Should we discard the needle stylet during EUS-FNA?

I read with interest a recent article comparing endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) of identical lesions with vs. without the needle stylet [1]. The authors’ conclusion ”Use of the stylet with EUS-FNA does not increase the yield for malignancy” is unconvincing due to methodological flaws.

The yields for malignancy of EUS-FNA with and without the needle stylet (S+ and S–, respectively) were compared exclusively for a subgroup of 46 (34 %) lesions selected among 135 lesions that were sampled using both techniques. The criterion used for including a lesion in the subgroup analysis was the performance of an equal number of S+ and S– FNA passes in the lesion. According to the study design, the sequence of FNA passes was ”S–, S–, S+, S–, S–, S+” for a lesion and ”S+, S–, S–, S+, S–, S–” for the following lesion (in alternation), and attempts at EUS-FNA were halted once on-site cytological analysis (available in > 95 % of cases) was positive for malignancy.

First, it is difficult to understand how lesions sampled in the ”S–, S–, S+, S–, S–, S+” sequence could ever have an equal number of S+ and S– passes as a maximum of 6 passes were performed (the first pass was S– in 21 of the 46 lesions included in the subgroup analysis).

Second, because (i) an equal number of S+ and S– passes was required for inclusion in the subgroup analysis and (ii) the procedure was halted when on-site cytological analysis was positive for malignancy, malignant lesions with a first, S+, pass negative for malignancy at on-site cytological analysis and a second, S–, pass positive for malignancy at on-site cytological analysis were favored for inclusion in the subgroup analysis. In line with this, the mean number of passes in the 46 lesions was 2.3. This is a major selection bias. Therefore, the authors’ comment that ”our results may be taken as if there had been randomized assignment as there is no reason to suspect that there could be any systematic pattern or bias in consecutive lesions“ is subject to criticisms. These would best be removed if complete results were provided in an on-line appendix to the article, detailing in a table for each of the 135 lesions: the passes performed (number and sequence; e. g., S+, S–, S–), the final diagnosis (malignant vs. benign), the result of on-site and of definitive cytological analysis, and whether the sample was included in the subgroup analysis or not. Otherwise, one should conclude that the stylet indeed increases the yield for malignancy of EUS-FNA because similar yields were obtained with and without the stylet despite that major selection bias.

Less important study inconsistencies are as follows:

The authors recognize that multiple statistical tests were performed without the necessary correction and justify this because multiple tests were related only to unimportant aspects of the article (”the P values for the secondary analyses, involving the subgroup of 46 lesions in which an equal number of passes were performed, are given for illustration only, so all P values are presented uncorrected for multiple testing”). However, the yield for malignancy, calculated for the subgroup of 46 lesions, is the primary outcome and is the basis of the authors’ conclusion. The study was insufficiently powered for the authors to draw their conclusion (46 cases were included although the authors calculated a sample size for the yield for malignancy – the primary outcome – of 50 cases). The McNemar test, not the chi-squared test, should have been used for the comparison of cytological diagnoses for samples obtained with vs. without the needle stylet from identical lesions. Insufficient details are given to allow for recalculating the sample size.

Finally, the authors state that their institutional review board (IRB) required two S– passes to be performed for each S+ pass ”because S+ is the standard of care”. This is intriguing and subject to criticism from an ethical perspective because the study design imposed by the IRB indeed impeded the ability to draw conclusions from the study and, hence, patients were unnecessarily placed at a theoretical risk (e. g., the risk of a false-negative result in the ”S–, S–, S+, S–, S–, S+” sequence).

In conclusion, use of the needle stylet should remain standard during EUS-FNA until well-designed studies show that it is unnecessary.

References

• 1 Sahai A V, Paquin S C, Gariépy G. A prospective comparison of endoscopic ultrasound-guided fine needle aspiration results obtained in the same lesion, with and without the needle stylet.  Endoscopy. 2010;  42 900-903