Patient characteristics

We enrolled 332 SRNS patients in our single-center study, including 219 boys and 113 girls (males: females 1:0.52, P = 0.044 < 0.05, with significant differences), with an average age of 69.9 ± 50.4 months old, who had used at least one type of genetic test. The baseline characteristics of all SRNS patients are summarized in Table 2. Supplementary Table 1 (see Additional file 1) lists the results of all detected causative mutations.

Table 2. Baseline characteristics of the study population.

Table 2 Baseline characteristics of the study populationGeneral information

We performed different sequencing in a single cohort of 332 SRNS children, and causative mutations detected by any method were counted as positive. A total of 100 SRNS-related causative mutations were detected, counting the positive rate of single-gene was 30.12% (100/332). The family history significantly increased the diagnostic rate compared to children without it (15.79% versus 4.61%, P = 0.01 < 0.05). Forty types of causative genes related to SRNS disease were found. The distribution of these genes is shown in Fig. 1, of which COL4A5 was the most detected, with a total of 15, followed by WT1, with a total of 13.

Fig. 1

The proportion of gene distribution in SRNS patients. Among them, COL4A5 accounted for the most, 4.52% (15/332), followed by WT1 at 3.92% (13/332), PAX2 at 2.10% (7/332), COQ8B at 1.80% (6/332), COL4A3, NPHS1, and NPHS2 all at 1.51% (5/332), TRPC6 0.90% (3/332); a total of 41 other genes, accounting for 12.35% (41/332)

Clinical manifestations with causative mutations

SRNS patients with causative mutations included 58 boys and 42 girls (males: females 1:0.72). The onset of illness ranged from 1 month to 15.2 years of age, and the median age in individuals in whom a causative mutation was detected was 68.2 ± 53.6 months old.

The proportions of patients with detected causative mutations were as follows: onset in the first 3 months of life (5/8, 62.5%); 3–12 months (9/19, 47.4%); aged 1–3 years (29/98, 29.6%); aged 3–6 years (18/80, 22.5%); aged 6–12 years (30/94, 31.9%); and aged 12–18 years (9/33, 27.3%). WT1 was more common in patients aged 3 months to 3 years old, and COL4A5 in children aged 1 to 12 years old. The results for the detection of causative mutations at different ages are shown in Fig. 2, and the specific distribution of these causative genes is shown in Fig. 3.

Fig. 2

Age of onset distribution (in years) for 332 SRNS patients. Numbers on bars represent the proportion of affected individuals in different age groups; Dark: patients with a causative mutation detected; Gray: patients without a causative mutation

Fig. 3

Number of patients with causative mutation detected per gene per age group. Numbers on bars represent proportion of causative mutations in different age group: onset in the first 3 months of life (5 patients); 3–12 months (9 patients); aged 1–3 years (29 patients); aged 3–6 years (18 patients); aged 6–12 years (30 patients); and aged 12–18 years (9 patients)

Clinical phenotype and renal biopsy

Among children with causative mutations, 70.0% (70 of 100) of SRNS patients had glomerular haematuria and/or renal insufficiency and/or hypocomplementaemia at onset, and 30 children (30.0%) did not have the above symptoms.

A total of 168 of 332 SRNS patients underwent renal biopsy at different ages but were not chosen by the age of 0–3 months old. The pathological phenotypes in each age group are shown in Fig. 4. Pathological changes were seen to be predominantly minimal change disease (MCD) and focal segmental glomerulosclerosis (FSGS), with MCD being higher than FSGS. A renal biopsy was performed in 42 of 100 individuals with disease-causing mutations, showing 13 patients with MCD, 18 patients with FSGS, 4 patients with mesangial proliferative glomerulonephritis (MsPGN), 1 patient with membrano-proliferative glomerulonephritis (MPGN), and 6 patients with other conditions. The distribution of different causative genes in different renal biopsy types is shown in Fig. 5.

Fig. 4

SRNS pathological types of different age groups. Other: Alport syndrome, various glomerulonephritis, membranous nephropathy

Fig. 5

The distribution of different causative genes in different renal biopsy types. A total of 42 children were detected with causative mutations and completed renal biopsies. Inner segments represent the numbers and fractions of renal biopsy types, specifically as follows: MCD, 13 out of 42; FSGS, 18 out of 42; Other1, 11 out of 42. Outer segments represent for each renal biopsy group the relative fraction of different causative mutations. We described the 5 specific causative mutations that accounted for a large share or occurred frequently, others are grouped in other2. The distribution is as follows: WT1 (1 out of 42), COL4A3 (1 out of 42), COL4A5 (1 out of 42), NPHS1 (2 out of 42) and other causative mutations (8 out of 42) were detected in MCD group. WT1 (1 out of 42), COL4A5 (2 out of 42), NPHS1 (1 out of 42), NPHS2 (2 out of 42) and other causative mutations (12 out of 42) were detected in FSGS group. COL4A3 (1 out of 42), COL4A5 (5 out of 42) and other causative mutations (5 out of 42) were detected in other renal pathological group. Other1: Alport syndrome, various glomerulonephritis, Membranous nephropathy. Other2: Other causative genes detected in patients with different pathological groups

Therapy

In this study, a total of 332 children were included, and 122 (36.7%) achieved complete remission of urinary protein, including 17 (5.1%) children who were monogenic positive and 105 (31.6%) with negative genetic testing. Partial remission of urinary protein occurred in 66 (19.9%) paediatric patients, of whom 10 (3.0%) were monogenic positive and 56 (16.9%) were genetic testing negative. During treatment, 123 children developed immunosuppression tolerance and did not meet the criteria for proteinuria remission, including 68 (20.5%) children who were monogenic positive and 55 (16.6%) children who were negative by genetic testing negative.

There were different options for treatment, as shown in Table 2. A total of 122 children achieved complete remission of urinary protein, 108 children chose steroids + TAC, 12 children chose steroids + mycophenolate mofetil (MMF)\cyclophosphamide (CTX), and 2 children did not choose immunosuppressive therapy. Among the TAC group, 15 were monogenic positive, and 93 were negative. Of the 12 MMF\CTX group children, only 2 were monogenic positive. In children with other non-immunosuppression therapies, 2 achieved complete remission and were both genetically negative, and most children did not achieve remission during the process of treatment.

Comparison of genetic testing modalities and costsProcess of acquisition

In our study, the choice of genetic tests can be divided into two different strategies. The details of the testing process are shown in Fig. 6. In the first strategy, all 46 panel-negative patients chose further WES, and 4 additional mutations were found, increasing the positive rate by 8.69% (4/46). Conversely, 117 of 195 WES-negative patients chose further WGS, and 5 additional mutations were found, increasing the positive rate by 4.27% (5/117).

Fig. 6

Flowchart for the selection of genetic testing strategies for SRNS patients. The first strategy was used in children who chose the panel first with 96 in total, 50 of whom had causative mutations, and 46 had negative results; all of the 46 panel-negative children chose to have further WES tests, and 4 additional causative mutations were detected in the end. Another group included 236 children who chose WES first, 41 of whom had causative mutations, and 195 had negative results. Of the 195 WES-negative children, 117 chose to be further tested for WGS, and finally, 5 additional causative mutations were detected

Analysis of additional detected genes

When analysing the results obtained by the first-selected panel or WES when the second genetic tests were not yet performed, we found that the overall causative mutation detection rates were 52.08% (50/96) and 17.37% (41/236), respectively. The types of causative genes were also analysed at the same time, and we found that a total of 19 types were detected by gene panel testing and 26 types were detected by WES. The top 3 causative mutations detected by the panel were COL4A5 (13 cases), WT1 (9 cases) and PAX2 (6 cases), accounting for 56.0%. The top mutation detected by WES was COQ8B (5 cases), followed by WT1 (3 cases) and NPHS2 (3 cases); all three genes accounted for 26.83%.

WES detected 4 additional causative mutations based on the negative panel outcome: TTC21B, COL4A5, WT1, and COQ8B. WT1 and COQ8B were detected in children whose first-selected mitochondria panel did not contain those two genes. Among the remaining 2 causative mutations, TTC21B detected 2 point mutations in the first panel, but subsequent Sanger sequencing found that both of the mutations were from this child’s mother, who did not show any phenotype of NS. However, later WES considered the incomplete penetrance and again concluded that it was a causative mutation. The COL4A5 mutation type was exon deletion. After asking the gene company, we found that the panel was not very good at detecting exon deletions in this case due to insufficient probe density at the time.

