OUP user menu

Is It Necessary to Evaluate Nuclei in HER2 FISH Evaluation?

Carlos López PhD, Barbara Tomás, Anna Korzynska PhD, Ramón Bosch MD, Maria T. Salvadó PhD, Montserrat Llobera MD, Marcial Garcia-Rojo MDPhD, Tomás Alvaro MDPhD, Joaquín Jaén MD, Marylène Lejeune PhD
DOI: http://dx.doi.org/10.1309/AJCPPXLYJVFGOV8I 47-54 First published online: 1 January 2013


A new method that simplifies the evaluation of the traditional HER2 fluorescence in situ hybridization (FISH) evaluation in breast cancer was proposed. HER2 status was evaluated in digital images (DIs) captured from 423 invasive breast cancer stained sections. All centromeric/CEP17 and HER2 gene signals obtained from separated stacked DIs were manually counted on the screen. The global ratios were compared with the traditional FISH evaluation and the immunohistochemical status. The 2 FISH scores were convergent in 96.93% of cases, showing an “almost perfect” agreement with a weighted k of 0.956 (95% confidence interval, 0.928-0.985). The new method evaluates at least 3 times more nuclei than traditional methods and also has an almost perfect agreement with the immunohistochemical scores. The proposed enhanced method substantially improves HER2 FISH assessment in breast cancer biopsy specimens because the evaluation of HER2/CEP17 copy numbers is more representative, easier, and faster than the conventional method.

Key Words
  • Breast neoplasms
  • In situ hybridization
  • Reproducibility
  • Sensitivity
  • Specificity

Amplification of the human epidermal growth factor receptor-2 (HER2) has been observed in 20% to 30% of invasive breast cancer13 and is associated with poor prog-nosis4,5 and poor response to treatments.6,7 In 1998, the US Food and Drug Administration (FDA) approved trastuzumab (Herceptin, Genentech, South San Francisco, CA), a recombinant humanized monoclonal antibody directed against the HER2 receptor, for patients with breast cancer with HER2 gene amplification. The clinical benefits of trastuzumab used alone or in combination with other treatments have been demonstrated in these patients.811 Nevertheless, accurate administration is necessary because of its cardiotoxicity12,13 and greater expense14,15 and because only patients with a clear evidence of HER2 gene amplification or HER2 protein overexpression are considered suitable candidates for receiving trastuzumab.16

Currently, HER2 status is examined using immunohistochemistry (IHC) and, in equivocal cases, with fluorescence in situ hybridization (FISH), which is considered the “gold stan-dard.”17,18 The FDA approved the DAKO HER2 FISH pharmDx kit (DAKO, Carpinteria, CA), which uses a dual-probe method and stains the centromeric region of chromosome 17 (CEP17) and copies of the HER2 gene. The most recent American Society of Clinical Oncology (ASCO)/College of American Pathologists (CAP) recommendations define the HER2 gene amplification classes by their HER2/CEP17 ratio—with a ratio of more than 2.2 classified as positive; less than 1.8 as negative; and 1.8 to 2.2 as equivocal.19 Adherence to the ASCO/CAP guidelines for HER2 testing has resulted in a stronger correlation between HER2 IHC and HER2 FISH results but also in a reduction in the number of inconclusive IHC cases.20,21

In current clinical practice, IHC remains an attractive technique because of its lower cost, great availability, and good concordance with FISH for almost all scores. On the other hand, FISH evaluation is a complicated, arduous, and highly time-consuming procedure.22 For adequate and accurate manual signal enumeration, the observer must discount necrotic areas and overdigested nuclei. Once an appropriate area is located, not all nuclei can be evaluated. Only well-isolated nuclei can be evaluated. Overlapping nuclei with ambiguous borders must be discarded. This is not always easy because of the high cell density of the invasive tumoral area of samples. Moreover, the observer must search a relatively large number of representative areas to evaluate enough individual nuclei. After that, in each tumoral area selected, the observer must move the fluorescent filter wheel several times to examine and quantify the various probe signals present in each evaluated nucleus. For each filter the observer must also refocus the microscope to examine all the signals in the different planes of each nucleus. After counting all the centromeric and gene signals in at least 20 nuclei,19 the global ratio with HER2/CEP17 is calculated in each sample.

