Molecular Detection of Lymph Node Metastases in Breast Cancer Patients: Results of a Multicenter Trial Using the One-Step Nucleic Acid Amplification Assay

Sentinel lymph node (SLN) biopsy for breast cancer is expected to become a standard surgical procedure in the near future, and accurate assessment of metastasis of SLNs is essential for making decisions about the avoidance of unnecessary axillary dissection and the provision of appropriate adjuvant treatment for patients. However, methods for the pathologic examination of SLNs to detect metastasis remain controversial (1–4). Although more detailed examination of SLNs can provide more accurate information about metastasis (5), to obtain more accurate results, a comparatively greater number of pathologic specimens need to be examined (6). This involves much time for preparation of the specimens and a heavy workload for pathologists to examine them, especially intraoperatively.

To overcome these problems, molecular detection of metastasis has been developed as one of the most promising methods for SLN examination. With this procedure, the whole lymph node can be examined during a short time without requiring much work for the pathologist. Very recently, several molecular-based metastasis detection procedures with proper calibration for clinical use have been developed and are expected to become alternatives to conventional pathologic examinations (7–10). The one-step nucleic acid amplification (OSNA) assay (Sysmex), an automated system for rapid and quantitative detection of cytokeratin 19 (CK19) mRNA with the reverse transcription loop-mediated isothermal amplification (RT-LAMP) method (11), has been shown to feature high specificity with a low false-positive rate (12). To assess the validity of this assay in clinical use, a multicenter clinical trial was conducted in Japan, concurrently with some single-institute studies in the Netherlands and Germany (13).

In this article, we report the results of the Japanese trial of the OSNA assay for detection of lymph node metastasis in breast cancer and discuss its validity for clinical use, especially for intraoperative SLN examination.

The two trials. The materials consisting of axillary lymph nodes were obtained from patients who underwent surgery for breast cancer between October 2005 and May 2006 at seven Japanese institutes and hospitals that had joined this study. Two of these institutes participated in trial 1, three in trial 2, and two in both. Patients were given the necessary information about the trial, and only the lymph nodes from patients who had given their consent were used. The trial consisted of two different protocols. In trial 1, designed to determine the specificity of the OSNA assay for detection of metastasis in comparison with that of detailed pathologic examination, clinically metastasis-negative (N0) lymph nodes with a maximum size between 4 and 8 mm were examined. Sampled lymph nodes were immediately divided into four pieces (a, b, c, and d) of 1 or 2 mm thickness with cutting devices developed by Tsujimoto et al. (12), and two of the four pieces (a and c) were examined with the OSNA assay in the laboratory of the participating institute concerned (Fig. 1). The two adjacent pieces (b and d) were sent to one of the three pathologists of the central committee, who examined them in the following manner. After the samples were fixed with formalin and embedded in paraffin, the two pieces were sliced sequentially at 0.2-mm intervals and a pair of sections with a thickness of 5 μm each was obtained from each level of slice. One of the paired sections was then stained with H&E, and the other was examined immunohistochemically for CK19 by using monoclonal antibody RCK108 (Dako). The pathologists examined the preps without access to information about the results of the OSNA assay. The pathologic diagnosis was negative when the specimen contained no tumor cells or only isolated tumor cells (ITC) and positive when micrometastasis or macrometastasis was found according to the criteria of the sixth edition of the tumor-node-metastasis classification of the International Union Against Cancer and the classification of the American Joint Committee on Cancer.

In trial 2, designed to determine the accuracy of the OSNA assay for detection of metastasis compared with that of routine pathologic examination, randomly sampled lymph nodes or SLNs with a maximum size between 4 and 8 mm were analyzed. The sampled lymph nodes were immediately divided into four pieces in the same manner as described above. Two pieces were examined with the OSNA assay, and the other two adjacent pieces were examined pathologically. Frozen sections of the SLNs were prepared and examined by a pathologist at the institute concerned. The remains of the SLN specimens and intact non-SLN specimens were sent to the central committee where three pairs of sections were obtained at the cut surface of each piece (Fig. 1) and examined with H&E and immunohistochemistry for CK19 in the same manner as in trial 1.

