Neuroradiological features of contrast-induced neurotoxicity: A systematic review and pooled analysis

Background: Contrast-induced neurotoxicity (CIN), is an increasingly recognised complication of endovascular procedures, presenting as a spectrum of neurological symptoms that mimic ischaemic stroke. The diagnosis of CIN remains a clinical challenge, and stereotypical imaging findings are not established. This study was conducted to characterise the neuroimaging findings in patients with CIN, to raise diagnostic awareness and improve decision making. Methods: We performed a systematic review of PubMed and Embase databases from inception (1946/1947) to June 2023 for reports of CIN following administration of iodinated contrast media. Studies with a final diagnosis of CIN, which provided details of neuroimaging were included. All included cases were pooled and descriptive analysis was conducted. Results: A total of 84 patients were included, with a median age of 64 years. A large proportion of patients had normal imaging (CT 40.8 %, MRI 53.1 %). CT abnormalities included cortical/subarachnoid hyperattenuation (42.1 %), cerebral oedema/sulcal effacement (26.3 %), and loss of grey-white differentiation (7.9 %). Frequently reported MRI abnormalities included brain parenchymal MRI signal change (40.8 %) and cerebral oedema (12.2 %), most commonly observed on FLAIR sequences (26.5 %). Characterisation of imaging findings according to anatomical location and clinical symptoms has been conducted. Conclusions: Neuroimaging is an essential part of the diagnostic workup of CIN. Analysis of the anatomical location and laterality of imaging abnormalities may suggest relationship between radiological features and actual clinical symptoms, although this remains to be confirmed with dedicated study. Radiological abnormalities, particularly CT, appear to be transient and reversible in most patients.


Introduction
In recent decades, the use of iodinated contrast agents has shifted the landscape of diagnostic imaging and interventional procedures.Despite the strides made, all investigations and procedures carry risk of complication, and contrast-induced neurotoxicity (CIN) has emerged as an increasingly recognised adverse event following administration of contrast media.CIN is characterised by a spectrum of neurological symptoms, that mimics typical stroke symptoms, including focal motor deficits, cortical blindness, aphasia, sensory deficit, and reduced consciousness [1].Consequently, CIN remains a diagnostic challenge, and is a poorly understood phenomenon, with a recent survey demonstrated that less than a third of clinicians were comfortable in diagnosing CIN [2].This is particularly concerning in neurointervention, where CIN may mimic ischaemic stroke.Therefore we conducted a systematic review of the literature in order to characterise the reported imaging findings associated with CIN, with the aim of raising diagnostic awareness and enhancing decision making in clinical practice.

Ethical approval
This study was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRSIMA) guidelines [3].Patient consent and ethical approval were not required for this study.

Search strategy
A literature search of Medline (1946 to June 2023) and Embase (1947 to June 2023) was conducted from inception.Key search terms including "contrast", "neurotoxicity", "encephalopathy", "blindness", "deficit", "diagnosis", "radiology", "imaging", "computed tomography", and "magnetic resonance" were used.Boolean operators were utilised to combine search terms as appropriate.Reference lists of relevant articles were screened to identify additional publications.

Eligibility criteria
Studies were selected on the basis of the following criteria: publications reporting (i) patients with a clinical diagnosis of CIN, (ii) with exclusion of other intracranial pathologies, (iii) and reporting of neuroimaging findings on either CT or MRI.Cases of patients <18 years of age, conference abstracts, case series from which individual patient data was not available, and non-English publications were excluded.Reports with clinical or radiological data that made it unclear if CIN was the primary explanation for symptoms (e.g.acute infarct or ischaemia, vasospasm), were also excluded.

Screening process
Two independent investigators (FM & DY) assessed studies for inclusion according to the eligibility criteria.Titles and abstracts were initially screening, followed by full text articles.Where consensus was not achieved, a third investigator (LAS) was consulted.The systematic review tool Covidence (https://www.covidence.org;Veritas Health Innovation, Melbourne, Australia) facilitated the screening process.All articles that met the eligibility criteria underwent data extraction.

