Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

A major limitation throughout the review is the frequent discussion of biomarker performance against traumatic encephalopathy syndrome (TES) rather than neuropathologically confirmed chronic traumatic encephalopathy (CTE). While the authors appropriately acknowledge this issue, the review could more explicitly distinguish findings related to TES classification from evidence related to confirmed CTE pathology, as TES remains an imperfect clinical construct with only moderate correspondence to postmortem diagnosis.

The discussion of tau biomarkers focuses primarily on established plasma p-tau epitopes (p-tau181, p-tau217, p-tau231), but emerging advances aimed at detecting disease-specific tau conformations, tau strains, seed amplification assays, and structurally distinct CTE-associated tau species are only briefly mentioned in the future directions section. Expanding this discussion would strengthen the review because the lack of CTE-selective tau detection appears to be one of the principal obstacles to biomarker development.

The authors appropriately introduce the pathognomonic lesion of CTE in the introduction; however, the review contains relatively little discussion of how each fluid biomarker relates to specific neuropathological features such as perivascular sulcal p-tau deposition, astrocytic pathology, neuroinflammation, axonal degeneration, or disease stage. Including a concise section linking biomarker findings to underlying pathology would improve biological coherence and help readers interpret mechanistic relevance.

The authors excluded biomarkers such as UCH-L1, YKL-40, TDP-43, and other candidates because of a limited CTE-specific evidence base. While this is a reasonable scope decision, the rationale would benefit from additional quantitative justification, such as specifying the minimum evidence threshold used for inclusion. This would make study selection more objective and clarify why particular biomarkers were omitted despite their biological relevance.

The manuscript notes that existing evidence is predominantly derived from male athletes but does not discuss potential biological sex differences in biomarker expression, response to repetitive head impacts, or susceptibility to neurodegeneration. Given the increasing participation of women in contact sports and the growing emphasis on sex-specific neuroscience research, a more detailed discussion of this knowledge gap would strengthen the review.

Several AUC values are reported throughout the manuscript, yet the clinical implications of these performance measures are not consistently discussed. For example, modest discrimination values in the 0.70 to 0.75 range are reported for multiple biomarkers, but readers are not provided with context regarding whether such performance would be considered acceptable, inadequate, or potentially useful in combination panels. Additional interpretation would improve clinical relevance.

The review correctly identifies the mismatch between short study durations and the decades-long latency of CTE, but there is limited discussion regarding the expected longitudinal evolution of individual biomarkers from acute exposure through chronic neurodegeneration. A more detailed conceptual framework describing how biomarkers may change across disease stages would provide valuable context for future longitudinal studies.

The review contains a useful comparative table but lacks a visual overview of the mechanistic relationships among repetitive head impacts, blood-brain barrier disruption, axonal injury, astroglial activation, microglial activation, tau pathology, and the associated fluid biomarkers. A schematic figure illustrating these pathways would substantially improve reader comprehension of the proposed biological cascade.

The concluding section states that the reviewed biomarkers provide a substantial foundation for future antemortem CTE detection. While broadly true, the available evidence remains heavily dependent on surrogate clinical diagnoses, cross-sectional studies, and relatively limited autopsy correlation. The conclusions would be more balanced if they further emphasized the preliminary nature of existing validation efforts and the absence of a biomarker with demonstrated clinical diagnostic utility for confirmed CTE.

