Platform Technologies

It has often been said that one of the key fea­tures that dis­tin­guishes “big pharma” from biotech is ac­cess to the lat­est tech­no­log­i­cal plat­forms to aid ef­fi­cient drug dis­cov­ery and de­vel­op­ment.

These plat­forms range from vast chem­i­cal li­braries, ultra high through­put screen­ing and huge ge­netic data­bases in dis­cov­ery, to pre­dic­tive tox­i­col­ogy plat­forms, cut­ting-edge ‘omics’ and even deep-seated knowl­edge of par­tic­u­lar ther­a­peu­tic areas in de­vel­op­ment.

All these ‘plat­forms’ have two things in com­mon: they can be used on any (or many) de­vel­op­ment can­di­date as­sets, and they cost huge sums to es­tab­lish in the first place, and in a few cases each time they are used as well.

Hence their re­stric­tion to the largest phar­ma­ceu­ti­cal com­pa­nies (and a few of the so-called ‘big biotechs’ that are, in many ways, in­dis­tin­guish­able from the old-guard pharma).  Only when you have hun­dreds of ac­tive pro­jects can you jus­tify the cost of cre­at­ing and op­er­at­ing these plat­forms.

Or so the mantra goes.  It is ac­cess to these plat­forms that keeps the big com­pa­nies ahead in the race to dis­cov­ery and de­velop the best med­i­cines (or at least coun­ter­bal­ance the dis­ad­van­tages of being large and slow mov­ing, de­pend­ing on your point of view).

But is that just an as­ser­tion? How much ev­i­dence is there to sup­port the propo­si­tion that the ef­fi­ciency gains due to these plat­forms out­strips the cost of cre­at­ing and main­tain­ing them?

Not for the first time, Drug­Baron takes the op­po­site view: that with­out proper met­rics of the cost ver­sus ben­e­fit of such plat­forms, they rapidly be­come the “black hole” into which bil­lions of R&D spend­ing dis­ap­pear.  Like an al­co­holic open­ing a bot­tle of vodka be­fore break­fast, the jus­ti­fi­ca­tion sounds good to them­selves but holds lit­tle sway with ra­tio­nal analy­sis.  Spend­ing on plat­forms pre­vents big pharma prop­erly ad­dress­ing the cost prob­lem in drug dis­cov­ery and de­vel­op­ment and un­less it is ad­dressed such plat­forms may ul­ti­mately prove to be their neme­sis rather than their sav­iour.

The so­lu­tion may lie in an un­usual kind of de-merger.

Plat­forms (or ‘as­set-in­de­pen­dent tech­nolo­gies’ as Drug­Baron prefers to call them, to cap­ture all kinds of ca­pa­bil­i­ties that can be lever­aged across many dif­fer­ent drug can­di­date as­sets rather than just dis­cov­ery tools that the term ‘plat­form’ im­me­di­ately brings to mind) are ubiq­ui­tous in mod­ern pharma.  They are the prod­uct of an arms race, to se­cure ac­cess to best ca­pa­bil­i­ties in key areas.

In the 80s and 90s, that race was fo­cused on screen­ing ca­pa­bil­ity – larger, more di­verse li­braries to­gether with faster, higher through­put screens.  Even­tu­ally, all the big play­ers (and even CROs) had suf­fi­cient ca­pa­bil­ity to en­sure that find­ing the right mol­e­cule was no longer lim­it­ing.  So for the last decade or so, the focus shifted to tech­nolo­gies to re­duce at­tri­tion in the clinic: bet­ter tar­get val­i­da­tion and im­proved pre­dic­tive tox­i­col­ogy.  The ge­net­ics bub­ble was per­haps the biggest and best known ex­am­ple of this com­pe­ti­tion play­ing out, but there are many other ex­am­ples: one pharma com­pany re­cently told Drug­Baron they had more than 6,000 pre­clin­i­cal ef­fi­cacy mod­els of dis­ease in-house; an­other has de­vel­oped the most de­tailed metabolomics plat­form for pre­dic­tive tox­i­col­ogy; while every major pharma has a pro­pri­etary plat­form for an­ti­body gen­er­a­tion dri­ving their bi­o­log­ics pipeline.

