Phar­ma's bro­ken busi­ness mod­el — Part 2: Scrap­ing the bar­rel in drug dis­cov­ery

Biotech Voic­es is a con­tributed col­umn from se­lect End­points News read­ers. Com­men­ta­tor Kelvin Stott reg­u­lar­ly blogs about the ROI in phar­ma. You can read more from him here.


In Part 1 of this blog, I in­tro­duced a sim­ple ro­bust method to cal­cu­late Phar­ma’s In­ter­nal Rate of Re­turn (IRR) in R&D, based on­ly on the in­dus­try’s ac­tu­al his­toric P&L per­for­mance.  Fur­ther, I showed that Phar­ma’s IRR has fol­lowed a rapid and steady lin­ear de­cline over 20 years, which is con­sis­tent with re­cent es­ti­mates from BCG and De­loitte, and can be ful­ly ex­plained by the Law of Di­min­ish­ing Re­turns as a nat­ur­al and un­avoid­able con­se­quence of pri­or­i­tiz­ing a lim­it­ed set of in­vest­ment op­por­tu­ni­ties while each new drug rais­es the bar for the next.  Fi­nal­ly, I showed that a sim­ple ex­trap­o­la­tion of this ro­bust lin­ear trend means that Phar­ma’s IRR will hit 0% by 2020, which im­plies that the in­dus­try is now on the brink of ter­mi­nal de­cline as it en­ters a vi­cious cy­cle of neg­a­tive growth with di­min­ish­ing sales and in­vest­ment in­to R&D.

Here in Part 2, I ex­plore the math­e­mat­i­cal re­la­tion­ship be­tween R&D pro­duc­tiv­i­ty, IRR and past and fu­ture P&L per­for­mance in more de­tail.  In par­tic­u­lar, I show how the lin­ear de­cline in IRR ac­tu­al­ly cor­re­sponds to an ex­po­nen­tial de­cline in nom­i­nal Re­turn on In­vest­ment (ROI) as a more di­rect mea­sure of R&D pro­duc­tiv­i­ty, which then leads di­rect­ly to ter­mi­nal de­cline in fu­ture P&L per­for­mance.  I then use this mod­el to run some what-if sce­nar­ios, to ex­plore how much we will need to im­prove nom­i­nal R&D pro­duc­tiv­i­ty/ROI in or­der to main­tain pos­i­tive P&L growth.  The re­sults show that we need a ma­jor break­through right now, in 2018, and even then we will face a pe­ri­od of sig­nif­i­cant con­trac­tion be­fore any re­cov­ery, while any­thing less would be too lit­tle, too late to save the in­dus­try from ter­mi­nal de­cline.

Fi­nal­ly, I iden­ti­fy the sin­gle lim­it­ing fac­tor that is ul­ti­mate­ly re­spon­si­ble for dri­ving the de­cline in R&D pro­duc­tiv­i­ty by the Law of Di­min­ish­ing Re­turns, and I ex­plain why many of Phar­ma’s past and cur­rent strate­gies (con­tin­u­ous im­prove­ment, new dis­cov­ery tech­nolo­gies, in-li­cens­ing, pre­ci­sion med­i­cine, big da­ta and AI, etc.) have all failed and will con­tin­ue fail­ing to ad­dress the un­der­ly­ing is­sue.  More­over, I pro­pose an al­ter­na­tive strat­e­gy that might just solve the prob­lem, but while I have my own spe­cif­ic ideas in this area (not shared here, sor­ry), I hope to stim­u­late more crit­i­cal strate­gic think­ing, self-re­flec­tion and open de­bate in or­der to re­fo­cus the in­dus­try’s at­ten­tion on de­vel­op­ing al­ter­na­tive so­lu­tions to tack­le the un­der­ly­ing is­sue be­fore it is too late.

Nom­i­nal ROI as a di­rect mea­sure of R&D pro­duc­tiv­i­ty

In Part 1, I showed that Phar­ma’s In­ter­nal Rate of Re­turn in any giv­en year x can be cal­cu­lat­ed by the fol­low­ing for­mu­la, based on­ly on the in­dus­try’s ac­tu­al his­toric P&L per­for­mance:

IRR(x) = [(EBIT(x+c) + R&D(x+c)) / R&D(x)]^(1/c) - 1

Where c is the in­dus­try av­er­age in­vest­ment pe­ri­od of 13 years, from an ini­tial R&D in­vest­ment to the re­sult­ing com­mer­cial re­turns.

More­over, I showed that Phar­ma’s his­toric IRR has fol­lowed a rapid and steady de­cline over 20 years, which fits the fol­low­ing lin­ear equa­tion al­most per­fect­ly (R^2 = 0.9916):

IRR(x) = -0.00912*(x-2020)

This means that Phar­ma’s IRR has been de­clin­ing at a steady rate of about 0.9% per year and is pro­ject­ed to hit 0% by 2020.  This ro­bust down­ward trend has re­cent­ly been con­firmed by yet an­oth­er da­ta point from De­loitte, which re­port­ed that Phar­ma’s IRR fell to a new record low of just 3.2% in 2017.

