Stone heterogeneity index as the standard deviation of Hounsfield units: A novel predictor for shock-wave lithotripsy outcomes in ureter calculi

We investigated whether stone heterogeneity index (SHI), which a proxy of such variations, was defined as the standard deviation of a Hounsfield unit (HU) on non-contrast computed tomography (NCCT), can be a novel predictor for shock-wave lithotripsy (SWL) outcomes in patients with ureteral stones. Medical records were obtained from the consecutive database of 1,519 patients who underwent the first session of SWL for urinary stones between 2005 and 2013. Ultimately, 604 patients with radiopaque ureteral stones were eligible for this study. Stone related variables including stone size, mean stone density (MSD), skin-to-stone distance, and SHI were obtained on NCCT. Patients were classified into the low and high SHI groups using mean SHI and compared. One-session success rate in the high SHI group was better than in the low SHI group (74.3% vs. 63.9%, P = 0.008). Multivariate logistic regression analyses revealed that smaller stone size (OR 0.889, 95% CI: 0.841–0.937, P < 0.001), lower MSD (OR 0.995, 95% CI: 0.994–0.996, P < 0.001), and higher SHI (OR 1.011, 95% CI: 1.008–1.014, P < 0.001) were independent predictors of one-session success. The radiologic heterogeneity of urinary stones or SHI was an independent predictor for SWL success in patients with ureteral calculi and a useful clinical parameter for stone fragility.

Factors affecting the one-session success and stone-free status. Univariate and multivariate logistic regression models were generated for one-session success and stone-free status. The multivariate model revealed that smaller stone size (OR 0.889, 95% Confidence Interval [CI]: 0.841-0.937, P < 0.001), lower MSD (OR 0.995, 95% CI: 0.994-0.996, P < 0.001), and higher SHI (OR 1.011, 95% CI: 1.008-1.014, P < 0.001) were independent predictors of one-session success. Similarly, stone size (OR 0.886, 95% CI: 0.839-0.933, P < 0.001), MSD (OR 0.996, 95% CI: 0.995-0.997, P < 0.001), and SHI (OR 1.008, 95% CI: 1.005-1.010, P < 0.001) also had an independent impact on one-session stone-free status ( Table 2).  Impact of SHI on SWL outcomes according to stone size and MSD. In patients with a stone size ≥ 10 mm, the one-session success rate was 50.2%, and SHI was significantly different between those with success and failure (279.92 ± 115.66 HU vs. 204.99 ± 85.85 HU, P < 0.001). However, there was no difference in SHI according to the success or failure in patients with a stone size < 10 mm (Table 3). Meanwhile, one-session success rates were 75.0% in patients with a MSD < 1000 HU versus 38.0% in patients with a MSD ≥ 1000 HU. In patients with a MSD ≥ 1000 HU, SHI was significantly higher in cases with one-session success than in the cases with failure (308.02 ± 91.87 HU vs. 251.48 ± 54.51 HU, P = 0.001; Table 4).

Discussion
In the current study, we introduced the concept of radiologic heterogeneity of a urinary stone based on a HU measurement in NCCT and demonstrated the clinical significance of SHI in the management of patients with a ureteral stone. To the best of our knowledge, this is the first report dealing with this novel clinical factor, and we revealed that SHI was an independent predictor of SWL outcomes in ureteral stones. In general, urinary stones with a larger stone size (i.e., >10 mm) or a higher MSD (i.e., >1000 HU) have been deemed to be resistant to SWL, and stone size seems to be the most influential factor in predicting SWL outcomes 1,15 . Our results are in agreement with prior studies. Nevertheless, in certain stones with a higher SHI, favorable outcomes can be expected even though a stone may possess conventionally unfavorable clinical features, such as a larger stone size or a higher MSD (Tables 3 and 4). We believe that SHI can be a useful clinical parameter for stone fragility and can play a complementary role for such a clinical prediction in addition to stone size and MSD. In addition, SHI can be readily measured using the currently available PACS without additional equipment. MSD has been widely used during the last decade as an important parameter to characterize urinary stones for both research and clinical practice 16 . However, MSD is only an arithmetical average that cannot represent the heterogeneity of stone composition, as shown in Fig. 1. Conversely, the standard deviation of a random variable, statistical population, data set, or probability distribution is the square root of its variance. SHI is an index  Table 2. Univariate and multivariate logistic regression models a on one-session success and stone-free status in total patients with a ureteral stone. SSD: skin-to-stone distance, MSD: mean stone density, SHI: stone heterogeneity index. a A multivariate logistic regression model with forward stepwise selection was performed.

