JP5183201B2 - Method and apparatus for analyzing amniotic fluid - Google Patents

Description

  This application claims priority from US Provisional Application No. 60 / 496,884, filed Aug. 21, 2003, the contents of which are incorporated by reference.

  The invention described and claimed in this application is not based on research funded by the US federal government or its institutions, and the US government has no rights to this patent application.

  The present invention relates to a method and apparatus for determining the risk of developing fetal health or fetal health. The present invention also relates to a method and apparatus for monitoring maternal health. The present invention further relates to a method and apparatus for analyzing amniotic fluid so as to determine one or more biological markers.

  Assessment of fetal weight at birth is an important part of prenatal management. Therefore, the result of prenatal management can be significantly improved by accurately determining the weight of the fetus before delivery. Thus, infants are particularly at risk of one of two extreme cases: (1) Giant's disease (also referred to as an unfair weight baby (LGA)) or (2) an unfair light weight baby (SGA) or intrauterine growth retardation (IUGR). If so, there is a need for a quick and easy way to estimate the weight of the fetus in the womb.

  Currently, the most reliable baby birth weight prediction device is ultrasound, and according to a recent review article, birth weight can be predicted within 300 g to 400 g (Table 12, Nahum eMedicine Journal). However, the authors also say that, as with other methods, it is still inaccurate, and reasonable strategies for achieving fetal weight prediction include clinical and sonic sources. It suggests that multiple predictions based still are used. The authors further stated that giantism cannot be easily predicted using ultrasound. Fetal size ultrasound diagnosis and palpation has less than 60% sensitivity to predict giantism, and false positives greatly exceed 40%. Similarly, predicting fetal weight with ultrasound of small fetuses less than 1800 grams often involves as much as 25% error. Disadvantages of ultrasound include the complexity and effort burden of methods often limited by suboptimal visualization of fetal organs. It also requires expensive equipment and skilled personnel. The latter requirement often hinders the use of modern methods in developing countries.

  In WO 2004/036359 published on April 29, 2004, birth weight is determined by combining the use of fetal ultrasound measurements and information about the mother. Current US and Canadian guidelines recommend that all pregnant women be tested for gestational diabetes mellitus (GDM) between 24-28 weeks. Priority screening is done if there are multiple risk factors, such as an older age of pregnancy, a higher pre-pregnancy weight, a member of a high-risk ethnic group, or a strong family history of diabetes, or Only if GDM has been diagnosed in the past or there has been a delivery of an infant with giantism. However, some studies recognize that “selective screening” based on these criteria may still underdiagnose GDM.

Current screening and diagnostic criteria for GDM are predicted based on observing increased perinatal and adult morbidity and mortality due to abnormal oral glucose tolerance test (OGTT) and concomitant gestational hyperglycemia The Increased glucose flux across the placenta is a prerequisite for stimulants that produce insulin in the uterus from growing islet cells and fetal hyperinsulinemia, thereby leading to fetal abdominal girth, giantosis, obesity and neonates It has been argued that it leads to increased hypoglycemia and diagnosis of GDM.
US Provisional Patent Application No. 60 / 496,884 WO2004 / 036359 A. Graps, “An Introduction to Wavelets”, IEEE Comput. Sci. Eng. 2, pp. 50-61 (1995) E. Aboufadel and S.M. Schlicker, “Discovering Wavelets” (John Wiley & Sons Inc., NY, 1999) J. et al. S. Walker, “A Primer on Wavelets and the Scientific Applications” (Chapman & Hall, Boca Raton, 1999)

  In a broad aspect of the invention, amniotic fluid is analyzed in situ without rupturing the amniotic membrane.

  In another broad aspect of the present invention, amniotic fluid is analyzed to ascertain information about the concentration of the matrix making up the fluid and / or other components without changing the components of the amniotic fluid.

  In yet another broad aspect of the invention, amniotic fluid analysis correlates with the risk that at least one of a mother and a child will develop a medical condition.

  In a more broad aspect of the invention, the prediction of birth weight by amniotic fluid analysis is improved by either making the prediction early in pregnancy or by making the prediction more accurate.

  The present invention provides a method for analyzing amniotic fluid, wherein the device is prepared for measuring one or more selected biological markers in amniotic fluid, and amniotic fluid in situ without inserting an instrument into the amniotic membrane. Is placed against the amniotic membrane to measure. The device is used to obtain measurement data that is processed to obtain values for one or more selected biological markers in the amniotic fluid.

  The present invention provides an apparatus for analyzing amniotic fluid in situ in pregnant women having amniotic fluid containing amniotic fluid without inserting an instrument into the amniotic membrane. The apparatus positions the device with respect to the amniotic membrane to measure one or more selected biological markers in the amniotic fluid and to measure the amniotic fluid in situ without inserting an instrument into the amniotic membrane. And a processing unit for processing the measurement data from the device to obtain the value of one or more selected biological markers in the amniotic fluid.

  The present invention provides a device for measuring one or more selected biological markers in amniotic fluid, such that the device measures amniotic fluid in situ without inserting an instrument into the amniotic membrane. , And measure the data to obtain the value of one or more selected biological markers in amniotic fluid, depending on the value A method of treating a pregnant mother and / or its fetus by determining at least one of a dietary change and a drug treatment is provided.

