A machine learning algorithm predicting risk of dilating VUR among infants with hydronephrosis using UTD classification

BACKGROUNDS

Urinary Tract Dilation (UTD) classification has been designed to be a more objective grading system to evaluate antenatal and post-natal UTD. Due to unclear association between UTD classifications to specific anomalies such as vesico-ureteral reflux (VUR), management recommendations tend to be subjective.

OBJECTIVE

We sought to develop a model to reliably predict VUR from early post-natal ultrasound.

STUDY DESIGN

Radiology records from single institution were reviewed to identify infants aged 0-90 days undergoing early ultrasound for antenatal UTD. Medical records were reviewed to confirm diagnosis of VUR. Primary outcome defined as dilating (≥Gr3) VUR. Exclusion criteria include major congenital urologic anomalies (bilateral renal agenesis, horseshoe kidney, cross fused ectopia, exstrophy) as well as patients without VCUG. Data were split into training/testing sets by 4:1 ratio. Machine learning (ML) algorithm hyperparameters were tuned by the validation set.

RESULTS

In total, 280 patients (540 renal units) were included in the study (73 % male). Median (IQR) age at ultrasound was 27 (18-38) days. 66 renal units were found to have ≥ grade 3 VUR. The final model included gender, ureteral dilation, parenchymal appearance, parenchymal thickness, central calyceal dilation. The model predicted VUR with AUC at 0.81(0.73-0.88) on out-of-sample testing data. Model is shown in the figure.

DISCUSSION

We developed a ML model that can predict dilating VUR among patients with hydronephrosis in early ultrasound. The study is limited by the retrospective and single institutional nature of data source. This is one of the first studies demonstrating high performance for future diagnosis prediction in early hydronephrosis cohort.

CONCLUSIONS

By predicting dilating VUR, our predictive model using machine learning algorithm provides promising performance to facilitate individualized management of children with prenatal hydronephrosis, and identify those most likely to benefit from VCUG. This would allow more selective use of this test, increasing the yield while also minimizing overutilization.

Spectrum of congenital anomalies of the kidney and urinary tract (CAKUT) including renal parenchymal malformations during fetal life and the implementation of prenatal exome sequencing (WES)

OBJECTIVES AND BACKGROUND

Congenital malformations of the kidney and urinary tract (CAKUT) have a prevalence of 4-60 in 10,000 livebirths and constitute for 40-50% of all end stage pediatric kidney disease. CAKUT can have a genetic background due to monogenetic inherited disease, such as PKD or ciliopathies. They can also be found in combination with extra-renal findings as part of a syndrome. Upon detection of genitourinary malformations during the fetal anomaly scan the question arises if further genetic testing is required. The purpose of this study was to determine the phenotypic presentation of CAKUT cases and the results of exome analysis (WES).

METHODS

This is a retrospective analysis of 63 fetal cases with a diagnosis of CAKUT or DSD at a single center between August 2018 and December 2022.

RESULTS

A total of 63 cases (5.6%) out of 1123 matched CAKUT phenotypes including renal parenchyma malformations. In 15 out of 63 WES analysis a pathogenic variant was detected (23.8%). In fetuses with isolated CAKUT the rate of detecting a pathogenic variant on exome sequencing was five out of 44 (11.4%). Ten out of 19 fetuses (52.6%) that displayed extra-renal findings in combination with CAKUT were diagnosed with a pathogenic variant.

CONCLUSIONS

WES provides an increase in diagnosing pathogenic variants in cases of prenatally detected CAKUT. Especially in fetuses with extra-renal malformations, WES facilitates a gain in information on the fetal genotype to enhance prenatal counselling and management.

Clinical characteristics of patients with prenatal hydronephrosis in early postnatal period: a single center retrospective study

BACKGROUND

The study aims to investigate the clinical characteristics of early postnatal period in children with prenatal hydronephrosis (HN) in our single center for 8 years.

STUDY DESIGN

The clinical data of 1137 children with prenatal HN from 2012 to 2020 were retrospectively analyzed in our center. Variables of our study mainly included different malformations and urinary tract dilation (UTD) classification, and main outcomes were recurrent hospitalization, urinary tract infection (UTI), jaundice, and surgery.

