Accuracy of ultrasonography for renal stone detection and size determination: is it good enough for management decisions?

Transcription

1 Upper Urinary Tract Accuracy of ultrasonography for renal stone detection and size determination: is it good enough for management decisions? Vishnu Ganesan*,, Shubha De*, Daniel Greene*, Fabio Cesar Miranda Torricelli* and Manoj Monga* *Glickman Urological Kidney Institute, and Lerner College of Medicine, Cleveland Clinic, Cleveland, OH, USA Objectives To determine the sensitivity and specificity of ultrasonography (US) for detecting renal calculi and to assess the accuracy of US for determining the size of calculi and how this can affect counselling decisions. Materials and Methods We retrospectively identified all patients at our institution with a diagnosis of nephrolithiasis who underwent US followed by non-contrast computed tomography (CT) within 60 days. Data on patient characteristics, stone size (maximum axial diameter) and stone location were collected. The sensitivity, specificity and size accuracy of US was determined using CT as the standard. Results A total of 552 US and CT examinations met the inclusion criteria. Overall, the sensitivity and specificity of US was 54 and 91%, respectively. There was a significant association between sensitivity of US and stone size (P < 0.001), but not with stone location (P = 0.58). US significantly overestimated the size of stones in the 0 10 mm range (P < 0.001). Assuming patients with stones 0 4 mm in size will be selected for observation and those with stones 5 mm could be counselled on the alternative of intervention, we found that in 14% (54/384) of cases where CT would suggest observation, US would lead to a recommendation for intervention. By contrast, when CT results would suggest intervention as management, US would suggest observation in 39% (65/168) of cases. An average of 22% (119/552) of patients could be inappropriately counselled. Stones classified as 5 10 mm according to US had the highest probability (43% [41/96]) of having their management recommendation changed when CT was performed. The use of plain abdominal film of kidney, ureter and bladder and US increases sensitivity (78%), but 37% (13/35) of patients may still be counselled inappropriately to undergo observation. Conclusions Using US to guide clinical decision-making for residual or asymptomatic calculi is limited by low sensitivity and inability to size the stone accurately. As a result, one in five patients may be inappropriately counselled when using US alone. Keywords nephrolithiasis, ultrasonography, tomography, X-ray computed, kidney, #KidneyStones Introduction Ultrasonography (US) is an accessible, relatively inexpensive imaging method that comes without the risks of exposure to ionizing radiation entailed by CT [1]. Stafford et al. [2] reported the ability to detect stones as small as 2 mm using US imaging in a porcine model more than 30 years ago. With an ability to demonstrate radiopaque and radiolucent stones, hydronephrosis, renal inflammation, ruptured fornices, ureteric jets and resistive index, US can provide valuable clinical information. Despite the wider availability of US units and increased bedside utilization [3], the national usage of US for renal colic had not significantly changed from 2000 to 2008, although the use of CT scans has increased dramatically [4]. An AUA best practice statement recommends follow-up imaging after endoscopic procedures with US or US/plain abdominal film of kidney, ureter and bladder (KUB) [5]. This allows the treating urologist to counsel patients on any residual fragments, while ruling out the presence of silent obstruction. The relative safety and low cost of US were noted to justify its use in the detection of the relatively rare but serious complications of silent obstruction attributable to residual stones or ureteric stricture formation. BJU Int 2017; 119: wileyonlinelibrary.com BJU International 2016 BJU International doi: /bju Published by John Wiley & Sons Ltd.

