Validation of an Accurate Automated Multiplex Immunofluorescence Method for Immuno-Profiling Melanoma

Abstract

Multiplex immunofluorescence staining enables the simultaneous detection of multiple immune markers in a single tissue section, and is a useful tool for the identification of specific cell populations within the tumour microenvironment. However, this technology has rarely been validated against standard clinical immunohistology, which is a barrier for its integration into clinical practice. This study sought to validate and investigate the accuracy, precision and reproducibility of a multiplex immunofluorescence compared with immunohistochemistry (IHC), including tissue staining, imaging and analysis, in characterising the expression of immune and melanoma markers in both the tumour and its microenvironment. Traditional chromogenic IHC, single-plex immunofluorescence and multiplex immunofluorescence were each performed on serial tissue sections of a formalin-fixed paraffin-embedded (FFPE) tissue microarray containing metastatic melanoma specimens from 67 patients. The panel included the immune cell markers CD8, CD68, CD16, the immune checkpoint PD-L1, and melanoma tumour marker SOX10. Slides were stained with the Opal^™ 7 colour Kit (Akoya Biosciences) on the intelliPATH autostainer (Biocare Medical) and imaged using the Vectra 3.0.5 microscope. Marker expression was quantified using Halo v.3.2.181 (Indica Labs). Comparison of the IHC and single-plex immunofluorescence revealed highly significant positive correlations between the cell densities of CD8, CD68, CD16, PD-L1 and SOX10 marker positive cells (Spearman's rho = 0.927 to 0.750, p < 0.0001). Highly significant correlations were also observed for all markers between single-plex immunofluorescence and multiplex immunofluorescence staining (Spearman's rho >0.9, p < 0.0001). Finally, correlation analysis of the three multiplex replicates revealed a high degree of reproducibility between slides (Spearman's rho >0.940, p < 0.0001). Together, these data highlight the reliability and validity of multiplex immunofluorescence in accurately profiling the tumour and its associated microenvironment using FFPE metastatic melanoma specimens. This validated multiplex panel can be utilised for research evaluating melanoma and its microenvironment, such as studies performed to predict patient response or resistance to immunotherapies.

Keywords: immunohistochemistry; immunotherapy; melanoma; multiplex immunofluorescence; multispectral imaging; pathology; predictive biomarker; tumour microenvironment.

FIGURE 1

Workflow for traditional chromogenic and immunofluorescence immunohistochemistry representing the steps taken from staining the TMA slides to image acquisition and analysis. TMA slides were stained on the Dako autostainer and then stained with chromogenic immunohistochemistry, single-plex immunofluorescence and multiplex immunofluorescence. Stained whole-slides were scanned with the Vectra 3.0 and high resolution images (20×) were acquired. Images were spectrally unmixed using inform. Analysis at the single-cell level was conducted using Halo Image Analysis Software.

FIGURE 2

Microphotographs of representative examples of staining with traditional chromogenic immunohistochemistry (left panels) and single-plex immunofluorescence (right panels) along with corresponding 20× sample regions from the same patient core for each marker. White arrows indicate positive staining. Antibodies stained using traditional chromogenic IHC and single-plex immunofluorescence show similar patterns of expression. Scale bars are 200 and 50 µm for the low-magnification (×7.5) and high magnification (×20) microphotographs respectively.

FIGURE 3

Correlation plots represent the correlation between the density of positive cells in traditional chromogenic immunohistochemistry and single-plex immunofluorescence for each marker in represented. Correlation analysis between the cell densities for each marker showed significant positive correlations. The strongest correlations were observed in CD16 (n = 59) and SOX10 (n = 61). The lowest correlations were observed in CD8 (n = 62). Data are Spearman’s correlation values with p < 0.05.

FIGURE 4

Microphotographs of representative examples of staining in the respective control tissue (a lymph node containing metastatic melanoma) for chromogenic immunohistochemistry (left panels), single-plex immunofluorescence (middle panels), and multiplex immunofluorescence (right panels) display comparable expression patterns for all markers. Scale bars are 50 μm at ×20 magnification.

FIGURE 5

Scatter plots comparing the average intensities of each marker in the single-plex immunofluorescence staining to the intensities in the three multiplex replicates. Single-plex immunofluorescence displayed higher marker intensities for CD68 (n = 60), CD16 (n = 59) and SOX10 (n = 60) compared to the corresponding multiplex immunofluorescence staining. PD-L1 (n = 56) single-plex and multiplex immunofluorescence staining demonstrated similar marker intensities. Single-plex immunofluorescence staining with CD68 demonstrated a larger variation in marker intensities compared to the multiplex immunofluorescence intensities. Multiplex 3 displayed lower cell intensities for markers CD8 (n = 60), CD16, and SOX10. One-way ANOVA with Tukey’s HSD (honestly significant differences) test, **** = p < 0.0001, ns, non-significant.

FIGURE 6

Microphotographs of representative examples of staining in the single-plex immunofluorescence (left panels) and the multiplex immunofluorescence (right panels) along with corresponding ×20 sample regions from the same patient core for each marker. Antibodies stained with single-plex immunofluorescence and multiplex immunofluorescence show similar patterns of expression. Scale bars are 200 and 50 µm for the low-magnification (×7.5) and high magnification (×20) microphotographs respectively.

FIGURE 7

Correlation plots compare the correlation between the cell densities for each marker in single-plex immunofluorescence and multiplex immunohistochemistry. An average of the cell densities of each marker across the three multiplex immunofluorescence sections was used to compare with the single-plex immunofluorescence marker cell density. Significant positive correlations between the single-plex immunofluorescence cell densities and the average cell density of the multiplex immunofluorescence were observed for all markers. CD8 displayed the strongest correlation between single-plex immunofluorescence and multiplex immunofluorescence. Data are Spearman’s correlation values with p < 0.05.

FIGURE 8

Microphotographs of representative examples of staining in multiplex immunofluorescence cores from the same patient core for each marker show similar patterns of expression in each multiplex. Composite images show the integration of markers on a single slide. Scale bars are 200 μm at ×7.5 magnification.

FIGURE 9

A correlation matrix heatmap comparing the relationship between the cell densities of each marker across the three consecutive multiplex immunofluorescence sections. Highly significant positive correlations were observed between the three Multiplexes for all markers. The strongest correlations were observed for PD-L1 between Multiplex 2 and Multiplex 3. The size of the circles represent the absolute value of the correlation coefficient. The colours of the circles represent the sign and magnitude of the correlation with shades of red representing positive and higher correlation coefficients and shades of blue representing negative and lower correlation coefficients. Data are Spearman’s correlation values with p < 0.05.

FIGURE 10

Microphotographs of representative examples of marker co-localisation in multiplex immunofluorescence, showing specific cell phenotypes. Composite images show the integration of markers on a single slide. White arrows indicate cell phenotypes which include PD-L1+ macrophages (CD68 positive, CD16 positive, and PD-L1 positive cells), CD8-expressing macrophages (CD68 positive, CD16 positive, and CD8 positive cells), PD-L1+ tumour cells (PD-L1 positive and SOX10 positive cells), and PD-L1+ cytotoxic T-cells (CD8 positive and PD-L1 positive cells). Scale bars are 20 µm (40×).