N (%) SHI (HU) P-value*
Stone size ≥ 10 mm (N = 229)  Table 3. Comparison of stone heterogeneity index (SHI) in one-session success and stone-free status according to stone size. SHI: stone heterogeneity index. *Student's two-sample t-test for stone heterogeneity index in each group.
presenting the radiological heterogeneity of a urinary stone defined as the standard deviation of HU in the region of interest, as aforementioned. Accordingly, SHI can represent the internal diversity of a stone, reflecting not only the heterogeneity of the stone's composition but also the structural and morphological heterogeneity of a stone. There can be several explanations for the difference in SHI of urinary stones. First, urinary stones are generally not monocrystals, which can be a cause for SHI. Jing et al. reported a prospective analysis of urinary calculi composition by infrared spectroscopy with 625 patients in eastern China 17 . They showed that 37.4% of urinary stones are pure stones, but most urinary stones were mixed (62.6%) in which calcium oxalate was the most commonly found major component. Second, the internal structure and morphology of stones can vary, even though the composition of minerals is similar, which can also contribute to such differences in SHI value. Urinary stones present with a variety of gross appearances according to their contour irregularities, such as smoothly round, spiculated, and mulberry stones 18 . Meanwhile, the internal structure showing a heterogeneity of composition or cracks were also detected using NCCT even in stones with the same attenuation; cross-sectional images of such a stone can differ from mottled to lamellar structures 19 . In addition, there can be some empty space within urinary calculi structural irregularities. This space can be filled with water or air, which might be an important cause of heterogeneity of the attenuation index. Conventionally, stone composition has been undoubtedly important in determining the efficacy of stone treatment, especially SWL. The most dramatic differences have been found with radiolucent uric acid calculi (easily fragmented with SWL) and relatively radiolucent cysteine calculi (often refractory to SWL), which is useful information in selecting stone treatment 20 . Accordingly, knowing the composition of urinary calculi is essential for deciding the optimal mode of treatment. Urine pH, the presence of crystals, urease-positive bacteria in the urine, a plain x-ray, and a history of urinary stones have long been used to predict the composition of stones 21 . During last two decades, the relationship between HU and stone composition has been investigated. Several studies demonstrated with an in vitro approach that stone composition could be predicted with high accuracy using HU and HU density (HU divided by the greatest transverse diameter) [22][23][24] . However, Toricelli et al. showed that there was an overlap between the HU values of cysteine and uric acid stones, making it difficult to differentiate these types of stones 25 . Such an overlap of values also precludes any more exact determination of stone composition by the MSD. In 2003, Williams et al. suggested that knowing the major composition of a stone may not allow adequate prediction of its fragility in lithotripsy treatment, and variations in internal stone structure, including secondary mineral composition, may be a significant cause of this variability in stone fragility 26 .
The relationship between CT parameters and SWL outcomes has also been extensively investigated, and representative studies are summarized in Table 5. In the NCCT era of urinary stone management, lots of interest has been raised about MSD and SSD as novel predictors of SWL outcomes. Most studies showed that MSD was significantly associated with SWL success, but only two studies showed no relationship between MSD and SWL outcomes 27,28 . El-Nahas showed that a MSD > 1000 HU was a significant independent predictor of SWL failure. Thus, they maintained that an alternative treatment should be offered for patients with a MSD > 1000 HU 5 . Interestingly, in cases with a MSD > 1000 HU, the success or stone-free groups demonstrated significantly higher SHIs than the failure group in our study (Table 4). However, the role of SSD as a predictor of SWL success remains controversial. Approximately half of the published studies have advocated the role of SSD in predicting SWL outcomes, but the other studies have failed to demonstrate a significant relationship between SSD and SWL success.
This study has some inherent limitations due to its retrospective design, which may have introduced sampling bias; however, we built a relatively large cohort for ureteral stone disease. In addition, to overcome this type of limitation, subjects of our study were only ureters stones to elucidate the impact of SHI on SWL outcomes more clearly. In renal stone, anatomical considerations including location of calyx and renal pelvic stone or infundibulopelvic angle can be another bias. Two different generating machines may be a bias, but there were no statistical difference in each period. Another concern is that the clinical significance of SHI may be limited due to its OR from the logistic regression analysis, which may be a major obstacle to using SHI in a clinical setting. However, multivariate analysis demonstrated that the predictive power for treatment outcomes was in order of stone size, SHI, and MSD based on odds ratios (Table 2). In addition, SHI was a significant predictor for successful outcomes in patients with stone sizes of ≥ 10 mm (Table 3), and there was a significant difference in SHI between N (%) SHI (HU) P-value* MSD ≥ 1000 HU (N = 100)  Table 4. Comparison of stone heterogeneity index (SHI) in one-session success and stone-free status according to mean stone density. MSD: mean stone density, SHI: stone heterogeneity index. *Student's twosample t-test for stone heterogeneity index in each group.
success and failure groups regardless of MSD (Table 4). Meanwhile, the relationship among stone size, MSD, and SHI is a very important and interesting issue. Correlation analyses demonstrated no relationship between stone size and SHI (r = 0.060; P = 0.114) while showing a positive correlation between stone size and MSD (r = 0.317; P < 0.001). Taken together, although the predictive power of SHI seems to be limited and weaker than stone size, SHI can play a complementary role in the prediction of treatment outcomes, similar to MSD. However, further studies with a prospective design are needed to confirm our observation on the relationship between SHI and SWL outcomes, and a clinically applicable cut-off value of SHI should be determined for the selection of proper candidates for SWL treatment. In addition, experimental studies in conjunction with chemical and structural analysis of urinary calculi would be helpful for a thorough understanding of the clinical significance of SHI.
In summary, the radiologic heterogeneity of a urinary stone or SHI was independently associated with SWL success in patients with ureteral calculi, thus SHI can be a useful clinical parameter for stone fragility. SHI may be affected by the compositional heterogeneity in urinary calculi, as well as their structural and morphological heterogeneity. SHI will play a promising role when determining a treatment modality in patients with a urinary stone, and especially when selecting the proper SWL candidates from the patients with a stone of large size or high MSD.

