Ultrasound Based Renal Parenchymal Area and Kidney Function Decline in Infants with Congenital Anomalies of the Kidney and Urinary Tract

Abstract

Background:

Congenital anomalies of the kidney and urinary tract (CAKUT) are the leading cause of chronic kidney disease (CKD) in children. Non-invasive imaging biomarkers that predict CKD progression in early infancy are needed.

Methods:

We performed a pilot study nested in the prospective Chronic Kidney Disease in Children cohort (CKiD) study to determine the association between renal parenchymal area (RPA) on first post-natal renal ultrasound and change in eGFR in children with CAKUT.

Results:

Among 14 participants, 78.6% were males, median age at time of ultrasound was 3.4 months (IQR 1.3, 7.9) and median total RPA z-score at baseline was −1.01 (IQR −2.39, 0.52). After a median follow up of 7.4 years (IQR 6.8, 8.2), eGFR decreased from a median of 49.4 mL/min/1.73m² at baseline to 29.4 mL/min/1.73m², an annual eGFR percent decline of −4.68%. Lower RPA z-scores were weakly correlated with higher annual decline in eGFR (Spearman correlation 0.35; 95%CI: −0.25, 0.76).

Conclusion:

This pilot study demonstrated feasibility of obtaining RPA from routine ultrasound and suggested that lower baseline RPA may be associated with a greater decline of eGFR over time. Further studies with larger patient cohorts are needed to confirm this association.

Introduction

Congenital anomalies of the kidney and urinary tract (CAKUT) are the leading cause for development of chronic kidney disease (CKD) in children worldwide (1–3). CAKUT is a group of disorders characterized by a wide range of anatomical aberrations and defects in the kidneys and the urinary tract. There are several mechanisms by which CAKUT causes CKD, which depend on the severity and type of the anomaly (4, 5). Nephron mass has been postulated as a key factor of CKD progression in renal agenesis as well as indicator of kidney health (6, 7). Discovering factors that predict CKD progression for children with CAKUT is important in order to identify those individuals at greatest risk of kidney function decline who could benefit from early interventions to reduce further deterioration of the kidney function and development of end stage kidney disease (ESKD) and its associated comorbidities (8, 9). Such biomarkers could also identify those individuals at lowest risk of CKD progression where non urgent interventions could be deferred.

A lower number of functioning nephrons at birth may increase the risk of future CKD progression for children with CAKUT due to the greater impact that progressive loss of nephrons would have on kidney function (10). Renal parenchymal area (RPA) is a measurement of the functional nephron mass and is obtained by subtracting the collecting system from the gross renal area in maximal length using conventional ultrasonography. Given that RPA subtracts the collecting system, which is often dilated in children with CAKUT, it has been postulated as a measurement of the functioning kidney tissue (11). Assessment of RPA has successfully identified high grade vesicoureteral reflux associated with or without urinary tract infections and ureteropelvic junction obstruction with worsening hydronephrosis (12–14). In addition, lower RPA has been associated with a higher risk of ESKD before 18 years of age among boys with posterior urethral valves (11). In this study, we explore RPA as an early biomarker (available immediately after birth) of CKD progression in children with CAKUT. We hypothesize that lower RPA will be associated with a greater decline of estimated glomerular filtration rate (eGFR) over time.

Materials and Methods

Study Design and Population:

This is a pilot study nested within the multicenter, prospective, observational Chronic Kidney Disease in Children (CKiD) cohort study (15). CKiD has enrolled 1095 children from 54 sites in North America over 15 years (15, 16). CKiD participants (median age of 11 years) have an eGFR of 30–90 ml/min/1.73m² at baseline with a median follow up of 5 years (15). After obtaining Institutional Review Board approval for this ancillary study, participants with CAKUT were identified from the CKiD database who were enrolled from Children’s Hospital of Philadelphia and Children’s Mercy Kansas City between August 2005 and October 2013. Inclusion criteria, in addition to a CAKUT diagnosis, consisted of having had a renal ultrasound within the first 10 months of life with images available for review, as well as an eGFR measured at the time of enrollment into CKiD and at the most recent follow-up site visit. The CONSORT diagram of participants is shown in Figure 1.

