Introduction

The Philadelphia chromosome (Ph), a derivative chromosome 22 with a chimeric BCR–ABL1 fusion derived mostly by a reciprocal t(9;22)(q34.1;q11.2), occurs in 95% of chronic myeloid leukemia (CML) [1], 2–10% of pediatric and 25% of adult acute lymphoblastic leukemia [2], and a small subset of acute myeloid leukemia [3]. BCR–ABL1 fusion is a diagnostic hallmark of CML, however, 5–10% of CML patients at presentation lack a Ph by karyotype analysis, but are shown to carry BCR–ABL1 as detected by either fluorescence in situ hybridization (FISH) or reverse transcription polymerase chain reaction (RT-PCR). Further studies have shown that most of these cases are caused by a three-way or even more complicated multi-way translocation involving chromosomes 9, 22, and one or more other chromosome(s). In these complex translocations, which are designated as “variant Ph” or “complex Ph” in the literature, BCR–ABL1 is still located on the derivative chromosome 22 or der(22) [4,5,6]. Up to 50% of these cases with variant or complex Ph exhibit a del(9q) of the derivative chromosome 9 or der(9), compared with that of 12–15% of cases with a classic Ph [7,8,9]. The del(9q) has been considered to be a marker for a poor prognosis in CML patients [7,8,9,10,11,12,13,14,15], but this observation is not supported by recent observations [16,17,18].

A small subset of CML cases exhibit a normally appearing chromosome 22, but positive for BCR–ABL1 fusion by FISH and/or RT-PCR, and these cases are often designated as “masked Ph” or “cryptic Ph” [4,5,6, 8]. Two underlying mechanisms have been previously proposed to explain this phenomenon: First, a cryptic insertion between chromosomes 9 and 22 may occur, manifested as either a partial ABL1 from a chromosome 9 inserted into the BCR locus on a chromosome 22, e.g., ins(22;9)(q11.2;q34.1q34.1) referred to as ins(22;9), forming a BCR–ABL1 located on the der(22) or vice versa, e.g., ins(9;22)(q34.1;q11.2q11.2, referred to as ins(9;22) with a BCR–ABL1 located on the der(9). This mechanism is also considered as a main route for this type of BCR–ABL1 in CML. Secondly, two or more sequential translocations between chromosomes 9 and 22 may take place that result in the affected chromosomes 9 and 22 having a normal appearance [8, 19]. Either a “masked Ph” or a “cryptic Ph” is apparently inadequate to describe those cases with BCR–ABL1 located on a chromosome 9; the term of Ph negative, BCR–ABL1 positive CML has been applied to more accurately describe these cases [20,21,22].

Fewer than 80 cases of CML with insertion-derived BCR–ABL1 have been reported since the first case was described in 1981 [23], mainly in the form of case reports or limited case series (up to nine cases) [6, 19, 21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55]. Some early studies performed only karyotyping or chromosomal analysis plus interphase FISH (i-FISH). However, these methods may be insufficient for determination of an insertion-derived BCR–ABL1 [23, 38, 40]. The frequent location of BCR–ABL1 varies from der(9) or der(22) in previous reports [6, 19, 20, 22, 47, 55]. The prognostic significance of insertion-derived BCR–ABL1 in CML patients remains unknown.

In this study, we present 41 new CML cases with insertion-derived BCR–ABL1 confirmed by various types of FISH tests. The clinical, cytogenetic, and molecular features of these cases are investigated and discussed.

Materials and methods

Case selection

We searched the database of the Clinical Cytogenetics Laboratory at The University of Texas MD Anderson Cancer Center, from May 1, 2001 through October 31, 2019, for all cases with a positive BCR–ABL1 by FISH that occurred as an insertion of either BCR into ABL1 or ABL1 into BCR. Within this time frame, the tyrosine kinase inhibitors such as imatinib, dasatinib, nilotinib, bosutinib, and ponatinib have been administrated to treat BCR–ABL1 positive patients. Cases with positive BCR–ABL1 FISH and/or RT-PCR result(s) but without further confirmation that the BCR–ABL1 originated by an insertion were excluded from this study. Clinicopathologic and laboratory data were collected by electronic medical chart review. This study was approved by the Institutional Review Board, and was conducted in accordance with the Declaration of Helsinki.