Data extraction
Data extraction was performed by two independent investigators (FM & DY) and was crosschecked.In the event of discrepancy, data was further examined, and if consensus was not reached, a third investigator (LAS) was consulted.Baseline patient data including age, sex, country, comorbidities, procedure, indication of procedure, contrast type, contrast volumes administered were collected.The clinical signs and symptoms of CIN, as well as specific CT and MRI findings, and discharge outcomes were extracted.

Quality assessment
Quality assessment was conducted on all included publications using a modified version of the Methodological Quality and Synthesis of Case Series and Case Reports eight-item questionnaire proposed by Murad et al. [4], which involves the domains of selection, ascertainment, causality, and reporting.Reporting items included in the modified tool were the type of contrast administered, if a figure of radiological findings was provided, exclusion of other pathologies, volume of contrast administered, and time course of CIN symptoms.Two independent investigators (FM & DY) independently assessed all articles according to the modified tool.

Statistical analysis
All included cases were pooled.Descriptive analysis was performed for patient demographics, comorbidities, procedural details, clinical symptoms of CIN, radiological findings and discharge outcomes.Fisher's exact and Pearson Chi-square tests were employed where appropriate.All statistical analyses were conducted with Prism 10 (GraphPad Software, San Diego, California).Statistical significance was defined as a p value ≤0.05.

Study selection
Our search strategy returned 706 articles, and after removal of duplicates 548 were screened by title and abstract (Fig. 1).Full-text screening was performed on 162 articles, with 93 excluded with reasons.In total, 69 articles  met the eligibility criteria, and were included for analysis, with a final pooled sample size of 84 patients.

CT findings
In the radiological assessment of CIN, 76 patients underwent CT imaging.When categorised according to the first CT performed follow CIN onset (range: 0.5-48 h), subarachnoid and cortical hyperattenuation secondary to contrast staining was seen in 42.1 % of patients, cerebral oedema or sulcal effacement was observed in 26.3 %, and loss of grey-white differentiation in 7.9 %.Thirty-one (40.8 %) reported no acute CT abnormalities.Unilateral hemispheric contrast staining (11.8 %) and cerebral oedema (11.8 %) were common findings, affecting a widespread mutlilobar region of the cerebral cortex.Bilateral occipital contrast staining was also commonly encountered in 9.2 %.The anatomical distribution of CT findings is summarised in Table 2. Comparison between patients who underwent intracranial and extracranial procedures demonstrated that CIN following cerebral procedures was associated with loss of grey-white differentiation (p=0.038) and cerebral oedema (p<0.001).Moreover, statistical analysis demonstrated that patients with hemiparesis were more likely to have abnormal imaging (p = 0.048).

