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

I am very curious if someone who's familiar with medical devices and different kinds of tests can explain the details of all of this.



As OopsCriticality points out above, it's incredibly difficult to judge the potential of Theranos' technology because they have so far refused to perform and/or publish peer-reviewed randomized clinical data comparing themselves to the standard of care.

I can add some nuance to the WSJ article as a veteran of medical devices and FDA inspections:

The WSJ article quotes the inspection as a "surprise". There are only two reasons this inspection should have been a surprise to Theranos. Either in their conversations with FDA they appeared to be less-than-forthcoming, or if they were not familiar with how the FDA operated. Pre-approval inspections (i.e. you submit an application to FDA and before they make a judgment they do an inspection of your site) are very common for new technology and in some cases mandatory. When exactly they show up you don't know, but you plan for them to show up, and you expect it within a certain window of time.

Theranos CEO Elizabeth Holmes has been vocal about the FDA exercising oversight over all "lab developed tests", i.e. novel testing methods developed by a laboratory for use only by that laboratory and not for sale to others. Currently, these tests are not regulated by FDA, but FDA's sister organization the Centers for Medicare and Medicaid Services (CMS), under a set of regulations called CLIA. CLIA regs are not as stringent as FDA oversight and are more about good laboratory practices as opposed to the safety and efficacy of any new technology. So, Theranos and Holmes are trying to hold themselves to a higher standard, which is admirable.

The caveat with doing that is that you should be familiar with what that standard is going to be. I am suspicious that we're getting the full story from the Journal article, because I find it bizarre that no one at Theranos thought that FDA would consider their "nanotainers" a medical device: when FDA gets involved, anything that is part of a diagnostic system gets their oversight. Someone with 3 months experience in medical device regulatory could have told Theranos this.


Theranos is secretive and doesn't behave like other diagnostic testing companies, and these deviations generate controversy.

Theranos have been very, very squirrelly on details about their testing platform, claiming that they need to keep them as trade secrets in order to maintain competitiveness. Ordinarily, the tests would be independently evaluated in the peer-reviewed literature, and through the literature is how scientists and physicians would assess the value of what Theranos is offering; without peer-reviewed literature, there is little to go on. Maybe it works, maybe it doesn't, only peer-reviewed literature counts so we don't know.

Theranos has an unusual board of directors for a diagnostic testing company, with surprisingly little medical experience overall. Holmes herself is a college dropout—it's very hard to compete in a highly specialized and technical field without the proper knowledge base and experience, and traditionally that would be with an M.D.+fellowship or Ph.D.+postdoc. In interviews, when things get the least bit technical, Holmes sounds like she doesn't know what she's talking about. Theranos' patents, of which Holmes is a named inventor, are (IMHO) vaguely written. Could Holmes be a diagnostics savant? It's possible, but I don't see credible evidence to suggest this is the case.

The popular media has hyped up the company technology a lot, but in reality microfluidics is not new to the diagnostics arena, and it remains unclear as to what exactly Theranos offers as an advancement on what others are doing. There are also many technical issues to be considered with fingersticks vs. venipuncture samples, and Theranos has never explained how they have overcome those issues other than unsubstantiated assurances that their tests match standard lab tests.

There is also a fundamental problem with the entire hypothesis underlying Theranos: which I very crudely summarize as more tests = better health. That hypothesis is sophomoric and does not hold under the current thinking of evidence-based medicine. One can over test, and it has real, proven consequences towards the health of the patient in terms of invasiveness, costs, and potential risks due to unneeded procedures.

Does Theranos have something unique? It's not impossible. My null hypothesis is that they do not, and until I see something in the peer-reviewed literature suggesting otherwise, I am thinking it's smoke and mirrors.

Edit: All that text and I didn't directly answer your question! Consider my bloviation proof of my credentials as an academic. The earlier WSJ article raised questions about irregularities in regulatory-mandated proficiency testing, a serious allegation. My interpretation of this new, linked article is that there might be something to the allegation.


With regards to microfluidics, I've always understood the problem mostly to do with business - nobody from the incumbents wants a Moore's law in an industry without a lot of potential growth.

And that story fits the "new disruptive startup" story well.


I spent a few years with a major player in the diagnostics business. I worked directly on their automated ELISA machines.

When I was there in the 1990s, they already had spent a decade or so of R&D on microfluidic technology. They saw it as a hockey stick, but the technical hurdles were (and continued to be) very large.

The idea that one very talented scientist could find a novel solution that this company, with dozens of dedicated biochemists, could not find is plausible. But I believe the reality is somewhat different.


Assuming we're talking veni-puncture based measurements ,can you please expand about the current technical hurdles of microfluidics, and what prevents them from being solved , even after 20 years+ of effort by the research community ?

And all those interesting research paper ,we're seeing from the academic community, aren't they targeting the issue ? or are mostly resume fillers ?


I'm not a biochemist, and it's been too long ago when I worked there to comment on what the current state of the art is in this area.

What I recall is that at the microfluidic level, a lot of the work becomes a mechanical issue not a biochemical one. You're working with very small samples in a disposable system (obviously cleaning a capillary between samples is practically impossible). Blood isn't the easiest substance to work with, especially when trying to move it through small channels.

On top of that, now you're trying to detect very small amounts of antigen in an even tinier sample size. Your technology needs to be incredibly sensitive but have a decent resolution as well. Some antigens you want a simple positive/negative outcome. But with some other tests you want a level or range (like thyroid).


Thanks you for the elaborate answer, I think this more than answered my question!


imho this sounds like the comment in HN about when dropbox first launched


Nobody can, because none of the relevant data has been made available.




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

Search: