Response

Abstract

In his letter entitled “Caveat lector,” Dr da Silveira states that there are significant discrepancies between our paper,1Hoda K.M. Rodriguez S.A. Faigel D.O. EUS-guided sampling of suspected GI stromal tumors.Gastrointest Endosc. 2009; 69: 1218-1223Abstract Full Text Full Text PDF PubMed Scopus (187) Google Scholar a recent publication by Sepe et al,2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar and an abstract by O'Neil et al,3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar which is commented on by two of the abstract's authors (Al-Haddad and Dewitt) in an editorial4Al-Haddad M. Dewitt J. EUS-guided sampling of suspected GI mesenchymal tumors: cells, cores or a combination?.Gastrointest Endosc. 2009; 69: 1224-1227Abstract Full Text Full Text PDF PubMed Scopus (8) Google Scholar regarding our article.Indeed, the primary endpoint of these 3 reports, the yield of EUS-guided FNA (EUS-FNA), is quite similar. Ours was an unselected consecutive series of 112 patients undergoing EUS-FNA evaluation of suspected GI stromal tumors (GIST). This is a population that a consultant endosonographer would be asked to evaluate, in contrast to both Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar and O'Neil et al,3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar who looked at highly selected cohorts of patients who had undergone both EUS-FNA and subsequent surgical resection. Theirs represents a small subset of relevant patients. For example, in the Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar article, the authors reviewed a pathology database of 460 resected patients with GIST to identify 37 patients (8%) who had had previous EUS-FNA. The O'Neil et al3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar study of 54 patients does not report the denominator. Nonetheless, the cytological yields (finding cells compatible with a mesenchymal tumor irrespective of immunohistochemistry) were similar among our study (83.9%), the Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar study (78.4%), and the O'Neil et al3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar study (86%). Immunohistochemical yields were also similar between our study (61.6%) and that of O'Neil et al3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar (54%-56%); Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar performed immunohistochemistry on only a small minority of their patients, which is a major limitation of their study.We acknowledge that, without an external criterion standard, we cannot calculate test characteristics such as sensitivity and specificity, but we strongly disagree that this threatens the validity of our findings, especially with regard to the primary endpoint of diagnostic yield. Furthermore, we disagree that Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar are able to calculate with any degree of confidence test properties of EUS-FNA for GIST. To make that calculation, the criterion standard must be applied independently from the test of interest, and there must be clinical doubt as to the diagnosis. Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar violate both of these principles, because, in the first place, it is likely that the EUS and FNA results were used to select patients for subsequent surgery, and, in the second place, by using a database of surgically and pathologically confirmed GIST, there can be no doubt of the diagnosis. This introduces bias that has the effect of falsely elevating the accuracy of the test being evaluated.5Jaeschke R. Guyatt G. Lijmer J. Diagnostic tests.in: Gutyatt G. Rennie D. Users' guides to the medical literature. AMA Press, Chicago2002: 121-140Google Scholar We also disagree that using statistical means to estimate test characteristics in the absence of a criterion standard is applicable to our data because we are looking at not one but several diseases (GIST, leiomyoma, neuroma, etc) for which the underlying prevalence rates are inadequately known.There does seem to be some discrepancy as to predictors of FNA yield, primarily related to tumor size and number of FNA passes. We found no associations with yield. We disagree that this is because of erroneous data analysis. Even a cursory examination of the size data in Figure 3 would show this. The univariate P values for size and FNA passes were 0.654 and 0.795, respectively; values so high that normally one would not even do multivariate statistics. Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar also found no relationship to number of passes but did report a negative association with size; that is, larger tumors (especially those >10 cm) had lower yields. The validity of their statistical analysis on these secondary endpoints is questionable, because they evaluated 19 variables in 37 patients, evaluated the size data with at least 3 tests, did not correct for multiple testing, and performed no multivariate analysis to adjust for confounders. The authors themselves state in their “Methods” section that these P values “should be taken as descriptive only.” The O'Neil et al3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar abstract, in fact, does not present any data on these predictors of yield. In the Al-Haddad and Dewitt4Al-Haddad M. Dewitt J. EUS-guided sampling of suspected GI mesenchymal tumors: cells, cores or a combination?.Gastrointest Endosc. 2009; 69: 1224-1227Abstract Full Text Full Text PDF PubMed Scopus (8) Google Scholar editorial, they state that they found positive associations among size, number of FNA passes, and yield, but they present no supporting data or statistical analysis.We suggest resolving the perceived “discrepancies” among these reports as follows: The cytological yield of EUS-FNA is 80% or more. The immunohistochemical yield exceeds 50%. The 3 reports disagree as to the importance of size, with our study showing no association with yield, the Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar study showing a negative association, and the Al-Haddad and Dewitt4Al-Haddad M. Dewitt J. EUS-guided sampling of suspected GI mesenchymal tumors: cells, cores or a combination?.Gastrointest Endosc. 2009; 69: 1224-1227Abstract Full Text Full Text PDF PubMed Scopus (8) Google Scholar study claiming a positive association. We conclude that size alone should not be used to decide on tissue sampling. The number of FNA passes to take is unclear but needs to be adequate, a determination that should be based upon the expertise of the endoscopist and, whenever possible, feedback from in-room cytology. In his letter entitled “Caveat lector,” Dr da Silveira states that there are significant discrepancies between our paper,1Hoda K.M. Rodriguez S.A. Faigel D.O. EUS-guided sampling of suspected GI stromal tumors.Gastrointest Endosc. 