Brevetoxin Metabolites: Emerging Toxins in French Shellfish Determined by LC-MS/MS and ELISA

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Brevetoxins metabolites: emerging toxins in French shellfish determined by LC-MS/MS and ELISA test

This manuscript represents a valuable addition to the literature that focuses on the analysis of brevetoxins in shellfish, which are of growing international interest and analytically challenging. Observations made from a previous study published in 2018 have been built upon and unsurprisingly additional brevetoxin metabolites have been found in shellfish, which answers a question this reviewer has had since the original publication.  

General comment – The brevetoxin LC-MS method does not attempt to measure all brevetoxins present in the samples, so the discrepancy between the LCMS and ELISA assays is not unexpected.

General comment - To determine whether the French safety threshold of 180 μg BTX-3 eq./kg of shellfish flesh was exceeded when using the LCMS method, the levels of the various BTX analogues observed in samples have simply been summed. A more thorough approach would be to apply relative response factors (for compounds where reference material is not available) and toxicity equivalence factors to allow conversion of results into BTX3 equivalents.

General comment – Comprehensive experimental MS-dependent parameter data has only been provided for BTX1. Do the trends observed for BTX1 apply to all BTXs monitored? This is important as only TypeB BTXs are observed in French shellfish, not Type A (ie BTX1).

General comment –The LCMS method developed for BTX analysis in shellfish has not been assessed for accuracy or precision in matrix, both key method validation criteria. LOD and LOQ assessment have been made but assessment of additional method performance characteristics is needed. Method validation is necessary for accurate quantitation of BTX in shellfish samples, especially if the authors are wanting to compare methods. It is known that recovery of some BTX analogues can be extremely poor. This will result in a large underestimation of levels in samples, and will likely explain some of the difference observed between the LCMS and ELISA results.  

Line 46-47: K.brevis was not confirmed as being responsible for the production of brevetoxins in New Zealand in 1992/93. In fact, TypeA brevetoxins were not observed which likely eliminates K.brevis as the microalgal-producer of the brevetoxins.   

Line 272: Reword sentence to make it clear that BTX1 was monitored but not observed in samples.  

Line 259-266: The approach to calculate brevetoxin concentrations in the total flesh is interesting and there is a comment that ‘remaining flesh’ was analysed and v little BTX was detected. How do you know this observation was not due to matrix suppression? This question can be explored by conducting fortification experiments on blank matrix and observing acceptable recoveries.

Line 280-286: The term ‘Total brevetoxin’ is misleading as it implies all brevetoxins in the sample are being monitored and totalled, when in fact it is just the sum of BTX-2, BTX-3, BTX-B5, and S-deoxy-BTX-B2.

Figure 11 – In addition to Fig11 it is suggested to include a plot comparing the ELISA vs LCMS results to allow an easy visual assessment of how the methods compare. I would imagine the results will correlate well but that the slope will indicate +ve bias for the ELISA results.  

Line 310-315: This text is correct but the underestimation of BTXs when using the LCMS assay could also be due to poor recovery is likely and should be included.

Line 328: Insert word “...for which some standards are commercially available.”

Line 332: MS parameters we optimised for BTX1, a TypeA BTX. It would be useful to know if the trends observed held for the other TypeB BTXs monitored.  

Line 340: Paragraph describes the application of HRMS for ID of BTXs for which no reference material exists. Was HRMS used in this study? If not, how were the BTXs identified without ref material?

Line 357: See previous comment about BTX recovery when using LCMS analysis. Antibody cross reactivity likely only part of the story.

Line 361: BTX1 and TypeA brevetoxins are absent in the French shellfish monitored as part of this project but they are observed in shellfish associated with K.brevis blooms in the US. It therefore important that TypeA brevetoxins are included in monitoring programmes.   

Line 374-377: Sentence needs to be reworded. Only BTX2 and BTX3 were detected because they were the only BTXs monitored!

Line 412: Please double check Australian reg guidelines for brevetoxins as this is not the reviewers understanding. MBA is not used.

Line 424-432: It is acknowledged that different BTXs have different polarities and that these characteristics will impact their extractability. Another way of saying this is that this will impact their recovery, which is an important analytical parameter that impacts method accuracy. Were BTX recoveries assessed from mussel DG as part of this study? If not, why not?

Line 506: How do you know that the ELISA result is overestimated, as it could be correct? It is more likely that the LCMS result is underestimated due to BTX recovery being <100% and only some BTXs being monitored.

