Deep Learning of Multimodal Ultrasound: Stratifying the Response to Neoadjuvant Chemotherapy in Breast Cancer Before Treatment.

BACKGROUND: Not only should resistance to neoadjuvant chemotherapy (NAC) be considered in patients with breast cancer but also the possibility of achieving a pathologic complete response (PCR) after NAC. Our study aims to develop 2 multimodal ultrasound deep learning (DL) models to noninvasively predict resistance and PCR to NAC before treatment. METHODS: From January 2017 to July 2022, a total of 170 patients with breast cancer were prospectively enrolled. All patients underwent multimodal ultrasound examination (grayscale 2D ultrasound and ultrasound elastography) before NAC. We combined clinicopathological information to develop 2 DL models, DL_Clinical_resistance and DL_Clinical_PCR, for predicting resistance and PCR to NAC, respectively. In addition, these 2 models were combined to stratify the prediction of response to NAC. RESULTS: In the test cohort, DL_Clinical_resistance had an AUC of 0.911 (95%CI, 0.814-0.979) with a sensitivity of 0.905 (95%CI, 0.765-1.000) and an NPV of 0.882 (95%CI, 0.708-1.000). Meanwhile, DL_Clinical_PCR achieved an AUC of 0.880 (95%CI, 0.751-0.973) and sensitivity and NPV of 0.875 (95%CI, 0.688-1.000) and 0.895 (95%CI, 0.739-1.000), respectively. By combining DL_Clinical_resistance and DL_Clinical_PCR, 37.1% of patients with resistance and 25.7% of patients with PCR were successfully identified by the combined model, suggesting that these patients could benefit by an early change of treatment strategy or by implementing an organ preservation strategy after NAC. CONCLUSIONS: The proposed DL_Clinical_resistance and DL_Clinical_PCR models and combined strategy have the potential to predict resistance and PCR to NAC before treatment and allow stratified prediction of NAC response.

Prospective assessment of pancreatic ductal adenocarcinoma diagnosis from endoscopic ultrasonography images with the assistance of deep learning.

BACKGROUND: Endosonographers are highly dependent on the diagnosis of pancreatic ductal adenocarcinoma (PDAC). The objectives of this study were to develop a deep-learning radiomics (DLR) model based on endoscopic ultrasonography (EUS) images for identifying PDAC and to explore its true clinical benefit. METHODS: A retrospective data set of EUS images that included PDAC and benign lesions was used as a training cohort (N = 368 patients) to develop the DLR model, and a prospective data set was used as a test cohort (N = 123 patients) to validate the effectiveness of the DLR model. In addition, seven endosonographers performed two rounds of reader studies on the test cohort with or without DLR assistance to further assess the clinical applicability and true benefits of the DLR model. RESULTS: In the prospective test cohort, DLR exhibited an area under the receiver operating characteristic curves of 0.936 (95% confidence interval [CI], 0.889-0.976) with a sensitivity of 0.831 (95% CI, 0.746-0.913) and 0.904 (95% CI, 0.820-0.980), respectively. With DLR assistance, the overall diagnostic performance of the seven endosonographers improved: one endosonographer achieved a significant expansion of specificity (p = .035,) and another achieved a significant increase in sensitivity (p = .038). In the junior endosonographer group, the diagnostic performance with the help of the DLR was higher than or comparable to that of the senior endosonographer group without DLR assistance. CONCLUSIONS: A prospective test cohort validated that the DLR model based on EUS images effectively identified PDAC. With the assistance of this model, the gap between endosonographers at different levels of experience narrowed, and the accuracy of endosonographers expanded.

Dual-Input Transformer: An End-to-End Model for Preoperative Assessment of Pathological Complete Response to Neoadjuvant Chemotherapy in Breast Cancer Ultrasonography.

Neoadjuvant chemotherapy (NAC) is the primary method to reduce the burden of tumor and metastasis; in the treatment of breast cancer, it may provide additional opportunities for breast-conserving surgery. Preoperative assessment of pathological complete response (PCR) to NAC is important for developing individualized treatment approaches and predicting patient prognosis. Compared to magnetic resonance imaging (MRI) and mammography, ultrasonography (US) has the advantages of simplicity, flexibility, and real-time imaging. Moreover, it does not require radiation and can provide multi-time acquisition of the tumor during NAC treatment. Recently, deep learning radiomics models based on multi-time-point US images for the prediction of NAC effectiveness have been proposed. To further improve the prediction performance, we carefully designed four supporting modules for our proposed dual-input transformer (DiT): isolated tokens-to-token patch embedding module, shared position embedding, time embedding, and weighted average pooling feature representation modules. The design of each module considers the characteristics of the US images at multiple time points. We validated our model on our retrospective US dataset composed of 484 cases from two centers whose consistency is not sufficiently high. Patients were allocated to training (n = 297), validation (n = 99), and external test (n = 88) sets. The results show that our model can achieve better performance than the Siamese CNN and the standard tokens-to-token vision transformer without using multi-time-point images. The ablation study also proved the effectiveness of each module designed for DiT.

