A 970 Hounsfield units (HU) threshold of kidney stone density on non-contrast computed tomography (NCCT) improves patients' selection for extracorporeal shockwave lithotripsy (ESWL): evidence from a prospective study
Idir Ouzaid, Department of Urology, Bichat–Claude Bernard Hospital, 46 rue Henri-Huchard, 75877 Paris, Cedex 18, France. e-mail: email@example.com
Study Type – Therapy (prospective cohort)
Level of Evidence 2b
What's known on the subject? and What does the study add?
Stone density on non-contrast computed tomography (NCCT) is reported to be a prognosis factor for extracorporeal shockwave lithotripsy (ESWL). In this prospective study, we determined that a 970 HU threshold of stone density is a very specific and sensitive threshold beyond which the likelihood to be rendered stone free is poor. Thus, NCCT evaluation of stone density before ESWL may useful to identify which patients should be offered alternative treatment to optimise their outcome.
• To evaluate the usefulness of measuring urinary calculi attenuation values by non-contrast computed tomography (NCCT) for predicting the outcome of treatment by extracorporeal shockwave lithotripsy (ESWL).
PATIENTS AND METHODS
• We prospectively evaluated 50 patients with urinary calculi of 5–22 mm undergoing ESWL.
• All patients had NCCT at 120 kV and 100 mA on a spiral CT scanner. Patient age, sex, body mass index, stone laterality, stone size, stone attenuation values (Hounsfield units [HU]), stone location, and presence of JJ stent were studied as potential predictors.
• The outcome was evaluated 4 weeks after the ESWL session by NCCT.
• ESWL success was defined as patients being stone-free (SF) or with remaining stone fragments of <4 mm, which were considered as clinically insignificant residual fragments (CIRF).
• Our survey concluded that 26 patients (52%) were SF, 12 (24%) had CIRF and 12 (24%) had residual fragment on NCCT after a one ESWL treatment.
• Stones of patients who became SF or had CIRF had a lower density compared with stones in patients with residual fragments [mean (sd) 715 (260) vs 1196 (171) HU, P < 0.001].
• The Youden Index showed that a stone density of 970 HU represented the most sensitive (100%) and specific (81%) point on the receiver-operating characteristic curve.
• The stone-free rate for stones of <970 HU was 96% vs 38% for stones of ≥970 HU (P < 0.001). A linear relationship between the calculus density and the success rate of ESWL was identified.
• The use of NCCT to determine the attenuation values of urinary calculi before ESWL helps to predict treatment outcome, and, consequently, could be helpful in planning alternative treatment for patients with a likelihood of a poor outcome from ESWL.
The incidence of nephrolithiasis is reported to be increasing across the world [1–3]. This increase is seen regardless of factors such as sex, race, and age. Obesity, diminished fluid and calcium consumption, increased oxalate, sodium and animal protein intake are considered to be among the most important environmental risk factors [4,5]. Due to its frequency, urolithiasis is of particular concern for health economics. An analysis of the 2009 French data, issued from the national coding system for in-hospital stays and interventions, using the term ‘urolithiasis’ revealed a total cost of >168 million €. Lotan et al.  examined treatment methods of renal colic in the Emergency Room in 10 countries in Europe and the USA. The costs ranged from $80 to $750 (American dollars). The greatest proportion of the total cost was related to radiological investigations in the Emergency Room (40.5%), followed by treatment costs (19.7%) . Therefore, scheduling the management of the disease is of utmost importance in decreasing the subsequent costs after diagnosis.
The introduction of ESWL in the early 1980s dramatically changed the management of urinary tract stones. At the same time, the treatment of kidney stones has been revolutionised by the development of new lithotripters, modified indications and treatment principles. Modern lithotripters are smaller and usually included in uro-radiological tables, thus, facilitating the application not only of the ESWL itself, but also of other diagnostic and ancillary procedures associated with it.
Most patients with nephrolithiasis are now imaged with non-contrast CT (NCCT). New predictors of ESWL success including stone attenuation are being identified with data provided by NCCT [9,10].
The objective of the present study was to prospectively assess the usefulness of measuring urinary calculi attenuation values by NCCT in predicting the outcome of treatment with ESWL.
PATIENTS AND METHODS
From January to June 2010, we prospectively examined 50 patients with urinary calculi undergoing ESWL. Our study involved adult patients (aged >18 years) with solitary renal stones of >5 mm and <20 mm and any body mass index (BMI). Exclusion criteria were uncontrolled coagulopathy, on-going UTI, and pregnancy.
