I graft-versus-host disease (GVHD) after allogeneic bone

Transcription

1 Acute Upper Gastrointestinal Graft-Versus-Host Disease: Clinical Significance and Response to Immunosuppressive Therapy By Daniel J. Weisdorf, Dale C. Snover, Robert Haake, Wesley J. Miller, Philip B. McGlave, Bruce Blazar, Norma K.C. Ramsay, John H. Kersey, and Alexandra Filipovich Recognized manifestations of acute graft-versus-host disease (GVHD) of the gastrointestinal (GI) tract include secretory diarrhea, abdominal pain, and, at times, hemorrhage. In a review of 469 patients undergoing allogeneic bone marrow transplantation (BMT) from matched sibling donors at our institution, we have recognized a syndrome of upper GI GVHD. This syndrome, presenting clinically as anorexia, dyspepsia, food intolerance, nausea, and vomiting, was recognized and confirmed histologically in 62 patients (1 3% by Kaplan-Meier projection) at the initiation of systemic GVHD therapy, a subset of the 197 patients developing grade II through IV GVHD. These 62 patients with upper GI GVHD were significantly older than the overall BMT population and older than the cohort with grade II through IV GVHD, as well. Of the 62 patients, 25 had upper GI GVHD accompanied only by limited (stage 1 and 2) skin GVHD: 13 others with upper GI GVHD plus limited skin involvement at initial presentation later progressed to more extensive multiorgan involvement; 24 others presented with upper GI along with other organ GVHD. This upper GI GVHD syndrome, first recognized at NVOLVEMENT of the gastrointestinal (GI) tract by I graft-versus-host disease (GVHD) after allogeneic bone marrow transplantation (BMT) is usually manifest clinically as voluminous secretory diarrhea accompanied by abdominal cramping, ileus, nutritional depletion, and, at times, hemorrage. The clinical staging system originally posed by investigators from Seattle reflects primarily the salt and water reabsorption defect in the distal small bowel and colon that results in diarrhea.** The histologic criteria for intestinal GVHD also are defined by the rectal mucosal histology. However, the secretory nature of the diarrhea with active small bowel electrolyte and solute secretion, as well as the protein exudation often accompanying GVHD enterocolitis, suggests that the major pathology may initially affect the upper small intestinal tract. We have recognized a syndrome of acute GVHD affecting primarily the upper GI tract presenting as anorexia, dyspepsia, and inability to eat, and have reviewed the clinical implications of upper GI GVHD in From the University of Minnesota Bone Marrow Transplantation Program, Minneapolis. Submitted October 2,1989; accepted April 2,1990. Supported in part by National Institutes of Health Grant POI CA21737, and by the Bone Marrow Transplant Research Fund. J.H.K. is the recipient of an Outstanding Investigator Award (CA 49721) from the National Cancer Institute. Address reprint requests to Daniel Weisdorf. MD, Department of Medicine, Box 480 UMHC, Harvard St at East River Rd, Minneapolis, MN The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. section 1734 solely to indicate this fact. o 1990 by The American Society of Hematology /90/ $3.00/0 our center in 1983, has been diagnosed with increasing frequency (22% f 5%) in the most recent Byear interval. The upper GI GVHD syndrome is more responsive to immunosuppressive therapy than grade II GVHD defined by Seattle criteria, with complete and continuing responses to treatment observed in 71 % f 17% (95% confidence interval) of those with the upper GI GVHD syndrome compared with only 37% + 10% complete responses in other patients with grade II GVHD (P =.002). Patients failing immunosuppressive therapy for upper GI GVHD often progress to symptomatic lower GI involvement, suggesting that this syndrome may be an earlier and perhaps more treatable manifestation of this unique intestinal immunopathology, which is followed by chronic GVHD in 74% of patients. While upper GI GVHD symptoms are nonspecific and require invasive histologic and microbiologic studies to confirm the diagnosis, we believe this syndrome has been underreported after allogeneic BMT and propose its recognition within the clinical GVHD scoring system by The American Society of Hematology. an extended series of allogeneic BMT