Prospective study examining the impact of cerebral angiography on quantitative pupillometer values in the interventional radiology suite

Abstract

Objectives

The purpose of this pilot study was to obtain baseline quantitative pupillometry (QP) measurements before and after catheter-directed cerebral angiography (DCA) to explore the hypothesis that cerebral angiography is an independent predictor of change in pupillary light reflex (PLR) metrics.

Design

This was a prospective, observational pilot study of PLR assessments obtained using QP 30 min before and after DCA. All patients had QP measurements performed with the NPi-300 (Neuroptics) pupillometer.

Setting

Recruitment was done at a single-centre, tertiary-care academic hospital and comprehensive stroke centre in Dallas, Texas.

Participants

Fifty participants were recruited undergoing elective or emergent angiography. Inclusion criteria were a physician-ordered interventional neuroradiological procedure, at least 18 years of age, no contraindications to PLR assessment with QP, and nursing transport to and from DCA. Patients with a history of eye surgery were excluded.

Main outcome measures

Difference in PLR metric obtained from QP 30 min before and after DCA.

Results

Statistically significant difference was noted in the pre and post left eye readings for the minimum pupil size (a.k.a., pupil diameter on maximum constriction). The mean maximum constriction diameter prior to angiogram of 3.2 (1.1) mm was statistically larger than after angiogram (2.9 (1.0) mm; p<0.05); however, this was not considered clinically significant. Comparisons for all other PLR metrics pre and post angiogram demonstrated no significant difference. Using change in NPi pre and post angiogram (Δpre=0.05 (0.77) vs Δpost=0.08 (0.67); p=0.62), we calculated the effect size as 0.042. Hence, detecting a statistically significant difference in NPi, if a difference exists, would require a sample size of ~6000 patients.

Conclusions

Our study provides supportive data that in an uncomplicated angiogram, even with intervention, there is no effect on the PLR.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Obtaining data pupillometer values prospectively before and after angiography is a recognised strength of the study design.

• The sampling strategy used was pragmatic and limited by staff availability.

• Sampling a diverse cohort has the advantage of providing greater generalisability.

• The results should be interpreted with caution as they were obtained from a single centre where quantitative pupillometry is standard of care.

Introduction

Assessment of the pupillary light reflex (PLR) is widely recognised as a fundamental tenet of the modern neurological examination for all critically ill patients.^{1 2} This simple non-invasive assessment can detect neurological deterioration, retinal axonal loss and autonomic dysfunction in critically ill patients but suffers from poor inter-rater reliability when assessed by clinicians.^{3 4} Although PLR assessment by human observers has limited reliability,⁵ quantitative pupillometry provides high reliability, precision and accuracy to provide information on acquired brain injury.^{6 7} Commensurate with the increasing pace at which pupillometry is being adopted into practice,^{8 9} there is a need to fully understand the limits of this technology. Specifically, there are no data examining how invasive neurological procedures impact pupillometer measurements. We propose that angiography may direct an effect on PLR measurements based on the invasive nature of the procedure along with prior studies suggesting that global cerebral dysfunction may result from intra-arterial iodinated contrast administration (such as contrast-induced encephalopathy) leading to diffuse cerebral oedema that can be potentially diagnosed using pupillometry.¹⁰ The purpose of this pilot study is to obtain baseline pupil measurements before and after catheter-directed cerebral angiography (angiography) to explore the hypothesis that angiography is an independent predictor of change in PLR metrics.

Although developed roughly 100 years ago, the success of Seldinger’s work in the 1950s spawned a steady increase in the type, frequency and availability of angiography globally.¹¹ The advantages of pupillometry are well documented in providing new metrics with higher accuracy, precision and reliability than subjective PLR assessment.^12–15 What remains unknown is whether changes in PLR metrics can be attributed to periprocedural events or whether these changes reflect new or worsening neurological deficits. Addressing this fundamental question will provide clinicians with knowledge that directly impacts practice and will provide researchers with data to determine the degree to which periprocedural events are confounding variables in pupillometry-related research.

