The role of piRNAs in predicting and prognosing in cancer: a focus on piRNA-823 (a systematic review and meta-analysis)

This report follows the PRISMA 2020 structure.

We examined the correlation between the role of piRNAs and any of the following types of survival analysis: Overall/cumulative survival (OS), Progression-free survival (PFS), Disease-free survival (DFS), Recurrence-free survival (RFS), Event-free survival (EFS), and Metastasis-free survival (MFS). OS measures the time from diagnosis to death from any cause [28]; PFS measures the time from first treatment to the identification of cancer progression or death from any cause [28]; DFS measures of time after treatment during which no sign of cancer is found [29]; RFS measures the time from cancer cure to the identification of cancer progression/recurrence [30]; EFS measures of time after treatment that a group of people in a clinical trial has not had cancer come back or get worse [31]; and finally MFS measures the time from diagnosis to the identification of a metastatic event [32].

We utilized MeSH terms and added extra search terms to broaden the scope of our search and include a wider range of relevant studies. This approach was intended to make our analysis more comprehensive and ensure that we capture all relevant literature on the topic.

We implemented the following criteria to exclude articles from our analysis: (i) articles that were not accessible in full-text electronically; (ii) articles published in languages other than English; (iii) comments, letters, editorials, protocols, guidelines, case reports, and review articles; (iv) in vitro or preclinical studies; (v) studies lacking sufficient outcome data; (vi) studies that solely investigated genetic alterations of piRNAs, such as polymorphisms or methylation patterns.

We retained studies that reported hazard ratios (HR) and standard error (SE) or confidence interval (CI) or provided life tables for comparing two groups of high- and low-expression levels of various piRNAs [33, 34]. Hence, those studies with more than two groups (e.g. comparing quartiles of expression levels) were excluded from review and meta-analysis [35, 36].

Our inclusion criteria for articles were as follows: (i) studies that examined piRNA expression among cancer patients and control groups; (ii) studies that included data on survival outcomes of patients; (iii) any study that involved quantitative analysis of piRNAs using methods such as quantitative PCR (qPCR), in situ hybridization (ISH), microarray, or sequencing was eligible for inclusion; iii) Studies with sufficient data to generate HR and 95% CI or Kaplan–Meier curves, and (iv) studies published in the English language.

We conducted a systematic search of four databases (PubMed, Scopus, Web of Science, and Wiley Online Library) spanning 8 years from 2015 to 2024. Two independent authors (MT and AK) performed the search to identify potentially eligible articles published in the English language. Our analysis focused on investigating the associations between expression levels of piRNAs and prognosis in human cancer. We used a combination of MeSH terms and subsequent keywords to refine our search strategy., including (“Overall Survival”, “Disease-Free Survival”, “Progression-Free Survival”, “Recurrence-free survival”, “Event-free Survival”, “Survival Analysis”, “Kaplan–Meier Estimate”) AND (“piRNA”, “piRNAs”, “PIWI-interacting ribonucleic acid”, “PIWI-interacting RNA”) AND (“cancer”, “carcinoma”, “tumor”, “neoplas*”, “tumor”, “malignan*”, “metastat*”, “metastas*”, “leukemia”, “leukemia”, “lymphoma”) were used to search the literature. Moreover, search terms that had no available MeSH terms included (“Liquid Biopsy Analysis”, “Prognostic Value”, “Prognostic Factor”, “Prognostic Indicator”, “Prognostic and Predictive Biomarkers”, “Recurrence Risk”, “Predictive for Outcome”, “recurrence*”,). The search was last updated to include articles published through February 1, 2024.

We used the EndNote 21 to remove duplicate records. We screened the titles and abstracts of articles to identify those that were relevant to our study. The full manuscript of each relevant article was then screened against our eligibility criteria. Any uncertainties were resolved. by two of the authors (MT and AK) The data was collected and saved in Microsoft Excel Spreadsheet Software 2021. If the information was unclear or confusing, like when a study mentioned different piRNA quantification methods without specifying which one was used for the survival analysis, we moved it to a separate sheet and marked it as "unclear." To ensure thorough and transparent data collection, we gathered the following details from each study: basic study information such as the title, authors, publication year, number of patients, piRNA quantification methods, follow-up duration, cancer type, and journal of publication. We also collected additional details from eligible articles, including the study cohort's country of origin, sample size, and the total number of piRNAs analyzed.

