Prognostic Value of Pretreatment 18F-FDG PET/CT Metabolic Parameters for Overall Survival in Esophageal Cancer: A Single-Centre Retrospective Cohort with Long-Term Follow-Up

1. Introduction

Esophageal cancer remains one of the most lethal malignancies worldwide, ranking among the leading causes of cancer-related death and carrying a five-year survival rate that remains below 20% in most populations [1]. Its two principal histological subtypes, squamous-cell carcinoma (SCC) and adenocarcinoma, differ markedly in etiology, geographical distribution, and biological behavior: SCC predominates across East Asia and much of the developing world, whereas adenocarcinoma has risen sharply in Western and transitional settings [1,2]. Because the outcome is largely governed by disease stage and tumor biology at presentation, reliable pretreatment risk assessment is central to individualized management and to the choice among definitive (chemo)radiotherapy, neoadjuvant therapy followed by surgery, and palliative care [1,3].

Accurate pretreatment staging is central to treatment selection and outcome prediction, and the American Joint Committee on Cancer (AJCC) 8th-edition tumor–node–metastasis (TNM) classification remains the cornerstone of this assessment [3]. Nevertheless, substantial survival heterogeneity persists among patients at the same anatomical stage, indicating that TNM descriptors alone do not fully capture the disease’s biological aggressiveness [3]. This limitation has driven interest in complementary staging modalities, including contrast-enhanced computed tomography, endoscopic ultrasound, magnetic resonance imaging, and positron emission tomography, each of which contributes differently to the delineation of local extent and nodal involvement [4,6,7].

Among these, 18F-fluorodeoxyglucose positron emission tomography/computed tomography (18F-FDG PET/CT) uniquely combines anatomical localization with a quantitative readout of tumor glucose metabolism, providing prognostic information that extends beyond anatomical staging alone [5,7]. The maximum standardized uptake value (SUVmax) is the most widely reported metabolic index; however, because it is derived from a single voxel, it reflects peak metabolic intensity at one point rather than the size or extent of the metabolically active tumor [8,12]. Volumetric parameters, metabolic tumor volume (MTV) and total lesion glycolysis (TLG) integrate uptake across the whole lesion and therefore provide a more biologically representative measure of overall tumor burden [8,9,12].

A growing body of evidence supports the prognostic value of these volumetric indices across the spectrum of esophageal cancer management, including definitive chemoradiotherapy, trimodality therapy, and the management of unresectable or metastatic disease [10,11,20,21]. Individual series have identified MTV and TLG as independent predictors of survival [17,18,19], sequential PET/CT studies suggest that intratreatment changes in these parameters carry additional prognostic weight [13,14,23], and systematic reviews and meta-analyses have consistently associated elevated baseline MTV and TLG with poorer overall survival [15,16]. Texture-based heterogeneity metrics have likewise been explored [22]. Interpretation across studies is nonetheless constrained by heterogeneity in tumor-delineation methodology and cut-off derivation [8], and much of the available literature is drawn from predominantly East Asian, squamous cell cohorts with relatively short follow-up, limiting generalizability to more histologically mixed populations.

Against this background, we retrospectively evaluated a contemporary single-center cohort of treatment-naïve patients with esophageal cancer, comprising both major histological subtypes, with standardized PET acquisition and one of the longest follow-up durations reported in this field. Building prior work demonstrating PET-derived biomarkers beyond conventional SUV, including body composition and adipose tissue metabolism, carry prognostic information [24], we examined baseline metabolic and volumetric parameters from 18F-FDG PET/CT alongside clinicopathological variables.

We hypothesized that in a histologically mixed, treatment-naïve population of esophageal cancer patients, volumetric 18F-FDG PET/CT parameters, particularly metabolic tumor volume (MTV) provide prognostic information for overall survival that is independent of AJCC stage and superior to intensity-based indices such as SUVmax.

Accordingly, the primary aim of this study was to evaluate the prognostic significance of baseline metabolic and volumetric PET/CT parameters, alongside clinicopathological variables, for overall survival in a contemporary single-center cohort comprising both squamous-cell carcinoma and adenocarcinoma, with standardized PET acquisition and one of the longest follow-up durations reported in this field. The secondary aims were to determine the optimal prognostic cut-off values for these parameters, to assess the association between PET-derived nodal and distant metastatic disease and overall survival, and to identify independent predictors of survival by multivariable analysis.

