Using 24-Hour Urinalysis to Predict Stone Type

Predictive Modeling of Urinary Stone Composition Using Machine Learning and Clinical Data: Implications for Treatment Strategies and Pathophysiological Insights

Purpose: Preventative strategies and surgical treatments for urolithiasis depend on stone composition. However, stone composition is often unknown until the stone is passed or surgically managed. Given that stone composition likely reflects the physiological parameters during its formation, we used clinical data from stone formers to predict stone composition. Materials and Methods: Data on stone composition, 24-hour urine, serum biochemistry, patient demographics, and medical history were prospectively collected from 777 kidney stone patients. Data were used to train gradient boosted machine and logistic regression models to distinguish calcium vs noncalcium, calcium oxalate monohydrate vs dihydrate, and calcium oxalate vs calcium phosphate vs uric acid stone types. Model performance was evaluated using the kappa score, and the influence of each predictor variable was assessed. Results: The calcium vs noncalcium model differentiated stone types with a kappa of 0.5231. The most influential predictors were 24-hour urine calcium, blood urate, and phosphate. The calcium oxalate monohydrate vs dihydrate model is the first of its kind and could discriminate stone types with a kappa of 0.2042. The key predictors were 24-hour urine urea, calcium, and oxalate. The multiclass model had a kappa of 0.3023 and the top predictors were age and 24-hour urine calcium and creatinine. Conclusions: Clinical data can be leveraged with machine learning algorithms to predict stone composition, which may help urologists determine stone type and guide their management plan before stone treatment. Investigating the most influential predictors of each classifier may improve the understanding of key clinical features of urolithiasis and shed light on pathophysiology.

Routine Urinary Biochemistry Does Not Accurately Predict Stone Type Nor Recurrence in Kidney Stone Formers: A Multicentre, Multimodel, Externally Validated Machine-Learning Study

Objectives: Urinary biochemistry is used to detect and monitor conditions associated with recurrent kidney stones. There are no predictive machine learning (ML) tools for kidney stone type or recurrence. We therefore aimed to build and validate ML models for these outcomes using age, gender, 24-hour urine biochemistry, and stone composition. Materials and Methods: Data from three cohorts were used, Southampton, United Kingdom (n = 3013), Newcastle, United Kingdom (n = 5984), and Bern, Switzerland (n = 794). Of these 3130 had available 24-hour urine biochemistry measurements (calcium, oxalate, urate [Ur], pH, volume), and 1684 had clinical data on kidney stone recurrence. Predictive ML models were built for stone type (n = 5 models) and recurrence (n = 7 models) using the UK data, and externally validated with the Swiss data. Three sets of models were built using complete cases, multiple imputation, and oversampling techniques. Results: For kidney stone type one model (extreme gradient boosting [XGBoost] built using oversampled data) was able to effectively discriminate between calcium oxalate, calcium phosphate, and Ur on both internal and external validation. For stone recurrence, none of the models were able to discriminate between recurrent and nonrecurrent stone formers. Conclusions: Kidney stone recurrence cannot be accurately predicted using modeling tools built using specific 24-hour urinary biochemistry values alone. A single model was able to differentiate between stone types. Further studies to delineate accurate predictive tools should be undertaken using both known and novel risk factors, including radiomics and genomics.

Pure uric acid stone prediction model using the variant coefficient of stone density measured by thresholding 3D segmentation-based methods: A multicenter study

Urinary stones are common urological diseases with increasing prevalence and incidence worldwide. Among the various types of stones, uric acid stones can be dissolved by oral chemolysis without any surgical procedure. Therefore, our study demonstrates that variant coefficient of stone density measured by thresholding a three-dimensional segmentation-based method from noncontrast computed tomography images can be used to identify pure uric acid stones from non-pure uric acid stones. This study provides a preoperative pure uric acid stone prediction model that could reduce invasive procedural treatments. The pure uric acid stone prediction model may offer optimized clinical decision-making for patients with urinary stones.

BACKGROUND AND OBJECTIVES

While most urinary stones are managed with interventional therapy, uric acid (UA) stones can be dissolved by oral chemolysis without invasive procedures. This study aimed to develop and validate a pure UA (pUA) stone prediction model using a variant coefficient of stone density (VCSD) measured by thresholding a three-dimensional (3D) segmentation-based method.