Figure 1:

Flow chart showing subject selection

Exposure:

RPA was measured by manually tracing the kidney contour and subtracting the area of the collecting system for those patients who had hydronephrosis using Image J (17), a Java based free NIH-developed software, Figure 2. RPA was measured on the ultrasound images that showed the kidney in the greatest longitudinal diameter. Image J was calibrated to the scale of the ultrasound before outlining the areas of interest. Measurements were obtained from four different readers that were blinded to each other and to any identifiable, clinical, or laboratory records relevant to the participants. Each reader obtained measurements of RPA using the previously described technique (13). Total RPA was defined as the sum of the RPA from both kidneys if present or from single kidney in the case of a solitary kidney. Final total RPA value was obtained by calculating the average total RPA measured from the four readers. Normalized RPA curves using the LMS method were used to calculate a z-score for the total RPA for each participant from previously published normative data (18, 19).

Figure 2:

Using ImageJ software we manually outlined the contour of the gross renal area (a). Renal collecting system (b) was then subtracted from the gross renal area to measure the RPA

Outcome:

The primary outcome was the annual percent change in eGFR for each participant. We used the creatinine-cystatin C based 2012 CKiD equation, which has been described previously (16, 20). CKiD participants have yearly follow-up visits. For this study, eGFR data was collected at baseline through the most recent follow-up site visit at the time of data retrieval in February 2020. Each participant’s annual percent change in eGFR was estimated by regressing log-transformed eGFR on time from baseline. The slope (β₁) from each individual’s regression model was used to calculate the annual percent change in eGFR given by (exp(β₁) – 1) × 100%. For participants who initiated renal replacement therapy (RRT) during the course of the study, eGFR data prior to initiation of RRT was used to characterize eGFR decline.

Patient characteristics:

Participant’s age at ultrasound was collected. Demographics including sex and race as well as medical and birth history including birth weight and gestational age were obtained from the baseline visit in the CKiD database. Growth variables including height, weight, body mass index (BMI), and body surface area (BSA), as well as clinical and laboratory renal parameters such as blood pressure (BP), serum creatinine, urine protein/creatinine ratio were prospectively recorded at baseline and follow-up visits. Details of CKiD study procedures have previously been described (16). BP classification was determined by 2017 Guidelines for Childhood Hypertension (21).

Statistical analysis:

Linear regression was used to quantify the association between the exposure of interest, total RPA z-score and the outcome of interest, eGFR annual percent change. We also calculated the Spearman correlation coefficient to describe the strength of association given the small sample size of this pilot study. In this non-parametric approach, both eGFR annual percent change and total RPA z-score were ranked, and these rank values were used as paired measurements in the correlation calculation. Statistical significance was determined using a p-value < 0.05, although the primary purpose of this pilot study was to determine effect size and direction of the association between RPA and change in eGFR.

Results

At data retrieval, 35 patients with CAKUT were identified in the CKiD database: 9 were enrolled at Children’s Hospital of Philadelphia and 26 were enrolled at Children’s Mercy Kansas City (Figure 1). Of the 35 CAKUT participants, 14 (37%) met inclusion criteria. 21 children were excluded as they did not have ultrasound images available for review. The predominantly (78.6%) male cohort with 30.8% prematurity (median gestational age of 33 weeks) had a median total RPA z-score of −1.01 (IQR −22.39, 0.52) at first post-natal ultrasound (Table 1).

Table 1:

Baseline characteristics of 14 CKiD participants with RPA measured within 10 months of age

Demographics
Age, yearsa	9 [7, 14]
Time from ultrasound, yearsa	8.2 [6.4, 13.3]
Male sex, n (%)	11 (78.6)
Black race, n (%)	2 (14.3)
Growth and Development, median [IQR]
Height, cm	125 [112, 158]
Height z-score	−0.93 [−1.44, −0.07]
Weight, kg	23.1 [19.1, 49.8]
BMI, kg/m2	16.2 [15.4, 18.3]
BSA, m2	0.89 [0.77, 1.49]
Birth History
Gestational age (GA), weeksa	38 [35, 40]
Premature, GA <36 weeks, n (%)	4 (30.8)
Birth weight, gramsa	2920 [2693, 3374]
Low birth weight <2500 g, n (%)	2 (15.4)
Primary Diagnosis, n (%)
Aplastic/hypoplastic/dysplastic kidneys	4 (28.6)
Congenital Urologic Disease (Bilateral Hydronephrosis)	2 (14.3)
Obstructive uropathy	2 (14.3)
Reflux nephropathy	2 (14.3)
Syndrome of agenesis of abdominal musculature	2 (14.3)
Medullary cystic disease/Juvenile nephronophthisis	1 (7.1)
Perinatal Asphyxia	1 (7.1)
Ultrasound Variables, median [IQR]
Age at ultrasound, months	3.41 [1.33, 7.93]
Total RPA, cm2	16.55 [13.84, 19.90]
Total RPA z-score	−1.01 [−2.39, 0.52]

^aResults reported as median [interquartile range] for continuous variables and n (%) for categorical variables. Missing include gestational age: n=1; birth weight: n=1.

