EA037137B1 - System and method for screening the probability of presence of lung cancer - Google Patents

Abstract

The invention relates to the field of medicine, namely oncology, and can be used for screening of the probability of presence of lung cancer or detection of the given oncological disease at an early stage. Screening of the probability of presence of lung cancer is based on measuring the level of biomarkers in a biological fluid sample obtained from a subject: HE4, ApoA2, CYFRA.21.1, Ddimer, ApoA1, TTR, B2M, CA125, hsCRP, CEA, sVCAM.1, CA15.3, and determining a patient's sex with subsequent treatment of a set of derived values using at least one classification model trained to determine the probability of presence of lung cancer.

Description

(54) METHOD AND SYSTEM FOR SCREENING DETERMINATION OF PROBABILITY OF LUNG CANCER

(31) 2018140406 RUSSIA (SECHENOVSKY (32) 2018.11.15 UNIVERSITY)) (RU) (33) (43) RU 2020.05.31 (72) Inventor: Glybochko Petr Vitalievich, (/ 1) (/ 3) Applicant and patent owner: FEDERAL STATE AUTONOMOUS Svistunov Andrey Alekseevich, Fomin Viktor Viktorovich, Kopylov Philip Yurievich, Sekacheva Marina EDUCATIONAL Igorevna, Parshin Vladimir INSTITUTION OF HIGHER EDUCATION FIRST Dmitrievich, Gitel Evgeny Pavlovich, Ragimov Aligeydar MOSCOW Alekperovich, Poddubskaya Elena STATE Vladimirovna (RU) ι MEDICAL UNIVERSITY (74) Representative: | NAMED AFTER I.M. Sechenova Kupriyanova O.I. (RU) MINISTRIES (56) RU-C1-2351936 HEALTH CARE RU-C2-23 97704 RUSSIAN FEDERATION WO-A2-2013048292 (SECHENOVSK UNIVERSITY) C. BRAMBILLA et al. Early detection of (FGAOU V FIRST MGMU IM. lung cancer: role ot biomarkers, European Respiratory Journal, 2003, 21: Suppl. 39, p. 36s-44s, <DOI: THEM. SECHENOVA MINISTRY OF HEALTH 10.1183 / 09031936.02.00062002>

037137 B1

037137 Bl (57) The invention relates to medicine, namely oncology, and can be used for screening to determine the likelihood of lung cancer or to detect this cancer at an early stage. The screening determination of the likelihood of having lung cancer is based on measuring the level of biomarkers in a sample of biological fluid obtained from a subject: HE4, ApoA2, CYFRA.21.1, Ddimer, ApoAl, TTR, B2M, CA125, hsCRP, CEA, sVCAM.l, CA15.3, and information about the patient's gender, followed by processing the data set using at least one classification model trained to determine the likelihood of having lung cancer.

The technical field to which the invention relates

The invention relates to medicine, namely oncology, and can be used for screening to determine the likelihood of lung cancer, including non-small cell lung cancer, or to detect this cancer at an early stage.

State of the art

Malignant tumors represent one of the most significant health problems not only in Russia, but throughout the world.

Cancer is the second most common cause of death in Russia. Average incidence of malignant neoplasms in 2016 amounted to 408.6 people. per 100,000 population. The average mortality rate is 201.6 people. per 100,000 population. The absolute number of deaths is 295,729 people. Cancer incidence is growing all over the world. Over the past 10 years, it has increased by more than 20%.

Non-small cell lung cancer ranks first in terms of prevalence among the male population of industrialized countries (Claudia Allemani, Hannah K Weir, Helena Carreira et al. Global surveillance of cancer survival 1995-2009: analysis of individual data for 25 676 887 patients from 279 population-based registries in 67 countries (CONCORD-2). Lancet 2014; (November 26). doi: http://dx.doi.org/ 10.1016 / S0140-6736 (14) 62038-9). The statistics of developed countries of the world indicate a steady increase in newly diagnosed cases of lung cancer in comparison with malignant tumors of any other localization.

Lung cancer in Russia also occupies a leading position in the structure of cancer morbidity and mortality. Over 63,000 people in Russia get lung cancer annually, including over 53,000 men. More than 20,000 patients, or 34.2%, at the time of diagnosis, have advanced stages of the tumor process, in which the results of treatment remain unsatisfactory. Analysis of surgical treatment failures showed that the most common cause of death in operated patients is hematogenous metastases (60-70%) and loco-regional recurrence (3040%).

Thus, the development of new available screening methods for the early diagnosis of lung cancer is a very urgent task.

A common screening strategy is an annual chest x-ray, especially in smokers. However, in a large clinical trial PLCO (Prostate, Lung, Colorectal and Ovarian Cancer screening) it has been shown that such screening does not affect mortality from lung cancer in the population examined (Oken et al., Screening by chest radiograph and lung cancer mortality: the Prostate, Lung, Colorectal, and Ovarian (PLCO) randomized trial. JAMA. 2011 Nov 2; 306 (17): 1865-73). This statement undoubtedly applies to the method of X-ray diagnostics of fluorography used in Russia. The study is performed in one projection and, undoubtedly, is even less informative than a chest x-ray.

Currently, the most effective method of screening diagnostics in the world is low-dose spiral computed tomography (SSCT). In a large clinical study, the National Lung Screening Trial (NLST), it was found that annual STCT leads to a 20% reduction in lung cancer mortality compared to annual chest x-ray (National Lung Screening Trial Research Team et al., Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med. 2011 Aug 4; 365 (5): 395-409). The technique is recommended for lung cancer screening in the United States. The American Thoracic Surgery Association recommends annual screening in the 50 to 79 age group for patients with 20 years of smoking history and additional comorbidities that increase their overall cancer risk by 5% over the next 5 years.

However, high-quality CT screening is possible only if highly qualified specialists and modern devices are available, which are available only in large medical institutions. It is also a negative fact that repeated studies are associated with the risk of additional radiation exposure.

As an alternative to the above instrumental imaging methods, diagnostic methods based on the determination of biochemical markers in biological tissues and patient fluids, for example, whole blood, serum or plasma, can be used. As such markers, for example, various antigens, proteins and metabolites secreted by malignant cells or formed during their death can be used. Thus, at present, for the diagnosis of lung cancer, the most widely used definitions are CYFRA 21-1 (fragment of cytokeratin 19) and CEA (cancer embryonic antigen) in blood plasma; other biomarkers are also known (Zamay et al., Current and Prospective Protein Biomarkers of Lung Cancer. Cancers 2017, 9, 155). It should be noted that the diagnosis of oncological diseases based on measurements of single biomarkers is not sufficiently reliable due to their low sensitivity. For example, the sensitivity and specificity of CYFRA 21-1 in the diagnosis of lung cancer is 43% and 89%, CEA - 69% and 68%, respectively (Zamay et al., Current and Prospective Protein Biomarkers of Lung Cancer. Cancers 2017, 9, 155). The use of multiplex diagnostic methods, implying an assessment of the risk of the presence of a disease based on measurements of several biomarkers, allows one to overcome this problem and achieve more reliable results.

So, for example, from KR-10-2016-0113444 (prototype) it is known to determine the presence of lung cancer by the markers of the following proteins measured in blood serum: HE4, RANTES, sVCAM-1, LRG1, CEA, CYFRA 21-1, ApoA2, ApoA1, TTR, B2M, CA125, CA19-9, hsCRP. In this case, the risk of having a disease is assessed using the logit regression method based on a set of measurements of the above biomarkers.

Despite the possibility of using a complex of markers in the technique that increase its diagnostic value in assessing the risk of developing lung cancer, there is a need to adapt the technique for different groups of subjects. In the literature, interracial differences in the molecular mechanisms of lung cancer have been noted, which casts doubt on the feasibility of using a single set of biomarkers for different races. Thus, it was shown that the incidence of mutations in the epidermal growth factor receptor is higher in the Asian population than in the Caucasian population, while the incidence of KRAS mutations is lower (MV Schabath, D. Chress, T. Munoz-Antonia. Racial and Ethnic Differences in the Epidemiology and Genomics of Lung Cancer. Cancer Control. 2016 Oct; 23 (4): 338346). Such interracial differences can be caused by both environmental factors (air pollution level), behavioral characteristics (dietary habits, smoking prevalence), and genetic predisposition (W. Zhou and D. C. Christiani. East meets West: ethnic differences in epidemiology and clinical behaviors of lung cancer between East Asians and Caucasians. Chin J Cancer. 2011 May; 30 (5): 287-292).

The claimed invention is based on the study of a new complex of markers, which makes it possible to increase the accuracy and reliability of determining the presence of the disease when screening lung cancer in a particular patient of the Caucasian population, the formation on this basis of a particular risk group and the identification of those patients who need an in-depth expensive examination to detect early stages of lung cancer.

Disclosure of invention

The technical problem solved by the present invention is to provide a more accurate method for determining the likelihood of having lung cancer in a Caucasian population.

The achieved technical result is an increase in the accuracy of screening detection of the presence of cancer in a particular patient of the Caucasian population, and already at the early stages of its development by identifying and taking into account the original set of biomarkers based on the results of the analysis of the serum or blood plasma fraction while accelerating the diagnosed conditions.

The technical result is achieved by implementing a method for screening determination of the likelihood of lung cancer, including measuring the level of biomarkers in a sample of biological fluid obtained from a subject: HE4, ApoA2, CYFRA.21.1, Ddimer, ApoA1, TTR, B2M, CA125, hsCRP, CEA, sVCAM. 1, CA15.3, as well as determining the sex of the patient, with subsequent processing of the set of obtained biomarker values using at least one classification model trained to determine a high or low probability of having lung cancer.

As classification models, the random forest method and / or linear discriminant analysis and / or support vector machine are used.

A trained classification model is obtained through the following steps:

form training and test samples of records of subjects with measured values of biomarkers HE4, ApoA2, CYFRA.21.1, Ddimer, ApoA1, TTR, B2M, CA125, hsCRP, CEA, sVCAM.1, CA15.3), including records of patients of different sex and age ;

teaching the classification model to identify a given pathology using the records of the training and test sample;

save the connections and weights of the trained classification model for subsequent determination of the likelihood of having lung cancer based on the processing of the measured data of the biomarkers of the subject.

When forming the training and test samples, they include records of subjects with identified pathology - the presence of cancer and the absence of lung cancer.

The technical result is achieved through the implementation of a screening system for determining the likelihood of lung cancer, including a module for inputting measured values of biomarkers of a subject;

data storage module configured to store training and test samples of the classification model, connections and weights of the trained classification model, records of subjects with measured values of biomarkers HE4, ApoA2, CYFRA.21.1, Ddimer, ApoA1, TTR, B2M, CA125, hsCRP, CEA, sVCAM.1, CA15.3, including records of patients of different sex and age;

a trained classification model module, configured to build and train at least one classification model to determine the presence of a given pathology by said markers taken from the data storage module;

diagnostic module adapted to process the entered biomarker values

- 2,037137 subjects using at least one trained classification model;

a data output module configured to obtain data on a high or low likelihood of having lung cancer.

The accuracy of the proposed multiplex method for diagnosing lung cancer is ensured through the use of a complex of 12 biomarkers and information about the patient's field, as well as through the use of several classification models with subsequent averaging of model results.

