Differentially Expressed Extracellular Vesicle-Contained microRNAs before and after Transurethral Resection of Bladder Tumors

Round 1

Reviewer 1 Report

This manuscript (cimb-1206810) tries to identify extracellular vesicles-contained miRNAs in both, urine and serum, as new diagnostic and recurrence-free survival biomarkers. Surprisingly, the authors only could find two consistent differentially expressed miRNAs (miR-451a and miR-486-5p) in urinary extracellular vesicles in presurgery samples from T1 patients compared to postsurgery samples. No miRNA was identified in serum samples or compared to non-cancer patients. Despite that, these results are important and relevant, and the manuscript is, in the general, well designed and written with some major concerns:

 

1. Yields of miRNA achieved with serum and exosome extraction kits (like exoRNeasy or miRNeasy kits) vary strongly between samples from different patients. Thus, it is recommended to use an exogenous control to ensure the reproducibility of the isolation method (DOI: 10.1016/j.molonc.2016.03.005). Without a spike-in miRNA use (exogenous control), the authors fail to demonstrate that variability between different samples are due to patient variability and not only due to yield variations from the extraction method. If this normalization cannot be done at this work, the authors must discuss it in the manuscript.

 

2. The authors only performed small RNA sequencing without any kind of posterior hit validation. The authors should validate miRNA expression (miR-451a and miR-486-5p) by an independent method, such as RT-PCR. It will be interesting also check the expression of miRNAs previously described as differentially expressed in bladder cancer (miR-375, miR-146a, miR-21, among others).

 

Minor concerns:

 

1. It will be interesting to add to discussion section, the possible target proteins of the two miRNAs, and related their functional role with the data obtained in Gene target predictions analysis.

 

2. The authors should put in Table 1 legend, the meaning of primary diagnosis acronyms.

Author Response

Reviewer 1 

 

We thank the reviewer for constructive feedback and we have addressed all of the comments below.  

 

Major concerns:

 

Point 1. Yields of miRNA achieved with serum and exosome extraction kits (like exoRNeasy or miRNeasy kits) vary strongly between samples from different patients. Thus, it is recommended to use an exogenous control to ensure the reproducibility of the isolation method (DOI: 10.1016/j.molonc.2016.03.005). Without a spike-in miRNA use (exogenous control), the authors fail to demonstrate that variability between different samples are due to patient variability and not only due to yield variations from the extraction method. If this normalization cannot be done at this work, the authors must discuss it in the manuscript 

 

Response 1. Spike-in miRNA is an important control of technical variability, and we appreciate that the reviewer raises this important question. We did discuss using spike-in miRNA but decided against applying them to our study. One of the leading reasons for avoiding spike-ins is that the amount of extracted miRNA was exceedingly low and all starting material was needed for the analysis, making quantification impossible. Addition of spike-in could therefore easily oversaturate the miRNA, leading to sequencing of only spike-ins. Small volumes of spike-in miRNA also lead to pipetting errors when spike-in miRNA is added.    

 

It is also a question of where to add spike-in miRNA when analyzing miRNA from EVs. It would have been ideal to add spike-in miRNA to the biofluid directly, to correct for all isolation steps, but because we are extracting EVs before we are extracting miRNA, spike-in miRNA added directly to the biofluid would have been lost. We could have normalized to the measured EV concentration, although EVs show a high degree of variance in miRNA content, making the relevance of this normalization questionable. Such a procedure would be interesting but requires optimization of the amount of spike-in miRNA to add relative to the exosome concentration. Alternatively, we could have added EVs with miRNA from a different species as a spike-in control, although this would require even more optimization. Unfortunately, further optimization was outside the resources of this project.  

 

Because addition of spike-in miRNA relative to EV counts occurs after an isolation step, it is arguably equally valid to normalize to the total amount of miRNA found by sequencing. The default normalization in the statistical analysis of most RNAseq analyis is to normalize to library size, i.e. the amount of detected miRNA in each sample, and with correction factor (here TMM as per standard in edgeR was used). Considering the limitations of adding spike-in miRNA to our current design, we decided that it was best to rely on this type of normalization. We do agree that this is a limitation of our study and have added this limitation to the discussion.   

