AUTHORS
Pauline Robbe, DPhil, Niamh Appleby2, Grigore-Aristide Gafencu3, Georgeta Ciuban4, Kate Ridout, PhD5, Dimitrios Vavoulis, PhD, MSc, BSc6, Melinda Kormany7, Nick York8, Renata Walewska, MD9, Andrew Pettitt10, Piers E M Patten, MB ChB PhD11, Oliver Miles12, Helen Marr13, Claire Hutchinson14, Nicolas Martinez-Calle, MD15, Scott Marshall, MD, PhD16, Simon Stern17, Murali Kesavan18, Toby A. Eyre, MBChB, DipMedEd, MRCP, MD19, Harriet Walter, MD,PhD,MRCP20, Wendy Harrison21, Charlotte Bomken22, Yasmin Reyal23, Savio Fernandes24, Timothy Coutts25, Anthony Cutts26, Helene Dreau27, Danmei Xu28, Anna Schuh, MD29
AFFILIATIONS
1 University of Oxford, Oxford, United Kingdom
2 University of Oxford, Oxford, United Kingdom
3 University of Oxford, Oxford, United Kingdom
4 University of Oxford, Oxford, United Kingdom
5 University of Oxford, Oxford, United Kingdom
6 University of Oxford, Oxford, United Kingdom
7 University of Oxford, Oxford, United Kingdom
8 Leukaemia Care, Worcester, United Kingdom
9 University Hospitals Dorset NHS Foundation Trust, Bournemouth, United Kingdom
10 University of Liverpool and The Clatterbridge Cancer Centre NHS Foundation Trust, Liverpool, United Kingdom
11 King's College Hospital NHS Foundation Trust, London, United Kingdom
12 Somerset NHS Foundation Trust (Musgrove Park Hospital), Taunton, United Kingdom
13 The Newcastle upon Tyne Hospitals NHS Foundation Trust, Newcastle upon Tyne, United Kingdom
14 University Hospitals Plymouth NHS Trust, Plymouth, United Kingdom
15 Nottingham University Hospitals NHS Trust, Nottingham, United Kingdom
16 South Tyneside and Sunderland NHSFT, Sunderland, United Kingdom
17 Epsom and St Helier University Hospitals NHS Trust, London, United Kingdom
18 Roche, Welwyn Garden City, United Kingdom
19 Oxford University Hospitals NHS Foundation Trust, Oxford, United Kingdom
20 University Hospitals of Leicester NHS trust and University of Leicester, Leicester, United Kingdom
21 James Paget University Hospitals NHS Foundation Trust, Great Yarmouth, United Kingdom
22 Northumbria Healthcare NHS Foundation Trust, Newcastle upon Tyne, United Kingdom
23 St George's University Hospitals NHS Foundation Trust, London, United Kingdom
24 Dudley Group NHS Trust, Dudley, United Kingdom
25 University of Oxford, Oxford, United Kingdom
26 University of Oxford, Oxford, United Kingdom
27 University of Oxford, Oxford, United Kingdom
28 University of Oxford, Oxford, United Kingdom
29 University of Oxford, Oxford, United Kingdom
ABSTRACT
Different scores using clinical and laboratory parameters have been proposed to predict progression risk of high-count monoclonal B-cell lymphocytosis (HC-MBL) and early-stage chronic lymphocytic leukaemia (CLL; Stage A), but all these models were derived predominantly from retrospective cohorts, and prospective validation in patients presenting through routine diagnostic pathways remains limited. To motivate change in routine practice, strong independent assessment of these classifications is required. Here, we performed a head-to-head comparison of four contemporary prognostic models IPS-E, CLL International Prognostic Index (CLL-IPI), CLL newly diagnosed patients without need of treatment (CLL-WONT), and Belgian Prognostic Index (CLL-BPI) in a prospective clinical study (n = 306), Oxford Pre-cancerous Lymphoproliferative Disorders (OxPLoreD), that recruited patients within three years of diagnosis from 15 centres across the United Kingdom between 2019 to 2023. We evaluated their ability to predict time to first treatment/time to progression (TTFT/TTP) for HC-MBL and early-stage CLL.
We calculated four scores for 306 individuals included in this study. Classification was ambiguous for 4 IPS-E, 33 CLL-IPI, 42 CLL-WONT and 28 CLL-BPI cases because of missing IGHV status (n=4), β2-microglobulin (n=30) or lactate dehydrogenase (LDH, n=58), despite these variables being required by the study protocol. (1) Overall performances were compared using concordance index (C-index) and agreement was assessed using quadratic weighted Cohen's κ. Prediction of TTFT/TTP was evaluated using Kaplan-Meier survival analysis and Cox proportional hazards regression. (2) Using IPS-E as reference, we quantified individuals classified to a different risk level (upgraded to a higher risk or downgraded to a lower risk) with other models and evaluated which biomarkers influenced this change in classification by multivariate logistic regression models and ranking by LR χ² statistics. (3) Clinical outcome of reclassified individuals was further investigated by comparing TTFT/TTP between unchanged and patients reclassified by Kaplan-Meier survival analysis and Cox proportional hazards regression.
The 4 classifications presented progressively shorter TTFT/TTP with increasing risk category (log-rank P<0.0001). Despite important differences in model complexity and constituent variables, IPS-E, CLL-WONT and CLL-BPI demonstrated comparable overall prognostic performance (C-index 0.750, 0.737, 0.735, respectively), whereas CLL-IPI consistently showed inferior discrimination (0.663). With IPS-E as the reference, agreement was poor with CLL-IPI (κ=0.29) and moderate with CLL-WONT and CLL-BPI (κ=0.59, 0.57, respectively). IPS-E achieved the largest separation between low- and high-risk groups (HR = 21.0 vs 18.5 for the second best which was the “Very high-risk" group in CLL-WONT, FDR < 0.01).
The IPS-E showed excellent prediction of low-risk. Relative to IPS-E, the proportions of patients reassigned to different risk categories was substantial: 55% for CLL-IPI, 37% for CLL-WONT and 38% for CLL-BPI. For CLL-IPI, despite TP53 status being the only significant markers associated with reclassification, reassigned patients did not demonstrate significantly different TTFT/TTP within the original IPS-E risk groups (Log rank p>0.1), suggesting that incorporation of TP53 did not improve prediction of treatment initiation beyond IPS-E in this early-stage cohort. Among patients classified as high risk by IPS-E, those downgraded by CLL-WONT (n=43) or CLL-BPI (n=40) had significantly longer TTFT/TTP than those remaining high risk: progression occurred in 18.6% downgraded vs 58.8% unchanged patients with CLL-WONT (FDR=0.01) and in 25.0% vs 51.9% with CLL-BPI (FDR=0.022). Patients downgraded by CLL-WONT were younger, had lower LDH and lacked del(11q) (FDR<0.01), whereas those downgraded by CLL-BPI had mutated IGHV (FDR<0.001). For IPS-E intermediate-risk patients (n=127), 50% were downgraded and 10% upgraded by CLL-IPI, 22% downgraded and 12% upgraded by CLL-WONT, and 40% downgraded and 13% upgraded by CLL-BPI. However, none of these reclassifications was associated with significantly different TTFT/TTP (all P>0.05).
The IPS-E is a simple model with excellent low-risk prediction. Incorporation of selected additional clinical and biological variables identifies a clinically distinct subgroup within the IPS-E high-risk category with substantially more favourable outcomes.
DISCLOSURES
Pauline Robbe, DPhil: No relevant conflicts of interest to declare.