Inflammation and CVD: Recent data heighten interest in effects, new strategies
For many years, cardiologists and basic science researchers suspected that inflammation plays a role in CVD, but definitive evidence proved elusive. That changed in August 2017 when findings from the CANTOS trial, presented at the European Society of Cardiology Congress and published in The New England Journal of Medicine, proved the inflammation hypothesis.
CANTOS evaluated whether canakinumab (Novartis), a fully human monoclonal antibody that targets interleukin-1 beta, would prevent CV events in MI survivors who were at increased risk for recurrent events due to persistent inflammation. Canakinumab has known anti-inflammatory effects and has approval for clinical use in rheumatologic disorders.
In CANTOS, 10,061 patients who randomly received one of three doses of canakinumab experienced a marked reduction of C-reactive protein over a median of 3.7 years of follow-up. The 150-mg dose conferred a 15% reduction in the primary endpoint of first occurrence of nonfatal stroke, nonfatal MI or CV death (see Table) and a 17% reduction in the key secondary endpoint of any component of the primary endpoint in addition to hospitalization for unstable angina resulting in urgent revascularization. Further analyses showed that canakinumab was also associated with reduced risk for lung cancer and cancer mortality, compared with placebo.
Cardiology Today assembled a panel of experts to discuss inflammation and CVD, the inflammation hypothesis, significance of the CANTOS results, the promise of new research, financial implications and how a focus on patients who show robust response to a therapy may be a cornerstone of CV medicine in the future.
Read on for insight from some of the leading experts in this area.
Carl J. Pepine, MD, MACC: When and where did the notion of inflammation relating to CVD begin?
Paul M. Ridker, MD, MPH, FACC, FAHA: The idea that inflammation was part of a broad disease process goes back a very long way, to Greek medicine. In fact, prior to the lipid hypothesis, there was a fair amount of interest in this and a group of vascular biologists and translational biologists who were thinking about this. During the explosion of the lipid hypothesis, those biologists kept at it, and it is a good thing that they did. I came in during the clinical/translational period about 20 years ago, but my colleague, Dr. Libby, was involved before that.
Peter Libby, MD: Dr. Ballantyne and I were working on inflammation in vascular biology at a time when cardiologists were not really embracing this concept. I was working on cytokines; Dr. Ballantyne was working on adhesion molecules. I think he had trouble getting funded because people questioned why a cardiologist would be interested in endothelial leukocyte adhesion molecules. I had difficulty getting published in the cardiology literature, but I continued working and observed that vascular wall cells — the endothelial and smooth muscle cells — were not only responders to families of pro-inflammatory mediators such as cytokines including IL-1 beta, but could actually produce these pro-inflammatory molecules. In 1986, my work that implicated IL-1 beta was published in the American Journal of Pathology because the cardiology literature was not interested in this topic.
Christie M. Ballantyne, MD, FACC, FACP, FAHA, FNLA: When I was a resident, it interested me that we had patients with terrible vascular problems, but we were only treating symptoms. It seemed strange that we had disparate risk factors of hyperlipidemia, hypertension and diabetes. How was it all connected? In the late 1970s and early 1980s, it became apparent that the connection was at the level of the vasculature. All these factors altered endothelial vascular function. One of the important elements in the vasculature is the regulation of inflammation. This is quite complicated but has been a very exciting area to follow.
Translation to clinical medicine
Pepine: What is involved in the management of inflammation, in terms of therapeutics?
Robert A. Harrington, MD, MACC, FAHA: The clinical and interventional communities became interested in the vascular response to injury during early days of balloon angioplasty for two reasons. No. 1, with balloon angioplasty, acute thrombotic risk needed to be dealt with, and there was not much to help with it in the late 1980s and early 1990s aside from heparin and aspirin. No. 2 was the restenosis problem, a more subacute issue that plagued interventional cardiologists for many years.
During those early days thinking about restenosis, there were many clinical trials, including of anti-inflammatory agents, corticosteroids and anti-adhesion molecules. That’s when the interventional community began to think about inflammation, but it was as an acute response to injury, not necessarily as the pathogenesis to atherosclerosis.
