Assessment of haematopoietic progenitor cell counting with the Sysmex^® XN-1000 to guide timing of apheresis of peripheral blood stem cells

Abstract

Background

Successful peripheral blood stem cell (PBSC) collection depends on optimal timing of apheresis, as usually determined by flow cytometry CD34-positive (⁺) cell count in peripheral blood (PB). Since this method is costly and labour-intensive, we evaluated the use of the Hematopoietic Progenitor Cell count programme on a Sysmex^® XN haematologic analyser (XN-HPC) as a rapid and inexpensive alternative for predicting CD34⁺ cell count in PB samples.

Materials and methods

Haematopoietic progenitor cell and CD34⁺ cell counts were compared using 273 PB samples collected from 78 healthy donors and 72 patients who underwent PBSC transplantation. We assessed the effectiveness of the XN-HPC in safely predicting pre-harvest CD34⁺ counts. The most efficient cut-off values of XNHPC were identified. We also evaluated the imprecision (coefficient of variation, CV) and functional sensitivity.

Results

Imprecision of the XN-HPC count was <6.3% on daily measurement of three levels of quality control material. Functional sensitivity was 8.9×10⁶/L. A cut-off value of ≥62×10⁶/L XN-HPC for multiple myeloma (MM) patients and ≥30×10⁶/L for all other subjects had both 100% specificity and 100% positive predictive value for identifying samples with CD34⁺ cells ≥20×10⁶/L. An XN-HPC threshold of <13×10⁶/L identified preharvest CD34⁺ cell count <10×10⁶/L with 100% sensitivity and 100% negative predictive value.

Discussion

The XN-HPC is a fast, easy and inexpensive test that can safely improve apheresis workflow thus possibly replacing other more expensive CD34 counts currently performed and promoting optimal timing of PBSC collection.

INTRODUCTION

Haematopoietic stem cell transplantation (HSCT) is a well-established treatment for haematologic cancers, solid tumours, and immunological or metabolic disorders^1–3. The 2016 activity survey report of the European Society for Blood and Bone Marrow Transplant (EBMT)⁴ showed a continued increase in the use of HSCT across Europe, with more than 43,636 transplants performed annually in 39,313 patients (42% allogeneic, 58% autologous) by 679 centres in 49 different countries. Stem cells are mostly collected from peripheral blood (PB) after stem cell mobilisation using haematopoietic growth factors either associated, or not, with chemotherapy^4,5.

Successful peripheral blood stem cell (PBSC) transplantation depends on infusion of an appropriate number of HSCs to achieve rapid and durable haematologic recovery. Both the evaluation of circulating HSCs and the decision to start apheresis are currently based on enumeration of CD34⁺ cells by flow cytometry, according to a protocol proposed by the International Society of Hematotherapy and Graft Engineering (ISHAGE)^6–8. Although well standardised, this method is expensive and labour-intensive; it also needs specific equipment and experienced staff.

Historically, white blood cell (WBC) count using automated haemocytometers has been used as a marker of bone marrow response to stem cell mobilisation because it is quick and cheap^9–11. Nevertheless, some studies showed poor correlations between WBCs and CD34⁺ cell count^12–14, which is then, therefore, potentially associated with inefficient collection and increased overall costs of PBSC for transplantation.

Automated applications have been recently developed on Sysmex^® analysers of the SE and XE series, specifically aimed at identifying haematopoietic progenitor cells (HPC)^15,16. Despite the fairly good correlations observed with the CD34⁺ cell yield in apheresis products^17–22, automated HPC counts were found to be of limited value since they were likely biased by the presence of co-existing immature granulocytes, thus resulting in a significant risk of under- or overestimation^18,23,24. Recently, Sysmex Corp. (Kobe, Japan) has developed another innovative application for the XN series analysers (the “XN-Stem Cells mode”). Preliminary evaluations showed comparable results between XN-HPC and CD34⁺ cell counts determined by flow cytometry^25–26. These results were confirmed in a recent evaluation performed using a larger cohort of patients²⁷ and in a multicentre study including three different facilities²⁸. Therefore, this study aimed to compare the XN-HPC count with the reference CD34⁺ count in mobilised PB samples collected from a heterogeneous population of donors and patients. Our main aim was to determine whether XN-HPC count can help assessing the optimal time of apheresis, thus replacing some of the CD34⁺ count systems currently used in PB samples.

