- Email: [email protected]
Video Capsule Endoscopy and Histology for Small-Bowel Mucosa Evaluation: A Comparison Performed by Blinded Observers
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2006;4:998 –1003
Video Capsule Endoscopy and Histology for Small-Bowel Mucosa Evaluation: A Comparison Performed by Blinded Observers FEDERICO BIAGI,* EMANUELE RONDONOTTI,‡ JONIA CAMPANELLA,* FEDERICA VILLA,‡ PAOLA ILARIA BIANCHI,* CATHERINE KLERSY,§ ROBERTO DE FRANCHIS,‡ and GINO ROBERTO CORAZZA* *First Department of Internal Medicine, §Biometry and Clinical Epidemiology, IRCCS Policlinico San Matteo, University of Pavia, Pavia, Italy; and ‡Gastroenterology and Endoscopy Unit, Department of Internal Medicine, Istituto di Ricovero e Cura a Carattere Scientifico (IRCCS) Ospedale Maggiore Policlinico, University of Milan, Milan, Italy
Background & Aims: In view of the excellent quality of the images obtained and its magnification capability, videocapsule endoscopy was proposed as a promising tool to evaluate the degree of duodenal villous atrophy. We studied whether the capsule can discriminate different degrees of mucosal damage caused by different conditions; we also evaluated interobserver and intraobserver variability in the assessment of villous atrophy with the capsule. Methods: Thirty-two patients underwent both gastroscopy with multiple duodenal biopsies and videocapsule endoscopy. Twenty-six had different forms of celiac disease with different stages of villous atrophy; 5 patients had irritable bowel syndrome and 1 had Crohn’s disease. Videocapsule findings were evaluated blindly by 3 observers. Histologic Marsh criteria and a specifically developed classification of videocapsule mucosal patterns were used to compare videocapsule findings and histology. Results: The study of the correlation between videocapsule and histologic findings showed a Kappa statistic of .45, .49, and .51 for observers 1, 2, and 3, respectively. The sensitivity was 90.5% for observer 1 and 95.2% for observers 2 and 3; the specificity was 63.6% for all observers. Conclusions: Videocapsule findings regarding the degree of intestinal mucosal atrophy show only moderate agreement with the histologic pattern; they have a very high sensitivity but a disappointing specificity. This method therefore cannot be proposed as an alternative to traditional biopsy examinations, but it suggests that a duodenal biopsy examination should be performed when an atrophic mucosal pattern is observed in patients undergoing videocapsule examination for other reasons.
ideocapsule endoscopy (VCE) is a minimally invasive method for the visual examination of the small intestine as indicated for the assessment of patients with obscure gastrointestinal bleeding,1,2 and of patients with other localized small-bowel lesions. VCE thus currently
V
is used to evaluate patients with polyposis syndromes,3,4 nonsteroidal anti-inflammatory drug injury,5 and Crohn’s disease.6 The optical system of the capsule has an 8-fold magnification capacity lens that makes the visualization obtained with VCE similar to that of dissection microscopy.7 In view of the excellent quality of the images obtained, this technique also has been proposed for evaluating patients suffering from diffuse enteropathy.8 Celiac disease (CD), a gluten-dependent chronic condition, is the most common enteropathy in the Western world.9,10 It is characterized by a variable degree of villous damage that Marsh11 classified into 5 different types. Briefly, Marsh type 0 lesion indicates normal histology. The mildest pathologic lesion, Marsh type 1, is characterized by an increased number of intraepithelial lymphocytes in an architecturally normal small-bowel mucosa; Marsh type 2 presents crypt hyperplasia but normal villi; finally, both Marsh types 3 and 4 are characterized by increasing degrees of villous atrophy. The first attempt to compare the atrophy of the celiac small intestinal mucosa that can be observed with VCE with traditional histology found that the sensitivity and specificity of VCE for CD were 100%, provided the procedure was performed by experienced staff.8 Despite these excellent results, the fact that this study was performed in only 10 celiac patients who all had a Marsh type 3 lesion should be taken into account. The aim of our study was to evaluate, in patients undergoing both VCE and a duodenal biopsy, the correlation between the degree of villous atrophy and VCE Abbreviations used in this paper: CD, celiac disease; CI, confidence interval; VCE, videocapsule endoscopy. © 2006 by the American Gastroenterological Association Institute 1542-3565/06/$32.00 doi:10.1016/j.cgh.2006.04.004
August 2006
findings; moreover, we also evaluated both the interobserver and intraobserver variations in assessing the degree of mucosal atrophy with VCE.
Patients and Methods Thirty-two patients (22 women; mean age, 45 y; range, 18 –71 y) were enrolled. Twenty-six of them were suffering from CD, 5 from irritable bowel syndrome, and 1 from Crohn’s disease (Table 1). The 26 celiac patients comprised 10 patients with untreated CD, 6 with refractory CD, 4 with ulcerative jejunoileitis complicating CD, 3 with treated CD, 2 with lymphoma complicating CD, and 1 patient with potential CD.12–14 It should be pointed out that all of the celiac patients had taken part in 2 independent European multicenter studies on the use of VCE in CD performed in Pavia and Milan.15,16 Both trials required that the patients underwent gastroscopy with duodenal biopsies and VCE; the trials had been approved by the local ethics committees. The 6 nonceliac patients underwent duodenal biopsies and VCE because of diarrhea and abdominal pain. Therefore, for the purpose of the study described in this article, we did not perform any further procedure in any of the patients. We simply re-evaluated both the histologic and the VCE findings. After informed consent was obtained, 3 or 4 duodenal biopsy specimens were obtained under sedation using an Olympus IT20 endoscope (Olympus, Shirakawa, Japan) with jumbo forceps. Biopsy specimens were formalin-fixed and paraffin-embedded. VCE had been performed within ⫾14 days of endoscopy in all patients but 5 (patients 20, 22, 27, 29, and 30 in Table 1), in whom VCE was performed later. Because none of these 5 patients changed their diet in the meantime, we believe that the delay did not affect the final outcome. VCE was performed with the Given Imaging System (M2A Capsule Endoscope; Given Imaging Ltd., Yoqneam, Israel), which comprises an ingestible capsule, a recorder, and a workstation. After an overnight fast, the patients swallowed the capsule with a glass of water. Eight hours later, the belt with the recorder and the sensor array were removed. Data were transferred from the recorder to the workstation. The patients who did not notice the passage of the capsule in the stool within 7 days had a plain abdominal radiograph performed to confirm capsule excretion. For the purpose of the present study, both the histologic and the VCE findings were re-evaluated strictly on a blind basis. The histologic slides were re-evaluated by 1 blinded observer (F.B.), the films, rendered anonymous and duplicated more than once, were analyzed together by 3 blinded observers. Observer 2 (E.R.) and observer 3 (F.V.) are highly experienced capsule endoscopists (⬎200 capsule endoscopies reviewed each); observer 1 (F.B.) previously had reviewed ⬃40 capsule endoscopies. Although it was suggested that VCE does not always visualize the descending duodenum,17 to evaluate this tract we analyzed a 10-minute video starting when the VCE passed through the pylorus.
