Chromatogram shape determination method

Abstract

To provide a method capable of easily determining the state of separation according to a shape of chromatogram without using indices such as a stage number used in liquid chromatography, the degree of separation, and an asymmetry coefficient.SOLUTION: A determining method is for similarity between reference and verification chromatogram by using a standard sample in liquid chromatography, and 1) performs offset processing so that output at a starting point of chromatogram becomes zero; and 2) multiplies a standard value (H) of a first component to be a reference of a standard sample and a ratio (H / S) of an output (S) after the first component prior setting and detecting an upper limit value (M) relative to a whole chromatogram and the standard value (H) of the first component to each time output of chromatogram and, in the case that it is beyond upper limit value (M), makes a new output value an upper limit value (M) and, in the case that it is below the upper limit value (M), makes the value obtained by multiplication a new output value, acquires standardization chromatogram, and evaluates similarity.SELECTED DRAWING: Figure 8

Description

[Technical Field]

[0001] The present invention relates to a technique that can distinguish changes in the pattern of a liquid chromatography chromatogram without prior experience or knowledge. [Background technology]

[0002] Liquid chromatography is a technique used to separate and quantify various mixtures. Separation occurs due to differences in the properties of the column (packing material) and sample components, and their interactions with the eluent, and each component is detected using various detectors. In qualitative analysis, the general method is to identify the component by taking advantage of the fact that each component elutes at a specific time under the same separation conditions. In quantitative analysis, the relationship between concentration and detector output (calibration curve) is obtained in advance using multiple standard samples with known concentrations, and the concentration is calculated using the calibration curve from the detector output obtained with an unknown sample.

[0003] Liquid chromatography can separate a variety of samples, including low molecular weight compounds, synthetic polymers, and proteins, but different "separation modes" based on different separation principles are used depending on the purpose. Separation modes such as "reverse phase chromatography" are often used for low molecular weight compounds, and "size exclusion chromatography" based on differences in molecular size is often used for synthetic polymers. For separating proteins, size exclusion chromatography and "ion exchange chromatography," which separates based on differences in charge, are often used.

[0004] In recent years, liquid chromatography has been increasingly used in clinical testing. One typical application is the measurement of HbA1c (hemoglobin S-A1c), a marker of diabetes. After hemolyzing and diluting a blood sample (patient sample), the sample is injected into a column packed with ion exchange resin. The hemoglobin is separated into various fractions using multiple eluents with different ionic strengths. The fraction is then detected using a visible absorption detector (around 415 nm). The HbA1c% is calculated from the amount of the S-A1c peak fraction (ion exchange chromatography). In practice, the HbA1c% is calculated from the ratio of the S-A1c peak area to the sum of the total peak area and a calibration curve prepared in advance from standard sample samples. To ensure the accuracy of the HbA1c% obtained here, it is important to constantly check whether the chromatogram shape is normal.

[0005] In typical HPLC, column performance and separation performance are expressed using parameters such as "column theoretical plate number," "resolution," "symmetry factor," and "peak symmetry" (see Figure 1). However, while these parameters are useful for "isocratic elution," in which the eluent composition is constant, they may not always accurately represent column performance and separation performance in "gradient elution," in which the eluent composition is changed gradually or in steps. This is because linear gradient and step gradient elution intentionally increase the elution power of component peaks that would normally elute, causing them to elute earlier (see Figure 3).

[0006] Measurement of HbA1c% using liquid chromatography involves switching between eluents with different ionic strengths for separation, and so presents similar challenges. Furthermore, the measurement time for the HPLC method used to measure HbA1c is extremely short, less than one minute, making it difficult to assess performance using the parameters mentioned above. Figure 2 shows the calculated plate numbers for a 2216-plate column, assuming elution over 10 minutes and elution over 1 minute.

[0007] For example, if the target component elutes in 10 minutes, the half-width is 0.5 minutes, and if it elutes in 1 minute, the half-width is 0.05 minutes. If the half-width fluctuates by 0.01 minutes, the plate number will fluctuate slightly from 2130 to 2307 in the 10-minute period, but will fluctuate significantly from 1539 to 3463 in the 1-minute period, making it unreliable to evaluate performance using the plate number.

[0008] When measuring HbA1c% using liquid chromatography, the various hemoglobin fractions (peaks) are not completely separated, making it difficult to calculate the resolution. Furthermore, the column plate number can only be calculated using the hemoglobin A0 peak, the last fraction, and its elution time is very short, at approximately 1 minute, so the calculated plate number is low to begin with and does not accurately represent separation performance. Therefore, currently, judgments are made based on the occurrence of phenomena such as abnormal measured values or high column pressure, or on the shape of the chromatogram. Automatic judgment is desirable. Summary of the Invention [Problem to be solved by the invention]

[0009] A method is provided that enables the state of separation to be easily determined from the shape of a chromatogram without using indicators such as the number of plates, resolution, and asymmetry coefficient used in liquid chromatography. [Means for solving the problem]

[0010] HbA1c% measured using liquid chromatography is calculated as the ratio of the S-A1c peak area to the total area of all hemoglobin fractions. HbA1c% is approximately 5-6%, and the S-A1c peak area% is also approximately 5-6%, with 85-90% being hemoglobin A0 and the remainder being peaks of other hemoglobin fractions. A typical chromatogram pattern is for the low-abundance "other fractions" to elute first, followed by the target S-A1c, and finally the main component A0. As can be seen in Figure 4, when a chromatogram is displayed at full scale, the fluctuations in the low-abundance S-A1c peak and the other fraction peaks are difficult to see, so it is often displayed or printed enlarged. Furthermore, although the eluent used for separation does not have significant absorption in the visible light region, it does have a slight background. This background is theoretically always constant, but it fluctuates slightly due to various environmental factors.

[0011] The present invention aims to numerically determine the shape of a chromatogram of a standard sample (or control sample) composed of multiple components, which is measured periodically to confirm analytical accuracy during routine analysis by liquid chromatography.

