bs_bs_banner
doi:10.1111/jgh.12651
GASTROENTEROLOGY
Clinicopathological features of patients with acute graft-versus-host disease of the upper digestive tract Kosuke Nomura,* Toshiro Iizuka,* Daisuke Kaji,* Hisashi Yamamoto,† Yasutaka Kuribayashi,* Ryusuke Kimura,* Akihiro Yamada,* Tsukasa Furuhata,* Satoshi Yamashita,* Daisuke Kikuchi,* Akira Matsui,* Toshifumi Mitani,* Osamu Ogawa,* Shu Hoteya,* Yasunori Ota,‡ Shuichi Taniguchi† and Mitsuru Kaise* Departments of *Gastroenterology and †Hematology, Toranomon Hospital, and ‡Department of Pathology, Research Hospital, The Institute of Medical Science, The University of Tokyo, Tokyo, Japan
Key words endoscopy, graft-versus-host disease, human stem cell transplantation, upper gastrointestinal tract. Accepted for publication 7 April 2014. Correspondence Dr Kosuke Nomura, Department of Gastroenterology, Toranomon Hospital, 2-2-2 Toranomon, Minato-ku, Tokyo 105-8470, Japan. Email: [email protected]
Abstract Background and Aims: Acute graft-versus-host disease (GVHD) occurring within 100 days post-transplant is one of critical factors influencing prognosis in transplant recipients. Among cases of GVHD of the gastrointestinal (GI) tract, GVHD rarely affects the upper GI. In this study, we retrospectively examined the frequency of upper GI GVHD and diagnostic accuracy. Patients and Methods: From among 868 patients who underwent allogeneic hematopoietic stem cell transplantation at our hospital between January 2005 and June 2012, 115 of whom underwent biopsy for upper GI symptoms. The endoscopic findings and histologic diagnosis from these 115 patients were retrospectively analyzed. Results: GVHD was histologically diagnosed in 85 patients overall (9.8% of all 868 transplant recipients). Although gastric mucosal exfoliation was not commonly observed, this endoscopic finding when used as a diagnostic predictor had both a specificity and positive predictive value (PPV) of 100%. When using redness, luster, and mucosal change as predictors, specificity and PPV were relatively high, suggesting that these gastric endoscopic findings are useful in the diagnosis of upper GI GVHD. Among the duodenal endoscopic findings, erosion as a diagnostic predictor had both a high specificity and PPV. The biopsy results often lead to a diagnosis of GVHD even in cases judged to be endoscopically normal. Conclusions: Among the gastric endoscopic findings, mucosal exfoliation, although rare, and redness, luster, and mucosal change are likely to be useful diagnostic predictors of upper GI GVHD. GVHD was frequently diagnosed in patients with endoscopically normal duodenum, suggesting that biopsies are important for definitive diagnosis.
Introduction Graft-versus-host disease (GVHD) results from immune responses where donor lymphocytes recognize histocompatibility antigens of the host (recipient) as foreign and launch an attack. Acute GVHD is defined as that occurring within the first 100 days posttransplant1 and it targets mainly the liver, gastrointestinal (GI) tract, and skin. On the other hand, chronic GVHD occurs more than 100 days post-transplant and its clinical symptoms resemble those of autoimmune disease. Chronic GVHD targets multiple organs (e.g. lungs, skin, and oral cavity), but lesions in the GI tract are rare. Both forms of GVHD are major complications that affect the prognosis of allograft recipients, and thus early diagnosis and treatment are necessary.2 However, GVHD can be complicated by various types of infections after hematopoietic stem cell transplan-
tation and by disorders caused via non-immune mechanisms, making differential diagnosis based solely on clinical symptoms difficult.3 GVHD of the digestive tract is diagnosed based on histological findings of biopsy specimens, which are obtained based on various endoscopic findings. The endoscopic features characteristic of GVHD of the lower GI tract, such as mucosal exfoliation and honeycomb-like patterns, have been reported in several studies.4,5 Many of these endoscopic findings are in good agreement with histological diagnosis, and thus identification of lesions associated with lower GI GVHD is not difficult. In contrast, endoscopic findings indicative of upper GI GVHD have not been well studied. Therefore, in this study, we retrospectively examined the frequency of upper GI GVHD that occurred within 100 days of hematopoietic stem cell transplantation and investigated its relationship with endoscopic findings.
Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872 © 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
1867
Acute upper GI graft-versus-host disease
K Nomura et al.
were used for GVHD diagnosis. A platelet count of ?50 000/mm3 was maintained in all patients before biopsy.
Methods Patients. Among 868 patients who had undergone allogeneic hematopoietic stem cell transplantation at our hospital between January 2005 and June 2012, acute GVHD of the digestive tract was suspected in 377 patients based on their clinical symptoms. Among them, 135 patients presented with upper GI symptoms, such as epigastralgia, vomiting, and loss of appetite, and underwent endoscopic examination. Histological diagnosis of biopsy specimens was performed in 115 of the 135 patients, and these 115 patients were the subjects of this retrospective study (Fig. 1). When upper GI endoscopy was performed twice or more within 100 days post-transplant, results of the first examination, not the later examinations, were analyzed in this study.
