Consensus guidelines for analysis and interpretation of the EGFR FISH assay

These consensus guidelines are the result of a global workshop convened to discuss key technical issues and standardise reading strategies for the EGFR FISH assay of NSCLC tumour tissue. All laboratories should also comply with all guidelines required by the regulatory agencies in their countries.

Sample preparation and storage guidelines

Pre-analytical variables can affect the final result of the EGFR FISH assay. For consistently robust and reproducible results, standardisation of the pre-analytical conditions and processes, as well as those of the assay itself, and interpretation of results are needed. Largely, the recommendations regarding pre-analytical variables are similar to the recommended guidelines for HER2 FISH testing in breast carcinoma.29

Fixation and storage of tumour specimens

The duration of fixation and the type of fixative used are key issues that could affect the viability of samples and modify subsequent FISH results. Samples should be fixed for greater than 6 h but no more than 48 h in 10% neutral buffered formalin/4% neutral buffered formaldehyde. The date and duration of fixation should be recorded for each specimen whenever feasible. Bouin’s, mercury-containing fixatives and any fixatives that would destroy DNA should be avoided. Ideally, tumour samples should be stored as blocks (formalin fixed, paraffin embedded) at room temperature, rather than as cut tissue sections.

Selection of tumour specimens

Tumour content of the specimen should be confirmed in a H&E-stained slide serial to the unstained sections prior to performing subsequent steps of FISH assays or prior to sending the sample to an external laboratory. It is preferable to use the diagnostic tumour blocks to ensure availability of a sufficient number of viable, morphologically intact tumour cells to fulfil the scoring requirements (see below) and histopathological representation of the patient’s tumour as a whole.

Sectioning of tissue

Positively charged slides should be used for mounting of tissue sections in order to avoid detachment of the tissue. The recommended thickness of tissue sections used for the EGFR FISH assay is 4 μM (range 1 μM), and the sectioning date should be recorded. If slides with sections are stored prior to testing, the date of sectioning and section storage conditions (room temperature, protected from light and humidity) should also be noted.

EGFR FISH assay

Laboratories may follow the standard operating procedures that have proved to be successful for formalin-fixed, paraffin-embedded tissue sections in their settings for performing the assay. Attention must be paid to particular steps in the protocol that may affect probe penetration and hybridisation to target DNA in order to optimise signal intensities. The deparaffinisation process must remove all traces of residual paraffin, therefore the solvents should be changed for fresh solvents on a regular and frequent basis according to the number of sections being processed. Modifications in the protease digestion protocol may be required for specific cases (eg, large surgical specimens), and small biopsies may require different digestion times, as well as freshly cut and previously stored slides. Samples with abundant mucinous cells or dense fibrous tissue may require longer digestion times than those with non-mucinous or more delicate stromal matrix.

After hybridisation, sealing the coverslips in place with rubber cement, finger nail polish, etc, is not necessary; however large coverslips should be used to avoid mixing of immersion oil and the 4′,6-diamidino-2-phenylindole (DAPI)/anti-fade mounting medium. The hybridised slides should be stored in the dark at −20°C to +4°C until analyses and review are completed. Microscope analysis is recommended to be performed within 1 week of the assay in order to guarantee the optimal fluorescence intensity.

Reading the EGFR FISH assay

Qualifications of personnel reading EGFR FISH

A properly trained reader who has undergone training in FISH should analyse the slides. The reader also should have had training on the pathological appearance of lung cancer, and should have easy access to assistance from an expert lung pathologist. The competency of all technical personnel reading and interpreting the results should be evaluated at least annually and continuous training should be ensured. The participation of the laboratory in quality assurance/quality control programmes is highly advised, as recommended in the HER2 testing guidelines.29

Quality assessment of the EGFR FISH specimen

Once the assay control slides have been assessed and passed quality criteria, the quality of each tumour specimen must be assessed to determine its acceptability for analysis. At least 50% of tumour cells have to pass the assessment test fulfilling the following criteria prior to the actual reading and scoring. (a) Verification of integrity of the tumour nuclear morphology using the DAPI filter. Within adequate specimens, the chromatin of tumour cells should be well defined and non-disrupted (fig 1A, B). The preparation should not be undertreated to the point of preventing clear identification of the nuclear border, or overtreated with chromatin missing from nuclear areas. Tumour nuclei should not be covered by a cloudy yellowish layer or obscured by autofluorescent structures. Examples of specimens that do not pass quality assessment are illustrated in fig 1C–F. (b) Verification of the quality of the hybridisation in tumour and non-tumour cells using single-pass and double-pass interference filters. Green signals (CEP7) should be bright, compact (occasionally slightly stringy or diffuse) oval shapes, and red (EGFR) signals should be bright, small round shapes (fig 1A, B). The CEP7 signal is usually larger and brighter than the EGFR red signal and they are commonly adjacent. Signals should not be fuzzy, very patchy or blotchy. Representative images of those conditions are provided in fig 2A–D. (c) Confirmation that the background of the specimen appears dark and free of fluorescent particles or haziness under inspection with single-pass and double-pass interference filters.

Selection of tumour areas and tumour nuclei

Using H&E-guided microscopy prior to FISH reading, the histopathological features of the tumour should be assessed, including identification and characterisation of invasive tumour area(s), cell size, cell shape, differentiation, growth pattern, and presence of inflammatory or necrotic areas, which should be avoided. The tumour foci should always be identified on the H&E-stained slide by a lung pathologist, independently of whether the FISH analysis is to be performed by a pathologist or not.

