RU2850819C1 - Method for predicting common survival in surgical treatment of patients with recurrent non-generalized resectable oral cancer - Google Patents

Abstract

FIELD: medicine.

SUBSTANCE: invention relates to medicine, specifically to oncology, and can be used to predict overall survival after surgical treatment for recurrent oral cancer. A morphological examination of the calcified tissues is performed. The following predictors are evaluated: type of progression: locoregional, local or regional; duration from previous treatment to relapse; prevalence in the primary focus (T); prevalence in regional lymph nodes (N); volume of resection - R1 or R2. The obtained data is entered into the online calculator "Medical prognostic model of patient survival in case of oral cancer relapse" and the overall survival is determined.

EFFECT: method allows for an objective assessment and prediction of the overall survival of patients with recurrent oral cancer when they undergo surgical treatment, improving the selection of patients for it, and helping in making decisions in case of an unfavorable prognosis of the inexpediency of the operation or its combination with other methods in order to achieve better results by evaluating a set of the most significant indicators.

1 cl, 20 dwg, 2 ex

Description

The invention relates to the field of information technology, namely to medicine, in particular to oncology, and can be used to predict progression-free and overall survival after surgical treatment for recurrent oral cancer, using nomograms and an online calculator created using a program for electronic computers, "Medical prognostic model for patient survival in case of recurrent oral cancer", located in electronic communication networks.

The incidence of malignant neoplasms of the oral cavity increased from 7970 cases in 2013 to 9819 cases in 2023, a 10-year increase of 16.14%. In 2023, 65.1% of patients were diagnosed with stages III and IV (Kaprin A.D., Starinsky V.V., Shakhzadova A.O., 2023). Oral cancer, despite visual localization, educational activities on oncological alertness, early diagnosis, the development of surgical, radiation and drug treatment methods, is characterized by a high number of relapses (up to 25-37.5% in the early stages and up to 50-60% in locally advanced disease). Progression in the primary lesion is observed in 30-63% of cases, in regional areas of the neck - in 24-51% of cases, locoregional - in 4-16% of cases (van Weert S. et al., 2021; Ionna F. et al., 2021; Shetty K.S.R. et al., 2022; Hassan N.H. et al., 2022; Hartl D.M. et al., 2023; Struckmeier A.K. et al., 2024). The best five-year overall survival rates are observed after surgical treatment; according to meta-analyses, they are 40.2%-56.75% (González-García R., 2016; Chang J.H. et al., 2017; Hamoir M.I. et al., 2017; Ionna F. et al., 2021; Li Y. et al., 2022; Zhou B. et al., 2024). However, when performing operations on locally advanced tumor processes in cicatricial and/or previously irradiated tissues, such interventions may be accompanied by an increased risk of complications, be accompanied by disability, and, in addition, with limited reserves of adjuvant treatment, may not lead to satisfactory oncological results. It is relevant to conduct studies to predict the results of treatment of patients with recurrent oral cancer in order to determine the optimal tactics. Recently, there has been increased interest in the graphical representation of predictive models and the creation of online calculators.

