1 INTRODUCTION

Adrenocortical carcinoma (ACC) is a rare and aggressive endocrine malignancy with high risk of relapse, poor survival and limited treatment options. The Surveillance, Epidemiology and End Results (SEER) database estimates that the annual incidence rate of ACC is approximately 0.72 per million cancer cases, resulting in 0.2% of all cancer deaths in the United States.¹ However, ACC shows highly aggressive biological behaviour with less than 35% of patients surviving 5 years after initial diagnosis.² Therefore, appropriate treatment is extremely important. The current preferred treatment of ACC is based on surgical resection of the primary tumour that is usually the first and most effective therapeutic strategy.^3-5 Currently, there are very few drugs to treat this disease and mitotane remains the only medication approved by the US Food and Drug Administration for ACC treatment.⁶ Thus, new treatment options and drug targets are urgently needed, especially for clinical management of patients with ACC who are resistant to mitotane.

Proteomics is a powerful tool for detecting unknown protein species, exploring absolute quantified protein abundance, and identifying biomarkers for pathogenic process.^{7, 8} Proteomics has been used widely to explore biomarkers for various diseases⁹ and the latest developments in proteomics have made it possible to conduct more comprehensive examinations of protein biomarkers in various cancers.¹⁰ For instance, Bouchal et al¹¹ used transcriptome and proteomic analysis to identify potential biomarkers associated with metastatic breast cancer, and several proteomic studies have focused on identifying new diagnostic biomarkers in patients with prostate cancer.^{12, 13} Thus far, very few studies have used a large-scale sequencing proteomic approach to identify potential protein biomarkers for ACC.¹⁴

Bioinformatics studies have generated large amounts of complex biological data through combinations of computer science, information technology and biology. For example, The Cancer Proteome Atlas (TCPA) database provides researchers with reverse-phase protein array (RPPA) data.¹⁵ The RPPA technique is a powerful proteomic approach for economical, sensitive and high-throughput evaluation of sizable numbers of selected protein markers, which made it possible to explore protein biomarkers using bioinformatics.^{16, 17} Because there is a big difference between adult patients and child patients with ACC, in this study, we focused only on adult patients. This study constitutes the first large-scale proteomic analysis combined with transcriptome data to describe the protein landscape of ACC in adult patients.

To explore novel protein biomarkers of potential prognostic value and develop a protein-derived predictive model in adult patients with ACC, we analysed the survival of proteins and constructed an integrated prognosis-related proteins model on risk assessment. Gene expression profiles also were analysed to reveal the underlying biological interaction networks. The goal of this study was to provide potential novel therapeutic targets and a high performing prognostic predictive model for clinical management of adult ACC.

2 MATERIALS AND METHODS

3 RESULTS

In this work, we aimed to explore new prognostic biomarkers for adult patients with ACC using proteomics and transcriptomics data. A flow chart of the methods used in this study is given in Figure S1.

3.1 Selection for candidate proteins with significant prognostic value

From the volcano plot (Figure 1A), 42 candidate protein biomarkers with P-values < 0.05 in both the Kaplan-Meier analysis and univariate Cox regression analysis were selected and are listed in Table 4. The Lasso Cox regression results for the selected proteins are shown in Figure 1B, C.

Table 4. Kaplan-Meier analysis and univariate Cox regression analysis of proteins (both P-value < .05)

