Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2012 Oct 1.
Published in final edited form as: Pharmacogenomics J. 2010 Jun 15;11(5):348–358. doi: 10.1038/tpj.2010.50

High Efficiency Genotype Analysis from Formalin-Fixed, Paraffin-Embedded Tumor Tissues

Matthew J Sikora 1, Jacklyn N Thibert 2, Janine Salter 3, Mitch Dowsett 3, Michael D Johnson 4, James M Rae 1,2
PMCID: PMC2996486  NIHMSID: NIHMS228101  PMID: 20548328

Abstract

Single nucleotide polymorphisms (SNPs) can be assayed using DNA isolated from archival formalin-fixed paraffin-embedded (FFPE) samples, making retrospective pharmacogenetic studies possible. Here we describe methods that significantly increase the number of SNP determinations possible using FFPE samples. Quantifying the amount of DNA amenable to PCR (amplification-quality DNA, AQ-DNA) allows a significant reduction in the amount of sample required for Taqman-based SNP assays. Optimizing AQ-DNA input increases PCR amplification efficiency and SNP determination accuracy. DNA was extracted from thirty-nine FFPE tumor sections and matched tumor and stromal cores, of the type used to generate tissue microarrays. Sections and tumor cores yielded sufficient AQ-DNA for over 1,000 SNP determinations. Seven SNPs were assessed, following individual assay optimization for minimal AQ-DNA. Genotypes from tumor cores for single SNPs were 92.3-100% concordant with those obtained from sections. Using these methods, the number of SNP genotypes that can be determined from single FFPE samples is greatly increased expanding the genetic association studies possible from limited archival specimens. The use of tumor cores is of particular importance since the harvesting of tumor cores has minimal impact on the utility of the donor blocks for other purposes.

Keywords: genotyping, paraffin-embedded, tumor samples, genetic polymorphism

Introduction

Until recently, clinical trials in cancer did not routinely collect and store patient DNA samples, thus limiting the ability to conduct pharmacogenetic analysis of many large, landmark clinical trials. Another potential source of patient DNA does, however, exist in the form of formalin-fixed, paraffin-embedded (FFPE) tumor samples that were collected for the majority of clinical trials in oncology. Previously, we and others have demonstrated that DNA can be obtained from these tumor blocks, that SNP genotypes can be reliably determined from this DNA, and that the genotypes of SNPs in genes of pharmacogenetic interest (primarily involved in drug metabolism) derived from the tumor samples match those of the germline DNA1-4. This technological advance demonstrated that FFPE samples could be used for pharmacogenetic analyses of historical prospective clinical trials, thereby allowing the existing wealth of large, carefully conducted clinical trials of chemotherapeutic agents to be mined for associations between inherited gene variants with drug toxicities and clinical outcomes.

Using DNA extracted from FFPE tumor samples for genotype analyses presents significant technical challenges due to the relatively low quantity and poor quality of the template DNA that is extracted from these samples. Standard DNA isolation methods used for FFPE samples typically produce severely sheared and fragmented DNA which is frequently not optimal for PCR-based genotyping5-8. All commonly used PCR-based genotyping assays have been designed to work using high quality DNA isolated from viable cells. Real-time PCR-based Taqman assays, for instance, involve the amplification of DNA segments 80-150 base pairs (bp) in length containing the SNP of interest9. In addition, 5-10μm FFPE tissue block sections typically only yield sufficient DNA for 30-50 Taqman-based genotyping reactions (unpublished observations). Furthermore, Taqman PCR reactions using DNA extracted from FFPE samples are frequently inefficient, and yield irregular fluorescence output curves, making allelic determination difficult or ambiguous. Common practice when confronted with DNA samples that are difficult to genotype is to increase the amount of template DNA used per reaction. However, we have observed that doing this with FFPE-derived DNA often results in worsening rather than improvement of the PCR reaction. A simple method to determine the optimal amount of a given sample of FFPE-derived DNA for each assay would be valuable.

In this report, we describe methods to analyze DNA harvested from FFPE materials that allows the assessment of overall quality and quantification of ‘amplification-quality DNA’ or ‘AQ-DNA’ (DNA fragments large enough for efficient PCR-based analysis). Using materials evaluated in this manor, we describe how minimizing the amount of input DNA in Taqman-based genotyping reactions significantly improves PCR amplification efficiency, increases the accuracy of allelic determinations, and greatly increases the number of genotyping assays that can be performed per sample. We also demonstrate that FFPE tumor cores of the type used to generate tissue microarrays (TMAs) can be used as a source of DNA for Taqman-based genotyping. The methodological approaches described herein facilitate the improved application of Taqman-based SNP genotyping to FFPE-derived DNA, significantly increasing the number of assays that can be conducted using what is a valuable and limited tissue resource.

Materials and Methods

Sample Preparation

Samples were selected randomly from the routinely formalin-fixed, paraffin-processed breast tumour archive at the Royal Marsden Hospital, with the prerequisites that they (i) were pre-2006 in order to comply with HTA (Human Tissue Act) legislation regarding patient consent and (ii) were suitable to TMA (tissue microarray) core acquisition from both the tumor and intratumoral stromal compartments. In practice these were predominantly invasive ductal carcinomas grade II/III. At the time of DNA extraction, tumor blocks ranged in age from 4-19 years, with a median age of 15 years. Specific FFPE samples used in each data figure are indicated in Supplementary Table S1.

Tumour and stromal areas were marked on an H&E stained slide and then transposed onto the associated paraffin block. 0.6mm cores were punched in the marked areas using the Beecher tissue arrayer MTA1 and the extracted cores were transferred to RNAse, DNase free tubes. The needle punch was cleaned with ethanol and allowed to dry fully between each core acquisition to prevent cross contamination. Two 10μm paraffin embedded sections were cut from the same paraffin block prior to core acquisition. A new blade was used for each patient sample to eliminate the potential for cross-contamination. Gloves were worn at all times.

DNA Extraction from FFPE-Tissue Samples

Our previously described method for extracting DNA from FFPE samples2 using the DNeasy Blood and Tissue Kit (Qiagen, Valencia, CA) was improved with minor modification to eliminate the use of solvents to de-paraffin samples. Briefly, samples (sections and cores, described above) were heated to 95°C for 15 minutes in 180μL Buffer ATL and then allowed to cool to room temperature for a minimum of 5 minutes. Twenty microliters of Proteinase K solution (20mg/mL) was added to the samples, which were then incubated at 55°C. After 8 hours, fresh Proteinase K solution (20μL) was added and the samples incubated for an additional 8 hours, then DNA extracted according to the manufacturer's instructions.

Quality Control Multiplex PCR

DNA quality was measured using a multiplex PCR-based method described by van Beers et al10, with minor modifications. In addition to primer pairs for the 100, 200, 300, and 400bp fragments of GAPDH used by van Beers et al, additional primer pairs were added for 500, 600, and 700bp fragments. Primer sequences are available upon request. The 100 and 500-700bp primers were used at a final concentration of 266nM, and the 200-400bp primers at 133nM. Reactions were incubated at 94°C for 1 minute, 56°C for 1 minute, and 72°C for 3 minutes, for 35 cycles. PCR products were analyzed by capillary electrophoresis using an Agilent 2100 Bioanalyzer with DNA 1000 electrophoresis chips (Agilent Technologies, Santa Clara, CA).

Quantification of Amplification-Quality DNA (AQ-DNA)

In order to determine the amount of AQ-DNA in extracted FFPE samples we quantified the 100bp fragment of GAPDH using SYBR Green real-time PCR. Briefly, 25μL reactions containing: forward primer (GTT CCA ATA TGA TTC CAC CC) and reverse primer (CTC CTG GAA GAT GGT GAT GG) at final concentrations of 250nM, 12.5μL of Platinum SYBR Green Master Mix (Invitrogen, Carlsbad, CA) and template DNA were incubated for 40 cycles at 95°C for 15 seconds and 56°C for 45 seconds using an iCycler real-time thermocycler (BioRad, Madison, WI).

DNA was isolated from fresh human lymphocytes (Blood and Tissue Kit, Qiagen) and the concentration was measured using a NanoDrop ND-1000 spectrophotometer (Thermo Scientific, Waltham, MA). SYBR Green real-time PCR (described above) was performed using a dilution series of lymphocyte DNA (in Ultra-Pure H2O, Invitrogen) from 1pg–1μg to generate standard curves. These standard curves were used to calculate the relative amount of AQ-DNA in FFPE-derived samples, compared to the high quality lymphocyte DNA. Standard curves were plotted as the cycle number at which SYBR Green fluorescence passed a software determined threshold (threshold cycle, Ct) versus DNA quantity (μg) using GraphPad Prism 4.03 (GraphPad Software, Inc., La Jolla, CA). AQ-DNA from FFPE samples was quantified by adding 1μL of each DNA sample per real-time PCR reaction. The threshold cycle for each FFPE sample was plotted on the standard curve to extrapolate the quantity of AQ-DNA, relative to the high quality lymphocyte DNA.

DNA Enzymatic Shearing

DNA was sheared enzymatically using an Enzyme Shearing Kit (Active Motif, Carlsbad, CA). Genomic DNA from lymphocytes at ∼100ng/μL was diluted 1:10 with Digestion Buffer and 50μL aliquots made. Two and half microliters of Enzyme Cocktail (1:10,000 in 50% glycerol) was added to each aliquot, and incubated for times ranging from 10 to 180 minutes. The reactions were stopped by adding 1μL of ice-cold 0.5M EDTA and incubation on ice for 15 minutes. DNA was purified from shearing reactions for subsequent PCR using Nucleotide Clean-up kits (Qiagen, Valencia, CA) according to the manufacturer's instructions.

Genotyping

FFPE tissue-extracted DNA was genotyped for known single nucleotide polymorphisms (SNPs) in the cytochrome P450 2D6 (CYP2D6) gene. The following SNPs were genotyped: 1846G>A (rs3892097), T1707del (rs5030655), A2549del (rs35742686), 2988G>A (rs28371725), 100C>T (rs1065852), and 4180G>C (rs1135840). Genotyping for these SNPs allows the identification of the most common variant alleles of CYP2D6: CYP2D6*2, *3, *4, *6, *10, and *41. SNPs were determined using Taqman Allelic Discrimination Assays (Applied Biosystems, Foster City, CA) according to the manufacturer's instructions as previously described 11, with minor modification. Reactions were carried out to 60 cycles to allow amplification of sub-nanogram quantities of DNA. Reactions were conducted using Genotyping Master Mix (Applied Biosystems) in an iCycler real-time thermocycler (BioRad). Samples were also genotyped for UGT2B7 802T>C (rs7439366) using a Taqman Alleic Discrimination Assay with minor modification. Reactions were conducted, using 50 pg of DNA as determined by our methods described above, for 50 cycles on an iCycler real-time cycler.

In order to minimize the potential for DNA cross-contamination (a concern when amplifying sub-nanogram quantities of DNA), all genotyping reactions were prepared in a designated template-free zone in a vertical laminar flow hood (AirClean 600, AirClean Systems, Raleigh, NC), with HEPA filtration and UV light.

Results

Multiplex PCR Identifies Samples of Insufficient Quality to Genotype

To assess the quality of DNA from FFPE tissue samples, we adapted a method described by van Beers et al. in which 7 amplicons of increasing size, from 100-700 base pairs, within the GAPDH gene are amplified by PCR.10 The sizes of the amplicons produced using FFPE-derived template DNA correlates with the degree to which the DNA has been sheared or fragmented. We used this assay to screen a panel of DNA samples isolated from FFPE tissue. We selected samples that had previously been genotyped for a panel of SNPs in the Cytochrome P450 2D6 (CYP2D6) gene successfully (thus considered higher quality, HQ DNA) and unsuccessfully (thus considered lower quality, LQ DNA). All 7 fragments are amplified from the high quality DNA control (lane 1). Samples that could not be genotyped (LQ) did not support amplification of any of the GAPDH fragments (shown in LQ samples, lanes 2-5). We observed that FFPE DNA samples that performed well using Taqman genotyping assays (HQ) supported amplification of at least the 100bp fragment, and larger fragments up to 400bp (Figure 1, lanes 6-9).

Figure 1.