References
Some landmark references in the CTE literature are not cited in the manuscript. While their omission does not necessarily undermine the review, the authors should consider either incorporating these foundational studies or clarifying the rationale for excluding them.
-clinicopathological study by McKee et al. (2009), which re-established CTE as a distinct neurodegenerative tauopathy and remains a foundational citation for the field.
McKee AC, Cantu RC, Nowinski CJ, Hedley-Whyte ET, Gavett BE, Budson AE, Santini VE, Lee HS, Kubilus CA, Stern RA. Chronic traumatic encephalopathy in athletes: progressive tauopathy after repetitive head injury. J Neuropathol Exp Neurol. 2009 Jul;68(7):709-35. doi: 10.1097/NEN.0b013e3181a9d503. PMID: 19535999; PMCID: PMC2945234.
-Brain 2013 study defining the pathological spectrum and staging of CTE.
McKee AC, Stern RA, Nowinski CJ, Stein TD, Alvarez VE, Daneshvar DH, Lee HS, Wojtowicz SM, Hall G, Baugh CM, Riley DO, Kubilus CA, Cormier KA, Jacobs MA, Martin BR, Abraham CR, Ikezu T, Reichard RR, Wolozin BL, Budson AE, Goldstein LE, Kowall NW, Cantu RC. The spectrum of disease in chronic traumatic encephalopathy. Brain. 2013 Jan;136(Pt 1):43-64. doi: 10.1093/brain/aws307. Epub 2012 Dec 2. Erratum in: Brain. 2013 Oct;136(Pt 10):e255. PMID: 23208308; PMCID: PMC3624697.
-JAMA 2017 clinicopathological series would strengthen historical context.
Mez J, Daneshvar DH, Kiernan PT, Abdolmohammadi B, Alvarez VE, Huber BR, Alosco ML, Solomon TM, Nowinski CJ, McHale L, Cormier KA, Kubilus CA, Martin BM, Murphy L, Baugh CM, Montenigro PH, Chaisson CE, Tripodis Y, Kowall NW, Weuve J, McClean MD, Cantu RC, Goldstein LE, Katz DI, Stern RA, Stein TD, McKee AC. Clinicopathological Evaluation of Chronic Traumatic Encephalopathy in Players of American Football. JAMA. 2017 Jul 25;318(4):360-370. doi: 10.1001/jama.2017.8334. PMID: 28742910; PMCID: PMC5807097.

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

Overall Assessment

This manuscript is a narrative critical review of five candidate fluid biomarkers for antemortem detection of chronic traumatic encephalopathy (CTE), synthesizing evidence mainly from the DIAGNOSE CTE Research Project and smaller RHI-exposed cohorts to argue for multi-marker panels, assay standardization, and longitudinal antemortem-to-autopsy validation. It is timely, and is organized consistently across analytes, with its most useful contribution being a systematic documentation that none of the five markers currently shows CTE-specific discrimination. That said, all five analytes were already known to be biologically non-specific in other disease contexts before this review began, so their failure to discriminate CTE was in large part foreseeable rather than a new empirical finding, which is a limitation inherent to the current state of these assays that revision cannot remove, though the manuscript can and should discuss it more directly. Several additional structural and factual issues, detailed below, are addressable through revision.

Major Issues

1. The Abstract and Conclusion overstate what the evidence supports. The Abstract (lines 29–32) and Conclusion (lines 533–538) describe the evidence as building toward CTE detection and, in the closing sentence, as being assembled "for the first time" with adequate rigor. Yet the body shows p-tau217 tracks Aβ-PET, not CTE (Section 2.2); GFAP and S100B have no standalone CTE-discriminating AUC (Table 1); and sTREM2's primary case–control comparison was null (Section 6.2). Please rewrite both to state plainly that none of the five analytes currently offers CTE-specific discrimination, and reframe the contribution around what was actually established.
2. The TDP-43/α-synuclein co-pathology axis is missing. The Introduction excludes TDP-43 as having a "thinner evidence base" (lines 107–110), but autopsy series (e.g., McKee et al., Brain 2013) show only ~63% of CTE cases are pure CTE, with comorbid TDP-43/motor neuron disease in ~12% and Lewy body disease in ~16%, and report that repetitive trauma promotes co-accumulation of tau, TDP-43, and α-synuclein; RHI-associated TDP-43 pathology is further supported by Nicks et al. (Acta Neuropathol 2023). Since NfL and GFAP are proteinopathy-blind and p-tau reads only the tau compartment, none of the reviewed markers can separate CTE from CTE-plus-comorbidity, which is also a plausible alternative explanation for findings such as the age-restricted NfL association (Section 3.2) or exposure-linked GFAP findings (Section 4.2). Please add a discussion of how these comorbidities limit interpretability across all five analytes.
3. The evidence base draws heavily on overlapping cohorts, and this non-independence is not disclosed. References 6–9, 12, and 14 all report analyses from the same DIAGNOSE CTE cohort (~230 participants), and ref 25 partially overlaps it, so patterns described as converging across "several analyses" (Section 7) are largely one cohort examined from different angles rather than independent replication. Please state, for each analyte, how much of its supporting evidence comes from this single cohort versus independent samples.
4. The TES-based validation discussed in Section 8.2 has circularity. Ref 25's multidomain model (AUC 0.91/0.84) uses cognitive performance to predict TES, but cognitive impairment is itself part of the TES diagnostic criteria (ref 1), risking incorporation bias. Please report the biomarker-only contribution if available, or explicitly caveat this limitation.

