and the hope is that this will ultimately reflect the subsequent biology directing therapies and

Transcription

1 FRANK C. SCIURBA, MD 1 So today we re going to be talking about the clinical presentation in biology and its variation in COPD. As you ll see there s much new information on the subgroups and characterization of COPD and the hope is that this will ultimately reflect the subsequent biology directing therapies and prognosis in this disease. So today we ll be covering the importance of phenotypic clarification, we ll be looking at the variation in clinical/physiologic and radiographic presentation, looking at the variation in comorbidity outcome in response to therapy based on phenotype and then looking at the variation in the biology to the extent that we understand it. This is the classic representation of the subgroups of COPD and while to some extent this still applies today, increasing exposure to the complexity of this disease let s us realize that the variation is much more complex than this. The typical separation of COPD into emphysema, chronic bronchitis and to whatever extent asthma overlaps does not come close to explaining the variation for instance in these two individuals. The individual on the left has disabling dyspnea, has a similar FEV1 to the individual on the right who s an aerobic instructor. Both individuals have severe emphysema, neither have bronchitis nor asthma and clearly this one to believe there s a fundamental difference in the biology, in the clinical progression of these two individuals that likely would warrant unique interventions and prognosis. COPD as we know is a disease of airflow obstruction that is caused both by the genetic milieu of he individual but interacting with environmental exposures most commonly in this country tobacco exposure but also other factors such as respiratory infection and environmental factors can play a role in the development progression. And as we ve come to learn not all COPD progresses the same

2 FRANK C. SCIURBA, MD 2 way and in fact that line is much more complex into the many different subversions with COPD depending on that person s unique genetics and environmental exposures. This unfortunately is the state of the art in COPD and that is that it is very complex with many different variations and subversions and the yellow labels on these circles represent attributes of patients with COPD for instance, disproportionate pulmonary vascular disease, disproportionate dyspnea, peripheral muscle dysfunction, small airways disease, emphysema, emphysema distribution which can vary independently within individuals and may in fact suggest unique biology. So why do we want to classify COPD and dissect the phenotypic subtypes? Well it allows again reclassification of antiquated disease syndromes and it allows hopefully individuals classified with more clearly defined phenotypes who represent more homogenous molecular and cellular processes and which may demonstrate similar prognosis and therapeutic response. This is the goal of the National Institute of Health with regards to the roadmap of personalized medicine, that we will be able to define the unique genetic attributes of an individual and treat individuals uniquely. This exemplifies a typical analysis in the past of a subgroup of patients with COPD defined by any common variable the specific variable, not necessarily specific, but apply an intervention to this patient population results in something like this so a therapeutic trial intervention as assessed by the FDA may result in a mean increase of X amount. If you in fact dissect this population into subgroups, and this is a theoretical representation of where we would like to be, different subgroups represented by the different circles may respond very

3 FRANK C. SCIURBA, MD 3 differently. While all those dots when they were gray moved in the same direction as the previous dots, in fact by dissecting the subgroups we see that in fact there is a group in red that has a dramatic response and should receive the therapy. The group in yellow has a more modest response, in white is nonresponsive and in fact we ve learned that there is a subgroup within this population of responders who in fact should not receive the drug because they exhibit deterioration. And we feel it s important to dissect out these subgroups. The hope is that these groups will have in fact unique molecular mechanisms and suggest in fact unique therapies for the disease. So let s expose what is this variation that we are talking about. Well here is another example of two individuals with similar FEV1s, that s the Forced Expiratory Volume in 1 second used typically to classify severity of COPD. The individual on the left has 80% of their lung destroyed with emphysema, whereas the individual on the right has less than 20%, yet they both have similar FEV1s, again leading one to believe that these two individuals may have unique biology resulting in a unique prognosis and warranting unique interventions. In fact we do know that the parameter of airflow obstruction as we define it, or FEV1, is made up of two very unique or affected by two very separate biological processes that plausibly may warrant unique therapies. Airway remodeling, fibrosis, increased airways resistance and parenchymal destruction, loss of lung elastic recoil may be very different biological processes both leading to airflow obstruction and individual with disproportionate or exclusive manifestations of one or the other in fact may be very, very different biologically.