Figure 1

Open in a new tab

Multiplex PCR analysis of high quality lymphocyte DNA (control) and FFPE tissue DNA samples (LQ1-4, HQ1-4). All 7 amplicons were amplified in the control (lane 1). No amplicons were amplified in FFPE samples that could not successfully be genotyped (LQ, lanes 2-5). Amplicons of 100bp and greater were amplified from FFPE samples that were successfully genotyped (HQ, lanes 6-9).

Quantification of Amplification-Quality DNA by Real Time PCR

Standard methods for quantifying DNA, such as UV absorbance, do not provide information regarding the degree of DNA fragmentation; therefore, we set out to measure the amount of “amplification-quality DNA” or “AQ-DNA” in DNA samples from FFPE tissues. We developed a quantitative-PCR based assay that determines the amount of fragments of 100bp or greater relative to a high molecular weight DNA standard harvested from viable lymphocytes. This technique uses SYBR Green-based amplification and quantification of the 100bp fragment of GAPDH from the quality control PCR described above. To validate that amplification of this fragment has a linear relationship with DNA quantity, a dilution series of high quality DNA obtained from lymphocytes was generated as described in Materials and Methods.

The linear range of threshold cycle vs. quantity of DNA was found to be 1pg – 10ng (Figure 2A). Interestingly, adding greater than 10ng of DNA to the PCR reaction caused a loss in the linear relationship, and caused inefficiency in PCR amplification (Figure 2B). Furthermore, addition of 1μg of DNA completely quenched the PCR reaction and blocked any amplification (indicated by a threshold cycle of 39 cycles). The dashed line in Figure 2B represents the linear regression from Figure 2A, to demonstrate the loss of linearity at >10ng of input DNA.

Figure 2.

Figure 2

Open in a new tab

SYBR Green quantitative PCR amplification of the 100bp amplicon of GAPDH. A dilution series of high quality lymphocyte DNA was used to generate a standard curve as described in Materials and Methods. Reactions were performed in triplicate, error bars represent ± SD. GAPDH Ct represents the cycle at which reporter fluorescence crosses a software-defined threshold (described in text). A, The linear range of threshold cycle vs. DNA quantity was 1pg – 10ng. Solid line, linear regression of DNA quantity vs. threshold cycle. B, >10ng of DNA/reaction caused a loss in the linearity and inefficiency of PCR amplification. 1μg of DNA quenched the PCR reaction and blocked amplification. The dashed line in B represents the linear regression from A.

Sample Quality by Multiplex PCR Correlates with Quantification of AQ-DNA

To demonstrate that quantification of the 100bp fragment of GAPDH correlates with sample fragmentation as measured by the multiplex PCR described above, lymphocyte DNA was enzymatically sheared as described in Materials and Methods. The Enzymatic Shearing Kit was specifically designed so that increasing incubation time results in progressive fragmentation of the DNA. Overall quality and quantity of AQ-DNA was evaluated as described above.

As shown in Figure 3, increased digestion time caused a progressive loss of the larger PCR amplicons, with loss of the 500-700bp amplicons by 60 minutes, and loss of the 400bp amplicon by 120 minutes. Over-digestion with high concentration enzyme cocktail caused a loss of all bands (data not shown). Increased digestion time also correlated with a loss of AQ-DNA as determined by amplification of the 100bp GAPDH fragment, with a progressive decrease in quantity of AQ-DNA from 0-120 minutes of digestion. AQ-DNA decreased from 17.5ng/μL in the control digestion, to ∼2ng/μL at >2hrs digestion. Further digestion up to 180 minutes with this dilution of enzyme cocktail appeared to have no further effect on fragment size or 100bp fragment quantification. Over-digestion with high concentration enzyme cocktail caused no 100bp amplicon signal to be seen with those samples, and decreased AQ-DNA to ∼0ng/μL (data not shown).

Figure 3.

Figure 3

Open in a new tab

Multiplex PCR analysis and AQ-DNA quantification (as described in Materials and Methods) of high quality lymphocyte DNA samples subjected to enzymatic shearing for increased time. Samples were digested and processed as described in Materials and Methods. All seven amplicons are present in the non-digested control (0′, lane 1), while larger amplicons are lost with increasing digestion time. Quantity of AQ-DNA also decreases with increasing digestion time.

Minimizing Input DNA Improves Genotyping Efficiency

Based on our quantification method described above, we attempted to establish the minimum amount of AQ-DNA required for assessment of genotypes using Taqman-based analyses. Ten FFPE-derived DNA samples were diluted from 1ng to 10pg of AQ-DNA per reaction, and genotyped for CYP2D6 1846G>A (CYP2D6*4). Results are shown in Table 1. Samples with wild-type genotypes for this SNP were successfully genotyped with only 10pg of input DNA. However, the mutant-specific probe in this assay required at least 50pg of input DNA to amplify successfully. Importantly, Sample J was genotyped as a homozygous mutant with 50pg of input AQ-DNA, but a heterozygote with 100pg or more of AQ-DNA. Thus, the minimum quantity of AQ-DNA required for genotyping the CYP2D6*4 allele was set at 100pg. Additional optimization was performed using other Taqman assays. The UGT2B7 T848C (UGT2B7*2) assay required only 50pg of AQ-DNA to reliably genotype samples. Other assays, including CYP2D6 A2549del (CYP2D6*3), required 150pg or more of AQ-DNA to generate reliable genotype data (data not shown).

Table 1.

CYP2D6*4 Genotypes as Determined with Varying Input AQ-DNA

Tumor Section
A B C D E F G H I J
10pg WT WT WT WT WT -- WT WT WT --
50pg WT WT WT WT WT Mut WT WT WT Mut
100pg WT WT WT WT WT Mut WT WT WT Het
500pg WT WT WT WT WT Mut WT WT WT Het
1ng WT WT WT WT WT Mut WT WT WT Het
Open in a new tab

Importantly, optimizing Taqman reactions by minimizing the amount of input AQ-DNA improved Taqman reaction efficiency, and allowed for less ambiguous genotype determinations (Figure 4). To determine patient genotype, endpoint fluorescence from the Taqman-based PCR for each allele-specific probe is assigned to an axis on a scatter plot. A positive or negative signal from each probe is used to assign patient genotypes. However, the fluorescence output curves generated in real-time are representative of efficient vs. inefficient amplification, and can be used to evaluate potential false positives and negatives.