  The present invention provides a device for analyzing a mother's amniotic fluid, using the device to obtain analytical data from the amniotic fluid and processing the analytical data to obtain a predictive value for risk. And a method for predicting the risk of developing medical symptoms in at least one of the children.

  The present invention provides an apparatus for predicting the risk of developing medical symptoms in at least one of a mother and her child. The apparatus includes a device for analyzing amniotic fluid and a processing unit for processing analytical data from the device to obtain a predictive value for the risk.

  The present invention can be applied to an animal having an amniotic membrane that can be used for amniotic fluid for analysis. In particular, the present invention can be applied to a human. The terms “patient”, “mother”, “child”, “fetus”, and similar terms for the subject and its body parts are used herein with respect to humans and non-humans, unless expressly stated otherwise. Shall.

  As used herein, the term “medical symptom” means a condition or possible condition related to the health of the mother, fetus or child. One example is the birth weight of the child, i.e., the birth weight, which includes an unacceptably light baby (SGA) or intrauterine developmental delay (IUGR), a normal gestational age child (AGA) or a healthy child, and There are unfair weight children (LGA) or giantism. Abnormal weight is recognized as a cause of various health complications. “Medical symptoms” also include showing that there are no problems such as AGA, and this information for pregnant women is a useful reconfirmation of maternal health. Another example is maternal diabetes, which is also known as maternal diabetes, more specifically gestational diabetes (GDM). Another example is preterm birth.

  As used herein, the term “biological marker” refers to one or more of glucose, lactate or other metabolic acidic substances, including but not limited to insulin, insulin-like growth factor (IGF) and its binding proteins. One or more biochemical indicators comprising the above proteins and / or one or more fatty acids. Biological markers can also include cells that can be identified and counted in amniotic fluid, whether in vivo or in situ, as well as other physical properties of amniotic fluid that can be measured, such as viscosity.

  The invention will be better understood by viewing the following detailed description of several embodiments with reference to the accompanying drawings, in which:

First Embodiment-In Vivo Amniotic Fluid NIR Raman Spectrum Analysis in Human and Correlation with Birth Weight In one embodiment of the present invention, 8-12 biochemistry in amniotic fluid predicts infant birth weight. Focus on the use of Raman spectral analysis to identify a panel of genetic markers. The advantages of this approach are as follows. (1) Only a small sample of amniotic fluid (μL) is required to measure all important biochemical components simultaneously, and more importantly, its chemical properties are determined by the amniotic fluid fluid matrix. This is a fetal health risk if too little (oligiohydramniois) or too much (polyhydramniois) is itself an important barometer of fetal health It becomes. Other chemical methods that require separate analysis for individual components are not only susceptible to concentration differences when volume is unstable, but this new compartment has not yet been developed for small quantities It is susceptible to deficiencies in the method of measuring components at, but this overcomes its limitations. More importantly, however, Raman spectral analysis is accurate to within 100-400 grams of final birth weight when performed early in the 15th week of pregnancy. This therefore provides the first medical potential for early intrauterine diagnosis of SGA and LGA. Furthermore, this method can be performed during normal amniocentesis and does not require chemical processing of the sample which requires additional labor. The method of this embodiment can be easily implemented in hospital, clinic and field environments with the development of two machines: when “fresh samples” are taken for rapid bedside processing. A machine that requires the use of a small hand-held Raman spectrometer to measure amniotic fluid droplets, and (2) the first to collect successive measurements and monitor the fetal growth and subsequent growth in the fetus Development of an intravaginal or intraabdominal fiber optic probe that provides a means for non-invasive use during pregnancy. The feasibility of the method of this embodiment allows more extensive use of amniotic fluid for normal fetal monitoring and is available with much better accuracy than current approaches.

  In this application, several components of amniotic fluid suitable for measurement have been identified. These components include, but are not limited to, glucose, a family of proteins (but not limited to insulin and two IGF binding proteins, ie IGF BP-1 and 3), several amino acids, and two metabolic properties There are acids (lactic acid and uric acid). Other ingredients for measurement include nitric acid and several fatty acids including trans-fatty acids that are found only in highly hydrogenated foods and may effectively limit the use of essential fatty acids necessary for fetal growth There is.

Amniotic fluid NIR Raman spectrometer Amniotic fluid from 68 women aged 14-16 weeks was measured. All patients signed the McGill University patient approval form for consent confirmation. After genetic testing, the remainder of all amniotic fluid samples were stored frozen. Near infrared Raman spectra were obtained using a Bruker Fourier transform near infrared Raman spectrometer. Each amniotic fluid sample was removed from the freezer and warmed to 20 ° C. The sample was then transferred into a 2 mm diameter glass tube mounted in the Raman system. The Raman system was maintained at 20 +/− 1 ° C. during the experiment. The Nd: YAG laser radiation emitting at 1064 nm was focused on the amniotic fluid sample. The Raman scattering shift from the sample was collected with an FT spectrometer and detected using a cooled NIR detector. The spectrum was scanned at 1 / second and a resolution of 8 cm-1 was obtained from a shift of 0-3750 cm-1. A total of 1800 scans was averaged for each sample. After Fourier transformation of the raw interferogram, the data was stored as 1919 data points spanning the 0-3750 cm-1 spectral range.