recipients. We report the clinical presentation, complications, and outcome of this syndrome of upper GI acute GVHD after BMT and propose its inclusion in the clinical staging of intestinal GVHD. PATIENTS AND METHODS We have reviewed the records of the University of Minnesota Bone Marrow Transplant database, which contains prospectively collected clinical records of all patients transplanted at our institution. Between January 1979 and October 1987 (unique patient nos. 77 through 797), 469 patients received an allogeneic marrow transplant from a matched sibling donor. Of these, 197 recipients developed grade I1 or greater acute GVHD, including 62 with involvement of the upper GI tract at the time of initial systemic therapy. Eight other patients who developed histologically confirmed upper GI GVHD more than 1 week after the initiation of systemic therapy for GVHD have been excluded. All patients have been followed for a minimum of 1 year post BMT. The patients clinical characteristics and their GVHD prophylaxis are shown in Table 1. They all received marrow from histmmpatible sibling donors and received GHVD prophylaxis including: methotrexate alone; methotrexate, horse anti-thymocyte globulin (ATG) and prednisone4; ex vivo T-lymphocyte depletion ; or other GVHD prophylaxis methods, which included methotrexate/cyclosporine/prednisone, methotrexate/okt3/prednisone, and other variant regimens6 The conditioning regimens used pretransplant and the supportive care techniques accompanying BMT in our institution have been reported previ~usly.~ ~ Acute GVHD was diagnosed by clinical criteria and confirmed histologically in over 95% of patients with skin GVHD, 95% with lower GI GVHD, 40% with liver GVHD, and 100% with upper GI involvement. Histologic findings diagnostic of GVHD without accompanying clinical organ symptomatology were not included in the clinical staging or analysis. Excluding the upper GI tract, the symptoms of acute GVHD were graded by standard clinical criteria? and in this report grade of GVHD refers only to clinical, not histologic, characteristics of GVHD. 624 Blood, Vol 76, No 3 (August 1). 1990: pp

2 UPPER GI GVHD 625 Table 1. Upper GI GVHD. Clinical Characteristics All patients Age median (range) Onset of acute upper GI GVHD (median: range) Original Diagnosis N = 62 (13%). 36 years (1 8-53) 32 days: N With Upper GI GVHD (% of those transplanted for the same diagnosis)t Aplastic anemia 7 (11) ALL 4 (4) ANL 17 (14) CML 33 (28) Other leukemia/malignancy 1 (2) GVHD prophylaxis N (%I$ Methotrexate/ATG/pred- 4 4 ) nisone Methotrexate alone l(1) T-lymphocyte depletion 14 (25) Other 3 (6) Abbreviations: ALL, acute lymphoblastic leukemia: ANL, acute nonlymphocytic leukemia: CML, chronic myelogenous leukemia. *Percent with upper GI GVHD of 469 patients receiving matched sibling allografts. tpercent with upper GI GVHD of patients with the same diagnoses receiving allografts. $Percent with upper GI GVHD of patients transplanted who received the same GVHD prophylaxis. The syndrome of upper GI acute GVHD was suspected in patients presenting with either nausea, vomiting, anorexia, epigastric pain or burning, and/or food intolerance, and was diagnosed only when confirmed histologically through endoscopic biopsy of gastric, duodenal, or rarely esophageal mucosa." The histologic findings of acute GVHD in the upper GI biopsies included single cell epithelial necrosis with associated karyorrhexic debris, dilatation of mucosal crypts or glands, and/or crypt abscesses or crypt obliteration." All biopsy sites were cultured for bacterial, viral, and fungal pathogens that might confound the diagnosis of GVHD, and the diagnosis of upper GI GVHD was not made if concurrent enteric infection was present. During this period, 132 of these BMT recipients underwent 216 endoscopic examinations with biopsy for upper GI symptoms. Within this group, 62 had biopsies diagnostic of GVHD by the above criteria. Only 1 of 15 biopsies performed before day +20 was positive (on day +17), and this patient had unequivocal extraintestinal GVHD at that and later time periods. Ten patients with initially negative studies later had repeat endoscopic biopsies diagnostic of GVHD. Twenty-five patients had this upper GI GVHD syndrome recognized in the absence of lower GI, hepatic, or extensive cutaneous