EUS-FNA with ROSE for pancreatic solid lesions: a single-center experience

Abstract

Background

Rapid On-Site Evaluation (ROSE) improves sample quality and diagnostic accuracy by allowing for immediate examination of collected specimens. In this paper we evaluated diagnostic value of Endoscopic Ultrasound Guided Fine Needle Aspiration (EUS-FNA) and ROSE for pancreatic space occupying lesions and the rapidity of ROSE.

Methods

We retrospectively analyzed 107 cases of pancreatic space-occupying lesions undergoing EUS-FNA performed by a single endoscopist at the Affiliated Cancer Hospital of Guizhou Medical University between December 2022 and October 2024. ROSE and cytological examination were conducted concurrently for all cases. Utilizing final clinical diagnoses as the reference standard, we evaluated the diagnostic performance—encompassing sensitivity, specificity, accuracy, along with positive and negative predictive values—of cytological examination, rapid on-site evaluation (ROSE), endosonography-guided fine-needle aspiration (EUS-FNA), and EUS-FNA + ROSE. Additionally, the reporting times for ROSE, cytology, and EUS-FNA pathological results were analyzed.

Results

Among 107 patients (62 males, 45 females; mean age 59.92 ± 10.98 years; mean lesion size 34.31 ± 28.36 mm), final diagnoses confirmed 87 malignant and 20 benign lesions. The diagnostic performance was as follows: Cytology: sensitivity 88.5%, specificity 100%, accuracy 90.7%, PPV 100%, NPV 66.7%;ROSE alone: sensitivity 87.3%, specificity 100%, accuracy 89.7%, PPV 100%, NPV 64.5%;EUS-FNA alone: sensitivity 89.6%, specificity 100%, accuracy 91.6%, PPV 100%, NPV 70.0%; EUS-FNA + ROSE: sensitivity 98.9%, specificity 100%, accuracy 99.1%, PPV 100%, NPV 95.2%; Statistical analysis revealed substantial agreement between ROSE (κ = 0.721), cytology (κ = 0.742), and EUS-FNA (κ = 0.764) with final diagnoses (all P < 0.05). The combined approach achieved near-perfect agreement (κ = 0.970, P < 0.05). Mean turnaround times differed significantly: ROSE (38.81 min) versus cytology (2748.25 min; ~1.9 days) and EUS-FNA pathology (5168.32 min; ~3.6 days). No procedure-related complications occurred.

Conclusions

ROSE compensates for limitations in puncture sampling and delayed cytology reporting by providing real-time preliminary diagnosis. When integrated into EUS-FNA procedures, the combined approach demonstrates high sensitivity (98.9%) and near-perfect diagnostic consistency (κ = 0.970). This synergy increases both puncture success rates and overall diagnostic accuracy, delivering evidence-based guidance for clinical decision-making.

Introduction

The incidence of pancreatic cancer has risen dramatically in recent years and is expected to increase further. According to the latest American Cancer Society projections [1], pancreatic cancer is expected to become the third leading cause of cancer-related deaths in the United States by 2025. Concurrently, 2022 data from China’s National Cancer Center indicates pancreatic cancer ranks as the eighth most common malignancy and the sixth leading cause of cancer mortality nationally [2]. Against the backdrop of population aging, dietary shifts, and rising psychosocial stress in China, pancreatic cancer incidence and mortality continue to escalate, with an overall 5-year survival rate of approximately 10% [3, 4]. The defining clinical features of this disease—comprising insidious initiation, a pronounced tendency for malignancy, and aggressive invasive behavior—lead to over four-fifths of cases manifesting unresectable locally advanced or metastatic conditions at initial diagnosis, rendering curative surgical intervention impossible [5]. As therapeutic strategies depend on precise pathological classification, definitive diagnosis requires multi-modal approaches: imaging (CT, MRI, PET, EUS) combined with tissue acquisition [6]. However, conventional histopathological analysis typically requires 3–5 days, prolonging hospitalization and increasing economic burdens. While intraoperative frozen sections offer rapid assessment, they demand substantial tissue specimens. Endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) has become a cornerstone for preoperative diagnosis of pancreatic space-occupying lesions. Current evidence demonstrates its diagnostic sensitivity of 85–95% and specificity of 95–98% [7, 8]. Beyond lesion staging, EUS-FNA enables procurement of cytopathological and histopathological specimens for biological characterization, establishing an evidence-based foundation for systemic therapy in patients deemed ineligible for curative resection [9, 10].Historically, specimen adequacy was assessed macroscopically through subjective parameters including tactile perception, flotation characteristics, number of passes, and fragment size [11, 12]. This approach frequently yielded suboptimal diagnostic material due to significant interobserver variability. The limitations of conventional evaluation prompted the development of Rapid On-Site Evaluation (ROSE) technology.

