Liquid chromatography using tandem mass spectrometry detection (LC/MS/MS) is the technique of choice for small molecule bioanalyses in clinical and preclinical studies. Sample-preparation steps are required to reduce or remove plasma proteins and other interferences while maintaining or concentrating the amount of the drug compound(s) of interest.
Plasma-preparation techniques include protein precipitation, liquid/liquid extraction, and solid-phase extraction. The first step in sample preparation is to transfer a small plasma aliquot into a secondary tube. Because plasma is a challenging liquid to pipette accurately and reproducibly, aliquotting has traditionally been done manually. This manual pipetting step may be performed hundreds or perhaps thousands of times to complete a single study.
Even in the presence of an anticoagulant such as heparin or EDTA, small clots can form in plasma samples. Also, due to the high surface tension of plasma, occasional bubbles may form on the surface of the sample. During manual pipetting, the laboratory analyst can observe the pipette tip closely to ensure that the desired volume of plasma is transferred.
Air bubbles are not usually problematic because the aspirate step is not started until the pipette tip is below the surface of the plasma. If a clot is encountered, the human eye can detect a stoppage of plasma flow in the clear pipette tip. With human observation, a high degree of accuracy and precision of manual pipetting can be ensured, but the process is slow.
Robotic liquid-handling systems have been applied to solve throughput issues in laboratory processes for years, and would be a logical solution to slow manual pipetting. But plasma’s bubbling and clotting issues are an obstacle. Potentially, a human observer could watch each step to ensure that the robot did not pipette an air bubble or get clogged. But this would eliminate any speed advantage the robot has over manual pipetting.
At Enthalpy Analytical, provider of contract LC/MS/MS bioanalytical services, we had concluded that robotics could not be trusted for plasma sample prep because there was no good way to monitor the pipetting steps. But recently, Hamilton Robotics introduced a new capability for its Microlab Star liquid-handling systems—total aspiration and dispense monitoring (TADM).
TADM monitors the air pressure in each of the pipetting heads in real time and verifies, with a traceable digital audit trail, that a sample has been transferred. Errors are reported in real time, and the system can simply record the errors or react to them.
Figure 1A shows that immediately after initiation of the aspiration step there is a significant drop in pressure. Over the course of the aspiration, the pressure follows a curve that is characteristic of the liquid entering the pipette tip. The pressure is also monitored during the dispense step, shown in Figure 1B, which begins with a pressure increase. As with the aspiration step, the pressure curve is liquid-specific. TADM curves (pressure versus time) are reproducible for a given liquid as these figures show, in which 10 aspirate and dispense steps for each liquid type are overlaid.
Once the TADM profile of a given liquid is established, curve limits can be set. The Microlab Star can be programmed to automatically handle events when these limits are exceeded. For example, our current procedure for TADM-curve aspirate errors instructs the robot to dispense the partially aspirated sample back to the tube and attempt to repipette. If the TADM boundaries are exceeded a second time, the robot moves on. A dialog box appears on the computer screen listing the well numbers for which there were TADM-curve errors that could not be resolved by repipetting.
Figure 2 shows the TADM curves for a 35 sample test run of plasma spiked with an analyte of interest. Two of the TADM curves deviated significantly from the average profile for plasma in both the aspirate and dispense steps. These errant curves, where there is no initial pressure drop during the aspirate step, are indicative of an encounter with an air bubble just above the surface of the plasma sample.
In this example, the robot was instructed not to repipette the samples. When the samples in this example were analyzed on an LC/MS/MS instrument the results provided quantitative substantiation of the errant TADM curves observed in Figure 2. In addition, because TADM provided traceable evidence that these two samples were pipetted incorrectly, they could be justifiably eliminated from data analysis.
We generally observe one failure out of every 1,000+ pipetting events; the example discussed in this article is unusual. Typically we centrifuge samples for at least 10 minutes to eliminate air bubbles and send clotted material toward the bottom of the tube. In the example above the samples were only centrifuged for about five minutes at 4,000 rpm.
With the full automation of our plasma-preparation methods we have been able to save a significant amount of time without any sacrifice in method accuracy and precision, ultimately delivering results to our customers much faster.