Someone with a PhD in microbiology explained to me (a layman) why what Theranos was claiming to do was impossible. She said you cannot test only a single drop of blood for certain things because what you are looking for literally may not be there. You need a full vial of blood to have a reliable chance of finding many things.
1. Is this simple but clear explanation basically correct?
2. If so, how could Theranos hoodwink investors for so long when possibly *millions* of well-educated people around the world knew that what they were claiming to do made no sense?
In: Biology
Your someone got the gist. Small sample size -> increased variance.
I remember when there was buzz about this. I’ve worked in biotech for years and began searching pubmed looking for novel microfluidics and high sensitivity breakthroughs across a variety of assyays. Needless say nothing jumped out and I thought it sounded pretty dodgy w/o knowing shit about Holmes.
A lot of the investors in Theranos weren’t experts in healthcare / biomedical investing. There wasn’t a single major bio/medtech fund invested in Theranos. Most of its investors were high brow political figured like Kissinger, Mattis, Devos, etc. for example, Moritz from Sequoia actively questioned the hype investing in Theranos and really tried to distance the rest of the Sand Hill VC funds from it.
I work in a reference lab that actually does some of those tests.
When Theranos was revealed. *Everyone* in my work unit from entry level lab tech to senior consultant called BS on the whole thing because it was glaringly flawed. We could do this because we knew the science of what they were trying to do inside and out.
Investors generally don’t have that scientific background. And they were disinclined to listen to lab scientists because the new tech was touted as being able to disrupt / destroy the lab industry, so we were perceived as automatically biased against it.
And frankly I have zero sympathy.
Edit to tag on my other comment:
Here’s an attempt at an oversimplified explanation related to what your friend said.
The most sensitive test for HIV can detect it at a sensitivity of about 14 viruses per ml, and it uses an input volume of 0.5 ml of blood plasma (note: this is insanely sensitive). Theranos claimed a comparable sensitivity, but boasted of a sample input volume of 0.02 ml. That is 1/50th of a ml.
If I split 1 ml equally into 50 tubes, but I only have 14 viruses in the original ml, then I can *at best* have 14 tubes with one virus in them. And then there will be zero viruses in the remaining 36 tubes. How will the test give an accurate result if 36/50 times, *there isn’t even a virus present to detect?*
Yes, for many things they were claiming to test, the amount of analyte in a drop of blood isessentially zero.
There are many reasons investor fell for it, probably the simplest/most obvious is FoMO/social proof: once a startup raises some money, it’s easier to raise more money, because new investors assume the old ones did their due diligence. Also, Elizabeth Holmes courted high-profile board members like Henry Kissinger, who made her seem credible (even though they weren’t bio/tech people).
A few other explanations:
1. Tech investors are often not tech experts, and many of their investments are *supposed* to be technically difficult and risky. As it turns out, many successful startups were build on ideas then considered technically difficult/impossible, investors are looking for that high-risk/high-reward startup that will become a unicorn. So “techincally risky” and “physically impossible” to an investor who is looking at dozens of deals at a time sound pretty similar. Also, many in the tech world have the attitude that most problems are solvable if you throw enough money at them. Theranos raised $700 million dollars–so even if they hadn’t figured it out, it’s easy to see why people might believe that they could.
2. Many investors in Theranos were not “traditional” Silicon Valley investors with track records of biotech/tech success; many were private investors who were not experienced in the space: the Walton family, Betsy DeVos and Rupert Murdoch account for half the money in. AFAIK, the most reputable SV investors were Tim Draper, who wrote an early check because he was a family friend, and Larry Ellison.
3. Theranos had that big deal with Walgreen’s and was working with the US military, so people assumed that their technology worked, because they were selling it and generating revenue with it. Of course, it turns out that they were serving Walgreens in a fraudulent way–but that was being covered up–and that Holmes was lying about the military contracts–but that’s an easy lie to believe when George Shultz is on your BoD.
I have a follow up question : what was her endgame? Grab the money and run? With her profile and all the attention, it seems highly unlikely. Regrerfully report after many hears of research that it doesn’t work? Again, with the media attention, I don’t think it was feasible. Or did she envision it as a scam with a much smaller scale, and it spinned out of control? Then why go after ultra high profile investors? Or did she at one point drank her own cool aid and honestly started to think it would somehow work out? She seemed highly intelligent, so this is also unlikely.
What was her aim in the long run? What was the “???” step before “profit! “?
