MtFFPECleaner: A high‐fidelity machine learning approach for artifact removal in high‐throughput mitochondrial DNA sequencing data of archival formalin‐fixed paraffin‐embedded specimens

Background: The current project aims to explore the possibility to use formalin fixed paraffin embedded (FFPE) samples for exome sequencing, which would open up a large collection of tissue samples for molecular studies. Usage of formalin is known to degrade and modify the DNA both during treatment and storage resulting in problems in downstream molecular analyses. The aim of the present study is to evaluate exome sequencing of FFPE samples by comparing data from FFPE normal tissue to corresponding snap frozen (SF) blood controls with focus on data output such as amount of data removed due to low sequencing quality, amount of data mapped to the genome, GC-content, and duplicate levels. Methods: In the current study we used the western Swedish biobank of neural tumors archived at the Sahlgrenska Hospital, Gothenburg. We performed exome sequencing of FFPE normal adrenal tissue samples and corresponding SF blood samples from five patients diagnosed with pheochromocytoma/paraganglioma. Two of the patients had known heterozygous germline mutations and were included to determine if these mutations could be found in both sample types. Libraries for SF samples were prepared according to protocol using Agilent Technologies SureSelect Human All Exon 50 Mb library prep kit. FFPE samples were prepared with slight modifications in the protocol, with additional PCR cycles used to increase amplification. 75 bp paired end reads were generated on Illumina HiScan SQ according to manufacturer's protocol (v.3). Results: Between 92-99% of the raw data were kept after removal of data due to low sequencing quality independent of sample type. For the SF samples 99-100% of the trimmed data mapped to the genome, the number for FFPE samples were slightly lower ranging from 87-98%. The GC content in the exome enriched DNA from FFPE and SF tissue were similar, ranging between 45-51%. Looking at the mapped data we found higher levels of duplicates in the FFPE samples compared to the SF samples, 19-78% vs. 5-15%. For one of the two patients with a known germline mutation the variant was detected at approximately the same frequency in both FFPE and SF sample (39% in FFPE vs. 44% in SF). For the other patient the variant were found at higher frequency in the FFPE sample compared to SF sample (80% vs. 50%). Conclusions: We saw no difference in amount data removed due to low sequencing quality between the five FFPE and SF samples evaluated in this study. Also, we saw only slightly lower mapping frequencies for the FFPE samples. The GC-content for both FFPE and SF samples were within the range of what's expected for exome sequencing data for libraries prepared with this kit and sequenced on Illumina HiScan SQ platform. Due to the additional PCR cycles used for FFPE samples we saw higher duplicate levels in these samples, suggesting that deeper sequencing is necessary for FFPE samples to get the same amount of unique reads as for the SF samples. A probable explanation to the higher allele frequency of the variant observed in the FFPE sample from one of the patients with a known germline mutation is the presence of contaminant tumor cells in the normal tissue sample, resulting in a shifted allele frequency. To conclude, SF samples are to prefer for exome sequencing. However, if SF material is not available FFPE tissue is a good alternative to increase the sample cohort when studying rare and complex diseases. Citation Format: Annica Wilzen, Heidi Ottesen, Anna Larsson, Bo Wangberg, Andreas Muth, Ola Nilsson, Frida Abel. Exome sequencing of FFPE material: An evaluation. [abstract]. In: Proceedings of the AACR Special Conference on Pediatric Cancer at the Crossroads: Translating Discovery into Improved Outcomes; Nov 3-6, 2013; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2013;74(20 Suppl):Abstract nr A11.

Comparison Of Single Nucleotide Mutations (SNVs) and Copy Number Variants (CNVs) Detection In Formalin Fixed Paraffin Embedded (FFPE) and Paired Frozen Tumor Tissues Using Target Capture and Sequencing Approach

Evaluation and Validation of Total RNA Extraction Methods for MicroRNA Expression Analyses in Formalin-Fixed, Paraffin-Embedded Tissues