Keep­ing these tech­nolo­gies “cut­ting edge” has be­come so ex­pen­sive that in­creas­ingly we hear pharma com­pa­nies talk­ing of “pre-com­pet­i­tive” ap­proaches to de­velop the next gen­er­a­tion.  A group of com­pa­nies might de­velop a plat­form ca­pa­bil­ity they then share.  The prin­ci­ple goal of such ini­tia­tives is to ac­cess even grander and more ex­pen­sive tools than in­di­vid­ual com­pa­nies could af­ford, rather than to dra­mat­i­cally cut costs (al­though shar­ing plat­forms rather than de­vel­op­ing the same thing in par­al­lel in each silo should at least keep a lid on ris­ing costs).

But through­out the last three decades, as spend­ing on these plat­form ca­pa­bil­i­ties has mush­roomed, there have been few voices ques­tion­ing their value.  After all, it seems plain ob­vi­ous that more and bet­ter tools will im­prove dis­cov­ery and de­vel­op­ment, lead­ing to more and bet­ter drugs and big­ger prof­its.

Drug­Baron does not ques­tion that as­ser­tion, but it is only half the equa­tion.

New in­for­ma­tion de­rived from these plat­forms has a value – but it also has a cost.  The ques­tion of whether the in­vest­ment in plat­form tech­nolo­gies is a boon or a bane for pharma lies in whether the ex­tent of the ben­e­fit out­weighs the cost of achiev­ing it.  Not just the mar­ginal cost of each ap­pli­ca­tion of the tech­nol­ogy, but an ap­pro­pri­ate slice of the fixed cost of cre­at­ing and main­tain­ing the plat­form.  After all, the cost of de­vel­op­ing the plat­form, even spread over a num­ber of pro­jects, is usu­ally by far the largest com­po­nent of the cost.

Drug­Baron can­not point to an analy­sis that ac­cu­rately es­ti­mates the in­cre­men­tal ben­e­fit of each new plat­form.  Such an analy­sis is im­pos­si­ble in a world that ap­proves only a few tens of new drugs each year.  The dataset is no where near rich enough for a re­fined as­sess­ment of the in­creased out­put of pharma as a re­sult of each new tech­no­log­i­cal plat­form – there are, after all, more plat­forms than an­nual ap­provals!

Nor is it suf­fi­cient to say that de­spite mas­sive in­creases in spend­ing on plat­forms gen­er­ally, pro­duc­tiv­ity (drugs ap­proved each year) has ac­tu­ally de­clined (at least through­out the pe­riod be­tween 2000 and 2010).  Pro­duc­tiv­ity might have been even worse with­out the new plat­forms.

In the end it has to re­main a mat­ter of judg­ment – how much ben­e­fit you be­lieve you get, and there­fore how much to spend on plat­form in­no­va­tion over­all, and in par­tic­u­lar which spe­cific areas and tech­nolo­gies you will ad­dress.

That the ben­e­fits can­not be quan­ti­fied ac­cu­rately is not the nub of the prob­lem though: that ho­n­our be­longs to the thorny issue of ac­count­ing for the costs.  Drug­Baron ar­gues that the dif­fer­ent way large and small com­pa­nies ac­count for these costs rep­re­sents the sin­gle biggest dif­fer­ence in their busi­ness mod­els.  And it re­ally mat­ters pre­cisely be­cause the in­cre­men­tal ben­e­fit de­rived from using such plat­form tools is in­her­ently im­pos­si­ble to demon­strate ac­cu­rately.

