Duodenal Endoscopic Findings and Histopathologic Confirmation of Intestinal Lymphangiectasia in Dogs
Work performed at the University of Wisconsin-Madison, School of Veterinary Medicine.
No outside funding was used to support this research.
Corresponding author: Daniel S. Foy, University of Wisconsin School of Veterinary Medicine, 2015 Linden Drive, Madison, WI 53706; email: email@example.com.
The diagnosis of intestinal lymphangiectasia (IL) has been associated with characteristic duodenal mucosal changes. However, the sensitivity and specificity of the endoscopic duodenal mucosal appearance for the diagnosis of IL are not reported.
To evaluate the utility of endoscopic images of the duodenum for diagnosis of IL. Endoscopic appearance of the duodenal mucosal might predict histopathologic diagnosis of IL with a high degree of sensitivity and specificity.
51 dogs that underwent upper gastrointestinal (GI) endoscopy and endoscopic biopsies.
Retrospective review of images acquired during endoscopy. Dogs were included if adequate biopsies were obtained during upper GI endoscopy and digital images were saved during the procedure. Images were assessed for the presence and severity of IL. Using histopathology as the gold standard, the sensitivity and specificity of endoscopy for diagnosing IL were calculated.
Intestinal lymphangiectasia (IL) was diagnosed in 25/51 dogs. Gross endoscopic appearance of the duodenal mucosa had a sensitivity and specificity (95% confidence interval) of 68% (46%, 84%) and 42% (24%, 63%), respectively for diagnosis of IL. Endoscopic images in cases with lymphopenia, hypocholesterolemia, and hypoalbuminemia had a sensitivity of 80%.
Conclusions and Clinical Importance
Endoscopic duodenal mucosa appearance alone lacks specificity and has only a moderate sensitivity for diagnosis of IL. Evaluation of biomarkers associated with PLE improved the sensitivity; however, poor specificity for diagnosis of IL supports the need for histopathologic confirmation.
World Small Animal Veterinary Association
Intestinal lymphangiectasia (IL) in dogs is a dilatation of lymphatic vessels within the gastrointestinal (GI) tract. IL may be a primary disease but is typically a secondary process in dogs.[1-8] The most common mechanisms associated with IL include increased lymphatic pressure caused by inflammatory or neoplastic intestinal diseases, and increased venous pressure at the level of the thoracic duct attributable to right-sided heart failure, pericarditis, or pericardial effusion.[1, 2, 4, 5, 7, 8] Lymphatic distention varies with severity and can be either focal or diffuse; clinical signs and laboratory abnormalities may also show significant variability. Laboratory abnormalities associated with IL include hypoalbuminemia, panhypoproteinemia, lymphopenia, hypocalcemia, hypocholesterolemia, and hypocobalaminemia.[1, 2, 5, 7, 8] These biomarkers are not specific, but merely suggestive of a protein-losing enteropathy (PLE) and other non-GI causes of hypoalbuminemia such as cutaneous losses, liver failure, or protein-losing nephropathy must be ruled out.[1, 2, 5, 7, 8]
Definitive diagnosis of IL is obtained through histopathologic evaluation of intestinal biopsies, which can be obtained surgically or endoscopically. Advantages of endoscopic biopsies include decreased invasiveness, expense, and patient risk relative to surgical biopsies, and the opportunity to obtain biopsies via direct magnified visualization of the mucosa.[1, 9] Because of the potential multifocal or segmental nature of IL, endoscopic biopsies may be nondiagnostic if unaffected areas are sampled. In particular, routine upper endoscopy might not permit biopsy of more distal segments of the small intestine.[3, 9, 11, 12] In addition, lesions located deep within the submucosal and muscularis layers of the intestinal wall are not easily sampled with instruments designed for endoscopic biopsies.[4, 8, 9, 12] .
Multiple studies have reported specific characteristics of the endoscopically visualized duodenal mucosa in humans and dogs with IL. In humans, these characteristics include scattered white spots at the tips of villi,[13-16] white nodules or macules, submucosal elevations,[13, 14] or yellow or white plaques.[16, 17] Dogs reportedly have similar characteristics, including the presence of white tipped villi,[1, 4, 7, 8, 11, 18] a white patchy appearance, multifocal white granular foci of the mucosa or occasionally lymphatic fluid within the intestinal lumen.[5, 7, 8, 18] Although these findings have been associated with IL, they can be present with other intestinal diseases such as inflammatory bowel disease, intestinal neoplasia,[1, 4, 5, 8] infectious diseases, and even in clinically unaffected individuals.[7, 14, 16, 17] .
