RESUMO
The utility of prognostic and predictive immunohistochemistry biomarkers in the context of cancer is plagued by inconsistent interpretation of results which can lead to poor rates of adoption or inappropriate use of novel therapeutic strategies. To monitor immunohistochemistry assay performance, a new on-slide control motif, Immunohistochemistry Critical Assay Performance Controls (ICAPC) was developed. We hypothesized that the use of these controls by the diagnosing pathologist to interpret BRAFV600E would result in reduced interobserver and intraobserver interpretation errors. A cross-sectional, sequentially obtained sample of surgical pathology cases stained for BRAFV600E was assembled from a single hospital in Vancouver, British Columbia. Half of the cases had normal on-slide controls and the remainder with ICAPC. Results from 6 independent and blinded readers were compared with each other and to the gold-standard pathologic diagnosis with the goal of demonstrating superior interrater agreement with ICAPC relative to standard on-slide controls. Cohen's κ was used to compute pair-wise reader agreements, whereas Fleiss' κ was used to compare to the gold standard. The implementation of ICAPC resulted in statistically significant improvements in the interobserver agreement of BRAF mutation status ascertained by BRAFV600E immunohistochemistry. Half of the readers demonstrated significant improvements in agreement with the gold-standard diagnosis with the addition of ICAPC. Across all readers, the mean increase in κ was 0.14 with a 95% confidence interval of 0.01-0.28 (P=0.04). This study demonstrates that the addition of ICAPC serves to significantly reduce interobserver variability in the assessment of BRAFV600E immunohistochemistry. As such, we recommend that this approach should be used as part of a comprehensive quality management strategy in the setting of histopathology.
Assuntos
Imuno-Histoquímica/métodos , Neoplasias/diagnóstico , Proteínas Proto-Oncogênicas B-raf/metabolismo , Neoplasias Colorretais/diagnóstico , Estudos Transversais , Humanos , Melanoma/diagnóstico , Variações Dependentes do Observador , Proteínas Proto-Oncogênicas B-raf/genética , Reprodutibilidade dos Testes , Neoplasias da Glândula Tireoide/diagnósticoRESUMO
BACKGROUND: Ampullary carcinomata (AC) can be separated into intestinal (IT) or pancreatobiliary (PB) subtypes. Although morphological, immunohistochemical and molecular differentiation of IT and PB have been well documented; the prognostic significance of histological subtype and whether patients with either subtype benefit from differential chemotherapeutic regimens remains unclear. METHODS: As part of a larger cohort study, patients who underwent resection for AC or pancreatic ductal adenocarcinoma (PDAC) were retrospectively identified. Clinicopathological covariates and outcome were obtained and MUC1, MUC2, CDX2 and CK20 were assessed with immunohistochemistry. RESULTS: Of 99 ACs, the resultant immunophenotypes indicated 48% and 22% were IT and PB, respectively. Thirty (30%) cases were quadruple negative (QN). Within the PDAC cohort (N = 257), the most prevalent immunophenotype was QN (53%). Subsequently, all QN ACs were classified as PB immunohistochemically yielding 47.5% and 52.5% classified as IT and PB, respectively. Involved regional lymph nodes and elevated T-stage were significantly associated with PB compared with IT AC (p = 0.0032 and 0.0396, respectively). Progression-free survival revealed inferior survival for PB versus IT AC (p = 0.0156). CONCLUSIONS: AC can be classified into prognostic groups with unique clinicopathological characteristics using immunohistochemistry. Immunophenotypical similarity of PB and PDAC suggests that treatment regimens similar to those used in PDAC should be explored.