Frozen section diagnosis for non-melanoma skin cancers: correlation with permanent section diagnosis

Multi-site cross-organ calibrated deep learning (MuSClD): Automated diagnosis of non-melanoma skin cancer

Although deep learning (DL) has demonstrated impressive diagnostic performance for a variety of computational pathology tasks, this performance often markedly deteriorates on whole slide images (WSI) generated at external test sites. This phenomenon is due in part to domain shift, wherein differences in test-site pre-analytical variables (e.g., slide scanner, staining procedure) result in WSI with notably different visual presentations compared to training data. To ameliorate pre-analytic variances, approaches such as CycleGAN can be used to calibrate visual properties of images between sites, with the intent of improving DL classifier generalizability. In this work, we present a new approach termed Multi-Site Cross-Organ Calibration based Deep Learning (MuSClD) that employs WSIs of an off-target organ for calibration created at the same site as the on-target organ, based off the assumption that cross-organ slides are subjected to a common set of pre-analytical sources of variance. We demonstrate that by using an off-target organ from the test site to calibrate training data, the domain shift between training and testing data can be mitigated. Importantly, this strategy uniquely guards against potential data leakage introduced during calibration, wherein information only available in the testing data is imparted on the training data. We evaluate MuSClD in the context of the automated diagnosis of non-melanoma skin cancer (NMSC). Specifically, we evaluated MuSClD for identifying and distinguishing (a) basal cell carcinoma (BCC), (b) in-situ squamous cell carcinomas (SCC-In Situ), and (c) invasive squamous cell carcinomas (SCC-Invasive), using an Australian (training, n = 85) and a Swiss (held-out testing, n = 352) cohort. Our experiments reveal that MuSCID reduces the Wasserstein distances between sites in terms of color, contrast, and brightness metrics, without imparting noticeable artifacts to training data. The NMSC-subtyping performance is statistically improved as a result of MuSCID in terms of one-vs. rest AUC: BCC (0.92 vs 0.87, p = 0.01), SCC-In Situ (0.87 vs 0.73, p = 0.15) and SCC-Invasive (0.92 vs 0.82, p = 1e-5). Compared to baseline NMSC-subtyping with no calibration, the internal validation results of MuSClD (BCC (0.98), SCC-In Situ (0.92), and SCC-Invasive (0.97)) suggest that while domain shift indeed degrades classification performance, our on-target calibration using off-target tissue can safely compensate for pre-analytical variabilities, while improving the robustness of the model.

Histological Stains in the Past, Present, and Future

Certain contemporary histology stains and methods are not the same as those used in the past. This progression has delved into the requirement for more precise, less complex, and efficient staining procedures. The objective of this study is to assess historical and contemporary stains and procedures, as well as the challenges surrounding their improvement. Carmine, hematoxylin, silver nitrate, Giemsa, trichome stain, Gram stain, and mauveine were among the first histological stains discovered in nature. Aside from their utility in the study of tissues at the time, they also laid the groundwork for the development of commercial dyes that are still in use today. Hematoxylin and eosin, Ziehl-Nielsen (ZN) stain, periodic acid-Schiff stain, and Grocott-Gomori methenamine silver stain are some of the most recently developed histological stains. The future of histological stains and processes appears to be influenced by technological advancements and the demand for cost-effective diagnostic approaches in the healthcare system. Thus, currently used histological stains appear to be economical, quick, and reliable tools for interpreting, archiving, and delivering essential diagnoses that could not be achieved by any other means.

Diagnostic Biopsy via In-Office Frozen Sections for Clinical Nonmelanoma Skin Cancer

BACKGROUND

Treatment of nonmelanoma skin cancer (NMSC) by Mohs surgery has traditionally relied on previous pathologic evaluation of paraffin-embedded tissue. Tissue processing by frozen sections allows for expedited diagnosis and treatment; however, data on its accuracy are limited.

OBJECTIVE

To measure the accuracy and outcomes of biopsy via frozen sections for clinical NMSC.

METHODS

Biopsies of clinical NMSCs processed via frozen sections with in-office diagnosis rendered by one Mohs surgeon were retrospectively reviewed by one board-certified dermatopathologist. Discordant diagnoses were re-read in blinded fashion by both physicians. If still discordant, final diagnosis was determined by consensus discussion. Inter-rater reliability was calculated using Cohen's kappa statistic.

