Stroma AReactive Invasion Front Areas (SARIFA) and the tumour immune microenvironment in colorectal cancer

This is a reply to Tapiainen et al.:

Tapiainen, V.V., Sirniö, P., Elomaa, H. et al. Stroma AReactive Invasion Front Areas (SARIFA), tumour immune microenvironment, and survival in colorectal cancer. Br J Cancer (2025). 10.1038/s41416-025-02972-z.

We read with great interest the article by Tapiainen et al. [1], which did not only validate our findings of Stroma AReactive Invasion Front Areas (SARIFA) as a strong prognostic predictor in colorectal cancer (CRC) [2–5] but also provided further evidence that SARIFA-positivity, that means the direct interaction between tumour cells and adipocytes at the invasion front, is associated with an altered immunity [6, 7]. Besides conventional immunohistochemistry (IHC), which we used in our previous studies, the authors analysed multiplex IHC using digital image analysis [1].

Interested in the immunologic implications of SARIFA, we aimed to validate and complement these findings by deploying the publicly available DFCI/Orion CRC cohort published by Wala et al. in 2024 [8], which provides spatial and targeted genomic analysis data of CRCs. Based on the publicly available representative routine H&E whole slide images (WSI), we established the SARIFA-status according to our biomarker definition, which was also used by Tapiainen et al. in their study [1]. SARIFA-positivity is defined as the direct contact between at least five tumour cells/one tumour gland and adipocytes at the tumour invasion front without intervening stromal or inflammatory reaction (Fig. 1a). The SARIFA-status could be reliably assessed in 35 samples (SARIFA-negative n = 17, SARIFA-positive n = 18) of the DFCI/Orion 2024 CRC cohort, with representative H&E as well as spatial immune profiling WSI available via the multiplex image viewer MINERVA and cBioPortal [9].

Fig. 1. SARIFA-positivity, which is defined as direct tumor-adipocyte interaction at the invasion front, is associated with an aggressive course of disease and an altered immunity in colorectal cancer.

a Example of a SARIFA-positive multiplex/H&E WSI from the DFCI/Orion cohort [8] (https://www.cbioportal.org/study/filesAndLinks?id=crc_orion_2024, accessed on the 4^th April 2025) displaying direct tumour-adipocyte interaction (upper panel) as well as a SARIFA-negative CRC with intervening desmoplastic stroma preventing a direct tumour-adipocyte interaction in this mucinous adenocarcinoma (lower panel). b Kaplan-Meier curve for overall survival stratified by SARIFA-status with p-value of log-rank test. c Kaplan-Meier curve for progression-free survival stratified by SARIFA-status with p-value of log-rank test. d–i Neighbourhood Cell Densities (cells/mm²) of different markers of CRCs from the DFCI/Orion cohort with p-values (t-test or Wilcoxon test, depending on distribution) stratified by SARIFA-status (violin plots visualizing the data distribution using a kernel density estimate; no negative cell counts were observed; the violins are crossing the zero-line due to the density estimation). CRC colorectal cancer, H&E hematoxylin-and-eosin, PFS progression-free survival, OS overall survival, DFCI Dana-Farber Cancer Institute, SARIFA Stroma AReactive Invasion Front Areas, WSI whole slide image.

In this comparably small cohort of CRCs, SARIFA-positivity was associated with a reduced overall and a significantly reduced progression-free survival (OS, log-rank p = 0.089, PFS, log-rank p = 0.011, Fig. 1b, c), proving again its prognostic value, which now has been validated in numerous studies [2–5].

Next, we then compared the available multiplex immunofluorescence cell densities between the SARIFA-positive and SARIFA-negative CRCs. Interestingly, we provide here evidence for fewer FoxP3-positive cells in SARIFA-positive CRCs (p = 0.005, Fig. 1f), which are considered regulatory T cells. In contrast to the situation in other entities, higher number of intrastromal FoxP3 cells, together with a higher number of intraepithelial CD8 T cells, have previously been linked to a better prognosis in CRCs of the SCOT trial [10]. We have also already shown that high numbers FoxP3-positive T cells at the invasion front are associated with a favourable prognosis in CRC [11]. CD8-positive T cells showed a trend wise higher cell density in SARIFA-negative CRCs (p = 0.096, Fig. 1h). For the macrophage marker CD68, there was no difference between SARIFA-positive and SARIFA-negative CRCs (p = 0.503, Fig. 1d), which is completely in line with the data from Tapiainen et al. [1]. In our previous studies, we observed higher numbers of CD68-positive macrophages specifically at the SARIFA-positive invasion front based on immunohistochemistry data in gastric as well as pancreatic cancer [12, 13]. For CD163, which is a marker of M2-polarisation in macrophages, there was a trend wise higher cell density in SARIFA-negative CRCs (p = 0.072, Fig. 1e) without statistical significance, and opposed to the data by Tapiainen et al.– nevertheless, the authors had a much higher sample number in their study, which makes their findings (including significantly higher numbers of M2-macrophages in SARIFA-positive CRCs) very reliable. There was also a trend for an increased PD1 expression in SARIFA-negative CRCs (p = 0.067, Fig. 1g) as well as a higher Ki67 index (p = 0.057, Fig. 1i). Even though it is hypothesized that highly proliferating cells show lower invasive potential, which could explain the potential link between higher Ki67 and SARIFA-negativity, we did not observe this association based on conventional immunohistochemistry data in larger gastric as well as colorectal cancer cohorts [3, 12]. The cell densities based on the other markers (CD20, CD31, CD45. CD45RO, E-Cadherin, PanCK, PD-L1, SMA) did not show SARIFA-dependent differences (each p-value above 0.05).

In their study, Tapiainen et al. could link SARIFA-positivity convincingly to a higher density of M2-like macrophages and CD14-positive HLA-DR-negative immature monocytic cells in their large CRC cohort [1], which once again highlights the immunologic implications of SARIFA-positivity in CRCs.

Our approach here has some limitations and only offers preliminary insights. We did not spatially resolve the marker expression (areas of direct tumour-adipocyte interaction, e.g. SARIFA, vs tumour centre vs superficial tumour surface). Moreover, the cohort (n = 35) was quite small.

In conclusion, SARIFA, as assessed from routine H&E morphology, provides valuable insights into the tumour microenvironment. Despite the inherent limitations of using a small cohort, we offer further insights into immune alterations, particularly the reduced number of FoxP3-positive regulatory T cells in SARIFA-positive CRCs, which should be further investigated. Given its simplicity, low cost, and minimal observer variability, we advocate for the routine implementation of SARIFA-status assessment in clinical diagnostics. Our findings complement and expand on those reported by Tapiainen et al. [1], reinforcing SARIFA’s role not only as a morphological marker but also as a reflection of tumour-immune dynamics.

Author contributions

NGR and BM conceived the study. NGR performed data analysis. BM and NGR interpreted the data and wrote the manuscript.

Competing interests

The authors declare no competing interests.