[0012] To make the process easier to understand, we will explain it based on a chromatogram obtained by measuring HbA1c using liquid chromatography.

[0013] Figure 4 shows a chromatogram of this. For ease of explanation, we will simplify it to three fraction peaks: "Others," "S-A1c," and "A0," in descending order of elution. The S-A1c peak is the peak to be quantified.

[0014] The basic procedure is to use a standard sample, memorize the shape of a reference chromatogram, and compare it with the shape of the chromatogram of a standard sample that is measured periodically.A major feature of the present invention is that various processes are performed in advance to obtain accurate chromatogram comparison results.

[0015] The first aspect of the present invention is a method for evaluating similarity after correcting and normalizing the entire chromatogram. The second aspect is a method for evaluating similarity after eliminating excessively large regions without normalizing the chromatogram. The processing flow is shown in Figure 8.

[0016] First, the first embodiment will be described. Chromatogram processing is performed according to the following procedure and method (processing on the left side of FIG. 8). 1) Offset processing 2) A specific peak is designated among the multiple components that make up the standard sample, and output correction is performed so that the output is the same. 3) If the corrected output is equal to or less than a certain value, the new output is left as the value after the offset processing, and if the corrected value is equal to or greater than the certain value, the output is set to the certain value.

[0017] A chromatogram is composed of a combination of (time, output). First, offset processing (1) is performed because it is difficult to compare chromatograms with different outputs at the starting point (time zero). The output value at the starting point (time zero) is subtracted (see Figure 5). Some recent chromatographs have a function that automatically zeros the output when a sample is injected, in which case this processing is not necessary. However, because the automatic zeroing function also varies depending on the model / machine, it is more preferable to perform offset processing here. Although the automatic zeroing function is useful, there are cases where the zero point is not exactly zero, resulting in a slightly positive or negative value, or the timing of the auto-zero execution is off, as shown in Figure 6.

[0018] Next, we will explain the correction methods 2) and 3). As mentioned above, the chromatogram obtained from HbA1c measurement has a small S-A1c peak, and the A0 peak is about 90%, resulting in an unbalanced chromatogram pattern. In this state, it is difficult to see changes in the S-A1c peak, which is the target of quantification, or in the peaks that appear before it. Therefore, the chromatogram is corrected and normalized using the following procedure.

[0019] First, the S-A1c peak is identified and its peak height (S) is obtained. The output of the S-A1c peak is corrected to be constant for all chromatograms. If the value to be constant is the standard value (H), the correction coefficient (f) is H / S. The output at each time after the offset processing is multiplied by the correction coefficient (f). If the multiplication result is below a constant value (M), the multiplied value is used as the new output value. If the multiplication result is above the constant value (M), the new output value is used as the M value. With this process, in the final normalized chromatogram, the output at the apex of the S-A1c peak will be H, and the output near the apex of the predominant A0 peak will be M.

[0020] Figure 7a shows a schematic representation of the chromatogram after offsetting, and Figure 7b shows a schematic representation of the chromatogram after normalization.

[0021] Table 1 shows the final normalized chromatogram data. These are two chromatograms with slightly different chromatographic shapes, which will be described later: the "reference chromatogram" and the "verification chromatogram." In the table, column (1) is the time, column (2) is the normalized output value of the "reference chromatogram," and column (3) is the normalized output value of the "verification chromatogram."

[0022] Column (2) is the normalized output value of the "reference chromatogram," and column (3) is the normalized output value of the "verification chromatogram." The time of the peak top of S-A1c in the reference chromatogram is t7, the time of the peak top of S-A1c in the verification chromatogram is t9, and the A0 peak is around t21 to t28.

[0023] In the normalized chromatogram, the output of the reference chromatogram at t7 is H, and the output of the verification chromatogram at t9 is H. In addition, the output is M around times t21 to t28.

[0024] Regarding the relationship between the standard value (H) of the S-A1c peak and the upper limit value (M) of the output, it is only necessary that H < M, and there is no restriction, but it is preferable that M is about three times the value of S. Also, in order to make it easier to discriminate the shape of the chromatogram, it is desirable to set M to 1, 10, 100, etc. in normalized values.

[0025]

Table 1

[0026] [[ID=eleven]] Next, the shape determination of the chromatogram will be described.

[0027] The shape determination evaluates the similarity between the "reference chromatogram" serving as a reference and the "verification chromatogram" for comparison. The sample used for verification should be stable with the same components and concentrations and no change over time. In the measurement of HbA1c, it is preferable to use the "control sample" commercially available as a lyophilized product.

[0028] Fig. 9 shows the flow of the shape determination of the chromatogram in the present invention. The reference chromatogram can use the chromatogram obtained when the state of the analyzer (HPLC) used for measurement is good and the performance of the eluent and the analytical column used for analysis is guaranteed. Since the analytical column has a lifespan, it is preferable to update the analytical column and use the chromatogram measured in a good state. This makes it possible to follow the change over time of the analytical column. The verification chromatogram is obtained during the routine measurement intervals. The acquisition interval can be determined considering the assumed lifespan of the analytical column and is arbitrary.

[0029] Also, the previous chromatogram of the verification chromatogram obtained during the routine measurement intervals may be treated as the reference chromatogram (see Fig. 10). In this case, sudden abnormalities can be discriminated. This is because the deterioration of the analytical column is considered to progress gradually.

[0030] Figure 11 shows a schematic diagram of the change in similarity. Figure 11a shows the initial chromatogram, and Figure 11b shows the case where the previous chromatogram is used as the reference.

[0031] The chromatogram comparison can be performed using any calculation formula that can evaluate similarity, but is not particularly limited. It is preferable to use "cosine distance." In the case of "cosine distance," a perfect match results in a value of 1.000, and the value decreases as the similarity decreases. The general formula is expressed by Equation 1.

[0032]

number

[0033] Applying this to the example in Table 1, we get number 2.

