Reply to Dumonceau

We thank Dr. Dumonceau for his extremely insightful comments. His points are well taken. In order to answer his queries we needed to review all our data. As we will show below, this further analysis only strengthens our impression that the stylet does not improve the results of EUS-FNA. Furthermore, since our publication, at least three other groups have presented data that confirm our results. Therefore, we stand by our conclusions.

Dr Dumonceau is completely correct to point out that the nature of the pass sequence starting with an S– pass automatically excluded all cases where the pass sequence started with S– from the 46 cases retained in our analysis. We congratulate him for this, because this inconsistency was missed by our team, as well as by several internal and external reviewers. We therefore reviewed our data files. The data from the data sheets were correctly entered in to the electronic database. However, there was a fault in the way the data columns were sorted on our spreadsheet, which led our research nurse to incorrectly classify 21 of the 46 cases retained for analysis as S– first when, in fact, all 46 cases were indeed S+ first.

Dr. Dumonceau’s concern regarding potential bias against S+ passes is justified. Were there cases that were excluded where the initial S– passes were negative, but the following S+ passes were positive? We reviewed the cases that were excluded from our analysis. Essentially, these were all cases where the pass sequence started with S– (the ”S– first” sequence) or cases where the pass sequence started with S+ (”S+ first” sequence), but where there were not an equal number of S+ and S– passes (e. g. S+, S–, S– or S+, S–, S–, S+, S–, S–). In the entire study series, there were 85 cases where the gold standard diagnosis was cancer. Of these 85 cases, 30 were included in the 46 cases retained for analysis. There were therefore 55 cases that were excluded and where the gold standard diagnosis was cancer. The mix of lesions was comparable to that of the 46 cases retained for analysis. In these 55 cases, the ”S– first” sequence was used in 42 and the ”S+ first” sequence in 13. In the 13 ”S+ first cases”, S+ was better (produced positive cytology when S– did not) in 4 / 13, was worse in 3 / 13, and was equal in 6 / 13. Based on this, we feel it is very unlikely that the study protocol biased the results in favour of the S– method. On the contrary, in the ”S– first” group, the on-site pathology was positive on the first S– pass in 22 / 42 (52 %) cases, and on the second S– pass in 9 / 42 (21 %). Therefore, the ”S– first” sequence was positive on the first or second pass in 73 % of cases, which excluded them from our analysis, thus possibly biasing our results actually against the S– method.

With regard to the statistical criticisms: as is sometimes the case, these are open to interpretation, and our methods were considered valid by the different internal and external statisticians consulted.

We agree that it may be difficult to understand why our institutional review board (IRB) stipulated two S– passes for every S+ pass. At our institution we had almost completely stopped using the stylet almost a year before our study began. In retrospect, we believe that the reasoning of our IRB can perhaps be explained by the fact that actually FNA without the stylet had become the ”standard of care” for EUS-FNA at our institution. As with many others, they also appear to have missed the inconsistency introduced by the ”S– first” sequence. This may explain why they required the sequence of two S– passes for each S+ pass.

Finally, we have indeed entirely stopped using the stylet since our study was completed. Since that time, we have performed approximately 6000 EUS-FNAs without the stylet and our results before and after ceasing to use the stylet use are the same. Our FNA procedures are also much simpler and faster (unpublished observations). Despite the fact that we are convinced that the stylet is useless for solid-lesion FNA (and may even make samples bloodier), we purposely kept our overall study conclusions conservative by stating that the value of the stylet is ”questionable.” We also encouraged other groups to perform further studies. Since our paper was submitted, four studies from three different groups (including two randomized trials) have also shown no benefit with the stylet [1] [2] [3] [4].

We again commend and would like to thank Dr. Dumonceau for his very astute observations, but we maintain that, despite its flaws, our study (including the post-study analysis needed to address his concerns) sheds much doubt on the value of the stylet for EUS-FNA. This appears to have been confirmed by multiple other studies, including two well-designed randomized controlled trials. Therefore, when considering whether or not to use the stylet, we believe it is important to note that all the available data where FNA with and without the stylet has been compared formally show no benefit for the stylet. If there are no data showing a clear benefit for the stylet, and only data showing the opposite, then why use it?

References

• 1 Devicente N M, Hawes R, Hoffman B et al. The yield of endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) is not affected by leaving out the stylet.  Gastrointest Endosc. 2009;  69 AB335
• 2 Chin M W, Coss A, Mcloughlin M et al. Stylet use does not affect adequacy of specimen of pancreatic EUS FNA: a prospective, single blinded, randomized, control trial.  Gastrointest Endosc. 2010;  71 AB285
• 3 Wani S B, Gupta N, Gaddam S et al. Is the use of stylet during endoscopic ultrasound (EUS)- guided fine needle aspiration (FNA) worth the effort? A comparative study of EUS-FNA with and without a stylet.  Gastrointest Endosc. 2010;  71 AB286
• 4 Rastogi A, Wani S, Gupta N et al. A prospective, single blinded, randomized controlled trial of endoscopic ultrasound (EUS) – guided fine-needle aspiration (FNA) with and without a stylet.  Am J Gastroenterol. 2010;  105 AB1418