4 FRANK C. SCIURBA, MD 4 We have ways now of actually quantitating the degree of airway abnormalities and parenchymal abnormalities using quantitative measures of typical CT scans. We take the lung density or the Hounsfield Units and their distribution and we can create graphs and quantification of the amount of emphysema, distribution of the emphysema. In this graph the areas of lowest density or greatest lung attenuation are represented in purple and we can see going in this histogram going from the apex to the base of the lungs we can see the proportion of each cut that has emphysema versus normal and get a sense of the distribution of emphysema within the lung in a very objective, quantifiable way that can go into databases. In addition we ve learned to assess the thickness of the airways, and here again are two individuals, the individual on the left has thinner airways, the individual on the right thicker airways that we can see visually. In fact the individual on the left has a higher and more normal FEV1 than the very low FEV1 in the individual on the right who has actually very little emphysema. We can apply quantitative algorithms to the airway walls and actually define the airway wall thickness or the percent of the external circumference that is in fact tissue and get a perspective of airway thickening. If we in fact take these two attributes from CT scan and plot them against each other we learn and on the horizontal axis the amount of emphysema or low attenuation units plotted against the wall area thickness or wall area percent it s really a scattergram. Individuals, and this is a cohort we have of smokers who we made these quantitative CT analyses on, we learn that there is really no relationship of these two parameters to each other. So there are two very different disease processes associated with smoking; however both of these separate processes are in fact independently associated with airflow obstruction. And you can see on this 3 x 3 chart that individuals with

5 FRANK C. SCIURBA, MD 5 increasing emphysema independent of wall thickness have a lower FEV1, individuals with increasing wall thickness independent of emphysema have a lower FEV1, and in fact those individuals in the upper right corner who have both the most emphysema and the thickest airway walls have the lowest FEV1, again supporting that these two processes of very different biology can both independently influence the classic parameter FEV1 which has defined the disease. But it can be even more complex than just the amount of emphysema. We notice variation in the distribution of emphysema. Again, the individual on the left and the right panels have the same amount of emphysema, but the individual on the left has holes spread throughout the lung, the individuals on the right has upper lobe collection, upper lobe dominant with relative preservation of the lower lobes and more focal severe emphysema destruction. Again leading one to believe perhaps there is a difference in the genetics, the biology, possibly prognosis and treatment in these two individuals. And we can even go beyond that just looking at different patterns, you can take these two individuals and it would be very difficult without some more complex algorithms that I think we need to develop to define objectively and quantitatively the difference in the individual on the left versus the individual on the right but we can see the individual on the left with more large discreet holes surrounded by thicker septa may appear at least visually to be different than the individual on the right. And again if we can get better at classifying these patterns we may be better at dissecting unique subgroups of patients with again unique prognosis.

6 FRANK C. SCIURBA, MD 6 We can go beyond the imaging and the radiology. Looking at functional performance, we ve learned and these are two separate exercise tests, on the left is incremental bicycle ergometry and on the right is 6 minute walk. And we see a relatively loose association between exercise performance and endurance with FEV1, the typical parameter classifying a patient really suggesting that there is great variation and many critical attributes resulting in variation in performance independent of the classically defined parameter of airflow obstruction or FEV1. In addition to looking cross-sectionally at the differences, we also notice that following patients longitudinally there is variation in progression in individuals, and so this is a recent New England Journal article from the Eclipse Cohort following patients over 3 years and looking at the decline in FEV1 and there are some individuals that have a fairly dramatic negative decline in FEV1, other individuals actually have slight improvement in FEV1 over time. And so this variation longitudinally is also of interest and has been very, very poorly analyzed. We have a cohort here we are following in Pittsburgh and now have 230 patients who we ve followed longitudinally. The top panel shows a relatively similar distribution as the Eclipse Cohort, over 2 years most individuals do decline but in fact some individuals stay stable or even improve their FEV1 over time. But we ve gone one step further and looked not just at FEV1 but we looked at change in the quantitative amount of emphysema, the bottom right panel, and visual scoring, quantitative visual scoring by one of our expert radiologists over time. And we see again variation in the progression in the emphysema over time. Some individuals actually decrease the amount of emphysema, many increase. What is most interesting though is that individuals how have a decline

