Potential value of high-throughput single-cell DNA sequencing of Juvenile myelomonocytic leukemia: report of two cases – npj Systems Biology and Applications

Patient 1

Patient 1 was diagnosed with JMML at the age of 3 years. Monosomy of chromosome 7 was detected by cytogenetic analysis. Targeted DNA sequencing revealed a missense variant c.181 G > T (p.D61Y) in the PTPN11 gene (VAF of 53%), but no variants were detected in other genes. The patient was put under watchful observation; however, an increase in leukocytes was observed within six months after diagnosis. According to bone marrow aspiration, the percentage of blast cells of myeloid cell line differentiation was 25%. Multiparameter flow cytometry (MFC) of bone marrow aspirate revealed the following antigen rates: 23% CD2+, 92% CD7+, 100% CD11а+, 32% CD11b+, 57% CD11с+, 27% CD13+, 35% CD15+, 60% CD33+, 71% CD34+, 100% CD45+, 32% CD64+, 100% CD117+, 32% HLA-DR+, and 6% CD79а+ cells. Thus, JMML transformation into sAML with coexpression of CD2 and CD7 was revealed (Fig. 1 and Supplementary Fig. 1). No additional cytogenetic abnormalities were found. ScDNA sequencing was performed, and two clones were found: the first clone harbored the single-nucleotide variant (SNV) PTPN11 c.181G>T (5.7% of cells), and the second clone showed the primary PTPN11 mutation and a comutation c.2602G>A (p.D868N) in the SETBP1 gene (88.3% of cells) (Table 1). A population of cells with wildtype (WT) PTPN11/SETBP1 status was also detected (6% of cells). The resulting pattern can represent progression of the sAML clone (secondary mutation in the SETBP1 gene) from the primary JMML clone with SNVs in the PTPN11 gene.

Considering the progression to sAML, HSCT was chosen as the only curative treatment option. Prior to HSCT, two courses of high-dose AML-like chemotherapy (FLAM: fludarabine, Ara-C, mitoxantrone and FLAE: fludarabine, Ara-C, VP-16) were administered for cytoreduction; however, complete remission was not achieved. Three months after sAML was diagnosed, transplantation was performed from a haploidentical parental donor with TCRαβ⁺ and CD19⁺ graft depletion. Conditioning regimen prior to myeloinfusion included treosulfan, melphalan, and fludarabine¹⁶. Engraftment was observed on day +18. Measurements of MRD in the bone marrow on day +30 by MFC showed persistence of a 5% leukemic population. ScDNA-seq revealed a clone with SNVs in the PTPN11 and SETBP1 genes (12.8% of cells) corresponding to the population detected by MFC. The remaining 87.2% of cells comprised the WT population for the genes described above. Interestingly, we also found the c.98C>G polymorphism (p.P33R) in the TP53 gene in homozygous and heterozygous states. Importantly, this polymorphism was homozygous in the tumor and a smaller part of the WT populations. In the WT population, 88.5% of cells were heterozygous and 11.5% homozygous according TP53 polymorphism. This SNV was also detected prior by targeted DNA sequencing at diagnosis and scDNA-seq before HSCT, but only in homozygous conditions, suggesting donor origin of cell population with heterozygous TP53 c.98C>G after HSCT. (Table 2). The obtained data were verified using the method of hierarchical clustering described by Xu et al. ¹⁷. This method allowed us to cluster all cells according to all SNVs into two fundamentally different cell populations, each of which differed in the zygosity of the TP53 c.98C>G polymorphism (Fig. 2A) that indicates a different origin of these cells. At the same time, leukemic cells differ from WT (TP53hom/het) ones (Fig. 2B), which is consistent with the point that the patient’s WT populations may origin due to residual normal hematopoiesis. To rule out the leukemic nature of the patient’s PTPN11wt/SETBP1wt cells, which could also be explained by the presence of chromosome 7 monosomy (by fluorescence in situ hybridization (FISH) data), copy number variation (CNV) analysis of sequencing data was performed. We did it first on cancer and then on WT cells, so that we could compare the two populations. Donor’s cells were used as a control. The tumor PTPN11mut/SETBP1mut/TP53hom cell clone exhibited loss of heterozygosity for the BRAF and EZH2 genes (Fig. 2C, D), which was consistent with FISH data. No copy number aberrations were detected in the patient’s PTPN11wt/SETBP1wt cells (10%) indicating their nonleukemic nature (Fig. 2D, E). These cells may represent a residual population of normal hematopoietic cells coexisting with tumor clones. FISH-plot and bar-plot analyses of disease evolution based on two time points (before and after HSCT) are shown in Fig. 2F.

