A novel detection methodology for HER2 protein quantitation in formalin-fixed, paraffin embedded clinical samples using fluorescent nanoparticles: an analytical and clinical validation study

The human epidermal growth factor receptor-2 (HER2) is a transmembrane tyrosine kinase receptor that plays an important role in regulating normal cell growth, differentiation and survival [1]. Over-expression of the ERBB2 gene or HER2 protein occurs in 15–20% of all invasive breast cancers and has an important bearing on prognosis, as majority of HER2-positive tumors are associated with a more aggressive clinical course and typically a poor outcome [2]. Fortunately, new targeted therapies can directly inhibit biomarker receptors including the drug trastuzumab, which is a humanized monoclonal antibody that directly targets HER2 over-expression by binding with high affinity to an extracellular epitope of the HER2 receptor. Targeting HER2 over-expression with trastuzumab in breast cancer has proven to be remarkably successful in clinical trials, which have demonstrated significant improvements in disease free survival and overall survival in breast cancer patients in the metastatic setting [3] and adjuvant setting [4, 5]. More recently, targeting HER2 over-expression has demonstrated an excellent pathologic response to therapy in the neoadjuvant setting [6, 7]. Clinical assays to assess the HER2 status in breast cancer patients that are being considered for targeted therapy include immunohistochemistry (IHC), which detects protein over-expression, and in situ hybridization (ISH), which detects gene amplification [8]. Both assays have been clinically validated for predicting which patients will benefit from treatment that targets HER2 over-expression [9, 10], and both assays have received FDA approval. With the development of new HER2-targeting drugs and the expanding options for targeting the HER2 pathway in breast cancer and other solid tumors [11–14], accurate and reliable HER2 testing to ensure that the right patients receive the right treatment is critical.

Both the IHC and ISH methodologies have limitations. Among “HER2-positive” tumors defined by consensus criterion [9, 15] there is a range of variability of results in terms of HER2 gene amplification and the semi-quantitative measurement of protein over-expression by the conventional ISH and IHC methods [16]. IHC utilizes chromogenic detection of the receptor protein and has a limited dynamic range where the test is linear, limiting the ability to obtain accurate quantitative results [17], which may be clinically useful. HER2 ISH analysis represents a surrogate for the over-expression of the HER2 receptor protein, and is more quantitative than IHC, utilizing fluorescent dyes to enumerate gene copy numbers; however, these fluorescent signals photo bleach and fade over time [18]. Given that the target of the drug trastuzumab is the HER2 receptor protein, novel detection systems that accurately and quantitatively detect HER2 protein on the membrane of tumor cells in formalin fixed, paraffin embedded (FFPE) clinical samples in a linear fashion over a broad dynamic range would be advantageous, and may provide clinically useful information. A number of new methodologies have been developed to obtain more quantitative HER2 results from FFPE breast cancer samples at the protein level, including AQUA technology [19] and HERmark [20, 21]. Both of these technologies have demonstrated promising clinical utility; however, they require special equipment and are currently only offered at central reference laboratories, limiting broader clinical applications. Various other methods, including enzyme-linked immunosorbent assay, fluorescence-activated cell sorting (FACS), and matrix-assisted laser desorption/ionization, have all been developed to quantify protein expression; however, these methods cannot simultaneously evaluate cellular morphology [22–25].

