Exploring the survival benefits of surgical treatment for pancreatic adenocarcinoma using the DeepSurv neural network model

Pancreatic adenocarcinoma (PaC) patients with positive lymph nodes (PLNs) have a dismal prognosis and lack a specific prognostic stage. This study aimed to construct a nomogram for the prediction of overall survival (OS) in these patients. A total of 1,340 patients screened from the Surveillance, Epidemiology, and End Results database were included and randomly divided at a ratio of 7:3 into a training set (n=940) and an internal validation set (n=400). Cox regression analyses were conducted to select independent predictors in the training set, and a nomogram was constructed. The model was verified in the internal validation set and in an external validation set, which comprised 64 patients from a Chinese institute. Six independent prognostic factors (age at diagnosis, tumor grade, lymph node ratio, T stage, radiotherapy, and chemotherapy) were identified in PaC patients with PLNs and were entered into the nomogram. The final model had a higher C-index for predicting OS than the American Joint Committee on Cancer-8th edition staging system (training set: 0.658 vs. 0.546; internal validation set: 0.661 vs. 0.546; external validation set: 0.691 vs. 0.581). The 1-, 2-, and 3-year area under the receiver operating characteristic curve values indicated better discrimination power for the established nomogram with respect to the prediction of OS in the training, internal validation, and external validation sets than for the American Joint Committee on Cancer-8th edition staging system. Furthermore, the nomogram performed well in both calibration and decision curve analyses (DCA) of clinical applicability. OS in PaC patients with PLNs was significantly distinguished among the three risk groups stratified according to the nomogram score (P<0.001). The well-calibrated nomogram was determined to be extremely efficient in predicting survival, and defining a high-risk population based on the nomogram score among PaC patients with PLNs after surgery.

BackgroundLymph node ratio (LNR; positive/harvested lymph nodes) was identified as overall survival predictor in several cancers, including pancreatic adenocarcinoma. It remains unclear if LNR is predictive of overall survival in pancreatic adenocarcinoma patients staged pN2. This study assessed the prognostic overall survival role of LNR in pancreatic adenocarcinoma patients in relation with lymph node involvement.MethodsA retrospective international study in six different centers (Europe and United States) was performed. Pancreatic adenocarcinoma patients who underwent pancreatoduodenectomy from 2000 to 2017 were included. Patients with neoadjuvant treatment, metastases, R2 resections, or missing data regarding nodal status were excluded. Survival curves were calculated using Kaplan-Meier method and compared using log-rank test. Multivariable Cox regressions were performed to find independent overall survival predictors adjusted for potential confounders.ResultsA total of 1,327 patients were included. Lymph node involvement (pN+) was found in 1,026 patients (77%), 561 pN1 (55%) and 465 pN2 (45%). Median LNR in pN+ patients was 0.214 [interquartile range (IQR): 0.105–0.364]. On multivariable analysis, LNR was the strongest overall survival predictor in the entire cohort [hazard ratio (HR) =5.5; 95% confidence interval (CI): 3.1–9.9; P<0.001] and pN+ patients (HR =3.8; 95% CI: 2.2–6.6; P<0.001). Median overall survival was better in patients with LNR <0.225 compared to patients with LNR ≥0.225 in the entire cohort and pN+ patients. Similar results were found in pN2 patients (worse overall survival when LNR ≥0.225).ConclusionsLNR appeared as an important prognostic factor in patients undergoing surgery for pancreatic adenocarcinoma and permitted to stratify overall survival in pN2 patients. LNR should be routinely used in complement to tumor-node-metastasis (TNM) stage to better predict patient prognosis.

