Fatigue Emerges as the Central Symptom in the First Month After Esophageal Cancer Surgery

Esophageal cancer is the seventh leading cause of cancer-related death worldwide, and China carries roughly half of the global burden. The primary curative treatment, surgical resection of the esophagus, saves lives but leaves patients with a constellation of symptoms arising from the profound anatomical reconstruction of the digestive tract. Now, a prospective cohort study from Harbin Medical University Cancer Hospital has mapped that symptom landscape with unusual precision, combining two complementary statistical frameworks to identify which symptoms cluster together and which single symptoms act as the hubs of the entire postoperative network. The findings, published in Frontiers in Gastroenterology, point to fatigue as the pivotal therapeutic target in the first month after esophagectomy.

The research team enrolled 378 consecutive patients who underwent curative-intent esophagectomy for esophageal cancer between April 2023 and March 2024. After accounting for twelve patients lost to follow-up and five who died within the first postoperative month, the final analytic cohort comprised 361 participants. One month after surgery, each patient completed two widely validated quality-of-life instruments: the European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire Core 30, known as QLQ-C30, and the esophageal cancer-specific module QLQ-OES18. Together, these questionnaires captured eighteen distinct symptoms, ranging from dysphagia and reflux to cough, insomnia, pain, and constipation. The timing was deliberate. The first postoperative month is a critical transition period during which patients progress from enteral to oral nutrition, and more than half experience complications at this stage, including pulmonary infections and anastomotic leaks.

To identify symptom clusters, the researchers turned to exploratory factor analysis, a classical statistical technique that detects latent constructs explaining the covariance among observed variables. Symptoms with factor loadings of at least 0.40 were retained, and factor retention was guided by Kaiser’s criterion, parallel analysis, and scree plot evaluation. The analysis proved feasible, with a Kaiser-Meyer-Olkin measure of 0.70 and a highly significant Bartlett’s test of sphericity. Five factors emerged, accounting for 52.2 percent of cumulative variance. The team cross-validated this solution using principal component analysis and hierarchical cluster analysis, but both alternatives produced either overly broad groupings or clusters with considerable symptom overlap and poorly defined boundaries. The five-factor structure from exploratory factor analysis exhibited the clearest distinctions and closest consistency with clinical observations, so it was retained.

The five clusters were labeled eating-related, breathing-related, pain-related, fatigue-speech-related, and excretion-related. The eating-related cluster, encompassing eating difficulty, reflux, choking when swallowing, appetite loss, nausea and vomiting, and dysphagia, was present in 49.0 percent of patients. The breathing-related cluster, combining cough, dyspnea, and insomnia, affected 47.0 percent. The pain-related cluster, comprising esophageal pain and generalized pain, was present in 48.0 percent. The fatigue-speech-related cluster, uniting fatigue, dry mouth, and trouble talking, showed the highest prevalence at 51.0 percent. The excretion-related cluster, consisting of constipation and diarrhea, was present in 50.0 percent. Notably, approximately 67.0 percent of patients exhibited two or more concurrent clusters, indicating a substantial comorbid symptom burden.

The clustering of fatigue alongside dry mouth and speech difficulties is clinically justified, the authors argue, because dry mouth and dysarthria frequently co-occur during and following chemoradiotherapy, attributable to shared pathophysiological mechanisms such as salivary gland impairment and cranial nerve damage. Given that radiotherapy directly exacerbates fatigue, grouping these symptoms makes sense. The cluster’s status as the most prevalent in the cohort suggests it warrants dedicated management protocols. The researchers also identified two high-risk subgroups for postoperative fatigue: patients aged 60 to 74 years had adjusted odds of 1.84, and those receiving neoadjuvant therapy had adjusted odds of 1.86, compared with their respective reference groups.

To determine core symptoms, the team applied network analysis, an emerging computational modeling framework originally developed in psychopathology to investigate how symptoms interact. The researchers constructed a mixed graphical model with sixteen symptom nodes specified as continuous variables on a 0 to 100 severity scale, conditioned on three covariates: surgical method, postoperative hospitalization duration, and neoadjuvant therapy. Regularization was applied via graphical LASSO with the penalty selected by ten-fold cross-validation, and edges were retained only if selected by the LASSO procedure. The strongest conditional dependencies were observed between esophageal pain and generalized pain, between eating difficulty and choking when swallowing, and between nausea and vomiting and reflux. Fatigue showed high predictability, with 36.9 percent of its variance explained by neighboring symptoms.

