Moving to a New Era of Clinical Trials

Frequently on rounds,  my colleagues argue that we should not do something to a patient since “there is no evidence that it works.”   This phenomenon of avoiding practice that has insufficient clinical trial evidence is often more common among young trainees in academic settings.    The practice of evidence-based medicine inherently involves integrating doctor’s experience, patient preferences and best available research. In an ideal world, every single question would have been tested in a clinical trial; but in reality this is not possible.   In fact, even the majority of recommendations in practice guidelines in cardiovascular disease are not supported by clinical trials.

Even for questions where there is a dire need for clinical trial evidence, such as adding new therapies to current standard of care or expanding or narrowing indications of existing therapies, multiple barriers remain.   Clinical trials are very expensive. In a recent analysis, the median cost of a clinical trial was estimated at 19 million U.S. dollars.   For pivotal cardiovascular disease trials, the numbers are much higher (north of 150 million USD) since those trials have to be larger and for longer duration to detect clinically meaningful outcomes (e.g. heart attack) and to also compare new interventions to current standards of care.   While cost is the biggest barrier, it is not the only one. Finding patients for trials has been a challenge that often leads to long periods of completion or even worse, aborting the trial due to inability to meet enrollment targets.  Even when patient enroll, they can easily lose interest and eventually dropout. Regulatory hurdles around accessing trial data add to the complexity. Even after successful trial completion, extensive inclusion and exclusion criteria_which often enable the trial to prove a positive outcome_ limit our ability to generalize the findings to many patients who do not fit those criteria.

A lot has changed in the world since we started doing randomized clinical trials in the mid 20th century, whether in science, health, technology, media, or even people’s behavior. Yet, we still do our trials the same.  It’s about time to move into a new era of clinical trials by thinking outside the box.  Smaller, smarter trials are possible. Analysis of big data from the real world could help drive hypotheses that we can test in clinical trials. For example in genomics a technique known as  Mendelian randomization uses genetic variation present as birth as a natural experiment to identify a causal relationship between an exposure and disease. Insights from big data could also help identify clusters of patients that are most likely to benefit from a treatment, have higher chance of having the outcome, or higher chance of having the side effect of the treatment, all of which could inform inclusion and exclusion criteria and increase the chance of having a smaller but more informative trial.

Another potential for innovating in design is by recruiting through direct-to-patient approaches. The Apple Heart Study recruited more than 400,000 people in a very short time and showed that this approach is possible. That should also be a coupled with  new approaches in statistical design as well as re-envisioning how we  ascertain outcomes, by capitalizing on the use of technology and patient engagement though ownership of their data. Changes in regulation are necessary to enable those innovations. Finally, despite the fact that clinical trials are so expensive, the value (yield divided by cost) has been low because we traditionally focused only on strong effects on the primary outcome. With appropriate data sharing of patient level data including those for negative trials, so much more could be learned.

As medical and scientific knowledge continues to increase, the cost of an incremental yield to health outcomes from new interventions will exponentially increase. Healthcare providers should be conscious of their practice of evidence-based medicine by always remembering that not all interventions necessarily require the highest level of evidence.  At the same time, we should re-envision our approaches to clinical research using the tools around us in today’s world to generate better evidence at lower cost.


Reaping The Benefits of a Fitness Program One Year After an Injury

One year ago, I had a back injury. It did not happen when I was skydiving nor when I was skiing on the black diamond trail, but rather when I was trying to get on a hammock! As silly as it sounded, I had to stay home for few weeks and suffer from pain and limited function for long period afterwards. As my pain started to become chronic, I started questioning whether this is the inflection point for me.

They say you don’t appreciate what you have until you lose it.  I have enjoyed good health all my life, so this was the first time I felt disabled.  It made me realize how vulnerable we are.

For the first time, I started to seriously think that my health should not be taken for granted and I should be doing an effort to maintain and improve it.  As a doctor for many years now, you would think that I should’ve learned my lesson earlier, seeing patients every day losing the health they enjoyed, even in more dramatic ways. But, somehow I didn’t.

