4.14. Multiplex Quantitative Immunofluorescence (MQIF) Staining and Analysis

Three PDAC patient cohorts from the Stanford Medical Center, National Cancer Centre Singapore (SingHealth CIRB 2012/879/B), and National Cancer Institute Singapore (DSRB 2012/00939) were obtained. The collection of Singaporean patient sections was approved by the Institutional Review Board of Singapore. PDAC human tissue FFPE sections were evaluated and confirmed of diagnosis by board-certified pathologists at each of these 3 institutions for analysis. PDAC human tissue FFPE sections (adjacent “normal” = 6; patient = 50) were stained in a 6-plex staining protocol (Perkin Elmer, Melbourne, Australia) in this sequence: First cycle—primary antibody anti-CD68 1:150 (Dako, PG-M1), followed by secondary antibody anti-mouse HRP (Dako, K4001), and then TSA-Cy3 1:100 (Perkin Elmer Life Sciences, FP1170), with microwave treatment using citrate pH 6 Antigen Retrieval Buffer (00-4955-58, eBioscience, San Diego, CA, USA). Second cycle—primary antibody anti-GLUT1 1:250 (Thermo Fisher Scientific, RB-9052-P1), followed by secondary antibody anti-rabbit HRP (Dako, K4003), and then TSA-Cy3.5 1:100 (Perkin Elmer Life Sciences, FP1484), with microwave treatment using citrate pH 6 Antigen Retrieval Buffer (eBioscience, 00-4955-58). Third cycle—primary antibody used was anti-CD163 1:200 (ab188571, Abcam, Cambridge, MA, USA), followed by secondary antibody anti-rabbit HRP (Dako, K4003), and then TSA-Cy5.5 1:100 (Perkin Elmer Life Sciences, FP1486), with microwave treatment using citrate pH 6 Antigen Retrieval Buffer (eBioscience, 00-4955-58). Fourth cycle—primary antibody used was anti-HK2 1:150 (Abcam, 104836), followed by secondary antibody anti-mouse HRP (K4001, Dako, Carpinteria, CA, USA), and then TSA-FITC 1:100 (Perkin Elmer Life Sciences, FP1168), with microwave treatment using Tris pH 9 (Dako, S2367). Fifth cycle—primary antibody used was anti-HIF1α 1:100 (Novus Biologicals, NB100-105), followed by secondary antibody anti-mouse HRP (Dako, K4001), and then TSA-Cy5 1:100 (Perkin Elmer Life Sciences, FP1171), with microwave treatment using citrate pH 6 Antigen Retrieval Buffer (eBioscience, 00-4955-58). After the fifth cycle, the sections were counterstained with DAPI (Sigma Life Science, D9542-5MG).

At least 15 fields of view were imaged per patient section using the Mantra Workstation (Perkin Elmer). Objective signal counts, tumor/stroma compartments, and cell phenotypes were defined and quantified in these stained sections using multispectral fluorescence analysis and spectral unmixing with the InForm software (PerkinElmer). Inform was trained by board-certified pathologists in Singapore to recognize tumor versus the stroma compartment on each stained section, based on tissue structures, morphology, size, and characteristics of nuclei. Subsequently, all analysis was based on the “Entire Cell Total Normalized Counts” generated by the Inform software. Using in-house codes, we studied the association of a cell subset (for e.g., CD68+GLUT1+) to survival and stage progression, by computing the percentage composition of the cell subset on the basis of individual patients. Cells captured from all the tissue sections of a patient were assigned to binary cell subsets, the target, and the complement (CD68+GLUT1- in this example) subsets.

Survival analysis on manually selected cell subsets was analyzed using the nearest-template prediction (NTP) method. Subsets that were confounded on either gender or ethnicity were excluded and the remaining subsets were then thresholded at ten-percentile intervals and used in a series of log-rank survival tests. The percentiles showing the largest differences in the survival curves were selected for interpretation and further analyses. All analysis and visualizations were done using the R statistical language (v3.3.1).