Preparation of DNA from cytological material

Effects of Fixation, Staining, and Mounting Medium on DNA Yield and Quality


  • Annika Dejmek MD, PhD,

    1. Department of Clinical Pathology, University and Regional Laboratories Region Skåne, Malmo, Sweden
    2. Department of Laboratory Medicine, Center for Molecular Pathology, Lund University, Lund, Sweden
    Search for more papers by this author
  • Nooreldin Zendehrokh PhD,

    1. Department of Clinical Pathology, University and Regional Laboratories Region Skåne, Malmo, Sweden
    Search for more papers by this author
  • Malgorzata Tomaszewska MSc,

    1. Department of Clinical Chemistry, University and Regional Laboratories Region Skåne, Malmö, Sweden
    Search for more papers by this author
  • Anders Edsjö MD, PhD

    1. Department of Clinical Pathology, University and Regional Laboratories Region Skåne, Malmo, Sweden
    2. Department of Laboratory Medicine, Center for Molecular Pathology, Lund University, Lund, Sweden
    3. Department of Clinical Pathology and Cytology, Sahlgrenska University Hospital, Gothenburg, Sweden
    4. Department of Pathology, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden
    Current affiliation:
    1. Department of Clinical Pathology and Cytology, Sahlgrenska University Hospital, Gothenburg, Sweden
    2. Department of Pathology, Sahlgrenska Academy, University of Gothenburg, Gothenburg, Sweden
    Search for more papers by this author

Corresponding author: Anders Edsjö, MD, PhD, Department of Clinical Pathology and Cytology, Sahlgrenska University Hospital, Gula Stråket 8, 413 45 Gothenburg, Sweden; Fax: (011) 46 (0)31 41 72 83;



Personalized oncology requires molecular analysis of tumor cells. Several studies have demonstrated that cytological material is suitable for DNA analysis, but to the authors' knowledge there are no systematic studies comparing how the yield and quality of extracted DNA is affected by the various techniques used for the preparation of cytological material.


DNA yield and quality were compared using cultured human lung cancer cells subjected to different preparation techniques used in routine cytology, including fixation, mounting medium, and staining. The results were compared with the outcome of epidermal growth factor receptor (EGFR) genotyping of 66 clinical cytological samples using the same DNA preparation protocol.


All tested protocol combinations resulted in fragment lengths of at least 388 base pairs. The mounting agent EcoMount resulted in higher yields than traditional xylene-based medium. Spray and ethanol fixation resulted in both a higher yield and better DNA quality than air drying. In liquid-based cytology (LBC) methods, CytoLyt solution resulted in a 5-fold higher yield than CytoRich Red. Papanicolaou staining provided twice the yield of hematoxylin and eosin staining in both liquid-based preparations. Genotyping outcome and quality control values from the clinical EGFR genotyping demonstrated a sufficient amount and amplifiability of DNA in both spray-fixed and air-dried cytological samples.


Reliable clinical genotyping can be performed using all tested methods. However, in the cell line experiments, spray- or ethanol-fixed, Papanicolaou-stained slides provided the best results in terms of yield and fragment length. In LBC, the DNA recovery efficiency of the preserving medium may differ considerably, which should be taken into consideration when introducing LBC. Cancer (Cancer Cytopathol) 2013;121:344–353. © 2013 American Cancer Society.


As personalized therapy continues to gain ground, molecular pathology (ie, molecular analysis) of tumor cells is rapidly becoming an integral part of clinical pathology.[1] Starting with the effect of imatinib on gastrointestinal stromal tumors, several tyrosine kinase inhibitors have been introduced for the targeted treatment of tumors. Because the response to treatment often is predicted by the presence or absence of activating mutations, genotyping of the tyrosine kinase family of genes and of genes encoding proteins in downstream signaling cascades has become an important part of molecular pathology.[2] The introduction of epidermal growth factor receptor (EGFR) and KRAS genotyping to predict response to treatment with therapeutic agents specifically targeting the EGFR in patients with non-small cell lung cancer[3-5] and colorectal carcinoma[6-8] served as a starting point for the clinical genotyping of solid tumors on a larger scale. Recently, the introduction of BRAF inhibition in patients with malignant melanoma[9, 10] has accelerated the need for sensitive and robust mutation analyses.[11]