The IRR de­fines an ef­fec­tive in­ter­est rate that pro­vides a more com­plete and ac­cu­rate mea­sure of re­turn on in­vest­ment over time, but R&D pro­duc­tiv­i­ty is best de­fined and more eas­i­ly un­der­stood as a sim­ple ef­fi­cien­cy ra­tio.  In par­tic­u­lar, the nom­i­nal Re­turn on In­vest­ment (ROI) in any year x mea­sures the ab­solute nom­i­nal val­ue of com­mer­cial re­turns vs orig­i­nal R&D in­vest­ment over the av­er­age in­vest­ment pe­ri­od c:

ROI(x) = (EBIT(x+c) + R&D(x+c)) / R&D(x)

As ex­plained in Part 1, note that the ul­ti­mate com­mer­cial re­turns in­clude not on­ly EBIT, but al­so fu­ture R&D spend­ing as an op­tion­al use of prof­its that re­sult from the orig­i­nal R&D in­vest­ment.

Now, by sub­sti­tut­ing this equa­tion in­to the orig­i­nal for­mu­la for IRR above, we can see that IRR is di­rect­ly re­lat­ed to the nom­i­nal ROI as fol­lows:

IRR(x) = ROI(x)^(1/c) - 1

And con­verse­ly:

ROI(x) = [1 + IRR(x)]^c

Fi­nal­ly, by sub­sti­tut­ing the his­toric lin­ear trend above in­to the IRR term of this equa­tion, and the in­dus­try av­er­age in­vest­ment pe­ri­od of 13 years in­to the c term, we get the fol­low­ing for­mu­la, which shows that nom­i­nal R&D pro­duc­tiv­i­ty/ROI cur­rent­ly stands at about 1.2 (i.e., we get on­ly 20% back on top of our orig­i­nal R&D in­vest­ment af­ter 13 years), is de­clin­ing ex­po­nen­tial­ly by about 10% per year, and will hit 1.0 (ze­ro net re­turn on in­vest­ment) by 2020:

ROI(x) = [1 - 0.00912*(x-2020)]^13 ≈ 0.899^(x-2020)

This re­sult is con­sis­tent with an ear­li­er re­port by Scan­nell et al., which shows that Phar­ma R&D pro­duc­tiv­i­ty (in terms of NMEs per $Bn R&D spend) has been de­clin­ing ex­po­nen­tial­ly by about 7.4% per year since 1950 (99% over 60 years).  Note that the 2.6% dif­fer­ence in the an­nu­al rate of de­cline must be ex­plained by a de­cline in the av­er­age com­mer­cial val­ue per NME, most like­ly due to di­min­ish­ing in­cre­men­tal ben­e­fit as each new drug rais­es the bar and re­duces the scope for im­prove­ment by the next, as well as in­creas­ing com­pe­ti­tion from gener­ics and me-too drugs.

The di­rect math­e­mat­i­cal re­la­tion­ship be­tween IRR, nom­i­nal R&D pro­duc­tiv­i­ty/ROI, and both past and fu­ture P&L per­for­mance is il­lus­trat­ed in the fol­low­ing 3 charts.  Note that the trends rep­re­sent­ed by red dot­ted lines in each chart are all ful­ly con­sis­tent with each oth­er ac­cord­ing to the for­mu­lae above, and fit close­ly with the his­toric P&L da­ta as well as re­cent IRR es­ti­mates from De­loitte.

Now we can see clear­ly, in re­al terms, just how fast R&D pro­duc­tiv­i­ty has been de­clin­ing.

Fur­ther­more, we can now use these for­mu­lae to pre­dict the im­pact of im­prov­ing nom­i­nal R&D pro­duc­tiv­i­ty/ROI on fu­ture P&L per­for­mance, ei­ther by con­tin­u­ous im­prove­ment or by mak­ing ma­jor tech­nol­o­gy break­throughs, in or­der to de­ter­mine just how much im­prove­ment is re­quired to main­tain pos­i­tive P&L growth and avoid ter­mi­nal de­cline.

Im­pact of con­tin­u­ous im­prove­ment in R&D pro­duc­tiv­i­ty

The ul­ti­mate goal of con­tin­u­ous im­prove­ment is to im­prove over­all R&D pro­duc­tiv­i­ty over an ex­tend­ed pe­ri­od of time, ei­ther by in­creas­ing the num­ber or com­mer­cial val­ue of new ap­proved drugs, or by de­creas­ing the R&D in­vest­ment re­quired to de­vel­op each new drug, or pos­si­bly a com­bi­na­tion of both.  In any case, change is slow and ef­fi­cien­cy is im­proved on­ly grad­u­al­ly by small amounts each year over many years.

So by how much do we need to in­crease nom­i­nal R&D pro­duc­tiv­i­ty/ROI each year in or­der to main­tain pos­i­tive P&L growth and avoid ter­mi­nal de­cline?  Is it 5%, 10%, 15% or even 20%?  And by when do we need to start mak­ing these an­nu­al im­prove­ments?  Has any­one even asked these ques­tions be­fore?

