1 Introduction
Ovarian cancer (OC) is 1 of the 3 most common gynecological tumors in women. It is the fifth most common cause of cancer death in women worldwide.[1] The prodromal manifestations of OC are always nonspecific, which makes it difficult to distinguish from other carcinomatosis.[2] Additionally, the current screening strategies for the diagnosis of early-stage OC, including transvaginal ultrasound, computed tomography, detection of tumor marker CA125, and detection of BRCA gene mutation, are apparently ineffective in reducing the mortality rate of OC.[2,3] Consequently, early stage diagnosis of OC is uncommon leading to a poor prognosis with a 5-year OS rate of < 30%.[4,5] Thus, there is an urgent requirement to detect a biomarker for tumor prognosis.
The mechanism of pathogenesis of OC is complex. Recently, a study found that the ciliated epithelial cells in the fallopian tube underwent periodic proliferation and differentiation during the menstrual cycle, which could be 1 of the mechanisms involved in the pathogenesis of OC.[6] In recent years, researchers have become interested in the role of cilia in various human diseases, including tumorigenesis. It has been shown that primary cilia are involved in cell-cycle regulation and further implicated in tumor progression.[7,8] Adenylate kinase 7 (AK7), a cytosolic isoform of adenylate kinase (AK) isoenzymes located on chromosome 14q32, is known to have a tissue-restricted expression pattern and is expressed in cilia-rich tissues in the epithelium.[9]
However, there is no direct evidence showing an association between AK7 and tumor progression and prognosis, possibly via the regulation of cilia function. In this study, we retrospectively analyzed the data from The Cancer Genome Atlas (TCGA) cohort and assessed the correlation between AK7 levels and clinicopathological symptoms of OC to evaluate the prognostic value (PV) of AK7 in OC. We also performed GSEA to explore relevant signaling pathways.
2 Materials and methods
2.1 Data extraction from the TCGA database
We extracted cancerous ovarian tissues (n = 308; OC group) and normal ovarian tissues (n = 88; CON group) from the TCGA database to determine AK7 levels and the PV by RNAseq (Illumina HiSeq). The high and low AK7 expression groups were classified based on the median value of AK7.
2.2 Statistical analysis
Statistical analysis was performed using R software (version 3.5.2). The c2 test assessed the correlation between AK7 levels and the clinical symptoms of OC. Kaplan–Meier curve was used to compare the OS between the AK7 expression groups. The independent PV of AK7 in OC was determined via Cox regression analyses. A value of P < .05 implied statistical significance.
GSEA . In GSEA, target genes are ranked according to predetermined gene sets based on the differential expression between the 2 sample groups, followed by the assessment of the position of the predetermined gene sets in the sorting table.[10] Here, we used GSEA 3.0 for patient data analysis. Permutation analysis was done to obtain normalized enrichment score (NES).
2.3 Ethics approval
This study did not require ethics approval since all clinical data were from public databases.
3 Results
3.1 Patient characteristics
Table 1 shows the demographic, clinical symptoms, and gene expression data of patients in the OC group.
Table 1 -
Demographic and clinical characteristics of TCGA cohort.
Characteristics
Numbers of cases
Age
<55
113 (36.69)
>=55
195 (63.31)
Subdivision
NA
17 (5.52)
Bilateral
212 (68.83)
Left
37 (12.01)
Right
42 (13.64)
Stage
NA
2 (0.65)
I
1 (0.32)
II
22 (7.14)
III
245 (79.55)
IV
38 (12.34)
Longest dimension
Large
124 (46.1)
Small
145 (53.9)
Lymphatic invasion
NA
180 (58.44)
NO
44 (14.29)
YES
84 (27.27)
Histologic grade
NA
2 (0.65)
G1
1 (0.32)
G2
37 (12.01)
G3
261 (84.74)
G4
1 (0.32)
GB
2 (0.65)
GX
4 (1.3)
New type
NA
145 (47.08)
Locoregional
4 (1.3)
Metastatic
1 (0.32)
Progression
12 (3.9)
Recurrence
146 (47.4)
Sample type
Primary Tumor
303 (98.38)
Recurrent Tumor
5 (1.62)
Vital status
Deceased
184 (59.74)
Living
124 (40.26)
AK7
High
154 (50)
Low
154 (50)
NA = not available, AK7 = Adenylate kinase, TCGA = the Cancer Genome Atlas .
