Analysis

Identification and validation of susceptibility modules and hub genes of adrenocortical carcinoma through WGCNA and machine learning

Yaoming Yang1 . Xinbao Wang1 . Liuqing Wu1 . Shihua Zhao1 . Ran Chen2 . Guoyong Yu1

Received: 24 September 2024 / Accepted: 15 April 2025

Published online: 03 May 2025

@ The Author(s) 2025 OPEN

Abstract

Purpose Adrenocortical carcinoma (ACC) is a rare and aggressive endocrine malignancy characterized by rapid progression, significantly impacting patients’ quality of life. Analyzing gene co-expression modules offers valuable insights into the molecular mechanisms driving ACC progression. In this study, we applied Weighted Gene Co-Expression Network Analysis (WGCNA) to identify gene co-expression modules associated with ACC progression.

Methods Before conducting WGCNA, differential gene expression and immune infiltration analyses were performed on the GSE90713 dataset (available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi). Dynamic tree cutting was utilized to identify co-expression modules, which were subsequently analyzed to determine their correlations and associations with traits. A total of 21 co-expression modules were identified, with the yellow module demonstrating a strong correlation with the progression of ACC. Enrichment analysis was carried out on differentially expressed genes, the yellow module, cross-module interactions, and the final hub genes to identify the associated Biological Processes (BPs) and pathways relevant to ACC. Additionally, the CIBERSORT algorithm was employed to predict immune cell infiltration in ACC.

Results The enrichment analysis revealed that pathways associated with cell division, protein synthesis, and metabolism play significant roles in the progression of ACC. Additionally, CDK1, AURKA, CCNB2, BIRC5, CCNB1, TYMS, and TOP2A were identified as key regulatory hub genes. Survival analysis further demonstrated that elevated expression levels of these genes in ACC tissues are significantly correlated with lower overall survival rates in patients, underscoring their critical involvement in ACC development and progression.

Conclusion This study sheds light on the mechanisms underlying ACC progression and highlights potential therapeutic targets. By identifying specific immune cell subtypes associated with ACC, the findings may aid in developing immune modulation therapies aimed at preventing or treating ACC.

Keywords Adrenocortical carcinoma . Prognostic genes . Weighted gene co-expression network

1 Introduction

Adrenocortical carcinoma (ACC) is a rare and aggressive endocrine malignancy, with an incidence rate of approximately two cases per million people. While ACC can manifest at any age, it demonstrates a bimodal age distribution, primarily affecting children under five years old and adults aged 40 to 50. The incidence is slightly higher in females compared to males [1-3]. Due to its varied morphology on imaging and the absence of specific

☒ Guoyong Yu, 18901133535@163.com | 1Dongzhimen Hospital, Beijing University of Chinese Medicine, Beijing 100007, China. 2School

of Traditional Chinese Medicine, Beijing University of Chinese Medicine, Beijing 100029, China.

Check for updates

(2025) 16:663

| https://doi.org/10.1007/s12672-025-02396-4

Discover

early symptoms, most cases of ACC are diagnosed at advanced stages (II or III), with a low survival rate ranging from 15 to 44%.

Currently, surgical resection remains the primary and most effective treatment option for ACC. However, even after complete resection, there is a substantial risk of recurrence, with rates ranging from 30 to 70% in patients who undergo radical surgery [1]. Prognostic factors influencing outcomes include tumor proliferative activity and excessive cortisol secretion. As a result, adjuvant therapy with mitotane is commonly administered to manage recurrence risks [4]. Despite advancements in imaging techniques, diagnostic methods for ACC are still limited, underscoring the need for simpler and more effective early diagnostic tools to improve patient outcomes [5]. Research has highlighted frequent mutations in the CTNNB1 and TP53 genes in ACC samples [6]. Specifically, about 80% of Brazilian ACC patients possess unique mutations in the TP53 gene [7]. This suggests that biomarkers hold significant potential for predicting patient survival and prognosis. Nevertheless, further evidence is required to identify additional biomarkers that can improve the diagnosis and treatment of ACC. ☒

Our study aimed to develop a polygenic signature and identify prognostic biomarkers that could enhance ACC prognosis prediction. We utilized transcriptomic sequencing data from the Gene Expression Omnibus (GEO) database to comprehensively identify and analyze genes implicated in adrenocortical carcinoma. Building on these findings, we explored the underlying molecular mechanisms to contribute to novel therapeutic strategies. Our research not only advances the understanding of ACC’s molecular landscape but also lays the foundation for targeted treatments, potentially improving patient outcomes in oncology.

2 Materials and methods

2.1 Data collection and processing

Gene expression profiles of adrenal cortex carcinoma (ACC) were obtained from the GEO database (https://www. ncbi.nlm.nih.gov/geo/). The selection criteria included: (1) adrenal cortex carcinoma, (2) samples containing both tumor and normal adrenal tissues, (3) array-based expression profiles, and (4) Homo sapiens data. Two GEO datasets were selected: GSE90713 [8] and GSE143383 [9].

2.2 Preprocessing and identification of differentially expressed genes

Differential expression analysis of gene expression data was performed using the ‘limma’ package in R language [10]. To address multiple hypothesis testing, we applied the False Discovery Rate (FDR) correction. This involved sorting the p-values of m genes and calculating FDR(i) =P(i) * m/i, while ensuring that any subsequent values greater than the preceding ones were equalized. Differentially Expressed Genes (DEGs) were identified using a threshold of |log2FC|> 1.5 and an FDR-adjusted p-value <0.05.

2.3 Immune infiltration analysis

To evaluate immune cell infiltration in ACC, the relative abundance of various immune cell subtypes was analyzed using the CIBERSORT deconvolution algorithm [11]. Gene expression profile data were used as input, along with the LM22 reference gene signature file, which encompasses expression patterns of 22 human hematopoietic cell types.

Subsequently, correlation analysis was performed to investigate the relationships among different immune cell subtypes and their associations with clinical characteristics. This analysis aimed to shed light on the potential roles these immune cells may play in the pathogenesis and progression of ACC.

2.4 Weighted gene co-expression network analysis

During the data preprocessing stage, outliers and outlying samples were first removed. Next, correlation coefficients between genes were computed to construct a correlation matrix. Using the scale-free network topology modeling

Discover

function in the WGCNA package, a soft threshold of 6 was selected to transform the correlation matrix into a weighted adjacency matrix. Based on this weighted adjacency matrix, the topological overlap matrix (TOM) was calculated. In the module identification stage, hierarchical clustering analysis was performed on the genes, employing the dissimilarity coefficient as a measure of similarity derived from the TOM matrix.

2.5 Functional enrichment analysis

Using the DAVID functional annotation tool, we performed Gene Ontology (GO) enrichment analysis and KEGG pathway enrichment analysis on the key gene modules identified through the WGCNA. In the GO enrichment analysis, we examined three aspects: biological processes (BP), cellular components (CC), and molecular functions (MF). The significance threshold for the enrichment analysis results was set at p <0.05. The results of the GO term and KEGG pathway enrichment analyses were subsequently visualized using the “ggplot2” package in the R programming language.

2.6 Construction and module analysis of protein-protein interaction networks

To further investigate the interaction between differentially expressed genes (DEGs) and key gene modules, we extracted the intersection genes and entered them into the STRING (Search Tool for the Retrieval of Interacting Genes) database to construct a protein-protein interaction (PPI) network. The resulting PPI network was then visualized using Cytoscape network visualization software.

Following this, we conducted a topological analysis of the PPI network using the node degree centrality method through the CytoHubba plugin. This analysis ranked the importance of genes within the network and identified the top 10 most critical and highly connected hub genes. These highly connected node genes were subsequently selected as candidate hub genes.

