Clinical triage decision vs risk scores in predicting the need for endotherapy in upper gastrointestinal bleeding.

BACKGROUND: Acute upper gastrointestinal hemorrhage (UGIH) is a common reason for hospitalization with substantial associated morbidity, mortality, and cost. Differentiation of high- and low-risk patients using established risk scoring systems has been advocated. The aim of this study was to determine whether these scoring systems are more accurate than an emergency physician's clinical decision making in predicting the need for endoscopic intervention in acute UGIH. METHODS: Patients presenting to a tertiary care medical center with acute UGIH from 2003 to 2006 were identified from the hospital database, and their clinical data were abstracted. One hundred ninety-five patients met the inclusion criteria and were included in the analysis. The clinical Rockall score and Blatchford score (BS) were calculated and compared with the clinical triage decision (intensive care unit vs non-intensive care unit admission) in predicting the need for endoscopic therapy. RESULTS: Clinical Rockall score greater than 0 and BS greater than 0 were sensitive predictors of the need for endoscopic therapy (95% and 100%) but were poorly specific (9% and 4%), with overall accuracies of 41% and 39%. At higher score cutoffs, clinical Rockall score greater than 2 and BS greater than 5 remained sensitive (84% and 87%) and were more specific (29% and 33%), with overall accuracies of 48% and 52%. Clinical triage decision, as a surrogate for predicting the need for endoscopic therapy, was moderately sensitive (67%) and specific (75%), with an overall accuracy (73%) that exceeded both risk scores. CONCLUSIONS: The clinical use of risk scoring systems in acute UGIH may not be as good as clinical decision making by emergency physicians.

In vivo characterization of pancreatic and lymph node tissue by using EUS spectrum analysis: a validation study.

BACKGROUND: Quantitative spectral analysis of the radiofrequency (RF) signals that underlie grayscale EUS images can be used to provide additional, objective information about tissue state. OBJECTIVE: Our purpose was to validate RF spectral analysis as a method to distinguish between (1) benign and malignant lymph nodes and (2) normal pancreas, chronic pancreatitis, and pancreatic cancer. DESIGN AND SETTING: A prospective validation study of eligible patients was conducted to compare with pilot study RF data. PATIENTS: Forty-three patients underwent EUS of the esophagus, stomach, pancreas, and surrounding intra-abdominal and mediastinal lymph nodes (19 from a previous pilot study and 24 additional patients). MAIN OUTCOME MEASUREMENTS: Midband fit, slope, intercept, and correlation coefficient from a linear regression of the calibrated RF power spectra were determined. RESULTS: Discriminant analysis of mean pilot-study parameters was then performed to classify validation-study parameters. For benign versus malignant lymph nodes, midband fit and intercept (both with t test P < .058) provided classification with 67% accuracy and area under the receiver operating curve (AUC) of 0.86. For diseased versus normal pancreas, midband fit and correlation coefficient (both with analysis of variance P < .001) provided 93% accuracy and an AUC of 0.98. For pancreatic cancer versus chronic pancreatitis, the same parameters provided 77% accuracy and an AUC of 0.89. Results improved further when classification was performed with all data. LIMITATIONS: Moderate sample size and spatial averaging inherent to the technique. CONCLUSIONS: This study confirms that mean spectral parameters provide a noninvasive method to quantitatively discriminate benign and malignant lymph nodes as well as normal and diseased pancreas.

Artificial neural network as a predictive instrument in patients with acute nonvariceal upper gastrointestinal hemorrhage.

BACKGROUND & AIMS: Triage of patients with acute upper gastrointestinal hemorrhage (UGIH) has traditionally required urgent upper endoscopy. The aim of this study is to evaluate the use of artificial neural network for nonendoscopic triage. METHODS: A cohort of 387 patients was used to train (n = 194) and internally validate (n = 193) the neural network, which was then externally validated in 200 patients and compared with the clinical and complete Rockall score. Two outcome variables were assessed: major stigmata of recent hemorrhage and need for endoscopic therapy. Patient cohort data from 2 independent tertiary-care medical centers were prospectively collected. Adult patients hospitalized at both sites during the same time period with a primary diagnosis of acute nonvariceal UGIH. RESULTS: In predicting the 2 measured outcomes, sensitivity of neural network was >80%, with high negative predictive values (92-96%) in both cohorts but with lower specificity in the external cohort. Both Rockall scores had adequate sensitivity (>80%) but poor specificity (<40%) at outcome prediction. Comparing areas under receiver operating characteristic curves, the clinical Rockall score was significantly inferior to neural network in both cohorts (/= 0.78), while in the external cohort, neural network performed similarly to the complete Rockall score (>/= 0.78). CONCLUSIONS: In acute nonvariceal UGIH, artificial neural network (nonendoscopic triage) performed as well as the complete Rockall score (endoscopic triage) at predicting stigmata of recent hemorrhage and need for endoscopic therapy, even when tested in an external patient population.

Hypervascular lesions of the pancreas: Think before you act.

