the impact of pre-analytical processing on staining quality for h&e, dual hapten, dual color in...

Methods 52 (2010) 287–300

Contents lists available at ScienceDirect

Methods

journal homepage: www.elsevier .com/locate /ymeth

The impact of pre-analytical processing on staining quality for H&E, dualhapten, dual color in situ hybridization and fluorescent in situ hybridization assays

Andrea Babic ⇑, Isabell R. Loftin, Stacey Stanislaw, Maria Wang, Rachel Miller, Stephanie M. Warren,Wenjun Zhang, Alexandria Lau, Melanie Miller, Ping Wu, Mary Padilla, Thomas M. Grogan,Lidija Pestic-Dragovich, Abigail S. McElhinnyVentana Medical Systems, Inc., Tucson, AZ 85755, USA

a r t i c l e i n f o

Article history:Available online 31 August 2010

Keywords:HER2ISHDual ISHFISHTissue fixationTissue processing

1046-2023/$ - see front matter � 2010 Elsevier Inc. Adoi:10.1016/j.ymeth.2010.08.012

Abbreviations: ASCO/CAP, American Society of ClAnatomical Pathologists; Dual ISH, dual color dualFFPE, formalin-fixed paraffin embedded; FISH, fluoreH&E, hematoxylin and eosin; IHC, immunohistochemtion; NBF, neutral-buffered formalin; Zn, zinc.⇑ Corresponding author.

E-mail address: [email protected] (A

a b s t r a c t

With the advent of personalized medicine, anatomic pathology-based molecular assays, including in situhybridization (ISH) and mRNA detection tests, are performed routinely in many laboratories and haveincreased in their clinical importance and complexity. These assays require appropriately fixed tissuesamples that preserve both nucleic acid targets and histomorphology to ensure reliable test results fordetermining patient treatment options. However, all aspects of tissue processing, including time until tis-sue fixation, type of fixative, duration of fixation, post-fixation treatments, and sectioning of the sample,impact the staining results. ASCO/CAP has issued pre-analytical guidelines to standardize tissue process-ing for HER2 testing in breast carcinoma specimens: 10% neutral-buffered formalin (NBF) with a fixationtime from at least 6 to 48 h [1]. Often, this recommendation is not followed to the detriment of stainingresults [2].

In this paper, we used a human breast carcinoma cell line (MCF7) generated as xenograft tumors as amodel system to analyze the effects of different pre-analytical conditions on ISH staining. We performedH&E, FISH and dual colorimetric HER2 ISH assays using specimens fixed across a range of times in six dif-ferent commonly used fixatives. Additionally, we investigated the effects of varying tissue section thick-ness, which also impacted the quality of ISH staining. Finally, we evaluated the effects of three differentdecalcifying solutions on human breast specimens, typically a treatment that occurs post-fixation forevaluating metastases to bone. The results indicate that time and type of fixation treatment, as well asappropriate tissue thickness and post-fixation treatment, all contribute to the quality of ISH stainingresults. Our data support the ASCO/CAP recommendations for standardized tissue processing (at least6 h in formalin-based fixatives and 4 lm section thickness) and indicate that certain fixatives andpost-fixative treatments are detrimental to molecular staining results.

� 2010 Elsevier Inc. All rights reserved.

1. Introduction

Anatomic pathology-based assays are increasing in their com-plexity and relevance as the field of personalized medicine pro-gresses. As diagnostic tests are developed for specific biomarkers,pathologists and oncologists have the potential to test the samespecimen with multiple assays to determine the best course oftherapy. Formalin-fixed, paraffin embedded (FFPE) tissue speci-

ll rights reserved.

inical Oncologists/College ofhapten in situ hybridization;scence in situ hybridization;istry; ISH, in situ hybridiza-

. Babic).

mens, when sectioned and placed onto glass slides and tested withspecific stains and assays, enable pathologists to observe histomor-phology and specific molecular characteristics of tumor tissue andindividual tumor cells. Therefore, pre-analytical conditions mustpreserve important histomorphological features and subcellularcomponents, including antigens and nucleic acids, the targets ofstains and assays. However, fixatives that produce the best histo-morphological appearance in tissue are not necessarily optimalfor preserving the molecular state of the cells. A balance betweenthese functions in the chemical composition of tissue fixatives isrequired for current histopathological practices.

In addition to type of fixative [1,3–6] studies have determinedthat other pre-analytical factors are critical for appropriate stainingresults for ISH and IHC-based assays. These include the time fromexcision to fixation and duration of fixation [1,7–15] the volume oftissue to the volume of fixative [1,9], and post-processing factors

http://dx.doi.org/10.1016/j.ymeth.2010.08.012

mailto:[email protected]

http://dx.doi.org/10.1016/j.ymeth.2010.08.012

http://www.sciencedirect.com/science/journal/10462023

http://www.elsevier.com/locate/ymeth

288 A. Babic et al. / Methods 52 (2010) 287–300

for specific specimen types [16]. Clearly, standardization of pre-analytical processing is required to obtain optimal, accurate andreproducible results from diagnostic assays. In recognition of theimportance of this idea, the American Society of Clinical Oncolo-gists/College of Anatomical Pathologists (ASCO/CAP) recently pub-lished their recommendations for several factors related tostandardization of HER2 testing in breast cancer patients, includingpre-analytical processing. For both IHC and ISH-based assays, therecommendation is for 10% neutral-buffered formalin (NBF) for atleast 6 h, with a maximum of 48 h [1]. Studies have verified thatstrict implementation of these guidelines is possible (even in largereference laboratories). Such standardization significantly de-creases the number of inconsistent/inconclusive test results forbreast marker IHC assays and FISH testing [2].

