.643

    • Per-protocol analysis: 178 days (95% CI, 84.6 to 271.4) versus 144 days (95% CI, 0 to 292.2), p=0.593

    Technical success: 100% (n=15) versus 93.3% (n=14), p>0.999

    Clinical success: 100% (n=15) versus 86.7% (n=13), p=0.483

    Among 13 people of each group who had the sequential procedure, more people had scheduled stent exchange after 3 months without PS occlusion in the RFA plus plastic stent group than in the plastic stent only group (69.2% vs 23.1%, p=0.018, power=81.49%).

    Subgroup analysis – total event-free stent patency:

    • people with strictures lengths of 11 mm or more on both side IHDs (13 people in each group): 175 days (95% CI, 146.1 to 203.9) versus 121 days (95% CI, 104.5 to 137.5), p=0.028

    • no statistically significant difference was found in people with different types of bismuth, in people with hilar CCA only, in people with mass-forming type CCA.

    No survival difference was shown in any subgroup analyses.

    Early adverse events (within 30 days) (RFA plus plastic stent versus plastic stent alone)

    • Cholangitis: 3 (20%) versus 5 (33.3%)

    • Cholecystitis: 1 (6.7%) versus 1 (6.7%)

    • Fever: 2 (13.3%) versus 3 (20.0%)

    • Abdominal pain, NRS: 3 (0 to 4) versus 0 (0 to 5)

    All p>0.05

    Pancreatitis, liver abscess and death (because of sudden myocardial infarction, so it was unrelated to the procedure) were reported in the stent alone group only, but no statistically significant difference was found.

    Late adverse events (more than 30 days) (RFA plus stent versus stent alone)

    • Cholangitis: 5 (33.3%) versus 5 (33.3%)

    • Cholecystitis: 1 (6.7%) versus 0

    • Liver abscess: 0 versus 2 (13.3%)

    All p>0.05

    All reported adverse events were mild to moderate except for 3 cases of severe late cholangitis related to stent occlusion in the stent only group. All adverse events were medically manageable and procedure-related death was not reported.

    Yang (2018)

    Total sample: n=65 (RFA plus plastic stent, n=32; plastic stent only, n=33)

    OS: RFA plus plastic stent versus plastic stent only

    • 6-month survival rate: 96.9% versus 81.8%, p=0.08

    • 9-month survival rate: 87.5% versus 24.2%, p<0.05

    • 12-month survival rate: 62.5% versus 12.1%, p<0.05

    • 15-month survival rate: 28.1% versus 3.0%, p<0.05

    • Overall mean survival time: 13.2 months (SD 0.6, 95% CI 11.8 to 14.2) versus 8.3 months (SD 0.5, 95% CI 7.3 to 9.3), p<0.001

    • Multivariable Cox regression analysis showed that RFA was the main protective factor affecting survival (HR 0.182; 95% CI, 0.08 to 0.322; p<0.001)

    • All people died by July 2017. The total follow-up duration of the study was 21 months.

    Stent patency: 6.8 months (95% CI, 3.6 to 8.2) versus 3.4 months (95% CI, 2.4 to 6.5), p=0.02

    No statistically significantly difference in the stent length between groups (8.3 cm versus 9.3 cm, p>0.05)

    KPA scores: RFA plus plastic stent versus plastic stent only

    • Preoperation: 82.9 (SD 9.3) versus 79.9 (SD 7.8), p=0.28

    • 1-month postoperation: 86.1 (SD 6.8) versus 72.4 (SD 8.2), p=0.02

    • 3-month postoperation: 71.4 (SD 7.1) versus 60.3 (SD 5.4), p=0.04

    • 6-month postoperation: 61.4 (SD 7.1) versus 48.2 (SD 6.2), p=0.03

    • 9-month postoperation: 58.2 (SD 11.5) versus 22.5 (SD 8.9), p<0.001

    Serum bilirubin (micromol/l): RFA plus plastic stent versus plastic stent only

    • Total bilirubin at baseline: 266.8 (SD 88.5) versus 245.9 (SD 76.2), p=0.23

    • Total bilirubin at 3 months: 39.2 (SD 6.3) versus 46.9 (SD 8.9), p=0.68

    • Direct bilirubin at baseline: 188.5 (SD 48.6) versus 169.5 (SD 58.4), p=0.43

    • Direct bilirubin at 3 months: 19.3 (SD 5.9) versus 22.6 (SD 7.8), p=0.58

    Postoperative adverse events:

