advertisement
Frequency doubling test (FDT): an interview with Chris Johnson
Chris Johnson
Q: The principle underlying the FDT is the supposed sensitivity of the M cells and even the My-cell subset in glaucoma. Do you yourself believe in the My-cell theory?
CJ: The presence of M cells in human retina is controversial. They are certainly present in the cat. But there is some controversy about it in the primate. To me, it is not a real critical thing whether it is My cells or just magnocellular (M cell) mechanisms; it really must be part of the M-cell pathway because of the very high rate of flicker.
Q: You feel certain that FDT uses the M cells or the My cells?
CJ: It is too high a rate of flicker for P cells to be able to follow that flicker. I am confident of that. I am also confident that it must be a non-linear mechanism.
Q: How important are M cells for glaucoma?
CJ: My bias in the past has been that this selective loss, whether it is fibrocytes or whether it is M-cell pathways, has been overrated. I ascribe this to the redundancy theory, meaning that if you have a sparse system (you can take any system that has sparse representation), you will notice changes earlier. Any type of test, be it SWAP, FDT or some other test, will be effective if there is sparse representation. We have done our second year of testing of high risk OH in Portland; we have about 200 patients who are tested with a whole battery of tests: flicker, FDT, SWAP, peripheral VA, etc. We have several P-cell functions and several M-cell function tests and they all use the same stimulus configuration (24-2 pattern). SWAP and FDT, which are at opposite ends of the M cell and P cell spectrum presumably, show very high correlation in terms of the location of damage. There may be a slight preferential loss for large diameter nerve fibres or for M cell mechanisms, but the overall damage is, I believe, fairly nonspecific to any particular subset of nerve fibers. The fact that some of these tests, like SWAP and FDT, correlate quite well suggests that that is the case.
Q: SWAP and FD use different channels?
CJ: SWAP is clearly a P cell function, whereas FDT is an M cell function. I believe the correlation between the two exists because those are very sparse systems and the predominant nature of the glaucomatous loss is non-selective. Both systems have only a small percentage of the total number of ganglion cells. With minimal overlap and minimal back-up, if you loose a small percentage of ganglion cells, you will notice it earlier than in the case of highly redundant systems. M cells are around 10% of the total nerve fiber population and P cells are the majority: about 80-90%. However, P cells that carry the blue-yellow information are only 5-7% of the total nerve fiber population. So, we are talking low percentages for nerve fibers involved in both SWAP and FDT, and that is why believe that redundancy is the main explanation for early detection of loss with these tests. Reduced redundancy is not inconsistent with there being preferential loss of large diameter fibers or other hypotheses.
Q: The variability of FDT is less than in conventional perimetry. If so, is it because of the large object size? Is it so because of the fatigue phenomenon, examination time being much shorter for FDT? Did you compare it to SITA which has about the same examination time?
CJ: First of all, we are very careful not to make a direct comparison between FDT and conventional perimetry, because we are measuring different things. What we found was that the increase in variability in damaged visual field areas was much less for FD than for standard perimetry. For standard perimetry, when you go from a normal area to a damaged area, the variability on test to re-test can increase to three to four times. So you have a 3-400% increase in variablility. In FDT, it is only a fraction of that, maybe a 30-40% increase, when you go from normal to damaged areas. So you have more reliable responses in the damaged areas.
Q: FDT uses a very large object size, doesn't it?
CJ: I think the object size is part of it. FDT uses ten degree targets. We have a non-commercial version that has a 4° target and uses a 24-2 pattern, and you get a slight increase in variability when you go to a smaller target. So, certainly target size is part of it.
Q: What about sensitivity and specificity?
CJ: In previous studies, we and others have found that FDT has high sensitivity and specificity for detecting all stages of glaucomatous visual field loss. The sensitivity and specificity is quite similar to that obtained with conventional automated perimetry using full threshold test strategies. At the present time, we are concentrating on early glaucoma and high risk ocular hypertensives. High risk means elevated IOP in addition to at least one other risk factor; in most cases it is a suspicious appearance of the optic nerve head. In that group we are finding that in a small percentage (approximately 10-15%) of cases we are getting repeatable abnormalities with FDT that do not appear with standard perimetry. The custom FDT with the 24-2 pattern and 54 locations we find to be more sensitive than the commercial version using 17 large stimuli. Let me come back to SITA. There is no reason you cannot use a SITA-like strategy for FDT. That is one of the things that we are currently using simulation procedures to develop for FDT. I believe that we can probably reduce the testing time for FDT, using a maximum likelihood type of procedure such as SITA, by at least 30-40%.
Q: You say it correlates reasonably well with SWAP?
CJ: More often than not. Deficits for SWAP and FDT tend to be in the same regions and they tend to show a similar extent of damage.
Q: Any idea about motion? Comparison? Motion should be using the same sort of system.
