JP4131744B2 - Discrimination of bacterial infections of platelet preparations using difference spectra - Google Patents

Description

  The present invention relates to a platelet preparation non-invasive test method for non-invasively detecting the presence or absence of bacterial contamination of a platelet preparation by near infrared spectroscopy, and a platelet preparation whose safety has been confirmed by the method.

As a method for measuring bacterial infection of a platelet preparation, there are a bacterial culture method (Non-patent document 1, Non-patent document 2) and a bacterial scan method (Non-patent document 3).
The former applies the principle that the oxygen pressure of the monitor decreases or the carbon dioxide pressure increases as the bacteria grow, and the latter stains the bacteria and scans them with a microscope.

  Patent Documents 1 to 6 are known as prior arts related to blood analysis.

  In Patent Document 1, a blood collection tube or blood collection bag containing a blood sample adjusted to a predetermined temperature is irradiated with near-infrared light separated in the range of 700 nm to 1100 nm from the outside to measure the transmitted light intensity, and then to the wavelength. A method of extracting information on chemical components and physicochemical properties using a calibration curve from a near-infrared absorption spectrum in which the absorbance is plotted against is disclosed.

  In Patent Document 2, as a method of determining whether or not a plasma sample representing blood plasma in a blood bag contains a light scattering substance interfering substance such as hemoglobin (Hb), bile pigment, or lipid particles, Irradiate light with a wavelength of 475-910 nm from the outside, measure the absorbance of light of different wavelengths of the plasma sample contained in the blood bag, and then calibrate this absorbance using standard measurements for interferents in the plasma sample A method for determining whether or not a light-scattering substance interferer is included by comparing with the obtained value has been proposed.

  Patent Document 3 proposes an optical measurement method for nondestructively measuring the survival rate of stored platelets in a platelet preparation stored in a blood bag. Specifically, the blood bag is placed in a horizontal state and sandwiched between flat plates to form a measurement platelet preparation layer that is a flow channel having a predetermined thickness, and the blood bag is divided into two reservoirs via this layer. Irradiate orthogonal light rays in a stationary state, measure the transmitted light of the platelet preparation layer for measurement to obtain a static light transmission intensity, and perform the same blood bag swinging at a predetermined flow rate as described above. A dynamic light transmission intensity is obtained by the measurement operation, and a rate of change of the light transmission intensity, which is a ratio of the difference value between the dynamic light transmission intensity and the static light transmission intensity, and the static light transmission intensity is calculated. The platelet ratio is calculated from the correlation coefficient between the change rate and% DISK indicating the normal platelet ratio in the blood bag. In addition, it is described that near-infrared light having a wavelength of 950 nm is used as the wavelength of light to be irradiated.

  Patent Document 4 proposes a method for comparing an unknown value with a calibrated reference value using a chemometric technique using multispectral analysis as a method for noninvasively determining the concentration of a blood analyte. ing. In this document, principal component analysis (PCA) is mentioned as a chemometric technique, and there is a significant change in the reflection intensity difference particularly in the range of the wavelength of the measuring light from about 2120 to 2180 nm, and this difference in reflection intensity increases the blood glucose level. And increase in a linear relationship.

  In Patent Document 5, two different light beams (wavelengths of 770 nm to 950 nm and wavelengths of 480 nm to 590 nm) are used as a method for noninvasively measuring the characteristics of blood in a blood vessel containing a mixture of liquid and blood cells. It is described that a mixture of liquid and blood cells contained in a human blood vessel is noninvasively measured by analyzing the ratio of the detected intensity of light.