Brief Description of Drawings

The invention is illustrated by drawings, where:

in fig. 1A is a scatter plot of patient age versus biomarker concentration. Points are individual measurements, lines are linear regression model predictions. The graphs show the values of the correlation coefficients calculated by the Pearson method and the P-values calculated by the Student's test; in fig. 1B shows a range diagram for assessing the significance of gender differences in biomarker concentrations. The graphs show the P-values obtained using the Student's t test. Data for women are shown in gray, data for men are shown in black;

in fig. 2 - ROC curves for assessing the predictive ability of individual biomarkers (the line type corresponds to the biomarker);

in fig. 3. - examples of decision trees obtained as a result of training the multifactorial classification algorithm random forest on experimental data for 12 biomarkers;

in fig. 4 - visualization of the results of dividing patients into 2 classes (healthy donors and patients with lung cancer) using linear discriminant analysis for 12 biomarkers;

in fig. 5 - examples of three-dimensional projections of dividing the pooled patient population into 2 classes (healthy donors and patients with lung cancer) using the support vector machine for 12 biomarkers;

in fig. 6 - the proportion of classifiers stratified by AUROC depending on the number of biomarkers included in them. The training was carried out using: A. Support vector method; B. Linear Discriminant Analysis;

in fig. 7 - ROC curves for assessing the predictive ability of various classification algorithms: A. The entire data set was used both for training the model and for its validation; B. 80% of the data was used for training the model, 20% for validation;

in fig. 8 is a block diagram of a system for assessing the likelihood of having lung cancer based on patient data;

in fig. 9 illustrates an algorithm for evaluating the likelihood of having lung cancer based on patient data.

Implementation of the invention

The initial group of biomarkers used in the diagnostic test to determine the likelihood of having lung cancer (LC) was obtained using a multivariate classification model. Such methods make it possible to find combinations of biomarkers with the greatest diagnostic potential. The mathematical model is trained on experimental measurements of a given set of biomarkers obtained from a mixed sample of healthy volunteers and patients with PD. The trained model can be used to assess the risk of having a disease in a patient based on the indicators of his biomarkers.

As part of the work carried out at the stage of developing a diagnostically significant set of indicators, the measurement data of 16 biomarkers were used (AFP, CEA, CA 19-9, CA 125, HE4, tPSA, CA 15-3, B2M, hsCRP, Ddimer, CYFRA 21-1, ApoA1, ApoA2, Apo B, TTR, sVCAM-1) obtained on a sample of healthy volunteers of the Caucasian population (n = 203, 104 women and 99 men 36-80 years old, mean age 53 years) and patients with lung cancer (n = 77 , 25 women and 52 men 36-80 years old, average age 62 years).

Statistical processing of experimental data and the development of classification models were carried out in the R {RDevelopmentCoreTeam environment (2007). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0}.

The first step was statistical analysis and data visualization. For healthy volunteers, the effect of gender and age on biomarker indices was assessed (Fig. 1).

Based on the results of the study, at this stage, it was concluded that there was no significant correlation between the age of patients and the indicators of most biomarkers, at the same time, there were significant gender differences in the indicators of CEA, CA 19-9, CA 125, HE4, Ddimer, ApoA1 and TTR.

At the next stage, the significance of differences in the levels of individual biomarkers between healthy volunteers and patients with lung cancer was assessed using the Student's test after normalization of the experimental data by log-transformation (Table 1).

- 3 037137

Table 1. Comparison of biomarker indices in blood plasma of healthy donors and patients with lung cancer

Biomarker Healthy donors Lung cancer patients P-values ¹ P-values ¹ (prototype) AFP, IU / ml 2.84003 2.857935 0.3 - ApoAl, g / l 1.59803 1.718442 4e-04 *** <2.2e -16 *** AroA2, g / l 0.2953 0.218312 1e-13 *** <2.2e-16 *** ApoB, g / l 1.026404 0.94039 0.008 ** - B2M, ng / ml 1477.291 1887.649 Ze-09 *** <2.2e-16 *** CA125, IU / ml 10.68987 44.81299 4e-06 *** 2.2e -14 *** CA15.3, IU / ml 15.12623 26.86701 0.004 ** - CA19.9, IU / ml 6.597143 13.31935 0.08 0.2829 CEA, ng / ml 1.856232 18.98655 2e-05 *** <2.2e-16 *** CYFRA.21.1, ng / ml 1.374517 5.257714 2e-11 *** <2.2e-16 *** Ddimer, ng / ml 119.7537 462.6623 3e-13 *** - HE4, pmol / l 51.42532 104.9732 8e-21 *** <2.2e-16 *** hsCRP, mg / l 1.770049 11.55325 5е-08 *** <2.2e-16 *** sVCAM.1, ng / ml 658.3713 815.4416 2e-04 *** 0.01928 * t.PSA, ng / ml 1.129717 1.735423 0.3 - TTR, mg / dl 25.64039 18.96104 2e-09 *** <2.2e-16 ***

