Abstract
The Assay Services Department at Genentech, Inc., is a service laboratory that performs drug level measurement and antibody testing in support of pre-clinical animal studies and human clinical studies. As the number of Genentech products has increased, so have the number of studies, resulting in an annual increase in the number of samples generated and an increased demand for assay support. Assay Services has addressed this by increasing the efficiency and productivity of sample handling and assaying through the automation of various procedures. All sample dilutions are now done by automated dilutors, reducing the number of reassays and virtually eliminating the repetitive stress of manual sample dilutions. In addition, two complete ELISA robot stations have been in use over the last two years that have increased throughput by increasing the number of plates per run (a robot can assay ten to fifteen 96-well microtiter plates per run), and by allowing assays to run overnight without requiring the presence of lab personnel. The net gain from the automation efforts has been to double the number of samples run without increasing the number of lab personnel.
This presentation was given at the 2000 International Symposium for Laboratory Automation and Robotics (ISLAR) held in Boston, MA, October 15–18, 2000. The full manuscript is available on CD-Rom and can be acquired by contacting Christine O'Neil, 508-497-2224; email
INTRODUCTION
Before a drug goes to market, extensive pre-clinical and clinical studies must be conducted to assess the efficacy and safety profiles for each drug. While some markers can be readily analyzed at contract labs with commercial kits or systems, novel drugs may need to have specific assays tailored for their detection. The Assay Services Department at Genentech, Inc., transfers assays from several method development groups and is responsible for the maintenance and qualification of these assays. Assay Services supports a multitude of pre-clinical animal studies and clinical human studies with drug level and antibody testing using an ELISA format. Assays may be qualified for one or more matrices, including serum, plasma, amniotic fluid, and urine, and one or more species, including human, monkey, rat, rabbit, and mouse. Over the years, as the number of samples and assays has steadily increased, Assay Services has turned to automation and robotics to increase efficiency and productivity. Automated sample dilutors and robotic ELISA systems, in addition to some changes in sample handling processes, have increased output and reduced the number of reassays.
REASSAYS
Reassaying of samples can be one of the most time and labor-consuming aspects of reporting sample results. Reassays in Assay Services stem from one of two causes. In the first case, the assay does not meet pre-determined assay quality criteria, and all the samples associated with that assay (or plate) need to be reassayed. In the second case, the assay is acceptable, but some samples do not meet pre-determined sample quality criteria. This type of re-assaying involves selecting a handful of samples out of several dozen and ensuring that not only are the correct samples chosen, but that the correct results are documented and reported. Any decrease in the number and frequency of either type of reassays directly increases the productivity of the lab.
Assay Services' original sample dilution method involved diluting samples to a single, targeted dilution and assaying in duplicate. This often resulted in samples falling outside the range of the standard curve and needing to be reassayed at a different dilution. Since a result of less than range (LTR) or greater than range (GTR) did not suggest how much a sample needed to be diluted, samples sometimes needed to be assayed multiple times to achieve a result within the reporting range of the assay.
In the second quarter of 1997, Assay Services implemented a new strategy for diluting samples. This strategy involved diluting samples to multiple dilutions, and assaying at least three dilutions. This method not only greatly increased the chances of obtaining a value within the assay range, it also allowed Assay Services to evaluate the linearity of the sample dilution. Non-linearity in sample dilutions can indicate either problems with the dilutions, assay performance, or matrix interference.
This change in sample dilution strategy caused a large drop in the percentage of reassay runs performed by Assay Services. Before the change, the average reassay rate across all assays was estimated at 50%. After the change, the average reassay rate across all assays decreased to 19%. Reassays are typically very inefficient uses of the assays, since they rarely fill a whole run yet require the same amounts of standards, controls, and manpower. Additionally, they use up very precious samples. Therefore, this 62% drop in the reassay rate represents a very large savings in terms of manpower and reagents. However, it also created some new issues for Assay Services due to the change in sample handling.
AUTOMATED SAMPLE DILUTIONS
The switch to multiple dilutions reduced the number of re-assays, but it also greatly increased the number of dilutions being performed by lab personnel. The repetitive motions of serially diluting large numbers of samples are fatiguing and can be potentially harmful. Also, manual dilutions vary in consistency from person to person. In one test, five research associates (RAs) were asked to dilute the same sample using the same dilution protocol. The samples were submitted to an assay and the results compared to the known concentration of the sample. While the sample results were generally within 20% of the target value, the results from different RAs differed by as much as 40%.