2 Accuracy of ultrasonography in renal stone size detection Stone presence and size play a critical role in the counselling of patients [6]. Our specific objective, therefore, was to evaluate the sensitivity of US in detecting stone presence as well as the accuracy in determining stone size, in comparison with the gold standard CT scan. Materials and Methods Study Design and Population The present study was a retrospective review of all patients from 1995 to 2012 who attended our large, urban academic tertiary centre and who had a diagnosis of kidney stone (International Classification of Disease code ICD ). We included only those patients who underwent renal US followed by abdominal CT within a 60-day window. All charts were reviewed to ensure spontaneous passage or surgical intervention did not occur in the interval between imaging methods, and such patients were excluded. Demographic details and indications for imaging were recorded. CT scans were independently reviewed by the investigators to obtain maximum stone diameter, polar location within the kidney, and skin to stone distance. CT scans were reviewed in the coronal and axial planes, and the largest diameter was used. In addition, the presence of hydronephrosis, renal masses, cysts and anatomical abnormalities was recorded. The largest stone was then chosen to be compared with findings from US imaging. As US images were saved at the discretion of the radiologist/us technologist, dictated measures were used for study comparison. Only renal stones were included. If a KUB was performed within 2 days of US for evaluation of nephrolithiasis, stone detection by the radiologist was recorded and, if so, investigators reviewed the KUB to obtain the maximum stone diameter. Stones were grouped, based on the longest axis diameter, into three categories according to clinical relevance in management: 4 mm (where observation would probably be recommended), 5 10 mm (where shockwave lithotripsy would probably be recommended) or >10 mm (where an endoscopic approach would probably be recommended). These groupings were based on previously reported practice patterns [7]. Analyses of sensitivity and specificity for detection of stones on US were calculated using CT as the standard reference. Statistical Analysis Descriptive statistics are reported as medians with interquartile ranges (IQRs) or means with standard deviations, and compared using the Wilcoxon signed-rank test because of the paired nature of the measurements. Categorical variables are described with proportions and compared using the chi-squared test. When paired categorical data were compared, such as when the sensitivity of KUB and US were compared with US alone, the McNemar test was used. Multivariable logistic regression was used to identify factors associated with detection of stone on US. Multivariable linear regression was used to identify factors associated with accuracy of stone size, as measured by CT. All results were considered significant at the level of a = The statistical software package R (Core development team, v.3) was used for the analyses. Results A total of 710 US and CT examinations were initially identified. After excluding patients who had passed their stones, who had undergone surgical intervention in the time between the two imaging methods, or were imaged by CT before the US, 486 patients remained for analysis, comprising 552 US and CT pairs. Cohort Characteristics The median (IQR) time from US to CT scan was 11 (5 23) days. Indications for the initial US included stone follow-up (31%), post-procedural follow-up (19%), renal failure (15.7%), LUTS (14.5%), haematuria (10.3%), cyst/mass follow-up (10.3%), infectious evaluation (8.3%) and hydronephrosis follow-up (6.9%). Patients in the cohort had a median (IQR) body mass index (BMI) of 28.4 ( ) kg/m 2.CT detected a significantly greater number of stones: 299 vs 184 on US (P < 0.001). The median (IQR) stone length on CT was 5 (3 8) mm vs 8 (6 12) mm on US, corresponding to a median overestimation by US of 1 ( 0.25 to 4.7) mm (P < 0.001), in patients where both US and CT detected stones. Sensitivity and Specificity of Ultrasonography The sensitivity and specificity of US in detecting any stone were 54% (95% CI 48 59) and 91% (95% CI 86 94%), respectively. The sensitivity of US stratified by size category (using sizes measured on CT as the standard) increased with larger stone sizes (Table 1; P < 0.001). There was no correlation between US sensitivity and location of the stone (Table 2; P = 0.47). On multivariable analysis, adjusting for the patient s BMI, stone size, and stone location; only stone size remained independently associated with detection of stone on US (P < 0.001). Because of the wide time span of the present study we investigated whether sensitivity and specificity differed between early and late studies using thresholds of before and after Before 2005, the sensitivity of US was 56% (95% CI 38 73) and specificity 89% (95% CI 74 97). After 2005, the sensitivity was 53% (95% CI 47 59) and specificity 91% (95% CI 86 94). The majority of the studies were performed after 2005 (87%). BJU International 2016 BJU International 465

