CN116543905A

CN116543905A - Systems and methods for predicting ovarian polycystic like changes (PCOM)

Info

Publication number: CN116543905A
Application number: CN202310518798.8A
Authority: CN
Inventors: 李蓉; 徐慧玉; 乔杰; 赵荣生; 杨丽; 张现化; 冯国双; 韩勇; 杨蕊; 史莉; 马聪聪
Original assignee: Peking University Third Hospital Peking University Third Clinical Medical College
Current assignee: Peking University Third Hospital Peking University Third Clinical Medical College
Priority date: 2023-05-09
Filing date: 2023-05-09
Publication date: 2023-08-04

Abstract

The present application relates to a system for predicting ovarian polycystic like changes (PCOM) that predicts conditions of ovarian polycystic like changes in polycystic ovarian syndrome patients taking into account T levels, 11KA4 levels, A4 levels, and 17OHP levels, or AMH levels, T levels, and age. The system can accurately predict polycystic change, is helpful for understanding the etiology of PCOM, can assist in the genotyping diagnosis of PCOS, and is helpful for judging the severity of PCOS phenotype pedigree.

Description

Systems and methods for predicting ovarian polycystic like changes (PCOM)

Technical Field

The present application relates to a system and method for assessing the probability of ovarian polycystic like changes (PCOMs) in a subject.

Background

Polycystic ovary syndrome (PCOS) is a reproductive and metabolic disorder, the most common cause of anovulatory infertility, with a incidence of 5-20% in women of childbearing age. Without other specific diagnosis, adults have at least two of three features for diagnosing the disease: hypersomnia, anovulation disorder (prolonged menstrual cycle, amenorrhea are its main manifestations), and ovarian polycystic like changes (PCOM). Thus, women with PCOS are divided into four phenotype groups based on a random combination of signs and symptoms. Most affected women are also associated with other metabolic dysfunctions, such as obesity and insulin resistance. The pathophysiology of polycystic ovary syndrome has not yet been elucidated, mainly due to the heterogeneity of phenotypes, elucidating the etiology and subtype diagnosis of the different PCOS phenotypes is essential for subsequent treatments. The present invention seeks to investigate the etiology of PCOM by modeling and provide tools for subtype diagnosis.

High androgens are considered to be one of the potential drivers of polycystic ovary syndrome. Some studies have shown that pubertal high androgens can be early signs of polycystic ovary syndrome. In addition, genetic aberrations in androgen receptor are found in women with polycystic ovary syndrome, which can increase circulating androgens and lead to abnormal folliculogenesis. There is also strong evidence that excessive exposure to androgens in the uterus can lead to the onset of polycystic ovary syndrome in adulthood.

Despite these population-based findings and basic studies, however, high androgens cannot be observed in clinical practice in all PCOS phenotypes. To explore the role of high androgens in PCOS, there are two problems to clarify: (1) Because of the phenotypic heterogeneity of PCOS, high androgens may be the primary cause of some characteristics, but not others. Thus, assessing the contribution of high androgens in PCOM may provide another perspective for understanding PCOS etiology. (2) Clinically, only typical androgens, mainly testosterone (T), are routinely tested in diagnosing polycystic ovary syndrome. In addition, a widely used measurement method, chemiluminescent immunoassay, has the problem of insufficient sensitivity and accuracy for female circulating androgen levels. It is currently internationally accepted that Mass Spectrometry (MS) based detection methods are gold standards for steroid (including androgen) detection.

A group of non-classical androgens, namely adrenal-derived 11 oxygen and androgens (11 oxyC 19), have recently been considered to have potential roles in PCOS. 11 beta-hydroxy androstenedione (11 OHA 4) is the most abundant 11oxyC19 androgen, 11 beta-hydroxy testosterone (11 OHT) being produced mainly in the adrenal gland. These metabolites are converted mainly in peripheral tissues to 11-keto androstenedione (11 KA 4), 11-keto testosterone (11 KT) and 11-keto-5α -dihydrotestosterone (11 KDH). There is evidence that 11KT and 11KDHT are equally potent in activating the androgen receptor and its target gene as classical androgens T and Dihydrotestosterone (DHT), which updates the understanding of androgen signaling. Thus, quantification of these various androgens may expand the understanding of the high androgen profile of women with polycystic ovary syndrome. The invention predicts PCOM which is one of the phenotypes of PCOS by bringing the traditional androgens and the new 11 oxygen and androgen series indexes into the model together, and simultaneously models the PCOM together by combining other PCOS related clinical common indexes so as to search the cause of PCOM on one hand, establish a PCOM prediction model on the other hand and assist in PCOS subtype diagnosis.

Disclosure of Invention

The inventors of the present application have developed a sensitive and accurate Mass Spectrometry (MS) based detection method for simultaneous detection of classical and 11 oxygen and androgens (11oxyC19 androgens). Here, the inventors have established methods and systems for predicting ovarian polycystic changes to more accurately predict ovarian Polycystic Changes (PCOMs).

In particular, the present application relates to the following:

1. a system for predicting ovarian polycystic like changes (PCOMs), comprising:

a data acquisition module for acquiring data of testosterone (T) levels, 11-keto androstenedione (11 KA 4) levels, androstenedione (A4) levels, and 17 OH-progesterone (17 OHP) levels of a subject; and

a module for calculating a probability of having an ovarian polycystic like change (PCOM) for calculating the above data information acquired in the data acquisition module, thereby calculating a probability (p) of having an ovarian polycystic like change in the subject.

