Specific characterization method of medicine crystal form quantitative method

Abstract

The invention relates to the technical field of medicine crystal form quantification, in particular to a specific characterization method of a medicine crystal form quantification method, which is characterized in that a characteristic derivative spectrum is compared with a regression coefficient curve of a quantitative model, and near-infrared absorbance data is multiplied by a correlation multiple suitable for drawing when a graph is drawn. According to the technical scheme, the mode that the characteristic derivative spectrum is compared with the regression coefficient curve of the quantitative model is adopted, the near-infrared absorbance data is multiplied by the correlation multiple suitable for drawing when the graph is drawn, the characteristic wave number range can be accurately judged, and specificity judgment is visual and reliable.

Description

Technical Field

[0001] The present invention relates to the technical field of drug crystal form quantification, and in particular to a specific characterization method for a drug crystal form quantification method. Background Art

[0002] Drug crystal form is one of the important factors affecting drug quality. The polymorphism of chemical raw materials will cause differences in solubility and dissolution, which directly affects the bioavailability and bioequivalence of solid oral drugs. Therefore, the qualitative analysis of the crystal form in existing raw materials or preparations can no longer meet the requirements. To ensure the efficacy of drugs, accurate polymorph quantitative and control analysis methods are needed. This has become one of the technical bottlenecks that restricts the integration of Chinese drugs with international standards, and is also an important factor affecting the consistency evaluation of quality and efficacy between domestic and original research products. Existing drug crystal form research methods mainly include X-ray diffraction, infrared and near-infrared spectroscopy, solid-state nuclear magnetic resonance, differential scanning calorimetry and Raman spectroscopy.

[0003] In order to determine the presence of crystalline APIs in the preparation, a quantitative method is required. Refer to ICHQ2A-Validation of Analytical Methods, USP <1039> Method development and validation are carried out according to the guidelines such as CHEMOMETRICS (chemometrics), ICHQ14-Analytical method development, near-infrared spectrophotometry guidelines (Chinese Pharmacopoeia 2020 Edition Part IV General Chapter 9104) and draft formula of near-infrared spectroscopy (second time). Among them, Chinese Pharmacopoeia 2020 Edition Part IV General Chapter 9104 stipulates that the specificity of the specificity model is usually expressed by the identification accuracy of known samples. Not only is it necessary to verify the identification accuracy of the authentic product, but it is also necessary to conduct challenging validation with samples that are similar in chemical structure or properties to the substances in the model to prove that the model can distinguish these substances. The method for describing specificity in CN111540417A is to weigh 500 mg of canagliflozin crystalline form I powder sample A, and then add 50 mg of canagliflozin crystalline form IV to obtain sample C. According to the near-infrared spectrum of sample C and the above model, it is predicted that the content of crystalline form III in C is -0.02%, that is, there is no interference in the determination between crystalline forms. Yang Yue et al. (2018) studied a quality control method for traditional Chinese medicine production based on near-infrared spectroscopy. The specificity of an NIR method refers to the model's ability to distinguish whether an analyte is a specific analyte. Comparing the sample Mahalanobis distance with a threshold can be used to evaluate the model's specificity. The value is calculated as follows: threshold = factor x rank MM, where rank is the principal component dimension of the PLS model; M is the number of samples in the calibration set; and factor is the coefficient, which in this study was taken as 2. This experiment used 15 batches of medicinal materials, including five batches of white peony root from different sources, as well as dried tangerine peel, tangerine peel, areca root, bergamot root, liquorice root, millettia reticulata, citron, corydalis yanhusuo, epimedium, and perilla stem, for validation. The threshold for the NIR quantitative model for paeoniflorin and paeoniflorin was 0.214. When the Mahalanobis distance of a sample is less than the threshold, the sample is considered specific to the model; when it is greater than the threshold, it is considered abnormal.

[0004] The above method uses Mahalanobis distance and threshold comparison to evaluate model specificity. This method does not comply with the guidelines of Chinese Pharmacopoeia 0403 and the FDA's Development and Submission of Near Infrared Analytical Procedures Guidance for Industry. In contrast, directly comparing the characteristic spectrum with the model's regression curve is more obvious, easier to understand, and can be used as an effective method for implementing these guidelines.

[0005] In view of this, the present invention is proposed. Summary of the Invention

[0006] The object of the present invention is to provide a method for characterizing the specificity of a drug crystal form quantitative method, which can accurately determine the characteristic wavenumber range and make the determination of specificity intuitive and reliable.