METHODS

Patients with urolithiasis treated at Keimyung University Dongsan Hospital between January 2017 and December 2020 were divided into training and internal validation sets, and patients from Kyungpook National University Hospital between January 2017 and December 2018 were used as an external validation set. Each stone was segmented by a thresholding 3D segmentation-based method using an attenuation threshold of 130 Hounsfield units. VCSD was calculated as the stone heterogeneity index divided by the mean stone density.

RESULTS

A total of 1175 urinary stone cases in 1023 patients were enrolled in this study. Of these, 224 (19.1%) were pUA stone cases. Among the potential predictors, thresholding 3D segmentation-based VCSD, age, sex, radio-opacity, hypertension, diabetes, and urine pH were identified as independent pUA stone predictors, and VCSD was the most powerful indicator. The pUA stone prediction model showed good discrimination, yielding area under the receiver operating characteristic curve of 0.960 (95% confidence interval (CI): 0.940-0.979, P < 0.001), 0.931 (95% CI: 0.875-0.987, P < 0.001), and 0.938 (95% CI: 0.912-0.965, P < 0.001) in the training, internal validation, and external validation sets, respectively.

CONCLUSIONS

VCSD measured using 3D segmentation was a decisive independent predictive factor for pUA stones. Furthermore, the established prediction model with VCSD can serve as a noninvasive preoperative tool to identify pUA stones.

Development and external validation of a machine learning-based model to classify uric acid stones in patients with kidney stones of Hounsfield units < 800

The correct diagnosis of uric acid (UA) stones has important clinical implications since patients with a high risk of perioperative morbidity may be spared surgical intervention and be offered alkalization therapy. We developed and validated a machine learning (ML)-based model to identify stones on computed tomography (CT) images and simultaneously classify UA stones from non-UA stones. An international, multicenter study was performed on 202 patients who received percutaneous nephrolithotomy for kidney stones with HU < 800. Data from 156 (77.2%) patients were used for model development, while data from 46 (22.8%) patients from a multinational institution were used for external validation. A total of 21,074 kidney and stone contour-annotated CT images were trained with the ResNet-18 Mask R-convolutional neural network algorithm. Finally, this model was concatenated with demographic and clinical data as a fully connected layer for stone classification. Our model was 100% sensitive in detecting kidney stones in each patient, and the delineation of kidney and stone contours was precise within clinically acceptable ranges. The development model provided an accuracy of 99.9%, with 100.0% sensitivity and 98.9% specificity, in distinguishing UA from non-UA stones. On external validation, the model performed with an accuracy of 97.1%, with 89.4% sensitivity and 98.6% specificity. SHAP plots revealed stone density, diabetes mellitus, and urinary pH as the most important features for classification. Our ML-based model accurately identified and delineated kidney stones and classified UA stones from non-UA stones with the highest predictive accuracy reported to date. Our model can be reliably used to select candidates for an earlier-directed alkalization therapy.

A New Parameter for Calcium Oxalate Stones: Impact of Linear Calculus Density on Non-Contrast Computed Tomography

Background and Objectives: Non-contrast computed tomography (NCCT) is widely used to evaluate urolithiasis. The NCCT attenuation, measured in Hounsfield units (HU), has been evaluated to predict stone characteristics. We propose a novel parameter, linear calculus density (LCD), and analyze variables from NCCT imaging to predict calcium oxalate (CaOx) stones, which are common and challenging to fragment. Materials and Methods: We retrospectively reviewed the medical records of patients with urolithiasis between 2014 and 2017. Among those, 790 patients were included. Based on the NCCT pre-treatment, the maximal stone length (MSL), mean stone density (MSD), and stone heterogeneity index (SHI) were obtained. In addition, the variation coefficient of stone density (VCSD = SHI/MSD × 100) and linear calculus density (LCD = VCSD/MSL) were calculated. In accordance with the stone analysis, the patients were divided into two groups (CaOx and non-CaOx groups). The logistic regression model and receiver operating characteristic (ROC) curve were used for predictive modeling. Results: In the CaOx group, the SHI, VCSD, and LCD were more significant than in the non-CaOx group (all p < 0.001). SHI (OR 1.002, 95% CI 1.001-1.004, p < 0.001), VCSD (OR 1.028, 95% CI 1.016-1.041, p < 0.001), and LCD (OR 1.352, 95% CI 1.270-1.444, p < 0.001) were significant independent factors for CaOx stones in the logistic regression models. The areas under the ROC curve for predicting CaOx stones were 0.586 for SHI, 0.66 for VCSD, and 0.739 for LCD, with a cut-point of 2.25. Conclusions: LCD can be a useful new parameter to provide additional information to help discriminate CaOx stones before treatment.

