The figures provided by MADRIGAL are on PP (per protocol). patients who ended the trial with a valid biopsy  and not on the ITT, (intention to treat) population as required in academic publication.

Regarding the 2/3 vs 1/3 ratio of patients during the trial and the baseline datas published during the last congress

the ITT population was  :


A basic analysis of results is then updated including ITT population

Those figures are estimations and should be confirmed as the results will be published.

NAS Score reduction

We can compare the NAS Score reduction by 2 pts with the two other phase 2b trial published using the same endpoint and on ITT population.

MGL 3196                  49%  vs    27%    RR = 1,83

OCA                            45%  vs    21%     RR = 2,14

ELAFIBRANOR*       48%  vs    21%     RR = 2,29 

*  ITT in 3 arms centers as defined in Ph3 protocol ( including unbalanced centers the values are 33% vs 21% RR=1,57)

Looking at the Relative Ratio values, one can see that MGL3196 is just behind OCA and ELAFIBRANOR* in terms of efficacy and as we will see below, the NAS Score reduction in the MADRIGAL trial could be mainly attributed to the reduction of the steatosis grade in the NAS Score components.

Nevertheless NAS Score reduction by 2 pts is not anymore a valid endpoint for NASH candidates.

NASH reversion

We can compare the NASH reversion endpoint with the two other phases 2b trial published using the same endpoint. 

because of strong differences in baselines caracteristics between trials, to compare similar populations we should use post analysis of GOLDEN and FLINT trials,  including balanced centers as they designed their Phase 3

FLINT definition of NASH reversion was not defined at start but in a post hoc study they applied the new definition on a sub group of patients with well-defined steatohepatitis at baseline. 

Because of that we will compare  here the figures based on PP populations and not ITT populations.

Results of NASH reversion according to new definition. Drug result vs placebo results

MGL 3196                  27%  vs    6%    RR = 4,50

OCA                            19%  vs    8%     RR = 2,37

ELAFIBRANOR*     26%  vs    5%     RR = 5,20

*  3 arms centers as defined in Ph3 protocol 

Looking at the Relative Ratio values, one can see that MGL3196 is just behind ELAFIBRANOR in terms of efficacy but as Elafibranor post hoc study was based only on balanced centers it is not easy to compare honestly and  one can say that MGL3196 is quite efficient on that point. 

What is interesting to look at is the ratio of patients resolving their NASH relative to their baseline NAS score

MADRIGAL explain that 65% of patients who resolved their NASH had a NAS score equal to 4 and that only 35% of patients were over NAS 4.

It is interesting because one of the recruitment criterias was a fat fraction mesured by MRI PDFF > 10%,  in facts,  the average fat fraction of liver at baseline published by MADRIGAL is over 20%  (20,7%) . It means that they recruited patients with advanced steatosis only (or mainly).

I found an article published by Z. Permutt T.‐A. Le M. R. Peterson E. Seki D. A. Brenner C. Sirlin R. Loomba

Correlation between liver histology and novel magnetic resonance imaging in adult patients with non‐alcoholic fatty liver disease – MRI accurately quantifies hepatic steatosis in NAFLD

In this article the correlation between fat fraction of the liver mesured by MRI PDFF and the  steatosis grade was established.

MRI‐PDFF (%)     Steatosis Grade

8.90                              1

16.3                              2

25.02                           3

As one can see, recruiting only patients with an average MRI‐PDFF (%) mesured fat just over 20% conduct to have a majority of patients between 2 and 3 steatosis grades

I tried to distribute the baseline caracteristics of patients in the MGL3196 arm, (it is an evaluation based on the published figures and the reality can be different).


As one can see in this simulation, 95% of patients had a steatosis grade >=2 at baseline.

My question is the following :

if MADRIGAL trial patients have mainly steatosis grade >=2 and  that 65% of patients resolving their NASH had a NAS score = to 4 it would imply that the main profile of the patients resolving their NASH was :

  • steatosis grade             2
  • ballooning grade          1 *
  • inflammation grade     1 *

* based on recruitment requirements

And it also mean that resolving their NASH regarding the new definition could be only the result of one point reduction of ballooning, nothing else. It would be interesting to have more data’s on that.

I the patients had mainly steatosis grade >=2 it could explain why reduction of fat in the liver announced at week 12 could help to drop easely the NAS score by 2 pts, by reducing the steatosis grade from 2 to 0 and with less direct action on inflammation. 

Again, i will look forward the detailled figures  as soon the results will be published .

With 23% of resolution on ITT population we are far from the MADRIGAL announcement "potential to resolve NASH in as little as 9 months in 30-40% of patients receiving only MGL-3196 » !  To do that, they’ll need to work on a selected sub population not yet defined.

G Divry

Notice that I am neither a physician nor a biologist or financial analyst, my point of view is only that of an enlightened amateur, so it must be taken for what it is, a questionable point of view

Annexe : published baseline data’s 

MGL baseline

WWW.NASHBIOTECHS.COM  -  Copyright G DIVRY 2015-2016  - Contact and TERMS OF USE