CAPSULE ENDOSCOPY VS HISTOLOGY
999
Table 1. Clinical and Pathologic Aspects of Study Patients Patient
Sex
Age
Diagnosis
1 2 3
M F M
44 60 40
4
F
40
5
F
33
6 7
F F
40 32
8 9
F F
41 42
10 11
F F
67 61
12
F
71
13
F
43
14
F
30
15
F
40
16 17
F M
41 38
18
M
25
19
M
29
20
F
33
21
F
44
22
F
45
23 24
M M
62 18
25
F
70
26
M
36
27
F
41
28
F
70
29
F
49
30
F
70
31
M
46
32
M
41
Lymphoma Ulcerative jejunoileitis Untreated celiac disease Untreated celiac disease Untreated celiac disease Ulcerative jejunoileitis Treated celiac disease Ulcerative jejunoileitis Treated celiac disease Lymphoma Refractory celiac disease Potential celiac disease Untreated celiac disease Untreated celiac disease Treated celiac disease Ulcerative jejunoileitis Refractory celiac disease Irritable bowel syndrome Irritable bowel syndrome Irritable bowel syndrome Untreated celiac disease Refractory celiac disease Crohn’s disease Untreated celiac disease Refractory celiac disease Untreated celiac disease Irritable bowel syndrome Refractory celiac disease Irritable bowel syndrome Refractory celiac disease Untreated celiac disease Untreated celiac disease
Marsh lesion
Diet
2 0/1 3
GCD GFD for 9 y GCD
3
GCD
3
GCD
2 2
GFD for 8 y GFD for 18 mo GFD for 3 y GFD for 5 y
0/1 0/1 3 3 0/1
GFD for 7 y GFD for 16 mo GCD
3
GCD
3
GCD
0/1
0/1
GFD for 6 mo GFD for 4 y GFD for 9 mo GCD
0/1
GCD
0/1
GCD
3
GCD
3
GFD for 13 y GCD GCD
2 3
0/1 2 3
GFD for 4 y
3
GCD
0/1
GCD
2 0/1
GCD, gluten-containing diet; GFD, gluten-free diet.
3
GFD for 4 y GCD
3
GFD for 16 mo GCD
3
GCD
1000
BIAGI ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 8
To assess the diagnostic accuracy of VCE small-bowel mucosa evaluation with respect to the histologic gold standard, the worldwide-used Marsh criteria and a VCE mucosal pattern classification that we developed were used. Regarding histology, Marsh criteria were separated into 3 groups: Marsh 0/1, Marsh 2, and Marsh 3/4.11 Because we believe that VCE cannot distinguish the debatable difference between Marsh 3 a, b, and c as developed by Oberhuber et al,18 we decided not to take this further classification into account. The most severe lesion was considered when different Marsh lesions were present on the same slide. As far as the VCE images of the mucosa are concerned, they were classified arbitrarily as normal when the mucosa presented a velvety appearance and the villi were recognizable (Figure 1), atrophic when the mucosa was smooth and with reticular furrows (Figure 2), and hypotrophic in intermediate cases (Figure 3). Again, the most severe pattern was taken into account when different aspects were present in the same patient.
Statistics Figure 2. VCE atrophic mucosa is smooth and has reticular furrows.
The Kappa statistic (for the grouped scores), Lin’s concordance correlation coefficient, and Bland and Altman’s limits-of-agreements procedures (for the continuous score) were used to assess the agreement between the histologic findings and VCE patterns, and the interobserver and intraobserver agreement. To calculate the sensitivity and specificity of VCE patterns, hypotrophic and atrophic images were considered as pathologic; similarly, Marsh 2 and Marsh 3 stages were considered pathologic lesions. The 95% confidence intervals (CIs) were computed using Stata 9 (StataCorp, College Station, TX).
Table 1 includes the histologic degree of villous damage found in all patients. Tables 2, 3, and 4 show the original results, respectively, obtained by observers 1, 2, and 3 for the correlation between VCE and the histologic findings. The statistic was .45 (moderate correlation) for observer 1. The Lin correlation coefficient was .86 (95% CI, .73–
Figure 1. VCE normal mucosa presents a velvety appearance and recognizable villi.
Figure 3. VCE hypotrophic mucosa shows an intermediate pattern between normal and atrophic mucosa.
Results
August 2006
CAPSULE ENDOSCOPY VS HISTOLOGY
Table 2. Correlation Between VCE and Histologic Findings for Observer 1
1001
Table 4. Correlation Between VCE and Histologic Findings for Observer 3
Histologic patterns
Histologic patterns
VCE patterns
Marsh 3
Marsh 2
Marsh 0/1
VCE patterns
Marsh 3
Marsh 2
Marsh 0/1
Atrophic Hypotrophic Normal
12 1 2
4 2 0
0 4 7
Atrophic Hypotrophic Normal
12 2 1
3 3 0
0 4 7
NOTE. Shaded areas show the number of observations in agreement between the 2 methods.