Streptavidin-coated phosphor integrated dot fluorescent nanoparticles (PID) are brightly fluorescent particles of controlled diameter that are 100 times brighter compared with commercially available Q-dots and 10,000 time brighter compared with conventional organic dyes [26] (Fig. 1). PID nanoparticles show high photostability, do not photo bleach or fade over time, and can be used as a novel immunofluorescence detection system to quantitatively measure proteins in FFPE tissue sections (Figs. 1 and 2). An image-processing method can be used to calculate the number of PID nanoparticles on images acquired by using a versatile optical system, with a read out that includes the average PID/cell (PID score/cells) and the average number of PID nanoparticles per unit area (100μm²) (PID score/ROI100μm²) [26]. Gonda et al., previously conducted an analytic and clinical validation study for the potential utility of PID particles in quantitating HER2 membrane protein in well characterized breast cancer cell lines that demonstrated a range of levels of HER2 expression. Briefly, their results showed significant analytic performance in parameters that included precision, repeatability, linearity, dynamic range and sensitivity [26]. In the current study, additional analytic validation of PID testing in six breast cancer cell lines was performed to provide further evidence and support for PID performance. Compatibility of PID testing with clinical specimens was then evaluated in a series of 108 well characterized breast cancer specimens, comparing HER2 immunofluorescence with PID to HER2 IHC with 3, 3′-diaminobenzidine (DAB)detection and HER2 ISH gene detection. Finally, we have investigated the potential clinical utility of PID quantification of HER2 protein in a cohort of HER2-positive breast cancers undergoing trastuzumab-based neoadjuvant therapy, looking for correlations with the pathologic response to treatment.

The development of effective targeted therapy for breast cancer patients who over-express HER2 represents a major advancement in medical oncology and has arguably played a large role in heralding in a new era of companion diagnostics and predictive cancer biomarker guided therapeutic decision making [3–6]. The clinical assays currently used to assess the HER2 status in patients being considered for targeted therapy include IHC and ISH, which both have limitations; furthermore, the accuracy and reliability of these methods in routine clinical practice for patient selection remains a subject of debate. In the current study, we have used HER2 receptor expression in breast cancer as a model to explore the potential clinical utility of a novel detection technology using streptavidin coated Phosphor Integrated Dot fluorescent nanoparticles (PID) for the quantitative measurement of proteins of clinical interest in breast cancer cell lines and routine clinical samples. In a series of analytic performance studies using well characterized breast cancer cell lines, our PID method demonstrated sensitive, accurate, precise and robust quantitative measurements of the HER2 protein. We next compared HER2 PID measurements against conventional IHC and ISH analysis in 108 well characterized breast cancer samples that were randomly selected to cover a range of HER2 results. Similarly, in this methods comparison study, HER2 PID measurements showed a strong correlation with conventional IHC and ISH methods in this series of well characterized clinical breast cancer samples. Of note, IHC showed a better correlation with PID measurements compared with ISH and there was a range of PID values seen for cases that were considered IHC 3+ (median PID/cell 204.4, SD 144.4) and ISH amplified cases (median PID/cell 57.1, SD137.1).

ISH analysis, which measures average HER2 gene copy number, is a surrogate for HER2 receptor over-expression and relies on the assumption that the amount of DNA copy numbers detected will accurately reflect the amount of protein that is eventually translated in tumor cells [20]. This is likely an oversimplification, given the intricacy of the potential regulatory influences and cell processes involved in the expression of receptor tyrosine kinases such as HER2. While gene amplification is considered to be the primary mechanism of HER2 over-expression in breast cancer [31, 32], the biologic regulation of HER2 expression in tumor cells is complex and multifaceted. Gene amplification may not always correlate quantitatively with HER2 protein expression. In fact, in the methods comparison portion of the current study, we noted a broad range of HER2 protein receptor values measured by PID-scores for cases that showed HER2 gene amplification with a subset of cases that had discordance between the ISH and PID results (Fig. 8). Similarly, other studies have reported discordant results between gene amplification and quantitative HER2 protein expression where 13% of central ISH positive tumors showed discordant low levels of HER2 protein [33]. The ability to directly and accurately measure the protein levels with PID-nanoparticles may potentially circumvent the impact of transcriptional or translational regulatory events that may influence HER2 receptor expression and subsequent disease progression [34, 35].

