Comprehensive analysis of an ATF/CREB family-based signature with regard to prognosis and immune feature in adrenocortical carcinoma: a cohort study
Kan Wu, MDª,#, Yige Jia, MDª,#, Yanxin Li, MDb,#, Jiayu Liang, MDª, Shengzhuo Liu, MDª, Yuchun Zhu, MDª, Xu Liu, MDc,*, Xiang Li, MDª,*, Zhihong Liu, MDª ;*
Background: The activating transcription factor (ATF)/cAMP response element-binding protein (CREB) family members play important roles in tumorigenesis and immunity across various cancers. However, their prognostic value and functional relevance in adrenocortical carcinoma (ACC) remain unrevealed.
Methods: Utilizing RNA-sequencing data from The Cancer Genome Atlas (TCGA), we evaluated the expression levels of 21 ATF/ CREB genes in ACC. A seven-gene prognostic model was constructed using LASSO and Cox regression analyses, with the TCGA cohort as the training dataset, and further validated in an external Gene Expression Omnibus (GEO, GSE10927) dataset and a West China Hospital (WCH) cohort by ATF4 immunohistochemical data from 78 samples. Kaplan-Meier and multivariable Cox analyses were performed to assess survival associations. siRNA knockdown and CCK-8 assays in SW13 cells evaluated the biological role of ATF4. Additionally, the relationship between the prognostic signature, immune checkpoints, and immune cell populations was examined.
Results: The ATF/CREB family-based signature significantly stratified ACC cases into high- and low-risk groups based on overall survival in the TCGA dataset. Furthermore, the signature remained an independent prognostic factor in multivariate analyses, and its clinical significance was well validated in different clinical subgroups and independent validation cohorts. Notably, functional assays confirmed that ATF4 knockdown suppressed ACC cell proliferation in vitro, supporting its oncogenic role. Additionally, high-risk signature patients exhibited distinct immune cell proportions and immune-suppressive states, including lower levels of CD8+ T cells, immune checkpoints, and major histocompatibility complex gene expression. Enrichment analyses linked the signature to immune regulation, cytokine signaling, and hormone metabolism pathways.
Conclusions: We established and validated the first ATF/CREB family-based prognostic model in ACC, integrating transcrip- tomic, pathological, and functional data. This signature provides prognostic insight and highlights potential for immune stratifica- tion and therapeutic targeting in ACC. However, the prediction capability of this signature for predicting prognosis and immunotherapy response warrants further validation.
Keywords: adrenocortical carcinoma, ATF/CREB transcription factors, immune checkpoints, prognostic model
Introduction
Adrenocortical carcinoma (ACC) is a rare and highly aggressive endocrine malignancy, with an estimated incidence of 0.5-2.0 cases per million people annually[1,2]. Despite complete resection of
localized disease at diagnosis, ACC patients harbor a significant risk of recurrence and metastasis, with a 5-year survival rate of less than 35%[3]. The aggressive behavior of ACC, along with its resistance to conventional therapies, including chemotherapy and radiotherapy, presents significant challenges in clinical management[3]. Although targeted therapies and immunotherapies are being explored as potential treatment options, the prognosis for ACC patients remains discouraging14), underscoring the fact that current treatments fail to entirely eradicate cancerous cells. In the era of personalized medicine, genomics studies have emphasized the importance of identifying transcription factors that may be
ªDepartment of Urology, Institute of Urology, West China Hospital, Sichuan University, Chengdu, China, Department of Urology, Institute of Urology, State Key Laboratory of Biotherapy and Cancer Center, West China Hospital, Sichuan University, Chengdu, China and “Institute for Breast Health Medicine, Cance Center, Breast Center, West China Hospital, Sichuan University, Chengdu, China #Contributed equally to this work.
Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.
*Corresponding authors. Address: Institute for Breast Health Medicine, Cance Center, Breast Center, West China Hospital, Sichuan University, 37 Guoxue Lane, Chengdu 610041, Sichuan, China. Tel .: +86 028-85120376. E-mail: lixiangscu87@163.com (X. Li); E-mail: zhihongliu031@163.com (Z. Liu); Breast Center, West China Hospital, Sichuan University, China. E-mail: liuxulzu@163.com (X. Liu).
Copyright @ 2025 The Author(s). Published by Wolters Kluwer Health, Inc. This is an open access article distributed under the terms of the Creative Commons Attribution-Non Commercial-No Derivatives License 4.0 (CCBY-NC-ND), where it is permissible to download and share the work provided it is properly cited. The work
cannot be changed in any way or used commercially without permission from the journal.
International Journal of Surgery (2025) 111:8169-8182 Received 18 February 2025; Accepted 21 June 2025
Supplemental Digital Content is available for this article. Direct URL citations are provided in the HTML and PDF versions of this article on the journal’s website, www.lww.com/international-journal-of-surgery.
Published online 8 July 2025
http://dx.doi.org/10.1097/JS9.0000000000002961
involved in tumor biological behavior, immune microenvironment, and therapeutic resistance, as well as may serve as valuable prog- nostic and therapeutic targets for ACC[5].
The activating transcription factor (ATF)/cAMP response ele- ment-binding protein (CREB) family is a group of transcription factors belonging to the basic leucine zipper family, sharing struc- tural similarities[6]. Currently comprising 21 members, these tran- scription factors exert regulatory control over numerous target genes by binding to specific DNA sequences in the promoter region17. As a consequence, they play pivotal roles in various biological processes across diverse cell types and tissues, including cell proliferation, differentiation, apoptosis, metabolism, immune responses/inflammation, and response to various cellular signals and stressors[6,8,9]. Due to their critical roles in gene regulation and cellular processes, dysregulation of ATF/CREB family mem- bers has been implicated in various diseases, including neurode- generative disorders, metabolic diseases, and cancer[7,10,11].
Currently, studies have revealed that the ATF/CREB family of transcription factors plays a multifaceted role in tumorigenesis and tumor progression across various cancer types[9,12-15]. Some ATF/ CREB family members serve as potential prognostic markers in specific cancers, where high expression is associated with poor patient outcomes and reduced survival[9,16]. Additionally, these transcription factors can influence immune checkpoint molecules, cytokines, and chemokines, thereby impacting tumor-immune interactions and responses to immunotherapies[17,18]. Moreover, the ATF/CREB family has been implicated in mediating resistance to various cancer therapies, including chemotherapy and targeted treatments[19,20]. However, their expression patterns and clinical significance in ACC have remained undisclosed. To address this gap, our study conducted a systematic investigation to unravel the expression details and clinical significance of ATF/CREB family members in ACC. Moreover, we developed and validated a risk score prognostic model based on the ATF/CREB gene family. Furthermore, we delved into the relationship between the iden- tified signature and the immune landscape in ACC. This work conforms to the principles of transparency and reproducibility in research reporting, in accordance with the TITAN guideline[21].
Materials and methods
Study population
The study included three independent cohorts of ACC patients. For the public cohorts, we initially screened 112 ACC cases from The Cancer Genome Atlas (TCGA) and the Gene Expression Omnibus (GEO) database. Inclusion criteria were as follows: (i) histologically confirmed ACC and (ii) availability of RNA sequencing data and complete clinical annotation, including survival information. Exclusion criteria were as follows: (i) non-ACC tumors and (ii) cases with incomplete survival data. After applying these criteria, 79 cases from TCGA (training cohort) and 24 cases from GEO-GSE10927 (validation cohort) were included. For the institutional cohort, we screened 89 ACC patients who underwent surgical resection at West China Hospital (WCH) between 2009 and 2019. Inclusion criteria were as follows: (i) histologically confirmed ACC and (ii) avail- ability of formalin-fixed, paraffin-embedded tissue blocks and complete clinical follow-up data. Exclusion criteria were as follows: (i) patients who received cytoreductive surgery and (ii)
HIGHLIGHTS
· Prognostic Model: A seven-gene ATF/CREB family signa- ture effectively stratifies ACC patients into high- and low-risk groups with distinct survival outcomes (HR 9.51, P < 0.001).
