The nature theory of Chinese medicine (CM) is a core basic theory of traditional Chinese medicine (TCM), and CM chemical fingerprint technology has been widely used in identification and research of CM. Following the previous study results, the UV spectrum data of 61 CMs with clear cold-hot nature, in which 30 CMs are “cold” medicines and 31 CMs are “hot” medicines, were analyzed. Based on the constructed CM fingerprint database, a retrieval scheme is studied to build a predictive identification model used to classify the cold-hot nature of CMs. According to the experimental results, this prediction model is best for the identification of the UV spectrum of petroleum ether solvents; compared with existing classical models, the constructed model has better predictive stability and extrapolation. The proposed prediction model is proven to be effective.

As one of the core theories of traditional Chinese medicine (TCM), the nature theory of Chinese medicine (CM) has been widely explored and studied in recent years. Under this theory, CMs are divided into four characteristics—cool, cold, hot, and warm—and the cold-hot nature is important (1,2). Treating a “hot” syndrome with a medicine with a “cold” nature, or treating a “cold” syndrome with a medicine with a “hot” nature illustrates that cold-hot medicine nature theory is an important basis for TCM treatment in recovering the balance between so-called yin and yang in the human body, and is of great significance for guiding the clinical practice of TCM (3).

The study of CM cold-hot nature has attracted a lot of attention recently. Jin and associates (4) constructed a “three-element” mathematical analysis model to study cold-hot medicine nature and to explore the biological performance of Chinese materia medica. Zhao and associates (5) introduced a novel cold/hot plate differentiating assay method to study the cold and heat properties of traditional CMs. Wan and colleagues (6) compared the intervention effects of four CMs with a cold or hot property on body temperature and temperature-sensitive transient receptor potential ion channel proteins of rats with yeast-induced fever. Liang and colleagues (7) integrated bioinformatics analysis and chemical structure analysis methods to identify the biological activity of Chinese herbal medicine properties. Wang and associates (8) utilized a self-organizing map to classify the cold-hot nature of CMs based on their constituent compounds. Fu and colleagues (9) investigated the presence of anticancer activity displayed by the cold-hot nature of traditional Chinese marine medicine with phylogenetic tree analysis. Fu and associates (10) explored bioinformatics methods to understand the cold, hot, and neutral nature of CMs.

The main basis for determining the cold or hot nature of any given medicine is the bioactivity of the CM and the material composition within the CM that determines its bioactivity. Therefore, material composition is the basis for the production of the cold or hot nature of a CM. At present, research on cold-hot nature is focused on the interrelationship between cold-hot medicine nature and material composition within CMs. Methodologies involving the construction of CM nature discriminant models, such as chemical fingerprinting techniques, are often used to analyze the relationship between material composition and prediction of the cold-hot nature of CMs.

CM cold-hot nature discrimination with artificial intelligence first utilizes chemical fingerprint technology to determine the original effects of a CM or the metabolomics method to represent the features of a CM, and then constructs an artificial intelligence prediction model to identify the cold or hot nature of CMs for which the cold or hot nature is unknown. This is of great significance to the development of new resources for TCM. Li and associates (11) analyzed 1725 types of CMs and represented the features of CMs with 150 original efficacy characteristics to build an artificial neural network (ANN) classification model to predict the cold or hot nature of CMs with an unknown cold or hot nature. Subsequently, Liu and associates (12) and Zhang and colleagues (13) explored partial least squares and principal component analysis-linear discriminant analysis models to identify the cold or hot nature of CMs. Nie and associates (14) utilized metabolomics data to characterize the features of CMs, and obtained better forecasting results in predicting the cold-hot nature with a random forest model. Long and colleagues (15) built a combination system for predicting the cold or hot nature of CMs that analyzed 284 CMs, and used a support vector machine (SVM) as a predictive model. Li and associates (16) combined nuclear magnetic resonance spectroscopy of proton 

H-NMR with pattern recognition techniques for cold-hot nature classification of Chinese medicinal herbs.

Thus, as we see, there are many studies that aim to identify or predict the cold or hot nature of CMs. However, most of these studies have adopted the classical and general artificial intelligence algorithms instead of building a proprietary prediction model for CM characteristic data, leading to insufficient forecasting results. In this study, the interrelationship between cold-hot medicine nature and the material composition of CMs is explored. An artificial intelligence prediction model that conforms to the characteristics of CM data is constructed to identify the cold or hot nature of CMs for which the cold or hot nature is unknown. First, UV fingerprint similarity is studied to pro- pose a retrieval scheme. Second, a predictive identification model is built with the proposed retrieval scheme. Finally, a number of experiments are used to verify the feasibility of the predictive model.

The UV spectral data of this study come from the 973 program ”Research of Basic Theories of CM,” The project selected 61 representative CMs with clear cold-hot nature, in which 30 CMs are “cold” medicines and 31 CMs are “hot” medicines. For example, 

 rhizome are hot medicines. Their natures are determined by referring to the Chinese Materia Medica or Shen Nong’s Herbal Classic. The project measured the absorbance of all 61 CMs at the wavelength range of 190–400 nm. UV spectral data were measured with four different solvents (chloroform, distilled water, absolute ethanol, and petroleum ether). Figure 1 shows the UV absorption spectra of 

Correlation similarity of random variables is usually represented by the correlation co-efficient; this includes Pearson’s correlation and Spearman’s correlation. Pearson’s correlation has been widely used in correlation calculations of functional connectivity of brain regions and similarity measurement of chemical fingerprints, such as spectra and chromatograms (17,18). This study intends to use Pearson’s correlation to calculate the similarity of the UV absorption spectra of different CMs. Higher values of Pearson’s correlation coefficient (close to 1.0) indicate that the corresponding material composition of CMs might exhibit higher similarity, which also implies that it may be the same for the cold-hot nature of CMs.

Pearson’s correlation is the most commonly used statistical calculation method that reflects the linear correlation between two variables. Correlation coefficient is used to measure the correlation between two variables, the well-known formula is as follows:

 represent the mean of the variables; cov(x,y) indicates the covariance of the two variables; and σ

, respectively. It can be seen from the formula that the correlation coefficient is the quotient of covariance and standard deviation of the two variables.

A Retrieval Scheme (RS) for Cold-Hot Medicine Nature Identification

By using the Pearson‘s correlation coefficient (PCC) to measure the similarity of UV spectra, a retrieval scheme based on the similarity measurement is proposed for cold-hot medicine nature prediction. For a CM where we do not know its cold-hot nature, we first measure the absorption degree of UV spectrum, and then calculate the similarity of the UV spectrum between this query CM and the CMs with a clear cold-hot nature in the database. The obtained PCCs are arranged from small to large to search for the “most similar” reference CMs in the data set. The K “most similar” CMs correspond to the reference CMs, with largest PCCs to the query CM. A cold nature probability value is calculated to measure the degree of cold nature of the query CM—the ratio of the sum of the PCCs of cold nature medicines to the sum of the PCCs of the 

 “most similar” CMs. The formula is as follows (with C being the number of cold nature medicines and H being the number of hot nature medicines):

With this retrieval scheme, when giving a threshold of 0.5, if is above 0.5, we classify this queried CM as cold; otherwise, it is deemed to be classified as hot.

