The relation of optic nerve sheath diameter (ONSD) and intracranial pressure (ICP) in pediatric neurosurgery practice - Part I: Correlations, age-dependency and cut-off values
- 295 Downloads
It is assumed that the width of the optic nerve sheath diameter (ONSD) is dependent on intracranial pressure (ICP) and pulsatility and thus constitutes a non-invasively accessible “window” for qualitative assessment of ICP. Data on the correlation to invasively measured ICP in children are scarce and have often been obtained from sedated patients in intensive care unit (ICU) or intraoperatively. We report on a mixed cohort of pediatric neurosurgical patients, ICP and ONSD measurements were available from both sedated and awake children, only a minority from ICU patients.
Seventy-two children were investigated. Ultrasound ONSD determination was performed immediately prior to invasive ICP measurement and the mean binocular ONSD was compared with ICP. The investigations were performed in children awake, sedated, or under general anesthesia.
In the entire patient cohort, the correlation between ONSD and ICP was good (r = 0.52, p < 0.01). Children > 1 year revealed a better correlation (r = 0.63; p < 0.01) and those ≤ 1 year did worse (r = 0.21). Infants with open fontanelle had no correlation. In the entire cohort, the best ONSD cut-off value for detecting ICP ≥ 15 and ≥ 20 mmHg was 5.28 and 5.57 mm (OR 22.5 and 7.2, AUC 0.782 and 0.733).
Transorbital ultrasound measurement of ONSD is a reliable non-invasive technique to assess increased ICP in children in every clinical situation; however, the impact of age and fontanelle status needs to be considered. ONSD thresholds enable qualitative first orientation regarding ICP categories with a very satisfying diagnostic accuracy.
KeywordsOptic nerve sheath diameter Non-invasive ICP assessment Ultrasound-awake investigation
In pediatric neurosurgery, much pathologies are associated with increased intracranial pressure (ICP). For initial diagnosis and follow-up evaluations of many underlying pathologies, repeated invasive ICP measurements are desirable and sometimes CT/MRI imaging is used as surrogates to indirectly assess ICP measurement and imaging; however, it is coming along with repeated risks of anesthesia, infection, and radiation in children. Thus, a reliable, clinical useful technique to non-invasively assess the ICP status in children is needed. Transorbital ultrasound measurement of the optic nerve sheath diameter (ONSD) is a promising method, since the optic nerve sheath (ONS) constitutes a diagnostic window into the intracranial compartment [1, 2].
The optic nerve (ON) is part of the central nervous system as it originates from the diencephalon during embryogenesis . The ON is surrounded by the optic nerve sheath, covering the optic subarachnoid space that is directly linked to the intracranial subarachnoid space . Changes of intracranial pressure are thus straightly transferred to the ONS [5, 6]. Under normal ICP conditions, there is only less cerebrospinal fluid (CSF) around the ON, whereas an ICP increase leads to a measurable swelling of the ONS due to CSF shifts into the expandable ONS .
In the current literature, studies correlating ultrasound-based ONSD measurement with invasively measured ICP were mainly performed in adult patients [8, 9, 10, 11]. Data that compare ultrasound ONSD with ICP values in children are sparse with merely small numbers of patients  or were obtained in children under general anesthesia or intraoperatively [13, 14] and therefore represent a limited set of situations presenting in pediatric neurosurgical and neurological pathologies.
This study investigates the correlation of ONSD and invasively measured ICP in a mixed cohort of pediatric patients awake, sedated, and anesthetized and aims to establish ONSD cut-off values, considering the influence of age and patency of anterior fontanelle.
Ultrasound of the ONSD is part of daily routine diagnostic workup in all children presenting with suspicion of ICP increase. All patients between newborn and 18 years, treated at the University Children’s Hospital of Tuebingen by Pediatric Neurosurgery or Pediatric Neurology between January 2016 and January 2018, were enrolled into the study if they underwent a diagnostic or surgical procedure including invasive ICP measurement.
The study was approved by the institutional ethics committee (project number: 180/2018B02).
The overall cohort included 72 patients with a mean age of 5.2 years. Fifty-one children were > 1 year, 21 ≤ 1 year. Ten patients presented with an open anterior fontanelle. Fifty children were male (69.4%), 22 were female (30.6%).
