matched pairs analysis using McNemar test was
matched-pairs analysis using McNemar test was carried out to test stroma expression differences. Pb.05 was considered statistically significant. Analyses were carried out using SPSS v24 (IBM, New York, NY) and SAS, v9.2 (SAS Institute, Cary, NC).
3.1. DepArray analysis reveals a small population of CK+/VIM+ co-expressing cells
Three different tissue areas were investigated for single and double-positive cells: the tumor center, tumor stroma and areas of tumor Relebactam at the invasion front. These areas were punched out from the corresponding tissue blocks of five cases and sent separately for DepArray analysis. The number of investigated double or single positive cells in each of the three compartments (center: CK only and CK/VIM; stroma: VIM only and CK/VIM; tumor budding: CK only and CK/VIM) was quantified.
Out of 5 samples, disaggregation was successful in four and of these, all could be successfully analyzed in the center, stroma and tumor budding regions except in sample 4 budding region. In areas of the tumor center and the stroma, only CK+ and VIM + single cells were identified, respectively. Tumor budding regions from three samples contained CK+ single cells while two different samples showed CK+/VIM+ double-positive cells. The frequency of CK+/ VIM+ double-positive cells was 10/337 (2.9%) in sample 1 and 49/173 (22%) in sample 3. Whole tissue slides of these cases stained for CK and VIM were also examined by fluorescent and confocal microscopy and found to contain CK+/VIM+ single cells or small tumor cell clusters (Fig. 1).
3.2. Next-generation sequencing of double-positive tumor cells
In order to verify that the double-positive cells were of tumor origin, we performed next-generation sequencing using an ion Torrent 50 gene cancer panel. Due to the low input of DNA, only one sample could be successfully sequenced. The mutational pro-file of the double-positive cells was compared to that of the cor-responding tumor center and to the proximal stroma (Table 1).
Four mutations were identified in the double-positive population: BRAFV600E missense, PTEN deletion, TP53 missense GNV and a second TP53 PNR missense mutation. Importantly, the mutated allele frequency was 100% for BRAF V600E in the CK+ cells of the tumor center and 100% in both the CK+ only and CK+/VIM+ cells of the tumor budding population but 0% in the proximal tumor stroma. Similar results were found for PTEN deletion and TP53 GNV with the same aberrations found in the tumor center as well as both CK+ and CK+/VIM+ tumor budding area cell collection, but not in the tumor stroma. The final mutation is TP53 PNR was mutated in all tissue areas suggesting its non-sporadic origin.
3.3. FACS sorting of single and double-positive tumor cells
Six tumor samples were sorted for CK+, VIM+ and CK+/VIM+ double-positive cells. On average the tissue contained 26% CK+ and 42% VIM+ cells (Fig. 2). To ensure maximal purity and exclude the possibility of doublets, the VIM+/CK+ double-positive cells were subsequently re-sorted. Using this protocol, of all CK-positive cells sorted, approximately 1.5% were determined to be double-positive for CK+/VIM+ (Table 2).
Fig. 1 Immunofluorescence staining of colorectal cancer using pan-CK (AE1/AE3) (red) and VIM (green). Red arrows point to CK/VIM co-expressing tumor buds.
Table 1 Mutational profile of double-positive cells in comparison to tumor center and stromal compartments following next-generation sequencing
Gene Ref Alt Stroma
Tumor center Mutation Variant
CK+VIM− CK+VIM+ CK+VIM−
PTEN G A 0%
Lost Lost Lost Gene deletion
Fig. 2 Gating in FSC-A/SSC-A (A, P1) was performed to remove debris, in FSC-A/FSC-H to exclude fused cells (B, P2), and DAPI staining was used to help select intact cells containing both a nucleus and cytoplasm (C, P3). In panel D, CK+ (P4), VIM+ (P5), and CK+VIM+ cell populations are shown.