Purpose: Leukemic stem cells (LSCs) may harbor important resistance to tyrosine kinase inhibitors in chronic myelogenous leukemia (CML). We identified Philadelphia chromosome (Ph)–positive CD34+CD38 bone marrow cells (here denoted LSCs) and addressed their response-predictive value in patients with CML (n = 48) subjected to nilotinib in the ENEST1st trial (NCT01061177).

Experimental design: Two flow cytometry–based cell sorting methods were used with multiparameter-directed CD45- (MPFC) and BCR-ABL1 probe-linked (FISH) identification of Ph-positive cells, respectively.

Results: We observed a positive correlation between the proportion of LSCs at diagnosis and established prognostic markers (blast count, spleen size, Sokal score, and hemoglobin). Conversely, a high LSC burden predicted for an inferior molecular response at 3 (MPFC and FISH), 6 (MPFC), 9 (FISH), and 15 months (FISH). During nilotinib therapy, the proportion of LSCs decreased rapidly. At 3 months, a median of only 0.3% LSCs remained among CD34+CD38 cells, and in 33% of the patients the LSC clone was not detectable anymore (FISH). The response kinetics was similar in LSC fractions as it was in the progenitor and unseparated bone marrow cell fractions.

Conclusions: The proportion of LSCs at diagnosis, as analyzed by two independent methodologies, reflects the biology of the disease and appeared as a prognostic and response-predictive marker in patients with CML subjected to first-line nilotinib therapy. Clin Cancer Res; 22(16); 4030–8. ©2016 AACR.

Translational Relevance

We assessed the translational relevance of leukemic stem cell (LSC) burden on therapy response in nilotinib-treated patients with chronic myelogenous leukemia (CML). We found that the proportion of LSCs at diagnosis, as analyzed by two independent methodologies [multiparameter flow cytometry (MPFC) and sorting plus FISH], not only reflects the biology of the disease but also appeared as a response-predictive marker in CML. This enables physicians to identify patients who respond more slowly to the therapy and could be at higher risk of transformation to more advanced disease phases. Moreover, MPFC has the additional benefit of enabling sorting of viable leukemic and normal stem cells into separate fractions. This permits functional evaluation as well as expression profiling and comparison of leukemic and normal stem cells.

The introduction of the tyrosine kinase inhibitor (TKI) imatinib and, more recently, of the second-generation TKIs nilotinib, dasatinib, and bosutinib marked a revolution in the treatment of patients with chronic myelogenous leukemia (CML). Most patients attain excellent cytogenetic and molecular responses to these drugs and remain free of progression to accelerated phase and blast crisis (1–3). However, in the majority of patients, cure of the disease will probably not be achieved by TKIs alone due to the persistence of leukemic stem cells (LSCs). According to in vitro studies, LSCs are insensitive to currently available TKIs (4–10). Patients with high Sokal or Euro scores, based on hematologic parameters, spleen size and age, and those with slow reduction of BCR-ABL1 transcripts in response to starting treatment are at increased risk of treatment failure (11–13).Moreover, our study groups have previously demonstrated that the composition of the CD34+CD38 stem cell compartment at diagnosis, as assessed by multiparameter flow cytometry (MPFC) or stem and progenitor cell sorting followed by FISH, not only predicts for later response, but also for hematologic toxicity during TKI treatment (14, 15).Reliable and early response prediction is of great value as it enables personalized tailoring of therapy before progression to advanced phases in patients who are deemed to respond poorly. In this study, we set out to assess whether analysis of the stem cell compartment at diagnosis and stem cell dynamics early after treatment initiation, such as the LSC reduction or, conversely, LSC persistence, has additional value in predicting outcome in nilotinib-treated patients with chronic phase CML (CP-CML). This translational project was performed as a substudy to the ENEST1st clinical study in which newly diagnosed CML-CP were treated with nilotinib 300 mg twice daily. The project is a collaborative effort of the Nordic CML Study Group, the Central European Leukemia Study Group (CELSG), and selected Dutch HOVON centers.

Patients

Patients who were treated in the multicenter and multinational phase IIIb clinical trial Evaluating Nilotinib Efficacy and Safety as First-Line Treatment (ENEST1st study) were eligible for the stem cell study (16). Patients could be included in this study when they had newly diagnosed chronic phase CML, were 18 years of age or older and had WHO performance status ≤ 2. Treatment consisted of nilotinib 300 mg twice daily. An adequate bone marrow aspirate and/or peripheral blood sample was required at diagnosis and after 1 and 3 months of therapy.

The study was performed in accordance with the declaration of Helsinki. All patients provided written informed consent. The study was approved by the local institutional review boards of all participating centers, and is registered at ClinicalTrials.gov (identifier NCT01061177).

Materials

For this study, MPFC and sorting followed by FISH were performed in parallel and independently in selected and qualified core laboratories (MPFC: Stem Cell Laboratory of the VU University Medical Center, Amsterdam, the Netherlands; sorting and FISH: Helsinki University Central Hospital, Helsinki, Finland, Department of Clinical Genetics, Lund University, Lund, Sweden and Karolinska University Hospital, Stockholm, Sweden).

The assessment of the CD34+CD38 compartment was performed by both methods at diagnosis and after 3 months of treatment and only by MPFC after 1 month of therapy.

Mononuclear cells were isolated from bone marrow and/or peripheral blood by density-gradient centrifugation using Ficoll-paque (Amersham Biosciences), followed by a red cell lysis step with lysis buffer (155 mmol/L NH4Cl, 10 mmol/L KHCO3, 0.1 mmol/L Na2EDTA, pH 7.4) for 10 minutes at 4°C.

Multiparameter flow cytometry

MPFC was used as previously described for separation of normal and leukemic CD34+CD38cells (14). In this study, no functional assessments of stem cell properties were performed. In previous work however, we could demonstrate that these cells have long-term culture initiating capacity (14). Although we are aware that the term (normal or leukemic) “candidate” stem cells may be more appropriate, for brevity, we will refer to the cells in the CD34+CD38 compartment as (either normal or leukemic) stem cells.

