3 Clinical evidence

3 Clinical evidence

Summary of clinical evidence

3.1 The key clinical outcomes for the transurethral resection in saline (TURis) system for transurethral resection of the prostate presented in the decision problem were:

  • hospital length of stay

  • procedural blood loss and blood transfusion

  • time to removal of urinary catheter postoperatively

  • transurethral resection syndrome

  • readmission for repeat procedures

  • duration of surgical procedure

  • healthcare‑associated infection

  • quality of life

  • device‑related adverse events.

3.2 The company identified a total of 1116 studies in their database searches, and presented 24 studies in their submission as relevant to the decision problem. These included 14 randomised trials, not all of which were published in full or in English, with a total of 3032 patients (Abascal Junquera et al. 2006; Akman et al. 2013; Chen et al. 2009, 2010; Fagerstrom et al. 2010, 2011; Goh et al. 2009, 2010; Gulur et al. 2010a, 2010b; Michielsen et al. 2007, 2010a, 2010b; Rose et al. 2007) and 10 observational studies (Bertolotto et al. 2009; Fumado et al. 2011; Giulianelli et al. 2012; Ho et al. 2007; Jun Hyun et al. 2012; Lee et al. 2011; Michielsen et al. 2010c, 2011; Petkov et al. 2011; Puppo et al. 2009).

3.3 The External Assessment Centre considered the 14 randomised trials described in the submission. It established that the 3 randomised studies and 2 observational studies published by Michielsen reported on various stages and subgroups of the same study population. It also considered that the 2 papers from Fagerstrom were based on the same study population, and that the 4 conference abstracts (Goh et al. 2009, 2010; Gulur et al. 2010a, 2010b) were based on the same study population. Two studies were not published in English but have English abstracts (Abascal Junquera et al. 2006; Rose et al. 2007). The External Assessment Centre considered that, of these, only the Rose et al. (2007) paper contained pivotal results and it obtained a translation of the paper; the other was not considered pivotal. A literature search by the External Assessment Centre identified 2 further randomised studies (Geavlete et al. 2011; Ho et al. 2006). In total the External Assessment Centre considered that there were 10 unique randomised studies (1870 patients) relevant to the decision problem, 9 published as papers (including 2 foreign language papers with English abstracts) and 1 abstract.

3.4 The company presented 10 observational studies, 5 of which were published in full and 5 of which were abstracts only. The External Assessment Centre established that the Michielsen et al. (2010 and 2011) studies reported on subgroups from the randomised study by Michielsen et al. published in 2007. A literature search by the External Assessment Centre identified 1 additional observational study (Shum et al. 2014). The External Assessment Centre considered that there were 4 published papers and 5 abstracts describing relevant observational studies. It agreed with the company's conclusion that the outcomes reported from the observational studies were consistent with those from the randomised trials. The observational studies are summarised in the assessment report and are not considered further here.

Randomised trials: published papers

3.5 Akman et al. (2013) reported a Turkish study of 286 men (143 in each group) randomised to have either TURis or monopolar transurethral resection of the prostate (TURP) who were followed‑up for 12 months. The mean procedure duration was 54.0 minutes for TURis and 58.7 minutes for monopolar TURP, p=0.03. The incidence of TUR syndrome was 0% for TURis and 1.5% for monopolar TURP (no p value reported). There was no statistically significant difference in the length of hospital stay for the TURis group compared with the monopolar TURP group (2.5 days compared with 2.7 days, no p value reported). The rate of blood transfusion was lower in the TURis group (2.4% compared with 6.2%) but the difference was not statistically significant (p=0.2). There were lower rates of clot retention (0.8% compared with 1.5%, p value not reported) and mean time to catheter removal (2.4 days compared with 2.6 days, p value not reported) for TURis.

3.6 The Chen et al. (2009) study was done in China on 45 men with symptomatic benign prostatic hypertrophy and a large prostate gland, randomised to have either TURis or monopolar TURP. Results were analysed for 40 men, with reasons given for withdrawals. The results showed that average procedure duration was shorter in the TURis group compared with the monopolar TURP group (88 minutes compared with 105 minutes, p=0.001). No men in the TURis group had TUR syndrome, compared with a 5% rate (n=1/19) in the monopolar TURP group. Fewer men had a blood transfusion in the TURis group (4.8% compared with 15.5%, p value not reported). There was no statistically significant difference between groups in the time to catheter removal (2.5 days compared with 3.4 days, p=0.11). However there was a statistically significant reduction in length of hospital stay for the TURis group (3 days compared with 4.2 days, p=0.001).

