Surveillance of Myelodysplastic Syndrome via Migration Analyses of Blood Neutrophils: A Potential Prognostic Tool

Autonomous migration is a fundamental ability of all immune cells. In the case of neutrophil granulocytes, referred to as neutrophils, migration is essential for central immunological mechanisms, such as mobilization (1, 2), homing (3–5), phagocytosis (6, 7), and tissue repair (8). Altered migration patterns of immune cells are among the most commonly observed cellular phenomena and are associated with acute and chronic diseases (9, 10). For example, reduced migration of neutrophils points to a severe disease course in sepsis (11, 12), whereas neutrophils in chronic inflammatory bowel diseases display accelerated migration (13).

These and additional studies carved out two main reasons why the determination of neutrophil migration patterns has the potential to be a novel prognostic biomarker. First, neutrophils are among the leading immune cells reacting initially to infections (14). Their migration is directed toward secreted pathogenic peptides at sites of pathogen intrusion, which, for example, involve receptors for formyl-methionyl-leucyl-phenylalanine (fMLP) or CXCL cytokines (15, 16). Among the latter group, the most potent ones are CXCR1 and CXCR2, the receptors for CXCL1 and CXCL8 (17). Collectively, enhanced neutrophil migration is a telltale signal for ongoing inflammation. Second, mature neutrophils have only a limited lifespan of several hours in peripheral blood (15) and are quickly replenished from bone marrow (1). Because of this high turnover, developmental defects of neutrophils manifest themselves rather quickly in the peripheral pool and, hence, functional deficiencies like altered migration might indicate hematopoietic diseases. In contrast, recovery to a normal migration pattern could indicate therapy success.

Examples of such conditions are myelodysplastic syndromes (MDS), a heterogeneous group of clonal hematopoietic stem cell disorders leading to hematopoietic insufficiency with an increased tendency for leukemic transformation (18). Various genetic alterations are detected in MDS, which transform the disease into acute myeloid leukemia, causing lethal collapse of hematopoiesis and immune deficiency. Additionally, patients suffering from MDS are more susceptible to infections because of impaired neutrophil degranulation, phagocytosis, and oxidative burst (19–21). Besides progression to acute leukemia, infections are indeed the most frequent causes of deaths, particularly in high-risk MDS cases (22). Recently, MAPK p38 emerged as a target of MDS therapy. Various reports suggest that p38 dysregulation drives alterations of immune responses and inflammatory signals during MDS progression (23–26). Thus, p38 inhibitors like ARRY-614, a p38/Tie2 dual inhibitor, are in clinical trials (27).

It is still challenging in clinical practice to attain a valid prognosis of the individual patient. Currently, the prognosis of MDS is calculated based on blood cell counts, medullary blast percentages, and cytogenetic results via the Revised International Prognostic Scoring System (IPSS-R) (28). However, determining these parameters is elaborate, requiring bone marrow aspiration. Hence, a fast surveillance, or pretest, prior to more invasive analytical procedures would facilitate clinical work. In light of the known neutrophil defects in MDS, we hypothesized that the degree of integrity of neutrophil migration might provide such a prognostic parameter for MDS. Two different principles to obtain clinical migration data are conceivable: 1) the counting of cells, which pass a certain distance during a given period of time (end point determination) or 2) the continuous observation of migrating cells using time-lapse video microscopy (2, 29, 30). End point analyses can be done in Transwell chambers or the end reservoirs of microcapillary systems (30, 31). Time-lapse observations are executed in transparent multiwell plates or sections of microcapillary systems and require single cell tracking (2, 32). Because live cell imaging provides more quantifiable parameters, such as speed, directionality, and the number of moving cells, it is a more versatile choice for clinical routine approaches.