Freshly isolated mononuclear cells were incubated with monoclonal antibodies for 15 minutes at room temperature, washed once in PBS containing 0.1% human serum albumin (HSA) and analyzed by flow cytometry. The fluorescent-labeled antibodies used were anti-CD34 phycoerythrin-Cy7 (PE-Cy7), anti-CD45 fluorescein isothiocyanate (FITC), anti-CD38 allophycocyanin (APC), anti-CD7-phycoerythrin (PE), anti-CD11b-PE, anti-CD56-PE, anti-CD90-PE and Via-Probe (7-amino-actinomycin D, 7AAD; all from BD Biosciences). IgG1-PE was used as a control for PE-labeled anti-CD7, -CD56 and -CD90, IgG2a-PE was used for PE-labeled anti-CD11b. PBS was used as a control for anti–CD34-PE-Cy7, CD38-APC and -CD45-FITC. 7-AAD was included to gate out apoptotic/dead cells before stem cell assessment. Data were acquired using a 3-laser FACSCanto or FACSFortessa flow cytometer (BD Biosciences) and analyzed using FACSDiva software (BD Biosciences).

Flow-cytometry gating strategy

The gating strategy we used in this study is presented in the Supplementary Fig. S1 and was as previously described (14). In short, the lowest 1% of CD38-expressing cells in the CD34+ compartment was designated as the stem cell compartment. In a CD34CD45 plot, these cells can display two different patterns. In the first, there are two separate populations visible, designated as CD34+CD45high with relatively high forward and side scatter light properties (FSC and SSC, respectively) and CD34+CD45low with lower FSC/SSC properties, respectively. As we previously showed (14), the latter consists of Ph-negative cells, always negative for aberrant markers like CD7, CD11b, or CD56 and low in CD90 expression, whereas the cells in the former population are almost all (>97%) Philadelphia-positive (Ph+) and may display CD7, CD11b or CD56 positivity and high CD90 expression. The second pattern involves patients with only one single CD34+CD45+ population. These cells exhibit relatively high FSC/SSC signals and express one or more of the aberrant lineage markers together with high CD90 expression, and were demonstrated to be Ph+. Moreover, as stated above, these cells showed clonogenic colony forming capacity by performing long-term culture assays (14). For a reliable distinction between malignant and benign stem cells, at least 10 clustered cells per population were required. To be able to achieve this, we aimed at analyzing at least 1 million mononuclear cells (MNC). On average, at diagnosis this would need less than 5 mL of peripheral blood or bone marrow.

Sorting and FISH

The protocol for sorting and FISH has been described previously (15). In short, CD34+ cells were enriched with paramagnetic beads (Miltenyi Biotech), stained with CD34 and CD38 antibodies (BD Biosciences) after which the primitive cell fractions were sorted by a flow cytometer (FACSAria; BD Biosciences) into two populations; the first representing 5% with the lowest expression of CD38 (CD34+CD38) and the second representing 80% with highest expression of CD38 (CD34+CD38+). Like the definition used for MPFC, Ph+ CD34+CD38 cells are further designated as LSCs and Ph+ CD34+CD38+ cells as leukemic progenitor cells.

After sorting, cytospin slides were made from both primitive cell fractions and additionally from unseparated whole bone marrow samples from which red cells were lysed with FACS lysing buffer (BD Biosciences). The proportion of Ph+ cells was determined by FISH using dual fusion dual color BCR-ABL1 probes (Vysis, Abbot).

By sorting and FISH, 1,000 cells were manually analyzed in each fraction, in contrast with conventional FISH that most often uses only 200 unselected cells. Therefore, the sensitivity increases considerably up to a detection limit of one Ph+ CD34+CD38 cell out of 100,000 to 1,000,000 mononuclear cells.

Before the trial start, quality control rounds were performed in the different stem cell laboratories to assess interlaboratory consistency. The same FISH sample slides were circulated among the laboratories and FISH slides were quantitatively evaluated. In addition, at two different time points, a bone marrow aspirate sample was taken from a newly diagnosed CML patient and an aliquot of the sample was sent to each participating stem laboratory to assess reproducibility for the sorting and FISH procedure. The results were compared between various centers and a >90% concordance was achieved among the different laboratories.

Statistical analysis

The primary objective of this study addressed the question whether the composition of the CD34+CD38 compartment at diagnosis and during follow-up, determined by either MPFC or sorting and FISH, has a predictive value for reaching MR4 at 18 months.

For analysis of MPFC results, samples were divided in two classes: The first showing detectable amounts of normal hematopoietic stem cells (nHSCs; in any quantity) and the second showing no detectable nHSCs at diagnosis. Conversely, at follow-up, a binominal distinction was made between samples with only nHSCs and samples where residual LSCs could be detected. Differences between those groups regarding a set of parameters were investigated using the Exact Wilcoxon Mann–Whitney rank-sum test.

For sorting and FISH analysis, outcome was given on a continuous scale, that is, the percentage of LSCs of all stem cells. Correlation coefficients were calculated using Kendall's tau-b statistic. Sokal and Euro-scores were calculated using original publications (12, 13). Molecular response was centrally analyzed in an EUTOS reference laboratory. Major molecular response (MR3) was defined as a BCR-ABL1 transcript level of ≤0.1%, MR4 as a measurable BCR-ABL1 transcript level of ≤0.01% or an undetectable transcript level with at least 10,000 ABL1 copies and MR4.5 as a measurable BCR-ABL1 transcript level of ≤0.0032% or an undetectable transcript level with at least 32,000 ABL1 copies, all according to the international scale. For calculation of mean BCR-ABL1 levels, undetectable transcript levels with at least 10,000 ABL1 copies were counted as zero. Statistical analyses were performed with the R software under consideration of (i) the Fisher's exact test for pairs of categorical variables, (ii) the Exact Wilcoxon Mann–Whitney rank-sum test for dichotomous-independent/continuous dependent variables, and (iii) the Kendall's rank correlation tau-b test for pairs of continuous random variables. A logistic regression model was used for predicting response for the percentage of Ph+ stem cells by FISH. For all analyses, P values <0.05 were considered to indicate statistically significant difference.

