Multi-Lineage BCR-ABL Expression in Philadelphia Chromosome-Positive Acute Lymphoblastic Leukemia Is Associated With Improved Prognosis but No Specific Molecular Features

Background Recently, various blood cell lineages expressing the BCR-ABL fusion gene in Philadelphia chromosome (Ph)-positive acute lymphoblastic leukemia (ALL) have been reported. However, the biological and clinical significance of these BCR-ABL lineages has not been established; therefore, we aimed to clarify the impacts of these different BCR-ABL-expressing lineages. Patients Multi-lineage BCR-ABL expression (multi-Ph) was defined as BCR-ABL expression outside of the B-lineage compartment, as determined by fluorescence in situ hybridization (FISH) in peripheral blood neutrophils and bone marrow clots, and flow cytometry-sorted polymerase chain reaction (PCR). We analyzed IKZF1 deletion patterns by PCR, examined gene expression profiles using RNA sequencing, and compared treatment outcomes across different BCR-ABL-expressing lineages. Results Among the 21 multi-Ph patients in our 59-patient cohort (36%), BCR-ABL expression was detected at the multipotential progenitor level. However, no IKZF1 deletion patterns or gene expression profiles were identified that were specific for multi-Ph. However, multi-Ph patients were found to have better survival rates than patients with uni-lineage BCR-ABL expression [event-free survival (EFS): 74 vs. 33%, P = 0.01; overall survival (OS): 79 vs. 44% at 4 years, P = 0.01]. In multivariate analyses, multi-Ph was identified as a good prognostic factor for both EFS and OS. Conclusion We confirmed that more than one-third of Ph+ALL patients could be classified as mutli-Ph. Although no specific molecular characteristics were identified for multi-Ph, this phenotype was associated with better treatment outcomes.


INTRODUCTION
Cells that express BCR-ABL outside of the B-lineage compartment have been reported in a proportion of patients with Philadelphia chromosome-positive (Ph+) B-precursor acute lymphoblastic leukemia (ALL) (1)(2)(3)(4). In most cases, the fusion gene BCR-ABL, which is the genetic abnormality associated with Ph+ALL, is expressed only in B-lymphocytes; however, in some cases, BCR-ABL was also expressed in the myeloid lineage, suggesting that these cells originated from a separate differentiation stage than B-lineage cells.
The expression of BCR-ABL by non-lymphocytic lineages is a relatively recent finding, and the biological and clinical significance of this finding has yet to be clarified. The differentiation stage during which the genetic abnormality occurs must be identified, and whether leukemic cells originating from various differentiation stages are molecularly heterogeneous must be determined. Furthermore, whether similar genetic abnormalities with different origins have varying influences on patient prognosis must also be determined. The occurrence of genetic abnormalities in cells other than leukemic cells may also affect the detection of minimal residual disease (MRD) (3,4), which is currently crucial for determining an optimal treatment strategy for Ph+ALL (4)(5)(6)(7). In this study, we confirmed two types of adult Ph+ALL that derive from different leukemic origins. In addition, we analyzed the genetic characteristics and therapeutic responses of Ph+ALL associated with genetic abnormalities in cells other than leukemic blast cells.

Patients
Fifty-nine adult Ph+ALL patients were deemed eligible for this retrospective study. Patients with de novo Ph+ALL were included if peripheral blood or bone marrow samples were available. No patients had any history of chronic myeloid leukemia (CML). Ph+ALL was confirmed through the detection of the Ph chromosome by fluorescence in situ hybridization (FISH) analysis and the positive detection of BCR-ABL fusion transcript expression by reverse-transcription-polymerase chain reaction (RT-PCR) analysis. The immunophenotyping of leukemic blast cells was performed using flow cytometry or immunohistochemistry, at individual hospitals (7). The copy numbers of BCR-ABL transcripts identified in each sample were normalized against the expression levels of GAPDH. The threshold for quantification was 50 copies/µg RNA, which corresponded to a minimal sensitivity of 10 −5 (8). In the present study, the inability to detect BCR-ABL by RT-PCR was defined as complete molecular remission (CMR). All patients were treated with tyrosine kinase inhibitor (TKI) ± conventional chemotherapy. No criteria were established for TKI selection or the performance of allogeneic hematopoietic cell transplantation (allo-HCT). Treatment strategies depended on each physician's judgments and the treatment policy of each hospital. Representative treatments were based on studies performed by the Japan Adult Leukemia Study Group for Ph+ALL [Ph+ ALL202 (9), Ph+ALL208 (10), or Ph+ALL213 (11)], and written informed consent was obtained from each patient upon treatment and sample collection. Each hospital's institutional review board approved the study.