Five areas representative of the histopathological features of the invasive part of the tumour must be selected. The selected areas are identified in the FISH slide with the DAPI filter, and within each area an average of 10 nuclei should be selected for analysis, for a total of 50 counted nuclei.

Selection of tumour nuclei for reading should be performed using high-magnification objectives according to the following criteria: (a) with the DAPI filter, identify tumour nuclei of average size or larger, with non-disrupted chromatin and non-overlapping borders (as illustrated in fig 3); (b) with the single green, single red or dual red/green filters, confirm that selected cells carry at least one copy of each EGFR and CEP7 signal.

The areas and level of heterogeneity in the tissue should be noted, if heterogeneity is present. Tumours may be heterogeneous in their susceptibility to protease (eg, depending on extracellular matrix composition) and the quality of the EGFR signal may be different among these areas.

Enumeration of signals

The number of EGFR and CEP7 signals in the selected nuclei should be counted with single-pass interference filters (one filter for each probe wavelength) at ×60 to ×100 magnification and recorded separately. In the signal enumeration, there are special situations that require careful attention, such as when signals are split, paired, or in triplets or clusters (figs 4 and 5). Doublet or triplet spots that are physically linked, ie, touching or linked by a thread, or adjacent (with a gap smaller than the diameter of the largest signal), are counted as one signal only. However, spots that are adjacent but separated by at least the diameter of the largest signal are counted as separate signals. The maximum number of signals to be counted in a single cluster is 15. If signals are >15 the cluster is considered as innumerable and reported as “16”. A number is still required for calculation of the overall EGFR/CEP7 ratio, but counting in this range becomes increasingly difficult.

Due to the variety of mechanisms of EGFR amplification and clustering patterns noted in NSCLC samples, recommendation is provided in table 2 on how to score and interpret clusters of EGFR and CEP7 signals in these unique cases. Raw data should always be carefully recorded and stored, as they may be found to be useful in future studies.

Number of readers for EGFR FISH

In its implementation at the University of Colorado (see Interpretation of the EGFR FISH results according to the Colorado scoring criteria), the EGFR FISH assay was scored by two independent readers, with a third reader involved when there was discordance between the first and second. Because of increased demand for efficiently testing large numbers of patients, a study was performed to establish a FISH evaluation procedure that would minimise the number of readers and reduce the number of specimens requiring multiple readers (L Morrison, personal communication 2008). The study utilised a mathematical approach to identify an equivocal range to determine if the sample was FISH positive or FISH negative. Based on the results of this study, the consensus recommendation was to set the high end of the equivocal range as 40% cells with ⩾4 signals, above which the first reader’s result classifies the specimen as EGFR FISH-positive. The low end was set at 25% of cells with ⩾4 signals, below which the first reader’s result classifies the specimen as EGFR FISH-negative. If the score of the first reader was within the equivocal or borderline range of >25% to <40% of cells with ⩾4 signals, then a second reader, blinded to the results of the first reader, was required. If the second reader reaches a positive score the specimen is concluded as FISH positive, and as FISH negative otherwise.

Interpretation of the EGFR FISH results according to the Colorado scoring criteria

A scheme for classifying NSCLC tumours as EGFR FISH-positive and FISH-negative was developed at the University of Colorado and has been used in multiple clinical studies.20 28

EGFR FISH-positive

(A) Specimens that have ⩾40% of cells displaying ⩾4 copies of the EGFR signal.

(B) Specimens that display EGFR gene amplification, defined according to one of the following criteria: (a) an EGFR to CEP7 ratio ⩾2 over all scored nuclei and calculated using the sum of EGFR divided by the sum of CEP7 when mean CEP7 per cell is ⩾2 copies; (b) the presence of gene cluster (⩾4 spots) in ⩾10% of tumour cells; (c) at least 15 copies of the EGFR signals in ⩾10% of tumour cells.

EGFR FISH-negative

Specimens that do not display gene amplification according to the criteria defined above and with <40% of cells displaying ⩾4 copies of the EGFR signal.

Future directions and conclusions

There are several areas that need further investigation. For instance, the application of the Colorado scoring system requires adaptation and subsequent validation for cytological specimens and smears since those specimens include whole cells and do not have truncated nuclei as in the tissue sections.

The unique biology of advanced-stage NSCLC also pose difficulties associated with tumour heterogeneity that might get unaccounted for in small biopsies or even in very large specimens. Potential differences may exist in the genomic status of EGFR in pretreated versus untreated patients and the vast majority of information currently available was generated in diagnostic specimens from patients who had not received prior treatment. The predictive value of EGFR FISH in primary tumours compared with their corresponding metastases is another issue that requires confirmation in future clinical trials. Yet to be determined are the potential benefits or drawbacks of the FISH technology versus the related methods of chromogenic in situ hybridisation/silver in situ hybridisation for evaluation of the copy number of the EGFR gene in NSCLC.

Last, the development of robust quality assurance programmes and regular proficiency and competency assessment of laboratories performing these assays are necessary to improve the consistency of assay performance and interpretation. It is our intention that these guidelines will provide a tool to help the clinical laboratories involved with the EGFR FISH test in NSCLC to accomplish these goals.