Endoscopic examination. Endoscopic findings were retrospectively determined for each patient as follows: for the stomach, linear redness, spotted redness, luster, erosion, ulceration, mucosal change, and mucosal exfoliation; and for the duodenum, redness, erosion, and ulceration (Fig. 2). The sensitivity, specificity, and positive predictive value (PPV) of each finding were then calculated against the histological diagnosis of GVHD, and also against that of ≥ grade III GVHD. Pathologic diagnosis of GVHD was performed according to the classification of GVHD severity proposed by Sale et al.6 (Table 1). When multiple biopsy specimens were obtained from the stomach and duodenum during a single endoscopic examination, the highest severity grade from among the specimens was used for this study. Because the diagnosis of GVHD is difficult in cytomegalovirus-positive specimens, biopsy specimens that did not indicate cytomegalovirus infection
Results Subjects were 66 men and 49 women (mean age 53.5 years; range, 19–74 years). The mean time of endoscopic examination was 53.9 days post-transplant (range, 20–99 days). Underlying diseases were acute leukemia (n = 67), malignant lymphoma (n = 23), myelodysplastic syndrome (n = 10), chronic myelogenous leukemia (n = 4), adult T-cell leukemia (n = 3), and other (n = 8). The sources of stem cells for transplantation were umbilical cord blood (n = 69), bone marrow (n = 30), and peripheral blood from related donors (n = 16). All patients received preventive therapy for GVDH with methotrexate (MTX) plus tacrolimus (FK506) or cyclosporine A (CyA) (n = 47), FK506 plus mycophenolate mofetil (MMF) (n = 41), FK506 or CyA only (n = 26), or other agents (n = 1) (Table 2). Non-steroidal anti-inflammatory drugs and aspirin, both of which may damage the mucosa, were orally administered to four and two patients, respectively, and a proton pump inhibitor or histamine H2-receptor antagonist was orally administered to all of the six patients. Biopsy specimens were taken from 3.3 sites/patient on average, and GVHD was histologically diagnosed in 85 patients, which accounted for 9.8% of all transplant recipients (n = 868) and 73.9% of the subjects of this study (n = 115). Fifty-one subjects (44.3%) had histologic severity grade I GVHD, while 22 (19.1%), 9 (7.8%), and 3 subjects (2.6%) had grade II, III, and IV GVHD, respectively. Complication by cytomegalovirus infection was found in 15 subjects (13.0%), 9 with GVHD, and 6 without GVHD.
Figure 1 Flow diagram of the study. CMV, ; GI, gastrointestinal; GVHD, graft-versus-host disease; HHSCT, homologous hematopoietic stem cell transplantation.
1868
Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872 © 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
K Nomura et al.
Acute upper GI graft-versus-host disease
a
b
c
d
e
f
g
h
i
j
k
Figure 2 Examples of each endoscopic finding. (a) Linear redness; (b) spotted redness; (c,j) erosion; (d) ulceration; (e) luster; (f,g) mucosal change; (h) mucosal exfoliation; (i) redness; (k) ulcer.
Table 1 Definition and histological grading system for acute gastrointestinal graft-versus-host disease Grade
Histological feature
I II
Single cell necrosis (apoptosis) noted on medium power Evidence of epithelial damage by crypt/glandular abscesses epithelial flattening or glandular/crypt dilation Dropout of one or more crypts/glands Total epithelial denudation
III IV
Multiple endoscopic findings were observed in many subjects. In the stomach, two findings were observed during a single endoscopic examination in 36 subjects, while three findings and four findings were observed in 35 and 21 subjects, respectively. In the duodenum, two findings were observed in 13 subjects, while three findings were seen in one subject. The most common gastric finding was erosion, which was present in 51.3% (59/115) of subjects. As for the other findings, luster was present in 50.4% (n = 58) of subjects, mucosal change in 49.6% (n = 57), linear redness in 27.8% (n = 32), spotted redness in 18.3% (n = 21), ulceration in 5.2% (n = 6), and mucosal exfoliation in 3.5% (n = 4). No findings were observed in 4.3%
(n = 5) of the subjects. The most common duodenal finding was redness, which was found in 32.2% (37/115) of subjects, followed by erosion and ulceration observed in 13.0% (15/115) and 2.6% (3/115) of subjects, respectively. Endoscopy did not detect abnormalities in the duodenal mucosa in 63.5% (73/115) of subjects. Comparisons of subjects with and without GVHD revealed no differences in individual findings; however, a significant difference was observed between the two groups when no abnormalities were evident in the stomach (Table 3). In the 85 patients histologically diagnosed with upper GI GVHD, when luster, mucosal change, and erosion were used as endoscopic gastric predictors, approximate diagnostic sensitivity was 50%, but specificity was high with mucosal exfoliation (100%), and erosion and linear/spotted redness (≥ 80%). The PPV was 100% when using gastric mucosal exfoliation as a predictor and was ≥ 75% when linear/spotted redness, mucosal change, and luster were used. No abnormalities were found in five patients, none of whom had GVHD; in other words, abnormal endoscopy findings were present in all GVHD patients (Table 3). Although data were re-evaluated using different combinations of endoscopic findings, again no significant differences were observed, indicating that none of the combinations were specific to GVHD. When
Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872 © 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
1869
Acute upper GI graft-versus-host disease
Table 2
K Nomura et al.
Patient characteristics
No. Mean age, years (range) Sex Male Female Mean days post transplantation Primary disease Acute leukemia Malignant lymphoma Myelodysplastic syndrome Chronic myelogenous leukemia Adult T-cell leukemia Others Stem cell source Cord blood stem cell transplantation Bone-marrow transplantation Allo-peripheral blood stem cell transplantation Previous treatment FLU-based regimen CY/TBI or CY/BU Preservation of GVHD CyA/FK506 + MTX FK506 + MMF CyA or FK506 Other
115 53.5 (19–74) 66 49 53.9 (20–99) 67 23 10 4 3 8 69 30 16 105 10 47 41 26 1
BU, busulfan; CY, cyclophosphamide; CyA, cyclosporin A; FK506, tacrolimus; FLU, fludarbine; MMF, mycophenolate mofetil; MTX, methotrexate; TBI, total body irradiation.