To demonstrate the effectiveness and feasibility of the abovementioned methods, extensive experiments were constructed to assess the prediction performance of the retrieval scheme in terms of stability accuracy and extrapolation accuracy for our algorithm and state-of-the-art prediction models, including ANN (19), extreme learning machine (ELM) (20), and SVM (21). ANN attempts to mimic the neural network of the human brain from the perspective of information processing, to establish a simple model that forms different networks according to different nodes and connections. The ELM algorithm is developed from the ANN algorithm; and in the classification process, ELM can project data from multidimensional space to label space, classifying the samples with similar descriptors into the same class. SVM is a binary classification model; the basic idea of SVM learning is to solve the separation hyperplane that divides the training data set correctly, and has the largest geometric interval. The experiments of performance assessment are in the context of the CM data set. The application allows a researcher to examine a CM substance with an unknown cold-hot nature by retrieving and studying similar UV spectra of CMs with a clear cold-hot nature before determining the true nature of the substance being classified. We first compared the medicine nature identification of UV spectra with different solvents, and then evaluated the model’s performance, including stability evaluation and extrapolation evaluation and comparing them with other prediction models.

There are two metrics used in our experiments to assess the performance of the proposed prediction model. The first metric, stability evaluation, is computed by the leave-one-out method (21) in the entire data set. The detailed steps are as follows: First, for the UV spectral data of 61 CMs in the database, one CM is selected as the testing data set, and the remaining 60 CMs in the data set serve as the training data set. Next, the top 

 most similar CMs are retrieved by computing UV spectra similarity between the query CM and each of the reference CMs in the training data set. Then, the cold nature probability of the query CM is calculated. The above process is repeated until the cold nature probability value of each CM is calculated. As a result, 61 cold nature probabilistic values are obtained. Stability evaluation applies the receiver operating characteristic (ROC) curve and prediction accuracy (ACC) to evaluate the proposed prediction model. The ROC curve is generated by varying the threshold of the cold nature probability. The area under the ROC curve (AUC) is computed to evaluate the prediction model. The larger the area, the more stable the model. The result of ACC illustrates the probability of correctly classified cold-hot nature of CMs. The formula of ACC is as follows:

th CMs after sorting the cold probability from small to large.

The second metric, extrapolation evaluation, indicates the extent to which cold nature CMs can be detected on basis of the CMs that are retrieved by RS. In these experiments, 61 CMs are randomly divided into the training set and the testing set. The raining set contains 40 CMs, with cold nature CMs and hot nature CMs each making up an equal portion. The remaining 21 CMs are used for testing. With the testing data, we search for the most similar 

 CMs from the training set, and calculate the cold nature probability values. These probabilities are applied to calculate the ROC curve and ACC value that are used for extrapolation evaluation. The experiments are repeated 10 times with randomly selected training data sets, and the experimental results are taken as the mean and variance of 10 experiments.

Performance Comparison with Different Solvents

As a fingerprint of CM, UV spectroscopy can identify complex substances in CMs, and be used to predict CM nature. This helps to quantitatively study the relationship between substance composition and medicine property.

In this study, the UV spectra of different solvents (chloroform, distilled water, absolute ethanol, and petroleum ether) were substituted into the constructed prediction model to analyze the influence of UV spectroscopy on the identification of medicine nature with different solvents. Spearman’s correlation of UV spectra with petroleum ether solvent is calculated as a comparative reference (denoted as “Spearman’s + PE”). Figure 2 shows the AUC curves for the medicine nature identification of the UV spectra with four solvents and Spearman’s + PE. The AUC value is calculated as a function of the number of retrieved similar reference CMs (

), and thus obtains a more comprehensive curve for the performance of the predictive model. From Figure 2, the UV spectra of petroleum ether solvent outperforms that of other three solvents for classification of cold-hot medicine nature, meaning that UV spectra of petroleum ether solvent can better identify CM nature. When 

 = 7, AUC value of petroleum ether solvent reaches the maximum value of 0.834. CM nature identification with Spearman’s + PE is inferior to that with Pearson’s correlation and petroleum ether solvent when 

 < 15, but outperforms that with other solvents. This illustrates Pearson’s correlation is more applicable to measure the similarity of UV fingerprints with petroleum ether. CM nature identification rates of absolute ethanol and distilled water solvents are poor; maximum AUC values are only 0.636 and 0.666, respectively. Therefore, UV spectra of absolute ethanol and distilled water solvents cannot be used as characteristic data alone to identify medicine nature. Table I displays the ACC values when AUC values of four solvents are the maximum, respectively. For UV spectra of petroleum ether solvent, the maximum AUC and ACC value are 0.834 and 0.754, respectively. However, model prediction accuracies of other three solvents or Spearman’s are all lower than 0.7, meaning that they are poor for identifying CM nature.

To comprehensively elaborate and verify the feasibility and stability of the proposed prediction model for identifying the CM nature, this study compares the prediction performance of our proposed model (the retrieval scheme, denoted as “RS”) with that of classical classifiers (SVM, ANN, ELM), or classifiers that have been reported in literatures to predict the CM nature. According to the results of the previous section, UV spectral data with petroleum ether solvent are utilized as the research object. Table II shows the stability evaluation comparison for RS and other three comparative algorithms for CM nature recognition. First, ANN and ELM are poor for classifying cold-hot medicine nature with UV spectral data. Second, prediction performance of SVM is better than that of ANN and ELM. Finally, the stability and recognition performances of the proposed model RS outperform that of the classical prediction models.

For extrapolation evaluation, Table III shows the performance comparison of RS and the baseline methods. According to Table III, the conclusion of extrapolation experiments is consistent with the conclusion of stable experiments. First, ANN and ELM are poor for identifying cold-hot medicine nature with UV spectral data. Second, prediction performance of SVM is higher than that of ANN and ELM, but lower than RS. Finally, the extrapolation performance of RS outperforms that of the classical prediction models.

Table IV displays statistically significant difference of AUC values between RS and other comparative algorithms at the 5% significance level. An unpaired t-test is applied to calculate the 

 value. According to the table, we can see the 

-values between RS and ANN; ELM are lower than 0.01, meaning that they have significant statistical difference. The 

-values between RS and SVM is 0.012, meaning that they have a statistically significant difference. In general, the data analysis illustrates that the extrapolation evaluation between RS and other comparative algorithms has a statistically significant difference.