Diagnoses encompassed hydrocephalus (41.7%), craniosynostosis (22.3%), pseudotumor cerebri (PTC) (13.8%), traumatic brain injury (TBI) (9.7%), tumor (5.6%), and other intracranial pathologies with suspected raised ICP (6.9%), such as cysts or subdural hematoma.
Forty percent of the patients were investigated intraoperatively under general anesthesia, 39% were awake, 14% were analog-sedated with propofol or midazolam for lumbar or reservoir puncture, and 7% were somnolent or comatose on the intensive care unit. Ketamin, which is known to possibly increase ICP in non-ventilated patients, was not used for sedation of study patients.
Ultrasound ONSD measurement
Invasive ICP measurement
ICP measurement was performed via intraparenchymal ICP probe (29%), closed extraventricular drainage (EVD) (7%), or puncture of the shunt reservoir (15.3%), the lumbar CSF space (8.4%) in children awake or analgo-sedated. During surgery, ICP was measured using a ventricular catheter or brain needle with connected manometer line (25%) or epidural probe (15.3%).
Data were tested for normal distribution using the Kolmogorov-Smirnov or Shapiro-Wilk test. Parametric data were reported as means and standard deviation (sd). The analyses were done using SPSS (PASW Statistics 18, IBM) statistical software. Depending on normality of distribution, the correlation of the variables was tested using Pearson’s or Spearman’s correlation coefficient. Receiver operating characteristic (ROC) curves were created to define ONSD cuf-off values with the best diagnostic accuracy for detecting ICP thresholds. Bland-Altman’s analysis was used to test agreement of ONSD between both eyes using mean differences and limits of agreement. Statistical significance was set at p < 0.05. The independent Student’s t test was used for comparing mean values.
Comparison of ultrasound ONDS values between left and right eye
For assessment of correlation and bias of ONSD between both eyes, mean ONSD of the left eye was compared with mean ONSD of the right eye of each patient (n = 72). Mean ONSD values of each eye were created out of at least three measurements. Mean ONSD in the left eye was 5.56 ± 0.85 mm and in the right eye 5.55 ± 0.89 mm (mean ± sd). The correlation of ONSD values between both eyes was extremely high (r = 0.93, p < 0.01). To test for a bias of left- and right-sided ONSD in the sense of a systematic difference, Bland-Altman analysis was performed. Mean bias between left- and right-sided ONSD was 0.016 ± 0.33 mm in favor of the left eye. Ninety-six percent of the values were within the limits of agreement. Two of three outliers were children with premature synostosis of the sagittal suture and one child presented with right-sided ventricular enlargement due to shunt failure, corresponding with right-sided ONSD enlargement.
Correlation of ultrasound-based ONSD and invasively measured ICP over the entire cohort
Mean ONSD and ICP values for different pediatric neurosurgical and neurological pathologies
Mean ONSD in mm ± sd
Mean ICP in mmHg ± sd
5.3 ± 0.9
12.4 ± 6
5.2 ± 0.4
13.1 ± 10.7
5.3 ± 0.9
15.3 ± 5.8
6.3 ± 0.4
22.3 ± 7.1
5.7 ± 0.8
12.3 ± 6.7
5.8 ± 1.1
9.9 ± 4.2
The overall cohort included 72 patients, 51 of whom were > 1 year and 21 were ≤ 1 year. In the entire cohort, median age was 5.2 years.
Over the entire cohort, ONSD values revealed a good correlation to invasively measured ICP values (r = 0.52, p < 0.01).
Effect of age and status of anterior fontanelle on ONSD-ICP correlation
Fifty-one children investigated in this study, were > 1 year (median: 6 years). 66.67% were male, 33.33% female. None of them had an open anterior fontanelle.
Fifty-five percent of these patients were investigated under awake conditions, 19.6% were sedated with propofol for reservoir or lumbar puncture, 17.7% were measured intraoperatively, and 7.7% were comatose in the ICU.
Twenty-one children included in this study were ≤ 1 year (age range newborn to 12 months, median; 6 months). 76.2% were male, 23.8% female. Twenty patients were investigated intraoperatively under general anesthesia and 1 child was in a vegetative state. Ten patients had an open anterior fontanelle (AF).