Analytical migration assays are not in clinical use for several reasons: 1) cell culture supplements, such as serum, collagen, or fibrin are impossible to standardize because of batch-dependent differences. 2) Until recently, the only possibility to determine migration patterns in time-lapse videos was manual cell tracking because the available solutions were not sufficiently reliable. However, manual cell tracking is extremely time-consuming and prone to bias. Taken together, because of the lack of fast and reproducible assays, immune cell migration analyses have not been applied for the benefit of patients.

To overcome these problems, we introduce in this study a standardized two-dimensional migration assay for neutrophils purified from peripheral blood. With this approach, we generated a database of mean neutrophil migration patterns for 111 representative individuals of the average German population. This database was used to calculate the average mean migration values. In addition, we determined migration patterns of neutrophils from patients suffering from various MDS subtypes. The responsiveness to fMLP negatively correlated with the disease category calculated via the IPSS-R. In fact, neutrophils of high-risk MDS patients showed the weakest response to fMLP, whereas patients with more beneficial IPSS-R were more responsive to fMLP. We followed the disease course of two patients suffering from MDS with excess blasts (EB) who displayed almost no neutrophil motility prior to therapy. Neutrophil migration recovered to normal levels during a successful 5-azacytidine treatment, whereas neutrophils of an unsuccessfully treated patient did not show improvement of motility during therapy. Thus, the assay may represent a suitable surveillance tool to indicate therapy success or failure. In both cases, p38 phosphorylation correlated with the incapability of the neutrophils to migrate. Collectively, this work suggests that routine analysis of neutrophil migration bears a potential in MDS surveillance. The identified defect in migration possibly represents an additional measure for the determination of MDS prognosis. Although not intended to replace the current diagnostic setup, a routine application of the introduced migration assay would assist in the surveillance of MDS during disease progression and therapy because of its minimally invasive, simple, and robust procedure.

Samples of blood in EDTA-supplemented tubes from participants of the Heinz Nixdorf Recall MultiGeneration (HNRM) study were provided by the Institute for Medical Informatics, Biometry and Epidemiology, University Hospital Essen (Essen, Germany). The main objective of the HNRM is to extend the Heinz Nixdorf Recall Study (HNRS) by recruiting children and spouses/partners of the original HNRS participants to explore whether indicators of atherosclerosis in the “first generation” participants are associated with respective indicators in their family members (children and partners). The objectives and the study design of the HNRS have been previously published (33). Approved by the relevant institutional ethics committees, the studies follow strict internal and external quality assurance protocols. Written informed consent was obtained from all participants. A total number of 111 blood samples were analyzed from participants of the HNRM recruited during a 9-mo period since February 2016 (18–90 y of age, mean 53.2; 61.9% women). For 99 samples, blood draw and analysis times were documented. Ninety-four measurements did not exceed storage times of more than 6 h. As flow cytometric analyses were not performed from the beginning, valid cytometry results exist for the last 50 samples. The ages of 77 HNRM volunteers were reported. A second cohort of 24 healthy volunteers was recruited at Ghent University Hospital (Ghent, Belgium) during a 3-mo period starting June 2017 (66.7% women). Blood samples were similarly obtained in EDTA tubes and processed. For the transports, a VACUETTE transport container (Greiner Bio-One, Kremsmünster, Austria), according to the UN 3373 regulation, was used. Blood samples from MDS patients (n = 26) treated at the Düsseldorf University Hospital (Düsseldorf, Germany) and participating in the Düsseldorf MDS-Registry from June 2017 to January 2018 were collected in EDTA-supplemented tubes and brought to the laboratory for analysis in time via a transport service. HNRM and MDS samples were measured on the same microscope.

