Histologic Accuracy of Mohs Micrographic Surgery James T. Highsmith, MD, MS,*†‡x M. Jason Highsmith, PhD,k¶# and Gary D. Monheit, MD†‡
BACKGROUND Mohs micrographic surgery enables the surgeon to maximize tumor removal while minimizing the amount of tissue being removed using advanced mapping techniques combined with microscopy. Interpretation of frozen section slides is vital to the success of Mohs micrographic surgery. OBJECTIVE Evaluate and confirm the congruence of frozen section slide interpretation between fellowshiptrained Mohs surgeons and dermatopathologists. METHODS
Retrospective cohort concordance study spanning 10 years (n = 1,720 cases).
RESULTS The concordance rate for frozen slide interpretation between fellowship-trained Mohs surgeons and dermatopathologists in this study was 99.5%. CONCLUSION This study demonstrates agreement between the interpretation of histologic frozen section slides when evaluated by fellowship-trained Mohs surgeons and dermatopathologists. The authors have indicated no significant interest with commercial supporters.
A
nnually within the United States, nonmelanoma skin cancer (NMSC) results in 4.9M cases and approximately 11,000 deaths.1–3 There are many treatment options for NMSC. Mohs micrographic surgery (MMS) is the treatment of choice in recurrent skin cancers and primary skin cancers that demand tissue preservation.4,5 Mohs micrographic surgery fellowship is a subspecialty designed to train board-certified dermatologists through an intensive, mentored, 1- to 2-year training experience in anatomy, surgery, tissue mapping, reconstruction, accurate histologic margin interpretation of frozen sections, and postoperative wound management. Fellowship training in MMS incorporates immediate feedback with mentored error correction. These training elements improve technique and minimize interpretive histologic errors which are important factors affecting recurrence rates.6 Previous
studies on this topic yielded concordance rates of 94.9% to 99.7%.4,5,7 Therefore, the purpose of this study was to confirm previous studies by conducting the largest histologic frozen section slide concordance study to date between fellowship-trained Mohs surgeons and fellowship-trained dermatopathologists.
Methods Two fellowship-trained Mohs surgeons, 9 fellowshiptrained dermatopathologists, and 11 MMS fellows produced the data set over a 10-year period. All cases in this study involved either basal cell carcinoma (BCC) or squamous cell carcinoma (SCC). It is the authors’ standard practice that a tumor layer is removed initially just before the first Mohs surgical layer to debulk the tumor as opposed to curettage. This tumor debulk layer and initial Mohs layer are
*Department of Dermatology (111H), James A. Haley Veterans’ Administration Hospital, Tampa, Florida; †Callahan Eye Hospital, University of Alabama at Birmingham, Birmingham, Alabama; ‡Total Skin and Beauty Dermatology Center, Birmingham, Alabama; xDermatology Surgery Institute, Lutz, Florida; kExtremity Trauma & Amputation Center of Excellence (EACE), U.S. Department of Veterans Affairs, Washington, District of Columbia; ¶Morsani College of Medicine, University of South Florida, Tampa, Florida; #319th Minimal Care Detachment, U.S. Army Reserves, Pinellas Park, Florida © 2017 by the American Society for Dermatologic Surgery, Inc. Published by Wolters Kluwer Health, Inc. All rights reserved. ISSN: 1076-0512 Dermatol Surg 2017;0:1–4 DOI: 10.1097/DSS.0000000000001352
·
·
1
© 2017 by the American Society for Dermatologic Surgery, Inc. Published by Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
HISTOLOGIC ACCURACY OF MMS
processed simultaneously by trained histotechnicians into frozen sections and stained with hematoxylin and eosin. Histopathologic interpretation was performed by 1 of the 2 Mohs surgeons or 1 of the 11 Mohs fellows under the supervision of the Mohs surgeon who also functioned as the pathologist. The tumor layer is read before reviewing the first Mohs micrographic surgical layer to visualize any distinct histological tumor feature. The senior Mohs surgeon reviewed each slide of every stage in concert with systematic tissue mapping. Additional surgical excision layers were performed in the same manner when deemed appropriate.
