Significance and Diagnostic Accuracy of Early S100B Serum Concentration after Aneurysmal Subarachnoid Hemorrhage
Round 1
Reviewer 1 Report
The authors provide a revised version of their manuscript, that is clearly improved. The rational, significance and methods of the study are clarified, and the majority of the questiones raised were adressed.
The introduction of Figure 3 is a major improvement, but it should be properly "introduced" in the text (and not only referenced after the pre-test / post-test probability discussions, as it is not directly related to that).
Authors should also clarify that this is a general proposal (and that patient management algorithm should not be based solely on S100B).
Authors still call the study "double-blind observational" and it is not clear for me at all the reason for this. I understand that the patients were blinded to the measurement - even if I don't see how this could affect their own outcomes - and , it is clear now that the team taking care of the patient did not have access to the S100B values, and do not take decisions based on this. However, this is a very unusual phrasing for an observational study.
I have no major comments to the current version of the manuscript.
Author Response
The introduction of Figure 3 is a major improvement, but it should be properly "introduced" in the text (and not only referenced after the pre-test / post-test probability discussions, as it is not directly related to that).
Authors should also clarify that this is a general proposal (and that patient management algorithm should not be based solely on S100B).
As suggested, we added an introduction to the Figure 3 at the end of the first paragraph of the discussion section which specifically discusses the figure. This figure is, as highlighted by the reviewer, our proposal of how to implement a patient management strategy using early S100B serum concentration. The decision making is not solely based on S100B but should be considered along with clinical examination and brain imaging.
Authors still call the study "double-blind observational" and it is not clear for me at all the reason for this. I understand that the patients were blinded to the measurement - even if I don't see how this could affect their own outcomes - and , it is clear now that the team taking care of the patient did not have access to the S100B values, and do not take decisions based on this. However, this is a very unusual phrasing for an observational study.
The term double-blind has been removed from the manuscript and changed to blinded in the abstract as well as in the experimental and discussion sections.
Reviewer 2 Report
The reviewer's comments and concerns have been addressed in the revised manuscript.
English editing is advised.
Author Response
we thank you for your work in proofreading our manuscript.
A new revision of the English was carried out by a native spearker.
Reviewer 3 Report
The authors adequately addressed comments in the revised manuscript.Author Response
we thank you for your work in proofreading our manuscript. The Figure 3 add much value to the manuscript.
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
The authors have designed an observational study to assess the role of early S100B measurements in assessing 3 days prognosis of patients with SAH.
The topic is relevant for clinical practice, and it could be informative for early prognostication.
However, the manuscript has some caveats - mostly related to readibility and internal logic consistency more than design or analysis.
The rational behind the introduction is difficult to follow and it does not clarify the hypothesis being studied.
The role and significance of EBI should be stated more clearly (in contrast to DCI, a well known contributor for bad outcome in SAH).
It is not provided a consistent rational on why S100B should be a good marker, and it's significance in brain injury.
The background does not feed enough information to understand the hypothesis.
Methods:
The use of the concept of double-blind observational study is not a common concept in clinical research and could lead to confusions about the design. The tudy design seems to be a pragmatical one with consecutive inclusion of patients. I believe it is important to discuss that the treating ICU team was blind to the S100B levels (that they did not influenced the therapeutic choices) and that they did not influence the assessment of the outcome (that is the mGCS, but, as the authors discuss it is a robust marker). Interestingly, the authors did not provide information about blinding for the primary outcome (mGCS) but only for a non-primary outcome (mRankin). Can they clarify this? As a summary it is hard to me to understand a rational to call this a "double-blind observational" study.
Results:
3.1 - Descriptive statitics of all groups, but no information regaring the outcome chose by the authors (mGCS). We can't even infer this, as no data of GCS a discharge is provided. Even if it is by definition 6 in the mild and moderate group, what about the severe? A score of 1 would be very different from a score of 5.
3.2 - There were 81 patients included. But weren't these the same patients discussed before? The first sentence may be confusing. Were the S100B levels correlated with the any kind of parenchimal lesion (for instance, brain edema as assessed by the SEBES scale)
3.3 - The concept of the gray zone deserves to be clarified.
3.4 - What is the median maximal serum S100B? At admission/day one? Or at any time during the stay. I don't understand the meaning of doing a comparison across all mRS categories. Usually, a cut-off of 2 is applied (<2 - no disability; >=2 disability). This would be more relatable to any relevant endpoint.
Discussion.
The discussion is structured in very confusing way.
GOS i introduced without any details about what it is. Aditionally, the paragraphs between line 313-321 are hard to follow. S100B during the first 2 weeks have the best diagnostic accuracy, then it has a poor value in the first 24 hours for predicting poor outcome, but then it can predict bad outcome?
Line 328 - Cerebral microanalysis must be a typo?
Limitations on the extrapolation of the data must also be recognized. Would we expect different cut-off values using different populations? Would this analysis be biased by the population it used? What would we expect?
Reviewer 2 Report
Balança et al – jcm-764054
Objective:
The authors present a prospective study on the value of serum S100B as a potential predictor of short-term outcome/EBI (day3) after aneurysmal subarachnoid hemorrhage (aSAH).
