Timothy C. Hain, MD. Page last modified: June 29, 2014 Return to testing index
ECochG is a variant of brainstem audio evoked response (ABR) where the recording electrode is placed as close as practical to the cochlea. We will use the abbreviation ECOG and ECochG interchangeably below. ECOG is preferable to us as it is shorter.
ECOG is intended to diagnose Meniere's disease, and particular, hydrops (swelling of the inner ear). ECOG may also be abnormal in perilymph fistula, and in superior canal dehiscence. The common feature connecting these illnesses is an imbalance in pressure between the endolymphatic and perilymphatic compartment of the inner ear.
ECOG can also be used to show that the cochlea is normal, in persons who are deaf. The cochlear microphonic of ECOG may be normal in auditory neuropathy (Santarelli and Arslan 2002) as well as other disorders in which the cochlea is preserved but the auditory nerve is damaged (Yokoyama, Nishida et al. 1999).
Finally, ECOG's have also been used to as a indicator of the temporary threshold shift that may follow noise injury (Nam et al, 2004).
The conventional explanation for the summating potential (SP) of ECOG is that there is a nonlinear response in Reissner's membrane caused by elevated endolymphatic pressure and distension. Another name for the elevated pressure is "Hydrops". A more thoughtful analysis would suggest that low pressure in the perilymph could also cause a bowed Reissner's membrane, and this is the reason that ECOG may also be positive in perilymphatic fistula and superior canal dehiscence.
It would also seem logical that ECOG done upright might be positive in CSF leak, as CSF leaks exhibit similar audiological patterns to Meniere's disease, but there evidently have not been any attempts to use this logic as yet. Perhaps this is because of the inherent difficulty in doing an upright ECOG.
Other sources of nonlinear responses are rarely considered and also may be responsible for generation of the SP in some situations.(Cheng et al. 1994).
|Figure 1: Normal ECOG. The ECOG is interpreted by comparing the height of the SP to the AP.|
|Figure 1: Equipment used to record an ECOG, a Bio-Logic Navigator Pro||ECOG electrode -- the blue tip is placed so that it is next to the ear drum. After an ECOG is done, there often is a little bit of blue gel that remains in the ear canal.|
The technique involves placing an electrode into the outer ear canal, so that it is close to the ear drum. Our preference here in Chicago is to use a commercial wick electrode. These are very expensive -- but they work nicely. Other variants include a gold sponge (Nicolet Tiptrode), a wire or spring placed in the ear canal, or a needle that transfixes the ear drum. In general, these are either less effective or more painful.
|Transtympanic ECOG - -a needle transfixes the ear drum (Gibson)|
Needle type ECOG's such as shown above have fallen out of favor by most clinicians, because they are generally judged to be unreasonably invasive and are painful too (Bonucci and Hippolito, 2009). Their main use would seem to be in a person who has a perforation or who is undergoing surgery, as in this case, the needle can be place through the perforation and provide a better signal without doing any additional damage to the ear. This situation doesn't come up often.
Complications of needle ECOGs done on intact eardrums mainly include perforation and pain(Ng, Srireddy et al. 2001). Nevertheless they are still used in some settings (Ge and Shea, 2002). In our experience, the wick electrodes work very well and it is simply not generally necessary to use an invasive methodology.
Note that electrodes that transfix the TM are said to be not equivalent to electrodes on the TM or in the canal, as the electrodes are different distances between the electrical generators of the responses -- the SP and AP. In other words, they are measuring "apples/oranges", and need to be evaluated independently. We find this line of logic a little hard to follow as we thought that the SP was a derivative of the AP.
|Cochlear microphonic response from a 1 KHz tone burst (5 ms rise-fall, 10 ms plateau) (Margolis, 1992)|
The ear is stimulated with alternating polarity clicks (although tone bursts can also be used). The objective is to record wave-1 (there are 5 waves), with greater accuracy and to detect the "summating potential", which is a shoulder on wave 1. In some instances, the cochlear microphonic is inspected. To see the cochlear microphonic, one must use a tone burst stimulus rather than a click (see above). For tone bursts, the 1000 Hz stimulus (16 msec rise and fall, 12 ms plateau). The cochlear microphonic is, naturally, much clearer when one uses needle electrodes that are closer to the cochlea.
A person with an intact cochlear microphonic but reduced or absent wave 1, might have a 8th nerve site of lesion. In other words, the sound gets to the cochlea (creating the microphonic), but doesn't make it into the brainstem (i.e. no waves 1-5).