Figure 4.

Figure 4

Figure 4

Open in a new tab

A,B, Real-time PCR fluorescence output curves for the wild-type allele probe of the CYP2D6*4 assay. A, 30 FFPE samples with reactions performed using optimal DNA quantities of AQ-DNA. Samples display efficient amplification and large separation between positive and negative results. B, 30 FFPE samples performed under standard conditions. Inefficient amplification results in minimal separation between positive and negative samples. C,D, Scatter plots of endpoint fluorescence for each allele-specific probe of the CYP2D6*4 assay. C, Optimized samples, and D, standard samples, were assigned genotypes as homozygous wild-type (red squares), heterozygous (blue diamonds), or homozygous mutant (yellow circles). Samples marked as a black ‘X’ could not be assigned a genotype.

Figures 4A and 4B show representative real-time fluorescence outputs for the wild type probe of the CYP2D6 1846G>A (CYP2D6*4) Taqman assay, while Figures 4C and 4D show endpoint fluorescence as a scatter plot of wild-type probe vs. mutant probe (VIC vs. FAM fluorophores), with samples' assigned genotypes. In Figures 4A and 4C, 30 FFPE DNA samples were genotyped using only 100pg of input AQ-DNA. Efficient PCR amplification (shown by Figure 4A) allowed for clear threshold cut-offs, with no samples with endpoint fluorescence at or near the fluorescence threshold (Figure 4C), or the fluorescence value at which point an allele determination is made. Twenty nine of 30 samples genotyped by this method could be assigned a genotype, and only one sample could not be reliably genotyped (96.7% success rate). Shown in Figures 4B and 4D are an additional 30 samples genotyped for the same SNP using ∼10ng of template DNA as determined by standard UV absorbance. Inefficient amplification caused clustering around the calculated threshold (Figure 4B, ∼60 fluorescence units), making genotype determination difficult. Only 21 of 30 samples could be assigned a genotype for this SNP (70% success rate). However, as is evident in Figure 4C, many of these genotypes are likely un-reliable due to the lack of separation between samples deemed positive and negative for the wild-type allele (VIC fluorophore, y-axis).

Tumor Block Cores Yield Equivalent Amounts of AQ-DNA versus Sections

We next set out to determine whether tissue from tumor cores of the type generated for TMA construction, or cores from adjacent stromal tissue, could be used in pharmacogenomic analyses. To compare DNA yield from FFPE tumor block sections and core punches, we obtained 39 matched sets of sections, tumor cores, and cores of adjacent stromal tissue (described in Materials and Methods). Samples were processed and AQ-DNA was quantified as described above. As shown in Figure 5A, sections yielded 1.25±1.05ng/μL of AQ-DNA, and tumor cores yielded 1.20±0.91ng/μL of AQ-DNA, indicating no significant difference in AQ-DNA yield between tumor sections and cores (p = 0.19). These yields provide enough DNA for an average of ∼1250 and ∼1200 genotyping reactions, respectively, based on 100μL of eluent per sample and 100pg AQ-DNA per reaction. However, stromal cores yielded only 0.39±0.52ng/μL of AQ-DNA (∼390 reactions on average), significantly less than both tumor cores and sections. Increased yield obtained from tumor sections correlated with increased yield from tumor cores (Figure 5B). Yield from neither sections nor tumor cores correlated with yield obtained from stroma cores (data not shown).

Figure 5.

Figure 5

Open in a new tab

Quantification of AQ-DNA from 39 matched section, tumor core and stroma core samples. Quantification was performed by real-time PCR as described in Materials and Methods. A, Bars represent average sample yield (ng/μL) ± SD. P values were determined by Student's T-test. Sections and tumor cores yielded equivalent AQ-DNA on average (p>0.05). However, stroma cores yielded significantly less AQ-DNA than other sample types (p < 0.0001 and p = 0.0005 vs. sections and tumor cores, respectively). B, Scatterplot of yield from tumor cores (y-axis) vs. yield from matched sections (x-axis). The dashed line represents the linear regression between yield from sections and matched tumor cores (p < 0.0001).

FFPE Tumor Block Age has Minimal Impact on AQ-DNA Yield

At the time of DNA extraction, 12 of 39 sample sets (30.8%) were from FFPE tumor blocks less than 10 years old; the remaining 27 (69.2%) were 10 years old or greater. We compared tumor block age to AQ-DNA yield for each sample type (Figure 6). For sections, there was a slight, but statistically significant (p=0.0025), decrease in AQ-DNA yield with increasing block age (Figure 6A; slope = -0.1ng/μL per year). The highest AQ-DNA yields from sections were obtained from blocks 4-6 years old. However, sections from blocks 14-19 years old routinely yielded AQ-DNA equivalent to blocks 6-13 years old. AQ-DNA yield from tumor cores (Figure 6B) and stroma cores (Figure 6C) had slight trends toward decreased yield with block age, but neither of these trends were statistically significant (p>0.05).

Figure 6.

Figure 6

Open in a new tab

AQ-DNA yield (ng/μL) for matched A, sections; B, tumor cores; and C, stroma cores; versus the age of the FFPE tumor block at the time of DNA extraction. AQ-DNA yield is shown on a log scale. P-value given is linear regression (dashed line) slope versus 0.

Tumor DNA from Sections or Cores is Higher Quality than Stroma Core DNA

Six matched samples from each type were chosen at random to assess overall sample type quality by multiplex PCR as described above. On average, DNA from tumor block sections was of greater quality than that from tumor cores or stroma cores (Table 2). The maximum amplicon size obtained by multiplex PCR from each sample is shown in Table 2. Section-derived DNA had a maximum amplicon size of 400bp across all 6 samples, and an average maximum amplicon of 250bp. Tumor and stroma core-derived DNA both had maximum amplicons of 300bp, with average maximum amplicons of 200bp and 150bp respectively. Further, while all 6 sets of section and tumor core-derived DNA produced amplicons of at least 100bp, 3 of the 6 stroma core-derived DNA samples produced no amplicons. Examples of multiplex PCR results for each sample type are shown for samples B and F in Figure S1. In each sample, section DNA produces the greatest amplicon size, while tumor and stroma cores do not amplify the larger amplicons.

