Similar to ccRCC, pRCC patients are usually diagnosed at an advanced stage or metastasis. The major treatment for local pRCC is surgical operation. However, approximately 40% patients reoccur after surgical resection, resulting in a worse prognosis [21]. In recent years, increasing researches of cancers focused on molecular characteristic in early diagnosis and prognosis improvement. Of note, lncRNAs have been revealed to function as main regulators in various biological processes [22, 23]. Notably, it was believed that cancer-related lncRNAs may be helpful for researchers to understand the mechanism of cancer progression and develop efficient measures to improve the prognosis [24]. And cancer-related lncRNAs have been described in some cancer types, such as esophageal adenocarcinoma [25] and prostate cancer [26]. In the present study, we obtained 1872 caner-related lncRNAs through integrating Immlnc and TCGA databases and screened 367 significant DECRLs in pRCC. Cox regression analysis suggested that 26 significant DECRLs were related to the survival of patients with pRCC, among which 10 lncRNAs (RP11-573D15.8, LINC01317, RNF144A-AS1, TFAP2A-AS1, LINC00702, GAS6-AS1, RP11-400K9.4, LUCAT1, RP11-63A11.1, RP11-156L14.1) were enrolled into the risk model to generate a cancer-related lncRNA signature in pRCC for survival prediction. More surprisingly, the signature was a prognostic risk factor from other clinical features in pRCC.
Based on this signature, survival curves suggested that patients in high risk group had a poorer survival rate than patients in low risk group. Additionally, ROC curves exhibited that the precision of this signature was more than 75% for survival prediction at different years. These results were validated in testing set and combination set and similar outcomes were observed, suggesting the signature was a forceful tool for predicting prognosis. What’s more, this cancer-related lncRNA signature was also a forceful tool for prognosis prediction in different classes of some clinicopathological features, including age, gender, and tumor stage.
Other prognostic signatures in pRCC was previously reported. For instance, Wang et al. [27] established a prognostic signature with 15 immune-related genes. Although prognosis of patients could be predicted by their signature, the accuracy of their prognostic signature in predicting prognosis in 1, 3 and 5 years was lower than that of ours (Table 3). In addition, Gao et al. [28] also grouped pRCC patients into training set and testing set. A five mRNAs signature was identified in training set and the accuracy of their signature was 0.82, which was lower than the accuracy in our training set. Although the result of testing set showed that patients could be divided into high and low risk groups by the five mRNAs signature, the accuracy of their signature in testing set was not assessed. Zhang et al. [29] constructed a risk model by 17 mutant genes. The AUC of their signature was 0.907 in 3 years, which was a considerable result. However, the result was not validated in their study and it was unknown whether the mutant-gene signature was efficient for predicting prognosis in 5 years. Similarly, a signature of four lncRNAs was constructed to predict prognosis [30]. In spite of significant accuracy of this signature displayed, a validation result was not provided in their study. Taken together, our cancer-related lncRNA signature might be more beneficial than the previous signatures mainly because of higher accuracy and validation of the findings.
Table 3
Comparison of the accuracy of the prognostic signatures from Wang et.al and ours in 1, 3 and 5 years.
| Sample set | Prognostic signature of Wang et. al | Prognostic signature of this study |
| 1 year | 3 years | 5years | 1 year | 3 years | 5years |
| Training set | 0.934 | 0.796 | 0.662 | 0.967 | 0.933 | 0.839 |
| Testing set | 0.756 | 0.695 | 0.714 | 0.901 | 0.788 | 0.884 |
| Combination set | 0.880 | 0.766 | 0.678 | 0.948 | 0.849 | 0.869 |
Among the lncRNAs in our cancer-related signature, RNF144A-AS1, now called GRASLND, was reported to serve as an important regulator in stem cell chondrogenesis [31]. RNF144A-AS1 could also facilitate the migration and invasion of bladder cancer cells [32]. LINC00702 was a newly identified lncRNA and was involved in the progression of several malignancies via tumorigenesis-associated pathways such as Wnt/β-catenin pathway and PTEN/PI3K-AKT pathway [33–36]. LUCAT1 was the only lncRNA widely investigated in human cancers among the ten lncRNAs. LUCAT1 could promote tumorigenesis and development in various cancers. Besides, it could promote anti-tumor drug resistance in some tumor types such as NSCLC and osteosarcoma [37, 38]. These results suggest that these lncRNAs are reasonable and of importance in human cancers. However, to our knowledge, they were firstly uncovered to be novel prognostic biomarkers in pRCC in this study. In consistent with previous researches [17, 18] in which RP11-63A11.1 was reported, we found that RP11-63A11.1 correlated with clinicopathological features and prognosis of pRCC patients. However, its role of RP11-63A11.1 in pRCC cells was incompletely understanding. We revealed that RP11-63A11.1 was decreased in pRCC cells. Furthermore, increased RP11-63A11.1 inhibited the proliferation and induced apoptosis of pRCC cells, indicating RP11-63A11.1 served as a tumor suppressor in pRCC. The roles of other lncRNAs including RP11-573D15.8, LINC01317, RP11-400K9.4, and RP11-156L14.1 were not reported in previous literatures. These lncRNAs were also firstly in the present study. When combined with the above-mentioned lncRNAs, these lncRNAs generate a powerful tool for survival prediction in pRCC. Their roles in human cancers worth more further investigations. Of note, RP11-63A11.1 may facilitate the further understanding of the development of pRCC.