Polo-like kinase 4 correlates with greater tumor size, lymph node metastasis and confers poor survival in non-small cell lung cancer.

BACKGROUND: Our study aimed to investigate the association of polo-like kinase 4 (PLK4) expression with tumor features as well as survival in non-small cell lung cancer (NSCLC) patients.
METHODS: Five hundred and sixty NSCLC patients who underwent pulmonary resection were recruited, and their tumor specimens were obtained. Immunohistochemistry (IHC) staining was performed to assess PLK4 expression in tumor specimen. Follow-up documents were reviewed, and the disease-free survival (DFS) and overall survival (OS) were evaluated.
RESULTS: According to IHC staining, there were 277 (49.5%) patients with PLK4 low expression and 283 (50.5%) patients with PLK4 high expression. PLK4 high expression was further classified into three different classes: high+, high++, high+++, and 122 (21.8%), 127 (22.7%), 34 (6.1%) patients were with PLK4 high+, high++, high+++ expression, respectively. Polo-like kinase 4 expression was correlated with larger tumor size, LYN metastasis, and higher TNM stage. As for survival, DFS and OS were lower in patients with PLK4 high expression compared with patients with PLK4 low expression. In addition, DFS and OS were the lowest in patients with PLK4 high+++ expression, followed by those with PLK4 high++ expression, PLK4 high+ expression, and then patients with PLK4 low expression. Univariate and multivariate Cox's proportional hazard regression model analyses further disclosed that PLK4 was an independent predictive factor for poor DFS and OS in NSCLC patients.
CONCLUSION: Our study preliminarily illuminates the clinical implication of PLK4 in NSCLC, while further studies are still needed to explicit the value of PLK4 in surveillance and treatment of NSCLC.

INTRODUCTION

As the leading cause of cancer‐related deaths worldwide, lung cancer is categorized into non‐small cell lung cancer (NSCLC) and small cell lung cancer with the former occupying 85% of all lung cancer cases.1 Since the initial presentation, NSCLC patients undergo multiple clinical and molecular examinations before implementation of treatment including clinical work‐up, biopsy, and molecular testing, and apart from surgery, platinum‐based chemotherapy, target therapy, and the novel immune therapy are administered accordingly.2 Although NSCLC is no longer incurable after decades of efforts, it is still premature to say that its prognosis is satisfying. Therefore, studies about novel biomarkers should not be shelved for deeper understanding of disease mechanism as well as better monitoring of disease progression and prognosis.

Polo‐like kinase 4 (PLK4), a prime member of protein kinase family that plays central role in cellular growth signaling pathways, is responsible for regulation of centrioles at the core of the centrosome.3 During cell mitosis, centrosome duplicates and becomes spindle pole at S phase, and inhibition of PLK4 leads to failure of centrosome duplication. In another hand, overexpression of PLK4 would drive centrosome amplification and cell cycle, which causes chromosome instability in cancer cells.4 In clinical studies, PLK4 is shown to be associated with higher pathological grade and poor prognosis in glioblastoma patients.5 As for in lung cancer, PLK4 inhibitor attenuated tumor growth via deregulating centriole duplication, mitotic defects, and deaths of lung cancer cells.6 Whereas for NSCLC, to be specific, the correlation of PLK4 with tumor development, progression, or prognosis is still unclear. Thus, we retrospectively analyzed data from 560 surgical NSCLC patients and investigated the association of PLK4 expression with tumor features as well as survival in NSCLC patients, so as to explore the clinical implication of PLK4 and its potential as biomarker for NSCLC.

MATERIALS AND METHODS

Subjects

Through reviewing the database of our hospital, 560 NSCLC patients who underwent pulmonary resection between January 2010 and December 2014 were included in this retrospective study. The inclusion criteria included the following: (a) pathologically confirmed diagnosis of primary NSCLC; (b) TNM stage I ~ III; (c) received pulmonary resection (including pneumonectomy, anatomic pulmonary resection, segmentectomy and wedge resection and sleeve lobectomy, etc); (d) no preoperative radiotherapy or chemotherapy; and (e) tumor specimen was accessible and available for immunohistochemistry (IHC) staining, while following patients were excluded: (a) complicated with other malignant tumors at initial diagnosis; (b) missing tumor specimen; and (c) missing clinical or follow‐up data. The Ethics Committee of our hospital approved this study, and the written informed consent or verbal agreement with tape recording was collected from patients or their guardians.