We have expanded these studies and investigated the impact ofa range of specific pre-analytical processing factors on the stainingresults from in situ hybridization assays. To control the variablesexamined, we used the human breast carcinoma cell line, MCF7,generated as xenograft tumors in mice. We analyzed the effectsof varying type and time of fixation on H&E staining as well asstaining with a newly developed, fully automated dual color, dualhapten HER2 ISH (Dual ISH) assay, and an existing HER2 FISH as-say. Specifically, we incubated the MCF7 tumors in commonly usedfixatives including 10% neutral-buffered formalin (NBF); zinc (Zn)formalin; alcoholic formalin; Prefer (a synthetic formalin substi-tute); Davidson’s AFA and Bouin’s (fixatives with formalin andacids); for a range of times, varying from 1 to 120 h. In addition,we analyzed the impact of varying the thickness of cut FFPE spec-imens (ranging from 2 to 8 lm) on ISH staining results, using MCF7tumors and human breast carcinoma specimens that span the dy-namic range of HER2 gene status. Finally, we examined the effectsof incubating formalin-fixed breast carcinoma specimens in decal-cifying solutions, which is performed prior to staining tumor sam-ples from bone. (Some cancers may metastasize to bone and thereis a need to test them with various stains and molecular assays.)Our results underscore the importance of implementing a stan-dardized pre-analytical processing system in pathology laborato-ries to ensure consistent and appropriate staining results formolecular diagnostic tests.

2. Materials and methods

2.1. MCF7 cell line growth and preparation for injection into SCID mice

MCF7 cells (P/N HTB-22, ATCC, Manassas VA) were grown insterile filtered Eagle’s MEM with L-glutamine (MediaTech Inc.,Manassas VA) supplemented with 10% FBS, 100 U/ml penicillin,100 lg/ml streptomycin, 0.1 mM non-essential amino acids(NEAA) (all Hyclone, Logan UT), 0.01 mg/ml insulin (Invitrogen,Carlesbad CA), and 1.0 mM sodium pyruvate (MediaTech Inc.,Manassas VA). MCF7 cells were maintained in a 5% CO2 humidified37 �C cell culture incubator in vented T75 or T150 flasks.

Culture flasks were inspected and growth media was changedevery 2–3 days depending on cell confluence at time of last feed-ing. Once reaching 80–100% confluence, MCF7 cells were harvestedfrom flasks for either passaging or injection into SCID mice using a1 � Dulbecco’s PBS rinse (Hyclone, Logan UT) followed by 0.25%Trypsin EDTA (Invitrogen, Carlesbad CA).

2.2. MCF7 xenograft production

Female immunodeficient Fox Chase SCID Outbred Mice be-tween 29 and 35 days old (Strain Code 332, Charles River Labora-tories, Wilmington MA) were received and quarantined forminimum of 7 days. After quarantine, mice were kept in micro-iso-

lation with care and maintenance done following best practices forimmunodeficient models.

Mice between 44 and 71 days old were readied for tumor pro-duction by intraperitoneal injection of a freshly made anestheticconsisting of a 1:1:5 ratio of Ketamine:Xylazine:Sterile water withthe dose of anesthetic mixture adjusted appropriately for mousetotal body weight. Once unconscious, mice were sterilely injectedwith a single 60-day release pellet of 17 beta-estradiol (InnovativeResearch of America, Inc., Sarasota FL) subcutaneously in neck areabetween shoulder blades using a trocar just large enough to con-tain the estradiol pellet. While unconscious, 1 � 107 MCF7 cellssuspended to a concentration of approximately 7 � 107 cells/mlin 1 � DPBS were sterilely injected subcutaneously in the right rearupper hip area using a 25 gauge needle.

After becoming fully conscious following injections, mice werereturned to a clean micro-isolation cage, monitored for healthyactivity, and cared for according to best practices for immunodefi-cient models. Growth of xenograft tumors were visually monitoreddaily with evident growth gently measured weekly using calipers.In 4–6 weeks, xenograft tumor growth had progressed to around500 mm3 at which time mice were sacrificed via cervical disloca-tion and MCF7 xenograft tumors immediately removed surgically.

2.3. Fixation of MCF7 xenograft tumors

Immediately following surgical removal of MCF7 xenograft tu-mors from SCID mice, tumors were cut into multiple pieces ofapproximately 25–50 mm2 by 5 mm thick, placed into appropri-ately labeled disposable plastic tissue cassettes, and then immedi-ately placed into approximately 30 ml per cassette of one of thefollowing fixatives:

Fixative
Formulation Source
Prefer
Glyoxal in water andethanol
Anatech

Davidson’s AFA
33% ethanol, 22% formalin,and 11.5% acetic acid
ElectronMicroscopySciences

Bouin’s fluid
Picric acid, 37–40%formalin, and acetic acid
Richard-AllenScientific

10% Neutral-bufferedformalin(NBF)

Formaldehyde, 37–40%:100 ml/Ldistilled or deionized water900 ml/Lsodium phosphate,monobasic: 4.0 g/Lsodium phosphate, di-basic(anhydrous): 6.5 g/L

Sigma-Aldrich

Zinc (Zn)formalin

Water: 94–95%formaldehyde: 3.5–4%methyl alcohol: 1.2%zinc sulfate: 1–2%sodium acetate: <1%

Richard-AllenScientific

10% Alcoholicformalin

Formaldehyde 37%, insodium phosphate mono-and di-basic, water, andethyl alcohol

BBCBiochemical

Often, pieces of each tumor were placed in different fixatives to ac-count for any potential intertumor or intermouse differences acrossmultiple fixatives. Xenograft tumors were fixed for: 1, 3, 6, 12, 24,and 48 h, and some specimens were fixed for 72 and 120 h. Follow-ing completion of scheduled fixation time, tumor cassettes were

A. Babic et al. / Methods 52 (2010) 287–300 289

removed from fixative and immediately rinsed in 200 ml of 70%EtOH per a maximum of five cassettes for between 1 h to overnight.Following this rinse, tumor cassettes were moved to a fresh 200 mlvolume of 70% EtOH for up to 1 week until automated embedding inparaffin wax.