Methods

This is a single-centre, prospective, observational pilot study of PLR assessments obtained using a pupillometer both 30 min before (to establish baseline) and after angiography (to assess for neurological deterioration). All procedures were per the ethical standards set forth by the Helsinki Declaration of 1975 and its subsequent addendums. The study was approved by the university institutional review board which determined that patients were exempt from written consent because routine serial PLR assessment with pupillometers is the standard of care at our institution.

Inclusion criteria were a physician-ordered interventional neuroradiological procedure. This includes patients scheduled, but not limited to angiography, embolisation, biopsy, vertebroplasty, thrombectomy and those admitted to the neurosciences intensive care unit (ICU) with a neurological or neurosurgical diagnosis, at least 18 years of age, no contraindications to PLR assessment with a pupillometer and nursing transport to and from angiography. Patients with a history of eye surgery were excluded. Pragmatic recruitment was used to identify the first 50 eligible patients. All patients had pupil measurements performed with the NPi-300 (Neuroptics) pupillometer with assessments being performed in the ICU and in the angiography suite. All pupillometer data were downloaded directly from the device to an electronic spreadsheet. Pupillometry data before surgery began was obtained in the holding area in our angiography suite or the Neurointensive care unit. Data after surgery were taken in the post-op area or the Neuro ICU. Lighting conditions are similar but were not recorded as part of our data collection. Prior studies suggest that ambient lighting conditions are unlikely to play a significant factor in affecting the accuracy of pupil size or reactivity measurements.⁶ Demographic and baseline data were extracted from the electronic medical record and entered into an electronic case report form. Data were uploaded into SAS V.9.4 for Windows for analysis.

This observational study reported the variation in paired pupillary reading before and after angiography. Data are reported as mean (SD) or frequency (per cent) unless noted otherwise. Paired t-test models were constructed to evaluate the differences between pre and post angiogram pupil metrics. The Fisher exact test was used to compare categorical variables. A p value of <0.05 was considered statistically significant for all analyses.

Patients and public involvement

Patients and public were not involved in any stage.

Results

Observations were recorded on 50 patients between January and March 2022. Patients were predominately male (60%), white (70%) and non-Hispanic (92%), with a mean age of 57.4 (18.8) years (table 1). The primary reason for angiogram was unruptured aneurysms (14 (28%) and haemorrhagic stroke (13 (26%)) which included aneurysmal subarachnoid haemorrhage and intracerebral haemorrhage. The mean modified Rankin scale scores on admission (1.1 (0.9)) were not significantly different at discharge (1.2 (1.4); p=0.714).

Table 1.

Baseline characteristics of patients

Characteristic	Patients
Age mean (SD)	57.4 (18.8)
Sex
Male (%)	30 (60%)
Female (%)	20 (40%)
Race
African American	8 (16%)
Asian	1 (2%)
Caucasian	35 (70%)
Native American	1 (2%)
Pacific Islander	0 (0%)
Other	5 (10%)
Ethnicity
Hispanic	4 (8%)
Not Hispanic	46 (92%)
Admission mRS
0	11 (22%)
1	28 (56%)
2	8 (16%)
3	2 (4%)
4	1 (2%)
Discharge mRS
0	15 (30%)
1	26 (52%)
2	3 (6%)
3	1 (2%)
4	3 (6%)
5	1 (2%)
6	1 (2%)
Diagnosis
Brain tumour	4 (8%)
Unruptured aneurysm	14 (28%)
Haemorrhagic stroke	13 (26%)
Ischaemic stroke	7 (14%)
Other	12 (24%)

Haemorrhagic stroke, subarachnoid haemorrhage and intracerebral haemorrhage; mRS, modified Rankin scale score.

Pupillometer readings were obtained within 30 min before the angiogram in all 50 patients, and 44 (88%) had follow-up readings obtained within 30 min of completing the angiogram. There were no procedural complications and none of the patients in our sample had clinically significant neurological changes during the perioperative period up to 2 hours post angiogram. Nurses were more likely to obtain post-angiogram pupil measurements for patients with lower baseline NPi values of 4.3 (0.38) compared with those with higher mean NPi values at baseline (4.6 (0.16); p<0.05).