In this study, the survival outcome was determined by utilizing data from various metrics to characterize the findings, including HR, Relative Risk (RR), Odds Ratio (OR), p-value, or Cox Regression, the type of analysis conducted (Univariable or Multivariable), the SE, and the number of censored participants during the follow-up period. HRs were aggregated using I² statistics to assess the heterogeneity among the relevant studies.

We conducted a meta-analysis on piRNA-823 to evaluate univariate OS. SE was calculated using the formula (upper limit of CI/lower limit of CI) / (2 × 1.96). Estimates were synthesized using a random-effects model and estimated via the restricted maximum-likelihood ratio method. Heterogeneity was evaluated using Q and I² statistics, with the 95% CI of I² also computed [37, 38].

To calculate the effect size and 95% CI, it was imperative for the included studies to provide adequate information. Studies were required to report the HR and its 95% CI or present survival curves that could be digitized using GetData Graph Digitizer software to extract the HR value. In cases where articles only presented survival curves without HR and its 95% CI, digitization was conducted to obtain the necessary data. Forest plots were utilized to display the meta-analysis outcomes, with these analyses performed using Review Manager (RevMan) [39] and for publication bias used Comprehensive Meta-Analysis (CMA) [40]. Data for each type of piRNA were pooled irrespective of cancer type.

The flowchart illustrates the search and selection strategy employed in the study (Fig. 1). We performed a systematic search for studies that investigated the potential role of piRNAs in the prognostic significance of cancer patients and identified 6,104 records using a detailed list of search terms (Fig. 1). After applying the exclusion criteria, a total of 1886 studies were excluded, resulting in 378 articles selected for further evaluation. Furthermore, 331 articles were excluded because of title and abstract screening criteria. Then 47 records remained eligible for full-text assessment. After full-text assessments, 20 studies meeting the search criteria were eligible for final review. Tables 1 and 2 provide an overview of the studies having eligible time-to-event analysis data and their main clinical characteristics.

After full-text assessments, 20 studies meeting the search criteria were eligible for final review. An overview of the studies having eligible overall survival analysis data and their main clinical characteristics are shown in Tables 1 and 2.

Out of the 20 studies analyzed, four focused primarily on RCC [26, 44, 50, 55], four on CRC [13, 24, 27, 53], three on Head and neck squamous carcinoma (HNSCC) [48, 49, 54], two on Gastric cancer (GC) [46, 47], one on Breast cancer (BC) [51], one on Diffuse large B cell lymphoma (DLBCL) [41], one on Ovarian cancer (OC) [42], one on Multiple myeloma (MM) [43], one on Glioblastoma (GBM) [45], one on NSCLC [25], and one on Hodgkin lymphoma (HL) [52].

Out of these 20 studies, nine records utilized high-throughput assays of RNA sequencing [13, 45, 47‐49, 51, 53‐55], seven used qRT-PCR [24‐26, 43, 46, 50, 52], three used Microarray analysis [27, 41, 44], and one used ISH [42] for quantification of piRNAs in patients biological specimen. The majority of these studies were conducted in China (nine studies) [13, 24, 25, 27, 41‐43, 46, 55], three studies followed by the USA [48, 49, 53], three in Canada [47, 51, 54], two in Czech Republic [26, 45], two in Germany [44, 50], and one in Spain [52].

The biological samples collected for analysis included serum (n = 5) [13, 24, 26, 43, 50], tissue (n = 11) [24‐27, 47‐50, 53‐55], Formalin-Fixed Paraffin-Embedded (FFPE) samples (n = 4) [41, 42, 51, 52], blood samples (n = 1) [43], fresh-frozen tissues (n = 2) [44, 45], gastric juice samples (n = 1) [46], bone marrow aspirates (n = 1) [43], and urine samples (n = 1) [26].

The selected articles were published between 2015 and 2023. The median follow-up time ranged from 25 to 250 months. We have summarized this information in Tables 1 and 2.

