Acute graft-versus-host disease (GVHD) is a major cause of morbidity and mortality following allogeneic stem cell transplant (SCT). The incidence of acute GVHD after allogeneic SCT, using human leukocyte antigen–matched sibling marrow and standard prophylactic immunosuppression, is 20% to 50% (1). Following skin GVHD, the gastrointestinal (GI) tract is the second most commonly involved organ system in acute GVHD and carries a poor prognosis (2,3). Acute GVHD is defined as GVHD occurring within the first 100 days following transplantation (4).
Diarrhea, nausea, abdominal pain, and vomiting are the primary manifestations of GI GVHD. Because these symptoms are nonspecific, a confirmatory biopsy is necessary to establish the diagnosis. The mucosal site with the highest diagnostic yield (upper and/or lower) varies between studies, with the majority coming from the adult literature. There is no consensus as to what portion of the GI tract is most likely to yield the diagnosis of GVHD in children. To avoid the risks of duodenal biopsies, including hemorrhage and intramural hematoma, some centers recommend gastric biopsies alone to establish the diagnosis of GI GVHD (5–7). Other centers prefer duodenal biopsies owing to their presumed higher diagnostic yield (8). Recent studies in adults support the traditional approach using rectosigmoid biopsies as the preferred diagnostic test (9–11) Our aim was to compare the relative contribution of endoscopic appearance and biopsies from upper and lower endoscopy for the diagnosis of acute GI GVHD in pediatric SCT recipients.
All pediatric patients younger than 18 years who had endoscopic evaluation within the first 100 days after allogeneic SCT between January 1999 and December 2009 were identified. Only patients undergoing their initial endoscopic evaluation for GI symptoms and having at least 1 biopsy obtained from the esophagus, stomach, duodenum, and/or rectosigmoid were included. We identified 48 patients who met these criteria. Demographic information, disease, symptoms prompting endoscopic evaluation, endoscopic findings, and histologic findings were extracted from the patient's electronic medical records. This retrospective study was approved by our institutional review board.
Diagnosis of GI GVHD
In addition to routine hematoxylin and eosin staining for pathological examination, biopsy specimens from all of the cases were subjected to immunohistochemical examinations for cytomegalovirus and submitted for viral culture. The specimens were reviewed by the on-call pediatric pathologist. Interpretation was aided by full access to the clinical record and knowledge of the patient's condition. For the purposes of the present study, the diagnosis of acute GI GVHD was based on the presence of glandular apoptosis in the absence of another inflammatory or infectious cause. We identified all of the patients in whom at least 1 GI mucosal biopsy was interpreted as being consistent with GVHD. All of the patients with diarrhea had Clostridium difficile toxin assays performed before endoscopy. Two patients were excluded because of positive cytomegalovirus tissue culture.
Descriptive statistics were calculated and presented as median and range. Differences between groups were evaluated with Kruskal-Wallis or Fisher exact test as appropriate. Statistical significance was defined by a P value <0.05. Statistical analyses were performed using SAS Learning Edition version 4.1 (SAS Institute, Cary, NC).
A total of 48 patients were identified who had endoscopic evaluation for GI symptoms, in whom biopsy specimens were obtained from the esophagus, stomach, duodenum, and/or rectosigmoid. The median age was 8 years (range 0.5–18). Leukemias were the most common underlying diseases (71%) (Table 1). Most patients received matched peripheral blood stem cells from siblings or other relatives. The median time of endoscopic biopsies following SCT was 54 days (range 21–98). None of our patients had endoscopic evaluation for GVHD within 3 weeks of SCT. This is particularly important because it is difficult to distinguish apoptosis due to treatment effects from that due to GVHD within 3 weeks of SCT (12–14).
The most common symptoms prompting endoscopic evaluation were diarrhea (70%), the combination of nausea and vomiting (67%), and abdominal pain (65%). Skin GVHD was present in 38% of patients (Table 2). There was no statistical difference between the symptoms and the presence or absence of GVHD. None of these clinical features were uniquely associated with either upper or lower GI tract involvement.
The most common endoscopic finding in patients with GVHD (at sites of GVHD) was normal mucosa seen in 25% of the gastric examinations, 57% of duodenal, and 50% of the rectosigmoid examinations (Table 3). Erosions or ulcers were seen in only 4% of the gastric, 29% of the duodenal, and 14% of the rectosigmoid examinations.
The number of biopsies per site per patient ranged from 1 to 8 (median 2). There was no statistical difference in the number of biopsies for different sites. GVHD was diagnosed in at least 1 site in 40 patients. The prevalence of biopsy-confirmed GVHD in patients suspected of having acute GI GVHD was 83%. A total of 22 patients had simultaneous upper and lower (rectosigmoid) endoscopy, 11 patients had only EGD, and 7 had only lower endoscopy. For data analysis, patients were divided into 2 main groups: patients who had simultaneous upper and lower endoscopy and patients who had upper endoscopy alone.