The OSNA assay. The OSNA assay for lymph nodes has been described in detail in a previous report (12). Briefly, pieces obtained from axillary lymph nodes were homogenized with 4 mL of a lysis buffer solution and centrifuged at 10,000 × g at room temperature. Two microliters of the supernatant were analyzed with the RD-100i system (Sysmex), an automated molecular detection system using a RT-LAMP method. A standard positive control sample containing 5 × 10³ copies/μL of CK19 mRNA and a negative control sample containing 0 copy/μL of CK19 mRNA were used for calibration in every assay. The results of the assay were expressed as the numbers of CK19 mRNA copies per microliter, and metastasis was assessed in accordance with the cutoff level determined by Tsujimoto et al. (12). That is, the lymph node was assessed negative when there were less than 2.5 × 10² copies/μL of CK19 mRNA and positive when there were 2.5 × 10² copies/μL or more.

Further examination for cases showing discrepancies between the OSNA assay and pathologic examination results. In trial 2, several nodes showed discrepant results for the OSNA assay and pathologic examination. After the trial period, such lymph nodes were subjected to further examination to determine the existence and localization of metastatic cells. For this purpose, the remaining pathologic specimen blocks were sectioned at 0.2-mm intervals and examined with H&E and immunohistochemistry for CK19 in the same manner as in trial 1. Furthermore, the lysate sample used for the OSNA assay was examined for CK19 protein expression by means of Western blotting analysis.

Statistical analysis. Sensitivity, specificity, and accuracy were determined by comparing the results of the OSNA assay and pathologic examination. The statistical program R 2.4.1¹⁸

Trial 1. A total of 149 axillary lymph nodes surgically obtained from 36 patients with early breast cancer (Table 1) constituted the materials for this study. Five nodes from one patient were excluded from the analysis because consent was withdrawn by the patient, as were 19 nodes in which at least one of the four pieces did not contain lymphatic tissue and one involving a technical error. The remaining 124 nodes from 34 N0 patients were then analyzed. Of the 104 nodes pathologically identified as negative, 101 were assessed as negative by the OSNA assay for a specificity of 97.1% (95% CI, 91.8-99.4%; Table 2). Of the three nodes that were pathologically identified as negative but assessed as positive in the OSNA assay, one was found to contain ITCs and another to have >0.3 ng/μL of CK19 protein in the remaining sample solution used for the OSNA assay. Of the 20 pathologically positive nodes, 19 were assessed as positive by the OSNA assay for a sensitivity of 95% (95% CI, 75.1-99.9%). One node assessed as negative by the OSNA assay was found to contain a micrometastasis.

Trial 2. A total of 551 axillary lymph nodes surgically obtained from 185 patients with early breast cancer (Table 1) constituted the materials of this study. Eight of the nodes from three patients were excluded because their consent was withdrawn. Twenty-six nodes from six patients who received neoadjuvant chemotherapy were also excluded, as well as 36 in which at least one of the four pieces of a node did not contain lymphatic tissue and 31 that did not meet the specifications of this study, such as the use of frozen materials for the assay. Of the 450 lymph nodes eligible for the analysis, 70 were assessed as positive and 348 as negative by both the OSNA assay and pathologic examination for an accuracy of 92.9% (95% CI, 90.1-95.1%; Table 3). Seventy of the 80 pathologically positive nodes were detected by the OSNA assay (sensitivity, 87.5%; 95% CI, 78.5-93.8%). On the other hand, 348 of the 370 (94.1%) pathologically negative nodes were assessed as negative in the OSNA assay, whereas 5.9% of the pathologically negative nodes were positive. These lymph nodes showing discrepant results were subjected to further analysis.