7 FRANK C. SCIURBA, MD 7 in FEV1 are not necessarily the same individuals who have increasing emphysema, and in fact measuring the emphysema quantitatively and using visual scoring often gets different patients who we classify as progressors. What s most interesting is that of those 230 patients only 12 had progression in all parameters, but over 120 had progression in any one parameter. Very importantly though progression in any one of these parameters, FEV1, visual emphysema progression or quantitative emphysema progression was associated with an increase in shortness of breath so each one of these parameters appears to have clinical relevance even though they don t decline associated with each other and this we feel is very important in classifying subgroups of patients who progress or don t progress. Now as we ve gotten better at measuring the variation in all the different attributes in COPD radiographically, clinical questionnaires, physiology we re now left with very large databases, not dissimilar from what we get with genetic analyses or gene expression or protein analyses, and we ve applied some of those classification, those clustering techniques that we learned in these genomic datasets to in fact these large clinical datasets. And so we applied this to our SCCOR dataset of nearly 700 patients and I came up with in fact 5 different clusters of patients who had certain attribute that were disproportionately high or disproportionately low in relationship to each other, and this is a heat map, again typically used in, in gene expression analyses used to classify the different clinical characteristics in showing how certain patients tend to group together with regards to certain characteristics. This may be one technique and this is really an evolving field, how do we subclassify these complex patients and complex datasets but this may be one approach that is being

8 FRANK C. SCIURBA, MD 8 used by others as well to try and define subgroups that may in fact have unique biology. We hope to relate these subgroups to eventually the genes, the molecules, other attributes to see if we can in fact find unique groups with unique biology. This is just another way of trying to subclassify patients looking at principal component analysis. Principal components represent the integration of a variety of variables and this is the plotting of the first and the second principle component showing that patients have very wide distribution but that in fact we can in fact find clusters of patients within those distributions that do seem to behave similarly. So this is a first step we hope to eventual more meaningful subclassification. So what does all this mean, what do we know now whether any of these sub characteristics suggest unique biology, comorbidity outcome or response to therapy? Well we do know that COPD while it is a disease defined in the lungs including airway inflammation, structural changes of emphysema, mucociliary dysfunction and small airways disease that in fact there is a significant systemic component in part related to inflammation that results in an association with weight loss, cachexia, skeletal muscle dysfunction, cardiovascular disease and other comorbidities such as osteoporosis, depression and cancer. Mechanistically how these are all linked we are unsure, but the association of comorbidities with the lung disease independently affects symptoms and quality of life. And it appears to be associated with COPD independent of any tobacco exposure, it seems to be related to the disease process in some ways.

