The Argument for Mitotic Rate-based Grading for the Prognostication of Adrenocortical Carcinoma
Thomas J. Giordano, MD, PhD
T he University of Michigan Health System (UMHS) has developed a so-called “Destination Program” for endocrine oncology, with special emphasis on patients with adrenal and thyroid tumors.4 The goal of this program was to provide an efficient and effective multidisciplinary clinical experience in which patients with these tumors can be examined by several specialists in the morning, have their radiology images and pathology slides presented by noon to a tumor board, have their case discussed by members of the board, and receive treatment recommenda- tions in the afternoon. The program was launched in November 2009 and has been very successful in terms of patient visits, especially for patients with adrenocortical tumors (ACTs) [whether it will be successful in substantially improving the health of patients with adrenocortical carcinoma (ACC) remains to be determined]. As the designated UMHS endocrine pathologist, I have had the opportunity to review 1 to 3 new cases of ACTs almost every week, mostly ACCs. The most surprising aspect of this experience for me was the low frequency of mitotic grading reporting (< 25%) in outside ACC surgical pathology reports. Thus, my goal here was to briefly present why surgical pathologists should provide prognostic information by routinely grading ACCs in adults by mitotic counts and even go as far as reporting the actual mitotic count.
Mitotic counting is not a new endeavor and has been a crucial tool for the diagnosis and prognostication of several tumors, including but not limited to neuroblastoma,2 smooth muscle tumors,1,7 and breast carcinoma.8 It consists of identifying and counting mitotic figures in a predetermined area of the tumor present on a histologic slide. Mitotic counting sounds easy and straightforward, but it actually has some associated difficulties, such as (1) what are the proper criteria for what constitutes a mitotic figure, (2) how much area or number of cells should be counted to get a representative sampling of the tumor, (3) should a tumor subclone with a higher mitotic rate be preferentially counted over subclones with lower rates, (4) how does the thickness of the histologic sections bias the results, (5) reproducibility between pathologists, and (6) variability of tumor cell density as a result of differences in amounts of stroma and/or necrosis. These are all valid concerns and, collectively, they have discouraged some pathologists from partaking in mitotic counting. However, despite these limitations and sources of error, there exists a good argument for grading ACCs in this manner.
Larry Weiss,10 through his study on ACTs in the 1980s, is well known for his development of a standardized approach (now called the Weiss system) for the evaluation of ACTs. This system, described in these pages over 25 years ago, uses various parameters to determine which ACTs have malignant potential (ie, those that are ACCs). However, in a subsequent study,11 the majority of these parameters either had no or marginal statistical association with poor patient outcome. The sole exception to this observation was mitotic rate, which showed a strong association
with patient outcome. Thus, based on the study, it was proposed to grade ACCs solely based on their mitotic rates. Tumors with 20 or less mitotic figures per 50 high- power fields (HPFs) were proposed to be “low-grade,” whereas ACCs with > 20 mitotic figures per 50 HPF were proposed to be “high-grade.” Although the Weiss system is useful for diagnostically challenging cases, Weiss grading of ACC by mitotic rate into low-grade and high-grade groups represents the more significant patho- logic contribution for these patients, in my opinion based on clinical experience and additional research.
Over the last several years, my laboratory has explored whether genome-wide gene expression data can assist the pathologic analysis of ACTs and ACCs, especially by providing diagnostic and prognostic data above and beyond that typically provided by morphologic evaluation. In a first study using DNA microarray analysis, we showed that ACC as a group was distinctly different from normal cortex and adrenocortical adenoma, allowing the derivation of robust gene expression profiles specific to ACC.5 In a recent study,6 genome-wide gene expression profiles were analyzed in the context of patient outcome data and in the context of tumor stage and mitotic rate data as well. Importantly, both stage and mitotic grade were associated with patient survival in the UMHS study cohort. The ACCs were further classified into 2 distinct clusters or classes defined solely by gene expression profiles. A study by the French ACC group yielded strikingly similar results,3 providing support for an ACC classification with 2 main groups. These ACC clusters were associated with patient survival and a strong relationship between ACC cluster and tumor grade based on mitotic rate was observed (P = 0.004 in a 2-sided Fisher exact test). The poor outcome ACC cluster consisted of mostly high-grade tumors (87.5%), whereas the better outcome ACC cluster consisted mostly of low-grade tumors (64.7%). Thus, the molecular classi- fication of ACC derived from gene expression profiles imperfectly reflected the mitotic grade of the tumors. In addition, bioinformatic analysis of the genes that differed between the 2 ACC clusters showed a striking and statistically significant enrichment of genes related to cell proliferation and mitosis. Finally, detailed survival analysis showed that the log2 of the mitotic rate was significantly associated with survival using Cox proportional hazard models (P = 0.027, Wald test). In essence, as the mitotic rate doubled, the relative risk of dying of ACC in the next 5-year period went up 4.7-fold. This last result suggests that a simple 2-grade approach (ie, low grade vs. high grade) is not sufficiently representing all the relevant data that are present in mitotic counts. In summary, the overall narrative derived from molecular profiling data fit exceptionally well with what is known about the pathology of ACC.
