Assessment of long-term stored specimens in the Siriraj Hospital colorectal cancer biobank for RNA sequencing and profiling

Participating patients and ethical considerations

Patients at least 18 years of age who were diagnosed with colorectal adenocarcinoma at Siriraj Hospital, Faculty of Medicine Siriraj Hospital, Mahidol University, Thailand, were included in this study. All participants were treated according to the National Comprehensive Cancer Network guidelines. This study was part of a research series approved by the Siriraj Institutional Review Board (COA No. Si156/2011, Si348/2019 and Si105/2021). Ethical considerations and informed consent have been followed during the collection of FF and FFPE tissues from a total of 75 participating individuals during two distinct timeframes, 2011–2012 and 2020–2021. Table 1 shows the demographic data of the participating patients.

Sample collection and storage

For FF tissues, within 30 min after surgical resection, the freshly obtained primary CRC tissues were promptly cut into 2–3 mm pieces. They were immediately washed with phosphate-buffered saline (pH ∼ 7.4) and then immersed in 1 mL of chilled RNAlater solution (Invitrogen, MA, USA) on ice, following the protocol outlined in our previous report [11]. Subsequently, the specimens were stored at −80 °C in the Siriraj CRC biobank until they were required for further use.

To prepare FFPE tissues, the resected primary CRC tissues were fixed in 10 % neutral buffered formalin within 60 min after surgical resection. The fixation process lasted for 90 min. Subsequently, the fixed tissues underwent dehydration using ascending grades of ethyl alcohol, including 95 % ethanol for 30 min (2 times) and absolute ethanol for 60 min (3 times). Following dehydration, the tissues underwent clearing by immersion in xylene for 60 min (2 times). For infiltration, the dehydrated and cleared tissues were then immersed in molten paraffin wax (maintained at 60–64 °C) for 60 min (3 times). After infiltration, the tissues were oriented in molds filled with molten paraffin wax (maintained at 60–65 °C) and allowed to solidify on a cold plate set at −5 °C. Finally, the resulting paraffin blocks containing the embedded tissue were stored at ambient temperature until further use.

Notably, all specimens were collected by surgeons and assessed by pathologists. The tissue collection, preparation, and storage protocols remained consistent across two distinct periods, spanning from 2011 to 2012 and from 2020 to 2021. All specimens were stored until all experiments were conducted in 2022. Therefore, the storage time for the specimens collected 2011–2012 and 2020–2021 was approximately 10.5 and 1.5 years, respectively. The difference in storage time between the two groups was approximately 9 years.

RNA-Seq

Total RNA was extracted from the stored FF tissues. A lysis buffer together with the lysing matrix Z (MP Biomedicals, Santa Ana, CA, USA) was applied to each FF tissue (30 mg) before homogenization using a FastPrep-24 5G instrument (MP Biomedicals, CA, USA) at a speed of 6.0 m/s for 40 s on and 3 min off cycles. The homogenization step was continued until a clear lysate was obtained. The RNeasy mini kit (Qiagen, Hilden, Germany) was used to extract total RNA according to the manufacturer’s instructions. Approximately 30 μL RNA eluates were obtained. Concentration and RNA integrity number (RIN) were assessed using the 5,400 fragment analyzer system (Agilent Technologies, CA, USA).

RNA library preparation and sequencing were performed by Novogene Co. Ltd. (Singapore) using human strand-specific mRNA (WBI-Quantification), directional mRNA library preparation (poly A enrichment) and 2 × 150 bp pair-end configuration (PE) with raw data of 8.0 Gb per sample. Briefly, isolation of mRNA from total RNA was achieved by utilizing magnetic beads coupled with poly-T oligos. Following fragmentation, the first strand of cDNA was synthesized via random hexamer primers, followed by the subsequent synthesis of the second cDNA strand using dUTP for directional library. The library’s quality assessment was conducted using Qubit and real-time polymerase chain reaction for quantification, as well as a bioanalyzer for size distribution analysis. The prepared libraries were subsequently pooled for sequencing on the NovaSeq platform (Illumina, San Diego, CA, USA), according to optimal library concentration and data amount. All samples were sequenced in the same batch. During the data quality control process, raw reads of the FASTQ format were processed before clean reads were obtained for downstream analyzes. We conducted quantification of gene expression levels, co-expression analysis, and enrichment analysis based on whole transcriptome RNA-Seq data. Cell deconvolution in each FF sample was calculated by python-based cellanneal software using the RNA-Seq data and using the signature of human primary cells from the human cell atlas [12].

