Usefulness of anti-phosphohistone H3 immunoreactivity to determine mitotic rate in gastrointestinal stromal tumors


  • Ahrong Kim and Dong Han Im contributed equally to this work.


Kyung Un Choi, MD, PhD, Department of Pathology, Pusan National University School of Medicine, Beomeo-ri, Mulgeum-eup, Yangsan 626-770, Korea. Email:

*Ahrong Kim and Dong Han Im contributed equally to this work.


Background and aim

The biologic behavior of gastrointestinal stromal tumors (GIST) is a topic of continuing controversy. The assessment of accurate mitotic figures is known to be one of the major indicators for patients with GIST. However, it is not always easy to search for mitotic figures (MFs) and count them accurately on hematoxylin and eosin stained (H&E) slides.


In this study, phopho-histone H3 (PHH3), which is a recently described specific mitosis marker, was immunohistochemically examined and compared to the mitotic count on H&E sections (H&E-MI) and Ki-67 expression (Ki-67 PI). A hundred cases of histologically confirmed GISTs were reviewed based on counting MF on H&E slides. PHH3 positive MFs (PHH3-MI) were counted in the same way, and the Ki-67 PI was calculated for each case.


A strong correlation was found between PHH3-MI and H&E-MI. Recurrence-free survival was correlated with risk category by National Institutes of Health (NIH) consensus criteria (P = 0.017), mucosal invasion (P = 0.005), H&E-MI (P = 0.002), Ki-67 PI (P = 0.005), and PHH3-MI (P = 0.000). None of these factors was an independent prognostic factor.


Among GISTs, PHH3 staining was primarily found to support grading by facilitating mitotic counting, and it might have a prognostic value in GISTs.


Gastrointestinal stromal tumors (GISTs) are the most common mesenchymal tumors of the gastrointestinal tract, and they typically express KIT and have the activation of KIT or platelet-derived growth factor receptor alpha through mutually exclusive gain-of-function mutations.[1, 2] About 70% of GISTs occur in the stomach, 20% occur in the small intestine, and less than 10% occur in the esophagus, although GISTs occur throughout the gastrointestinal tract. GISTs have a wide clinical spectrum, from benign to frankly malignant. Tumor size, mitotic rate, and location can be used to predict the risk of recurrence in GIST patients.[3-7] As a result of major recent advances in understanding the biology of GISTs, the National Institutes of Health (NIH) suggested a consensus approach to diagnosis and morphologic prognostification.[8] The proposed scheme for estimating metastatic risks in GISTs is based on tumor size and mitotic count.

Proliferation activity provides prognostic data for different types of tumors as well as GISTs. However, the assessment of an accurate mitotic count is frequently subject to interobserver and intraobserver variability. This unsatisfactory situation has led many authors to try to define more objective indices of mitotic activity, including immunohistochemical markers of cell proliferation, such as Ki-67, MIB-1, proliferating cell nuclear antigen (PCNA) and so on,[9-12] but none of these has shown consistent improvement over the more accurate mitotic count. In recent years, it has been reported that anti-phosphohistone-H3 (PHH3) antibody might be a mitosis specific marker for the classification and grading of various types of tumors.[13-18] In mitotic cells, histone H3 is phosphorylated at serine 10 (Ser10) and Ser28, and the antibody PHH3 specifically detects histone H3 when phosphorylated at Ser28.[19, 20] PHH3 is consistently expressed in cells undergoing mitosis, while it is not expressed in interphase cells.

The aim of this study was to evaluate the usefulness of PHH3 immunostaining as a fast and reliable method of mitotic figure detection and consider its potential prognostic value regarding GISTs.


Tissue samples

Formalin-fixed, paraffin-embedded samples were obtained from 100 patients with GISTs who underwent surgical resection at Pusan National University Hospital between 2004 and 2009. The cases were histologically classified as GISTs based on the previously known histopathologic spectrum and their c-kit positivity. They were categorized on the basis of a proposed scheme for defining the risk of aggressive behavior in GISTs by the NIH consensus approach (Table 1). Clinical information and follow-up data were obtained from the medical records and a computerized prospective database. Pathologic parameters in this study included tumor site, tumor cell type, tumor size, mitotic count, and risk category according to the NIH consensus criteria. The patients in this study were all curatively resected. Recurrence-free survival (RFS) was calculated from the date of surgery to the date of tumor relapse or progression. The patients visited the hospital every 6 months on average and endoscopy or the computed tomography was performed if clinically indicated. All patients gave informed consent before specimen collection, according to institutional guidelines.

