Touch imprint cytology: a practical step toward standardized rapid on-site evaluation Authors: Seong Ji Choi Published: 2026 Endoscopic ultrasound-guided tissue acquisition (EUS-TA) has become the cornerstone of pathological diagnosis of solid pancreatic lesions. For decades, rapid on-site evaluation (ROSE) has been used in conjunction with EUS-TA to confirm specimen adequacy, optimize the number of needle passes, and provide a preliminary diagnosis. However, its clinical value has been debated, and this debate has intensified in the era of fine-needle biopsy. Randomized trials suggest that ROSE may not significantly improve diagnostic accuracy,1,2 and the 2024 American Society for Gastrointestinal Endoscopy guidelines do not recommend its routine use.3 Nevertheless, ROSE remains integral to the workflow at many centers worldwide, and practice patterns vary considerably across institutions. This gap between evidence and practice raises a question that has received less attention: Before debating the value of ROSE, should we not ask how ROSE is being performed? In the current issue of Clinical Endoscopy, Kan et al.4 attempted to answer this question. This question deserves attention because practice has outpaced standardization. Current guidelines, including those of the Korean Society of Gastrointestinal Endoscopy, describe ROSE using direct smear preparations stained with Diff-Quik,5 and the World Health Organization framework standardizes the diagnostic terminology applied to such cytologic specimens.6 Yet neither addresses how different preparation techniques compare with one another, a question that has rarely been examined. In a retrospective, single-center study of 385 patients who underwent EUS-TA for pancreatic lesions, the authors compared touch imprint cytology (TIC), spray cytology, and instrument rinse cytology within a fixed workflow. The concordance between TIC-based ROSE and final histology reached 85.1% (κ=0.575), with an overall diagnostic accuracy of 92.7%, sensitivity of 92.3%, and specificity of 96.9% relative to the final diagnosis. Within the same procedural workflow, TIC outperformed spray and instrument rinse cytology in terms of accuracy and adequacy. Multivariate analysis identified small lesion size (<16 mm) and the use of 25-gauge needles as independent predictors of discordance. Taken together, these findings suggest that TIC is a practical, reproducible, and accurate option for on-site cytological assessments. TIC is a simple and practical approach to specimen preparation. The technique samples the surface of the aspirated core without disaggregating it, so most of the specimen remains intact for subsequent histological processing.7 Spray cytology disperses and crushes the specimen, whereas rinse cytology recovers only depleted residual material. Therefore, TIC preserves this tissue-sparing advantage, while minimizing the time and complexity of the procedure. Evidence for TIC has been demonstrated in non-pancreatic settings, such as breast cancer,8 but a comparative assessment of pancreatic EUS-TA is lacking. Head-to-head comparison of TIC with spray cytology and instrument rinse cytology within a single workflow further enhances the clinical utility of these findings. In addition, this tissue-sparing property is becoming increasingly relevant as adequate cellularity determines the feasibility of next-generation sequencing in pancreatic cancer.9 However, this interpretation has two limitations. Most importantly, the on-site read was performed only during the first pass, and the read determined whether additional passes were obtained. As a result, TIC effectively accessed the best available tissue, whereas spray and rinse cytology were limited to the remaining material. The apparent superiority of TIC over spray and rinse cytology may partly reflect this procedural advantage, rather than a true difference in diagnostic performance. What is clear from this study is that TIC functions as an excellent adequacy gatekeeper, guiding the decision to obtain additional passes. Second, the negative predictive value was only 54.4%. In a cohort in which most lesions are malignant, a negative or non-diagnostic TIC result does not exclude malignancy, and additional passes or alternative sampling strategies may be considered for clinically suspicious lesions. The discordance analysis identified the clinical conditions in which this limitation was most relevant. Lesion size and needle gauge were independently associated with concordance between TIC and final histology in multivariate analysis. Concordance reached 87.8% in lesions of at least 16 mm, but decreased to 74.7% in smaller lesions, consistent with the established relationship between lesion size and EUS-TA diagnostic yield.10 Similarly, the use of a 22-gauge needle yielded 89.1% concordance compared with 79.2% for a 25-gauge needle. In small lesions or when a 25-gauge needle is used, on-site reading warrants more caution, and a single reassuring impression should not preclude additional passes. The fact that the number of passes was inversely associated with diagnostic accuracy supports this observation; lesions that demand more sampling are those that remain difficult to diagnose. This broader implication applies beyond the preparation methods. Much of the debate surrounding ROSE has assumed that it is a uniform intervention. However, its practice varies substantially across institutions in terms of specimen preparation, staining, and operator expertise. The limited benefits of ROSE observed in recent randomized trials may partly reflect the lack of standardized specimen preparation across studies. Standardizing specimen handling with an evidence-supported method, such as TIC, could improve the overall value of ROSE, particularly in centers where it is performed routinely.7 Furthermore, the integration of TIC interpretation with artificial intelligence may reduce the dependence on on-site cytotechnologists and broaden the accessibility of high-quality ROSE. In conclusion, Kan et al.4 provided useful data to support the use of TIC as a reliable and practical specimen preparation technique for ROSE in pancreatic EUS-TA. Prospective multicenter validation is needed to confirm whether standardized TIC-based ROSE improves clinical outcomes, including diagnostic turnaround time, procedure-related costs, and avoidance of repeat procedures across different practice settings. This study provides a useful reminder that the value of ROSE depends not only on whether it is performed, but also on how it is performed. Footnotes Conflicts of Interest The author has no potential conflicts of interest. Funding None. REFERENCES