Combining the Albumin-to-fibrinogen ratio and pathologic factors predicts survival in surgically treated patients with esophageal squamous cell carcinoma

Patients

The research comprised 414 individuals diagnosed with stage I–III ESCC who received surgical treatment at the First Hospital of Jilin University between January 2014 and December 2022. The inclusion criteria were as follows: (1) confirmed diagnosis of ESCC through endoscopy and pathology, (2) not receiving any neoadjuvant oncological treatment before surgery, (3) no prior history of malignancies, no acute or chronic infections such as bacterial, fungal, or tuberculosis, and no hematological system diseases or rheumatoid immunity-related diseases, (4) surgery was performed to achieve R0 resection, and (5) availability of complete perioperative and follow-up data. Patients who died from factors unrelated to esophageal cancer, patients who had a combination of other primary cancers, and patients who had received neoadjuvant antitumor therapy were excluded. All participants underwent radical esophagectomy along with systematic lymph node dissection, and staging was conducted in accordance with the 8th edition of the AJCC TNM staging system [15] for esophageal cancer. The research protocol was approved by the Research Ethics Committee of the First Hospital of Jilin University (No. 2024-1157). As this was a retrospective study, informed consent was exempted.

Data collection

The clinicopathological data included sex, age, body mass index (BMI), tumor site, tumor size, and TNM stage. Blood was collected before breakfast 2 weeks before surgery. Hematological tests included neutrophils, lymphocytes, monocytes, platelet count, globulin, albumin, and fibrinogen. The AFR is defined as the ratio of albumin content to fibrin content, while the LMR is the lymphocyte count divided by the monocyte count. The PLR represents the platelet count over the lymphocyte count, and the NLR is calculated by dividing the neutrophil count by the lymphocyte count. The SII is obtained by multiplying the platelet count by the neutrophil count and then dividing by the lymphocyte count. Additionally, the albumin-to-globulin ratio (AGR) is the albumin level compared to the globulin level, and the PNI is calculated as the albumin concentration in grams per liter plus five times the lymphocyte count expressed in 10⁹ per liter.

Follow-up

Follow-up assessments were performed from the time of surgery until January 2023 or until the patient’s death. Overall survival was defined as the interval from the date of surgery to either death attributable to esophageal cancer or the last follow-up visit. The initial follow-up occurred 1-month post-surgery, with subsequent evaluations scheduled every 3 months during the first year. In the second year, follow-ups were conducted biannually, and after 5 years, annual assessments were implemented.

Statistical analysis

Data analysis for this study was conducted using SPSS 26.0 and R 3.6.1 software RMS package. Medians and quartiles were used to represent the measurement information. The count data was expressed as a percentage. To determine the optimal cut-off points for the AFR and other biomarkers within the preoperative patient group, X-tile 3.6 software was utilized. Utilizing the identified optimal cut-off values, patients were divided into two distinct categories: those with elevated AFR levels and those with decreased AFR levels. Following this classification, the relationship between AFR and various clinical characteristics was analyzed employing either the chi-square test or Fisher’s exact test.

The Kaplan-Meier method was used to create survival curves, and log-rank tests were performed to assess the differences between these curves. Additionally, both univariate and multivariate analyses were carried out on clinical variables using the Cox proportional hazards regression approach. The hazard ratio (HR) and its corresponding 95 % confidence interval (CI) were computed to evaluate the associations between prognostic factors and survival duration. Based on the independent risk factors identified in the multivariate analysis, a nomogram prediction model was developed using the R package RMS to estimate the probability of OS at 3 and 5 years for patients undergoing surgical intervention for ESCC. Furthermore, the model’s discrimination and accuracy were assessed through the generation of the concordance index (C-index), receiver operating characteristic (ROC) curves, decision curve analysis (DCA), and calibration curves, all performed using R 3.6.1. p<0.05 was considered statistically significant.

Patient and tumor characteristics

This study included a total of 414 patients diagnosed with stage I–III ESCC, with their clinical characteristics detailed in Table 1. Among the participants, 184 patients (44.4 %) were under the age of 60, while 230 patients (55.6 %) were 60 years or older. The cohort comprised 22 female patients (5.3 %) and 392 male patients (94.7 %). Additionally, 292 patients (70.5 %) reported a history of smoking, and 311 patients (75.1 %) indicated alcohol consumption. Notably, 65 patients (15.7 %) had a prior diagnosis of hypertension, and 26 patients (6.3 %) had a history of diabetes mellitus. The proportions of patients with tumor length <3 cm and ≥3 cm were 39.6 and 60.4 %, respectively. The majority of tumors were located in the lower esophagus (64.7 %), followed by the middle esophagus (29.2 %). Pathologic differentiation showed that the proportions of poor, moderate, and well-differentiated were 21.3, 74.2, and 4.5 %, respectively. Pathological TNM staging revealed that 17.1 % of patients (71 individuals) were classified as stage I, 26.1 % (108 individuals) as stage II, and the majority, 56.8 % (235 individuals), as stage III. In T staging, 92 patients (22.2 %) were in the T1 stage, 65 patients (15.7 %) were in the T2 stage, and 257 patients (62.1 %) were in the T3 stage. In N staging, 163 patients (39.4 %) were in the stage N0, 151 patients (36.5 %) were in the stage N1, and 100 patients (24.1 %) were in the stage N2. There were 220 (53.1 %) and 166 (40.1 %) patients with lymphangio vascular invasion and perineural invasion, respectively. Follow-up data revealed that 223 patients (53.9 %) underwent adjuvant therapy after surgery.

