Evaluation of three stone-scoring systems for predicting SFR and complications after percutaneous nephrolithotomy: a systematic review and meta-analysis

Background Clinical studies assessing the feasibility and accuracy of three stone scoring systems’s (SSSs: Guy’s stone score, CROES nomogram and S.T.O.N.E nephrolithometry scoring system) have reported contradictory outcomes. This systematic evaluation was performed to obtain comprehensive evidence with regard to the feasibility and accuracy of three SSSs. Methods A systematic search of Embase, Pubmed, Medline, and the Cochrane Library was conducted to identify studies that compared three SSSs up to Mar 2018. Patients were categorized according to stone free (SF) and no-stone free (NSF), Outcomes of interest included perioperative variables, stone-free rate (SFR), and complications. Results Ten studies estimating three SSSs were included for meta-analysis. The results showed that SF patients had a significantly lower proportion of male (OR = 1.48, P = 0.0007), lower stone burden (WMD = -504.28, P < 0.0001), fewer No of involved calyces (OR = -1.23, P = 0.0007) and lower proportion of staghorn stone (OR = 0.33, P < 0.0001). Moreover, SF patients had significantly lower score of Guy score (WMD = -0.64, P < 0.0001), but, S.T.O.N.E. score (WMD = -1.23, P < 0.0001) and a higher score of CROES nomogram (WMD = 29.48, P = 0.003). However, the comparison of area under curves (AUC) of predicting SFR indicated that there was no remarkable difference between three SSSs. Nonetheless, Guy score was the only stone scoring system that predicted complications after PCNL (WMD = -0.29, 95% CI: − 0.57 to − 0.02, P = 0.03). Conclusions Our meta-analysis indicated that the three SSSs were equally, feasible and accurate for predicting SFR after PCNL. However, Guy score was the only stone scoring system that predicted complications after PCNL. Electronic supplementary material The online version of this article (10.1186/s12894-019-0488-y) contains supplementary material, which is available to authorized users.


Background
The recommended treatment option for renal calculi and staghorn calculi is percutaneous nephrolithotomy (PCNL) according to the guidelines of the European Association of Urology (EAU) [1]. PCNL has increasingly been used over the past few decades and may continue in the future [2,3]. However, PCNL outcomes among the authors are different, because of the vast heterogeneity in the methods for clinical and academic characterization of nephrolithiasis besides the evaluation of surgical outcomes. So assessing the preoperative factors that affect SFR and complications is critical.
The Guy's stone score, the Clinical Research Office of the Endourological Society(CROES) nomogram and the S.T.O.N.E.(stone size, tract length, obstruction, number of involved calices and essence) stone score are seen as predictors of stone-free status (SFS) and complications after PCNL [4][5][6]. The widespread use of a standardized stone scoring system is very precious for counseling patient, clinical decision, and assessment of outcomes, in addition to improving academic reporting [7]. However, no universally accepted stone scoring system for predicting SFR and complications after PCNL exists. Comparison of the SSSs in different clinical studies indicated some advantages as well as disadvantages of one nomogram to another for different variables. Hence, we performed a systematic review of the literature with a meta-analysis of the available published literature to compare the feasibility and accuracy of three SSSs in predicting PCNL outcomes concerning SFR and complications.

Study selection
According to the Cochrane Handbook recommendations, a systematic review of published literature was performed [8]. To identify all studies published up to Dec 31, 2018, which assessed the feasibility and accuracy of three SSSs. The following MESH search headings were used: "comparative studies", "Guy", "CROES", "S.T.O.N.E", "percutaneous nephrolithotomy", "stone free rate", and "complication".

Inclusion and exclusion criteria
All studies included in this meta-analysis satisfied the following requirements: (a) compare the two or three SSSs, (b) report the outcomes of two or three SSSs, (c) document the surgery as PCNL, (d) document indications for PCNL with renal stones. Studies were excluded if: (a) the article did not meet the inclusion criteria, (b) no outcomes were mentioned or the parameters were impossible to analyze the three SSSs from the published findings.

Data extraction and outcomes of interest
Two of the authors (JKH and SF) extracted data from the included studies including: author identification, country, publication years, study design, age, and the number of patients. All disagreements about eligibility were resolved by consensus through author discussions. The outcomes, including SFR and, overall complications, were extracted to compare between three SSSs. Overall complications were graded based on the Clavien-Dindo system [9].

Study quality assessment
In accordance with the criteria of Centre for Evidence-Based Medicine in Oxford, we evaluated the level of evidence (LOE) of the included ten studies. Furthermore, Jaded Score was applied to evaluate the methodological quality of RCTs [10]. while the Newcastle-Ottawa Scale (NOS) assessed the methodological quality of non-RCTs observational studies [11]. Besides, JKH and ZJG evaluated the quality of the articles and discrepancies were rechecked and resolved by discussion.