In a small com­pany – a very small com­pany, such as the as­set-cen­tric drug de­vel­op­ment com­pa­nies pi­o­neered by Index Ven­tures – there are no plat­forms (there may even be no peo­ple), just a sin­gle drug can­di­date.  The com­pany op­er­ates on its sin­gu­lar asset by ac­cess­ing sup­pli­ers of­fer­ing each re­quired process (whether they are out-sourced drug de­vel­op­ers or CROs).  This al­lows ac­cess to many dif­fer­ent plat­form tech­nolo­gies, though in many cases not the “plat­inum grade” tools pro­pri­etary to pharma.

But even the gold-stan­dard tools avail­able to the fee-for-ser­vice clients can be very ex­pen­sive.  Ask for a GWAS study for a par­tic­u­lar dis­ease, and the cost will be seven fig­ures if not eight.  It’s not sur­pris­ing – the fee-for-ser­vice provider had to in­vest heav­ily to cre­ate and main­tain that plat­form, just as the pharma com­pa­nies do.

Yet the arms-length na­ture of this re­la­tion­ship yields a crit­i­cal, but often over­looked, ben­e­fit: the cost of ac­cess­ing the plat­form is com­pletely trans­par­ent to the pro­ject leader.  Up­front, he knows the na­ture of the in­for­ma­tion he will re­ceive from ap­ply­ing the plat­form to his asset, and he knows the price he will have to pay.  While its im­pos­si­ble to know the value of the in­for­ma­tion once its de­liv­ered, at least he has a fight­ing chance of mak­ing a sound judg­ment know­ing the cost so clearly.

Vir­tual drug de­vel­op­ment is, then, noth­ing more than a se­ries of these cost:ben­e­fit judg­ments by the drug de­vel­oper.

Con­trast that with the sit­u­a­tion in­side any large pharma com­pany.  Most of the processes the pro­ject man­ager re­spon­si­ble for any given drug can­di­date asset needs to ac­cess are avail­able in-house.  Bet­ter still, the plat­form tech­nolo­gies avail­able to him are likely the best that money can buy.

But his judg­ment as to which one to apply, and in what order, to de-risk the asset he is work­ing on, and move it to­wards the clinic, or to­wards ap­proval, is ham­pered by one crit­i­cal omis­sion: the price of ac­cess­ing each.  Since the plat­form is in­ter­nal, the task of ac­cess­ing the tool is not for­mu­lated as a trans­ac­tion.  Crit­i­cally, now nei­ther the ben­e­fit of the in­for­ma­tion nor the cost of ob­tain­ing it are trans­par­ent to the de­ci­sion-maker.

Of course, its dif­fi­cult to es­ti­mate the “in­ter­nal trans­fer pric­ing” in any case.  The mar­ginal cost of using the plat­form for each asset should be ac­ces­si­ble (but, ac­tu­ally, is rarely known), but the cost of cre­at­ing the plat­form in the first place – usu­ally the largest com­po­nent of the cost – is dif­fi­cult to ap­por­tion among the sub­se­quent users over time.

The same pharma com­pany with 6,000 pre­clin­i­cal ef­fi­cacy mod­els also told Drug­Baron that they do not know the mar­ginal cost of run­ning any given one.  They know very well the ag­gre­gate cost of the 250 staff run­ning the mod­els, and the total con­sum­ables bill for the unit – but could not even es­ti­mate the mar­ginal cost of run­ning one in­di­vid­ual study.   Any at­tempt to guess the cost of de­vel­op­ing and val­i­dat­ing any one model, so that could be added to the mar­ginal cost of run­ning it, was be­yond the realms of pos­si­bil­ity.

The con­trast is stark then: a re­searcher in a vir­tual as­set-cen­tric biotech with a can­di­date drug to treat lupus can ac­cess sev­eral dif­fer­ent pre­clin­i­cal mod­els of the dis­ease from a sup­plier such as Rx­Cel­er­ate.  As he works out his ideal study de­sign, he knows quickly the cost im­pli­ca­tion – should he try his drug can­di­date in two mod­els in par­al­lel? Should he add an extra dose group? Or a range of dif­fer­ent end-points?  With each op­tion, comes a trans­par­ent cost di­rectly at­tached to the dif­fer­ent in­for­ma­tion he will re­ceive.  That cost:ben­e­fit judg­ment can then be com­pared with other sim­i­lar judg­ments he must make in par­al­lel to have the best chance of de­liv­er­ing a clin­i­cal can­di­date for the bud­get he has avail­able.