ROSE constitutes an intraoperative quality control procedure wherein an endoscopist or cytopathologist utilizes rapid staining techniques to microscopically evaluate puncture/biopsy specimens. This enables specimen quality classification, preliminary diagnosis, and procedural guidance [13]. Performed in operating rooms, interventional suites, or endoscopy units, ROSE involves three sequential steps: sample preparation, staining, and microscopic analysis. Preparation is optimized through macroscopic examination, smear techniques, and specimen sorting—enhancing sample adequacy and diagnostic yield. Immediate intraprocedural sample processing reduces turnaround time, diagnostic completion time, and costs while minimizing repeat procedures and invasive follow-up investigations [14]. ROSE demonstrably improves adequacy rates (by 15–30%), sensitivity (by 10–25%), and reduces required passes (by 1.3–2.1) for FNA across pancreatic, pulmonary, thyroid, hepatic, and head/neck lesions [13, 15–17]. This study assessed the diagnostic performance of EUS-FNA with ROSE for solid pancreatic lesions and the reporting timeliness of ROSE.

Materials and methods

Clinical data

This retrospective investigation enrolled 107 patients with suspected pancreatic solid lesions undergoing EUS-FNA at Guizhou Medical University Cancer Hospital from December 2022 to October 2024. Relevant studies [18–22]indicate that rapid on-site cytological examination, even when conducted by endoscopists, can enhance the diagnostic rate of pancreatic solid lesions. Our procedures were conducted by expert endoscopists (>10 years’ EUS experience), with contraindications excluded through preprocedural clinical assessment. And the endoscopy center has collaborated with other hospitals in Guizhou Province to conduct basic cytopathology interpretation training guided by the pathology department, covering smear preparation, Liu staining interpretation, and cell sufficiency criteria. To meet diagnostic requirements for cell quantity and integrity, specimens must contain at least six representative cell clusters (≥ 10 cells per cluster) without significant blood contamination or tissue necrosis. Discrepancies during on-site review are referred to the pathology department for verification. ROSE results undergo periodic pathology review and comparison with cytology and pathology findings to ensure diagnostic consistency and accuracy. Concurrent cytology, ROSE, and histopathological specimen analysis were performed for each case. Ethical approval was granted by Guizhou Medical University Affiliated Cancer Hospital’s Ethics Committee (Approval No.: FZ 2025-05-119). The study protocol strictly adhered to the CONSORT guidelines. Among 117 initially screened patients, 10 were excluded due to unmet eligibility criteria (incomplete data or coagulation disorders), resulting in 107 patients for final analysis. (Fig. 1).

Fig. 1.

Flowchart of patient selection. A total of 117 patients were assessed for eligibility; 10 patients were excluded (4 due to incomplete data, 2 refused EUS-FNA, and 4 had coagulation disorders with INR > 1.5). Finally, 107 patients were included in the analysis

Inclusion criteria, exclusion criteria and diagnostic criteria

Inclusion criteria

(i)Imaging-confirmed highly suspicious solid pancreatic space-occupying lesions; (ii) Complete clinical data records; (iii) Meeting indications for EUS-FNA per ASGE guidelines; (iv) Provision of written informed consent by legally competent patients.

Exclusion criteria

(i)Severe comorbidities precluding invasive procedures (ASA class ≥ IV); (ii) Patient refusal or incomplete data documentation; (iii) Non-standardized pathological processing of samples; (iv) Coagulopathy (INR > 1.5 or platelet count < 50 × 10^9/L); (v) Pregnancy; (vi) Indeterminate diagnosis after multidisciplinary review.