[0007] In a first aspect, the present invention provides a specific characterization method for the quantitative method of drug crystal forms, which uses characteristic derivative spectra to compare with the regression coefficient curve of the quantitative model, and when drawing the graph, the near-infrared absorbance data is multiplied by a relevant multiple suitable for drawing.

[0008] Specifically, when drawing a graph, the horizontal axis is the wave number, the vertical axis is the absorbance of the model, and the absorbance of the derivative spectrum of the characteristic spectrum is multiplied by the corresponding coefficient so that the two graphs are at the same absorbance level. This multiple is determined according to the degree of comparison of the graphs.

[0009] The method of the present invention can be used as a specific characterization method in the methodological verification of any near-infrared quantitative method.

[0010] Preferably, the drug is cefditoren pivoxil granules.

[0011] In a second aspect of the present invention, a method for methodological validation of a drug crystal form quantitative method is provided, comprising: preprocessing the near-infrared spectrum using standard normal variable transformation (SNV) and second-order derivatives, and establishing a calibration model using partial least squares (PLS); through the establishment of the calibration model, evaluating different influencing conditions, determining method parameters, and performing methodological validation; the methodological validation includes specificity characterization; the specificity characterization uses a characteristic derivative spectrum to compare with the regression coefficient curve of the quantitative model, and when drawing the graph, multiplying the near-infrared absorbance data by a relevant multiple suitable for drawing.

[0012] Preferably, the near-infrared spectrum is obtained by scanning a series of calibration samples specially designed according to the requirements of the guiding principles using a Fourier transform near-infrared spectrometer.

[0013] Preferably, the calibration sample spectrum is used to establish a standard normal variable transformation and second-order derivative for preprocessing the near-infrared spectrum, and a partial least squares method is used for quantitative modeling.

[0014] Preferably, the different influencing conditions include: wavelength range, processing parameters, crystal content concentration, and the impact of moisture on durability and particle size evaluation.

[0015] Preferably, the methodological validation further includes: characterization of method linearity, method accuracy, method repeatability and intermediate precision.

[0016] Preferably, the drug is cefditoren pivoxil granules.

[0017] The third aspect of the present invention provides the application of the specific characterization method of the above-mentioned drug crystal form quantitative method or the methodological verification method of the above-mentioned drug crystal form quantitative method for near-infrared quantitative detection of the crystal raw material content or crystal form impurities in cefditoren pivoxil granules.

[0018] Preferably, the crystal content of the active ingredient in the cefditoren pivoxil granules is calculated by using the model and the unknown sample spectrum.

[0019] A fourth aspect of the present invention provides a method for quantitatively determining the crystal form of cefditoren pivoxil granules, comprising the following steps:

[0020] 1) Preparation of calibration model samples: Develop a near-infrared quantitative model calibration sample design containing concentration variables for four main components. The key feature of this design is that the concentration changes of the four components are independent of each other to ensure that they are linearly uncorrelated during modeling;

[0021] 2) Preparation of validation model samples: Following the established formula ratio, only the crystalline content and the amorphous raw material ratio were varied;

[0022] 3) Collect the spectrum set of raw materials and the spectrum set of calibration samples;

[0023] 4) Preprocessing the spectrum set with different parameters;

[0024] 5) development of multivariable models;

[0025] 6) predictions of different models on the validation model samples;

[0026] 7) Determine the best model;

[0027] 8) Methodological validation;

[0028] 9) Predict unknown samples.

[0029] Preferably, during the preparation of the calibration sample in step 1), the concentrations of the other five components in the cefditoren pivoxil granule formula remain unchanged. This is because their contents are relatively low and do not have a significant impact on the quantitative model. Their impact on quantitative analysis can also be eliminated by other means, such as selecting an appropriate spectral preprocessing method and the wavelength range used for modeling. The linear independence of the samples is to avoid the problem of collinearity.

[0030] Preferably, the spectrum set collected in step 3) is to demonstrate the effectiveness of spectral preprocessing. When using the spectral preprocessing step, care must be taken to avoid introducing artificial noise or losing important information. Preprocessing uses SNV (standard normal variate transformation) + second-order derivative to preprocess the near-infrared spectrum and use partial least squares (PLS) modeling.

