ISOPT Clinical Hot Topic Panel Discussion on Uveitis and Inflammation

Ron Neumann: We are very happy to continue now with the Uveitis Hot Topics. We will start with what seems to be one of the hottest topics, not today, but over a long period of time: how we are measuring activity and severity of uveitis and how we are able to define it in clinical studies.

Lately, there were some remarks made about the role of flare in assessing uveitis severity and activity. Traditionally cells were considered the major component of uveitis activity and flare was considered a byproduct of longstanding damage to blood vessels. Currently, changing levels of flare seems to become more relevant in the assessment of uveitis activity. Having said that, Quan, as a leader in uveitis clinical studies can you share some thoughts with us about how to conduct proper clinical trials for uveitis and whether we can consider flare as an outcome.

Quan Dong Nguyen: Thank you, Ron and Sarah, for putting together this wonderful meeting in the Holy Land. It is wonderful.

We know that there have been several uveitis trials that have started within the last 17 years since I left fellowship, and very few of them had made it far because of the number of challenges we have.

There are three different aspects that are very important in planning and conducting clinical trials: one is the target population condition. Unlike age-related macular degeneration (AMD), we don't have thousands of patients to enroll into a trial, so our target population is small.

The second point is inclusion criteria which are also very difficult because of the dispute whether uveitis is one entity. For example, we can consider the vast differences between VKH and sarcoid. Is it proper to consider both entities as candidates in the same study? Thirdly, outcome measures are still under discussion. Regulatory bodies such as the FDA only accept certain endpoints so far, because we as a community, have not been able to demonstrate what are the appropriate parameters that should be followed over short and long periods of time.

Thus, challenges exist both in the design and the conduct of trials: the different etiologies of disease mechanism that may present with similar clinical manifestations. One etiology may be different than the other. Infection may also trigger and stimulate autoimmune disease. Autoimmune diseases have such wide variety of etiologies and natural history that are not well known.

Therefore, the question we need to consider is whether there will be ever be a clinical trial design that will satisfy a group of uveitis experts like us. Probably if one asks any of our five distinguished colleagues sitting here today, each one of them would have a specific goal in mind of what the trial should look like, and that continues to be a challenge.

What is missing from our group here? First of all, we need to have a better understanding of disease mechanism in different entities for example, sarcoid. Is it the same as VKH? How is it different from Adamantiades-Behçet disease? As clinician scientists, we do not yet have a very clear understanding of the disease pathophysiology.

Second, there's also a need to have better agreements among uveitis specialists, because if one were to ask the group from SUN, the group from IUSG, AUS or IOIS, each of them may not always come up with the same definition for disease, and that's just our fault. Therefore, before we come to the regulatory authority, whether the FDA or the EMA, we need to come up with better definitions and agreements for endpoints.

Ron just referred to the flare meter, and why we have not incorporated it in any of our clinical trials thus far. It is because we have not yet demonstrated reproducibility and utility in clinical activities.

Also, outcome measures in uveitis clinical trials have been very difficult. Vitreous haze, which is something that the EMA and the FDA have been using, has not always been well accepted by our community. We know that visual acuity is not always an accurate reflection of uveitic activity. Many of us feel that macular thickness in the setting of edema, is an anatomic endpoint, and not always a true representation of inflammation.

What are other possible parameters? One of the parameters we're missing, for example, is macular sensitivity. Why can't we incorporate retinal sensitivity into uveitis trials? One of our colleagues from Stanford, Yasir Sepah, has published on this topic to describe how macular sensitivity and fixation pattern have good correlation and can be used as potential markers in patients with maculopathy secondary to uveitis, for example.

Other parameters that we can consider include reading speed and visual field, which may be considered as endpoints in appropriate settings.

Thus, the gap in planning and conducting uveitis studies is to develop trials for real-world patients. We may need to step beyond patients that are easier to recruit in order to finish enrollment for a trial, but rather those who have unmet needs. Thus, we should not conduct trials that are just to find a role for a new pharmacologic agent. Rather, we should define carefully the clinical and the etiologic diagnosis in the trials.

In addition, we have different devices that can and should be employed in trials. The laser flare meter is certainly a possibility. The flare meter has been promoted by many colleagues, especially those in Asia and Europe, but have not yet been incorporated into clinical trials.

