Phospho-Histone H3 as a Tool for Improved Mitotic Counting and Risk Assessment in Gastrointestinal Stromal Tumors: A Study of 241 Cases

Introduction

Gastrointestinal stromal tumor (GIST), a rare mesenchymal tumor of the digestive tract, occurs predominantly in middle-aged and older individuals, with an estimated annual incidence of 10–15 cases per 100 000 population [1]. Common symptoms include abdominal discomfort, early satiety, and gastrointestinal bleeding, although many cases are discovered incidentally. Approximately 85–90% of gastrointestinal stromal tumors have oncogenic KIT or platelet-derived growth factor receptor (PDGFRA) mutations (accounting for 75–80% and about 10%, respectively), which leads to continuous activation of the kinase, thereby promoting tumor cell proliferation at an early stage of tumor formation [2]. Diagnosis mainly relies on enhanced CT/MRI imaging examinations, endoscopic ultrasound-guided biopsy, and immunohistochemistry (eg, CD117, DOG1). In treatment, localized GIST is mainly treated with complete surgical resection. For advanced patients with unresectable or metastatic cases, tyrosine kinase inhibitors (eg, imatinib) for targeted drug therapy can significantly improve prognosis [1]. GIST exhibits highly heterogeneous biological behavior, ranging from indolent growth to aggressive metastasis [3]. Accurately assessing the risk of recurrence and metastasis is crucial for guiding clinical treatment and prognosis determination. Currently, the revised risk stratification (RS) system released by the National Institutes of Health (NIH) in 2008 is widely used in clinical practice. This RS system uses tumor location, size, and mitotic count (MC) as the core evaluation indicators [4].

Among them, the mitotic count is a key pathological parameter reflecting the proliferative activity of tumor cells [5]. However, the traditional hematoxylin-eosin (HE) staining method relies on pathologists’ manual identification and counting under a microscope. This process not only requires patience but is also susceptible to interference from factors such as apoptotic cells, deeply stained chromatin, and tissue compression, resulting in differences among observers and within observers, especially when the counting results are close to the critical value of RS [6]. This can directly affect clinical risk assessment and treatment decisions. Over the years, to reduce the subjectivity and identification difficulty of MC-HE, much time and effort have been spent on training pathologists to ensure the accuracy and consistency of the counts. However, there are still some differences in interpretation among the pathologists.

In recent years, phospho-histone H3 (PHH3) has been identified as a specific organizational marker during the mitotic process [7]. This is because histone H3 is one of the basic proteins that constitute the core nucleosome of eukaryotic cell chromatin [8]. The 10th serine (Ser10) at the amino-terminal tail of histone H3 undergoes phosphorylation, which is a highly conserved and crucial event for cells entering the mitotic phase [9]. This modification begins in the prophase of mitosis, reaches its peak during chromatin condensation, and persists throughout metaphase and anaphase, until it is rapidly dephosphorylated by specific phosphatases at the end of cell division [10]. Therefore, the expression of PHH3 shows a distinct “all or nothing” pattern throughout the cell cycle, with almost no detection in interphase cells and strong positivity only in the cell nuclei of mitotic cells [11]. This strict spatiotemporal expression pattern allows PHH3 to be specifically and sensitively detected in mitotic cells via immunohistochemistry, establishing it as an ideal marker for mitotic figure counting [12]. Studies have shown that it has potential advantages in evaluating the proliferative activity of various tumors (eg, breast cancer, meningioma, melanoma) [6,12–17]. It can significantly improve the accuracy and efficiency of identifying mitotic images [18]. At present, few studies have reported on the correlation between PHH3 and GIST. Jin et al conducted PHH3 staining on 119 GIST patient samples and compared 3 counting methods (MC-HE [mitotic count on hematoxylin-eosin], MC-PHH3 [mitotic count on phospho-histone H3], and computer-assisted MC-PHH3). They found that MC-PHH3 was a reliable and superior alternative to MC-HE, suitable for risk assessment of GIST. Computer-assisted PHH3 image analysis counting could serve as an effective alternative to MC-PHH3, but further validation of its clinical significance and standardization is required [19]. Shin et al studied 77 patients of primary GIST. They found that MC-HE was strongly correlated with MC-PHH3 and had high inter-observer consistency, and could be an effective auxiliary tool for mitotic count in GIST. In some cases (eg, poor fixation or cell degeneration), PHH3 can more accurately reflect tumor proliferation activity, and can even identify high-risk cases that were underestimated by traditional methods [20]. Uguen et al studied 61 cases of GIST patients and found that in the assessment of recurrence risk, PHH3 and Ki-67 immunohistochemistry were more reproducible and faster than traditional HE staining. The number of mitotic figures detected by PHH3 was higher, which might change the RS of some borderline tumors [21]. Kemmerling et al found that Ki-67 immunohistochemistry can more accurately assess the proliferative activity of GIST compared to PHH3, and proposed a new threshold based on Ki-67, which enables more rapid and reliable prognosis assessment [22]. Erhan et al conducted a retrospective study on 98 cases of GIST and concluded that PHH3 can be used for mitotic assessment of newer specimens, but it cannot replace HE staining, but it is not applicable to long-term archived specimens, and its risk stratification lacks prognostic predictive value [23]. PHH3 shows good inter-observer consistency and can identify mitotic figures more easily than HE. Approximately 38% of cases had an increased grading after using PHH3. A PHH3 count of ≥7 was significantly associated with a poorer overall survival rate, while Ki-67 was less correlated with mitotic count and survival rate [24]. These studies provide a foundation for research on PHH3 in GIST. However, the sample sizes of most studies were small. Currently, there have been no relevant clinical studies in China. Whether it can become a reliable method for assisting in mitotic counting, how consistent the method is among observers, and whether the counting efficiency has improved all require further verification in large-scale studies. Therefore, this study aimed to evaluate RS in 241 patients with GIST using the cell proliferation marker PHH3 and cell mitosis counts.

