Wednesday, November 23, 2016

RNA-Sensing USB Stick: Promise Despite the Hype

Earlier this month the newsfeeds were abuzz over a new USB-stick style nucleic acid testing device. Based on technology from the United Kingdom firm DNA Electronics, this device is intended to make testing for infectious diseases portable and inexpensive.  Since the pulse of news was triggered by a publication in a journal, I dove in to see where things really stand.  The device is interesting, but alas the paper describes a prototype far from ready to deploy.  Also interesting to ponder is how this device might stack up against other devices emerging for the portable diagnostics marker, such as Oxford Nanopore's MinION.
A fact that was missing from most textbook diagrams is that the polymerization of nucleic acids kicks off a proton ion each time a nucleotide is incorporated.  A number of clever scientists were aware of this and worked to develop minaturized pH sensors to detect these hydrogen ions.  Ion Torrent is the best known, but they had actually been beaten to the patent office by a UK firm called DNA Electronics (DNAe).  Ion licensed DNAe's intellectual property, so no fireworks.

DNAe and collaborators have now published in the Open Access journal Scientific Reports on using their electronics to detect HIV in a device that plugs into a USB port.  RNA was purified outside the device and then added.  The reaction/test chambers contain the elements of a nucleic acid amplification assay called LAMP (Loop-mediated isothermal AmPlification; I guess LMIA doesn't roll of the tongue).  Isothermal means a much simpler device than a PCR assay, which would require thermocycling.  LAMP operates using at least four different primers and a strand-displacing polymerase; the best explanation I've seen is a short video on New England Biolab's site.  Various primers push off the results of other primers, and also create stem-loop structures.  Self-priming from the stem-loop ends generates long copies of the original and ultimately concatamers, which in turn are amplified until there is a big gooey mess of copied DNA.  RT-LAMP works off of RNA but along the same lines; initial copies of the RNA are made by a reverse-transcriptase.

So the DNAe device amplifies the target if it is present, with heaters on the gadget maintaining an optimal temperature of 65C.  Successful amplification generates hydrogen ions, which can be detected by the electronic sensor.  The device and a schematic of the electronics can be seen below.  Each device appears to have 4 sets of sensing chambers, each with three independent sensing sub-chambers.
RT-LAMP can be quite fast.  Using a SYBR green fluorescence assay, even samples with only 10 input HIV molecules have nearly completed exponential amplification after 30 minutes (NTC=Negative Control).

Tested again in a tube with a pH assay, the RT-LAMP approach shows nearly 100% sensitivity if >1000 copies of the target are in the input, around 90% if 50-1000 copies are present and around 40% sensitivity with less than 50 copies (left panel below).  Promising intra-assay variability is seen on clinical samples (right panel below)

On the chip, things get a bit messier in terms of the pH, as seen below.  Still, a useful signal is apparent between the positives (left) and negatives (right).  However, in their tabulated data the pH assay was considerably less sensitive than the pH tube assay for detecting the target, dropping from 41.20% to 21.20% for less than 50 input molecules, 88.75% to 76.10% for 50-1000 molecules and 95.00% to 88.80% for 1000 inputs (despite fewer than 1000 samples in each category, the paper reports 4 significant digits, a seeming violation of the usual rules).  This is despite the system being run with a positive result called if any of the three chambers are positive.  However, both the tube and chip assays had no false positives, though apparently only 18 negative controls were run.  Nobody likes running negatives, but in a real-life setting the negatives will probably greatly outnumber positives, so even a modest false positive rate can completely degrade the positive predictive value of the test.  In the paper, tests were run for 50 minutes.  While I'm nit-picking, neither plot below has unit labels on the X-axis!