2. The system of item 1, further comprising:

a grouping module, wherein a default grouping parameter with ovarian polycystic like change (PCOM) is prestored in the grouping module, and the calculated probability (p) with ovarian polycystic like change (PCOM) is grouped according to the grouping parameter, so that the risk of the subject with ovarian polycystic like change is grouped.

3. The system of claim 1 or 2, wherein the subject is a polycystic ovary syndrome patient.

4. The system according to item 1 or 2, wherein,

the testosterone (T) level refers to the testosterone concentration of the subject detected on any one of the days in the menstrual cycle of the subject,

The 11-keto androstenedione (11 KA 4) level refers to the concentration of 11-keto androstenedione in the subject measured on any day of the menstrual cycle of the subject,

the androstenedione (A4) level refers to the concentration of androstenedione in the subject measured on any day of the menstrual cycle of the subject,

the 17 OH-progesterone (17 OHP) level refers to the concentration of 17 OH-progesterone in a subject measured on any day of the subject's menstrual cycle.

5. The system according to item 1 or 2, wherein,

in the module for calculating the probability of having an ovarian polycystic change, a formula for calculating the probability (p) that a subject has an ovarian polycystic change, fitted based on data of testosterone (T) level, androstenedione (A4) level, 11-keto androstenedione (11 KA 4) level, and 17 OH-progesterone (17 OHP) level of the subject in the existing database, is pre-stored.

6. The system according to item 5, wherein,

the formula is the following formula one:

p＝1/(1+e ^{-(a+b*T+c*A4+d*11KA4+f*11OHP)} ) (equation I)

Where p is the calculated probability that the subject has ovarian polycystic changes, a, b, c, d, f is the non-unity parameter, T is testosterone level, A4 is androstenedione level, 11KA4 is 11-keto androstenedione level, and 17OHP is 17 OH-progesterone level.

7. The system according to item 6, wherein,

a is selected from any numerical value in-0.194322-3.1655263;

b is selected from any numerical value in 0.1138558-3.0958606;

c is any numerical value selected from-0.109174 to 0.2678637;

d is any numerical value selected from-1.320305 to-0.163216;

f is selected from any number from-0.281078 to 0.0827339.

8. The system according to item 2, wherein,

the basis of the groups prestored in the grouping module is as follows:

when the calculated probability of the subject having ovarian polycystic like changes (PCOM) (p) <20%, the risk of the subject having ovarian polycystic like changes is low risk;

when 20% less than or equal to the calculated probability (p) of a subject having ovarian polycystic like changes (PCOM) is <40%, the risk of the subject having ovarian polycystic like changes (PCOM) is lower;

when 40% or less of the calculated probability (p) that the subject has ovarian polycystic like changes (PCOM) is <90%, the risk that the subject has ovarian polycystic like changes (PCOM) is a higher risk;

when the calculated probability (p) of a subject having an ovarian polycystic like change (PCOM) is ≡90%, the risk of the subject having an ovarian polycystic like change (PCOM) is high.

9. A system for predicting ovarian polycystic like changes (PCOMs), comprising:

A data acquisition module for acquiring data of anti-mullerian hormone (AMH) levels, testosterone (T) levels, and age of a subject; and

a module for calculating a probability of having an ovarian polycystic like change (PCOM) for calculating the above data information acquired in the data acquisition module, thereby calculating a probability (p) of having an ovarian polycystic like change (PCOM) for the subject.

10. The system of item 9, further comprising:

a grouping module, in which default ovarian Polycystic Change (PCOM) grouping parameters are pre-stored, and the calculated probability (p) of having ovarian Polycystic Change (PCOM) is grouped according to the grouping parameters, so as to group the risk of having ovarian Polycystic Change (PCOM) of the subject.

11. The system of claim 9 or 10, wherein the subject is a polycystic ovary syndrome patient.

12. The system according to item 9 or 10, wherein,

the anti-mullerian hormone (AMH) level refers to the AMH concentration of a subject detected on any day of the subject's menstrual cycle,

the testosterone (T) level refers to the testosterone concentration of the subject detected on any day in the menstrual cycle of the subject.

13. The system according to item 9 or 10, wherein,

In the module for calculating the probability of having an ovarian polycystic like change (PCOM), a formula for calculating the probability of having an ovarian polycystic like change (p) fitted based on data of the anti-mullerian hormone (AMH) level, testosterone (T) level, and age of the subject in the existing database is stored in advance.

14. The system of item 13, wherein,

the formula is the following formula II:

p＝1/(1+e ^{-(a+b*AMH+c*T+d*age)} ) (equation II)

Where p is the calculated probability of the subject having ovarian polycystic like changes (PCOM), a, b, c, d is the no-unit parameter, AMH is the anti-mullerian hormone (AMH) level, T is testosterone level, and age.

15. The system of item 14, wherein,

a is selected from any numerical value in-2.33502-2.6238419;

b is selected from any numerical value in 0.4500201-0.7267895;

c is any numerical value selected from-0.349974 to 0.6939225;

d is selected from any value in-0.132286 to 0.0077054.