BACKGROUND: The "classic" CT appearance of pancreatic neuroendocrine tumors (PNETs) is a solid, hypervascular lesion; however, non-PNET diagnoses may appear similar. In addition, some PNETs have a "non-classic" appearance. METHODS: Demographic, imaging, endoscopic ultrasound-fine needle aspiration (EUS/FNA) results, and pathology data were analyzed for patients who underwent pancreatectomy for suspected or confirmed diagnosis of PNET from our institutional database. RESULTS: Forty-three patients with a hypervascular lesion on CT had pancreatectomy for a pre-operative diagnosis of PNET. Final pathology revealed PNET in 30 (70%) and non-PNET diagnoses in 13 (30%). EUS/FNA had a sensitivity of 82% for the pre-operative diagnosis of PNET in patients with "classic" CT. Of 13 non-PNET diagnoses, 7 were benign. Among a total of 41 patients with a final diagnosis of PNET, 11 (27%) had "non-classic" CT (5 hypodense solid lesions, 3 isodense solid lesions, and 3 cystic lesions). Among these patients, EUS/FNA had a sensitivity of 100% in diagnosing PNET. CONCLUSIONS: Consideration of non-PNET diagnoses is important for patients with hypervascular lesions on CT. Appropriate pre-operative evaluation will optimize treatment plans.

Characterization of pancreatic cancer and intra-abdominal lymph node malignancy using spectrum analysis of endoscopic ultrasound imaging.

This study assessed the ability of spectral analysis of endoscopic ultrasound (EUS) RF signals acquired in humans in vivo to distinguish between (1) benign and malignant intraabdominal and mediastinal lymph nodes and (2) pancreatic cancer, chronic pancreatitis, and normal pancreas. Mean midband fit, slope, intercept, and correlation coefficient from a linear regression of the calibrated RF power spectra were computed over regions of interest defined by the endoscopist. Linear discriminant analysis was then performed to develop a classification of the resulting spectral parameters. For lymph nodes, classification based on the midband fit and intercept provided 67% sensitivity, 82% specificity, and 73% accuracy for malignant vs. benign nodes. For pancreas, classification based on midband fit and correlation coefficient provided 95% sensitivity, 93% specificity, and 93% accuracy for diseased vs. normal pancreas and 85% sensitivity, 71% specificity, and 85% accuracy for pancreatic cancer vs. chronic pancreatitis. These promising results suggest that mean spectral parameters can provide a non-invasive method to quantitatively characterize pancreatic cancer and lymph malignancy in vivo.

Endoscopic Doppler US probe for the diagnosis of gastric varices (with videos).

BACKGROUND: Gastric varices may appear similar to enlarged gastric folds and submucosal neoplasms at endoscopy. A simple endoscopic method to diagnose variceal blood flow without formal EUS could be clinically useful. OBJECTIVE: To demonstrate the use of Doppler US (DOP-US) in the diagnosis of gastric varices. DESIGN: Case series. SETTING: A tertiary-care U.S. academic medical center. PATIENTS: Eight patients with findings of gastric submucosal lesions of uncertain etiology on EGD. INTERVENTIONS: EGD with DOP-US examination, with or without standard EUS. MAIN OUTCOME MEASUREMENTS: Presence or absence of audible DOP-US signal and EUS findings for gastric submucosal lesions. RESULTS: DOP-US demonstrated a reproducible continuous venous hum in 5 cases of gastric varices (confirmed by EUS in 2 cases). A sixth case of gastric varices demonstrated pulsatile flow with DOP-US (confirmed by EUS). In 1 case of a GI stromal tumor (GIST) in the stomach, no signal was heard when the lesion itself was examined by DOP-US. In a final case of Menetrier's disease, no signal was heard when the giant gastric folds were examined. CONCLUSIONS: DOP-US can help differentiate gastric varices from other gastric submucosal lesions. The use of DOP-US may obviate the need for EUS to confirm gastric varices when the EGD diagnosis is uncertain.

EUS spectrum analysis for in vivo characterization of pancreatic and lymph node tissue: a pilot study.

BACKGROUND: EUS is limited by variability in the examiner's subjective interpretation of B-scan images to differentiate among normal, inflammatory, and malignant tissue. By using information otherwise discarded by conventional EUS systems, quantitative spectral analysis of the raw radiofrequency (RF) signals underlying EUS images enables tissue to be characterized more objectively. OBJECTIVE: Our purpose was to determine the feasibility of using spectral analysis of EUS data for characterization of pancreatic tissue and lymph nodes. DESIGN AND SETTING: A pilot study of eligible patients was conducted to analyze the RF data obtained during EUS by using spectral parameters. PATIENTS: Twenty-one subjects who underwent EUS of the esophagus, stomach, pancreas, and surrounding intra-abdominal and mediastinal lymph nodes. MAIN OUTCOME MEASUREMENTS: Linear regression parameters of calibrated power spectra of the RF signals were tested to differentiate normal pancreas from chronic pancreatitis and from pancreatic cancer as well as benign from malignant-appearing lymph nodes. RESULTS: The mean intercept, slope, and midband fit of the spectra differed significantly among normal pancreas, adenocarcinoma, and chronic pancreatitis when all were compared with each other (P < .01). On direct comparison, mean midband fit for adenocarcinoma differed significantly from that for chronic pancreatitis (P < .05). For lymph nodes, mean midband fit and intercept differed significantly between benign- and malignant-appearing lymph nodes (P < .01 and P < .05, respectively). LIMITATIONS: Small sample population and spatial averaging inherent to this technique. CONCLUSIONS: Mean spectral parameters in EUS imaging can provide a noninvasive method to discriminate normal from diseased pancreas and lymph nodes.