Fluorescence In Situ Hybridization of the BCR-ABL Fusion Gene
To assess the expression of the BCR-ABL fusion gene outside of leukemic blast cells, we performed FISH analyses on two different specimens: peripheral blood (PB) neutrophils (PN-FISH) (12) and bone marrow clot sections (BM-FISH) (13). A Vysis LSI BCR/ABL ES Dual-Color Translocation Probe (Abbott Laboratories, Abbott Park, IL, USA) was hybridized with PB and BM samples (2,14). The denaturation of the chromosome/probe, hybridization, and washing were performed according to the manufacturer's instructions. Multi-Ph was defined as the detection of BCR-ABL fusion gene signals in segmented neutrophils, including of PB mononuclear cells and PB neutrophils. Similarly, the detection of BCR-ABL fusion gene signals in myeloperoxidase-positive cells and polynuclear megakaryocytes in 10% formalin-fixed paraffin-embedded bone marrow clot sections by BM-FISH was defined as multi-Ph.

RT-PCR Analysis
Total RNA was extracted from sorted cells, using an RNeasy MiniKit or a QIAamp RNA Blood Mini Kit (QIAGEN, Hilden, Germany), and reverse transcribed, using the SuperScript II Reverse Transcriptase Kit (Thermo Fisher Scientific, Waltham, MA, USA), according to the manufacturer's instructions. BCR-ABL transcript expression levels were quantitated based on a real-time fluorescence detection method using an ABI Prism7300 Sequence Detection System and Taqman ® Gene Expression Assay probe (Applied Biosystems, Foster City, CA, USA). GAPDH served as a control for cDNA quality. To detect IKZF1 isoforms, IKZF1 transcripts were amplified from cDNA, as previously described (16). BCR-ABL fusion transcripts that were detected outside of the lymphoid compartment, using a flow cytometry-sorted PCR method (FCM-PCR), were classified as multi-Ph.

RNA Sequencing
All cryopreserved samples from which high-quality RNA could be extracted were submitted to RNA sequencing. We prepared non-directional sequencing libraries, using a NEBNext Ultra II RNA Prep Kit for Illumina and a NEBNext Poly(A) mRNA Magnetic Isolation Module (New England Biolabs, Ipswich, MA, USA), according to the manufacturer's instructions. We ran the prepared libraries using a HiSeq 2500 next-generation sequencing platform (Illumina, San Diego, CA, USA) to obtain an average of 100 million reads per sample. The obtained reads were aligned against the hg19 reference genome, using TopHat2 (17), and fragments per kilobase of exon per million mapped reads (FPKM) values were calculated using Cufflinks (18). For clustering analysis, we used HTSeq (19) for raw read counting and DESeq2 (20) for variance stabilizing transformation. Cluster 3.0 (21) was used to filter genes with default parameters. Finally, we used pyclust (22) to perform average linkage hierarchical clustering using Pearson's correlation coefficient. Bootstrapping was performed for 1,000 permutations to calculate the bootstrap probability and to obtain approximately unbiased values.