16. The system of item 10, wherein the system further comprises, in combination,

the basis of the groups prestored in the grouping module is as follows:

17. A method of predicting ovarian polycystic like changes (PCOMs), comprising:

a data acquisition step of acquiring data of testosterone (T) levels, 11-keto androstenedione (11 KA 4) levels, androstenedione (A4) levels, and 17 OH-progesterone (17 OHP) levels of a subject; and

a step of calculating a probability of having an ovarian polycystic like change (PCOM) for calculating the above-described data information acquired in the data acquisition step, thereby calculating a probability (p) that the subject has an ovarian polycystic like change.

18. The method of item 17, further comprising:

a grouping step in which a default ovarian Polycystic Change (PCOM) grouping parameter is pre-stored, and the calculated probability (p) of having an ovarian Polycystic Change (PCOM) is grouped according to the grouping parameter, thereby grouping the risk of having an ovarian polycystic change in the subject.

19. The method of claim 17 or 18, wherein the subject is a polycystic ovary syndrome patient.

20. The method of item 17 or 18, wherein,

21. The method of item 17 or 18, wherein,

in the step of calculating the probability of having an ovarian polycystic change, a formula for calculating the probability (p) that the subject has an ovarian polycystic change, which is fitted based on data of testosterone (T) level, androstenedione (A4) level, 11-keto androstenedione (11 KA 4) level, and 17 OH-progesterone (17 OHP) level of the subject in the existing database, is stored in advance.

22. The method of item 21, wherein,

the formula is the following formula one:

p＝1/(1+e ^{-(a+b*T+c*A4+d*11KA4+f*11OHP)} ) (equation I)

23. The method of item 22, wherein,

a is selected from any numerical value in-0.194322-3.1655263;

b is selected from any numerical value in 0.1138558-3.0958606;

c is any numerical value selected from-0.109174 to 0.2678637;

d is any numerical value selected from-1.320305 to-0.163216;

f is selected from any number from-0.281078 to 0.0827339.

24. The method of item 18, wherein,

the basis of the pre-stored grouping in the grouping step is as follows:

25. A method of predicting ovarian polycystic like changes (PCOMs), comprising:

a data acquisition step of acquiring data of anti-mullerian hormone (AMH) level, testosterone (T) level and age of a subject; and

and a step of calculating the probability of having an ovarian Polycystic Change (PCOM) by calculating the above-described data information acquired in the data acquisition step, thereby calculating the probability (p) of having an ovarian Polycystic Change (PCOM) in the subject.

26. The method of item 25, further comprising:

a grouping step in which a default ovarian Polycystic Change (PCOM) grouping parameter is pre-stored, and the calculated probability (p) of having an ovarian Polycystic Change (PCOM) is grouped according to the grouping parameter, thereby grouping the risk of having an ovarian Polycystic Change (PCOM) in the subject.

27. The method of claim 25 or 26, wherein the subject is a polycystic ovary syndrome patient.

28. The method of item 25 or 26, wherein,

29. The method of item 25 or 26, wherein,

in the step of calculating the probability of having an ovarian polycystic like change (PCOM), a formula for calculating the probability of having an ovarian polycystic like change (p) fitted based on data of the anti-mullerian hormone (AMH) level, testosterone (T) level, and age of the subject in the existing database is stored in advance.

30. The method of item 29, wherein,

the formula is the following formula II:

p＝1/(1+e ^{-(a+b*AMH+c*T+d*age)} ) (equation II)

31. The method of item 30, wherein,

a is selected from any numerical value in-2.33502-2.6238419;

b is selected from any numerical value in 0.4500201-0.7267895;

c is any numerical value selected from-0.349974 to 0.6939225;

d is selected from any value in-0.132286 to 0.0077054.

32. The method of item 26, wherein,

the basis of the pre-stored grouping in the grouping step is as follows:

Effects of the invention

The present application establishes two mathematical models for predicting ovarian polycystic like changes (PCOM) in polycystic ovarian syndrome patients, namely, taking into account T level, 11KA4 level, A4 level and 17OHP level, or taking into account AMH level, T level and age.

Polycystic ovary syndrome (PCOS) is known for ovarian polycystic like changes (PCOM) in ultrasound in women with menoxenia and hyperandrogenism. However, PCOS diagnostic criteria recommended by the National Institutes of Health (NIH) in 1990 did not include the phenotype of ovarian polycystic change under ultrasound (PCOM). The importance of PCOM in diagnostic criteria is again underscored by the recent group of the PCOS consensus seminar sponsored by the deer Tedan European Society of Human Reproduction and Embryo (ESHRE)/American Society of Reproduction Medicine (ASRM). However, the inclusion of PCOM into the definition of PCOS remains controversial. In fact, PCOM is found in 16-25% of normal women, although it is more present in women with diluted menses and hyperandrogeny. Several findings indicate that ovulated females with PCOM are probably at the mildest end of the PCOS phenotype lineage.

In summary, ovarian polycystic like changes (PCOMs) are one of the diagnostic indicators of PCOS. However, the etiology of PCOM is not yet clear. On the one hand, the prediction of PCOM is helpful for knowing the cause of PCOM, on the other hand, the typing diagnosis of PCOS can be assisted, the severity of PCOS phenotype pedigree can be assisted to judge, and the current situation that no model for analyzing and predicting ovarian Polycystic Change (PCOM) exists in the field at present is solved.

Drawings

FIG. 1 shows the ROC curve of model 1;

fig. 2 shows the ROC curve of model 2.

