This study paper optimized a pressure cycling technology (PCT)-sequential window acquisition of all theoretical fragment-ion spectra (SWATH) MS workflow for the analysis of formalin-fixed, paraffin-embedded (FFPE) specimens and compared results in case-matched FFPE and frozen specimens. Additionally, the paper investigated potential effects of storing FFPE punches versus section on peptide/protein profiles. The method was then used to compare the proteome of prostate tumor and benign FFPE specimens and identify prognostic markers for diffuse large B-cell lymphoma (DLBCL).
High-throughput proteomic analysis of FFPE tissue samples facilitates tumor stratification.
Zhu Y, Guo T, Zhang Q, Sun R, Wang B, Yi X, Wu Z, Gao H, Cai X, Ruan G, Zhu T, Xu C, Lou S, Yu X, Gillet L, Blattmann P, Saba K, Fankhauser CD, Schmid MB, Rutishauser D, Ljubicic J, Christiansen A, Fritz C, Rupp NJ, Poyet C, Rushing E, Weller M, Roth P, Haralambieva E, Hofer S, Chen C, Jochum W, Gao X, Teng X, Chen L, Zhong Q, Wild PJ, Aebersold R, Weiss T
Purpose
Conclusion
The optimized protocol included a 30 min incubation in 0.1 M Tris/HCl (pH 10.0) at 95°C, a 60 min liquid chromatography gradient and 48 variable SWATH windows. The peptide and protein profiles of case-matched FFPE and frozen specimens were very strongly correlated (r=0.91); in unsupervised clustering, the FFPE and frozen specimens were mixed and did not cluster by preservation method. Three proteins were found to be significantly upregulated (>2-fold, P<0.05) in frozen tumors relative to benign specimens. All three proteins were supported by the literature and two, AGR2 (4-fold higher in tumor) and POSTN (2-4 fold higher in tumor), were found to be positively correlated with patient survival (P = 0.008 and P=0.009). Immunohistochemistry (IHC) confirmed the increased expression of POSTN in tumor specimens and IHC levels were strongly correlated with MS data. The 3 proteins and an additional 21 proteins were found to be differentially regulated in FFPE tumor relative to benign specimens. Levels of 7 identified proteins (3 proteins in both specimen types and 4 proteins in solely FFPE) were capable of distinguishing tumor and benign specimens with an area under the curve (AUC) of 0.983 for frozen specimens and 0.977 for FFPE specimens. When FFPE specimens collected pre-2010 (mean storage duration of 7.1 years) were compared with those collected after 2010 (mean storage duration of 4.6 years), the total ion chromatograms differed, but recently collected and older specimens were still strongly correlated at the peptide and protein levels (r=0.9735 and r=0.9815, respectively). The proteome map was very strongly correlated between specimens stored as sections and punches (r=0.95). Unsupervised cluster analysis did not group specimens based on storage format or storage duration. The CV in protein abundance was slightly higher for punches than sections (4.13-5.04% versus 3.52-4.13%), which the authors attributed to the increased diversity of tissue regions in a punch. There was no observed effect of storage duration for FFPE punches or sections on the proteome.
Using their optimized method, the authors were able to develop a 12-protein panel that had AUCs of 1.0 and 0.991 for identifying prostate cancer in two different cohorts. Interestingly, the proteome as a whole was only modestly correlated between FFPE prostate tumors in the two cohorts (r=0.514), which the authors attribute, in part, to racial differences (Swiss versus Chinese). Additionally, the authors identified proteins that could partially separate low- and high-stage cancers in principal component analysis in each cohort separately, but both groups overlapped with intermediate-grade tumor specimens, and the overlapping proteins between cohorts did not fully separate low- and high-grade tumor specimens. The authors’ optimized method was successfully applied to identify 17 proteins that could be used to distinguish extracerebral DLBCL from primary central nervous system lymphomas (PCNSL), two of these proteins (ZAP70 and GFAP) had high AUCs (0.8847 and 0.8524, respectively). Five of the 17 proteins showed a trend toward predicting survival in the discovery cohort and another cohort, with IHC staining confirming the observed increased expression in eDLCBL.