Data processing The amniotic fluid Raman spectrum was preprocessed to reduce the effects of fluctuations in laser intensity. In particular, each spectrum was normalized to the Si—OH Raman emission at 2500 cm −1. Similarly, the spectrum was smoothed by a moving average boxcar smoothing function at 15 points so as to suppress spurious noise during measurement.

Haar Transform Haar Transform (HT) is the oldest form of wavelet analysis. This predicts a given signal on the orthogonal component of the basis function. Data included in a time window of 0 <τ <1 is decomposed according to a father wavelet φ (τ), a mother wavelet ψ (τ), and successive daughter wavelets ψ _{n, k} (τ). Here, n and k represent scaling and translation, respectively.

Similarly, each daughter wavelet can be decomposed into a sum of two sun wavelets ψ _nk (τ) with corresponding positive and negative weights and associated scaling and translation. It is interesting to note that all daughter wavelets can be decomposed into a sum of sun wavelets, ie a compressed and moved version of the father wavelet. For example, when ψ = φ _1,0 −φ _1,1 , a Haar transform can be performed with a basis component consisting only of 0 and 1, which can be performed experimentally with a spectral filter. it can. In order to measure wavelet coefficients, it is useful to represent wavelets as a matrix. For example, it can be expressed father, mother, and the first generation of daughter wavelets as A _2.

However, each column corresponds to a wavelet and each row represents a Haar wavelet value when the time window is divided into four equal segments. Decomposing four equal wavenumber binary Raman signals into wavelet coefficients leads to the mathematical problem of the following matrix: The coefficient vector is calculated so that a Raman spectral profile is obtained when doubling to A _2. Must-have. The final vector wavelet coefficients are ordered to start with the lowest resolution wavelet (father wavelet) and proceed to higher spectral resolution. The matrix A ₂ can be further expanded to include earlier generation daughter wavelets or sun wavelets, thereby extending the analysis to high frequency levels.

  More detailed information on Haar transforms can be found in A. Graps, “An Introduction to Wavelets”, IEEE Comput. Sci. Eng. 2, 50-61 (1995). Aboufadel and S.M. Schlicker, “Discovering Wavelets” (John Wiley & Sons Inc., NY, 1999); S. See, Walker, “A Primer on Wavelets and the Scientific Applications” (Chapman & Hall, Boca Raton, 1999).

  Using the FTNIR Raman instrument to calculate the Haar coefficient of an experimentally collected distribution is the first step in data analysis. Sun Wavelet's Haar transform calculation was performed using a tailored program written by Matlab (The MathWorks Inc., Natick, Mass.) That iteratively calculates totals and differences. In the calculation, the length of the input data needs to be a power of two lengths. Sun wavelet coefficients of up to 1024 ha were obtained and arranged from low resolution to high spectral resolution.

Stepwise Multiple Linear Regression Inverse least square regression can be used to estimate extrinsic parameters for a given sample from the preprocessed Raman spectrum. However, it is almost certain that not all 1024 wavelets are required since HT provides a sparse representation signal. Stepwise multiple linear regression is an established method of selecting a subset of variables that are most correlated with the quantity of interest. The general purpose of the genetic algorithm was to identify wavelet combinations that best describe a given data set according to Equation 5.
Y = α ₀ + α ₁ X ₁ + α ₂ X ₂ +... + Α _n X _a (5)
However, Y is the dependent variable (birth weight), the X _1, X ₂ ... X _n are independent variables (i.e., wavelet _{coefficients), α 0, α 1 ...} α n is inverse least of a set of X This is a coefficient obtained by square regression. The combination of wavelets that best estimates Y was determined according to the following scheme.
1. 1. Set the range of HT coefficients used by Stepwise. 2. Select the number of wavelets included in the model. 3. Evaluate the fitness of each model. 4. Increase the number of wavelets included in the model and repeat steps 2-4. Choose the optimal number of variables 6. Evaluate the model using an independent dataset. Repeat steps 1-6 while changing the range of HT coefficients used in the stepwise method

1. Set the range of HT coefficients used by the stepwise MLR: The goal of STEPMLR was to determine a small subset of wavelets that correlate with birth weight. Similarly, low resolution (large wavelength range) components are preferred from the perspective of developing the identification of spectral components related to fetal growth and to simplify future measurements. Thus, in addition to choosing to optimize the estimation among all Haar-Sun wavelets by the stepwise method, the algorithm also uses lower spectral resolution wavelets than 512, 256, 128, 64, and 32 wavelets. Only executed with.
2. Select the number of wavelets to include in the model: Start with one wavelet, ie, one X in Equation 5, and incrementally. The maximum number of wavelets was set according to the number of wavelet coefficients available for stepwise selection. In all cases, the maximum number of wavelets used was set to 10.
3. The fitness of each model is evaluated: For each model in the population, the coefficients α ₁ to α _{n of} Equation 5 are used by inverse least squares regression using a calibration set with known values of absorption or scattering determined from sample preparation. Was calculated. An estimate of birth weight was obtained by fitting the determined α _n parameter and the Haar coefficient of the test set to Equation 5, and the standard error of calibration (SEC) was calculated. Thus, a better model involves a smaller SEC.
4). Repeat steps 2-3 by increasing the number of wavelets contained in the model: In step 2, select the maximum number of wavelets.
5. Choosing the optimal number of variables: A prediction error sum of squares (PRESS) diagram was created by graphing the number of SECs versus wavelets in the model. The number of wavelets of the model at the minimum PRESS value is represented by h. The model selected had a minimum number of wavelets so that the PRESS of the model did not significantly exceed the PRESS of the h wavelet model based on the f-test with 95% confidence.
6). Assess the model using an independent data set: The “optimal” model was evaluated by estimating the birth weight value of the validation set using the calibration coefficients from the independent data set, the calibration set. R ² and coefficient of variation (CV) were used as indices of model effectiveness.

