Formalin-fixed, paraffin-embedded (FFPE) primary and/or metastatic tumor surgical biopsy specimens were prospectively obtained from 14 patients with PDAC, 14 patients with breast cancer, and 40 patients with other solid tumor malignancies as part of an IRB-approved Molecular Tumor Board study at the Inova Schar Cancer Institute (ISCI). Written informed consent was obtained from all patients. All experimental protocols involving human data in this study were in accordance with the Declaration of Helsinki.

Serial tissue thin sections (8 µm) were sectioned by microtome (median = 2 sections, range 2–8), placed onto polyethylene naphthalate (PEN) membrane slides (Leica Microsystems, Wetzlar, Germany), and stained with hematoxylin and eosin (H&E). One representative section (5 µm) per specimen block was mounted onto charged glass slides and H&E stained. Representative slides (the H&E section on glass and one LMD slide per specimen block) were scanned using the Aperio ScanScope XT slide scanner (Leica Microsystems). Micrographs of the representative sections on glass were reviewed by board-certified pathologist in Aperio eSlide Manager and Aperio ImageScope (Leica Microsystems) to annotate and confirm tumor areas of tumor epithelium for LMD enrichment.

LMD enrichment of tumor epithelium (5–10 µm2) was performed on the LMD7 (Leica Microsystems). The LMD enriched tumor samples for RPPA analysis were collected into cylindrical pressure cycling technology (PCT; Pressure Biosciences, Inc., South Easton, MA, USA) microtubes. A lysis/extraction buffer containing final concentrations of 10% tris(2-carboxyethyl)phosphine (TCEP, 50 mM; ThermoFisher Scientific, Inc., Waltham, MA, USA), 22.5% tris hydrochloride (Tris–HCl, 225 mM; ThermoFisher Scientific, Inc.), 4% sodium dodecyl sulfate (SDS, v/v; ThermoFisher Scientific, Inc.), 10% glycerol (v/v; ThermoFisher Scientific, Inc.), and 37.5% MilliQ type I water was added at a ratio of 2.5 µl buffer per mm2 LMD tissue. The microtubes were capped with PCT microcaps and placed into 0.5 ml PCR tubes, briefly centrifuged, and stored at − 80℃ until sample lysis. Representative images before and after LMD were captured using the Aperio AT2 slide scanner (Leica Microsystems).

The LMD enriched tumor samples were heated at 95 ℃ for 20 min, briefly centrifuged, and heated at 80 ℃ for 2 h. After heating, the samples were stored at 4 ℃ for one minute and then centrifuged for 18,000 rcf for 15 min. The lysate supernatants containing the denatured proteins were transferred to fresh low protein binding o-ring screw-top tubes (Agilent Technologies, Savage, MD, USA) and stored at − 80 ℃ until shipment on dry ice to Theralink Technologies, Inc. (Golden, CO, USA) for RPPA analysis.

RPPA analysis was performed as previously described [32] using a 32-marker, CLIA certified and CAP accredited commercial calibrated assay panel for examination of the total and phosphoprotein abundances of several targets with known relevance in solid tumors [29, 33]. The CLIA and CAP accredited RPPA assay is a calibrated immunoassay format that has been previously described [34,35,36], and uses a series of calibrators and controls along with a reference population data set of levels of expression of each of the 32 analytes including total and phosphorylated HER2, phosphorylated HER3, etc. derived from LMD enriched tumor epithelium over 400 FFPE breast cancer tumor biopsy samples (100 of each HER2/HR subtype). The HER2 expression status of the 400 tumors was derived from central lab based HER2 and HR IHC and/or FISH analysis and each experimental sample is interpolated/extrapolated to the reference population using the calibrator sets which are printed on every slide. Under CAP guidelines and in keeping with antibody validation criteria outlined by the RPPA society and now widely used by the RPPA community [37], the RPPA assay workflow includes extensive validation of all antibodies by western blotting, peptide competition and using positive and negative lysates derived from ligand stimulated cell lines. Of note, the CLIA IHC HER2 assay (NeoGenomics, Laboratories, Inc., Fort Myers, FL, USA) uses the Ventana 4B5 clone, which is validated for use as the breast and gastric companion diagnostic. The CLIA RPPA HER2 assay (Theralink Technologies, Inc.) uses the Invitrogen SP3 clone (rabbit recombinant antibody, (catalog MA5-14509, dilution 1:400). However, while the RPPA and IHC assays use different clones, the reference population used by the RPPA CLIA laboratory to qualify the RPPA assay results were derived from 400 breast tumors whose HER2 IHC was previously determined using the 4B5 clone and the results were > 95% correlated [7].

Briefly, sample lysates for RPPA analysis were printed onto nitrocellulose backed slides in technical triplicates alongside all requisite controls and calibrators. Prior to staining, nitrocellulose slides were pre-treated with ReBlot (MilliporeSigma, Rockville, MD, USA) and a blocking reagent (I-Block; Applied Biosystems, Waltham, MA, USA). Slides were incubated with primary antibodies for 30 min at room temperature, followed by incubation with a secondary antibody. Each staining run included a single negative control slide (only antibody diluent; no primary antibody). Protein detection was amplified via horseradish peroxidase mediated biotinyl tyramide deposition and visualized using a fluorescent probe (LI-COR Biosciences, Inc., Lincoln, NE, USA). Images of the stained RPPA slides were captured on an InnoScan 710-AL (Innopsys, Inc., Chicago, IL, USA). Individual spots were detected using the InnoScan software program, Mapix, and spots with non-standard morphology and/or staining were flagged and removed from analysis. The average median intensity value was calculated for each sample on the array for each analyte as well as the negative control slide. Background subtracted intensity values for each sample were fit to an analyte-specific calibrator and total protein normalized. Resulting values were compared to a population reference to determine patient sample percentile and quartile score. The RPPA data is provided in the report as both a continuous variable percentile-based output (0–100) as well as a quartile-based output (0–3) based on the relationship of the patient-specific value compared to the population reference. The 0–3 scoring output is meant to provide interpretive context to physicians who are most used to looking at CLIA IHC-based scoring outputs for protein biomarkers used in patient treatment decision making (e.g. PD-L1, HER2, etc.).

Clinical NGS (specifically, DNA-seq) analysis was performed using the Tempus xT 648-gene panel [38] (Tempus Labs, Inc., Chicago, IL, USA), prioritizing the same FFPE tissue specimen as was used for the LMD-RPPA analysis whenever possible. Matched specimens (ie. the same specimen block) were available for 46 patients for analysis by both RPPA and DNA-seq. Different specimen blocks were used for RPPA versus DNA-seq for 20 patients. Different specimen blocks were used for RPPA versus DNA-seq for 20 patients. The samples for DNA-seq were prepared from tissue sections with a median of 50% tumor cellularity after macrodissection/microdissection (range = 20–90% tumor cellularity).

Descriptive statistics was used to summarize the patients' demographics. Fisher's Exact tests were used to compare the RPPA abundances of HER2Total, pHER2Y1248, and pHER3Y1289 between groups. SAS software (v9.4; SAS Institute, Cary, NC, USA) was used for statistical analysis. A significance threshold of p < 0.05 was considered statistically significant. Boxplots of RPPA normalized intensity scores for HER2Total, pHER2Y1248, and pHER3Y1289 were generated using BoxPlotR [39].