Peripheral Blood CD26+ Leukemia Stem Cells Monitoring in Chronic Myeloid Leukemia Patients from Diagnosis to Response to TKIs: Interim Results of a Multicenter Prospective Study (PROSPECTIVE FLOWERS)

Background

We already showed that in CML pts peripheral blood CD34+/CD38-/CD26+ cell population represent a “CML specific” leukemia stem cell (LSC) circulating compartment. Indeed, we demonstrated that CD26+LSCs are measurable by flow-cytometry in 100% of CML pts at diagnosis the latter representing an alternative and rapid diagnostic tool. In addition, in a cross-sectional study we were able to spot peripheral blood CD26+LSCs also in about 65% of CML during TKI treatment regardless of type and length of TKI treatment and degree of molecular response. However, no prospective data are available regarding the behavior of PB CD26+LSCs in terms of rate and timing of reduction during TKI therapy and the correlation, if any, with the attainment of response according to ELN guidelines. Interestingly, even CML patients in stable TFR may harbor circulating CD26+LSCs thus suggesting a probable active role of the immune system in the control of residual disease. One hypothesis could reside in the presence or absence on the LSCs of molecules (such as PD-L1) able to hamper an anti-leukemic T cell response. From Jan 2018 we conducted a prospective multicenter Italian study including CML pts at diagnosis treated and managed by each of 15 participating center according to ELN guidelines. We here present the first interim analysis after a median time of treatment of 12 mos.

Aims

The main goals of this study were to prospectively monitoring PB CD26+LScs in CML pts during TKI treatment and to correlate the behavior of LSCs with molecular response. In a proportion of pts PD-L1 expression on CD26+ LSCs at diagnosis was also evaluated.

Methods

At diagnosis and during TKI treatment, pts have been centrally evaluated in Siena lab for flow-cytometry PB CD26+ LSCs (+3, +6, +12, +18, +24 mos) and PD-L1 expression (at diagnosis). At each time point molecular BCR-ABL/ABL^IS ratio was monitored locally in each center.

Results

176 consecutive CML pts (IMA 92; NILO 61; DASA 23) were enrolled in the study so far (table 1). PB CD26+LSCs were measured at time 0 (baseline) in all 176 CML pts and in 165/176 (94%), 142/176 (81%) and 112/176 (71%) at +3, +6, +12 mos of TKI treatment, respectively. Median CD26+LSCs absolute number/µL at baseline was 6.96/µL (range 0.0126-64429), at +3 mos 0.0137/µL (range 0-6,49), at +6 mos 0.0056/µL (range 0-1.188), and at +12 mos 0.0112/µL (range 0-0.1824). No significant correlation between number of CD26+LSC, degree of response and BCR-ABL copies was found (Table 2). Interestingly, median CD26+LSCs at diagnosis was found significantly higher in NILO and DASA treated pts (12, 48/µL and 17,48/µL, respectively) than in IMA pts (4,58/µL). So far, 20/176 (11.4%) pts switched to different TKIs, due to failure/suboptimal response: of note, median CD26+ LSCs of this cohort at diagnosis was the highest (23.12/µL).

Starting from Jun 2019, 44/176 (25%) CML pts have been evaluated also for PD-L1 expression at diagnosis: of these, 23/44 (52%) resulted PD-L1 positive and 21/44 (48%) resulted negative with a median of CD26+LSCs of 15.39/µL (range 1.28-635.5) and 4.45/µL (range 0.234-113.9), respectively.

Conclusions

After a sensible drop observed at 3 mos of any TKI treatment, CD26+LSCs are fluctuating and measurable at low level in most of pts (> 65%) even at 18 and 24 mos. We confirmed no correlation between the absolute number of persisting CD26+LSCs and BCR-ABL copies. However, pts with failure or suboptimal response showed the highest level of CD26+ at diagnosis. CD26+LSCs were found PD-L1+ in about half of 44 pts tested. At diagnosis higher CD26+LSCs number, PD-L1 positivity or both may correlate with a lower probability to achieve an optimal response; interim data of this first report will be presented; enrolment and follow up are ongoing.