Preoperative radiological evaluation included plain abdominal radiograph of the kidneys, ureters and bladder and NCCT at 120 kV and 100 mA on a spiral CT scanner (Light speed VCT™, General Electric Company, Fairfield, Connecticut, USA) performed by the same uro-radiologist (P.F.). Laboratory tests included coagulation and urine analysis. Before the procedure, urine cultures were obtained, and, if positive, appropriate antibiotics were prescribed for 1 week. Urine cultures were repeated to document sterile urine. The ESWL procedure was performed under i.v. sedation on an outpatient basis with a Dornier DoliS lithotripter (Dornier, Medtech, Munich, Germany). Patients were delivered 3000 shocks at 90 shocks/min to each stone. During the procedure, pain was reported using a pain visual scale (PVS). After the procedure, patients were prescribed analgesics (1 g paracetamol, 4 times/day for 1 week if needed).
Patient age, sex, BMI, stone laterality, stone size, stone attenuation values (Hounsfield units [HU]), stone location, and presence of JJ stent were recorded. The primary endpoint was to determine the best attenuation value threshold to predict outcomes after ESWL. Institutional Review Board approval was obtained. The patients included in the study were informed and given a leaflet of the procedure issued by the French association of Urology, and all consented to be included in the study.
Preoperative consultation and follow-up was performed by one surgeon (sd). All patients were scheduled a medical meeting with a second NCCT 4 weeks after the ESWL to evaluate success and complications. ESWL success was defined as patients being stone free (SF) or with remaining stone fragments of <4 mm, which were considered as clinically insignificant residual fragments (CIRF). Remaining fragments of ≥4 mm were considered as treatment failures.
Descriptive data are presented as the mean (sd). The Student t-test was used to compare continuous variables, e.g. age and stone size. Fisher's exact test was used for categorical variables, a two-tailed P < 0.05 was considered to indicate statistical significance. The Youden index is a single statistic that captures the performance of a diagnostic test. The Youden Index was used to determine the most sensitive and specific threshold of stone density predicting the outcome of ESWL. At this threshold of density, the area under the receiver–operating characteristic (ROC) curve (AUC) is the most important. A cross validation (internal validation) was performed to limit the bias and standard error. This cross validation is achieved in two steps. First, a 1000 bootstrapping is used to obtain a CI for this threshold. Second, a Jackknife test is used to validate the observed performances. Factors with a significant effect on success rates according to univariate analysis were further analysed through multivariate analysis using logistic regression adjusted on the calculated threshold.
In all, 26 (52%) patients were SF, 12 (24%) had CIRF, and 12 (24%) had residual fragment on the NCCT at 4 weeks after one ESWL treatment. The patients' characteristics are shown in Table 1. The three groups were similar for age, BMI, and stone size. The ESWL procedures were equally tolerated in the different groups. In all, 27 (48%), 19 (38%), and four (8%) of the patients reported a PVS <2, 2–5, and >5, respectively. No complications have been reported.
Table 1. The patients' characteristics
Mean (sd) age, years
Stone size, mm
Stone density, HU
Stones of patients rendered SF or with CIRF (success) had a lower density than the stones of patients with significant (failure) residual fragments (715 (260) vs 1196 (171) HU, P < 0.001). Other factors potentially predicting the outcome included stone location and the presence of a JJ stent at ESWL (Table 2). There was a linear relationship between the calculus density and the success rate of ESWL. The Youden Index showed that a stone density of 970 HU represented the most sensitive (100%) and specific (81%) point on the ROC curve corresponding to the largest AUC (0.94) as shown in Fig. 1. After the internal cross validation, our results were still consistent (Table 3). Predicting factors in the univariate analysis remained statistically significant after the multivariate analysis (Table 4). Table 5 shows significant results when outcome of the ESWL is stratified on the threshold of density at 970 HU.
Table 2. Univariate analysis of factors predicting success after ESWL at 4 weeks
Mean (sd) age, (years)
Stone size, mm
Stone density, HU
Table 3. Observed and Jackknife (cross-sectional validation) performance of NNCT at the 970-HU threshold for the success of ESWL
Positive predictive value
Negative predictive value
Likelihood ratio (positive test)
Likelihood ratio (negative test)
Table 4. Multivariate analysis of factors predicting success after SWL at 4 weeks
Odds ratio (95%CI)
Stone density, HU
Stone size, mm
Table 5. Stratified outcome of the ESWL on the threshold of density at 970 HU showed significant results
After 6 months follow-up, patients with CIRF were offered ‘watchful waiting’ treatment. Among the 12 patients with significant fragments, eight were treated with a second ESWL, three had a rigid ureteroscopy because of fragments migration to the pelvic ureter, and one had a flexible ureteroscopy for a remaining UUT fragment.
ESWL can remove up to 90% of stones in adults . The success rate for ESWL depends on the efficacy of the lithotripter, the location of the stone mass (pelvic or calyceal), the composition (hardness) of the stones and the patient's habits. Each of these factors has an important influence on the re-treatment rate and the final outcome of ESWL .