Deep learning radiomics of ultrasonography for comprehensively predicting tumor and axillary lymph node status after neoadjuvant chemotherapy in breast cancer patients: A multicenter study.

BACKGROUND: Neoadjuvant chemotherapy (NAC) can downstage tumors and axillary lymph nodes in breast cancer (BC) patients. However, tumors and axillary response to NAC are not parallel and vary among patients. This study aims to explore the feasibility of deep learning radiomics nomogram (DLRN) for independently predicting the status of tumors and lymph node metastasis (LNM) after NAC. METHODS: In total, 484 BC patients who completed NAC from two hospitals (H1: 297 patients in the training cohort and 99 patients in the validation cohort; H2: 88 patients in the test cohort) were retrospectively enrolled. The authors developed two deep learning radiomics (DLR) models for personalized prediction of the tumor pathologic complete response (PCR) to NAC (DLR-PCR) and the LNM status (DLR-LNM) after NAC based on pre-NAC and after-NAC ultrasonography images. Furthermore, they proposed two DLRNs (DLRN-PCR and DLRN-LNM) for two different tasks based on the clinical characteristics and DLR scores, which were generated from both DLR-PCR and DLR-LNM. RESULTS: In the validation and test cohorts, DLRN-PCR exhibited areas under the receiver operating characteristic curves (AUCs) of 0.903 and 0.896 with sensitivities of 91.2% and 75.0%, respectively. DLRN-LNM achieved AUCs of 0.853 and 0.863, specificities of 82.0% and 81.8%, and negative predictive values of 81.3% and 87.2% in the validation and test cohorts, respectively. The two DLRN models achieved satisfactory predictive performance based on different BC subtypes. CONCLUSIONS: The proposed DLRN models have the potential to accurately predict the tumor PCR and LNM status after NAC. PLAIN LANGUAGE SUMMARY: In this study, we proposed two deep learning radiomics nomogram models based on pre-neoadjuvant chemotherapy (NAC) and preoperative ultrasonography images for independently predicting the status of tumor and axillary lymph node (ALN) after NAC. A more comprehensive assessment of the patient's condition after NAC can be achieved by predicting the status of the tumor and ALN separately. Our model can potentially provide a noninvasive and personalized method to offer decision support for organ preservation and avoidance of excessive surgery.

Ultrasound radiomics in personalized breast management: Current status and future prospects.

Breast cancer is the most common cancer in women worldwide. Providing accurate and efficient diagnosis, risk stratification and timely adjustment of treatment strategies are essential steps in achieving precision medicine before, during and after treatment. Radiomics provides image information that cannot be recognized by the naked eye through deep mining of medical images. Several studies have shown that radiomics, as a second reader of medical images, can assist physicians not only in the detection and diagnosis of breast lesions but also in the assessment of risk stratification and prediction of treatment response. Recently, more and more studies have focused on the application of ultrasound radiomics in breast management. We summarized recent research advances in ultrasound radiomics for the diagnosis of benign and malignant breast lesions, prediction of molecular subtype, assessment of lymph node status, prediction of neoadjuvant chemotherapy response, and prediction of survival. In addition, we discuss the current challenges and future prospects of ultrasound radiomics.

Deep learning radiomics based on contrast-enhanced ultrasound images for assisted diagnosis of pancreatic ductal adenocarcinoma and chronic pancreatitis.

BACKGROUND: Accurate and non-invasive diagnosis of pancreatic ductal adenocarcinoma (PDAC) and chronic pancreatitis (CP) can avoid unnecessary puncture and surgery. This study aimed to develop a deep learning radiomics (DLR) model based on contrast-enhanced ultrasound (CEUS) images to assist radiologists in identifying PDAC and CP. METHODS: Patients with PDAC or CP were retrospectively enrolled from three hospitals. Detailed clinicopathological data were collected for each patient. Diagnoses were confirmed pathologically using biopsy or surgery in all patients. We developed an end-to-end DLR model for diagnosing PDAC and CP using CEUS images. To verify the clinical application value of the DLR model, two rounds of reader studies were performed. RESULTS: A total of 558 patients with pancreatic lesions were enrolled and were split into the training cohort (n=351), internal validation cohort (n=109), and external validation cohorts 1 (n=50) and 2 (n=48). The DLR model achieved an area under curve (AUC) of 0.986 (95% CI 0.975-0.994), 0.978 (95% CI 0.950-0.996), 0.967 (95% CI 0.917-1.000), and 0.953 (95% CI 0.877-1.000) in the training, internal validation, and external validation cohorts 1 and 2, respectively. The sensitivity and specificity of the DLR model were higher than or comparable to the diagnoses of the five radiologists in the three validation cohorts. With the aid of the DLR model, the diagnostic sensitivity of all radiologists was further improved at the expense of a small or no decrease in specificity in the three validation cohorts. CONCLUSIONS: The findings of this study suggest that our DLR model can be used as an effective tool to assist radiologists in the diagnosis of PDAC and CP.

Herbal Formula Modified Bu-Shen-Huo-Xue Decoction Attenuates Intervertebral Disc Degeneration via Regulating Inflammation and Oxidative Stress.

OBJECTIVE: This study aims to clarify the potential mechanism of modified Bu-Shen-Huo-Xue decoction (MBSHXD) in treating intervertebral disc degeneration (IDD) with methods of network pharmacology and molecular docking. METHODS: An MBSHXD and IDD-related common target gene set was established through TCMSP, UniProt, and two disease gene databases. GO and KEGG enrichment analysis and protein-protein interaction (PPI) networks were performed through the R platform and STRING to discover the potential mechanism. Molecular docking between the active ingredients and the core genes is used to calculate the binding energy. RESULTS: A total of 147 active ingredients and 79 common genes (including 10 core genes, TNF, VEGFA, IL6, MAPK3, AKT1, MAPK8, TP53, JUN, MMP9, and CXCL8) were identified. The results of GO and KEGG enrichment analysis showed that MBSHXD plays an essential role in regulating inflammation and oxidative stress. The meaningful pathways are the AGE-RAGE signaling pathway in diabetic complications, the IL-17 signaling pathway, the TNF signaling pathway, the PI3K-Akt signaling pathway, the MAPK signaling pathway, and apoptosis. In addition, the PPI network and molecular docking further demonstrated the roles that nine bioactive ingredients of MBSHXD play in IDD treatment through their interference with core target proteins. CONCLUSION: This study reveals that MBSHXD has the characteristics of a "multi-component, multi-target, and multi-pathway" in the treatment of IDD by regulating inflammation and oxidative stress, and network pharmacology may provide a feasible method to verify the molecular mechanism of MBSHXD for IDD by combining with molecular docking.

Usefulness of new shear wave elastography in early predicting the efficacy of neoadjuvant chemotherapy for patients with breast cancer: where and when to measure is optimal?

BACKGROUND: The aim of this study was to investigate the diagnosis performance of new shear wave elastography (sound touch elastography, STE) in the prediction of neoadjuvant chemotherapy (NAC) response at an early stage in breast cancer patients and to determine the optimal measurement locations around the lesion in different ranges. METHODS: One hundred and eight patients were analyzed in this prospective study from November 2018 to December 2020. All patients completed NAC treatment and underwent STE examination at three time points [the day before NAC (t0); the day before the second course (t1); the day before third course (t2)]. The stiffness of the whole lesion (G), 1-mm shell (S1) and 2-mm shell (S2) around the lesion was expressed by STE parameters. The relative changes (∆stiffness) of STE parameters after the first and second course of NAC were calculated and shown as the variables [Δ(t1) and Δ(t2)]. The diagnostic accuracy of STE was evaluated by means of receiver operating characteristic curve analysis. RESULTS: The ∆stiffness (%) including ∆Gmean(t2), ∆S1mean(t2) and ∆S2mean(t2) all showed significant differences between pathological complete response (pCR) and non-pCR groups. ∆S2mean(t2) displayed the best predictive performance for pCR (AUC = 0.842) with an ideal ∆stiffness threshold value - 26%. CONCLUSIONS: Measuring the relative changes in the stiffness of surrounding tissue or entire lesion with STE holds promise for effectively predicting the response to NAC at its early stage for breast cancer patients and ∆stiffness of shell 2 mm after the second course of NAC may be a potential prediction parameter.