Urinary supersaturation in a Randomized trial among Individuals with Nephrolithiasis comparing Empiric versus selective therapy (URINE): design and rationale of a clinical trial

Clinical guidelines disagree on whether the identification of abnormal urine chemistries should occur before starting diet and medication interventions to prevent the recurrence of kidney stone events. We describe the rationale and design of the Urinary supersaturation in a Randomized trial among Individuals with Nephrolithiasis comparing Empiric versus selective therapy (URINE) study, a randomized trial comparing two multi-component interventions to improve urinary supersaturation. Participants are randomized (1:1 ratio) to the empiric or selective arm. The target sample size is 56 participants. Adults ≥ 18 years of age with idiopathic calcium stone disease and two symptomatic stone events within the previous 5 years. Exclusion criteria include systemic conditions predisposing to kidney stones and pharmacologic treatment for stone prevention at baseline. Participants in the empiric arm receive standard diet therapy recommendations, thiazide, and potassium citrate. Participants in the selective arm receive tailored diet and nutrient recommendations and medications based on baseline and 1-month follow-up of 24-h urine testing results. The primary endpoints are urinary supersaturations of calcium oxalate and calcium phosphate at 2 months of follow-up. Secondary endpoints include side effects, diet and medication adherence, and changes in 24-h urine volume, calcium, oxalate, citrate, and pH. Short-term changes in urinary supersaturation may not reflect changes in future risk of stone events. The URINE study will provide foundational data to compare the effectiveness of two prevention strategies for kidney stone disease.

Machine Learning Prediction of Kidney Stone Composition Using Electronic Health Record-Derived Features

Objectives: To assess the accuracy of machine learning models in predicting kidney stone composition using variables extracted from the electronic health record (EHR). Materials and Methods: We identified kidney stone patients (n = 1296) with both stone composition and 24-hour (24H) urine testing. We trained machine learning models (XGBoost [XG] and logistic regression [LR]) to predict stone composition using 24H urine data and EHR-derived demographic and comorbidity data. Models predicted either binary (calcium vs noncalcium stone) or multiclass (calcium oxalate, uric acid, hydroxyapatite, or other) stone types. We evaluated performance using area under the receiver operating curve (ROC-AUC) and accuracy and identified predictors for each task. Results: For discriminating binary stone composition, XG outperformed LR with higher accuracy (91% vs 71%) with ROC-AUC of 0.80 for both models. Top predictors used by these models were supersaturations of uric acid and calcium phosphate, and urinary ammonium. For multiclass classification, LR outperformed XG with higher accuracy (0.64 vs 0.56) and ROC-AUC (0.79 vs 0.59), and urine pH had the highest predictive utility. Overall, 24H urine analyte data contributed more to the models' predictions of stone composition than EHR-derived variables. Conclusion: Machine learning models can predict calcium stone composition. LR outperforms XG in multiclass stone classification. Demographic and comorbidity data are predictive of stone composition; however, including 24H urine data improves performance. Further optimization of performance could lead to earlier directed medical therapy for kidney stone patients.

Predictors of Uric Acid Stones: Mean Stone Density, Stone Heterogeneity Index, and Variation Coefficient of Stone Density by Single-Energy Non-Contrast Computed Tomography and Urinary pH