Protein	P value (KM)	P value (unicox)	HR
P27_pT157	.004	.000	0 (0-0.001)
ERALPHA_pS118	.007	.001	0 (0-0.007)
NRAS	.002	.002	0 (0-0.041)
X1433BETA	.018	.022	0.001 (0-0.34)
CMET	.001	.001	0.003 (0-0.077)
SNAIL	.020	.002	0.009 (0-0.183)
CD31	.039	.004	0.01 (0-0.23)
ARAF_pS299	.001	.003	0.02 (0.002-0.254)
PRDX1	.004	.002	0.021 (0.002-0.238)
XBP1	.006	.003	0.026 (0.002-0.291)
P70S6K_pT389	.003	.004	0.027 (0.002-0.311)
PKCDELTA_pS664	.023	.003	0.036 (0.004-0.324)
PARPCLEAVED	.044	.034	0.039 (0.002-0.781)
CIAP	.007	.010	0.055 (0.006-0.498)
P53	.012	.025	0.071 (0.007-0.713)
RAB25	.006	.010	0.075 (0.01-0.542)
JAB1	.007	.001	0.08 (0.018-0.347)
ECADHERIN	.006	.004	0.133 (0.033-0.527)
INPP4B	.002	.019	0.137 (0.026-0.722)
ERALPHA	.004	.050	0.164 (0.027-0.999)
SMAC	.003	.002	0.188 (0.065-0.543)
PKCALPHA_pS657	.024	.001	0.281 (0.133-0.594)
EPPK1	.027	.011	0.343 (0.151-0.779)
FASN	.039	.015	1.943 (1.139-3.316)
CYCLINB1	.000	.001	2.198 (1.407-3.434)
PEA15	.028	.033	2.209 (1.066-4.577)
TFRC	.001	.000	2.771 (1.681-4.567)
YAP_pS127	.014	.030	2.984 (1.109-8.031)
PARP1	.002	.002	3.23 (1.553-6.72)
ERK2	.000	.000	5.629 (2.168-14.618)
RBM15	.008	.008	6.531 (1.64-26.011)
FIBRONECTIN	.001	.000	7.007 (2.517-19.511)
DVL3	.002	.010	10.978 (1.754-68.72)
IGF1R_pY1135Y1136	.012	.047	14.431 (1.033-201.583)
TSC1	.001	.001	16.639 (2.965-93.39)
JAK2	.042	.043	18.739 (1.098-319.658)
BCL2A1	.003	.005	20.533 (2.432-173.352)
CASPASE3	.042	.006	51.697 (3.028-882.701)
CABL	.007	.008	68.124 (2.962-1566.613)
BID	.038	.010	77.659 (2.787-2164.284)
RAD50	.005	.019	158.062 (2.325-10744.279)
TAZ	.013	.003	768.971 (10.143-58295.167)

3.2 Construction of the IPRPs model

In the univariate Cox regression analysis (Figure 2A), the pathological stage (P < .001), pTstage (P < .001), pMstage (P = .001) and risk score of the IPRPs model (P < .01) were associated with shorter overall survival. However, in the multivariate Cox regression analysis, only risk score (P < .05) was significantly correlated with worse outcome (Figure 2B). C-index (0.939, 95% CI:0.916-0.962) and NRI (0.235, 95% CI:0-0.597) indicated that our model is stable. These results indicate that our IPRPs model has independent prognostic significance. The risk score with AUC of 0.933 indicates the diagnostic accuracy and consistent predictive ability of our IPRPs model (Figure 2C).

3.3 Survival analysis of the IPRPs model in the TCPA cohort

Kaplan-Meier survival curves (Figure 3A) revealed that high abundances of fatty acid synthase (FASN) (P = .039), fibronectin (FN) (P < .001), transferrin receptor (TFRC) (P < .001) and tuberous sclerosis 1 (TSC1) (P < .001) indicated a worse outcome. The formula used to predict overall survival was generated by multivariate Cox regression models as integrated risk score = 2.743 × FN abundance (ref. Low) + 0.781 × FASN abundance (ref. Low) + 1.091 × TFRC abundance (ref. Low) + 3.043 × TSC1 abundance (ref. Low). The heat map shows that the abundances of FASN, FN, TFRC and TSC1 in the high-risk group were higher than they were in the low-risk group (Figure 3B). The survival time of the high-risk group was significantly shorter than that of the low-risk group (P < .001), and the increased risk score corresponded to shorter survival (Figure 3C–E).

