Overexpression of Circular PRMT1 Transcripts in Colorectal Adenocarcinoma Predicts Recurrence and Poor Overall Survival

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript investigates the impact of circ-PRMT1 overexpression on colorectal adenocarcinoma and its prediction of recurrence and poor overall survival. It was found that circ-PRMT1 expression is significantly elevated in colorectal adenocarcinoma tissue samples. Moreover, higher circ-PRMT1 levels were significantly associated with shorter disease-free survival (DFS) and overall survival (OS), particularly in patients with TNM stage II or III colorectal adenocarcinoma. The conclusions of the study are of substantial clinical relevance. However, I have a few minor concerns:

• The manuscript does not clearly explain how circ-PRMT1 was selected as the research target.

• Since PRMT1 itself is a cancer-related target, is there a correlation between the expression levels of circ-PRMT1 and PRMT1? This aspect is missing in the current study and should be addressed.

• In line 118, subsection 2.1 is titled “circ-PRMT1 expression is lower in colorectal adenocarcinoma tissue specimens than in their adjacent non-cancerous tissues.” This contradicts the previous statement that circ-PRMT1 is overexpressed in colorectal adenocarcinoma. Is the expression level in colorectal adenocarcinoma tissue actually lower or higher? This needs clarification.

• Lines 147–148: The authors should clearly define the criteria used to distinguish between circ-PRMT1 overexpression and low circ-PRMT1 levels.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Thank you for the opportunity to review the manuscript entitled ‘Overexpression of circular PRMT1 transcripts in colorectal adenocarcinoma predicts recurrence and poor overall survival’ by Kakoropoulus et al.

The manuscript investigates the prognostic value of circular PRMT1 (circ-PRMT1) RNA expression in colorectal adenocarcinoma using quantitative PCR on tumour and matched normal tissue samples, with follow-up analysis of patient survival. While the exploration of circular RNAs as cancer biomarkers is timely and significant, the present study is fundamentally limited by serious statistical, methodological, and interpretive flaws. These issues substantially undermine the reliability of the reported conclusions and, in their current form, preclude recommendation for publication.

 

Below I address in points 1-6 some of the issues:

 

1. A central weakness in the manuscript lies in the handling, analysis, and reporting of the primary expression data. The circ-PRMT1 expression levels, as shown in Table 2, are profoundly skewed, with a median of 2.38 and a mean of 74.46 in cancer tissue, and a median of 0.23 and a mean of 21.70 in normal tissue. This >30-fold difference between mean and median signals the presence of extreme outliers, which are further highlighted by the reported ranges (tumour: 0.061–750.48; normal: 0.015–229.64). Despite this, the manuscript predominantly reports mean ± SEM, a statistical summary that is entirely inappropriate for non-normally distributed, outlier-rich data. This presentation is misleading, overstates the magnitude of differential expression, and does not accurately communicate the underlying data structure.

The only statistical comparison provided is the Wilcoxon signed-rank test for paired tumour-normal samples. While non-parametric, this test alone cannot compensate for poor data visualization and inappropriate summary statistics. No distributional plots—such as box plots, violin plots, or density plots—are provided to allow the reader to assess the degree of skew, the frequency and influence of outliers, or the real magnitude of the reported effect. Figure 1 employs a paired-line plot, which, although showing directionality of change, conceals the critical aspects of data dispersion and outlier-driven differences.

Furthermore, there is no evidence of a sensitivity analysis to determine how robust the reported differences are to the inclusion or exclusion of extreme values. Nor is there a discussion of potential technical or biological reasons for such extreme outliers in expression. This lack of transparency is a serious shortcoming that must be addressed to ensure confidence in the main result.

 

2. The study claims to have developed a sensitive and specific assay for circ-PRMT1 using divergent primers. However, no data are presented to demonstrate that this assay specifically detects circular RNA and not the linear transcript. In the field of circRNA research, it is standard to provide evidence of specificity, such as treatment with RNase R (which degrades linear but not circular RNA), or sequencing of the PCR product to confirm back-splice junctions. Neither is provided here. The claim of specificity, therefore, remains unsubstantiated, and the reported expression values may be confounded by linear PRMT1 contamination.

