RU2638788C1 - Method for preoperative evaluation of progression development risk after radiofrequency thermal ablation of colorectal cancer hepatic metastases in case of cytoreductive surgery - Google Patents

Abstract

FIELD: medicine.

SUBSTANCE: method for preoperative evaluation of the risk of disease progression after radiofrequency thermal ablation of colorectal cancer hepatic metastases in case of cytoreductive surgery is proposed. The level of cancer-embryonic antigen, the presence of mutations in the KRAS gene and the index of metastatic liver damage are determined. The probability of disease progression is calculated by the formula. At P value of less than or equal to 0.37, a low risk of disease progression is determined, and at a value greater than 0.37, a high risk is determined.

EFFECT: invention provides a preoperative evaluation of the risk of disease progression during the first year and can be used in oncology units to determine the tactics of treating patients diagnosed with stage IV colorectal cancer with synchronous multiple bilobar metastases to the liver.

6 tbl, 2 ex

Description

, при значении Р меньше или равного 0,37 - риск прогрессии заболевания низкий, при значении больше 0,37 риск высокий.The invention relates to medicine, in particular to surgical hepatology, can be used in the treatment of metastatic liver damage. A method for preoperatively assessing the risk of progression of a disease after radiofrequency ablation of colorectal cancer metastases in the liver during cytoreductive surgery, including determining before surgery the level of cancer-embryonic antigen (X1), the presence of mutations in the KRAS gene (X2), the metastatic liver lesion index (X3), which differs calculating the probability of disease progression by the formula:

, with a value of P less than or equal to 0.37 - the risk of disease progression is low, with a value greater than 0.37, the risk is high.

The invention relates to medicine, namely to surgical hepatology, can be used to determine the tactics of treatment of metastatic liver damage.

One of the important stages in the treatment of patients with stage IV colorectal cancer is the removal of liver metastases, since the liver is one of the first organs in the hematogenous dissemination of colorectal cancer metastases (Gantsev Sh.Kh., 2008). As a result, the frequency of its defeat reaches 70%. In 25% of cases, the liver is the only organ affected by metastases (Ramia J.M. et al., 2011, Vishnevsky V.A. et al., 2010, Kaprin A.D. et al., 2014). Despite numerous studies to date, there is no clear indication for a particular volume of liver surgery in patients with single and multiple bilobar metastases, depending on the extent of the lesion. This is associated with a high risk of further disease progression after surgical treatment.

A known method for predicting the risk of progression of the disease after surgical treatment of patients with liver metastases according to computer or magnetic resonance imaging. The principle of this technique is to assess the number, size and localization of metastatic formations in the liver, on the basis of which the prognosis of the disease is made (G. G. Karmazanovsky et al. Metastatic liver cancer: materials of the Plenum of the Board of the International Society. . - 2010 .-- S. 57-58).

The disadvantage of this method is that only the volume of tumor tissue and the localization of metastases are evaluated, but the possible immunological response of the patient’s body is not taken into account, therefore, the risk of progression after radiofrequency thermal ablation (RFA) cannot be predicted.

The predecessor of the proposed method according to its technical nature is the criteria for assessing the response of solid tumors (Responist Evaluation Criteria In Solid Tumors - RECIST), adopted in 2000 by WHO and revised in 2008. This method is based on a change in the linear dimensions of the metastatic formation determined by radiation research methods, in the process of chemotherapy.

The disadvantage of this method is the assessment of changes in the linear dimensions of metastatic lesions in the liver in response to chemotherapy. This method does not give a complete assessment of the volume of metastatic liver damage at the “pre-treatment” stage, this method does not take into account the number of detected metastatic lesions, does not take into account the data of immunological blood markers, and does not take into account the presence of gene mutations.

The objective of the invention is to develop a method for preoperatively assessing the risk of progression after radiofrequency thermoablation of multiple bilobar metastases of colorectal cancer in the liver during cytoreductive surgery to remove the primary tumor by analyzing preoperative computed tomography data and the level of blood tumor markers, determining the metastatic index damage to the liver and the presence or absence of mutations in the KRAS gene, before starting any treatment method.

This problem is solved due to the fact that prior to the operation, the level of cancer-embryonic antigen in the venous blood, the index of metastatic liver damage (UTI) and the presence of mutations in the KRAS gene are determined, and the probability of disease progression is calculated by the formula:

where P is the probability that the disease will progress after radiofrequency thermal ablation of multiple bilobar metastases of colorectal cancer in the liver during cytoreductive surgery to remove the primary tumor in fractions of units; X1 - the value of cancer-embryonic antigen; X2 - the presence or absence of mutations in the KRAS gene; X3 - index of metastatic liver damage; and when a P value of less than or equal to 0.37 is obtained, the risk of developing a process progression is low, and when a P value is greater than 0.37, the risk of developing a process progression is high.

