PDF
ePub
Citation
Print
INTRODUCTION
Breast cancer is a biologically heterogeneous disease with diverse natural history and different clinical behaviors. As a result, there has been an increasing shift toward a personalized approach for breast cancer management. To provide the right treatment based on the patient's underlying tumor biology, tumor biomarkers such as estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) are routinely used to determine the optimal and most effective therapy. Expression of hormone receptors is observed in approximately 70% of invasive breast cancers and acts as a strong predictive biomarker. Patients expressing ER and/or PR may probably benefit from endocrine therapy [1], which is highly effective and less toxic than chemotherapy. Patients with tumors containing as few as 1% invasive cells are known to gain clinical benefits from endocrine therapy [2]. The guidelines of the American Society of Clinical Oncology/College of American Pathologists (ASCO/CAP) recommend a cutoff of >1% ER/PR-positive tumor cells to distinguish “positive” from “negative” cases [3]. Therefore, careful examination and accurate assessment of immunohistochemistry (IHC) positivity is absolutely vital to avoid missing potentially significant focal staining.
The HER2 (also referred to as ERBB2) gene, located at chromosome 17q12-21, encodes a transmembrane receptor tyrosine kinase of the human epidermal growth factor receptor family [4] and is overexpressed in approximately 15% to 20% of breast cancers [5]. HER2 (also called epidermal growth factor receptor ErbB-2) signaling pathway promotes cell proliferation and survival, and HER2 overexpression is associated with an aggressive tumor phenotype and reduced survival [6]. Patients with HER2 amplification usually overexpress HER2 protein and display worse prognosis than patients with normal HER2 levels [47]. Trastuzumab, a monoclonal antibody against the extracellular domain of HER2 protein, is the first HER2-targeted agent that showed efficacy as a single agent or in combination with chemotherapy in patients with HER2-positive invasive breast cancer [8910]. Other targeted agents such as lapatinib, a reversible dual HER1/2 tyrosine kinase inhibitor, and pertuzumab were also developed and have improved the prognosis of patients with HER2-positive breast cancer [1112]. In recent years, treatment with trastuzumab has become the standard therapy for patients with HER2-positive breast cancer [13] and HER2 testing is recommended at the time of diagnosis for all breast cancers [14]. Although trastuzumab is effective, it can be toxic in some patients; therefore, accurate assessment of HER2 overexpression is mandatory to identify patients who may benefit from this targeted therapy.
In 2013, ASCO/CAP recommended two diagnostic methods for the determination of HER2 status: IHC and in situ hybridization (ISH). IHC uses a protein antibody to detect HER2 protein, whereas ISH employs a DNA probe with a fluorescent, chromogenic, or silver detection system to determine the number of HER2 gene copies [1516]. Although the ASCO/CAP guidelines provide relatively clear instructions for the assessment of HER2 status, inter-laboratory or inter-observer variability may occur owing to the technical issues or difficulty in the interpretation of test results. Interpretations of the results of IHC or ISH for HER2 status by pathologists may show inter- or intra-observer differences, as both methods involve semi-quantitative or qualitative analyses. Therefore, quantitative methods are needed for the determination of HER2 expression level to achieve good agreement with the HER2 status among pathologists.
In this direction, the possibility of using quantitative real-time reverse transcription polymerase chain reaction (qRTPCR) as an ancillary test to complement the more subjective methods of IHC and fluorescence in situ hybridization (FISH) was raised. Several studies have reported the performance of qRT-PCR to determine HER2 status [171819]. These studies concluded that molecular approaches such as qRT-PCR are powerful and reliable quantitative methods for HER2 status assessment and may complement IHC or FISH for optimal patient treatment. In particular, in the era of personalized molecular target therapy, a high-throughput and cost-effective screening tool such as qRT-PCR becomes even more important for deciding cancer treatment strategies. An optimal assay for molecular studies should be sensitive and specific as well as easy to perform, readily interpretable, and reliable. qRT-PCR is relatively easy to perform and cost-effective, but the requirement for reverse transcription and PCR amplification may lead to potential error or bias in target gene assessment.
NanoString nCounter® gene expression system is a recently developed RNA-based technology that allows digital quantification of multiplexed target molecules through the use of color-coded barcodes. It can provide discrete counts of target mRNA transcripts using a small amount of total RNA without the need for amplification [20].
