NEW YORK (GenomeWeb) – A team led by researchers from the Pacific Northwest National Laboratory and the National Cancer Institute's Clinical Proteomic Tumor Analysis Consortium (CPTAC) has found that formalin-fixed, paraffin-embedded (FFPE) tumor tissue specimens are likely suitable for mass spec-based proteomic work.
Detailed in a study published this month in Clinical Proteomics, the findings indicate that FFPE tissue samples could prove particularly useful for verification experiments where researchers are looking to narrow biomarker candidates identified in discovery work to smaller validation panels, said Karin Rodland, senior author on the paper and director of the Precision Medicine Innovation CoLaboratory at PNNL as well as an associate professor at Oregon Health & Science University.
This is significant in that many banked samples available for research use have been stored as FFPE specimens, Rodland said. However, some FFPE repositories have been reluctant to provide samples for proteomic experiments due to questions about whether FFPE specimens could be effectively analyzed using conventional shotgun mass spec workflows.
The challenge in analyzing FFPE tissues has been twofold, Rodland said. The first issue is extracting proteins from the samples at sufficiently high yields for analysis. The second issue is the fact that formalin fixing causes protein crosslinking, which can make it difficult to get good mass spec identifications.
"There have been commercial kits available for 12 to 15 years for extracting proteins out of FFPE blocks, and on the face of it, the yields of protein from the FFPE blocks is not that bad," she said. "But with the mass spec technologies of 12 to 15 years ago, the rate of identification was very low. You just didn't get good [proteome] coverage out of FFPE blocks. And the assumption was that the formalin crosslinking was causing you to lose identifications."
However, improvements in mass spec technology have provided instruments with higher sensitivity and better resolution that are capable of working with smaller amounts of sample, Rodland noted. The search software used to make peptide identifications from mass spectra has also improved.
These advances have raised the possibility that mass spec approaches might now be more effective for analyzing FFPE tissue than they were in the past, Rodland said. "So, basically, the NCI came to us … and the concept was, let's just try it [with modern equipment] to see how deep we can go and how good the coverage is that we can get."
Rodland and her colleagues analyzed 60 patient samples taken from the NCI's Surveillance, Epidemiology, and End Results (SEER) residual tissue repositories, which contain samples from more than 100,000 cancer patients, along with detailed demographic information, data on tumor characteristics, treatment, survival, and cause of death. The 60 samples ranged in their time of storage from seven to 32 years.
The researchers used 10-plex TMT labeling and divided each sample into six fractions, each of which they then ran on a 100-minute nanoLC gradient upfront of analysis on a Thermo Fisher Scientific Q-Exactive Plus instrument. For phosphopeptide analysis, they used IMAC enrichment. They found that all 60 samples provided sufficient material for proteome-wide protein expression analysis and 18 of the 60 samples provided enough material for phosphopeptide work.
The experiment identified and quantified a total of 8,582 proteins and 8,073 phosphopeptides across the SEER sample set, indicating that FFPE tissue is amenable to mass spec proteomics analysis. Protein identifications were reduced compared to the identifications possible in comparable optimal cutting temperature (OCT) compound-embedded specimens. Compared to OCT samples, "peptide, protein, and phosphopeptide identifications were reduced by 50, 20, and 76 percent, respectively," the authors noted.
The decrease in identifications was not correlated with the time of storage of an FFPE sample, Rodland noted. However, certain classes of proteins were more affected than others.
For instance, DNA-binding proteins and chromatin proteins were underrepresented in FFPE samples compared to OCT samples.
"That makes sense, because in those cases, the formalin fixation would have fixed the proteins to the DNA and probably made them being less effectively extracted," Rodland said, adding that "further due diligence" was needed to investigate whether any other classes of proteins were underrepresented in the FFPE samples.
The fact that there does appear to be a bias against certain kinds of proteins suggests FFPE samples may not be well suited for initial discovery work, Rodland said, "because you don't know what's missing in the FFPE block compared to a flash-frozen tissue."
"But once you have, say, a preliminary panel that might have 100 candidates, and you want to narrow that down to 10 or fewer candidates, and you want to verify the performance of your biomarker panel in a statistically significant number of patient samples with clinical outcomes data, [FFPE] is extremely well fitted to that particular application and will be a huge resource," she said.
As Rodland and her co-authors noted, much proteomic biomarker work has been done in fresh frozen or OCT samples but these are expensive and difficult to obtain, particularly in the large numbers desired to ensure high-quality statistics. Additionally, because these samples have in many cases been taken more recently, they may not have the long-term treatment and outcome data that is available for banked FFPE tissues like those provided by the SEER repositories.
Rodland said that to date, the SEER repositories have been reluctant to release their FFPE samples for mass spec-based proteomics research.
They "had to be convinced that there was utility in doing that," she said. "It's a precious resource, and you're not going to loan it out to people if they're not going to get reliable, reproducible results out of it. And the feedback I'm getting is that this study did convince that community that if they released some of their FFPE materials for proteomic studies, there will be good and valid results out of those studies that would be useful to the community and a good use of the resource."