  Patent Document 6 describes near-infrared spectroscopic analysis of a blood vessel wall to which a chemometric method suitable for predicting characteristics such as the chemical composition of a structure is applied based on the spectral response of the structure. Specifically, as a chemometrics method for suppressing unnecessary signals in the collected spectrum, specifically, partial least square discriminant analysis (PLS-DA), Mahalanobis distance and large residual (augmented Residual) Principal component analysis method (PCA / MDR) used, Principal component analysis method using K-nearest neighbor, Principal component analysis method using Euclidean distance, soft independent modeling by class analogy (SIMCA), boot A bootstrap error-adjusted single-sample technique (BEST) is mentioned.

JP 2002-122537 A JP-T-2001-514744 JP 9-318624 A JP 2006-126219 A Special table 2003-508765 gazette JP 2005-534415 A Ortolano J, et al. Detection of bacteria in WBC-reduced PLT concentrates using percent oxygen as a marker for bacteria growth.Transfusion 43 (9): 1276-1285, 2003. Brecher M E, et al. Monitoring of apheresis platelet bacterialcontamination with an automated liquid culture system: a university experience.Transfusion 43: 974-978 (7), 2003. Schmidt M. et al: A comparison of three rapid bacterial detection methods under simulated real-life conditions.Transfusion 46 (8): 1367-1373, 2006.

  Patent Documents 1 to 5 disclose non-invasive measurement of the near-infrared spectrum of a sample contained in a blood collection bag. Further, as disclosed in Patent Document 4, it is known to compare with a reference value in order to measure the concentration of a specific substance in a sample. In addition, chemometric methods, that is, mathematical or statistical techniques are generally used in spectrum analysis.

However, regarding the measurement of platelets, there is a problem that cannot be solved by the general method described above.
That is, the manufactured platelet preparations do not all have the same composition, and there are slight differences in chemical and physical properties (individual differences). This slight difference greatly fluctuates the spectrum of the platelet preparation. For this reason, in order to capture minute changes in the components in the platelet preparation due to the presence or absence of bacterial infection, it is necessary to reduce the influence of individual differences in the platelet preparation, but the above-mentioned prior art cannot solve this problem.

  The present invention relates to a platelet preparation non-invasive test method for non-invasive and rapid detection of bacterial infection of a platelet preparation without taking out a sample for examination of the platelet preparation from a blood collection bag. Analytical methods for clearly identifying

  In order to achieve the above object, the present invention provides a near-infrared spectrum of a platelet preparation contained in a blood collection bag over time from immediately after production of the platelet preparation by the transmission method or interactance method without removing the platelet preparation from the blood collection bag. Measure and analyze the time course of the spectrum of the bacterially infected sample and the healthy sample by spectroscopic techniques, and compare the time course of the spectrum of the unknown sample with that of the bacterially infected sample and that of the healthy sample. In the platelet preparation non-invasive test method for identifying the presence or absence of the platelet preparation, the spectrum used for the analysis was a difference spectrum based on the spectrum first measured after the preparation of the platelet preparation.

The method of determining the presence or absence of bacterial infection of the unknown sample is based on setting the degree of infection of a healthy sample to 0 (zero),
There is a method of discriminating from the estimated value of the degree of infection obtained by applying a regression model created by PLS regression with a bacterial infection sample being 1 to an unknown sample. When the estimated value of the infection level is less than a certain value (for example, 0.4), the unknown sample is determined as a healthy sample, and when the estimated value is equal to or greater than the certain value, the unknown sample is determined as a bacterial infection sample.

  Since the blood collection bag is difficult to transmit light, using near-infrared rays in the wavelength band normally used (about 2500 nm) causes large noise and tends to cause measurement errors. Therefore, it is preferable to use near-infrared light having a short wavelength range of 700 nm to 1100 nm having a transmission power 10 to 100 times that of a normally used wavelength band.

  In addition, since the spectrum of platelet preparations varies greatly depending on the chemical and physical characteristics (individual differences) of platelet preparations, in order to capture minute changes in the components of platelet preparations due to the presence or absence of bacterial infection, In order to reduce the influence of individual differences, the spectrum used for the analysis is a difference spectrum based on the spectrum first measured after the manufacture of the platelet preparation. Prior to spectrum analysis, the difference spectrum can be preprocessed by any one or combination of moving average, differentiation, MSC processing, and the like.