Studies (2)
Purpose
This study optimized the duration of alkaline hydrolysis and the liquid chromatography gradient (30 min, 45 min or 60 min) and SWATH window configurations (8 tested) for proteomic analysis of FFPE tissue punches. Replicate FFPE tissue punches (0.6 mm x 3 mm) of a single resected prostate cancer specimen were deparaffinized in heptane and rehydrated before acidic hydrolysis in 0.1% formic acid for 30 min at 30°C. To test the effect of alkaline hydrolysis steps, specimens were then placed in 0.1 M Tris/HCl (pH 10.0) at 95°C for 15 min, 30 min or 45 min before pressure cycling technology (PCT) based lysis in lysis buffer, protein extraction and Lys-C digestion. Proteins and peptides were then separated using a liquid chromatography gradient length of 30 min, 45 min or 60 min and analyzed by mass spectrometry using 8 different SWATH window configurations (30, 39, 48, 58, 67, 75, 85 or 93 variable SWATH windows).
Summary of Findings
The peptide yield from FFPE prostate increased when the duration of the incubation in 0.1 M Tris/HCl (pH 10.0) at 95°C increased from 15 min to 30 min but did not increase further when extended to 45 min; the number of peptides identified was also highest when the incubation was 30 min. The authors report that the peptide yield obtained per mg of tissue was comparable to prior studies using frozen prostate specimens when the 30 min incubation was used. The number of peptides and proteins identified generally increased with liquid chromatography gradient length and was highest with a 48 variable SWATH window. When a 30 min gradient was used, 19% fewer peptides and 8% fewer proteins were identified than with a 60 min gradient.
Biospecimens
TissueProstatePreservation Types
FormalinDiagnoses
NeoplasticCarcinomaPlatforms
ProteinLC-MS or LC-MS/MSPre-analytical Factors
Classification Pre-Analytical Factor Value(s) Analyte Extraction and Purification Analyte purification Liquid chromatography gradient of 30 minLiquid chromatography gradient of 45 minLiquid chromatography gradient of 60 minAnalyte Extraction and Purification Hydrolyzation Incubation in 0.1 M Tris/HCl (pH 10.0) at 95°C for 15 minIncubation in 0.1 M Tris/HCl (pH 10.0) at 95°C for 30 minIncubation in 0.1 M Tris/HCl (pH 10.0) at 95°C for 45 minLC-MS or LC-MS/MS Specific Detection method 30, 39, 48, 58, 67, 75, 85 or 93 variable SWATH windows
Purpose
This study compared the (PCT)-SWATH mass spectrometry proteome of case-matched FFPE and frozen specimens and investigated the potential effects of FFPE specimen storage duration and format (punch versus section). The method was then used to compare the proteome of prostate tumor and benign specimens and identify prognostic markers for diffuse large B-cell lymphoma (DLBCL). To compare frozen and FFPE specimens, punches of resected case-matched frozen (~800 µg) and FFPE (~900 µg) prostate specimens from 24 patients with prostate cancer and 24 with benign prostate conditions were used (stored for 4-8 years before analysis). To investigate the effects of storage, case-matched 5 µm thick sections and punched tissue biopsies (1 x 1 x 0.5 mm) of three benign prostatic hyperplasia specimens that were stored for 1, 5 10 and 15 years were used (further details of storage not provided). To identify prostate cancer signatures, 3 punches of each of 58 prostate cancer and 10 benign prostate FFPE specimens (stored for 2 years) were analyzed. To see if this method could be used prognostically for DLBCL, 2-3 punches (three 0.5 x 0.6 x 3 mm) of 41 FFPE DLBCL tumors (19 with primary central nervous system lymphomas (PCNSL), 20 with extracerebral DLBCL (eDLBCL) and 2 with intravascular lymphoma) were used in an initial study; five 10 µm thick sections of 52 FFPE eDLBCL specimens from a second cohort were used for the prognostic analysis. Specimens were deparaffinized in heptane and rehydrated before acidic hydrolysis in 0.1% formic acid for 30 min at 30°C and alkaline hydrolysis in 0.1 M Tris/HCl (pH 10.0) at 95°C for 30 min. Specimens were lysed in lysis buffer, underwent extraction and Lys-C digestion using a pressure cycling technology (PCT) based method. Proteins and peptides were then separated using liquid chromatography gradient with a length of 60 min and were analyzed by mass spectrometry using 48 variable SWATH windows. Expression of select proteins was confirmed by immunohistochemistry.
