408
B. Igne and E. W. Ciurczak
18.4.3 Summary
The variety of applications in drug substance and drug product manufacturing
shows the versatility and usefulness of the technology for the development of robust
processes to deliver patients with safe and efficacious medicines. It also shows the
breadth and depth that a PAT scientist must have to successfully deliver a NIRS
analytical method.
18.5 Raw Material Identification
Raw material identification has been made possible by the flexibility of the hardware,
particularly the handheld systems. In essence, developing a raw material identification method relies on the creation of a relevant spectral library and the determination
of a set of criteria that can determine the class of a new sample. A comprehensive
review on the topic is available [66].
The development of a spectral library is not a simple task. Similar to developing
a quantitative method, the spectral library must include spectra representative of the
variability that will be encountered during the use of the method. If future samples are
expected to vary in physical properties or have a range of chemical properties, they
should be included in the library. Failure to do so will falsely identify valid samples
as not belonging to the expected population, thus requiring the need for investigation
and updating the library and model. While it is never possible to foresee all the sources
of variability that will be encountered, it is recommended to consider various lots
and chemical/physical property ranges to define each class as precisely as possible.
Once a representative library is created, an algorithm is used to discriminate
between classes. A large variety of methods is available [67].
Control over the raw material characteristics has also been investigated for
biopharmaceutical products. The performance of basal powders on process performance and product quality was used to identify classes amongst spectra [68]. The
grouping of the samples was attributed to the blend uniformity, impurities, and heat
sensitivity during the milling process.
18.6 Summary
From the first usage of NIRS in the field to its now prominent position for process
monitoring and control, near-infrared spectroscopy has come a long way in hardware,
software, and modeling. The pharmaceutical industry is now able to rely on the
technology to release medicine to patients without having to sample for off-line
analyses. The sensitivity of NIR light to the matrix, which makes it a tool of choice
for raw material identification and counterfeit detection [69], can be an issue for
B. Igne and E. W. Ciurczak
18.4.3 Summary
The variety of applications in drug substance and drug product manufacturing
shows the versatility and usefulness of the technology for the development of robust
processes to deliver patients with safe and efficacious medicines. It also shows the
breadth and depth that a PAT scientist must have to successfully deliver a NIRS
analytical method.
18.5 Raw Material Identification
Raw material identification has been made possible by the flexibility of the hardware,
particularly the handheld systems. In essence, developing a raw material identification method relies on the creation of a relevant spectral library and the determination
of a set of criteria that can determine the class of a new sample. A comprehensive
review on the topic is available [66].
The development of a spectral library is not a simple task. Similar to developing
a quantitative method, the spectral library must include spectra representative of the
variability that will be encountered during the use of the method. If future samples are
expected to vary in physical properties or have a range of chemical properties, they
should be included in the library. Failure to do so will falsely identify valid samples
as not belonging to the expected population, thus requiring the need for investigation
and updating the library and model. While it is never possible to foresee all the sources
of variability that will be encountered, it is recommended to consider various lots
and chemical/physical property ranges to define each class as precisely as possible.
Once a representative library is created, an algorithm is used to discriminate
between classes. A large variety of methods is available [67].
Control over the raw material characteristics has also been investigated for
biopharmaceutical products. The performance of basal powders on process performance and product quality was used to identify classes amongst spectra [68]. The
grouping of the samples was attributed to the blend uniformity, impurities, and heat
sensitivity during the milling process.
18.6 Summary
From the first usage of NIRS in the field to its now prominent position for process
monitoring and control, near-infrared spectroscopy has come a long way in hardware,
software, and modeling. The pharmaceutical industry is now able to rely on the
technology to release medicine to patients without having to sample for off-line
analyses. The sensitivity of NIR light to the matrix, which makes it a tool of choice
for raw material identification and counterfeit detection [69], can be an issue for