Twenty-two of the 40 (55%) patients in whom the diagnosis of GI GVHD was made had simultaneous upper and lower endoscopic biopsies (Fig. 1). In 17 patients, GVHD was identified in both upper and lower endoscopic biopsies. The sensitivity and negative predictive values for diagnosing GVHD for the rectosigmoid and upper endoscopic biopsies were equal and were 77% and 61%, respectively. The specificity and PPV could not be calculated because there was not a separate criterion standard by which the GVHD diagnosis was made. All of the patients had rectosigmoid biopsies, except 4 who had full colonoscopy, including rectosigmoid biopsies. Two patients had GVHD in the left and right side of the colon, and the other 2 had GVHD diagnosed only in rectosigmoid biopsies.
Of the 22 patients with simultaneous upper and lower endoscopy, 5 patients were diagnosed only by upper endoscopy and 5 patients were diagnosed only by lower endoscopy (Table 4). Of note, all of the patients with isolated upper GVHD had vomiting compared with only 1 patient with isolated lower GI GVHD; however, all of the patients with isolated lower GVHD had diarrhea compared with only 2 patients with isolated upper GVHD (Table 4). Thirty-three of the 40 patients had upper endoscopy. Twenty-eight of 33 (85%) patients had GVHD. Twenty-four of 33 had positive gastric biopsies for GVHD (86% sensitivity). Fourteen of 33 had positive duodenal biopsies (50% sensitivity). A total of 13 patients had positive stomach and negative duodenal biopsies, whereas in only 3 cases was the reverse true (nonsignificant P = 0.06). Six patients had positive esophageal biopsies. In 1 patient, esophageal biopsies showed GVHD, whereas the other biopsies were negative (Table 5). The patient with the isolated esophageal GVHD had nausea, vomiting, and abdominal pain similar to the other 6 patients.
None of our patients had bleeding after the procedure requiring intervention. Two patients had bacteremia during the week following endoscopy that can be attributed to their immunosuppressed state because their neutrophil and T cell count were still low.
Given the broad differential diagnosis and nonspecific nature of the clinical presentation, tissue biopsy is usually required to establish the diagnosis of acute GVHD following SCT. Although endoscopy with biopsy is commonly used in the evaluation of suspected GI GVHD, the best diagnostic approach in children has not been standardized. The extent to which GVHD involves the GI tract is a related topic of debate.
In adults, some authors suggest that GI involvement is selective and advocates specific sites as the most sensitive for diagnosing GVHD (6,15). Other authors argue that the process is pan intestinal (16). Recent adult data showed that in GVHD, rectosigmoid biopsies had higher sensitivity and negative predicted value than gastric or duodenal biopsies (11). Ross et al (11) identified 112 patients who had simultaneous endoscopic biopsies of the stomach, duodenum, and rectosigmoid within the first 100 days following SCT. Eighty-one percent of the patients had GI GVHD. Rectosigmoid biopsies had the highest sensitivity and negative predictive value for diagnosing GI GVHD, at 95.6% and 84%, respectively. The sensitivities of gastric and duodenal biopsies versus rectosigmoid were 72.5% (P < 0.0001) and 79.2% (P = 0.0018), respectively. Other studies have suggested that UGI biopsies are superior to rectal or rectosigmoid biopsies in the diagnosis of GVHD. In a prospective study of patients with the onset of diarrhea and UGI symptoms between day 20 and 100 posttransplant, 26 patients were identified with biopsy-proven intestinal GVHD by upper endoscopy and sigmoidoscopy. Biopsies were taken from the stomach, duodenum, and rectosigmoid. Gastric biopsy was positive in 85%, whereas duodenal and rectosigmoid biopsy were each positive in 58% (17).
In a pediatric retrospective study, Khan et al (18) found that acute GVHD was found on sigmoid biopsy in 38% compared with 17% and 21% for gastric and duodenal biopsies. Unlike our study, the present study did not include simultaneous biopsy from the upper and lower GI tract, making the comparison not possible.
In contrast, we found upper endoscopy and flexible sigmoidoscopy had similar sensitivities in 17 of 22 patients with simultaneous upper and lower endoscopy; GVHD was identified in both upper and lower endoscopic biopsies.
We and others found that the type of the clinical features were not uniquely associated with either upper or lower GI tract involvement, which questions the practice of having symptom-directed endoscopic evaluation (11,15). The cause of these symptoms is likely to be multifactorial and includes conditioning and neutropenia. Although looking to the small subgroup of patients (Table 4) with GVHD diagnosed only by upper endoscopy had predominant vomiting compared to the patients with GVHD diagnosed only by lower endoscopy and had diarrhea as a predominant symptom, which may suggest a correlation of the clinical features and the distribution of GVHD in the GI tract. Our data suggest that it would be reasonable to start with a flexible sigmoidoscopy or suction rectal biopsy. Both techniques are equal in diagnosing GVHD, although guided biopsies by sigmoidoscopy can have a higher yield when there is an infectious cause and the ability to observe post-biopsy bleeding. This approach can be helpful especially in patients who may tolerate an unsedated procedure with diarrhea and then do an upper endoscopy only if that is negative and clinical suspicion remains high.
