Genetic determinants of renal scarring in children with febrile UTI

Background Febrile urinary tract infections (UTIs) are among the most severe bacterial infections in infants, in which a subset of patients develops complications. Identifying infants at risk of recurrent infections or kidney damage based on clinical signs is challenging. Previous observations suggest that genetic factors influence UTI outcomes and could serve as predictors of disease severity. In this study, we conducted a nationwide survey of infant genotypes to develop a strategy for infection management based on individual genetic risk. Our aims were to identify genetic susceptibility variants for renal scarring (RS) and genetic host factors predisposing to dilating vesicoureteral reflux (VUR) and recurrent UTIs. Methods To assess genetic susceptibility, we collected and analyzed DNA from blood using exome genotyping. Disease-associated genetic variants were identified through bioinformatics analysis, including allelic frequency tests and odds ratio calculations. Kidney involvement was defined using dimercaptosuccinic acid (DMSA) scintigraphy. Results In this investigation, a cohort comprising 1087 infants presenting with their first episode of febrile UTI was included. Among this cohort, a subset of 137 infants who underwent DMSA scanning was subjected to gene association analysis. Remarkable genetic distinctions were observed between patients with RS and those exhibiting resolved kidney involvement. Notably, the genetic signature indicative of renal scarring prominently featured mitochondrial genes. Conclusions In this nationwide study of genetic susceptibility to RS after febrile UTIs in infancy, we identified a profile dominated by mitochondrial polymorphisms. This profile can serve as a predictor of future complications, including RS and recurrent UTIs. Graphical abstract A higher resolution version of the Graphical abstract is available as Supplementary information Supplementary information The online version contains supplementary material available at 10.1007/s00467-024-06394-6.


Introduction
Urinary tract infection (UTI) is a common bacterial infection in childhood, affecting approximately 2% of all infants [1].Urosepsis and acute pyelonephritis (APN) are the most severe forms of UTI, leading to systemic infections with kidney damage.One potential consequence of APN in childhood is renal scarring (RS).Estimates of the frequency of RS in children vary across populations due to inconsistencies in the timing of acute imaging and follow-up examinations [2].Nevertheless, population-based studies suggest that 10-26% of children with febrile UTI may suffer from RS [3][4][5][6].The presence of renal involvement can be diagnosed early in some patients post-infection, while in others, scarring is only detected during follow-up, implying that the scarring process can be acute or gradual, possibly influenced by recurrences and other predisposing factors [7][8][9][10].
The pathogenesis of RS is subject to debate, as it encompasses parenchymal anomalies, dysplasia (CAKUT), and acquired scarring due to the complex interaction of virulent bacteria and the host's immune response [11,12].While vesicoureteral reflux (VUR) increases the risk of RS, scarring can also develop in the absence of VUR [13].Therefore, VUR must be considered when exploring susceptibility

Patients
Infants under one year of age with a first episode of febrile UTI were eligible for inclusion.Exclusion criteria included overt urogenital malformations or ongoing catheter therapy.Children were enrolled from emergency units when antibiotic treatment was initiated due to a clinical diagnosis of suspected UTI.All participants received treatment according to local practice, in line with the Swedish guidelines for managing febrile UTI in children under 2 years of age [24].Diagnostic criteria for febrile UTI were a temperature exceeding ≥ 38.0°C and a positive urine culture indicating the presence of a single uropathogenic bacterium (any growth in urine obtained through suprapubic aspiration, ≥ 10 4 cfu/mL in catheterized urine, or ≥ 10 5 cfu/mL in mid-stream clean-catch urine).DNA samples were collected upon inclusion.Data regarding clinical and laboratory findings were recorded at each of the 29 collaborating centers and entered into a database at the University of Gothenburg at two time points: after the initial evaluation and one year following the infection.The results of follow-up examinations, prophylactic antibiotic treatment, surgical interventions, and new episodes of febrile UTI were reported.