· Multicohort Validation: Prognostic performance was validated across public datasets (TCGA, GEO) and an institutional cohort (WCH, n = 78) via ATF4 protein expression.
· Functional Evidence: siRNA knockdown of ATF4 in SW13 ACC cells significantly inhibited cell proliferation, support- ing its oncogenic role and functional relevance in ACC.
· Immunotherapy Relevance: Low-risk patients exhibit higher immune checkpoint and MHC expression, indicating a potentially more favorable response to immunotherapy.
· Biological Insight: The signature associates with immune regulation, hormone metabolism and cytokine signaling, implicating mechanistic roles in ACC progression and immune evasion.
cases with incomplete survival data. A total of 78 patients were included in the WCH cohort. The flow diagram of patient selection is shown in Figure 1.
Public ACC datasets
We enrolled 103 ACC cases from two publicly available datasets in this study. First, the RNA sequencing data, corresponding clinical annotations, and survival information of 79 ACC samples in the TCGA were accessed through the Cancer Genomics Browser of the University of California Santa Cruz (UCSC) (), and used as the training set. In addition, the GSE10927 dataset was downloaded from the GEO (https://genomecancer.ucsc.edu), which comprised the mRNA expression data, clinical annota- tions, and survival information of 24 ACC patients, as the public validation sets. Prior to analysis, we subjected the mRNA expres- sion data from the GEO microarray to preprocessing. Log2 transformation and quantile normalization were applied to ensure data uniformity and comparability across samples. For genes interrogated by more than one probe, the mean expression was chosen as the representative expression value.
Construction of ATF/CREB family-based signature
To explore the prognostic potential of the ATF/CREB family genes in ACC, we first performed univariate Cox proportional hazards regression analysis to assess the correlation between gene expression levels and overall survival (OS) in the TCGA- ACC cohort. This method was selected due to its suitability for modeling time-to-event data while accounting for censored observations. Genes significantly associated with OS (P < 0.05) were retained for further selection. Subsequently, to reduce over- fitting and enhance model interpretability, we employed the Least Absolute Shrinkage and Selection Operator (LASSO) Cox regression model to identify the most predictive markers among the OS-related genes. Tenfold cross-validation was per- formed to determine the optimal penalty parameter (lambda) that minimized the partial likelihood deviance. Finally, the final risk score for each patient was calculated as a linear combination
1. ACC sample inclusion
Exclusion criteria:
with incomplete survival data (GEO=9, WCH=9) Insufficient tissue quality (WCH=7) cytoreductive surgery (WCH=6)
Three independent cohorts (TCGA=79, GEO=33, WCH=78)
Included individuals (TCGA=79, GEO=24, WCH=78)
2. Construction of ATF/CREB family based-signature (Training cohort: TCGA)
10
GO/KEGG (Down)
.
Coefficients
-
€
1
8
.
==
-
LI Nom
1
$
=
Log 4 (Pad))
4
-3
1
1
Prognostic significance
Cox regression analysis
LASSO Cox regression model
Immune landscape
Biological pathways
3. Validation of ATF/CREB family based-signature (Validation cohort: GEO, WCH)
SW13
SIATF4-1 siATF4-2 siATF4-3
SW13
1
1.0
SINC
low nsk
high risk
ATF4
2.5-
Relative absorbance
SINC
Survival probability
0.8
IHC
2.0
siATF4-1
Į
High
Dem
0.6
HR = 2.72 (0.99 - 7.52)
GAPDH
siATF4-2
P= 0.053
1.5
- siATF4-3
.
1.8
0.4
ATT 4 ROW
-ATFA low -ATF4 high
SW13
1.0
HR = 1.85 (1 06 = 3.22]
0
HR - 2:47 (1:33 - 4 500
ATF4 miNA salafive expression
Low
€
€
T
0.5
Alve
0.2
8
0
1
5
8
9
10
04
-
0.0
Years
~
0
2
4
6
8
Stage
Weiss
Score
Status
02
T
8 9 10
*
siA194-2
Time (days)
Mª
GSE10927
Years
Yeas
WCH
Functional validation
of the expression levels of the selected genes, weighted by their corresponding LASSO-derived coefficients. Based on the median risk score as the cutoff, patients were stratified into high- and low-risk groups. The Kaplan-Meier method was applied to estimate survival curves for both the high- and low-risk groups, and compared by a log-rank test.
Clinical specimens and immunohistochemical analysis
These specimens were obtained from 78 independent patients who had received medical care at a unique tertiary center
(WCH). All patients provided written informed consent for pathological tissue samples and clinical data collection, and the research protocol was approved by our institutional review board. Detailed clinicopathologic data, follow-up, and survival information were recorded for all patients.
Formalin-fixed, paraffin-embedded (FFPE) tissues from our study cohort were used for immunohistochemical staining using anti-ATF4 rabbit polyclonal antibody (dilution ratio 1:100; FNab00662; FineTest; Wuhan, China), following established standard protocols as previously described[22]. A semiquantitative scoring system was used to quantify the ATF4 staining based on
the intensity and extent of staining. The staining intensity was scored on a scale of 0-2 (0 = no staining, 1 = low staining, 2 = strong staining). The percentage of tumor cells showing posi- tive staining was visually scored on a scale as follows: 0 for 0% of tumor cells stained, 1 for 1-25%, 2 for 26-50%, 3 for 51-75%, and 4 for 76-100% stained. The staining intensity was multiplied by the percentage of positively stained tumor cells to derive a semiquantitative score. As a result, the final score for ATF4 staining ranged from 0 to 8. A cutoff of 4 was set to distinguish absent or weak ATF4 staining from strong ATF4 staining. Tumors with a score >4 were categorized as high ATF4 staining, while those with a score ≤4 were classified as low ATF4 staining. The ATF4 immunostaining was blindly assessed by two indepen- dent observers; the evaluations of the two observers were com- pared, and the mode was adopted for statistical analysis.
siRNA transfection and cell proliferation assay
The human ACC cell line SW13 was obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). Cells were cultured in RPMI-1640 medium (Gibco, Thermo Fisher Scientific) supplemented with 10% fetal bovine serum, 100 U/mL penicillin, and 100 µg/mL streptomycin. Cells were maintained at 37℃ in a humidified atmosphere containing 5% CO2. The culture medium was replaced every 2-3 days, and cells were subcultured upon reaching approximately 80% con- fluence using 0.25% trypsin-EDTA.
To investigate the biological function of ATF4 in ACC pro- gression, three small interfering RNAs (siRNAs) targeting ATF4 (designated as siATF4-1, siATF4-2, and siATF4-3) and a non- targeting negative control siRNA were used for gene knock- down experiments. The specific sequences of the siRNAs are provided in Supplementary Digital Content Figure S3, Available at http://links.lww.com/JS9/E625 These siRNAs were synthesized by Tsingke Biotechnology Co., Ltd., and transfected into SW13 cells using Lipofectamine 3000 (Invitrogen, USA) following the manufacturer’s protocol. The knockdown effi- ciency of ATF4 was assessed 48 h post-transfection by both quantitative real-time polymerase chain reaction (qRT-PCR) and Western blotting. Total RNA was extracted using the HiPure Total RNA Mini Kit (Magen), and cDNA was synthe- sized using a reverse transcription kit (Thermo Fisher Scientific). Protein expression of ATF4 and GAPDH was detected by Western blot using specific antibodies (ATF4: FineTest; GAPDH: Cell Signaling Technology).
