Keywords
intraoperative neuromonitoring - facial nerve - cerebellopontine angle - motor evoked
potentials - vestibular schwannomas
Introduction
Neurosurgeries undertaken for the resection of vestibular schwannomas[1]
[2] or of other cerebellopontine angle lesions such as meningiomas,[3] or for trigeminal neuralgia[4] or hemifacial spasms[5] can be complicated by deterioration of the facial nerve function. In all cases,
not only the proximity of the lesion to the nerve but also the size of the lesion
increase the risk of facial nerve injuries.[6]
For these surgeries, intraoperative monitoring techniques applied to the facial nerve
have been developed using three different modalities: electromyography (EMG), direct
electrical stimulation, and motor evoked potentials (MEPs).
EMG recording has been applied for continuous monitoring of the nerve activity[7]
[8]
[9] and analyzed with the objective of predicting the postoperative functions of the
facial nerve.[10]
[11] EMG has been applied in our center as a background monitoring technique of the possible
mechanical stimulations of the nerve only.
Direct electrical stimulation of the facial nerve, introduced in otolaryngology[12]
[13] and neurosurgery,[14] helps comfort the location of the facial nerve throughout the surgery. It has been
acknowledged to be effective in limiting the risks of postoperative facial paresis.[15]
[16]
MEPs of the facial nerve help check the integrity of the motor pathway from the primary
motor cortex to the muscles of the contralateral hemiface. The efficacy of MEPs is
dependent on the alarm criterion applied to trigger an alarm and the consecutive adaptation
of the surgical strategy to prevent an irreversible damage to the facial nerve. The
choice of the alarm criterion is still a matter of debate. A 50% decrease in MEP amplitude
has been reported as reliable in some studies[17]
[18] but not strict enough in others.[19]
[20]
[21]
In this context, the objective of this retrospective study was to analyze the validity
of the alarm criterion applied at the time of surgery, that is, a reproducible 50%
decrease in MEP amplitude in neurosurgeries performed for vestibular schwannomas or
for other lesions of the cerebellopontine angle.
Although intraoperative monitoring should be considered an “intervention”-like technique
and hence its efficacy ideally evaluated with control groups, as discussed by Howick
et al,[22] this is actually not performed for ethical considerations. In addition, as discussed
by Howick et al,[22] intraoperative monitoring cannot present false-positive cases. This point is of
major importance. Indeed, in case an alarm has been raised and no new deficit was
observed postoperatively, it cannot be determined if that was due to or not due to
the alarm raised during the surgery. Hence, specificity and positive predictive value,
both necessitating the identification of false-positive cases, should not be calculated
for intraoperative neuromonitoring techniques and in particular for MEPs. The computation
of sensitivity and negative predictive values for MEPs remains effective to postoperatively
evaluate an alarm criterion, in the single objective to avoid positive cases in the
procedures to come.
Methods
Patients
This retrospective study included patients who undertook a surgery of the cerebellopontine
angle, with corticobulbar MEPs in our center, between 2015 and 2018. The participation
to the study was offered to all patients who undertook such a surgery. They all gave
their signed agreement to be part of this study anonymously. The local ethics committee
approved this study (CCER, 2018–02029).
Thirty-three patients participated in the study: 16 patients suffered from a vestibular
schwannoma (7 males, 9 females; mean age 53 years, standard deviation 11.5 years),
while 17 patients suffered from various other cerebellopontine angle lesions (5 meningiomas,
4 hemifacial spasms, 6 trigeminal neuralgias, 1 epidermoid cyst, and 1 dural arteriovenous
fistula; 7 males, 10 females; mean age 59 years, standard deviation 13.4 years). These
lesions were reached by a retrosigmoid approach.
Anesthesia
General anesthesia was performed with propofol and sufentanil: induction was reached
with propofol concentrations of 4.5 to 5.0 mg/mL and sufentanil concentrations of
0.3 to 0.4 ng/mL. Concentrations were then adjusted throughout the operation according
to the needs of the patient. No inhaled agents were used during the resection.