X-tile determines cut-off values for clinical indicators

As shown in Figure 1, X-tile software was used to determine the optimal thresholds for preoperative nutritional and inflammatory biomarkers for all patients; the optimal cut-off value for AFR was 9.0. According to Supplementary Figure S1, the optimal cut-off values for NLR, PLR, LMR, SII, PNI, albumin, and AGR were 2.0, 123.3, 4.8, 826.6, 46.3, 39.4 and 1.4, respectively.

Figure 1:

The cutoff values of preoperative AFR in 414 patients with ESCC using X-tile software. (A) Tile plots graphically represent the data, with different values grouped as cutoffs for statistical tests. The data with the smallest p-value can be considered the best cutoffs, manifested as the brightest pixel dots on the tiled plot. (B) The histogram shows the distribution of the number of patients on either side of the cutoff value.

Correlation analysis between AFR and clinicopathologic characteristics

In this study, AFR was categorized into two groups based on the optimal cut-off value: the low-level group (AFR≤9.0) and the high-level group (AFR>9.0). This classification facilitated an examination of the relationship between AFR and various clinicopathological characteristics, as illustrated in Table 2. A low AFR demonstrated a significant association with tumor length, T-stage, N-stage, pathologic stage, and choroidal infiltration (all p<0.05). In addition, low AFR was significantly associated with anastomotic leakage resulting from surgical treatment (p<0.001). However, AFR was not correlated with sex, age, smoking, alcohol consumption, tumor location, degree of differentiation, and perineural invasion (all p>0.05).

Survival analysis

The overall median follow-up duration for the entire cohort was 56 months, with a range spanning from 2 to 118 months. The median survival duration was recorded at 42 months (95 %-CI: 31.85–52.15). By the end of the follow-up timeline, 224 individuals, accounting for 54.1 % of the cohort, had succumbed to factors related to EC. The observed OS rates were 84.1 % at 1 year, 54.6 % at the 3-year mark, and 43.3 % at 5 years.

Our results showed a more significant trend of decreasing survival curves in patients with low-level AFR. Moreover, the 5-year OS rate was markedly higher in patients exhibiting elevated levels of the AFR compared to those with low AFR levels (49.7 % vs. 10.1 %, p<0.001). Meanwhile, high-level SII, high-level PLR, high-level NLR, and low-level LMR showed poorer OS (all p<0.005), but there was no statistical significance in OS for different levels of albumin, AGR, and PNI (all p>0.05). The survival curves are depicted in Figure 2.

Figure 2:

Cumulative 5-year survival curves for ESCC patients. (A) AFR, (B) LMR, (C) NLR, (D) PLR, (E) SII, (F) PNI, (G) AGR, and (H) albumin. AFR, Albumin-to-fibrinogen ratio; LMR, lymphocyte-to-monocyte ratio; NLR, neutrophil-to-lymphocyte ratio; PLR, platelet-to-lymphocyte ratio; SII, systemic immune-inflammation index; PNI, prognostic nutritional index; AGR, albumin-to-globulin ratio.

As shown in Table 3, univariate analysis revealed that gender, smoking history, hypertension, tumor length, lymphangio vascular invasion, perineural invasion, pathological stage, T stage, N stage, degree of differentiation, NLR, PLR, SII, LMR, and AFR significantly affected the prognosis of patients who had undergone surgery for ESCC (all p<0.05). Conversely, factors such as age, alcohol consumption, BMI, cardiovascular disease, diabetes mellitus, tumor location, albumin levels, AGR, and PNI did not demonstrate any significant association with patient prognosis. The multivariate analysis identified independent risk factors influencing patient prognosis, including the AFR, lymphangio vascular invasion, pathological stage, and tumor differentiation degree (all p<0.05).

Subgroup analysis

The patients were categorized into two groups based on tumor stage, as illustrated in Figure 3. For patients in stages I and II, the survival curves revealed that those with a low AFR had a significantly lower 3-year OS rate compared to those in the high AFR group (20.0 % vs. 72.0 %, p<0.001). Similarly, among stage III patients, the survival curves indicated that the low AFR group also exhibited a significantly diminished 3-year OS rate relative to the high AFR group (29.2 % vs. 47.9 %, p<0.001).

Figure 3:

Survival curves for subgroup analysis of ESCC patients. AFR survival curves in patients with (A) stage I + II and (B) stage III ESCC.

A nomogram predicts survival risk based on AFR

The independent risk factors derived from the multifactorial COX-based analysis were incorporated into the nomogram. This model allows for the assessment of the impact of each independent factor on patient survival at 1, 3, and 5 years after surgery, as depicted in Figure 4. Among them, the C-index of the model was 0.737 (95 %-CI: 0.700–0.776).

Figure 4:

Nomogram predicts survival risk based on AFR and pathological factors in patients with ESCC.

The constructed model underwent validation. The internally validated calibration curves indicated that the 3- and 5-year OS probabilities predicted by the nomogram closely aligned with actual observations, as illustrated in Figure 5A and B. The DCA revealed that the model’s net gain in predicting 5-year OS was superior to that of other individual prediction models, including TNM staging and histological differentiation. The time-dependent ROC curve also showed that the nomogram incorporating AFR outperformed the models constructed from pathological factors alone in predicting 5-year OS in patients with ESCC, as shown in Figure 5C and D.

Figure 5:

Validation of nomogram accuracy and effectiveness. (A) 3-Year OS correction curve, (B) 5-year OS correction curve, (C) the decision curve analysis, (D) the time-dependent receiver operating characteristic curve. LVI, lymphangio vascular invasion.