Introduction of Issues for Excimer Laser Guidance

Скачать 330.02 Kb.
НазваниеIntroduction of Issues for Excimer Laser Guidance
Размер330.02 Kb.
1   2   3   4   5   6   7   8

And I was trying to think  maybe you could do some sort of test of the slope, but I am not sure that that's a good thing to do, either, because for one thing, the larger the study, the more likely you are to have a statistically significantly different than zero slope, and that's not the issue here.

I think the issue is that you don't want  I don't know where the 1.0 D came from, and we can argue about the number  but you don't want to be losing 1.0 D or more of efficacy of the procedure, at least not without knowing that it exists.

And I did take some exception to a comment that was made that, oh, we don't follow these people for after a year, and it works out okay. Well, yes, it works out okay because we don't know whether it works out okay or not. So we don't know what's going on later.

So at some point, somebody is going to do a long term follow up of these patients, because obviously, from a public health point of view, that is important. It may be, though, that looking at the slope of this difference is an important direction to go to say that you are reaching stability and that that slope ought to be less than one tenth of a diopter, or that slope out to be less than two tenths of diopter. If you still have a drop off of 0.25 D every three months, that sounds pretty concerning to me.

DR. McCULLEY: And I guess that's what Morris was asking  what would that slope be. But one of the problems with the mean and why we need or wanted the other information on the individual patient is that the mean can be screwed up by somebody going a +1.0 and somebody else going a  1.0.

DR. FERRIS: Oh, absolutely. For efficacy, you absolutely have to look at those individual patients, but for this question, I think it may be important to look at the mean to get some idea of whether there is an on the average recovery  the sort of thing that was seen in PERK, where on the average, there tended to be a drift in one direction or another  and I think the only way to assess that is looking at means.

DR. McCULLEY: Well, we do that. The question is how much of a change in mean from one 3 month point to the next is acceptable or unacceptable  where is the cut off  because we do look at the means.

DR. FERRIS: I understand.

DR. McCULLEY: What would you suggest would be a  

DR. FERRIS: Well, I am a little bit concerned about doing this by the seat of my pants  

DR. McCULLEY: Yes, okay.

DR. FERRIS:   but someone between one tenth and two tenths of a diopter. If there is still that level of drop off over a 3 month period, and it is consistent  it is different when it is bouncing around  but when you look at these intervals, and there is a consistent drop off in that range, that would make me feel much more concerned than if it were bouncing around, if it went down, and then it went up, and then it went back down again. That third back down is probably a combination of errors between the two previous ones. It is the trend over time, and maybe that's the issue, that the average trend over time can't be more than one tenth or two tenths of a diopter.

DR. McCULLEY: You've got to consider what points you have on the curve.


DR. McCULLEY: I mean, they are going to have 3 months and 6 months.

DR. FERRIS: Well, if you just have 6 months  

DR. McCULLEY: Well, no  we may have 12  

DR. FERRIS: Yes. I think you need the 12 for this particular calculation. Even that would be  

DR. McCULLEY: Well, we haven't required that in the past.

Dr. Matoba?

DR. MATOBA: I may not understand the issue here, but Dr. Bullimore said that reproducibility of a refraction is about 0.50 D to 0.75 D, but in that case, it is random, and it is as likely to go positive as negative; isn't that correct?


DR. MATOBA: So then, that variability is as likely to cancel out as to add to the postop change in refraction. So with that effect, Dr. Ferris, your comments?

DR. FERRIS: No. For the slope, you are right, that on average, it averages out, and you have a mean change of zero. When you are saying you have to have 95 percent within 1.0 D or 95 percent within 0.50 D, that's when the random error comes into play, because a certain number of those are due to chance, not due to treatment effect. So it depends which one of those two pieces you are looking at, whether the error is going to be important or not.

DR. McCULLEY: Dr. Bullimore?

DR. BULLIMORE: I think the data that you want is out there, Morris, and it's just a matter of whether through a homework assignment or something  we go to the literature or look at past PMAs, because you are allowed to do that, we are not  and see what the standard of care is, and what has the bar been set at by previous studies and previous procedures.

DR. McCULLEY: I don't think they are allowed to look at past PMA data in developing guidelines.