Ridker: That is a very important point. In the mid-1990s, there were assumptions that the ischemia was driving the inflammation. The CRP story came out of literature showing that after acute coronary ischemia, CRP levels would go up. The problem was, it was a chicken-and-egg issue. Was the ischemia driving the inflammation, in which case it might not be modifiable, or was the inflammation there years in advance of the ischemia, in which case this was a risk factor or risk marker?
That is where my group came in. Our first paper in this field was in 1997. We established that CRP elevation was there years in advance of men having their first-ever MI. They had no underlying ischemia or other illness. Then, in one of those serendipitous-luck issues, that was a trial of aspirin, and it showed an interaction between an antiplatelet, anti-inflammatory drug, and the outcome. That began an exploration of whether other agents had this property. The second paper in this field showed that if you had an underlying pro-inflammatory response and you were on a statin, you had a better reduction in events. That was in 1998, which eventually led to the JUPITER trial, which found that if a patient had a high CRP level, they probably ought to be on a statin. What JUPITER accomplished was getting many more people on a statin. Whether the inflammation piece was answered was secondary, because rosuvastatin is an LDL-lowering drug that has anti-inflammatory properties.
Ballantyne: Inflammation is complex process, and it also turns out a lot of the things we measure have tremendous biological variability and are not easy to measure. Dr. Ridker’s team went through a lot of different markers and determined that CRP is a very stable analyte: easy to measure and well-standardized high-sensitivity CRP assays are available.
This can have major influence on how a field moves. For example, we had many different lipoprotein(a) assays that were not good, which set that field back approximately 20 years.
Challenge of identifying targets
Pepine: Let’s jump forward about 15 years. What was the decision behind targeting interleukin-1?
Libby: There were monoclonal antibodies that neutralized IL-1 beta selectively. I had spent 30 years mining IL-1 from a vascular biology perspective. I knew that Dr. Ridker was designing studies to look at anti-inflammatory agents, so I catalyzed some discussions that led to a trial where we targeted IL-1 beta with a selective monoclonal antibody.
Ridker: We looked at the entire inflammatory cascade. In some way, because CRP is at the bottom of the cascade, it ended up being not the target, but a downstream integrated function of everything that it was going on. It has been a blessing that it worked out that way.
In these targeted anti-inflammatory trials, the idea was: Can we give the target anti-inflammation therapy to the people with an underlying biologic problem? Not just people with a high Framingham risk score or coronary artery calcium score. This was a guiding principle behind the CANTOS trial.
Harrington: This is the critical point in clinical trials. You must have both risk and modifiable risk. This has become a central tenet in biomarker-chosen clinical trials: Making sure that you are identifying something that portends bad outcomes, yet is modifiable.
Expected and unexpected findings
Pepine: For me, the main CV results of CANTOS were expected, but some of the other findings were surprising.
Ridker: We started two trials simultaneously: CANTOS and the NHLBI-funded CIRT trial of low-dose methotrexate. CIRT, which is progressing a bit slower, is designed to evaluate a very inexpensive generic approach that is routinely used to treat rheumatoid arthritis. That trial now has 4,500 mostly U.S. patients randomized.
CANTOS included 10,000 patients from 39 countries, with almost 1,000 physicians involved. Its fundamental finding was that the 150-mg and 300-mg doses of canakinumab yielded the same reductions in IL-6 and CRP. There was no change at all in LDL or HDL and a trivial change in triglycerides. In the absence of LDL reduction, there was still a 15% reduction in major adverse CV events and a 17% reduction in major adverse CV events plus unstable angina resulting in urgent revascularization.
What was really interesting was trying to figure out who benefits. The main results proved the inflammation hypothesis of atherosclerosis and opened up an enormous amount of research into the biology of other agents that target these pathways. They have been controversial in terms of who might be treated, who we can afford to treat and how does the biology inform us of that process.