Find standard for diagnosis

For patients undergoing surgical treatment, postoperative histopathological examination serves as the definitive diagnosis. For patients not undergoing surgery (e.g., those with negative cytological or histological findings, or with surgical contraindications), the definitive diagnosis is based on comprehensive clinical reference criteria. These include histopathological or cytological confirmation of malignancy obtained from EUS-FNA specimens, along with clinical, laboratory, and imaging findings (CT/MRI, EUS, or PET-CT, and serum CA19-9) during ≥ 6 months of follow-up. Lesions were classified as malignant if cancer cells were detected or if definite progression consistent with pancreatic cancer was observed. Lesions were classified as benign if no malignant pathological evidence was present and no tumor evidence emerged during follow-up. All EUS-FNA specimens were reviewed by an experienced pathologist to standardize interpretation of indicator test results.

Instruments and research methods

Equipment

Procedures utilized an OLYMPUS GF-UCT260 convex linear echoendoscope and COOK Medical (USA) 22G fine-needle aspiration needles (3 passes per lesion).

Operative protocol

Following comprehensive preoperative assessment and informed consent, all procedures were performed under general anesthesia. Lesion characteristics—including location, dimensions, morphology, and vascular relationships—were evaluated via endoscopic ultrasound. Under continuous multiparameter vital signs monitoring, real-time EUS with Doppler guidance facilitated precise needle positioning.Each lesion underwent 30 standardized aspiration cycles under constant negative pressure (5–10 mL). Puncture specimens were processed per protocol: 2–4 smears were prepared per pass, with half subjected to rapid Liu’s staining for ROSE. The staining protocol comprised: (i) Application of Liu A solution (0.5–0.8 mL, 30 s); (ii) Addition of Liu B solution (volume = 2×Liu A); (iii) Gentle airflow mixing (1.5–2 min); (iv) Rinsing and drying.

ROSE-guided optimization & specimen handling

The standard for sufficient cell quantity in ROSE typically refers to the number and integrity of cells meeting diagnostic requirements. Generally, at least six representative cell clusters (≥ 10 cells per cluster) are required, with no significant blood contamination or tissue necrosis. Procedures concluded when ROSE confirmed diagnostic adequacy. Post-procedural specimen processing included: (i) Ethanol-fixed (95%) smears for cytopathology; (ii) Formalin-fixed (4% neutral buffered) tissue cores for histology; Prophylactic broad-spectrum antibiotics, hemostatic agents, and proton pump inhibitors were administered. Patients underwent 72-hour vital signs monitoring for complication assessment.

When ROSE shows positive for malignant tumors but negative for malignant tumor pathology in the study, the rescue plan is as follows: [1] Perform additional EUS-FNA under ROSE guidance; [2] If diagnosis remains uncertain, proceed with percutaneous biopsy under CT or ultrasound guidance; [3] Surgical resection for histopathological confirmation, if clinically feasible. Patients not undergoing invasive surgery require follow-up for at least 6 months of clinical follow-up to establish a definitive diagnosis. Importantly, in this study, no cases were diagnosed as malignant based solely on ROSE results. The final diagnosis for all malignant cases was confirmed by histological (cell block) or cytological (slide) analysis, or by consistent clinical and imaging progression during follow-up.

Statistical method

Diagnostic efficacy (sensitivity, specificity, accuracy, PPV, NPV) and consistency (kappa coefficients) of the four specimen processing methods were quantitatively assessed against the gold standard. Statistical analyses were performed using SPSS 27.0 (IBM Corp., Armonk, NY). Continuous variables are presented as mean ± standard deviation (x̄ ± s), while categorical data comparisons employed paired Wilcoxon signed-rank tests. A two-tailed P < 0.05 defined statistical significance.

Outcome measures

We assessed the diagnostic agreement of ROSE, cytology, EUS-FNA, and EUS-FNA + ROSE against the final diagnosis using kappa statistics. Diagnostic accuracy metrics (sensitivity, specificity, PPV, NPV, accuracy) were quantitatively compared. Additionally, turnaround times—defined as the interval from specimen acquisition to final report issuance—were analyzed for ROSE, cytology, and EUS-FNA assessments. Surgical resection with pathological examination is considered the gold standard for determining diagnostic accuracy. However, since most patients do not undergo surgery due to negative results or unsuitability for surgery, at least 6 months of follow-up is required.