Background: Pancreatic adenocarcinoma (PAAD) is one of the most aggressive tumors of the digestive tract, with low surgical resection rate and insensitivity to radiotherapy and chemotherapy. Existing evidence suggests that regulation of ferroptosis can induce PAAD cell death, inhibit tumor growth, and may synergistically improve the sensitivity of other antitumor drugs. However, there is little of systematic research on iron metabolism-related genes in PAAD. In this study, a risk-score system of PAAD iron metabolism-related genes was designed and tested, and verified to be robust.Materials and Methods: The TCGA database was used to download 177 PAAD patients’ message RNA (mRNA) expression profiles and clinical characteristics. By identifying dysregulated iron metabolism-related genes between PAAD related tissues and adjacent normal tissues, univariate Cox proportional hazards regression and LASSO regression algorithm were used to establish prognostic risk-score system and construct nomogram to estimate the 1-, 2-, 3-year survival in PAAD patients. Finally, selected genes were validated by quantitative PCR (q-PCR).Results: A 9-gene related to iron metabolism risk-score system of PAAD was constructed and validated. The clinicopathological characteristics of age, histologic grade, pathologic stage, T stage, residual tumor, and primary therapy outcome were all worse in patients with a higher risk-score. Further, immunohistochemistry results of SLC2A1, MBOAT2, XDH, CTSE, MOCOS, and ATP6V0A4 confirmed that patients with higher expression are more malignant. Then, a nomogram with 9-gene risk score system as a separate clinical factor was utilized to foretell the 1-, 2-, 3-year overall survival rate of PAAD patients. Results of q-PCR showed that 8 of the 9 genes screened were significantly up-regulated in at least one PAAD cell line, and one gene was significantly down-regulated in three PAAD cell lines.Conclusion: To conclude, we generated a nine-gene system linked to iron metabolism as an independent indicator for predicting PAAD prognosis, therefore presenting a possible prognostic biomarker and potential treatment targets for PAAD.

Pancreatic adenocarcinoma (PAAD) is one of the most malignancies with a poor prognosis. The immune phenotype of pancreatic cancer is remarkably complex, presenting a significant challenge to the efficacy of immunotherapy. Tertiary lymphoid structures (TLSs) are active zones composed of various immune cells, which reflect immune responses against cancer. TLSs have emerged as critical biomarkers for predicting immunotherapy outcomes, potentially reflecting alterations in the metabolic characteristics of the host. Accumulating evidence indicates that 18F-fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT) may shed light on cancer immunology. The tumor environment is known to be associated with the metabolic activity and antitumor immunity of immune cells. A total of 266 patients with PAAD who underwent 18F-FDG PET/CT followed by pancreatectomy were enrolled in this study, and the presence of TLSs was identified in 66 patients by HE staining, and fluorescent multiplex immunohistochemistry (mIHC) was employed to analyze the cellular composition of TLSs. Elevated maximum standard uptake values (SUVmax) showed significant correlation with TLS presence in PAAD patients. However, SUVmax value showed no correlation with overall survival in PAAD patients. Immunohistochemistry (IHC) staining demonstrated a statistically significant correlation between higher SUVmax levels and increased CD4 + and CD8 + tumor-infiltrating lymphocytes. In PAAD patients with TLS, CD4⁺ T-cell infiltration levels showed a positive correlation with SUVmax. In our study, a TLS signature containing 12-chemokine was proposed to determine high TLS score and low TLS score group. The high TLS score group exhibited increased infiltration of naïve/memory/activated B cells, central memory/activated CD4⁺/CD8⁺ T cells, Tregs, Th1/Th2/Th17 subsets, and neutrophils. Furthermore, GLUT3 and GLUT10 expression consistently correlated with TLS scores in both TCGA and CPTAC cohorts. GLUT3 and GLUT10 were found to be upregulated in the high TLS score group. A more inflamed immune infiltrative landscape was characterized in the high TLS score group with a high proportion of multiple immune cells. Increased FDG uptake is often induced by overexpression of glucose transporters (GLUT) and glucometabolic genes. Overexpression of GLUT3 and GLUT10 was found to be positively correlated with higher infiltration levels as well as immune activation of CD4+ and CD8+ T cells. To conclude, our study showed an association between SUVmax on 18F-FDG PET/CT and TLS presence in PAAD. FDG uptake may reflect local immune responses and glucose metabolism associated with TLS presence in PAAD. 18F-FDG PET/CT may serve as a non-invasive and aided tool for assessing the presence of TLS in tumor immune environment and predicting antitumor immunity in PAAD patients.Supplementary InformationThe online version contains supplementary material available at 10.1007/s00262-025-04205-x.