The correlation between ONSD and ICP was poor for patients ≤ 1 year (r = 0.21). After stratification of this cohort into subgroups regarding open AF, patients ≤ 1 year with open AF (n = 10, median age 4 months) revealed no correlation (r = 0.057), those with closed AF (n = 11, median age 6 months) correlated better (r = 0.4). Considering that, an open AF seems to mainly compromise ONSD-ICP correlation.
Diagnostic accuracy of ONSD cut-off values in entire cohort
For application of ultrasound ONSD determination in the daily routine care of children, it is essential to know which ONSD values indicate ICP levels. To address this issue, ONSD cut-off values at different ICP thresholds were calculated for the entire cohort.
ONSD cut-off value for detecting ICP ≥ 10 mmHg
Nineteen children had an ICP < 10 mmHg, with a mean ICP of 6.17 ± 2.39 mmHg (sd) and mean ONSD of 4.77 ± 0.65 mm (sd). The group of patients with ICP ≥ 10 mmHg included 53 individuals with a mean ICP of 17.2 ± 6.2 mmHg and a mean ONSD of 5.7 ± 0.81 mm. Both ONSD and ICP values were significantly higher in the group with ICP ≥ 10 mmHg (p < 0.001). Detailed values for ONSD and ICP over the entire cohort are summarized in supplementary material (Table 1.1.)
The best ONSD value for detecting ICP ≥ 10 mmHg was 5.24 mm (sensitivity 79.2%, specificity 84.2%, PPV 93.3%, NPV 59.3%, AUROC 0.835, OR 20.4).
ONSD cut-off values for entire patient cohort, patients ≤ and > 1 year
ICP value (in mmhg)
ONSD cut-off value (in mm)
> 1 year
≤ 1 year
Influence of age on validity of ONSD cut-off values
As correlation of ONSD and ICP was strongly impaired by age and especially open AF, it is assumed that age will also affect validity of ONSD cut-off values at different ICP thresholds.
To evaluate this issue, ONSD cut-off values at ICP thresholds 5, 10, 15, 20, 25, and 30 mmHg were calculated for patients > and ≤ 1 year.
Fifty-one individuals of our cohort were > 1 year; all of these had a closed anterior fontanelle. Distribution of the patients to different ICP groups, ONSD and ICP mean values as well as differences of ONSD and ICP values regarding ICP groups are summarized in the supplementary material (Table 1.2).
Twenty-one of the included patients were ≤ 1 year, 10 of whom had an open AF and none of them was investigated awake. The lowest ICP value in this group was at 5 mmHg and none of these patients had an ICP > 30 mmHg. ONSD cut-off values at ICP thresholds 10, 15, 20, and 25 mmHg were 4.65, 4.99, 4.99, and 5.74 mm, respectively. Distribution of the patients to different ICP groups, ONSD and ICP mean values as well as differences of ONSD and ICP values regarding ICP groups are summarized in the supplementary material (Table 1.3).
ONSD cut-off values and statistic validity for children ≤ 1 year are summarized in Table 2 but all were worse than in the entire cohort and much worse than in children > 1year of age.
This study reports on the second largest pediatric cohort in the current literature where transorbital ultrasound–measured ONSD values are compared with invasively measured ICP values. It is moreover the first ONSD/ICP study that was performed also in children under awake conditions and in a mixed cohort of neurosurgical and pediatric neurological pathologies. Additionally, it is the only study where invasive ICP measurement was performed by variable approaches and where ONSD cut-off values were defined for ICP thresholds above 20 mmHg.
ONSD does not show a significant unilateral accentuation in entire cohort
As tilting of the ultrasound probe could cause an incomplete representation of the ONS and thus lead to inaccurate measurements, three measurements of each eye were performed and a mean right and left ONSD was created, from which the mean ONSD value of both eyes resulted.
Both systematic measurement errors and anatomical variations are possible reasons for unilateral ONSD deviations [1, 4]. Cases of unilateral papilledema and ONSD dilatation under conditions of general ICP increase are rare and in the current literature mainly described in patients with idiopathic intracranial hypertension [15, 16] or unilateral occlusion of a venous sinus .
In our cohort, we found three patients with asymmetric ONSD values. Two children, aged 6 months, were investigated during surgery for sagittal synostosis. The ONSD difference was 0.9 in favor of the right eye (mean ICP 20 mmHg) in one and 0.8 mm in favor of the left eye (mean ICP 16.7 mmHg) in the other. The third patient, 1.6 years old, presented with shunt failure due to occlusion of the right-sided ventricular catheter with ONSD difference of 0.9 mm in favor of the right eye (mean ICP 16.2 mmHg).
All the other patients presented with a minimal difference of ONSD values. The correlation between left- and right-sided ONSD was outstanding with r = 0.93, p < 0.01 and well in agreement with the recent literature . The mean ONSD difference between both eyes was minimal (0.016 mm), excluding a systematical unilateral measurement error.
In cases of unilateral ONSD dilatation and optic disc swelling at ICP increase, the underlying pathophysiological background is not yet completely understood. Asymmetrical intraocular pressures , anatomical anomalies of the ONS, and decreased ONS elasticity [1, 16, 19] as well as an optic compartment syndrome  are possible reasons under discussion.
ONSD and ICP correlate well independently of wakefulness and mode of ICP measurement
Several previous studies in adult [10, 11, 20, 21, 22, 23] and pediatric patients [12, 13] revealed a good correlation of ultrasound-based ONSD and invasive measured ICP but especially data obtained from children are scarce and mainly performed under general anesthesia  or with representation of a limited cohort of pathologies . Furthermore, methods of ICP measurement were limited to ICP probe, lumbar puncture, or presence of ICP increase and were qualitatively assessed (guessed) based on CT imaging [25, 26] or clinical signs of raised ICP . In our study, about half of the patients received ONSD and ICP measurements under awake conditions and ICP measurements were performed in various ways. Thus, the study reflects clinical reality.
Overall ONSD-ICP correlation was reasonable (r = 0.52, p < 0.01) and well comparable to the latest similar study in sedated ICU patients, all measured with an intracranial probe , although a majority of our cohort was awake and ICP measurement was performed by various means. Additionally, ONSD measurement was performed in an axial plane with a large linear transducer in our study, compared with sagittal investigation using a hockey stick in the comparative study.
This shows that ONSD measurement per se is robust and the correlation to ICP does not seem to be strongly influenced by the way of ICP determination.
The correlation of ONSD and ICP is mainly affected by existence of open fontanelle
When allocating our cohort into subgroups regarding age and open anterior fontanelle, patients > 1 year revealed the best ONSD-ICP correlation (r = 0.63, p < 0.01). In patients ≤ 1 year, children with open AF had no correlation to ICP (r = 0.057), whereas in patients ≤ 1 year and with closed AF, an acceptable correlation (r = 0.4) was found. Another recent study also showed a poorer correlation in patients with open AF .
Our results strongly suggest that in children with open AF, the relationship between ONSD and ICP is not linear, due to a very compliant skull in cases of ICP increase. After fontanelle closure, there is still a more compliant skull bone and more mobile sutures, which is the most likely reason for poorer ONSD-ICP correlation in children ≤ 1 year with closed AF compared with older patients .
ONSD cut-off values are age-dependent and low ONSD values can reliably exclude increased ICP
According to the literature, ONSD values < 4 mm in children ≤ 1 year and values < 4.5 mm in children > 1 year are assumed as normal , values above 5 mm in children > 4 years are assumed as abnormal .
Evaluating the entire cohort of this study, ONSD cut-off values for ICP thresholds 5, 10, 15, 20, 25, and 30 mmHg were 5.24, 5.24, 5.28, 5.57, 5.75, and 5.91 mm. For children > 1 year, slightly increased values were found (Table 2). Similar values were described in the previous literature [13, 14] and also in a recent study, where MRI-based ONSD was compared with ICP in children with craniosynostosis, a significant age-dependency of ONSD values was shown .
In patients > 1 year, the best ONSD cut-off value for detecting ICP values ≥ 10 mmHg, what can be supposed as the limit to pathology, was 5.28 mm, with a sensitivity of 92.1%, specificity of 100%, and AUROC of 0.981. This borderline cut-off ONSD was also found in the entire cohort, marking ICP values between 10 and 15 mmHg. In patients >1 year, diagnostic accuracy of ONSD cut-off values was best, followed by values defined for entire cohort. According to the reduced correlation of ONSD and ICP in the subgroup of patients ≤ 1 year, the validity of ONSD cut-off values was also strongly reduced in this age group (Table 2). These results once again indicate that the relationship of ONSD and ICP is not strictly linear and affected by other factors, one of which is age, or rather age-dependent skull compliance.