To evaluate cell proliferation, the Cell Counting Kit-8 (CCK-8, Dojindo, Japan) assay was conducted. Transfected SW13 cells were seeded into 96-well plates and incubated for 0, 4, and 8 days. At each time point, 10 µL of CCK-8 reagent was added to each well and incubated for 2 h at 37℃. Absorbance at 450 nm was mea- sured using a BioTek microplate reader (BioTek Instruments, USA).
Biological process and pathway enrichment analysis
The Kyoto Encyclopedia of Genes and Genomes (KEGG) path- way enrichment analysis and Gene Ontology (GO) analysis of the ATF/CREB family-based signature related genes from the TCGA- ACC dataset were performed by using the “ClusterProfiler” pack- age. Then, the enriched pathway, including GO terms (biological process, cellular component, molecular function) and KEGG pathway, was plotted by using the “ggplot2” package.
Immune feature analysis
The profiling of immune cell infiltration across different groups was assessed by the CIBERSORT technique[23]. CIBERSORT is a sophisticated computational algorithm widely used in cancer research to estimate the abundance of distinct immune cell populations within complex tissues based on gene expression values. These characterizations have demonstrated high concor- dance with ground-truth estimations across various cancer types. Additionally, we extracted the expression values of immune checkpoint molecules and major histocompatibility complexes (MHCs) from TCGA database, and then compared immune-related gene expression between different groups.
Statistical analyses
Differences in clinicopathological variables between groups were analyzed using the chi-square test or Fisher’s exact test, as appro- priate, based on expected frequencies. The Mann-Whitney U-test was utilized to investigate the immune cell-type fractions, immune checkpoint molecules, and MHC expression levels between groups. Correlations between different variables were assessed using the Spearman method. Independent prognostic factors were deter- mined by the Cox proportional hazards regression model, and estimated as hazard ratios (HRs) with corresponding 95% confi- dence intervals (CIs). Only cases with complete gene expression and clinical data were included. Samples with missing survival or clinical annotation were excluded; no imputation was applied. Data analysis and the generation of figures were performed using the R software (version v4.0.3, the R Foundation for Statistical Computing, 2020). For all statistical analyses, a significance level of P < 0.05 was set.
Results
Cohort characteristics
The demographic and clinical characteristics of the three cohorts are summarized in Table 1 and Supplementary Digital content Table 1, Available at http://links.lww.com/JS9/E626. The TCGA cohort included 79 ACC patients with a median age of 49 years, and 61% were female. The GEO cohort (GSE10927) consisted of 24 ACC patients with a median age of 47 years, and 71% were female. The WCH cohort included 78 ACC patients with a median age of 44 years, and 58% were female. Other clinical features such as tumor stage, laterality, hormonal secretion status, Weiss score, and Ki67 index were also recorded (Table 1 and Supplementary Digital content Table 1, Available at http://links.lww.com/JS9/ E626).
The landscape and prognostic significance of the ATF/ CREB family genes in ACC
The mRNA expression data of 21 well-defined ATF/CREB family genes were obtained from the TCGA-ACC cohort. First, we compared the expression levels of these 21 genes between ACC and normal samples to investigate whether these genes exhibited aberrant expression profiles in ACC. Fourteen ATF/ CREB genes displayed significant differences in expression between ACC and normal tissue samples (Fig. 2A). Furthermore, we examined the expression correlation network of ATF/CREB family members (Supplementary Digital Content
| Table 1 Clinicopathological characteristics of study populations | ||||||||
|---|---|---|---|---|---|---|---|---|
| Characteristics | TCGA cohort (n = 79) | WCH cohort (n = 78) | ||||||
| ATF/CREB risk score | n (%) | ATF4 protein expression | ||||||
| n (%) | Low | High | P value | Low | High | P value | ||
| Age (years) | 0.432 | 0.857 | ||||||
| <50 | 40 (51) | 22 (55) | 18 (46) | 49 (63) | 26 (62) | 23 (64) | ||
| ≥50 | 39 (49) | 18 (45) | 21 (54) | 29 (37) | 16 (38) | 13 (36) | ||
| Gender | 0.128 | 0.916 | ||||||
| Male | 31 (39) | 19 (47) | 12 (31) | 33 (42) | 18 (43) | 15 (42) | ||
| Female | 48 (61) | 21 (53) | 27 (69) | 45 (58) | 24 (57) | 21 (58) | ||
| ENSAT stage | 0.029 | 0.007 | ||||||
| Low (I, II) | 46 (60) | 28 (72) | 18 (47) | 55 (71) | 35 (83) | 20 (56) | ||
| High (III, IV) | 31 (40) | 11 (28%) | 20 (53%) | 23 (29) | 7 (17%) | 18 (44) | ||
| Laterality | 0.581 | 0.432 | ||||||
| Left | 45 (57) | 24 (60) | 21 (54) | 31 (40) | 15 (36) | 16 (44) | ||
| Right | 34 (43) | 16 (40) | 18 (46) | 47 (60) | 27 (64) | 20 (56) | ||
| Hormone secretion excess | 0.010 | 0.028 | ||||||
| No | 26 (35) | 19 (49) | 7 (20) | 32 (41) | 22 (52) | 10 (28) | ||
| Yes | 48 (65) | 20 (51) | 28 (80) | 46 (59) | 20 (48) | 26 (72) | ||
| Weiss grade | 0.105 | 0.268 | ||||||
| Low [1-3] | 14 (23) | 10 (31) | 4 (14) | 15 (19) | 10 (24) | 5 (14) | ||
| High [4-9] | 47 (77) | 22 (69) | 25 (86) | 63 (81) | 32 (76) | 31 (86) | ||
| Ki67 index | 0.001 | 0.025 | ||||||
| Low | 39 (49) | 27 (68) | 12 (31) | 41 (53) | 27 (64) | 14 (39) | ||
| High | 40 (51) | 13 (32) | 27 (69) | 37 (47) | 15 (36) | 22 (61) | ||
ACC, adrenocortical carcinoma; ENSAT, European Network for the Study of Adrenal Tumours; TCGA, The Cancer Genome Atlas; WCH, West China Hospital. Bolded values signify that the P-value is statistically significant (typically P < 0.05).
Figure S1, Available at http://links.lww.com/JS9/E625). A majority of the members exhibited a robust positive correlation with each other. These findings underscore the potential relevance of the ATF/CREB family in the context of ACC and suggest a potential role in the tumorigenesis of ACC.
Subsequently, we used 79 TCGA-ACC patients with complete prognostic information to evaluate the impact of ATF/CREB family gene expression profile on patient survival. The univariate Cox regression analysis was utilized to assess the potential association between the expression levels of ATF/CREB family members and OS in ACC patients. Our analysis revealed that 11 genes within the ATF/CREB family were significantly correlated with OS (P < 0.05, Fig. 2B and Supplementary Digital Content Figure S2, Available at http://links.lww.com/JS9/E625). Among these genes, 10 genes (ATF2, ATF3, ATF4, ATF5, ATF7, BATF, CREB3, CREM, JDP2, and CREBZF) were identified as unfavorable prognostic indicators, with HR greater than 1. Conversely, one gene, CREB3L3, was identified as a “protective” factor due to its HR being less than 1.