The demographic and clinical parameters of all patients are shown in Table 1. Median follow-up for all patients was at least 24 months. MPFC of the CD34+CD38 cell compartment (bone marrow or peripheral blood) was available for 48 patients at diagnosis (Supplementary Fig. S2 presents the consort diagram). Four patient samples were not evaluable due to either insufficient cell numbers in the CD34+CD38 compartment or indistinct flow patterns. Of the remaining 44 patients, residual nHSCs were detectable by MPFC in 18 patients (41%), in either bone marrow and/or peripheral blood. In 5 of these patients, only nHSCs were detected with MPFC, thus without the presence of LSCs. In 3 of these, FISH was performed as well, and all had low LSC burden (1.2%, 14.4% and 29.5%). In the samples of the other 26 patients (59%) nHSCs could not be identified within the detection limits of the assay. In 28 of 30 patients who had both bone marrow and peripheral blood samples evaluable, the stem cell patterns (either residual nHSCs or no detectable residual nHSCs) were identical between both compartments. Only two patients had inconsistent patterns in bone marrow and peripheral blood: in one patient residual nHSCs were detected in peripheral blood, whereas bone marrow revealed only LSCs, whereas in the second patient the opposite was observed.

Table 1.

Demographic and clinical data at diagnosis

AllMPFCSorting + FISH
n 50 48 46 
Sex (M:F) 33:17 31:17 30:16 
Age, y 59 (20–82) 58 (20–75) 58 (27–82) 
WBC (109/L) 61 (3.7–307) 62 (3.7–307) 76 (9.7–307) 
Platelet count (109/L) 377 (98–1,485) 388 (98–1,485) 402 (98–1,485) 
Spleen size (cm) 0 (0–30) 0 (0–30) 0 (0–30) 
Sokal score 
 Low (%) 30 29 30 
 Int. (%) 34 33 35 
 High (%) 26 27 25 
 Unknown (%) 10 10 10 
Euro score 
 Low (%) 40 40 40 
 Int. (%) 36 35 35 
 High (%) 10 10 13 
 Unknown (%) 14 15 13 
AllMPFCSorting + FISH
n 50 48 46 
Sex (M:F) 33:17 31:17 30:16 
Age, y 59 (20–82) 58 (20–75) 58 (27–82) 
WBC (109/L) 61 (3.7–307) 62 (3.7–307) 76 (9.7–307) 
Platelet count (109/L) 377 (98–1,485) 388 (98–1,485) 402 (98–1,485) 
Spleen size (cm) 0 (0–30) 0 (0–30) 0 (0–30) 
Sokal score 
 Low (%) 30 29 30 
 Int. (%) 34 33 35 
 High (%) 26 27 25 
 Unknown (%) 10 10 10 
Euro score 
 Low (%) 40 40 40 
 Int. (%) 36 35 35 
 High (%) 10 10 13 
 Unknown (%) 14 15 13 

NOTE: Data for white blood counts (WBC), platelet counts, spleen sizes, Sokal and Euro scores are shown as medians with ranges in parentheses.

Abbreviations: F, female; int, intermediate; M, male; n, number; yr, years.

To assess whether the total LSC burden as measured by MPFC differed between patients with or without residual nHSCs, we measured the percentage of LSCs from the total MNCs (the number of designated LSCs divided by the total number of MNCs): LSC burden was higher in patients without detectable residual nHSCs compared to those with residual nHSCs (mean 0.018% vs. 0.003%, respectively; median 0.011% vs. 0.001%, respectively, range 0.0007%–0.11% and 0.0002%–0.02%, P < 0.001).

For sorting and FISH, 46 diagnostic patient samples were available, of which 39 were assessable (Supplementary Fig. S2). The average percentage of Ph+ cells within the CD34+CD38 population in the diagnostic samples was lower than in the progenitor and total bone marrow fraction (85% vs. 96% vs. 96%, respectively, P = 0.023 and P = 0.12; Fig. 1A). Figure 1B presents the proportion of LSCs analyzed with the sorting and FISH method in patients with and without detectable residual nHSCs by MPFC method.

Figure 1.

The proportion of Ph+ cells at diagnosis in the stem and progenitor cell compartments and whole bone marrow. A, the proportion of Ph+ cells analyzed by sorting and FISH in stem (CD34+CD38), progenitor (CD34+CD38+) and whole bone marrow compartments. The mean and median percentages of Ph+ cells were 85% and 98% in the stem cell fraction, 96% and 99% in the progenitor fraction and 96% and 99% in the whole bone marrow, respectively. Black lines represent mean and dashed lines median values. B, concordance between results of MPFC (detectable nHSCs and undetectable nHSCs) and sorting and FISH (percentage of Ph+ stem cells) at diagnosis (n = 34). Black lines represent median values. BM, bone marrow.

Figure 1.

The proportion of Ph+ cells at diagnosis in the stem and progenitor cell compartments and whole bone marrow. A, the proportion of Ph+ cells analyzed by sorting and FISH in stem (CD34+CD38), progenitor (CD34+CD38+) and whole bone marrow compartments. The mean and median percentages of Ph+ cells were 85% and 98% in the stem cell fraction, 96% and 99% in the progenitor fraction and 96% and 99% in the whole bone marrow, respectively. Black lines represent mean and dashed lines median values. B, concordance between results of MPFC (detectable nHSCs and undetectable nHSCs) and sorting and FISH (percentage of Ph+ stem cells) at diagnosis (n = 34). Black lines represent median values. BM, bone marrow.

Close modal

LSC burden by MPFC and sorting and FISH at diagnosis correlates with clinical and laboratory characteristics

Patients with residual nHSCs by MPFC and patients with low LSC burden by sorting and FISH had lower blood and marrow blast counts (P < 0.001 and P = 0.044 for MPFC; P < 0.001 and P = 0.003 for sorting and FISH, respectively), smaller spleen size (P < 0.001 for MPFC; P = 0.006 for sorting and FISH), lower platelet counts (P = 0.027 for MPFC; NS for sorting and FISH), lower Sokal scores (P < 0.001 for MPFC; P = 0.035 for sorting and FISH), lower WBC counts (P = 0.006 for sorting and FISH; NS for MPFC) and higher hemoglobin levels (P = 0.010 for MPFC; P = 0.001 for sorting and FISH; Fig. 2).

Figure 2.