Imaging
Infants diagnosed with febrile UTI underwent ultrasound examinations (n = 1081) to identify urinary tract dilatation and rule out significant malformations.DMSA scintigrams were conducted to diagnose kidney involvement and were analyzed locally in a selected group of children with one or more risk factors, such as CRP ≥ 70 mg/L, creatinine > 30 μmol/l, or non-E.coli growth in urine culture.Decreased kidney uptake in examinations conducted after the acute infection, regardless of the time interval from the infection, was considered as kidney involvement (first DMSA positive).Acute pyelonephritis (APN) was defined as decreased kidney uptake in examinations performed within seven days from diagnosis.In children with kidney involvement on the first DMSA scan, RS was diagnosed through a second DMSA scan conducted ≥ 6 months after the infection, based on persistent kidney involvement or reduced differential function < 45%.If the first DMSA scan was normal, and the infants did not experience any new episodes of febrile UTI, they were considered to have normal kidneys during the follow-up period [3].Voiding cystourethrography (VCUG) was performed in selected patients, primarily in those with dilatation on kidney ultrasound or reduced differential function on DMSA.Vesicourethral reflux (VUR) was graded 1-5 following the recommendations of the International Reflux Study in Children [25].The maximum VUR grade was reported for children who underwent more than one VCUG examination and the highest VUR grade for those with bilateral VUR.VUR grades 0-2 are defined as no VUR or non-dilating VUR, while VUR grades 3-5 are defined as dilating VUR.Individuals diagnosed with kidney involvement at the first DMSA scan, who also had follow-up examination, were included in calculations of RS in comparison to VUR grade and UTI recurrences.

Laboratory parameters and exome genotyping
CRP, creatinine, urine dipstick, and urine culture were included in the initial evaluation, as stipulated in the Swedish national guidelines.Blood samples were collected for a genetic evaluation of RS in children with UTI.DNA was extracted from heparinized peripheral blood using the QIAamp DNA Blood mini kit, and samples were analyzed via exome genotyping using Illumina Infinium Exome bead chip technology.

Statistical and bioinformatics analysis
SPSS software version 27 was used to analyze the clinical data.Patient criteria were analyzed by Fisher's exact test for group comparisons with two dichotomous variables, the χ 2 test for ordered categorical variables and for large sample sizes an independent t test for continuous variables and a non-parametric Mann-Whitney's test for skewed or smaller sample sizes.The post hoc Kruskal-Wallis test was used to compare more than two independent groups regarding continuous or ordered variables.p < 0.05 was considered significant.Significant values (p values of overlap) were calculated by Fisher's exact test, and p values were adjusted for multiple testing.
The primary outcome was the genetic association with RS following acute febrile UTI.DNA association analysis was performed using allelic frequency tests and odds ratio (OR) calculations to compare patient groups based on the second DMSA outcome: (1) RS compared to resolved APN and (2) RS compared to first DMSA negative (i.e., no APN).For the secondary outcomes, we also compared (3) VUR grade 0-2 compared to VUR grade 3-5 and (4) recurrent UTI compared to no recurrences for the DNA association analysis.
For all association analyses, OR values and p values were calculated for each single nucleotide polymorphism (SNP) using Fisher's exact test using the fisher.testfunction in R v.4.1.3[26].Variants with p values < 0.01 were considered significant.To summarize the association analysis results on the gene level for each gene, the SNP with the lowest p value was selected as the representative SNP (< 0.005).The genes with representative SNPs were selected to be visualized with heatmaps.The heat maps were generated in R using the pheatmap v.1.0.12R-package [27].The sample ordering for the heat maps was done based on DMSA status and columns on mean log2 ORs for DMSA-positive group.Furthermore, a principal component analysis (PCA) plot, using the prcomp function in R, was used to show clusters of samples based on their similarity.Variants located in pseudoautosomal regions (PARs) of the XY chromosome were included, while non-PARs were omitted to avoid gender biases in the analysis due to hemizygous male genotypes.SNPs were annotated with chromosomal location, gene name, reference allele/alternative allele, and gene type using Variant Effect Predictor (VEP) [28].To estimate the prevalence of the SNPs in the Swedish population, the allele frequencies were retrieved from the SweGen database [29].
Pathway analysis was performed using IPA (QIAGEN Inc., https:// www.qiage nbioi nform atics.com/ produ cts/ ingen uity-pathw ay-analy sis) of SNPs and included differences in gene associations with p values (p < 0.005).Genes from the IPA Knowledge database were used as a background reference, and the results analyzed by IPA were presented in terms of canonical pathways, major regulators, upstream regulators, disease, and toxicity functions.

Patient characteristics
Among the 1087 infants diagnosed with febrile UTI, an initial DMSA scan was administered to 624 infants (Figure 1A), and subsequently, a DNA association analysis was carried out on a subset of 137 infants.Table 1 displays the age, gender distribution, maximum temperature, and maximum CRP levels for infants not subjected to DMSA scan (group I), infants with a first DMSA examination but not included in genetic evaluation (group II), and infants examined by a first DMSA and included in genetic evaluation (group III).Infants included in DNA evaluation were representative for children subjected to a first DMSA scan.