into a data set using Microsoft Excel (Microsoft Corp., Redmond, WA). All discordant slides were obtained and rereviewed with 1 independent dermatopathologist at a different institution. This dermatopathologist completed residency training in clinical and anatomic pathology as well as a fellowship in dermatopathology. Documentation of the discordant reviews was recorded onto QC forms and kept in the MMS laboratory. All discordant patient records were reviewed. The outcomes of these reviews were recorded into the data set.
In the quality control (QC) program, the histotechnicians selected every tenth slide from Mohs cases each week and were not aware of the type of tumor present on the slide when submitting them for quality assurance. The slide submitted may be from the tumor debulking layer with tumor present or any of the MMS layers. The histotechnician selecting the slide recorded the documented findings of the Mohs surgeon as noted on the intraoperative Mohs map. Hence, the MMS surgeon was blinded as to which slides were reviewed in the QC program.
All procedures were conducted and data collected in accordance with the Declaration of Helsinki.
The selected slides were then sent for analysis to an independent dermatopathology private practice for review. After submission, the slide diagnoses were recorded by the dermatopathologists who were blinded to the MMS surgeon’s interpretation. The interpretation of each slide was recorded on a standardized QC form which was kept on file in the MMS laboratory. Every month, the histotechnician received and reviewed the QC forms for concordance. If discord occurred, the histotechnician alerted the MMS surgeon who then reviewed the slides with the dermatopathologist to elucidate the nature of the discordance and its effect on the patient’s treatment. The QC form precluded descriptive annotations for diagnosis but contained a comments section to allow a written record of interpretation or communication between the Mohs surgeon and dermatopathologist. A retrospective analysis was performed on a cohort of 1,720 consecutive cases recorded in the QC program between 2005 and 2014. Discord data were converted
2
Results
The authors’ study resulted in the analysis of 1,720 patient cases (Table 1). Initially, there were 21 disparate cases. However, 8 of these cases differed only in an additional actinic keratosis was identified without other discrepancies. Using the previously established concordance protocol regarding epidermal squamous atypia, these cases were then reclassified as concordant.4,5,7 The remaining 13 disparate cases were blindly re-evaluated by an independent dermatopathologist. In 3 of the 13 cases, the Mohs surgeon identified a tumor but the initial dermatopathologist did not. However, on rereview, the dermatopathologist did locate tumor (1 BCC and 1 SCC) in 2 of these cases but not in the third case which resulted in 2 more concordant cases. In the remaining 10 cases, the Mohs surgeon did not identify tumor but the initial
TABLE 1. Interpretation of Frozen Section Slides for Mohs Micrographic Surgery Dermatopathologist Tumor
No Tumor
Mohs surgeon Tumor No tumor
1,185
1
7
527
1,186 534
1,192
528
1,720
Sensitivity: 99.40% (95% CI: 98.80%–99.80%). Specificity: 99.80% (95% CI: 99.00%–100.00%). CI, confidence interval.