This study was performed to assess the usefulness of NanoString nCounter® gene expression system for the quantification of mRNA expression in archived formalin-fixed, paraffin-embedded (FFPE) invasive ductal carcinoma samples as well as to correlate the mRNA expression level of ER, PR, and HER2 by NanoString nCounter® gene expression system with the results of IHC/FISH and qRT-PCR.
METHODS
Patients and samples
A total of 240 cases of surgically resected invasive breast cancer from 1995 to 2012 were retrospectively retrieved from the computerized records system at Samsung Medical Center. Among them, 45 cases were HER2-positive breast cancer patients treated with adjuvant trastuzumab, 143 cases included triple-negative breast cancer, and 52 cases were ER/PR-positive breast cancer. All patients provided written informed consent.
Two pathologists (E.Y.C. and J.H.) reviewed the pathologic findings to determine the following variables: histological subtype, tumor size, nuclear grade, histological grade according to the modified Bloom-Richardson grading system, presence of lymphovascular invasion, and immunohistochemical profiles of ER, PR, Ki-67, and HER2 without the knowledge of NanoString, qRT-PCR, and HER2 FISH results.
The Institutional Review Board of Samsung Medical Center, Seoul, Korea approved our study protocol and waived the need for informed consent, as the study was conducted using archival tissues with retrospective clinical data (IRB number: 2017-07-088). All investigations were conducted according to the principles of the Declaration of Helsinki.
Immunohistochemistry and fluorescence in situ hybridization
We determined the IHC positivity for ER and PR protein according to ASCO/CAP guidelines of a threshold of 1% and by Allred score [21] using antibodies specific for ER (clone 6F11; Novocastra, Newcastle upon Tyne, UK) and PR (clone 16; Novocastra) [3]. HER2 status was evaluated using a specific antibody (HercepTest; Dako, Glostrup, Denmark) and/or by FISH. HER2 expression was also defined according to ASCO/CAP guidelines [15]. IHC grades of 0 and 1 for HER2 were defined as a negative result, while grade 3 was recorded as a positive result. Amplification of HER2 was confirmed using FISH, if a sample was rated 2+ by IHC.
FISH was performed using a dual-color DNA-specific probe kit from PathVision™ (Vysis LSI® HER2 SpectrumOrange™ and CEP17 SpectrumGreen™; Abbott Moleclular, Des Plains, USA) in cases with equivocal HER2 IHC results (2+). A total of 50 nuclei per sample were evaluated under a fluorescence microscope (Zeiss Axioskop, Oberkochen, Germany) using filter sets recommended by Vysis (4´,6-diamidino-2-phenylindole [DAPI]/Spectrum Orange dual bandpass, DAPI/Spectrum Green dual bandpass). All overlapping nuclei were excluded. The gene copy number ratio was determined according to ASCO/CAP guidelines [15]. After assessment of expression profiling by IHC/FISH, cases were classified into molecular subtypes [22]. In particular, ER or PR+/HER2− tumors with low proliferation index were considered luminal A, high-grade ER or PR+/HER2− tumors and ER or PR+/HER2+ tumors were considered luminal B, whereas ER and PR−/HER2+ tumors were considered HER2-enriched subtype. Triple-negative tumors were considered basal-like [23].
RNA extraction
All available hematoxylin and eosin-stained slides from FFPE primary breast tumor tissue were reviewed. Areas containing representative invasive tumor were carefully marked on the stained slide and microdissected from 2 to 4 paired 4-µm-thick unstained FFPE sections with a pointed surgical blade. Non-tumor areas were removed by manual microdissection. Total RNA was extracted using the high pure RNA paraffin kit (Roche Diagnostic, Mannheim, Germany) according to the manufacturer's instruction. The yield and purity of RNA was evaluated using the NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Rockland, USA). All samples had a total RNA concentration greater than 50 ng/µL and were, therefore, included in the analysis.
