ORIGINAL ARTICLE

The Oxygen Reactivity Index and Its Relation to SensorTechnology in Patients with Severe Brain Lesions

Julius Dengler • Christin Frenzel • Peter Vajkoczy •

Peter Horn • Stefan Wolf

Published online: 7 March 2012

� Springer Science+Business Media, LLC 2012

Abstract

Background The oxygen reactivity index (ORx) has been

introduced to assess the status of cerebral autoregulation

after traumatic brain injury (TBI) or subarachnoid hemor-

rhage (SAH). Currently, there is some controversy about

whether the ORx depends on the type of PbrO2-sensor

technology used for its calculation. To examine if the probe

technology does matter, we compared the ORx and the

resulting optimal cerebral perfusion pressures (CPPopt) of

simultaneously implanted Licox (CC1.SB, Integra Neuro-

science, France) and Neurovent-PTO (Raumedic,

Germany) probes in patients after aneurysmal SAH or

severe TBI.

Methods Licox and Raumedic probes were implanted

side by side in 11 patients after TBI or SAH. ORx and

CPPopt were recorded continuously. The equivalence of

both probes was examined using Bland–Altman analyses.

Results The mean difference in ORx was 0.1, with Licox

producing higher values. The limits of agreement regarding

ORx ranged from -0.6 to +0.7. When both probes’ ORx

values were compared in each patient, no specific pattern in

their relationship was seen. The mean difference in CPPopt

was 0 mmHg with limits of agreement between -16.5 and

+16.4 mmHg.

Conclusions Owing to the rather limited number of

patients, we view the results of this study as preliminary.

The main result is that Licox and Raumedic showed con-

sistent differences in ORx and CPPopt. Therefore, ORx

values of both probes cannot be interchanged and should

not be viewed as equivalent. This should be taken into

consideration when discussing ORx data generated by

different PbrO2 probe types.

Keywords ORx � Licox � Raumedic �Cerebral tissue oxygenation � PbrO2 �Cerebral autoregulation

Introduction

Impaired cerebral autoregulation (CA) is believed to be the

main reason for the poor prognosis of patients suffering

from severe traumatic brain injury (TBI) or aneurysmal

subarachnoid hemorrhage (SAH). Minimal-invasive neur-

omonitoring techniques seem to improve outcome after

severe cerebral insult and are therefore also used to quan-

tify CA, for which a number of different indices were

introduced. The pressure reactivity index (PRx), which

correlates the intracranial pressure (ICP) to the mean

arterial pressure, was shown to be of prognostic value in

patients with TBI [1, 2]. Over a course of time, alternative

indices have been introduced. One of these is the oxygen

reactivity index (ORx), which reflects the correlation

between spontaneous changes in the partial pressure of

brain tissue oxygen (PbrO2) and changes in cerebral per-

fusion pressure (CPP) [3]. The ORx is based on the

assumption that changes in cerebral blood flow, which

occur during disturbed CA, directly affect PbrO2. It ranges

between +1 and -1, with negative values representing

good CA and positive values impaired CA.

Recently, there has been some controversy about the

role of ORx in the measurement of CA, especially when

comparing it to simultaneous measurements of the PRx

[4–6]. While Jaeger et al. were able to show that ORx and

PRx correlate, Radolovic et al. could not confirm these

J. Dengler (&) � C. Frenzel � P. Vajkoczy � P. Horn � S. Wolf

Department of Neurosurgery, Charite, Universitaetsmedizin

Berlin, CVK, Augustenburger Platz 1, 13553 Berlin, Germany

e-mail: julius.dengler@charite.de

123

Neurocrit Care (2013) 19:74–78

DOI 10.1007/s12028-012-9686-0

findings. One valuable point in the ensuing discussion of

this conflict is that both groups used different PbrO2

probes, each relying on a different measuring technique:

the Licox probe (CC1.SB, Integra Neuroscience, France),

used by Jaeger et al., measures an electrical current that is

created by oxygen diffusion through an electrolyte cham-

ber. In the Neurotrend probe (Codman, Johnson & Johnson,

Raynham, USA), used by Radolovich et al., oxygen dif-

fuses into a Ruthenium dye and causes measurable changes

in its fluorescence intensity. Both these author groups agree

that these principal differences might have translated into

the observed differences in ORx. It was also suggested that

not only in the case of Neurotrend, but also in that of the

recently introduced Neurovent-PTO probe (Raumedic,

Germany) care should be taken when comparing its ORx

values to those of the Licox probe.

