A new tool for assessing an individual's risk of recurrence of gastrointestinal stromal tumours following surgery is an important complement to current predictive models and will help with the development of more accurate selection of high-risk patients who are most likely to benefit from additional treatment. The study, published Online First in the Lancet Oncology, indicates that the new prognostic maps could reduce costs and minimise unnecessary side-effects in the 60% of patients who are likely to be cured by surgery alone.
At present many patients with GIST, the most common soft tissue cancer of the intestinal tract, are given adjuvant imatinib therapy despite many having a good chance of being cured by surgery alone. Therefore, accurately predicting the risk of recurrence for each patient and identifying those most likely to benefit from additional treatment is vital to ensure that those at high risk of recurrence can be aggressively treated, while those at low risk are not over treated and exposed to unnecessary side-effects.
Current risk-prediction methods, developed using the established prognostic factors of tumour size and mitotic count, are widely used but their accuracy at predicting the risk of recurrence is unknown.
In this study, a team led by Heikki Joensuu from Helsinki University Central Hospital, Helsinki, Finland, amassed a database on 2560 patients with operable GIST who did not receive adjuvant therapy by pooling individual patient data from population-based cohorts. They assessed key prognostic factors for recurrence free survival (RFS), compared the performance of three widely used risk-prediction methods (the National Institute of Health [NIH], the modified consensus criteria, and the Armed Forces Institute of Pathology [AFIP] criteria), and developed a new method for estimating the risk of GIST recurrence.
Findings showed that most (59.9%) patients were cured by surgery alone, and would not benefit from adjuvant therapy.
Comparison of the three main risk-stratification models found that they all estimated 10-year risk of GIST recurrence fairly accurately. The modified NIH consensus criteria were the best at identifying a single group of patients with unfavourable prognosis that were most likely to benefit from adjuvant therapy.
Since tumour size and mitosis count showed a non-linear association with the risk of recurrence, the researchers developed new prognostic maps to more accurately portray the continuous and non-linear nature of these variables, taking into account tumour location and rupture (identified as an independent risk factor).
The resulting prognostic maps provided the most accurate prognosis for individual GIST patients compared with the conventional models that stratified patients into a few broad groups.
The authors conclude: "In practice, the modified consensus criteria identify well those patients who have little risk of GIST recurrence and who may thus not be candidates for adjuvant therapies. There are, however, patients whose risk borders the high-risk group or who wish their risk to be estimated using another method and perhaps more individually. The prognostic heat maps and contour maps can be illustrative and helpful in such cases."*
In a Comment, Anette Duensing from the University of Pittsburgh Cancer Institute, Pittsburgh, USA, says that the results: "Give clinicians a solid foundation, method, and reason to separate the subset of high-risk patients who are likely to benefit from adjuvant therapy from those who will do just as well without it. This personalised approach will ultimately reduce costs and side-effects in patients who are cured by surgery alone, and allow a focus on high-risk patients who need more intense treatment."
She adds: "Although the current risk-prediction schemes will probably remain standard for pathologists, the contour maps will be crucial for oncologists when discussing individual risks with patients, since they are graphic and easy to explain."
Professor Heikki Joensuu, Helsinki University Central Hospital, Helsinki, Finland. T) 358)-40-72-10-438 E) email@example.com
Professor Anette Duensing, University of Pittsburgh Cancer Institute, Pittsburgh, USA. T) 412-623-5870 E) firstname.lastname@example.org
Notes to Editors
*Quote direct from author and cannot be found in text of Article