Скачать 330.02 Kb.

DR. BULLIMORE: On relatively short notice, I was provided with blank overhead sheets and a pen, just in case they were needed. I really do want to speak just to the questions raised by Morris and talk about what aspects of astigmatism I personally believe should be reported. Must like spherical corrections, in my mind, the most important thing is the percentage of the cylinder that is corrected, so in my mind, the most important thing will be to give the preoperative mean astigmatism and the postoperative mean astigmatism, and compare the two. That is based on past experience, which I am not allowed to talk about; that's what the panel seems to have based their judgments on. So what do we need to do beyond that? Well, there are some concerns about, I think, vector analysis, and vector analysis is just a tool that helps us understand not so much the way the laser or other device worked, but perhaps the way it didn't work, and there are good reasons for also including some vector analysis in the summary data. Let's just take an individual patient, and ignoring axis for the moment, if a person is +2.00 D of cylinder before and +.50 afterwards, I think we'd all agree that the reduction in astigmatism is 1.50 D, and comparing the post as a function of the pre, we've got a 1.50 D correction, and that's 75 percent of what they started with. That is important information to go in the PMA, and that's what I want to see. Now, because we don't have axis information, we don't know what has really gone on here. For example, had this been axis 90 before and axis 90 afterward, obviously, then, without the need for vectors, we figure out that +1.50 is the change. But if the axis before is 90, and the axis afterward is 180, clearly, the total refractive change induced by the laser is +2.50. What is those axis are not so nicely aligned at 90 or at 180 that's when we need vectors. And I think it is useful to include vector analysis to actually determine the amount of surgically induced refractive cylinder. And really, the surgically induced refractive cylinder is defined by the agency and maintained in what I put down in my notes is really the vector difference between these two cylinders. Now, I don't want to lecture for 10 minutes on vectors, so I am going to sit down and quit while I am ahead. What else do you want to know? DR. McCULLEY: Is everyone comfortable with the issues of vector analysis and the calculation of the surgically induced correction is that the agency's definition? SIRC is vector difference? DR. WAXLER: Yes. DR. McCULLEY: That's correct. And you are leaving it to the sponsor to use whatever program or formula presumably, whatever program one uses, the difference between the vectors is a real number, but there may be different software packages to get there. But that really isn't the issue, or shouldn't be the issue. The difference is what is the issue. DR. WAXLER: I think I'll try without Malvina here, but that may be hazardous also. I think one of the issues is well, what you have already addressed but the other issue is, again, is there a line, is there an acceptable amount, is there an amount of inducement that is acceptable or not, because we have to make actual decisions on particular documents. We can't dither around. We have documents in there, and it has certain numbers. Is there an amount of change that is beyond the pale that, clinically, you would find unacceptable? DR. BULLIMORE: Again, in my report to you, Morris, I did a very cursory review of the recent literature just to see what was being reported out there, and of the 7 studies that I could readily access on the National Library of Medicine database, the average percentage reduction in cylinder that's just ignoring vectors, ignoring everything, just taking the average pre and the average post you come up with 70 percent. That, if you like, is the standard of care that people are reporting or at least getting published. Maybe the people who report 30 percent reduction don't get published, so there may be some inherent bias in the literature. Again, we can't talk about specific PMAs, but in my Table 1 in my notes, the data I used there, if you like, percentage of efficacy was 79 percent. That's a pretty impressive number. I would be happy to entertain a number of 70 percent or lower to go in the guidance documents. One thing that is important to remember is when you are correcting astigmatism, you have, if you like, two degrees of freedom what the laser or device does and how accurately it is aligned. And when you start playing around with vectors, you find that if you are off by 10 degrees, even if you make the appropriate cylindrical correction or the appropriate astigmatic correction, you lose a third of your effectiveness or efficacy. You are down to 66 point whatever percent just by being 10 degrees off. If you are off by 5 degrees, you are still only just batting above 83 percent. And if you think about what is involved, there measurement of the preoperative cylinder, alignment of the patient, and then alignment of the device. There