In the pharma com­pany, by con­trast, the costs are opaque, so the re­searcher – not un­rea­son­ably – sim­ply de­signs the best study de­sign he is able to.  Cost im­pli­ca­tions are only in­di­rectly im­pact­ing him: if se­nior man­age­ment cut the total ca­pac­ity of the unit de­liv­er­ing pre­clin­i­cal mod­els, they may ask him to re­fine (trim) his study de­sign to fit with the new ca­pac­ity con­straint.  But by and large, the avail­able ca­pac­ity is al­lo­cated with­out any con­sid­er­a­tion for where it will be most use­ful.

That is not to say that ei­ther one of them, the vir­tual de­vel­oper or the big pharma sci­en­tist, is bet­ter at mak­ing the right de­ci­sion about the size and de­sign of his pre­clin­i­cal study – but only the vir­tual de­vel­oper has the cost im­pli­ca­tions of each de­sign change in mind to as­sist him.

This di­chotomy is the sin­gle biggest rea­son why pharma can­not copy as­set-cen­tric drug de­vel­op­ment.  Such a large frac­tion of their R&D costs are con­sumed by the plat­forms, that even as they squeeze the di­rect costs of the pro­ject teams work­ing on an asset the im­pact on the over­all bud­get is neg­li­gi­ble.

And the con­se­quences of this opaque cost prob­lem go much wider.  How do you de­cide which plat­forms were worth their in­vest­ment, given the ac­knowl­edged prob­lem of quan­ti­fy­ing ben­e­fit?  If the plat­form is set up by an in­de­pen­dent ser­vice com­pany, the an­swer to that is easy: if they can turn a profit sell­ing ac­cess to the plat­form then the in­vest­ment in de­vel­op­ing and main­tain­ing it was worth­while.  But if the price they have to charge to re­coup the in­vest­ment ex­ceeds the price clients are will­ing to pay for the in­for­ma­tion the plat­form de­liv­ers, then ar­guably de­vel­op­ing that par­tic­u­lar plat­form was a drain on R&D pro­duc­tiv­ity.

With ac­cess to plat­forms not treated as a fi­nan­cial trans­ac­tion in­ter­nally within pharma, there is no such guid­ance as to which plat­forms are “worth” the cost of hav­ing them.  Fur­ther­more, the de­ci­sion to cre­ate a new plat­form is a com­pe­ti­tion be­tween ideas for a fixed bud­get for tech­no­log­i­cal in­no­va­tion.  In the small com­pany uni­verse, the de­ci­sion for a ser­vice provider to in­vest in cre­at­ing and main­tain­ing a new ser­vice is a very trans­par­ent re­turn-on-in­vest­ment cal­cu­la­tion.

Is there a so­lu­tion for pharma?

Out-sourc­ing solves the prob­lem for generic ca­pa­bil­i­ties – there is no rea­son why pharma can­not ac­cess all the same processes as the vir­tual as­set-cen­tric biotech, and ac­cess them from the sam­ple sup­pli­ers at sim­i­lar prices.  Doing so, may (or may not) cost more than de­liv­er­ing the same ser­vice in­ter­nally – al­though it will be dif­fi­cult to make that com­par­i­son of course, since the true cost of de­liv­er­ing the same ser­vice in­ter­nally is al­most never ac­tu­ally known.   But the ad­van­tage of the cost trans­parency, and the im­proved judg­ment of the cost:ben­e­fit ratio of his de­ci­sion-mak­ing, more than out­weighs the profit the ser­vice-provider needs to earn on his cap­i­tal.