Construction of the ATF/CREB family-based signature with ACC in the TCGA cohort
Following the filtering process to exclude genes without prog- nostic significance, a total of 11 genes were identified for further analysis. Utilizing the TCGA-ACC cohort, we applied the LASSO Cox regression model to identify a subset of seven genes with the most substantial impact on patient survival, including ATF3, ATF4, BATF, CREB3, CREM, CREB3L3, and CREBZF (Figs. 3A and 3B). Next, we constructed a risk formula based on the expression levels of these seven genes and
their corresponding regression coefficients, as follows: Risk score = (0.1322) * ATF3 + (0.6195) * ATF4 + (0.1538) * BATF + (0.0075) * CREB3 + (0.0771) * CREM + (-0.6828) * CREB3L3 + (0.2369) * CREBZF. The expression profiles of the seven selected genes and the corresponding risk scores were visually represented in Figure 3C. This allowed us to classify patients into either the high-risk group (n = 39) or the low-risk group (n = 40) based on the optimal cut-off point of 6.2335. Compared to patients in the low-risk group, those in the high- risk group exhibited significantly worse OS outcomes (HR 9.51, 95% CI 3.44-26.26, P < 0.001, Fig. 3D).
The association between the risk score and clinicopathological characteristics is summarized in Table 1. The high-risk group tended to have advanced stage diseases, hormonal hypersecretion features, and high Ki67 index. Sankey diagram also further revealed the strong correlations between the risk score and tumor stage, hormone secretion status, and survival outcomes (Fig. 3E). Given the clinical significance of disease staging in treatment selection and prognostication, we further examined the performance of the ATF/CREB family-based signature in patients with different pathologic stages within the TCGA training set. Strikingly, irrespective of the pathologic stage, a high-risk score was consistently associated with significantly unfavorable outcomes (stage I/II: log-rank test P < 0.001; stage III/IV: log-rank test P = 0.017; Figs 3F and 3G).
Validation of the ATF/CREB family-based signature in independent cohorts
To validate the reliability and reproducibility of the ATF/CREB family-based signature in the ACC patients, we calculated the risk
A
10
$[.
8
The expression levels Log2 (FPKM+1)
#[
6
$[
H
Normal
E Tumor
4
2
0
ATF1
ATF2
ATF3
ATF4
ATF5
ATF6
ATF7
BATF
BATF2
BATF3
CREB1
CREB3
CREB5
CREM
CREB3L1
CREB3L2
CREB3L3
CPEB4
JDP2
ATF6B
CREBZF
B
| Names | Hazard Ratio(95% CI) | P value |
|---|---|---|
| ATF1 | 1.93 (0.90-4.12) | 0.091 |
| ATF2 | 2.26 (1.05-4.87) | 0.038 |
| ATF3 | 2.39 (1.10-5.20) | 0.028 |
| ATF4 | 4.80 (2.02-11.41) | <0.001 |
| ATF5 | 3.93 (1.67-9.28) | 0.002 |
| ATF6 | 1.68 (0.79-3.56) | 0.175 |
| ATF7 | 3.56 (1.58-7.99) | 0.002 |
| BATF | 3.24 (1.42-7.38) | 0.005 |
| BATF2 | 0.94 (0.45-1.98) | 0.867 |
| BATF3 | 0.93 (0.44-1.95) | 0.842 |
| CREB1 | 1.64 (0.77-3.46) | 0.199 |
| CREB3 | 3.49 (1.52-8.00) | 0.003 |
| CREB5 | 1.29 (0.61-2.71) | 0.507 |
| CREM | 5.32 (2.18-13.03) | <0.001 |
| CREB3L1 | 0.54 (0.25-1.17) | 0.119 |
| CREB3L2 | 0.76 (0.35-1.63) | 0.48 |
| CREB3L3 | 0.19 (0.08-0.46) | <0.001 |
| CPEB4 | 0.68 (0.32-1.45) 1 | 0.319 |
| JDP2 | 2.98 (1.33-6.64) | 0.008 |
| ATF6B | 0.75 (0.35-1.61) | 0.456 |
| CREBZF | 3.12 (1.40-6.96) | 0.005 |
0
5
10
Figure 2. Expression and prognostic significance of ATF/CREB family genes in adrenocortical carcinoma (ACC). (A) Expression of 21 ATF/CREB family genes in normal liver tissue (GTEx data) and tumor tissue (TCGA dataset). Box plots indicate the interquartile range of values. The rows in the boxes indicate median values and the asterisks above indicate P-values (+P < 0.05, ** P < 0.01, *** P < 0.001). (B) Univariate Cox regression analysis was used to explore the relationship between ATF/CREB family genes and overall survival in patients with ACC.
score for each ACC patient in the independent GSE10927 datasets using the same risk formula. Subsequently, patients in the valida- tion set were also stratified into high- and low-risk groups based on the optimal cut-off point for risk score. As expected, risk scores showed significant associations with tumor stage, Weiss score, and survival status of ACC patients (Fig. 4A). The Kaplan-Meier sur- vival analysis also demonstrated that patients in the high-risk group
had a notable trend toward increased mortality risk compared to cases in the low-risk group (HR 2.72, 95% CI 0.99-7.52, P = 0.053, Fig. 4B).
To assess the clinical applicability and prognostic signifi- cance of the ATF/CREB family-based signature, we conducted further validation by examining the expression of ATF4 pro- tein in the WCH cohort comprising 78 ACC FFPE tissues using
A
B
0
6
7
9
9
10
10
9
9
9
9
9
9
9
8
7
7
6
6
6
4
2
1.0
ATF4
2
CREBZF
Coefficients
Partial Likelihood Deviance
0.5
CREM
BATF
AT233
=
0.0
ATF5
JDP2
10
-0.5
ATF2
a
-1.0
CREB3L3
8
0
1
2
3
4
5
-6
-5
-4
-3
-2
C
L1 Norm
D
Log(2)
RiskType
1.00
Log-Tank P = 1.46-05
8
High_risk
;HR(+high’groups)-9.505
Low_risk
Overall survival probability
95%CI(3, 441, 26.2 5)
groups=High groups
0.75
groups=Low groups
Riskscore
6
0.50
4
0.25
Status
0.00
Median time:3.2
Alive
groups=High groups groups=Low groups
39
20
5
0
0
0
10
Dead
40
30
19
10
4
2
0
2.5
5
7.5
10
12.5
Time (years)
Time
E
5
I
Cortisol
Low
0
Alive
CREBZF
II
CREB3L3
CREM-
CREB3
None
BATF
III
ATF4-
High
ATF3
Dead
IV
Other
z-score of expression
F
-2-1012
G
Stage
Hormone
Score
Status
100
100
Survival probability (%)
Survival probability (%)
75
75
50
50
25
+
low risk
25
+
low risk
+
high risk
+
high risk
0
p < 0.0001
0
p
=
O
017
0
2
4
6
8
10
12
14
0
2
4
6
8
10
12
14
Number at risk
Number at risk
low risk
28
26
17
13
7
3
2
0
low risk
11
9
4
1
1
1
0
0
high risk
18
13
6
2
0
0
0
0
high risk
20
8
2
0
0
0
0
0
0
2
4
6
8
10
12
14
0
2
4
6
8
10
12
14
Time (years)
Time (years)
A
B
1
1.0
low risk
Low
high risk
Survival probability
0.8
II
High
Dead
HR = 2.72 (0.99 - 7.52)
0.6
P = 0.053
III
High
0.4
Low
IV
Alive
0.2
0
1
2
3
4
5
6
7
8
9
10
Stage
Weiss
Score
Status
Years
C
D
100 um
E
1.0
F
- ATF4 low
1.0
- ATF4 high
ATF4 low
-ATF4 high
0.8
HR = 1.85 (1.06 - 3.22)
0.8
HR = 2.47 (1.33 - 4.59)
P = 0.030
P = 0.004
DFS (%)
OS (%)
0.6
0.6
0.4
0.4
0.2
0.2
1
2
3
4
5
6
7
8
9
10
1
2
3
4
5
6
7
8
9
10
Years
Yeas
immunohistochemistry. The ATF4 protein expression showed a significant heterogeneity in ACC samples (Figs 4C and 4D). Strong ATF4 expression was detected in 36 out of 78 samples (46.2%), while the remaining cases exhibited weak or absent ATF4 protein expression (Table 1). Strong ATF4 protein expression was notably associated with hormone excess, advanced stage, and high Ki67 index (Table 1). Patients with strong ATF4 protein expression had a shorter disease-free survival (DFS) (HR 2.72, 95% CI 0.99-7.52, P = 0.053, Fig. 4E), and OS (HR 2.72, 95% CI 0.99-7.52, P = 0.053, Fig. 4F), consistent with the transcriptomic findings.