NOTE. Shaded areas show the number of observations in agreement between the 2 methods.
.99), and the Bland and Altman difference in means was ⫺.09 (95% limits of agreement ⫺2.60 to 2.42). The Kappa statistics for observers 2 and 3 were very similar (.49 and .51, respectively). As far as sensitivity and specificity are concerned, the sensitivity was 90.5% for observer 1 (95% CI, 80.3%– 100%) and 95.2% for observers 2 and 3 (95% CI, 87.9%–100%); the specificity was 63.6% for all 3 observers (95% CI, 47.0%– 83.3%). The positive predictive value for the VCE atrophic pattern was 100% for all the observers. The negative predictive value for the VCE normal pattern was 77.8% for observer 1 and 87.5% for observers 2 and 3. Table 5 shows the values of the Kappa statistic relative to the intraobserver and the interobserver agreement for VCE findings. The former ranged from substantial to almost perfect, whereas the latter ranged from moderate to substantial. The most interesting result is the observation that, when a VCE pattern was considered to be suggestive of atrophy, it was always histologically abnormal. On the other hand, 4 Marsh 3 histologic lesions were mistaken for a VCE normal pattern. Only 1 patient was misinterpreted as such by both observers 1 and 3 (patient 30 in Table 1). The slide of this patient was reviewed. It showed a very unusual patchy lesion with almost architecturally normal villi adjacent to a flat mucosa (Figure 4). The 6 Marsh 2 patients always were considered pathologic at VCE. Finally, for all 3 observers, a VCE hypotrophic pattern could be related to any histologic aspect.
Although our intention was to evaluate the duodenal mucosa, the entire small bowel obviously had been observed. It is difficult to measure reliably the extent of the lesions along the small bowel.
Table 3. Correlation Between VCE and Histologic Findings for Observer 2
Discussion In this experiment, we studied 32 patients. Ten of them were affected by untreated CD, 12 by complicated CD, and 3 by treated CD. We opted to select these patients because it was logical to expect that they would offer us the entire spectrum of intestinal mucosa damage. In a recent article, VCE was shown to have 100% sensitivity and 100% specificity for the diagnosis of CD when compared with histology.8 In our series the sensitivity also was high (range, 90.5%–95.2%) but the specificity was very disappointing (63.6%). There are several reasons to explain the differences between the results of that study and our own. Petroniene et al8 studied 10 celiac patients who all had been selected on the basis of a Marsh type 3 lesion and 10 controls with a normal mucosa. Because 9 of the 10 celiac patients had a subtotal/total villous atrophy, we believe that Petroniene et al8 studied only the 2 far ends of the spectrum. The larger sample we studied more realistically represents the entire spectrum of villous damage (in our series, ⬍50% of patients had Marsh 3 atrophy). Again, Petroniene et al8 took into account VCE patterns with and without villi. Although it is true that VCE patterns showing normal villi (Figure 1) can be distinguished easily from Table 5. Original Results and Kappa Statistics for Intraobserver and Interobserver Agreement Results
Kappa statistics
Histologic patterns VCE patterns
Marsh 3
Marsh 2
Marsh 0/1
Atrophic Hypotrophic Normal
9 5 1
1 5 0
0 4 7
NOTE. Shaded areas show the number of observations in agreement between the 2 methods.
Intra observer agreement Observer 1 Observer 2 Observer 3 Inter observer agreement Observer 1 vs observer 2 Observer 1 vs observer 3 Observer 2 vs observer 3
18/21 ¡ ⫽ .77 20/21 ¡ ⫽ .93 20/21 ¡ ⫽ .93 21/32 ¡ ⫽ .49 26/32 ¡ ⫽ .70 25/32 ¡ ⫽ .67
1002
BIAGI ET AL
Figure 4. Duodenal biopsy specimen from patient 30 who was misdiagnosed by both observers 1 and 3. Histology showed a patchy lesion with almost architecturally normal villi adjacent to a flat mucosa.
VCE patterns showing the absence of villi (Figure 2), practice suggests that these 2 aspects are just the far ends of the same spectrum. For this reason, we believed it was necessary to include a VCE intermediate pattern (Figure 3). This last pattern was the main reason for our results. Patient 30 highlights a condition that may explain why the VCE missed the correct diagnosis. In this patient, histology showed a very unusual patchy lesion that could not be detected by VCE. Although they are uncommon, these patients do exist and a duodenal biopsy is an absolute must for them to be diagnosed. By the same token, the patients with Marsh 2 histologic changes who were diagnosed by VCE as having hypotrophic or atrophic mucosa may represent cases in which the biopsy missed more severe damage that existed somewhere else in the duodenum. In fact, by definition, Marsh 2 lesions would not include villous changes that conceivably might be recognized by VCE. This could be one of the reasons why our study showed a moderate correlation between the VCE scores and the histologic scores (Kappa statistic, .45–.51). Thus, VCE provides only partial indications on the degree of intestinal mucosa atrophy: it recognizes the presence or absence of mucosal atrophy but does not allow a correct evaluation of its degree. Importantly, all atrophic VCE patterns actually were abnormal histologically and this probably explains the very high sensitivity we obtained. Similarly to upper gastrointestinal endoscopy,19 –21 the VCE results suggested performing a duodenal biopsy when an atrophic VCE pattern is observed in those patients undergoing VCE for other reasons. In this study, we created a 3-grade scale to evaluate the degree of mucosal atrophy with VCE. By using our scale, we observed an excellent intraobserver variability
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 8
(Kappa statistic, .77–.93). This probably means that the evaluation of mucosal atrophy using our 3-grade scale is an easy and reliable method. On the other hand, we observed a relatively higher interobserver variability (Kappa statistic, .49 –.70). This could mean that the correct evaluation of villous atrophy is a difficult task needing experience on VCE or that our proposed 3-grade scale needs extensive validation and further standardization. Four main conclusions can be drawn from our study. We confirmed a high sensitivity of VCE in identifying mucosal atrophy in patients with a complete spectrum of mucosal atrophy. However, a correct evaluation of the degree of mucosal atrophy does not seem possible in all cases. Nevertheless, the observation of an atrophic VCE pattern is strongly suggestive for histologic lesions. On the other hand, a duodenal biopsy and celiac antibodies always are necessary to achieve a proper diagnosis and to allow a differential diagnosis between different enteropathies.22 The low specificity of VCE that we observed suggests that the observation of a normal VCE pattern cannot exclude villous atrophy. It recently was suggested that patients with positive celiac serology who are unable or unwilling to undergo an endoscopic biopsy for diagnosis of CD may benefit from VCE.23 Because our aim was to compare histology with VCE pattern, we believe that this article cannot provide useful data on that point. However, in a patient who underwent VCE for other reasons, and in whom a VCE atrophic pattern was seen, the detection of positive endomysial antibodies would be strong evidence in support of CD. The 100% positive predictive value of VCE atrophic pattern for histologic damage and the 100% specificity of endomysial antibodies will practically allow the diagnosis of CD. In this particular case, it could be taken into account whether a gluten-free diet could be commenced without taking a duodenal biopsy specimen. Similarly to Petroniene et al,8 our results confirm the relevance of experience. Therefore, the evaluation of small-bowel mucosa should be performed only by very experienced VCE endoscopists.
