A nomogram for predicting the risk of residual stone fragments after ureteroscopy

Introduction

Urolithiasis is becoming a significant worldwide source of morbidity with prevalence rates varying from 1% to 20% (1). Rapid technological and surgical advancements have granted urologists with more mini-invasive treatment choices to remove stones in urological tracts, such as extracorporeal shock-wave lithotripsy (ESWL), ureteroscopy (URS), and retroperitoneal laparoscopic ureterolithotomy (RPLU) (2,3). URS is a convenient, efficient, and minimally invasive surgical procedure for the management of urolithiasis and is widely adopted in the removal of urinary tract stones (4,5). After these active monitoring procedures, unsatisfactory outcomes needing further management are still possible, with a possibility of 10.1–15.4%, especially postoperative residual stone fragments (6,7). If left untreated, patients with residual stone fragments may experience stone-related events such as stone growth, obstruction, or urinary tract infections and have to eventually undergo secondary surgical intervention (8-10).

To deal with postoperative stone burdens, predictive tools distinguishing high-risk patients who are likely to face residual stone fragments preoperatively are essential in practical clinical scenarios. Nomograms can provide excellent individualized disease-related risk estimations due to their high accuracy and satisfactory discriminating characteristics, and they are widely used in estimating cancer prognosis or other health outcomes (11,12). Imamura et al. established the first nomogram to predict the clinical postoperative outcomes of URS and compare the nomogram with ESWL nomograms (13). De Nunzio et al. further constructed a new nomogram based on data from a Mediterranean population (14).

Based on the previous studies, we aimed to establish a new nomogram to make a personalized prediction of the possibility of residual stone fragments after URS in Asian countries, especially in the Chinese population, and to attempt to incorporate new variables to improve the performance of the model in the aspects of discrimination and calibration. We present the following article in accordance with the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) reporting checklist (available at https://tau.amegroups.com/article/view/10.21037/tau-22-609/rc).

Methods

Study population

The data of 277 patients with urolithiasis treated with URS from September 2021 to May 2022 were collected. All patients had accepted normative and professional clinical care in Shanghai General Hospital, Shanghai Jiao Tong University (Shanghai, China). According to the radio of 7:3, these patients were randomly divided into a training group including 186 cases and an independent testing group including 91 cases. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was granted a waiver from approval by the Ethical Committee of the Shanghai General Hospital and informed consent was provided by all the patients. The diagnosis of ureteral stones was confirmed by non-contrast computed tomography (NCCT).

We collected the base information of all cases, including age, gender, body mass index (BMI), body temperature, hypertension, and diabetes mellitus. Patients underwent a series of detailed preoperative examinations including blood routine tests, complete urine tests, and urine cultures. The imaging parameters of stones were calculated from the preoperative NCCT by well-experienced uroradiologists. We measured the length of transverse and longitudinal diameters of the stones to calculate the stone size. Ureteral wall thickness (UWT) was defined as the maximum thickness of the ureteral wall and was obtained by measuring the thickness of the low-density soft tissue around the calculi. The location of stones was categorized by 0 = ureteropelvic junction, 1 = proximal ureter, 2 = middle ureter, and 3 = distal ureter. We also distinguished patients with single and multiple calculi. The severity of hydronephrosis was graded according to computed tomography (CT) image features. Grade 1 represented splitting of the renal pelvis only or not, grade 2 represented uniform dilation of the major and minor calyces, and grade 3 represented uniform dilation of the major and minor calyces accompanied with renal parenchymal thinning (15). Surgery-related variables were also recorded, such as the preoperative placement of double-J stents, the choice between rigid or flexible ureteroscope, and the operation time. Residual stone fragments were defined as remnant fragments bigger than 2 mm, which might need further intervention.

