Toward harmonized phenotyping of human myeloid-derived suppressor cells by flow cytometry: results from an interim study

Abstract

There is an increasing interest for monitoring circulating myeloid-derived suppressor cells (MDSCs) in cancer patients, but there are also divergences in their phenotypic definition. To overcome this obstacle, the Cancer Immunoguiding Program under the umbrella of the Association of Cancer Immunotherapy is coordinating a proficiency panel program that aims at harmonizing MDSC phenotyping. After a consultation period, a two-stage approach was designed to harmonize MDSC phenotype. In the first step, an international consortium of 23 laboratories immunophenotyped 10 putative MDSC subsets on pretested, peripheral blood mononuclear cells of healthy donors to assess the level of concordance and define robust marker combinations for the identification of circulating MDSCs. At this stage, no mandatory requirements to standardize reagents or protocols were introduced. Data analysis revealed a small intra-laboratory, but very high inter-laboratory variance for all MDSC subsets, especially for the granulocytic subsets. In particular, the use of a dead-cell marker altered significantly the reported percentage of granulocytic MDSCs, confirming that these cells are especially sensitive to cryopreservation and/or thawing. Importantly, the gating strategy was heterogeneous and associated with high inter-center variance. Overall, our results document the high variability in MDSC phenotyping in the multicenter setting if no harmonization/standardization measures are applied. Although the observed variability depended on a number of identified parameters, the main parameter associated with variation was the gating strategy. Based on these findings, we propose further efforts to harmonize marker combinations and gating parameters to identify strategies for a robust enumeration of MDSC subsets.

Keywords: Flow cytometry; Myeloid-derived suppressor cells; Phenotyping; Proficiency panel.

Fig. 1

Exemplary gating strategy included in the panel’s guideline. The figure shows the gating strategy proposed in the guidelines and made of three sequential steps, allowing the identification of the 10 putative MDSC subsets. Step 1 consists in doublets exclusion followed by dead-cell exclusion. Step 2 allows the identification of three MDSC subsets (MDSC1-2-8), and from a consecutive gate, the other seven defined MDSC populations are identified in step 3

Fig. 2

Analysis of the impact of a dead-cell marker on MDSC quantification. Black bars indicate average %CV from analyses performed without exclusion of dead cells, while white bars refer to analyses considering only live cells (n = 18 evaluable laboratories reporting data for the 8-color panel). Intra-laboratory variance is shown in panel a, while panel b reports inter-laboratory variance. c Average %CV (n = 14 evaluable laboratories reporting data for the 8-color panel) of the frequencies of the ten MDSC subsets without normalization (black bar) and normalized on the number of singlets (yellow bar), lymphocytes (blue bar) and lymphocytes + monocytes (red bar)

Fig. 3

Values of a representative subsets of monocytic (MDSC4 – panel a), granulocytic (MDSC5 – panel b) and immature (MDSC10 – panel c) MDSCs are shown as examples. Frequencies were normalized on the count of lymphocytes + monocytes for each laboratory performing either one 8-color panel or three 4-color panels for MDSC identification; the two paired histograms refer to the two independent experimental runs performed by each laboratory (black bars = first run, gray bars = second run). Missing values, not reported by single laboratories, were labeled with “m”

Fig. 4

a Average frequencies (n = 18 evaluable laboratories reporting data for the 8-color panel) of MDSC subsets normalized on lymphocytes + monocytes ( n%) identified either excluding or not the dead cells. Wilcoxon matched-pair test (***P < 0.001, ****P < 0.0001). b Average %CV (n = 18 evaluable laboratories reporting data for the 8-color panel) of the frequencies of MDSC subsets identified either using a homogenous gating strategy (white bar) or not (black bars)