Troubles (Part 1)
Laboratory training is a rare, expensive, and time-consuming process, especially in research and development or R&D. You’re not just repeating someone else’s perfected Standard Operating Protocol, it is up to you to design all the steps and make sure they are “foolproof”.
The same applies in any type of higher-order learning; it’s not enough to copy the perfect version of events that someone else has mastered, you have to know all of the ways something can go wrong so you can reverse engineer and troubleshoot any situation. Troubleshooting experiments is a complicated topic, and we rarely have time to cover it amidst our jam-packed curriculum at University. Over the next few weeks I am trying to break this down with videos (and blogs) for my microbiology students using different lab techniques as the context.
Today the focus will be on antibiotic sensitivity testing - the experiments that will flag any new superbugs that will cause havoc in the population because none of the available drugs we have can kill them.
Grand Designs
Let’s imagine that we have isolated a new bacteria, and want to test to see which antibiotics will work against it. Using disk diffusion methods, we can spread the microbe across the entire surface of an agar plate, add a whole series of antibiotic discs, incubate, then measure the zone of inhibition (ZOI) around each disc. The bigger the ZOI, the better the antibiotic works in killing the bacteria. If this ZOI is smaller than a standardised value determined by diagnostic labs across the world, we can consider the bug resistant to the antibiotic.
Let’s pause here for a second and look at 3 experimental design parameters.
How do we know how long to incubate the plates for?
Each of the antibiotic discs added is at a given concentration. How did they figure out the best concentration to use?
The standardised value that determines if a ZOI is resistant or sensitive to an antibiotic. How was this determined?
Incubation Time
To figure out the best incubation time, we’ll need to setup a series of disk diffusion assays where the only variable that changes is the incubation time.
We can pick a series of timepoints - let’s say 6, 12, 24, 48, 96 hours - and setup 3 repeats for each timepoint. You want to check for the level of bacterial growth and the size of any ZOI that appears. Ideally you want the experiment to be done in the shortest amount of time that still gives you valid results.
The next point is how reproducible these results are. We’ve set up 3 replicates (or repeats) for each timepoint, and we can measure the zones of inhibition present for each repeat and assess how much variability there is from plate to plate. You can plot this out using standard deviation or standard error of the mean, and decide what is a reasonable amount of acceptable variability - say plus or minus 5-10%.
Drug concentration
How about what concentration of antibiotics to use? It’s hard to monitor more than one variable at a time, so let’s say the last series of experiments told us the best timepoint to use (24 hours), and now everything will be done using one timepoint. The next variable will be the concentration of antibiotic - we can dilute the drug out across a series of gradually decreasing concentrations, and see how this affects the size of the ZOI. If the drug concentration is too high it will kill all the bacteria on the plate, removing the presence of a measurable zone. If it’s too low, then no ZOI will appear anyway.
Like Goldilocks, a series of experiments will let you find the best conditions better than any one single setup.
Resistance Thresholds
Lastly - once we measure the zone of inhibition, and compare it against the set threshold for resistance or sensitivity, this begs the question - who decided these thresholds? It takes a lot more work than 3 repeats to figure this out - at least thousands of repeats in labs all across the world using the same reference strains of microbes were needed to establish these guidelines.
This isn’t work that a single scientist can do, and it is organisations like EUCAST that generate guidelines that are regularly updated to be able to identify the next wave of resistant superbugs.
The Cycle
None of these parameters were landed on by luck - they were systematically tested again and again until the most optimal version of the experiment was figured out. Eventually you will be able to transform a sprawling, branching web of experimental combinations down to a linear reproducible protocol that someone else can follow and get very similar results.
Once scientists optimise one lab technique, we move on to the next one - the cycle continues.
Jack.