Statistical analysis
All analyses were conducted by Review Manager 5.3 (Cochrane Collaboration, Oxford, UK). Continuous and dichotomous variables were analyzed by weighted mean differences (WMDs) and odds ratios (ORs). All analysis results were reported with 95% CIs. I 2 and X 2 statistics were applied to evaluate the quantity of heterogeneity, and when I 2 >50%, the evidence was considered to have substantial heterogeneity, the random-effects (RE) model would be applied, otherwise, the fixed effects (FE) model was applied. Egger's test and funnel plot evaluated the publication bias. Sensitivity analyses estimated the influence of studies with a high risk of bias on the overall effect.

Quality of the studies and level of evidence(Table 1)
In this meta-analysis, the NOS quality assessment method of the observational studies [11], and the US Preventive Services Task Force grading system were applied to evaluate the quality of all studies [10]. Included studies were all level 3b. Also, the demographic variables of the three SSSs were extracted from included articles ( Table 1).

Outcomes of perioperative variables
Outcomes of three scoring systems for predicting complications after PCNL ( Table 5) Pooled data of reported that the three SSSs in predicting post-PCNL complications, the forest plot indicated that Guy score was the only stone scoring system for predicting post-PCNL complications (WMD = -0.29, 95% CI: − 0.57 to − 0.02, P = 0.03) ( Table 5, Fig. 5). No association between the two other scoring systems (CROES nomogram and S.T.O.N.E. score) and post-PCNL complications ( Table 5, Fig. 5).

Discussion
Widely applicable and straightforward tools will highly improve patient counseling, clinical decision making, assessment of operation outcomes and academic study after PCNL for renal stones [22,23]. These can allow reliable and accurate comparisons of treatment safety and efficacy, and facilitate the meaningful comparison of clinical studies [24]. We considered that the commonest and validated SSSs (Guy score, CROES nomogram, and S.T.O.N.E. score) could be predictive of SFR and complications after PCNL.
Both similar and divergent variables between the three SSSs has to be known. The Guy stone score consists of four grades based on stone burden and patient anatomy [7,25]. The CROES nomogram is highly generalizable based on global data and uniquely grades risk across a continuous scale rather than dividing stones of varying complexity into discrete groups [7,25]. The S.T.O.N.E. stone score stratifies patients into low-, moderate-, and high-risk groups, and it is more useful for decision making [7,25]. The three SSSs included different parameters, however, the stone location, stone count and staghorn calculi were pivotal variables in the three SSSs [7].
On the other hand, Guy stone score included renal anatomy but not stone burden, and this was a difference from CROES nomogram and S.T.O.N.E. score. The CROES nomogram included prior treatment as well as operation volume, and these variables had remarkable relationships to the SFR. However, CROES nomogram lacked imageology information on hydronephrosis and calyceal abnormalities. The S.T.O.N.E. score comprised stone size, tract length, obstruction, number of involved calices and essence, and it had greater feasibility and accuracy than any of the individual variables alone. In our meta-analysis, the results showed preoperative variables gender, stone burden, number of involved calyces and staghorn calculi were remarkably correlating with SFR after PCNL for kidney calculi.
However, there is still no widely accepted stone scoring system for the prediction of outcomes after PCNL, and contradictions between different authors exist concerning the prediction of outcomes by the SSSs. Some experts reported that three SSSs were efficacy and equally predictive of SFR by estimating and comparing the three SSSs in 246 patients after PCNL [17]. Tailly et al. reported that three SSSs have similar predictive accuracy of SFS by comparing the three SSSs in 586 patients after PCNL, but no association between three SSSs and complications [20]. Moreover, Bozkurt et al. reported that both the Guy and CROES  SFR for post-PCNL, and the three SSSs were not associated with complications [19]. In our meta-analysis, the results indicated that the three SSSs were remarkably associated with SFS and equally predictive of SFR, but only Guy score was a predictor of complications. However, some critical limitations exist in the three SSSs. Firstly, stone burden and density as the important parameters dide not reflect in Guy's score. Moreover, it is also failed to describe procedure difficulties as well as clinical variability. Secondly, the CROES nomogram is also did not reflect stone density and lacked important variables affecting the outcomes, including imaging information on hydronephrosis and pelvicalyceal abnormalities [5,26]. Moreover, the CROES nomogram was complex in the clinical applications [17]. Thirdly, the limitations of S.T.O.N.E. score were validated with a small cohort which may limit its widespread usage [4,7]. Similarly, several limitations existed while analyzing and interpreting results in our meta-analysis. Firstly, to identify prognostic factors, we acknowledge that other variables, such as surgeon experience and advanced surgical instruments, may need further investigation. These factors in the future may need to be incorporated in multicenter and larger samples clinical applications. Secondly, there existed heterogeneities of studies, some studies in this meta-analysis had the risk of selection bias. Lastly, all patients were evaluated for SFR after PCNL by KUB but not by CT, which may have overstated the SFR. Therefore, a need to develop more accurate and practical SSS to assess the relationship between the SSS and SFR, complications.In conjunction, our meta-analysis thus provides some up to date conclusions for the advantages and disadvantages of three SSSs in predicting of SFS and complications.

Conclusions
The Guy score, CROES nomogram and S.T.O.N.E. score were equally accurate predictive of SFR in patients undergoing PCNL, but the Guy score is the only SSS for predicting complications.