3.2 AK7 expression and association with clinicopathological variables
We found a substantially downregulated AK7 levels in the OC group than the CON group (P < .05). Furthermore, there was a marked difference in AK7 levels based on patient age (Fig. 1 ). Patients with OC were classified into high and low AK7 expression groups. Table 2 describes their clinicopathological parameters and OS. We found that low AK7 levels were correlated with patient age (P = .0093).
Figure 1: AK7 expression in OC. Boxplots show the difference in AK7 expression grouped by stage, histological grade, new type, vital status, subdivision, lymphatic invasion, and patient age. AK7 = adenylate kinase 7.
Table 2 -
Association of AK7 mRNA expression in ovarian cancer tissues with clinicopathologic variables.
AK7 mRNA expression
Parameter
Variable
N
High
%
Low
%
χ 2
P value
Age
<55
113
68
(44.16)
45
(29.22)
6.7652
.0093
>=55
195
86
(55.84)
109
(70.78)
Subdivision
Bilateral
212
111
(74.5)
101
(71.13)
4.0109
.1346
Left
37
22
(14.77)
15
(10.56)
Right
42
16
(10.74)
26
(18.31)
Stage
I
1
1
(0.65)
0
(0)
2.2183
.5284
II
22
12
(7.79)
10
(6.58)
III
245
125
(81.17)
120
(78.95)
IV
38
16
(10.39)
22
(14.47)
Longest dimension
Large
124
60
(43.8)
64
(48.48)
0.4212
.5164
Small
145
77
(56.2)
68
(51.52)
Lymphatic invasion
No
44
18
(27.27)
26
(41.94)
2.4315
.1189
Yes
84
48
(72.73)
36
(58.06)
Histologic grade
G1
1
0
(0)
1
(0.66)
5.5678
.3506
G2
37
24
(15.58)
13
(8.55)
G3
261
126
(81.82)
135
(88.82)
G4
1
1
(0.65)
0
(0)
GB
2
1
(0.65)
1
(0.66)
GX
4
2
(1.3)
2
(1.32)
New type
Locoregional
4
4
(4.82)
0
(0)
5.3073
.1506
Metastatic
1
0
(0)
1
(1.25)
Progression
12
5
(6.02)
7
(8.75)
Recurrence
146
74
(89.16)
72
(90)
Sample type
Primary Tumor
303
151
(98.05)
152
(98.7)
0
1
Recurrent Tumor
5
3
(1.95)
2
(1.3)
Vital status
Deceased
184
92
(59.74)
92
(59.74)
0
1
Living
124
62
(40.26)
62
(40.26)
AK7 = Adenylate kinase 7, N = number.
3.3 Low AK7 expression as an independent prognostic factor for poor OS
Low AK7 levels were related to poor OS (P = .019; Fig. 2 ), especially in those with late-stage OC (P = .014) but not early-stage OC (P = .62); G3/G4 grade (P = .011) but not G1/G2 grade (P = .97); old age (P = .018) but not young age (P = .83; Fig. 2 ). The results of the univariate analysis showed that patient age and AK7 levels were related to poor OS (Table 3 ). Further multivariate analysis estimated the independent PV of low AK7 levels for poor OS of OC (HR: 1.34, 95% CI: 1-1.8, P = .048; Table 3 ).
Figure 2: The PV of AK7 in patients with OC. Kaplan-Meier curves for the survival of patients with OC based on AK7 expression in cancerous ovarian tissues. AK7 = adenylate kinase 7.
Table 3 -
Univariate and multivariate analyses of overall survival in patients with ovarian cancer.
Univariate analysis
Multivariate analysis
Parameters
Hazard Ratio
CI 95
P value
Hazard Ratio
CI 95
P value
Age
1.63
1.19–2.24
.003
1.57
1.14–2.16
.005
Subdivision
0.84
0.67–1.04
.101
Stage
1.09
0.8–1.5
.581
Longest dimension
1.12
0.82–1.52
.485
Lymphatic invasion
1.02
0.85–1.23
.798
Histologic grade
1.12
0.88–1.42
.349
New type
0.99
0.63–1.55
.951
Sample type
0.43
0.11–1.73
.235
AK7
1.41
1.06–1.89
.02
1.34
1–1.8
.048
AK7 = adenylate kinase 7.