The technical result of this proposal is to assess the risk of progression after radiofrequency thermoablation of multiple bilobar metastases of colorectal cancer in the liver at the “preoperative” stage for the choice of tactics for the treatment of multiple metastases of colorectal cancer in the liver.

The proposed method is as follows. Before performing a cytoreductive surgery to remove the primary tumor with simultaneous RFA of multiple bilobar metastases in the liver, the level of cancer-embryonic antigen in the venous blood is determined. Determine the presence or absence of mutations in the KRAS gene from a biopsy obtained by endoscopic examination of the intestine. Computed tomography of the liver is performed and a UTI is calculated. The obtained data is substituted into the formula:

where P is the probability that the disease will progress after radiofrequency thermal ablation of multiple bilobar metastases of colorectal cancer in the liver during cytoreductive surgery to remove the primary tumor in fractions of units; X1 - the value of cancer-embryonic antigen; X2 - the presence or absence of mutations in the KRAS gene; X3 - index of metastatic liver damage; and when a P value of less than or equal to 0.37 is obtained, the risk of developing a process progression is low, and when a P value is greater than 0.37, the risk of developing a process progression is high.

Since it is technically impossible to model a dichotomous dependent variable, taking into account an infinite number of predictor values (X), the regression problem is formulated as follows: instead of a dichotomous variable, a continuous variable is predicted with values in the interval from 0 to 1 for any predictor values using the logit transformation:

where P is the probability that the disease will progress after radiofrequency thermal ablation of multiple bilobar metastases of colorectal cancer in the liver during cytoreductive surgery to remove the primary tumor in fractions of units; a value of 2.71 is the basis of natural logarithms;

- формула множественной линейной регрессии, где X1 - значение раково-эмбрионального антигена; Х2 - наличие или отсутствие мутаций в гене KRAS; Х3 - индекс метастатического поражения печени.

- multiple linear regression formula, where X1 is the value of the cancer-embryonic antigen; X2 - the presence or absence of mutations in the KRAS gene; X3 - index of metastatic liver damage.

The model included 77 patients who underwent RFA.

The creation of a mathematical model was aimed at determining the risk factors for relapse in the RF area and the appearance of new metastases for abdominal CT with contrast enhancement after thermal ablation. The case coding was carried out as follows:

a) during the first year of observation after surgery, there was no relapse or the appearance of new metastases - 0 (N = 40);

b) during the first year of observation after surgery, a relapse or the appearance of new metastases was revealed - 1 (N = 37).

The model included as predictors:

1. Age;

2. Sex;

3. T (2,3,4);

4. N (0,1,2);

5. STI;

6. Differentiation (high-, moderate-, low-differentiated tumors);

7. The value of CA-242 before surgery in units / ml;

8. The CEA value before surgery in ng / ml;

9. The value of CA19-9 before surgery in units / ml;

10. The presence of mutations in the KRAS gene.

Predictor coding was carried out as follows:

In this study, the mathematical model was built in the logistic regression module according to the Wald algorithm, with step-by-step inclusion in the SPSS program. When using this method, predictors by significance level are automatically included in the model step by step. The relative contribution of individual predictors was expressed by the value of the following criteria:

1. Statistics Wald;

2. X2;

3. The value of the standardized regression coefficient.

As a result, a model was obtained that included the 3 most significant predictors (X1-X3):

X1 - CEA

X2 - KRAS mutations

X3 - IMPP

The parameters of the mathematical model for predicting the development of relapse in the first year of observation after RFA.

B1-3 - regression coefficients;

B0 is a constant with a negative sign;

p is the level of significance of differences;

expB is the exponent of the regression coefficients.

All 3 predictors are statistically significantly different from 0 (significance is checked using Wald statistics). The number of degrees of freedom is 1 if the hypothesis of equality of the zero coefficient of an ordinary or index variable is tested, and for a categorical variable, it is equal to the number of values without unity (the number of corresponding index variables). In the table, the coefficients of all variables were at the level of 5%, which indicates their significance. Given the results of Wald’s statistics and standardized regression coefficients, the largest influence on the formation of metastases in the first year after RFA is performed by the UTI before surgery, an increase of this indicator by one by 19.1% increases the likelihood of relapse, in second place - CEA - an increase by one leads increased risk by 3.3%, in 3rd place - mutations of the KRAS gene.

, где

Thus, the mathematical model calculated by the formula:

where

(since B0 is a countant with a negative sign)

as follows:

, где

where

The quality of approximation of the mathematical model (the correspondence of the mathematical model to real data) is estimated using the combined tests.

