Constellation Pharma (CNST) Thesis

Catalyst: Dec 9, 2019 - ASH presentation w/ updated patient numbers

Back in 2008, when BET inhibitors were novel and exciting science, Constellation Pharma (CNST)
spawned out of the bright fire of epigenetic research with promises of ridding the world of
ailments. 11 years later (during which nearly all BET inhibitors had the time to disappoint their
creators), and once-divorced by Roche, CNST hangs onto the hopes that someone out there has
still not heard of a BET inhibitor. But I’m not certain how much longer their secret can be kept.

Epigenetics: booms and busts

Epigenetics are the next natural and decidedly very lucrative contender area - after genetics - to
address with drugs. Epigenetics are, in essence, the governance of gene activation and silencing.
A “regulatory body”, in the form of various proteins and enzymes, that decides the fate of which
genes will be turned on and off throughout various tissues and cells. These epigenetic pathways
are very often de-regulated in disease, and targeting them is logical and promising.

Where there is promise and excitement though, there is delirium and frenzy, certainly when it
comes to the biotech market. Although CNST’s lead drug CPI-0610 would probably frown upon
being cited in such company, everyone must surely remember a novel class of epigenetic-slash-
chemotherapy drugs called Histone deacetylase (HDAC) inhibitors, like Spectrum
Pharmaceuticals’ (SPPI) Belinostat. Way back when, HDAC inhibitors were all the rage - over 13(!)
were in development, it was a Big Pharma must-have circa 2010. HDAC excitement was real and
frenzied - as real as the sheer disappointment these drugs delivered: the total worth of the HDAC
market today is roughly $97.5 million. After all Big Pharma called it quits, some smaller players
like Spectrum, for whom HDACs were the sole purpose of existence, were left holding the bag.
And this story sounds strangely familiar.

I am not definitively saying a BET inhibitor is like an HDAC inhibitor: one inhibits histone
deacetylation, the other prevents binding to acetylated histone proteins ☺. But, a quick look
through literature will find that the two are strangely similar in terms of the downstream genetic
effects they induce. Some scientists might agree:

The uncanny similarities don’t necessarily end there. BET inhibitors, too, managed to lewdly
seduce pharma - 10+ have been in development recently. Coincidentally, they have also delivered
roughly similar clinical results to the HDACs: little more than gnarly toxicity alongside woeful
efficacy. While many would describe the BET inhibitor party as now familiarly waning, CNST raves
on in the corner. We will try to find out why.

BET proteins

Without going into too much detail, the BET proteins are a step in the chain of events which take
place in preparation for expression of various genes. They bind to acetylated (“open”) chromatin,
summoning transcription factors and helping them express (but also repress) various genes. This
simplified graphic is helpful (the string is DNA, the scallops are histones around which it is wound;
Ac = acetylated lysine residues)

Source

Currently, four BET proteins are known to man - BRD2, BRD3, BRD4 and BRDT, all with very
highly conserved bromodomains - the domains which specifically bind to acetylated chromatin:

Source

It is these bromo-domains which BET inhibitors, like CPI-0610 target. By inhibiting the binding of
the BET proteins to acetylated chromatin, these drugs inhibit the downstream pathways of gene
expression, and thereby the production of various, potentially harmful/deregulated proteins (but
also many others we do not know about).

BET inhibitors might be “a little too early” - or just another “close but no cigar” in science

Everyone who talks about epigenetic inhibitors loves to talk about “undruggable” targets.
Mentioning the undruggable “MYC oncogenes” gets people very excited, but it’s a bit of coy
deception.

The premise for such parallels lies in the fact that some oncogenes (cancer-inducing genes) and
their regulators cannot be “drugged” - for one reason or another - such as structural instability of
the target protein. Continuing on this thought, blocking the machinery of gene regulation at the
histone level, should surely be able to control the menacing oncogenes, such as the nefarious
MYC.