Further examination of discrepant cases from trial 2. The 10 nodes that were pathologically positive but negative in the OSNA assay (false negative) and the 22 nodes that were pathologically negative but positive in the OSNA assay (false positive) were subjected to further analysis. The results are summarized in Table 4. In eight of the 10 false-negative nodes, uneven localization of the tumor cells was found in the remnants of the nodes. However, two nodes with pathologically identified macrometastasis remained negative in the OSNA assay because tumor cells in these nodes showed faint expression of CK19, as was confirmed by immunohistochemical staining for CK19.

In 5 of the 22 false-positive nodes, some foci of tumor cells were found in the remnants of the nodes. Another eight of these nodes were not found to contain tumor cells in the pieces remaining after the pathologic examination, but lymphatic vascular invasions were detected in the main tumors of these nodes, whereas the lysate of two of them preserved for the OSNA assay contained a significant amount of CK19 protein. However, further analysis of the remaining nine nodes showed no pathologic or clinical signs of metastasis after additional sectioning.

The results based on the further analysis are shown in Table 5. The final accuracy of the OSNA assay based on the results of further examination was 93.1% (95% CI, 90.0-95.5%), the final sensitivity was 87.7% (95% CI, 78.5-93.9%), the final specificity was 94.3% (95% CI, 95.3-98.8%), the final positive predictive value was 77.2% (95% CI, 67.2-85.3%), and the final negative predictive value was 97.2% (95% CI, 95.5-98.9%). The OSNA assay detected 94.1% of the lymph node metastases larger than 2 mm (macrometastasis).

Because several studies have shown the feasibility of molecular detection of micrometastasis in the lymph nodes using reverse transcription-PCR (14–17), a lot of markers, such as CK19, mammaglobin, carcinoembryonic acid, MUC1, prolactin-induced protein, have been examined for their sensitivity and specificity for detection of metastasis, and a variety of combinations of these markers have been proposed for clinical use (18–23). One of these, a combination of CK19 and mammaglobin, showed both high sensitivity and specificity for detection of lymph node metastasis of breast cancer (23), and the system using these two markers showed high reliability and is now being used clinically (7–10).

The OSNA assay is also a molecular-based metastasis detection system, which uses CK19 as a single marker. CK19, a representative epithelial marker widely expressed in human cancers, is considered to be a promising marker with high sensitivity for detection of lymph node metastasis from various cancers. Reverse transcription-PCR for CK19, on the other hand, is sometimes unreliable because of the presence of pseudogenes (24) and contamination of benign epithelial cells (25). To overcome this problem, the OSNA assay adopted the RT-LAMP method developed by Notomi et al. (11). The amplification is processed isothermally by means of six primers and can detect mRNA of CK19 quantitatively without interference by pseudogenes. The assay can differentiate contamination of a few benign epithelial cells and the presence of ITCs from clinically significant tumor metastasis by using a verified cutoff value (12). The assay can be done with on-the-spot preparation and easy operation because homogenization of a lymph node in the lysis buffer takes only 90 seconds and centrifugation of the sample 1 minute, whereas placement of the supernatant and reagents into the detector does not require extraction and purification of mRNA to synthesize cDNA, both of which are necessary for the reverse transcription-PCR method. The mRNA is automatically amplified in a RD-100i gene amplification detector (Sysmex) in 16 minutes, and stable results are provided without being affected by the size of the sample (maximum of 600 mg for one assay; ref. 12).

Trial 1 was designed to determine its specificity because high specificity is needed to avoid unnecessary axillary dissection when the assay is used for SLN examination. The specificity of the assay in our study was 97.1% (95% CI, 91.8-99.4%), with only three nodes judged positive in the OSNA assay but negative pathologically, but two of these nodes were found to contain ITCs in the specimen used for pathology. The specificity of the OSNA assay thus reaches 99.0% when these cases with ITCs are excluded.

