ALBOM Study
What it is
ALBOM is our study characterizing Human Serum Albumin (HSA) isoforms as biomarkers of chronic liver disease (CLD) staging, using top-down LC-HR-MS with absolute quantification. The study uses an internal calibration approach (equine myoglobin internal standard) to quantify 10 intact HSA isoforms in clinical plasma samples across the full spectrum of liver fibrosis.
Publication
“Human Serum Albumin Profiling by Top-Down Analysis Enables Multi-Class Liver Fibrosis Staging: A Cross-Platform Validation Study” El Balkhi S*, Berrah R*, Sauvage FL, Le Du L, Rahali MA, Lakis R, Marquet P, Saint-Marcoux F, Loustaud-Ratti V, Carrier P (*equal contribution) Published in Scientific Reports (2026) — DOI 10.1038/s41598-026-57614-y; accepted 8 June 2026; Open Access (CC BY-NC-ND 4.0) Full wiki analysis: el-balkhi-2025
Study design
- Type: Prospective, single-centre, cross-sectional
- Period: Jan 2021 – Jan 2023, CHU Limoges (Hepatology Dept.)
- Sample type: Human plasma (residual from routine care)
- n: 172 CLD patients (F0/F1–F4_C) + 82 healthy controls
Cohort breakdown
| Stage | n | % total |
|---|---|---|
| Controls | 82 | — |
| F0/F1 | 36 | 20.9% |
| F2 | 23 | 13.0% |
| F3 | 30 | 17.0% |
| F4_A (Child-Pugh A) | 37 | 21.5% |
| F4_B (Child-Pugh B) | 26 | 15.1% |
| F4_C (Child-Pugh C) | 20 | 11.6% |
Etiologies by stage (Table 1)
| Etiology | F0/F1 | F2 | F3 | F4_A | F4_B | F4_C |
|---|---|---|---|---|---|---|
| MASH | 13 | 13 | 22 | 12 | 3 | 0 |
| ALD | 1 | 0 | 0 | 9 | 16 | 14 |
| HBV | 10 | 1 | 2 | 2 | 0 | 0 |
| HCV | 2 | 3 | 1 | 5 | 0 | 0 |
| AIH | 3 | 2 | 4 | 1 | 0 | 1 |
| ALD+MASH | 0 | 0 | 0 | 2 | 5 | 4 |
⚠️ MASH dominates early stages; ALD dominates F4_B/C (~85% alcohol-related in decompensated cirrhosis)
Staging methods used per stage (Table 2)
| Stage | FibroScan | Biopsy | Both |
|---|---|---|---|
| F0/F1 | 35 (97%) | 4 (11%) | 3 |
| F2 | 23 (100%) | 5 (22%) | 5 |
| F3 | 28 (93%) | 12 (40%) | 10 |
| F4_A | 25 (68%) | 5 (13%) | 2 |
| F4_B | 3 (11%) | 6 (23%) | 1 |
| F4_C | 4 (20%) | 8 (40%) | 1 |
FibroScan largely unavailable for F4_B/C → clinical consensus or biopsy required
Analytical method
- Method: Validated top-down LC-QTOF-MS (published as Lakis et al. [ref 22])
- Sample prep: 1:50 dilution in 0.9% NaCl + equine myoglobin (4 g/L) as IS
- Column: C4 reverse-phase; gradient elution
- Ionization: ESI-QTOF
- Deconvolution range: 66,000–68,000 Da
- Platforms: Bruker timsTOF Pro2 (P1) + Sciex TripleTOF 5600+ (P2)
- Quantification: Absolute (g/L) — internal calibration to equine Mb IS; high-MW glycated isoforms calibrated from related species slope
Main targets — 10 isoforms quantified
| Isoform | PTM(s) | Δmass (Da) | Disease pattern |
|---|---|---|---|
| Native HSA | None (Cys34 free) | 0 | Progressive ↓; precipitous F4_B/C |
| HSA-DA | N-terminal truncation (–Asp) | −115 | ↓ all stages vs controls |
| HSA-L | C-terminal truncation | variable | No significant change |
| HSA+CYS | Cysteinylation Cys34 | +119 | Biphasic — peak F4_A, ↓ F4_B/C |
| HSA+SO₃H | Sulfonylation (irreversible) | +48 | ↓ F4_B/C |
| HSA+GLYC | Glycation (mono, Lys) | +162 | Biphasic — peak F3/F4_A, ↓ F4_B/C |
| HSA-DA+CYS | Truncation + cysteinylation | −115 +119 | ↓ F4_B/C |
| HSA+CYS+GLYC | Cysteinylation + glycation | +281 | Biphasic — peak F3 (1.9 g/L) |
| HSA+2GLYC | Double glycation | +324 | Biphasic — peak F4_A |
| HSA+CYS+2GLYC | Double glycation + cysteinylation | +443 | Progressive ↑ — end-stage marker |
Key results
Native HSA performance
Controls: 12.2 g/L → F4_B: 4.1 g/L, F4_C: 4.2 g/L AUC: F0/F1 = 0.67, F4_B = 0.99, F4_C = 0.89
Best individual diagnostic ratios (normalized to native)
| Ratio | Stage | Sensitivity | Specificity |
|---|---|---|---|
| HSA+CYS/Native | F4_C vs controls | 65% | 99% |
| HSA+GLYC/Native | F4_B vs controls | 85% | 100% |
| HSA+GLYC/Native | F4_C vs controls | 70% | 99% |
| HSA+CYS+GLYC/Native | F4_B vs controls | 77% | 99% |
OrdinalForest classifier performance
| Platform | n_test | QWK | 95% CI |
|---|---|---|---|
| P1 Bruker timsTOF Pro2 | 46 | 0.862 | 0.735–0.923 |
| P2 Sciex TripleTOF 5600+ | 49 | 0.916 | 0.822–0.964 |
| FIB-4 (comparator) | — | 0.188–0.229 | — |
Overall triage accuracy: 81.5% (LC-TOF + Clinical) vs 59.3% (FIB-4) → +26 pp
Cross-platform reproducibility
- McNemar p = 0.149 (no significant difference between instruments)
- Jaccard similarity index of errors = 0.696 (~70% shared misclassifications → biologically driven)
Status
- Published — Scientific Reports 2026, DOI 10.1038/s41598-026-57614-y (accepted 8 June 2026, Open Access)
- Multicentric validation: MALAHBAR study (NCT06318949) — ongoing; 560 patients enrolled / 8 French CHUs (target >700); SEB test + isoform signature validation expected 2027
Connections to CQFD-PTM
ALBOM data was generated using the CQFD-PTM pipeline — ALBOM is the clinical validation study for the pipeline. The absolute quantification method (Lakis et al.) feeds directly into the pipeline’s quantification module.
Next steps / open questions
- Multicentric MALAHBAR study (560 patients / 8 CHUs, target >700) to confirm predictive capacity and externally validate across centres — expected 2027
- Etiology-specific isoform analysis (MASH vs ALD vs HCV subtypes)
- Optimization of Platform 1 baseline correction for high-mass glycated species (>67,500 Da)
- Potential downstream development of targeted LC-MRM-MS or immunoassay for CYS/Native and GLYC/Native ratios
- Formal proficiency testing scheme for multicenter deployment