References 1. O’Loughlin C, Barkin JS. Wireless capsule endoscopy: summary. Gastrointest Endosc Clin N Am 2004;14:229 –237. 2. Pennazio M, Santucci R, Rondonotti E, et al. Outcome of patients with obscure gastrointestinal bleeding after capsule endoscopy: report of 100 consecutive cases. Gastroenterology 2004;126: 643– 653. 3. Schulmann K, Schmiegel W. Capsule endoscopy for small bowel surveillance in hereditary intestinal polyposis and non-polyposis syndromes. Gastrointest Endosc Clin N Am 2004;14:149 –158. 4. Parsi MA, Burke CA. Utility of capsule endoscopy in Peutz-Jeghers syndrome. Gastrointest Endosc Clin N Am 2004;14:159 –167.
August 2006
5. Chutkan R, Toubia N. Effect of nonsteroidal anti-inflammatory drugs on the gastrointestinal tract: diagnosis by wireless capsule endoscopy. Gastrointest Endosc Clin N Am 2004;14:67– 85. 6. Lo SK. Capsule endoscopy in the diagnosis and management of inflammatory bowel disease. Gastrointest Endosc Clin N Am 2004;14:179 –193. 7. Petroniene R, Dubcenco E, Baker JP, et al. Given capsule endoscopy in celiac disease. Gastrointest Endosc Clin N Am 2004;14: 115–127. 8. Petroniene R, Dubcenco E, Baker JP, et al. Given capsule endoscopy in celiac disease: evaluation of diagnostic accuracy and interobserver agreement. Am J Gastroenterol 2005;100:685– 694. 9. Trier JS. Celiac sprue and refractory sprue. In: Feldman M, Scharschmidt BF, Sleisinger MH, eds. Sleisinger & Fordtran’s gastrointestinal and liver disease. 6th ed. Philadelphia: WB Saunders, 1997:1557–1573. 10. Corazza GR, Andreani ML, Biagi F, et al. The smaller size of the “coeliac iceberg” in adults. Scand J Gastroenterol 1997;32: 917–919. 11. Marsh MN, Crowe PT. Morphology of the mucosal lesion in gluten sensitivity. Baillieres Clin Gastroenterol 1995;9:273–293. 12. Biagi F, Corazza GR. Defining gluten refractory enteropathy. Eur J Gastroenterol Hepatol 2001;13:561–565. 13. Biagi F, Lorenzini P, Corazza GR. Literature review on the clinical relationship between ulcerative jejunoileitis, coeliac disease and enteropathy associated T-cell lymphoma. Scand J Gastroenterol 2000;35:785–790. 14. Biagi F, Corazza GR. Clinical features of coeliac disease. Dig Liver Dis 2002;34:225–228. 15. Krauss NG, Cellier C, Collin P, et al. Evaluation of capsule en-
CAPSULE ENDOSCOPY VS HISTOLOGY
16.
17.
18.
19.
20.
21. 22. 23.
1003
doscopy in celiac disease patients with ongoing symptoms on gluten-free diet. A prospective, blinded European multicentre trial. Gastroenterology 2005;128(Suppl 2):A81. De Franchis R, Riccioni ME, Cave D, et al. Videocapsule endoscopy for the diagnosis of celiac disease: preliminary results from a multicenter international study. Gastroenterology 2005; 128(Suppl 2):A82. Culliford A, Daly J, Diamond B, et al. The value of wireless capsule endoscopy in patients with complicated celiac disease. Gastrointest Endosc 2005;62:55– 61. Oberhuber G, Granditsch G, Vogelsang H. The histopathology of coeliac disease: time for a standardized report scheme for pathologist. Eur J Gastroenterol Hepatol 1999;11:1185–1194. Brocchi E, Corazza GR, Caletti G, et al. Endoscopic demonstration of loss of duodenal folds in the diagnosis of celiac disease. N Engl J Med 1988;319:741–744. Brocchi E, Corazza GR, Brusco G, et al. Unsuspected celiac disease diagnosed by endoscopic visualization of duodenal bulb micronodules. Gastrointest Endosc 1996;44:610 – 611. Brocchi E, Tomassetti P, Misitano B, et al. Endoscopic markers in adult coeliac disease. Dig Liver Dis 2002;34:177–182. Katz AJ, Grand RJ. All that flattens is not “sprue.” Gastroenterology 1979;76:375–377. Cellier C, Green PH, Collin P, et al. ICCE Consensus for celiac disease. Endoscopy 2005;37:1055–1059.
Address requests for reprints to: Dr. Federico Biagi, 1st Department of Internal Medicine, IRCCS Policlinico San Matteo, 27100 Pavia, Italy. e-mail: [email protected]; fax: (39) 0382-502618.
