Comparison of stone-free rates following shock wave lithotripsy, percutaneous nephrolithotomy, and retrograde intrarenal surgery for treatment of renal stones: A systematic review and network meta-analysis

Objectives To perform a systematic review and network meta-analysis comparing stone-free rates following retrograde intrarenal surgery (RIRS), extracorporeal shock wave lithotripsy (SWL), and percutaneous nephrolithotomy (PCNL) treatments of renal stones. Materials and methods Clinical trials comparing RIRS, SWL, and PCNL for treatment of renal stones were identified from electronic databases. Stone-free rates for the procedures were compared by qualitative and quantitative syntheses (meta-analyses). Outcome variables are shown as risk ratios (ORs) with 95% credible intervals (CIs). Results A total of 35 studies were included in this network meta-analysis of success and stone-free rates following three different treatments of renal stones. Six studies compared PCNL versus SWL, ten studies compared PCNL versus RIRS, fourteen studies compared RIRS versus SWL, and five studies compared PCNL, SWL, and RIRS. The quality scores within subscales were relatively low-risk. Network meta-analyses indicated that stone-free rates of RIRS (OR 0.38; 95% CI 0.22–0.64) and SWL (OR 0.12; 95% CI 0.067–0.19) were lower than that of PCNL. In addition, stone-free rate of SWL was lower than that of RIRS (OR 0.31; 95% CI 0.20–0.47). Stone free rate of PCNL was also superior to RIRS in subgroup analyses including ≥ 2 cm stone (OR 4.680; 95% CI 2.873–8.106), lower pole stone (OR 1.984; 95% CI 1.043–2.849), and randomized studies (OR 2.219; 95% CI 1.348–4.009). In rank-probability test, PCNL was ranked as No. 1 and SWL was ranked as No. 3. Conclusions PCNL showed the highest success and stone-free rate in the surgical treatment of renal stones. In contrast, SWL had the lowest success and stone-free rate.


Introduction
Urinary tract calculi, one of the most common benign urological diseases, is seen in 12% of patients and has a recurrence rate of approximately 50% [1,2]. Factors that may play an important role in the increase of urinary tract stone disease include increases in diagnosis of metabolic syndrome, lifestyle changes, dehydration, lack of water intake, and low urine volume [3]. Furthermore, recent studies have shown that the worldwide increase of renal colic and renal stones is affected by seasonal changes, particularly the hot season, and that global warming is capable of increasing the incidence of renal stones [4]. In particular, renal uric acid stones show a tendency to increase in hot and dry climates because of the reduction of urine excretion and urine pH [5].
The European Association of Urology (EAU) Urolithiasis Guidelines suggest that the primary treatment of renal stones <2 cm should include extracorporeal shock wave lithotripsy (SWL) and retrograde intrarenal surgery (RIRS) and that the primary treatment for renal stones >2 cm should include percutaneous nephrolithotomy (PCNL) [6]. In cases of 1-2-cm lower pole renal stones, RIRS or PCNL is recommended if there are unfavorable factors in SWL. In comparison with PCNL and RIRS, SWL plays a pivotal role in the treatment of urinary tract stones because it is the only interventional treatment with non-invasive properties [7]. In contrast with SWL, RIRS can perform stone dusting and fragmentation under endoscopic direct vision and has the advantage of being able to directly remove the fragmented stone using a stone basket [8]. PCNL is the standard treatment for large, renal stones (>2 cm) and can also be considered as a treatment option for large stones with resistance to shock waves [9]. Though prospective studies and a meta-analysis of the three treatments along with their advantages and disadvantages have been reported, a network meta-analysis that compares all three treatments at the same time has not yet been reported. Network meta-analysis is a research method that can compare multiple treatments using direct comparison and indirect comparison methods [10][11][12]. Therefore, we performed a systematic review and a network meta-analysis analysis that compares the success as well as the stone-free rates of SWL, RIRS, and PCNL. outcome parameters also included complication rate; (vi) the full text of the study was available in English. This report was prepared in compliance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (accessible at http://www.prismastatement.org/) [13]. The protocol for this study is shown in S1 Table.

