Spectroscopy can be used to analyze process streams for proteins and analytes of interest in real-time. The problem is that many spectroscopic techniques, including NIR and Raman, rely on models to separate signals from background noise.
Developing these models takes time and money, according to Nirrin Technologies CEO, Bryan Hassell, who told GEN, “They are built empirically, from large, designed experiments that span every condition the method will ever encounter. They are expensive to build, tied to the process and instrument they were built on, and to be rebuilt when the process, scale, or instrument changes.”
Pure component analysis
An alternative is the model-free approach “pure-component analysis,” continued Hassell. “In PCA each component has a known spectrum, and the measured spectrum is resolved into the concentrations of those components directly. There is no empirical model to develop and no DOE campaign behind it. A method built in development runs the same way in manufacturing, and on a different system at a different site.”
Hassell pointed to Nirrin’s new tunable laser near-infrared (TL-NIR) platform, which will be showcased at the Bioprocess International (BPI) East conference in Boston. to illustrate the potential benefits of pure-component analysis.
“It is a near-infrared measurement platform built around a tunable laser instead of a broadband lamp. The laser puts far more optical power into each individual wavelength, which gives a much higher signal-to-noise ratio than conventional NIR,” he explained. “Spectroscopic methods return spectra that must be interpreted through a model before anyone can act on them. TL-NIR returns concentrations. That is the number an engineer already makes decisions on, available in seconds or continuously, rather than a signal that needs deciphering first.”
But eliminating models is just one of the potential benefits, according to Hassell, who said “For protein analysis, a TL-NIR measurement is five times faster than variable-pathlength UV, and requires no dilution, no sample prep and no pathlength adjustment. The larger savings are upstream of that. Companies avoid the model-building campaigns that Raman and broadband NIR methods require, and they can bring in-house measurements that are currently sent out to HPLC, which removes both the cost per sample and the days of waiting for results.”
At present, five to 10 “top biopharmaceutical companies” use TL-NIR systems in process development, pointed out Hassell, and said the plan now is to raise awareness about the potential benefits of using the technology in commercial production.
“We are at the beginning of bringing this into manufacturing environments. These processes have been in place for a long time, and the industry will not convert to data-driven manufacturing overnight. We start in process development, where the technology is proven and validated and where transferability is demonstrated, and we are now working with key customers on manufacturing deployment from there.”