The ATF/CREB family-based signature is an independent risk factor for patients with ACC
To evaluate whether the ATF/CREB family-based signature serves as an independent prognostic factor, a multivariate Cox regres- sion model was conducted in the TCGA-ACC cohort. Our results demonstrated that the risk scores derived from the ATF/CREB family-based signature were independently correlated with both OS (HR 6.53, 95% CI 1.91-22.27, P= 0.003) and DFS (HR 4.56, 95% CI 2.06-10.09, P < 0.001) (Table 2). Moreover, to further validate the independence of ATF/CREB family-based signature in the clinical setting, we conducted Cox regression analyses in the WCH cohort. Consistently, ATF4 protein expression demon- strated independent prognostic value for both DFS and OS in ACC patients, after adjusting for various prognostic factors such as age, sex, tumor stage, and Ki67 index (Table 2).
Functional validation of ATF4 in ACC cells
To validate the biological function of ATF4 in ACC progression, we performed siRNA-mediated knockdown experiments in
SW13 cell line (Supplementary Digital Content, Figure S3A, Available at -http://links.lww.com/JS9/E625). Three distinct siRNAs targeting ATF4 (siATF4-1, siATF4-2, siATF4-3) signif- icantly suppressed ATF4 mRNA and protein levels, as con- firmed by qRT-PCR and Western blotting (Supplementary Digital Content, Figures S3B and S3C Available at http://links. lww.com/JS9/E625). Moreover, CCK-8 proliferation assays revealed that ATF4 knockdown markedly reduced the growth of SW13 cells over an 8-day period (Supplementary Digital Content Figure S3D, Available at http://links.lww.com/JS9/ E625). These findings support the oncogenic role of ATF4 in ACC and reinforce the biological relevance of the ATF/CREB- based prognostic signature.
Immune cell infiltration and immune landscape of the ATF/ CREB family-based signature
Considering the link between the ATF/CREB family-based signa- ture and immune-related pathways, we further sought to delve deeper into the relationship between the risk score and immune cell infiltration as well as immune profiles in TCGA-ACC patients. First, we investigated the connection between the seven key prog- nostic genes and immune cell infiltration (Fig. 5A). Correlation analysis revealed that these genes were associated with the abun- dance of multiple immune cells, with ATF4, CREB3, CREM, and CREBZF negatively linked to immune cell infiltration, while BATF and CREB3L3 were positively linked to the infiltration of immune cells. Furthermore, we proceeded to compare high- and low-risk patient groups in terms of the composition of immune cell popula- tions (Fig. 5B). More specifically, the low-risk group displayed a higher abundance of cytotoxic T cells, T helper cells, natural killer (NK) cells, CD8+ T cells, and macrophages. In contrast, the
| Variables | TCGA | WCH | ||||
|---|---|---|---|---|---|---|
| HR | 95% CI | P value | HR | 95% CI | P value | |
| OS | ||||||
| Age | 1.19 | 0.52-2.74 | 0.685 | 1.13 | 0.46-2.78 | 0.576 |
| ≥50 vs >50 years Gender | 1.68 | 0.70-4.05 | 0.245 | 1.14 | 0.61-2.13 | 0.690 |
| Male vs female Stage | 3.16 | 1.13-8.82 | 0.028 | 2.32 | 1.28-3.23 | 0.035 |
| III/IV vs I/II Ki67 index | 3.84 | 1.17-12.63 | 0.027 | 1.98 | 1.04-3.77 | 0.039 |
| High vs low Risk scoreª | 6.53 | 1.91-22.27 | 0.003 | 2.02 | 1.02-4.00 | 0.043 |
| High vs low DFS | ||||||
| Age | 0.92 | 0.47-1.80 | 0.808 | 0.77 | 0.42-1.41 | 0.395 |
| ≥50 vs >50 years Gender | 0.65 | 0.33-1.31 | 0.233 | 1.13 | 0.64-2.02 | 0.670 |
| Female vs male Stage | 2.55 | 1.15-5.68 | 0.022 | 1.98 | 1.08-3.42 | 0.041 |
| III/IV vs I/II Ki67 index | 1.88 | 0.80-4.39 | 0.147 | 2.94 | 1.07-3.50 | 0.028 |
| High vs low Risk scoreª | 4.56 | 2.06-10.09 | <0.001 | 2.23 | 1.13-4.78 | 0.045 |
| High vs low | ||||||
CI, confidence interval; DFS, disease-free survival; HR, hazard ratio; OS, overall survival; TCGA, The Cancer Genome Atlas; WCH, West China Hospital.
aIn the WCH cohort refers to the protein expression level of ATF4.
Bolded values signify that the P-value is statistically significant (typically P < 0.05).