Because the ECOG is a test of the cochlea, and one can go deaf from things further in than the cochlea, it is possible to get an ECOG in a deaf person. For this reason, it is not logical to avoid doing ECOG's in people who are deaf, as this presupposes one knows the site of lesion.
|Figure 2: Abnormal ECOG on right side. Note the higher SP on the right.|
ECOG results are reported as an SP/AP ratio. A very conservative criterion for abnormality is a ratio of 0.5 or greater is considered abnormal. When attempting to diagnose Meniere's, a criterion of 0.41 might be a better choice.
While some labs use 0.35 as their upper limit, (Chung et al, 2004), considering all of the technical problems with ECOG (see below), we think it best to be more conservative than this as 0.35 is very close to the average SP/AP ratio reported by many authors (see below). The sensitivity of the SP/AP ratio to Meniere's disease depends on the criterion that you use for "positive". The higher the criterion, the lower the sensitivity and higher the specificity. With ECOG, there is no "clean" pick where a "normal" limit is very sensitive and specific, at the same time.
The sensitivity is said to be higher if it is performed during a symptomatic period (Devaiah et al, 2003), but it is difficult to get patients scheduled for the test when they are symptomatic. Also, some authors disagree that sensitivity is increased when patients are symptomatic (Levine, Margolis et al. 1998). ECOG is vulnerable to operator bias (see below), which adds to the confusion.
It would seem logical that one could use a normal hearing ear as the baseline, and look for differences between ears. This is rarely done.
|Figure 3: Noisy (useless) ECOG on L, obtained in person with severe hearing loss on L. Note that this ECOG is recorded "upside down" compared to the other traces in this document. They simply reversed the electrode input settings. We prefer it shown the other way around. A good thing about this recording is that the operator did not discard traces that he/she did not like. One can look at the left side and say - -this has no value. Sometimes operators are "sneaky" and eliminate noise traces that don't have much resemblance to an ECOG. This causes trouble.|
|0.25||.05||29||Aso et al (1991)|
|0.34||Coats et al, 1984|
|0.25||0.40||Ferrarro et al (1983)|
|0.27||0.51||Filipo et al (only 80 dB HL), 1989|
|0.19||0.37||Kitahara et al (1981)|
|0.225||0.43||28||Mori et al (1987)|
|0.31||0.11||0.42||10||Levine et al (1992)|
|0.32||0.14||100 ears||Wilson and Bowker (2002)|
In Meniere's disease, the central dogma states that the endolymphatic pressure is elevated, causing bowing of Reissner's membrane, and a resulting increased size of the SP/AP. The main difficulty with Meniere's disease is signal:noise. Noise is always a problem with ECOG. Signal is also a problem in Meniere's due to reduced size of the cochlear potential, from cochlear damage due to the disease itself.
|0.41||0.15||168||Aso et al (1991)|
|0.31||0.12||10||Campbell et al(1992).|
For TT ECOG, the literature suggests a wider separation between normals and hydropic ears, but TT ECOG is simply not practical in most patients whose symptoms from the TT ECOG might easily exceed their symptoms from their underlying condition.
|Meta analysis of ET ECOG, From Wuyts et al, 1997.|
The summary graph above is from Wuyts et al.
Recently Oh et al (2014) found ECOG testing including both SP/AP ratio and SP area ratio, of no value in Meniere's disease.
In SCD, there is an opening in the superior semicircular canal, which presumably lowers pressure in the perilymphatic compartment, due to a "third" window. One would think that this would be much more prominent upright than supine. Several authors have reported that the ECOG is elevated in persons with SCD, and that it normalizes with surgery(Arts, Adams et al. 2009, Adams, Kileny et al. 2011).
In PLF there is an opening in the oval or round window. Like the situation in SCD, this can cause the pressure in the perilymphatic compartment to be low. One would think that this could be more prominent upright than supine as well as instability in the SP/AP ratio, from run to run or with Valsalva.
Campbell and Abbas (1993) reported that posture affects the SP/AP in patients with PLF. They used an unusual protocol - -6000 Hz tone bursts. When they did a similar study in animals (1994), they found high variability prevented statistical significance.
Gibson (1992) using transtympanic studies during surgery reported good results. Gibson used the criteria of an increase of 15% of the AP with or without a decrease in the SP during a period of raised intrathoracic pressure. In other words, improvement with raised pressure. Similar results were reported by Saas and Densert, again using TT method (1997).