Extrapolation evaluation is used to provide prediction examples. Two retrieval sets returned by RS are showed in Table V. The query CM (first row) and their top 

 = 7 retrieved reference CMs are calculated by PCC. Perfect retrieval reference CMs would result in a ranked order of CMs with a monotonically decreasing PCC metric. In the first column, for the query hot nature CM Fructus Piperis Alba, seven retrieved reference CMs are all hot nature. Its cold nature probability is 0, indicating the query CM Fructus Piperis Alba is more likely to be hot nature. In the second column, for the query cold nature CM Dianthi Herba, six retrieved CMs are cold nature and one is hot nature; its calculated cold nature probability is 0.8586, indicating the query CM Dianthi Herba is more likely to be cold nature. According to the prediction examples, it can be inferred that the similarity of UV spectra can characterize the similarity of cold-hot medicine nature. All CM retrieval results are provided in Table V, according to the leave-one-out method.

We evaluated the total prediction performance of RS using a leave-one-CM-out method with a threshold of 0.5. The prediction of ROC curve is shown in Figure 3. The AUC value is 83.4%. The total classification accuracy is 75.4% (46/61). Both kinds of the medicines obtained approximately 75% identification rates; 23.3% (7/30) of cold nature medicines are misclassified as hot nature medicines, whereas 25.8% (8/31) of hot nature medicines are misclassified as cold nature class. Overall, the results illustrate relatively good prediction performance for a complex classification problem.

This study investigates the feasibility of identifying CM nature with a retrieval scheme based on the similarity of UV spectral data. We demonstrate that it is a reasonable method to determine the cold-hot medicine nature according to medicine nature of the similar CMs. This study has many unique characteristics and experimental observations. First, to realize CM nature prediction, a 61 reference CM UV spectral data set is assembled in which each of the CM is clearly identified as cold or hot nature. The conclusion is that it is feasible to discriminate CM nature.

Second, cold-hot nature is an important part of CM nature theory. In this study, we explore the interrelationship between cold-hot medicine nature and material composition within CM. The UV spectra are used to represent material composition. Experiment results have demonstrated that material composition within CM is related to cold-hot medicine nature, which can be used for medicine nature-type identification and classification. Meanwhile, it can be inferred that material composition is the basis for the manifestation of CM nature.

Third, according to UV spectral data characteristics of CM, we designed a retrieval scheme for predicting cold-hot nature of CM. The PCC value is applied to measure the similarity of UV spectra. Experimental results show that our specific model RS is superior to existing classical models. The potential explanations are as follows: i) The RS model is a classifier designed specifically for UV spectral data of traditional Chinese medicine; ii) On the basis of the theory that material composition is the basis for the manifestation of CM nature, RS fully mines the relationship between material composition and CM nature with high accuracy.

Fourth, in this study, the robustness of the proposed RS model needs to be demonstrated by independent studies to support future clinical applications. For an intelligent forecasting system, the ultimate goal is to aid researchers in reading and interpreting UV spectral data. The system is not sufficiently effective if it has low robustness for an independent testing data set. The proposed RS model has a high robustness if the UV spectral data of other CMs have the identical wavelength range. In the future, other CMs will be collected to verify the robustness of the model.

Despite the advantages discussed above, there are some limitations to our research. First, in this study, UV spectra is used to study material composition within CM. However, infrared (IR) spectra and liquid chromatography technologies can also be used to study material composition within CM. Medicine nature prediction methods with these fingerprint data are one of the emphases for study in the future. Second, we explore the similarity of UV spectra to predict the cold-hot nature of CM. The fingerprint data of CM is high dimensional and small sample data. The study of predictive models that fit the characteristics of such fingerprint data is the focus of follow-up research. Third, since this paper focuses on the similarity of UV spectra, UV spectrum features are not studied in depth. Consequently, fusion with other effective spectrum features will be explored to further improve the medicine nature discrimination performance. Finally, the identification of CM nature based on fingerprints is very complicated. In this study, 61 CMs are extracted UV spectra. However, it is not enough for training a discrimination model. Extraction of more CM UV spectral data is also planned.

In this study, a cold-hot medicine nature predictive model that conforms to the characteristics of TCM is proposed. Unlike the other existing classical classification models, this identification model is based on the retrieval scheme, discriminating the unknown CM nature with the retrieved similar nature CMs. The experimental results show that the model can better predict the cold-hot nature of CMs based on the UV spectral data.

The authors declare no competing financial interest.

The research is supported by the national key basic research development program (973 Program) (No. 2007CB512601); National Natural Science Foundation of China (No. 81473369); Key research and development plan of Shandong province (No. 2016CYJS08A01-1).

(4) R. Jin, B. Zhang, X.-Q. Liu, S.-M. Liu, X. Liu, L.-Z. Li, Q. Zhang, and C.-M. Xue, 

(5) Y. Zhao, L. Jia, J. Wang, W. Zou, H. Yang, and X. Xiao, 

(6) H.-Y. Wan, X.-Y. Kong, X.-M. Li, H.-W. Zhu, X.-H. Su, and N Lin, 

(7) F. Liang, L. Li, M. Wang, X. Niu, J. Zhan, X. He, C. Yu, M. Jiang, and A. Lu, 

(8) M. Wang, L. Li, C. Yu, A. Yan, Z. Zhao, G. Zhang, M. Jiang, A. Lu, and J. Gas- teiger, 

(9) X. Fu, X. Song, X. Li, K.K. Wong, J. Li, F. Zhang, C. Wang, and Z. Wang, 

(10) X. Fu, L.H. Mervin, X. Li, H. Yu, and J. Li, 

(11) Y. Li, X. Li, F.-Z. Xue, and Y.-X. Liu, 

(12) W.-H., Liu, Y. Li, Y.-J. Ji, P. Wang, Y.-Q. Zhang, and F.-Z. Xue, 

(13) X.-X. Zhang, Y. Li, Y.-J. Ji, P. Wang, Y.-Q. Zhang, and F.-Z. Xue, 

(14) B. Nie, Z.-L. Hao, B. Gui, Z. Wang, J.-Q. Du, G.-L. Wang, and X. Zhang, 

(15) W. Long, P. Liu, J. Xiang, X. Pi, J. Zhang, and Z. Zou, 

(16) H. Li, Q. Xu, J., Zhang, and H. Xu, 

2017 International Conference on Medical Science and Human Health

(17) L. Geerligs, C. Can, and R.N. Henson, 

(18) J.H. Christensen, J. Mortensen, A.B. Hansen, and O. Andersena, 

(19) L. Tao, T. Runtao, Y. Xinlan, L. Sun, Y. He, P. Xie, S. Ma, 

(20) E. Malar, A. Kandaswamy, D. Chakravarthy, and A.G. Dharan, 

(21) G. Wei, H. Ma, W. Qian, and M. Qiu, 

 are with the Shandong University of Traditional Chinese Medicine, in Jinan, China. Direct correspondence to: 

Articles

Peer-Reviewed Research

A model has been developed to predict the “cold” or “hot” nature of Chinese medicines based on UV spectral data.

Cold-Hot Nature Identification of Chinese Medicine Based on an Ultraviolet Chemical Fingerprint

The first part of this article focused on the fundamental principles of a scanning quadrupole and how measurement protocol can be optimized based on the data quality objectives of the analysis and in particular the compromise between limit of detection and sample throughput. Part 2 looks at different ways to assess detection capability when processing continuous signals generated by a conventional sample introduction system, and also transient signals produced by different types of sampling methods such as chromatographic separation techniques and nanoparticle size studies.

Detection capability is a term used to assess the overall detection performance of an inductively coupled plasma (ICP)–mass spectrometry (MS) instrument. There are a number of different ways of looking at detection capability, including instrument detection limit (IDL), limit of quantitation (LOQ), elemental sensitivity, background signal, method detection limit, and background equivalent concentration. Of these four criteria, the IDL is generally thought to be the most accurate way of assessing instrument detection capability, but it’s important to emphasize that it is not considered an assessment of real-world limit of quantitation. It is often referred to as signal-to-background noise, and for a 99% confidence level is typically defined as 3× standard deviation (SD) of n replicates (

 = ~10) of the sample blank and is calculated in the following manner:

However, there are slight variations of both the definition and calculation of IDLs, so it is important to understand how they are being calculated if a comparison is to be made. So, for example, are they being run in single-element mode, using extremely long integration times (5–10 s) to achieve the highest-quality data, or in multielement mode using multielement standards? Whatever measurement protocol is used, it is generally acknowledged that the achievable LOQ of an analysis is approximated by multiplying the instrument detection limit by at least a factor of 10 and with some difficult samples, by a factor of 100. So, when comparing instrument detection capability, it is important to know and understand what measurement regime is being used.

A more realistic way of calculating analyte detection limit performance in your sample matrices is to use the method detection limit (MDL). The MDL is broadly defined as the minimum concentration of analyte that can be determined from zero with 99% confidence. MDLs are calculated in a similar manner to IDLs, except that the test solution is taken through the entire sample preparation procedure before the analyte concentration is measured multiple times. This difference between MDL and IDL is exemplified in EPA Method 200.8 where a sample solution at 2–5 times the estimated IDL is taken through all the preparation steps and analyzed. The MDL is then calculated in the following manner:

” value for a 95% confidence level and specifies a standard deviation estimate with 

Both IDL and MDL are very useful in understanding the capability of an ICP- MS system. However, regardless of the method used to compare instrument detection limits, it is essential to carry out the test using realistic measurement times that reflect your analytical situation. For example, if you are determining a group of elements across the mass range in a digested rock sample, it is important to know how much the sample matrix suppresses analyte sensitivity, because the detection limit of each analyte will be impacted by the amount of suppression across the mass range. On the other hand, if you are carrying out high-throughput multielement analysis of drinking or wastewater samples, you probably need to be using relatively short integration times (1–2 s per analyte) to achieve the desired sample throughput. Or if you are dealing with a laser ablation transient peak that perhaps lasts just a few seconds, it is important to understand the impact that the time has on detection limits compared to a continuous signal generated with a conventional nebulizer. In other words, when comparing detection limits, it is critical that the tests rep- resent your real-world analytical situation.

Elemental sensitivity is also a useful assessment of instrument performance, but it should be viewed with caution. It is usually a measurement of background-corrected intensity at a defined mass, and is typically specified as counts per second (cps) per concentration (ppb or ppm). However, unlike detection limit, raw intensity usually does not tell you anything about the intensity of the background or the level of the background noise. It should be emphasized that instrument sensitivity can be enhanced by optimization of operating parameters such as RF power, nebulizer gas flows, torch-sampling position, interface pressure, and sampler or skimmer cone geometry, but usually comes at the expense of other performance criteria, including oxide levels, matrix tolerance, or background intensity. So, be very cautious when you see an extremely high sensitivity specification, because it could be that the instrument has been optimized for sensitivity, and, as a result, the oxide or background signal might also be high. For this reason, it is unlikely there will be an improvement in detection limit unless the increase in sensitivity comes with no compromise in the level of the background.

It is also important to understand the difference between background and background noise when comparing specifications (the 

 is a measure of the stability of the background and is defined as the square root of the background signal). Most modern quadrupole instruments today are capable of 100–200 million cps per ppm of a mid-mass isotope such as rhodium (

) and <1–2 cps at a background mass where there are little or no spectral features (usually at 220 amu), However, some commercial instruments offer what is called a high sensitivity option with sensitivity in the order of 1 billion cps/ppm. This is a very attractive feature for many application areas, particularly when characterizing small inclusions on the surface of a geological mineral using laser ablation. However, when solution nebulization is being carried out, the additional sources of noise related to the sample introduction system (mentioned earlier) could have a negative impact on the background and the increased sensitivity might not produce a significant improvement in the detection capability.

Background Equivalent Concentration (BEC)

Another useful figure of merit that is being used more routinely nowadays is background equivalent concentration (BEC), defined as the intensity of the background at the analyte mass, expressed as an apparent concentration and is typically calculated in the following manner:

BEC is considered a more realistic assessment of instrument performance in complex sample matrices (especially if the analyte mass sits on a high continuum or background feature), because it gives an indication of the level of the background, defined as a concentration value. Detection limits alone can sometimes be misleading because they are influenced by the number of readings taken, integration time, cleanliness of the blank, and at what mass the background is measured—and are rarely achievable in a real-world situation. Figure 1 emphasizes the difference between detection limit and BEC. In this example, 1 ppb of an analyte produces a signal of 10,000 cps and a background of 1000 cps. Based on the calculations defined earlier, the BEC is equal to 0.11 ppb, because it is expressing the background intensity as a concentration value. On the other hand, the detection limit is 10 times lower because it is using the SD of the background (or the noise) in the calculation. For this reason, BECs are particularly useful when it comes to comparing the detection capabilities of techniques such as cool or cold plasma and collision–reaction cell interface technology, because it gives a very good indication of how well the polyatomic spectral interference has been reduced.

Peak measurement protocol wouldn’t be complete without mentioning that some applications involve a signal that is transient in nature. Most ICP-MS applications involve a continuous signal generated by a sample that is being aspirated into the plasma by a conventional pneumatic nebulizer and spray chamber. However, for some analyses, the sample is being delivered over a finite period of time dictated by the sampling system. Common ICP-MS applications that generate transient signals include:

electrothermal vaporization (ETV) of small volume samples

flow injection analysis of samples with high dissolved solids

high performance liquid chromatography (HPLC) separation systems for speciation studies.

Depending on the sampling technology being used, the duration of the transient event can be anything from a few milliseconds (nanoparticle studies) to up to 10 min (HPLC separations). For that reason, it is critical to optimize the measurement time to achieve the best multielement S/N ratio in the sampling time available. This is demonstrated in Figure 2, which shows the temporal separation of a group of elemental species in a typical chromatographic transient peak. The plot represents signal intensity against mass over the time period of the chromatogram. When carrying out elemental speciation studies using ICP-MS, chromatography transient peaks typically last a few minutes, so if 3–4 species of the same element are being measured in duplicate, as is fairly common when characterizing different species of arsenic in water samples, there has to be a realistic understanding of the real-world LOQs, based on the available measurement time. And the more elements being separated the more challenging the application. So, to get the best detection limits for this group of multielement species exemplified in Figure 2, it is very important to spend all the available time quantifying the analyte peaks of interest (1).