The authors are modest in describing the results as "proof-of-concept", but that certainly isn't how the press treated it.  Not only were the results not likely to be deemed sufficient good for production, but the system requires purified RNA.  Nucleic acid purification in the field is a challenging area that is receiving great attention these days, but is far from a done deal. There is also the question of whether the system's reagents would require cold storage, though the number of room temperature formulations of molecular biology reagents continues to grow.  But these caveats and problems still to be solved didn't stop multiple outlets from declaring the group had a device that can "test HIV levels in under 30 minutes".  Fox Health News added to this the ignominy of labeling DNA Electronics as a "US Company", whereas all signs point to it being a UK company.  I'm too lazy to find all the news items; just search for "DNA Electronics" and set a time filter for November 2016 and you'll find lots of them.

The papers briefly mentions the idea of multiplexing tests, which is a bit confusing.  Presumably they are suggesting that the same RNA sample could be applied to each of the four testing elements to run four separate tests.  I don't see a way to run multiple tests in the same tube and distinguish their results; multiple LAMP assays in the same tube would generate a combined signal. 

The obvious comparator for this device is Oxford Nanopore's MinION sequencer, which has been used in the field for RNA virus assays.  Both are designed to be plugged into (and powered by) a USB port on a supporting computer.  Both require upstream nucleic acid purification.  

As far as the molecular biology and such, the RT-LAMP sensing chip appears to just require a simple mixing of reagents; whether these could be room temperature stabilized and pre-packaged within the device remains to be seen.  Or perhaps the ingredients would be mixed with purified sample and then injected.  Half of all positive samples can be called in about 21 minutes; 50 minutes was considered the endpoint of the assay.  In contrast, MinION has a new 5 minute library prep, but that requires two pipettings.  Reagents haven't yet been made room temperature stable (which for field use in the tropics, really means more like 50C stable).  The flowcell requires precise pipetting to prime which in turn requires the operator's attention over a 15 minute timeframe and one more pipetting to load.  But MinION is really sequencing and can therefore detect a wide range of possible targets and give detailed information on them.  Such details may be useful for tracking resistance mutations and other variations of interest. MinION needs more input material (200ng for the new rapid kit); perhaps LAMP could be used as an upstream booster.  

Comparing time-to-result on the DNAe device to MinION is complicated.  MinION would appear to require an upstream amplification step, so that time would need to be baked in.  MinION also requires the priming, though that could in theory be done in parallel with the DNA extraction (though in practice that is probably too many balls in the air for a single operator).  So most likely DNAe would have an advantage, though the degree of which really can't be assessed.

Cost would be the other key variable, particularly for a test intended to be deployed on a large scale in countries with very limited healthcare budgets.  No cost estimate was given for the DNAe device, but the electronics appear (to this untrained eye) to be relatively simple ones.  Still, that could get pricey; I wish I had a good way to estimate.  Current MinION flowcells are around $500 in bulk, so for a workable cost structure assays would need to be multiplexed, which adds to assay complexity and potentially cost.  

Overall, DNA Electronics appears to have an interesting prototype for an electronic single-use nucleic acid detector.  I always find it unfortunate that these sorts of results are hyped and distorted in the press, as it breeds a broad contempt for technology.  The device shows promise, but there is a long road ahead to truly turn it into DNAe's target of a high quality, inexpensive infectious disease platform for developing countries.


Anonymous said...

If they want to commercialise anything with a LAMP assay they'll also need a license for LAMP, which is notoriously difficult to get. LAMP kit reagents can also be tricky in the field as they don't travel well, even lyophilised. Definitely work to be done before they can bring anything to market, for sure.

Andi B├╝hlmann said...

Great article

LAMP licenses may be difficult to license- I dont know. But in our experience the chemistry travels very well, even in liquid. If the license is too costly, the company has also shown previously to be able to do qPCR on those chips. So I would guess in low price testing situations this may have a future. However, does anybody know the estimated cost of such a chip?

Anonymous said...

Interesting perspective here, thanks for posting Keith. And not really the focus, but LAMP reagents are very sturdy and portable, and a license can't be that hard to obtain. NEB, Lucigen, Optigene (and Eiken) all sell LAMP mixes and Meridian has 7+ FDA cleared LAMP assays. Plenty of papers doing field/direct LAMP with visual detection, so the worry about the materials seems not warranted.