 

We correct for variations from the extraction process by repeating the RNA sequencing in the replication run from different aliquotes from the same patients. Considering that we find miR-451a and miR-486-5p to be differentially expressed in both runs this increases the likelihood that these changes are from biological differences and not technical noise from the extraction process.  

 

Point 2. The authors only performed small RNA sequencing without any kind of posterior hit validation. The authors should validate miRNA expression (miR-451a and miR-486-5p) by an independent method, such as RT-PCR. It will be interesting also check the expression of miRNAs previously described as differentially expressed in bladder cancer (miR-375, miR-146a, miR-21, among others).  

 

Response 2. We thank the reviewer for this comment and agree that in general it would have been beneficial to perform an alternative technology such as RT-PCR to validate our sequencing results. Given the limited amount of patient material available, we decided to perform a replication sequencing run rather than RT-PCR. The rationale for this was to both do a confirmation of the initial results, while at the same time allowing for an exploratory path, potentially identifying additional candidate biomarkers, as we also increased the number of T1 patients (from 10 to 14) with a complete miRNA profile. Considering that we found miR-451a and miR-486-5p in both sequencing runs, this increases our confidence in the results for these miRNA.  

 

As can be seen in figure 2, miR-375, miR-146a and miR-21 are not differentially expressed in our study. We can only speculate that this is because of differences in extraction method, patient population or other reasons. Hopefully, future studies with more patients would be able to find more differentially expressed miRNA of which the mentioned miRNA could be present. 

  

Minor concerns:  

 

Point 1. It will be interesting to add to discussion section, the possible target proteins of the two miRNAs, and related their functional role with the data obtained in Gene target predictions analysis. 

 

Response 1. We appreciate this comment and have updated the discussion section with references to literature, which finds some of the same targets as we find in our gene target prediction in figure 5.A. Our discussion of these references also highlights the potential biological implication of changes to these genes.  

 

Point 2. The authors should put in Table 1 legend, the meaning of primary diagnosis acronyms.  

 

Response 2. Thank you, we have added the meaning of the acronyms to the table legend for Table 1.  

 

 

 

Reviewer 2 Report

Brief Summary:

The aim of the study by Stromme & Heck et al., was to identify potential bladder cancer-specific microRNA (miRNA) biomarkers in urine and blood extracellular vesicles (EV) of bladder cancer patients undergone transurethral resection. The authors performed next generation sequencing of EV-contained miRNA isolated from either the urine or serum of 41 patients with non-muscle invasive bladder cancer pre- and post-surgery, as well as from non-cancer patients (NCP) with benign disease. They identified two urinary EV-contained miRNAs, miR-451a and miR-486-5p, significantly upregulated in pre-surgery samples from 10 patients with T1 stage disease compared to post-surgery samples where patients had disease recurrence, but not to NCP samples. An additional sequencing run confirmed these results, but also revealed no significant difference in the expression of serum EV or supernatant miRNAs. Although descriptive, this a concise study identifying miR-451a and miR-486-5p as potential biomarkers for bladder cancer recurrence after surgical resection of T1 stage tumors. However, some issues should be addressed for the manuscript to be considered for publication.

Strengths of the study:

• miRNA profiling of pre- and post-transurethral resection bladder tumors
• Revealing the feasibility of non-invasive sequencing approaches for prognostic/diagnostic purposes of bladder cancer recurrence

Weaknesses of the study:

• Small number of samples may preclude the robust identification of more miRNAs and/or their correlation to clinical patient parameters, such as sex or smoking.
• Discrepancies in the results between different approaches of miRNA identification in urine and serum.

