NX210c Demonstrates Therapeutic Potential to Restore Blood–Brain Barrier in a QSP Model of Relapsing–Remitting Multiple Sclerosis

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This manuscript presents a quantitative systems pharmacology (QSP) model evaluating the therapeutic potential of NX210c in relapsing-remitting multiple sclerosis (RRMS), focusing on blood-brain barrier (BBB) restoration and immune modulation through in silico simulations. Although the approach uses computational modeling with no experimental verification, several critical shortcomings must be addressed.

1. The study does not align with the special issue's scope, which emphasizes cellular and molecular mechanisms of BBB dysfunction specifically in dementia contexts. RRMS, an autoimmune demyelinating disorder without primary dementia pathology, falls outside this dementia-focused scope, despite incidental BBB involvement.
2. As a purely theoretical in silico model without experimental corroboration – relying instead on redesigned prior QSP frameworks and existing datasets – it oversimplifies the complex, multifaceted pathophysiology of the BBB and MS. This renders simulated outcomes speculative and disconnected from empirical validation, which again contravenes the special issue's emphasis on preclinical, in vitro, and animal or organoid experimental models.
3. The validation process is insufficiently rigorous, as it draws from unspecified preclinical and clinical datasets without demonstrating robust sensitivity analyses or accounting for inter-patient variability in MS progression, potentially leading to unreliable predictions of BBB markers.
4. The manuscript overstates translational implications, such as predicting zero relapses over two years in combination therapy scenarios, based on unverified assumptions that ignore known MS complexities and risk unsubstantiated clinical optimism.
5. Novelty is reduced, as the work merely extends a previous QSP model without introducing new mechanistic insights or data, reiterating established knowledge on BBB-targeted therapies in neurological disorders.
6. Conclusions extend findings to "ageing-related BBB dysfunction" in RRMS without supporting experimental or clinical evidence, potentially conflating distinct pathologies and overlooking the exploration of senescence in dementia.
7. Inherent limitations of QSP modeling are only partially addressed, undermining the model's utility for go/no-go decisions or dosing optimization.

Comments on the Quality of English Language

The English could be improved to more clearly express the research.

Author Response

Comment 1: The study does not align with the special issue's scope, which emphasizes cellular and molecular mechanisms of BBB dysfunction specifically in dementia contexts. RRMS, an autoimmune demyelinating disorder without primary dementia pathology, falls outside this dementia-focused scope, despite incidental BBB involvement.

Response 1: We thank the reviewer for pointing this out. This appears to be due to a mistake regarding the selection of the special issue. We confirm that the manuscript was not intended for the dementia-focused special issue and apologize for any confusion caused. We have raised this point with the Editor, as a result the manuscript has now been transferred to the issue “Molecular Mechanisms and Regulation in Blood-Brain Barrier”, which scope is aligned with the objectives of our study. We also would like to mention that we have reviewed the entire manuscript to improve clarity throughout the text. High-quality figures have been transmitted to the Editor for your review.

Comment 2: As a purely theoretical in silico model without experimental corroboration – relying instead on redesigned prior QSP frameworks and existing datasets – it oversimplifies the complex, multifaceted pathophysiology of the BBB and MS. This renders simulated outcomes speculative and disconnected from empirical validation, which again contravenes the special issue's emphasis on preclinical, in vitro, and animal or organoid experimental models.

Answer 2: We appreciate the reviewer’s concern and would like to clarify that the current work is not purely theoretical. The QSP framework has been validated against clinical data from prior studies (e.g. Delétage 2021, Lemarchant 2022), and the NX210c’s drug action incorporated into the model is supported by in vitro experimental evidence (Greene 2024). Our approach demonstrates how multiple data sources—internal and external, including in vivo, in vitro, and clinical datasets—can be integrated within a mechanistic modeling platform to generate hypothesis-driven insights and expedite inference for future studies. While the model does not replace experimental work, it complements it by providing a systems-level perspective that may inform study design and prioritization of disease indications to study in clinical programs. We have better emphasized this point in the revised version to clarify the role and scope of the modeling approach (lines 281-287, 384-396, 515-520).