Table 2.

Maximum Amplicon Sizes from Multiplex PCR of Matched Samples

Maximum Amplicon Size
Sample Section Tumor Core Stroma Core
A 300 300 300
B 400 300 300
C 200 200 300
D 100 100 0
E 200 100 0
F 300 200 0
Avg. Maximum Amplicon 250 200 150
Open in a new tab

To assess the ability to determine SNP genotypes using DNA from each tumor block sample type, the matched sample sets were genotyped for the UGT2B7*2 allele (802T>C). Samples were quantified as described above before genotyping, and reactions were optimized at 50pg AQ-DNA per reaction (data not shown). Figures 7A, 7B, and 7C show initial real-time fluorescence outputs for the wild type probe of the UGT2B7 802T>C Taqman assay, for sections, tumor cores, and stroma cores, respectively. Thirty eight of 39 tumor section samples were successfully genotyped, with only 1 inefficient PCR amplification curve that requires re-genotyping (97.4% call rate). Thirty five of 39 tumor core samples were successfully genotyped (89.7% success rate). The endpoint fluorescence difference between samples above and below threshold fluorescence was decreased in tumor core-derived DNA versus section-derived DNA, and 4 amplification curves ended near the calculated threshold. This difference was decreased further in the stroma core-derived DNA, resulting in successful genotypes for only 30/39 samples (76.9% call rate). Samples that could not be assigned a genotype after this initial run were re-optimized and re-assessed.

Figure 7.

Figure 7

Figure 7

Open in a new tab

A-C, Real-time PCR fluorescence output curves of matched FFPE samples using the UGT2B7 802T>C Taqman assay (mutant allele probe shown). A, Section-derived DNA samples displayed efficient amplification, with clear separation between positive and negative results. B, Tumor core-derived DNA samples and C, stroma core-derived DNA samples did not all clearly pass fluorescence thresholds. D-F, Scatter plots of endpoint fluorescence for each allele-specific probe of the UGT2B7 802T>C assay. D, Section-derived, E, Tumor core-derived, and F, Stroma core derived samples were assigned genotypes as homozygous wild-type (red squares), heterozygous (blue diamonds), or homozygous mutant (yellow circles). Samples marked as a black ‘X’ could not be assigned a genotype.

Figures 7D, 7E, and 7F show endpoint fluorescence as a scatter plot of wild-type probe vs. mutant probe (VIC vs. FAM fluorophores), with samples' assigned genotypes, for sections, tumor cores, and stroma cores, respectively, following re-assessment. Thirty nine of 39 (100%) tumor section samples were successfully genotyped (Figure 7D), while 38 of 39 (97.4%) tumor cores (Figure 7E) and stroma cores (Figure 7F) were successfully genotyped. We have previously shown that genotypes from sections are 100% concordant with genotypes obtained from germ line DNA2. We assessed the agreement between genotypes (concordance) for the CYP2D6 gene (requiring identification of 6 SNP genotypes, listed in Materials and Methods) and the UGT2B7*2 allele (data not shown). Thirty two of 39 (82.1%) tumor core samples yielded identical CYP2D6 genotypes to section samples. However, 5 of 7 discordant genotypes were likely the results of poor AQ-DNA yields (< 0.1ng/μL) that were identified prior to genotype assessment. The two remaining samples differed in only a single SNP genotype (4180G>C; rs1135840). Only 27 of 39 stroma cores yielded identical CYP2D6 genotypes to the section samples, but half of the discordant samples (6 of 12) could be identified by low sample yield prior to genotyping. 3 of 12 differed in only a single SNP genotype. There was a higher degree of concordance between sections and the other sample types for the single SNP UGT2B7*2 allele. Thirty six of 39 (92.3%) tumor core genotypes or stroma core genotypes matched those obtained from section DNA, and 33 of 39 (84.6%) of samples had matching genotypes from all 3 sample types. The 3 tumor core samples that had discordant results versus their matched section for the UGT2B7*2 allele had the lowest AQ-DNA yields of all tumor cores (< 0.05ng/μL).

Discussion

Archival formalin-fixed, paraffin-embedded (FFPE) tumor tissue is an invaluable potential source of DNA for retrospective pharmacogenetic analyses from large clinical trials, and a number of groups have begun to use this approach for studies in breast cancer3, 12, 13, colon cancer1, 14, and leukemia4. Further, since our initial report2, advances in genotyping using Taqman-based assays and MALDI-TOF mass spectrometry15 have been reported, and recently, high quality genomic copy number analysis using DNA from FFPE tissue16 has been described. Despite the increasing use of FFPE samples for genotype analysis and improved methods, several key issues still significantly inhibit the wider application of this approach. The blocks from historical trials with long follow up are a finite and valuable resource and, quite appropriately, access to these materials is tightly controlled. It is very important, therefore, that sample preparation and assay methods be rigorously optimized in order to maximize the amount of information that can be obtained from the minimum amount of tissue. Access to sections of blocks from important clinical trials is being further complicated by the increasing application of tissue microarray (TMA) technology. IHC-based methods can be applied very efficiently to TMA sections providing high quality data regarding the presence of tumor antigens within all the cases of entire clinical trials with a minimum expenditure of reagents and tumor tissues. As a result, there is an increasing trend toward the preparation of TMAs from the blocks from clinical trials, with the result that access to entire sections of each block is even more limited.

In this study we, therefore, set out to achieve two goals: 1) to develop an approach to maximizing the genotyping assay yield that can be achieved from a given amount of DNA prepared from FFPE materials, and 2) to evaluate the practicality and utility of isolating DNA suitable for genotyping studies from cores of the type used to prepare TMAs. Our hope was that by demonstrating the tremendous potential value of DNA harvested from TMA cores, we would encourage investigators contemplating the production of TMAs to harvest additional tumor cores at the time of TMA manufacture for the preparation of DNA for future genetic analysis.