Clinical Validation of Somatic Mutation Detection by the OncoScan CNV Plus Assay

BackgroundArchived formalin fixed paraffin embedded (FFPE) samples are valuable clinical resources to examine clinically relevant morphology features and also to study genetic changes. However, DNA quality and quantity of FFPE samples are often sub-optimal, and resulting NGS-based genetics variant detections are prone to false positives. Evaluations of wet-lab and bioinformatics approaches are needed to optimize variant detection from FFPE samples.ResultsAs a pilot study, we designed within-subject triplicate samples of DNA derived from paired FFPE and fresh frozen breast tissues to highlight FFPE-specific artifacts. For FFPE samples, we tested two FFPE DNA extraction methods to determine impact of wet-lab procedures on variant calling: QIAGEN QIAamp DNA Mini Kit (“QA”), and QIAGEN GeneRead DNA FFPE Kit (“QGR”). We also used negative-control (NA12891) and positive control samples (Horizon Discovery Reference Standard FFPE). All DNA sample libraries were prepared for NGS according to the QIAseq Human Breast Cancer Targeted DNA Panel protocol and sequenced on the HiSeq 4000. Variant calling and filtering were performed using QIAGEN Gene Globe Data Portal. Detailed variant concordance comparisons and mutational signature analysis were performed to investigate effects of FFPE samples compared to paired fresh frozen samples, along with different DNA extraction methods.In this study, we found that five times or more variants were called with FFPE samples, compared to their paired fresh-frozen tissue samples even after applying molecular barcoding error-correction and default bioinformatics filtering recommended by the vendor. We also found that QGR as an optimized FFPE-DNA extraction approach leads to much fewer discordant variants between paired fresh frozen and FFPE samples. Approximately 92% of the uniquely called FFPE variants were of low allelic frequency range (< 5%), and collectively shared a “C > T|G > A” mutational signature known to be representative of FFPE artifacts resulting from cytosine deamination. Based on control samples and FFPE-frozen replicates, we derived an effective filtering strategy with associated empirical false-discovery estimates.ConclusionsThrough this study, we demonstrated feasibility of calling and filtering genetic variants from FFPE tissue samples using a combined strategy with molecular barcodes, optimized DNA extraction, and bioinformatics methods incorporating genomics context such as mutational signature and variant allelic frequency.

Archived formalin-fixed paraffin-embedded (FFPE) specimens represent excellent resources for biomarker discovery, but it has been a major challenge to study gene expression in these samples due to mRNA degradation and modification during fixation and processing. We have developed methods for small RNA and whole transcriptome analysis of FFPE samples on the Ion Torrent PGM™ System using Ion total RNA-Seq Kit v2 and RiboMinus Eukaryote System v2. Paired FFPE and fresh-frozen RNA samples from lung adenocarcinoma tissue were used in this study. Small RNA libraries were prepared using the Ion Total RNA-Seq kit v2 with modified protocol, followed by sequencing on PGM™ system. Results show that small RNA extracted from FFPE sample was successfully converted to small RNA library. Although percentage of reads mapped to miRBase was slightly lower when compared to fresh-frozen sample, the overall miRBase coverage and detection sensitivity were comparable. For whole transcriptome analysis, RiboMinus Eukaryote System v2 kit was used to deplete rRNA from FFPE samples, followed by library construction using the Ion Total RNA-Seq kit v2 with a modified protocol. Significant reduction in rRNA reads was observed after sequencing. High RefSeq correlations were observed between fresh-frozen and FFPE samples. In conclusion, a full spectrum of gene expression data, from both small RNA and whole transcriptome, were generated from FFPE samples using Ion Total RNA-Seq kit v2 with modified protocols. High correlations were observed between paired fresh-frozen and FFPE samples, demonstrating the reliability of the modified workflow for gene expression profiling on FFPE samples. This study provides feasibility for systematic gene expression analysis on FFPE samples from cancer and other diseases. Citation Format: Jian Gu. Expression profiling of paired formalin-fixed, paraffin-embedded (FFPE) and fresh-frozen tissue samples on Ion Torrent PGMTM. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 4127. doi:10.1158/1538-7445.AM2013-4127

