“Errata” does anyone read them? Revisiting mitotic counts in mast cell tumors

In this issue of Veterinary Pathology, the authors of the publication “Mitotic index is predictive for survival for canine cutaneous mast cell tumors” ¹⁷ published a short erratum that has the potential for significant impact on the histopathological evaluation of canine cutaneous mast cell tumors (MCT).^11,17 An error in calculating the size of the area in which mitotic figures (MFs) were counted resulted in the area reported as 2.7 mm², but it was 2.37 mm². The configuration of the microscope used for counting MFs in this study was a 40× objective and 10× oculars with a 22-mm field number (FN 22 mm). The area within 1 field of view (FOV) for a microscope with this configuration is 0.237 mm². All steps in counting MFs in each tumor were performed correctly and are detailed in the Materials and Methods section of the article. ¹⁷ The erratum also indicated that the descriptor “mitotic index” (MI) should be replaced with “mitotic count” (MC).

What is the “significance” of this correction? The corrected area, 2.37 mm², is the standardized area now used for counting MFs in tumors of animals with glass slides or whole-slide imaging (WSI).^7,13,14,16 Therefore, the outcome data, metastatic rate, and survival times correlated to MC in the 2007 manuscript could be applied to a canine cutaneous MCT evaluated with today’s methods.^13,16 If we want to employ a published grading system or use thresholds for evaluating tumor parameters in diagnostic cases, we must replicate the authors’ methodology precisely, necessitating, of course, that the methodology is sufficiently described in the publication. The 2007 manuscript is the only veterinary oncology publication prior to 2016 that reported the area in which MFs were counted using a standardized international unit for area (millimeters-squared, mm²).^4,6,14,17 The importance of using standardized international units to define the size of the area in a FOV is so others can replicate the exact size, with microscopes, WSI, or artificial intelligence.^{1,3,4,6,12,16,18} MFs should be defined, ⁷ reported as the number/X mm² (not per 10 high-power fields [hpf]), and as MC not MI.

The term hpf, which is reported in numerous oncology publications and is outdated, can vary nearly 6-fold in size depending on the microscope configuration, including the objectives and ocular lens FN mm.^4,8,10 Conversion tables for different microscope configurations, and method for calculation of the FOV area for microscopes and WSI are published.^5,13,14 Hpf and low-power field are not standard units of area and should be abandoned as measures of area and replaced with standardized international units for area, mm².^4,6,10 Standardization of the MC is essential to permit the application of information from published studies. If the published methods are not followed, the data are not diagnostically useful.

Unfortunately, when the area or diameter in the FOV is not defined in an original manuscript beyond hpf or 10 hpf, it is not possible to convert the data to mitotic figures per mm². This leaves clinicians and editors in the precarious position of being uncertain of how to use published data that did not provide these details and what needs to be required in future publications. Conversion from hpf to mm² could be accomplished if the configuration of the microscopes used in the original studies was known or the slides and blocks were still available. This dilemma would not exist if WSI had preceded microscopes as WSI requires an area to be defined in mm² and the images would be archived for review.^10,12,16,18 Journals should require authors to describe the methods used to determine MC in sufficient detail so that others can replicate the method, which includes defining the area in which MC was performed in a standardized international unit for area, such as mm². Using and reporting 10 hpf to define an area in which MC was performed should stop, in journals, textbooks, and pathology reports.

Does the 2007 MCT publication have clinical applicability? Absolutely. The purpose of the study was to determine if MC could be used to predict specific outcomes and survival metrics in dogs with cutaneous MCT and it did. ¹⁷ The authors studied 148 dogs, a large study by veterinary standards, and reported that for 99 dogs with known outcomes: “Our data indicate that MI (MC) is strongly predictive of biologic behavior and overall survival and should therefore be included as an additional prognostic when determining therapeutic options for canine patients with cutaneous MCTs.” ¹⁷ The authors used MC to predict metastases and survival times (ST) regardless of grade (3 tier system). ¹⁵ Dogs with an MC > 5 had a median ST of 2 months (19 dogs), while dogs with an MC ≤ 5 had a median ST of 70 months (80 dogs). Furthermore, this system divided grade II tumors ¹⁵ into 2 groups: dogs with an MC > 5 had a median ST of 5 months (7 dogs) while dogs with an MC < 5 had median ST of 70 months (72 dogs). Moreover, for dogs with grade III MCTs possessing an MC <5, the median survival time was not reached, suggesting that MC can also be used to identify a subset of grade III tumors likely to have long STs. ¹⁷ The investigators used a simple approach to produce a system that is discriminating, clinically useful, and that diagnostic pathologists can use in their workflow.

Others should read the 2007 manuscript for details not summarized here and read the 2009 letter to the editor in which another group validated the 2007 study using 57 different patients and various statistical procedures. ⁹ These authors concluded that their investigation “underlines the prognostic importance of the mitotic count for dogs with cutaneous MCTs.” ⁹ However, they did not define the area beyond “10 high-power fields.” Therefore, we contacted these authors, and they used a 40× objective and 10X FN 20 mm oculars, which equates to an area of 0.20 mm² in 1 FOV. Future investigations should attempt to validate both studies and avoid creating single digit thresholds. It may be “statistically sound” to create MC thresholds such as “6,” however, principles of biology would predict that 1 count on either side of the threshold number should not be used (alone) to influence prognostic implications. A more prudent approach would be to stratify MCs as done by both MCT studies cited here and that is also used in grading canine and human soft tissue tumors. Then assign numerical scores to each group and compare statistically known outcomes to MC scores.

New parameters and new data are the way forward, but first, we need to standardize our routine methods. All our early oncology studies need to be repeated with standardized methods for the parameters evaluated and for the outcomes assessed.^2,16 Investigators should also try to include new standardized methods to determine if 1 or more add to the predictability of our routine parameters. It is not clear that MCs performed in the invasive front, random counts, skipping to multiple hot spots, or using different size areas for different tumors will be the most predictive method. Finding the method that recognizes the most MFs in a tumor seems like a logical goal; however, the goal of counting MFs in tumors is not to find the average number of MFs or the greatest number of MFs/unit area, but the goal is to find the number of MFs in a selected region of a tumor that correlates with 1 or more outcomes. To determine this, we need large numbers of cases and teams of oncologists and pathologists that use standardized approaches and introduce new parameters. What an opportune time to be a young investigator interested in oncology and improving the care of animals with cancer. The erratum published in this journal may be the spark we need to standardize our methods, abandon the terms hpf and low-power field, replace MI with MC, report counts using mm², and to explain to our clinical colleagues that we do not know the area in which MFs were counted in the manuscripts you may be using to guide prognoses or select therapies. ¹⁵ The principles reported in those manuscripts are sound, but the exact thresholds reported cannot be accurately correlated to modern day methods. It is past time to stop reporting MI in 10 hpfs and report the MC in mm² in a defined sized area (for now, 2.37 mm²) and region of the tumor, as done in 2007.^11,17