Kidney involvement in first DMSA and renal scarring at follow-up DMSA scan
A total of 201 out of 623 patients (32.3%) showed signs of kidney involvement.Among those who underwent the first DMSA scan within seven days, 57 out of 90 (63.3%) were diagnosed with kidney involvement and classified as having APN.Furthermore, RS was identified in 63 out of 129 (43.2%) patients who underwent a second DMSA scan ≥ 6 months from infection (see Figure 1A).

Gene association in relation to renal scarring
Out of 90 children who underwent their first DMSA scan within 7 days of infection, exome genotyping was performed on 56 patients (32 with positive and 24 with negative DMSA results) (Supplementary Figure 1A).We compared significant SNPs (with a cutoff of p < 0.005) and calculated individual OR values for each variant in the population, then sorted them into heat maps.The first analysis focused on SNPs associated with RS (n = 7) in comparison to those associated with resolved APN (n = 25) (Figure 2A).This analysis revealed a distinct separation of variants, with alternative alleles being more prevalent in the scarring group.Group distinctions were further demonstrated in a PCA plot (Figure 2B).In a second analysis of significant SNPs associated with RS (n = 7) and the no-APN group (n = 24) (with the same p < 0.005 cutoff), we also observed a clear genetic differentiation between the two groups (Supplementary Figure 1B).Table 1 The clinical characteristics of the core population of infants diagnosed with febrile urinary tract infection (UTI) were studied, with a total of 1087 subjects (N=1087).Three groups were evaluated based on the selection for DMSA examination and DNA analysis

Association of mitochondrial genes found in renal scarring
Gene association analysis identified 582 SNPs significantly more common in RS, with the 25 most significant ones listed in Table 2.Among these, 22 SNPs are linked to mitochondrial function, including ND-4-6, ND4L, COX I-III, ATPase6, and ATPase8, all playing roles in mitochondrial respiratory complexes.
These significant SNPs from DNA association analysis comparing children with RS to children with resolved APN were further analyzed using IPA (Table 3).Canonical pathway analysis revealed ten pathways associated with RS (cutoff p < 0.005 and adjusted p value < 0.05).IPA comparing RS to resolved APN highlighted those genes causing mitochondrial dysfunction and oxidative phosphorylation showed the strongest association with scarring (cutoff p < 0.005 and adjusted p value < 0.05).This was followed by chondroitin sulfate biosynthesis, dermatan sulfate synthesis, and FGF signaling pathways (Supplementary Table 1).Among the most significant upstream regulators, IPA of RS compared to resolved APN groups (p = 0.005, adjusted p value < 0.05) confirmed the dominance of mitochondrial genes (Supplementary Table 2).
Further examination of these mitochondrial genes revealed a predominant association with various forms of cancer rather than inflammatory conditions (Supplementary Table 3).

Involvement of mitochondrial genes distinguishes infants with renal scarring
To investigate whether infants with RS can be genetically distinguished from children with a first negative DMSA, we conducted another DNA comparison.We identified 524 polymorphic genes associated with RS (p < 0.005).The strongest association was found in the ubiquitously expressed SH3BP4, which is considered to function as a transferrin receptor involved in cell growth regulation, proliferation, and autophagy [30].Additionally, 22 polymorphic SNPs were of mitochondrial origin, similar to the genes found in the RS analysis above (see Table 2B).Using IPA, we identified 12 canonical pathways associated with RS (p < 0.005 adjusted to p < 0.05), with the most significant pathways related to mitochondrial dysfunction and oxidative phosphorylation (see Supplementary Table 4).The dominance of mitochondrial involvement was further confirmed by a combined analysis of common polymorphic genes among infants with RS compared to infants with resolved APN and infants with a first negative DMSA.When comparing the 150 most significant SNPs (p < 0.005) in these two groups, we found that 22 out of 50 shared genes were involved in mitochondrial function (see Table 4).

Gene association in relation to VUR
In this analysis, we compared 54 infants with VUR grades 0-2 to 28 infants with VUR grades 3-5.Infants included in the DNA evaluation were randomly selected and demonstrated a similar distribution of dilating VUR (34.1%) as described above in children subjected to both VCUG and DMSA.As in the RS analysis, we observed a clear distinction between the groups of VUR grades 0-2 and VUR grades 3-5.This separation is visualized in the heatmap and PCA plots (Figure 2C, D).
DNA analysis identified 893 polymorphic SNPs associated with dilating VUR (p < 0.005).A mitochondrial profile was evident in 22 of the 50 most significant genes.Polymorphic genes previously linked to CAKUT, such as GRIP1, were also associated with children with dilating VUR [31].The top 50 genes associated with dilating VUR are listed in Supplementary Table 5. IPA canonical pathway analysis identified mitochondrial dysfunction (Supplementary Table 6), and the upstream regulators were dominated by mitochondrial genes (Supplementary Table 7).

