Biotech Dan's  Instablog

Biotech Dan
Send Message
Post-graduate student in chemical + bioengineering, get my kicks by learning about things that interest me! Looking to use my expertise and knowledge to better communicate biotech trade ideas to the average investor.
  • The Value In APPY1 18 comments
    Feb 6, 2014 9:17 PM | about stocks: APPY

    Hi,

    I feel there is a lot of misconception over APPY1. I have seen many comments questioning the usefulness of such a device. When posting my article on APPY (check my profile if you are inclined), I did not anticipate such a backlash from the community.

    I am going to discuss the comments that I have seen regarding APPY1 here:

    Robert posted on another article:

    A normal/low WBC count, which is already performed on the patient, has a NPV of 96% in children. Who needs another inaccurate test.
    The APPY1 test is worthless in my opinion.

    I posted a lengthy response within that SA analysis on APPY after researching and accessing the article which I assume he crept up on.

    If you would like the full response, please find the article here: seekingalpha.com/article/2000261-venaxis...-march

    For here, I will start by saying Robert conveniently left out a few important facts. One is that this NPV of 96% in children actually refers to toddlers aged 1-3.9 years old. For children aged 4-11.9, the NPV is 89.5%. The next thing that was left out was just what this NPV refers to. It isn't a 96% true negative result for ALL toddlers. It is a 96% NPV for toddlers who have a low or medium white blood cell count. Obviously this is not every single toddler, and so if the test were applied to ALL toddlers - the NPV would drop. Second, Robert only attached a value for NPV and not any other statistical result. There is a reason APPY1 reports more than just one statistical figure. I outline within my lengthier comment why JUST NPV is not an accurate representation of a test accuracy at all. It turns out the PPV (accuracy in predicting positive events) was 42%. Lastly, the study was not conducted in an appropriate manner - so it is on uneven grounds and cannot even consider to be compared to APPY1. The diagnostic tool from APPY is tested in a blinded study. The study Robert posted was performed unblinded, and the article authors reported within their literature paper that multiple sources of bias did exist. This included purposefully selecting patients who looked sicker and more likely to "better perform" within their study.

    Within my article, rootbeer posted:

    I agree with your article on PGNX. I disagree with this article. The standard of care for diagnosing appendicitis is a physical exam, followed by a white blood cell count, followed by ultrasound and/or CT. In children, we use ultrasound in an attempt to visualize the appendix. If it is seen and it is increased in size, the work up is done. If it is not seen, as is frequently the case, we proceed to CT. If the appendix is not seen by CT, then the surgeon is forced to use best judgement in either proceeding to surgery or close monitoring if WBC and physical exam are unremarkable. If appendix is seen and is normal, the work up is done. If the appendix is seen and is enlarged, inflamed, etc, the patient goes to surgery. Pregnant women are diagnosed with ultrasound and/or MRI. The reason that imaging exams are used in the work up of suspected appendicitis is to avoid a missed case of appendicitis. Even one missed case of appendicitis in a child can be devastating to the child and eventually all the physicians involved in the case. The test offered by Venaxis will have little to no value here in the US. The threat of malpractice for a missed case of appendicitis forces the ER physicians and the surgeons to request any and every test to rule in and rule out appendicitis. The test offered by Venaxis does not do a good enough job of ruling it out.

    I am a fan of his posting, and his comments strike me with intelligence.

    However, inspection of his profile reveals he is a radiologist. As such, he will always have an inherent bias to CT scanning. You are always going to unknowingly uphold your own work in the highest regard - it is only natural.

    I also responded to his post within my article. The lengthy procedure for diagnosing appendicitis, in my opinion, already outlines the need for a speedier and more efficient test. We have already looked at the advantage of APPY1 to white blood cell count testing (see above). If ANYTHING, APPY1 would provide a SOLID replacement for white blood cell count within this diagnosis procedure. It is better suited for diagnosing appendicitis (and negative cases), and it even uses white blood cell count as one of the measurements within the test. This alone should prove that APPY1 has a place within the diagnostic testing for appendicitis - a bearish investor has unknowingly admitted a proper placement for it.

    As I've also mentioned, I feel his wording further stresses the need for further testing. "In children, we use ultrasound in an attempt to visualize the appendix." ATTEMPT. The ultrasound itself is open to extreme interpretation. The process of running the ultrasound test + making sense of the data is a job in itself. This easily takes up valuable employee time. APPY1 is a quick test that is not biased by human interpretation of what "could be" or "perhaps is".

    rootbeer goes on to claim that imaging is used so that appendicitis is not missed. But CT imaging (perhaps the gold standard) has a diagnostic accuracy of 95-98% (emedicine.medscape.com/article/363818-ov...). Compared to APPY1 (mean NPV of 97%, sensitivity 97%), there is no statistical significance between CT imaging and APPY1 when a negative result is present.