    • RFA plus plastic stent: 6.3% (2/32)

    • Plastic stent only: 9.1% (3/33)

    • p=0.67

    RFA plus plastic stent group:

    • Acute cholangitis: 6.3% (2/32), both people improved after nasobiliary drainage through the bile duct and intravenous antibiotic therapy.

    Plastic stent only group:

    • Acute cholangitis: 3.0% (1/33)

    • Acute pancreatitis: 3.0% (1/33)

    • Haemorrhage of the incision margin of the papilla: 3.0% (1/33)

    All people were successfully managed endoscopically. No serious adverse events were found.

    Xia (2021)

    Total sample: n=883 (RFA plus stent [plastic or metal stent], n=124; stent [plastic or metal stent] only, n=759; following PSM, 124 in RFA plus stent and 496 in stent only)

    Clinical success: RFA plus stent versus stent only

    • Unmatched: 92.7% (115/124) versus 85.2% (647/759), p=0.024, 95% CI (difference), 0.022 to 0.128

    • Matched: 92.7% (115/124) versus 88.7% (440/496), p=0.190, 95% CI (difference), -0.014 to 0.094

    Time to recurrent biliary obstruction (median, IQR): RFA plus stent versus stent only

    • Unmatched: 7.2 months (IQR 5.4 to 11.3) versus 6.0 months (IQR 3.6 to 9.7), p=0.016, 95% CI (difference), -3.636 to -0.576

    • Matched: 7.2 months (IQR 5.4 to 11.3) versus 6.4 months (IQR 3.8 to 10.2), p=0.121, 95% CI (difference), -3.203 to 0.130

    Total number of intervention (mean):

    • Unmatched: 1.6 (SD 0.9) versus 1.5 (SD 0.9), p=0.391, 95% CI (difference), -0.097 to 0.247

    • Matched: 1.6 (SD 0.9) versus 1.5 (SD 0.9), p=0.163, 95% CI (difference), -0.050 to 0.296

    OS (median): RFA plus stent versus stent only

    • Unmatched: 9.5 months (95% CI, 7.7 to 11.3) versus 5.5 months (95% CI, 5.0 to 5.9), p<0.001, 95% CI (difference), 1.600 to 3.400

    • Matched: 9.5 months (95% CI 7.7 to 11.3) versus 6.1 months (95% CI 5.6 to 6.6), p<0.001, 95% CI (difference), 1.288 to 4.936

    By the final follow up, 789 (89.4%) people died of disease progression.

    In multivariable Cox proportional hazard models, RFA plus stent (HR 0.552, 95% CI 0.438 to 0.697, p<0.001), TNM stage IV (HR 1.2073, 95% CI 1.026 to 1.420, p=0.023), re-procedural total bilirubin level more than 200microμmol/l (HR 1.265, 95% CI 1.085 to 1.474, p=0.003), other antitumour therapy (HR 0.908, 95% CI 0.866 to 0.951, p<0.001), and use of metal stent (HR 0.751, 95% CI 0.640 to 0.880, p<0.001) were found to be independent predictors of OS.

    In stratified analyses, the improved OS was only demonstrated in the subgroup of extrahepatic CCA (n=463 [RFA plus stent in 79 people and stent alone in 384 people], 11.3 months [95% CI, 10.2 to 12.4] versus 6.9 months [95% CI, 6.0 to 7.8], p<0.001), but not in the subgroups of gallbladder cancer, hepatocellular carcinoma, intrahepatic CCA, pancreatic cancer and other metastatic cancers (all p>0.05). The survival benefits were noted only in the people with non-metastatic CCA (11.5 versus 7.4 months, p<0.001).

    Adverse events: RFA plus stent versus stent only

    • Unmatched: 18.5% (23/124) versus 19.5% (148/759), p=0.804, 95% CI (difference) -0.085 to 0.066

    • Matched: 18.5% (23/124) versus 15.9% (79/496), p=0.481, 95% CI (difference) -0.047 to 0.099

    Post-ERCP pancreatitis: RFA plus stent versus stent only

    • Unmatched: 8.9% (11/124) versus 5.1% (39/759), p=0.095, 95% CI (difference) -0.016 to 0.090

    • Matched: 8.9% (11/124) versus 5.4% (27/496), p=0.155, 95% CI (difference) -0.020 to 0.089

    Cholangitis: RFA plus stent versus stent only

    • Unmatched: 6.5% (8/124) versus 14.2% (108/759), p=0.017, 95% CI (difference) -0.128 to -0.028

    • Matched: 6.5% (8/124) versus 10.3% (51/496), p=0.194, 95% CI (difference) -0.090 to 0.013

    Cholecystitis: RFA plus stent versus stent only

    • Unmatched: 4.8% (6/124) versus 0.1% (1/759), p<0.001, 95% CI (difference) 0.009 to 0.086

    • Matched: 4.8% (6/124) versus 0.2% (1/496), p<0.001, 95% CI (difference) 0.008 to 0.085

    More acute cholecystitis occurrences in the RFA plus stent group, which might be related to cystic duct injury caused by RFA. All people were cured by enhanced antibiotic therapy or percutaneous gallbladder drainage.

    Bleeding: RFA plus stent versus stent only

    • Unmatched: 1.6% (2/124) versus 1.2% (9/759), p=0.691, 95% CI (difference) -0.017 to 0.025

    • Matched: 1.6% (2/124) versus 1.2% (6/496), p=0.722, 95% CI (difference) -0.018 to 0.026

    Perforation: RFA plus stent versus stent only

    • Unmatched: 0.8% (1/124) versus 0% (0/759), p=0.140, 95% CI (difference) -0.008 to 0.024

    • Matched: 0.8% (1/124) versus 0% (0/496), p=0.200, 95% CI (difference) -0.008 to 0.024

    One person in the RFA plus stent group developed minor duodenal perforation induced by a plastic stent and was successfully closed using endoscopic clipping.

    No RFA-related bleeding, perforation, or mortality occurred.

    Cha (2021)

    420 (3 RCTs and 5 retrospective observational studies; RFA plus stent, n=190; stent only, n=230)

    Comparative outcomes - RFA plus stent versus stent only:

    • Pooled HR for OS: 0.47; 95% CI, 0.34 to 0.64; I2=44% (8 studies)

    • Pooled HR for stent patency: 0.79; 95% CI, 0.57 to 1.09; I2=7% (2 RCTs and 2 observational studies)

    No significant publication bias was noted in the funnel plot for overall survival (Egger's test p=0.2869).

    Adverse events:

    • Abdominal pain: n=3

    • Pancreatitis and hyperamylasaemia: n=3

    • Cholangitis regardless of symptoms, n=34

    • Cholecystitis: n=8

    All these complications were treated with antibiotics and conservative therapy, and no procedure-related mortality was reported.

    One case of intestinal perforation with pneumothorax occurred in 1 study; but the authors commented that it was not related to the RFA procedure but rather to the scope device.

    No study showed statistically significant differences in terms of adverse events between the RFA plus stent group and the stent only group.

    Procedure technique

    All 6 primary studies detailed the procedure technique and devices used. The most common catheter used was Habib EndoHPB, followed by ELRAcatheter. Four RFA generators were used.

    This procedure was done under sedation or monitored anaesthesia. Biliary dilation could be done at the endoscopist's discretion. RFA was usually done with a power of 7 to 10 w for 90 sections with a subsequent cooling period of 60 seconds to reduce adverse events. The ablation was done in a stepwise manner, covering the stricture from the proximal to the distal edge. More than 1 ablation could be done in a single session. After RFA, plastic or metal stents were inserted. Repeated RFA sessions could be done (Gao 2021; Kang 2021b; Yang 2018; Xia 2021).

    Efficacy

    The meta-analysis of the 6 primary studies was done, and the pooled results of stent patency and OS can be found in the appendix.

    Clinical success

    Clinical success was reported in 4 studies, with its rate ranging from 88% to 100%. When comparing RFA plus stent with stent only, no statistically significant difference was found in general.

    Gao (2021) did not find any statistically significant difference in clinical success for jaundice control between the RFA plus plastic stent group (n=87) and the plastic stent only group (n=87; 92% versus 91%, p>0.99). Similarly, Kang (2021a) reported that the clinical success rate did not have a statistically significant difference between the RFA plus SEMS group (n=24) and the SEMS only group (n=24; 88% versus 83%, p=1.000). Kang (2021b) also did not see any statistically significant difference in clinical success between the RFA plus plastic stent group (n=15) and the plastic stent only group (n=15; 100% versus 87%, p=0.483).

    In a non-randomised comparative study of 883 people, Xia (2021) noted that the rate of clinical success was statistically significantly higher in the RFA plus stent group (n=124) than the stent only group (n=759; 93% versus 85%, p=0.024). But after PSM no statistically significant difference was found between groups (93% of 124 people versus 89% of 496 people, p=0.190).

    Stent patency

    Stent patency was described in 6 studies. There were conflicting results in the duration of stent patency or the patency rate between RFA plus stent and stent only. But no statistically significant difference was found, except for the Yang (2018) study.

    Gao (2021) did not find any statistically significant difference in the cumulative stent patency duration after the second procedure between the RFA plus plastic stent group and the plastic stent group (3.7 months versus 4.1 months, p=0.674; HR, 1.069; 95% CI, 0.782 to 0.460).

    Albers (2022) did not report any statistically significant difference in the patency rate between the RFA plus SEMS group (n=42) and the SEMS only group (n=44) after 3 months (73% versus 82%, p=1.0) and 6 months (33% versus 52%, p=0.6). The patency rate in the subgroup with extrahepatic strictures showed no statistically significant difference after 3 and 6 months between groups. So, the patency rate in the subgroup with pancreatic cancer (SEMS only, n=18; RFA plus SEMS, n=16) showed no statistically significant difference after 3 and 6 months.

    Kang (2021a) also did not find any statistically significant differences in the median duration of stent patency (132 days versus 116 days, p=0.440) and the 90‑day stent patency rate (58% versus 46%, p=0.386) between the RFA plus SEMS group and the SEMS only group.

    Kang (2021b) reported that ITT analysis of 30 people (15 in each group) did not show any statistically significant difference in the total event-free stent patency between the RFA plus plastic stent group and the plastic stent only group (178 days versus 122 days, p=0.154). But the authors found that in people with each stricture length of 11 mm or longer on both sides, stent patency was statistically significantly longer in the RFA plus plastic stent group than in the plastic stent only group (175 days versus 121 days, p=0.028). More people had elective exchange to SEMS without plastic stent occlusion in the RFA plus stent group than in the stent only group (69% versus 23%, p=0.018).

    Yang (2018) reported that the duration of stent patency was statistically significantly longer in the RFA plus plastic stent group than the plastic stent only group (6.8 months versus 3.4 months, p=0.02), and stent length was comparable between groups.

