Henoch Schönlein purpura (HSP) is an immunoglobulin A–mediated hypersensitivity vasculitis, characterized by association of skin, joint, gastrointestinal, and renal symptoms. Gastrointestinal abnormalities related to HSP have been shown to commonly resolve under adequate medical management. Nonetheless, unnecessary exploratory laparotomy has been reported, inasmuch as the clinical presentation of patients with such abnormalities varies from vague to severe abdominal symptoms and shock, with or without the other symptoms of HSP (1,2). Ultrasound has proved to be effective as first-line screening in children with gastrointestinal symptoms (3). Ultrasound sensitivity for the detection of bowel abnormalities in children with HSP varies from 50% to 100%, with an unknown specificity (4–8). Pathologically, digestive involvement of the bowel is a dynamic process, which includes inflammatory involvement of small vessels within the lamina propria and submucosa, causing edema, necrosis, and hematoma (9). Thus, ultrasound findings for these pathological features include, with variable importance, both features of mural edema and inflammation and features of hematoma. However, to the best of our knowledge, the descriptions provided in the literature for gastrointestinal abnormalities in HSP neither include this potential polymorphism nor attempt to evaluate its significance.
The goal of our study was to describe the polymorphism and to assess the clinical significance of intestinal ultrasound findings in children with gastrointestinal symptoms of HSP.
PATIENTS AND METHODS
Our institutional review board approved the study and waived patients' written informed consent because of its retrospective nature. Our hospital database was searched for children first admitted between January 1998 and December 2005 with a diagnosis of HSP made on the basis of the combination of palpable purpura and at least 1 of the following features: diffuse abdominal pain, any biopsy specimen showing predominant immunoglobulin A deposition, arthritis or arthralgia, and renal involvement with hematuria or proteinuria (10,11). Abdominal ultrasound performed at the onset of symptoms and clinical data were available for 43 patients with gastrointestinal symptoms related to HSP: 24 boys and 19 girls, 3 to 14 years of age (average age 6.3 ± 2.3 years). The reason for requesting abdominal ultrasound was abdominal pain without any of the other symptoms of HSP in 11 (25.6%) patients, whereas a clinical presentation suggested HSP in association with gastrointestinal symptoms in 25 (58.1%). Abdominal ultrasound was performed in the 7 (16.3%) remaining patients having HSP without any abdominal symptoms, to exclude asymptomatic abdominal involvement before the beginning of corticoid therapy. As a rule in our institution, stool analysis, including bacterial culture; blood studies; upper gastrointestinal endoscopy; and biopsies are performed in symptomatic patients with negative ultrasound findings (ie, 5 patients in this study). Biopsies were not performed because at this time the diagnosis was elucidated. Endoscopic findings used to confirm HSP were mucosal petechiae and/or submucosal edema in the absence of peptic ulcers. The printed abdominal ultrasound films included scans performed through the 9 abdominal quadrants, with graded compression (12). Relevant images from pathological structures, if any, were also identified by location. Examinations were performed by an operator with at least 5 years of experience in pediatric radiology, using 5 to 7 MHz or 5 to 12.5 MHz multifrequency ultrasound arrays on the same ultrasound machine platform (ATL HDI 3000, upgraded HDI 5000, Philips, Eindhoven, the Netherlands). Additional scans performed with color or power Doppler requested by the referring clinician were available for 14 patients. The standard presets for such examinations in our laboratory include elevated levels of both the pulse repetition frequency and the gain. With these settings, only the normally vascularized bowel have no color or power Doppler signal.
Ultrasound data were reviewed in consensus by 2 radiologists who had, respectively, 15 and 5 years of experience in pediatric gastrointestinal radiology, blinded to the clinical data of the patients. For the purpose of this review, the ultrasound appearance of the intestinal wall was 5-layered, including the 2 outermost echoic interfaces and, from lumen to serosa, the hypoechoic mucosa, the hyperechoic submucosa, and the hypoechoic muscularis propria. For thickened bowel wall (≥4 mm), the reviewers were asked to designate, if possible, the involved layer(s). Reviewers were asked to assign each patient's examination to the following grades: group 0, normal findings; group I, differentiated wall thickening; group II, pseudodifferentiated wall thickening; and group III, dedifferentiated wall thickening. A description of abnormal grades is given in Table 1 and illustrated in Figure 1. Patients with mixed findings were assigned to the highest grade. Color or power Doppler data were evaluated for the absence or presence of abnormal intestinal vascularization, which was categorized as central, peripheral, or diffuse.
