September 1, 2017 (Vol. 37, No. 15)
Angelo DePalma Ph.D. Writer GEN
New Sensors Are Enhancing the Monitoring of Diverse Parameters While Reducing the Need for Extractive Sampling
During its formative years, bioprocessing for protein therapeutics borrowed sensor technologies from the medical device industry for rudimentary measurements of dissolved oxygen (DO), carbon dioxide, and a few other parameters of critical importance to cultures of cells and microorganisms.
As the industry and its production base evolved towards greater complexity and operational diversity, the need for pinpoint control over process parameters led to the emergence of purpose-designed sensors for all significant parameters related to culture health and productivity.
Today, biomanufacturers monitor most critical process parameters in real time through inline sensors that feed into control systems and maintain acceptable culture conditions. One of the least well-controlled parameters is the feed rate; optimization of feed rate is essential for culture health, to produce high product titers, and ensure product quality.
Conventionally, feed requirements are estimated through extractive sampling and offline analysis for glucose, lactate, ammonia, product secretion rates, viable cell counts, and cell density. The inputs are many and the analyses time-consuming, which limits their utility for answering, in real time, the question, “to feed or not to feed?”
Continuous sampling has not lived up to its promise due to costs, contamination risk, and in some instances, the loss of substantial process volumes.
Precise Nutrient Needs
Ranger®, a system from Stratophase for automated and adaptive, non-extractive, closed-loop control, takes a novel approach to determining a culture’s feed needs by responding to the cells’ precise nutrient demands. Ranger consists of a steam-in-place (SIP) refractive index (RI) sensor probe that is deployed similarly to a pH or oxygen probe. It works in glass, stainless-steel, or plastic single-use bioreactors, in which the sensor element is an integral part of the vessel wall and is gamma-irradiation-compatible. The sensor connects through a fiber-optic cable to the Ranger manager, which interrogates the sensor and receives signals from it. Ranger relies on measurements of metabolic kinetics, which prompt the Ranger manager to signal addition of feeds. Figure 1 shows a typical setup of the Ranger system.
“Ranger’s feed-on-demand protocol eliminates most of the issues related to conventional feed control,” says Stratophase CEO Simon Saxby. “It does not measure conventional process parameters like pH or DO, but rather, responds to the rate of change in medium composition reflected in the RI.”
It continually interrogates the cells by measuring the effect on a culture caused by additions of feeds or specific feed components, thereby simultaneously determining the optimum feed concentration, composition, and feeding rate required to maintain that concentration. “You can essentially ask a culture a set of questions to determine if a certain change was helpful or not, and obtain an answer in less than an hour,” notes Saxby. That, with the ability to monitor the effects of several different additions or even changes to the wider culture environment (i.e., temperature, pH, or DO) can reduce design-of-experiment (DOE) times and compress process development to less than twelve weeks, according to Saxby.
Robust single-use sensors are available, but are not typically integrated into single-use bioreactor platforms. According to Barbara Paldus, Ph.D., CEO of Finesse Solutions (part of Thermo Fisher Scientific), one issue that has delayed the adoption of single-use sensors is a double standard on the measurement. “Single-use sensors are expected never to drift, whereas their electrochemical counterparts are often standardized once per day. The same issue exists with doing a two-point calibration prior to the process. The expectation is that the single-use sensors will never require calibration. These are both unrealistic expectations.”
Yet single-use sensors for pH, DO, and pressure monitoring in cell culture is on the rise. Single-use sensors are in fact the fastest-growing segment of this market. A 2017 report from Transparency Market Research estimates the 2016 market for these products is at $420 million. With robust 13.4% annual growth, sales of single-use sensors could reach $1.3 billion by 2025.
According to Dr. Paldus, interest is growing in small-scale single-use bioreactors. As a result, vendors are developing small-scale single-use bioreactors with integrated optical sensors for research and process development. “On the upstream side, integrated single-use sensors and intelligent automation are desirable, but price has become a touch-point with customers,” points out Dr. Paldus.
Optical sensors allow a noninvasive and highly precise measurement that complements other sensor types such as electrical sensors for temperature, cell viability (capacitance), or media conductivity. Optical sensors quantify analytes either directly (e.g., Raman) or through use of a substrate (fluorophore for pH or DO). “Their benefits compared with electrochemical sensors for DO is not consuming the substance being measured, and the fact that they do not easily saturate,” Dr. Paldus explains. “They also possess a very high measurement dynamic range, which is useful for determining cell density, where several logarithmic orders can be measured by one sensor, provided that the optics are well designed.”
Finesse Solutions launched the G3Lite+ SmartController for cell culture and fermentation in 2016. G3Lite+ extends the capabilities of the company’s previous controllers to all major brands of single-use bioreactors up to 2,000 L of operating volume, and single-use fermenters (30 L to 300 L). It works with a virtual who’s who of single-use systems, including MilliporeSigma’s Mobius® CellReady, Thermo Scientific™ HyPerformaTM TK SUB, GE Xcellerex XDR, and Pall PadReactor® single-use vessels.
Dr. Paldus notes that universal sensors and controls help bioprocessors achieve the highest utilization of space, capital, labor, and capability. “When only one controller is required, irrespective of the brand, labs or production facilities can be designed for the number of concurrent vessels being run, rather than for all brands of vessels. For contract manufacturers employing a variety of bioreactors, this represents a significant cost savings. Additionally, staff need only learn one user interface and operating system. This is particularly important if certain vendor brands are only used infrequently, for specific processes, and the operators can forget how to use that vendor’s controller.”
Software has become a crucial component at every level of bioprocessing. Process-development laboratories, the testbeds of manufacturing methodologies, have become more sophisticated, enhanced with instrumentation to acquire data that contribute to DOE, sampling, supervisory control, and data analysis. John Poppleton, marketing director at Applikon Biotechnology, notes the software that manages these instruments and functions operates independently, resulting in data fragmentation, superficial data analysis, and inefficient workflows. Applikon’s response has been the Lucullus PIMS (process information management system), a workflow software package that incorporates planning and data analysis for upstream operations (Figure 2).
Capabilities include bioprocess control, automated information extraction, and one-step data analysis in a single piece of software. By focusing on the whole process, Lucullus PIMS supports process analytic technology (PAT) and quality-by-design (QbD) objectives.
“When you walk into a lab, you see all types of analytical tools and controls. But the supervisory software has not kept up,” Poppleton says. “Each datastream operates independently, funneling information into multiple databases and software packages, leading to inefficiencies. We saw a real need to organize these functions into one workflow, our PIMS: a single database that supports data sharing and massively improves workflow.”
The consequence of data fragmentation is lost opportunities. Poppleton relates the experience of one Applikon customer, who managed a busy process-development laboratory. “He found that he could only perform a quarter of the data analysis he would have liked to do. At the end of the day, his ability to define manufacturing-worthy processes was superficial.”
PAT and QbD have been huge drivers for adopting sensors and controls in bioprocessing. Full realization of PAT and QbD benefits is achievable only to the extent that relevant data are captured and subject to analysis. “A number of parameters are available only offline [and] after sampling,” Poppleton says. Overcoming implementation hurdles (i.e., sampling itself, sample prep, analysis, reporting) depends on automating these activities, which is precisely what PIMS facilitates.
Because non-contact sensing options eliminate the issue of product contact, bioprocessors want to give them a closer look. Sonotec has carved out niches in ultrasonic sensors for flow and bubble detection. Like many “biosensor” companies, Sonotec began as a provider of components for medical devices, so its sensor products have a reputation for accuracy and robustness.
The company’s Sonoflow series operates through transit-time ultrasound, which differs from Doppler ultrasound. “Transit-time ultrasound has been used in the medical device industry since the late 1980s, particularly as vascular probes during organ transplantation,” notes Alison Sedler, director of business development at Sonotec. Transit-time devices send an ultrasonic beam upstream and downstream of flow. The signal moves faster with, and slower against, the flow. The integrated difference in velocity, multiplied by the tubing cross-section, is a measure of flow. Unlike Doppler ultrasound, particulate matter is not needed for measurement, and Sonoflow sensors can be used on water.
