Assessment of HER2 status in breast cancer: overall positivity rate and accuracy by fluorescence in situ hybridization and immunohistochemistry in a single institution over 12 years: a quality control study

Since the FDA approval of Herceptin, the determination of HER2 status became routine in processing breast cancer specimens [1‐3]. HER2 status can be assessed on both surgical specimens and core biopsies using national and international guidelines [3, 4]. The FDA approved methodologies include the assessment of the protein level by immunohistochemistry (IHC) or gene copy count on the DNA level by in situ hybridization technology (ISH) [2‐5]. Current guidelines give clear specifications as to quality assurance, participation in proficiency testing and defined scoring systems [3, 5, 6]. Furthermore, ASCO/CAP guidelines include the achievement of a concordance level of at least 95% for both positive and negative tumors when using IHC and ISH technology [3, 5, 6]. As a result of more than a decade intensive research in HER2 testing in breast cancer, it became apparent that testing results can be crucially influenced by pre-analytical and interpretation issues. Concordance between IHC and ISH results vary from 50-100% in the literature [1, 2, 6‐9]. Economic issues of cost-effectiveness of IHC or ISH as primary test have been the issue of several previous studies and are still debated [10, 11].

We are not aware of any similar studies on systematical analysis of HER2 positivity-rate and accuracy in a solely routine diagnostic setting using three different approaches as IHC primary testing with complimentary FISH assay, FISH tests as primary testing and double testing using IHC and FISH on a large diagnostic breast cancer cohort.

In this study we demonstrate that concordance-rate between IHC and FISH HER2 status in breast cancer significantly improved over 12 consecutive years and FISH only HER2 testing resulted in a constant amplification-rate. We followed three different approaches in a single institution using IHC assays complemented with FISH, FISH only testing and double IHC/FISH testing in an exclusively diagnostic setting examining 7714 consecutive breast cancers.

Accuracy of HER2 testing in breast cancer is of enormous importance as a subsequent therapy decision for or against chemotherapy and Herceptin therapy is directly subjected to the HER2 status determined by the pathologists [3].

When using FISH methodology as primary testing between 2005 and 2010, we observed a constant positivity-rate from 16-17%, which slightly dropped to 13-14% after implementing the modified ASCO guidelines in 2008.

There are only a very few papers on large consecutive cohorts available in the literature, addressing the stability and changes of HER2 positivity over several years of testing [14‐16]. Two large cohorts (n > 1000) described that HER2 positivity-rate dropped from 21-26% to 11-14% when tests were performed consecutively on primary breast cancers from 2003 and 2012 [14, 16]. These two studies used one sole diagnostic approach: Bilous et al. followed primary IHC testing with complementary FISH assays in a multicenter setting, Vergara et al. conducted a double analysis (IHC and FISH) on each case in a single institution. We used three different diagnostic algorithms over 12 years in a single institution and could show that HER2 IHC positivity-rate was fairly variable, whereas FISH HER2 positivity-rate remained constant over the years. Similarly to the data above, a slight decrease in HER2 positive cases were observed immediately after implementation of the modified ASCO/CAP guidelines 2008 also in our collective [3, 14, 16]. In another smaller cohort no differences were detected when HER2 positivity-rate was compared between 2003/2004 and 2008/2009, though the case number enrolled in this study was rather small (n < 1000) [15].

The decrease in HER2 positivity-rate is potentially to explain by the introduction of mammography screening with improved detection of early breast cancers [15, 16]. Screen detection resulted in a shift to a different patient population with less HER2 positive cases in early breast cancer and with younger age at diagnosis [15, 16]. Organized mammography screening in some cantons of Switzerland (started in the 1999’s) influenced disease stage distribution and had an impact on improved mortality [17]. This is clearly different in metastatic settings with no relevant change in HER2 positivity based on literature data [15, 16].

