Elizabeth Holmes presents the new miniLab to AACC 2016 in Philadelphia
View Elizabeth Holmes’ AACC 2016 Keynote Presentation in full, here…. (read my analysis, below)
After the CMS’s announcement of immediate jeopardy (read the letter here) on July 7th, 2016, I was not sure Elizabeth Holmes, CEO of Theranos, would be found on the AACC 2016 stage. She came and she presented. I applaud her moxie. However, in reply to a question about the Edison instrument and the operation of Theranos’ labs on patient samples to-date, Holmes replied, “In the appropriate forum, we’ll address those. But today we’re hoping to be able to engage on a scientific exchange on this [new] platform.” It is a most-unsatisfactory answer when the clinical chemistry community assembles to hear exactly those answers.
The presentation was an largely a marketing announcement of Theranos’ new system, miniLAB ( Mechanism of Action (MOA) video here), which has not been cleared by the FDA. It is “For Investigational Use Only” (FIUO). #fail
My thoughts on miniLab are detailed here…
The Sample Collection Device (SCD) was purported by Holmes to reduce pre-analytical error such as cleansing-interferents, hemolysis, clotting, and short sampling. The device is a Rube Goldberg of collection technologies including two capillary tubes, internal puncture needles, plungers, nanotainers, and both EDTA and Lithium Heparin anticoagulants, which we know are both required to be expansive in the resultant test menu. Some tests cannot be done on LiHep samples and vice versa. Conspicuously absent in the training video, there is no mention of a mixing step by the user. Without any data presented, one wonders if Theranos has reduced any pre-analytical errors.
The collection system does not appear to be optimized for human factors usability. There are a lot of steps involved and so a lot of things can go wrong. Nothing was presented to suggest that site prep, technique dependencies, interstitial fluid dilution, adequate distal perfusion, hemolysis, anti-coagulant homogenization, cell settling, or other error-vectors have been eliminated, or even reduced from capillary draws.
The worst part is the 170 µl sample size. An average hanging blood drop of peripheral whole blood is 20 µl. It does not seem, to this former phlebotomist, Theranos understands the challenge of fingerstick-sampling from a frail, dehydrated, 80 year old lady or her 84 year old, obese husband with congestive heart failure and gross peripheral edema. #fail
The miniLab “cartridge” includes a weird assortment of microtainers, pipette tips, reagents, slides, chambers, reaction vessels, and sample wells all covered by a hinged door. Apparently the cartridge can be customized to the panel. The panel configuration, according to Holmes, is done during manufacturing. With over 120 assays across a half dozen methodologies planned, real world panel-selection is definitely not a feature of miniLab. The combinatorial math for patient panels, need for cold chain storage, wasted storage space (enormous interior dead space), and proper cartridge selection by the user, all seem to be overlooked in this design. #fail
The miniLab analyzer itself is a panoply of methodologies. The demonstration videos were shot on breadboard models. The interior componentry changed throughout the presentation in sequential videos. Where did the Hemotology slide stage go when the Thermocycler was being demonstrated? Inside miniLab, you can find a material handling robot and pipetter, cameras and lights, optics, a thermocycler for PCR, a sonicator for lysing, a centrifuge, a fluorometer, a luminometer, and more. In watching Mechanism of Action (MOA) video, I was reminded of the hours (mintes?) of fun watching the Beckman CX7 mix samples in the tube on the sample wheel, circa 1993.
Cost, Service/Cleaning, and Mechanical Interference
In viewing the inside of the analyzer, the complexity of cartridge staging, sample handling, preanalytical sample prep, and a half-dozen methodologies, all inside one box, defy DFM (Design for Manufacturing) best practices. Everything appears to be hand assembled, which WILL lead to out-of-box failures and infuriating service challenges. The cost of assembly, service, and components will be enormous for this device in this class – meant to be used at the point of care, not in a Core Laboratory. Its cost and price will both be prohibitive. (The same concerns apply to the aforementioned SCD, Sample Collection Device).
The interior surfaces are replete with crevices. Although the sample is moved with disposable pipette tips, the sonicator and the centrifuge both appear open-topped. The potential for aerosolization is high, the chassis is vented, and the device simply does not appear to be able to be cleaned or sterilized.
One wonders with a sonicator and a centrigue in the same box, if Theranos has truly worked out the issues with cell-settling, and mechanical interferences in the same instrument as cameras and optics (spectrophotometry and mirrors). One wonders, with the thermocycler in the same box as ELISA reactions (may or may not be used on the same patient sample), if Theranos has truly worked out the thermodynamic issues with sample and reaction temperatures. One wonders, with software managing the calibration, quality control, image analysis, spectrum analyis, and transfer to a remote site, if Theranos has truly worked out the challenges of multi-threaded software working in real-time on a critical system which cannot be directly monitored by trained technician. It would have been great to see a comprehensive data presentation, but to be fair, 45 minutes was too short for these questions to be answered.
miniLab is meant to be a distributed system with pre-analytical and analytics steps occurring in a distributed lab environment, at the point of care. The miniLab front end will communicate with a server-based backend, the TVA (Theranos Virtual Analyzer). In TVA, result approval will be managed by a “certified laboratory”. This distributed testing model was tried by Medical Automation Systems with the original RALS system at UVA (University of Virginia) in the early 90’s. It was unappealing to most labs and most end-users. The turnaround time benefit of a point of care test is perturbed by waiting for an unseen entity in the Core Lab to review and approve test results before release back to the originating instrument. And as aforementioned, the trained Medical Technologist has no ability to gauge sample integrity, reagent integrity, or instrument operations. MT’s don’t like that environment very much – neither do Clinical Chemists. We know that for certain. #fail
Theranos’ miniLab is an integration of every imaginable technology from sampling (pre-analytical), methodologies (analytics), and closed loop approval and data transfer technologies (post-analytical). It does not appear to be an appealing or even feasible integration on many levels. It lacks best in class analytics, optimization of point of care principles (namely short turnaround times), serviceability, cost per sample, device complexity, or high-margin business model. I give Theranos a D- on SCD, miniLab, and TVA and predict a short arc into oblivion.