are three sources of error, and if those combine in a destructive way, and you end up with 10 percent, which is not unreasonable, you are only going to correct two thirds of the astigmatism. So when you see numbers in the literature of 70 percent, they are believable. When you see numbers like 79 percent, they are very impressive, very impressive, and I don't think we should necessarily put the bar too high. DR. McCULLEY: Would you have CMD/IRC and SIRC/IRC percentage the same, or different percentages? DR. BULLIMORE: No. They should be very different. I think that in terms of our primary evaluation of did everybody get this; I know the panel got them the numbers I wrote down, I'll write them down again. The preop was 1.42 these are means the postop was 0.30, so the CMD, which is the this is FDA language cylinder magnitude difference, was 1.12. That's just simple arithmetic there. And then, the percentage, CMD/IRC, assuming that the IRC, the intended refractive cylinder, was the same as the preop, it's just this as a percentage of this. And in the data that I used in my example, that was 79 percent. Now, if you want to now calculate the SIRC, in the dataset which I had access to, I calculated that to be 1.36 D for a preop or intended refractive correction of 1.42. We are probably going to judge the device's efficacy based on this number. We don't need immediately to resort to vector analysis to help us judge whether this is an effective device. In the same way, we can make judgments on spherical correction based on their percentage effectiveness, or the number of people between +/ 0.50. We can get most of our gut reaction from this. It becomes something of an intellectual exercise to figure out whether they don't hit 100 percent because of axis problems or because the laser just doesn't do that well. And vector analysis can help you figure that out, and with this dataset that I had access to, when you work out the spherical induced refractive change, that's the mean astigmatic change induced by the device. You've got 1.36, which is very close to that which was intended. So on average, the device is doing what it purports to be doing, and either the errors are due to alignment errors, or there is some inherent variability in the device such that sometimes it hits a little under, sometimes it hits a little over. DR. McCULLEY: So your recommendation CMD/IRC would be 70 percent based on literature DR. BULLIMORE: Yes. DR. McCULLEY: and your CIRC/IRC? DR. BULLIMORE: I don't think we need a recommendation for that. It's an academic issue. DR. McCULLEY: Okay. DR. BULLIMORE: If this is good, let's approve it, and let's go home. DR. McCULLEY: Okay. So you've heard that Dr. Bullimore is saying he doesn't think you need a CIRC/IRC. I'm going to take them in the order in which I saw hands. Dr. Bradley? DR. BRADLEY: It seems to me it depends a lot on what the FDA really wants from these numbers. You might think that from the patient's point of view, the bottom line is how astigmatism do they end up with, and if that's the case, it's purely the magnitude of the astigmatism, that one number, that you need. But personally, I think the FDA would be doing the patient population a huge disservice by not demanding an appropriate vector analysis, and I'll give you an example of why. As Dr. Bullimore just pointed out, there are two reasons you could end up with the wrong result. One, you corrected too much or too little astigmatism, or two, you corrected it at the wrong axis. An example he gave is if you were off by 30 degrees you had the right amount of correction, but you ere off by 30 degrees you end up with only half the correction. One solution could be to just do it again or double the amount. Well, you end up with a larger error by doing that. You don't end up you don't result in the appropriate correction, but you end up with a worse correction. So I think the only way you are going to know this is if you do the appropriate vector analysis. You will not know the source of your error, and therefore you cannot improve the procedure, and ultimately, I think the patient population needs the procedure to be optimized, and that cannot be done without vector analysis. DR. McCULLEY: Dr. Bullimore, and then Dr. Belin, then Dr. Ferris. DR. BULLIMORE: Just a point of clarification. I agree with everything Dr. Bradley said. I am not suggesting that vector analysis not be done. I am only suggesting that in terms of a guidance document, the only number that we need to give is this whatever it is called CMD/IRC value. They should do the vector analysis. That is important information for primary reviewers and the public. I am just suggesting that they don't have to meet a target value for anything else other than that. DR. McCULLEY: Before we go to Dr. Belin, could you suggest a target or be thinking about a target, and I'll come back to you in a minute. Dr. Belin? DR. BELIN: yes. I just want to bring up so that we all understand I don't disagree with anything that was just said, but we have changed the way we are now looking at the results from what