Comparative analysis of the immune microenvironment in histological subtypes of lung and breast cancer using a tissue microarray (TMA) comprising invasive margin (IM) and tumor center (TC)

M Cumberbatch (1), L Memeo (1), C Womack (1), M Bhagat (1), WH Kim (2).
(1) TriStar Technology Group, LLC, Washington, DC 20006, USA, (2) Seoul National University Medical College, Seoul, South Korea.

Logo (1)


Immunotherapy is proving successful in several tumour settings, although many cancer types remain less responsive. Pre-existing tumour infiltrating lymphocytes and PD-L1 expression associate with response to immunotherapy and cancers have been classified into response subtypes according to these biomarkers. To enable a simultaneous comparative assessment of the immune microenvironment in multiple tumour indications, we have constructed a multi-tumour TMA comprising matched cores for each donor from IM and TC.


In this study, we employed a TMA comprising cores taken from formalin fixed paraffin embedded (FFPE) non-treated surgical resections for donors diagnosed with squamous NSCLC (SCC; n=13), adeno NSCLC (ADC; n=12), small cell lung cancer (SCLC; n=9), oestrogen receptor positive breast cancer (ER+BC; n=12), Herceptin positive BC (Her2+BC; n=12) and triple negative BC (TNBC; n=12). For each case, a core (1mm) was taken from the IM and TC, total n=144 cores. Serial sections were stained by immunohistochemistry for immune biomarkers: CD3 (2GV6), CD8 (SP57), CD4 (SP35), CD20 (L26), Foxp3 (236A/E7), PD-1 (D4W2J), CD163 (2G12), CD68 (PG-M1), PD-L1 (22C3). Immune infiltrates were analysed by digital image analysis (CellProfilerTM) and PD-L1 was scored by pathologist to deliver both the tumour proportion score (TPS) and the combined positivity score (CPS).


In the first instance, the mean (±SE) of all cores (IM and TC) for each disease indication was calculated. In line with published literature, NSCLC (SCC and ADC) and TNBC were more highly infiltrated (Figure 1). Interestingly, where CD4+ cell counts were higher than anticipated, it was clear that cells with macrophage/dendritic morphology were detected in addition to T cells; an observation that was more prevalent in the more highly infiltrated tumour types.

Figure 1

Immune cell infiltration in histological subtypes of breast and lung cancers


Figure 1. Serial sections (4μm) were stained for CD3, CD8, CD4 and CD20. Digital images were analysed using CellProfiler™ and results displayed as cells/mm² (A). A low magnification image of a stained TMA is shown in B (CD3); representative images for each marker are displayed in C (NSCLC SCC).

Figure 2

Suppressive populations and PD-L1 in histological subtypes of breast and lung cancers


Figure 2. Serial sections (4μm) were stained for Foxp3, PD-1 and PD-L1. Foxp3 and PD-1 were analysed digitally using CellProfiler™ and results displayed as cells/mm2 (A). PD-L1 was scored by pathologist to eliver Tumour Proportion Score (TPS) and Combined Positivity Score (CPS) (B). Low magnification image of PD-L1 stained TMA (C); representative images of staining (D).

Immunosuppressive populations (Foxp3+ and PD-1+ cells) were generally higher in lung cancer (Figure 2A), with the highest prevalence of PD-1 in squamous lung cancer (SCC). While tumour PD-L1 (TPS) was absent in ER+BC and Her2+BC, the contribution from infiltrating immune cells (CPS) was greatest in SCC and TNBC (Figure 2B). Analyses of macrophage markers revealed a reciprocal profile for CD163:CD68 in lung cancer vs breast cancer with the M2-like CD163+ cell population exceeding CD68+ macrophages in breast cancer, whereas the converse was observed for lung cancer.

Figure 3

Distribution of macrophage/monocyte populations in breast and lung cancers

Figure 3. Serial sections (4μm) were stained for CD68 and CD163 and digital images analysed using CellProfilerTM.Results are displayed as cells/mm2 (A). The balance of M2-like macrophage/monocyte marker CD163 to CD68 was determined for IM and TC, and the ratio displayed in (B). The higher number in BC reveals a more M2-like macrophage/monocyte population in this disease setting compared with lung cancer. (C) representative images.


Immune infiltrates were reduced in TC compared with IM for breast cancer (Figure 4A), and distinct case-by-case immune microenvironments were revealed within each disease indication (Figure 4B).

Figure 4

Analyses of immune infiltrates in IM versus TC for breast and lung cancers


Figure 4. Data for each histological subtype were averaged for IM  and TC  to explore potential differences in immune infiltrates in these distinct tumour locations (A). Although general trends were revealed, distinct immune profiles were evident on a case by case basis, two patient examples are illustrated (B).


  • Although ER+ and Her2+ breast cancers were less infiltrated, these data reveal a more M2-like monocytic/macrophage phenotype in these histological subtypes compared with TNBC or lung cancer.
  • Using an immuno-oncology focussed TMA we have revealed distinct immune profiles in multiple tumour subtypes simultaneously, providing a valuable basis against which to profile novel immune targets.