Chromosomal analysis

As reported previously [56, 57], conventional G-banded chromosomal analysis was performed routinely on unstimulated 24 and 48 h bone marrow (BM) aspirate cultures using standard techniques. Usually, 20 metaphases are analyzed for each case, and the final results are reported following the International System for Human Cytogenomic Nomenclature guidelines [58]. A complex karyotype (CK) is defined as ≥3 chromosomal abnormalities with at least one being a structural abnormality. A balanced ins(22;9) or ins(9;22) is usually considered as one chromosomal abnormality, whereas unbalanced der(9) and der(22) are usually considered as two separate chromosomal abnormalities.

Fluorescence in situ hybridization (FISH) analysis

The Vysis LSI BCR–ABL1 ES Dual Color Fusion probe set (referred to as “ES probe”) (Abbott Molecular, Des Plaines, IL) was employed as a routine BCR–ABL1 fusion test in our laboratory. This assay is capable of distinguishing the major breakpoint (M, p210 BCR–ABL1 transcript) and minor breakpoint (m, p190 BCR–ABL1 transcript) when a typical signal pattern is observed: two-red, one-green, and one-fusion (2R1G1F) for p210 and one-red, one-green, and two-fusion (1R1G2F) for p190 [59]. For cases with atypical signal pattern(s), especially those cases with a suspicion of concomitant del(9q) in the der(9), the Vysis BCR/ABL1/ASS1 Tri-Color DF FISH probe set (referred to as “tricolor probe”) (Abbott Molecular, Des Plaines, IL) was performed as a reflex test [59]. Both types of BCR–ABL1 FISH probe sets were fully validated before their utilization for clinical services.

Several types of FISH tests were performed in this study, encompassed as “intensive FISH studies”. i-FISH: analysis of 200 or more (up to 500 if necessary) interphase cells with various FISH signal patterns, including both normal and abnormal signal patterns, which likely represent the heterogeneity of abnormal clones in a specimen. Metaphase FISH (m-FISH): analysis of FISH signals on metaphase cells, which usually provides the chromosomal location of each signal. Map-back FISH (mb-FISH): performing FISH tests on previously G-banded/karyotyped slide(s) so that a correlation of m-FISH and karyotype results can be established. The latter test is extremely useful for identification/confirmation of FISH signal location in cases with un-identifiable chromosome(s) (marker chromosomes) and/or cryptic chromosomal abnormalities. Whole chromosome painting (wcp): a mixture of FISH probes targeting a whole chromosome is applied to confirm/exclude a subtle rearrangement between two or even among three or more chromosomes. All of these types of FISH tests were performed by following standard protocols as reported previously [59].

Quantitative BCR–ABL1 real-time RT-PCR assay

A multiplex real-time RT-PCR assay was employed to quantitatively assess BCR–ABL1 levels. This assay simultaneously detects the common BCR–ABL1 transcripts e14a2, e13a2, and e1a2 [60]. Briefly, RNA was extracted from BM or peripheral blood (PB) specimens using Trizol reagent (Gibco-BRL, Gaithersburg, MD) according to the manufacturer’s instructions. Reverse transcription (RT) was performed on total RNA (1 µg) using random hexamers and superscript II reverse transcriptase (Gibco-BRL). The resulting cDNA was then subjected to PCR to amplify BCR–ABL1 transcripts on an ABI PRISM 7700 Sequence Detector (Applied Biosystems, Foster City, CA) [60]. The quantitative BCR–ABL1 mRNA levels were expressed as the percent ratio of BCR–ABL1 to ABL1 transcript levels. The sensitivity of this assay is between 1 in 10,000 and 1 in 100,000.

Morphological evaluation

Both diagnostic PB and BM samples for each case were reviewed. The white blood cell counts, hemoglobin (Hb) and platelet counts, percentages of blasts and neutrophils in the PB, and the percentages of blasts and granulocytes in the BM as well as the BM cellularity were collected.