Be­fore we use the for­mu­lae above to cal­cu­late the im­pact of con­tin­u­ous im­prove­ment on fu­ture P&L per­for­mance, con­sid­er that any im­prove­ments must be ap­plied to the cur­rent base­line.  In oth­er words, we must coun­ter­act the cur­rent an­nu­al de­cline in R&D pro­duc­tiv­i­ty be­fore we can start in­creas­ing over­all R&D pro­duc­tiv­i­ty in ab­solute terms.  On that ba­sis, the ex­pect­ed im­pact of con­sis­tent­ly im­prov­ing nom­i­nal R&D pro­duc­tiv­i­ty/ROI by 5%, 10%, 15% or 20% each year from 2018 is shown in the fol­low­ing charts:

What we can see is that im­prov­ing R&D pro­duc­tiv­i­ty by 5% or even 10% each year from 2018 would slow, but not re­verse the cur­rent de­cline in nom­i­nal ROI and IRR.  More­over, it would make vir­tu­al­ly no dif­fer­ence to the pro­ject­ed ter­mi­nal de­cline in P&L per­for­mance.  Even a 15% an­nu­al in­crease in R&D pro­duc­tiv­i­ty would bare­ly be enough to avoid ter­mi­nal de­cline, and the in­dus­try’s sales and prof­its would still fall by al­most 50%.

In fact, we would need to in­crease nom­i­nal R&D pro­duc­tiv­i­ty/ROI by at least 20% each year to re­verse the pro­ject­ed de­cline in P&L per­for­mance, and even then, the in­dus­try’s sales and prof­its would fall by about one third be­fore they be­gin to pick up again in 2030.  This is be­cause it will take sev­er­al years for any im­prove­ment in R&D pro­duc­tiv­i­ty to trans­late in­to in­creased sales and prof­its due to the long in­vest­ment pe­ri­od.  In oth­er words, the next 10 years of P&L per­for­mance are al­ready large­ly de­ter­mined by the past and cur­rent low lev­els of R&D pro­duc­tiv­i­ty, and there is now very lit­tle we can do about this.

So by when do we need to start mak­ing these an­nu­al im­prove­ments?  The charts be­low show the im­pact of im­prov­ing nom­i­nal R&D pro­duc­tiv­i­ty/ROI by 20% per year from 2018, 2020, 2022 or 2024:

In short, we need to start im­prov­ing nom­i­nal R&D pro­duc­tiv­i­ty/ROI by 20% per year right now, from 2018, be­cause the longer we wait the less im­pact it will have to avoid ter­mi­nal de­cline.

A 20% sus­tained an­nu­al in­crease in R&D pro­duc­tiv­i­ty is a very high tar­get in­deed, which would re­quire in­creas­ing the num­ber or av­er­age com­mer­cial val­ue of new ap­proved drugs by 20% each year, or de­creas­ing the R&D in­vest­ment re­quired to de­vel­op each new drug by 20% per year.  So is it achiev­able?  Could any, or even all of Phar­ma’s strate­gies for con­tin­u­ous im­prove­ment ever make this much im­pact?  Con­sid­er that none of Phar­ma’s past ef­forts at con­tin­u­ous im­prove­ment has made any dif­fer­ence at all to the rapid and steady de­cline in R&D pro­duc­tiv­i­ty over the last 60 years.  In fact, the im­pact of Phar­ma’s past ef­forts is al­ready in­clud­ed in the cur­rent de­clin­ing base­line, so how rea­son­able is it to ex­pect that any of Phar­ma’s cur­rent strate­gies for con­tin­u­ous im­prove­ment will in­crease R&D pro­duc­tiv­i­ty by an ad­di­tion­al 20% each year, on top of what we have been able to achieve in the past?

I will leave this ques­tion open for read­ers to re­flect, mean­while let us now con­sid­er the po­ten­tial im­pact of a ma­jor break­through in R&D pro­duc­tiv­i­ty.

Im­pact of a ma­jor break­through in R&D pro­duc­tiv­i­ty

Un­like con­tin­u­ous im­prove­ment, which re­quires mak­ing in­cre­men­tal an­nu­al im­prove­ments in R&D pro­duc­tiv­i­ty over many years, new tech­nolo­gies have the po­ten­tial to make a sig­nif­i­cant im­pact on R&D pro­duc­tiv­i­ty with­in a short time­frame, and pos­si­bly even with­in a sin­gle year.  Now, we have seen many break­through tech­nolo­gies in drug dis­cov­ery over the years, and not one of these has made any dif­fer­ence to the rapid and steady de­cline in R&D pro­duc­tiv­i­ty, but still let us con­sid­er: What if we could im­prove R&D pro­duc­tiv­i­ty now in 2018 by 100%, 200%, 300%, or even 400%?  What would be the im­pact on pro­ject­ed P&L per­for­mance?

Be­fore we run the cal­cu­la­tions, we must con­sid­er that a ma­jor break­through may pro­vide a one-time jump in R&D pro­duc­tiv­i­ty from the cur­rent base­line, but R&D pro­duc­tiv­i­ty would then con­tin­ue to de­cline at the cur­rent rate of 10% per year be­cause no im­prove­ment is sus­tain­able in the long term due to the Law of Di­min­ish­ing Re­turns.  On that ba­sis, the im­pact of in­creas­ing nom­i­nal R&D pro­duc­tiv­i­ty/ROI by 100%, 200%, 300% or 400% is shown in the charts be­low:

Here we can see that a 100% in­crease (two-fold im­prove­ment) in R&D pro­duc­tiv­i­ty would de­lay the tail end of ter­mi­nal de­cline by on­ly 5 years, while a 200% in­crease (three-fold im­prove­ment) would de­lay ter­mi­nal de­cline by about 10 years, but would not avoid it, and the in­dus­try’s sales and prof­its would still de­cline from their peak in the next cou­ple of years.  Even a 400% in­crease (five-fold im­prove­ment) in R&D pro­duc­tiv­i­ty would on­ly de­lay ter­mi­nal de­cline by 20 years, but at least the in­dus­try’s sales could reach a new high­er peak af­ter a short dip.