Intestinal lymphangiectasia might be diagnosed based on endoscopic appearance[9, 19] or that endoscopic appearance is highly suggestive of IL.[7, 18] However, no prior study has evaluated the relationship between the endoscopic appearance and histopathologic diagnosis of IL in dogs. The purpose of this study was to retrospectively evaluate the utility of endoscopic images of the duodenal mucosa to predict the presence of IL. We hypothesized that endoscopic findings would predict histopathologic diagnosis of IL with a high degree of sensitivity and specificity. In addition, we postulated that sensitivity and specificity would improve when evaluating the subset of dogs presenting with other biomarkers associated with IL including hypoalbuminemia, panhypoproteinemia, hypocholesterolemia, hypocalcemia, hypocobalaminemia, and lymphopenia.
Materials and Methods
Criteria for Case Selection
Medical records of dogs admitted to the University of Wisconsin Veterinary Medical Teaching Hospital from November 2007 to March 2010 were reviewed for inclusion in the study. Dogs were considered for inclusion if upper GI endoscopy was performed and intestinal tissue was obtained for histopathologic evaluation. Two internists (JAG, DSF), blinded to all patient data, independently reviewed and scored digital images saved during endoscopy. Sixty-two cases met the initial inclusion criteria. Seven cases were excluded because there were fewer than 4 images with sufficient quality available for scoring. Three cases were excluded because biopsy samples were nondiagnostic, attributable to insufficient quality or number of biopsies, or handling artifacts. One case was excluded because of disparate endoscopic grades between evaluators.
Signalment, presenting clinical signs, relevant physical exam findings, total protein, albumin, globulin, cholesterol, calcium, ionized calcium, folate, cobalamin, and lymphocyte count were recorded when available within 30 days of endoscopic examination. Biomarker abnormalities were determined based on the reported reference intervals. Endoscopy was performed with Olympus 1M GIF-160,1 Olympus 2.1M SIF-100,2 and Olympus 1M GIF-XP160.3 The number of biopsy samples evaluated was reported in all 51 cases. The mean and median numbers of duodenal biopsies evaluated were 10.9 and 11, respectively.
Before evaluation of the duodenal mucosa images, a grading scale for severity of IL was established based on appearance of white foci, mucosal granularity, and active lymphatic discharge. For each image, the location within the GI tract was noted and the images were scored based on severity of IL on a scale of 0–3. IL scores assigned by the 2 evaluators were averaged and a final IL score was assigned to each patient. Table 1 summarizes the grading scale criteria, scores, interpretation, and number of cases with each IL score. Cases were excluded if the difference in endoscopic score between evaluators was >1.
Table 1. Endoscopic IL score criteria, scoring, interpretation, and results
|Absence of white foci||0||0||Normal||12|
|Presence of rare pinpoint white foci with or without mucosal irregularity or granularity||1||0.5–1||Mild IL||17|
|Presence of diffuse pinpoint white foci and mucosal irregularity or granularity||2||1.5–2||Moderate IL||12|
|Presence of pinpoint to coalescing white foci, mucosal irregularity or granularity, and active lymphatic drainage||3||2.5–3||Severe IL||10|
A board certified pathologist (CMB) evaluated all GI biopsies. The pathologist was blinded to all patient data, including the retrospective assessment and scoring of the duodenal mucosa, and the previously determined histopathologic diagnosis. Before evaluation of biopsies, the pathologist established the following criteria for histologic diagnosis of IL: diagnosis of IL was based on the pathologist's ability to consistently identify the lumen of dilated lacteals that contained finely, granular eosinophilic (proteinaceous) material. The severity of lacteal dilatation, and therefore IL, was graded according to guidelines established by WSAVA. Considering the most severely affected villi in each case, a grade of mild, moderate, or marked IL was assigned if the width of the central lacteal as percentage of the width of the lamina propria was 25–50%, 50–75%, or greater than 75%, respectively. A final histopathologic diagnosis of IL was considered the gold standard for calculation of sensitivity and specificity.
