Enemies of pharmaceutical innovation

I was asked this week what the main culprits are that get in the way of innovation in pharma. It’s easier to pull out a few than to go in depth across all of the organisational and process errors, so I will start with two, and put three more into next week’s post.

Some of these have their defenders, but that is usually a defence of convenience. It’s often hard to change, even when something isn’t working.

Last year, I put an axe through my foot. It was an entirely self-inflicted injury, by literal definition. It was also stupidly preventable - I had been chopping wood for longer than I should have, and?very?stupidly I was doing it in shorts and with only Crocs rather than boots. I was lucky - of all the ways that could have turned out, it wasn’t too bad. The thing is, it’s the only time I’ll do it - it’s a harsh lesson.?

If only pharma would recognise its injuries, and the axe it swings above its own head.

Phase transition metrics

We’ve come to accept a default pre-clinical → phase I → phase II → phase III, etc., view of ‘how things are done around here.’ And, because of that, and the organisational structures that make people responsible for different parts of that chain, incentive structures and metrics have sprung up, McKinsey-style, to recognise ‘success’ in each part.?

That would be OK, if you believed that these are just ever-harder hurdles towards the best place that the drug could go. In that scenario, your ‘success’ at phase I could genuinely be said to be a useful step on your journey. You’d measure how many drugs you put into phase I that went on to start a phase II study as a Key Performance Indicator. Job done, and high fives all around.

But, if you measure it, and incentivise it, you get more of it. You get more of it by making it easier. And, making phase I and phase II easy is surely the exact opposite of what you want to do, rationally? This is the period when you really should be looking to find out not just that your drug isn’t terrible, but that it might be great. This is the time to try your best to kill it. Departments that hand over just-good-enough products to the next in line are doing the eventual recipient of the pass-the-parcel game no favours. And, unfortunately, because the organisational rewards are oriented towards progression-seeking, the behaviour you get is fast-as-possible unwrapping, but no-one rattling the parcel to see if anything is there. Removing the last layer reveals no gift. Sometimes that revelation is in phase II, sometimes, expensively in phase III, or, most expensively, on market.

I wrote more about phase transition?here .

Indication Selection

In the ‘seems like a good idea, how can it not make sense?’ category, many companies now do something they call ‘indication selection’ for assets in early phase.?

Imagine you’re a chef who wants to open a restaurant, and you go through a process to establish where, and what kind. And your process results in ‘North America’, or even ‘New York City’. And that’s?all?it does. You’ll have looked at numbers that show the value of the restaurant business in North America, success rates for new restaurants there, versus, say, Switzerland. Perhaps it compared New York City to Columbus, Ohio. And, so armed, you head off with hope in your heart to your flight to JFK, your investors’ cheque in your pocket.

That is analogous to ‘indication selection’ processes as they’re currently run. All the information you need to know - is the corner of 6th and 49th better than the middle of East 21st, between 5th and 6th? - is regarded as too complex. Would a high end Italian do better than a mid-market American farm to table at either of those places? Or any of the thousand other sites you could have chosen? What is the business model for those two kinds of restaurant (wastage, price points, staff costs) or the other hundred kinds of restaurant you could be? The break-even year, if ever, for that?specific?kind of restaurant? How good a chef are you, really? Your ragu might be amazing when served to your friends, but how would you do in the Michelin 3 star world you’d love to occupy?

There are variables at both ends: what kind of restaurant, and where. Making assumptions about the first, and then leaving the resolution on the second at a city level, rather than a Google StreetView level, would include lots of work, to little useful outcome.

In pharma, there are variables at both ends: what kind of product, and where.

Let’s say you’re trying to evaluate whether Non-Small Cell Lung Cancer is ‘better’ than Colorectal Cancer as a place to set up shop, following the ‘if I can make it there, I can make it anywhere’ adage. Well, what exactly are you offering, in terms of value proposition (the kind of effect, the effect size, the value-for-money proposition, in-combination or not, etc.)? Could you do both, and if so, in which order? Some of that analysis might be possible. But, that’s not enough. Would you launch in 3rd line, and hope to go towards neo-adjuvant in time, with or without a biomarker. Is your value proposition in 3rd line as good as the one in earlier lines? (Analagous, perhaps, to ‘is 9th and 23rd the right place to establish whether a high end nose-to-tail farm-to-table restaurant would succeed in the Upper East Side? If it failed, would you know any more about whether it might have succeeded there? Or if it succeeds, is it obvious that you can make the move across the city?)

Is an indication in relapsing remitting multiple sclerosis ‘better’ for your molecule than one in primary progressive MS? How about if it would work in both, and your process traded them off against each other? Or if you looked at MS, and overestimated how likely you were to beat Ocrevus in a head to head? Or if, in your ‘exhaustive’ review of potential indications, you missed one?

The ‘indications’ for drugs might be broad, but in the real world they have positions?within?the indications, and real, embedded and potential competition - huge complexity to map, even if it was for today’s world, never mind the world in 10 years’ time. The path to those markets is not necessarily linear, and the probability distribution along the path almost impossible to predict with accuracy.

‘Indication selection’ as a consequence has the impression of logic, of trade-offs, of analysis. It is necessary, but insufficient.