I will dis­cuss be­low how we might be able to achieve such a break­through in R&D pro­duc­tiv­i­ty, but as­sum­ing we could in­crease R&D pro­duc­tiv­i­ty by 400%, by when would we need to achieve it?  How much time do we have left to de­vel­op and im­ple­ment such a break­through?

The fol­low­ing charts show the ex­pect­ed im­pact of in­creas­ing nom­i­nal R&D pro­duc­tiv­i­ty/ROI by 400% in 2018, 2020, 2022, or in 2024:

Here again, the bot­tom line is that we need a ma­jor break­through right now, in 2018, be­cause the longer we wait the less im­pact it will have to save the in­dus­try from ter­mi­nal de­cline.

Now, in or­der to eval­u­ate how we might achieve this, we need to take an­oth­er look at the Law of Di­min­ish­ing Re­turns to un­der­stand ex­act­ly what is dri­ving this trend so that we can fi­nal­ly fig­ure out how to ad­dress the un­der­ly­ing is­sue.

An­oth­er look at the Law of Di­min­ish­ing Re­turns

In Part 1 of this blog, I showed that the lin­ear de­cline in IRR can be ful­ly ex­plained by the Law of Di­min­ish­ing Re­turns as a nat­ur­al and un­avoid­able con­se­quence of pri­or­i­tiz­ing a lim­it­ed set of in­vest­ment op­por­tu­ni­ties.  In par­tic­u­lar, I demon­strat­ed that pri­or­i­tiz­ing a lim­it­ed set of ran­dom in­vest­ment op­por­tu­ni­ties by their IRR over time pro­duces a per­fect lin­ear de­cline in IRR, which pass­es right through 0%, ex­act­ly as we have seen with Phar­ma’s R&D pro­duc­tiv­i­ty.  More­over, the IRR plot of pri­or­i­tized in­vest­ment op­por­tu­ni­ties fol­lows a per­fect lin­ear de­cline re­gard­less of their ini­tial dis­tri­b­u­tion.

In fact, the on­ly con­di­tion re­quired to guar­an­tee that a se­quence of in­vest­ments fol­lows the Law of Di­min­ish­ing Re­turns in this way, is that the to­tal num­ber and/or po­ten­tial val­ue of in­vest­ment op­por­tu­ni­ties is ul­ti­mate­ly lim­it­ed.  In essence, there must be some crit­i­cal lim­it­ing fac­tor, which is both ex­haustible and in short sup­ply.

So what could be the ul­ti­mate lim­it­ing fac­tor in Phar­ma R&D?  It is cer­tain­ly not the num­ber of po­ten­tial new drugs it­self, since the num­ber of pos­si­ble drug-like mol­e­cules has been es­ti­mat­ed to ex­ceed the num­ber of atoms in the en­tire so­lar sys­tem.

And it is not the un­met clin­i­cal need or po­ten­tial val­ue of new drugs, since we spend more each year on health­care for our grow­ing and age­ing pop­u­la­tion.  In­deed, there ap­pears to be no end to hu­man suf­fer­ing, and we will al­ways get sick and die at least once in our lives, de­spite med­ical progress.

The re­al an­swer, as I ex­plain be­low, is that we are rapid­ly run­ning out of vi­able new drug tar­gets that could pos­si­bly be ad­dressed with ex­ist­ing ap­proach­es and tech­nolo­gies.

A di­min­ish­ing pool of vi­able new drug tar­gets

Ul­ti­mate­ly, all drugs work by in­ter­act­ing with at least one spe­cif­ic mol­e­cule or “drug tar­get” in the body.  Fur­ther­more, all such drug tar­gets must sat­is­fy all of the fol­low­ing cri­te­ria in or­der to pro­vide a vi­able source of ef­fec­tive new drugs:

  1. Clear cor­re­la­tion or re­la­tion­ship with hu­man dis­ease
  2. Can be tar­get­ed with small mol­e­cules or large pro­teins
  3. Not al­ready ex­ploit­ed by ex­ist­ing ap­proved drugs
  4. Not al­ready test­ed and failed due to mech­a­nism of ac­tion
  5. Com­mer­cial­ly vi­able, linked to a clear un­met need

Ac­cord­ing to the Hu­man Pro­tein At­las, there are 19,613 pro­teins en­cod­ed by the hu­man genome.  Of these, 14,545 (74%) have no known link or re­la­tion­ship with dis­ease, which rules them out as po­ten­tial new drug tar­gets be­cause they fail to meet cri­te­ri­on 1 above.  Per­haps these pro­teins are non-es­sen­tial, as any de­fi­cien­cies can be com­pen­sat­ed by oth­er pro­teins or path­ways; or per­haps they are es­sen­tial, how­ev­er any de­fi­cien­cies are lethal be­fore birth so they nev­er have the chance to cause any dis­ease.  In any case, we have no rea­son to be­lieve that tar­get­ing these pro­teins will do any­thing for any known hu­man dis­ease.

