Outcome pRognostication of Acute Brain Injury With the NeuroloGical Pupil indEx

Introduction Pupillary examination, and in particular pupillary light reactivity (PLR), is fundamental for intensive care unit (ICU) monitoring and follow-up of patients with acute brain injury (ABI), and has both diagnostic and prognostic values. Secondary cerebral insults, e.g. elevated intracranial pressure (ICP) may alter midbrain function and cause abnormalities in pupil size, symmetry and PLR. Sustained or newfound pupillary abnormalities are associated with a worse outcome, and indeed PLR is a robust validated predictor in several prognostic models, such as the CRASH (Corticosteroid Randomization after Significant Head Injury) and the IMPACT (International Mission for Prognosis and Analysis of Clinical Trials) scores. In current clinical practice however, pupillary examination is performed using a manual, hand-held light source (e.g. pen torch), implying that the evaluation of pupillary size and reactivity remains essentially based on a visual qualitative assessment. This traditional approach has several limitations, such as limited precision (especially in patients with small pupil size), significant intra- and inter-observer variability, differences in ambient light exposure between measurements, or the technique used to direct the stimulus (i.e. intensity, proximity, duration and orientation of the light source).

The use of quantitative, automated, infrared technology for pupillary examination has long been used in ophthalmology and anesthesiology research. Its interest in neurocritical care has progressively grown, in parallel with the advancements in device technology. In this regard, the use of the noninvasive NPi®-200 pupillometer (Neuroptics, Laguna Hills, CA, USA) allows the measurement of a series of dynamic pupillary variables (including the percentage pupillary constriction, latency, constriction velocity, and dilation velocity), which can be integrated into an algorithm, to compute the Neurological Pupil index (NPi). The NPi is a proprietary scalar index with values between 0 and 5 (with a 0.1 decimal precision), an NPi value < 3 indicating an abnormal pupillary reactivity. Importantly, the NPi is not influenced by sedation-analgesia, at the doses used in neurocritical care practice, and by mild hypothermia.

Preliminary single-center data recently demonstrated that abnormal NPi is associated with worse outcome in patients with traumatic and hemorrhagic ABI, and can be a useful adjunct for ICP monitoring and therapy. There is currently a great need for quantitative tools to predict early prognostication in ABI patients, and the NPi appears of potential great value. For this purpose, large multicenter studies are required. The Investigators recently conducted an international multicenter study that demonstrated the prognostic value of NPi in the setting of early prognostication of ABI following cardiac arrest. Here, the Investigators aim at evaluating the prognostic value of the NPi in patients with ABI following traumatic brain injury (TBI), aneurysmal subarachnoid hemorrhage (SAH) or intracerebral hemorrhage (ICH) at risk of secondary ICP elevation.

Research questions

The Investigators hypothesize that:

1. Abnormal NPi (defined as NPi <3) are strongly predictive of poor GOS-E (1-4) at 6 months after the acute event.
2. NPi=0 is strongly predictive of mortality (GOS 1).
3. Abnormal NPi is predictive of a higher ICP 20 index (number of end-hourly measures of ICP >20 mm Hg divided by the total number of measurements, multiplied by 100) and a greater burden of interventions needed to control ICP (measured by the Therapy Intensity Level scale for ICP management, TIL 4).

Methods

Prospective, observational, international cohort study focused at identifying the relationship of NPi with:

• long-term outcome defined as 6-months mortality and neurological recovery, measured with the extended Glasgow Outcome Score, GOS-E;
• intracranial hypertension.

Sample size calculation No formal sample size calculation has been performed due to the exploratory nature of the study. However, the Investigators expect to recruit a total of 420 patients, 140 per pathology for whom GOSE will be performed (i.e. TBI, ICH, SAH), over a 12-months period. Therefore, the twelve participating centres will contribute, based on their potentiality of recruitment, with a minimum of 20 patients.

Screening All patients admitted to the participating ICUs in coma after ABI will be screened daily and entered into a screening log . Each ICU will recruit eligible patients for 12 consecutive months and collect data for each recruited patient daily in an expanded electronic case report form (eCRF). Both common-data elements and etiology-specific data will be collected.

Demographics and Medical History Demographic characteristics and medical history information will be extracted from patients' medical records including gender, age, co-morbidities, diagnosis, timeline and clinical presentation of acute brain injury. All NPi and ICP monitoring data, as well as additional neuro-monitoring and neuroimaging data will be extracted from patients' medical records too and documented in the eCRF.