¹ - comparison according to Student's test after normalization of experimental data;

* - P-values <0.05, * * - P-values <0.01, * ** - P-values <0.001.

Based on the analysis, it was concluded that there were no significant differences in the concentrations of AFP, CA19.9, t.PSA between healthy volunteers and patients with lung cancer. Also, within the framework of this study, a significant difference was noted in the concentrations of Ddimer, ApoB and CA15.3 not included in the prototype, as well as a more significant difference in sVCAM.1 compared to the prototype study.

The logistic regression method was used to assess the diagnostic value of individual biomarkers. In these statistical models, the relationship between the concentration of a biomarker and the likelihood of having a disease was considered (equation 1):

where P (Y) is the probability of having a disease, b ₀ and b ₁ are coefficients determined from experimental data, X is a predictor (biomarker concentration).

The predictive power of logistic models was assessed using ROC analysis, which involves determining the sensitivity, specificity and accuracy of a method relative to a test or general dataset. For this, the value of the threshold probability that determines the presence of the disease varied from 0 to 1 with a given step, for each step, the proportion of correctly diagnosed cases of the disease (sensitivity) (Se), correctly identified cases of absence of the disease (specificity) (Sp) was calculated, and also the total proportion of correctly diagnosed cases, both the presence and absence of disease (accuracy) (Acc), (equations 2-4):

S _e = fn) 100% (2)

Sp = + pp) 100% (3)

Acc = (TN + TP) / (TN + FP + TP + FN) 100% (4) where TR - correctly classified positive result (correctly diagnosed disease), FP - false positive result (misdiagnosed disease), TN - correctly classified negative result (correctly diagnosed absence of disease), FN - false negative result (misdiagnosed absence of disease).

The resulting set of sensitivity and specificity values was used to construct the ROC curve. The area under the ROC curve (AUROC) was used as an integral indicator of the quality of the models: the predictors with the maximum predictive ability show the highest AUROC values. The results of the ROC analysis are shown in FIG. 2 and in table. 2.

- 4 037137

Table 2. Diagnostic value of selected biomarkers

I Biomarker 1 Specificity,% I Sensitivity,% 1 Accuracy,% AUROC HE4 90 82 ί. 88 ... , 0. ^ '’L 'AroA2 97 74 90 0.88 CYFRA.21.1 91 61 ί. ⁸² __. 0.83 ... Ddimer 67 84 71 0.82 M. MM ^ M .... · H “MM · M MMM "" Μ ^ mmmm Μ. Μ Μ ApoAl 81 70 f 78 0.81 TTR 85 60 '78 0.8 B2M 58 ¹ ; 83 ! 65 0.76 CA125 82 56 75 0.73 Mn * MMMMMMM. "Μ m m. ^ M ^ MMMM · · -. - a ^ MM MM ^ M · ΜΜ Μ —..... ..Μ · Μ Μ -. • · Μ · Μ · · · hsCRP 85 56 ; 77 0.72 4 CEA 83 52 75 0.68 sVCAM.1 88 45 ί 76 θ CA15.3 69 49 64 0.6 M l> «ΜΜΜΜ MMHM—- MM M • - ·· Μ · Μ Μ Μ Μ Μ Μ Μ ^ Μ i ApoB. 53 66 J __5θ __ 0.6 CA19.9 71 47 64 0.59 AFP 48 69 1 54- 0.56 t.PSA | 86 | 40 1 70 0.56