Methods
Patient cohort. Medical records were obtained from a consecutive database of patients who had undergone the first session of SWL between November 2005 and December 2013 in the Severance Hospital, Seoul, Korea. During this period, a total of 1,519 patients were registered in our database. Inclusion criteria for the current study were a 4-to 20-mm stone in the ureter and radiopaque calculi on a plain x-ray. Patients who did not undergo a NCCT scan were excluded. Ultimately, 604 patients with ureter calculi were eligible for the current analyses. The Institutional Review Board of Severance Hospital approved this study protocol (Approval No. 4-2014-0465).

Shock-wave lithotripsy. SWL was performed with an electroconductive lithotripter (EDAP Sonolith
Praktis, Technomed, Lyon, France) until 2011; after 2012, the device was replaced by an electromagnetic generative lithotripter (Dornier Compact Delta II lithotripter, Dornier MedTech GmbH, Wessling, Germany). All patients were treated under fluoroscopic guidance. The number of shock waves per SWL session varied from 2500 to 4000 at a rate of 60 to 90 shock waves per minute with a launch intensity ranging from 16 to 55 MPa. Table 5. Review of the literature on the relationship between stone characteristics and shock-wave lithotripsy outcomes. a Success group vs. failure group. b Stone-free vs. residual stone (p < 0.05 in both MSD and SSD). c Stone-free vs. residual stone (p > 0.05 in both MSD and SSD). SSD was measured from the center of the stone to the level of the skin at 30°. d Stone-free vs. residual stone (p > 0.05 in both MSD and SSD). e Stone-free vs. residual stone (p > 0.05 in MSD and p < 0.05 in SSD). f Stone size ≤ 10 mm vs. > 10 mm. (Fig. 2). A successful SWL treatment of ureteral and renal calculi was defined as those patients who were rendered stone-free or had asymptomatic, clinically insignificant residual fragments ≤ 3 mm in the largest stone diameter 2 weeks after a single SWL treatment 29 , as measured by a simple x-ray without the need for auxiliary measures within a 3-month follow-up period. Stone-free status was defined as when a simple X-ray analysis determined that patients had a calcification-free 2-week period after a single SWL treatment.
Statistical analysis. Statistical comparisons of continuous variables from the patient demographic information were carried out using either a Student's or Welch's two-sample t-test or the Wilcoxon rank-sum test. Categorical variables were compared using Pearson's chi-squared test. Univariate and multivariate logistic regressions with a binomial method were carried out for significant factors of one-session success and stone-free status. Statistical analyses were performed using R software (version 3.0.3, R Foundation for Statistical Computing, Vienna, Austria; http://www.r-project.org).