I work around 80 hours a week between my clinical duties, research work, and other career-building activities. My career always took priority over any effort to maintain my health, such as regularly working out at a gym.  If I ever took time off to disconnect from work, I felt guilty by my lack of productivity. If I ever felt burnt out, I powered through, and then rewarded myself with an unhealthy meal, such as a burger, in front of the TV.

The following day in clinic, I would spend hours preaching my patients how they should eat healthier and exercise at least 3 hours a week!

While this approach served my career pretty well, it was clearly taking a toll on my well-being and the back injury was nothing but a wake up call; so I decided to take action. My priority was improving my cardiovascular fitness, improving my core strength, building more muscle, and losing waist fat.  It was clear to me that if I were to achieve those goals, I really need to prioritize working out at the gym over work.

That meant that even in the face of many passed deadlines and endless work commitments, I would still go to the gym at least four days a week for at least an hour each time.

I knew I needed motivation, so I looked for the nicest and most convenient gym, which unfortunately was also the most expensive. I then convinced my wife and a close friend to join me, but promised myself that I would maintain my schedule without them. Because I was never a gym regular, I felt I would benefit from professional guidance, so I hired a personal trainer.  Michael is 10 years younger than me but probably 100 times more fit. He has a bachelors degree in exercise physiology and has gone through rigorous training of how to become trainer, which I’ve come to discover requires incredible talent in terms of interpersonal skills, ability to motivate, and solid understanding of the science of fitness.

A year later, I’m still a regular at the gym and I achieved all my goals. I feel healthier, more fit and stronger than I ever was.  More importantly, I discovered benefits of fitness that I never thought possible.  My fitness program made me:

1- … healthier 

A year later, I feel healthier than ever. I am able to stand or work on a computer for long hours without having any neck or back aches.

2- .. look better

With the belly fat gone and the shoulders, arms and chest larger, I never looked any better in shirts I’ve had for years.

3- … a better doctor

I finally can preach what I do. I also learned a ton about exercise physiology and how to motivate behavior change.

4- … more productive

Although gym hours took time that otherwise I could spend working, I became more productive in the hours that I am working. An hour at the gym boosts my motivation and focus and allows me to produce more in shorter period of time.

5- … happier

Overall, achieving a more balanced lifestyle made me happier.


Five Lessons From My Heart Attack Patients

In only few months, I leave my clinic where I have been seeing patients with heart disease for the past three years. It was not until I started discussing with them the transition to a new cardiologist that I appreciated the unique relationship we have built.  It has only been few years since we first met in in the Emergency Department in the middle of the night as they were having a heart attack, but somehow this journey feels longer.  I witnessed their heart attacks change their lives in many different ways, as they learned to cope, recover and carry on in life with a story to tell. Being one of the characters in their stories, I learned five lessons about life that will stay with me.


1- You can make any terrible event a wake up call or the beginning of the end.

A big heart attack or sudden cardiac arrest is arguably the most terrible event one could experience. As my patients recovered from such an event and came to see me in clinic, I could tell that the event changed them, either to the better or to the worse. While some lost weight, started training regularly, left their high stress jobs, or decided to travel the world, others became even more sedentary, gained weight, and started lamenting their bad luck and “missed opportunity” for good health.  It was striking to observe those two trajectories of opposite directions. Like my heart attack patients, when you are faced with a terrible event in life, you can either use it as a wake up call to do better afterwards or fall off a ledge and spiral down. You decide your own fate.


2- A supporting family is worth all the medicine of the world.

One patient after another, it became clear to me that the presence of a supporting family member that cares for you during difficult times is worth the most state-of-the-art medicine. It is that person that ensures that your food is low in salt and has no butter, that you do your daily exercise and not miss your medication, that you come to your appointments and ask all the right questions, or that simply hold your hand and tell you it’s going to be okay when things turn sour.


3- You’re as old as you think you are.

I met a 90 year-old woman who had a big heart attack and went for the most aggressive therapies. A year later she still shows to my clinic fully groomed and cheerful telling me she picked up dancing and life could not be any better.  I also met a 50 year-old man who after a small heart attack gave up on enjoying life or hoping for better future and couldn’t be convinced otherwise.


4- Faith, hope, and courage are your best friends when you’re not in control.

