Introduction
In the United States in 2023, lung cancer was the second most commonly diagnosed cancer and the leading cause of cancer death.1 Non-small cell lung cancer (NSCLC) accounts for approximately 85% of lung cancers and includes adenocarcinoma, lung squamous cell carcinoma (LSCC), and large-cell carcinoma.2 These subtypes differ in histologic and molecular characteristics. For example, adenocarcinoma is often associated with mutations in the epidermal growth factor receptor and anaplastic lymphoma kinase (CD246) genes, while LSCC more often develops in smokers and is not typically associated with these mutations.2 Therefore, the study of the pathogenesis and diagnosis of these subtypes is often conducted separately.2–4 Stage III LSCC is heterogeneous and is divided into IIIA (T1N2M0, T2N2M0, T3N1M0, T4N0M0, T4N1M0), IIIB (T1N3M0, T2N3M0, T3N2M0, and T4N2M0), and IIIC (T3N3M0 and T4N3M0).5
Patients with lymph node lesions in the contralateral lung (N3) are not candidates for surgical treatment (IIIC and partially IIIB stages). The remaining stage III LSCC patients undergo tumor removal with R0 surgery, which ensures complete tumor excision along with lymph node dissection. Patients with tumor characteristics T1N2M0, T2N2M0, T3N1M0, and T3N2M0 undergo adjuvant (postoperative) polychemotherapy to suppress remaining cancer cells. Those with T4N0M0, T4N1M0, and T4N2M0 receive neoadjuvant polychemotherapy before surgery.6 Despite these efforts, the risk of relapse due to micrometastases after surgery remains high. Tumor recurrence is diagnosed in 20% of stage I LSCC patients during the first year after treatment,3 and in 35–40% of stage III patients.4
All patients undergo medical examinations every three months for two years post-treatment. If the patient is asymptomatic, computed tomography (CT) is performed at six and 12 months after treatment.6 However, relapses may occur earlier and initially be asymptomatic.
Circulating metabolites and serum metabolic profiles may serve as informative tools for the diagnosis and prognostic stratification of patients with NSCLC. The use of blood concentrations of cytokeratin fragment 19 (CYFRA 21-1), carcinoembryonic antigen (CEA), squamous cell carcinoma antigen, neuron-specific enolase, tissue polypeptide antigen, and others is being explored for this purpose.7–15 Among these markers, CYFRA 21-1 and CEA have been examined in relation to overall and relapse-free survival, particularly in early-stage (I–II) cohorts.9,10,13 However, these markers have low sensitivity and specificity, and studies have only examined stage III LSCC patients in mixed groups with early stages.14,15
Based on previous biomarker studies, we selected serum CYFRA 21-1, the percentage of lymphocytes expressing CXCR1, and the percentage of monocytes expressing CXCR2 for exploratory longitudinal assessment in patients with stage III LSCC.
Recently, we demonstrated the role of biomarkers in predicting relapse in stage I-II LSCC after surgical treatment. Differences in SCC antigen concentrations, the percentages of lymphocytes with CXCR2, and the percentages of monocytes with CD44v6 across postoperative intervals were reported to classify recurrence status with accuracies ranging from 68.4% to 89.5%. Subsequent regression analysis and the development of a combined model that included these parameters increased the apparent classification accuracy to 96.5% (specificity – 95.6%, sensitivity – 100%). These results indicated the usefulness of the combined model in stage I-II LSCC patients as a potential adjunct for postoperative monitoring.16,17
Thus, surgical treatment of stage III LSCC is associated with chemotherapy. Despite these efforts, some patients develop tumor relapses, mostly within the first year after surgery, which are associated with high mortality. Earlier identification of relapse may facilitate timely clinical evaluation; therefore, reliable criteria for postoperative monitoring remain important.8–10,12–15 Although biochemical criteria have not yet found practical application due to their nonspecificity and insufficient sensitivity, numerous studies continue to explore new biomarkers.