This problem is even more challenging when the transient peak is lasting only a few milliseconds such as signals generated by metal-based nanoparticles studies. Nanoparticles, which are defined as particles of less than 100 nm diameter, are now being used in a variety of products, including scratch-proof eyeglasses, crack-resistant paints, transparent sunscreens, deodorants, stain-repellent fabrics, and ceramic coatings for solar cells using a variety of metals such as gold, silver, titanium, antimony, iron, and silicon. Unfortunately, they can get into the environment from water treatment plants, effluents, and industrial wastes and have a negative impact on surrounding ecosystem, affecting the health of animals and plants.

So instead of dealing with a continuous signal generated by metallic ions in solution, the nanoparticles generate a rapid transient signal, as shown in Figure 3. Figure 3a shows a constant signal generated by positively charged ions in solution. Because the signal is continuous it does not matter when the measurement dwell time is initiated to measure and quantify the analyte mass. However, Figure 3b shows a small number of metal nanoparticles suspended in a sample solution, which will generate pulses of charged ions. So the timing of the measurement is absolutely critical, because the individual pulses could be missed if the dwell time isn’t initiated at the right time. Depending on the size of the nanoparticles, the transient event often last less than a millisecond. So, in order to fully characterize the peak, dwell times in the order of 50–100 microseconds are required, with settling times as short as possible. In fact, when attempting to quantify the size of the nanoparticles from the number and intensity of the pulses, no settling time is the preferred option to maximize the measurement duty cycle (2).

So, to successfully carry out an analysis that generates a transient peak using a scanning quadrupole-based ICP-MS instrument, a great deal of flexibility is required to allow very short dwell and settling times to maximize the quality of the generated data. For this reason, the quadrupole scanning or settling time and the time spent measuring the analyte peaks must be optimized to achieve the highest signal quality, which basically involves optimizing the number of sweeps, selecting the best dwell time, and using short settling times to achieve the highest measurement duty cycle and maximize the peak S/N ratio over the duration of the transient event. If this is not done, there is a high likelihood that the quality of the generated data could be compromised. In other words, the data quality objectives of the analysis would need to be reassessed, because it is unrealistic to expect to carry out multielement or species quantitation at detection limit levels when dealing with such a short transient signal.

Although the cost of an ICP-MS system has dropped considerably over the past 38 years, it is still considered a fairly complex technique, requiring a very skilled person to operate it. Manufacturers of ICP-MS equipment are constantly striving to make the instrumentation simpler to operate, offer better interference reduction capabilities, design easy-to-use-software, and configure hardware components that are less complicated to clean and maintain. You have only got to look at the newer triple and multi-quadrupole instruments on the market today to realize that they all do an excellent job of maximizing the capabilities of the technique. However, even with all these enhanced problem-solving tools, it is still not a technique you can place in the hands of a complete novice and expect to get high quality data, particularly when it comes to analyzing difficult matrices.

For these reasons, the ICP-MS marketplace appears to be separating into two distinct categories. One is for more advanced users, which includes high performance techniques to reduce or minimize interferences and enhance detection capabilities in complex samples. The second category is targeted at more routine, high throughput applications. These instruments will be relatively simple to operate and easy to maintain and include built-in, standard operating methods or procedures to ensure that the desired application methodology is “bullet-proof” for the less-experienced users.

As the price for this “routine” category is slowly being eroded, it will most definitely attract more application areas to the technique, and as a result novice ICP-MS practitioners will be called upon to run the instrumentation. The challenge with opening up the technique to new application areas, such as those found in the pharmaceutical and cannabis industries, is that many of these operators will have very little experience developing ICP-MS methods. This makes it all the more reason why there is still a need for information in the public domain about the fundamental principles of how the technique works, its strengths and weaknesses, and how it’s applied to solve real-world analytical problems. This concern was the major incentive for writing this column on optimizing measurement protocol as well as giving a talk at a 

 magazine virtual symposium focusing on trace element detection limits and how they are impacted by the demands of the application (3). My goal is to educate and help ICP-MS practitioners with limited knowledge of the technique to better understand its analytical capabilities, and in particular, the balance between maximizing productivity and optimizing detection capability. I hope I have achieved that goal.

(1) K.R. Neubauer, P.A. Perrone, W. Reuter, and R. Thomas, 

Trace Element Analysis in the Real World: Practical Considerations Driving ICP-MS Detection Performance

, “Symposium on Trace Element Detection Limits: What Every Spectroscopist Should Know,” Spectroscopy Magazine/Milestone Inc. webcast, September 23, 2021. http://www.spectroscopyonline.com/spec_d/trace.

Column: Atomic Perspectives

How can you assess real-world detection capability of an ICP-MS system?

Impact of Measurement Protocol on ICP-MS Data Quality Objectives: Part II

The new PIC/S data guidance document PI-041 was published in July 2021. This column is the second part of a review looking at the expectations for computerized systems, outsourcing, regulatory actions, and remediation of data integrity failures. Spoiler alert: You won’t like this column if you use hybrid systems.

The final version of the PIC/S guidance on Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments with a document number PI-041 (1) was published in July of 2021. The document scope covers all activities under the remit of both Good Manufacturing Practice (GMP) and Good Distribution Practice (GDP) regulations. This column will only focus on laboratories working to GMP.

The last “Focus on Quality” column reviewed the first part of the document covering scope, data governance, organizational influences on data integrity (DI) management, general DI principles, and specific DI considerations for paper records (2). In this second part of the review, we focus on the requirements for computerized systems and take a high-level look at outsourcing, regulatory actions, and remediation of data integrity failures.

Specific DI Considerations for Computerized Systems

Section 9 of PI-041 is focused on specific considerations for ensuring data integrity for computerized systems (1). Section 9 is probably the most pertinent section for Spectroscopy readers because the majority of spectrometers have a workstation attached with application software to control the instrument and acquire, process, report, and store spectroscopic data. PI-041 section 9 consists of 10 subsections, as shown in Figure 1. This column covers the key sections from this guidance, starting with the Pharmaceutical Quality System, and the role it should (but often does not) play in controlling computerized systems.

Role of the PQS in Controlling Computerized Systems

To ensure overarching control of computerized systems and the processes automated by them, the Pharmaceutical Quality System (PQS) needs to ensure the following:

Evaluation and control of computerized systems to meet GMP regulations (essentially Chapter 4 and Annex 11 [3, 4])

Knowledge of the data integrity risks and vulnerabilities of each system and the ability to manage them.

Knowledge of the impact each system has on product quality and the criticality of the system

Ensuring that data generated by a system is protected from accidental and deliberate manipulation

An assessment that suppliers of computerized systems have an ad- equate understanding of applicable regulations, and that understanding is reflected in the applications that they market

An assessment of the criticality and risks to the data and metadata generated by each system. Data process mapping is one way to understand how the system operates, the meta-data used to acquire and manipulate data, how the data are stored, and the access allowed to each user role.

These areas are discussed in more detail in the remaining subsections in this part of the PI-041 guidance.

