Fibroblast activation protein identifies Consensus Molecular Subtype 4 in colorectal cancer and allows its detection by 68Ga-FAPI-PET imaging

Abstract

Background: In colorectal cancer (CRC), the consensus molecular subtype 4 (CMS4) is associated with therapy resistance and poor prognosis. Clinical diagnosis of CMS4 is hampered by locoregional and temporal variables influencing CMS classification. Diagnostic tools that comprehensively detect CMS4 are therefore urgently needed.

Methods: To identify targets for molecular CMS4 imaging, RNA sequencing data of 3232 primary CRC patients were explored. Heterogeneity of marker expression in relation to CMS4 status was assessed by analysing 3-5 tumour regions and 91.103 single-tumour cells (7 and 29 tumours, respectively). Candidate marker expression was validated in CMS4 peritoneal metastases (PM; n = 59). Molecular imaging was performed using the 68Ga-DOTA-FAPI-46 PET tracer.

Results: Fibroblast activation protein (FAP) mRNA identified CMS4 with very high sensitivity and specificity (AUROC > 0.91), and was associated with significantly shorter relapse-free survival (P = 0.0038). Heterogeneous expression of FAP among and within tumour lesions correlated with CMS4 heterogeneity (AUROC = 1.00). FAP expression was homogeneously high in PM, a near-homogeneous CMS4 entity. FAPI-PET identified focal and diffuse PM that were missed using conventional imaging. Extra-peritoneal metastases displayed extensive heterogeneity of tracer uptake.

Conclusion: FAP expression identifies CMS4 CRC. FAPI-PET may have value in the comprehensive detection of CMS4 tumours in CRC. This is especially relevant in patients with PM, for whom effective imaging tools are currently lacking.

Conflict of interest statement

The authors declare no competing interests.

© 2022. The Author(s).

Figures

Fig. 1. FAP mRNA expression identifies CMS4 CRC.

a Volcano plot showing differentially expressed genes between CMS2 and CMS4 subgroups that encode plasma membrane proteins. All individual genes and accompanying P and fold-change values are presented in Supplementary Table S1. b Violin plot showing FAP mRNA expression in the CMS subgroups in a large composite cohort consisting of 3232 primary tumours [1]. c Scatter plot showing the correlation between FAP mRNA expression and the CMS4-identifying geneset in the random forest CMS classifier (n = 143 genes; CMS4(RF)) in the CMS-3232 cohort. d Receiver-operating characteristic (ROC) curves showing the sensitivity and specificity of using FAP mRNA levels to distinguish between CMS4 and either all other tumours in the cohort (left curve) or all tumours except CMS1 (right curve). In clinical practice, the vast majority of CMS1 tumours are routinely identified with MSI tests. The area under the curve (AUC) values are shown, as a measure of diagnostic accuracy of the FAP test. e Kaplan–Meier curves showing relapse-free survival probabilities in tumour subgroups defined by FAP expression quartiles, n = 805 patients. f Pie charts showing the CMS distribution of tumour subgroups defined by FAP expression quartiles in the same tumour groups as in (e) with (left panel) and without (right panel) patients with CMS1 tumours. The vast majority of patients with CMS1 tumours will be identified by MSI testing as part of routine diagnostic procedures.

Fig. 2. FAP identifies CMS4 regions in heterogeneous tumours.

a Dot plot showing FAP expression levels in multiple [–5] distinct regions of the primary tumours of seven individual CRC patients (BOSS1 cohort) analysed by RNA sequencing. CMS status, identified by applying the random forest classifier, is colour-coded. b Scatter plot showing the correlation of FAP mRNA levels with CMS4 probabilities of each of the tumour regions across the entire dataset. The ROC curve is not shown because the discriminative ability of FAP mRNA levels (CMS4 versus all other regions) is perfect in this dataset (AUC = 1).

Fig. 3. FAP is expressed in myofibroblasts, proliferating endothelial cells, and a subset of tumour cells.

a Box plots showing FAP expression in single cells from colorectal tumours and adjacent normal colon tissue with annotated cellular identities [17]. b Distribution of FAP-expressing cells per cell type in normal and tumour tissue. c UMAP plot of all stromal cell types showing the distribution of FAP-positive cells for normal colon cells and each individual CMS. d Gene expression dot plot showing FAP expression distribution within the stromal subtypes for each CMS. Dot size indicates the number of stromal cells as a percentage of the total cells sequenced. Dot colour indicates the mean scaled FAP expression within the cell type. e Gene expression dot plot showing the gene expression of inflammatory CAF (iCAF) and myofibroblast CAF (myCAF) markers across different stromal cell types showing that FAP expression is highly selective for the myofibroblast cell type. Dot size indicates the number of stromal subtypes as a percentage of the total stromal cell fraction. Dot colour indicates the mean scaled expression within the cell type. f Correlation matrix showing the correlation between FAP expression, iCAF markers, myCAF markers and three distinct stromal subtypes in pancreatic cancer (C stroma, A stroma, F stroma) [24].

Fig. 4. CMS4 Peritoneal Metastases uniformly express high levels of FAP.

a Dot plot showing relative FAP expression levels in primary CRC tumours in the TCGA cohort (n = 592) versus primary tumours with peritoneal involvement (n = 35) and the peritoneal metastases derived from them (n = 59). b Dot plot showing FAP expression levels in peritoneal metastases and their matched primary tumours. CMS subtypes are colour-coded. Tissue type is annotated by shape. c Scatter plot showing the correlation of FAP mRNA levels with CMS4 probabilities of all tumours in the primary tumour/peritoneal metastasis cohort. d Immunohistochemistry for FAP expression in peritoneal metastasis. See Supplementary Fig. S5 for a comprehensive overview of FAP protein expression in peritoneal metastases. e Boxplot of FAP expression quantification on IHC slides of colorectal liver metastasis (CRLM, n = 21) and peritoneal metastasis (PM, n = 19). FAP expression is measured as the number of FAP-positive cells and corrected for tumour size using an EpCAM staining of the sequential slide. f Boxplot of the percentage of FAP-positive stromal cells of colorectal liver metastasis (CRLM, n = 21) and peritoneal metastasis (PM, n = 19). g Boxplot of the percentage of FAP-positive tumour cells of colorectal liver metastasis (CRLM, n = 21) and peritoneal metastasis (PM, n = 19).

Fig. 5. FAPI-PET detects peritoneal metastases.

a Immunohistochemistry for FAP on peritoneal metastases tissue that was resected one year prior to FAPI-PET imaging. b Immunohistochemistry of EpCAM-positive tumour cells in the peritoneum show a strong ZEB1 and HTR2B staining. c FAPI-PET image shows tracer accumulation in pelvic peritoneal depositions and along the abdominal peritoneal lining (white arrows). See also Supplementary Fig. S6. d FAPI-PET image (same coronal section, same scaling)) after 2 months of chemotherapy shows a reduction of FAPI tracer uptake in the pelvic lesion and the serosal lesion and a complete loss of diffuse FAPI tracer uptake along the peritoneum.

Fig. 6. Heterogeneous FAPI uptake in metastases outside the peritoneum.

FAPI-PET was performed in 15 CRC patients with distant extra-peritoneal metastases at various sites. The dot plot shows average SUVmax values for individual patients. Each dot represents a single lesion and each metastasis site is colour-coded.