Nomograms for determining three- and five-year tumor-specific and overall survival are known [Nie Z., Zhao P., Shang Y., Sun B. Nomograms to predict the prognosis in locally advanced oral squamous cell carcinoma after curative resection. BMC Cancer. 2021 Apr 7; 21(1): 372. doi: 10.1186/s12885-021-08106-x]. The authors included 269 patients with primary squamous cell carcinoma of the oral cavity, stages III and IV, in the study. Cox proportional regression models were used to identify independent prognostic factors of tumor-specific and overall survival. Statistically significant factors were used to create nomograms based on the R programming language. The reliability of the nomogram was assessed using the C-index values and calibration curves. For tumor-specific survival, the following were significantly significant predictors in multivariate analysis: age (OR=1.05, 95% CI: 1.029-1.071, p<0.001), Kaplan-Feinstein comorbidity index greater than 2 (OR=1.706, 95% CI: 1.109-2.624, p=0.015), prevalence of pT at the primary site (p=0.004) (T2 – OR=3.122, 95% CI: 1.252-7.782, p=0.015; T3 – OR=5.073, 95% CI: 2.077-12.392, p<0.001; pT4 – OR=4.428, 95% CI: 1.762-11.129, p=0.002), the number of metastatic lymph nodes ((p=0.009), 1-2 – OR=1.432, 95% CI: 1.003-2.044, p=0.048; 3-4 – OR=1.679, 95% CI: 1.097-2.569, p=0.017; more than 5 – OR=2.413, 95% CI: 1.349-4.318, p=0.003), systemic inflammatory index ≥535.5 (OR=2.214, 95% CI: 1.289-3.805, p=0.004). For overall survival in multivariate analysis, the following independent significant predictors were: age (OR = 1.023, 95% CI: 1.012-1.033, p < 0.001), Kaplan-Feinstein comorbidity index greater than 2 (OR = 1.458, 95% CI: 1.137-1.871, p = 0.003), pT ((p < 0.001); T2 - OR = 2.128, 95% CI: 1.303-3.477, p = 0.003; T3 - OR = 3.157, 95% CI: 1.945-5.124, p < 0.001; T4 - OR = 4.316, 95% CI: 2.674-6.966, p<0.001), the number of metastatic lymph nodes ((p=0.009); 1-2 – OR=1.357, 95% CI: 1.013-1.819, p=0.041; 3-4 – OR=1.549, 95% CI: 1.106–2.171, p=0.011; ≥5 – OR=1.786, 95% CI: 1.202-2.653, p=0.004), systemic inflammatory index ≥535.5 (OR=1.599, 95% CI: 1.25-2.047, p<0.001). A nomogram for three- and five-year tumor-specific and overall survival was created and presented in the publication (Fig. 1). The C-score values were 0.712 for overall survival and 0.709 for tumor-specific survival. All patients were divided into three subgroups based on the total score calculated using the nomograms. For overall survival, patients in the low-risk group had a score of 145 points or less, in the intermediate-risk group, 145-190 points, and in the high-risk group, more than 190 points. For tumor-specific survival, the scores for the same groups were 102 points or less, 102-141 points, and more than 141 points, respectively.

The disadvantages of this prognostic scale are:

- performing statistical analysis on patients with primary localization at stage III and IV tumors;

- lack of an online calculator.

A nomogram for determining three- and five-year overall survival is known [Zhang X.Y., Xie S., Wang D.C., Shan X.F., Cai Z.G. Prognosis and Nomogram Prediction for Patients with Oral Squamous Cell Carcinoma: A Cohort Study. Diagnostics (Basel). 2023 May 17; 13(10): 1768. doi: 10.3390/diagnostics13101768]. The authors included 432 patients with primary squamous cell carcinoma of the oral cavity who underwent surgical treatment in the study. Cox proportional regression models were used to identify independent prognostic factors for overall survival. Further construction and validation of the nomogram were performed using R 4.0.1 software and the Survival, RMS, Hmisc, and Foreign packages. The reliability of the nomogram was assessed using C-index values and calibration curves. After multivariate analysis, the following were significant predictors of mortality: body mass index over 24 (OR=0.434, 95% CI: 0.192-0.978, p=0.044), pain level (moderate pain (6-8 points) – OR=2.301, 95% CI: 1.222-4.334, p=0.01; severe pain (9-10 points) – OR=5.193, 95% CI: 2.269-11.885, p<0.0001), tumor G3 (OR=2.393, 95% CI: 1.264-4.528, p=0.007), stage N (N1 – OR=2.164, 95% CI: 1.187-3.944, p=0.012; N2b – OR=2.090, 95% CI: 1.158-3.771, p=0.014; N2c – OR=3.134, 95% CI: 1.324-7.420, p=0.009; N3b – OR=14.848, 95% CI: 1.523-144.762, p=0.02). A nomogram for predicting three- and five-year overall survival was created and presented (Fig. 2). The C-index values for the 3-year and 5-year prediction model in this study were 0.782 and 0.770. Patients were divided into low, intermediate and high risk groups. Using the Kaplan-Meier method and the log-rank test, a statistically significant difference in patient survival (p<0.001) was demonstrated depending on the risk group.

The disadvantages of this prognostic scale are:

- performing statistical analysis on patients with primary localization of oral cavity cancer only,

- lack of an online calculator.