nCounter RNA profiling

Total RNA was extracted from the stored FFPE tissues (two sections with 5 μm thickness) using a high purity FFPE RNA isolation kit (Roche Diagnostics, Indianapolis, IND, USA), according to the manufacturer’s instructions. Approximately 30 μL RNA eluates were obtained and the RNA concentration was analyzed using a Nanodrop 8,000 spectrophotometer (Thermo Fisher Scientific, MA, USA). Besides, the 18S fragment size was analyzed by the Agilent 2,100 Bioanalyzer and RNA 6000 Nano Chip (Agilent Technologies, CA, USA).

Gene expression profiling was performed using the nCounter PanCancer Progression Panel, which encompasses 770 cancer-associated genes (NanoString Technologies, WA, USA). Each sample, comprising 300 ng of RNA, was hybridized involving capture probes and reporter probes via CodeSet hybridization, conducted at 65 °C for a duration of 18 h. Subsequently, the hybridized specimens were processed on the automated nCounter Prep Station (NanoString Technologies, WA, USA), and the nCounter cartridges obtained were interpreted using the nCounter Digital Analyzer (NanoString Technologies, WA, USA). Data processing was performed with the nSolver Analysis Software version 4.0 (NanoString Technologies, WA, USA), with raw counts being normalized against the geometric mean counts of 11 housekeeping genes (HRNP1, RPL27, RPL9, RPL6, RPL30, OAZ1, PTMA, RPS29, UBC, RPS12, and RPS16). Cell deconvolution for each FFPE sample was determined using python-based cellanneal software, the nCounter RNA data, and the signature of human primary cells from the human cell atlas [12].

Statistical analysis

IBM SPSS Statistics (SPSS Inc., IL, USA) or Prism (GraphPad Software, Inc., CA, USA) was used to perform statistical analyzes using appropriate methods, such as the T test, the Mann-Whitney U test, the Chi-square test, and Fisher’s exact test to determine differences between the 2011–2012 and 2020–2021 groups. A p-value of <0.05 was considered statistically significant.

Assessment of long-term stored FF specimens from CRC patients for RNA-Seq

In this section, we conducted an examination of FF CRC primary tissues collected during two distinct time periods, 2011–2012 and 2020–2021, with the objective of evaluating their molecular integrity and suitability for RNA-Seq, a high-throughput technique critical for understanding gene expression patterns and molecular mechanisms in cancer [13]. Our analysis involved 30 FF tissue specimens collected from a cohort of 30 CRC patients (Figure 1A). The tissues collected during the two specified time frames, which had an approximate 9-year difference in storage time, were subjected to RNA extraction and subsequent RNA-Seq to assess their suitability for downstream molecular analyzes.

Figure 1:

Assessment of long-term stored FF specimens from CRC patients for RNA-Seq. (A) Patient participation chart. (B) RNA concentration. (C) RIN value. (D) Raw reads. (E) Clean reads. (F) Tumor cell fraction. p-Value (p) was calculated using the Mann-Whitney U test.

First, the quality control processes of the samples and sequencing data were focused. As shown in Figure 1B and C, the RNA extracted from both time frames exhibited consistent concentration and integrity. There were no significant differences in concentration and RIN values between both groups, indicating optimal RNA preservation, despite the extended storage durations (Figure 1B and C). Furthermore, in terms of sequencing data quality, the data obtained from both sample groups were consistent in raw and clean reads (Figure 1D and E). From cell deconvolution, FF tissues contained tumor cells (colorectal adenocarcinoma) with an average of approximately 28 % (Supplementary Figure S1 and Figure 1F). There was no significant difference of tumor cell content in both groups of FF tissues (Figure 1F).