A pre-specified analysis asked if we can predict benefit the traditional way. The answer was absolutely not; everybody in the trial who had residual inflammatory risk benefited. It turned out that the biologic response to the drug, the magnitude of inflammation inhibition achieved after one dose for reduction of IL-6 or CRP, distinguishes two very different groups of people. We simply divided them at the average reduction in IL-6 and CRP before they got the second dose, and tracked them. If you had a robust anti-inflammatory response, you saw large reductions in risk. That is where all the benefit sits. People with a less robust response still had a reduction in risk, but it was not statistically significant.
In robust responders, the risk reduction never went away no matter which adjustments were made. I think that was because of the underlying biology: lower is better, just like with the LDL hypothesis, and biology really matters. We only targeted people with a pro-inflammatory response. But within that group, the response to the drug was different, and that is not something we typically take into account.
Ballantyne: When we look at on-treatment results in statin trials, outcomes are always influenced by adherence. The unique thing in CANTOS is that all patients got the first dose (given subcutaneously), and then the response was measured. There was a lot of science behind choosing the IL-6 receptor as a pathway. This is a novel approach toward determining who should be getting an expensive biological.
Ridker: Another piece was that certain tumors were more prone to initiation, progression and metastasis in the setting of an inflamed tumor micro-environment. There are 25 years of literature on this in oncology. The most prevalent, of the tumors affected by an inflamed tumor micro-environment, happens to be non-small cell lung cancer. This should not be that big a surprise to cardiologists because if you activate the same NLRP3 inflammasome in the lung, that is going to drive this. Crystalline structures activate that, including silica and asbestos, as well as cigarette smoking and breathing diesel fuel. All of those are associated with increased risk for lung cancer.
In CANTOS, there were a lot of current and former smokers, so we knew we would see a lot of lung cancer. Cardiologists are aware that CRP is a pretty good predictor of future vascular risk. Our oncology colleagues view CRP as a predictor of cancer risk for the same reason. We established an oncology review committee on day 1. There was a 51% reduction in all-cause cancer mortality, almost entirely driven by non-small cell lung cancer, which fits the biology. Interestingly, there, we did not see a dose ceiling.
We have to interpret this carefully. This was chemoprevention, which is not the way we treat patients with cancer. Everyone was free of cancer at baseline, but in a group of 10,000 people, some are going to have underlying tumors that are undiagnosed.
The fact that inflammation is turning out to be a disease process modifiable for multiple diseases also makes sense. We had a highly significant reduction in incident arthritis and in incident osteoarthritis. The drug also eliminated gout, which makes sense because gout is the canonical NLRP3 disease.
Harrington: I must stress that CANTOS patients were very well-treated: high use of antiplatelet agents, statins, other anti-anginals, revascularization, ACE inhibitors and angiotensin receptor blockers. The LDL was 81 mg/dL, lower than observed at baseline in FOURIER.
Libby: I would like to raise two points. No. 1, like any powerful drug, we can see a downside: the infections. No. 2, the cancer result is exploratory. It was not a primary pre-specified endpoint, although when Dr. Ridker and I learned of it, neither of us were surprised. My lab showed in 1994 that IL-1 regulates activity of an enzyme, MMP-2, that degrades the basement membrane that tumor cells have to traverse to invade and metastasize.
Pepine: We should talk about the infection issue, as well as using the response of the first dose of canakinumab as a predictor of who might be best suited for this expensive therapy to reduce mortality by almost one-third.
Ridker: Not unexpectedly, any anti-inflammatory like this is going to have some hazards along the infectious side. There was a modest, statistically significant increase in fatal infections. Interestingly, it was not opportunistic infections or reactivation of tuberculosis. It was run-of-the-mill staph and strep infections occurring in older individuals. If this is going to become a therapy, cardiologists will be more likely to send someone home with azithromycin, like rheumatologists do when they prescribe a tumor necrosis factor inhibitor.
Importantly, infection risk was not related to magnitude of CV and mortality risk reduction. It’s a bit idiosyncratic.
What could happen is that a patient “samples” the first dose to see what happens. If you get a large reduction in the IL-6 and CRP pathway, perhaps you think about using the drug long-term, knowing you might maximize the benefit but the risk will not be any higher. The cost picture would be radically different because you would know you would get the benefits. For less robust responders, we would know the therapy would be less effective medically and from a cost perspective. The hazard would be the same, so perhaps you would not give the drug to those patients.
It’s a challenge from a regulatory perspective because that is not what is typically done. But in this era of expensive monoclonal antibodies, maybe we have reconsider how to allocate drugs that do work, keeping in mind who will benefit the most.
Pepine: This is a different approach from what we do now in prevention, which is to treat almost everybody. Treating on a prediction of who might benefit the most based on the anti-inflammatory response to an initial attempt at therapy would be novel.
Ballantyne: You had to have an elevated CRP level to get into the trial. So this would essentially be two layers of precision medicine.
Harrington: The traditional way of looking for an enhanced subgroup is at baseline, because the power of randomization persists. There has always been a challenge in analyses of responders vs. non-responders, which Paul’s team tried to address from an analytical perspective in CANTOS. There is some subtlety here that is going to challenge the way people think about enhanced risk in this two-step process.
It will also be interesting to see how regulators will grapple with this if Novartis takes this forward. The cost question will be difficult to answer, but responder/non-responder analysis is what everyone wants but carries some challenges in interpretation since it is a post-randomization analysis.
Ridker: The epidemiological challenge is fascinating. I think we missed the boat on this 15 to 20 years ago. Yes, it’s more pristine to have a pre-randomization biomarker. But the fact is, we don’t have any because they don’t work. Also of note is that we adjusted for everything we could think of, and then did a causal inference analysis, but the point estimates did not move. When you have a powerful drug and a powerful biomarker, the drug is going to swamp residual confounding. On-treatment LDL matters, because we have a powerful drug lowering LDL, and the analogy holds here.
Libby: We wring our hands a lot about dead ends in CV drug development. To move past this, we need to think more like they do in oncology, where it is routine to evaluate therapies by biomarkers. If we’re going to get out of these dead ends, we need to be more proactive about embracing personalization.
Pepine: Also, more adaptive clinical trial designs.
Ballantyne: In cancer, you do imaging, and if it’s a non-responder, you change therapies.
Ridker: Cardiologists are already doing this; we just don’t admit it. When we start trials of BP targets, if a therapy does not work, we don’t keep the patient on that drug.
Harrington: We’ve done strategy trials for years. What Paul is describing here is a strategy. But now, it’s using a strategy of does the biomarker change. If yes, you go down a certain pathway. We are used to doing that, both clinically and in clinical trials.
Pepine: What are the implications for the future?
Libby: We need to discuss the financial issue. Canakinumab is marketed as an orphan drug for a handful of patients who have rare genetic diseases related to gain of function of the NLRP3 inflammasome that generates active IL-1 beta. It happens to be triggered also by cholesterol crystals and asbestos and silica.
None of us know what the price will be if a broader indication receives approval by regulatory authorities in different jurisdictions. Many have extrapolated from the orphan drug price, but I would be astonished if that were the cost going forward.
Harrington: Which is exactly what I said in the editorial I wrote accompanying the main CANTOS findings in NEJM, that the drug was priced according to the tenets of orphan drug pricing. But now that it might be used to treat a common disease, the pricing is going to have to be restructured to reflect a common disease market. The analysis can be done based on your response analysis, using contemporary cost-effectiveness parameters, to determine what the price should be.
Ridker: What we found was that you end up with two different numbers needed to treat. If they pursue this indication, they will lose orphan drug status. The NNT in our robust responders for the combined primary endpoint is only 16. The NNT for less-robust responders is 57. It’s still statistically significant, but it is a major difference. That is an important issue, whatever the cost is.
Harrington: What we as a society have accepted is somewhere between $50,000 and $100,000 per quality-adjusted life-year. This is not that different from discussions about CAR (chimeric antigen receptor) T-cell immunotherapy (Kymriah, Novartis): If you respond, you pay, but if you don’t respond, perhaps you don’t pay.
Libby: We all agree that there will have to be innovative financial models and policies to move forward in getting effective therapies to the right patients.
Harrington: There is a lot of discussion in the literature about value-based reimbursement.
Pepine: The counter to that is the responders pay for everybody, even the tests upfront.
Harrington: It’s an interesting question, and it may be different depending on type of health care system that you are in. It could be a population-level question for pricing reimbursement. An organization like Kaiser Permanente may think of it that way. An organization like United Healthcare may think of it that way with an accountable care organization population but maybe not with a fee-for-service population.
Libby: We all agree that this is a great victory, and that we are at a big tipping point. When CANTOS results were presented, some people said this was like being in 1994 for statins.
But how do we move forward? We have one target, IL-1 beta, which has been validated. But the world of inflammation is broad. Dr. Ridker mentioned the CIRT trial that is testing methotrexate. There are also trials of colchicine and other agents on the horizon.
Ridker: Dr. Libby and I are interested in taking this from chronic secondary prevention into acute coronary ischemia. The study I’m eager to do is to take the same agent if we can, because we know its safety profile, but to ask our interventional colleagues if they are willing to give an IV dose of canakinumab at the time of primary angioplasty. Then we can do a large, simple pragmatic trial contacting patients every 3 months and give another dose, and create a quick 12-month outcome trial. In that setting of acute ischemia there is an enormous burst of inflammatory response.
Libby: Also pertaining to ACS, we did a preclinical study showing that the mouse equivalent of canakinumab could reduce the inflammatory response to leukocyte mobilization following coronary ligation. So, there is a preclinical study that shows the outcome benefit with canakinumab in the setting of acute ischemia.
Pepine: What about stroke?
Ridker: The stroke benefit in CANTOS was modest at best. Even in the on-treatment analysis, the stroke benefit was greater in responders, but not overwhelming. The difference may have to do with the blood-brain barrier.
Harrington: Another area of interest would be cardiopulmonary bypass, where we have struggled for years to figure out a way to suppress poor outcomes that occur from this procedure.
Ridker: That would be a very exciting area. Remember too that dialysis is an extremely unusual situation where risk is extremely high, LDL does not predict the risk, CRP is a spectacular predictor of the risk and we have very little to offer. So we need to think about whether we can give an anti-inflammatory agent to patients on the dialysis membrane, which is a pro-inflammatory response, three times per week. There will be a lot of interest in that.
Ballantyne: Another area of interest is vasculitis, although uncommon, many of those patients have very high CRP levels and extremely severe atherosclerosis.
New pathways forward
Pepine: What is your take-home message?
Harrington: This is a big step scientifically and clinically in terms of how we’re going to understand and take care of this group of patients. I want see two next steps. No. 1, learning more about the data to determine if there is a path forward for this drug in clinical practice. No. 2, what are the other trials to come with other agents.
Libby: The CANTOS story is a victory for biology. Some of us are very keen to see it adopted clinically, both in chronic and acute coronary disease. But there are many other targets in the pathway that we think merit consideration for interventions. This is a first, important step and one that I hope will come to clinical reality. I hope it is the first of many therapies that will target inflammation and open new pathways to reduce the burden of residual risk with current therapies.
Ridker: At ESC, the congratulations that came from the investigators of 4S was exceptional. In 1994, 4S was the first piece of evidence telling us we need to think about lipid lowering in a very serious way. But it took 15 to 20 years to figure out how to do it best. I think that’s where we are with inflammation. CANTOS has opened the floodgates providing proof that the inflammation hypothesis works. It’s going to take a while for us to figure out the best drug. This is very exciting.
Ballantyne: Putting this in context with everything else happening in primary and secondary prevention, we are in a new era. We are going to use quantitative risk assessment tools for both primary and secondary prevention, which will include ischemic events such as MI and stroke but also HF. And then, we will use biomarkers as part of that process. Afterwards we will individualize therapy depending on risk, using biomarkers to help guide us. This strategy for inflammation can be relevant to the use of other therapies including diabetes.
