Multiplex immunohistochemistry accurately defines the immune context of metastatic melanoma

Abstract

A prospective study explored the heterogeneous nature of metastatic melanoma using Multiplex immunohistochemistry (IHC) and flow cytometry (FACS). Multiplex IHC data quantitated immune subset number present intra-tumoral (IT) vs the tumor stroma, plus distance of immune subsets from the tumor margin (TM). In addition, mIHC showed a close association between the presence of IT CD8⁺ T cells and PDL1 expression in melanoma, which was more prevalent on macrophages than on melanoma cells. In contrast, FACS provided more detailed information regarding the T cell subset differentiation, their activation status and expression of immune checkpoint molecules. Interestingly, mIHC detected significantly higher Treg numbers than FACS and showed preferential CD4⁺ T cell distribution in the tumor stroma. Based on the mIHC and FACS data, we provide a model which defines metastatic melanoma immune context into four categories using the presence or absence of PDL1⁺ melanoma cells and/or macrophages, and their location within the tumor or on the periphery, combined with the presence or absence of IT CD8⁺ T cells. This model interprets melanoma immune context as a spectrum of tumor escape from immune control, and provides a snapshot upon which interpretation of checkpoint blockade inhibitor (CBI) therapy responses can be built.

Figure 1

Multiplex IHC (mIHC) defines the precise location of immune subsets in metastatic melanoma. FFPE sections of melanoma were stained by OPAL “T-cell Panel” for SOX10, PDL1, CD3 (red), CD8, CD4 and FOXP3. Stained sections were imaged on the Vectra Automated Imaging System and composite images displayed for Mel TIL026 ((A) pelvic lymph node) and MelTIL015 ((B) visceral). Using the low power image, high-powered Fields (HPFs) were selected from specific tumour regions including intra-tumoral (IT), tumour margin (TM) and stromal regions (S). Data displayed for the T cell panel shows the distribution of tumour cells (purple), PDL1 (membranous yellow) on either melanoma or infiltrating APCs, CD8+ T cells (green), CD4+ T cells (pink) and T regulatory T cells (orange nucleus). Cell segmentation and phenotyping (Supplementary Fig. 1) enabled quantitation expressed as counts and percentage of total cells per high-powered field (HPF) for different T cell subsets (A,B). Cell distance measurements (Supplementary Fig. 2) enabled detailed distance analysis of T cell subsets from the tumour margin (boxed graph). In (C) collated data for all Mel-TIL samples in the study cohort is depicted, including the % of T cell subsets (CD8⁺, CD4⁺ and Tregs) in three tumour regions (IT, TM and S) as well as the average of these regions. Statistical analysis (Mann Whitney) was performed across the cohort, significantly different data is represented by *(p < 0.05) and ***(p < 0.005).

Figure 2

Multiplex IHC reveals the broader immune context of metastatic melanoma. FFPE sections of metastatic melanoma from MelTIL024, s.c. arm metastasis (A) and MelTIL026, pelvic lymph node metastasis (B) were stained by OPAL for the “T-cell Panel” including SOX10 (melanoma marker), PDL1, CD3, CD8, CD4 and FOXP3 and in the “Pan-Immune Panel” SOX10, PDL1, CD3, CD20, CD68 and CD11c. Stained sections were imaged on the Vectra Automated Imaging System and composite HPF images displayed. The T cell Panel showed the distribution of tumour cells (purple), PDL1 (membranous yellow) on either melanoma or infiltrating APCs, CD8+ T cells (green), CD4+ T cells (pink) and T regulatory T cells (orange nucleus). The pan-immune panel showed distribution of tumour cells (purple), PDL1 (membranous yellow) on either melanoma or infiltrating APCs, CD3+ T cells (red), CD20+ B cells (orange), CD68+ macrophages (white cytoplasmic), and CD11c+ likely dendritic cells (green membranous). High-powered fields were selected based on highest TIL density, and cell populations measured using inform v2.2 software. Immune context quantitation from the T cell panel and pan immune panel was recorded and plotted in Graph Pad PRISM as counts per HPF and % of total cells for each resected metastasis. Collated data for all Mel-TIL samples in the study cohort is depicted in (C) including the % of T cells (CD3), B cells (CD20), macrophages (CD68) and CD11c⁺ (likely dendritic cells) in three tumour regions (IT, TM and S) as well as the average of these regions. Statistical analysis (Mann Whitney) was performed across the cohort, significantly different data is represented by *(p < 0.05), **(p < 0.01) and ****(p < 0.0001).

Figure 3

Melanoma TILs. Melanoma metastases were analysed by FACS (gating strategy as per Supplementary Fig. 6) to characterize the T cells present. Viable cells were assessed for (A) % TILs (HLA-ABC⁺CD45⁺), and (B) T cell subsets. In addition, T cell compartments in (C) CD4⁺ T cells, and (D) CD8⁺ T cells; immune checkpoints and activation markers in (E) CD4⁺ T cells, and (F) CD8⁺ T cells, and PD-1 expression by T cell compartments in (G) CD4⁺ T cells, and (H) CD8⁺ T cells. Finally, relative expression of IL-7Rα and PD-1 was assessed in (I) CD4⁺ T cells, and (J) CD8⁺ T cells. Data is presented as %+ for each marker; statistical analysis (Mann Whitney) was performed across the cohort, significantly different data is represented by *(p < 0.05), **(p < 0.01), ***(p < 0.001) and ****(p < 0.0001).

