Cell-free DNA chromosome copy number variations predict outcomes in plasma cell myeloma

Dear Editor, Testing for measurable residual disease (MRD) in persons with plasma cell myeloma (PCM) after therapy correlates with therapy outcomes including progression-free survival (PFS) and survival (reviewed in refs. [1, 2]). Most MRD-testing in this setting uses multi-parameter ﬂ ow cytometry (MPFC) but prediction accuracy is imperfect with C-statistics of only about 0.67 for PFS and 0.76 for overall (OS) [3]. Recently, MRD-testing includes next-generation sequencing (NGS) and/or next-generation ﬂ ow cytometry (NGFC) [4, 5]. These approaches are more sensitive compared with MPFC but there are no data on C-statistics for these tests. Prediction inaccuracies arise from diverse sources such as inadequate sensitivity and/or speci ﬁ city, imprecision and sampling biases resulting from non-homogeneous bone marrow involvement and/or extra-medullary PCM [6, 7]. More accurate methods to predict PFS and OS are needed. So-called liquid biopsy using cell-free


Dear Editor,
Testing for measurable residual disease (MRD) in persons with plasma cell myeloma (PCM) after therapy correlates with therapy outcomes including progression-free survival (PFS) and survival (reviewed in refs.[1,2]).Most MRD-testing in this setting uses multi-parameter flow cytometry (MPFC) but prediction accuracy is imperfect with C-statistics of only about 0.67 for PFS and 0.76 for overall (OS) [3].Recently, MRD-testing includes next-generation sequencing (NGS) and/or next-generation flow cytometry (NGFC) [4,5].These approaches are more sensitive compared with MPFC but there are no data on C-statistics for these tests.Prediction inaccuracies arise from diverse sources such as inadequate sensitivity and/or specificity, imprecision and sampling biases resulting from nonhomogeneous bone marrow involvement and/or extra-medullary PCM [6,7].More accurate methods to predict PFS and OS are needed.
So-called liquid biopsy using cell-free DNA (cfDNA) is a non-invasive technique for diagnosis, evaluation, and/or monitoring of diverse cancers including hematological cancers [8,9].We developed an ultra-sensitive chromosome aneuploidy detector (UCAD) to profile genome-wide chromosomal instability through low-coverage wholegenome sequencing (LC-WGS) of cfDNA and used it to detect and monitor MRD in 242 longitudinal plasma cfDNA samples from 68 newly-diagnosed subjects with PCM treated at Shanghai Changzheng Hospital between March 2018 and December 2021 enrolled in a prospective trial (NCT04122092).We also compared the accuracy of this technique with MRD-testing by NGFC using the EURO-flow 8color two-tube method.Diagnosis and response were assessed according to the International Myeloma Working Group (IMWG) criteria [10].Subjects had blood and bone marrow sampling at diagnosis and progression to determine chromosomal aberrations.Induction therapy was bortezomib-based regimen, including lenalidomide, bortezomib and dexamethasone (VRD) or bortezomib, cyclophosphamide and dexamethasone (CBD).29 subjects subsequently received an auto-transplant and posttransplant therapy with lenalidomide with or without bortezomib.We collected plasma samples at baseline (T 0 ), at the end of induction cycles 2, 4, and 6 and at disease progression (Fig. 1A).Transplant recipients had additional samples taken immediately pretransplant.Total genomic DNA and cfDNA were isolated from plasma using the Circulating Nucleic Acid kit (Qiagen, Valencia, CA, USA).Segment copy number and tumor fraction (TFx) were derived using the customized UCAD workflow.Cell-free DNA from plasma samples was analyzed using the Illumina X 10 system.The detailed procedure of Low-coverage whole-genome sequencing and additional related methods details are provided in the Supplementary Method.
PFS and OS were estimated by using the Kaplan-Meier method.Data were analyzed with SPSS22.0 or R software version 3.4.3(R Foundation for Statistical Computing, Auckland, New Zealand).P-values (two-tailed) less than 0.05 were set as the threshold for statistical significance.
27 subjects had a negative NGFC-test for MRD. 10 were cfDNA CNV-negative at simultaneous testing and 17, -positive (Table S5).NGFC-negative subjects had significantly longer PFS, not reached but must exceed 43 months, compared with 26 months (16, 35 months) in subjects NGFC-positive (HR = 2.87 [1.39, 5.90], P = 0.003; Table S3).In NGFC-negative subjects those cfDNA CNVnegative had a median PFS not reached but must exceed 43 months compared with those who were -positive, not reached but must exceed 35 months (P = 0.09).In NGFC-positive subjects those also CNV-positive had a median PFS of 19 months (7,32 months) compared with those who were -negative, not reached but must exceed 47 months (P = 0.023; Fig. 1F).Corresponding 3-year OS rates were 100% in subjects negative of both tests, 81% in subjects NGFC-negative but cfDNA CNVpositive and 65% in subjects positive in both tests (Fig. 1G).CNVpositive subjects with a NGFC-negative test had a HR of 3.63 (0.75, 17.5; P = 0.09) compared with those who were negative in both tests.Amongst subjects who were NGFC-test negative there were no significant differences in baseline covariates between those cfDNA CNV-positive and -negative (Table S6).
Liquid biopsy techniques in PCM include quantifying blood plasma cells, concentration of cfDNA or circulating myeloma DNA (ctDNA) and analyzing mutation topography by next-generation sequencing [11][12][13].There are few data on the quantitation of CNVs in cfDNA samples [14,15].We used an UCAD to identify CNVs in cfDNA plasma samples, and obtained longitudinal collections which allowed us to monitor CNV changes of PCM in response to treatment.We found dynamic cfDNA CNVs detection was a better predictor of prognosis than baseline cfDNA CNVs, and dynamics cfDNA-CNV analysis during therapy was more sensitive than interim MRD responses by NGFC.Our study has important limitations.1st, we had few subjects requiring validation of our findings.2nd, CNVs based on LC-WGS do not provide potentially important cytogenetic data such as t(4;14), t(11;14) and t(14;16).
In summary, our data indicate changes from baseline levels of cfDNA-CNVs after therapy increases prediction accuracy of PFS and OS in subjects with newly-diagnosed PCM and is more accurate than MRD-testing by NGFC.Combining plasma cfDNA-CNV analysis with standard approaches for MRD detection may usefully contribute to the prognostic analysis of PCM.If validated, this approach may help physicians with therapy decision-making.

Fig. 1
Fig. 1 Time course of cfDNA aberrations.A Study flow diagram.Plasma samples were collected before, during and at the end of induction therapy.MRD was measured after 4 cycles of induction therapy, before an auto-transplant, or at the time of suspected complete remission every year thereafter.B Significant chromosomal aberrations (|Z| ≥ 3) detected before and during therapy.C cfDNA CNV changes (measured by maximum |Z| values) in progression free subjects and disease-progressed subjects.D, E Kaplan-Meier curves showing PFS (D) and OS (E) of cfDNA CNV-positive or -negative subjects.F, G Kaplan-Meier curves showing the PFS (F) and OS (G) of subjects based on combining cfDNA-CNV and interim NGFC MRD-test results.cfDNA cell-free DNA, CNVs copy number variations, MRD measurable residual disease, PCM plasma cell myeloma, BM bone marrow, NGFC next-generation flow cytometry, PFS progression-free survival, OS overall survival.