Procedures

All surgical procedures were monitored and implemented by the same professional urologist who had more than 15 years of ureteroscopic lithotripsy experience. Under general anesthesia, the URS was carried out in the lithotomy position. The rough distance between the stones and opening of ureters were evaluated according to the pre-operative examinations and body surface location. In the beginning, a safe guidewire was placed into the ureter to gain access to the stone through the bladder for stone localization. Then, a rigid ureteroscope was placed into the ureter at the lower end of the stones over the sensor guidewire, which was subsequently removed. We tended to utilize flexible ureteroscope and ureteral access sheaths to deal with upper ureteral calculi or renal calculi. Before the insertion of flexible ureteroscope, coaxial ureteric access sheaths were routinely placed into the ureter to gain access to the upper urinary tract. In the ureteroscopic field of view, the appearance of calculi, the situation of the ureteral wall, and the degree of urine opacity were observed and subsequently the lithotripsy was performed. The power was generated from a holmium laser generator (Lumenis company, San Jose, CA, USA) and transmitted through a 200/250 µm laser fiber. The power was usually set at 1.2–2.0 W and the frequency was set at 24–40 Hz, with flexible adjustment as necessary. Physiologic saline was used for irrigation to maintain a clear field of view during operation. Subsequently, the stone pieces were broken into fragments in the ureteral lumen for removal from the urinary system with a stone extractor. Procedures were completed with the retrograde dwelling of a double-J stent which would be removed through cystoscopy under local anesthesia on postoperative week 4. The record of the operation started at the beginning of general anesthesia and ended at successful placement of the catheters. All patients underwent NCCT on postoperative week 4 to examine for the presence of the residual stone fragments.

Statistical analysis

Out of all 277 cases, 186 were randomly identified as a training group to establish the nomogram, whereas the other 91 cases served as an independent dataset to validate the model. To explore the significant differences of variables between the patients with residual stone fragments and those without fragments, the Wilcoxon rank-sum test and Fisher’s exact test were adopted. To assess the significance of each variable, t-test was used for continuous variables and Fisher’s exact test for categorical variables.

We then conducted multivariate stepwise logistic regression analysis under the threshold of P<0.1 to select valuable predictive factors of stone-residual in the training group and also recorded the odds radio (OR) and 95% confidence interval (CI). Akaike’s information criterion (AIC) was calculated and compared between each factor. We hypothesized that the model which had the minimal AIC value with the fewest number of variables was the final model. Receiver operating characteristic (ROC) curves were plotted to assess the discrimination of the model. The Hosmer-Lemeshow (HL) test was utilized to evaluate the calibration of the model. We further drew calibration plots to show the predictive probability and decision curve analysis (DCA) was conducted to demonstrate the net benefit of the model.

Based on previous analysis, we successfully formulated a nomogram to predict the risk of residual stone fragments. All statistical analysis was conducted with Stata 15.0 software (Stata Corp., College Station, TX, USA). A P value <0.05 was considered statistically significant. The process of our statistical analysis is shown in Figure 1.

Figure 1 Characteristics of enrolled patients and analysis process of the study. AIC, Akaike’s information criterion; AUC, area under the curve; HL, Hosmer-Lemeshow.

Results

Among the 277 enrolled patients, including 200 (72.2%) males and 77 (27.8%) females, 87 (31.4%) cases had residual stone fragments after URS and the remaining 190 (68.6%) patients were confirmed as stone-free. Among them, diabetes occurred in 20 (7.2%) patients and hypertension occurred in 38 (13.7%) patients. We used median and interquartile range [IQR] to describe the distribution trends of quantitative variables, such as age 49.1 [38, 61] years, white blood cell (WBC) 6.91 [5.53, 9.07] ×10⁹/L, and transverse diameter of stones 7.07 [5.59, 9.33] mm. The characteristics of all variables are shown in Table 1. There were no statistically significant differences between the training and the testing group.

The numbers of stone-free cases and stone-residual cases in the training group were 128 (68.8%) and 58 (31.2%), respectively, and in the testing group were 62 (68.1%) and 29 (31.9%), respectively. According to the univariate Cox regression analysis, 11 variables with P≤0.05 were recognized as being associated with residual stone rate (RSR). The results are shown in Table 2. The logistic model which had the minimal AIC value of 213.734 was identified as the final model (Table 3). It included 5 variables: WBC (P=0.061), hypertension (P=0.068), transverse diameter of stones (P=0.048), stone location (P=0.074), and hydronephrosis (P=0.021).