Population characteristics such as age and weight were reported as medians and ranges. Prevalence of IL in the study population was calculated based on histopathologic results from GI biopsies. The age and weight of patients with and without a histopathologic diagnosis of IL were compared by the Mann-Whitney-U test. Sensitivity and specificity of endoscopic IL scores were calculated by means of histopathologic diagnosis of IL as the gold standard. True positives were defined as endoscopic images with an IL score of ≥0.5 and a histopathologic diagnosis of IL. True negatives were defined as endoscopic images with an IL score of <0.5 and absence of a histopathologic diagnosis of IL. False positives were defined as endoscopic images with an IL score of ≥0.5 without a histopathologic diagnosis of IL whereas false negatives were defined as endoscopic images with an IL score of <0.5 with a histopathologic diagnosis of IL. Sensitivity and specificity were also calculated for the subgroup of patients with concurrent biomarkers associated with PLE including hypoalbuminemia, panhypoproteinemia, hypocholesterolemia, hypocalcemia, and lymphopenia. For all statistical analyses, a P-value of <.05 was considered significant. A Pearson correlation coefficient was calculated to assess the relationship between IL scores derived endoscopically and histopathologically.
Fifty-one cases met the inclusion criteria. There were 25 castrated males, 22 spayed females, 2 intact males, and 2 intact females. The median age at the time of biopsy was 7.3 years (range 1.5–13.8 years) and the median weight was 19.0 kg (range 3.2–62.8 kg). Frequently represented breeds included Labrador Retrievers (11/51), Cocker Spaniels (4/51), Yorkshire Terriers (3/51), and German Shepherds, Cockapoos, Siberian Huskies, Boxers, Standard Poodles, and Rat Terriers (2/51 of each breed).
Pertinent laboratory data were available for the majority of the study population: albumin (51/51), total protein (51/51), cholesterol (49/51), calcium (48/51), and lymphocyte count (45/51). Other biochemical data available include cobalamin (19/51) and ionized calcium (8/51). Patients with hypocobalaminemia (7) and ionized hypocalcemia (5) were deemed to have insufficient numbers for calculation of sensitivity and specificity so these parameters were excluded from further analysis.
Based on final histopathologic diagnosis, the prevalence of IL in the study population was 25/51 (49%). Of the 25 cases with IL there were 9 castrated males, 13 spayed females, 2 intact males, and 1 intact female. The median age at the time of biopsy was 7.2 years (range 4.0–13.5 years) and the median weight was 18.0 kg (range 3.2–62.8 kg). Frequently represented breeds with IL included Labrador Retrievers (5/25), Cocker Spaniels (4/25), and Yorkshire Terriers (3/25). The cases without IL included 16 castrated males, 9 spayed females, and 1 intact female. The median age at the time of biopsy was 7.2 years (range 1.5–13.8 years) and the median weight was 22.8 kg (range 7.0–55.2 kg). The 2 populations did not differ significantly with respect to age (P = .48) or weight (P = .42).
For each dog included in the study, a minimum of 4 adequate images of the duodenum were required. The mean and median numbers of duodenal endoscopic images evaluated per patient were 10 and 9, respectively (range 4–44). Table 2 summarizes the sensitivity and specificity of all endoscopic still image IL scores for predicting a histopathologic diagnosis of IL. In addition, results for subsets with various biomarker abnormalities associated with PLE are included. There were no statistically significant differences in total protein, albumin, globulin, cholesterol, total calcium, or lymphocyte count when the dogs with a severe endoscopic IL score were compared to those with a moderate IL score, mild IL score, or normal appearance. There was no statistically significant relationship between the endoscopic IL and histopathologic IL scores (r = 0.16).