We analyzed the capacities of pertinent parameters (determined by single-energy non-contrast computed tomography [NCCT]) and urinary pH to predict uric acid stones. We reviewed the medical records of 501 patients whose stones were removed surgically or passed spontaneously between December 2014 and April 2016. Qualifying participants (n = 420) were stratified by the nature of the stone (calcium oxalate, uric acid, or infectious). Based on NCCT, we determined maximal stone length (MSL), mean stone density (MSD), and stone heterogeneity index (SHI) using Hounsfield units (HU) and calculated the variant coefficient of stone density (VCSD = SHI/MSD × 100). Urinary pH was also ascertained. Mean patient age was 55.55 ± 15.46 years. MSD (448.59 ± 173.21 HU), SHI (100.81 ± 77.37 HU), and VCSD (22.58 ± 10.55) proved to be significantly lower in uric acid versus other types of stones, as did urinary pH (5.33 ± 0.56; all p < 0.001). Receiver operating characteristic (ROC) curves depicting predictability of uric acid stones yielded area under ROC curve (AUC) values for MSD, SHI, VCSD, and urinary pH of 0.806 (95% CI: 0.761⁻0.850), 0.893 (95% CI: 0.855⁻0.931), 0.782 (95% CI: 0.726⁻0.839), and 0.797 (95% CI: 0.749⁻0.846), respectively, with corresponding cutpoints of 572.3 HU, 140.4 HU, 25.79, and 6.0. Among these four parameters, SHI was verifiably (DeLong's test) the most effective predictor of uric acid stones (all p < 0.001). Compared with MSD, VCSD, and urinary pH, SHI may better predict uric acid stones, using a cutpoint of 140.4 HU.

Impact of dual energy characterization of urinary calculus on management

INTRODUCTION

Dual energy CT (DECT) is a recent technique that is increasingly being used to differentiate between calcium and uric acid urinary tract calculi. The aim of this study is to determine if urinary calculi composition analysis determined by DECT scanning results in a change of patient management.

METHOD

All patients presenting with symptoms of renal colic, who had not previously undergone DECT scanning underwent DECT KUB. DECT data of all patients between September 2013 and July 2015 were reviewed. Urinary calculi composition based on dual energy characterization was cross-matched with patient management and outcome.

RESULTS

A total of 585 DECT KUB were performed. 393/585 (67%) DECT scans revealed urinary tract calculi. After excluding those with isolated bladder or small asymptomatic renal stones, 303 patients were found to have symptomatic stone(s) as an explanation for their presentation. Of these 303 patients, there were 273 (90.1%) calcium calculi, 19 (6.3%) uric acid calculi and 11 (3.4%) mixed calculi. Of those with uric acid calculi, 15 were commenced on dissolution therapy. Twelve of those commenced on dissolution therapy had a successful outcome, avoiding need for surgical intervention (lithotripsy or stone retrieval). Three patients failed dissolution therapy and required operative intervention for definitive management of the stone.

CONCLUSION

Predicting urinary tract calculi composition by DECT plays an important role in identifying patients who may be managed with dissolution therapy. Identification of uric acid stone composition altered management in 15 of 303 (5.0%) patients, and was successful in 12, thereby avoiding surgery and its attendant risks.

The influence of maternal and paternal history on stone composition and clinical course of calcium nephrolithiasis in subjects aged between 15 and 25

Our aim was to compare the influence of maternal history of stones (MHS) and paternal history of stones (PHS) on composition of calculi and disease course in a group of patients with calcium nephrolithiasis (CN) aged between 15 and 25, the age range with the maximal influence of family history on disease expression. One-hundred thirty-five patients (68 F) with CN and one stone-forming parent were retrospectively selected from the database of our outpatient stone clinic, and categorized according to MHS or PHS. Data about stone disease course and composition of passed calculi, determined by chemical analysis or Fourier-transformed infrared spectrophotometry, were collected together with information on blood chemistry and 24-h urinary profile of lithogenic risk. The characteristics of disease course and stone composition were compared using logistic regression tests adjusted for age, sex, and BMI or analysis of covariance where appropriate. Patients with MHS (n = 46) had significantly higher urinary calcium/creatinine ratio and ammonium, a higher prevalence of urological treatments (57 vs 27 %, p < 0.001) and mixed calcium oxalate/calcium phosphate stone composition (69 vs 35 %, p = 0.002) than those with PHS. At multivariate logistic regression models, MHS was independently associated with urological treatments (OR 4.5, 95 %CI 1.9-10.7, p < 0.001) and the formation of calculi with mixed calcium oxalate/calcium phosphate composition (OR 5.8, 95 %CI 1.9-17.9, p = 0.002). The method of stone analysis did not affect this result. In conclusion, in subjects aged 15-25, MHS is associated with mixed calcium stones and with a higher risk for urological procedures, and should be, therefore, considered in the management of urolithiasis.

Biochemical composition of urolithiasis from stone dust - a matched-pair analysis

OBJECTIVE

To determine if the biochemical composition of a renal calculus can be measured from 'dust' obtained during laser fragmentation.

