Liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) has become a well-established technique in clinical laboratories around the world. However, for laboratories with minimal experience, taking the leap and establishing LC-MS/MS services can be daunting. Unlike dedicated automated clinical analyzers, LC-MS/MS platforms have wide-ranging applications and broad flexibility that pose both benefits and challenges for clinical laboratories.
One drawback of the flexibility inherent in LC-MS/MS is that each clinical laboratory often uses a unique set of methods that relies on vendor application teams and service engineers. However, these support groups often operate in isolation, and even the best service contract rarely provides complete peace of mind when the system isn’t operating properly and racks of extracted patient samples await processing.
Laboratories can counterbalance this limitation and assure high-quality testing by creating an in-house early detection system for suboptimal instrument performance—one that ties into a preventive maintenance program that is tailor-made to the applications run in that laboratory. System suitability tests (SST) are an essential part of this quality system.
SSTs serve primarily to provide confidence that a system is in a suitable state before a batch is submitted, but they fulfill other purposes as well. SSTs also act as troubleshooting guides for suboptimal systems, and they enable longitudinal assessment of parameters to shape future preventive maintenance.
How to Develop System Suitability Tests
Many clinical laboratory professionals get their first exposure to an SST when a service engineer runs a generic version after an instrument installation or major service. In such cases the SST solution usually comprises a mixture of generic compounds (typically acidic, basic, polar, and non-polar). The engineer injects the solution into a defined column using simple mobile phases and assesses response to ensure the LC and MS components of the platform meet the vendor-set performance qualification.
While useful, this type of SST has limited utility when assessing system performance on a day-to-day basis in a clinical laboratory. Assay-specific SSTs run immediately prior to submitting a batch best fulfill this role, and the remainder of this article will focus on these specific SSTs.
A specific SST material is made up of the target analyte(s), internal standard(s), and extraction/reconstitution solvent for the assay. For example, an estradiol SST could contain 10 pg/mL estradiol and 10 pg/mL 13C3- estradiol internal standard in a 40% methanol solution. A laboratory would run the SST before submitting a sample batch to ensure that each component of the system (mobile phases, column, pumps, auto-sampler, mass spectrometer, acquisition method, etc.) meet the in-house performance criteria for that method. Ideally a lab would make the SST material in bulk, then aliquot and store it for quick and easy use.
Labs usually run the SST in combination with reagent blank samples as a batch in the following order: 1) reagent blank, 2) reagent blank, 3) SST, 4) reagent blank (carryover blank).
Choosing SST Parameters
Clinical laboratories can include many parameters when assessing the performance of an SST sample, and experience with the specific assay helps determine what the acceptance criteria should be set to. Where possible, investigate the SST response from failed validation batches: These can act as a guide to where criteria should be set. If this data isn’t available, then an educated guess will suffice for initial values and, as a lab gains experience, adjusted to make the acceptance criteria more suitable.
In my own experience, checking the peak intensity, peak shape, retention time, and the initial LC back pressure has been sufficient prior to batch submission. However, for longitudinal assessment purposes, additional parameters recorded by the acquisition software—peak symmetry, signal to noise, plate count, and the LC back pressure trace over the course of the SST sample—have proven useful.
In addition, if auto-sampler imprecision is an issue, expanding the SST batch to include multiple SST injections provides a reproducibility check, with the trade-off being a longer batch to run before a decision can be made to proceed.
When first establishing an SST, a common question is, “what concentration should my SST be set to?” There is no right or wrong answer to this and, again, experience will act as a guide.
If the method has a challenging lower limit of quantitation (LLoQ), then setting an SST around this level (1x or 1.2x the LLoQ) makes sense. Alternatively, if an assay is prone to carry-over then an SST concentration at the upper limit of quantitation would help assess whether this is an issue. When this kind of information is not known, I usually suggest beginning with an SST concentration around 1.5x or 2x LLoQ to provide a degree of confidence in the assay’s sensitivity.
Beginning with this concentration also provides a signal large enough to distinguish a case of a missing peak from a case of a severe loss of sensitivity. The laboratory would then review concentration during the lifetime of the assay to ensure that it is suitable.
I also recommend that laboratories keep for easy reference a copy of a typical SST chromatogram, the LC back pressure trace, and the key acceptance criteria. These key inputs could be stored electronically—such as a shortcut on the instrument desktop computer—or even posted physically on a clipboard beside the instrument. In addition to the SST, chromatograms from the reagent blanks, injected with the SST, should be assessed to ensure that they do not contain analyte peaks.