Search strategy
A literature search of all publications before 31 June 2016 was performed using EMBASE and PubMed. Additionally, a cross-reference search of eligible articles was performed to identify studies that were not found during the computerized search. The proceedings of appropriate meetings were also searched. Combinations of the following MeSH terms and keywords were used: extracorporeal shock wave lithotripsy, shock wave lithotripsy, percutaneous nephrolithotomy, nephrolithotomy, percutaneous, flexible ureteroscopy, flexible ureterorenoscopy, retrograde intrarenal surgery, renal stone, urolithiasis, rate, and stone-free (S2 Table).

Data extraction
Two researcher (DYC and DHK) screened all titles and abstracts identified by the search strategy. Two other researchers (HDJ and JKK) independently evaluated the full text of each paper to determine whether it met the inclusion criteria. Disagreements were resolved by discussion until a consensus was reached or by arbitration mediated by another researcher (JYL).

Quality assessment for studies
When the final group of articles was agreed upon, two researchers independently examined the quality of each article using the Downs and Black checklist. The Downs and Black checklist was developed for the purpose of quality assessment of both randomized and nonrandomized studies of health interventions [14]. The checklist consists of five subscales: reporting, internal validity bias, internal validity confounding, external validity, and power. Because six items in the original list were related to intervention, randomization, and power calculation, and not all of the studies examined were randomized studies, the scores for these six items were counted as zero, as suggested in a previous study [15]. Therefore, the maximum quality score was 31 points. A higher score was considered to be an indicator of a good quality study.

Heterogeneity tests
Heterogeneity of included studies was examined using the Q statistic and Higgins' I 2 statistic [16]. Higgins' I 2 measures the percentage of total variation due to heterogeneity rather than chance across studies. Higgins' I 2 was calculated as follows:

Ethics statement
The study was exempt from requiring the participants' written informed consent because this is systematic review and network meta-analysis. The approval of the Institutional Review Board was also exempted.

Statistical analysis
Outcome variables measured at specific time points were compared in terms of odds ratios (OR) or mean differences with 95% CIs using a network meta-analysis. Analyses were based on non-informative priors for effect sizes and precision. Convergence and lack of auto-correlation were confirmed after four chains and a 50,000-simulation burn-in phase. Finally, direct probability statements were derived from an additional 100,000-simulation phase. The probability that each group had the lowest rate of clinical events was assessed by Bayesian Markov Chain Monte Carlo modeling. Sensitivity analyses were performed by repeating the main computations with a fixed-effect method. Model fit was appraised by computing and comparing estimates for deviance and deviance information criterion. All statistical analyses were performed with R (R version 3.5.1, R Foundation for Statistical Computing, Vienna, Austria; http://www.r-project.org) and the associated meta, netmeta, pcnetmeta, and gemtc packages for pairwise and network meta-analyses.

Eligible studies
The database search retrieved 35 articles covering 237 studies for potential inclusion in metaanalysis. Eight articles were excluded according to the inclusion/exclusion criteria; three had no data on stone-free rate, three were reviews, and two reported case series. The remaining 35 articles were included in the qualitative and quantitative syntheses using pairwise and network meta-analyses (Fig 1).

Quality assessment
The results of quality assessment based on the Downs and Black checklist are shown in Table 1. The median of the total quality scores was 14.8. Overall, the quality scores within subscales were relatively low. In most studies, external validity was not satisfactory for both significant and insignificant groups.