A
B
aDC
*
*
B cells
*
*
*
CD8 T cells
*
*
*
Low_risk High_risk
Cytotoxic cells
*
*
*
*
*
*
DC
*
*
*
Eosinophils
0.6
iDC
*
*
*
*
Macrophages
*
*
*
*
*
Mast cells
*
*
*
*
*
Neutrophils
*
*
*
*
*
*
* p < 0.05
0.4
NK CD56bright cells
*
*
*
*
*
Cor
NK CD56dim cells
1.0
*
*
*
*
*
NK cells
0.5
*
*
*
PDC
0.0
*
*
*
0.2
T cells
*
*
*
*
-0.5
T helper cells
*
*
-1.0
Tcm
*
*
Tem
*
*
0.0
TFH
*
*
*
*
*
*
CD4_naive
CD8_naive
cytotoxic
Tex
Tr1
nTreg
iTreg
Th1
Th17
Tfh
Tcm
Tem
NKT
MAIT
DC
B_cell
Monocyte
Macrophage
NK
Neutrophil
Tgd
CD4_T
CD8_T
Tgd
Th1 cells
*
*
*
*
Th17 cells
*
*
*
*
*
Th2 cells
*
*
*
*
*
TReg
*
ATF3
ATF4
CREB3
CREM
CREBZF
BATF
CREB3L3
C
Low_risk High_risk
10
5
&
0
CD274
CD276
TNFSF4
CD200
CD27
IDO2
TNFSF15
ICOS
TNFRSF4
TIGIT
IDO1
CD160
NRP1
CTLA4
VTCN1
TNFRSF25
KIR3DL1
CD86
TNFSF9 TNFSF18
HAVCR2
CD200R1
TNFRSF9
CD80
CD70
LGALS9
CD244
LAIR1
PDCD1
TNFRSF8
LAG3
CD44
CD28
CD48
ICOSLG
HHLA2
CD40LG
TNFRSF18
BTLA
PDCD1LG2
D
E
riskScore
M
*
*
M
*
*
*
*
*
*
M
*
*
*
Ac
*
HLA.G *
*
*
*
*
*
*
*
*
*
*
*
* *
HLA.F
*
*
*
K
*
M
*
*
V
¥
*
K
*
*
*
*
y
*
*
Low_risk High_risk
HLA.E *
*
*
*
¥
M
Me
*
*
¥
*
Ne
S *
*%
*
*
M
*
*
HLA.A
*
*
*
W
*
*
¥
*
*
*
*
S
*
*
*
*
EN
¥
15
HLA.B
M
S
M
*
*
Me
3
M
M
M
M
2
*
*
S
W
1
3
*
HLA.C
*
*
*
*
*
*
M
*
*
*
2
2
*
★
*
*
*
HLA.DRB1 *
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
€
de
*
*
* p < 0.05
HLA.DRB5
»
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
10
Cor
HLA.DRA
*
M
M
*
*
*
*
*
*
M
*
M
M
*
**
*%
A
34
54
*
1.0
HLA.DQA1
*
M
0.5
*
*
*
*
M
*
*
M
*
2
*
*
M
A
*
»
*
0.0
HLA.DQA2
*
*
*
*
*
*
*
*
*
*
*
>
*
*
*
*
*
>
*
-0.5
5
HLA.DQB1
*
*
*
*
¥
3
*
S
*
₾
S *
>
*
S
:
*
-1.0
HLA.DQB2
*
*
*
*
*
S
*
*
S
*
*
x
*
%
*
3
5
*
HLA.DPA1
*
*
*
*
*
*
*
*
*
ME
*
*
%
**
¥4
M
SA
S
*
HLA.DPB1 *
*
M
*
M
M
*
*
*
*
*
*
M
*
*
*
*
MA
S
*
0
R
HLA.G
HLA.F
HLA.E
HLA.A
HLA.B
HLA.C
HLA.DRB1
HLA.DRB5
*
*
7
HLA.DRA
HLA.DQA1
HLA.DQA2
HLA.DQB1
HLA.DQB2
HLA.DPA1
HLA.DPB1
HLA.DOA
HLA.DOB
HLA.DMA
HLA.DMB
HLA.DOA
*
*
*
*
*
*
*
*
*
*
*
*
*
HLA.DOB
*
*
>
*
*
*
*
*
>
>
*
*
M
3
*
HLA.DMA
*
3
3
S
3
3
N
3
3
2
3
*
3
*
3
>
2
*
HLA.DMB
1
3
MA
M
JA
K
M
M
3
2
>
1
3
*
M
riskScore
HLA.G
HLA.F
HLA.E
HLA.A
HLA.B
HLA.C
HLA.DRB1
HLA.DRB5
HLA.DRA
HLA.DQA1
HLA.DQA2
HLA.DQB1
HLA.DQB2
HLA.DPA1
HLA.DPB1
HLA.DOA
HLA.DOB
HLA.DMA
HLA.DMB
high-risk group exhibited a higher prevalence of central memory T cells, dendritic cells, memory B cells, and neutrophils. The close association between the risk score and immune cell infiltration sparked our curiosity about another set of immune-related prog- nostic markers in ACC-immune checkpoint genes. Therefore, we next embarked on characterizing the interactions between the risk score and 46 immune checkpoint genes in the TCGA-ACC cohort (Supplementary Digital Content Figure S4, Available at http://links. lww.com/JS9/E625). Most immune checkpoint molecules exhib- ited a significantly positive correlation with one another. As expected, the expression levels of most immune checkpoint genes were significantly higher in the low-risk score group compared to the high-risk score group (Fig. 5C). And, the majority of immune checkpoint genes exhibited an inverse correlation with the risk score (Supplementary Digital Content Figure S4, Available at http://links.lww.com/JS9/E625). Next, to better understand ATF/ CREB family-based signature-related immune profiles, we exam- ined the expression pattern of MHC molecules in the high- and low-risk groups. Similarly, the mRNA levels of most MHC genes were significantly higher in the low-risk group in comparison to the high-risk group (Fig. 5D). Additionally, interaction analysis also underscored a negative correlation between the risk score and MHC gene expression (Fig. 5E).
ATF/CREB family-based signature related biological processes
The prognostic implications and immune profile of the ATF/ CREB family-based signature in ACC patients led us to investigate signature-related biological pathways. We first identified 1077 genes differentially expressed between the low-risk group and high-risk group in the TCGA cohort. Among which, there were 611 down-expressed genes and 466 up-expressed genes (Fig. 6A). As expected, the expression levels of these core genes displayed significant differences between the two risk groups (Fig. 6B). Furthermore, the GO/KEGG (Down) enrichment analysis revealed significant enrichment in diverse pathways, including “Humoral immune response,” “Regulation of hormone levels,” and “Drug metabolism-cytochrome P450” (Fig. 6C). In contrast, the GO/KEGG (Up) enrichment analysis indicated significant enrichment in pathways related to “Signaling receptor activator activity,” “Receptor ligand activity,” and “Cytokine-cytokine receptor interaction” (Fig. 6D).
Discussion
Surgical resection, radiation, and chemotherapy remain the cur- rent mainstay of ACC treatment[3]. However, even after com- plete resection, ACC patients face a high risk of recurrence and metastasis. Moreover, conventional therapies have demon- strated limited efficacy against ACC. Although some studies suggest the potential effectiveness of targeted therapies and immunotherapies in ACC[4], achieving a transformative impact in treatment necessitates further in-depth research and clinical trials. Thus, identifying robust prognostic markers and novel therapeutic targets remains a priority.
Advancements in next-generation sequencing technology have led to the identification of numerous prognostic markers and therapeutic targets, contributing to our understanding of cancer biology. However, reliable biomarkers derived from the intrinsic tumor microenvironment for immunotherapy response
and prognosis in ACC are still scarce. In this study, we identified and validated a novel ATF/CREB family-based gene prognostic signature. This study is the first comprehensive investigation of the immune profile and prognostic implications of the ATF/ CREB signature by the gene expression data of 21 well-defined ATF/CREB family members from the TCGA-ACC dataset. Utilizing univariate Cox regression analysis and LASSO Cox regression model, a robust seven-gene based prognostic signa- ture was constructed. The validation of this signature was per- formed on an independent public cohort and 78 FFPE tissue cases by immunohistochemistry. The ATF/CREB family-based signature was identified as an independent prognostic factor for ACC patients and demonstrated significant associations with OS in various clinical subgroups. In addition, through various immune profile analyses, we observed a significant relationship between the ATF/CREB family-based signature and different tumor-infiltrating immune cells. Notably, the signature score displayed an inverse correlation with various immunotherapy- related biomarkers, suggesting that low-risk patients might bene- fit more from immune checkpoint inhibitor-based immunothera- pies. To the best of our knowledge, this study represents the most comprehensive evidence demonstrating the prognostic accuracy of the ATF/CREB family-based signature in ACC patients. With further validation, this signature could provide a more compre- hensive understanding and facilitate the precise implementation of immunotherapy in ACC patients.
