Demographic and Clinical Variables
Table 1 summarizes the clinical and biochemical characteristics of the INS study cohort (n = 105, 58% males) and healthy controls (CTRL) (n = 19, 94% males). The study cohort includes SSNS (n = 80) and SRNS (n = 25).
Table 1
Clinical characteristics of the enrolled patients at the time of collection.
Parameters | Healthy children CTRL | Active INS disease (n = 71) | SSNS | SRNS | | Inactive INS disease (n = 34) | SSNS | SRNS |
| (n = 19) | | Onset (n = 17) | Rel (n = 37) | (n = 17) | | | Rem (n = 26) | Rem (n = 8) |
Demographic and clinical characteristics | | | | | | | | | |
Sex, n (%) | | | | | | | | | |
Male | 18, (94.7) | 43, (60.6) | 11, (64.7) | 24, (64.9) | 8, (47.1) | | 18, (52.9) | 13, (50) | 5, (62.5) |
Female | 1, (5.3) | 28, (39.4) | 6, (35.3) | 13, (35.1) | 9, (52.9) | | 16, (47.1) | 13, (50) | 3, (37.5) |
Age (yr), median (IQR) | 6 (2–11) | 9, (4–15) | 4 (3–13) | 8 (5–12) | 15 (12-18.5) a, b, c | | 11, (8-14.25) | 11 (6-13.25) | 15 (8.25–17.75) |
Age at onset (yr), median (IQR) | N/A | 4, (3–12) | 4 (3–13) | 4 (2–5) | 9 (4–15) c, d | | 5, (3–8) | 4 (2.37–7.25) | 7 (5.25–12.50) |
Drugs treatment at the time of collection, n (%) | | | | | | | | | |
Immunosuppressants | 0, (0) | 25, (35.2) | 0, (0) | 14, (37.8) | 11, (64.7) | | 30, (88.2) | 24, (92.3) | 6, (75) |
Others | 0, (0) | 4, (5.6) | 0, (0) | 0, (0) | 4, (23.5) | | 0, (0) | 0, (0) | 0, (0) |
NT | 19, (100) | 42, (59.1) | 17, (100) | 23, (62.2) | 2, (11.8) | | 4, (11.8) | 2, (7.70) | 2, (25) |
Biochemical parameters | | | | | | | | | |
Protein-to-creatinine ratio, median (IQR) | N/A | 5.85, (1.86–8.7) f | 8.5 (4.96–9.27) d, e | 3.7 (1.60–8.52) d, e | 2.68 (0.96–8.78) d | | 0.15, (0.12–0.18) | 0.14 (0.11–0.16) | 0.31 (0.19–0.63) |
Urine creatinine (g/L), median (IQR) | 0.77 (0.42–1.23) | 0.99, (0.59–1.6) | 0.8 (0.47–1.56) | 1.0 (0.63–1.64) | 0.98 (0.57–1.88) | | 1.19, (0.8–1.54) | 1.05 (0.57–1.48) | 1.49 (1.02–1.80) |
Serum creatinine (mg/dL), median (IQR) | 0.46 (0.29–0.56) | 0.45, (0.31–0.68) | 0.31 (0.26–0.50) e | 0.40 (0.32–0.59) | 0.84 (0.56–1.06) a, b, c | | 0.52, (0.46–0.62) | 0.51 (0.43–0.57) | 0.67 (0.47–0.95) |
eGFR (mL/min), median (IQR) | 107 (86–132) | 118, (91–145) | 138 (101.5-163.5) | 122 (106.8-149.8) | 74 (58.50–106) b, c, d | | 115, (93–126) | 117 (96.45–126.8) | 99 (78-119.5) |
Immunoglobulins (mg/dL), median (IQR) | | | | | | | | | |
IgG | N/A | 383, (148–618) f | 150 (93–230) c, d, e | 487 (365.5-719.5) | 571 (186.5–679) | | 695, (500.3–919) | 695 (465-849.3) | 763 (644.8–1128) |
IgM | N/A | 127, (96.3–184) f | 121 (100-201.5) d | 127 (72–163) | 162 (114.8–194) d | | 66, (38–110) | 57 (37.5–112) | 82 (58.25–118.8) |
IgA | N/A | 92, (72.3-138.8) | 103 (57-168.5) | 83 (59–129) | 92 (69-192.5) | | 98, (46–151) | 71 (31.5–110) | 143 (95.50–234) |