The present data confirmed often reported risk factors of failure: stone location, the presence of a JJ stent at ESWL and high density stones. However, in the present study stone size, assessed as a continuous variable, was not a predictive factor of outcome. We presume this due to either the limited size of the study population or the homogeneity of the stone size within it (type one error). Thus, given that current published data supports stone size as a predictive factor of outcome during ESWL, it has been included in the multivariate analysis [9,10,13,14].
The clearance rate of stones located in the pelvis is higher than those located in the calyces. The clearance rate for upper pole stones is faster than for stones in the lower pole. Many kidney stones are located in the lower calyx and the most effective way to treat these stones is vigorously debated. In up to 35% of patients treated with ESWL, the lower calyces are incompletely cleared of disintegrated stone material. Percutaneous nephrolithotomy (PCNL) and ureteroscopic removal of the stones have been suggested to clear lower calyx stones. However, the results are contradictory. On the one hand, a relatively high morbidity is associated with PCNL. On the other hand, a multicentre randomised comparison between ESWL and ureteroscopic removal of stones from the lower calyceal system failed to show a significantly better result with ureteroscopy . Based on these outcomes, ESWL is recommended for stones with a diameter of <20 mm, despite the lower clearance rate of lower calyx fragments.
In the present study, as in previous reports, the absence of a ureteric stent is known to be a significant determinant of SF status . However, stents usually prevent obstruction and loss of ureteric contraction when large stones are to be treated. As recommended, we inserted an internal stent before ESWL only when stones were >20 mm . Conversely, a ureteric stent was inserted in 10 (20%) of our patients with stones of <20 mm for two reasons. First, if pain relief could not be achieved by medical means (NSAIDs and opioid) . Second, some patients were initially managed in other institutions by a ureteric stent and then referred to our department. If a stent was inserted, it was removed when the patient was rendered SF or CIRF.
As the stone density on NNCT has been reported to be correlated with ESWL outcome , the measurement of mean stone density by NCCT plays an important role in predicting stone disintegration. Stones with medium density of >1000 HU are less likely to be disintegrated . A threshold of 750 HU has also been reported .
Herein, we propose a new threshold of 970 HU, calculated using the Youden index on the outcome of a prospective cohort, to predict successful ESWL treatment. This threshold, obtained by measurements by a single radiologist, has been validated internally using the Jackknife test on the observed performance. Both the sensitivity and specificity of the test did not vary to a large extent (sensitivity 100% vs 91% and specificity 81% vs 81%). Moreover, further stratified analysis of the cohort based on this threshold showed highly significant results. Given the size of the present cohort, concerns about the power of the study may arise. A posteriori, calculation of the power of the study based on the observed means of density in the success and failure groups in a two-tailed fashion retrieved a >90% power. Therefore, we presume the consistency of the present results.
The use of the CIRF as a reasonable measurement of success may potentially be controversial. However, this issue has already been evaluated. Osman et al.  reported CIRF cleared spontaneously within few weeks in 78.6% of the patients, and did not recur within the 5-year follow-up of his study. Furthermore, fragment size of ≥4 mm and a history of recurrent stone disease were reported to be significant independent predictors of a clinically significant outcome . Considering this, the updated European Association of Urology guidelines recommend only ‘reasonable follow-up’ for asymptomatic CIRF .
One of the limitations of the present study is related to the timing of the outcome evaluation. As additional fragments may pass up to 3 months after the procedure, the endpoint evaluation at 4 weeks may seem short. This may overestimate our failure rate. Nevertheless, that some reports have shown that stones of 4–6 mm pass spontaneously within a period of 39 days after the procedure in 95% of cases, contribute to limit this bias . Besides, among the present patients, seven had a stone density of >970 HU and had a successful outcome. This means that any of the predictive factors taken separately cannot identify all patients who are likely to benefit from ESWL and exclude those that will have a poor outcome. A modern approach should combine all the discriminators including stone location, size, skin–stone distance, BMI, and stone density. A nomograms and neural networks approach may help to improve patients' selection [20,21] and should be further investigated. However, such models should be validated externally with different set of data than those used originally to develop them.
Finally, although CT is associated with greater radiation exposure and costs than plain radiography, it has greater sensitivity in detecting residual fragments after SWL than plain radiography . A density calculation improves patient selection for ESWL and selected patients are likely to become SF or have CIRF. For others, alternative treatments, including PCNL or ureteroscopy, could improve care by avoiding repetitive treatments and ultimately this could potentially be more cost-effective.
In conclusion, the results of the present study confirmed that stone location and the presence of a ureteric stent are associated with a poor outcome of ESWL. It also showed that the use of NCCT for determining the attenuation values of urinary calculi before ESWL helps to predict treatment outcome. A stone density of >970 HU suggests a poor chance of stone disintegration with a DoliS lithotripter. The use of this threshold could help to plan alternative treatment in patients with the likelihood of ESWL failure.
The authors are grateful to Dr A. Kettaneh for the statistical and methodological contribution to this work.