| Part Ⅰ: Detachable string magnetically maneuvered capsule endoscopy for detection of esophagogastric varices in patients with cirrhosisBackground and Objective: Esophagogastric varices(EGVs)are present in approximately 50% of patients with cirrhosis,which are a major cause of morbidity and mortality due to the risk of variceal hemorrhage.International clinical guidelines strongly recommend screening and periodic surveillance for EGVs by esophagogastroduodenoscopy(EGD).However,EGD is an unpleasant invasive procedure that requires conscious sedation or general anesthesia in most cases,resulting in low compliance to the screen program in cirrhotic patients.The detachable string magnetically maneuvered capsule endoscopy(dsMCE),an ingestible capsule enclosed within the sleeve that could be maneuvered under the string in esophagus and under the magnetic force in stomach,was designed for active esophageal and gastric examination.At the same time,ds-MCE with long battery life enables further evaluation of the portal hypertensive enteropathy(PHE)in small-bowel.This study investigated the diagnostic performance of ds-MCE in detecting EGVs and portal hypertensive gastropathy(PHG)using EGD as the reference standard.Methods: In this prospective,multicenter,diagnostic accuracy study conducted at 14 medical centers,patients with cirrhosis were recruited to undergo ds-MCE first,followed by EGD within 48 hours in blinded fashion.The primary outcome was the sensitivity and specificity of ds-MCE in detecting EGVs compared with EGD.Secondary outcomes were the following: the diagnostic accuracy of ds-MCE in detection of EVs,large EVs,red signs of EVs,high-risk EVs,GVs,fundal GVs(GOV2 and IGV1)and PHG compared with the EGD;the incidence of PHE in small bowel under ds-MCE;the feasibility and duration assessment of procedures;patient satisfaction assessment and safety evaluation.Results:(1)From January 2021 through November 2022,a total of 607 cirrhotic patients enrolled in the study.Of the 607 patients,25 were dropped out for different reasons.Thus,607 patients were included in the safety evaluation,and 582 patients were included in the accuracy analysis since they had both ds-MCE and EGD procedures performed.(2)With EGD considered the gold standard,the ds-MCE had a sensitivity of 97.52%(95% confidence interval [CI],95.50 to 98.65)and a specificity of 97.75%(95% CI,94.37 to 99.12)for the detection of EGVs.The positive predictive value(PPV)and negative predictive value(NPV)were 98.99%(95% CI,97.44 to 99.61)and 94.57%(95% CI,90.29 to 97.02),respectively,and the overall accuracy was 97.59%(95%CI 96.00%,98.56%).(3)Using EGD as the reference standard,the sensitivity,specificity of ds-MCE for the presence of EVs were96.35%(95%CI,93.97 to 97.82),97.98%(95%CI,94.92 to 99.21).(4)Using EGD as the reference standard,with regard to red sign in EVs,the sensitivity and specificity of ds-MCE were 96.08%(95%CI,92.45 to 98.00),97.62%(95%CI,95.54 to 98.74).(5)Using EGD as the reference standard(varix diameter ≥5mm is defined as the large EV),the luminal circumference threshold of 25% was applied for discriminating the size of EVs using dsMCE.The sensitivity and specificity of ds-MCE for identifying the presence of large EVs were 48.22%(95%CI 42.14%,54.36%),99.39%(95%CI 97.81,99.83).(6)Based on the results of the presence of red sign and identified large EVs using 25% luminal circumference threshold,the sensitivity and specificity of ds-MCE for detection of high-risk EVs were82.78%(95%CI 77.86,86.80),98.38%(95%CI 96.27,99.31).(7)With EGD considered the gold standard,the ds-MCE had a sensitivity of 96.23%(95%CI,93.00 to 98.01)and a specificity of 97.08%(95%CI,94.72 to 98.41)for the detection of GVs.(8)Using EGD as the reference standard,the sensitivity and specificity of ds-MCE for identifying fundal GVs(GOV2 and IGV1)were 92.17%(95%CI,85.79 to 95.83)and 99.57%(95%CI,98.45 to99.88).(9)Using EGD as the reference standard,the sensitivity and specificity of ds-MCE for identifying PHG were 95.27%(95%CI,90.94 to 97.58),98.31%(95%CI,96.54 to99.18),respectively.(10)The videos of the whole small bowel were recorded in 510 patients and 72 capsules failed to reach colon before the end of the life of the capsule’s battery.Portal hypertensive enteropathy(PHE)was found in 333(65.29%)patients.(11)The median esophageal transit time was 4.74 minutes(interquartile range [IQR],3.12–7.15).The median gastric examination time was 15.78 minutes(IQR,8.57-23.70).The median gastric transit time was 1.12 hours(IQR,0.68-1.85).The median small bowel transit time was 5.30 hours(IQR,4.13-6.78).The total recording time for ds-MCE was 12.83 hours(IQR,11.66-13.63).The mean EGD procedure time was 6.11±2.48 minutes.