Nomogram and scoring system for predicting stone-free status after extracorporeal shock wave lithotripsy in children with urolithiasis

Authors


Correspondence: Hiep T. Nguyen, Department of Urology, Hunnewell-353, Children's Hospital Boston, 300 Longwood Avenue, Boston, MA 02115, USA.

e-mail: hiep.nguyen@childrens.harvard.edu

Abstract

What's known on the subject? and What does the study add?

  • Extracorporeal shock wave lithotripsy is often considered to be the first-line treatment method for the majority of urinary tract stone disease in children. The stone clearance rate in children treated with ESWL is higher than that in adults. Recently, nomograms for several diseases, e.g. for specific cancers, have been developed and validated in large patient populations. They have become very popular predictive tools that provide the most objective, evidence-based, and individualized risk estimation. These nomograms have gained acceptance as useful guides in clinical practice for use by physicians and patients. In adults, a nomogram has been created to predict stone-free outcome after ESWL; however, to our knowledge none has been developed for children with urolithiasis.
  • This is the first study-generated nomogram table and scoring system for predicting the stone-free rate after ESWL in children. This predictive tool could be useful for clinicians in counselling the parents of children with urolithiasis and in recommending treatment.

Objective

  • To determine the stone-free rate after extracorporeal shock wave lithotripsy (ESWL) and its associated factors to formulate a nomogram table and scoring system to predict the probability of stone-free status in children.

Patients and Methods

  • A total of 412 children (427 renal units [RUs]) with urolithiasis were treated with ESWL using a lithotriptor between 1992 and 2008.
  • Cox proportional hazards regression was used to model the number of treatment sessions to stone-free status as a function of statistically significant demographic characteristics, stones and treatment variables.
  • A bootstrap method was used to evaluate the model's performance. Based on the multivariate model, the probabilities of being stone-free after each treatment session (1, 2 and >3) were then determined.
  • A scoring system was created from the final multivariate proportional hazard model to evaluate each patient and predict their stone-free probabilities.

Results

  • Complete data were available for 395 RUs in 381 patients.
  • Of the 395 RUs, 303 (76.7%) were considered to be stone-free after ESWL.
  • Multivariate analysis showed that previous history of ipsilateral stone treatment is related to stone-free status (hazard ratio [HR]: 1.49; P = 0.03).
  • Stone location was a significant variable for stone-free status, but only in girls.
  • Age (HR 1.65, P = 0.02) and stone burden (HR 4.45, P = 0.002) were significant factors in the multivariate model.

Conclusion

  • We believe that the scoring system, and nomogram table generated, will be useful for clinicians in counselling the parents of children with urolithiasis and in recommending treatment.
Abbreviations
RU

renal unit

KUB

plain abdominal film of kidney, ureter and bladder

US

ultrasonography

HR

hazard ratio

Introduction

Extracorporeal shock wave lithotripsy is often considered to be the first-line treatment method for the majority of urinary tract stone disease in children [1]. The stone clearance rate in children treated with ESWL is higher than that in adults [2]. While ESWL is less invasive compared with other methods, e.g. ureterorenoscopy and percutaneous nephrolithtotomy, it has limitations in treatment efficacy. Treatment outcome after ESWL depends on several factors, including the type of lithotriptor, stone characteristics (i.e. number, size, composition and location), renal anatomy and function.

Recently, nomograms for several diseases, e.g. for specific cancers, have been developed and validated in large patient populations. They have become very popular predictive tools that provide the most objective, evidence-based and individualized risk estimation. These nomograms have gained acceptance as useful guides in clinical practice for use by physicians and patients. A nomogram has been created to predict stone-free outcome after ESWL in adults [3], but, to our knowledge, none has been developed for children with urolithiasis. The aim of the present study was to determine the stone-free rate after ESWL and its associated factors to formulate a nomogram table and scoring system to predict the probability of stone-free status after treatment in children with urolithiasis.

Patients and Methods

Patient Database and Evaluation

After institutional approval, we retrospectively evaluated all children ≤17 years old, (412 patients, 427 renal units [RUs]) with stones, consecutively treated between March 1992 and February 2008 at our institution. Patients with abnormal renal anatomy (i.e. horseshoe kidney, pelvic kidney, rotation anomaly), non-opaque stones and/or previous history of recurrent cystine stones were excluded from the study, leaving 381 patients (395 RUs) with complete data for evaluation.

