Prediction for the prognosis of diffuse axonal injury using automated pupillometry

Objective: Previous studies have reported various predictive indicators of diffuse axonal injury (DAI), but no consensus has not been reached. Although the efficiency of automated pupillometry in patients with consciousness disorder has been widely reported, there are few reports of its use in patients with DAI. This study aimed to investigate the significance of pupillary findings in predicting the prognosis of DAI. Patients and Methods: We included patients admitted to our center with a diagnosis of DAI from June 1, 2021 to June 30, 2022. Pupillary findings in both eyes were quantitatively measured by automated pupillometry every 2 hours after admission. We statistically examined the correlations between automated pupillometry parameters, the patients ’ characteristics, and outcomes such as the Glasgow Outcome Scale Extended (GOSE) after 6 months from injury, the time to follow command, and so on. Results: Among 22 patients included in this study, five had oculomotor nerve palsy. Oculomotor nerve palsy was correlated with all outcomes, whereas Marshall computed tomography (CT) classification, Injury severity score (ISS) and DAI grade were correlated with few outcomes. Some of the automated pupillometry parameters were significantly correlated with GOSE at 6 months after injury, and many during the first 24 hours of measurement were correlated with the time to follow command. Most of these results were not affected by adjustment using sedation period, ISS or Marshall CT classification. A subgroup analysis of patients without oculomotor nerve palsy revealed that many of the automated pupillometry parameters during the first 24 hours of measurement were significantly correlated with most of the outcomes. The cutoff values that differentiated a good prognosis (GOSE 5 – 8) from a poor prognosis (GOSE 1 – 4) were constriction velocity (CV) 1.43 (AUC = 0.81(0.62 – 1), p = 0.037) and maximum constriction velocity (MCV) 2.345 (AUC = 0.78 (0.58 – 0.98), p = 0.04). The cutoff values that differentiated the time to follow command into within 7 days and over 8 days were percentage of constriction 8 (AUC = 0.89 (0.68 – 1), p = 0.011), CV 0.63 (AUC = 0.92 (0.78 – 1), p = 0.013), MCV 0.855 (AUC = 0.9 (0.74 – 1), p = 0.017) and average dilation velocity 0.175 (AUC = 0.95 (0.86 – 1), p = 0.018). Conclusions: The present results indicate that pupillary findings in DAI are a strong predictive indicator of the prognosis, and that quantitative measurement of them using automated pupillometry could facilitate enhanced prediction for the prognosis of DAI.


Introduction
Diffuse axonal injury (DAI) is caused by high-energy trauma with acceleration-deceleration forces, especially rotational acceleration, which produces mechanical stress leading to the severing of axons and consequent axonal degeneration [1][2][3][4].Patients with DAI generally present with severe consciousness disorder in the acute phase and higher brain dysfunction in the chronic phase.However, its sequela varies from consciousness disorder to mild memory disorder and is difficult to predict.Some reports suggest that magnetic resonance imaging (MRI) findings, the Glasgow Coma Scale (GCS), and so on can predict the prognoses of patients with DAI, or that some biomarkers can be predictive factors for the prognosis of DAI in terms of molecular pathology [1,[5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20].
Automated pupillometry is a noninvasive tool that monitors the neurological condition of patients and has recently been used increasingly frequently used in neurocritical care.Previous studies have demonstrated the utility of automated pupillometry mainly in patients with traumatic brain injury with elevated intracranial pressure and in those after cardiac arrest, but also in patients with consciousness disorders such as nonconvulsive status epileptics or delirium [21][22][23][24][25][26][27][28][29][30].Nevertheless, there are no reports in which automated pupillometry has been used purely in patients with DAI, only in DAI subgroups in traumatic brain injury [25].
In the present study, we quantitatively measured the pupillary light reflex in patient with DAI and investigated the feasibility of the parameters as predictive indicators, along with the other possible predictive indicators mentioned in previous studies.

Patients
Patients with a diagnosis of DAI who had been admitted to the Emergency Medicine and Critical Care department at Saitama Medical University from June 1, 2021 to June 30, 2022 were retrospectively screened for inclusion in this study.We did not exclude cases with coexistence of traumatic intracranial hemorrhage other than DAI or cases with coexistence of multiple trauma in the torso and extremities, but we did exclude cases with previously known ophthalmic disease.