CJ: It all depends on whether the FDT is truly a subset of M cells or not. If it is, it should do a little bit better than motion. And, although we are not evaluating motion at the present time, we are evaluating high frequency flicker, which is believed to be the same mechanism that is involved in motion perception. Flicker does not seem to be, at least in our initial studies, as sensitive as FDT. And, in fact, a couple of years ago, we looked at presenting just the spatial component or the temporal component alone of the FDT test in comparison to the FDT stimulus, and the FDT stimulus did better than the spatial component alone or the temporal component alone for detection of glaucomatous visual field loss.
Q: Any idea about comparison to morphological data, RNFL, disc?
CJ: Not at the present time. As far as optic disc topography is concerned, we are also taking these measures in our current study. We are in the process of doing those types of evaluations and analyzing the data. In another 6-12 months, we should have some information about the relationship of FDT to morphologic characteristics of the optic disc and retinal nerve fiber layer.
Q: What specificity do you reach for threshold procedures?
CJ: For threshold procedures, it is about 96%; that was intentional. We developed both the threshold and screening procedures for FDT to optimize specificity. Sensitivity in early glaucoma is about 85-90%, and is above 96% for moderate and advanced glaucoma. There are two screening procedures; one of them is designed to have high specificity and presents targets at the 1% probability level. Studies with that test show that it has specificity levels of about 98%, with a sensitivity of about 75-80%. This procedure may miss early defects. We also developed a test procedure that presents targets at the 5% probability levels. In comparison to the 1% probability level screening test, the specificity of this procedure decreases to about 90% and the sensitivity increases to 85-90%. I believe that the 1% probability level screening test is most appropriate when you are doing mass screening, because you want to optimize specificity. The 5% probability level is probably more appropriate for clinical practice, where there is a greater suspicion of a possible eye problem and greater sensitivity for detecting early losses is more important. Both screening tests work well, and they both take about 45 seconds for a normal eye, and between 45 and 80 sec. for an abnormal eye.
Q: Do you know what raised or lowered IOP dose to the test?
CJ: I do not believe that this has been examined. We have not noticed any significant relationship between IOP and FDT threshold values in our OH population.
Q: Did you see in this group of OH patients that a patient was normal on one test occasion and abnormal on another, and could you see some differences in those particular patients? Clinical correlations?
CJ: It occurred. We need more information and some more years to really be able make any definitive statements.
Q: Does FD have a role in mass screening?
CJ: We currently have funding for a three-year study during which we will be screening more than 10,000 people per year. We will be targeting at-risk populations such as African-Americans, native Americans, and individuals in rural areas that don't have access to traditional health care. So we will be examining how effective it is as a mass screening tool. I think that this is a useful visual function screening test; it is portable, it is quick and easy to perform, it is resistant to the effects of blur and pupil size, and it has high sensitivity and specificity. Detection of visual field loss from glaucoma and other eye diseases is clearly one of the functions of this test procedure. In addition, I think that there are some indications that it may be useful for monitoring progression of visual field loss as well. It correlates well with the HFA for distinguishing different levels of glaucomatous visual field loss. Also, the variability in damaged visual field areas is not as great for FDT as for conventional perimetry. Longitudinal follow-up will of FDT in glaucoma patients will be necessary to determine.
Q: You do not exclude the possibility that a technician could use this test for the follow-up of progression?
CJ: For the follow-up of progression, a stimulus presentation pattern with a larger number of test locations, such as 24-2 pattern, would probably be desireable. There would clearly give a much better characterization of visual field defects than the 17 target pattern. In combination with the SITA-like strategy, I can see where it may be possible to have a more sensitive test that also has low variability and is able to monitor progression effectively. I think this seventeen target test may be a bit difficult for monitoring progression because of the limited spatial characterization of defective areas.
Q: How early do we really want to detect glaucoma?
CJ: One of the things I was going to mention is that, for a long time, the emphasis in developing new tests has been on the earlier detection of glaucoma. I see now that there is a shift in the emphasis of test development from trying to find the earliest losses to finding better methods of detecting a change in people who already have damage: in other words, more effective methods of determining progression. We are just now beginning to analyze SWAP data from the OH treatment study. About 350 patients are included in the study and approximately half of them are treated and half of them untreated. If it can be demonstrated from the OHTS SWAP study that treatment of an individual with an early abnormality such as a SWAP deficit is beneficial, then I think early detection is important. If there is no long-term benefit to treating someone with an early deficit, then early detection of an abnormality is not as important. Hopefully, we will have some preliminary answers to these questions in the near future.
E.L.G.
Address: Chris A. Johnson, PhD, Devers Eye Institute, Legacy Clinical Research and Technology Center, 1225 NE Second Avenue, PO Box 3950, Portland, OR 97208-3950, USA