Minor Issues

1. Lines 260–263: Ref 26 received FDA marketing authorization via the De Novo pathway; "FDA-cleared" is not the precise regulatory term and should be corrected to "FDA-authorized" or similar.
2. Lines 159–161: The reported null association between p-tau217 and TES/certainty/RHI metrics comes from a longitudinal, repeated-measures analysis (time × group interaction), which is not made explicit and currently reads as a single cross-sectional comparison.
3. Line 187: The "AUCs approximately 0.71–0.74" range is stated without a specific supporting citation for the 0.71 endpoint; only 0.742 (line 168) is traceable to a specific analysis. Please clarify the source of the lower bound.
4. Line 408 directs readers to Section 7 for further discussion of panel-based approaches, but Section 7 itself defers detailed development to Section 8.2 (line 427); please make this signposting consistent.
5. Line 512 begins with "Third" with no corresponding "First"/"Second" earlier in the paragraph.
6. The disclaimer that a null result doesn't establish an accumulation threshold is repeated near-verbatim four times (lines 288–291, 353–355, 434–435, 511–512); state once and cross-reference.
7. Table 1 lists CSF as a sample matrix for S100B, but no RHI-specific CSF S100B data appear anywhere in the text (Section 5.2 reports serum only).
8. Section 5.3's mechanistic hypothesis (lines 366–368) is attributed to ref 16, a Letter to the Editor rather than primary research.
9. Section 4.2 discusses ref 14, which concerns psychiatric symptoms and generic neurodegeneration/microglial markers, not GFAP specifically, despite being placed in the GFAP section.
10. Reference-list formatting is inconsistent: refs 3 and 14 render the journal name as "Alzheimer s & Dementia" (missing apostrophe) while refs 9, 20, and 21 use "Alzheimer's & Dementia" correctly.

Author Response

Please see the attachment. 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Satisfactory

Author Response

I thank the reviewer for their time and for the constructive feedback provided during the first and second rounds, which helped me improve the manuscript. I am pleased that the revisions were found satisfactory. No further changes were made in response to this report. 

Reviewer 2 Report

Comments and Suggestions for Authors

The revision is substantially improved. The remaining issues are factual consistency and the tone of newly added speculative material. I recommend minor revision.

Main Points

1. Exclusion of total tau. Section 1 states that no t-tau study outside EV/CSF proteomic contexts met the inclusion criteria. Alosco et al., Alzheimers Dement (Amst) 2017;7:33–40 (plasma t-tau in former NFL players) and Stern et al., J Alzheimers Dis 2016;51:1099–1109 (plasma exosomal tau) do meet them. Please include these studies or revise the rationale.
2. Independence of evidence. Section 7 treats the EV studies [10,11] and the sTREM2 study [20] as independent sources, but all derive from the DETECT cohort, as do the two t-tau studies above. Section 7 also says the Refs column of Table 1 flags DIAGNOSE CTE studies, which the table does not do. Independent rugby data (Graham et al., Brain 2025: higher plasma p-tau217 in ex-players and more TES among those with elevated p-tau217; companion JNNP 2025 analysis: lower plasma GFAP) bear on the p-tau217 and GFAP conclusions.
3. Abstract. “Short study windows are also poorly matched” has lost its original hedge (“may”) and conflicts with Sections 5.2, 7 and 9. “Most evidence derives from a single cohort” holds for p-tau, NfL and GFAP but not for S100B or sTREM2.
4. sTREM2. The source analysis [20] reported bidirectional mediation (sTREM2 mediating exposure → t-tau, and t-tau mediating exposure → sTREM2, with a feedback loop). The Abstract and Section 6.2 present a one-directional exposure → microglia → tau sequence.
5. Table 2 and Section 8.6. These new elements are hypothesis-generating and go beyond the evidence reviewed.
   • GFAP returning to baseline “within days” is cited to [13,18], but neither reports GFAP time courses.
   • The sTREM2 time course is speculation in the Discussion of [20], not data.
   • “The evidence … supports a three-phase framework” should read “is compatible with”.
   Please shorten both items and label them as hypotheses.
6. Conclusions. “A direction supported by recent multidomain TES analyses” conflicts with the incorporation-bias caveat in Section 8.2.

Minor Points

1. Section 1 (line 47): the sentence beginning “have defined the pathognomonic lesion…” lacks a subject.
2. Line 481: “summerizes.”
3. Section 6.3 (line 469): “Section 7” should read Section 8.2.
4. Section 8.2: AD is described as a “biomarker-poor field.”
5. References 32, 33 and 35 still carry “[to be verified]” placeholders.