The primary outcome of the study represented by the annual percent change in eGFR was found to be −4.68%. eGFR declined from a baseline value of 49.4 mL/min/1.73m² to 29.4 mL/min/1.73m² over a median time of 7.4 years. Figure 3 describes the linear regression results between total RPA z-scores and the annual percent change in eGFR for the 13 participants that had ultrasound images that were analyzable (1 participant was excluded from analysis as incomplete ultrasound images were available). An overall line of best fit was represented by the equation eGFR annual % change = 1.22 × Total RPA z-score – 4.16. A slight positive correlation between RPA and eGFR change over time was revealed by the slope 1.22 (95%CI: −0.86, 3.30; p=0.22). This result suggests that lower total RPA z-scores were associated with faster decline in eGFR. Specifically, the nonparametric Spearman rank correlation between total RPA z-score and eGFR percent change is 0.35 (95%CI: −0.25, 0.76; p-value = 0.27). Urine protein and BP also deteriorated over the course of the study (Table 2).

Figure 3:

Relationship of estimated GFR annual percent change and RPA z-score measured on ultrasound of 13 participants. The baseline estimated GFR stages are depicted by color: blue shows baseline eGFR ≥ 60 ml/mi/1.73m2 and red depicts baseline eGFR < 60 ml/mi/1.73m2. eGFR annual %change = 1.22 × Total RPA z-score – 4.16, slope 1.22 (95%CI: −0.86, 3.30; p=0.22)

Table 2:

Clinical and laboratory characteristics at baseline and follow-up visits of the 14 CKiD participants with renal parenchymal area measured within 10 months of age

Clinical variablesa	Baseline visit	Most recent visit
eGFR, ml/min/1.73m2	49.4 [47.4, 67.3]	29.4 [22.1, 63.8]
Time from baseline to most recent, years	n/a	7.4 [6.8, 8.2]
eGFR annual % change	n/a	−4.68 [−6.59, −1.04]
Serum creatinine, mg/dL	1.02 [0.80, 1.23]	2.96 [1.22, 4.03]
Urine protein/creatinine	0.37 [0.12, 0.88]	1.19 [0.70, 1.81]
Systolic BP percentile	79 [59, 89]	34 [13, 81] (n=9*)
Diastolic BP percentile	80 [54, 98]	46 [33, 96] (n=9*)
BP stage, n (%)
Normal	8 (57.1)	7 (50.0)
Elevated	1 (7.1)	1 (7.1)
Hypertension	5 (35.7)	6 (42.9)

^aResults reported as median [interquartile range] for continuous variables and n (%) for categorical variables. Missing include urine protein/creatinine: n=1.

^*9 participants were <18 at follow-up visit and therefore BP percentiles obtained. eGFR based on 2012 creatinine- cystatin C CKiD equation.

Discussion

In this pilot study, we demonstrated the feasibility of obtaining RPA from archived images from the CKiD database. We also demonstrated that RPA z-scores and annual percent change in eGFR showed a slightly positive correlation such that participants with lower total RPA z-scores show a trend of greater annual decline in eGFR. The concept of this plausible relationship sets the stage for future larger cohort studies to further investigate this relationship.

Current kidney biomarkers include urine and plasma measurements, which can be hard to interpret in infants given ongoing kidney development. eGFR tends to be lower in the first two years of life regardless of patient’s gender, weight, and height (26). This phenomenon is explained by the transplacental transfer of maternal serum creatinine during pregnancy (reflective of high serum creatinine at birth), in addition to ongoing kidney maturation associated with low renal clearance, and increase in the muscle mass of the rapidly growing infants (26, 27). At the same time, even though baseline proteinuria and systolic BP have been identified as independent risk factors for CKD progression in children with CAKUT, as we noted on our pilot cohort, not all infants with CAKUT present with significant proteinuria or elevated BP. Moreover, an accurate BP can also be challenging to obtain in this age group because of poor tolerance of the measurement procedure (28). This highlights the importance of an imaging biomarker such as RPA, that not only represents the functioning renal reserve at a single point in time, but a measurement that can be monitored overtime without added risk of radiation exposure or the invasiveness of a blood draw and provide objective information to caregivers and health care providers.

