Pupillary dilation velocity is reduced in intensive care unit patients with septic shock

Septic shock is common in the intensive care unit (ICU). The pathophysiology is poorly understood but prolonged sympathetic activation leading to autonomic dysfunction may be involved. Pupillary light response (PLR) is a fast, inexpensive, noninvasive way to measure autonomic nervous system function. The aim of the study was to observe dilation velocity of the PLR (PLRdil.vel.) in patients with and without septic shock and explore whether other factors influenced the possible association. We hypothesized that the presence of septic shock in intensive care patients is associated with changes in sympathetic autonomic tone, which can be observed as changes in PLRdil.vel.


Editorial Comment
Sympathetic dysregulation is common in septic shock, likely linked to severity and prognosis.
This study reveals reduced pupillary dilation velocity in septic patients, suggesting potential for future research on its connection to mortality and interventions guided by objective measures of sympathetic dysregulation.

| INTRODUCTION
2][3][4][5][6][7] Septic shock may lead to prolonged dysregulation of the sympathetic autonomic nervous system. 8Many patients in septic shock also suffer from delirium, which by itself may be associated with autonomic nervous system dysregulation. 3,8[14][15] The pupil diameter is regulated by a balanced interplay between the two parts of the autonomic nervous system.[18] The aim of the study was to observe PLR dil.vel. in patients with and without septic shock and explore whether other factors influenced the possible association.We hypothesized that the presence of septic shock in intensive care patients is associated with changes in sympathetic autonomic tone that can be observed as changes in PLR dil.vel. .

| MATERIALS AND METHODS
This was a prospective observational cohort study conducted from 1 October 2018, to 21 January 2020.We screened patients admitted to the 12-bed general mixed ICU at Copenhagen University Hospital -North Zealand.
Inclusion criteria were acute ICU admission and age ≥18 years.
Patients were included as early as possible after ICU admissionusually within 24 h.Exclusion criteria were a state of permanent incompetence (i.e., dementia or mental retardation), nonapplicability for delirium assessment (i.e., due to coma or language barriers), withdrawal from active therapy or brain death, positive urine human chorionic gonadotropin (hCG) or plasma-hCG in fertile women, a nonobtainable consent according to national regulations, coercive measures, alcohol-induced delirium (delirium tremens) or prone position ventilation.All patients or next-of-kin gave written informed consent.The study was approved by the regional Committee on Health Research Ethics (protocol no.H-18030538).
The study was paused for 4 months (June-September) because of concern that warm weather during summer could influence the measurements.
Pupillary dilation velocity (PLR dil.vel. ) was measured each day from inclusion until ICU discharge, both in the morning and the afternoon with a portable pupillometer PLR-3000™ from NeurOptics ® .The pupillometer automatically tracked the pupil contraction and dilation and recorded PLR dil.vel in 5 s following a white light stimulus of 50 μW lasting for 800 ms.The right eye was measured first.Measurements exceeding defined limits were excluded.0][21] Ambient light was measured concomitantly with PLR dil.vel.
Septic shock was diagnosed according to the current Sepsis-3 definition 4 and patients were categorized as being in septic shock when they met all criteria.They were considered out of septic shock when a noradrenaline infusion was no longer required as judged by the clinical team caring for the patient.The individual patient could change from one categorization to the other, for example when they were no longer in septic shock.The category was taken into account with every measurement (two times a day) and the mixed models take this into account.
We collected data on doses of intravenously administered noradrenaline and adrenaline in patients receiving these two vasopressors.To have a measure of both exogenously and endogenously noradrenaline and adrenaline, blood samples were taken each morning from inclusion and throughout a patient's ICU stay for analysis of the two metabolites methoxynoradrenaline and methoxyadrenaline.
Plasma from centrifuged whole blood samples were frozen for shipment to the Department of Clinical Biochemistry at Hvidovre Hospital where the metabolites were measured by mass spectrometry. 22rticipants were assessed twice per day with the Confusion Assessment Method for the ICU (CAM-ICU) and were considered to be in delirium when the CAM-ICU was positive. 23Patients who were intermittently comatose, where not assessed during coma but assessed before and after.
From the medical records, we collected the demographic characteristics age, sex, mechanical ventilation, Sequential Organ Failure Assessment score (SOFA) and Simplified Acute Physiology Score 3 (SAPS3), ICU length of stay, alcohol, and smoking habits.

| STATISTICS
For the baseline demographic data patients were grouped based on if they ever during the study had septic shock or not.A two-sided Student's t-test was made for age and SAPS 3. Nonparametric Wilcoxon rank sum test was made for length of stay and SOFA.Fisher's exact test was used for the categorical variables sex, mechanical ventilation, and alcohol and smoking habits.
A linear mixed model was used for the analysis of association between PLR dil.vel and septic shock at a given point of time.The model was constructed with the presence or absence of septic shock as a fixed effect in relation to PLR dil.vel.
The two random effects side (right or left eye) and measurement occasion were nested to each patient (i.e., patients were also a random effect set one level above side and measurement occasion) allowing for multiple measurements from individual patients to be used.Measurements from both eyes were included concomitantly in the models as no systematic effect between the right and the left eye All CI given are 95% confidence intervals.A p < 0.05 was considered significant.Due to rounding, some differences and CI boundaries are reported as zero.
All statistics were done using RStudio version 2023.03.1, build 446. 24Specifically for the linear mixed model, the lmerTest package 25 was used.The biostatistical consulting service at the Department of Biostatistics at the University of Copenhagen was consulted regarding the construction and interpretation of the linear mixed models.

| Demographics
One hundred patients were included, see Figure 1.Ninety-one had PLR measurements (mean age 67.7 years [SD = 13.3],63.7% males) and of these 35 had septic shock at some point (see Table 1).
The patients with septic shock had on average a longer ICU stay compared with those without (5.0days longer (CI [2.5; 8]).A larger proportion of the patients with septic shock were mechanically ventilated at some point during the study period compared with patients without septic shock ( p = 0.01).
Expectedly patients with septic shock had a higher SOFA score (CI [2; 5]) and higher SAPS3 scores (CI [1.0; 12.9]) than those who never had septic shock.There was a significantly larger part of the septic shock patients who experienced delirium than those never having septic shock ( p = 0.02) There was no difference in sex, age, alcohol intake or smoking status between the study groups.

| Pupillary light response
Septic shock was associated with a slowed PLR dil.vel.The mean of PLR dil.vel.was significantly lower in the presence of septic shock than in the absence of septic shock.PLR dil.vel.measurements taken in the presence of septic shock had a mean of 0.6 mm/s (CI [0.5; 0.7]) and measurements taken in the absence of septic shock had a mean of 0.9 mm/s (CI [0.8; 1.0]).The difference in means was À0.3 mm/s (CI [À0.2; 0.4], p < 0.001).Figure 2   Analysis of all the covariables was performed (see Supporting information S1).There was no significant association between PLR dil.vel.and any of the covariables, as well as no significant interaction between any of the covariables and septic shock with three exceptions.Being on a mechanical ventilator had significant association with a slower PLR dil.vel.(CI [À0.2;À0.1]), however not when coupled to septic shock in an interaction term.Current smoking also had a significant association with PLR dil.vel.when in an interaction term with septic shock (CI [À0.7;À0.2]).Noradrenaline infusions were significantly associated with a slower PLR dil.vel.(CI [À0.54;À0.1]); however, not when coupled to septic shock in an interaction term (CI [À0.1; 1.31]).The number of patients and measurements varied between the analyses as only measurement occasions containing both data on PLR dil.vel.and the covariable in question were used.
Comorbidities were tested for possible influence on the PLR dil.vel as well.'Infection' ( p = 0.001) and 'neurological' ( p = 0.02) showed a significant uneven distribution between patients in septic shock and patients not in septic shock but showed no convincing effect on PLR dil.vel ("infection" CI [À0.

| DISCUSSION
We found that septic shock was associated with a slower PLR dil.vel .This is in line with the established theory of prolonged sympathetic activation leading to autonomic dysregulation in septic shock. 8,26This dysregulation leads to a severely decreased sympathetic tone on the blood vessels and need for external noradrenaline to sustain an acceptable blood pressure.
In this explorative study, we observed that noradrenaline infusions were associated with a slower PLR dil.vel.This is contrary to our understanding that increased noradrenaline should be associated with a faster dilation velocity, as noradrenaline is the neurotransmitter in the sympathetic innervation of the pupillary dilator muscles.
Circulating catecholamines might affect the dilator and sphincter muscles of the pupil, 9 and larger pupils have been associated with higher concentrations of noradrenaline. 13An interpretation of our findings is that septic shock is associated with a depletion of endogenously noradrenaline and hence a slower pupillary dilation velocity.
We speculate that under normal physiological circumstances (i.e., in the absence of disease), infusion of noradrenaline would result in a faster pupillary dilation velocity.However, in the presence of septic shock, endogenously noradrenaline is depleted.In our study popula-