In patients with colorectal carcinoma or malignant melanoma, diagnosis is usually based on formalin-fixed, paraffin-embedded (FFPE) material. Conversely, in the diagnosis of lung cancer, cytology plays an important role. The traditional method of diagnosing lung cancer is by bronchial brushings and washings combined with forceps biopsies obtained at fiber bronchoscopy when possible, or transthoracic fine-needle aspiration if the tumor is situated within the periphery of the lung.[12, 13]

In a clinical setting, a cytological specimen may be the only material available for the determination of EGFR mutation status. One example is a case in which the cytological material obtained at bronchoscopy may be diagnostic for cancer, whereas the corresponding biopsy may contain only normal tissue or necrotic material, rendering it unevaluable. In some cases, obtaining biopsy material is considered too hazardous due to the risk of bleeding. Moreover, in patients with advanced lung cancer, surgery is usually not performed and cytological material may be the only material available for EGFR mutation analysis when it is most needed. In addition, in patients with locally advanced or metastatic disease, using cytological samples for these tests makes it possible to avoid noncurative surgery.

Molecular analysis requires optimal processing and good sample quality. Genotyping of tumor samples can be performed using various methods, each with their own limitations. Because of the heterogenic nature of most analyzed samples, a key parameter for successful analysis is the percentage of mutant alleles needed for mutation detection. The level of detection of the techniques varies from 1% to 5% mutant alleles for highly sensitive clinical methods such as amplification refractory mutation system (ARMS) polymerase chain reaction (PCR) or pyrosequencing to 20% to 30% for dideoxysequencing.[14, 15] Two other key parameters are the quality and quantity of extracted DNA needed. The majority of clinical molecular pathology assays are designed for fragment lengths varying between 100 base pairs (bp) and 400 bp and typically require 1nanogram (ng) to 50 ng per analysis.

The article by Pirker et al states the importance of appropriate sampling techniques, the necessity of standardizing tumor specimen handling including fixation, and the need to validate the tests used to assess EGFR status.[16] Similar recommendations were proposed by Eberhard et al.[17]

The analysis of mutational status in tumor cells has primarily been performed on FFPE material, but analysis of cytological samples has started to make its way into clinical practice. As stated in a review by da Cunha Santos et al, many studies of EGFR status have shown that cytological material is suitable for EGFR status analysis.[18] In a recent commentary, Knoepp and Roh stated their preference for direct smears versus other cytopathological preparation methods as ancillary techniques because they allow for morphological evaluation and the selection of areas with a sufficient amount of relevant cells.[19]

The techniques used to fix and prepare cytological material vary considerably. The methods used in clinical cytology have been developed to ensure optimal morphology, but to the best of our knowledge no systematic studies have been performed to evaluate their effect on the submicroscopic level. The studies analyzing EGFR status in cytological material have been performed in different types of samples, and processed using different cytological techniques and different detection methods.[20-29] Therefore, no conclusions regarding the efficiency of DNA recovery using the different techniques can be drawn. To address this question, the techniques of interest need to be applied to identical material, using a fixed set of methods for the assessment of quantity and quality.

Using this approach, we set out to test the effect of the different types of standard fixation and preparation methods currently in use in our laboratory on DNA quantity and quality, using the human lung cancer cell lines A549 (EGFRwt) and NCI-H1650 (EGFRexon 19 del).


Cell Culture

Two human lung cancer cell lines, A549 (American Type Culture Collection [ATCC] no. CCL-185 [EGFRwt]) and NCI-H1650 (ATCC no. CRL-5883 [EGFRexon 19 del]) were used. The cell lines were maintained in RPMI-1640 supplemented with glutamine, 10% fetal bovine serum, and 1% penicillin-streptomycin in a humidified incubator with 5% carbon dioxide at 37°C. The cell lines were monitored for altered morphology. Each comparison experiment was replicated on 2 separate cell harvests.

Fixation and Mounting

To emulate cytological material, the cells were trypsinized and centrifuged for 1 minute at 110 g, washed with phosphate-buffered saline (PBS), and centrifuged again, after which the supernatant fluid was removed.