Dr. Wang?

DR. WANG: I just want to bring in one historical perspective. If you look at the published PERK study, over 20 to 25 years, there is expected in the 20th century to be about a 1.5 to 2.0 D hyperopic shift, so if you divide by 20 years, you have about 0.1 D per year, or 0.05 for 6 months. The point of this is that I understand Jim's point that it is impractical to look too long, but err on the conservative side, because as we found out, PRK has much smaller slope and has been drifting.

DR. McCULLEY: Other questions, comments?

Dr. Stulting?

DR. STULTING: I am sorry to be a little bit out of order, but you were struggling with some data about refractive change, and we spent a good bit of time looking at this, and I have some real data to offer, and I can present it to you in public forum so you can reference it.

The first was from the multivariate study of our LASIK data. Essentially, what we did was to separate out the refraction dependent or the attempted correction dependent component of variance from that which was apparently refractive independent. In other words, we looked at the Y intercept. And that number was 0.50 D for standard deviation.

There is another dataset that we looked at as well, and that was contributed to us by Peterson et al., and it was from the Nidek study. It was not the operated eyes, but the unoperated eyes, that were measured 3 to 6 months apart, and that was presented at ARVO.

We did a similar analysis on that, looking at the standard of deviation, and that was 0.4 D.

So those are two independent datasets, one on operated myopic eyes, the other on unoperated myopic eyes, with similar ranges of refraction, and the numbers for the standard deviation were in very good agreement, 0.4 and 0.5. So it turns out that looking at +/  1.0 D is the 95 percent confidence interval, roughly, based on both of those studies. If you look at +/  0.50 D, that means you are going to disqualify patients for refractive surgical studies, a third of them, just based on chance variation in their refraction alone.

Similar numbers that we got for variance in astigmatic magnitude were 0.3 D standard deviation. So you may want to consider those numbers for your deliberations.

DR. McCULLEY: Is that published now, Doyle?

DR. STULTING: The Peterson numbers were not analyzed by them; we analyzed them with their permission. And those data have been submitted as well, and they were in the same information that Morris mentioned, so the FDA got them a couple of weeks ago.

DR. McCULLEY: That is where I was leading.

DR. STULTING: Yes, sir.

DR. McCULLEY: Thank you.

Having reopened the floor to Dr. Stulting, in fairness, is there anyone else in the audience who would like to approach the podium for additional comments?

[No response.]

DR. McCULLEY: Other panelists, questions, comments?

Dr. Belin?

DR. BELIN: Just again  I think this is what Dr. Matoba was mentioning  I believe Dr. Stulting was again talking about error of individual refractions, not mean population drift, and those are two different things.

DR. McCULLEY: Yes; that was what I understood as well.

Is that correct, Doyle? A nod from the back. Thank you. Let the record show he nodded.

Dr. Ferris?

DR. FERRIS: That is that the standard deviation of 5 percent of these repeat refractions would be more than a diopter and 5 percent less than a diopter. This isn't standard error of the mean. This is  

DR. McCULLEY: He said standard deviation.

DR. FERRIS: I understand. I just want to make sure. So that's saying that 5 percent would be more than a diopter.

DR. McCULLEY: Correct. That's my understanding, and let the record show Dr. Stulting is nodding again.

DR. FERRIS: Or, no  2.5 percent more and 2.5 percent less. Five percent outside  of course, the mean is going to be stable. But he's saying that by chance, you are going to find 2.5 percent 1.0 diopter above and 2.5 percent 1.0 diopter below  no?

DR. McCULLEY: Dr. Stulting is nodding his head once again.

DR. FERRIS: Well, there is somebody else shaking their head, so that's  

DR. McCULLEY: We were talking about Dr. Stulting's data.

DR. FERRIS: Oh, all right.

DR. McCULLEY: Dr. Odrich?

DR. ODRICH: Mark Odrich, VISX.

That's assuming no cretosis [ph.]; that's assuming a bell shaped curve. And that's the point  these are not always distributed in bell shaped curves. So there are some other variables involved. Certainly it is 2.5 above and below with no cretosis. But again, that's the error ratio.

DR. McCULLEY: I could ask him to define "cretosis," but I think I can assume what it is  I'm not sure.

