Several chromosomal aberrations detected in addition to the pathognomonic Philadelphia
chromosome (Ph) at diagnosis confer a poor prognosis in chronic myeloid leukemia (CML)
chronic phase (CP) patients and herald earlier progression to accelerated phase (AP)
or blast crisis (BC), and CML-related death.
1-3
Their prognostic significance has been established both at diagnosis and when emerging
in the course of the disease. Since not all clones have the same clinical relevance,
several classifications have been proposed in the recent literature to define additional
cytogenetic aberrations (ACA) presenting a higher risk of inferior outcomes.
4-8
The conventional classification in “major” and “minor route” ACA was based on their
prevalence and appeared to be too restricted to cover all “high-risk” ACA (HR-ACA).
Besides four major-route ACA (trisomy 8, isochromosome 17q, additional Ph chromosome
and trisomy 19 while excluding loss of Y), five other HR-ACA were identified (trisomy
21, 3q26.2 rearrangements, monosomy 7/7q-, 11q23 rearrangements, and complex karyotypes)
in a recent study of CML-CP patients.
8
In this study, their presence often preceded an increase in blast percentage and thereby
anticipated progression. However, one study did not observe a prognostic impact of
trisomy 8 or an additional Ph chromosome when occurring as a single ACA and only heralded
inferior outcomes when in combination with other concurrent ACA.
5
These discrepant results may be due to low observation numbers at diagnosis as HR-ACA
remain relatively rare and are detected in less than 3% of de novo CML-CP patients.
8-10
Consequently, the cohort sizes of previous studies of patients with HR-ACA have been
relatively small and verification of findings is necessary. Here, we aim to assess
the prevalence of ACA at diagnosis and their clinical impact in a Dutch nationwide
patient cohort, with a focus on the recently proposed HR-ACA classification.
8
In addition, we intend to assess the relation of HR-ACA to the EUTOS long-term survival
(ELTS) score at diagnosis and to assess the impact of chromosomal aberrations on hematological
toxicity (hemtox) of first-line tyrosine kinase inhibitor (TKI) treatment.
Data were derived from a real-world population-based CML registry in the Netherlands
(PHAROS-CML registry combined with HemoBase) covering a nationwide patient cohort
diagnosed with CML between 2008 and 2014.
11
We included all adult CML-CP patients with an evaluable cytogenetic assessment at
diagnosis. HR-ACA were defined following Hehlmann et al. (+8, i(17q), +Ph, +19, +21,
3q26.2, -7/7q-, 11q23.2 and complex karyotype; present in Ph-positive cells).
8
Other ACA in Ph-positive cells were classified as low-risk ACA (LR-ACA). The emergence
of chromosomal aberrations was also assessed during the first 24 months of TKI treatment,
including clonal chromosomal aberrations in Ph-negative cells (CCA/Ph-). AP and BC
were defined as described in the ELN recommendations.
12
Hemtox was defined as de novo anemia, thrombocytopenia and/or leukopenia CTC grade
3 or higher, emerging during firstline TKI therapy.
Survival analysis was performed with Kaplan-Meier estimates and the log-rank test
was used to compare subgroups. Progression-free survival (PFS) was defined as the
time from diagnosis until progression to AP/BC or death. Patients were censored at
last follow-up visit. CML-related death was defined as death preceded by CML progression
and was assessed using the cumulative incidence competing risk (CICR) method in which
death of any other cause was considered as a competing event. Response milestones
(complete hematological response [CHR], complete cytogenetic response [CCyR] and major
molecular response [MMR]) were defined in accordance with the ELN recommendations.
12
The achievement of CCyR, MR2.0 (BCR::ABL1 <1%IS) or MMR was assessed with the CICR
method in which progression or death were considered as a competing event. A Cox proportional
hazards model was used to assess different predictors for PFS including age, ELTS
score (as a numeric variable) and the presence of HR-ACA at diagnosis. The Χ
2
test was used to assess differences in hemtox across subgroups, only considering complete
cases. The Medical Ethics Committee of the Erasmus Medical Center in Rotterdam approved
this study and the exemption from informed consent. The study was conducted in accordance
with the Declaration of Helsinki. A total of 398 CML-CP patients were included in
this analysis. Thirty ACA (8%) were detected at diagnosis of which 15 were HR-ACA
(4%) (Figure 1). The most frequent HR-ACA were trisomy 8 and an extra copy of Ph chromosome.
Loss of the Y chromosome (-Y) as a solitary additional aberration in Ph-positive cells
was observed in ten patients and was not designated as ACA since several studies did
not report any clinical impact of this aberration.
5,8
Patients with HR-ACA at diagnosis were younger than patients without HR-ACA, with
a median age of 49 years (interquartile range [IQR], 34-61 years) versus 57 years
([IQR], 43-68 years) at diagnosis, respectively (P=0.198). Other baseline characteristics
were comparable between subgroups, including the ELTS score at diagnosis and the use
of second generation TKI as first line treatment. There was no statistically significant
association between ELTS categories and the presence of HR-ACA using the Χ
2
test (P=0.168), nor was there a significant difference in the mean ELTS score in patients
with or without HR-ACA using the Student's t-test (P=0.400).
Figure 1.
Inclusion flowchart and prevalence of (additional) cytogenetic aberrations found in
the Pharos-HemoBase chronic myloid leukemia patient population at diagnosis. AP/BC:
accelerated phase or blast crisis; ACA: additional cytogenetic aberrations; HR: high-risk;
LR: low-risk; kar: karyotype.
