Monitoring of minimal residual disease in early T‐cell precursor acute lymphoblastic leukaemia by next‐generation sequencing

Abstract

Early T-cell precursor acute lymphoblastic leukaemia (ETP-ALL) is a recently identified stem-progenitor-originated malignancy with a high risk of treatment failure (Coustan-Smith et al, 2009). Diversified genetic alterations have been observed at the onset of ETP-ALL, whereas common disease-specific mutations have not been defined. Mutations with myeloid features such as FLT3 and NRAS are more often found in ETP-ALL than in other T-ALLs, whereas prototypical T-ALL lesions, such as CDKN2A/B deletions and NOTCH1 mutations, are less frequent (Haydu & Ferrando, 2013). Thus, identification of clonal and sub-clonal mutations in each case and subsequent monitoring of minimal residual disease (MRD) by utilizing Next-Generation Sequencing (NGS) are expected to improve the diagnosis and survival of ETP-ALL patients. Monitoring of MRD, crucial for planning therapeutic strategy and preventing relapse of leukaemia, focuses on particular regions of the genome in which leukaemia-specific gene mutations are localized, and prior identification of disease-related mutations is necessary for this. Therefore, NGS is a powerful means for MRD detection. A 33-year-old woman with no subjective symptoms was found to possess a considerable number of circulating blasts (Table S1) at the time of enrolment into a cohort study by the Tohoku Medical Megabank organization (ToMMo) and referred to Tohoku University Hospital. Initial physical examination and laboratory data were unremarkable except for an increased serum lactate dehydrogenase level. However, her bone marrow was hypercellular and 91·8% of the cells were myeloperoxidase-negative blasts (Fig 1A). Based on the characteristics of the blasts (see supplementary document), the patient was diagnosed with ETP-ALL. After obtaining informed consent from the patient, whole genome sequencing (WGS) was performed to compare her admission samples of blast-rich blood DNA and buccal mucosa DNA. WGS analyses were conducted using our standard protocols (Katsuoka et al, 2014; Motoike et al, 2014). This comparison identified approximately 500 discordant calls in exons as candidate somatic mutations during leukaemogenesis. Representative data are shown in Fig 1B. In addition, we detected several chromosomal gains and losses in primary leukaemic cells (Fig S1). The discordant calls contained non-synonymous mutations in the genes known to be frequently altered in ETP-ALL blasts (Zhang et al, 2012). For further analyses, we chose 12 somatic mutations that were known as oncogenic mutations (Fig 1B). The variant calls for these 12 mutations were verified by the integrative genome viewer and polymerase chain reaction (PCR)-direct sequencing (Fig 1C, D). The initial treatment with hyper-CVAD (fractionated cyclophosphamide/vincristine/doxorubicin/dexamethasone) regimen failed (bone marrow blast count = 57·2%: Fig 2A). Therefore, treatment was switched to the HAM (high-dose cytarabine/mitoxantrone/dexamethasone) regimen, which achieved hypoplastic bone marrow without residual blasts (Fig 2A). The patient then received a conditioning regimen (cyclophosphamide and total body irradiation) prior to an allogeneic peripheral blood stem cell transplant (allo-PBSCT) from a human leucocyte antigen-identical male sibling donor. Neutrophil engraftment was confirmed on Day 11 and 99·8% donor chimerism of peripheral blood leucocytes was verified on Day 14 after transplantation (Fig 2A and data not shown). She has maintained complete haematological remission (CHR) for more than 1 year following allo-PBSCT (Table S1). We performed NGS analyses of subclonality utilizing a paired sample of the leukaemic cells obtained at two separate time points; one at admission and one just prior to the initiation of HAM therapy (referred to as T0 and T1, respectively: Fig 2A). In these analyses, we employed 9 sets of PCR amplicon-sequencing with MiSeq sequencer (Illumina, San Diego, CA, USA). It has been reported that the raw error rate of MiSeq sequencing is generally around 0·1% (Loman et al, 2012). Mutant reads of the 9 genes were detected at similar ratios in both T0 and T1 samples (Table S2), suggesting that there is no sub-clonal