When an assay is known to have closely eluting assay interferences, such as 3-epi-25OH Vitamin D3 in a 25OH Vitamin D3 assay, these substances should be included in the SST material to ensure sufficient chromatographic resolution before submitting a sample batch. When closely eluting unknown interferences are present in patient samples, such as with some total urine metanephrines assays, it is good practice to use a representative patient sample as the SST, either stored as extracts, when stable, or extracted fresh on the day when instability is an issue.
When Things Go Wrong
SSTs can fail acceptance criteria in many ways, such as the peak seeming to be in the wrong place, being too small, or missing. In my experience most of these issues have a simple cause that the laboratory can easily resolve, including but not limited to:
- The auto-sampler is sampling from the wrong vial/there is no vial present/the vial is empty
- The wrong type of sample plate has been placed in the auto-sampler
- The wrong method has been submitted
- There is a leak in the LC or the LC is not connected to the MS
- The wrong column/mobile phase/ion source is being used
To seasoned operators these possible causes may seem obvious; however, when an SST fails and there is time pressure to get the system up and running, even experienced operators can miss them. This is why it’s worth having a checklist as a first pass before starting any extensive troubleshooting.
When troubleshooting begins I often refer to the mantra championed by John Dolan in his “LC Troubleshooting” column in LCGC North America—divide and conquer (July 2004;22:618-622). The key to this strategy is using a decision tree to begin troubleshooting based on the information provided by the SST and the LC back pressure trace.
For example, a peak that is eluting late, combined with a lower than expected LC back pressure, suggests that there is an issue with solvent delivery or mobile composition. This immediately rules out the mass spectrometer and directs the trouble-shooter toward looking for a leak or checking that the mobile phase has been made correctly. If the same issue occurred but the LC back pressure was fluctuating in a rhythmic fashion, then a further level on the decision tree could direct attention specifically to the check valves on the pump heads.
The same approach holds true for assessing reagent blank samples, where we commonly see two types of issues: the carryover blank has a peak in it (injection 4) and all the blanks have a peak in them (injection 1, 2, and 4).
The first issue appears to be carryover and would lead the operator initially to investigate the auto-sampler (wash solvent levels or composition, needle seal, etc.). The second issue would suggest system contamination, and the troubleshooting scientist might start with the mobile phases or wash solvents.
Building troubleshooting decision trees is an incremental process based on experience, with each issue used as a learning opportunity with the symptoms and solutions recorded and later added to the tree.
When the SST Is Fine and Things Still Go Wrong
When problems occur despite a well thought out SST, it can be a very frustrating experience. Yet this provides a chance to fine-tune the preventive maintenance program. For example: An increasing workload for 25OH vitamin D analysis led us to repurpose an LC-MS platform that previously ran small batch, esoteric assays.
On one occasion the 25OH vitamin D SST generated peaks that were late eluting but within the in-house acceptance criteria. Subsequently we submitted two plates of extracted samples. During the sample analysis the retention time drifted further backward until the peaks had slipped off the back of the chromatogram window and were being diverted to waste, causing the batch to fail.
As retention time is independent of the mass spectrometer, the operator began troubleshooting with a focus on the LC system, where it was revealed that a pump seal wash bottle had overflowed due to a worn pump seal. A check of past SST chromatograms demonstrated a subtle but continuous retention time shift over the preceding days along with a slight decrease in LC back pressure—indicators of a gradual pump seal failure. This experience highlighted an issue with the maintenance schedule for the repurposed system which was no longer suitable now that the platform was experiencing a much higher workload.
The lesson learned: With careful record keeping and retrospective analysis of SST data, our laboratory was able to tighten the acceptance criteria and increase the maintenance frequency to prevent a repeat occurrence and enable a more robust service.
While LC-MS instruments have become more robust and easier to use over the years, these platforms remain considerably more challenging to maintain than automated clinical analyzers. However, with the appropriate use of SSTs and good record keeping combined with an evolving troubleshooting and preventive maintenance scheme, clinical laboratories can offer high- quality testing that benefits from the selectivity and sensitivity that this technology has to offer.
Michael Wright is the Scientific Director at LGC–Drug Development Solutions in Fordham, Cambridgeshire, UK. +Email: [email protected]
CLN's Focus on Mass Spectrometry is supported by Waters Corporation.