Prognostic significance of tumor length in patients receiving esophagectomy for esophageal cancer

• Published in: Journal of surgical oncology Vol. 116; no. 8; pp. 1114 - 1122
• Main Authors: Hollis, Alexander C., Quinn, Lauren M., Hodson, James, Evans, Emily, Plowright, James, Begum, Ruksana, Mitchell, Harriet, Hallissey, Mike T., Whiting, John L., Griffiths, Ewen A.
• Format: Journal Article
• Language: English
• Published: United States Wiley Subscription Services, Inc 15-12-2017
• Subjects: Cancer surgery; Esophageal cancer; esophagectomy; Esophagus; Gastrointestinal surgery; Pathology; Patients; recurrence; survival; tumor length; recurrence; esophagectomy; esophageal cancer; tumor length; survival
• ISSN: 0022-4790; 1096-9098
• DOI: 10.1002/jso.24789

Aims
We investigated the prognostic value of tumor length measurements acquired both from pre‐operative imaging and post‐operative pathology in esophageal cancer.
Methods
Tumor lengths were examined retrospectively for 389 esophagectomy patients with respect to Endoscopy, EUS (Endoscopic Ultrasound), CT and PET‐CT, and pathology. Correlations between the measurements on the different approaches were assessed, and associations between tumor length and survival were analyzed.
Results
Only the tumor lengths assessed on pathology were found to be significantly associated with overall (P = 0.001) and recurrence free (P < 0.001) survival on univariable analysis. The median overall survival was 47.1 months in those patients with tumor lengths <3.0 cm, falling to 19.6 and 18.0 months in those with 3.0‐4.4 and 4.5+ cm tumors, respectively, demonstrating a reduction in patient survival at a tumor length of around 3 cm. Tumor length on pathology was significantly correlated with tumor differentiation and both T‐ and N‐categories. After accounting for these factors, tumor length on pathology was a significant independent predictor of recurrence‐free (P = 0.016), but not overall (P = 0.128) survival.
Conclusions
Tumor lengths on pathology were found to be the most predictive of patient outcome. However, after accounting for other tumor‐related factors, tumor length only resulted in a marginal improvement in predictive accuracy.