To solve some of these problems, several automated and semiautomated procedures have been developed to evaluate FISH HER2 status.2326 Most of these published procedures followed 2 different steps: one for detecting spots and the other for detecting nuclei.24,26,27 However, the problem of the presence of overlapping nuclei still exists in these procedures, and the correct separation of touching nuclei is not always optimal. This part of the analysis increases the complexity of the automated methods and needs to be improved despite the promising results.23,27 So although the samples may stain well, the major problem is to isolate a sufficient number of nuclei for both manual and automated HER2 FISH evaluation. Therefore, to obtain a sufficient number of nuclei, the total number of areas to evaluate must be increased, thereby increasing the period of sample evaluation as well.

The aim of our study was to modify the FISH HER2 status evaluation without considering the nuclei, thus leading us to evaluate a high number of breast cancer areas faster, more easily, and more efficiently than the traditional method.

Materials and Methods

Patient Selection Criteria and Tissue Microarray Construction

The workflow of the study profile is outlined in Figure 1. After obtaining approval from the institutional research and clinical ethics committees, 423 invasive breast cancer samples evaluated between 1990 and 2000 were collected from the archives of the Hospital de Tortosa Verge de la Cinta (Tortosa, Spain). Different tissue microarrays (TMAs) were constructed to evaluate the large number of samples of the study. This method has been used previously in HER2 evaluation.28,29 Our laboratory used a TMA construction technique without prefabricating recipient blocks, as described by Chen and Zhou.30

Figure 1

Overview of the study profile for immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH) staining. DI, digital image; TMA, tissue microarray.

Sample Stain

TMA paraffin blocks were used to cut sections containing all the tissue cores embedded in the recipient block. All TMA sections were prepared for IHC and standardized as previously described.31 The sections of TMAs were stained using the DAKO HER2 FISH pharmDx kit in accordance with the manufacturer’s instructions. The specific hybridization produces a distinct red fluorescent signal at each HER2 gene locus and a distinct green fluorescent signal at each chromosome 17 centromere. Finally, the slides were counterstained and mounted with fluorescence mounting medium containing 4',6-diamidino-2-phenylindole (DAPI) and coverslipped.

FISH Digital Image Acquisition and Stacking

FISH TMA slides were observed using an Axioskop 2 plus fluorescence microscope (Carl Zeiss Microscopy, Jena, Germany) equipped with a light filter wheel for DAPI (XF06 Vivid Standard set), fluorescein isothiocyanate (FITC; XF202 M-Fish set), and Texas red (XF207 M-Fish set) (Omega Optical, Brattleboro, VT). Tissue sections were viewed with oil immersion at ×100 with an objective NPL Fluotar (numerical aperture = 1.32). Three representative areas of each case were captured manually with the Leica DFC300FX camera (Leica Microsystems Digital Imaging, Cambridge, England) coupled to the microscope and controlled with Leica Application Suite v.2.8.1 computer software (Leica Microsystems Digital Imaging). Captured images were saved in uncompressed tagged image file format at a resolution of 1,392 × 1,040 pixels in RGB 24 true-color format.32 The same ranges of illumination values were used for each filter to allow good visual evaluation of probe signals.

After selecting a representative area, a single image of the cell nuclei was captured with the DAPI filter. Then, to visualize all the fluorescent signals above and below the focal plane (unfocused spots), a “manual focusing” was accomplished by z-axis scanning. For this purpose, the lens was shifted by 4 steps in the z-axis, and 4 serial digital images (DIs) in thick fluorescent-stained sections were captured with the FITC filter and another 4 were captured with the Texas red filter, each separated by 0.5 μm of focus perfectly aligned in the z-axis (z-stacks). This “manual focusing” allows the capture of all the FISH signals distributed throughout the tissue sections.

