Correlation of In Situ HER2 RNA Expression With HER2 Immunohistochemistry and Fluorescence In Situ Hybridization Categories in Breast Cancer

We used formalin-fixed, paraffin-embedded (FFPE) specimens from 259 patients with BC, collected between 2006 and 2019 by the Department of Pathology of National Taiwan University Hospital (Taipei, Taiwan), covering all HER2 IHC/FISH statuses with adequate sample numbers in each category. All IHC 2+ cases had available reflex FISH results from medical charts, classified into G1 to G5 as based on the updated 2018 ASCO/CAP guideline.⁴  Additional HER2 FISH tests were performed on all IHC 3+ cases (n = 28) and selected IHC 1+ (n = 39) and IHC 0 (n = 30) cases. Other clinicopathologic information was obtained from the original pathology reports. Estrogen receptor (ER) or progesterone receptor (PR) positivity was defined by using a 1% cutoff. Hormone receptor (HR) positivity and negativity was defined as positivity for either ER or PR or negativity for both ER and PR, respectively. The study protocol was approved by the Institutional Review Board of National Taiwan University Hospital (approval No. 202201031RINB).

HER2 FISH tests were performed with the PathVysion HER2 DNA Probe Kit (Abbott, Abbott Park, Illinois) according to the manufacturer’s instructions. HER2 ISH was performed with the RNAscope FFPE 2.5 kit (Advanced Cell Diagnostics, Hayward, California) according to manufacturer’s guidelines. Briefly, FFPE sections were treated with citrate buffer and protease before being hybridized sequentially with HER2 probes (RNAscope Probe-Hs-ERBB2), preamplifier, amplifier, and label probes. Hybridization signals were detected by DAB (3,3′-diaminobenzidine) staining. The slides were visualized by microscopy (×400 magnification), and at least 40 tumor cells were evaluated for each case. isHRE was semiquantitatively categorized into 6 RNAscope scores on the basis of a modification of previously described criteria^18,19 : score 0 (<1 individual dot/tumor cell), score 1 (1–3 individual dots/tumor cell), score 2 (4–9 individual dots/tumor cell), score 3 (10–15 individual dots/tumor cell), score 4 (>15 individual dots/tumor cell and ≤50% clustered dots), and score 5 (>15 individual dots/tumor cell and >50% clustered dots, or diffuse dense cytoplasmic signal). For all cases with a score of 1 or lower, RNAscope with a probe for the housekeeping gene peptidylprolyl isomerase B (PPIB) (Advanced Cell Diagnostics) was performed to evaluate RNA integrity. The method for PPIB scoring was the same as for HER2 scoring. Cases with a PPIB score of 1 or lower were considered to have degraded RNA and were excluded from subsequent analyses. All slides were independently scored by 2 breast pathologists (H.-C.L., Y.-H.L.) who had no knowledge of the clinicopathologic features at the time of scoring. An additional pathologist (H.W.H.) joined to review cases with discrepant scores. A final consensus was reached and recorded for all cases.

For each case, tumor part from 6 slides (10 μm) were macrodissected. RNA extraction was performed with RNeasy FFPE kits (QIAGEN, Hilden, Germany). RNA was reverse transcribed into cDNA with MMLV (Moloney murine leukemia virus) reverse transcriptase kits (Protech, Taipei, Taiwan). Quantitative reverse transcription–polymerase chain reaction (qRT-PCR) was performed in triplicate on a QuanStudio 7 Flex Real-time PCR using Power SYBR Green PCR Master Mix according to the manufacturer’s instructions (Applied Biosystems, Foster City, California). Normalized values were calculated by using the delta-delta cycle threshold (dCT) method. HER2 gene expression in all samples was normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH). The primers used in this study are listed in Supplemental Table 1 (see the supplemental digital content containing 2 tables and 2 figures at https://meridian.allenpress.com/aplm in the March 2024 table of contents).

Data processing, analyses, and plotting were conducted with GraphPad Prism 6 and 9 (GraphPad Software Inc, San Diego, California) and IBM SPSS Statistics for Windows version 19 (IBM, Armonk, New York). κ coefficient analysis was used to assess interobserver agreement for HER2 RNAscope score evaluations. Pearson χ² tests or Fisher exact tests were used for comparison of categorical variables when appropriate. Statistical significance of differential HER2 expression was determined by unpaired 2-tailed t tests. Statistical significance was defined as P < .05.