Results

All 107 patients successfully underwent EUS-FNA with concurrent ROSE, completed under real-time imaging guidance without major complications. Specimens obtained were subjected to cytological and histological examinations. The cohort comprised 62 males and 45 females, with a mean age of 59.92 ± 10.98 years and mean lesion size of 34.31 ± 28.36 mm. According to the gold standard, confirmed 87 malignant and 20 benign lesions. Under ROSE microscopy, normal pancreatic cells are characterized by regular acinar masses, abundant cytoplasmic granules, small uniform nuclei, a clear background, and the absence of dysplastic or necrotic cells (Fig. 2). Conversely, pancreatic tumors observed under ROSE microscopy present as clusters of neoplastic epithelial cells exhibiting marked cellular enlargement, pleomorphism in size and shape, enlarged hyperchromatic nuclei, and disorganized architecture (Fig. 3).

Fig. 2.

ROSE of normal pancreatic tissue (Liu’s stain, ×40). A 45-year-old female; B 57-year-old male. Images demonstrate normal pancreatic acinar cells exhibiting uniform nuclear morphology, homogeneous chromatin distribution, and cytoplasm abundant in zymogen granules

Fig. 3.

Malignant cytological features of pancreatic tumors on rapid on-site evaluation (ROSE) (Liu’s stain, ×40). C 49-year-old male (neuroendocrine tumor); D 69-year-old male (adenocarcinoma). The micrographs reveal cohesive clusters of neoplastic cells demonstrating architectural disarray, marked cellular pleomorphism, increased nuclear-to-cytoplasmic ratios, hyperchromatic nuclei with irregular contours, and coarse chromatin distribution

Agreement analysis (κ)

Statistical analysis revealed substantial diagnostic agreement between final diagnosis and: ROSE (κ = 0.721, P < 0.05); Cytology (κ = 0.742, P < 0.05); EUS-FNA (κ = 0.764, P < 0.05). Notably, EUS-FNA + ROSE achieved near-perfect concordance (κ = 0.970, P < 0.05) (Table 1). To further illustrate the discrepancies between ROSE and final pathology and analyze the compensatory role of ROSE in false-negative cases, we have listed detailed information for selected cases with inconsistencies (Table 2). As shown in Table 2, five representative false-negative cases are presented, in which ROSE yielded negative results but the final histopathology confirmed malignancy (including adenocarcinoma and neuroendocrine tumor). The remaining six false-negative cases exhibited similar cytological and histopathological characteristics and were therefore not listed separately to avoid redundancy.

Table 1.

Diagnostic performance of different modalities against gold standard (n = 107)

Diagnostic Modality	Gold Standard Result	True Positive (TP)	False Negative (FN)	False Positive (FP)	True Negative (TN)	Kappa(95% CI)	P value
ROSE	Positive	76	11	0	20	0.721 (0.62–0.82)	< 0.05
Cytology	Positive	77	10	0	20	0.742 (0.64–0.84)	< 0.05
EUS-FNA	Positive	78	9	0	20	0.764 (0.67–0.86)	< 0.05
EUS-FNA + ROSE	Positive	86	1	0	20	0.970 (0.94-1.00)	< 0.05

Gold standard (malignant vs. benign). “Positive” indicates malignancy, while “negative” indicates benignity. and No. false positives were observed (FP = 0)

Table 2.