3.4 AK7-related signaling pathway
The results of GSEA showed a marked difference in the enrichment of MSigDB Collection (NOM P < .05; Table 4 ). The essential signaling pathways, including EMT, apical junction, TGF-b signaling, UV response, and myogenesis, were chosen based on NES. These signaling pathways were all enriched in low AK7 expression phenotype (Table 4 and Fig. 3 ).
Table 4 -
GSEA enrichment plot in low ABCB9 phenotype.
Gene set
ES
NES
NOM P -value
HALLMARK_epithelial mesenchymal transition
0.64390194
1.8245728
.019880716
HALLMARK_UV response
0.41891807
1.5772356
.029940119
HALLMARK_TGF-beta signaling
0.48144048
1.5725749
.03952569
HALLMARK_myogenesis
0.43896723
1.5351363
.043052837
HALLMARK_apical junction
0.3967964
1.5181613
.042
ES = enrichment score, NES = normalized enrichment score, NOM = nominal.
Figure 3: Enrichment plots from gene set enrichment analysis.
4 Discussion
Several complex factors interact to influence the pathogenesis and progression of OC, including genetic factors, reproductive factors, environmental factors, and so on.[11] Among these pathogenic factors, cilia were found to be involved in ovarian tumorigenesis. Cilia are sensory and motor organelles extending from the cellular surface and have long been considered as a degraded organelle.[12] Accumulating evidence has shown that primary cilia structure dysfunction may cause a series of multisystemic developmental disorders known as ciliopathies and multifactorial human diseases, including cancer.[13,14] It has been shown that primary cilia have a dual role in regulating tumorigenesis, whereas the loss of primary cilia and abnormally activated cilia regulation of hedgehog signaling pathway has been associated with the progression and prognosis of various tumors, including pancreas, breast, prostate, ovarian cancer, and so on.[15–17] Shpak performed bioinformatics analysis to study the gene expression patterns of cilia in OC and identified 354 cilia genes abnormally expressed in OC tissues, indicating an important role of ciliary disruption in the development of OC.[14] However, there is still a lack of in vivo experimental data to verify this hypothesis.
The maintenance of proper cilia structure and function requires ATP hydrolysis, which is done by the AK family.[9] Based on recent studies, AK7, a cytosolic human AK isoform, was found to be associated with ciliary homeostasis. The mutation of AK7 in primary ciliary dyskinesia (PCD) was found in both humans and murine species.[18,19] However, there is still a lack of research on the association between AK7 and human cancer. It has been speculated that ciliary dysfunction mediates the effects of AK7 on the genesis, progress, and prognosis of different types of cancer, including OC.
Here, we detected a substantially downregulated expression of AK7 in cancerous ovarian tissues than with normal ovarian tissues, which was also associated with patient age based on the analysis of high and low AK7 groups. Milara identified a downregulated AK7 expression at both RNA and protein levels and found that it was associated with PCD.[18] Consistently, Angeles observed the phenotypes of PCD in AK7-deficient mice.[19] These findings are consistent with our hypothesis that lower AK7 expression may cause ciliary structure disorder or dysfunction, further affecting the progression and prognosis of OC.
Also, patients who expressed lower levels of AK7 in OC tissues were more prone to have a poorer prognosis, particularly those with late-stage disease, G3/G4 grade, and old age. Also, late-stage diagnosis is related to a poor prognosis. Thus, AK7 might act as a novel indicator of OC prognosis. Moreover, the onset age of epithelial ovarian cancer, the most common and malignant type of OC, was 62 years and older patients with lower AK7 levels had a worse prognosis, suggesting a valuable role of AK7 in the diagnosis of epithelial ovarian cancer.[2] Cox regression analysis revealed that low AK7 levels had a significant PV in patients with OC.
Our investigations are focused on the identification of novel biomarkers of cancers to track their onset and development. This is the first study to detect a relationship between AK7 levels and OS in patients with OC. The results of our study have shown that downregulated AK7 levels are related to poor OS in OC and has an independent PV in OC. These results provide a new insight that AK7 plays a valuable role in the prognosis of OC, which might be mediated by ciliary structure disorder and function and lays a foundation for further investigation.
