Mitotic Cell Count

One of the cheapest, yet most straightforward, means of analyzing cell proliferation is determination of the mitotic figure count, and many studies have attempted to standardize the counting of mitoses in surgical pathology.⁴ Similar approaches also have been made for central nervous system (CNS) neoplasms.⁵ Typically, the mitotic count is performed in a given area of the tissue specimen and is reported as the number of mitoses per 10 high-power (magnification) fields (HPFs). Usually, an HPF represents the image obtained with the use of a 40× objective and a 10× ocular piece, which yields 400× magnification of the area. The mitotic count is often calculated in 10 such HPFs and frequently repeated on other sections/slides to give a more realistic estimate. This method provides a fairly reproducible index and has been used successfully as a prognostic parameter for many CNS and non-CNS neoplasms. For example, the mitotic cell count has been very helpful in distinguishing grade I from grade II meningiomas.

There are technical and methodologic challenges with the simple method of mitotic count, such as the effects of delay in fixation, the duration of fixation, and the counting method. Thus, without an established standard for mitotic cell count, it is not surprising to find significant variations among studies of a given neoplastic entity, especially in those with low mitotic counts. Nevertheless, mitotic count has been the most commonly used method for assessing cell proliferation. It may be possible to develop algorithms to provide a much more objective assessment of mitotic count with the advent of virtual microscopy and whole-slide imaging, and this prospect may enhance the practical value of counting mitotic figures.

Immunohistochemical Markers of Cell Proliferation

The use of antibodies directed against the well-known elements of cell proliferation has steadily been replacing mitotic cell counting techniques over the past two decades. With the discovery of robust antibodies that are suitable for use in formalin-fixed, paraffin-embedded material, immunohistochemistry is poised to provide a more practical and less subjective assessment of proliferating cells. However, many of the available markers and methods still require significant refinement.

The most common and reliable immunohistochemical markers for cell proliferation are the Ki-67 antibodies, developed against the same-named protein. The name of the antigen is derived from the city of origin (Kiel) and the number of the antibody clone in a 96-well plate. The target protein is present exclusively in the nuclei of proliferating cells at all active phases of the cell cycle and is absent in cells in the G₀ phase. Despite abundant evidence connecting this molecule to cell proliferation, the specific function of the protein is still elusive.

Formalin-fixed, paraffin-embedded tissues provide the optimal material for Ki-67 immunohistochemistry. There are a number of well-characterized monoclonal antibodies against the Ki-67 protein that have different qualitative and quantitative staining characteristics. MIB-1 antibody appears to have higher sensitivity for detecting Ki-67 antigen than do the other antibodies available. These differences become important, especially when different studies are compared or when a specific cutoff value is sought.

One of the commonly used markers of cell proliferation is proliferating cell nuclear antigen (PCNA). PCNA is a protein associated with DNA polymerase delta and is involved in control of DNA replication through the enzyme’s availability during elongation of the leading strand. The more commonly used antibodies for PCNA have been developed with the use of frozen tissue, and the antibodies still perform better in fresh or frozen material. Many of the antibodies developed for use in paraffin-embedded tissue have been polyclonal, with poor specificity, and have found limited use in clinical practice.

Other less common markers used for the detection of proliferating cells include JC1, an antibody that recognizes a nuclear antigen in proliferating cells, and others such as antibodies against cell cycle checkpoint molecules, including cdc2 p34, cdc20, and CDK1. Most of these markers have been useful in research studies, whereas their clinical utility has been limited.

Another set of markers typically restricted to mitotic cells are antibodies such as MPM-2 and PHH3, which have been helpful in determining the mitotic rate.⁶ Phosphorylation of tyrosine 3 of histone H3 is highly conserved among many species, and PHH3 antibodies have been used successfully in paraffin-embedded tissue to detect mitotic cells.⁷ Although this marker has only recently been introduced and validation data are limited, it may find better clinical use in the future. PHH3 markers are distinctly different from those such as Ki-67 in that they do not label all cells in the cell cycle and highlight only mitoses.

Markers that have been used in the past include fluorescence-labeled antibodies, special stains such as the silver stain for argyrophilic nucleolar organizer regions (AgNOR), and bromodeoxyuridine labeling.⁸ Although these markers have been useful for research applications in the past, they are of little practical value today.

Interpretation of Proliferation Markers

Whether one uses sections stained routinely with hematoxylin and eosin or those stained immunohistochemically, quantification of the proliferation rate requires a standardized evaluation method. The current methods are inherently semiquantitative because variables such as staining intensity and the number and type of cells counted are subject to marked variation. Determination of mitotic count is further confounded by the variability of the visual field in a given microscope. Typically, 400× HPFs vary from 0.18 to 0.25 mm², and this affects the total number of cells present in the HPF of a given microscope. Therefore, it is necessary to state the specific area of the HPF when reporting the mitotic count. This information is frequently omitted from manuscripts, a failure that is often attributable to lack of input from a pathologist.

Evaluation and reporting of immunohistochemical markers of proliferation are highly variable among different studies. The most common reporting method is the percentage of positively staining cells among a total of 1000 cells of interest. The percentage of positive cells is referred to as the labeling index. In general, antibodies against nuclear proteins are better suited for calculating a labeling index. This is typically done by counting tumor cells only. Yet other cell types may also be inadvertently counted, depending on the experience of the observer and the complexity of the tissue. One such source of variability is positive staining in cells of activated lymphocyte or macrophage lineage within tumors, which may be counted in the positive column.

Three significant challenges that can further complicate the current methods of evaluation should be emphasized. First, immunohistochemical stains may vary from one batch to another, and unless all stains are performed on the same day with the same controls, they are likely to show variation. This is critical when the labeling indices from different laboratories are compared. Second, it is important to determine a visual threshold of staining intensity above which the cells would be scored as positive. Unless such a practice is performed with actual optical measurements of staining intensity, the results are influenced by individual observation bias. Third, because the observer often seeks the area with the highest staining intensity and almost all tumors show marked regional variation in the distribution of positive cells, delineation of the area of interest is critical for reproducibility of the measurement. Furthermore, the results could vary depending on how many cells are counted, even if the same area were selected. Figure 99-1A demonstrates a graphic example of the influence of counting a different number of cells starting from the area of highest staining. The percentages in the figure represent the labeling indices if all cells within the respective circles are to be counted. As in this example, increasing the total number of cells counted may alter the labeling index. Figure 99-1B and C also highlight the fact that by counting only the area of highest intensity, it is possible to get a similar labeling index from two slides that actually have marked variation in overall labeling.

FIGURE 99-1 A, Graphic representation of potential variations in the Ki-67 labeling index based on the total number of cells counted. Each percentage represents the ratio of positive nuclei (black dots) to the total number of cells (black and white dots) within a given circle. Because the counting begins in an area of maximal staining, including a larger number of cells in the count is likely to alter the labeling index number considerably. B and C, Graphic example of the caveat in counting a single area of maximalm staining to obtain the Ki-67 labeling index. When counting the maximal area of staining (red circles), both examples would give a similar labeling index, although Example 1 has a much lower overall index than Example 2 does.

Although such challenges appear difficult to resolve, improved digital quantification and immunohistochemical analytic methods with increased reliability and standardization are emerging.⁹ We should expect to see much better validity and reliability for methods that take advantage of such advances.

Pilocytic and Pilomyxoid Astrocytomas

Pilocytic astrocytoma (PA) and its more aggressive variant pilomyxoid astrocytoma (PMA) are indolent, circumscribed tumors that are more common in children and young adults. PMA has been recognized as a variant with a higher rate of recurrence and dissemination, but its prognosis is still much more favorable than that of infiltrating gliomas. Nevertheless, markers of proliferation have not been useful in distinguishing between PA and PMA.

Numerous studies on proliferation markers for PA have been published, and the results are variable. Currently, no consensus exists on the typical proliferation rate of PA or its prognostic significance. A cutoff value for mitotic rate in either PA or PMA is not available.

There is substantial variation among PAs in terms of proliferation indices, and this does not appear to correlate with any of the known clinical parameters. The practical value of determining the proliferation rate in PA has been questioned in a number of studies.^10–12 These studies have failed to find any correlation between the proliferation rate as measured by labeling indices and the clinical outcome of patients with PA. Most studies report a mean Ki-67 labeling index of less than 2%, although a minority of tumors may have higher indices.^10–12

Similar proliferation rates have been reported for PMA and PA, but data on the proliferation index for PMA are limited.^13–16 Hopefully, future studies will better characterize the actual significance of cell proliferation in relation to the biologic behavior of PMA.