The closest (prototype) is the created nomogram for determining three- and five-year overall survival and disease-free survival [Xu X.L., Xu J.H., He J.Q., Li Y.H., Cheng H. Novel prognostic nomograms for postoperative patients with oral cavity squamous cell carcinoma in the central region of China. BMC Cancer. 2024 Jun 14;24(1):730. doi: 10.1186/s12885-024-12465-6]. The study included 607 patients with squamous cell carcinoma of the oral cavity after surgical treatment. Univariate and multivariate Cox regression analyses were performed. Nomograms were created using the R programming language. The reliability of the nomogram was assessed using the C-index values, calibration curves and ROC analysis. Significant predictors of the risk of death were: stage III (p=0.002), stages IVA and IVB (p=0.001), the presence of perineural invasion (p<0.001), smoking after surgery (p=0.001), ECOG 2 (p=0.014), systemic inflammation index ≥1433 (p=0.004), platelet-to-lymphocyte ratio ≥185 (p=0.009), Charlson comorbidity index ≥6 (p<0.001), absence of adjuvant radiotherapy (p=0.011), absence of adjuvant chemotherapy (p=0.018). Significant independent factors for disease-free survival were stage III (p=0.017), stages IVA and IVB (p=0.006), presence of perineural invasion (p<0.001), postoperative smoking (p=0.005), ECOG 2 (p=0.006), systemic inflammatory index ≥1433 (p=0.008), platelet/lymphocyte ratio ≥185 (p=0.011), Charlson comorbidity index ≥6 (p<0.001), absence of adjuvant radiotherapy (p=0.027), absence of adjuvant chemotherapy (p=0.032). A nomogram predicting three- and five-year overall survival and disease-free survival was generated and presented (Fig. 3). The C-index values for the prediction model of overall and disease-free survival were 0.691 and 0.674 in the training group and 0.722 and 0.680 in the control group, respectively. All patients in the postoperative period were classified into three risk groups for death and disease occurrence depending on the score. For overall survival at low risk, the score was 165.31 and below, at medium risk – from 165.32 to 237.45 points, at high risk – more than 507.37 points; for disease-free survival for the same groups, the score was less than 165.31, from 157.47 to 241.66 and more than 482.00, respectively. The authors have developed web calculators:

- for overall survival: https://xxlchxjh.shinyapps.io/DynNomapp_oral_cancer_OS/;

- for disease-free survival: https://xxlchxjh.shinyapps.io/DynNomapp_oral_cancer_DFS/.

The disadvantages of this prognostic scale are:

- performing statistical analysis on patients with primary localization only.

The technical result of the proposed invention is the development of a method for predicting the progression-free and overall survival of patients with recurrent oral cancer during surgical treatment through the use of nomograms and the creation of an online calculator, which will allow for a more objective assessment of the results of the operation.

The technical result is achieved by the fact that, just as in the known method, statistically significant independent unfavorable survival factors are identified during single-factor and multi-factor analyses, which are used to create nomograms and an online calculator.

The peculiarity of the claimed method is that a morphological study of the preserved tissues is carried out and the following predictors are assessed:

- type of progression: locoregional, local or regional;

- the period from previous treatment to relapse;

- prevalence at the primary site (T);

- prevalence in regional lymph nodes (N);

- resection volume – R1 or R2;

The obtained data are entered into the online calculator “Medical prognostic model of patient survival in case of recurrent oral cancer” and the overall survival is determined.

The method is explained with a detailed description, clinical examples and illustrations, which show:

Fig. 1 – Nomograms for determining three- and five-year overall (A) and tumor-specific (B) survival (Nie Z., Zhao P., Shang Y., Sun B., 2021). Abbreviations: OS – overall survival, CSS – tumor-specific survival, KFI – Kaplan-Feinstein comorbidity index, SII – systemic inflammation index.

Fig. 2 – Nomograms for determining three- and five-year overall survival (Zhang X.Y., Xie S., Wang D.C. et al., 2023). Abbreviations: BMI – body mass index, OPMDs – potentially malignant diseases of the oral cavity.

Fig. 3 – Nomograms: a) for determining three- and five-year overall survival, b) and disease-free survival. Example of calculation for a specific patient (Xu X.L., Xu J.H., He J.Q., 2024). Abbreviations: OS – overall survival, DFS – disease-free survival, PLR – platelet-to-lymphocyte ratio, SII – systemic inflammation index, ACCI – age-adjusted Charlson Comorbidity Index.

Fig. 4 – Nomogram for calculating progression-free survival after salvage surgery for recurrent oral cancer. Abbreviations: PNI – perineural invasion, PFS – progression-free survival.