But out-sourc­ing is not an op­tion for the plat­inum-grade pro­pri­etary tools that ar­guably de­fine pharma com­pa­nies.  The point of these tech­nolo­gies is that are sup­posed to be de­liv­er­ing com­pet­i­tive ad­van­tage.  So how can they si­mul­ta­ne­ously be pro­pri­etary and yet have a trans­par­ent cost as­so­ci­ated with them?  In prin­ci­ple, it should be pos­si­ble to cre­ate a rea­son­ably ac­cu­rate “in­ter­nal trans­fer price” by ac­count­ing dif­fer­ently for the cost cen­tres (on a much more gran­u­lar basis than at pre­sent).  But given the size and com­plex­ity of a very large pharma com­pany that may be harder to achieve in prac­tice than it sounds.

A bet­ter so­lu­tion may be to spin out the plat­form tech­nolo­gies into a sep­a­rate com­pany.  Such a move will force the cal­cu­la­tion of an ap­pro­pri­ate costed trans­fer price for each ser­vice.  Some of the new com­pa­nies tech­nolo­gies may be generic enough to sell to third party cus­tomers, like any other out-sourced drug de­vel­op­ment house, but oth­ers may be re­stricted in their sale to the orig­i­nat­ing pharma com­pany, pre­serv­ing those pro­pri­etary cut­ting-edge tech­nolo­gies per­ceived to be es­sen­tial for com­pet­i­tive ad­van­tage.  The strate­gic part­ner­ship be­tween the two would also allow a dis­cus­sion about where the plat­forms com­pany should be mak­ing new in­vest­ment to cre­ate im­proved tools – tools that the pharma would be pre­pared to pay a cer­tain price to ac­cess.

Such a move may sound bizarre – how could such a re-or­ga­ni­za­tion do any­thing other than in­crease ad­min­is­tra­tion costs? Does it not run counter to the en­tire the­ory of economies of scale that un­der­pinned the M&A trail which cre­ated the global pharma com­pa­nies in the first place?

The propo­si­tion for con­sid­er­a­tion by the read­ers of this ar­ti­cle is that such a move would im­prove R&D pro­duc­tiv­ity to such a large de­gree that it would leave any extra costs vis­i­ble in only the last dec­i­mal place.  The in­abil­ity to prop­erly as­sess the cost as­so­ci­ated with each new process un­der­taken on a pro­ject is a daily hand­i­cap across hun­dreds of multi-mil­lion dol­lar pro­jects in­side each com­pany.  If it is not ad­dressed some­how, the as­sump­tion that plat­form tech­nolo­gies boost out­put by more than their cost will be tested in a very painful ex­per­i­ment – one which kills the com­pa­nies.

Pharma com­pa­nies are just be­gin­ning to rec­og­nize that they are too large.  Some are even split­ting, try­ing to dis­en­tan­gle parts of the busi­ness that show no syn­ergy (such as Ab­bott’s split into a drug and a health­care busi­ness, or mooted splits for Pfizer’s an­i­mal health di­vi­sion or GSK’s con­sumer prod­ucts di­vi­sion).  These “de-merg­ers” do lit­tle to ad­dress the op­er­at­ing prob­lems that come from scale – the busi­nesses being sep­a­rated are op­er­a­tionally in­de­pen­dent any­way.  Split­ting Ab­bott in two pro­vided more flex­i­bil­ity for the in­vestors to de­cide which part of the busi­ness might offer the higher re­turn on cap­i­tal in the fu­ture (which is a good thing) but it did noth­ing to make ei­ther busi­ness unit op­er­ate bet­ter or more ef­fi­ciently, or to im­prove its fun­da­men­tal re­turn on cap­i­tal.

A braver de-merger would sep­a­rate plat­form tech­nolo­gies from drug can­di­date as­sets.  Who will be the first to try? And if you are not quite brave enough to con­sider this sur­gi­cal op­tion, then a healthy dose of pric­ing trans­parency for cre­ation and use of in­ter­nal plat­form tech­nolo­gies is a must.

*Drug­Baron is the blog from David Grainger, PhD, part­ner at Index, bio­pharma con­sul­tant, and in­terim CEO of XO1 Ltd.