diagnostic use of gastric findings was re-examined in 12 subjects with ≥ grade III GVHD, which is conventionally considered nonresponsive, the PPVs for use of mucosal exfoliation and ulceration were 75% and about 50%, respectively, but the PPVs for other gastric findings were low (Table 4). Use of endoscopic duodenal findings as predictors of upper GI GVHD also had low sensitivities, but specificities ≥ 90% were achieved for ulceration and erosion. PPVs were ≥ 75% when erosion, normal mucosa, and redness were used as predictors (Table 3). Different combinations of endoscopic findings were evaluated separately, but again no significant differences were noted. PPVs were ≤ 10% when using any of the duodenal findings of the 12 subjects with ≥ grade III GVHD on histology (Table 4). Treatment was effective in 73.0% (62/85) of GVHD patients. Specifically, treatment was effective in patients with histologic grade I and II GVHD (80.4%, 41/51 and 77.3%, 17/22, respectively), but less effective in patients with histologic grade III GVHD (44.4%, 4/9) and not effective in those with histologic grade IV GVHD (0.0%, 0/3). When the effectiveness of therapy was examined in relation to endoscopic findings, the proportion of responders was low (50.0%) in patients with mucosal exfoliation and gastric ulceration (7/85), but was high (79.7%) in patients without these findings (78/85).
Discussion Acute GVHD occurs in 30–60% of patients who undergo allogeneic bone marrow transplantation.7 It mainly damages the liver, GI 1870
tract, and skin, and it is characterized clinically primarily by maculopapular rash, nausea/vomiting, diarrhea, and jaundice. GI lesions in acute GVHD have been evaluated in patients with lower GI symptoms, in other words, with severe diarrhea. However, some GVHD patients present only with upper GI symptoms (e.g. nausea and vomiting), not diarrhea. Weisdorf et al. propose that upper GI GVHD is a disease specifically targeting the upper GI tract and is thus distinct from lower GI GVHD.8 It was reported that upper GI GVDH occurs in 10–26% of allogeneic transplant recipients in Western countries,9 compared with 11% (22/205) of Japanese recipients.10 In the present study, upper GI GVHD was diagnosed in 9.8% (85/868) of all of the recipients, which is in a good agreement with previous report. In upper GI GVHD, lesions in the esophagus are extremely rare, but the stomach and duodenum are susceptible sites.3 Although previous studied have reported various endoscopic findings as characteristic of GVHD, such as mucosal edema, spotted redness, erosion, ulceration, and bleeding, the number of cases studied was small. In a study of 34 upper GI GVHD patients by endoscopy, Roy et al. found gastric inflammation such as redness and erosion was the most common finding, observed in 65% of patients, and that mucosal abnormalities were absent in 18% of patients.11 Terdiman et al. also reported that nonspecific findings such as redness and erosion were the most commonly findings in the 12 patients they assessed.3 In the present study, we examined 115 subjects and found that when gastric mucosal exfoliation, despite not being a common endoscopic finding, was used as a diagnostic predictor, both specificity and PPV were 100%. When using redness (linear or spotted), luster, and mucosal change as predictors, specificities and PPV were also relatively high, albeit not significantly. These results suggest that these endoscopic findings will be useful in the diagnosis of upper GI GVHD. This study also examined mucosal change, which has not been well examined in previous studies. Among 115 subjects, 57 presented with honeycomb-like, granular, and wrinkling changes in the gastric mucosa. In good agreement with the results in lower GI GVHD reported by Mashiba et al.,5 our histologic examination confirmed edema of the lamina propria and deceases in crypts, which is likely to explain the above endoscopic findings. These mucosal changes are useful findings, with a specificity of 63.3% and a PPV of 80.7%, although sensitivity was low. Among endoscopic findings in the duodenum, erosion had high specificity and high PPV, but low sensitivity as a diagnostic predictor. Unexpectedly, the biopsy findings of mucosal specimens with no apparent abnormalities were similar to those of erosion. This is in clear contrast to the gastric findings that none of the five patients without mucosal abnormalities had GVHD. As abnormalities are usually found in the stomach in upper GI GVHD, gastric biopsy is recommended in such cases. On the other hand, the duodenum was often endoscopically normal, although the reason is not known, suggesting that these organs need to be examined individually. Acute GVHD occurs predominantly in the skin, liver, and GI tract. Occurrence of lower GI GVHD is usually preceded by, or accompanied by, skin GVHD. Upper GI GVHD, on the other hand, often occurs without GVHD affecting other organs. For example, according to a report by Weisdorf et al., approximately 30% of upper GI GVHD patients did not have GVHD affecting other organs,8 similar to approximately 35.3% of our patients (n = 30). Thus, upper GI endoscopic examination is encouraged Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872
© 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
K Nomura et al.
Table 3
Acute upper GI graft-versus-host disease
Summary of outcomes
Total Stomach Linear redness Spotted redness Erosion Ulceration Luster Mucosal change Mucosal exfoliation No abnormality Duodenum Redness Erosion Ulcer No abnormality
n
GVHD
Non-GVHD
P-value
Sensitivity (%)
Specificity (%)
PPV (%)
115
85
30
32 21 59 6 58 57 4 5
26 18 41 4 46 46 4 0
6 3 18 2 12 11 0 5
0.27 0.17 0.27 0.68 0.18 0.10 0.23 < 0.01
30.6 21.2 48.2 4.7 54.1 54.1 3.5 0
80.0 90.0 40.0 93.3 60.0 63.3 100 83.3
81.3 85.7 69.4 66.7 79.3 80.7 100 0
37 15 3 73
27 12 1 55
7 3 2 18
0.38 0.56 0.10 0.65
31.8 14.1 1.2 64.7
66.7 90.0 93.3 40.0
73.0 80.0 33.3 75.3
GVHD, graft-versus-host disease; PPV, positive predictive value.
Table 4
Summary of outcomes; grade III or IV
Total Stomach Linear redness Spotted redness Erosion Ulceration Luster Mucosal change Mucosal exfoliation No abnormality Duodenum Redness Erosion Ulcer No abnormality
n
GVHD (grade III or IV)
Sensitivity (%)
Specificity (%)
PPV (%)
115
12
32 21 59 6 58 57 4 5
3 1 7 3 6 6 3 0
25.0 8.3 58.3 25.0 50.0 50.0 25.0 0
71.8 80.6 49.5 97.1 49.5 50.5 99.0 83.3
9.4 4.8 6.8 50.0 10.3 10.5 75.0 0
37 15 3 73
4 1 0 7
33.3 8.3 0 58.3
68.0 86.4 100 35.9
10.8 6.6 0.0 9.6
GVHD, graft-versus-host disease; PPV, positive predictive value.
for the diagnosis of upper GI GVHD even without preceding or accompanying GVHD of other organs. Biopsies are essential to diagnose GVHD, and the common biopsy sites are the skin, liver, and GI tract. While the diagnostic yield of skin biopsies has been reported to be 46%, that of upper or lower GI biopsies is higher (58%), indicating the usefulness of GI endoscopy in diagnostic biopsies,7 with the diagnostic yield of upper GI endoscopy reported as being higher than that of lower GI endoscopy.11,12 In this study, patients with ≥ grade III GVHD did not respond well to steroid therapy (Table 4), suggesting that findings characteristic of ≥ grade III GVHD may be indicative of the efficacy of treatment or prognosis. The number of patients with ≥ grade III GVHD examined in this study was small, but nevertheless use of mucosal exfoliation confirmed by endoscopy had extremely high
specificity and PPV (both 100%). Indeed, only 50.0% of patients who presented with mucosal exfoliation responded to treatment. Early therapeutic intervention, without waiting for pathological examination of biopsy specimens, may be necessary for these patients. There are several limitations to this study. First, this was a retrospective single-center study examining a limited number of patients. Second, one endoscopist was responsible for endoscopic interpretation. Lastly, we performed endoscopy in selective, but not all, patients who had undergone allogeneic hematopoietic stem cell transplantation and thus sampling bias exists. Because examination was limited to symptomatic patients, it was impossible to compare them with non-symptomatic patients. A prospective study with a large sample size is warranted. In conclusion, we examined 115 patients who underwent upper GI endoscopy within 100 days of allogeneic hematopoietic stem cell transplantation at our hospital. GVHD was histologically diagnosed in 9.8% of all recipients (n = 868). Redness, luster, and mucosal change were common endoscopic findings in the stomach of upper GI GVHD patients. It was found that patients presenting with mucosal exfoliation, an infrequent endoscopic finding, are likely to have ≥ grade III GVHD. Even when endoscopic findings of the duodenum were normal, GVHD of the upper GI was frequently diagnosed, suggesting that biopsies are necessary for definitive diagnosis.
References 1 Ferrera JL, Levine JE, Reddy P, Holler E. Graft-versus-host disease. Lancet 2009; 373: 1550–61. 2 Ferrara JL, Deeg HJ. Graft-versus-host disease. N. Engl. J. Med. 1991; 324: 667–74. 3 Terdiman JP, Linker CA, Ries CA, Damon LE, Rugo HS, Ostroff JW. The role of endoscopic evaluation in patients with suspected intestinal graft-versus-host disease after allogeneic bone marrow transplantation. Endoscopy 1996; 28: 680–5. 4 Ponec RJ, Hackman RC, McDonald GB. Endoscopic and histologic diagnosis of intestinal graft-versus-host disease after marrow transplantation. Gastrointest. Endosc. 1999; 49: 612–21.
Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872 © 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
1871
Acute upper GI graft-versus-host disease
K Nomura et al.
5 Mashiba T, Usami A, Soga Y et al. Endoscopic findings of intestinal graft-versus-host disease following allogenic bone marrow and peripheral blood stem. Cells Transplant. Gastroenterol. Endosc. 2003; 45: 929–33. 6 Sale GE, Shuman HM, McDonald GB, Thomas ED. Gastrointestinal graft-versus-host disease in man. A clinicopathologic study of the rectal biopsy. Am. J. Surg. Pathol. 1979; 3: 291–9. 7 McDonald GB, Shulman HM, Sullivan KM, Spencer GD. Intestinal and hepatic complications of human bone marrow transplantation. Part I. Gastroenterology 1986; 90: 460–77. 8 Weisdorf DJ, Snover DC, Haake R et al. Acute upper gastrointestinal graft-versus-host disease: clinical significance and response to immunosuppressive therapy. Blood 1990; 76: 624–9.
1872
9 Przepiorka D, Weisdorf DJ, Martin P et al. 1994 Consensus conference on acute GVHD grading. Bone Marrow Transplant. 1995; 15: 825–8. 10 Kakugwa Y, Mineishi S, Saito D, Fujii T, Takaue Y, Shimoda T. The features and prognostic significance of endoscopic findings in gastrointestinal tract graft versus host disease after hematopoietic stem cell transplantation. Blood 2002; 100: 444B–5. 11 Roy J, Snover D, Weisdorf S, Mulvahill A, Filipovich A, Weisdorf D. Simultaneous upper and lower endoscopic biopsy in the diagnosis of intestinal graft-versus-host disease. Transplantation 1991; 51: 642–6. 12 Cox GJ, Matsui SM, Lo RS et al. Etiology and outcome of diarrhea after marrow transplantation: a prospective study. Gastroenterology 1994; 107: 1398–407.
Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872 © 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
Copyright of Journal of Gastroenterology & Hepatology is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
doi:10.1111/jgh.12651
GASTROENTEROLOGY
Clinicopathological features of patients with acute graft-versus-host disease of the upper digestive tract Kosuke Nomura,* Toshiro Iizuka,* Daisuke Kaji,* Hisashi Yamamoto,† Yasutaka Kuribayashi,* Ryusuke Kimura,* Akihiro Yamada,* Tsukasa Furuhata,* Satoshi Yamashita,* Daisuke Kikuchi,* Akira Matsui,* Toshifumi Mitani,* Osamu Ogawa,* Shu Hoteya,* Yasunori Ota,‡ Shuichi Taniguchi† and Mitsuru Kaise* Departments of *Gastroenterology and †Hematology, Toranomon Hospital, and ‡Department of Pathology, Research Hospital, The Institute of Medical Science, The University of Tokyo, Tokyo, Japan
Key words endoscopy, graft-versus-host disease, human stem cell transplantation, upper gastrointestinal tract. Accepted for publication 7 April 2014. Correspondence Dr Kosuke Nomura, Department of Gastroenterology, Toranomon Hospital, 2-2-2 Toranomon, Minato-ku, Tokyo 105-8470, Japan. Email: [email protected]
Abstract Background and Aims: Acute graft-versus-host disease (GVHD) occurring within 100 days post-transplant is one of critical factors influencing prognosis in transplant recipients. Among cases of GVHD of the gastrointestinal (GI) tract, GVHD rarely affects the upper GI. In this study, we retrospectively examined the frequency of upper GI GVHD and diagnostic accuracy. Patients and Methods: From among 868 patients who underwent allogeneic hematopoietic stem cell transplantation at our hospital between January 2005 and June 2012, 115 of whom underwent biopsy for upper GI symptoms. The endoscopic findings and histologic diagnosis from these 115 patients were retrospectively analyzed. Results: GVHD was histologically diagnosed in 85 patients overall (9.8% of all 868 transplant recipients). Although gastric mucosal exfoliation was not commonly observed, this endoscopic finding when used as a diagnostic predictor had both a specificity and positive predictive value (PPV) of 100%. When using redness, luster, and mucosal change as predictors, specificity and PPV were relatively high, suggesting that these gastric endoscopic findings are useful in the diagnosis of upper GI GVHD. Among the duodenal endoscopic findings, erosion as a diagnostic predictor had both a high specificity and PPV. The biopsy results often lead to a diagnosis of GVHD even in cases judged to be endoscopically normal. Conclusions: Among the gastric endoscopic findings, mucosal exfoliation, although rare, and redness, luster, and mucosal change are likely to be useful diagnostic predictors of upper GI GVHD. GVHD was frequently diagnosed in patients with endoscopically normal duodenum, suggesting that biopsies are important for definitive diagnosis.
Introduction Graft-versus-host disease (GVHD) results from immune responses where donor lymphocytes recognize histocompatibility antigens of the host (recipient) as foreign and launch an attack. Acute GVHD is defined as that occurring within the first 100 days posttransplant1 and it targets mainly the liver, gastrointestinal (GI) tract, and skin. On the other hand, chronic GVHD occurs more than 100 days post-transplant and its clinical symptoms resemble those of autoimmune disease. Chronic GVHD targets multiple organs (e.g. lungs, skin, and oral cavity), but lesions in the GI tract are rare. Both forms of GVHD are major complications that affect the prognosis of allograft recipients, and thus early diagnosis and treatment are necessary.2 However, GVHD can be complicated by various types of infections after hematopoietic stem cell transplan-
tation and by disorders caused via non-immune mechanisms, making differential diagnosis based solely on clinical symptoms difficult.3 GVHD of the digestive tract is diagnosed based on histological findings of biopsy specimens, which are obtained based on various endoscopic findings. The endoscopic features characteristic of GVHD of the lower GI tract, such as mucosal exfoliation and honeycomb-like patterns, have been reported in several studies.4,5 Many of these endoscopic findings are in good agreement with histological diagnosis, and thus identification of lesions associated with lower GI GVHD is not difficult. In contrast, endoscopic findings indicative of upper GI GVHD have not been well studied. Therefore, in this study, we retrospectively examined the frequency of upper GI GVHD that occurred within 100 days of hematopoietic stem cell transplantation and investigated its relationship with endoscopic findings.
Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872 © 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
1867
Acute upper GI graft-versus-host disease
K Nomura et al.
were used for GVHD diagnosis. A platelet count of ?50 000/mm3 was maintained in all patients before biopsy.
Methods Patients. Among 868 patients who had undergone allogeneic hematopoietic stem cell transplantation at our hospital between January 2005 and June 2012, acute GVHD of the digestive tract was suspected in 377 patients based on their clinical symptoms. Among them, 135 patients presented with upper GI symptoms, such as epigastralgia, vomiting, and loss of appetite, and underwent endoscopic examination. Histological diagnosis of biopsy specimens was performed in 115 of the 135 patients, and these 115 patients were the subjects of this retrospective study (Fig. 1). When upper GI endoscopy was performed twice or more within 100 days post-transplant, results of the first examination, not the later examinations, were analyzed in this study.