    Closing Statements

    To all those remaining bearish on the APPY1 diagnostic tool, perhaps you could better enlighten me why? If your child was waiting in the ER with abdominal pain, which would you prefer - sending them immediately to a CT scan resulting in high doses of radioactivity, and a diagnostic accuracy of 95-98%, or to attempt to avoid that situation with APPY1?

    Disclosure: I am long APPY.

    Stocks: APPY
Back To Biotech Dan's Instablog HomePage »

Instablogs are blogs which are instantly set up and networked within the Seeking Alpha community. Instablog posts are not selected, edited or screened by Seeking Alpha editors, in contrast to contributors' articles.

Comments (18)
Track new comments
  • B_Banzai
    , contributor
    Comments (539) | Send Message
     
    Thanks Dan. The more I learn, the more I like about APPY. The insight of seeing APPY1 as an improvement over the existing blood test is an excellent one. Does APPY1 report a white blood cell count as a separate output or does it only provide an integrated result? In other words is it a "Two for one" test? Or do you have to run a separate test if you want to check white blood cell count even though APPY1 is negative for appendicitis?
    9 Feb 2014, 02:55 PM Reply Like
  • Biotech Dan
    , contributor
    Comments (111) | Send Message
     
    Author’s reply » I have no idea whether it is a separate output or not. That would be a good question for management. The key is that it uses white blood cell count within its algorithm to check for appendicitis - so that alone proves it to be more valuable (and accurate) than a standard white blood cell test.

     

    Also, I see no reason why they couldn't list the results for each variable they test.
    9 Feb 2014, 10:55 PM Reply Like
  • Joulian
    , contributor
    Comments (5) | Send Message
     
    Thanks Dan. I think your research and analysis are insightful and helpful.

     

    I am long APPY myself.

     

    My concern is that a drop in sensitivity results might be bad news for the Company.

     

    What do you think are the odds of the 2010 results repeating here?
    11 Feb 2014, 03:45 AM Reply Like
  • Biotech Dan
    , contributor
    Comments (111) | Send Message
     
    Author’s reply » The upcoming results for sensitivity is impossible to predict. The studies are blinded as is, so really no one has any idea.

     

    But it isn't "new" technology, in a way. APPY1 builds on prior knowledge. It uses a combination of tests that are already known to be indicator of appendicitis. Through a combination of these tests, the software can output a predicted result. This result can obviously be fine tuned as well.
    12 Feb 2014, 10:12 PM Reply Like
  • Joulian
    , contributor
    Comments (5) | Send Message
     
    Thanks Dan. I was wondering whether you have any insight to the requirements for the European CE and Futility Analyses? Do you think these are reasonably indicative of the sensitivity results for the APPY1 test?

     

    I am concerned that any result less than 97% sensitivity might send the stock south bound or even crashing...

     

    I disagree with what commentators have said about the malpractice angle. They have alluded to the APPY1 test not being able to do a good enough job of ruling appendicitis out, given its less than 100% sensitivity and negative predictive values. Legal malpractice arises from substandard medical care or the failure to provide reasonable care to the patient. We all know that there are currently no tests (CT scan and Ultrasound included) offering 100% accuracy in ruling out appendicitis. An ER physician relies reasonably on the high sensitivity and npv of any test to make a sound judgment (together with other information at hand). When I consider what has been reported about the likely adoption rate (75%) by ER doctors of the APPY1 test, I think this clearly indicates the APPY1 product has passed the reasonable standard care test. The converse applies as well: an ER Physician might well be negligent for failing to use APPY1 (assuming it is approved and widely adopted) and thus causing harm and damages in the particular case. By the way, I have a legal background.

     

    17 Feb 2014, 05:48 AM Reply Like
  • Jeremy LaKosh
    , contributor
    Comments (209) | Send Message
     
    Hi Dan:

     

    I'm still thinking about APPY1 and how the CT scan relates to its relevance. Do you know of any comparable products that have come to market that have the similar purpose of reducing dependence/usage on CT scans? Sorry to be a pest about this, but I would like to see if there are markets out there already based on this alternative concept. Please let me know if you need me to clarify.
    20 Feb 2014, 01:10 AM Reply Like
  • Biotech Dan
    , contributor
    Comments (111) | Send Message
     
    Author’s reply » Hi,

     

    APPY1 has shown positive feedback in recent arms of the Futility Analysis with respect to FDA clearance within the USA.

     

    I understand your concern over the sensitivity issue, and so feel as if I should outline two major things.

     

    1) Equities, particularly biotech stocks, can be risky. If you do not want to take on a large amount of risk, the greatest play would be to cash in on the run-up to data release. I would not advise you hold through data release in March if you are worried about losing an irreplaceable amount of funds. Not to scare you, but you should be aware that there are ways to make the trade while avoiding the data release. Of course, this leaves you out of a potential SIGNIFICANT upside...
    2) About your worries of sensitivity: this is a diagnostic test, and not a "traditional" pharmaceutical product. Bear with me, this might be a little bit difficult to explain.
    The test operates primarily through measuring specific variables, and then using a method to combine these results into a readout of appendicitis+ or appendicitis-. If this makes sense to you, then you should understand that it does not just operate ONE way. What I mean by this, is this method can be adjusted. It is proprietary, yes, but ultimately must use a certain threshold for these measured variables to determine a diagnosis. If you are still with me, basically it is going to check variables A, B and C. For each variable, it will compare the measured amount to a "cutoff value". For example, is A < "certain amount"? If yes, then this is a sign that appendicitis is negative. Repeat for B and C. These results combine to determine a diagnosis. Of course, this is extremely over simplified. The actual algorithm used is likely much more complicated and a result of much research, applying specific weightings to A, B, and C, possibly accounting for synergistic effects of A+B, B+C, etc.
    The takeaway message from this is that adjustments can be made. If we make our conditions MORE strict (i.e. A <<<< "cutoff value"), then the "predicted" result for a negative appendicitis will be more accurate. If you understand this, you will see that basically we are imposing stricter conditions in order to come up with a negative appendicitis test result. Of course, the drawback to this, is people might have "very negative results" but no longer quite "negative enough", thus increasing the amount of false positives.
    In this way, we can see how sensitivity is in essence a non-issue. 97% is not something that just happened by chance, it is likely a target Venaxis set out for as an acceptable measure. We could increase the sensitivity by imposing stricter conditions, but this will also have an overall effect of increasing false positives (and not necessarily in a linear relationship (i.e. 1 for 1)). The inverse is also true, we could relax conditions and have fewer false positives, but perhaps result in more false negatives.
    From this long-winded message, the main idea here is that even a drop in sensitivity at this point would not be "bad" per se in the eyes of a knowledgeable investor. We know it works (i.e. the use of these specific biomarkers CAN predict negative appendicitis) and thus the actual sensitivity with which we hope to achieve CAN be resulted. It is not set in stone by any means.
    I can try to better explain this if you wish. Overall, I see no reason why the sensitivity should drop.
    20 Feb 2014, 01:33 AM Reply Like
  • Biotech Dan
    , contributor
    Comments (111) | Send Message
     
    Author’s reply » Sorry, I do not know of any "similar products" that are aiming to replace or prevent CT usage.

     

    I do know that minimizing CT usage is something we are more aware of today. Much like going "green" in industry, the reduction in radioactive substances (especially in medicine) is gaining momentum in recent years.

     

    Furthermore, the selling point, or sales pitch, is to minimize CT usage through the use of APPY1. However, there is no foreseeable reason why APPY1 can not also be used in conjunction with a CT scan. We all know CT scans are not perfect. A quick literature search reveals that a significant percentage of people who are flagged as "appendicitis positive" through a CT scan, are later revealed to be perfectly healthy (once the surgeon has cut the patient up and noticed no signs of inflammation or burst appendix).
    20 Feb 2014, 01:38 AM Reply Like
  • Biotech Dan
    , contributor
    Comments (111) | Send Message
     
    Author’s reply » Sorry, I mentioned positive feedback for APPY1 but forgot to expand.

