Article Text

Download PDFPDF
Prognostic significance of PAK4 expression in gastric cancer
  1. Kenta Kobayashi1,
  2. Mikito Inokuchi1,
  3. Yoko Takagi2,
  4. Sho Otsuki1,
  5. Yoshitaka Fujimori1,
  6. Yuya Sato1,
  7. Yoshimitsu Yanaka1,
  8. Kyoko Higuchi1,
  9. Tomoki Aburatani1,
  10. Chiharu Tomii1,
  11. Hiroyuki Uetake2,
  12. Kazuyuki Kojima1,3,
  13. Tatsuyuki Kawano1
  1. 1Department of Gastrointestinal Surgery, Tokyo Medical and Dental University, Tokyo, Japan
  2. 2Department of Surgical Specialties, Tokyo Medical and Dental University, Tokyo, Japan
  3. 3Center for Minimally Invasive Surgery, Tokyo Medical and Dental University, Tokyo, Japan
  1. Correspondence to Dr Mikito Inokuchi, Department of Gastrointestinal Surgery, Tokyo Medical and Dental University, 1-5-45, Yushima, Bunkyo-ku, Tokyo 113-8519, Japan; m-inokuchi.srg2{at}


Aims p-21 activated kinase (PAK) 4, part of the six PAK families, plays an important role in growth factor signalling, cytoskeletal remodelling, gene transcription, cell proliferation and oncogenic transformation. However, the clinical significance of PAK4 in gastric cancer has yet to be fully elucidated. PAK4 expression was evaluated, and the correlations of PAK4 expression with clinicopathological features and outcomes in gastric cancer were examined.

Methods Gastric adenocarcinomas obtained from 217 patients who underwent gastrectomy were analysed. PAK4 expression was evaluated using immunohistochemical staining.

Results PAK4 overexpression was found in 95 (43.8%) of 217 tumours . High PAK4 expression was significantly correlated with clinicopathological variables related to tumour progression, including depth of invasion, metastatic lymph nodes, pathological stage, distant metastasis or recurrent disease. High PAK4 expression was significantly associated with poorer disease-specific survival (DSS) (p<0.001) and relapse-free survival (RFS) (p<0.001). On multivariable analysis, PAK4 was an independent prognostic factor for DSS (HR 2.5 (95% CI 1.4 to 4.7), p=0.003) and RFS (HR 2.8 (95% CI 1.4 to 5.6), p=0.004). Even in stage II and III disease, PAK4 was an independent prognostic factor for RFS (HR 2.2 (95% CI 1.1 to 4.5), p=0.029).

Conclusions PAK4 may become a new prognostic factor in patients with gastric cancer.


Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Gastric cancer continues to have a high incidence and mortality worldwide. Based on the GLOBOCAN 2012 estimates, about 952 000 cancer cases and 723 000 cancer deaths were estimated to have occurred. Although the incidence of gastric cancer has gradually decreased in recent years, it is the third most common cause of cancer death.1 Even patients with gastric cancer who undergo standard resection including D2 lymphadenectomy have a poor 5-year survival.2 Combined chemotherapy consisting of fluorouracil and platinum is a standard regimen for metastatic gastric cancer. Trastuzumab, a monoclonal antibody against human epidermal growth factor receptor 2 (HER2), has been used as a molecular targeted drug for metastatic gastric cancer with high expression of HER2 protein. Trastuzumab plus standard chemotherapy substantially improved overall survival in such patients.3 However, only 10–20% of all patients with gastric cancer have HER2 overexpression.4 ,5 Ramucirumab, a monoclonal antibody against vascular endothelial growth factor receptor 2 (VEGFR2), has recently been approved for metastatic gastric cancer. In two phase III studies of second-line chemotherapy, ramucirumab prolonged overall survival.6 ,7 Tyrosine kinase receptors, such as HER2 and VEGFR2, or their downstream molecules are critically important targets for anticancer therapy. It is therefore crucial to investigate other biomarkers of gastric cancer that may serve as future targets of molecular-targeted therapy.

The p-21 activated kinases (PAKs) are a family of effector proteins for the family of Rho GTPases, consisting of Cdc42, Rac1 and Rho-A, known to affect various cellular functions required for cellular morphogenesis such as differentiation and migration.8 ,9 PAKs contain six mammalian isoforms, PAK1–6, and they regulate both cell motility and cell survival pathways and diverse cellular activities. PAKs are components of the mitogen-activated protein kinase (MAPK), JUN N-terminal kinase (JNK), steroid hormone receptor and nuclear factor κβ (NFκβ) signalling pathways, which have all been associated with oncogenesis. PAKs are divided into two groups based on their domain architecture. Group 1 contains PAK1–3 and Group 2 contains PAK4–6.10 In mammals, PAK4 is widely expressed in both embryonic and adult tissues, with high expression levels in the prostate, testis and colon.11 PAK4 overexpression has been detected in a number of tumour cell lines and in primary tumours.12 However, the clinical significance of PAK4 expression has not been sufficiently demonstrated in gastric cancer. The purpose of this study was to investigate PAK4 expression in human gastric cancer and to evaluate the relationships between PAK4 expression and clinicopathological features.