Table 1. Characteristics of patients with gastrointestinal stromal tumors (GISTs)
  1. HPF, high power field; SMA, smooth muscle actin.

  2. †According to the National Institutes of Health consensus criteria.

 Jejunum and ileum14
Tumor cell histologySpindle88
Tumor size≤2 cm26
 >2 cm and ≤5 cm59
 >5 cm and ≤10 cm12
 >10 cm3
Mitosis≤5/50 HPFs77
 >5 and ≤10/50 HPFs15
Risk categoryVery low21

Immunohistochemistry for Ki-67 and PHH3

Hematoxylin and eosin (H&E)-stained slides were reviewed in all cases. The most representative sections in each case were selected for immunohistochemical staining, obtaining a consecutive section after the H&E sections. Each slide was deparaffinized and rehydrated according to standard procedure and treated with 0.01 mol/L sodium citrate buffer (Ventana-Bio Tek Solutions, Tucson, AZ, USA) in a microwave pressure cooker at 120°C for 15 min. Immunohistochemical staining was performed via standard techniques using the UltraVision LP detection kit (Neomarkers, Fremont, CA, USA). The rabbit polycolonal antibody PHH3 (1:100; Cell Signaling, Danvers, MA, USA) and monoclonal antibody Ki-67 (MIB-1, 1:200; Dakocytomation, Carpinteria, CA, USA) were used as primary antibodies. Sections of lymph node were used as a positive control. Negative controls were performed by omitting the primary antibody.

Mitotic count

We used an Olympus Bx 51 microscope (Olympus, Melville, NY, USA). The mitotic counts on the H&E sections (H&E-MI) were measured by counting the mitotic figures per every 50 consecutive high power fields (HPFs) in the area of the highest mitotic activity. Mitotic rates were categorized as follows: mitotic rates ≤ 5/50HPFs were considered low, and mitotic rates > 5/50HPFs were considered high. The indices of PHH3 immunostained section were also measured in the same way. We counted the PPH3 mitosis index (MI) with the morphological features of prophase, metaphase, anaphase, and telophase, which are characterized by the strong and dense staining of chromatin clumps. PPH3-positive nuclei with fine granular staining and intact membranes were excluded from the quantification, because they correspond to cells in interphase. Ki-67 labeling indices were calculated by counting ki-67 positive cells among every 1,000 tumor cells in the area with the highest number of positive cells.

Statistical analysis

Statistical analysis was performed using SPSS version 17.0 (SPSS, Chicago, IL, USA). The associations between the variables were assessed using Pearson's χ2 test. RFS was calculated by the Kaplan–Meier log-rank test. Multivariate analysis was used to assess the independent prognostic value of factors by the Cox regression test. P-values less than 0.05 were regarded as statistically significant.


A summary of clinicopathologic data is shown in Table 1. Patients’ ages ranged from 27 to 80 years (mean, 57.5 years) at diagnosis. According to the NIH consensus criteria, very low risk tumors were 21 in number, low risk tumors were 49 in number, intermediate risk tumors were 15 in number, and high risk tumors were 15 in number. Very low risk and low risk tumors did not show any recurrence or metastasis during the follow-up period. Intermediate and high risk tumors were included and 30 patients were eventually retained for the survival analysis. Six patients (n = 30) showed disease recurrence, including local recurrence or metastasis, after a follow-up ranging from 24 to 76 months (mean, 40.9 months). RFS was 80%. None of them died of disease.

The staining of PHH3 was evaluated by counting of positive nuclei in tumor cells. The phopho-histone H3 positive mitotic figures (PHH3-MI) was readily identified, allowing for the fast identification of the mitotically active areas for counting. Because not only mitotic figures were stained with anti-PHH3 antibody but also any prophase nuclei, the latter must be excluded from counting. Similarly, apoptotic nuclei and crushed cells could also be distinguished from the mitotic figures on PHH3 stained slides (Fig. 1).