Based on the final model, we constructed a nomogram to predict the RSR. ROC analysis demonstrated that it had a great predictive accuracy of 0.7203 (95% CI: 0.6437–0.8051) (Figure 2A) and HL test revealed that it had strong calibration ability (P=0.34) (Figure 2B). In addition, the nomogram presented a net benefit in a range of probability between 20% and 80% through the DCA (Figure 2C). We further assessed the nomogram with an independent external dataset of which area under the curve (AUC) was 0.7280 (95% CI: 0.6054–0.8507) (Figure 3A) and the P value of the HL test was 0.65 (Figure 3B). It also presented a benign net benefit of 20% to 70% (Figure 3C). The nomogram is presented in Figure 4.

Figure 2 Performances of the predictive model on the training group. (A) ROC curves of the nomogram in the training group. The nomogram had good discriminative power with an area under the ROC curve of 0.7203. (B) Calibration plots with local regression non-parametric smoothing lines of the present nomogram in the training group. (C) DCA demonstrated the model presented a net benefit in a range of probability between 20% to 80% in the training group. ROC, receiver operating characteristic; AUC, area under the ROC curve; DCA, decision curve analysis; CI, confidence interval; Pr, probability.

Figure 3 Performances of the predictive model on the testing group. (A) ROC curves of the nomogram in the testing group. The nomogram had good discriminative power with an area under the ROC curve of 0.7280. (B) Calibration plots with local regression non-parametric smoothing lines of the present nomogram in the testing group. (C) DCA demonstrated the model presented a net benefit in a range of probability between 20% and 70% in the testing group. ROC, receiver operating characteristic; DCA, decision curve analysis; CI, confidence interval; Pr, probability.

Figure 4 Nomogram predicting RSR after ureteroscopic lithotripsy. UPJ, ureteropelvic junction; RSR, residual stone rate.

Discussion

Our primary aim in the study was to establish a personal predictive nomogram for RSR and evaluate the performance of the nomogram from aspects of discrimination and calibration with an independent dataset. Briefly, our nomogram included 5 variables: WBC, hypertension, transverse diameter of stone, stone location, and hydronephrosis.

Estimating stone burden is a critical component of the management of renal stone disease, which can consistently and significantly affect decision-making. In general, the size of stones has a negative correlation with stone clearance effectiveness (16). In this study, we evaluated stone burden through measuring the length of diameters of stones and found the transverse diameter of calculi had a significant relationship with RSR (P=0.048), which agreed with previous studies (17,18). More severe stone burden frequently incurred longer operation time, more complex operation procedures, and eventually lead to higher RSR.

Previous studies have reported that hypertension is associated with the incidence of urolithiasis. Vodanović et al. confirmed that arterial hypertension was an independent predictor of current urolithiasis (P<0.05) (19). Several cohort studies have revealed possible bidirectional associations between hypertension and nephrolithiasis (20-22). Stone disease influenced the level of blood pressure through multiple pathophysiological mechanisms like inflammation and oxidative stress. Thus, we attempted to unearth any potential connection between the risk of residual stone fragments and hypertension. Our analysis implied that patients with hypertension were more likely to have residual stone fragments after URS (P=0.068). Though no clear evidence of less satisfactory treatment results of patients with hypertension was found, enrolling hypertension in the final nomogram would improve clinical management and primary prevention of urolithiasis patients with hypertension.