Correlation between LSC burden and clinical parameters. At the diagnosis, LSCs were measured by MPFC (A–E) and sorting and FISH (F–J), and results were correlated with biological disease variables. Panels show peripheral blood blast percentages (A and F,P < 0.001 and P < 0.001), BM blast percentages (B and G,P = 0.044 and P = 0.003), spleen sizes (C and H,P < 0.001 and P = 0.006), platelet counts (D and I,P = 0.027 and P = 0.33), and Sokal scores (E and J,P < 0.001 and P = 0.035). In F to J patients with detectable nHSCc by MPFC have been marked with open dots () and patients with undetectable nHSCs with black dots (). Gray dots () mark patients with unclassifiable or absent MPFC results. Black lines represent median values. BM, bone marrow; PB, peripheral blood.

Figure 2.

Correlation between LSC burden and clinical parameters. At the diagnosis, LSCs were measured by MPFC (A–E) and sorting and FISH (F–J), and results were correlated with biological disease variables. Panels show peripheral blood blast percentages (A and F,P < 0.001 and P < 0.001), BM blast percentages (B and G,P = 0.044 and P = 0.003), spleen sizes (C and H,P < 0.001 and P = 0.006), platelet counts (D and I,P = 0.027 and P = 0.33), and Sokal scores (E and J,P < 0.001 and P = 0.035). In F to J patients with detectable nHSCc by MPFC have been marked with open dots () and patients with undetectable nHSCs with black dots (). Gray dots () mark patients with unclassifiable or absent MPFC results. Black lines represent median values. BM, bone marrow; PB, peripheral blood.

Close modal

LSC burden at diagnosis correlates with molecular response

Patients with residual nHSCs, and thereby lower LSC levels at diagnosis as detected by MPFC attained deeper molecular responses at 3 and 6 months after starting nilotinib treatment (P = 0.020 and P = 0.024, respectively; Fig. 3A) and attained MR3 earlier than patients with only LSCs, although the difference did not reach statistical significance. Although all evaluable patients in both MPFC groups achieved a BCR-ABL1 level <10% at 3 months, only 50% of patients (10/20) without detectable nHSCs at diagnosis reached a BCR-ABL1 level of ≤1% at 3 months, as compared with 100% of the 15 patients with residual nHSCs at diagnosis (P = 0.001). In addition, all patients of the latter group achieved a BCR-ABL1 level ≤1% at 6 months, whereas only 78% of patients (18/23) with only LSCs at diagnosis did (P = 0.046). We were unable to determine the relative decrease of BCR-ABL1 levels during the first 3 months as BCR-ABL1 values at diagnosis were only available for a few patients.

Figure 3.

Correlation of LSCs with molecular response. BCR-ABL1 transcript levels were measured every 3 months after the start of nilotinib treatment. All patients were followed for at least 24 months, were from 3 patients response data were not available. A, mean BCR-ABL1 levels during treatment of patients with (gray bars) and without (black bars) residual nHSCs at diagnosis measured by MPFC. Error bars represent SEs of the mean. P values for the difference between groups for 3 and 6 months response were 0.020 and 0.024, respectively. B, correlation between LSC burden at diagnosis measured by sorting and FISH and molecular response at 3 (P = 0.023), 9 (P = 0.007), 12 (P = 0.066), and 18 months (P = 0.082). Patients with detectable nHSCc by MPFC have been marked with open dots () and patients with undetectable nHSCs with black dots (). Gray dots () mark patients with unclassifiable or absent MPFC results. BM, bone marrow.

Figure 3.

Correlation of LSCs with molecular response. BCR-ABL1 transcript levels were measured every 3 months after the start of nilotinib treatment. All patients were followed for at least 24 months, were from 3 patients response data were not available. A, mean BCR-ABL1 levels during treatment of patients with (gray bars) and without (black bars) residual nHSCs at diagnosis measured by MPFC. Error bars represent SEs of the mean. P values for the difference between groups for 3 and 6 months response were 0.020 and 0.024, respectively. B, correlation between LSC burden at diagnosis measured by sorting and FISH and molecular response at 3 (P = 0.023), 9 (P = 0.007), 12 (P = 0.066), and 18 months (P = 0.082). Patients with detectable nHSCc by MPFC have been marked with open dots () and patients with undetectable nHSCs with black dots (). Gray dots () mark patients with unclassifiable or absent MPFC results. BM, bone marrow.

Close modal

Twenty-five percent of patients without detectable residual nHSCs at diagnosis versus 14% of patients with residual nHSCs at diagnosis did not attain MR3 at 12 months (P = 0.48). After 18 months, 22% of patients without detectable residual nHSCs versus 19% of patients with residual nHSCs did not attain MR3 (P = 0.82). There was no significant difference between the groups with respect to achieving MR4 at 18 months (primary endpoint), although a slightly higher proportion of patients with residual nHSCs at diagnosis achieved this endpoint (50% vs. 39%, P = 0.53).

LSC burden at diagnosis as a predictive factor for molecular response could also be validated with the sorting and FISH method. Patients with lower LSC burden achieved superior molecular responses at 3 (P = 0.023) and 9 (P = 0.007) months (Fig. 3B). A similar tendency toward a better response in patients with lower LSC burden at diagnosis was observed at later time points (6, 12, 15, 18, and 21 months), although statistical significance was only reached at 15 months (P = 0.034, at other time points P values were between 0.05 and 0.08). FISH analysis also predicted milestones; all patients not achieving MR3 at 12 or 18 months had a high LSC burden at diagnosis (>80% Ph+ cells; Fig. 3B). However, high LSC content did not preclude attaining superior molecular responses.

On the basis of the logistic regression analysis, the proportion of LSCs did not significantly impact the likelihood of achieving MR4 at 18 months (P = 0.136). However, the odds for a patient with a hypothetical LSC value of zero being in MR4 at 18 months was as high as 26.6, and for every 1% increase in the LSC number, the odds decreased by 1.04. Therefore, the odds for a patient reaching MR4 at 18 months decrease with increasing LSC number.

