Introduction
Contamination by Gram-negative bacteria is one of the major reasons for sterile products recalls in the United States [1]. Endotoxins are major structural components in the cell wall of Gram-negative bacteria [2]. If present in injectable biopharmaceuticals, they can trigger reactions leading to extreme morbidity, i.e., fever, or mortality [2, 3]. Some of the potential sources of endotoxin contamination are water, excipients such as solvents, thickening agents, chelating agents, antioxidants, reducing agents, preservatives, buffers, bulking agents, and special additives [2].
Endotoxin analysis is performed by three basic technologies, which are based upon the highly sensitive reagent limulus amoebocyte lysate (LAL). These technologies are the Gel-Clot, Endpoint analysis, and Kinetic assays [4-6]. A previous study published by our laboratory demonstrated the validation of a portable endotoxin detection system with a handheld spectrophotometer for the analysis of biopharmaceutical samples under current good manufacturing conditions (cGMP) [3]. Another study demonstrated the capacity of the system to analyze cell therapy products [7]. The portable endotoxin detection system is based upon the kinetic chromogenic detection of pyrogens by measuring color intensity as related to the endotoxin concentration present in a given sample. Polystyrene cartridges contained all the reagents necessary to perform the test. Samples analysis was completed within 15 minutes. The portable system was a point of source test which allowed the rapid test of the samples nearby the sampling site. However, because the reader has only one slot, the throughput was one sample every 15 minutes. A regular biopharmaceutical endotoxin laboratory receives a very high volume of daily and weekly samples of water, in-process, and finished products, which are tested as soon as possible. Production runs are constantly changing and continuously monitored to provide important information on the capabilities of the different processes to reduce pyrogen loads. There is a limited throughput capacity with the endotoxin detection system when compared to traditional 96-well kinetic and chromogenic endotoxin test. However, a new system, e.g., a multi-cartridge endotoxin detection system, has been developed which analyzes up to 5 samples at a time. Furthermore, the traditional kinetic and chromogenic systems require intensive training, standard curves, accessories, and reagents, which are not needed in the multi-cartridge endotoxin detection system. However, no study has been reported on the evaluation of the multi-cartridge endotoxin detection system to analyze biopharmaceutical samples. Nor the quantitation of a given endotoxin concentration spiked into biopharmaceutical samples has been ascertained. The major objective of this study was to evaluate the system for the analysis of biopharmaceutical samples.
Materials and Methods
System Description
The multi-cartridge endotoxin detection system is based upon kinetic chromogenic detection of pyrogens by measuring color intensity related to endotoxin concentrations in a sample. The multi-cartridge endotoxin detection system is a reader used to perform kinetic LAL testing. The multi-cartridge endotoxin detection system is a rapid test system which is comprised of a test cartridge along with a spectrophotometer. The system measures color intensity directly related to the endotoxin concentration in a sample. There is a linear relationship between endotoxin concentration and absorbance values. 
The software is configured by the manufacturer to run a two-step kinetic chromogenic assay, utilizing the multi-cartridge endotoxin detection system test cartridges. The system utilizes existing FDA-licensed LAL formulations. It provides quantitative LAL test results in approximately 15 minutes. Polystyrene cartridges contain precise amounts of LAL reagents, chromogenic substrates, and control standard endotoxin (CSE) loaded into channels. The cartridges contain 2 sample channels and 2 spiked channels (3). The unit has 5 sample slots or bays and can analyze 5 different samples simultaneously (Figure 1).
The endotoxin measuring software is installed on a standard Dell PC computer station running the Windows XP operating system. The multi-cartridge endotoxin detection system kinetic reader is connected directly to a data management computer with a USB cable where the data are managed by a data management software called the endotoxin measuring software. The software provides the ability to configure the user access level to meet specific needs. A username and password are required to access the functions available. It provides high levels of security by logging system activities to ensure that data integrity and change histories are maintained. The endotoxin measuring software can be used to generate reports, view product information and test results, store data, and keep information/data secure using system access levels, user accounts and user passwords.
An audit trail report can be printed and archived manually. The computer is connected to a remote server located in the server room. All the test results generated are saved directly to a server in form of a plate file. The data results generated in the file are not editable. All the data can be saved to any location by changing the fi le path. The test result reports can be printed, saved, and retrieved from the server.
