Defining a New “Mayo Stage IIIc” Ultra Poor Risk Category in Systemic AL Amyloidosis: Incorporating Echocardiographic Global Longitudinal Strain to the European Modified Mayo Staging System

Background: Systemic light chain (AL) amyloidosis is a protein misfolding disorder leading to progressive organ failure. Risk stratification has been robustly performed using cardiac biomarkers (troponin T and NT-proBNP) and light chain burden (difference in involved and uninvolved free-light chain, dFLC). Recently, myocardial global longitudinal strain (GLS) has also been shown to be prognostic. We aimed to develop a new staging system incorporating GLS and to identify those with the poorest risk.

Methods: Patients enrolled in a prospective ALchemy study at the UK National Amyloidosis Centre treated with a bortezomib-based regimen from 2010–2022 were analysed. Ethic approval was obtained (REC: 09/H0715/58). Diagnosis of AL amyloidosis was confirmed centrally by histology and typed with immunohistochemistry or mass spectrometry, or if not available, for patients with biopsy confirmed amyloidosis and cardiac involvement alone, if they also had a negative DPD-Tc99m bone scan. Overall survival (OS) estimates were generated using the Kaplan-Meier method and groups compared using Cox regression. Restricted mean survival time was performed where proportional hazards were violated. Statistical analyses were conducted using STATA v18 (STATAcorp, Texas).

Results: 1390 patients (828 male, 562 female) were treated between 2010-2019 with a first-line bortezomib-based regimen. Median age at diagnosis was 66 years (range 29-89) with 2 organs (range 1-5) involved: 904 (65%) cardiac and 955 (70%) renal involvement. Median dFLC was 189mg/L (range:0-15989), NT-proBNP 1565ng/L (range 12-93602), high-sensitivity troponin T 54ng/L (range 0-742), GLS -14.4% (range -3--28). At baseline, using the European modified Mayo classification, 233, 481, 462 and 210 patients were classified as having stage I, II, IIIa, IIIb disease, respectively. Ninety percent were treated with bortezomib-cyclophosphamide-dexamethasone and 6% with daratumumab-bortezomib combination. At a median follow-up of 72 months (95% CI 68-74), estimated median OS was 88 months (95% CI 72-112), 5-year OS was 55% (95% CI 52-57). On multivariable analysis factors independently predictive of OS were: NT-proBNP (>332 v <332ng/L HR 2.40 [95% CI 1.26-4.60], p=0.008; >8500 v <332ng/L HR 4.68 [95% CI 2.31-9.48], p<0.001), troponin >50ng/L (HR 1.92 [95% CI 1.38-2.69], p<0.001) and GLS >-9% (HR 1.84 [95% CI 1.35-2.50], p<0.001); dFLC >180mg/L (HR 1.12 [95% CI 0.86-1.47], p=0.41) was not independently predictive of OS. There was no interaction between GLS and NT-proBNP (p=0.71).

On six-month landmark analysis, the dFLC-180mg/L threshold had a lower sensitivity in predicting 1-year OS for survivors with complete haematological response (CR), compared with without CR (62% vs 36%). In those <CR, a baseline dFLC<180mg/L predicted a 5 month longer 5-year restricted mean survival time compared with dFLC >180mg/L (95% CI 1.4-7.9, p=0.005), however did not predict significantly different survival in patients who achieved a CR (95% CI -2.1-6.0, p=0.35).

A new model incorporating GLS into the traditional European modified Mayo score identified an ultra-high risk ‘stage IIIc’ (GLS >-9%, NT-proBNP >8500ng/L, troponin >50ng/L) with median OS of 5 months (95% CI:3-11). Hazard ratios for stage II, IIIa-b, IIIc were 2.25 (95% CI 1.22-4.15), 4.91 (95% CI 2.79-8.68), 14.59 (95% CI 7.55-28.17) respectively (n=650, Harrel’s C 0.67). Internal validation by bootstrapping 1000 samples confirmed model robustness (shrinkage factor 0.94). Specificity for 6-month OS for stage IIIc v IIIb was 97% v 90% with 6-month OS 41% (95% CI 26-56) v 55% (95% CI 48-62).

Conclusion: In the bortezomib-treatment era, the addition of GLS to the European modified Mayo staging system identifies those at highest risk of death, ‘stage IIIc’. This model requires external validation particularly in the new treatment era.