CDKN1C as a prognostic biomarker correlated with immune infiltrates and therapeutic responses in breast cancer patients

Abstract Breast cancer (BC) prognosis and therapeutic sensitivity could not be predicted efficiently. Previous evidence have shown the vital roles of CDKN1C in BC. Therefore, we aimed to construct a CDKN1C‐based model to accurately predicting overall survival (OS) and treatment responses in BC patients. In this study, 995 BC patients from The Cancer Genome Atlas database were selected. Kaplan‐Meier curve, Gene set enrichment and immune infiltrates analyses were executed. We developed a novel CDKN1C‐based nomogram to predict the OS, verified by the time‐dependent receiver operating characteristic curve, calibration curve and decision curve. Therapeutic response prediction was followed based on the low‐ and high‐nomogram score groups. Our results indicated that low‐CDKN1C expression was associated with shorter OS and lower proportion of naïve B cells, CD8 T cells, activated NK cells. The predictive accuracy of the nomogram for 5‐year OS was superior to the tumour‐node‐metastasis stage (area under the curve: 0.746 vs. 0.634, p < 0.001). The nomogram exhibited excellent predictive performance, calibration ability and clinical utility. Moreover, low‐risk patients were identified with stronger sensitivity to therapeutic agents. This tool can improve BC prognosis and therapeutic responses prediction, thus guiding individualized treatment decisions.

of BC patients. [13][14][15] Genes are closely associated with cell cycle and apoptosis, thus playing pivotal roles in tumour progression.
Numerous genome variants have been reported to be associated with BC survival outcomes and treatment responses. [16][17][18] Antitumor medicine can decrease the risk of recurrence and BC mortality, 19,20 but its application was limited by the uncertain effectiveness and common adverse effects. 21 Although a vast majority of methods had been generated to monitor the therapeutic responses, 22,23 they could not identify the patients who can benefit from some specific drugs clinically. Hence, the construction of a novel tool for precise prediction of BC prognosis and therapeutic responses is required.
CDKN1C, encoding the Cyclin-dependent kinase inhibitor p57 Kip2 , is a paternally imprinted gene on chromosomal band 11 p15.5. Its encoded protein blocks the substrate interaction domain on cyclins and prevents binding of ATP and catalytic activity, thus mediating cyclin/CDK complex inhibition and negatively regulating cell proliferation. 24 It can also cause cell cycle arrest via binding and inhibition of PCNA. 25 As a tumour suppressor gene, CDKN1C is implicated in various human cancers and Beckwith-Wiedemann Syndrome. 26 Previous studies have tried to investigate the connection between CDKN1C and BC. Downregulation and hypermethylation of CDKN1C have been acknowledged prevalent in BC, which are related to a deterioration of prognosis. 27,28 With respect to therapeutic application, Y Ma et al. have revealed transcriptional upregulation of CDKN1C correlated with CDK inhibitors. 29 Interestingly, some antioxidant agents and wellness interventions were also reported to increase the expression levels in BC cells. 30,31 In contrast, through epigenetic mechanisms, CDKN1C can be suppressed by methylation and histone deacetylation, 32 multiple micro-RNAs and lncRNAs, 33,34 and specifically ERα signalling in hormone-responsive BC cells. 35 These observations all support the implication of CDKN1C in BC tumorigenesis.
Despite the fact that BC harbouring lower levels of CDKN1C tended to present with poor survival outcomes, 36 its role in BC progression and prognostic evaluation remained largely unknown. Therefore, we aimed to ascertain the CDKN1C expression and its relationship with prognosis in BC. Besides, the association between CDKN1C expression and enriched gene sets and pathways, as well as tumour immune microenvironment (TIM), were investigated in BC patients.
Recently, nomogram is widely conducted as a personalized tool to predict prognosis intuitively and precisely in various cancers. [37][38][39][40][41][42] Because this tool can rapidly calculate through easyto-use digital interfaces and more easily acquire prognostic information compared with traditional TNM staging. Moreover, nomograms can integrate biological and clinicopathological parameters to establish a prognostic model that generates a possibility of survival outcome.
Thus, to improve the accuracy of survival and therapeutic sensitivity assessment for BC patients, a novel prediction model integrating the expression of CDKN1C was established. We aimed to build a CDKN1C-based nomogram to predict overall survival (OS) and therapeutic responses in BC patients.