Table 2. Test properties utilizing endoscopic images for the diagnosis of IL in all cases and subsets with biomarkers associated with IL as compared to final histopathologic diagnosis
|Total||51 ||68.0 (46.4, 84.2)||42.3 (24.0, 62.8)|
|Hypoalbuminemia||17 ||73.3 (44.8, 91.1)||-|
|Panhypoproteinemia||16 ||73.3 (44.8, 91.1)||-|
|Hypocholesterolemia||16 ||73.3 (44.8, 91.1)||-|
|Lymphopenia||18 ||80.0 (44.2, 96.5)||25.0 (4.5, 64.4)|
|Hypocalcemia||15 ||71.4 (42.0, 90.4)||-|
|6 ||80.0 (29.9, 98.9)||-|
Final histopathologic diagnoses included normal cell infiltration (14/51) where 6/14 had a diagnosis of IL, mild inflammatory cell infiltrate (27/51) where 12/27 had a diagnosis of IL, moderate inflammatory cell infiltrate (9/51) where 7/9 had a diagnosis of IL, and severe inflammatory cell infiltrate (1/51) where 0/1 had a diagnosis of IL. Intestinal neoplasia was diagnosed in 4/51 cases with 1/4 cases also diagnosed with IL. A histopathologic diagnosis of diffuse, mild to moderate lymphoplasmacytic and eosinophilic duodenitis with edema and crypt abscesses was made in the 1 case excluded attributable to endoscopic scoring difference >1.
Results of this study, which contained a population with an IL prevalence of 49%, showed the sensitivity and specificity of endoscopic appearance for diagnosis of IL to be 68% and 42%, respectively. Therefore, these results indicate that gross endoscopic appearance of the duodenal mucosa lacks the sensitivity and particularly the specificity to accurately predict IL.
In cases with biochemical abnormalities typical of IL, the sensitivity of evaluated images improved; however, specificity decreased when evaluating the subpopulation with lymphopenia alone. Endoscopic findings in the subset of patients with combined hypoalbuminemia, hypocholesterolemia, and lymphopenia were moderately sensitive for detection of IL. Although there were improvements in sensitivity when evaluating the subpopulations with biochemical abnormalities, the modest degree of improvement coupled with the decrease in specificity in the subpopulation with lymphopenia necessitates intestinal biopsy for histopathologic diagnosis of IL.
In people, IL is the most common cause of scattered white spots observed within the intestine, but a similar endoscopic appearance can also be detected in chronic nonspecific duodenitis and giardiasis or a combination of any of the three etiologies. Therefore, it is reasonable to consider that dogs with scattered white spots noted endoscopically could have disease ranging from nonspecific duodenitis to IL.
There are important limitations to this study that warrant further discussion. First, our study population was derived from patients treated at a referral institution, which might draw from a more severely or chronically affected population. Therefore, the prevalence of IL is likely higher in our study population compared to the overall canine population, which would affect the predictive values of IL scores. Although our demographic precludes use of IL scores for prediction of IL in the overall population of dogs, the test values of IL scores calculated in this study remain clinically relevant. IL scoring would more likely be utilized in dogs with GI disease severe enough to require endoscopy and biopsy to determine diagnosis and treatment, rather than dogs with mild signs or GI disease that responds to dietary modification alone.
Retrospective review of still, as opposed to live, endoscopic images might have limited our ability to assign accurate IL scores, in particular when attempting to define focal versus diffuse disease. In addition, we relied upon the endoscopist to obtain representative images, but had no means of verifying that this occurred. However, as still images are routinely captured to represent significant findings and abnormalities, this disadvantage should be minimal. In an effort to eliminate variability between evaluators, endoscopic still image review was also elected in lieu of written nonstandardized endoscopic report review. This decision resulted from the variability in reported findings and lack of standardization of endoscopic reports as the study period encompassed an institutional transition to a standardized WSAVA endoscopic report. An advantage of retrospective image review was the blinding of the evaluators to all other information including signalment, clinical signs, laboratory data, and final diagnosis. Even though the evaluators were blinded, they maintained excellent agreement given that only 1 case was excluded attributable to a major deviation in endoscopic score.
Because of the retrospective nature of the study, not all biomarker parameters were measured for each dog. Although hypocobalaminemia has been associated with PLE and negative outcomes the limited number of dogs with documented hypocobalaminemia in this study (n = 7) did not allow us to draw conclusions regarding its utility for diagnosis of IL. Similarly, very few dogs with a documented ionized hypocalcemia (n = 5), or combined hypocholesterolemia, hypoalbuminemia, and lymphopenia (n = 6) were included in this study. In future studies, larger sample sizes within each subset will help determine the true utility of combining endoscopic findings with laboratory abnormalities to make a presumptive diagnosis of IL.