Deep learning radiomics of ultrasonography can predict response to neoadjuvant chemotherapy in breast cancer at an early stage of treatment: a prospective study.

OBJECTIVES: Breast cancer (BC) is the most common cancer in women worldwide, and neoadjuvant chemotherapy (NAC) is considered the standard of treatment for most patients with BC. However, response rates to NAC vary among patients, which leads to delays in appropriate treatment and affects the prognosis for patients who ineffectively respond to NAC. This study aimed to investigate the feasibility of deep learning radiomics (DLR) in the prediction of NAC response at an early stage. METHODS: In total, 168 patients with clinicopathologically confirmed BC were enrolled in this prospective study, from March 2016 to December 2020. All patients completed NAC treatment and underwent ultrasonography (US) at three time points (before NAC, after the second course, and after the fourth course). We developed two DLR models, DLR-2 and DLR-4, for predicting responses after the second and fourth courses of NAC. Furthermore, a novel deep learning radiomics pipeline (DLRP) was proposed for stepwise prediction of response at different time points of NAC administration. RESULTS: In the validation cohort, DLR-2 achieved an AUC of 0.812 (95% CI: 0.770-0.851) with an NPV of 83.3% (95% CI: 76.5-89.6). DLR-4 achieved an AUC of 0.937 (95% CI: 0.913-0.955) with a specificity of 90.5% (95% CI: 86.3-94.2). Moreover, 19 of 21 non-response patients were successfully identified by DLRP, suggesting that they could benefit from treatment strategy adjustment at an early stage of NAC. CONCLUSIONS: The proposed DLRP strategy holds promise for effectively predicting NAC response at its early stage for BC patients. KEY POINTS: • We proposed two novel deep learning radiomics (DLR) models to predict response to neoadjuvant chemotherapy (NAC) in breast cancer (BC) patients based on US images at different NAC time points. • Combining two DLR models, a deep learning radiomics pipeline (DLRP) was proposed for stepwise prediction of response to NAC. • The DLRP may provide BC patients and physicians with an effective and feasible tool to predict response to NAC at an early stage and to determine further personalized treatment options.

Usefulness of New Shear Wave Elastography Technique for Noninvasive Assessment of Liver Fibrosis in Patients with Chronic Hepatitis B: A Prospective Multicenter Study.

PURPOSE: To explore the usefulness of liver stiffness measurements (LSMs) by sound touch elastography (STE) and sound touch quantification (STQ) in chronic hepatitis B (CHB) patients for staging fibrosis. METHODS: This prospective multicenter study recruited normal volunteers and CHB patients between May 2018 and October 2019. The volunteers underwent LSM by STE and supersonic shear imaging (SSI) or by STQ and acoustic radiation force impulse imaging (ARFI). CHB patients underwent liver biopsy and LSM by both STE/STQ. The areas under the receiver operating characteristic curves (AUCs) for staging fibrosis were calculated. RESULTS: Overall, 97 volunteers and 524 CHB patients were finally eligible for the study. The successful STE and STQ measurement rates were both 100 % in volunteers and 99.4 % in CHB patients. The intraclass correlation coefficients (ICCs) for the intra-observer stability of STE and STQ (0.94; 0.90) were similar to those of SSI and ARFI (0.95; 0.87), respectively. STE and STQ showed better accuracy than the aspartate aminotransferase-to-platelet ratio index (APRI) and fibrosis-4 index (FIB-4) (AUC: 0.87 vs 0.86 vs 0.73 vs 0.77) in staging cirrhosis. However, both STE and STQ were not superior to APRI and FIB-4 in staging significant fibrosis (AUC: 0.76 vs 0.73 vs 0.70 vs 0.71, all P-values > 0.05). CONCLUSION: STE and STQ are convenient techniques with a reliable LSM value. They have a similar diagnostic performance and are superior to serum biomarkers in staging cirrhosis in CHB patients.

Sirtuins: Potential Therapeutic Targets for Defense against Oxidative Stress in Spinal Cord Injury.