Automated tissue embedding was done using a Renaissance Tis-sue Processor (Ventana Medical Systems) using a graded ethanolseries (Reagent grade EtOH, VWR, West Chester PA) followed byxylene washes (Histology Grade Xylene, VWR) and lastly heatedparaffin washes (McCormic Scientific LLC).

The automated tissue embedding procedure program consistedof the following parameters:

Station
Time Solution Temperature(�C)
Vacuum(psi)

Pressure(psi)

3
1:00 70%EtOH
37
15 7
4
1:00 80%EtOH
37
15 7
5
1:00 95%EtOH
37
15 7
6
1:00 95%EtOH
37
15 7
7
1:00 100%EtOH
37
15 7
8
1:00 100%EtOH
37
15 7
9
1:00 Xylene 45 15 7 10 1:00 Xylene 45 15 7 11 1:00 Paraffin 60 15 7 12 1:00 Paraffin 60 15 7 13 1:00 Paraffin 60 15 7
Stirrer speed
10 Level lower Lockout Off Delay station 3 Delay heat and stir Off
2.4. Decalcification

Post-fixation of normal breast tissue in 10% NBF, the tissuespecimen was sectioned further into small and large pieces by apathologist. MCF7 xenografts were obtained as previously men-tioned in Section 2.3. Each piece of tissue was stored in 70% ethanoland was washed three times in distilled water prior to placing thetissue into the decalcifying solutions. The tissues were placed incassettes and submerged into approximately 100 ml of one of threedecalcifying solutions: 0.5 M EDTA (Invitrogen and Gibco), FormicAcid Decalcifying Solution (100 g/L sodium citrate, dehydrate,and 250 ml/L formic acid 88%) (Kraijian, J.T. Baker), or Thermo Sci-entific Richard Allan Decalcifying Solution (HCl 3–4%,<0.1% tetraso-dium EDTA and water) (Thermo Scientific) and stored at 2–8 �C.Small breast tissue was placed in decalcifying solutions for 0.5, 1,2, or 3 h and large breast tissue was placed in the decalcifying solu-tions for 4, 8, 16, 24, 48, or 72 h. Once the time point was reached,the cassette was removed from the decalcifying solution andrinsed three times in distilled water. Each block was stored in70% ethanol, and processing was performed as described above.Fresh sections of the blocks, cut at 4 lm thickness, were stainedwith the HER2 Dual ISH assay.

2.5. Testing the impact of tissue thickness on the slide

Three different human invasive breast carcinoma specimensand three different xenograft tumor specimens (Calu-3, ZR751and MCF7) that span the dynamic range of HER2 gene status were

used in this part of the study. The FFPE specimens were sectionedat different thicknesses (2, 4, 6, and 8 lm) using a calibrated micro-tome and stained in duplicate with the HER2 Dual ISH assay. Onereader quantified the HER2 and chromosome 17 copy numbers ineach specimen and the data were analyzed for mean copy numbersand the variability in the counts (%CV).

2.6. Brightfield in situ hybridization (HER2 Dual Hapten, Dual ColorISH: Dual ISH)

Paraffin tissue sections (4 lm thick) were mounted on Super-frost Plus slides (VWR, Catalog No. 48311–703) and stained withthe newly developed, fully automated HER2 Dual ISH assay on aBenchMark� XT slide stainer with the recommended procedure(Ventana Medical Systems, Inc., Tucson, Arizona). After the auto-mated staining procedure, the stained slides were rinsed withtap water containing DAWN� (Proctor & Gamble Company, Cincin-nati, Ohio) and then rinsed with distilled water. Slides were driedat 45–65 �C for at least 15 min and cover-slipped using the Tis-sue-Tek Film automated coverslipper (Sakura). HER2 Dual ISHand H&E images were taken using an Olympus BX51 microscopeand Olympus DP72 camera.

2.7. FISH staining

HER2 manual FISH staining was performed using Abbott/VysisPATHVYSION HER2 DNA Probe Kit (Catalog No. 30–161060).Briefly, slides were baked overnight at 56 �C, deparaffinizedthrough three 10-min incubation in CitriSolv Clearing agent (FisherScientific, Pittsburgh, PA, Catalog No 22–143-975), dehydratedwith two 5-min immersions in 100% EtOH, and then air dried.Specimens were treated with pre-treatment solution at 80 �C for30 min followed by protease treatment for 15 min at 37 �C. Slideswere next dehydrated in a graded alcohol series (70%, 85%, and100%) and air dried.

Ten microliter of probe mixture was added and then the hybrid-ization area was cover-slipped (22 � 22 mm glass cover slip) andsealed with rubber cement immediately. Specimens and probeswere co-denatured at 73 �C for 5 min and hybridized at 37 �C over-night in a sealed and humidified chamber (StatSpin, Inc., West-wood, MA). Second day, the rubber cement and the cover slipwere gently removed and the slides were immersed in 2 � SSC/0.3% NP-40 at 73 �C for 2 min, air dried in the dark, and cover-slipped with DAPI counter stain. FISH photographs were takenusing SPOT CCD microscope digital camera (Diagnostic Instru-ments, Inc.) using Zeiss fluorescence microscope equipped withappropriate filters.

3. Results

The first set of studies was designed to assess the impact of dif-ferent fixatives and the time of fixation on staining quality of H&E,FISH, and Dual ISH assays. Since there has been increasing discus-sion around the need for standardization of HER2 testing in breastcarcinoma [1,17], we used the HER2 ISH assays as our test system.MCF7 xenograft tumors were used as a model because the humancarcinoma cells are relatively easy to generate as tumors in SCIDmice, and one tumor can be divided so that different sections canbe analyzed for multiple conditions. Additionally, the MCF7 cellline exhibits non-amplified (single copy) HER2 gene status. There-fore, any impact of pre-analytical processing on the target DNA inthe tissue specimen would easily be observed as a weakening orloss of the single copies of HER2 and chromosome 17 that aredetectable by FISH and Dual ISH.


