AzurRx

Pharma (AZRX)

Consensus

The current SOC for steatorrhea (fatty stool) due to pancreatitis is porcine lipase.
Many companies have attempted to develop recombinant, bacterial or yeast-
derived lipases but every program so far has run into a brick wall & failed in
comparator studies versus the pig lipase.

AZRX in-licensed a yeast-derived lipase (MS1819) from a lab in France that was
part of the wine&cheese institute. It was discovered in 1998! They have tested
the “stability of the lipase” in pH studies and decided that this lipase will be much
more suitable to withstand human digestion. However, it seems to be not so
simple. The human pancreatic lipase’s sequence is vastly different from 1819;
moreover, human lipolysis is a multi-step process which begins in the stomach
(with gastric lipase) and which requires a sequence of digestion by several
different enzymes. This is why porcine-derived lipase, which is very closely
related to the human one, is able to achieve improved coefficient of fat
absorption (CFA) of ~20-35% at best. Finally, natural pancreatic lipases are able
to cleave a multitude of substrates – very long-chain fatty acids, phospholipids,
galactolipids, cholesterol esters, etc. It is unlikely that a yeast-derived lipase has
this level of multifunctionality, considering that it has evolved in a microbe.

The original French owner of MS1819, Mayoly-Spindler, ran a Ph2a study in
2011, in patients with pancreatitis, and got very modest improvement in CFA.
Vs-baseline results were roughly +9.8% (porcine lipase usually gets ~+35%). A
dodgy and very persistent fact, which has now become a running theme/joke, is
that AZRX keeps using per-protocol (PP) numbers, rather than intent-to-
treat (ITT) in virtually all the data they release. In the French study,
according to AZRX data (which are the only data currently available) three
patients were thrown out “due to protocol violations”, which is of course a code
word for “we didn’t like these patients”. It is possible that, like with allll the other
bacterial lipases tested prior, some patients were blatant non-responders &
AZRX did not like them. Imo, this is the real reason for the rampant per-protocol-
ism. Of interest, I have spoken to someone who was involved with pre-IPO AZRX,
and managed to gaze upon the full French data (which is now AWOL), describing
it as utter rubbish.

On the rampant per-protocolism: sometimes AZRX admit to it, sometimes they
don’t bother in their investor slides. In the only other study AZRX ran in 2018
(the Phase 2a study), MS1819 showed roughly +16.3% improvement vs baseline
in ITT population (which is very modest, particularly for a single-arm open-
label study where patients changed their diet for the study). Here, AZRX threw
out 5 out of 11 patients in their favorite per-protocol machinations.
Amongst whatever patients were “kept” for PP, AZRX touted a single patient
whose CFA improvement was +57%, of course not realizing that they are in this
way admitting the effect in the other 5/11 PER PROTOCOL patients was non-
existent. How did 5 patients (45% of the population!) violate the protocol and
why were they removed? We will never know.
Of note – PP numbers in a study are not accepted by the FDA, naturally. So the
meaningless-ness of the data AZRX have presented is self-explanatory.

Interestingly, after the “very promising outcome” of the Phase2a study AZRX
announced on Sept 24, 2018, Mayoly gave full rights to the enzyme to AZRX in
December 2018, for $250,000 + restricted stock. Clearly, Mayoly thought highly
of the blockbuster-potential data AZRX released.

Perhaps it was also the fact that the main AZRX patents for MS1819 expire in
2019 and 2020.

In my opinion, the ongoing Ph2b OPTION study is going to be the final moment of
truth for AZRX. It seems as though FDA insisted that AZRX run a head-to-head
study vs porcine lipase early on, because they have run into a lot of problems
with bacterial lipases for this exact reason in the past. Anthera’s Sollpura, for
instance, managed to pass 4 large trials and an AdCom and a CRL before
blatantly failing in a head to head study. The current trial design is an open-label
6 week crossover study in 30 patients very similar to the infamous Sollpura one:

Due to the cross-over nature of the study, I think it is going to be very difficult to
fiddle with per-protocol numbers, as patients essentially serve as their own
controls here. Ie – throwing out poor responders from the active arm will fiddle
with their results while they were in the placebo arm, and it will be difficult to
find a match whose disposition will affect both arms equally in the interests of
AZRX.

But it won’t be for lack of trying. Below is the sequence of announcements from
AZRX as to when the data from the OPTION study would be released:

19 Dec 2018: Study initiation announced, “initial data in 2019”
Q1 2019 filing: “topline data in summer of 2019”
July 2019 PR: “topline data in summer of 2019”, along with announcement of
initiation of a (extremely strange/useless/bewildering) combination study of
MS1819+porcine lipase
August 2019 (Q2 results): “topline data in September 2019”

Thus, at about the time that OPTION study results should have been ready (July
2019), AZRX delayed the data to September and announced the initiation of yet
another, brand new study with MS1819 – a combination study with porcine
lipase. As a reminder, as per AZRX, the purpose of a recombinant lipase is to
compensate for the following:

• Some patients are not willing to consume pork products
• The porcine lipase comes in a very large dose/many tablets. Here is a
picture from one of AZRX’s older investor presentations (it has been
removed from the July 2019 one):

o

Perhaps someone out there knows which one of these dire needs a combination
study of MS1819+pork lipase is targeting; despite my best efforts I have been
unable to solve this mystery. “I know what we need in order to address the pill
burden of pork lipase! Adding five recombinant lipase pills to it!” – unlikely to
have said anyone ever.

If any doctor reading this has any possible explanation for what kind of need a
combination of pork lipase + a recombinant one is meeting, I would absolutely
love to hear from you.

Bottom line: we are nearing the end of September with no data in sight from a
30-patient study which should have finished by June/July. 4 months to analyze
30-patient data sounds like a lot of highly laborious analysis is taking place.
AZRX seem to be very meticulous with seeking out the best per protocol analysis
combinations; but I remain skeptical that a cross-over study design is very
accommodating to their favorite exercise of throwing away patients they didn’t
like.

Regardless, one fine September day when a press release finally appears from
AZRX – look out for the magic words, “per protocol”. Perhaps four months of
hard statistical labour yielded a magic combination of numbers which granted
AZRX the possibility of using the words “statistical significance”. However, for
anyone reading, keep in mind that the worlds “per protocol” also render the
results utterly useless.