DR.MESSING: Would the panelists come up, and this is open for discussion now. I guess we have an abbreviated discussion session but please anyone who has questions come up to the microphone?

PARTICIPANT: This is a wonderful series of talks to kind of look into the future. Let me ask I guess all of you up there the baseline, the truth is histology showing cancer.

I am wondering, and I am sure many of you thought about this but we don't in our individual settings, do you envision a time where that will not be the indicator of cancer, where someone will come in; he will a blood test, etc., identifying based on one of these future assays indicating that there is cancer, maybe prostate cancer and don't even worry about the biopsy because the specificity is so good that we won't wait to see the clinical biopsy information, so that everything is based on all of your studies seeing there is a cancer there in tissue? Is that really in the future what we are going to be finding?

DR. SIDRANSKY: I will take a first crack at that. It reminds me of what we used to say in our fellowship, "Tumor is the rumor and tissue is the issue."

I believe that all the tests that we are currently developing are going to augment standard histopathology for a long time to come. It is always when I look back at all the genetic and epigenetic changes that pre-neoplastic lesions as well as cancer cells have, I am always amazed that really the turn of the century we solidified morphologic diagnosis to give us the information that it provides us which I think is amazing that the genetic and epigenetic changes are the ones that ultimately dictate how cells look and remarkably you know pathologists have sat under the microscope for years, figured out these other changes without knowing anything about the underlying genetic changes and actually were able to give us very useful information.

I think it remains the standard, and I don't see us changing that standard for a long time to come. I see most of these tests as potentially augmenting those morphologic changes that are evident and I think the first place that we are probably going to make the biggest difference is in that huge gap between totally normal and totally cancer, whether you call it you know some sort of intermediate lesion or pre-cancer, I think that is where a lot of these new markers are going to help in the near future.

DR. MESSING: The treatments we have now are also fairly barbaric, even the minimally invasive ones and I think that that goes along with the other side of the coin but on the other hand we already have that to some degree as an experience because all of us deal with the positive urine cytology and we don't know where it is coming from, and that has a very, very high specificity and so the likelihood of a false positive does exist but it is quite rare.

We wind up chasing our tail after that but very few of us treat enormously because of that. We may put medicines in the bladder but we don't take the bladder out because of it.

So, I think I am just sort of echoing what David said.

DR. PETRICOIN: I could probably follow that. I always kid my colleague, Lance Liotta that he is going to invent himself out of being a pathologist, but I don't think that will be the case. I think as David alluded to perhaps even the pathologist was the first pattern recognition expert and because of that I think you just have new tools to do more high-dimensional pattern analysis, but I do think that in the future we won't so much classify them by histology as we will by molecular portrait of perhaps genomic and proteomic and metabonomic end points and that portrait will be the decider of how that patient should receive treatment and be followed and then provide maybe the surrogates for following efficacy.

So, I think the pathologist will become more of a molecular diagnostician than a classical tissue-under-the-microscope-trained scientist.

PARTICIPANT: I have a question about serum-based markers. There is a lot going on in the serum besides what proteins and DNA are floating around like whether or not an individual is a smoker, what they have eaten before the blood test was drawn, whether or not they fasted before it is drawn and that is sort of thing.

How sensitive do you think these basic methodologies are to be able to ignore all of that and not be influenced by it and if they are too sensitive to that how many prospective trials do we need to do to sort of sort that out? You know, should you eat a fat meal before you have this? Do you have to be fasting? Should you not be fasting? Should you do your normal life style and that kind of stuff? Where is that all going to lead?

DR. SIDRANSKY: Let me take a crack at that, also, because it is very interesting. We just published a study using MALDI for head and neck and lung cancer, and it was interesting. We wanted to see that kind of over fit where we are over predicting for certain baseline characteristics.

Importantly because we use smoking in our controls there was a significant amount of smokers and that was not a problem but alcohol which is not surprising since heavy alcohol drinkers probably upset their proteins in the blood and even though it isn't statistically significant did result in over fitting. I think serum is much,much tougher than I think what anybody believes right now.

When I look at the molecular tests we have developed in tissues they are extremely reproducible which is I think getting at the source. When we look at urine and saliva it is moderately reproducible because I think we are also getting a large sample from the source. When we did for example our methylation markers in serum there is a lot more noise out there and I think what you are alluding to is exactly what is going ont.

Ideally we will be able to make some sort of a weighing or a correlation with these things in the background so that they will kind of get out of the equation and I think it is one of the biggest challenges that we have but I think the notion that we can take a drop of blood and look at protein, DNA or whatever and be extremely sensitive and specific requires a lot more study than what we have currently done.

DR. MESSING: Mike?