2.7 Establishment of prognostic model and survival analysis

The Random Forest (RF) algorithm [12] was employed in the R programming language for feature selection of candidate hub genes derived from the PPI network analysis. The objective was to identify core genes exhibiting significant differential expression, which could serve as potential molecular biomarkers for the early diagnosis and prognosis prediction of ACC.

To assess the clinical relevance of the identified core genes, a receiver operating characteristic (ROC) curve analysis [13] was performed using the “ROC curve” package in R. The area under the curve (AUC) values were calculated for these genes, with higher AUC values indicating a greater potential for effective classification and diagnosis.

3 Results

3.1 Identification of differentially expressed genes

The GSE90713 microarray dataset comprises ACC samples from 42 patients and highlights the expression of CXCR4 in ACC. The GSE143383 dataset, derived from a comprehensive analysis of multiple studies, includes data from 416 patients registered in the ACC Registry in Germany, 45 patients from the COMETE study, 91 patients from the TCGA study, and 43 patients from the NIH study. Exon sequencing was performed on 43 tumor samples and 25 matched normal ACC samples, identifying significant gene mutations, including TP53 and CTNNB1. This dataset also served as a validation cohort, which consisted of five normal samples and 58 tumor samples. The search date for the data was January 28, 2024 (Fig. 1).

DEGs were identified from the GEO dataset GSE90713 using R analysis. The dataset comprised 5 normal samples and 58 tumor samples. A heatmap displaying the top 50 genes based on variance from this GEO dataset is presented in Fig. 2B. The criteria for identifying DEGs were established as |log2FC|> 1.5 and an adjusted p-value <0.05. In total, 288 genes were identified, including 71 that were upregulated and 217 that were downregulated (Fig. 2A).

This analytical strategy integrates classical differential analysis methods with rigorous correction for multiple hypothesis testing, facilitating the precise identification of genes that exhibit significant and robust expression differences between normal and diseased states. This approach also controls the false positive rate in statistical analyses, laying the foundation for further investigation into their biological functions and molecular mechanisms.

Discover

Fig. 1 Technology roadmap

GEO

Downlodad of GSE90713

limma

Identification of DEGs

Highly related gene clusters (modules) — yellow

Crossover genes

DAVID GO/KEGG and Immune cell infiltration analysis

STRING PPI network Cytoscape

DAVID GO/KEGG

Hub Gene

RF

Final Hub Gene

Downlodad of GSE 143383

Validation

3.2 Weighted gene co-expression network analysis and identification of key modules

Key gene modules in ACC were identified using WGCNA. The height threshold was set at 130, resulting in the exclusion of GSM2411098 from the hierarchical clustering tree (Fig. 3A). Based on scale independence and mean connectivity, the soft-threshold power was established at 6 (Fig. 3B). Utilizing the dynamic tree cut package, a total of 21 modules were identified (Fig. 3C). Module-trait relationships were analyzed to determine the correlation between the identified modules and tumorigenesis (Fig. 3D). Notably, the yellow module exhibited a significant correlation with tumorigenesis, with a coefficient of 0.34 (Fig. 3).

This unsupervised clustering analysis effectively grouped highly co-expressed genes into distinct gene modules, thereby laying the groundwork for subsequent investigations into the relationships between these modules and phenotypic traits.

3.3 Exploration of hub genes

The yellow module genes were overlapped with DEGs to identify the crossover genes (Fig. 2C), resulting in a total of 56 intersecting genes.And a grouped dot plot (Tumor vs. Normal) is to display the expression levels of the 56 overlapping genes(Fig. 2D).For PPI network analysis, we utilized the STRING database. The CytoHubba software was employed to screen for key genes within the PPI network based on node degree, leading to the identification of 10 hub genes (Fig. 4B). These top 10 genes were selected, and the intersecting genes were recognized as candidate hub genes.

Next, we applied Recursive Feature Elimination (RFE) in conjunction with the RF algorithm to identify and retain the most influential features. RFE is a feature selection method that recursively considers progressively smaller subsets of features to determine the most significant ones. We implemented a tenfold cross-validation setup, with accuracy maximization as the evaluation criterion. This approach enabled us to identify the genes most critical for the classification of ACC.

To validate the effectiveness of the selected features, we plotted a line graph showing accuracy against the number of features (Fig. 4C), which indicated the optimal number of features that achieved the highest accuracy. This analysis confirmed the performance of the selected feature set in distinguishing between normal and ACC samples (Fig. 4).

Discover

Fig. 2 Identification and Characterization of Differentially-Expressed Genes (DEGs). A Volcano plot depicting DEGs in the GSE90713 dataset. The plot illustrates the fold-change (logFC) on the x-axis and the negative log 10 of the p-value (-log10(p-value) on the y- axis. Blue points represent down-regulated genes, red points represent up-regulated genes, and gray points represent genes with stable expression. B Heatmap showing hierarchical clustering of the top 50 DEGs with the highest standard deviation in the GSE90713 dataset. The heatmap is color-coded to represent the expression levels of genes in tumor and normal samples, as indicated by the color key. C Venn diagram depicting the overlap between DEGs and genes within the MEyellow module. The Venn diagram also includes a bar chart below it, which shows the size of each gene list (MEyellow and DEGs) and the number of elements specific to or shared by the two lists. D A grouped dot plot of the expression levels of the 56 overlapping gene

group

4

group

MEGF10

SPINK13

Tumor

NSG1

NEFL

2

Normal

A

25

B

ANO4

CTNNA2

CALB1

WIF

<CNH2

0

SEX

20

PLA

2G1B

SMI

-2

CDH2

-log10(Pvalue)

15

change

4Sb382

CYP1162 182

-4

down

ALDH1A1

10

stable

GST

SLC27A2

up

ADGRY1

5

ABLIM1

RES2

LUM

O

RG4

SLC16A9

0

I

RGFD

<2163

-6

-4

-2

0

2

PTX3

logFC

FABP

100A

100A8

NMT

ADH18

104

MEyellow

DEGS

TGDS

MI1!

APOÈ

C

D

APOC1

group

Control

ACC

1023

56

232

54

#

12

-

gene expression level

8

Size of each list

Y

4

1079

1079

539 5

258

6

0

MEyellow

DEGS

Number of elements: specific (1) or shared by 2, 3. … lists

TIMP3

CREG1

SLC40A1

PREC1

CONB2

NUPRI

SLCO2A1

TYMS

AURKA

GBP2

STMINT

Var

MADØL1

BUB18

TOPZA

APIS1

SKAT

CONAZ

NETO2

CONB1

UBEZC

ANULN

COH2

SEMAJS

ZNF367

GINST

UBEZT

ENCI

CDKNJ

APOBEC3B

USPS3

SLCARAS

RAB34

NORG4

HOKA13

CYBORI

CENPF

CENPK

CTHAGI

NOCAO

CKS2

RARRES2

GUINN

1255

2 (56)

Through PPI network analysis of the intersection between DEGs and key module genes, we not only clarified their mutual regulatory relationships at the molecular level but also uncovered potential core regulatory genes. This provides new insights for further research into their crucial roles in the pathogenesis and progression of the disease.

Utilizing the RF algorithm for feature selection, seven key genes associated with the progression of ACC were identified: CDK1, AURKA, CCNB2, BIRC5, CCNB1, TYMS, and TOP2A. These genes are believed to play significant regulatory roles in the development and progression of ACC.