 Comments

Abstract: Clear and concise. Comments:

• The authors should outline the importance of their study, since it is the first one comparing human patient samples pre- and post-surgery for EV-containing miRNA sequencing. [minor]
• The authors mention that they used 14 T1 stage bladder tumor samples. However, they actually only sequenced and analyzed 10. This should be corrected to avoid misleading the readers. [major]

Introduction: Overall, complete background on diagnosis/prognosis, EVs and miRNA in bladder cancer. The authors outline their study design and goal.

 

Materials and Methods: The EV isolation and sequencing methods are adequately described. Comments:

• For the clinical samples, the follow-up time for the post-surgery matched samples for each T1 patient is missing. Cite Table S3. [minor]
• A statement on the ethical approval for the use of patient material (informed consent) should be included or this issue clarified if it not applicable. [major]

Results: This section is well designed and is data are presented in a clear manner, with some discrepancies to be addressed. Comments:

• The authors mention that there were 14 T1-stage bladder cancer patients in this study. However, they do not clarify why they used only 10 for their analyses as well as why there are only 11 such patients present in Table 1. These discrepancies have to be clarified in text. [major]
• It is unclear in how many patients out of the 10 T1-stage analyzed, the 11 urinary EV and 12 serum identified miRNAs were upregulated in. [minor]
• In Figures 2D and F, the authors show that the expression of almost all of the identified miRNAs are not significantly deregulated between T1 pre and NCP samples, but they show a differential expression of all in panels 2E and G. They should explain the rationale behind this and/or correct it, since it is misleading. [major]
• In Figure 3 the authors identify a different set of deregulated miRNAs in the additional sequencing run of the same samples in urine EVs, compared to what they show in Figure 2. They should explain this discrepancy. Is it a reproducibility issue? [major]

Discussion: Overall, the authors discuss the findings, limitations and future implications of their study well. However, some issues need to be addressed:

• How do the authors explain the lack of miRNA identification in serum EVs? The EV are supposed to shield miRNAs in circulation and their analysis should in principle, yield some miRNA identification. How do they explain the presence of miRNAs in the supernatant but not serum EVs? [major]
• The authors should comment on the difference in EV size between urine and serum. Does this difference in size have any impact on the type and/or amount of miRNA these vesicles can carry? [major]
• Different miRNAs were identified in urine and serum. The authors could discuss on why that may happen. [minor]
• miRNA-451a is upregulated in pre-surgery urine EVs compared to post-surgery in this study. The authors should discuss, why that may be happening and what the functional implication may be for tumor growth. Since miRNA-451a is known to be a tumor suppressor, why would its upregulation contribute to a more aggressive bladder phenotype? On the same lines, what do the authors mean by “selective incorporation” of miRNA-451a? Selective in what context and to what effect? [major]

Author Response

We thank the reviewer for constructive comments, which have been adressed below

 

 

Comments 

Abstract: Clear and concise. Comments: 

The authors should outline the importance of their study, since it is the first one comparing human patient samples pre- and post-surgery for EV-containing miRNA sequencing. [minor]  

 

Thank you for the comment and we have added your suggestion to our abstract.  

 

• The authors mention that they used 14 T1 stage bladder tumor samples. However, they actually only sequenced and analyzed 10. This should be corrected to avoid misleading the readers. [major]

 

We apologize for the confusion. 10 T1 patients were sequenced in the initial sequencing run, In the replication run we validated our results by re-sequencing these 10 patients. Additionally, we added 4 more T1 patients in the replication run giving us 14 unique T1 patients. This is best shown in figure 1.C and we updated the abstract to clarify our setup.   

 

Introduction: Overall, complete background on diagnosis/prognosis, EVs and miRNA in bladder cancer. The authors outline their study design and goal. 

 

Materials and Methods: The EV isolation and sequencing methods are adequately described. Comments: 

• For the clinical samples, the follow-up time for the post-surgery matched samples for each T1 patient is missing. Cite Table S3. [minor] 
•  

We thank the reviewer for the input, and we have added the information accordingly in the Clinical Samples information (Material & Methods).  