Although accuracy and reproducibility of the PID assessment of HER2 was confirmed in the analytic validation and method comparison portions of this study, pre-analytical variables have the real potential to affect the quality of quantitative HER2 results. A previously published investigation of cold ischemic time on HER2 assay results showed that cold ischemia time up to 3 h has no deleterious effect on the detection of ERBB2 via in situ hybridization or IHC (Porter, et al., Mod Path 2013). This is certainly a potential limitation to any quantitative protein assay and clinical data is clearly needed to understand the relationship between quantitative HER2 protein expression measurements, pre-analytical variables and clinical outcomes in breast cancer patients treated with HER2-targeted therapy, especially in institutions that do not closely monitor pre-analytical variables.

As a pilot study, we sought to investigate the potential clinical utility of evaluating HER2-PID scores in the pretreatment biopsies from 34 HER2-positive carcinomas that had undergone neoadjuvant trastuzumab-based chemotherapy. Our goal was to assess correlations of HER2-PID scores with the pathologic response to treatment. The pathologic response to neoadjuvant chemotherapy has been shown to be prognostic and has been proposed as a surrogate endpoint for prediction of long term clinical benefit [29, 36]. Patients who achieved a pCR in our study had a significantly higher mean and median PID score/cell compared with patient who did not achieve a pCR (p = 0.011, Fig. 6a.). The mean and median PID scores also showed an inverse correlation with the RCB-class, which separates patients into groups based on the degree of the pathologic response seen after neoadjuvant therapy (ranging from zero or pCR to RCB class-3 representing a poor response, Fig. 6b.). Although limited by retrospective analysis, a small sample size, and lack of randomization or uniform therapeutic regimens, our findings suggest that quantitative HER2 expression levels may provide improved specificity in the selection of patients who are more likely to benefit from HER2-targeted therapy.

Conventional immunohistochemistry (IHC) with 3.3’diaminobenzidine chromogen detection is extensively used as a diagnostic tool in pathology for the detection of a wide range of proteins in FFPE clinical samples. Conventional IHC is a powerful technique, allowing the simultaneous evaluation of tissue morphology as well as the localization of specific proteins within different tissue compartments; however there are limitations in its application for predictive biomarkers in cancer. The chromogenic detection employed by IHC has a restricted dynamic range where the test is linear, limiting the ability to obtain accurate and quantitative results. The intensity of the chromogenic staining depends on the enzymatic activity of the detection system being employed and is significantly influenced by the reaction time, temperature and the substrate concentration, thus limiting the quantitative sensitivity of IHC [17, 31].

It is clear that, among breast cancers defined as “HER2-positive” by conventional testing methods, there is a wide range of variability in terms of the level of gene amplification and/or protein over-expression in tumor cells. The likelihood that a more quantitative analysis of HER2 protein expression could improve prediction of the response to HER2-targeted therapy has led to an exploration of methods that are more accurate and quantitative for protein expression levels. Numerous attempts have been made to develop such quantitative methodologies for the measurement of predictive biomarkers, such as HER2 in breast cancer samples. Jensen et al. reported on a novel amplification system (qIHC) that enables quantification of proteins directly in FFPE samples by counting dots [30] which showed the ability to measure HER2 protein accurately and over a large dynamic range in cell lines and human breast cancer tissues. This study showed a good correlation between their qIHC method and standard HER2 IHC measurements in 44 breast cancer clinical specimens; however, no data on clinical utility was reported. Miyashita et al. has developed an IHC technique with quantum dot conjugated trastuzumab for measuring HER2 that correlated proportionally with HER2 gene copy number assessed by ISH (R = 0.83). Furthermore, the level of expression of HER2 measured by this method correlated with time to progression after trastuzumab therapy in 37 HER2 positive cancers (R = 0.69) [37]. Nuciforo et al. used selected reaction monitoring mass spectrometry (SRM-MS) to quantify HER2 protein levels in FFPE breast cancer samples [38]. They showed that the absolute HER2 levels measured by SRM-MS were significantly correlated with both IHC and ISH and that their quantitative measurement of HER2 was superior to IHC and ISH for predicting outcome after treatment with HER2-targeted therapy in the adjuvant and metastatic settings [38]. Cheng et al. used quantitative immunofluorescent measurements (AQUA technology) to assess HER2 protein in core biopsies from 27 HER2-positive breast cancers enrolled in a preoperative clinical trial using trastuzumab-based chemotherapy [39]. They showed a significant correlation between high levels of HER2 measurements (mean 10,251) and pCR compared with patients without a pCR (mean 4766, p = 0.0021). This finding is strikingly similar to what is being reported in the current study and suggests that the accurate measurement of HER2 protein may help determine the likelihood of response in the neoadjuvant setting. Similarly, many other studies investigating the response to neoadjuvant HER2 targeted therapy have shown that high HER2 levels [16, 40], HER2/CEP17 ratio > 6 [41], high baseline tumor infiltrating lymphocytes [42] and HER2-enriched molecular subtype [43] are all associated with a significantly higher pCR rate. Each of these reported predictors of pathologic response to neoadjuvant therapy is likely to be associated with higher levels of HER2 receptor protein expression in tumor cells as well as possible immune mechanisms of action for HER2-targeted therapy [42].