DERMATOLOGIC SURGERY
© 2017 by the American Society for Dermatologic Surgery, Inc. Published by Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
HIGHSMITH ET AL
dermatopathologist did record positive findings of tumor. The reviewing dermatopathologist identified small foci of tumor in 7 of these cases (6 were BCC and 1 was SCC) leading to 3 more concordant cases. Ultimately, 1,185 cases were determined to have a tumor (true positive) and 527 were determined not to have a tumor (true negative) by both the fellowshiptrained Mohs surgeon and dermatopathologist. Disagreement was reached in 8 cases. The Mohs surgeon identified no tumor in 7 cases, and the dermatopathologist conversely identified a tumor. Oppositely, the Mohs surgeon identified tumor in 1 case where the dermatopathologist did not. This results in a concordance rate of 99.5% {Kappa (SE; 95% confidence interval [CI]) = 0.99 (0.004; 0.98–0.99)}. The strength of agreement at this level is very good. In addition, sensitivity (95% CI) was 99.4% (98.8%–99.8%) and specificity (95% CI) was 99.8% (99.0%–100.0%) for the ability of the Mohs micrographic surgeon to identify tumor (or not) relative to reference diagnosis by a dermatopathologist. The negative predictive value (95% CI) was 98.7% (97.3%–99.5%). The 8 discordant cases were further evaluated and followed. Seven of these cases involved the head and neck and 1 on the lower leg. Histologic discord occurred in 2 cases which debated inflammation versus tumor (BCC in 1 case and SCC in the other). The remaining 6 cases involved a diagnosis of BCC or normal adnexal structures. One of these cases was deemed not clinically pertinent as discord only occurred in the biopsy layer and was not present on the actual Mohs layers. The other 7 cases were deemed clinically pertinent, as the discrepancy occurred within the first 6 sections of Mohs specimens. All 7 of these patients elected to be followed clinically, and none of the patients opted for re-excision, topical chemotherapy, or other physical modalities. All patients were being followed clinically at the time of this report without further clinical concerns of recurrence. Follow-ups range in duration from 6 months to 7 years (mean 35 months).
volume spent viewing horizontal frozen sections coupled with timely feedback from their mentor. Mohs micrographic surgery fellowship is designed to train physicians who have completed a 3-year dermatology residency which requires background training in histologic interpretation for board certification. This histology background serves as a foundation prerequisite because frozen section analysis is crucial to MMS. Most fellowship programs begin viewing slides on a multiheaded microscope. This affords the fellow the opportunity to view slides with the program director for immediate teaching opportunities. With time and experience, the fellow will begin to interpret slides independently. However, each slide will be evaluated by the mentor along with the fellow’s analysis for histologic and mapping accuracy. This process repeats daily during the MMS fellowship. A pilot study showed a histologic error rate of
BACKGROUND Mohs micrographic surgery enables the surgeon to maximize tumor removal while minimizing the amount of tissue being removed using advanced mapping techniques combined with microscopy. Interpretation of frozen section slides is vital to the success of Mohs micrographic surgery. OBJECTIVE Evaluate and confirm the congruence of frozen section slide interpretation between fellowshiptrained Mohs surgeons and dermatopathologists. METHODS
Retrospective cohort concordance study spanning 10 years (n = 1,720 cases).
RESULTS The concordance rate for frozen slide interpretation between fellowship-trained Mohs surgeons and dermatopathologists in this study was 99.5%. CONCLUSION This study demonstrates agreement between the interpretation of histologic frozen section slides when evaluated by fellowship-trained Mohs surgeons and dermatopathologists. The authors have indicated no significant interest with commercial supporters.
A
nnually within the United States, nonmelanoma skin cancer (NMSC) results in 4.9M cases and approximately 11,000 deaths.1–3 There are many treatment options for NMSC. Mohs micrographic surgery (MMS) is the treatment of choice in recurrent skin cancers and primary skin cancers that demand tissue preservation.4,5 Mohs micrographic surgery fellowship is a subspecialty designed to train board-certified dermatologists through an intensive, mentored, 1- to 2-year training experience in anatomy, surgery, tissue mapping, reconstruction, accurate histologic margin interpretation of frozen sections, and postoperative wound management. Fellowship training in MMS incorporates immediate feedback with mentored error correction. These training elements improve technique and minimize interpretive histologic errors which are important factors affecting recurrence rates.6 Previous
studies on this topic yielded concordance rates of 94.9% to 99.7%.4,5,7 Therefore, the purpose of this study was to confirm previous studies by conducting the largest histologic frozen section slide concordance study to date between fellowship-trained Mohs surgeons and fellowship-trained dermatopathologists.