Real-time quantitative reverse transcription polymerase chain reaction
RNA from archival specimens was measured by real-time qRT-PCR using the TaqMan gene expression assays (Applied Biosystems Inc., Foster City, USA) according to the manufacturer's instructions. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) gene was used as an endogenous control (Applied Biosystems; assay ID, HS99999905_m1). RT-PCR was conducted using the high capacity cDNA reverse transcription kit (Applied Biosystems) according to the manufacturer's instructions. PCR amplification of exons 20 and 21 of HER2 was performed using the oligonucleotide primers 5´-GGAGCTGAG GAAGGTGAAGGT-3´ (forward) and 5´-GATCCAGAT GCCCTTGTAGACTGT-3´ (reverse). After incubation for 10 minutes at 95℃, amplification was performed in a thermocycler (Applied Biosystems) with 40 cycles of denaturation at 95℃ for 30 seconds, annealing at 60℃ for 30 seconds, extension at 72℃ for 60 seconds, and a final extension at 72℃ for 10 minutes. Each complementary DNA sample was analyzed in triplicates with ABI PRISM 7500HT fast real-time PCR (Applied Biosystems). CT, the fractional cycle number at which the amount of amplified target reached a fixed threshold, was determined and the mRNA expression level of HER2 was measured using the 2−ΔΔCt(ΔΔCt=ΔCtarget gene−ΔCGAPDH) method.
NanoString nCounter® system assays
The nCounter®assay was performed using the NanoString nCounter® Analysis System (NanoString Technologies, Seattle, USA). Hybridization reactions were performed according to the manufacturer's instructions. An nCounter® CodeSet (NanoString Technologies) containing a biotinylated capture probe for target genes and six housekeeping genes (CLTC, GAPDH, GUSP, HPRT1, PGK1, and TUBB) and reporter probes attached to color-barcode tags according to the nCounter® CodeSet design was hybridized to 200 ng of total RNA for 18 hours at 65℃. Samples were processed using an automated nCounter® Prep Station. Hybridized samples were purified and immobilized in a sample cartridge for data collection, followed by the quantification of target mRNA in each sample using the nCounter® Digital Analyzer. For each reaction, 600 fields of view were counted. Quantified expression data were analyzed using nSolver analysis software (NanoString). The resulting counts were normalized to the average counts for all control spikes in each sample. After performing image quality control using a predefined cutoff value, we excluded the outlier samples using a normalization factor based on the sum of positive control counts greater than 3-fold. The counts of the probes were then normalized using the geometric mean of the six housekeeping genes (CLTC, GAPDH, GUSP, HPRT1, PGK1, and TUBB). Data from all 240 patients with breast cancer were included in the statistical analysis.
Statistical analysis
We used the Spearman rank correlation coefficient test for the comparison of (1) HER2 gene expression level between NanoString and qRT-PCR and (2) HER2 gene expression level by NanoString and HER2 average gene copy number by FISH. Jonckheere-Terpstra test and Cohen kappa statistic were used to compare the NanoString or qRT-PCR results for HER2 expression with HER2 IHC results. The correlation between ESR1 and PgR gene expression by NanoString and IHC profiles for ER and PR protein expression was evaluated by the Mann-Whitney test and Cohen kappa statistic. A value of p<0.05 was considered statistically significant. Statistical analyses were performed using SPSS software version 18.0 (SPSS Inc., Chicago, USA). To set the cutoff value (low vs. intermediate/high) of the NanoString counts for specific genes (ESR1, PgR, and HER2), receiver operator characteristic (ROC) curves were generated by dichotomizing IHC data. ROC curves were analyzed using SPSS and MedCalc software (MedCalc Software, Mariakerke, Belgium).
RESULTS
Clinicopathological characteristics of 240 breast cancer cases
The general clinicopathological characteristics of 240 breast cancer patients are presented in Table 1. The median age at diagnosis was 46 years (range, 24–74 years). Invasive ductal carcinoma was found in 91.3% of cases and lymph node involvement was present in 42.9% of cases. High histological and nuclear grades were found in 155 patients (64.6%). Of 240 cases, 37 (15.4%) and 17 (7.1%) showed HER2 positive (3+) and equivocal (2+) results on IHC, respectively. ER-positive and PR-positive tumors were noted in 57 (23.8%) and 53 (22.1%) cases, respectively. According to the results of IHC/FISH profile for 240 cases, 48 cases were ER or PR(+)/HER2(−), 21 cases were ER or PR(+)/HER2(+), 24 cases were ER or PR(−)/HER2(+), and 143 cases were ER or PR(−)/HER2(−). Two of 240 cases were ER or PR(−) but showed HER2 equivocal results in both IHC and FISH. Two of 240 cases were ER or PR(+) but failed to show adequate results for HER2 FISH.