Fig. 5 – Nomogram for calculating overall survival after salvage surgery for recurrent oral cancer. Abbreviations: OS – overall survival.

Fig. 6 – Heat matrix for multicollinearity testing. Abbreviations: LVI – lymphovascular invasion, PNI – perineural invasion.

Fig. 7 – Odds ratio estimates with 95% CI for the studied predictors of progression after salvage surgery included in the model.

Fig. 8 – Validity of the final progression-free survival model using Schoenfeld residual analysis to verify the proportional hazards assumption.

Fig. 9 – Calibration curve of the 3-year progression-free survival model.

Fig. 10 – Odds ratio estimates with 95% CI for the studied predictors of mortality after salvage surgery included in the model.

Fig. 11 – Validity of the final overall survival model using Schoenfeld residual analysis to verify the proportional hazards assumption.

Fig. 12 – Calibration curve of the overall survival model for a 3-year period.

Fig. 13 – Calculation of progression-free survival in patient Ch. using the generated nomogram.

Fig. 14 – Calculation of progression-free survival for patient Ch., in the online calculator.

Fig. 15 – Calculation of overall survival for patient Ch. using the created nomogram.

Fig. 16 – Calculation of overall survival for patient Ch., in the online calculator.

Fig. 17 – Calculation of progression-free survival in patient G. using the generated nomogram.

Fig. 18 – Calculation of progression-free survival for patient G., using the online calculator.

Fig. 19 – Calculation of overall survival in patient G. using the created nomogram.

Fig. 20 – Calculation of overall survival for patient G. in the online calculator.

The method is carried out as follows.

To determine progression-free survival in a patient, the following unfavorable predictors are assessed: locoregional progression; the presence of perineural invasion; resection in the amount of R1 or R2. For overall survival, the following are assessed: locoregional progression; the time from the previous treatment to relapse is less than 12 months; the prevalence of rpT4a in the primary lesion, the prevalence of N3b in regional lymph nodes; resection in the amount of R1, 2. Then, progression-free and overall survival are determined using nomograms and an online calculator created using the program for electronic computers "Medical prognostic model of patient survival in case of recurrent oral cancer", located in electronic communication networks at the address: https://kalk-290-oral-cavity.shinyapps.io/rocsc/ - to determine the prognosis of progression-free and overall survival.

In the nomogram, the first line represents an integral line with values from 0 to 100; the following lines represent unfavorable predictors, on which a value is selected and a perpendicular line is drawn to intersect the first line, resulting in a score. Progression-free survival is assessed using three predictors:

- type of progression: 1 – locoregional, 0 – local or regional;

- perineural invasion: 1 – absent, 0 – present;

- R resection: 1 – R 1 or 2, 0 – R 0.

Overall survival is assessed by five predictors:

- type of progression: 1 – locoregional, 0 – local or regional;

- time to progression: 1 – up to 12 months, 0 – after 12 months;

- prevalence of the primary lesion rpT4a: 1 – corresponds, 0 – rpT0-3;

- prevalence in lymph nodes rpN3b: 1 – corresponds, 0 – rpN0-3a;

- R resection: 1 – R 1 or 2, 0 – R 0.

Next, the total score is determined, marked on the line labeled “total score,” and a perpendicular line is drawn down to the intersection with the lines indicating one- and two-year survival rates, and the corresponding indicators are obtained.

The calculation was made on the basis of a study including 290 patients after surgical treatment for recurrent malignant neoplasms of the oral mucosa, corresponding to the prevalence of rpT1-4aN0-3bM0; rpT0N1-3bM0.

In the studied cohort, progression was recorded in 160 cases, and death occurred in 109 cases.

The influence of nine unfavorable predictors was studied, such as: history of radiation therapy, locoregional recurrence, progression within 12 months after previous treatment, prevalence of recurrent tumor in the primary rpT4a lesion, prevalence of recurrent tumor in the rpN3b lymph nodes, presence of lymphovascular and perinerval invasion, depth of invasion, resection in the volume of R1 or R2. During the data analysis, 0.44% of missing values were identified, covariates with a Pearson correlation coefficient >0.75 and with zero variance were not identified (Fig. 6).

The development of predictive models was performed using statistical analysis using the R programming language (version 4.3.1; R Foundation for Statistical Computing, Vienna, Austria) and the R studio interactive development environment.