Author Response

Response to Reviewer 2 (Round 2)

I thank the reviewer for the careful second reading and for identifying the remaining inconsistencies. All points have been addressed as detailed below. Changes made in this round are highlighted in turquoise in the revised manuscript, first-round changes remain in yellow and green.

The reference list has been renumbered in order of first appearance in this round. Reference numbers below follow the revised manuscript, where the reviewer cited a number from the previous version, the former number is given in parentheses.

Main Points

1. Exclusion of total tau. The reviewer is correct: both Alosco et al. (2017) and Stern et al. (2016) meet the inclusion criteria, and the statement in Section 1 was inaccurate. Total tau has been removed from the list of excluded analytes, and Section 1 now states that t-tau is not treated as a separate analyte but that the two qualifying plasma studies are reviewed alongside p-tau. A new paragraph in Section 2.2 summarizes both studies (new references 11 and 12), including their sample sizes, the null group comparison and exposure association for plasma t-tau, the group separation reported for exosomal tau, and their limitations. Both are also added to the p-tau row of Table 1.
2. Independence of evidence. I agree. The provenance paragraph in Section 7 now states that the EV studies [16,17] (formerly [10,11]), the sTREM2 study [18] (formerly [20]) and the two plasma tau studies [11,12] all derive from the DETECT cohort and are not independent of one another, and that most of the evidence reviewed therefore rests on two American football cohorts. Additionally, the incorrect statement about the Refs column has been removed, a footnote to Table 1 now lists the source cohort of every supporting study, and the text refers to that footnote. The independent rugby data have been added (new references 19 and 20): the p-tau217 findings and the association with TES in Section 2.2, with a qualifying sentence in Section 2.3; the GFAP findings in a new paragraph in Section 4.2, with a corresponding sentence in Section 4.3. Table 1, Section 8.5 and the Abstract have been updated accordingly.
3. Abstract. Both sentences have been revised. The hedge has been restored ("Short study windows may also be poorly matched"), consistent with Sections 5.2, 7 and 9. The cohort statement now reads: "most evidence for p-tau, NfL and GFAP derives from one cohort, with the total tau, extracellular-vesicle and sTREM2 findings from a second." The p-tau217 and GFAP statements in the Abstract have also been qualified considering the rugby data.
4. sTREM2. Section 6.2 now reports the bidirectional mediation described in [18] (formerly [20]): sTREM2 mediating the association between exposure and t-tau, t-tau mediating the association between exposure and sTREM2, and the feedback loop between the two. It states explicitly that these cross-sectional data do not support a one-directional sequence. The Abstract now describes sTREM2 as reflecting microglial activation "that is bidirectionally related to tau after RHI exposure," and the sTREM2 row of Table 1 has been corrected to match.
5. Table 2 and Section 8.6. Both have been shortened and are now explicitly labelled as hypotheses.

• The statement that GFAP returns to baseline "within days" has been removed, together with its citation (formerly [13,18]). Section 8.6 now states that no GFAP time course has been measured in an RHI cohort.
• The sTREM2 time course is now identified as speculation from the Discussion of [18] (formerly [20]), not data, and the corresponding Table 2 entry reads "Unknown; no time-course data."
• "Supports a testable three-phase framework" now reads "is compatible with, but does not establish, a three-phase working hypothesis."
• The Table 2 caption now begins "Hypothesized relationship" and states that the second and third columns are untested hypotheses; the third column heading has been changed to "Hypothesized relation to CTE stage (untested)."

6. Conclusions. The phrase has been revised to: "a direction explored in a recent multidomain TES analysis across two athlete cohorts, although its reported discrimination is subject to the incorporation bias discussed in Section 8.2." The parallel sentence in Section 4.3 has been qualified in the same way.

Minor Points

1. Section 1: the subject has been restored ("Neuropathological diagnostic criteria have defined the pathognomonic lesion...").
2. Section 7: corrected to "summarizes."
3. Section 6.3: the cross-reference now reads Section 8.2.
4. Section 8.2: the sentence no longer describes AD as biomarker-poor; it now reads "As in AD, where a multi-protein CSF panel outperformed individual established biomarkers across multiple cohorts [38]."
5. The references formerly numbered 32, 33 and 35 (now 23, 39 and 41) have been completed and the placeholders removed.

Additional changes

• References are now numbered in order of first appearance, and the reference list has been reordered to match.
• The Abstract has been shortened to bring it closer to the journal's length guideline; its content is otherwise as described under Main Point 3.
• DETECT and t-tau have been added to the Abbreviations list.