Heterogeneity and inconsistency assessment and publication bias
Forest plots of the pairwise meta-analysis of SWL, PCNL, and RIRS are shown in Figs 3, 4 and 5, respectively. There was no heterogeneity between PCNL and RIRS; however, there was heterogeneity between PCNL and SWL and between SWL and RIRS in each study. Thus, random-effect models were applied using the Mantel-Haenszel method for PCNL and SWL analysis and SWL and RIRS comparison (Figs 4 and 5). After selection of effect models, little heterogeneity was noted in L'Abbe plots and radial plots (Figs 6 and 7).
In node-splitting analysis, no inconsistency was demonstrated in direct, indirect, or network comparison (Fig 8). A net-heat plot showed that there was also little inconsistency in the whole network (Fig 9).
The Begg and Mazumdar rank correlation tests for each analysis showed no evidence of publication bias in the present meta-analysis between PCNL and SWL (P = 0.697). However, Egger's regression intercept tests revealed a slight publication bias (P = 0.041). According to a rank correlation test (P = 0.520) and regression tests (P = 0.771), there was no publication bias       in PCNL and RIRS. Also, no publication bias was shown for SWL versus RIRS in the rank correlation test (P = 0.421) and regression test (P = 0.855). However, there was little publication bias from funnel plots in each comparison (Fig 10).  Network meta-analysis of stone-free rates following SWL, PCNL, and RIRS for renal stones

Pairwise meta-analysis of SWL, PCNL, and RIRS for stone-free rate
Pooled data that were used to compare the stone-free rate between PCNL and RIRS showed a significantly higher stone-free rate with PCNL (OR 2.493; 95% CI 1.708-3.637; P<0.001; Fig  3). Network meta-analysis of stone-free rates following SWL, PCNL, and RIRS for renal stones P<0.001; Fig 4). The stone-free rate of SWL was lower than that RIRS (OR 0.352; 95% CI 0.223-0.557; P<0.001; Fig 5).

Network meta-analysis of SWL, PCNL, and RIRS for stone-free rate
In network meta-analyses, the stone-free rate of RIRS was lower than that of PCNL (OR 0.38; 95% CI 0.22-0.64), the stone-free rate of SWL was lower than that of PCNL (0.12; 95% CI 0.067-0.19), and the stone-free rate of SWL was lower than that of RIRS (OR 0.31; 95% CI 0.20-0.47) (Fig 9). In the rank-probability test, PCNL was ranked as No. 1 and SWL was ranked as No. 3 (Fig 11). The P-score test using a frequentist method to rank treatments in the network demonstrated PCNL (P-score 1.0) was superior to RIRS (P-score 0.5) and SWL (Pscore 0) in stone-free rate [56]. Network meta-analysis of stone-free rates following SWL, PCNL, and RIRS for renal stones

Subgroup analyses using stone size, location of renal stone, and study design
In � 2 cm stones, seven studies were included.