During the first 24 months of TKI treatment, one or more follow-up cytogenetic assessments
were done in 257 patients. In these patients, four patients (2%) had newly emerging
ACA in the context of disease progression, and 31 patients (12%) developed CCA/Ph-.
Most frequent CCA/Ph- were –Y (n=12), +8 (n=11) and -7/7q- (n=4). Transition to myelodysplasia
or acute myeloid leukemia was not observed during further follow-up of these patients.
Five-year PFS for patients with HR-ACA, with LR-ACA or without ACA was 60% (95% confidence
interval [CI]: 40-91), 87% (95% CI: 71-100) and 85% (95% CI: 81-89), respectively,
with a median follow-up duration of 5 years (IQR, 4-8 years) (Figure 2A). Of note,
in patients with ACA, all events of progression or death occurred within 3 years from
time of diagnosis. After further stratification based on HR-ACA and the ELTS score
at diagnosis, an inferior PFS was noted in patients with HR-ACA in combination with
an intermediate or high ELTS score (Figure 2B). In line with PFS results, the cumulative
incidence of CML-related mortality was higher in patients with HR-ACA than patients
without HR-ACA (13% vs. 3% at 5 years; P<0.032). No difference in PFS was observed
for patients with solitary –Y or with emerging CCA/Ph- compared to patients without
aberrations (Online Supplementary Figure S1). Again, when specifically assessing non
–Y CCA/Ph-, no difference in PFS was noted (graph not shown; P=0.703).
In a univariable Cox regression analysis, age, ELTS score and the presence of HR-ACA
were predictive for PFS, with a hazard ratio (HR)=1.06, 95% CI: 1.04-1.08; HR=2.09,
95% CI: 1.39-3.15 and HR=2.81; 95% CI: 1.22-6.49, respectively (Online Supplementary
Table S1). We fitted a multivariable model with ELTS score and HR-ACA, and excluded
age since it is already part of the ELTS score calculation. The HR for PFS of HR-ACA
and ELTS score were HR=3.13, 95% CI: 1.34-7.31 and HR=2.06, 95% CI: 1.37-3.11, respectively.
Regarding the achievement of the ELN response milestones, CHR at 90 days was achieved
in 80% versus 87% of patients with versus without HR-ACA, respectively (P=0.428).
The cumulative incidence of CCyR or MR2.0 at 6 months was 10% (95% CI: 0-30) versus
38% (95% CI: 32-43) in patients with versus without HR-ACA, respectively (P=0.261).
The cumulative incidence of MMR at 12 months was 22% (95% CI: 0-51) versus 50% (95
% CI: 44-57) in patients with versus without HR-ACA, respectively (P=0.045). Of note,
all HR-ACA patients who eventually presented with disease progression, failed to achieve
the MR2.0 or MMR ELN milestone in time.
In a final exploratory analysis, we assessed the occurrence of hemtox on first-line
tyrosine kinase inhibitor (TKI) treatment. Patients with HR-ACA at diagnosis had significantly
more hemtox than those without any ACA (39% vs. 16%; P=0.030), while this difference
was not observed for patients with LR-ACA (10% vs. 16%; P=0.607). Patients with CCA/Ph-
emerging during the first 24 months of TKI treatment, also experienced more hemtox
than patients without CCA/Ph- (32% vs. 16%; P=0.026). In CCA/Ph- patients, hemtox
was mostly observed in case of +8 and/or -7/7q-(7/15, 47%), and in lesser extent in
case of –Y (2/12, 17%). In both groups (HR-ACA and CCA/Ph- patients) the difference
in hemtox was mainly due to an increased incidence of thrombocytopenia, with or without
concomitant anemia or leukopenia.
Figure 2.
Progression-free survival Kaplan-Meier estimates. (A) Subgroups based on the presence
of low-risk (LR) or high-risk (HR) additional cytogenetic aberrations (ACA). (B) Subgroups
based on the presence of high-risk ACA and the EUTOS long-term survival (ELTS) score.
In conclusion, our results support the recently proposed ACA risk classification.
8
HR-ACA at diagnosis were associated with inferior responses, and a significantly higher
probability of progression and (CML-related) death, while patients with LR-ACA had
a PFS comparable to that of other CML-CP patients. Furthermore, HR-ACA at diagnosis
remained independently predictive for PFS in a multivariable regression model including
ELTS score, which is in line with a previous analysis.
10
In contrast with HR-ACA, the emergence CCA/Ph- did not have an impact on PFS in our
cohort. The prognostic significance of this entity remains controversial, more specifically
for non –Y CCA/Ph-.
13,14
Additionally, our data suggest that patients with HR-ACA at diagnosis or with CCA/Ph-
emerging during TKI treatment, have a higher risk of TKI-related hemtox. CCA/Phmight
interfere with normal (Ph-) hematopoiesis, predisposing to TKI-related hemtox. This
is in line with previous studies showing an increased risk of development of myelodysplastic
syndrome from a CCA/Ph- clone.
15,16
Taken together, follow-up cytogenetic evaluation after diagnosis is warranted in case
of failure to achieve molecular milestones in order to evaluate clonal progression,
17
and also in case of hematological toxicity to evaluate emergence of CCA/Ph-, even
when molecular response is optimal. Our results on their own should be interpreted
with caution since the number of patients with HR-ACA and CCA/Ph- was low. However,
our study contributes to the accumulating evidence that implies that patients with
HR-ACA at diagnosis, particularly with a high ELTS, may benefit from a more aggressive
treatment strategy with a second-generation TKI and an earlier switch to allogeneic
stem cell transplantation if the response to TKI is unsatisfactory or results in significant
hematological toxicity.
Supplementary Material
Supplementary Appendix