The QP values met the assumption of being approximately normally distributed and therefore the primary null hypothesis was explored using paired t-test. Models were constructed to explore each PLR metric (table 2). The only statistically significant difference was noted in the pre and post left eye readings for the minimum pupil size (a.k.a., pupil diameter on maximum constriction). The mean maximum constriction diameter prior to angiogram of 3.2 (1.1) mm was statistically larger than after angiogram (2.9 (1.0) mm; p<0.05); however, this was not considered clinically significant. Comparisons for all other PLR metrics pre and post angiogram demonstrated no significant difference (table 2). Using change in NPi pre and post angiogram (Δpre=0.05 (0.77) vs Δpost=0.08 (0.67); p=0.62), we calculated the effect size as 0.042. Hence, detecting a statistically significant difference in NPi, if a difference exists, would require a sample size of ~6000 patients.

Table 2.

Pupil metrics before and after cerebral angiogram

Variable	Pre angiogram (n=50)	Post angiogram (n=50)	P value*
NPi
Left eye	4.29 (0.42)	4.33 (0.38)	0.612
Right eye	4.25 (0.75)	4.25 (0.77)	0.902
Absolute difference in left and right eye NPi values	0.05 (0.77)	0.08 (0.67)	0.621
Resting pupil diameter
Left eye	3.21 (1.07)	2.19 (0.49)	0.054
Right eye	3.07 (0.98)	2.24 (0.53)	0.699
Latency
Left eye	0.24 (0.05)	0.25 (0.05)	0.262
Right eye	0.25 (0.05)	0.29 (0.19)	0.140
Constriction velocity
Left eye	1.74 (0.90)	1.54 (0.92)	0.113
Right eye	1.58 (0.82)	1.52 (0.80)	0.737
Constricted pupil diameter
Left eye	3.22 (1.07)	2.91 (0.95)	0.028
Right eye	3.07 (0.98)	3.02 (0.94)	0.608
% change in pupil diameter
Left eye	23.21% (9.73)	22.5% (10.85)	0.635
Right eye	22.67% (9.75)	23.31% (10.93)	0.737
Dilation velocity
Left eye	0.72 (0.39)	0.65 (0.41)	0.259
Right eye	0.72 (0.39)	0.64 (0.35)	0.345
PLR categorisation*			0.321
Left eye
Abnormal	1 (2%)	0 (0%)
Normal	49 (98%)	50 (100%)
Right eye
Abnormal	2 (4%)	2 (4%)
Normal	48 (96%)	48 (96%)

*The pupillary light reflex (PLR) is considered abnormal if the NPi value is <3.0.

Discussion

Pupil constriction is largely driven by mechanism such as smooth muscle contraction which is not impacted by normal-dose neuromuscular blockade agents and normal-dose narcotic agents. Based on our own anecdotal experiences, current understanding of PLR physiology, and limited literature, we had expected to find clinically relevant differences in the ipsilateral NPi that would correlate with the underlying pathology or invasive effects of angiography.^16–18 We expected QP to identify post-angiography patients developing either cerebral oedema (as anecdotally reported in contrast-associated encephalopathy^{19 20}) or expansion of intracranial lesions.²¹ Research examining pupil reactivity in the perioperative and sedated patient is inconclusive with some studies suggesting that changes are associated with nociceptive afferent stimulation,^{17 22} and others finding that pupil size change is the most common PLR effect.²² Jolkovsky et al ²³ found that although there were significant differences in pupil size and constriction velocity, compared with healthy controls, intoxicated patients had no difference in NPi values.

