Differential expression and diagnostic significance of P53, MutS homologs 2, tropomyosin‐4 in alpha‐fetoprotein‐negative hepatocellular carcinoma

Abstract Background Current study aimed to explore the value of P53, MutS homologs 2 (MSH2), and tropomyosin‐4 (Tm‐4) combined with inflammatory factors, life‐history traits in the differential diagnosis of alpha‐fetoprotein‐negative hepatocellular carcinoma (AFP‐Negative HCC). Methods A testing cohort including 280 AFP‐Negative HCC patients and 300 controls was included. Three external validation cohorts from 3 centers were used to assess the novel logistic regression model including 400 AFP‐Negative HCC patients and 400 controls. Results Compared with the control group, the levels of P53, MSH2, and Tm‐4 protein in si‐P53 group, si‐MSH2 group, and si‐Tm‐4 group were significantly reduced (P < .05). The P53, MSH2, Tm‐4, neutrophil to lymphocyte ratio (NLR), monocytes to lymphocyte ratio (MLR), hypersensitive C‐reactive protein (hs‐CRP), tumor necrosis factor‐α (TNF‐α), interleukin 6 (IL‐6) levels, and the smoking, drinking, and occupational exposure to chemicals rates in patients were significantly higher than those in controls (P < .05). ROC analyses showed that the area under curve (AUC) of NLR, MLR, hs‐CRP, TNF‐α, IL‐6, P53, MSH2, Tm‐4, drinking, smoking, and occupational exposure to chemicals were 0.798, 0.803, 0.560, 0.644, 0.808, 0.681, 0.830, 0.694, 0.582, 0.581, and 0.567, respectively. A novel logistic regression model was built and has a high value in identifying AFP‐Negative HCC with AUC of 0.917, sensitivity of 85.2%, and specificity of 88.3%. In the validation cohorts, this model also showed good diagnostic efficiency (AUC = 0.898 with Dazu Branch cohort, AUC = 0.924 with Jinshan Branch cohort, and AUC = 0.907 with Liangping Branch cohort). Conclusion Current model has potential significance for the noninvasive diagnosis of AFP‐Negative HCC.


| INTRODUC TI ON
Hepatocellular carcinoma (HCC) is the most common hepatic malignancy and has no obvious symptoms even in the middle stage. 1,2 Currently, the main mode of HCC screening is alpha-fetoprotein (AFP). 3 However, studies have shown that elevated serum AFP is observed in only 60% patients. 3 Proteomics is an important part of post-genomic research. With the help of proteomics research technology, the pathogenesis of tumors can be further revealed at the protein level, and tumor-specific markers and specific antigens can be screened and identified. Recent studies have pointed out that abnormal expression of P53, MutS homologs 2 (MSH2), and tropomyosin-4 (Tm-4) was closely related to the occurrence and metastasis of various cancers. [4][5][6][7][8] P53 is considered to be an important tumor suppressor protein, but reports have pointed out that the P53 gene in many cancer cells will mutate and lose cellular gene repair function. 4 Wang et al 4 pointed out that P53 gene was significantly increased in bladder cancer tissues, and high expression of P53 protein was associated with poor prognosis in patients. Recently, researchers found that the oncogene long non-coding RNA H19 can play a competitive endogenous ribonucleic acid to inhibit miR-29 expression, thereby reducing the inhibitory function of miR-29 on P53 gene expression and promoting the cancer metastasis. 5 MSH2 is an important type of mismatch repair protein. Hinrichsen et al 6  It is also well known that inflammation causes many cancers, especially liver cancer. 9,10 HCC is closely related to inflammation and usually occurs on the basis of chronic liver injury. 11,12 In this study, we developed a novel logistic regression model based on P53, MSH2, and Tm-4 combined with inflammatory factors and life-history traits to easily estimate AFP-Negative HCC patients and validated this model in 3 external validation cohorts.

| Ethical approval
Informed patient consent was obtained in accordance with ethics committees of The First Affiliated Hospital of Chongqing Medical University (EA20150017).