Yes, it may be exactly like setting the house on fire to kill a spider, but it’s technically not wrong,
and you get to say that your drug targets “MYC” and avoid talking about the hundreds of other
genes it affects. It is a self-fulfilling prophecy, too, because under the premise of a general “lights
out” drug, all scientific hypotheses can be corroborated. Once you’ve turned off the lights at the
mains, you can start asking, “is the chair dark?” (ie - are the MYC genes down-regulated?) -
“Yes!”. And so on, for virtually any target of your choosing.

This was the animated talk of proponents of HDACs inhibitors, just like it was the talk with DNMT
inhibitors, etc. BET protein research is no different: we’ve thrown a grenade at gene regulation,
now let’s try to find the spider’s carcass in the dark.

A quick search through literature will tell you BET proteins have already been found to regulate
anything from adiposity, to antioxidant gene expression, and then some - because they are
general epigenetic regulators. Their downstream effects are probably on hundreds genes.

To complicate matters further, recent research, including apt warnings by Gerald Denis - a pioneer
in the discovery of BET proteins - suggests that much of the responsibility for oncogenic
expression actually falls mainly on one BET protein - BRD4. It is for this reason that AstraZeneca
recently opted for a BRD4-specific BET drug, AZD5153, which has much higher affinity for BRD4
than for the other three BET targets. This is probably one of the remaining BET assets that anyone
should care about. CPI-0610 is - you guessed it, a (not very good) pan-BET inhibitor, which is so
2010 - potentially making it Gen 1.0 before it even had the chance to pretend it does something
meaningful.

As a side note - CNST have recently tried to implicitly re-brand CPI-0610 as a BRD4-specific
drug. This is (like many other things) untrue (page S5).

In a publication titled, “Clinical trials for BET inhibitors run ahead of the science”, Gerald
Denis remarks on the familiar crippling hype which had seized the entire BET field, including
questionable science, poorly understood downstream effects, and - of course - a gold stampede
for a clinical exit ahead of unanswered questions. To quote this veteran BET researcher on one,
not insignificant aspect,

“recent excitement has centered on the transcriptional co-activator activity of BET

proteins, particularly for cell cycle and proliferation genes in cancer, but it has long been
clear that BET proteins also function as co-repressors of transcription, depending on the
signal transduction, cellular and gene context. Clinical trials are proceeding without much
regard for numerous important genes that will obviously be transcriptionally activated
upon inhibition of BET bromodomain proteins”

Seemingly not everyone is a proponent of nuking spiders.

BET inhibitors - everyone has got one now, but they probably want their money back
(Roche certainly does)

Virtually every pharma co big and small jumped onto the BET opportunity. While everyone has
now got one, no one has so far had anything much to show for it, other that BET inhibitors have
naturally wreaked havoc on people’s bodies by means of serious adverse events.

If you were to follow up on many of the BET inhibitors from a few years’ back - such as Incyte’s
(the developer of Jakafi & accepted authority on myelofibrosis - more on this later) ’54239, many
have by now been de-prioritized, or put on hold in wait for better science in the field.

Below is a brief summary of the performance of BET inhibitors to date:

BET BRD4- MYC Results/safety to date Current status

Inhibitor BD1 IC50
IC50/Ki (nM)
(nM)

Mivebresib BRD2, 100 R/R solid tumors: 72 patients. 38.5% SD,61.5% PD

(ABBV-075) BRD4, Ki mPFS - 1.8 months.

Abbvie = 1-2.2

BRD3 Ki Grade 3/4 TEAEs were reported in 52 pts (72.2%); No ongoing

thrombocytopenia (30.6%) and anemia (15.3%) were trials
= 12.2
most common; Dose-limiting toxicities included
thrombocytopenia, fatigue, aspartate aminotransferase
elevation, gastrointestinal bleed and hypertension.

R/R AML: 41 patients. mOS -2.6 months

All patients experienced a treatment-emergent adverse

event (AE), most commonly (≥40% patient incidence),
fatigue (56%), dysgeusia (46%), decreased appetite
(44%), diarrhoea (42%), nausea (42%), vomiting (42%).
40/41 patients had grade ≥3 AEs (febrile neutropenia
(37%), anemia (34%) and thrombocytopenia (32%).
33 patients had serious AEs, most commonly febrile
neutropenia (19%).