A Hierarchical clustering heatmap of SNVs after HSCT. Сells are divided into two clusters that differ in the zygosity status of the polymorphism (TP53:chr17:7579472:G/C) in the TP53 gene. These data indicate the different origin of cells from these two clusters. B UMAP analysis after HSCT. Two subsets of tumor cells are separated from clusters of other cells. The presence of other cells in one of the tumor subsets may be due to ADO in tumor clones. C CNV profile of all detected regions in PTPN11mut/SETBP1mut/TP53hom tumor clone. The BRAF and EZH2 genes on chromosome 7 show a decrease in ploidy (marked in red), which is consistent with monosomy 7 detected in the patient. D CNV-related genes on chromosome 7 (BRAF and EZH2) for all clones. Hierarchical clustering heatmap shows loss in the BRAF and EZH2 genes only in tumor clones. BCOR gene was used as a control. Patient “normal” “TP53hom” clone has no any CNVs in comparison with the donor “TP53het” clone, which excludes its leukemia/pre-leukemic nature. E CNV profile of all detected regions in “TP53hom” clone. Patient “TP53hom” clone has no BRAF and EZH2 gene loss. F FISH-plot analysis of clonal evolution and subclone distribution (%). The plot shows the clonal evolution pattern of JMML. Each color represents a clone. The patient showed residual leukemic cells at MRD time point (day 30 after HSCT) with emergence of donor-derived clone. Moreover, patient’s “normal” clone for analyzed SNVs was detected across both time points.

The patient was treated with palliative chemotherapy, including targeted therapy. However, the patient died of sepsis/organ failure due to progressive disease.

Patient 2

JMML was diagnosed in the second patient at the age of 1 month. Genetic analysis of the bone marrow using targeted DNA sequencing revealed a missense mutation c.38 G > A (p.G13D) in the KRAS gene (VAF of 48%). Considering the absence of adverse risk factors, low-dose chemotherapy was initiated, and a partial response was achieved. At the age of 1.5 years, disease progression occurred with an increased leukocyte count and splenomegaly. Cytogenetic analysis showed monosomy on chromosome 7. The percentage of blast cells in the bone marrow was 12%. A further increase in the blast cell count was observed and transformation into sAML was confirmed (bone marrow blast count is 50%).

We performed scDNA-seq of a bone marrow sample, which revealed two clones with heterozygous (87% of cells – clone sAML/JMML No1) and homozygous (5.3% of cells – clone sAML/JMML No2) KRAS c.38 G > A mutations, as well as a population of WT cells (7.7%; Table 3). No additional genetic aberrations, including copy number alterations, were detected (Fig. 3A). Prior to HSCT, the patient received two high-dose chemotherapy (FLAM: fludarabine, Ara-C, mitoxantrone, and FLAE: fludarabine, Ara-C, VP-16) to reduce the tumor burden. Myeloinfusion was performed from a partially compatible related donor with TCRαβ+ and CD19+ graft depletion. The conditioning regimen included treosulfan, melphalan, plerixafor, and venetoclax. Engraftment was observed on day +12. There were no signs of MRD persistence by MFC in the posttransplantation period. The level of donor chimerism determined by STR analysis was not less than 99%. However, at 180 days after HSCT, the level of lineage-specific chimerism in the CD34⁺ population was 13.7% of the patient’s cells while the whole marrow donor chimerism was 99%. Using scDNA-seq, we genotyped 6090 cells, three of which carried the KRAS c.38 G > A mutation in a heterozygous state (0.05%; Table 3). The average ADO value (4–5%) by one SNV was significantly higher than that suggested by the analysis of the patient’s own chimerism. Considering this fact, two polymorphisms with different zygosity in donor and patient cells were used simultaneously to distinguish genotyped cells by origin, reducing the probability of combined ADO in 1.5–2 orders in both analyzed alleles. The FLT3 c.2541+58 A > G (V1) and FLT3 c.2053+85_2053+88del (V2) polymorphisms were chosen as markers, which were heterozygous in donor cells and homozygous in patient cells, including tumor cells, as found in the sample before HSCT. The results of the clonal analysis are shown in Table 4. The presence of clones with different zygosity according to V1/V2 hom/het variants and a clone with the V1wt/V2wt genotype allowed us to predict with great accuracy the appearance of cells with the patient’s genotype (V1hom/V2hom) as a result of ADO. Thus, we can assume that the actual number of cells with the patient genotype was approximately 15-20 ( ~ 0.45-0.5%) among all 3239 cells reliably genotyped by these variants, which is consistent with the results of STR analysis when extrapolated to the entire bone marrow population.

A CNV spectrum before HSCT. Hierarchical clustering heatmaps shows no differences in CNV between tumor clones and WT clone. B FISH-plot analysis of clonal evolution and subclone distribution. The plot shows the clonal evolution pattern. Each color represents a clone. HSCT led to a decrease in the tumor clones and the WT clone. At the same time, a small percentage of cells originating from the patient remained.

According to the obtained findings and due to the high risk of leukemia recurrence, the patient underwent immunotherapy with donor lymphocyte infusion. Complete donor and lineage-specific chimerism was achieved. At the last follow up, the patient was in complete remission. FISH-plot and bar-plot analyses of disease evolution based on two time points (before and after HSCT) are shown in Fig. 3B.