Before surgery, all patients were evaluated by plain abdominal film of kidney, ureter and bladder (KUB), IVU, urine analysis, urine culture, serum biochemistry and coagulation tests. Antibiotics were administered when urine culture was positive. The procedure was performed under i.v. sedation (midazolam 0.1 mg/kg and alfentanyl 10 μg) in 354 (90%) of the patients, while 41 (10%) had general anaesthesia. Treatment was performed with a Siemens Lithostar lithotriptor (Siemens Medizinische Technik, Erlangen, Germany). All patients were treated in the supine position, and shielding of the lung fields or gonads was not used. ESWL was performed by one urologist (N.T.) who decided on the level of energy and number of shock waves for each case. In general, treatment was initiated at 13 kV, which was gradually increased by 0.3 kV to the maximum level deemed appropriate for the patient; it was terminated when complete fragmentation of the stone(s) was noted on fluoroscopy. If there was persistent or incomplete fragmentation of the stone(s) noted on fluroscopy after the maximum number of shocks had been delivered, a repeat ESWL session was performed at least 2 weeks after the previous one. All patients were evaluated by IVU and ultrasonography (US) 12 weeks after the last session. They were designated as stone-free or as having residual stones (any evidence of persistent stone fragments irrespective of size). No adjunctive measures, e.g. mechanical inversion, diuretic therapy or percussion were used to facilitate fragment passage. Treatment with ESWL was regarded as a failure if no fragmentation was noted after the third session.

Statistical Analysis

The number of treatment sessions to reach stone-free status was analysed as our major outcome. Stone-free status was considered as an event, while patients who did not reach stone-free status but no more sessions followed were considered as censored at the last session they had. We analysed the data using survival models. Univariate and multivariate Cox proportional hazards models were used to analyse the demographic, stone and treatment variables. Stone burden was calculated by measuring the stone area on the plain film in cm2. In the presence of more than one stone, burden was calculated by adding the areas of each stone. The patients were stratified according to stone burden: group 1 (G1): stones ≤1 cm2; group 2 (G2): 1.1–2.0 cm2; and group 3 (G3): >2.0 cm2. Stone location was categorized as follows: renal pelvis; upper/middle calyx; lower calyx; upper ureter; middle/lower ureter; or multiple locations. The resulting multivariate model was used to construct a nomogram table predicting the cumulative probability of stone-free status according to the number of treatment sessions. We also conducted a bootstrapping procedure to internally validate our model. The function ‘validate’ in the ‘Design’ library of The R Foundation for Statistical Computing was used to calculate the bootstrap overfitting-corrected c-index to measure the predictive ability of our models. c = 0.5 means a random prediction, and c = 1 means perfect prediction [4]. Our bootstrapping algorithm was based on 200 replicates.

In addition, a scoring system was created from the final multivariate proportional hazard model to evaluate each patient and predict their stone-free probabilities. The regression coefficients from the model were multiplied by a positive number, and then the results were rounded to integers as the point for each variable. The score system was then categorized for easy use into three levels: low, medium and high, and put into a Cox proportional hazard model. Predicted probabilities of stone-free were calculated based on the model for each low, medium and high group. Observed probabilities of stone-free based on Kaplan–Meier methods were also calculated for each group based on 200 bootstrappings, and then compared with the predicted ones for internal validation of the scoring system. Statistical analyses were performed using SAS (version 9.2) and The R Foundation for Statistical Computing.

Results

The median (range) patient age was 8 (1–17) years. Patient characteristics, stone location and burden, and history of previous treatment are shown in Table 1. The median stone burden was 0.8 cm2. Ancillary procedures (JJ stent placement and percutaneous nephrostomy tube) were performed in 38 RUs (9.6%). The median number of shock waves and energy level used were 1600 and 17.2 kV, respectively.

