Hacker News new | comments | ask | show | jobs | submit login

How did a company like Theranos get so "far" with all signs early and present pointing to deep, fundamental problems? I refuse to believe investors simply failed to perform due diligence.



I heard from a friend who is fairly prominent in the microfluidics space (doing lab work with much smaller amounts of material than ordinarily required) that initially it was a drug delivery company. Delivering compounds in a reliable time released way is challenging and valuable, and I think their initial foray was in that direction. Apparently it didn't work well early on and they pivoted in this direction, so maybe the VCs were not expert in the space post pivot.

More likely I think they were also enamored with what looked like promising results. I've personally seen situations where technical due diligence on lab tech was poorly done and therefore wrong. Unless the VC has a fairly strong science background/scientific advisors in a related sector, it's super easy to make something look like it's working when it's actually totally busted.


Microfluidics is a sexy field. About ten years ago, it was nearing the peak of its first hype peak. This brought a lot of attention to the field and corresponded to the first time people started talking about precision medicine and "lifestyle" diagnostic tools.

Theranos was at the right time with a very intriguing idea, but one that was ultimately not a model for reality. They were able to get lots of capital at this time in preparation to develop a product that ultimately never materialized (and more fundamentally, may not be physically possible).


There are a couple of companies doing the same thing. They seem legit, and are founded by PhDs and university research scientists. If you look at their websites, you will find details about their technology, publications in scientific journals, professional conference talks & awards, etc. But they are not well known or hyped as much as Theranos. Do you really think the technology is not physically possible?

http://www.cbc.ca/news/canada/british-columbia/uvic-blood-te...

http://www.siscapa.com

http://www.genalyte.com


Oops, missed this comment. If anyone is reading still reading this...

To be honest, I don't know. I think it is possible to get a lot of information out of a single drop of blood. But there are issues with sampling--blood from the capillaries in my finger is very different from blood coming back from my intestine.

Maybe more fundamentally: a lot of "interesting" things are specific proteins. These can be at extremely low concentration--on the order of nanomolar all the way down to femtomolar. It's very difficult to amplify such a signal -- DNA is easy to amplify but proteins are not. At such low levels, the wrong drop of blood may have a dramatically different meaning, due to stochastic effects.


I think they ran in stealth mode for as long as possible. I remember finding out about them and not seeing a lot of information other than a few job postings some time back. Seems like once they started their move into daylight, the Wall Street Journal has been on a mission to take them down.


Maybe the WSJ has been tough on Theranos but there are other journalists who were big boosters, who now feel like they were misled -

http://fortune.com/2015/12/17/how-theranos-misled-me-elizabe...

Once one journalist finds one problem and does the legwork and prints it, others who know of problems find that journalist because they know he/she will do the legwork and get the story out.

The company then puts their side out to journalists they think might not put them in a bad light, who maybe get special access if they don't ask the toughest questions.

There's a natural triage of journalists in these battleground stocks/stories into pro- and anti- factions, even if the journalists are doing the fairest job they can with the info they get.

Not to say the company, the competition and other interested parties don't feed tips, leaks and tough questions to one side or another.


It's not obvious that the diagnostic assays they were running couldn't scale down.

These assays are performed regularly on large samples no problem, but the sensitivity and accuracy of the assay becomes erratic when you only have a couple micro liters of blood.

Either they couldn't develop proper assays (an engineering problem) or microliters of blood don't contain enough substrates for their assays (a major lapse in knowledge)


It's more like the method of drawing the blood was thought to be the problem. A venous draw is going to be pretty clean, in that whatever tissue got stuck in the aperture of the needle is going to be dwarfed by the quantity of blood taken; I've read that the actual assaying machines use very little of the blood, which makes sense because the reagents used are expensive, plus to the extent you're using optical scanning to read the results (what else might you use?), thinner will tend to be better (note most of the preceding is WAGs on my part from basic principles).