In summary, there are challenges in the design and implementation of clinical trials in uveitis, which is owed to the nature and the presentation of disease, and the need to measure the structural and functional endpoints. Thus far, the various recent and ongoing clinical trials are meeting these challenges by two ways. One, they try to evaluate the new treatment or treatments that have not been used in patients, and two, they try to employ innovative outcomes.

We have been handicapped for a long time. I left fellowship in 2001. Since then, over the last 17 years, I have seen a number of clinical trials, and so far there are very few that have made it all the way to the end to be approved because a number of them have failed near the end of the chain.

The session just before us discussed about geographic atrophy (GA). Retina colleagues are having difficulty getting drugs to be approved for GA, because perhaps the endpoints have not been well defined, and the disease mechanisms have not been fully elucidated. For uveitis, I believe it's also a problem, too, because we don't have endpoint or endpoints that all of us are completely satisfied. Hopefully, during the next 45 minutes, we'll be able to reach consensus!

Ron Neumann: Thank you Quan, you raised a very interesting point that is always in front of us. I mean, when does a study fail, did it fail because the actual drug did not perform, or did it fail because we did not design it correctly, with all the layers of clinical trials, including endpoints, of course.

So, Marc, why don't you talk to us about the flare? And then we will discuss together the various ways of measuring an effect.

Marc D. de Smet: I suggested this topic, because the measurement of cells and flare is a major issue when you're looking at designing or running clinical studies in uveitis. For one of the drugs that was mentioned earlier, the trials failed in part because the measurement of intraocular inflammation could not be sufficiently well assessed.

Recently, the SATURN trial used a novel flare measurement in the posterior segment using both clinical observation and photographic documentation. And there also, a significant discrepancy was observed between the two methods. Ultimately, if you're trying to do studies, you want to be able to document in an objective, reproducible manner meaningful clinical parameters. It is a significant issue for all of us interested in achieving better care in uveitis patients.

In uveitis, we are dealing with (to use a biblical term) the Tower of Babel and this, I think, was very well outlined by Quan. We treat multiple diseases, each of which are different in etiology, course, and outcomes. They have therefore different endpoints regarding progression, and response to treatment. Furthermore, we assess these diseases, using non-validated scales—where each incremental step may or may not represent a clinically significant worsening of the inflammation. In all its years of existence, the flare/cell scale which goes back to Hogan and Kimura, was never validated against disease. It was used from its inception as a qualitative assessment.

We use it because it helps us to evaluate uveitis severity. It is also oddly weighted with most uveitis entities falling in the 1 to 2 range and very few reaching a higher level.

There are few grading steps. Having only four levels limits the ability to discern fine changes in inflammation. Hence, adjudicating finer changes in inflammation. In studies, we are often left evaluating an all or none response. Each uveitis expert also interprets the scale in his own way, using a slit beam and intensity that can be variable from one exam to the next.

The SUN Working Group attempted to standardize our approach starting by defining what can be considered as active or inactive uveitis. What should be considered as an improvement in activity, or what defines a remission? Also it took a good look at previous scales and provided further standardization.

For cells, the consensus was to measure within a 1 mm by 1 mm max intensity slit beam, the number of cells present. Trace was 1 to 5 cells, 1+ from 6 to 15. Counting these numbers is possible. When you get to 25, 50 or more, it becomes impossible to count that number, particularly since they move around. In fact, to be perfectly honest, beyond 10, I'm uncomfortable giving a number and I start making a mental estimate. Would it not be better to have a “comparative scale” or even take a high-resolution picture?

With regards to flare, our definition is vague, particularly between none and faint (0 and 1+). The difference between faint and moderate is also unclear as for the latter, iris and lens details are clearly visible. Hogan and Kimura were more specific in terms of their description.

Yeo et al. came out in 2016 with a study in which they asked 65 uveitis experts how they use the SUN scale. 69% of these experts did use a slit beam that is 1 by 1 mm, but the remaining experts were using a larger beam.

If you look at interobserver variations and I guess John could tell us a lot more about it, exact matches for anterior chamber cells are present in only 58% of cases. But for vitreous haze, a measurement that could be considered more difficult to perform as it requires the use of an indirect ophthalmoscope, we get better concordance between observers around 69%.

I believe the reason for this concordance is that to quantitate vitreous haze, we use a photographic scale, while for anterior chamber cells we use a physician's judgment. The use of a photographic grading scale insures a better standardization of the assessment and reduces interobserver variability.