Gastric cancers from 217 patients who underwent gastrectomy with lymph node dissection between 2003 and 2007 at Tokyo Medical and Dental University Hospital were included in this study. Surgical approaches included both laparotomy and laparoscopy. The study was conducted in conformity with the Declaration of Helsinki13 and approved by the Institutional Review Board of the Tokyo Medical and Dental University. Written informed consent was obtained from all patients in this study. Each tumour was classified according to the tumour node metastasis (TNM) classification recommended by the Union for International Cancer Control (UICC). Tumours were histopathologically diagnosed based on the Japanese Classification of Gastric Carcinoma V.14.14 Papillary and tubular adenocarcinoma were classified into differentiated type, and poorly differentiated adenocarcinoma, signet-ring cell carcinoma and mucinous adenocarcinoma were classified into undifferentiated type. All patients were followed up after surgery. Examinations for recurrent disease were carried out every 3–6 months by serum tumour marker assays and diagnostic imaging (eg, oesophagogastroduodenoscopy, CT, ultrasonography or MRI). The median follow-up was 60 months (3–111 months). A total of 75 (35%) patients died, 50 (23%) had recurrent disease and 11 (5%) died of other causes. Patients who experienced distant metastases or recurrence received chemotherapy with S-1 (oral fluoropyrimidine consisting of tegafur, gimeracil and oteracil potassium; Taiho Co, Tokyo, Japan) alone or in a combined regimen. Twenty patients (9%) received adjuvant chemotherapy with S-1.


Well-preserved, representative, formalin-fixed, paraffin-embedded blocks were used for immunohistochemical staining. As the first antibody, a polyclonal rabbit antibody against PAK4, which had been used in other studies, was purchased from Abcam (Cambridge, UK).15 ,16 Secondary antibody conjugated to peroxidase-labeled polymer (Histofine Simple Stain MAX PO (MULTI); Nichirei Co, Tokyo, Japan) was applied. Blocks were cut into 4 μm thick sections. All available H&E-stained slides of the surgical specimens were reviewed. The sections were deparaffinised and dehydrated using a graded series of ethanol solutions. Antigen retrieval treatment was then carried out at 98°C (microwave processor, MI-77; Azumaya, Tokyo, Japan) for 8 min in pH 6.0, 10 mmol/L sodium citrate buffer (Mitsubishi Chemical Medience Corporation, Tokyo, Japan). Subsequently, the slides were cooled gradually after microwave irradiation until the buffer temperature fell below 45°C. Endogenous peroxidase was blocked with 3% hydrogen peroxide in methanol. The primary antibody was applied at 1:500 dilutions and the slides were incubated at 4°C overnight, followed by the secondary antibody for 30 min at room temperature. 3,3-Diaminobenzidine tetrahydrochloride substrate solution (Histofine Simple Stain DAB Solution; Nichirei Co) was then applied for colour development. The slides were counterstained with 1% Mayer's haematoxylin. Strongly stained gastric specimens obtained from the same block were set as positive controls to reduce the bias from the staining conditions with each immunohistochemistry (IHC), and matched negative controls were stained without primary antibody.

Interpretation of immunohistochemistry

The stained slides were evaluated by two separate investigators (KK and YT) who were blinded to patients’ outcomes. The investigators counted whole staining cancer cells of representative cross-sectional slices. For evaluating discretely distributed cancer cells, such as poorly differentiated adenocarcinoma, they counted at least three fields per section of the invasive front of each tumour. The staining intensity was scored into the following three grades: 0, negative to weak; 1, positive; and 2, strongly positive. The extent of staining (extensity) was then scored into four grades: 0, 0–25% of tumour areas stained; 1, 26–50%; 2, 51–75%; and 3, 76–100%. Composite scores were derived by addition of the intensity score and the extensity score for statistical analysis with respect to each patient. Composite scores ≥4 were defined as high expression and scores <4 were defined as low expression. Discrepant evaluations among the investigators were re-examined simultaneously using a double-headed microscope and one monitor to achieve consensus.

