Read on Twitter
B4 jumping to too many conclusions about the summary sentence in recent @JACCJournals article:
"Plaque burden, not stenosis per se, is the main predictor of risk for CVD events and death", read the paper carefully as well as the EDITORIAL BELOW.
https://www.jacc.org/doi/10.1016/j.jacc.2020.10.030#.X89Z9X5lmaA.twitter
"Plaque burden, not stenosis per se, is the main predictor of risk for CVD events and death", read the paper carefully as well as the EDITORIAL BELOW.
https://www.jacc.org/doi/10.1016/j.jacc.2020.10.030#.X89Z9X5lmaA.twitter
Why is the sentence misleading?
Because the study didn't actually directly measure EITHER plaque burden or severity of stenosis.
Instead, it used plaque burden surrogate (CAC, not actually plaque burden), & used disease extent w/binary cutoff (50%) for presence of "stenosis".
Because the study didn't actually directly measure EITHER plaque burden or severity of stenosis.
Instead, it used plaque burden surrogate (CAC, not actually plaque burden), & used disease extent w/binary cutoff (50%) for presence of "stenosis".
Aside from the issues with reporting CAC from a CTA study as "plaque burden" (I won't go further there bc I'm not an expert in CT), what was ACTUALLY measured as a surrogate of "stenosis" was # of vessels with obstructive disease (defined in a binary fashion as >50% disease).
So what's wrong with that?
First, it is clear from studies of coronary flow that stenosis severity prognosticates better at greater degrees of stenosis severity (if a binary cutpoint is used, it probably ought to be used at the severity at which flow is reduced).
First, it is clear from studies of coronary flow that stenosis severity prognosticates better at greater degrees of stenosis severity (if a binary cutpoint is used, it probably ought to be used at the severity at which flow is reduced).
Most would agree that there isn't much difference between a true 60% and a 30% stenosis prognostically, but there is likely a difference between a true 80% and a 50%, even if we are treating all of these medically.
But beyond that, what was done here was a comparison of CAC and "extent of disease as defined by a stenosis severity of 50%".
So that's fair. Except that you need to have enough patients w/greater extent of severe disease to determine if that is less prognostically important.
So that's fair. Except that you need to have enough patients w/greater extent of severe disease to determine if that is less prognostically important.
Well, the population of patients were those referred for CT selecting out a lower risk population by definition (only 13% with typical symptoms in Table 1), with:
**<2% prevalence of 3-vessel disease (and <7% prevalence of 2- OR 3-vessel disease).**
**<2% prevalence of 3-vessel disease (and <7% prevalence of 2- OR 3-vessel disease).**
Here is Table 2 that shows the distribution of disease in the population. Look at the bottom rows. The vast majority of patients had either no CAD or non-obstructive disease.
Further, CAC and extent of disease were collinear.
So if you take a covariate with more spread (CAC) and compare it to a covariate with very little spread and throw both in the model (or do a stratified analysis), is it any surprise that CAC was more predictive?
So if you take a covariate with more spread (CAC) and compare it to a covariate with very little spread and throw both in the model (or do a stratified analysis), is it any surprise that CAC was more predictive?
Notably, even with such a low prevalence of 2 and 3 VD, the extent of disease WAS actually predictive of events.
Bottom line is that this is a very interesting paper that shows that CAC can further risk-stratify patients in a low-risk population with non-severe disease on average, which is a very similar message to the manuscript published about a month ago also in @JACCJournals
But to me the sentence that has been tweeted around stretches what data actually showed.
True multivessel CAD is a prognostically bad thing to have (and BTW usually ALSO has significant plaque burden if we actually measure it).
If you don't believe me, just read the editorial.
True multivessel CAD is a prognostically bad thing to have (and BTW usually ALSO has significant plaque burden if we actually measure it).
If you don't believe me, just read the editorial.