3.7 Chen et al. (2010) reported a separate study of 100 men in China randomised to have either TURis or monopolar TURP. There was no statistically significant difference in procedure duration in the TURis group compared with the monopolar TURP group (59 minutes compared with 60 minutes, p=0.82) or weight of tissue resected (40 g compared with 38.9 g, p=0.31). No patient in either group had TUR syndrome. One man in the TURis group and 3 men in the monopolar TURP group needed a blood transfusion (2% compared with 6%, p=0.62).

3.8 The Fagerstrom et al. (2009 and 2011) studies were performed in Sweden on 202 men randomised to have either TURis or monopolar TURP. Results were analysed for 185 men, with reasons given for withdrawals. Results showed that there was no statistically significant difference between the TURis and monopolar TURP group in mean procedure time (62 minutes compared with 66 minutes, p not significant) or weight of tissue resected (27.3 g compared with 26.3 g, p not significant). No patient developed TUR syndrome in the TURis group, but 3 did so in the monopolar TURP group. A statistically significantly lower proportion of men in the TURis group had a blood transfusion (4% compared with 11%, p<0.01). Median time to catheter removal was the same in both groups (20 hours), and the length of stay in hospital was similar (51 hours compared with 52 hours). There was a statistically significant reduction in the rate of readmission in the TURis group (n=5/98 compared with n=14/87, p<0.011).

3.9 The Geavlete et al. (2011) study involved 510 men in Romania who were randomised to 3 study arms (170 in each arm). Results are reported here for the TURis and monopolar TURP arms (340 patients), but not for the bipolar plasma vaporisation of the prostate arm which was considered to be outside the scope. Statistical analysis was performed on the difference between the 3 groups and is not reported here. The average procedure duration was 52.1 minutes in the TURis group and 55.6 minutes in the monopolar TURP group. No men had TUR syndrome in the TURis group compared with 3 men (1.8%) in the monopolar TURP group. In the TURis group 3 men (1.8%) needed a blood transfusion, compared with 11 men (6.5%) in the monopolar TURP group. In the TURis group 2 men (1.2%) had clot retention compared with 7 men (4.1%) in the monopolar TURP group. The mean time to catheter removal was 46.3 hours (range 36–72 hours) in the TURis group compared with 72.8 hours (range 48–96 hours) in the monopolar TURP group. In the TURis group length of stay in hospital was 3.1 days compared with 4.2 days in the monopolar TURP group.

3.10 The Ho et al. (2007) study was performed in Singapore on 48 men randomised to TURis and 52 men randomised to monopolar TURP. There was no statistically significant difference in mean procedure duration between the groups (59 minutes for TURis compared with 58 minutes for monopolar TURP) or in the weight of tissue resected (29.8 g TURis compared with 30.6 g monopolar TURP). There was a statistically significantly lower rate of TUR syndrome in the TURis group compared with the monopolar TURP group (0 men compared with 2 men, p<0.005). One patient in each group needed a blood transfusion. In the TURis group 3 men had clot retention compared with 2 men in the monopolar TURP group; this difference was not statistically significant.

3.11 The Michielsen et al. (2007) study recruited patients between January 2005 and June 2006 in Belgium. However, recruitment into the study continued until August 2009, leading to subsequent papers reported as randomised (Michielsen et al. 2010a, 2010b) and observational studies (Michielsen et al. 2010c, 2011). In total 550 patients were included in the study; 285 in the TURis group and 265 in the monopolar TURP group, but some outcomes were reported on smaller groups. There was no significant difference between the TURis group (n=263) and monopolar TURP group (n=255) in mean procedure duration (52.1 minutes compared with 50.9 minutes, p=0.357) or mean weight of tissue resected (17.6 g compared with 19.2 g, p=0.173). TUR syndrome did not occur in the TURis group and occurred twice (0.8%) in the monopolar TURP group (p value not reported). In the TURis group (n=118) 4 men (3.4%) needed a blood transfusion compared with 1 patient (0.8%) in the monopolar TURP group (n=120, p=0.211). There was no statistically significant difference in mean length of hospital stay: 3.72 days in the TURis group (n=263) and 3.89 days in the monopolar TURP group (n=255, p=0.773). No patients in the TURis group (n=118) and 2 patients in the monopolar TURP group (n=120) needed a repeat procedure because of incomplete resection (p value not reported).