Combined tests for mathematical model coefficients

A measure of the similarity function is the negative value of the doubled logarithm of this function - 2LL. The introduction or removal of predictors leads to a change in this function, the difference of these functions is denoted as an X-square and in our case is significant. Cox and Snell R-squared, Neigelkirk's R-squared are measures of certainty. These are pseudo-coefficients of determination, obtained on the basis of the relationship of the likelihood function of models with only a constant and with all coefficients. They indicate that part of the variance that can be determined using logistic regression. The measure of certainty according to Cox and Snell has the disadvantage that a value of 1 is theoretically unattainable. This disadvantage is eliminated due to the modification of this measure by the Neijelkerk method. After 4 steps, the most adequate model was obtained: criterion X - the square is 80.317 (p = 0.000), the Neijelkerk determination coefficient is 86.4%, this percentage of the variability of the variable can be explained using the used predictors.

The following is a classification table calculated on the basis of the above model. A comparison was made of the results of 77 patients obtained by performing CT of the abdominal cavity in order to identify relapse and progression of the metastatic process (observed results) and the predicted results obtained using the mathematical model. The coding of the dependent variable was carried out as follows:

a) during the first year of observation after surgery, relapse or the appearance of new metastases was not - 0;

b) during the first year of observation after surgery, a relapse or the appearance of new metastases was revealed - 1.

When evaluating the results obtained using the model, the separation point (cut-off threshold) for the occurrence and not the occurrence of an event was 0.5. If the value turned out to be less than 0.5, then we can assume that the event (in our case, the occurrence of a disease progression by 1 year) will not occur, otherwise the event is supposed to occur.

Split Point 0.5

Of the total number of patients, 77 based on the calculations, the development of progression could be predicted in 37 patients - of these, the diagnosis was confirmed by CT in 34 patients (true positive result - PI), the remaining 3 were false positive (LP) result. Progression for 1 year according to the results of the model was not detected in 40 observations, of which 3 patients with diagnosed CT metastases (false negative result - LO), the remaining 37 patients did not have new metastases and relapses (true negative result - IO). Given the data obtained, the sensitivity (Se), specificity (Sp) and the accuracy of this mathematical model were calculated at a cut-off point of 0.5.

Sensitivity (Se) = (PI / (PI + LO)) * 100 = 34 / (34 + 3) = 91.9%

Specificity (Sp) = (IO / (IO + PL)) * 100 = 37 / (37 + 3) = 92.5%

Accuracy = ((PI + IO) / (PI + LP + LO + IO)) * 100 = 34 + 37/34 + 3 + 3 + 37 = 92.21%

With a cut-off point of 0.5 for the model as a whole, 92.21% of all examined and 91.9% of those who really metastasized during the year were correctly recognized.

To achieve maximum total sensitivity and model specificity, it is necessary to determine the optimal cutoff threshold. To determine the optimal choice of the cutoff threshold, an ROC curve was constructed and analyzed.

The SPSS computer program created a table of predicted results obtained using a mathematical model, which included P values for 93 patients. The range of values was from 0 to 1.

Then, using the SPSS computer program, the sensitivity (se) and specificity (sp) values were calculated for each predicted value (P), which the program took as the cutoff threshold.

Coordinate table for ROC curve

Yellow lines are marked with a cut-off point of 0.50, which was used by default, and the corresponding sensitivity and specificity, and a cut-off point of 0.37, which corresponded to the maximum sensitivity and specificity for this group of patients. The following is a classification table calculated on the basis of a study of 77 patients using abdominal CT with contrast enhancement and the predicted results using a mathematical model with a cut-off point of 0.37.

Classification table for a training sample with a cutoff threshold of 0.37

Given the data obtained, the sensitivity (Se), specificity (Sp), and the accuracy of this mathematical model were calculated at a cutoff point of 0.37.

Sensitivity (Se) = (PI / (PI + LO)) * 100 = 36 / (36 + 1) * 100 = 97.3%.

Specificity (Sp) = (IO / (IO + PL)) * 100 = 37/40 * 100 = 92.5%.

Accuracy = ((PI + IO) / (PI + LP + LO + IO)) * 100 = 37 + 36/36 + 1 + 3 + 37 = 94.8%.

Thus, changing the separation point allowed us to increase the sensitivity of the model to 97.3% while maintaining the level of specificity. The high sensitivity of the method allows us to identify the maximum number of patients who may develop relapse, so with a separation point of 0.5, 2 patients escaped our attention and were assigned to the low-risk group, while they had relapses of 1 year, a change in the separation point fixed this error.

To finally determine the quality of the mathematical model according to the coordinate table for the ROC curve, a ROC curve was constructed (the values of Se are plotted along the ordinate axis and the value = 100% -Sp along the abscissa axis). The area under the curve (AUC) is 0.982 ± 0.011, which corresponds to the excellent quality of the model.