MATERIALS AND METHODS

Patient population and mobilised blood samples

Patients and donors included in this study were referred to the Haematology Bone Marrow Transplant Centre (HBMTC) of our institution for autologous or allogenic PBCS transplantation (University Hospital of Verona, Italy). A total of 273 mobilised PB samples collected from 78 healthy donors and 72 patients who underwent PBSC transplantation between March 2016 and May 2018 were randomly selected. Patients’ and donors’ characteristics, as well as the number of PB samples in each group, are listed in Table I.

Table I.

Characteristics of patients and donors included in the study, number of mobilised peripheral blood (PB) samples and values distribution of HPC and CD34-positive (+) cell counts

Groups	N. of patients	N. of samples	CD34+×106/L Median (range)	XN-HPC×106/L Median (range)	p-valued
ALL	150	273	47.4 (0.0–631.0)	55.0 (0.0–583.0)	0.0019*
Donorsa	78	97	73.0 (3.5–297.0)	75.0 (9.0–263.0)	0.318
Lymphomasb	31	82	27.8 (0.0–631)	24.0 (0.0–583.0)	0.076
MM	37	83	34.7 (0.0–214.0)	56.0 (0.0–196.0)	<0.0001*
Other diseasesc	4	11	6.0 (0.5–149.6)	7.0 (0.0–135)	0.625
ALL without MM	113	190	51.7 (0.0–631.0)	53.5 (0.0–583.0)	0.858

^aDonors: 25 related (33 samples); 53 unrelated (64 samples).

^bIncludes Hodgkin’s lymphomas (7 patients, 14 samples) and non-Hodgkin’s lymphomas (21 patients, 68 samples).

^cIncludes Ewing Sarcoma (2 patients, 5 samples), neuroblastoma (1 patient, 5 samples), medulloblastoma (1 patient, 1 sample).

^dWilcoxon-test: comparison between median values of HPC and CD34 + counts.

^*p<0.05 was considered statistically significant. Significant values in italic.

ALL: acute lymphpblastic leukemia; MM: multiple myleoma; HPC: haematopoietic progenitor cells.

Peripheral blood stem cells were mobilised following the HBMTC protocol, by using granulocyte-colony stimulating factor (G-CSF), cytostatic chemotherapy and G-CSF, or G-CSF and Plerixafor (Mozobil, Sanofi US, Bridgewater, NJ, USA). Patients received the appropriate high-dose cytostatic chemotherapy according to their diagnosis. The cytostatic treatment was followed by administration of G-CSF at a dose of 5 μg/kg/day for patients with lymphomas and 5 μg/kg/day for patients with multiple myeloma (MM). Healthy donors were only administered G-CSF at a dose of 10 μg/kg/day. The CD34⁺ assessment was performed starting from 3–4 days after the first administration of G-CSF in healthy donors and after 10–15 days after chemotherapy. Apheresis was initiated when the number of CD34⁺ cells in PB reached at least 20×10⁶/L or was between 10–20×10⁶/L in some specific cases, according to patient characteristics.

All PB samples included in the study were anticoagulated with K₃EDTA and analysed by CD34⁺ flow cytometry and XN-Stem Cells mode within 2 hours (h) of collection.

The study was carried out in accordance with the Declaration of Helsinki and the terms of local legislation. All analyses on the XN-Stem Cells mode were performed using pre-existing samples, and results were not used for clinical decision-making on patient management.