Gene mutation analysis

Gene mutation analysis was performed using DNA extracted from BM aspirate samples in a subset of patients. Due to technologic advances and updating of assay equipment over time, various techniques were employed for mutation analysis: Sanger sequencing for ABL1 and other (e.g., JAK2 and KIT) mutation(s) and next-generation sequencing (NGS) using 28-gene, 53-gene, and 81-gene panels, where ABL1, JAK2, and KIT were constantly included in these panels. The NGS-based mutation analysis was performed using the Illumina MiSeq (Illumina, San Diego, CA) sequencer [61]. FLT3 gene mutation analysis, including internal tandem duplication, ITD, and D835 point mutation, was assessed by PCR followed by capillary electrophoresis on a Genetic Analyzer (Applied Biosystems, Foster City, CA) [61].

Statistical analysis

Overall survival (OS) was defined as the time from the first diagnosis of CML to death or the last follow-up date in this study. The Kaplan–Meier curves were applied to estimate unadjusted OS durations. The Log-rank (Mantel–Cox) test and the Gehan–Breslow–Wilcoxon test were used to compare OS between groups. A student t test was applied to perform all univariate analyses. A chi-square (X2) test was applied to compare the frequencies of different groups. A result was considered statistically significant if p < 0.05. All computations were conducted in GraphPad Prism 8.

Results

Clinical and laboratory findings, treatment response, and outcomes

In this study, 41 cases were identified as Ph negative, BCR–ABL1 positive, in which the BCR–ABL1 fusion was derived from an insertion confirmed by a combination of various FISH tests (i-FISH, m-FISH, mb-FISH, and/or wcp). The study group included 20 men and 21 women with a median age of 47 years (range 12–78 years) at time of initial diagnosis. Thirty-one patients had chronic phase (CML-CP), one accelerated phase (CML-AP), and nine blast phase (CML-BP), respectively. Among patients with CML-CP, one patient (case #2) had concurrent metastatic melanoma and another patient (case #34) had a history of essential thrombocythemia. All CML patients received therapy with at least one tyrosine kinase inhibitor (TKI). Fourteen (34.1%) patients received additional chemotherapy and/or interferon-alpha (IFN-α) prior to or in combination with TKIs and four patients (9.8%) also received allogeneic hematopoietic stem cell transplant.

The National Comprehensive Cancer Network criteria were applied to evaluate the outcomes of the CML patients in this study [62]. Three patients were lost for follow-up and their outcomes are unknown. Continuous follow-up was available for the remaining 38 patients. The median followed-up interval was 63 months (range, 11–242 months) after the initial CML diagnosis. Response to TKIs, chemotherapy, or other interventions led to complete molecular response (CMR, n = 6), major molecular response (MMR, n = 8), complete cytogenetic response (CCyR, n = 4), partial cytogenetic response (PCyR, n = 5), relapse (n = 4), and death (n = 11) respectively (Table 1). Among the 31 patients with CML-CP, 2 died; 1 patient with CML-AP, and 8 patients with CML-BP died of disease. One of the patients who died with CML-CP had concurrent metastatic melanoma.

Table 1 General information and outcomes of patients in this study.
Full size table

The most recent hematologic laboratory findings in these groups of CML patients were compared (Table 2). Statistically significant differences in the percentages of blasts and granulocytes in the BM, and percentages of blasts and neutrophils, Hb levels and platelet counts in the PB were observed between these two groups consistent with their CML status and clinical presentation. No significant differences in male/female ratio and/or median age were observed in these two groups. The mean survival/follow-up was 96.6 months (range, 12–242 months) in the CML-CP group and 43.1 months (range, 11–201 months) in the CML-AP + CML-BP group. A significant difference in OS was observed between these two groups (p < 0.0001).

Table 2 Comparison of the most recent hematological changes in 38 CML cases by disease phases (CP-CML vs. AP-CML + BP-CML).
Full size table

Chromosomal analysis

Chromosomal analyses of all 41 cases revealed a variety of karyotypes in this cohort: Group 1 (n = 19), normal diploid karyotype; Group 2 (n = 5), abnormal karyotype with morphologically normal chromosomes 9 and 22; Group 3 (n = 2), apparent ins(9;22) which could mimic the classic t(9;22) but presented with different i-FISH and m-FISH signal patterns (see below); and Group 4 (n = 15), abnormal karyotype with apparent abnormal chromosomes 9, der(9) and/or abnormal chromosomes 22, der(22). Even the abnormalities in chromosomes 9 and/or 22 in group 4 cases might mimic a classic t(9;22), mostly depending on the resolution of available metaphases for analysis, but these results did provide a clue for further investigation to confirm or exclude the presence of BCR–ABL1. All cases in Group 4 had a CK whereas cases in the other three groups had a non-CK (except case# 36 in the Group 2) if determined by conventional cytogenetics only. Correlation between the karyotypes and BCR–ABL1 positivity confirmed by FISH testing (see below) showed that 17 (41.5%) cases had apparent aberrations involving chromosome 9 and/or 22 or non-cryptic der(9)/der(22) (Group 3 + Group 4), whereas 24 (58.5%) cases carried cryptic der(9)/der(22) (Group 1 + Group 2) in this study.