Data acquisition
DAI was diagnoded using 3 T MRI, with fluid-attenuated inversion recovery, diffusion-weighted imaging, and susceptibility-weighted imaging.Classification of DAI grade was also carried out by MRI based on Adams classification and previous studies [6,10,31].We used an NPi-200 pupillometer (NeurOptics, Laguna Hills, CA, USA) for the quantitative measurement of variables of pupillary light reflex.This device measures variables of pupillary light reflex; maximum pupil size before (Size) and after (MIN) light irradiation, percentage of change of the pupil diameter before and after constriction (CH), average constriction velocity (CV), maximum constriction velocity (MCV), average dilation velocity (DV), latency between the initiation of the light irradiation and the onset of constriction (LAT), and compute the Neurological pupil index (NPi) integrating these seven parameters using a proprietary algorithm (Table 1).We recorded these parameters in both eyes every 2 hours after admission for a maximum of 14 days.The measurements were conducted in ambient light during the daytime and in the dark during the nighttime.In the cases in which patients presented with anisocoria on admission despite a lack of previously known ophthalmic disease, we considered it oculomotor nerve palsy and continued measuring both eyes.In those cases, the ophthalmologist confirmed that those patients did not have an optic nerve disorder as soon as they were able to undergo ophthalmic examinations.Except for cases with oculomotor nerve palsy, if the automated pupillometer was unable to track the pupil because of eye blinks, closures, or eye movements, or unable to display many of eight parameters, the measurement was discarded and repeated.The median values of the eight parameters during the total measurement period and the first 24 hours after admission were adopted for the statistical analyses.
All patient data on admission were collected from their medical records including age, sex, GCS, Marshall computed tomography (CT) classification [32], existence of abnormal posture, Injury severity score (ISS), with or without blood transfusion, hemoglobin level (Hb), requirement of vasopressor, and existence of oculomotor nerve palsy.In cases in which patients needed to be sedated for treatment, we recorded the duration as the sedation period in which sedative agents and/or opioids were used.The primary outcome was the Glasgow Outcome Scale Extended (GOSE) after 6 months from injury, which was assessed through face-to-face or telephone interview with the patients or their family.The secondary outcomes were the time to follow command, the duration of mechanical ventilation, intensive care unit (ICU) stays, hospital stays, and GCS at the time of hospital discharge.The time to follow command was defined as the period until the patient was able to obey simple order with the eyes open.In the cases in which patients underwent permanent tracheostomy and did not show any sign of being able to speak at the time of hospital discharge, we assessed their GCS verbal score as 1 and then calculated the total score.

Statistical analysis
Concerning the statistical analysis for the correlations between the explanatory variables and GOSE and the GCS at the time of hospital discharge, ordinal logistic regression and the Mann-Whitney's U test were used.Simple regression analysis and the Mann-Whitney's U test were also used to examine the correlation between the explanatory variables and the time to follow command, ventilator days, and the duration of ICU and hospital stays.The degree of correlation was measured by Spearman's rank correlation coefficient, and the degree of difference was measured by effect size 'r'.Multiple regression analysis and ordinal logistic regression were adjunctively used to eliminate the influence of confounding factors as much as possible.A p-value < 0.05 was considered statistically significant.A correlation between an automated pupillometry parameter and an outcome was considered significant when observed in both eyes.All the statistical analyses were conducted using RStudio (version 2022.12.0-353;RStudio, Boston, MA, USA; available at: https://rstudio.com/).
Informed consent to participate in this study was obtained in the form of an opt-out choice on the institute website prior to the study.This study was approved by our institutional ethics committee board and performed in accordance with the ethical standards of the 1964 Declaration of Helsinki and its later amendments.
Significant correlations were found between GOSE and sex, GCS on admission, Hb on admission, and oculomotor nerve palsy.Concerning the automated pupillometry parameters, CV, MCV, DV, and LAT were significantly correlated with the primary outcome (Table 3).Among the patients' characteristics, GCS on admission, abnormal posture, blood transfusion, Hb, vasopressor use on admission, and oculomotor nerve palsy were correlated with many outcomes, oculomotor nerve palsy was the strongest predictive indicator, being significantly correlated with all outcomes (Table 2).None of the outcomes were correlated with Marshall CT classification, and three and one outcomes were significantly correlated with ISS and DAI grade, respectively.Among the automated pupillometry parameters, the median CH, CV, MCV, and DV values were corelated with many outcomes; above all, the median DV was significantly correlated with all outcomes (Table 3).This trend was also observed in the median parameters during the first 24 hours of measurement, and these parameters were correlated with almost all outcomes except GOSE.
Most notably, the median CH, CV, MCV, and DV values during the first 24 hours of measurement were strongly correlated with the time to follow command.
Multiple regression analysis and ordinal logistic analysis were conducted to adjust the influence of confounding factors to the automated pupillometry parameters that were significantly correlated with outcomes, adding the Marshall CT classification, ISS, and sedation period as explanatory variables (Table 4a-c).When adjusted by ISS, the significant correlations that had been observed in the simple regression analysis were not observed in two of the four median parameters in the primary outcome, and in approximately 60% (27/45) of the median parameters in the secondary outcomes.All of the median values that had been correlated with the primary outcome in the simple regression analysis were still significantly correlated when adjusted by sedation time; however, in 17% (7/41) of the median values, the significant correlation with the secondary outcomes that had been observed in the simple regression analysis were no longer seen.The results of the simple regression analysis were not affected when adjusted by the Marshall CT classification.
In addition, simple regression analyses were performed in the same way in the subgroup without oculomotor nerve palsy, because this can affect the automated pupillometry parameters (Table 5).Although no significant correlation with the primary outcome was observed, the median CH, CV, MCV, and DV values during the first 24 hours of measurement were significantly correlated with the time to follow command, the ventilator days, and duration of ICU and hospital stays.
We defined GOSE 1-4 as a poor prognosis and 5-8 as a good prognosis, and drew receiver operating characteristic (ROC) curves to identify the cutoff values for the median automated pupillometry parameters that differentiated the two (Table 6).In addition, the cutoff values for the median automated pupillometry parameters during the first 24 hours of measurement were also identified by an ROC curve that differentiated the time to follow command into within 7 days and over 8 days, because these were especially correlated with the time to follow command.