Discussion
The use of minimally invasive techniques like SWL, PCNL, and RIRS, has developed dramatically despite the continued high incidence and recurrence of urinary tract stone disease. [57]. The minimally invasive techniques for treatment of renal stones, have continuously improved over the last 30 years, and new procedures are being introduced as a result of the combination of instruments and technology that is now taking place. Since Fernstrom and Johansson introduced PCNL as the surgical treatment for patients with large and complex renal calculi for the first time in 1976 [58], PCNL has been considered as the standard surgery for the treatment of renal stones >2 cm [9]. The procedure was developed in the sequential order of tubeless PCNL, supine PCNL, and mini-PCNL [59][60][61]. Further changes in the PCNL procedure led to the recent development of endoscopic combined intrarenal surgery (ECIRS) [62]. The first Network meta-analysis of stone-free rates following SWL, PCNL, and RIRS for renal stones experience of SWL was reported in 1984, when Chaussy and his colleagues performed SWL on 852 patients [63]. Until recently, the advancement of patient selection, shock wave delivery, and the new lithotripter design were the reasons why SWL is was still the primary treatment for non-lower pole renal stones <2 cm [7]. RIRS has achieved rapid development since the 1990's when the holmium:yttrium aluminum garnet (YAG) laser system was introduced [64]. The development of the recently introduced small-aperture digital video scope (Flex-Xc; Karl Storz Endoskope, Tuttlingen, Germany, URF-V2; Olympus Corp, Tokyo, Japan) and the single-use video scope (LithoVue; Boston Scientific, Marlborough, MA, USA) has led to the popularization of RIRS by improving both the image quality as well as durability [65,66]. In most cases of non-symptomatic kidney stones, observation is sufficient. However, treatment is recommended in cases in which stones are continuously increasing in size, there is a high risk of additional stone formation, there is obstruction due to the stones, infection, pain, or hematuria, or stones are >1.5 cm. Treatment is also recommended if it is desired with regard to the patient's social situation [67]. As mentioned earlier, the EAU guideline suggests SWL and RIRS for the primary treatment of renal stones <2 cm, and PCNL for the primary treatment for stones >2 cm. In general, PCNL is more invasive than RIRS and SWL and has Network meta-analysis of stone-free rates following SWL, PCNL, and RIRS for renal stones relatively large complications related to hemorrhaging. Though the procedure of SWL is relatively safe, there is a possibility of repeated treatment. RIRS is also expanding in use due to the gradual development of related systems, but there can be technical difficulties and surgical complications may occur. Hence, there are advantages and disadvantages for each interventional treatment, and it is extremely important to find and perform the best treatment for the individual patient with the renal stones.
Perhaps stone-free rate is one of the first things to consider when choosing among treatments that have their own advantages and disadvantages. This report is the first of a network meta-analysis on the success or stone-free rates of SWL, PCNL, and RIRS. A pairwise metaanalysis comparing each method has already been reported several times. In the pairwise meta-analysis of PCNL and RIRS reported in 2015, the complication rate (OR 1.61; 95% CI 1.11-2.35), hemoglobin drop (MD 0.87; 95% CI 0.51-1.22), and the hospital stay (MD 1.28; 95% CI 0.79-1.77) of RIRS showed better results than PCNL [68]. However, the stone-free rate of PCNL was higher than that of RIRS (OR 2.19; 95% CI 1.53-3.13, P<0.001). In our study, Network meta-analysis of stone-free rates following SWL, PCNL, and RIRS for renal stones the pairwise meta-analysis of PCNL and RIRS showed better results of PCNL in terms of the stone-free rate (OR 2.31; 95% CI 1. 45-3.67). Either in the network meta-analysis, RIRS showed a lower stone-free rate than PCNL (OR 0.36; 95% CI 0.19-0.68). In another study, Zhang and colleagues performed pairwise meta-analyses of SWL, PCNL, and RIRS for the lower pole renal stone, and found that PCNL shows a higher stone-free rate than SWL and RIRS, and there is no difference in the stone-free rates of SWL and RIRS (OR 1.97; 95% CI 0.98-3.95) [69]. Our results also show PCNL had the best stone-free rate, but the results for SWL and RIRS differ between our study and that of Zhang et al. These authors argue that residual fragments should be considered more seriously for the lower pole stone than for other locations because gravity plays a crucial role in the clearance of the residual stone fragments. In particular, they predict that the increase in laser dusting without stone extraction in the mini-PCNL and RIRS treatments will play a role in lowering the stone-free rate to values similar to that for the fragments clearance using SWL, and that this prediction explains why the stone-free rate does not differ between SWL and RIRS treatments in their study. Donaldson et al reported meta-analysis on clinical effectiveness of SWL, RIRS and PCNL for lower pole stone [70]. They concluded that PCNL and RIRS were superior to SWL in clearing the stones within 3 months. In their study, they used pair-wise meta-analysis for the outcomes in patients with only lower pole stone. We also performed subgroup