Among the 38 CML cases with follow-up, 23 (60.5%) cases had a cryptic der(9)/der(22), and the remaining 15 (39.5%) cases had an apparent der(9)/der(22). Interestingly, patients with CML-AP and CML-BP seemed to be more prone to exhibit cryptic der(9)/der(22) than patients with CML-CP, although this difference was not statistically significant (8/10 vs. 15/28, p = 0.28) (Table 3). OS analysis also did not show significant differences between these two groups (Fig. 1a). Cases with a CK (n = 14, all cases in Group 4 and case #36) and a non-CK (n = 24, cases in Groups 1–3 except case #36) (p = 0.97) also did not show a significant difference in OS (Fig. 1b), suggesting that karyotype results alone are inadequate to predict prognosis.

Table 3 Aberrations involving chromosomes 9 and/or 22 in 38 CML cases.
Full size table
Fig. 1: Overall survival comparison of CML patients with cryptic vs. non-cryptic der(9)/der(22) and complex karyotype (CK) vs. non-CK.
figure 1

a Cryptic der(9)/der(22) (n = 23) vs. non-cryptic der(9)/der(22) (n = 15). The mean survival/follow-up lengths were 78.2 months (range 12–242 months) in the group with cryptic aberration and 96.8 months (range 11–200 months) in the group with non-cryptic aberration respectively. No statistically significant difference for overall survival was observed between these two groups (p = 0.21). b Complex karyotype (CK) (n = 14) vs. non-CK (n = 24). The mean survival/follow-up lengths were 87.6 months (range 11–187 months) in the group with complex karyotype (CK) and 84.4 months (range 12–242 months) in the group with non-CK respectively. No statistically significant difference for overall survival was observed between these two groups (p = 0.97).

Full size image

FISH analysis

As mentioned above, the ES probe is used routinely for FISH testing for BCR–ABL1 in our laboratory. All 41 cases were tested positive with the ES probe, which were initially considered to be discordant with the karyotypes which did not show a Ph. Therefore, m-FISH or ideally mb-FISH was performed, which allowed direct observation and documentation of apparent FISH signal(s) on the metaphases necessary to conclude, at least at the first-time FISH analysis, that a cryptic chromosomal insertion was responsible for BCR–ABL1 in these cases [47, 59]. All cases had mb-FISH performed that also helped to identify the exact location of BCR–ABL1 in each case. The BCR–ABL1 was located on chromosome 22 in 28 (68.3%) cases, chromosome 9 in 12 (29.3%) cases, and chromosome 19 in 1 (2.4%) case. The latter case (case #22) had an ins(9;22) that formed BCR–ABL1 which was likely sequentially relocated to chromosome 19 through a translocation.

Twenty-six (63.4%) cases in this study exhibited a signal pattern of 2R1G1F using the ES probe, the same as that of the p210 transcript derived from the classic t(9;22). In the remaining 15 cases, regardless of the karyotype results, their signal patterns were considered as atypical for either p210 or p190 (1R1G2F) transcripts; the results most likely suggested a more complex chromosomal rearrangement than a simple and balanced insertion, for which mb-FISH with the ES probe could not further confirm [59, 63]. Therefore, mb-FISH with a tricolor probe was performed in eight cases [55]. In another seven cases, wcp was performed due to the complexity of chromosomal abnormalities simultaneously involving chromosomes other than 9 and/or 22. A balanced three-way or multiple-way translocation involving chromosomes 9 and 22 was suspected initially, but an insertion was finally determined to be the cause of BCR–ABL1 and complexity of chromosomal abnormalities, along with other simultaneous or sequential rearrangement in these cases. After correlating the chromosome analysis and all FISH test results in each case, the underlying causes of the atypical signal patterns mentioned above were attributed to more complicated chromosomal abnormalities undetected by conventional cytogenetics and interpreted as one the following: a cryptic deletion of 9q including ASS1 gene; a cryptic der(9)ins(9;22) with two normal chromosomes 22; and gain of an extra der(9) or an extra der(22) with BCR–ABL1 (see Supplementary Information, including the Supplementary Figure for detailed interpretation for each case).