PATIENTS AND METHODS

This pilot study was set in a tertiary referral hospital between 2011 and 2013. Stone dust was aspirated through the ureteroscope during lasering and a stone fragment also retrieved. Both samples were analysed by Fourier transform infrared spectroscopy. Pairs of stone (standard) and dust were compared. They were deemed to match if both were of the same pure biochemical composition or if the predominant constituent was the same in mixed compositions, as this would not alter subsequent management.

RESULTS

Paired specimens were obtained from 97 ureteroscopies. The dust specimen was sufficient for analysis in 66/97 (68%) cases. Of these, the composition matched that of the stone in 49/66 (74%) cases. In 12/66 (18%) the biochemistry differed only in the relative proportions of each constituent, whilst 5/66 (8%) showed a complete mismatch. The overall sensitivity was 51% and specificity 97%. A limitation of the study is the small number of some stone types analysed (<5 each cystine, atazanavir, mixed uric acid/calcium oxalate).

CONCLUSION

We have demonstrated in this pilot study successful proof of principle. Further work is required initially to improve the number of sufficient dust specimens. This technique may offer an option when a stone cannot be retrieved ureteroscopically.

Protein expression of urate transporters in renal tissue of patients with uric acid nephrolithiasis

URAT1 and GLUT9 are two primary urate transporters involved in the renal urate handling. Renal urate underexcretion was reported in uric acid stone formers (UASF) in previous clinical studies. The aim of this study was to investigate the clinical features and possible impact of protein expression of URAT1 and GLUT9 in renal tissues of patients with uric acid (UA) nephrolithiasis. 23 UASF, 27 patients with calcium oxalate (CaOx) stones, and 22 normal controls were enrolled in this study. Clinical data revealed that older age of onset, high plasma UA concentration, low urinary PH, and relative renal urate underexcretion were associated with UASF. By immunohistochemical or western blotting analysis, a significant increase in the relative expression quantity of URAT1 in renal tissue of UASF was found compared to patients with CaOx nephrolithiasis and normal controls. In conclusion, our results suggested that upregulated URAT1 protein expression might contribute to the relative urate underexcretion from the kidney of UASF.

Nomogram to predict uric acid kidney stones based on patient's age, BMI and 24-hour urine profiles: A multicentre validation

INTRODUCTION

We performed a multicentre validation of a nomogram to predict uric acid kidney stones in two populations.

METHODS

We reviewed the kidney stone database of two institutions, searching for patients with kidney stones who had stone composition analysis and 24-hour urine collection from January 2010 to December 2013. A nomogram to predict uric acid kidneys stones based on patient age, body mass index (BMI), and 24-hour urine collection was tested. Receiver-operating curves (ROC) were performed.

RESULTS

We identified 445 patients, 355 from Cleveland, United States, and 90 from Sao Paulo, Brazil. Uric acid stone formers were 7.9% and 8.9%, respectively. Uric acid patients had a significantly higher age and BMI, as well as significant lower urinary calcium than calcium stone formers in both populations. Uric acid had significantly higher total points when scored according to the nomogram. ROC curves showed an area under the curve of 0.8 for Cleveland and 0.92 for Sao Paulo. The cutoff value that provided the highest sensitivity and specificity was 179 points and 192 for Cleveland and Sao Paulo, respectively. Using 180 points as a cutoff provided a sensitivity and specificity of 87.5% and 68% for Cleveland, and 100% and 42% for Sao Paulo. Higher cutoffs were associated with higher specificity. The main limitation of this study is that only patients from high volume hospitals with uric acid or calcium stones were included.

CONCLUSION

Predicting uric acid kidneys stone based on a nomogram, which includes only demographic data and 24-hour urine parameters, is feasible with a high degree of accuracy.

Metabolic evaluation of urinary lithiasis: what urologists should know and do

INTRODUCTION

Urolithiasis is a complex medical entity and regroups several different types of stones, each caused by a multitude of dietary imbalances or metabolic anomalies. In order to better assess the stone-forming patient, urologists should be competent in performing a thorough metabolic work-up.

MATERIALS AND METHODS

We reviewed the litterature in order to provide an appropriate overview of the various components of the metabolic evaluation, including stone analysis, biochemistry tests, and urine collection.

CONCLUSION

Performing a metabolic evaluation allows precise intervention in order to treat and mainly prevent stone disease.