Detailed Description

Anti-mullerian hormone (AMH), a hormone secreted by granulosa cells of ovarian follicles, is produced by women who begin at week 36 in the fetal period, and the greater the number of follicles in the ovaries, the higher the concentration of AMH; on the contrary, when the follicle is gradually depleted with age and various factors, the AMH concentration is also reduced, and the AMH gradually approaches 0 as the follicle approaches the climacteric.

Testosterone (T) is a typical androgen secreted by the testes of men or the ovaries of women, and the adrenal glands also secrete small amounts of testosterone, and has the effects of maintaining muscle strength and quality, maintaining bone mass density and strength, refreshing, improving physical performance, and the like.

Continuous variable: in statistics, variables can be classified into continuous variables and classified variables according to whether the variable values are continuous. The variable which can be arbitrarily valued in a certain interval is called a continuous variable, the numerical value of the variable is continuous, and two adjacent numerical values can be infinitely divided, namely, infinite numerical values can be taken. For example, the size of the parts to be produced, the height, weight, chest circumference, etc. measured by the human body are continuous variables, and the values can only be obtained by measuring or metering methods. In contrast, the value can only be calculated by natural number or integer unit, and is discrete variable. For example, the number of enterprises, the number of workers, the number of equipment, etc. can be counted only by the number of measuring units, and the value of the variable is generally obtained by a counting method.

Classification variables refer to variables in terms of geographic location, demographics, etc., which function to group survey respondents. The descriptive variable describes the distinction of a client group from other client groups. Most of the classification variables are also referred to as descriptive variables. Classification variables can be classified into two major categories, unordered classification variables and ordered classification variables. Wherein the unordered classification variable (unordered categorical variable) refers to the lack of degree and order of distinction between the classified or attributes. It can be classified into (1) two categories, such as sex (male, female), drug response (negative and positive), etc.; (2) multiple classifications, such as blood type (O, A, B, AB), occupation (worker, farmer, business, school, soldier), etc. And there is a degree of distinction between the ordered classification variables (ordinal categorical variable) of each class. For example, urine saccharification test results are shown as-, ± ++, ++ + classifying; the curative effect is classified according to cure, obvious effect, improvement and ineffective. For the ordered classified variables, the number of observation units of each group should be counted according to the hierarchical order, and the frequency table of the ordered variables (each class) should be compiled, and the obtained data is called class data.

The variable types are not constant, and various variables can be converted according to the needs of research purposes. For example, the primary numerical variable of the amount of hemoglobin (g/L) can be analyzed according to two classification data if the primary numerical variable is classified into two types according to the normal and the low of hemoglobin; if the blood glucose level is classified into five grades according to severe anemia, moderate anemia, mild anemia, normal hemoglobin increase, the blood glucose level can be analyzed according to grade data. Classification data may also be quantified, for example, the nausea response of a patient may be represented as 0, 1, 2, 3, and analyzed in terms of numerical variable data (quantitative data).

Logistic regression (logistics regression), which is a generalized linear regression analysis model, is commonly used in the fields of data mining, automatic disease diagnosis, economic prediction and the like. For example, risk factors for causing diseases are studied, and the probability of occurrence of a disease is predicted from the risk factors. Taking gastric cancer disease analysis as an example, two groups of people are selected, one group is gastric cancer group and the other group is non-gastric cancer group, and the two groups of people have different signs, life patterns and the like. Thus, the dependent variable is gastric cancer, and the value is "yes" or "no", and the independent variable may include a number of factors such as age, sex, eating habits, helicobacter pylori infection, and the like. The arguments may be either continuous or categorical. Weights for the independent variables can then be obtained by logistic regression analysis, so that it can be roughly understood which factors are risk factors for gastric cancer. And meanwhile, the possibility of cancer of a person can be predicted according to the risk factors according to the weight. The dependent variables of the logistic regression may be classified into two categories or into multiple categories.

The data fitting model as used herein is a logistic regression model that penalizes the absolute magnitudes of the coefficients of the regression model based upon the value of λ. The greater the penalty, the closer to zero the estimate of the weaker factor, so only the strongest predicted variable remains in the model.

Minimum absolute shrinkage and selection operator regression (often referred to simply as Lasso regression) is a method of compression estimation that is based on the idea of reducing the set of variables (reduced order). The method can compress the coefficients of the variables and change some regression coefficients into 0 by constructing a penalty function, thereby achieving the purpose of variable selection. The algorithm utilizes a penalty function to improve the model prediction capability, and the algorithm can solve the problems of high dimensionality and collinearity by using a 1-norm constraint, and can enable an established model to have sparsity, namely the algorithm has the effect of automatically carrying out wavelength selection in modeling.

Five-fold cross-validation, or five-fold cross-validation, is a common test method used to test algorithm accuracy. The data set was divided into five parts at the time of verification, four of which were alternately used as training data, and 1 part as test data, and the test was performed. Each test gives a corresponding correct rate (or error rate). The average value of the accuracy (or error rate) of the result of 5 times is used as an estimate of the accuracy of the algorithm, and it is generally also necessary to perform five-fold cross-validation (e.g., 5 times 5-fold cross-validation) multiple times, and then calculate the average value as an estimate of the accuracy of the algorithm.