To examine whether the probe type does matter when

discussing ORx, we evaluated simultaneous ORx values

from a patient cohort of a previously published prospective

trial in which two minimal-invasive PbrO2 probes (Licox

and Raumedic) were implanted side by side in patients

after SAH or TBI [7]. We now aimed to evaluate whether

ORx values for both probes differ, given the fact that ORx

is a relative value that does not so much depend on the

absolute value of PbrO2 but rather on its changes in relation

to CPP. Since ORx-guided therapy intends to optimize the

CPP to a level where intact CA is most likely (CPPopt), we

also aimed to examine whether a CPPopt could be deducted

from the ORx measurements and whether this CPPopt

would be the same for both probes.

Methods

The study was approved by the ethics committee of the

Charite, Berlin and was conducted between March 2009

and May 2010.

The trial’s methods and study protocol were described

previously [7]. In brief, all patients either suffered from

severe TBI (GCS <9) or high-grade SAH (WFNS >4)

and were treated with analgesia, sedation, and mechanical

ventilation. Each patient received an external ventricular

drainage.

Multimodal monitoring began on day 1 of severe brain

trauma or directly after either surgical or endovascular

aneurysm treatment. It included continuous monitoring of

the mean arterial pressure (MAP), ICP, and CPP. Also,

PbrO2 was measured using two probe types. First, a Licox

probe was positioned into the more traumatized hemi-

sphere in TBI patients or on the side of the aneurysm in

SAH patients at 12 cm dorsal to the nasion and 6 cm lateral

to the midline (to reach the MCA territory) and at a depth

of 25 mm. Next, a Raumedic probe was placed within

20 mm from the Licox probe using a separate burr hole.

Correct probe positioning was checked by CT scan. A

separate part of the study compared both PbrO2 probes

during dynamic phases of monitoring, for which two types

of interventions were conducted about once per day. One

was a hyperoxygenation challenge, during which the FiO2

was increased by 20 percent for 10 min. The other inter-

vention was a hypertensive challenge with an increase in

MAP by 20 mmHg for 10 min using intravenous titration

of norepinephrine.

Data Analysis

All data–including phases of intervention–were recorded

continuously with dedicated software (ICMplus, University

of Cambridge, UK). Data pooling and evaluation were

carried out using the statistical environment R 2.13.1 [8].

To calculate the ORx, the moving correlation coefficients

of CPP and PbrO2 were continuously averaged over 1 h. To

determine CPPopt, we averaged the ORx over 24 h for CPP

in steps of 5 mmHg, analogous to similar calculations in

relation to the PRx [9]. To exclude possible effects of

differences in the probes’ run-in time, all the measurements

of day 1 were discarded. The limits of agreement for ORx

and CPPopt of both probes over the entire monitoring per-

iod were visualized and calculated using the approach

suggested by Bland and Altman [10]. Repeated measure-

ments per individual were taken into consideration with

random effects, as proposed [11, 12].

Results

We included 11 patients (7 males/4 females, mean age

53 ± 8 years) in this study. Five patients suffered from

aneurysmal SAH with a WFNS score of 5 and a Fisher

Grade 3 in three cases and Grade 4 in two cases. Four

patients underwent surgical clipping while 1 patient

received coil embolization. The other 6 patients suffered

from TBI with initial GCS values of 3 in four cases, a GCS

of 6 in one case and a GCS of 8 in one case. No surgical

intervention was conducted on the TBI patients. The mean

distance between both probes’ tips, verified by CT scan,

was 13 ± 5 mm. Mean time of PbrO2 monitoring per

patient was 7.2 ± 2.4 days. The ORx values were calcu-

lated continuously from both probes’ PbrO2 measurements.