Material and Methods

ETHICS STATEMENT:

This study was a retrospective pathological analysis. All the pathological specimens and clinical data used were archived and de-identified. It did not involve any additional intervention of the patients or risks of privacy leakage. It has been reviewed and approved by the ethics committee of our unit, and the requirement of informed patient consent was waived. The study protocol was approved by the Ethics Review Committee of the Yichang Central People’s Hospital (Approval Number 2024–42).

GENERAL INFORMATION:

The participants of this study were 241 patients with GIST (from January 2020 to December 2023), all of whom underwent radical surgery at our hospital. All cases were independently re-reviewed by 2 senior pathologists (with senior professional titles) for the HE-stained slides, and a consensus diagnosis was reached. Immunohistochemical staining showed 100% positivity for CD117 and DOG-1 in tumor cells. For tumors with partial positivity for CD117 or DOG-1, we conducted KIT and PDGFRA gene mutation tests to confirm the diagnosis. The diagnosis strictly followed the current consensus and standard procedures in the field of pathology to ensure the accuracy of the diagnosis for all included cases. Eligible participants needed to meet the following criteria: 1) Before surgery, no radiotherapy, chemotherapy, targeted therapy or other intervention measures affecting the biological behavior of the tumor were received; and 2) Complete clinical and pathological data were available, including age, sex, tumor location, size, whether it was ruptured, presence or absence of necrosis, pathological type, and immunohistochemical results. We excluded patients with concurrent malignant tumors in other systems, patients who had received any treatment before surgery, and patients with recurrent GIST.

Material and Methods

PROCESSING AND STAINING OF TISSUE SPECIMENS:

All surgical specimens were fully fixed with 4% neutral formaldehyde solution, then underwent conventional dehydration, clearing, and wax infiltration before being embedded in paraffin. Serial sections were prepared with a thickness of 4 μm. Two sections were made for each specimen, which were used for the following staining: 1) Conventional HE staining: Standardized staining was performed using the Roche V600 fully automatic staining machine, and all staining reagents were provided by Roche Company; 2) PHH3 immunohistochemical staining using the Envision 2-step method. The procedure was as follows: The sections were subjected to antigen retrieval under high temperature and high pressure (in a pressure cooker), and the retrieval solution was citrate buffer with a pH of 6.0. We added rabbit polyclonal anti-human PHH3 antibody (Santa Cruz Company, working concentration 1: 200) at 25°C for 2-h incubation, then added Envision secondary antibody and incubated at 25°C for 30 min, followed by diaminobenzidine staining, re-staining the cell nucleus with hematoxylin, and sealing with neutral gum. Known PHH3-positive sections were used as positive controls, and phosphate-buffered saline was used as a negative control instead of the primary antibody.

MITOTIC COUNT:

The mitotic counts were performed independently by 3 experienced pathologists in a blinded fashion. To ensure the accuracy of the results, all assessors were trained in the interpretation of mitotic figures before the assessment. The criteria for MC-HE were deeply stained and dense chromatin, presenting as spiky, rod-like, ring-like, dumbbell-like, or separate chromosome clusters (Figure 1A–1F), excluding pale staining of the nucleus, smooth perinuclear area (without spiky-like changes), eosinophilic cytoplasm, lymphocytes, vascular endothelial cells, necrotic zone cells, and impurities (Figure 2A–2I). The criteria for MC-PHH3 were brown to black-brown staining of the cell nucleus, with irregular nuclear morphology and rough edges (with spiky-like changes) of positive cells (Figure 3A–3F). Excluding non-specific staining such as only nuclear ribosome staining, cytoplasmic staining, non-specific staining of the necrotic area, and staining of inflammatory cells (Figure 4A–4I).