We're not the only ones using scales to assess severity of ocular inflammation. Eaton published in the J Ocul Pharmacology and Therapeutics in 2017, the “SPOT” system used by ophthalmic veterinarians when performing preclinical toxicity assessments. During their evaluation, they quantitate not only intraocular inflammation, using a more elaborate description of flare, they also look at conjunctival hyperemia and chemosis. It is interesting to note that we also see in many cases of the anterior segment uveitis changes in the conjunctiva, and yet we don't grade this at all. A more elaborate description of what each flare grade represents improves consistency, but not as much as can be achieved by comparing photographs of specific grades of flare.

The vitreous haze scale makes use of such images. The original grading scale by Nussenblatt was created by defocusing the view of the retina, making the image progressively blurrier. As mentioned earlier, despite its simplicity, it provides a relatively consistent interobserver assessment of haze, but as the anterior cell/flare grading scale, it lacks discriminatory power. A newer scale developed in Miami by Janet Davis uses nine levels of vitreous haze instead of 4. This scale still needs to be “field tested,” but I suspect this will be a step forward from the Nussenblatt scale.

Flare can also be quantitated using technology. The laser flare photometer was introduced nearly two decades ago for this purpose. Technology in principle can provide us with better discriminatory power using a continuous scale and better interobserver reproducibility. Agarwal et al., attempted to correlate laser flare photometry with the SUN classification. From an analysis of 50 eyes, they observed that laser/flare photometry provided a continuous variable and was better able to distinguish levels of inflammation. The discriminatory power of the SUN classification was limited, but could be improved by adding an intermediate step between 0 and 1. Without the introduction of this intermediate step, there was considerable overlap in the flare photometry measurements between patients assigned to grades O and 1. However, in their study, the state of anterior chamber, in particular the presence of a cataract was not taken into account, and yet it is known to influence basal flare values. Hence, if we wish to use flare photometry in clinical studies, also for this device, we would have to establish standards for the measurement and interpretation of results. We would have to define basal flare photometry levels with regards to age, the presence of cataract, and since chronic flare can lead to a persistent high value over time, some standardization in the number of measurements needed to determine baseline values.

Would other technologies be appropriate? One drawback of laser/flare photometry is the need for a dedicated machine. Anterior segment swept-source optical coherence tomography (OCT) also provides a means of assessing both cells and flare in the anterior chamber. If existing OCT machines could be adapted by use of an add-on lens to measure flare in the anterior chamber, this would be a significant advantage for many clinicians, and facilitate the conduct of clinical trials. An exploratory paper using an anterior swept source OCT gave striking results with good correlation between the measurements and the SUN clinical grades. There was also little overlap in the OCT values from one clinical grade to the next indicating that OCT may be better at assessing finer differences in disease severity as compared to laser/flare photometry, where overlap between grades was present.

So, what would be the ideal way to look at cell and flare? Well, we need a measurement that is robust, reproducible, correlated with disease severity. We need a simple test that we can use in many different settings, and that is not exorbitant in cost.

Devices would be nice, but one should not dispel a clinician's ability to score inflammation. As indicated above, validation of the scoring system is important, and the use of reference images. I was made recently aware of a UK digital patient record which required ophthalmologists to score diabetic retinopathy based on the same photographic images that were used in the ETDRS studies. Despite scoring while seeing patients in busy practices, the scores were similar to those achieved using independent graders.

In uveitis, we have scales for both the anterior segment as well as a posterior segment, but they're not ideal right now. Not as explicit or as well tested as those for diabetic retinopathy, but with a little effort, we can get there. While dedicated machines may be nice, we can also just use a clinical evaluation, against photographic comparators. Comparators that provide significant distinctive steps between each severity level. May be this is where technology can help us best to establish the ideal scales we as clinicians can use in our practices and clinical studies. Thanks.

Ron Neumann: I would like to bring it now to all our discussants. Let us talk about the whole story of flare in the anterior chamber. And I wonder, just before we go into any depth, how many of you consider flare as one of the parameters to follow when deciding to change therapy or in any way even guide your therapy? Manfred, please.

Manfred Zierhut: I agree with people like Janet Davis and some Europeans that the flare is probably a very important parameter to check in juvenile idiopathic arthritis (JIA), because here, if you see zero cells and a lot of flare, I would increase the aggressiveness of my treatment, but only in JIA. Actually, in other types of uveitis I would definitely not.