Table 4a
Result of multivariable regression analysis assigning each outcome as an objective variable and each of the parameters and the Marshall CT classification as explanatory variables.studies have investigated the feasibility of DAI grade as a predictive indicator; however, no consensus has been reached [5,6,[9][10][11]15,34].Although some correlations between DAI grade and outcomes were observed, its significance as a predictive indicator was not proven sufficiently in this study.More importantly, for what we believe is the first time, the present study investigated and suggested the possible efficiency of automated pupillometry parameters as a novel predictive indicator of DAI.
In this study, GCS, abnormal posture, blood transfusion, Hb and vasopressor use on admission were significantly correlated with many outcomes, which is consistent with previous reports [6][7][8][9][10]15,33].Above all, oculomotor nerve palsy was such a strong predictive indicator that it was correlated with all outcomes, a finding that is also consistent with a previous report suggesting that pupillary abnormalities were a predictive indicator for the prognosis of DAI [11,15].Additionally, significant correlations were observed between automated pupillometry parameters, especially CH, CV, MCV and DV, and many of the outcomes in this study, which suggests the importance of pupil observation in patients with DAI.
It is worth noting that oculomotor nerve palsy and sedative agents and opioids could have affect the automated pupillometry parameters in this study.Previous studies have reported that sedative agents such as propofol or dexmedetomidine and opioids can affect pupillary constriction and dilation, resulting in changes in automated pupillometry parameters [35][36][37][38][39][40].Although the NPi is considered to be hardly affected by these agents, some studies have reported that the NPi may be affected by anesthesia, particularly in critically ill patients [35][36][37][38]41].
In the present study, multiple regression and ordinal logistic analyses were conducted using the sedation period as an explanatory variable to adjust for its effect on automated pupillometry parameters; however, many of the median values were significantly correlated with outcomes independent from the sedation period.In addition, subgroup analysis was performed to eliminate the effect of oculomotor nerve palsy on automated pupillometry parameters.As a result, many of the median parameters were significantly correlated with the majority of secondary outcomes, suggesting that automated pupillometry parameters can directly predict the prognosis of patients with DAI without reflecting the effect of oculomotor nerve palsy.
It is also worth noting that the inclusion of cases with the coexistence of traumatic intracranial hemorrhage other than DAI, as well as cases with the coexistence of multiple trauma in the torso and extremities could have affected the results [33,42].Therefore, multiple regression and ordinal logistic analyses were performed using the Marshall CT classification and ISS as explanatory variables.As a result, the median automated pupillometry parameters were found to be significantly correlated with outcomes totally independent of the Marshall CT classification.By contrast, approximately 60% of the results of the simple regression analysis were affected by adjustment using ISS.To interpret this finding, we considered a confounding factor in which a patient with DAI tends to have a high ISS because high score is set for DAI effected on the results of analysis.
Therefore, the findings of the present study suggest the effectiveness of automated pupillometry parameters as a predictive indicator for the prognosis of DAI, independent of coexistent oculomotor nerve palsy or other coexistent traumas, including intracranial hemorrhage.In addition, the fact that the median parameters during the first 24 hours of