For each sample, the mitotic figures were counted from both the HE-stained sections and the PHH3 immunohistochemical sections. Each observer conducted counting using the 2 methods at different time points, and the observers were unaware of the previous counting results. Spearman rank correlation analysis was used to evaluate the correlation between the 2 methods. The consistency among observers was evaluated using the intra-class correlation coefficient (ICC). Firstly, the entire section was observed under a low-power microscope (4× or 10×) to identify the tumor area with the densest mitotic figures, which was designated as the counting area (or “hotspot”). In the selected “hotspot” area, the counting area was 5 mm², equivalent to 21 consecutive high-power microscope fields (40× objective lens, eyepiece field diameter 22 mm). The mitotic figures counted by each pathologist on each section and the time required for counting (in seconds) were recorded. Finally, the MC-HE value and MC-PHH3 value for each sample were taken as the average of the counts made by the 3 pathologists.

RISK STRATIFICATIONS:

According to the revised standards of the NIH in 2008, combined with the maximum diameter of the tumor, its location, and the count of mitotic figures (using the MC-HE and MC-PHH3 values, respectively), each case of GIST was classified into 4 risk levels: extremely low, low, medium, and high. The specific operation process was as follows: We recorded the maximum diameter and the primary location of each tumor, obtained the numerical values of mitotic figures based on HE and PHH3, respectively, then, based on the corresponding table in the NIH standard, for each case, we determined 2 risk levels independently using MC-HE and MC-PHH3. The difference between the 2 reflected the change in risk level.

STATISTICAL ANALYSIS:

Data analysis was conducted using SPSS 20.0 statistical software. Measurement data are expressed as median (interquartile range). Spearman rank correlation analysis was used to evaluate the correlation between MC-HE and MC-PHH3, as well as the consistency among different observers. The χ² test was used to analyze the relationship between the count of mitotic figures (referring to the NHI standard and divided into high and low groups at a boundary of 10/5 mm²) and clinical pathological parameters (eg, age, sex, tumor size, location, rupture, necrosis, risk grade). The consistency among observers was evaluated using the intra-class correlation coefficient (ICC). Corresponding 95% confidence intervals were computed based on Fisher transformation. The Wilcoxon signed-rank test was used to compare the time differences in the counts of MC-HE and MC-PHH3. A difference was considered statistically significant at P<0.05.

Results

CLINICOPATHOLOGICAL FEATURES OF 241 GISTS:

A total of 241 patients with primary GIST were included in this study. The clinicopathological features are detailed in Table 1. There were 129 male patients (53.5%) and 112 female patients (46.5%). The average age of the patients was 63 years (range: 37–85 years). The average maximum diameter of the tumors was 5.52 cm (range: 0.2–22.1 cm). The most common tumor location was the stomach (159 cases, 66.0%), followed by the small intestine (73 cases, 30.3%) and the large intestine (9 cases, 3.7%). Tumor rupture occurred in 61 cases (25.3%). The pathological types included spindle cell type (132 cases, 54.8%), epithelioid type (13 cases, 5.4%), and mixed type (96 cases, 39.8%). There were 49 cases (20.3%) with tumor necrosis and 192 cases (79.7%) with no necrosis. According to the NIH RS criteria based on MC-HE, there were 48 cases (19.9%) with very low risk, 73 cases (30.3%) with low risk, 22 cases (9.1%) with medium risk, and 98 cases (40.7%) with high risk. Based on MC-PHH3, there were 47 cases (19.5%) with very low risk, 73 cases (30.3%) with low risk, 21 cases (8.7%) with medium risk, and 100 cases (41.5%) with high risk.

RELATIONSHIP BETWEEN MITOTIC COUNT AND CLINICOPATHOLOGICAL PARAMETERS:

Using the mitotic count of 10/5 mm2 as the cut-off value, the cases were divided into a low mitotic group (≤10) and a high mitotic group (>10). The analysis results showed that with both MC-HE and MC-PHH3, a high mitotic count was significantly correlated with larger tumor diameters (P<0.01), tumor rupture (P<0.01), higher risk grade (P<0.01), and tumor necrosis (P<0.01). However, the mitotic count (MC-HE or MC-PHH3) was not significantly correlated with patient age, sex, tumor location, or histological type (P>0.05) (Table 1).