| LIMITATIONS
A limitation of the study was that only one researcher was available for measurements, leading to missed inclusions and missed data whenever the researcher was not available.Patients were mostly included during the daytime Monday to Friday and patients who were admitted and discharged or died during a weekend were missed.This may lead to a systematic error since more patients end up in the ICU during the weekends because of general less experienced staffing.
8][29][30] Whether this has actual implications in our study is unclear but should be remembered as a possible confounder.
As mentioned in the Section 2, the study was paused 4 months during because of concern that warm weather during summer could influence the measurements.No specific room temperature interval was a priori decided for, as the building's temperature control unit was believed to keep the room temperature within a comfortable range-however, during the summer the temperature control unit was not able to keep the ICU constantly cooled (temperatures above 28 C was measured) and we therefore decided to stop measurements when this happened, as we speculated that abnormally warm ambient temperature might affect the autonomic nervous system.
The cohort received many kinds of therapeutic interventions.We expected sedation to possibly influence the measurements.Some medications used for light or heavy sedation (haloperidol, benzodiazepines, dexmedetomidine, opioids, propofol) are known to influence the PLR although not described in specifics. 10,14,31,32One study found that opioids only cause miosis but that the PLR was otherwise unaffected, 33 However, the correlation has yet to be investigated for a more precise pupillometry.We did not collect the specific data of sedative medication, but our results may translate to a clinical context as mechanical ventilation is a clinically applicable proxy variable, as all mechanical ventilated patients received some level of sedation.We found that being mechanically ventilated were associated with a slower PLR dil.vel.It is however a major limitation of this study that we do not have data on all medications that potentially could influence PLR dil.vel.
Furthermore, some of the drugs used in the ICU have known sympathomimetic effects, that is, noradrenaline, adrenaline, fenoterol, salbutamol, and various β-blockers.Haloperidol causes miosis 14 but possibly also influences other parameters of pupillometry.There was an ongoing blinded randomized clinical trial of low-dose haloperidol effect on delirious patients in the ICU, Agents Intervening against Deliriumin the Intensive Care Unit (AID-ICU). 34The delirium covariate might be influenced by the above-mentioned factors as a bias of an unknown magnitude.

| CONCLUSION
Septic shock may be associated with changes in sympathetic autonomic tone which is supported by the findings from this study that septic shock was associated with a slower dilation velocity in the pupillary light reflex.Further studies should examine if the pupillary dilation velocity may serve as surrogate marker for changes in sympathetic autonomic nervous system activity in intensive care patients in septic shock.If so, future interventional studies should test if use of the pupillary dilation velocity may be used for earlier detection of septic shock, which could mean earlier institution of treatment measures for this condition.

AUTHOR CONTRIBUTIONS
The manuscript was drafted by Stine Uhrenholt.All authors have contributed to the design of the study, discussing and revising the manuscript.Stine Uhrenholt and Signe Maria Linér has collected data.
the two research nurses, Lone Valbjørn and Sanne Lauritzen for their help in including patients in the study.The authors declare no conflicts of interest.
was found (confidence intervals [CI À0.1; 0.1]) in a preliminary model with side (right or left eye) as the only fixed effect.Time was represented by measurement occasions with up to two occasions per day per patient.The occasions happened with different actual time intervals randomly dispersed within and among patients.No measurement occasions were during the weekends and some work-week days due to absence of the research team.Several covariables were tested in additional linear mixed models with each covariable added as a fixed effect in turn to the model described above.These covariables had values that were either static for each patient throughout the study or had values that were dynamic over time.The covariables with static values were sex, age, current smoking, alcohol intake, and SAPS3 score.The covariables with dynamic values were ambient room light, mechanical ventilation, delirium (CAM-ICU positive), SOFA scores, adrenaline and noradrenaline infusions, body temperature, and plasma values of methoxyadrenaline, noradrenaline, lactate, and C-reactive protein.Dynamic values were matched with the PLR dil.vel.value for a particular measurement occasion in the linear mixed models.For all linear mixed models, only measurement occasions containing both data on PLR dil.vel.and the covariable in question were used.
shows the PLR dil.vel.measurements in relation to the presence or absence of septic shock and in relation to time.

F I G U R E 1
Consort diagram.a Other reasons: Four included in earlier admittance, one prone position ventilation, and one could not cooperate to measurements after first assessment.b Death before written consent was obtained and no next of kin.c Did not have PLR data, light stimulus did not induce response in pupil that the pupillometer was able to detect.PLR, pupillary light response.

a
If in septic shock/delirium at some point throughout the ICU stay.b If never septic shock throughout the ICU stay.c Alcohol intake above recommended by Danish Health Authorities in 2020 (84 g of pure alcohol for women/168 g for men).

F I G U R E 2
Dilation velocity of the pupillary light response (PLR dil.vel. ) and septic shock.(A) PLR dil.vel.divided if there was no septic shock or septic shock at each measurement occasion.Each point represents a single measurement.Note that the x-axis has been truncated at 30 days.(B) The means of PLR dil.vel.at each measurement occasion.
tion, noradrenaline infusion solely in the presence of septic shock was nearly significantly associated with an increased dilation velocity.It is possible that both septic shock and noradrenaline infusions are associated with changes in PLR dil.vel.independently of each other.PLR dil.vel.may serve as surrogate marker for changes in sympathetic autonomic nervous system activity in intensive care patients in septic shock and used as such in future studies on the role of the sympathetic autonomic nervous system in the pathology of septic shock.To the best of our knowledge, no other group has studied whether septic shock influences the dilation velocity of PLR and no other studies have examined the possible influence of intravenously noradrenaline infusions on the PLR dil.vel .