GVHD, and would have been classified by standard criteria' as overall grade 0 (n = 9) or grade I (n = 16) acute GVHD. They were treated with systemic immunosuppressive therapy just as patients with grade I1 or greater acute GVHD based on their refractory upper GI symptoms. Patients were assigned therapy for acute GVHD according to prospectively designed treatment protocols. These treatments included either prednisone 60 mg/m2/d with a slow tapering schedule over 15 weeks, or prednisone 60 mg/m'/d with a variant tapering schedule. Additional immunosuppressive therapies for acute GVHD included high-dose methylprednisolone (30 mg/kg intravenously twice per day x 5 days); ATG/prednisone (ATG 15 mg/kg twice per day x 5 days plus prednisone 40 mg/m2/d); and other therapies including cyclosporine, anti-t-cell immunotoxins, or anti-lymphocyte globulin. Complete response (CR) was defined as the resolution of all symptomatology of acute GVHD before the onset of chronic GVHD (cgvhd). Patients were censored at death, relapse, second BMT, or at the onset of cgvhd in the analysis of time to CR. Transient or partial responses were not included in the analysis and only complete and continuing response to GVHD therapy was recorded as a CR. Statistical analyses were performed using BMDP-83 software'' and standard statistical methods. The development of CR or cgvhd were assessed using the Kaplan-Meier product limit method with 95% confidence limits derived from the standard errors. Time-dependent covariate analysis was used for nonbaseline characteristics." Differences between groups were assessed with the Mantel-Cox test statistic. Multivariate analysis was performed with the Cox model using known or suspected covariates in stepwise multivariate fashion." RESULTS Of the 469 patients receiving histocompatible allografts, 62 (13% f 4%, 95% confidence interval [CI] by Kaplan- Meier projection) developed histologically confirmed acute GVHD involving the upper GI tract at the time of initial systemic therapy for acute GVHD. As shown in Table 1, the patients with upper GI GVHD were all adults and were older (median 36 years) than other patients undergoing BMT (median 17 years; P <.001). They also were older than other patients who developed grade I1 through IV acute GVHD without upper GI involvement (median 23 years; P 5.001). Univariate analysis demonstrated an increased risk of upper GI GVHD in older patients (or those with older donors) (P <.0001) and in those with chronic myelogenous leukemia (P <.0001) both within the entire patient group and in the years 1983 through 1987 when upper GI GVHD was recognized more consistently. No GVHD prophylaxis technique was more highly associated with upper GI GVHD. However, when stepwise multivariate analysis was performed for the whole patient group, or for the years 1983 through 1987, only older patient age was identified as a significant and independent risk factor for upper GI GVHD. Each decade of patient age added a 2.2-fold increased relative risk of upper GI GVHD (P <.0001). At our center the diagnosis of upper GI acute GVHD has been more common in recent years. As shown in Table 2, while lower GI disease was recognized in 5% to 15% of the patients receiving histocompatible sibling allografts over the 9 years analyzed, upper GI GVHD was never diagnosed until 1983, but since that time has been recognized consistently and more frequently than lower GI involvement. Excluding the years 1979 through 1982 when upper GI GVHD was probably underdiagnosed, the observed incidence of upper GI GVHD is 62 of 304 sibling donor allografts; 22% f 5% by Kaplan-Meier projection. Most patients who developed upper GI involvement with acute GVHD had other organ GVHD as shown in Table 3. Of the 62 patients with upper GI involvement, 19 had no other organ involved at the time of initial therapy and 12 others had only limited cutaneous GVHD. Upper GI involvement often occurred without lower GI GVHD. Only 11