2009; 69: 1218-1223Abstract Full Text Full Text PDF PubMed Scopus (187) Google Scholar a recent publication by Sepe et al,2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar and an abstract by O'Neil et al,3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar which is commented on by two of the abstract's authors (Al-Haddad and Dewitt) in an editorial4Al-Haddad M. Dewitt J. EUS-guided sampling of suspected GI mesenchymal tumors: cells, cores or a combination?.Gastrointest Endosc. 2009; 69: 1224-1227Abstract Full Text Full Text PDF PubMed Scopus (8) Google Scholar regarding our article. Indeed, the primary endpoint of these 3 reports, the yield of EUS-guided FNA (EUS-FNA), is quite similar. Ours was an unselected consecutive series of 112 patients undergoing EUS-FNA evaluation of suspected GI stromal tumors (GIST). This is a population that a consultant endosonographer would be asked to evaluate, in contrast to both Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar and O'Neil et al,3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar who looked at highly selected cohorts of patients who had undergone both EUS-FNA and subsequent surgical resection. Theirs represents a small subset of relevant patients. For example, in the Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar article, the authors reviewed a pathology database of 460 resected patients with GIST to identify 37 patients (8%) who had had previous EUS-FNA. The O'Neil et al3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar study of 54 patients does not report the denominator. Nonetheless, the cytological yields (finding cells compatible with a mesenchymal tumor irrespective of immunohistochemistry) were similar among our study (83.9%), the Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar study (78.4%), and the O'Neil et al3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar study (86%). Immunohistochemical yields were also similar between our study (61.6%) and that of O'Neil et al3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar (54%-56%); Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar performed immunohistochemistry on only a small minority of their patients, which is a major limitation of their study. We acknowledge that, without an external criterion standard, we cannot calculate test characteristics such as sensitivity and specificity, but we strongly disagree that this threatens the validity of our findings, especially with regard to the primary endpoint of diagnostic yield. Furthermore, we disagree that Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar are able to calculate with any degree of confidence test properties of EUS-FNA for GIST. To make that calculation, the criterion standard must be applied independently from the test of interest, and there must be clinical doubt as to the diagnosis. Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar violate both of these principles, because, in the first place, it is likely that the EUS and FNA results were used to select patients for subsequent surgery, and, in the second place, by using a database of surgically and pathologically confirmed GIST, there can be no doubt of the diagnosis. This introduces bias that has the effect of falsely elevating the accuracy of the test being evaluated.5Jaeschke R. Guyatt G. Lijmer J. Diagnostic tests.in: Gutyatt G. Rennie D. Users' guides to the medical literature. AMA Press, Chicago2002: 121-140Google Scholar We also disagree that using statistical means to estimate test characteristics in the absence of a criterion standard is applicable to our data because we are looking at not one but several diseases (GIST, leiomyoma, neuroma, etc) for which the underlying prevalence rates are inadequately known. There does seem to be some discrepancy as to predictors of FNA yield, primarily related to tumor size and number of FNA passes. We found no associations with yield. We disagree that this is because of erroneous data analysis. Even a cursory examination of the size data in Figure 3 would show this. The univariate P values for size and FNA passes were 0.654 and 0.795, respectively; values so high that normally one would not even do multivariate statistics. Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar also found no relationship to number of passes but did report a negative association with size; that is, larger tumors (especially those >10 cm) had lower yields. The validity of their statistical analysis on these secondary endpoints is questionable, because they evaluated 19 variables in 37 patients, evaluated the size data with at least 3 tests, did not correct for multiple testing, and performed no multivariate analysis to adjust for confounders. The authors themselves state in their “Methods” section that these P values “should be taken as descriptive only.” The O'Neil et al3O‘Neil J. Al-Haddad M. Leblanc J. et al.Endoscopic ultrasound-guided fine needle aspiration of suspected mesenchymal tumors of the gastrointestinal tract: correlation with surgical pathology [abstract].Gastrointest Endosc. 2008; 67: AB207-AB208Abstract Full Text Full Text PDF Google Scholar abstract, in fact, does not present any data on these predictors of yield. In the Al-Haddad and Dewitt4Al-Haddad M. Dewitt J. EUS-guided sampling of suspected GI mesenchymal tumors: cells, cores or a combination?.Gastrointest Endosc. 2009; 69: 1224-1227Abstract Full Text Full Text PDF PubMed Scopus (8) Google Scholar editorial, they state that they found positive associations among size, number of FNA passes, and yield, but they present no supporting data or statistical analysis. We suggest resolving the perceived “discrepancies” among these reports as follows: The cytological yield of EUS-FNA is 80% or more. The immunohistochemical yield exceeds 50%. The 3 reports disagree as to the importance of size, with our study showing no association with yield, the Sepe et al2Sepe P.S. Moparty B. Pitman M.B. et al.EUS-guided FNA for the diagnosis of GI stromal cell tumors: sensitivity and cytologic yield.Gastrointest Endosc. 2009; 70: 254-261Abstract Full Text Full Text PDF PubMed Scopus (138) Google Scholar study showing a negative association, and the Al-Haddad and Dewitt4Al-Haddad M. Dewitt J. EUS-guided sampling of suspected GI mesenchymal tumors: cells, cores or a combination?.Gastrointest Endosc. 2009; 69: 1224-1227Abstract Full Text Full Text PDF PubMed Scopus (8) Google Scholar study claiming a positive association. We conclude that size alone should not be used to decide on tissue sampling. The number of FNA passes to take is unclear but needs to be adequate, a determination that should be based upon the expertise of the endoscopist and, whenever possible, feedback from in-room cytology. EUS-guided FNA for GI stromal tumors: caveat lectorGastrointestinal EndoscopyVol. 71Issue 3PreviewTo the Editor: Full-Text PDF