Line 508-516: To support risk management the LCMS method needs to be properly validated to an appropriate standard. In its current form it is not.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This study is devoted to the standardization of brevitoxin determination in shellfish tissue. This article is important for the detection of marine toxins, including for aquaculture. I believe the data obtained by the authors are suitable for publication in the Marne Drugs Journal. I have only a few minor comments.

Minor

Please explain the phrase "Brevetoxins of algal origin can undergo metabolic transformation in shellfish." (Line 58). In Lines 31-37, the authors described that dinoflagellate produces "brevetoxins" and "A wide variety of brevetoxin analogues have been identified to date." So the diversity of brevetoxins depends both on "metabolic transformation in shellfish" and on direct synthesis in the dinoflagellate Karenia brevis? Please clarify this in the MS to avoid misunderstanding about the diversity of brevetoxins. And include information on the exact (or approximate) number of brevetoxin analogues produced by the dinoflagellate Karenia brevis (line 33-32).

Line 40 “Karenia brevis”. In line 31, full name of this dinoflagellate was mentioned. Below Authors have to use only short name “K. brevis” for each MS section. Please, do this all over the MS.

Line 51 “or the enzyme-linked immunosorbent assay (ELISA),”. This phrase is strange. Above, Authors wrought about mouse units (Line 49: “of 20 mouse units per 100 g of tissue”). But ELISA use Ig and extracts from shellfish tissues. Please, rephrase this sentence.

Line 65-66. The authors wrought "the primary brevetoxin metabolites formed in shellfish flesh, derived from the algal toxin BTX-2." Figure 2 also contains "BTX-2." So, if BTX-2 is metabolized, then logically it shouldn't be found in shellfish? Please clarify.

Line 77-78 “BTX-3 was eliminated within two weeks of exposure”. Could Authors explain this phrase? Is BTX-3 produced by algae or by biotransformation in shellfish? In Figure 2 BTX-3 is metabolite which is identified in shellfish.

Line 342-343 – Please, set references.

Line 344-345 - Please, set references.

Line 359. “133%,” “127%,” “102%”—these values look strange. Perhaps the authors should choose a more appropriate unit of measurement.

Line 363 “(Pierce and Henry, 2008)” – use numeric references.

Lines 369-372. Where did the authors get this data? Please clarify.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript deals with an interest subject: the presence of brevetoxins in European waters.

In general it is clear, well-structured and written. An important part deals with the methodological optimization that was correctly performed, but that has the problem that it is restricted to a precise equipment (it is anyway interesting as an orientation for other devices)c . However, being this part important, the fragmentation spectra, at least of the toxins for which reference materials exist, but preferably also for any other, should be reported.

Some other changes should also be done:

1. Results contain many descriptions that should be in the Material and Methods section. The journal structure (with material and methods at the end) favors the inclusion of pieces of that section to make the reading easier. However, I think that at least the parts that are not subjected to modifications in the development of the work should be moved to Material and Methods.
2. 2. S-desoxy is used in this case but deoxy is used in other places. Homogenize the notation.
3. Fig 3 y-scale is given in relative units but it is not clear relative to what.  It will be better to use units which allow for direct comparison of the different spectra.
4. 256 Cite here the places of “other locations”
5. 270. I suppose the other characteristics (as water losses) have been used because only molecular masses with a unit resolution MS seems insufficient.
6. Ln 272. Why only four were quantified? Were they simply not measured, were not detected, reference materials were not available?. Could they not be quantified relative to other compound (even being imprecise)?.
7. 10. Legend is unnecessary.
8. 11. Replacing this figure for a regression would make the explanation much easier. Additionally, analyzing the effect of LC-MS/MS quantified toxin on the concentration measured by ELISA could contribute substantially to find the basis of the discrepancies (including crossreactivity) and to quantity how much of the response is not due to the MS analyzed toxins.
9. Ln 361-362. “BTX-1-like toxins are absent”. Are absent of were absent from the samples? In the former case, include the corresponding citations.
10. Pierce and Henry 2008 is not correctly cited (formally) and it is not in the references list.
11. Ln 374 -375 Make clear if they were not detected or they were not looked for.
12. Ln 389-390. A full stop should not be used if the next line makes reference to “these sample” (additionally, samples should be used).
13. Ln 396-397. It is unclear how the fact that tested negative rules out the possibility of matrix effects.
14. Ln 450 and 485. It seems that the whole body was extracted, but previously it was said that only digestive gland was extracted.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

General comment – The brevetoxin LC-MS method does not attempt to measure all brevetoxins present in the samples, so the discrepancy between the LCMS and ELISA assays is not unexpected.