[0031] Preferably, the development of the multivariate model in step 5) is based on the fact that the derivative spectra of the amorphous raw material, the crystalline raw material, and the excipients have strong characteristic absorption peaks, and the two main components also have characteristic absorption peaks, which makes it possible to establish a quantitative model. The specific characterization can clarify the wavelength (wavenumber) range.

[0032] Preferably, in step 6), the optimal model and parameters are determined by prediction of the verification sample.

[0033] Preferably, the unknown samples predicted in step 9) include day 0 and stability samples.

[0034] The present invention has at least the following beneficial effects:

[0035] The technical solution of the present invention adopts a method of comparing the characteristic derivative spectrum with the regression coefficient curve of the quantitative model. When drawing the graph, the near-infrared absorbance data is multiplied by a relevant multiple suitable for drawing. This can accurately determine the characteristic wavenumber range, and the judgment of specificity is intuitive and reliable. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0037] Figure 1 The original near-infrared spectrum of the T1 spectrum set provided by the present invention;

[0038] Figure 2 The T2 spectrum set after SNV processing provided by the present invention;

[0039] Figure 3 The T3 correction spectrum set provided by the present invention;

[0040] Figure 4 The wavelength range of the spectrum provided by the present invention is 9945–4344 cm -1 picture;

[0041] Figure 5 The wavelength range of the spectrum provided by the present invention is 8741–5941 cm -1 picture;

[0042] Figure 6 The PLS2 model diagram provided by the present invention;

[0043] Figure 7 A proprietary diagram provided for the present invention;

[0044] Figure 8 This is a linear regression curve diagram provided by the present invention. DETAILED DESCRIPTION

[0045] It should be noted that the following detailed descriptions are illustrative and intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present application belongs.

[0046] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular also includes the plural. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, they indicate the presence of features, steps, operations, devices, components and / or combinations thereof.

[0047] The following will clearly and completely describe the technical solutions of the present invention in conjunction with the embodiments. Obviously, the embodiments described are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0048] Example

[0049] The information of the near-infrared spectrometer used in this embodiment is as follows:

[0050] Instrument Fourier transform near-infrared spectrometer

[0051] Model MAP

[0052] Spectral type diffuse reflection

[0053] Monochromator principle Fourier transform

[0054] Modeling spectrum wavelength range 8741-5941cm -1

[0055] Detector InGaAs 2.6um

[0056] Sample glass container glass bottle

[0057] Spectral resolution 2cm -1 (0.3nm at 1250nm)

[0058] Wave number accuracy is better than 0.05cm -1 (0.01nm at 1250nm)

[0059] Wave number accuracy is better than 0.1cm-1 (0.02nm at 1250nm)

[0060] Absorbance accuracy 0.1% T

[0061] Sample spectrum scan times 128

[0062] Spectral preprocessing: 1) SNV (Standard Normal Variate)

[0063] 2) Second-order derivative (S. Golay, 7 / 7 / 2)

[0064] 1. Prepare samples and spectral sets for calibration and validation models

[0065] Table 1 shows the ratio of main raw materials and auxiliary materials for calibration model. Table 2 shows the ratio of main raw materials and auxiliary materials for validation model. The preparation process of samples is consistent with the production process, but the batch size is different.

[0066] Table 1 Calibration model samples

[0067]

[0068]

[0069] Table 2 Validation batches

[0070]

[0071] The calibration samples were collected through four sampling points set in the drying step of the preparation process. The moisture content of the calibration samples obtained at different sampling points was measured, and 40 calibration samples were obtained.

[0072] For convenience, spectral data can be defined as different spectral sets:

[0073] The wavelength of the spectrum is in wave numbers cm -1 To express it, the spectral data of each specific wave number is also considered as an x ​​variable, and the concentration of each component corresponding to it is considered as a y variable.

[0074] Original Calibration Spectral Set (T1) – This spectral set includes spectra from 32 calibration particle samples from eight batches, including DOEs 1-8, for a total of 160 near-infrared spectra. The spectra in this spectral set are original, unprocessed spectra.

[0075] Corrected spectrum set after SNV preprocessing (T2) - This spectrum set is obtained after the T1 spectrum set is calculated after SNV preprocessing. During this preprocessing step, the entire wavelength range of the near-infrared spectrum is covered.