Introduction: The early molecular detection of the colorectal dysplasia-carcinoma transition may enhance the classification of colonic tissue samples. Recently, a colorectal cancer-specific transcript set including COL12A1, CXCL2, CA7, Il1B, MMP3, IL8 was identified on colonic biopsy samples using whole genomic microarrays (WG MA). Aims: Our aim was to evaluate the applicability of these markers on independent biopsy and also on fresh frozen and formalin-fixed, paraffin-embedded (FFPE) tissue specimens from healthy and CRC patients by expression arrays and RT-PCR. Furthermore, automated RNA isolation was introduced and evaluated for increased sample number in one run. Materials and methods: Total RNA from 3 normal and 3 CRC FFPE specimens were analyzed by using Affymetrix U133Plus 2.0 WG expression arrays. Total RNA was isolated from 30 biopsy samples stored in RNALater (15 CRC, 15 normal), 20 fresh frozen surgically removed colonic tissues (10 CRC, 10 adjacent normal) and 60 FFPE (30 CRC, 30 adjacent normal) samples with the automated MagNA Pure 96 Cellular RNA Large volume kit and reverse transcription was done using Transcriptor First Strand cDNA Synthesis Kit (Roche). Gene expression analysis was performed with real-time PCR using RealTime ready assays and the LightCycler 480 system (Roche). Results: The FFPE specimens could be clustered correctly in 100% by the results of the microarray analysis. According to the gene expression levels of the set of 11 transcripts, the biopsy samples could be distinguished by 93,3% sensitivity and 93,3% specificity by RT-PCR. The discriminatory power of the marker set was proved to be high also on fresh frozen and FFPE samples (sens:&amp;gt;90 %, spec: &amp;gt;80 %). The automated RNA isolation technique has not influenced the classification power of the markers, but decreased the workaround time by 100% (4 hours to 2 hours). Conclusion: The analyzed set of markers in an automated environment could correctly characterize the healthy and the tumorous colonic tissue samples on biopsy, fresh frozen and also on FFPE tissue samples by MAs and RT-PCR. On the basis of these results, these markers can enhance the automated, early and differential detection of colorectal cancer. Citation Format: Bela Molnar, Alexandra Kalmar, Barnabás Wichmann, Orsolya Galamb, Sandor Spisak, Ferenc Sipos, Kinga Tóth, Katalin Leiszter, Arpad V. Patai, Andrea Schöller, Zsolt Tulassay. mRNA biomarkers of colorectal cancer development tested on biopsy, fresh frozen and formalin-fixed, paraffin embedded (FFPE) specimens in an automated workflow. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 3506. doi:10.1158/1538-7445.AM2013-3506

Introduction: Personalizing cancer care relies on accurate detection of actionable genomic aberrations in tumor cells. Conventionally, this strategy relies on analysis of primary tumor samples, which are often temporally and biologically distinct from recurrent, metastatic or treatment-resistant disease. As an alternative, CTCs offer real-time cancer tissue for analysis that may more accurately represent the current state of a patient's disease. In this pilot, we used CTCs as source material for targeted NGS across a range of malignancies. Methods: Under IRB approval, blood samples from patients with advanced cancer were labeled with EpCAM ferrofluid and placed into the LiquidBiopsy® platform (Cynvenio Biosystems, Inc.), an immunoaffinity-based microfluidic device tailored to query genomic events. CTCs were identified by CK, CD45 and DAPI expression. A matched WBC pellet served as a control representing germline sequence. Amplicon libraries were generated using Life Technologies AmpliSeq 2.0 and sequenced with an Ion Torrent sequencer. When available, matched formalin fixed paraffin embedded (FFPE) primary tumor tissue from the same patient was analyzed in parallel. Somatic single nucleotide variants (SNV) present in CTCs or FFPE samples but not in WBC were identified. Results: CTCs were detected in 18 of 19 patients with advanced prostate (8), breast (6), renal cell (2), bladder (1), lung (1) and rectal (1) cancer (CTC median 54, range 15-421). Germline SNPs were consistently detected across WBC, CTC and FFPE samples. Significant SNVs (occurring in &amp;gt; 1% of DNA in a sample) were found in 6 of 18 patient CTC samples (range 1-6 SNVs/sample, frequency 1.1%-11.9% with 620X-14,422X sequence coverage depth). Numerous SNVs were identified in all 9 matched primary tumor FFPE specimens but did not correlate with the SNVs identified in the CTCs. Conclusion: This pilot demonstrates the feasibility of using CTCs as a real-time disease relevant substrate for NGS to identify personalized genomic targets. A high number of CTCs were detectable across malignancies, and CTC germline variants correlated with matched WBC controls. Cancer relevant SNVs were detected in a third of patients even using the relatively narrow primary tumor derived AmpliSeq platform. The FFPE specimens generated a high number of SNVs but did not correlate with CTC profiles, likely reflecting biological disparity between early localized tumors and advanced metastatic disease, as well as genomic artifacts introduced into primary tumor specimens by FFPE preservation. These data demonstrate the feasibility and potential biological and technical advantages of CTCs over traditional FFPE samples for genomic analysis in the pursuit of personalized cancer medicine. Citation Format: Stephen V. Liu, Paul W. Dempsey, William Strauss, Yucheng Xu, Tong Xu, Jessamine Winer-Jones, Tim J. Triche, David I. Quinn, Janine McMurdie, Andre Defusco, Amir Goldkorn. Targeted next-generation sequencing (NGS) of circulating tumor cells (CTCs) and matched primary tumors. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr LB-227. doi:10.1158/1538-7445.AM2013-LB-227