The ATF/CREB family members are promising prognostic markers for patient stratification in ACC. To identify potential prognostic genes in the ATF/CREB family, we systematically analyzed a comprehensive panel of ATF/CREB family genes in the TCGA-ACC cohort. Most of the significant genes were found to be “high-risk” factors, consistent with the known functions of the ATF/CREB family in promoting cell prolifera- tion, angiogenesis, and metastasis. Finally, we filtered out seven genes, namely ATF3, ATF4, BATF, CREB3, CREM, CREB3L3, and CREBZF, to establish the prognostic signature. ATF3 and ATF4 are particularly intriguing members of this signature. In certain cancer types, such as breast cancer, lung cancer, and prostate cancer, ATF3 has been identified as a tumor suppressor gene, inhibiting cell proliferation and inducing cell cycle arrest or apoptosis to suppress tumor growth[8,24,25]. On the contrary, ATF3 has also been reported to have oncogenic properties in specific contexts, promoting cancer cell survival, migration, invasion, and metastasis in cancers like melanoma and non- small cell lung cancer[26,271. Similarly, ATF4 has emerged as a pivotal regulator in cancer biology, showing diverse roles in different cancer types[9]. In several cancer types, ATF4 functions as an oncogenic transcription factor, promoting cell prolifera- tion, migration, invasion, survival, and drug resistance, thereby contributing to tumor progression and metastasis[20,28,29]. Moreover, ATF4 plays a role in cancer cell metabolism, enabling adaptive responses to stress conditions like hypoxia and nutrient deprivation, thereby helping cancer cells survive harsh microenvironments[30]. In the context of ACC, our findings indicated that high ATF4 expression is associated with poor clinical outcomes and reduced survival rates. Our functional validation confirmed that ATF4 knockdown inhibits prolifera- tion in ACC cells, reinforcing its oncogenic role in this malig- nancy. However, it is worth noting that the expression details and functions of all molecules in this signature in ACC remain to be fully elucidated, warranting further investigation.
A
B
Up
Not sig
Down
group stage
group
30
CREB3L3
High_risk
Low_risk
-Log 10 (P.adj)
CREB3
stage
20
CREM
I/II
IIIAV
ATF4
Unknown
10
CREBZF
2
1
0
ATF3
0
-5
0
5
BATF
-1
Log2 (Fold Change)
-2
C
GO/KEGG (Down)
D
GO/KEGG (Up)
humoral immune response
pattern specification process
antibacterial humoral response
.
regionalization
BP
regulation of hormone levels
axon guidance
blood microparticle
muscle myosin complex
collagen-containing extracellular
matrix
CC
myosin filament
CC
stereocilium bundle
Ontology
myosin II complex
Ontology
BP
☐ BP
CC
CC
serine-type endopeptidase activity
MF
signaling receptor activator
MF
KEGG
activity
KEGG
serine hydrolase activity
MF
receptor ligand activity
MF
serine-type peptidase activity
DNA-binding transcription activator
activity, RNA polymerase
Il-specific
Drug metabolism - cytochrome
P450
Cytokine-cytokine receptor interaction
Metabolism of xenobiotics by
KEGG
Maturity onset diabetes of the
KEGG
cytochrome P450
young
Retinol metabolism
Neuroactive ligand-receptor
interaction
T
T
0
3
6
9
0
2
4
6
-Log 10 (P.adj)
-Log 10 (P.adj)
The ATF/CREB family of transcription factors plays a significant role in immune responses by modulating gene expression involved in various aspects of the immune system191. One important aspect of the ATF/CREB family is involved in immune cell development and differentiation, with ATF3 and ATF4 regulating the differentiation of CD8+ T cells and T helper cells, respectively[31,32]. Moreover, the ATF/CREB family can regulate the expression of immune check- point molecules, which are critical regulators of immune responses and tolerance[17,33,34]. Dysregulation of immune checkpoints, such
as PD-L1, CTLA-4, and TIM-3, can promote immune escape and cancer progression1351. Additionally, the ATF/CREB family can regulate the expression of MHC molecules on cancer cells[9]. MHC molecules play a vital role in presenting tumor antigens to immune cells, enabling the recognition and elimination of cancer cells[36]. For instance, ATF4 has been shown to regulate the expres- sion of MHC class I molecules, essential for presenting antigens to cytotoxic T cells[371. Furthermore, the ATF/CREB family can mod- ulate the production of immunosuppressive factors and cytokines
within the tumor microenvironment, creating an immune-suppres- sive milieu[38,39]. This hampers the function of effector immune cells, such as T cells and NK cells, leading to tumor immune escape[40]. In the context of the present study, enrichment analyses showed that the ATF/CREB family-based signature is associated with immune responses and cytokine interactions. Moreover, the immune correlation analysis revealed that the ATF/CREB family- based risk score is negatively associated with immune cell infiltra- tion, immune checkpoints, and MHC molecule expression. These findings align with previous studies that link ATF/CREB dysregula- tion to impaired anti-tumor immunity and poor immunotherapy response. Overall, our results suggest that the ATF/CREB-based signature may serve not only as a prognostic marker but also as a potential stratification tool for immunotherapy suitability in ACC. Patients classified into the low-risk group, with higher expression of immune checkpoint molecules and MHC genes, may derive greater benefit from immune checkpoint inhibitor- based therapies.
Although the ATF/CREB family-based signature shows pro- mise as an effective and independent prognostic determinant, along with its potential for predicting immunotherapy responses in ACC patients, there are certain limitations that warrant acknowledged. First, the analysis was based primarily on retro- spective samples. Therefore, further validation using prospective samples is imperative. Secondly, due to the limited availability of frozen ACC samples, we were unable to conduct additional experiments to validate the mRNA expression of certain factors in the WCH-ACC cohort. Hence, the prognostic predictive power of the signature may be somewhat constrained. Nevertheless, the signature could provide valuable insights into the immune microenvironment profile. Finally, while siRNA knockdown of ATF4 demonstrated in vitro functional relevance, in vivo studies are required to fully elucidate its role in tumor progression and therapeutic response. Consequently, well-pow- ered prospective studies that delve into the functional aspects of the ATF/CREB family members are warranted to elucidate their roles in ACC pathogenesis and potential therapeutic targeting.
Conclusions
In conclusion, this study represents a novel and robust ATF/ CREB family-based prognostic signature for ACC. Through comprehensive bioinformatic, pathological, and functional vali- dation, we demonstrate that the signature reliably predicts patient survival and reflects the tumor immune landscape. The model stratifies ACC patients into subgroups with distinct prog- nostic outcomes and immune characteristics, potentially guiding personalized therapeutic decisions. As the first to establish and validate such a model in ACC, our work provides new insight into the immunological role of the ATF/CREB family and high- lights ATF4 as a potential therapeutic target. With further pro- spective validation and functional exploration, this signature holds promise as a clinically applicable tool for risk assessment and immunotherapy stratification in ACC.
Ethical approval
This study was approved by the Medical Ethics Committee of West China Hospital, Sichuan University (20211068A).
Sources of funding
This work was supported by National-funded Postdoctoral Researcher Program (GZC20231821), the Sichuan Provincial Natural Science Foundation (2025ZNSFSC1917), Postdoctoral Research Fund of West China Hospital, Sichuan University (2024HXBH146), China Postdoctoral Science Foundation (2024M752256), and the National Natural Science Foundation of China (82403074).
Author contributions
KW, YGJ, YXL: data operation, writing original paper; KW, YGJ, JYL, YXL, XuL: confirmation, precise analysis; KW, ZHL, SZL: methodology; YZ, ZHL, XL: project administration; YJ, YL, KW: software; YJ, YXL, KW: data curation; YZ, XL: providing suggestion; KW, ZHL, XuL: conceptualization, fund- ing acquisition; All authors have read and approved the final manuscript.
Conflicts of interest disclosure
The authors have no potential conflicts of interest to disclose.
Research registration unique identifying number (UIN)
This study is a retrospective study.
Guarantor
Dr Liu and Li had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Provenance and peer review
Not commissioned, externally peer-reviewed.
Data availability statement
All data generated or analyzed in this study are available upon reasonable request.
Consent
This manuscript does not contain personal and/or medical infor- mation about an identifiable living individual. All anonymous patients provided a written informed consent.
References
[1] Allolio B, Fassnacht M. Clinical review: adrenocortical carcinoma: clin- ical update. J Clin Endocrinol Metab 2006;91:2027-37.
[2] Else T, Kim AC, Sabolch A, et al. Adrenocortical carcinoma. Endocr Rev 2014;35:282-326.