3.12 The Rose et al. (2007) study was published in German and the External Assessment Centre obtained an English translation. It included 38 men who had TURis and 34 men who had monopolar TURP (the remainder had treatment for bladder cancer) in Germany. Mean procedure duration was longer in the TURis group than in the monopolar TURP group (55 minutes compared with 35 minutes, p=0.005), but the mean weight of tissue resected tended to be greater in the TURis group (42 g compared with 31 g, p value not reported). No men had TUR syndrome in either group. The mean time to catheter removal was longer in the TURis group (64 hours compared with 49 hours, p value not reported) and the TURis group had a higher rate of readmission because of haemorrhage (n=4/38 compared with n=1/34, p value not reported).

3.13 The Abascal Junquera et al. (2006) study was published in Spanish with an English abstract that had limited information on the statistical analysis. The External Assessment Centre considered that the study did not provide additional important data and the paper was therefore not translated. In this study 45 men were prospectively randomised, with 24 men having TURis and 21 men having a TURP procedure using a monopolar system. TURis was a slightly quicker procedure compared with monopolar TURP (39.7 minutes compared with 42.7 minutes) based on a similar resection weight (13 g for TURis compared with 12.6 g for monopolar TURP). The time to removal of the catheter was similar between the groups (2.92 days for TURis compared with 3.1 days for monopolar TURP, not statistically significant) as was the length of hospital stay (3.63 days for TURis compared with 3.67 days for monopolar TURP).

Randomised trials: abstracts

3.14 The Goh et al. (2009 and 2010); and Gulur et al. (2010a and 2010b) conference abstracts relate to the same multicentre study (country not reported). In this study, 210 men with benign prostatic obstruction were randomly allocated to TURis (n=110) or monopolar TURP (n=100). The study reported a similar procedure duration for TURis compared with monopolar TURP (38 minutes compared with 35 minutes, not statistically significant). There were no cases of TUR syndrome in the TURis group and 3 (3%) in the monopolar TURP group (p value not reported). Men in the TURis group tended to have a shorter time to catheter removal (48 hours compared with 52 hours, p=0.97), and a shorter hospital stay (90 hours compared with 103 hours, p=0.06) but neither result was statistically significant.

Meta-analysis of evidence

3.15 The company presented fixed‑effect meta‑analyses of the randomised studies for procedure‑related outcomes between TURis and monopolar TURP for TUR syndrome, clot retention, procedure duration, time to catheter removal, length of hospital stay and procedural blood loss. The results are described in sections 3.17–3.22 with further details in the assessment report on pages 81–98. A summary of the results is presented in table 1.

3.16 The External Assessment Centre did not agree with the included studies used for some outcomes in the company meta‑analyses. It did revised meta‑analyses with changes in the selected studies, investigated additional outcomes and explored using either fixed‑ or random‑effects methods. The results of the External Assessment Centre revised meta‑analyses are shown in table 1.

Table 1 Results of company's meta-analyses and the External Assessment Centre revised meta-analyses (all fixed effects)

Outcome

Company's meta‑analysis

External Assessment Centre's revised meta‑analysis

Studies (n)

Relative risk for TURis (95% CI)

Studies (company studies)

Relative risk for TURis (95% CI)

TUR syndrome

6

0.28 (0.08 to 1.02)

6 (2)

0.18 (0.05 to 0.62)

Blood transfusion

3

0.36 (0.16 to 0.80)

6 (3)

0.35 (0.19 to 0.65)

Clot retention

2

0.63 (0.21 to 1.90)

5 (2)

0.55 (0.26 to 1.15)

Studies (n)

Mean difference for TURis (95% CI)

Studies

Mean difference for TURis (95% CI)

Hospital stay (days)

3

−0.52 (−0.74 to −0.30)

2 (2)

−0.19 (−0.46 to 0.07)

Time to removal of catheter (days)

3

−0.23 (−0.38 to −0.08)

2 (2)

−0.09 (−0.25 to 0.06)

Procedure time (minutes)

4

−1.68 (−4.18 to 0.81)

5 (4)

−1.36 (−3.70 to 0.98)

CI, confidence interval; TURis, transurethral resection in saline; TUR, transurethral resection.