Methods Two fellowship-trained Mohs surgeons, 9 fellowshiptrained dermatopathologists, and 11 MMS fellows produced the data set over a 10-year period. All cases in this study involved either basal cell carcinoma (BCC) or squamous cell carcinoma (SCC). It is the authors’ standard practice that a tumor layer is removed initially just before the first Mohs surgical layer to debulk the tumor as opposed to curettage. This tumor debulk layer and initial Mohs layer are
*Department of Dermatology (111H), James A. Haley Veterans’ Administration Hospital, Tampa, Florida; †Callahan Eye Hospital, University of Alabama at Birmingham, Birmingham, Alabama; ‡Total Skin and Beauty Dermatology Center, Birmingham, Alabama; xDermatology Surgery Institute, Lutz, Florida; kExtremity Trauma & Amputation Center of Excellence (EACE), U.S. Department of Veterans Affairs, Washington, District of Columbia; ¶Morsani College of Medicine, University of South Florida, Tampa, Florida; #319th Minimal Care Detachment, U.S. Army Reserves, Pinellas Park, Florida © 2017 by the American Society for Dermatologic Surgery, Inc. Published by Wolters Kluwer Health, Inc. All rights reserved. ISSN: 1076-0512 Dermatol Surg 2017;0:1–4 DOI: 10.1097/DSS.0000000000001352
·
·
1
© 2017 by the American Society for Dermatologic Surgery, Inc. Published by Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
HISTOLOGIC ACCURACY OF MMS
processed simultaneously by trained histotechnicians into frozen sections and stained with hematoxylin and eosin. Histopathologic interpretation was performed by 1 of the 2 Mohs surgeons or 1 of the 11 Mohs fellows under the supervision of the Mohs surgeon who also functioned as the pathologist. The tumor layer is read before reviewing the first Mohs micrographic surgical layer to visualize any distinct histological tumor feature. The senior Mohs surgeon reviewed each slide of every stage in concert with systematic tissue mapping. Additional surgical excision layers were performed in the same manner when deemed appropriate.
into a data set using Microsoft Excel (Microsoft Corp., Redmond, WA). All discordant slides were obtained and rereviewed with 1 independent dermatopathologist at a different institution. This dermatopathologist completed residency training in clinical and anatomic pathology as well as a fellowship in dermatopathology. Documentation of the discordant reviews was recorded onto QC forms and kept in the MMS laboratory. All discordant patient records were reviewed. The outcomes of these reviews were recorded into the data set.
In the quality control (QC) program, the histotechnicians selected every tenth slide from Mohs cases each week and were not aware of the type of tumor present on the slide when submitting them for quality assurance. The slide submitted may be from the tumor debulking layer with tumor present or any of the MMS layers. The histotechnician selecting the slide recorded the documented findings of the Mohs surgeon as noted on the intraoperative Mohs map. Hence, the MMS surgeon was blinded as to which slides were reviewed in the QC program.
All procedures were conducted and data collected in accordance with the Declaration of Helsinki.
