Who Replaces Invitae?

The future of genomic diagnostic labs

A non-biotech friend asked me yesterday, “So, who replaces Invitae now that they’ve gone bankrupt?”

The question both caught me off guard and spawned a mini-monologue.

It caught me off guard because nobody is going to replace Invitae simply because they filed for bankruptcy. In their press release, they state they’re going to continue without disrupting operations. I don’t know what that means for two years from now, but I know Invitae isn’t disappearing overnight. Invitae provides genuinely valuable services to millions of patients and thousands of clinicians. Personally, I hope they continue pushing the mission of providing genetic information to billions of people indefinitely. 

Nevertheless, the question also prompted me to verbalize my thoughts on what the next generation genomic diagnostic lab(s) will need to look like in order to be profitable and therefore sustainable. We need three umbrella updates:

Blended [short | long]read WGS

The short-read vs. long-read conversation has been going on for years, but it seems clear that the tradeoffs between these approaches are complementary enough that we shouldn’t limit ourselves to one or the other. A sequence-once approach using a blended [short | long]read WGS assay has to be the backbone technology. Medically relevant regions in the genome are often also difficult to sequence regions and the two technologies are each better at capturing different subsets of these regions. The foreseeable technical and economic updates to either approach are unlikely to change that dynamic. 

    “You’re running two assays, it’s going to be too expensive!”

The price of a blended assay (or any longreads assay) is always a primary point of contention. And it’s a particularly salient topic in the context of Invitae, whose philosophy was always to slice the costs of products as much as possible and then some. Even outside of Invitae, it’s all too common to hear people talk about how a given genetic test is “too expensive.” 

I’m putting “too expensive” in quotes because the follow-up conversation is usually about how to reduce the cost of the test. But cost is only half of the equation, and it gets an undue amount of airtime. The taboo that too many people are avoiding saying is, 

    “This test isn’t valuable enough.”

Tests are considered too expensive because they often don’t provide sufficient perceived value. I would pay $2500 for a WGS test if I knew that it had a 90% likelihood of leading to a five-year life extension. Or, an insurance company would happily pay for $100M worth of sequencing costs if it were demonstrably true that it was going to save them $200M across the lifetime of the sequenced cohort. There are many hurdles here: a majority of individuals in a population are going to be given negative reports; VUSs are everywhere, even proving a positive ROI in a sequenced cohort is nontrivial, etc. 

To be clear, I say this as an optimist about the future of whole genome sequencing. I want to live in a world where society enthusiastically pays a thousand dollars to sequence a newborn because it’s undebatable that the expected lifetime value of sequencing is multiples of the initial investment. My point is, I wish we’d spend a little less time talking about how to make sequencing cheaper and a little more time talking about how to make it more valuable. 

Drug Development

Using genomic data to fuel drug development needs to be a front-and-center priority from day one. This is as important for the state of healthcare as it is for the diagnostic lab.

Drug development from genomic data isn’t a new idea. It’s something Invitae actively does, as evidenced by their recent partnership with BridgeBio. But it’s never been a top priority for them. It’s also something 23andMe does, as evidenced by this recent scathing Quartz article. But they’ve ruined their reputation, in no small part because of their approach to selling customer information.

The idea exists, but the execution thus far is mediocre. The best model I’m aware of for delivering this value includes:

1. Being transparent: Tell everyone, as clearly and as simply as possible, “If you’re okay with it, we’re going to sell your information to pharma and use your information to help develop new therapies.”

2. Doing the hard work: Build an in-house team as early as possible to do the heavy lifting of developing leads. Architecting the business’ data models in a way that’s conducive to drug discovery, creating specialized analysis tools and pipelines, and optimizing the investigation process cycles to minimize feedback loops are all critical to increasing the consistency of turning raw genomic insights into useful pharmaceutical insights.

3. Sharing the profits: One of the main reasons 23andMe is under fire is because customers feel like the company is “double dipping.” As the customer, you pay them to give them your data, then they take your information and sell it to a third party that could potentially put your privacy at risk. Patients own their genomic information. The only equitable way to profit from someone else’s data is to share that profit with them.

As much as people want to know why they’re ill or what condition their child has, their real priority is to be healthy. Future diagnostic labs need to play a more active role in the creation of novel therapeutics.

Complexity Reduction

I would love to watch a lab attempt to profitably return a million clinical grade reports per year with a team size of 150–200 people. I bet the right team could comfortably pull it off. 

A lab spinning up today would have two key advantages that Invitae didn’t have access to:

• A roadmap: I’m talking about classic “second mover advantage” here. Invitae, in many ways, was in uncharted territory and was truly innovative. But innovation is difficult and resource-consuming. A new lab would be able to leverage recent industry standards to quickly unlock certain economies of scale that took Invitae many years to achieve. 

• AI tooling: Every role filled by these 150 theoretical people would contain tasks that would be partially or fully automated in ways that weren’t possible a year ago. Software engineers building internal tooling, sales leads creating marketing campaigns, clinical geneticists interpreting reports—all of these people could be markedly more productive than 14 years ago. 

Large companies have a complexity overhead that inevitably starts to calcify their development flexibility and velocity. Invitae was a victim of their organization's size in the later years, and it’s not clear to me that it was avoidable given their ambition, the current state of technology, and the market they were helping to create. 

2024 might be the first time in history that a small team could provide clinical-grade genomics at scale.