(12)For the comparison of ds-MCE and EGD without sedation,the median satisfaction score of ds-MCE was 3(IQR,3-4),which is higher than EGD of 2(IQR,2-2)(P<0.001).In the subgroup of patients in whom EGD was performed under general anesthesia,the median satisfaction score of ds-MCE was 3(IQR,3-4),which is higher than EGD of 3(IQR,2-3)(P<0.001).(13)Overall,6(0.99%)adverse events were reported within the study.Two episodes of EV bleeding occurred during EGD examination procedure.The 2 adverse events occurring with the capsule.One adverse event mainly related to the gastric preparation procedure.One episode during follow-up.Conclusions: The ds-MCE is an accurate,safe,well tolerated method for detecting EGVs and PHG in patients with cirrhosis and enables a more comprehensive evaluation of gastrointestinal changes as PHE can be recorded.However,the 25% circumferential cut-off criterion of ds-MCE to identify large esophageal varices is of low sensitivity,and further exploration of capsule endoscopic esophageal variceal size grading criteria needs to be explored.Part Ⅱ: Optimal luminal circumference threshold for identification of large esophageal varices based on capsule endoscopy imagesBackground and Objective: The 25% circumferential cut-off criterion of ds-MCE to identify large esophageal varices is of low sensitivity.Based on the first part of this thesis,this study further explored the relationship between the proportion of the capsule picture frame circumference occupied by the largest esophageal varix and the size of the largest esophageal varix under EGD(5 mm as the cut-off value).We aimed to find out the optimal luminal circumference threshold to distinguish large esophageal varices(EVs)based on capsule endoscopy images and identify the diagnostic performance of ds-MCE in detection large EVs and high-risk EVs based on the optimal luminal circumference threshold.Methods: The enrolled 582 subjects of the first part of this thesis were divided to training cohort and external validation cohort.The Youden Index,defined as [(sensitivity+specificity)-1],was calculated to determine the best cutoff for the proportion of the capsule picture frame circumference occupied by the largest varix present derived from the training cohort.We assessed diagnostic accuracy of grading of EVs and diagnostic accuracy of identification of high-risk EVs and high-risk EGVs on the basis of the above best cutoff in the external validation cohort.The subgroup analysis results of the screening and surveillance population and the subgroup analysis results of the compensated cirrhosis and decompensated cirrhosis population were also performed.Besides,we compared the diagnostic accuracy for detecting the large EVs and high-risk EVs using different percentage of luminal circumference from CE that has been reported in previous studies in the external validation cohort.Results:(1)A total of 582 patients enrolled in this study.There were 334 patients assigned in the training cohort and 248 patients assigned in the validation cohort.(2)In the training cohort,a receiver-operating characteristic(ROC)curve was generated based on the percentage of the luminal circumference from ds-MCE predicting large(as opposed to none/small)varices from the EGD(gold standard).The optimal luminal circumference threshold based on the Youden method was 18.175% and the Youden’s J index is 0.945.(3)For internal validation based on training cohort,the luminal circumference threshold of 18%,which is close to 18.175% and in integral value,was applied for discriminating the size of EVs using ds-MCE.The performance of 18% luminal circumference threshold for identifying large EVs in the training cohort showed a sensitivity of 98.35% and a specificity of 97.61%.The performance of 18% luminal circumference threshold for identifying highrisk EVs in the training cohort showed a sensitivity of 96.84% and a specificity of 96.59%.(4)In the external validation cohort,the sensitivity,specificity of ds-MCE to distinguish large EVs from small or absent EVs using a 18% luminal circumference threshold were 96.69%,92.91%.Based on the results of the presence of red sign and identified large EVs using 18% luminal circumference threshold,the sensitivity,specificity of ds-MCE for detection of high-risk EVs in external validation cohort were 96.97%,93.97%,respectively.