For conventional cytology, the cell sediment was smeared onto slides and fixed using either air-drying, immersion in 96% ethanol or by an isopropanol-based spray fixative (Histolab, Gothenburg, Sweden). Both ethanol-fixed and spray-fixed slides were stained according to the Papanicolaou (Pap) method and hematoxylin and eosin staining using Harris hematoxylin, and the air-dried slides were stained using May-Grünwald-Giemsa (MGG) staining.

For liquid-based cytology (LBC), 2 different fixatives were used: CytoLyt solution (Cytyc Corporation, Marlborough, Mass) and CytoRich Red collection fluid (Fisher Scientific UK Ltd, Loughborough, UK). After washing with PBS, the cells were transferred to 1 tube containing CytoLyt solution and 1 tube containing CytoRich Red and left for 30 minutes. After removal of the supernatant fluid, the cells fixed with CytoLyt solution were spun (10 minutes at 970 g), the supernatant fluid was removed, and the sediment was smeared onto the slides. The slides were placed into 96% ethanol for 10 minutes before staining was performed according to the Pap method.

After fixation in CytoRich Red, the cell suspensions were centrifuged at 970 g for 10 minutes. The supernatant fluid was removed, 10 mL of deionized water was added, and the suspension was centrifuged at 970 g for 10 minutes. After the supernatant fluid had been decanted, the cell pellet was resuspended in 1 mL of deionized water. SurePath PreCoat slides (TriPath Imaging Inc, Burlington, NC) were placed in slide racks, a settling chamber (Settling Chamber 240; TriPath Imaging) was placed on each slide and locked into place, 100 μL of the cell suspension was placed in the settling chamber, and the cells were allowed to settle for 12 minutes. The slides were then placed in 95% ethanol for 20 minutes and stained according to the Pap method.

The slides were mounted with 2 different mounting media: the non–xylene-based EcoMount (Biocare Medical LLC, Concord, Calif) and the xylene-based Pertex (Leica Biosystems, Buffalo Grove, Ill).

Cell Counting and Lysis

To remove the coverslips, the slides mounted with EcoMount were placed in Histo-Clear (AGTC Bioproducts Ltd, Hessle, UK) and those mounted with Pertex were placed in xylene. After removal of the coverslips, the slides were rehydrated in a graded ethanol series (99%, 95%, and 70%) for 1 minute each and then air dried.

To estimate the number of cells within a defined area on the slides, the slides were examined using light microscopy (× 200; Olympus BX45 microscope; Olympus Corporation, Center Valley, Pa). In an area of the slide considered to be homogenous, cell density was estimated by manually counting 6 areas, each measuring 1 mm2. Based on the cell density, an estimate of the slide area necessary for a given total cell number was calculated and marked on the slide; the coverslips were removed; the cells in the excess area were scraped away; and the cells within the selected area were lysed directly on the slide with 180 μL of Buffer ATL and tissue lysis buffer (both from Qiagen, Hilden, Germany) and transferred to 1.5-mL Eppendorf tubes. Experiments were performed with lysates nominally containing 5000, 10,000, 20,000, 40,000, and 80,000 cells.

DNA Preparation and Quantitation

For the preparation of genomic DNA, the QIAamp DNA Micro Kit (Qiagen) was used according to the manufacturer's instructions but with the proteinase K incubation step prolonged to an overnight incubation. DNA quantity was assessed using a Beckman DTX 880 multimode detector (Beckman Coulter, Brea, Calif) for measuring fluorescence-labeled, double-stranded genomic DNA using the PicoGreen dsDNA Quantitation Reagent Kit (Molecular Probes, Eugene, Ore). All samples were analyzed in triplicate.

PCR-Based Analyses

Fragment lengths were assessed using endpoint PCR with 40 cycles on an ABI GeneAmp 9700 PCR system using Perkin Elmer-PCR buffer and AmpliTaq gold polymerase (all from Applied Biosystems, Foster City, Calif). The primers used amplified stretches of genomic DNA of increasing lengths (209 bp, 388 bp, 578 bp, and 760 bp) and were chosen from already extensively tested primer pairs directed against the KRAS, SRGAP2 (SLIT-ROBO Rho GTPase-activating protein 2), and EGFR genes, respectively (Table 1). EGFR genotyping was performed according to the manufacturer's instructions using the Therascreen EGFR PCR Kit (Qiagen) and an Applied Biosystems 7900HT PCR system.