The z-stacking was previously used in several studies to avoid the need to focus up and down through the section during the fluorescent signal evaluations and also to improve the focus of the 1-plane signals.3335 In the current study, to obtain definitive composite best-focus images, the “extended depth of field” (EDF) tool of Image-Pro Plus 5.0 software (MediaCybernetics, Rockville, MD) was used to stack the 4 FITC DIs into a single image and the 4 Texas red DIs into another. The focus settings used in the EDF were of “maximum intensity” so that the pixel from the plane having the highest intensity at the current pixel location was selected to be part of the stacked image. This “extended focusing” procedure yielded 3 single images: 1 of nuclei (DAPI), 1 with all stacked centromeric signals (FITC), and 1 with all stacked HER2 gene signals (Texas red).

Conventional Evaluation of IHC and FISH HER2 Status

Two independent observers of the Hospital de Tor-tosa Verge de la Cinta conducted the IHC evaluation in accordance with ASCO/CAP criteria.19 Conventional FISH results were evaluated from the acquired DIs. For this purpose, DAPI and the stacked FITC and Texas red images were merged into a single frame using the “merge images” tool of the Image-Pro Plus software. This allows the 3 images of the same area to be viewed differently through the same frame as with the fluorescent microscope. This technique also benefits from the absence of a fading effect and from the ability to observe the dots in the same area always at the same intensity. In each DAPI DI, a trained observer sought nuclei that were well defined, isolated, and not overdigested. Using the program’s “area-of-interest” tools, the observer delineated each representative nucleus with a circle. The circles formed around the objects remained visible during the display of images on the same frame. Using the “sequence toolbar,” the user switched to the FITC DI frame to count the centromeric signals and then to the Texas red DI to count the HER2 gene signals in the circular area of each nucleus. Following DAKO’s guidelines, the signals were counted in a minimum of 20 nuclei for each sample. The ratio of interest was established as the number of HER2 gene signals divided by the number of centromeric CEN17 signals, in accordance with ASCO/CAP guidelines.19

New Method for FISH HER2 Status Evaluation

The new method is based on 3 single steps: image capture, image stacking, and dot counting, excluding the selection of isolated nuclei. The first 2 steps were described earlier. The areas evaluated for each case are the same as those used in the traditional evaluation. The new method uses DAPI DIs to detect only representative tumoral areas and counts all dots regardless of whether they belong to a nuclei or not. For each case, all the green signals in the FITC-stacked DIs and the red signals in the Texas red–stacked DIs were counted manually with the help of the “manual tag” tool in Image-Pro Plus. This allowed each counted signal to be marked by the observer, thereby avoiding accidental repeat counting of the same signal and ensuring that all were counted. The tallies of FITC and Texas red signals were increased automatically in the program’s object counter. The ratio was calculated for each area, dividing the number of HER2 gene signals by the number of CEN17 centromeric signals. For each case, the definitive ratio was calculated as the average of the 3 ratios of the representative areas. As with conventional FISH evaluation, HER2 amplification was scored according to ASCO/CAP guidelines.19

Statistical Analysis

Agreement between measures of HER2 status was analyzed using SPSS 18.0 software (SPSS, Chicago, IL). The agreement between the 2 IHC scores, between the 2 modalities of the FISH scoring method, and between both FISH modalities and the IHC scores were estimated using the λ statistic,36,37 after the appropriateness of this method was established with the McNemar test. Results are presented as the λ statistic ± 95% confidence intervals (CIs). The quadratic weighted k was also used to assess agreement between more than 2 categories of the interobserver IHC and inter-FISH methods for HER2 status evaluation.38

Receiver operating characteristic curve analysis was used to compare the accuracy of the 2 IHC scores and the new FISH scoring method, using the conventional FISH scoring evaluation as the standard. The sensitivity and specificity of the new method were also evaluated. Area-under-the-curve values were compared with the method of Delong et al39 using the statistical software R 2.10.1, with P values of less than .05 indicating significance.


Interobserver IHC HER2 Scoring Agreement

The comparison of the IHC scores from the 423 cases evaluated by the 2 observers showed “almost perfect” interob-server agreement (weighted λ = 0.906; 95% CI = 0.872-0.940) Table 1. Among the scores of the 2 observers, 91.5% were convergent: 76.6% of negative cases (0/1+), 11.3% of positive cases (3+), and 3.5% of equivocal cases (2+). Discrepancies between the 2 observers’ scores (8.5% of cases) included negative vs equivocal scores in 32/36 cases (88.9%) and positive vs equivocal scores in 4/36 cases (11.1%). Eliminating the equivocal cases (2+) from the scores of both observers (n = 372), the interobserver variability showed perfect agreement (λ = 1.000; 95% CI = 1.000-1.000).