Specimen and clinical data collection

All patients' tumor specimens were obtained from the sample storage room of our hospital, which were formalin‐fixed and paraffin‐embedded. Meanwhile, the corresponding clinical data were collected from the case database. The clinical data mainly consisted of demographic features (including age, gender, history of smoke, and drink), complications (such as hypertension, hyperlipidemia, and diabetes), tumor characteristics at initial diagnosis (including pathological grade, tumor size, lymph node [LYN] status, and TNM stage), and carcinoembryonic antigen (CEA) level at initial examination.

IHC assessment

Immunohistochemistry staining was performed for assessment of PLK4 expression in the tumor specimen. Briefly, the specimens were cut into 4‐μm slices and baked overnight in an air oven. The slices were then deparaffinized using xylene and hydrated using graded ethanol. Next, the slices were quenched with fresh hydrogen peroxide to inhibit endogenous tissue peroxidase activity. Then, slices were placed in antigen retrieval citra buffer and brought up to boil for epitope retrieval. After blocked with normal goat serum, the slices were incubated with primary antibody PLK4 Rabbit Polyclonal Antibody (Invitrogen™) overnight at 4°C, followed by horseradish peroxidase‐conjugated goat anti‐rabbit immunoglobulin G antibody (Abcam). Subsequently, diaminobenzidine (DAB) was used for chromogenic reaction, and the hematoxylin was used for counterstaining. Ultimately, the slices were viewed and photographed on a microscope (Leica). A semi‐quantitative scoring method was used for IHC staining assessment, which was performed as described in previous study.7 The staining density was scored as: 0, 0%; 1, <25%; 2, 26 ~ 50%; 3, 51 ~ 75%; and 4, >75%. And the staining density was scored as: 0 (no staining), 1 (weak staining), 2 (moderate staining), and 3 (strong staining). Immunohistochemistry score was calculated thorough multiplying the density score by the intensity score. The cutoff point of three of IHC score was used to identify PLK4 high and low expression, that is, IHC score >3 was defined as PLK4 high expression; IHC score 0 ~ 3 was defined as PLK4 low expression. And the PLK4 high expression was further classified into three grades: IHC score 4 ~ 6, high+ expression; IHC score 7 ~ 9, high++ expression; and IHC score 10 ~ 12, high+++ expression.7

Postoperative treatment and follow‐up

Postoperative treatments of patients were decided by attending doctor with consideration of clinical status and TNM stage. Generally, stage I patients received close observation, chemotherapy, or radiation therapy (RT); stage II patients were treated with chemotherapy or chemoradiotherapy+ chemotherapy; stage III patients were given chemotherapy with RT or chemoradiotherapy+ chemotherapy. The chemotherapy regimen was mainly cis‐platinum‐based, and the RT was administered in accordance with the principles recommended by guidelines. Follow‐up was conducted regularly, and the last follow‐up date was June 30, 2019. Follow‐up documents were reviewed, and the disease‐free survival (DFS) and overall survival (OS) were evaluated.

Statistical analysis

Statistical analysis and figures plotting were performed on SPSS 24.0 software (IBM) and GraphPad Prism 7.01 software (GraphPad Software). Data were expressed as mean ± standard deviation (SD), median and interquartile range (IQR), or number (percentage). Clinical data comparison between PLK4 high expression and low expression patients was determined by the Student's t test, chi‐square test, or Wilcoxon rank sum test. The DFS was calculated from initial therapy to disease recurrence, disease progression, or death. The OS was calculated from initial therapy to death. Both DFS and OS were displayed using Kaplan‐Meier curves, and the difference between/among groups was determined by the log‐rank test. Factors predicting DFS and OS were analyzed by univariate and forward stepwise multivariate Cox's proportional hazard regression models. All tests were two‐sided, and P value <.05 was considered as significant.

