Thread by @Jbauml

I am going through the CROWN trial presentation, expertly presented at #ESMO20 by @bensolomon1. Also kudos to @EnriquetaFelip @JulienMazieres @TonyMok9 and the rest of the authors (those were the ones I could find on twitter). This is an important study, and analysis is complex

Here, of course, is the slide everyone is talking about. HR 0.28 for PFS favoring lorlatinib. This is undoubtedly an impressive HR, but how do these data compare to prior series and studies? #LCSM

Trying to compare apples to apples as much as I can, here is the BIRC PFS for brigatinib and PFS for alectinib (primary endpoint for that was not BIRC, but investigator assessed PFS). HR is clearly better for lorlatinib, but what about medians?

Median is NE for lorlatinib with 18.3 months median FU. Median for alectinib was 25.7 months and 24 months for brigatinib. As a result, I think it is too early to say if that median is going to be better for lorlatinib - numbers get really small out at the 24 month mark #LCSM

As expected, when you use PFS by investigator, lorlatinib looked even better, HR goes down to 0.21. Still, numbers are quite small at 2 years. Comparing crizotinib, median for criz is 9.1 months in CROWN vs 11 months in ALTA 1L and 11.1 months for ALEX. That's a big difference

Predictably, there is no difference in OS with lorlatinib (just like there wasn't with alectinib or brigatinib)

Going into this presentation, the AE profile was my concern based on prior use of lorlatinib. Patients are going to be on these drugs for a long time. Even minor toxicities are meaningful. Gr3/4 AE rate on lorlatinib is 72%, compared to 41% on alectinib and 61% on brigatinib

What did these toxicities look like? Most common was related to lipid changes. Still unclear if this makes an impact on CV endpoints (hopefully someone is looking into that!). The cognitive effect being 20% is lower than I had expected, but this has a major impact on QOL

I must say I find the bar graph on the L a bit misleading. We know that having a response is associated with better QOL. Responses are longer with lorlatinib. As a result, the late divergence of the curves on the R is expected.

To show that bar graph on the L, which makes it look like QOL is much better with lorlatinib, is a bit much. Curves were overlapping, with early seemingly better QOL with criz, followed by a shift around C7. Presumably pts who had to stop lorla were off by then

Lets start with the easy answer - the study aimed to evaluate if lorlatinib is better than crizotinib. Yes. It clearly is. Is lorlatinib the new SOC? I would advise caution at this point. It is an option (once approved). It has higher tox than the other agents and less FU

It should be discussed with patients when they present with metastatic ALK+ NSCLC. Each of the agents has a distinct tox profile and we can and should involve our patients in this decisionmaking. Congrats to the investigators.

I also want to mention here how amazing it is that a subtype of NSCLC that occurs in 3-7% of pts has now had multiple Ph3 trials completed. That is a triumph of so many groups, but I think we must specifically thank the patients and advocates that helped us get here.

The lung cancer advocacy community is amazing, and it is so wonderful to work with them. It is clear there is so much work left to do @ALKLungCancer @alk_fusion @EGFRResisters @ros1cancer @kRasKickers @METCrusaders