Video Capsule Endoscopy and Histology for Small-Bowel Mucosa Evaluation: A Comparison Performed by Blinded Observers FEDERICO BIAGI,* EMANUELE RONDONOTTI,‡ JONIA CAMPANELLA,* FEDERICA VILLA,‡ PAOLA ILARIA BIANCHI,* CATHERINE KLERSY,§ ROBERTO DE FRANCHIS,‡ and GINO ROBERTO CORAZZA* *First Department of Internal Medicine, §Biometry and Clinical Epidemiology, IRCCS Policlinico San Matteo, University of Pavia, Pavia, Italy; and ‡Gastroenterology and Endoscopy Unit, Department of Internal Medicine, Istituto di Ricovero e Cura a Carattere Scientifico (IRCCS) Ospedale Maggiore Policlinico, University of Milan, Milan, Italy
Background & Aims: In view of the excellent quality of the images obtained and its magnification capability, videocapsule endoscopy was proposed as a promising tool to evaluate the degree of duodenal villous atrophy. We studied whether the capsule can discriminate different degrees of mucosal damage caused by different conditions; we also evaluated interobserver and intraobserver variability in the assessment of villous atrophy with the capsule. Methods: Thirty-two patients underwent both gastroscopy with multiple duodenal biopsies and videocapsule endoscopy. Twenty-six had different forms of celiac disease with different stages of villous atrophy; 5 patients had irritable bowel syndrome and 1 had Crohn’s disease. Videocapsule findings were evaluated blindly by 3 observers. Histologic Marsh criteria and a specifically developed classification of videocapsule mucosal patterns were used to compare videocapsule findings and histology. Results: The study of the correlation between videocapsule and histologic findings showed a Kappa statistic of .45, .49, and .51 for observers 1, 2, and 3, respectively. The sensitivity was 90.5% for observer 1 and 95.2% for observers 2 and 3; the specificity was 63.6% for all observers. Conclusions: Videocapsule findings regarding the degree of intestinal mucosal atrophy show only moderate agreement with the histologic pattern; they have a very high sensitivity but a disappointing specificity. This method therefore cannot be proposed as an alternative to traditional biopsy examinations, but it suggests that a duodenal biopsy examination should be performed when an atrophic mucosal pattern is observed in patients undergoing videocapsule examination for other reasons.
ideocapsule endoscopy (VCE) is a minimally invasive method for the visual examination of the small intestine as indicated for the assessment of patients with obscure gastrointestinal bleeding,1,2 and of patients with other localized small-bowel lesions. VCE thus currently
V
is used to evaluate patients with polyposis syndromes,3,4 nonsteroidal anti-inflammatory drug injury,5 and Crohn’s disease.6 The optical system of the capsule has an 8-fold magnification capacity lens that makes the visualization obtained with VCE similar to that of dissection microscopy.7 In view of the excellent quality of the images obtained, this technique also has been proposed for evaluating patients suffering from diffuse enteropathy.8 Celiac disease (CD), a gluten-dependent chronic condition, is the most common enteropathy in the Western world.9,10 It is characterized by a variable degree of villous damage that Marsh11 classified into 5 different types. Briefly, Marsh type 0 lesion indicates normal histology. The mildest pathologic lesion, Marsh type 1, is characterized by an increased number of intraepithelial lymphocytes in an architecturally normal small-bowel mucosa; Marsh type 2 presents crypt hyperplasia but normal villi; finally, both Marsh types 3 and 4 are characterized by increasing degrees of villous atrophy. The first attempt to compare the atrophy of the celiac small intestinal mucosa that can be observed with VCE with traditional histology found that the sensitivity and specificity of VCE for CD were 100%, provided the procedure was performed by experienced staff.8 Despite these excellent results, the fact that this study was performed in only 10 celiac patients who all had a Marsh type 3 lesion should be taken into account. The aim of our study was to evaluate, in patients undergoing both VCE and a duodenal biopsy, the correlation between the degree of villous atrophy and VCE Abbreviations used in this paper: CD, celiac disease; CI, confidence interval; VCE, videocapsule endoscopy. © 2006 by the American Gastroenterological Association Institute 1542-3565/06/$32.00 doi:10.1016/j.cgh.2006.04.004
August 2006
findings; moreover, we also evaluated both the interobserver and intraobserver variations in assessing the degree of mucosal atrophy with VCE.
Patients and Methods Thirty-two patients (22 women; mean age, 45 y; range, 18 –71 y) were enrolled. Twenty-six of them were suffering from CD, 5 from irritable bowel syndrome, and 1 from Crohn’s disease (Table 1). The 26 celiac patients comprised 10 patients with untreated CD, 6 with refractory CD, 4 with ulcerative jejunoileitis complicating CD, 3 with treated CD, 2 with lymphoma complicating CD, and 1 patient with potential CD.12–14 It should be pointed out that all of the celiac patients had taken part in 2 independent European multicenter studies on the use of VCE in CD performed in Pavia and Milan.15,16 Both trials required that the patients underwent gastroscopy with duodenal biopsies and VCE; the trials had been approved by the local ethics committees. The 6 nonceliac patients underwent duodenal biopsies and VCE because of diarrhea and abdominal pain. Therefore, for the purpose of the study described in this article, we did not perform any further procedure in any of the patients. We simply re-evaluated both the histologic and the VCE findings. After informed consent was obtained, 3 or 4 duodenal biopsy specimens were obtained under sedation using an Olympus IT20 endoscope (Olympus, Shirakawa, Japan) with jumbo forceps. Biopsy specimens were formalin-fixed and paraffin-embedded. VCE had been performed within ⫾14 days of endoscopy in all patients but 5 (patients 20, 22, 27, 29, and 30 in Table 1), in whom VCE was performed later. Because none of these 5 patients changed their diet in the meantime, we believe that the delay did not affect the final outcome. VCE was performed with the Given Imaging System (M2A Capsule Endoscope; Given Imaging Ltd., Yoqneam, Israel), which comprises an ingestible capsule, a recorder, and a workstation. After an overnight fast, the patients swallowed the capsule with a glass of water. Eight hours later, the belt with the recorder and the sensor array were removed. Data were transferred from the recorder to the workstation. The patients who did not notice the passage of the capsule in the stool within 7 days had a plain abdominal radiograph performed to confirm capsule excretion. For the purpose of the present study, both the histologic and the VCE findings were re-evaluated strictly on a blind basis. The histologic slides were re-evaluated by 1 blinded observer (F.B.), the films, rendered anonymous and duplicated more than once, were analyzed together by 3 blinded observers. Observer 2 (E.R.) and observer 3 (F.V.) are highly experienced capsule endoscopists (⬎200 capsule endoscopies reviewed each); observer 1 (F.B.) previously had reviewed ⬃40 capsule endoscopies. Although it was suggested that VCE does not always visualize the descending duodenum,17 to evaluate this tract we analyzed a 10-minute video starting when the VCE passed through the pylorus.