Table 1. Primer Sequences Used for Fragment Length Analysis
PrimerSequence 5′ to 3′Amplicon Length
  1. Abbreviations: bp, base pair; EGFR, epidermal growth factor receptor; ex, exon; SRGAP2, SLIT-ROBO Rho GTPase-activating protein 2.

EGFR_ex18_FtgtaaaacgacggccagtATGAGCTGGCAAGTGCCGTG578 bp

Statistical Analysis

Data were evaluated using the general linear model (GLM) routine of Minitab 12 statistical software (Minitab Inc, State College, Pa) for multiple comparisons with the Tukey correction.


Taking advantage of the high sensitivity and wide dynamic range of fluorescent labeling of double-stranded DNA,[30, 31] we first assessed DNA yields.

When comparing the xylene-based mounting medium Pertex with the water-soluble EcoMount (Fig. 1), both of which are used in clinical practice, significantly higher yields were observed for EcoMount (GLM with cell batch as a random factor; P=.024). Consequently, EcoMount was used as the mounting medium in all subsequent experiments.

Figure 1.

Comparison of yields using EcoMount (Biocare Medical LLC, Concord, Calif) or Pertex mounting medium (Leica Biosystems, Buffalo Grove, Ill) is shown. Samples were air dried and stained with May-Grünwald-Giemsa staining. The means and standard error of the mean from 2 batches with 3 replicates of each case are shown.

The differences between air drying (“air”) and isopropanol-based spray fixation (“spray”) are shown in Figure 2. Significantly higher yields were observed with the spray (GLM with cell batch as a random factor; P=.007)

Figure 2.

Comparison of yields using fixation with isopropanol-based spray (Papanicolaou staining) or air drying (May-Grünwald-Giemsa staining). EcoMount (Biocare Medical LLC, Concord, Calif) was used as the mounting medium. The means and standard error of the mean from 2 batches with 3 replicates of each case are shown.

The yields after fixation using ethanol or isopropanol-based spray are compared in Figure 3. Even when correcting for batch, there was no difference noted between the 2 fixation methods.

Figure 3.

Comparison of yields using fixation with isopropanol-based spray (“spray”) or ethanol (EtOH) (90%) is shown. Cells were stained using the Papanicolaou method and EcoMount (Biocare Medical LLC, Concord, Calif) was used as the mounting medium. The means and the standard error of the mean from 2 batches with 3 replicates of each case are shown.

As shown in Figure 4, there was no significant difference between the slides that underwent fixation with CytoLyt solution and those that underwent fixation with ethanol, but CytoRich Red gave a significantly lower yield than the others (P=.0001). For all 3 fixation agents used there was also a significantly higher yield with Pap staining (3-way GLM with cell batch as a random factor; P=.007) (Fig. 4).

Figure 4.

Comparison of yields are shown between 2 liquid-based methods and ethanol (EtOH) fixation, each with nominally 40,000 cells, stained with hematoxylin and eosin (HTX) and Papanicolaou (PAP) and with EcoMount (Biocare Medical LLC, Concord, Calif) used as the mounting medium. According to the manufacturer's instructions, precoated slides were used for cells in CytoRich Red, whereas routine slides were used for the other fixations.

In addition to yield, we characterized the prepared genomic DNA in terms of fragmentation as assessed by PCR amplification of amplicons with increasing lengths. DNA was amplified and separated using gel electrophoresis. The presence or absence of a visible band of the correct size was assessed and the band intensity was semiquantified (Table 2).

Table 2. Presence of Visible Bands of DNA Fragments in Extracts With Different Fixations (Median Value of 2 Batches With 3 Replicates Is Shown)a
Nominal Cell CountAirSprayCytoRich Red
209 bp388 bp578 bp760 bp209 bp388 bp578 bp760 bp209 bp388 bp578 bp760 bp
  1. Abbreviation: bp, base pair.