NPi Data and Daily eCRF The daily eCRF Data Capture will be completed for NPi (collected every 4 hours) and for ICP (matched to NPi), every day, from admission up to day 7 . Data collected will also include additional ICP derived variables and interventions.

Outcome measures Glasgow Outcome Scale-Extended (GOS-E) as main outcome will be collected at ICU/hospital discharge and at 6 months). Data on the cause of death will be collected as well. The GOS-E at the End-of-Study will be collected via telephone-structured interviews to patients and/or family members using a validated questionnaire.

Statistical methods No formal sample size calculation has been performed due to the exploratory nature of the study. However, the Investigators expect to recruit a total of 420 patients, 140 per pathology (i.e. TBI, ICH, SAH), over a 12-months period.

Therefore, the twelve participating centres will contribute, based on their potentiality of recruitment, with a minimum of 20 patients.

Data will be summarised by counts and percentages and quartiles or means and standard deviation, as appropriate, for qualitative and quantitative characteristics, respectively.

Unsupervised and supervised methods will be applied with explorative purposes using the individual NPi longitudinal measurements. For example, the pattern recognition of longitudinal profiles and the cluster trajectory analyses will be used in order to identify patterns of NPi trajectories associated with prognosis. NPi trends will be also described graphically and modelled by longitudinal mixed models using splines.

A Cox and a logistic model will be then applied to evaluate the association between the NPi process with the 6-month mortality and the neurological recovery (GOSE≤4 vs GOSE>4) at 6 months, respectively. This will be done considering NPi in categories identifying different potential patterns in NPi profiles or using summary measures that have been already introduced in this context, such as the percentage of NPi<3 observed in the time interval of observation or the area under the trajectory in time. The two eyes will contribute to these analyses with the worst result. This analysis will be done overall on a multivariable model that will also explore a potential interaction effect with the different pathologies, while the improvement with respect to the standard risk factors (e.g. the components of the IMPACT model for TBI patients) will be evaluated by multivariable models on each of the three specific pathologies.

Lastly, the association between NPi and mortality will be evaluated by the use of shared frailty joint models, in which NPi trend in left and right eye will be evaluated by a multivariate mixed model, which outcome will be used to model mortality. Model-based mortality prediction will be computed based on individual NPi trends. The same kind of approach will be used for the binary outcome.

Potential Risks and Benefits Risks The ORANGE study is observational. It does not introduce any interventional procedure. The data is extracted from the patients' medical records and does not affect local standard of care. Hence, the study does not add any interventional risk to the patients recruited.

Confidentiality breach is a potential risk, which will be addressed by patients' data codification. A unique code will be generated by the e-CRF and allocated to every site and every enrolled patient.

Benefits No individual benefit is expected for the study participants. Nevertheless, the results of the study could possibly help us improving the knowledge for a better medical care for similar patients in the future and the generation of hypotheses for further collaborative research.

Data Collection ICUs willing to participate will register electronically and collect data via an electronic Case-Report Form (RedCAP platform). An online training module will be developed to help data collectors in completing the study eCRF. Data collection will be web based, permitting conditional Data Collection screens, i.e. data collectors will be automatically guided as to which sections to complete based on data entered indicating whether Inclusion Criteria are met.

Ethical standards This study is conducted in compliance with the protocol version 3.0 , the current version of the Declaration of Helsinki, the ICH-Good Clinical Practice (GCP) guidelines, as well as local relevant legal and regulatory requirements of every site.

Ethics committee Each investigator will notify the relevant ethics committee, in compliance with the local legislation and rules.

The approval of the protocol (if required by local authorities) must be obtained before any participant is enrolled. Any amendment to the protocol will require review and approval by the 3 chief investigators before the changes are implemented to the study.

Lack of capacity and Delayed Consent Patients recruited in this study will not be able to provide informed consent at the time of recruitment. The responsible clinical/research staff will act as Consultee and consent eligible patients after discussion with the next-of-kin.

If the patient has a Power of Attorney or a Legal tutor or an, he/she will act as Consultee and will be asked to consent/decline participation to the study on legal behalf of the patient.

If patients have Advance Decision Plan including participation in research studies the Plan will be respected and recruitment pursued/abandoned accordingly.

At follow-up, patients who have regained capacity will be asked to provide Informed Consent and will be given the possibility to:

• Provide Informed Consent for the acute data and follow-up.
• Deny research participation and request destruction of acute data collected.