RESULTS

Among the 1137 children with prenatal HN in our center, 188 cases (16.5%) were followed-up in early postnatal period, and 110 cases (58.5%) were found malformations. The incidence of recurrent hospitalization (29.8%) and UTI (72.5%) were higher in malformation, but the incidence of jaundice (46.2%) was higher in non-malformation(P < 0.001). Furthermore, UTI and jaundice were higher in vesicoureteral reflux (VUR) than those in uretero-pelvic junction obstruction (UPJO) (P < 0.05). Meanwhile, Children with UTD P2 and UTD P3 were prone to recurrent UTI, but UTD P0 was prone to jaundice (P < 0.001). In addition, 30 cases (16.0%) of surgery were all with malformations, and the surgical rates of UTD P2 and UTD P3 were higher than those of UTD P0 and UTD P1 (P < 0.001). Lastly, we concluded that the first follow-up should be less than 7 days, the first assessment should be 2 months, and the follow up should be at least once every 3 months.

CONCLUSION

Children with prenatal HN have been found many malformations in early postnatal period, and with high-grade UTD were more prone to recurrent UTI, even to surgery. So, prenatal HN with malformations and high-grade UTD should be followed up in early postnatal period regularly.

Outcomes of intermediate-risk hydronephrosis in pediatric patients

PURPOSE

Hydronephrosis is a common antenatal diagnosis and is present in approximately 1-5% of pregnancies. The urinary tract dilation (UTD) classification system was introduced in 2014 and stratifies post-natal hydronephrosis risk into three groups: low-risk (P1), intermediate-risk (P2), and high-risk (P3). Recommendations for P3 hydronephrosis have been established, whereas those for P1 and P2 UTD are often left to the discretion of providers with P1 considered low-grade and less concerning significant pathology. Given the obscure nature of P2 hydronephrosis, we sought to determine the natural history and outcomes of pediatric patients with P2 hydronephrosis within a single institution.

MATERIALS AND METHODS

Children <18 years old diagnosed with hydronephrosis between January 2015 and December 2018 were identified by ICD-9 and ICD-10 codes. Patients with P1 hydronephrosis, P3 hydronephrosis, known vesicoureteral reflux, complex anomalies (ex. Posterior urethral valve), neurological impairments, neurogenic bladder secondary to spinal abnormalities, and <6 months of follow-up were excluded. The development of urinary tract infection (UTI; ≥100 000 CFU/mL of bacterial growth, UA > 10 WBCs/hpf with fever >38C), need for surgical intervention (impaired renal function, worsening hydronephrosis, and/or delayed drainage on diuretic renography), and stability of hydronephrosis were collected retrospectively.

RESULTS

Eighty-seven patients [105 renal units (RU)] were included. Twenty-six patients (30%) were female and 61 (70%) were male. Of the male patients, 30 (49%) of them were circumcised. The median age at initial evaluation was 1 month, and the median duration of follow-up was 13 months. Thirty-four (32%) RU had complete resolution, 24 (23%) improved to P1 hydronephrosis, 33 (31%) remained stable, and 14 (13%) progressed to P3 hydronephrosis. The median duration to resolution and improvement was 8.5 months and 5 months, respectively. Eleven (11%) RU required surgical intervention, 10 of which underwent pyeloplasty, with a median duration to intervention of 9 months. Fifty-five patients (63%) received antibiotic prophylaxis (amoxicillin or sulfamethoxazole-trimethoprim) for a median duration of 5 months. Nine patients (10%) developed a UTI, 3 of which were taking antibiotic prophylaxis at the time of infection.

CONCLUSIONS

Intermediate-risk hydronephrosis diagnosed in the pediatric population will either improve, resolve, or remain stable during 1-year follow-up in 87% of RU. Only 11% of RU required surgical intervention, and 19% of patients developed a UTI in the absence of antibiotic prophylaxis. These findings will assist with counseling parents concerning the importance of follow-up imaging and monitoring for UTI. However, the low risk of surgical intervention is encouraging and should be discussed with the children's caretakers.

