COMMENTARY

Jun 5, 2020 This Week in Cardiology Podcast

John M. Mandrola, MD

Disclosures

June 05, 2020

Please note that the text below is not a full transcript and has not been copyedited. For more insight and commentary on these stories, subscribe to the This Week in Cardiology podcast.

In This Week’s Podcast

For the week ending June 5, 2020, John Mandrola, MD comments on the following news and features stories.

COVID Statistics

When I recorded last week there were 1.75 million cases in the United States. Today there are roughly 1.9 million. The rate of growth of 1.08x is the same as last week. US deaths last week were 102,000 and this week they are 109,000; this rate of rise identical to last week’s, at 1.06x.

Justice

This is a cardiology podcast and I will discuss cardiology topics today but it seems tone-deaf and irreverent to drone on without acknowledging the widespread protests about racial injustice.

It’s impossible for people like me to feel the injustices felt by racial minorities. My friend Dr. Andrew Foy, a cardiologist and researcher in Hershey, Pennsylvania, has a unique point of view of the black and Hispanic communities because he promotes boxing. Andrew sent me a slide he made and posted on social media showing the percentage of black and Hispanic graduates of medical school vs the percentage of blacks and Hispanics in the US population. It showed that the percentage of black and Hispanic US medical graduates is 11%. But the percentage of black and Hispanic people in US population is 32%. Andrew’s point is that medical education leaders are not doing enough to close the gap.

Impella

The FDA issued an emergency use authorization for Impella RP for COVID-19 patients with right heart failure. This is classic COVID-19 science—which is not a good thing. Here is a quote from the letter: "Based on extrapolation of data from the approved indication and reported clinical experience, FDA has concluded that the Impella RP may be effective at providing temporary right ventricular support for the treatment of acute right heart failure or decompensation caused by COVID-19 complications, including PE [pulmonary embolism]...."

I’ve previously discussed the lack of RCT-level data in support of the approved indication for Impella, but let’s today focus on the above noted “reported clinical experience.” One is a case report of 59-year-old woman with COVID-19 and right heart failure due to an acute PE. In 2018, the same group has also reported a series of 5 patients with acute PE and right heart failure treated with the EKOS Endovascular System which delivers low intensity ultrasound for the purpose of facilitated thrombolysis using low dose fibrinolytic therapy. These patients all survived. The endpoints of the case series were hemodynamic measures.

Other data cited in support of the Impella RP emergency use authorization (EUA) include a link to an attempted systematic review from Japanese authors who wrote this is in the abstract: “The results evidenced that available data are currently scant to definitively assess the real role Impella RP® in patient with acute PE and refractory shock.”

We could easily have an answer with mechanical circulatory support: the FDA or CMS simply says, if you want reimbursement for this device, do it as part of an adaptive RCT. If it is as good as people think, this would be a small trial.

Cardiology Research in JAMA Internal Medicine

In a meta-research analysis looking at RCTs in interventional and structural cardiology from 2008 – 2019, the authors, led by academic heart surgeon Mario Gaudino, studied the association between design, outcome, and reporting of these trials with industry funding.

  • About half the trials had industry sponsors. Surprisingly, only about half used hard endpoints. And, typical of cardiac RCTs, the mean follow-up was short at one year.

  • Almost one in five trials was not prospectively registered.

  • One major discrepancy existed between the registered and published outcomes in 38.0% of the trials.

  • Commercial sponsorship was associated with statistically significant findings favoring the sponsored treatment; 64% vs 49% with a p-value of 0.02.

  • Spin—language that detracts from a nonsignificant primary endpoint—was present in 65%.

  • Spin was significantly more likely in industry sponsored trials; 80% vs 54%.

This sort of meta-research is not minutia; it is important in the translation of medical evidence at the bedside.

Another Look at EXCEL

JAMA Internal Medicine also published a Bayesian reanalysis of the EXCEL trial, an RCT comparing left main PCI vs CABG in patients with amenable anatomy. Before you get out of sorts about complexities of the word Bayesian, let’s first see how basic it is. First, children are natural Bayesian learners. They constantly use prior information to adjust their behavior. Doctors, too, are natural Bayesians. They also use prior information and adjust the probabilities of something given the prior information.

Bayesian interpretation of trial results differ from the standard null hypothesis significance testing using p-values in that Bayesians estimate the probability of the hypothesis given the results, while we are much more used to the frequentists’ approach of estimating the probability of the results observed given the hypothesis—which is usually no difference or null.

In the 5-year EXCEL results, the paper reports that in the primary endpoint, a composite of MACE, the difference was 2.8% higher for PCI. But the confidence intervals ranged from -0.9% reduction to a 6.5% increase and the p-value was nonsignificant. This nonsignificant difference led the authors to conclude that PCI and CABG performed equally well. And then the conclusion was codified in guidelines.

The equivalence was also problematic because the odds ratio for mortality was 38% higher in the PCI arm (difference, 3.1%; 95% CI, 0.2%-6.1%). Recall that the authors of EXCEL argue that the mortality signal is post-hoc subgroup that is underpowered and likely due to chance.

Prof James Brophy from McGill University took the data from EXCEL and previous studies and did a Bayesian analysis. His results are highly notable, not just for deciding on PCI or CABG in left main disease, but more importantly for the interpretation of any clinical trial.

Brophy first took the EXCEL data on its own. You can call this a noninformative prior. He then took an effect size, in this case an increase in 1% mortality with PCI, and used the trial data to calculate its probability. Brophy found a 99% probability that total mortality for patients with left main CAD is increased with PCI and 94% probability that the increased risk exceeds 1%. He did the same for the primary endpoint of MACE: corresponding probabilities for an increase in MACE were 95% and that it was greater than 1% was 87%; and for major adverse cardiac and cerebrovascular events (MACCE), 99.9% and higher than 99%.

Brophy then did the analysis again using the previous studies. For mortality, the probability that PCI increased mortality at all was 85% and that it increased it more than 1% was 47%. For MACE, the probability that PCI increased events was 96% and that it increased it more than 1% was 86%.

One other point that both Brophy and editorialist Sanjay Kaul from Cedars Sinai brought up was that the 3-year EXCEL results were analyzed as a non-inferiority design with the noninferiority margin set at 4.2 percentage points. But then they changed the analysis method at 5 years to superiority. In the 5-year results, the 2.8% higher rate of MACE with PCI had an upper bound of 6%, which would have exceeded the non-inferior margin. Thus, had the EXCEL authors stayed true to their analytic method, PCI would be deemed not noninferior to CABG.

Lancet and NEJM Retractions

Two of the most respected medical journals have retracted two major observational trials. The Lancet pulled the hydroxychloroquine (HCQ) observational study that got great press and actually caused the World Health Organization to pause the HCQ RCT. And the New Engand Journal of Medicine pulled the study from the same authors on underlying cardiovascular disease and increased risk of COVID-19. That study did not find a harmful association of ACE inhibitors or ARBs.

Retraction of these papers can be directly attributed to sleuth-like investigations from people on Twitter, some with content expertise but others armed only with curiosity. This highlights the value of post-publication peer review. And it should send a warning to healthcare journalists that making conclusions from published papers—especially in the COVID space—is fraught. Even if these papers were pristine, they were still nonrandomized observational studies and should not be used to shred equipoise or make clinical decisions.

Comments

3090D553-9492-4563-8681-AD288FA52ACE
Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as:

processing....