3 Ganesan et al. Accuracy of Stone Size Measurement and Location Our analysis showed that US overestimated stone sizes for size groups 0 4 mm and 5 10 mm (P < 0.001). There was no significant difference for stones >10 mm, although US tended to underestimate (P = 0.05; Table 3). On multivariable linear regression, no statistically significant association was found between the accuracy of US and BMI or stone location. With regard to location, in particular the lower pole, the CT location was concordant with US location in 67% (95% CI 54 79) of cases. We also conducted an analysis that included a time variable in the model, dichotomizing to studies carried out before 2005 or after 2005; we found no significant association with accuracy (P = 0.59). How Stone Size Measurement on Ultrasonography Affects Counselling Decisions Because stone size measurements affect counselling decisions, we analysed how counselling would be changed based on which imaging method was used. In 3% of cases, US reported a >10 mm stone when there was no stone on CT. By contrast, US did not detect 30% of stones >10 mm, a reflection of its low sensitivity. Table 1 Sensitivity of ultraonography, stratified by stone size. Stone size Sensitivity (95% CI), % 2 mm 28 (18 40) 4 mm 37 (29 46) 5 10 mm 64 (55 73) >10 mm 70 (55 83) To further assess the impact of imaging method, we assumed that, for patients with no stones or stones 4 mm, observation would be recommended while for patients with stones 5 mm, an intervention would be recommended. Under these assumptions, in a situation where, with the use of CT, observation would have been recommended, the results of US would have recommended an intervention in 14% of patients (95% CI 11 8). Conversely, where with the use of CT an intervention would have been recommended, using US, 39% (95% CI 31 47) of those patients would have undergone observation (Table 4). On average, therefore, 22% (95% CI 18 25) of patients could have been inappropriately counselled by relying on US alone. We then repeated the analysis, grouping by size categories provided by US. This was to answer the question, given that a stone on US is within a certain size group, what is the probability that performing a CT would change management for that stone. We found that stones classified as 5 10 mm by US had the highest probability (43%) of having the management recommendation changed when a CT was performed (Table 5). Plain Abdominal Film of Kidney, Ureter and Bladder and Ultrasonography Overall, 86 patients (16%) underwent both KUB and US, performed in addition to CT. The sensitivity of KUB and US combined was 78% (95% CI 64 89) and increased with size: 73% (39 94%) for 0 4 mm, 77% (56 91%) for 5 10 mm and 89% (52 99%) for >10 mm. Overall, the addition of KUB Table 4 Concordance by stone size groups. US size group CT size group Table 2 Sensitivity of ultraonography by stone location. Stone location Sensitivity (95% CI), % Upper calyx 45 (32 60) Middle calyx 55 (43 66) Lower calyx 54 (45 64) Pelvis 60 (45 74) No stone 0 4 mm 5 10 mm >10 mm No stone mm mm >10 mm US, ultrasonography. Table 5 Probability that CT would change counselling recommendation when size on US is available. Table 3 Stone sizes measured by ultrasonography compared with CT. Stone size group Size on CT* Size on US* Difference US CT* 0 4 mm 3(2 4) 6 (4 8) 4 (3 5) < mm 7 (6 9) 8 (6 12) 1 (0 5) <0.001 >10 mm 14 (11 18) 12 (10 14) 2 ( 7 to 1) 0.05 US, ultrasonography. *Sizes reported as median (interquartile range). **Measurements compared using the Wilcoxon signed-rank test. P** US size group Probability CT would change counselling*, % No stone mm mm 43 >10 mm 21 US, ultrasonography. *We defined counselling change as going from observation to intervention or intervention to observation. For analysis purposes we assumed that patients with no stones or stones <4 mm will be recommended to undergo observation and patients with stones >5 mm will be recommended an intervention. 466 BJU International 2016 BJU International

4 Accuracy of ultrasonography in renal stone size detection significantly increased the sensitivity for detecting a stone compared with US alone (P < 0.001). We then compared accuracy of stone measurement from KUB with CT. A total of 31 patients remained who had a stone detected on both KUB and CT. The median (IQR) stone size on KUB was 6 ( ) mm and on CT was 7.0 ( ) mm. There was no significant difference in measured size between KUB and CT (P = 0.11). If we base counselling decisions on the combination of KUB and US, using the size from KUB when available, 23% (95% CI 13 38) of patients could be inappropriately advised to undergo intervention, while 37% (95% CI 21 55) would be inappropriately counselled to be observed. Discussion Accurate detection and measurement of renal calculi is essential for guiding management decisions and clinical decision-making. A non-contrast helical CT has high specificity and sensitivity and is considered to be the gold standard for the diagnosis of kidney stones [8]. To reduce radiation exposure, US is often used in both the initial assessment and follow-up of patients with kidney stones. Several studies comparing US with CT have shown low sensitivity (24 69%) and high specificity (82 90%) for US [9 11]. A few studies have examined the accuracy of size measurements on US and have found that US tends to lead to an overestimate of the size compared with CT [9,12]; however, these studies have been limited by small sample