Figure 1.

Gastrointestinal stromal tumor of stomach in the high risk category by National Institutes of Health (NIH) consensus criteria. Hematoxylin and eosin staining shows cellular fasicles of spindle cells with mitotic figures (arrows) (a, ×400). Phopho-histone H3 immunostaining allows for the rapid identification of the highest mitotic activity (b, ×400) and shows the exclusion objects of apoptotic figures (arrowheads) (c, ×400) and interphase nuclei (arrowheads) (d, ×400).

Gastrointestinal stromal tumors of very low risk and low risk categories (n = 70) show no staining or less than two positive nuclei on PHH3 immunostained slides. The detailed description of the mitotic count performed on H&E stained slides of GISTs of the intermediate and high risk categories (n = 30), PHH3-MI, and Ki-67 expression (Ki-67 PI) is summarized in Table 2. In the intermediate risk group, the mitotic index on H&E stained slides ranged from 1 to 7 mitoses/50 HPFs, with a mean of 4.5 mitoses/50 HPFs. In the high risk group, H&E-MI ranged from 2 to 30, with a mean of 12.8 mitoses/50 HPFs. PHH3-MI was 1 to 8 with a mean of 3.9 in the intermediate risk group, and it was 1 to 38 with a mean of 13.5 mitoses/50 HPFs in the high risk group. Ki-67 PI was 1 to 10% (mean, 3.1%) in the intermediate risk group and 2 to 50% (mean, 12%) in the high risk group. A strong correlation was found between PHH3-MI and H&E-MI (correlation coefficient, 0.932). Ki-67 PI was also correlated with H&E-MI (correlation coefficient, 0.869).

Table 2. H&E-MI, PHH3-MI, Ki-67 PI, and tumor size with follow up in intermediated and high risk group tumors
Case No.Risk categoryH&E-MIPHH3-MIKi-67 PISize (cm)SiteFollow-up (months)
  1. H&E-MI, mitotic count on H&E sections; Ki-67 PI, Ki-67 expression; MI, mitotic index; NED, no evidence of disease; proliferation index; PHH3-MI, phopho-histone H3 positive mitotic figures

1Intermediate6432.5Stomach24, NED
2Intermediate7824.7Stomach28, NED
3Intermediate1119.2Duodenum28, NED
4Intermediate1157.2Stomach29, NED
5Intermediate3255.2Stomach30, NED
6Intermediate74102.5Stomach36, NED
7Intermediate7322.4Stomach36, NED
8Intermediate2116.3Jejunum36, NED
9Intermediate7553.7Stomach36, NED
10Intermediate6753.3Stomach37, NED
11Intermediate1116Jejunum40, NED
12Intermediate6224.8Jejunum41, NED
13Intermediate6353Stomach46, NED
14Intermediate1108Stomach47, NED
15Intermediate7303.2Stomach76, NED
16High118102.3Stomach25, NED
17High121082Rectum28, NED
18High6628.5Stomach31, local recurrence
19High3038203.7Stomach31, liver metastasis
20High21514.1Stomach35, NED
21High2430257Stomach40, liver metastasis
22High1415152.2Stomach44, NED
23High3040507Stomach48, liver metastasis
24High6526.4Stomach49, NED
25High42010.5Jejunum50, NED
26High1510104.5Stomach50, NED
27High6508Ilium50, NED
28High6235.5Stomach50, NED
29High1112152.5Rectum52, local recurrence
30High15151512.3Ilium73, local recurrence

When the risk category was re-evaluated on the basis of PHH3-MI and tumor size, five cases were shifted from the high risk group to the intermediate group (cases 17, 24, 26, 27, and 28), and seven cases from the intermediate group to the low risk group (cases 1, 6, 7, 9, 12, 13, and 15). According to these criteria, six patients showing disease recurrence were still included in the high risk group (Table 3).