Now of the 5,068 pro­teins that have any link to dis­ease, 3,131 (16% of all hu­man pro­teins) are con­sid­ered to be “un­drug­gable”, ei­ther be­cause they have no ob­vi­ous pock­et ca­pa­ble of bind­ing small mol­e­cule drugs, or be­cause they are in­tra­cel­lu­lar and thus in­ac­ces­si­ble to large pro­teins that can­not pen­e­trate the cell mem­brane.  We must rule out these pro­teins as po­ten­tial new drug tar­gets be­cause we cur­rent­ly have no way to tar­get them, so they fail to meet cri­te­ri­on 2 above.

This leaves on­ly 1,937 po­ten­tial drug tar­gets (10% of all hu­man pro­teins), but 672 of these have al­ready been ful­ly ex­ploit­ed as proven drug tar­gets by cur­rent ap­proved drugs.  Once a new drug tar­get is first iden­ti­fied and ex­ploit­ed by an orig­i­nal first-in-class drug, any “me-too” drugs that fol­low tend to pro­vide lit­tle, if any in­cre­men­tal ben­e­fit or val­ue to pa­tients, and prof­it most­ly by tak­ing mar­ket share from the orig­i­nal drug.  In essence, drug tar­gets are an ex­haustible re­source rather like oil: once we have tapped its po­ten­tial val­ue, it’s gone; we can’t have our cake and eat it.  There­fore, we must al­so rule out these pro­teins as po­ten­tial new drug tar­gets, sim­ply be­cause they are no longer new, and they fail to meet cri­te­ri­on 3 above.

So now we are left with on­ly 1,265 po­ten­tial new drug tar­gets:

At first glance, it seems that we have more than twice as many po­ten­tial new drug tar­gets left to find and ex­ploit as those we have al­ready ex­ploit­ed, so we should not be over­ly con­cerned about run­ning out any time soon.  But what about the oth­er two cri­te­ria, 4 and 5?  How many of these po­ten­tial drug tar­gets have al­ready been test­ed but failed to yield any drugs due to mech­a­nism of ac­tion?  How many have not yet been test­ed, but are still un­like­ly to yield any drugs?  And how many will yield on­ly drugs that are not com­mer­cial­ly vi­able in any case?

Now this is where the num­bers get a bit fuzzy be­cause they are not wide­ly re­port­ed (or at least I could not eas­i­ly find them), but we can make some very rough es­ti­mates.

First, let’s say that about 50% of all drug tar­gets we have ever ful­ly test­ed pro­duced at least one ap­proved drug, while the oth­er 50% failed to de­liv­er any drug at all, due to fun­da­men­tal rea­sons (e.g., safe­ty) based on mech­a­nism of ac­tion.  Giv­en that we now have ap­proved drugs for 672 drug tar­gets, this would im­ply that we have al­ready ful­ly test­ed a sim­i­lar num­ber of drug tar­gets with­out ever pro­duc­ing any drug, so we can rule these out as po­ten­tial new drug tar­gets be­cause they are not new, and fail to meet cri­te­ri­on 4 above.  Fur­ther­more, we can rule out an­oth­er 50% (297) of the re­main­ing 593 untest­ed drug tar­gets be­cause they are un­like­ly to de­liv­er new drugs for the same fun­da­men­tal rea­sons.

Now we are left with on­ly 296 po­ten­tial drug tar­gets, but how many of these will pro­duce drugs that are com­mer­cial­ly vi­able?  It has been es­ti­mat­ed that on­ly about 25% of new ap­proved drugs man­age to ful­ly re­cov­er their own R&D costs and make any com­mer­cial re­turn.  Many of those that fail com­mer­cial­ly are me-too drugs that com­pete for the same drug tar­get, but many are al­so nov­el first-in-class drugs that com­pete with oth­er drugs act­ing by dif­fer­ent mech­a­nisms to tar­get the same dis­ease, or that tar­get dis­eases with in­suf­fi­cient clin­i­cal need.

So let’s as­sume that 50% (148) of the re­main­ing 296 po­ten­tial drug tar­gets are not com­mer­cial­ly vi­able (i.e., do not meet cri­te­ri­on 5 above), and we are now left with on­ly 148 po­ten­tial new drug tar­gets, com­pared with 672 that we have al­ready ex­ploit­ed with ex­ist­ing ap­proved drugs:

Again, this is just a rough es­ti­mate based on some crude as­sump­tions, but still it is clear that we are rapid­ly run­ning out of vi­able new drug tar­gets that meet all 5 cri­te­ria above.  We are lit­er­al­ly scrap­ing the bar­rel for the last re­main­ing drug tar­gets, and chances are we are al­ready work­ing on all these re­main­ing tar­gets in di­rect com­pe­ti­tion with each oth­er.  Now is it re­al­ly any won­der that R&D pro­duc­tiv­i­ty has been de­clin­ing so rapid­ly by the Law of Di­min­ish­ing Re­turns?

Lim­it­ed po­ten­tial im­pact of Phar­ma’s cur­rent strate­gies

Giv­en that we are rapid­ly run­ning out of vi­able new drug tar­gets, it is easy to see why Phar­ma’s R&D pro­duc­tiv­i­ty has been de­clin­ing so rapid­ly by the Law of Di­min­ish­ing Re­turns.  More­over, it is easy to see why none of Phar­ma’s past ef­forts has made any dif­fer­ence, and why none of its cur­rent strate­gies will make any dif­fer­ence, ei­ther: They do not ad­dress the un­der­ly­ing is­sue.