    In a systematic review and meta-analysis of 420 people who had RFA plus stent or stent only (3 RCTs and 5 retrospective observational studies; RFA plus stent, n=190; stent only, n=230), Cha (2021) described that no statistically significant difference was found in the duration of stent patency between groups (HR, 0.79; 95% CI 0.57 to 1.09, I2=7%; 2 RCTs and 2 observational studies).

    OS

    OS was reported in all 7 studies and the evidence suggested a trend in favour of RFA plus stent. A statistically significant level was achieved in 3 studies (Gao 2021; Yang 2018; Xia 2021) and in a systematic review and meta-analysis (Cha 2021).

    Gao (2021) found that the median OS was statistically significantly higher in the RFA plus plastic stent group than the plastic stent only group (14.3 months versus 9.2 months; HR, 0.488; 95% CI, 0.351 to 0.678; p<0.001). A survival benefit was also shown in people with CCA (13.3 months versus 9.2 months; HR, 0.546; 95% CI, 0.386 to 0.771; p<0.001).

    Kang (2021a) did not find any statistically significant difference in median OS between the RFA plus SEMS group and the SEMS only group (244 days versus 180 days, p=0.281).

    Albers (2022) described that the overall mortality after 3 and 6 months did not differ statistically significantly in both groups (RFA plus SEMS, 24% and 42%, p=0.56; SEMS, 27% and 50%, p=0.99). Similar results were found in the cohort with extrahepatic strictures (37 people in each group) and in the pancreatic cancer subgroup (SEMS only, n=18; RFA plus SEMS, n=16). The addition of RFA did not impact OS (HR, 0.72; p=0.389 for RFA plus SEMS).

    Yang (2018) reported that the mean OS time was statistically significantly longer in the RFA plus plastic stent group than the plastic stent only group (13.2 months versus 8.3 months, p<0.001). The survival rates were statistically higher in the RFA plus plastic stent group than the plastic stent only groups at 9 months (88% versus 24%), 12 months (63% versus 12%) and 15 months (28% versus 3%; all p<0.05) but not at 6 months (97% versus 82%, p=0.08). Multivariable Cox regression analysis showed that RFA was the main protective factor affecting the survival of people (HR 0.182, 95% CI, 0.08 to 0.322; p<0.001).

    Kang (2021b) described that ITT analysis of 30 people did not show any statistically significant difference in OS between the RFA plus plastic stent group and the plastic stent only group (230 days versus 144 days, p=0.643).

    Xia (2021) reported that the median OS was statistically significantly longer in the RFA plus stent group than the stent only group (9.5 months versus 5.5 months, p<0.001). After PSM, people who had RFA plus stent also showed statistically significantly longer OS than those who had stent alone (9.5 months versus 6.1 months, p<0.001). In multivariable Cox proportional hazard models, RFA plus stent was found to be an independent predictor of OS (HR, 0.552; 95% CI, 0.438 to 0.697; p<0.001), In stratified analyses, the improved OS was only showed in the subgroup of extrahepatic CCA (11.3 months versus 6.9 months, p<0.001) but not in the subgroups of other cancers (all p>0.05). The survival benefit was noted only in the people with non-metastatic CCA (11.5 months versus 7.4 months, p<0.001).

    In the systematic review and meta-analysis by Cha (2021), when comparing RFA plus stent with stent only, the pooled HR for OS was 0.47 (95% CI, 0.34 to 0.64, I2=44%; 8 studies). No significant publication bias was noted in the funnel plot for overall survival (Egger's test p=0.2869).

    Quality of life

    Quality of life was measured using KPS in 2 RCTs, suggesting a better quality of life after RFA plus stent than stent only. Gao (2021) reported that the postprocedural KPS scores were statistically significantly higher in the RFA plus plastic stent group than the plastic stent only group until 9 months (all p<0.05). Similarly, Yang (2018) described that the postoperative KPS scores of people in the RFA plus plastic stent group were statistically significantly higher than those of the plastic stent only group after 1 month (86.1 versus 72.4), 3 months (71.4 versus 60.3), 6 months (61.4 versus 48.2), and 9 months (58.2 versus 22.5, all p<0.05). The authors of both RCTs described that the preoperative KPS scores were comparable between groups.

