Cystic phenotype and chronic kidney disease in autosomal dominant Alport syndrome

ABSTRACT

Background

Autosomal dominant Alport Syndrome (ADAS), also known as thin basement membrane disease (TBMD), is caused by pathogenic variants in the COL4A3 and COL4A4 genes. A cystic phenotype has been described in some patients with TBMD, but no genetic studies have been performed. We conducted a genetic and radiologic investigation in a cohort of ADAS patients to analyze the prevalence of multicystic kidney disease (MKD) and its association with chronic kidney disease (CKD).

Methods

This was a retrospective single-center cohort study. Thirty-one patients showing pathogenic or likely pathogenic variants in COL4A3 or COL4A4 from a cohort of 79 patients with persistent microscopic hematuria were included. Mean follow-up was 9.4 ± 9.6 years. The primary objective of the study was to determine the prevalence of MKD in the cohort of ADAS patients. Secondary objectives were to determine risk factors associated with an estimated glomerular filtration rate (eGFR) <45 mL/min/1.73 m² at the time of genetic and radiologic evaluation and to investigate the coexistence of other genetic abnormalities associated with familial hematuria and cystic kidney disease.

Results

MKD was found in 16 patients (52%). Mean number of cysts per kidney was 12.7 ± 5.5. No genetic abnormalities were found in a panel of 101 other genes related to familial hematuria, focal segmental glomerulosclerosis and cystic kidney disease. A greater number of patients with MKD had an eGFR <45 mL/min/1.73 m² (63% vs 7%, P = .006) and more advanced CKD than patients without MKD. The annual rate of eGFR decline was greater in patients with MKD: –1.8 vs 0.06 mL/min/1.73 m²/year (P = .009). By multivariable linear regression analysis, the main determinants of eGFR change per year were time-averaged proteinuria (P = .002) and MKD (P = .02).

Conclusion

MKD is commonly found in ADAS and is associated with a worse kidney outcome. No pathogenic variants were found in genes other than COL4A3/COL4A4.

Graphical Abstract

Open in new tabDownload slide

INTRODUCTION

Autosomal dominant Alport syndrome (ADAS) is a genetic disorder caused by pathogenic variants in the genes encoding the α3 (COL4A3) and α4 (COL4A4) chains of type IV collagen [1–5]. The most characteristic clinical manifestation is a persistent microscopic hematuria and until recently it was considered a benign disorder (familial benign hematuria). However, several studies have shown that a non-negligible number of patients develop proteinuria, chronic kidney disease (CKD) and even kidney failure [6–13]. Diffuse thinning of the glomerular basement membrane is the histologic hallmark of this entity; however, superimposed lesions of focal segmental glomerulosclerosis are frequently observed, particularly in patients who develop CKD and kidney failure. In fact, pathogenic variants in COL4A3/COL4A4 genes have been shown to be one of the most common causes of focal segmental glomerulosclerosis (FSGS) [11, 14–16].

Recent studies have shown that pathogenic variants in the COL4A3/COL4A4 genes are a frequent finding in patients with kidney failure of unknown cause. However, the clinical phenotype and the factors involved in the progression to kidney failure of some ADAS patients remain to be fully defined. A study from our group reported that more than half (9 out of 16) of patients with persistent hematuria and a histologic diagnosis of thin basement membrane disease (TBMD) had bilateral kidney cysts [17]. The presence of cysts was associated with greater proteinuria and worse kidney function. However, genetic testing was not performed in the aforementioned study. Another investigation found pathogenic variants in COL4A4 in three patients with bilateral kidney cysts and CKD [18]. These preliminary results suggest that the finding of multicystic kidney disease (MKD) is relatively frequent in ADAS and is associated with a poorer kidney prognosis. However, studies with a comprehensive genetic study to exclude the involvement of other genes associated with cystic kidney diseases, and with a long follow-up to analyze the association of the cystic phenotype with worse kidney outcomes, are needed to confirm these data.

The aim of this study was to perform a complete genetic and radiologic investigation to define the prevalence of MKD in ADAS, to search for the possible presence of concomitant genetic abnormalities associated with familial hematuria, FSGS and cystic kidney diseases, and to analyze the association of MKD with the risk of developing CKD. Moreover, since the finding of kidney cysts may be more common in patients with CKD [19, 20], we performed a kidney ultrasound study in a cohort of patients with immunoglobulin A nephropathy (IgAN) of similar age and kidney function to that of the ADAS cohort, as well as in a small group of patients with X-linked Alport syndrome.