Details of cases with discrepancies between ROSE and final pathological diagnosis

Case No	Age/Sex	Lesion Location	ROSE Result	Final Histopathology	Type of Discordance	Probable Cause	Remarks (Additional Pass/Final Confirmation)
1	62/M	Head	Negative	NET	False Negative	Limited on-site expertise	Confirmed post-resection
2	75/F	neck	Negative	Adenocarcinoma	False Negative	Mild atypia	Confirmed post-resection
3	56/M	Head	Negative	Adenocarcinoma	False Negative	Insufficient material	Confirmed after re-sampling
4	62/F	Head	Negative	Adenocarcinoma	False Negative	Mild atypia	Confirmed post-resection
5	54/F	Body	Negative	Adenocarcinoma	False Negative	Improper needle trajectory	Confirmed after repeat pass

The table lists 5 representative false-negative cases (out of 11 total cases) that illustrate the types of discrepancies observed between ROSE and final histopathology. The remaining six cases showed similar features and were not listed separately to avoid redundancy. All false-negative cases were ultimately confirmed as malignant through surgery or additional tissue sampling. No false-positive cases were identified

Diagnostic efficacy of various diagnostic methods

among 107 pancreatic space-occupying lesions, cytological diagnosis identified 87 malignant and 20 benign cases. This method demonstrated sensitivity of 88.5%, specificity of 100%, accuracy of 90.7%, positive predictive value (PPV) of 100%, and negative predictive value (NPV) of 66.7%. Using identical validation methodology, diagnostic performance parameters—including sensitivity, specificity, PPV, NPV, and accuracy—were systematically evaluated for all modalities. Notably, the EUS-FNA + ROSE approach achieved superior sensitivity (98.9%) and accuracy (99.1%) (Table 3).

Table 3.

Breaking efficiency analysis (%)

Methods	Sensitivity(%) (95% CI)	Specificity(%) (95% CI)	Accuracy(%) (95% CI)	PPV (%) (95% CI)	NPV (%) (95% CI)
Cytology	88.5 (79.0–94.6)	100 (83.2–100)	90.7 (83.1–95.6)	100 (94.0–100)	66.7 (47.2–82.7)
ROSE	87.3 (77.4–93.8)	100 (83.2–100)	89.7 (81.9–94.9)	100 (94.0–100)	64.5 (45.4–80.8)
EUS-FNA alone	89.6 (80.2–95.4)	100 (83.2–100)	91.6 (84.1–96.3)	100 (94.2–100)	70.0 (50.6–85.3)
EUS-FNA + ROSE	98.9 (93.9–100)	100 (83.2–100)	99.1 (95.1–100)	100 (95.8–100)	95.2 (76.2–99.9)

Differences in diagnostic efficacy between cytology and ROSE

In 107 patients, the sensitivity of cytology was 88.5% and the accuracy was 90.7%. The sensitivity of ROSE was 87.3% and the accuracy was 89.7% (Table 4).

Table 4.

Cytology vs. ROSE

ROSE	cytology		Total
	Positive	Negative
Positive	69	7	76
Negative	8	23	31
Total	77	30	107

Comparison of reporting time among ROSE, cytology and EUS-FNA

Following specimen collection, the reporting times were calculated from the first specimen to the final report. The reporting time for ROSE ranged from 7 to 78 min, with a mean of 38.81 min. In contrast, cytology reporting times ranged from 408 to 13,034 min (mean: 2,748.25 min; approximately 1.9 days), and EUS-FNA reporting times ranged from 1,245 to 14,289 min (mean: 5,168.32 min; approximately 3.6 days). The differences in reporting times among the three methods were statistically significant (P < 0.05) (Table 5).

Table 5.

Comparison of ROSE with cytological and EUS-FNA pathological report time

Time	x̄ ± s (min)	t value	P value
ROSE	38.81 ± 17.95	-8.98	< 0.05
Cytology	2748.25 ± 2256.31
EUS-FNA	5168.32 ± 2611.47

Discussion

The integration of ROSE in EUS-FNA procedures has been validated by multiple prospective multicenter studies [17, 23–25], demonstrating its capacity to significantly enhance specimen adequacy and improve diagnostic accuracy for malignancies. Nevertheless, some scholars contend that ROSE does not substantially increase malignancy detection rates [26]. Notably, a study [27] reported excellent concordance between ROSE and final cytological assessment (κ = 0.84, 95% CI: 0.802–0.877). Although final interpretations exhibited marginally higher accuracy than ROSE (95.8% vs. 93.9%), this difference lacked statistical significance. Supporting these findings, Rapid staining techniques, including Liu’s method [28], significantly reduce procedural time while maintaining diagnostic accuracy comparable to conventional stains.