percentage obtains, let's say, 90 percent, gets 90 percent of a correction to just a single percentage. So we are saying here, whether it's 70 percent or 79 percent, if we do the same thing for myopia, we say 90 percent of correction, that means half of them could be 80 and half of them could be 100. So we have changed the way DR. BULLIMORE: You have got to deal with them differently, because with myopia, you have undercorrection and overcorrection. What I am suggesting here is a number that takes into account both of these. Let's take 70 percent, so that patients on average get within 30 percent of their cylinder correction. That's the same as saying +/ 30 percent for sphere. So they're not entirely different. DR. McCULLEY: If we're doing this for you, and you're having to talk while we're discussing it, it doesn't do you much good. Did you hear what was just said well, you couldn't have, because you were talking to each other. Do you have a question that you want to interject here? DR. EYDELMAN: Yes. DR. McCULLEY: Okay. DR. EYDELMAN: Malvina Eydelman, FDA. We perfectly agree internally with Dr. Bullimore's recommendation of using CMD/IRC as the guideline, but including vector analysis for additional information without necessarily adding a specific target for it. We just had two additional questions within Dr. Bullimore's analysis whenever you are ready. DR. McCULLEY: Okay. Fred Rick? DR. FERRIS: Rick Ferris. I guess I'd like to make the same comment, perhaps, or a similar comment to what I made for the overall myopia guidance. That is, it seems to me that this question of efficacy, we ought to think of as it's like approvable and nonapprovable; it is effective, or it is not effective. Well, life to me I guess I live in a world of grays I understand that you have to do effective or not effective. Well, it is clear that any of the devices that we're talking about, if we're going to talk about a ration of 0.7, it's doing something, and the real measure of its efficacy is what it did. Show me what it did, and then I'll know how effective your machine is. It seems to me and I recognize you can't talk about previous data, but we do that all the time, at least in medical in new drugs and that is that it has to be equivalent to the drug that's on the market. So the guidance, it seems to me, is going to be a floating document. Right now, 0.7 is a reasonable goal; people have shown that you can do that much if you can't do that much, don't bring your device to us but 0.7 10 years from now might seem laughable. So it seems to me that the guidance ought to change. For right now, the 0.7 seems perfectly reasonable, and we want to see how they finally wound up, not just this fraction, but how did you do compared to how you wanted to do. And I'd say the same thing for myopia, that the overall guidelines are there for lower levels. We said, well, we'll relax it a little bit; show us what you do, and unless it's ridiculous, we're likely to say, well, that becomes the new benchmark, and then the guidelines ought to change on the drug that's available today. DR. McCULLEY: Do you disagree with the 70 percent? DR. FERRIS: No, and I say the reason for the 70 percent is that it's documented, that that is what can be done, so if you are going to come in with less than that, it doesn't seem to make a lot of sense. DR. McCULLEY: Okay. Ming, then Gary, and if anyone else wants to wave their hand, I'll write you name down and get to you. Ming? DR. WANG: Ming Wang. Food for thought for the panel and for the audience: Two patients, both with uncorrected vision. One is minus one plus 2 cylinder 180, the other is minus 6 plus 2. The point is should we be thinking about the amount of residual cylinder after correction if the corrected cylinder is, say, 0.7 or whatever, in relationship to the sphere part, giving, for example, the intrinsic technical difficulty of high degree ablation. Just as sphere part is more inaccurate, the cylindrical correction would be also inaccurate. So perhaps conceptually, not talking about percentages, but conceptually, the cylindrical correction should also be, say, a total spherical equivalent dependent concept. DR. McCULLEY: I'm not sure of the answer to that. I don't know that we've talked about it. Maybe in a moment, when we have another brief break, industry can make some comment based on their data and bigger data base. Gary? DR. RUBIN: I wanted to go back to the point that I must have missed that you raised a minute ago, that if we are only talking about CMD/IRC, that's different from saying a certain percentage of people fall within a CMD/IRC of "x". Mark, you said that it isn't different or, I'm not sure. Do I make myself clear here? DR. BULLIMORE: Yes. DR. McCULLEY: They would report is in percentage of patients. It would be in the data. DR. BULLIMORE: You could do that, but again, you are trying to end up with either two percentages or a percentage and a number. I am a lumper rather than a splitter, so I thought overall, if the device reduces 60, 70, 80 percent of astigmatism, that's what I want to know. Obviously, in