Hematologic adverse events were only observed in patients with high LSC burden

During the 24 months observation period, hematologic adverse events (cytopenias) occurred in 8 of 39 patients from whom LSC analysis had been done with sorting and FISH method. Interestingly, all patients experiencing any hematologic toxicity (grade 2–4) had a high LSC burden (>89% Ph+ cells) at diagnosis (Fig. 4). However, high LSC burden at diagnosis did not predict the occurrence of hematologic toxicity, as 72% of patients who had >89% of LSCs at diagnosis did not have any hematologic toxicity. The results were also concordant when analyzed with the MPFC method as in six of seven patients with hematologic toxicity during nilotinib treatment no residual nHSCs could be detected.

Figure 4.

Correlation of LSCs with the occurrence of hematologic toxicity during nilotinib therapy. Grade 2 to 4 hematologic toxicity (hem) during nilotinib therapy was defined according to standard clinical trial procedures and patients were divided in 2 groups based on the occurrence of any hematologic toxicity. The proportion of LSCs was measured by sorting and FISH in patients with and without hematologic toxicity. The median percentage of LSCs did not significantly differ between the groups (P = 0.16, Mann–Whitney test), but all patients with hematologic toxicity had >89% of LSCs at the diagnosis. Abbreviations: hem AE, hematologic adverse events.

Figure 4.

Correlation of LSCs with the occurrence of hematologic toxicity during nilotinib therapy. Grade 2 to 4 hematologic toxicity (hem) during nilotinib therapy was defined according to standard clinical trial procedures and patients were divided in 2 groups based on the occurrence of any hematologic toxicity. The proportion of LSCs was measured by sorting and FISH in patients with and without hematologic toxicity. The median percentage of LSCs did not significantly differ between the groups (P = 0.16, Mann–Whitney test), but all patients with hematologic toxicity had >89% of LSCs at the diagnosis. Abbreviations: hem AE, hematologic adverse events.

Close modal

LSCs rapidly disappear during nilotinib treatment

In vitro studies have suggested that TKIs do not effectively target LSCs or that their survival is BCR-ABL1 independent.4-10 Here we show by using two independent methods, that TKI treatment in vivo rapidly reduces LSC burden in the bone marrow. At 1 and 3 months after starting the treatment, 45 and 43 patient samples, respectively, were available for MPFC. From both of these time points, samples were evaluable by MPFC in 34 patients (Supplementary Fig. S2). Although at diagnosis LSCs were detectable by MPFC in 89% of patients (39/44), the level dropped to 18% (6/34 evaluable patients) at 1 month and 12% (4/34 evaluable patients) at 3 months. Notably, in all four patients with persisting LSCs at month 3, no detectable nHSCs were discovered by MPFC at diagnosis. One of these patients never reached MR3, two eventually attained a good response (MR3 at 9 and 12 months, respectively) and in one patient, the molecular response was unknown.

For sorting and FISH, at 3 months 15 patient samples were evaluable out of 23 bone marrow samples received (Supplementary Fig. S2). Residual LSCs were detected in the majority of patients (67%); however, the levels were relatively low (mean 10%, median 0.3%), and only three patients (20%) had more than 5% Ph+ CD34+CD38 cells (Supplementary Fig. S3). Two of three patients with >5% of LSCs by FISH had also detectable LSCs by MPFC method, and one patient was not evaluable. One of these three patients lost MR3 at 18 months, one obtained a good response, and one patient was lost to follow-up. Similarly, in the progenitor and overall bone marrow compartments, the percentage of Ph+ cells was low at the 3-month time point (Supplementary Fig. S3).

Because of the rapid LSC elimination kinetics during nilotinib therapy and the small patient numbers with persisting LSC at the early time points after treatment initiation, we were unable to properly analyze the association between molecular response and persisting LSCs at month 1 and 3.

LSCs in CML are quiescent, self-renewing cells that derive from nHSCs after acquiring the chromosomal translocation t(9;22)(q34;q11). As a consequence of the translocation, LSCs are characterized by deregulated cell-cycle activity, reduced apoptosis and adhesion, evasion of innate immunity and longevity (17, 18). As LSCs have been suggested to persist during TKI treatment, they may act as a reservoir for resistant clones and eventually be responsible for progression to advanced phases. In this clinical study, we used two independent methods (MPFC and stem cell sorting and FISH), which are able to identify putative CML stem cells as shown in previous studies by long-term clonogenic colony-forming assays (14). We were now able to demonstrate that identification and quantification of these leukemic cells in patients with newly diagnosed CML-CP reflect the biology of the disease, and also allows for the prediction of important response milestones during therapy with the second-generation TKI nilotinib. Our findings may help in identification of a subgroup of patients where more aggressive up front treatment is warranted.

The proportion of LSCs markedly differed between individual patients with CML at diagnosis, although all patients were in the chronic phase. The LSC burden correlated well with peripheral blood and bone marrow blast percentage, hemoglobin, spleen size, and Sokal risk score. This result is in agreement with our previous studies done both with the MPFC and sorting and FISH (14, 15). Interestingly, in the current study, we were unable to detect LSCs at diagnosis by MPFC in five patients. This could be due to limited sensitivity of the MPFC assay in case LSC numbers are low, and indeed, with the sorting and FISH analysis, we could establish that these patients had the lowest proportion of Ph+ cells.

Achievement of an early molecular response at 3 and 6 months has recently been included in the updated ELN criteria (19). Several studies have shown that the BCR-ABL1 level at these time points is a good predictor for later major molecular response, overall survival and progression-free survival (11, 20–25). Although all patients in this study reached the ELN-defined molecular goal of a BCR-ABL1 <10% according to the international scale at 3 months, BCR-ABL1 levels at that time point were significantly lower for patients who had detectable residual nHSCs at diagnosis than in those without detectable nHSCs.

At later time points during the therapy (after 12 months), patients with high initial LSC burden reached similar MR3 rates as patients with lower LSC burden. However, it was interesting to note that all patients not achieving MR3 at 12 or 18 months had >80% of Ph+ cells in CD34+CD38 fraction at the diagnosis. Whether detection of residual nHSCs represents earlier disease stage is unknown, but it may explain better treatment results in this patient group. Similarly, Landberg and colleagues (26) have studied the LSC burden by flow cytometry using IL1RAP as a marker for CML stem cells and found that the proportion of LSCs closely correlates with therapy outcome.

