2026 ASH®
Annual Meeting & Exposition December 12-15, 2026 ▪ New Orleans, Louisiana

Paper Details

Abstracts, authors, affiliations, and disclosures.

Saturday, December 12, 2026
09:30 a.m. - 11:00 a.m.
MCCNO - New Orleans Theatre A
Technology and Procedures, Assays, Molecular Testing, Omics Technologies, Pathology, Computational Biology, Profiling, Artificial Intelligence (AI), Machine Learning, Biological Processes, Genomics, Immune Mechanism, Diseases, Lymphoid Malignancies, Lymphomas, Measurable Residual Disease, Hodgkin Lymphoma, Non-Hodgkin Lymphoma, B Cell Lymphoma, Aggressive Lymphoma, Research, Fundamental Science, Translational Research, Clinical Research, Real-World Evidence, Study Population, Human, Adult, Pediatric, Young Adult
PAPER

69 Frenchconnect, the first nationwide multiplatform network for harmonized ctDNA monitoring in lymphoma: Prognostic validation of interim phased variants in 292 real-world patients with LBCL from the realysa cohort

Alexis Claudel, MD, Elodie Bohers, PhD2, Cathy Quelen, PhD3, Gauthier Decool, MD4, Sarah Huet, PhD, PharmD5, Aurélie Caillault-Venet6, Kaddour Chabane7, Nesrine Trabelsi8, Maël Louis9, Bruno Tesson10, Fanny Cherblanc11, Aurélien Belot12, Cedric Portugues13, Laetitia Vercellino, MD14, Constance Enriquez15, François Lemonnier Sr, MD, PhD16, Pierre Sujobert, MD PhD17, Vincent Camus, MD, PhD18, Fabrice Jardin, MD, PhD19, Camille Laurent, MD, PhD20, Emmanuel Bachy I, MD, PhD21, Hervé Ghesquières, MD, PhD22, Bertrand Nadel23, Pierre-Julien Viailly24, Marie-Helene Delfau25
1 GHU Henri Mondor, Assistance Publique–Hôpitaux de Paris, Créteil, France
2 Centre Henri Becquerel, Rouen, France
3 Université de Toulouse, Toulouse, France
4 CHU de Lille, Lille, France
5 Hospices Civils de Lyon, Pierre-Bénite, France
6 CHU de Lille, Lille, France
7 Hospices Civils de Lyon, Pierre-Bénite, France
8 GHU Henri Mondor, Assistance Publique–Hôpitaux de Paris, Créteil, France
9 Institut Carnot, Paris, France
10 LYSARC, Pierre-Bénite, France
11 LYSARC, Pierre-Bénite, France
12 LYSARC, Pierre-Bénite, France
13 LYSARC, Pierre-Bénite, France
14 Hôpital Saint-Louis, Assistance Publique–Hôpitaux de Paris, Paris, France
15 Institut Carnot, Paris, France
16 GHU Henri Mondor, Assistance Publique–Hôpitaux de Paris, Créteil, France
17 Hospices Civils de Lyon, Pierre-Bénite, France
18 Centre Henri Becquerel, Rouen, France
19 Centre Henri Becquerel, Rouen, France
20 Université de Toulouse, Toulouse, France
21 Hospices Civils de Lyon, Pierre-Bénite, France
22 Hospices Civils de Lyon, Pierre-Bénite, France
23 Institut Carnot, Paris, France
24 Centre Henri Becquerel, Rouen, France
25 GHU Henri Mondor, Assistance Publique–Hôpitaux de Paris, Créteil, France
Introduction

Circulating tumor DNA (ctDNA) is a promising early response biomarker in aggressive large B-cell lymphoma (LBCL), but its standardization across centers and its added value over interim PET remain incompletely established in real-world multiplatform settings.

The FrenchConnect network harmonizes ctDNA workflows across French platforms, from pre-analytical processing to bioinformatics, for identifying and monitoring markers including phased variants (PVs).

To validate this framework, 292 REALYSA patients treated with an R-CHOP-like regimen were studied at diagnosis (D0) and before cycle 3 (C3D1).