CD34⁺ cell count

CD34⁺ cells were quantified with a single platform method using the ISHAGE protocol with BD Stem Cell Enumeration kit (BD Biosciences, San Jose, CA, USA) on a BD FACScalibur flow cytometer (BD Biosciences). The local laboratory participates in an external quality control (QC) programme for CD34 Stem Cell Enumeration (UK NEQAS, Sheffield, UK).

XN-HPC count

Automated HPC counts were performed using a single Sysmex^® XN-1000 analyser (Sysmex Corp.), up-graded with the XN-Stem Cells mode application. The HPCs are identified in the white pathological and progenitor cells (WPC) channel, which is specifically designed to detect pathological cells such as myeloblasts and abnormal lymphocytes²⁵. With the XN-Stem Cells mode, HPC counting has been improved through detection of fluorescence staining of cell nucleic acids and optimisation of temperature and reaction time of reagents in the WPC channel. The analysis is performed using 190 μL of PB samples, or apheresis products. The mean value of four separate HPC counts is directly reported by the analyser in approximately 4 min, with no need for manual gating, pre-treatment, or sample washing.

The HPC counts were reported as absolute value (HPC#) and as a percentage (HPC%, calculated with respect of total WBC count). Both are reportable diagnostic parameters, CE-IVD marked for European countries. Results can be monitored daily using the XN-Chex control blood (Sysmex Corp.), a 3-level control material used for the daily QC of all the parameters determined by the Sysmex^® XN analyser.

The XN-Stem Cells mode uses fluorescent flow cytometry for quantification of HPC# and HPC%. Cells are first treated with a lysing agent (the WPC reagent) which permeabilises the cell membrane of WBCs, which is then followed by the addition of a fluorescent dye for staining DNA. Flow cytometry analysis allows cells to be categorised according to forward-scatter light (FSC), side-scatter light (SSC), and fluorescence intensity (SFL), which reflect cell size, cell granularity/internal complexity and cell DNA content, respectively. Fluorescence intensity of cells depends on their relative permeability to WPC reagent. Immature stem cells differ from more mature progenitor cells by membrane lipid composition, which makes them more resistant to permeabilisation by WPC reagent. As a result, stem cells are less stained compared to other cell populations, as the fluorescent dye cannot easily enter the cell and bind to the DNA. Stem cells are further differentiated according to FSC (medium to low) and SSC (low), as medium size elements with low granularity and modest intracellular complexity. All these HPC detection patterns have been largely improved on XN-Stem Cells mode, through modification of temperature and reaction time of reagents in the WPC channel.

Between-day precision, reproducibility and functional sensitivity

Between-day precision of XN-HPC count was assessed by daily analysis for 40 days of a single lot of three levels of Sysmex^® XN-Chex control material. Mean values, standard deviation (SD), and co-efficient of variation (CV) were calculated.

Reproducibility (i.e., within-day imprecision) was assessed by measuring ten replicates of six samples with different cell concentrations (i.e., 6.9–169.6×10⁶/L XN-HPC cells). The mean value of each sample was plotted against the CV, and functional sensitivity was then calculated from a power regression analysis as minimum value yielding an analytical imprecision (CV) ≤20%.

Comparison studies

The distribution of XN-HPC and CD34⁺ cell counts was evaluated on all samples and in each group of patients and donors (Table II), and compared with Wilcoxon’s test for paired data. p<0.05 was considered statistically significant. XN-HPC and CD34⁺ cell counts were also compared with Passing-Bablok regression analysis, Pearson’s correlation coefficient, and Bland-Altman plots. Statistical significance was based on 95% confidence intervals (CI): a significant proportional or systematic difference in Passing-Bablok regression analysis was considered to be that at which the numbers 1 and 0 were not included in the 95% CI of either slope or intercept. In Bland-Altman plots, absolute differences were plotted against the results of the CD34⁺ cell count. A significant bias was considered to be that at which the 0 value was not included in the 95% CI.

Table II.

Comparison between XN-HPC and CD34+ cell counts.