Compared with proximal stones, the stone free rate was higher in distal stones (23). In our study, 78.9% of distal stones were removed clearly whereas only 66.5% of proximal stones had no residual fragments. Although the rate was somewhat lower than the treatment outcomes published in the European Association of Urology (EAU) Guidelines, which could be explained by the limitations of treatment options, we still certified that the location of stones obviously influenced the RSR, with statistical differences (P=0.006). Although some previous studies have remained suspicious of the relationship between the degree of hydronephrosis and RSR (24,25), hydronephrosis was identified as a significant factor for predicting RSR in a range of other surgical outcome predictive models (26,27). Overall analysis in our study indicated a significant difference in the role of hydronephrosis for RSR and retreatment rate (P=0.002). For surgical outcomes of ESWL, Abe et al. considered that pyuria served as an independent predictive factor (28). In our univariate analysis and logistic regression analysis, infectious-related factors such as WBC also showed great importance. No certain association between positive infectious-related factors and poor surgical outcomes has been found yet. Our findings can be explained by that the turbid urine related to serious infection can interfere surgical field and affect the stone-free rate. Besides, it is worth noting that severe hydronephrosis was often accompanied by impacted stones and increased the risk of urinary tract infection (29,30).

In clinical scenarios, rigid and flexible URS were both feasible options for the treatment of ureteral stones. Galal et al. concluded that flexible URS contributed to longer operative time and higher stone-free rate (31). Similarly, Yoshida et al. considered that high UWT was associated with poor endoscopic findings and adverse surgical outcomes in patients undergoing URS (32). However, our cohort study found no statistically significant difference between these factors and postoperative RSR, which may be explained by the discrepancy in the data source.

When evaluating the predictive accuracy of a model, discrimination, calibration, clinical benefit, and internal and external validity were significant aspects (33). Large numbers of nomograms have been developed to serve as predictive models in urology, especially in the oncological and cancer fields (34,35). Certainly, models to predict RSR after URS are also important (36). Imamura et al. constructed the first nomogram including the length, number, location of the stone, and the presence of pyuria based on a Japanese population (13), but the model showed plain discrimination abilities and poor calibration and lacked clinical net benefits with subsequent validation (37). De Nunzio et al. established another predictive model which included the following factors: hydronephrosis, length of stones, stone location, and number of stones (14); the model still needs further validation before actual clinical application. With the independent validation dataset, we examined the performance of our nomogram by ROC curves, HL test, and DCA. The results illustrated that our nomogram had great discrimination, favorable calibration, benign clinical benefit, and a convenient model presentation. It has been widely accepted that the predictive accuracy of a nomogram is inevitably affected by the general characteristics of the population (33). Thus, we aimed to construct a nomogram based on the Chinese population to achieve better application abilities. Our nomogram could distinguish high-risk patients who would have residual stone fragments and require secondary surgical intervention after URS. The nomogram could also offer auxiliary guidance for urologists with excellent clinical application value preoperatively.

Our study had some limitations. Firstly, the simple presentation of the nomogram sometimes limited its application; more flexible app- or web-based tools are required for further dissemination. Secondly, the single-center data source somewhat influenced the reliability of the model, more validation data from other institutions is necessary to enhance the predictive value in our further study. Nonetheless, our study established a nomogram to identify high-risk patients of stone residual fragments who still require secondary intervention.

Conclusions

We successfully constructed a nomogram including 5 factors to evaluate the risk of RSR after URS with superior discrimination, excellent calibration abilities, and great clinical benefit. With further improvement and validation, the nomogram has high potential application to pre-emptively identify high-risk patients who may experience postoperative stone burden and improve stone management in the future.

Acknowledgments

Funding: This work was supported by The National Natural Science Foundation of China (No. 81771564) and Zunyi Municipal Science and Technology Bureau (No. 2018[192]).

Footnote

Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://tau.amegroups.com/article/view/10.21037/tau-22-609/rc

Data Sharing Statement: Available at https://tau.amegroups.com/article/view/10.21037/tau-22-609/dss

Peer Review File: Available at https://tau.amegroups.com/article/view/10.21037/tau-22-609/prf