MRI findings
MRI was performed in 49 patients between 15 min and 11 days following CIN onset, with the majority demonstrating normal MR imaging (53.1 %).Signal change was the most common finding which was reported in 20 patients (40.8 %).Unilateral signal change was observed in 22.4 %, and bilateral signal change in 18.4 %.MRI evidence of cerebral oedema was also seen in 6 patients (12.2.The anatomical distribution of MRI findings is summarised in Table 3. Signal change was most commonly cortical, with 80 % (16/20) of signal change involving the cortex.There was no statistically significant difference in the MRI findings between patients who underwent cerebral and non-cerebral procedures (Table 4).Furthermore, statistical analysis demonstrated there to be no association between MRI abnormality and individual symptoms.
When classified according to sequence, FLAIR hyperintensity was observed in 26.5 % and DWI hyperintensity with normal or raised ADC values was reported in 18.4 % of patients.T2 hyperintensity was described in 10.2 % of patients who underwent MRI, whilst T1 hypointensity was reported in 2 patients (4.1 %).
Unilateral hemiparesis was experienced by 23 patients, with 22 undergoing CT imaging.Hyperattenuation was seen in 54.5 %, cerebral oedema in 45.5 %, and loss of grey-white differentiation in 2 patients (9.1 %).A total of 4 patients (18.2 %) had no acute CT changes.All CT findings were either contralateral or bilateral to the side of hemiparesis.MRI was performed in 16 patients with unilateral hemiparesis.Signal change was observed in 50.0 % of patients consisting of FLAIR hyperintensity (n = 7), DWI hyperintensity with normal or reduced ADC values (m = 7), T2 hyperintensity (n = 5) and T1 hypointensity (n = 1).Cerebral oedema was observed in 31.3 %, whilst 6 patients (37.5 %) were seen to have no acute abnormalities.Signal change was most commonly located contralateral to the side of motor deficit (6/8 patients; 75.0 %).In both MRI and CT studies, the frontal/frontoparietal lobes were frequently specified as the anatomical location of abnormalities.
A total of 24 patients were reported to have reduced consciousness, and 23 underwent CT imaging.Cortical hyperattenuation was reported in 47.8 % and cerebral oedema in 21.7 % of patients.An absence of acute abnormalities was reported in 43.5 % of CT studies.MRI was conducted in 13 patients.Signal change was reported in 38.5 %, unilateral cerebral oedema in 15.4 %, and 53.8 % were observed to have normal MRI.
Aphasia was reported in 21 patients, all of whom underwent CT imaging.Cerebral oedema was observed in 38.1 %, cortical hyperattenuation in 28.6 %, and loss of grey-white differentiation in 19.0 %, with an absence of abnormalities reported in 42.9 %.Apart from 2 patients with bilateral hyperattenuation, all CT changes were left sided.Thirteen patients with aphasia underwent MRI.Signal change was observed in 4 patients (30.8 %) (4, DWI hyperintensity; 2, FLAIR hyperintensity; 1, T2 hyperintensity), and cerebral oedema in 2 (15.4 %).Eight patients (61.5 %) were observed to have normal MRI.All observed MRI abnormalities were left sided.In the 8 patients reported to have generalised seizures, unilateral hyperattenuation was reported in 4 (50 %) patients, and bilateral hyperattenuation in 3 (37.5 %).Cerebral oedema was also noted in 50.0 %.In addition, 2 patients suffered from focal seizures.Both patients were reported to have contralateral hyperattenuation on CT, and one was found to have cerebral oedema contralateral to seizure symptoms.Of patients with generalised seizures, an MRI was performed in four.All patients (100.0 %) demonstrated signal change on MRI including DWI hyperintensity (n = 3), FLAIR hyperintensity (n = 2) and T2 hyperintensity (n = 1).Two patients (50.0 %) were observed to have cerebral oedema on MR imaging.

Serial imaging
In the 45 patients with abnormal findings on initial CT, a total of 26 underwent repeat imaging (range: 6 h-7 days).Nineteen patients (73.1 %) had normal follow up CT studies, and the remaining patients were observed to have residual imaging abnormalities, including hyperattenuation (15.4 %) and oedema (15.4 %).Of 23 patients who were observed to have abnormal MRI, repeat studies were performed for five patients (range: 2 days-17 days).Two patients (40.0 %) were observed to have complete resolution of MR abnormalities, and the remaining 3 patients (60.0 %) had residual cortical signal change, albeit improving.

Follow up imaging
A total of 9 patients underwent clinical follow up with dedicated neuroimaging after hospital discharge (range: 1 week-1 year).MRI was performed for seven patients, CT for one, and both a CT and MRI for one patient.All imaging abnormalities resolved in all patients.

Quality analysis
Of the included 69 publications, 7 (10.1 %) specified a selection criterion and reported the incidence at their institution.The type of contrast was reported by 61 papers (88.4 %), and 64 publications (92.8 %) reported the volume of contrast administered.All 69 articles provided radiological exclusion of other intracranial pathologies.A total of 60 articles (87.0 %) provided at least one figure of CT or MR imaging, and 57 (82.6 %) specified the duration of CIN symptoms.It was deemed that 40 articles (58.0 %) provided sufficient details that were clinically applicable.

Discussion
Clinically, the manifestation of CIN is heterogenous, and includes cortical blindness, confusion, hemiparesis, reduced consciousness,   most commonly in the frontal and occipital lobes.
The findings of our study are reflected in a recent review of CIN following neuroendovascular procedures, which reported cortical and subarachnoid hyperattenuation, and gyriform T2 hyperintensity, as the most common CT and MRI findings respectively [1].Moreover, our review builds on this knowledge by analysing the anatomical distribution of imaging abnormalities.Additionally, our evaluation may suggest a relationship between radiological findings and clinical symptoms.

Relationship between clinical symptoms and radiological features 4.1.1. Cortical blindness
Cortical blindness was the most frequently reported symptom.Almost half of the patients with cortical blindness had normal CT and MRI findings.In patients with acute imaging abnormalities, however, the occipital lobe seemed to be commonly involved.Whether in the context of widespread mutlilobar changes or in isolation, the occipital lobe appeared to be a predominant focus of CT hyperattenutation, MR signal change and cerebral oedema.It has been postulated that the occipital lobe is particularly susceptible to contrast-induced neurotoxicity due to differences in the autonomic control of arteries in the posterior circulation [74][75][76].This may explain the prevalence of cortical blindness in the symptomatology of CIN, as well as the frequency of occipital changes observed in neuroimaging.

Hemiparesis
The findings of our study may indicate a pattern of imaging abnormalities which predominantly appear contralateral to the side of the motor deficits.Moreover, the frontal lobe emerges as the most frequently affected anatomical region in both CT and MRI.These findings support the notion of a strong link between imaging findings and clinical symptoms, with motor weakness frequently observed with frontal abnormalities with likely involvement of the motor cortex.Moreover, statistical analysis demonstrated hemiparesis to be significantly associated with abnormal neuroimaging (specifically CT).The reason for this is unclear, although it may be that patients with hemiparesis experience more extensive manifestations of CIN, leading to the observed correlation between hemiparesis and abnormal CT findings.Further investigation is required to elucidate this relationship.

Aphasia
Aphasia forms a recognised part of the clinical manifestation of CIN.In patients with aphasia in this study, all observed imaging abnormalities were seen to affect the left side.Anatomically, the left frontal and left frontoparietal lobes were the most common location of abnormalities including cortical enhancement, cerebral oedema and signal change on MRI.Although unable to be confirmed in the current study, these radiological findings may be related to Broca's area (Brodmann area 44 and 45) which is typically located in the left frontal lobe [77].Patients with dysfunction of Broca's area would be expected to suffer from an expressive aphasia, and may be related the symptoms exhibited in CIN.

Clinical implications
Although it remains to be confirmed with statistical studies, the observations of this review may provide clinicians with a guide for decision making.Cortical blindness was classically associated with abnormalities in the occipital lobe, hemiparesis was associated with findings in the contralateral frontal lobe, and left-sided frontal changes were commonly seen in patients with aphasia.
An important consideration not be overlooked is the large proportion  of patients with completely normal imaging, which suggests that an absence of imaging abnormalities does not preclude a diagnosis of CIN.As such, the diagnosis of CIN requires thorough neurological examination, and cannot be made on the basis of radiological findings alone.In spite of this, the role of imaging is vital in CIN, due to the close symptomatology it shares with other neurological conditions, most notably ischaemic stroke.We would advocate for all patients with suspected CIN to undergo dedicated neuroimaging to adequately exclude ischaemic stroke, intracranial haemorrhage and other acute neurological pathologieswithout which, a diagnosis of CIN cannot be safely made.
Furthermore, serial imaging, particularly CT, demonstrated a large proportion of abnormalities resolving within a matter of days.This phenomenon of reversibility bears potential clinical significance, particularly when the objective is to comprehensively exclude alternative intracranial pathologies.The extent to which acute CT findings, such as cortical and subarachnoid enhancement or cerebral oedema, might obscure concurrent radiological pathologies remains a point of uncertainty.Nevertheless, if clinical suspicions regarding the presence of other disease processes persists in the days following the initial insult, repeat imaging may hold merit.Such follow-up imaging serves a dual purpose: firstly, to assess any potential resolution of previously observed abnormalities, and secondly, to investigate the potential presence of underlying pathologies that may have eluded detection during the initial imaging assessment.