The receiver operating characteristic curve (receiver operating characteristic curve, ROC curve for short), also called the sensitivity curve (sensitivity curve). The reason for this is that each point on the curve reflects the same sensitivity, they are all responses to the same signal stimulus, but are the results obtained at several different decision criteria. The receiver operation characteristic curve is a graph formed by taking the probability of the frightening as a horizontal axis and the probability of the hitting as a vertical axis, and a curve drawn by different results obtained by different judging standards under the specific stimulation condition.

In this application, ovarian polycystic change refers to sum of Dou Luan bubbles number >24. Wherein the sinus follicle count is between day 2 and day 5 of the menstrual cycle, ultrasonically scanning each ovary and counting the number of follicles between 2 and 10 millimeters in diameter.

The present application provides a system for predicting ovarian polycystic like changes comprising:

and a module for calculating the probability of ovarian polycystic sample change, wherein the module is used for calculating the data information acquired in the data acquisition module, so as to calculate the probability (p) of the ovarian polycystic sample change of the subject. In the module for calculating the probability of ovarian polycystic like changes, data from the subject's testosterone (T) levels, 11-keto androstenedione (11 KA 4) levels, androstenedione (A4) levels, and 17 OH-progesterone (17 OHP) levels are used to calculate the probability (p) that the subject has ovarian polycystic like changes. Specifically, in the module for calculating the probability of polycystic ovarian morphology ovarian polycystic like changes, testosterone (T) levels, 11-keto androsterone androstenedione (11 KA 4) levels, androstenedione (A4) levels, and 17 OH-progesterone (17 OHP) levels of the subject are used as continuous variables.

Wherein the testosterone (T) level refers to the testosterone concentration of the subject detected on any day in the menstrual cycle of the subject. The 11-keto androstenedione (11 KA 4) level refers to the concentration of 11-keto androstenedione in the subject measured on any day of the menstrual cycle of the subject. The androstenedione (A4) level refers to the concentration of androstenedione in the subject measured on any day of the menstrual cycle of the subject. The 17 OH-progesterone (17 OHP) level refers to the concentration of 17 OH-progesterone in a subject measured on any day of the subject's menstrual cycle. Testosterone concentration, 11-keto androstenedione concentration, 17 OH-progesterone concentration can be measured using any method known in the art, such as chromatography, chemical analysis. For example, detection may also be performed using a detection method using liquid chromatography in combination with tandem mass spectrometry, and specific detection steps may be referred to the method described in CN115541776 a.

In the module for calculating the probability of having an ovarian polycystic change, there is stored in advance a formula for calculating the probability of having an ovarian polycystic change (p) fitted based on data of testosterone (T) level, 11-keto androstenedione (11 KA 4) level, androstenedione (A4) level, and 17 OH-progesterone (17 OHP) level of the subject in the existing database. And grouping the subjects with a probability (p) of ovarian polycystic like change according to a grouping criterion.

In the present application, the existing database refers to a database composed of subjects who can acquire a treatment or who have previously received a treatment satisfying inclusion and exclusion criteria described below, and there is no contract in terms of the sample size of the database, but the larger the sample size of the database, the better, for example, 100 subjects, 200 subjects, 300 subjects, preferably 400 subjects or more, more preferably 500 subjects or more can be used.

The module for calculating the probability of having an ovarian polycystic change calculates the probability (p) that the subject has an ovarian polycystic change using the following equation (one):

p＝1/(1+e ^{-(a+b*T+c*A4+d*11KA4+f*11OHP)} ) (equation I)

In the module for calculating the probability of ovarian polycystic like changes, the testosterone (T) level, 11-keto androstenedione (11 KA 4) level, androstenedione (A4) level, and 17 OH-progesterone (17 OHP) level of the subject, and the a, b, c, d, f values are taken into the formula to directly calculate.

Further, a is any value selected from-0.194322 to 3.1655263, preferably 1.4856023; b is any value selected from 0.1138558 to 3.0958606, preferably 1.6048582; c is any value selected from-0.109174 to 0.2678637, preferably 0.0793451; d is any value from-1.320305 to-0.163216, preferably-0.741761.

The method comprises the steps that default ovarian polycystic sample changing grouping parameters are prestored in a grouping module, and grouping basis prestored in the grouping module is as follows: when the calculated probability of the subject having ovarian polycystic like changes (PCOM) (p) <20%, the risk of the subject having ovarian polycystic like changes is low risk; when 20% less than or equal to the calculated probability (p) of a subject having ovarian polycystic like changes (PCOM) is <40%, the risk of the subject having ovarian polycystic like changes (PCOM) is lower; when 40% or less of the calculated probability (p) that the subject has ovarian polycystic like changes (PCOM) is <90%, the risk that the subject has ovarian polycystic like changes (PCOM) is a higher risk; when the calculated probability (p) of a subject having an ovarian polycystic like change (PCOM) is ≡90%, the risk of the subject having an ovarian polycystic like change (PCOM) is high.

In another specific embodiment of the present application, the present application also relates to a method of predicting ovarian polycystic like changes in a polycystic ovarian syndrome patient, comprising: a data acquisition step of acquiring data of testosterone (T) levels, 11-keto androstenedione (11 KA 4) levels, androstenedione (A4) levels, and 17 OH-progesterone (17 OHP) levels of a subject; and a step of calculating the probability of having an ovarian polycystic change by calculating the above data information acquired in the data acquisition step, thereby calculating the probability (p) of having an ovarian polycystic change in the subject.

As described above, for the specific details of the steps performed in the methods of the present application, the steps performed by the modules of the system of the present application described above may be referenced for the acquisition, grouping, and processing of data for testosterone (T) levels, 11-keto androstenedione (11 KA 4) levels, androstenedione (A4) levels, and 17 OH-progesterone (17 OHP) levels of a subject.

The present application also provides another system for predicting ovarian polycystic like changes, comprising:

and a module for calculating the probability of ovarian polycystic sample change, wherein the module is used for calculating the data information acquired in the data acquisition module, so as to calculate the probability (p) of the ovarian polycystic sample change of the subject. In the module for calculating the probability of ovarian polycystic changes, data of anti-mullerian hormone (AMH) levels, testosterone (T) levels, and age of the subject are used to calculate the probability (p) that the subject has an ovarian polycystic change. Specifically, in the module that calculates the probability of polycystic ovarian morphology ovarian polycystic changes, the subject's anti-mullerian hormone (AMH) levels, testosterone (T) levels, and age are used as continuous variables.

Wherein, the anti-mullerian hormone (AMH) level refers to the concentration of anti-mullerian hormone in venous blood of a female subject on any day of menstrual cycle. For example, the subject has a menstrual cycle of 28 days, the anti-mullerian hormone (AMH) level may be the anti-mullerian hormone concentration in venous blood on day 1 of the menstrual cycle, the anti-mullerian hormone concentration in venous blood on day 10 of the menstrual cycle, or the anti-mullerian hormone concentration in venous blood on day 28 of the menstrual cycle. The testosterone (T) level refers to the testosterone concentration of the subject detected on any day in the menstrual cycle of the subject. In the module for calculating the probability of having an ovarian polycystic change, a formula for calculating the probability of having an ovarian polycystic change (p) fitted based on data of the anti-mullerian hormone (AMH) level, testosterone (T) level, and age of the subject in the existing database is stored in advance. And grouping the subjects with a probability (p) of ovarian polycystic like change according to a grouping criterion.

The module for calculating the probability of having an ovarian polycystic change calculates the probability (p) that the subject has an ovarian polycystic change using the following equation (two):

p＝1/(1+e ^{-(a+b*AMH+c*T+d*age)} ) (equation II)

Where p is the calculated probability that the subject has ovarian polycystic like changes, a, b, c, d is the no-unit parameter, AMH is the anti-mullerian hormone (AMH) level, T is testosterone level, and age.

In the module for calculating the probability of ovarian polycystic like changes, the calculation can be directly performed based on the anti-mullerian hormone (AMH) level, testosterone (T) level and age of the subject, and the value of a, b, c, d brought into formula two.

Further, a is any value selected from-2.33502 to 2.6238419, preferably 0.1444109;

b is any value selected from 0.4500201 to 0.7267895, preferably 0.5884048;

c is any value selected from-0.349974 to 0.6939225, preferably 0.1719744;

d is selected from any of-0.132286 to 0.0077054, preferably-0.06229.

In another specific embodiment of the present application, the present application also relates to a method of predicting ovarian polycystic like changes in a polycystic ovarian syndrome patient, comprising: a data acquisition step of acquiring data of anti-mullerian hormone (AMH) level, testosterone (T) level and age of a subject; and a step of calculating the probability of having an ovarian polycystic change by calculating the above data information acquired in the data acquisition step, thereby calculating the probability (p) of having an ovarian polycystic change in the subject.

As described above, for details of the steps performed in the methods of the present application, the steps performed by the modules of the system of the present application described above may be referenced for the acquisition of data on the subject's anti-mullerian hormone (AMH) level, testosterone (T) level, and age, grouping, and manner of treatment.

The system of the present application establishes two mathematical models for predicting ovarian Polycystic Changes (PCOM) in polycystic ovarian syndrome patients, respectively, i.e., predicting the conditions of ovarian polycystic changes in polycystic ovarian syndrome patients taking into account T level, 11KA4 level, A4 level, and 17OHP level, or taking into account AMH level, T level, and age. Wherein testosterone (T) levels, 11-keto androstenedione (11 KA 4) levels, androstenedione (A4) levels, and 17 OH-progesterone (17 OHP) levels are detected primarily using mass spectrometry, and if there are existing clinical mass spectrometry experimental facilities, mass spectrometry data can be obtained for testosterone (T) levels, 11-keto androstenedione (11 KA 4) levels, androstenedione (A4) levels, and 17 OH-progesterone (17 OHP) levels, but without detection of AMH indicators, models established at T levels, 11KA4 levels, A4 levels, and 17OHP levels can be used. If the detection of AMH is facilitated and conventional testosterone detection is enabled, PCOM prediction can be performed using a model built with AMH levels, T levels and age with higher detection accuracy.

Examples

Selection of experimental data

In this example, case data of an infertile woman seeking Assisted Reproductive Technology (ART) treatment at a third hospital of Beijing university was used. Patients diagnosed with PCOS and/or commonly associated metabolic disorders, i.e., obesity and insulin resistance, are included. Finally 462 women aged 20-44 years (30.1.+ -. 4.0 years) were enrolled. After the exclusion of AFC or non-recorded menstrual cycles, 458 women were included in the ovarian polycystic like change (PCOM) analysis.