The purpose of the present investigation was to evaluate longitudinal changes in CYFRA 21-1, the percentage of lymphocytes expressing CXCR1, and the percentage of monocytes expressing CXCR2 in relation to one-year recurrence status after surgery for stage III LSCC.
Materials and methods
Research object
The study involved 48 patients (29 men and 19 women) admitted to the clinic of the N.N. Alexandrov National Cancer Center of Belarus between 2021 and 2022. The inclusion criteria for patients were as follows: newly diagnosed stage IIIA or IIIB LSCC. The exclusion criteria were the presence of metachronous or secondary cancer, and patient refusal to participate in the study. During the first year of observation, there were no patient dropouts. Patients with T1N2M0, T2N2M0, T3N1M0, or T3N2M0 underwent surgical resection of the tumor (surgical volume – R0) followed by 4 courses of adjuvant polychemotherapy consisting of a combination of vinorelbine (V) 25–30 mg/m2 and cisplatin (C) 80 mg/m2. Patients with T4N0M0, T4N1M0, or T4N2M0 received two courses of neoadjuvant chemotherapy consisting of V + C, followed by surgical resection of the tumor and two additional courses of adjuvant polychemotherapy with V + C. The post-treatment monitoring algorithm for all patients included a physical examination every three months for the first year after surgery. In the absence of complaints and symptoms of the disease based on the results of a physical examination, a CT scan was performed at six and 12 months after surgery. This was the routine surveillance schedule used in the cohort. Information on the development of relapse in the examined patients after surgery was obtained based on CT data from the Cancer Register of the Republic of Belarus (N.N. Alexandrov National Cancer Center of Belarus).
Ethical approval and consent
The study was approved by the Biomedical Ethics Committee of Belarusian State Medical University (Protocol No. 2, April 10, 2021). All patients provided written voluntary consent to participate in the study in accordance with the Declaration of Helsinki.
Study design and sample collection
Biomarker measurements were scheduled before treatment and at three weeks, three months, and six months after surgery. Four recurrences occurred before the six-month time point; consequently, the nominal six-month measurements for those patients were post-recurrence observations. Blood was collected from the cubital vein of patients on an empty stomach into a vacutainer with EDTA-K2 as an anticoagulant. To obtain serum, blood was collected into a tube with thrombin and separating gel.
Analysis of samples
The concentration of CYFRA 21-1 antigen in blood serum was determined using the electrochemiluminescent method on a Cobas e411 automated analyzer (Roche Diagnostics GmbH, Germany) using original Elecsys CYFRA 21-1 kits (Roche Diagnostics GmbH, Germany). To assess the concentrations of CXCR1 and CXCR2 receptors in leukocyte cells, a Navios flow cytometer (Beckman Coulter, USA) was used. For this, 100 µL of blood and a solution containing a mixture of antibodies with fluorescent labels were placed in a test tube: CD181 (CXCR1)-PE-Cy5 (BioLegend, USA), CD182 (CXCR2)-PE (BioLegend, USA), and CD45-Pacific Orange (Exbio, Czech Republic). After 15 min of incubation in the dark with antibodies containing a fluorescent label, 1 mL of VersaLyse lysis solution (Beckman Coulter, France) was added to the mixture. Antibodies were fixed on the cell surface using IQTest 3 fixing solution (Beckman Coulter, France).
Statistical analysis
The three biomarkers were evaluated individually. Combined-model results were excluded because the available equation did not reproduce the tabulated patient-level scores.
Recurrence-free survival was measured from surgery to computed tomography-confirmed recurrence or censoring at one year. Associations between each biomarker or its change and recurrence-free survival were explored using univariable and multivariable Cox proportional hazards models; the multivariable models included the three biomarkers simultaneously. Analyses of biomarker changes were restricted to the three-week-to-three-month interval, which ended before the earliest recorded recurrence. Six-month measurements were summarized descriptively only.
