Breast adenocarcinoma liver metastases, in contrast to colorectal cancer liver metastases, display a non-angiogenic growth pattern that preserves the stroma and lacks hypoxia

Abstract

Although angiogenesis is a prerequisite for the growth of most human solid tumours, alternative mechanisms of vascularisation can be adopted. We have previously described a non-angiogenic growth pattern in liver metastases of colorectal adenocarcinomas (CRC) in which tumour cells replace hepatocytes at the tumour-liver interface, preserving the liver architecture and co-opting the sinusoidal blood vessels. The aim of this study was to determine whether this replacement pattern occurs during liver metastasis of breast adenocarcinomas (BC) and whether the lack of an angiogenic switch in such metastases is due to the absence of hypoxia and subsequent vascular fibrinogen leakage. The growth pattern of 45 BC liver metastases and 28 CRC liver metastases (73 consecutive patients) was assessed on haematoxylin- and eosin-stained tissue sections. The majority of the BC liver metastases had a replacement growth pattern (96%), in contrast to only 32% of the CRC metastases (P<0.0001). The median carbonic anhydrase 9 (CA9) expression (M75 antibody), as a marker of hypoxia, (intensity x % of stained tumour cells) was 0 in the BC metastases and 53 in the CRC metastases (P<0.0001). There was CA9 expression at the tumour-liver interface in only 16% of the BC liver metastases vs 54% of the CRC metastases (P=0.002). There was fibrin (T2G1 antibody) at the tumour-liver interface in only 21% of the BC metastases vs 56% of the CRC metastases (P=0.04). The median macrophage count (Chalkley morphometry; KP-1 anti-CD68 antibody) at the interface was 4.3 and 7.5, respectively (P<0.0001). Carbonic anhydrase 9 score and macrophage count were positively correlated (r=0.42; P=0.002) in all metastases. Glandular differentiation was less in the BC liver metastases: 80% had less than 10% gland formation vs only 7% of the CRC metastases (P<0.0001). The liver is a densely vascularised organ and can host metastases that exploit this environment by replacing the hepatocytes and co-opting the vasculature. Our findings confirm that a non-angiogenic pattern of liver metastasis indeed occurs in BC, that this pattern of replacement growth is even more prevalent than in CRC, and that the process induces neither hypoxia nor vascular leakage.

Figure 1

Breast cancer liver metastasis, replacement growth pattern (haematoxylin and eosin stain): the tissue architecture of the liver is preserved within the tumour tissue. There is close contact between tumour cells and hepatocytes at the interface and no inflammation.

Figure 2

Colorectal cancer liver metastasis, desmoplastic growth pattern (haematoxylin and eosin stain): a rim of desmoplastic stroma separates the liver parenchyma from the tumour tissue. A dense inflammatory cell infiltrate is present in the stroma nearby the liver parenchyma.

Figure 3

Carbonic anhydrase 9 immunostaining of a colorectal cancer liver metastasis: strong staining (brown; score 3+) of the tumour cells. Constitutive expression by bile duct epithelium (internal positive control).

Figure 4

Immunostaining of fibrin (brown) in a desmoplastic colorectal cancer liver metastasis: fibrin deposits mainly in the liver parenchyma surrounding the metastasis.

Figure 5

Anti-CD68 immunostaining demonstrating numerous macrophages (brown) in the desmoplastic rim surrounding a colorectal cancer liver metastasis. The Kupffer cells in the liver parenchyma are also immunoreactive.

Figure 6

Expression of LYVE-1 (brown, immunostaining) by sinusoidal endothelial cells within the liver parenchyma at the interface with a replacement-type breast cancer liver metastasis (arrow heads). Sinusoids engulfed by tumour cells express LYVE-1 at the tumour–liver interface (arrows) and lose this expression towards the centre of the metastasis (towards the right on the microphotograph).

Figure 7

Carbonic anhydrase 9 immunostaining of a breast cancer liver metastasis: replacement growth and no immunostaining (constitutive expression by bile duct epithelium (internal positive control) was present in the section (not shown)).

Figure 8

Immunostaining of fibrin (brown) in a replacement breast cancer liver metastasis: no staining at the tumour–liver interface (internal positive control: fibrin in a sinusoidal blood vessel (arrow)).

Figure 9

Anti-CD68 immunostaining demonstrating the Kupffer cells in the liver parenchyma. No macrophages at the tumour–liver interface of this breast cancer metastasis with replacement growth.

Figure 10

Expression of LYVE-1 (brown, immunostaining) by sinusoidal endothelial cells in the liver parenchyma surrounding a desmoplastic colorectal liver metastasis (arrows). No staining within the tumour tissue.