To our knowledge, nobody has done this study. We would expect a very high SP/AP ratio and position dependence, as the fenestration procedure is just another type of canal dehiscence.
Although ECOG is the only current method of detecting increased inner ear pressure, and the only reasonable non-invasive method of detecting a PLF, ECOG testing is difficult, has a questionable rationale for Meniere's disease, lacks reliable norms, lacks standards, and is vulnerable to operator bias.
The main technical problem plaguing ECOG is noise (see above). It requires getting a small electrode deep inside a small cavity (the external ear canal). The ear must be clean. There must be a functioning cochlea or nothing will be recorded.
Practically, it is rare to get an ECOG on someone with more than a 40 DB sensorineural hearing loss, but on the occasions that this occurs, the information is very valuable as it shows that the cochlea is functioning and is not likely to be the cause of the hearing loss.
Considerable patience is required. The test takes at least an hour. Certain electrodes have a better signal to noise ratio than others. We have not had much luck with the gold foil "tiptrodes". Gelled cotton wick electrodes, work better. While needle electrodes have the best signal to noise ratio, these are generally poorly tolerated and we have abandoned them.
Rationale for test: The Bayesian problem in Meniere's. As about 6% of the population has hydrops on autopsy (Honrubia, 1999; Rauch et al, 2001), one would expect that 6% of the otherwise normal population would have a positive ECOG. Considering that only 1/2000 people have Meniere's disease (0.2%), one would expect a high number of false-positives in ECOG testing.
From this perspective, the main utility of ECOG testing should be to detect hydrops, and to rule out Meniere's (as if one doesn't have hydrops, then Meniere's shouldn't be the diagnosis). Furthermore, if one doesn't have hydrops, then the low-dose gentamicin protocol for Meniere's disease might be something to avoid.
The main difficulty with this logic is that ECOG testing is difficult, and often uninterpretable in the very population that is of interest (persons with Meniere's), because of hearing loss.
So, to summarize, good judgment, experience, and clinical knowledge are a must to properly interpret ECOGs.
Lack of reliable norms: It is important when interpreting ECOG to consider the noise level, which is generally assessed by obtaining multiple trials. If they are all similar, then the standard deviation should be small and the result is likely to be correct. If they vary widely, the reliability of the average SP/AP ratio may be questionable. Similarly, if multiple trials are done and the SP/AP ratio varies wildly, one's confidence in the result should be lower than if it is very tight. At the present writing, typically little nor no consideration is given in reports to the confidence intervals for ECOG results. We would like to see some research done on this, and a tightening up of standards for reporting of ECOG's.
Practically, standards for ECOG testing are sketchy, and ECOG is vulnerable to operator bias . Roland and Roth (1997) reported that that there was "significant interinterpreter difference among SP/AP ratios calculated from the same tracings.
This does not even include the even more pernicious effect of operator bias. We have observed some labs acquire multiple ECOG runs, select out the ones that look "similar", average them, and delete the rest. The individuals who acquire ECOG's are often understandably reluctant to have the outside world see their "bad" runs. Unfortunately, this procedure can easily create a "response" from random noise, and also gives the operator (generally an unblinded audiologist who is given the working diagnosis of Meniere's disease) the opportunity to create abnormal results out of noise. We advise a methodology where all ECOG runs are at least displayed, so that the interpreter can determine the noise level.
|Figure 4A. Audiogram with reduced high frequency hearing on right||4B. ECOG for individual shown on the left -- nothing obtained on the right side due to high frequency hearing loss (presumably cochlear).|
Persons with poor high-frequency hearing , such as most older individuals, are likely to have higher noise levels, and therefore the limit of normal for their ECOG should be set higher. It is also quite possible for there to be no ECOG at all on the side of hearing loss -- as shown above in figure 3.
This creates a "catch-22" situation with ECOG. ECOG is generally not considered unless there is a hearing loss, and hearing loss makes ECOG less reliable. The best candidate for ECOG is probably someone with a monaural low-frequency sensorineural hearing loss, such as is often found in early Meniere's disease. The preservation of high-frequency responses makes the ECOG more likely to be valid and the low-frequency hearing loss also makes Meniere's more likely. Unfortunately, ECOG results are rarely reported as a function of hearing (e.g. Chung et al, 2004), making it difficult to interpret most studies.