Traditional Bacterial Endotoxin Test
The Gel-Clot [4] reagents were composed of buffers and LAL of 0.03 EU/ mL sensitivity. Control standard endotoxin (CSE) was diluted to 1.0, 0.06, 0.03, 0.015, and 0.0075 EU/mL, respectively. Dilutions were prepared daily in endotoxin free sterile water for injection from reconstituted CSE stock solutions (10,000 EU/mL). Product positive control (PPC) samples were prepared by spiking no more than 10% of the total sample volume with 10 EU/mL. Water for injection (WFI) samples from different sampling locations were analyzed at 1:1 dilutions while 1:10 dilutions of products in endotoxin-free sterile WFI were tested. All glassware used in the study was depyrogenated for a minimum of 1 hour at 250°C. Samples (0.10 mL) were added to depyrogenated tubes containing 0.10 mL of lysate. After addition, samples were incubated for 1 hour at 37°C. The end point value was the lowest concentration of endotoxin at which the lysate formed a solid gel-clot upon inversion [4].
Multi-cartridge Endotoxin Detection System Test Procedure
Quadruplicate 0.025 mL aliquots of WFI and products described above were loaded into the four sample reservoirs of the cartridge. All the 5 sample bays were used for the analysis (Figure 1). The reader draws and mixes the sample with the LAL reagent in two wells (sample wells) and positive product control in the other two wells (spike wells). Cartridges with LAL sensitivity of 0.01 EU/mL were used for all multi-cartridge endotoxin detection system analysis. The cartridges were calibrated by the vendor and documented to be able to quantify endotoxin concentrations ranging from 1.00 to 0.01 EU/ml. The sample is then incubated at 37°C, inside the reader, and combined with the chromogenic substrate. After mixing, the optical density of the wells are measured and analyzed against an internally archived standard curve. The archived standard curve specific for each batch of cartridges is constructed using the log of the reaction time against the log of the concentration [8]. The sample and spike values are calculated by interpolation of the standard curve using the reaction times. The system simultaneously performs testing in duplicate and averages the results. Acceptance criteria for a valid assay is an archived curve correlation coefficient of > -0.980, a positive product control (spike) recovery of 50-200%, a coefficient of variation (CV) of <25% for both the negative control channels and positive control channels, respectively.
Quantitation of Endotoxin in Samples
To determine the capability of the system to detect endotoxin in biopharmaceutical samples, WFI and product samples were spiked with 0.10 EU/ml of the Reference Standard Entotoxin (RSE). All 5 bays were used in the assays. Several dilutions of the RSE with a concentration of 1 EU/mL were prepared to deliver a 0.10 EU/ml endotoxin spike into each WFI and product samples. The assay was repeated 3 times using different analysts on different days.
Results and Discussions
To determine the applicability of the multi-cartridge endotoxin detection system for the analysis of biopharmaceutical samples of WFI and finished products, 3 different analysts performed a total of three tests. The reproducibility, robustness, and accuracy of the assay were determined by performing the assays on different days, using different reagents, samples, and sample preparations. During the preliminary studies it was observed that some spike recovery values were >200% (data not shown). This is often caused by the presence of glucans introduced by pipettes or pipette tips with filters/cotton. Some samples also exhibited coefficient of variation failures caused by pipetting, small bubbles (sometimes not visible). Because of the sensitivity of the assay, only calibrated pipettes were used and the pipette tips were replaced with tips capable of dispensing 25 μL-1000 μL with an endotoxin detection level of 0.005 EU/ml. Once the tips were changed the consistency and reproducibility of the values for both the coefficient of variation and spike recovery were always within acceptable ranges. 
When WFI samples were analyzed by the multi-cartridge endotoxin detection system, all 3 analysts reported similar results. All the WFI samples demonstrated an endotoxin value of less than 0.01 EU/ml (Table 1). The recovery of the spiked values, e.g., spike recovery, was within acceptable levels, e.g., 50-200%, (Table 1). Values ranged from 76 to 188%. The CV for both the negative product control channels and positive product control channels were within acceptable limits. The values ranged from 0 to 2.4% and 0.5 to 11.2, respectively. Evidently the assay was robust, reproducible, and accurate. The Gel-Clot results for all samples demonstrated an endotoxin value of less than 0.03 EU/ml. No endotoxin was detected by both methods.