Receiver operating characteristic (ROC) analysis was restricted to biomarker changes from three weeks to three months after surgery. Sensitivity, specificity, positive and negative predictive values, and overall accuracy were calculated from true-positive, true-negative, false-positive, and false-negative classifications using generally accepted formulas.18 The threshold value was selected to optimize sensitivity and specificity on the ROC curve. Because thresholds and performance were derived and evaluated in the same 48-patient cohort, all estimates are apparent and exploratory. Baseline characteristics and serial biomarker summaries were presented descriptively without significance testing; P < 0.05 was used for the Cox models.
Results
Baseline characteristics are summarized descriptively in Table 1. No inferential comparison was retained because the small groups and the available report did not specify the statistical tests used.
Table 1Baseline characteristics of patients with and without recurrence
| Characteristics | Recurrence group (n = 17) | No-recurrence group (n = 31) |
|---|
| Age (mean ± SD), years | 55.3±19.3 | 58.1 ± 18.5 |
| Gender, n (%) | |
| Males | 10 (58.8) | 19 (61.3) |
| Females | 7 (41.2) | 12 (38.7) |
| Male smoking status, n (%) | | |
| former | 2 (20.0) | 3 (15.8) |
| current | 8 (80.0) | 15 (78.9) |
| never | 0 (0.0) | 1 (5.2) |
| Female smoking status, n (%) | | |
| former | 1 (14.3) | 2 (16.7) |
| current | 5 (71.4) | 8 (66.6) |
| never | 1 (14.3) | 2 (16.7) |
Recurrence occurred in 17 of 48 patients within one year after surgery: one patient at 3.8 months, three additional patients by six months, and 13 patients between six months and one year (Table 2).
Table 2Levels of CYFRA 21-1 and percentages of lymphocytes with CXCR1, and percentages of monocytes with CXCR2 in patients with recurrent LSCC before treatment and after surgery
| Patients | Index | Before treatment | After surgery
| Time to recurrence, months |
|---|
| Three weeks | Three months | Six months |
|---|
| 1 | CYFRA 21-1, g/l, ×10−6 | 6.64 | 5.29 | 6.41 | 8.07 | 3.8 |
| CXCR1, %, lymphocytes | 3.83 | 2.27 | 2.72 | 3.42 | |
| CXCR2, %, monocytes | 2.19 | 1.2 | 1.46 | 1.91 | |
| 2 | CYFRA 21-1, g/l, ×10−6 | 7.94 | 3.59 | 4.75 | 6.23 | 4.3 |
| CXCR1, %, lymphocytes | 5.94 | 2.37 | 2.77 | 3.31 | |
| CXCR2, %, monocytes | 1.83 | 1.01 | 1.24 | 1.71 | |
| 3 | CYFRA 21-1, g/l, ×10−6 | 5.99 | 3.14 | 4.30 | 5.82 | 5.1 |
| CXCR1, %, lymphocytes | 3.73 | 1.86 | 2.42 | 3.07 | |
| CXCR2, %, monocytes | 1.7 | 1.41 | 1.63 | 2.12 | |
| 4 | CYFRA 21-1, g/l, ×10−6 | 7.32 | 3.64 | 4.72 | 6.16 | 5.4 |