[0076] Corrected spectrum set (T3) after SNV and second-order derivative preprocessing - This spectrum set is obtained after the T2 spectrum set is preprocessed by the second-order derivative. The derivative spectrum calculation uses the S.Golay method and the 7 / 7 / 2 smoothing mode. The full wavelength range of the near-infrared spectrum is covered during this preprocessing step.

[0077] Corrected spectrum set (T4) after SNV and second-order derivative preprocessing - This spectrum set is obtained after the T2 spectrum set is preprocessed by the second-order derivative. The derivative spectrum calculation uses the S.Golay method and 5 / 5 / 2 smoothing mode. The full wavelength range of the near-infrared spectrum is covered during this preprocessing step.

[0078] Corrected spectrum set (T5) after SNV and second-order derivative preprocessing - This spectrum set is obtained after the T2 spectrum set is preprocessed by the second-order derivative. The derivative spectrum calculation uses the S.Golay method and the 9 / 9 / 2 smoothing mode. The full wavelength range of the near-infrared spectrum is covered during this preprocessing step.

[0079] Corrected spectrum set (T6) after SNV and second-order derivative preprocessing - This spectrum set is obtained after the T2 spectrum set is preprocessed by the second-order derivative. The derivative spectrum calculation uses the S.Golay method and the 11 / 11 / 2 smoothing mode. The full wavelength range of the near-infrared spectrum is covered during this preprocessing step.

[0080] Corrected spectral set preprocessed with SNV and second derivatives (T7) - This spectral set was preprocessed in the same way as the T3 spectral set, but with the addition of spectra from DOE6 and DOE12, for a total of 200 NIR spectra.

[0081] Spectral preprocessing:

[0082] The effectiveness of spectral preprocessing is demonstrated using NIR spectra from spectral sets T1, T2, and T3. Care must be taken when using spectral preprocessing steps to avoid introducing artifacts or losing important information.

[0083] Figure 1 The raw NIR spectra of the T1 spectral set are shown. As expected, they exhibit broad absorption bands and large variations in absorbance values ​​(baseline shifts), making them unsuitable for direct use in the development of quantitative models due to the well-known properties of NIR spectroscopy itself, the light scattering effects of the particle sample, and other physical factors.

[0084] Figure 2 The image shows the T2 spectrum set after SNV (Standard Normal Variate) processing. This preprocessing step covers the entire wavelength range of the near-infrared spectrum. This preprocessing step can effectively remove the influence of baseline offset on the spectrum in the T1 set.

[0085] Figure 3 The image shows a T3-corrected spectrum set, obtained by calculating the second-order derivative of the T2 spectrum set. The derivative calculation covers the entire wavelength range, using the S. Golay method and a 7 / 7 / 2 smoothing scheme (number of points on the left / number of points on the right / degree of the polynomial). Derivative calculations can significantly increase spectral resolution.

[0086] 2. Multivariate model development

[0087] Figure 4 and Figure 5 The NIR spectra of the four main component reference samples were preprocessed using the same preprocessing method as the T3 spectrum set. They are: 1) amorphous cefditoren pivoxil (API-AM), 2) sucrose (Sucrose), 3) sweetener (Sweetener) and 4) crystalline cefditoren pivoxil (API-Crystal). Figure 4 The spectral wavelength range is 9945-4344cm -1 , Figure 5 The spectral wavelength range is 8741-5941cm -1 ,exist Figure 5 In the wavelength range, the derivative spectra of crystalline cefditoren pivoxil and sucrose have strong characteristic absorption peaks, and the other two main components also have characteristic absorption peaks, which makes it possible to establish a quantitative model.

[0088] The key variables examined during the development of the quantitative model were 1) the wavelength range of the calibration spectral set, 2) calibration spectral set preprocessing parameters (derivative calculation parameters), 3) the concentration range of crystalline cefditoren pivoxil, 4) the water content of the calibration sample, and 5) the impact of particle size (grinding) on ​​modeling. Variables 4) and 5) will be discussed in the section on robustness of the quantitative model.

[0089] 3. Impact of wavelength range on quantitative model

[0090] To evaluate the impact of wavelength range on quantitative models, spectral data at different wavelength ranges were extracted from calibration spectrum set T3 for modeling. Table 3 lists four characterization parameters that can be used for preliminary evaluation of quantitative models PLS1-4. Based on these parameters, models PLS2-4 used a wider wavelength range and produced better quantitative models. PLS1, on the other hand, used a narrower wavelength range. Although it included the characteristic absorption peaks of crystalline cefditoren pivoxil, this model exhibited poor quantitative analysis capabilities due to its inability to completely eliminate interference from other components.