The study used a multivariate Cox proportional hazards regression model with an iterative algorithm for backward stepwise selection of predictors based on assessing the importance of each variable using residual chi-square statistics and minimizing the Akaike information criterion (AIC).

The concordance index (C-index) was used to assess the accuracy and effectiveness of the predictive model. Model performance was assessed through rigorous internal bootstrap validation (400 iterations). Visual representation of the predictive models was achieved by constructing nomograms and developing an interactive online calculator using the Shiny software package in the R statistical programming environment.

Initially, univariate and multivariate analyses were performed for progression-free and overall survival.

In univariate analysis, statistically significant predictors of progression-free survival were: the prevalence of recurrent process in the primary rpT4a lesion (OR=2.00; 95% CI: 1.46-2.75, p<0.001), the prevalence of recurrent process in the N3b lymphatic pathways (OR=1.73; 95% CI: 1.21-2.45, p=0.002), locoregional progression (OR=2.34; 95% CI: 1.69-3.24, p<0.001), the depth of tumor invasion (OR=1.01; 95% CI: 1.00-1.03, p=0.0004), lymphovascular invasion (OR=2.22; 95% CI: 1.58-3.11, p<0.001), perineural invasion (OR=2.14; 95% CI: 1.56-2.93, p<0.001), positive resection margin (R1, R2) (OR=2.68, 95% CI: 1.90-3.78, p<0.001). In multivariate analysis, the following factors were statistically significant: locoregional progression (OR=1.61; 95% CI: 1.07-2.42, p=0.022), lymphovascular invasion (OR=1.50; 95% CI: 1.04-2.18, p=0.031) and perineural invasion (OR=1.45; 95% CI: 1.02-2.05, p=0.037), positive resection margin (R1, R2) (OR=2.25, 95% CI: 1.55-3.28, p<0.001).

The formula for the final model of progression risk:

h _i (t) = h ₀ (t) × exp ^{(0.781 × X1 + 0.492 × X2 + 0.904 × X3)} ,

Where:

h _i (t) is the predicted risk of progression in the i-th patient over a certain period of time,

h ₀ (t) – baseline risk of progression over a certain time period t (intercept),

X1 – type of progression (1 – locoregional, 0 – local or regional),

X2 – perineural invasion (1 – absent, 0 – detected),

X3 – resection radicality (1 – R 1 or 2, 0 – R0).

The C-index of the model was 0.7.

Figure 7 shows the estimated hazard ratios with 95% confidence intervals for the predictors of progression after salvage surgery included in the model.

When testing the proportionality of hazards in the Cox model using Schoenfeld residual analysis, no statistically significant evidence of violation of the proportional hazards assumption was found for all variables in the model (individual test p>0.05), or for the model as a whole (global test p>0.05) (Fig. 8).

The adjusted C-index of the model as a result of rigorous internal validation was 0.692. When creating a calibration curve for the progression-free survival model over a three-year period, the average absolute calibration error was 0.04 (the average prediction error between the predicted and true probabilities), that is, on average, the model predictions deviate from the actual values by 4%) (Fig. 9).

Statistically significant factors for overall survival in univariate analysis were the following: progression within 12 months after previous treatment (OR=2.14; 95% CI: 1.29-3.55, p=0.003), prevalence of recurrent process in the primary lesion rpT4a (OR=2.34; 95% CI: 1.58-3.47, p<0.001), prevalence of recurrent process in the lymphatic pathways N3b (OR=2.31; 95% CI: 1.55-3.44, p<0.001), locoregional progression (OR=2.93; 2.01-4.28, p<0.001), depth of tumor invasion (OR=1.02; 95% CI: 1.01-1.03, p<0.001), LVI (OR=2.09; 95% CI: 1.40-3.13, p<0.001) and PNI (OR=2.39; 95% CI: 1.62-3.52, p<0.001), positive resection margin (R1, R2) (OR=2.53, 95% CI: 1.68-3.82, p<0.001). In multivariate analysis, the following factors were statistically significant: progression within 12 months after previous treatment (OR=1.77; 95% CI: 1.03-3.05, p=0.040), locoregional progression (OR=1.67; 95% CI: 1.03-2.70, p=0.036), the prevalence of recurrent process along the N3b lymphatic pathways (OR=1.66; 95% CI: 1.03-2.68, p=0.037), positive resection margin (R1, R2) (OR=2.50, 95% CI: 1.60-3.91, p<0.001).