Pleomorphic Xanthoastrocytoma

The typical pleomorphic xanthoastrocytoma (PXA) is a low-grade neoplasm (World Health Organization [WHO] grade II) in which mitotic figures are observed only occasionally.¹⁷ Even though the WHO 2007 does not describe a grade III PXA, numerous studies highlight rare examples with the histologic characteristics of PXA and aggressive clinical behavior. Although rare mitoses are considered usual for PXA, the aggressive examples are reportedly more mitotically active and even show vascular endothelial proliferation or necrosis, or both. Rare case reports have alleged that because anaplastic transformation is often accompanied by increased mitotic activity, this finding should be considered to be a negative prognostic indicator.¹⁸ In a comprehensive review, Giannini and coauthors reported that up to 18% of cases in their study had mitotic counts greater than 5 per 10 HPF. The median Ki-67 (MIB-1) labeling index was 0.6%, and 79% of all cases had a labeling index of greater than 2%.¹⁹ Currently, it is advisable to consider an increased Ki-67 labeling index as worrisome for recurrence of PXA, especially if accompanied by additional aggressive histologic features.

Subependymal Giant Cell Astrocytoma

Subependymal giant cell astrocytoma (SEGA) is a WHO grade I neoplasm that typically occurs in the setting of the tuberous sclerosis complex. Most SEGAs harbor rare mitotic figures, frequently not reaching a quantifiable level, and the Ki-67 (MIB-1) labeling index is generally less than 3%.²⁰ Because most studies on SEGA are either small series or case reports, the prognostic value of proliferative markers in these neoplasms is not clear.

Diffuse Astrocytoma, World Health Organization Grade II

Countless studies have analyzed cell proliferation indices in diffuse astrocytomas. A cursory PubMed keyword search for clinicopathologic studies of cell proliferation in astrocytomas yields in excess of 300 publications, and a substantial number of these studies include WHO grade II diffuse astrocytomas. Most such studies suggest that the Ki-67 (MIB-1) labeling index positively correlates with histologic grade, as well as with mitotic activity and survival.²¹ In WHO grade II diffuse astrocytoma, the mitotic count and cell proliferation indices are significantly lower than those in grade III astrocytomas. Some suggest that even within grade II astrocytomas, patients with a higher MIB-1 labeling index have a shorter survival time.²² Others support this suggestion by reporting a positive correlation between Ki-67 labeling indices and the probability of diffuse astrocytomas recurring.²³ In addition to cell proliferation markers, the mitotic marker PHH3 was also suggested to be an independent prognostic marker for astrocytomas, and it appears that PHH3 staining correlates well with mitotic count and Ki-67 labeling.⁷ However, some studies have found no correlation between survival and the Ki-67 labeling index.²⁴ The variations in methodology and the issues mentioned previously often prevent accurate comparison of these studies and their conclusions.

Most studies of WHO grade II diffuse astrocytoma report a mean labeling index of less than 2%, but the range of staining has been much greater.²⁵ Sampling is a serious concern when determining an appropriate proliferation index for diffuse astrocytomas. Small biopsy specimens with only low-grade features of high-grade astrocytoma may yield scores that are much higher than expected. A more realistic interpretation of the labeling index requires knowledge of the radioimaging characteristics of the tumor, the site, the size of the biopsy sample, and knowledge of the staining and evaluation technique. When all of the aforementioned variables are optimal, one can realistically expect a labeling index of around 2% of tumor cells in diffuse astrocytoma. Much higher labeling indices should prompt scrutiny and raise suspicion that the sample may have been obtained from a higher grade astrocytoma.

Anaplastic Astrocytoma and Glioblastoma

Anaplastic astrocytomas (AAs) contain mitoses by virtue of their definition, but there is no clearly acceptable threshold for mitotic count in AAs. Some authors consider a single mitotic figure in a small biopsy sample sufficient for the designation of AA. Such tumors with a “solitary” mitosis have reportedly lower Ki-67 labeling indices and result in significantly longer patient survival times than do AAs with more than one mitotic figure.²⁶ Mean values reported for the Ki-67 labeling index in AAs have been around 10% but have ranged between 1% and 30%.^25,27 One caveat for Ki-67 labeling in AAs is the significant overlap of values reported for either diffuse astrocytoma or glioblastoma. It seems reasonable to consider the 5% to 10% range as being consistent with AA, and tumors with indices greater than this range should be suspected of being inadequately sampled glioblastomas.

Cell proliferation indices often demonstrate considerable variation within an individual AA. Some studies suggest that deletions of cell cycle checkpoint genes such as CDKN2/p16 in some high-grade astrocytomas may have a more deleterious effect on cell cycle control and might result in higher labeling indices. It is not clear whether such additional genetic alterations and increased cell proliferation have any effect on survival in patients with AA. In summary, although any labeling index can be considered consistent with the diagnosis of AA, the range expected should be between 5% and 10%.

