+ Watch MELA
on My Watchlist
A medical device company which focuses on designing and development of a non-invasive, point-of-care instrument to assist in the early diagnosis of melanoma.
Lots of interesting pitches already on MELA, so I won't rehash the basics. But I'm very confused by the results of the pivotal study. We know the study analyzed 1831 lesions that were already targeted for biopsy based on clinical assessment by dermatologists. Of these 1831 lesions, 127 turned out to be melanomas and 114 were melanomas suitable for endpoint analysis. Either way, MelaFind gave a positive result for 98% of melanomas, indicating 98% sensitivity. This cannot be compared with the sensitivity of dermatologist's clinical assessments as we don't know anything about the lesions that the dermatologists chose not to biopsy and were never entered in the study.Here's where things get extra sketchy. The specificity (likelihood that a non-melanoma would be correctly identified as a non-melanoma) was only 9.5%. In other words more than 90% of non-melanomas were identified as melanomas by MelaFind. Since the vast majority of lesions in the study are non-melanomas, the bottom line is that MelaFind will have a minimal impact in terms of reducing biopsies.To look at this another way, without MelaFind 1831 lesions would have been biopsied. With MelaFind, 1667 lesions would have been biopsied. Therefore 164 biopsies would have been avoided at the price of missing two melanomas. How good do those numbers look now? And of course we have no idea how MelaFind might perform in the broad population of lesions where the decision to biopsy hasn't been made yet. It's very possible that the poor specificity of MelaFind might actually lead to an increase in negative biopsies in a lower probability pretest population.MELA has chosen to obfuscate the specificity problems by declaring that the specificity of the dermatologists was only 3.7%. So who cares if MelaFind only has a specificity of 9.5%, it's still three times better than the dermatologists, right? But how did they arrive at that number of 3.7% given that the study only included lesions deemed suspicious by the dermatologists? We don't know anything about the lesions that were examined but not biopsied. We don't know how many there were or how many of those turned out to be melanoma. I've looked for the slides or primary data for the MelaFind pivotal trial and I can't find it. Just endless repetitions of the 98% sensitivity and the comparison between MelaFind and dermatologist specificity numbers. However, I did find a published meta-analysis comparing the accuracy of dermatologists and primary care physicians in diagnosing melanoma here: http://www.ncbi.nlm.nih.gov/pubmed/11735713. In that analysis the specificity of dermatologists was 70-89% and primary doctors was 70-87%. Quite a difference from 3.7%, no?Now I'll be the first to admit that I hate statistics, and frankly I suck at them. These numbers could be analyzed in a much better way using Bayes theorem and crap like that. If anyone notes major errors in my analysis or wants to take it to a higher level, feel free to chime in and please be as rude as possible. Until then I'm firmly red and possibly buying puts before the FDA panel November 18.
I BELIEVE THAT WE MAY HAVE LOST THE GAME AT LEAST MFOR NOW.THEY WILL REVEW THE ADDED INFO IN JANUARY AND MAY APPROVE THE DEVICE.IN THE MEAN TIME THE CO IS RUNING OUT OF CASH.I STILL BELIEVE THAT THIS DEVICE WILL BE APPROVED BUT IT COULD BE IN THE HANDS OF OTHERS BY THEN.LAST CHANCE IN JANUARY.HANG ON AND GOOD LUCK.
BATS data provided in real-time. NYSE, NASDAQ and NYSEMKT data delayed 15 minutes.
Real-Time prices provided by BATS. Market data provided by Interactive Data.
Company fundamental data provided by Morningstar. Earnings Estimates, Analyst Ratings and Key Statistics provided by Zacks.
SEC Filings and Insider Transactions provided by Edgar Online.
Powered and implemented by Interactive Data Managed Solutions. Terms & Conditions