RPA represents a potential anatomical marker that is readily available immediately after birth. Prior studies show that RPA correlates with kidney volume determined by MRI, even in the presence of hydronephrosis (where the renal collecting system is markedly dilated) (11, 13, 18, 29). Therefore, as RPA reflects the functional renal tissue, lower RPA could potentially predict a greater decline of eGFR over time and identify patients at greatest risk of CKD progression. This is the first investigation of RPA using prospectively collected data and is consistent with previously data reported in children with posterior urethral valves where RPA < 12.4cm² was associated with tenfold increased risk of developing ESKD in childhood (11). By identifying such a risk factor, one could consider the early institution of therapeutic interventions (such as institution of clean intermittent catheterization) designed to prevent or at lease slow further deterioration of eGFR. Given non-invasive nature of ultrasound and the findings of this study, we propose that RPA is worthy to be further investigated as a potential post-natal indicator of disease severity which can aid in more objective prognostication and information for caregivers.

This pilot study has several limitations. First, misclassification may exist. Even though only CAKUT patients were included, not all primary diagnoses were specific and secondary diagnoses were specified. This highlights the reason why no patients with a diagnosis of posterior urethral valves are included in our cohort, which are likely classified as “obstructive uropathy” or “bilateral hydronephrosis”. Despite this, our participant’s z- scores show that nearly all of the 14 RPA measurements of our cohort fall below average of their corresponding ages in the general population, consistent with CAKUT phenotype. Second, this pilot study was not powered to detect a statistically significant association between RPA and change in eGFR. In addition, our sample size could be influenced by outlier data points. However, this pilot study generated an estimate of the magnitude and direction of the association between RPA and eGFR change that could be used to design future larger studies. Conceptually, a primary challenge is assessing the association between RPA measured within the first year of life with change in kidney function evaluated at different ages later in life. Age is associated with accelerated disease progression and we also know that individuals also have unique trajectories, modified by a constellation of factors, with RPA potentially being one. A larger study is needed to appropriately adjust for disease duration with the goal of understanding the extent to which early life RPA is associated with kidney disease. An additional limitation is the manual measurement of RPA. Ultrasound images have lower resolution when compared to computed tomography and Magnetic Resonance Imaging, which makes the manual tracing of the kidney contour harder to do with high level of accuracy (29). We used Image J as this is a free NIH-developed software and has been previously used in children with CAKUT in prior RPA studies (11, 18). Nevertheless, given the results suggest a trend of worsening eGFR with a lower RPA in a prospective study, a different semi-automated approach proven to have a faster RPA measurement with higher validity and lower inter-reader variability will be explored in future larger studies (18, 29).

Linking RPA assessments within the first year of life to disease progression at heterogeneous ages at enrollment is a unique study design with key epidemiologic considerations for bias. Perhaps of most concern is “horse-racing bias”(30) in which, under the assumption that RPA is a causal factor for disease progression, investigating GFR at study entry after the disease progression already occurred is “analogous to peeking at the order of horses halfway through a race”(31). Adjusting for age at entry is therefore crucial because in this setting of a congenital condition: failing to account for duration of disease (which is equivalent to age) will potentially yield biased estimates of the relationship between RPA and progression. This is also why it is unwise to adjust for disease severity (GFR) at enrollment (baseline), since the putative effect of RPA on disease severity will have occurred prior to study enrollment. Regression models that characterize both GFR level at entry and subsequent decline will address this. This pilot data was useful to better understand this phenomena and plan accordingly in order to justify sufficient number of participants across the age spectrum at enrollment and design analytic plans for valid estimates.

The imaging research for evaluating CKD progression continues to evolve. While novel imaging modalities such as contrast enhanced ultrasound which uses a non-nephrotoxic microbubbles based contrast agent has shown variable findings in adults with CKD when compared to controls (32, 33), anatomical features from conventional kidney ultrasound have shown great potential for predicting CKD progression (34, 35). With the goal to identify a worldwide accessible tool that is a reproducible and reliable imaging biomarker for children, and based on our recent experience, we now plan to investigate RPA in a larger CKiD cohort, in combination with biological biomarkers, to further evaluate its potential role as a predictor of CKD progression.

Conclusion

This CKiD ancillary study explored RPA as a potential indicator of CKD progression. A possible trend of lower RPA associated with faster eGFR decline was shown, though not strongly. Larger studies at this point are planned to evaluate RPA and other imaging biomarkers for CKD progression.

Abbreviations:

CAKUT

Congenital Anomalies of Kidney and Urinary Tract

CKiD

Chronic Kidney Disease in Children Cohort Study

RPA

Renal Parenchymal Area

CKD

chronic kidney disease

eGFR

estimated glomerular filtration rate

ESKD

end-stage kidney disease

RRT

Renal Replacement Therapy

BP

blood pressure