Groups	r valuea	Passing Bablok regression		Bland-Altman difference plot (×106/L)
		Slope (95%CI)	Interceptb (95%CI)	Mean bias (95%CI)	95% limits of agreement (mean bias ±1.96 SD)
ALL	0.926	0.99 (0.93–1.05)	2.97 (1.00–5.00)	2.80 (−0.29–5.99)	−48.12–53.72
Donors	0.849	0.88 (0.77–1.00)	11.97 (3.84–18.21)	0.10 (−6.46–6.65)	−63.63–63.83
Lymphomas	0.976	0.90 (0.80–0.99)	1.15 (−0.03–3.00)	−4.68 (−9.06-−0.30)	−43.78–34.42
MM	0.887	1.17 (1.00–1.32)	7.03 (0.35–11.09)	14.30 (10.01–18.59)	−24.21–52.81
Other diseasesc	0.975	0.88 (0.48–1.07)	1.21 (−0.43–4.82)	−4.34 (−12.25–3.57)	−27.41–18.73
ALL without MM	0.937	0.96 (0.91–1.02)	2.03 (0.03–3.41)	−2.22 (−6.06–1.62)	−54.83–50.39

^ar value: Pearson’s coefficient correlation.

^bIntercept: 95% Confidence Interval (CI): ×10⁶/L.

^cEwing sarcoma, neuroblastoma, medulloblastoma.

MM: multiple myeloma; ALL: acute lymphoblastic leucaemia; SD: standard deviation.

Diagnostic accuracy

The diagnostic accuracy of XN-HPC was assessed by the area under the curve (AUC) in receiver operating characteristic (ROC) analysis. Sensitivity (SE), specificity (SP), negative predictive value (NPV), and positive predictive value (PPV) were calculated at the corresponding XN-HPC count which best predicted a PB CD34⁺ cell count ≥20 CD34⁺×10⁶/L (i.e., the threshold used for starting apheresis in the local institution). We also identified the cut-off values of XN-HPC count capable of maximising its efficiency to be used as a “rule-in” and “rule-out” test for starting apheresis (i.e., showing 100% SP and PPV, or 100% SE and NPV). Furthermore, considering that apheresis can start at a PB CD34⁺ cell count of 10–20×10⁶/L, we evaluated the best XN-HPC cut-off for ruling out a CD34⁺ cell count <10×10⁶/L.

We finally evaluated the kinetics of XN-HPC and CD34⁺ cells in 31 patients (16 lymphomas, 1 Ewing Sarcoma, and 1 neuroblastoma, 13 multiple myelomas) for whom at least three measurements were available on consecutive days, thus exploring whether the changes in the two parameters were similar over time.

Statistical analysis

Statistical analysis was performed using Analyse-it software version 5.11.3 (Analyse-it software Ltd., Leeds, UK) and Microsoft Excel 2010.

RESULTS

Between-day precision, reproducibility and functional sensitivity

In the reproducibility study, the imprecision ranged between 4.3 and 22.9% in the range of XN-HPC values tested (i.e., 6.9–169.6×10⁶/L). The between-day imprecision calculated using three levels of XN-Chex control material was 3.6%, 5.3% and 6.3% for XN-HPC counts of 168.5×10⁶/L, 70.3×10⁶/L and 30.8×10⁶/L, respectively. The mean values of the 40 daily analyses were very similar to the assigned HPC values, with relative bias ranging from −0.6% for the low concentration to 2.7% for the high concentration. Similar results were obtained with all the lots of XN-Chex material used during the evaluation (data not shown). The final calculated functional sensitivity was 8.9×10⁶/L.

Comparison of mobilised peripheral blood samples

The distribution of HPC and CD34⁺ cell counts (median values and range) in all samples of all groups are shown in Table I. Statistically significant differences were observed in MM patients, in whom the median XN-HPC value was 1.6-fold higher than the CD34⁺ cell count. No significant differences were observed in other groups, including all samples without MM patients.