PARTICIPANT: As a continuation of this discussion two things, well, three things, don't we run the risk of being too sensitive? While I appreciate Ian's comments focusing on what is likely to be significant in an individual's lifetime versus ignoring that that is not likely to be significant, isn't that a more primarily important focus because the worst thing that I think you can do is diagnose a cancer in an individual who is unlikely to be impacted by that cancer other than having a psychological burden of knowing that they have that word?

We run the risk I think of being too sensitive without having the adequacy of knowledge as to not only when to treat but how to treat effectively without introducing the possible side effects.

The second issue is also in a way philosophical. If we know that we haven't reached a phase in an understanding of a particular marker that is useful, we do have the two markers that were discussed today, BTA, BTA stat and NMP and we are all aware of the difficulties involved in their practical application.

Is there not a mechanism for decreasing or eliminating their usage? We talk about them being approved, but from what Ian was talking about that approval may not have satisfied criteria that we now recognize as being more important before approving them for general clinical usage.

DR. THOMPSON: Let me just give an overview and then let them answer that. You know we use markers today. We use clinical diagnostic markers. Microscopic hematuria is a marker for an intervention. PSA is a marker that we currently use with its own associated foibles, and PSA is an imperfect marker as well just based upon the anxiety that it tends to cause, the same sort of thing in interventions.

The remarkable thing that we have the opportunity to do now is to look for mechanism-related markers and markers that may be related to prognosis to move forward into that ideal world where you actually find disease that we need to find to treat.

DR. PETRICOIN: I would, also, think that there is a tremendous opportunity, and one of the things that we are pursuing is coupling serum type proteomic analysis with diagnostic imaging to save unnecessary biopsies.

The spiral CT for lung nodules which has a very good sensitivity but crappy specificity if we had a serum-based test to save unnecessary thoracotomies we could have a clinical impact, and that is something that we are pursuing with imaging companies today, the same with unnecessary prostate biopsies.

So, I think that you are absolutely correct. It is always the intended use. The Food and Drug Administration is always getting into the labeling end of things. It is the intended use that you really need to hone in on. You have to have your training and testing sets obviously reflect that, but just to diagnose cancer say in a prostate without having some idea about the clinical indolence or aggressiveness at this point in time with the new therapeutics coming on the market I would say is more ethically reprehensible than anything else because you want that tie in to have a good understanding of how to take that patient forward throughout their entire course.

DR. MESSING: I can answer finally to the last part and I don't have a regulatory answer to it but the marketplace has done a reasonably good job. Maybe I am entirely wrong. I don't hold stock in either company, but I don't see BTA or NMP22 beating pathway to the upper levels of Wall Street.

So, I don't think we use it currently that much and that may be enough of a thing rather than just banning them which may not be appropriate.

PARTICIPANT: Two questions. That was a great session. Ian, everyone I think proteomics is where it is at, but the results have been seen to be very operator dependent and when I sort of looked through your list it looked like these excellent people at NIH were going to do all this laser capture and all preparation. Then you send it out to others maybe to check, but the real issue is what happens if you send these samples out to slightly lesser places, get them to do the laser capture and preparation and then see if your SELDIs are all reproducible because I think that has been the problem.

The next question just is looking into the future. If proteomics is going to do away with DNA microarrays where does metabolomics fit into this? Will metabolomics sort of allow us to get away from proteomics or are we going to have an omics that goes in, you know, one to the other to try to answer these questions, if you would just look into your crystal ball?

DR. THOMPSON: I don't want to answer the second question, but the first one is well taken and that is one of the biggest concerns about SELDI is that it may not be, you know if the humidity changes you get a different peak pattern, but the issue is that it performs better than the existing tests in the populations in which it has been tested.

So, we have an obligation to see whether or not it is possible to do so and that is how the program was set up, to ensure that you actually can reliably take it out to other laboratories, generate the same peaks, get the same assignment using the algorithm.

Ideally you know these sorts of technologies, and I am not a technology person, may not be in your office or they may not actually use SELDI. There are reasons for using potentially SELDI versus some of the higher fidelity mass specs and again I am not a mass spec person but having witnessed great arguments between the two, there are reasons to explore the possibilities of both, some of which is cost, but the short answer is the right way to test it is to find out whether or not it is portable, to find out if it actually generates the data in a more descript, because right now we have papers that suggest that it is the degree of assignment, the performance characteristics of the test are superior to what we currently use.

PARTICIPANT: Yes, but on your very nice, you know, one, two, three, four, five, I mean you very much make a point we are going to tell you can you use this in your clinical practice. I mean that is what you said.

So, you know, if we are going to use it in our clinical practice, it is not going to be practical that every single SELDI is done by Lance Liotta .