     

    Here: http://bit.ly/MdWLlu

     

    The data is still blinded to Venaxis (and thus, us), so all we can tell is that an external body has deemed the data to be positive, or at least positive enough to warrant a continuation of the study. Whether this is a sensitivity of 97%, 99%, or 79% - I don't know.
    20 Feb 2014, 01:42 AM Reply Like
  • B_Banzai
    , contributor
    Comments (539) | Send Message
     
    Dan - That was VERY helpful for me. Hopefully, it was for others as well.
    20 Feb 2014, 09:36 AM Reply Like
  • Biotech Dan
    , contributor
    Comments (111) | Send Message
     
    Author’s reply » Out of curiosity: what was helpful about it?
    20 Feb 2014, 09:42 AM Reply Like
  • Joulian
    , contributor
    Comments (5) | Send Message
     
    Thank you so much Dan. Your explanation was very helpful. I have to admit that I don't know enough about algorithms, but I think you were trying to tell me that the APPY1 test has a way to ascertain the highly probable incidence of appendicitis.

     

    There aren't enough details in the testing by the external Data and Safety Monitoring Board to help me know how well APPY1 is going to perform. I guess I can't put too much weight on APPY1 clearing the futility analyses.

     

    I understand the inverse correlation between sensitivity and false positives, but if I read the CEO correctly, he's aiming for high sensitivity and high specificity. I have to admit that the science in securing these objectives is beyond me!

     

    I don't know if you will agree with this thinking. The real value of the APPY1 diagnostic tool is in yielding a quick (and not so dirty) result that would help the ER physician allocate limited resources, in the limited time that he has, given the inverse correlation between sensitivity and false positives. In other words, save the CT Scan for the highest risk cases, which also helps the physician make the call to dispense the CT Scan for the lowest risk cases. There is value in time saved and management of CT Scan usage. In other words, the Company should be aiming for the highest sensitivity possible, and not be too concerned with false positives (i.e. the next gold standard).
    20 Feb 2014, 09:45 AM Reply Like
  • Biotech Dan
    , contributor
    Comments (111) | Send Message
     
    Author’s reply » I'll try to address each thing you mentioned.

     

    First, you're totally right. There isn't much information from the DSMB. All we know is that it should be positive considering the trials are continuing.

     

    Now as for the test:
    You are entirely correct that high sensitivity and high specificity are a goal (as well as negative predictive value).

     

    The issue is, these both have a trade-off. They are like a yin and yang. We'll reduce the complexity of APPY1 and pretend it only measures ONE variable, which we will call A. It is known, and has been proven through research, that a high level of A (say, 10mg/L or more) is indicative of appendicitis. Therefore, we set a cut-off value for A at 10 mg/L. Of course, it isn't this simple... As humans, we are all very different. The levels of A can vary in people following a normal distribution (if you are unfamiliar with this, see this: http://bit.ly/1hyGTZz). Thus if we look at the extreme tails of the distribution, it can be estimated that a fraction of the population exist outside the norms - and ALWAYS have high levels of A (>10 mg/L). This fraction of the population will always have a positive appendicitis result, even when they do not have it. Thus this percentage would be the false positive. If we increase our cutoff, we reduce the false positives and thus increase our specificity. If you are still following, this comes at a risk. Increasing our cutoff will reduce the sensitivity (i.e. people with ACTUAL positive results will have a harder time reaching this cutoff... it's possible they will have elevated A, but not quite high enough for the test). Thus by reducing our false positives (increasing specificity), we are increasing the amount of false negatives (decreasing sensitivity). Again, this is a simplified example. It accounts for one variable and doesn't even take into consideration the fact that the A variable can and likely will vary on a day-to-day basis for any particular individual.

     

    Now here is where "maximizing sensitivity and specificity" comes in. I've already shown you how adjusting the "cutoff" can change the sensitivity and specificity of the result. Imagine me, as a diagnostic tool. If I say "negative" to everyone in the ER, I have a 100% specificity (since there are no false positives, everything is negative to me). If I change this to guessing, saying positive to half the people and negative to half the people, and the condition occurs in exactly half the people, then my new sensitivity/specificity will be 50%/50%.

     

    Generally, a diagnostic test that has a 1-for-1 trade off between sensitivity and specificity (i.e. from 100%/0% to 50%/50%) is considered useless. There is no use to it. This is what the CEO of Venaxis is talking about. We want to maximize these two numbers, while realizing that increasing one results in a trade off for the other. What this actual trade off for APPY1 is beyond me. Obviously it is not 1 for 1, and it likely is not linear either (i.e. proportional). Perhaps the optimal solution is shown when sensitivity is 97% (i.e. we also get the most bang for our back for specificity), or perhaps 97% sensitivity is the desired amount that Venaxis wants (since this is a high level of negative predictive value). I do not know, but I am trying to explain is that it CAN be adjusted. So if we discover within this new result that sensitivity is no longer 97% for APPY1, obviously there are variables that can be adjusted to allow for a sensitivity of 97% (i.e. change cutoff values for A, B or C, or alter algorithm in many other different ways).

