Evaluation of cell-free DNA as a biomarker in RRMM

After treatment, patients (pts) with multiple myeloma (MM) may achieve complete response (CR). However, even in the cases of CR, a pt will eventually relapse. This is due to the fact that a tiny number of malignant plasma cells (PCs) remain untreated in the bone marrow (BM) of MM pts. These residual cells, which will eventually proliferate and lead to relapse, are collectively known as minimal residual disease (MRD). Monitoring MRD is a good means for controlling disease progression and re-treating pts in a timely manner.

Methods to evaluate MRD include multiparametric flow cytometry (MFC) and next-generation sequencing (NGS) on PCs from the BM. BM PC biopsies are burdensome and the two existing procedures (BM aspiration and BM trephine) do not always provide the same results.

Giulia Biancon and Silvia Gimondo, both from the National Cancer Institute (IRCCS) in Milan, Italy, and colleagues, proposed that analysis of clonal heavy immunoglobulins (IGH) in cell-free DNA (cfDNA) from plasma can be used to estimate MM tumor burden. To examine this hypothesis, they performed a pilot study with 25 pts who were taking part in a randomized phase III clinical trial and had achieved CR during the course of the trial. The trial compared the activity of bortezomib (bor)/cyclophosphamide (cyclo)/dexamethasone (dex) vs lenalidomide (len)/cyclo/dex as second-line therapy after first relapse. During the study, BM and peripheral blood (PB) samples were collected. The study was published in The Journal of Molecular Diagnostics in August 2018.

Study Design:

• Number of pts = 25
• Number of male pts = 16
• Median age = 65 (range, 41–75)
• BM and PB sample collection points: at study entry; after three, six, and nine cycles of treatment, and at follow-up time points
• Previous treatments: bortezomib = 9 pts; high dose melphalan = 16 pts
• Median follow-up = 17 months (range, 2–82)
• Median levels of CD138⁺ PCs in the BM (at accrual) = 8.8% (range, 0.05–80%)
• Circulating CD138⁺ PCs in the PB = none
• Sequencing using the Ion PGM bench-top system generated an average of 101,455 reads (± 79,377 SD)

Key Data:

Identification and characterization of rearranged clonal IGH in MM PCs

• Sequencing using the Ion PGM bench-top system generated an average of 101,455 reads (± 79,377 SD)
• Mean number of sequences that passed quality filters = 47,755 (± 32,787 SD)
• The clonality of samples was assessed by the percentage of reads with the same V(D)J usage and a unique complementarity-determining region 3 (CDR3) nucleotide sequence
• Median frequency of most represented clone = 50.1% (range, 1.4–86.7%)
• To define a clone as a tumor clonotype, frequency threshold was set to > 5%
• Sequencing method identified a myeloma clonotype in 22/25 pts
• In 3/25 pts IGH frequencies were < 5%
• In 22/22 pts: non-random IGH V(D)J gene segment usage of the tumor clonotype
• VH –families expressed at highest frequencies = IGHV3; IGHV4; IGHV1; IGHV2
• VH-segments encountered most frequently = VH3-23 (18%, 4/22 pts); VH1-69; VH2-5; VH3-30; VH3-48; VH3-7 (each at 9%)
• IGHD segments found most frequently = IGHD6-19; IGHD6-13
• Frequency of IGHJ subgroups = IGHJ4 over-represented (45% observed vs 16.7% expected, assuming a random choice for JH, P < 0.001, Fisher’s exact test); IGHJ2 and IGHJ3 under-represented (5%, P < 0.001)

Validation studies

• Comparison of NGS data with traditional Sanger sequencing: in 22/22 pts complete agreement of results (V(D)J identity and CDR3 sequence of tumor clonotype); in the 3 pts with IGH sequences < 5% frequency, Sanger sequencing identified a polyclonal IGH repertoire
• Amplification with a different primer set: consistent results were obtained when a different forward primer was used
• Evaluation of sensitivity of the sequencing method: a series of samples generated by serial 10-fold dilutions of DNA extracted from two cell lines (Namalwa and JVM-2) with a known clonal IGH rearrangement; the clonal sequence was identified at dilutions as low as 10^-5

