Thread by @realDaveFeldman

1/ #Study #MegaThread

A new study dropped with a trove of data I've long waited for around #CCTA/#CAC scans and #LDL #Cholesterol

✅ >23k studied (!)
✅ Largest sample of CCTA w/ ≥190 #LDL to date (!)
✅ Very uniform study population

Let's unpack...
jamanetwork.com/journals/jaman…

2/ First, be sure to check out this short thread from the lead author, @MaBMortensen.

2 quick notes:
a) While there'll be lots of data appreciated by LDL skepticism, @MaBMortensen maintains LDL-C is still "an important causal RF"
b) Usual epi caveats, etc

https://twitter.com/MaBMortensen/status/1492506722094788615?s=20&t=L182O8x95El-K2bSRkK2Xg

3/ Okay, so if you've followed me a while, you know just how incredibly thankful I am of studies that seek to avoid common risks of selection bias (Even if entirely unintended).

This study had the distinct advantage of categorically scooping its population directly...

4/ "All adults" undergoing CCTA from 1/1/2008 to 12/31/2017. Exclusion was known CAD at time of CCTA and missing info on pretest LDL-C.

Translation: A complete, clearly defined group of all adults with exclusions just about everyone would likely agree on. (In two sentences 👏)

5/ One critique I'd have is that it is unclear how they are determining detected "obstructive" vs "non-obstructive" plaque. Is it a % stenosis? A particular threshold of non-calc plaque volume?

Naturally, I'd always pull for something more objectively quantifiable when possible.

6/ Interesting baseline nums w/ Table 1:

Highest LDL (≥190) had the highest % history of heart disease and smoking (red outline), however, lowest LDL-C (<77) had highest % hypertension and diabetes (blue outline).

Take special note of the purple statin use line at the bottom..

7/ ... We see there are 34.5% in the LDL-C ≥ 190 group at baseline taking statins. That definitely seems suggestive there are likely a number with monogenetic FH given the higher LDL-C persistence to these levels.

(Really would love to see all groups stratified by statin use.)

8/ Now let's jump to Figure 3 as I'd like to step you through it. I'm going to fade out all but the brown "no plaque" bars. As you can see -- these are extremely comparable to each other in associated event rates, whatever the LDL-C levels. (overlay purple lines mine)

9/ "Overall, absence of plaque was associated with low event rates across all LDL-C levels (6.3 [95% CI, 5.6-7.0] per 1000 person-years), with any detectable plaque (calcified or noncalcified) being associated with 2 to 3 times higher event rates."

10/ Now get this -- next to the brown bar will be the event rates for "nonobstructive plaque" but with a CAC=0, which you'll notice is pretty comparable to the "no plaque" anyway.

That's a pretty powerful endorsement for CAC of 0, particularly given the LDL-C spectrum as well.

11/ Adding in the remaining bars shows it gets a lot more complicated from there. However, also note the lowest LDL-C and highest LDL-C show the worst outcomes at all levels of CAC where there are obstructive plaque levels...

12/ (Naturally, I *really* would love if @MaBMortensen stratified by HDL ≥ 50 and TG ≤ 100 to confirm/disconfirm how much atherogenic dyslipidemia associated with much of these outcomes across this LDL spectrum. Just something for the Suggestion Box)

13/ Okay, okay, there's actually a lot more to talk about with this study-- just read it already.

Key takeaways, IMO:

✅ Hello-- physical measurements of disease (ie CCTA) still suggest much stronger predictive power than lipid levels (shouldn't surprise anyone anymore)

14/

✅ But that doesn't mean lipid levels aren't relevant (Hence our coming #LMHRstudy- see CitizenScienceFoundation.org/study)

✅ With that said, #LDL #Cholesterol levels have far less association w/ future events in this study where CAC=0, even where LDL ≥ 190 mg/dL. That's a big deal.

15/ (One last technical caveat -- you may already be aware I like seeing pre-adusted/modeled data reporting before post which we get to an extent with Table 1, but I also prefer more data visualization pre-Cox PH as well, although I likewise acknowledge I'm weird that way.)

16/ Mega grats again to @MaBMortensen, @MichaelJBlaha, @miguelcainzos23, @khurramn1 and team for impressive paper. I'm excited to see what comes next. 👏👏👏

a/ Addendum thread:

https://twitter.com/realDaveFeldman/status/1492876806180859905?s=20&t=CN9UaZTrRIV4EWYh6llfsQ

b/ Addendum thread:

https://twitter.com/realDaveFeldman/status/1492884043167256579?s=20&t=CN9UaZTrRIV4EWYh6llfsQ