    Safety

    The pooled results of cholangitis and cholecystitis can be found in the appendix.

    Overall postoperative adverse events

    In general, the overall postoperative adverse events were comparable between people with RFA plus stent and people with stent only across studies.

    Gao (2021) did not find a statistically significant difference in early (within 30 days; 28% versus 20%, p=0.211) or late adverse events (39% versus 37%, p>0.99) between the RFA plus plastic stent group and the plastic stent only group.

    Albers (2022) did not report a statistically significant difference in adverse events between the RFA plus SEMS group and the SEMS only group within the first 30 days after the procedure (11% versus 2%, p=0.18).

    Kang (2021a) described that the early complication (within 7 days) rates were not statistically significantly different between the RFA plus SEMS group and the SEMS only group (4% versus 13%, p=0.609), and there were no late complications (7 to 30 days) in both groups.

    Yang (2018) did not find a statistically significant difference in the incidence of postoperative adverse events between the RFA plus plastic stent group and the plastic stent only group (6% versus 9%, p=0.67).

    Xia (2021) found that the overall complication rates were comparable between the RFA plus stent group and the stent only group before and after PSM (unmatched: 19% versus 20%; matched: 19% versus 16%). The authors also found that the total post-ERCP complication rates were comparable between the RFA plus stent group and the stent only group before and after PSM (9% versus 5%).

    In the systematic review and meta-analysis (Cha 2021), no study showed any statistically significant difference in adverse events between the RFA plus stent group and the stent only group.

    Cholangitis

    Cholangitis was described in all studies and its incidence was generally comparable between the RFA plus stent group and the stent only group.

    Gao (2021) did not find a statistically significant difference in cholangitis (acute cholangitis: 11.5% versus 10.3%, p=0.808; delayed cholangitis, 38% versus 37%, p=0.875) between the RFA plus plastic stent group and the plastic stent only group within 30 days after the procedure. Kang (2021b) did not report any statistically significant difference in cholangitis between the RFA plus plastic stent group and the plastic stent only group within the first 30 days (20% [n=3] versus 33% [n=5]) and more than 30 days after the procedure (33% [n=5] versus 33% [n=5]).

    Albers (2022) reported cholangitis in 1 person in the RFA plus SEMS group but 0 in the SEMS only group within the first 30 days after the procedure. Similarly, Kang (2021a) reported cholangitis in 1 person in the RFA plus SEMS group but 0 in the SEMS only group within 30 days after the procedure. This case resulting in septic shock and death was reported as an unlikely related severe adverse event.

    Yang (2018) reported acute cholangitis in 2 people (6.3%) in the RFA plus plastic stent group and 1 (3%) in the plastic stent group only after the procedure.

    Xia (2021) reported that the incidence of cholangitis was statistically significantly lower in the RFA plus stent group than the stent only group (6.5% [8/124] versus 14% [108/759], p=0.017) before PSM. But this statistically significant difference was not found after PSM (7% [8/124] versus 10% [51/496], p=0.194).

    In the systematic review and meta-analysis, Cha (2021) reported cholangitis regardless of symptoms in 34 people after RFA plus stent insertion.

    Cholecystitis

    Cholecystitis was reported in 4 studies and in 1 systematic review and meta-analysis. Gao (2021) and Xia (2021) found a statistically significantly higher incidence of acute cholecystitis in the RFA plus stent group than the stent only group. This finding was not supported by Kang (2021b) and Albers (2022).

    Gao (2021) reported that the number of people experiencing acute cholecystitis was statistically significantly higher in the RFA plus plastic stent group than the plastic stent only group (10% versus 0%, p=0.003) within 30 days after the procedure. Of the 9 people with acute cholecystitis, 7 had hilar CCA, 1 had ampullary cancer, and 1 had distal CCA coexisting with gallstones. Xia (2021) reported that the incidence of acute cholecystitis was statistically significantly higher in the RFA plus stent group than the stent only group (6 versus 1, p<0.001) before and after PSM.