The modified definition for cut-offs for HER2/CEP17 ratios and for score 3+ IHC presumably also played a role in the improved congruency between IHC and FISH technology in our cohort [3]. Similarly to our own results, two larger studies also reported a drop in HER2 positivity and simultaneously a better congruence after implementing the modified guidelines from 2008 [16, 18]. Interestingly, the positivity-rate by FISH when using this assay as the primary test in our institution (n = 4843), resulted in a rather constant positivity-rate prior to and after applying the modified ASCO criteria. Additionally to strict adherence to the criteria mentioned above, it is important to note, that both IHC and FISH HER2 assays are subjected to inter-observer variability without thorough expertise [2, 13, 19]. There are clear recommendations in the literature as to how many HER2 assays one pathologist should analyze per year and to limiting the number of pathologists who are involved in HER2 diagnostic service [2]. We believe that constant FISH HER2 positivity over many consecutive years in our institution can be also connected to the small number of max 2 pathologists who exclusively interpreted FISH signals.

Standardization of technical issues as pre-analytical processing, time to fixation and the use of divergent fixatives are considered as further reasons for incongruent test results [6, 20‐22]. Based on established literature data it can be assumed that the way and duration of antigen retrieval and fixation can lead to inconsistent and incorrect IHC results [6, 20‐22]. FISH (and ISH) technology is less prone to fixation and laboratory errors than IHC tests, making FISH assay more reproducible if signal interpretation is carried out by experienced pathologists [6, 12, 13]. As reported and recommended by Tubbs and Hicks several years ago, the choice for FISH as primary HER2 testing should be considered if discordance issues with IHC and FISH assays and/or inter-observer interpretation difficulties occur in the given institution [12, 13]. Our choice for FISH only testing in our institution from 2005 was triggered exactly by these considerations. The poor performance and concordance between IHC and FISH assays 2001–2004 in our institution were potentially due to pre-analytical issues and laboratory as well as interpretational difficulties. We assume that the semi/manually conducted HER2 IHC stains with the highly sensitive and less specific CB11 antibody combined with signal interpretational difficulties in a large group of pathologist analyzing HER2 IHC stains, led to this poor performance [2, 23, 24]. It is well known that fixation can be significantly affect test results in IHC as variable fixation can make HER2 overexpressed protein undetectable (false negative) or native HER2 protein more detectable (equivocal or non-interpretable) [6, 21, 22].

According to current ASCO guidelines there is no need for complementary FISH testing at IHC score 0/1+ cases [3]. We experienced FISH amplification in up to 6% of score 0/1+ samples in 2001–2004 with a considerable improvement of <1% in 2011–2012. Data on FISH amplification in score 0/1+ cases is inconsistent in the literature as FISH testing is not done systematically. In this IHC group, Gown reported a 0.8% FISH amplification-rate. In contrast, Martin et al. reported HER2 amplification-rate by FISH in up to 14% in IHC score 0/1+ cases [25, 26]. We assume that IHC score 0/1+ with FISH amplification represent most likely false negative results due to laboratory errors and fixation issues. Furthermore we believe that the improvement in this IHC group 2011–2012 is a consequence of automatic and standardized pre-analytical and interpretation issues, in a similar matter as the improvement in score 3+ cases from 2011.

Interestingly, the percentage of technically not interpretable or equivocal cases also showed a decrease over the last years [18]. In a large cohort, Middleton et al. reported a drop from 10 to 3.4%, which is quite similarly to our own observations (3.6% to 1.6%) [18].

The need for standardization and quality controlled HER2 testing were the subject already addressed in the early 2000s, when HER2 testing in breast cancer became a predictive marker for Herceptin therapy [1, 23, 24, 27, 28]. Useful practical recommendations became available from the early papers dealing with HER2 testing as to the use of commercially available and validated antibodies, internal validation, fixation issues, reporting and proper training of pathologists participating in HER2 testing [1, 23, 24, 27, 28].

Even from these early years there was a concern about incorrect HER2 results, the importance of all points mentioned above did not loose from their merit until now. Recent papers on HER2 testing add to the recommendations mentioned above, the need for pathology institutions to participate in external national or international quality assurance and proficiency programs together with develop national/international guidelines for HER2 testing [5, 26, 29‐35]. During the last years, pitfalls in HER2 testing as polysomy and co-amplification of HER2/CEP17 were explored and published, which pathologists must be also aware of, when reporting HER2 status in breast cancer [36‐38].

The question whether there is a best methodology as the gold standard for HER2 testing remains controversial [2, 3, 6]. Whatever methodology is used in the given pathology institution, standardized pre-analytical, analytical and scoring issues need to be guaranteed in order to provide accurate HER2 test results in the diagnostic setting.