Statistical analysis

The χ2 test was used to test possible associations between PAK4 expression and clinicopathological factors. Kaplan–Meier curves were plotted using a log-rank test for univariable analysis of disease-specific survival (DSS) and relapse-free survival (RFS). Eighteen patients with stage IV disease were excluded from the analysis of RFS. Spearman's test was used to confirm correlation coefficients among clinicopathological prognostic factors selected by univariable analysis. Multivariable Cox proportional hazards regression models were used to assess the prognostic significance of PAK4 expression and of several clinicopathological factors. The log–log plot of survival was used to assess graphically whether the assumption of proportional hazards was reasonable. IBM SPSS Statistics V.22 software (IBM, Armonk, New York, USA) was used for statistical analysis. p Values <0.05 were considered significant.


Immunohistochemical findings of PAK4

PAK4 overexpression was mainly detected in the cytoplasm of cancer cells (figure 1). Although positive expressions were found in some nuclei of cancer cells, only cytoplasmic staining was counted towards the extensity score. Almost all fibroblasts, smooth muscle and muscularis mucosae were uniformly stained. Metaplastic intestinal epithelium located next to early cancer lesions was weakly stained. The deeper sites invaded by cancer cells tended to be stained more strongly than the surface sites (figure 1). Ninety-five (43.8%) of the 217 patients had high PAK4 expression. PAK4 expression was also evaluated in metastatic lymph nodes and then compared with expression in the primary lesions (figure 1). PAK4 expression in metastatic tumours in lymph nodes was significantly correlated with that of primary tumours (p=0.033). The concordance rate between these two lesions was 56.3% (table 1).

Table 1

Correlation of PAK4 expression between primary tumours and metastatic lymph nodes

Figure 1

Representative photographs of immunostaining of p-21 activated kinase 4 (PAK4) in (A) normal gastric epithelium with negative intensity, (B) primary gastric carcinomas with weak intensity, (C) with positive intensity and (D) with strongly positive intensity (magnification ×200). Representative photographs comparing the staining intensity of PAK in (E) a metastatic lymph node and (F) a primary lesion, both showing weak intensity staining. (G) A metastatic lymph node and (H) a primary lesion with strong intensity are shown (magnification ×200). (I) Representative photograph of intratumoral phenotypic PAK4 heterogeneity (magnification ×200); the upper inset shows a surface site of the tumour and the lower inset shows a site of deep invasion (magnification ×40).

Correlations between PAK4 expression and clinicopathological variables

High PAK4 expression was significantly associated with deeper tumour invasion (T1 vs T2–T4, p<0.001), extent of lymph node metastases (N0 vs N1–N3, p<0.001), advanced tumour stage (I vs II–IV, p<0.001) and the presence of distant metastases or recurrent disease (presence vs absence, p<0.001) (table 2). High PAK4 expression in the primary tumour was found in 51 (75%) of 68 patients with distant metastases or recurrent disease. PAK4 expression was not significantly different between the differentiated and undifferentiated types.

Table 2

Correlations of PAK4 expression with clinicopathological features

Survival analysis

High PAK4 expression was significantly associated with poorer DSS and RFS (both p<0.001). The 5-year DSS rate of patients with low PAK4 expression was 88.6%, while that of patients with high PAK4 expression was 50.1% (figure 2). Upper stomach lesion, undifferentiated type of cancer, deeper local invasion of primary tumour and positive lymph node metastases were significantly associated with poorer DSS on univariable analysis. Before multivariable analysis it was confirmed that these prognostic factors were not strongly correlated with each other. Multivariable analysis of clinicopathological features affecting DSS adjusted for established clinical prognostic factors (ie, histopathology, depth of invasion, lymph node involvement) indicated that high PAK4 expression, depth of cancer invasion and lymph node metastases were independent prognostic factors (HR 2.5 (95% CI 1.4 to 4.7), p=0.003; HR 7.3 (95% CI 1.7 to 31.6), p=0.008; HR 5.2 (95% CI 2.3 to 11.7), p<0.001; table 3). Similarly, multivariable analysis of RFS indicated that high PAK4 expression, depth of cancer invasion and lymph node metastases were independent prognostic factors (HR 2.8 (95% CI 1.4 to 5.6), p=0.004; HR 6.5 (95% CI 1.5 to 28.6), p=0.012; HR 5.2 (95% CI 2.2 to 12.8), p<0.001).

Table 3

Prognostic factors on univariable and multivariable Cox proportional hazards regression models for DSS in the study group as a whole

Figure 2

Disease-specific survival of all patients.

In patients with stage II or III disease (n=95), high PAK4 expression was significantly associated with poorer DSS and RFS on univariable analysis (p=0.002 and p=0.01, respectively; figure 3). On multivariable analysis adjusted for histopathology and clinical stage, PAK4 was an independent prognostic factor of RFS (HR 2.2 (95% CI 1.1 to 4.5), p=0.029; table 4).