Figure 4

Melanoma stromal distribution of CD4⁺ T cells is a key factor for discrepant TIL data between mIHC and FACS. Melanoma metastases were analysed by FACS and mIHC (as per methods) and the T cell data compared. %TILs were measured by FACS as %(CD45+ HLA− ABC+) of all viable cells (as depicted in Supplementary Fig. 6) and %TILs by mIHC as the %CD3⁺DAPI⁺ cells of total DAPI per HPF, and expressed as an average of multiple HPFs. Shown in (A) collated data for the cohort for %TILs measured via FACS vs mIHC, and divided into groups based on metastatic site; (B) a direct comparison of paired data for %TILs measured via FACS vs mIHC. Also, the % T cell subsets (CD8⁺, CD4⁺, Treg) measured by FACS vs mIHC in (C) for the whole cohort. A comparison of paired data (measured via FACS vs mIHC) in (D) %CD8⁺ T cells, (E) %CD4⁺ T cells, and (F) Treg. Finally CD8:Treg ratio was compared for mIHC vs FACS and grouped according to tissue origin (G). Statistical analysis (Mann Whitney) was performed across the cohort comparing mIHC and FACS data, significantly different data is represented by *(p < 0.05) and ****(p < 0.001). Also a Spearman correlative analysis was performed for (B) and (D–F), the r value is provided in each graph.

Figure 5

Multiplex IHC describes which cells express PDL1 and their precise location within metastatic melanoma. OPAL stained FFPE sections were imaged on the Vectra Automated Imaging System and cell phenotyping assigned for PD-L1 expression. Colour separation was performed on Inform Software v2.2 to create a composite image (A) where CD3 (red), CD20 (orange), CD68 (white), CD11c (green), SOX10 (purple), PDL1 (yellow), DAPI (blue). The composite image was edited to display DAPI and PDL1 with (B) CD68 and CD11c or (C) SOX10. This combination of images enabled appropriate phenotyping of the melanoma and/or the antigen-presenting cells for PDL1 expression. Using this strategy, all melanoma samples were examined for PDL1 expression. The PDL1 score by mIHC staining (0 = <1%; 1 = >1%, <5%; 2 = >5%, <10%; 3 = >10%) was compared to TIL percentage by mIHC (D) and FACS (E). The PDL1 score was then compared to the intratumoral (IT) CD8 (F) and CD4 (G) percentages by mIHC and FACS. Collated data depicted the location of PD-L1 staining for melanoma (diffuse, peripheral or negative) (H) and/or macrophages (intra-tumoral, peripheral, both or negative) (I) this was plotted against % IT TIL by mIHC. Also correlation between PDL1⁺ cell location plotted against % IT CD8, CD4 and Tregs TILs by mIHC (J) as well as the comparison to % stromal CD8, CD4 and Tregs (K). Statistical analysis (Mann Whitney) was performed across the cohort comparing mIHC and FACS data, significantly different data is represented by *(p < 0.05).

Figure 6

Conceptual diagram to depict metastatic melanoma immune context. Immune context of metastatic melanoma can be categorised into four distinct categories. Categories 1–3 are ‘T cell inflamed’ and characterised by IT CD8⁺ T cells and Treg, plus CD8⁺ T cells at the TM; CD4⁺ T cells, B cells and CD11c⁺ (likely dendritic cells) are in the tumor stroma. Category 1 also has PDL1⁺ melanoma and PDL1⁺ macrophages (IT and at TM). In Category 2, melanoma cells are PDL1⁻, but macrophages are PDL1⁺ (IT and/or TM). In category 3, melanoma and macrophages cells are PDL1⁻. Category 4 includes the T cell excluded (4A) where all immune cells are in the stroma and both melanoma and macrophages are PDL1⁻, or immune desert (4B) where no immune cells are present.

Figure 7

Using metastatic melanoma immune context data to understand checkpoint blockade inhibitor responses. Two metastatic melanoma samples (MelTIL015 and MelTIL013) had divergent responses to CBI. The immune context of each melanoma biopsy was examined by multiplex IHC (A,B) and FACS (C,D). Multiplex IHC data is shown for the T cell panel for MelTIL015 (A) and MelTIL013 (B) including a scanned low power image of the whole biopsy, plus an image at the TM or IT. FACS data was analyzed (as per Supplementary Fig. 6) and T cell differentiation compartments represented in pie chart format for MelTIL015 and MelTIL013 (C). In addition, TIL immune checkpoint (PD-1 and TIM-3), activation markers (OX-40) and IL-7Rα expression were represented in column charts for MelTIL015 and MelTIL013 (D).