3.17 The company included 6 studies presenting results assessing the risk of TUR syndrome (Abascal Junquera et al. 2006; Akman et al. 2013; Chen et al. 2010; Goh et al. 2010; Michielsen et al. 2011; Rose et al. 2007). The company applied a continuity correction to account for the zero event rate in all TURis arms, replacing nil values with 0.5. They found a non‑statistically significant lower pooled relative risk in favour of TURis of 0.28 (95% confidence interval [CI] 0.08 to 1.02). The External Assessment Centre repeated the company's meta‑analysis, excluding 4 studies: 3 studies in which there were no cases of TUR syndrome in either arm, and the results from the conference abstract by Goh et al. (2010). The External Assessment Centre added data from 4 randomised studies that the company did not include (Ho et al. 2006; Chen et al. 2009; Fagerstrom et al. 2011; Geavlete et al. 2011). This revised meta‑analysis found a statistically significant effect in favour of TURis: relative risk 0.18 (95% CI 0.05 to 0.62, p=0.006), corresponding to a number needed to treat to prevent 1 case of TUR syndrome compared with monopolar TURP of 50.

3.18 The company's meta‑analysis of trials presenting data on blood transfusion gave a pooled relative risk of 0.52 (95% CI 0.26 to 1.04) in favour of TURis based on 4 studies (Akman et al. 2013; Chen et al. 2010; Fagerstrom et al. 2011; Michielsen et al. 2007). The company re‑ran this analysis, excluding Michielsen et al. (2007) because a higher proportion of procedures were carried out by trainee surgeons in the TURis arm of that study. This gave a pooled relative risk of 0.36 (95% CI 0.16 to 0.80) in favour of TURis. The External Assessment Centre agreed with this approach and repeated the analysis, adding data from 3 further studies (Chen et al. 2009; Ho et al. 2006; Geavlete et al. 2011). The result was a statistically significant effect in favour of TURis with a relative risk of 0.35 (95% CI 0.19 to 0.65, p=0.0008). The External Assessment Centre calculated the number needed to treat to prevent 1 case of blood transfusion compared with monopolar TURP) as 20.

3.19 For clot retention, the company's meta‑analysis included 2 studies (Akman et al. 2013; Michielsen et al. 2007) and found a relative risk in favour of TURis of 0.63 (95% CI 0.21 to 1.90; not statistically significant). The External Assessment Centre re‑ran the meta‑analysis adding 3 further studies (Chen et al. 2010; Geavlete et al. 2011; Ho et al. 2006) giving a revised pooled relative risk of 0.55 (95% CI 0.26 to 1.15, p=0.11).

3.20 For length of hospital stay, the company conducted a meta‑analysis on 3 trials presenting data on length of hospital stay (Akman et al. 2013; Chen et al. 2009; Michielsen et al. 2011) which revealed a pooled mean difference between the groups (TURis minus monopolar TURP) of −0.52 days (95% CI −0.74 to −0.30, p=0.0001). The External Assessment Centre examined the impact of the study by Chen et al. (2009), which was a source of significant heterogeneity and considered that it should be excluded. The External Assessment Centre calculated a pooled mean difference in length of hospital stay between the groups (TURis minus monopolar TURP) of −0.19 days (95% CI −0.46 to 0.07, p=0.16) which was not statistically significant.

3.21 The company included 3 randomised studies (Akman et al. 2013; Chen et al. 2009, Michielsen et al. 2010) in its analysis of mean time to removal of the urinary catheter and reported a significantly shorter time in favour of TURis of −0.23 days (95% CI −0.38 to −0.08). The External Assessment Centre excluded the Chen et al. (2009) study because it introduced significant heterogeneity to the analysis and presented a result based on 2 studies (Akman et al. 2013; Michielsen et al. 2010) which gave a non‑statistically significant pooled mean difference (TURis minus monopolar TURP) for time to catheter removal of −0.09 days (95% CI −0.25 to 0.06).

