It has been demonstrated that reactive stromal formation in solid tumours is associated with disease progression and poor outcome. Genes have been identified that are involved in biological processes such as angiogenesis and alterations in the extracellular matrix, including desmoplasia. (Gao et al, 2011; Planche et al, 2011; Berdiel-Acer et al, 2014; Dakhova et al, 2014; Duss et al, 2014).

The presence of stromal cells located in the interior of the tumour, surrounded by small groups or nests of tumour cells, is partly determinative of its (pre) metastatic capacity. Over the last decade, the tumour-stroma ratio (TSR) has gained significant interest in the disease prediction of patients with breast, colon, oesophageal, lung and cervical cancer. The elegance of the parameter is the use of conventional Hematoxylin & Eosin-stained slides for histopathological microscopy analysis. The use of a simple cut-off value, proven to be applicable for multiple solid tumour types, distinguishes between stroma-high and stroma-low tumours, of which the stroma-high tumours are independently associated with a relatively worse prognosis. (Mesker et al, 2007, 2009; Courrech Staal et al, 2010; West et al, 2010; Beck et al, 2011; Courrech Staal et al, 2011; de Kruijf et al, 2011; Ahn et al, 2012; Huijbers et al, 2012; Moorman et al, 2012; Wang et al, 2012, 2013; Dekker et al, 2013; Liu et al, 2014; Park et al, 2014).

The TSR has demonstrated the highest prognostic power when looking at the population of triple-negative breast tumours. For this group of patients, the prognostic hazard ratio (HR) for disease recurrence was reported as high as 4.12 (P=0.006) and 3.0 (P=0.0034) for patients harbouring stroma-rich tumours. (10,13). Furthermore, within our own data set, oestrogen receptor-positive patients also show a significant relapse-free survival (RFS) difference in the disadvantage for stroma-producing tumours (RFS P=0.001, HR 1.8). Similar results were observed in the POP study (de Kruijf et al, 2011; Dekker et al, 2013).

Now, Downey et al present the analysis of 118 female breast cancers with stromal formation resulting in a relatively favourable prognosis. These data are in contrast with formerly obtained data on breast and other solid cancers scoring the TSR parameter as described by our group (and subsequently validated by others). (Mesker et al, 2007, 2009; Courrech Staal et al, 2010; West et al, 2010; Beck et al, 2011; Courrech Staal et al, 2011; de Kruijf et al, 2011; Ahn et al, 2012; Huijbers et al, 2012; Moorman et al, 2012; Wang et al, 2012, 2013; Dekker et al, 2013; Liu et al, 2014; Park et al, 2014). In the rest of this letter, we will describe methodological differences between the method used by Downey et al and previously published reports, which might underlie the differences in results.

First, Downey et al evaluate only one 9 mm2 area at the tumour’s leading or non-leading edge. This area was selected with the emphasis that the advancing ‘front’ of a tumour may be more proliferative and the metabolic activity of tumour cells in this area is not compromised by a potential lack of nutrients. This method of TSR scoring of a given tumour underestimates the heterogeneity within the tumour concerning stromal production (Zhang et al, 2014). It is our personal experience that a solid tumour can be very heterogeneous for desmoplastic characteristics. Estimation of the TSR as indicated by our group entails evaluation of the complete tumour area after which the TSR is determined based on the intratumoural area with the highest degree of stromal formation (Mesker et al, 2007, 2009; Courrech Staal et al, 2010; de Kruijf et al, 2011; Courrech Staal et al, 2011; Ahn et al, 2012; Huijbers et al, 2012; Moorman et al, 2012; Wang et al, 2012, 2013; Dekker et al, 2013; Liu et al, 2014; Park et al, 2014). For colorectal cancer, it has been shown that the deepest invasive part of the tumour is the most determinative for tumour progression and almost invariably demonstrates the highest stromal formation (if any). For breast cancer this is not applicable, possibly in part because these tumours do not progress through adjacent, consecutive tissue layers as is the case in colorectal tumours (mucosa, submucosa, muscle layers and so on). As such, the area with the highest amount of desmoplasia cannot be predicted and thus warrants evaluation of all available microscopic slides. Also for cervical, non-small lung and oesophageal cancer confirmatory data was observed (Courrech Staal et al, 2010, 2011; Wang et al, 2012, 2013; Liu et al, 2014).

Downey’s study also did not indicate whether patients were pre-treated with radio, chemo or endocrine therapy. The studies by Moorman et al (2012) and de Kruijf et al (2011) excluded patients with neoadjuvant therapy as therapy influences the tissue arrangement including desmoplasia. Furthermore, no clinical-pathological data with respect to the proportion of stroma was provided, and neither were data for univariate analysis.

West et al (2010) used an identical approach as Downey for colorectal cancer, for this study an area of the luminal region was selected, resulting in a comparable cut-off points and survival data within stages I–III as given for the conventional TSR scoring. For this study no patients with pre-operative chemo or radiotherapy were included.

In the current setting we do not think that the reported method of Downey et al validates the approach of the TSR as it was only based on a subselected tumour area. The previously described TSR by our group is determined on the distribution of the stroma within the complete tumour including areas with heterogeneity and highly aggressive stromal formation. These contrasting reports illustrate the use of stringent criteria for scoring intratumoural stromal formation in order to reliably integrate the TSR into clinical decision-making.