In trial 2, the accuracy of the assay was examined in comparison with a routine method for pathologic examination for SLNs used in Japan, which is similar to a protocol with three sections from each 2- to 3-mm slice of the node recommended by the consensus meeting (26). The rate of concordance between the assay and the pathologic examination was 92.9% (418 of 450; 95% CI, 90.1-95.1%), and the results indicated that the lymph node metastasis detection capability of the OSNA assay was statistically equal to that of the pathologic examination for three sections cut from slices obtained at 2-mm intervals when discrepancies caused by differences in the samples used for the assay and the pathologic examination are taken into consideration. This is so because a sample for molecular examination needs to be homogenized, it cannot be used for pathologic examination, so that studies comparing the two modalities using different pieces of a sample must therefore of necessity include some cases producing discrepant results caused by uneven localization of tumor foci. Actually, 32 nodes (7.1%) with discrepant results were found. In eight of the 10 false-negative nodes, only a few of the serial sections contained metastasized foci of the tumor. These small metastases might have been detected by the OSNA assay if the whole lymph node had been used for the assay. In addition, some tumor cell clusters or ITCs were found in the remaining specimens of 5 of the 22 false-positive nodes after additional sectioning, which had not been detected by routine pathologic examination using 2-mm interval sections. In another eight false-positive nodes, no tumor cells were found in the pieces remaining after the pathologic examination, but lymphatic vascular invasions were observed in the main tumors of these nodes. In addition, the lysate of two of them preserved for the OSNA assay contained a significant amount of CK19 protein. These nodes may thus have harbored some foci of tumor cells in the piece used for the OSNA assay.

On the other hand, two nodes from different patients accounted for the false-negative cases with a very weak expression of CK19 mRNA. The primary tumors of the patients also showed negative staining for CK19 as confirmed by immunohistochemistry. The actual incidence of tumors with low CK19 expression remains unclear. Bartek et al. (27) reported an incidence for breast cancer of 0%, but Parikh et al. (28) of 20.5% for young patients. The OSNA assay is more sensitive than immunohistochemistry so that the incidence of false-negative results caused by low expression of CK19 can be expected to become exceptional in clinical use. In fact, another lymph node obtained from one of the two patients was positive in the OSNA assay, so that only 1 of the 185 patients (0.5%) was identified as negative by the assay. However, this ratio should be confirmed with more lymph nodes and patients.

The findings of the further analysis of discrepant cases in trial 2 prompted a reanalysis of the data, resulting in the final specificity of the OSNA assay compared with the pathologic examination becoming 94.3%, the final accuracy 93.1%, and the final negative predictive value 97.2%. These ratios indicate that the OSNA assay can accurately detect node-negative cases. As for pathologically positive lymph nodes (diameter of metastasis, >0.2 mm), 87.7% could be detected by the OSNA assay, whereas 94.1% of macrometastases (>2 mm) were assessed as positive. Our results were similar to recently reported results reported by Visser et al. (13), who compared the OSNA assay with pathologic examination at five levels at 0.25-mm intervals of each piece used for pathology.

In conclusion, the OSNA assay showed high specificity, accuracy, and negative predictive value compared with conventional pathologic examination for the detection of lymph node metastasis of breast cancer. It provides satisfactory results in a short time and with easy procedure. The OSNA assay can thus be used as an alternative tool for examining metastasis in SLNs. However, for the time being, it is recommended to use the assay together with pathologic examination for minimal numbers of specimens until additional clinical trials with more lymph nodes and patients show that prognostic outcomes determined by the assay are equal or superior to those determined by conventional pathology.

Y. Tamaki, F. Akiyama, T. Kaneko, K. Tsugawa, M. Tsujimoto, and N. Matsuura: honorarium, Sysmex Corp. S. Noguchi and N. Matsuura: Advisory Board, Sysmex Corp. M. Tsujimoto, T. Kaneko, and N. Matsuura: travel grant, Sysmex Corp.

We thank all the staff of the clinical and pathologic laboratories at the participating institutes for their technical support.