The significance for bioprocessing is the potential for ultrasonic sensors in single-use equipment. “Coriolis meters work fine with traditional stainless-steel piping,” Sedler explains. “They’re accurate and come in a variety of designs suitable for pipes. But flexible tubing requires a different clamping mechanism.” Because Sonoflow sensors operate on the outside of tubing and have no contact with process fluids, using them entails no cleaning or related validation issues. “When the run is over, you remove the sensor, discard the tubing, and prepare the sensor for the next run by wiping the sensor lumen, if needed.”
Another plus involves scalability—a big issue in biotech. Sonoflow designs handle from 20 mL/min to 200 L/min, so are appropriate for bench, pilot, and full-scale manufacturing. Sonotec also sells inline sensors for low-flow measurement during discovery and process development. With flow accuracy to 0.5 mL/min, these devices are made from polyether-ethylketone (PEEK), which is generally regarded as safe. The inline sensors may be autoclaved or steam-in-place sterilized. The same control software operates both inline and clamp-on sensors. Reusable sensors make sense for R&D purposes; however, says Sedler, “Throwing single-use sensors away can get costly.”
Non-contact sensing for biomass is similarly underutilized. BugLab’s cofounder Martin Debreczeny, Ph.D., notes that in this market, disposable sensors and deployment of sensors on single-use bioreactors are hot-button issues. “Manufacturers are struggling to produce a robust sensor that operates effectively for the entire run and provides reproducible results. We’re still not at the point where we can use the sensor once and throw it away.”
Disposability and reproducibility are not, however, issues for BugLab, because its devices work outside the bioreactor, according to Dr. Debreczeny. “We can measure biomass through a bag, without direct contact. For us, the challenge is obtaining an accurate measurement through either a glass bioreactor wall, or through single-use plastic.” BugLab sells an immersible sensor probe as well.
BugLab biomass sensors work through reflectance—the scatter of light off cells back into the detector. Issues to overcome include the optical properties of the bioreactor material’s composition, thickness, and optical quality. “We got around that by optimizing the sensor geometry, choosing the optimal source/detector separation and wavelength, and assuring the optical collimation is appropriate,” notes Dr. Debreczeny. Collimation refers to narrowing the divergence of a light beam.
State-of-the-Art in Single-Use Sensors
Single-use sensors for measurements such as pressure, temperature, conductivity, and UV absorbance align with other single-use technologies such as bags, tubing, and filters to form complete assemblies. The availability of single-use sensors is necessary to drive implementation of single-use technology from development through to GMP manufacturing.
Single-use sensors have gained widespread acceptance by offering the same ease of use and convenience as other single-use assembly components, while providing the accuracy and robustness of traditional measurement techniques.
To be accepted in the highly regulated biopharmaceutical industry, all single-use sensors must meet the requirements of biocompatibility (USP Class VI), bioburden reduction steps (gamma and/or autoclave compatibility), and manufacture in a clean and controlled environment (Class 7 cleanroom), according to Dennis C. Annarelli, Ph.D., technical and quality manager, PendoTECH. Accuracy is built into single-use sensors through tight manufacturing tolerances and in-process testing, as calibration is not possible once sensors are installed in assemblies.
“Single-use pressure sensors are ubiquitous throughout the bioprocess manufacturing process,” says Dr. Annarelli. “They find use in every step in downstream processing from primary clarification to fill and finish. They are also used extensively to monitor gas pressure in bioreactors. Single-use conductivity sensors are critical for effective concentration/diafiltration in TFF, and single-use UV is used to monitor and control chromatographic separations.”
Future developments in single-use sensors will include expanded offerings to accommodate the wide range of process connections that are needed in the biopharmaceutical industry, he continues. Additionally, qualification testing will be needed to demonstrate applicability in continuous processing where sensors will need to remain accurate and drift-free for up to 60 days in use, he points out.
“Finally, cell- and gene-therapy requirements for single-use sensors will provide unique challenges and opportunities for new types and sizes of sensors for this exciting new therapeutic area,” notes Dr. Annarelli.