CfDNA extraction and amplification

• No linear correlation was found between plasma volume and cfDNA concentration
• CfDNA was collected in two different tube types: K2-EDTA tubes (BD Vacutainer) and cfDNA BCT tubes (Streck, Inc. Omaha) and the quality and quantity of cfDNA was compared
• When plasma cells were processed within three hours, both tube types showed similar performances though slightly higher concentrations of cfDNA were observed from plasma samples collected in cfDNA BCT tubes
• When plasma cells were processed after three hours, cfDNA stability was guaranteed only when cfDNA BCT tubes were used
• Plasma storage conditions (fresh or frozen samples) did not impact cfDNA recovery: median cfDNA quantity from 1 ml of plasma = 7.8 ng after immediate processing; 7.5 ng after < 6 months storage; 7.4 ng after > 6 months storage
• The available 1–3 ml plasma collected in K2-EDTA tubes in the present study was considered sufficient for extraction and downstream amplification

Sequencing of IGH gene rearrangements in plasma samples

• Samples available at study entry = genomic tumor cell DNA (tcDNA) and plasma cfDNA from 22 pts with a defined myeloma clonotype
• IGH cfDNA sequencing generated an average of 210,056 reads (± 161,520 SD)
• Mean number of sequences that passed quality filters = 77,188 (± 55,058 SD)
• Percentage of predominant clones in MM plasma samples = 1.2–18% (n = 22 pts)
• No predominant clones were observed among control plasma samples: mean 0.006 % (range, 0.002–1%) (n = 10 controls)

Prognostic value of clonotypic IGH cfDNA

• Median frequency of PCs on BM biopsies (BMB) before therapy = 40% of BM leukocytes (range, 10–95%)
• No significant correlation between PC frequency and the clonotypic cfDNA frequencies (r = 0.098, P = 0.6724)
• No significant correlation between paraprotein levels and clonotypic cfDNA frequencies (r = -0.018, P = 0.9504)
• Moderate positive correlation between serum-free light chain (sFLC) ratios and clonotypic cfDNA frequencies (r = 0.4567, P = 0.2552)
• No significant correlation between BMB PCs and paraprotein levels (r = 0.289, P = 0.295)
• No significant correlation between BMB PCs and sFLC ratios (r = -0.032, P = 0.8465)
• Significant correlation between plasma sample NGS and BM PCs infiltration by MFC (r = 0.713, P = 0.0002)
• Significant correlation between plasma sample NGS and tcDNA NGS (r = 0.45, P = 0.0354)
• Levels of the tumor-associated clonotype in cfDNA distinguish between groups of pts with different prognosis (cut off value = 4.7% of total reads)
• Pts with ≥ 4.7% (n =12) of the tumor-associated IGH sequence prior to therapy had significantly inferior progression-free survival (PFS) than pts with levels < 4.7% (n = 10) (median values = 268 vs 990 days; hazard risk (HR) = 3.507, P = 0.04988, Log-rank test)
• Higher numbers of PCs calculated by MFC were also associated with poorer PFS (median percentage as cut-off value = 8.8%) (median values = 268 vs 545 days; HR = 3.745, P = 0.04655, Log-rank test)

Disease monitoring by clonotypic IGH cfDNA

• Longitudinal analysis of clonal IGH cfDNA = 70 plasma samples analyzed (22 pts)
• Tumor-associated clonotype could be tracked over time in plasma samples
• MRD monitoring using MFC: 1 x 10⁶ cellular events; median frequency of detection of residual cells = 0.00065 tumor cells out of the total events analyzed (range, 0.00000668–0.027)
• MRD-negativity = < 50 aberrant-phenotype PCs were detected; 6 pts achieved MRD-negativity
• High-level correlation between cfDNA NGS and MFC data (r = 0.5831, P = 0.0044)
• Prolonged PFS for the 6 pts with MRD-negativity (P < 0.001); frequencies of the cfDNA rearrangement < 10^-5
• Mean PFS ± SD in relation to clonotypic cfDNA frequencies: for frequencies < 10^-5 = 714 ± 327 days (n = 6 pts); for frequencies ≥ 10^-5–≤ 10^-4 = 325 ± 75 days (n = 9 pts); for frequencies > 10^-4 = 143 ± 59 (n = 7 pts)

Conclusions

This report evaluates for the first time the clinical significance of cfDNA using an IGH deep-sequencing method. Analysis of cfDNA gives rise to results similar to those obtained using MFC. Future studies can expand these promising results by exploring further the sensitivity of cfDNA as a MM biomarker.

References

Biancon G. et al.Non-invasive molecular monitoring in multiple myeloma patients using cell-free tumor DNA: a pilot study. The Journal of Molecular Diagnostics. 2018 Aug 27. pii: S1525-1578(18)30063-1. DOI: 10.1016/j.jmoldx.2018.07.006. [Epub ahead of print].