Birabresib BRD2-4 Advanced Solid Tumors: 47 patients. 6% PR. 2 completed

(MK-6828 IC50 = trials, 2
Merck 92-112 Of 46 patients, 38 (83%) had treatment-related adverse terminated
events (diarrhea, 17 [37%]; nausea, 17 [37%]; anorexia, trials, 1
14 [30%]; vomiting, 12 [26%]; thrombocytopenia 10 withdrawn trial,
[22%]). For cohort A, four of 19 (21%) evaluable patients 1 active
had DLTs at 80 mg once daily (grade 3
thrombocytopenia [n = 3], ALT/hyperbilirubinemia [n
= 1]) and two of three had DLTs at 100 mg once daily
(grade 2 anorexia and nausea with treatment delay > 7
days [n = 1], grade 4 thrombocytopenia [n = 1]).
 BET BRD4- MYC Results/safety to date Current status
Inhibitor BD1 IC50
IC50/Ki (nM)
(nM)

INCB054329 BRD2/1: Solid tumors: 50pts; Lymphoma: 4pts

Incyte 44
PR - 1 patient; diagnosed with new malignancy
BRD2/2: 5
shortly after. SD >6mo - 3 patients; SD <6mo - 25 pts Deprioritized &
BRD3/1: 9
current status
BRD3/2: 1
Doses >20 mg BID were not tolerated beyond cycle 1. unknown
BRD4/1: At the nontolerated dose of 22.5 mg BID, G3/4
23
thrombocytopenia was observed in 2 out of 3 pts in
BRD4/2: 3 cycle 2.

Treatment-related G ≥3 TEAEs were

thrombocytopenia (n=5 [9%]); neutropenia (n=2
[4%]); anemia, elevated aspartate aminotransferase,
hyponatremia, and hypophosphatemia (n=1 [2%] each).
Three pts (6%) had 4 treatment-related serious TEAEs
(thrombocytopenia, n=3; anemia, n=1).

RG6146 BRD4: 37 R/R DLBCL: 19 patients. 37% PD. ITT ORR: 11% Dead.
Roche Discontinued
Grade (G) 3/4 treatment‐related toxicities included Oct 2019
thrombocytopenia (n = 2, both G4), neutropenia
(n = 1, G4), fatigue (n = 1, G3), hyperglycemia (n = 1,
G4), hyperbilirubinemia (n = 1, G3), and hyperuricemia
(n = 1, G3)

35 300 R/R Solid tumors: 72 patients. Of 61 evaluable

patients from dose escalation, 43% SD, 57%PD.
Molibresib mPFS 1.8 months 2 Active studies
(I-BET762) - NUT
Most common grade 3/4 TEAEs related to carcinoma;
GSK
mivebresib were thrombocytopenia (35%) and advanced
anemia (6%). Dose-limiting toxicities included lymphoma
thrombocytopenia (2 mg daily; 4.5 mg M-W-F),
gastrointestinal bleed (2 mg daily), hypertension (2–3 mg
4/7)

NHL: 27 patients. CR- 1%, SD - 19%

Common Grade 3 and Grade 4 related events
included thrombocytopenia (n = 19 [70%]), as well as
anemia, asthenia, and increased blood bilirubin (n = 2
[7%] each). Among all subjects, 11 (41%) required dose
reduction for toxicity

AZD5153 BRD4- 33 R/r malignant solid tumor & lymphoma: 28

AstraZeneca specific: patients. 53.6% w/ disease progression R/r solid
- BRD4 1.7 tumors:
specific 25% of pts experienced treatment-related Grade recruiting
≥3 AEs, which were thrombocytopenia and
fatigue (7.1% each), and anemia, diarrhea, and
platelet count decreased (3.6% each). Dose-
limiting toxicities of thrombocytopenia (1 pt) and
diarrhea with herpetic rash
 BET BRD4- MYC Results/safety to date Current status
Inhibitor BD1 IC50
IC50/Ki (nM)
(nM)

CPI-0610 BRD2/1: 180 R/r lymphomas: 64 patients. 2pt CR, 3 pts PR, Trying its best
120
5pt SD. The most frequent treatment related
BRD2/2: 18

BRD3/1: adverse events were thrombocytopenia (45%),

170
fatigue (34%), nausea (27%).
BRD3/2: 25
DLTs - 4 patients, 1 patient w/ grade 3 neutropenic
BRD4/1: fever at 80-mg. Grade 4 thrombocytopenia: 3
120
patients, 1 on 230-mg 1 on 300-mg, 1 on 225-mg.
BRD4/2:18
2/3 patients recovered within 2 weeks.