The results of comparitive studies are summarised in Table II. A good agreement was observed between the two methods in all the 273 samples (r=0.93; y=0.99+2.97; mean bias of 2.80×10⁶/L) (Table II and Figure 1A). A strong correlation was also found in the 82 samples collected from lymphoma patients (r=0.98), despite a modest underestimation of XN-HPC noted in Passing-Bablok regression analysis (95% CI of slope: 0.80–0.99) and Bland-Altman plot (95% CI of mean bias: −9.06 to −0.30×10⁶/L) (Table II and Figure 1C). In healthy donors, the correlation was fairly good (r=0.85), but a significant bias was noted (95% CI of intercept: 3.84–18.21×10⁶/L). The XN-HPC values tended to be slightly higher in samples with CD34⁺ cell counts <100.0×10⁶/L, and lower in samples with CD34⁺ cell counts ≥100.0×10⁶/L (Table II and Figure 1B). However, these differences were mostly irrelevant at the 20.0×10⁶/L CD34⁺ cell threshold. In the 11 “Other diseases” samples collected from 4 different patients (2 with Ewing sarcoma, 1 with neuroblastoma and 1 with medulloblastoma), the correlation was optimal (r=0.98). However, due to the limited number of samples a clear statistical interpretation of data was not feasible. The most relevant differences were observed in MM patients, with statistically significant overestimation of XN-HPC counts (slope 1.17; 95% CI: 1.00–1.32; mean bias 14.30×10⁶/L; 95% CI: 10.01–18.59×10⁶/L) (Table II and Figure 1D).

Figure 1.

Correlation of the Sysmex XN haematologic XN-HPC analyser and CD34-positive (⁺) cells count by Passing-Bablok regression analysis.

Solid and dashed lines represent the regression line and 95% Confidence Interval (CI) of the slope, respectively. Grey line represents the identity line (y=x). Correlations between XN-HPC and CD34⁺ cells are shown for (A) all the 273 peripheral blood (PB) samples, (B) healthy donors (n=97), (C) lymphomas (n=82), (D) multiple myeloma (n=83). See Table II for data of Passing-Bablok regression analysis.

Diagnostic accuracy

Overall, 207 of 273 mobilised PB samples (75.8%) had a CD34⁺ cell count ≥20.0×10⁶/L, and were therefore considered positive. The ROC curve analysis of the XN-HPC count yielded an excellent AUC (0.97; 95% CI: 0.95–0.99). At a cut-off of 20.0×10⁶/L XN-HPC count, 259 of 273 samples (94.9%) were correctly classified. SE and SP were 98.1 and 84.8%, respectively (Table III). In the four misclassified CD34⁺ positive samples (with CD34⁺ cell count 21.3–28.0×10⁶/L), XN-HPC values ranged between 13–19×10⁶/L. In the ten misclassified CD34⁺ cells, negative samples with counts were 4–18×10⁶/L while the XN-HPC counts were 22–78×10⁶/L; seven of these samples were from MM patients.

Table III.

Results of receiver operating characteristics curve analysis

Groups	Total samples	Cut-off CD34+a	N. of positivesb	Cut-off XN-HPC	AUC (95% CI)	SE	SP	NPV	PPV
ALL	273	≥20.0	207	≥20.0c	0.97 (0.95–0.99)	98.1	84.8	93.3	95.3
Donors	97	≥20.0	94	≥30.0d	0.98 (0.94–1.01)	94.7	100	37.5	100
Lymphomas	82	≥20.0	46	≥26.0d	0.99 (0.98–1.00)	84.8	100	83.7	100
MM	83	≥20.0	64	≥62.0d	0.92 (0.85–0.99)	54.7	100	39.6	100
Other diseases	11	≥20.0	3	≥20.0d	1.00 (n.a.-n.a)	100	100	100	100
ALL without MM	190	≥20.0	143	≥30.0d	0.99 (0.99–1.000)	88.8	100	74.6	100
ALL	273	<20.0	66	<13.0e	0.97 (0.95–0.99)	100	68.2	100	90.8
ALL	273	<10.0	48	<10.0f	0.98 (0.97–1.00)	99.1	81.3	95.1	96.1

^aCut-off value of CD34⁺ cell count (×10⁶/L).

