In vitro culture systems demonstrated that BCR-ABL1 can transform immature hematopoietic cells, some fibroblast cell lines, and hematopoietic cell lines rendering them growth factor-independent. In addition, several groups reported that a CML-like disease could be induced in mice transplanted with bone marrow infected with a
BCR-ABL1 retrovirus. In contrast, mutant isoforms of BCR-ABL1 carrying inactivating mutations in the SH1 domain, or mutants lacking the BCR coiled coil domain, did not induce leukemia. All these studies [
27‐
30], conducted around the 90s, converged to demontrate that BCR-ABL1 is indeed the causative agent of CML and fostered the search for small molecule inhibitors. On the other hand, evidences have also been brought that challenge this view. There are marked strain differences in disease induction after
BCR-ABL1 retroviral expression, suggesting that the genetic background may influence the ability of the oncogene to initiate CML [
29]. Even more interestingly, a conditional knock-in mouse in whom the human
BCR-ABL1 cDNA was knocked into the endogenous mouse
Bcr locus so that it could be conditionally expressed with different tissue-specific Cre transgenes under the added control of the native
Bcr regulatory elements, was found not to develop leukemia during its lifetime, despite expression of a constitutively active BCR-ABL1 tyrosine kinase was observed in the hematopoietic progenitors [
31]. The authors thus postulated that i) physiologic BCR-ABL1 expression may be insufficient for development of a CML-like disease; ii) in the retroviral or transgenic models, non-physiologic, very high levels of BCR-ABL1 expression due to multiple copies of the oncogene and expression from a very active retroviral promoter, non-specificity of expression timing and locale and maybe also random insertion-site mutations could artificially select for disease development [
31]. This study was published in 2013, but the idea that additional cooperating events might be required for the induction of CML was, indeed, not new. Between the 80s and the 90s, initial evidences were brought in support of the existence of a putative event preceding the acquisition of
BCR-ABL1 at least in a proportion of patients. Studies of X chromosome inactivation and glucose-6-phosphate dehydrogenase genotype had raised the hypothesis that clonal hematopoiesis might precede the acquisition of the Ph chromosome [
32,
33]. In addition, starting from the 90s, five reports had been published about the detection of
BCR-ABL1 transcripts in circulating leukocytes of up to 65% of healthy individuals when using sensitive polymerase chain reaction (PCR)-based assays [
34‐
38]. Overall, 380 samples have been analyzed in these studies.
BCR-ABL1 was detected in cord blood and newborns (up to 40%), children and adolescents (up to 56%), adults (20–59 yrs.; up to 65%) and elderly (> 60 yrs.; up to 65%). For unknown reasons, the e1a2 rearrangement (leading to p190
BCR-ABL1) was much more frequently detected than the e13a2 or e14a2 rearrangements (leading to p210
BCR-ABL1). It might be argued that in all the studies a nested reverse transcription (RT)-PCR strategy was used to enhance sensitivity, although such an approach has the known drawback of being more prone to contamination. Unfortunately, there is no follow-up information available for BCR-ABL1-positive cases. The latency period between acquisition of the Ph chromosome and overt clinical development of CML is unknown and it is likely to be highly variable. Atomic bomb survivors could develop CML up to 40 years later. On the other hand, there are reports of children > 1 year of age who were diagnosed with CML [
39]. In spite of the technical issues, these data, together with case reports of patients with detectable Ph chromosome in their bone marrow cells but otherwise asymptomatic (with a follow-up of few years only, however) [
40,
41] raise, among others, the hypothesis that other events are needed before a true malignant expansion can occur and overt CML may develop. Mathematical models predict that 2 or more genetic hits in the hematopoietic stem cells may be needed for CML to develop [
42,
43]. Although CP CML has long been considered a genetically homogeneous entity, the power of next generation sequencing (NGS) is now changing this view. A few years ago, targeted NGS-based resequencing of the 25 most commonly mutated genes in myeloid leukemias/myelodysplasias revealed
ASXL1,
TET2,
RUNX1,
DNMT3A,
EZH2 and
TP53 mutations in 5 out 15 chronic phase CML patients at diagnosis [
44]. In the same study, analysis of individual hematopoietic colonies showed that the great majority of mutations were part of the Ph + clone. However, targeted resequencing of subsequent samples during TKI treatment revealed that the
DNMT3A mutation found in the Ph + cells of a patient at diagnosis was also present in the Ph- clone, implying that it preceded
BCR-ABL1 acquisition. [
44] Now we know that
DNMT3A,
TET2 and
ASXL1 mutations, among others, may indeed be found in healthy elderly individuals, where they correlate with the risk of hematologic cancer and all-cause mortality (‘CHIP’, clonal hematopoiesis of indeterminate potential) [
45‐
47]. Such mutations are thought to represent the first hit, leading to a clonally expanded pool of pre-leukemic hematopoietic stem cells from which overt leukemia may subsequently evolve through the acquisition of additional, disease-shaping genetic lesions [
48]. Most recently, a NGS-based screen of 92 myeloid-associated genes in 300 serial samples from 100 CP CML patients at diagnosis and after TKI therapy showed evidence of DNMT3A, TET2, ASXL1, BCOR and CREBBP mutations in both diagnosis and follow-up samples, despite response to TKI therapy and
BCR-ABL1 transcript clearance [
49]. This further indicates that up to 10% of CML patients may have CHIP-related mutations and reinvigorates earlier hypotheses of a multistep pathogenesis of CML – arising, at least in some cases, from pluripotent stem cells of a pre-existing Ph- clone that enjoys a growth advantage.
Prospective serial screening of healthy individuals to determine whether the presence of the BCR-ABL1 oncogene in their blood predicts for future CML development would be of great interest. To this purpose, the use of digital PCR would enable to conjugate high sensitivity with a more precise and accurate count of BCR-ABL1 transcripts. However, because CML occurs at a frequency of 1–2 cases per 100,000 per year, a very large cohort would be needed, together with analysis of an equal number of individuals without detectable BCR-ABL1 transcripts.