RESULTS

Two hundred ninety-seven lesions from 208 patients were included. Correlation between in-office and final diagnosis was 0.876 indicating "almost perfect" concordance. Sensitivity and specificity of in-office diagnosis for detecting malignancy were 98.1% and 94.4%. Seven cases (2.0%) had a clinically relevant change in final diagnosis, but appropriate treatment had been rendered. Two benign lesions (0.7%) initially diagnosed as malignant underwent excision.

CONCLUSION

In-office biopsy via frozen sections is highly accurate in confirming NMSC. This practice may speed diagnosis and treatment thus improving outcomes and patient satisfaction.

Muffin Technique Micrographic Surgery for Non-melanoma Skin Cancer

Background: Mohs micrographic surgery (MMS) is the gold standard treatment for high-risk facial non-melanoma skin cancer. However, patients' access to MMS is limited by cost. The muffin technique micrographic surgery (MTMS) is an alternative micrographic technique wherein the entire excised margin is evaluated post-operatively by a pathologist using paraffin-embedded material. Herein, we describe the implementation and the preliminary results of MTMS in an academic dermatology center.

Objective: To describe the MTMS and outline its efficacy and safety in a real-world clinical academic setting.

Methods: A retrospective chart review was conducted of all patients with basal cell carcinoma (BCC) and squamous cell carcinoma (SCC) who underwent MTMS at the University of Alberta Dermatology Center from June 2016 until July 2019.

Results: A total of 69 patients were included (64 BCCs and 5 SCCs). 68.1% of surgeries had clear margins following the first incision, 100% after second round re-excisions. There were no observed cases of tumor recurrence after a median 40 months of follow-up. There were no major adverse events or complications.

Conclusions: MTMS is a superior alternative to simple excision of skin cancer by providing full margin control and residual tumor mapping.

Sensitivity and specificity of frozen section diagnosis in orbital and adnexal malignancies

Purpose:

To analyze the diagnostic accuracy of frozen section in orbital and adnexal malignancies.

Methods:

A total of 55 cases between January 2006 and December 2011 for which intraoperative frozen section was performed for various orbital and adnexal lesions were included in the study. The frozen section diagnosis was compared with the permanent section diagnosis. Margin clearance was also compared between the two. Data were analysed using SPSS version 14. Odds ratio and cross-tabulation was used to perform the analysis.

Results:

The mean age at presentation was 51.46 ± 20 years. Eyelid was the most common site of involvement. Out of 55 cases, diagnosis was deferred in four cases (7.27%) on frozen section. Among 51 cases, 44 (86%) cases were concordant, whereas 7 (13%) cases were discordant. The sensitivity and specificity of frozen section compared to permanent section for diagnosis of malignancy was found to be 87.2% and 87.5%, respectively. The sensitivity and specificity of frozen section for diagnoses of basal cell carcinoma was found to be 100%, while it was 83.3% and 100% respectively for sebaceous gland carcinoma and 87.5% and 94.9% respectively for squamous cell carcinoma.

Conclusion:

Frozen section had high sensitivity and specificity when compared with permanent section for all three parameters studied. It is an important intraoperative tool that is increasingly being used in histopathological examination of ophthalmic lesions. However, it should not be used as a substitute for the permanent section and critical decisions based on it are best avoided.

Frozen Section Evaluation of Margin Status in Primary Squamous Cell Carcinomas of the Head and Neck: A Correlation Study of Frozen Section and Final Diagnoses

Frozen section is routinely used for intraoperative margin evaluation in carcinomas of the head and neck. We studied a series of frozen sections performed for margin status of head and neck tumors to determine diagnostic accuracy. All frozen sections for margin control of squamous carcinomas of the head and neck were studied from a 66 month period. Frozen and permanent section diagnoses were classified as negative or malignant. Correlation of diagnoses was performed to determine accuracy. One thousand seven hundred and ninety-six pairs of frozen section and corresponding permanent section diagnoses were obtained. Discordances were found in 55 (3.1%) pairs. In 35 pairs (1.9%), frozen section was reported as benign, but permanent sections disclosed carcinoma. In 21 cases, the discrepancy was due to sampling and in the remaining cases it was an interpretive error. In 20 cases (1.1%), frozen section was malignant, but the permanent section was interpreted as negative. Frozen section is an accurate method for evaluation of operative margins for head and neck carcinomas with concordance between frozen and permanent results of 97%. Most errors are false negative results with the majority of these being due to sampling issues.