Using Different Vaf Cut-Offs, Myeloid NGS Analysis of 924 Paired Peripheral Blood/Bone Marrow Samples Yielding 6944 Result Pairs Reveal Peripheral Blood As Safe Alternative Sample to Bone Marrow, Irrespective of the Degree of Neutropenia or Circulating Blast Counts

Background

Next generation sequencing (NGS) has become indispensable for monitoring of patients (pts) with myeloid neoplasias. Guidelines require bone marrow (BM) evaluations for the above, which are only performed in ~50% of pts outside of clinical trials [Mukherjee S, Oncologist 2023; Pleyer L, AJH, 2023], indicating a need for surrogate samples. We previously demonstrated highly concordant NGS results between 240 paired BM and peripheral blood (PB) samples [Jansko-Gadermeir B, Cancers 2023]. This updated analyses with ~4 times as many samples analyses the potential impact of PB or BM blast percentage and various degrees of neutropenia on concordance, at 3 different variant allele frequency (VAF) cutoffs.

Methods

AmpliSeq for Illumina Myeloid Panel (40 genes) and VariantStudio v3.0 were used. Sensitivity was set to 1% VAF. All variants were visualized in integrated genome viewer to exclude sequencing artefacts. Manual review acc. to standard operating procedures [Barnell EK, Genet Med 2019] was performed. VAF cutoffs of 5%, 1% as well as an experimental 0.1% (with the provision of ≥10 alternate reads) were used. Spearman's correlation and sensitivity analyses (Kendall's Tau) were assessed using SAS®9.3.

Results

924 paired BM and PB samples collected from 654 pts, 42% of which had ≥2 (range 2-9) serial sample pairs. Longitudinal analyses of serial sample pairs sequenced at different timepoints make sequencing artefacts in our cohort unlikely. Median no. of days between sampling was 0.0. PB blasts were 0% in 665/924 (72%) and neutropenia <1.0, <0.5 or <0.1 G/L was present in 29, 15, and 9% of sample pairs.

Amplicon mean coverage across all samples was >12600x. Minimum read depth of targeted regions was 500x in 99.4%, and between 100 and 497x in 0.6% of cases.

6954 result pairs were found. Among these, a total of 3899, 5647 and 6156 mutations were identified using a VAF cutoff of 5%, 1% or ≥0.1%. Of these, 235/3899 (6%), 444/5647 (8%) and 202/6156 (3%) were discordant between the BM and PB.

Among discordant mutations 46/235 (20%), 288/455 (63%) and 72/202 (36%) were found only in the PB, whereas 189/235 (80%), 167/455 (37%) and 130/202 (64%) where found only in the BM using ≥5%, ≥1% and ≥0.1% VAF cutoffs.

For VAF cutoffs of ≥5%, ≥1% or ≥0.1%: strong or moderate correlation between mutations found in PB vs BM was observed (r=0.831, r=0.525, r=0.717); very strong correlations were found between PB_VAF and BM_VAF (r=0.940, r=0.929, r=0.937); median (IQR) difference between PB_VAF and BM_VAF of concordant mutations was 2.8 (0.0-8.8)%, 1.3 (-0.2-6.6)%, 1.1 (-0.2-5.7)%.

Negligible or weak correlation was found between the BM_VAF and the BM blast count (r=0.090, r=0.098, r=0.119) or the PB_VAFand the PB blast count (r=0.147, r=0.146, r=0.149) for VAF cutoffs ≥5, ≥1 or ≥0.1%.

Correlation between PB_VAF and BM_VAF in subgroups with (r=0.946, r=0.945, r=0.948) or without circulating blasts (r=0.931, r=0.912, r=0.925) remained very strong, irrespective of VAF cutoff used.

Correlation between PB_VAF and BM_VAF in subgroups with (r=0.873, r=0.878, r=0.887) or without (r=0.967, r=0.950, r=0.958) neutropenia, as defined by an absolute neutrophil count (ANC) <1.0 G/L was strong. This remained true for varying degrees of neutropenia: ANC <0.5 G/L (r=0.852, r=0.876, r=,0.889) or ANC <0.1 G/L (r=0.810, r=0.859, r=0.892) for VAF cutoffs of ≥5, ≥1 or ≥0.1%.P values for all correlations listed above were <0.0001.

High concordance (100, 100, 100%), positive predictive value (PPV) (97.6, 90.0, 97.6%), negative predictive value (NPV) (99.7, 99.0, 98.9%), sensitivity (90.6, 94.0, 95.8%), specificity (99.9, 98.3, 99.4%), and between NGS analyses of paired BM/PB samples was observed for the differing VAF cutoffs applied.

Conclusions

Differing degrees of neutropenia, BM or PB blast percentage, and VAF cutoffs have little or no impact on the concordance, PPV, NPV, sensitivity, or specificity of myeloid NGS results of paired PB/BM samples. This is in line with the fact that all PB cells except for lymphocytes have been identified as part of the MDS/AML clone. Thus, PB samples can be reliably used as an alternative to BM samples irrespective of ANC or PB blast counts, rendering a bone marrow evaluation for the purpose of monitoring mutations unnecessary. We use 1% VAF cutoff in our reports. Use of unique molecular identifiers to reduce the rate of false positive variant calls and aid in discerning authentic biologic signals is planned to increase sensitivity.

Pleyer:AbbVie: Honoraria; Otsuka: Honoraria; BMS: Honoraria. Melchardt:Abbvie, Roche: Honoraria. Greil:Celgene, Roche, Merck, Takeda, AstraZeneca, Novartis, Amgen, BMS, MSD, Sandoz, Abbvie, Gilead, Daiichi Sankyo, Sanofi: Honoraria; Celgene, Novartis, Roche, BMS, Takeda, Abbvie, Astra Zeneca, Janssen, MSD, Amgen, Merck, Gilead, Daiichi Sankyo, Sanofi: Consultancy; Celgene, Roche, Merck, Takeda, AstraZeneca, Novartis, Amgen, BMS, MSD, Sandoz, Abbvie, Gilead, Daiichi Sankyo: Research Funding; Roche, Amgen, Janssen, AstraZeneca, Novartis, MSD, Celgene, Gilead, BMS, AbbVie, Daiichi Sankyo: Other: Travel, accommodations, expenses; Novo Nordisk, Lilly: Divested equity in a private or publicly-traded company in the past 24 months.