So for me, flare definitely plays a role, in JIA associated anterior uveitis. There are some problems with the SUN criteria to define activity and inactivity. If you have 1+ cell in an eye, with a pressure of 30, then this is highly active, and actually suggests a diagnosis of a virally induced anterior uveitis. So, in these situations, the pressure is a sign of activity, not so much the amount of cells.

In contrast to this situation: If you have a +1 cell finding in Fuchs' uveitis, you may say “Well, wonderful. See you in a year.”

I think there are some very good studies. We recently finished a very successful one recruiting 41 intermediate uveitis only. So why not using this example? I agree: that seems to be quite a small group. In our clinic we see ∼1,000 new uveitis patients per year. Approximately 60% have intermediate uveitis. So it's not so rare and it should provide sufficient number to include into studies.

One of the things I want to comment to Marc: it's very nice to use scales for vitreous haze and cells. The major problem I see is measuring exactly your view to the posterior segment. And this does not portray the complete vitreous inflammation, especially not in intermediate uveitis, where a lot of inflammation happens more peripheral. So, we should have measurements which allow you to count the whole vitreous inflammation. In addition, the amount of snowballs for me do not reflect high activity. I may see patients probably with 20/20 once a year, they have the same snowballs, so I don't treat them, definitely not. But it's very important for reaching the correct diagnosis.

So, I would encourage people from the industry to be a little bit more patient. Work with the good people, and not necessarily only in Europe and America, definitely not. There are fantastic activities going on in Asia and other places. As a second point I may suggest: include subgroups.

To compare a birdshot with intermediate uveitis is ridiculous. You are having two different animals, and no surprise that you will get different outcomes. I must agree that the Humira studies were probably the best ones we had officially. Some of the others were a complete disaster, and their design ruined the results initially.

Ron Neumann: John, what is your view and let us talk specifically about, what happens when flare changes? How—do you take that into consideration?

John H. Kempen: I don't have a flare meter, so I must rely upon clinical evaluation.

Ron Neumann: And neither most of us, I think.

John H. Kempen: Yes. When we did the agreement study with the flare grading scale, one of the things that I saw is that if you're an enterprising trainee, you could just grade it 1+ every time, you'll nearly always be within one step of the other clinician, because the large majority of cases would be grade 0, 1+ or 2+. Since I started practicing in Africa, I see more 3+ cases, but not too many.

The point of this anecdote is that I don't think this scale is precise enough to make much judgment because +2 is really a huge amount of flare, and +1 is much less and improving from 2+ to 0 is probably nearly impossible. I think there must be some better way of measuring flare than this scale if you're going to use it for clinical practice, especially since cases don't necessarily go to 0 when they are quiet.

Ron Neumann: So, your point is that we're actually missing a good machine that would allow us to make a an accurate measurement that would be affordable to most specialists, not something that is limited to only two, three places in the world?

John H. Kempen: I don't know much about the machine, because I've never had it. The people who have the machine seem to think that it's very useful in certain circumstances, like Manfred was describing.

Quan Dong Nguyen: I think that has been the problem, because certainly as far as I know, there's only one company in the world which is currently manufacturing a flare meter that is available for commercial use, and that's Kowa in Japan.

In the United States, there is less than a dozen, probably less than 6 clinical centers that have the flare meters from Kowa. We are fortunate to have one at the Byers Eye Institute at Stanford. It has been difficult for us to be able to get others to know about it because very few colleagues, at least in the United States, are familiar with it.

I am beginning to incorporate the flare meter and flare measurement into my clinical practice and in the uveitis clinical trials that we are launching across the United States. But it's hard because so few people have the device. As Marc said, it's about $US50,000 to 60,000. And one of the things people always ask is: how are we going to get this measurement reimbursed if we purchase the flare meter. If we get an OCT, at least we can charge for it. If we get a flare meter, right now, we don't have a way to recover the cost. Therefore, flare meters are mainly at academic centers in the US.

Therefore, getting the community to get accustomed to the flare meter has been a challenge for us because the majority do not have the device and thus cannot appreciate what it can do and what information it can give and how useful it is.

Michal Kramer: It would be nice to have a flare meter for clinical trials, and maybe someday in clinical practice (when it is more affordable). However, I think we are missing the main issue, which in my opinion, building an activity scale, like the rheumatologists have, where flare would be one part of it.