3.4 Validation of the prognostic value of the IPRPs model in the FUSCC cohort

Representative IHC plots for the ACC samples are displayed in Figure 4A–E (Abundances of A: Ki-67, B: Fatty acid synthase (FASN), C: Fibronectin (FN), D: Tuberous sclerosis 1 (TSC1) and E: Transferrin receptor (TFRC)). The Ki-67 protein abundance and high-risk (HR) score (Figure 4F, G) were both significantly correlated with worse outcome for patients in the FUSCC cohort (P = .005, HR = 3.787; P = .003, HR = 3.947). The IPRPs model predicted the prognosis better than the Ki-67 protein in the FUSCC cohort. A high-risk score was significantly correlated with higher Stage, T stage and N stage (Figure 4H, I).

3.5 External validation of the IPRPs model and comparison with other biomarkers using gene expression data

In the TCGA cohort (Figure 5A–F), the IPRPs model showed stronger ability for predicting prognosis than the expression levels of CTNNB1 (beta-catenin gene), IGF2 and TP53 (P = .005, HR = 3.061; P = .012, HR = 2.768; P = .162, HR = 0.574; P = .033, HR = 2.336), whereas the MKI67 (Ki-67 protein gene) and NR5A1 (SF-1 protein gene) expression levels had stronger predictive ability than the IPRPs model (P < .0001, HR = 9.238; P = .003, HR = 4.084). In the GSE10927 cohort (Figure 5G–K), which lacked IGF2 expression data, the IPRPs model showed better ability for predicting prognosis than the expression levels of TP53, CTNNB1, NR5A1 and MKI67 (P = .0898, HR = 2.318; P = .73, HR = 1.187; P = .16, HR = 1.983; P = .36, HR = 1.57; P = .22, HR = 1.824). AUC, C-index and NRI of various biomarkers were listed in Table 5, and it indicated that IPRPs model may act better than other biomarkers in RPPA data and IHC.

Table 5. Evaluating biomarkers of ACC in multiple cohorts

Biomarker evaluation	AUC				C-index				NRI
Cohorts	TCGA (RPPA)	TCGA (RNAseq)	GSE10927 (RNA seq)	FUSCC IHC)	TCGA (RPPA)	TCGA (RNAseq)	GSE10927 (RNAseq)	FUSCC (IHC)	TCGA (RPPA)	TCGA (RNAseq)	GSE10927 (RNAseq)	FUSCC (IHC)
CTNNB1	0.4	0.704	0.732	-	0.564	0.659	0.657	-	ns	-0.02	0.06	-
IGF2	-	0.313	-	-	-	0.575	-	-	-	ns	-	-
TP53	0.489	0.87	0.571	-	0.676	0.659	0.551	-	-0.02	ns	ns	-
MKI67	-	0.862	0.786	0.621	-	0.853	0.671	0.694	-	0.37	ns	ns
SF-1	-	0.71	-	-	-	0.665	-	-	-	-0.07	ns	-
Risk Score	0.933	0.885	0.705	0.649	0.939	0.789	0.657	0.72	-	-	-	-

3.6 Significantly involved pathways of the IPRPs

The top 100 genes that were most significant positively and negatively correlated with the risk score are depicted in a heat map (Figure 6A). Besides an ACC progressive phenotype, the GSEA indicated that significant alteration of the IPRPs model involved chromosome separation, metaphase-anaphase transition of the cell cycle and protein modification by small protein removal. Hub genes with prognostic implications associated with the IPRPs were involved mainly in regulation of cell-cycle pathways (Figure 6B–D).

3.7 Identification of DEGs associated with the IPRPs

A significant difference was detected between the gene expression in high-risk and low-risk groups as shown in the heat map (Figure 7A). A PPI network of the DEGs was constructed and the identified hub genes were CENPM, NDC80, DLGAP5, SPC25, CENPF, ZWILCH, AURKB, CENPA, CDC20, CCNA2, KIF4A, BUB1B, CCNB2, UBE2C, AURKA, PLK1, CASC5, RANGAP1, BIRC5, CEP55, NEK2, SGOL2, KIF18A, CCNB1, SKA1, RRM2, ASPM, SGOL1, KIF2C, CDCA8, CENPI, KIF11, BUB1, CDCA5, CDK1, SPC24, SPAG5 and NUF2 (Figure 7B,C). The functional enrichment analyses indicated the hub genes were enriched mainly in cell cycle, mitotic nuclear division, chromosome, centromeric region and microtubule binding (Table 6 and Figure 7D, E).