Table 4b
Result of multivariable regression analysis assigning each outcome as an objective variable and each parameter and the Injury Severity Score as explanatory variables.GCS; Glasgow Coma Scale, GOSE; Glasgow Outcome Scale-Extended, ICU; intensive care unit, NPi; Neurological pupil index, NS; not significant, Size; maximum diameter, MIN; minimum diameter, CH; percent change, CV; constriction velocity, MCV; maximum constriction velocity, DV; dilation velocity, LAT; latency of constriction.measurement were strongly correlated with the time to follow command, ventilator days, and the duration of ICU stays suggests the possibility that automated pupillometry can be used to predict the prognosis of patient with DAI at the early stage of the treatment and to facilitate the determination of a treatment strategy for coexistent traumas in other parts of the body.
This study has some limitations.The main limitation is the small number of cases, which reduced the statistical power.Therefore, although the results of the present study were consistent with previous studies reporting correlations between pupillary abnormalities and the prognosis of DAI [11,15], the present study could not identify the correlations between automated pupillometry parameters and the prognosis of patients with DAI with sufficient scientific rigor.In addition, the present study could not identify the causal relationship between the prognosis of DAI and the pupillary light reflex because it is complicatedly regulated by the autonomic nerve system.Although previous reports have found that patients with DAI have an increased incidence of paroxysmal sympathetic hyperactivity, an abnormality of the autonomic nerve system [43][44][45], little remains known about how abnormalities of the autonomic nerve system affect the pupillary light reflex in patients with DAI.
Although previous studies have reported the utility of the NPi in patients with acute brain injuries, such as traumatic brain injury, nonconvulsive status epileptics, and post-cardiac arrest syndrome [21][22][23][24][25][26][27][28][29][30], it is difficult to assess the validity of the algorithm used to calculate the NPi.The scientific validity of the NPi in terms of its correlation with the outcomes of DAI cannot be discussed or evaluated because the algorithm for calculation has not been clarified.
In addition, in the present study, we included patients with multiple traumas, including head trauma, as opposed to only patients with "pure" DAI, and this could have produced various confounding factors.Although many of the median automated pupillometry values were correlated with outcomes even after adjustment using multiple regression and ordinal logistic analyses, the possibility of some unexpected confounding factors cannot be denied.These limitations should be carefully considered when interpreting the present results.

Conclusions
In the present study, we investigated predictive factors for the prognosis of DAI, along with a novel factor, automated pupillometry parameters, the significance of which in DAI has not been previously examined.Although oculomotor nerve palsy strongly influences the prognosis of DAI, we also found that pupillary findings could directly reflect the prognosis of DAI.Therefore, the quantitative measurement of pupillary findings in DAI using automated pupillometry could facilitate more accurate predictions of the prognosis of DAI and aid the development of enhanced treatment strategies.

Financial and material support
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Table 4c
Result of multivariable regression analysis assigning each outcome as an objective variable and each parameter and sedation time as explanatory variables.GCS; Glasgow Coma Scale, GOSE; Glasgow Outcome Scale-Extended, ICU; intensive care unit, NPi; Neurological pupil index, NS; not significant, Size; maximum diameter, MIN; minimum diameter, CH; percent change, CV; constriction velocity, MCV; maximum constriction velocity, DV; dilation velocity, LAT; latency of constriction.

Table 5
Correlations between the outcomes of patients without oculomotor nerve palsy and the median automated pupillometry parameters.GCS; Glasgow Coma Scale, GOSE; Glasgow Outcome Scale-Extended, ICU; intensive care unit, NPi; Neurological pupil index, NS; not significant, Size; maximum diameter, MIN; minimum diameter, CH; percent change, CV; constriction velocity, MCV; maximum constriction velocity, DV; dilation velocity, LAT; latency of constriction.

Table 6
Cutoff values that differentiated a poor prognosis (Glasgow Outcome Scale-Extended 1-4) from a good prognosis (Glasgow Outcome Scale-Extended 5-8), and those that differentiated the time to follow commands within 7 days and 8 days or over.

Table 1
Description of the automated pupillometry parameters.

Table 2
Characteristics and outcome of the patients, and their correlations.