The original examination reports were searched by an independent investigator for localization, maximum wall thickness, and length of abnormal bowel segments identified on the films. Note was also made of any associated relevant finding. The same investigator also searched the clinical data to determine which type of treatment the patients received (eg, supportive, surgical, corticoids). The main clinical symptoms leading to corticoid administration were categorized as abdominal or extraabdominal. The mode of administration and the dose and the duration of the therapy were also extracted and stored. Records were also searched for biological and clinical follow-up data.
The pooled diagnostic value of ultrasound in diagnosing gastrointestinal involvement of HSP (grades I–III) was calculated according to the absence or presence of clinical symptoms as the standard of reference. Continuous data were expressed as mean ± SD. The characteristics of patients with different ultrasound findings were compared by the means of a univariate analysis of variance ANOVA. Multiple testing was used for intergroup comparisons. P < 0.05 indicated statistical significance.
Thirty-six patients had abdominal symptoms of HSP. Five (13.9%) had no abnormality at ultrasound, whereas in the remaining patients the abnormalities were graded I, II, and III in, respectively, 6 of 36 (16.6%) patients, 22 of 36 (61.1%) patients, and 3 of 36 (8.3%) patients. The results in 5 patients with no intestinal abnormalities at ultrasound were false negative because the intestinal involvement was further confirmed by upper gastrointestinal endoscopy in 3 patients and the presence of occult blood in the stools of the remaining patients. The results in all 7 patients without abdominal symptoms were true negatives because abdominal ultrasound in all was unremarkable. The sensitivity, specificity, positive predictive value, and negative predictive value of ultrasound for the diagnosis of gastrointestinal involvement of HSP were, respectively, 83.3%, 100%, 100%, and 53.8%.
In patients with abdominal symptoms, the average length of hospitalization was 11.82 ± 8.89 days, and the average duration of abdominal symptoms was 6.85 ± 6.65 days. In groups 0 to III, the average duration of abdominal symptoms was 2.20 ± 2.06, 5.67 ± 1.88, 6.29 ± .94, and 17.67 ± 2.66 days, respectively, and the average length of hospitalization was 4.80 ± 2.96, 9.17 ± 2.70, 11.46 ± 1.35, and 24.67 ± 3.82 days, respectively. The average duration of both hospitalization and abdominal symptoms in group III was significantly higher (P < 0.05) than in the other groups. In the 31 patients with positive ultrasound findings, the total number of abnormal intestinal segments at ultrasound was 61, with an average length of 14.05 ± 5.58 cm. Of the abnormal segments, 28 of 61 (45.9%) involved the jejunum, 18 of 61 (29.5%) the duodenum, 10 of 61 (16.4%) the ileum, 3 of 61 (4.9%) the stomach, and 2 of 61 (3.3%) the colon. In the 14 patients who had undergone color or power Doppler ultrasound, peripheral hyperemia was detected in 10 (71.5%) and diffuse hyperemia in 4 (28.5%).
Intraabdominal complications were diagnosed or suspected at the first ultrasound examination, then confirmed by further explorations (when needed) in 1 patient in group II (duodenal involvement leading to mild dilatation of the biliary tree and elevated pancreatic enzymes) and 2 patients in group III (small bowel intussusception and intestinal perforation). No other intraabdominal complication occurred in the study group at initial and follow-up examinations. Only the patient with pneumoperitoneum caused by ileal perforation was treated surgically, and the remaining patients were successfully treated with intravenously corticoids. Abdominal symptoms led to intravenous corticoids in 13 of 36 patients (36.1%), including 0 of 5 (0%) in group 0, 1 of 6 (16.7%) in group I, 10 of 22 (45.4%) in group II, and 2 of 3 (66.7%) in group III. Corticoids were given at the dosage of 2 mg/kg/day until the cessation of symptoms in all cases, with an average duration of 2.71 ± 0.91 days (range 2–5 days). Four other patients in the study group received corticoid therapy for extraabdominal symptoms (arthralgia, nephritis, or both): 1 in group 0, 1 in group I, and 2 in group II. Figure 2 shows the percentage of patients in each group treated with corticoid for both intraabdominal and extraabdominal symptoms.
