Impact of the OncotypeDX score and HER2 RNA PCR levels on HER2-low IHC levels in primary and metastasized tumors

Breast cancer (BC) is the most prevalent diagnosed malignancy worldwide [1]. It is a highly heterogeneous disease consisting of biological entities that differ in clinicopathologic features, susceptibility to treatments, and prognosis. Clinicians manage their decision-making approach based upon 3 main subtypes derived from immunohistochemistry (IHC) and fluorescent situ hybridization (FISH) analyses: luminal or HR-positive/HER2-negative (hormone receptor-positive/human epidermal growth factor receptor 2-negative), HER2-enriched and triple-negative or HR-negative/HER2-negative [2]. Overexpression of the HER2 subtype occurs in approximately 25% of BC cases [3]. HER2 is primarily divided into HER2-positive (IHC = 3 + or 2 + and FISH amplified) and HER2-negative (IHC = 0/1 + or 2 + and FISH not amplified). Targeted anti-HER2 therapies enabled an improvement in the prognosis of HER2-positive BC patients [4]. This dichotomic classification, however, benefited only a minority of BC patients, specifically, the HER2-positive subgroup [5]. Most HER2-negative BCs still express HER2 on cancer cells to some extent [6], differentiated HER2-low (IHC = 1 + or 2 + and FISH not amplified), which is now targetable, from HER2-0 (IHC = 0) [7].

Novel anti-HER2 agents for HER2-low BC, such as trastuzumab deruxtecan (TDXd; Enhertu, Daiichi Sankyo, Inc.), could improve prognosis for some patients currently excluded from HER2-targeted therapy, thereby expanding the reach to a much larger share of BC patients [7]. Patients diagnosed with HR + /HER2-"negative" early-stage BC undergo the OncotypeDX test to determine which patients will benefit from chemotherapy and to what degree. The OncotypeDX test quantifies the expression of 21 genes by means of real-time reverse transcription-polymerase chain reaction (qRT-PCR). One of them is the HER2 gene [8].

Despite its wide use by pathologists, the IHC test has some limitations. One factor that contributes to the discrepancy between labs is poor adherence to 2018 HER2 testing recommendations and faulty technical execution of the test. Another limitation might stem from insufficient sensitivity to accurately detect low levels of HER2 expression [6]. In the past, distinguishing between HER2 IHC 0 and IHC 1 + scores was not a primary concern for pathologists due to limited clinical impact. The introduction of targeted therapies such as TDXd has shifted this perspective, emphasizing the clinical significance of this differentiation. Accurate classification of HER2-low status now plays a vital role in guiding treatment decisions for optimal patient care. Consequently, there may be a need for careful reassessment of previous pathological reports.

All of these restraints might undermine the reliability of the test and consequently reduce the sensitivity of the HER2-low classification. Here, this study aimed to identify parameters that are associated with HER2-low levels in primary and relapsed tumors. We specifically sought to determine whether the RT-PCR test that measures HER2-mRNA expression could identify patients who could be classified as HER2-low.

We characterized several factors that differentiate the HER2-low group from the HER2-0 group. Our results showed a positive correlation between a higher RT-PCR score and a HER2-low status, suggesting that RT-PCR could potentially enhance the prediction capability of the traditional IHC test in borderline cases. HER2-low patients are expected to express higher levels of mRNA compared with HER2-0 patients, and their results should therefore be more commonly found at the higher end of the RT-PCR scale. This partially explains the overlap between IHC subgroups (Fig. 1 a and b) and enables us to focus upon the area of the scale where there is a high percentage of patients classified as HER2-low. We further showed that the correlation between HER2 RT-PCR and IHC results, stratified by tumor size groups, was enhanced for larger tumors (Fig. 1e). These results reinforce the notion that HER2 RT-PCR can be a good indicator of HER2 IHC status, especially in tumors larger than 2 cm, but it does not contribute to the distinction in tumors smaller than 1 cm. Indeed, prior studies have yielded analogous findings, indicating a positive association between higher RT-PCR scores and HER2-low status. However, those studies were constrained by much smaller cohort sizes [13, 14].

We also found that larger tumors were significantly associated with HER2-low status. Given that HER2 belongs to the epidermal growth factor receptor family, whose activation results in cell proliferation and tumorigenesis [15], it is not surprising that tumors that overexpress HER2 will be characterized by accelerated proliferation and growth in size compared with tumors that do not overexpress HER2. A comparison of ER IHC levels with HER2-0 and -low status also linked higher ER IHC levels to HER2-low status. This finding raises the possibility of crosstalk between HER2 and HR pathways and may contribute to HER2-low phenotype variation, as observed in other studies [5, 16].

We showed a significant increase in HER2 expression compared with primary BC in distant metastases (Fig. 2), and that finding correlates with tumors being more aggressive, as reported by others [5, 17]. A comparison of the HER2 evolution between primary and matched distant-staged BCs revealed that the majority of patients (65.9%) gained or maintained some degree of HER2 expression. HER2 pathway expression has also been suggested as a mechanism of resistance to hormonal treatment, thereby explaining HER2 gain (even at low levels) in metastatic samples [18]. Interestingly, we showed that based upon the OncotypeDX score category, the low-risk group (< 11) had a higher HER2 IHC score than the medium-risk group (11–25). Despite not reaching a level of significance, brain metastases were more common among patients with increased HER2 vs. no-increase HER2 patients. Several studies have demonstrated that patients with HER2-positive BC are more commonly affected by brain metastases than HER2-negative/HR + patients [19]. This correlation may also be consistent with HER2-low BC compared with HER2-0 BC. Further research with a larger paired cohort could elucidate the mechanism underlying this finding.