Based on the results of statistical data analysis and assessment of the predictive ability of univariate logistic models, biomarkers were selected, which were subsequently included in multivariate classification models. The criteria for the inclusion of biomarkers were pval <0.005 (Table 1) and AUROC> 0.6 (Table 2). Thus, to construct classification models, experimental measurements of 12 biomarkers (HE4, ApoA2, CYFRA.21.1, Ddimer, ApoA1, TTR, B2M, CA125, hsCRP, CEA, sVCAM.1, CA15.3) were selected, and information on the field the patient.

The development of multivariate classification models was the final stage of the study. Various machine learning methods (random forest, linear discriminant analysis, support vector machine) were used in the current task. The estimation of the parameters of the models (training) was carried out on the combined data obtained on healthy volunteers and patients with lung cancer, and was aimed at minimizing the predictive errors of the algorithm. A detailed description of the methods used is presented in the book (Bishop CM, Pattern recognition and machine learning. Springer. 2006).

The random forest (RF) method involves creating a collection of cross-validated decision trees. Each of these trees is trained on a subsample of data, which includes information only on a part of biomarkers and observations, and is validated on a subsample that was not used for its construction (bagging). Based on the predictions of each of the constructed decision trees, the patient is assigned to one of the groups (healthy donors or patients with lung cancer), the final prediction of the classifier is determined by the majority of votes of the constructed trees (see Fig. 3 A, B).

The use of linear discriminant analysis (LDA) involves the search for a linear combination of biomarkers - discriminants that provide the best division of the entire population of subjects into healthy volunteers and patients with lung cancer. The linear discriminant can be calculated: z (x) = β ^ + ... + enxn, where xi are the concentrations of the i-th biomarker, βί are the model coefficients. This problem is solved by finding the axis, the projection on which provides the maximum ratio of the total variance of the linear combination of biomarkers in the sample to the sum of the variances of the linear combination of biomarkers within the classes (see Fig. 4).

Table 3. Values of linear coefficients for discriminant (LDAcomponent 1)

Factor Coefficient SEX (M) 1.78E-01 CEA -3.53E-03 CA125 7.98E-03 ΗΕ4 2.48E-02 CA15.3 -9.30E-03 Β2Μ 3.23E-04 hsCRP -1.58E-03 Ddimer -3.00E-05 CYFRA.21.1 -5.24E-02 ApoAl 1.94E-02 ApoA2 -8.02E + 00 TTR -6.34E-02 sVCAM.1 9.86E-05

- 5 037137

The use of the support vector machine (SVM) assumes finding an (n-1) -dimensional hyperplane dividing the n-dimensional space of biomarker values into two classes. Let there be a training set (x ₁ , y ₁ ), ... (x _n , y _n ), xi € R ⁿ , yi e {-1,1}, where xi is a vector of biomarker values, and yi determines the patient's identity to the class. The classifying function can be defined as F (x) = sign ((w, x) + b), where w is the normal vector to the dividing hyperplane, b is an auxiliary parameter, and the function can take values 1 or -1 depending on the class of the object ... Algorithm training implies the search for such a hyperplane that provides the smallest empirical classification error and maximizes the distance between the biomarker values of patients belonging to different classes (see Fig. 5).

At the first stage of constructing multivariate models, the diagnostic value of various combinations of biomarkers from the above group was studied. For this, all possible combinations, including from 2 to 12 biomarkers, were used to build classification models (4803 variants). Pooled data obtained from healthy volunteers and patients with lung cancer and methods of linear discriminant analysis and support vectors were used for training. The developed models were ranked in accordance with their predictive potential, assessed by the AUROC indicator (Fig. 7, Table 4).

As seen in FIG. 6, complex tests including 11-12 biomarkers have the highest predictive power, while for a relatively small proportion of classifiers that include combinations of 2-3 biomarkers, the AUROC indicator is more than 80%.

The final phase of constructing classifiers was their validation.

The pooled data obtained from healthy volunteers and patients with lung cancer were randomly divided into training and test samples. The estimation of the parameters of the models (training) was carried out on a training set and was aimed at minimizing the predictive errors of the algorithm. Validation of trained models consisted in assessing their predictive ability on a test sample. The predictive power of multivariate classification models was assessed using ROC analysis, as was done previously for individual biomarkers (Fig. 7, Table 4).

Table 4. Diagnostic value of trained multivariate classification models

Method Sensitivity,% Specificity,% Accuracy,% Threshold probability,% AUROC The entire dataset was used to both train the model and validate it. RF 100 100 100 fifty one LDA 69 97 89 fifty 0.93 SVM 79 98 92 fifty 0.99 80% of the data was used for training the model, 20% for validation RF 79 100 93 fifty 0.99 LDA 58 95 83 fifty 0.94 SVM 74 95 88 fifty 0.98

The final classification models are trained algorithms that predict the likelihood of having lung cancer based on experimental measurements of patient biomarkers, taking into account gender differences.

The final decision - determining the likelihood of having lung cancer - is calculated as the median of the lung cancer probabilities calculated in 3 classification models (RF, LDA SVM) trained on the entire sample of patients (see, for example, Kittler J, Hatef M, Duin RPW et al , On Combining Classifiers. IEEE Transactions on Pattern Analysis and Machine Intelligence, VOL. 20, NO. 3, MARCH 1998 226-39.)

To implement the proposed method, software was developed that allows, based on the data of a particular patient (gender and the results of biomarker measurements), to calculate the likelihood of having lung cancer. A block diagram of an embodiment of the invention is shown in FIG. 8.

The computer-implemented system consists of (1) an interface that includes a device for inputting patient data (gender and results of biomarker measurements) and outputting calculation results (the likelihood of lung cancer); (2) a memory block containing trained classifier models and software products necessary to work with them (R portable, Google Chrome Portable), and (3) about

- 6 037137 gram module, with the help of which the program code necessary for the exchange of data between the interface and the memory unit is implemented. The shiny package was used to create the GUI (Winston Chang, Joe Cheng, JJ Allaire, Yihui Xie and Jonathan McPherson (2017). Shiny: Web Application Framework for R. R package version 1.0.5.https: //CRAN.R- proiect.org/package-shiny) based on the R {RDevelopmentCoreTeam (2007) environment. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0}. To work with this package, you need the R portable and Google Chrome portable software stored in the memory block. To work with the proposed models, the following packages are required: (1) RandomForest (A. Liaw and M. Wiener (2002). Classification and Regression by randomForest. R News 2 (3), 18-22); (2) MASS (Venables, W. N. & Ripley, B. D. (2002) Modern Applied Statistics with S. Fourth Edition. Springer, New York. ISBN 0-387-95457-0); (3) e1071 (David Meyer, Evgenia Dimitriadou, Kurt Hornik, Andreas Weingessel and Friedrich Leisch (2017). E1071: Misc Functions of the Department of Statistics, Probability Theory Group (Formerly: E1071), TU Wien. R package version 1.6- 8.https: //CRAN.Rproject.org/package=e 1071).

An algorithm for assessing the likelihood of having lung cancer based on patient data is shown in FIG. nine.

Patient data are entered through the interface and fed as input variables to the developed models, in each of which the probability of having lung cancer is calculated. Further, based on the results of the model predicted, the average value is calculated, which is displayed in the output window.

The diagnostic multiplex panel for assessing the risk of lung cancer includes biomarkers that showed the maximum predictive potential in the study (Fig. 2, Table 2): HE4, ApoA2, CYFRA.21.1, Ddimer, ApoA1, TTR. In addition, the claimed complex includes additional biomarkers with a lower predictive potential, however, significantly different between healthy volunteers and patients with lung cancer (Table 1): B2M, CA125, hsCRP, CEA, sVCAM.1 and CA15.3 in the studied population.