Commonly used formalin-based fixatives, a synthetic formalinsubstitute, and formalin-based fixatives containing additionalcomponents were used for this study. We generated microarraysof tumors fixed in each fixative for 1, 3, 6 (the minimum recom-mended time per ASCO/CAP), 12, 24, and 48 h (the maximum rec-ommended time per ASCO/CAP). The fixation microarrays werestained with H&E, HER2 FISH, and HER2 Dual ISH. Each core wasevaluated for HER2 and chromosome 17 signal quality, non-spe-cific background staining, and overall nuclear morphology.

Tissue staining results were scored as pass or fail, where passingindicated that both HER2 and chromosome 17 signals were clearlyenumerable based on adequate signal intensity and lack of inter-fering non-specific background staining. A failing score was givenif signal intensity or excessive background staining inhibited enu-meration of the specific HER2 and chromosome 17 staining. Over-all nuclear morphology also was evaluated for each specimen.

3.1. Overall nuclear integrity evaluation

The MCF7 microarray fixation slides were stained with theHER2 Dual ISH assay, which includes hematoxylin counterstainingto enable visualization of the entire tissue specimen using standardbrightfield microscopy. We first evaluated the microarrays underlow power using brightfield microscopy to determine the impacton overall counter stain intensity, as a measure of nuclear integrity.Fig. 1 illustrates the counter stain intensity on MCF7 xenograftsfixed for varying lengths of time in 10% NBF, the recommended fix-ative for ISH-based assays. Tumors that were ‘‘under fixed” (<6 h)were characterized by weak hematoxylin counterstaining thatwas readily noticeable with low power (4�) (Fig. 1). In contrast,specimens fixed for > 6 h in 10% NBF exhibited increasingly intensehematoxylin staining and the nuclei appeared intact and distinct.Weak hematoxylin staining can indicate that nuclear integrity isaffected, while stronger staining correlates with intact nuclei. Thisis consistent with the finding that H&E staining is compromisedwhen the tissue is fixed in 10% NBF for 6 8 h [18], and that ‘‘incon-clusive” testing results were improved when fixation of breast

Fig. 1. MCF7 xenografts fixed in 10% NBF, imaged at 4 �magnification. Weaker hematoxThe staining and nuclear integrity were improved once the fixation time was increased

specimens was a minimum of 6 h [2]. These data are consistentwith ASCO/CAP guidelines for fixing tissue for a minimum of 6 hin NBF.

3.2. Impact of fixation types and times on H&E, FISH, and Dual ISHstaining results

3.2.1. 10% NBFWe stained the fixative arrays with H&E, FISH, and Dual ISH. We

evaluated ISH staining quality (including background staining andnuclear morphology) and whether single copies of HER2 and chro-mosome 17 were easily detectable in the specimens that werefixed at the indicated time points. Ten percent NBF fixative exhib-ited excellent performance overall, with both HER2 FISH and DualISH assays yielding enumerable staining and overall intact tissuemorphology for most time points (Fig. 2b and c). Strikingly, how-ever, specimens fixed for < 6 h in 10% NBF appeared to have weakersingle copy staining for Dual ISH, and the nuclear integrity wascompromised but still intact. Some background staining wasapparent with Dual ISH in the earlier time points, but this didnot affect enumeration of specific signals. The H&E staining exhib-ited a strong contrast for all time points (Fig. 2a). In conclusion,overall staining quality for both FISH and Dual ISH was accept-able/passable for all time points examined, but was optimal in sig-nal intensity and nuclear integrity when the fixation timewas > 6 h.

3.2.2. Alcoholic formalinAlcoholic formalin, another commonly used formalin-based fix-

ative, also showed acceptable performance overall. For FISH, HER2and chromosome 17 staining were distinct and nuclear stainingwas acceptable for later times, but relative signal intensity was im-pacted at early times (Fig. 3b). Overall, Dual ISH performed betterover the range of time points, with a consistent and similar trendas 10% NBF (Fig. 3c). The tissue morphology was acceptable andHER2 and chromosome 17 staining was strong and distinct for alltime points, although fixation times > 6 h were optimal. There

ylin staining is apparent in the specimens that were fixed in 10% NBF for 1 and 3 h.to 6, 12, 24, or 48 h.

Fig. 2. H&E, FISH, and Dual ISH staining at 60 �magnification on 10% NBF fixed tissues. (a) H&E staining was acceptable at all time points. (b) HER2 FISH staining wasacceptable for HER2 and chromosome 17 intensity with negligible background. HER2 was detected as orange and chromosome 17 as green. (c) HER2 Dual ISH also performedwell with distinct HER2 (black) and chromosome 17 staining (red), minimal background, and morphology of the nuclei was acceptable. For both FISH and Dual ISH staining,the nuclei appeared to be over-digested at time points < 6 h, consistent with the weaker hematoxylin staining observed in Fig. 1 above. This correlated with slightly weakerstaining intensity overall for both ISH assays.


Fig. 3. H&E, FISH, and Dual ISH staining at 60 � on alcoholic formalin-fixed tissues. (a) H&E staining was acceptable at all time points. (b) FISH assay staining was acceptablefor both HER2 and chromosome 17. HER2 was detected as orange and chromosome 17 as green. (c) Dual ISH also performed well with distinct HER2 and chromosome 17staining. There was poor contrast in the counter stain, except for the 48 h time point; however, it did not affect the performance of HER2 or chromosome 17 staining. Therewas also good tissue morphology. HER2 was detected as black and chromosome 17 as red.



was minimal background at times < 12 h and the nuclear counter-staining was weaker at time points < 24 h; however, the H&E stain-ing exhibited a strong contrast for all time points (Fig. 3a).