Centrality analysis identified fatigue as the symptom with the highest strength centrality, with a relative strength of 2.83, followed by pain at 1.69 and insomnia at 1.58. Constipation exhibited the lowest strength centrality at 0.42. Critically, fatigue also demonstrated the highest bridge strength, at 1.09, confirming its dual role as both a core postoperative symptom and a pivotal bridge node interconnecting symptom clusters. Network stability was assessed through 1000 bootstrap samples, yielding a correlation stability coefficient of 0.52 for strength and expected influence centrality, which indicates good stability. Bootstrap analysis confirmed high edge weight accuracy, with narrow 95 percent confidence intervals closely aligning with the estimated weights.

The researchers then examined whether these symptom profiles predicted survival. Using a one-month landmark design to avoid immortal time bias, follow-up was counted from postoperative day 31, and the primary endpoint was one-year all-cause mortality. Of the 361 patients, 293 with complete one-year follow-up data were included in the primary analysis. During a median follow-up of 335 days, sixteen deaths occurred. Kaplan-Meier analysis showed no statistically significant differences in one-year survival between patients with or without any symptom cluster or core symptom. In univariable and multivariable Cox regression models, none of the symptom clusters or core symptoms were significantly associated with increased one-year mortality risk. The authors attribute this null finding partly to insufficient statistical power, given the limited number of mortality events and wide confidence intervals, and partly to contemporary improvements in esophageal cancer outcomes. A recent Swedish population-based study including 2291 patients who underwent esophagectomy between 2000 and 2020 demonstrated substantial survival improvements over time, with the adjusted hazard ratio for five-year all-cause mortality falling to 0.57 when comparing the most recent surgery period with the earliest.

The study has limitations that warrant cautious interpretation. It was conducted at a single tertiary hospital in China, which may limit generalizability to other populations or healthcare settings. The cohort was predominantly male, at 96.7 percent, reflecting the higher incidence of esophageal cancer in men but limiting applicability to female patients. Symptom presence was defined using a cutoff of at least 3 on the four-point Likert scale, which may underestimate subclinical burden in patients scoring 2. The cross-sectional nature of the network analysis cannot establish causal relationships between symptoms. The authors recommend longitudinal studies with larger, more diverse, and sex-balanced cohorts to validate the findings and delineate the temporal evolution of postoperative symptoms.

Nevertheless, the clinical implications are significant. Dysphagia showed the highest frequency at 93.1 percent, and fatigue the highest severity, with a mean score of 33.33 on the standardized scale. Evidence indicates that 30 to 50 percent of patients experience acute or chronic post-esophagectomy pain, and persistent insomnia may exceed twelve months after surgery. Yet 66 percent of patients do not report fatigue symptoms to clinicians, and 43 percent lack awareness of management strategies. The authors suggest that symptom network centrality metrics may serve as useful research endpoints in oncology practice, with routine clinical data applied to quantify symptom severity, construct symptom networks, and calculate centrality indices to guide evidence-based symptom management. Future studies should develop time-resolved dynamic symptom networks to track centrality trajectories and clarify causal interrelationships between symptoms. In an era of improving survival, precision management targeting fatigue represents a critical strategy for optimizing quality of life, distinct from survival-focused interventions, and reframes clinical priorities from solely extending survival to improving how patients actually live after esophageal cancer surgery.

Subject of Research: Health

Article Title: Symptom clusters and core symptoms in early postoperative esophageal cancer patients

Article References: Zhong, M., Wang, C., Liu, X., Wang, Y., Qi, X., & Xie, S. (2026). Symptom clusters and core symptoms in early postoperative esophageal cancer patients. Frontiers in Gastroenterology, 5, Article 1766306. https://doi.org/10.3389/fgstr.2026.1766306

Image Credits: AI Generated

DOI: 10.3389/fgstr.2026.1766306

Keywords: esophageal cancer, esophagectomy, symptom clusters, network analysis, fatigue, postoperative care, survival analysis, quality of life, Symptom, clusters, core, symptoms