A validation study of a new molecular diagnostic assay: The Dartmouth-Hitchcock Medical Center experience with the GeneSearch™ BLN assay in breast sentinel lymph nodes

Abstract

Background

Sentinel lymph node (SLN) processing remains variable in terms of performing multiple tissue levels and immunohistochemical (IHC) or PCR-based assays. A rapid and reliable molecular pathology assay, as an adjunct to routine SLN processing, could minimize and standardize the histologic evaluation needed for an accurate and clinically significant diagnosis. We compared the recently FDA-approved Veridex GeneSearch™ Breast Lymph Node (BLN) Assay (Veridex, LLC; Warren, NJ), a real time reverse transcriptase-polymerase chain reaction assay that is designed to detect metastases > 0.2 mm, with our standard lymph node processing.

Materials and methods

The GeneSearch™ BLN assay evaluates RNA expression data for three target genes (mammaglobin, cytokeratin 19, and internal control porphobilinogen deaminase), and provides a qualitative (positive/negative) result. In 59 patients, the assay was performed on SLN tissue that would normally be deep within the tissue block and not routinely evaluated histologically. Two 1 -mm slices from the outer node portions were submitted fresh for RNA extraction; the remaining tissue was submitted for routine histology.

Results

Of the 59 patients, the assay determined 43 as true negative, eight as true positive, one as false-negative, three as false-positive, and four as invalid. Assay sensitivity was 88.9%, specificity 93.5%.

Discussion

The sensitivity of the assay sampling from the outer node tissue was high (88.9%) and identical to that validated in the large registration study in which half of the node was assessed as alternate slices (87.6%). Our protocol uses this assay as an adjunct to traditional histologic evaluation, to reduce and standardize the number of tissue sections needed for thorough SLN evaluation, and to enhance our ability to bank RNA.

Introduction

While the need for widespread molecular diagnostic testing in medicine today is not in doubt, the validation and implementation of such testing remains complex. In an era of cost and reimbursement constraint, it is important to consider how cheaper, clinically relevant molecular testing can replace established tests rather than be developed in addition to a current traditional test. This is only possible after a thorough validation of the molecular test against the current gold standard with some adaptations for unique laboratory workflow and staffing. This manuscript details our experience validating the GeneSearch™ BLN Assay in Breast Sentinel Lymph Nodes.

Sentinel lymph node (SLN) status is the most important prognostic indicator in patients with breast cancer. Despite recommendations from the College of American Pathologists (CAP) and the Association of Directors of Anatomic and Surgical Pathology (ADASP), the processing of SLN remains variable in terms of performing multiple tissue levels and immunohistochemical (IHC) or PCR-based assays (Fitzgibbons, 2000, Silverberg, 2001). The recently FDA-approved GeneSearch™ BLN Assay (Veridex, LLC, Warren, NJ) uses a reverse transcriptase real time PCR (RT²-PCR) assay to evaluate SLN for breast cancer metastases using cytokeratin 19 (CK 19) and mammaglobin in evaluation of 50% of the lymph node. Human mammaglobin has been shown to be over-expressed in breast cancer, and has breast-specific expression (Leygue, 2999, Ouellette et al., 2004, Watson, 1999, Zehentner and Carter, 2004). Studies assessing genetic markers for identifying clinically significant breast cancer metastases in SLNs showed that the optimal two-gene combination was mammaglobin and CK 19, with 90% sensitivity and 94% specificity (Backus et al., 2005). This assay has been validated to detect metastatic foci > 0.2 mm.

Pathologists often have reservations about validating new molecular diagnostic tests because the tissue required to perform the test may interfere with current standard of care tissue sectioning and interpretation. However, there are molecular diagnostic assays available that can be of great assistance to pathologists, and potentially lessen the workload and provide results in a more timely and cost effective manner.

In order to visualize how useful a molecular method could be to our practice, we reviewed our experience with SLN at Dartmouth-Hitchcock Medical Center during 2003 and 2004. SLN biopsy was performed in 80% (n = 222) of breast cancer patients at the time of the primary tumor excision (50%), mastectomy (21%), re-excision for positive margins (9%) or as a separate procedure (20%), usually preceding immediate breast reconstruction. Of the 222 SLN patients, 31% (n = 68) had one or more positive nodes. Of these 68 patients, 28% (n = 18) did not have completion axillary dissection. Of those that did, 50% (n = 25) required a separate surgical procedure. Only 21% (n = 14) of patients with positive SLN had subsequently positive axillary nodes. The majority of deposits (26%) were in the micrometastasis size range (> 0.02 to ≤ 0.2 cm) while 19% were > 0.2 to ≤ 0.5 cm, and 25% were > 0.5 to ≤ 1.0 cm. Only 3% (n = 2) of SLN showed CK-only positive cells without H&E confirmation and these patients were treated as node-negative. Nodal deposits were present in all tissue histology levels in 76% of cases. Extranodal extension was seen in 22% of nodes. Given the poor sensitivity and specificity of the technique, frozen sections on sentinel nodes are rarely performed at our institution. Identifying SLN metastasis with a molecular method could decrease the number of histologic sections and IHC needed to render an accurate diagnosis.

The GeneSearch™ Breast Lymph Node Assay instructions for use recommends serially sectioning SLN along the long axis into equal size and numbers of sections and submitting alternating sections for PCR and frozen sections, respectively. However, our pathologists wished to maintain our current histology sampling while still evaluating the assay, so we decided to examine the outer node portions that normally would be embedded deep within the block, unevaluated by histology. In order to validate our chosen alternative sampling of the SLN, we took tissue blocks from 10 positive SLNs, melted them down and after turning 180° re-embedded them so that tissue could be cut from the outer most portions of the nodes. We found that we were able to still find metastatic disease in this outer node tissue, even for metastasis deposit sizes as small as 0.1 mm (Table 1). We, therefore, concluded that this sampling would be a valid alternative to sampling half the node which includes the center of the node currently used for our routine tissue levels and cytokeratin immunostudies.

The aims of this study were two-fold: first, to determine the assay's specificity, sensitivity, positive predictive value, and negative predictive value in patient samples using an alternative sampling method compared to performance of the larger U.S. registration trial; and second, to demonstrate how to overcome pathologists' reservations to consider such an assay by validating if nodal tissue that is consistently left unobserved, within the paraffin tissue block, can be used to perform the BLN Assay.