3.4 Validation of prognostic model and survival analysis

The seven hub genes were subjected to ROC validation using R code. The results (Fig. 5A) demonstrated that the area under the curve (AUC) values for “CDK1,""AURKA,""CCNB2""BIRC5""CCNB1,""TYMS,” and “TOP2A” were 0.993, 0.982, 0.975, 0.979, 0.968, 0.926, and 0.972, respectively. These findings indicate a strong association of these genes with the progression and prognosis of ACC. To further validate the discriminatory power of the prognostic model using an independent dataset, we employed dataset GSE143383 for additional analysis. The results showed that the AUC values for the same seven genes CDK1, AURKA, CCNB2, BIRC5, CCNB1, TYMS, and TOP2A were 0.968, 0.905, 0.986, 0.965, 0.961, 0.993, and 0.961, respectively (Fig. 5B). These findings were consistent with the results of the prognostic model established in dataset GSE90713, further confirming the close relationship of these genes with ACC prognosis and progression. Overall, our analysis indicates that the prognostic model based on these seven key genes exhibits robust discriminatory ability for evaluating the prognosis and disease progression of ACC patients (Fig. 5). To further ensure the reliability of the validation results and avoid unreliable outcomes due to overfitting, we conducted an Out-Of-Bag (OOB) error estimation for the

Discover

Fig. 3 Weighted Gene Co-expression Network Analysis (WGCNA). A Construction of a clustering tree for sample clustering to detect outliers. The tree also shows the merging of modules with similar expression profiles. B Determination of the soft-thresholding power. The left panel shows the scale independence, and the right panel shows the mean connectivity as a function of the soft threshold (power). C Gene cluster dendrogram. The variance of the genes is in the top 25%. Each branch of the dendrogram represents a gene, and each color below the dendrogram represents a co-expression module. D Correlation of gene modules with Adrenocortical Carcinoma (ACC). The heatmap shows the correlation coefficients between different gene modules and the trait (ACC), with colors indicating the strength and direction of the correlation. E Association between modules and disease phenotype. The scatter plot shows the relationship between module eigengenes in the yellow module and the disease phenotype, with a correlation coefficient (COR) of 0.31 and a p-value of 2E-23

Sample clustering to detect outliers

Scale independence

Mean connectivity

A

1

120

CO

Scale Free Topology Model Fit, signed R

1.0

10 12 14 16 18

20

3000

GSM2411098

0.8

8

9

100

3

GSM2411092

0SU2411093

GSM2411093

GSM2411068

GSM2411113

2009411108

0.6

2000

80

GSM2411099

GSM2411094

GSM2411118

GSM2411059

Mean Connectivity

2

0.4

1000

Height

GSM2411076

60

GSM2411065

GSM2411095

GSM2411074

GSM12411067

GSM2411075

GSM2411101

GSM2411096

GSM2411069

2

GSM2411081

GSM2411047 GSM2411097

GSM2411078

05M24111037

GSM2411105

GSM2411102

GSM2411116

GSM2411091

GSM2411120

GSM2411082

GSM2411104

GSM2411100

GSM2411117

0.2

500

40

3

GSM2411086

0.0

1

o

5 6 7 8 910

12

14

16

18

20

GSM2411079

T

20

GSM2411064

GSM2411070

GSM2411083

GSM2411085

5

10

15

20

5

10

15

20

GSM2411073

Soft Threshold (power)

Soft Threshold (power)

0

GSM2411061

GSM2411062

GSM2411110

GSM2411111

GSM2411109

GSM2411115

GSM2411112

GSM2411114

GSM2411090

GSM2411089

GSM2411087

GSM2411088

GSM2411106

GSM2411107

Cluster Dendrogram

Module-trait relationships

C

1.0

D

MEmidnightblue

-0.11

0.9

(50-40)

0.11

(50-40)

ME purple

-0.049

0.049

(Se-11)

(Se-11)

0.8

MEgrey60

-0.14

(3e-79)

0.14

(3e-79)

Height

MElightgreen

-0.050

0.059

(4e-16)

(40-16)

0.7

MEblack

-0.11

0.11

(80-53)

MEsalmon

-0.035

(86-53)

(2e-06)

0.035

0.6

(2e-06)

8

MEyellow

-0.34

(0)

0.34

MEmagenta

-0.033

0.033

0.5

(5e-06)

(5e-06)

MEtan

0 82 TOY

-0.02

(0)

MElightyellow

-0.092

0.092

Zona

(28-36)

Module colors

MEred

(te-14)

(le-14)

0

MEcyan

-0.037

(46-07)

0.037

0.22

(40-07)

-0.22

E

Module membership vs. gene significance cor=0.31, p=2e-23

MEblue

(9.90009999999990e-203)

(9.9900999999990Ge-203)

MEgreenyellow

-0.084

(1e-30)

0.084

(1e-30)

MEroyalblue

-0.065

(6e-19)

0.065

0.8

(66-19)

MEbrown

-0.0097

0

(0.2)

0.0097

(0.2)

Gene significance for body weight

MEturquoise

-0.061

0.061

(40-17)

(48-17)

0.6

MElightcyan

-0.1

0.1

(De-46)

-0.26

(06-46)

MEgreen

(6e-280)

0.26

(60-280)

MEpink

-0.28

(3e-281)

0.26

0.4

-0.05

(36-281)

MEgrey

(7e-12)

0.05

(7e-12)

1

normal

tumor

0.2

0.0

0.0

0.2

0.4

0.6

0.8

Module Membership in turquoise module

random forest model. The results showed that the OOB error rate was 3.23%, which is a relatively low level. This indirectly reflects that the model has good generalization ability, thereby enhancing our confidence in the reliability of the model.

This integrated analytical strategy, which combines machine learning algorithms, differential analysis, and ROC curve assessment, facilitates the precise identification of differentially expressed core molecular biomarkers with clinical translational potential from high-throughput expression profile data. It provides critical insights for the early diagnosis and prognosis prediction of ACC while establishing a foundation for further exploration of the mechanistic roles of these biomarkers in the onset and progression of ACC.

Subsequently, we performed survival analysis on the identified hub genes using GEPIA2 (http://gepia2.cancer-pku. cn/#survival) and calculated the expression levels of these hub genes in ACC (Fig. 6).

3.5 Enrichment analysis

KEGG and GO enrichment analyses were conducted on the DEGs. The KEGG pathway analysis showed significant enrichment of these genes in several pathways, including “systemic lupus erythematosus,“steroid hormone biosynthesis,” and “Staphylococcus aureus infection” (Fig. 7A). In the GO-Biological Process (BP) category, these genes were associated with processes such as”regulation of complement activation,” “prostaglandin biosynthesis,” and “positive regulation

Discover

Fig. 4 Protein-Protein Interaction (PPI) Network Analysis. A PPI network of genes from the yellow module and differentially-expressed genes (DEGs). Nodes represent genes, and edges represent protein-protein interactions. B Identification of the most significantly up- regulated genes. The color-coding of the nodes indicates the score of the genes, with deeper colors representing higher scores. C Application of the random forest algorithm for gene selection. The plot shows the accuracy (using repeated cross-validation) as a function of the number of genes. The algorithm determined that selecting 7 genes yielded the best results

EVENSER

ONASKLAD

MORE

PLEXICO

ECHOCA

KAH&#

KCTDNI

ZWF263

A

%

-

ALCAMO

CUKR15

VIFEVIA

B

KIF11

HEYZAP

GARDA

CCNA2

cal

AoFIM

Stai

NOWOŚĆ

1000

Cauta:

-HA

ALEPA

-

O

COLOCA

NEIDE

CCNB

NE

ADRIVE

PBK

SCALA

E

TYMS

MOC#

9

AURKĄ

CDK1

-

BIRC5

PAGE

WWWESZ

p

CONPK

3

CCNB

HLADNA

TOP2A

C

CRI

NORGE

4

2

LAAR

Accuracy (Repeated Cross-Validation)

0.985

W

ELCZTAL

P

A

S

-AK

FE

9

%

on

0.980

A488-2

S

A

KOR;

AREAC

W

U

VEST

G

DAFL

ET

0.975

TENHO

DUSFRE

-

-

Q

0

VECHID

.

