Breast carcinoma can be classified as either Estrogen Receptor (ER) positive or negative by immunohistochemical phenotyping, although ER expression may vary from 1 to 100% of malignant cells within an ER + tumor. This is similar to genetic variability observed in other tumor types and is generally viewed as a consequence of intratumoral evolution driven by random genetic mutations. Here we view cellular evolution within tumors as a classical Darwinian system in which variations in molecular properties represent predictable adaptations to spatially heterogeneous environmental selection forces. We hypothesize that ER expression is a successful adaptive strategy only if estrogen is present in the microenvironment. Since the dominant source of estrogen is blood flow, we hypothesized that, in general, intratumoral regions with higher blood flow would contain larger numbers of ER + cells when compared to areas of low blood flow and in turn necrosis.
This study used digital pathology whole slide image acquisition and advanced image analysis algorithms. We examined the spatial distribution of ER + and ER- cells, vascular density, vessel area, and tissue necrosis within histological sections of 24 breast cancer specimens. These data were correlated with the patients ER status and molecular pathology report findings.
ANOVA analyses revealed a strong correlation between vascular area and ER expression and between high fractional necrosis and absent ER expression (R2 = 39%; p < 0.003 and R2 = 46%; p < 0.001), respectively). ER expression did not correlate with tumor grade or size.
We conclude that ER expression can be understood as a Darwinian process and linked to variations in estrogen delivery by temporal and spatial heterogeneity in blood flow. This correlation suggests strategies to promote intratumoral blood flow or a cyclic introduction of estrogen in the treatment schedule could be explored as a counter-intuitive approach to increase the efficacy of anti-estrogen drugs.