| CXCR1, %, lymphocytes | 4.78 | 2.27 | 2.70 | 3.25 | |
| CXCR2, %, monocytes | 1.62 | 1.29 | 1.52 | 1.98 | |
| 5 | CYFRA 21-1, g/l, ×10−6 | 6.56 | 4.21 | 5.23 | 6.67 | 6.1 |
| CXCR1, %, lymphocytes | 4.38 | 2.17 | 2.53 | 3.12 | |
| CXCR2, %, monocytes | 1.42 | 1.25 | 1.48 | 1.93 | |
| 6 | CYFRA 21-1, g/l, ×10−6 | 7.27 | 3.47 | 4.65 | 6.19 | 6.3 |
| CXCR1, %, lymphocytes | 4.28 | 2.17 | 2.81 | 4.14 | |
| CXCR2, %, monocytes | 2.19 | 0.7 | 0.97 | 1.42 | |
| 7 | CYFRA 21-1, g/l, ×10−6 | 7.20 | 3.62 | 4.60 | 5.95 | 7.2 |
| CXCR1, %, lymphocytes | 3.63 | 1.76 | 2.11 | 2.67 | |
| CXCR2, %, monocytes | 2.16 | 1.49 | 1.73 | 2.18 | |
| 8 | CYFRA 21-1, g/l, ×10−6 | 6.42 | 3.95 | 4.66 | 6.09 | 7.9 |
| CXCR1, %, lymphocytes | 4.18 | 2.27 | 2.52 | 3.07 | |
| CXCR2, %, monocytes | 1.36 | 0.75 | 1.01 | 1.47 | |
| 9 | CYFRA 21-1, g/l, ×10−6 | 6.21 | 3.96 | 4.64 | 5.92 | 8.4 |
| CXCR1, %, lymphocytes | 4.13 | 2.06 | 2.40 | 2.98 | |
| CXCR2, %, monocytes | 2.32 | 1.44 | 1.65 | 2.15 | |
| 10 | CYFRA 21-1, g/l, ×10−6 | 5.85 | 3.29 | 3.61 | 4.96 | 8.9 |
| CXCR1, %, lymphocytes | 3.68 | 2.37 | 2.52 | 3.26 | |
| CXCR2, %, monocytes | 2.12 | 1.06 | 1.29 | 1.77 | |
| 11 | CYFRA 21-1, g/l, ×10−6 | 6.03 | 3.29 | 4.24 | 5.80 | 9.3 |
| CXCR1, %, lymphocytes | 3.83 | 1.81 | 2.14 | 2.73 | |
| CXCR2, %, monocytes | 1.5 | 1.1 | 1.33 | 1.81 | |
| 12 | CYFRA 21-1, g/l, ×10−6 | 6.29 | 3.92 | 4.37 | 6.30 | 10.1 |
| CXCR1, %, lymphocytes | 3.32 | 1.66 | 1.84 | 2.67 | |
| CXCR2, %, monocytes | 2.54 | 1.2 | 1.48 | 1.91 | |
| 13 | CYFRA 21-1, g/l, ×10−6 | 7.84 | 3.24 | 3.54 | 4.86 | 10.6 |
| CXCR1, %, lymphocytes | 3.98 | 1.66 | 1.78 | 2.44 | |
| CXCR2, %, monocytes | 2.33 | 1.37 | 1.61 | 2.09 | |
| 14 | CYFRA 21-1, g/l, ×10−6 | 4.14 | 3.09 | 4.06 | 5.67 | 10.8 |
| CXCR1, %, lymphocytes | 2.47 | 1.61 | 1.93 | 2.69 | |
| CXCR2, %, monocytes | 1.69 | 1.16 | 1.42 | 1.85 | |
| 15 | CYFRA 21-1, g/l, ×10−6 | 4.57 | 3.57 | 4.43 | 5.93 | 11.0 |
| CXCR1, %, lymphocytes | 3.70 | 1.80 | 2.15 | 2.70 | |
| CXCR2, %, monocytes | 1.99 | 1.02 | 1.28 | 1.73 | |
| 16 | CYFRA 21-1, g/l, ×10−6 | 6.58 | 3.29 | 4.15 | 5.64 | 11.2 |
| CXCR1, %, lymphocytes | 4.25 | 1.85 | 2.20 | 2.75 | |
| CXCR2, %, monocytes | 1.96 | 1.32 | 1.55 | 2.01 | |
| 17 | CYFRA 21-1, g/l, ×10−6 | 5.12 | 3.68 | 4.54 | 6.03 | 11.7 |
| CXCR1, %, lymphocytes | 3.55 | 1.95 | 2.30 | 2.85 | |
| CXCR2, %, monocytes | 1.92 | 1.13 | 1.39 | 1.84 | |
Patients 1–4 experienced recurrence at 3.8, 4.3, 5.1, and 5.4 months, respectively; their nominal six-month biomarker measurements were therefore obtained after recurrence.
Three weeks after surgery, all three biomarkers decreased sharply (Table 2), approaching the threshold values shown in Table 3. Among the 17 patients who later experienced recurrence, CYFRA 21-1 remained above its threshold in 11, CXCR1-positive lymphocytes in nine, and CXCR2-positive monocytes in 11.
Table 3Threshold values and exploratory associations of three-week biomarker levels with recurrence-free survival in stage III LSCC (Cox proportional hazards models)
| Index | TV | Univariate model
| Multivariate model
|
|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value |
|---|
| CYFRA 21-1, g/l, ×10−6 | 3.30 | 1.069 (1.005–1.133) | 0.035 | 1.062 (1.003–1.121) | 0.039 |
| CXCR1, %, lymphocytes | 1.90 | 1.059 (1.008–1.110) | 0.037 | 1.055 (1.005–1.105) | 0.041 |
| CXCR2, %, monocytes | 1.10 | 1.032 (1.009–1.055) | 0.038 | 1.027 (1.006–1.048) | 0.043 |
At three weeks after surgery, each biomarker was associated with recurrence-free survival in exploratory Cox proportional hazards models (Table 3); the reported confidence intervals did not include 1 and all P-values were below 0.05.
In patients without recurrence during one year, median values at three and six months were similar to those at three weeks after surgery (Table 4). In the group classified as having recurrence within one year, the markers increased across the nominal three-week, three-month, and six-month measurements (Tables 4 and 5). For four patients, however, the six-month measurement occurred after recurrence; comparisons involving six months therefore describe longitudinal differences by eventual recurrence status rather than pre-recurrence trajectories.
Table 4Serial biomarker levels by one-year recurrence status in patients with stage III LSCC
| Index | Recurrence | Before treatment | After surgery
|
|---|
| Three weeks | Three months | Six months |
|---|
| CYFRA 21-1, g/l, ×10−6 | No | 4.19 [3.51;5.13] | 2.96 [0.90; 3.79] | 2.97 [1.01; 3.81] | 2.99 [1.03; 3.86] |
| Yes | 6.42 [5.99;7.20] | 3.59 [3.29; 3.92] | 4.54 [4.24; 4.66] | 5.95 [5.80; 6.19] |
| CXCR1, %, lymphocytes | No | 2.45 [1.75;3.20] | 1.70 [0.95;2.00] | 1.75 [1.00; 2.10] | 1.80 [1.05;2.20] |
| Yes | 3.83 [3.68;4.25] | 1.95 [1.80; 2.27] | 2.40 [2.14; 2.53] | 2.98 [2.70; 3.25] |
| CXCR2, %, monocytes | No | 1.50 [0.55;2.45] | 0.90 [0.30;1.40] | 0.95 [0.35;1.65] | 1.00 [0.50;1.95] |
| Yes | 1.92 [1.62;2.16] | 1.20 [1.06; 1.32] | 1.46 [1.29; 1.55] | 1.91 [1.77; 2.01] |
Table 5Descriptive changes in biomarker levels across measurement intervals in patients with stage III LSCC
| Index | Recurrence | Before treatment to three weeks | After surgery
|
|---|
| Three weeks – three months | Three months – six months | Three weeks – six months |
|---|
| CYFRA 21-1, g/l, ×10−6 | No | 1.27 [0.91; 2.03] | 0.02 | 0.03 [0.02; 1.19] | 0.04 [0.01; 1.29] |
| Yes | 2.56 [2.25; 3.58] | 0.95 [0.71; 1.08] | 1.49 [1.43; 1.54] | 2.38 [2.33; 2.58] |
| CXCR1, %, lymphocytes | No | 1.25 [1.00; 1.85] | 0.03 [0.01;0.35] | 0.03 [0.01; 0.45] | 0.05 [0.01;0.70] |
| Yes | 1.91 [1.66; 2.21] | 0.35 [0.32; 0.40] | 0.59 [0.55; 0.70] | 0.92 [0.90; 1.01] |
| CXCR2, %, monocytes | No | 0.55 [0.35; 0.75] | 0.02 [0.01;0.19] | 0.03 [0.01; 0.37] | 0.05 [0.01;0.60] |
| Yes | 0.79 [0.53; 0.97] | 0.24 [0.23; 0.26] | 0.46 [0.45; 0.48] | 0.71 [0.69; 0.71] |
Changes in each biomarker from three weeks to three months after surgery were associated with recurrence-free survival in exploratory Cox analyses (Table 6). Because four recurrence events preceded six-month sampling, intervals ending at six months were excluded from the Cox analysis.
