This article has been updated to correct the QuickMIC error rates, which were reported inaccurately in a table in the cited study.
NEW YORK – Swedish microfluidics firm Gradientech plans this year to launch clinical performance studies for its QuickMIC antibiotic susceptibility testing system as it pursues a CE-IVD mark for the platform.
The company has also begun collaborating with Jeffrey Bender, an infectious disease physician at Children's Hospital Los Angeles, as scientific advisor with the aim of laying the ground for a US regulatory submission, said CEO and Cofounder Sara Thorslund.
Additionally, last week the firm along with collaborators at Uppsala University published a study in mBio detailing the performance of a more highly multiplexed version of the QuickMIC technology.
The QuickMIC system uses Gradientech's microfluidics technology to establish a series of antibiotic gradients in which clinicians can grow their samples of interest to gauge their susceptibility to these drugs. According to the company, the system generates antibiotic susceptibility data in as little as two hours, significantly faster than the 24 or more hours typically required by the culture-based methods commonly used in clinical AST work.
Gradientech was founded in 2009 as a spinout from Uppsala University and began to develop the QuickMIC system around 2012. It also offers microfluidic-based, cell-based assays on its CellDirector research platform.
The system uses agarose gels embedded with the bacteria of interest and treated with a gradient of different antibiotic concentrations. These gels are then monitored using automated microscopy to assess microcolony growth, allowing clinicians to generate minimum inhibitory concentration (MIC) values for the microorganism-bacteria pairings to assess susceptibility.
Use of the concentration gradient provides higher resolution data on bacteria growth, which allows the platform to calculate MIC values more rapidly.
In the mBio paper, the authors used the QuickMIC platform to test 21 clinical isolates. They tested Escherichia coli (6 strains) and Klebsiella pneumoniae (5 strains) against amikacin, ceftazidime, and meropenem; and Staphylococcus aureus (10 strains) against gentamicin, ofloxacin, and tetracycline. They compared the MICs generated by the QuickMIC to values generated by conventional broth microdilution, finding that they were in categorical agreement in 86 percent of cases (meaning they agreed as to whether the isolate fell into the European Committee on Antimicrobial Susceptibility Testing breakpoint categories of susceptible (S), sensitive with increased exposure (I), or resistant (R).)
The agreement between the methods was 80 percent, 100 percent, and 80 percent for gentamicin, ofloxacin, and tetracycline, respectively and 100 percent, 82 percent and 74 percent for amikacin, ceftazidime, and meropenem, respectively. The system's very major error rate (meaning when a resistant strain is classified as susceptible) was 26.2 percent, while its major error rate (susceptible strains classified as resistant) was 3.7 percent, and its minor error rate (resistant strains misclassified as sensitive with increased exposure or vice versa) was 9 percent.
The authors noted several challenges with the system, one being the need to control for the different speeds with which different antibiotics diffused through the gel and form a stable concentration gradient.
"We have been working more on that and doing quite extensive testing to try to characterize how these diffusion times affect the results," said Christer Malmberg, lead development engineer at Gradientech. "As of today, we have run up to 24 different antibiotics, and of those, with 22 we had no problems with diffusion times and our analyses."
Thorslund added that the company has also been working to boost the usability and reliability of the platform, improving automation and ease-of-use to make a system better suited to a clinical lab environment.
Upping the level of multiplexing to enable better throughput has also been a key point of emphasis, Malmberg said.
"We took the approach to miniaturize and simplify the instrument so that you can buy multiple instruments and stack them together," he said. "You can scale your throughput by having smaller or larger stacks—two instruments or four or eight or 12 connected as one system."
He said that this modular approach was developed to help the company address the needs of different size labs.
"There are many small-scale laboratories that don't want to have such a big system, they don't have that many samples coming in," he said. "At the same time, there are large labs that need more capacity. So, we have one machine but it can be adapted to varied markets."
While the QuickMIC system offers higher resolution of measurements within the concentration ranges captured by the antibiotic gradient, this resolution comes at the cost of a narrower overall range.
The mBio authors noted that this could prove a challenge in some cases as "there are significant differences in the antibiotic concentration range needed for efficient AST testing depending on the bacterial species."
They suggested that a relatively straightforward way to address this issue would be to use multiple gradients spanning the required concentration ranges.
Malmberg noted the system's tradeoff between range and resolution.
"We have an accurate linear gradient over the clinically relevant interval, but we do lose information in very low or very high antibiotic ranges," he said. "But we believe that tradeoff is worth it to get the increase in resolution."
In addition to speed, the increased resolution also improves reproducibility, particularly with results that are close to the testing breakpoints, he said. "With traditional two-fold dilution AST testing, bacteria that end up close to the breakpoints, referred to as 'challenging' strains, can test as susceptible one day and resistant the next just based on the variation in the method."
In a 2018 interview, Thorslund said the company hoped to secure the CE-IVD mark by the spring of 2019. This week she said it did not have a timeline for when it hoped to obtain the CE-IVD mark but said that it was planning to start clinical performance trials for the system this fall that would run for around four to six months and expected an CE-IVD following the trials.