To address both of these issues, Assay Services began the process of automating sample dilutions in 1996. After experimenting with several systems, the Hamilton Microlab AT 2 Plus (Hamilton Company, Reno, Nevada) was identified as the principal dilutor to be used in Assay Services and went into department wide usage in May 1997. All sample dilutions are currently being performed on the Hamilton dilutors.
The Hamilton Microlab AT 2 Plus system consists of an automated dilutor and its control software, Sunrise Plus version 3.2. The dilutor has a 12-channel pipetting arm that uses disposable positive displacement tips. Up to 96 primary sample tubes, either 12×75mm roundbottom or 1.5mL conical microcentrifuge tubes, can be loaded directly onto the Hamilton dilutor. Up to four boxes of 96 micronic tubes can be loaded on the secondary rack for multi-step sample dilutions (the fourth rack was made by Genentech's machine shop to increase the capacity of the Hamilton dilutor). Multiple diluents can be loaded in different sized containers on the reagent rack. Liquid level detection is available on the primary and reagent racks; bubbles and clots trigger warning messages that must be resolved before diluting can continue.
Hamilton dilution protocols are programmed and maintained by Assay Services RAs who have programmer level access to the protocols. All other personnel have user level access, which allows them to run the dilution protocols, but does not allow them to modify or write new protocols. The protocols are divided into assay-specific protocols and generic protocols. Generic protocols are programmed for the maximum number of samples and for common final dilution endpoints (e.g., 1/10, 1/20, 1/40, 1/80). They may be used across the department for multiple projects. A typical dilution protocol can have up to 400 lines of code, but copying and editing capabilities of the software simplify writing new protocols.
Hamilton Microlab AT 2 Plus as configured for use in Assay Services
Before releasing the Hamilton for department-wide use, the accuracy of the dilutions was tested using reference material of known concentration spiked into several different matrices. The mean represents a total of 12 replicates and the percent recovery is based on the expected concentration of the spike. All recoveries were within 10% of the expected concentration.
Spike recoveries of reference material in varying matrices diluted on Hamilton
Concentration is in ng/mL.
% Recovery = [(obtained mean cone - expected cone) / expected cone] × 100
To evaluate the linearity of the Hamilton dilutions, the results of serial dilutions were corrected for dilution and the mean calculated. The percent difference was determined by subtracting the lowest value from the highest value, dividing by the mean, and multiplying by 100. A percent difference of less than 20% was used to determine acceptable linearity. All matrices tested were linear in their dilutions. Whenever a new matrix is introduced into an individual assay, the linearity of diluted samples is tested.
Linearity of Hamilton diluted samples in varying matrices.
Concentration is in ng/mL.
% Difference = [(highest value - lowest value) / mean] × 100
The precision and accuracy of the Hamiltons is tested on a regular basis. Low, mid, and high volumes of liquid dye are pipetted across an ELISA plate by all 12 channels. The percent CV of six replicates of each volume is calculated for each channel and must be less than or equal to 3% at low, mid and high volumes. The accuracy is assessed by comparison to a calibrated single channel pipet and must be less than or equal to 5% at all volumes. This check is done on a monthly basis. On a daily basis, a modified version of this test is done to check accuracy. Replicates at low, mid and high volumes of liquid are compared to historical data.
Over the last three years, the Hamilton dilution protocols have been modified several times in an effort to decrease the amount of sample handling and tube labeling required of the RAs. The current method places the final dilutions in the sample dilution boxes in the exact conformation used by the ELISAs so that diluted samples can be taken off the Hamilton and pipetted directly to the ELISA plate. Control dilutions have long been a part of the sample dilution protocols and standard curve dilutions have recently been added for many assays. The incorporation of automated sample dilutions has reduced the amount of repetitive manual pipetting, decreased the amount of bench time for the RAs by approximately 60%, and increased the reproducibility of the sample dilutions.
ROBOTIC ASSAY SYSTEMS
Over the last three years, as the number of Genentech research products has increased, the number of studies, trials, and samples have also increased. Many assays are now considered high volume assays because hundreds of samples are being assayed each month. To increase efficiency and decrease RA pipetting and bench time, Assay Services brought the first of two robotic assay ELISA systems (RAS) on-line in the second quarter of 1998. Each system consists of a robot arm, a 96-well dispensing station, a reagent dispensing station, a platewasher, a plate-shaker, a microplate reader, and a carousel hotel for holding tips and plates. Each system is capable of performing all steps of the assay. This includes plate washing, sample, standard, and control addition, plate shaking and incubating, reagent addition, and reading of the plates. Samples, standards, and controls are prepared as described above on the Hamilton dilutors then loaded onto the robot system.
