GASTROENTEROLOGY 2011;141:1629 1637

Efcacy of Thalidomide for Refractory Gastrointestinal Bleeding From Vascular Malformation

ZHIZHENG GE, HUIMIN CHEN, YUNJIE GAO, WENZHONG LIU, CHUNHONG XU, HONGHONG TAN, HAIYING CHEN, WEI WEI, JINGYUAN FANG, and SHUDONG XIAO
Division of Gastroenterology and Hepatology, Shanghai Jiao-Tong University School of Medicine Renji Hospital, Shanghai Institute of Digestive Disease; Key Laboratory of Gastroenterology & Hepatology, Ministry of Health, (Shanghai Jiao-Tong University) Shanghai, China CLINICAL AT

BACKGROUND & AIMS: Patients with recurrent bleeding from gastrointestinal vascular malformations are a challenge to treat. We investigated the long-term efcacy and safety of thalidomide for refractory bleeding from gastrointestinal vascular malformations in an open-label, randomized study. METHODS: Eligible patients were randomly assigned to groups that were given either 100 mg thalidomide (n 28) or 400 mg iron (n 27, controls), daily for 4 months; patients were followed for at least 1 year (mean, 39 months). Bleeding was dened by a positive result from an immunoassay fecal occult blood test. The primary end point was the effective response rate, dened as the proportion of patients in whom bleeding episodes had decreased by 50% in the rst year of the follow-up period. The secondary end points included the rates of cessation of bleeding, blood transfusion, overall hospitalization, and hospitalization for bleeding. We also quantied yearly bleeding episodes, bleeding duration, levels of hemoglobin, and yearly requirements for transfusions of red cells, numbers of hospitalizations for bleeding, and hospital stays. Plasma levels of vascular endothelial growth factor were measured in the group given thalidomide. RESULTS: Rates of response in the thalidomide and control groups were 71.4% and 3.7%, respectively (P .001). All secondary end points differed significantly different between groups; thalidomide was more effective. No severe adverse effects were observed, although minor side effects were common among patients in the thalidomide group. Levels of vascular endothelial growth factor were signicantly reduced by thalidomide (P .001). CONCLUSIONS: Thalidomide is an effective and relatively safe treatment for patients with refractory bleeding from gastrointestinal vascular malformations. Mechanisms of thalidomide activity might involve vascular endothelial growth factor. Keywords: Treatment; Angiodysplasia; Blood Vessels; Angiogenesis Inhibitor; Clinical Trial. astrointestinal vascular malformation (GIVM), such as angiodysplasia and gastric antral vascular ectasia (GAVE), became an increasingly recognized cause of obscure gastrointestinal bleeding and iron-deciency anemia, particularly in the elderly.13 Approximately 20% of patients present with iron-deciency anemia and/or stools that are intermittently positive for occult blood. Massive

bleeding, which occurs in about 12%27% of patients, needs emergency treatment.4 6 Therefore, patients are often bothered by the endless investigation, frequent hospitalization, repeated blood transfusions, and long-term clinical follow-up, and thus, their quality of life is usually compromised to a certain extent. At present, conventional therapies including angiographic embolization, local endoscopic ablation, and surgical resection for GIVM are often ineffective in preventing recurrent gastrointestinal (GI) bleeding, and may occasionally be associated with severe complications such as ischemic bowel necrosis or even mortality.711 The inefcacy is partially because the exact locations of lesions are usually hard to determine, especially for multicentral small bowel lesions.8 10 Therefore, an effective and relatively safe pharmaceutical agent is urgently needed.12 Although hormonal therapy using estrogen and progestagen was previously considered promising, a recent randomized study found that such a treatment was ineffective.13 Somatostatin and its analog, octreotide, have been reported to reduce blood loss from intestinal angiodysplasias; however, a nal conclusion on its efcacy has not been drawn.14 During the past several years, there have been several case reports demonstrating encouraging clinical outcomes of thalidomide in the prevention of recurrent GI bleeding from angiodysplasia or GAVE.1521 Thalidomide has been reported to exhibit an inhibitory effect against neoangiogenesis by suppressing the expression of vascular endothelial growth factor (VEGF).22 VEGF is strongly expressed within colonic angiodysplasias and may be involved in the pathogenesis of GIVM.23 Thus, it has been proposed that the therapeutic efcacy of thalidomide for angiodysplasia might be attributed to the inhibition of VEGF; however, this hypothesis has never been explored.18 Therefore, inspired by these ndings, we performed an exploratory study to determine the long-term efcacy and
Abbreviations used in this paper: CE, capsule endoscope; EGD, esophagogastroduodenoscopy; GAVE, gastric antral vascular ectasia; GI, gastrointestinal; GIVM, gastrointestinal vascular malformation; iFOBT, immunoassay fecal occult blood test; VEGF, vascular endothelial growth factor. 2011 by the AGA Institute 0016-5085/$36.00 doi:10.1053/j.gastro.2011.07.018

1630

GE ET AL

GASTROENTEROLOGY Vol. 141, No. 5

safety of thalidomide in the prevention of recurrent bleeding due to GIVM and possible role of VEGF.

Methods Study Population

Consecutive patients who were diagnosed with GIVM at the Department of Gastroenterology, Renji Hospital, from November 2003 to November 2007 were eligible for the screening for the study. Patients must have met the following inclusion criteria to enter the study: (1) patients were between 40 and 85 years old; (2) female patients were postmenopausal, with post-tubal ligation, or females with childbearing potential using some form of birth control (eg, contraceptive ring or latex condoms); (3) patients had recurrent or refractory bleeding due to GIVM. Refractory bleeding was dened as bleeding either unresponsive to (ie, recurrent bleeding after at least 3 sessions of esophagogastroduodenoscopy (EGD) or coloscopic therapies, or at least 1 session of double-balloon endoscope therapy or primary medical approaches, such as hormonal or/and somatostatin therapies) or unsuitable for (ie, in the presence of multiple lesions, lesions inaccessible by endoscopic therapy or surgical ectomy, or patients unwilling or unable to undergo surgery due to poor health condition) conventional therapies; and (4), all patients had been conrmed as having GIVM (angiodysplasia and/or GAVE) without obvious infectious, neoplastic, or other specic lesions by endoscopic examination including EGD, colonoscopy, and capsule endoscope (CE) or double-balloon endoscope (which was performed only when CE was negative). Angiodysplasia at endoscopy was characterized by focal or diffused venous/capillary lesions presenting as bright red ectatic vessels or pulsatile red protrusions, with surrounding venous dilatation or patchy erythema with or without oozing, and endoscopic appearance of GAVE was indicated by longitudinal antral folds converging on the pylorus, containing visible columns of tortuous red ecstatic vessels.24,25 Patients were excluded from the study if they had any of the followings: (1) cirrhotic or portal hypertension gastropathy; (2) severe comorbidities of cardiac, pulmonary, renal, liver, hematological, rheumatologic disorders, or uncontrollable diabetes mellitus or hypertension; (3), a history of severe bilateral peripheral neuropathy or seizure activity, thromboembolic disease, known thalidomide or iron allergy, or prior treatment of gastrointestinal bleeding with thalidomide; (4), treatment with any dose of systemic or oral topical corticosteroids or aspirin, nonsteroidal anti-inammatory drugs, antiplatelet drugs, anticoagulants, or Chinese medications (with salicylates), gingko, or echinacea, or other putative immunomodulators or antiangiogenic agents; (5) currently pregnant or lactating; and (6) currently undergoing systemic cancer chemotherapy or receiving radiation.26

of 4. Within each block, the number of patients allocated to each of 2 treatments was equal. Eligible patients were consecutively assigned a random number in chronological order that allocated him or her to receive either thalidomide or iron. Emphatically, all patients continued to be followed up until the last patient had been followed up for 1 year after the treatment, and then the study was completed.