Table 3. Risk group by NIH consensus criteria and new criteria based on tumor size and PHH3-MI
 Size (cm)H&E-MI (per 50 HPFs)No. of casesDisease recurrence (%)
  1. H&E-MI, mitotic count on H&E sections; HPF, high power field; NIH, National Institutes of Health; PHH3-MI, phopho-histone H3 positive mitotic figures.

NIH consensus criteria   
High>5>5156 (40%)
 Any size>10  
Criteria based on tumor size and PHH3-MI  
High>5>5106 (60%)
 Any size>10  

The RFS was correlated with risk category by the NIH consensus criteria (P = 0.017; 95% confidence interval, 0.002 to 14.505), mucosal invasion (P = 0.005; 95% CI, 6.199 to 15.801), H&E-MI (P = 0.002; 95% CI, 0.000 to 24.305), Ki-67 PI (P = 0.005; 95% CI, 0.000 to 55.722), and PHH3-MI (P = 0.000; 95% CI, 7.605 to 14.396) (Table 4, Fig. 2).

Table 4. Correlation between disease recurrence and clinicopathologic characteristics in intermediate and high risk groups of gastrointestinal stromal tumors
CharateristicsNo.Disease recurrence (%)P- value
  1. HPF, high power field; Ki-67 PI, Ki-67 expression; PHH3-MI, phopho-histone H3 positive mitotic figures.

  2. Percentage of disease progression cases per number.

SexMale143 (21.4)0.732
 Female163 (18.8) 
SiteStomach214 (19.0)0.600
 Nonstomach92 (22.2) 
Size≤5 cm132 (15.4)0.651
 >5 cm114 (36.4) 
Tumor cell histologySpindle244 (16.7)0.125
 Epithelioid51 (20.0) 
 Mixed11 (100) 
Tumor size≤2 cm10 (0)0.651
 >2 cm and ≤5 cm142 (14.3) 
 >5 cm and ≤10 cm123 (25.0) 
 >10 cm31 (33.3) 
Mitosis≤5/50 HPFs90 (0)0.002
 >5 and ≤10/50 HPFs131 (7.7) 
 >10/50 HPFs85 (62.5) 
PHH3-MI≤5/50 HPFs140 (0)0.000
 >5 and ≤10/50 HPFs91 (11.1) 
 >10/50 HPFs75 (71.4) 
Ki-67 PI<10%201 (5.0)0.005
 ≥10%105 (50.0) 
Risk categoryIntermediate150 (0)0.017
 High156 (40.0) 
Mucosal invasionNo273 (11.1)0.005
 Yes33 (100) 
Tumor necrosisNo234 (17.4)0.603
 Yes72 (28.6) 
Figure 2.

Recurrence free survival (RFS) of gastrointestinal stromal tumors. The RFS was correlated with risk category by National Institutes of Health (NIH) consensus criteria (a, P = 0.017), H&E-MI (b, P = 0.002), Ki-67 PI (c, P = 0.005), and PHH3-MI (d, P = 0.000). H&E–MI, mitotic count on H&E sections; HPF, high power field; Ki-67 PI, Ki-67 expression; PHH3–MI, phopho–histone H3 positive mitotic figures.


The pathologic assessment of malignancy in GISTs is difficult unless local invasion or metastases are apparent. Cytologic atypia, epithelioid type, vascular invasion, coagulative necrosis, and mucosal invasion have been reported to be correlated with an adverse outcome.[5-7] In addition, tumors with c-kit mutations of exon 11, the loss of heterozygosity in the region of chromosome 1p34–1p36, the inhibition of apoptosis resulting from the c-kit gene, and bcl2 expression behave more aggressively than those without these.[20-24] However, by far the most consistently useful parameters are tumor size and mitotic count.[4-7, 10, 25]

To estimate the prognostic significance of the classification schemes of GISTs, the diagnostic criteria used were those proposed by the NIH GIST Workshop (2001) based on tumor size and mitotic count. The NIH criteria include four risk groups; very low risk (<2 cm and <5 mitoses/50 HPF), low risk (2–5 cm and ≤5 mitoses/50 HPF), intermediate risk (≤5 cm and 6–10 mitoses/50 HPF or 5–10 cm and ≤5 mitoses/50 HPF), and high risk (>5 cm and >5 mitoses/50 HPF or >10 cm regardless of mitotic activity or >10 mitoses/50 HPF regardless of the tumor size). In one large study with long-term follow ups,[6] no case in the very low risk group led to tumor-related mortality, and fewer than 3% of patients in the low risk group had progressive disease. Tumors measuring 2–5 cm with >5 mitoses/50 HPFs were aggressive, with a 12–15% tumor-related mortality rate. High risk group tumors were more aggressive, with a 49–86% tumor-related mortality rate.