^bNumber of CD34⁺ positive samples.

^cCut-off value of XN-HPC count (×10⁶/L) with best overall agreement to predict PB CD34⁺ ≥20.0×10⁶/L.

^dCut-off of XN-HPC count (×10⁶/L) that optimally predicts PB CD34⁺ ≥20.0×10⁶/L with 100% SP and PPV.

^eCut-off of XN-HPC count (×10⁶/L) that optimally predicts PB CD34⁺ <20.0×10⁶/L.

^fCut-off of XN-HPC count (×10⁶/L) that optimally predicts PB CD34+ <10.0×10⁶/L.

ROC: receiver operating characteristics; SE: sensitivity (%); SP: specifity (%); NPV: negative predictive value (%); PPV: positive predictive value (%); ALL: acute lymphoblastic leucaemia; MM: multiple myeloma.

ROC curves were then generated in each group of samples to identify the XN-HPC cut-off associated with 100% of both SP and PPV for predicting PB CD34⁺ cells counts ≥20.0×10⁶/L (Table III). The cut-offs of the XN-HPC count was 62×10⁶/L in MM patients and ranged from 20 to 30×10⁶/L in the other groups. Using specific cut-offs, 166 of 207 (80.2%) CD34⁺ cell positive samples were correctly identified. The NPV ranged between 54.7% in MM patients up to 100% in the “Other disease” group (Table III). Using a cut-off of ≥62×10⁶/L for MM patients and of ≥30×10⁶/L for all other groups, 162 of 207 CD34⁺ cell positive samples (78.3%) were correctly identified.

The best cut-off of the XN-HPC for predicting the 66 poor mobiliser samples (<20.0×10⁶/L CD34⁺ cells) was <13.0×10⁶/L, showing 100% of both SE and NPV. At this cut-off, 45 of 66 samples (68.2%) could be accurately identified (Table III). Similarly, an XN-HPC count <10.0×10⁶/L identified 39 of 48 (81.2%) samples with <10.0×10⁶/L CD34⁺ cells, showing optimal SE (99.1%) and NPV value (95.1%) (Table III). Only 2 of 41 samples with an XN-HPC count <10.0×10⁶/L had more than 10.0×10⁶/L CD34⁺ cells (18×10⁶/L and 17×10⁶/L, respectively), and they both originated from lymphoma patients.

The kinetics of XN-HCP and CD34⁺ cell counts were very similar in almost all patients, with comparable changes over the study period. Figure 2A shows the XN-HPC and CD34⁺ cell values (mean±standard error of the mean, SEM) in 16 lymphoma, 1 Ewing sarcoma and 1 neuroblastoma patients. Mean values and changes over time of XN-HPC and CD34⁺ cell counts were overall highly comparable. Figure 2B shows a similar analysis in 13 MM patients. As predicted, mean XN-HPC values were significantly higher than CD34⁺ cell count, although changes were comparable over a period of days.

Figure 2.

CD34-positive (⁺) and kinetics of the Sysmex^® XN haematologic XN-HPC analyser showing CD34⁺ (white squares) and XN-HPC (black circles) cell count performed on successive days starting initiation of mobilisation.

Counts are expressed as means ± 1 Standard Error of Mean. Kinetics are shown for (A) 15 lymphomas, 1 Ewing sarcoma, 1 neuroblastoma patient, and (B) 13 multiple myeloma patients.