In our study, ileal biopsies were obtained in only 4 of 51 cases, an insufficient number to evaluate the utility of ileal biopsies for diagnosis of IL. Endoscopic lesions might be more readily detected in ileal biopsies.[8, 22] These findings suggest that upper and lower GI endoscopy should be performed whenever possible to increase the chance of an accurate diagnosis. Failure to consistently obtain ileal biopsies might have underestimated true positive and overestimated false positive results, thus affecting sensitivity and specificity of endoscopic IL scores in the diagnosis of IL.
There are also limitations associated with interpretation of GI biopsies, and in particular, endoscopic biopsies. A previous report demonstrated that significant interindividual variability exists between pathologists evaluating GI biopsy samples. This variability might increase further based on the quality of endoscopic biopsy; the sensitivity of diagnosing IL with endoscopic biopsies significantly increases when better quality samples are obtained. To optimize diagnostic utility and avoid interobserver variability in this study, a single pathologist evaluated multiple biopsy specimens and all slides from each case. This approach allowed consistent interpretation of microscopic findings but might have decreased accuracy compared to slide review by multiple pathologists. In addition, although WSAVA has established criteria for characterizing lacteal dilatation in GI biopsy samples, there is no current standard for the histopathologic diagnosis of IL. Therefore, the WSAVA guidelines for lacteal dilatation were used to establish criteria for diagnosis of IL in this study. Utilizing these standards, the number of dogs diagnosed with IL (n = 25) following blinded histopathologic review increased slightly when compared to the number of dogs with lacteal dilatation described in the original histopathology reports (n = 21). IL was graded as mild in all cases (n = 4) in which lacteal dilatation was identified by the study pathologist but not in the original histopathology report. This observation illustrates inconsistency in how a histopathologic diagnosis is rendered based on the described microscopic findings.
Endoscopic biopsy could be inferior to surgical biopsies for diagnosis of IL for 2 main reasons. First, the thickness of endoscopic biopsy samples is limited. Dilated lymphatic vessels that may be present between the mucosa and submucosa are inconsistently accessed with endoscopic biopsies.[7, 8] In 1 study evaluating surgical full-thickness biopsies, 76% of IL was found transmurally. Submucosal dilatation of lymphatics is a necessary finding for the diagnosis of IL; therefore, full-thickness intestinal biopsies might be required for definitive diagnosis of IL. Second, studies have documented the segmental nature of IL, and endoscopy could allow limited access to affected areas.[3, 9, 11, 17] Despite these limitations, endoscopy has the noteworthy benefit of minimizing patient risk. Many dogs with IL are severely hypoalbuminemic and some have demonstrated a greater risk for life-threatening surgical complications in hypoalbuminemic patients, although others dispute these findings.[26, 27] Despite conflicting results of studies, in many cases endoscopic biopsies might be deemed the safest, and therefore the best diagnostic option to minimize potential risks from a more invasive procedure in a compromised patient. A final potential benefit of endoscopy is the ability to target observed mucosal lesions for biopsy. Ideally, comparison of both endoscopic findings and surgically obtained biopsies should be performed. However, because of the retrospective nature of this analysis, both techniques were rarely utilized in conjunction in a single dog and a meaningful analysis could not be conducted.
Future studies further evaluating the correlation between GI endoscopic appearance and histopathologic diagnosis of IL should include a prospective blinded review of real-time endoscopic video paired with use of standardized WSAVA endoscopy reports. Although description of lacteal dilatation has been published, standardized criteria for histopathologic diagnosis of IL, severity of IL, and description of specific lesions, including their locations and any associated characteristics, require establishment. Evaluation of IL scores with concurrent assessment of other biomarker abnormalities such as hypoalbuminemia, panhypoproteinemia, hypocholesterolemia, hypocalcemia, lymphopenia, and hypocobalaminemia should include an adequate sample size to determine how these biomarkers affect the utility of IL scores for the diagnosis of IL. Finally, future studies should incorporate at least 2 pathologists for slide review and comparison of results.
In conclusion, although endoscopy is moderately sensitive for detection of IL, it is not specific. When the dog has concurrent biomarkers associated with PLE, endoscopic sensitivity improves slightly but potentially at the cost of specificity. Therefore, although endoscopic appearance of IL is supportive of the diagnosis, multiple adequate endoscopic biopsies of affected regions are recommended for definitive diagnosis of IL.
Olympus America, Inc, Center Valley, PA
Olympus America, Inc
Olympus America, Inc