Whole-genome gene expression profiling of formalin-fixed, paraffin-embedded tissue samples

Abstract

Background: We have developed a gene expression assay (Whole-Genome DASL), capable of generating whole-genome gene expression profiles from degraded samples such as formalin-fixed, paraffin-embedded (FFPE) specimens.

Methodology/principal findings: We demonstrated a similar level of sensitivity in gene detection between matched fresh-frozen (FF) and FFPE samples, with the number and overlap of probes detected in the FFPE samples being approximately 88% and 95% of that in the corresponding FF samples, respectively; 74% of the differentially expressed probes overlapped between the FF and FFPE pairs. The WG-DASL assay is also able to detect 1.3-1.5 and 1.5-2 -fold changes in intact and FFPE samples, respectively. The dynamic range for the assay is approximately 3 logs. Comparing the WG-DASL assay with an in vitro transcription-based labeling method yielded fold-change correlations of R(2) approximately 0.83, while fold-change comparisons with quantitative RT-PCR assays yielded R(2) approximately 0.86 and R(2) approximately 0.55 for intact and FFPE samples, respectively. Additionally, the WG-DASL assay yielded high self-correlations (R(2)>0.98) with low intact RNA inputs ranging from 1 ng to 100 ng; reproducible expression profiles were also obtained with 250 pg total RNA (R(2) approximately 0.92), with approximately 71% of the probes detected in 100 ng total RNA also detected at the 250 pg level. When FFPE samples were assayed, 1 ng total RNA yielded self-correlations of R(2) approximately 0.80, while still maintaining a correlation of R(2) approximately 0.75 with standard FFPE inputs (200 ng).

Conclusions/significance: Taken together, these results show that WG-DASL assay provides a reliable platform for genome-wide expression profiling in archived materials. It also possesses utility within clinical settings where only limited quantities of samples may be available (e.g. microdissected material) or when minimally invasive procedures are performed (e.g. biopsied specimens).

Figure 1. WG-DASL assay reproducibility with variable RNA inputs.

Raw intensity scatterplots (all 24526 probes) are shown for WG-DASL assay replicates (replicate 1, x-axis; replicate 2, y-axis) for (A) 200 ng total RNA derived from a formalin-fixed, paraffin-embedded (FFPE) ovarian tumor; (B) 100 ng intact Raji RNA; and (C) 200 ng artificially degraded Raji RNA; (D) The fold-change correlation between intact (x-axis) and artificially degraded (y-axis) cell line RNAs was calculated as the Log₂ of the fold-change between the Raji and MCF-7 sample intensities. All detected probes (Detection p-value <0.01) are plotted.

Figure 2. Comparison between fresh-frozen (FF) and FFPE samples.

Matched FF and FFPE samples for lung tumor (TUM) and normal adjacent tissue (NAT) were labeled and hybridized using the WG-DASL assay. (A) Assays were performed in quadruplicate and the average number of detected probes plotted for each sample (error bars ± SEM). (B) The overlap of detected probes between the matched FF and FFPE samples for both the NAT and TUM tissues was calculated as the percentage of detected FFPE probes also detected in the corresponding FF sample. (C) The fold-change correlation between FF and FFPE samples was calculated as the Log₂ of the fold-change between the NAT and TUM tissues intensities. All detected probes (Detection p-value <0.01) are plotted. (D) Lists of differentially expressed probes were generated by comparing replicates of the NAT and TUM tissues with a false discovery rate (FDR) cutoff of <5%. The overlap of differentially expressed probes between the FF and FFPE matched samples for both the NAT and TUM tissues was calculated as the percentage of differentially expressed FFPE probes also detected in the corresponding FF sample.

Figure 3. Assay fold-change detection.

(A) Total RNAs from Raji (R) and MCF-7 (M) cells were mixed at various ratios, with each mixture having a combined input of 100 ng. After labeling and hybridizing, the raw data underwent unsupervised hierarchical clustering using all 24526 probes shown in the cluster dendrogram. Units on the x-axis are Pearson's correlation coefficient (R). (B) Using a set of differentially expressed probes, derived from comparisons between the two intact RNA samples (Raji and MCF-7), the number of differentially expressed probes identified in different Raji/MCF-7 mixtures was calculated and plotted as a percentage of that identified in the 100% Raji sample. Values on the x-axis correspond to the percentage of Raji RNA in each mixture, while the values on the y-axis represent the percentage of Raji-specific probes detected in each mixture at p <0.05. For example, more than 98% of the Raji-specific probes were detected in the 25% Raji mixture (1.3-fold change) and higher Raji mixtures. (C) Using a set of differentially expressed probes, obtained from comparisons between two FFPE samples (DFCI #121 and DFCI #123), the number of differentially expressed probes identified in the different DFCI #121/#123 mixtures was calculated and plotted as a percentage of that identified in the 100% DFCI #121 sample. Values on the x-axis correspond to the percentage of DFCI #121 RNA in each mixture, while the values on the y-axis represent the percentage of DFCI #121-specific probes detected in each mixture at p<0.05. For example, more than 80% of the DFCI #121-specific probes were detected in the 33% DFCI #121 mixture (1.5-fold change) and higher DFCI #121 mixtures.

Figure 4. WG-DASL assay performance with low RNA inputs.

(A) Raw intensity scatterplots are shown for assay replicates for 250 pg Raji RNA and (B) also for a correlation between 250 pg and 100 ng Raji RNA. (C) Lists of differentially expressed probes were generated by comparing replicate data for Raji and MCF-7 cells using a cutoff of p = 0.001. The overlap of differentially expressed probes (Raji vs. MCF-7) between the 250 pg and 100 ng RNA inputs was calculated as the percentage of differentially expressed probes identified in the 250 pg sample which was also identified in the 100 ng sample.

Figure 5. Validation with in vitro transcription-based and qRT-PCR assays.

(A) Logarithmic fold-changes in transcript abundance between two cancer cell lines (Raji and MCF-7) were compared between the WG-DASL assay (100 ng total RNA input; x-axis) and an in vitro transcription (IVT)-based array platform (100 ng total RNA input; y-axis); data were quantile normalized and all common detected probes (Detection p-value <0.01) were plotted. (B) Comparison of the WG-DASL assay (100 ng total RNA input; x-axis) and qRT-PCR (y-axis) with intact RNA and (C) the WG-DASL assay with lower intact input (250 pg total RNA; x-axis) and qRT-PCR (y-axis). The qRT-PCR fold-change data are derived from Ct values for 24 common transcripts.