The method we developed allows us to combine and order the above factors for the progression of the disease after performing radiofrequency thermoablation of synchronous metastases of colorectal cancer in the liver, simplifies and systematizes the approaches to the analysis of the obtained data, which allows us to draw up the correct treatment plan.

The proposed method is illustrated by the following examples.

Patient N., 55 years old. Diagnosis: "Cancer of the sigmoid colon cT2NxM1 (hepar). 2 a C. gr. "

окружности кишки, деформирующая просвет кишки; гистологическое исследование - аденокарцинома; КТ печени: Структура печени неоднородная: в третьем сегменте определяется два гиперэхогенных образования диаметром 18 и 23 мм, в четвертом и шестом сегментах два образования схожей структуры диаметром 25 и 22 мм соответственно. Заключение: множественные метастазы в печень.Instrumental examinations: fibrocolonoscopy - an exophytic tuberous tumor, which occupies 25 cm from the anus, is determined

intestinal circumference, deforming the intestinal lumen; histological examination - adenocarcinoma; CT of the liver: The liver structure is heterogeneous: in the third segment, two hyperechoic formations with a diameter of 18 and 23 mm are determined, in the fourth and sixth segments two formations of a similar structure with a diameter of 25 and 22 mm, respectively. Conclusion: multiple liver metastases.

The metastatic liver damage index (LTI) was calculated: (1.8 + 2.3 + 2.5 + 2.2) * 4 = 35.2. Blood test data for cancer-embryonic antigen = 9.1 ng / ml. Mutations in the KRAS gene were not detected.

Substitute the obtained data in the formula:

where P is the probability that the disease will progress after radiofrequency thermal ablation of multiple bilobar metastases of colorectal cancer in the liver during cytoreductive surgery to remove the primary tumor in fractions of units; X1 - the value of cancer-embryonic antigen; X2 - the presence or absence of mutations in the KRAS gene; X3 - index of metastatic liver damage.

The result is less than 0.37, which corresponds to a low risk of progression of the disease.

The operation was performed: resection of the sigmoid colon, radiofrequency thermal ablation of four metastases in the liver. Discharged 12 days after surgery in satisfactory condition. He passed 4 courses of chemotherapy according to the FOLFOX regimen. When passing the control examination after 6 months, there is no data on the progression of the disease, which confirms the reliability of the calculations.

Patient E., 63 years old. Diagnosis: "Colorectal cancer cT3NxM1 (hepar). 2 a C. gr. "

Instrumental examinations: fibrocolonoscopy - an exophytic large-tuberous tumor is determined 12 cm from the anus, occupying 2/3 of the intestinal circumference, deforming and narrowing the intestinal lumen; histological examination - adenocarcinoma; CT of the liver: The structure is heterogeneous: in the second, third, fourth, fifth, eighth segments, hyperechoic formations with a diameter of 22 to 30 mm in diameter are determined. Conclusion: multiple liver metastases.

The metastatic liver damage index (LTI) was calculated: (2.5 + 2.2 + 3.0 + 2.6 + 2.3 + 2.4) * 6 = 90. Blood test data for cancer-embryonic antigen = 36.7 ng / ml. Mutations in the KRAS gene were detected.

Substitute the obtained data in the formula:

where P is the probability that the disease will progress after radiofrequency thermal ablation of multiple bilobar metastases of colorectal cancer in the liver during cytoreductive surgery to remove the primary tumor in fractions of units; X1 - the value of cancer-embryonic antigen; X2 - the presence or absence of mutations in the KRAS gene; X3 - index of metastatic liver damage.

The result is more than 0.37, which corresponds to a high risk of progression of the disease.

The operation was performed: rectum resection, radiofrequency thermal ablation of six liver metastases.

Discharged on the 13th day after the operation in satisfactory condition. She completed 4 courses of chemotherapy according to the FOLFOX regimen. At 6 months after surgery, during the follow-up examination, the patient was diagnosed with the progression of the disease in the form of new metastases to the liver, retroperitoneal lymph nodes, which confirms the reliability of the calculations.

The method is intended for preoperative risk assessment of the progression of the disease during the first year and can be used in oncology units to determine the treatment tactics of patients diagnosed with colorectal cancer of stage IV with synchronous multiple bilobar liver metastases.

Claims (1)

A method for preoperatively assessing the risk of progression of a disease after radiofrequency ablation of colorectal cancer metastases in the liver during cytoreductive surgery, including determining before surgery the level of cancer-embryonic antigen (X1), the presence of mutations in the KRAS gene (X2), the metastatic liver lesion index (X3), which differs calculating the probability of disease progression by the formula: P = 1 / (1 + 2.71 ^{- (0.032 * X1 + 4.67 * X2 + 0.175 * X3-15.74)} ), with a P value less than or equal to 0.37 - risk disease progression is low, with a value greater than 0.37, the risk is high .