You are right. This is what I already mentioned in the first version of the manuscript (lines 388-389):

« Given the differences in detection principles and metabolite polarity, high concordance between LC-MS/MS and ELISA results was not expected ».

Ok but it is an unusual way to say it. Poor BTX2 recovery is a common observation and is due to it chemically reacting during sample preparation, which is not related to the detection principle or metabolite polarity.

General comment - To determine whether the French safety threshold of 180 μg BTX-3 eq./kg of shellfish flesh was exceeded when using the LCMS method, the levels of the various BTX analogues observed in samples have simply been summed. A more thorough approach would be to apply relative response factors (for compounds where reference material is not available) and toxicity equivalence factors to allow conversion of results into BTX3 equivalents.

I completely agree with you regarding the consideration of the relative toxicity of the different analogs to determine the contamination level of a shellfish sample. Unfortunately, for brevetoxins, there are no relative toxicity factors for each analog that would allow for the conversion of results into BTX3 equivalents. Consequently, we had no choice but to add the concentrations of the analogs found.

Understood, but you do not report in BTX3 equivalents which are the units used in the French safety threshold. Please make this clear.

General comment – Comprehensive experimental MS-dependent parameter data has only been provided for BTX1. Do the trends observed for BTX1 apply to all BTXs monitored? This is important as only TypeB BTXs are observed in French shellfish, not Type A (ie BTX1).

In the publication by Amzil et al. (2021), we previously validated the LC-MS/MS method for analysing lipophilic multitoxins (including BTX2 and BTX3: type B). The experimental conditions of the LC-MS/MS method presented here are similar to those used in Amzil et al. (2021). BTX1 was shown as an illustrative example, but all MS-dependent parameters were optimized for the other available standards (BTX2, BTX3, BTX-B5, and S-desoxy BTX-B2) (See Supplementary Materials).

Great to hear. You should say in the manuscript that BTX1 was included as an illustrative example and that the observation made for this compound were consistent for other BTX analogues, with info provided in the supplementary material.

General comment –The LCMS method developed for BTX analysis in shellfish has not been assessed for accuracy or precision in matrix, both key method validation criteria. LOD and LOQ assessment have been made but assessment of additional method performance characteristics is needed. Method validation is necessary for accurate quantitation of BTX in shellfish samples, especially if the authors are wanting to compare methods. It is known that recovery of some BTX analogues can be extremely poor. This will result in a large underestimation of levels in samples, and will likely explain some of the difference observed between the LCMS and ELISA results.  

You are right. Indeed, we plan to conduct a full validation of the method, particularly its accuracy and precision, as soon as the currently unavailable standards become available.

I would like to reiterate that the method presented in this manuscript is similar to the one initially developed for the analysis of lipophilic multitoxins (including BTX2 and BTX3) by LC-MS/MS. Doping experiments were already performed during the single laboratory validation of the LC-MS/MS multitoxins method according to the NF V03-110 guideline through construction of an accuracy profile.

We must point out that only the BTX3 could be validated because the BTX2 transformed too quickly upon contact with the shellfish matrix. The average recovery rate of BTX3 was between 100.4% and 104.1% depending on the concentration level tested, and the variation in intra-laboratory reproducibility is less than 5%. (Amzil et al., 2021).

Again, please include what you have said above in the manuscript as it needs to be acknowledged.

Line 46-47: K.brevis was not confirmed as being responsible for the production of brevetoxins in New Zealand in 1992/93. In fact, TypeA brevetoxins were not observed which likely eliminates K.brevis as the microalgal-producer of the brevetoxins.   

You are right. Correction made in the manuscript. “The species K. cf brevis in New Zealand, the first observations of this species in the Pacific, with slight morphological differences from K. brevis »(lines 50-51).

I would suggest the following wording……“Although the responsible species was not definitively identified at the time the event is linked to a Karenia‑type gymnodinioid species capable of producing brevetoxins.”

Line 272: Reword sentence to make it clear that BTX1 was monitored but not observed in samples.  

I am sorry, line 272 is not about the dosage of BTX1. Perhaps there's a line number error?