Cell proliferation indices for glioblastomas are typically higher than those for AAs. Despite the significant overlap in values for these two tumor types, the mitotic rate and Ki-67 labeling index seem to correlate with patient outcomes in glioblastomas.^28,29 Yet this finding has not been widely reproducible. Some suggest using a cutoff value of 15% for a typical glioblastoma, but the significant variations stemming from manual counting methods influence this threshold. It is prudent to expect a very aggressive clinical course in histologically confirmed glioblastoma, regardless of the labeling index.

Although it is possible that some glioblastomas may have Ki-67 labeling indices similar to those of low-grade astrocytomas, this assumption should be approached with great caution. A critical issue involves tissues obtained from patients after radiotherapy or chemotherapy, or both. Glioblastomas from patients treated with radiotherapy are likely to show deceptively low labeling indices. In such cases, the proliferation indices do not appear to be of practical benefit, and the results are more likely to reflect a technical problem, sampling bias, or interpretive error.

Oligodendrogliomas

Proliferation indices and the mitotic count in oligodendroglial neoplasms depend on the tumor grade, as is seen in the case of infiltrating astrocytomas. By definition, WHO grade II oligodendrogliomas can have “occasional mitoses,” but a specific cutoff point for mitotic rate is not defined. Conversely, the finding of “abundant” mitoses is compatible with an anaplastic (WHO grade III) oligodendroglioma, but this designation also lacks a specific threshold. Proliferation markers such as Ki-67 exhibit a labeling index lower than 5% for grade II oligodendrogliomas and much higher for anaplastic tumors. In some studies, cell proliferation indices have correlated positively with disease-free survival in patients with grade II oligodendroglioma.^30,31 Some reports have found a positive correlation for Ki-67 labeling index and survival, but not for mitotic count,^32,33 whereas others have found no correlation for either.^34,35 The challenges of variability in determining the labeling index mentioned in the first section of this chapter apply especially to oligodendrogliomas.³⁶ These variations can partly explain the discrepancies in the literature. Furthermore, variations in classification criteria or pathologic interpretation may also play a significant role inasmuch as some studies report unusually poor 5- and 10-year survival probability for patients with oligodendrogliomas.³⁷ Although determination of the Ki-67 labeling index is highly dependent on the method and the observer, many studies imply a more aggressive behavior for tumors with indices greater than 5%, and this threshold should be considered carefully when reporting or grading oligodendrogliomas.

Ependymomas

Ependymomas are classified as WHO grade II or III neoplasms based on a set of histologic features, and “mitotic activity” is considered one of the criteria in the grading scheme. However, a specific cutoff point for mitotic rate has not been established for grading ependymomas. Two major caveats in studies that focus on the proliferation rate in ependymomas and its prognostic value have been the limited size and the diversity of the study populations. Many studies have included wide age ranges, with both adult and pediatric patients, and they often span a long period. In addition, only a small number of studies actually control for variables such as the extent of resection; tumor grade, stage, and location; and other confounding issues.³⁸ Nevertheless, many well-designed studies have identified histologic grade and proliferation rate (mitotic count, Ki-67 labeling) as independent prognostic indicators for patients with ependymomas.^39–43

The challenge in the case of pediatric and adult ependymomas is identification of a cutoff value for the mitotic count beyond which the risk for recurrence or progression is compatible with the diagnosis of an anaplastic (WHO grade III) ependymoma. Although some of the larger studies do not report a particular value,⁴⁰ many others consider 5 mitoses per 10 HPFs as a prognostically relevant cutoff point for ependymomas.^42–45 Even though there are some variations in the conclusions of the latter group of studies, it is fair to assume that tumors with less than 5 mitoses per 10 HPFs are much less likely to behave in an aggressive manner.

The proliferation index as measured by Ki-67 immunohistochemistry demonstrates even wider variations in studies of ependymoma. In the study by Bennetto and associates, the range of the Ki-67 labeling index in 74 childhood posterior fossa ependymomas was found to be 0.1% to 68.8%.⁴⁶ This study has reported a prognostically significant cutoff labeling index value of 25%; other studies have reported significant differences at 1%,³⁹ 4%,⁴⁷ 9%,⁴⁵ and 20%.⁴⁸

In short, although it appears that there is a significant tendency for anaplastic or clinically aggressive ependymomas to demonstrate high proliferation indices, it is prudent to consider a conclusion of Prayson that “MIB-1 (Ki-67) immunoreactivity labeling indices may be useful in concert with other histologic features, but are by no means absolutely predictive of tumor behavior by themselves.”⁴⁷ I believe that with increased standardization, quantification, whole-slide imaging, and digitized analysis of histologic and immunohistochemical stains, it may be possible to reduce the wider variations observed in the literature and consider a cutoff value that may be of more practical value.