In vitro data have shown that TKI therapy is unable to eliminate LSCs, which is at least partly due to BCR-ABL1–independent survival of the quiescent cells (4, 6, 9, 10, 27–30). In concordance with the in vitro data, it has been shown that in almost all patients in sustained complete molecular response on the RNA level after discontinuation of imatinib, the BCR-ABL1 rearrangement DNA can still be detected in genomic DNA or in selected CD34+CD38 cells (10, 31). In our study, in most of the patients, the majority of LSCs were rapidly eliminated following 3 months of nilotinib treatment, with relatively high levels of LSCs persisting in only a subset of the patients. No other previous study has evaluated the disappearance of LSCs during nilotinib treatment, but in a study by Defina and colleagues (32), the persistence of CD34+ progenitor cells was examined. Ph+ cells were only detected in one of 20 patients who had been treated with nilotinib for a longer time (median 22 months), and surprisingly, in none of five first-line patients who had been on nilotinib treatment for only 3 months (32). In contrast, in our study cohort residual Ph+ stem cells were detected in 67% of patients at 3 months. The limited number of CD34+ cells analyzed in the former study (median 100 cells) may at least in part explain the differences between these two studies as the level of Ph+ cells found in our study was very low (median 0.3% based on 1,000 cells analyzed). This low level is in accordance with our previous study analyzing first-line imatinib- and dasatinib-treated patients (15) where we also found that TKI therapy rapidly eliminated the majority of LSCs (median 0.2% at 3 months). Similarly to our previous studies (14, 15), therapy-related hematologic toxicity was observed in patients who had high LSC percentage at diagnosis. We postulate that this may mirror a smaller residual nHSC pool, which is insufficient in maintaining the normal hematopoiesis even though the leukemic progenitors and LSCs are efficiently eliminated during the initial phase of nilotinib treatment.

Although slightly different definitions of the stem cell compartments were used, the results of these two independent stem cell quantification methods (MPFC and sorting and FISH) correlated quite well. MPFC has the additional benefit of enabling sorting of viable leukemic and normal stem cells into separate fractions. This permits functional evaluation as well as expression profiling and comparison of LSCs versus nHSCs. This may facilitate the identification of novel targets on LSCs and ultimately enable the development of LSC-specific targeted therapy concepts. As to the detection of residual disease during therapy, the cell sorting and FISH method appears to be more sensitive than MPFC.

In conclusion, we here demonstrate that leukemic burden of the CD34+CD38 compartment correlates with biologic disease parameters and may predict the achievement of later molecular response to nilotinib therapy in first-line–treated patients with CML already at the diagnosis. Furthermore, the proportion of leukemic cells rapidly decreases in the CD34+CD38 stem cell compartment upon start of nilotinib therapy together with a disappearance of leukemic cells from more mature progenitor and whole bone marrow fractions.

All authors received research support from Novartis for this study. J. Richter reports receiving speakers bureau honoraria from Novartis and is a consultant/advisory board member for Ariad and Novartis. G. Barbany reports receiving a commercial research grant from Novartis. T. Fioretos has ownership interest (including patents) in and is a consultant/advisory board member for Cantargia and reports receiving commercial research grants from Bristol-Myers Squibb, Cantargia, and Novartis. F.J. Giles reports receiving a commercial research grant from and is a consultant/advisory board member for Novartis. B.T. Gjertsen reports receiving a commercial research grant and speakers bureau honoraria from Novartis and is a consultant/advisory board member for Ariad. A. Hochhaus and D. Wolf report receiving commercial research grants from and are consultants/advisory board members for Ariad, Bristol-Myers Squibb, Novartis, and Pfizer. G.J. Schuurhuis reports receiving commercial research support from Becton Dickinson. G. Ossenkoppele reports receiving a commercial research grant from Novartis and is a consultant/advisory board member for Ariad, Bristol-Myers Squibb and Novartis. K. Porkka reports receiving commercial research grants from Bristol-Myers Squibb, Novartis, and Pfizer. S. Mustjoki reports receiving commercial research grants from Ariad, Novartis, and Pfizer and speakers bureau honoraria from and is a consultant/advisory board member for Bristol-Myers Squibb, Novartis and Pfizer. No potential conflicts of interest were disclosed by the other authors.

Conception and design: J. Richter, F. Giles, A. Hochhaus, D. Wolf, G. Ossenkoppele, K. Porkka, J. Janssen, S. Mustjoki

Development of methodology: J. Richter, G.J. Schuurhuis, L. Stenke, J. Janssen, S. Mustjoki

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): J. Richter, G. Barbany, T. Fioretos, F. Giles, B.-T. Gjertsen, A. Hochhaus, G.J. Schuurhuis, L. Stenke, S. Thunberg, D. Wolf, G. Ossenkoppele, J. Janssen, S. Mustjoki

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): N. Thielen, J. Richter, M. Baldauf, B.-T. Gjertsen, A. Hochhaus, G.J. Schuurhuis, S. Sopper, L. Stenke, J. Janssen, S. Mustjoki

Writing, review, and/or revision of the manuscript: N. Thielen, J. Richter, M. Baldauf, G. Barbany, T. Fioretos, F. Giles, B.-T. Gjertsen, A. Hochhaus, G.J. Schuurhuis, S. Sopper, L. Stenke, S. Thunberg, D. Wolf, G. Ossenkoppele, K. Porkka, J. Janssen, S. Mustjoki

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): F. Giles, B.-T. Gjertsen, S. Sopper, L. Stenke, J. Janssen, S. Mustjoki

Study supervision: F. Giles, B.-T. Gjertsen, A. Hochhaus, G.J. Schuurhuis, G. Ossenkoppele, K. Porkka, J. Janssen, S. Mustjoki

The authors thank the personnel at the participating stem cell laboratories at the VU University Medical Center, (Amsterdam, the Netherlands), Hematology Research Unit (Helsinki, Finland), Department of Clinical Genetics, Lund University (Lund, Sweden), and Department of Medicine, Karolinska University Hospital (Stockholm, Sweden) for their expert technical assistance. Novartis is acknowledged for financially supporting this study. The authors also thank patients, study nurses, and contributing investigators (listed below) for their participation in this trial:

Tobias Gedde-Dahl, Oslo University Hospital, Oslo, Norway

Henrik Hjorth-Hansen, Trondheim University Hospital, Trondheim, Norway

Perttu Koskenvesa, Helsinki University Central Hospital Cancer Center, Helsinki, Finland

Alois Lang, Hospital Feldkirch, Feldkirch, Austria

Carin Lassen, Skåne University Hospital, Lund, Sweden

Kirsi Latvala, Helsinki University Central Hospital Cancer Center, Helsinki, Finland

Hanna Lähteenmäki, Helsinki University Central Hospital Cancer Center, Helsinki, Finland

Werner Linkesch, Medical University of Graz, Graz, Austria

Waleed Majeed, Stavanger University hospital, Stavanger, Norway

Claes Malm, Linköping University hospital, Linköping, Sweden

Kristina Myhr Eriksson, Sunderby Sjukhus, Luleå, Sweden

Lotta Ohm, Karolinska University Hospital, Stockholm, Sweden

Anton Schattenberg, Radboud University Nijmegen Medical Centre, Nijmegen, the Netherlands

Willem Smit, Medisch Spectrum Twente, Enschede, the Netherlands

Jesper Stentoft, Aarhus University Hospital, Aarhus, Denmark

Ulla Olsson-Strömberg, Uppsala University Hospital, Uppsala, Sweden

Josef Thaler, Wels-Grieskirchen Hospital, Wels Austria

Jan Van Droogenbroeck, Academisch Ziekenhuis Sint Jan, Brugge, Belgium

Edo Vellenga, Universitair Medisch Centrum Groningen, Groningen, the Netherlands

Gregor Verhoef, University Hospital Gasthuisberg, Leuven, Belgium.