The mean clinical and biological follow-up was 46.61 ± 22.50 months (range 12–96 months). Forty patients were free of symptoms at their last follow-up. At this writing, the 3 remaining patients have persisting proteinemia and microscopic hematuria. Of these patients, all had both intestinal and renal involvement at the onset of the disease; 2 were in group II and 1 in group III. Once diagnosis of HSP made, relapses occurred 15 times in 12 patients, with a mean interval of 13.37 ± 9.41 weeks after onset. Relapses consisted of the same symptoms present during the first episode in all patients except for 1 patient, who experienced arthralgia as a new symptom. Patients who experienced relapse with intestinal involvement were 2 in group I, 7 in group II, and 3 in group III.
Gradation of Ultrasound Appearances and Potential Clinical Importance
With high-resolution ultrasound, digestive involvement of HSP has been included in the differential diagnosis of bowel wall thickening, with uniform intraseries descriptions. These descriptions vary across series and include submucosal thickening, eccentric wall thickening, and loss of differentiation (4,7,8). In light of the present study, these uniform descriptions seem to be incomplete, inasmuch as differentiated (grade I), pseudodifferentiated (grade II), and dedifferentiated (grade III) wall thickening were present in our series in 6, 22, and 3 patients, respectively, of the 31 patients with positive ultrasound findings. We have described and proposed a gradation for digestive involvement of HSP based exclusively on ultrasound findings.
Normal bowel wall consists of 5 layers, including the outermost echoic thin mucosal and serosal interfaces, with submucosa appearing as a continuous homogeneous echoic structure, <2 mm thick, delineated by 2 hypoechoic layers: internally the lamina propria and externally the muscularis propria (13). Grade I abnormalities result of infiltration of 1 or many superficial layers with respect to the wall stratification. These abnormalities are not specific for a disease and have been described in many conditions, including inflammatory bowel diseases, pseudomembranous colitis, infectious colitis, ischemic colitis, infarction, graft-versus-host disease, and mesenteric venous obstruction (14,15). In the absence of pathological correlation in the present study, grade II abnormalities are more difficult to explain. Extravascular blood has various and heterogeneous reflectivity, but it mainly has an echoic appearance. Given that, we hypothesize that an extra-echoic layer internal to the submucosa may be secondary to deep mucosa and submucosa bleeding. This results in the so-called pseudodifferentiated appearance. Such an uncommon finding in the absence of coagulation disorder and trauma would be highly suggestive of hemorrhagic vasculitis. By contrast, loss of differentiation of the wall (grade III) is a relatively uncommon pattern that has been reported only with intense infiltrative or necrotic disease processes involving the wall, with no respect to compartmentalization of the layers (16). As a result, both grade II and grade III findings that reflect abnormalities of the ultrasound structure of the bowel wall are relatively uncommon.
Provided a proper recognition of grade II and grade III findings, which occurred in 22 of 36 (61%) symptomatic patients in our study, ultrasound could help patients with a potentially poor abdominal prognosis to benefit from early management of their vasculitis. Indeed, in the present study, these groups had the longer average duration of symptoms and hospitalization, especially group III (P < 0.05). Such early recognition and management has been shown to reflect positively on the outcome (17). Despite the false-negative results of ultrasound examinations in 5 of 36 (14%) symptomatic patients, further diagnostic procedures are questionable because patients in this group had the shorter average duration of symptoms and hospitalization in this study. Moreover, intraabdominal complications potentially requiring a change in management occurred only in groups II and III. As a result, the significance of a negative ultrasound result in the setting of abdominal pain in children with HSP could be predictive of a positive outcome.
Clinical Issues Associated With Intestinal Involvement of HSP
Abdominal symptoms are present in nearly two thirds of children with HSP and may precede the other symptoms for many days in a significant number of cases (1). The 2 most important clinical issues associated with digestive involvement of HSP are the onset or the worsening of abdominal symptoms in children with enough symptoms to complete the diagnosis of HSP and isolated abdominal pain of unknown origin. The first issue falls beyond the scope of the present study, although complications potentially requiring change in management were at least suspected at ultrasound in 3 of 3 (100%) patients. The second issue has been generically addressed by previous studies evaluating ultrasound in both the assessment of acute abdominal pain and distinguishing between surgical and nonsurgical causes of abdominal pain in children (3,15). The sensitivity of ultrasound in detecting gastrointestinal abnormalities in children with HSP was high [31/35 (86%)] and within the range (50%–100%) of that in published reports (4–8). The specificity of ultrasound findings in HSP is unknown because of the low incidence of the disease. In our study, the specificity in selected patients with known diagnoses of HSP was high (100%), but further studies are needed for assessment in a general population.
Several limitations should be acknowledged. First, the retrospective review of clinical and imaging data led to several unavoidable biases, including interpretation of bowel abnormalities on printed films and lack of data for appreciation of both abdominal symptoms and their duration, follow-up analysis, and evaluation of comorbid factors. Second, in this retrospective study over a 7-year period, the ultrasound findings may have affected the patients' treatment, resulting in an uncontrollable bias. Third, because of the lack of endoscopic and pathological correlation, we used clinical symptoms as the standard for the presence or absence of intestinal involvement. Such a reference is probably well correlated to the presence of intestinal involvement, but poorly to the parietal extent and the number sites of involvement. Last, the relatively low number of patients in this study, although comparable to those in previous studies, may have underpowered the significance of differences between patient groups.
In conclusion, there is a polymorphism of bowel wall ultrasound findings in children with HSP. The histological causes of this polymorphism remain unknown, although dedifferentiated bowel wall thickening has a poor clinical prognosis compared with the other findings.
1. Choong CK, Beasley SW. Intra-abdominal manifestations of Henoch-Schönlein purpura. J Paediatr Child Health 1998; 34:405–409.
2. Kraft DM, McKee D, Scott C. Henoch-Schönlein purpura: a review. Am Fam Physician 1998; 58:405–408.
3. Vasavada P. Ultrasound evaluation of acute abdominal emergencies in infants and children. Radiol Clin North Am 2004; 42:445–456.
4. Ozdemir H, Isik S, Buyan N, et al
. Sonographic demonstration of intestinal involvement in Henoch-Schönlein syndrome. Eur J Radiol 1995; 20:32–34.
5. Connolly B, O'Halpin D. Sonographic evaluation of the abdomen in Henoch-Schönlein purpura. Clin Radiol 1994; 49:320–323.
6. Kagimoto S. Duodenal findings on ultrasound in children with Schönlein-Henoch purpura and gastrointestinal symptoms. J Pediatr Gastroenterol Nutr 1993; 16:178–182.
7. Couture A, Veyrac C, Baud C, et al
. Evaluation of abdominal pain in Henoch-Schönlein syndrome by high frequency ultrasound. Pediatr Radiol 1992; 22:12–17.
8. Bomelburg T, Claasen U, von Lengerke HJ. Intestinal ultrasonographic findings in Schönlein-Henoch syndrome. Eur J Pediatr 1991; 150:158–160.
9. Shirahama M, Umeno Y, Tomimasu R, et al
. The value of color Doppler ultrasonography for small bowel involvement of adult Henoch-Schönlein purpura. Br J Radiol 1998; 71:788–791.
10. Mills JA, Michel BA, Bloch DA, et al
. The American College of Rheumatology 1990 criteria for the classification of Henoch-Schönlein purpura. Arthritis Rheum 1990; 33:1114–1121.
11. Ozen S, Ruperto N, Dillon MJ, et al
. EULAR/PReS endorsed consensus criteria for the classification of childhood vasculitides. Ann Rheum Dis 2006; 65:936–941.
12. Puylaert JB. Mesenteric adenitis and acute terminal ileitis: US evaluation using graded compression. Radiology 1986; 161:691–695.
13. Tio TL, Tytgat GN. Endoscopic ultrasonography of normal and pathologic upper gastrointestinal wall structure: comparison of studies in vivo and in vitro with histology. Scand J Gastroenterol Suppl 1986; 123:27–33.
14. Siegel MJ, Friedland JA, Hildebolt CF. Bowel wall thickening in children: differentiation with US. Radiology 1997; 203:631–635.
15. Gorg C, Wollenberg B, Beyer J, et al
. High-resolution ultrasonography in gastrointestinal graft-versus-host disease. Ann Hematol 2005; 84:33–39.
16. Baud C, Saguintaah M, Veyrac C, et al
. Sonographic diagnosis of colitis in children. Eur Radiol 2004; 14:2105–2119.
17. Trujillo H, Gunasekaran TS, Eisenberg GM, et al
. Henoch-Schönlein purpura: a diagnosis not to be forgotten. J Fam Pract 1996; 43:495–498.