The very nature of tech startups dictates that many investors either have to stay out entirely OR accept that they’ll routinely be assessing companies via rules-of-thumb about how businesses operate in general rather than through their technical knowledge of the products the companies create. That’s part of how you sometimes hear cliches about having invested in a culture or people more than having invested in a specific product. So Theranos’ big stupid lies were obviously a huge part of their hype train but part of the magic is that once you have enough inroads established you can start marketing those relationships as the bedrock of your company as much or more than your big stupid lie of a product because there’s already people who are predisposed to largely ignore technical assessments in general. It’s thus super dumb that she was able to hype her way to public relationships with Kissinger & Co. then ride *that* hype to Walgreens and Safeway but once you have a presence with Walgreens and Safeway I definitely see how you could ride that relationship to another few waves of investors.
Here’s an attempt at an oversimplified explanation related to what your friend said.
The most sensitive test for HIV can detect it at a sensitivity of about 14 viruses per ml with an input volume of 0.5 ml (note: this is insanely sensitive). Theranos claimed a comparable sensitivity, but boasted of a sample input volume of 0.02 ml. That is 1/50th of a ml.
If I split 1 ml equally into 50 tubes, but I only have 14 viruses in the original ml, then I can *at best* have 14 tubes with one virus in them. And then there will be zero viruses in the remaining 36 tubes. How will the test give an accurate result if 36/50 times, *there isn’t even a virus present to detect?*
As a scientist who researches and develops diagnostic tests, I’ll try to concisely answer your first question.
When developing any diagnostic test, it all comes down to sensitivity and specificity. To achieve that, there is a lot of optimisation needed to make sure you accurately detect and measure the one thing you’re looking for.
A box that is able detect and measure hundreds of metabolites from a drop of blood using a single assay is not impossible as this is what mass spectrometers are used for. However, these machines are very complex and expensive. That’s why they are more useful in measuring levels of things that are normally too low and require high sensitivity. On the other hand, they would be a waste of resource for measuring something like cholesterol or blood sugar.
More importantly, there is no one single sample preparation technique for measuring everything in the blood. Some metabolites may be stuck to proteins, some metabolites are inside red blood cells and need to be pulled out, some metabolites are very similar in shape/chemistry, etc. Then again, not everything in the blood is detected using the same detection technique. All that means is a one size fits all approach condensed in a small box that has both a sample prep robot and a detector is practically impossible, but also not something you want to have. It is a maintenance and quality control nightmare.
Because investors are desperately looking for what is called an investment *unicorn*. A company that most people underestimate, but that becomes exceptionally profitable and makes its investors a once in a lifetime return on investment.
Someone explained to me that during the Industrial Revolution, there were a lot of opportunities for investors to make fortunes. There were no cars for example, and suddenly there were car makers. You would have made a fortune investing in a product that wasn’t there before but that everybody needed.
The last that we saw something similar was with the GAFAs (Google Apple Facebook and Amazon). There wasn’t anything like them before. Employees who received company shares early on became millionaires once that the company traded publicly.
Anyway, back to Theranos. They presented themselves as a possible unicorn: they were alleging to be creating a revolutionary product that everyone would need. That’s all it was to get investors salivating and throwing money at them.
>Is this simple but clear explanation basically correct?
Not really.
An irony from all this is that scientific and commercial research into blood “microsampling” has intensified in part due to Theranos, with many promising results.
From [Nature](https://www.nature.com/articles/s41587-022-01242-0):
>**Advances in technology and changes to healthcare during the pandemic may finally realize the vision of patient-centric blood testing espoused by disgraced Theranos CEO Elizabeth Holmes.**
>[…] four years after Theranos’s lights went out, advances in blood analysis — including volumetric adsorptive microsampling (VAM), integrated point-of-care (POC) devices, wearables and telemedicine — are reaching an inflection point that could make Holmes’s fantasy attainable.
From [Stanford](https://systemx.stanford.edu/news/2023-01-20-000000/%E2%80%98theranos-works%E2%80%99-stanford-researchers-say-they%E2%80%99ve-measured-thousands):
>‘Theranos that works’: Stanford researchers say they’ve measured thousands of molecules from a single drop of blood
>Now, in a research paper published Thursday, Stanford researchers say they’ve accomplished what Theranos was unable to do: they’ve developed a new approach that can measure thousands of molecules with about one drop of blood. […]
>**It’s a bold claim, but one that Stanford medicine professor John Ioanndis, who was one of the first to publicly question Theranos in a column for the Journal of American Medicine Association, says has scientific backing**. Ioanndis is not affiliated with Snyder’s study.
Recent paper [from Australia](https://www.medicalrepublic.com.au/what-if-theranos-was-real/105514):
>Anyway, the point of this mini-rant is that just this month, 20 years after the founding of Theranos, [a paper has been published](https://pubs.acs.org/doi/10.1021/acs.analchem.3c05152) announcing the feasibility of using a capillary microsample to measure more than 100 lipoproteins and other parameters using metabolic phenotyping, with results on par with venous sampling.
Latest Answers