Myelofibrosis - see below

There are other BET inhibitors in development by smaller pharma or private companies, below is a
list for the purpose of demonstrating their sheer abundance (most certainly not complete). Status
of all these drugs is also not well known - many have already failed or been put on hold.

• NUE7770 (Nuevolution) - disappeared since 2017

• RVX297 (Resverlogix) - also not seen since 2017

• ZEN-3694 (Zenith Pharma) - no updates since 2018

From the table above it is obvious that, while CPI—0610 has roughly comparable binding
constants to the graveyard of fellow BET inhibitors, it is for some reason the “safest” in its class.
This would be great news had this not been a scientific oxymoron, because it is by the virtue of its
sufficient binding to the BET proteins that CPI-0610 should really be exhibiting the class AEs.
Unless, of course, it is under-dosed (which makes it useless), or it doesn’t work for some other
reason, perhaps PK/PD-related.

Where has CPI-0610 actually been since 2008?

Failing and being spun by CNST. We have virtually all the Big Pharma players trying their luck with
a BET inhibitor. And then we have Constellation Pharmaceuticals. Has anybody ever heard of a
BET inhibitor making headlines at a conference? No, because one response out of 64 is not great
reading material.

Big Pharma have no existential necessity to shamelessly market a drug class which turned out to
be not so good - they re-prioritize and move on. For Constellation Pharmaceuticals, unlike for
GlaxoSmithKline, though, this BET inhibitor is all that this company’s got (no, their other drug is
actually worse). So they keep slyly gritting on.

Were you to look up “CPI-0610”, you would find copious popular science literature on this
"breakthrough new compound”, and all other kinds of propaganda across all mediums, only to
discover that you have been subjected to befouling clickbait. Below is why:

(via Biomedtracker)

Not only has CNST tried to throw CPI-0610 at any indication that might stick, their response rates
have fared hardly better than.. really anything else that equally doesn’t work. So that’s what kept
CNST busy all these 11 years, despite them having you believe they were partaking in some
quantum leap.

That and being dumped by Roche for another, better looking BET inhibitor. This BET asset,
RG6146, which so embarrassingly upstaged CPI-0610 and which Roche acquired for $420
million, has officially been discontinued as of October 2019. Some might start sensing a theme
unravel here.

Myelofibrosis - a star is born

Now, while CPI-0610 has not worked thus far, it magically "started working” around the beginning
of summer 2019. CNST's current lardy market cap is a result of a series of events, but mostly
engineered endpoints to make CPI-0610 look like it does anything and some luck (and possibly
some heavy drinking), I shall try to break them down below.

Myelofibrosis (MF) is a type of bone marrow cancer which disrupts the production of red blood
cells, leading to severe anemia and enlarged spleen (due to the spleen's efforts to compensate for
the ill-workings of the marrow), among other things.

Reduction of spleen volume is an FDA-accepted, and commonly used pivotal endpoint in

myelofibrosis studies. The only approved therapies for MF are Incyte’s JAK inhibitor ruxolitinib
(Jakafi) and Celgene's JAKi fedratinib (Inrebic), on the basis of the patients’ SVR35 response (how
many patients had a 35%+ reduction in spleen volume). For Jakafi, roughly 30-40% of all patients
achieve an SVR35, but some are non-responders, while others stop responding after a while.
There is unmet need, for sure. So this is where CNST barges in.