Background: Recently, RNA-binding proteins (RBPs) were reported to interact with target mRNA to regulate gene posttranscriptional expression, and RBP-mediated RNA modification can regulate the expression and function of proto-oncogenes and tumor suppressor genes. We systematically analyzed the expression of RBPs in pancreatic adenocarcinoma (PAAD) and constructed an RBP-associated prognostic risk model.Methods: Gene expression data of normal pancreatic samples as well as PAAD samples were downloaded from TCGA-PAAD and GTEx databases. Wilcoxon test and univariate Cox analysis were, respectively, applied to screen differential expression RBPs (DE-RBPs) and prognostic-associated RBPs (pRBPs). Functional enrichment was analyzed by GO, KEGG, and GSEA. Protein–protein interaction (PPI) network was constructed by STRING online database. Modeling RBPs were selected by multivariate Cox analysis. Kaplan–Meier survival and Cox analysis were applied to evaluate the effects of risk score on the overall survival of PAAD patients. ROC curves and validation cohort were applied to verify the accuracy of the model. Nomogram was applied for predicting 1-, 3-, and 5-year overall survival (OS) of PAAD patients. At last, modeling RBPs were further analyzed to explore their differential expression, prognostic value, as well as enrichment pathways in PAAD.Results: RBPs (453) were differentially expressed in normal and tumor samples, besides, 28 of which were prognostic associated. DE-RBPs (453) are functionally associated with ribosome, ribonuclease, spliceosome, etc. Eight RBPs (PABPC1, PRPF6, OAS1, RBM5, LSM12, IPO7, FXR1, and RBM6) were identified to construct a prognostic risk model. Higher risk score not only predicted poor prognosis but also was an independent poor prognostic indicator, which was verified by ROC curves and validation cohort. Eight modeling RBPs were confirmed to be significantly differentially expressed between normal and tumor samples from RNA and protein level. Besides, all of eight RBPs were related with overall survival of PAAD patients.Conclusions: We successfully constructed an RBP-associated prognostic risk model in PAAD, which has a potential clinical application prospect.

Dysregulation of Hippo signaling is observed in pancreatic adenocarcinoma (PAAD). Moreover, overactivation of YAP is crucial for tumor progression. Although the inhibitory phospho-cascade is functional, the reason for YAP hyperactivation in PAAD remains unclear. Recent studies have revealed that the ubiquitin modification of YAP also plays an important role in the Hippo/YAP axis and cancer progression. To gain a better understanding of the potential mechanisms underlying the ubiquitination and deubiquitination of YAP, we carried out siRNA screening for critical deubiquitinases in PAAD. By using a deubiquitinase (DUB) library, we identified valosin-containing protein-interacting protein 1 (VCPIP1) as an important effector of YAP function and PAAD progression. Inhibition of VCPIP1 hampered PAAD progression via Hippo signaling. Clinical data revealed that VCPIP1 was elevated in PAAD and correlated with poor survival in PAAD patients. Biochemical assays demonstrated that VCPIP1 interacted with YAP, inhibiting K48-linked polyubiquitination and thereby increasing YAP stability. YAP directly binds to the VCPIP1 promoter region, enhancing its transcription in PAAD. Our study revealed a forward feedback loop between VCPIP1 and Hippo signaling in PAAD, indicating that VCPIP1 is a potential therapeutic drug target in PAAD.