Treatment and Concomitant Therapies

Patients in the thalidomide group received 25 mg thalidomide (Pharmaceutical Co., Ltd. of Changzhou, China) orally 4 times daily for 4 months, and those in the iron-controlled group received 100 mg iron (ferrous succinate tablets; Pharmaceutical Co., Ltd., Nanjing, China) orally 4 times daily at the same time points. Patients assigned to the thalidomide group were refrained from iron ingestion during the 4-month course of treatment, regardless of the hemoglobin levels. During the treatment period, patients were hospitalized and closely monitored for the rst week or a few days longer when required. In the case of an adverse event, the study medication was temporarily interrupted or permanently discontinued based on the seriousness of the event or tolerability of the patient. Concomitant therapies, such as blood transfusions and other symptomatic treatments, were performed in both groups as necessary. During the whole period of the study, blood transfusion (ie, packed red cell transfusion) was indicated and recorded when the hemoglobin level fell below 7.0 g/dL (2 U were administered for 6.1 g/dL hemoglobin 7.0 g/dL, 3 U for 5.1 g/dL hemoglobin 6.0 g/dL, and 4 U for hemoglobin 5.0 g/dL).27 Iron supplementation (100 mg, 4 times per day) was allowed for patients in both groups when the hemoglobin levels fell between 7.1 g/dL and 11.0 g/dL during the 1-year observation period and subsequent follow-up periods.14

CLINICAL AT

Data Collection
Two qualied research nurses collected data on the demographics, medical and family histories at screening, and all clinical data, including any adverse events during the whole study, and 2 qualied doctors were responsible for physical examinations, treatment, and follow-up. During the whole period of study, patients were asked to record the characteristics of their feces (ie, color, volumes, frequency of defecation, and softness) daily in a planned diary. All patients were requested to visit the hospital monthly during the observation and follow-up periods and biweekly during the treatment period for data collection. At each scheduled visit, a physical examination, complete blood counts, and routine iFOBT (using monoclonal colloidal gold color technology) were performed. In patients positive for a routine iFOBT, with or without any signs of overt bleeding, iFOBT was performed daily until there were 3 consecutive days of a negative test, in order to ensure whether or not bleeding truly occurred, whether or not bleeding was stopped, and how long it lasted. In addition, during the whole study period, iFOBT was performed daily in the same way in patients with any signs of overt bleeding (such as hematemesis, melena, hematochezia, or a dramatic drop in hemoglobin). Serum chemical laboratory tests, blood coagulation tests, and hepatic and renal tests were performed at the time of enrollment, randomization, and after the treatment. Neuropathy and other adverse events were assessed monthly since the treatment started. A urine pregnancy test for the -subunit of human chorionic gonadotropin was performed for female patients with childbearing potential at randomization, biweekly

Study Design and Randomization

We performed an open-label, randomized, parallel controlled study that consisted of 3 major periods: a 1-year observation period before randomization, a 4-month treatment period, and a subsequent at least 1-year follow-up period. Recruited patients were entered into the treatment period and randomized only when they had 6 or more bleeding episodes dened by a positive immunoassay fecal occult blood test (iFOBT) in the observation period. Randomization was performed through the proc plan procedure of SAS (SAS Institute Inc, Cary NC) using the method of randomly permuted blocks

November 2011

THALIDOMIDE FOR BLEEDING FROM VASCULAR MALFORMATION

1631

during the treatment period, and 1 month after treatment in the thalidomide group.

Assessment of Efcacy and Safety

The primary objective of the present study was to determine the effective response rate that was dened as the proportion of patients in whom bleeding episodes decreased by 50% within the rst-year of follow-up after treatment, compared with the bleeding episodes in the observation year prior to treatment. The secondary end points included the rates of cessation of bleeding, blood transfusion, overall hospitalization, and hospitalization for bleeding. We also quantied the changes (average values in the follow-up period vs the corresponding values in the observation period) in yearly bleeding episodes, bleeding duration and levels of hemoglobin, yearly requirements for transfusions of red cells, yearly numbers of hospitalizations for bleeding and hospital stay. All these secondary objectives were evaluated at the end of the study. A bleeding outcome was dened as being present by a positive iFOBT, and a bleeding episode was dened as being stopped when there were 3 consecutive days of a negative test after a positive iFOBT. The duration of a bleeding episode was dened as the period of time between the rst day of a positive iFOBT and the rst day of the 3 consecutive days of negative iFOBT. All bleeding episodes described in the present study were only attributed to GIVM without any other possible source of bleeding. In cases with GAVE or angiodysplasia in the duodenum or colon, EGD or colonoscopy was repeated 12 months after the treatment period by the same experienced endoscopist to evaluate the lesion. Adverse events included any unfavorable changes in health, including occurrence or worsening of any clinical symptoms, signs, or abnormal laboratory ndings during the study. Finally, plasma VEGF was measured before and after the 4-month treatment course in the thalidomide group by a technician who was blinded to the study. Peripheral venous blood (3 mL) was collected and centrifuged (3000 rpm at 4C, 10 minutes) and plasma were extracted and stored at 70C for no more than 4 months. Plasma VEGF levels (pg/mL) were determined in duplicate using sandwich enzyme-linked immunosorbent assay kits (R&D Systems, Inc., Minneapolis, MN).

In addition, a sensitivity analysis excluding patients with GAVE was performed to determine whether the results for patients with GIVM remained the same, as the present study mainly addressed the management of small bowel angiodysplasia. Data were reported as mean standard deviation/median (range) for continuous variables and number (%) for categorical variables. Continuous variables were compared using a 2-sample independent t test, paired t test, or Wilcoxon rank-sum test, where appropriate. Categorical variables were compared using 2 test or Fishers exact test. Statistical analysis was performed by a biostatistician blinded to the treatment assignment with the SPSS 13.0 software package (SPSS Inc. Chicago, IL) or SAS 9.2 software (SAS Institute Inc). All reported P values and 95% condence intervals were 2-sided, and a P value .05 was considered statistically signicant.

Results Demographical and Clinical Characteristics

A total of 80 patients conrmed as having GIVM were screened and 2 patients who did not complete the 1-year observation were excluded from the study. Thus, 78 patients were enrolled and completed the 1-year observation, during which 19 patients had 6 bleeding episodes and 59 had 6 bleeding episodes. Of these 59 patients eligible for randomization, 2 developed severe concomitant diseases and had to be withdrawn and another 2 patients refused to continue. Fifty-ve patients were nally randomized to receive thalidomide (n 28) and iron supplementation (n 27). Two patients in the thalidomide group refused to take study medications after 4 weeks of treatment due to slight leukopenia and unexplained somnolence; both were followed up for 12 months after discontinuation of the treatment. One subject in the iron-controlled group was lost to follow-up after 8 months because of personal reasons. A total of 52 patients (26 in each group) completed the study per protocol (Supplementary Figure 1). Of 55 eligible patients, the median age was 58.0 years (range, 40.0 82.0 years) and most (80.0%, 44 of 55) patients were 50 years of age or older. There were 46 (83.6%) female patients; 16 had childbearing potential and were on birth control, and the remaining were either postmenopausal (n 24) or received post-tubal ligation (n 6). Multiple lesions were present in 27 (49.1%) patients, which were found in the jejunum (n 7) or ileum (n 5) or both (n 15). Angiodysplasia was identied in 52 (94.6%) patients by CE (n 48) or subsequent doubleballoon endoscope after a negative CE (n 4). GAVE was identied in 3 (5.5%) patients who had a history of at least 3 failed sessions of endoscopic argon plasma coagulation treatment. Median follow-up period of all 55 patients was 39 months (range, 8 52 months). The demographic baselines were similar between the thalidomide group and the iron-controlled group (Table 1). During the observation year, 85.7% (24 of 28) patients in the thalidomide group and 92.6% (25 of 27) in the control group presented with both occult and overt bleeding, whereas 14.3% (4 of 28) of patients and 7.4% (2 of 27)

Statistical Analysis
To our knowledge, no such similar study has been reported for our reference to determine the sample size. Alternatively, we performed a preliminary unpublished study in which 20 patients (median age, 52 years; range, 4379 years; 16 female, 4 male) with GIVM, median yearly bleeding episodes of 4 (range, 29) received either 4-month thalidomide (n 10) or iron supplementation (n 10). The primary objective (same denition as in the present study) was achieved in 80% and 10%, respectively. An equally divided sample of 11 patients was necessary to give the study a type I error (2-sided) of 5% and a power of 90%. Assuming a drop-off rate of 10% in the study, we needed a minimum sample size of 13 patients per group (calculated with nQuery 5.0). To ensure an adequate power of later stratied analysis, the sample size was approximately doubled to be 26 in each group. Analysis of the responses and adverse events were performed on all randomized patients according to the intention-to-treat principle. Analysis was also performed for the primary end point in patients who completed the study per protocol set.

CLINICAL AT

1632

GE ET AL

GASTROENTEROLOGY Vol. 141, No. 5

Table 1. Demographic and Clinical Characteristics of 55 Patients Before Randomization

Characteristic Age 50b 50b Sexb Male Female Comorbiditiesb Coronary artery heart disease Hypertension Diabetes mellitus Aortic valve stenosis Asthma History of bleeding (y)c Endoscopic technique conrming the diagnosisb EGD CE DBE Unresponsive to endoscopic therapy Patients with GAVE refractory to EGD therapy Patients with GIVM refractory to DBE therapy Unsuitable for DBE therapy Multiple lesions Older than 75 years old Lesions inaccessible to DBE Laboratory parameters Total WBC (109/L) PLT (109/L) GPT (IU/L) Total bilirubin (mol/L) Serum creatinine (mg/dL) PT-INR (s) APTT (s) MCV (fL) MCHC (g/L) (y)a Thalidomide group (n 28) 58.82 12.22/57.5 (4182) 21 (75.0) 7 (25.0) 4 (14.3) 24 (85.7) 2 (7.1) 3 (10.7) 4 (14.3) 2 (7.1) 1 (3.6) 5.57 4.14/5 (121) Iron-controlled group (n 27) 58.96 10.46/58 (4079) 23 (85.2) 4 (14.8) 5 (18.5) 22 (81.5) 2 (7.4) 2 (7.4) 3 (11.1) 2 (7.4) 0 (0.0) 5.04 4.43/3 (120)

CLINICAL AT

2 (7.1) 24 (85.7) 2 (7.1) 2 (7.1) 2 (7.1) 13 (46.4)d 3 (10.7) 9 (21.4) 6.44 1.57/6.2 (4.59.6) 206.45 53.17/194.00 (114.00346.00) 24.81 13.50/26.75 (5.8050.00) 9.16 4.42/8.10 (3.5018.00) 77.55 20.95/78.25 (49.50130.00) 1.05 0.15/1.04 (0.791.43) 27.49 6.44/27.45 (14.6040.00) 76.80 3.97/78.40 (67.8080.70) 295.86 15.26/300.00 (265.00318.00)

1 (3.7) 24 (88.9) 2 (7.4) 1 (3.7) 1 (3.7) 14 (51.9) 1 (3.7) 10 (18.5) 6.59 1.74/6.2 (4.19.7) 197.71 60.09/188.00 (105.00353.00) 25.20 14.11/22.10 (6.4051.60) 9.14 4.12/8.80 (3.0017.80) 88.93 24.27/85.80 (51.10138.50) 1.10 0.15/1.05 (0.911.45) 28.07 5.99/28.60 (12.9041.90) 75.24 4.41/76.50 (67.9080.60) 299.00 13.30/304.00 (266.00319.00)

NOTE. Measurement data are expressed as mean standard deviation/median (range). Categorical data are expressed as n (%). There were no signicant differences in any variables between the 2 groups. APTT, activated partial thromboplastin time; DBE, double-balloon endoscope; GPT, glutamic-pyruvic transaminase; MCV, mean corpuscular volume; MCHC, mean corpuscular hemoglobin concentration; PLT, platelet count; PT-INR, prothrombin time-international normalized ratio; SD, standard deviation; WBC, white blood cell. aData were compared using a 2-sample independent t test. bCategorical variables were compared using 2 test or Fishers exact test. cData were compared using the Wilcoxon rank-sum test. dOne patient with GAVE refractory to EGD therapy was older than 75 years old.

of patients, respectively, in the 2 groups presented only occult bleeding. There was no signicant difference in the bleeding episodes, average bleeding duration (days), hospitalization rate for bleeding, hospitalization number for bleeding, average hospital stay (days), blood transfusion rate, transfused red cell requirements, and average hemoglobin levels between the 2 groups (Table 2).

Clinical Efcacy
According to the intention-to-treat principle, the effective response rate was 71.4% (20 of 28) in the thalidomide group and 3.7% (1 of 27) in the iron-controlled group (P .001) (Table 3). Per-protocol analysis revealed the effective response rate of 76.9% (20 of 26) in the thalidomide group and 3.8% (1 of 26) in the iron-controlled group (P .001).

During the follow-up periods, the rate of cessation of bleeding was 46.4% (n 13) in the thalidomide group and 0% in the iron-controlled group (P .001). The rate of blood transfusions was signicantly decreased from 50.0% (n 14) to 10.7% (n 3) in the thalidomide group, while slightly decreased from 51.9% (n 14) to 48.1% (n 13) in the iron-controlled group (P .002). Yearly bleeding episodes decreased by 9.36 4.31 in the thalidomide group and 1.41 2.74 in the ironcontrolled group (P .001). Average bleeding duration deceased by 5.21 2.96 days in the thalidomide group and 0.81 5.10 days in the iron-controlled group (P .001). Yearly transfused red cell requirements deceased by 1585.71 446.97 mL in the thalidomide group and 28.57 106.90 mL in the iron-controlled group (P

November 2011

THALIDOMIDE FOR BLEEDING FROM VASCULAR MALFORMATION

1633

Table 2. Baseline Characteristics of 55 Patients in 1-Year Observation

Characteristic Bleedinga Bleeding episodes (n) Bleeding duration (days) Hospitalization Hospitalization rates for bleedingb Hospitalization number for bleedinga Hospital stays (days) Transfusion requirements Patients with blood transfusionb Transfused red cell requirements (mL)c,d Average hemoglobin level (g/dL)d
aData

Thalidomide group (n 28) 13.96 3.38/14 (719) 8.82 3.08/9 (520) 28 (100) 3.04 1.00/3 (15) 6.64 2.39/7 (314) 14 (50.0) 1928.57 763.02/1800 (8003400) 6.40 1.79/7.18 (2.699.30)

Iron-controlled group (n 27) 13.96 3.47/14 (721) 8.74 4.50/7 (321) 27 (100) 2.74 1.16/3 (15) 6.63 2.27/6 (312) 14 (51.9) 1985.71 766.47/1900 (8003400) 5.99 1.53/5.60 (3.79.99)

NOTE. Measurement data are expressed as mean standard deviation/median (range). Categorical data are expressed as n (%). were compared using the Wilcoxon rank-sum test. bData were compared using the 2 test. cData were derived from 14 patients in each group who were transfusion-dependent before treatment. dData were compared using a 2-sample independent t test.

.001). Average hemoglobin level was increased by 3.06 2.08 g/dL in the thalidomide group and 0.01 1.32 g/dL in the iron-controlled group (P .001). Other secondary outcomes, including overall hospitalization rate, rate of hospitalization for bleeding, yearly hospitalization number for bleeding, and average hospitalization stay (days) were all decreased signicantly in the thalidomide group compared to the iron-controlled group (all P .001) (Table 3). During the follow-up period, 4-month interval bleeding episodes progressively decreased from 4.90 0.62 to 2.30 0.45 after treatment until month 8, and then remained low afterward in the thalidomide group. However, there was such no change in the iron-controlled group (Figure

1). Furthermore, there were no signicant differences in the sex distribution (males 5.0% [1 of 20], vs 37.5% [3 of 8]; P .058), transfusion dependency (60.0% [12 of 20] vs 25.0% [2 of 8]; P .209), and distribution of the lesions (multiple 50.0% [10 of 20] vs 37.5% [3 of 8]; P .686) between responders (n 20) and nonresponders (n 6) after thalidomide treatment. Three patients with GAVE (2 in the thalidomide group and 1 in the iron-controlled group) were monitored by EGD. Before treatment, GAVE lesions were characterized with raised and reddened areas with distinctly irregular margins in the eld of endoscopic view (Figure 2A). One year after thalidomide treatment, the vascular lesion was signicantly reduced in size or dis-

Table 3. Clinical Efcacy of Thalidomide vs Iron Supplement in the Prevention of GI Bleeding Due to GIVM in 55 Patients (Intention-To-Treat Population)
Variable Primary end point Patients with effective responsea Second end points Patients with cessation of bleedinga Change in yearly bleeding episodesb Change in average bleeding duration (days)b Patients dependent on blood transfusiona Change in yearly transfused red cell requirement (mL)b,c Overall hospitalization rateb Hospitalization rate for bleedingb Change in yearly hospitalization number for bleedingb Change in average hospital stay (days)b Change in average hemoglobin level (g/dL)b Thalidomide group (n 28) 20 (71.4) Iron-controlled group (n 27) 1 (3.7) P value .000 Difference (95% CI) 67.7% (54.7%85.9%) 46.4% (28.0%64.9%) 8.5 (6.011.0) 4.0 (2.07.0) 37.4% (59.5% to 15.4%) 1500 (1800 to 1200] 53.6% (72.0% to 35.1%) 60.7% (42.6% to 78.8%) 3.0 (2.04.0) 4.5 (3.06.0) 2.74 (3.72 to 1.75)

13 (46.4) 9.36 4.31/9.5 (15, 0) 5.21 2.96/4.5 (10, 0) 3 (10.7) 1585.71 446.97/1600 (2400, 800) 13 (46.4) 11 (39.3) 2.39 1.52/2.5(4, 3) 4.5 2.49/4 (11, 0) 3.06 2.08/2.33(0.3, 7.62)

0 (0.0) 1.41 2.74/2 (7, 5) 0.81 5.10/1 (11, 10) 13 (48.2) 28.57 106.90/0 (400, 0) 27 (100) 27 (100) 0.48 1.09/1 (2, 2) 0.19 2.13/0 (5, 4) 0.01 1.32/0.2 (3.54, 2.83)

.000 .000 .000 .002 .000

.000 .000 .000

.000 .000

NOTE. Continuous data are expressed as mean standard deviation/median (range). Categorical data are expressed as n (%). Yearly data were calculated in the following formula: total values of the variables during the whole follow-up period were divided by the average number of follow-up months and multiplied by 12. aCategorical variables were compared using the 2 test or Fishers exact test. bData were compared using Wilcoxon rank-sum test. cData were derived from 14 patients in each group who were transfusion-dependent before treatment.

CLINICAL AT

1634

GE ET AL

GASTROENTEROLOGY Vol. 141, No. 5

Figure 1. Four-month interval bleeding episodes during the observation (months 12 to 0), treatment (months 04), and scheduled follow-up (months 416) periods in thalidomide and iron-controlled groups.

appeared in 2 patients with GAVE in the thalidomide group, compared to the raised and reddened areas with distinctly irregular margins of the lesion before treatment (Figure 2B). However, no signicant change was found in the patient in the iron-controlled group after 1 year of follow-up.

groups, respectively (P .007). More than one adverse event occurred in 14 and 5 of these patients, respectively (P .023) (Supplementary Table 5). Two patients in the thalidomide group discontinued treatment due to leukopenia and somnolence, as described. Fatigue was the most common adverse event in both the thalidomide (32.1%) and iron-controlled groups (11.1%). Constipation (25.0% vs 11.1%), dizziness (21.4% vs 7.4%), and abdominal distension (3.6% vs 3.7%) were observed in both thalidomide and the iron-controlled groups. Peripheral edema (n 4), bitter taste (n 2), leukopenia (n 1), thrombopenia (n 1), numb limb (n 1), bradycardia (n 1), somnolence (n 1), headache (n 1), hands tremble (n 1), tinnitus (n 1), rash (n 1), pruritus (n 1), dry eye (n 1), blurred vision (n 1), increased vaginal discharge (n 1), and herpes zoster (n 1) were reported in the thalidomide group only. Stomach upset (n 3), diarrhea (n 1), nausea (n 1), and vomiting (n 1) were reported in the iron-controlled group only. There was no signicant difference in the frequency of any individual adverse events between the 2 groups. All adverse events were mild-to-moderate and were either resolved after symptomatic treatment or disappeared 4 to 8 weeks after the treatment. No pregnancy occurred in 16 childbearing potential females and no severe adverse events such as deep venous thrombosis were seen during the study.

CLINICAL AT

Sensitivity Analysis Excluding Cases With GAVE

In the 52 patients with small bowel angiodysplasia, the effective response rate was 69.2% and 3.8%, respectively, in the thalidomide and the iron-controlled groups (P .001) (Supplementary Table 1). During the follow-up period, the rate of cessation of bleeding was 42.3% and 0%, respectively, in the 2 groups (P .001). Results for other secondary end points were almost identical to those obtained in the 55 patients with GIVM (Table 3, Supplementary Tables 1 4).

Plasma Expression Levels of VEGF

After the 4-month thalidomide treatment, the plasma concentration of VEGF decreased signicantly (from 118.24 34.60 pg/mL to 58.31 9.87 pg/mL; P .001) (Figure 3). In addition, the reduction was more signicant in responders than nonresponders (67.10 27.84 pg/mL vs 36.03 26.00 pg/mL; P .023) (Figure 3). In the 2 GAVE patients, VEGF levels were decreased from 135.25 6.72 pg/mL to 63.25 8.13 pg/mL after thalidomide treatments.

Adverse Events
Adverse events were observed in 20 (71.4%) and 9 (33.3%) patients in thalidomide and the iron-controlled

Discussion
This exploratory study showed that thalidomide is an effective and relatively safe therapy preventing recur-

Figure 2. Endoscopic images of vascular lesions in a 65-year-old patient with gastric antral vascular ectasia before (A) and after (B) thalidomide treatment. (A) Before thalidomide treatment, multiple longitudinal stripes of red vessels, limited to the gastric antrum and radiating in a spoke-like fashion from the pylorus to the antrum (arrows) are shown. (B) After thalidomide treatment, these lesions are either improved or disappear.

November 2011

THALIDOMIDE FOR BLEEDING FROM VASCULAR MALFORMATION

1635

Figure 3. Plasma levels of VEGF in the thalidomide group in relation to effective response. In the thalidomide treatment, plasma concentration of VEGF was decreased signicantly. The reduction was more signicant in patients with an effective response than in those without. *P .05; **P .01.

rent bleeding as dened by positive iFOBT test due to GIVM, with an effective response rate of 71.4% (20 of 28), whereas the rate was 3.7% (1 of 27) in the iron-controlled group (intention-to-treat; P .001) within 1-year followup. The rate of bleeding cessation was signicantly higher in the thalidomide group than in the iron-controlled group (46.4% vs 0%; P .001) during the whole follow-up period. No serious adverse events were found. It has been reported that bleeding due to angiodysplastic lesions stops spontaneously in approximately 50%90% of cases.13,19 However, spontaneous cessation of bleeding due to GIVM did not occur in any of the patients in the iron-controlled group. This might be attributed to our strict inclusion criterion requiring at least 6 bleeding episodes a year, which could minimize the possible bias caused by spontaneous cessation of bleeding. Additionally, most of our patients had a long-term bleeding history and previous bleeding episodes from angiodysplasia, which is a risk factor for recurrent bleeding.28 Thus, bleeding in these patients is less likely to be stopped spontaneously without therapies. Thalidomide may serve as a reasonable option for these patients. The efcacy of thalidomide in preventing recurrent bleeding due to GIVM was previously reported in several case reports and retrospective series.1521 However, to our knowledge, no randomized controlled trial assessing the long-term efcacy and safety of thalidomide in treating recurrent bleeding due to GIVM has been reported. The natural history of bleeding due to GIVM is known to be variable,29 usually episodic with remissions and recurrences, which makes it difcult to evaluate the efcacy of any therapy in short-term. In some studies, the increase in hemoglobin level and decrease in transfusion requirement were used to represent the improvement of bleeding status.1521 However, the hemoglobin level usually represents the instantaneous degree of acute bleeding, as it can be easily inuenced by blood transfusion, iron supplementa-

tion, or recurrent bleeding. Transfusion is not always required in patients with bleeding due to GIVM. In our study, only 50.9% (28 of 55) of patients were transfusiondependent. Therefore, we included the hemoglobin level or transfusion dependency as the secondary end points instead of the primary end point. On the other hand, bleeding episode, which directly reects the status of GI bleeding and the response to medications and is thus reliable, was used as the primary end point in this longterm study. In addition, iFOBT, which was used for determine the bleeding episode, is an objective and reliable measurement that represents bleeding and can easily be performed in the clinic and accepted by the patients. The optimal dosage of thalidomide for antiangiogenic treatment is unclear. Previous studies used thalidomide dosages of 100 400 mg per day.1521 Considering the older age of our study population (average age, 58.8 years), we chose a low dosage (100 mg daily) to achieve hemostasis while minimizing toxicity and consequent residual effects. Thalidomide carries the risk of serious adverse events, such as deep vein thrombosis, and common adverse events such as fatigue, constipation, dizziness, peripheral neuropathy, and edema. In our study, fatigue, constipation, dizziness, and peripheral edema were observed frequently, while there was no case with peripheral neuropathy. A new thalidomide analogue (lenalidomide), which is reported to be a more potent immunoregulator with possibly fewer side effects,30,31 may be evaluated for the prevention of GI bleeding due to GIVM. GAVE, a peculiar form of angiodysplastic lesion conned to the gastric antrum, was rst described as antral vascular ectasia in 1953 by Rider et al24 and the phrase watermelon stomach was coined in 1984 by Jabbari et al.25 In our population, a decrease in vascular lesions was observed in 2 patients treated with thalidomide, but not in the 1 treated with iron. However, the small sample sizes of these groups do not allow us to make a statistical conclusion. We included GAVE cases in the present study because we assumed that GAVE and small bowel angiodysplasia may share similar elements in vascularization or angiogenesis, although it is believed that GAVE does not share the exact same pathophysiology with small bowel angiodysplasia. Another reason we included GAVE in the present study was that it can be re-examined by an endoscope easily to monitor the efcacy of the treatment. In order to specically address the management of small bowel angiodysplasia, a sensitivity analysis excluding the 3 patients with GAVE was performed, which demonstrated that the overall results remained virtually unchanged in patients with small bowel angiodysplasia. The precise pathogenesis of GIVM has not been fully claried. Age-related vessel degeneration is a known causative or associated factor. High intestinal wall tension may precipitate local vascular ectasia, with chronic obstruction of the submucosal veins leading to capillary dilatation and precapillary sphincter incompetence.1,2 More impor-

CLINICAL AT

1636

GE ET AL

GASTROENTEROLOGY Vol. 141, No. 5

tantly, abnormalities or disorders of angiogenesis or angiogenic regulation are believed to play a major role. This is particularly true for abnormalities related to VEGF signal transduction, because VEGF is the strongest angiogenic factor expressed in the vascular wall.15,18 Evidence of VEGF and its messenger RNA has been found in colonic angiodysplasia patients.23 In vivo studies have shown that unregulated VEGF expression can lead to formation of disorganized fragile vessels susceptible to rupture.32 Thalidomide is an immunomodulating, antiangiogenic, antitumor, and anti-inammatory drug notorious for its teratogenic effects.22,33,34 The antiangiogenic properties may imbue thalidomide with a potential value in the treatment of GAVE, angiodysplasia, or other vascular lesions that share common developmental mechanisms, such as high expression levels of VEGF or other angiogenic factors. Bauditz et al17 previously demonstrated bleeding cessation with reduced VEGF levels in 6 patients with severe intestinal bleeding after thalidomide treatment (100 300 mg per day). Our study showed that plasma VEGF levels after the treatment were decreased signicantly in both patients with angiodysplasia and 2 patients with GAVE, compared with levels before treatment. Moreover, the decrease in VEGF levels is more signicant in responders than in nonresponders. Further extensive investigation on the exact pathogenic role of the VEGF pathway and its regulations in GIVM and treatment-associated mechanisms of thalidomide is required. The major limitation of the present study is that we failed to carry out a double-blinded placebo design. The primary and most secondary outcomes were objective; therefore, the ndings were minimally susceptible to reporting bias. The challenge we faced was to choose an appropriate control group. We acknowledge that iron supplementation as a control could most likely favorably impact some of our secondary end point variables, such as the hemoglobin and transfusion requirement, so that it can be regarded as a positive control agent rather than blank (placebo) control agent. Besides this, the timedosage effects of thalidomide were unable to be studied, especially in the older population we had enrolled. Therefore, well-designed studies are required to conrm the ndings.

References
1. Regula J, Wronska E, Pachlewski J. Vascular lesions of the gastrointestinal tract. Best Pract Res Clin Gastroenterol 2008;22: 313328. 2. Hodgson H. Hormonal therapy for gastrointestinal angiodysplasia. Lancet 2002;359:1630 1631. 3. Ell C, Remke S, May A, et al. The rst prospective controlled trial comparing wireless capsule endoscopy with push enteroscopy in chronic gastrointestinal bleeding. Endoscopy 2002;34:685 689. 4. Plummer JM, Gibson TN, Mitchell DI, et al. Emergency subtotal colectomy for lower gastrointestinal haemorrhage: over-utilised or under-estimated? Int J Clin Pract 2009;63:865 868. 5. Dodda G, Trotman BW. Gastrointestinal angiodysplasia. J Assoc Acad Minor Phys 1997;8:16 19. 6. Marwick T, Kerlin P. Angiodysplasia of the upper gastrointestinal tract. Clinical spectrum in 41 cases. J Clin Gastroenterol 1986; 8:404 407. 7. Barnert J, Messmann H. Diagnosis and management of lower gastrointestinal bleeding. Nat Rev Gastroenterol Hepatol 2009;6: 637 646. 8. Di Girolamo G, Malavasi A, Cassisa L, et al. Chronic bleeding caused by small bowel angiodysplasia: a problem of recognition and diagnosis. Ital J Gastroenterol 1992;24:468 469. 9. Marshall JK, Hunt RH. Hormonal therapy for bleeding gastrointestinal mucosal vascular abnormalities. Eur J Gastroenterol Hepatol 1997;9:521525. 10. Jarbandhan S, van der Veer WM, Mulder CJ. Double-balloon endoscopy in the diagnosis and treatment of hemorrhage from retrovalvular angiodysplasias. J Gastrointest Liver Dis 2008;17: 333334. 11. Gady JS, Reynolds H, Blum A. Selective arterial embolization for control of lower gastrointestinal bleeding: recommendations for a clinical management pathway. Curr Surg 2003;60:344 347. 12. Brown SG, Swain CP, Storey DW, et al. Endoscopic laser treatment of vascular anomalies of the upper gastrointestinal tract. Gut 1985;26:1338 1348. 13. Junquera F, Feu F, Papo M, et al. A multicenter, randomized, clinical trial of hormonal therapy in the prevention of rebleeding from gastrointestinal angiodysplasia. Gastroenterology 2001; 121:10731079. 14. Junquera F, Saperas E, Videla S, et al. Long-term efcacy of octreotide in the prevention of recurrent bleeding from gastrointestinal angiodysplasia. Am J Gastroenterol 2007;102:254 260. 15. Heidt J, Langers AM, van der Meer FJ, et al. Thalidomide as treatment for digestive tract angiodysplasias. Neth J Med 2006; 64:425 428. 16. Shurafa M, Kamboj G. Thalidomide for the treatment of bleeding angiodysplasias. Am J Gastroenterol 2003;98:221222. 17. Bauditz J, Schachschal G, Wedel S, et al. Thalidomide for treatment of severe intestinal bleeding. Gut 2004;53:609 612. 18. Kirkham SE, Lindley KJ, Elawad MA, et al. Treatment of multiple small bowel angiodysplasias causing severe life-threatening bleeding with thalidomide. J Pediatr Gastroenterol Nutr 2006;42: 585587. 19. Dabak V, Kuriakose P, Kamboj G, et al. A pilot study of thalidomide in recurrent GI bleeding due to angiodysplasias. Dig Dis Sci 2008;53:16321635. 20. Dunne KA, Hill J, Dillon JF. Treatment of chronic transfusiondependent gastric antral vascular ectasia (watermelon stomach) with thalidomide. Eur J Gastroenterol Hepatol 2006;18:455 456. 21. Kamalaporn P, Saravanan R, Cirocco M, et al. Thalidomide for the treatment of chronic gastrointestinal bleeding from angiodysplasias: a case series. Eur J Gastroenterol Hepatol 2009;21:1347 1350. 22. DAmato RJ, Loughnan MS, Flynn E, et al. Thalidomide is an inhibitor of angiogenesis. Proc Natl Acad Sci U S A 1994;91: 4082 4085.

CLINICAL AT

Conclusions
Thalidomide is effective and relatively safe for refractory bleeding from GIVM. Mechanisms of thalidomide activity might involve VEGF.

Supplementary Material
Note: To access the supplementary material accompanying this article, visit the online version of Gastroenterology at www.gastrojournal.org, and at doi: 10.1053/j.gastro.2011.07.018.

November 2011 23. Junquera F, Saperas E, de Torres I, et al. Increased expression of angiogenic factors in human colonic angiodysplasia. Am J Gastroenterol 1999;94:1070 1076. 24. Rider JA, Klotz AP, Kirsner JB. Gastritis with veno-capillary ectasia as a source of massive gastric hemorrhage. Gastroenterology 1953;24:118 123. 25. Jabbari M, Cherry R, Lough JO, et al. Gastric antral vascular ectasia: the watermelon stomach. Gastroenterology 1984;87: 11651170. 26. Jacobson JM, Greenspan JS, Spritzler J, et al. Thalidomide for the treatment of oral aphthous ulcers in patients with human immunodeciency virus infection. National Institute of Allergy and Infectious Diseases AIDS Clinical Trials Group. N Engl J Med 1997; 336:14871493. 27. Practice Guidelines for Perioperative Blood Transfusion and Adjuvant Therapies: an updated report by the American Society of Anesthesiologists Task Force on Perioperative Blood Transfusion and Adjuvant Therapies. Anesthesiology 2006;105:198 208. 28. Selinger CP, Ang YS. Gastric Antral vascular ectasia: an update on clinical presentation, pathophysiology and treatment. Digestion 2008;77:131137. 29. Richter JM, Christensen MR, Colditz GA, et al. Angiodysplasia: natural history and efcacy of therapeutic interventions. Dig Dis Sci 1989;34:15421546. 30. Maier SK, Hammond JM. Role of lenalidomide in the treatment of multiple myeloma and myeloysplastic syndrome. Ann Pharmacother 2006;40:286 289. 31. Richardson P, Jagannath S, Hussein M, et al. Safety and efcacy of single-agent lenalidomide in patients with relapsed and refractory multiple myeloma. Blood 2009;114:772778. 32. Lee RJ, Springer ML, Blanco-Bose WE, et al. VEGF gene delivery to myocardium: deleterious effects of unregulated expression. Circulation 2000;102:898 901. 33. Franks ME, Macpherson GR, Figg WD. Thalidomide. Lancet 2004; 363:18021811.

THALIDOMIDE FOR BLEEDING FROM VASCULAR MALFORMATION

1637

34. Sharma NL, Sharma VC, Mahajan VK, et al. Thalidomide: an experience in therapeutic outcome and adverse reactions. J Dermatolog Treat 2007;18:335340.

Received September 16, 2010. Accepted July 18, 2011. Reprint requests Address requests for reprints to: Zhi-Zheng Ge, MD, PhD, Division of Gastroenterology and Hepatology, Shanghai Jiao-Tong University School of Medicine Renji Hospital, Shanghai Institute of Digestive Disease; Key Laboratory of Gastroenterology & Hepatology, Ministry of Health (Shanghai Jiao-Tong University), Shanghai, China, 145 Shan-dong Zhonglu, Shanghai 200001, China. e-mail: zhizhengge@yahoo.com.cn; fax: 86-21-68383015. Acknowledgments Drs Zhi-Zheng Ge and Hui-Min Chen contributed equally to this paper. All procedures were in accordance with the Declaration of Helsinki. Informed consent was taken from all patients and the Institute Ethics Committee approved the study protocol. The full trial protocol can be accessed from Dr. Zhi-Zheng Ge via email to zhizhengge@yahoo.com.cn. The name of trial registry is Long-term Effects of Thalidomide for Recurrent Gastrointestinal Bleeding Due to Vascular Malformation. The Registration number at ClinicalTrials.gov is NCT00964496. Conicts of interest The authors disclose no conicts. Funding The study was not supported by any pharmaceutical funding. The study medications were provided by the Pharmacy Department of Renji Hospital. All data and analyses were conducted at Renji Hospital.

CLINICAL AT

1637.e1

GE ET AL

GASTROENTEROLOGY Vol. 141, No. 5

Supplementary Table 1. Clinical Efcacy of Thalidomide vs Iron Supplement in the Prevention of Bleeding in 52 Patients With Small Bowel Angiodysplasia
Variable Primary end point Patients with effective responsea Secondary end points Patients with cessation of bleedinga Change in yearlyd bleeding episodesb Change in average bleeding duration (days)b Patients dependent on blood transfusiona Change in yearlyd transfused red cell requirement (mL)b,c Overall hospitalization rateb Hospitalization rate for bleedingb Change in yearlyd hospitalization number for bleedingb Change in average hospital stay (days)b Change in average hemoglobin level (g/dL)b Thalidomide group (n 26) Iron-controlled group (n 26) P value Difference (95% CI)

18 (69.2) 11 (42.3) 8.96 4.22/9.0 (15 to 0) 4.96 2.92/4.0 (10 to 0) 3 (11.5) 1600.00 381.39/1600 (2400 to 1000) 13 (50.0) 11 (42.3) 2.39 1.52/2.5 (4 to 3) 4.35 2.48/4 (11 to 0) 2.88 2.04/2.25 (0.3 to 7.62)

1 (3.8) 0 (0.0) 1.38 2.79/1.5 (7 to 5) 0.81 5.20/0.5 (11 to 10) 13 (50.0) 28.57 106.90/0 (400 to 0) 26 (100) 26 (100) 0.48 1.09/1 (2 to 2) 0.08 2.10/0 (5 to 4) 0.05 1.34/0.20 (3.54 to 2.83)

.000 .000 .000 .000 .006 .000 .000 .000 .000 .000 .000

65.4% (46.2%84.6%) 42.3% (23.3%61.3%) 8.5 (6.011.0) 4.0 (2.07.0) 38.5% (61.3% to 15.7%) 1500 (1800 to 1200) 50.0% (69.2% to 30.8%) 57.7% (76.7% to 38.7%) 3.0 (2.04.0) 4.5 (3.0, 6.0) 2.74 (3.72 to 1.75)

Continuous data are expressed as mean standard deviation/median (range). Categorical data are expressed as n (%). CI, condence interval. variables were compared using Fishers exact test. bData were compared using the Wilcoxon rank-sum test. cData were derived from 14 patients in iron-control group and 12 patients in thalidomide group who were transfusion dependent before treatment. dYearly data were calculated in the following formula: the total values of the variables during the whole follow-up period were divided by the average number of follow-up months and multiplied by 12.
aCategorical

November 2011

THALIDOMIDE FOR BLEEDING FROM VASCULAR MALFORMATION

1637.e2

Supplementary Table 2. Demographic and Clinical Characteristics of 52 Patients With Small Bowel Angiodysplasia Before Randomization
Characteristic Age (y)a 50b 50b Sex Male Female Comorbiditiesb Coronary artery heart disease Hypertension Diabetes mellitus Aortic valve stenosis Asthma History of bleeding (y)c Endoscopic technique conrming the diagnosisb CE DBE Unresponsive to DBE therapy Unsuitable for DBE therapy Multiple lesions Older than 75 years old Lesions inaccessible to DBE Laboratory parameters Total WBC (109/L) PLT (109/L) GPT (IU/L) Total bilirubin (mol/L) Serum creatinine (mg/dL) PT-INR (s) APTT (s) MCV (fL) MCHC (g/L) Thalidomide group (n 26) 59.62 12.15/60 (4182) 20 (76.9) 6 (23.1) 3 (11.5) 23 (88.5) 2 (7.1) 3 (10.7) 4 (14.3) 2 (7.1) 1 (3.6) 5.69 4.27/5 (121) Iron-controlled group (n 26) 58.54 10.43/57.5 (4079) 22 (84.6) 4 (15.4) 5 (19.2) 21(80.8) 2 (7.4%) 2 (7.4) 3 (11.1) 2 (7.4) 0 (0.0) 5.00 4.52/3 (120)

24 (92.3) 2 (7.7) 2 (7.1) 13 (46.4) 3 (10.7) 9 (21.4) 6.44 1.62/5.9 (4.59.6) 210.18 53.36/196.50 (114.00346.00) 23.65 13.22/25.10 (5.8050.00) 8.67 4.14/7.50 (3.5018.00) 77.42 21.77/73.90 (49.50130.00) 1.05 0.16/1.03 (0.791.43) 26.98 6.19/26.85 (14.6039.40) 76.76 4.07/78.40 (67.8080.70) 296.96 14.94/300.00 (265.00318.00)

23 (88.5) 2 (7.7) 1 (3.7) 14 (51.9) 1 (3.7) 10 (18.5) 6.67 1.73/6.4 (4.19.7) 198.89 60.95/188.00 (109.00353.00) 24.60 14.03/22.05 (6.4051.60) 9.36 4.04/8.90 (3.0017.80) 87.32 23.24/83.95 (51.10138.50) 1.10 0.16/1.05 (0.911.45) 27.85 6.00/28.00 (12.9041.90) 75.17 4.48/76.40 (67.9080.60) 299.77 12.94/304.00 (266.00319.00)

NOTE. Measurement data are expressed as mean standard deviation/median (range). Categorical data are expressed as n (%). There were no signicant differences in any variables between the 2 groups. APTT, activated partial thromboplastin time; CE, capsule endoscope; DBE, double-balloon endoscope; GPT, glutamic-pyruvic transaminase; MCHC, mean corpuscular hemoglobin concentration; MCV, mean corpuscular volume; PLT, platelet count; PT-INR, prothrombin time-international normalized ratio; WBC, white blood cell. aData were compared using a 2-sample independent t test. bCategorical variables were compared using Fishers exact test. cData were compared using the Wilcoxon rank-sum test.

1637.e3

GE ET AL

GASTROENTEROLOGY Vol. 141, No. 5

Supplementary Table 3. Baseline Characteristics of 52 Patients With Small Bowel Angiodysplasia in 1-Year Observation
Characteristic Bleedinga Bleeding episodes Average bleeding duration (days) Hospitalization Hospitalization rates for bleedingb Hospitalization number for bleedinga Average hospital stays (days) Transfusion requirements Patients with blood transfusionb Transfused red cell requirements (mL)c,d Average hemoglobin level (g/dL)c
aData

Thalidomide group (n 26) 13.92 3.51/14 (719) 8.85 3.20/9 (520) 26 (100) 3.04 1.00/3 (15) 6.65 2.45/7 (314) 12 (46.2) 2000.00 48.33/1800 (10003400) 6.56 1.70/7.34 (2.989.30)

Iron-controlled group (n 26) 13.85 3.48/13.5 (721) 8.81 4.58/7 (321) 26 (100) 2.69 1.16/3 (15) 6.65 2.31/6 (312) 14 (53.8) 1985.71 766.47/1900 (8003400) 6.03 1.55/5.64 (3.79.99)

NOTE. Measurement data are expressed as mean standard deviation/median (range). Categorical data are expressed as n (%). were compared using the Wilcoxon rank-sum test. bData were compared using 2 test. cData were compared using a 2-sample independent t test. dData were derived from 14 patients in iron-control group and 12 patients in thalidomide group who were transfusion-dependent before treatment.

Supplementary Table 4. Adverse Events Observed in 52 Patients With Intestinal Angiodysplasia in the Studyb
Adverse event Patients with at least one adverse eventa Patients with more than two adverse eventsa Constipationc Diarrhea Stomach upsetc Abdominal pain Abdominal distension Nausea Vomiting Leukopenia Thrombopenia Fatiguec Dizzinessc Bitter tastesc Peripheral edemac Bradycardia Somnolence Headache Hands tremble Tinnitus Rash Pruritus Dryness of eye Blurred vision Increase in vaginal discharge Herpes zoster infection
aP

Supplementary Table 5. Adverse Events in Total 55 Patients With GIVM Observed in the Study
Thalidomide group (n 28), n (%) 20 (71.4) 14 (50.0) 7 (25.0) 0 (0.0) 0 (0.0) 0 (0.0) 1 (3.6) 0 (0.0) 0 (0.0) 1 (3.6) 1 (3.6) 9 (32.1) 6 (21.4) 2 (7.1) 1 (3.6) 4 (14.3) 1 (3.6) 1 (3.6) 1 (3.6) 1 (3.6) 1 (3.6) 1 (3.6) 1 (3.6) 1 (3.6) 1 (3.6) 1 (3.6) 1 (3.6) Iron-controlled group (n 27), n (%) 9 (33.3) 5 (18.5) 3 (11.1) 1 (3.7) 3 (11.1) 1 (3.7) 1 (3.7) 1 (3.7) 1 (3.7) 0 (0.0) 0 (0.0) 3 (11.1) 2 (7.4) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0)

Thalidomide group Iron-controlled group (n 26), n (%) (n 26), n (%) 19 (73.1) 13 (50.0) 6 (23.1) 0 (0.0) 0 (0.0) 0 (0.0) 1 (3.8) 0 (0.0) 0 (0.0) 1 (3.8) 1 (3.8) 9 (34.6) 6 (23.1) 2 (7.7) 3 (11.5) 1 (3.8) 1 (3.8) 1 (3.8) 1 (3.8) 1 (3.8) 1 (3.8) 1 (3.8) 1 (3.8) 1 (3.8) 1 (3.8) 1 (3.8) 9 (34.6) 5 (19.2) 3 (11.5) 1 (3.8) 3 (11.5) 1 (3.8) 1 (3.8) 1 (3.8) 1 (3.8) 0 (0.0) 0 (0.0) 3 (11.5) 2 (7.7) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0) 0 (0.0)

Adverse event Patients with at least 1 adverse eventa Patients with 2 adverse events Constipationb Diarrhea Stomach upsetb Abdominal pain Abdominal distension Nausea Vomiting Leukopenia Thrombopenia Fatigueb Dizzinessb Bitter tastesb Numb limb Peripheral edemab Bradycardia Somnolence Headache Hands tremble Tinnitus Rash Pruritus Dryness of eye Blurred vision Increase in vaginal discharge Herpes zoster infection
aThe

P value .007 .023 .295 .491 .111 .491 1.000 .491 .491 1.000 1.000 .101 .252 .491 1.000 .111 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

.05, compared between the 2 groups. bThe number of patients is less than the sum of the numbers of patients with individual adverse events because some patients experienced 2 or more events. cAdverse events with a frequency of 5% within at least 1 arm.

number of patients is less than the sum of the numbers of patients with individual adverse events because some patients experienced 2 events. bAdverse events with a frequency of 5% within at least 1 arm.

November 2011

THALIDOMIDE FOR BLEEDING FROM VASCULAR MALFORMATION

1637.e4

Supplementary Figure 1. Flow diagram. ITT, intention-to-treat principle; PP, per-protocol.