Three-dimensional static-fluid MR urography with gradient- and spin-echo (GRASE) at 3.0T: comparison of image quality and diagnostic performance with respiratory-triggered fast spin-echo (FSE)

Purpose

To compare the performance of 3D MRU based on a breath-hold gradient- and spin-echo (BH-GRASE) technique with conventional 3D respiratory-triggered FSE (RT-FSE) sequence in patients with urinary tract dilation.

Methods

We prospectively included 90 patients with urinary tract dilation who underwent both 3D BH-GRASE and RT-FSE MRU at 3T. The acquisition time of two MRU sequences was recorded. Three readers blinded to the protocols reviewed the image quality using a five-point scale and assessed the diagnostic performance related to urinary tract dilation. The relative contrast ratio (CR) between the urinary tract and adjacent area was measured quantitatively.

Results

Acquisition time was 14.8 s for BH-GRASE MRU and 213.6 ± 52.2 s for RT-FSE MRU. The qualitative image analysis demonstrated significant equivalence between the two MRU protocols. 3D BH-GRASE MRU better depicted bilateral renal calyces than RT-FSE MRU (p < 0.05). The CR values of the urinary tract were lower on BH-GRASE MRU compared with RT-FSE MRU (p < 0.05). There were excellent agreements in the assessment of urinary tract dilation between BH-GRASE and RT-FSE MRU, including the dilated degree, obstructive level, and obstructive imaging features (inter-sequence κ = 0.924–1).

Conclusion

3D BH-GRASE MRU significantly decreased the acquisition time and achieved comparable image quality, urinary tract visualization, and diagnostic performance with conventional 3D RT-FSE MRU. Breath-hold 3D MRU with GRASE may provide a feasible evaluation of urinary tract dilation.

Graphical abstract

2021 update on the urinary tract dilation (UTD) classification system: clarifications, review of the literature, and practical suggestions

In 2014, a multidisciplinary consensus on the classification of pre- and postnatal urinary tract dilation (UTD classification) was developed. Its goal was to provide a standardized system for evaluating and reporting urinary tract dilation both in the prenatal and postnatal periods. In this review, we summarize insights learned from the implementation of the UTD classification system since its inception, providing clarifications on common points of confusion. In addition, we review current literature in the clinical validation of the UTD classification system to provide credence for its use in managing fetuses and children with urinary tract dilation.

A Retrospective Analysis to Evaluate Role of the New UTD Classification System in Prenatal Prediction of Severity and Postnatal Outcome in Antenatally Diagnosed Urinary Tract Dilatation Abnormalities

Background/Purpose of the Study

Foetal urinary tract dilation (UTD) abnormalities affect 1-5% of all pregnancies. However, exact incidence is difficult to estimate because of different terminologies used to define the condition and different grading systems to define its severity antenatally as well as postnatally worldwide. In order to overcome this problem, the new UTD classification system has been introduced in the year 2014 so as to have universal approach for diagnosis and management of UTD globally. Indian data about clinical utility of the UTD classification system and its role in prenatal prediction of severity of renal disease are lacking. The present study aims to investigate clinical utility of new UTD classification system in foetal UTD abnormalities and to evaluate the role of UTD classification system in antenatal prediction/prognostication of severity of UTD abnormalities.

Methods

We conducted a single-centre retrospective study between April 2014 and January 2017, which included 70 infants with antenatally diagnosed UTD delivered in our hospital and managed in our paediatric unit postnatally. Pre- and postnatal ultrasound findings were noted, and UTD-A and UTD-P classification were applied retrospectively in all cases as per criteria defined in the new UTD classification. Postnatal outcome in all cases was evaluated in terms of need for immediate postnatal urosurgical intervention, presence of persistent UTD pathology and severity of renal impairment in relation to their pre- and postnatal UTD A and P risk categories.

Results

None from UTD A1 risk group in the last prenatal scan showed significant postnatal UTD abnormality. In contrast to this, UTD A2-3 risk group in the last prenatal scan had persistent postnatal UTD pathology in 70% cases. All infants with abnormal postnatal UTD diagnosis were identified prenatally as UTD A2-3 (high risk). Nine infants (12.8%, n = 70) who needed urosurgical intervention postnatally were categorised as UTD A2-3 prenatally and UTD P3 postnatally.