3.2.3. Zinc formalinZn formalin fixative also exhibited good overall performance for

both ISH assays. For HER2 FISH, staining appeared generally weak-er at early time points (<6 h) (Fig. 4b). HER2 Dual ISH assay stain-ing results and morphology were acceptable across all fixationtime points, although there was fainter nuclear counter stain at fix-ation times < 24 h (Fig. 4c). Some background staining was appar-ent at the earlier time points, but it did not impact the ability todistinguish the specific HER2 and chromosome 17 staining. H&Estaining was acceptable for all time points (Fig. 4a).

3.2.4. Davidson’s AFAIn contrast to the fixatives described above, AFA fixative showed

obvious poor performance with the HER2 FISH assay in both qual-ity of staining and tissue morphology for all time points as shownin Fig. 5b. For the HER2 Dual ISH assay, single copy for both probeswas detectable only for short fixation times: 1, 3, 6, and 12 h(Fig. 5c). Strikingly, the signal intensity greatly diminished overtime, becoming undetectable after 12 h in AFA. This may be dueto the degradation of the target DNA over time in the AFA fixative.This trend was in contrast to the results from 10% NBF, alcoholicformalin, and zinc formalin where > 6 h time points were optimal.The H&E staining exhibited a strong contrast for all time points(Fig. 5a).

3.2.5. PreferPrefer fixative showed overall poor performance with HER2

FISH staining for both quality of staining and tissue morphology.At early time points, FISH staining results were virtually undetect-able (Fig. 6b). After 24 and 48 h, FISH staining was detectablealthough relatively weak compared to the results from 10% NBF.Overall tissue morphology improved at 24 and 48 h compared tothe earlier time points. For the Dual ISH assay, the trend also wasthat the early time points had signal intensity that would fail,but > 12 h in the fixative yielded improved staining intensity(Fig. 6c). Although the staining results were detectable at the latertime points, they were weaker compared to results from 10% NBF.Overall, the tissue morphology was acceptable and there was min-imal background for all time points. The H&E staining for 1, 3, 6,and 12 h was normal with good contrast, while the 24 and 48 htime points had weaker contrast (Fig. 6a).

3.2.6. Bouin’sBouin’s fixative exhibited a trend similar to the AFA fixative re-

sults. Bouin’s fixative resulted in very poor performance with theHER2 FISH assay (Fig. 7b). The tissue morphology was poor qualityfrom 6 h on. All times failed to exhibit appropriate single copydetection for both probes. For the HER2 Dual ISH assay, therewas detectable staining for both HER2 and chromosome 17 onlyat the 1 and 3 h time points, although staining was relatively weakcompared with results from 10% NBF (Fig. 7c). Subsequent fixationtimes in Bouin’s failed due to weak to undetectable staining inten-sity. Furthermore, the intensity of the counter stain was weak in alltime points. These data indicate that similar to the trend with AFA,longer fixation times in Bouin’s are detrimental to ISH staining. TheH&E staining for time points up to 24 h was normal with good con-trast, while the 48 h time point had weaker contrast (Fig. 7a).

3.3. Extended fixation time points

The study described above indicates that the formalin-basedfixatives 10% NBF, alcoholic formalin and Zn formalin yielded

optimal ISH staining results across a wide range of time points.However, consistent staining for both FISH and Dual ISH was ob-served when ASCO/CAP guidelines were followed:>6 h of fixationtime maintained nuclear integrity and optimal signal intensity. Incontrast, for Bouin’s and AFA fixatives, this trend was reversedand longer fixation times resulted in a significant decrease in stain-ing intensity or no detectable staining. This likely is due to the fix-atives affecting the target DNA sequences over time. Preferexhibited better performance in staining results for HER2 DualISH compared with FISH, with longer times exhibiting stronger sig-nal intensity overall.

We next investigated whether extending the fixation times be-yond the maximum ASCO/CAP guidelines (48 h) would impact ISHstaining results. We chose 10% NBF, alcoholic formalin, Zn forma-lin, Prefer, and AFA and fixed MCF7 xenograft tumors in these fivefixatives for 72 (3 days) or 120 h (5 days), and stained the speci-mens with the HER2 Dual ISH assay. We did not repeat the stainingwith the FISH assay since overall, FISH and Dual ISH results fol-lowed similar trends in staining results. The results are summa-rized in Table 1.

The HER2 Dual ISH staining results were acceptable in sectionsfixed for 72 and 120 h with 10% NBF. Both HER2 and chromosome17 probes were detectable and nuclear morphology was accept-able, with the nuclei appearing dark blue and intact. Both alcoholicand Zn formalin also exhibited acceptable Dual ISH staining. How-ever, the HER2 Dual ISH signal was slightly weaker than the chro-mosome 17 signal in the alcoholic formalin-fixed sections. This islikely due to differences in target area and detection systems usedto detect each probe. Prefer exhibited poor performance at 72 and120 h, with very weak to no detectable staining for either probe.Not surprisingly, AFA had the poorest performance with no detect-able staining at 72 or 120 h. Thus, 10% NBF, alcoholic formalin andZn formalin fixatives still exhibited appropriate staining even after3 and 5 days of fixation, but prefer and AFA fixation likely affectedthe target DNA in the tissue over time.

3.4. Tissue thickness

The studies outlined above support the idea that control of bothfixative type and time in the fixative are critical for appropriate ISHassay results. The next experiments were performed to assess theimpact of tissue thickness on HER2 Dual ISH staining. Three humanbreast carcinoma specimens, as well as xenograft control slides(containing three different cores: Calu-3 which is amplified;ZR751, which is non-amplified; and MCF7, also non-amplified)were sectioned at 2, 4, 6, and 8 lm and stained with HER2 DualISH in duplicate. Human cases 1 and 2 were non-amplified,although case 2 contained > 2 copies of HER2 in the tumor nuclei.Human case 3 was overtly amplified with clusters of HER2 detect-able. The copy numbers for HER2 and chromosome 17 in each car-cinoma specimen were quantified by one reader (20 nuclei perspecimen) and the raw counts were analyzed for their statisticalvariability across the sections (%CV). The sections also were evalu-ated for overall Dual ISH staining quality. The results are shown be-low in Table 2.