CD+55

C

-

:

0.970

APCACCAD

-

Chatise

STORI

S

2

4

6

8

10

Number of genes

Fig. 5 Receiver-Operating Characteristic (ROC) Curve Analysis and Area-Under-the-Curve (AUC) Calculation. A ROC curve for the hub gene. The AUC values for different genes are shown in the legend. Higher AUC values indicate a better potential for classification and diagnosis, with AUC>0.6 considered to validate the results. B ROC curves for validation models. Similar to panel A, the legend lists the AUC values for various genes, highlighting their discriminatory power in the validation models

A

1.0

1.0

0.8

0.8

Sensitivity

0.6

Sensitivity

0.6

CDK1-auc0.968

0.4

CDK1-auc0.993

AURKA-auc0.905

AURKA-auc0.982

0.4

CCNB2-auc0.975

CCNB2-auc0.986

BIRC5-auc0.965

0.2

BIRC5-auc-0.979

CCNB1-auc0.968

0.2

CCNB1-auc0.961

TYMS-auc0.926

TYMS-auc0.993

0.0

TOP2A-auc0.972

0.0

TOP2Aauc-0.961

-0.5

0.0

0.5

1.0

1.5

1 - Specificity

-0.5

0.0

0.5

1.0

1.5

1 - Specificity

of tumor necrosis factor production” (Fig. 7B). In the GO-Cellular Component (CC) category, notable enrichment was observed in the “insulin-like growth factor ternary complex,""insulin-like growth factor binding protein complex,” and “high-density lipoprotein particles” (Fig. 7C). For the GO-Molecular Function (MF) category, significant enrichment was noted in “monooxygenase activity,“iron ion binding,” and “insulin-like growth factor binding” (Fig. 7D).

Subsequently, we performed GO and KEGG analyses on the genes within the yellow module. The KEGG analysis indicated that these genes were enriched in pathways such as “viral life cycle-HIV-1,“terpenoid backbone biosynthesis,” and “steroid biosynthesis” (Fig. 7E). In the GO-BP category, these genes were involved in processes including “mitotic chromosome separation""mitotic cell cycle,” and “double-strand break repair via homologous recombination” (Fig. 7F). In the GO-CC category, these genes showed significant enrichment in the “spindle,“kinetochore,” and “nucleus” (Fig. 7G). Lastly, in the GO-MF category, they exhibited significant enrichment in “single-stranded DNA binding,""ATP-dependent single-stranded DNA helicase activity” and “RNA binding” (Fig. 7H).

Discover

Fig. 6 Survival Analysis and Disease-related Expression of Hub Genes using GEPIA2. A Overall survival curves and expression levels of CDK1 in Adrenocortical Carcinoma (ACC). The box-plot on the left shows the expression of CDK1 in ACC samples, and the Kaplan-Meier curve on the right shows the survival differences between the low-and high-expression groups. B Overall survival curves and expression levels of AURKA in ACC. Similar to panel A, it includes a box-plot for expression and a Kaplan-Meier curve for survival analysis. C Overall survival curves and expression levels of CCNB2 in ACC. The visual representation follows the same format as in panels A and B. D Overall survival curves and expression levels of BIRC5 in ACC. The box-plot and Kaplan-Meier curve are presented to show expression and survival associations. E Overall survival curves and expression levels of CCNB1 in ACC. It provides a comparison of expression and survival outcomes for different expression levels of CCNB1. F Overall survival curves and expression levels of TYMS in ACC. The combination of box- plot and Kaplan-Meier curve illustrates the relationship between TYMS expression and patient survival. G Overall survival curves and expression levels of TOP2A in ACC. The figure shows the expression distribution of TOP2A in ACC and its impact on overall survival through the Kaplan-Meier curve

A

B

C

Overall Survival

Overall Survival

-

Overall Survival

9

·

Low AURKA Group

=

9

-

Lo CON52

Low COKY Group

6

High AURKA GIOLD

High CONB2 Group

4

High COKI

2

Lag:918 00 00018

.

Logrank prp:30-06

Expression - log.(TPM + 1)

UHPC)=0.00051

8

Expression-log(TPM +1)

Percent survival

P(HR)=1:26-05

Percent survival

Expression - log:(TPM +1)

Percent survival

-

P(HR)=0.00011

0.6

.

0.6

0.6

+

0.4

0.4

0.4

1

=

2

2

2

3

A

-

-

8

-

5

%

0

50

100

150

0

50

100

150

0

50

100

150

0

Months

o

Months

0

Months

ACC

ACC

(num(T)=77; num(N)=128)

[num(T)=77; num(N)=128)

ACC

(num(T)=77; num(N)=128)

CDK1

AURKA

CCNB2

D

E

F

:

Overall Survival

8

Overall Survival

-

Overall Survival

0

Low BIRCE GINp

:

Low CONB1 Group

9

Low TYMS Group

High BỊAOS GROUP

High CONB1 Group

Hiện TYMS Group

00

O

Logrink gu6.5e-de

.

0.8

Expression - log,(TPM + 1)

Expression -log:(TPM-1)

4

Expression -log.(TPM+1}

-

Percent survival

Percent survival

Percent survival

0

0.6

0.6

Je

+

0.4

0.4

+

0,4

*

1

4

02

3

3

1

2

00

8

0

50

100

150

0

50

100

150

0

50

100

150

-

D

Months

-

Months

-

Months

(num(T)=77; num(N)=128)

ACC

ACC

(num(T)=77; num(N)=128)

ACC

BIRC5

CCNB1

[num(T)=77; num(N)=128)

TYMS

G

Overall Survival

:

Low TOPZA Group

High TOP2A Group

8

Logrank pnQ 00014

Expression - log.(TPM+ 1)

Percent survival

P(HR)-G.00047

0.6

.

.

0.

9;

0

:

5

A

0

50

100

150

-

Months

(num(T)=77;num(N)=128)

ACC

TOP2A

GO and KEGG enrichment analyses were performed on the intersecting genes between the yellow module and DEGs. The KEGG pathway analysis revealed that these genes were significantly enriched in several pathways, including “progesterone-mediated oocyte maturation,""the p53 signaling pathway,” “oocyte meiosis,""HTLV-I infection,""HIV-1 infection,""cellular senescence,” and “the cell cycle” (Fig. 71). In the BP category, these genes were involved in processes such as “female meiosis I spindle assembly,""regulation of cyclin-dependent protein serine/threonine kinase activity” and “positive regulation of mitotic spindle assembly and mitotic spindle organization” (Fig. 7J).

In the CC category, significant enrichment was observed in “spindle poles,” “spindle microtubules,” “spindles,” and “nuclei” (Fig. 7K). For the MF category, these genes were notably enriched in “protein serine/threonine kinase activity” “protein kinase binding,” and “protein kinase activity” (Fig. 7L).