Below is a description of one of the clinical examples of the application of the method, which confirms the possibility of implementing the invention with the achievement of a technical result.

Example 1. Patient K., 54 years old. Smokes 35 years.

In January 2018, in connection with complaints of weakness and rapid fatigability, he turned to the polyclinic at his place of residence.

Was examined by a therapist. The recommended intake of vitamins, complete blood count, in which clinically significant abnormalities were not identified.

The patient was asked to take part in the Oncopoisk program.

The patient was examined as part of the program. The following blood serum test results were obtained: AFP 2.4 IU / ml, CEA 2.1 ng / ml, CA 19-9 3.6 IU / ml, CA 125 9.7 IU / ml, HE4 110.2 pmol / l , tPSA 0.65 ng / ml, CA 15-3 19.2 IU / ml, B2M 2154 ng / ml, hsCRP <0.08 ng / ml, D-dimer 51.0, CYFRA 21-1 1.28 ng / ml, Apo A-1 1.38 g / l, Apo A2 0.289 g / l, Apo B 1.15 g / l, TTR (prealb) 25.0 mg / dl, sVCAM-1 812 ng / ml, Rantes 40784 pg / ml, VEGFR1 135 pg / ml.

When processing the results obtained by the claimed method, a high probability of lung cancer was revealed.

Contrast-enhanced X-ray CT was performed. Revealed a formation in the lower lobe of the right lung 13x12 mm, with uneven heavy contours, non-uniformly accumulating contrast agent. Mediastinal lymph nodes are not enlarged. The patient is hospitalized for surgical treatment. Performed video-assisted thoracoscopy, resection of the lower lobe of the right lung, mediastinal lymph node dissection. Gistol.№ highly differentiated adenocarcinoma of the lung. In 5 remote l / nodes - no signs of metastatic growth.

Claims (5)

1. A method of screening for determining the likelihood of lung cancer, including measuring the level of biomarkers in a sample of biological fluid obtained from a subject: HE4, ApoA2, CYFRA.21.1, Ddimer, ApoA1, TTR, B2M, CA125, hsCRP, CEA, sVCAM.1, CA15 .3, as well as determining the sex of the patient, followed by processing the set of obtained biomarker values using at least one classification model trained to determine a high or low likelihood of lung cancer. 2. The method according to claim 1, characterized in that the classification models are random forest and / or linear discriminant analysis and / or support vector machines. 3. The method according to claim 1, characterized in that the trained classification model is obtained by implementing the following steps: form training and test samples of records of subjects with measured values of bio- 7 037137 markers HE4, ApoA2, CYFRA.21.1, Ddimer, ApoA1, TTR, B2M, CA125, hsCRP, CEA, sVCAM.1, CA15.3, including records about patients of various gender and age; teaching the classification model to identify a given pathology using the records of the training and test sample; save the connections and weights of the trained classification model for subsequent determination of the likelihood of having lung cancer based on the processing of the measured data of the biomarkers of the subject. 4. The method according to claim 3, characterized in that during the formation of the training and test samples, records of subjects with identified pathology are included - the presence of cancer and the absence of lung cancer. 5. A screening system for determining the likelihood of lung cancer, including a module for entering the measured values of the subject's biomarkers HE4, ApoA2, CYFRA.21.1, Ddimer, ApoA1, TTR, B2M, CA125, hsCRP, CEA, sVCAM.1, CA15.3; a data storage module configured to store training and test samples of the classification model, connections and weights of the trained classification model, records of subjects with measured biomarker values, including records of patients of different sex and age; a trained classification model module, configured to build and train at least one classification model to determine the presence of a given pathology by said markers taken from the data storage module; a diagnostic module configured to process the inputted biomarker values of the subject using at least one trained classification model; a data output module configured to obtain data on a high or low likelihood of having lung cancer.