Construction and Validation of a Protein Prognostic Model for Lung Squamous Cell Carcinoma

A protein-based prognostic model for pancreatic ductal adenocarcinoma: Construction and validation

BACKGROUND

Probing relevant proteomic biomarkers may facilitate effective pancreatic adenocarcinoma (PDAC) diagnosis, treatment and prevention. Here, we developed a protein-based prognostic model for PDAC by using relevant proteomic biomarkers data from The Cancer Genome Atlas (TCGA).

METHODS

We obtained PDAC's proteomic and clinical data from TCGA and used various analytical tools to identify differentially expressed proteins between normal and cancer tissues. We constructed our protein-based prognostic model and confirmed its accuracy using receiver operating characteristic curve and Kaplan-Meier survival analyses. We elucidated clinical factor-signature protein correlations by clinical correlation assessments and protein coexpression networks. We also used immunohistochemistry (protein expression assessment), Gene Set Enrichment Analysis (protein role identification) and CIBERSORT (infiltrating immune cell distribution assessment).

RESULTS

CIITA, BRAF_pS445, AR, YTHDF2, IGFBP2 and CDK1_pT14 were identified as PDAC-associated prognostic proteins. All risk scores calculated using our model provided 1-, 3-, 5-year survival probability at 70 % accuracy. The reliability of our model was validated by the GEO as well. In high- and low-risk groups, age, sex, T- and N- stage disparities were significant, and prognostic and coexpressed proteins correlated. PDAC tissues demonstrated significant CDK1_pT14 overexpression but significant BRAF_pS445, YTHDF2, and IGFBP2 underexpression. Downstream proteins of BRAF were validated by IHC. Low-risk tissues demonstrated more naïve B cells, eosinophils, activated NK cells and regulatory T cells, whereas high-risk tissues demonstrated more activated memory T cells, monocytes, neutrophils, dendritic cells and resting NK cells.

CONCLUSIONS

Our protein-based prognostic model for PDAC, along with six signature proteins, might aid in predicting PDAC prognosis and therapeutic targets.

Protein expression profiling identifies a prognostic model for ovarian cancer

Background

Owing to the high morbidity and mortality, ovarian cancer has seriously endangered female health. Development of reliable models can facilitate prognosis monitoring and help relieve the distress.

Methods

Using the data archived in the TCPA and TCGA databases, proteins having significant survival effects on ovarian cancer patients were screened by univariate Cox regression analysis. Patients with complete information concerning protein expression, survival, and clinical variables were included. A risk model was then constructed by performing multiple Cox regression analysis. After validation, the predictive power of the risk model was assessed. The prognostic effect and the biological function of the model were evaluated using co-expression analysis and enrichment analysis.

Results

394 patients were included in model construction and validation. Using univariate Cox regression analysis, we identified a total of 20 proteins associated with overall survival of ovarian cancer patients (p < 0.01). Based on multiple Cox regression analysis, six proteins (GSK3α/β, HSP70, MEK1, MTOR, BAD, and NDRG1) were used for model construction. Patients in the high-risk group had unfavorable overall survival (p < 0.001) and poor disease-specific survival (p = 0.001). All these six proteins also had survival prognostic effects. Multiple Cox regression analysis demonstrated the risk model as an independent prognostic factor (p < 0.001). In receiver operating characteristic curve analysis, the risk model displayed higher predictive power than age, tumor grade, and tumor stage, with an area under the curve value of 0.789. Analysis of co-expressed proteins and differentially expressed genes based on the risk model further revealed its prognostic implication.

Conclusions

The risk model composed of GSK3α/β, HSP70, MEK1, MTOR, BAD, and NDRG1 could predict survival prognosis of ovarian cancer patients efficiently and help disease management.