Table 1. Characteristics of patients, stones and treatments
Patient characteristics 
No. of patients/RUs381/395
Median (range) age8.0 (1–17)
Girls, n (%)178 (47)
Boys, n (%)217 (53)
No. of solitary kidneys9
Previous history of ipsilateral stone treatment, n (%)69 (18)
Open surgery33
Percutaneous nephrolithotomy15
ESWL11
Ureterorenoscopy2
Combination8
Stone characteristics 
Left kidney, n (%)164 (42)
Right kidney, n (%)231 (58)
Stone location, n 
Pelvis/upper ureter178
Calix (upper/middle/lower)111
Mid or lower ureter59
Multiple47
No. of stones 
Single, n (%)348 (88)
Multiple, n (%)47 (12)
Stone burden median (range), cm20.8 (0.2–16)
≤1.0 cm2257 (65)
1.1–2.0 cm297 (25)
>2.0 cm241 (10)
Treatment characteristics 
Median (range) no. of shock waves1600 (180–3500)
Median (range) generator energy, kV17.2 (14.8–18.4)
No. of ESWL session(s) 
Median (range)1 (1–12)
1 session, n (%)243 (62)
2 sessions, n (%)86 (22)
3 sessions, n (%)41 (10)
≥4 sessions, n (%)25 (6)
Ancillary procedures, n (%) 
JJ stent29 (7)
Percutaneous nephrostomy tube9 (2)
Anaesthesia, n (%) 
General41 (10)
i.v. analgesia354 (90)

Of the 395 RUs, 303 (76.7%) were stone-free (median number of sessions = 1). Of the patients rendered free of stones, 207 (68%) were treated in a single ESWL session, 62 (21%) after two sessions and 34 (11%) underwent three or more sessions (Fig. 1). Univariate hazard ratios for each variable are listed in Table 2, while the multivariate Cox proportional hazards model results are provided in Table 3. On multivariate analysis, a previous history of ipsilateral stone treatment was associated with stone-free status (hazard ratio [HR] = 1.49; P = 0.03). In addition, stone location was a significant variable for stone-free status, but only in girls. Girls with stone(s) in the renal pelvis and upper ureter required the least number of treatments to become stone-free, while those with stones located in multiple locations required the most (HR = 3.12, P < 0.001). Age was also a significant variable associated with stone-free status with a HR 1.65 (<5 years old vs. >10 years old, P = 0.002) and 1.28 (5–10 years old vs >10 years old, P = 0.002). Stone burden was also associated with stone-free status after ESWL (HR = 4.45 for 0–1 cm2 vs. >2 cm2 and 2.11 for 1–2 cm2 vs. >2 cm2). The resulting multivariate model was used to construct a nomogram table predicting the cumulative probability of stone-free status according to the number of treatment sessions (Figs 2 and 3).

Figure 1.

Stone-free rate according to the number of ESWL sessions.

Figure 2.

Nomogram table for boys predicting the cumulative probability of stone-free status according to number of treatment sessions.

Figure 3.

Nomogram table for girls predicting the cumulative probability of stone-free status according to number of treatment sessions.

Table 2. Univariate table for stone-free status
VariableHR (95% CI)P
Gender Female vs. Male1.25 (0.99–1.56)0.06
History of stone treatment (No vs. Yes)1.67 (1.20–2.32)0.002
Age <0.001
≤5 years2.19 (1.64–2.94)
>5 and ≤10 years1.44 (1.07–1.93)
>10 yearsReference
Stone burden, cm2 <0.001
0–13.95 (2.55–6.11)
1–21.75 (1.09–2.79)
>2Reference
Stone location <0.001
Pelvis or Upper Ureter2.87 (1.90–4.33)
Calix2.24 (1.45–3.46)
Mid or Lower Ureter2.75 (1.72–4.39)
MultipleReference
Table 3. Multivariate analysis for stone-free status
VariableHR (95% CI)P
History of stone treatment (No vs. Yes)1.49 (1.05–2.13)0.03
Age 0.002
≤5 years1.65 (1.20–2.26)
>5 and ≤10 years1.28 (1.50–1.10)
>10 yearsReference
Stone burden, cm2 <0.001
0–14.45 (2.77–7.15)
1–22.11 (1.66–2.67)
>2Reference
Interaction of gender and stone location 0.003
Girls 0.0002
Stone location 
Pelvis or upper ureter3.12 (1.61–6.05)
Calix2.58 (1.25–5.34)
Mid or lower ureter1.39 (0.67–2.88)
MultipleReference
Boys
Stone locationNA0.86

A bootstrap method was used to evaluate the model performance (Table 4). A scoring system ranging from 0 to 12 was created from the final multivariate proportional hazard model to evaluate each patient and predict their stone-free probabilities (Table 5). Finally, Table 6 shows that probabilities of being stone-free after ESWL after each treatment session (1, 2 and >3) can be predicted based on the score system. A lower score was correlated with significantly higher stone-free rates (i.e. it was associated with low risk of treatment failure). A calibration plot shows that fraction probability and predicted probability of stone-free status by score group (Fig. 4).