[0034]

number

[0035] In this way, by calculating a value indicating the similarity between the periodically acquired verification chromatogram and the reference chromatogram, it becomes possible to numerically determine the difference in the shape of the chromatogram. This method is particularly useful for chromatograms obtained by measuring HbA1c using liquid chromatography. By normalizing the chromatograms before evaluating their similarity, the effect of the large A0 peak is reduced. Furthermore, by correcting the target S-A1c peak output to a constant value, fluctuations near the S-A1c peak can be more easily captured. Using this method, if the similarity value falls below a preset value, a message indicating a significant difference in chromatogram shape can be displayed, or the HPLC system can be shut down.

[0036] Next, the second embodiment will be described. In the second embodiment, the chromatogram is not normalized, and the similarity is evaluated after removing excessively large regions. The chromatogram is processed according to the following procedure and method (processing on the right side of Figure 8). 1) Offset processing 2) Only when the output after offset is equal to or greater than a certain value, the output is set to the certain value.

[0037] A chromatogram is composed of a combination of (time, output). First, the offset processing of 1) is omitted here because it is the same as the first mode described above. Next, we will explain the correction method 2). As mentioned above, in the chromatogram obtained by HbA1c measurement, the S-A1c peak to be quantified is small, and the A0 peak is about 90%, resulting in an unbalanced chromatogram pattern.

[0038] If the output after offset processing falls below a certain value (M), that output value is used. If the output after offset processing exceeds a certain value (M), the new output value becomes the value of M. With this process, the output near the apex of the predominant A0 peak in the final normalized chromatogram becomes M.

[0039] FIG. 7a is a schematic diagram of the chromatogram after offsetting, and FIG. 7c is a schematic diagram of the chromatogram after this processing.

[0040] Table 2 shows the final normalized chromatogram data. These are two chromatograms with slightly different chromatographic shapes: the "reference chromatogram" and the "verification chromatogram." In the table, column (1) is the time, column (2) is the output value of the "reference chromatogram" with the upper limit cut off, and column (3) is the output value of the "verification chromatogram" with the upper limit cut off.

[0041] The time of the peak top of S-A1c in the reference chromatogram is t7, the time of the peak top of S-A1c in the verification chromatogram is t9, and the A0 peak is around t21 to t28. In the chromatograms with the upper limit cut, the output becomes M around times t21 to t28 in both the reference chromatogram and the verification chromatogram.

[0042] [Table 2]

[0043] In this second embodiment, evaluation is performed using a chromatogram with the upper portion of the A0 peak cut off. If the components and concentrations of the standard samples are the same, a certain degree of similarity evaluation can be performed without normalizing the S-A1c peak. The upper limit of the output (M) is not limited to a value greater than the S-A1c peak output of a typical sample. A value 2 to 10 times the S-A1c peak is preferable. If the upper limit (M) is too large, fluctuations around the S-A1c peak cannot be captured. The method for evaluating the similarity of chromatograms is the same as in the first embodiment.

[0044] Figure 12 is a diagram showing a schematic diagram of changes in a chromatogram using the method of the present invention. Multiple normal distribution peak functions are synthesized and plotted. The number of column plates is gradually reduced by seven stages from the initial state. Figure 12a shows the chromatogram after offsetting, in which Peak_4 mimics the S-A1c peak and Peak_5 mimics the A0 peak. Figure 12b shows the first mode, in which Peak_4 is normalized by 30, and if the output exceeds 100 after correction, the output is set to 100. Figure 12c shows the second mode, in which if the output after offsetting simply exceeds 40, the output is set to 40. The right diagram shows changes in cosine distance (similarity).

[0045] In the right figure, #1 is the initial state, and #2 to #7 have the number of stages reduced. In the first mode of Fig. 12b, the output of Peak_4 remains the same even when the number of stages is reduced. In the second mode of Fig. 12c, the output of Peak_4 decreases as the number of stages is reduced. Regarding the cosine distance (similarity), in both modes, the value becomes smaller as the number of stages is reduced, meaning that it can be seen that the similarity decreases. [Effects of the Invention]

[0046] The present invention makes it possible to infer the characteristics of a sample from the estimated chromatogram pattern. [Brief explanation of the drawings]