Comment 3: The validation process is insufficiently rigorous, as it draws from unspecified preclinical and clinical datasets without demonstrating robust sensitivity analyses or accounting for inter-patient variability in MS progression, potentially leading to unreliable predictions of BBB markers.

Answer 3: The model relies on preclinical and clinical information explicitly referenced (Bourdès 2021, Greene 2024, Dublin 2022), which serves a qualitative and semi-quantitative model qualification purpose, consistent with the intended context of use of a mechanistic QSP framework. The BBB module was calibrated and verified against published in vitro and in vivo evidence on tight-junction protein modulation and permeability changes, while the RRMS disease layer inherits its qualification from the previously validated MS TreatSim framework. In the revised manuscript, we have explicitly detailed the role of each dataset and clarified which components are newly introduced versus inherited (lines 384-396).

Inter-patient variability is intrinsically represented in the model through the use of virtual populations. Heterogeneity in immune repertoire, autoimmune activation propensity, thymic output, and BBB baseline integrity is explicitly encoded via parameter distributions and stochastic processes. The results shown already represent cohort-level behavior emerging from this variability rather than single deterministic trajectories. We have clarified this aspect in the Methods section, emphasizing that variability is not an a posteriori adjustment but a structural property of the model (lines 403-405, 477-479).

Regarding sensitivity analysis, the aim of this study was not parameter identification or predictive uncertainty quantification at the individual-patient level, but a mechanistic plausibility assessment and comparative scenario analysis. The robustness of the conclusions is demonstrated by the consistency of treatment effects across heterogeneous virtual patients and across multiple endpoints (BBB integrity markers and relapse-related outputs). Importantly, the manuscript does not claim patient-specific prediction of BBB biomarkers, but rather mechanistically grounded, population-level insights on  the translation of  early biomarkers of effect into downstream clinical response. We have strengthened this point in the revised Discussion (lines 308-312, 336-346).

Comment 4: The manuscript overstates translational implications, such as predicting zero relapses over two years in combination therapy scenarios, based on unverified assumptions that ignore known MS complexities and risk unsubstantiated clinical optimism.

Answer 4: We have expanded the manuscript to more explicitly acknowledge the inherent scope and limitations of the present QSP framework and to clarify the intended interpretation of the simulation results. Specifically, we have now emphasized that the model is not designed to provide precise quantitative predictions regarding the magnitude of NX210c’s effects, nor does it fully capture the full spectrum of biological variability or uncertainty across the RRMS population. Instead, we highlight that the current model serves primarily as a hypothesis‑generating tool intended to support early scientific exploration of plausible mechanistic pathways and treatment effects, guiding early-development go/no‑go decisions, rather than as an exact quantitative approach for dosing optimization. We also note that future extensions of the model, supported by additional preclinical and clinical data, will allow for more quantitative parameter estimation, uncertainty quantification, and refined predictive performance. The newly added text can be found in the Discussion and Conclusion sections (lines 286-287, 328-346, 515-520).

Comment 5: Novelty is reduced, as the work merely extends a previous QSP model without introducing new mechanistic insights or data, reiterating established knowledge on BBB-targeted therapies in neurological disorders.

Answer 5: We respectfully disagree and believe that this concern arises from an under-emphasis of the novel elements introduced in the current work. While the RRMS backbone builds upon the previously published MS TreatSim model, the present study introduces a new mechanistic BBB integrity module explicitly modeling tight-junction protein dynamics and permeability proxies, which was not present in earlier work. Crucially, BBB integrity is here mechanistically coupled to immune-cell trafficking into the CNS, thereby linking molecular-level BBB markers to downstream clinical proxies such as relapse dynamics. Moreover, NX210c is implemented with a mechanism of action targeting BBB stabilization, enabling prospective in silico assessment of BBB biomarkers together with disease-level outcomes under monotherapy and combination scenarios. This integrative BBB–RRMS coupling and the focus on BBB-targeted therapeutic effects of a new drug already in development represent the main conceptual and methodological novelty of the work. Explicative text has been added in Introduction and Methods sections (lines 103-110, 389-396, 451-461).

Comment 6: Conclusions extend findings to "ageing-related BBB dysfunction" in RRMS without supporting experimental or clinical evidence, potentially conflating distinct pathologies and overlooking the exploration of senescence in dementia.