terms of reviewing the PMA, looking at the variance about that is an important consideration, so if you've got an average 75 percent reduction in cylinder, but half the patients only get a half reduction, and the other half get a full reduction, that is important information, too. We seem to have enough difficulty dealing with astigmatism to begin with, so taking it to the lowest common denominator in terms of a single number was the approach I was proposing but I am open to changing that, and I am here to advise, much like everybody else is. DR. McCULLEY: Have you come up with a recommendation, or leave it soft, as the FDA has now jumped in and let us know they are happy with it soft? DR. BRADLEY: Two recommendations. First, for those marginally dyslexic members of the panel like myself, this acronym based conversation is very difficult to follow but I have tried to convert acronyms into real world vectors, and it seems to me it is possible to define the tolerable error in terms of a vector length. I am having trouble giving a number to that, but I could certainly think on that a little longer. DR. McCULLEY: I don't think we are going to get any closer right now, so rather than beat on that dog, why don't we ask that Dr. Bradley provide that additional insight to you for you to take under consideration later; okay? DR. WAXLER: That's good enough. Thanks. DR. McCULLEY: Now, are there other you had two. Did that cover both? You were just fussing about the acronyms; that was number one. DR. BRADLEY: Yes. DR. McCULLEY: Okay. Any other comments from the panel at this point? We are not necessarily done. Dr. Eydelman? DR. EYDELMAN: I just wanted to clarify whether this 70 percent CMD/IRC is the panel's recommendation and if that's applicable regardless of the range of astigmatism in the application or in the PMA. In the literature, we see much less accuracy once very small cylinders are trying to be achieved, so if the population is skewed, you will get different outcomes. DR. BULLIMORE: Yes. The 70 percent in terms of changing astigmatism divided by the preoperative that's in plain English I think is based on what's out there, and yes, you are absolutely right, if you are trying to correct 0.50 or 0.75, you're not going to hit that target rate, and I think it behooves the sponsor to go off the meaningful levels of astigmatism in order to meet that benchmark and not be frivolous in a correction of 0.50 D of cylinder. Such a huge proportion of the population has cylinders of less than 4, and most of those have cylinders of less than 2, I think there's no reason to lump it into categories right now. But this is intended as guidance and not necessarily as a line in the sand. DR. McCULLEY: I think it would be the same thing, except going the opposite way, as we said with sphere become more tolerant as you get into the lower ranges because it is harder to peg it. You had two things. DR. EYDELMAN: Yes. If I may bring you back to Point 2 in Dr. Bullimore's review, he was referring to the table attached on page 24. That table was created due to your request at the last panel in October in trying to give the residual astigmatic error some clinical meaning, and we believe we created what you recommended. At that time, I believe the idea was that even though some of the cells in this table will have very little significance, others will have greater significance, and our hope was that you would attempt at some point to fill in percentages for that table that would be acceptable. DR. McCULLEY: Dr. Bullimore? DR. BULLIMORE: Basically, what I said in my review was that this is my least favorite table, in part because the data I have seen in the past have been difficult to interpret. And I'll soften that. I think there is useful information to be gained by this table. The problem is you've got a table with the order of 25 cells in it, and if we recall recent PMAs and I can't remember them, so I'll think of a number where we have 100 or so patients, we've got 4 people in each cell on average. And then, asking us to come up with an acceptable number is not reasonable. I would be willing to put a gray zone and say nobody should fall in this area. For example, nobody with a residual I can't even think in this DR. McCULLEY: We need more time to think about that, I think, and more data to see over time. DR. BULLIMORE: Yes. DR. McCULLEY: I actually kind of like that table. DR. BULLIMORE: I do, too. DR. McCULLEY: It was my favorite table. DR. BULLIMORE: Well, it's yours. You can keep it. DR. McCULLEY: Any other comments from the panel? [No response.] DR. McCULLEY: Dr. Wang brought up a question which let me see if I can restate it is the accuracy of correcting a 2 D cylinder dependent on the degree of sphere being attempted in correction? I have not been aware of that. Is there? Mark? MR. ODRICH: Mark Odrich, VISX. Having conducted some of these trials, I'd like to suggest that the error of the measurement begins to get in the way, and when you start looking at a vector, you begin to realize that the error of the measurement for the lower amounts of