Public Data about Direct Costs of Genomic Testing

A colleague asked me about direct costs of genomic testing, e.g. costs of good sold.  There are a number of public sources that contribute to what we know about this.

One over-arching consideration is scale.  Economies of scale are very important in the laboratory, especially with next generation sequencing.   Running equipment to capacity is the most efficient approach to both capital equipment and reagents.

AMP MICROCOSTING
One approach, dating to Spring 2016, is the Association for Molecular Pathology's microcosting paper, Sabatini et al.   A Genomeweb article on the publication is here.   The ten page publication is open access at Journal of Molecular Diagnostics (18:319-28), here.   The authors surveyed several labs on several typical genomic procedures.   For many of the tests, costs per sample clustered in the $700-$1000 range, but several large assays had outlying cost estimates up to $2000-3400.  It's interesting that for some tests sample prep and library costs are about the same as "sequencing" costs.

One critical flaw in the paper is a lack of allocation for most overhead.  In some ways, it's like estimating the cost of an airline ticket by the cost of fuel, crew salary, and perhaps the food, without the capital equipment, airports, actual airline, etc.   Many members of AMP work embedded inside hospital-based labs where huge amounts of overhead are outside the actual lab bench chemistry budget - the building, staff, H.R., utilities, cleaning and security, upper management, everything. 

PUBLIC FINANCIALS OF PUBLICLY HELD LABS

Much data is available from the public financials of labs with quarterly investor calls and SEC reports.  There's one caveat, and that publicly held labs are incented to make their "cost of goods sold" as low as possible, within the limits of generally accepted accounting procedures.  They're proud to say their cost of goods sold is falling from 19% to 18% to 17%, but it also means that just as little as possible will be loaded onto the COGS number and all possible expenses will be accrued somewhere else.   To repeat the analogy I mentioned above, it's a little like looking at airline costs based on gas and crew salaries in the cabin.

That said, let's touch on the COGS of big labs like Labcorp and Quest.  In 2016, Labcorp had $9.6B revenue and $6.4B cost of revenue, so cost of revenue was 66% of revenue.  [*]   In 2016, Quest had $7.5B revenue and $4.6 cost of revenue, so cost of revenue was 61% of revenue.  So, all things average out, the average $10 test had average $6.10 direct cost.

If we switch to research oriented companies, Myriad Genetics had $771M revenue and $171M cost of revenue, so cost of revenue was 22% of revenue.  So, all things averaged out, the average $3000 test had $660 cost.  [**]   Invitae doesn't work for the simple calculation, since they are nowhere near profitable.  They had $25M revenue and $28M costs in 2016.   But they also had a $100M loss.  If they had broken even, they would have had by definition $125M revenue but they would have been selling considerably more than the $28M of tests they sold, so it's hard to directly extrapolate.

WHEN PUBLIC LABS SAY: "HERE, WRITE THIS DOWN, THIS IS MY DIRECT COST PER TEST"

There's another window, in that at least two publicly held genetics company feature their COGS in investor calls, so you get the number first-hand per test and you don't have to back it out from volume.  In their August 2017 investor call, gene panel test company FULGENT said it had cost per test of $485.  It refers to this as "the lowest cost in the industry."   The August 2017 investor call at gene panel test company INVITAE gives their cost per sample as $345,  which has "improved approximately 30%" from the prior year. 


_________________


[*] One source says COGS and Cost of Revenue are interchangeable: here.
[**]  In a different field, MAAA tests, Genomic Health had $328M revenue and $57M costs, or 17%.  In the pharma field, Gilead had $30B revenue and $4.2B costs, or 14%.  For recent articles on Gilead's revenue structure, WSJ here, LATimes here.