I’m not entirely sure why a drug whose well-documented class effect is serious anemia was
thought of as a great candidate to treat.. anemia. Perhaps the reasoning was that since CPI-0610,
for mechanistically worrisome reasons, does not exhibit class AEs, their lack could be therapeutic.

MANIFEST study

CNST decided to try CPI-0610 in first- and second-line MF, with or without Jakafi. This study was
titled MANIFEST (NCT02158858), with a total target enrolment of 271 patients.

CNST’s first “breakthrough” probably occurred around ASCO-time, in June of 2019, when the
company presented a poster calling CPI-0610 “best in class” (not even kidding). The poster
showed the following study design:

Not FDA endpoints

CNST has been partaking in some endpoint creativity

The protocol featured SVR response (not the standard SVR35 - not an approvable/relevant
endpoint), and Transfusion Dependent (TD) - to Transfusion Independent (TI) (TD-TI) conversion
(definitely not an approvable endpoint) as primary endpoints. It was pretty clear from the start
that CPI-0610 was not even going to succeed on CNST’s own “proprietary in-house” SVR
parameters, so CNST focused on the conversions.

Let's have a look at those. Many MF patients require periodic blood transfusions due to severe
anemia (~1+/month). While Jakafi effectively reduces spleen volume in MF, it tends to worsen
anemia and does little to help take patients off blood transfusions. Image from Sierra Oncology
(SRRA) for some clarity:

TD-TI conversion is not a approvable endpoint, and many companies would kill for it to be

Anyone familiar with the Sierra Oncology story, and with their lead asset momelotinib (and in fact
with many other companies, like CTIC and GERN, who have developed MF drugs in the past), will
know that if the TD-TI endpoint were real, we would certainly know about it by now. Because
SRRA spent countless efforts discussing a regulatory path forward with the FDA for their pivotal
MOMENTUM Phase 3 study (check out SRRA “regulatory clarity” update pres here).

Let this SRRA regulatory update presentation also be the guide for anyone thinking CNST can
score “accelerated approval” or any such other fly-by-night talk travelling around in these times. I
think, even the current FDA has a bare minimum standard of efficacy requiring things to not be
placebos in order to get approved.

Anyway - while SRRA have most probably aspired for TD-TI conversion endpoints, their wishes
were not granted. For SRRA, anemia reduction, haemoglobin improvements, and transfusion
independence (which were of course, better than CNST’s on all accounts, because momelotinib is
an active drug) were the selling points throughout the entire development path of momelotinib.
SIMPLIFY-2 patients (n=104) here were most comparable to the r/r patients CNST recruited in
MANIFEST:

 That’s nearly 50% of patients becoming, and staying transfusion-independent on momelotinib

monotherapy. Just imagine what CNST must have showed, to warrant a market cap 49 times
that of SRRA. Here you go (ASH abstract, arms 1-2 2019):

What is this

6% “best” SVR35
responders

TSS: total symptom score

Before getting to the really obnoxious stuff, let’s note CNST's self-appointed “selling point”: the
TD-TI conversions. 4/13 in the combination arm, equalling to ~31% in a tiny, preliminary patient
update.

Recall momelotinib had a 46.6% response in 104 patients as monotherapy, and SRRA were still
not allowed to use TD-TI conversion as a pivotal endpoint, probably because this is a
heterogeneous parameter that does not accurately predict improvement. And even if it were -
what good would a 31% (which will likely become half of that with 2x patients), conversion rate be
in the current clinical setting?

Play silly games, win silly prizes

As for the rest of this convoluted chart - for anyone wondering, no of course “best response” is
not a thing. If it were, studies would be run until a patient has had an amazing momentary
response, for an hour, or for a week, and would all succeed all the time. MF is a heterogeneous
disease with significant fluctuations, so choosing to record best responses is like running a
depression study and only measuring patients’ responses on the day they won the lottery.
Especially for a Symptom Score endpoint.

Does a 17% “best” SVR reduction, or a 39% “best” symptom reduction, mean anything at all, or
is it just a transient artefact, brought about by daily changes, and recorded only at its heights over
an extended time?

