FTC Slams Illumina-PacBio Merger, but Illumina Not Quitting Yet

Tonight I was intending to finally get out my summary of technical notes from the Oxford Nanopore Community Meeting, but yesterday the U.S. Federal Trade Commission issued a press release that they believe the proposed acquisition of Pacific Biosciences by Illumina would be grossly anticompetitive and cannot be approved in any form.  A more detailed report is promised but hasn't surfaced yet.  Curiously, not only did this ultimately move PacBio's stock very little, but today Illumina and Pacific Biosciences filed matching SEC documents that Illumina will continue to infuse cash into PacBio through March of next year.  Also, I should note that someone left a passionate defense of regulators in the comments on a prior piece, noting that the FTC decision shows that the CMA was justified in opposing the deal and not simply acting with a parochial eye on Oxford Nanopore.
That cash infusion is really important for PacBio, as in February there is due a $20 million loan payment and Illumina will cover that.  That's a big relief, since the last quarterly statement showed only $53M in cash and equivalents.  There's a $98 million breakup fee due PacBio should (when?) the merger terminate.  Illumina's gifts do come with a big string attached -- should PacBio be bought or if it raises $100 million or more in a single financing, Illumina gets repaid.  That's perhaps a poison pill a big acquirer can tolerate, but may really put a crimp on refinancing.  One would hate to see PacBio reduced to taping a SMRTcell to the wall at an art gallery just to stay afloat.

The fact that Illumina didn't just walk away is curious.  Most observers feel that the combined opposition of the FTC and the UK's CMA means there is no way this deal can proceed.  The FTC action isn't binding at this point, but they can seek an injunction and I believe they would probably get it.  The FTC hasn't scheduled an actual legal proceeding until August of 2020, so if Illumina sticks at it PacBio will be twisting in the wind for at least another six months.  

It should also be noted that the stock price barely fell.  In aftermarket PacBio was down about 10%, but that mostly was recovered today.  I hate trying to read stock price tea leaves, but given an $8 offer from Illumina it makes some sense.  One way to read the ultimate recovery is that the current price of PacBio is roughly what the market thinks the company is worth independent of an Illumina merger.  In other words, should Illumina walk away the price wouldn't fall much.  But, there may be optimists who still see some path for Illumina to get past two different regulatory agencies in its path. The price isn't actually much higher than the pre-acquisition price, before the Sequel II was launched so successfully.

Illumina Holding On For Plan B?

One theory that has bandied about online for awhile is that Illumina has a plan B, which is to execute some sort of co-marketing deal with PacBio.  Such a deal would probably involve an equity investment to pump new cash into PacBio and ensure that further development of the platform and the building of clinical applications occurs.  Not walking away makes sense then as a buying time maneuver, keeping other parties at bay. 

But the question there is whether such a deal would pass muster with regulators, particularly now that their level of attention has been heightened.  While such a deal wouldn't unite the two companies IP portfolios, or Illumina's legal group with PacBio's patent estate, too broad a deal could be seen as accomplishing the same restriction of trade as a full merger.  Not that I'm qualified to go beyond speculating on this; for once in my life I'm wishing I had some antitrust law in my formal resume.  Perhaps a narrowly focused deal, allowing Illumina to co-develop and co-promoter for clinical applications, might slide by.

If Not Illumina, Who Might Buy PacBio?

If the deal doesn't go through, then what?  The list of companies with experience in the general molecular biology sector and sufficient heft is not long.  I am a believer that BGI/MGI would not be allowed to acquire PacBio due to the current U.S.-Chinese trade environment; many small biotechs have floundered in part from the CFIUS (Committee on Foreign Investment in the United States) scotching deals.  On the other hand, trade policy with regard to China under the current administration sometimes resembles the output of a random number generator, so nothing is guaranteed.

I actually found a recent list of the top twenty molecular diagnostics companies; that would be a reasonable first draft of a list of candidate acquirers.  They are Roche, Philips Healthcare, Abbott, Danaher (which recently absorbed GE Healthcare, which was also on the list),  Siemens Healthineers, Canon, ThermoFisher Scientific, FujiFilm, Biomerieux, Sysmex, Ortho Clinical Diagnostics, BD (Becton Dickinson), Bio-Rad, Hologic, Konica Minolta, Grifols, DiaSorin, Varex and Exact Sciences.  Throw in genomics plays Agilent and QIAGEN and we have a full list.  I'd be shocked if there aren't good arguments for some other companies; I'm not an M&A type and don't think about this routinely. And who knows?  Perhaps someone not currently in molecular diagnostics might jump in.

But that list can be winnowed down quickly.  Several of those companies appear to be focused on imaging-type diagnostics and not much in the molecular genetics space: Philips, Siemens, Canon, FujiFilm, Konica Minolta.  Biomerieux is very focused on infectious diseases, not likely a strong play for PacBio.  Exact's main product is a colorectal cancer screening assay, which would be the wrong sort of sales force (general practioners)-- assays using a Sequel family instrument are likely going to be in rare diseases or oncology.  DiaSorin is quite small, with a market cap of $6.5 billion and only $0.76 billion in 2018 diagnostics revenues.  Varex is even smaller, at about $1B market cap and imaging focused.  Sysmex is a Japanese company that appears to be focused on blood analysis, so probably not a good candidate. Grifols is a Spanish company that appears focused on blood testing, so again probably not a good sales force fit.