Furthermore, the PCR workflow involves a pre-amplification step of 15 cycles prior to quantitative PCR. Pre-amplification is known to introduce quantitative bias, especially if efficiencies differ between targets or across samples. There are no details provided regarding validation of this step, nor are primer efficiency or standard curve data included to demonstrate quantitative accuracy. This omission further weakens confidence in the quantification and comparability of circ-PRMT1 expression.

Normalization of qPCR data is conducted using GAPDH as a reference gene. While GAPDH is commonly used, its expression stability between tumour and normal tissues, or across treatment subgroups, is not demonstrated or referenced. This is particularly important in colorectal cancer, where GAPDH can be variably expressed. Without such validation, normalization may introduce systematic bias.

 

3. The manuscript claims that circ-PRMT1 overexpression is an independent prognostic biomarker for disease-free and overall survival in colorectal adenocarcinoma. This assertion is drawn from a single-centre, retrospective cohort of 210 patients, with no external validation or prospective replication. In the absence of an independent dataset, these findings must be regarded as preliminary and hypothesis-generating rather than practice-changing or clinically actionable. So- appropriate contextualization and claims are to be considered in the text.

Additionally, the study does not benchmark circ-PRMT1 expression against established biomarkers (e.g., KRAS, MSI status, or consensus molecular subtypes) or demonstrate whether it adds incremental prognostic value over current clinical models. The manuscript fails to present receiver operating characteristic (ROC) analysis or report area under the curve (AUC) values that could contextualize the diagnostic or prognostic performance of circ-PRMT1. Nor does it translate hazard ratios into risk stratification tools or actionable cutoffs for clinical decision-making.

The cutoff used to dichotomize circ-PRMT1 expression for survival analysis (30th percentile, 0.40 RQU, determined by X-tile software) is arbitrarily chosen and not biologically justified or validated in any independent cohort. No sensitivity analysis is performed to show that this threshold is robust, and no assessment of sensitivity or specificity is provided. This introduces substantial risk for overfitting and false discovery.

 

4. The manuscript presents univariate and multivariate Cox regression models to support the independent prognostic value of circ-PRMT1. While the models include TNM stage, grade, location, and treatment, they omit potential confounders such as comorbidities, adjuvant therapy details, and molecular features that are known to impact colorectal cancer outcomes. The subgroup analyses, including those for TNM stage II (n=84) and III (n=63), are likely underpowered, yet results are interpreted without any correction for multiple hypothesis testing. Furthermore, survival endpoints are defined according to available follow-up data, but 12 patients are excluded due to missing follow-up with no further discussion or sensitivity analysis to assess possible selection bias. The impact of missing data is not explored, which is critical for any retrospective cohort study.

 

5. There is no mention of batch effects or quality control in the RNA extraction, reverse transcription, or qPCR steps. Variability across experimental runs could easily confound results, especially with low-abundance targets such as circRNAs. No mention is made of measures taken to randomize, blind, or otherwise reduce experimental bias. Likewise, no details are given on how the authors ensured the integrity and purity of RNA samples, nor is there evidence of replicate concordance beyond mention of duplicate qPCR wells. The potential for technical artifacts to generate false positives or to exaggerate the apparent upregulation of circ-PRMT1 is not addressed. Given the profound skew in the data and the technical challenges of circRNA quantification, these omissions are serious.

 

6. The discussion section makes broad claims about the clinical utility of circ-PRMT1 as a biomarker and even as a therapeutic target (e.g., via antisense oligonucleotides or siRNAs). These statements are unsubstantiated: no functional data (knockdown, overexpression, or mechanistic experiments) are provided, and no demonstration is given that circ-PRMT1 causally influences tumour biology or patient outcome. Such speculation is inappropriate without supporting experimental evidence and should be omitted or heavily qualified. The manuscript also fails to acknowledge its key limitations, including the retrospective, single-centre design, potential cohort bias, lack of validation, and significant technical and statistical shortcomings. A transparent discussion of these issues is expected in any manuscript submitted to a leading journal.

 

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The authors have adequately addressed my queries