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tau.amegroups.com/article/view/10.21037/tau-22-609/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was granted a waiver from approval by the Ethical Committee of the Shanghai General Hospital and informed consent was provided by all participants.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Türk C, Petřík A, Sarica K, et al. EAU Guidelines on Diagnosis and Conservative Management of Urolithiasis. Eur Urol 2016;69:468-74. [Crossref] [PubMed]
2. Desai M, Sun Y, Buchholz N, et al. Treatment selection for urolithiasis: percutaneous nephrolithomy, ureteroscopy, shock wave lithotripsy, and active monitoring. World J Urol 2017;35:1395-9. [Crossref] [PubMed]
3. Torricelli FC, Marchini GS, Pedro RN, et al. Ureteroscopy for management of stone disease: an up to date on surgical technique and disposable devices. Minerva Urol Nefrol 2016;68:516-26. [PubMed]
4. Li J, Xiao J, Han T, et al. Flexible ureteroscopic lithotripsy for the treatment of upper urinary tract calculi in infants. Exp Biol Med (Maywood) 2017;242:153-9. [Crossref] [PubMed]
5. Secker A, Rassweiler J, Neisius A. Future perspectives of flexible ureteroscopy. Curr Opin Urol 2019;29:113-7. [Crossref] [PubMed]
6. Fiori C. Life-threatening complications: "the dark side" of ureteroscopy. Minerva Urol Nefrol 2017;69:521. [PubMed]
7. Aboumarzouk OM, Monga M, Kata SG, et al. Flexible ureteroscopy and laser lithotripsy for stones >2 cm: a systematic review and meta-analysis. J Endourol 2012;26:1257-63. [Crossref] [PubMed]
8. Rippel CA, Nikkel L, Lin YK, et al. Residual fragments following ureteroscopic lithotripsy: incidence and predictors on postoperative computerized tomography. J Urol 2012;188:2246-51. [Crossref] [PubMed]
9. Ito H, Kuroda S, Kawahara T, et al. Preoperative factors predicting spontaneous clearance of residual stone fragments after flexible ureteroscopy. Int J Urol 2015;22:372-7. [Crossref] [PubMed]
10. Jing S, Gai Q, Zhao X, et al. Physical therapy in the management of stone fragments: progress, status, and needs. Urolithiasis 2018;46:223-9. [Crossref] [PubMed]
11. Iasonos A, Schrag D, Raj GV, et al. How to build and interpret a nomogram for cancer prognosis. J Clin Oncol 2008;26:1364-70. [Crossref] [PubMed]
12. Balachandran VP, Gonen M, Smith JJ, et al. Nomograms in oncology: more than meets the eye. Lancet Oncol 2015;16:e173-80. [Crossref] [PubMed]
13. Imamura Y, Kawamura K, Sazuka T, et al. Development of a nomogram for predicting the stone-free rate after transurethral ureterolithotripsy using semi-rigid ureteroscope. Int J Urol 2013;20:616-21. [Crossref] [PubMed]
14. De Nunzio C, Ghahhari J, Lombardo R, et al. Development of a nomogram predicting the probability of stone free rate in patients with ureteral stones eligible for semi-rigid primary laser uretero-litothripsy. World J Urol 2021;39:4267-74. [Crossref] [PubMed]
15. Liu DB, Armstrong WR 3rd, Maizels M. Hydronephrosis: prenatal and postnatal evaluation and management. Clin Perinatol 2014;41:661-78. [Crossref] [PubMed]
16. Barreto L, Jung JH, Abdelrahim A, et al. Medical and surgical interventions for the treatment of urinary stones in children. Cochrane Database Syst Rev 2019;10:CD010784. [Crossref] [PubMed]
17. Zhang Y, Li J, Zhang D, et al. Nomograms predicting the outcomes of endoscopic treatments for pediatric upper urinary tract calculi. Int J Urol 2021;28:295-301. [Crossref] [PubMed]
18. Xiao Y, Li D, Chen L, et al. The R.I.R.S. scoring system: An innovative scoring system for predicting stone-free rate following retrograde intrarenal surgery. BMC Urol 2017;17:105. [Crossref] [PubMed]