Monitoring
The facial MEPs were performed with transcranial electrical stimulation (train of
4 anodic pulses, 400 Hz, 0.4-ms phase duration, no averaging; NimEclipse system, Medtronic,
Minneapolis, Minnesota, United States). One subdermal anode electrode was placed either
between C5 and C3 or between C6 and C4[23] (international 10/10 EEG system) on the contralateral side of the facial nerve to
monitor, with the cathode at the vertex (DME1001, Medtronic Xomed Inc., Jacksonville,
Florida, United States). The location of anodes was therefore approximately similar
to that of the anodes already reported for MEPs of the facial nerve.[23] Stimulation made of a single pulse was applied from time to time to verify that
the stimulation did not directly excite the facial nerve distally.[24]
The MEPs were measured using subdermal needles (DSN2282, Medtronic Xomed Inc., Jacksonville,
Florida, United States) positioned in different muscles of the lower hemiface (mentalis
or orbicularis oris, nasalis) ipsilateral to the side of the surgery. MEPs from the
ipsilateral hand (thenar muscles) were also recorded to detect possible changes in
systemic parameters, such as changes in mean blood pressure. MEPs were filtered (bandpass:
80–2,000 Hz) to improve their reproducibility, attenuating the low-frequency part
of the MEP signal.
During surgery, the alarm criterion was defined as a reproducible 50% decrease in
MEP amplitude, resistant to a 10% increase in stimulation intensity.
Direct electrical stimulation was applied using a monopolar probe (Inomed Medizintechnik
GmbH, Emmendingen, Germany). Each stimulation consisted of a biphasic pulse of 0.2 ms
per phase, with maximum intensity of 0.7 mA, delivered every second. Motor responses
were collected for the same ipsilateral muscles (mentalis or orbicularis oris, nasalis)
and for the frontalis muscle.
Facial Outcome
To correlate electrophysiological measurements with the clinical outcome, the HBS[25] measuring facial paresis was applied. The scores were evaluated preoperatively and
7 days and 3 months postoperatively at the time of standard clinical consultations.
This score is graded from 1 to 6, with 1 being normal facial function and 6 a total
facial nerve palsy. It includes the movements of the forehead, the eyes, and the mouth.
Statistics
The patients were classified into different categories according to the possible changes
in the amplitude of facial MEPs and the possible postoperative changes in their HBS.
The postoperative distribution of the cases among true-positive (TP; alarm raised,
reproducible 50% decrease in MEP amplitude, resistant to a 10% increase in stimulation
intensity, at the end of monitoring and new deficits), true-negative (TN; alarm raised
or not at the time of surgery, decrease in MEP amplitude of <50% or none at the end
of monitoring and no deficit), and false-negative (FN; alarm raised or not at the
time of surgery, decrease in MEP amplitude of <50% or none at the end of monitoring
but new deficits) cases were computed for both groups of patients.
The confidence intervals of sensitivity and negative predictive values were calculated
using the MedCalc software (https://www.medcalc.org/calc/diagnostic_test.php).
Results
Facial MEPs were obtained in 31 out of 33 patients. Facial MEPs could not be obtained
without concomitant distal stimulation of the facial nerve in two patients. In one
patient with hemifacial spasm, the HBS was unreliable in the context of preoperative
Botox treatment.
The facial MEPs obtained in patient P10 are described in [Fig. 1]. The patient suffered from a left vestibular schwannoma (Koos grade III), with intrapetrous
extension into the middle and internal ear, and with preoperative cophosis. The MEP
loss (white arrow in [Fig. 1]) occurred while MEPs were interrupted for direct electrical stimulation during intradural
resection. Resection was stopped at that time, leaving small residue. Note that although
every MEP trace in [Fig. 1] is the response to one single stimulation obtained without averaging responses,
the MEP amplitudes of the orbicularis oris muscle (“OrOr + -OrOr-”) were reproducible.
