NEW YORK (GenomeWeb News) – A study appearing in PLOS One last month supports the notion that DNA and RNA sequencing from formalin-fixed, paraffin-embedded (FFPE) samples can provide informative polymorphism and gene expression information.
Even so, the analysis also emphasized the importance of critiquing nucleic acid extraction and library preparation methods and understanding the systematic errors and biases associated with them.
"The main message to take home is that you should test different library kits and purification kits, because they differ quite a lot," the study's first author Jakob Hedegaard, a molecular medicine researcher at Aarhus University, told In Sequence.
Hedegaard and colleagues from Aarhus University, AROS Applied Biotechnology, and CLC Bio did RNA sequencing and exome sequencing on both FFPE and fresh-frozen samples from three normal human tissues and three cancerous tissue types. The head-to-head methods comparison was "quite a natural thing to do, given all the [FFPE] samples lying around in different archives," Hedegaard said.
"We want to be able to go back to available samples and use them for clinical studies," he said, noting that it would also be beneficial to refine FFPE sequencing protocols to the point where it's no longer necessary to collect new fresh-frozen tissue samples from patients.
In general, though, DNA and RNA tend to be trickier to extract from FFPE than from fresh-frozen samples, the study's authors noted, explaining that "nucleic acids are more difficult to extract from FFPE tissue because of the need to remove the paraffin and to counteract covalent protein-DNA interactions that result from the fixation process."
There are other potential complications when dealing with FFPE samples that can stem from subtle differences in the protocol used to prepare tissues and fix DNA that can introduce nucleic acid degradation or chemical modification, they noted.
On the heels of other published studies examining sequencing methods and workflows associated with FFPE and other samples with low nucleic acid quality or quantity, the team decided to examine methods for sequencing RNA and protein-coding DNA sequences from FFPE samples representing bladder, prostate, and colon cancers, and normal colon, liver, and tonsil tissue.
The DNA and RNA extraction and library prep methods described in the study are not representative of the full suite of techniques available nowadays, Hedegaard said, noting that the study began a few years ago.
After doing a survey of the commercially available nucleic acid extraction and sequencing library preparation methods on the market at the time, he and his colleagues did more extensive testing to compare several FFPE DNA and RNA extraction methods in the context of exome sequencing and RNA sequencing.
They also did a head-to-head comparison of sequence data generated from FFPE samples from four of the tissue types that had been stored for different stretches of time, up to around 20 years.
Starting from 23 fresh-frozen tissue samples, 35 FFPE samples, and 38 paired fresh-frozen and FFPE sets representing the same six tissue types, the team used kits from Qiagen, Macherey-Nagel, AmpTec, and Ambion (now part of Thermo Fisher Scientific) to extract DNA or RNA from the fresh-frozen and FFPE samples.
For FFPE DNA extraction steps, for example, the researchers tested Qiagen's QIAmp DNA FFPE tissue kit and the Nucleospin FFPE DNA extraction kit from Machery-Nagel. RNA in the FFPE samples was extracted using Qiagen's miRNeasy FFPE, AmpTec's ExpressArt FFPE RNAready, and/or Machery-Nagel's Nucleospin FFPE RNA kits.
Following initial experiments pointing to higher RNA yields and quality with the Qiagen and AmpTec FFPE kits, the researchers turned to those two extraction methods for their follow-up FFPE RNA sequencing analyses.
For FFPE exome sequencing experiments, they prepared Illumina sequencing libraries with Illumina's TruSeq kit before sequencing protein-coding sequences with the Illumina HiSeq 2000. The RNA sequencing libraries were prepared using ScriptSeq kits from Illumina's Epicentre and analyzed using a CLC Bio genomics pipeline.
Results from the analyses indicated that it can be especially tricky to prepare exome sequencing libraries from FFPE DNA.
In contrast to RNA sequencing libraries from FFPE samples, which the team successfully prepared for all of the fresh-frozen and FFPE samples, the exome sequencing library prep produced usable libraries for just 29.5 percent of the FFPE samples.
Problems with exome library preparation appeared to stem primarily from amplification problems after the sequencing adaptor ligation step in the library prep process that can be traced back to DNA degradation and/or modifications caused by FFPE sample fixation and storage, the study authors noted.
Hedegaard noted that other published studies did not detect amplification difficulties with FFPE DNA, though those were not based on FFPE samples as old as those described in the current analysis, which went back two decades in some cases.
The genomic DNA also appeared more prone to increasing degradation in FFPE samples stored for longer and longer periods of time. In addition, the insert sizes in exome sequencing libraries prepared for FFPE samples tended to be shorter than those in libraries from fresh-frozen samples, particularly for the oldest FFPE samples.
Even so, the group saw between 70 and 80 percent concordance between variants detected by sequencing the exomes of fresh-frozen samples and those found by exome sequencing on matched FFPE samples stored for three years or less.
On the RNA sequencing side, the team saw similar gene expression profiles and fresh-frozen samples and matched FFPE samples — even for FFPE samples stored up to two decades — though FFPE samples consistently displayed differential expression at almost 1,500 genes.
"We observed this systematic set of affected genes that also seemed to be affected in the same direction across different tissues," Hedegaard said.
The reason for that pattern remains to be determined, he noted, as does the reduced representation of protein-coding exon sequences that the team saw in FFPE samples, which generated far fewer of those exonic reads than reads coming from introns.
The latter findings hint that a yet unknown FFPE-related process may leach mature RNA transcripts from the cell while leaving more immature transcripts relatively unscathed, Hedegaard noted.
Because transcripts became more and more difficult to map to reference sequences when older FFPE samples were used, the study's authors suggested that sequencing methods that generate longer RNA sequence reads may be beneficial when dealing with FFPE samples.
The current analysis was done using Illumina instruments, but Hedegaard said he would expect to see the same results with other sequencing platforms, though some library prep kits could offer advantages for dealing with FFPE DNA or RNA.
For instance, FFPE-related problems DNA amplification problems could theoretically be diminished by kits designed for dealing with low quality samples, such as the Kapa Biosystems Hyperprep kit or Illumina kits aimed at low-quality inputs.
The group is continuing to evaluate new extraction and sample prep kits as they become available, including purification kits designed to simultaneously extract DNA and RNA from the same sample. The researchers are also working to fine tune their informatics approach in an effort to make analysis more efficient.