Spinal cord injury (SCI) is one of the most incapacitating neurological disorders. It involves complex pathological processes that include a primary injury and a secondary injury phase, or a delayed stage, which follows the primary injury and contributes to the aggravation of the SCI pathology. Oxidative stress, a key pathophysiological event after SCI, contributes to a cascade of inflammation, excitotoxicity, neuronal and glial apoptosis, and other processes during the secondary injury phase. In recent years, increasing evidence has demonstrated that sirtuins are protective toward the pathological process of SCI through a variety of antioxidant mechanisms. Notably, strategies that modulate the expression of sirtuins exert beneficial effects in cellular and animal models of SCI. Given the significance and novelty of sirtuins, we summarize the oxidative stress processes that occur in SCI and discuss the antioxidant effects of sirtuins in SCI. We also highlight the potential of targeting sirtuins for the treatment of SCI.

Fusion imaging-guided radiofrequency ablation for residual hepatocellular carcinoma invisible on ultrasound after transcatheter arterial chemoembolization.

PURPOSE: This study aimed to investigate the technical efficiency and therapeutic response of fusion imaging (considered as virtual navigation) between contrast-enhanced ultrasound (CEUS) and contrast-enhanced computed tomography/magnetic resonance imaging (CECT/CEMRI) for the guidance of radiofrequency ablation (RFA) in patients with residual hepatocellular carcinoma (HCC) after transcatheter arterial chemoembolization (TACE). METHODS: For this prospective study, 98 patients with residual HCC lesions after TACE treatment were enrolled between June 2017 and December 2020. All the lesions were invisible on conventional ultrasound scans. Percutaneous RFA was performed using either CEUS (CEUS group, 52 lesions) or virtual navigation (VN group, 46 lesions) guidance. The lesion display rate, disease-free survival rate, local recurrence rate, overall survival rate and complication incidence were calculated and compared. RESULTS: Fusion imaging had a significant impact on the RFA outcomes (hazard ratio, 2.629; 95% confidence interval, 1.256-5.505; p = .01). The median disease-free survival time of the VN group was significantly higher than that of the CEUS group (10.9 vs. 8.8 months; p = .007). The local recurrence rates after 3, 6 and 12 months in the VN group were significantly lower than those in the CEUS group (p = .014, .002 and .011). The minor complication rate was not significantly different between the two groups. CONCLUSIONS: CEUS-CECT/CEMRI fusion imaging for guiding RFA enables an efficient and useful therapy of inconspicuous HCC lesions after TACE. The novel solution prolongs the disease-free survival time and reduces the long-term local recurrence of residual lesions treated when using virtual-navigation (VN)-guided RFA.

Quantitative Ultrasound Elastography Methods in Focal Liver Lesions Including Hepatocellular Carcinoma: From Diagnosis to Prognosis.

ABSTRACT: The ability of ultrasound elastography to diagnose focal liver lesions and determine their prognoses including hepatocellular carcinoma (HCC) is unclear. At present, radiofrequency ablation and liver resection are the most common treatments for HCC. However, the survival rate remains disappointing because of recurrences and postoperative liver failure, necessitating the development of noninvasive approaches. There is currently no systematic definition of an elastic technique for measuring liver stiffness to predict the recurrence of HCC after radiofrequency ablation and postoperative liver failure. In this review, recent advances in ultrasound elastography for the diagnosis and prognosis of focal liver lesions are discussed including HCC.

Comparison of two-dimensional shear wave elastography, magnetic resonance elastography, and three serum markers for diagnosing fibrosis in patients with chronic hepatitis B: a meta-analysis.

BACKGROUND: Two-dimensional shear wave elastography (2D-SWE), magnetic resonance elastography (MRE), aspartate transaminase-to-platelet ratio index (APRI), fibrosis index based on 4 factors (FIB-4), and King's score have been proposed for diagnosing fibrosis. METHODS: Literature databases were searched until October 1st, 2020. The summary area under the receiver operating characteristic curve (AUROC), the summary diagnostic odds ratios, and the summary sensitivities and specificities were used to assess the performance of these noninvasive methods for staging fibrosis. RESULTS: Our final data contained 72 studies. The prevalence of significant fibrosis, advanced fibrosis, and cirrhosis was 58.3%, 36.2%, and 20.5%, respectively, in chronic hepatitis B (CHB). For 2D-SWE and MRE, the summary AUROCs were 0.89 and 0.97, 0.95 and 0.97, and 0.94 and 0.97 for significant fibrosis, advanced fibrosis, and cirrhosis, respectively. The summary AUROCs using APRI and FIB-4 for detecting significant fibrosis, advanced fibrosis, and cirrhosis were 0.76 and 0.75, 0.74 and 0.77, and 0.77 and 0.82, respectively. The summary AUROCs of King's score for detecting significant fibrosis and cirrhosis were 0.77 and 0.83, respectively. CONCLUSION: MRE and 2D-SWE may show the best diagnostic accuracy for predicting fibrosis in CHB. Among the three serum markers, King's score may be more useful for diagnosing fibrosis.