RESULTS

Patients' characteristics

The mean age of 128 (22.9%) female and 432 (77.1%) male NSCLC patients was 61.5 ± 10.4 years (Table 1). The number of patients at pathological grade G1, G2, and G3 was 81 (14.5%), 360 (64.3%), and 119 (21.2%), respectively. In addition, the mean tumor size was 5.3 ± 2.2 cm, and 193 (34.5%) patients had LYN metastasis. As for TNM stage, there were 193 (34.5%), 167 (29.8%), and 200 (35.7%) patients at TNM stage I, II, and III, respectively. Other detailed characteristics were listed in Table 1.

Table 1

Clinical characteristics of NSCLC patients

Items	NSCLC patients (N = 560)
Age (y), mean ± SD	61.5 ± 10.4
Gender, No. (%)
Female	128 (22.9)
Male	432 (77.1)
History of smoke, No. (%)	308 (55.0)
History of drink, No. (%)	219 (39.1)
Hypertension, No. (%)	211 (37.7)
Hyperlipidemia, No. (%)	174 (31.1)
Diabetes, No. (%)	93 (16.6)
Pathological grade, No. (%)
G1	81 (14.5)
G2	360 (64.3)
G3	119 (21.2)
Tumor size (cm), mean ± SD	5.3 ± 2.2
LYN metastasis, No. (%)	193 (34.5)
TNM stage, No. (%)
I	193 (34.5)
II	167 (29.8)
III	200 (35.7)
CEA (ng/mL), median (IQR)	6.8 (3.1‐27.3)

Abbreviations: CEA, carcinoembryonic antigen; IQR, interquartile range; LYN, lymph node; NSCLC, non‐small cell lung cancer; SD, standard deviation.

Expression of PLK4 in NSCLC

The IHC staining images of PLK4 in NSCLC tumor specimen were shown in Figure 1. There were 277 patients with PLK4 low expression and 283 patients with PLK4 high expression. Additionally, PLK4 high expression was further classified into high+ expression (IHC score 4 ~ 6), high++ expression (IHC score 7 ~ 9), and high+++ expression (IHC score 10 ~ 12), with 122, 127, and 34 patients presenting PLK4 high+, high++, and high+++ expression, respectively.

Figure 1

PLK4 expression in NSCLC. IHC staining of PLK4 low expression, and the three grades of the PLK4 high expression: high+ expression; high++ expression; and high+++ expression. PLK4, polo‐like kinase 4; NSCLC, non‐small cell lung cancer

Correlation of PLK4 expression with patients' characteristics

Polo‐like kinase 4 expression was correlated with larger tumor size (P < .001), LYN metastasis (P < .001), and higher TNM stage (P < .001) in NSCLC patients, and however, it was not associated with other characteristics (All P > .05) (Table 2).

Table 2

Comparison of clinical characteristics between PLK4 high and low expression patients

Items	PLK4 expression		P value
	Low (n = 277)	High (n = 283)
Age (y), mean ± SD	61.6 ± 10.3	61.3 ± 10.6	.735
Gender, No. (%)			.252
Female	69 (24.9)	59 (20.8)
Male	208 (75.1)	224 (79.2)
History of smoke, No. (%)	157 (56.7)	151 (53.4)	.430
History of drink, No. (%)	106 (38.3)	113 (39.9)	.687
Hypertension, No. (%)	100 (36.1)	111 (39.2)	.446
Hyperlipidemia, No. (%)	84 (30.3)	90 (31.8)	.706
Diabetes, No. (%)	40 (14.4)	53 (18.7)	.173
Pathological grade, No. (%)			.078
G1	43 (15.5)	38 (13.4)
G2	186 (67.2)	174 (61.5)
G3	48 (17.3)	71 (25.1)
Tumor size (cm), mean ± SD	4.9 ± 2.0	5.6 ± 2.2	<.001
LYN metastasis, No. (%)	75 (27.1)	118 (41.7)	<.001
TNM stage, No. (%)			<.001
I	116 (41.9)	77 (27.2)
II	83 (30.0)	84 (29.7)
III	78 (28.1)	122 (43.1)
CEA (ng/mL), median (IQR)	6.2 (2.8‐19.9)	8.2 (3.3‐30.5)	.105

Comparison was determined by Student's t test, chi‐square test, or Wilcoxon rank sum test.