Immunosuppressant drugs: prednisone, mycophenolate, tacrolimus, Others: ramipril, NT = not treated. IQR = interquartile range. N/A = not applicable. Age: SRNS vs. (a) CTRL (p < 0.01), (b) patients with SSNS onset (p < 0.001) and (c) SSNS Rel (p < 0.01). Age at the onset: SRNS vs. (c) SSNS Rel (p < 0.01) and (d) SSNS Rem (p < 0.05). Protein-to-creatinine ratio: SSNS Onset vs. (d) SSNS Rem (p < 0.001) and (e) with SRNS Rem (p < 0.001); SNSS Rel vs. (d) SSNS Rem (p < 0.001) and (e) SRNS Rem (p < 0.05); SRNS vs. (d) SSNS Rem (p < 0.001); Active INS vs. (f) Inactive INS (p < 0.001). Serum Creatinine: SSNS Onset vs. (e) SSNS Rem (p < 0.05); SRNS vs. (a) CTRL (p < 0.01), (b) SSNS onset (p < 0.001) and (c) SSNS Rel (p < 0.001). eGFR: SRNS vs. (b) SSNS onset (p < 0.001), (c) SSNS Rel (p < 0.001) and (d) SNSS Rem (p < 0.05). IgG: SNSS Onset vs. (c) SSNS Rel (p < 0.01), (d) vs. SSNS Rem (p < 0.001) and (e) SRNS Rem (p < 0.001); Active INS vs. (f) Inactive INS (p < 0.001). IgM: SSNS Onset and SRNS vs. (d) SSNS Rem (p < 0.05); Active INS vs. (f) Inactive INS (p < 0.001). |
The INS population was then subdivided into five groups: Group 1 SSNS patients at the onset of the disease (n = 17, SSNS Onset); -Group 2 SSNS at relapse (SSNS Rel, n = 37); -Group 3 SRNS patients with persistent proteinuria (above 0.5 mg/mg) (SRNS, n = 17); -Group 4 SSNS in remission (SSNS Rem, n = 26); -Group 5 SRNS patients who achieved complete response to second-line treatments (SRNS Rem, n = 8). A control group of age-matched children, with no kidney-related or immunological disease, was included in the study (CTRL). In selected experiments, children in groups 1, 2, and 3 were selected as patients with active INS (n = 71), while children in groups 4 and 5 were in the remission phase of the disease (inactive INS, n = 34). In our cohort, the age was homogenously distributed with a median of 8.5 years (IQR: 4–13) in SSNS and 6 years (IQR: 2–11) in CTRL, while SRNS patients had a higher median age of 15 years (IQR: 10–18). eGFR was reduced (74 mL/min) in SRNS compared to SSNS subgroups (138 and 122 mL/min for SSNS Onset and SSNS Rel, respectively). Serum IgG and IgM levels varied according to disease phase 24, as expected.
Qualitative and quantitative evaluation of urine EVs in INS.
EVs isolated from the urine of INS children (uEVs) exhibited a heterogeneous size and preserved membranes, confirmed by TEM analysis (Fig. 1A). Super-resolution microscopy analysis revealed tetraspanin distribution on single vesicles in both CTRL and INS patients (Fig. 1B). Quantitative analysis showed significantly higher CD63 + and CD81 + uEVs in active INS (62,500 ± 23,300 and 65,700 ± 21,000 respectively, P < 0.05 and P < 0.01) compared to CTRL (22,100 ± 10,200, and 22,000 ± 9,600) and inactive INS (only for CD81, 28,800 ± 10,500, P < 0.05) (Fig. 1C). Capture separation revealed different co-expression patterns among tetraspanins (Fig. 1D), particularly significant for CD9/CD63 and CD81/CD63 in the active phase compared to CTRL (P < 0.05 and < 0.01, respectively). EVs co-expressing CD63/CD81 also showed a different distribution between the active and inactive INS stages (P < 0.05). Triple-positive uEVs were less abundant but significantly increased in the active phase when captured with CD9 (P < 0.001 vs CTRL and P < 0.01 vs inactive INS) and CD81 antibodies (P < 0.05 vs both groups).
