Interobserver variability of laryngeal mucosal premalignant lesions: a histopathological evaluation.

Department of Otorhinolaryngology and Head and Neck Surgery, Radboud University Nijmegen Medical Center, Nijmegen, The Netherlands.
Modern Pathology (Impact Factor: 6.36). 04/2011; 24(7):892-8. DOI: 10.1038/modpathol.2011.50
Source: PubMed

ABSTRACT The objective of this study is to measure interobserver variability in the classification of laryngeal mucosal premalignant lesions by reassessing the histopathology of previously diagnosed cases and to determine the possible therapeutic consequences of disagreement among observers. Histopathological assessment of 110 laryngeal mucosal premalignant lesions was done by three pathologists. Each slide had to be classified according to the World Health Organization, Squamous Intraepithelial Neoplasia, and the Ljubljana Squamous Intraepithelial Lesions systems. After the independent assessment, a joint meeting took place. To assess the relation between histopathological grading and subsequent clinical management, we created a two- and a three-grade system besides one comprising all options. For all analyses, the SAS/STAT statistical software was used. The highest unweighted κ-values concerning the all-options system are observed for the Squamous Intraepithelial Neoplasia classification (0.28, 95% confidence interval 0.23-0.33), followed by the World Health Organization and Ljubljana classifications. For the two-grade system the Ljubljana classification shows the highest unweighted κ-values (0.50, 95%, 0.39-0.61), followed by the World Health Organization and Squamous Intraepithelial Neoplasia classifications. For the three-grade system, the unweighted κ-values are similar. The implementation of weighted κ-values led to higher scores within all three classification systems, although these did not exceed 0.55 (moderate agreement). Given the high level of consensus, simultaneous pathological assessment may be said to provide added value in comparison with independent assessment. In the current study, no clear tendency is observed in favor of any one classification system. The proposed three-grade system could be an improved histopathological tool because it is easier to correlate with clinical decision making and because it yields better unweighted κ-values and proportions of concordance than the all-options system. Furthermore, clinical management could benefit from assessment by more than one pathologist in suspected cases of dysplasia or carcinoma.

  • [Show abstract] [Hide abstract]
    ABSTRACT: To verify the applicability, reproducibility and predictive value of a proposed unified classification (amended Ljubljana classification) of laryngeal squamous intraepithelial lesions (SILs). Six internationally recognised experts and three pathologists from Ljubljana contributed to this study by evaluating a set of laryngeal SILs using the new system: low-gradeSIL, high-grade SIL and carcinoma in situ (CIS). The overall agreement among reviewers was good. Overall unweighted and weighted κ-statistics and 95% confidence intervals were 0.75 (0.65 - 0.84) and 0.80 (0.71 - 0.87), respectively. The results were stratified between the international reviewers and the Ljubljana pathologists. The former had a good overall agreement, while the latter had very good agreement. Kaplan-Meier survival curves showed a significant difference (p<0.0001) between patients with low- and high-grade SILs; 19/1204 patients with low-grade SILs and 30/240 patients with high-grade SILs progressed to malignancy in 2 to 15 years and in 2 to 26 years, respectively. The proposed modification to the LC classification provides clear morphological criteria to define the prognostic groups. The criteria facilitate better inter-observer agreement than previous systems and the retrospective follow-up study demonstrates a highly significant difference in the risk of malignant progression between low and high grade SILs. This article is protected by copyright. All rights reserved.
    Histopathology 04/2014; · 3.30 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The histopathology of premalignant laryngeal lesions does not provide reliable information on the risk of malignant transformation, hence we examined new molecular markers which can easily be implemented in clinical practice.Dual-target fluorescence in situ hybridisation (FISH) for chromosome 1 and 7 centromeres was performed on tissue sections of laryngeal premalignancies in 69 patients. Chromosome instability was indicated by numerical imbalances and/or polysomy for chromosomes 1 and 7. Additionally, immunostainings for p53, Cyclin D1 and (p)FADD expression were evaluated. Malignant progression was recorded. Eighteen patients with carcinoma in situ (CIS) were treated after diagnosis and excluded from follow-up.Chromosome instability was strongly associated with a high risk of malignant transformation, especially in lower grade lesions (hyperplasia, mild and moderate dysplasia; odds ratio = 8.4, p = 0.004). Patients with lesions containing chromosome instability showed a significantly worse 5-year progression-free survival than those with premalignancies without chromosome instability (p = 0.002). Neither histopathology nor the protein markers predicted progression in univariate analysis, although histopathological diagnosis, p53 and FADD contributed positively to chromosome instability in multivariate analysis.Chromosome instability is associated with malignant progression of laryngeal premalignancies, especially in lower grade lesions. These results may contribute to better risk counselling, provided that they can be validated in a larger patient set.
    Pathology 03/2014; · 2.62 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: A better understanding of the reasons for diagnostic variability may help reduce the phenomenon. In preparation for a study on interpretation of breast specimens (B-Path), a panel of three experienced breast pathologists reviewed 336 cases to develop consensus reference diagnoses. After independent assessment, cases coded as diagnostic discordance were discussed at consensus meetings. Using qualitative data analysis techniques, transcripts of 16 hours of consensus meetings for a subset of 201 cases were analyzed. The reasons for diagnostic variability could be attributed to three overall root causes: 1) pathologist-related, 2) diagnostic coding/study methodology-related and 3) specimen-related. Most pathologist-related root causes were due to professional differences in pathologists' opinions about whether the diagnostic criteria for a specific diagnosis were met, most frequently in cases of atypia. Diagnostic coding/study methodology root causes were primarily miscategorizations of descriptive text diagnoses, which led to development of a standardized electronic diagnostic form (BPATH-Dx). Specimen-related causes included artifacts, limited diagnostic material and poor slide quality. After re-review and discussion, a consensus diagnosis could be assigned in all cases. In conclusion, diagnostic variability is related to multiple factors, but consensus conferences, standardized electronic reporting formats and comments on suboptimal specimen quality can be used to reduce diagnostic variability. This article is protected by copyright. All rights reserved.
    Histopathology 02/2014; · 3.30 Impact Factor