16 The justifications

for this sample size are based on rationale about feasibility, precision about the mean, and variance. 16 Median bleeding times were 41.5 seconds (IQR 27.25-67.5 seconds) for Autophagy inhibitor FNA compared with 7.5 seconds (IQR 5.5-10.25 seconds) for CB and 7.5 seconds (IQR 5.5-10 seconds) for TC biopsy specimens. Bleeding time was significantly longer for FNA compared with CB (P = .0006) and was indifferent between CB and TC biopsy specimens (P = .86) ( Fig. 3). The median scoring for artifacts was 5.5 (IQR 2-6) for FNA compared with 2 (IQR 2-2) for CB and 2 (IQR 0.5-2.75) for TC biopsy specimens. CBs showed fewer artifacts than did FNAs (P = .016) and were comparable to TC biopsy specimens (P = .53) ( Fig. 4). Retrieval of CBs with a sheath did not result in more artifacts compared with direct puncture CBs (CB-1) (cryo vs cryo + sheath 2.53: P = .16, cryo vs cryo + sheath 1.75: P = .074, cryo vs cryo + sheath 1.6: P = .27) ( Fig. 4). Transduodenal CBs displayed more artifacts than did direct puncture CBs (P = .028). Histopathologic assessability was given a median score of 1 (IQR 1-2) for FNA compared with 6 (IQR 6-6) for CB and 6 (IQR 6-6) for TC biopsy specimens. The histologic assessability of CBs (CB-1) was superior over FNAs (P < .0001) and as good as that of TC biopsy specimens (P = .98) and transduodenal CBs (P = .54)

( Fig. AZD6738 5). The use of sheaths decreased the histologic assessability in comparison with direct puncture CB (CB-1) (cryo vs cryo + sheath 2.53: P = .0088, cryo vs cryo + sheath 1.75: P = .0023, cryo vs cryo + sheath 1.6: P = .0076) ( Fig. 5). CB specimens (CB-1) were larger than FNA biopsy specimens (P those = .010) but smaller than TC biopsy specimens (P = .0011) ( Fig. 6). Smaller biopsy specimens also were obtained when CB specimens were retrieved by transduodenal puncture (P = .0005) or with sheaths (cryo vs cryo + sheath 2.53: P < .0001, cryo vs cryo + sheath 1.75: P = .0001, cryo vs cryo + sheath 1.6: P < .0001) ( Fig. 6). Sample histology images are provided in Figure 7. Handling of the CB probe with standard endoscopic equipment was performed

without technical difficulties (no increased stiffness through cooling of the probe, no abnormal friction between the probe and the channel, maneuverability was not different in comparison to a 19-gauge FNA needle based on subjective impressions of the 3 examiners). Tissue could be extracted with a single pass of the CB probe for transgastric and transduodenal EUS-CB punctures in all cases. During EUS the frozen tissue appears with a discrete hyperechogenic signal and can be discriminated from the surrounding tissue endosonographically because of its different density. This can be seen as echo enhancement in the EUS image. The echo enhancement lasts as long as freezing is activated. As soon as the freezing process is deactivated, the visible EUS effect disappears.