When patients are waiting for a high risk surgery or intervention, they simply are not in control of their fate. I found that those that fared well had three unique characteristics. First, they had faith in themselves, their doctors, or God. Second, they always hoped for the best. Third, they had the courage to face a difficult reality when things don’t go as well.  Whenever you’re not in control, let faith, hope, and courage always be your best friends.


5- Write your best story today because you never know when it ends.

While I learned so much from the stories of those patients who made it to my clinic, many did not, and their stories ended in the hospital. As you go through life, live every day to the fullest and write your best story page by page, because you really never know when it ends.


Behind Our Backs: A Flurry of Complementary Health Approaches 

As a cardiologist who trained in a quaternary care hospital, I am used to treating the sickest patients, such as those with large heart attacks, shock and cardiac arrest.  When I go to my weekly clinic, I have to suddenly shift my focus. Much healthier people walk through the door and we spend the majority of the time discussing preventive strategies to reduce their risk of future heart events through prescription treatments and lifestyle changes.

In my clinic, I am exclusively focused on treating or preventing heart disease using a defined armamentarium of evidence-based approaches that I’ve learned over my years of training.  As healthcare providers, we set a cut-off of patient conditions and respective treatments that are “doctor-worthy.” Those are health complaints that are serious enough for us to treat, and their treatments have met high thresholds of evidence to make recommendation guidelines. The reality however is that health is not merely the absence of disease, and patient priorities regarding their health are not always aligned with our recommended item list. They realize that their priorities might be “non-doctor worthy,” so they turn behind our backs to online and community resources for guidance.

If you’re a healthcare provider, the next time you review a patient’s medication list, I encourage you to look at the number of naturals, vitamins, and supplements on it. When I did this exercise myself, I found that around 90% of my patients take at least one non-prescribed item, and often several of them. I then researched the statistics and found that my patients are not far off from the general U.S. population. More than two-thirds of Americans take a vitamin, mineral or supplement. Nearly half of older Americans take vitamins and minerals. Almost 18% of adults take a natural product, including the 7.8% of Americans who take fish oil.  This does not include complementary therapies such as acupuncture, massage, and mind-body practices which are used by more than 30% of adults.

This flurry of complementary health approaches is happening behind our backs. As a result, people are left unguided and sometimes misguided by a flourishing market. For example, there are thousands of ingredients, each being packaged and marketed in hundreds of products. For a single health condition, people can choose from a list of nearly a thousand products. The result is a fruit salad containing the effective and the ineffective, the safe and the unsafe, the appropriate and the inappropriate…



Turning our backs is not the answer. Complementary health approaches could be powerful resource to help with patient’s wellness. Integrating those approaches into mainstream medicine is key. This is why many top academic centers now have integrative medicine departments, and the NIH dedicates an agency for scientific research on the subject.

Guidance is critical for three reasons. First, it ensures that people only use complementary approaches when appropriate. This means that they do not replace treatments by their doctors with less potent or effective approaches and they do not not delay seeking medical care when necessary. Second, it is important to distinguish ingredients and practices with the highest level of evidence for effectiveness for a condition (the minority), from those with evidence for lack of effectiveness or those with insufficient evidence (the majority). Third, guidance regarding safety of naturals, vitamins, and supplements as well as their interactions with prescription medications would help avoid detrimental consequences.

Climbing the ladder starts with a first step. I recently started asking my patients about their vitamins, naturals, and supplements, including why they take them, how they learned about them, and what are the results they’re achieving. You should do the same. You’ll be surprised!


Polygenic Scores in Cardiovascular Disease 

For many decades, we knew that cardiovascular disease and several of its risk factors are heritable. This justifies why we often ask our patients about their family history, but is that truly the best measure for someone’s genetic liability to develop a disease?

Diseases such as myocardial infarction, type 2 diabetes, and atrial fibrillation are called complex traits. This is because their genetic liability comes from the summation of weak effects of many single nucleotide variations across the 3 billion nucleotides in our genome. This is different from monogenic diseases such as hypertrophic cardiomyopathy, which is due to a single but highly potent nucleotide variation in the cardiac sarcomere gene.