For every “uveitic disease” entity there are different weights for different signs, depending on their relative risk in causing visual impairment. I think we should try to come up with a numerical scale, like other disciplines in medicine (for example, Glasgow coma scale).

In this setting, a flare meter would be a more accurate tool, to provide an exact numerical data, rather than the variable inter-observer evaluation.

An important end point that we also currently use is visual acuity, and it may sometimes be misleading. A dense vitritis just in front of the posterior pole correlates with visual acuity (VA), however a focus of retinitis that does not yet involve the fovea may have a higher risk effect of vision loss but will look “better” if VA will be the sole parameter.

Quan Dong Nguyen: Michal, do you believe that, as Manfred was saying, the flare assessment (and hence the flare meter) maybe relevant only for certain subsets of uveitis?

Michal Kramer: Yes.

Quan Dong Nguyen: I agree with the fact that flare assessment/flare meter is very appropriate for anterior uveitis, not just for JIA, but also for other types of anterior uveitis. Our team at Byers and Stanford is looking at the role of flare assessment and flare meter in other types such as pan-uveitis.

Michal Kramer: But I think that, for instance, the visual acuity really does not give the whole picture. I mean, if we have a dense vitreous opacity which lies down there, and it looks very active, or there is a retinitis, and it does not affect the visual acuity, if our endpoint is visual acuity, then we're missing so much. So, we must come up with some scale that would be adapted for every disease entity, basically, and quantitate it so it will give us a more quantitative scale.

Marc D. de Smet: So, let me be more direct with regards to that. There are probably 40 different entities in uveitis. When was the last time in say cancer chemotherapy, a trial was initiated not to treat a specific cancer but all cancers at the same time?

Ron Neumann: Let me challenge you here. There are trials that for multiple sclerosis, and I would urge us to think that multiple sclerosis is also a sack of many different diseases. Nevertheless, including all comers to a study give played to positively in the case of Multiple Sclerosis (MS) if we consider the high number of new drugs that had been developed lately for this disease.

Marc D. de Smet: I agree, but we also put into uveitis infections, we combine inflammation targeting myelin with vasculitis such as Behcet's disease. You won't tell me that what works for MS, should also work for neuro-Behcet's!

Ron Neumann: No, but when you are talking about the noninfectious uveitis as a group—excluding the few outliers such as Bechet it may play to our benefit. In multiple sclerosis the clinician sees much less details than we do in our clinical examination. When we see all the fine details in our clinical examination, we diverge our group to many subgroups, and therefore we try to make our studies more and more specific.

Marc D. de Smet: I agree, but I think we're running about 30 years behind rheumatologists. Rheumatology 40 years ago defined various entities with clear endpoints, and we haven't done that so far in uveitis. It is not that we're less motivated than rheumatologists, we had less means.

We should define now entities as well as we can in terms of diagnostic criteria, but also in terms of what is an adequate clinical response. I see no problem in testing different uveitis entities with the same drug, so long as for each we know what represents an acceptable response. Right now, for intermediate uveitis it's fairly well defined. It is not surprising that several recent clinical studies focused specifically on this disease. It is a very select group of patients, and an adequate response in other entities is not the same.

John H. Kempen: I would like to just comment regarding the outcomes. The basic concept of this is about classification and misclassification. If we have some way of classifying a patient as active or not active, and it's not severely biased, then if the drug works, the classification scheme should work. If you have a small misclassification, if it's nondifferential between treatment arms, then it's only going to slightly depress the measure of association (e.g., Hazard Ratio). With a little bit of additional sample size to account for non-differential misclassification's effect on the measure of association, we should be able to detect the outcome.

In my mind, the AbbVie trials were good trials, but I don't think they were necessarily that much better than the other trials in terms of the outcome selected. I think that it's more that the drug was more efficacious, so the effect could be detected. I think that a wide variety of different designs would have found that the drug works.

As long as we have some way of measuring activity or inactivity (or whatever we're going to measure), and it's not severely biased, then the trial should find differences between active treatment and comparator for that outcome once the drug actually works and the sample size is sufficient. The issue about whether the drug only works for certain types of uveitis is another matter—a very important matter.