In addition to quality control processes, CRC-specific gene signature, coexpression, and enrichment analyzes were performed. The expression of four genes (BDNF, CTNNB1, GSK3B and PTGS2), which were reported to be upregulated in primary tumor tissues of CRC compared to adjacent normal tissues [14] was analyzed. Interestingly, comparative analysis between the two timeframes demonstrated a concordance in the four-gene CRC signature (Figure 2A). Figure 2B shows the Venn coexpression diagram, indicating the number of genes that are uniquely expressed and co-expressed in the two groups. Both groups co-expressed approximately 93 % of the genes analyzed (Figure 2B). As a result of enrichment analysis by the collection of Kyoto Encyclopedia of Genes and Genomes (KEGG), ribosome biogenesis, spliceosome and antifolate resistance pathways were significantly altered in the observed gene expression patterns of the two groups (Figure 2B).

Figure 2:

Analysis of RNA-Seq data from long-term stored FF CRC specimens. (A) Transcript per million value of four CRC-signature genes. p-Value (p) was calculated using the Mann-Whitney U test. (B) Venn diagram of coexpression analysis (top) and table of KEGG: Kyoto Encyclopedia of Genes and Genomes enrichment analysis (bottom).

This investigation underscores the remarkable molecular suitability of long-term stored FF CRC primary tissues. Despite storage over an extended period, these specimens retain RNA suitable for RNA-Seq analyzes. Consistent concentration, integrity, sequencing read generation, gene signature expression, and gene coexpression of the two groups affirmed the suitability of approximately 10-year-stored FF primary tissues for exploring CRC biology. However, the pathways related to ribosome biogenesis in eukaryotes, spliceosome, and antifolate resistance should be noted when the samples were collected in the different periods. This finding further emphasizes the lasting value of biobank resources to advance oncological research.

Assessment of long-term stored FFPE specimens from CRC patients for RNA nCounter profiling

For this section, we performed an assessment of FFPE CRC primary tissues collected during two distinct time periods, 2011–2012 and 2020–2021. Our objective was to evaluate the molecular suitability of long-term stored FFPE tissues for nCounter RNA profiling, a promising technique for quantifying specific RNA transcripts and elucidating gene expression patterns in cancer [15]. The study involved 58 FFPE CRC primary tissues (36 tissues in 2011–2012 and 22 tissues in 2020–2021) collected from a cohort of 45 CRC patients (Figure 3A). In some patients, two or more archived FFPE blocks were used (Figure 3A). These FFPE tissues were processed over specified timeframes to extract RNA for subsequent nCounter RNA profiling, with the aim of assessing their suitability for CRC-related molecular analysis.

Figure 3:

Assessment of long-term stored FFPE specimens from CRC patients for RNA nCounter profiling. (A) Patient participation chart. (B) RNA concentration. (C) 18S fragment size. (D) Total raw counts. (E) Total normalized counts. (F) Tumor cell fraction. p-Value (p) was calculated using the T-test or Mann-Whitney U test.

Despite prolonged storage periods, the RNA extracted from both time frames did not show significant differences in yield or concentration (Figure 3B). However, the reduction of 18S fragment size in the older samples was observed (Figure 3C). Following the hybridization and counting of 770 target genes in the Nanostring PanCancer panel, gene counts and CRC-specific genes were analyzed. The results of the RNA nCounter profiling showed that the FFPE tissues from 2011 to 2012 significantly produced lower raw counts than the FFPE tissues from 2020 to 2021 (Figure 3D). However, the total of normalized counts was comparable between the two groups after normalization by the housekeeping genes (Figure 3E). Based on cell deconvolution analysis, FFPE tissues showed an average tumor cell content of approximately 22 % (Supplementary Figure S2 and Figure 3F). No significant difference in tumor cell content was observed between the two groups of FFPE tissues (Figure 3F).

Five genes, including COL1A1, COMP, ETV4, SPP1 and INHBA, were then analyzed, since they were included in the Nanostring PanCancer panel and reported to be differentially expressed in tissues of colon adenocarcinoma versus normal tissues of the colon [16]. The results of the nCounter RNA profiling also demonstrated that the FFPE tissues from both time periods produced comparable normalized counts of these CRC-specific genes (Figure 4). These 5 CRC-related genes also did not exhibit significant differences between the two storage periods in FF tissues (Supplementary Figure S3). This comparative analysis between 2011-2012 and 2020–2021 specimens revealed consistent normalized expression profiles of these CRC-specific genes, reaffirming the utility of approximately 10-year-stored FFPE primary tissues for investigating gene expression patterns in CRC.

Figure 4:

Normalized counts for five CRC-signature genes in the PanCancer progression panel. p-Value (p) was calculated using the Mann-Whitney U test.