analyses with lower pole stone data using Bayesian network meta-analysis and the results of our study also demonstrated similarities to those by Donaldson et al., but we reaffirmed the superiority of PCNL and RIRS using network metaanalysis. In EAU guidelines, in lower pole stone, PCNL and RIRS should be recommended as the first-line treatment [6]. In our analysis, the reason why RIRS showed a higher stone-free rate than SWL was because our research included all renal stones regardless of their location, whereas the analysis performed by Zhang and colleagues included only lower pole renal stones. Network meta-analysis of stone-free rates following SWL, PCNL, and RIRS for renal stones Furthermore, our results may differ from those of their research because a higher number of studies were included in our meta-analysis. The technical development of RIRS can be another reason for the differing results. A recent survey of 414 surgeons indicates that the dusting technique using high-power holmium laser is popular and that this technique is judged to be a help in improving the stone-free rate of RIRS [71]. The lower pole stone has been reported to be used in 55.8% of cases of translocation using the stone basket. In the case of RIRS and even focusing on the lower pole stones, stones <2 cm may increase the stone-free rate through translocation [72]. There was no difference in the stone-free rate (RR 0.95; 95% CI 0.88-1.02, P = 0.15) shown in the pairwise meta-analysis of RIRS and PCNL for renal stones >2 cm reported by Zheng et al [73]. This is quite different from our meta-analysis results because Zheng and colleagues did not provide a clear quality assessment, there was a factor of publication bias, and it is presumed that the suitability of the effect model was not evaluated using the Labble plot. These conflicting results indicate that additional research is still needed.
Finally, without factoring the size and location of renal stones, the results presented in our study show that PCNL treatment resulted in the highest stone-free rate and SWL exhibited the lowest stone-free rate. Our study is unique in that three treatments were analyzed simultaneously using a network meta-analysis model. Furthermore, our study is judged to have great value because it is the first study to derive the superiority of a treatment using the rank test and because only studies with low bias and high quality were included in the analysis using quality assessment. Especially, in large stone (> 2 cm) and lower pole stone, PCNL can be superior to RIRS and SWL. EAU guidelines also recommended PCNL as the first-line treatment in large stone and lower pole stones. So far, the success rates of RIRS and SWL seem to not exceed that of PCNL. Based on our results, further research for treatments with higher stone-free rates will be necessary in the future.
The recently presented ECIRS is a treatment comprising a combination of PCNL and RIRS and is predicted to be capable of achieving a higher stone-free rate [74]. PCNL and RIRS should be the mainstay of interventional therapy for patients with renal stones. However, for some patients with bilateral disease, ECIRS may also be an effective treatment rather than bilateral PCNL or RIRS [75,76]. Although PCNL is the most effect interventional therapy with the highest stone-free rate, careful patient selection is required because of the high invasiveness of this treatment. Indeed, recent reports highlight the advantage of reduced invasiveness in mini-PCNL and ultramini-PCNL treatments [77] and successful results in treatments with ECIRS performed with mini-PCNL [78]. In summary, PCNL is the most effective treatment, and RIRS is able to compensate for a lower stone-free rate than PCNL. For patients with a low stone-free rate in the recently presented nephrolithometry score [79], increasing the stone-free rate by using ECIRS should be the goal of interventional therapy in the future [76].
A limitation of our study is that no subgroup analysis was performed on the size and location of the renal stones. In the event that a subgroup analysis is performed, there is a possibility it may lead to different outcomes because the recommended treatments vary depending on the size and location of the renal stones. Some degree of publication bias was also a limitation of this study. However, Sutton et al. reviewed 48 articles from the Cochrane Database of Systematic Reviews and showed publication or related biases were common within the sample of meta-analyses assessed [80].
Another limitation is that the results reflected only the efficacy aspect of the stone-free rate and did not take into account the safety aspect of the treatments. Discriminating between merits and drawbacks of the treatment for a patient is clearly an important decision. Further studies that address these limitations are needed in the future.

Conclusions
PCNL for renal stones resulted in the highest success and stone-free rate and ranked the highest of the treatments analyzed. In contrast, SWL ranked the lowest of the treatments because of its lowest success and stone-free rates. The complexity of individual patients considered in this meta-analysis may have played a role in the results. Future analyses should include patient selection criteria such as renal stone location.
Supporting information S1 Table. PRISMA NMA checklist of Items to include when reporting a systematic review involving a network meta-analysis.