 

• A statement on the ethical approval for the use of patient material (informed consent) should be included or this issue clarified if it not applicable. [major] 

 

We apologize for the confusion. The ethical approval information can be found at the end of the clinical sample information (material & methods) and is also stated under the ‘’Informed Consent Statement’’ section.   

 

Results: This section is well designed and is data are presented in a clear manner, with some discrepancies to be addressed. Comments: 

• The authors mention that there were 14 T1-stage bladder cancer patients in this study. However, they do not clarify why they used only 10 for their analyses as well as why there are only 11 such patients present in Table 1. These discrepancies have to be clarified in text. [major] 

 

Thank you for the observation and we apologize for the confusion. We have updated Table 1 to outline patient characteristics describing both their primary diagnosis as well as adding an additional column ‘’Diagnosis at time of inclusion (TOI)’’. We have therefore removed the (TOI) from the ‘’Primary Diagnosis’’ column as this is now denoted in the last column in Table 1.  

 

Table S3 also outlines the sample details including the diagnosis at time of inclusion in the study for each of the 41 patients. These have now been corrected and we apologize for the confusion. There are 14 T1 patients and 27 Ta patients in our study. We have further clarified these details in the result section (under Patient characteristics). While updating our tables, we discovered accidental copy errors in Tables 1 and S3 for some of the patient diagnoses which occurred when making our tables originally. We apologize for the error. 

 

• It is unclear in how many patients out of the 10 T1-stage analyzed, the 11 urinary EV and 12 serum identified miRNAs were upregulated in. [minor] 

 

We appreciate your comment. We have now added a table (table 2) in the manuscript file highlighting the number of patients where the maximum counts per thousand value was in the T1 cancer sample for each miRNA identified from the Urine EV and serum supernatant in the analysis presented in Figure 2. 

 

• In Figures 2D and F, the authors show that the expression of almost all of the identified miRNAs are not significantly deregulated between T1 pre and NCP samples, but they show a differential expression of all in panels 2E and G. They should explain the rationale behind this and/or correct it, since it is misleading. [major]  

 

We apologize for the confusion. We wished to show all results (completeness) by including non-significant values. We have corrected the scales in Figure 2D, 2F and 3C, to highlight non-significant values.   

 

• In Figure 3 the authors identify a different set of deregulated miRNAs in the additional sequencing run of the same samples in urine EVs, compared to what they show in Figure 2. They should explain this discrepancy. Is it a reproducibility issue? [major] 

 

We apologize for the confusion. To increase the statistical resolution, the analysis in figure 2 is based on the complete data set (10 patients sequenced twice and 4 patients sequenced in the replication run) with patient, biosource and batch as variables in the statistical model. When comparing run 1 and run 2 in figure 3 we subset our data to focus solely on the 10 patients sequenced in both runs. The differences in the data chosen for each analysis would necessarily affect which miRNA are detected, particularly near the threshold of significance. However, for the miRNA far away from the threshold (miR-486-5p and miR-451a) we do detect them no matter which subset of the data is chosen for the analysis. We have changed the text when introducing the different analysis in the results to make this more clear.  

 

Discussion: Overall, the authors discuss the findings, limitations and future implications of their study well. However, some issues need to be addressed: 

• How do the authors explain the lack of miRNA identification in serum EVs? The EV are supposed to shield miRNAs in circulation and their analysis should in principle, yield some miRNA identification. How do they explain the presence of miRNAs in the supernatant but not serum EVs? [major] 

 

We apologize for the confusion. Figure 2B shows no differentially expressed miRNA from serum EVs. However, we did find many hundreds of miRNA that was not differentially expressed from serum EVs. We cannot explain why we did not find any differentially expressed miRNA in serum EVs but our results show that, with our approach, miRNA from serum EVs does not have enough signal compared to the noise and further research with a potentially larger sample size is needed to detect differentially expressed miRNA from serum EVs. We can only speculate as to why serum miRNA has a lower signal to noise ratio than urine EVs. Potentially, urinary EVs might have a better signal to noise because they are sourced close to the bladder cancer itself, while serum miRNA originates from multiple organ systems. We have updated the discussion with this point.  