DR. STARK: Jim, why don't you define it for those who might not know?

DR. McCULLEY: No, no, no. I think that means that these are not shaped like a bell, that they have a little, weird squiggly at the end, or an upturn or something  that would be my guess. That'll teach me to try to be a smart alek this late in the day.

Any other questions or comments from the panel?

[No response.]

DR. WAXLER: Are you ready for the second part, or did you answer the second part?

DR. McCULLEY: I'm sorry, I forgot about that. Go ahead. I just forgot about it.

DR. WAXLER: Okay. This has to do with when you  according to the guidance, where we say you measure at 3 months and you measure again at 6 months, and if you find that there is no mean difference, you then take that to understand that the point of stability is at 3 months, or it is at the middle or it is at 6 months  so I would like to hear some discussion; that would be helpful to us.

DR. McCULLEY: Well, it would be helpful for me to know from where you are coming with that question. You have to have the 6 month to define the 3 month. So what is the  

DR. BULLIMORE: I think you have to say that stability has been established at 6 months.

DR. McCULLEY: But based on establishing at 6, it occurred at 3.

DR. BULLIMORE: Well, I chose my words very carefully  you need the 6 month data point to say that, so stability has been established at 6 months.

DR. McCULLEY: Right. And in retrospect, occurred at 3. What's the point?

DR. WAXLER: The point of this exercise  and it may be pointless  but the point of this exercise is when do you count the confirmatory next evaluation? Do you start counting from the 3 month point and get another confirmatory point, or do you get a point at 7 months or 8 months  

DR. McCULLEY: Do you mean if you want to confirm what you think you have confirmed at 6, when do you have to do it the next time?

DR. WAXLER: Well, the question is do you go to the next longer time interval. Do you have to go out another 3 months in order to confirm what you found as no difference between 3 and 6.

DR. McCULLEY: All right. At 6 months, by our definitions, you have stability established based on the values at 3 versus 6 months. If you want to confirm that further, at what point should the next exam take place  would one more month be acceptable, or would one need to go to a full 3 month from the 6 month point, in your example?

DR. WAXLER: And that issue has come to us, and that's why I am asking you the question. What do you think?

DR. McCULLEY: I guess my question would be why is it being asked if you've established stability.

DR. ROSENTHAL: This is Dr. Rosenthal. I'll tell you why it's being asked. Firstly, it has a labeling issue, because in the labeling they say "point of stability." The other thing is that in many of the tables that we request, it says "point of stability." That is what we use in the Summary of Safety and Efficacy, the "point of stability." And is the point of stability at 3 months, or is the point of stability at 6 months?

I intuitively felt that it was at 6 months, but they established between 3 and 6, and therefore, the data should be shown at 6 months  but maybe I am wrong.

DR. McCULLEY: I don't know  wasn't it Reagan who got in trouble for saying "point in time"?

DR. SUGAR: But when you define it as a change of less than 1.0 D between 3 and 6 months, it is therefore stable in that interval; it became stable at the beginning of that interval. And if you want to confirm that, you have to look at another 3 months, because your definition includes 3 months.

DR. MACSAI: Right.

DR. McCULLEY: I agree. Stability was reached at 3 months, but you had to have 6 months to establish that it did.

DR. MACSAI: And 9 months to confirm.

DR. ROSENTHAL: This is Dr. Rosenthal. So, when you ask for a table at "point of stability," do you pick 3 months or 6 months?

DR. SUGAR: Three months.


DR. SUGAR: Three months is the point at which it became stable.

DR. McCULLEY: But you don't know that unless you have 6 months. That's the problem with "point."

Dr. Macsai?

DR. MACSAI: If you go along with the point that Morris brought up earlier about the slope, then, it is a little bit of a moot discussion, because you need the 3, 6, and 9 to calculate if you are talking about the mean difference. So if you are going to redefine the terms to be dependent on the slope of the mean differences, you need at least two points  in other words, three measuring points, 3, 6, 9  to calculate that slope  and hopefully, longer.

DR. McCULLEY: This is a labeling issue. Rather than begging off the point, to get rid of the point and say stability was established at 6 months  it wasn't established until 6 months  Rick, do you have a better play on words?