Correlation of Both FISH HER2 Status Evaluations With IHC Scores of 2 Observers

Table 1 shows the scores obtained with the new and the conventional methods for evaluating FISH HER2 status. Taking into consideration only positive and negative scores of IHC and conventional FISH evaluations, the λ values were within the almost perfect agreement range for both observers: λ = 0.939 (95% CI = 0.886-0.992) for observer 1, and λ = 0.906 (95% CI = 0.841-0.971) for observer 2. Comparing the new FISH scoring method with IHC, λ values were 0.949 (95% CI = 0.900-0.998) for observer 1 and 0.901 (95% CI = 0.832-0.970) for observer 2. Table 2 shows the correlation of the IHC scores of both observers with the results of both FISH evaluation methods.

Agreement Between the Conventional and New FISH HER2 Evaluation Methods

The comparison of the FISH scores from the 423 cases evaluated with the conventional and new methods showed almost perfect agreement Table 3, with a weighted λ value of 0.956 (95% CI = 0.928-0.985). When the equivocal cases were excluded, the agreement remained almost perfect but with a higher k value (k = 0.988; 95% CI = 0.964-1.000). As illustrated in Table 3, the scores obtained from the 2 types of FISH evaluation are convergent in 96.93% of the cases (410/423). Table 4 shows the results of the 4 scoring methods in the 13 discordant cases. Only 1 case (0.24%) in the study showed complete disagreement because the conventional method detected amplification whereas the new method did not. The other 12 cases (2.8%) showed partial disagreement, with values found between 1.8 and 2.2, thus making the results very borderline in almost all cases.

View this table:
Table 1
View this table:
Table 2

The accuracy of the new FISH scoring method compared with the conventional FISH score was 97.50% when the equivocal cases were included and 99.00% when excluded Figure 2. Under the latter circumstances, the new method had a sensitivity of 97.96% (95% CI = 0.878-0.999) and a 100% specificity (95% CI = 0.987-1.000).


The work presented herein is a proposal to modify the manual procedure (not automated) of HER2 FISH assessment using DIs. The FDA and the ASCO/CAP criteria for FISH HER2 evaluation require the centromeric and HER2 signal numbers to be simultaneously evaluated in a minimum of 20 well-separated nuclei. Our study shows that it is possible to quantify all the centromeric and HER2 gene signals in a single tumoral field (DI) independent of individual and overlapping nuclei and that the results are similar to those obtained with the conventional method.

Excluding the equivocal IHC cases, our results demonstrate complete agreement between HER2 expression (IHC) and HER2 amplification status as determined using the conventional and new FISH methods. In general, documented concordance rates have fallen well below the 95% threshold demanded by the new ASCO/CAP guidelines, with many studies demonstrating concordance rates (excluding 2+ cases) closer to 80% to 90%.4041 These variations arise principally from the broad differences in methods, instrumentation, and experience of the testing laboratories. Image analysis is known to reduce interobserver variability among pathologists evaluating HER2 IHC and to produce better concordance with HER2 FISH.42,43 The good results obtained with the new method demonstrate that the number of background dots with similar shape and size is lower than the number of probe dots. The method presented here is based on the assumption that if there are some background dots with similar shape and size as the dots of the probes, they do not affect the final ratio calculation; if they do, it is not of clinical importance.

View this table:
Table 3
View this table:
Table 4
Figure 2

Receiver operating curves (ROCs) comparing the 2 immunohistochemistry analyses (A) and the new fluorescence in situ hybridization (FISH) scoring modalities (B) with conventional FISH evaluation. These ROC curves report the sensitivity and specificity of FISH HER2 amplification (ratio >2.2).