Roche has a bad history with PacBio, having yanked the rug out from under their prior partnership, but PacBio's cash position will be encouragement to let bygones-be-bygones.  The research genomics crowd sees Roche as somewhere genomics technologies go to die, having failed to keep Nimblegen and 454 competitive and never getting Genia out of the gates.  Still, a huge player.

ThermoFisher is another company with deep roots and a checkered history in genomics, with a boring but profitable fluorescent Sanger business, an Ion Torrent business run the same way and SOLiD in their list of extinct technologies.  It's also rumored that Thermo is pursuing buying QIAGEN, which would take both of those off the table. At around $8 billion market capitalization, QIAGEN is probably just big enough to think about buying PacBio, but they're also coming off the abandoment of GeneReader and going through a CEO transition; much more likely to be merger fodder than predator.  

Hologic is a fun wild card option; I really wasn't familiar with them but they do have qPCR-based assays for gene fusions, and gene fusions are an area where long reads clearly beat short reads for sensitivity.  Ortho, which is a branch of Johnson & Johnson, is another out-there possibility; as far as I am aware they've traditionally been in antibody-type blood tests.  Abbott is another huge player in traditional blood tests but not a player in genomics.  Certainly any of the monsters could try their hand at genomics with PacBio, given they have abundant resources.

That leaves Bio-Rad, BectonDickinson, Danaher and Agilent from my list.  

BioRad at 10.76 billion is big enough.  They've not been super innovative in the genomics space, mostly just consolidating a lot of droplet microfluidics IP to support their Droplet Digital PCR business and hammer at 10X Genomics.  

BectonDickinson hasn't been a big player in genomics, but has a broad portfolio of molecular diagnostics.  Danaher (with the inclusion of GE Healthcare) is similar.  Agilent has mostly stayed a reagents vendor, but their work with LaserGen offers the interesting possibility of combining short reads and long reads in the same portfolio.  All of these have the heft to take on PacBio: Agilent is the wee company at only $25 billion market cap; BD is more than twice that and Danaher a bit shy of twice BD.

What PacBio Should Do

Okay, here's my free advice, worth every penny I'm charging for it.  PacBio as a standalone company,  or whomever buys them, needs to drive in a few  directions.

One is to push the yield of the devices even higher, as higher yield equals lower cost.  If a clinical-quality human genome could be reliably obtained from a single flowcell, that matters.  Ideally this will mean increasing the number of ZMWs on a flowcell, but better sample prep and more data collection per ZMW never hurt.

PacBio still has a lead over Oxford Nanopore on quality due to a much smaller and more manageable systematic error issue.  Nanopore is working hard on closing that gap (more on this at some point, maybe not until the New Year), but the technical difficulties are significant -- particularly for repeat arrays much longer than the length of a nanopore channel.

But perhaps more important is to identify clinical opportunities that could be generated rapidly and get them into proper trials.  Given the high capital cost of an instrument, that's probably going to mean things like oncology or rare diseases (as noted before) where patients can be concentrated at relatively few academic centers.  For oncology, it's probably fusion proteins; for rare diseases it's going to be repeat expansion disorders.  Clinical markets are conservative and hard to get into because of that, but that also means it's a barrier to entry for competitors.

There may also need to be a rethink of key aspects of the instruments.  For example, you can load up a Sequel with multiple flowcells -- but only as a batch.  There's no hot-swapping.  That's not an arrangement that works well in a clinical setting.  I haven't a clue how nightmarish it would be to redesign the flowcell feeding mechanism to allow hot swapping, but the rewards in the clinical space are potentially huge.

Illumina's Plan C?

If Illumina can't get access to PacBio, then there is probably no path past the regulators to buy any long read sequencing platform.  Of course, right now the only real long read platform other than PacBio is Oxford Nanopore, and Illumina and ONT have antimatter-matter level bad blood between them.  10X Genomics' linked read technology, or other linked read technologies launching from various startups such as Universal Sequencing Technology, could be an option -- but again, would regulators permit it.  It may be that the only way for Illumina to have a long read product is to develop one internally (long rumored, also long rumored to be moribund) or buy into one or more startups that don't yet have a product.

Alternatively, Illumina could decide to pivot to providing assays easily converted to any long read platform.  If sequencing itself becomes a commodity, it makes sense to try to move upstream if value will be given to the assay developers.  Of course, their fraught relationship with ONT won't make generic long read assay development easy.

A third possibility is just to pivot away entirely, focusing on delivering genomic information.  That seemed to be the direction of the company a few years ago when launching Helix and Grail, but they certainly haven't gone full thrust that way.

Or, one more option: drop a bomb on the market in the form of some crazy advance in short read technology.  Obviously getting a breakthrough out of the lab isn't easy and it runs the risk of cannibalizing their existing business, but if NovaSeq got another 4X bump in performance it could mute the claims that SBS technology is out of steam.  Is that even possible?  I don't know, but then again I thought it was stunning they could get 2x100 out on the original HiSeq.