CORRELATION BETWEEN THE OVERALL RESULTS OF MC-PHH3 AND MC-HE, AND CORRELATION WITHIN DIFFERENT RISK STRATIFICATIONS:

Among the 241 cases of GIST, the median of MC-HE was 3.00/5 mm2, and the median of MC-PHH3 was 4.00/5 mm2. Spearman correlation analysis showed that there was a highly positive correlation between the 2 in all cases (r=0.879, P<0.01). In different risk stratification subgroups, MC-PHH3 also showed a good correlation with MC-HE (Table 2).

COMPARISON OF RS BASED ON MC-PHH3 AND MC-HE:

The risk stratification based on MC-PHH3 was consistent with that based on MC-HE in most cases (229/241, 95.0%). Changes in risk levels were observed in 12 cases (5.0%) (Table 3). Specifically, 8 cases experienced an upgrade: 1 case from very low risk to intermediate risk, 2 cases from low risk to intermediate risk, and 5 cases from intermediate risk to high risk. Four cases experienced a downgrade: 2 cases from high risk to intermediate risk, 1 case from high risk to low risk, and 1 case from intermediate risk to low risk. None of the 12 cases with changed risk levels experienced recurrence during the follow-up period.

CONSISTENCY AMONG OBSERVERS AND COUNTING EFFICIENCY:

The inter-observer consistency analysis revealed that the ICC for MC-HE count was 0.648 (95% confidence interval [CI]: 0.534–0.748), while the ICC for MC-PHH3 count was 0.710 (95% CI: 0.608–0.795), indicating that MC-PHH3 exhibited higher consistency among different observers (Table 4). Additionally, the average time spent by the 3 pathologists in identifying and counting mitotic figures on MC-PHH3-stained slides was significantly shorter than that on HE-stained slides (median time: 83 s vs 175 s, P<0.01), resulting in an efficiency increase of approximately 52.6% (Table 5). The intra-observer variability analysis also showed a high correlation between each observer’s own MC-HE and MC-PHH3 results (Figure 5A–5C).

Discussion

STUDY LIMITATIONS AND FUTURE PROSPECTS:

First, this was a retrospective, single-center study lacking an external validation cohort, which may affect the generalizability of the results. Second, while PHH3 staining performs well in most cases, its limitations should still be noted: for instance, in areas with abundant inflammatory cells, poor fixation, or tumor cell degeneration, non-specific staining may occur, leading to an overestimation of the count. Therefore, in practical applications, the tissue morphology should be considered, and counting should be avoided in areas with dense inflammation or necrosis. In necessary cases, multiple pathologists can jointly review to improve the accuracy of the results. Although the inter-observer consistency was high, the mitotic count still has some subjectivity. Third, we did not conduct follow-up analysis on the PHH3 count and the patient’s long-term prognosis, and its clinical prognostic value still requires further prospective research to verify. Moreover, this study did not employ an independent standard verification method to assess the accuracy of the HE or PHH3 count results. Therefore, this study can only reflect the consistency level between the 2 counting methods, but cannot determine the true accuracy of either method in evaluating tumor proliferation activity.

Although this study confirmed the auxiliary value of PHH3 in the mitotic count and RS of GIST through manual counting, we also recognize the important role of image analysis techniques in quantitative pathology. Digital pathology and artificial intelligence-assisted analysis systems can achieve rapid scanning of entire sections, with automatic identification and counting, significantly improving work efficiency and reducing human errors. However, the automated recognition algorithms for PHH3 immunohistochemistry are still under development, especially in distinguishing true mitosis from non-specific staining, which requires further optimization. Future research should combine full-section digital scanning with machine learning algorithms to establish more accurate and repeatable automatic counting models, further enhancing the objectivity and clinical applicability of GIST risk assessment.

Conclusions

A retrospective analysis of 241 patients with gastrointestinal stromal tumors (GIST) demonstrated that PHH3 immunohistochemistry closely corresponds with traditional hematoxylin and eosin (HE)-based mitotic counting. PHH3 enables faster and more consistent mitotic counts among observers, supporting its application as a practical tool for risk assessment. Both PHH3 and HE-based methods demonstrated strong concordance in risk classification and exhibited similar associations with established pathological features, indicating that PHH3 can improve mitotic evaluation within current NIH guidelines. Nevertheless, these findings should be interpreted with caution due to the single-center study design, lack of external validation, and absence of long-term outcome data. Technical factors such as staining quality and interpretative variability also remain important considerations. Despite these limitations, PHH3 is a valuable adjunct to enhance the reliability and efficiency of pathological assessment in GIST.