Accuracy of the IASLC/ATS/ERS histological subtyping of stage I lung adenocarcinoma on intraoperative frozen sections

Abstract

Histological subtyping of surgically resected lung adenocarcinoma has been shown to be of prognostic significance, and limited surgical resection has been proposed as a treatment of choice for early-stage lung adenocarcinoma. The accuracy of histological subtyping has been recently assessed in the surgical resection and small biopsy specimens; however, the accuracy of intraoperative subtyping on frozen sections remains relatively unknown. The aim of this study was to determine diagnostic accuracy and interobserver variability in histological subtyping of lung adenocarcinoma on intraoperative frozen sections. Overall, 112 consecutive cases of surgically resected stage I lung adenocarcinoma were reviewed independently by three pathologists. Histological patterns (acinar, lepidic, papillary, micropapillary, and solid) and mucinous variant were recorded in 5% increments for each intraoperative frozen and permanent sections. Primary and secondary histological patterns were assigned in each case. Kappa scores were calculated to evaluate agreement between pathologists in the assessment of histological subtype on intraoperative frozen sections versus permanent sections. Overall agreement between intraoperative frozen and permanent sections was moderate for primary pattern (69.7% of cases), with kappa scores ranging from 0.43 to 0.58, with more consistent agreement for stage IA tumors. Kappa scores for the secondary pattern ranged from 0.16 to 0.32. Acinar and solid patterns were most likely to be correctly identified as primary growth patterns. Micropapillary pattern was recognized in only 11-55% of cases. The main reasons for discrepancies between intraoperative frozen and permanent sections were inadequate sampling and poor quality of frozen sections. Our study suggests that it is difficult to predict the primary adenocarcinoma pattern on a single representative frozen section. This observation suggests a potential impact on the extent of frozen section sampling by pathologists at the time of intraoperative consultation, if surgical management of stage I lung adenocarcinoma will be guided by its histological subtype.

Relationship of epidermal growth factor receptor activating mutations with histologic subtyping according to International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society 2011 adenocarcinoma classification and their impact on overall survival

“…showed discrepancies between patterns in intraoperative frozen and permanent sections and attributed it to inadequate sampling and poor quality of frozen sections. [ 42 ] Moreover, there are no publications which have correlated biopsy patterns with patterns in the resected specimens.…”

Clinical impacts of a micropapillary pattern in lung adenocarcinoma: a review

“… 102 Trejo et al have also reported a lower consistency between intraoperative frozen sections and permanent sections in MPA compared with other histological patterns in stage I lung ADC. 103 Therefore, at present, permanent sections, as a postoperative diagnosis, are the only reliable method to identify the presence of MPP, rather than preoperative imaging, small biopsies, and cytology or frozen sections during an operation. 31 , 104 …”

Challenges for real‐time intraoperative diagnosis of high risk histology in lung adenocarcinoma: A necessity for sublobar resection

“…On the other hand, precise histologic subtyping, in particular detection of high grade histology is generally challenging to date. Errors in sampling and/or interpretation have been reported as major causes of frozen section misdiagnosis as lung ADC usually consists of multiple histologic subtypes . In addition, even a small amount of high‐grade histology should be detected because it contributes to an unfavorable prognosis after sublobar resection as above .…”

“…On frozen section slides, alveolar spaces are frequently collapsed, which can cause difficulty in evaluating invasion . Bittar et al . reported that the concordance rate was unsatisfactory mainly due to sampling errors and poor frozen section quality.…”