The data show that none of the tissue thickness conditionstested affected the expected HER2 gene status of the specimens(non-amplified or amplified) and the %CVs for the vast majorityof specimens were < 10%. However, it appears that for case 2 (mul-tiple copies of HER2), less mean copies of both HER2 and chromo-some 17 were apparent in the 2 lm section, and more wereapparent in the 8 lm sections, although these differences werenot statistically significant. It was noticed that sections with thick-nesses > 6 lm exhibited vacuoles of paraffin in the nuclei at agreater extent than the thinner sections (‘‘nuclear bubbling”),although it did not interfere with the ability to enumerate the

Fig. 4. H&E, FISH, and Dual ISH staining at 60 � on Zn formalin-fixed tissues. (a) H&E staining was acceptable at all time points. (b) FISH assay staining was acceptable forboth HER2 and chromosome 17; however, at time points < 6 h, staining was generally weaker, but HER2 and chromosome 17 staining were distinct. HER2 was detected asorange and chromosome 17 as green. (c) Dual ISH also performed well with distinct HER2 and chromosome 17 staining. There was good overall tissue morphology for all timepoints and some background staining at the earlier time points, but it did not affect HER2 or chromosome 17 staining.


Fig. 5. H&E, FISH, and Dual ISH staining at 60 � on Davidson’s AFA fixed tissues. (a) H&E staining was acceptable at all time points. (b) FISH assay had unacceptable HER2 andchromosome 17 staining and poor tissue morphology. HER2 was detected as orange and chromosome 17 as green. (c) Dual ISH performed better at earlier time points (up to12 h) with distinct HER2 and chromosome 17 staining but weaker Dual ISH staining at 24 h and no staining at 48 h. HER2 was detected as black and chromosome 17 as red.


Fig. 6. H&E, FISH, and Dual ISH staining at 60 � on Prefer fixed tissues. (a) Although H&E staining was less intense compared with other fixatives and decreased in contrast atthe 24 and 48 h time points, staining was still acceptable. (b) FISH assay staining was not acceptable for either HER2 and chromosome 17. HER2 was detected as orange andchromosome 17 as green. (c) ISH staining was evident in all time points. SISH staining performed reasonably well with distinct HER2 staining at time points > 6 h. The counterstain was weak for all time points, except at 48 h. The overall tissue morphology was acceptable for all time points. HER2 was detected as black and chromosome 17 as red.


Fig. 7. H&E, FISH, and Dual ISH staining at 60 � on Bouin’s fixed tissues. (a) H&E staining was acceptable for all time points, except for the 48 h time point, which was weakerin contrast. (b) FISH assay staining had poor quality of staining and tissue morphology for all time points. HER2 was detected as orange and chromosome 17 as green. (c) DualISH staining was acceptable for both HER2 and chromosome 17 staining at the 1 and 3 h time points. For time points > 3 h, there was no Dual ISH staining. The intensity of thecounter stain was weak for both FISH and Dual ISH at most time points.


Table 1Fixatives used to process MCF7 tumors for 72 h or 120 h. P, passing(acceptable) single copy detection for both HER2 and chromosome 17; F,failing (weak or not detectable staining for HER2 and chromosome 17).

Fixative 72 h (3 days) 120 h (5 days)

10% NBF P PAlcoholic formalin P PZn formalin P PPrefer F FAFA F F

Table 2Results from staining breast carcinoma specimens sectioned at different thicknesseswith the HER2 Dual ISH assay. Three human breast cases and three xenograft celllines were sectioned at 2, 4, 6, or 8 lm and staining results were quantified by onereader. The raw counts for HER2 and chromosome 17 were analyzed for variance byevaluating %CVs. The coefficient of variation (CV) is used to measure the consistencyof gene copy numbers across tissues and thicknesses. The larger the CV value, the lessconsistent the gene copy number within that tissue specimen. A lower CV valueindicates the precision of the copy numbers. The %CV is defined as the standarddeviation of counts for a given tissue specimen divided by the mean multiplied by100. The mean HER2 and chromosome 17 copy numbers from 20 nuclei per specimen,as well as the mean ratio, also are provided.

HER2 CH17 Ratio

Tissue Mean %CV Mean %CV Mean %CV

2 lmCalu-3 17.3 7.9 2.6 2.0 6.8 5.9MCF7 1.4 1.8 1.9 9.4 0.8 7.6ZR-75–1 2.5 1.0 1.9 0.0 1.4 1.0Human case 1 1.9 5.4 1.7 3.0 1.1 2.5Human case 2 3.4 1.5 2.1 3.6 1.6 2.1Human case 3 17.1 1.8 1.9 2.6 9.0 0.9




Table 3Effects of post-fixation treatment of xenografts and normal breast tissuespecimens (small or large) in three different decalcification solutions:EDTA, formic acid, or HCl. The xenografts and different sized breastspecimens were formalin-fixed and subsequently incubated for varyingtimes in decalcification solutions. They were embedded in paraffin andimmediately stained with the HER2 Dual ISH assay. P, passing results forsingle copy detection of HER2 and chromosome 17; F, weak or non-detectable single copy detection.

Time (h) EDTA Formic acid HCl

MCF7 xenograft specimen4 P P F8 P P F

24 P P F48 P P F

Small breast tissue specimen0.5 P P P1 P P P2 P P F3 P P F

Large breast tissue specimen4 P P F8 P P F

16 P P F24 P P F48 P P F72 P P F


signals in those specimens. Four micron sections were thereforeoptimal for the overall ISH staining results.