For the entire HNRM cohort of individuals, neutrophils were isolated from 3 ml EDTA-supplemented blood via density centrifugation using Polymorphprep (Axis-Shield Diagnostics, Oslo, Norway). Polymorphprep was overlaid with blood in a 15-ml tube at a 1:1 ratio and centrifuged at 450 relative centrifugal force for 30 min without a break. Serum and PBMCs were discarded, and polymorphonuclear cells were collected and washed with 10 ml sterile PBS (Biochrom, Berlin, Germany). Erythrocytes were removed via 10 min incubation at room temperature in lysis buffer (155 mM NH₄Cl, 10 mM KHCO₃, 0.1 mM EDTA in distilled H₂O). Afterwards, purified neutrophils were washed with 10 ml sterile PBS, resuspended in sterile hematopoietic progenitor growth medium (HPGM; Lonza, Basel, Switzerland), and counted using a Cellometer Auto T4 (Nexcelom Bioscience, Lawrence, MA).

Because Polymorphprep isolation is error prone for various reasons, neutrophils from the volunteers at Ghent University, from an additional 14 volunteers in Essen, and from the MDS patients were isolated via magnetic negative isolation with the MACSxpress Whole Blood Neutrophil Isolation Kit (Miltenyi Biotec, Bergisch Gladbach, Germany) to ensure a comparable procedure. For magnetic depletion of erythrocytes, the MACSxpress Erythrocyte Depletion Kit (Miltenyi Biotec) was used according to the manufacturer’s instructions. Afterwards, purified neutrophils were washed with 10 ml sterile PBS, resuspended in sterile HPGM, and counted using a Cellometer Auto T4. Of note, both isolation techniques did not activate the neutrophils (Supplemental Fig. 1A). Hence, the isolation of comparable populations via Polymorphprep and magnetic negative isolation was ensured. Furthermore, the outcome of the migration assay was not affected by the purification method of the neutrophils (Supplemental Fig. 2A).

Neutrophils were seeded in a hydrophobic μ-Plate 96 Well (ibidi, Martinsried, Germany) at a density of 8250 cells per well in 198 μl sterile HPGM supplemented with sterile Serum Replacement 3 (final concentration 0.3×; Sigma-Aldrich, Munich, Germany). Cells were stimulated at the indicated concentrations with either 2 μl fMLP (Sigma-Aldrich; Abcam, Cambridge, U.K.; Tocris Bioscience, Bristol, U.K.) or 2 μl human recombinant CXCL1 (R&D Systems, Minneapolis, MN; PeproTech, Hamburg, Germany; Miltenyi Biotec) or 2 μl human recombinant CXCL8 (R&D Systems; PeproTech; Miltenyi Biotec). Because fMLP, CXCL1, and CXCL8 were reconstituted in PBS, the addition of 2 μl of sterile PBS served as control. The plates were centrifuged at 50 relative centrifugal force for 3 min and incubated at 37°C, 5% CO₂ for 20 min prior to microscopy.

For samples from Germany (HNRM and MDS patients), microscopy plates were imaged in a Leica DMI6000 B (Leica Microsystems, Wetzlar, Germany) with a motorized stage, 20× magnification, and rate of one frame/8 s for 1 h at 37°C, without CO₂. Samples from the Belgian cohort were imaged in an identical fashion on a CellM Live Cell Imaging system comprising an IX81 microscope (Olympus, Berchem, Belgium) with motorized stage using 16× magnification. Generated Multi-TIFF files were converted to *.mov files. With these files, automated segmentation was performed using the Automated Cellular Analysis System (MetaVi-Harmony software, MetaVi Labs, Austin, TX). Parameters presented in this manuscript are speed including and excluding nonmoving cells. Recruitment represents the difference of the percentage of moving cells upon stimulation and the PBS control.