Table 6Exploratory associations between recurrence-free survival and biomarker changes from three weeks to three months after surgery in stage III LSCC (Cox proportional hazards models)
| Index | Time interval | Univariate model
| Multivariate model
|
|---|
| HR (95% CI) | P-value | HR (95% CI) | P-value |
|---|
| CYFRA 21-1, g/l, ×10−6 | Three weeks to three months | 1.123 (1.013–1.233) | 0.041 | 1.114 (1.008–1.220) | 0.038 |
| CXCR1, %, lymphocytes | Three weeks to three months | 1.023 (1.003–1.043) | 0.033 | 1.023 (1.001–1.045) | 0.031 |
| CXCR2, %, monocytes | Three weeks to three months | 1.036 (1.004–1.068) | 0.031 | 1.031 (1.002–1.060) | 0.028 |
ROC analysis explored how changes in the individual biomarkers from three weeks to three months after surgery classified one-year recurrence status in this cohort (Table 7). This window ended before the earliest recorded recurrence. Intervals ending at six months were excluded from predictive-performance analyses because four recurrences occurred before six-month sampling.
Table 7Apparent within-cohort performance of biomarker changes from three weeks to three months after surgery for classifying one-year LSCC recurrence status
| Index | TV | SE | SP | PPV | NPV | AUC | ACC |
|---|
| From three weeks to three months |
| CYFRA 21-1, g/l, ×10−6 | 0.95 | 76.5 | 74.2 | 61.9 | 85.2 | 0.726 | 75.0 |
| CXCR1, %, lymphocytes | 0.35 | 64.7 | 77.4 | 61.1 | 80.0 | 0.698 | 72.9 |
| CXCR2, %, monocytes | 0.24 | 70.6 | 71.0 | 57.1 | 81.5 | 0.681 | 70.8 |
For the three-week-to-three-month window, which ended before the earliest recorded recurrence, apparent within-cohort accuracy ranged from 70.8% for CXCR2, %, monocytes to 75.0% for CYFRA 21-1.
The nominal six-month measurements were retained in Tables 2, 4, and 5 solely to describe the observed longitudinal course. They were not used to estimate prospective predictive performance.
Discussion
Our study focused on three blood biomarkers: serum CYFRA 21-1, the percentage of lymphocytes expressing CXCR1, and the percentage of monocytes expressing CXCR2. The analysis reported here treats these biomarkers individually because the available combined-model equation did not reproduce the tabulated patient-level scores.
In our current investigation, while tracking the dynamics of changes in the determined parameters within one year after surgery, several trends attracted attention. The first is that within three weeks after surgical resection of the tumor, the values of all indicators in all patients decreased to values comparable to the TV. However, in some patients, the amplitude of the decrease did not reach the TV, remaining above this value. The majority of those patients (76.5%) subsequently developed tumor relapse during the year of observation.
Other researchers have also observed a decrease in blood concentrations of CYFRA 21-1 and CEA, although to different degrees. In some patients, it decreased compared to the levels before treatment but did not reach the threshold values.8–10 In these studies, patients had early stages of NSCLC and underwent only surgical treatment, so the phenomenon of a sharp decrease in the levels of these indicators after surgery was due to the resection of tumor tissue. In our study, patients received neoadjuvant and adjuvant therapy in addition to surgical treatment, which also aimed at destroying tumor cells and reducing their metabolites.