To determine whether HER2 RNAscope scoring could be used as a semiquantitative analysis of the HER2 levels, we evaluated the correlation between HER2 RNAscope scores and HER2 RNA levels measured by qRT-PCR (dCT [CtGAPDH – CtHER2]) in 113 BC cases classified as HER2 IHC 0/1 (n = 54), IHC 2+ (n = 40), and IHC 3+ (n = 19). There were significantly higher HER2 RNA levels in IHC 3+ versus IHC 2+ cases (P < .001) and IHC 2+ versus IHC 0/1+ cases (P = .004) (Figure 1, A) as well as in HER2-positive (IHC 3+ or IHC 2+/FISH⁺) versus HER2-negative BCs (P < .001) (Figure 1, B), supporting the use of qRT-PCR in estimating isHRE in BCs. Next, we demonstrated a positive correlation between HER2 RNAscope scores and isHRE, evidenced by significantly higher dCT values in cases with a score of 5 versus 4, 4 versus 3, and 3 versus 2 (Figure 1, C). The difference in dCT between score 1 and score 2 cases was not statistically significant. Similarly, we demonstrated significantly higher RNAscope scores in HER2-positive BCs than in HER2-negative BCs (P < .001) (Figure 1, D). These results indicate that RNAscope scoring could be applied as a semiquantitative method to evaluate isHRE.

A total of 259 cases covering all HER2 IHC/FISH categories were evaluated by HER2 RNAscope. A PPIP control was conducted by RNAscope in all 43 cases with a score of 1, and 25 additional randomly selected cases (score 2 [n = 18] and score 3 [n = 7]). Two of the cases with a HER2 score of 1 had a PPIB score of 1 or lower and were excluded. Two cases with heterogeneous RNAscope patterns were also excluded. The remaining 255 cases were used for further analysis and the clinicopathologic features are shown in the Table. The HER2 IHC 2+/FISH results were as follows: G1 (23.4%), G2 (18.4%), G3 (15.8%), G4 (24.7%), and G5 (17.7%). The subtypes observed were HR⁺/HER2⁻ (56.5%), HR⁺/HER2⁺ (26.3%), HR⁻/HER2⁺ (9.0%), and HR⁻/HER2⁻ (8.2%). The overall κ coefficient was 0.827 (P < .001). Representative images demonstrating HER2 RNAscope scoring are shown in Figure 2, A through S.

To investigate isHRE among BCs of various HER2 IHC/FISH categories, we correlated RNAscope scores with HER2 IHC/FISH results for 255 BC cases. Significantly higher RNAscope scores were observed in HER2 IHC 3+ than IHC 2+ cases (P < .001) and in IHC 2+ than IHC 0/1+ cases (P < .001) (Figure 3, A). Within the IHC 2+ FISH subgroups, significantly higher scores were observed in G1 than G2 (P < .001), G3 (P < .001), G4 (P < .001), and G5 (P < .001). G2 cases had scores lower than G3 (P = .05) and G4 (P = .02) but were not significantly different from G5. The scores between G3, G4, and G5 were not significantly different. Despite the overall higher scores in IHC 2+/G1–5 cases than in IHC 0/1+ cases, the IHC 2+/G2 scores were not significantly different from those of IHC 0/1+. Because G1 and G3 are the only 2 IHC 2+/FISH categories currently considered as FISH positive, we further compared the scores between these 2 categories. Since all G3 cases had HER2 copy numbers between 6 and 10, we also limited the comparison to G1 cases with HER2 copy numbers in that range. We observed significantly higher scores in G1 than G3 cases when HER2 copy numbers were controlled for (Figure 3, B). Together, these findings demonstrate an overall positive correlation between HER2 RNAscope scores and HER2 IHC status (0/1+, 2+, and 3+). Additionally, scores were significantly higher in G1 cases and lower in G2 and G5 cases, among IHC 2+/FISH groups.

Next, we investigated the impact of HER2 FISH parameters, namely HER2 and CEP17 copy numbers and HER2:CEP17 copy number ratios, on isHRE in the 216 BCs with available HER2 FISH results. The frequency of each HER2 FISH parameter in various HER2 IHC/FISH categories is shown in Supplemental Figure 1. For each HER2 IHC/FISH category, we separated cases into those with greater than or equal to or those with less than the median HER2 or CEP17 copy number or HER2:CEP17 copy number ratio, respectively, and compared their HER2 RNAscope scores (Supplemental Table 2). We observed that cases with higher HER2 copy numbers, defined as greater than or equal to the median HER2 copy number, were significantly correlated with higher scores in G1, IHC 3+ and IHC 0/1+, but not G2 to G5 HER2 categories (Figure 4, A, and Supplemental Figure 2, A). In contrast, with the exception of the IHC 0/1+ category, higher CEP17 copy numbers were not correlated with higher scores in HER2 categories (Figure 4, B, and Supplemental Figure 2, B). Cases with a higher HER2:CEP17 ratio were correlated with higher HER2 RNAscope scores only in the IHC 2+/G1 HER2 subgroup (Figure 4, C, and Supplemental Figure 2, C). Together, these findings demonstrate that the impact of HER2 FISH parameters on isHRE varies between HER2 categories.