Al­most all of Phar­ma’s past and cur­rent strate­gies are de­signed to im­prove R&D pro­duc­tiv­i­ty in one or more of the fol­low­ing ways:

  1. In­crease the ef­fi­cien­cy by which we iden­ti­fy vi­able new drug tar­gets that meet all 5 key cri­te­ria list­ed ear­li­er
  2. In­crease the ef­fi­cien­cy by which we iden­ti­fy safe and ef­fec­tive new drugs against those tar­gets iden­ti­fied in 1 above
  3. In­crease the qual­i­ty and ex­pect­ed com­mer­cial val­ue of those drugs iden­ti­fied in 2 above

For ex­am­ple, mol­e­c­u­lar bi­ol­o­gy, ge­nomics, pro­teomics and bioin­for­mat­ics have been de­vel­oped to in­crease the ef­fi­cien­cy of tar­get dis­cov­ery by im­prov­ing our un­der­stand­ing of hu­man bi­ol­o­gy and dis­ease, while oth­er tech­nolo­gies like ra­tio­nal drug de­sign, chem­in­for­mat­ics, com­bi­na­to­r­i­al chem­istry and high through­put screen­ing have been de­vel­oped to in­crease the ef­fi­cien­cy of drug dis­cov­ery by ex­plor­ing new chem­i­cal space.  Mean­while, open in­no­va­tion and in-li­cens­ing have been de­vel­oped to source new drugs and tech­nolo­gies more ef­fi­cient­ly than in­ter­nal in­no­va­tion.  Pre­ci­sion med­i­cine with bio­mark­ers and re­al-world ev­i­dence has been de­vel­oped to in­crease the clin­i­cal ben­e­fit and com­mer­cial val­ue of new drugs in spe­cif­ic pa­tient pop­u­la­tions.  Now there is a big push with big da­ta, ma­chine learn­ing and AI to make sig­nif­i­cant im­prove­ments in all these ar­eas.  And of course, con­tin­u­ous im­prove­ment has been Phar­ma’s fa­vorite long-term strat­e­gy to im­prove over­all ef­fi­cien­cy.

Note that none of these strate­gies can in­crease the over­all num­ber of vi­able new drug tar­gets that meet the 5 key cri­te­ria above.  In­stead, they are sim­ply de­signed to ex­ploit the re­main­ing pool of vi­able new drug tar­gets more ef­fi­cient­ly, which iron­i­cal­ly, will on­ly ac­cel­er­ate its de­ple­tion.

These strate­gies have not worked, and will not work, be­cause they do not ad­dress the un­der­ly­ing is­sue: We are rapid­ly run­ning out of vi­able new drug tar­gets that can be tar­get­ed by clas­sic small mol­e­cule drugs or large ther­a­peu­tic pro­teins.

So how can we ad­dress this prob­lem to im­prove R&D pro­duc­tiv­i­ty?

An al­ter­na­tive ap­proach to im­prove R&D pro­duc­tiv­i­ty

Ul­ti­mate­ly, the on­ly way we can break free from the Law of Di­min­ish­ing Re­turns is to in­crease the num­ber of vi­able new drug tar­gets; and the on­ly way we can do this is to re­move or re­lax at least one of the 5 key cri­te­ria list­ed ear­li­er.

At first, it seems that all these cri­te­ria are ab­solute crit­i­cal re­quire­ments for any new drug tar­get.  For ex­am­ple, if there is no clear link with hu­man dis­ease, or if there is no clear un­met need, then there is no vi­able drug tar­get.  Fur­ther­more, if we have al­ready test­ed a drug tar­get and it failed for safe­ty rea­sons, or if we have al­ready ful­ly ex­ploit­ed it with ex­ist­ing ap­proved drugs, then we can­not ex­ploit it fur­ther.  And fi­nal­ly, if we can’t hit a spe­cif­ic drug tar­get with small mol­e­cules or large pro­teins, then we can’t de­vel­op an ef­fec­tive drug against that tar­get.

Or can we?  Are we re­al­ly lim­it­ed to us­ing small mol­e­cules and large pro­teins as drugs to tar­get spe­cif­ic pro­teins and treat dis­eases more gen­er­al­ly?

Small mol­e­cules have the great ben­e­fit that they can pen­e­trate cell mem­branes to reach po­ten­tial drug tar­gets with­in the cell, but on the oth­er hand, they re­quire a clear bind­ing pock­et with­in the tar­get pro­tein, oth­er­wise they have the wrong size and shape to bind ef­fec­tive­ly and specif­i­cal­ly to flat pro­tein sur­faces.  Mean­while, large ther­a­peu­tic pro­teins such as an­ti­bod­ies can form much stronger, more spe­cif­ic in­ter­ac­tions with such flat pro­tein sur­faces, but they are gen­er­al­ly un­able to pen­e­trate cell mem­branes and get in­to the cell.  Thus by lim­it­ing our po­ten­tial drug reper­toire to small mol­e­cules and large pro­teins, we are ef­fec­tive­ly lim­it­ing our pool of po­ten­tial new drug tar­gets to ex­tra­cel­lu­lar pro­teins, or in­tra­cel­lu­lar pro­teins that have a clear bind­ing pock­et.  At the mo­ment, we have no means to tar­get in­tra­cel­lu­lar pro­teins that have no clear bind­ing pock­et, yet there are thou­sands of these “un­drug­gable” pro­teins en­cod­ed by the hu­man genome.