 

• The authors should comment on the difference in EV size between urine and serum. Does this difference in size have any impact on the type and/or amount of miRNA these vesicles can carry? [major] 

 

Thank you for highlighting this point. The method we employ for the isolation of EVs corresponds with two studies (one in urine and one in serum) now referenced within the ‘’Results – EV characterisation’’. The average size of EVs obtained in those two articles are different between the two biosources but both fall within the size range expected for small-EVs. Our EVs show the presence of small-EVs within the expected range for both urine and serum.  

 

As the size of small-EVs are dependent on the size of the multivesicular bodies from which they are derived, they span over a size range distinguishing them from other EVs. The amount of RNA present in serum and urine small-EVs has been compared by others (doi: 10.1098/rstb.2013.0502) and the RNA has been found to vary between urine and serum small-EVs. One can speculate that this variation in RNA content could be due to small-EVs size or due to different cargo incorporation representative of their parental cell. This is an area where further investigation is required.  

 

Different miRNAs were identified in urine and serum. The authors could discuss on why that may happen. [minor]  

 

Thank you for this comment. The presences of different miRNAs in our three biosources is probably due to the secretion of EVs from different cell types around the body. The cargo of EVs reflect that of their parental cell of origin and would explain the different miRNA packaging in EVs from urine compared to EVs from serum. We have expanded on this in our discussion. Other studies have found differences in exosomes from urine and serum (Li et al, doi: 10.1098/rstb.2013.0502). The different repertoire of serum miRNA could be explained by the large amounts of circulating stable miRNA. These accumulate in the blood (serum/plasma) from all parts of the body without being directly dependent on the EV biogenesis process as in the case of miRNA in EVs.  

 

• miRNA-451a is upregulated in pre-surgery urine EVs compared to post-surgery in this study. The authors should discuss, why that may be happening and what the functional implication may be for tumor growth. Since miRNA-451a is known to be a tumor suppressor, why would its upregulation contribute to a more aggressive bladder phenotype? On the same lines, what do the authors mean by “selective incorporation” of miRNA-451a? Selective in what context and to what effect? [major] 

 

Thank you for this comment. We have added text to the discussion referencing studies that identify potential targets for miRNA-451a and which also investigate how this miRNA could influence cancer growth.  We agree that it is may seem slightly paradoxical  that a tumor suppressive miRNA, miR-451a,  is upregulated presurgery when there is viable tumor tissue present compared to postsurgery when the patients is clinically recurrence-free. In the discussion we reference literature which find that miR-451a is enriched in EVs derived from oral cancer cells and lung cancer cells, as compared to their cells of origin, potentially as a mechanism for the cancer cells to get rid of this tumor suppressive miRNA. This highlights the fact that it is theoretically possible to find tumor suppressors upregulated in urine EVs from patients with an aggressive bladder cancer.  We have also referenced a study where miR-451a is upregulated in the urine of muscle-invasive bladder cancer compared to healthy controls, and it may well be that a significant fraction of this miR-451a i in fact is EV-contained, although the study did not investigate this. We have clarified the text about selective incorporation of miR-451a in EVs to make our point clearer.  

 

Round 2

Reviewer 1 Report

This manuscript (cimb-1206810) tries to identify extracellular vesicles-contained miRNAs in both, urine and serum, as new diagnostic and recurrence-free survival biomarkers. Surprisingly, the authors only could find two consistent differentially expressed miRNAs (miR-451a and miR-486-5p) in urinary extracellular vesicles in presurgery samples from T1 patients compared to postsurgery samples. No miRNA was identified in serum samples or compared to non-cancer patients. Despite that, these results are important and relevant, and the manuscript is, in the general, well designed and written.

Reviewer 2 Report

The authors have addressed the majority of my comments and the manuscript has significantly improved. Specific discrepancies have been clarified adequately either in text or response to reviewers.