| Study samples from TCGA database
A total of 995 BC samples with specific CDKN1C expression levels were screened from the Cancer Genomes Atlas (TCGA) data portal. Patients without complete follow-up data or whose survival period was shorter than 1 month were excluded. Other clinical and pathological characteristics included in our analysis were as follows: age at diagnosis, T, N and TNM stage, tumour subtype and survival time. In the light of the optimal cut-off value of CDKN1C expression, study samples were classified into two groups with 786 patients in the low-expression group and 209 patients in the high-expression group.

| Differential expression and survival analysis of CDKN1C
First, differential gene expression analysis of CDKN1C was performed based on TCGA database via a Sangerbox tool, including 1098 BC and 113 normal breast tissues. In order to assess the effects of differentially expressed CDKN1C on prognosis, Kaplan-Meier survival analysis was utilized to estimate the OS of the TCGA patients. Subsequently, univariate and multivariate analyses were formulated to evaluate the prognostic effects of CDKN1C and other potential risk factors.

| Gene set enrichment analysis (GSEA) and immune infiltrates analysis of CDKN1C
GSEA was executed to investigate the functions of CDKN1C.
HALLMARK gene sets and KEGG pathways were considered as significantly enriched function annotations (p < 0.05, enrichment score >2.0). Furthermore, Through Tumor Immune Estimation Resource (TIMER) was applied to explore the association between CDKN1C expression and six essential TIICs. In order to determine whether the TIM differs markedly in low/high CDKN1C expression group, we To profile the variation of CDKN1C in BC, cBioportal was used to analyse the BC samples in TCGA Pan-cancer Atlas. The CDKN1C genetic alteration in Chinese BC patients was also analysed to make a comparison. Acquired from patients who were diagnosed as invasive BC at the GDPH, 589 BC samples underwent next-generation sequencing. It was approved by the Ethics Committee of GDPH and informed consents were obtained from all patients. Besides mRNA levels, differential protein expression between normal and BC tissues was validated by immunohistochemistry (IHC) staining obtained from the human protein atlas (HPA) database. HPA database retrieves transcriptomics data from TCGA and generates proteomics data. Therefore, using IHC analysis based on tissue microarrays, the transcriptomes of different human cancer types were visualized.

| Construction and evaluation of CDKN1Cbased prognostic nomogram
To assist in clinical decision making, an applicable and quantitative model is required for predicting OS for BC patients. In terms of the multivariate analysis above, CDKN1C, age, TNM stage and tumour subtype were proved to be independent prognostic factors in BC survival. Therefore, we introduced a prognostic model integrating CDKN1C expression level and other clinicopathological factors. The area under the curve (AUC) of the receiver operating characteristic (ROC) curve for OS was formulated to assess the discrimination of the CDKN1C-based model. As for its calibration ability, a calibration curve was drawn to verify. Finally, considering the potential for clinical utility, decision curve analysis (DCA) was used to assess the clinical practicability of the CDKN1C-based nomogram.

| Therapeutic responses estimation in BC patients
In the light of the optimal cut-off value of CDKN1C-based nomogram score, BC patients were divided into the high-risk and lowrisk groups. High-risk patients were characterized with higher scores and, therefore, worse predicted survival outcomes. Based on Genomics of Drug Sensitivity in Cancer, 'Prophetic' package was used to predict the therapeutic sensitivity. 6 common therapeutic agents for BC treatment were included. Their IC 50 was estimated between two groups.

| Statistical analysis
Descriptive analysis was conducted for clinicopathological features of included BC patients. Kaplan-Meier curve and log-rank test were adopted to plot and compare the survival curves. Univariate and multivariate analyses were used to verify the independent risk factors and construct a risk score formula and nomogram. Timedependent ROC curve analysis was exploited to evaluate the predictive accuracy of CDKN1C-based nomogram. The calibration ability of the CDKN1C-based nomogram was estimated using the calibration curve. Calibration plot was carried out to test the agreement between model-predicted and actual outcome. The appropriate cutoff values of CDKN1C expression level and CDKN1C-based nomogram score were confirmed using X-tile software, version 3.6.1 (Yale University, New Haven, CT, USA). 43,44 Statistical analyses were performed using R (Version 4.0.5) and a p-value <0.05 was considered statistically significant.

| Identification of CDKN1C signature in BC prognosis
On the transcriptomic level, TCGA database analysis found that CDKN1C was significantly overexpressed in the normal tissue, compared with multiple tumours, such as BC, bladder urothelial carcinoma, kidney carcinoma and lung adenocarcinoma (Figure 1). Since the aberrant low expression of CDKN1C in BC, we further explored its prognostic value. In accordance with previous findings, Kaplan-Meier survival analysis uncovered that BC patients with decreased levels of CDKN1C had a shortened OS (p = 0.00022, Figure 2A). The distribution of CDKN1C and survival status of the BC patients were shown in Figure 2B, indicating that its expression was positively cor-

| GSEA and genetic alteration analysis of CDKN1C
After exploring the correlation between CDKN1C expression levels and prognosis, GSEA was performed to clarify the biologic role of

| Relationship between CDKN1C expression and TIICs
TIMER ( Figure 5A), we observed that B cells were negatively correlated with CDKN1C (p = 9.47 × 10 −6 ). Simultaneously, a positive correlation existed between its expression and CD4+ T cells (p = 3.16 × 10 −4 ). It is noteworthy that with augmentation in CDKN1C expression, the tumour purity was significantly lower, indicating higher levels of TIICs. As shown in Figure 5B

| Development and assessment of CDKN1Cbased prognostic model
Now, that CDKN1C level is related with survival outcomes probably due to the biological process and immune microenvironment, it may assist in prognosis prediction. Consequently, we established a nomogram incorporating the CDKN1C expression, age, TNM staging and tumour subtype aiming to predict the OS in BC patients

| The role of nomogram in prediction of therapy sensitivity in BC patients
Finally, therapeutic response prediction was performed to compare BC patients in the low-risk and high-risk groups, with low and high nomogram scores respectively. In Figure 8, the estimated IC 50 of methotrexate, doxorubicin, paclitaxel, cisplatin, vinorelbine were significantly reduced in the low-risk group, which indicated better response to these therapeutic agents. Oppositely, lapatinib sensitivity was moderately better, when the nomogram scores were higher indicating worse prognosis.

| DISCUSS ION
Identification of a novel predictive signature is urgent for survival outcomes and therapeutic selection in BC survivors. CDKN1C, known as a BC suppressor, is transcriptionally and translationally expressed in the myoepithelial layer in BC. 45   In summary, a novel CDKN1C-based nomogram was developed to estimate the survival outcome of BC patients, which reflected good predictive accuracy and outperformed the TNM staging alone.
At the same time, we can find out the patients who may maximally benefit from specific antitumor agents, thus reducing the burden of overtreatment. Our study provided new insights into the role of CDKN1C, and facilitate prognosis and therapeutic responses prediction.

ACK N OWLED G EM ENTS
This study was supported by the National Natural Science Foundation of China (grant number 82002928). This study was supported by the Doctor Launch Fund of Guangdong Provincial People's hospital (grant number 2020bq11).

CO N FLI C T O F I NTE R E S T
The authors declare that they have no competing interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data sets used and analysed during the current study are available from the corresponding author on reasonable request.