3.5. Decalcification

Finally, we also took the opportunity to examine the effects ofspecific post-fixation processing that occurs for certain specimensin clinical practice. For histological analysis, bone-containing spec-imens are incubated in decalcification solutions to remove calcifiedbone remnants [19]. It can be difficult to cut sections of biopsies ofmetastases to the bone, thus, decalcification solutions are designedto soften the tissue to allow cutting [19]. Since some metastaticbreast specimens may require post-fixation decalcification, weinvestigated the effects of commonly used decalcification solu-tions, including 0.5 M EDTA (a calcium chelator), a solution con-

taining formic acid, and a solution containing hydrochloric acid(HCl), on Dual ISH staining (Table 3). Not only were MCF7 xeno-grafts used for this study, but also small (�2 mm2) and larger(�10 mm2) normal human breast tissue specimens were chosento mimic a range of tumor size in clinical settings. The tissues werefixed in 10% NBF, and then incubated in decalcifying solutions fordifferent time points prior to embedding and staining with HER2Dual ISH. (Small breast tissue specimens were incubated in decal-cifying solution for less time compared to large breast tissue.)

The results demonstrated that HER2 Dual ISH staining for theMCF7 xenografts was not affected by 0.5 M EDTA or formic acidat any of the time points. More importantly, HER2 Dual ISH stain-ing for neither small nor large breast tissue was obviously im-pacted by incubation in 0.5 M EDTA post-fixation, even at longertime points (up to 3 h for small breast tissue and 72 h for the largerspecimen). This also was true for incubation of the breast speci-mens in the formic acid solution, as they both exhibited distinctand enumerable signals for both HER2 and chromosome 17. In con-trast, however, there was weak or no detectable HER2 Dual ISHstaining for both tissue specimens and MCF7 xenografts whendecalcified with the HCl solution over time. The MCF7 xenograftshad no HER2 or chromosome 17 signal even at the earliest timepoint of 4 h. For the small breast specimen, staining was acceptableat very early incubation periods in HCl (0.5 and 1 h only), but thesignal decreased in intensity significantly by 2 h. For the largerbreast specimen, incubation from 4 to 72 h resulted in no detect-able staining for either probe. These data indicate that HCl incuba-tion likely destroys the target DNA at a relatively quick rate, evenpost-fixation. In summary, when decalcification methods are re-quired to analyze tissue biopsies metastasized to the bone, theyshould be verified by the laboratory first to determine whetherthey are compatible with ISH testing. Some acids (formic acid) thatare generally weaker in their pKa values compared with other acids(HCl) are compatible with ISH, but the relative acid concentrationin the solution may also be critical.

4. Discussion

The impact of multiple pre-analytic factors, including type offixative, time from excision to fixation, the volume of tissue to


the volume of fixative, duration of fixation, post-processing factors,and the thickness of the tissue specimen cut onto the glass micro-scope slide, are critical for optimal staining and analysis by H&E,IHC and ISH. There have been studies that address various aspectsof the pre-analytical process on various diagnostic tests, but com-plete data sets on the impact of fixative types and times for molec-ular assays are more limited. Here, we evaluated the impact of sixdifferent fixatives across a range of fixation time points on ISH as-say performance and tissue morphology. Commonly used fixativesinclude 10% NBF, various zinc formalins, glyoxal-dialdehyde fixa-tives (such as Prefer), and various alcohols (reagent alcohol, isopro-pyl alcohols, and ethanol), as well as proprietary fixatives. Weselected six different fixatives and used MCF7 tumors as a modelsystem to generate enough samples to test their impact across awide range of fixation times. Fixation times had obvious effectson H&E staining, HER2 FISH and Dual ISH stained results. The im-pact of the different fixatives on optimal performance of HER2 FISHand Dual ISH assays are summarized in Table 4.

Our data indicate that the best fixatives for preserving bothHER2 FISH and Dual ISH signal intensity across a wide range of fix-ation times were 10% NBF, zinc formalin and alcoholic formalin.However, time also played a major role in the quality of the overallstaining results, with under-fixation (<6 h) in these fixatives exhib-iting an apparent loss of nuclear material that could lead to weak-ened signal. This is consistent with a recent study that showed thatfixing breast core biopsies for a minimum of 6 h in NBF, using adedicated tissue processor, significantly improved the staining suc-cess rates at a large reference laboratory for both IHC and ISH [2].Our results with extended fixation times yielded acceptable stain-ing in 10% NBF, alcoholic formalin and zinc formalin even when thespecimens were fixed up to 5 days. Taken together, our data indi-cate that appropriate single copy HER2 gene detection is ensuredwhen specimens are fixed in formalin-based fixatives (not contain-ing acids) for at least 6 h, which is in agreement with the ASCO/CAPguidelines. The results also indicate that ‘‘under-fixation” is detri-mental to ISH staining results, while ‘‘over-fixation” is not as prob-lematic in these specific fixatives.

The impacts of fixative and time of fixation were most apparentwhen the fixative contained acidic components (AFA and Bouin’s),where staining was adequate only at the early fixation times (gen-erally,<12 h). Consistent with these findings, studies utilizingshrimp tissue found that fixation in Davidson’s AFA for P 6 daysresulted in false negative results using cDNA ISH. Fixation for thesame length of time in a neutral fixative resulted in appropriatestaining, indicating that the loss of signal in the over fixed David-son’s tissue is due to acid hydrolysis of nucleic acids [16]. Thus,it is not recommended that fixatives containing acidic componentsbe used for ISH-based assays. This was also observed when weinvestigated the effects of decalcifying solutions post-NBF fixation.Decalcifying solutions containing strong acidic components (HCl)greatly diminished the signal intensity for HER2 and chromosome

Table 4Summary of appropriate ISH staining results with different fixatives.