This functional enrichment analysis contributes to a comprehensive interpretation of the biological functions of the genes within the relevant modules from a systems biology perspective. It elucidates the key biological processes, molecular functions, and metabolic regulatory networks in which these genes are involved, providing vital insights into

Discover

Fig. 7 Functional Enrichment Analysis of Differentially-Expressed Genes (DEGs), Yellow Module Genes, and Crossover Genes. A Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of DEGs. The plot shows the enriched KEGG pathways, with the x- axis representing the enrichment factor, the y-axis listing the pathway names, and the color and size of the points indicating the p- value and the number of genes in the pathway, respectively. B, C, D The top 10 functions enriched by DEGs in the Biological Process (BP), Cellular Component (CC), and Molecular Function (MF) analyses, respectively. Each plot presents the most significantly enriched functions in the corresponding category for DEGs. E KEGG enrichment analysis of genes in the yellow module. Similar to panel A, it visualizes the enriched KEGG pathways for the yellow-module genes. F, G, H The top 10 functions enriched by yellow-module genes in the BP, CC, and MF analyses, respectively. These plots highlight the key functional categories for the yellow-module genes. I KEGG enrichment analysis of crossover genes. The plot shows the enriched KEGG pathways specific to the crossover genes. J, K, L The top 10 functions enriched by crossover genes in the BP, CC, and MF analyses, respectively. They provide an overview of the important functional aspects of the crossover genes in different ontological categories

KEGG Enrichment Results

KEGG Enrichment Results

KEGG Enrichment Results

A

Systemic lupus erythematnous

E

Wral litt tyde - HIV-1

Terpencid backbone berninithepit

·

PValue

I

Progesterone-mediated porte maturation

Count

Court

+ 3

PI3K-Fit signaling pathway

Polycomb regressie complex

0.03

853 signaling patrway

.

Pertussis

· 20

5

Metabolism of xenotions by cytochiame P450

20

Nucisontoglaamic transport

0.01

Docte melapis

.

Term

V

Focal achopion

Fluid shear stress and amerssclerosis

PValue

Motor protsind

Human T-cell leukemia virus 1 infection

Mismatch repar

Grated cardiomyopathy

0.04

Human T-oel leukema wus 1 infection

Count

0.00

Homologous recemtinason

10

Human immuno deficiency virus 1 infection

PValue

Cortisol synthesis and sterett on-

Coronavirus disease - COVID-19-

0.02

Fancon anemia pathway

15

007

Complement and coagulation cascades

0.01

20

25

0.02

Cell cycha

30

0.01

Natohollt liver dienase

Base excision rtgar

CHlicyde

10

20

30

15

20

25

30

3

4

$

Count

6

7

.

Count

Court

BP Enrichment Results

BP Enrichment Results

BP Enrichment Results

B

regulation of complement activation

F

.

mitotic sister chromatid segregation

PValan

J

spinde assemblyinvolved in female meiosis

+

prostaglandn biosynnete procesa

Count

mitotic cet cycle

44-08

34-08

regulation of cydin dependent protein serne/threonine kinase actur

PValue

positive regulation oftumor necrosis factor production

.

6

M

double strand break regairvia homatopaus recombinaton

20-08

positive regulation of mitote cell cycle spindle assembly checkpoint

.

4e-04

35-44

pialive regalitos of amaçin muacie cel pegillaration

10

Ofi sapkcaton

14-00

mote spinde organization

20-04

poalive regulation of Eibrottaut pegillaration

V

minic ipade assembly checkpoint

18-04

postive sagalition of ERX1 and ERK2 cascade

PValue

chromcapme segregation

Count

mâuic call cyde phase trana tion

complement atăvation, dassical pattway

20-04

chaostarel bosinchani: procesa

.

Court

compiamant activation

cellularresponse to DNA damage almilus

GMi Tassiton of miviše out cada

cholesterol matad dik: procasa

10-04

044 division

50

cremesome segregation

12

10

G2-1-startid hormone biosyntetic procasa

.

70

cet division

4

6

Count

¥

10

20

40

5

Count

10

Count

15

CC Enrichment Results

CC Enrichment Results

CC Enrichment Results

C

niukn-the grouin Sicher summary compila

*

G

spindle pole

.

K

spnde pole

.

Court

inquin-le growth factor binding protein compila

.

Court

-

Court

spnde microtubule

* 5

20

100

· 10

40

nadespasm

.

200

spindle

.

15

edescelular space

6/0

miste spindle

300

400

nadius

·

20

.

citracoluter region

PValue

Term

Hnetodtore

25

E

30

emacelibr mata

PValue

50-04

edracellular tapcome

54-00

PValue

40-04

oftplasm

40-08

endoplasmic reticulum

chromasemt, centrameri: region

3e-08

aydin-dapondant grottain kinase haloanzyme pamgiai

36-04

col surface

chromopart

14-48

condensed chromasams outer emotachar

20-44

Hond mioopartadie

centroname

chromosome, contromaticragion

.

10-04

0

20

40

60

Count

0

100

250

Count

300

400

10

Court

20

30

MF Enrichment Results

MF Enrichment Results

L

MF Enrichment Results

D

monpangtnase achit

H

single-stranded DNA binding .

protein serinastresseine kinase acuity

Count

iron ion binding

Count

single-stranded DNA-dependent ATP-dependent DNA helicase achty

.

PValue

* 2

1.00-00

3

Insulin like growth factor binding

₱ 10

protein kinase binding

20

RNA binding

4

idaritical protein binding

30

protein binding

6:00-06

puiten kinase actuily

.

.

E

hama bnăng

PValue

E

mưutubule bìnướng

2 50-00

E

Bananacăn bnăng

mostubule binding

0.0020

unacasaar milti studunil conilituire

0.0015

Count

cddin-dippedent profitin soinetteroning knote engulstor acairty

PValue

calagin bnăng

0.0010

200

0.04

8.0006

400

cyclin dipon dent profein serinethreening kinase actuator actuity

.

003

catalytic acuity

ATPase actuity

600

002

caloum ion binding

ATP bnăng

chromatin binding

10

20

Court

30

6

200

400

Count

600

2

»

4

7

Count

·

5

their roles in the pathogenesis and progression of the disease. Moreover, visualizing the enrichment results enhances the clarity of the analysis findings.

3.6 Immune infiltration analysis of differentially expressed genes

Immune infiltration analysis was performed on the identified differentially expressed genes (Fig. 8A). This analysis revealed significant differential levels of M2 Macrophages, activated Dendritic cells, T follicular helper cells, and Monocytes in ACC tissues. These findings suggest that these immune cell subsets actively participate in the ACC tumor microenvironment, supporting the hypothesis that immune responses are crucial for the progression of ACC.

Subsequently, we conducted a correlation analysis among these immune cell types (Fig. 8B). The results indicated a positive correlation between activated Dendritic cells and T follicular helper cells, while a negative correlation was observed between activated Dendritic cells and M2 Macrophages. Additionally, T follicular helper cells showed a negative correlation with M2 Macrophages. In contrast, Monocytes did not exhibit significant correlations with the other cell types (Fig. 8).

This analytical approach combines advanced immune cell scoring algorithms with traditional correlation analysis, enabling a systematic characterization of immune cell infiltration patterns in the ACC tumor microenvironment. These insights are important for investigating the mechanistic roles of immune cells in the occurrence and development of ACC.

Discover

Fig. 8 Association between Immune-cell Infiltration and Signature Genes. A Comparison of immune-cell infiltration levels between the Adrenocortical Carcinoma (ACC) cohort and the healthy cohort. The box-plots display the cell composition of various immune-cell types, with the green and red colors representing ACC and normal samples, respectively. B Correlation analysis among immune cells. The heatmap shows the pairwise correlation coefficients between different immune-cell subpopulations. This analysis aims to explore the inter- relationships among immune-cell subpopulations and their potential associations with clinical features

T cells CD4 memory resting

A

TME Cell composition

B

Group ACC Normal

Monocytes

Dendritic cells resting

Macrophages M1

Monocytes

0.4

Dendritic cells resting

T cells CD4 memory resting

Cell composition

B cells memory

Macrophages M2

T cells CD8

NK cells activated

Mast cells resting

T cells CD4 memory activated

Macrophages M1

0.3

B cells memory

Mast cells activated

.

