Prognostic Factors Influencing Surgical Management and Outcome of Gastrointestinal Stromal Tumours

Br J Surg. 2003 Mar;90(3):332-9. doi: 10.1002/bjs.4046.


Background: The purpose of this study was to review surgical experience with gastrointestinal stromal tumours (GISTs) at a single tertiary university hospital, and to identify morphological and genetic prognostic markers of tumour progression.

Methods: Forty-eight GISTs from 39 patients were reviewed retrospectively. The prognostic significance of DNA copy number changes, measured by comparative genomic hybridization (CGH), and morphological markers in low-risk and high-risk tumours were investigated.

Results: Significantly more patients died from disease after incomplete tumour resection than after complete primary resection (P = 0.020). Tumour size of 5 cm or greater, mitotic count of 2 or more, and proliferative activity greater than 10 per cent were significantly associated with a shorter recurrence-free survival (P = 0.020, P = 0.001 and P = 0.002 respectively). Patients with low-risk tumours had a significantly better outcome than those with high-risk GISTs, both in terms of overall and recurrence-free survival (P < or = 0.001). CGH performed on 16 tumours revealed fewer DNA sequence copy number changes in low-risk than in high-risk GISTs. Non-progressive GISTs contained significantly fewer genetic alterations than recurrent or metastatic tumours (P < 0.001). Only tumours with more than five changes showed disease progression.

Conclusion: Complete surgical resection is the most important means of cure for GISTs. DNA copy number changes are related to the behaviour of these tumours and may serve as additional prognostic markers.

MeSH terms

  • Adult
  • Aged
  • Female
  • Gastrointestinal Neoplasms / surgery*
  • Humans
  • Immunohistochemistry
  • Male
  • Middle Aged
  • Neoplasm Recurrence, Local
  • Neoplasms, Connective Tissue / surgery*
  • Nucleic Acid Hybridization
  • Prognosis
  • Retrospective Studies
  • Survival Analysis
  • Tomography, X-Ray Computed / methods