Abbreviations: CEA, carcinoembryonic antigen; IQR, interquartile range; LYN, lymph node; NSCLC, non‐small cell lung cancer; SD, standard deviation.

Correlation of PLK4 expression with survival in NSCLC patients

Disease‐free survival was lower in patients with PLK4 high expression compared with patients with PLK4 low expression (P < .001) (Figure 2A). Furthermore, patients with PLK4 high+++ expression presented the lowest DFS, followed by those with PLK4 high++ expression, PLK4 high+ expression, and then patients with PLK4 low expression (P < .001) (Figure 2B). Similarly, OS was decreased in patients with PLK4 high expression compared with patients with PLK4 low expression (P < .001) (Figure 2C). Besides, patients with PLK4 high+++ expression had the lowest OS, followed by patients with PLK4 high++ expression, PLK4 high+ expression, and patients with PLK4 low expression (P < .001) (Figure 2D). Furthermore, patients were subdivided according to their TNM stage, and the correlation of PLK4 expression with survival was explored in TNM stage I, II, and III subgroups, respectively. PLK4 high expression was correlated with shorter DFS in TNM stage I subgroup (P < .001) (Figure 3A) and that in TNM stage III subgroup (P < .0001) (Figure 3C), but not in TNM stage II subgroup (P = .171) (Figure 3B). And PLK4 high expression was associated with lower OS in TNM stage I subgroup (P = .002) (Figure 3D) and TNM stage III subgroup (P < .001) (Figure 3F), but not TNM stage II subgroup (P = .202) (Figure 3E).

Figure 2

Negative correlation of PLK4 with survival in NSCLC patients. Comparing DFS between PLK4 high expression and PLK4 low expression patients (A). Comparing DFS among PLK4 high+ expression, high++ expression, high+++ expression, and low expression patients (B). Comparing OS between PLK4 high expression and PLK4 low expression patients (C). Comparing OS among PLK4 high+ expression, high++ expression, high+++ expression, and low expression patients (D). PLK4, polo‐like kinase 4; NSCLC, non‐small cell lung cancer; DFS, disease‐free survival; OS, overall survival

Figure 3

Negative correlation of PLK4 with survival in NSCLC patients subdivided by TNM stage. The correlation of PLK4 with DFS in patients at TNM stage I (A), II (B) and III (C). The correlation of PLK4 with OS in patients at TNM stage I (D), II (E), and III (F). PLK4, polo‐like kinase 4; NSCLC, non‐small cell lung cancer; DFS, disease‐free survival; OS, overall survival

Factors predicting DFS

Univariate Cox's proportional hazard regression disclosed that PLK4 high expression (P < .001), age (>60 years) (P = .039), high pathological grade (P = .008), larger tumor size (P < .001), LYN metastasis (P < .001), advanced TNM stage (P < .001), and higher CEA level (P = .003) were correlated with poor DFS in NSCLC patients (Table 3). Further multivariate Cox's proportional hazard regression exhibited that PLK4 high expression (P < .001), high pathological grade (P = .040), LYN metastasis (P < .001), and higher CEA level (P = .004) were independent predictive factors for short DFS in NSCLC patients.

Table 3

Univariate and multivariate Cox's proportional hazard regression model analyses of factors predicting DFS

Items	Cox's proportional hazard regression
	P value	HR (95%CI)
Univariate Cox's proportional hazard regression model
PLK4 high expression	<.001	1.766 (1.460‐2.136)
Age (>60 y)	.039	1.220 (1.010‐1.474)
Gender (male)	.209	1.160 (0.920‐1.463)
History of smoke	.370	0.918 (0.761‐1.107)
History of drink	.927	0.991 (0.818‐1.200)
Hypertension	.741	0.968 (0.798‐1.174)
Hyperlipidemia	.945	1.007 (0.823‐1.233)
Diabetes	.183	0.839 (0.648‐1.086)
Pathological grade (G3)	.008	1.352 (1.081‐1.691)
Tumor size (>5 cm)	<.001	1.402 (1.160‐1.695)
LYN metastasis	<.001	2.533 (2.083‐3.079)
TNM stage (III)	<.001	1.801 (1.488‐2.181)
CEA (>5 ng/mL)	.003	1.333 (1.101‐1.613)
Multivariate Cox's proportional hazard regression model
PLK4 high expression	<.001	1.673 (1.380‐2.027)
Pathological grade (G3)	.040	1.266 (1.011‐1.586)
LYN metastasis	<.001	2.417 (1.985‐2.942)
CEA (>5 ng/mL)	.004	1.330 (1.098‐1.611)

Abbreviations: CEA, carcinoembryonic antigen; CI, confidence interval; DFS, disease‐free survival; HR, hazard ratio; LYN, lymph node.

Factors predicting OS

As for factors predicting OS, PLK4 high expression (P < .001), high pathological grade (P = .005), larger tumor size (P < .001), LYN metastasis (P < .001), advanced TNM stage (P < .001), and higher CEA level (P < .001) were correlated with unfavorable OS in NSCLC patients (Table 4). Further multivariate Cox's proportional hazard regression showed that PLK4 high expression (P < .001), high pathological grade (P = .018), LYN metastasis (P < .001), and higher CEA level (P < .001) were independent predictive factors for short OS in NSCLC patients.

Table 4

Univariate and multivariate Cox's proportional hazard regression model analyses of factors predicting OS

Items	Cox's proportional hazard regression
	P value	HR (95%CI)
Univariate Cox's proportional hazard regression model
PLK4 high expression	<.001	1.801 (1.459‐2.222)
Age (>60 y)	.626	1.053 (0.856‐1.295)
Gender (male)	.902	0.985 (0.770‐1.259)
History of smoke	.410	0.917 (0.746‐1.127)
History of drink	.931	1.009 (0.818‐1.246)
Hypertension	.688	0.957 (0.773‐1.185)
Hyperlipidemia	.485	0.923 (0.736‐1.156)
Diabetes	.187	0.824 (0.619‐1.098)
Pathological grade (G3)	.005	1.416 (1.113‐1.801)
Tumor size (>5 cm)	<.001	1.605 (1.306‐1.973)
LYN metastasis	<.001	3.184 (2.578‐3.932)
TNM stage (III)	<.001	1.738 (1.409‐2.144)
CEA (>5 ng/mL)	<.001	1.662 (1.339‐2.062)
Multivariate Cox's proportional hazard regression model
PLK4 high expression	<.001	1.660 (1.342‐2.052)
Pathological grade (G3)	.018	1.340 (1.051‐1.707)
LYN metastasis	<.001	3.001 (2.426‐3.711)
CEA (>5 ng/mL)	<.001	1.629 (1.312‐2.023)

Abbreviations: CEA, carcinoembryonic antigen; CI, confidence interval; HR, hazard ratio; LYN, lymph node; OS, overall survival.

DISCUSSION

In our study, we observed that PLK4 high expression was correlated with larger tumor size, LYN metastasis, and higher TNM stage, and it was an independent predictive factor for poor DFS and OS in NSCLC patients.

Polo‐like kinases are important regulators in cell cycle and DNA damage response, which are shown to be closely related to pathogenesis of cancers. In vitro study reports that PLK4 promotes tumor cell proliferation, survival, invasion, and migration in central nervous system neuroblastoma and increases tumor growth.8 And in pancreatic cancer, the inhibitor of PLK4 results in a significant reduction of tumor initiating cells.9 In clinical studies, PLK4 is associated with higher incidence of LYN metastasis, distant metastasis, or surrounding recurrence in breast cancer patients.10 Whereas in hepatocellular carcinoma, low PLK4 expression is correlated with advanced clinical stage, higher AFP, and larger tumor size.11 As for in lung cancer, inhibitor of PLK4 has been shown to cause mitotic defects and cell death, which suppress tumor growth, and however, studies regarding the direct implication of PLK4 in tumor progression are still lacking. In the present study, we assessed the correlation of PLK4 with clinicopathological characteristics of NSCLC patients and disclosed that PLK4 high expression was correlated with larger tumor size, LYN metastasis, and higher TNM stage, which was in accordance with the findings from previous studies in other solid tumors. The explanations might be that PLK4 high expression might favor cell proliferation by regulating cell cycle and interfere with cell response to DNA damage, which promoted tumor progression.4 Thus, PLK4 high expression was correlated with unfavorable clinicopathological features in NSCLC patients. Besides, inhibition of PLK4 was previously shown to cause defects in cell mitosis and induce cell death in NSCLC cells via regulating centriole duplication, which implied that PLK4 suppressed cell apoptosis and contributed to tumor growth in NSCLC.6 Thus, PLK4 was associated with greater tumor size and higher TNM stage in NSCLC patients.

Clinically, PLK4 is shown to predict prognosis of cancer patients. PLK4 is overexpressed in colorectal cancer tissues compared with adjacent normal intestinal mucosa and is of potential to predict poor prognosis in colorectal cancer patients.12 High PLK4 expression is correlated with poor DFS and OS in breast cancer patients; meanwhile, it is a negative predictive factor for treatment response to taxane‐based neoadjuvant chemotherapy.10 Study on epithelial ovarian cancer also exhibits that PLK4 is positively correlated with LIN28 homolog A, with which the co‐expression of PLK4 is associated with poor prognosis of epithelial ovarian cancer patients.13 In general, a strong correlation has been observed between PLK4 and survival, which indicates the prognostic significance of PLK4 in different cancers, and however, the detailed association of PLK4 with prognosis in NSCLC patients still needs further investigation. Therefore, we compared DFS and OS between patients with PLK4 high and low expression and observed that PLK4 high expression was correlated with poor survival; furthermore, Cox's proportional hazard regression model revealed that PLK4 was an independent predictive factor for shorter DFS and OS in NSCLC patients. This could be due to that: (a) PLK4, which regulated centrosome duplication, might promote chromosomal instability and carcinogenic cell activity, thereby facilitate the progression of NSCLC and contribute to poor survival in NSCLC patients. (b) PLK4 was correlated with larger tumor size, LYN metastasis, and higher TNM stage, which were predictors for poor prognosis in NSCLC patients. In addition, as previously shown in breast cancer, PLK4 was responsible for drug resistance to taxane‐based chemotherapy, which was likely to be resulted from α‐tubulin and β‐tubulin mutations via PLK4‐γ‐tubulin axis.10 Therefore, it could be speculated that PLK4 might interfere with treatment response in NSCLC patients by inducing drug resistance, while further investigation was needed for validation.

This study revealed the correlation of PLK4 with clinicopathological features and prognosis in NSCLC patients, whereas the detailed mechanism of PLK4 in pathology of NSCLC needed further exploration. In addition, since only NSCLC patients at TNM stage I ~ III were recruited and patients were restricted to one hospital, there might be bias in sample selection. Furthermore, patients included in this study were with initial NSCLC, while the implication of PLK4 in recurrent NSCLC was not investigated. Therefore, additional studies are needed to enforce our findings.

In summary, PLK4, a prime member of protein kinase family, presents correlation with unfavorable tumor features and survival in NSCLC patients. This preliminary finding illuminates the clinical implication of PLK4 in NSCLC, while further works are surely necessary to explicit its value in surveillance and treatment of NSCLC.