DISCUSSION

The timing of apheresis is a critical issue for the safe and efficient collection of sufficient PBSCs for transplantation. Many variables can compromise the collection of PBSCs, including age, underlying pathologies, previous chemotherapy, and mobilisation regimen^29–34. It is now recognised that the number of CD34⁺ cells in PB measured the morning of collection remains the best predictor of adequate PBSC yield. Although flow cytometry, using the ISHAGE protocol, is the “gold standard” for quantifying CD34⁺ cells, this method requires approximately 1 h, is expensive, and must be performed by experienced operators. Moreover, a single analytical platform is commercially available, and CD34⁺ cell count is therefore frequently plagued by a long turnaround time when multiple samples need to be analysed, thus complicating patient management and making organisation difficult, prolonged bed occupancy and disruption of timetable of the clinical staff. For these reasons, initiation of apheresis procedure is occasionally postponed, which in turn leads to processing of the collected PBSC product being delayed.

All studies evaluating other surrogate markers, including WBC count^9–11 or cell morphology parameters from the Unicel^® DxH800 (Beckman Coulter, Brea, CA, USA) and Advia^® 2120i (Siemens Healthcare Diagnostics, Deerfield, IL, USA)³⁵, proved unable to predict accurate timing of apheresis collection due to their poor correlation with CD34⁺ cell counts. Other studies using HPC enumeration with the Sysmex^® SE-9500 and XE-2100 analysers produced better data, but the correlation with CD34⁺ cells in PB remained modest^{18,20,24,35,36} and was characterised by inconclusive HPC counts over a broad range of values. This meant these had to be confirmed by flow cytometry analysis in a large proportion of cases^18,20,24,37.

The XN stem cell mode of the Sysmex^® XN analyser is the most recent application for HPC counting. The optical detection system, combined with staining of intracellular nucleic acids after permeabilisation of cell membranes using a surfactant agent, provides a more reliable detection of HPC cells compared to previous Sysmex^® SE/XE systems^25,26. Furthermore, the number of cells analysed has increased by 4-fold; a remarkable improvement in the accuracy of HPC counts at lower values. All studies published to date have reported improved correlation of XN-HPC with CD34⁺ cells quantified by flow cytometry^25–28, both in PB, during apheresis collection, and in apheresis product^27,28.

In this study, we evaluated the XN-HPC count in 273 mobilised blood samples collected from 78 donors and 72 patients who underwent autologous or allogenic PBCS transplantation. Our study focused on the possibility of using XN-HPC count to optimise the timing of PBSC harvest, thus safely replacing some expensive CD34⁺ cell counts performed daily in PB samples.

We first assessed the imprecision of the XN-HPC count, obtaining CV values <7.0% in the daily QC on control materials and a functional sensitivity of 8.9×10⁶/L. The correlation between XN-HPC and CD34⁺ cells was very good. More specifically, the correlation in the total 273 samples tested was 0.93; similar to that earlier reported by Grommé et al. (0.92)²⁸ and by Peerschke et al. (0.88)²⁷. In samples from lymphoma patients, solid tumours and donors, the correlations were 0.976, 0.975 and 0.849, respectively; this was significantly better than results previously published using Sysmex^® SE/XE analysers reporting values between 0.44 and 0.78^{18,20,24,35,36}.

We observed significant differences between XN-HPC and CD34⁺ cell counts in samples collected from MM patients. Despite a good correlation (r=0.89), the median value of XN-HPC count was 1.6-fold higher than the CD34⁺ cells. Results of this kind were not reported in studies published by Peerschke and Grommé, although both had previously assessed a significant number of MM patients (approximately 43 and 45% of the total, respectively). Indeed, both authors reported correlation data between XN-HPC and CD34⁺ cell counts only in the whole set of PB samples, and this is probably why these authors failed to find the significant differences in MM patients seen in our study. However, other studies using the Sysmex^® SE/XE confirmed our findings in those samples collected from MM patients^24,36. Although the specific reasons underlying these differences are still not fully understood, one possible cause may be the presence of the so-called MM stem cells, or myeloma-initiating cells (MIC), which exhibit tumour-initiating potential, self-renewal, and resistance to chemotherapy^37–39. These cells, or other CD34⁻ cells mobilised after administration of G-CSF or plerixafor, could not be efficiently separated from CD34⁺ cells during HPC analysis, while also blood cell precursors (including some CD34⁻ cells)^40–44 are detected by Sysmex^® analysers in the same area in which HPC are enumerated¹⁶. However, besides the differences observed in cell enumeration, the kinetics of XN-HPC and CD34⁺ cells in the 13 MM patients was comparable to that encountered in all other patients, with HPC counts changing over time in parallel with the CD34⁺ cells count.