Comparison between NanoString and IHC/FISH for ER, PR, and HER2 expression
To evaluate the concordance of NanoString nCounter® and conventional methods for determining the molecular subgroup of breast cancer, we compared the NanoString gene expression level for ESR1, PgR, and HER2 with IHC results for ER, PR, and HER2 proteins and FISH results for HER2 gene. The expression level of each gene using NanoString nCounter® showed a good correlation with levels of the corresponding protein biomarkers by IHC (p<0.001) (Figure 1).
The samples with ER-negative results on IHC showed nCounter® levels from 5.03 to 491.01 (median, 28.11), while those with ER-positive results on IHC displayed nCounter® levels from 10.58 to 4,594.15 (median, 426.8).
The samples with PR-negative results on IHC showed nCounter® levels from 2.35 to 219.32 (median, 18.14), while those with PR-positive results on IHC displayed nCounter® levels from 5.66 to 4,190.71 (median, 74.02). A statistically significant difference was observed between mean nCounter® levels of negative and positive ER/PR IHC groups (p<0.001) (Figure 1A and 1D, 1B and 1E, respectively).
The samples with HER2 IHC score 0 showed nCounter® levels from 147.18 to 2,302.25 (median, 508.97) and the samples with HER2 IHC score 1 showed nCounter® levels from 306.42 to 4,293.16 (median, 1,077.85). The equivocal cases with HER2 IHC (2+) had nCounter® levels ranging from 1,207.76 to 10,127.69 (median, 1,948.67), while the positive cases with HER2 IHC (3+) showed nCounter® levels from 2,229.67 to 72,436.31 (median, 11,666.14). A statistically significant difference was observed between mean nCounter® levels of negative, equivocal, and positive HER2 IHC groups (p<0.001) (Figure 1C, F).
The gene expression level for HER2 using NanoString nCounter® showed a good correlation with the gene amplification status for HER2 (p<0.001) (Figure 2A-C).
Comparison between HER2 gene expression by qRT-PCR and HER2 expression by IHC/FISH
We evaluated the correlation between HER2 gene expression level by qRT-PCR, HER2 IHC results, and FISH data. The gene expression level (relative quantitation [RQ]) for HER2 using qRT-PCR showed a good correlation with HER2 protein expression results with IHC (p<0.001, box plot not shown) but not with HER2 gene amplification status with FISH (p=0.124) (Figure 2D-F).
Estimation of cutoff values for gene expression using ROC curves
Comparisons between gene expression and IHC data showed good overall agreement with a high area under the curve (AUC) for ESR1/ER (AUC=0.939), PgR/PR (AUC=0.796), and HER2/HER2 (AUC=0.989) (p<0.001). The cutoff values of each gene, estimated with consideration of sensitivity and specificity for the detection of specific genes, are shown in each graph (Figure 3).
Concordance of NanoString level and IHC/FISH expression
We compared the results between NanoString level and IHC/FISH for ER, PR, and HER2 expression (Table 2). Using NanoString cutoff for ER and PgR expression, we found that the concordance rate of ER and PR expression status by IHC and NanoString method was 82.1% and 79.6%, respectively (Cohen kappa=0.596 and 0.461, respectively).
Using NanoString cutoff for HER2 expression, we found that the concordance rate of HER2 expression status by IHC and NanoString method was 94.2% when we excluded the cases with equivocal HER2 expression results for IHC (Cohen kappa=0.833).
Comparison between the results of IHC, FISH, and NanoString counts for HER2
The general distribution of HER2 for 240 specimens is presented in Figure 4. FISH analysis was performed for 25 cases, including all 17 equivocal cases by IHC and eight controls that had relatively ambiguous NanoString counts in comparison with other cases. Adequate results were obtained for 23 cases. Two of the equivocal cases failed to produce appropriate FISH results due to poor sample condition.