Data Integrity and Compliance Starts with System Purchase

Do you have a death wish to keep perpetuating data integrity problems? If you are a DI team member, a CSV expert, or a consultant, you may have answered “Yes.“ However, if you work in the laboratory and must trudge through miles of procedural controls to ensure data integrity, the answer should be a resounding “No.” See the discussion on the hidden factory in your laboratory for a detailed description of the pain that paper can cause (5). But that lovely shiny new spectrometer that has just been delivered has arrived to resolve problems has the same old issues you have with the current one—a lack of data integrity controls. If you want to make money out of data integrity, buy shares in paper companies. You focused on the instrument and not the software during the selection process, you naughty people!

To avoid perpetuating data integrity problems, it is essential that any new analytical instruments and computerized systems have adequate technical controls in the software to protect the electronic records. Clause 9.1.4 of PI-041 notes that laboratories must ensure that suppliers have an adequate understanding of regulations and data integrity requirements, and, as noted in 9.3 item 1, specific attention should be paid to the evaluation of data integrity controls (1). Purchase of inadequate instruments is still a problem, as shown by Smith and McDowall, who analyzed over 100 FDA citations for infrared instruments. Over 40% of the citations were because of lack of software controls present before the instrument was purchased, such as no security, conflicts of interest, or no audit trail (6). Although there is the regulatory expectation that one should not purchase inadequate systems, as laboratories focus on the instrument and not the software data, integrity problems will be perpetuated.

Don’t forget how the Purchasing department can screw up data integrity. You may have spent time and effort in the evaluation of a specific instrument, and the software and the capital expenditure justification goes through the approval process. You await delivery of a shiny new red spectrometer but a yellow one you rejected due to inadequate software turns up instead. The gnomes of Purchasing have struck again, saying, “I got this one $1,000 cheaper on eBay!” The excuse goes along the lines “Its only an FT-IR, isn’t it?” Purchasing needs to be brought into the data integrity loop; otherwise what hope is there for your organization?

Clause 9.2.2 make an excellent point that a system may be validated, but the data may still be vulnerable (1); for more discussion on this point, please read a previous “Questions of Quality” column on this topic (7).

Section 9.3 has the first mention of the expectation and potential risk of not meeting regulatory expectations, with six topics for validation and maintenance (1). Listed below are the expectations and my comments; for the risks of not meeting them, you’ll have to read the document itself:

Purchasing systems that have been properly evaluated, including their data integrity controls, and validating them for intended use is a key expectation that most people will ignore totally. We will revisit this later. Software configuration settings must be documented, as this is a key part of validation. This section also covers legacy systems and where data integrity controls are lacking “additional controls,” meaning SOPs and training should be implemented.

There should be an inventory of all computerized systems, with risk assessments and data vulnerability assessments. Now, we have an inconsistency with clause 3.7 that states that this guide is not mandatory or enforceable under law. However, there is a legal requirement in Annex 11 for an inventory in clause 4.3, as well as application of risk management in clause 1 (3). Data process mapping to identify data integrity vulnerabilities, and then remediating them, is common sense in today’s regulatory environment (8).

For new systems, a Validation Summary Report (VSR) is expected, along with a shopping list of items that need to be included, including a list of all users, configuration details, etc. (1). The writers of this document appear to have forgotten the middle word of what is being written: summary. A VSR is a concise document that should point to where such information can be found in the suite of validation documents. Who in their right mind would include a list of all current users in a VSR, especially for a large networked application? “Bonkers” is the politest word I can use.

The validation and testing approach outlined is based on Annex 15 on Qualification and Validation (9), rather than Annex 11 for Computerized Systems (3). As such, it is a poor and apparently inflexible approach being focused on manufacturing rather than laboratory systems, as it lists user requirement specifications (URS), design qualification (DQ), factory acceptance test (FAT), site acceptance test (SAT), installation qualification (IQ), operational qualification (OQ), and performance qualification (PQ) documents (1). How many laboratories perform FAT and SAT testing in their validations? This approach ignores clause 2.5 in Annex 15 allowing the merger of documents (9), and, furthermore, does not reference clause 1 of Annex 11 (3) regarding the use of risk management to determine the extent of validation work required. The list omits the configuration specification ,and, if custom software is being developed, functional and design specifications. In all, a poor set of expectations.

Periodic review is another concept not mandatory or enforceable under law. However, we now reach the point where this becomes audience participation time in a British pantomime, where the evil regulatory villain shouts to the audience it’s not enforceable by law—and the laboratory audience shout back, “Oh, yes it is!” Have you ever read Annex 11 clause 11 on periodic evaluation (3)?

We now come to the point in PI-041 where the regulatory enthusiasm for regular system updates (such as, for example, patches and application upgrades) meets the buffers of pharmaceutical reality—requirements for timely update of security patches (1). We assess this in a separate section. 

Another pantomime moment arrives when we consider section 9.4 and data transfer.

The “not mandatory” or “enforceable under law” statement is inconsistent with item 1 about the need to validate data transfer. This is a specific requirement of clause 5 of Annex 11 (3).

A good point regarding PI-041 is when software is updated, the laboratory should confirm that old data can still be read (1). This may be in the existing data format, or may require migration to a new data format, but this is another specific requirement in clause 4.8 of Annex 11 (3) that is legally enforceable.

Now we come to what we say in English is the pièce de résistance: When conversion to the new data format is not possible, the old software should be maintained. Aargh! Computer museum, here we come! The problem is that the life of the data exceeds the life of the instrument that created them. One way of keeping a system on life support could be virtualization, provided that the software does not have a dongle or requires the instrument to be connected to operate. Expectation 3 also looks at the problem further, and suggests conversion to a different file format while acknowledging that data could be lost as a result, but that the process needs to be controlled and validated. The problem is that spectroscopic data are mainly dynamic, and guidance documents expect dynamic data to remain dynamic (1,10–12).

Section 9.5 covers five pages as it sets the expectations for system security, appropriate access to user functions, configuration to ensure data integrity, network protection, and restrictions on use of USB devices (1). Rather than go though the shopping list of non-mandatory expectations, I’ll cherry pick the key items from my perspective. In doing this, I’m making the assumption that you know about individual user identities with appropriate access privileges. Please read the guidance document to gain the full picture of the regulatory expectations here.

Strict segregation of duties is essential so that administrators are independent of laboratory users, which is reflected in several other regulatory guidances (1,10–12). However, one of the common questions that arises during training is if a laboratory has a standalone system with only a few users, how should this be achieved? The PIC/S guidance recognizes that this is a potential problem for smaller organizations, and suggests that a user can have either two user identities or roles:

However, if this option is selected, it is critical that the audit trail records the respective activities to demonstrate no conflict of interest (1).

The format and use of passwords must be controlled. This subject was discussed at length in a previous “Focus on Quality” column (13); don’t have complex passwords that are difficult to remember but string easy to remember words together. Doing so will save those groveling calls to the help desk saying you have forgotten your password and locked your account.