[0091] In order to more accurately evaluate the quantitative models in Table 3, these quantitative models were used to predict the content of crystalline cefditoren pivoxil in three validation sample batches. Each validation sample had 20 near-infrared spectra of samples from four sampling points. Table 4 lists the prediction results of the quantitative models, and the average, standard deviation, and relative standard deviation of the prediction results were calculated using the 20 results of each validation sample batch as a set of data. In addition, the average deviation between the predicted average and the standard addition amount was calculated:

[0092] Relative deviation = 100 × A / B

[0093] Where A is the absolute value of the difference between the predicted mean and the standard addition amount, and B is the standard addition amount. From the prediction results of the quantitative model, the prediction results of PLS2 have a smaller relative deviation, so it was selected as the model for further optimization and verification ( Figure 6 ).

[0094] Table 3 Corrected spectral wavelength range and quantitative model characterization parameters

[0095]

[0096] Table 4 Prediction results of the quantitative models for validation samples obtained in different wavelength ranges

[0097]

[0098]

[0099] 4. Influence of spectral preprocessing parameters on quantitative models

[0100] In the two steps of preprocessing the near-infrared spectrum, different parameters used in the derivative calculation will have a certain impact on the accuracy of the quantitative model. The parameters involved in the S.Golay method used in this report are the number of points for smoothing calculation and the number of polynomial powers. For example, 7 / 7 / 2 means that a square curve is used for smoothing calculation, with 7 smoothing points on the left and right. If the number of smoothing points is too large, it will affect the separation between spectral peaks and thus affect the resolution of the quantitative model, while if the number of smoothing points is too small, the influence of spectral noise cannot be removed. In order to evaluate the effect of different spectral preprocessing parameters on the quantitative model, spectral data of the same wavelength range were intercepted from the calibration spectrum sets T3, T4, T5 and T6 for modeling (8741-5941cm -1 Table 5 lists the four characterization parameters of the quantitative models PLS5-7 and the parameters of PLS2 for comparison. From these parameters, there is no significant difference between the four quantitative models.

[0101] PLS5-7 were also validated using validation samples. Table 6 lists the prediction results of these quantitative models for the validation samples, using the same calculation method as in Table 4. The results show that PLS5 had large average biases for quantitative prediction of validation samples 1 and 3, at 16.29% and 12.0%, respectively. The average biases of the prediction results of PLS2 and PLS6-7 for all three validation sample batches were less than 10%.

[0102] Table 5 Second-order derivative calculation parameters and quantitative model characterization parameters

[0103]

[0104] Table 6 Prediction results of the validation samples obtained by the quantitative model with different second-order derivative calculation parameters

[0105]

[0106]

[0107] 5. Effect of crystalline cefditoren pivoxil concentration range on quantitative model

[0108] When evaluating the impact of the concentration range of crystalline cefditoren pivoxil on the quantitative models, the effect of the number of DOE batches involved in each model on the results was ignored, and only the concentration range factor was considered. Comparing the characterization parameters of the three quantitative models in Table 7 does not allow for a judgment on the superiority of these models. However, the prediction results for the validation samples in Table 8 show that PLS8 has a large average deviation (62.3%) in the quantitative prediction of validation sample 1, while the average deviations of the prediction results of PLS2 and PLS9 are both less than 10%.

[0109] Table 7 Concentration range of crystalline cephalosporins and quantitative model characterization parameters

[0110]

[0111]

[0112] Table 8 Prediction results of the quantitative model for validation samples obtained in different crystalline cephalosporin concentration ranges

[0113]

[0114] In summary, by evaluating the effects of the spectral wavelength range of the calibration spectrum set, the preprocessing parameters of the calibration spectrum set, and the concentration range of crystalline cefditoren pivoxil on the quantitative model, PLS2 showed good accuracy in determining the crystalline content of cefditoren pivoxil and was selected as the quantitative model for this method after comprehensive consideration.

[0115] Further investigation on the durability of the model

[0116] 1. Impact of wavelength range on durability

[0117] When testing the effect of wavelength range on durability, 10 data points were added or subtracted at both ends of the wavelength range according to the PLS2 model. The new wavelength range and analysis results are shown in Table 9. The average deviation results of the three validation samples show that the quantitative model was not significantly affected by the small changes in the test wavelength range.