     

    As for your understanding of the use of APPY1 - you are entirely correct again in my eyes. APPY1 is a quick and not dirty way of diagnosing a negative appendicitis result. I can expand upon this further if you want, though I've done so a lot in past posts of mine.
    20 Feb 2014, 11:45 AM Reply Like
  • B_Banzai
    , contributor
    Comments (539) | Send Message
     
    Dan in reply to your question "Out of curiosity: what was helpful about it? "

     

    I work with "random" or stochastic processes in my line of work. So I saw your earlier comments in this light. Your further description in terms of probability distributions helped clarify it even further.

     

    I'm not in the auto industry, but my work is similar to trying to establish a vibration test level for a car radio. Let's say I want the radio to last 10 years. Before I put that radio in a car, I want to make sure it is not a "lemon" because production quality will vary from radio to radio as they comes off the line. How severe should my vibration test be? If I set the level too high, none of the radios will pass. If I set it too low, some will fail before the 10 year goal. Further complicating things is that each car will experience a different amount of vibration over it's life time based on how it is used. So we have variability in radio production quality and variation in vibration exposure.

     

    I do the best I can at picking the right vibration threshold, but I end up rejecting some radios that would have exceeded the 10 year life in the majority of cars that hit the road and I end up passing some radios that get married to cars that see extremely harsh vibration levels over their lives and the radios do not last 10 years. Hopefully that example helps the engineers among us!
    21 Feb 2014, 09:56 AM Reply Like
  • Biotech Dan
    , contributor
    Comments (111) | Send Message
     
    Author’s reply » Ah, very nice!

     

    I'm glad I could be so helpful.
    21 Feb 2014, 06:06 PM Reply Like
  • Jeremy LaKosh
    , contributor
    Comments (209) | Send Message
     
    Dan:

     

    I appreciate all the work you've done in describing this product and I'm currently evaluating whether or not to become a potential investor.

     

    Do you know how the units are sold? Are they disposable test strips that feed into a machine like a blood sugar test? I'm trying to understand the product so I can estimate the market size.

     

    Additionally, I'm trying to figure who avoids a CT scan as a result of this test. I spoke with a nurse the other day who told me that a doctor is more apt to want to see the problem, hence the possible use of a CT scan no matter what the APPY1 test results are (according to her opinion). Can you add anything to this? Thanks so much for your time.
    12 Feb 2014, 08:56 PM Reply Like
  • Biotech Dan
    , contributor
    Comments (111) | Send Message
     
    Author’s reply » Hi,

     

    It is exactly as you've described. They are disposable units that feed into another machine.

     

    As for your other paragraph - well it is all speculation. The fact is, not everyone who enters the ER with abdominal pain has appendicits. I think the opinion of the nurse may be misconstrued, it is very possible someone could go for a CT scan even with a negative APPY1 test result.

     

    However, as the case already is, not everyone who enters the ER with abdominal pain ends up with a CT scan. It is VERY clear that a CT scan is not necessary. It's more of a quick, dirty way of pinpointing the problem - without using more conservative diagnosis. For example, if we use APPY1, then we can be more willing to take conservative diagnosis and avoid the CT scan. So, I'm not sure if I'm being very clear, but there already exists people who visit the ER and avoid a CT scan. This group of people can grow even larger if we implement a simple system such as APPY1.

     

    Let me know if you have further questions.
    12 Feb 2014, 10:10 PM Reply Like
  • Joulian
    , contributor
    Comments (5) | Send Message
     
    Hi Dan,

     

    I wanted to thank you for your sterling analysis of APPY's market potential thus far.

     

    The results of the Pivotal Clinical Trial of APPY1 Test are imminent. Do you have any contingency responses to the various result scenarios? For instance, would you maintain your long-term bullish stance if the sensitivity and specificity results were similar to the 2012 test results (97% and 43% respectively)?

     

    The bullish case will speak for itself, should the aforementioned results improve significantly.

     

    At which point would you turn less bullish or bearish? South of 94% for sensitivity and 40% for specificity?
    8 Mar 2014, 08:37 AM Reply Like
Full index of posts »
Latest Followers

StockTalks

More »

Latest Comments


Most Commented
Instablogs are Seeking Alpha's free blogging platform customized for finance, with instant set up and exposure to millions of readers interested in the financial markets. Publish your own instablog in minutes.