Table 4

Prognostic factors on univariable and multivariable Cox proportional hazards regression models for RFS in stage II and III patients

Figure 3

Relapse-free survival of patients with stage II and III disease.


The results of this study suggest that high PAK4 expression may play a critical role in tumour progression, metastasis and survival in gastric cancer. PAK4 may be a valuable prognostic marker. PAK4-positive patients with stage II or III disease may postoperatively require more effective drugs such as PAK4 inhibitors, because PAK4 expression was identified as an independent prognostic factor in such patients.

PAK4 expression in human samples or cell lines of gastric cancer has been investigated in other studies.15 ,17 ,18 These studies pointed to an underlying oncogenic potential of PAK4. However, few studies showed correlations between PAK4 expression and prognosis in patients with gastric cancer. In one study, PAK4 overexpression was found in 4 (8.1%) of 49 metastatic gastric cancer specimens from patients who received palliative chemotherapy, and patients with gastric cancer with high PAK4 expression had a slight tendency towards poor outcomes.15

PAK4, the first identified member of group 2 PAK, is a target for Cdc42 and undergoes autophosphorylation on Ser-474.11 PAK4 was found to phosphorylate and activate Raf, firmly placing PAK4 in a Ras–Raf–MAPK pathway.19 Cancer cells that have high PAK4 levels also have mutations in genes in the senescence pathway, such as p53, thus allowing them to undergo transformation rather than senescence.19 PAK4 interacts with Smad2/3 and phosphorylates Smad2/3, thereby impairing the transforming growth factor β1 (TGF-β1) signalling pathway, resulting in escape from TGF-β-mediated growth inhibition in gastric cancer cells.18 DiGeorge critical region 6 (DGCR6L), which may be involved in cancer cell migration and is necessary for PAK4 forming a complex with β-actin, interacts directly with PAK4 and enhances PAK4-mediated migration of human gastric cancer cells through LIMK1.17

PAK4 knockdown by siRNA treatment induces apoptosis in the gastric cancer cell line that has a high level of PAK4 on western blotting.15 PAK4 siRNA transfection reverts cisplatin-resistant sublines from gastric cancer cell lines to their former state as cisplatin cytotoxic-sensitive.20 It is therefore tempting to speculate that PAK4 could be an attractive candidate for molecular targeted therapy. In fact, some PAK4 inhibitors were identified as agents suppressing tumour progression. PF-3758309, developed as an ATP-competitive inhibitor of PAK4, suppresses migration and invasion of a human lung cancer cell line.21 LCH-7749944 suppresses the proliferation of gastric cancer cells and mediates cell cycle arrest involving cyclin D1 repression in a human gastric cancer cell line.22 Unfortunately, a phase I study of a PAK inhibitor showed that it had low bioavailability.23 No relationship has been found between PAK4 and tyrosine kinase receptors targeted by molecular anticancer drugs such as HER2 and VEGFR2. In our series, 20 patients (9%) had overexpression of HER2, although PAK4 expression was not associated with HER2 expression. A PAK4 inhibitor may be a candidate anticancer drug in gastric cancer without expression of HER2.

Intratumoral heterogeneity in gastric cancer is one of the problems that may complicate molecular targeted therapy. Nishida et al24 showed both intratumoral phenotypic and genotypic HER2 heterogeneity in gastric cancer tissue prepared by tissue microarray techniques. In this study, PAK4 tended to be more strongly stained in the invasive front of the tumour than in the surface site. Therefore, PAK4 expression was assessed only in the invasive front of the tumour.

In this study, gene expression or alteration of PAK4 was not shown. However, in the Oncomine database, high expression of PAK4 mRNA levels and copy number gain of the PAK4 gene is shown in some datasets of human gastric cancers.25 The clinical significance of PAK4 gene expression needs to be determined in the future.

In conclusion, PAK4 was significantly associated with tumour progression and poor survival in gastric cancer. PAK4 expression was an independent prognostic factor. These findings suggest that PAK4 may potentially serve as a target of molecular targeted therapy of gastric cancer.

Take home messages

  • High expression of PAK4 is related to cancer progression.

  • High expression of PAK4 is significantly associated with poor survival in patients with gastric cancer. On multivariable analysis, PAK4 was an independent prognostic factor.

  • These results suggest that PAK4 could be a candidate target for molecular targeted therapy.



  • Handling editor Cheok Soon Lee

  • Contributors KK and MI were responsible for drafting the manuscript. KK and YT contributed to immunohistochemistry. KK and MI contributed to analysis and interpretation of data. KK, MI, HU and TK contributed to conduct the study. All authors read and approved the final manuscript.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.