Increased expression of zinc transporter ZIP4, ZIP11, ZnT1, and ZnT6 predicts poor prognosis in pancreatic cancer

Erlotinib, in combination with gemcitabine, has shown clinical benefits in pancreatic adenocarcinoma patients. The presence of EGFR mutations and increased EGFR copy numbers in pancreatic adenocarcinoma was explored. Sixty-six pancreatic cancer patients were included in the analysis. The EGFR mutation was analyzed by DNA sequencing of exons 18-21 in the tyrosine kinase domain. KRAS mutation was analyzed by sequencing codons 12, 13, and 61. Quantitative real-time polymerase chain reaction was performed to analyze the copy number of EGFR. In the current study the EGFR mutation was harbored in only 1 (1.5%) of the 66 inoperable or metastatic pancreatic adenocarcinoma patients. Amino acid substitution was detected in exon 20 of the EGFR gene. Increased EGFR copy numbers (> or =3.0 per cell) were detected in 26 (41%) patients. There was only 1 patient, who had a highly increased EGFR copy number (> or =6.0 per cell), who died, 2.1 months from the date of diagnosis. The EGFR amplification did not significantly influence survival in pancreatic adenocarcinoma patients (P = .935). Thirty-two (49%) of the 65 pancreatic adenocarcinomas examined harbored a point mutation in codons 12 (n = 31) and 61 (n = 1) of the KRAS gene. The presence of a point mutation in codon 12 adversely influenced survival of pancreatic cancer patients (P = .030). The incidence of somatic mutations in the tyrosine kinase domains of EGFR was very low and the increased gene copy number of EGFR did not significantly influence survival.

PurposePancreatic adenocarcinoma (PAAD) is a devastating disease with high mortality and morbidity. Matrix metalloproteinase 28 (MMP28) has been associated with carcinogenesis of many human cancers. However, little is known about the potential prognostic value and underlying regulatory mechanisms of MMP28 in PAAD.MethodsThe relationship between MMP28 expression level and various clinicopathological parameters was analyzed in TCGA-PAAD cohorts. MMP28-correlated genes in the TCGA-PAAD cohort were identified and enrichment analysis according to the Gene Ontology and Kyoto Encyclopedia of Genes and Genomes was conducted using LinkedOmics. Protein–protein interaction and transcription factors-miRNA co-regulatory networks were constructed with the use of NetworkAnalyst. Then, the distribution of immune cells related to MMP28 expression in blood was analyzed using the Human Protein Atlas, and the tumor microenvironment of PAAD was analyzed by the TIMER 2.0 database. To investigate the biological function of MMP28 in PAAD, siRNA was constructed to knock down the MMP28 gene in vitro.ResultsHigh MMP28 expression is associated with poor overall survival and disease-free survival in PAAD patients. The expression of MMP28 in PAAD is most significantly correlated with KRT19, IL1RN, and ANXA2 genes. Network analysis revealed that MIR-181 family, TAFs, and CDC6 are potential regulators of MMP28. Furthermore, naive CD4+ T cell, naive CD8+ T cell, and mucosal-associated invariant T cell enrichment in blood were correlated with MMP28 expression. Furthermore, high MMP28 expression was correlated with a decrease in B cell, naive CD4+ T cell, naive CD8+ T cell, and endothelial cell presence in the tumor microenvironment in PAAD. Finally, genetic knockdown of MMP28 could restrain the proliferation, migration, and invasion of PAAD cells.ConclusionOur findings indicate that high MMP28 expression in PAAD is associated with cancer progression, invasion, and metastasis. Hence, MMP28 might serve as an independent prognostic biomarker and a prospective therapeutic target for PAAD.