The formula for the final model of the risk of death:

h(t) = h ₀ (t) × exp ^{(0.618 × X1 + 0.685 × X2 + 0.617 × X3 + 0.584 × X4 + 0.982 × X5)} ,

Where:

h _i (t) is the predicted risk of death in the i-th patient over a certain period of time,

h ₀ (t) – baseline risk of death for a certain time period t (intercept),

X1 – type of progression (1 – locoregional, 0 – local or regional),

X2 – time to relapse (1 – less than 12 months, 0 – more than 12 months),

X3 – prevalence of relapse in the primary rpT focus (1 – rpT4a, 0 – rpT0-3),

X4 – prevalence along regional lymphatic pathways rpN (1 – rpN3b, 0 – rpN0-3a),

X5 – radicality of resection (1 – R1 or 2, 0 – R0).

The C-index of the model was 0.72.

Figure 10 shows the estimated odds ratios with 95% confidence intervals for the predictors of mortality after salvage surgery included in the model.

When testing the proportionality of hazards in the Cox model using Schoenfeld residual analysis, no statistically significant evidence of violation of the proportional hazards assumption was found for all variables in the model (individual test – p>0.05), as well as for the model as a whole (global test – p>0.05) (Fig. 11).

The adjusted C-index of the model as a result of rigorous internal validation was 0.709.

When creating a calibration curve for the overall survival model over a three-year period, the mean absolute calibration error was 0.033 (the mean prediction error between the predicted and true probabilities), meaning that, on average, the model predictions deviate from the actual values by 3.3% (Fig. 12).

The invention is illustrated by clinical examples.

Clinical example 1

Patient Ch., 34 years old, with a clinical diagnosis: tongue cancer, stage II, type T2N0M0. ChRT, 70 Gy, 08-10/2016. Continued tumor growth, type T2N0M0.

From August to October 2016, chemoradiation therapy was administered at a total dose of 70 Gy. A follow-up examination in December 2016 revealed continued tumor growth.

On January 18, 2017, the patient underwent surgery in the following volume: left hemiglossectomy, selective supraomohyoid lymphadenectomy on the left side of the neck with removal of groups I-III lymph nodes, microsurgical plastic surgery with a radial autograft.

According to the routine morphological examination: the left half of the tongue measures 6.5 x 3 x 2.5 cm, on the lower surface of the tongue there is an ulcerated area with undermined edges measuring 2.2 x 1 cm. On a cut, in the projection of the ulcerated area there is a dense tumor node measuring 1.5 x 2.3 x 1.3 cm. Microscopic examination: in the tongue tissue there is an infiltrative growth of squamous cell carcinoma with ulceration, keratinization, foci of necrosis, lymphocytic infiltration. Perineural and lymphovascular invasion are not determined. Therapeutic pathomorphism is grade 1. The tumor infiltrates the native muscles of the tongue with an invasion depth of up to 1 cm. There is no tumor growth at the resection margins. In 30 examined lymph nodes, there are no cancer metastases, sinus histiocytosis.

This patient lacks all the unfavorable predictors of progression-free survival: progression only in the primary lesion, no perineural invasion, and R0 resection.

The nomogram calculations are as follows. The integrated scale for all predictors is 0 points. Therefore, the overall score is 0 points, which we mark on the "overall score" scale, draw a perpendicular line down to the intersection with the one- and two-year progression-free survival lines, and obtain the corresponding figures (70% and 64%, respectively) (Figure 13).

We calculated progression-free survival using an online calculator by entering the relevant parameters into the online version at the link (Fig. 14). One-, two-, three-, and five-year progression-free survival rates were 70.1%, 64.3%, 60.1%, and 53.2%.

The patient has one unfavorable factor for overall survival - progression within 12 months, the rest are absent: progression only in the primary lesion, rpT2, rpN0, resection in the volume of R0.

Calculations according to the nomogram look like this.

For the "time to relapse less than 12 months" factor, select 1 on the corresponding line and draw a perpendicular line to intersect with the first integral line, obtaining a score of 70. For all other factors, select 0 points. The overall score is 70. Mark this on the "overall score" line and draw a perpendicular line downward to intersect with the one- and two-year survival lines, obtaining the corresponding figures (89% and 80%, respectively) (Fig. 15).