The selected slides were then sent for analysis to an independent dermatopathology private practice for review. After submission, the slide diagnoses were recorded by the dermatopathologists who were blinded to the MMS surgeon’s interpretation. The interpretation of each slide was recorded on a standardized QC form which was kept on file in the MMS laboratory. Every month, the histotechnician received and reviewed the QC forms for concordance. If discord occurred, the histotechnician alerted the MMS surgeon who then reviewed the slides with the dermatopathologist to elucidate the nature of the discordance and its effect on the patient’s treatment. The QC form precluded descriptive annotations for diagnosis but contained a comments section to allow a written record of interpretation or communication between the Mohs surgeon and dermatopathologist. A retrospective analysis was performed on a cohort of 1,720 consecutive cases recorded in the QC program between 2005 and 2014. Discord data were converted
2
Results
The authors’ study resulted in the analysis of 1,720 patient cases (Table 1). Initially, there were 21 disparate cases. However, 8 of these cases differed only in an additional actinic keratosis was identified without other discrepancies. Using the previously established concordance protocol regarding epidermal squamous atypia, these cases were then reclassified as concordant.4,5,7 The remaining 13 disparate cases were blindly re-evaluated by an independent dermatopathologist. In 3 of the 13 cases, the Mohs surgeon identified a tumor but the initial dermatopathologist did not. However, on rereview, the dermatopathologist did locate tumor (1 BCC and 1 SCC) in 2 of these cases but not in the third case which resulted in 2 more concordant cases. In the remaining 10 cases, the Mohs surgeon did not identify tumor but the initial
TABLE 1. Interpretation of Frozen Section Slides for Mohs Micrographic Surgery Dermatopathologist Tumor
No Tumor
Mohs surgeon Tumor No tumor
1,185
1
7
527
1,186 534
1,192
528
1,720
Sensitivity: 99.40% (95% CI: 98.80%–99.80%). Specificity: 99.80% (95% CI: 99.00%–100.00%). CI, confidence interval.
DERMATOLOGIC SURGERY
© 2017 by the American Society for Dermatologic Surgery, Inc. Published by Wolters Kluwer Health, Inc. Unauthorized reproduction of this article is prohibited.
HIGHSMITH ET AL
dermatopathologist did record positive findings of tumor. The reviewing dermatopathologist identified small foci of tumor in 7 of these cases (6 were BCC and 1 was SCC) leading to 3 more concordant cases. Ultimately, 1,185 cases were determined to have a tumor (true positive) and 527 were determined not to have a tumor (true negative) by both the fellowshiptrained Mohs surgeon and dermatopathologist. Disagreement was reached in 8 cases. The Mohs surgeon identified no tumor in 7 cases, and the dermatopathologist conversely identified a tumor. Oppositely, the Mohs surgeon identified tumor in 1 case where the dermatopathologist did not. This results in a concordance rate of 99.5% {Kappa (SE; 95% confidence interval [CI]) = 0.99 (0.004; 0.98–0.99)}. The strength of agreement at this level is very good. In addition, sensitivity (95% CI) was 99.4% (98.8%–99.8%) and specificity (95% CI) was 99.8% (99.0%–100.0%) for the ability of the Mohs micrographic surgeon to identify tumor (or not) relative to reference diagnosis by a dermatopathologist. The negative predictive value (95% CI) was 98.7% (97.3%–99.5%). The 8 discordant cases were further evaluated and followed. Seven of these cases involved the head and neck and 1 on the lower leg. Histologic discord occurred in 2 cases which debated inflammation versus tumor (BCC in 1 case and SCC in the other). The remaining 6 cases involved a diagnosis of BCC or normal adnexal structures. One of these cases was deemed not clinically pertinent as discord only occurred in the biopsy layer and was not present on the actual Mohs layers. The other 7 cases were deemed clinically pertinent, as the discrepancy occurred within the first 6 sections of Mohs specimens. All 7 of these patients elected to be followed clinically, and none of the patients opted for re-excision, topical chemotherapy, or other physical modalities. All patients were being followed clinically at the time of this report without further clinical concerns of recurrence. Follow-ups range in duration from 6 months to 7 years (mean 35 months).
volume spent viewing horizontal frozen sections coupled with timely feedback from their mentor. Mohs micrographic surgery fellowship is designed to train physicians who have completed a 3-year dermatology residency which requires background training in histologic interpretation for board certification. This histology background serves as a foundation prerequisite because frozen section analysis is crucial to MMS. Most fellowship programs begin viewing slides on a multiheaded microscope. This affords the fellow the opportunity to view slides with the program director for immediate teaching opportunities. With time and experience, the fellow will begin to interpret slides independently. However, each slide will be evaluated by the mentor along with the fellow’s analysis for histologic and mapping accuracy. This process repeats daily during the MMS fellowship. A pilot study showed a histologic error rate of