Ac­cord­ing to the Hu­man Pro­tein At­las, 3,131 (about 16%) of all pro­teins en­cod­ed by the hu­man genome are “un­drug­gable” pro­teins that have a clear link with dis­ease, but can’t be tar­get­ed with ei­ther small mol­e­cules or large pro­teins be­cause they are in­tra­cel­lu­lar and have no clear bind­ing pock­et.  This com­pares with on­ly 1,937 drug­gable tar­gets, of which 672 have al­ready been ful­ly ex­ploit­ed with ex­ist­ing ap­proved drugs, and per­haps on­ly 148 re­main vi­able as ex­plained above.  There­fore, we could po­ten­tial­ly in­crease the to­tal num­ber of vi­able new drug tar­gets by as much as 20 fold, if on­ly we could find an ef­fec­tive way to tar­get them.  So how can we do this?

First, it is clear that small mol­e­cules do not have the size and shape re­quired to bind ef­fec­tive­ly and specif­i­cal­ly to large and flat pro­tein sur­faces.  They are sim­ply un­able to com­pete with the tight and spe­cif­ic bind­ing that oc­curs be­tween dif­fer­ent pro­tein mol­e­cules with­in the cell, which is why we have nev­er been able to de­vel­op an ef­fec­tive small mol­e­cule in­hibitor of any known pro­tein-pro­tein in­ter­ac­tion.  There­fore, we are forced to use large mol­e­cules in or­der to com­pete ef­fec­tive­ly with these strong in­ter­ac­tions, but this leaves us with the oth­er prob­lem: How to get such large mol­e­cules in­to cells in the first place?

If on­ly we could find a re­li­able way to get large mol­e­cules in­to cells, then we could po­ten­tial­ly tar­get thou­sands of dif­fer­ent pro­teins and pro­tein-pro­tein in­ter­ac­tions that are cur­rent­ly be­yond reach with­in the cell.  So again, how to achieve this?

The cell mem­brane is no­to­ri­ous­ly dif­fi­cult to pen­e­trate, es­pe­cial­ly by large mol­e­cules, but na­ture has shown that it can be done.  For ex­am­ple, sev­er­al large macro­cyclic an­tibi­otics and bac­te­r­i­al tox­in pro­teins are known to cross the cell mem­brane.  So can we adapt these mol­e­cules to act as drugs once they get in­to the cell?  Or bet­ter still, can we un­der­stand how they get in­to cells in the first place and ap­ply these prin­ci­ples to de­sign a whole new class of cell-pen­e­trat­ing ther­a­peu­tic pro­teins that could be adapt­ed to bind tight­ly and specif­i­cal­ly to any tar­get pro­tein in the cell?  I have my own spe­cif­ic ideas that I would like to pur­sue in this re­gard, but hope­ful­ly it is clear by now that get­ting large mol­e­cules in­to cells is per­haps the on­ly way to ad­dress the re­al un­der­ly­ing is­sue of de­clin­ing R&D pro­duc­tiv­i­ty.  This prob­lem is too im­por­tant to re­ly on just one idea, so we need to pur­sue as many po­ten­tial so­lu­tions as pos­si­ble, in or­der to re­verse the de­cline in R&D pro­duc­tiv­i­ty and save the in­dus­try from ter­mi­nal de­cline, be­fore it is too late.

In sum­ma­ry, Phar­ma R&D pro­duc­tiv­i­ty is de­clin­ing by the Law of Di­min­ish­ing Re­turns be­cause we are rapid­ly run­ning out of vi­able new drug tar­gets that can be in­ter­cept­ed by small mol­e­cules or large pro­teins.  None of Phar­ma’s past or cur­rent strate­gies to im­prove R&D pro­duc­tiv­i­ty has worked be­cause they do not ad­dress the un­der­ly­ing is­sue, and the on­ly way to solve this prob­lem is to de­vel­op com­plete­ly new modal­i­ties that can ad­dress cur­rent­ly “un­drug­gable” tar­gets with­in the cell.

It is still not too late, but time is run­ning out very fast.

It’s fi­nal­ly over: Bio­gen, Ei­sai scrap big Alzheimer’s PhI­I­Is af­ter a pre­dictable BACE cat­a­stro­phe rais­es safe­ty fears

Months after analysts and investors called on Biogen and Eisai to scrap their BACE drug for Alzheimer’s and move on in the wake of a string of late-stage failures and rising safety fears, the partners have called it quits. And they said they were dropping the drug — elenbecestat — after the independent monitoring board raised concerns about…safety.

We don’t know exactly what researchers found in this latest catastrophe, but the companies noted in their release that investigators had determined that the drug was flunking the risk/benefit analysis.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 59,700+ biopharma pros reading Endpoints daily — and it's free.

It's not per­fect, but it's a good start: FDA pan­elists large­ly en­dorse Aim­mune's peanut al­ler­gy ther­a­py

Two days after a fairly benign review from FDA staff, an independent panel of experts largely endorsed the efficacy and safety of Aimmune’s peanut allergy therapy, laying the groundwork for approval with a risk evaluation and mitigation strategy (REMS).

