RCT2: flexible ureterorenoscopy versus percutaneous nephrolithotomy – baseline, trial results and clinical effectiveness

RCT2 for stones > 10 mm: flexible ureterorenoscopy versus percutaneous nephrolithotomy

Trial recruitment

RCT2 compares FURS with PCNL as kidney stone treatments for larger stones (> 10 mm, ≤ 25 mm). We recruited 159 participants from 39 centres. The median number of participants per centre was 2 (IQI 1–3). We randomised 73 and 86 participants into each of the care pathways FURS and PCNL respectively. Participants were recruited between 9 May 2016 and 31 March 2021, with final follow-up completed in January 2023 (see Appendix 5, Table 38 and Figure 17).

Recruitment was paused due to the COVID-19 pandemic between 16 March 2020 and 1 July 2020. Only 20 sites had the capacity to reopen to recruitment after the recruitment pause. A further 9 participants were recruited after the study reopened to recruitment (6% of the total 159 recruited).

Participant flow

The CONSORT diagram for RCT2 (Figure 4) shows that 545 potentially eligible patients were screened and 386 were excluded. Reasons for exclusion are shown in Figure 4. Of those excluded 60 were deemed unsuitable (surgical reasons) for FURS or PCNL, 50 were ineligible due to stone size or position, and 193/386 (50%) of patients that were excluded, indicated a preference for treatment as a reason for declining to take part. There were no post-randomisation exclusions in RCT2.

FIGURE 4 Consort flowchart – RCT2. a, See Table 11. AUC, area under the curve. (Figure not reproduced here; see the original article.)

The majority [128, (80%)] of participants received their intervention and reached their 12-week post-intervention time point prior to the COVID-19 restrictions (Table 9). However, 27% (43/159) participants completed their follow-up to 12 months post randomisation after COVID-restrictions started.

Baseline characteristics

The randomised groups were mostly well balanced at baseline (Table 10). The mean age of participants was 58 years, and most were male. This trial was restricted to those with larger kidney stones (> 10 mm). Stone measures from detailed imaging (including stone Hounsfield unit density, as a marker of ‘hardness’, and skin-to-stone distance) were assessed at baseline after randomisation and were similar in both treatment arms, see Appendix 5, Table 40. In the FURS group, 29% of the participants had undergone previous stone treatments compared to 43% in the PCNL group. The level of stone-related pain that participants experienced on the day (scale 0–10) completed for the baseline questionnaire was low with median 0 days (IQI 0–3). Pain in the previous week was experienced by 44% of the participants reported. Participants reported taking pain medication on a median of 0 days (IQI 0–7) in the previous 7 days.

The health status and QoL measures taken at baseline these were comparable in both treatment groups for all the measures.

Care pathway and treatment received

The allocated treatment compared to the received treatment is described in Table 11. A total of 61 (83.6%) received their allocated pathway in the FURS arm and 64 (74.4%) received their allocated pathway in the PCNL arm. A total of 8 (11%) participants allocated to FURS had PCNL while 16 (18.6%) allocated to PCNL had FURS. A total of 10 participants did not receive any active treatment: 4 in the FURS arm and 6 in the PCNL arm. Most of these decided on conservative management with their clinician. For those participants, we used the date when conservative management or the care plan was agreed to start their 12-week post-intervention period.

The waiting times to initial treatment (see Table 11) were very similar in both treatment groups, both being surgical procedures (mean around 27 weeks) but with an equally large SD indicating a slight skewed distribution. Length of hospital stay for first intervention treatment differed slightly between treatment groups.

Primary outcome

Figure 5 shows plots of the mean EQ-5D score and the change in EQ-5D score from baseline over time. A similar pattern was seen for the EQ-5D-VAS scores (see Appendix 6, Figure 19).

FIGURE 5 Health status plots over time – RCT2. Time points B and 0 are baseline and pre intervention. Areas (a) and (b) do not necessarily represent 1 week from week 1 to 12; all are 1-week intervals (EQ-5D score: observed and differences since baseline). (Figure not reproduced here; see the original article.)

The PO and treatment comparison effect sizes are presented in Table 12; in the FURS arm 57/73 (78.1%) had sufficient measures and 63/86 (73.3%) in the PCNL arm had sufficient measures. The model effect estimates should be considered along with the original minimal clinical important difference (MCID) the study was powered to be around 0.07 (based on 0.3 SD difference and baseline data). The difference between groups consistently favours PCNL.