CAPSULE ENDOSCOPY VS HISTOLOGY
999
Table 1. Clinical and Pathologic Aspects of Study Patients Patient
Sex
Age
Diagnosis
1 2 3
M F M
44 60 40
4
F
40
5
F
33
6 7
F F
40 32
8 9
F F
41 42
10 11
F F
67 61
12
F
71
13
F
43
14
F
30
15
F
40
16 17
F M
41 38
18
M
25
19
M
29
20
F
33
21
F
44
22
F
45
23 24
M M
62 18
25
F
70
26
M
36
27
F
41
28
F
70
29
F
49
30
F
70
31
M
46
32
M
41
Lymphoma Ulcerative jejunoileitis Untreated celiac disease Untreated celiac disease Untreated celiac disease Ulcerative jejunoileitis Treated celiac disease Ulcerative jejunoileitis Treated celiac disease Lymphoma Refractory celiac disease Potential celiac disease Untreated celiac disease Untreated celiac disease Treated celiac disease Ulcerative jejunoileitis Refractory celiac disease Irritable bowel syndrome Irritable bowel syndrome Irritable bowel syndrome Untreated celiac disease Refractory celiac disease Crohn’s disease Untreated celiac disease Refractory celiac disease Untreated celiac disease Irritable bowel syndrome Refractory celiac disease Irritable bowel syndrome Refractory celiac disease Untreated celiac disease Untreated celiac disease
Marsh lesion
Diet
2 0/1 3
GCD GFD for 9 y GCD
3
GCD
3
GCD
2 2
GFD for 8 y GFD for 18 mo GFD for 3 y GFD for 5 y
0/1 0/1 3 3 0/1
GFD for 7 y GFD for 16 mo GCD
3
GCD
3
GCD
0/1
0/1
GFD for 6 mo GFD for 4 y GFD for 9 mo GCD
0/1
GCD
0/1
GCD
3
GCD
3
GFD for 13 y GCD GCD
2 3
0/1 2 3
GFD for 4 y
3
GCD
0/1
GCD
2 0/1
GCD, gluten-containing diet; GFD, gluten-free diet.
3
GFD for 4 y GCD
3
GFD for 16 mo GCD
3
GCD
1000
BIAGI ET AL
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 8
To assess the diagnostic accuracy of VCE small-bowel mucosa evaluation with respect to the histologic gold standard, the worldwide-used Marsh criteria and a VCE mucosal pattern classification that we developed were used. Regarding histology, Marsh criteria were separated into 3 groups: Marsh 0/1, Marsh 2, and Marsh 3/4.11 Because we believe that VCE cannot distinguish the debatable difference between Marsh 3 a, b, and c as developed by Oberhuber et al,18 we decided not to take this further classification into account. The most severe lesion was considered when different Marsh lesions were present on the same slide. As far as the VCE images of the mucosa are concerned, they were classified arbitrarily as normal when the mucosa presented a velvety appearance and the villi were recognizable (Figure 1), atrophic when the mucosa was smooth and with reticular furrows (Figure 2), and hypotrophic in intermediate cases (Figure 3). Again, the most severe pattern was taken into account when different aspects were present in the same patient.
Statistics Figure 2. VCE atrophic mucosa is smooth and has reticular furrows.
The Kappa statistic (for the grouped scores), Lin’s concordance correlation coefficient, and Bland and Altman’s limits-of-agreements procedures (for the continuous score) were used to assess the agreement between the histologic findings and VCE patterns, and the interobserver and intraobserver agreement. To calculate the sensitivity and specificity of VCE patterns, hypotrophic and atrophic images were considered as pathologic; similarly, Marsh 2 and Marsh 3 stages were considered pathologic lesions. The 95% confidence intervals (CIs) were computed using Stata 9 (StataCorp, College Station, TX).
Table 1 includes the histologic degree of villous damage found in all patients. Tables 2, 3, and 4 show the original results, respectively, obtained by observers 1, 2, and 3 for the correlation between VCE and the histologic findings. The statistic was .45 (moderate correlation) for observer 1. The Lin correlation coefficient was .86 (95% CI, .73–
Figure 1. VCE normal mucosa presents a velvety appearance and recognizable villi.
Figure 3. VCE hypotrophic mucosa shows an intermediate pattern between normal and atrophic mucosa.
Results
August 2006
CAPSULE ENDOSCOPY VS HISTOLOGY
Table 2. Correlation Between VCE and Histologic Findings for Observer 1
1001
Table 4. Correlation Between VCE and Histologic Findings for Observer 3
Histologic patterns
Histologic patterns
VCE patterns
Marsh 3
Marsh 2
Marsh 0/1
VCE patterns
Marsh 3
Marsh 2
Marsh 0/1
Atrophic Hypotrophic Normal
12 1 2
4 2 0
0 4 7
Atrophic Hypotrophic Normal
12 2 1
3 3 0
0 4 7
NOTE. Shaded areas show the number of observations in agreement between the 2 methods.