3.22 The company's meta‑analysis of trials presenting data for procedure duration included 4 papers (Akman et al. 2013; Chen et al. 2010; Fagerstrom et al. 2011; Michielsen et al. 2010), and found a non‑significant mean difference (TURis minus monopolar TURP) of −1.68 minutes (95% CI −4.18 to 0.81). The External Assessment Centre agreed with the exclusion of Michielsen et al. (2007) in the company's initial analysis but considered the addition of 2 further studies (Chen et al. 2009; Ho et al. 2006). After the External Assessment Centre explored the heterogeneity of the meta‑analysis calculations, it presented a result based on 5 studies, which gave a non‑statistically significant pooled mean difference in procedure time in favour of TURis of −1.36 minutes (95% CI −3.70 to 0.98, p=0.26).

3.23 The External Assessment Centre examined 3 further outcomes that were not included in the company's meta‑analysis. For readmission because of haemorrhage, data from 3 randomised studies were used (Fagerstrom et al. 2011; Geavlete et al. 2011; Rose et al. 2007) and the result was a non‑statistically significant lower rate for TURis, with a relative risk of 0.53 (95% CI 0.22 to 1.25, p=0.15). The External Assessment Centre also conducted a meta‑analysis on urethral strictures and bladder neck contractures because this was highlighted as a potential concern with TURis by expert advisers. This analysis included 5 studies (Ackman et al. 2013; Chen et al. 2010; Fagerstrom et al. 2011; Geavlete et al. 2011; Michielsen et al. 2011) and found no statistically significant difference between the groups, with a relative risk of 1.08 (95% CI 0.70 to 1.69, p=0.72). The third additional outcome considered by the External Assessment Centre was repeat procedure because of incomplete resection. This analysis included 3 studies (Fagerstrom et al. 2011; Geavlete et al. 2011; Michielsen et al. 2011) and found no statistically significant difference between the groups: relative risk 0.76 (95% CI 0.42 to 1.40, p=0.38).

Committee considerations

3.24 The Committee considered that the evidence demonstrated the clinical equivalence of TURis and monopolar TURP for prostatic resection. The Committee noted there was evidence showing that the TURis system reduces the risk of TUR syndrome and reduces patients' need for blood transfusion as compared with monopolar TURP.

3.25 The Committee considered length of hospital stay derived from the meta‑analyses by the company and by the External Assessment Centre. It discussed the rationale for excluding the Chen et al. (2009) study. The External Assessment Centre confirmed that it excluded the Chen et al. (2009) study because it was the source of significant heterogeneity in the meta‑analysis results. However, the External Assessment Centre stated that it did not differ in terms of methodological quality from the 2 included studies. The Committee noted that all the trials were based outside the UK and heard expert advice that local policies on healthcare reimbursement and hospital‑specific catheter guidelines could have an effect on length of hospital stay. The Committee concluded that there was a possibility that TURis would result in shorter hospital stays, but that clinical trial data were inconclusive.

3.26 The Committee discussed readmission to hospital after resection and noted that this outcome was not included in most of the clinical trials. However, it noted a non‑statistically significant lower rate of readmission because of bleeding for TURis compared with monopolar TURP in the data from 3 trials included in a meta‑analysis. The Committee also noted that the readmission rate reported in the Fagerstrom et al. (2011) study showed a statistically significant reduction in the TURis group compared with the monopolar TURP group (n=5/98 compared with n=14/87, p<0.011). In addition, it heard expert advice based on experience of the use of TURis in the NHS, which suggested that there was indeed a reduction in readmissions due to bleeding seen in clinical practice. Based on the evidence, the Committee concluded that it was plausible that TURis would result in lower readmission rates, although the evidence was not definitive.

3.27 The Committee considered the other outcomes from the meta‑analysis and noted no statistically significant differences between TURis and monopolar TURP in procedure time, time to catheter removal, the incidence of clot retention and incidence of urethral stricture or bladder neck contracture.

  • National Institute for Health and Care Excellence (NICE)