Methods

REALYSA is a nationwide real-world cohort of 6,015 lymphoma patients from 35 centers. We selected 292 patients with diffuse large (n=247), high-grade (n=21) or primary mediastinal (n=24) B-cell lymphoma and plasma at both D0 and C3D1; patients progressing before C3D1 were not sampled.

cfDNA was sequenced on five platforms using three capture chemistries (Agilent, Twist, Qiagen) with the same 130-kb panel covering 302 PV-enriched regions, and analyzed with one centralized pipeline (GenerateReports).

Informative UMIs were unique molecules spanning both positions of doublets identified in baseline plasma; residual burden was their mutant fraction, and the per-sample limit of detection (LoD) the burden detected with 95% probability (Krupka, JCO 2026). Because a threshold must be attainable on every platform, measurable residual disease (MRD) at C3D1 was pre-specified as a burden of 1E-4 or more, reported only for samples reaching that LoD. Major molecular response (MMR) was a 2.5-log reduction or undetectable disease (Kurtz, JCO 2018). Interim PET was performed a median of 47 days after baseline and Deauville-scored (positive, 4-5). Survival analyses were landmarked at C3D1.

Results

Of 292 patients, 274 (93.8%) had technically evaluable baseline cfDNA (median 94 PV doublets, IQR 32-166), with a median burden of 2.17% (IQR 0.68-5.78) associated with Ann Arbor stage and elevated LDH (both p<0.001) and with total metabolic tumor volume (41% SUVmax; Spearman rho=0.525; n=181).

At C3D1, 1E-4 was the lowest threshold attainable in at least 80% of samples on every platform; 240 samples (87.6%) reached it (median LoD 8.3E-6, 12-fold below the threshold; IQR 3.3E-6-2.0E-5), and analyses were restricted to these. Excluded samples had fewer doublets (median 6), lower baseline burden (median 0.48%) and similar EFS (HR 0.77 [0.35-1.68]).

Of these, 63 (26.2%) were MRD-positive (median 0.041%, IQR 0.017-0.170). After a median follow-up of 44 months (reverse Kaplan-Meier; 66 EFS, 45 TTP, 62 PFS and 37 OS events), MRD positivity was associated with shorter event-free survival (EFS; 36-month 58.6% vs 81.3%; HR 2.42 [1.48-3.95]), time to progression (TTP; HR 3.15 [1.75-5.65]), PFS (HR 2.54 [1.54-4.20]) and overall survival (OS; HR 2.61 [1.37-4.99]; all p<0.005), and remained so after adjustment for IPI 4-5 (n=209 with available IPI; EFS HR 2.23 [1.35-3.69], p=0.002; OS HR 2.02 [1.04-3.92], p=0.038). Disease detectable above the per-sample LoD but below the threshold was not associated with EFS (HR 0.82 [0.40-1.70]), unlike burden above it (HR 2.29 [1.36-3.88]). MRD positivity was comparable across platforms (19-35%), with no platform effect unadjusted (p=0.31) or adjusted for baseline burden (p=0.62), and stratified estimates were unchanged (EFS HR 2.35 [1.41-3.89]).

In detectable samples, the mean log10 reduction (2.51) converged on the pre-specified 2.5-log threshold. MMR was achieved in 177 (73.8%), agreed with the burden-based call (kappa 0.72), and its absence predicted shorter EFS (HR 1.91 [1.16-3.14]) and OS (HR 2.33 [1.21-4.47]; both p=0.011). Among 155 patients with interim PET (29 EFS events), the 21 (13.5%) positive by both had shorter EFS (HR 3.54 [1.61-7.78]; p=0.002), and ctDNA remained prognostic in a bivariate model with PET (HR 2.21 [1.05-4.66]; likelihood-ratio p=0.04).

Conclusion

This study demonstrated the feasibility of harmonized cfDNA analysis across French centers, with a burden-based MRD readout transferable across platforms, and prognostic separation confined to burden above the threshold. MRD positivity at C3D1 was associated with clinical outcomes independently of the IPI, and added prognostic information to interim PET, providing the analytical basis for phased-variant ctDNA monitoring in early response-adapted strategies in LBCL. This network is now prospectively deployed in the LYSA RT4 study (NCT07514715).

Alexis Claudel, MD: No relevant conflicts of interest to declare.