Owing to multiple large genome-wide association studies, a better understanding of the human genome and advances in statistical genomics methods, this polygenic liability of disease could now be measured in a single individual to predict where he or she falls on the genetic risk spectrum of disease. Over the past year, two high profile papers in Nature Genetics and JACC showed that polygenic scores can accurately predict coronary artery disease, as well as other diseases such as Type 2 Diabetes and Atrial Fibrillation.

A polygenic score is a number that is normally distributed in the population. Where your score is located on that bell-shaped curve determines your risk or protection from disease. While most of us will be in the middle, the unlucky person who happens to be in the upper tail of the distribution will carry several fold increased risk of CAD compared to the rest of us. This could be your 45 year-old patient, non-smoker with an LDL of 120mg/dL and no clinical risk factors who presents with a STEMI. On the other hand, I always wonder whether my 95 year-old grandpa who died of lung cancer with the cigarette in his mouth but a healthy heart actually was on the bottom tail of the CAD polygenic score distribution?

Just like the systolic blood pressure and the LDL cholesterol, polygenic scores are continuous measures of risk that require drawing cut-offs in order to practically classify patients and treat accordingly.  Defining those thresholds and determining how we act on them will be key for the successful implementation of those scores in clinical care. Anytime we draw thresholds to use in screening or treatment, there will be issues of sensitivity, specificity, outcomes, and value. After all, even cutoffs for risk factors that we’ve understood for decades, such as blood pressure and LDL cholesterol, continue to be debated from one guideline update to the other. The power of the polygenic scores is their wide availability and low cost (around $50 once for all diseases) as well as their ability to classify risk at an early stage in life (practically from birth) before clinical risk factors start appearing, which creates an opportunity to target disease early on before it develops.

The journey to clinical implementation of polygenic scores in cardiovascular disease still requires several steps. First, the current predictive ability of those scores declines in populations of non-European ancestry.  Development of scores that carry similar predictive ability across different ethnicities will be crucial to avoid widening healthcare disparities.  Second, prospective trials testing specific thresholds and interventions will be necessary to prove that implementation of those scores can lead to positive outcomes. For example, would targeting lower LDL thresholds in people with high polygenic scores reduce their risk of CAD?  Third, behavioral psychology studies could inform how young and healthy people react to information regarding their score. Ideally, you would want someone with higher score to engage in good lifestyle activities to mitigate his/her risk but also someone with a low score to not get falsely reassured and pick up poor lifestyle choices.  Fourth, clinical trials of existing and new therapies could be better informed with polygenic score stratification to pick up missed opportunities of benefit. One example would be that people at the upper tail of risk could benefit from treatment that when applied to the average population does not show a benefit.

Precision medicine in cardiovascular disease is happening and polygenic scores are one opportunity to prevent disease early on by targeting specific people at risk. Unlike the heritability informed by a positive family history which rarely changes management, the quantitative aspects of polygenic scores and our ability to validate their impact on outcomes prospectively and in different settings will change how we care for patients in the near future.



Deep Learning in Cardiology

Thirteen years ago in my first anatomy class of Med School, the instructor asked us to make sure our learning is “deep.”

“You need to memorize the names of every single tiny nerve and muscle, because they all will be on your exam. One day you could be surgeons and if you cut out the wrong structure, you can kill someone!” he asserted as we all stood there in fear.

Later in Med School, we were told that half of what we’re learning will be wrong by the time we practice. But one thing we were not told is that the way we learn and the entire premise of what makes a good doctor would also change. For example, “deep learning” itself now means something different to me and to most healthcare professionals. If you are reading this article so far, then you are likely to have seen the term floating around in medical journals.

Deep learning is a type of machine learning in which the computer uses multiple layers of processing to extract features from otherwise vague data input, such as an ECG or a slice of an MRI.  Each layer uses outputs from the previous layer. Through deep learning, the computer simulates the neural network of the brain and is able to learn and make sense of abstraction.

Working at the Broad Institute of Harvard and MIT allowed me to recently be part of a team that uses deep learning to solve important problems in cardiology.  Over several weeks, cardiologists, scientists, and machine learning experts worked in teams to train computers on deep learning models so that they understand data such as medications, ECGs, genomic architecture, and cardiac MRIs of tens to hundreds of thousands of people.