D. Wolf was supported by Öwas supported by iversi Bank (Jubiläumsfondsprojekt Nr. 14781) and S. Mustjoki by Academy of Finland, Finnish Cancer Institute, and Finnish Cancer Organizations.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Deininger
M
,
O'Brien
SG
,
Guilhot
F
,
Goldman
JM
,
Hochhaus
A
,
Hughes
TP
, et al
International randomized study of interferon vs. STI571 (IRIS) 8-year follow up: sustained survival and low risk for progression or events in patients with newly diagnosed chronic myeloid leukemia in chronic phase (CML-CP) treated with imatinib
.
Blood
2009
;
114
:
1126
.
2.
Rosti
G
,
Palandri
F
,
Castagnetti
F
,
Breccia
M
,
Levato
L
,
Gugliotta
G
, et al
Nilotinib for the frontline treatment of Ph(+) chronic myeloid leukemia
.
Blood
2009
;
114
:
4933
8
.
3.
Cortes
JE
,
Jones
D
,
O'Brien
S
,
Jabbour
E
,
Ravandi
F
,
Koller
C
, et al
Results of dasatinib therapy in patients with early chronic-phase chronic myeloid leukemia
.
J Clin Oncol
2010
;
28
:
398
404
.
4.
Graham
SM
,
Jorgensen
HG
,
Allan
E
,
Pearson
C
,
Alcom
MJ
,
Richmond
L
,
Holyoake
TL
. 
Primitive, quiescent, Philadelphia-positive stem cells from patients with chronic myeloid leukemia are insensitive to STI571 in vitro
.
Blood
2002
;
99
:
319
25
.
5.
Copland
M
,
Hamilton
A
,
Elrick
LJ
,
Baird
JW
,
Allan
EK
,
Jordanides
N
, et al
Dasatinib (BMS-354825) targets an earlier progenitor population than imatinib in primary CMLbut does not eliminate the quiescent fraction
.
Blood
2006
;
107
:
4532
9
.
6.
Jorgensen
HG
,
Allan
EK
,
Jordanides
NE
,
Mountford
JC
,
Holyoake
TL
. 
Nilotinib exerts equipotent antiproliferative effects to imatinib and does not induce apoptosis in CD34
+
CML cells
.
Blood
2007
;
109
:
4016
9
.
7.
Konig
H
,
Holtz
M
,
Modi
H
,
Manley
P
,
Holyoake
TL
,
Forman
SJ
,
Bhatia
R
. 
Enhanced BCR-ABL kinase inhibition does not result in increased inhibition of downstream signaling pathways or increased growth suppression in CML progenitors
.
Leukemia
2008
;
22
:
748
55
.
8.
Konig
H
,
Holyoake
TL
,
Bhatia
R
. 
Effective and selective inhibition of chonic myeloid leukemia primitive hematopoietic progenitors by the dual Src/Abl kinase inhibitor SKI-606
.
Blood
2008
;
111
:
2329
38
.
9.
Corbin
AS
,
Agarwal
A
,
Loriaux
M
,
Cortes
J
,
Deininger
MW
,
Druker
BJ
. 
Human chronic myeloid leukemia stem cells are insensitive to imatinib despite inhibition of BCR-ABL activity
.
J Clin Invest
2011
;
121
:
396
409
.
10.
Chu
S
,
McDonald
T
,
Lin
A
,
Chakraborty
S
,
Huang
Q
,
Snyder
DS
,
Bhatia
R
. 
Persistence of leukemia stem cells in chronic myelogenous leukemia patients in prolonged remission with imatinib treatment
.
Blood
2011
;
118
:
5565
72
.
11.
Marin
D
,
Ibrahim
AR
,
Lucas
C
,
Gerrard
G
,
Wang
L
,
Szydlo
RM
. 
Assessment of BCR-ABL1 transcript levels at 3 months is the only requirement for predicting outcome for patients with chronic myeloid leukemia treated with tyrosine kinase inhibitors
.
J Clin Oncol
2012
;
30
:
232
8
.
12.
Sokal
JE
,
Cox
EB
,
Baccarani
M
,
Tura
S
,
Gomez
GA
,
Robertson
JE
, et al
Prognostic discrimination in "good-risk" chronic granulocytic leukemia
.
Blood
1984
;
63
:
789
99
.
13.
Hasford
J
,
Pfirrmann
M
,
Hehlmann
R
,
Allan
NC
,
Baccarani
M
,
Kluin-Nelemans
JC
, et al
A new prognostic score for survival of patients with chronic myeloid leukemia treated with interferon alfa. Writing committee for the collaborative CML prognostic factors project group
.
J Natl Cancer Inst
1998
;
90
:
850
8
.
14.
Janssen
JJ
,
Deenik
W
,
Smolders
KG
,
van Kuijk
BJ
,
Pouwels
W
,
Kelder
A
, et al
Residual normal stem cells can be detected in newly diagnosed chronic myeloid leukemia patients by a new flow cytometric approach and predict for optimal response to imatinib
.
Leukemia
2012
;
26
:
977
84
.
15.
Mustjoki
S
,
Richter
J
,
Barbany
G
,
Ehrencrona
H
,
Fioretos
T
,
Gedde Dahl
T
, et al
Impact of malignant stem cell burden on therapy outcome in newly diagnosed chronic myeloid leukemia patients
.
Leukemia
2013
;
27
:
1520
6
.
16.
Hochhaus
A
,
Rosti
G
,
Cross
NC
,
Steegmann
JL
,
le Coutre
P
,
Ossenkoppele
G
, et al
Frontline nilotinib in patients with chronic myeloid leukemia in chronic phase: results from the European ENEST1st study
.
Leukemia
2016
;
30
:
57
64
.
17.
Rice
KN
,
Jamieson
CH
. 
Molecular pathways to CML stem cells
.
Int J Hematol
2010
;
91
:
748
52
.
18.
Nicholson
E
,
Holyoake
T
. 
The chronic myeloid leukemia stem cell
.
Clin Lymphoma Myeloma
2009
;
9
Suppl 4
:
S376
S81
.
19.
Baccarani
M
,
Deininger
MW
,
Rosti
G
,
Hochhaus
A
,
Soverini
S
,
Apperley
JF
, et al
European LeukemiaNet recommendations for the management of chronic myeloid leukemia: 2013
.
Blood
2013
;
122
:
872
84
.
20.
Hughes
TP
,
Hochhaus
A
,
Branford
S
,
Müller
MC
,
Kaeda
JS
,
Foroni
L
, et al
Long-term prognostic significance of early molecular response to imatinib in newly diagnosed chronic myeloid leukemia: an analysis from the International Randomized Study of Interferon and STI571 (IRIS)
.
Blood
2010
;
116
:
3758
65
.
21.
Hanfstein
B
,
Muller
MC
,
Hehlmann
R
,
Erben
P
,
Lauseker
M
,
Fabarius
A
, et al
Early molecular and cytogenetic response is predictive for long-term progression-free and overall survival in chronic myeloid leukemia (CML)
.
Leukemia
2012
;
26
:
2096
102
.
22.
Marin
D
,
Hedgley
C
,
Clark
RE
,
Apperley
J
,
Foroni
L
,
Milojkovic
D
, et al
Predictive value of early molecular response in patients with chronic myeloid leukemia treated with first-line dasatinib
.
Blood
2012
;
120
:
291
4
.
23.
Neelakantan
P
,
Gerrard
G
,
Lucas
C
,
Milojkovic
D
,
May
P
,
Wang
L
, et al
Combining BCR-ABL1 transcript levels at 3 and 6 months in chronic myeloid leukemia: implications for early intervention strategies
.
Blood
2013
;
121
:
2739
42
.
24.
Hochhaus
A
,
Hughes
TP
,
Saglio
G
,
Guilhot
F
,
Al-Ali
HK
,
Rosti
G
, et al
Outcome of patients with chronic myeloid leukemia in chronic phase (CML-CP) based on early molecular response and factors associated with early response: 4-year follow-up data from Enestnd (Evaluating Nilotinib Efficacy and Safety in Clinical Trials Newly Diagnosed Patients)
.
Blood
2012
;
120
:
167
.
25.
Jain
P
,
Kantarjian
HM
,
Nazha
A
,
Jabbour
E
,
Quintas-Cardama
A
,
Benjamini
O
, et al
Early molecular and cytogenetic responses predicts for significantly longer event free survival (EFS) and overall survival (OS) in patients (pts) with newly diagnosed chronic myeloid leukemia (CML) in chronic phase (CP)—an analysis of 4 tyrosine kinase inhibitor (TKI) modalities (standard dose imatinib, high dose imatinib, dasatinib and nilotinib
.
Blood
2012
;
120
:
70
.
26.
Landberg
N
,
Hansen
N
,
Askmyr
M
,
Ågerstam
H
,
Lassen
C
,
Rissler
M
, et al
IL1RAP expression as a measure of leukemic stem cell burden at diagnosis of chronic myeloid leukemia predicts therapy outcome
.
Leukemia
2016
;
30
:
255
8
.
27.
Jamieson
CH.
Chronic myeloid leukemia stem cells
.
Hematology Am Soc Hematol Educ Program
2008
;
436
42
.
28.
Kavalerchik
E
,
Goff
D
,
Jamieson
CH
. 
Chronic myeloid leukemia stem cells
.
J Clin Oncol
2008
;
26
:
2911
5
.
29.
Hamilton
A
,
Helgason
GV
,
Schemionek
M
,
Zhang
B
,
Myssina
S
,
Allan
EK
, et al
Chronic myeloid leukemia stem cells are not dependent on Bcr-Abl kinase activity for their survival
.
Blood
2012
;
119
:
1501
10
.
30.
Morotti
A
,
Panuzzo
C
,
Fava
C
,
Saglio
G
. 
Kinase-inhibitor-insensitive cancer stem cells in chronic myeloid leukemia
.
Expert Opin Biol Ther
2014
;
14
:
287
99
.
31.
Ross
DM
,
Branford
S
,
Seymour
JF
,
Schwarer
AP
,
Arthur
C
,
Bartley
PA
, et al
Patients with chronic myeloid leukemia who maintain a complete molecular response after stopping imatinib treatment have evidence of persistent leukemia by DNA PCR
.
Leukemia
2010
;
24
:
1719
24
.
32.
Defina
M
,
Ippoliti
M
,
Gozzetti
A
,
Abruzzese
E
,
Castagnetti
F
,
Crupi
R
, et al
Evaluation of residual CD34(+) Ph(+) progenitor cells in chronic myeloid leukemia patients who have complete cytogenetic response during first-line nilotinib therapy
.
Cancer
2012
;
118
:
5265
9
.