Answer 6: As pointed out in response 1, we don’t specifically target dementia with this work but focus on general decline of the BBB with time and to explore the application in RRMS of a new drug with potential to restore the BBB. The Conclusions has been slightly rephrased (row 533-35).

Comment 7: Inherent limitations of QSP modeling are only partially addressed, undermining the model's utility for go/no-go decisions or dosing optimization.

Answer 7: Please refer to answer 4.

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript describes efforts to establish NX210c, a novel therapeutic peptide, as a promising molecule in restoring BBB integrity, at both preclinical and clinical levels. Consequently, it exemplifies translational potential merit toward its administration in MS populations and across various central nervous system (CNS) pathologies with overlapping mechanisms. The specific study evaluated the therapeutic potential of NX210c in patients with relapsing-remitting MS (RRMS), using a known quantitative systems pharmacology (QSP) model, currently redesigned to capture the dynamic interplay between BBB integrity and immune system activity. Adoption of the employed model with the parameterization, entering the long equation, led to the input of a number of parameters that had been drawn out of literature studies, essentially formulating the patients with the specific pathology for which the study was designed. The BBB was essentially put to test with specified conditions affecting specific genes and gene products responsible for the permeability, with the latter known to work for healthy individuals compared to disease-carrying patients.

The design was described adequately by the authors and, when transferred into the QSP model for validation, it was formulated so as to reflect the dosology employed to induce effects on the reduction of BBB leaking and its electrical activity. The run QSP model simulations project the potential of NX210c upon ageing-related BBB dysfunction RRMS. The application suggests a cost-effective tool for a well-calculated choice, supporting the refinement of clinical strategies and optimizing dosage. The approach offers a well-configured assessment that merits further consideration for handing disease administration for patients suffering of pathologies heavily involving the BBB.

The work, albeit not in vivo, offers a significant body of information, with a wealth of information that proves credible in light of the currently existing literature. The results are screened and the interpretation offers significant insight of practical value that might be of use at the clinical level.

In the process of the specific work, there are several points of concern that need to be addressed by the authors prior to any consideration. These points are remarked upon and require specific responses and clarification in the form of revisions.

1. Figure 2 consists of a collection of subfigures for the testing conducted. Unfortunately, the clarity of the figures is very low. The resolution is low and the various subfigures can barely be recognized for the information they carry. The figures in their entirety cannot be read. That should be rectified.
In addition, the subset of Figures (C) reflects the permeability of the brain endothelial monolayer following NX210c treatment for two experiments as in [19]. There is no additional information on the two experiments. Beyond what is stated as being in line with those two experiments, the authors should provide a concise description of the two experiments and what the two experiments represent in the milieu of experimental activity in this phase of the work.
2. Analogous comments hold for Figures 3 and 4. The Figures are illegible. The authors should make sure that the figures are clear so that the preview process takes place to the benefit of the work done.
3. In lines 195-197, the statement “Moreover, a dose-dependent protective effect of NX210c – i.e. mitigating the decline in BBB function over time – is observed in both healthy and RRMS populations, as compared to the respective placebo group.” should be re-written to read “Moreover, a dose-dependent protective effect of NX210c, i.e. mitigating the decline in BBB
function over time, is observed in both healthy and RRMS populations, compared to the respective placebo group.”.
4. In lines 386-388, the statement “The compound then transform into its cyclic version, NX210c, for which frequent PK samples were obtained up to 180 minutes post dosing [20].” should be revised to read “The compound then transformed into its cyclic form, NX210c, for which frequent PK samples were obtained up to 180 min post dosing [20].”.
5. In Figure 7, it is stated that administration of NX210c every 6 months suggests reduction (according to the graphs provided) of the relapses in patients ((A) Relapse activity without treatment. (B) Relapse 242 activity following repeated NX210c treatment cycles). Several questions arise as a result of the graphs shown in the Figure:
a)It is stated that “… repeated NX210c dosing cycles (every 6 months) led to a substantial and sustained reduction in both the frequency and severity of relapses over the simulated decade: from 7 to 4 clinical relapses (+2 subclinical).”. If two relapses are subclinical, then the term should be clarified in terms of the pattern shown in the figure.
b)If subclinical relapses are reduced post-administration of NX210c, then are there any subclinical relapses in the case of the untreated patients? How does that entail significance in the observed reduction?
c)There is a reduction of relapses when NX210c is administered at the dose and time intervals opted for. To that end, one can observe that the pattern of remaining relapses in the treated patients is the same (and part of) as the original one for the untreated patients. What is the meaning of that? What is the significance of that in the subsequent route of RRMS patients? Does the pattern itself signify merit in the course of pathology and thus the patient?
d)The pattern of remaining relapses is almost an identical part of the original pattern in the untreated patients. The intensity of the relapse incidents, however, remains the same. What does that imply in terms of the future in the patient’s profile? Along the same lines of thought, what could possibly be the origin of the pattern, thereby allowing for further inquiry into its potential merit in the course of the pathology or the possibility of ameliorating the condition of the patient?
6. Since IJMS is a journal on molecular sciences and the NX210c is a molecule, one would expect to see a) the structure of the linear and cyclical form of the peptide, and b) the chemistry of its interactions with targets that lead to the amelioration of the profile of the patients studied. A meager and very general elaboration was provided in the manuscript that pertains to the interaction mode(s) of the peptide, as a result of which all of the observations shown in the diagrams emerged. That clarification should be provided in a descriptive manner.