When analyzing statistic accuracy of different ONSD cut-off values, it can be shown that higher ONSD values come along with higher negative predictive values. This phenomenon manifests for entire cohort independently of age (Table 2) and allows important conclusions regarding the application of ONSD in the daily routine. For the entire cohort, ONSD values < 5.28 mm can thus exclude ICP ≥ 15 mmHg at about 90%, in children > 1 year, and ONSD values < 5.75 can exclude ICP ≥ 20 mmHg at almost 100%. In contrast to that, positive predictive values are high for low ONSD values and become poorer with higher ONSD values, implying that high ONSD values can be associated with low ICP. In cases of former prolonged episodes of ICP increase, a proposed theory of hysteresis  possibly due to compartmentation of the ONS  or due to stiffness of the ONS as a result of ICP increase  might be the underlying mechanism of persisting ONSD increase.
Moreover, when comparing ICP and ONSD values according to distinctive pathologies, it was obvious that patients with pseudotumor cerebri presented the highest ONSD and ICP values whereas children with hydrocephalus, TBI, and synostosis had almost similar and lower values. Tumor patients and children with other intracranial pathologies such as arachnoid cysts had similar ICP levels compared with patients with hydrocephalus and TBI, but considerable larger ONSD values (Table 1). Padayachy et al. investigated the same number of synostosis-patients but found higher ONSD and ICP values for this cohort . An earlier study with a smaller number of patients described lower ONSD values, similar to ours .
These results suggest that pathology itself might affect width of ONS independently of ICP and that effects on ONSD are not linear.
We, furthermore, divided patients in subgroups regarding different ICP values (e.g., patients with ICP ≥ 10 mmHg) and compared the mean differences of ONSD and ICP (Delta-ONSD and Delta-ICP) among different ICP subgroups (see supplementary material Tables 1.1.–1.3.). It was remarkable that Delta-ICP values between different subgroups were more separated with higher significance compared with Delta-ONSD for entire cohort independently of age. This finding is also arguing against an absolute linear relationship between ONSD and ICP.
In summary, ONSD shows a good correlation to invasive measured ICP and age-dependent ONSD cut-off values can be used for first-line estimation of ICP in the daily routine care of children to indicate or avoid further diagnostic procedures. However, validity and reliability of a general ONSD-ICP correlation and its use in clinical practice have to be critically discussed, especially in view of the fact that width of ONS is not only affected by ICP, but probably also by several other factors. In our opinion, it is therefore essential to analyze intra-individual ONSD-ICP correlations as well as dynamic changes of ONSD and ICP within therapy. It is furthermore necessary to investigate if wakefulness and type of ICP measurement impair ONSD-ICP correlations. These questions have been addressed in the second part of this study.
We are deeply indebted to Juergen Beck, MD PhD, University of Bern and Freiburg, who inspired our interest in ultrasound determination of ONSD and to Llewellyn Padayachy, MD PhD, University of Cape Town and Pretoria, for many fruitful discussions over the years.
Compliance with ethical standards
Conflict of interests
The authors have no potential conflict of interests to declare.
- 13.Padayachy LC, Padayachy V, Galal U, Gray R, Fieggen AG (2016) The relationship between transorbital ultrasound measurement of the optic nerve sheath diameter (ONSD) and invasively measured ICP in children : part I: repeatability, observer variability and general analysis. Child Nerv Syst 32:1769–1778CrossRefGoogle Scholar
- 14.Padayachy LC, Padayachy V, Galal U, Pollock T, Fieggen AG (2016) The relationship between transorbital ultrasound measurement of the optic nerve sheath diameter (ONSD) and invasively measured ICP in children. : Part II: age-related ONSD cut-off values and patency of the anterior fontanelle. Child Nerv Syst 32:1779–1785CrossRefGoogle Scholar
- 24.Ozturk Z, Atalay T, Arhan E, Aydin K, Serdaroglu A, Hirfanoglu T, Havali C, Akbas Y, Yalinbas D (2017) The efficacy of orbital ultrasonography and magnetic resonance imaging findings with direct measurement of intracranial pressure in distinguishing papilledema from pseudopapilledema. Child Nerv Syst 33:1501–1507CrossRefGoogle Scholar