3 626 WEISDORF ET AL Table 2. Diagnosis of Upper GI Acute GVHD: Effect of Year of BMT Year of Upper GI GVHD Lower GI GVHD BMT N* N (%It N (%)$ (10) 13 (23) 16 (26) 17 (26) 10 (17) Total (13) 45 (10) *Number of patients receiving histocompatible sibling allografts and thus at risk for acute GVHD each year. tpercent of those patients at risk for acute GVHD developing upper GI (or lower GI) GVHD. patients with upper GI involvement when initial therapy was instituted had involvement of the lower GI tract manifest as significant diarrhea. Similarly, significant liver disease was uncommon (only 8 of 62), while skin disease was present in 37 of these patients. Twenty-five patients with upper GI involvement at presentation never subsequently developed stage I11 or IV skin GVHD, nor any lower GI or liver involvement, and thus were treated systemically exclusively on the basis of upper GI with stage I skin GVHD (n = 16) or with no skin involvement (stage 0; n = 9). Thirteen patients with upper GI and limited skin GVHD later progressed to multiorgan involvement, and 24 others had upper GI plus other organ involvement at onset therapy for GVHD. Outcome after immunosuppressive therapy for acute GVHD. After initiating immunosuppressive therapy for Table 3. Organ Involvement at Onset of Systemic Therapy for Acute GVHD Organ Skin Lower GI Liver Stage. N (%)t N (%I N (%) 0 25 (40) 51 (81) 1 12 (19) 4 (6) 2 14 (24) 1 (1) 3 11 (17) 4 (6) 4 2 (3) Overall Grade 0 I II 111 IV Grade Discounting Upper GI* N (%I 19 (31) 12 (19) 19 (30) 10 (16) 2 (3) Grade Including Upper GIS N (%I N = 62, all with upper GI GVHD. *According to Seattle criteria (not including UGI). tpercent of 62 patients with UGI GVHD. $Grade as treated clinically. UGI GVHD was considered t grade II. the 62 patients with upper GI involvement, 31 patients experienced complete and continuing responses (CR) and all symptoms of acute GVHD resolved, resulting in a Kaplan- Meier product limit estimate of 58% CR (95% CI 44% to 72%). Thirteen patients failed to develop adequate responses and progressed to require secondary immunosuppressive therapy. Two patients died before achieving CR, 4 relapsed, and 12 of the 64 developed chronic GVHD before all acute GVHD symptomatology had resolved, and thus these 18 were censored from analysis of CR to GVHD. Lower GI GVHD often developed or worsened in patients with upper GI GVHD failing initial systemic therapy for GVHD. Of 51 patients with no lower GI GVHD at initial therapy, seven subsequently developed lower GI involvement. Similarly, of 11 patients with lower plus upper GI GVHD at onset of therapy, two developed subsequent progression of lower GI GVHD. Six of the 13 patients with resistant GVHD requiring secondary immunosuppressive therapy developed progressive lower GI involvement as the manifestation of their refractory GVHD. To assess the clinical significance of upper GI acute GVHD, we compared posttransplant outcomes (CR to immunosuppressive therapy for GVHD, the development of cgvhd, and overall survival) for 38 patients whose upper GI GVHD was the indication for grade I1 scoring and thus systemic therapy (upper GI positive with only skin stage 0 or I disease), with 98 other patients with grade I1 GVHD (discounting their upper GI disease in the scoring). As shown in Table 4, despite initiation of systemic GVHD therapy at a similar time post BMT, a CR to immunosuppressive therapy was obtained significantly more frequently in the patients with upper GI syndrome (71%) compared to those with overall grade I1 acute GVHD (37%) (P =.0009). However, this upper GI skin acute GVHD was followed by the development of cgvhd in 74% of the patients, not significantly different than the 65% incidence of cgvhd observed in the other patients with grade I1 disease. By comparison, from this same patient cohort of 469 sibling donor allografts, only 13% 5 6% of 226 patients with no acute GVHD subsequently developed cgvhd (de novo), and in the cohort Table 4. Effect of Upper GI Involvement in Grade II GVHD Grade II by Upper GI Plus Skin Stage 011 Only. Overall Grade Ilt N P Day of initial systemic therapy for GVHD (median; range) 37; ; NS CR$ (% * 95% CI) 71 * r Chronic GVHD (%) 74r r Survival (% at 5 years post 8MT) 45 r r At initial systemic therapy of acute GVHD. Abbreviation: NS, not significant. *Includes 25 patients who never developed grade I1 by Seattle criteria and 13 who later progressed to 5 stage Ill skin or GI or liver involvement. tas in Table 2. Grade II GVHD by original Seattle criteria (excludes upper GI from scoring). $Kaplan-Meier projections (*95% Cl) of CR of GVHD, chronic GVHD, and survival. Mantel-Cox test of significance between groups.