DR. THOMPSON: These are seven different institutions around the United States that are actually validating that, and that is to specifically answer your question whether or not you can actually take it out and use it elsewhere.

DR. PETRICOIN: I, also, want to follow up on the metabonomics side of things. I mean Labcorp and Quest have licensed the US Government technologies around proteomic pattern diagnostics and they are going to be offering a home brew test using electrospray-based metabonomic profile next year for ovarian cancer. They are going to follow that up with prostate cancer.

Now, how they are going to validate this home brew remains to be seen. As a Food and Drug Administration employee I would like to see these under full FDA rigorous clinical trials, but nonetheless under home brew people like Quest and Labcorp I am sure and people like Richard Bender are very good at doing these types of analysis and the metabonomic-type studies seem to be holding their own as well.

So, I think there is lots of room in the ocean for people to swim here.

PARTICIPANT: I think one of the really interesting questions is what happens to people over time. I mean are you going to be, you know, we all think that we develop a cancer and some of those cancers don't clinically develop into something significant. They may go away and so I think an expansion of Eric's point is that we need natural history of people before we use proteomic diagnosis but getting to the point that was raised this morning how many of us would counsel a patient to go on finasteride for chemoprevention, we may be able to use this to develop a risk group, saying, "Yes, this person should be on a chemoprevention protocol," but I am fearful of using this diagnostically until we know what the longitude is in terms of over time will someone be positive today, negative in 3 months because their immune system worked.

DR. THOMPSON: Actually that is one of the charges to the Data Monitoring and Coordinating Center of EDRN is to develop actually the science of evaluating the mathematics of evaluating predictive performance characteristics because when you close a trial for example the microsatellite instability trial, what happens if it is positive today and your cystoscopy which is done at the same time is negative; is that patient at risk of going on further?

So, there is an entire science of the math that has to be developed to evaluate these markers. Good point.

DR. SIDRANSKY: I just want to say that I think you have to be very careful with a nihilistic approach because it sounds very reasonable, but I would say that two arguments I think in my mind always override. One of them was the terrible tragedy that happened at the Mayo Clinic where kids with osteosarcoma was randomized to receive or not receive chemotherapy because nihilists decided that improvement in surgical techniques were responsible for the improved survival and they lost 60 kids. So I think you have to be very careful with a nihilist approach, and the other one is I would ask you directly if your PSA is elevated or rising would you really be so objective as to sit down and say, "I am just going to wait and see what the natural history is"? and I think the point is that when the opportunity arises you can't take a nihilist approach. What you need to do is you need to rigorously test it and try to figure out how best to use it but not to, I think just to scorn that approach I think is also very dangerous.

PARTICIPANT: But I don't think I am being nihilistic. I think we know what the natural history of a rising PSA is. We haven't established what the natural history of this test is. All I am saying is we have to show that natural history. I am not saying that we should ignore it.

DR.MESSING: One more question, and then we will be going to lunch. So, Paul?

PARTICIPANT: This has to do with validation of serum-based markers. I have listened to these arguments from the early days of AFP and HCG and PSA but I have still never quite understood them and that is in the face of well-documented and large serum banks why you need prospective studies and can you give me an example where a prospective study showed something that the retrospective large serum bank studies did not show that made a difference.

DR. PETRICOIN: I would just say from the Food and Drug Administration's point of view it is all about the intended use of what you are going to come into the agency, if this is going to be for a clinical trial of what you are going to label your test as being able to do.

If you can't power the labeling to do that from your prospective trial then you are not going to be able to do it. I mean there is no necessary reason that you would have to have a naive naked prospective trial out there to validate but a lot of times what happens is that people will come in to the agency with an indication that they don't have power under the prospective trial to actually answer and that ends up happening more often than not, unfortunately.

DR. SIDRANSKY: I would just like to add to that that in fact the FDA is very willing to look at retrospective analysis of a prospectively collected serum if there are no issues there, but we have found for example that some of these DNA tests require much larger volumes than are really available.

So, traditionally while only a few microliters is needed to do simple ELISA we traditionally need right now between 1 to 2 mls of serum to be able to extract enough DNA to do methylation. So, that really doesn't set the stage to adequately use that many of the prospectively collected samples and I think that no matter how much you think in advance of what you want oftentimes new technology changes and I think unfortunately it remains a problem.

DR. LINEHAN: Thank you very much, Dr. Messing and the phenomenal panel who talked about this really exciting work. It was a great session.

We are now going to break for lunch. Also, we have lunch and poster session 2 on the ground floor here and you can pick up your lunches in the lobby. Be sure to be back at one-fifteen. We are going to do the Huggins Medal presentation at one-fifteen. Thank you.

TOP