Carcinomas from either pulmonary or extrapulmonary sites can metastatise to supraclavicular lymph nodes. Immunohistochemistry (IHC) is invaluable to comment on the possible primary site. However, the optimal number of antibodies to be tested is debatable. Seven antibodies were tested on 135 metastatic supraclavicular lymph node biopsies to propose a "best minimal" IHC panel to infer lung carcinoma, incorporating the principles of "evidence-based medicine." The 135 cases were divided into the following: category I (110 cases), wherein the primary was in the lung based on histologic analysis (Ia, n = 14 [12.7%]), cytologic analysis (Ib, n = 43 [39.1%]), or strong clinicoradiologic evidence (Ic, n = 53 [48.2%]), and category II (25 cases) with a histologically proven extrapulmonary primary site. Categories Ia and Ib were together designated as the "control group," and category Ic was designated as the "test group." The antibodies tested were cytokeratin (CK 7, CK20), epithelial membrane antigen, carcinoembryonic antigen, thyroid transcription factor-1 (TTF-1), surfactant protein B (SPB), and vimentin. Results of both individual and panels of antibodies were statistically evaluated. The sensitivity and specificity of single antibodies for inferring a lung primary was as follows: CK7 (90%/56%), CK20 (98%/40%), epithelial membrane antigen (90.9%/4%), carcinoembryonic antigen (80.9%/36%), TTF-1 (62.7%/100%), SPB (65.6%/100%), and vimentin (60.9%/60%). The highest sensitivity (85%) and specificity (100%) were seen with a 4-antibody panel: CK7, CK20, TTF-1, and SPB. This panel revealed the highest binomial probability (.8), for diagnosing lung cancer. The results were validated using a "split sample method," and a high concordance was noted between the control and test groups. To conclude, such evidence-based validated studies analyzing IHC results would be invaluable to guide the practice of surgical pathology in the future.
Keywords: Evidence-based medicine; Lung; Metastasis; Minimal IHC panel; Supraclavicular lymph node.
Copyright © 2014 Elsevier Inc. All rights reserved.