AI in spotting high-risk characteristics of medical imaging and molecular pathology.

Medical imaging provides a comprehensive perspective and rich information for disease diagnosis. Combined with artificial intelligence technology, medical imaging can be further mined for detailed pathological information. Many studies have shown that the macroscopic imaging characteristics of tumors are closely related to microscopic gene, protein and molecular changes. In order to explore the function of artificial intelligence algorithms in in-depth analysis of medical imaging information, this paper reviews the articles published in recent years from three perspectives: medical imaging analysis method, clinical applications and the development of medical imaging in the direction of pathological molecular prediction. We believe that AI-aided medical imaging analysis will be extensively contributing to precise and efficient clinical decision.

Pre-operative Detection of Liver Fibrosis in Hepatocellular Carcinoma Patients Using 2D Shear Wave Elastography: Where to Measure?

The aim of this study was to pre-operatively investigate the diagnostic performance of 2D shear wave elastography (2D-SWE) for staging liver fibrosis and inflammation in patients with hepatocellular carcinoma (HCC) who then undergo surgery and to determine the optimal locations for measurement. In total, 106 patients were enrolled in this prospective study from March 2017 to May 2018. Two-dimensional SWE was used to measure liver stiffness (LS) in each patient 0-1, 1-2 and 2-5 cm from the tumor border (groups 1, 2 and 3, respectively). Spearman's correlation was used to evaluate the relationships between LS and hepatic fibrosis and between LS and inflammation. Receiver operating characteristic curve (ROC) analysis was used to evaluate the diagnostic accuracy of 2D-SWE. The technical success rate of SWE in tissue distant from the tumor (group 3) was significantly higher than that in peri-tumoral tissue (groups 1 and 2) (p < 0.001). Moreover, the area under the ROC for diagnosing cirrhosis (F4) and severe inflammation (A3) was higher for group 3 than for groups 1 and 2. Our results suggest that 2D-SWE is a helpful approach to assessment of hepatic fibrosis in HCC patients before hepatic resection. We found that to achieve a superior success rate and preferable diagnosis accuracy for patients with HCC, LS measurement should be performed 2-5 cm from the tumor margin.

Endoscopic Ultrasonography-Guided Versus Percutaneous Drainage for the Recurrent Pancreatic Fluid Collections.

BACKGROUND Ultrasonography-guided percutaneous drainage for pancreatic fluid collections is associated with a high recurrence rate and endoscopic ultrasonography (EUS)-guided drainage is a valuable approach. Our aim was to compare the efficacy and safety of percutaneous and EUS-guided drainage for the recurrent pancreatic fluid collections. MATERIAL AND METHODS A retrospective analysis of percutaneous-guided and EUS-guided procedures for pancreatic fluid collections drainages at a single tertiary care center between February 2017 and May 2018 was performed. Treatment success, adverse events, recurrence, need for surgery, length of hospital stays, and number of follow-up computed tomography (CT) scan were assessed. RESULTS A total of 119 pancreatic fluid collections treated with initial percutaneous drainage were included in this study and 35 patients had recurrent pancreatic fluid collections. Recurrent patients were classified based on drainage method: EUS-guided drainage (18 patients) and the second percutaneous drainage (17 patients). EUS-guided drainage revealed a shorter length of hospital stays (P<0.001), less re-intervention (P=0.047), fewer number of follow-up CT scans (P=0.006) compared with the initial percutaneous drainage. Furthermore, we also compared the clinical outcomes between the EUS-guided drainage and the second percutaneous drainage for the recurrent PFC after initially failed percutaneous drainage. EUS-guided drainage showed higher clinical success (P=0.027), shorter length of hospital stays (P<0.001), less re-intervention (P=0.012), fewer number of follow-up CT scan (P<0.001) and less recurrence P=0.027) compared to the second percutaneous drainage procedure. CONCLUSIONS EUS-guided drainage is an effective and appropriate method to treat the recurrent pancreatic fluid collections after initially failed percutaneous drainage procedure, with the advantage of higher clinical success, shorter length of hospital stays, less re-intervention, fewer number of follow-up CT scan and less recurrence compared to the percutaneous drainage.