Researchers at the Max Planck Institute for Molecular Genetics in Berlin have shown that they can use next-generation sequencing to analyze DNA from formalin-fixed paraffin-embedded samples stored for as long as 18 years, and that they can detect several types of genetic variation, including copy number alterations and certain SNPs, in FFPE tumor samples.
The ability to sequence FFPE tumor samples would open up large tumor collections from clinical trials for sequence analysis and might help researchers identify genetic alterations that are linked to disease progression or response to cancer therapy. However, to date sequencing such samples has been difficult because the formaldehyde compromises the nucleic acids in the tissue.
"The majority of human tumor samples are currently stored as FFPE tumors, so methods to permit the large-scale analysis of such samples will be of great benefit," said Victor Velculescu, director of cancer genetics at the Ludwig Center at Johns Hopkins University, in an e-mail message. Velculescu was not involved in the study.
FFPE tumors, because they are readily available in large numbers from hospital tissue banks, are especially useful for researchers who want to analyze rare tumors, or a subset of patients with certain attributes, for example responders and non-responders to therapy, or patients with advanced disease, according to Michal Schweiger, a group leader in the department of vertebrate genomics at the MPI and the first author of the study, which was published in PLoS ONE earlier this month.
Another reason why it would be helpful to be able to sequence FFPE tumors — rather than snap-frozen tumors that are used in many tumor-sequencing studies today — is that formalin fixing provides an easy way to collect new samples that is compatible with work in the clinic, said Schweiger, who is also a resident in medical clinical genetics at the Charité Hospital in Berlin.
"Doctors don't have time to think about how to conserve the tissue they take out," she said. "Everything has to be easy and fast. They can't deal with liquid nitrogen or other difficult conservation techniques. This technique is very easy — they just have to take it out and put it into a solution."
However, there is "considerable variability in sample fixation and storage conditions between different pathology departments," cautioned Mike Stratton, deputy director of the Wellcome Trust Sanger Institute who directs the institute's cancer genome project, and "the impact of this variation on the state of the DNA, and hence the practicality of large-scale use of FFPE material will have to be assessed exhaustively before it can be seriously entertained." In particular, researchers will need to analyze what fraction of FFPE samples can be turned into usable sequencing libraries, and what the data's sequence error rate is.
FFPE samples have posed problems for genomic analyses because they contain fragmented DNA and crosslinks between DNA, RNA, and proteins, so that molecular techniques such as PCR or in situ hybridization often fail. "Even conventional PCR-based resequencing of DNA from FFPE samples is fraught with artifacts and unpredictability," Stratton said.
But "since the next-generation sequencing techniques rely on short fragments, it was quite tempting to speculate that it would be possible" to use them on FFPE samples, according to Schweiger.
For their study, the MPI researchers divided a normal breast cancer sample into several parts, which they stored by either snap-freezing or formaldehyde fixation, using various ischemic and fixation times.
They then sequenced the samples on an Illumina Genome Analyzer with 36-base or 40-base unpaired reads, generating on the order of 3 million reads per sample, and aligned these reads to the NCBI human reference genome. They found that the number of reads and their distribution across the genome was similar for snap-frozen and FFPE tissues, "proving that fresh FFPE tissues can be used for high-throughput resequencing approaches," according to the study.
[ pagebreak ]
They also sequenced snap-frozen and FFPE-stored tissue samples that were 14 and 18 years old and found that tissue storage time "has a minor influence on sequence quality."
The researchers then sequenced two breast tumor samples — one snap-frozen, the other FFPE-preserved — and analyzed them for copy-number alterations as well as nucleotide exchange rates. The results showed that "the data obtained from both preparation methods is of comparable quality," they wrote.
According to Schweiger, it remains to be seen if paired-end sequencing is also possible with FFPE tissue, given the short DNA fragments it yields. "It might be difficult," she said, although her team has not explored it yet.
Johns Hopkins' Velculescu said the study is "a very nice example" of the use of next-generation sequencing methods to analyze copy number changes in FFPE tumors. However, "it would be important in the future to see how well somatic point mutations can be detected in FFPE tumors using these approaches," he added.
Based on their results, Schweiger and colleagues have recently embarked on sequencing tumor samples from clinical trials. Their current study, which involves sequencing approximately 100 samples, focuses mainly on gastrointestinal tumors, aiming to define which ones metastasize, and whether responders to chemotherapy can be separated from non-responders based on genomic attributes.
The ability to select FFPE samples from the pathology department means that researchers can select samples according to precise criteria, for example age or gender of the patient, tumor stage, or other underlying diseases. "You can set up a much better cohort of patients to analyze," Schweiger said. "That makes it much easier to do the statistical analysis later — that's a huge advantage."
Though useful for analyses of existing samples from clinical trials, the study may have limited significance for ongoing large-scale tumor-sequencing projects. The International Cancer Genome Consortium, for example, will largely focus on frozen samples because it wants to combine analyses of DNA variation, DNA methylation, and the transcriptome in the same sample. "But it is possible that for rare tumors, archived tissues will be needed," Tom Hudson, president and scientific director of the Ontario Institute for Cancer Research, told In Sequence by e-mail.
According to Hudson, the ability to sequence FFPE samples from past clinical trials "will undoubtedly be useful to discover new correlations between clinical outcome (and drug responses) and tumor mutations," and "the current study will help realize this potential."