    But Kang (2021b) did not find any statistically significant difference in cholecystitis between the RFA plus plastic stent group and the plastic stent group within the first 30 days (7% [n=1] in each group) and more than 30 days (7% [n=1] versus 0%) after the procedure. Albers (2022) reported cholecystitis with gallbladder perforation in 1 person in the RFA plus SEMS group but not in the SEMS only group within the first 30 days after the procedure.

    In the systematic review and meta-analysis, Cha (2021) reported cholecystitis in 8 people after RFA plus stent insertion.

    Pancreatitis

    Pancreatitis was seen in 2 studies and in 1 systematic review and meta-analysis.

    Gao (2021) did not find a statistically significant difference in post-ERCP pancreatitis (mild; 5% versus 6%, p>0.99) between the RFA plus plastic stent group and the plastic stent only group within 30 days after the procedure. Albers (2022) reported pancreatitis in 1 person in the RFA plus SEMS group but 0 in the SEMS only group within the first 30 days after the procedure. Cha (2021) reported pancreatitis and hyperamylasaemia in 3 people.

    Perforation

    Perforation was reported in 2 studies, with 1 case in each study after RFA plus stent insertion.

    Xia (2021) did not find a statistically significant difference in the incidence of perforation between the RFA plus stent group and the stent only group before and after PSM (1 versus 0). This case was a minor duodenal perforation induced by a plastic stent and was successfully closed using endoscopic clipping. Albers (2022) reported cholecystitis with gallbladder perforation in 1 person in the RFA plus SEMS group but 0 in the SEMS only group within the first 30 days after the procedure.

    Liver abscess

    Gao (2021) saw 1 person with liver abscess in the RFA plus plastic stent group but not in the plastic stent only group after 30 days following the procedure.

    Bleeding

    Bleeding was described in 3 studies and its incidence was comparable between the RFA plus stent group and the stent only group.

    Gao (2021) did not find a statistically significant difference in bleeding (1% versus 3%, p=0.621) between the RFA plus plastic stent group and the plastic stent only group within 30 days after the procedure. Albers (2022) reported bleeding in 1 person in the RFA plus SEMS group and 1 in the SEMS only group within the first 30 days after the procedure. Xia (2021) reported that the incidence of bleeding was comparable between the RFA plus stent group and the stent only group before and after propensity score matching (2% versus 1%).

    Pain

    Pain was reported in 2 studies and 1 systematic review and meta-analysis.

    Kang (2021a) did not find any statistically significant difference in postprocedural pain between the RFA plus SEMS group and the SEMS only group (median 6.0 in each group; for VAS 7 or more, 13% [n=3] versus 33% [n=8], p=0.086). Kang (2021b) reported the median score of abdominal pain (NRS) 3 (IQR 0 to 4) in the RFA plus plastic stent group and 0 (IQR 0 to 5) in the plastic stent only group within the first 30 days after the procedure. Cha (2021) reported abdominal pain in 3 people.

    Anecdotal and theoretical adverse events

    Expert advice was sought from consultants who have been nominated or ratified by their professional society or royal college. 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).

    In addition to the adverse events reported in the literature, they did not list anecdotal or theoretical adverse events.

    Three 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.

    Validity and generalisability

    The key evidence includes 5 RCTs, 1 non-randomised comparative study and 1 systematic review and meta-analysis. All 6 primary studies were conducted outside the UK. Most studies had a follow up of 4 to 32 months, although the observational study reported the data for 7 years. A mixture of cancer types was included in most studies, but the most common tumour was CCA.