A study of different biopharmaceutical products showed similar results by the 3 analysts. The concentration of endotoxin in all products was less than 0.01 EU/ml (Table 2). The recovery of the spiked values, e.g., spike recovery, was within acceptable levels, e.g., 50- 200%, (Table 2). Values ranged from 86 to 151%. The CV for both the negative product control channels and positive product control channels were within acceptable limits. The values ranged from 0 to 2.6% and 0.0 to 15.5, respectively. The Gel-Clot results for all samples demonstrated an endotoxin value of less than 0.03 EU/ml. These results indicated that no endotoxins were detected by both methods. The presence of endotoxin in biopharmaceutical samples can be detrimental to the sample quality and stability. Several protein purification steps during the manufacturing of biopharmaceuticals do not remove pyrogenic substances [9]. In some situations there is an established endotoxin limit which provides a clear indication of the process performance during protein purification. If any pyrogenic substance is present in the samples, quantitation of the pyrogen will provide an accurate estimation of the type of load which might indicate the degree of the lack of process control.
To determine the endotoxin concentration present in artificially contaminated samples, 3 analysts spike a final concentration of 0.10 EU/mL to WFI samples (Table 3). The test results are shown in Table 3. The average endotoxin value detected by all 3 analysts was 0.120 EU/ ml. Endotoxin values ranged from 0.079 to 0.178 EU/ml. The recovery of the spiked values, e.g., spike recovery, was within acceptable levels, e.g., 50-200%, (Table 3). Values ranged from 70 to 161%. The CV for both the negative product control channels and positive product control channels were within acceptable limits. The values ranged from 0.8 to 13% and 0.0 to 9.4, respectively.
When samples of biopharmaceutical products were spiked with similar concentrations of endotoxin, the average value was 0.110 EU/ml (Table 4). Endotoxin values ranged from 0.063 to 0.183 EU/mg. The recovery of the spiked values, e.g., spike recovery, was within acceptable levels, e.g., 50-200%, (Table 4). Values ranged from 78 to 168%. The CV for both the negative product control channels and positive product control channels were within acceptable limits. The values ranged from 1.0 to 17.4% and 0.7 to 13.3, respectively. The quantitation studies demonstrated that the values obtained by the different analysts were reproducible and that the assays were robust. A previous study demonstrated that archived standard curves were reproducible and reliable for the quantitation of endotoxin [8]. That study has shown that there were no differences between archived standard curves and daily standard curve prepared for quantitative endotoxin testing [8]. In this study we spiked only one concentration of endotoxin to verify the endotoxin spiked even though the cartridges were already calibrated by the vendor. Furthermore, the cartridges used in this study have been approved and licensed by the FDA for endotoxin testing [7]. The endotoxin spiked was always detected and quantified with values obtained within acceptable ranges by different analysts, on different days, and using different endotoxin preparations. When there was no endotoxin the system did not detect any (Tables 1 and 2). However, when an endotoxin level was present, quantitation was accurate, reproducible, and robust (Table 3 and 4). 
A larger side-by-side study was performed to determine the capability of the multi-cartridge endotoxin detection system to analyze a high volume of WFI samples and to compare the results with the Gel-Clot test. Fifty-eight samples were analyzed with the Gel-Clot and the multi-cartridge endotoxin detection system (Table 5). The Gel-Clot test was performed using a lysate sensitivity of 0.03 EU/mL while the multi-cartridge endotoxin detection system was performed with a lysate sensitivity of 0.01 EU/mL. The USP limit for endotoxin concentration in WFI is 0.25 EU/mL. All water samples were found to be negative by using the Gel-Clot test and multicartridge endotoxin detection system (Table 5). Endotoxin values were way below the required limits with less than 0.03 EU/mL for the Gel-Clot and less than 0.01 EU/mL for the multi-cartridge endotoxin detection system. Although the multi-cartridge endotoxin detection system system was 3 times more sensitive than the Gel-Clot, no inhibition or enhancement of the reactions were detected. Each multi-cartridge endotoxin detection system assay delivered results within 15 minutes while the Gel-Clot took 1 hour to complete. This represented a 75% reduction in the time needed to evaluate the presence or absence of pyrogens in WFI and finished products. The theoretical throughput with the multi-cartridge endotoxin detection system was 160 samples analyzed in an 8 hour shift. Therefore, the multi-cartridge endotoxin detection system can process approximately 800 samples a week.