(5)In the subgroup analysis of compensated cirrhosis based on external validation cohort,the sensitivity,specificity of ds-MCE for detection of high-risk EVs using a 18% luminal circumference threshold were 93.33% and 98.36%.In the subgroup analysis of decompensated cirrhosis based on external validation cohort,the sensitivity,specificity of ds-MCE for detection of high-risk EVs using a 18% luminal circumference threshold were 98.04% and 89.09%.(6)In the subgroup analysis of screening population based on external validation cohort,the sensitivity,specificity of ds-MCE for detection of high-risk EVs using a 18% luminal circumference threshold were 93.33% and 96.61%.In the subgroup analysis of surveillance population based on external validation cohort,the sensitivity,specificity of ds-MCE for detection of high-risk EVs using a 18% luminal circumference threshold were 98.85% and 91.23%.Conclusions: Our data support a new circumference threshold of 18% for the identification of large EVs under capsule endoscopy.Based on the circumference threshold of 18%,ds-MCE presented with high sensitivity and specificity for diagnosis of large EVs and high-risk EVs.Thus,the ds-MCE is an accurate alternative to EGD for identifying highrisk EVs.Part Ⅲ: Incidence and predictive factors of portal hypertensive enteropathy in cirrhotic patients: a capsule endoscopy studyBackground and Objective: Portal hypertensive enteropathy(PHE)is a mucosal abnormality noted in small intestine in portal hypertensive patients and can lead to gastrointestinal bleeding and anemia.Capsule endoscopy(CE)is an ideal noninvasive modality in identifying such lesions but rarely performed.There is no large-sample clinical study exploring PHE based on the CE in Chinese patients with liver cirrhosis.We aimed to determine the prevalence of PHE and its distribution in small bowel using CE in a Chinese cirrhotic patient population and correlate its presence with clinical findings.Methods: The population included in the analysis of this study were patients enrolled in the first part of this thesis,who underwent both EGD and ds-MCE,and completed the whole small bowel examination under CE.The small-bowel findings by CE were categorized as inflammatory-like,vascular lesions and active bleeding.The relationships between the presence of small-bowel changes and patients’ clinical characteristics were evaluated.Results:(1)A total of 510 patients completed the examination of the entire small bowel.Among the 510 patients,349 were male patients,161 were female patients,331 had EVs,237 had high-risk EVs,199 had GVs,and 149 had PHG.(2)Among the 510 patients who completed the whole small intestine examination under CE,PHE was found in 333(65.29%)patients.There were 273(53.53%)patients presented with inflammatory-like lesions,with erythema in 204(40.00%),villous edema in 185(36.27%),erosion in 39(7.65%),ulcerations in 19(3.73%),polypoid lesions in 13(2.55%).There were 252(49.41%)patients presented with vascular lesions,with varices in 180(35.29%),red spot in 123(24.12%),angioectasia in 58(11.37%).Spontaneous bleeding was found in 3(0.59%).(3)In the proximal small intestine,PHE was found in 274(53.73%),with inflammatory-like lesions in 227(44.51%),vascular lesions in 122(23.92%)and spontaneous bleeding in 1(0.20%).In the middle small intestine,PHE was found in 229(44.9%),with inflammatorylike lesions in 121(23.73%),vascular lesions in 175(34.31%)and spontaneous bleeding in 1(0.20%).In the distal small intestine,PHE was found in 243(47.65%),with inflammatorylike lesions in 121(23.73%),vascular lesions in 131(25.69%)and spontaneous bleeding in 1(0.20%).The incidence of inflammatory-like lesions of the proximal small intestinal mucosa was higher than that of the middle and distal small intestines(44.51% vs.23.73% vs.25.69%;P<0.001).The incidence of proximal small bowel vascular lesions was lower than that of the middle and distal small intestines(23.92% vs.34.31% vs.37.06%;P<0.001).(4)The presence of EVs(OR,3.45;P<0.01),GVs(OR,1.81;P = 0.02),PHG(OR,1.98;P = 0.02)were independent predictors of PHE.(5)The presence of EVs(OR,3.18;P<0.01),GVs(OR,1.68;P = 0.03),PHG(OR,2.06;P<0.01)were independent predictors of inflammation-like lesions.(6)The presence of splenomegaly(OR,1.76;P =0.02)and EVs(OR,3.51;P <0.01)were independent predictors of vascular lesions.Conclusion: The ds-MCE can identify PHE.PHE were identified in more than 60% of cirrhotic patients in our study.The inflammation-like lesions have a higher incidence in the proximal small intestine,and vascular lesions have a higher incidence in the middle and distal small intestine.The presence of EVs,GVs,PHG were independent predictors of PHE. |
PDF Full Text Request