Table 6. GO and KEGG pathways enrichment analysis of hub genes

Term	Description	Count in gene set	P value
GO:0 140 014	Mitotic nuclear division	21	1.26E-30
GO:0 000 280	Nuclear division	22	1.81E-28
GO:0 048 285	Organelle fission	22	1.61E-27
GO:0 000 775	Chromosome, centromeric region	24	5.20E-41
GO:0 000 776	Kinetochore	21	1.48E-37
GO:0 098 687	Chromosomal region	25	1.04E-36
GO:0 008 017	Microtubule binding	8	1.65E-08
GO:0 035 173	Histone kinase activity	4	2.65E-08
GO:0 015 631	Tubulin binding	8	1.83E-07
hsa04110	Cell cycle	8	7.26E-12
hsa04914	Progesterone-mediated oocyte maturation	7	1.13E-10
hsa04114	Oocyte meiosis	7	7.02E-10

• Abbreviations: DEGs, differentially expressed genes; GO, Gene Ontology; KEGG, Kyoto Encyclopedia of Genes and Genomes.

4 DISCUSSION

The prognosis of ACC is poor because most patients with ACC have locally advanced or metastatic diseases and cannot be treated by surgery. Approximately 66% of patients with localized diseases experience recurrence and usually require systematic treatment.^{28, 29} Although there are diagnostic and prognostic molecular detection methods for ACC, including IGF2, p53, and the Wnt/β-catenin and PI3K signalling pathways, they have not been well applied in morphological evaluation, auxiliary diagnosis, or prognostic modelling of ACC.³⁰ Early diagnosis and appropriate treatment play key roles in the management of ACC; thus, effective biomarkers are urgently needed.³¹ Proteomics has unique advantages and our study is the first large-scale proteomic analysis of ACC with RPPA data. We found that FASN, FN, TFRC and TSC1 abundance levels were of high prognostic value. To explore the underlying biological mechanism, we performed a GSEA of high-risk and low-risk groups and the results indicated that the most significant pathways associated with candidate proteins included chromosome separation, metaphase-anaphase transition of cell cycle and protein modification by small protein removal. These pathways are worth further study.

FASN is a key enzyme in mammals that is needed for ab initio palmitic acid synthesis. In most normal non-adipose tissues, the abundance and activity of FASN are largely inhibited by adequate dietary fat, but in many human cancers, FASN abundance and activity are abnormally increased and are associated with poor prognosis.³² The increased abundance of FASN potentially confers tumour cells an advantage in survival and growth.³³ For instance, Ueda et al³⁴ found that FASN expression promoted cell survival and growth of tumour cells in gestational trophoblastic neoplasms, and Nguyen et al³⁵ found that increased intratumoral FASN expression led to more aggressive prostate cancers. In this study, we found that a high abundance of FASN also was significantly correlated with worse prognosis of ACC. Previous studies have established the anti-tumour effects of the first-generation FASN inhibitors.^{36, 37} Thus, FASN may be a potential therapeutic target in ACC.

FN is a large extracellular matrix protein in bones, which can combine with itself and collagen to form a network.³⁸ Studies have shown that the abundance of FN in breast cancer is higher than in normal tissues and FN abundance is significantly related to the invasiveness of the disease.³⁹ Knowles et al⁴⁰ found that FN matrix formation was associated with kidney tumour cell spreading. Besides the prognostic value of FN in ACC, we also found that the abundances of JAB1, SCD1 and PRDX1 were negatively correlated with FN abundance, whereas the abundances of HEREGULIN, TIGAR and BID were positively correlated with FN abundance. Thus, FN is also a candidate target for new therapeutic drugs.