In this study, the diagnostic findings of cytology, ROSE, and EUS-FNA demonstrated good agreement with the final diagnosis. The Kappa value for the combined diagnostic results of ROSE and EUS-FNA was significantly higher than that for cytology alone, ROSE, or EUS-FNA. This suggests that combining EUS-FNA + ROSE can significantly improve the diagnostic accuracy for malignant tumors. The sensitivity and concordance rates of cytology and ROSE for diagnosing pancreatic diseases were 88.5% vs. 87.3% and 90.7% vs. 89.7%, respectively; however, the difference did not reach statistical significance (P = 1.000 >0.05). The specificity and positive predictive value for both methods in diagnosing pancreatic diseases were 100%. It is noteworthy that the sensitivity, concordance rate, and negative predictive value of cytological diagnosis in this study were marginally higher than those of ROSE. This slight advantage may be attributable to the concurrent use of both conventional smear and liquid-based cytology (LBC) processing for cytological specimens. Although conventional smears are simpler to perform, they are susceptible to contamination by blood and gastrointestinal mucus, potentially leading to cell clumping and obscuring tumor cells. Furthermore, manual smearing techniques can introduce operational errors causing cellular damage, which may increase the false-negative rate and consequently impact the accuracy of pathological diagnosis [29]. Conversely, LBC offers advantages in managing interfering substances and provides immediate fixation to prevent cell swelling and lysis. Nevertheless, sample preparation and processing in LBC may alter cellular morphology and spatial arrangement, potentially resulting in dispersed cancer cells and a lack of characteristic background features. These changes can complicate diagnosis and increase the risk of misdiagnosis or missed diagnosis [30]. Combining both techniques can effectively leverage their complementary strengths to enhance diagnostic accuracy, albeit at a higher cost in terms of both time and resources [31]. Additionally, a study by VAN et al. [32] reported comparable diagnostic efficacy between liquid-based cytology and rapid on-site assessment. Previous literature [17, 33] has indicated that ROSE demonstrates superior efficacy compared to conventional smear cytology in diagnosing pancreatic diseases, significantly improving diagnostic accuracy and specificity.

EUS-FNA + ROSE showed superior diagnostic performance for pancreatic space-occupying lesions compared with cytology, ROSE, or EUS-FNA achieving very high sensitivity, specificity, and overall accuracy. ROSE provided immediate preliminary results in about 39 min, far faster than cytology or pathology, and helped ensure sample adequacy, thereby reducing inconclusive results and repeat procedures. By preventing unnecessary re-biopsies, ROSE lowered procedural risks, hospitalization costs, and diagnostic delays. Its rapid turnaround and real-time quality control also translated into direct cost savings and more efficient treatment planning. Although actual costs and hospitalization data were not provided, it is clear that prospective cost-effectiveness studies must be conducted in the future to quantify its economic impact and support optimized clinical resource allocation. Besides, EUS-FNB is gaining increasing popularity worldwide. However, overall, EUS-FNA combined with ROSE remains the preferred technique for EUS-guided tissue acquisition, with ROSE being a critical factor influencing diagnostic outcomes [34]. In China, a large number of endoscopists learn EUS-FNA and FNB techniques annually, and more endoscopy centers are equipped with these facilities. Nevertheless, the number of qualified cytopathology personnel cannot meet the demand. Some centers compensate for the lack of on-site cytotechnologists or cytopathologists by having endoscopists perform ROSE [13, 23]. Several studies suggest that when FNA is combined with ROSE, there is no significant difference in diagnostic rates compared to FNA alone versus FNB [35, 36]. Larger studies are needed to determine whether EUS-FNB without ROSE could become the standard for tissue sampling in solid pancreatic masses (SPMs) [37]. Although FNB alone yields diagnostic results comparable to EUS-FNA + ROSE, FNB + ROSE is still considered to enhance diagnostic rates [16, 38, 39]. Furthermore, according to a cost-effectiveness analysis by Yen et al., the average cost of EUS-FNB alone was $45 higher than that of EUS-FNA + ROSE [35].