Fixative HER2 Dual ISHappropriate results

HER2 FISHappropriate results

10% NBF >6 to 48 h (and extendedtimes up to 120 h)

>6 to 48 h

Alcoholic formalin >6 to 48 h (and extendedtimes up to 120 h)

>6 to 48 h

Zinc formalin >6 to 48 h (and extendedtimes up to 120 h)

>6 to 48 h

AFA <12 h NonePrefer >12 to 48 h but not

extended beyond 48 h24 and 48 h

Bouin’s 3 h or less None

17. Weaker acids, such as formic acid, along with EDTA, did not sig-nificantly impact Dual ISH results. Thus, time and fixative are crit-ical for appropriate ISH staining results, and treatments containingacidic components at any concentration should be tested thor-oughly by the laboratory before being used for molecular assays.

The effects of the different fixation times are likely due not onlyto the rate at which the fixative reacts with the tissue, but also thetime of diffusion across the tissue [7]. It is intuitive that ‘‘under-fix-ation” in 10% NBF could result in suboptimal tissue morphologyand molecular characteristics, which was apparent from the resultsof our studies. ‘‘Over-fixation” in 10% NBF did not impact HER2Dual ISH staining results on MCF7 tumors. However, we have ob-served that over-fixation of many human tissue specimens can re-sult in weaker ISH signals. ISH signal strength can often beimproved in these instances by harsher pre-treatment conditions(including increased protease and cell conditioning times). Theharsher pre-treatment presumably breaks formalin cross-linksand enables probe and antibody penetration (unpublishedobservations).

Finally, we determined that tissue thickness is also importantfor proper ISH staining results. Although the range of thicknessestested in this study did not impact the HER2/chromosome 17 ratiosquantified for human breast carcinoma specimens, thicker sectionstended to exhibit residual paraffin (‘‘nuclear bubbling”) within thesection. Previously, it was determined that incomplete removal ofparaffin can inhibit probe and antibody penetration and contributeto staining artifacts [7]. In fact, we have observed that thickerbreast tissue sections often require harsher protease treatmentand/or cell conditioning to obtain adequate staining (unpublishedobservations). It should also be noted that we have observed that‘‘nuclear bubbling” may also occur due to under-fixation (<6 h)which produces a less discrete, ‘‘fuzzier” bubble. Thus, 4 lm sec-tions, for sections fixed a minimum of 6 h in formalin, should bethe standard for ISH assays.

There are other factors that we did not analyze here that arealso known to contribute to the overall quality and reproducibilityof staining results from IHC and ISH. For example, the interval be-tween the acquisition of the tissue and its fixation should be asshort as possible to obtain consistent staining results [7]. Thesestudies, combined with the results presented here, not only con-firm that the guidelines set by ASCO/CAP provide conditions thatare on par for optimal IHC and ISH staining results, but will serveas an instructional and troubleshooting tool for tissue processinglaboratories to illustrate the effects of varying the ISH fixationguidelines.

References

[1] A.C. Wolff, M.E. Hammond, J.N. Schwartz, K.L. Hagerty, D.C. Allred, R.J. Cote, M.Dowsett, P.L. Fitzgibbons, W.M. Hanna, A. Langer, L.M. McShane, S. Paik, M.D.Pegram, E.A. Perez, M.F. Press, A. Rhodes, C. Sturgeon, S.E. Taube, R. Tubbs, G.H.Vance, M. van de Vijver, T.M. Wheeler, D.F. Hayes, Arch. Pathol. Lab. Med. 131(2007) 18–43.

[2] L.P. Middleton, K.M. Price, P. Puig, L.J. Heydon, E. Tarco, N. Sneige, K. Barr, M.T.Deavers, Arch. Pathol. Lab. Med. 133 (2009) 775–780.

[3] F.L. Carson, C. Hladik, Histotechnology: A Self-Instructional Text, third ed.,American Society for Clinical Pathology Press, Illinois, 2009.

[4] L. Tyrrell, J. Elias, J. Longley, Am. J. Dermatopathol. 17 (1995) 476–483.[5] L.M. Weiss, Y.Y. Chen, J. Histochem. Cytochem. 39 (1991) 1237–1242.[6] G.J. Nuovo, S.J. Silverstein, Lab. Invest. 59 (1988) 720–724.[7] T. Khoury, S. Sait, H. Hwang, R. Chandrasekhar, G. Wilding, D. Tan, S. Kulkarni,

Mod. Pathol. 22 (2009) 1457–1467.[8] N. Graem, K. Helweg-Larsen, Acta. Pathol. Microbiol. Scand. A 87A (1979) 375–

378.[9] R.D. Start, M.S. Flynn, S.S. Cross, K. Rogers, J.H. Smith, Virchows. Arch. A Pathol.

Anat. Histopahtol. 419 (1991) 475–477.[10] R.D. Start, S.S. Cross, J.H. Smith, J. Clin. Pathol. 45 (1992) 546–547.[11] K.W. Hasson, J. Hasson, H. Aubert, R.M. Redman, D.V. Lightner, J. Virol.

Methods 66 (1997) 227–236.[12] Y. Ha, C. Chae, J. Vet. Diagn. Invest. 21 (2009) 649–654.


[13] T.J. O’leary, C.B. Fowler, D.L. Evers, J.T. Mason, Biotech. Histochem. 84 (2009)217–221.

[14] T. Oyama, Y. Ishikawa, M. Hayashi, K. Arihiro, J. Horiguchi, Breast Cancer 14(2007) 182–188.

[15] S.A. Bogen, K. Vani, S.R. Sompuram, Biotech. Histochem. 84 (2009) 207–215.

[16] T.D. Hewitson, B. Wigg, G.J. Becker, Methods Mol. Biol. 611 (2010) 3–18.[17] F. Schmitt, Adv. Ther. 26 (2009) S1–S8.[18] R.W. Dapson, Biotech. Histochem. 82 (2007) 133–140.[19] J.C. Alers, P.J. Krijtenburg, K.J. Vissers, H. van Dekken, J. Histochem. Cytochem.

47 (1999) 703–710.

the impact of pre-analytical processing on staining quality for h&e, dual hapten, dual color in...

Documents