Author contributions
Conceptualization : XZ, LZ.
Data curation : LZ, YL.
Formal analysis : XZ, YJ.
Funding acquisition : LZ.
Investigation : LZ, XZ, YJ, ZY.
Methodology : YL, YJ.
Project administration : XZ, YL.
Resources : YL, ZY, YG, YZ.
Software : YJ.
Supervision : YZ.
Validation : YZ, YG.
Visualization : ZY, YG.
Writing – original draft : XZ.
Writing – review & editing : LZ.
References
[1]. Siegel RL, Miller KD, Jemal A. Cancer statistics, 2018. CA Cancer J Clin 2018;68:7–30.
[2]. Orr B, Edwards RP. Diagnosis and treatment of ovarian cancer. Hematol Oncol Clin North Am 2018;32:943–64.
[3]. Dorigo O, Berek JS. Personalizing CA125 levels for ovarian cancer screening. Cancer Prev Res (Phila) 2011;4:1356–9.
[4]. Lachmann P. Cancer survival in Australia, Canada, Denmark, Norway, Sweden, and the UK. Lancet 2011;377:1149.
[5]. Vaughan S, Coward JI, Bast RC Jr, et al. Rethinking ovarian cancer: recommendations for improving outcomes. Nat Rev Cancer 2011;11:719–25.
[6]. Rohozinski J, Diaz-Arrastia C, Edwards CL. Do some epithelial ovarian cancers originate from a fallopian tube ciliate cell lineage? Med Hypotheses 2017;107:16–21.
[7]. Radford R, Slattery C, Jennings P, et al. Carcinogens induce loss of the primary cilium in human renal proximal tubular epithelial cells independently of effects on the cell cycle. Am J Physiol Renal Physiol 2012;302:F905–16.
[8]. Gradilone SA, Pisarello MJL, LaRusso NF. Primary cilia in tumor biology: the primary cilium as a therapeutic target in cholangiocarcinoma. Curr Drug Targets 2017;18:958–63.
[9]. Panayiotou C, Solaroli N, Karlsson A. The many isoforms of human adenylate kinases. Int J Biochem Cell Biol 2014;49:75–83.
[10]. Subramanian A, Tamayo P, Mootha VK, et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles. P Natl Acad Sci USA 2005;102:15545–50.
[11]. Meinhold-Heerlein I, Hauptmann S. The heterogeneity of ovarian cancer. Arch Gynecol Obstet 2014;289:237–9.
[12]. Badano JL, Mitsuma N, Beales PL, et al. The ciliopathies: an emerging class of human genetic disorders. Annu Rev Genomics Hum Genet 2006;7:125–48.
[13]. Yang Y, Ran J, Liu M, et al. CYLD mediates ciliogenesis in multiple organs by deubiquitinating Cep70 and inactivating HDAC6. Cell Res 2014;24:1342–53.
[14]. Shpak M, Goldberg MM, Cowperthwaite MC. Cilia gene expression patterns in cancer. Cancer Genomics Proteomics 2014;11:13–24.
[15]. Han YG, Kim HJ, Dlugosz AA, et al. Dual and opposing roles of primary cilia in medulloblastoma development. Nat Med 2009;15:1062–5.
[16]. Cervantes S, Lau J, Cano DA, et al. Primary cilia regulate Gli/Hedgehog activation in pancreas. P Natl Acad Sci USA 2010;107:10109–14.
[17]. Hassounah NB, Bunch TA, McDermott KM. Molecular pathways: the role of primary cilia in cancer progression and therapeutics with a focus on Hedgehog signaling. Clin Cancer Res 2012;18:2429–35.
[18]. Mata M, Lluch-Estelles J, Armengot M, et al. New adenylate kinase 7 (AK7) mutation in primary ciliary dyskinesia. Am J Rhinol Allergy 2012;26:260–4.
[19]. Fernandez-Gonzalez A, Kourembanas S, Wyatt TA, et al. Mutation of murine adenylate kinase 7 underlies a primary ciliary dyskinesia phenotype. Am J Respir Cell Mol Biol 2009;40:305–13.