This could be obtained thanks to filters (80–1,500 Hz) applied to each MEP to improve
the signal-to-noise ratio.
Fig. 1 Facial motor evoked potentials (MEP) (filtering 80–1,500 Hz) recorded in patient
P10 during resection of a vestibular schwannoma. The reproducible loss (–100%) of
the orbicularis oris MEP was resistant to an 8% increase in stimulation intensity
(white arrow). Three months postsurgery, the patient did not yet recover from this new facial
paresis.
Among the group of 15 patients with vestibular schwannomas and with contributive facial
MEPs ([Table 1]), at 3 months postsurgery, no change in HBS was observed in 6 patients for whom
no change in MEP amplitude or a decrease in MEP amplitude of <50% was observed at
the time of surgery (6 TN); an increase in the HBS was observed in 7 patients for
whom a reproducible decrease of >50% in MEP amplitude, resistant to a 10% increase
in stimulation intensity, was observed (7 TP); an increase in HBS was observed in
2 patients for whom no decrease in MEP amplitude or a decrease of <50% in MEP amplitude
was observed (2 FN). [Table 2] indicates sensitivities and negative predictive values computed postoperatively
for different percentages of decrease in MEP amplitudes, in the group of patients
with vestibular schwannomas. With the standard alarm criterion of 50% decrease in
MEP amplitude, at 3 months postoperatively, facial MEPs presented a sensitivity of
77.8% (confidence interval [CI]: 40.0–97.2%) and a negative predictive value of 75.0%
(CI: 46.0–91.3%). Postoperative analyses suggest better sensitivity of 88.9% (CI:
51.7–99.7%) and negative predictive value of 85.7% (CI: 48.6–97.4%) with an alarm
criterion of 30% decrease in the MEP amplitude.
Table 1
Individual changes observed in patients with vestibular schwannomas
Patients
|
% change in MEP mentalis/nasalis muscle
|
Stimulation amplitude at baseline (mA)
|
Alarm (Y/N)
|
Action trigged
|
HB score, pre-operative
|
HB score, day 1
|
HB score, day 7
|
HB score, month 3
|
Outcome, month 3
|
P1
|
–100/–100
|
68
|
Y
|
None (iatrogenic section)[a]
|
1
|
6
|
6
|
6
|
TP
|
P2
|
0/20
|
106
|
N
|
NA
|
1
|
1
|
1
|
1
|
TN
|
P3
|
0/NC
|
70
|
N
|
NA
|
1
|
3
|
2
|
1
|
TN
|
P4
|
–100/–100
|
63
|
Y
|
None (expected[b])
|
3
|
6
|
6
|
6
|
TP
|
P5
|
42/NC
|
76
|
N
|
NA
|
2
|
3
|
2
|
2
|
TN
|
P6
|
–33/NC
|
76
|
N (50% criterion)
|
NA
|
1
|
4
|
4
|
4
|
FN
|
P7
|
33/300
|
NA
|
N
|
NA
|
1
|
2
|
2
|
1
|
TN
|
P8
|
–70/–20
|
63
|
Y
|
Resection stopped
|
1
|
4
|
4
|
2
|
TP
|
P9
|
0/0
|
66
|
N
|
NA
|
1
|
4
|
4
|
3
|
FN
|
P10
|
–100/NC
|
76
|
Y
|
(cf. “Results” section)
|
1
|
4
|
4
|
4
|
TP
|
P11
|
NA/–100
|
90
|
Y
|
Adaptation of the resection strategy
|
1
|
3
|
3
|
2
|
TP
|
P12
|
NC/–50
|
73
|
Y
|
1
|
2
|
4
|
4
|
TP
|
P13
|
–8/–100
|
69
|
Y
|
1
|
2
|
2
|
2
|
TP
|
P14
|
–17/0
|
50
|
N
|
NA
|
1
|
2
|
2
|
1
|
TN
|
P15
|
0/NC
|
76
|
N
|
NA
|
5
|
5
|
5
|
5
|
TN
|
Abbreviations: FN, false negative; HB score, House–Brackmann score; MEP, motor evoked
potential; N, no alarm given; NA, not adapted; NC, noncontributive; TN, true negative.