The participants were routinely subjected to an Oral Glucose Tolerance Test (OGTT). AFC was measured using two-dimensional ultrasound (Hitachi Aloka, tokyo, japan). Follicles between 2 mm and 10 mm are considered sinus follicles. The participants provided written informed consent. Approval by the third medical ethics committee of the university of Beijing has been obtained (2019-014-02).

Serum steroid analysis

Classical androgens and their precursors, including T, A, dehydroepiandrosterone (DHEA) and 17 OH-progesterone (17 OHP), and four 11oxyC19 androgens (11 OHA4, 11OHT, 11KA4 and 11 KT) were measured and quantified using a sensitive high performance liquid chromatography-differential tandem mass spectrometry (HPLC-DMS/MS) instrument. The method consists of a 20A high performance liquid chromatography system (Shimadzu, chiyoda-ku, tokyo, japan) and a 5500QTrap mass spectrometer (AB SCIEX, framingham, MA, USA), and Selexion is used as a DMS component internally.

The serum sample is prepared by a dispersed magnetic solid phase extraction method (DMSPE), and nano particles (Fe3O4@GO) with a core-shell structure of magnetic graphene oxide are used as an adsorbent. An isotopic internal standard is added to the sample prior to extraction to reduce errors caused by extraction. After DMSPE, androgens were derivatized with Girard's Reagent P to increase sensitivity.

Chromatographic separation was performed on a Agilent Poroshell EC C18 column. A gradient mobile phase consisting of methanol and ammonium formate buffer was used, both solvents containing formic acid. To further distinguish the isomers and reduce background noise, DMS was introduced and parameters for each androgen were optimized. Tandem mass (MS/MS) detection electrospray sources employing positive ion mode and Multiple Reaction Monitoring (MRM) conditions for each compound were optimized.

Biochemical analysis

Plasma glucose was measured using the glucose oxidase-phenol aminophenol method (Merit Choice bioengineering company, beijing, china). Insulin was measured by chemiluminescence (Beckman DxI800, brea, calif., USA). Insulin resistance balance model assessment (HOMA-IR) was calculated using the following formula: fasting glucose (mmol/L) x fasting insulin (mU/L)/22.5.

Statistical analysis

The result variable is PCOM. Polycystic ovary is defined as the anterior follicle number of the left and right ovaries > 24. Normally distributed variables are represented by mean and standard deviation, while non-normally distributed variables are represented by median and quartile ranges. Independent sample t-tests or nonparametric tests were used to analyze the continuous variable. A Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression is applied with five-fold cross validation to construct the predictive model. The area under the receiver operating characteristic curve (AUC) was used to evaluate the performance of the model. All analyses in this example were performed using SAS JMP Pro (version 14.2; SAS Institute, cary, NC, USA), with P <0.05 considered statistically significant.

Clinical characteristics of patients

For PCOM analysis, a total of 336 patients (336/458, 73.4%) were diagnosed as having PCOM. Univariate analysis of clinical features of patients with or without PCOM is shown in table 1. Patients with PCOM were slightly younger than those without PCOM (PCOM and non-PCOM groups were 29.5±3.6 years of age vs 31.7±4.5 years of age, P <0.001, respectively). The differences in Body Mass Index (BMI) and HOMA-IR between the two groups were not statistically significant (P > 0.05). The anti-mullerian hormone (AMH), inhibin B and four androgens in the classical pathway (T, A, DHEA and 17 OHP) were statistically different between the PCOM group and the non-PCOM group (P < 0.05).

Variable selection

The clinical characteristics of the patients with or without ovarian polycystic changes collected in the examples are shown in table 1.

TABLE 1 clinical characterization of patients with or without ovarian polycystic changes

Wherein, BMI in Table 1 represents body mass index; AMH stands for anti-mullerian hormone; SHBG represents sex hormone binding globulin; HOMA-IR represents an insulin resistance balance model assessment; t represents testosterone; a4 represents androstenedione; DHEA represents dehydroepiandrosterone; 11OHA4 represents 11 beta-hydroxyandrostenedione; 11OHT means 11 β -hydroxytestosterone; 11KA4 represents 11-ketoenedione; 11KT represents 11-ketotestosterone; 17OHP represents 17 OH-progesterone.

Construction of a model

The present application built 2 models, model 1 and model 2.

In the construction of model 1, the T level, 11KA4 level, A4 level, and 17OHP level of the subject were used as variables.

Data from 458 subjects for PCOM analysis were first split into training and validation sets at a ratio of 80% to 20%. In the training set, the best model is determined using a Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression with five-fold cross validation. Raw data of the probability of the subject having PCOM and its corresponding predicted data are calculated. The contribution of each variable in the first model is shown in table 2. The main effect refers to the contribution of a single index to a model, and the total effect is the sum of the contributions of the interaction of the single index and other indexes to the model. The results in table 2 show that each of the most important contributors is T, with a weight of 51.6%. Other contributors are 11KA4, A4 and 17OHP, weighted 21.5%, 5.5% and 3.1%, respectively. The results of FIG. 1 show that the AUC of this model is 0.835 (95% confidence interval [ CI ] 0.797-0.873) in the training set and 0.824 (95% CI 0.746-0.902) in the validation set.

TABLE 2

Based on the confirmed model 1, a formula for calculating the probability (p) that the subject has PCOM, that is, formula one, which can calculate the probability (p) that the subject has PCOM based on data of the T level, 11KA4 level, A4 level, and 17OHP level of the subject, can be obtained.

