PET is performed by injecting a patient with a radiolabeled tracer which is concentrated by the body in certain metabolically active tissues. As radioactive decay occurs, emissions are measured with a scanner and a threedimensional image representing relative uptake of the tracer is produced. 2-[fluorine 18] fluoro-2-deoxy-D-glucose (FDG) labeled glucose is used most frequently as the tracer, and this paper will assume the use of FDG unless otherwise indicated. As f luorine-labeled glucose is transported into metabolically active cells, it is phosphorylated and trapped, ensuring that continued dissipation and transport do not dilute the signal. These biochemical properties make FDGPET a useful modality for measuring glucose demand as a surrogate for metabolically active tissues such as cancer. In several gastric cancer histologies, however, the metabolic differential between tumor and normal tissue is not as stark as with other malignancies, making the conceptual utility of PET less clear. Mucinous carcinoma, signet ring cell carcinoma, and poorly differentiated adenocarcinomas typically have less prominent FDG uptake (7, 8).

The American Joint Committee on Cancer (AJCC) staging system is widely used for the characterization of disease burden and prognosis in gastric cancer. Based on a TNM system, the 7^th edition of AJCC guidelines designate tumor characteristic staging (T) as follows: T1 when tumor invades lamina propria or muscularis mucosae, T2 when tumor invades muscularis propria, T3 when tumor penetrates subserosal tissue without further invasion, and T4 when tumor invades visceral peritoneum or adjacent structures (9). Because surgical treatment is a major prognostic factor, effort to accurately determine the invasiveness of a gastric lesion is crucial. CT-determined T staging agreed closely with pathologic staging in early studies but was subsequently shown to have disappointing accuracy. EUS is a more accurate method for determination of pre-operative T stage and was directly compared with CT in a study by Botet (10). However, evolving technologies produce ever-increasing resolution of CT imaging, and thinsection scans with multiplanar reformation and contrast suggest the comparative value between CT and EUS is not static (11).

PET may predict response to preoperative chemotherapy in gastric cancer. Ott et al. showed that a 35% decrease in uptake between pre-chemotherapy and PET scan taken 2 weeks after initiation of therapy predicted response with accuracy of 85%. Two year survival rate was 90% in responders and 25% in non-responders using this criteria with p=0.002 (19). Uptake decrease during therapy is a continuous variable and different thresholds have been determined by other investigators. For example, Shah et al found that a 45% cutoff comparing uptake after 35 days was the best value to separate responders from nonresponders and predict outcome (20). In evaluating response to treatment for esophageal carcinoma, studies have shown marked variability (from 10-80%) in the cutoff values determined retrospectively, and it seems likely that gastric cancer may have comparable variability (21).

Disease recurrence frequently occurs locally in sites that have lost characteristic anatomic features due to surgery. In such cases early detection may allow for better salvage therapy and may be assisted with the use of PET. Glucose metabolism is typically low in scar tissue and high in recurrent tumor. CT remains central in the characterization of post surgical changes and post-treatment monitoring, however, equivocal findings can be better characterized with the added metabolic information of PET. Unfortunately, the same limitations of PET previously discussed apply in this circumstance; specifically, only certain histologies exhibit sufficient uptake necessary for useful sensitivity, and spatial resolution is limited by the current technological limitations of the modality.