One hundred thousand cells were washed with PBS and stained with the following Abs: CD15 VioBlue (dilution: 1:100, clone: VIMC6; Miltenyi Biotec), CD16 FITC (dilution: 1:100, clone: REA423; Miltenyi Biotec), fMLPR1 Alexa Fluor 647 (final dilution: 1:100, clone: 5F1; BD Biosciences, San Jose, CA), CXCR1 PE (dilution: 1:100, clone: 8F1; Miltenyi Biotec), CD11b-PE (dilution: 1:50, clone: REA713; Miltenyi Biotec), CD66b-FITC (dilution: 1:10, clone: REA306; Miltenyi Biotec), CD62L-PE-Vio770 (dilution: 1:100, clone:145/15; Miltenyi Biotec), and CXCR2 PE-Vio770 (dilution: 1:20, clone: REA208; Miltenyi Biotec). After 15 min in the dark at 4°C, the suspensions were diluted 1:1 in PBS and analyzed using a MACSQuant VYB (Miltenyi Biotec).

Cell lysis was done in TPNE buffer (300 mM NaCl, 1 mM EDTA, 1% Triton X-100 in 1× PBS) and SDS–PAGE, and Western blot was performed as described previously (34–36). Primary Abs included p38 MAPK (D13E1) (no. 8690; Cell Signaling Technology) and phospho-p38 MAPK (Thr180/Tyr182) (D3F9; Cell Signaling Technology), and the secondary Ab was anti-rabbit IgG, HRP-linked Ab (7074; Cell Signaling Technology).

Experiments involving human material were performed with approval of the local ethical committees. Numbers of ethical approvals/registers are as follows: 15-6686-BO (HNRM samples, Essen, Germany), 3768 (MDS samples, Düsseldorf, Germany), UZG 2017/0214 (Ghent samples, Ghent, Belgium).

Statistical analyses were performed via GraphPad Prism (GraphPad Software, San Diego, CA) and IBM SPSS (IBM SPSS Statistics 24, Armonk, NY).

Because live cell imaging provides the most quantifiable parameters compared with other migration analyses, we developed a two-dimensional imaging assay to determine cell migration. This avoids the use of three-dimensional hydrogels, a possible source of heterogeneity. For the two-dimensional assay, however, the neutrophils needed to adhere on a transparent plastic surface. Comparing different plastic dishes (data not shown), we identified hydrophobic polycarbonate plastic as the ideal surface. To ensure identical conditions, neutrophils were only cultured in defined precomposed media supplemented with human serum replacement. To provide quick and reliable data, an autotracking tool (Harmony) was used. The software tracks individual cells from time-lapse videos and generates trajectory plots (Fig. 1A, Supplemental Video 1) in addition to calculating further migration parameters, such as speed (micrometers per minute) including and excluding nonmoving cells, moving cells (percentage), and others. For all future experiments, the optimal concentrations of the respective stimuli were determined via titration (Fig. 1B). Increasing CXCL1 or CXCL8 concentrations enhanced the speed dose-dependently to a certain plateau level. At 100 ng/ml CXCL8, we identified a strong statistical difference to the PBS control, which did not improve even at higher concentrations. The same concentration was chosen for CXCL1 because literature points to the triggering of neutrophils at this concentration (15, 16), although only mild but statistically insignificant differences were found. Stimulation with 10 nM fMLP resulted in the strongest difference compared with the PBS control, and at higher or lower concentrations, migration dropped (Fig. 1B). Previous reports indicate that MAPK p38 plays a central role in the regulation of neutrophil migration (37). Interestingly, the phosphorylation of p38 negatively correlated with cell speed upon fMLP stimulation in our assay, and the cultivation conditions themselves apparently induced strong p-p38 compared with freshly isolated neutrophils (Fig. 1C). Hence, optimal neutrophil migration upon fMLP triggering was associated with a low p-p38 level in our assay. Thus, in all subsequent experiments, we used 10 nM fMLP and 100 ng/ml CXCL1 or CXCL8. The migration result accorded with CD62L shedding and CD11b and CD66b upregulation (Supplemental Fig. 1B). Because reagents from various suppliers may vary in their effectiveness because of different production processes, neutrophil migration triggered by three different suppliers for fMLP, CXCL1, or CXCL8 was analyzed. Interestingly, we did not detect significant differences between the manufacturers (Fig. 1D). To find out whether neutrophil migration is a reproducible parameter in individuals, we measured neutrophil migration patterns three times, independently in three different volunteers, and then calculated the minimal and maximal fold differences between tests (Fig. 1E). Of note, although individuals differed between each other, the fold differences of the repeated individual measurements never exceeded 1.14 or fell below 0.76, putting the assay in the range of reproducible blood count parameters. The following can be used for comparison: According to the adapted criteria from the International Society for Laboratory Hematology, the WBC count can range between 4 × 10³/μl and 30 × 10³/μl (7.5-fold difference) before further analyses in terms of leukopenia or leukocytosis are considered. Neutrophil frequencies can range between 1 × 10³/μl and 20 × 10³/μl (20-fold difference) (38).