Macrophages M2

.

B cells naive

T cells CD8

T cells gamma delta

.

.

0.2

NK cells activated

·

.

.

Mast cells resting

T cells CD4 naive

·

B cells naive

.

-

.

0.1

Mast cells activated

*

·

.

T cells CD4 memory activated

·

·

.

.

T cells gamma delta

·

-

0.0

.

.

.

T cells CD4 naive

·

.

.

Macrophages M2

NK cells activated

Dendritic cells activated

T cells follicular helper

cells regulatory (Tregs)

T cells CD8

T cells regulatory (Tregs)

Monocytes

B cells naive

T cells CD4 memory resting

NK cells resting

Macrophages MO

T cells follicular helper

Mast cells resting

Macrophages MO

T cells CD4 naive

Plasma cells

Dendritic cells resting

Neutrophils

B cells memory

T cells CD4 memory activated

T cells gamma delta

NK cells resting

Macrophages M1

Mast cells activated

Eosinophils

Dendritic cells activated

NK cells resting

-

-

.

-

.

.

Plasma cells

Macrophages MO

.

·

.

- .

.

.

.

.

T cells follicular helper

.

.

.

·

T cells regulatory (Tregs)

.

. ☒

.

.

.

·

·

·

Plasma cells

.

.

·

Dendritic cells activated

..

·

.

.

-

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

4 Discussion

This study utilized the GEO dataset GSE90713 and integrated it with bioinformatics analysis to identify hub genes and prognostic markers crucial for the development and progression of ACC. By employing WGCNA, a systems biology tool, we identified DEGs and key modules within gene networks based on clustering and biological functional annotation results. An immune infiltration analysis of these DEGs revealed that immune responses play a significant role in ACC, underscoring the potential value of immunotherapy in its treatment. Further analysis, which examined the overlap between ke1y module genes and DEGs, involved PPI network analysis to identify 10 hub genes. Using the Random Forest (RF) method for validation, we identified seven key genes: CDK1, AURKA, CCNB2, BIRC5, CCNB1, TYMS and TOP2A. We developed a prognostic model based on these key genes and validated its accuracy in predicting ACC prognosis through receiver operating characteristic (ROC) curves. Consequently, these genes show promise as potential biomarkers and prognostic indicators for the diagnosis and treatment of ACC. To further validate our findings, we conducted an independent dataset validation using the GSE143383 GEO dataset, which revealed a strong correlation between these seven key genes and ACC prognosis and progression, thereby bolstering our research outcomes.

CDK1, a critical protein kinase that regulates the G2/M phase transition of the cell cycle, exhibits abnormal expression in various tumors. Previous studies have indicated its involvement in the transition from the G2 phase to mitosis, often demonstrating enhanced or overexpressed activity in cancer cells [14-16]. Overactivation of CDK1 can disrupt cell mitosis and increase genomic instability, thereby promoting tumor initiation and progression.A study employing immunohistochemistry examined ACC samples and discovered a significant elevation in the expression of CDK1 protein in ACC (p<0.01), corroborating the accuracy of our findings [17]. Previous research has linked CDK1 to recurrent tumors, suggesting its potential role in tumor recurrence and treatment resistance, which may also be relevant to ACC [18]. In our study, we observed elevated expression levels of CDK1, AURKA, CCNB2, and CCNB1 in ACC. AURKA, a member of the Aurora kinase family, plays a crucial role in regulating cell division during mitosis. It is overexpressed in multiple cancers and is involved in cancer cell proliferation, epithelial-mesenchymal transition (EMT), metastasis, apoptosis, and the self-renewal of cancer stem cells, thereby contributing to tumorigenesis [19]. Moreover, AURKA expression correlates positively with tumor immune infiltration in various cancers [20]. Other studies have found that compared to normal adrenals, the expression of AURKA in ACC is 5.2 times higher (p <0.001). Experimental evidence indicates that the inhibition of Aurora kinase and the Wnt/B-catenin pathway reduces the growth of adrenocortical cancer cells [21]. Furthermore, it has been established in prior research that AURKA is overexpressed in childhood adrenocortical tumors, aligning with the results of our study [22] CCNB1 and CCNB2 are members of the cyclin B protein family, with CCNB1 being particularly important for regulating G2/M phase progression of the

Discover

cell cycle. Scholars have identified CCNB1 as a specific marker that distinguishes between adrenocortical carcinoma and adenoma, and CCNB2 was detected in formalin-fixed paraffin-embedded ACC samples [28]. CCNB1 inhibits separase activation, leading to delayed chromatid separation, while CCNB2 induces excessive activation of AURKA, resulting in accelerated centrosome separation and chromosome lagging. These findings suggest a causal relationship between the overexpression of B-type cyclins and tumor pathophysiology [23-26].Our findings further underline their critical role in ACC and suggest a causal relationship between the overexpression of B-type cyclins and the invasive behavior of ACC tumors. BIRC5, one of the inhibitors of apoptosis proteins (IAPs), inhibits apoptosis and promotes cell proliferation. It is essential for the mitotic checkpoint and is involved in cell differentiation, proliferation, and invasion. Although no progress has been made in linking BIRC5, TYMS, and TOP2A to ACC specifically, their further exploration in ACC research remains worthwhile.BIRC5 is highly expressed in most tumors, leading to poor prognosis for cancer patients [29, 30]. Thymidylate synthase (TYMS) is pivotal in thymidine biosynthesis and is located on chromosome 18p; it is a fundamental substrate for DNA synthesis.It is known that TYMS expression is elevated in renal cell carcinoma [31]. Its overexpression in malignant tumor cells is associated with adverse clinical outcomes in cancer. Topoisomerase 2a (TOP2A), a subtype of type II topoisomerase (TOP2), is essential for maintaining genomic integrity during various dynamic processes, such as transcription, replication, and cell division. Crucial for cell division, TOP2A is highly expressed during mitosis and is necessary for chromosome condensation and segregation during mitotic cell division.

ACC is a rare but highly heterogeneous malignancy with poor prognosis and high recurrence rates. Therefore, identifying key genes associated with the occurrence, development, and invasion of ACC is crucial for improving diagnosis, prognosis, and treatment. Recently, DLK1 (Delta-like non-canonical Notch ligand 1), a marker of adrenocortical stem/progenitor cells, has been shown to be re-expressed in ACC and correlated with tumor malignancy and recurrence risk [32]. This suggests that stem cell-related genes may play a significant role in the tumor biology of ACC. Research indicates that DLK1 expression in ACC is linked to stem cell characteristics, and DLK1 + cells exhibit higher steroidogenic potential and clonogenic capacity. CDK1 and CCNB1 play pivotal roles in the G2/M phase transition of the cell cycle, while BIRC5 is associated with cell survival and stem cell properties. Abnormal activation of these genes may endow ACC cells with stem cell-like characteristics, promoting tumor invasion and recurrence. Studies also highlight the correlation between DLK1 expression levels and recurrence-free survival in ACC patients, with its soluble ectodomain detectable in patient serum, aiding in the diagnosis and follow-up of ACC. This underscores DLK1’s potential as a prognostic and diagnostic marker. In our study, the high expression of CDK1 and AURKA is often associated with tumor aggressiveness and poor prognosis. We also discuss the expression levels of seven core genes in ACC and patient survival rates, further validating their prognostic value. This suggests that future research could explore the possibility of combining these genes with DLK1 to enhance the accuracy of prognostic predictions. Additionally, inhibiting DLK1 function may suppress the stem cell characteristics and invasive capabilities of tumor cells. The core genes in our study, such as CDK1 and CCNB1, which are key regulators of the cell cycle, have the potential to become therapeutic targets, and inhibitors of these genes are already undergoing clinical trials in various cancers. Future research can also investigate the interactions between these core genes and DLK1 to identify strategies for combination therapy. Continued exploration of related outcomes is encouraged, such as studying the functions of core genes in ACC cells using gene-editing techniques like CRISPR-Cas9 and exploring the interaction networks between these genes and DLK1. Furthermore, clinical trials can be conducted to evaluate the efficacy of inhibitors targeting these genes in the treatment of ACC.