[0118] Table 9 Effect of wavelength range on durability

[0119]

[0120] 2. Impact of smoothing calculation parameters on durability

[0121] To test the effect of smoothing parameters on robustness, minimal changes were made to the parameters used in the PLS2 model. The new parameters and analysis results are shown in Table 10. The average deviations of the three validation samples indicate that the smoothing parameters have a modest impact on the quantitative model, but do not significantly alter the analytical results. The quantitative model's sensitivity to the smoothing parameters may be due to the low absolute content of crystalline cefditoren pivoxil in the formulation and the high number of other interfering ingredients.

[0122] Table 10 Effect of smoothing calculation processing parameters on durability

[0123]

[0124] 3. Effect of moisture content of calibration sample on durability

[0125] This report has taken into account the effect of water content in the sample on the quantitative model when designing the calibration sample. Therefore, when the PLS2 model predicts unknown samples, its accuracy is not affected by different water contents in the sample within a certain range. This can be demonstrated by the prediction results of the verification sample.

[0126] Table 11 shows the mean, standard deviation, and relative standard deviation of the predicted results for each validation batch at each sampling point, calculated for each of the five predicted results. The mean deviation between the predicted mean and the standard addition was also calculated. The results showed no significant correlation between the mean predicted value and the mean deviation and the moisture content of the validation samples.

[0127] Table 11 Prediction results of PLS2 model for different water content test samples

[0128]

[0129] 4. Impact of sample grinding on durability

[0130] Table 12 shows the mean, standard deviation, and relative standard deviation of the predicted results for validation batch 1, calculated for each sampling point and for each of the five predicted results. The mean deviation between the predicted mean and the standard addition was also calculated. The results indicate that grinding the unknown sample significantly increases the deviation in the predicted results, making it inappropriate to include this step in actual analytical procedures.

[0131] Table 12 Prediction results of PLS2 model for ground and unground test samples

[0132]

[0133]

[0134] Further validation of the PLS2 model using near-infrared methods

[0135] The experimental results of the near-infrared method development part showed that the method discussed in this report for determining the content of active ingredient crystals in cefditoren pivoxil granules is feasible. It was further validated according to the guidelines of relevant methods. The validation parameters included method specificity, method linearity, method accuracy, method repeatability and intermediate precision.

[0136] 1. Method specificity

[0137] After searching, we found that there are different ways to characterize the specificity of the method validation. As explained in the technical background, the method with higher spectral recognition and more intuitive way to demonstrate the specificity of the PLS model is to compare the near-infrared absorption spectrum of the measured component with the regression coefficient curve of the quantitative model (Regression Coefficients, see Figure 6 ). Figure 7 The red curve is the regression coefficient curve, and the blue curve is the derivative spectrum of cefditoren pivoxil crystals. Its preprocessing method is the same as that of spectrum set T3. Its wavelength range is 8741-5941 cm-1, with a total of 364 absorbance data. The wavelength difference between two adjacent points is 7.69 cm-1. To facilitate plotting and comparison with the regression coefficient of the quantitative model, each absorbance data of the near-infrared absorption spectrum is multiplied by a coefficient of 200,000. Figure 7 The horizontal axis is the number 1-400, which is used to replace the wave number, and the vertical axis is the regression coefficient value.

[0138] Figure 7 Displayed in the wavelength range 6187-5987cm -1 (332-358), the characteristic absorption peaks of cefditoren pivoxil crystals have a good coincidence with the regression coefficient curve of the quantitative model, showing that the PLS2 model has good specificity.

[0139] 2. Method linearity

[0140] According to this method, the average predicted value of 6 validation sample batches (Table 13) and the theoretical addition amount of cefditoren pivoxil crystals were regressed and calculated to obtain Figure 8 The regression curve shows that the method is linear in the range of 1-7% crystal content. The slope of the regression curve is 0.968 and R2 is 0.994.

[0141] Table 13 Near infrared prediction verification sample crystal content

[0142]

[0143] 3. Method reproducibility

[0144] The repeatability of the method was tested using six samples from sampling point 4 of validation batch 1. The prediction results are shown in Table 14. Five near-infrared spectra were measured for each sample, and the average predicted value was obtained from the prediction results. The grand mean was then calculated based on the six average predicted values. The RSD calculated from the grand mean and the SD was 4.48%, indicating good repeatability of the method.

