PDK2: a novel diagnostic and prognostic biomarker for liver 2 cancer

Background: Liver cancer (LC) is a common malignancy with very high morbidity. Pyruvate dehydrogenase kinases (PDKs) are regulators of mitochondrial pyruvate dehydrogenase complexes (PDCs) and play an important role in regulating cellular energy metabolism. In this study, The Cancer Genome Atlas (TCGA) database was used to analyze the expression of PDK2 mRNA in LC, and to explore the value of PDK2 in the diagnosis and prognosis of LC. Methods: The TCGA database, containing the clinical data of 373 LC patients, includes 27 information on PDK2 expression values. The receiver operating characteristic (ROC) curve of PDK2 was drawn to evaluate its diagnostic ability. Patients were divided into PDK2 high- and 29 low-expressing groups by threshold levels. The Chi-square test was used to evaluate the 30 correlation between PDK2 levels and clinicopathological characteristics. The Kaplan-Meier 31 estimator and Cox regression analysis were performed to assess the effect of PDK2 levels on 32 survival outcomes. 33 Results: PDK2 expression in LC tissue was lower than that in normal liver tissues. According 34 to the area under the curve (AUC) value calculated by ROC, PDK2 has a considerable diagnostic value for LC prognosis. The decreased expression of PDK2 is associated with 36 clinical parameters, such as histologic grade ( P =0.0001), radiation therapy ( P =0.0490), vital status ( P =0.0240), and overall survival (OS) ( P =0.0222). Multivariate analysis shows that decreased PDK2 level is an independent risk factor for predicting poor prognosis in LC. Conclusions: PDK2 has a significant impact on the prognosis of LC and is a potential biomarker for the diagnosis and prognosis of LC.


45
Liver cancer (LC) is one of the most common malignancies and has attracted worldwide 46 attention. The morbidity of LC is higher than that of any other cancer, irrespective of gender [1]. LC is characterized by occult onset and rapid progression, together with a high degree of 48 invasiveness, metastasis and recurrence, with the result that the overall prognosis is extremely 49 poor [2]. Studies have shown that the five-year survival rate of LC is more than 70% if it is 50 diagnosed early, but is reduced to 10% if diagnosed at an advanced stage [3,4]. Since most 51 diagnoses occur in the middle or late stages, patients are only able to choose between 52 interventional therapy, ablation therapy, radiation therapy and chemotherapy. Surgical 53 resection is the choice of less than 20% of LC patients [5]. Therefore, there is a critical need to phosphorylation. Therefore, an increase or decrease in PDKs expression will have a significant 64 impact on glucose metabolism and cell bioenergetics [8]. Current research has found that PDKs 65 are closely related to cancer cell metabolism, cancer drug resistance and cell steatosis [9][10][11]. 66 PDKs include four subtypes, PDK1, PDK2, PDK3, and PDK4. Among them, PDK2 is the 67 most widely distributed and is expressed in almost all tissues [12]. At present, it is generally 68 believed that PDK2 is closely related to proto-oncogenes, transcription factors and growth 69 factors in tumorigenesis and development [13] 99 Boxplots were prepared to examine the correlation between PDK2 expression and different 100 histological features ( Figure 1). The expression of PDK2 mRNA in LC tissue was lower than 101 that in normal liver tissue (P=1.1e -10 ). In addition, PDK2 expression was clearly related to 102 histological type (P=0.0019), histologic grade (P=0.00043) and radiation therapy (P=0.0089).

104
The diagnostic value of PDK2 in LC 105 The ROC curve shows that the AUC value of all LC patients is 0.781, suggesting that PDK2 106 has considerable diagnostic ability ( Figure 2). The subsequent subgroup analysis showed that 107 the AUC of stages I, II, III, and IV were 0.774, 0.760, 0.816, and 0.844, respectively, suggesting 108 that the diagnostic sensitivity and specificity were acceptable. These results indicate that PDK2 109 has a high degree of accuracy in predicting the prognosis of LC.

111
Based on the threshold expression value, all patients were divided into high-and low-112 expressing groups and the correlations between PDK2 expression and clinical parameters were 113 examined ( Table 1). The Chi-square test indicated that PDK2 expression was clearly correlated 114 with histologic grade (P = 0.0001), radiation therapy (P=0.0490), vital status (P=0.0240) and OS (P =0.0222). 117 The Kaplan-Meier curve was drawn to determine the correlation between PDK2 levels and 118 patient survival (Figure 3). Patients were grouped as above. The statistical significance of the 119 differences was calculated by log-rank test. The results showed that PDK2 levels were 120 significantly associated with OS (P <0.0001) and RFS (P=0.0032). It is not difficult to conclude 121 that decreased PDK2 levels can predict unsatisfactory OS and RFS in all patients. Next, the 122 same method was used to explore the effect of PDK2 expression on patient prognosis in 123 subgroup analysis. 125 Subgroup analysis showed that PDK2 levels have remarkable predictive value for OS ( Figure   126 4). This was seen for PDK2 levels in relation to early clinical stage (I/II) (P <0.0001), early 127 and late histologic stage (G1/G2 and G3/G4, P=0.00032 and P =0.0027, respectively), gender 128 (P=2e -04 for males and P=0.0025 for females) and older and younger age (P=0.00024 and 129 P=0.0013, respectively). In these subgroups, decreased hepatic PDK2 levels predict 130 unsatisfactory OS. 132 Subgroup analysis showed that PDK2 levels in early clinical stages (I/II) (P=0.00097), early ]. This study confirmed that hepatic PDK2 levels were lower in cancerous tissue than in 152 normal tissue. PDK2 mRNA expression was found to be related to histologic grade, radiation 153 therapy, vital status, and OS of LC patients. We also found that PDK2 plays an important role

184
To our knowledge, this is the first study to investigate the prognostic significance of PDK2 185 in LC. We report that low PDK2 expression may be an independent risk factor for poor  192 In summary, we found that PDK2 levels were down-regulated in LC and were related to several 193 clinicopathological features of patients. Patients with low PDK2 expression in LC have 194 unsatisfactory OS and RFS. Low expression of PDK2 was also found to be an independent risk 195 factor for poor LC prognosis. We conclude that PDK2 has a significant impact on the prognosis 196 of LC and is a potential biomarker for the diagnosis and prognosis of LC. In future study, we 197 plan to conduct complex experiments to further explore the mechanisms behind these findings 198 and the application of PDK2 levels in predicting the prognosis of LC patients in the clinic.   207 There is no need for ethical review as all information is in the public domain. 209 All data published here are under the consent for publication.

210
Availability of data and materials 211 All data were obtained from The Cancer Genome Atlas (TCGA) database 212 (https://cancergenome.nih.gov). 214 The authors declare that they have no competing interests.