  According to the present invention, it is possible to eliminate an individual difference for each blood collection bag and perform an accurate measurement when non-invasively examining the presence or absence of bacterial infection of the platelet preparation contained in the blood collection bag. Therefore, for example, the period of use of the platelet preparation can be extended from 3 days to one week, and effective use of the platelet preparation that has been discarded after expiration can be achieved while ensuring safety.

  The best mode for carrying out the present invention will be described below. This inspection method includes (1) spectrum measurement, (2) difference spectrum calculation, and (3) identification by spectrum analysis.

  (1) In the spectrum measurement process, the spectrum (sample thickness: 15 mm) of the platelet preparation contained in the blood collection bag stored at a constant temperature (for example, 24 ° C.) while being oscillated by a constant temperature shaker is shown in FIG. It is measured by the transmission method on the left or the interactance method on the right.

  That is, in the transmission method, in the sample chamber 2, the dispersed light 3 is irradiated to the platelet preparation 1 placed in the bag through the incident light optical fiber bundle 5, and the light 4 transmitted through the platelet preparation 1 is transmitted. It is detected by a detector through the optical fiber bundle 6 for light.

  In the interactance method, the dispersed light 8 is irradiated to the platelet preparation 7 via the interactance optical fiber bundle 10, and the light 9 diffusely reflected inside the platelet preparation 7 is detected via the interactance optical fiber bundle 10. Detected by the instrument.

  In the actual measurement, it is not necessary to measure the spectrum by both the transmission method and the interactance method, and the measurement may be performed by either method.

  After the spectrum measurement, return the sample to a constant temperature shaker for storage. This measurement is repeated for 3 days at regular time intervals, for example, every 6 hours. In the case of actual measurement, it is possible to continue the measurement for one week.

  (2) In the process of calculating the difference spectrum, in order to remove the effects caused by the difference in chemical and physical properties (individual differences) of the initial (during manufacture) platelet product from the spectrum, The difference spectrum is calculated by subtracting the spectrum for each sample. For the calculation of the difference spectrum, a spectrum that has been preprocessed (moving average, differentiation, MSC processing, etc.) may be used.

  FIG. 2 shows an example of the second derivative spectrum. The strong absorption observed in the vicinity of 970 nm is due to water, and the difference in intensity is due to individual differences. The results of principal component analysis using this second derivative spectrum are shown in FIG. In the figure, ◆ are samples from healthy areas (target areas), ▲ are samples from Serratia bacteria-infected areas, and ■ are samples from Staphylococcus epidermidis-infected areas. In the plane composed of the axes of PC1 and PC2, the respective samples overlap, and the healthy sample and the infected sample cannot be distinguished. Since the spectrum change due to bacterial infection is very small, it is considered that the spectrum was buried among individual differences in the sample.

  The method of determining the presence or absence of bacterial infection in the unknown sample is to set the infection degree of the healthy sample to 0 (zero) and subtract the second derivative value of the spectrum first measured from the second derivative value of the spectrum measured over time. The example of the difference spectrum obtained by this is shown in FIG. In the difference spectrum, the shift of the spectrum due to the difference in chemical and physical characteristics (individual difference) of each platelet preparation is removed, and the subtle component change due to bacterial infection is emphasized.

  The result of principal component analysis using this difference spectrum is shown in FIG. In the figure, ◆ are samples from healthy areas (target areas), ▲ are samples from Serratia bacteria-infected areas, and ■ are samples from Staphylococcus epidermidis-infected areas. In the plane composed of the PC1 and PC2 axes, the sample 11 in the healthy area (target area) is concentrated on the left side of the scatter diagram, and the sample 12 in the Serratia bacteria-infected area and the sample 13 in the Staphylococcus epidermidis-infected area are samples in the healthy area. Widely distributed on the right side. Samples far from the distribution area of samples in the healthy area can be excluded as bacterial infections.