Traditionally, peanut allergies are managed by avoidance, but the threat of accidental exposure cannot be nullified. Some allergists have devised a way to dose patients off-label with peanut protein derived from supermarket products to wean them off their allergies. But the idea behind Aimmune’s product was to standardize the peanut protein, and track the process of desensitization — so when accidental exposure in the real world invariably occurs, patients are less likely to experience a life-threatening allergic reaction.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 59,700+ biopharma pros reading Endpoints daily — and it's free.

Lisa M. DeAngelis, MSKCC

MSK picks brain can­cer ex­pert Lisa DeAn­ge­lis as its next CMO — fol­low­ing José Basel­ga’s con­tro­ver­sial ex­it

It’s official. Memorial Sloan Kettering has picked a brain cancer expert as its new physician-in-chief and CMO, replacing José Baselga, who left under a cloud after being singled out by The New York Times and ProPublica for failing to properly air his lucrative industry ties.

His replacement, who now will be in charge of MSK’s cutting-edge research work as well as the cancer care delivered by hundreds of practitioners, is Lisa M. DeAngelis. DeAngelis had been chair of the neurology department and co-founder of MSK’s brain tumor center and was moved in to the acting CMO role in the wake of Baselga’s departure.

Penn team adapts CAR-T tech, reengi­neer­ing mouse cells to treat car­diac fi­bro­sis

After establishing itself as one of the pioneer research centers in the world for CAR-T cancer therapies, creating new attack vehicles to eradicate cancer cells, a team at Penn Medicine has begun the tricky transition of using the basic technology for heart repair work.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 59,700+ biopharma pros reading Endpoints daily — and it's free.

Tal Zaks. Moderna

The mR­NA uni­corn Mod­er­na has more ear­ly-stage hu­man da­ta it wants to show off — reach­ing new peaks in prov­ing the po­ten­tial

The whole messenger RNA field has attracted billions of dollars in public and private investor cash gambled on the prospect of getting in on the ground floor. And this morning Boston-based Moderna, one of the leaders in the field, wants to show off a few more of the cards it has to play to prove to you that they’re really in the game.

The whole hand, of course, has yet to be dealt. And there’s no telling who gets to walk with a share of the pot. But any cards on display at this point — especially after being accused of keeping its deck under lock and key — will attract plenty of attention from some very wary, and wired, observers.

“In terms of the complexity and unmet need,” says Tal Zaks, the chief medical officer, “this is peak for what we’ve accomplished.”

Moderna has two Phase I studies it wants to talk about now.

Endpoints News

Keep reading Endpoints with a free subscription

Unlock this story instantly and join 59,700+ biopharma pros reading Endpoints daily — and it's free.

Rit­ter bombs fi­nal PhI­II for sole lac­tose in­tol­er­ance drug — shares plum­met

More than two years ago Ritter Pharmaceuticals managed to find enough silver lining in its Phase IIb/III study — after missing the top-line mark — to propel its lactose intolerance toward a confirmatory trial. But as it turned out, the enthusiasm only set the biotech and its investors up to be sorely disappointed.

This time around there’s little left to salvage. Not only did RP-G28 fail to beat placebo in reducing lactose intolerance symptoms, patients in the treatment group actually averaged a smaller improvement. On a composite score measuring symptoms like abdominal pain, cramping, bloating and gas, patients given the drug had a mean reduction of 3.159 while the placebo cohort saw a 3.420 drop on average (one-sided p-value = 0.0106).

Ear­ly snap­shot of Ad­verum's eye gene ther­a­py sparks con­cern about vi­sion loss

An early-stage update on Adverum Biotechnologies’ intravitreal gene therapy has triggered investor concern, after patients with wet age-related macular degeneration (AMD) saw their vision deteriorate, despite signs that the treatment is improving retinal anatomy.

Adverum, on Wednesday, unveiled 24-week data from the OPTIC trial of its experimental therapy, ADVM-022, in six patients who have been administered with one dose of the therapy. On average, patients in the trial had severe disease with an average of 6.2 anti-VEGF injections in the eight months prior to screening and an average annualized injection frequency of 9.3 injections.

Alex Ar­faei trades his an­a­lyst's post for a new role as biotech VC; Sanofi vet heads to Vi­for

Too often, Alex Arfaei arrived too late. 

An analyst at BMO Capital Markets, he’d meet with biotech or pharmaceutical heads for their IPO or secondary funding and his brain, trained on a biology degree and six years at Merck and Endo, would spring with questions: Why this biomarker? Why this design? Why not this endpoint? Not that he could do anything about it. These execs were coming for clinical money; their decisions had been made and finalized long ago.

Arde­lyx bags its first FDA OK for IBS, set­ting up a show­down with Al­ler­gan, Iron­wood

In the first of what it hopes will be a couple of major regulatory milestones for its new drug, Ardelyx has bagged an FDA approval to market Ibsrela (tenapanor) for irritable bowel syndrome.

The drug’s first application will be for IBS with constipation (IBS-C), inhibiting sodium-hydrogen exchanger NHE3 in the GI tract in such a way as to increase bowel movements and decrease abdominal pain. This comes on the heels of two successful Phase III trials.