NOTE. Shaded areas show the number of observations in agreement between the 2 methods.
.99), and the Bland and Altman difference in means was ⫺.09 (95% limits of agreement ⫺2.60 to 2.42). The Kappa statistics for observers 2 and 3 were very similar (.49 and .51, respectively). As far as sensitivity and specificity are concerned, the sensitivity was 90.5% for observer 1 (95% CI, 80.3%– 100%) and 95.2% for observers 2 and 3 (95% CI, 87.9%–100%); the specificity was 63.6% for all 3 observers (95% CI, 47.0%– 83.3%). The positive predictive value for the VCE atrophic pattern was 100% for all the observers. The negative predictive value for the VCE normal pattern was 77.8% for observer 1 and 87.5% for observers 2 and 3. Table 5 shows the values of the Kappa statistic relative to the intraobserver and the interobserver agreement for VCE findings. The former ranged from substantial to almost perfect, whereas the latter ranged from moderate to substantial. The most interesting result is the observation that, when a VCE pattern was considered to be suggestive of atrophy, it was always histologically abnormal. On the other hand, 4 Marsh 3 histologic lesions were mistaken for a VCE normal pattern. Only 1 patient was misinterpreted as such by both observers 1 and 3 (patient 30 in Table 1). The slide of this patient was reviewed. It showed a very unusual patchy lesion with almost architecturally normal villi adjacent to a flat mucosa (Figure 4). The 6 Marsh 2 patients always were considered pathologic at VCE. Finally, for all 3 observers, a VCE hypotrophic pattern could be related to any histologic aspect.
Although our intention was to evaluate the duodenal mucosa, the entire small bowel obviously had been observed. It is difficult to measure reliably the extent of the lesions along the small bowel.
Table 3. Correlation Between VCE and Histologic Findings for Observer 2
Discussion In this experiment, we studied 32 patients. Ten of them were affected by untreated CD, 12 by complicated CD, and 3 by treated CD. We opted to select these patients because it was logical to expect that they would offer us the entire spectrum of intestinal mucosa damage. In a recent article, VCE was shown to have 100% sensitivity and 100% specificity for the diagnosis of CD when compared with histology.8 In our series the sensitivity also was high (range, 90.5%–95.2%) but the specificity was very disappointing (63.6%). There are several reasons to explain the differences between the results of that study and our own. Petroniene et al8 studied 10 celiac patients who all had been selected on the basis of a Marsh type 3 lesion and 10 controls with a normal mucosa. Because 9 of the 10 celiac patients had a subtotal/total villous atrophy, we believe that Petroniene et al8 studied only the 2 far ends of the spectrum. The larger sample we studied more realistically represents the entire spectrum of villous damage (in our series, ⬍50% of patients had Marsh 3 atrophy). Again, Petroniene et al8 took into account VCE patterns with and without villi. Although it is true that VCE patterns showing normal villi (Figure 1) can be distinguished easily from Table 5. Original Results and Kappa Statistics for Intraobserver and Interobserver Agreement Results
Kappa statistics
Histologic patterns VCE patterns
Marsh 3
Marsh 2
Marsh 0/1
Atrophic Hypotrophic Normal
9 5 1
1 5 0
0 4 7
NOTE. Shaded areas show the number of observations in agreement between the 2 methods.
Intra observer agreement Observer 1 Observer 2 Observer 3 Inter observer agreement Observer 1 vs observer 2 Observer 1 vs observer 3 Observer 2 vs observer 3
18/21 ¡ ⫽ .77 20/21 ¡ ⫽ .93 20/21 ¡ ⫽ .93 21/32 ¡ ⫽ .49 26/32 ¡ ⫽ .70 25/32 ¡ ⫽ .67
1002
BIAGI ET AL
Figure 4. Duodenal biopsy specimen from patient 30 who was misdiagnosed by both observers 1 and 3. Histology showed a patchy lesion with almost architecturally normal villi adjacent to a flat mucosa.
VCE patterns showing the absence of villi (Figure 2), practice suggests that these 2 aspects are just the far ends of the same spectrum. For this reason, we believed it was necessary to include a VCE intermediate pattern (Figure 3). This last pattern was the main reason for our results. Patient 30 highlights a condition that may explain why the VCE missed the correct diagnosis. In this patient, histology showed a very unusual patchy lesion that could not be detected by VCE. Although they are uncommon, these patients do exist and a duodenal biopsy is an absolute must for them to be diagnosed. By the same token, the patients with Marsh 2 histologic changes who were diagnosed by VCE as having hypotrophic or atrophic mucosa may represent cases in which the biopsy missed more severe damage that existed somewhere else in the duodenum. In fact, by definition, Marsh 2 lesions would not include villous changes that conceivably might be recognized by VCE. This could be one of the reasons why our study showed a moderate correlation between the VCE scores and the histologic scores (Kappa statistic, .45–.51). Thus, VCE provides only partial indications on the degree of intestinal mucosa atrophy: it recognizes the presence or absence of mucosal atrophy but does not allow a correct evaluation of its degree. Importantly, all atrophic VCE patterns actually were abnormal histologically and this probably explains the very high sensitivity we obtained. Similarly to upper gastrointestinal endoscopy,19 –21 the VCE results suggested performing a duodenal biopsy when an atrophic VCE pattern is observed in those patients undergoing VCE for other reasons. In this study, we created a 3-grade scale to evaluate the degree of mucosal atrophy with VCE. By using our scale, we observed an excellent intraobserver variability
CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 8
(Kappa statistic, .77–.93). This probably means that the evaluation of mucosal atrophy using our 3-grade scale is an easy and reliable method. On the other hand, we observed a relatively higher interobserver variability (Kappa statistic, .49 –.70). This could mean that the correct evaluation of villous atrophy is a difficult task needing experience on VCE or that our proposed 3-grade scale needs extensive validation and further standardization. Four main conclusions can be drawn from our study. We confirmed a high sensitivity of VCE in identifying mucosal atrophy in patients with a complete spectrum of mucosal atrophy. However, a correct evaluation of the degree of mucosal atrophy does not seem possible in all cases. Nevertheless, the observation of an atrophic VCE pattern is strongly suggestive for histologic lesions. On the other hand, a duodenal biopsy and celiac antibodies always are necessary to achieve a proper diagnosis and to allow a differential diagnosis between different enteropathies.22 The low specificity of VCE that we observed suggests that the observation of a normal VCE pattern cannot exclude villous atrophy. It recently was suggested that patients with positive celiac serology who are unable or unwilling to undergo an endoscopic biopsy for diagnosis of CD may benefit from VCE.23 Because our aim was to compare histology with VCE pattern, we believe that this article cannot provide useful data on that point. However, in a patient who underwent VCE for other reasons, and in whom a VCE atrophic pattern was seen, the detection of positive endomysial antibodies would be strong evidence in support of CD. The 100% positive predictive value of VCE atrophic pattern for histologic damage and the 100% specificity of endomysial antibodies will practically allow the diagnosis of CD. In this particular case, it could be taken into account whether a gluten-free diet could be commenced without taking a duodenal biopsy specimen. Similarly to Petroniene et al,8 our results confirm the relevance of experience. Therefore, the evaluation of small-bowel mucosa should be performed only by very experienced VCE endoscopists.