DNA isolated from FFPE tissues is notoriously challenging to work with due to degradation, shearing, and chemical modification which can cause significant difficulty with downstream applications5-8. This includes PCR-RFLP or Taqman-based assays, since both assays require minimum fragment lengths to be amplified efficiently for analysis2, 5, 17-19. Simple UV absorbance based assessment of DNA quality and quantity does not adequately predict performance in PCR-based applications20. This is likely the result of small DNA fragments that inflate UV absorbance readings, in spite of being too small for efficient PCR. Ideally, genotyping reactions should be set up with just sufficient sample to provide enough DNA template of adequate fragment length to allow efficient amplification. This would minimize the amount of sample used and ensure optimal efficiency of the genotyping assay, thereby maximizing the number of genotyping reactions that can be conducted with a given sample. In this report we describe the strategy we have employed to optimize genotyping assays by: 1) determining the amount of “amplification quality DNA” or “AQ-DNA” in FFPE-derived samples relative to a high-quality DNA calibrator, and 2) establishing the optimum amount of AQ-DNA required for individual assays. Previous reports have proposed or described various strategies for assessing the quality of DNA extracted from FFPE materials for subsequent assay10, 20-23. In fact, a similar approach of quantifying amplifiable DNA was used successfully by Farrand and colleagues to identify samples amenable to loss-of-heterozygosity analyses22. However, this is the first time that the systematic quantification of amplification competent DNA has been applied to the optimization of Taqman-based genotyping assays.

In this report, we demonstrate that a multiplex PCR using 100-700bp amplicons from the GAPDH gene can identify FFPE tissue-derived DNA samples that are too fragmented to genotype. Since the majority of samples amenable to genotype analysis demonstrated amplification of at least the 100bp amplicon in this multiplex analysis, the 100bp amplicon was chosen as a marker of AQ-DNA. Additionally, assessing AQ-DNA based on a minimum fragment size of 100bp is in line with previous reports regarding optimal amplicon size in genotype assessment and other PCR-based analyses from archival tissue 5, 6, 15, 22, 24, 25. We show that the amount of AQ-DNA in FFPE-derived samples, relative to a high quality DNA control, can be quantified based on real-time quantitative PCR-based amplification of this amplicon (Figure 2). Optimization of Taqman-based genotyping assays using this quantification showed that using minimal quantities of AQ-DNA (as little as 50pg per reaction) actually improves genotyping efficiency (Figure 4). Based on this optimization and quantification, sample yield increased to an average of ∼1250 assays (at 100pg/assay) from a single 10μm section (Figure 5). However, it is important to note that the minimal amount of AQ-DNA required for reliable genotyping varied for each individual Taqman-based assay, ranging from 50-150pg in the Taqman assays tested in this report. Importantly, as our measurements of AQ-DNA are based on the use of high-quality DNA as a calibrator, individual assessment of AQ-DNA necessary per reaction will be likely be laboratory-dependent, based on the calibrator used. However, the end results of increased efficiency and increased sample yield will still apply. Individual genotyping assays will also require specific optimization for ideal amplification of both the mutant and wild-type probe sets in each assay. Our optimization of the CYP2D6 1846G>A assay (Table 1) shows that while some genotypes can be assessed using as low as 10pg of input AQ-DNA, some allele-specific Taqman reagents may not amplify efficiently until more AQ-DNA is added to the reaction. This may be most important in the case of heterozygote genotypes. As seen in Table 1, Sample J would have been assessed as a homozygous mutant if only 50pg of DNA were used. Since only one copy of the mutant gene was present per genome, additional DNA was required to achieve optimal amplification of this gene. Taqman-based assays must be optimized to allow for the amplification of both the mutant or wild-type allele for heterozygotes, and should ideally be performed with known controls of each potential genotype.

We also evaluated the usefulness of FFPE block tumor cores and stroma cores, of the type used to generate tissue microarrays (TMAs), in performing genotype analysis. We observed that tumor tissue cores yield a comparable amount of AQ-DNA to a matched section, however, matched stroma cores yielded considerably less AQ-DNA versus both sections and tumor cores (Figure 5). DNA from tumor cores have slightly decreased but comparable quality versus sections as determined by multiplex PCR, however, stroma core-derived samples showed significantly decreased quality (Table 2, Figure S1). These initial observations with DNA obtained from tumor cores suggested that such samples may serve as a useful source of DNA for SNP genotype analysis. We then evaluated real-time PCR amplification efficiency and genotyping success rates for each sample type (Figure 7). To determine genotype accuracy, we compared genotypes across sample types, since we and others have reported that genotypes obtained from tissue sections were 100% concordant with matched germline samples2-4, 12, 13. In spite of decreased success in initial genotype assessment in both tumor and stroma cores, it was possible to determine a genotype from the majority of all sample types with repeated optimization and assessment. However, in this study we did observe a decreased concordance between section and tumor core genotypes for the six CYP2D6 SNPs (82.1%) and the UGT2B7*2 allele (92.3%). It is important to note, however, that for all of the UGT2B7*2 discordances, and 5 of the 7 CYP2D6 discordances, the samples had been identified prior to genotyping as having very low AQ-DNA yields and thus unlikely amenable to genotyping. Further, the remaining two discordances in CYP2D6 genotype were due to a single SNP genotype. This SNP is used by our laboratory as a secondary confirmation of other SNP assays, and an inaccurate genotype from this SNP would not alter the predicted CYP2D6 phenotype of the patient as determined by CYP2D6 activity score (data not shown)26, 27. Genotype discordance is likely due to both low AQ-DNA yields and decreased DNA quality as observed in Figure S1 and Table 2. Assay optimization (input AQ-DNA per assay, cycling conditions, etc.) specifically for tumor core-derived samples is likely necessary and will potentially overcome these discordances. Further, whereas standard operating procedure for genotyping from sections has been to include repeat assays for 10-20% of samples to verify genotypes, increasing validation assays may be necessary for genotyping from tumor cores. This would likely reduce genotyping errors caused by decreased DNA quality. We have been unable to address the decrease in DNA quality, but not total AQ-DNA yield, from tumor cores versus sections. These differences potentially arise from sample processing from the original FFPE block; the shearing generated by either microtome slicing or core punching likely alters how nuclei are exposed and the efficiency of cross-linking removal. Further studies will be necessary to optimize DNA extraction specifically from tumor cores to obtain high AQ-DNA yield and high overall quality. Though tumor cores will potentially serve as a useful source of DNA with increased optimization, stroma cores appear to be unsuitable for reliable SNP genotype analysis by Taqman-based methods, due largely to poor quantity and quality DNA yielded and subsequent assay failure. The unsuitability of stroma cores as a source of DNA may be unique to FFPE samples from breast tumors, as these cores likely contain a high proportion of adipose tissue and therefore have a relatively lower cellular density compared to the nearby tumor tissue. The usefulness of FFPE stroma cores as a source of genomic DNA for pharmacogenomic analyses will most likely need to be individually evaluated for specific tumor types.