4 UPPER GI GVHD 627 of 44 patients with only stage 1 or 2 skin disease (overall grade I), only 50% 2 26% later developed cgvhd (P =.01 compared to overall grade I1 with upper GI GVHD). Finally, survival comparisons demonstrated similar, though slightly better, survival in patients with the limited acute upper GI GVHD compared to the others with grade I1 disease by conventional scoring criteria (P =.11)(Table 4). GI involvement overall, either upper or lower, was not associated with more frequent CR for acute GVHD than was seen in patients with no GI involvement. As shown in Table 5, the rate of achieving CR was the same in patients with or without GI GVHD. However, patients with upper GI disease had a striking and significantly greater likelihood of CR to therapy than patients with no upper GI GVHD or than those with lower GI involvement. Patients with upper (but not lower) GI involvement, compared to those with lower (but not upper) GI involvement, had a significantly greater CR rate to GVHD therapy. Upper GI GVHD, despite responsiveness to immunosuppressive therapy, reflected the systemic nature of the GVHD syndrome in that upper GI involvement was frequently associated with both the development of cgvhd and overall survival similar to that seen in patients with lower GI GVHD. Patients with upper (but not lower) GI disease developed cgvhd 76% 2 15% of the time, similar to those with lower (but not upper) GI involvement who had a 71% 21% incidence of cgvhd (P =.98). Their overall survival was also similar (41% 19% v 45% * 17%, respectively; P =.35). Proposed clinical grading to include upper GI GVHD. Our analysis has shown that symptomatic upper GI GVHD is a frequent complication of allogeneic BMT. In addition, even with prompt initiation of systemic immunosuppressive therapy we have observed that upper GI GVHD is frequently followed by cgvhd and compromised overall survival. Therefore, we contend that upper intestinal acute GVHD should be recognized within the clinical organ staging and overall grading of acute GVHD. Shown in Table 6 is our proposed modification of the widely accepted Seattle criteria for the clinical scoring of acute GVHD. Because of the Table 5. Response to Therapy of Upper GI Versus Lower GI Acute GVHD CR GI Involvement* N (% f 95% CII P Any GI (upper OT lower) No GI GVHD Upper GI No upper GI Lower GI No lower GI Upper GI (no lower GI) Lower GI (no upper GI) 96 45* f f f f k 14 Kaplan-Meier projection of CR rate of GVHD (+95% CI) for each patient group shown. 'At initial systemic therapy of acute GVHD. Table 6. Proposed Clinical Staging/Grading of Acute GVHD Including the Upper GI Tract Clinical Stage GI GVHD ml diarrhea/d and/or anorexia, nausea. ++ z 1,000 ml diarrhea/d and/or food intolerance, vomiting+ +++ > 1,500 ml diarrhea/d Severe abdominal pain, with or without ileus ++++ Overall Clinical Grade of Acute GVHD I + to + + skin involvement; no GI involvement; no liver involvement; no decrease in clinical performance II + to skin rash; + lower GI; + to + + upper GI involvement; + liver: mild decrease in performance to skin rash; + + to lower GI; + to liver (w both); marked decrease in performance IV + + to organ involvement with extreme decrease in clinical performance Modified with permission of the New England Journal of Medicine (292:895, " *Upper GI GVHD requires histologic confirmation for clinical staging or grading. nonspecific presenting symptoms, we feel that the diagnosis of upper GI involvement with GVHD requires histologic confirmation. Upper GI involvement is to be added to the intestinal stage score, though it cannot be quantitated in severity. The overall GVHD grading system should also account for upper GI involvement (as shown in Table 6) since we feel its diagnosis is an indication for systemic immunosuppressive therapy. DISCUSSION GI disturbances accompanying allogeneic BMT frequently compromise patients' comfort, nutrition, and survival after BMT. The clinical manifestations of intestinal dysfunction are often