The next observed pattern was an increase in measured biomarkers after the initial decrease. Most patients showing this pattern (89.5%) belonged to the one-year recurrence group; for four patients, recurrence preceded the nominal six-month measurement. Other researchers have also noted that serum concentrations of CYFRA 21-1 and CEA in patients with NSCLC and resected tumors often increase further after a decrease. Moreover, such dynamics are associated with the development of relapse.13–15 Based on these data, the researchers concluded that monitoring serum CEA concentrations in patients with NSCLC predicts relapse with a sensitivity of 74.7% and specificity of 69.8%.13 Comparable results were obtained for CYFRA 21-1, where the sensitivity and specificity of measuring this marker for response to treatment were 79.1% and 60.6%, respectively.14 Moreover, according to another study, of five serum tumor markers (CYFRA 21-1, CEA, neuron-specific enolase, and carbohydrate antigen 125 and 19–9), only CYFRA 21-1 was the most sensitive for predicting response to chemotherapy, and an increase in its level after an initial decrease correlates with a high likelihood of tumor relapse.15 Prior studies have evaluated CYFRA 21-1 for recurrence monitoring after lung cancer resection and for treatment-response assessment in NSCLC.19,20 Although these receptors have not previously been evaluated in circulating blood-cell populations for LSCC monitoring, CXCR2-related signaling has been implicated in the lung cancer tumor microenvironment.21–23
The three-week-to-three-month interval ended before all recorded recurrences and was therefore retained for exploratory Cox and ROC analyses. Measurements at six months described later longitudinal patterns, but four of these values were obtained after recurrence and were therefore excluded from prognostic modeling.
For changes from three weeks to three months, apparent within-cohort accuracy ranged from 70.8% to 75.0% across the three individual biomarkers. These estimates were derived and evaluated in the same small cohort and were reported without confidence intervals; they should therefore be regarded as hypothesis-generating rather than clinically validated performance.
This study has several limitations. The sample size was small (48 patients, including 17 with recurrence), recruitment was limited to one center, and follow-up was limited to one year. Thresholds and performance were derived and evaluated in the same cohort without confidence intervals or independent validation, increasing the risk of optimism. Four patients experienced recurrence before the nominal six-month biomarker measurement; those values were retained only for descriptive longitudinal summaries and cannot characterize pre-recurrence trajectories for all participants. The available combined-model equation did not reproduce the tabulated patient-level scores, so combined-model findings were excluded. Future studies require prespecified landmark populations, complete reporting of statistical procedures, and external validation in larger multicenter cohorts.
Conclusions
This exploratory study observed longitudinal changes after surgery in CYFRA 21-1, the percentage of lymphocytes expressing CXCR1, and the percentage of monocytes expressing CXCR2. The biomarkers decreased soon after surgery and subsequently increased in the group that experienced recurrence within one year; however, six-month measurements were post-recurrence for four patients.
These findings are hypothesis-generating and do not establish a clinical prediction tool or support treatment decisions based on the biomarkers. Future studies should use prespecified landmark analyses that include only patients recurrence-free at the end of each measurement window, report uncertainty for all performance estimates, and validate findings in independent multicenter cohorts.
Declarations
Acknowledgement
We appreciate Professor Victor T. Malkevich, the head of the Laboratory of Thoracic Oncopathology at N.N. Alexandrov National Cancer Center of Belarus, for his kind and fruitful cooperation and discussion of the study design.
Ethical statement
The study was approved by the Biomedical Ethics Committee of Belarusian State Medical University (Protocol No. 2, April 10, 2021). All patients provided written voluntary consent to participate in the study in accordance with the Declaration of Helsinki.
Data sharing statement
Data supporting the research article are available from the corresponding author at ataganovich@gmail.com.
Funding
The work was carried out with financial support from the Ministry of Health of the Republic of Belarus (grant number: 2.17/20220385).
Conflict of interest
The authors declare no conflicts of interest.
Authors’ contributions
Conceptualization (ADT, VIP); formal analysis, methodology, and original draft writing (ADT, MMK); investigation (MMK, EMB, OVG); review and editing (ADT, MMK, VIP). All authors have made a significant contribution to this study and have approved the final manuscript.