19. Vodanović M, Lucijanić M, Zupančić Šalek S, et al. Prevalence of and risk factors for urolithiasis in Croatian patients with hemophilia. Int J Hematol 2021;113:656-61. [Crossref] [PubMed]
20. Lin BB, Huang RH, Lin BL, et al. Associations between nephrolithiasis and diabetes mellitus, hypertension and gallstones: A meta-analysis of cohort studies. Nephrology (Carlton) 2020;25:691-9. [Crossref] [PubMed]
21. Shang W, Li Y, Ren Y, et al. Nephrolithiasis and risk of hypertension: a meta-analysis of observational studies. BMC Nephrol 2017;18:344. [Crossref] [PubMed]
22. Rezaee ME, Ward CE, Pollock M, et al. Association between multiple chronic conditions and urolithiasis. Int Urol Nephrol 2017;49:1361-7. [Crossref] [PubMed]
23. Ur Rehman MF, Adnan M, Hassan A 3rd, et al. Comparison of Ureteroscopic Pneumatic Lithotripsy and Extracorporeal Shock Wave Lithotripsy for Proximal Ureteral Calculi. Cureus 2020;12:e7840. [Crossref] [PubMed]
24. He Z, Yin S, Duan X, et al. Does the presence or degree of hydronephrosis affect the stone disintegration efficacy of extracorporeal shock wave lithotripsy? A systematic review and meta-analysis. Urolithiasis 2020;48:517-26. [Crossref] [PubMed]
25. Kaya C, Kaynak Y, Karabag A, et al. The Predictive Role of Abdominal Fat Parameters and Stone Density on SWL Outcomes. Curr Med Imaging Rev 2020;16:80-7. [Crossref] [PubMed]
26. Zhu Z, Wang S, Xi Q, et al. Logistic regression model for predicting stone-free rate after minimally invasive percutaneous nephrolithotomy. Urology 2011;78:32-6. [Crossref] [PubMed]
27. Kadihasanoglu M, Erkan E, Yucetas U, et al. Does preoperative hydronephrosis affect the stone-free rate of micro-percutaneous nephrolithotomy? Arch Esp Urol 2019;72:406-14. [PubMed]
28. Abe T, Akakura K, Kawaguchi M, et al. Outcomes of shockwave lithotripsy for upper urinary-tract stones: a large-scale study at a single institution. J Endourol 2005;19:768-73. [Crossref] [PubMed]
29. Legemate JD, Wijnstok NJ, Matsuda T, et al. Characteristics and outcomes of ureteroscopic treatment in 2650 patients with impacted ureteral stones. World J Urol 2017;35:1497-506. [Crossref] [PubMed]
30. Özbir S, Can O, Atalay HA, et al. Formula for predicting the impaction of ureteral stones. Urolithiasis 2020;48:353-60. [Crossref] [PubMed]
31. Galal EM, Anwar AZ, El-Bab TK, et al. Retrospective comparative study of rigid and flexible ureteroscopy for treatment of proximal ureteral stones. Int Braz J Urol 2016;42:967-72. [Crossref] [PubMed]
32. Yoshida T, Inoue T, Omura N, et al. Ureteral Wall Thickness as a Preoperative Indicator of Impacted Stones in Patients With Ureteral Stones Undergoing Ureteroscopic Lithotripsy. Urology 2017;106:45-9. [Crossref] [PubMed]
33. Kleinrouweler CE, Cheong-See FM, Collins GS, et al. Prognostic models in obstetrics: available, but far from applicable. Am J Obstet Gynecol 2016;214:79-90.e36. [Crossref] [PubMed]
34. Rocco B, Sighinolfi MC, Sandri M, et al. A novel nomogram for predicting ECE of prostate cancer. BJU Int 2018;122:916-8. [Crossref] [PubMed]
35. Wang Z, Zhou Y, Guan C, et al. The impact of previous cancer on overall survival of bladder cancer patients and the establishment of nomogram for overall survival prediction. Medicine (Baltimore) 2020;99:e22191. [Crossref] [PubMed]
36. Anan G, Kudo D, Matsuoka T. What are the predictors of residual stone after ureteroscopy for urolithiasis? Transl Androl Urol 2022;11:1071-3. [Crossref] [PubMed]
37. De Nunzio C, Bellangino M, Voglino OA, et al. External validation of Imamura nomogram as a tool to predict preoperatively laser semi-rigid ureterolithotripsy outcomes. Minerva Urol Nefrol 2019;71:531-6. [Crossref] [PubMed]