MATERIALS AND METHODS

Study patients

Adult patients with persistent microscopic hematuria of glomerular origin (three or more dysmorphic red blood cells per field in at least 90% of urine sediments) and routine outpatient follow-up at Hospital Universitario 12 de Octubre between 1985 and 2021, were selected for the study. All eligible patients who agreed to participate underwent genetic testing (as described in the following section), and only those with pathogenic or likely pathogenic variants in COL4A3 or COL4A4 and belonging to different families were finally included in the study. Those with known secondary causes of hematuria, those without available follow-up and those who did not consent genetic testing, were excluded.

The study was approved by the Institutional Review Board of Hospital Universitario 12 de Octubre (19/407) and was conducted in accordance with the tenets of the Declaration of Helsinki. All participants provided written informed consent.

Clinical, laboratory and radiologic data

Baseline and follow-up data were collected from medical records, including demographics, comorbidities (e.g. hypertension, type 2 diabetes, obesity), previous history of macroscopic hematuria bouts, hearing loss or ocular abnormalities. Family history of hematuria, hearing or eye abnormalities, presence of other concomitant kidney diseases, and history of CKD or kidney replacement therapy were also recorded. Laboratory data included serum creatinine (SCr), estimated glomerular filtration rate (eGFR), 24-h urine proteinuria and the number of red blood cells per high-power field (RBC/h.p.f.) in urinary sediment, at baseline and during follow-up. Treatments received during follow-up were also recorded.

eGFR was calculated using the CKD Epidemiology Collaboration formula. The slope of eGFR was calculated for each patient and presented as annual rates of kidney function loss. Average proteinuria was determined for every 6 months of follow-up for each patient. A time-averaged (TA) proteinuria was then calculated for each patient, representing an average of the mean proteinuria measurements in each 6-month period. The same method was used to calculate TA hematuria, with the number of RBC/h.p.f. recorded at each visit.

Kidney biopsy had been performed in five patients. Diagnosis of TBMD was defined by a thickness of the glomerular basement membrane <150 nm in at least 50% of its surface.

A kidney ultrasound was performed in all study patients. Data on kidney size, as well as the number and size of cysts, were recorded. MKD was defined by the presence of five or more bilateral kidney cysts. Patients with MKD underwent magnetic resonance imaging or abdominal computed tomography to accurately characterize and quantify the kidney cysts.

Genetic studies

A comprehensive genetic study was performed on all study patients by the Department of Genetics at Hospital Universitario 12 de Octubre. In all, 101 genes were analyzed, including COL4A genes and a complete panel of other genes related to familial microhematuria, FSGS and cystic kidney diseases, as described in Supplementary data, Table S1.

Studies were performed using whole-exome next-generation sequencing (xGen Exome Panel v1.0 kit). Paired-end sequencing (2 × 75 bp) was performed on the Illumina NextSeq 550 sequencer (San Diego, CA, USA). Bioinformatic analysis was performed using the Karma tool, an in-house pipeline integrating BWA (v0.7.17) and Bowtie2 (v2.4.1) for sequence alignment to the reference genome (hg19 assembly), GATK (v4.1) and VarDict (v1.7.0) for genotyping, ExomeDepth (v.1.10) for CNV detection, AnnotSv (v2.4) for CNV annotation and ANNOVAR (v2018Apr16) for variant annotation. Karma follows the validation recommendations of the Association for Molecular Pathology [21]. Genetic variants were filtered according to alignment and genotyping quality metrics. Variants identified from alignments with low mapping quality, variants with strand bias, variants with a frequency >3% in the gnomAD population database (v2.1.1), and variants classified as benign and probably benign according to the ClinVar database were not evaluated. Clinical interpretation and final classification of the identified variants were performed using information extracted from the gnomAD population database; the genetic variant databases ClinVar, Leiden Open Variant Database (LOVD) and Human Gene Mutation Database (HGMD); the protein domain and structure databases Uniprot, PFAM and Prosite; and the hereditary disease databases OMIM, Orphanet and GeneReviews. Variants were classified as pathogenic, likely pathogenic and variants of uncertain significance according to the recommendations of the American College of Medical Genetics and Genomics (ACMG) [22].