The findings must be taken in context with the intent of angiography to diagnose or treat cerebral ischaemia. Delayed neurological improvement represents a phenomenon in which there is an absence of neurological improvement during the immediate period following acute ischaemic stroke treatment.²⁴ Although our study did not collect data on delayed neurological improvement following the angiogram, changes in PLR have repeatedly been found to associated with neurological recovery and decline in the periprocedural patient.^{25 26} Quantitative pupillometers have the advantage of high reliability of a subtle neurological exam finding (changes in PLR) which have been suggested as a biomarker of neurological recovery or decline.^{27 28} PLR recovery after cardiac arrest or global cerebral hypoxaemia predicts recovery, but the temporality of this relationship is not well defined.²⁹ Moreover, because vascular lesions can often be unilateral, the ability to objectively detect trends and differences in left and right PLR may have distinct advantages in understanding acute changes in neurological function.13 15 30 31

The results add to the body of evidence that a well-designed large-scale clinical trial is needed to resolve conflicting reports on the use of pupil reactivity as a biomarker of neurologic injury. Recent evidence demonstrates that NPi is correlated with intracranial pressure change in multiple studies,^32–36 but a recent study found no correlations between changes in pupil metrics and outcomes of cerebral autoregulation in critically ill patients.³⁷ Therefore, there is little consensus on the use of measurement of pupillary function as a technique of non-invasive neuromonitoring follow-up in critically ill patients after undergoing an angiogram. PLR can be associated with an early detection of intracranial disorders, and angiography is one test to identify these maladies.³⁸

The finding of difference in maximum constriction size (despite being only in the left eye, and the difference being ~0.3 mm) is worthy of discussion. Pupil diameter and reactivity depend on the intact and coordinated functioning of the sympathetic and parasympathetic nervous systems, including the hypothalamus, brainstem, and upper cervical ganglions and their output.³⁹ It is not unreasonable that the statistically significant difference simply represents a spurious finding. Constriction diameter was not the primary hypothesis, and controlling for multiple testing would result in failure to reject the null. Another possibility is that the nurses were purposeful in timing their assessments with the lights on, and the brighter lights resulted in more constriction.⁴⁰

Our study provides supportive data that in an uncomplicated angiogram, even with intervention, there is no effect on the PLR. This may be difficult to quantify as this was a pilot study so the number of patients with specific pathologies were small. However, with our preliminary data and an effect size calculation at 0.042, a population of 5959 patients would be required to detect any subtle differences. A previous study has also supported the accuracy and stability of PLR metrics when no neurological change has occurred.⁴¹ By assessing the discrete elements of the PLR before and after angiography, our study provides insights for future research as the values did not statistically.

Limitations

There are several limitations to our study that might affect its generalisability. First, this is a single-centre study with limited cohort of patients that might be different from other centres. Second, a 30-min time window was selected in anticipation that patients were at risk for acute neurological worsening and 24% of the readings were outside the 30-min window due to logistics involved in patient transport. Room turnover times are a common metric for measuring operating room and procedural efficiency and may have impacted the ability of staff to focus on postprocedural neurological assessment.⁴² Third, this is an observational study, and we are limited by our initial patient sampling as they were screened solely for the procedure (ie, angiography) and not for pathology. Our data set did include varying pathologies, such as haemorrhage, stroke and aneurysm, but there were both emergent and elective cases that might undermine the effect of certain pathology if they are not represented in enough numbers. Another limitation is our inability to exclude the effects of sedative medications used intraoperatively that may have a pharmacological effect on the pupils.

Conclusion

These results demonstrate that a routine angiography is not associated with a change in PLR metrics measured with quantitative pupillometry. The study also provides compelling evidence in favour of the feasibility of using pupillometers in the angiography suite. Therefore, any and all postprocedural changes in the PLR exam should be considered indicative of possible new or worsening neurological deficit until ruled out through comprehensive inquiry. These data should serve as baseline for such future and ongoing research in that they suggest there is no need to control for angiography in multivariable analyses where PLR metrics are an independent or dependent variable.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

This study involves human participants but UT Southwestern Institutional Review Board exempted this study. The enrolling university’s IRB deemed this project as exempt. Therefore, a consent form was not required.