If we think about the outcome we are going to use, we should consider how the drug would be used in the future. In general, I think that people are going to decide if they like the drug or not based on whether they think it controls their uveitis. Most of us think we can tell if the uveitis is controlled. And so, it could perhaps be fairly simple to ask for the main outcome in our trials, is the uveitis controlled or not at the key follow-up visits? Or if the goal is corticosteroid-sparing, has that goal been met by the key time point?

I think success in our fundamental clinical objectives is really what we want to look for, because that's also going to support the use of the drug. If the drug is causing us to say that the uveitis is inactive, then we're going to like it in our clinical practice. Such a drug is likely to do well in the market. If the drug is not having too much effect, and we're sort of splitting hairs as to whether it was showing a positive effect or not, then it's probably not a very strong drug.

Ron Neumann: Just to come back to Quan's point before, if we are happy with the studies or not, as uveitis experts, I think that we would all agree that when the question is 0 or 1, and we go from 1 to 0, we are all happy. We are struggle a lot however, about severity of the disease at any certain time point. And there, where we are hitting issues of vagueness.

Manfred Zierhut: Yes. Probably a bit provocative, but positive results of such studies that include all types of posterior uveitis suggest that we don't know anything about the pathogenesis of these diseases. There's no one here who believes that VKH has a lot of similarity to Behcet's disease.

And if these both disorders behave nicely to the treatment, then this is probably only for a short time. The more we know about the pathophysiology of these diseases, the more specific our treatment will become.

Summarizing everything as “posterior segment uveitis” and giving them all such an unspecific treatment may result in immediate good results. But the future will tell you that you must work with subgroups. So Behcet's disease needs its own treatment, as we have shown in our interferon alpha studies. Never compare that with the simple signs of inflammation in intermediate uveitis.

Ron Neumann: With that in mind Manfred we seem to aim at orphan indications and niche development. Kerstin, please join us with your regulatory expertise.

Kerstin Wickström: Okay, thank you. By tradition we have accepted vitreous haze for posterior uveitis and cells and flare for the anterior ones. I fully agree with you with regards to JIA where you want to get rid of every sign of inflammation in these children. It is also agreed that there are difficulties with haze for uveitis involving the posterior part. As mentioned, in the Humira file, for example, they had a composite endpoint evaluating time to treatment failure. I think that is a good step forward. Of course, we want to have objective, reliable measures as free of subjectivity as possible, for example with regards to grading. The basis should be randomized, controlled and double-masked trials. This can take out some of the uncertainties of the gradings.

Ideally, if you have evaluated inflammation with a fancy machine in the trial, you still want to be able to follow the outcome in clinical practice. This is somewhat in an analogy with Lucentis when it was first approved. It was suggested that the retreatment criteria should be based on OCT or visual acuity. But only a few universities clinical had OCT, so initially, re-treatment criteria had to be based on visual acuity only.

So, when the drug comes out in clinical practice at least the majority of clinics should be able to follow the treatment effect.

Marc D. de Smet: Yeah. I don't quite agree, in all honesty, because I think clinical trials, first of all cost a lot. That's one of the bottlenecks. And, you know, clinical trials should never necessarily be the base on how you treat patients in clinical practice. Clinical studies have one aim, that is to demonstrate whether a drug works or it doesn't. And for that you need good solid criteria. What you use in clinical practice it is another story.

Kerstin Wickström: Yes, okay. Perhaps I was a bit unclear. It is fine with such objective measures in the setup of a clinical trial. But when it comes to the product label that tells you how to monitor the patient, it can't be based on those machines.

Marc D. de Smet: I agree with you.

Kerstin Wickström: And therefore, you need also to show us that what you measure in the trial is representative of what you're going to measure in clinical setup. So you have your laser flare meter, but you'll also do some manual evaluation, so you need both.

Marc D. de Smet: Yes and no. Depending on the situation, a manual count may be fine, but in others an automated measurement may be better if it can better discriminate between levels of disease. What is needed though are studies allowing a comparison between the two scales in clinical settings.

Where I think the flare meter is particularly useful is for example the measurement of flare. I can't measure flare very well with my eyes, but a machine can. When objective measures are critical, a device that measures flare as continuous variable has advantages. So I agree with you, but at the same time, it's not for the investigator to define or validate the parameters. Maybe you can ask a drug company to do these multiple comparisons because they need them for their label.

Kerstin Wickström: No, no, it's not, of course. And if you have a multisite trial, as I said, we would like to have an objective measure, but sometimes we can't get it. We understand that a flare meter is not available at every investigator's site, so it would have to be realistic, even for us as regulators.