The concordance rate of HER2 expression status by IHC and NanoString method was 94.2% and the discordance rate was 5.8% when we excluded the cases with equivocal HER2 expression results for IHC.
One case showed HER2 negativity (1+) in invasive cancer but HER2 positivity in ductal carcinoma in situ (DCIS) by IHC. In the invasive area, FISH result was equivocal and the NanoString count was 4,293.16.
Among the 17 cases that were equivocal for HER2 by IHC, five showed negative FISH results and their median NanoString count was 1,948.67 (range, 1,425.16–2,330.70). All these cases were considered positive by NanoString level. Two of the 17 cases showed equivocal results by FISH and their median NanoString count was 1,266.14 (range, 1,207.76–1,324.52). All these cases were considered negative by NanoString level. Eight of the 17 equivocal HER2 cases by IHC showed positive FISH results and had a median NanoString count of 4,859.76 (range, 1,589.55–10,127.69). All these cases were considered positive by NanoString level. Two cases of equivocal HER2 by IHC failed to yield adequate FISH results. Therefore, seven of 15 cases (46.7%) that were HER2 equivocal by IHC showed discrepancy between NanoString and FISH results.
Of the 15 HER2 equivocal cases by IHC that had FISH results, seven cases (46.7%) revealed an elevated CEP17 count, indicative of polysomy. Five of these seven cases that showed FISH negative/equivocal results had an elevated CEP17 count. The median HER2 and CEP17 copy numbers of FISH-negative cases (n=5) were 3.26 (range, 2.84–3.38) and 3.1 (range, 2.46–3.78), respectively, while the median HER2 copy number of FISH-equivocal cases (n=2) was 4.43 (range, 4.22–4.64). All cases of CEP17 copy number of FISH-equivocal cases were 4.08. Among the eight HER2 equivocal cases on IHC with FISH-positive results, two showed elevated CEP17 count with median HER2 and CEP17 copy numbers of 9.09 (range, 5.30–11.60) and 4.89 (range, 1.20–6.26), respectively.
Comparison of NanoString and qRT-PCR for HER2 gene expression
We performed comparison between the sensitivity and specificity of NanoString nCounter® and qRT-PCR for gene expression level. NanoString counts for the HER2 gene showed high correlation with RQ levels generated by qRT-PCR (p<0.001) (Figure 5A). Although both methods showed good overall agreement between gene expression and IHC data with a high AUC for HER2/HER2 (AUC of 0.989 for NanoString and 0.880 for qRT-PCR; p<0.001 for each method), the AUC of the NanoString method was greater than that of qRT-PCR. Pairwise comparison of the two ROC curves showed a statistically significant difference (p<0.001) (Figure 5B). As AUC is a measure of the overall value of a diagnostic test, these data indicate that NanoString has better overall performance as a diagnostic test than qRT-PCR.
DISCUSSION
In this study, we described an ancillary method for the detection of the molecular status of invasive breast cancer using NanoString-based gene expression technology. This method relies on the direct digital detection of gene expression. The main goal of this study was to compare the results of NanoStringbased gene expression with those of the conventional IHC/FISH methods for the assessment of protein/mRNA expression. We specifically focused on the three most common molecular markers in breast cancer: ER, PR, and HER2. These three biomarkers are regarded as targets for specific treatment in breast cancer and have important value in molecular classification with respect to the survival rate of patients with breast cancer [23].
At present, the combination of IHC/FISH is routinely used for the detection of ER, PR, and HER2 status. We evaluated the accuracy of NanoString-based assay by comparing it with the conventional biomarker detection method. The sensitivity and specificity of NanoString-based assay as compared with that of IHC was 94.7% vs. 78.7% for ER, 67.9% vs. 83.4% for PR, and 96.3% vs. 93.6% for HER2. Although we have compared these methods using only three biomarkers, the sensitivity and specificity of NanoString-based assay appears to be significantly reliable [2425]. However, some discrepancies were reported in the ER/PR protein expression status by IHC and the gene expression status by NanoString, which may be associated with the presence of the normal breast tissue adjacent to the tumor. Therefore, careful microdissection of invasive cancer is essential for the evaluation of ER/PgR gene status using NanoString method. The discordance rate of HER2 level was low, as HER2 protein overexpression was found only in tumor tissues. Some discrepancy in HER2 protein expression and gene expression was also observed in a relatively small number of cases. A majority of discordant cases were HER2 equivocal by IHC/FISH. One of them contained approximately 1% HER2 negative (1+) invasive cancer and approximately 99% HER2 (3+) DCIS. FISH result was equivocal in the invasive area. Some of the HER2-positive DCIS components may have been included in the microdissected samples of invasive carcinoma in the course of the experiment. With the exception of HER2 equivocal cases by IHC/FISH, 12 out of 240 cases were inconsistent and IHC/FISH (−)/NanoString (+). This observation may be related to the technical factors associated with the experiment and interpretation of raw data [25].