Background: Copy number (CN) and somatic mutation studies of cancer hold great promise for the discovery of reliable biomarkers that can predict clinical outcomes. One of the challenges in such study is the majority of banked samples are formalin-fixed paraffin embedded (FFPE). Due to DNA degradation, FFPE samples generally perform poorly with CN technologies. However, MIP technology works well on FFPE samples, requiring only small stretches of intact genomic DNA (∼40 bp) in a 75-ng input. We have applied this 330,000-plex platform on thousands of FFPE samples from various tissues, e.g. breast, colon and brain, with 90% overall pass rate. Experimental Design: MIP probes were synthesized and screened for the performance. The selection criteria were: (1) good quantitative performance in reproducibility and dynamic range; (2) non-redundant whole genome coverage. Our platform employs a measurement comparing adjacent markers across the genome: this median of absolute pairwise distribution, or MAPD, is a reliable metric for assessing sample quality1. For array design, the best sequence was chosen based on empirical screening data. Results: Using only 75 ng of genomic DNA input, we have obtained both good genotyping and CN data, which is offered as a service-only product under the name “Oncoscan FFPE Express” by Affymetrix. For FFPE samples, good concordance between normal and tumor sample pairs is also observed1. Using an arbitrary cutoff of MAPD ≤ 0.6, a 2000-FFPE sample project has an overall pass rate &amp;gt; 92%. In addition to CN information, data for 400 frequent occurred somatic mutations in cancer listed in the Sanger COSMIC database are also generated in the same assay. They range from single nucleotide change to 10 bp insertion-deletion. Preliminary validation data by different platforms show 10% sensitivity (10% tumor with 90% normal) from a few key mutations (unpublished). While more validation is needed for every mutation interrogated, the current data show great promise of a complete solution for cancer mutation spectrum by a single assay. Conclusion: We have developed a powerful 330K MIP-CCN platform that works well on both frozen and archival FFPE samples. All three types of mutation (CN, SNP and somatic) can be interrogated by the same assay, offering unprecedented opportunity for retrospective study where only FFPE samples are available, and ongoing clinical trials where only small quantity of biopsy samples are accessible.

BackgroundA major challenge facing DNA copy number (CN) studies of tumors is that most banked samples with extensive clinical follow-up information are Formalin-Fixed Paraffin Embedded (FFPE). DNA from FFPE samples generally underperforms or suffers high failure rates compared to fresh frozen samples because of DNA degradation and cross-linking during FFPE fixation and processing. As FFPE protocols may vary widely between labs and samples may be stored for decades at room temperature, an ideal FFPE CN technology should work on diverse sample sets. Molecular Inversion Probe (MIP) technology has been applied successfully to obtain high quality CN and genotype data from cell line and frozen tumor DNA. Since the MIP probes require only a small (~40 bp) target binding site, we reasoned they may be well suited to assess degraded FFPE DNA. We assessed CN with a MIP panel of 50,000 markers in 93 FFPE tumor samples from 7 diverse collections. For 38 FFPE samples from three collections we were also able to asses CN in matched fresh frozen tumor tissue.ResultsUsing an input of 37 ng genomic DNA, we generated high quality CN data with MIP technology in 88% of FFPE samples from seven diverse collections. When matched fresh frozen tissue was available, the performance of FFPE DNA was comparable to that of DNA obtained from matched frozen tumor (genotype concordance averaged 99.9%), with only a modest loss in performance in FFPE.ConclusionMIP technology can be used to generate high quality CN and genotype data in FFPE as well as fresh frozen samples.