Next, an example of analysis using PLS regression is shown. A calibration model for estimating the degree of infection is created by performing PLS regression based on the difference spectrum in a calibration model creation sample in which the infection degree of a healthy sample is 0 (zero) and that of a bacterial infection sample is 1. Apply the difference spectrum of the sample for evaluation with known infectivity to the created calibration model and confirm the performance of the model.

  FIG. 6 shows a scatter diagram at the time of calibration model evaluation. In the scatter diagram, when the estimated value of the infection level is less than 0.4, the sample is determined to be a healthy sample, and when the estimated value of the infection level is 0.4 or more, the sample is determined to be a bacterial infection sample.

Next, an analysis example using SIMCA will be shown. A SIMCA model is created based on the difference spectrum of a healthy sample. The produced SIMCA model is applied to an unknown sample, and the distance of the unknown sample from the center of the SIMCA model is obtained. The distance is compared with the residue standard deviation of the healthy sample, and when the distance is larger than the residue standard deviation, it is judged as an infected area and when it is smaller, it is judged as a healthy area.

  FIG. 7 shows a Coomans plot in which a healthy sample and an infected sample (Bacillus-infected sample) are identified by the SIMCA model. The horizontal axis represents the distance from the healthy sample SIMCA model center, and the vertical axis represents the distance from the infected sample SIMCA model center. The sample in the healthy area 16 is distributed in the upper left in the figure, and the sample 17 in the infected area is distributed in the lower right. That is, it can be seen that samples infected with Bacillus bacteria are distributed at a distance from the sample group in the healthy area.

  The platelet product non-invasive test system according to the present invention can be used for quality control of platelet products at blood centers such as the Japanese Red Cross by automation.

Spectrum measurement (cross-sectional view) by transmission method (left) and interactance method (right) Diagram showing second derivative spectrum The figure which shows the principal component analysis result which uses the second derivative spectrum Diagram showing difference spectrum Diagram showing results of principal component analysis using difference spectrum The figure which shows the analysis result by PLS regression Diagram showing the results of SIMCA analysis

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Platelet preparation contained in blood collection bag 2 ... Sample chamber 3 ... Incident light 4 ... Transmitted light from sample 5 ... Optical fiber bundle 6 for incident light ... Transmitted light Optical fiber bundle 7 ... Platelet preparation 8 in blood collection bag ... Spectral incident light 9 ... Diffuse reflected light 10 from sample ... Interactance optical fiber bundle 11 ... Healthy area (target area) ) Sample 12 ... Sample 13 of Serratia fungus-infected area ... Sample 14 of Staphylococcus epidermidis infected area ... Healthy area 15 ... Infectious area 16 ... Healthy area 17 ... Infectious area

Claims (1)

The near-infrared spectrum of the platelet product in the blood collection bag is measured non-invasively over time without removing the platelet product from the blood collection bag, and the time course changes in the spectrum of bacterially infected samples and healthy samples are analyzed. In a non-invasive platelet preparation method for determining the presence or absence of bacterial infection in an unknown sample by comparing the time course of the spectrum of the sample with that of a bacterially infected sample and a healthy sample,
The first near-infrared spectrum is measured at the time of manufacture of the unknown sample, and the second derivative of the spectrum obtained by the first measurement is subtracted from the second derivative of the spectrum of the unknown sample measured over time. This difference spectrum is used as a control value and applied to a PLS regression calibration model in which the infectivity of healthy samples is 0 (zero) and the infectivity of bacterially infected samples is 1. If the estimated value of infectivity is less than 0.4, it is unknown. A platelet product non-invasive test method characterized in that a sample is discriminated as a healthy sample, and an unknown sample is discriminated as a bacterial infection sample at 0.4 or more.

Images