Correlations between uEVs/mL in INS children’s urine and kidney function parameters (uCR, uPr/Cr, eGFR) were investigated (Fig. 2A). Spearman’s analysis revealed a significant positive correlation between uEV numbers and uCr levels (R = 0.29, P < 0.01), as well as with the uPr/uCr (R = 0.22, P < 0.05) ratio. uEV dimensions negatively correlated only with the uPr/uCr levels (R=-0.27, P < 0.01) (Fig. 2B). These associations persisted when INS patients were separated from controls (Additional File 1: Figure S1). NTA analysis (Fig. 2C) showed increased uEV numbers (uEVs/uCr) in SSNS Onset compared to CTRL (P < 0.01), and the other groups and clinical time-points (P < 0.001). No gender-based differences were observed in uEV numbers and size in INS (Fig. 2C, D). EV size distribution decreased only between SRNS and SSNS in remission (P < 0.05, SSNS Rem vs SRNS Rem) (Fig. 2D).
uEV surface marker characterization in different INS groups.
uEV surface marker profile was evaluated by flow cytometry in seventy-four INS patients. CD9-CD63-CD81 EV expression positively correlated with uPr/uCr (Fig. 3A), with CD9 most significantly associated with the active state. Tetraspanin levels were higher in active INS compared to controls and inactive INS (CD9 and CD63 P < 0.05 vs CTRL and P < 0.001 vs inactive INS; CD81 P < 0.001 vs inactive INS) (Fig. 3B).
Global EV-surface antigens distribution in each INS, analyzed through PCA, distinctly separated SSNS in relapse from onset and the SRNS group (Fig. 3C). Unsupervised hierarchical clustering confirmed the clustering of INS patients in the active phase (Fig. 3D). Marker frequency analysis revealed exclusive markers in INS children (CD25, CD20, CD11c, CD2, CD49e, CD62p, and CD42a) with higher frequencies during the active phase than during remission (Frequency > 50%). SSNS showed prominent immune markers (CD25, CD11c, CD20, CD1c, CD40) and adhesion molecules (CD49e), while SRNS exhibited higher levels of adaptive immune-related proteins (CD4, CD8, CD19), monocyte markers (CD11c, CD2, CD1c), and adhesion molecules (CD49e, CD146) compared to the SSNS group (Fig. 3E).
Identification of uEV-associated markers to distinguish the active phase of INS disease.
Nineteen core EV-surface markers were identified as potential discriminants between active and inactive INS (Fig. 4a), with eight effectively distinguishing the active phase from CTRL (Fig. 4A). Among them, the three markers that achieved the best discrimination were CD41b, CD105, and CD29 with an AUC > 0.88 (Fig. 4B). These core markers, excluding CD24, CD1c, CD11c, CD25, and CD40, and with the addition of CD19 and CD69 molecules, exhibited a significant positive correlation with proteinuria levels in INS children (Table 2), suggesting an association between kidney damage and uEVs of distinct cellular origin. Seven markers effectively separated SSNS at the onset from CTRL (Supplementary material; Table S1), with CD41b achieving the highest performance (AUC > 0.9). The same cluster of markers except for CD20 and CD42a, but including CD326, CD9, CD133, CD63, CD81, and CD24 differentiated between SSNS Rel and SSNS Rem groups (Supplementary material; Table S2), with CD29 showing the best AUC. When the SRNS group was analyzed, almost half of the markers were differentially expressed between patients with persistent proteinuria or in remission after second-line immunosuppressive treatments (best AUC for CD9, CD19, and CD146) (Supplementary material, Table S3).
Table 2
Urine-EV biomarkers and their correlation to pathological proteinuria. P value < 0.05 was considered significant.