Whole transcriptome profiling is a promising technique in adrenal studies; however, whole transcriptome profiling of adrenal disease using formalin-fixed paraffin-embedded (FFPE) samples has to be further explored. The aim of this study was to evaluate the utility of transcriptome data from FFPE samples of adrenocortical tumors. We performed whole transcriptome profiling of FFPE and fresh frozen samples of adrenocortical carcinoma (ACC, n = 3), aldosterone-producing adenoma (APA, n = 3), and cortisol-producing adenoma (CPA, n = 3), and examined the similarity between the transcriptome data. We further examined whether the transcriptome data of FFPE samples could be used to distinguish tumor types and detect marker genes. The number of read counts was smaller in FFPE samples than in fresh frozen samples (P < 0.01), while the number of genes detected was similar (P = 0.39). The gene expression profiles of FFPE and fresh frozen samples were highly correlated (r = 0.93, P < 0.01). Tumor types could be distinguished by consensus clustering and principal component analysis using transcriptome data from FFPE samples. In the differential expression analysis between ACC and APA-CPA, known marker genes of ACC (e.g., CCNB2, TOP2A, and MAD2L1) were detected in FFPE samples of ACC. In the differential expression analysis between APA and CPA, known marker genes of APA (e.g., CYP11B2, VSNL1, and KCNJ5) were detected in the APA of FFPE samples. The results suggest that FFPE samples may be a reliable alternative to fresh frozen samples for whole transcriptome profiling of adrenocortical tumors.

Targeted metabolomics in formalin-fixed paraffin-embedded tissue specimens: Liquid chromatography-tandem mass spectrometry determination of acidic metabolites in cancer research

The aim of this study was to identify and quantify lncRNAs in formalin-fixed paraffin-embedded (FFPE) samples from stage II colorectal cancer (CRC) patients. Colorectal cancer (CRC) is the third most common cancer worldwide, and has high metastasis and recurrence rates. Use of adjuvant chemotherapy in stage II CRC patients is challenging, and new biomarkers are required to identify patients with high probability of relapse. Long noncoding RNAs (lncRNA) play an important role in cancer, awakening especial interest as potential novel cancer biomarkers and therapeutic targets. Resected tissues and biopsy preserved as formalin-fixed, paraffin-embedded (FFPE) samples are the most abundant supply for translational research. Unfortunately, long-chain RNA molecules, such as lncRNAs, may suffer degradation in those conditions. This fact coupled with their low expression levels challenges their detection with current techniques. Thus, in this study we developed a new approach which couples target enrichment and RNAseq to enhance detection of lncRNAs. Our three main goals are: I) evaluate the efficiency of this methodology on FFPE samples, II) investigate the lncRNAs involved in stage II CRC, and III) check for the presence of lncRNA differentially expressed in tumors of individuals who relapsed after surgery and not. We designed a custom-made target enrichment based on Nimblegen (Roche) technology to sequence a total of 8,048 lncRNAs (258 of interest for CRC). We tested 36 paired stage II CRC tumor FFPE tissues and their adjacent normal tissue, 12 from patients who relapsed after surgery. We quantified the expression of the targeted lncRNA and performed a differential expression analysis. Our approach detected expression of numerous lncRNA from FFPE samples. Importantly, we validated 5 target lncRNAs expression values by real time qPCR using two independent housekeeping genes, and found significant correlation between RNAseq (log2 (TPM)) to the qPCR (deltaCT) values. In addition, we identified a total of 83 lncRNAs which are differentially expressed between normal and tumoral samples. This list is enriched in genes previously reported as being involved in CRC, but includes many novel candidates of unknown function. Finally, the expression profile of tumors from patients with and without relapse seems quite similar, although we identified 5 lncRNAs that are significantly differentially expressed. Further validation in a larger cohort is needed to determine whether they could have prognostic value. In summary, our study shows that SeqCap Enrichment System is a good strategy to determine lncRNAs expression in FFPE tissue. Of note, we identified a list of new lncRNAs that are deregulated in stage II CRC tumor. Further research on these lncRNAs may lead to novel clinical applications in cancer biogenesis and prognosis. Citation Format: Anna Brunet-Vega, Cinta Pegueroles, Susana Iraola-Guzmán, Laia Vilà, Alex Casalots, Paula Ribera, Ismael Macias, Hrant Hovhannisyan, Ester Saus, Carles Pericay, Toni Gabaldón. A novel approach of target enriched next generation sequencing reveals differences in expression of lncRNAs in stage II colorectal cancer formalin fixed paraffin embedded (FFPE) tumors [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1396.