Pathophysiology
The basic pathophysiological mechanisms of CIN can be broken down into two stages.Firstly, the integrity of the typically impermeable blood-brain barrier (BBB) is compromised.This disruption can be attributed to various factors, including hypertension [77], stroke and the direct effects of contrast agents themselves [78].Secondly, this breach allows the passage of contrast media into neural tissue, leading to neurotoxic sequelae.
The findings of this study indicate a correlation between neuroimaging findings, particularly CT evidence of cerebral oedema and loss og grey-white differentiation, and patient who have undergone intracranial procedures.This finding may be explained by the increased physical manipulation of the cerebrovasculature during such procedures, which potentially induces greater shear stress and cellular damage to the BBB.Furthermore, the direct injection of contrast agents into the cerebrovasculature may lead to increased passage of contrast across the BBB.This may ultimately lead to changes in oncotic pressure gradients, with the efflux of hyperosmolar contrast agents, culminating in cerebral oedema as observed on CT.
In this review, radiological changes were frequently observed in the frontal and occipital lobes.A potential hypothesis for this finding is that these regions typically contain eloquent areas of the cortex, leading to a higher likelihood of symptomatic manifestation when involved.It is also possible that the mechanisms of BBB breakdown and contrast passage occur in other regions of the brain, but remain symptomatically "silent".Ultimately, further study would be beneficial to improve understanding of the clinical symptoms of CIN and its relationship to the underlying pathophysiological processes.

Study strengths
This review has a number of strengths.Firstly, to ensure reliability in our findings, a very strict eligibility criteria was adopted and stringently applied.Cases with insufficient investigation of other pathologies, or in which a diagnosis of CIN was unsure, were ultimately excluded.The patient cohort represents a widespread of demographics, with cases originating from over 20 countries and a spectrum of ages.Additionally, by including cases from a variety of contrast-requiring procedures, the applicability of these findings are extended to clinicians from a broad range of specialties.

Study limitations
There exist several limitations of this study.Due to the heterogeneity in literature, several relationships were unable to be explored, including the association between imaging findings and clinical course, prognosis and outcomes.The majority of included studies were case reports without a specific focus on radiological findings.As such, there was a significant range in the reporting of imaging features, with several reports not specifying location and laterality.Moreover, extraction of data from this heterogenous reporting of imaging proved challenging, and may be oversimplified in our final synthesis.These restrictions prevented meaningful analysis into the association between symptom severity and factors such as contrast dosage, which may have provided applicable clinical insights.In addition, although comparisons between imaging findings and clinical symptoms were explored, they are merely observations of patterns in the reported literature, and do not represent statistical correlation.

Gaps in knowledge and future directions
Future investigation into the diagnostic features of CIN is certainly warranted.Firstly, rigorous quantitative studies are warranted to confirm the observed relationships between clinical symptoms and radiological features.Additionally, further research should focus on standardising the reporting of imaging features in order to improve data consistency and facilitate more comprehensive analyses.Reporting of imaging abnormalities (or lack thereof) from a range of imaging modalities and sequences, will help improve clinical applicability of findings.Serial imaging at predefined time points may provide insight into the appearance and subsequent resolution of abnormalities.Beyond neuroimaging, investigation into other diagnostic factors would provide substantial benefit.Very few studies reported CSF findings, biochemistry and other basic investigations.Ultimately, further study would provide a more robust foundation for clinicians to rely on in the diagnostic workup of a patient with suspected CIN.

Conclusion
In CIN patients, a large proportion of CT and MRI studies were normal.In patients with abnormalities, however, subarachnoid/cortical hyperattenuation on CT, signal change on MRI, and cerebral oedema were the most common findings.The anatomical location and laterality of abnormalities appear to have some relationship with actual clinical symptoms, although this remains to be confirmed with dedicated study.The majority of imaging abnormalities appear to be temporary and reversible.This review ultimately demonstrates the need for ongoing investigation into the radiological features of CIN, and their implications on presentation, clinical course and outcomes.

Funding statement
None (not applicable).

Table 1
Clinical Characteristics.

Table 2
Radiological Characteristics described on CT imaging.
*Based on first CT performed following CIN onset (range: 0.5-48 h).F.P.Mariajoseph et al.aphasia and seizures.The most common CT findings included cortical/ subarachnoid hyperattenuation and cerebral oedema.The most common MRI abnormality was cortical signal change, which was demonstrated on a range of MR sequences.A large proportion of imaging did not show acute abnormalities and all patients who underwent postdischarge follow up imaging were noted to have no abnormalities.The anatomical location of imaging changes in patients with abnormalities was relatively widespread, involving all anatomical lobes, but appearing

Table 3
Radiological Characteristics described on MR imaging.

Table 4
Clinical and radiological findings in intracranial vs extracranial procedures.

Table 5
CT findings according to clinical symptoms.
(continued on next page)

Table 6
MRI findings according to clinical symptoms.