Conclusion

We found increased frequency of complications and urosurgical interventions in all infants with antenatal UTD A2-3 grades in the last prenatal scan in comparison with those with UTD A1 grades who showed complete resolution (100%) postnatally. Antenatal UTD classification may be useful in antenatal prediction and prognostication of postnatal severity, especially in high-risk cases (i.e. UTD A2-3).

Rater reliability of postnatal urinary tract dilation consensus classification

BACKGROUND

A multidisciplinary urinary tract dilation (UTD) classification system was published in 2014 to standardize definitions and renal/bladder ultrasound image interpretation.

OBJECTIVE

To evaluate intra- and inter-rater reliability of this system on postnatal RBUS.

MATERIALS AND METHODS

Renal/bladder US of 60 infants (<12 months) with urinary tract dilation were anonymized, retrospectively reviewed and scored twice using the UTD classification system by a pediatric urologist and four pediatric radiologists. Exams included supine and prone images of each kidney. Raters recorded the anterior posterior renal pelvis diameter in each position; and when present calyceal dilation (central and peripheral), ureteral dilation, parenchymal and bladder abnormalities. A UTD score was given to each kidney based on these components.

RESULTS

Intraclass correlation (ICC) of anterior posterior renal pelvis diameter measures was 0.99 (P<0.001). Intra-rater reliability for the anterior posterior renal pelvis diameter of each kidney was high, with ICC >0.95 (P<0.001). Inter-rater kappa values for UTD scores of both kidneys ranged from 0.60 to 0.77 (P <0.001). Intra-rater kappa values for UTD scores of both kidneys ranged from 0.74 to 0.92 (P <0.001). Of the six categories comprising the UTD score, discrepancy between raters was highest for interpretation of central and peripheral calyceal dilation.

CONCLUSION

Present inter- and intra-rater reliability findings were similar to those previously reported for grading systems for urinary tract dilation. Across these studies, kappa values are generally lower than the 0.8 cut-off advocated for medical measures. Here, calyceal dilation commonly determined the kidney UTD score and was also the source of greatest discrepancy between raters. Improving consistency of calyceal dilation interpretation may improve UTD consensus score reliability.

A practice pattern assessment of members of the Society of Pediatric Urology for evaluation and treatment of urinary tract dilation

INTRODUCTION

Over the last decade the literature, including a multidisciplinary consensus statement, has supported a paradigm shift in management of urinary tract dilation, yet the impact on practice patterns has not been well documented.

OBJECTIVE

This study aims to elucidate specific practice patterns for treatment of prenatal unilateral urinary tract dilation and to assess surgical intervention patterns for ureteropelvic junction obstruction.

STUDY DESIGN

An online survey was distributed to 234 pediatric urologists through the Society of Pediatric Urology. The survey was composed of five clinical case scenarios addressing evaluation and management of unilateral urinary tract dilation.

RESULTS

The response rate was 71% (n = 168). Circumcision status, gender, and grade were significant factors in recommending prophylactic antibiotics for newborn urinary tract dilation. Prophylactic antibiotic use in the uncircumcised male and female was twice that of a circumcised male for grade 3 (Table). This difference was minimized for grade 4. Use of VCUG was high for circumcised males with grade 3 or 4 (Table). The choice of minimally invasive surgery for ureteropelvic junction repair increased with age from 19% for a 5-month-old, 49% for a 2-year-old, and 85% for a 10-year-old. Notably, 44% of respondents would observe a 10-year-old with intermittent obstruction. Retrograde pyelography was recommended in conjunction with repair in 65% of respondents. Antegrade stent placement was the most common choice (38-47%) for urinary diversion after pyeloplasty. Regarding postoperative imaging, only 5% opted for routine renal scan whereas most would perform renal ultrasound alone.

