Cohen, Oliver C.

Abstract Background: Systemic AL amyloidosis is an incurable disorder caused by misfolded light chain fibrils. Survival is determined by the extent of cardiac involvement and response to therapy. Mayo 2012 (stage I-IV) and European modified Mayo 2004 (stage I-IIIB) prognostic scores were developed in a historic treatment era, prior to widespread bortezomib and daratumumab use. These no longer discriminate those with the poorest outcomes. We presented a new model sub-stratifying IIIB into a high risk, ‘IIIC’ group, incorporating longitudinal strain (LS) to the European modified score (IIIC: LS >-9% + NT-proBNP >8500ng/L + high-sensitivity troponin T [hs-TnT] >50ng/L) in the bortezomib-era (2015-2019) last year (Khwaja et al, ASH 2024). Here, we present an independent external validation of the IIIC model in the modern treatment era. Methods: The ‘IIIC model’ was derived from a large real-world cohort of adults from the UK National Amyloidosis Centre (2015-2019). Performance was externally validated across Europe (Greece, Italy, Netherlands, Switzerland), USA (2015-2024) and UK (2020-2024). Survival estimates were generated from diagnosis to death or last follow-up using the Kaplan-Meier method and groups compared using Cox regression (using Stata v18). Stage IIIB and IIIC were defined as follows IIIB: LS <-9%, NT-proBNP >8500ng/L, hs-TnT >50ng/L and IIIC: LS >-9%, NT-proBNP >8500ng/L, hs-TnT >50ng/L, respectively. Stage I, II and IIIa were defined as previously described in the European modification of Mayo 2004 staging system. Results: 2493 patients were analysed from 2015-2024 (derivation, n=573; validation: Europe, n=374, USA n=458, UK, n=1088). In the derivation cohort, LS ≥-9% and standard cardiac biomarkers (NT-proBNP 332ng/L and 8500ng/L and hs-TnT>50ng/L) were selected as independent prognostic factors in a multivariable model. In the European and UK validation cohort (n=1462), median age at diagnosis was 67 years (range 35-95) with 1014 (69%) cardiac and 869 (60%) renal involvement. Median difference in involved and uninvolved free light chain (dFLC) was 181 mg/L (range 0-24933), NT-proBNP 1911ng/L (range 42-70000), hs-TnT 45ng/L (range 3-1221) and LS -14.7% (range -3- -28). In the US cohort median NT-proBNP was 1764 (range 13-71000) and hs-TnT 44 (6-687). Applying the new model staging, the patient stages were across Europe and UK were stage I: 224 (15%), II: 627 (43%), IIIA: 396 (27%), IIIB: 139 (10%) and IIIC: 76 (5%), respectively. 592 (41%) were treated with first-line daratumumab-based regimens. In the US dataset stage I: 103 (22%), II: 176 (38%), IIIA: 131 (29%), IIIB: 48 (10%) and IIIC: 16 (3%), respectively. 234 (51%) were treated with first-line daratumumab-based regimens. At a median follow-up of 30 months (95% CI 28-32) in the European and UK validation cohort, estimated median overall survival (OS) was not reached and 2-year OS was 74% (95% CI 72-77). Median survival for stages I-II, IIIA, IIIB and IIIC were not reached (NR), 67, 31 and 10 months, respectively (p<0.001). Hazard ratios for stage II, IIIA, IIIB, IIIC were 2.01 (95% CI 1.32-3.07), 3.51 (95% CI 2.31-5.35), 5.67 (95% CI 3.60-8.91) and 11.12 (95% CI 6.93-17.87), respectively (p<0.001). This was confirmed in the US cohort with a median survival for respective stages were NR, NR, NR, 8 and 2 months (p<0.001). External validation in the European and UK cohorts exhibited good predictive performance: 12-month calibration slope was 1.00 (95% CI 0.81-1.20, p=0.98), Harrel's C was 0.90 (95% CI 0.76-1.00), Royston's D 1.07, R2D 0.21 (95% CI 0.15-0.28). In 1462 patients with dataset available, stage IIIC discriminated the poorest outcome despite a greater proportion treated with first-line daratumumab v IIIB in the validation cohort (59% v 40%, p=0.008). 12-month OS for IIIC v IIIB was 44% (95% CI 32-55) v 69% (95% CI 60-76), specificity for 12-month OS 97% v 92% and diagnostic odds ratio 6.62 v 2.36. Conclusions: This study validates the new staging system from stage I-IIIC with robust identification of an ultra-poor prognostic group (Stage IIIC) in a contemporarily treated patient cohort. Despite novel therapies, the OS remains <12 month in the IIIC group and also poor in stage IIIB with median OS 31 months compared to a median of >5 years for all other stages. This staging system can be considered as a new standard of care for risk stratification of AL amyloidosis. Novel approaches are required for patients with advanced risk.

ImportanceKidney light chain (AL) amyloidosis is associated with a risk of progression to kidney replacement therapy (KRT) and death. Several studies have shown that a greater reduction in proteinuria following successful anticlonal therapy is associated with improved outcomes.ObjectiveTo validate graded kidney response criteria and their association with kidney and overall survival (OS).Design, Setting, and ParticipantsThis retrospective, multicenter cohort was conducted at 10 referral centers for amyloidosis from 2010 to 2015 and included patients with kidney AL amyloidosis that was evaluable for kidney response and who achieved at least hematologic partial response within 12 months of diagnosis. The median follow-up was 69 (54-88) months. Data analysis was conducted in 2023.ExposureFour kidney response categories based on the reduction in pretreatment 24-hour urine protein (24-hour UP) levels: complete response (kidCR, 24-hour UP ≤200 mg), very good partial response (kidVGPR, >60% reduction in 24-hour UP), partial response (kidPR, 31%-60% reduction), and no response (kidNR, ≤30% reduction). Kidney response was assessed at landmark points (6, 12, and 24 months) and best kidney response.Main Outcomes and MeasuresCumulative incidence of progression to KRT and OS.ResultsSeven-hundred and thirty-two patients (335 women [45.8%]) were included, with a median (IQR) age of 63 (55-69) years. The median (IQR) baseline 24-hour proteinuria and estimated glomerular filtration rate was 5.3 (2.8-8.5) g per 24 hours and 72 (48-92) mL/min/1.73m2, respectively. In a competing-risk analysis, the 5-year cumulative incidence rates of progression to KRT decreased with deeper kidney responses as early as 6 months from therapy initiation (11%, 12%, 2.1%, and 0% for kidNR, kidPR, kidVGPR, and kidCR, respectively; P = .002) and were maintained at 12 months and 24 months and best kidney response. Patients able to achieve kidCR/kidVGPR by 24 months and at best response had significantly better OS compared with kidPR/kidNR. Kidney progression, defined as a 25% or greater decrease in estimated glomerular filtration rate, was associated with cumulative incidence of progression to KRT and OS.Conclusions and RelevanceThe results of this cohort study suggest that graded kidney response criteria offers clinically and prognostically meaningful information for treating patients with kidney AL amyloidosis. The response criteria potentially inform kidney survival based on the depth of reduction in 24-hour proteinuria levels and demonstrate an OS advantage for those able to achieve kidCR/kidVGPR compared with kidPR/kidNR. Taken together, achievement of at least kidVGPR by 12 months is needed to ultimately improve kidney and patient survival.