Percentages of MEP changes are indicated first for the mentalis and second for the
nasalis muscle; they were computed from the amplitude of responses at the time of
baseline and at the end time of resection. Stimulation intensity at baseline is indicated
(mA); Y: warning was given based on MEPs; N: no warning based on MEPs; action trigged
in case of warning is indicated.
a Before direct electrical stimulation.
b The deterioration of the facial nerve was expected in this redo surgery for a giant
schwannoma with foraminal invasion.
Table 2
Postoperative evaluation of sensitivities and of negative predictive values for different
decreases in motor evoked potential (MEP) amplitudes, computed in the group of patients
with vestibular schwannomas
MEP decrease (%)
|
True negative
|
True positive
|
False negative
|
Sensitivity (%)
|
Negative predictive value (%)
|
10[a]
|
5
|
8
|
1
|
88.9
|
83.3
|
20
|
6
|
8
|
1
|
88.9
|
85.7
|
30
|
6
|
8
|
1
|
88.9
|
85.7
|
40
|
6
|
7
|
2
|
77.8
|
75.5
|
50
|
6
|
7
|
2
|
77.8
|
75.0
|
60
|
6
|
6
|
3
|
66.7
|
66.7
|
70
|
6
|
6
|
3
|
66.7
|
66.7
|
80
|
6
|
5
|
4
|
55.5
|
60.0
|
90
|
6
|
5
|
4
|
55.5
|
60.5
|
100
|
6
|
5
|
4
|
55.5
|
60.0
|
The numbers of true positive (TP; permanent MEP decrease above that threshold at the
end of resection and new deficits), true negative (TN; MEP decrease under that threshold
at the end of resection and no deficit), and false negative (FN; MEP decrease under
that threshold at the end of resection but new deficits) are indicated with sensitivity
(TP/(TP + FN)) and with negative predictive value (TN/(TN + FN)).
a With this criterion, one case would be unclassified (equivalent to one case of false
positive).
Among the group of 15 patients with other cerebellopontine angle lesions and with
contributive facial MEPs ([Table 3]), at 3 months postsurgery, no change in HBS was observed in 14 patients for whom
no change in MEP amplitude or no decrease in MEP amplitude of >75% was observed at
the time of surgery (14 TN). A deterioration in the HBS of one grade was observed
in one patient for whom a 75% decrease in MEP amplitude was observed (1 TP).