As our AQ-DNA measurements were all performed on freshly extracted DNA, we were unable to evaluate changes in AQ-DNA values following long term storage of extracted DNA. However, we have observed that when measuring AQ-DNA from a series of FFPE tissue derived DNA that had been in storage at -20°C for ∼2 years that these samples did not appear to contain lower amounts of AQ-DNA than other freshly extracted samples (data not shown). Based on these data, and the comparison of AQ-DNA yields from FFPE blocks of varying ages (Figure 6), we do not anticipate that increased storage time or sample age will have a major impact on AQ-DNA yields.

These data show that through the technical advances presented, yield and genotyping efficiency of DNA from FFPE tissues can be improved. These advances should greatly increase the number of genotype analyses that can be performed using minimal amounts of valuable archival tissue from clinical trials, reducing tissue waste and increasing cost-effectiveness. Our observation that cores of tumor tissue are a potentially useful source of DNA for pharmacogenomic analyses may also make tissue procurement from clinical trials much easier. With the construction of TMAs becoming more widespread, pathologists now have the option of simply reserving an additional punch for DNA extraction. We expect that these advances will dramatically increase the usefulness of FFPE tissue for genotyping, and further provide an additional source of samples, in TMA cores. This will allow for much simpler sample procurement, and greatly reduce the amount of tissue required for any pharmacogenomic analyses.

Supplementary Material

Fig S1

Figure S1. Multiplex PCR of two matched sample sets of FFPE sections, tumor cores, and stroma cores (see Table 2). Multiplex PCR was performed as described in Materials and Methods. All 7 amplicons were amplified in the high quality control (Ctrl, lane 1). Sections of both samples had the highest amplicon sizes, with other samples showing decreased maximum amplicon sizes.

Table S1

Acknowledgments

We thank Dr. Sharoni Jacobs for assistance in developing the quality control multiplex PCR. MD and JS are grateful for funding from Breakthrough Breast Cancer and the Royal Marsden NIHR Biomedical Research Centre.

This work was supported in part by the Breast Cancer Research Foundation grant N003173 and by U-01 GM61373 and T-32 GM007767 from the National Institute of General Medical Sciences, Bethesda, MD.