Iron is a basic trace element involved in cell metabolism, division and proliferation, and iron also has been considered as an important factor in the development of cancer.⁴¹ TFRC is a cell surface receptor that is responsible for transferrin-mediated iron uptake; thus, TFRC may play a key role in the energy supply of cancer cells.⁴² Shpyleva et al⁴³ found a high abundance of TFRC in breast cancer, and TFRC antibodies have been used to inhibit tumour growth.⁴⁴ We found mutual inhibition between TFRC and SMAC, and that the abundances of X1433ZETA, ERK2 and CYCLINB1 were positively correlated with TFRC abundance. Modulation of PPIs is a promising new idea in drug development^{45, 46}; thus, the design of TFRC inhibitors based on the interaction modes may create new therapeutic drugs.

TSC1, in a complex with tuberous sclerosis 2, inhibits the nutrient-mediated or growth factor-stimulated phosphorylation of S6K1 and EIF4EBP1 by negative regulation of mTORC1 signal transduction.^{47, 48} We also found interactions between various types of proteins and TSC1. Among them, PARP1 abundance showed the highest correlation with TSC1 (correlation coefficient = 0.742) and it attracted our attention because of its key role in DNA repair.⁴⁹ Maintaining the integrity of the genome is the basis of cell survival, and PARP inhibitors kill tumours mainly by inhibiting DNA repair and destroying the genomes of tumour cells.⁵⁰ Inhibiting PARP1 also may inhibit TSC1, suggesting a potential strategy for the treatment of ACC.

Moon et al also established a model for predicting prognosis using RPPA data. They focused on the patients with distant metastasis. But the C-index of their model (maximum:0.86) is much lower than ours (0.939). Guo J et al identified 9 hub genes (CCNB1, CDK1, TOP2A, CCNA2, CDKN3, MAD2L1, RACGAP1, BUB1 and CCNB2) with prognostic value. But the data they used were different from our research. They only focused on gene expression data, which is usually considered unstable than protein data. And they just identified 9 hub genes with prognostic value without any further validation. In our study, we used protein data to establish a model for predicting prognosis and validate its value successfully in multiple cohorts. The main strength of this study lies in the first attempt to explore the prognostic role of protein biomarkers based on quantitative proteomic analysis of ACC in adult patients. An IPRPs model was constructed with AUC values equal to 0.933 and our results show that it distinguished itself from previous prognostic predictive models of ACC.

This study had several limitations. The validation of IRPPs model in the transcriptome dataset may lead some bias as the model derived from the proteomic data. The nature of retrospective research limits the clinical value of this work. Further validation cohorts in multicentre or prospective studies are needed to verify the findings. And the more advanced ACC patients in FUSCC cohort may lead to unbalanced baseline. However, it is difficult to conduct randomized controlled trials for ACC because of the rarity of these tumours. There is also an urgent need for in vitro and in vivo experiments to explore potential effective functions of IPRPs and reveal the underlying mechanisms.

5 CONCLUSION

We constructed an IPRPs model for predicting the prognosis of adult patients with ACC using proteomic data, gene expression data and real-world data. The prognostic model showed a stronger predictive value for prognosis than other biomarkers (eg Ki-67 and beta-catenin) in multi-cohorts. Our results distinguished FASN, FN, TFRC and TSC1 from previously identified tumour promoters and revealed novel prediction model IPRPs that outperformed the currently established prognostic parameters for anticipating disease course and better clinical management of adult ACC.

CONFLICT OF INTERESTS

The authors declare no competing interests.

AUTHORS' CONTRIBUTIONS

All authors: Work present carry out in collaboration. YDW, ZHL and QYY: Define the theme of the study and discussed analysis, interpretation and presentation. TX and XWH: Manuscript draft; data analysis; development of the algorithm; and explanation of the results. Aihetaimujiang, WHK and WFN: Participation in the collection of relevant data and manuscript draft. CDL and LWJ: Help to perform the statistical analysis. SGH: Help to revise the manuscript and provide guiding suggestions. All the authors: Read and approval of the final manuscript.

ETHICAL APPROVAL

The Ethics approval and consent to participate of the current study was approved and consented by the ethics committee of Fudan University Shanghai Cancer center.