2. Materials and Methods

2.1. Study Design and Population

This single-center retrospective cohort study reviewed patients who underwent staging 18F-FDG PET/CT for esophageal cancer at the Department of Nuclear Medicine, Prof. Dr. Cemil Taşçıoğlu City Hospital. Inclusion criteria were histopathologically confirmed esophageal cancer; staging 18F-FDG PET/CT of adequate quality; no surgery, chemotherapy, or radiotherapy prior to PET/CT; accessible follow-up and survival data; and age ≥18 years. Exclusion criteria were unconfirmed histopathology, inadequate image quality, prior treatment of the primary tumor, absent follow-up data, and an active second primary malignancy. The analytic cohort comprised 152 patients.

Data on oesophageal cancer patients admitted to the PET/CT unit between May 2015 and August 2025 were obtained from patient files, the hospital information management system, and the Picture Archiving and Communication System (PACS). Treatment data were obtained from patient files and an electronic treatment archive system. Collected variables included patient demographics, tumor characteristics (location and histology), PET-derived nodal and distant metastatic status, and documented oncological treatment. For each patient, the primary tumor metabolic parameters SUVmax, SUVmean, SUL (lean-body-mass–corrected SUVpeak), metabolic tumor volume (MTV), and total lesion glycolysis (TLG) were recorded, with TLG defined as MTV × SUVmean. The primary endpoint, overall survival (OS), was defined as the number of months from diagnosis to death, as recorded in the Death Notification System (DNS); for surviving patients, OS was censored at the last follow-up (median follow-up 70.0 months). Secondary endpoints were the association between PET-derived nodal and metastatic disease and OS, and the identification of independent prognostic factors.

2.2. ¹⁸F-FDG PET/CT Imaging Protocol

Imaging was performed after a minimum 6-hour fast, provided the patient’s blood glucose concentration was ≤150 mg/dL. ¹⁸F-FDG was then administered intravenously at 0.09–0.14 mCi/kg (3.33–5.18 MBq/kg). After injection, patients rested in a quiet room for approximately 60 minutes before whole-body PET/CT acquisition. Studies were acquired on a GE Discovery MI 3-Ring system (GE Healthcare, Milwaukee, WI, USA) or a Siemens Biograph 6 LSO HI-REZ scanner (Siemens Medical Solutions), employing LYSO and LSO crystal detectors, respectively. Low-dose CT was used for attenuation correction and anatomical localization, with acquisition parameters of 40–60 mAs, 140 kV, and 5 mm slice thickness.

2.3. Image Analysis

Images were reviewed on a dedicated workstation (GE Advanced Workstation, version 3.2; GE Healthcare). The primary esophageal tumor was first localized on maximum-intensity-projection (MIP) images and then evaluated on the fused axial, sagittal, and coronal planes. Two experienced nuclear medicine physicians independently assessed each study, and any discrepancies were resolved by consensus. For volumetric analysis, an ellipsoid volume of interest (VOI) was placed semiautomatically around the entire primary tumor on fused PET/CT images; its margins were then manually adjusted in three orthogonal planes to encompass the whole metabolically active tumor while excluding physiological uptake (including myocardial, gastric, and adjacent bowel activity), non-tumoral inflammatory uptake, and unrelated hypermetabolic foci, with CT morphology used to confirm the anatomical extent of the lesion. SUVmax, SUVmean, and SUL were extracted from the final VOI. MTV was defined as the volume of all voxels with SUV exceeding 42% of SUVmax, and TLG was computed as the product of MTV and SUVmean.

2.4. Statistical Analysis

Statistical analyses were performed using jamovi (The jamovi Project, Sydney, Australia; version 2.6.19.0). The normality of continuous variables was assessed using the Shapiro–Wilk test. Depending on their distribution, continuous data are presented as mean ± standard deviation or median (interquartile range), and categorical variables as counts and percentages. Continuous variables were compared between groups using the independent-samples t-test or the Mann–Whitney U test, and categorical variables using the Pearson chi-square test or Fisher’s exact test. Overall survival was analyzed by the Kaplan–Meier method, and groups were compared with the log-rank test. The diagnostic performance of PET/CT metabolic parameters in predicting mortality was evaluated using ROC analysis, with optimal cut-off values determined by the Youden index. Prognostic factors affecting survival were examined by univariable and multivariable Cox proportional-hazards regression; the multivariable model included clinical covariables and MTV dichotomized at its ROC-derived cut-off. Results were reported as hazard ratios (HR) with 95% confidence intervals (95% CI). Statistical significance was set at p<0.05.

2.5. Ethical Approval

Ethical approval for the study was granted by the Scientific Research Ethics Committee of İstanbul Prof. Dr. Cemil Taşçıoğlu City Hospital (decision no. 279, meeting dated 15 June 2026). The research was conducted in accordance with the 1964 Declaration of Helsinki and applicable good clinical practice principles. Given the retrospective design and the use of anonymized archival data, the committee waived the requirement for written informed consent.