It is funny, too, because CNST love to tell the market at conferences that there is “insufficient
followup” for certain patients, when their very own “endpoints” do not respect end-of-followup
readings.

My guess is - CNST resorted to this endpoint because these “responses” were simply not
sustained. And this would not be the first time this happens with drugs that do not work in MF.

But, fine, even if we were to play-pretend that FDA guidelines didn’t exist - according to these
spurious rules CNST created for themselves - CPI-0610 still does nothing as monotherapy here.
There is hardly any % SVR or TSS difference between CPI-0610 monotherapy and Jakafi
combination. So what is the added value of CPI-0610 in second-line MF?

Might these weak “improvement signals” CNST tried their best to engineer just be good old
placebo response? It does look like it. Note the “conference progression” below:

Endpoint ASCO poster ASH abstract ASH update

Number of patients 16 31 ~40

Spleen volume -19.2% -17%

reduction - median
“best”

TSS >50% responders 6/11 (55%) 11/28 (39%)

Patient Global 100% 85%

Impression Chance
(PGIC)

Increase in 100% 50%

haemoglobin >1.5mg/
dl, week 12,
monotherapy

As is to be expected from fluffy conference updates, CPI-0610's “best responses" began to wane
at alarming speeds with increasing numbers of patients, reminiscent of statistically normalised
inter-patient fluctuations. I have left the “ASH update” column blank until December 9, for anyone
desiring further proof that CPI-0610 is not going to cure myelofibrosis.

But CNST says there are other improving parameters, such as cytokines and fibrosis!

For anyone making this point, I would like to ask how can the universal indicators of disease (such
as SVR, TSS, etc), be getting worse, while some cherry-picked biomarkers be improving? It
simply does not work that way.

And anyway - BET inhibition is clearly known to lead to inhibition of all kinds of genes, which
include cytokine expression. IL-8 reduction, which CNST loves to talk about, is a known class
effect of BET inhibitors. Roche’s now-dead drug's cousin, JQ1, has exhibited strong cytokine
reduction metrics, in particularly IL-8.

All this says is that BET proteins are being inhibited (though for CPI-0610 that might already be a
feat). The patients are still not getting better - so what use is even touting this, other than for the
purpose of some well-meaning deception?

For reference, I will throw in how CPI-0610 recently compared to actual drugs (not “best
responses”, either) in similar patient refractory populations:

Nr of pts Prior tx Duration Spleen reduction TSS reduction

CPI-0610
“Best” SVR35-
CNST
31 JAKi
12 monotherapy: 0%
“Best”: 39%
ASH interim evaluable refractory/ weeks
intolerant “Best” SVR35 - w/rux:
15%

“Best” Median SVR:

-17%

Fedratinib
31% responders with 27%
CELG
97 JAKi r/r or 6 months >35% reduction
Ph 2 intolerant

Tagraxofusp
43% with >29% 45%
STML
23 JAKi failed variable reduction

Ph 1/2
or intolerant 21% with >45%
interim
reduction

For additional light, I would beckon anyone to search any recent ASCO/ASH myelofibrosis
abstract to find that absolutely any drug tried in this indication fared better than CPI-0610, using
real endpoints and real followup times.

Without a doubt, if CNST-0610 works in this myelofibrosis population, CNST for the life of them
have not managed to show any of this, and certainly have not shown it to any FDA-approvable
standard. I shall wait with trepidation for anyone willing to point out to me any parameter, thus far,
which could be construed as objective, testable proof that CPI-0610 is improving these patients’
conditions.

MANIFEST Arm 3: Jakafi-naive patients

Arm 3 of MANIFEST - an ASH abstract which contained data on a grand total of four patients,
was the straw that turned this strange bender into full-on debauchery. Recall from the protocol
above, Arm 3 is the cohort of Jakafi-naive patients, where CPI-0610 is being tested in
combination with Jakafi.

These were the words (and chart) which added roughly 1 billion dollars in valuation to what really
was another perishing BET inhibitor - that is $32 million per word. Just don’t tell JK Rowling:

“All 4 (100%) pts on treatment for ≥ 12 weeks achieved ≥35% spleen volume reduction
(median: -52.4%, [range -68.7%, -42.7%]) and all 4 pts (100%) achieved ≥50%
improvement in TSS (median best change: -79.35% [range -90.2%, –70.1%]) (Fig. 1).”