| Biomarkers detection
Biomarkers tested in this study include inflammatory factors (neu-

| Western blot
The protein concentration of each sample was measured using the bicinchoninic acid (BCA) protein concentration measurement kit. A12004-1) was used, and bands were measured using an enhanced chemiluminescence (ECL) kit. SPSS 19.0 was performed. Data were presented as mean ± standard deviation (SD), or median [interquartile range (IQR)], or number of cases (%). Receiver operating characteristic (ROC) curve analysis was also used. The formula for calculating the number of diagnostic test samples is n = ( 2 ∕ ) 2 (1 − P)P where α is set to .05, δ is set to .05, and the P is set to 90%. P < .05 means the difference is statistically significant.

| Specific verification of anti-P53, anti-MSH2, and anti-Tm-4 protein antibodies and levels of P53, MSH2, and Tm-4 protein in AFP-Negative HCC and adjacent tissues
The results of specific verification of anti-P53, anti-MSH2, and anti-Tm-4 protein antibodies are shown in Figure 2A. Compared with the control group, the levels of P53, MSH2, and Tm-4 protein in si-P53 group, si-MSH2 group, and si-Tm-4 group were significantly reduced (P < .05). It is suggested that each antibody has high specificity, does not cross-link with other proteins, and meets the experimental requirements. Western blot results showed that P53, MSH2, and Tm-4 protein levels in cancer tissues were significantly higher than those in adjacent tissues (P < .05, Figure 2B-E).

| Expression of P53, MSH2, Tm-4, inflammatory factors, and life-history traits in testing cohort
There were 580 subjects in the testing cohort. The subjects' P53, MSH2, and Tm-4 protein levels, clinical information, inflammatory factors, and life-history traits are presented in Table 1. The P53, MSH2, Tm-4, NLR, MLR, hs-CRP, TNF-α, IL-6 levels, and the smoking, drinking, and occupational exposure to chemicals rates in the AFP-Negative HCC patients were significantly higher than those in controls (P < .05, Figure 3A-H).

| Logistic regression analyses and ROC analyses in testing cohort
Logistic regression analyses results are presented in Table 2. NLR  Table S1).

| A novel logistic regression model for AFP-Negative HCC based on the testing cohort
The regression model for AFP-Negative HCC was: Logit  Table S2. The estimated probability was .316 (Figure 4 and Table S1).

| Multicenter validation of the logistic regression model
The validity of our logistic regression model was assessed in 3 external validation cohorts. The subjects' data are presented in Table S3,   Table S4, and Table S5. There were 800 subjects in the validation cohorts.
In the Dazu Branch cohort, the sensitivity/specificity for AFP-  Abbreviations: hs-CRP, hypersensitive C-reactive protein; IL-6, interleukin 6; MLR, monocytes to lymphocyte ratio; MSH2, MutS homologs 2; NLR, neutrophil to lymphocyte ratio; PLR, platelet to lymphocyte ratio; Tm-4, tropomyosin-4; TNF-α, tumor necrosis factor-α. has high specificity for the diagnosis of primary liver cancer. 24,25 Previous studies have shown that MSH2 is closely related to the development of various liver diseases, especially HCC. 26,27 It has been reported in China that the expression of MSH2 is gradually up-regulated during the development of liver disease. MSH2 is hardly expressed in normal liver cells, but may increase slightly in the presence of acute inflammation or more severe liver fibrosis in the liver, and the expression of MSH2 may rapidly increase and reach a high level when it develops into HCC. 28 Therefore, the level of MSH2 is considered to be closely related to the process of HCC Therefore, we built a novel logistic regression model containing NLR, MLR, hs-CRP, TNF-α, IL-6, P53, MSH2, Tm-4, drinking, smoking, and occupational exposure to chemicals. It showed higher diagnostic efficiency (AUC = 0.917, sensitivity = 85.2%, specificity = 88.3%) than any single parameter. Of course, the model that combined different parameters has been reported. 30 However, we found that current model has better AUC (0.917) than the model of combing AFP and DCP (AUC = 0.910). 30  In summary, current model including P53, MSH2, Tm-4, inflammatory factors, and life-history traits might more effectively improve the diagnostic efficiency of AFP-Negative HCC.

ACK N OWLED G EM ENTS
None.

AUTH O R CO NTR I B UTI O N S
AH researched literature and conceived the study. XG and AH were involved in gaining ethical approval, patient recruitment, and data analysis. XG wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

E TH I C A L A PPROVA L
The ethics committee of The First Affiliated Hospital of Chongqing Medical University approved this study (EA20150017).