Based on the comments and the remarks made above, the manuscript should undergo revisions and submitted for re-evaluation.

Author Response

Comment 1: Figure 2 consists of a collection of subfigures for the testing conducted. Unfortunately, the clarity of the figures is very low. The resolution is low and the various subfigures can barely be recognized for the information they carry. The figures in their entirety cannot be read. That should be rectified. In addition, the subset of Figures (C) reflects the permeability of the brain endothelial monolayer following NX210c treatment for two experiments as in [19]. There is no additional information on the two experiments. Beyond what is stated as being in line with those two experiments, the authors should provide a concise description of the two experiments and what the two experiments represent in the milieu of experimental activity in this phase of the work.

Comment 2: Analogous comments hold for Figures 3 and 4. The Figures are illegible. The authors should make sure that the figures are clear so that the preview process takes place to the benefit of the work done.

Answer 1 & 2: We thank the reviewer for pointing this out. The images included in the automatically generated pdf manuscript file are indeed low resolutions of the figures submitted as a separate zip file. We have resent such a folder containing high quality figures to the Editor for transmission to the Reviewers for this second round of revisions. We have also amended the text with details about the experimental work we used as a basis for model validation (lines 150-152).

Comment 3: In lines 195-197, the statement “Moreover, a dose-dependent protective effect of NX210c – i.e. mitigating the decline in BBB function over time – is observed in both healthy and RRMS populations, as compared to the respective placebo group.” should be re-written to read “Moreover, a dose-dependent protective effect of NX210c, i.e. mitigating the decline in BBB
function over time, is observed in both healthy and RRMS populations, compared to the respective placebo group.”.

Answer 3: We have revised the sentence according to your suggestion (lines 200-203).

Comment 4: In lines 386-388, the statement “The compound then transform into its cyclic version, NX210c, for which frequent PK samples were obtained up to 180 minutes post dosing [20].” should be revised to read “The compound then transformed into its cyclic form, NX210c, for which frequent PK samples were obtained up to 180 min post dosing [20].”.

Answer 4: We have revised the sentence according to your suggestion (lines 439-441).

Comment 5: In Figure 7, it is stated that administration of NX210c every 6 months suggests reduction (according to the graphs provided) of the relapses in patients ((A) Relapse activity without treatment. (B) Relapse 242 activity following repeated NX210c treatment cycles). Several questions arise as a result of the graphs shown in the Figure:
a)It is stated that “… repeated NX210c dosing cycles (every 6 months) led to a substantial and sustained reduction in both the frequency and severity of relapses over the simulated decade: from 7 to 4 clinical relapses (+2 subclinical).”. If two relapses are subclinical, then the term should be clarified in terms of the pattern shown in the figure. 

b)If subclinical relapses are reduced post-administration of NX210c, then are there any subclinical relapses in the case of the untreated patients? How does that entail significance in the observed reduction? 