LDL receptor-related proteins (LRPs) are transmembrane receptors involved in endocytosis, cell-signaling, and trafficking of other cellular proteins. Considerable work has focused on LRPs in the fields of vascular biology and neurobiology. How these receptors affect cancer progression in humans remains largely unknown. Herein, we mined provisional databases in The Cancer Genome Atlas (TCGA) to compare expression of thirteen LRPs in ten common solid malignancies in patients. Our first goal was to determine the abundance of LRP mRNAs in each type of cancer. Our second goal was to determine whether expression of LRPs is associated with improved or worsened patient survival. In total, data from 4,629 patients were mined. In nine of ten cancers studied, the most abundantly expressed LRP was LRP1; however, a correlation between LRP1 mRNA expression and patient survival was observed only in bladder urothelial carcinoma. In this malignancy, high levels of LRP1 mRNA were associated with worsened patient survival. High levels of LDL receptor (LDLR) mRNA were associated with decreased patient survival in pancreatic adenocarcinoma. High levels of LRP10 mRNA were associated with decreased patient survival in hepatocellular carcinoma, lung adenocarcinoma, and pancreatic adenocarcinoma. LRP2 was the only LRP for which high levels of mRNA expression correlated with improved patient survival. This correlation was observed in renal clear cell carcinoma. Insights into LRP gene expression in human cancers and their effects on patient survival should guide future research.

Pancreatic adenocarcinoma (PAAD) is a leading cause of tumor-related mortality. Identifying potential prognostic risk genes is crucial for predicting the overall survival of PAAD patients. In this study, we constructed and validated a 24-gene risk score. This risk score stratifies PAAD patients into low-risk and high-risk groups. The model demonstrated excellent prognostic accuracy at different follow-up times (1-year AUC: 0.81, 2-year AUC: 0.85, 3-year AUC: 0.92). PAAD patients from 3 GEO datasets were categorized into low-risk and high-risk groups, with survival analysis revealed significant differences in survival rates between the 2 groups (P < .01). Multivariate analysis identified 2 independent risk factors, namely, N stage (HR 2.026, 95% CI 1.139-3.603, P = .016) and the 24-gene risk score (HR 0.239, 95% CI 0.148-0.385, P < .001). The performance of the nomogram in the TCGA database is commendable (AUC for 1-year, 2-year, and 3-year survival rates = 0.76, 0.77, and 0.86, respectively). In essence, our work establishes a 24-gene risk score and nomogram to facilitate clinicians in predicting the prognosis of individual PAAD patients.

Postoperative recurrence of pancreatic adenocarcinoma (PAAD) remains a major challenge. This study aims to establish and validate a lipid metabolism-related prognostic model to predict recurrence in PAAD patients. The TCGA-PAAD database was used to establish a training cohort, which was validated using the ICGC database and multiple center cohorts. A prognostic model based on LASSO Cox regression and a nomogram was developed and further validated. Among 196 lipid metabolism-related genes, four were selected for the prognostic model. Patients were stratified into high- and low-risk groups based on the risk score. Univariate and multivariate Cox regression analyses showed that tumor site, T stage, N stage, M stage, and risk score were significantly associated with progression-free interval (PFI). High-risk patients had worse PFI, overall survival (OS), and disease-specific survival (DSS) (all P < 0.05). Time-dependent ROC and decision curve analyses confirmed the superior diagnostic capacity of the nomogram. GSEA revealed enrichment in G2M checkpoint, glycolysis, estrogen response, and hypoxia pathways for the high-risk group. Additionally, high-risk scores correlated with poor immune infiltration, gene mutations, and tumor mutational burden (TMB). Single-cell analysis suggested that risk genes interact with various cell types to promote PAAD progression. A novel lipid metabolism-related prognostic model was developed and validated to predict recurrence and survival in PAAD patients, with strong accuracy and stability.