[3] Gaujoux S, Mihai R, Ensat E. European society of endocrine surgeons (ESES) and European network for the study of adrenal tumours (ENSAT) recommendations for the surgical management of adrenocor- tical carcinoma. Br J Surg 2017;104:358-76.
[4] Araujo-Castro M, Pascual-Corrales E, Molina-Cerrillo J, et al. Immunotherapy in adrenocortical carcinoma: predictors of response, effi- cacy, safety, and mechanisms of resistance. Biomedicines 2021;9:304.
[5] Clay MR, Pinto EM, Fishbein L, et al. Pathological and genetic stratifi- cation for management of adrenocortical carcinoma. J Clin Endocrinol Metab 2022;107:1159-69.
[6] Hai T, Hartman MG. The molecular biology and nomenclature of the activating transcription factor/cAMP responsive element binding family of transcription factors: activating transcription factor proteins and homeostasis. Gene 2001;273:1-11.
[7] Cui A, Ding D, Li Y. Regulation of hepatic metabolism and cell growth by the ATF/CREB family of transcription factors. Diabetes 2021;70: 653-64.
[8] Persengiev SP, Green MR. The role of ATF/CREB family members in cell growth, survival and apoptosis. Apoptosis Int J Program Cell Death 2003;8:225-28.
[9] Chen M, Liu Y, Yang Y, et al. Emerging roles of activating transcription factor (ATF) family members in tumourigenesis and immunity: implica- tions in cancer immunotherapy. Genes Dis 2022;9:981-99.
[10] Lonze BE, Ginty DD. Function and regulation of CREB family transcrip- tion factors in the nervous system. Neuron 2002;35:605-23.
[11] Vlahopoulos SA, Logotheti S, Mikas D, et al. The role of ATF-2 in oncogenesis. Bioessays 2008;30:314-27.
[12] Mir KB, Faheem MM, Ahmad SM, et al. ß-(4-fluorobenzyl) Arteannuin B induced interaction of ATF-4 and C/EBPB mediates the transition of breast cancer cells from autophagy to senescence. Front Oncol 2022;12: 1013500.
[13] Shen H, Gu X, Li H, et al. Exploring prognosis, tumor microenviron- ment and tumor immune infiltration in hepatocellular carcinoma based on ATF/CREB transcription factor family gene-related model. J Hepatocell Carcinoma 2023;10:327-45.
[14] Cui J, Yin Z, Liu G, et al. Activating transcription factor 1 promoted migration and invasion in lung cancer cells through regulating EGFR and MMP-2. Mol, Carcinog 2019;58:1919-24.
[15] Ghoneim C, Soula-Rothhut M, Blanchevoye C, et al. Activating tran- scription factor-1-mediated hepatocyte growth factor-induced down- regulation of thrombospondin-1 expression leads to thyroid cancer cell invasion. J Biol Chem 2007;282:15490-97.
[16] Wu DS, Chen C, Wu ZJ, et al. ATF2 predicts poor prognosis and promotes malignant phenotypes in renal cell carcinoma. J Exp Clin Cancer Res CR 2016;35:1.
[17] Liu H, Kuang X, Zhang Y, et al. ADORA1 inhibition promotes tumor immune evasion by regulating the ATF3-PD-L1 Axis. Cancer Cell 2020;37:324-39.
[18] Liu B, Chen P, Xi D, et al. ATF4 regulates CCL2 expression to promote endometrial cancer growth by controlling macrophage infiltration. Exp Cell Res 2017;360:105-12.
[19] Zhao W, Sun M, Li S, et al. Transcription factor ATF3 mediates the radioresistance of breast cancer. J Cell Mol Med 2018;22:4664-75.
[20] Igarashi T, Izumi H, Uchiumi T, et al. Clock and ATF4 transcription system regulates drug resistance in human cancer cell lines. Oncogene 2007;26:4749-60.
[21] Agha RA, Mathew G, Rashid R, et al. Transparency in the reporting of Artificial INtelligence - the TITAN guideline. Prem J Sci 2025;10:100082.
[22] Wu K, Liu Z, Liang J, et al. Expression of glucocorticoid receptor (GR) and clinical significance in adrenocortical carcinoma. Front Endocrinol Lausanne 2022;13:903824.
[23] Newman AM, Liu CL, Green MR, et al. Robust enumeration of cell subsets from tissue expression profiles. Nat Methods 2015;12:453-57.
[24] Avraham S, Korin B, Aviram S, et al. ATF3 and JDP2 deficiency in cancer associated fibroblasts promotes tumor growth via SDF-1 transcription. Oncogene 2019;38:3812-23.
[25] Wang Z, Xu D, Ding HF, et al. Loss of ATF3 promotes Akt activation and prostate cancer development in a Pten knockout mouse model. Oncogene 2015;34:4975-84.
[26] Hao ZF, Ao JH, Zhang J, et al. ATF3 activates Stat3 phosphorylation through inhibition of p53 expression in skin cancer cells. Asian Pac J Cancer Prev 2013;14:7439-44.
[27] Song X, Lu F, Liu RY, et al. Association between the ATF3 gene and non-small cell lung cancer. Thorac Cancer 2012;3:217-23.
[28] Guan X, Chen S, Liu Y, et al. PUM1 promotes ovarian cancer prolifera- tion, migration and invasion. Biochem Biophys Res Commun 2018;497: 313-18.
[29] Mu N, Lei Y, Wang Y, et al. Inhibition of SIRT1/2 upregulates HSPA5 acetylation and induces pro-survival autophagy via ATF4-DDIT4- mTORC1 axis in human lung cancer cells. Apoptosis Int J Program Cell Death 2019;24:798-811.
[30] Ye J, Kumanova M, Hart LS, et al. The GCN2-ATF4 pathway is critical for tumour cell survival and proliferation in response to nutrient deprivation. EMBO J 2010;29:2082-96.
[31] Ji LS, Sun XH, Zhang X, et al. Mechanism of follicular helper T cell differentiation regulated by transcription factors. J Immunol Res 2020;2020:1826587.
[32] Mukherjee D, Bercz LS, Torok MA, et al. Regulation of cellular immunity by activating transcription factor 4. Immunol Lett 2020;228: 24-34.
[33] Liu Q, Ou Q, Shen L, et al. BATF potentially mediates negative regula- tion of PD-1/PD-Ls pathway on T cell functions in mycobacterium tuberculosis infection. Front Immunol 2019;10:2430.
[34] Chang LC, Chen TP, Kuo WK, et al. The protein expression of PDL1 is highly correlated with those of eIF2a and ATF4 in lung cancer. Dis Markers 2018;2018:5068701.
[35] Andrews LP, Yano H, Vignali DAA. Inhibitory receptors and ligands beyond PD-1, PD-L1 and CTLA-4: breakthroughs or backups. Nat Immunol 2019;20:1425-34.
[36] Dersh D, Hollý J, Yewdell JW. A few good peptides: MHC class I-based cancer immunosurveillance and immunoevasion. Nat Rev Immunol 2021;21:116-28.
[37] Sano R, Reed JC. ER stress-induced cell death mechanisms. Biochim Biophys Acta 2013;1833:3460-70.
[38] Thomas DA, Massagué J. TGF-beta directly targets cytotoxic T cell functions during tumor evasion of immune surveillance. Cancer Cell 2005;8:369-80.
[39] Avni D, Ernst O, Philosoph A, et al. Role of CREB in modulation of TNFalpha and IL-10 expression in LPS-stimulated RAW264.7 macrophages. Mol Immunol 2010;47:1396-403.
[40] Batlle E, Massagué J. Transforming growth factor-ß signaling in immu- nity and cancer. Immunity 2019;50:924-40.