Identification and Validation of a Prognostic 5-Protein Signature for Biochemical Recurrence Following Radical Prostatectomy for Prostate Cancer

Background: Biochemical recurrence (BCR) is an indicator of prostate cancer (PCa)-specific recurrence and mortality. However, there is a lack of an effective prediction model that can be used to predict prognosis and to determine the optimal method of treatment for patients with BCR. Hence, the aim of this study was to construct a protein-based nomogram that could predict BCR in PCa.

Methods: Protein expression data of PCa patients was obtained from The Cancer Proteome Atlas (TCPA) database. Clinical data on the patients was downloaded from The Cancer Genome Atlas (TCGA) database. Lasso and Cox regression analyses were conducted to select the most significant prognostic proteins and formulate a protein signature that could predict BCR. Subsequently, Kaplan–Meier survival analysis and Cox regression analyses were conducted to evaluate the performance of the prognostic protein-based signature. Additionally, a nomogram was constructed using multivariate Cox regression analysis.

Results: We constructed a 5-protein-based prognostic prediction signature that could be used to identify high-risk and low-risk groups of PCa patients. The survival analysis demonstrated that patients with a higher BCR showed significantly worse survival than those with a lower BCR (p < 0.0001). The time-dependent receiver operating characteristic curve showed that the signature had an excellent prognostic efficiency for 1, 3, and 5-year BCR (area under curve in training set: 0.691, 0.797, 0.808 and 0.74, 0.739, 0.82 in the test set). Univariate and multivariate analyses indicated that this 5-protein signature could be used as independent prognosis marker for PCa patients. Moreover, the concordance index (C-index) confirmed the predictive value of this 5-protein signature in 3, 5, and 10-year BCR overall survival (C-index: 0.764, 95% confidence interval: 0.701–0.827). Finally, we constructed a nomogram to predict BCR of PCa.

Conclusions: Our study identified a 5-protein-based signature and constructed a nomogram that could reliably predict BCR. The findings might be of paramount importance for the prediction of PCa prognosis and medical decision-making.

Subjects: Bioinformatics, oncology, urology.

Identification of a Novel Protein-Based Signature to Improve Prognosis Prediction in Renal Clear Cell Carcinoma

Background

Clear cell renal cell carcinoma (ccRCC) is one of the most common types of malignant adult kidney cancer, and its incidence and mortality are not optimistic. It is well known that tumor-related protein markers play an important role in cancer detection, prognosis prediction, or treatment selection, such as carcinoembryonic antigen (CEA), programmed cell death 1 (PD-1), programmed cell death 1 ligand 1 (PD-L1), and cytotoxic T lymphocyte antigen 4 (CTLA-4), so a comprehensive analysis was performed in this study to explore the prognostic value of protein expression in patients with ccRCC.

Materials and Methods

Protein expression data were obtained from The Cancer Proteome Atlas (TCPA), and clinical information were downloaded from The Cancer Genome Atlas (TCGA). We selected 445 patients with complete information and then separated them into a training set and testing set. We performed univariate, least absolute shrinkage and selection operator (LASSO) Cox analyses to find prognosis-related proteins (PRPs) and constructed a protein signature. Then, we used stratified analysis to fully verify the prognostic significance of the prognostic-related protein signature score (PRPscore). Besides, we also explored the differences in immunotherapy response and immune cell infiltration level in high and low score groups. The consensus clustering analysis was also performed to identify potential cancer subgroups.

Results

From the training set, a total of 233 PRPs were selected, and a seven-protein signature was constructed, including ACC1, AR, MAPK, PDK1, PEA15, SYK, and BRAF. Based on the PRPscore, patients could be divided into two groups with significantly different overall survival rates. Univariate and multivariate Cox regression analyses proved that this signature was an independent prognostic factor for patients (P < 0.001). Moreover, the signature showed a high ability to distinguish prognostic outcomes among subgroups, and the low score group had a better prognosis (P < 0.001) and better immunotherapy response (P = 0.003) than the high score group.

Conclusion

We constructed a novel protein signature with robust predictive power and high clinical value. This will help to guide the disease management and individualized treatment of ccRCC patients.