[0145] Table 14 Method repeatability

[0146]

[0147] 4. Method accuracy

[0148] The accuracy of the method was tested using samples from 1) validation batch 4, sampling point 4; 2) validation batch 5, sampling point 4; and 3) validation batch 1, sampling point 4. Ten samples were collected from each batch, and five near-infrared spectra were obtained for each sample, for a total of 50 spectra. The predicted values ​​for each sample in each validation batch were averaged, and then a grand average was calculated from the 10 average predicted values. The average deviations (%) calculated from the grand average and the standard addition amount were 3.3%, 0.95%, and 1.13%, respectively. Table 15 shows that the method has good accuracy at 1%, 2%, and 3% crystal contents.

[0149] Table 15 Method accuracy

[0150]

[0151]

[0152] 5. Intermediate precision of the method

[0153] The intermediate precision of the method was tested at sampling point 4 in validation batch 1. Two experimenters each took one sample and measured its near-infrared spectrum on the first, second, and third days. Five near-infrared spectra were obtained for each sample each time. The prediction results are shown in Table 16. The average predicted value was obtained from the prediction results, and then the grand average was calculated based on the six average predicted values. The grand average of the two experimenters showed that the method had good intermediate precision.

[0154] Table 16 Method intermediate precision

[0155]

[0156] Conclusion: The experimental results of method specificity, method linearity, method accuracy, method repeatability and intermediate precision discussed above show that the near-infrared method for the determination of the active ingredient crystalline content in cefditoren pivoxil granules discussed in this report meets the requirements of various guidelines and can be used for the purpose of quality control of relevant drugs.

[0157] Further prediction was performed on the samples accelerated for 6 months, and the results are shown in Table 17.

[0158] Table 17 Prediction results for 6-month accelerated samples

[0159]

[0160]

[0161] Conclusion: The crystal content of samples accelerated for 6 months was less than 1%.

[0162] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the above embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A specific characterization method for a quantitative method of a drug crystal form, characterized in that: The characteristic derivative spectrum is compared with the regression coefficient curve of the quantitative model, and the near-infrared absorbance data is multiplied by a relevant multiple suitable for drawing when drawing the graph.

2. The specific characterization method of the drug crystal form quantitative method according to claim 1, characterized in that: The drug is cefditoren pivoxil granules.

3. A method for methodological validation of a drug crystal form quantitative method, characterized in that: The method comprises the following steps: pre-processing the near-infrared spectrum by using standard normal variable transformation and second-order derivative, establishing a calibration model by using partial least squares (PLS); evaluating different influencing conditions, determining method parameters, and performing methodological verification by establishing the calibration model; the methodological verification comprises specificity characterization; the specificity characterization comprises comparing the characteristic derivative spectrum with the regression coefficient curve of the quantitative model, and multiplying the near-infrared absorbance data by a relevant multiple suitable for drawing when drawing the graph.

4. The methodological verification method of the drug crystal form quantitative method according to claim 3, characterized in that: The near-infrared spectrum is obtained by scanning a designed calibration sample series using a Fourier transform near-infrared spectrometer.

5. The methodological verification method of the drug crystal form quantitative method according to claim 3, characterized in that: The calibration sample spectrum was used to establish the standard normal variable transformation and second-order derivative to preprocess the near-infrared spectrum, and the partial least squares method was used to quantify the model.

6. The methodological validation method of the drug crystal form quantitative method according to claim 3, characterized in that: The different influencing conditions include: wavelength range, processing parameters, crystal content concentration, and the impact of moisture on durability and particle size evaluation.

7. The methodological validation method of the drug crystal form quantitative method according to claim 3, characterized in that: The methodological validation also includes the characterization of method linearity, method accuracy, method repeatability and intermediate precision.

8. The methodological validation method for the quantitative method of drug crystal forms according to claim 3, wherein the drug is cefditoren pivoxil granules.

9. Use of the specific characterization method for the quantitative method for drug crystal forms according to any one of claims 1 to 2 or the methodological validation method for the quantitative method for drug crystal forms according to any one of claims 3 to 8, characterized in that: Used for near-infrared quantitative detection of crystal raw material content or crystal impurities in cefditoren pivoxil granules.

10. The use according to claim 9, characterized in that The crystal content of the active ingredient in cefditoren pivoxil granules was calculated using the model and the unknown sample spectrum.

Images