Results Three separate calibrations were performed to estimate birth weight from the amniotic fluid Raman spectrum. First, birth weights were estimated using spectra associated with all samples. The results are shown in FIG. An estimate of birth weight within 500 grams was obtained for all samples. Dividing the samples into groups less than 3500 grams and more than 3500 grams gave significantly better results. The results of two calibrations are shown in FIGS. 1d and 1f. As can be seen from the figure, an estimation with an error of about 200 grams and only one outlier was seen in each group. This is significantly better than any current method.

Second and Third Embodiments—In Situ Probe Including Optical Spectrometer In the second embodiment, intravaginal spectrum measurements were made during 2-5 regular ultrasounds during pregnancy. Early measurements have opportunities for therapeutic or nutritional treatment. However, the sensitivity of early pregnancy measurements may be reduced by cervical tissue thickness (˜4 cm). In contrast, in intravaginal measurements performed in late pregnancy, interference with cervical tissue (less than 1 mm) is less because it fades throughout pregnancy, but there are fewer opportunities for medical treatment. It will be readily appreciated that an ultrasound image of the amniotic membrane can be used to help position the probe or to confirm that the probe measures amniotic fluid without interfering with the fetus. If desired, an optical spectrometer can be incorporated into the ultrasonic intravaginal probe.

  Particularly for the estimation of IUGR and giantism, the spectral region important for predicting birth weight in situ measurements is that obtained with frozen samples ex vivo as described above depending on the intervening tissue, temperature, and pH of AMF. Can be expected to be slightly different. Specific regression models developed at various pregnancy times are compared to “gold” standard measurements obtained by ultrasound.

  Vaginal NIR measurements can be used to find a regression relationship between the Raman spectrum and birth weight. Similarly, optical attenuation spectra in the NIR or IR region can be measured and correlated for birth weight prediction and according to other medical conditions of the invention. Raman scattering measurements provide excellent analytical information about amniotic fluid, but optical attenuation or absorption measurements are more efficient when the measurement is done in situ and the depth of the amniotic fluid measurement makes it more difficult It is expected to be.

  As shown in FIG. 2, the probe has a tip with a light source and an optical detector. Alternatively, it is preferable to integrate a suitable light source and detector into the probe tip, but in the presently preferred embodiment, the optical fiber relays light between the remote light source and the detector to the tip.

  As shown in FIG. 3, the optical probe is operably coupled to a spectral analyzer that controls the light source and detector to obtain the desired spectral information. As described above, such an analyzer is a Raman scattering analyzer. The resulting spectral data is received by a correlator that calculates risk values for medical conditions based on calibration data that specifies how spectral information is correlated to the risk values. In addition, or alternatively, the correlation can be performed to determine amniotic fluid biochemical markers or other component concentration values.

  In a third embodiment, as shown in FIGS. 4 and 5, the adapted probe can have a non-invasive tip with a light source and two optical detectors. As shown, the first detector “recognizes” significantly different path lengths through tissue between nearby and deeper tissues, and the second detector “recognizes shorter path differences. " By subtracting the intensity data measured by the two detectors, information about deeper tissue, ie amniotic fluid, can be obtained. The probe is adapted to make optical contact with the patient's abdomen. In the case of an intravaginal probe, the device of FIG. 4 is also preferred, but the depth of tissue or fluid to be measured is not large and therefore the separation between the first detector and the second detector need not be large.

  More specifically, in the case of a NIR Raman system suitable for intravaginal measurements, a NIR Raman system commercially available from Ocean Optics can be adapted for amniotic fluid measurements. The system laser can be a low power (50 mW) laser at 785 nm. Choose a shorter wavelength compared to 1064 nm used in the first embodiment, since the scattering is proportional to λ-4 and the low light output can be used for non-invasive in situ measurements. Is desirable. Similarly, this near-infrared region can easily pass through the cervical tissue predicted by the measurement. The Raman spectrometer detector is a highly sensitive cooled CCD detector that provides a robust system for portable use. Conventional measurements suggest that a relatively low resolution Raman spectrum is sufficient for regression modeling of birth weight. Adjusting the entrance slit of the spectrometer provides a convenient means of optimizing in-situ measurement resolution and signal strength. From preliminary measurements, it is expected that in vivo spectrum acquisition will take about 3 minutes.

  Both laser excitation and Raman scattering are transmitted to the patient by a bundle of customized optical fibers. The bundle consists of separate illumination and collection fibers that focus on the same location in the tissue. Using this confocal arrangement, the exact location of the tissue can be isolated for three-dimensional quantitative measurements. At the distal end of the illumination fiber, a small short wavelength transmission optical fiber is placed to eliminate unwanted scattering from the illumination fiber. A second long wavelength transmission filter is placed in front of the collection fiber to isolate the Raman shift signal transmitted to the spectrometer. The diameter of the fiber bundle is less than 2 mm. The confocal optical probe is only 2 mm × 5 mm with an ultrasonic scan head.