  2. a

    Scoring: strong band=2; distinct band=1; indefinite=0; no band=−1.


In the air-dried, MGG-stained samples, amplicon sizes of 388 bp could be consistently amplified, while amplification of a 578-bp amplicon proved to be difficult, with only a barely visible band in some of the samples with 40,000 cells. In the spray-fixed, Pap-stained samples, bands from the longest 760-bp amplicon could be observed in most samples, even with 5,000 cells as the input. CytoRich Red preserved the 578-bp amplicon well but not the 760-bp amplicon. The differences in yield in favor of ethanol fixation and Pap staining was thus reflected in higher DNA quality measured as fragment lengths.

To further test the amplifiability of air-dried and spray-fixed cytological material, we used an ARMS EGFR genotyping assay as part of our routine for clinical samples with a low percentage of neoplastic cells (Table 3). In the assay, increasing amounts of cells were assayed for 28 different EGFR mutations. To mimic a clinical setting, DNA from cells with an EGFR exon 19 deletion was mixed in a 1:20 ratio with DNA from the EGFRwt cell line used for the initial experiments. The EGFR deletion was detected in all samples with the exception of those in which DNA from the EGFRwt cell line only was used as a negative control for the assay. However, the control Ct, used to assess whether the specified level of detection of the assay (1% mutated alleles at a control Ct of 26) was reached, was consistently lower in DNA from the spray-fixed cytological material (Table 3). For samples with 5000 to 20,000 cells, the difference exceeded 2 Ct steps, thereby lowering, in theory, the needed percentage of mutated alleles needed for detection 22 times when using ethanol-fixed material.

Table 3. EGFR Genotyping of Cells Fixed With Air Drying (MGG-Stained) or Isopropanol-Based Spray (Pap-Stained)a
 Air-Dried, MGG-Stained
A549 (EGFRwt)95%95%95%95%100%
H1650 (EGFRexon19 del)5%5%5%5%
No. of cells500010,00020,00040,00040,000
Mean control Ct32.1731.1830.0028.8328.86
EGFR statusExon 19 delExon 19 delExon 19 delExon 19 delWT
 Spray-Fixed, PAP-Stained
  1. Abbreviations: Ct, control value; del, deletion; EGFR, epidermal growth factor receptor; MGG, May-Grünwald-Giemsa; Pap, Papanicolaou; WT, Wild type.

  2. a

    DNA from cell lines with a deletion in exon 19 of EGFR, diluted 1:20 with DNA from EGFR wt cells, were analyzed with a CE-IVD-marked (Conformité Eropéenne, In Vitro Diagnostics), allele-specific polymerase chain reaction analysis kit. The results were compiled from 2 experiments with an identical genotyping outcome and the control Ct values were expressed as the mean between the values of the 2 experiments.

A549 (EGFRwt)95%95%95%95%100%
H1650 (EGFRexon19 del)5%5%5%5%
No. of cells500010,00020,00040,00040,000
Mean control Ct29.5428.7327.5527.1627.18
EGFR statusExon 19 delExon 19 delExon 19 delExon 19 delWT

Finally, we set out to compare the Ct values noted in the cell line experiments with those observed in the analyses of clinical cytological samples and discover whether the differences between the fixation/staining methods that differed the most under controlled conditions also could be detected in our highly heterogenous clinical material. Twenty-seven samples that were previously genotyped with the same EGFR genotyping assay used in the cell line experiments and for which fixation and staining details were known were retrieved from our files and the genotyping results were compared. Twenty of these cases were air-dried, demonstrating a mean control Ct of 26.8. The 7 ethanol-fixed samples had a mean control Ct of 26.6. The cases were too few to assess whether the genotyping outcome is reasonable, but examining all EGFR assays on cytological assays performed to date (n=66), the percentage of EGFR mutations (12%) appears to be similar to that of histological material and within the range expected for a Western European population.[32]


The finding that the measured DNA amounts were very similar at different nominal cell counts gives confidence to the way in which the nominal cell counts were established. The somewhat lower yields at the 5000-cell level are likely to be caused by adsorption of DNA to surfaces and similar losses. The absolute amounts per nominal cell were of a reasonable order. There was a modest variability from experiment to experiment due to variables that cannot be explained. However, our results are based on direct comparisons of the different preparation methods on the same culture batches and although the average values of DNA in 2 batches are reported for simplicity in Figures 1 to 4, the effect of the batch was included as a random factor in the statistical analysis.