    In addition to the RoB assessment detailed in table 2, samples were often not adequately powered for key efficacy outcome measures. Sample size calculation was done in 4 RCTs but for different efficacy outcomes. 2 RCTs (Albers 2022; Kang 2021a) were reasonably powered for stent patency while the other 2 RCTs (Gao 2021; Yang 2018) were reasonably power for OS. Kang (2021b) was a pilot RCT with a small sample size. Therefore, statistical power was generally not desirable for key efficacy outcome measures across studies.

    Stent patency was one of the key efficacy outcomes. This outcome was least affected by tumour types, concomitant systemic antitumour therapies or comorbidities (Albers 2022). Although there were conflicting results, in general, the evidence did not show any statistically significant difference in the duration of stent patency or the patency rate between the RFA plus stent group and the stent only group. This indicates no benefit in stent patency after RFA regardless of stent types.

    However, the mechanism of stent occlusion differs principally between plastic stents and SEMS. Albers (2022) stated that tumour ingrowth is the main reason for malfunction of SEMS, resulting in stent occlusion. The occlusion of plastic stents is mainly caused by bacterial biofilm formation, biliary sludge, and duodenobiliary reflux. Endoscopic RFA preceding stent insertion aims at reducing tissue overgrowth. Therefore, the effect of RFA on the patency of different stents needs to be explored further.

    OS was another key efficacy outcome. Across all the included studies, the evidence suggested a survival benefit in favour of RFA plus stent. It is noted that RFA is more like a local therapy than a systemic therapy and OS could be affected by many factors, such as stent patency, systemic control in combining with antitumour therapies (especially chemotherapy), cancer types and stages, and repeated RFA sessions.

    Other antitumour treatments (such as chemotherapy) were allowed concomitantly or after the index procedure in all but 1 primary study (Yang 2018). However, the proportion of people who actually had antitumour therapies, in particular chemotherapy, was generally low across studies but comparable between the RFA plus stent group and the stent only group within studies. So, the survival benefit may be more presentative of people who were forgoing systemic treatments.

    In respect of cancer type, the most common type of cancer was CCA. A survival benefit was generally directed towards people with MBO, mainly caused by extrahepatic CCA (especially in those without metastasis). Extrahepatic CCA mostly grows along the bile duct wall within limited thickness, thus the endoscopic RFA can reduce the maximal tumour load differing disease's progression (Xia 2021). For MOB caused by other cancers (such as pancreatic cancer), the survival benefit was uncertain.

    More than 1 RFA session was allowed in 4 studies, and of these, 3 studies reported a statistically significant survival benefit after RFA plus stent compared with stent only (Gao 2021; Yang 2018; Xia 2021). It is noted that 1 study (Kang 2021b) had only 2 of 13 people were clinically eligible for a second RFA session, indicating the difficulty to repeat RFA in MHBO, due to some clinical obstacles such as active cholangitis. Although the results were indicative, the survival benefit after repeated RFA procedures was shown.

    In terms of safety outcomes, no statistically significant difference was found in the overall postoperative adverse events, cholangitis and bleeding between people with RFA plus stent and people with stent only across studies. Regarding cholecystitis, 2 studies (Gao 2021; Xia 2021) found that its incidence was statistically significantly higher in people with RFA plus stent than people with stent only, suggesting extra precaution would be necessary during ablation. The incidences of pancreatitis, perforation and liver abscess were generally low.

    In conclusion, most evidence focused on primary RFA, with CCA (mainly hilar and distal CCA) being the most common tumour. The evidence suggested that endoscopic bipolar RFA confers a survival benefit to people with MBO (mainly caused by CCA, particularly in those without metastasis). However, the evidence failed to exhibit a benefit for stent patency which is the primary aim of the procedure. Although endoscopic RFA, in addition to stent insertion, could improve quality of life but this was measured using KPS, so no data on quality of life using a conventional tool. More well-designed studies are warranted to evaluate the effects of endoscopic bipolar RFA in MBO caused by different tumour entities, stages and locations, in different RFA application protocols and in secondary RFA. To date, no ongoing trials specifically for this procedure have been identified.