The method of mitotic counting is subjective and time-consuming and is associated with problems of reproducibility, although that is a routine procedure in assessing the grade of malignancy in tumors and has been reported to be the most important prognostic factor in GISTs. The subsequent challenge is to accurately distinguish true mitotic figures from karyorrheic debris and pyknotic nuclei in H&E sections.[26] There have been many attempts to identify immunohistochemical markers, that could be helpful in mitotic figure assessment and offer numerous practical benefits and reproducibility.[17, 18, 27-32] Many reports have shown various scoring systems using proliferative activity immunomarkers such as MIB-1, AgNOR, PCNA, bcl-2, and p53.[4, 9-12, 33, 34]

Recent studies have documented a good correlation between histone H3 phosphorylation and mitosis, with an antibody selective for the Ser10-PHH3 amino-terminus. In mitotic cells, histone H3 is phosphorylated at Ser10 or Ser28, and there is a good correlation between H3 phosphorylation and mitotic chromosome condensation initiated during the early phase. Phosphorylation of the Ser10 residue of histone H3 reaches a maximum during mitosis and is strongly correlated with the G2 to M transition, as compared to Ki-67, which stains the cells in all phases of the cell cycle, except for the G0 phase.[20] No phosphorylation of histone H3 has been observed during apoptosis, which is also the reason why PHH3 is considered to be a mitosis-specific marker. In many studies, PHH3 immunostaining has been attempted for breast carcinoma, astrocytoma, meningioma, and melanocytic tumors in which the mitotic index has been demonstrated to have diagnostic or prognostic importance, and it has been demonstrated to be a useful method of enhancing mitotic figure recognition.[13-16, 28, 29, 35]

This study showed a very good correlation between the H&E-MI and PPH3-MI and several advantages of PPH3 immunostaining in GISTs. It allows the clear and rapid identification of mitotic figures with little background staining. As a consequence, PPH3 immunostaining improves mitotic counting. Accordingly, the risk category of GISTs was shifted from the high risk category to the intermediate risk category in five cases and from the intermediate risk category to the low risk category in seven cases. PPH3 immunostaining made it possible to downgrade these cases by distinguishing apoptotic or crushed cells that were counted on H&E sections from mitotic figures. Six cases showing disease recurrence were retained in the high risk category when the criteria based on PHH3-MI and tumor size were applied. Using a combination of tumor size and PHH3-MI to evaluate risk could be a more reliable criterion to predict disease recurrence by facilitating mitotic counting. Ki-67 PI has been known to be a good marker of proliferative cells. However, it is uncertain how many of the cells expressing Ki-67 will actually undergo mitosis, and KI-67 PI is usually higher than H&E-MI.[13, 16] PHH3 immunostaining better highlighted mitotic figures as compared to Ki-67 immunostaining.

In previous studies, PHH3 immunostaining has been demonstrated to have prognostic importance. In meningioma, mitotic count is one of the most reliable predictors of meningioma recurrence like GISTs, and PHH3 immunostaining may be a sensitive and useful marker for miningioma grading based on the mitotic count.[16] The RFS of GIST cases in this study demonstrated significant correlations with NIH consensus criteria (P = 0.017), mucosal invasion (P = 0.005), H&E-MI (P = 0.002), Ki-67 PI (P = 0.005), and PHH3-MI (P = 0.000). However, none of the above factors show independent prognostic significance, probably due to the limited number and short follow-up period of cases included in this study.

In conclusion, the study shows that PHH3 immunostaining is a useful tool that could assist in the correct interpretation of mitotic figures and that PHH3-MI could help in the more accurate evaluation of risk categorization based on tumor size and mitotic count according to the NIH consensus criteria for GISTs.