e18556 Background: The results of comprehensive genome-wide characterization of lung cancer have recently been reported. However, the concordance of driver mutations between formalin-fixed paraffin-embedded (FFPE) and surgically resected snap-frozen samples is unclear. We conducted the Shizuoka Lung Cancer Mutation Study to analyze driver mutations in patients with lung cancer. Methods: Based on the biobanking system in conjunction with the clinic including the pathology lab, we developed a multiplexed mutational panel designed to assess 23 mutations in 9 genes (EGFR, KRAS, BRAF, PIK3CA, NRAS, MEK1, AKT1, PTEN and HER2), EGFR, MET, PIK3CA, FGFR1 and FGFR2 amplifications, and EML4-ALK translocations, using pyrosequencing plus capillary electrophoresis, qRT-PCR, and RT-PCR, respectively.DNA concentration in FFPE samples was analyzed by measuring absorbance at 260 nm (A260) in a spectrophotometer. Results: Between July 2011 and July 2012, sixty-five lung cancer patients with both FFPE and snap-frozen samples were included in this study. Patient characteristics were as follows: median age (range) 70 (38-92) years; female 68%; never-smoker 34%; histology adenocarcinoma/squamous cell carcinoma/small cell carcinoma/other 52/29/6/13 %. FFPE samples included 64 transbronchial biopsies and one pleural biopsy. We detected driver mutations in 58% of all cases. Mutations found: EGFR 28%, KRAS 12%, PIK3CA 6%, NRAS 2%, HER2 5%, EGFR amplification 5%, PIK3CA amplification 12%, FGFR1amplification 2%. Complete concordance of driver mutations between FFPE and snap-frozen samples was shown in 65%. Median DNA concentration in FFPE samples was 72.7 ng/ul (range; 2.4 - 472.4 ng/ul). Exploratory analyses using an ROC curve, done to evaluate the useful cutoff of DNA concentration in FFPE samples, showed low AUC (AUC; 0.5326). Conclusions: These results suggest that the rate of complete concordance of driver mutations between FFPE and snap-frozen samples might be acceptable, and DNA concentration in FFPE samples might not be correlated with concordance.

e15716 Background: Although fresh-frozen samples are currently the gold standard when performing mass spectrometry-based metabolomics, clinical workflows commonly produce formalin-fixed paraffin-embedded (FFPE) tissues. Fixation in formalin and embeddement in paraffin offers a number of advantages, such as mitigating the risks of infectious agents and preserving the architectural components of the tissue. The latter is important for pathological assesement of cancer, where changes in tissue architecture can be used for diagnosis and to guide treatment decisions. Methods: A concern of using FFPE for metabolomics is that it chemically modifies metabolites and lipids. Thus, while metabolomics data can be generated from FFPE specimens, there remains a major question about its reliability. In this study, we optimized a sample preparation method for use in FFPE metabolomics, then validated the approach on 12 tissue samples from colon cancer patients comparing the results to healthy nearby adjacent tissue (NAT). The analysis was performed on matched fresh frozen and FFPE tissues. By comparing the data, we establish a panel of metabolites and lipids that can be reliably profiled from FFPE tissues by using our workflow. Results: A total of 946 unique metabolites and lipids were measured from FFPE samples using our next-generation metabolomics platform. Across all assays, the coefficient of variation (CV) values were less than 10%. More than 50% of the unique metabolites and lipids from FFPE samples were also identified in matched fresh-frozen tissues. Molecules measured from both tissue types spanned multiple chemical classes ranging from fatty acids to central carbon metabolites. Next, changes in metabolite abundance across cancer and NAT tissues were used to assess the reliability of our FFPE metabolomics workflow in producing biologically relevant findings. Among the metabolites and lipids that were measured in both platforms, a large fraction showed consistent fold changes between fresh frozen and FFPE specimens. As an example, metabolites in glycolysis and the TCA cycle were found to be altered with a similar statistical magnitude in both fresh frozen and FFPE samples. These results are consistent with expected changes associated with the Warburg effect. Conclusions: This study reveals that metabolite profiling in FFPE tissues can effectively identify biologically significant compounds and pathways, offering a new tool for discovery research in cancer.