The gross endoscopic findings of GI GVHD can present with a wide range, including normal mucosa, mild edema, erythema, erosion, and ulcer. These findings are not sufficiently typical to confirm a diagnosis of GI GVHD. Histological examination is always required (11). In contrast to the results of others (19) that endoscopic diagnosis predicted histologic diagnosis in both acute and chronic GI, in our study normal endoscopy was the most frequent finding, suggesting that the patients at our institution may undergo endoscopy earlier as soon as symptoms suggestive of GVHD develop.
The present study has several limitations. This is a small retrospective study. A total of 11 patients had only upper endoscopy and 7 patients had only lower endoscopy. In these patients, it was impossible to compare the yield of upper endoscopy and lower endoscopy. A prospective study with larger numbers of patients with simultaneous upper and lower endoscopy is needed to clarify and to define further the yield of mucosal biopsies taken from various sites in the GI tract.
In conclusion, we found that rectosigmoid and combined upper endoscopic biopsies are equally sensitive for the diagnosis of acute GI GVHD in children. This sensitivity is not altered by the presence of different GI symptoms. Although not statistically significant, gastric biopsies are more sensitive than duodenal biopsies for the diagnosis of acute GVHD. Because flexible sigmoidoscopy can be done unsedated in appropriate patients at the bedside without anesthesia, it can be performed first to identify GI GVHD. If GVDH is found on rectosigmoid biopsy, then upper endoscopy would not be needed.
1. Weisdorf DJ, Anasetti C, Antin JH, et al. Allogeneic bone marrow transplantation for chronic myelogenous leukemia: comparative analysis of unrelated versus matched sibling donor transplantation. Blood
2. Gratwohl A, Brand R, Apperley J, et al. Graft-versus-host disease and outcome in HLA-identical sibling transplantations for chronic myeloid leukemia. Blood
3. Przepiorka D, Weisdorf D, Martin P, et al. 1994 consensus conference on acute GVHD grading. Bone Marrow Transplant
4. Couriel D, Caldera H, Champlin R, et al. Acute graft versus host disease: pathophysiology, clinical manifestations, and management. Cancer
5. Iqbal N, Salzman D, Lazenby AJ, et al. Diagnosis of gastrointestinal graft-versus-host disease. Am J Gastroenterol
6. Ponec RJ, Hackman RC, McDonald GB. Endoscopic and histologic diagnosis of intestinal graft-versus-host disease after marrow transplantation. Gastrointest Endosc
7. Lipson SA, Perr HA, Koerper MA, et al. Intramural duodenal hematoma after endoscopic biopsy in leukemic patients. Gastrointest Endosc
8. Terdiman JP, Linker CA, Ries CA, et al. The role of endoscopic evaluation in patients with suspected intestinal graft versus host disease after allogeneic bone marrow transplantation. Endoscopy
9. Oomori S, Takagi S, Kikuchi T, et al. Significance of colonoscopy in patients with intestinal graft-versus-host disease after hematopoietic stem cell transplantation. Endoscopy
10. Thompson B, Salzman D, Steinhauer J, et al. Prospective endoscopic evaluation for gastrointestinal graft-versus-host disease: determination of the best diagnostic approach. Bone Marrow Transplant
11. Ross WA, Ghosh S, Dekovich AA. Endoscopic biopsy diagnosis of acute gastrointestinal graft vs host disease. Rectosigmoid biopsies are more sensitive than upper gastrointestinal biopsies. Am J Gastroenterol
12. Snover DC, Weisdorf SA, Vercellotti GM, et al. A histopathologic study of gastric and small intestinal graft versus host disease following allogeneic bone marrow transplantation. Hum Pathol
13. Bombi JA, Nadal A, Carreras E, et al. Assessment of histopathologic changes in the colonic biopsy in acute graft-versus-host disease. Am J Clin Pathol
14. Shidham VB, Chang CC, Shidham G, et al. Colon biopsies for evaluation of acute graft-versus-host disease (A-GVHD) in allogeneic bone marrow transplant patients. BMC Gastroenterol
15. Wu D, Hockenberry DM, Brentnall TA, et al. Persistent nausea and anorexia after marrow transplantation: a prospective study of 78 patients. Transplantation
16. Vogelsang GB, Lee L, Bensen-Kennedy DM. Pathogenesis and treatment of graft-versus-host disease after bone marrow transplant. Annu Rev Med
17. Cox GJ, Matsui SM, Lo RS, et al. Etiology and outcome of diarrhea after marrow transplantation: a prospective study. Gastroenterology
18. Khan K, Schwarzenberg SJ, Sharp H, et al. Diagnostic endoscopy in children after hematopoietic stem cell transplantation. Gastrointest Endosc
19. Cruz-Correa M, Poonawala A, Abraham SC, et al. Endoscopic findings predict the histologic diagnosis in gastrointestinal graft-versus-host disease. Endoscopy
Keywords:Copyright 2012 by ESPGHAN and NASPGHAN
human stem cell transplantation; gastrointestinal; graft-versus-host disease