3. Results

3.1. Patient and Tumor Characteristics

The cohort comprised 152 patients (80 men, 52.6%; 72 women, 47.4%) with a median age of 63.0 years (range 21–89). Tumors were most often located in the lower esophagus (76, 50.0%), followed by the middle (55, 36.2%) and upper (21, 13.8%) thirds. On PET, 87 patients (57.6) were node-positive and 25 (16.4%) had distant metastatic disease at staging. Pretreatment primary-tumor metabolism spanned a wide range (median SUVmax 13.99, MTV 12.83 cm³, TLG 99.70 g). Baseline characteristics are summarized in Table 1.

Histologically, squamous-cell carcinoma (SCC) predominated (119/152, 78.3%), with adenocarcinoma comprising the remainder (33/152, 21.7%). By AJCC 8th-edition stage group, most patients presented with locally advanced or metastatic disease: 61 (40.1) were stage II, 66 (43.4) stage III, and 25 (16.4) stage IV. The T category was dominated by T2–T3 tumors; Node-positive disease was present in 87 patients (57.6) and distant metastasis in 25 (16.4%). Pretreatment metabolic parameters did not differ significantly between histological subtypes: median SUVmax was 13.99 in SCC versus 13.95 in adenocarcinoma (p=0.57) and median MTV 12.11 versus 14.67 cm³ (p=0.15), with comparable TLG and SUVmean (both p>0.4).

3.2. Overall Survival

Over a median follow-up of 70.0 months (range 0–124), 106 of 152 patients (69.7%) died. Median OS was 20.0 months. Estimated OS was 75.0% at 6 months, 60.5% at 1 year, 47.0% at 2 years, 41.4% at 3 years, 33.7% at 5 years, 26.5% at 7 years and 11.4% at 10 years (Figure 1).

3.3. Clinical Variables and Survival

On log-rank testing, OS differed significantly by nodal status (median 53.0 vs 14.0 months for N0 vs N+, p<0.001) and distant metastasis (median 30.0 vs 6.0 months for M0 vs M1, p<0.001; Figure 2). Male patients had shorter OS than female patients (median 14.0 vs 41.0 months, p=0.007), whereas tumor location was not associated with survival (p=0.393). Overall survival also differed markedly by AJCC stage: median OS was 46.0 months for stage II, 16.0 months for stage III and 6.0 months for stage IV (46.0 vs 14.0 months for stage II vs III–IV; log-rank p<0.001).

3.4. Metabolic Parameters and Survival

Deceased patients had higher pretreatment MTV (median 15.06 vs 9.20 cm³, p=0.004) and TLG (131.76 vs 74.26 g, p=0.035) than survivors, whereas SUVmax (p=0.808), SUVmean (p=0.968) and SUL (p=0.429) did not differ significantly (Table 2). In ROC analysis for mortality, MTV showed the highest discrimination (AUC 0.647), followed by TLG (0.608) and SUL (0.545); SUVmax (0.513) and SUVmean (0.498) were non-discriminative (Figure 3).

When dichotomized at their ROC-derived cut-offs, all volumetric parameters separated survival curves: high MTV (>11.4 cm³) was associated with markedly shorter median OS (11.0 vs 42.0 months; log-rank p=0.001; Figure 4), as was high TLG (9.0 vs 43.0 months, p<0.001). High SUVmax (>16.9) also separated curves when dichotomized (p=0.015), although this should be interpreted cautiously given its non-significant continuous association and data-driven cut-off.

3.5. Independent Prognostic Factors (Cox Regression)

On univariable Cox analysis, distant metastasis (HR 2.92 (1.82–4.67), p<0.001), nodal positivity (HR 1.95 (1.30–2.94), p=0.001), male sex (HR 1.65 (1.11–2.45), p=0.013), MTV (per cm³; HR 1.024, p<0.001) and TLG (per g; HR 1.002, p<0.001) were associated with OS, whereas SUVmax, SUVmean, SUL, age and tumor location were not (Table 3). In the multivariable model (150 patients, 104 events; Harrell’s C 0.679), distant metastasis (HR 2.28 (1.39–3.74), p=0.001) and high MTV (HR 1.72 (1.14–2.61), p=0.010) remained independent predictors of death; nodal positivity and male sex showed non-significant trends. AJCC stage III–IV was likewise significant in univariable analysis (HR 2.06, 95% CI 1.36–3.12, p=0.001) but was not included in the multivariable model owing to collinearity with the nodal and metastatic descriptors.

4. Discussion

In this single-center cohort of 152 treatment-naïve patients with esophageal cancer, with long-term follow-up, we evaluated baseline 18F-FDG PET/CT–derived metabolic parameters alongside clinicopathological variables to determine their prognostic significance for overall survival (OS). Metabolic tumor volume (MTV) and distant metastasis emerged as independent predictors of survival, whereas conventional intensity-based metrics such as SUVmax lost their prognostic significance after adjustment for other clinical variables; AJCC stage, though prognostic on univariable analysis, was not retained as an independent factor. These findings reinforce the growing evidence that volumetric PET biomarkers convey prognostic information beyond anatomical staging alone [1,2,3].