The fiber bundle is attached to a low-cost ultrasound imaging system equipped with an intravaginal ultrasound scan head (Medison A-600) so that it can determine the sampling location in vivo. . Optical fiber, using a mechanical indentation vaginal probe, are clipped in the retaining ring of Teflon so as to maintain a known sampling position. Within the patient's body, a condom is placed over the ultrasound / optical probe to provide a hygienic environment. The position of spectrum acquisition is determined using a tissue phantom. In addition to being placed at the location of spectral measurements, intravaginal ultrasound images provide information about the uterine and membrane shapes that are useful for spectral comparisons. In addition to making and calibrating the system with known composition samples, the components using the purified component spectrum present in the majority of amniotic fluid so that a comparison between the spectral regions used in the regression is made Can be provided.

  It will be appreciated that although the vaginal probe embodiments described above are intended for use in women, they can also be adapted for use in other mammals.

  It will also be appreciated that non-optical analyzers can similarly be used to collect in situ information about amniotic fluid components that can be correlated with medical condition risk. For example, MRS can provide detailed analytical information about chemical composition. Physical parameters such as amniotic fluid viscosity, which can be measured in situ by ultrasound, can also be used alone or in combination with other optical or non-optical analyzers to measure risk or one or Further biological markers can be measured. Amniotic fluid components that change as a function of predicted birth weight are believed to affect viscosity, and therefore it is expected that there will be a correlation between viscosity and birth weight.

  For an optical spectrometer, a suitable wavelength region is about 200 nm to 400 μm. Amniotic fluid dry samples can be analyzed over this range, while ex vivo whole samples or in situ amniotic fluid are analyzed using wavelengths that are not excessively absorbed by intervening tissues or the amniotic fluid itself. For example, water absorbs a large range from 2 μm to 50 μm, and the presence of water in amniotic fluid makes this range essentially unusable for in situ measurements.

Fourth and Fifth Embodiments-Monitoring Gestational Diabetes Mellitus (GDM) Using Iterative Non-invasive Amniotic Fluid Analysis The purpose of this study is to: 1) Describe the prevalence of GDM in a population of elderly women who are at high risk of age who receive regular amniocentesis for genetic testing. 2) Increased glucose, insulin or insulin-like growth factor binding protein (IGF BP) type 1 in amniotic fluid (AF) pre-existing at the time of normal amniocentesis (range 12-22 weeks) in these women is 24 Shown to be diagnosed with GDM at ~ 28 weeks. 3) Use multiple regression to establish that there is an association between these amniotic fluid indices and subsequent GDM diagnosis. 4) Use probability maps to demonstrate specific amniotic fluid concentrations of glucose, insulin, and IGF BP1 that predict an increased risk of GDM.

  Design, recruitment and consent: From 1998 to 2002, pregnant women undergoing regular amniocentesis at the St Mary's Hospital Center in Montreal, Canada were invited to participate in this prospective study. The consent signature allowed the researchers to obtain amniotic fluid after genetic testing from the Montreal Children's Hospital and to view medical charts. Inclusion criteria (single pregnancy) and exclusion criteria (multiple pregnancy, genetic abnormalities) were applied, resulting in 1008 participants. By medical examination, GDM status, pregnancy age, pre-pregnancy weight and height, ethnicity, number of births, smoking (n = 888-928), 25 weeks (n = 70) and 35 weeks (n = 149) Information on fetal weight, infant birth weight, sex, and gestational age (n = 928) predicted by ultrasound was obtained. The gestational age was based on physician estimates using LMP and certain hospital protocols. The completeness of each data subset depended on the availability of information in medical charts and questionnaires. Ethical approval was obtained from Institutional Review Boards of McGill, Montreal Children's Hospital, and St Mary's Hospital Centre.

  Biochemical analysis: Amniotic fluid samples stored at −80 ° C. were analyzed for glucose, insulin, and IGF type BP1. Insulin (n = 718) uses a Beckman Access ultrasound analysis system, a one-step immunoenzymatic assay that adds a monoclonal anti-insulin complex, antibodies coated with paramagnetic particles, and a chemiluminescent substrate to the reaction vessel. And analyzed. Insulin was measured within 0.03 to 300 ul pmol / L. Glucose (n = 662) was adapted to the Abbott Laboratories (North Chicago, IL) analysis kit (No. 6082) for use in a microplate reader, and IGF BP1 type (n = 876) was Diagnostic Systems Laboratories (Nor. DSL kit 10-7800 (Webster, Texas) was used for analysis after fitting by ELISA.

  Statistical analysis: All data analysis was performed using SAS (Version 8.02, SAS Inc., Cary, NC) with statistical significance P <0.05. All non-normal distribution data were converted using square roots: pregnancy weight, BMI, ethnicity, number of births, amniocentesis week, smoking, infant birth weight, 35 week fetal weight, and amniotic fluid glucose, insulin, and IGF -BP1 type. The biochemical comparison of GDM and non-GDM mothers included maternal BMI, ethnicity, number of births, and week of amniocentesis as covariates. GDM and birth weight as dependent variables and multiple regressions of previously established predictors included in the model were also verified using forward and backward regression. Due to the co-linearity between IGF BP1 type, insulin, and glucose, each was incorporated into a separate regression model. GDM and non-GDM maternal data were modeled separately using a mixture of Gaussian distributions using a program written in Matlab V6.1 (Mathworks, Inc.) called Bayesnet by Ian McNabbey of Cambridge University. . The empirical probability of developing GDM with amniotic fluid biomarkers for insulin and glucose was calculated using the Bayesian weights of the Gaussian distribution determined from the measured data. Then, as shown in FIGS. 6 and 7, respectively, contour plots of the probability of developing GDM were determined for IGF-BP1 type and insulin variations associated with glucose during AF.