Marc D. de Smet: That's true. On the other hand, if you don't have an objective measure, it increases the number of patients by probably 10- to 100-fold.

Kerstin Wickström: Agree that variability may well increase. And it is for the academia (and companies) to drive the science forward. It's not the regulators that do that.

Ron Neumann: And Marc, another question that relates to flare, Manfred just pointed out that he's only thinking about it with JIA patients. But how many times do we have a B27 person coming to you and telling you, “I feel that something is wrong. Upon examination they are completely normal. You don't see anything.” That said, some of them comes the next week with +2 or more cells. Maybe if we have a flare meter that is really good and much more sensitive than our eyes, we might see something ahead of time.

Marc D. de Smet: Ron, simply believe him, that's all, and treat him immediately when he feels something. That's number one. You don't need a flare meter for this one, yes?

Ron Neumann: But neither you nor the patient know. I feel that a sensitive machine that would tell me that patient have an increase in his flare level (still below my ability to evaluate clinically would be very helpful in guiding my treatment). Should the patient be already treated, should he be treated hourly? Four times daily?

Marc D. de Smet: Every half an hour.

Ron Neumann: Every half an hour without seeing anything?

Marc D. de Smet: If he feels absolutely, yes.

Ron Neumann: All right. I must say that I have experienced such patients complaining and not developing recurrence, so I am not so positive. Many complaints may be derived from ocular surface and not be part of the uveitis complex.

Manfred Zierhut: That person has experienced a lot of these things before. He has probably a doctor who didn't believe him. He said, “There's nothing. Go home.” Two days later, he comes with hypopyon. These people know about these things, and you can trust them. They're sitting around the television, and say, “Oh, my God, it starts again.”

And that's the point where my patients start immediately half an hour or an hour. It depends a little bit what happened before. If they can get rid of this one, and this is very, very successful. The next day they go to the doctors, and you may see probably 1- or 2-plus cells which were not there. This is an acute uveitis. That's a beautiful disease, yes? The chronic one, that's a real problem.

Ron Neumann: When I know the patient really well and it is a recurrent story as you portray Manfred that is another story, still I would be happier to have an objective measurement showing a difference before embarking on highly demanding therapy. Yes, Michal, please.

Michal Kramer: This brings me to the second topic with is an unmet need in the field of clinical trials and this is the chronic anterior uveitis. We are focused on this entity in children with or without JIA, but it is pretty much neglected in adults, and frequently looked upon as the simple disease.

If we would like to estimate the risk of vision loss in these patients, we again should aim to establish a dedicated activity scale. Such a scale should consider also their therapy burden and the risk of complications resulting in visual impairment, such as, the presence of posterior synechiae, cataract, cystoid macular edema, etc.

In clinical trial we should be very accurate. If it helps us to get conclusions in a shorter period with much more accuracy, that's great, because this is the goal of clinical trial. It's not to treat patients, just to help us as soon as we can to have this drug on the market. And that's different thinking.

That's why I believe that some of the trials did fail, because of a suboptimal design. Like, for instance, the definition of an activation. Looking at small fluctuations, are these really flares? If you're counting the small fluctuations, probably it will not show any difference between study groups.

And, dividing for different disease entities. There was as an attempt to do a Behcet's trial, and I agree, we all feel very sorry that it failed, because we could have had a drug that is good for this and may not be working for others. So yes, some diseases within the uveitis spectrum seem to stand out and require special attention.

Quan Dong Nguyen: I can see that if I were to sit on the other side, let's say, at the FDA or the EMA, how much challenges I will have in reviewing pharmacologic agents for uveitis. As uveitis experts, during the last 45 minutes, we have not ourselves come up with an agreement yet, so how do we expect the EMA or the EUAO or the FDA to agree on something. During the last 45 minutes, as I'm listening, I'm still not sure whether we are agreeing or not agreeing! Thus, it would be very difficult. I think that's why Dr. Chambers with the FDA has said that so far, he only accepts haze, or in the case of the VISUAL study, time to failure, as endpoints, because it has been difficult to define and accept other endpoints.

How do we propose that the uveitis community come up with something that colleagues can agree upon? We have not reach consensus so far.