Recurrent UTI
In our study, 136 out of 622 infants investigated by a first DMSA (21.9%) experienced recurrent febrile UTIs.Among the 129 children who underwent a second DMSA, RS was more common in the group with recurrent UTIs compared to children without UTI recurrences (75.0%compared to 37.1%, Figure 1C, p < 0.001).

Gene association in relation to recurrent febrile UTI
In our DNA association analysis comparing infants with reported recurrences of febrile UTI (n = 36) to infants without UTI recurrences (n = 77), we identified 814 polymorphic SNPs.The top 25 genes associated with UTI recurrences are presented in Supplementary Table 8.A trend of group separation was observed, although it was less distinct than the clear separation observed in the previous RS compared to resolved APN analysis (Figure 2 E-F).This group showed heterogeneity with no dominant immunity genes.The most related genes were Peli2, linked to IL-1 family signaling, and the transcription factor ZNF195, which has a kidney association (www.genec ards.org).When the allele counts of the representative SNPs for these 25 genes were compared against the SweGen database, it was found that many of them exhibited significantly different allele proportions also on the populations level.
IPA of the canonical pathways identified 46 pathways related to recurrent UTI, including genes associated with mitochondrial dysfunction (Supplementary Table 9).IPA-Tox analysis revealed gene expression changes associated with UTI recurrence genes, mainly related to liver and heart diseases, but also nephrosis (Supplementary Table 10) [32].Network analysis indicated an association with connective tissue disorders and organismal abnormalities among the ten most significant diseases and functions annotated by IPA.

Common risk genes of RS, dilating VUR and recurrent UTI
To identify possible shared risk genes, a Venn analysis was performed on the significant genes associated with RS, dilating VUR, and recurrent UTI.We identified 36 common genes (p < 0.005) considered as RS susceptibility genes, of which 23 are mitochondrial (Figure 3).IPA of all the common genes of RS, VUR grades 3-5, and recurrent UTI was also conducted.Affected canonical pathways are presented in Supplementary Table 11 and include mitochondrial dysfunction.Annotation of toxic functions describes major organ involvement, with cardiac diseases being most dominant, followed by liver and mitochondrial diseases, in addition to renal fibrosis and glomerular disease (Supplementary Table 12).