Both robot systems were custom built by SCITEC (Wilmington, Delaware), for Assay Services and are running CLARA and ROSCOS software. Assay parameters (incubation times, volume of dispensing, number of washes), are set by Assay Services system coordinators. Users do not have access to change these parameters.
RASI consists of the following pieces of equipment (see Figure 2)
CRS robotic arm (CRS Robotics Inc., Ontario, Canada)
Zymark RapidPlate 96-well pipettor (Zymark Corporation, Hopkinton, Massachusetts)
SLT SpectraFluor Plus (Tecan US, Durham, North Carolina)
SLT PW 96 Platewasher (Tecan US, Durham, North Carolina)
Cavro reagent dispensing unit (Tecan US, Durham, North Carolina)
Microplate shaker (SCITEC Inc., Wilmington, Delaware)
Carousel hotel for microplates, tip boxes (SCITEC Inc., Wilmington, Delaware)
Microplate lid station (SCITEC Inc., Wilmington, Delaware) RASII consists of the following pieces of equipment (see Figure 3)
Mitsubishi Robotic System Model RV-E2 (Mitsubishi Electric Automation Inc., Vernon Hills, Illinois)
SciCLONE 96-well pipettor (SCITEC Inc., Wilmington, Delaware)
SLT SpectraFluor Plus (Tecan US, Durham, North Carolina)
SLT PW 96 Platewasher (Tecan US, Durham, North Carolina)
Cavro Reagent Dispensing unit (Tecan US, Durham, North Carolina)
Microplate shaker (SCITEC Inc., Wilmington, Delaware)
Carousel hotel for microplates, tip boxes (SCITEC Inc., Wilmington, Delaware)
RASI configuration for Assay Services
Depending on the incubation times, each RAS is capable of running 10–15 plates per assay run. Assays can also be set up late in the day and run overnight without the presence of lab personnel. This increases the number of assays that can be performed daily. Additionally, each RAS is equipped with an alarm function and pager that pages the user in the event of an error.
In 1998, two assays were put on the robot. In 1999, 13 assays were running on the robots, and in 2000, 17 assays were running on the robot. Before an assay is qualified for the robot, extensive testing is done to ensure that the manual and robot methods are equivalent. Because the robots require that all solutions be diluted in advance, conjugate concentrations and substrate incubation times often have to be optimized for the robot method. After the assay is optimized for the robot, control quantitation for the manual and robot methods is compared to determine equivalency. Table 3 shows the results of low, mid and high controls for two different assays. Quantitation is similar between manual and robot with percent differences of less than 5% at all levels.
Control comparison between robot and manual method.
Bench time for the RA running a robotic assay has been reduced by 80%. The remaining time is spent doing sample, standard, and control preparation and data reduction. However, not all assays have sufficient volume of samples to convert to a robotic method. To reduce the strain of repetitive pipetting for these assays, a SciCLONE 96-well pipettor (SCITEC Inc., Wilmington, Delaware), is available as a stand-alone unit for sample and reagent addition.
PRODUCTIVITY GAINS
Assay Services maintains and runs more than 40 assays on a regular basis. About half of these can be run on the robot. The reduction in reassays with the change in sample dilution strategy, the gain in accuracy and precision of the sample dilutions with the Hamilton automated sample dilutors, and the gain in efficiency and throughput with the robot ELISA systems, have all contributed to an overall increase in productivity. Table 4 shows the increase in number of samples that have been supported by Assay Services.
Sample support for Assay Services.
From 1998 to 1999, Assay Services was able to assay nearly twice the number of samples with essentially the same number of resources. For 2000, Assay Services is on track to support a similar number of samples as 1999, but for 2001, a large increase in the number of samples is expected.
SUMMARY
‘Do more with less!’ is the ultimate goal for many labs. The Assay Services Department at Genentech, Inc., has addressed the increasing demand for assay support by automating assay steps wherever possible. A combination of changes in sample dilution strategy, automation of sample dilutions, and robotization of ELISAs has led to a doubling of output with the same amount of resources. In addition, the reduction in manual pipetting has decreased the injury risk for all lab personnel. In the future, Assay Services will continue to explore additional methods to further increase productivity.
ACKNOWLEDGEMENTS
Sara Ruiz, Christopher Morrow, Ihsan Nijem, Eric Torres