ROC curve analysis showed excellent diagnostic performance of XN-HPC ≥20×10⁶/L for predicting timing of apheresis. At this cut-off (i.e., that used in our institution for starting apheresis), the AUC of XN-HPC count was excellent (0.97; 95% CI: 0.95–0.99) with 259 of 273 PB samples correctly classified, thus displaying a significantly better diagnostic accuracy than that reported in previous studies using Sysmex^® XE analysers^18,20,24,36.

In order to optimise the clinical usefulness of the XN-HPC count, for each group we studied we identified the XN-HPC cut-off values capable of efficiently predicting (i.e., with 100% of both SP and PPV) a number of PB CD34⁺ cells ≥20.0×10⁶/L. By using a cut-off of 62×10⁶/L for MM patients and 30×10⁶/L for all other groups, 78.2% of CD34⁺ positive samples (i.e. 162 of 207) were correctly identified. This leads us to conclude that the XN-HPC count is an excellent rule-in test for assessing when the healthy donor or the patient is adequately mobilised, thus avoiding having to perform a specific CD34⁺ cell count.

The efficiency of a safe identification of whether donors or patients are not adequately mobilised is equally important. An XN-HPC count <13.0×10⁶/L could identify 45 of 66 (68.2%) samples with <20.0×10⁶/L CD34⁺ cells, with a 100% value for both SE and NPV. When the CD34⁺ cell cut-off was lowered to <10.0×10⁶/L, an XN-HPC count <10.0×10⁶/L correctly predicted 39 of 48 (81.2%) poor mobiliser samples, maintaining remarkable values of both SE (99.1%) and NPV (95.1%). Only two of the 41 samples with XN-HPC count <10.0×10⁶/L had a CD34⁺ cells count ≥10.0×10⁶/L, but very near to the cut-off level. Therefore, XN-HPC count may also be considered an excellent rule-out test for apheresis initiation, and, even in this case, CD34⁺ cell analysis can be safely avoided.

CONCLUSION

Our study confirms that XN-HPC count is strongly correlated to CD34⁺ cell count, and could be a useful surrogate test to assess optimal timing for PBSC collection. The Sysmex^® XN-Stem Cell mode has several advantages compared to CD34 flow cytometry: it is fast, simple, less expensive, available 24/7 as part of a full blood count analysis, well standardised, and does not require experienced operators. Furthermore, the XN analyser also provides additional useful parameters such as immature platelet fraction, nucleated red blood cells, immature granulocyte count or immature reticulocyte fraction, which can be used to improve patient monitoring, even in the pre- and post-apheresis phases^45–49.

The XN-Stem Cell mode can substantially improve the apheresis workflow by promoting rapid decision-making for scheduling PBSC harvest by replacing a significant number of CD34 cell counts performed on PB samples. Based on our data, its efficiency is definitely satisfactory, since this analysis may predict timing of apheresis collection with 100% SP and 100% PPV. It can also efficiently detect samples with CD34⁺ cell counts both <10.0×10⁶/L and <20.0×10⁶/L.

Further studies will be needed to validate our XN-HPC cut-off values to optimise apheresis timing, based on a larger number of donors and patients with cancer, undergoing chemotherapy, or with other conditions which are known to influence PBSC mobilisation. However, it is very unlikely that the XN-HPC count will completely replace flow cytometry assessment of CD34⁺ cells, since there are some differences in the cell populations detected by these two tests.