[0047] [Figure 1] FIG. 1 is a diagram showing a method for calculating parameters that serve as indicators of separation in liquid chromatography. [Figure 2] This is a diagram showing the fluctuations in separation index and "plate number" in liquid chromatography. The left figure shows the case where the elution time is assumed to be 10 minutes, and the right figure shows the case where the elution time is assumed to be 1.0 minute. [Figure 3] These are diagrams showing the separation patterns in liquid chromatography: (a) isocratic elution, (b) linear gradient elution, and (c) step gradient elution. [Figure 4] FIG. 1 is a diagram schematically showing the characteristics of a chromatogram obtained by HbA1c measurement by liquid chromatography. [Figure 5] FIG. 1 is a diagram schematically showing the effect of offset on a chromatogram obtained in HbA1c measurement by liquid chromatography. [Figure 6] FIG. 1 is a diagram schematically showing the difference in offset between chromatograms obtained by HbA1c measurement by liquid chromatography. [Figure 7] FIG. 1 is a diagram schematically showing the normalization of a chromatogram obtained by HbA1c measurement by liquid chromatography, which is one of the features of the present invention. [Figure 8]1 is a flowchart showing the flow of normalization of a chromatogram obtained by HbA1c measurement by liquid chromatography, which is one of the features of the present invention. [Figure 9] 1 is a flowchart showing a first flow of determining the shape of a chromatogram obtained by HbA1c measurement by liquid chromatography, which is one of the features of the present invention. [Figure 10] 10 is a flowchart showing a second flow of determining the shape of a chromatogram obtained by HbA1c measurement by liquid chromatography, which is one of the features of the present invention. [Figure 11] 1A and 1B are diagrams showing a schematic diagram of the change in shape of a chromatogram obtained by HbA1c measurement by liquid chromatography, which is one of the features of the present invention, where FIG. 1A shows the first shape determination and FIG. 1B shows the second shape determination. [Figure 12] 1A and 1B are diagrams showing a schematic diagram of the change in shape of a chromatogram obtained by HbA1c measurement by liquid chromatography, which is one of the features of the present invention, where FIG. 1A shows the first shape determination and FIG. 1B shows the second shape determination. [Figure 13] FIG. 1 is a diagram showing the flow path configuration of GHbVIII manufactured by Tosoh Corporation, used in Examples 1 and 2. [Figure 14] FIG. 1 shows an example of the measurement results of GHbVIII (Tosoh Corporation) used in Examples 1 and 2. [Figure 15] 1 shows the procedure for processing chromatograms performed in Example 1. Panel a shows the chromatogram after offset processing, panel b shows the chromatogram after peak height (output) correction, and panel c shows the final chromatogram obtained. [Figure 16] FIG. 1 shows the flow of the chromatogram shape comparison carried out in Example 1. [Figure 17]These figures show chromatograms of a control sample (Level 1) measured to confirm column performance during continuous measurement of patient samples (whole blood) in Example 1. In the figures, the dashed line shows the chromatogram of the initial control sample, and the solid line shows the chromatogram of the control sample measured periodically. Figure a shows the chromatogram of the 1797th measurement, Figure b shows the 14113th measurement, Figure c shows the 19700th measurement, Figure d shows the 21341st measurement, Figure e shows the 22755th measurement, and Figure f shows the 23192nd measurement. [Figure 18] This figure shows a chromatogram after correction, which was performed to confirm changes in the chromatogram of the control sample (Level 1) in Example 1. In the figure, the dashed line shows the chromatogram of the control sample in the initial state, and the solid line shows the chromatogram of the control sample measured periodically. The S-A1c peak was set to 15, and if the result of dividing by the correction coefficient exceeded 100, the output value was set to 100. [Figure 19] These figures show changes in the chromatogram of the control sample in Example 1. Figure a shows the similarity (cosine distance) of the chromatogram of the control sample measured over time, when the chromatogram of the initial control sample (Level 1) was used as a reference. Figure b shows the change in the number of plates of the A0 peak. Figure c shows the fluctuations in the similarity (cosine distance) and number of plates up to 15,000 times. [Figure 20] These figures show the changes in the chromatogram of the control sample in Example 1. Figure a shows the similarity (cosine distance) of the chromatogram of the control sample measured over time, when the chromatogram of the initial control sample (Level 1) was used as a reference. The S-A1c peak was set to 15, and if the result of dividing by the correction coefficient exceeded 100, the output value was set to 100. [Figure 21] FIG. 1 is a diagram showing a chromatogram calculated in Example 1 with 19,700 measurements and a distance of 0.9183, in a normal print format. [Figure 22] FIG. 1 shows a step gradient for GHbVIII (manufactured by Tosoh Corporation) used in the Examples. [Figure 23] FIG. 10 is a diagram showing the flow of shape comparison in Example 1 when a chromatogram from a previous measurement is used as reference data. [Figure 24] These figures compare the results of Example 1 when the initial chromatogram was used as the reference data and when the chromatogram from the previous measurement was used as the reference data. Figure a shows the initial chromatogram, and Figure b shows the chromatogram from the previous measurement as the reference data. In both cases, the left figure shows a chromatogram that has been subjected to offset processing only (the lower figure is enlarged), and the right figure shows a chromatogram that has been subjected to normalization processing to evaluate similarity. [Figure 25] FIG. 1 shows the change (similarity) of a chromatogram when the chromatogram of the previous measurement is used as reference data in Example 1. [Figure 26] FIG. 1 shows the flow of chromatogram shape comparison performed in Example 2. [Figure 27] These figures show chromatograms of a control sample (Level 1) measured to confirm column performance during continuous measurement of patient samples (whole blood) in Example 2. In the figures, the dashed line shows the chromatogram of the initial control sample, and the solid line shows the chromatogram of the control sample measured periodically. Figure a shows the chromatogram of the 1797th measurement, Figure b shows the 14113th measurement, Figure c shows the 19700th measurement, Figure d shows the 21341st measurement, Figure e shows the 22755th measurement, and Figure f shows the 23192nd measurement. [Figure 28] FIG. 10 is a diagram showing the changes (similarity) in the chromatogram of the control sample in Example 2. [Figure 29] FIG. 1 shows changes in the chromatogram of a control sample in Example 1. [Figure 30] FIG. 1 is a diagram showing the application of the method of the present invention to size exclusion chromatography (SEC), illustrating an example of separation of a general standard sample. [Figure 31] 1 is a diagram showing a schematic diagram of the fluctuation of a chromatogram when the method of the present invention is applied to size exclusion chromatography (SEC), where FIG. 1 shows the fluctuation of the chromatogram and FIG. 1 b shows the change in column performance. [Figure 32] FIG. 1 is a diagram showing the application of the method of the present invention to ion chromatography, and is an example of separation of a standard sample in general anion analysis. [Figure 33] These are diagrams showing the variations in chromatograms when the method of the present invention is applied to ion chromatography. Figure a shows a normal separation pattern, Figure b shows a case where a specific component is contaminated, and Figure c shows a case where an unknown component is contaminated. [Figure 34] 1 is a diagram showing a schematic diagram of the fluctuation of a chromatogram when the method of the present invention is applied to ion chromatography, where FIG. 1A shows the fluctuation of the chromatogram and FIG. 1B shows the change in column performance. [Figure 35] FIG. 1 is a schematic diagram showing flow paths for size exclusion chromatography and ion chromatography. DETAILED DESCRIPTION OF THE INVENTION

[0048] Examples of the present invention will be described below, but the present invention is not limited to these examples in any way.