Discussion
In our population, RS was linked with a mitochondrial gene profile, distinguishing it from the profile of children with resolved APN and those without any kidney involvement in the first DMSA.The association of polymorphic mitochondrial genes with RS in children has not been previously demonstrated, but the clinical relevance is supported by research describing the role of mitochondria in acute kidney injury and in chronic kidney diseases such as focal segmental glomerulosclerosis and tubulopathy [33][34][35].Since mitochondrial genes affect most organs, they play a role in the heterogeneity of disease phenotypes in children [36].A recent publication describes a threshold level of mutated mitochondrial DNA, which varies between different tissues depending on energy demand.This makes the brain, heart, and kidneys especially sensitive [35], which could possibly explain why many annotations by IPA link diseases and toxic functions to these organs.Additionally, in our study, canonical pathway analysis further supports the association of mitochondrial engagement with RS.The pathogenesis of scarring could involve sensitivity to infection due to affected mitochondrial function.The mitochondrial role in immunity encompasses the production of reactive oxygen species (ROS) and other signaling molecules such as mitochondrial constituents and metabolic products in the progression of inflammatory disorders [37][38][39].For example, the leakage of mtDNA is known to increase the levels of ROS, leading to hyperactivation of inflammatory responses resulting in kidney damage and pathology [33].This is intriguing given previous studies showing that a distinctive genetic APN profile is related to hyperinflammation in combination with inhibited lymphocyte maturation (unpublished observations).In our population, we also found XRCC associated with RS.This gene is important for DNA repair and has previously been linked to kidney failure risk [37][38][39], thus connecting RS susceptibility to kidney failure.
In this study, we focused on identifying genetic traits associated with susceptibility to RS following febrile UTI.We observed a clear genetic distinction between patients with RS and those with resolved APN.We specifically chose to include only children who had undergone two DMSA scans.Furthermore, since participants were managed according to the local routines of each center with variable access to acute DMSA scan, the number of DNA samples from patients with a first DMSA examination within seven days from UTI diagnosis was relatively low.However, we still discovered a highly significant group of genes associated with RS, which strengthens our results.
We observed associations between dilating VUR, recurrent UTI, mitochondrial polymorphisms, and dysfunction.These findings bridge the genetic results from RS to the   [48].
The choice of significance level significantly influences bioinformatic findings.Using a lower threshold (p < 0.05), genes previously linked to kidney disease (TINAG and COL4A1) were found among common risk genes [49][50][51].A literature comparison of CAKUT-related genes identified five common genes reported in all three groups-RS, VUR grades 3-5, and recurrent UTI [52].Furthermore, PAX2 was identified as a shared gene between RS and dilating VUR.Mutations in the PAX2 gene have previously been implicated in kidney abnormalities, including VUR [53][54][55].The hepatocyte nuclear factor 1β (HNF1β), associated with kidney pathology in CAKUT patients, was detected in the RS group [56][57][58].Finally, vascular endothelial growth factors (VEGF A and C) were significant in RS (p < 0.05).VEGF gene variants have previously been suggested as potential genetic markers for VUR and UTI [59,60].
The interpretation of our findings presents a considerable challenge.Our primary objective was to identify potential genetic determinants in children predisposed to RS. Clinically, there is an overlap between scarring due to congenital malformations and infection-induced RS, but in our material, we are not able to separate these conditions with certainty.We believe that these infants are representative of the group of children where kidney defects are detected after febrile UTI and hypothesize that concurrent factors like acute inflammation and CAKUT may contribute to the scarring susceptibility after febrile UTI.In our investigation, we have unveiled a cluster of genes associated with crucial biological processes, including metabolism, mitochondrial function, cell differentiation, and developmental pathways.Notably, the pronounced involvement of mitochondrial function stands out in our results.While additional research involving the genetic underpinnings of RS remains imperative, our current findings posit that mitochondrial function plays a pivotal role in rendering individuals more susceptible to RS when carrying specific mitochondrial function-affecting polymorphisms.It is noteworthy to mention that bilateral RS leads to enduring complications such as hypertension, a reduced GFR, and an increased risk of progressing to kidney failure [61][62][63][64].In a follow-up study of patients after childhood UTI, Jacobson et al. found that even unilateral RS is associated with increased renin activity levels, diastolic hypertension and decreased GFR compared to controls, but small unilateral scars are not associated with long-term complications [65,66].Our findings substantially advance our understanding of the fundamental mechanisms contributing to the pathogenesis of RS.
The absence of reliable RS predictive markers can lead to unwarranted and potentially harmful diagnostic procedures in infants followed up after febrile UTI, increasing their exposure to ionizing radiation and incurring avoidable costs.Hence, the identification of susceptible individuals is of paramount importance to facilitate personalized and effective management strategies in the future.The development of novel tools for the early detection of RS susceptibility is imperative.As suggested by Khan et al. for CKD, the consideration of implementing a polygenic risk score is a prospective avenue for identifying children at risk of experiencing long-term complications following a febrile UTI [67].

Fig. 1 A
Fig.1A Flow chart depicting the inclusion criteria and outcomes of DMSA scans, identifying infants with kidney involvement after the first DMSA and renal scarring (RS) after the second DMSA evaluations (one result is missing* and one result is excluded due to nephrectomy**).B Bar chart illustrating a higher incidence of RS in infants diagnosed with VUR grade 3-5 compared to those with VUR grades 0-2 (p < 0.001).C Bar chart indicating that RS is more prevalent among children who experienced UTI recurrences (p < 0.001)

Fig. 2 A
Fig. 2 A Heatmap illustrating the segregation of SNPs based on an allelic frequency test (p < 0.005) and calculation of odds ratio (OR), comparing infants with renal scarring (RS) after acute pyelonephritis (APN) (n = 7) and infants with resolved APN (n = 25).B Principal component analysis plot depicting the genetic differences between the

Table 2
Genes associated with renal scarring (RS).Polymorphic genes identified through DNA association analysis (allelic frequency test p < 0.005) show a prevalence of mitochondrial genes among the most significant ones linked to RS in both analyses, when compared to resolved APN

Table 3
Genes associated with renal scarring (RS) were compared between cases of RS and those without acute pyelonephritis (no APN) indicated by a negative first DMSA examination.Mitochondrial genes dominate among the most significant genes associated with RS in both analyses