References 1. O’Loughlin C, Barkin JS. Wireless capsule endoscopy: summary. Gastrointest Endosc Clin N Am 2004;14:229 –237. 2. Pennazio M, Santucci R, Rondonotti E, et al. Outcome of patients with obscure gastrointestinal bleeding after capsule endoscopy: report of 100 consecutive cases. Gastroenterology 2004;126: 643– 653. 3. Schulmann K, Schmiegel W. Capsule endoscopy for small bowel surveillance in hereditary intestinal polyposis and non-polyposis syndromes. Gastrointest Endosc Clin N Am 2004;14:149 –158. 4. Parsi MA, Burke CA. Utility of capsule endoscopy in Peutz-Jeghers syndrome. Gastrointest Endosc Clin N Am 2004;14:159 –167.
August 2006
5. Chutkan R, Toubia N. Effect of nonsteroidal anti-inflammatory drugs on the gastrointestinal tract: diagnosis by wireless capsule endoscopy. Gastrointest Endosc Clin N Am 2004;14:67– 85. 6. Lo SK. Capsule endoscopy in the diagnosis and management of inflammatory bowel disease. Gastrointest Endosc Clin N Am 2004;14:179 –193. 7. Petroniene R, Dubcenco E, Baker JP, et al. Given capsule endoscopy in celiac disease. Gastrointest Endosc Clin N Am 2004;14: 115–127. 8. Petroniene R, Dubcenco E, Baker JP, et al. Given capsule endoscopy in celiac disease: evaluation of diagnostic accuracy and interobserver agreement. Am J Gastroenterol 2005;100:685– 694. 9. Trier JS. Celiac sprue and refractory sprue. In: Feldman M, Scharschmidt BF, Sleisinger MH, eds. Sleisinger & Fordtran’s gastrointestinal and liver disease. 6th ed. Philadelphia: WB Saunders, 1997:1557–1573. 10. Corazza GR, Andreani ML, Biagi F, et al. The smaller size of the “coeliac iceberg” in adults. Scand J Gastroenterol 1997;32: 917–919. 11. Marsh MN, Crowe PT. Morphology of the mucosal lesion in gluten sensitivity. Baillieres Clin Gastroenterol 1995;9:273–293. 12. Biagi F, Corazza GR. Defining gluten refractory enteropathy. Eur J Gastroenterol Hepatol 2001;13:561–565. 13. Biagi F, Lorenzini P, Corazza GR. Literature review on the clinical relationship between ulcerative jejunoileitis, coeliac disease and enteropathy associated T-cell lymphoma. Scand J Gastroenterol 2000;35:785–790. 14. Biagi F, Corazza GR. Clinical features of coeliac disease. Dig Liver Dis 2002;34:225–228. 15. Krauss NG, Cellier C, Collin P, et al. Evaluation of capsule en-
CAPSULE ENDOSCOPY VS HISTOLOGY
16.
17.
18.
19.
20.
21. 22. 23.
1003
doscopy in celiac disease patients with ongoing symptoms on gluten-free diet. A prospective, blinded European multicentre trial. Gastroenterology 2005;128(Suppl 2):A81. De Franchis R, Riccioni ME, Cave D, et al. Videocapsule endoscopy for the diagnosis of celiac disease: preliminary results from a multicenter international study. Gastroenterology 2005; 128(Suppl 2):A82. Culliford A, Daly J, Diamond B, et al. The value of wireless capsule endoscopy in patients with complicated celiac disease. Gastrointest Endosc 2005;62:55– 61. Oberhuber G, Granditsch G, Vogelsang H. The histopathology of coeliac disease: time for a standardized report scheme for pathologist. Eur J Gastroenterol Hepatol 1999;11:1185–1194. Brocchi E, Corazza GR, Caletti G, et al. Endoscopic demonstration of loss of duodenal folds in the diagnosis of celiac disease. N Engl J Med 1988;319:741–744. Brocchi E, Corazza GR, Brusco G, et al. Unsuspected celiac disease diagnosed by endoscopic visualization of duodenal bulb micronodules. Gastrointest Endosc 1996;44:610 – 611. Brocchi E, Tomassetti P, Misitano B, et al. Endoscopic markers in adult coeliac disease. Dig Liver Dis 2002;34:177–182. Katz AJ, Grand RJ. All that flattens is not “sprue.” Gastroenterology 1979;76:375–377. Cellier C, Green PH, Collin P, et al. ICCE Consensus for celiac disease. Endoscopy 2005;37:1055–1059.
Address requests for reprints to: Dr. Federico Biagi, 1st Department of Internal Medicine, IRCCS Policlinico San Matteo, 27100 Pavia, Italy. e-mail: [email protected]; fax: (39) 0382-502618.