Figure 4.

Calibration plot shows that fraction probability (solid curve) and predicted probability (dashed curve).

Table 4. Validation of the final model using the bootstrap method
 Mean predicted stone-free probability, %Observed K-M stone-free probability, %Bootstrap-corrected C statistics
Session 152.652.30.78
Session 273.372.00.85
Session 384.082.00.86
Table 5. Scoring system
AgePoints
≤5 years0
>5 and ≤10 years1
>10 years2
Stone burden 
0–10
1–22
>24
History of stone treatment 
No0
Yes1
Gender 
Male2
Female (according to stone location) 
Pelvis or Upper Ureter0
Calix1
Mid or Lower Ureter2
Multiple3
Table 6. Score and stone-free probability
ScorePredicted stone-free probability, % (95% CI)Observed stone-free probability,% (mean, 2.5% quantile and 97.5% quantile from 200 bootstrap samples)
Session 1Session 2Session 3Session 1Session 2Session 3
  1. aOnly one stone-free patient here and three censored, from a total of 44 in this group.
Low (0–2)76.5 (72.6, 79.1)94.1 (91.5, 96.0)98.7 (97.3, 99.4)75.7 (69.8, 83.1)95.7 (91.9, 98.9)97.7 (93.4, 100)
Medium (3–6)44.7 (40.3, 48.8)68.6 (63.7, 72.9)83.2 (78.2, 87.0)48.0 (41.9, 54.0)67.8 (60.1, 74.0)81.4 (73.5, 87.5)
High (>6)17.0 (12.4, 21.3)30.5 (22.6, 37.6)42.8 (32.1, 51.9)2.5a (0.0%, 9.1)24.5 (12.3, 38.0)45.2 (29.3, 60.1)

Complications occurred in 27 patients. Only one patient required hospitalization owing to high fever. Steinstrasse developed in 26 patients; the stone(s) was originally located in the renal pelvis in 18 patients, the upper calyx in two patients, the lower calyx in two patients and multiple locations in four patients. When analysing according to the stone burden, steinstrasse occurred in four patients (15%) of G1 (stones ≤ 1 cm2), in 14 (54%) of G2 (1.1–2.0 cm2), and in eight (31%) of G3 (>2.0 cm2). All patients were successfully treated with repeat ESWL monotherapy without using ancillary procedures such as JJ stenting.

Discussion

Extracorporeal shock wave lithotripsy is considered to be a first-line treatment method in the management of paediatric urolithiasis [1, 5, 6]. It has been reported to be safe and effective even in treating stone diseases in low birth weight infants and for staghorn calculi in younger children [7, 8]. In spite of the popular use of ESWL, it has yet to be well defined which children with stone disease would benefit from ESWL treatment and the number of sessions required to achieve stone-free status. The ability to predict the effectiveness of ESWL is essential in determining the most appropriate treatment for patients. In urological oncology, nomogram tables for several diseases, e.g. prostate and bladder cancer, have been developed and validated in large patient populations. These nomograms have gained acceptance as useful guides in clinical practice for both physicians and patients. A nomogram for predicting stone-free outcome after ESWL in adults has been developed [3], but no such nomogram exists for ESWL in children. We believe that the results of the present study provide a useful tool for the physician to chose the appropriate treatment based on objective evidence rather than an individual subjective assessment.

The present analysis led to several interesting findings. We observed that a previous history of ipsilateral stone treatment negatively impacts on stone-free status, consistent with that observed in a study by Nelson et al. [9]. Although the rationale for this finding is unclear, it is speculated that previous stone treatment can cause scarring and prevent good propulsive peristalsis and adequate contractions, resulting in subtle delays in urinary drainage which then can impede on the subsequent passage of the stone fragments after ESWL treatment [10].

In addition, we observed that age at presentation was a significant variable in predicting stone-free status after ESWL. Age could be classified into three groups (<5, 5–10 and >10 years), with the highest probability of stone-free status for the younger age group and the lowest probability for the older age group. In the literature, some previous studies had similarly findings [11], while others found no significant impact of age on stone clearance [12, 13]. However, in general, it is believed that stone clearance rate after ESWL is higher in the children than in adults [2] and that younger children passed renal stones at a higher rate than older ones [14]. A potential explanation for this observation is that younger children have greater compliance in their urinary tracts and a shorter distance to pass the stone(s) [2].