DR. FERRIS: This is Rick Ferris.

The stability is driving me crazy, because I think, as Dr. Wang pointed out, it could be inherently unstable over a 20 year period, and we were not able to measure it over a one year period. So to say that you have established stability seems pretty grandiose. What you have established is that there isn't any gross instability.

DR. McCULLEY: So what you're saying is we should never have used the word "stability"; we should have said that it was determined by 6 months that there was no greater than a "blank" change in refractive error.

DR. FERRIS: I'm not sure that labeling change will fly.

DR. McCULLEY: Anyone else?

DR. ROSENTHAL: May I just ask the panel, therefore, you say the stability has been established at 3 months. If you want to look at the day  say you were a doctor who wanted to look at the data that the agency has put in the public arena  would you rather see the 3 month data or the 6 month data if you had your choice to see one or the other?

DR. McCULLEY: The gut response from me is to say the longer out, the better, but that's gut  that's my gut.

DR. ROSENTHAL: Yes, okay. That may be why I chose 6 months, because my gut response was you want to see it the further out it is.

DR. BELIN: If this is your graph, although over a 3  to 6 month period, you may have a slope that you define as acceptable, you cannot say that at the 3 month period, that slope is no longer acceptable. You need two points  if you want to have anything less than the 6 month, then you have to do an exam at 4 months or at 5 months. Between 3 and 6, you met criteria for whatever you define as stability, but from 3 to 4, you haven't met it unless you have a 4 month point.

DR. McCULLEY: Okay. Well, I don't know, that made sense to me. Basically, if you look at that curve, you are saying that stability was truly reached there, as he drew it, at 4 months. But you don't know it; so it took the 6 month to know.

DR. McCULLEY: Other comments? I hope we haven't been too helpful.

DR. WAXLER: Thank you very much.

DR. McCULLEY: Any other questions, comments?

Morris, do you have anything else?


DR. McCULLEY: Dr. Rosenthal, do you have anything else?

DR. ROSENTHAL: No. Thank you very much.

DR. McCULLEY: Let me remind everyone that tomorrow we begin at 9 a.m. In your packet is material that has been provided by Dr. Michael Lemp, who is going to speak in the public session. You might want to look it over. It was in the packet that we got on arrival today. And be sure to look through the material that Donna provided to us as well.

Sara, do you have anything before we adjourn?

MS. THORNTON: I just want to thank the three presenters who made the gargantuan effort to answer Morris' questions, and we would like to see you tomorrow, fresh.

DR. McCULLEY: We stand adjourned.

[Whereupon, at 5:18 p.m., the proceedings were adjourned, to reconvene on Friday, October 23, 1998, at 9 o'clock a.m.



507 C Street, N.E.

Washington, D.C. 20002

(202) 546 6666
1   2   3   4   5   6   7   8


Introduction of Issues for Excimer Laser Guidance iconInside the States North Carolina Appraisal Board Issues Guidance on bpos [Posted Sept. 22, 2010] Around the Industry

Introduction of Issues for Excimer Laser Guidance iconIntroduction to the Issues: Genetically modified organisms

Introduction of Issues for Excimer Laser Guidance iconLaser-Ablation Rates Measured Using x-ray Laser Transmission

Introduction of Issues for Excimer Laser Guidance iconFinancial issues, about accommodation, employment, health and welfare, visa or immigration issues

Introduction of Issues for Excimer Laser Guidance iconProfessional: Laser & Skin Surgery Center of New York

Introduction of Issues for Excimer Laser Guidance iconProfessional: Laser & Skin Surgery Center of New York

Introduction of Issues for Excimer Laser Guidance iconLaser cleaning of stone materials: an overview of current research

Introduction of Issues for Excimer Laser Guidance iconVersion for paid-up members of the uk and North American rc laser Organizations

Introduction of Issues for Excimer Laser Guidance iconUltrashort laser pulse phenomena : fundamentals, techniques, and applications on

Introduction of Issues for Excimer Laser Guidance iconSimilarity for ultra-relativistic laser plasmas and the optimal acceleration regime

Разместите кнопку на своём сайте:

База данных защищена авторским правом © 2014
обратиться к администрации
Главная страница