Endoscopic examination. Endoscopic findings were retrospectively determined for each patient as follows: for the stomach, linear redness, spotted redness, luster, erosion, ulceration, mucosal change, and mucosal exfoliation; and for the duodenum, redness, erosion, and ulceration (Fig. 2). The sensitivity, specificity, and positive predictive value (PPV) of each finding were then calculated against the histological diagnosis of GVHD, and also against that of ≥ grade III GVHD. Pathologic diagnosis of GVHD was performed according to the classification of GVHD severity proposed by Sale et al.6 (Table 1). When multiple biopsy specimens were obtained from the stomach and duodenum during a single endoscopic examination, the highest severity grade from among the specimens was used for this study. Because the diagnosis of GVHD is difficult in cytomegalovirus-positive specimens, biopsy specimens that did not indicate cytomegalovirus infection
Results Subjects were 66 men and 49 women (mean age 53.5 years; range, 19–74 years). The mean time of endoscopic examination was 53.9 days post-transplant (range, 20–99 days). Underlying diseases were acute leukemia (n = 67), malignant lymphoma (n = 23), myelodysplastic syndrome (n = 10), chronic myelogenous leukemia (n = 4), adult T-cell leukemia (n = 3), and other (n = 8). The sources of stem cells for transplantation were umbilical cord blood (n = 69), bone marrow (n = 30), and peripheral blood from related donors (n = 16). All patients received preventive therapy for GVDH with methotrexate (MTX) plus tacrolimus (FK506) or cyclosporine A (CyA) (n = 47), FK506 plus mycophenolate mofetil (MMF) (n = 41), FK506 or CyA only (n = 26), or other agents (n = 1) (Table 2). Non-steroidal anti-inflammatory drugs and aspirin, both of which may damage the mucosa, were orally administered to four and two patients, respectively, and a proton pump inhibitor or histamine H2-receptor antagonist was orally administered to all of the six patients. Biopsy specimens were taken from 3.3 sites/patient on average, and GVHD was histologically diagnosed in 85 patients, which accounted for 9.8% of all transplant recipients (n = 868) and 73.9% of the subjects of this study (n = 115). Fifty-one subjects (44.3%) had histologic severity grade I GVHD, while 22 (19.1%), 9 (7.8%), and 3 subjects (2.6%) had grade II, III, and IV GVHD, respectively. Complication by cytomegalovirus infection was found in 15 subjects (13.0%), 9 with GVHD, and 6 without GVHD.
Figure 1 Flow diagram of the study. CMV, ; GI, gastrointestinal; GVHD, graft-versus-host disease; HHSCT, homologous hematopoietic stem cell transplantation.
1868
Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872 © 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
K Nomura et al.
Acute upper GI graft-versus-host disease
a
b
c
d
e
f
g
h
i
j
k
Figure 2 Examples of each endoscopic finding. (a) Linear redness; (b) spotted redness; (c,j) erosion; (d) ulceration; (e) luster; (f,g) mucosal change; (h) mucosal exfoliation; (i) redness; (k) ulcer.
Table 1 Definition and histological grading system for acute gastrointestinal graft-versus-host disease Grade
Histological feature
I II
Single cell necrosis (apoptosis) noted on medium power Evidence of epithelial damage by crypt/glandular abscesses epithelial flattening or glandular/crypt dilation Dropout of one or more crypts/glands Total epithelial denudation
III IV
Multiple endoscopic findings were observed in many subjects. In the stomach, two findings were observed during a single endoscopic examination in 36 subjects, while three findings and four findings were observed in 35 and 21 subjects, respectively. In the duodenum, two findings were observed in 13 subjects, while three findings were seen in one subject. The most common gastric finding was erosion, which was present in 51.3% (59/115) of subjects. As for the other findings, luster was present in 50.4% (n = 58) of subjects, mucosal change in 49.6% (n = 57), linear redness in 27.8% (n = 32), spotted redness in 18.3% (n = 21), ulceration in 5.2% (n = 6), and mucosal exfoliation in 3.5% (n = 4). No findings were observed in 4.3%
(n = 5) of the subjects. The most common duodenal finding was redness, which was found in 32.2% (37/115) of subjects, followed by erosion and ulceration observed in 13.0% (15/115) and 2.6% (3/115) of subjects, respectively. Endoscopy did not detect abnormalities in the duodenal mucosa in 63.5% (73/115) of subjects. Comparisons of subjects with and without GVHD revealed no differences in individual findings; however, a significant difference was observed between the two groups when no abnormalities were evident in the stomach (Table 3). In the 85 patients histologically diagnosed with upper GI GVHD, when luster, mucosal change, and erosion were used as endoscopic gastric predictors, approximate diagnostic sensitivity was 50%, but specificity was high with mucosal exfoliation (100%), and erosion and linear/spotted redness (≥ 80%). The PPV was 100% when using gastric mucosal exfoliation as a predictor and was ≥ 75% when linear/spotted redness, mucosal change, and luster were used. No abnormalities were found in five patients, none of whom had GVHD; in other words, abnormal endoscopy findings were present in all GVHD patients (Table 3). Although data were re-evaluated using different combinations of endoscopic findings, again no significant differences were observed, indicating that none of the combinations were specific to GVHD. When
Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872 © 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
1869
Acute upper GI graft-versus-host disease
Table 2
K Nomura et al.
Patient characteristics
No. Mean age, years (range) Sex Male Female Mean days post transplantation Primary disease Acute leukemia Malignant lymphoma Myelodysplastic syndrome Chronic myelogenous leukemia Adult T-cell leukemia Others Stem cell source Cord blood stem cell transplantation Bone-marrow transplantation Allo-peripheral blood stem cell transplantation Previous treatment FLU-based regimen CY/TBI or CY/BU Preservation of GVHD CyA/FK506 + MTX FK506 + MMF CyA or FK506 Other
115 53.5 (19–74) 66 49 53.9 (20–99) 67 23 10 4 3 8 69 30 16 105 10 47 41 26 1
BU, busulfan; CY, cyclophosphamide; CyA, cyclosporin A; FK506, tacrolimus; FLU, fludarbine; MMF, mycophenolate mofetil; MTX, methotrexate; TBI, total body irradiation.
