Ultrasensitive Circulating Tumor DNA MRD Status Predicts Treatment Failure & Complements PET/CT throughout Treatment for Early and Advanced Stage Classic Hodgkin Lymphoma

Background:

While PET-CT has been successfully integrated to guide risk-adapted therapies in classic Hodgkin lymphoma (cHL), it has imperfect sensitivity and specificity and does not measure disease at the molecular level. Circulating tumor DNA (ctDNA) profiling of cHL has shown significant diagnostic and prognostic utility pre-treatment and for post-treatment surveillance (Alig et al., Nature 2024). Among ctDNA MRD monitoring tools, tracking so called phased somatic variants is among the most sensitive methods (Kurtz et al., Nature Biotech 2021). Here, we assess the prognostic value of ultra-sensitive ctDNA MRD for cHL during and after therapy.

Methods: We assembled a cohort of newly diagnosed cHL patients (pts) enriched for older individuals (median age 62y), participating in the prospective clinical trial LYSA-PVAB or the real-world cohorts (REALYSA, BIO-LYMPH) undergoing first-line chemotherapy ± radiatiotherapy. We stratified pts into Early (I-II) vs Advanced Stage (III-IV or IIB with risk factors [large mediastinal mass and/or extranodal lesion(s)]). Pre-treatment plasma specimens were profiled to identify phased variants and used to monitor molecular residual disease (MRD) at subsequent timepoints by phased variant enrichment and detection sequencing (PhasED-seq). For Early Stage, we assessed ctDNA at pre-treatment and C3D1. For Advanced Stage, we assessed ctDNA at C3D1, C5D1, and end of treatment (EOT). We then assessed concordance with PET-CT and the predictive value of detectable ctDNA at these timepoints for outcomes including progression free survival (PFS) and time to relapse (TTR).

Results:

We identified 181 pts evaluable for MRD, including 27% with Early Stage vs 73% with Advanced Stage cHL. For Early Stage cases (median age 61y), plasma specimens were evaluable at pre-treatment (100%) and C3D1 (92%). ABVD-based regimens were most common (67%) and 70% received consolidative radiation. Among this cohort, 20% were MRD-positive at C3D1, which was highly prognostic for inferior PFS (HR 9.9, 95%-CI [1.8-54.5]) and TTR (7.7 [1.3-47]). MRD+ status at C3D1 remained strongly associated with worse PFS (7.7 [1.2-50]) after controlling for post-C2 PET (PET2) complete response (CR), and when limiting to cases without post-C2 treatment escalation (6.7 [1.12-40.6]). Among 31 pts in CR at PET2, 22% (n=7) were MRD+, of which 29% (n=2) experienced relapse after therapy. Conversely, among non-CR pts at PET2 with undetectable ctDNA (n=5), none progressed after therapy. At C3D1, using time-dependent ROC curves, the AUC for PhasED-seq predicting PFS at C3D1 was 79%.

For Advanced Stage cases (median age 64y, 80%>60y), samples were evaluable at pre-treatment (100%), C3D1 (55%), C5D1 (47%), and EOT (56%). The most common regimens were PVAB- (53%), ABVD- (20%), or BEACOPP-based (19%). MRD+ rates were 49% at C3D1, 24% at C5D1, and 38% at EOT. MRD+ status at C3D1 predicted significantly inferior PFS (2.8 [1.1-6.9]) and TTR (4.6 [1.5-14]). When comparing interim molecular response rates after 2 cycles, BEACOPP-like therapy had significantly lower MRD+ rates (17%) than PVAB (50%) or ABVD-like (68%) regimens (p=0.001). Among pts >60y, MRD+ status at C3D1 remained predictive of PFS (4.5 [1.6-12]) and TTR (4 [1.2-13]). MRD+ status at C5D1 again predicted inferior PFS (12 [4.7-28]) and TTR (10 [3.7-28]). At C5D1, 80% (4/5) of PET CR pts who were MRD+ experienced relapse. At EOT, MRD+ status also predicted inferior PFS (4.8 [2-11.8]). In pts achieving CR at EOT PET, MRD+ status predicted inferior PFS (6.2 [2.1-18.6]) and TTR (7.4 [1.8-30.9]). At EOT, using time-dependent ROC curves for PFS, PhasED-seq had a sensitivity of 78%, specificity of 87%, and AUC of 83%. The AUC for PhasED-seq predicting PFS was lower at C3D1 (68%) and at C5D1 (76%).

Conclusions: Detection of ctDNA MRD during therapy and at EOT is highly predictive of treatment failure in cHL. While this is evident as early as after 2 cycles in early stage, MRD status has the highest predictive values at later time points in advanced stages, including the highest performance at EOT. We observed frequent relapse in pts with interim MRD+ despite PET CR, suggesting complementary utility for molecular MRD in addition to PET-CT in both Early and Advanced Stage cHL. These data emphasize the utility of ctDNA for MRD in cHL pts with the potential to be integrated into novel risk-adapted study designs.