The insights I gained were incredible.  Just like medical students perform better on their exam if they learn “deeper,” the longer you train a computer model, the more it learns and the better it performs in predicting – but it does so at much faster rates than any doctor could ever match. For example, in only two days, we trained a computer to read the ejection fraction from a cardiac MRI as good as a doctor would. Using the MRI, the computer could also predict with reasonable precision the presence of hypertension and coronary artery disease, without knowing anything else about the patient. The power of computer vision is beyond imagination. While it could take you a full day to read 100 ECGs, a well-trained deep learning model could read them in only few seconds. It could also identify patterns in the data that the human eye could not discern, which might or might not be biologically or clinically relevant.

As data availability and computing power continue to grow, we will be seeing more and more applications of deep learning in cardiology.  While we do so, we should stay mindful that human supervision and our role as doctors in charge of our patient’s health is more important than ever. This requires us to understand how computers work and how those models are built through working with multidisciplinary teams. If we do this right, we can probably do less deep learning ourselves by delegating to computers, and gain a whole lot of extra time that we can invest in taking care of our patients.



What Can Cardiology Learn from Impressionism?

A Sunday on La Grande Jatte — 1886 Georges Seurat

At the end of three inspiring days at the American Heart Association Scientific Sessions (AHA18) in Chicago, I took advantage of my late night return flight to spend the afternoon at the Art Institute of Chicago. The museum has one of the finest collections of impressionist paintings, and I’m a big fan.

Impressionist artists in the mid-1870s in France challenged the artistic traditions of their predecessors. They depicted spontaneity in their paintings by capturing moments of daily life of regular people, focusing on nature, and using bright colors and rapid brush strokes.

The mid-1880s marked the end of Impressionism. Fathers of the movement started challenging the very basic conventions they helped establish. Claude Monet, for example, traveled outside Paris and instead of painting spontaneous moments, his paintings started reflecting thoughtful and deliberate approach with series of paintings of the same subject to reflect all the level of detail. In the last impressionist exhibition in Paris in 1886, Georges Seurat exhibited A Sunday a La Grande Jatte which used pointillism, a scientific technique of painting deliberately distinct from the more intuitive approach of impressionists. It was a challenge to the impressionist movement and marked its end and the start of a post-impressionist era.

The American Heart Association brings breakthrough science in cardiovascular disease to the art of cardiology practice. At the Scientific Sessions every year, you get to see practice-changing clinical trials, which are often the result of at decade or more of pre-clinical and clinical development. Despite that excitement, adoption of new evidence-based therapies remains slow.  While economic, drug-specific, and prescriber-specific characteristics play a role, we are sometimes shackled by our habits.

Impressionists revolted on the habits of the past and brought a whole new approach to painting. When they no longer needed it a decade later, they evolved quickly into new techniques. One painting, A Sunday a La Grande Jatte, marked the end of an era.

We can learn from that.

If new anti-diabetic drugs such as GLP-1 receptor agonists and SGLT2 inhibitors are showing cardiovascular benefits, maybe we should take more ownership of diabetes management.   If the new cholesterol guidelines recommended lower LDL cut-offs for statin initiation, we should be more proactive about re-evaluating all our clinic patients. And if angiotensin receptor-neprolyhsin inhibitor reduces death in heart failure compared to ACE inhibitor, then we should probably we using it more.   There is often a delay in diffusion of scientific sessions research to clinical practice. We should be always conscious that any delay of implementing new scientific findings to patient care is a  missed opportunity to save lives.

When AHA was founded in 1925, Dr. Paul Dudley White, one of the co-founders, commented, “We were living in a time of almost unbelievable ignorance about heart disease.”  Thankfully nowadays, we have gone so far from that, as we even discuss cutting edge cardiovascular science such as systems-based approaches to drug development, nanotech monitoring in the ICU, and development of an anti-atherosclerosis vaccine.

Like the impressionists, we should continue to challenge the past and present every day, but also free ourselves of habit when necessary,  as we strive “to be a relentless force for a world of longer, healthier lives.”