diagnostic use of gastric findings was re-examined in 12 subjects with ≥ grade III GVHD, which is conventionally considered nonresponsive, the PPVs for use of mucosal exfoliation and ulceration were 75% and about 50%, respectively, but the PPVs for other gastric findings were low (Table 4). Use of endoscopic duodenal findings as predictors of upper GI GVHD also had low sensitivities, but specificities ≥ 90% were achieved for ulceration and erosion. PPVs were ≥ 75% when erosion, normal mucosa, and redness were used as predictors (Table 3). Different combinations of endoscopic findings were evaluated separately, but again no significant differences were noted. PPVs were ≤ 10% when using any of the duodenal findings of the 12 subjects with ≥ grade III GVHD on histology (Table 4). Treatment was effective in 73.0% (62/85) of GVHD patients. Specifically, treatment was effective in patients with histologic grade I and II GVHD (80.4%, 41/51 and 77.3%, 17/22, respectively), but less effective in patients with histologic grade III GVHD (44.4%, 4/9) and not effective in those with histologic grade IV GVHD (0.0%, 0/3). When the effectiveness of therapy was examined in relation to endoscopic findings, the proportion of responders was low (50.0%) in patients with mucosal exfoliation and gastric ulceration (7/85), but was high (79.7%) in patients without these findings (78/85).
Discussion Acute GVHD occurs in 30–60% of patients who undergo allogeneic bone marrow transplantation.7 It mainly damages the liver, GI 1870
tract, and skin, and it is characterized clinically primarily by maculopapular rash, nausea/vomiting, diarrhea, and jaundice. GI lesions in acute GVHD have been evaluated in patients with lower GI symptoms, in other words, with severe diarrhea. However, some GVHD patients present only with upper GI symptoms (e.g. nausea and vomiting), not diarrhea. Weisdorf et al. propose that upper GI GVHD is a disease specifically targeting the upper GI tract and is thus distinct from lower GI GVHD.8 It was reported that upper GI GVDH occurs in 10–26% of allogeneic transplant recipients in Western countries,9 compared with 11% (22/205) of Japanese recipients.10 In the present study, upper GI GVHD was diagnosed in 9.8% (85/868) of all of the recipients, which is in a good agreement with previous report. In upper GI GVHD, lesions in the esophagus are extremely rare, but the stomach and duodenum are susceptible sites.3 Although previous studied have reported various endoscopic findings as characteristic of GVHD, such as mucosal edema, spotted redness, erosion, ulceration, and bleeding, the number of cases studied was small. In a study of 34 upper GI GVHD patients by endoscopy, Roy et al. found gastric inflammation such as redness and erosion was the most common finding, observed in 65% of patients, and that mucosal abnormalities were absent in 18% of patients.11 Terdiman et al. also reported that nonspecific findings such as redness and erosion were the most commonly findings in the 12 patients they assessed.3 In the present study, we examined 115 subjects and found that when gastric mucosal exfoliation, despite not being a common endoscopic finding, was used as a diagnostic predictor, both specificity and PPV were 100%. When using redness (linear or spotted), luster, and mucosal change as predictors, specificities and PPV were also relatively high, albeit not significantly. These results suggest that these endoscopic findings will be useful in the diagnosis of upper GI GVHD. This study also examined mucosal change, which has not been well examined in previous studies. Among 115 subjects, 57 presented with honeycomb-like, granular, and wrinkling changes in the gastric mucosa. In good agreement with the results in lower GI GVHD reported by Mashiba et al.,5 our histologic examination confirmed edema of the lamina propria and deceases in crypts, which is likely to explain the above endoscopic findings. These mucosal changes are useful findings, with a specificity of 63.3% and a PPV of 80.7%, although sensitivity was low. Among endoscopic findings in the duodenum, erosion had high specificity and high PPV, but low sensitivity as a diagnostic predictor. Unexpectedly, the biopsy findings of mucosal specimens with no apparent abnormalities were similar to those of erosion. This is in clear contrast to the gastric findings that none of the five patients without mucosal abnormalities had GVHD. As abnormalities are usually found in the stomach in upper GI GVHD, gastric biopsy is recommended in such cases. On the other hand, the duodenum was often endoscopically normal, although the reason is not known, suggesting that these organs need to be examined individually. Acute GVHD occurs predominantly in the skin, liver, and GI tract. Occurrence of lower GI GVHD is usually preceded by, or accompanied by, skin GVHD. Upper GI GVHD, on the other hand, often occurs without GVHD affecting other organs. For example, according to a report by Weisdorf et al., approximately 30% of upper GI GVHD patients did not have GVHD affecting other organs,8 similar to approximately 35.3% of our patients (n = 30). Thus, upper GI endoscopic examination is encouraged Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872
© 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
K Nomura et al.
Table 3
Acute upper GI graft-versus-host disease
Summary of outcomes
Total Stomach Linear redness Spotted redness Erosion Ulceration Luster Mucosal change Mucosal exfoliation No abnormality Duodenum Redness Erosion Ulcer No abnormality
n
GVHD
Non-GVHD
P-value
Sensitivity (%)
Specificity (%)
PPV (%)
115
85
30
32 21 59 6 58 57 4 5
26 18 41 4 46 46 4 0
6 3 18 2 12 11 0 5
0.27 0.17 0.27 0.68 0.18 0.10 0.23 < 0.01
30.6 21.2 48.2 4.7 54.1 54.1 3.5 0
80.0 90.0 40.0 93.3 60.0 63.3 100 83.3
81.3 85.7 69.4 66.7 79.3 80.7 100 0
37 15 3 73
27 12 1 55
7 3 2 18
0.38 0.56 0.10 0.65
31.8 14.1 1.2 64.7
66.7 90.0 93.3 40.0
73.0 80.0 33.3 75.3
GVHD, graft-versus-host disease; PPV, positive predictive value.