[0049] To verify the effectiveness of the present invention, the following verification was performed using a glycohemoglobin analyzer GHbVIII manufactured by Tosoh Corporation. This device is an HbA1c measurement device based on the principle of liquid chromatography and has the configuration shown in Figure 13. A patient sample (whole blood) (10) is aspirated using a sampling needle (9) and hemolyzed / diluted with a hemolyzing / diluting solution (2) at a dilution port (8). The sample is then injected into an analytical column (12) using a sample injection mechanism (6), separated using multiple eluents (1), and detected by a detector (13). When measuring a standard sample, the lyophilized product is manually diluted, placed in a rack (11), and then directly injected into the analytical column for measurement. A line filter (15) is installed before the analytical column to prevent deterioration of the analytical column.

[0050] The separation mode was "standard mode," and the analytical column, eluent, and hemolysis / washing solution were all proprietary. A control sample for the glycohemoglobin analyzer provided by Tosoh Corporation was used as the standard sample. The control sample was dissolved in 0.5 mL of purified water for the primary dissolution, and then diluted 1 / 51 with the dedicated diluent for the secondary dissolution before measurement. Figure 14 shows an example of a standard measurement result. The measurement time was 1.00 minutes, and six fraction peaks were obtained: A1A, A1B, F, L-A1c, S-A1c, and A0. "S-A1c" was the fraction peak for quantitative analysis.

[0051] As shown in Figure 16, chromatograms were obtained by measuring a control sample on a new analytical column, then measuring 200 to 800 patient samples (whole blood), then re-measuring the control sample, and repeating this process until a final total of 23,000 samples was measured, and changes in the shape of the chromatograms of the control samples were measured. If an increase in column pressure was observed, the filter was replaced as appropriate. Example 1 shows the results of measuring changes in the shape of the chromatogram after normalizing the chromatogram, which is a first embodiment of the present invention. Example 2 shows the results of measuring changes in the shape of the chromatogram without normalizing the chromatogram, which is a second embodiment of the present invention.

[0052] Example 1 According to the first embodiment of the present invention, chromatogram normalization was performed, and the change in the shape of the chromatogram of the control sample was measured. The procedure for normalizing the chromatogram was as follows. 1) Obtain a chromatogram 2) Offset processing (output at time zero is set to zero) 3) Identify the S-A1c peak and obtain the peak height (output value) (S) 4) Calculate the correction factor (f) as follows: Specified value (H) / S-A1c peak height (S) (Specified value (H) was set to 15.) 5) Multiply each output value in 2) by the correction factor (f) Corrected output <100: Use the value of 5) as the output value. Corrected output >=100: Set the output value to 100.

[0053] An example of calculation is shown in Table 3. Here, the chromatogram of a control sample (level 1) obtained at the time of column renewal is used as the reference, and the chromatogram of a control sample (level 1) obtained after measuring whole blood samples approximately 700 times is used as the verification data.

[0054] Column (1) shows the time (minutes), column (2) shows the output of the reference data, column (3) shows the output of the validation data, column (4) shows the output of the reference data after offset processing, column (5) shows the output of the validation data after offset processing, column (6) shows the output of the reference data after normalization processing, and column (7) shows the output of the validation data after normalization processing.

[0055] Figure 15 shows an example of how the chromatogram changes after processing. (4) is the value obtained by subtracting 99.90 from (2), and (5) is the value obtained by subtracting 95.10 from (3). Figure 15a shows the chromatogram after offset processing.

[0056] The elution time of S-A1c in the reference data is 0.5600 minutes, and the output is 13.94 (shaded area). The correction coefficient is 15.00 / 13.94 = 1.076. The elution time of S-A1c in the validation data is 0.5650 minutes, and the output is 13.58 (shaded area). The correction coefficient is 15.00 / 13.58 = 1.105.

[0057] (6) is the value obtained by multiplying (4) by a correction coefficient of 1.076, and any data exceeding 100 is set to 100 (shaded area). (7) is the value obtained by multiplying (5) by a correction coefficient of 1.105, and any data exceeding 100 is set to 100 (shaded area). Figure 15b shows a chromatogram obtained by simply multiplying the correction coefficient, and Figure 15c shows the chromatogram after normalization, with the output value set to 100.

[0058] Column (8) shows the product of the output after normalization of the reference data and the output after normalization of the validation data, column (9) shows the square of the output after normalization of the reference data, and column (10) shows the square of the output after normalization of the validation data. The bottom rows of columns (8), (9), and (10) show the sums of each. The cosine distance (similarity) is calculated using equation 3. In this case, the cosine distance is calculated as 0.9945.

[0059]

number

[0060] [Table 3]

[0061] Figure 17 shows the change in the chromatogram (data obtained only after offset processing) of a control sample (level 1) measured between whole blood sample measurements. Figure 18 shows the change in the chromatogram (data obtained after the normalization processing of the present invention) of a control sample (level 1) measured between whole blood sample measurements. In both chromatograms, the dashed line indicates the initial chromatogram, and the solid line indicates the verified chromatogram.

[0062] Figures 17a and 18a show chromatograms of the control sample (Level 1) after 1,797 whole blood runs, Figures 17b and 18b after 14,113 whole blood runs, Figures 17c and 18c after 19,700 whole blood runs, Figures 17d and 18d after 21,341 whole blood runs, Figures 17e and 18e after 22,755 whole blood runs, and Figures 17f and 18f after 23,192 whole blood runs. There are no differences in the chromatogram shapes between Figures 17a, 18a, 17b, and 18b. The S-A1c peak has become slightly broader in Figures 17c and 18c. The S-A1c peak elutes earlier in Figures 17d and 18d, revealing a shoulder. In Figures 17e, 18e, 17f, and 18f, the A0 peak becomes bimodal. Figures 17c and 18c onward can be seen as an abnormality in the shape of the chromatogram. However, in normal results displays (printouts of measurement results), the chromatogram is generally reduced in size (especially the time axis), making it difficult to identify Figures 17c and 18c as abnormal (Figure 21 shows the chromatogram in Figure 17c in a format that matches the normal results display (printouts of measurement results)).

[0063] When the method of the present invention was used to determine the chromatogram shape based on the distance (similarity) to the reference chromatogram, the results were 0.9774 for Figures 17a and 18a, 0.9564 for Figures 17b and 18b, 0.9183 for Figures 17c and 18c, 0.8680 for Figures 17d and 18d, 0.8301 for Figures 17e and 18e, and 0.81824 for Figures 17f and 18f, allowing the chromatogram shapes to be expressed numerically. Table 4 shows some of the distances (similarity) between the periodically acquired validation chromatograms (Level 1 and Level 2) and the reference chromatograms.

[0064] Figure 19a is a plot of the number of measurements of a whole blood sample on the horizontal axis and the distance (similarity) on the vertical axis (Level 1). Figure 19b is a plot of the number of measurements of a whole blood sample on the horizontal axis and the number of steps of the A0 peak on the vertical axis. As shown in Figure 19a, when the distance (similarity) is taken on the vertical axis, a monotonically decreasing sigmoid function-like curve number 4 is obtained, and the change in the shape of the chromatogram can be expressed.

[0065]

number

[0066] When the vertical axis is the number of steps of the A0 peak, the graph also resembles a monotonically decreasing sigmoid function, but there is a large amount of variation. Furthermore, when the number of measurements is 15,000 or more, there are many points that deviate significantly. This is because the shape of the A0 peak becomes multi-modal, making it impossible to accurately calculate the number of steps.

[0067] The GHbVIII manufactured by Tosoh Corporation used in this verification performs separation by switching between three types of eluents (Buffer_1, 2, 3). A1A-A1B, etc. are separated with Buffer_1, F-S-A1c with Buffer_2, and A0 is eluted with Buffer_3 (see Figure 22). Buffer switching is controlled by time. This is because any change in the actual flow rate over time or suddenly can have a significant impact on the separation and, as a result, on the calculated plate number value.

[0068] Figure 19c shows the variability (CV%) up to 15,000 runs, where the shapes appear relatively consistent. This also demonstrates that the method of the present invention more accurately represents the chromatogram shape numerically, with a variation in distance (similarity) that is half to one-third smaller than the number of plates, demonstrating the superiority of this method. With this measurement system, it can be inferred that a calculated distance (similarity) of 0.94 to 0.95 can be used as a threshold to determine the quality of a chromatogram. Figure 20 shows a verification chromatogram (control level 1) obtained from an actual sample overlaid over the chromatogram obtained after 23,000 runs. The evolution of the S-A1c and A0 peaks can also be seen. From 10,000 to 15,000 samples, the chromatogram is highly similar to the reference chromatogram. After that, the elution time of the S-A1c peak becomes earlier and more multimodal, and the rear portion of the A0 peak becomes more multimodal.

[0069] [Table 4]

[0070] Up to this point, we have shown an example in which the chromatogram of the control sample (level 1) at the time of column update was used as the reference, but the reference data is not limited to the initial state. For example, one or more previous chromatograms of the validation data may be used as the reference (see Figure 23). An example is briefly shown below. The chromatogram uses the data group described above.

[0071] The cosine distance was calculated using the chromatogram of a control sample (Level 1) after approximately 12,000 measurements of a whole blood sample as verification data, and the previous data (after approximately 1,100 measurements of a whole blood sample) as reference data. Table 5 shows only the calculated values. For comparison, the top row shows the case where the initial data was used as the reference (method described above), and the bottom row shows the case where the previous data was used as the reference.

[0072] When the initial data is used as the reference, the cosine distance (similarity) is calculated to be 0.9804, and when the previous data is used as the reference, the cosine distance (similarity) (D) is calculated to be 0.9328.

[0073] Figure 24 shows chromatograms when the initial data is used as the reference and when the previous data is used as the reference. In both cases, the left figure shows the chromatogram after offset processing, and the right figure shows the chromatogram after normalization processing used to calculate the distance. From the figure, it appears that the validation data is more similar to the previous value (Figure 24b). However, the distance compared to the initial chromatogram is 0.9804, closer to 1.000, and is judged to be more similar. Enlarging the chromatogram after offset processing in the left figure shows that the elution time of the A0 peak differs significantly from the previous chromatogram. However, there is little difference in the shape of the A0 peak compared to the initial chromatogram. As a result, when comparing with the previous value, the distance is calculated to be small at 0.9328, and the similarity is judged to be low.

[0074] Figure 25 shows the cosine distance (similarity) calculated for all the chromatograms used above, using the previous values as reference data. As can be seen, the cosine distance is close to 1.000. This is because even if the shape of the chromatogram changes over time, the similarity to the previous chromatogram remains almost the same.

[0075] However, data like that in Table 5 (approximately 12,000 measurements of whole blood samples) shows extremely low calculated values of 0.95 or less. The circled points in Figure 25 are examples of this. This suggests that rather than capturing changes over time in the column, eluent, etc., a sudden abnormality has occurred. For example, it is conceivable that air was suddenly trapped in the liquid delivery pump, causing a drop in the actual flow rate for that measurement only.

[0076] In this way, to capture changes over time in the column, eluent, etc., it is preferable to use the initial chromatogram as reference data, and to capture sudden abnormalities, it is preferable to use the previous chromatogram as reference data.

[0077] [Table 5]

[0078] Example 2 In the second embodiment of the present invention, the chromatogram normalization was not performed, and the change in shape of the chromatogram of the control sample was measured. The chromatogram processing procedure was as follows. 1) Obtain a chromatogram 2) Offset processing (output at time zero is set to zero) 3) Compare the output value with the specified value Output value <40: Use the value of 2) as the output value. Corrected output >=40: Set the output value to 40.

[0079] Figure 26 shows the changes in the chromatogram due to the above processing. Table 6 shows an example of calculation. Here, the chromatogram of the control sample (level 1) at the time of column renewal is used as the reference, and the chromatogram of the control sample (level 1) obtained after measuring whole blood samples approximately 700 times is used as the verification data.

[0080] (1) Column is time (minutes), (2) Column is reference data output, (3) Column is validation data output, (4) Column is reference data output after offset processing, (5) Column is validation data output after offset processing, (6) Column is reference data output after upper limit cut processing, (7) Column is validation data output after upper limit cut processing, (4) is the value obtained by subtracting 99.90 from (2), and (5) is the value obtained by subtracting 95.10 from (3).

[0081] The elution time of S-A1c in the reference data is 0.5600 minutes, and the output is 13.94 (shaded area). The correction coefficient is 15.00 / 13.94 = 1.076. The elution time of S-A1c in the validation data is 0.5650 minutes, and the output is 13.58 (shaded area). The correction coefficient is 15.00 / 13.58 = 1.105.

[0082] (6) is the value obtained by multiplying (4) by a correction coefficient of 1.076, and any data exceeding 100 is set to 100 (shaded area). (7) is the value obtained by multiplying (5) by a correction coefficient of 1.105, and any data exceeding 100 is set to 100 (shaded area).

[0083] Column (8) shows the product of the output after normalization of the reference data and the output after normalization of the validation data, column (9) shows the square of the output after normalization of the reference data, and column (10) shows the square of the output after normalization of the validation data.

[0084] The bottom rows of columns (8), (9), and (10) show the totals for each. The cosine distance (similarity) is calculated using equation 5. In this case, the cosine distance is calculated as 0.99212.

[0085]

number

[0086] [Table 6]

[0087] Figure 27 shows the changes in the chromatogram (data after normalization according to the present invention) of a control sample (Level 1) measured between measurements of whole blood samples. In both chromatograms, the dashed line indicates the initial chromatogram, and the solid line indicates the verified chromatogram.

[0088] Figure 27a shows the chromatograms of the control sample (Level 1) after 1,797 whole blood measurements, Figure 27b after 14,113 whole blood measurements, Figure 27c after 19,700 whole blood measurements, Figure 27d after 21,341 whole blood measurements, Figure 27e after 22,755 whole blood measurements, and Figure 27f after 23,192 whole blood measurements. Figures 27a and 27b show no differences in chromatogram shape. In Figure 27c, the S-A1c peak has become slightly broader. In Figure 27d, the elution of the S-A1c peak has accelerated, revealing a shoulder. In Figures 27e and 27f, the A0 peak has become bimodal. Figures 27c and after can be seen as abnormalities in the chromatogram shape. However, in normal result displays (printouts of measurement results), the chromatogram is generally reduced in size (especially the time axis), making it difficult to identify Figure 27c as abnormal (Figure 21 displays the chromatogram of Figure 17c in a format that matches normal result displays (printouts of measurement results)).

[0089] When the method of the present invention is used to determine the shape of a chromatogram based on the distance (similarity) to a reference chromatogram, the values calculated are 0.9809 for Figure 27a, 0.9703 for Figure 27b, 0.9515 for Figure 27c, 0.8801 for Figure 27d, 0.8450 for Figure 27e, and 8361 for Figure 27f, which allows the shape of the chromatogram to be expressed numerically.

[0090] Figure 28 is a plot of the number of measurements of a whole blood sample on the horizontal axis and the distance (similarity) on the vertical axis. When the distance (similarity) is plotted on the vertical axis, a monotonically decreasing sigmoid function-like curve number 6 is obtained, which can represent the change in the shape of the chromatogram.

[0091]

number

[0092] In this measurement system, it can be estimated that the quality of a chromatogram can be determined if the calculated distance (similarity) is set at a threshold of approximately 0.95.

[0093] Figure 29 shows the verification chromatogram (control level 1) obtained from the actual sample over 23,000 runs. The changes in the S-A1c and A0 peaks can be seen here as well. From 10,000 to 15,000 samples, the chromatogram is highly similar to the reference chromatogram, but after that, the elution time of the S-A1c peak becomes earlier and more multi-peaked, and the latter part of the A0 peak also becomes more multi-peaked.

[0094] [Table 7]

[0095] Example 3 The present invention is particularly effective for chromatograms of glycohemoglobin, but similar effects can be obtained with other HPLC methods, and the present invention is not limited to chromatograms of glycohemoglobin. For example, it can also be applied to size exclusion chromatography (SEC). SEC is one type of liquid chromatography and has a configuration as shown in Figure 35.

[0096] In the case of SEC, low molecular weight compounds such as dicyclohexyl phthalate (DCHP) are often used to calculate the plate number, which is an index used to evaluate performance. Because this is a single component, the peak width is very narrow, making it unsuitable for routine monitoring of column deterioration.

[0097] To monitor column deterioration on a daily basis, changes in separation performance can be accurately captured by using a relatively low molecular weight oligomer, which is used when creating a calibration curve. Figure 30 shows the molecular weight marker "A-500" (polystyrene, Mw: 5.89 x 10) manufactured by Tosoh Corporation. 2 This is an example of separation of a compound (Mw / Mn 1.19) on TSKgel SuperHz (3000 + 3000 + 2000 + 2000) (6.0 mm I.D. x 15 cm x 4 columns). As shown, approximately five fractions are separated according to their degree of polymerization. Using such samples for daily performance evaluations allows for clear monitoring of column deterioration. In the case of SEC, column deterioration leads to a decrease in the number of plates, resulting in a decline in resolution. Furthermore, elution times can fluctuate if the column is not delivered accurately due to an abnormality in the equipment, particularly the delivery pump.

[0098] To simulate the column deterioration process during SEC, we created and verified pseudo-chromatograms by combining multiple normal distribution peak functions to produce a separation pattern similar to that of the A-500. Figure 31 shows six pseudo-chromatograms. The thick line in the figure represents the reference chromatogram, and these are examples in which the number of plates was reduced by 10%. As the number of plates decreases, the resolution also decreases. Figure 30b shows the similarity values relative to the initial chromatogram. As this shows, it is possible to determine the degree of deterioration even with SEC.

[0099] Example 4 Furthermore, it can also be applied to ion chromatography, which is a type of liquid chromatography and has a configuration as shown in Figure 35.

[0100] Example: In ion chromatography, when preparing a calibration curve, a multi-component mixed reagent supplied by the reagent manufacturer is often diluted 1:1. Standard samples with preset dilution ratios can be used to monitor column deterioration. Column deterioration in ion chromatography can result in tailing, leading, or peak splitting. Furthermore, elution times can fluctuate if the instrument malfunctions, particularly the delivery pump, and the solution is not delivered accurately. Figure 32 shows an example of separation of an anion mixed standard solution (manufactured by Fujifilm Wako Co., Ltd.) on a Tosoh Corporation TSKgel SuperIC-AZ (4.6 mm I.D. x 15 cm) column. Ion chromatography is prone to contamination from the environment and manual handling of instruments, which often poses problems. Figure 33 shows a schematic pseudo-chromatogram of a sample with contamination. Figure 33a shows a normal standard sample mixture (7 components), Figure 33b shows a sample with specific component contamination, and Figure 33c shows a sample with unknown component contamination. In ion chromatography, chlorine contamination is likely to occur from the equipment or operator, resulting in a pattern like that shown in Figure 33b. Furthermore, contamination from the environment often results in a pattern like that shown in Figure 33c. This method is useful for determining whether or not such contamination has occurred.

[0101] As an example, to simulate the deterioration process of a column in ion chromatography, we created and verified pseudo-chromatograms by combining multiple normal distribution peak functions to form a standard anion separation pattern. Figure 34 shows seven pseudo-chromatograms. The thick line in the figure represents the reference chromatogram, and these are examples in which the number of plates was reduced by 10%. As the number of plates is reduced, the peak width broadens, and the resolution of peaks 6-7, which mimic divalent ions, also decreases. Figure 34b shows the similarity values relative to the initial chromatogram. This demonstrates that the degree of deterioration can be determined even with ion chromatography.

[0102] As can be seen from Examples 1 and 2, when samples of the same type and concentration are always used, it is possible to capture fluctuations in the shape of the chromatogram to a certain extent without normalizing the chromatogram. However, when freeze-dried samples are used, as in the case of a glycohemoglobin analyzer, accurate concentration adjustment can be difficult, and normalizing the chromatogram as in Example 1 is preferable because it allows for more accurate discrimination. [Explanation of symbols]

[0103] 1. Eluent 2. Hemolysis / Dilution Solution 3. Degassing device 4. Flow path switching mechanism 5. Liquid transfer pump 6. Sample injection mechanism 7. Measuring syringe 8. Dilution port 9. Sampling Needle 10. Specimen 11. Sample rack 12.Analytical column 13. Detector 14.Thermostat 15. Line filter 16. Drain 17. Eluent A 18. Eluent B (SEC only) 19. Degassing device A 20. Degasser B (SEC only) 21. Liquid transfer pump (sample side) 22. Liquid delivery pump (reference side, SEC only) 23.Sample injection mechanism 24. Sampling needle 25. Specimen 26.Analytical column 27. Reference analytical column (SEC only) 28. Detectors for SEC (refractometers, etc.) 29. Detectors for ion chromatography (conductivity meters, etc.) 30. Constant temperature bath 31. Suppressor (Ion Chromatography Only)

Claims

1. A method for determining the similarity of a chromatogram to be verified with a reference chromatogram that serves as a basis for comparison, using a standard sample consisting of multiple components in liquid chromatography, the method comprising the steps of: 1) Perform offset processing so that the output at the start point (time zero) of the two chromatograms is zero, 2) Among the multiple components constituting the standard sample, the standard value (H) of the first component that serves as the reference and the upper limit value (M) for the entire chromatogram are set in advance; multiplying the output of each time of the two chromatograms by the ratio (H / S) of the output (S) after offset of the first component detected in the two chromatograms to the normalized value (H) of the first component; if the result exceeds the upper limit (M), the new output value is set to the upper limit (M); if the result is equal to or less than the upper limit (M), the value obtained by the multiplication is set to the new output value; and obtaining a normalized chromatogram; A method for assessing the similarity between reference and validation chromatograms.

2. A method for determining the similarity of a chromatogram to be verified with a reference chromatogram that serves as the basis for comparison, using a standard sample consisting of multiple components in liquid chromatography, comprising: Before the determination, the two chromatograms are subjected to the following processing: 1) Perform offset processing so that the output at the start point (time zero) of the two chromatograms is zero, 2) An upper limit value (M) for the entire chromatogram is set in advance, and if the output (S) after the offset exceeds the upper limit value (M), the new output value is set to the upper limit value (M), and if the output (S) is equal to or less than the upper limit value (M), the output value after the offset is set to the new output value, and a corrected chromatogram is obtained; A method for assessing the similarity between reference and validation chromatograms.

3. A method for determining the similarity of the two chromatograms by calculating a cosine distance from each time and output of the normalized reference chromatogram and verification chromatogram obtained as claimed in claim 1.

4. A method for determining the similarity of the two chromatograms by calculating a cosine distance from each time and output of the normalized reference chromatogram and verification chromatogram obtained as claimed in claim 2.

5. A method for determining the similarity of chromatograms obtained by using a standard sample consisting of two or more components in liquid chromatography, using the method of claim 3.

6. A method for determining the similarity of chromatograms obtained by using a standard sample consisting of two or more components in liquid chromatography, using the technique of claim 4.

7. A method for determining the similarity of chromatograms in measuring glycohemoglobin (HbA1c) using liquid chromatography, using the technique of claim 3, and using the S-A1c peak as the first component.

8. A method for determining the similarity of chromatograms in measuring glycohemoglobin (HbA1c) using liquid chromatography, using the technique of claim 4, with the S-A1c peak as the first component.

Images