Survival Analysis
Primarily, we analyzed event-free survival (EFS), from the date of diagnosis to the date of any event, including death by any cause, failure to achieve remission after induction therapy, relapse in any site, and second malignancy (23)(24)(25); for patients without an event, the date of the last visit was used. EFS probabilities were estimated using the Kaplan-Meier method, and P-values were calculated using a log-rank test (26,27). We also analyzed overall survival (OS), from the date of diagnosis to the date of death or the date of the last visit. Univariate and multivariate analyses were performed using a Cox proportional hazard regression model (28). A backward stepwise procedure was used to develop a final model, based on a P-value threshold of 0.05. The covariates considered were patient sex (male vs. female), patient age (<65 y vs. ≥65 y), white blood cell count at diagnosis (< 30,000/µl vs. ≥ 30,000/µl), BCR-ABL transcript [(e1a2 vs. b2a2 or b3a2 (b2a2/b3a2)], additional cytogenetic abnormality (no vs. yes), type of TKI (imatinib vs. dasatinib), allo-HCT (no vs. yes) and BCR-ABL lineage (uni-lineage vs. multi-lineage). A significance level of P <0.05 was used for all analyses.

Expression of BCR-ABL Outside of Leukemic Blast Cells
In this study, patients were classified as multi-Ph through at least one of three approaches (PN-FISH, BM-FISH, and FCM-PCR). All other patients were defined as uni-lineage Ph leukemia (uni-Ph). PN-FISH was performed for 19 patients ( Figure 1A), BM-FISH was performed for 49 patients ( Figure 1B), and FCM-PCR was performed for six patients. A total of 21 patients (36%) were identified as multi-Ph; a discrepancy between detection methods was observed in only one patient (PN-FISH: uni-Ph, BM-FISH: multi-Ph). In two patients, BCR-ABL signals were also detected in megakaryocytes ( Figure 1B-2). Four multi-Ph individuals expressed e1a2, whereas the remaining 17 individuals with multi-Ph expressed b2a2/b3a2. The frequency of multi-Ph was significantly higher among patients expressing b2a2/b3a2 than those expressing e1a2 (81 vs. 11%, P < 0.001, Table 1).
Fraction Expressing BCR-ABL and IKZF1 Deletion FCM-PCR was evaluated in six patients (4 with e1a2 and 2 with b2a2/b3a2). BCR-ABL fusion transcripts were detected in the LAIP 20and LAIP 20+ compartments, as expected, but the HSCs/MPP and Pro compartments were also BCR-ABL-positive in four patients (2 with e1a2 and 2 with b2a2/b3a2).
IKZF1 deletion was detected in five of six evaluated patients (4 with e1a2 and 1 with b2a2/b3a2). No deletion patterns were found associated with multi-Ph ( Figure 2). In three patients, IKZF1 deletion was detected in the HSCs/MPP and Pro compartments, as well as in the LAIP 20and LAIP 20 + compartments, which was consistent with the findings for BCR-ABL-positive compartments.

Gene Expression According to Ph Lineage
The expression profiles of leukemic blast cells were assessed by performing a hierarchical clustering analysis, with bootstrapping, in ten patients (uni-Ph: 5, multi-Ph: 5). Blast cells derived from patients with multi-Ph were not grouped into a cluster that was distinguishable from those derived from patients with uni-Ph, suggesting that the global expression profiles of uni-Ph and multi-Ph blast cells were similar (Figure 3, Supplementary Table).

Treatment Response According to Ph Lineage
Both EFS and OS were significantly higher in patients with multi-Ph than in patients with uni-Ph (EFS: 74 vs. 33%, P = 0.01; OS: 79 vs. 44% at 4 years, P = 0.01, Figure 4). However, the CMR rate following the first induction therapy was significantly lower in patients with multi-Ph than in those with uni-Ph (11 vs. 56%, P = 0.003). In two patients with multi-Ph, BM-FISH at complete hematological remission showed positive BCR-ABL signals only outside of leukemic blast cells. In multivariate analyses, both BCR-ABL lineage and age were identified as significant prognostic factors for both EFS and OS ( Table 2). Neither the type of TKI nor allo-HCT were found to be significant prognostic factors.