In conclusion, Our analysis of the differential expression of these key genes in primary and recurrent ACC tumors provides valuable insights into their potential roles in tumor recurrence.The high expression levels of these key genes in ACC are closely linked to the onset and progression of the disease, indicating their potential as promising biomarkers and predictive factors for ACC diagnosis and treatment. Future studies should investigate the specific mechanisms by which these genes operate in ACC and evaluate their potential as therapeutic targets.

We conducted KEGG and GO enrichment analyses on DEGs in ACC tumors to elucidate their biological functions and involvement in signaling pathways. Our findings revealed significant enrichment of DEGs in various biological pathways associated with tumor development, thereby illuminating the molecular mechanisms underlying ACC. Notably, the enrichment of DEGs in the steroid hormone biosynthesis pathway underscores their critical roles in the pathogenesis of ACC. Given the essential regulatory functions of steroid hormones in cell growth, differentiation, and metabolism, dysregulated expression of these genes may drive malignant transformation and tumor proliferation.

Furthermore, the GO biological process (GO-BP) analysis emphasized the involvement of DEGs in regulating complement activation and prostaglandin biosynthesis, both of which are pivotal in immune response and inflammation.

Discover

This suggests that DEGs may modulate the tumor microenvironment through immune and inflammatory responses, thereby promoting tumor progression.

The KEGG analysis of genes within the yellow module revealed enrichment in pathways such as the HIV-1 lifecycle. Although this pathway may appear unrelated to the direct pathological mechanisms of ACC, these findings could be connected to viral infection and host immune responses, providing a fresh perspective on the tumor immune microenvironment in ACC. Additionally, GO cellular component (GO-CC) and GO molecular function (GO-MF) analyses elucidated the roles of these genes in maintaining cellular structure and function, including spindle assembly and RNA binding, both of which are crucial for cell division and gene expression regulation.

Moreover, the analysis of intersecting genes between the yellow module and DEGs highlighted enrichment in pathways related to oocyte maturation and cell cycle regulation, suggesting their potential roles in reproductive and developmental processes. Aberrant expression of these genes may disrupt cell cycle regulation, thereby promoting tumor cell proliferation. Furthermore, the GO-MF analysis indicated significant enrichment of these genes in protein kinase activity, underscoring their potential roles in cellular signaling and regulation. Targeting these genes could provide new therapeutic strategies to overcome platinum resistance in ACC tumors.

In conclusion, our analysis not only highlighted the significant roles of DEGs and the genes within the yellow module in pathways and processes relevant to ACC tumors, but it also identified several potential molecular targets and avenues for future research. Subsequent studies can explore the mechanisms by which these genes contribute to the onset and progression of ACC tumors, as well as evaluate their viability as therapeutic targets.

The results of the immune infiltration analysis conducted in this study revealed significant differences in the infiltration levels of immune cell subtypes, including M2 macrophages, activated dendritic cells, follicular helper T cells, and monocytes, within the ACC tumor microenvironment. These findings support the hypothesis regarding the critical role of the immune response in the progression of ACC and align with the theoretical foundations of tumor immunology. M2 macrophages secrete various factors that promote tumor growth, such as VEGF and MMPs, which facilitate tumor cell migration, invasion, and angiogenesis [33, 34]. Furthermore, they suppress the activity of immune effector cells, including T cells and NK cells, thereby weakening the body’s anti-tumor immune response [35-37], which favors the malignant progression of the tumor. In contrast, activated dendritic cells and follicular helper T cells play crucial roles in initiating and regulating anti-tumor immune responses [38, 39], guiding T cells and B cells in their fight against cancer [40]. Monocytes may exhibit a dual role in the tumor microenvironment [41]. These results highlight the complex immune regulatory mechanisms involved in ACC pathogenesis. Further correlation analysis revealed a positive correlation between activated dendritic cells and follicular helper T cells, while both exhibited a negative correlation with M2 macrophages. Additionally, follicular helper T cells demonstrated a negative correlation with M2 macrophages. This suggests the existence of a regulatory network characterized by mutual inhibition or cooperation among different immune cell subtypes. For instance, dendritic cells and follicular helper T cells may undergo synergistic activation while simultaneously inhibiting M2 macrophages. These intricate regulatory relationships may play a significant role in the immune pathology of ACC.

The immune analysis results from this study confirm the critical role of the immune microenvironment in the development of ACC and provide initial insights into the potential regulatory networks and molecular mechanisms involved. A deeper understanding of this complex immune regulatory network will enhance our knowledge of ACC pathogenesis and support the development of immunotherapeutic strategies, thereby offering new treatment options for patients.

In summary, while our study has yielded significant discoveries, it also faces the limitations mentioned earlier. Addressing these limitations in future research endeavors will be essential for strengthening the theoretical foundations and practical guidance for precision diagnosis and treatment in ACC. However, our study has several limitations. First, data sourcing is restricted to the GEO database’s GSE90713 and GSE143383 datasets, which may introduce bias and lack general representativeness. Second, experimental validation is lacking; our findings are primarily derived from bioinformatics analysis, necessitating further experimental verification to substantiate the roles and clinical significance of these potential key genes and regulatory pathways in ACC. Third, the prospective application of these biomarkers requires additional evaluation. Despite identifying seven core genes with prognostic potential, their feasibility and accuracy as ACC biomarkers for diagnostics, molecular subtyping, and personalized treatment still need thorough evaluation and refinement. Lastly, the explanation of the immune regulatory network is not comprehensive. While we provide preliminary insights into the intricate tumor immune microenvironment of ACC, deeper studies into molecular mechanisms are imperative for a comprehensive elucidation of the complexities involved. To fully realize the potential of these genes as biomarkers, further preclinical research and large-scale clinical validation are necessary.

Discover

5 Conclusion

In conclusion, this study conducted a comprehensive analysis of DEGs in ACC. A total of 288 DEGs were identified, among which seven genes-CDK1, AURKA, CCNB2, BIRC5, CCNB1, TYMS, and TOP2A-were pinpointed as potential key biomarkers. These genes demonstrated significant differential expression in ACC tissues, and their significance was further corroborated through validation with independent datasets. As potential biomarkers for ACC, these genes hold great promise in providing new molecular targets for early diagnosis, prognosis assessment, and treatment monitoring. They can serve as the basis for the development of novel molecular diagnostic reagents, thereby enabling more precise and timely detection of ACC. Moreover, these genes offer valuable insights into the molecular subtyping and personalized treatment of ACC, paving the way for the exploration of new targeted therapeutic strategies. Looking forward, future research should concentrate on investigating the expression characteristics of these molecular markers across different stages and subtypes of ACC, as well as examining their correlation with clinical outcomes and prognosis. Such endeavors will contribute to establishing a more robust foundation for precision medicine in ACC, ultimately enhancing our understanding and management of this complex disease.

Acknowledgements Not applicable.