Mean ORx differed by 0.1 with Licox producing higher

values. The Bland–Altman analysis resulted in limits of

agreement between -0.6 and +0.7 (Fig. 1), which com-

prises the majority of all the possible ORx values (range:

-1 to +1). When both probes’ ORx values were compared

within each group of pathology (SAH and TBI) and in each

Neurocrit Care (2013) 19:74–78 75

123

single patient, no specific pattern in the relationship of both

probes could be observed.

CPPopt was calculable in 77% of cases for Licox, and in

66% for Raumedic. Figure 2 shows the Bland–Altman

analysis for both probes regarding CPPopt. Mean difference

in CPPopt was 0 mmHg with limits of agreement ranging

between -16.5 and +16.4 mmHg.

Figure 3 gives examples of our observations regarding

CPPopt. Even though in some patients there were days at

which we observed rather good agreement (Fig. 3a), we

were not able to establish a CPPopt in other cases (Fig. 3b).

Discussion

The main message of this study is that Licox and Raumedic

did not produce agreeing values for ORx or CPPopt in our

patient cohort. The disagreement between both probes was

so consistent and so large that it does not in any way support

the view that both probes can be interchanged or can be

viewed as equivalently measuring ORx or CPPopt [13].

The question regarding the comparability of two dif-

ferent PbrO2 probe types was recently raised when Jaeger

et al. [4, 6] showed that ORx correlates with PRx and

therefore allows for an assessment of CA while Radolovic

et al. [5] observed the opposite. When commenting on

potential reasons for this conflict, Radolovic et al. point out

Fig. 1 Bland–Altman plot comparing ORx values of Licox and

Raumedic. For each of the 102,435 data points the average value of

both probes is plotted against their mean difference. Limits of

agreement range from -0.6 to +0.7. A Random Effect model was

used to account for repeated measurements per individual

Fig. 2 Bland–Altman plot comparing CPPopt of Licox and Raumed-

ic. For each of the 55 data points, the average values of both probes

are plotted against their mean differences. Limits of agreement range

from -16.5 to +16.4 mmHg. A Random Effect model was used to

account for repeated measurements per individual

Fig. 3 CPPopt is defined as the CPP level where ORx shows a

minimum from all ORx values of a given day. The example in

a shows CPPopt for Raumedic at 75 mmHg, while CPPopt for Licox

would be 80 mmHg. In b the CPPopt for Raumedic is 90 mmHg,

while ORx values for Licox do not show a detectable minimum;

therefore, no CPPopt could be determined. Dashed and dotted lines

show means and 95% CIs of spline functions, respectively, for each

probes’ ORx values, used for mathematical approximation of CPPopt

76 Neurocrit Care (2013) 19:74–78

123

that both studies examined different pathologies (exclu-

sively TBI vs. exclusively SAH) with probes in different

locations (only right hemisphere vs. hemisphere of

pathology). Jaeger adds that the relation between ORx and

PRx might be more complicated than expected, since PRx

depends on changes in blood volume, while ORx depends on

changes in blood flow. In the discussion, both groups agree

on two further valid points: 1) There was a significant dis-

parity in ORx monitoring time that was evaluated: while

Jaeger et al. examined a mean period of 7.4 days, Radolovic

et al. present data on the first 24 h of neuromonitoring. Jaeger

states that PbrO2 probes need a couple of hours of run-in

time, which consequently has an impact on data on day 1 but

not on other days. 2) The specific measurement principles of

both probes might account for differences in read-out: to

quantify oxygen diffusion, the Licox probe, which was used

by Jaeger et al., measures an electrical current while the

Neurotrend probe, used by Radolovich et al., measures

fluorescent activity.

The results of the study which we presented here support

the view that this last point might be crucial. Earlier in vitro

studies have shown that—in the case of Licox versus

Neurotrend—reaction times differ significantly [14, 15].

Interestingly, in the case of Licox versus Raumedic, which

is the probe we compared to Licox, in vitro data suggest

equivalence of both in terms of absolute values and in

reaction times [16]. However, in vivo data on Licox versus

Raumedic show otherwise: while a porcine model rendered

partial equivalence with differences in reaction times [17];

two studies in humans show limits of agreement too large

to declare both probes as equivalent, with one of these

studies showing significant differences in reaction times

[7, 18].