astigmatism is much more significant than for the higher, so it is actually counterintuitive. It is Heisenberg [ph.] rolling in his grave. What actually occurs is that when you have higher degrees of cylinder, it is more accurately portrayed, and there is a lower amount of noise. The paper that is always cited is Zadnik [ph.] in 1992, but the problem with that is that she limited the people coming into her repeatability of ocular measurements to somewhere between 5 and 6 D, and I don't quite remember where, with smaller amounts of cylinder. But we started to look, as a secondary role through Columbia University, at repeatability of measurements as you get higher, and in fact this discussion of myopia and astigmatism, both efficacy criteria are fraught with problems once you don't take into account the errors of the measurements. So I think they are actually interrelated issues, but I think the interesting thing is that actually, the magnitude is easier to correct as it gets higher, because your measuring ability is more reproducible, so DR. McCULLEY: That wasn't the issue, Mark. MR. ODRICH: I'm saying technically, that that affects our technical ability. DR. McCULLEY: That wasn't the issue. In correcting the same degree of cylinder pick a 250 that should be fairly easy to get the axis MR. ODRICH: A 6 and a 1. DR. McCULLEY: A 1 plus 2.50, 6 plus 2.50. MR. ODRICH: That would be laser by laser, I would imagine, but I can tell you for the VISX it is no more difficult. There is nothing technically more difficult. But what is more difficult is that measurement, and that's why I'm saying to you DR. McCULLEY: What's more difficult is the 0.50 versus the 2.50. MR. ODRICH: It's the quantification as a vector. That's what the clinicians are doing. So I think it does answer that question. DR. McCULLEY: Okay. Dr. Belin? DR. BELIN: i was going to comment on something else, but just on that last comment, Mark did imply that that's going to be laser dependent, and again, I don't want to make this laser or procedure dependent, but it clearly will be dependent, because different lasers go to different ways of correcting cylinder at a certain level. A plano minus 3 is treated differently than a minus three minus three the way you correct cylinder. But the fact that a company for a procedure decides to treat it differently doesn't mean we are going to set different efficacy variables for it. I wanted to make a comment earlier that I think vector analysis my specialty, you can talk about all day I think that when push comes to shove, you are going to end up finding that at about a 70 percent CMD/IRC is going to turn out to be roughly the same as about a 50 percent reduction vector analysis. You will find out that it will be about the same. I'll let Dr. Bradley do that. DR. BULLIMORE: I think it actually goes the other way. I think you'll find that the surgically induced refractive change is actually going to be bigger than the change in the absolute cylinder values. DR. McCULLEY: It was in your example. DR. BULLIMORE: Yes. DR. McCULLEY: Now that we're all confused again so Dr. Bradley is going to work on this more to make further recommendation to you. Does the panel have any further comments on this particular item? [No response.] DR. McCULLEY: Do you have any further questions for us on this item? [No response.] DR. McCULLEY: Then, we can go on to the last item. Doyle, I'm sorry, I didn't see you. Dr. Stulting. DR. STULTING: Doyle Stulting, Emory University. To address a point that was raised that you asked about, which was the effect of spherical correction on astigmatic change, we just completed a multivariate analysis and in fact submitted the paper to the FDA, but apparently, it didn't get distributed. And there is an interaction term of the sphere on the astigmatic correction, so indeed it does have an effect, at least in our dataset. DR. BULLIMORE: Which way does it go? DR. STULTING: It increases the variability. It was an ANOVA, analysis of variation, so the sphere increases the variability of the astigmatic correction. DR. McCULLEY: As you go up. DR. STULTING: Yes. DR. McCULLEY: So I guess we did not have that data, and it does then support the issue that, at least in terms of discussions that I remember from the panel, brings up a new issue. DR. WAXLER: Correct. That paper that Doyle spoke about came to late for me to distribute to hardly anybody, actually. DR. McCULLEY: Okay. So that's something that you will take under consideration and advisement for the future. DR. WAXLER: Your pleasure. DR. McCULLEY: Next point? DR. WAXLER: Are you ready to go to the next issue? DR. McCULLEY: Yes. DR. WAXLER: The next issue, Dr. Stark will speak to, and that is can the amount of residual corneal tissue in refractive surgery be reduced to 250 microns without an unreasonable risk of endothelial damage that is, endothelial cell loss or ectasia. If yes, then, what should be the new amount of residual tissue that should remain to provide a reasonable assurance of safety? What precautions should be taken to minimize compromising endothelial integrity? 