Assessment of metastatic risk of gastric GIST based on treatment-naïve CT features

https://doi.org/10.1016/j.ejso.2016.03.032Get rights and content

Abstract

Objective

To study whether the CT features of treatment-naïve gastric GIST may be used to assess metastatic risk.

Methods

In this IRB approved retrospective study, with informed consent waived, contrast enhanced CT images of 143 patients with pathologically confirmed treatment-naïve gastric GIST (74 men, 69 women; mean age 61 years, SD ± 14) were reviewed in consensus by two oncoradiologists blinded to clinicopathologic features and clinical outcome and morphologic features were recorded. The metastatic spread was recorded using available imaging studies and electronic medical records (median follow up 40 months, interquartile range, IQR, 21–61). The association of maximum size in any plane (≤10 cm or >10 cm), outline (smooth or irregular/lobulated), cystic areas (≤50% or >50%), exophytic component (≤50% or >50%), and enhancing solid component (present or absent) with metastatic disease were analyzed using univariate (Fisher's exact test) and multivariate (logistic regression) analysis.

Results

Metastatic disease developed in 42 (29%) patients (28 at presentation, 14 during follow-up); 23 (16%) patients died. On multivariate analysis, tumor size >10 cm (p = 0.0001, OR 9.9), irregular/lobulated outline (p = 0.001, OR 5.6) and presence of a enhancing solid component (p < 0.0001, OR 9.1) were independent predictors of metastatic disease. On subgroup analysis, an irregular/lobulated outline and an enhancing solid component were more frequently associated with metastases in tumors ≤5 cm and >5–≤10 cm (p < 0.05).

Conclusion

CT morphologic features can be used to assess the metastatic risk of treatment-naïve gastric GIST. Risk assessment based on pretreatment CT is especially useful for patients receiving neoadjuvant tyrosine kinase inhibitors and those with tumors <5 cm in size.

Introduction

Gastrointestinal stromal tumors (GIST) are the most common mesenchymal tumor of the gastrointestinal tract.1, 2 Within the gastrointestinal tract, GIST occurs most frequently in the stomach (40–60%) followed by the small intestine (25–35%), colorectum (5–15%) and esophagus (<1%).3, 4, 5, 6

Risk stratification of GIST is currently based on size, mitotic count and tumor location. The NIH 2002 consensus defined high risk GIST as1 size >5 cm and mitotic rate >5/50 HPF (high power field),2 size >10 cm with any mitotic rate or3 any size with a mitotic rate > 10/50 HPF.7 The Armed Forces Institute of Pathology divides GIST into very low risk, low, moderate and high risk based on tumor size and mitotic count with high risk for gastric GIST defined as size >5 cm, >5/50 HPF.8 In a series of 1765 gastric GIST 86% of tumors >10 cm and with >5/50 mitoses per HPF metastasized.9 Gastric GIST has a lower risk of metastatic disease than a non-gastric GIST of the same size and mitotic count.10

Advent of tyrosine kinase inhibitors has radically changed the management and prognosis of patients with GIST.11, 12 Patients with large primary tumors are often treated with neoadjuvant tyrosine kinase inhibitors limiting the assessment of an accurate mitotic count in the final resection specimen because of post-therapy changes. The mitotic count cannot always be accurately determined on biopsy specimen because of small sampling size, sampling error and tumor heterogeneity.13 This understanding of intratumoral heterogeneity has led to increasing interest in correlating the imaging morphology with tumor genomics and cancer biology. Since most patients with GIST undergo a pretreatment CT, risk-assessment based on the CT features would be helpful. Since stomach is the most common site of GIST, we focused this exploratory study on gastric GIST. Our purpose was to study if CT features of treatment-naïve gastric GIST can be used to assess its metastatic risk.

Section snippets

Study population

In this institutional review board-approved, HIPAA –compliant, retrospective study performed at a tertiary cancer center, informed consent was waived. The electronic radiology database was searched from January 1998 to October 2012 to identify all the patients with pathologically proven treatment-naïve gastric GIST in whom pretreatment contrast-enhanced CT was available for review. A total of 143 patients met the inclusion criteria and were included in the study (74 men, 69 women; mean age 61

Results

Table 1 summarizes the patient characteristics and Table 2 shows the CT features of the 143 GISTs included in this study. The mean tumor size in this study was 7.6 cm (SD 5.9). A total of 57 (40%) tumors had an irregular/lobulated outline and 83 (58%) tumors were >50% exophytic in location. A total of 80 (56%) tumors demonstrated the presence of an enhancing solid component.

Surgery was performed in 127/143 patients (89%), with the majority undergoing partial gastrectomy (n = 88/127, 69%) or

Discussion

The risk stratification of gastric GIST is currently based on the size of the tumor and the mitotic count. However, in patients receiving neoadjuvant tyrosine kinase inhibitors the assessment of the mitotic count in the final resection specimen is not reliable. Furthermore, assessment of mitotic count on pretreatment biopsy is also not reliable. Risk stratification using morphological features from a treatment-naïve CT would therefore be helpful in planning of treatment and follow-up strategy.

Conflict of interest statement/Diclosures

Ailbhe C O'Neill, Vikram Kurra, Sree Harsha Tirumani, Jyothi P Jagannathan, Akshay D Baheti, Jason L Hornick and Nikhil H Ramaiya have no disclosures.

Suzanne George has consulting and research funding from Bayer, Novartis, Pfizer and ARIAD.

This research project was not funded; however, Atul Shinagare is now supported by RSNA Research Scholar Grant, #RSCH1422. The grant support began after this research project was completed, but before submission of the manuscript to EJSO.

References (23)

  • V.P. Koshenkov et al.

    Adjuvant therapy of gastrointestinal stromal tumors

    Curr Opin Oncol

    (2012)
  • Cited by (25)

    View all citing articles on Scopus
    View full text