The HERmark Breast Cancer Assay (Monogram Biosciences, South San Francisco CA) provides another method for precise quantitation of total HER2 protein expression and HER2 homodimers in FFPE breast cancer specimens [20]. When compared against centrally performed IHC and ISH results from 237 breast cancer cases, HERmark showed a 98% concordance with IHC positive and negative assay values [20]. For 94 IHC equivocal cases, 67 and 39% concordance results were seen between HERmark and ISH amplified and non-amplified results respectively. Yardley et al. has examined the prognostic significance of quantitative measurement of HER2 by HERmark in a cohort of 192 breast cancer cases enrolled in a multicenter collaborative biomarker study, most of whom received no HER2-targeted therapy [44]. While the HERmark assay results correlated well with routine HER2 testing, there was a range of levels of expression of HER2 protein measured by HERmark. When the HERmark and local test results were discordant (local testing negative, HERmark discordantly high or local testing positive, HERmark discordantly low) HERmark more accurately predicted overall survival. This result suggests that the quantitative measurement of HER2 protein may more accurately reflect the underlying tumor biology in cancers that are labeled as “HER2-positive” by routine testing in terms of prognosis. Lipton et al. has reported on the potential role for quantitative measurement of HER2 protein by HERmark for predicting outcome after treatment with trastuzumab in metastatic breast cancer patients. In this study of 98 patients who had central ISH and HER2 total expression results, Cox proportional hazards multivariate regression identified HER2 total expression as an independent predictor of time to progression (HR = 0.29, P = 0.0015) and overall survival (HR = 0.19, P < 0.001). These authors concluded that a subset of patients with HER2 gene amplification by ISH express low levels of HER2 protein and have a reduced response to HER2-targeted therapy, similar to ISH-negative cases [33].

PID nanoparticles demonstrate great potential for the quantitative measurements of proteins of clinical interest in routine clinical samples with morphologic confirmation of the tissue being studied. Using HER2 expression in breast cancer as a model, we have shown that PID measurements demonstrated sensitive, accurate and precise quantitation of HER2 protein in six breast cancer cell lines, is highly correlated with conventional HER2 testing methodology in clinical samples, and may be useful for predicting pathologic response to trastuzumab-based neoadjuvant therapy. Several major limitations are present in the current study, which could potentially confound these analyses, including 1) the small sample size that was used to evaluate the correlation between the PID scores and the pathologic response to neoadjuvant therapy, 2) the retrospective nature of the study, and 3) the non-uniformity of the treatment regimens received. Thus our findings should be considered preliminary and hypothesis-generating until they can be confirmed in additional larger, well-controlled neoadjuvant studies for patients with HER2-positive breast cancer. That being said, our data suggests that the quantitative measurement of HER2 protein levels, where the drug is targeted, may offer an advantage over conventional testing in the selection of patients who will be the most likely to benefit from HER2-targeted therapy, which is a conclusion that is supported by other reports in the literature. Further studies in a larger patient cohort undergoing neoadjuvant therapy are warranted.