Among the 107 patients undergoing EUS-FNA, no serious complications occurred. Only a small number of patients experienced mild adverse events, such as fever, minor bleeding, or abdominal pain; these symptoms resolved completely with symptomatic treatment. Meta-analyses [40–42] report that the overall incidence of complications associated with EUS-FNA is low, with the majority being mild to moderate in severity. Although rare, serious complications can occur; however, prompt detection and appropriate management through follow-up and intervention are typically effective.

In Tables 1, 8 cases were classified as false negatives in EUS-FNA but as positive in ROSE. These discrepancies primarily relate to sample adequacy and differences in on-site immediate interpretation. ROSE enables immediate on-site evaluation and may detect small numbers of malignant cells in smears with limited material or poor cell preservation—conditions that may not be adequately reflected in EUS-FNA examinations. Tumor heterogeneity, blood contamination, or cell loss during processing can lead to false-negative EUS-FNA reports. Conversely, in certain cases, cytology/histology final assessment had access to more extensive material, enabling EUS-FNA to detect malignancy when ROSE remained negative. Collectively, this bidirectional pattern indicates that ROSE and EUS-FNA serve complementary roles. This variation falls within the normal range observable in real-time cytology assessment and does not represent a systemic bias. Overall diagnostic performance remained unaffected, with combined EUS-FNA and ROSE demonstrating the highest concordance with the gold standard (κ = 0.97).

This study has several limitations. Firstly, as a single-center retrospective analysis, it may limit the generalizability of the findings. Secondly, not all patients underwent surgical histopathological examination. The final diagnosis in non-surgical cases relied on composite clinical reference criteria, where some patients’ final diagnosis included positive results from this EUS-FNA. Consequently, inclusion bias exists, potentially leading to a slight overestimation of diagnostic performance. This reflects routine clinical decision-making practices but warrants cautious interpretation from a methodological perspective. Finally, the reference standard of at least 6 months of follow-up (referencing patients’ laboratory or imaging studies) represents a practical yet potentially imperfect approach. The study also had limited sample size and lacked subgroup and confounding factor analyses. Despite these limitations, this study reflects real-world clinical practice and underscores the need for multicenter, prospective studies with larger cohorts and standardized protocols to validate and extend these findings. Future multicenter, prospective studies are required to validate the robustness and transferability of these results.

Our study represents the first single-center investigation in Guizhou Province to systematically evaluate the diagnostic efficacy of EUS-FNA combined with ROSE for pancreatic solid lesions, providing baseline data and actionable quality control procedures for the region. Although the retrospective design and limited innovation of the study, its findings validate the practical value of ROSE in enhancing needle aspiration success rates and diagnostic accuracy, offering reference for clinical practice under similar medical conditions.

Conclusions

In summary, ROSE provides immediate cytological feedback during EUS-FNA procedures, effectively reducing turnaround time and ensuring sample adequacy. When integrated with EUS-FNA, ROSE significantly improves diagnostic accuracy. This combined approach offers clinicians a rapid, reliable, and scientifically grounded basis for diagnosis and treatment planning. Broader validation through multi-center studies and standardized protocols is warranted to establish ROSE as a routine component of pancreatic lesion evaluation. While lacking high innovation, multicenter validation, and economic metrics, it serves as valuable regional practice experience and baseline data, laying the groundwork for subsequent multicenter, prospective, or economic evaluation studies.

Authors’ contributions

Conceptualization, W.Z. and Q.L.; methodology, W.Z., T.D. and P.L.; writing—original draft preparation, Q.L., R.X.S. and T.T.H.; writing—review and editing, W.Z., T.D. and Q.L.; supervision, W.Z.; funding acquisition, W.Z. and P.L. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by Guizhou Provincial Science and Technology Foundation (Grant Number: gzwkj2021-128).

Data availability

The datasets generated during and/or analyzed during the current study are not publicly available due to including personal medical and life information. But the datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the principles of the Declaration of Helsinki. This single-center retrospective study was approved by the Ethics Committee of Affiliated Tumor Hospital of Guizhou Medical University (Approval No.: FZ 2025-05-119). The requirement for informed consent was waived by the Ethics Committee of Affiliated Tumor Hospital of Guizhou Medical University, due to the retrospective.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.