Marc D. de Smet: I believe that's what Doug Jabs is trying to do by defining what the various entities really are and what their features are. And the other, which is what you brought up, is time to failure. Maybe what we should look at in uveitis is not so much specific measurements for improvement, but instead, since these are recurrent diseases, the time interval until there is a relapse. Each disease has a slightly different pattern. But relapse is a very objective endpoint and may be the most objective one. But we have never investigated uveitis responses along those lines.

Quan Dong Nguyen: Time to failure has been, because of the success of the VISUAL studies, employed in other ongoing international uveitis studies. But my thinking is that, yes, time-to-failure is a good approach, but for that approach in a trial, one must start all study subjects on systemic corticosteroid, which is not always appropriate. I know that one of the reasons why I myself like clinical trials is because I have options not to give patients systemic corticosteroid.

But in some current studies, because the sponsors want to be successful, they just follow what the VISUAL studies did and start every study subject on high-dose systemic corticosteroids.

There are caveats in such trials that use time-to-failure. The VISUAL studies demonstrate that adalimumab can be beneficial for patients with non-infectious uveitis. Yes, the trials showed that adalimumab worked. However, in clinical practice, colleagues would not necessarily use it as the first line of treatment until they have employed other agents because of the study designs.

Ron Neumann: Dear panelists, I would like to thank you very much, the audience and yourself, for this discussion, and like usually uveitis discussions, we do not come to agreement. Yes, Kerstin, please. Please come down.

Kerstin Wickström: Just a quick comment on the last one. Back to the rheumatology, arthritis field again, and we're talking about maintenance therapy. We talk about induction therapy. If you have an acute uveitis, of course you want a good drug that's nonsteroidal to get the patient by it. So there are two phases, an induction phases and a maintenance phase and both should be explored.

Ron Neumann: Thank you very much. John, last remark.

John H. Kempen: So I'd like to add a comment as well. For anterior uveitis, we have many successes in developing drugs using anterior chamber cell grade as an outcome. Anterior chamber cells correspond very well with what we think of in deciding whether the inflammation is responding or controlled or not. So, for anterior uveitis, which is a large majority of uveitis cases, we already have a good system in place that has a track record of success with drug development.

I think with the other forms of uveitis, we should give credence to outcomes like overall control of inflammation. This “time to failure” kind of concept to some extent addresses that. I do not like the idea that losing visual acuity is part of failure, because somebody could have gotten intra-vitreous corticosteroids a month and a half before enrollment in the study. Then they develop a cataract during the study, and now it's counted as a failure which may have nothing to do with the treatment. It's true that such an event may end up being balanced across treatment groups, but why do we need it? The immediate goal of treatment is to control inflammation without causing unnecessary side effects. It is true that one of the key reasons for controlling inflammation is in order to preserve vision, but preserving vision is an indirect effect of the treatment. So to me, it seems like controlling inflammation would be the ideal outcome for a study we want to complete quickly, to make the treatment available.

Regarding heterogeneity of effect across different types of uveitis, we're a little bit confused because corticosteroids work for basically all the forms of ocular inflammation, and so we expect all the other drugs to work for all of them, too. Nevertheless, they won't necessarily work equally well for all of them. That sais, many types of ocular inflammation are so uncommon that it will be hard ever to do sufficient trials to sort that out, and we probably will have to rely on post-approval clinical outcomes.

In summary, it seems to me that control of inflammation should be what we're monitoring and tells us whether we're succeeding with our therapy. In terms of implementation afterwards, control of inflammation should correspond well with our therapeutic goal in managing the patient. I would argue that we can largely agree on control or lack of control of inflammation (and the related concept of corticosteroid sparing), and ought to be able to do trials using that outcome. We may however be prudent to inflate the sample sizes for heterogeneity of effect across different types of posterior and panuveitis.

We do have a problem in deciding on whether an eye is active or not when there are merely 1 or 2 cells per high-powered field in the anterior chamber. But if we differentiate between very active and minimally active, I think we can have very good agreement in discriminating that for almost any uveitic disease.

Manfred Zierhut: So, final words from my side, it seems that biologics are very effective, especially if you can concentrate your studies on special entities. Look for the JIA KIT study. This was stopped, yes, it was so effective. So we have these situations. If you have the right group with the right fantastic, quickly effective drug, you see miracles. We see that with other biologicals in macular edema, get rid of macular edema in a week or in 2 weeks. This is effective, yes? Other things, not so much.

Ron Neumann: Thank you very much all panelists and audience.