Objectives and definitions

The primary objective of the study was to determine the prevalence of MKD in the cohort of ADAS patients. Secondary objectives were to determine risk factors associated with an eGFR <45 mL/min/1.73 m² at the time of genetic and radiologic evaluation, to investigate the coexistence of genetic abnormalities associated with familial hematuria, FSGS and cystic kidney disease, and to compare the prevalence of MKD in ADAS patients with that of other chronic kidney diseases. For the latter purpose, 24 patients with biopsy-proven IgAN of similar age and kidney function to ADAS patients were selected. IgAN patients underwent kidney ultrasound using the same technique and recording the same data as the ADAS patients.

Baseline was defined as the first visit to the outpatient clinic and follow-up, as the interval between the first visit and the performance of the genetic and radiologic assessment (study entry). Kidney failure was defined as an estimated eGFR <15 mL/min/1.73 m² or the need for renal replacement therapy.

Control group

In order to compare the prevalence of MKD in ADAS patients with that of other chronic kidney diseases, 30 patients with biopsy-proven IgAN of similar age and kidney function to ADAS patients were selected. IgAN patients underwent kidney ultrasound using the same technique and recording the same data as the ADAS patients. Variants of COL4A3/COL4A4/COL4A5 genes were screened in 11 of these IgAN patients. Finally, a similar kidney ultrasound study was performed in 10 patients with X-linked Alport syndrome (pathogenic variants in COL4A5 genes).

Statistical analyses

This is a retrospective, observational cohort study. Descriptive statistics are presented as mean and standard deviation, or median and interquartile ranges (IQR) for continuous variables, and frequencies or percentages for categorical variables.

Comparisons of continuous variables between two groups were assessed by using the unpaired t-test or the Mann–Whitney U test, where appropriate. Chi-squared test or Fisher's exact test were used for categorical variables.

Multivariable linear regression analysis was used to analyze the main determinants of percentage change in eGFR over follow-up time.

A P-value <.05 was considered to be significant. All P-values are reported two-sided. Analyses were performed using IBM SPSS Statistics 24.0 (IBM Corp., Armonk, NY, USA).

RESULTS

Cohort description

During the study period, data from 79 patients with persistent microscopic hematuria were retrieved, of whom 18 were excluded due to lack of follow-up and 9 refused to participate in the study. Of the 52 remaining patients, 21 were additionally excluded because genetic testing found no genetic abnormalities or only variants of unknown significance in COL4A3/COL4A4 (Fig. 1). Clinical and radiological data for these 21 patients are shown in Supplementary data, Table S3. Thus, the final study group comprised 31 Caucasian patients with pathogenic or likely pathogenic variants in COL4A3 or COL4A4. Mean baseline age was 50 ± 12 years, and 15 patients (48%) were males.

Table 1 summarizes the main clinical and biochemical characteristics at baseline, and at the time of genetic and radiologic study. All patients presented microscopic hematuria and six of them (19%) referred previous episodes of gross hematuria. Four cases (12%) presented unilateral or bilateral hearing loss and one patient (3%) had retinal abnormalities. Five patients (16%) had a family history of hearing loss and two (7%) of ocular abnormalities. Thirteen patients (41%) had first-degree relatives with persistent microhematuria and 14 (45%) with CKD, which had required kidney replacement therapy in 10 cases (32%). Three patients had a family history of cystic kidney disease in which the etiology had not been specified.

At the time of genetic and radiologic study [after a mean follow-up of 5.9 years (IQR 3–11.1)], mean SCr was 1.3 ± 0.7 mg/dL and mean eGFR 67±31 mL/min/1.73 m² (Table 1). An eGFR <60 mL/min/1.73 m² was found in 14 patients (45%) and an eGFR<45 mL/min/1.73 m² in 11 (36%). There were only two patients (6%) with an eGFR <15 mL/min/1.73 m². Mean slope of eGFR was –0.57 mL/min/1.73 m²/year. Median 24-h proteinuria was 0.32 g (IQR 0.1–0.6), being ≥1 g/day in six patients (19%). None showed nephrotic-range proteinuria. Microscopic hematuria persisted during follow-up in all patients, with a TA hematuria of 14 (IQR 1.8–30) RBC/h.p.f.