Summary of Findings
The total ion chromatogram normalized for peptide mass was 15% higher for frozen than FFPE specimens, which the authors attributed to ion suppression and incomplete hydrolysis due to formalin fixation. However, peptide and protein profiles of case-matched FFPE and frozen specimens were very strongly correlated (r=0.91); additionally, FFPE and frozen specimens were mixed in unsupervised clustering. Three proteins were found to be significantly upregulated (>2-fold, P<0.05) in frozen tumors relative to benign specimens. All three of these proteins were supported by the literature and two, AGR2 (4-fold higher in tumor) and POSTN (2-4 fold higher in tumor), were positively correlated with patient survival (P = 0.008 and P=0.009). IHC confirmed the increased expression of POSTN in tumor specimens and IHC levels were strongly correlated with MS data. All three of these proteins were also upregulated in FFPE tumors relative to benign specimens, but an additional 8 proteins were upregulated and 13 were downregulated in FFPE tumors compared to benign specimens, the involvement of 4 of these proteins were supported by literature evidence. Levels of the 7 identified proteins (3 were present in both frozen and FFPE and 4 were only present in FFPE) were able to distinguish tumor and benign specimens with an area under the curve of 0.983 for frozen specimens and 0.977 for FFPE specimens. When FFPE specimens collected pre-2010 (mean storage duration of 7.1 years) were compared with those collected after 2010 (mean storage duration of 4.6 years), differences in the total ion chromatogram were observed between specimens, but new and old specimens were still strongly correlated at the peptide and protein level (r=0.9735 and r=0.9815, respectively). The proteome map was very strongly correlated between specimens stored as sections and punches (r=0.95). Unsupervised cluster analysis did not group specimens based on storage format or duration. The CV in protein abundance was slightly higher for punches than sections (4.13-5.04% versus 3.52-4.13%), which the authors attributed to the increased diversity of tissue types in a punch. There was no observed effect of storage duration for punches or sections on the proteome.
A total of 241 proteins were upregulated (fold change >2, adjusted P<0.05) and 89 downregulated in 58 prostate cancer specimens relative to 10 benign tumors, and this number increased when the threshold for significance was reduced. Importantly, these included the seven proteins identified when frozen and FFPE specimens were compared. However, the proteome as a whole was only modestly correlated between FFPE tumors in this cohort and the other cohort (r=0.514), which the authors attributed in part to racial differences (Swiss versus Chinese). The authors developed a 12-protein panel (the selected protein was differentially expressed in prostate cancer compared to benign specimens in both cohorts), which had an AUC of 1.0 and 0.991 for identifying prostate cancer in the two cohorts. Additionally, the authors identified proteins that could partially separate low- and high-stage cancers in principal component analysis in each cohort separately, but both groups overlapped with intermediate-grade tumor specimens, and the overlapping proteins between cohorts did not fully separate low- and high-grade tumor specimens. The authors’ method was successfully applied to identify 17 proteins that could be used to distinguish eDLBCL from PCNSL, two of which (ZAP70 and GFAP), had high AUCs (0.8847 and 0.8524, respectively). Five of the 17 proteins showed a trend toward predicting survival in the discovery cohort and another cohort, and IHC staining confirmed the increased expression in eDLCBL via MS.
Biospecimens
TissueLymph NodeTissueProstatePreservation Types
FrozenFormalinDiagnoses
NeoplasticLymphomaNeoplasticBenignNeoplasticCarcinomaPlatforms
ProteinLC-MS or LC-MS/MSProteinImmunohistochemistryPre-analytical Factors
Classification Pre-Analytical Factor Value(s) Preaquisition Diagnosis/ patient condition Benign prostate tumorPCNSLProstate CancereDLBCLPreaquisition Patient race ChineseSwissPreaquisition Prognostic factor High stageIntermediate stageLow stageBiospecimen Preservation Type of fixation/preservation Formalin (buffered)FrozenStorage Storage conditions PunchSectionStorage Storage duration 1 year10 year15 years5 yearPost- 2010 (mean storage 4.6 years)Pre-2010 (mean storage 7.1 years)