sizes. In the present large study of ~552 CT/US pairs, we found that the sensitivity of US was low, at 54% overall, but increases with larger stone sizes, ranging from 28 to 73%. The specificity of US remains high at 91%. This sensitivity was not significantly affected by location of the stone or the patient s BMI. This is consistent with previously reported data. In a series of 123 US and CT examinations, Fowler et al. [9] found that the sensitivity of US for any calculi was 24%, with a specificity of 90%. Similarly to the present study, the sensitivity of US increased with size, from 13% for stones 3.0 mm to 71% for stones >7.0 mm. Ray et al. [12], in a series of 71 comparisons, found US to overestimate size by a mean of 1.8 mm. Unal et al. [10] found a sensitivity and specificity of 69 and 87% for US when stone confirmation was determined using stone passage, ureteroscopy or retrograde pyelography. Ather et al. [13] report a sensitivity and specificity of 81 and 100% for US in a small series of 34 patients. The accuracy of stone sizes measured on US vs CT likewise varied by size group. In general, US significantly overestimated the stone size for stones of 0 10 mm, with a trend toward underestimating at sizes >10 mm. This again supports findings reported in previous studies. In their series, Fowler et al. [9] found size measurements on US were concordant with CT in 79% of cases and US tended to overestimate by an average of mm. Dunmire et al. [14] showed similar findings using an in vitro water bath model, with a mm mean overestimation, depending on the type of US device and settings [14]. Groupings used in this analysis ( 4 mm, 5 10 mm, >10 mm) are common size considerations when counselling patients on stone intervention [7]. While sensitivity, specificity and accuracy of stone size are important considerations; at the patient level, the question to consider is: does it change management? Under these size groupings, we found that, by using US, up to 22% of patients may be counselled differently than if CT was used to confirm stone size. Furthermore, findings from the present study suggest that for the patient who has undergone US, those with stones particularly in the 5 10 mm range, as measured by US, may benefit from follow-up with CT because management can change in 43% of cases within this size group. Location also plays a role when counselling patients about intervention. A lower pole stone location significantly affects outcomes with shockwave lithotripsy, and may affect outcomes with ureteroscopy [15,16]. In the present study, the concordance in identifying a lower pole location between US and CT was poor, further confounding the ability to render management recommendations based on US alone. The addition of KUB has been shown to significantly increase the sensitivity of US, with one study showing a sensitivity of 96% [17]. The addition of KUB increased sensitivity from 54 to 78% across all stones, again with increasing sensitivity for larger stones. Notably, the combination of US and KUB missed 11% of stones >10 mm in size. KUB also showed a better accuracy in stone size measurements compared with US, but with only 31 patients with both a stone on KUB and CT, we may be underpowered to detect a difference. Even with the addition of KUB, approximately a third of patients may be inappropriately counselled for observation, with this primarily driven by the lower sensitivity. The sensitivity of US in detecting renal stones is low compared with CT. While this improves with increasing stone size, our data suggest that US misses 27% of even largesized stones (>10 mm). There are several reasons for this. To identify a stone on US, there needs to exist a hyperechoic focus with an acoustic shadow. The acoustic shadow can be impaired because of the impedance of intervening tissue or inappropriate selection and balance of the transducer power and focal length [18]. In addition, given the variety of settings and techniques, US is more operator-dependent than CT. An inexperienced operator may not use the ideal settings to pick up the acoustic shadow or may not scan the entire kidney. The variable accuracy of US reflects the significant variety of equipment, settings, techniques and observer skills used to produce the results. Two common techniques for stone BJU International 2016 BJU International 467

5 Ganesan et al. detection, increasing US depth and gain, were found in an in vitro study to overestimate stone size [14]. In the same study, Dunmire et al., found that overestimation of stone sizing could be eliminated using standardized stone detection software [14]. In another study by Dunmire et al. [19] the measurement of stone shadow width as opposed to the stone itself led to significantly more accurate stone measurements [19]. Recently, in a prospective study, detection of stone twinkling using colour Doppler US led to good detection rates of ureteric stones in an emergency room setting [20]. Given the heterogeneity of US techniques, personnel and equipment, it is not surprising that there is variability in the stone sizing relative to the CT scan. While there are techniques available to improve US accuracy, the present study provides valuable insight as it represents the cumulative results across a wide variety of practice patterns in a typical high-volume centre. This analysis represents the largest reported series of patients comparing both US and CT for stones within the kidney. Shortcomings of this study include its retrospective nature, and the potential for undocumented stone passage/removal or changes in position during follow-up imaging. We attempted to control for this by identifying such events on chart and excluding those patients. In addition, the US measurements of renal calculi were not performed under standardized conditions, as multiple US technicians and radiologists performed the studies. Settings, techniques and experiences among ultrasonographers vary and therefore so do the resulting sensitivity and accuracy of measurements; however, this reflects a real-world clinical setting and the data with which treatment decisions are made. Changes in stone positioning between imaging dates could also affect the measurements being considered. Finally, our analyses of how US vs CT changes counselling rests on the assumption that stones 0 4 mm will be observed and stones 5 mm will require an intervention. We realize that the decision to observe or treat a stone depends on more than size alone and as such we may overestimate the change in management rate. In conclusion, the appeal of the low cost of US, its accessibility, functional assessments and lack of ionizing radiation make it an ideal imaging method for renal pathology. The use of US for guiding management decisions for residual or asymptomatic calculi is limited because of its low sensitivity for smaller stones and its inability to accurately size the stone. As a result, one in five patients may be inappropriately managed when using US alone. Conflict of Interest None declared. References 1 Brenner DJ, Hall EJ. Computed tomography an increasing source of radiation exposure. N Engl J Med 2007; 357: Stafford SJ, Jenkins JM, Staab EV, Boyce I, Fried FA. Ultrasonic detection of renal calculi: accuracy tested in an in vitro porcine kidney model. JCU 1981; 9: Dalziel PJ, Noble VE. Bedside ultrasound and the assessment of renal colic: a review. Emerg Med J 2013; 30: Hyams ES, Korley FK, Pham JC, Matlaga BR. Trends in imaging use during the emergency department evaluation of flank pain. J Urol 2011; 186: Fulgham PF, Assimos DG, Pearle MS, Preminger GM. Clinical effectiveness protocols for imaging in the management of ureteral calculous disease: AUA technology assessment. J Urol 2013; 189: Burgher A, Beman M, Holtzman JL, Monga M. Progression of nephrolithiasis: long-term outcomes with observation of asymptomatic calculi. J Endourol 2004; 18: Bandi G, Best SL, Nakada SY. Current practice patterns in the management of upper urinary tract calculi in the north central United States. J Endourol 2008; 22: Dalrymple NC, Verga M, Anderson KR et al. The value of unenhanced helical computerized tomography in the management of acute flank pain. J Urol 1998; 159: Fowler KAB, Locken JA, Duchesne JH, Williamson MR. US for detecting renal calculi with nonenhanced CT as a reference standard. Radiology 2002; 222: Unal D, Yeni E, Karaoglanoglu M, Verit A, Karatas OF. Can conventional examinations contribute to the diagnostic power of unenhanced helical computed tomography in urolithiasis? Urol Int 2003; 70: Ulusan S, Koc Z, Tokmak N. Accuracy of sonography for detecting renal stone: comparison with CT. J Clin Ultrasound JCU 2007; 35: Ray AA, Ghiculete D, Pace KT, Honey RJD. Limitations to ultrasound in the detection and measurement of urinary tract calculi. Urology 2010; 76: Ather MH, Jafri AH, Sulaiman MN. Diagnostic accuracy of ultrasonography compared to unenhanced CT for stone and obstruction in patients with renal failure. BMC Med Imaging 2004; 4: 2 14 Dunmire B, Lee FC, Hsi RS et al. Tools to improve the accuracy of kidney stone sizing with ultrasound. J Endourol 2015; 29: Albala DM, Assimos DG, Clayman RV et al. Lower pole I: a prospective randomized trial of extracorporeal shock wave lithotripsy and percutaneous nephrostolithotomy for lower pole nephrolithiasis-initial results. J Urol 2001; 166: Pearle MS, Lingeman JE, Leveillee R et al. Prospective, randomized trial comparing shock wave lithotripsy and ureteroscopy for lower pole caliceal calculi 1 cm or less. J Urol 2005; 173: Mitterberger M, Pinggera GM, Pallwein L et al. Plain abdominal radiography with transabdominal native tissue harmonic imaging ultrasonography vs unenhanced computed tomography in renal colic. BJU Int 2007; 100: King W, Kimme-Smith C, Winter J. Renal stone shadowing: an investigation of contributing factors. Radiology 1985; 154: Dunmire B, Harper JD, Cunitz BW et al. Use of the acoustic shadow width to determine kidney stone size with ultrasound. J Urol 2016; 195: Abdel-Gawad M, Kadasne RD, Elsobky E, Ali-El-Dein B, Monga M. A prospective comparative study between color doppler ultrasound with twinkling and non-contrast computed tomography in the evaluation of acute renal colic. J Urol 2016; 196: BJU International 2016 BJU International

6 Accuracy of ultrasonography in renal stone size detection Correspondence: Manoj Monga, Glickman Urological and Kidney Center, Cleveland Clinic, 9500 Euclid Avenue, Q10-1, Cleveland, OH 44195, USA. Abbreviations: IQR, interquartile range; KUB, plain abdominal film of kidney, ureter and bladder; US, ultrasonography. BJU International 2016 BJU International 469