In conclusion, rapid monitoring of product testing and water can alert production personnel of potential problems before they become critical. Corrective actions can be taken as soon as possible to reduce pyrogen load and levels of endotoxin during manufacturing of biopharmaceuticals. The release of WFI and products in biopharmaceutical operations requires endotoxin testing to provide information to support the safety and efficacy of the processes in the removal of pyrogenic substances.
References
1. Jimenez, L. Microbial diversity in pharmaceutical product recalls and environments. PDA Journal of Pharmaceutical Science and Technology 61, 383-399, (2007).
2. Williams, K. Endotoxin: Relevance and control in parenteral manufacturing. In Microbial Contamination Control in the Pharmaceutical Industry. Luis Jimenez (Editor). Marcel Dekker Publishers, New York, N.Y., Chapter 8, 183-249, (2004).
3. Jimenez, L., Rana, N., Travers, K., Tolomanoska, V., Walker, K. Evaluation of the Endosafe® Portable Testing System™ for the rapid analysis of biopharmaceutical samples. PDA Journal of Pharmaceutical Science and Technology 64, 211-221, (2010).
4. United States Pharmacopeia. Bacterial endotoxin test. U.S. Pharmacopeia, Mack Publishing Company, Easton, Pennsylvania, 29, 2521-2524, (2006).
5. McCullough, K.Z., Weidner-Loeven, C. Variability in the LAL test: comparison of three kinetic methods for the testing of pharmaceutical products. J Parenter Sci Technol. 46, 69-72, (1992).
6. Copper, J.F., Hochstein, D.H., Seligman, E.B. The limulus test for endotoxin (pyrogen) in radiopharmaceuticals and biologicals. Bull. Parenteral Drug Assoc. 26, 153-162, (1972).
7. Gee, A.P., Sumstad, D., Stanson, J., Watson, P., Proctor, J., Kadidlo, D., Koch, E., Sprague, J., Wood, D., Styers, D., McKenna, D., Gallelli, J., Griffin, D., Read, E.J., Parish, B., Lindblad, R. A multicenter comparison study between the Endosafe® PTS™ rapid-release testing system and traditional methods for detecting endotoxin in cell-therapy products. Cytotherapy. 10,427-35, (2008).
8. Tsuchiya, M. Advantages of archived standard curves for the quantitative bacterial endotoxins test. BioProcess International 7, 152-153, (2009).
9. Zhang, J. Manufacture of mammalian cell biopharmaceuticals. In Manual of Industrial Microbiology and Biotechnology, 3rd ed., R. Baltz et al. (eds.). ASM Press, Washington, DC. Chapter 13. pp. 179-195, (2010).
Author Biographies
Dr. Luis Jimenez currently holds positions of Assistant Professor of Microbiology at Bergen Community College and Adjunct Professor of Biology at Fairleigh Dickinson University, both located in northern New Jersey, United States of America. He is also a consultant for biotechnological applications to environmental, clinical, and industrial problems.
Dr. Jimenez has 17 years of industrial experience in the biotechnology and pharmaceutical industry. He has developed or participated in the development of several products for clinical and environmental applications. Dr. Jimenez’s expertise in these areas is reflected by 52 publications, 8 book chapters, 1 book, 1 patent, and 75 presentations. He completed his Ph.D. in Environmental Microbiology at the University of Puerto Rico and performed his dissertation research under the supervision of Dr. Terry Hazen at the Savannah River Plant in South Carolina under a predoctoral fellowship in Bioengineering and Microbiology from the Department of Energy of the United States and the National Institutes of Health. He completed postdoctoral studies in Environmental Biotechnology at the University of Tennessee, Knoxville under the supervision of Dr. Gary Sayler.