Finally, we sought to compare HER2 genomic status by FISH, and HER2 RNA expression status by isHRE and qRT-PCR, between HER2-low and HER2-negative cases with available HER2 FISH and qRT-PCR data. We observed significantly higher HER2 copy numbers, HER2:CEP17 copy number ratios, HER2 RNAscope scores, and dCT values in HER2-low than in HER2-negative BCs (Figure 5, A through D). When HER2-low cases were stratified into IHC 2+/FISH-negative and IHC 1+ groups and analyzed separately, all these HER2 parameters remained significantly higher in IHC 2+/FISH-negative than in IHC 0 cases. However, none of these 4 parameters were significantly different between IHC 1+ and IHC 0 cases (Figure 5, A through D). Together, these findings demonstrate heterogeneous HER2 statuses within the HER2-low BCs: HER2 copy number, HER2:CEP17 ratio, isHRE, and dCT were significantly higher in IHC 2+/FISH-negative than in IHC 0 BCs, but comparable between IHC 1+ and IHC 0 BCs.

In the present study, we demonstrate significantly higher dCT in score 5 versus 4, score 4 versus 3, and score 3 versus 2 cases. There was no significant difference in dCT between score 1 and score 2 cases. Compared to HER2 RNAscope, which measures isHRE, qRT-PCR measures RNA levels from dissected tissue in toto that contains both tumor and nontumor cells and might not clearly distinguish the slightly higher HER2 RNA expression in tumor cells over background nontumor cells. This may partly explain the lack of a significant difference in HER2 RNA expression between score 1 and 2 cases. Nevertheless, our finding of an overall positive correlation between HER2 RNAscope scores and HER2 RNA levels quantified by qRT-PCR supports RNAscope scoring as a validated semiquantitative method to evaluate isHRE. Through RNAscope scoring, we demonstrate isHRE in samples from BC cases with various HER2 IHC/FISH categories with a κ coefficient of 0.827 between observers. We also demonstrate the impact of HER2 FISH parameters on isHRE and the isHRE of HER2-low and HER-negative BCs. These results support the conclusion that HER2 RNA ISH by RNAscope could be a useful, convenient, reproducible, and tissue-relevant diagnostic adjunct to evaluate the HER2 expression status in BCs.

We show significantly higher HER2 RNAscope scores in IHC 3+ than IHC 2+ BCs, and in IHC 2+ than IHC 0/1+ BCs, illustrating the correlation between isHRE and in situ HER2 protein expression. Among the IHC 2+ BCs, we show that G1 cases had the highest RNAscope scores among the 5 FISH groups tested. Although IHC 2+/FISH G1 and G3 groups are currently considered to be HER2 positive,⁴  the significantly lower RNAscope scores in G3 than G1 BCs, even when their HER2 copy numbers were controlled for, and the comparable scores between G3 and G4 cases suggest that despite both being considered HER2 positive, IHC 2+/FISH G1 and G3 BCs may have different HER2 RNA expression levels. HER2 RNA expression determined by RNA sequencing was demonstrated as a significant predictor of pCR for anti-HER2 therapy.⁵  Our finding underscores the need to reevaluate HER2 expression and anti-HER2 therapeutic efficacy in the IHC 2+/FISH G3 group, in light of the critical role of HER2 RNA levels in anti-HER2 treatment.⁵ 

HER2 RNAscope scores in IHC 2+/FISH G2 BCs were significantly lower than in G1 BCs, but not different from G5 cases, supporting the revision of ISH G2 as HER2 negative in the 2018 ASCO/CAP guideline. However, Rakha et al⁶  recently reported a pCR rate of 27% in IHC 2+/FISH G2 cases, which was not significantly different from that of IHC 2+/FISH G1 cases, although HER2 RNA expression levels were unevaluated. While more cases are needed to draw a conclusion regarding the eligibility for HER2-targeted therapy in IHC 2+/FISH G2 cases, HER2 RNA expression should be considered with respect to anti-HER2 therapeutic efficiency in this rare HER2 FISH group. In addition, despite HER2 RNAscope scores in G4 BCs being higher than those in FISH G2 BCs, they were significantly lower than the scores in FISH G1 cases. These findings support the revision of ISH G4 from equivocal to negative ISH, with respect to HER2 RNA expression.⁴ 

