Pretreatment staging by endoscopic ultrasonography does not predict complete response to neoadjuvant chemoradiation in patients with esophageal carcinoma

Abstract

I read the excellent article by Mallery et al.1 in the September issue of Cancer with great interest. Their finding that pretreatment endoscopic ultrasonography (EUS) staging does not reliably predict a response to chemoradiation in esophageal carcinoma patients raises one very obvious question: Is their any need to perform EUS in patients with esophageal carcinoma who will be treated with multimodality therapy? Endosonographers, and I happen to be one, can make the argument that EUS permits reliable identification of patients with T1N0 cancers who might not be candidates for multimodality therapy. However, these patients are few and far-between. EUS staging of T1N0 cancers might not be that much more accurate than the combination of clinical presentation (i.e., a cancer discovered inadvertently or during Barrett surveillance) and diagnostic endoscopy. Endosonographers can also argue that EUS permits the identification of patients with celiac lymph node disease classified as distant metastases according to the TNM staging system. Furthermore, EUS-guided fine-needle aspiration allows confirmation of the diagnosis of distant lymph node involvement. However, celiac lymph node disease not identifiable with computed tomography occurs only in a minority of patients with esophageal carcinoma, and thoracic surgeons often consider patients who have cancers involving the gastroesophageal junction to have resectable disease, even when the disease has spread to the celiac lymph nodes. A priori, one assumes that early stage disease will be more likely to respond to neoadjuvant therapy. Therefore, the findings reported by Mallery et al.1 initially are somewhat surprising. One possible explanation for their findings could be that their pretreatment EUS staging was not very accurate. This is likely not the case because even a test with a low staging accuracy would be expected to demonstrate some correlation with response if early stage were predictive of response. Although EUS staging can vary depending on the skill of the endosonographer, in general it is very accurate. Also, the study investigators have considerable experience and skill in performing EUS. The overall complete response rate of 40% reported in the study is a little on the high side. One wonders if this reflects the fact that a single pathologist did not review the specimens. The identification of residual cancer in the presence of necrosis often depends on the diligence of the pathologist and the thickness of the histologic sections. However, even if some of the complete responders were reclassified as partial responders on pathologic review, it is doubtful that the results of the study would change. There is one other factor to consider when interpreting the results of their study. The authors found a complete response in 60% of tumors classified as T2 by EUS, 33% of tumors classified as T3 by EUS, and 63% of tumors classified as T4 by EUS.1 The numbers of patients in the T2 and T4 groups were low (<10 in each group), limiting the power to detect any differences. At our institution (unpublished results), we too have found a greater complete response rate to chemoradiation of cancers classified as T4 by EUS than cancers classified as T3 (40% vs. 20%). Most T2 cancer patients at our center do not receive multimodality therapy. Presumably, greater proportions of T4 cancers than T3 (or T2) cancers have distant metastases. Because cancers with distant metastases are automatically not eligible for EUS or multimodality therapy, they would not have been included in this study. Therefore, the biology of T4 cancers selected for neoadjuvant treatment may be quite different from the biology of T3 (or T2) cancers. The basic question of whether EUS should be performed in patients undergoing multimodality therapy, however, remains to be answered. Conventional EUS criteria for determining T and N classification are inaccurate following chemoradiation. We have previously demonstrated that EUS measurement of tumor size reduction following chemoradiation is correlated with pathologic response.2 Surgical resection is associated with considerable morbidity and mortality and should not be performed on patients who have a limited prognosis. For EUS to be useful, it has to provide prognostic information. Perhaps EUS assessment of neoadjuvant response by measurement of tumor size will prove to be the answer. Amitabh Chak M.D.*, * University Hospitals of Cleveland, Case Western Reserve University, Cleveland, Ohio