Identification of a costimulatory molecule-based signature to predict prognostic risk of pancreatic adenocarcinoma

Figures & data

Table 1. Primers used in RT-PCR.

Genes	Primers
FAS-F	TCTGGTTCTTACGTCTGTTGC
FAS-R	CTGTGCAGTCCCTAGCTTTCC
TNFSF10-F	TGCGTGCTGATCGTGATCTTC
TNFSF10-R	GCTCGTTGGTAAAGTACACGTA
TNFSF9-F	GGCTGGAGTCTACTATGTCTTCT
TNFSF9-R	ACCTCGGTGAAGGGAGTCC
EDA-F	AGATGGCCCAGTTAAAAACAAGA
EDA-R	CAGGTGGTCCCATAACAGTTG
NGFR-F	CCTACGGCTACTACCAGGATG
NGFR-R	CACACGGTGTTCTGCTTGT

Table 2. Univariate Cox analysis of costimulatory molecule genes in TCGA Cohort of PAAD.

Official symbol	Aliases	Family	HR	95%CI	P value
TNFSF10	TRAIL, CD253	TNFSF	1.6098	(1.26–2.05)	0.0001***
TNFRSF10A	TRAILR1, CD261	TNFRSF	1.4514	(1.16–1.81)	0.0003***
TNFRSF21	DR6, CD358	TNFRSF	1.5780	(1.20–2.07)	0.0003***
TNFRSF10B	TRAILR2, CD262	TNFRSF	1.4555	(1.13–1.88)	0.0020**
EDA	EDA-A1, EDA-A2	TNFSF	0.7487	(0.61–0.92)	0.0045**
FAS	TNFRSF6, CD95	TNFRSF	1.4495	(1.11–1.88)	0.0049**
TNFSF11	RANKL, CD254	TNFSF	1.1895	(1.04–1.36)	0.0065**
VTCN1	B7-H4	B7	1.1285	(1.02–1.25)	0.0147*
HHLA2	B7-H5	B7	1.1234	(1.01–1.25)	0.0175*
TNFRSF11B	OPG	TNFRSF	1.1487	(1.02–1.29)	0.0196*
LTBR	TNFRSF3	TNFRSF	1.4377	(1.07–1.93)	0.0210*
EDAR	EDA-A1R	TNFRSF	1.1347	(1.01–1.27)	0.0214*
TNFSF9	4-1BB-L, CD137L	TNFSF	1.1588	(1.01–1.33)	0.0338*
NGFR	TNFRSF16, CD271	TNFRSF	0.8796	(0.78–0.99)	0.0341*
TNFSF13	APRIL, CD256	TNFSF	1.2395	(0.99–1.55)	0.0439*
TNFRSF12A	FN14, TWEAKR, CD266	TNFRSF	1.2179	(0.99–1.50)	0.0501*

***P value < 0.001, **p value < 0.01, *p value < 0.05.

Figure 1. Identification of the costimulatory molecule-based signature (CMS). The distribution of risk score, survival status, and the five-gene expression panel (A). Area under the curve (AUC) curves of different models (B).

Figure 2. Reverse Transcription-Polymerase Chain Reaction (RT-PCR) used to detect the differential expression of five significant correlation genes in PAAD tissue samples and paracancer tissue samples. **P < 0.01 and ***P < 0.001.

Table 3. The result of model construction (4-fold cross validation).

	AUC	Median Y	Formula
Cvlist1	0.51	6.5646	Risk score = (0.6514*TNFRSF21) + (−0.5062*EDA) +(0.5338*FAS) + (−0.2941*TNFRSF12A)
Cvlist2	0.73	4.7656	Risk score = (0.5758*TNFSF10) + (−0.5364*EDA) + (0.5161*FAS) + (−0.2346*TNFSF9) + (−0.1601*NGFR)
Cvlist3	0.68	2.0939	Risk score = (0.4518*TNFSF10) + (−0.2934*EDA) + (−0.1204*NGFR)
Cvlist4	0.60	6.2182	Risk score = (0.4316*TNFRSF21) + (−0.6319*EDA) +(0.6467*FAS) + (−0.1761*TNFSF9)
Cvlist all	0.70	6.4433	Risk score = (0.2835*TNFSF10) + (0.2798*TNFRSF21) + (−0.4860*EDA56070298854)

Figure 3. Kaplan–Meier curve analysis. The overall survival rate in the high- and low-risk groups of the subgroup, with total patients (A), female patients (B), male patients (C), patients with age ≥ 60 (D), patients with age < 60 (E) and patients at early stage (F), were estimated based on the risk score.

Figure 4. Evaluation of the value of CMS. Survival analysis in both total samples (A), valid samples (B), and GSE57495 dataset (C). The proportional-hazards assumption of the nomogram model (C). The area under the curve (AUC) of genes in model and previous studies (E).

Table 4. Univariable and multivariable Cox regression analysis of the costimulatory molecule-based signature and survival in TCGA dataset.

	Univariable analysis			Multivariable analysis
Variable	HR	95%CI	P value	HR	95%CI	P value
Risk score	1.64	(1.3113–2.0620)	7.76E-06***	1.66	(1.2645–2.1828)	2.70E-04***
Age (≥ 60 or < 60)	1.03	(1.0075–1.0496)	6.20E-03***	1.03	(1.0060–1.0494)	1.20E-02*
Lymph count	1.00	(0.9765–1.0195)	8.38E-01	1.01	(0.9842–1.0364)	4.51E-01
Histologic	1.33	(1.0437–1.6946)	2.49E-02*	1.05	(0.7984–1.3807)	7.27E-01
M (0,1, X)	0.92	(0.7495–1.1369)	4.53E-01	0.96	(0.7737–1.1963)	7.28E-01
N (0,1,2, X)	1.46	(1.0039–2.1328)	4.62E-02*	1.90	(1.1263–3.2014)	1.61E-02*
T (1,2,3,4)	1.13	(0.8342–1.5409)	4.23E-01	1.04	(0.5502–1.9795)	8.96E-01
Gender (Male or Female)	0.82	(0.5448–1.2257)	3.29E-01	0.79	(0.5158–1.2068)	2.74E-01
Race	0.07	(1.0590–0.9666)	9.99E-01	2.06	(0.5235–8.0712)	3.02E-01
Tumor stage (I, II, III, IV)	1.21	(0.8435–1.7429)	3.09E-01	0.79	(0.4488–8.9766)	5.21E-01

***P value < 0.0001, **p value < 0.01.

Figure 5. Differential gene expression analysis in CMS. Principal component analysis described the transcriptome distribution of the sample as a whole (A). Volcano plot of differentially expressed genes (B). GO and KEGG analyses of the highly expressed genes (C) and lowly expressed genes (D). The number of DEGs is shown on the X axis, while GO terms or KEGG pathway terms are shown on the Y axis.

Figure 6. CMS-related immune cell infiltration and inflammatory activities. The average proportion of immune cells in the samples (A). Significant differences in the proportion of three immune cells between high- and low-risk patients (B). *P < 0.05 and **P < 0.01.

Figure 7. CMS-related somatic mutation. The mutant genes in the low- (A) and high-risk groups (B).

Data availability statement

The data that support the findings of this study are available in UCSC Xena of TCGA database at http://xena.ucsc.edu, reference (Zweig et al. 2008), as well as GSE57495 dataset in GEO database at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE57495 (doi: 10.1371/journal.pone.0133562), and the raw data of RT–PCR are available at https://figshare.com/s/9de1b322ce2af8fa3b84.