Equation one: p=1/(1+e) ^{-(a+b*T+c*A4+d*11KA4+f*11OHP)} )

Where p is the calculated probability that the subject has an ovarian polycystic change, a is 1.4856023, b is 1.6048582, c is 0.0793451, d is-0.741761, f is-0.099172, T is testosterone level, A4 is androstenedione level, 11KA4 is 11-keto androstenedione level, and 17OHP is 17 OH-progesterone level.

Grouping results in the verification set based on the constructed model, and combining clinical experience to obtain the following grouping basis:

when the calculated probability of the subject having ovarian polycystic like changes (PCOM) (p) <20%, the risk of the subject having ovarian polycystic like changes is low risk; when 20% less than or equal to the calculated probability (p) of a subject having ovarian polycystic like changes (PCOM) is <40%, the risk of the subject having ovarian polycystic like changes (PCOM) is lower; when 40% or less of the calculated probability (p) that the subject has ovarian polycystic like changes (PCOM) is <90%, the risk that the subject has ovarian polycystic like changes (PCOM) is a higher risk; when the calculated probability (p) of a subject having an ovarian polycystic like change (PCOM) is ≡90%, the risk of the subject having an ovarian polycystic like change (PCOM) is high.

In model 2, AMH levels, T levels, and age were used as variables.

Data from 458 subjects for PCOM analysis were first split into training and validation sets at a ratio of 80% to 20%. In the training set, the best model is determined using a Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression with five-fold cross validation. Raw data of the probability of the subject having PCOM and its corresponding predicted data are calculated.

As shown in FIG. 2, the AUC of this model increased slightly to 0.869 in the training set (95% CI 0.834-0.904) and 0.850 in the validation set (95% CI 0.777-0.923).

Based on the confirmed model 2, a formula for calculating the probability (p) that the subject has PCOM, i.e., a formula two, which can calculate the probability (p) that the subject has PCOM based on the data of the AMH level, T level, and age of the subject, can be obtained.

Formula II: p=1/(1+e) ^{-(a+b*AMH+c*T+d*age)} )

Where p is the calculated probability that the subject has an ovarian polycystic change, a is 0.1444109, b is 0.5884048, c is 0.1719744, d is-0.06229, AMH is anti-mullerian hormone (AMH) level, T is testosterone level, and age.

Both model 1 and model 2 are effective in predicting the probability that a subject will have ovarian polycystic changes. Wherein testosterone (T) levels, 11-keto androstenedione (11 KA 4) levels, androstenedione (A4) levels, and 17 OH-progesterone (17 OHP) levels in model 1 are detected primarily using mass spectrometry, if the laboratory has an off-the-shelf clinical mass spectrometry laboratory, mass spectrometry data can be obtained for testosterone (T) levels, 11-keto androstenedione (11 KA 4) levels, androstenedione (A4) levels, and 17 OH-progesterone (17 OHP) levels, but without detection of an AMH index, model 1 can be used. If the laboratory has an AMH index and is capable of routine testosterone detection (where testosterone was initially detected by mass spectrometry, and also can be detected using chemiluminescence methods with accuracy comparable to the mass spectrometry level), PCOM prediction can be performed using model 2 with higher detection accuracy.

In this application, two potential confounding factors of obesity and insulin resistance are considered in the process of recruiting participants. High incidence of obesity/overweight and insulin resistance was found in each group, which corresponds to most PCOS patients. At the same time, because of the prevalence of high androgens in these metabolic disorder patients, this patient population is selected to ensure sensitivity and accuracy of androgen measurement. In this case, both models of the present application have good predictive power (AUC > 0.8) for PCOM.

The particular embodiments described above are illustrative only, and not limiting. Those skilled in the art, having the benefit of this disclosure, may make numerous forms, and equivalents thereof, without departing from the scope of the invention as defined by the claims.

Claims

1. A system for predicting ovarian polycystic like changes (PCOMs), comprising:

2. The system of claim 1, further comprising:

4. The system according to claim 1 or 2, wherein,

5. The system according to claim 1 or 2, wherein,

6. The system of claim 5, wherein,

the formula is the following formula one:

p＝1/(1+e ^{-(a+b*T+c*A4+d*11KA4+f*11OHP)} ) (equation I)

7. The system of claim 6, wherein,

a is selected from any numerical value in-0.194322-3.1655263;

b is selected from any numerical value in 0.1138558-3.0958606;

c is any numerical value selected from-0.109174 to 0.2678637;

d is any numerical value selected from-1.320305 to-0.163216;

f is selected from any number from-0.281078 to 0.0827339.

8. The system of claim 2, wherein,

the basis of the groups prestored in the grouping module is as follows:

9. A system for predicting ovarian polycystic like changes (PCOMs), comprising:

10. The system of claim 9, further comprising:

12. The system according to claim 9 or 10, wherein,

13. The system according to claim 9 or 10, wherein,

14. The system of claim 13, wherein,

the formula is the following formula II:

p＝1/(1+e ^{-(a+b*AMH+c*T+d*age)} ) (equation II)

15. The system of claim 14, wherein,

a is selected from any numerical value in-2.33502-2.6238419;

b is selected from any numerical value in 0.4500201-0.7267895;

c is any numerical value selected from-0.349974 to 0.6939225;

d is selected from any value in-0.132286 to 0.0077054.

16. The system of claim 10, wherein,

the basis of the groups prestored in the grouping module is as follows: