Clinical Features and Treatment of Hematopoietic Stem Cell Transplantation- Associated Gastric Antral Vascular Ectasia

Clinical features and treatment of hematopoietic stem cell transplantation- associated gastric antral vascular ectasia

K. Ohashi1, M. Sanaka2,Y.Tu2, N. Egawa2, K. Ohashi3, N. Funata3, Y. Okuyama4, K. Hiruma4, Y. Tanaka1, S. Mori1, H. Akiyama1 andH. Sakamaki 1

1Hematology Division, Tokyo Metropolitan Komagome Hospital, Tokyo, Japan; 2Gastroenterology Division, Tokyo Metropolitan Komagome Hospital, Tokyo, Japan; 3Pathology Division, Tokyo Metropolitan Komagome Hospital, Tokyo, Japan; and 4Blood Transfusion Service, Tokyo Metropolitan Komagome Hospital, Tokyo, Japan

Summary: showing capillary dilation, thromboses, and fibromuscular hyperplasia. Tobin et al3 reportedthat the male gender, Gastric antral vascular ectasia (GAVE) may occur after busulfan containing conditioning regimens, and veno- hematopoietic stem cell transplantation (HSCT) and occlusive disease (VOD) of the liver, are correlated with cause severe and prolonged gastric bleeding. The under- the development of HSCT-GAVE, yet its etiology still lying pathology of transplant-associated GAVE (HSCT- remains obscure andstandardtreatment has not been GAVE) is poorly understood and an effective therapeutic established. In the present study, we retrospectively strategy has not been established yet. We retrospectively reviewedthe clinical charts of 230 consecutive allogeneic reviewed the medical records of 230 consecutive allogeneic transplant recipients for evidence of HSCT-GAVE. transplant recipients in our institution between January 1997 and June 2002. We identified five patients who developed HSCT-GAVE (2.2%). Four patients had Patients and methods bleeding from HSCT-GAVE and one patient had HSCT-GAVE discovered incidentally. The clinical fea- We retrospectively reviewed230 patients (141 males, 89 tures of these patients were similar in that they all received females; median age 35 years, range 15–58) with hemato- conditioning treatment with busulfan and had history of logical diseases who underwent allogeneic transplantation thrombotic microangiopathy. Furthermore, treatment (201 bone marrow transplantations: 78 related, 123 with a beta-blocker apparently improved the outcome of unrelated; 25 related peripheral blood stem cell transplan- HSCT-GAVE in three patients. tations; four unrelatedcordbloodstem cell transplanta- Bone Marrow Transplantation (2003) 32, 417–421. tions) in our institution between January 1997 andJune doi:10.1038/sj.bmt.1704137 2002. Underlying diseases were chronic myeloid leukemia Keywords: gastric antral vascular ectasia; hematopoietic (n ¼ 60), acute nonlymphoidleukemia ( n ¼ 65), acute stem cell transplantation; thrombotic microangiopathy; lymphoidleukemia ( n ¼ 56), myelodysplastic syndrome beta-blocker; busulfan (n ¼ 22), severe aplastic anemia (n ¼ 11), non-Hodgkin’s lymphoma (n ¼ 9), myelofibrosis (n ¼ 5), multiple myeloma (n ¼ 1), andEwing’s sarcoma ( n ¼ 1). Preparative therapy Gastrointestinal complications are common following was given basically according to the primary disease and hematopoietic stem cell transplantation (HSCT);1 however, type of transplant. Generally, patients with lymphoid gastric antral vascular ectasia (GAVE) is not well malignancy were mainly conditioned with a total body characterizedin terms of its association with HSCT and irradiation (TBI, 12 Gy) containing regimen, including only a few reports of GAVE in transplant recipients are cytarabine (8 g/m2) andcyclophosphamide(120 mg/kg). available. HSCT-GAVE was first described by Marmaduke On the other hands, patients with myeloid malignancy were et al 2 in 10 patients who receivedallogeneic bone marrow conditioned with a non-TBI containing regimen that included transplantation. Although the incidence of HSCT-GAVE busulfan (16 mg/kg) andcyclophosphamide(120 mg/kg). remains uncertain, bleeding, characterized as profuse and Plasma concentration of busulfan was not monitored. recurrent from hemorrhagic mucosa in the antrum,3 occurs 20–90 days post-transplant. GAVE is diagnosed by the endoscopic appearance of gastric antrum and biopsies Results

Diagnosis of HSCT-GAVE Correspondence: Dr K Ohashi, Hematology Division, Tokyo Metropo- litan Komagome Hospital, 3-18-22 Honkomagome, Bunkyo-ku, Tokyo, Of the 230 transplant recipients, 30 patients with signs of 113-8677, Japan gastrointestinal symptoms underwent upper endoscopic Received15 November 2002; accepted18 March 2003 examination andfive patients (2.2%) were ultimately Clinical features and treatment of HSCT-GAVE K Ohashi et al 418 diagnosed with HSCT-GAVE, a median of 95 days (range, a 65–208 days) after transplantation. All five patients receivedupper endoscopyon occasion of hematemesis (n ¼ 3), melena (n ¼ 1), or epigastric pain (n ¼ 1) (Table 1). Two patients developed hypovolemic shock at the onset of GAVE. Diagnosis of HSCT-GAVE was basedon the endoscopic appearance of red patches or spots in a diffuse array in the antrum of the stomach5 (Figure 1a). Two patients (patient nos. 1 and5) underwent endoscopic biopsy for histological diagnostic proof, which showed compatible pathological findings such as dilatation of mucosal capillaries with focal thrombosis andectasia (Figure 1b andc). In three patients (patient nos. 2, 3, and4), thrombocytopenia preventedus from obtaining endoscopic biopsy specimens; however, the obvious endoscopic characteristics of GAVE excluded the possibilities of antral ulceration, erosion, infection, or graft-versus-host disease. b Clinical feature of HSCT-GAVE The patients’ characteristics are shown in Table 2. The group was composedby four males andone female with a median age of 51 years (range, 32–51 years). Their underlying diseases were CML (n ¼ 3), myelodysplastic syndrome (n ¼ 1), andacute myeloidleukemia ( n ¼ 1). All were conditioned with a busulfan-containing regimen and receivedbone marrow from unrelateddonors( n ¼ 4) or a relateddonor( n ¼ 1). The incidence of GAVE in patients with busulfan regimen (n ¼ 134) was higher than that in patients with non-busulfan regimen (n ¼ 96): five patients (3.7%) andzero patients (0%), respectively ( P ¼ 0.0057). HSCT-GAVE has been associatedwith hepatic VOD; 2 however, in this group only patient no.1 hada prior VOD episode. Although one patient (patient no. 3) developed c ascites 1 month before the onset of GAVE, his clinical symptoms did not meet the VOD criteria.6 Rather, all patients exhibitedthe typical signs of thrombotic microangiopathy (TMA) 20–53 days (median 33 days) before the onset of GAVE (Table 2). Clinical diagnosis of TMA was made by the presence of intravascular hemolysis in association with redcell fragmentation on the peripheral blood smear. Individual laboratory findings are shown in Table 2. Elevatedserum LDH with a decreasein hemoglobin andincreasedreticulocytes, as well as a dropin the platelet count, suggests the presence of microangio- pathic hemolytic anemia.

Treatment of HSCT-GAVE

Effective treatment for patients with HSCT-GAVE has not Figure 1 (a) Characteristic endoscopic appearance in HSCT-GAVE. been established as yet: the bleeding did not respond to Fresh bloodactively oozes from patchy erythema developing in the entire omeprazole andplatelet transfusion as in our cases antrum. (b)and(c) Endoscopic biopsy of antral mucosa in HSCT-GAVE shows ectatic subepithelial capillaries (B, arrow) andﬁbrin thrombus in (Table 1). We initially pursueda therapeutic approach to superﬁcial mucosal capillaries (C, arrowhead). reduce the symptoms because of complication of portal hypertension in one patient (patient no. 1). The patient underwent abdominal angiography during a bleeding corpus. She was administered the cardioselective beta- episode, and the hepatic venous pressure gradient measured blocker metoprolol at 60 mg/day orally in three-divided during procedure was approximately 13 mmHg. The doses.7 A cardioselective beta-blocker was chosen because diagnosis was HSCT-GAVE, rather than portal hyperten- of her history of bronchial asthma. Her bleeding stopped or sive gastropathy, due to the typical endoscopic appearance slowed appreciably within a couple of days due to the of diffuse red spots in the antrum but not in the fundus or greatly reduced red blood cell transfusion requirement.

Bone Marrow Transplantation Table 1 Clinical feature of HSCT-GAVE andresponse to treatment

Patient GAVE no. Day of Initial Ptt counts at Duration of LFT at onset of GAVE Treatment Resolution of Outcome onset symptom onset of bleeding (days) GAVE bleeding ( Â 104 /ml Method of Amount of Evidence of AST/ALT ALP PT% T bil Initial Metoprolol APC (times Causes diagnosis MAP portal (IUA) (IUA) (mg/dl) intervention (treatment and day of death transfusion hypertension period) performed (units) APC)

1 Day 95 Hematemesis 0.3 Day 94–157 42/43 88 92 0.6 PC transfusion 60 mg — Yes/completea Alive with (64 days) PPI (oral) (day 155–188) hypovolemic Local thrombin shock Tranexamic acid

Visual and 58 Angiography on — histological day 155 HPVG 13 mmHg

2 Day 208 Hematemesis 1.6 Day 208–222 19/17 88 100 0.1 PC transfusion 60 mg — Yes/complete Dead Epigastralgia (15 days) PPI (oral) (day 213–217) (day 236) Local thrombin (day 220–224) Visual 29 No 80 mg Pulmonary (day 225–233) hemorrhage

3 Day 86 Hematemesis 5.7 Day 84–87 36/37 455 85 0.4 PPI (oral) 60 mg — Yes/complete Alive (4 days) (day 86–122)

Visual 12 Yes HSCT-GAVE Ohashi of K treatment and features Clinical

4 Day 118 Melena with 1.1 Day 118–172+b 83/113 430 100 1.7 PC transfusion — Five courses Yes/incomplete Dead

hypovolemic (55days+) PPI (oral) al et shock

Visual 15 No First APC Infection on day 123, (fungal) secondAPC on day 137, thirdAPC on day 146, fourth APC on day 151, ﬁfth APC on day 158

5 Day 65 Epigastric 6.4 No 41/16 207 96 1.0 PPI (oral) — One course Yes/incompletec Alive pain. Severe appetite loss

oeMro Transplantation Marrow Bone Visual and 4No First APC on histological day 96

LFT=liver function test, AST=aspartate aminotransferase (normal range 11–32), ALT=alanine aminotransferase (normal range 7–43), ALP=alkaline phosphatase (normal range 34–110), PT%=prothrombin time% (normal range 82–132), T bll=total bilirubin (normal range 0.2–1.1), PPI=proton pump inhibitor (omeprazole), APC=argon plasma coagulation, PC=platelet concentration. a Complete means that endoscopic appearances after treatment show complete improvement. b Patient (no. 4) was transferredto another hospital after day172, andtherefore there was no subsequent information about his clinical course except for ﬁnal outcome. cPatient (no. 5) refusedto receive a follow-up endoscopy. 419 oeMro Transplantation Marrow Bone 420

Table 2 Characteristics of patients who developed HSCT-GAVE

Patient Age/ sex Type of transplant Stomatitis aGVHD TMA Other condition a no. grade (day of onset) HSCT-GAVE of treatment and features Clinical Diagnosis HLA Disparity Yes/ no Grade Treatment Yes/no Laboratory ﬁndings of onset of TMA Treatment (disease status) (day of (day of onset) onset) Conditioning Affected organ HB Plt RET LDH Cr Fragmentationb GVHD prophylaxis (g/dl) ( Â 104 /ml) (%) (g/dl) (g/dl) (%) 1 32/F Unrelated BMT III (day 50) Yes (day 27) I Local steroid Yes (day 60) 10.1 0.9 85 626 1.9 Not counted FFP VOD (Day 50) CML (CPI) + match (Day 50) Helmet cell (+) CyA tapered BU/CY/TLI Skin CyA/MTX Ohashi K 2 44/M Unrelated BMT IV Yes (day 11) III mPSL Yes (day 176) 7.2 1.2 42 570 1.5 29 FFP CML (CP1) DRB1 mismatch Pulse

BU/CY/TLI Skin ATG al et FKMTX Gut 3 51/M UnrelatedBMT III No Yes (day 41) 10.0 5.8 62 795 1.6 40 FFP Ascites match Heparin (day 50) CML (CP1) BU/CY/TLI CyA/MTX 4 51/M RelatedBMT IV Yes (day 34) II mPSL Yes (day 66) 9.9 0.3 87 648 1.9 31 FFP match AML (CR1) BU/CY Skin CyA tapered CyA/MTX 5 51/M UnrelatedBMT III No Yes (day 46) 8.4 2.7 25 294 1.3 Not counted No match Helmet cell (+) MDS (refractory) BU/CY/TLI CyA/MTX

aGVHD=acute graft-versus-host disease, TMA=thrombotic microangiopathy, RET=reticulocyte (normal range 2–20), LDH= lactate dehydrogenase (normal range 120–220), Cr=creatinine (normal range 0.4–1.0), CML=chronic myeloidleukemia, AML=acute myeloidleukemia, MDS=myelodysplasticsyndrome,CP1=chronic phase 1, CR1=complete remissio n 1, BU=busulfan, CY=cyclophosphamide, TLI=total lymphoidirradiation,MTX=methotrexate, CyA=cyclosporine, FK=FK506, mPSL=methylprednisolone,ATG=antithymocyte globulin, VOD=v eno-occlusive disease. a Stomatitis was graded according to Japanese society of clinical oncology criteria. b Fragmentation=redcell fragmentation on peripheral bloodsmear. Clinical features and treatment of HSCT-GAVE K Ohashi et al 421 Furthermore, metoprolol was taperedoff without re- eventually leadto development of vascular ectasia. The bleeding approximately one month after the start of the fibrin thrombi and/or ectasia in mucosal vessels observed therapy. This successful experience promptedus to treat the histologically in GAVE patients support this hypothesis. next two patients (patient nos. 2 and3) with the same Third, although the pathological role of portal hypertension medication, even though they did not have obvious signs of in the development of classical GAVE is controversial,14 the portal hypertension. Surprisingly, both patients showed documented case of portal hypertension and the efficacy of complete responses following treatment with the short-term the beta-blocker observedin our patients suggests a possible beta-blocker; however, one patient (patient no. 2) devel- association between portal hypertension andHSCT-GAVE. oped rebleeding after the treatment was stopped, and then In conclusion, five of the 230 patients who underwent returned on up to 120 mg/day doses (Table 1). He eventually HSCT developed GAVE at our institution. Development of died on day 236 due to pulmonary hemorrhage without GAVE may be associatedwith busulfan administration, further evidence of melena. All three patients experienced endothelial injury, and portal hypertension. Treatment with no major complication during a prescribed period. Thus, a beta-blocker was well toleratedandall patients achieveda treatment with the short-term beta-blocker appears to have durable response. Although further study of this approach therapeutic potential, but is still questionable. Therefore, is warranted, beta-blocker therapy should be considered as we evaluatedan alternative therapeutic strategy, andtwo a treatment for patients with HSCT-GAVE. patients (patient nos. 4 and 5) underwent endoscopic ablation using argon plasma coagulation (APC)8,9 instead of receiving the short-term beta-blocker. One patient References requiredrepeatedmanipulations even though partial hemostasis was attained, suggesting the limitation of APC 1 Singh C, McDonaldGB. Intestinal andhepatic complications as a single therapeutic tool for HSCT-GAVE. The small of bone marrow andstem cell transplantation. In: Taylor MB, number of patients treatedprohibitedfurther evaluation of Gollan JL, Steer ML, Wolfe MM (eds). Gastrointestinal the efficacy of this new strategy. Emergencies. Williams & Wilkins: Baltimore, 1997, pp 1–21. 