The 2023 Cholesterol Guidelines

A few days ago, the long-anticipated 2018 AHA/ACC Cholesterol Clinical Practice Guidelines were released at the American Heart Association Scientific Sessions 2018 in Chicago.

The update from 2013 was viewed favorably in the cardiology community, as it reflected a large body of evidence that has accumulated since, specifically the recommendations for targeting LDL< 70 mg/dL in secondary prevention and using non-statin lipid-lowering medications (ezetimibe and PCSK9 inhibitors) with proven incremental reduction in cardiovascular events. In primary prevention, the recommendations for the use of coronary calcium score to decide on statin therapy in intermediate risk patients and the use of several ASCVD risk enhancers in borderline risk patients also reflected a decade of accumulating evidence.

In fact, many cardiologists feel that the new 2018 guidelines finally reflect what they already practice or would like to practice. I definitely feel this way, but are guidelines always meant to come that late after the evidence? Also, should guidelines be static documents at specific time intervals? When writing guidelines that will be used by millions around the globe, it is crucial to strike the right balance in being timely in providing guidance for clinicians but also cautious in not providing premature recommendations based on low levels of evidence, which could result in harm. This is not an easy job and the authors of the current guidelines successfully achieved this balance, in my opinion.

At Scientific Sessions 2018 where the new guidelines were released and made headlines in the morning, new science was being presented in the afternoon showing that these guidelines might already be outdated! The REDUCE-IT trial, which showed that icosapent ethyl 4g/day reduced major adverse cardiovascular events by 25%, was only one example.

I could not but reflect: What will be in the next cholesterol guidelines? How outdated will our current guidelines be if we wait another five years? And if new treatments will target triglycerides and inflammation, should we even change the name to “Atherosclerosis Management Guidelines”?

Here are my predictions for the next set of guidelines:

  • The LDL target cutoffs will be shifted downwards by 20-30mg/dL. In the highest risk patients we will be talking about LDL targets of <50mg/dL for secondary prevention and <70mg/dL for primary prevention. There is accumulating evidence that “lower is better” and that very low LDL (~20mg/dL) is safe, so as we become comfortable with targeting <70mg/dL in the coming few years, it would be reasonable to move the needle even lower.
  • Polygenic Risk Scores (PRS) will be used to risk-stratify patients <40 years of age and target a fraction of the population with high polygenic risk score who would benefit from statin therapy despite their LDL not being  >160mg/dL. The predictive ability of the polygenic risk score for CAD is already established and retrospective data show that statin therapy can attenuate the risk of CAD in those with highest polygenic risk score. Establishing the value of implementing PRS in clinical practice will require prospective randomized trials, and we are likely to see that in the near future.
  • New non-statin therapies to target ASCVD will emerge and have a major role in treatment. Icosapent Ethyl is leading the way, but other triglyceride lowering agents are also promising, specifically inhibitors of Angiopoietin-like 3 (ANGPTL3) and Apolipoprotein C-3 (APOC3). Antisense oligonucleotide inhibitors of  apolipoprotein(a) successfully reduced Lp(a) levels in a Phase 2 trial presented at this year’s scientific sessions. Future phase 3 trials will test whether lowering Lp(a) will reduce CV events. If proven, we might see more emphasis on Lp(a) screening and treatment cut-offs in the next guidelines. Finally and most importantly, the role of heightened inflammation in ASCVD risk is clear. While low-dose methotrexate did not reduce ASCVD outcomes in the CIRT trial, targeted anti-cytokine therapy with canakinumab did improve outcomes in the patients selected for high hsCRP in the CANTOS trial. The next guidelines will likely recommend routine hsCRP screening in secondary prevention to identify patients with residual inflammatory risk (high hsCRP, low LDL) who could benefit from anti-cytokine therapy.
  • And then there’s the atherosclerosis vaccine! A “Cutting Edge in Cardiovascular Science” presentation at Scientific Sessions 2018 by Dr. Klaus Ley highlighted that this is possible in mice. Will it be possible to safely manipulate the adaptive immune system in humans to  create an atherosclerosis vaccine? The answer is probably yes, but it would be wishful thinking to hope for it in the next guidelines.


Those are my predictions. What are yours?