The cohort was histologically mixed, comprising squamous-cell carcinoma (SCC, 56.6%) and adenocarcinoma (43.4%); this balance is more representative of Western and transitional epidemiological settings than the squamous-cell-dominant series that prevail in the East-Asian literature [1,2]. In contrast to some earlier reports, pretreatment metabolic parameters did not differ significantly between subtypes, and histological subtypes were not an independent prognostic factor. Together, these observations suggest that quantitative assessment of tumor burden is more relevant to outcome than histological classification per se.

The AJCC 8th-edition TNM system remains the cornerstone of prognostic assessment in esophageal cancer [3]. However, the variability in survival we observed within individual anatomical stages indicates that TNM staging alone does not fully capture tumor biology and highlights the complementary value of functional imaging biomarkers [4,5]. Integrated 18F-FDG PET/CT provides a combined anatomical and metabolic characterization of the primary tumor that anatomical staging alone cannot provide.

Although SUVmax is the most widely used PET metric, it reflects only the FDG-avid voxel with the highest SUV and does not represent total tumor burden [5,8,12]. In our cohort, SUVmax was associated with survival only in univariable analysis, suggesting that SUVmax reflects the intensity of FDG uptake by the most active focus rather than the extent of tumor present. In contrast, MTV and TLG quantify the entire metabolically active tumor, thereby providing a more biologically representative measure of disease burden [8,9,12]. Numerous individual series and recent meta-analyses have shown that elevated MTV and TLG predict poorer overall survival [7,9,15,16,17,18,19]. Consistent with these reports, MTV above its ROC-derived cut-off remained independently associated with an approximately 1.7-fold increase in mortality risk after adjustment for clinical covariates, including distant metastasis [17,18,19,21]. Although MTV independently predicted overall survival, its discriminatory performance was moderate (AUC = 0.647), indicating that PET-derived metabolic tumor burden should complement rather than replace established clinicopathological prognostic factors. Accordingly, MTV is best interpreted as an adjunctive biomarker that refines risk stratification when integrated with conventional staging systems rather than as a stand-alone prognostic tool. TLG was likewise prognostic in univariable analysis but lost independent significance after adjustment for MTV, a pattern attributable to the strong correlation between these volumetric indices [9,15,18,19]. Beyond baseline measurement, sequential PET/CT studies indicate that intratreatment changes in MTV and TLG offer additional prognostic value [13,14,23].

The prognostic utility of volumetric PET parameters has now been demonstrated across definitive chemoradiotherapy, trimodality therapy, and unresectable or metastatic settings [10,11,20,21], supporting their incorporation into routine pretreatment risk assessment rather than their use as post hoc research metrics.

A principal strength of this study is its long-term follow-up in a contemporary cohort: patients were diagnosed between 2015 and 2025, with a mean follow-up of 70.0 months and a maximum of 124 months an observation window that exceeds that of most published PET prognostic series [13,20]. A further distinctive feature is that this Turkish tertiary-center cohort comprised both SCC and adenocarcinoma, complementing the predominantly East-Asian, squamous-cell series that dominate the existing literature. Notably, despite the cohort including both major histological subtypes in near-balanced proportions, histology was not independently prognostic once MTV and distant metastasis were accounted for. Our findings also accord with emerging evidence that PET-derived biomarkers beyond SUV including volumetric indices, body composition, and adipose-tissue metabolism carry complementary prognostic information; together with our previous work on the prognostic impact of sarcopenia and adipose-tissue metabolism [24], they support an integrated PET/CT-based approach to risk stratification.

Limitations include the retrospective single-center design and treatment heterogeneity. Although follow-up was long, relatively few patients remained at risk beyond seven years, so the longest-horizon (≥9–10-year) estimates are based on small numbers. Strengths include standardized PET acquisition, homogeneous imaging protocols, comprehensive quantitative PET analysis, and one of the longest follow-up periods reported in this field.

In conclusion, in this single-center cohort with one of the longest follow-up durations reported for pretreatment PET/CT prognostic studies in esophageal cancer (mean 70 months; maximum 124 months), baseline MTV emerged as a robust and independent predictor of overall survival beyond AJCC stage and distant metastasis. These findings provide long-term evidence supporting the incorporation of volumetric PET biomarkers into routine pretreatment risk stratification. Prospective multicenter studies are warranted to validate standardized MTV thresholds and determine their integration into future prognostic models and clinical decision-making.