We conducted an analysis of twenty eligible studies to investigate the role of piRNA expression levels as prognostic markers in different cancer patients. Our findings indicate that six piRNAs, including piRNA-57125 [44], piRNA-651 [25, 52], FR237180 [54], piRNA-017724 [13], piRNA-34536 [50], and piRNA-51810 [50], have a significant prognostic impact on survival prediction. Conversely, the expression levels of twelve different piRNAs, such as piRNA-823 [26, 27, 43], piRNA-1245 [46, 53], piRNA-30924 [44], piRNA-38756 [44], piRNA-30473 [41], (piRNA-009051 and piRNA-021032) [51], piRNA-54265 [24], piRNA-1742 [55], FR-222326 [47], piRNA-58510 [49], and piRNA-3537 [49], were found to be lower and associated with poor survival outcomes. It is important to note that only univariate OS analysis was conducted in this study which is illustrated in Fig. 2. Moreover, we refrained from performing a pooled analysis due to the potential for misleading interpretations when combining various piRNA data.

Figure 3 depicts the forest plot of HR for seven piRNAs studied across four records of patients with renal RCC [26, 44, 50, 55]. Our analysis revealed that higher expression levels of three piRNAs, namely piRNA-57125 [44], piRNA-51810 [50], and piRNA-34536 [50], were significantly associated with better OS in RCC patients. Conversely, the overexpression levels of four other piRNAs, including piRNA-823 [26], piRNA-1742 [55], piRNA-38756 [44], and piRNA-30924 [44], were significantly linked to poorer outcomes in patients with RCC.

We also investigated the prognostic impact of piRNAs on survival prediction in CRC [13, 24, 27, 53] patients based on four records. We found that a higher expression level of only one piRNA, piRNA-017724 [13], was significantly associated with better survival and outcome. On the other hand, the higher expression levels of three different piRNAs: piRNA-823 [27], piRNA-54265 [24], and piRNA-1245 [53], were significantly associated with poorer survival outcomes (Fig. 4).

We performed a meta-analysis of three records in CRC [27], MM [43], and RCC [26] involving 836 patients to explore the association between overall survival and piRNA-823 expression. Pooled HRs showed that overexpression levels of piRNA-823 expression were associated with poorer OS (HR = 3.82; 95% CI, [1.81, 8.04]; P = 0.0004,) (Fig. 5). Our meta-analysis has attempted to combine multiple studies that are known to be heterogeneous in terms of cancer type. Our estimates of heterogeneity metrics have wide 95% CI indicating heterogeneity (I² = 70%).

Sensitivity analyses were conducted to investigate methodological heterogeneity in these studies and assess the impact of individual study data on the overall outcome. No significant influence of any single study on the overall outcome was identified (as shown in Fig. 6). Additionally, a potential publication bias was examined using the Funnel plot, revealing no apparent publication bias in the included studies.

Our meta-analysis offers several advantages. 1) It is the first study to systematically evaluate the correlation between piRNA expression and survival outcomes in cancer patients. 2) Our findings hold significant value, indicating that the sample size included in this meta-analysis was adequate. 3) This study showed high expression levels of piRNA-823 expression in cancer patients cause a greater risk of experiencing poor OS. 4) The included studies varied in the cancer pathological type that were analyzed, which often resulted in substantial heterogeneity between studies in the strength of the predictive effect.

Although our meta-analysis has several strengths, it is important to acknowledge some of its limitations, which include: 1) This meta-analysis was limited to the evaluation of univariate OS due to the available studies that could be included in the analysis. 2) Survival analysis requires suitable data for PFS, DFS, and RFS. However, the lack of such data poses a significant challenge in this regard. 3) Obtaining raw data is crucial for survival prediction and analysis. However, the lack of access to such data poses a significant challenge, and researchers often have to rely on data obtained from the K-M chart. 4) The insufficiency of adequate data has resulted in the absence of meta-analyses on other piRNAs in various types of cancers. 5) Due to the potential for misleading interpretations when combining various piRNA data, a pooled analysis was not performed.

We detected 33 piRNAs in this study, mostly in serum or plasma. This may be because blood samples from cancer patients are more accessible. On the other hand, saliva, urine, and other body fluids are seldom used in research.