Table 3
Individual changes observed in patients with other cerebellopontine angle lesions
Patients
|
Lesion type
|
% change in MEP mentalis/nasalis muscle
|
Stimulation amplitude at baseline (mA)
|
Alarm (Y/N)
|
Action trigged
|
HB score, preoperative
|
HB score, day 1
|
HB score, day 7
|
HB score, month 3
|
Outcome, month 3
|
P16
|
Trigeminal neuralgia
|
200/0
|
44
|
N
|
NA
|
1
|
1
|
1
|
1
|
TN
|
P17
|
Meningioma
|
–28/150
|
58
|
N
|
NA
|
1
|
1
|
1
|
1
|
TN
|
P18
|
Epidermoid cyst
|
122/NC
|
86
|
N
|
NA
|
1
|
1
|
1
|
1
|
TN
|
P20
|
Trigeminal neuralgia
|
33/NC
|
67
|
N
|
NA
|
1
|
1
|
1
|
1
|
TN
|
P21
|
Trigeminal neuralgia
|
33/NC
|
157
|
N
|
NA
|
1
|
1
|
1
|
1
|
TN
|
P23
|
Meningiomaa
|
0/–50
|
50
|
Y
|
Adaptation of the resection strategy
|
2
|
2
|
2
|
1
|
TN
|
P24
|
Trigeminal neuralgia
|
NC/–25
|
81
|
N
|
NA
|
1
|
1
|
1
|
1
|
TN
|
P25
|
Meningioma
|
–20/–16
|
59
|
N
|
NA
|
2
|
2
|
2
|
1
|
TN
|
P26
|
Trigeminal neuralgia
|
0/0
|
113
|
N
|
NA
|
1
|
1
|
1
|
1
|
TN
|
P27
|
Fistula
|
–14/NC
|
131
|
N
|
NA
|
1
|
1
|
1
|
1
|
TN
|
P28
|
Hemifacial spasm
|
–50/60
|
65
|
Y
|
Pause
Irrigation
Release
|
3
|
3
|
2
|
2
|
TN
|
P29
|
Hemifacial spasm
|
–50/6.7
|
72
|
Y
|
1
|
1
|
1
|
1
|
TN
|
P30
|
Hemifacial spasm
|
–75/NC
|
62
|
Y
|
3
|
4
|
4
|
4
|
TP
|
P31
|
Meningiomaa
|
–25/NC
|
93
|
N
|
NA
|
1
|
1
|
1
|
1
|
TN
|
P32
|
Meningioma
|
–16.7/0
|
75
|
N
|
NA
|
1
|
1
|
1
|
1
|
TN
|
Abbreviations: HB score, House–Brackmann score; MEP, motor evoked potential; NA: not
applicable; NC, noncontributive MEP; TN, true negative; TP, true positive.
Lesion types; aintrapetrous extension; percentages of motor evoked potentials (MEP) changes are indicated
first for the mentalis and second for the nasalis muscles; they were computed from
the amplitude of responses at the time of baseline and at the end of resection. Stimulation
intensity at baseline is indicated (mA); Y: warning was given based on MEPs; N: no
warning based on MEPs; action trigged in case of warning is indicated.
[Table 4] indicates sensitivities and negative predictive values computed postoperatively
for different percentages of decrease in MEP amplitudes, in the group of patients
with other cerebellopontine angle lesions. With the standard alarm criterion of 50%
and a 70% decrease in MEP amplitude as alarm criterion, at 3 months postoperatively,
facial MEPs presented a sensitivity of 100.0% (CI: 2.5–100.0%) and a negative predictive
value of 100.0%. Nevertheless, given the small size of the group with other cerebellopontine
angle lesions, the accepted 50% decrease in MEP amplitude as alarm criterion should
not be loosened.
Table 4
Postoperative evaluation of sensitivities and of negative predictive values for different
decreases in motor evoked potential (MEP) amplitudes, computed in the group of patients
with other cerebellopontine angle lesions
MEP decrease (%)
|
True negative
|
True positive
|
False negative
|
Sensitivity (%)
|
Negative predictive value (%)
|
10[a]
|
6
|
1
|
0
|
100.0
|
100.0
|
20[a]
|
9
|
1
|
0
|
100.0
|
100.0
|
30[a]
|
12
|
1
|
0
|
100.0
|
100.0
|
40[a]
|
12
|
1
|
0
|
100.0
|
100.0
|
50[a]
|
15
|
1
|
0
|
100.0
|
100.0
|
60[a]
|
15
|
1
|
0
|
100.0
|
100.0
|
70[a]
|
15
|
1
|
0
|
100.0
|
100.0
|
80
|
15
|
0
|
1
|
0.0
|
93.7
|
90
|
15
|
0
|
1
|
0.0
|
93.7
|
100
|
15
|
0
|
1
|
0.0
|
93.7
|
The numbers of true positive (TP, MEP decrease above that threshold at the end of
resection and new deficits), true negative (TN, MEP decrease under that threshold
at the end of resection and no deficit) and false negative (FN, MEP decrease under
that threshold at the end of resection but new deficits) are indicated with sensitivity
(TP/(TP + FN)) and with negative predictive value (TN/(TN + FN)).
a With this criterion, cases would be unclassified (equivalent to false-positive cases).