In this study we show that the impact of HER2 FISH parameters on isHRE varied among FISH groups. Among IHC 2+/FISH G1 to G5 cases, we show that both higher HER2 copy number and higher HER2:CEP17 copy number ratio, defined as greater than or equal to the median, correlated with significantly higher isHRE in FISH G1, but not in FISH G2 to G5 cases. Higher HER2 copy number was also correlated with higher isHRE in IHC 3+ BCs. Because both IHC 2+/FISH G1 and IHC 3+ BCs were associated with significantly higher HER2 expression among the HER2 IHC/FISH categories, our finding suggests that the positive impact of HER2 copy number and HER2:CEP17 ratio on isHRE was evident in cases with higher HER2 RNA levels. In fact, response to neoadjuvant chemotherapy plus anti-HER2 targeted therapy occurs more frequently in tumors with a HER2 copy number higher than 11.5 and HER2:CEP17 ratios above 3.7.²⁰  In addition, in a study of HER2-targeted therapy without chemotherapy, none of the 11 patients with a HER2 copy number below 10 and/or HER2:CEP17 ratio below 4 achieved pCR, compared to 29% of patients with a HER2 copy number above 10 and/or HER2:CEP17 ratios above 4.²¹  Considering that HER2 RNA expression level was demonstrated as a significant predictor of pCR to anti-HER2 targeted therapy,⁵  our findings are in line with the aforementioned observation, suggesting a positive impact of HER2 copy number and HER2:CEP17 ratio on isHRE, which in turn correlates with therapeutic efficacy.

Although the increases in HER2 copy numbers that result from CEP17 polysomy can lead to HER2 protein overexpression in HER2-nonamplified BC cases,^22–26  the impact of CEP17 copy number on isHRE among various HER2 IHC/FISH categories remains unclear. We demonstrate that isHRE did not significantly differ between cases with greater than or equal to or cases with less than the median CEP17 number in IHC +3 cases and all 5 IHC 2+/FISH groups, suggesting that compared to HER2 copy number and HER2:CEP17 copy number ratio, CEP17 copy numbers do not impact HER2 RNA expression.

We also found that both isHRE and HER2 genomic status (HER2 copy number and HER2:CEP17 copy number ratio) were significantly different between HER2-low and HER2-negative BCs. When HER2-low cases were stratified into IHC 2+/FISH-negative and IHC 1+ groups and analyzed separately, all HER2 parameters remained significantly higher in IHC 2+/FISH-negative than in HER2-negative cases. None of these parameters were significantly different between IHC 1+ and IHC 0 BCs. Our finding is supported by a recent study by Shu et al²⁷  that showed that HER2 1+ and 0 tumors had similar HER2 mRNA levels. The significantly different HER2 parameters found between IHC 0 and IHC 2+/FISH-negative but not IHC 1+ BCs in the present study highlight the heterogeneity in HER2 status within HER2-low–positive BCs. Moreover, a recent phase 3 randomized trial demonstrated clinical benefit on HER2-low–positive BCs,¹⁶  but no significant difference in survival was observed between IHC 2+/FISH-negative and IHC 1+ cases upon further analysis. Although the diagnosis of HER2 IHC in this study was achieved by consensus among the pathologists, the lack of difference between these HER2 IHC/FISH groups in these studies may reflect the inability of pathologists to reliably distinguish 0 from 1+ and 1+ from 2+/FISH-negative cases by IHC. In fact, IHC scoring accuracy for HER2-low–positive and HER2-negative cases among pathologists was shown to be poor,²⁸  but may be improved through some form of assistance such as online training modules. Nevertheless, since HER2 RNA expression was a significant predictor of pCR for anti-HER2 therapy in HER2-positive BCs,⁵  our finding may suggest that comparing the clinical benefit of novel antibody-drug conjugates in HER2 IHC 0 and 1+ cases is warranted.

In summary, our results present the isHRE status among BC cases of various HER2 IHC/FISH categories and demonstrate the impact of HER2 FISH parameters on HER2 expression. We found heterogeneous HER2 status within HER2-low BCs where the isHRE of IHC 0 BCs was significantly lower than that of HER2 IHC 2+/FISH-negative BCs, but not that of IHC 1+ BCs. Our findings may be relevant for anti-HER2 treatment decisions in view of the critical role of HER2 RNA expression in predicting anti-HER2 therapeutic efficacy.

We would like to thank Uni-edit (www.uni-edit.net) for editing and proofreading this manuscript.