We have recently expanded these observations using a larger cohort of 91 patients with ACC and have shown that mitotic grading provides useful information for stage 2 ACCs (ie, those >5cm and confined to the adrenal gland). Accordingly, a revision of the European Network for the Study of Adrenal Tumors staging system for ACC to incorporate mitotic grading was proposed.9
The results from these studies are also consistent with the overall morphology of ACC. Unlike many other cancers, such as colorectal and pancreatic adenocarcino- mas, that have a more complicated biology due to contributions from stromal and inflammatory cells, ACCs are typically devoid of significant desmoplasia and inflammatory cells and consist largely of tumor cells and blood vessels. Accordingly, ACC pathology and behavior more purely reflects the biology of its tumor cells compared with many other carcinomas.
It is fair to ask how mitotic grade information can be used clinically. In general, tumor grade provides information regarding the pace of the disease and that plays a large role in determining a treatment plan. Currently at our institution, a treatment plan is not finalized until the mitotic grade of the tumor is determined. In the setting of a stage II ACC that was been adequately resected, treatment options for a low- grade tumor might include watchful waiting or mitotane therapy. In contrast, if the tumor is high grade, a more aggressive approach that includes radiation to the surgical bed might be considered. In more advanced disease, knowledge of the grade of the tumor allows individualization of the timing and components of the systemic therapy. Grading has become an essential and vital part of the process.
In conclusion, despite the inherent difficulties and sources of error in mitotic counting, the available pathologic and molecular studies on ACCs with mitotic rate data have produced statistically significant results that uniformly show that tumor cell proliferation plays a dominant role in the biology of ACC. I believe there is enough primary pathologic and molecular data derived over the last 25 years to justify using mitotic rate grading to classify ACCs into low-grade and high-grade types, as well as going the extra step of reporting the raw mitotic count data in routine practice. Although there is much excitement regarding the application of personalized molecular pathology for patients with ACC, it is true that something as simple as assessing the growth rate of a tumor by mitotic counting still represents a strong benchmark against which future molecular assays must be compared.
I was contacted by a young endocrine surgeon at another academic institution while preparing this study. She had recently resected an adrenal tumor that was correctly diagnosed as ACC by an excellent pathology department. However, despite a detailed pathology report, she was uncertain about the likely clinical course of this tumor, and therefore asked me to provide a second pathology opinion. This shows that there is a real clinical need to provide grade information for ACC. If we do not routinely provide the same, our clinical colleagues will find a way to get it.
In practical terms, 50 fields are counted at × 400 using an Olympus BX40 microscope using a routine 5-um-thick hematoxylin and eosin-stained section. ACC subclones with the highest mitotic rate, when present, as is often the case with these large tumors, are preferentially
counted. In my experience, the behavior of the tumor reflects the highest grade subclone present in the tumor.
ACKNOWLEDGMENTS
The author thanks his numerous colleagues in the UMHS Endocrine Oncology program. The author gives a special thanks to Henry Appelman for useful discussions about mitotic counting and review of the article.
REFERENCES
1. Bell SW, Kempson RL, Hendrickson MR. Problematic uterine smooth muscle neoplasms: a clinicopathologic study of 213 cases. Am J Surg Pathol. 1994;18:535-558.
2. Burgues O, Navarro S, Noguera R, et al. Prognostic value of the International Neuroblastoma Pathology Classification in Neuroblastoma (Schwannian stroma-poor) and comparison with other prognostic factors: a study of 182 cases from the Spanish Neuroblastoma Registry. Virchows Arch. 2006;449: 410-420.
3. de Reynies A, Assie G, Rickman DS, et al. Gene expression profiling reveals a new classification of adrenocortical tumors and identifies molecular predictors of malignancy and survival. J Clin Oncol. 2009;27:1108-1115.
4. Giordano TJ. Adrenocortical tumors: an integrated clinical, pathologic, and molecular approach at the University of Michigan. Arch Pathol Lab Med. 2010;134:1440-1443.
5. Giordano TJ, Thomas DG, Kuick R, et al. Distinct transcriptional profiles of adrenocortical tumors uncovered by DNA microarray analysis. Am J Pathol. 2003;162:521-531.
6. Giordano TJ, Kuick R, Else T, et al. Molecular classification and prognostication of adrenocortical tumors by transcriptome profil- ing. Clin Cancer Res. 2009;15:668-676.
7. Jeffers MD, Oakes SJ, Richmond JA, et al. Proliferation, ploidy and prognosis in uterine smooth muscle tumours. Histopathology. 1996;29:217-223.
8. Lehr HA, Hansen DA, Kussick S, et al. Assessment of proliferative activity in breast cancer: MIB-1 immunohistochemistry versus mitotic figure count. Hum Pathol. 1999;30:1314-1320.
9. Miller BS, Gauger PG, Hammer GD, et al. Proposal for modification of the ENSAT staging system for adrenocortical carcinoma using tumor grade. Langenbecks Arch Surg. 2010;395: 955-961.
10. Weiss LM. Comparative histologic study of 43 metastasizing and nonmetastasizing adrenocortical tumors. Am J Surg Pathol. 1984;8:163-169.
11. Weiss LM, Medeiros LJ, Vickery AL Jr. Pathologic features of prognostic significance in adrenocortical carcinoma. Am J Surg Pathol. 1989;13:202-206.