The current study is targeted toward the reuse of routinely acquired cytological material. Obviously, the choice of the routine sample preparation method in each laboratory is a cumulative result of experience, not least of which is familiarity with the morphological features of the cells in a specific type of sample preparation. To facilitate laboratory routines and ensure optimal use of the specimen, an ideal procedure should be useful for ancillary molecular techniques without compromising the morphological quality. The introduction of different fixatives depending on the type of test to be performed would be a considerable disadvantage because it would lead to time-consuming verification of quality with respect to morphology for each new protocol introduced.

One part of sample preparation, which does not interfere with morphological assessment, is the mounting medium that is essential for making permanent slides. We tested 2 different mounting media, Pertex and EcoMount. Pertex is a traditional mounting medium based on xylene, an aromatic hydrocarbon. The toxicity of xylene has been known since the 1970s and its occupational hazards are well documented.[33] EcoMount is a low-hazard, organic, polymer-based mounting medium used as a substitute for xylene-based media. We found that the use of EcoMount improved the yield of DNA (Fig. 1). Thus, it is preferable for reasons of quality, environment, and workplace hygiene.

In our cell line experiments, fixation using spray and ethanol were found to be equally efficient in the extraction of DNA, and both were significantly better than air-dried specimens (Figs. 2 and 3). MGG is the traditional routine staining for air-dried cytological material. Thus, it cannot be determined whether the poorer results noted with air-dried MGG-stained specimens are due to the fixation, staining, or a combined effect of both.

Ethanol-fixed or spray-fixed samples are usually stained with hematoxylin and eosin or are stained according to the Pap method. In the traditionally wet-fixed sample preparation, we did not observe any difference in the DNA yield between these 2 stains (Fig. 4). The finding that Pap staining is compatible with DNA extraction has been amply demonstrated in reports of successful DNA extraction from Pap-stained cervical cytology smears (Pap smears) since testing for the human papillomavirus was introduced early in the screening process for precancerous cervical lesions. It is known that long-term storing induces DNA degradation but the method of fixation appears to be more important than storage time.[34] There are conflicting claims concerning the stability of DNA in Pap-stained material. Boulet et al performed DNA extraction on 10-year-old archival Pap smears using 3 different commercial DNA isolation techniques and found that all provided DNA that was suitable for amplification and that there was no difference noted in the amount of amplified DNA between the methods.[35] Concurrently, Jacobs et al managed to extract amplifiable DNA from 13-year to 14-year-old Pap-stained cervical smears.[36] Conversely, Killian et al compared Diff-Quik–stained and Pap-stained archival fine-needle aspiration smears that were older than 10 years and found that Diff-Quik was superior in terms of DNA preservation and integrity.[37]

From a clinical standpoint, it can be argued that a 10-year–long storage time is not relevant for the selection of patients who would benefit from targeted lung cancer therapy but, for research purposes, DNA degradation over time can of course be of importance.

LBC is well established in gynecological cytology and is becoming increasingly used in other types of cytological specimens. It has the advantage of reducing background material such as blood and debris and offers the possibility of dividing the sample equally on several generally identical slides.

LBC has several potential advantages when the material is being used for molecular techniques. The homogenous cell distribution facilitates assessment of the percentage of malignant cells in a specimen in which benign or normal cells predominate. Furthermore, multiple slides can be prepared from the same specimen and therefore a routinely stained preparation does not have to be destroyed as is the case when cells are scraped off routinely stained smears, which might have medicolegal consequences.

Different types of LBC use different preservation media with different compositions.[38-40] A considerable number of studies of DNA extraction in LBC preparations have been performed and have demonstrated that both the specific variant of fixative and liquid-based preparation technique used might affect the usefulness of the method for DNA retrieval.[41] Thus, each fixative has to be tested and no general conclusions can be drawn regarding LBC.

