Table 3.
Serum proteins selected for 6MWT regression model and performance in correlations
| Target | UniProt | Entrez gene symbol | Elastic net selections (of 100 cross-validations) | Correlation serum protein versus clinical severity (P-value) | Biopsy mRNA normal versus DMD fold change (P-value) | Correlation mRNA versus histopathology severity score (P-value) |
|---|---|---|---|---|---|---|
| Angiopoietin-1 | Q15389 | ANGPT1 | 98 | 0.53 (0.005) | −1.4 (ns) | 0.15 (ns) |
| CBG | P08185 | SERPINA6 | 98 | 0.44 (0.018) | 1.8 (0.017) | 0.35 (ns) |
| CTAP-III | P02775 | PPBP | 98 | 0.0044 (ns) | n/a | n/a |
| ERBB4* | Q15303 | ERBB4 | 98 | 0.52 (0.005) | 6.2 (9.5e−06) | 0.66 (0.004) |
| TGF-b2 | P61812 | TGFB2 | 98 | 0.44 (0.018) | −1.7 (0.001) | 0.018 (ns) |
| 4EBP2 | Q13542 | EIF4EBP2 | 97 | 0.42 (0.021) | −1.1 (ns) | 0.54 (0.013) |
| AMNLS | Q9BXJ7 | AMN | 97 | 0.22 (ns) | −2.3 (ns) | 0.35 (ns) |
| EPHA3* | P29320 | EPHA3 | 97 | 0.38 (0.04) | −1.1 (ns) | −0.033 (ns) |
| Nidogen | P14543 | NID1 | 97 | 0.16 (ns) | −4.8 (5.2e−12) | −0.52 (0.016) |
| SOD* | P00441 | SOD1 | 95 | −0.32 (ns) | 1.2 (0.039) | 0.58 (0.008) |
| PDGF-AA | P04085 | PDGFA | 93 | 0.23 (ns) | 1.2 (ns) | 0.43 (ns) |
| ON | P09486 | SPARC | 74 | 0.28 (ns) | −3.6 (5.4e−14) | −0.54 (0.013) |
| BCL6 | P41182 | BCL6 | 71 | 0.31 (ns) | 1 (ns) | 0.39 (ns) |
| CNTF | P26441 | CNTF | 71 | 0.48 (0.013) | 1.2 (0.039) | 0.64 (0.004) |
| Ubiquitin+1 | P62979 | RPS27A | 71 | −0.35 (0.051) | −1.5 (9.3e−07) | −0.6 (0.007) |
| MAPK5 | Q8IW41 | MAPKAPK5 | 66 | 0.32 (ns) | 1.4 (2.3e−05) | 0.12 (ns) |
| PLPP* | Q96GD0 | PDXP | 61 | −0.19 (ns) | 1.4 (0.039) | −0.28 (ns) |
List of proteins identified as explanatory variables from multivariate model for 6MWT distance along with repeatability of signal in multivariate model cross-validations, Spearman correlation of log-transformed mRNA values with severity (severe, moderate or mild histological severity in that order from DMD and BMD samples), fold change from mRNA (healthy controls, numerator versus DMD subjects, denominator) and Spearman correlation of log-transformed SOMAscan
values with 6MWT distance. All P-values were adjusted using Benjamini–Hochberg (FDR) correction alongside the set of proteins for TTSTAND velocity with any P-value higher than 0.05 noted as ns. No ‘_at’ probeset was found for CTAP-III. Starred rows are proteins that are shared between the TTSTAND and 6MWT models.