What was wrong with 12-

week time point again?

I don’t know whether anyone is laughing or crying at this point, but both reactions are equally
appropriate - I imagine particularly if you happen to be an oncologist. No, 4 patient studies with
standard of care drugs do not warrant 1.5 billion dollar valuations. But certainly if no one has
paused to reason with the mechanics of a drug “working” in Jakafi-naive patients, but utterly
failing in Jakafi-refractory patients, or in all other indications where it was tried, I shall withhold my
expectations.

If anyone is wondering whether the baseline characteristics of the patients in this study simply
made them more likely to respond to Jakafi - the answer is a sturdy yes. These patients were
healthier across all parameters, their spleens were over half the size of the Phase 3 Jakafi spleens,
and 100% of them had the JAK2 V617F mutation:

Baseline value Jakafi - Phase 3 MANIFEST Arm 3: CPI-0610 +

Jakafi

dose baseline platelet count

200×109 per liter or less: 15mg 125mg - starting dose. Final dose
twice daily
unclear
baseline platelet count
200×109 per liter or more: 20mg
twice daily

Primary endpoint SVR35 at week 48 SVR

Secondary endpoint SVR35 at week 24 TSS

Number of patients Jakafi: 146

11
BAT: 73

Age 67 (35-83) 71 (52-76)

Risk Category (%) Intermediate-2: 40

Int-1: 18

High: 60 Int-2: 64

High: 18

 Baseline value Jakafi - Phase 3 MANIFEST Arm 3: CPI-0610 +

Jakafi

ECOG (%) 0: 40
</=1:90.9%
1: 53

2: 7

3: 1

Myelofibrosis subtype (%) Primary: 53

primary: 72%
Post-polycythemia vera: 33

Post-essential thrombocytemia:
14

Spleen volume Median: 2408

1379 (580-2807)
451-7766

Hemoglobin <10g/dl (%) 45 81.8

Median neutrophil count 11.3

(x10^9/liter)

Median platelet count (x10^9/ 244 368

liter)

History of leukocyte count 38

>25x10^9/liter (%)

Circulating blasts >1% (%) 76

JAK2 V617F mutation (%) 75 100

On a prognosis scale, patients with such baseline characteristics are decidedly more likely to be
Jakafi responders. Note that CNST did not disclose the baseline characteristics of the four 1-
billion-dollar patients, but rather of 11 patients who have been enrolled until that point - thus the
baselines of the 4 could be even more favorable.

Also note that once again CNST did not like the 12-week cutoff point for TSS response (though it
was finally to their standards on the SVR front), so it is more than likely that TSS responses were
different at the actual time point that mattered. I guess they just wanted it to look perfect. While
this may not seem significant, it is an indication of the kind of business this company runs.

Not so fast

Coming back to HDAC inhibitors, I would encourage anyone to browse through ASH and ASCO
abstracts on small patient numbers with MEI Pharma’s (MEIP) HDAC inhibitor pracinostat. ASH
2013, for instance, showed a >50% spleen length reduction in 28/38 (74%) of Jakafi-naive
patients on Jakafi+pracinostat. In 2015, these were the numbers (graph below), alongside
haematological and fibrosis improvements much like CPI-0610, and 80% symptom response
response. Just 28% of these patients were JAK2 V617F-positive.

Great - why isn’t pracinostat approved for MF yet? That's a good question. In 2018, a study
was finally designed in such a way that the added effect of pracinostat to Jakafi could be tested.
In this trial, Jakafi was given to 25 patients for 12 weeks, and pracinostat was added at that time.
These were the results:

“The rate of spleen response (by palpation) was 74%, and that of symptom response 80%. Most
responses occurred prior to pracinostat initiation. Three patients experienced improvement in
bone marrow fibrosis, and one a near-complete molecular response after two years on study
treatment.

These findings do not support continued development of pracinostat in myelofibrosis.”

