BACKGROUND: Vesicoureteral reflux (VUR) is a common urological disorder in children. The development of breakthrough urinary tract infections (bUTI) lead to important morbidity and often triggers surgical correction. To date, various models have been developed to offer better prognostication for children with VUR, all hampered by subjectivity in VUR grading and severity assessment. Herein, we aimed to enhance bUTI prediction in children with VUR, employing a large multi-institutional dataset that incorporates novel quantitative measures for VUR severity and contrasting it with analyses with traditional grading methods.
METHODS: Four pediatric centers across Canada, USA, and India were queried for clinical information and voiding cystourethrogram (VCUG) data in children with VUR. Eligible participants had primary VUR with >=14 months of follow-up on prophylactic antibiotics and no surgical interventions. Each VCUG was graded for severity with traditional classification (by multiple raters) and by quantitative VUR (qVUR) features including ureteral tortuosity and dilatation. Using multivariable and Cox proportional hazards regression, the risk of bUTI was estimated. Models were trained and externally validated to provide individualized survival curves. Model performance by concordance index and area under the receiver-operator-characteristic (AUROC), calibration, and net benefit was compared between baseline clinical risk, VUR-based, and qVUR-based models.
RESULTS:Overall, 684 children were included (308 training, 376 validation); of which 109 (16%) experienced a bUTI within the study period. Age >12 months (HR 0.64, 95%CI 0.44, 0.93; p=0.02) and male sex (HR 0.60, 95%CI 0.41, 0.89; p=0.01) were associated with decreased risk of bUTI. There was no significant difference with bilateral VUR versus unilateral VUR (HR 0.93, 95%CI 0.64, 1.35; p=0.69). High-grade VUR (grade 4-5) was associated with an increased risk of bUTI (HR 1.48, 95%CI 1.01, 2.15; p=0.04). Ureteral dilation was associated with increased risk of bUTI (HR 1.58-1.73, p=0.01), while ureteral tortuosity was not significantly associated with bUTI risk. Together, the qVUR-based cox regression model performed with a c-index of 0.64, compared to 0.57 for VUR-based models. In children under 1 year of age, the binary AUROC was 0.75 with qVUR-based model. The qVUR-based model stratified patients into low-, moderate- and high-risk groups. The latter were at significantly elevated risk (HR 2.17 95%CI 1.45, 3.23), and this provided significantly improved calibration and net benefit, while appearing to not be biased by age, sex, or laterality (Figure).
CONCLUSIONS:
These findings suggest that personalized bUTI risk is associated with VUR severity, which can be forecasted with modest model performance. qVUR measures remain clinically significant and offers new objective metrics to determine VUR severity over traditional grading. This can improve the detection of new clinical associations in VUR. Our data shows that qVUR offers promising improved risk prognostication and can select patients at increased risk for bUTI who may benefit from a more proactive surgical strategy.