In expectation 1, it is noted that a system should generate a list of successful and unsuccessful login attempts, with dates and times plus a session length if a user has successfully logged in. For the latter expectation in a document that is purportedly non-mandatory and not legally enforceable requirements, this is going too far—you can work this out by log-on and log-off times.

There are many security expectations reading as a list of system requirements that most users won’t have read, or even bother with, during selection of an instrument and its data system. For example, can a system generate a list of users with user identification and their role? Most networked systems can generate a similar list, but not many standalone systems can, as this does not feature on the horizon of supplier requirements. If you use spreadsheets, forget it!

Network protection, including antivirus software, intrusion detection, and firewalls, is covered in table 9.5.3; as noted earlier, operating security patching was mentioned in table 9.3.6 (1). Personally, there should have been a separate section on cybersecurity that merges both these elements instead of separating them.

Table 9.5.3 discusses electronic signatures and notes that records must be permanently linked to their respective record, along with the date and time of signing (1). Again, in contrast to clause 3.7, this is a legal requirement lifted directly from clause 14 of Annex 11 (3). An extension to the requirements of Annex 11 is the expectation to revoke an electronic signature, but show on the record that it is now unsigned. Finally, there is encouragement to use biometric and digital signatures, I have doubts this will be implemented on small laboratory systems until the market requires it, and the technology reduces in price. The placing of this discussion is strange: security. I would have positioned it as a way of improving data integrity and process efficiency by eliminating paper in Section 9.8, as many systems are still hybrid. Hybrid systems will be discussed later. Another spoiler alert! You won’t like the comments.

There is a specific section on the control of USB devices, either as sticks or thumb drives, but also cameras, smartphones, and other similar equipment (1). This section is to ensure that malware is not introduced into an organization, and devices are not used to copy and manipulate data. See the discussion on the use of USB sticks in an earlier “Focus on Quality” column this year (14).

Audit Trails Are Critical for Data Integrity

In contrast with the security section, the one for audit trails is shorter, which is surprising given how critical an audit trail is for ensuring traceability of actions and data integrity. The first expectation is that a laboratory purchases a system with an audit trail. However, there is an alternative option available in EU GMP Annex 11 clause 12.4 that management systems for data and for documents should be designed to record the identity of operators entering, changing, confirming, or deleting data, including date and time (3). Therefore, if an application can meet 12.4 requirements, then an audit trail is not required.

The guidance reiterates that proper system selection should ensure that software should have an adequate audit trail that users must verify the functionality of during the validation (1). Some applications may have more than one audit trail, and it is important to know which ones are important for monitoring changes to data. A good point is that the guidance suggests looking at entries for changes to data with a focus on anomalous or unauthorized activities to allow review by exception. With software having more than one audit trail, the review of any noncritical audit trails can be performed in periodic reviews at longer time intervals.

The guidance reiterates the issue with proper selection of software, as systems that prevent users from changing parameters or permission settings can reduce the need for detailed review of all audit trail entries. This allows a laboratory to implement an effective review by exception approach—but only where there are effective controls that have been implemented, documented, and validated.

The guidance is very clear, along with the WHO and MHRA guidances (10,11), that laboratories must know the criticality of data generated by any computerized system, as well as any vulnerabilities (1). Data process mapping (8) is suggested as a way of identifying vulnerabilities such as shared user identities, or data files stored in directories that can deleted outside of the application with no audit trail entry. The regulatory expectation is that these vulnerabilities should be fixed using technical controls, although the guidance acknowledges that legacy systems may require procedural controls to ensure data integrity before these are updated or replaced. Clause 9.2.2 makes the point that a system may be qualified, calibrated, and validated, but there is no guarantee that the data contained within it are adequately protected (1), hence the need to identify and mitigate any data vulnerabilities.

Don’t Touch the System – It’s Validated!

The traditional approach to validation of a computerized system in a regulated laboratory is once validated, no changes to the system are made. In contrast, in the security section the PIC/S guidance takes a more pragmatic approach on page 37, and suggests that computerized systems should be updated in a timely manner, including security patches and new application versions (1). This is good in principle, but there needs to be a different mindset within industry to commit and follow this in practice. For example, it is easier to update an application incrementally with minor re-validation rather than wait until the application goes out of support, and panic as a full validation and possible data migration project may be required. However, the detail provided here appears to be more of an inspector’s checklist with little flexibility.

Sections 9.7 to 9.9 cover the operational use of a computerized system from data entry, through data review, data storage, and finally destruction of the e-records (1). Yet again, there are regulatory expectations listed that are regulatory requirements, such as:

9.7.1 Critical data entered manually should be verified by a second person or a validated process (clause 6 of Annex 11) (3)

9.7.2 Changes to data need to be authorized and controlled. I’m not certain that “authorized” is the correct word, as the regulations merely use the word “reviewed.” However, these are regulations in Chapter 4.9 (4), Chapter 6.17 vii (15), and Annex 11.9 (3).

9.8.1 requires a risk assessment to identify all GMP critical data (1). Whoa! We performed an analysis, but, before we review the data, we need to identify critical data. We are in the process of reviewing data, and we need a risk assessment? This should have been undertaken during evaluation and validation!

There is the requirement to write an SOP to review audit trails (1). This is wrong. Look around your laboratory; how many different systems are there? There are basic regulatory requirements for audit trails, but there is no standard implementation. You won’t have one SOP for audit trail review; you will have as many review procedures as you have computerized systems (assuming that each one has an audit trail). A far better approach would be to have a single SOP for second person review that calls the appropriate procedure or work instruction for audit trail review, as shown in Figure 2. This is an easier way to review audit trails rather than have a generic SOP that will be useless, as it will not have enough detail for each system’s audit trail review.

An expectation is the need for quality oversight of audit trail review that should be incorporated into the Chapter 9 self-inspections (16) in an organization.

Back to regulations disguised as expectations; most of the items in section 9.9 can be derived from clause 7 of Annex 11 (3):

Backup must include all data and associated metadata, and be capable of recovery

In the absence of alternatives, backup using USB devices must be carefully controlled. Backup must be performed automatically and must not use either drag and drop or copy and paste. The backup must not allow data deletion (1), which implies that it should be performed by an independent person.

There is an assumption that data will be archived offline; however, networked systems can archive data on-line by making the data read only, and only accessible by designated users.

For those that use hybrid systems, Section 9.10 does not make for comfortable reading; such systems are not encouraged, and should be replaced as soon as possible (1). They also require additional controls, due to the complexity of the record set and increased ability to manipulate data. This is essentially the same as the stance taken by the WHO guidance (10).

A potentially onerous requirement is a detailed system description of the entire system that outlines all major components, how they interact, the function of each one, and the controls to ensure the integrity of data. How many hybrid systems do you have in your laboratory, including spreadsheets?

To read further about hybrid systems, and why they need to be replaced, see the following references (17–19).