Of the 15 HER2 equivocal cases by IHC that had FISH results, elevated CEP17 count (polysomy) was observed in five of seven cases that showed FISH-negative/equivocal results and two of eight cases with FISH-positive results. Although the clinical implications of elevated CEP17 count (polysomy) are ambiguous, some studies have linked elevated CEP17 count with adverse clinical features and HER2 overexpression [2627]. In addition, studies have shown that elevated CEP17 copy number may account for trastuzumab response in tumors with a normal HER2:CEP17 ratio [2829]. This is the reason some pathologists suggest the use of mean HER2 copy number rather than HER2:CEP17 ratio to determine HER2 status in presence of CEP17 co-amplification [16]. According to this recommendation, the absolute quantification of HER2 gene expression is very helpful for the determination of HER2 status. A recent study [30] reported an association between high levels of HER2 and achievement of a pathologic complete response with neoadjuvant treatment. This suggests that accurate measurement of HER2 may predict the likelihood of response in the presurgical setting. In this direction, NanoString assay represents a strong alternative tool to evaluate the HER2 status, as it is quantitative, reproducible, and easy to perform. Many studies have focused on the quantification of HER2 expression by qRT-PCR and there is a good correlation between qRT-PCR and conventional IHC/FISH methods [18]. However, AUC of NanoString method was observed to be greater than that of qRT-PCR in our study. As AUC is a measure of the overall value of a diagnostic test, our data suggest that NanoString method displays better overall performance as a diagnostic test than qRT-PCR. In addition, qRT-PCR is more sensitive to RNA quality and requires a gene amplification process, which may be a cause of error in the determination of gene status.
Until recent years, oncologists would require an appropriate tumor specimen to detect various potential molecular markers for personalized cancer treatment. Thus, there is a need for an efficient biomarker profiling assay that can be performed with a limited tumor sample. Hence, high-throughput gene screening methods such as NanoString-based assay are useful for comprehensive cancer genome study of individual patient. In particular, NanoString method can profile 250 genes associated with breast cancer using only 200 ng of total RNA. Moreover, NanoString assay has outstandingly high sensitivity, reproducibility, and a wide dynamic range [20].
Although, NanoString-based assay offers several advantages, there are some limitations associated with this method. In comparison to IHC/FISH or qRT-PCR, this technology is relatively expensive mainly due to the acquisition and maintenance of devices. However, this method can be considered as cost-effective owing to its ability to analyze a variety of genes at once with a limited sample. Aside from its high cost, it is difficult for this method to present the delicate tumor characteristics with genetic complexity of tumor, including intratumoral heterogeneity, copy number variations (polysomy), structural alteration in the target gene, and structure and function of proteins predicted to be altered by genetic alteration [25]. However, this method may be useful by specifically focusing on the expression levels of most commonly screened, targetable markers such as ER, PR, and HER2 in breast cancer.
Our results revealed high concordance between NanoString-based gene assay and IHC/FISH test for ER, PR, and HER2 in breast cancer. Therefore, NanoString-based assay may be a feasible screening tool for initial identification of breast cancer patients that can undergo treatment with specific targeted agents.
In conclusion, our results suggest a high concordance between IHC assessment of ER, PR, and HER2 and NanoString-based multigene assay. The NanoString-based multigene assay showed high specificity and sensitivity in the evaluation of the molecular status of breast cancer. Based on our study, the comprehensive NanoString-based panel of 250 genes may be a useful and efficient high-throughput screening tool for the identification of genetic alterations of potential therapeutic targets, including ER, PR, and HER2, in the era of personalized medicine.
XML Download