| Author | Message | ||
     Robert A. Forbes |
Fellow Pharmaceutical NIR Users: The USP NIR chapter, now official since August 1, says: "The performance verification standard matches the format of the samples in the current analysis as closely as possible and uses the same background spectrum." It further says: "Quantitative measurement of this external performance verification standard checks both the wavelength calibration and the photometric accuracy." My questions for the community are first, what does this mean? Does this imply a saved background spectrum (file) is used or just use of the same reference material? Second, does this require establishing a quantitative calibration for some reference attribute of the PVS or can qualitative approaches (e.g. wavelength distance method) be used? Finally, and most importantly, what are people using for a PVS? Best Regards, Bob Forbes | ||
| Gary Ritchie |
Hi Rob, I'll let you know what I am doing in my laboratory and well learn together if I am in compliance, hows that? 1. The performance verification standard matches the format of the samples... Means that if your sample presentation is done using a probe, then you must present your standard the same way; if you are using some other sample presentation, you must present your standard the exact same way. 2. ...and uses the same background spectrum, means that if you use a 99% spectralon background which was presented to the instrument the exact same way as you will measure your standard (If you are measuring wavelengths using a wavelength standard or noise or photometric accuracy using the other 4 or 5 standards to obtain those values)then your final background measurement for your standard should be what you report your final PVS for wavelength accuracy usually. I have checked everything against the 99% standard in reflection mode because it turns out that in order to run the carbon doped spectralon standards from 99%, 40%, 20%, 10% and 2% and get your observed values to match the reported standard values, you have to run this series against the 99% standard. 4. We use NIST traceable SRM 1920 for wavelength accuracy and NIST traceable SRM carbon doped spectralon for all of our reflectance work. Transmission is another story and we are still working that one out. I have been running this way since 1995 using the USP specifications with great success. Failures have been detected usually where the slopes for photometric accuracy fail and this has indicated to us that our system was sensitive to sample presentation in one case, which resulted in our hardware being repaired or total detector failure which also resulted in a total replacement. Both of these issues were caught by our due dilagence in running our performance tests on system start up. I have never failed any system yet for wavelength accuracy, but when it has detected a peak failed, it was uuually on the high end, and it usually dissapeared when the system warmed up a little longer. Hope this helps some. Gary Ritchie | ||
| Su-Chin Lo (Suchin) |
Since Bob asked several questions regarding to this new USP NIR measurement in pharmaceutical lab, I also have some questions here: 1. Should the pharmaceutical company need to fully implement the practice in their lab during the routine analysis?? I keep hearing from NIR vendors saying that the Monograph is not finalized yet..... 2. How many standards are required in order to run the "performance verification" test ?? On page 2904 (USP Second Supplement) Frequency of Testing, "The performance verification includes wavelength uncertainty, absorbance-scale linearity and high-light-level noise tests", but on the other section (page 2905-2906 Performance Verification", also indicates it is possible to use "a single external performance verification standard...on a continuing basis". 3, FT instrument related questions: How does the FT-NIR noise criterion generated ?? Says 16 cm-1 resolution in 300 cm-1 segment vs. 10 nm bandwidth in 100 nm segment from grating instrument. Different data points may be generated by two different kind of instrument(grating vs. FT) and does the noise specifications is also working reasonable on the FT instrument ?? Thanks..... Su-Chin Lo | ||
| Gary Ritchie |
Su-Chin, I would be pleased to clear up some of your concerns. 1. I keep hearing from NIR vendors saying that the Monograph is not finalized yet..... I have been applying the USP specifications for all of my NIR instrument performance tests since 1995 when they first appeared, and the specifications have not changed since then, nor will they change in the percieved future until a further revision appears. The current and the proposed revision soon to follow have kept the same specifications. 2. How many standards are required in order to run the "performance verification" test ?? 6. One wavelength verification standard and at least 5 photometric standards. 3. FT instrument related questions: How does the FT-NIR noise criterion generated ?? Says 16 cm-1 resolution in 300 cm-1 segment vs. 10 nm bandwidth in 100 nm segment from grating instrument. The final noise specifications were selected based on a round robin experiment comparing the measurement performance of several instruments from various vendors. As far as I can tell (I am using FT and dispersive instruments in my lab, all specifications are met with both types of instruments. I approximate the range selection for noise tests on the FT but cover the range 1203nm to 2222 nm after wavenumber to wavelength conversion. Yes, I know it's not a linear conversion. The idea here is to establish OQ/PQ protocols for controlling / assessing your instruments performance over time. Once the test is established as close to the tolerances specified in the monograph, you should adhere to these parameters for as long as your instrument is in service generating GMP data. Understand? So the fact that the bandwidths and resolutions are not exactly the same is not an issue. The noise performance on the various instruments vary from very very low to just at the specification limits but not exceeding them. I think that was the general idea behind the round robin study; to ensure that the final settled upon limits do not edge out any particular instrument or vendor. Hope this clears up some of your questions. Gary Ritchie |