Via the HNRM study, we determined neutrophil migration patterns of 111 individuals of a representative cross section of the German population to generate, to our knowledge, the first comprehensive database of neutrophil 2D migration patterns. The distributions passed normality tests (Fig. 2A). The steady-state speed of nonstimulated neutrophils averaged 9.03 ± 0.35 μm/min and was increased up to 15.47 ± 0.31 μm/min by fMLP, the most potent inducer of migration (Fig. 2A).

Assaying migration dynamics using classical experimental setups is notorious for grave variations among laboratories (39, 40). To this end, neutrophils from HNRM volunteers were purified via density gradient centrifugation. However, pipetting of the neutrophil phase correctly critically depends on the experience and skills of the experimenter. As the transferability of the assay to other laboratories had to be ensured, negative magnetic isolation of neutrophils was implemented, eliminating the user- dependent variability of density gradient centrifugation of cells as a major source of variance between laboratories. To test the robustness of our approach, the migration assay was performed using magnetic neutrophil isolation in laboratories in Germany (Essen) and Belgium (Ghent) with neutrophils donated from local volunteers using local instrumentation. The analyses of the generated movies were centralized on the same tracking system. Neutrophil migration from donors in Germany or Belgium showed the same tendencies, as the chosen concentrations of fMLP induced the highest speed, followed by CXCL8, CXCL1, and PBS (Fig. 2B). The distributions of mean speeds of control-treated neutrophils in the Belgian cohort was, however, slightly but significantly higher. The different speeds upon fMLP, CXCL8, and CXCL1 triggering was not statistically significant between the Belgian and German cohorts, supporting the comparability of the assay results between different laboratories.

At this point, it was unclear whether neutrophil migration was affected by circadian rhythms or age, as described for mice (41). If true, they could influence migration, which would compromise a clinical application. To test their influence, the time of day the blood was drained from HNRM volunteers and their age was correlated to the speed of neutrophils. However, via Pearson tests, we could not detect a correlation of age or draining time to the speed of neutrophils (Fig. 2C).

As neutrophils display a short lifespan and high turnover rate (15, 42), their survival and metabolic activity could be affected upon storage in the EDTA-supplemented tubes used for blood draw. Several HNRM blood samples were not processed immediately but were stored at room temperature for later analyses, delaying the sample preparation from 2 to 6 h. We analyzed the impact of this preparation time on migration (Fig. 2D), and within 6 h we detected no correlation with neutrophil migration via Pearson tests. To corroborate this finding on an individual level, blood was drained in the morning from three volunteers, and neutrophils were isolated after 2, 4, and 6 h of storage, and neutrophil migration patterns were subsequently determined. Again, we detected no correlation of the storage time with neutrophil migration via Spearman tests (Fig. 2E). Taken together, the assay tolerated blood storage times of a maximum of 6 h to yield results equal to directly analyzed samples.