When GMP work is outsourced, such as API production, con- tract manufacture, or contract analysis, the same principles of data integrity are expected. After an adequate assessment of the facility, technical ability, and approaches to data integrity, requirements for DI and escalation of data integrity issues to contract giver need to be included in agreements between the two parties, as required by Chapter 7 (20). Limitations of summary reports were discussed in Part 1 of this review, as well as ways of reviewing analytical data remotely by video conference (2).

Section 11 has a section where the various PIC/S regulations and guidance documents are mapped to the ALCOA+ criteria. Of most use is a classification of data integrity deficiencies in four categories (1):

Critical: Impact to product with actual or potential risk to patient health

Major: Impact to product with no risk to patient health 

Major: No impact to product, evidence of moderate failure

Other: No impact to product, limited evidence of failure

Examples given for each of the above deficiency categories are given to help during inspections. It may be prudent to adopt the same approach for classifying DI findings in internal audits or supplier assessments.

The last section is guidance of how an organization should investigate data integrity failures. The process is shown in Figure 3 that is taken from Section 12.1.1.1 in PI-041 (1). What the guidance does not state is that a data integrity investigation is a very detailed process and time-consuming process. From personal experience, it is unlikely that any root cause will be determined at stage 2 of the investigation (interviewing employees). Root cause determination comes at stage 6 of Figure 3, when the investigation team has a complete picture of the extent and depth of the data integrity problems.

In addition to identification of data integrity problems, a data integrity investigation will also identify poor data management practices. Don’t ignore them, they need to be identified, classified, and remediated in parallel, as they could be the source of the next data integrity investigation. These will also be included in the data investigation report.

PI-041 makes the assumption that all data integrity problems are caused by evil people. This is not so. I have been involved with one investigation that was caused by a software error (18).

The only other data integrity guidance that goes into detail about investigations is the FDA data integrity guidance in Question 18 (12) that calls out the agency’s Application Integrity Policy (21), which is broadly similar to that shown in Figure 3.

Some of the outcomes of an investigation could be:

Data in a product license submission are falsified

Data manipulation resulting in a risk to patient health, necessitating a batch recall

Is there a material impact that requires the company to inform one or more regulatory authorities? The classic data integrity case is that of AveXis, a subsidiary of Novartis, and data integrity lapses found during the development of the biological Zolgensma for treating muscular atrophy. The company was granted a product license for the product, and then told the FDA there were data integrity problems. Understandably, the FDA were not happy campers, and in July 2019 went on site and subsequently issued a 483 (22). The issue is that a company needs to have a good understanding of all of the data integrity issues, and have at least a draft report available before approaching a regulatory authority to discuss the lapses.

Section 12 ends with a discussion on the indicators of improvement following an onsite inspection to verify that the CAPA actions have been effective. Indicators of improvement are listed as (1):

A thorough and open investigation of the issue, and how to implement effective CAPA actions

Having open communications with clients and other regulators

Evidence of open communication of the DI expectations, and processes for open reporting of issues and opportunities for improvement across the organization

Resolution of the DI violations, and remediation of computerized system vulnerabilities

Implementation of data integrity policies with effective training

Implementation of routine data verification practices

The best advice I can offer is to ensure that DI principles are infused throughout the organization, the PQS, and computerized systems, and your staff are trained in the DI policies. The effort and cost, along with loss of reputation with regulatory authorities, is not worth any efforts to save money. The cost of noncompliance is always much more expensive than the cost of compliance (23).

Columns: Focus on Quality

We continue our explanation of the new PIC/S data guidance.

Are You Ready for the Latest Data Integrity Guidance? Part 2: Computerized Systems, Outsourcing, Quality, and Regulatory Oversight

We investigated in detail the Raman scattering of barium fluorohalides BaFX (X = Cl, Br, I) in an effort to characterize changes in structure and chemical bonding. The overall picture that emerges from the BaFX Raman data is that with the compositional change from X = Cl → Br → I, electron density shifts from the axial Ba-X bond to the equatorial Ba-X bonds with compositional change to the heavier halide, and that the charge distribution becomes more delocalized in the halide layers. This change in electron density indicates an increase in covalency of the Ba-X bonds from Cl to I.

Alkaline earth halides have long been known to function well as host lattices for optically active dopants (1–5). As such, they have found uses as luminescent materials (1), X-ray phosphors (2), and as scintillators (1). Also, alkaline earth fluorohalides have found applications such as X-ray storage phosphors, and have brought about the development of computed or digital radiography (3–6). Computed radiography is a means of imaging the internal structure of the human body, and has become widely used in medicine for the diagnosis of human disease. The storage phosphors employed in computed radiography systems are the europium-doped barium fluorohalides, BaFX:Eu

; whereas here X = Br or a Br/I solid-solution.

The compounds in the BaFX series (X = Cl, Br, I) are isostructural, having the PbFCl structure-type, and all are known to be X-ray storage phosphors when doped with europium. However, it is noteworthy that the Eu

) do not all perform comparably. In particular, the performance of the X-ray storage phosphor is highly dependent upon the chemical composition of the host lattice (7,8). We investigated in detail Raman scattering across this series of compounds in an effort to characterize the changes in structure and chemical bonding.

The barium fluorohalides form with the matlockite type structure, and belong to the D4h crystallographic point group (9). The irreducible representations of the lattice vibrations derived from a group theoretical treatment are:

where IR and R represent infrared active and Raman active modes, respectively. The set of acoustic modes is comprised of one 

 mode. Therefore, the set of optical phonons is given by:

and we should expect a total of six bands to appear in the Raman spectrum of BaFX. The polarizability tensors for the Raman active modes are:

Raman spectra of the BaFX compounds have previously been reported and the bands assigned to specific lattice vibrational modes (10–12). Powder Raman spectra of BaFCl, BaFBr, and BaFI are shown in Figure 1. Each spectrum consists of two 

 Raman bands, as predicted by group theory. The Raman band positions from Figure 1, the lattice vibrational modes from which they arise, and the directions of the atomic motions for the specific vibrational modes, as reported by Haeuseler (11), are indicated in Table I. The notations “parallel” and “anti-parallel” are used here to denote the motion of the halide plane relative to that of the Ba atom, which is positioned immediately below the halide plane. An example of this distinction is illustrated by the parallel and anti-parallel motions of the 

 crystallographic axis is shown in the figure; the halide plane lies in the 

 crystallographic plane. The BaFBr Raman bands at 211.0 cm

 are attributed to motions of the F atoms, and, therefore, the parallel and anti-parallel halide designations do not apply to these bands. Those modes constituted primarily of Ba and X motion will be referred to as 

We explain how to characterize barium fluorohalides with Raman spectroscopy.

December 2021

Cold-Hot Nature Identification of Chinese Medicine Based on an Ultraviolet Chemical Fingerprint

Impact of Measurement Protocol on ICP-MS Data Quality Objectives: Part II

Are You Ready for the Latest Data Integrity Guidance? Part 2: Computerized Systems, Outsourcing, Quality, and Regulatory Oversight

Raman Spectroscopy and Chemical Bonding of BaFCl, BaFBr, and BaFl

Challenges of Spectrofluorometry Part 3: Sample-Specific Concerns