Issue link: http://www.e-digitaleditions.com/i/477622
12 March 2015 Tablets & Capsules which includes a large number of uncorrelated vari- ables—by linear transformation into a new set of uncor- related variables called principal components (PCs). The PCs are structured so that the first few retain most of the spectral variation contained in all the original vari- ables [1]. PCR is required in order to properly distinguish between subtle physical differences of raw materials. In general chapter <1119> "Near Infrared Spectro photo - metry," USP 37 discusses method validation using NIR with respect to qualitative aspects, such as raw material identification. Specifically, it requires presenting both pos- itive and negative challenges to the identification method to demonstrate specificity. A positive challenge entails the analysis of material that was independently received (e.g., the same material from three different lots) but that was not used to create the spectral library itself. Subsequently, this material must pass the test for identification. A nega- tive challenge can also involve materials received on site that are similar to the library compounds in chemical structure, but not identical. These challenge materials should fail the identification test [2]. In-process testing NIR has been used to determine percentage assay of drug in core (uncoated) tablets and in the powder blends used to make drug products. While penetration depths of 2 to 4 mil li meters are possible with NIR (depending on the wave- length), methods typically use the diffuse reflectance mode for tablets and blends contained in glass vials. As with inden- tifying raw materials, NIR requires calibration standards, which are specially prepared tablets or blends that contain the exact amounts of drug or analyte. For example, three cal- ibration standards for an assay method of cores might include drug concentrations of 90 percent, 100 percent, and 105 percent of the label claim. These three standards would be used for determining system suitability before the actual core tablets or blend samples were analyzed. To generate a chemometric model, which is used to cor- relate the NIR spectral response with the amount of drug in tablet cores and blend samples, more extensive reference or "training" sets are required. For example, 10 reference standard blends spanning a range of 70 percent to 115 per- cent of the target drug concentration might be prepared, and a portion might be compressed into tablets. Ideally the tablets used in this calibration set would be compressed at the same hardness. Next, partial least squares or principal component analyses algorithms would be applied to the calibration set, and a mathematical model would be cre- ated that relates the NIR spectra to the concentration of drug in either the tablets or the blends. For core tablets, the sample is typically set directly on the instrument's sapphire window, and at least one repli- cate is performed on each side (left and center photos in next column). Powder blends are placed in a glass vial in an amount sufficient to cover the bottom in a layer of at least 4 to 5 millimeters. Replicates are obtained by shak- ing or spinning the vial between readings. In a more sub- tle use of NIR spectroscopy, Cogdill et al. [3] correlated tablet hardness to changes in NIR spectra. Their study is important because it reinforces the need to establish the training or calibration sets discussed above using carefully controlled physical parameters. Finished-product testing Moisture uptake of capsules. NIR has also been used to monitor the moisture uptake of intact hard gelatin cap- sules [1]. These tests employed a specially designed coni- cal reflector [1, 4] that positions the capsule along the axis of rotation of the conical reflector, thereby maximiz- ing diffuse reflectance (signal) while minimizing specular reflectance (Figure 1). The study exposed capsules to water inside a desiccator for as long as 120 hours, and their weight gain was monitored to determine water uptake. The NIR spectra of the capsules showed a strong correlation to increases in capsule mass due to their expo- sure to moisture (Figure 2). Predicted weight gain was calculated using the PCR values from the NIR spectra against the gravimetric values. Determining the water content of intact gelatin capsules using NIR saves time, effort, and expense over traditional methods of moisture analysis, such as Karl Fischer titration. The diffuse reflectance sample window of an FT NIR analyzer (Antaris II). From left to right: Open cell; tablet placed directly on window; and vial with formulation blend covering the window. Figure 2 NIR spectra of capsules show a strong correlation to increase in capsule mass after exposure to moisture (22°C at 100% RH). Adapted from Cogdill et al. [3]. Predicted weight gain (mg) Actual weight gain (mg) 0 10 20 30 40 50 + + + + + + + + + + + + + + + + + + + + + + + + + + + + 40 30 20 10 0