The "area curve" SP/AP ratio has recently been suggested as a more sensitive method of detecting Meniere's disease (Deviah et al, 2003). In this method, the SP area is determined from response onset (also called "baseline"), to the first point after the AP where the response returns to baseline. In other words, it actually includes not only the SP but the entire AP. The AP area is determined from the onset of the AP through it's peak, and to the first opposite polarity peak. Although called the SP/AP area ratio, more accurately, it is the SP+AP/AP area ratio. It is difficult to say whether this method improves the diagnostic performance of ECOG, as one needs to know both sensitivity and specificity. It is possible that with this method one has just traded sensitivity for specificity. Only a small number of controls were reported. Time will tell whether this method is useful.
Higher stimulus rate ECOG's seem logical. With more stimuli, one can average out more noise, and get more accurate recordings. Wilson and Bowker (2002) studied many normal subjects. They found that the SP/AP ratio increases with stimulus frequency, going from (for example) an average of about 0.3 for conventional slow repetition rates to roughly 0.71 at 51 reps/sec., 0.86 at 101, and 0.81 at 151 reps/second. Variability across subjects became even greater at higher rep rates than at low rates. The authors did not report the # of repetitions used, and it is likely that they used the same # of reps for the slow and high speed rate. If this was the technique, we think that they may have "thrown the baby out with the bath water", as the whole point of high rep rates is to obtain more reps in the same amount of time, and reduce the effects of noise. So it appears to us that this study needs to be done over, with more reps.
Gamble and others (1999) reported that salt-loaded ECOG may be useful in patients who have normal ECOG's but a history suggestive of Meniere's. Similarly it has been suggested that ECOG may be useful in detecting allergic Meniere's disease. ECOG is performed before and after challenge with an allergen (Noell et al, 2001). The status of the challenge tests is presently uncertain.
Coats and Alford (1981) reported that ECOG results in 11 ears residing in persons diagnosed with Meniere's disease improved with administration of hypertonic glycerol. This test is rarely performed.
The author's experience with ECOG is mixed. It is a difficult test to perform, and poorly paid for by insurance. It is neither 100% sensitive nor 100% specific. In other words, you can have Meniere's disease but still have a normal ECOG, and you can also have an abnormal ECOG but not have Meniere's disease. We think it is a modestly useful test, most helpful in situations where other tests have failed to produce a clear answer as well as in situations where one is planning an invasive treatment such as gentamicin. It not indicated as a screening test because of the false-positive problem.
ECOG sometimes sometimes documents "Hydrops" (i.e. high pressure) , in a person who does not have a hearing loss. When this occurs, we will generally treat using the same protocols as we use for Meniere's disease. Whether or not this is useful is presently unknown.
Advice for patients: The most important consideration is who/where the test is done.
ECOG's are usually done in subspecialty otology practices. For example, we think that our practice in Chicago is one of just a few that do them well in a large metropolitan area. Pick a clinical setting where ECOG testing is done routinely.
We suggest avoiding having ECOG done in community settings, as well as having ECOG's done by persons with little experience in neurophysiology. It is not the sort of test that one would want someone to do who just finished their training, or who learned it last week.
Technicians can certainly do ECochG's very well and it is not necessary to have an audiology or "AuD" degree to do an ECochG. Experience, aptitude and training (roughly 8 hours) is what is needed to do a good job.
ECOG is also not a test that can reasonably be done in a setting that cannot remove ear wax (ie. by the technician that does evoked responses in a Neurology testing center). We don't think that ECochG's done using needle electrodes are generally a good idea, and we suggest refusing the test unless the reason to use a needle (over a "wick" electrode) is justified (for example, a perforation would be an acceptable reason).
Be sure that you get copies of the ECochG, including all traces. A type-written report is not sufficient (as there is no way to determine quality). You have to also trust the audiologist or technician that did the ECOG did not discard traces that they didn't like. In other words, ECOG is not the kind of test that you have done at your local hospital because the HMO doesn't cover it. Either do it right -- at a very experienced place -- or don't do it at all.
ECochG is usually covered by health insurance, e.g. Aetna and BCBS. ECochG may be indicated for diagnosis of hearing loss, but the main reason is for vertigo.
ECochG has been well studied. In Pubmed, there are approximately 345 publications with Electrocochleography in their title, going through 2013.
We think that studies of ECOG in CSF leak would be very logical, and we would like to see research done on this. We would expect that the SP/AP ratio would be higher near upright than supine.