DISCUSSION
In this study, we used multiple approaches to confirm the heterogeneity of adult Ph+ALL according to BCR-ABL expression lineages (1,2). BCR-ABL signals were observed outside of leukemic blast cells in 36% of patients with Ph+ALL.
We detected BCR-ABL signals not only in myeloid cells but also in megakaryocytes, which suggested the involvement of early differentiation stages in multi-Ph, including those that occur close to HSCs (29).
Three techniques were used to identify multi-Ph in this study: PN-FISH, BM-FISH, and FCM-PCR. Therefore, even in cases for which multi-Ph was difficult to determine using one method, the complementary use of multiple approaches allowed for the identification of multi-Ph. Most cells were classified as leukemic blast cells at diagnosis, and morphological judgments alone were not sufficient for the identification of fractions other than leukemic blast cells. Myeloperoxidase double staining was effective in such cases. Although FCM-PCR is a sensitive method, the number of cases we analyzed remained small because cryopreserved cells tended to feature too many dead cells to obtain an adequate signal.
To clarify the biological differences between uni-Ph and multi-Ph, we analyzed the deletion pattern of IKZF1, the most frequent genetic abnormality associated with Ph+ALL (16). However, no distinct IKZF1 deletion patterns were associated with multi-Ph. Gene expression analysis also failed to identify any clusters specific to multi-Ph. In addition to the limited number of samples included in this analysis, the lack of differences in gene expression may be associated with the use of stored samples, which were obtained at diagnosis and primarily contained leukemic blast cells. Because the only difference between uni-Ph and multi-Ph is the process through which the patient develops leukemia, any differences in the expression profile may disappear once a cell differentiates into a leukemia blast cell. Therefore, the analysis of other cell types or different differentiation stages between uni-Ph and multi-Ph may be necessary to observe genetic phenotypes. Survival rates were significantly different between patients with uni-Ph and those with multi-Ph, consistent with a previous smallscale study (2). The better survival rate observed for patients with multi-Ph may be due to differences in the cell differentiation stage affected by TKI treatment between the two lineages, instead of specific genetic differences in leukemic cells. Leukemic blast cells are thought to derive from short-term HSC (ST-HSC) or the MPP level of development in multi-Ph, whereas they derive from common lymphoid progenitors (CLP)/lymphoid-primed multipotent progenitors (LMPP) in uni-Ph (30,31). TKI treatment, alone, demonstrated a weak effect on the elimination of quiescent ST-HSCs/MPPs; however, when combined with chemotherapy, TKI may activate quiescent HSCs (32), resulting in the better eliminating of leukemic stem cells (33). In contrast, uni-Ph ALL, which is thought to originate from CLPs/LMPPs, includes ALL derived from aggressive B-1 progenitor cells, which is associated with poor TKI treatment outcomes (34,35). This study has several limitations. First, the number of samples that could be analyzed was limited. Because all available samples were used, the selection was not arbitrary; however, a limited selection of samples was available for data collection, such as few samples from remission cases. In addition, the MRD data were only based on PCR for BCR-ABL, and no MRD data associated with Immunoglobulin (Ig)/T-cell receptor (TCR) gene rearrangements were collected in this study. In Japan, Ig/TCR analysis is not commonly performed to detect MRD in adult Ph+ALL patients. Discrepancies in MRD data between BCR-ABL expression and Ig/ TCR analyses can indicate the presence of BCR-ABL expression outside of lymphoblasts (3,4). The histopathological distinction between blast cells and other blood cells using tissue FISH could complement the lack of Ig/TCR data by detecting the expression of BCR-ABL in cell types other than leukemic blast cells.
In conclusion, we confirmed BCR-ABL expression outside of leukemic blast cells in more than one-third of the Ph+ALL patients examined in this study. Although unique genetic patterns could not be identified, these patients revealed good prognoses following treatment, including TKI treatment.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Institutional review board of Nagoya University. The patients/participants provided their written informed consent to participate in this study.