References

  • 1.Marsh S, Mallon MA, Goodfellow P, McLeod HL. Concordance of pharmacogenetic markers in germline and colorectal tumor DNA. Pharmacogenomics. 2005;6(8):873–877. doi: 10.2217/14622416.6.8.873. [DOI] [PubMed] [Google Scholar]
  • 2.Rae JM, Cordero KE, Scheys JO, Lippman ME, Flockhart DA, Johnson MD. Genotyping for polymorphic drug metabolizing enzymes from paraffin-embedded and immunohistochemically stained tumor samples. Pharmacogenetics. 2003;13(8):501–507. doi: 10.1097/00008571-200308000-00008. [DOI] [PubMed] [Google Scholar]
  • 3.Schneider BP, Skaar TC, Sledge GW, Badve S, Li L, Flockhart DA. Analysis of angiogenesis genes from paraffin-embedded breast tumor and lymph nodes. Breast Cancer Res Treat. 2006;96(3):209–215. doi: 10.1007/s10549-005-9003-0. [DOI] [PubMed] [Google Scholar]
  • 4.Weiss JR, Baer MR, Ambrosone CB, Blanco JG, Hutson A, Ford LA, et al. Concordance of pharmacogenetic polymorphisms in tumor and germ line DNA in adult patients with acute myeloid leukemia. Cancer Epidemiol Biomarkers Prev. 2007;16(5):1038–1041. doi: 10.1158/1055-9965.EPI-06-0964. [DOI] [PubMed] [Google Scholar]
  • 5.Farragher SM, Tanney A, Kennedy RD, Paul Harkin D. RNA expression analysis from formalin fixed paraffin embedded tissues. Histochem Cell Biol. 2008;130(3):435–445. doi: 10.1007/s00418-008-0479-7. [DOI] [PubMed] [Google Scholar]
  • 6.Greer CE, Peterson SL, Kiviat NB, Manos MM. PCR amplification from paraffin-embedded tissues. Effects of fixative and fixation time. Am J Clin Pathol. 1991;95(2):117–124. doi: 10.1093/ajcp/95.2.117. [DOI] [PubMed] [Google Scholar]
  • 7.Iwamoto KS, Mizuno T, Ito T, Akiyama M, Takeichi N, Mabuchi K, et al. Feasibility of using decades-old archival tissues in molecular oncology/epidemiology. Am J Pathol. 1996;149(2):399–406. [PMC free article] [PubMed] [Google Scholar]
  • 8.Romero RL, Juston AC, Ballantyne J, Henry BE. The applicability of formalin-fixed and formalin fixed paraffin embedded tissues in forensic DNA analysis. J Forensic Sci. 1997;42(4):708–714. [PubMed] [Google Scholar]
  • 9.Hui L, DelMonte T, Ranade K. Genotyping using the TaqMan assay. Curr Protoc Hum Genet. 2008;Chapter 2 doi: 10.1002/0471142905.hg0210s56. Unit 2 10. [DOI] [PubMed] [Google Scholar]
  • 10.van Beers EH, Joosse SA, Ligtenberg MJ, Fles R, Hogervorst FB, Verhoef S, et al. A multiplex PCR predictor for aCGH success of FFPE samples. Br J Cancer. 2006;94(2):333–337. doi: 10.1038/sj.bjc.6602889. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Jin Y, Desta Z, Stearns V, Ward B, Ho H, Lee KH, et al. CYP2D6 genotype, antidepressant use, and tamoxifen metabolism during adjuvant breast cancer treatment. J Natl Cancer Inst. 2005;97(1):30–39. doi: 10.1093/jnci/dji005. [DOI] [PubMed] [Google Scholar]
  • 12.Goetz MP, Rae JM, Suman VJ, Safgren SL, Ames MM, Visscher DW, et al. Pharmacogenetics of tamoxifen biotransformation is associated with clinical outcomes of efficacy and hot flashes. J Clin Oncol. 2005;23(36):9312–9318. doi: 10.1200/JCO.2005.03.3266. [DOI] [PubMed] [Google Scholar]
  • 13.Xie B, Freudenheim JL, Cummings SS, Singh B, He H, McCann SE, et al. Accurate genotyping from paraffin-embedded normal tissue adjacent to breast cancer. Carcinogenesis. 2006;27(2):307–310. doi: 10.1093/carcin/bgi215. [DOI] [PubMed] [Google Scholar]
  • 14.Mort R, Mo L, McEwan C, Melton DW. Lack of involvement of nucleotide excision repair gene polymorphisms in colorectal cancer. Br J Cancer. 2003;89(2):333–337. doi: 10.1038/sj.bjc.6601061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Jaremko M, Justenhoven C, Abraham BK, Schroth W, Fritz P, Brod S, et al. MALDI-TOF MS and TaqMan assisted SNP genotyping of DNA isolated from formalin-fixed and paraffin-embedded tissues (FFPET) Hum Mutat. 2005;25(3):232–238. doi: 10.1002/humu.20141. [DOI] [PubMed] [Google Scholar]
  • 16.Wang Y, Carlton VE, Karlin-Neumann G, Sapolsky R, Zhang L, Moorhead M, et al. High quality copy number and genotype data from FFPE samples using Molecular Inversion Probe (MIP) microarrays. BMC Med Genomics. 2009;2:8. doi: 10.1186/1755-8794-2-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Johnson NA, Hamoudi RA, Ichimura K, Liu L, Pearson DM, Collins VP, et al. Application of array CGH on archival formalin-fixed paraffin-embedded tissues including small numbers of microdissected cells. Lab Invest. 2006;86(9):968–978. doi: 10.1038/labinvest.3700441. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18.Koch I, Slotta-Huspenina J, Hollweck R, Anastasov N, Hofler H, Quintanilla-Martinez L, et al. Real-time quantitative RT-PCR shows variable, assay-dependent sensitivity to formalin fixation: implications for direct comparison of transcript levels in paraffin-embedded tissues. Diagn Mol Pathol. 2006;15(3):149–156. doi: 10.1097/01.pdm.0000213450.99655.54. [DOI] [PubMed] [Google Scholar]
  • 19.Andreassen CN, Sorensen FB, Overgaard J, Alsner J. Optimisation and validation of methods to assess single nucleotide polymorphisms (SNPs) in archival histological material. Radiother Oncol. 2004;72(3):351–356. doi: 10.1016/j.radonc.2004.07.006. [DOI] [PubMed] [Google Scholar]
  • 20.Lehmann U, Kreipe H. Real-time PCR analysis of DNA and RNA extracted from formalin-fixed and paraffin-embedded biopsies. Methods. 2001;25(4):409–418. doi: 10.1006/meth.2001.1263. [DOI] [PubMed] [Google Scholar]
  • 21.Siwoski A, Ishkanian A, Garnis C, Zhang L, Rosin M, Lam WL. An efficient method for the assessment of DNA quality of archival microdissected specimens. Mod Pathol. 2002;15(8):889–892. doi: 10.1097/01.MP.0000024288.63070.4F. [DOI] [PubMed] [Google Scholar]
  • 22.Farrand K, Jovanovic L, Delahunt B, McIver B, Hay ID, Eberhardt NL, et al. Loss of heterozygosity studies revisited: prior quantification of the amplifiable DNA content of archival samples improves efficiency and reliability. J Mol Diagn. 2002;4(3):150–158. doi: 10.1016/S1525-1578(10)60696-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Gilbert MT, Sanchez JJ, Haselkorn T, Jewell LD, Lucas SB, Van Marck E, et al. Multiplex PCR with minisequencing as an effective high-throughput SNP typing method for formalin-fixed tissue. Electrophoresis. 2007;28(14):2361–2367. doi: 10.1002/elps.200600589. [DOI] [PubMed] [Google Scholar]
  • 24.Godfrey TE, Kim SH, Chavira M, Ruff DW, Warren RS, Gray JW, et al. Quantitative mRNA expression analysis from formalin-fixed, paraffin-embedded tissues using 5′ nuclease quantitative reverse transcription-polymerase chain reaction. J Mol Diagn. 2000;2(2):84–91. doi: 10.1016/S1525-1578(10)60621-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Abrahamsen HN, Steiniche T, Nexo E, Hamilton-Dutoit SJ, Sorensen BS. Towards quantitative mRNA analysis in paraffin-embedded tissues using real-time reverse transcriptase-polymerase chain reaction: a methodological study on lymph nodes from melanoma patients. J Mol Diagn. 2003;5(1):34–41. doi: 10.1016/S1525-1578(10)60449-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Rae JM, Sikora MJ, Henry NL, Li L, Kim S, Oesterreich S, et al. Cytochrome P450 2D6 activity predicts discontinuation of tamoxifen therapy in breast cancer patients. Pharmacogenomics J. 2009 doi: 10.1038/tpj.2009.14. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Gaedigk A, Simon SD, Pearce RE, Bradford LD, Kennedy MJ, Leeder JS. The CYP2D6 activity score: translating genotype information into a qualitative measure of phenotype. Clin Pharmacol Ther. 2008;83(2):234–242. doi: 10.1038/sj.clpt.6100406. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Fig S1

Figure S1. Multiplex PCR of two matched sample sets of FFPE sections, tumor cores, and stroma cores (see Table 2). Multiplex PCR was performed as described in Materials and Methods. All 7 amplicons were amplified in the high quality control (Ctrl, lane 1). Sections of both samples had the highest amplicon sizes, with other samples showing decreased maximum amplicon sizes.

NIHMS228101-supplement-Fig_S1.eps (859.3KB, eps)
Table S1
NIHMS228101-supplement-Table_S1.eps (947.6KB, eps)

RESOURCES