DISCUSSION

Practice patterns seen for use of prophylactic antibiotics are in agreement with the literature, which promotes selective use in those at highest risk for urinary tract infections. Interestingly, use of aggressive screening was not concordant with this literature. Several studies have indicated an increased usage of robotic pyeloplasty; however, results indicate that minimally invasive surgery is not preferred in those younger than 6 months. Study limitations include use of clinical case scenarios as opposed to actual clinical practice.

CONCLUSION

Practice patterns for prophylactic antibiotic use for neonatal urinary tract dilation are dependent on gender, circumcision status, and grade. The use of minimally invasive surgery for ureteropelvic junction repair increased with patient age, with 50% preferring this modality at 2 years.

The antenatal urinary tract dilation classification system accurately predicts severity of kidney and urinary tract abnormalities

BACKGROUND

Urinary tract dilation (UTD) is a commonly diagnosed prenatal condition; however, it is currently unknown which features lead to benign and resolving or pathologic abnormalities. A consensus UTD classification system (antenatal UTD classification, UTD-A) was created by Nguyen et al. in 2014 [1], but has not yet been validated.

OBJECTIVE

To evaluate the ability of the UTD-A system to identify kidney and urinary tract (KUT) abnormalities, assess whether UTD-A can predict severity of KUT conditions, and perform a cost analysis of screening ultrasound (US).

METHODS

A retrospective single-center study was conducted at an academic medical center. Inclusion criteria were: neonates in the well or sick nursery who had a complete abdominal or limited renal US performed in the first 30 days of life between January 01, 2011 and December 31, 2013. Data were collected on prenatal US characteristics from which UTD-A classification was retrospectively applied, and postnatal data were collected up to 2 years following birth.

RESULTS

A total of 203 patients were identified. Of the 36 abnormal postnatal KUT diagnoses, 90% were identified prenatally as UTD A1 or UTD A2-3. The remaining 10% developed postnatal KUT abnormalities due to myelomeningocele, such as VUR or UTD, which were not evident prenatally. Overall sensitivity and specificity of the UTD-A system was 0.767 (95% CI 0.577, 0.901) and 0.836 (95% CI 0.758, 0.897), respectively, when resolved UTD was counted as a normal diagnosis. Postnatal diagnoses differed by UTD-A classification as shown in the Summary fig. Of all the obstructive uropathies, 90.9% occurred in the UTD A2-3 class and none occurred in UTD-A Normal. Rate of postnatally resolved UTD was significantly higher in the UTD A1 group (78%) compared with UTD A2-3 (31%) or UTD-A Normal (12%, all P < 0.001). There was a notable trend towards more UT surgeries, UTI, and positive VUR among UTD A2-3 patients, but statistical significance was limited by a small number of patients.

CONCLUSIONS

This study found that the UTD-A classification system revealed important differences in the severity of UTD abnormalities. With repeated validation in larger cohorts, the UTD-A classification may be used to offer a prognosis for parents regarding prenatally diagnosed KUT conditions. Larger prospective studies should be designed to validate whether the UTD-A system can predict postnatal events related to UTD morbidity such as need for UT-related surgery or UTI.

Six of one, half a dozen of the other: A measure of multidisciplinary inter/intra-rater reliability of the society for fetal urology and urinary tract dilation grading systems for hydronephrosis

INTRODUCTION

The urinary tract dilation (UTD) classification system was introduced to standardize terminology in the reporting of hydronephrosis (HN), and bridge a gap between pre- and postnatal classification such as the Society for Fetal Urology (SFU) grading system. Herein we compare the intra/inter-rater reliability of both grading systems.

MATERIALS AND METHODS

SFU (I-IV) and UTD (I-III) grades were independently assigned by 13 raters (9 pediatric urology staff, 2 nephrologists, 2 radiologists), twice, 3 weeks apart, to 50 sagittal postnatal ultrasonographic views of hydronephrotic kidneys. Data regarding ureteral measurements and bladder abnormalities were included to allow proper UTD categorization. Ten images were repeated to assess intra-rater reliability. Krippendorff's alpha coefficient was used to measure overall and by grade intra/inter-rater reliability. Reliability between specialties and training levels were also analyzed.