Purpose: Population-based studies indicated that prognosis of pancreatic adenocarcinoma (PAC) is worse in black patients compared to other races. Nonetheless, survival probabilities can change over time based on number of years (yr.) already survived by patients; a concept called conditional survival. This study explored the dynamic changes in risk according to patient characteristics, particularly race, on survival of PAC patients using cancer-specific survival (CSS) estimates. Methods: The Surveillance, Epidemiology, and End Results (SEER) database was queried for data on adult patients with non-metastatic PAC, diagnosed between 1988 and 2010. Patient characteristics, such as age, race, tumor grade, and stage were collected at the time of diagnosis. CSS probabilities, as well as Cox proportional hazard ratios (HRs), were computed at the time of diagnosis (Actuarial CSS and baseline HR), and after already surviving 1 to 6 yr. after diagnosis (Conditional CSS and HR). Harrell’s concordance index (C-index) was used to measure the cross-validation accuracy of the Cox models. Results: Our search retrieved data on 20,491 patients, with a mean age at diagnosis of 67.2 yr. Most of the patients were White (81.6%), followed by Black (12%) and Asians/Pacific Islander (6.4%). The stage was T1-2N0M0 in 15.9%, T3-4N0M0 in 41.8%, and T1-4N1M0 in 42.3% of patients. The 3-yr actuarial CSS calculated from time of diagnosis was significantly different across racial groups, at 11%, 10%, and 13% for Whites, Blacks, and Asians, respectively (P &amp;lt; 0.01). Conversely, for patients who already survived 1 yr. after diagnosis, the probability of surviving an additional 2 yr. was similar across races, at 26.2%, 27.1%, and 29.9%, for Whites, Blacks, and Asians, respectively (P = 0.218). As patients survived for longer periods of time following diagnosis, conditional CSS estimates increased similarly across different races; for White, Black, and Asian patients who already survived 3 yr. after diagnosis, the probability of surviving an additional 2 yr. was 62.6%, 60.5%, and 62.1%, respectively (P = 0.532). In multivariate cox models, the prognostic effect of race lost significance if patients already survived ≥1 yr. after diagnosis (Baseline HR = 1.114, 95%CI [1.045- 1.187], mean C-index = 67%; conditional HR at 1 yr = 1.015, 95%CI [0.919- 1.12], mean C-index = 60%). The prognostic effect of tumor grade, site, and age lost significance if patients already survived ≥2, ≥4, and ≥6 yr. after diagnosis, respectively. Tumor stage maintained its prognostic significance over time (conditional HR at 6 yr = 1.522, 95%CI [1.049- 2.208], mean C-index = 59%). Conclusion: Racial disparities in survival outcomes exist at the time of diagnosis for PAC patients. However, the survival impact of these disparities does not seem to persist over time. Other variables, such as age, tumor grade, stage, and treatment received should be taken into account when predicting future prognosis of PAC patients who have already survived ≥ 1 yr. after diagnosis. Citation Format: Anas M Saad, Maha AT Elsebaie, Mohamed Amgad, Muneer J Al-Husseini, Kyrillus S Shohdy, Omar Abdel-Rahman. Racial disparities in pancreatic adenocarcinoma survival. Do they exist for patients who already survived their first year? [abstract]. In: Proceedings of the Twelfth AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2019 Sep 20-23; San Francisco, CA. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2020;29(6 Suppl_2):Abstract nr B110.

Pancreatic cancer is a common malignant tumor of the digestive system, with insidious onset, difficult early diagnosis, easy metastasis, and poor prognosis. N6-methyladenosine (m6A) and long non-coding RNA (lncRNA) play important roles in the prognostic value and immunotherapy response of pancreatic adenocarcinoma (PAAD). Therefore, it is crucial to recognize m6A-related-lncRNAs in PAAD patients. In this study, m6A-related lncRNAs were obtained by coexpression analysis. Univariate, the Least Absolute Shrinkage, and Selection Operator (LASSO) and multivariate Cox regression analyses were performed to construct m6A-related lncRNA prognostic models. Kaplan&amp;ndash;Meier analysis, principal component analysis, feature-rich annotation, and nomogram were used to analyze the accuracy of risk models. Potential drugs targeting this model are also discussed. A prognostic model based on m6A-related lncRNAs was constructed, potential drugs targeting this m6A-related lncRNAs feature were discovered, and the relationship with immunotherapy response was studied. Finally, a nomogram was established to predict survival in PAAD patients. This m6A-based lncRNAs risk prognostic model may be promising for clinical prediction of prognosis and immunotherapy response in PAAD patients.

Enzyme replacement improves survival among patients with pancreatic cancer: Results of a population based study