Table 4
Summary of outcomes; grade III or IV
Total Stomach Linear redness Spotted redness Erosion Ulceration Luster Mucosal change Mucosal exfoliation No abnormality Duodenum Redness Erosion Ulcer No abnormality
n
GVHD (grade III or IV)
Sensitivity (%)
Specificity (%)
PPV (%)
115
12
32 21 59 6 58 57 4 5
3 1 7 3 6 6 3 0
25.0 8.3 58.3 25.0 50.0 50.0 25.0 0
71.8 80.6 49.5 97.1 49.5 50.5 99.0 83.3
9.4 4.8 6.8 50.0 10.3 10.5 75.0 0
37 15 3 73
4 1 0 7
33.3 8.3 0 58.3
68.0 86.4 100 35.9
10.8 6.6 0.0 9.6
GVHD, graft-versus-host disease; PPV, positive predictive value.
for the diagnosis of upper GI GVHD even without preceding or accompanying GVHD of other organs. Biopsies are essential to diagnose GVHD, and the common biopsy sites are the skin, liver, and GI tract. While the diagnostic yield of skin biopsies has been reported to be 46%, that of upper or lower GI biopsies is higher (58%), indicating the usefulness of GI endoscopy in diagnostic biopsies,7 with the diagnostic yield of upper GI endoscopy reported as being higher than that of lower GI endoscopy.11,12 In this study, patients with ≥ grade III GVHD did not respond well to steroid therapy (Table 4), suggesting that findings characteristic of ≥ grade III GVHD may be indicative of the efficacy of treatment or prognosis. The number of patients with ≥ grade III GVHD examined in this study was small, but nevertheless use of mucosal exfoliation confirmed by endoscopy had extremely high
specificity and PPV (both 100%). Indeed, only 50.0% of patients who presented with mucosal exfoliation responded to treatment. Early therapeutic intervention, without waiting for pathological examination of biopsy specimens, may be necessary for these patients. There are several limitations to this study. First, this was a retrospective single-center study examining a limited number of patients. Second, one endoscopist was responsible for endoscopic interpretation. Lastly, we performed endoscopy in selective, but not all, patients who had undergone allogeneic hematopoietic stem cell transplantation and thus sampling bias exists. Because examination was limited to symptomatic patients, it was impossible to compare them with non-symptomatic patients. A prospective study with a large sample size is warranted. In conclusion, we examined 115 patients who underwent upper GI endoscopy within 100 days of allogeneic hematopoietic stem cell transplantation at our hospital. GVHD was histologically diagnosed in 9.8% of all recipients (n = 868). Redness, luster, and mucosal change were common endoscopic findings in the stomach of upper GI GVHD patients. It was found that patients presenting with mucosal exfoliation, an infrequent endoscopic finding, are likely to have ≥ grade III GVHD. Even when endoscopic findings of the duodenum were normal, GVHD of the upper GI was frequently diagnosed, suggesting that biopsies are necessary for definitive diagnosis.
References 1 Ferrera JL, Levine JE, Reddy P, Holler E. Graft-versus-host disease. Lancet 2009; 373: 1550–61. 2 Ferrara JL, Deeg HJ. Graft-versus-host disease. N. Engl. J. Med. 1991; 324: 667–74. 3 Terdiman JP, Linker CA, Ries CA, Damon LE, Rugo HS, Ostroff JW. The role of endoscopic evaluation in patients with suspected intestinal graft-versus-host disease after allogeneic bone marrow transplantation. Endoscopy 1996; 28: 680–5. 4 Ponec RJ, Hackman RC, McDonald GB. Endoscopic and histologic diagnosis of intestinal graft-versus-host disease after marrow transplantation. Gastrointest. Endosc. 1999; 49: 612–21.
Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872 © 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
1871
Acute upper GI graft-versus-host disease
K Nomura et al.
5 Mashiba T, Usami A, Soga Y et al. Endoscopic findings of intestinal graft-versus-host disease following allogenic bone marrow and peripheral blood stem. Cells Transplant. Gastroenterol. Endosc. 2003; 45: 929–33. 6 Sale GE, Shuman HM, McDonald GB, Thomas ED. Gastrointestinal graft-versus-host disease in man. A clinicopathologic study of the rectal biopsy. Am. J. Surg. Pathol. 1979; 3: 291–9. 7 McDonald GB, Shulman HM, Sullivan KM, Spencer GD. Intestinal and hepatic complications of human bone marrow transplantation. Part I. Gastroenterology 1986; 90: 460–77. 8 Weisdorf DJ, Snover DC, Haake R et al. Acute upper gastrointestinal graft-versus-host disease: clinical significance and response to immunosuppressive therapy. Blood 1990; 76: 624–9.
1872
9 Przepiorka D, Weisdorf DJ, Martin P et al. 1994 Consensus conference on acute GVHD grading. Bone Marrow Transplant. 1995; 15: 825–8. 10 Kakugwa Y, Mineishi S, Saito D, Fujii T, Takaue Y, Shimoda T. The features and prognostic significance of endoscopic findings in gastrointestinal tract graft versus host disease after hematopoietic stem cell transplantation. Blood 2002; 100: 444B–5. 11 Roy J, Snover D, Weisdorf S, Mulvahill A, Filipovich A, Weisdorf D. Simultaneous upper and lower endoscopic biopsy in the diagnosis of intestinal graft-versus-host disease. Transplantation 1991; 51: 642–6. 12 Cox GJ, Matsui SM, Lo RS et al. Etiology and outcome of diarrhea after marrow transplantation: a prospective study. Gastroenterology 1994; 107: 1398–407.
Journal of Gastroenterology and Hepatology 29 (2014) 1867–1872 © 2014 Journal of Gastroenterology and Hepatology Foundation and Wiley Publishing Asia Pty Ltd
Copyright of Journal of Gastroenterology & Hepatology is the property of Wiley-Blackwell and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.