Methods:
S100B was assessed on day 0 or day 1 following the ictus.
Outcome at day 3 was assessed using the Glasgow Coma Scale Motor Score (GCS-MS), with the latter being used to define early brain injury (EBI) as follows:
- EBI-mild group: M-GCS =6 at arrival and GCS =6 at day 3,
- EBI-moderate group: M-GCS<6 at arrival and GCS= 6 on day 3
- EBI-severe group: M-GCS < 6 at day 3 regardless of the M-GCS on arrival.
Results:
The maximal S100B value at admission or day 1 was significantly higher in the EBI-severe group than both in the EBI-moderate and the EBI-mild group.
A maximal S100B value between days 0-1 had an AUC of 86.7% to predict M-GCS<6 on day 3.
Conclusion:
They suggest closely monitoring patients with a high early serum S100B value. The claim that S100B might become a surrogate marker of EBI-severity if results are confirmed by an independent trial.
Although the topic is of interest to a special audience, the methods, as well as the statistical analysis, fail some important features, such as correction for confounders. As such, the results and the conclusion are not reliable. Despite the strength of the prospective protocol and the detailed patient characterization, this study is too small to draw definitive conclusions.
In addition, the manuscript is poorly written and makes it difficult to follow, which is a major drawback. Please also check spelling (aneurysm, not aneurism; Hijdra score, not Hidjra score, etc...).
I feel that the presentation of the paper should be improved in order to increase the readability and accessibility of the paper. The study suffers from a number of additional major limitations:
Additional concerns:
Despite being a prospective trial, the study is poorly designed.
- It fails to adhere to the current definition of what EBI is, also from a pragmatic point of view.5
- The GCS-MS used is accurate in the assessment of traumatic brain injury.3 As for aSAH, the World Federation of Neurological Surgeons (WFNS) grade is the test of choice. 1
- Following the initial ictus, patients’ condition may be confounded by seizure, hydrocephalus or sedation.2 It is not quite clear to me, how the authors did incorporate and valued these parameters. In addition, a surgical approach, even the insertion of an EVD and the presence of intracerebral hemorrhage due to the intervention interfere with S100B values. Although listed, they are not weighted correctly. Due to the fact that the majority of poor grade SAH patients also display an ICH, the data must be interpreted with caution in relation to EBI. Whether high levels of S100B can be attributed solely to the severity of the SAH as rated by the GCS-MS grade or also to the occurrence of ICH itself fails to be answered.
- It is stated that the physicians and nurses in charge of the patients were blinded to S100B levels. How? Was access to laboratory values not granted to them? Please specify.
- The necessary number of participants was calculated using the Obuchowski method. A minimum total population of 64 patients including 8 cases (patients with a M-GCS<6 at day 3) was necessary to detect an AUC >0.8, with a power of 0.8, an alpha risk of 0.05. This is contradictory to what was estimated at the time the study was registered: https://clinicaltrials.gov/ct2/show/NCT02890004. Please discuss. Also, be aware that begin of study and recruitment doesn’t coincide. See also next two comments.
- The small number of patients with severe EBI limits the extrapolation to poor grade aSAH patients in general, which, as correctly stated in the limitations, would require a larger validation cohort. Why did the authors not continue including patients in their prospective trial as it was foreseen at the time of the study registration?
- There were 10 missing measurements of S100B at both the admission on day 1. Hence, as for a prospective trial, the study should include results on the 61 patients, not the 81 patients reported on, although this interferes with the power or the study.
- Lines 317: it’s not correct that DCI is known to be associated with an increase of S100B. Pooled analysis did not find any association.4 However, prognosis is dependent by EBI and DCI, and is associated with a S100B increase.6
Given these shortcomings and the poor study design, I won't recommend your article for publication in JCM.
References:
- Report of World Federation of Neurological Surgeons Committee on a Universal Subarachnoid Hemorrhage Grading Scale. J Neurosurg 68:985-986, 1988
- Fung C, De Marchis GM, Katan M, Seiler M, Arnold M, Gralla J, et al: Copeptin as a marker for severity and prognosis of aneurysmal subarachnoid hemorrhage. PLoS One 8:e53191, 2013
- Healey C, Osler TM, Rogers FB, Healey MA, Glance LG, Kilgo PD, et al: Improving the Glasgow Coma Scale score: motor score alone is a better predictor. J Trauma 54:671-678; discussion 678-680, 2003
- Lai PM, Du R: Association between S100B Levels and Long-Term Outcome after Aneurysmal Subarachnoid Hemorrhage: Systematic Review and Pooled Analysis. PLoS One 11:e0151853, 2016
- Suzuki H: What is early brain injury? Transl Stroke Res 6:1-3, 2015
- Weiss N, Sanchez-Pena P, Roche S, Beaudeux JL, Colonne C, Coriat P, et al: Prognosis value of plasma S100B protein levels after subarachnoid aneurysmal hemorrhage. Anesthesiology 104:658-666, 2006
Reviewer 3 Report
This article is methodologically sound, with potentially important clinical prognostic decision pathways using GCS-M and S100B levels. Authors may consider inclusion of clinical prognostication algorithm chart.