CD44 Loss in Gastric Stromal Tumors as a Prognostic Marker

Elizabeth Montgomery, Susan C. Abraham, Cyril Fisher, Mari Robinette Deasel, S. S. Amr, Salwa S. Sheikh, Michael House, Keith Lilliemoe, Michael Choti, Malcolm Brock, David T. Ephron, Mariana Zahuruk, Amy Chadburn

Research output: Contribution to journalArticlepeer-review

40 Scopus citations


Background: The adhesion molecule CD44 (CD44s; CD44H) and its isoforms (CD44v3-6 and v9) are preferentially expressed by different cell types. These transmembrane glycoproteins are involved in cell-cell and cell-matrix interactions and in cell trafficking and, thus, may play a role in tumor metastasis and/or local invasion. The expression pattern of CD44s and variant isoforms, particularly CD44v6 and CD44v9, of some neoplasms, including soft tissue tumors, correlates with clinical course and outcome. The clinical behavior of gastrointestinal stromal tumors (GIST) is site specific; however, other reliable predictors of clinical outcome have not been identified. Thus, the prognostic value of CD44s and isoform expression in GIST were evaluated by immunohistochemistry of tissue microarrays. Design: Paraffin-embedded formalin-fixed tissue cores (129: 103 GIST and 26 normal stomach smooth muscle) from 33 patients with clinical outcome data were collected and used for the construction of the tissue microarrays. One to five tissue cores from each patient specimen were evaluated (mean = 3 tissue cores/patient). Array slides were stained with anti-CD44s (CD44H) and with antibodies to v3, v4, v5, v6, and v9 isomers. CD44s and isoform expression and staining intensity were scored semiquantitatively without knowledge of patient identity or outcome: 0 = no; 1 = weak; 2 = moderate; 3 = moderate to strong; 4 = strong. The scores of multiple cores from the same GIST were averaged; the nonneoplastic smooth muscle was similarly graded. CD44s and isoform expression and intensity were compared with outcome. Results: The 33 patients with gastric GIST, 0.8 to 30 cm in size, were followed for 1 to 111 months with a median follow-up of 7 months (mean 17.5 months). The overall median survival was 25 months. Nine of the 33 (27%) patients had metastases, 9 (27%) had recurrent disease, and 9 (27%) died of disease (9-111 months; mean 39 months; median 23 months). All 18 patients with GIST CD44s expression >2+ were alive at last follow-up (1-62 months; median 3.5 months; mean 11 months). More than half (53%) of patients with GIST CD44s expression ≤2+ died (9-111 months; median 23 months; mean 38 months); the median follow-up of the surviving patients with CD44 expression ≤2 was 5 months (2-22 months; mean 6.5 months; log rank P = 0.07). The majority of tumors were variably positive CD44v3 and CD44v4, but there was minimal staining (number of cases and/or expression level) with antibodies directed against the v5, v6, and v9 isomers. Conclusion: These preliminary results suggest that although gastric GISTs variably express CD44s and variants, only the expression of CD44s correlates with clinical outcome with loss of CD44s positivity correlating with poor clinical outcome.

Original languageEnglish (US)
Pages (from-to)168-177
Number of pages10
JournalAmerican Journal of Surgical Pathology
Issue number2
StatePublished - Feb 1 2004


  • CD44
  • Gastrointestinal stromal tumors
  • Prognostic markers

ASJC Scopus subject areas

  • Anatomy
  • Surgery
  • Pathology and Forensic Medicine


Dive into the research topics of 'CD44 Loss in Gastric Stromal Tumors as a Prognostic Marker'. Together they form a unique fingerprint.

Cite this