c)There is a reduction of relapses when NX210c is administered at the dose and time intervals opted for. To that end, one can observe that the pattern of remaining relapses in the treated patients is the same (and part of) as the original one for the untreated patients. What is the meaning of that? What is the significance of that in the subsequent route of RRMS patients? Does the pattern itself signify merit in the course of pathology and thus the patient?

d)The pattern of remaining relapses is almost an identical part of the original pattern in the untreated patients. The intensity of the relapse incidents, however, remains the same. What does that imply in terms of the future in the patient’s profile? Along the same lines of thought, what could possibly be the origin of the pattern, thereby allowing for further inquiry into its potential merit in the course of the pathology or the possibility of ameliorating the condition of the patient?

Answer 5: a) We have revised the text to clearly state whether we refer to clinical or subclinical relapses (lines 234-241). Formal definitions of the threshold used to determine (sub)clinical relapses are included in lines 216-218 and 481-484.

b) Our primary objective was to evaluate NX210c’s impact on clinically relevant disease activity rather than subclinical relapses. We have clarified this focus in the revised manuscript and provided a mechanistic explanation regarding relapse severity (see lines 235-251). This distinction underscores that the observed effect pertains to clinically meaningful outcomes, which are most relevant for therapeutic significance.

c-d) The reviewer’s observation is correct and, importantly, mechanistically informative. In the model, NX210c primarily acts upstream by improving BBB integrity and reducing the probability of immune-cell infiltration into the CNS. Consequently, relapses that are predominantly driven by BBB permeability alterations are preferentially prevented. The remaining relapses observed under treatment therefore correspond to a subset of virtual patients characterized by stronger intrinsic autoimmune activation, for whom relapse triggering is less BBB-limited.

This explains why the temporal pattern of residual relapses appears as a subset of the untreated pattern: it reflects differential patient-level susceptibility rather than a uniform treatment failure. Far from being a limitation, this behavior indicates emergent responder/non-responder structure within the virtual cohort, supporting the biological plausibility of the model and motivating stratification strategies based on baseline inflammatory or BBB-related features.

Regarding relapse intensity, the model represents relapse severity through acute inflammatory damage once a relapse is triggered. Since NX210c primarily modulates relapse occurrence rather than the downstream inflammatory cascade, the intensity of breakthrough relapses is not expected to change substantially. Clinically, this corresponds to a treatment effect that reduces relapse frequency and increases time in remission, even if the severity of rare breakthrough events remains comparable. Such an outcome is highly relevant for long-term disease burden.

We have expanded the text to explicitly articulate this interpretation and clarify its implications for disease course and patient stratification (lines 244-251, 317-324, 451-454).

Comment 6: Since IJMS is a journal on molecular sciences and the NX210c is a molecule, one would expect to see a) the structure of the linear and cyclical form of the peptide, and b) the chemistry of its interactions with targets that lead to the amelioration of the profile of the patients studied. A meager and very general elaboration was provided in the manuscript that pertains to the interaction mode(s) of the peptide, as a result of which all of the observations shown in the diagrams emerged. That clarification should be provided in a descriptive manner.

Answer 6: We have expanded the Introduction to strengthen the description of the peptide’s mechanistic rationale, and provide the context underlying the model predictions while remaining aligned with the current state of scientific knowledge: “Although the precise molecular mechanism of action of NX210c is not fully elucidated, previous studies [see previous references in this section] have shown that the peptide enhances TJP expression and improves endothelial barrier stability, leading to reduced permeability and decreased likelihood of autoimmune cell infiltration into the CNS. These downstream effects are consistent with the observed mitigation of relapse‑related oligodendrocyte damage in the present model” (lines 111-116).

Additional information which the mechanistic model is based on is stated in lines 309-319, and has been expanded for further clarity (lines 447-454).

We note that NX210c’s molecular structure was previously published in the referenced study by Delétage et al. (Neuroscience 2021).

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

The authors have conveniently addressed most of the questions and comments, consequently the manuscript has been improved in precision.

Reviewer 2 Report

Comments and Suggestions for Authors

The revised version of the manuscript is significantly improved over the originally submitted version.  The manuscript can now be considered.