The per-protocol (treatment as received) analyses are presented in Appendix 7, Table 56. The treatment effects were similar to the ITT analyses. The suite of multiple imputation approaches is described in Appendix 2, Description of different scenarios for multiple imputation and presented in Appendix 7, Figure 26. The results were similar to the original ITT analyses.

Secondary outcomes

Appendix 2 gives definitions and derivations of the secondary outcomes both for clinical outcomes (stone clearance and additional treatments) and the patient reported outcomes (pain, health status and QoL). Briefly to reiterate: stone clearance was further defined as complete clearance of the target stone from the urinary tract defined, and no further action or observation required for that stone; acceptable clearance, where observation is required but no intervention planned; and unacceptable clearance, where further intervention will be required.

Stone clearance and additional treatments by 12 weeks post intervention

At 12 weeks post intervention, the FURS care pathway had fewer participants with complete stone clearance than PCNL (47.9% vs. 70.9%) with the overall effect indicating that for these larger stones PCNL is superior to FURS [OR: 0.27 (0.16, 0.43); p < 0.001] (Table 13). Further, the FURS group had a larger risk of having had additional treatments at this point compared to PCNL [RR: 3.96 (2.07, 7.57); p < 0.001].

Stone clearance and additional treatments by 12 months post randomisation

For most (80.5%) participants in this trial, the 12-month post randomisation measures were after the 12-week post-intervention time point as intended in the original design. There were a few participants [18 (12%)] who had not had their intervention (or set nominated treatment date in the case conservative management) until after they had been in the trial for 12 months. Twelve (8%), reached their 12-month post-randomisation time point somewhere between intervention and the 12-week post-intervention period. This happened in both arms (see Table 13). The individuals still waiting for treatment were classed as requiring treatment unless evidence of their stone passing was given. Again though, as with the 12-week results, FURS had a lower rate of stone clearance compared to PCNL [OR: 0.24 (0.16, 0.35); p < 0.001] and an increased rate of requiring additional treatment and/or had additional treatment [RR: 2.78 (1.69, 4.57); p < 0.001]. These were assessed in the context of where in the timeline the 12-month measures fell, taking into account that some were only half-way through their allowed care pathway.

Complications

The number of treatment-related complications was small: 16/73 (22%) participants randomised to FURS and 17/86 (20%) randomised to PCNL experiencing a complication. There were 17 complications (Table 14) in each arm with 2/73 (3%) in the FURS arm and 4/86 (5%) in the PCNL arm graded as Clavien–Dindo grade 3 or above.

Patient-reported pain, health status and quality of life

The patient-reported weekly pain and health status (EQ-5D score) are summarised in detail in Appendix 7, Tables 58–69. Appendix 7 also summarises the pre-intervention (see Table 57) and 12-month post-randomisation (see Table 70) pain, health status (EQ-5D score) and QoL (SF-12) measures.

The weekly pain measures (Table 15) reflect the health status plots (see Figure 5) – the pain measures indicate very slightly higher pain in the first weeks for those allocated to FURS compared to those on PCNL but then decrease. The secondary outcomes of health status, measured by the EQ-5D score and the EQ-5D-VAS scale, and QoL as measured by the SF-12 scores (physical and mental domains) are also reported Table 15. Both EQ-5D-5L measures show similar patterns with initial detriment seen in FURS followed by improvement after 2 to 5 weeks, statistically comparable in the PCNL group. The SF-12 measures do not show any difference by 12 weeks.

Subgroup analyses

We explored the potential impact of three subgroup factors reported at baseline: BMI, stone density as measured by the Hounsfield unit, and stone size (the definitions are in Appendix 2, Subgroup analyses and technical factors definitions). Appendix 7 gives the plots of the three PO models (main, Figure 27, sensitivity 1, Figure 28 and sensitivity 2, Figure 29) and the secondary models (AUC from baseline, Figure 30, and the clinical measure of stone clearance, Figure 31) with 99% CIs. The AUC outcomes (Figures 27–30) all favour PNCL, but not statistically and span the original overall effect estimate. This is also the case for stone clearance at 12 weeks (Figure 31), although there the division into the stone density categories maintained some significance but gave little additional information.

Technical factors

We report the results for the main outcomes and covariates in the PO models (main and two sensitivities), the secondary AUC model (AUC from baseline to 12 weeks post intervention) and also for stone clearance at 12 weeks for the technical factors in Appendix 8 (see Tables 71–75). These were not a between treatment arm comparison (since they are within a specified treatment) but were summaries of each of the outcomes and covariates for each technical factor across the two RCTs. They were restricted to time of the first intervention.