9 FRANK C. SCIURBA, MD 9 In fact if you look at what patients with COPD die from, patients with more moderate disease, Gold I, Gold II in fact rarely die of COPD, which are the tan bars. They most commonly die of other conditions, lung cancer, cardiovascular disease in disproportionate numbers to what would be expected just from tobacco exposure. But even patients with more advanced disease, Gold III and IV, FEV1s less than 50% predicted, while some of these patients do die of respiratory failure related to their COPD, in fact many of these patients also have death from lung cancer, cardiovascular disease in disproportionate numbers. so these comorbidities are critical not only for symptoms and disability but for resulting mortality in these patients. So we talked about differences in the individual on the left who has a similar FEV1 as the individual on the right, but the individual on the left has more emphysema than the individual on the right. What does that mean? Well there is an interesting natural experiment suggesting environmental exposures that may cause one pattern versus another pattern. While in this country tobacco smoke is the most common cause of COPD in fact in the Third World and these are two examples, in Madagascar and Ethiopia of indoor biomass exposure often in women who cook in poorly ventilated homes. And these individuals also get COPD, but the pattern at least in one preliminary study suggests it may be very different. Individuals with biomass or cooking exposure versus tobacco exposure tend to have less emphysema read by a radiologist and more airway dominant disease than those with tobacco exposure. And so the patterns appear to be different. We also know that from the NHANES Epidemiologic Study that individuals with similar FEV1s and who have COPD but are never smokers actually have less of a mortality rate than individuals who

10 FRANK C. SCIURBA, MD 10 get COPD with smoking. What we don t know for sure is that these never smokers here also have that pattern of airway dominance and not emphysema, but it certainly is a hypothesis that some of the indirect data might support that emphysema that obstruction associated with emphysema may in fact have a poorer prognosis than obstruction associated with just airway disease. There is other data to support this. This is a study from the UK, McAllister, who showed an association between pulse wave velocity, a marker of arterial stiffness, and in the right upper panel FEV1, or airflow obstruction, and there is a correlation. However if you look at it plotted against emphysema the left upper panel in fact the correlation is significantly higher. And in fact if you look in the bottom right panel and look at quartiles of arterial stiffness in fact as the arteries get stiffer there is a much stronger association with severity of emphysema than with severity of airflow obstruction, so it appears again in this example that it s the emphysema and not just the airflow obstruction that drives the vascular comorbidities. We see a similar phenomenon with lung cancer association, and if you look at for instance Gold III- IV patients, the top two rows, and this is from a cross-sectional lung cancer screening cohort, we visually identified whether emphysema was present or not. And of the 154 individuals who were severe air flow obstruction, Gold III-IV, who had visual emphysema 10% of them, 15 had a diagnosis of lung cancer whereas none, 0 of the 63 individuals with advanced COPD, Gold III-IV who did not have visual emphysema, none of them developed cancer. Similar to individuals with tobacco exposure who didn t smoke, there is about a 3-fold odds ratio or not who didn t smoke but who do not have airflow obstruction, there is a 3-fold odds ratio in those that have visual emphysema

11 FRANK C. SCIURBA, MD 11 versus those that do not. So again it appears that the emphysema disproportionate to the airflow obstruction drives the comorbidity of lung cancer. And then finally Dr. Bonn, working in our SCCOR Cohort, identified in fact a relationship between emphysema and osteoporosis as might be suggested just by visually looking at the loss of matrix in these disease processes. And in fact if you look at the association of osteoporosis measured by DEXA scan and FEV1 or airflow obstruction, there really isn t an increased odds ratio just from having COPD, but if you look at the presence of emphysema moderate and severe or even trace and mild emphysema, there is a significant 2 to 3-fold increased risk of having osteoporosis. So emphysema, not obstruction results in an increased risk of osteoporosis almost certainly related to similar biology driving both of these processes. What do we know about phenotypes suggesting response to therapies? Well really the first dramatic exhibit of this came out of the National Emphysema Treatment Trial, a trial which randomized patients to lung volume reduction surgery versus control. And you can see the top left panel a group of patients who have upper lobe dominant disease and low exercise who have lung reduction surgery there is a dramatic reduction in mortality rate or decline in quality of life in patients who had lung reduction versus controls, whereas if patients had in fact lower lobe dominant non-upper lobe emphysema or more diffuse emphysema and preserved exercise tolerance this sub-phenotype compared to this phenotype had no benefit from lung reduction surgery from a mortality standpoint. And so this really was the groundwork to suggesting that in fact variation in phenotype that s not

12 FRANK C. SCIURBA, MD 12 classically measured just by looking at FEV1 can result in difference in outcome to therapy. We hope to find more of these examples. What do we know about the variation in biology in this disease? We ve shown some of the variation in clinical presentation. We know that in fact there is association with certain phenotypic distributions and unique genetic variance or polymorphisms. We know in Alpha 1-antitrypsin deficiency that in fact this results in a lower lobe dominant pattern of disease, a sub-phenotype. Interestingly in a group of patients who have glutathione S-transferace or microsomal epoxide hydrolase enzymes that in fact metabolize the toxins in tobacco smoke, that pattern is associated with upper lobe dominance. So we are beginning to see some of these biological associations with the phenotypes. But we know there is incredible variation in some of the biology, for instance Dr. Morris at our institution identified that a percentage of patients have in fact colonization, this is from tissue taken from lung transplant patients, colonization with Pneumocystis. We can look at any variety of organisms, and there is a variation in the environmental exposures to these organisms and the host response that may result in unique phenotypes. By measuring these variations, by measuring the biology it may help us to meaningfully subclassify these patients. Recently there has been an identification that some patients with emphysema and COPD have autoimmunity. This first paper that came out in Nature by Dr. Lee again showed that some patients with COPD or emphysema have elevated anti-elastin antibodies, whereas it was rare to nonexistent

13 FRANK C. SCIURBA, MD 13 in the controls. But this didn t occur in all patients, suggesting that there is a sub-phenotype of a subgroup of patients that have anti-elastin antibodies that may be meaningful in disease progression. Dr. Duncan in our institution with Dr. Feghali-Bostwick has identified in fact a proportion of patients with autoimmunity, or in this case represented by various patterns of ANA. And whether these beget unique prognoses need to be further examined. We do know that one autoantibody against 130 kda antigen tends to be associated with a low body mass index, really some very early validation that in fact these autoantibodies may be associated with unique clinical patterns. These need to be examined certainly more thoroughly. We examined a subgroup of patients who had airway dysfunction and emphysema classifying those quantitative parameters we discussed earlier and in fact showed various associations with peripheral protein markers depending on whether there was simply airflow obstruction or wall thickness or emphysema and that these inflammatory markers did associate differently with the different subphenotypes suggesting biological meaning. Dr. Tedrow working with lung tissue in Dr. Kaminski s laboratory took a group of patients who had severe emphysema, lung tissue from these patients and compared them to patients who had more minimal emphysema with tobacco exposure, and found over 400 genes that were differentially expressed in the lung tissue in patients with more emphysema, again suggesting that there are in fact unique targets, unique biology that we need to work out and perhaps discover perhaps treatable targets in this disease.

14 FRANK C. SCIURBA, MD 14 Interestingly, we have in our SCCOR Cohort looked at, and this is an example of four markers, Pentraxin, PTX3, THBS1, SPD and CCP, and in fact identified that each of these markers have interesting and unique attributes. Without going into too much detail, some of the markers are associated just cross-sectionally with severity of disease, some associated only with emphysema cross-sectionally but not obstruction, some of these markers may be associated cross-sectionally but didn t predict progression of disease. Others did predict progression of the various phenotypes. Some are associated only in former smokers but not current smokers and so these markers may be very useful in identifying sub-populations of the biology, the disease that may be meaningful for future therapy and direction. So I appreciate your attention and I realize that while we have now begin to come a long way at really identifying the incredible variation and in fact convincing ourselves that it is meaningful, we have a lot more work to do to really associate the variation and clinical pattern with the unique biology but we do believe that this will eventually result in meaningful subclassification predicting unique prognosis, hopefully identifying unique targets that will in the hoped personalized medicine approach treat patients individually and more effectively. So I appreciate your time, I tend to end my lectures with this Larson comment that we ve come a long, long way but we have a long way to go in this disease process. So thank you.