Table 2 displays the main characteristics of patients according to eGFR, under/above eGFR 45 mL/min/1.73 m² at the time of study. Patients with eGFR <45 mL/min/1.73 m², had significantly greater degree of proteinuria (0.98 vs 0.29 g/24 h, P = .003) and TA proteinuria (0.6 vs 0.18 g/24 h, P = .002). Moreover, the presence of MKD was significantly more frequent in those with eGFR <45 mL/min/1.73 m² (50% vs 30%, P = .001). No significant differences were found in the frequencies of pathogenic variants in COL4A3/4 genes, nor in the concomitant treatment with renin-angiotensin-aldosterone system blockade.

A kidney biopsy had been performed in five patients (16%). Light microscopy was normal except for mild mesangial hypercellularity in three cases (60%). FSGS lesions (15%–50% of the glomeruli), associated with mild interstitial fibrosis and tubular atrophy, were found in three patients (60%). All kidney biopsies showed a thin glomerular basement membrane, and negative immunofluorescence staining for all immune-reactants.

Genetic findings

Sixteen patients (52%) had pathogenic or likely pathogenic variants in COL4A4, 14 patients (45%) in COL4A3 and 1 patient (3%) in both genes. Variant categories were missense in 24 cases (77%), splicing in 4 (13%), frameshift in 2 (6%) and large deletion in 1 (3%). Supplementary data, Table S2 provides a detailed description of each individual genetic variants, together with the main clinical features and the presence or absence of MKD. No genetic abnormalities were found in other panels including genes related to familial microscopic hematuria, FSGS and cystic kidney disease.

Radiologic imaging

MKD was found in 16 patients (52%) (Fig. 2). Mean number of cysts per kidney was 12.7 ± 5.5. The largest cyst was 5 cm and 65% of the cysts were ≥5 mm. Twenty-four patients (77%) had normal-size kidneys (10–12 cm), whereas small-size kidneys (<10 cm) were observed in six patients (19%) and slightly enlarged kidneys (12.5 cm) in one patient (3%). Cysts were mainly located in the cortical and subcortical areas. Previous ultrasound was only available in one patient with MKD who had an SCr of 1.8 mg/dL at the time of the study. In a kidney ultrasound performed 5 years earlier, when the SCr was 1.1 mg/dL, bilateral kidney cysts were already present.

Comparison between patients with and without MKD

Table 3 shows the characteristics of patients with and without MKD. MKD patients were older (65.9 ± 8.7 vs 52.8 ± 11 years, P = .01) and had a significantly worse kidney function than patients without MKD (eGFR 50.2 ± 26.4 vs 83.3 ± 26.1 mL/min/1.73 m², P = .004). A greater number of patients with MKD had an eGFR <45 mL/min/1.73 m² (63% vs 7%, P = .006) and more advanced CKD than patients without MKD. The annual rate of eGFR decline was also greater in patients with MKD: –1.8 (IQR –3.2 to –0.5) vs 0.06 (IQR –0.9 to 0.6) mL/min/1.73 m²/year (P = .009) (Fig. 3). TA proteinuria was higher in MKD (Table 3), whereas TA hematuria was similar. No differences in other clinical characteristics, family history or type of genetic abnormalities (Table 3 and Supplementary data, Table S2) were observed. The number of patients treated with renin-angiotensin-aldosterone system blockers was similar in both groups.

As shown in Table 4, there were no differences between patients with or without MKD in the proportion of patients with pathogenic variants in COL4A3 and in COL4A4 genes. Likewise, no differences were found in the type of pathogenic variants (missense, splicing, frameshift or gross deletion) between patients with or without MKD.

Determinants of percentage change in eGFR over time

By multivariable linear regression analysis, the main determinants of percentage change in eGFR per year were (Table 5): TA proteinuria (β = –0.51; P = .002) and MKD (β = –0.35; P = .02).

Control group

Table 6 displays the main clinical and biochemical characteristics of IgAN patients, and the comparison with the ADAS cohort. There were no differences in age, kidney function or the degree of proteinuria. However, MKD was significantly more frequent in ADAS patients compared with IgAN patients (52% vs 10%, P = .001) (Table 6). MKD was also more common among ADAS patients with eGFR <45 mL/min/1.73 m² (91% vs 13%, P = .001). No pathogenic variants in COL4A3/COL4A4/COL4A5 genes were found in the 11 IgAN patients in whom a genetic study was performed.