Ridker: One last point: This is by no means challenging or undermining the LDL hypothesis. Everybody in CANTOS was already on a high-intensity statin. Residual cholesterol risk and residual inflammatory risk concern two very different patient groups that by and large don’t overlap. If a patient’s LDL is still > 100 mg/dL after high-intensity statin therapy, that is residual cholesterol risk and the patient should be considered for ezetimibe or perhaps a PSCK9 inhibitor. There is no competition here. The patients in CANTOS had LDL 70 or 75 mg/dL, but elevated CRP. The reason they were having residual events is a different biology. CANTOS is the first shot at that different biology. Aggressive LDL lowering must remain the bedrock of prevention, along with diet and exercise. CANTOS represents a move forward, not a substitute for anything.
Pepine: I thank all of you for this fascinating discussion. It is clear that this is just the beginning of a novel approach to CVD.
- Galis ZS, et al. Circ Res. 1994;doi:10.1161/01.RES.75.1.181.
- Harrington RA. N Engl J Med. 2017;doi:10.1056/NEJMe1709904.
- Libby P, et al. Am J Pathol. 1986;124:179-185.
- Ridker PM, et al. Circulation. 1998;doi:10.1161/01.CIR.98.9.839.
- Ridker PM, et al. Lancet. 2017;doi:10.1016/S0140-6736(17)32247-X.
- Ridker PM, et al. Lancet. 2017;doi:10.1016/S0140-6736(17)32814-3.
- Ridker PM, et al. N Engl J Med. 1997;doi:10.1056/NEJM199704033361401.
- Ridker PM, et al. N Engl J Med. 2001;doi:10.1056/NEJM200106283442601.
- Ridker PM, et al. N Engl J Med. 2008;doi:10.1056/NEJMoa0807646.
- Ridker PM, et al. N Engl J Med. 2017;doi:10.1056/NEJMoa1707914.
- Sager HB, et al. Circulation. 2015;doi:10.1161/CIRCUALTIONAHA.115.016160.
- For more information:
- Christie M. Ballantyne, MD, FACC, FACP, FAHA, FNLA, can be reached at 6665 Travis St., Suite 320, Houston, TX 77030; email: firstname.lastname@example.org.
- Robert A. Harrington, MD, MACC, FAHA, can be reached at 300 Pasteur Drive, Stanford, CA 94305; email: email@example.com.
- Peter Libby, MD, can be reached at Brigham and Women’s Hospital, 77 Avenue Louis Pasteur, NRB 741G, Boston, MA 02115; email: firstname.lastname@example.org.
- Carl J. Pepine, MD, MACC, can be reached at Cardiology Today, 6900 Grove Road, Thorofare, NJ 08086; email: email@example.com.
- Paul M. Ridker, MD, MPH, FACC, FAHA, can be reached at Center for Cardiovascular Disease Prevention, Brigham and Women’s Hospital, 900 Commonwealth Ave., Boston, MA 02215; email: firstname.lastname@example.org.
Disclosures: Ballantyne reports he is a consultant for Novartis and has received a research grant from Novartis paid to his institution. Harrington reports he receives grants from AstraZeneca, Bristol-Myers Squibb, GlaxoSmithKline, Janssen, Merck, Novartis, Regado, Sanofi Aventis and The Medicines Company, receives personal fees from Amgen, Gilead, The Medicines Company, Merck, MyoKardia and WebMD, receives other support from Element Science and MyoKardia, and serves on the board of directors for Scanadu and SignalPath. Libby declines all personal payments from pharmaceutical and device companies; he reports he has received research support from Novartis. Pepine reports no relevant financial disclosures. Ridker reports he received research grants from Kowa and Novartis, is listed as a co-inventor on patents related to use of inflammatory biomarkers in CVD licensed to AstraZeneca and Siemens, and serves as a consultant/advisory board member for Janssen and Novartis.