RESULTS

Overall inter-rater reliability was slightly higher for SFU (α = 0.842, 95% CI 0.812-0.879, in session 1; and α = 0.808, 95% CI 0.775-0.839, in session 2) than for UTD (α = 0.774, 95% CI 0.715-0.827, in session 1; and α = 0.679, 95% CI 0.605-0.750, in session 2). Reliability for intermediate grades (SFU II/III and UTD 2) of HN was poor regardless of the system. Reliabilities for SFU and UTD classifications among Urology, Nephrology, and Radiology, as well as between training levels were not significantly different.

DISCUSSION

Despite the introduction of HN grading systems to standardize the interpretation and reporting of renal ultrasound in infants with HN, none have been proven superior in allowing clinicians to distinguish between "moderate" grades. While this study demonstrated high reliability in distinguishing between "mild" (SFU I/II and UTD 1) and "severe" (SFU IV and UTD 3) grades of HN, the overall reliability between specialties was poor. This is in keeping with a previous report of modest inter-rater reliability of the SFU system. This drawback is likely explained by the subjective interpretation required to assign grades, which can be impacted by experience, image quality, and scanning technique. As shown in the figure, which demonstrates SFU II (a) and SFU III (b), as assigned by a radiologist, it is possible to make an argument that either of these images can be classified into both categories that were observed during the grading sessions of this study.

CONCLUSION

Although both systems have acceptable reliability, the SFU grading system showed higher overall intra/inter-rater reliability regardless of rater specialty than the UTD classification. Inter-rater reliability for SFU grades II/III and UTD 2 was low, highlighting the limitations of both classifications in regards to properly segregating moderate HN grades.

Inter-rater reliability of the APD, SFU and UTD grading systems in fetal sonography and MRI

INTRODUCTION

Antenatal hydronephrosis (ANH) is frequently detected on screening obstetric ultrasonography. Common ANH grading systems include the anterior-posterior diameter (APD) and the Society for Fetal Urology (SFU) grading system. Recent developments in the management of ANH include the use of fetal magnetic resonance imaging (MRI), and a new grading system - Urinary Tract Dilation (UTD). This study reviewed patients who underwent fetal MRI and ultrasound, and compared the grading systems across these imaging modalities.

MATERIALS AND METHODS

Patients who underwent paired fetal MRI and ultrasound studies between January 2012 and January 2014 were included. Two pediatric urologists and a pediatric radiologist reviewed the studies. Data collected included APD, SFU grade, and UTD grade. Fleiss' kappa statistic determined the inter-rater reliability (IRR) of the SFU and UTD grading within each imaging modality. Intra-class correlation assessed the consistency of the APD measurements.

RESULTS

Forty-seven patients and 88 renal units were evaluated. Median gestational age was 22 weeks. Kappa values of the SFU grading system indicated fair IRR for ultrasound imaging and moderate IRR for MRI imaging, while the UTD grading system reached moderate IRR for both. The IRR of the SFU grading system was improved with the use of MRI, while the UTD grading system was no different. The APD intraclass correlation coefficient improved significantly when measured by MRI. As the ultrasound SFU grade increased, the odds of the MRI SFU grade being scored higher increased by a factor of 3.7. There was no difference between ultrasound and MRI when using the UTD grading system.

DISCUSSION

This study was the first to assess the UTD system in a cohort of patients who underwent paired ultrasound and MRI studies. The results suggested that the UTD system might improve IRR, compared with the SFU system. The use of fetal MRI may improve the IRR of the SFU grading system. It also found that the proportion of SFU grades was affected by the imaging modality, raising the possibility that MRI 'overcalls' the SFU grade, compared with ultrasound. This difference was not observed using the UTD grading system. The most important limitation was the selection bias favoring complex pathology with severe ANH diagnosed at an early gestational age.

CONCLUSIONS

In this unique cohort, the UTD system improved IRR when compared to the SFU grading system. Fetal MRI improved the IRR of the SFU grading system, and improved the APD intraclass correlation. The SFU grading was likely to be higher when assessed by MRI vs ultrasound, but the UTD grade was not affected by the imaging modality.