We calculated overall survival using the online calculator by entering the relevant parameters into the online version via the link (Fig. 16). One-, two-, three-, and five-year overall survival rates were 89.1%, 79.6%, 75.6%, and 68.7%. Currently, the progression-free and overall survival rates for this patient after surgical treatment are 96 months.

Clinical example 2

Patient G., 51 years old, with a clinical diagnosis: cancer of the mucous membrane of the floor of the mouth, type T2N0M0, stage II. RT, 70 Gy, 03-05/2014. Continued tumor growth, metastases to the lymph nodes of the neck on the left, type T4aN3bM0.

At the P.A. Herzen Moscow Oncology Research Institute, 70 Gy of RT was performed in March–May 2014. A follow-up examination in September 2014 revealed continued tumor growth and metastases to the left lymph nodes of the neck.

On October 8, 2014, an operation was performed in the following volume: resection of tissues of the floor of the mouth, tongue, lower jaw, radical modified nerve-sparing lymphadenectomy on the left side of the neck with removal of groups I-V of lymph nodes, selective lymphadenectomy on the right side of the neck with removal of groups I-III of lymph nodes, microsurgical plastic surgery of the oral cavity with a cutaneous-fascial radial autograft.

According to the planned morphological examination: fragments of the alveolar process of the mandible with overall dimensions of 7 × 5.5 × 3 cm with 11 teeth, soft tissues of the floor of the mouth, and a fragment of the lower third of the tongue are determined in a single block with overall dimensions of 7 × 5.5 × 3 cm. An ulcerated tumor measuring 3 × 2 × 1.5 cm is located in the area of the floor of the mouth with the transition to the tongue. Microscopic examination reveals moderately differentiated squamous cell carcinoma with weak signs of therapeutic pathomorphism, foci of perineural growth and lymphovascular invasion. After decalcification, invasion into bone tissue and a positive resection margin are determined. Metastases from squamous cell carcinoma are found in five lymph nodes, two of which have extracapsular spread.

The nomogram calculations are as follows. The patient has all the unfavorable factors for progression-free survival. The integrated score is as follows: 86 points for locoregional progression, 54 points for perineural invasion, and 100 points for R1 resection. The total score is 240. This is marked on the "overall score" scale. A perpendicular line is drawn from this mark down to the intersection with the one- and two-year progression-free survival lines, yielding the corresponding figures (less than 5%) (Fig. 17).

We calculated progression-free survival using an online calculator by entering the relevant parameters into the online version at the link (Fig. 18). One-, two-, three-, and five-year progression-free survival rates were 4.4%, 2.0%, 1.1%, and 0.4%.

This patient has all the unfavorable factors for overall survival. Using the integrated scale, we obtain the following score: 63 points for locoregional progression, 70 points for the 12-month progression period, 63 points for rpT4a primary lesion spread, 60 points for rpN3b, and 100 points for R1 resection. The total score is 356. We plot this value on the "overall score" scale. We draw a perpendicular line from it down to the intersection with the one- and two-year overall survival lines, obtaining the corresponding figures (15% and less than 2.5%, respectively) (Fig. 19).

We calculated overall survival using the online calculator by entering the relevant parameters into the online version via the link. One-, two-, three-, and five-year overall survival rates were 15.4%, 2.3%, 1.0%, and 0.2% (Fig. 20). The patient's progression-free survival was 2 months, and the overall survival was 10 months.

The proposed method in clinical practice, through the use of nomograms and an online calculator, allows for the assessment and prediction of progression-free and overall survival in patients with recurrent oral cancer undergoing surgical treatment, improves patient selection for it, and helps in making decisions in the case of an unfavorable prognosis about the inappropriateness of surgery or its combination with other methods in order to achieve better results.

Claims (7)

A method for predicting overall survival in surgical treatment of patients with recurrent non-generalized resectable oral cancer, including an assessment of unfavorable prognostic factors, characterized in that a morphological study of the resected tissues is performed and the following predictors are assessed: - type of progression: locoregional, local or regional; - the period from previous treatment to relapse; - prevalence at the primary site (T); - prevalence in regional lymph nodes (N); - resection volume - R1 or R2; The obtained data are entered into the online calculator “Medical prognostic model of patient survival in case of recurrent oral cancer” and the overall survival is determined.