  Population characteristics: According to comparisons between GDM and non-GDM subpopulations, GDM mothers are shorter in height, have higher pregnancy weights and BMI, and overweight or obesity is only 26% in non-GDM mothers In contrast, it was 54% among GDM mothers. For healthy, non-GDM mothers, the mean birth weight was 3396 ± 19 g, compared to 3515 ± 52 g for GDM mothers. However, only 4000% of GDM offspring and 12% of non-GDM offspring exceeded 4000g, and using the percentile of birth weight to correct gender and gestational age, the mother of GDM 23% of infants born in the United States exceeded 90% (LGA), and the non-GDM population had only 10% LGA. Both classifications resulted in 3-4% of GDM mothers giving birth to IUGR or SGA infants. The weight of the fetus was not different until 25 weeks, but by 35 weeks it was 134 grams more in GDM mothers. At 35 weeks, gestational age (β-factor (β) = 215 g) and GDM (β = 54 g; p = 0.450) (except BMI, smoking and infant gender) are independent predictors of fetal weight. As entered. This difference decreases to 119 grams by maturity, at which time GDM (β-factor = 165 g), smoking (β = −111 g), infant sex (β = 124 g) and gestational age (β = 135 g), And pregnancy height and weight (β = 750 g and β = 7.50 g, respectively) as an independent predictor of infant weight. In this study population (n = 928) of elderly mothers (37.8 ± 0.1 years, 26-45 years), the onset of GDM was 12%.

  Biochemical measurements: Regardless of BMI, ethnicity, number of births, and inclusion of the amniocentesis week, GDM mothers had higher amniotic glucose concentrations and lower amniotic fluid IGF BP1 than non-GDM mothers. Interestingly, amniotic fluid insulin did not differ when these covariates were incorporated. However, all three biochemicals of amniotic fluid were entered as independent predictors of GDM, but only amniotic fluid IGF-BP-1 was entered as a negative predictor of birth weight. Using a probability map to visualize the risk of GDM, it can be shown that if either amniotic fluid glucose or insulin is high and the concentration in other amniotic fluid is low, the risk of GDM exceeds 90% did it. Furthermore, low amniotic fluid IGF BP-1 in the presence of high glucose is also associated with a risk of GDM greater than 90%.

  In this study, it is clear that amniotic fluid glucose, insulin, and IGF BP-1 concentrations are already high in women later diagnosed with GDM, and these amniotic fluid components may act as early predictors of GDM It was made. This conclusion is revolutionary by: 1) High AF glucose and low IGF BP-1 have been demonstrated to be associated with later GDM diagnoses and various birth weight pups in women of various BMI categories. 2) Probability diagrams could be used to predict the risk of subsequent GDM diagnosis for each AF glucose, insulin and IGF BP-1 concentration by week 15 of gestation. This GDM risk profile assessment using amniotic fluid samples obtained at the time of regular amniocentesis for genetic testing prior to the current screening and diagnostic protocol by week 10 is a high concentration measured in early pregnancy. Based on the presence of urinary amniotic fluid insulin and glucose, GDM mothers demonstrated that growing fetuses were exposed to a glucose-rich environment fairly early. The current protocol effectively screens higher maternal BMI and effectively attempts to minimize high birth weight in GDM offspring by reducing abdominal girth, giantosis and obesity related to the fetus, It does not help reduce fetal disease associated with early in-utero glycation and protein glycosylation, which are reported to exist by the third trimester. If amniotic fluid glucose rises very early in pregnancy, the amniotic fluid glucose diffuses throughout the skin of the unkeratinized fetus until 20-24 weeks, and damage to the fetus is greater than previously predicted This could lead to exposure of the growing fetal pancreas to glucose in the early amniotic fluid, predisposing to increased risk of β-cell exhaustion later in life, and higher BMI later in life The risk of developing diabetes and GDM increases and the risk of adult disease increases.

  In this population, the prevalence of GDM was 12%. This incidence is higher than the CDA report (ie, 8-18%) for multi-ethnic groups, including indigenous people, and higher than the ADA report (7%). This is surprising considering that the average age of mothers who have undergone regular amniocentesis for genetic testing is higher than those shown as risk factors by CDA and ADA (ie, 26 to 35 years) None, but provides the first report of the incidence in this high-risk population of older women. It should be noted that the proportion of Asians is higher in the GDM population (37%) than in the non-GDM population (18%), but this observation also indicates that Asian GDM prevalence is higher in other reports. I support it. Interestingly, GDM occurs with the same frequency in women with a BMI less than 25 as women with a BMI greater than 25. As conventional screening methods emphasize high pregnancy weight as a risk factor, this data relates to why GDM is underdiagnosed when as many normal and low-weight individuals are also susceptible. Give some insights. Interestingly, most women gave birth to non-giant pups when associated with being GDM and having a birth weight of 165 grams. While the traditional Pedersen hypothesis linked fetal hyperglycemia and unduly babies to increased birth weight, it was also observed that GDM mothers gave birth to SGA and AGA as well. Other than that, our multi-ethnic population was non-smoking and the incidence of IGUR was lower than the normal population, but the incidence of AGA and giantism was generally similar to the Canadian and US populations.