As shown in Table 7, all patients with X-linked Alport syndrome had a normal or small kidney size. Four of them (40%) showed MKD, with a number and size of cysts similar to that of patients with ADAS. All X-linked Alport syndrome patients with MKD showed advanced CKD (mean eGFR 15.5 ± 11.4 mL/min/1.73 m²).

DISCUSSION

ADAS, formerly known as familial benign hematuria or TBMD, is one of the most common kidney diseases, affecting approximately 1% of the population [13, 23, 24]. Despite this high prevalence, ADAS remains a frequently unrecognized disorder and many aspects of its phenotype remain largely unknown. Several studies [6–13] have shown that a substantial proportion of ADAS patients develop CKD and even kidney failure, while others maintain normal kidney function throughout life. The factors underlying the development of CKD in ADAS remain incompletely understood.

The most salient findings of our study are the presence of MKD in more than half of the patients with ADAS and the association of MKD with a worse kidney outcome. Furthermore, the comprehensive genetic analysis performed (including genes related to other cystic kidney diseases) excluded the presence of pathogenic variants in genes other than COL4A3/COL4A4.

The clinical and laboratory data of our patients were otherwise typical of ADAS. All the cases presented with persistent microscopic hematuria and, in a minority of cases, hearing loss or retinal abnormalities were observed. A substantial proportion of patients reported a family history of CKD and the need for kidney replacement therapy in relatives. Our study also confirms the occurrence of CKD in a significant proportion of ADAS patients. Although the rate of decline in kidney function was slow overall, the proportion of patients with an eGFR <60 mL/min/1.73 m² almost doubled from baseline to the last visit, after a mean follow-up of 9 years. The presence of MKD was clearly associated with impaired kidney function in our cohort. Patients with MKD had a significantly higher TA proteinuria and a significantly faster kidney function decline compared with patients without MKD. The proportion of patients with an eGFR <45 mL/min/1.73 m² at the end of follow-up was nine times higher in patients with MKD.

Some previous studies had provided preliminary data on the presence of kidney cysts in ADAS patients. Our group found a high prevalence of MKD in patients with TBMD (53%), but a genetic analysis was not performed in this study [17]. Gulati et al. [18] described four patients with pathogenic variants in COL4A4 and bilateral kidney cysts. No mutations in PKD1 and PKD2 genes were detected in these patients. Pierides et al. [11] reported multiple kidney cysts in four families presenting with persistent microscopic hematuria and heterozygous mutations in COL4A3. Our study clearly shows that MKD is present in more than half of ADAS patients and that such association is found in both patients with heterozygous COL4A3 or COL4A4 mutations. On the other hand, our study also rules out the existence of genetic abnormalities in PKD1 or PKD2, or in other genes involved in the development of cystic kidney diseases, familial hematuria or FSGS.

The pathogenic mechanisms underlying this cystic phenotype of ADAS and its association with worse kidney outcomes are unclear. Six different α-helical isoforms (α1–α6) of type IV collagen, encoded by their respective COL4A1–A6 genes, form α1α1α2 and α3α4α5 heterotrimers, which are a major component of the glomerular and tubular basement membranes [25, 26]. Although the α1α1α2 and α3α4α5 isoforms are not uniformly distributed in the glomerular and tubular basement membranes [27], abnormal function of a single gene encoding type IV collagen isoforms may result in abnormally reduced collagen content along the glomerular or tubular basement membranes and the formation of cysts of glomerular or tubular origin. In support of this hypothesis, experimental studies have shown the occurrence of glomerulocystic kidney disease and cystic dilatation of Bowman's capsule and tubules in animals with mutations in different genes encoding for type IV collagen proteins [28, 29]. Of note, kidney cysts are a well-recognized feature of HANAC syndrome, caused by heterozygous mutations in COL4A1 [30]. A cystic phenotype has been reported in isolated patients with mutations in COL4A5 [18, 31]. In our control group of patients with X-linked Alport syndrome, we found a significant proportion (40%) of cases with MKD, with radiological characteristics similar to those of patients with ADAS. Interestingly, the presence of MKD was associated with advanced degrees of CKD, as occurred in previously reported patients [31]. Further studies are needed to conclusively analyze the presence of MKD in X-linked Alport syndrome and its possible association with a worse kidney outcome. On the other hand, it could be hypothesized that more severe mutations in COL4A3/COL4A4 genes could lead to a lower type IV collagen content in glomerular and tubular basement membranes, resulting in an increased propensity for cyst formation and increased glomerular and tubulointerstitial damage. However, the type of gene mutations in COL4A3/COL4A4 did not differ between patients with and without MKD in our study.