When comparing CytoRich Red collecting fluid with CytoLyt solution, we found that the latter was superior in terms of the yield of suitable DNA (Fig. 4). Malapelle et al managed to extract DNA that was suitable for EGFR mutation analysis in CytoLyt-fixed clinical cytological samples from patients with non-small cell lung cancer.[42] It has also been shown that PreservCyt solution (Cytyc Corporation), a similar fixative that is suitable for long-term storage, preserves nucleic acids (RNA and DNA).[38, 43, 44] We did not post-fix in PreservCyt solution as we processed the specimens immediately after they had been stored in CytoLyt solution for 30 minutes.

Both PreservCyt and CytoLyt solution are methanol-based, whereas CytoRich Red contains a small amount (< 1%) of formaldehyde. The poor results obtained with CytoRich Red may be explained by the presence of the formaldehyde because formaldehyde may cause DNA degradation and modification by cross-linking of cytosine residues on either strand either by itself[45] or in combination with subsequent exposure to ethanol.[46]

The quality of the extracted DNA, as measured by fragment analysis (Table 2), demonstrated that all the tested cytological techniques achieved fragment lengths typically used for molecular pathology, in which the tests are designed to handle the fragmented and modified DNA extracted from FFPE tissue. However, spray fixation was found to be superior for preserving up to the tested limit of 760 bp. Air-dried specimens gave the shortest fragment lengths. A possible bias could be that the cell line cells were washed with PBS, which is not common for routine clinical samples. We took care to limit the time of PBS exposure to minimize the extent of hydration of the cells and no morphological changes were observed in cells either before or after PBS treatment, indicating that the cells were not affected.

Surprisingly in view of the low yield, CytoRich Red demonstrated DNA preservation that was nearly as good as that of spray-fixation.

The genotyping results, with the possibility of detecting the most common activating EGFR mutation under all tested conditions and an expected genotyping outcome of clinical EGFR analyses in samples treated according to different protocols (n=66 samples), appear to further strengthen the conclusion that DNA from all the cytological material analyzed was of sufficient quantity and quality for reliable clinical genotyping. In a controlled cell line setting, a consistent difference in amplifiability as measured by control Ct could be noted in favor of ethanol-fixed cells, a potentially crucial feature in cases with few cells and for those samples with a low percentage of neoplastic cells.

In the relatively few clinical cases (n=27 cases) analyzed with ARMS quantitative PCR for which there was detailed information concerning fixation and staining, we noted no statistically significant difference in control Ct values between the different fixation methods or staining procedures used. However, a difference would have to be larger than that measured in the cell line experiments to compensate for the considerable differences in cell number, percentage of neoplastic cells, ploidy, and EGFR amplification status that was observed in clinical material and is difficult to compensate for. In addition, in the clinical analyses, a compensation for low DNA concentrations is performed when possible, loading a higher volume of the sample in the genotyping analysis. A difference in control Ct values due to yield, the likely prime explanation of the Ct differences noted in the cell line experiments, would thus only be expected in a minority of cytological samples with such a low cell content that the maximum volume of DNA is analyzed. Regardless of the relative differences noted between the various preparation methods of cytological material, even the air-dried cytological specimens have been reported to demonstrate an amplifiability that is superior to that of the FFPE material in the current clinical series (data not shown).


Although reliable genotyping can be performed using all tested methods, fixation, staining, and mounting medium all appear to affect the yield and quality of extracted DNA. In our cell line experiments, Pap-stained ethanol-fixed/spray-fixed slides provided better results when compared with the other fixation and staining methods. Although the extrapolation of cell line-based results to clinical samples should be done with caution, we argue that the differences noted call for a structured validation of the extracted DNA from differently processed cytological material. Xylene-based mounting media should be avoided because of their occupational hazard and can be replaced with a polymer-based medium, resulting in an even higher yield. When introducing LBC, the DNA recovery efficiency of the preserving medium should be taken into consideration.


Supported by AstraZeneca. The company has had access to the results but has not been involved in data interpretation or the writing of the article.


Dr. Edsjö has performed consulting work for Amgen, AstraZeneca, and Roche.