Author contributions All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Y. Y. X W. L.W S.Z. R.C and G.Y. The first draft of the manuscript was written by Y.Y and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Funding Basic and Clinical Research on Nephritis Anti-Failure Granules in Chronic Renal Interstitial Tubular Disease (601071)

Data availability The datasets [GSE 90713 and GSE 143383] for this study can be found in the [GEO database] (https://www.ncbi.nlm.nih. gov/geo/).

Ethical approval and consent to participate Not applicable.

Consent for publication Not applicable.

Competing interests The authors declare no competing interests.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativeco mmons.org/licenses/by-nc-nd/4.0/.

References

1. Fassnacht M, et al. European Society of Endocrinology clinical practice guidelines on the management of adrenal incidentalomas, in collaboration with the European network for the study of adrenal tumors. Eur J Endocrinol. 2023;189(1):G1-g42.

2. Rodriguez-Galindo C, et al. Treatment of Pediatric Adrenocortical Carcinoma With Surgery, Retroperitoneal Lymph Node Dissection, and Chemotherapy: The Children’s Oncology Group ARAR0332 Protocol. J Clin Oncol. 2021;39(22):2463-73.

3. Crona J, Beuschlein F. Adrenocortical carcinoma - towards genomics guided clinical care. Nat Rev Endocrinol. 2019;15(9):548-60.

4. Fassnacht M, et al. European society of endocrinology clinical practice guidelines on the management of adrenocortical carcinoma in adults, in collaboration with the European network for the study of adrenal tumors. eur J Endocrinol. 2018;179(4):G1-g46.

5. Zheng S, et al. Comprehensive pan-genomic characterization of adrenocortical carcinoma. Cancer Cell. 2016;29(5):723-36.

6. Fiorentini C, et al. Molecular drivers of potential immunotherapy failure in adrenocortical carcinoma. J Oncol. 2019;2019:6072863.

7. Riedmeier M, et al. Adrenocortical carcinoma in childhood: a systematic review. Cancers. 2021;13(21):10.

8. Weiss ID, et al. Screening of cancer tissue arrays identifies CXCR4 on adrenocortical carcinoma: correlates with expression and quantification on metastases using 64Cu-plerixafor PET. Oncotarget. 2017;8(43):73387-406.

9. Fojo T, et al. Metastatic and recurrent adrenocortical cancer is not defined by its genomic landscape. BMC Med Genomics. 2020;13(1):165.

10. Ritchie ME, et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7): e47.

Discover

11. Chen B, et al. Profiling tumor infiltrating immune cells with CIBERSORT. Methods Mol Biol. 2018;1711:243-59.

12. Breiman L. Random Forests. Mach Learn. 2001;45(1):5-32.

13. Rosner, B., Fundamentals of Biostatistics. London : Cengage Learning EMEA, , ed. i. edition. 2011.

14. Prevo R, et al. CDK1 inhibition sensitizes normal cells to DNA damage in a cell cycle dependent manner. Cell Cycle. 2018;17(12):1513-23.

15. Ravindran Menon D, et al. CDK1 interacts with Sox2 and promotes tumor initiation in human melanoma. Cancer Res. 2018;78(23):6561-74.

16. Ying X, et al. CDK1 serves as a novel therapeutic target for endometrioid endometrial cancer. J Cancer. 2021;12(8):2206-15.

17. Wu A, et al. Transmissible gastroenteritis virus targets paneth cells to inhibit the self-renewal and differentiation of Lgr5 intestinal stem cells via Notch signaling. Cell Death Dis. 2020;11(1):40.

18. Butz H, et al. Systematic investigation of expression of G2/M transition genes reveals CDC25 alteration in nonfunctioning pituitary adenomas. Pathol Oncol Res POR. 2017;23(3):633-41.

19. Du R, et al. Targeting AURKA in cancer: molecular mechanisms and opportunities for Cancer therapy. Mol Cancer. 2021;20(1):15.

20. Lu K, et al. Comprehensive pan-cancer analysis and the regulatory mechanism of AURKA, a gene associated with prognosis of ferroptosis of adrenal cortical carcinoma in the tumor micro-environment. Front Genet. 2022;13: 996180.

21. Maria AG, et al. Inhibition of aurora kinase A activity enhances the antitumor response of beta-catenin blockade in human adrenocortical cancer cells. Mol Cell Endocrinol. 2021;528: 111243.

22. Borges KS, et al. Spindle assembly checkpoint gene expression in childhood adrenocortical tumors (ACT): overexpression of aurora kinases A and B is associated with a poor prognosis. Pediatr Blood Cancer. 2013;60(11):1809-16.

23. Ikeya A, et al. CCNB2 and AURKA overexpression may cause atypical mitosis in Japanese cortisol-producing adrenocortical carcinoma with TP53 somatic variant. PLoS ONE. 2020;15(4): e0231665.

24. Yu S, Ma J. Spindle and kinetochore-associated complex is associated with poor prognosis in adrenocortical carcinoma. J Surg Res. 2022;277:50-9.

25. Nam HJ, van Deursen JM. Cyclin B2 and p53 control proper timing of centrosome separation. Nat Cell Biol. 2014;16(6):538-49.

26. Yanxiu L, et al. Advances in the study of cyclin B1 in malignancies. J Chongqing Med Univ. 2021;41:1-4.

27. Soon PSH, et al. Microarray gene expression and immunohistochemistry analyses of adrenocortical tumors identify IGF2 and Ki-67 as useful in differentiating carcinomas from adenomas. Endocr Relat Cancer. 2009;16(2):573-83.

28. Iwahashi N, et al. Whole transcriptome profiling of adrenocortical tumors using formalin-fixed paraffin-embedded samples. Front Endocrinol. 2022;13: 808331.

29. Ye HB, et al. Bioinformatics analysis of BIRC5 in human cancers. Ann Transl Med. 2022;10(16):888.

30. Liu Z, et al. A viroporin-like 2B protein of duck hepatitis A virus 1 that induces incomplete autophagy in DEF cells. Poult Sci. 2021;100(10): 101331.

31. Mizutani Y, et al. Significance of thymidylate synthase activity in renal cell carcinoma. Clin Cancer Res Off J Am Assoc Cancer Res. 2003;9(4):1453-60.

32. Mariniello K, et al. Dlk1 is a novel adrenocortical stem/progenitor cell marker that predicts malignancy in adrenocortical carcinoma. BioRxiv. 2024;10:12.

33. Condeelis J, Pollard JW. Macrophages: obligate partners for tumor cell migration, invasion, and metastasis. Cell. 2006;124(2):263-6.

34. Zumsteg A, Christofori G. Corrupt policemen: inflammatory cells promote tumor angiogenesis. Curr Opin Oncol. 2009;21(1):60-70.

35. Liu KX, Joshi S. “Re-educating” tumor associated macrophages as a novel immunotherapy strategy for neuroblastoma. Front Immunol. 2020;11:1947.

36. Fu LQ, et al. The roles of tumor-associated macrophages in tumor angiogenesis and metastasis. Cell Immunol. 2020;353: 104119.

37. Gao J, Liang Y, Wang L. Shaping polarization of tumor-associated macrophages in cancer immunotherapy. Front Immunol. 2022;13: 888713.

38. Hansen M, Andersen MH. The role of dendritic cells in cancer in Seminars in immunopathology. Berlin: Springer; 2017.

39. Gutiérrez-Melo N, Baumjohann D. T follicular helper cells in cancer. Trends in cancer. 2023;9(4):309-25.

40. Gardner A, Ruffell B. Dendritic cells and cancer immunity. Trends Immunol. 2016;37(12):855-65.

41. Olingy CE, Dinh HQ, Hedrick CC. Monocyte heterogeneity and functions in cancer. J Leukoc Biol. 2019;106(2):309-22.

Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Discover