Awareness of the frequent presence of MKD among ADAS patients is important to avoid diagnostic confusion with other hereditary cystic kidney diseases, particularly autosomal dominant polycystic kidney disease (ADPKD). In addition to the genetic testing, which definitely resolves equivocal cases, kidney size and the presence of persistent microhematuria are critical in differentiating the two entities. Large kidneys are a typical finding in ADPKD patients and, indeed, there is a strong relationship between kidney volume and poor kidney outcomes. In contrast, the vast majority of ADAS patients have normal or reduced kidney size. In our cohort, all patients had a normal or slightly reduced kidney size, except for one case with a slightly increased kidney size. Cysts in ADPKD are typically more numerous and larger than in ADAS and are frequently accompanied by cysts in the liver and sometimes in other organs, while extrarenal cysts have not been described in ADAS. On the other hand, the presence of persistent microscopic hematuria in ADAS and its familial nature is another important differential feature. Persistent microhematuria is not a characteristic finding in patients with ADPKD, and even less showing a familial aggregation. In conclusion, ADAS should be suspected in any patient with persistent microhematuria and bilateral kidney cysts. Family history and genetic testing will provide the definitive diagnosis.

The presence of kidney cysts in the normal population increases with age, although in most cases they are isolated simple cysts without clinical significance. CKD patients have a greater predisposition to develop kidney cysts [19, 20, 32]. Thus, it could be argued that the presence of MKD in our ADAS cohort could be a non-specific phenomenon related to the development of CKD. However, the radiologic study of a control cohort of biopsy-proven IgAN patients, similar in age and kidney function to the ADAS cohort, showed a significantly lower frequency of MKD compared with ADAS patients. Although pathogenic variants in COL4A3/COL4A4/COL4A5 genes have been reported in some sporadic IgAN patients [33], the analysis of these genes was normal in the 11 IgAN patients from our control group in whom this analysis was performed.

Several limitations should be acknowledged in our study. Due to its observational design, no causal relationship can be established between the appearance of kidney cysts and CKD development and the lack of sequential kidney ultrasounds precluded to observe whether the formation of cysts precedes kidney function decline. Previous radiologic studies were only available in one patient, who had bilateral kidney cysts before the decline in kidney function. Further studies are required to confirm this interesting observation. On the other hand, our study is the first to analyze the cystic phenotype of patients with ADAS in a relatively large number of cases and with a complete genetic study.

In conclusion, our study shows that more than half of ADAS patients develop MKD and that this cystic phenotype is associated with a worse kidney outcome. A comprehensive analysis of genes involved in cystic kidney diseases, familial hematuria and FSGS allowed us to exclude the presence of pathogenic variants in genes other than COL4A3/COL4A4, suggesting that defective collagen content in the glomerular or tubular basement membranes may be involved in the pathogenesis of cyst formation.

ACKNOWLEDGEMENTS

We acknowledge the support from Sociedad Española de Nefrología (SEN).

FUNDING

Work in this study was supported by the Sociedad Española de Nefrología (S.E.N.) (PI 2021/137).

AUTHORS’ CONTRIBUTIONS

Research idea and study design: T.B.-B., A.M.S., F.C.-F., M.P.; data acquisition: T.B.-B., A.M.S.; data analysis/interpretation: M.T.S.-C., C.P.-M., I.A.C., F.D.-C., H.T., M.A., C.C.-C., A.S., J.F.Q.-E., J.M.L.R.; statistical analysis: T.B.-B., F.C.-F.; supervision or mentorship: A.M.S., E.G., G.F.-J., F.C.-F., M.P.; all authors revised the paper and approved the final version of the manuscript.

DATA AVAILABILITY STATEMENT

The data underlying this article will be shared on reasonable request to the corresponding author.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

REFERENCES

Author notes