Discussion
The standard alarm criterion of 50% decrease in MEP amplitude did not appear reliable
for the monitoring of the facial nerve in all cerebellopontine angle surgeries. It
appears that different alarm criteria should be used in the monitoring for patients
undergoing vestibular schwannoma surgery and for patients with other cerebellopontine
angle lesions. Variations in facial MEPs observed during vestibular schwannoma resections
must be carefully monitored because a reproducible decrease of 30% in MEP amplitude,
resistant to a 10% increase in stimulation intensity, can already be assigned to a
long-term facial nerve dysfunction. This observation may be due to the fact that vestibular
schwannomas are most often intertwined with fibers of the facial nerve. This is similar
to the application of different alarm criteria for transcranial as opposed to direct
cortical stimulation during the resection of insular, pre-, or postcentral lesions.[26] Dong et al[24] and Matthies et al[17] also found a lower alarm criterion, that is, a 35% decrease in MEP amplitude, prone
to trigger more warnings than with a 50% decrease in MEP amplitude, in similar and
larger groups of patients. This low alarm criterion is in line with the previous observation
of the high rates of poor facial nerve outcomes in the cases with a 50% decrease in
MEP amplitude.[20]
[21] Unfortunately, these previous analyses of facial MEP performances were not enough
acknowledged in the domain of intraoperative neuromonitoring.
Sensitivity and negative predictive values were found to be higher than those reported
recently by Tawfik et al.[19] This observation can be explained by the fact that a single alarm criterion was
applied across different lesion types. This criterion was a 50% decrease in MEP amplitude,
which might not be low enough for vestibular schwannomas. Indeed, a decrease in MEP
amplitude of more than 30% can already be associated with an increase in the HBS and
hence with a facial nerve deficit. This alarm criterion could have led to a lower
sensitivity, with FN cases, and could have contributed to the conclusion that neuromonitoring
did not decrease the rate of facial nerve deterioration.[19]
Seidel et al[16] reported a low complication rate using continuous direct electrical stimulation
during the resection of vestibular schwannomas. In the present series, direct electrical
stimulation was usually performed temporarily and was not always done before the observation
of MEP decreases. The use of continuous direct electrical stimulation in our series
could have contributed to better protecting the facial nerve of at least one accidental
event and to further lowering the facial nerve deterioration rate.
Also, the fact that the alarm criterion at the time of the surgery was a 50% decrease
in MEP amplitudes and not yet a reproducible 30% for vestibular schwannomas could
have contributed to not giving early enough the information that continuing the resection
would present a high risk of irreversible facial nerve deficits.
Limitations
This is a retrospective study in a group of patients of limited size. Larger group
sizes would help narrower the confidence intervals.
For ethical reasons, no control group was included in the study as is the case for
the majority of studies conducted in the domain of intraoperative neuromonitoring.
Conclusion
The analysis of facial nerve MEPs suggested that the most reliable alarm criterion
to be applied was different for surgeries of vestibular schwannomas as compared with
surgeries of other cerebellopontine angle lesions. As introduced by Dong et al[24] and recalled by Matthies et al,[17] a reproducible 30% decrease in MEP amplitudes, resistant to a 10% increase in stimulation
intensity and prone to trigger early warnings, was verified for vestibular schwannomas.
Given the small size of our group with other cerebellopontine angle lesions, the established
50% decrease in MEP amplitude as an alarm criterion should not be loosened to a less
strict criterion for these surgeries. The stricter alarm criterion found for the resection
of vestibular schwannomas, which is the highest risk surgery for the facial nerve,
suggests that its application could improve the reliability of facial MEPs and render
facial MEPs contributive to the decrease of the principal complications of this surgery.