Lines 283-287 list the metabolites targeted by LC-MS/MS, including BTX1. 

I made the following clarification (line 292): « Of all the targeted metabolites by LC-MS/MS, only….. » Line 259-266:

Ok

The approach to calculate brevetoxin concentrations in the total flesh is interesting and there is a comment that ‘remaining flesh’ was analysed and v little BTX was detected. How do you know this observation was not due to matrix suppression? This question can be explored by conducting fortification experiments on blank matrix and observing acceptable recoveries.

This observation is not due to matrix suppression. as mentioned above; fortification experiments on blank matrix were already performed during the single laboratory validation of the LC-MS/MS multitoxins method according to the NF V03-110 guideline through construction of an accuracy profile.

The average recovery rate of BTX3 was between 100.4% and 104.1% depending on the concentration level tested, and the variation in intra-laboratory reproducibility is less than 5%. (Amzil et al., 2021).

The LC-MS/MS method for lipophilic multitoxins (including BTXs) developed previously is very similar to the one developed specifically for BTXs presented in this manuscript.

Ok but was this work performed on the ‘remaining flesh’?

Line 280-286: The term ‘Total brevetoxin’ is misleading as it implies all brevetoxins in the sample are being monitored and totalled, when in fact it is just the sum of BTX-2, BTX-3, BTX-B5, and S-deoxyBTX-B2.

You are correct. The term "total" has been removed (line 301). Furthermore, the measured metabolites are given in parentheses.

Great. Thanks.

Figure 11 – In addition to Fig11 it is suggested to include a plot comparing the ELISA vs LCMS results to allow an easy visual assessment of how the methods compare. I would imagine the results will correlate well but that the slope will indicate +ve bias for the ELISA results.  

Following the reviewer’s suggestion, we have added a new figure comparing ELISA and LC-MS/MS results. The methods show a strong correlation (r² = 0.97), with ELISA displaying a positive bias relative to LC-MS/MS, as reflected by the regression slope of 1.98 and intercept of 4.95. 

Good to see inclusion of the new Figure 12.

Line 310-315: This text is correct but the underestimation of BTXs when using the LCMS assay could also be due to poor recovery is likely and should be included.

The fortification experiments on blank matrix were already performed during the single laboratory validation of the LC-MS/MS multitoxins method according to the NF V03-110 guideline through construction of an accuracy profile. The average recovery rate of BTX3 was between 100.4% and 104.1% depending on the concentration level tested, and the variation in intra-laboratory reproducibility is less than 5%. (Amzil et al., 2021). However, it is possible that some BTX metabolites may not have the same yield as BTX-3. Indeed, we must point out that only the BTX3 could be validated because the BTX2 transformed too quickly upon contact with the shellfish matrix. Therefore, we will proceed with experimental doping plans for metabolites for which standards are available.

Again, say this in the manuscript. It is a stand-alone piece of research and important information like this needs to be communicated not assumed.

Line 328: Insert word “...for which some standards are commercially available.”

Done (lines 360-361)

Great.

Line 332: MS parameters we optimised for BTX1, a Type A BTX. It would be useful to know if the trends observed held for the other Type B BTXs monitored.  

Although BTX-1 (Type A) is described in detail as an illustrative example, MS parameters were optimized for all available BTX standards (BTX-1, BTX-2, BTX-3, BTX-B5, and S-desoxy BTX-B2). In addition, chromatographic conditions had already been evaluated during the development of a previously established LC-MS/MS method for lipophilic multitoxins, including BTX-2 and BTX-3.

Comparable MS-dependent trends were observed across all Type B BTXs monitored, which are the relevant analogues detected in French shellfish.

Great. Please say this in the manuscript.

Line 340: Paragraph describes the application of HRMS for ID of BTXs for which no reference material exists. Was HRMS used in this study? If not, how were the BTXs identified without ref material?

HRMS was not used directly in this study (specified in the manuscript: Line 377). 

For brevetoxin metabolites lacking available reference materials (BTX-B3a, BTX-B3b, BTX-B1, BTXB2, BTX-B4a, BTX-B4b, and BTX-6), identification was based on a molecular mass database. An extensive optimization process was carried out to define mass spectrometry parameters and chromatographic conditions for the targeted brevetoxin metabolites. This included the use of reference fragmentation information reported in the literature, particularly for metabolites without commercially available standards but with known molecular masses (Hort et al., 2021). 