  GDM is currently diagnosed at 24-28 weeks. As noted above, this study revealed that AF glucose had already increased by 15 weeks of gestation in the GDM subpopulation. Several studies have previously suggested that maternal fasting and 2 h plasma glucose levels are positively related to birth weight and that glucose is free to cross the placental barrier by accelerated diffusion. Meanwhile, another study reports that AF insulin is superior to AF glucose as a predictor of impaired maternal glucose intolerance. One study actually showed that AF glucose was not associated with fetal hyperinsulinemia before 23 weeks of gestation. Our amniotic fluid study, with a large number of cases used and controlled conventional confounding factors, showed in a series of multiple regressions that amniotic fluid glucose and insulin were associated with GDM. Insulin fluid insulin measured early in pregnancy is probably not a major growth factor in early pregnancy, but it accumulates throughout this period and acts in late pregnancy, so neither predicts birth weight. It was. Regarding glucose, perhaps GDM mothers who had been treated in the past were able to predict GDM but could hardly predict birth weight because of the varying birth weights of the babies who gave birth. Importantly, however, the complexity of AF glucose and insulin predicting GDM was evident by creating a probability map. Probability contour plots demonstrated that the relationship was clearly not linear and that the risk of GDM exceeded 90% when either amniotic fluid glucose or insulin was high and the other concentration was low. In addition, each risk profile was non-linear. Therefore, the most important is the separation that exists between glucose and insulin values. A large difference between the two predicted that it was likely to develop GDM in the future, demonstrating that fetal hyperinsulinemia and increased amniotic fluid glucose both predict later GDM development.

  Another interesting predictor of GDM was that low amniotic fluid IGF BP-1 in the presence of high glucose was associated with a risk of greater than 90% GDM. Traditionally, IGF BP-1 was inversely related to birth weight, but we used a larger number of cases, and the lower IGF BP-1 resulted in a 54 g increase in fetal weight by 35 weeks They showed that full-term birth weight is associated with an increase of 164 g. In previous studies, this was either due to higher levels of growth hormone and / or increased blood IGF-1 levels, or secondary increases in insulin due to increased food intake, both of which are placenta It has been said to inhibit the production of IGF BP-1. Active IGF-1 increase stimulates fetal growth more late in pregnancy and is responsible for the increase in fetal weight by 35 weeks in non-GDM vs GDM pups as already demonstrated there is a possibility.

  In conclusion, high AF glucose, insulin and low IGF BP-1 were shown to predict GDM, where GDM ultimately predicts fetal birth weight. Our results demonstrate that the growing fetus of a GDM mother is already exposed to a “diabetes-induced risk profile” prior to the current GDM diagnosis, so Guarantees earlier screening and treatment to minimize damage. In addition, it was observed that almost 50% of GDM mothers had a BMI of less than 25, and many infants were not born as unweighted infants, so overemphasis on BMI as a screening criterion is an underdiagnosis of GDM. It may cause

  In a fifth embodiment, the present invention is used to repeat amniotic fluid measurements during pregnancy after a medical symptom of GDM to monitor the effects of health and diet and / or therapeutic treatment. The method is shown in FIG. Combining in situ measurements with the data obtained during amniocentesis is optional, but doctors who are not accustomed to interpreting or trusting the in situ analysis of amniotic fluid should be confident that this confirmation It will be fully understood. Similarly, the present invention allows in situ measurements that can be performed prior to 12 weeks gestation of a woman, and thus well before an amniocentesis can be performed safely, as shown at 10 weeks in the figure. However, the physician can also choose to begin monitoring maternal health in late pregnancy according to the present invention. While embodiments of the present invention are illustrated in the GDM example, it will be appreciated that the same applies to monitoring fetal health as well as birth weight. It is recognized that changing the mother's diet can affect the final birth weight and symptoms of GDM risk. It is believed that detecting the risk of developing GDM early and, as a result, changing meals early is effective in reducing the actual outcome of developing GDM. Exercise and drug treatment can also be applied according to medical guidelines.