2 Marmaduke DP, Greenson JK, Cunningham I et al. Gastric vascular ectasia in patients undergoing bone marrow transplantation. Am J Clin Pathol 1994; 102: 194–198. Discussion 3 Tobin RW, Hackman RC, Kimmey MB et al. Bleeding from gastric antral vascular ectasia in marrow transplant patients. The clinical features seen in our patients are mostly Gastrointest Endosc 1996; 44: 223–229. consistent with those reportedin recent reviews, 2,3 and 4 Burak KW, Lee SS, Beck PL. Portal hypertensive gastropathy may propose several causal factors for the development andgastric antral vascular ectasia (GAVE) syndrome. Gut of HSCT-GAVE. First, five (2.2%) of the 230 patients 2001; 49: 866–872. developed GAVE and all of our patients received oral 5 Gostout CJ, Viggiano TR, Ahlquist DA et al. The clinical and busulfan for myeloablation. It is especially noteworthy that endoscopic spectrum of the watermelon stomach. J Clin no cases of GAVE were observedin patients who Gastroenterol 1992; 15: 256–263. et al conditioned with TBI regimen. The anatomic distribution 6 McDonaldGB, HindsMS, Fisher LD . Veno-occlusive disease of the liver and multiorgan failure after bone marrow of the GAVE lesions suggested direct mucosal damage transplantation: a cohort study of 355 patients. Ann Intern 2 from the oral administration of high-dose busulfan. Med 1993; 118: 255–267. A variety of factors can cause intestinal dysfunction 7 Westaby D, Bihari DJ, Gimson AE et al. Selective andnon- including stress, anxiety, some medications, and meal selective beta receptor blockade in the reduction of portal ingestion.10 Thus, drugs orally given to hospitalized pressure in patients with cirrhosis andportal hypertension. patients may retain in the proximal gut longer than in the Gut 1984; 25: 121–124. distal gut, suggesting that the stomach and duodenum may 8 Gostout CJ, Ahlquist DA, Radford CM et al. Endoscopic laser be more extensively exposedto busulfan than the ileum and therapy for watermelon stomach. Gastroenterology 1989; 96: colon. This is the possible reason why the busulfan-related 1462–1465. GAVE exclusively involves the stomach andproximal 9 Bjorkman DJ, Buchi KN. Endoscopic laser therapy for watermelon stomach. Laser Surg Med 1992; 12: 478–481. duodenum. It also suggests that vascular lesion is not 10 Sanaka M, Kuyama Y, Shimomura Y et al. Gastric emptying of limitedto the gastric antrum, but vascular ectasia can liquids is delayed by co-ingesting solids: a study using salivary extend into the duodenal bulb. In addition, we propose that paracetamol concentrations. JGastroenterol2002; 37: 785–790. high plasma concentrations of busulfan may contribute to 11 Buggia I, Locatelli F, Regazzi MB, Zecca M. Busulfan. Ann the development of HSCT-GAVE. All of our patients Pharmacother 1994; 28: 1055–1062. developed severe mucositis immediately after busulfan 12 Yeager AM, Wagner Jr JE, Graham ML et al. Optimization of administration (Table 2), suggesting higher busulfan levels busulfan dosage in children undergoing bone marrow trans- in patients with GAVE.11,12 Second, all of the patients had plantation: a pharmacokinetic study of dose escalation. Blood experienceda prior TMA episodeandtwo patients devel- 1992; 80: 2425–2428. et al opedascites approximately 1 month before the onset of 13 Catani L, Gugliotta L, Vianelli N . Endothelium and bone marrow transplantation. Bone Marrow Transplant 1996; 17: HSCT-GAVE. These potentially predisposing symptoms 277–280. 13 may reflect endothelial injury and local thrombosis, and 14 Vincent C, Pomier-Layrargues G, Dagenais M et al. Cure of these events might be relatedto the pathogenesis of GAVE. gastric antral vascular ectasia by liver transplantation despite Specially, local endothelial injury and subsequent local persistent portal hypertension: A clue for pathogenesis. Liver thrombosis may obstruct submucosal bloodvessels and Transplant 2002; 8: 717–720.

Bone Marrow Transplantation