Due to the additional chromosomal abnormalities detected by intensive FISH tests, we reclassified all cases into two new groups based on their complexity of chromosomal abnormalities defined by correlating both karyotype and FISH test results: 1. Cases harboring simple chromosomal abnormalities (SCAs, n = 15) that presented with a normal karyotype or a balanced ins(9;22) or ins(22;9) confirmed by FISH tests; 2. Cases harboring complex chromosomal abnormalities (CCAs, n = 23) that presented with an obvious CK or a normal karyotype or even a non-CK but with unbalanced ins(9;22) or ins(22;9) and other abnormalities confirmed by FISH, such as gain of extra copy of der(9) or der(22). These cases possibly could be categorized as a CK, providing that these chromosomal abnormalities could have been detected by conventional cytogenetics. Many cases previously assigned to Groups 1 and 2 (Table 3) were now categorized in the CCAs group, e.g., cases #4, #7, #8, #18,#38 and #41 with a normal karyotype and cases #11, #24 and #36 with cryptic der(9)/der(22) and additional chromosomal aberration(s). Interestingly, cases #12 and #39 with an apparent ins(9;22) previously assigned in Group 3 were now in the SCAs group due to a balanced ins(9;22) confirmed by FISH tests. Statistical analysis showed a significant difference in the OS between these newly assigned groups (SCAs vs. CCAs) in 38 cases with clinical follow-up (Fig. 2). Therefore, we conclude that conventional karyotype results cannot reflect the entirety of chromosomal aberrations and thus should not be used, by themselves, to stratify patient risk and/or predict outcomes. Instead, the complexity of chromosomal aberrations (SCAs vs. CCAs) detected by FISH should be applied as a predictive marker. Our study also demonstrated the clinical relevance of FISH tests in the workup of CML patients.

Fig. 2
figure 2

Overall survival comparison of CML patients with simple chromosomal abnormalities (SCAs, n = 15) vs. complex chromosomal abnormalities (CCAs) (n = 23) according to a correlation of both chromosomal analysis and FISH test results. The mean survival/follow-up lengths were 102 months (range 12–242 months) in the SCAs group and 75 months (range 11–201 months) in the CCAs group respectively. A statistically significant difference for overall survival was observed between these two groups (p = 0.004).

Full size image

RT-PCR analysis for BCR–ABL1

All cases tested were positive by RT-PCR for BCR–ABL1 transcripts: e13a2 in 18 (43.9%) cases, e14a2 in 14 (34.1%) cases, e14a2 + e13a2 in 7 (17.1%) cases, e1a2 in 1 (2.4%) case, and undetermined in 1 (2.4%) case. Therefore, 95.1% of cases resulted in a p210 transcript, and 1 case (case #36) had a p190 (e1a2) transcript. The quantitative real-time RT-PCR Assay for BCR–ABL1 played an essential role for determining treatment responses in this study, especially for those cases with a cryptic der(9) and/der(22), where cytogenetic response criteria are not applicable.

Gene mutation data

In this study, 23 cases were tested for ABL1 mutations: 18 by Sanger sequencing and 5 by NGS. Five (12.2%) cases (cases #5, #8, #19, #28 and #37) had mutations: p.E255K (cases #8 and #19), p.L298V (case #5), p.T315I (case #28) and p.P465S (case #37). Among them, 2 patients (one was CML-AP, case #8; another was CML-BP, case #19) died, whereas the remaining three (all CML-CP) achieved a MMR (case #5), CCyR (case #37), and PR (case #28), respectively. Due to the small size of cases with ABL1 mutations and application of two methods with different coverage of ABL1 gene in the mutation testing, the association of ABL1 mutations and outcomes cannot be analyzed in this cohort. It also remains unknown whether cases with an insertion-derived BCR–ABL1 are prone to ABL1 mutation(s). Eight cases were tested for FLT3 and all were negative for FLT3 ITD and D835 mutations. Two cases (cases #23 and #25) were tested for JAK2 and both were negative. Case #40 had an ASXL1 p.G646fs.