So far, the only available comparative in vivo data on

ORx examined Licox versus Neurotrend in six patients

after SAH or TBI: ORx measured by Licox was signifi-

cantly higher than in Neurotrend [5]. Our study is the first

to directly compare ORx measured by Licox and Raumedic

and reaches a much similar result: Licox produces higher

mean ORx values.

Thanks to the ongoing debate about ORx we feel that we

were able to avoid some potential pitfalls to strengthen our

comparative study.

First of all, we present simultaneous data of both probes in

the same patient. Furthermore, interventional challenges–as

described above–were conducted to make sure that both

probes were subjected to dynamic changes of CPP and FiO2.

For the study presented here, we did not evaluate these

phases separately as the interventions were shorter than the

time over which the ORx is averaged (1 h).

Also, we compared both probes in the same pathology

per patient (either TBI or SAH), so that differences in

pathologies cannot be blamed for differences in ORx. Next,

we implanted both probes into the same cerebrovascular

region: that is the frontal lobe of the hemisphere of pathology.

Furthermore, since both probes were placed simultaneously,

there was equal duration of ORx monitoring with a mean of

7.2 (±2.4) days. Possible measurement problems during the

probes’ run-in times could be excluded, since data from day 1

were discarded.

In a previous study, we found significant differences in

absolute PbrO2 values between both probes [7]. Never-

theless, since the ORx only accounts for relative changes in

PbrO2 and CPP such absolute differences in PbrO2 might

not matter when calculating the ORx. However, looking at

our results, we conclude that this is not the case. These

differences in ORx might be caused by differences in

reaction times of both probes [7]. Finally, one might argue

that even though both probes differ both in absolute PbrO2

and ORx values there might still be a chance that at least

they agree on a CPPopt, which is simply the lowest point

of a U-shaped relation between the ORx and the CPP,

and therefore also independent of absolute ORx values.

However, even regarding the CPPopt we found limits of

agreement too large to declare equivalence [13]. This is

relevant since the clinician intends to use the CPPopt to

guide therapy. Therefore, the result that both probes do not

measure the same CPPopt is somewhat puzzling, as it is

difficult to imagine that there is more than one optimal

CPP.

The strength of our study is that it is the first one to

directly compare the ORx of Licox and Raumedic in vivo.

However, a number of limitations have to be mentioned.

First, we present a rather heterogeneous patient selection

that includes both SAH and TBI. Also, the rather limited

number of patients (n=11) in this study does not allow for

definite results or a reliable report on outcome. Further-

more, even though the probes were located in the same

cerebrovascular region, there might still be significant

metabolic differences within one territory which might

affect one probe more than the other.

It is important to mention that our results do not allow

for an assessment as to which probe is better suited for

ORx monitoring. Also, even if we suggest that Licox and

Raumedic are not equivalent, we still cannot say why Barth

et al. [19], who used the same probe as Jaeger et al. (Licox)

in the same pathology (SAH) over a comparable period of

time, did not find a correlation between ORx and PRx. As

there are no in vivo data on the agreement between the two

Licox probes when simultaneously placed into the same

vascular territory, this conflict remains yet to be resolved.

However, regarding the ongoing discussion about the

influence of the PbrO2 probe type on the true value of ORx,

we share the view that the probe type does matter.

Neurocrit Care (2013) 19:74–78 77

123

Conclusion

Although the number of patients was rather limited in this

study, we feel that the results suggest that the ORx depends

on the type of PbrO2 probe used for its measurement. Our

results support the view that different studies on the topic

of ORx are directly comparable only if they apply the same

type of PbrO2 probe.

Disclosure The authors are not subject to any conflict of interest.

The study was exclusively funded by the Department of Neurosurgery

of the Charite, Berlin.

References

1. Sorrentino E, Diedler J, Kasprowicz M, et al. Critical thresholds

for cerebrovascular reactivity after traumatic brain injury. Neur-

ocrit Care. 2011. doi:10.1007/s12028-011-9630-8.