We also observed that, in children, the probability of being stone-free was dependent on the stone burden. Our overall stone-free rate was 76.7% but it was 82.4% for stones ≤10 mm, 69.0% for stones 11–20 mm, and 58.5% for stones >20 mm. Muslumanoglu et al. [15] observed similar stone-free rates of 87.8%, 75.5% and 56.7%. By contrast, Ather and Noor [5] reported higher rates of 97%, 88%, and 90%. This discrepancy is attributable to the fact that in the study by Ather and Noor: (i) the stone burden reported was based on the maximum single diameter, rather than on measurement of two dimensions (cm2); (ii) patients with residual stone fragments of ≤3 mm in diameter were still considered to be stone-free; and (iii) the lack of consistency in type of postoperative imaging (KUB, US or both). The stone-free rate based on a KUB could be 10–15% greater than when nephrotomography was used [16]. In the present study, all patients were evaluated with both IVU and US 12 weeks after the last session to accurately establish stone-free status.

To our knowledge, no study has previously shown a significant interaction between gender, location and stone-free status in the children. We observed that stone location is a significant determinate for stone-free status only in girls. The probability of being stone-free after ESWL in a boy was independent of location, and was the same as for a girl with mid/lower ureteric stones. By comparison, for girls, the pelvis and upper ureter required the least number of sessions to achieve stone free status, while stones located in more than one location required the most. Several studies in adults have shown that the location (especially lower pole) and the number of stones are significant determinates of stone-free rates 3 months after ESWL [3, 17]. In children, some studies had similar findings [5], but others failed to find a statistically significant relationship between stone-free rates and location when comparing lower-pole stones with other intrarenal stones or intrarenal to ureteric stones [12, 13, 18]. The aetiology of the gender difference in impact of location on stone-free rates after ESWL is not known. We speculate that it may originate from gender differences in a variety of patient characteristics such as in the aetiology of stone disease, dietary/environmental factors and stone composition. Defining the underlying responsible factor(s) for gender-related differences requires more detailed data, which are not available in the present study.

We created a scoring system from the final multivariate proportional hazard model to evaluate each patient and predict their stone-free probabilities. Our scoring system is based on the five most significant clinical factors: previous history of ipsilateral stone treatment, age, gender, stone burden and stone location. The calculated probability of stone-free status by using this scoring tool may be more effective to choose the most suitable treatment option.

The present study has several limitations. First, our data does not include stone composition that could be a significant predictor of the stone-free rate and needed for constructing more accurate nomograms. However, in clinical practice, stone composition is often not known at the time of determining treatment; therefore, our nomogram tables and scoring system may more applicable in the clinical setting. Nevertheless we do recognize that, because stone composition in children may have wide geographic variations, our nomogram could consequently be only specific for the population from which the data was obtained. Thus, our scoring system needs to be validated in several populations in order for it to be more widely applicable in various populations. Second, our study group includes only those who presented with radio-opaque stones, and hence, our nomogram is specific for predicting stone-free rates after ESWL treatment in radio-opaque stones. Third, the present data were retrospective; although the data were collected longitudinally, they were verified retrospectively, which could have introduced error. Finally, all the treatment was done by one urologist (N.T.) at a single institution using a second-generation single machine. Despite a small decrease in the success rate, a major advantage of the second-generation lithotriptor is to decrease the rate of treatment requiring general anaesthesia. Currently many believe that second/ third or newer generation machines perform similarly with a higher retreatment rate than with the original HM-3 (Dornier MedTech America, Kennesaw, Georgia) [9]. In the present study, the bootstrap method was used for internal validation to reduce overfit bias; however, it will also be necessary to perform external validation to investigate the differences in institutions or treatment machines. We believe that multi-institutional testing of these data will serve to validate them and further demonstrate their clinical usefulness.

In conclusion, the present analysis shows that stone-free status for children with urolithiasis depends on previous history of ipsilateral stone treatment, age, gender, and stone burden and location. Interestingly, we found that stone location is only significant in girls. We believe that the nomogram table and scoring system developed in the present study could be useful for clinicians in counselling the parents of children with urolithiasis and in recommending treatment. However, multi-institutional testing will be needed to validate our findings and the utility of the scoring system.

Conflict of Interest

None declared.

Ancillary