Evaluating the discordant cases between FISH and the 2 IHC evaluations, we found means of 8.98% false-positive (IHC 3+ and unamplified FISH) and 0.60% false-negative (IHC 0 or 1+ and amplified FISH) results with the conventional method. On the other hand, comparing the results of the new FISH method with those of IHC, we found means of 9.96% false-positive and 0.30% false-negative results. These results are consistent with the majority of the published studies that followed the ASCO/CAP recommendations44,45 or are within the range of, or have better values than, the results reported by Sauter et al46 (5%-22% false-positive and 2%-8% false-negative scores). It has been suggested that the false-positive results could be the consequence of extra copies of chromosome 17 (polysomy 17) in some of these cases.47,48 Nevertheless, some authors have proposed that chromosome 17 polysomy should be regarded as HER2 negative.49,50 The new scoring method presented here does not differentiate between negative results and polysomy because their ratios were always less than 1.8. However, these and other published studies48,51 indicate that patients with polysomy would not benefit from trastuzumab therapy, so they should be treated as HER2 negative.

Although the ASCO/CAP guidelines recommend evaluating at least 20 nuclei,19 other authors suggest a range of 20 to 80 nuclei.17,52,53 With the traditional FISH method, only 20 isolated nuclei were evaluated in each field or DI, whereas the new method enabled all the signals to be evaluated in each DI. The analyses of the 3 DIs correspond with the evaluation of at least 60 nuclei per case, but they have not been taken into account to calculate the ratio. This simplified procedure takes the same time to carry out the HER2 status as the traditional one, but it has the advantage in that at least 3 times as many nuclei can be evaluated. So we would expect the final result obtained with this method to be more representative. In the future, the possible integration of this method in automated procedures should allow the elimination of the problem encountered with the nuclei evaluation as mentioned previously. The results obtained in the present work also appear to be more objective because the DIs can be tagged to ensure that all the signals are counted accurately. Finally, the images do not show a fading effect, so the areas can be reevaluated a long time after the sample was stained. This new method could be tested using dual-color chromogenic in situ hybridization (CISH) as a viable alternative to FISH.54,55 However, CISH has not yet been validated as a primary test.

In summary, the new method for the quantification of HER2 status in breast cancer tissues presented in this study enables HER2/CEN17 ratios to be evaluated independent of nucleus density or distribution, avoiding the problem of nuclear clusters, saving time searching for better areas, and avoiding several filter wheel changes in each of the nuclei evaluated. Another advantage of this method is that it can be incorporated into any commercial image analysis system that uses standard DI processing techniques. Moreover, because of its simplicity (only 3 steps are necessary), this method may be useful for resident physicians and researchers working with FISH. Even so, this new method should be tested further in independent laboratories to confirm its efficiency and assess its possible limitations.


Upon completion of this activity you will be able to:

  • provide the American Society of Clinical Oncology/College of American Pathologists guideline values that define the HER2 gene amplification classes and the HER2 protein expression status.

  • resolve discordance between fluorescence in situ hybridization (FISH) and immunohistochemical HER2 status.

  • describe the different steps of a digital analysis procedure for FISH HER2 scoring method.

  • compare the accuracy of conventional and new FISH HER2 procedures.

The ASCP is accredited by the Accreditation Council for Continuing Medical Education to provide continuing medical education for physicians. The ASCP designates this journal-based CME activity for a maximum of 1 AMA PRA Category 1 Credit ™ per article. Physicians should claim only the credit commensurate with the extent of their participation in the activity. This activity qualifies as an American Board of Pathology Maintenance of Certification Part II Self-Assessment Module.

The authors of this article and the planning committee members and staff have no relevant financial relationships with commercial interests to disclose.

Questions appear on p 125. Exam is located atwww.ascp.org/ajcpcme.


  • Acknowledgments: We thank María del Mar Barbera, Vanesa Gestí, Ainhoa Montserrat, Verònica Echevarría, Marc Iniesta, Anna Carot, and Rosa Cabrera. We also thank the tumor bank of the Hospital de Tortosa Verge de la Cinta within the “Xarxa de bancs de tumors de Catalunya” for collecting the human tumor samples.

  • This study was supported by grants from the Ministerio de Ciencia e Innovación of Spain (FIS 08/1527; J. Jaén); a grant from the Fundación Mutua Madrileña FMMA 2008-2009 (R. Bosch); and 5 awards from the Short Term Scientific Mission of the Cost Action IC0604 Euro-Telepath (COST-STSM-IC0604 -4046, -5043, -5044, -5045, -5046).


View Abstract