Discussion

To our knowledge, this study is the largest series of Ph negative, BCR–ABL1 positive CML cases with the BCR–ABL1 derived from an insertion. As shown in this study, an insertion-derived BCR–ABL1 occurs in CML rarely, in <1% of CML cases at our institution. Chromosomal insertions that result in BCR–ABL1 can take place in multiple ways as observed in this study. Approximately 70% of cases exhibited BCR–ABL1 located on a chromosome 22, which was either a cryptic der(22), or “cryptic Ph” or non-cryptic der(22) or “masked Ph”. The remaining cases had BCR–ABL1 located on a cryptic or non-cryptic der(9) and in one patient, a der(19). Although this is different from some earlier reports in which the authors reported more frequent or even predominant BCR–ABL1 located on der(9) [6, 22, 55], our results are similar to the sum of all cases reported previously (Table 4); BCR–ABL1 locations reported in 78 patients in the literature have been identified on der(22) in 49 (62.8%), der(9) in 25 (32.1%), der(1)in 1 (1.3%) and simultaneously on both der(9) and der(22) in 3 (3.8%). Therefore, including the cases in this report, 77 of 119 (64.7%) cases with an insertion-derived BCR–ABL1 possess a cryptic or a masked Ph. Our data also demonstrate that BCR–ABL1 location does not correlate with OS in CML patients (p = 0.41, Fig. 3).

Table 4 Summary of cases with insertion-derived BCR–ABL1 fusion in the literature.
Full size table
Fig. 3
figure 3

Overall survival comparison of CML patients with BCR–ABL1 fusion located on der(22) (n = 26) vs. non-der(22) (n = 12, including chr. 9, n = 11 and chr. 19). The mean survival/follow-up lengths were 75.6 months (range 11–242 months) in the der(22) group and 107 months (range 12–201 months) in the non-der(22) group respectively. No statistically significant difference for overall survival was observed between these two groups (p = 0.41).

Full size image

Sizes of the inserts can vary, which may be closely related to the morphological appearance of the chromosomes with the BCR–ABL1 or those involved chromosomes without the BCR–ABL1. For examples, two cases (cases #12 and #39) with an ins(9;22) as well as a few cases reported previously with either ins(9;22) or ins(22;9) (Table 4) [36, 38, 53] exhibited morphologically abnormal der(22) and/or der(9) which might mimic the classic t(9;22), implying that a large insert changed the appearance of involved chromosome(s). By intensive FISH mapping using many probes targeting ABL1, BCR, and their flanking regions, Virgili et al. [47] identified inserts carrying 3′ABL1 and flanking region or 5′ BCR and flanking region of sizes of 720 kb to 3.9 MB in 6 Ph negative, BCR–ABL1 positive cases with a normal karyotype. Valle et al. [45] reported an insert of ~5.7 MB in a case also with a normal karyotype. A classic t(9;22) usually involves an exchange of approximately 10 MB of 9q (from the ABL1 to the 9q telomere) and ~25 MB of 22q (from BCR to the 22q telomere).

Two mechanisms involved in the formation of a cryptic Ph were proposed in previous studies: an insertion of 3′ABL into 5′BCR or vice versa (“one-step”) and two sequential translocations (“two-step) [8, 19]. However, more cases with different mechanism(s) have been identified subsequently (“multi-step”) [6, 47, 55], e.g., simultaneous/sequential insertion + translocation involving chromosomes 9 and 22; additional insertion and/or translocation involving chromosome(s) other than 9 and/or 22; and gain or amplification of BCR–ABL1 [64] (Table 4). In this study, no reciprocal translocation between chromosomes 9 and 22 has been observed and/or confirmed by various FISH tests. Therefore, we suggest that an insertion-derived BCR–ABL1 fusion is likely (Supplementary Information). Although many cases in this study were intensively investigated by i-FISH, mb-FISH and/or wcp, the exact mechanism(s) (one-step vs. two-step vs. multi-step) cannot be completely determined due to the complexity of FISH results, even in some cases with a normal karyotype. Previous studies have suggested that concomitant deletion of 9q, del(9q) and/or del(22q) with ins(22;9) or ins(9;22) or even classic t(9;22) might be a predictive marker of a poorer prognosis in CML patients, in the era when interferon-alpha (INF-α) and hydrea were the mainstays of treatment [11, 12]. The role of these cytogenetic changes in the era of TKIs remains controversial [14, 15, 55], due to few patients reported previously treated with TKIs, as well as the uncertain status of del(9q) and/or del(22q) in many of those cases. In this study, eight cases were tested with a tricolor FISH test, which may be helpful to exclude/confirm a del9q including ASS1 and its flanking region, but this test is not informative for cases with a potential del(9q) involving genes/loci beyond the target of ASS1/ABL1 probe.

