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

Paper Details

Abstracts, authors, affiliations, and disclosures.

Monday, December 14, 2026
04:30 p.m. - 06:00 p.m.
MCCNO - New Orleans Theatre C
Treatment Considerations, Combination Therapy, Therapy Sequence, Diseases, Lymphoid Malignancies, Lymphomas, Measurable Residual Disease, Hodgkin Lymphoma, Research, Translational Research, Clinical Research, Clinical Trials, Epidemiology, Study Population, Human, Adult
PAPER

1022 Dynamic circulating tumor DNA and an integrated radiogenomic risk model predict outcomes in pediatric high-risk Hodgkin lymphoma: A liquid-biopsy study of COG AHOD1331

Shuyu Shi, PhD, Lianna J Marks, MD2, Hrishikesh Srinagesh, MD, MSCR3, Lena Osswald, MD4, Stefan Alig, MD5, Maya Strohband, BS6, Xiaoman Kang7, Mari Olsen8, Ragini Adams9, Chih Long Liu, PhD10, Feng Tian, PhD11, Lisa G Roth, MD12, Kara M. Kelly, MD13, Robert Tibshirani, PhD14, Tyler J Bradshaw15, Maximilian Diehn, MD, PhD16, Steve Yoon-Ho Cho, MD17, Sharon M Castellino Jr, MD, MSc18, Ash A. Alizadeh, MD, PhD19
1 Stanford University, Stanford, California, United States
2 Stanford University, Stanford, California, United States
3 Stanford University, Stanford, California, United States
4 Stanford University, Stanford, California, United States
5 Stanford University, Stanford, California, United States
6 Stanford University, Stanford, California, United States
7 Stanford University, Stanford, California, United States
8 Stanford University, Stanford, California, United States
9 Stanford University, Stanford, California, United States
10 Stanford University, Stanford, California, United States
11 Stanford University, Stanford, California, United States
12 NYU Langone and Perlmutter Cancer Center, New York, New York, United States
13 Roswell Park Comprehensive Cancer Center, Buffalo, New York, United States
14 Stanford University, Stanford, California, United States
15 University of Wisconsin-Madison, Madison, Wisconsin, United States
16 Stanford University, Stanford, California, United States
17 The University of Texas MD Anderson Cancer Center, Houston, Texas, United States
18 Emory University, Aflac Cancer and Blood Disorders Center, Children’s Healthcare of Atlanta, Atlanta, Georgia, United States
19 Stanford University, Stanford, California, United States

Background: Children with high-risk classic Hodgkin lymphoma (cHL) are treated with response-adapted therapy guided by interim FDG-PET, yet interim PET imperfectly identifies who will relapse. The prognostic value of interim PET has been challenged in recent trials incorporating antibody-drug conjugates such as Brentuximab vedotin (Bv) and immune checkpoint inhibitors. Circulating tumor DNA (ctDNA) enables noninvasive, quantitative molecular response assessment. Extending our baseline genomic landscape of COG AHOD1331, we tested whether on-treatment (interim, cycle 3 day 1 [C3D1]) ctDNA response predicts progression-free survival (PFS) and improves risk stratification beyond interim PET and baseline tumor burden.

Methods: Patients 2-21 yrs. with newly diagnosed high-risk cHL enrolled on COG AHOD1331 (randomized to standard ABVE-PC vs Bv-AVEPC; Castellino et al NEJM 2022) had plasma centrally profiled at baseline and after 2 cycles (C3D1) by PhasED-Seq. Interim molecular response was defined as major molecular response (MMR, ≥2.5-log10) (Kurtz et al 2018 JCO). Centrally reviewed interim PET after 2 cycles (rapidly (RRL) vs slow (SRL) responding lesions based on Deauville scores (RRL≤3 vs SRL=4-5)) and baseline total metabolic tumor volume (TMTV) were captured. PFS associations used Kaplan–Meier log-rank and Cox log-likelihood methods. On-treatment predictors were evaluated in an interim C3D1 model, and combined into a dynamic multivariable radiogenomic model integrating baseline and on-treatment risk factors.

Results: Among 399 of 587 patients eligible for AHOD1331, 1021 blood specimens were available either before or after 2 planned per protocol therapy cycles, with 98% of patients/specimens evaluable after QC. This evaluable subset was representative of the AHOD1331 ITT population.

At baseline (n=390), higher pretreatment ctDNA levels were significantly associated (all p<0.05), with B-symptoms (across stages, and when restricted to Stage IV), LMA, ESR≥50, and TMTV (Spearman rho=0.37; p=4.49e-13). Baseline ctDNA levels were not significantly associated with stage, sex, age, or treatment arm. Among baseline clinical, radiographic parameters, and pretreatment ctDNA levels, only radiographic TMTV retained prognostic value in both univariate and multivariate models for PFS.

In contrast, at interim C3D1 evaluation (n=238 available and no event dropoff, p=0.63), ctDNA MMR strongly predicted PFS across both arms in a landmark analysis (no MMR hazard ratio (HR) 6.35 (2.90-13.88, p<0.0001), and in each arm (Bv HR 5.48 [1.48-20.26], p<0.005; Standard HR 6.65 [2.49-17.76], p<0.0001). Interim PET SRL was associated with PFS for the standard but not the experimental arm (Bv HR 1.37 [0.45-4.19], p=0.58; Standard HR 3.32 [1.57-7.04], p<0.001 in this study, and as also noted previously). Interim PET and interim ctDNA were significantly associated but only modestly concordant (Cohen’s κ 0.26), and ctDNA response added prognostic value beyond interim PET, despite neither significantly differing between treatment arms. Combining interim PET and ctDNA assessments improved prognostication for a PFS event (3-tier and 2×2 composites, log-rank P<0.001).

Finally, we compared baseline and interim factors for prognostic value in integrated models considering TMTV (log10), interim PET status, and ctDNA MMR status. In stepwise models, interim ctDNA MMR was the most significant single predictor across arms (ΔAIC=-14.91, p<0.001), and model performance was further improved by adding baseline TMTV (ΔAIC=-14.51, p<0.001) and interim PET (ΔAIC=-2.03, p<0.001). Tertiles of this dynamic multivariable radiogenomic model score defined low-, intermediate-, and high-risk groups with 4-year PFS of 97%, 92%, and 75%, respectively (log-rank P<0.001); these risk groups and prognostic discrimination were preserved within both arms (Bv-AVEPC P=0.005; ABVE-PC P=0.003).

Conclusions: In the largest ctDNA outcomes analysis of high-risk cHL, interim ctDNA response at C3D1 is a strong, treatment-arm-independent predictor of PFS that adds considerably to PET. A bias-free radiogenomic model integrating baseline metabolic tumor burden with interim imaging and molecular response yields accurate, clinically interpretable PFS risk, supporting prospective evaluation of ctDNA-guided, response-adapted therapy in childhood Hodgkin lymphoma.

Shuyu Shi, PhD: No relevant conflicts of interest to declare.