To maximize the likelihood of detecting these compounds in the absence of standards, the selected MS/MS transitions were intentionally non-specific (e.g. neutral loss of ammonium adduct and water loss). While this approach allows the screening of a broad range of brevetoxin metabolites, the formal identification of compounds without reference standards cannot be unambiguously confirmed.

Great to hear about this approach. Please describe this in the manuscript as it is both relevant and interesting.

Line 357: See previous comment about BTX recovery when using LCMS analysis. Antibody cross reactivity likely only part of the story.

You are right. That is what we say in the manuscript (lines 428-435)

Ok.

Line 361: BTX1 and TypeA brevetoxins are absent in the French shellfish monitored as part of this project but they are observed in shellfish associated with K.brevis blooms in the US. It therefore important that TypeA brevetoxins are included in monitoring programmes.   

A priority list of brevetoxin analogues (types A & B) was established for targeted screening, based on available reference materials and toxicity data. This list includes BTX-1, BTX-2, BTX-3, BTX-B1, BTX-B2, S-deoxy-BTX-B2, BTX-B3a, BTX3b, BTX-B4a, BTX4b and BTX-B5.

Line 374-377: Sentence needs to be reworded. Only BTX2 and BTX3 were detected because they were the only BTXs monitored!

Done (line 412)

Ok

Line 412: Please double check Australian reg guidelines for brevetoxins as this is not the reviewers understanding. MBA is not used.

Upon verification, the mouse test (MBA) is indeed used in Australia (see

https://www.primesafe.vic.gov.au/wp-content/uploads/2024/03/Victoria-Shellfish-Quality Assurance-Program-VSQAP-Oct-2023.pdf

In legislation you are correct but in practice I do not believe this would be applied. I have been in touch with my Australian colleague who is involved with marine toxin testing, and they made the following comment “unfortunately the domestic standard is still in mouse units, but this is not available in Australia. If it was, it would be hard to stop them using it, but in practice everyone would rely on the chemical methods.”

Line 424-432: It is acknowledged that different BTXs have different polarities and that these characteristics will impact their extractability. Another way of saying this is that this will impact their recovery, which is an important analytical parameter that impacts method accuracy. Were BTX recoveries assessed from mussel DG as part of this study? If not, why not?

As mentioned earlier in the responses, the BTX recovery rates were previously evaluated using a validated LC-MS/MS multitoxin method, which included BTX-2 and BTX-3. The extraction solvents used in that method are identical to those applied in the BTX-specific method described in the present manuscript. Mean recovery rates for BTX3 ranged from 100.4% and 104.1with intra-laboratory reproducibility below 5%. (Amzil et al., 2021). Although recovery was not re-evaluated for all BTX analogues in the present study due to limited availability of reference materials, additional recovery experiments for other BTX analogues will be undertaken as standards become available

Ok. Some brevetoxin reference material is available via Merck and other will soon be available via NRC.   

Line 506: How do you know that the ELISA result is overestimated, as it could be correct?

It is more likely that the LCMS result is underestimated due to BTX recovery being <100% and only some BTXs being monitored.

The term “overestimation” was used to describe the higher concentrations obtained by ELISA relative to the targeted LC-MS/MS analysis. However, we agree that this difference can also be interpreted as an underestimation by LC-MS/MS, as the latter is limited to a defined list of targeted BTX analogues and may not capture additional metabolites detected by ELISA. In this context, the observed discrepancy reflects differences in analytical coverage and detection principles between the two methods, rather than an absolute bias of one method over the other. This point has been clarified in the revised manuscript.

The following sentence " The ELISA test results show that brevetoxins were consistently quantified, but with systematic overestimation, in all samples where BTX metabolites had been quantified by LCMS/MS."  has been replaced by the following sentence: “The ELISA test results show consistently higher brevetoxin concentrations compared to those obtained by targeted LC-MS/MS analysis, reflecting differences in analytical coverage and detection principles between the two methods. (Lines 516-519).

Great.

Line 508-516: To support risk management the LCMS method needs to be properly validated to an appropriate standard. In its current form it is not.

You are absolutely right. We will continue to optimize the LC-MS/MS method and fully validate it by integrating the missing BTX standards as soon as they become available.

Please make this clear in the manuscript. The method has been developed but requires further validation to have confidence in the results generated and reported.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

Comments and Suggestions for Authors

Review comments have been addressed to an acceptable standard.