Five additional causative mutations were detected by WGS based on negative WES, including four species, which were NIPBL, COL4A3, TRNL1, and COL4A4. Two TRNL1 mutations were located in mitochondria. The remaining 3 causative mutations were all missense mutations. The COL4A3 mutation was not found in the first WES but was found with WGS; however, we later decided to reanalyse the WES data and finally reached a positive result for COL4A3, similar to the WGS results. Generally, WES will not miss the COL4A4 mutation. In the case of COL4A4, we consulted the genomic company about this matter and learned that the patient was sequenced at the first genomic company, then copied the sequencing date results to the second genomic company and went WES pure data analysis. WES failed to find the mutation for the first time probably due to poor sequencing quality at the first company, but later WGS detected the COL4A4 mutation at the second genomic company successfully. The last NIPBL mutation was initially identified by WES, but it was not considered a causative mutation. Subsequent WGS added parental verification and eventually designated it as a spontaneously mutated disease-causing gene.

Economic-benefit ratio

The Children’s Hospital of Chongqing Medical University is the National Clinical Research Center for Child Health and Disorders, located in southwest China. A comprehensive look at the quotes for genetic tests from two genetic companies commonly found in this region showed that a single renal disease panel cost approximately $253, a single WES cost approximately $393, and a single WGS cost approximately $562. If Trios were chosen for family line verification, the price was higher: up to $815 for Trios WES and $1,405 for Trios WGS.

For genetic testing benefits when first genetically tested, the first-chosen panel positive rate was 52.08% (50/96) and cost approximately $466 for every 1%, and the first-chosen WES positive rate was 17.37% (41/236) and cost approximately $5,339 for every 1%. All 46 panel-negative individuals selected WES to continue testing, and an 8.69% (4/46) additional positive rate was generated by the additional selection of WES with a total additional cost of $18,078 or $2,080 for every 1% boost; 4.27% (5/117) was generated by further WGS based on WES-negative children with a total additional cost of $65,754 or $15,399 for every 1%. In comparison, WES costs approximately 1/7th the price of WGS for every 1% increase in pathogenicity detection.

Model analysis and cost comparison

We compared the two different models along with the actual cost analysis model and showed them in Fig. 7. Figure 7b shows that the early genetic testing model has reduced the cost per diagnosis of the late genetic testing model by 79% ($464); Fig. 7d shows that the early genetic testing model is 88% ($934) lower than the real-life cost of the 40-patient diagnostic pathway; Fig. 7f shows that the actual cost of a twice genetic test is about 89% ($949) more to diagnose than only once. These results suggested that early and appropriately selected genetic testing can save the cost of diagnosing single-gene-positive SRNS.

Fig. 7

Cost analysis. (a) Modeling diagnostic trajectories for SRNS patients with suspected single gene mutations. tier 1 includes baseline investigations, tier 2 and tier 3 include increasingly complex and/or costly investigations. Diagnostic tests are based on current guidelines and local clinical practice. In the late genetic testing model, patients first undergo Tier 1–3 investigations, followed by genetic testing. In the early genetic testing model, patients go through only Tier 1 and then directly to genetic testing. If positive results were not obtained from the test, further genetic testing would be performed until a final diagnosis was reached. In both models, the prices for all checks were the same as the actual local prices. (b) Comparison of the average cost per diagnosis for late versus early genetic testing models in the study population. (c) Comparison of real-life diagnostic procedure costs with early genetic testing model. Retrieve all costs incurred by the real-life diagnostic procedure for 40 children. Compare the average cost incurred by the real-life diagnostic pathway to the cost of the early genetic testing model. (d) Average real-life versus early genetic testing model cost. (e) The real genetic test model 1 was the cost of the first genetic test to determine the single gene cause in the 40 patients described above, for a total of 36 children diagnosed with the first genetic test. The real genetic test model 2 was the cost of the remaining 4 children who had a second genetic test to determine a single gene cause. (f) Comparison of the actual costs of the real genetic testing model 1 with the real genetic testing model 2