Neutrophils respond with directed movement to a chemotactic gradient by sensing concentration changes along their cell body (43). It would be conceivable that at homogeneously distributed receptor ligands, the cell migration would correlate with the amount of triggered receptors. If this was the case, the expression levels of receptors for fMLP, CXCL1, or CXCL8 would correlate with migration, thereby obviating the need to measure it. To test this hypothesis, we determined the expression of fMLPR, CXCR1, and CXCR2 (Fig. 3A). The mean fluorescence intensities (mfi) of the analyzed receptors remained independent of draining time and age of the HNRM volunteers (Fig. 3B). Analogous to the speed analyses, the mfi were independent of the time for sample preparation, both for the volunteers of the HNRM (Fig. 3C) and on the level of three individuals (Fig. 3D). To test a potential correlation of receptor expression and migratory capacity, we plotted the mfi of the receptors relative to the migration observed in response to their ligands (Fig. 3E). Interestingly, the expression did not correlate with the responsiveness; thus, flow cytometry–based receptor analysis alone cannot substitute migration analyses, suggesting a pivotal role of intracellular signal processing for cell motility.

MDS is a very heterogeneous condition, presenting with frequently observed defects in neutrophil function (18). We chose a cohort of MDS patients from the Düsseldorf MDS-Registry (Düsseldorf, Germany) to determine neutrophil migration patterns and to correlate the results of our assay to the current prognosis scoring in a double-blinded fashion. Because the samples were not taken at the initial diagnosis, all patients received various therapies, such as 5-azacytidine, α-phthalimido glutarimide, or erythropoietin at the time of neutrophil migration analyses. Thus, we were only able to retrospectively analyze potential correlations.

In this analysis, we indeed found a clear drop in neutrophil speed correlating with the MDS subtype. Thereby, the most severe forms of MDS, MDS-EB-I and MDS-EB-II, exhibited statistically significant impaired migration patterns (Fig. 4A, 4B) compared with the less severe forms MDS–multilineage dysplasia (MLD)–ring sideroblasts (RS), MDS-MLD, MDS–single-lineage dysplasia (SLD)–RS, and MDS-SLD. We examined the power of neutrophil migration analyses for MDS scoring by correlating of current prognostic gold standard, the IPSS-R, and the corresponding disease categories to neutrophil migration speed (Fig. 4C). Pearson correlation analyses revealed a strong relationship of migration and the disease prognosis (Fig. 4C), revealing the poorest migration with the most severe prognosis scores. In addition, we also correlated the speed to the routinely determined blood count parameters to determine whether, for example, anemia, lymphopenia, or C-reactive protein (CRP) levels as indicators for infection had an impact on neutrophil migration. However, except for the absolute neutrophil count (ANC), none of these parameters displayed a correlation with the neutrophil speed (Fig. 4D). Similarly, neutrophil purity and quality among the different patients was controlled via flow cytometric analyses of CD15 and CD16 (Supplemental Fig. 2B).

fMLP-stimulated neutrophils displayed the most prominent distribution between MDS-EB-I and -II patients and the controls (Fig. 4A, 4B). Interestingly, the percentage of moving cells was elevated in the less severe forms of MDS compared with the MDS-EB-I and -II cases (Fig. 5A). Neither CXCL1, CXCL8, nor the PBS control could serve in the same fashion as a readout out to identify the most severe cases (Fig. 5A), pointing to fMLP triggering alone as the best stimulus for prognosis.

In healthy donors, we had observed the stability of migration patterns over time (Fig. 1E). To investigate, whether this assay could, hence, be used to measure therapy success in MDS, we determined migration patterns in two MDS-EB-II cases (Fig. 5B, 5C), both treated with 5-azacytidine. Patient number 21 did not show improvement of the clinical condition, and the disease turned into MDS–acute myeloid leukemia in its terminal stage. During the entire monitoring time, a low migration efficiency of neutrophils was maintained with negligible differences between the measurements. In contrast, another MDS-EB-II case, patient number 8, was identically treated but displayed a recovery of neutrophil migration to a normal pattern and improvement of the overall clinical condition during therapy (Fig. 5B, 5C).