And that was the end. But it took over 10 years, and that is surely hopeful for CNST. This
company can plough on for as long as endpoints continue to have no regulatory or clinical value,
and no trial of CPI-0610+Jakafi is ever run in a head-to-head Jakafi trail. And the Kool Aid is
continued to be imbibed.

Perhaps a final clue as to what those four Arm 3 patients may have really be responding to is in
the safety,

As of 27 June 2019, the most common treatment-emergent adverse events (TEAE)

observed in ≥ 2 pts include anemia (1 grade 3), fatigue (all ≤ grade 2), and non-cumulative
reversible thrombocytopenia (all ≤ grade 2).

Isn’t.. CPI-0610 supposed to protect MF patients against anemia? It seems to have failed 50%+
of the time. Does anyone know what causes anemia though? Jakafi. And BET inhibitors that work.

Doesn’t Incyte have a BET inhibitor? Why aren’t they doing this?

Excellent question. Not only does Incyte have a BET inhibitor, this inhibitor is more specific and
generally appears better than CPI-0610 (see binding/inhibition constants of INCB054329 in table
earlier).

Incyte is also the developer of Jakafi, and probably knows myelofibrosis better than anyone (my
bet is - better than CNST). They are actively looking for combinations to expand the Jakafi
franchise - so they must be jumping through fire hoops for this combination opportunity. Except,
of course, that they aren’t.

Already as far back as Q1 2018 earnings call, CMO had this to answer about INCY’s BET:

Incyte were “still looking for a way forward” with a Jakafi/BET combination, at the beginning 2018
- which was the last time anyone had heard of this story. Whether this endeavour has already
joined the necropolis of fantastical BET ideas by now is anyone's guess, but let's place our bets:

Does CNST know something INCY doesn’t, or vice versa? I know what mine is.

The Arm 3 Expectations Debate

This, of course, should have never escalated to this point, and it is hard to believe we are now
engaging in this race numbers game but when in Rome..

Arm 3 had investors most bedazzled, so CNST is currently in the hot seat about what can be
cooked up for ASH on December 9th. Expectation downgrades are already about, and sell-side is
throwing 60% response rates around in attempts to “expectation-trip” this event.

The 100% SVR35 response was surely grandiose, but this response is unlikely to hold up, even
for patients extremely likely to respond to Jakafi.

So the most important question now centres around what response rate would be able to sustain
the current binge. The ASH update will feature 10 patients. A 7-8/10 (70-80%) response rate
would begin to sober people up, but, judging by the current comical situation, the bender would
be far from over. At 60%, my faith-in-humanity guess is some earnest believers in CPI-0610 would
start heading home. At 50%, which is basically the unequivocal Jakafi response rate for this
patient population (though looking at pracinostat, it is actually higher in such small studies), the
game should be over.

Epilogue

BET inhibitors were once very promising, but there has just been no pudding proof of their
effectiveness. They are by now just a little passé, passed from indication to indication like a blunt
hammer looking for a nail that would finally yield.

What’s yielded in the current times, much to the luck of CNST, is not myelofibrosis, but rather
inebriated sentiment of investors who have time-travelled into 2017 - where small numbers of
patients at exultant conference updates spark up salacious benders in negative-rate times.

But, like everything else, this too shall pass.

How I would trade CNST

In terms of trading the ASH update, while I trust the placebo-like qualities of CPI-0610, I do not
trust the driving skills of CNST holders, and much less the current market. The float is small and
concentrated into large positions of holders with unwavering determination, and borrow is scarce.
So I would tread carefully as anything can happen with such small patient numbers and investors
indulging in great amounts of psychedelics.

Do I believe that CPI-0610 is not going to cure myelofibrosis? Pretty much. But do I have high
enough conviction that its ASH presentation on December 9th is going to be the moment of truth?
Sadly not. While I will be short for ASH, I think ultimately patience will pay off with CNST as the
smoke dissipates over this phenomenal phantasmagoria with growing patient numbers.

On the bright side though - we probably haven’t seen an opportunity like this since IFRX.