In this issue, Andtbacka et al. report their experience with using the combination of preoperative tyrosine kinase inhibition (TKI) for locally unresectable or metastatic gastrointestinal stromal tumor (GIST). These two types of advanced GIST are quite distinct from a biologic and therapeutic perspective. Therefore, it would have been preferable to consider them separately, instead of combining the data in several places in the manuscript (e.g., Tables 1 and 3). Locally unresectable and metastatic GIST will be addressed sequentially here. There are several potential reasons to use neoadjuvant TKI prior to exploring a patient with locally unresectable GIST. The most obvious reason is to increase the likelihood of complete gross resection of the tumor. In this report, the median decrease in tumor volume was 85% in the 11 patients with locally unresectable GIST. All patients were able to undergo complete gross tumor removal (although two did not have preoperative tumor shrinkage). While it is often difficult to establish precise criteria for what constitutes a locally unresectable primary tumor, the authors definition of ‘‘significant involvement of a single organ with tumor size ‡5 cm or extension to adjacent organs’’ is inadequate and raises doubt about the baseline resectability of the tumors. The other reason to use neoadjuvant TKI is to minimize the sacrifice of normal tissue. For instance, a person with a low rectal GIST may otherwise require an abdominoperineal resection, but after TKI therapy the operation may be converted to a coloanal or low anterior resection. While the authors state vaguely that the surgeons felt that the operations were ‘‘less extensive’’ because of neoadjuvant TKI therapy, it would have been more meaningful to estimate how the operations were actually altered. There may be other benefits to neoadjuvant TKI therapy besides reducing normal tissue loss, such as decreasing intraoperative blood loss from these normally hypervascular tumors. The optimal duration of neoadjuvant TKI therapy has not been established. The authors treated the patients until consecutive CT scans showed no further change in size, density, or vascularity, which occurred at a median of 11.9 (2.8–15) months. However, the decision should be based on when the rate of response diminishes to the point that the consulting surgeon feels that the operation should be performed. In our experience, the incremental tumor response is minimal after six months of therapy, but waiting for consecutive CT scans to show no change at all may actually take a while. The reason to operate sooner rather than later is that the median time to imatinib mesylate resistance is less than two years. It is important to know the potential risks of neoadjuvant therapy. Overall, about 15% of patients demonstrate primary resistance to imatinib from the initiation of treatment. Thus, there is a small risk that a patient who has a marginally resectable tumor may lose the opportunity to undergo an attempt at tumor resection. For these patients, early (within a month) cross-sectional imaging is important to determine if Received May 9, 2006; accepted May 23, 2006; published online September 7, 2006. Address correspondence and reprint requests to: Ronald P. DeMatteo, MD; E-mail: dematter@mskcc.org