Interestingly, we detected increased p-p38 levels in freshly isolated neutrophils of both MDS-EB-II patients compared with a healthy control (Fig. 5D), pointing to a preactivation of p38 in the course of MDS, which is corroborated by literature addressing the role of p38 in MDS (26). Because we detected a correlation between the highest migration speed and low p-p38 levels in healthy individuals (Fig. 1C), we aimed to compare neutrophil migration of both MDS-EB-II patients with the p-p38 status. As neutrophils for Western blot analysis were not triggered, the migration of untreated neutrophils (PBS) was taken for comparison. The successfully treated case number 8 displayed a certain increase of p-p38 levels, but neutrophil speed was only minimally and statistically not significantly reduced compared with the controls. In contrast, the unsuccessfully treated case number 21 revealed a much stronger p-p38 signal and a statistically significant difference in neutrophil speed. Both results underline the correlation of p-p38 with neutrophil motility and suggest a functional relationship of random neutrophil migration with the phosphorylation status of p38.

Migration patterns of leukocytes in health and disease have been investigated for decades. The compelling evidence demonstrates a promising clinical potential for migration analyses (44–46). However, utterly not standardized and time-consuming assay conditions made a clinical application practically illusive. At best, each laboratory generated its own set of assays with standard values serving as an internal database. But those values were hardly transferable between different studies, much less among different research teams.

With the two-dimensional neutrophil migration assay developed and validated in this study, which requires only a minimum number of reagents and is serum-free, thereby minimizing supplement variations, we generated a comprehensive database of neutrophil migration patterns of 111 participants of the HNRM study. This high number of individuals strengthens the quality of the generated value. The application of the protocol to a cohort in Belgium provided comparable tendencies. However, statistical differences in PBS-incubated neutrophils between the Ghent cohort and the HNRM were found but not upon fMLP, CXCL8, and CXCL1 stimulation. These mild variances point to differences among the local German and Belgian populations. Alternatively, small differences in handling of the reagents and instrumentation might contribute to statistically significant differences.

Because the assay provides results within hours after blood draw, it is applicable as a useful surveillance tool. Because of this low time requirement, critical clinical interventions might still be possible if the migration pattern indicates an immediate necessity. One condition fitting into this concept is septic shock because neutrophil migration is heavily altered hours before clinical signs (11). Thus, monitoring neutrophil migration patterns in patients with increased sepsis risks, for example burn victims, would lead to a clinical intervention on time. This idea focuses on the interpretation of migration patterns in acute conditions. The present study demonstrated its usefulness in MDS, a group of heterogeneous neoplasias gaining various genetic alterations due to genetic instability, whose causes of initiation are not completely understood (47, 48). The incidence is four to five out of 100,000 people per year and >30 out of 100,000 are over the age of 70 (49). The disease is staged by blood cell counts, bone marrow blast percentages, and cytogenetic results, leading to the calculation of the IPSS-R as a prognostic parameter (28). We showed that neutrophil migration correlated with the IPSS-R of MDS patients despite heterogeneous pathophysiological causes. It is likely that other nonmalignant diseases that depend on proper neutrophil function, such as sepsis or sterile inflammation, modulate neutrophil motility. Therefore, for any prognostic approach, the impact of possible interfering factors needs to be determined to evaluate the usefulness of the assay, for example at times of infections. In this study, it remained unclear to what extent neutrophil migration is affected by comorbidities, therapies, or infections. Of note, we found that only the ANC showed a correlation with the neutrophil migration patterns but not the CRP level, which is a classical marker for inflammation. Taken together, our results suggest that a more comprehensive study involving a larger group of MDS patients is required to fully answer the question regarding specificity and stability of the obtained migration patterns.