2. Czosnyka M, Balestieri M, Steiner L, et al. Age, intracranial

pressure, autoregulation, and outcome after brain injury. J Neu-

rosurg. 2005;102:450–1.

3. Jaeger M, Schuhmann MU, Soehle M, Meixensberger J. Con-

tinuous assessment of cerebrovascular autoregulation after

traumatic brain injury using brain tissue oxygen pressure reac-

tivity. Crit Care Med. 2006;34(6):1783–8.

4. Jaeger M, Schuhmann MU, Soehle M, Nagel C, Meixensberger J.

Continuous monitoring of cerebrovascular autoregulation after

subarachnoid hemorrhage by brain tissue oxygen pressure reac-

tivity and its relation to delayed cerebral infarction. Stroke. 2007;

38(3):981–6.

5. Radolovich DK, Czosnyka M, Timofeev I, et al. Reactivity of

brain tissue oxygen to change in cerebral perfusion pressure in

head injured patients. Neurocrit Care. 2009;10(3):274–9.

6. Jaeger M. Different indices to assess cerebrovascular autoregulation

have different dynamic properties. Neurocrit Care. 2010;13(1):

163–5.

7. Dengler J, Frenzel C, Vajkoczy P, Wolf S, Horn P. Cerebral

tissue oxygenation measured by two different probes: challenges

and interpretation. Intensive Care Med. 2011;37(11):1809–15.

8. The R project for statistical computing R 2.13.1. 2011.

http://www.r-project.org. Accessed 20 September 2011.

9. Steiner LA, Czosnyka M, Piechnik SK, et al. Continuous moni-

toring of cerebrovascular pressure reactivity allows determination

of optimal cerebral perfusion pressure in patients with traumatic

brain injury. Crit Care Med. 2002;30(4):733–8.

10. Bland JM, Altman DG. Statistical methods for assessing agree-

ment between two methods of clinical measurement. Lancet.

1986;1(8476):307–10.

11. Myles PS, Cui J. Using the Bland–Altman method to measure

agreement with repeated measures. Br J Anaesth. 2007;99(3):

309–11.

12. Bland JM, Altman DG. Agreement between methods of mea-

surement with multiple observations per individual. J Biopharm

Stat. 2007;17(4):571–82.

13. Critchley LA, Critchley JA. A meta-analysis of studies using bias

and precision statistics to compare cardiac output measurement

techniques. J Clin Monit Comput. 1999;15(2):85–91.

14. Haitsma I, Rosenthal G, Morabito D, Rollins M, Maas AI,

Manley GT. In vitro comparison of two generations of Licox and

Neurotrend catheters. Acta Neurochir Suppl (Wien). 2008;102:

197–202.

15. Hoelper BM, Alessandri B, Heimann A, Behr R, Kempski O.

Brain oxygen monitoring: in vitro accuracy, long-term drift and

response-time of Licox- and Neurotrend sensors. Acta Neurochir

(Wien). 2005;47:767–74.

16. Purins K, Enblad P, Sandhagen B, Lewen A. Brain tissue oxygen

monitoring: a study of in vitro accuracy and stability of Neurovent-

PTO and Licox sensors. Acta Neurochir (Wien). 2005;152(4):

681–8.

17. Orakciouglu B, Sakowitz OW, Neumann JO, Kentar MM, Un-

terberg A, Kiening KL. Evaluation of a novel brain tissue

oxygenation probe in an experimental swine model. Neurosur-

gery. 2010;67(6):1716–22.

18. Huschak G, Hoell T, Hohaus C, Kern C, Minkus Y, Meisel HJ.

Clinical evaluation of a new multiparameter neuromonitoring

device: measurement of brain tissue oxygen, brain temperature,

and intracranial pressure. J Neurosurg Anesthesiol. 2009;21(2):

155–60.

19. Barth M, Woitzik J, Weiss C, et al. Correlation of clinical out-

come with pressure-, oxygen-, and flow-related indices of

cerebrovascular reactivity in patients following aneurysmal SAH.

Neurocrit Care. 2009;12(2):234–43.

78 Neurocrit Care (2013) 19:74–78

123