difference among the nine mutated genes selected. To monitor MRD, we then prepared genomic DNA from the bone marrow cells three months after allo-PBSCT (T2 sample), and collected further bone marrow samples at 6, 9 and 12 months after transplantation (T3, T4 and T5, respectively). These DNA samples were subjected to MiSeq sequencing of the NRAS G13D. We found as few as 27 mutant NRAS (G13D) allele reads out of 269,439 total reads in T2 samples and T3 to T5 showed almost similar results (Fig 2B). The frequencies of mutant allele in T2 to T5 samples were almost equivalent to that in the negative control sample (33/262,921 total reads of the mutated NRAS sequence were detected). Therefore, we deduced that the patient had achieved molecular CHR based on the reported criteria (Beldjord et al, 2014) after her second intensive chemotherapy, and maintained this condition for over 1 year following allo-PBSCT. In this study, abnormal blood smear data was obtained at the baseline survey of a prospective cohort and the data were not anonymized to our clinician. In this way, we were able to provide these incidental findings identified during the cohort study to the participant (Jarvik et al, 2014). This approach has been planned in the ToMMo project and approved by our Institutional Review Board. We believe that, in this manner, prospective cohort research is able to contribute to the potential problems of participants. We executed extensive NGS-based genomic analyses to monitor MRD. Determining candidate target genes by WGS prior to chemotherapy seems to be a powerful approach for monitoring leukaemia MRD, especially if disease-specific mutations have not been identified. Indeed, while some somatic alterations have been identified in each ETP-ALL case (Haydu & Ferrando, 2013; Neumann et al, 2013), characteristic mutations common for ETP-ALL have not been identified at present. It is known that leukaemic genomes are highly heterogeneous. Whereas leukaemia often consists of multiple subclonal leukaemic cells, the present study has identified a unique leukaemic clone possessing nine target mutations. Based on this experience, we surmise that WGS analysis for identification of multiple target genes assures precise estimation of the subclonality and identification of an appropriate target for MRD monitoring. The MiSeq sequence approach exploited in this study has an advantage over the digital droplet PCR or related methods in that it does not limit the type of mutation sequences. In conclusion, the combination of initial WGS analysis and NGS follow-up provides a highly promising means to search for candidate genes responsible for MRD in genetically heterogeneous landscapes, such as ETP-ALL. The NGS follow-up approach is also effective for MRD monitoring. The present result supports the potential of WGS and combinatorial NGS analyses in the personalized treatment of leukaemic diseases. The authors would like to thank Ai Hachiya, Nozomi Hatanaka, Natsumi Konno, Kiriko Nozoe, Yukie Oguma, Ayako Okumoto, Noriko Takahashi, Naomi Inagaki, Keiko Tateno, Shin Ito and Satoshi Nishikawa for technical assistance. We would like to thank members of Tohoku Medical Megabank Organization (ToMMo) at Tohoku University for seminal contribution to the establishment of the genome cohort and biobank, and for help with the genome analyses. The member list of ToMMo at Tohoku University is available at http://www.megabank.tohoku.ac.jp/english/a150601/. This work was supported in part by the MEXT Tohoku Medical Megabank Project, Japan Agency for Medical Research and development, AMED, and the Centre of Innovation Program from Japan Science and Technology Agency, JST. JYa, HH and MY designed the study. XP, SS, FK, YK, ID and RSa performed DNA analyses. NN, YS, KK and MN performed the bioinformatics. NFuk, SH, YO, AK, HK, FN, AH, and HH contributed the samples and clinical data. NFus, and RSh helped to draft the manuscript. XP, JYa, SS, SH, NFuk, HH and MY wrote the manuscript. The authors declare no conflict of interest. Fig S1. molecular karyotyping of the blast cells at the onset. Table S1. peripheral blood counts and follow-up bone marrow FISH data. Table S2. subclonality analyses. Tables S3 and S4. Information of PCR conditions. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.