A complete characterization of the mechanism(s) involved in BCR–ABL1 fusion and the del(9q) and/or del(22q) status in each case usually requires intensively FISH mapping of BCR, ABL1, and their flanking regions [6, 47], or using other technologies with a genome-wide coverage, such as array-based comparative genomic hybridization [53, 65]. Genome-wide assays are often impractical in the clinical setting, especially for cases with extensive tumor heterogeneity. It is more important to explore the complexity of chromosomal aberrations involved/caused by the formation of BCR–ABL1 n this cohort of cases. A FISH test usually analyzes 200–500 interphase cells at a resolution of approximately 100 kb, whereas conventional karyotyping routinely analyzes 20 selected metaphase cells at a much lower resolution of 5–10 Mb. Therefore, FISH testing is considered more sensitive and specific than conventional cytogenetics for detection of targeted chromosomal abnormalities, e.g., the BCR–ABL1 and related chromosomal abnormalities in this cohort. As demonstrated in this study, cryptic chromosomal abnormalities or even some apparent chromosomal abnormalities presenting as minor clone(s) can be easily overlooked. Therefore, the results of conventional cytogenetics alone showing a normal karyotype, or even an abnormal karyotype but with cryptic der(22) and/or der(9), or misinterpretation of subtle changes in the karyotype can be misleading. FISH testing in these cases can clarify the findings and these results could have direct clinical impact, in terms of risk stratification or prognosis.

In this study, almost all CML cases presented a p210 BCR–ABL1 transcript, except one with a p190 BCR–ABL1 transcript and another case with a variant BCR–ABL1 transcript that was not characterized [21]. We have observed a high frequency of co-existing e14a2 and e13a2 BCR–ABL1 transcripts in this study, ~16%, which was reported in a single case with an insertion-derived BCR–ABL1 previously [35]. A co-existence of e14a2 and e13a2 BCR–ABL1 transcripts is caused by a polymorphism within BCR gene [66], however, it is unknown whether the patients in this cohort are prone to carry BCR polymorphisms. Some hotspot ABL1 mutations have been reported to be associated with TKI resistance in CML patients. It is necessary to point out that TKIs have dramatically evolved in the past two decades, and patients in this cohort have received various types of TKIs, either solely or subsequently according to the availability and specific indication(s) of each TKI. No statistical analysis for efficiencies of various TKIs has been performed for this cohort.

In summary, this study represents the largest cohort of CML cases with an insertion-derived BCR–ABL1 fusion. A systemic investigation of clinical, laboratory and special testing information in this cohort reveals that approximately 40% of cases exhibited an apparent der(9) and/or der(22), with clues for a BCR–ABL1 rearrangement, although these aberrations did not mimic a classic t(9;22). Therefore, chromosomes carrying an insertion-derived BCR–ABL1 are not always cryptic. The locations of BCR–ABL1 and sizes of the inserts were different among all cases in this cohort, indicating the complexity of this type of chromosomal rearrangement. In addition, the insert size is related to the morphology of affected chromosomes, and the FISH signal patterns may imply for a balanced or unbalanced insertion and/or a mixture of subclones. Our data further show that many CML cases with a normal karyotype actually possess complex chromosomal abnormalities that can be shown by various FISH tests, and that the presence of complex chromosomal abnormalities shown by FISH predict poorer OS. Therefore, any cases with a potential insertion-derived BCR–ABL1 fusion warrant intensive FISH studies, at least at the time of initial diagnosis, and a normal or noncomplex karyotype obtained by conventional cytogenetics in these cases is misleading in terms of risk stratification, response categorization and clinical follow-up. We suggest that FISH testing is mandatory for the workup of cases of CML with BCR–ABL1 fusion and that these results are relevant for patient management.