The HER2 treatment paradigm changed with the emergence of TDXd as the only FDA-approved medication to treat HER2-low patients. Drug eligibility lies entirely upon IHC test findings, and incorrect classification may result in missing patients who qualify for its use. Technical limitations (staining techniques), interobserver differences, intratumoral heterogeneity, or insufficient sensitivity can undermine the reliability of IHC as the sole test for classification. Indeed, the HER2-enriched subtype was infrequent and similarly distributed in HER2-low and HER2-0 BCs [20]. HER2 intratumoral heterogeneity has been linked to low-grade or HER2 IHC = 2 + BCs in several reports [21‐24]. Another study showed that HER2 intratumoral heterogeneity was also more frequent in HER2-low cases than in HER2-positive cases, and those authors emphasized that smaller tumor samples are more prone to inaccurate assessment of HER2 status [23]. Moreover, primary tumors are larger than metastatic biopsies and therefore are less susceptible to false assessment.

Intratumoral heterogeneity of HER2 has an effect in another direction as well. Internalization of TDXd releases a payload of DXd in the target cell in addition to having a bystander effect, which causes death of neighboring cells regardless of their HER2 status [7]. Bearing intratumoral heterogeneity in mind, even patients classified by IHC as HER2-0 might potentially benefit from the drug. Tissue heterogeneity, technical limitations and bystander effects point to whether it is appropriate to insist upon taking a sample from a metastatic tumor such as in DESTINY-Breast04 [7], which is smaller, or rely on sample results from the larger primary tumor. ESMO guidelines emphasize new biopsies upon metastatic relapse [25]. However, when infeasible, treatment aligns with primary tumor traits. ASCO also allows oncologists to rely on primary samples, given that metastatic samples are more susceptible to pre-analytic conditions than primary breast tissue samples [26]. The decision-making process should also evaluate the prognostic costs of false positive and false negative classifications in terms of a patient who will needlessly receive the medication or a suitable patient who will be ruled out.

Artificial intelligence (AI) offers a promising solution for the inaccuracies of IHC/FISH testing, enhancing precision and inter- and intraobserver reproducibility while potentially reducing the need for molecular testing [27, 28]. Nevertheless, Wu et al. (2023) highlight that AI's impact on accuracy was most pronounced among novice pathologists [27]. Similarly, Palm et al. (2023) suggest AI's potential in distinguishing positive and negative HER2 samples, which encountered challenges in HER2 IHC 0/1 + classification [28]. A potential implication of our observation could be by implementation of the OncotypeDX with HER2 levels by AI to overcome pathologists' subjectivity. AI thus represents a significant advancement, hinting that a multidisciplinary approach could potentially provide the most accurate solution.

This study has several limitations. First, we used only paired samples to characterize the evolution of HER2. It is possible that we would have obtained more significant results had the paired cohort been larger. In addition, this is a single-center study with no access to follow-ups from other institutions. Although it may create some degree of bias, the rates of metastatic recurrence and the long median follow-up duration are similar to those described in other studies. Furthermore, our results are consistent with those of other reported analyses in terms of the median follow-up time [29] and HER2 enrichment [5, 17]. Another limitation stems from the majority of metastatic tumor biopsies having been collected during/after ET, which potentially altered the HER2 expression result. However, ET duration was practically identical among the HER2 increase vs. no-increase HER2 groups. Indeed, digital evaluation of images can sometimes lead to higher estimates compared to microscopic images. Nevertheless, in our research, the number of digital samples was limited, and no statistical difference was observed.

Evaluation of HER2 status in bone metastases via IHC is susceptible to underestimating HER2 scores due to the decalcification process [30]. Therefore, prioritizing nonbone sites for biopsies is favorable. Indeed, we observed that 50% of bone samples were classified by IHC as 0. Nonetheless, considering that bone is the most likely site for breast metastases to reach, we believe it is advisable not to overlook data originating from bone altogether.

Differences among antibodies can affect HER2 assessment. Ventana's 4B5 (which was predominantly used in our institution) is less sensitive than DAKO HercepTest (polyclonal and new monoclonal) [31]. Rüschoff et al. (2022) [32] found that DAKO HercepTest (new monoclonal) was more specific than Ventana's 4B5 and DAKO HercepTest (polyclonal). We claim that HER2 IHC = 0/1 + by Ventana might still benefit from TDXd because Ventana's 4B5 might underestimate the true extent of HER2-low tumors.

The new HER2-low classification and the implications of using ADCs in this setting have revolutionized breast oncology. A debate surrounds whether HER2-low status affects BC biology, necessitating gene expression analysis, or if the receptor's presence alone suffices as a therapeutic target for ADC binding, irrespective of pathway activation. In SABCS 2022, both perspectives are presented [33]. Should the latter perspective hold true, it would call into question that patients with HER2-0 status by IHC paired with pathway activation through HER2 RT-PCR would gain benefits from therapies such as TDXd. Additionally, clinical evidence for ADCs beyond traditional IHC remains limited.

Taken together, our study findings demonstrate that HER2-low status is associated with a higher RT-PCR score, larger tumor size, and higher ER expression. We also showed enrichment of HER2 in the metastatic setting. Revisiting primary BC settings could complement the imperfect widely applied IHC test, which would be especially relevant to borderline cases. These findings may not only enable us to more accurately characterize HER2-low but also provide HER2-low patients with more appropriate treatment, potentially expanding the patient population eligible for novel anti-HER2 treatment.