At the present state, our data suggest that the determination of neutrophil migration patterns can assist with the verification of the IPSS-R. Therefore, it could be used as a surveillance tool during the clinical course of the disease by preserving costs and reducing redundant invasive procedures, such as repetitive bone marrow analyses, cytogenetic analyses, and mutation screening by microarray or deep sequencing. Whether neutrophil migration pattern analyses provide a rational benefit in the choice of the treatment of MDS patients requires clinical studies on a greater scale and duration than was possible in the current study. It will have to involve repetitive measurements of the patients from initial diagnosis over longer periods of time. It is especially of interest to address how neutrophil migration changes during the disease progression under different therapies and how the assay results differ compared with other neoplastic diseases. The repetitive measurement of two MDS-EB-II cases in this study already points to the applicability of the assay as a therapy surveillance tool because the migration patterns did and did not improve upon therapy success or failure, respectively. A strong direct impact of 5-azacytidine on neutrophil migration is unlikely because of the opposing outcomes of the therapy and neutrophil migration over time. However, future studies need to include more individuals to correlate beneficial or disadvantageous outcomes of therapy and to determine the impact of possible interfering conditions. Taken together, a closer approximation of the correlation between neutrophil migration patterns, the incidence of infections, and the IPSS-R would lead to a fine-tuning of the predictive power of the assay.

Remarkably, the highest p38 phosphorylation was detected in the PBS control, which migrated the slowest. A causal connection between high p-p38 and reduced migration has not been formally demonstrated by our analysis and, interestingly, the available literature points to an opposing function of p38: the requirement of phosphorylated p38 for migration (50). A closer look at these studies reveals, however, that this role of p38 was determined via chemotactic (directed) assays, only (37, 51, 52). But chemotactic migration requires the reorientation of the cytoskeleton in relation to the source of the gradient. Strikingly, accumulation of p-p38 in the plasma membrane directed toward fMLP was reported (37). Hence, it is conceivable that p38 accumulates to a specific side of the cell in directed migration assays in general, leading to an asymmetric accumulation of p38 substrates, which drive migration in a certain direction. In random migration assays, these substrates are equally distributed on the subcellular level. Accidental protein accumulations determine the migration vector individually on the single cell level. Hence, the correlation of low p38 phosphorylation with the strongest migration could be specific for the developed assay or for random neutrophil migration in principle, suggesting a comparison of the role of p38 in random and directed migration in future studies. To this end, it remains unclear how p38 activation is triggered in the PBS control and which other signaling proteins may be involved. It is conceivable that adhesion proteins contribute to this activation, as cells adhered in the control as well. In this line, p38 might also control the strength of cell adhesion. A point of origin for future mechanistic studies is the higher level of p-p38 in the steady state of both MDS-EB-II patients. Because strong p-p38 signals correlated with the slowest neutrophil migration patterns, constant high p-p38 levels might inhibit neutrophil motility. However, because complex cellular outcomes like migration require a number of active kinases (53) (and not just p38), it is possible that other factors affected in MDS are the leading cause of impaired neutrophil migration.

Another interesting finding was the absence of a correlation of receptor expression and migration, suggesting that intracellular proteins play a more pivotal role in controlling the strength of neutrophil migration. In contrast, we observed that intermediate concentrations of fMLP triggered the strongest cell migration, whereas low and high concentrations led to reduced migration. Hence, the fMLP receptor can sense different concentrations of its ligand. A more detailed analysis of fMLP receptor thresholding might reveal insights into the way the receptor senses low or high doses of its ligand.

In summary, leukocyte migration assays bear the potential to be novel prognostic tools that test a central function of immune cells. Because of decades of research in leukocyte migration and, thus, heavily substantiated knowledge of associations between migration and multiple diseases, a huge foundation of experience and experimental data already exists. Thus, only standardized assays, eventually in combination with novel high-throughput devices are required to bring migration analyses to the next level as a clinical application. The assay we developed constitutes a good starting point and warrants further steps toward this important goal.

We thank all participating and supporting persons of the HNRM study, particularly the dedicated personnel at the Heinz Nixdorf study center and the investigative group.

The authors have no financial conflicts of interest.