Whereas a fingerstick produces a few drops of blood that flow to the surface moving past the tissue and cells that were cut getting to the blood vessels, and they'll contribute unrelated proteins, lipids from cell membranes, etc. etc. On its face, many many people have said this is so problematic real evidence that it worked was needed; heck, even a theory would have been nice, but neither were forthcoming.


> A venous draw is going to be pretty clean, in that whatever tissue got stuck in the aperture of the needle is going to be dwarfed by the quantity of blood taken

From the many blood draws I have done, standard practice is to fill a tosser vial before collecting the real samples in order to remove these contaminants.


From the ones done on me for bog standard tests in the last dozen years or more, that wasn't done, but the extra gunk in the vial just might be intended to grab the contaminants as well as stabilize the blood. Or maybe these more standard tests allow for whatever amount of contamination is in the drawn blood, suppose it's 2-3 orders of magnitude less.


They don't throw out the first vial right there in front of you.


When they only take one vial, they'd better not be throwing it out and sending my doctor and myself "results"!

Now, it's possible I'm mis-remembering this, although the protocol includes showing me the vial after the draw so I can confirm it's labeled correctly and I tend to remember additional vials, since that manipulation often hurts more than anything else. I'll pay attention, heck, ask the phlebotomist in a few months during my next annual checkup.


...optical scanning to read the results (what else might you use?)

There's been some research into using piezoelectric elements coated with antigen or other reactive surfaces.

A positive result (meaning antibodies are attaching themselves to the substrate) would change the resonant frequency of the surface.

http://www.sciencedirect.com/science/article/pii/S0003267000...


There are a couple of companies doing the same thing. They seem legit, and are founded by PhDs and university research scientists. If you look at their websites, you will find details about their technology, publications in scientific journals, professional conference talks & awards, etc. But they are not well known or hyped as much as Theranos.

http://www.cbc.ca/news/canada/british-columbia/uvic-blood-te...

http://www.siscapa.com

http://www.genalyte.com


It was obvious to a great number of other diagnostics firms that this technology was hard to scale down.

I was at one of those firms 25 years ago and they were working on it back then. Either they abandoned the effort, or are still trying, but they haven't launched a single microfluidic immunoassay.


Like open source not making all bugs shallow, start-up capitalism doesn't guarantee an accurate assessment.


And with this specific industry it would be cheaper to not do the deep assessment (medical compliance work is $$$ and you can only lose 1x your investment, if it is real you can make 100x).


But if it is 100x more likely to not be real then it is worth investing the money to do the confirmation. This is true for medical and a lot more. Any company with a scientific foundation should have due diligence in the form of test accuracy, precision and reliability before a penny is invested. Anyone investing in a company where the technology is a scientific breakthrough should insist on confirmation tests. If those are forged then it becomes a criminal issue and you don't get to hide behind an officer level position.

I think with this specific industry testing and compliance is more important in due diligence.


She had an inside track to a lot of investors (and big name board members) which helped her immensely. Add on a nice demo and it's not hard to imagine.

According to the WSJ she was childhood friends with Tim Draper's daughter and that's where she got her first round.

http://www.wsj.com/articles/at-theranos-many-strategies-and-...

> The first $1 million came from Tim Draper, a founder of Draper Fisher Jurvetson, through two of his funds. Ms. Holmes was a childhood friend of his daughter and came to him “with extraordinary energy and brilliance,” he says.

At that point you have a very bright founder who interviews well and is backed by Draper. Everyone else assumes everyone else did their research.


A combination of greed, badly wanting to win the VC lottery, and not wanting the obligation of knowing whoever holds their stock is a whale.

If you've ever invested in something and found out you were about to lose it, part of you flirts with the idea of selling while you still have plausible deniability. You know that if you ask too many questions, you'll lose that plausible deniability. Of course it doesn't really make a difference, you know damn well, but it's part of the grieving process.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: