Risk Assignment Using the Myeloma Genome Project Panel: A Comparison to Standard of Care Genetic Testing within the UK-MRA RADAR Trial

Background.

A subset of patients with multiple myeloma (MM), termed “high-risk”, have high rates of early relapse and short survival. The definition of high-risk (HR) has evolved with treatment advances and improved genetic characterisation of patient cohorts. New genetic insights require a test that can detect translocations, copy number alterations (CNA) and small variants for risk assignment. This test must correctly identify the HR cohort and be widely and sustainably deployable in healthcare systems.

The Myeloma Genome Project (MGP) Panel is a targeted DNA panel developed to include loci for CNA, translocations and genetic variants (PMID:35522533). It incorporates key disease drivers, therapeutic targets and prognostic features, including detection of genomic double/triple hits and bi-allelic inactivation of tumour suppressor genes via structural change, CNA and somatic variation.

The UK-MRA RADAR study is a national, multi-centre, risk-adapted, response-guided multi-arm, multi-stage phase II/III trial which will recruit 1400 patients with newly diagnosed MM eligible for autologous stem cell transplant. Genetic risk classification is performed as standard of care (SoC) testing using Fluorescence in situ hybridization test (FISH)/ multiplex ligation-dependent probe amplification (MLPA). HR is defined as two or more genetic abnormalities incorporating t(4;14), t(14;16), t(14:20), del(17p), del(1p) and gain(1q). Participants who cannot be confirmed as standard risk (SR) or HR are classified as unable to determine (UTD) and treated as SR.

The collection of central samples within RADAR affords a unique opportunity to compare the performance of FISH/MLPA to MGP in assigning risk status. We report the first results of this analysis, with more to follow.

Methods.

DNA was extracted from CD138-magnetic bead enriched MM cells in baseline bone marrow samples at a central lab. Paired germline DNA was obtained from peripheral blood. MGP libraries were prepared in an NHS clinical laboratory from samples that met QC metrics and sequenced as previously reported. SoC testing was done across 23 regional NHS labs using FISH/MLPA. Data were analysed to identify 20% Cancer Clonal Fraction (CCF) of high-risk features as per requirements for the SoC results used in RADAR.

Genetic risk classification was derived as per the RADAR definition and agreement between the two methods assessed using Cohen’s Kappa. Additional measures (sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV)) were derived using SoC classification as the reference group, in a complete case analysis.

Finally, a recently proposed revised IMWG HR definition was applied to the MGP results, and the classification of patients compared descriptively.

Results.

To 1^st July 2024, 876 patients have been recruited to RADAR. Of the first 93 central samples received, we report comparison of the 83 who met QC criteria for performing MGP sequencing/analysis. This set was similar to the overall RADAR population in terms of age, ethnicity, paraprotein type and HR proportion, with numerically more females and ISS1 in the MGP set.

In the reported set SoC risk assignment was 63/83 (75.9%) SR, 15/83 (18.1%) HR and 5/83 (6.0%) UTD. MGP-based risk assignment using the RADAR definition was 65/83 (78.3%) SR, 15/83 (18.1%) HR and 3/83 (3.6%) UTD. Cohen’s Kappa between the two was 0.64 (95% CI: 0.46, 0.83), with MGP re-designating 2 HR cases to SR, and 3 SR cases to HR.

78/83 (94.0%) participants had confirmed classification (SR or HR) in the SoC method. In these 78 samples with SoC as the reference, the sensitivity of MGP was 80%, the specificity was 97%, the PPV 86% and NPV 95%.

Risk assignment by MGP using the new IMWG definition was 48/83 (58%) SR, 32/83 (39%) HR and 3/83 (3.6%) UTD. TP53 variants were present in 8/83 (9.6%), co-existing with other HR events in 6/83 (7.2%).

Discussion.

These initial results suggest there is substantial agreement between the SoC and MGP methods when classifying HR according to the RADAR trial definition. RADAR’s ‘2-hit’ HR definition predates the anticipated IMWG criteria, where single TP53/del(17p) events also define HR status, thus HR incidence by IMWG criteria is higher. MGP detects TP53 variants, but RADAR SoC tests do not. We continue to assess and develop MGP assay performance against SoC, in consideration of its deployment for clinical use.