FIG. 6 shows the general spectral and wavelength regions selected for birth weight estimation using NIR Raman measurements of amniotic fluid for a wide range of birth weights ranging from 1 kg to 5.3 kg. It is a figure which shows the correlation degree of an estimated birth weight and an actual birth weight regarding the case group between the birth weight of a wide range ranging from 1 kg to 5.3 kg. FIG. 5 shows the general spectral and wavelength regions selected for birth weight estimation using amniotic fluid NIR Raman measurements for a narrow range of birth weight ranging from 1 kg to 3.5 kg. It is a figure which shows the correlation degree between an estimated birth weight and an actual birth weight regarding the case group of the birth weight of the narrow range ranging from 1 kg to 3.5 kg. FIG. 6 shows the general spectral and wavelength regions selected for birth weight estimation using amniotic fluid NIR Raman measurements for a wide range of birth weights ranging from 3.5 kg to 5.3 kg. It is a figure which shows the correlation degree between an estimated birth weight and an actual birth weight about the case population of a wide birth weight ranging from 3.5 kg to 5.3 kg. FIG. 2 shows an intravaginal optical Raman spectrometer operating in the NIR or IR region. FIG. 3 is a block diagram of an apparatus according to the embodiment of FIG. It is a figure which shows schematically the abdominal probe which uses a light absorption spectrometer. FIG. 5 is a block diagram of an apparatus according to the embodiment of FIG. FIG. 2 is a contour plot of the probability of developing gestational diabetes (GDM) as a function of glucose value in the amniotic fluid (X-axis) and insulin (Y-axis). FIG. 5 is a contour plot of the probability of developing gestational diabetes (GDM) as a function of amniotic fluid glucose level (X-axis) and intrauterine growth factor binding protein type 1 or IGF-BP1 (Y-axis). It is a flowchart which shows the treatment method of GDM by 5th Embodiment.

Claims (21)

Providing a device for measuring one or more selected biological markers in amniotic fluid, wherein the device is a Raman near infrared spectrometer;
Placing the device against the amniotic membrane so that the amniotic fluid is measured in situ without inserting an instrument into the amniotic membrane, the placing step directing the spectrometer to analyze the amniotic fluid through the abdominal wall Process, including process,
Using the device to obtain measurement data, and processing the measurement data to obtain values of one or more selected biological markers in amniotic fluid. A method for measuring the amniotic fluid concentration of one or more selected biological markers.   The method of claim 1, further comprising obtaining an ultrasound image of the amniotic membrane during the placing step so as to indicate or confirm that the device measures amniotic fluid without interfering with the fetus.   3. The method of claim 1 or 2, wherein the one or more selected biological markers include glucose.   4. The method of any one of claims 1-3, wherein the one or more selected biological markers comprises at least one of insulin and IGF-BP1. a) preparing a device for analyzing maternal amniotic fluid, wherein the device is a Raman near infrared spectrometer;
b) using a device to obtain analytical data from the amniotic fluid, wherein the amniotic fluid is analyzed without treatment to separate or concentrate its components, the step of using the device comprising the amniotic fluid through the abdominal wall Including the step of orienting the spectrometer to analyze
c) depending on inverse least squares regression, the equation coefficients are calculated (5), by substituting the analysis data, and a step <br/> to obtain an estimate of birth weight, child in utero To calculate the birth weight of a child.   6. The method of claim 5, wherein the predicted value indicates that the child is born with either a high birth weight or a low birth weight.   7. The method of claim 5 or 6, wherein the analytical data is spectral data that directly correlates with body weight, thereby not using the value of a specific biochemical marker to obtain an expected value. d) storing analysis data;
e) obtaining subsequent data relating to body weight; and f) improving the correlation data using the stored analytical data and weight data for subsequent use in step c. The method described in 1.   9. The method according to any one of claims 5 to 8, wherein using the device comprises positioning the device relative to the amniotic membrane so that the amniotic fluid is measured in situ without inserting an instrument into the amniotic membrane. .   The method according to any one of claims 5 to 9, wherein the mother is a human and steps a to c are first performed about 12 weeks of gestation. The method according to any one of claims 5 to 10 , wherein steps a to c are repeated at least three times during pregnancy. A device that is a Raman near-infrared spectrometer for analyzing amniotic fluid,
Predict the risk of developing medical symptoms in the mother and / or its child, with an optical coupler adapted to place the device in relation to the amniotic membrane, so that the amniotic fluid is measured in situ without inserting an instrument in the amniotic membrane An apparatus for predicting the risk of developing medical symptoms in a pregnant mother and / or its child, comprising a processing unit for processing analytical data from the device to obtain a value. The apparatus according to claim 12 , wherein the analytical data comprises a spectrum of amniotic fluid. 14. The device of claim 13 , wherein the spectrum directly correlates with symptoms, thereby not using a specific biochemical marker value to obtain an expected value. 15. A device according to claim 13 or 14 , wherein the coupler is adapted to position the spectrometer to analyze amniotic fluid through the abdominal wall. 15. A device according to claim 13 or 14 , wherein the coupler is adapted to position the spectrometer to analyze amniotic fluid through the cervical canal. 15. A device according to any one of claims 12 to 14 , wherein the coupler is adapted to operate in contact with the woman at a location near the amniotic membrane. 18. A device according to any one of claims 12 to 17 , wherein the optical coupler is incorporated into an intravaginal probe. The apparatus of claim 18 , wherein the intravaginal probe also functions as an ultrasound device. 20. Apparatus according to any one of claims 12 to 19 , wherein the optical coupler comprises a light source and two optical detectors. A device that is a Raman near-infrared spectrometer for measuring one or more selected biochemical markers in amniotic fluid,
A coupler adapted to position the device relative to the amniotic membrane so as to measure the amniotic fluid in situ without inserting an instrument into the amniotic membrane, and one or more selected biochemicals in the amniotic fluid Analyzing amniotic fluid in situ in a pregnant patient with amniotic fluid containing amniotic fluid without inserting an instrument into the amniotic membrane, including a processing unit for processing measurement data from the device to obtain a marker value System for.

Images