Thread by @enricoferroMD

In our @JAMA_current editorial, we tackle a timely question in #CardioTwitter: should we recommend #apixaban or #rivaroxaban as first-line anticoagulant for #Afib?The two #DOACs are not interchangeable for bleeding or stroke. Let's review all RCTs & RWE🧵
jamanetwork.com/journals/jama/…

Why should you care? Why now? 1) >70% of #Medicare pts on anticoagulation are taking #DOACs, practically replacing warfarin 2) Generic DOACs will enter the market in 2022, massively ⬆️ patient access & use 3) Insurance companies may not cover all DOACs in their formularies 💵💊

RCTs first: in ROCKET-AF (n=14K pts), riva was NON-INFERIOR to warfarin for stroke (per-protocol)🧠, w/ similar rates of major/non-major bleed🩸. In ARISTOTLE (n=18K pts), apixa was SUPERIOR to warfarin for stroke (ITT)🧠 AND w significantly⬇️ major bleed🩸, AND even⬇️mortality.

Before you ask 🚨: we cannot just take conclusions from RCTs to claim apixa>riva. Why? ROCKET-AF had higher-risk pts ⚠️ (mean CHADS 3.5 vs 2.1 in ARISTOTLE), of which >50% had prior stroke 🧠(vs <20% in ARISTOTLE), and only 55% of the time at goal INR in warfarin arm (vs 62%) 💊

But yes, one can leverage the common comparator in ROCKET and ARISTOTLE (i.e. warfarin) to run adjusted indirect comparisons. A subgroup analysis in pts with CHADS≥3 found similar stroke rates 🧠 among #DOACs, but significantly ⬇️ bleeding 🩸 with #apixaban. @PORTAL_Research

What about real-world evidence?The largest meta-analysis (605K pts) suggests lowest bleeding risk🩸w apixa (followed by dabiga, then riva), and also lowest stroke risk🧠w apixa (followed by riva, then dabiga) –though this stroke benefit is not consistently found in all studies.

The latest real-world analysis in @JAMA_current studied 580K Medicare pts w/ #Afib, prescribed #apixaban vs #rivaroxaban in 2013-2018. Apixa was associated w/ significantly⬇️ bleeding🩸AND stroke🧠AND mortality. Seemingly, another win on all fronts for apixaban w real world data.

Inevitably, there is residual confounding – no data on med adherence, lab values to monitor kidney function, concurrent use of anti-platelet agents, or clinical appropriateness of dose reduction. The robustness of the study findings to many sensitivity analyses is reassuring.

Statistical significance is nice, but two questions need an answer first 🧐: A) Why does #apixaban consistently have lowest bleeding risk, but not always lowest stroke risk of all #DOACs? B) Can we explain why/how apixa can have BOTH ⬇️ stroke AND bleeding risk than peer DOACs💊?

A) Bleeding events🩸are way more common than strokes🧠. Also, bleeding events cluster in first 30 days after #DOAC initiation, while strokes are longitudinally distributed over observation periods. Thus, most studies may be sufficiently powered to detect bleeding, but not stroke.

B) All #DOACs have half-life of 12 hrs, yet #apixaban is dosed twice daily (BID), while #rivaroxaban is dosed once daily (QD). Thus, riva has larger fluctuations in blood levels, reaching higher supra-therapeutic peaks📈 (=bleed🩸) AND lower sub-therapeutic troughs📉(=stroke🧠)!

But wait... shouldn't #rivaroxaban QD dosing lead to ⬆️ adherence for patients? Intuitively, QD dosing leads to ⬆️adherence (i.e. total number of pills taken💊), but the TIMING ⌚️of missed doses also matters! And BID dosing may be MORE FORGIVING of occasionally missed doses! How?

🚨Pharmacological modeling shows that missing a single dose of once-daily #rivaroxaban produces a critically low trough level that is equivalent to missing ⚠️ 2-3 consecutive doses of twice-daily #apixaban, which is statistically (& practically) was less common! Not as intuitive.

🚨Key Take-Homes🚨
1) RCTs of apixa & riva cannot be compared due to key patient differences
2) Indirect comparisons of RCTs show⬇️bleeding w apixa (over riva)
3) Real-world data also show⬇️bleeding w apixa, some suggest also⬇️stroke w apixa (over riva)
4) BID dosing may⬇️stroke.

As @ZimetbaumP @kardiologykazi and I argue in @JAMA_current, we now have 3 choices: 1) run an ambitious riva vs apixa RCT, which may need >85K pts to detect a bleeding difference💵 2) run yet another observational study 3) allow the current evidence to inform clinical practice.

Based on the latest observational data from @JAMA_current, adding to 10 yrs of RCT & real-world data suggesting #apixaban has significantly ⬇️ bleeding risk than #rivaroxaban in #Afib, are we ready to make a decision & update guidelines? @AHAScience @ACCinTouch @escardio @US_FDA

Thank you @ZimetbaumP and @kardiologykazi for your guidance in writing this @JAMA_current editorial, and @BIDMChealth @BIDMCVFellows @BidmcCvi @SmithBIDMC @rwyeh for helping me refine the skills needed to combine RCT and real-world data to inform clinical decision making.