Cellular Origin | Markers | R | p-value |
Immune cell compartment | CD19 | 0.43 | < 0.001 |
CD20 | 0.38 | < 0.001 |
HLA-DR | 0.62 | < 0.001 |
CD69 | 0.29 | < 0.01 |
CD86 | 0.26 | < 0.05 |
Endothelial/platelet activation | CD62p | 0.32 | < 0.01 |
CD41b | 0.71 | < 0.001 |
CD42a | 0.33 | < 0.01 |
CD105 | 0.44 | < 0.001 |
CD142 | 0.23 | < 0.05 |
Adhesion cell activation | CD29 | 0.68 | < 0.001 |
CD326 | 0.59 | < 0.001 |
MCSP | 0.32 | < 0.01 |
Stem/Progenitor cell activation | CD133 | 0.43 | < 0.001 |
SSEA-4 | 0.48 | < 0.001 |
ROR1 | 0.44 | < 0.001 |
Identification of a candidate uEV panel to discriminate patient’s sensitivity to the steroid therapy.
uEVs from SSNS Rel and SRNS patients were compared to identify biomarkers that differentiated significantly the two groups. Markers encompassing both innate (CD11c, CD209, and CD1c) and adaptive immune responses (CD19, CD4, and CD8), along with those involved in angiogenesis/adhesion and stemness processes (CD31, CD44, CD146, and ROR1) were identified (Fig. 5A). CD146 emerged as a significant predictor for SRNS (Fig. 5B). However, the best separation between SRNS and SSNS Rel groups was achieved by the combination of CD19-CD44-CD8 (AUC = 0.87) (Fig. 5C and Table 3).
Table 3
Diagnostic performance of individual, combined uEV markers and proteinuria in steroid sensitivity classification.
Markers | AUC | Sensitivity | Specificity | ACC |
CD19-CD44-CD8 | 0.873 | 0.833 | 0.913 | 0.886 |
CD146 | 0.844 | 0.75 | 0.826 | 0.8 |
CD44 | 0.797 | 0.833 | 0.826 | 0.829 |
CD19 | 0.777 | 0.917 | 0.696 | 0.771 |
CD8 | 0.75 | 0.583 | 0.913 | 0.8 |
uPr/uCr | 0.569 | 0.583 | 0.696 | 0.657 |
Generation of INS expression signature using the serum and urinary EVs from the same patient.
EVs enriched in the serum (sEVs) were also characterized in eighty-three INS patients. NTA analysis revealed significantly higher sEV numbers in SSNS at onset (7.87e11 ± 3.13e11 part/mL, P < 0.01) and relapse compared to CTRL (6.50e11 ± 2.92e11 part/mL vs 3.79e11 ± 1.38e11, P < 0.05) (Fig. 6A). sEVs were also increased in SRNS with persistent proteinuria compared to CTRL (P < 0.001) and SSNS Rem (P < 0.05). No differences in sEV size were observed among groups or after gender-related INS separation (Fig. 6B). CD9 expression in sEVs decreased during the active phase compared to CTRL and inactive INS (P < 0.01 and < 0.05, respectively) (Fig. 6C). Three markers on serum EV surfaces were identified as potential discriminants for active INS (not shown), with lower discriminating power compared to urine in distinguishing INS patients from controls. The combination of serum and urine markers from the same patient was able to reach a statistical correlation as observed in Fig. 6D, with CD3 and CD29 exhibiting significant positive/negative relations in both matrices (R = 0.25, P = 0.047 and R=-0.26, P = 0.039, Pearson correlation) (Fig. 6D). Likewise, the combination of sCD146 and uCD29-uCD41b separated active INS patients from healthy children with higher sensitivity (SE ~ 1) compared to single markers; the combination of CD41b in serum and urine distinguished SSNS at onset from CTRL, and in SSNS, uCD326, and sCD8 effectively differentiated disease activity (Table 4).
Table 4
Diagnostic performance of uEV and sEV markers in pediatric INS between the different analyzed groups.
s-uEVs Markers | AUC | Sensitivity | Specificity | ACC | Groups |
sCD146-uCD29-uCD41b | 1 | 1 | 1 | 1 | Active vs CTRL |
sCD146 | 0.8 | 0.6 | 1 | 0.676 |
uCD29 | 0.914 | 0.867 | 1 | 0.89 |
uCD41b | 0.933 | 0.867 | 1 | 0.89 |
sCD41b-uCD41b | 1 | 1 | 1 | 1 | |
sCD41b | 0.964 | 0.875 | 1 | 0.933 | Onset vs CTRL |
uCD41b | 0.911 | 0.875 | 1 | 0.933 | |
sCD8-uCD326 | 1 | 1 | 1 | 1 | SSNS Rel vs SSNS Rem |
sCD8 | 0.641 | 0.533 | 0.778 | 0.625 |
uCD326 | 0.97 | 0.867 | 1 | 0.917 |