nonspecific, rendering the differential diagnosis between chemoradiotherapy effects, infections, or GVHD difficult without detailed microbiologic and histologic assessment. At our center we have recognized a clinical syndrome of upper GI disturbance characterized by the nonspecific symptomatology of nausea, vomiting, bloating, and food intolerance, which, when accompanied by histologic demonstration of characteristic mucosal changes in gastric or duodenal mucosa and not confounded by enteric infection, we have characterized as upper GI GVHD." This syndrome, occurring even more frequently than GI GVHD with diarrhea, represents an additional and perhaps earlier manifestation of GI GVHD. Symptomatic upper GI GVHD represents significant systemic disease that requires immunosuppressive therapy to provide resolution of symptoms, and despite such intensive immunosuppressive therapy it is frequently followed by more extensive GVHD in other organs. Additionally, the majority of patients with isolated upper GI GVHD develop cgvhd, similar to the rate in other patients with

5 628 WEISDORF ET AL grade I1 through IV acute GVHD and more frequently than patients with grade 0 or I acute GVHD. Upper GI GVHD appears highly responsive to immunosuppressive therapy. Prompt recognition of the clinical syndrome, early endoscopic biopsy, and initiation of treatment may result in a greater likelihood of CR to GVHD therapy and better overall survival. We have often observed that patients failing therapy for upper GI GVHD subsequently manifest significant diarrhea (seen in 7 of 14 treatment failures), implying that the upper GI clinical syndrome may be an earlier rather than a distinct manifestation of intestinal involvement. The recognition of upper GI GVHD in our center since 1983 has been accompanied by somewhat less frequent diagnosis of lower GI GVHD. We suggest that treatment of the upper GI syndrome may have either prevented or preemptively treated the more extensive intestinal involvement before it resulted in large volume diarrhea. The frequent recognition of upper GI GVHD in older patients might reflect a biologic predisposition to more GVHD in this population. These observations will require confirmation from other centers. Because this diagnosis requires endoscopic exam and biopsy, our observed incidence of 22% must represent a minimum rate of upper GI involvement with GVHD. Our clinical protocols for investigating upper GI symptoms recognized a similar incidence as that reported prospectively from Seattle. Spencer et all4 reported unsuspected acute upper GI GVHD in 13 of 44 patients investigated in a prospective study of unexplained nausea and vomiting after BMT. Eight additional patients had acute upper GI GVHD complicated by intestinal infection. Six of the 13 with uncomplicated GVHD progressed to significant diarrhea, but seven remained at stage 0 lower GI GVHD. Discordant findings between simultaneous rectal and upper GI biopsies were frequent. In a recent review of the clinical characteristics of GVHD, Vogelsang et all5 acknowledged the frequency of nausea and vomiting accompanying gut GVHD, but emphasized only GI GVHD manifest by significant diarrhea. Our previous studies have emphasized that upper intestinal, and in particular gastric and duodenal biopsies, may be necessary to diagnose GVHD in patients with upper GI complaints only, with no diarrhea, and normal rectal biopsy specimens." We also recognized the mimicry of these symptoms and histologic changes by the effects of cytoreductive therapy early posttransplant and by intestinal viral infection, particularly cytomegalovirus (CMV). Modern techniques for more sensitive and more rapid detection of CMV-related antigen in the biopsies may minimize this potential confusion. Previous pathologic studies have recognized GVHD involvement throughout the GI One study based on rapidly examined, pre-autolysis autopsy material recognized the frequent concordance of the diagnosis and grading of GVHD in multiple sites within the GI tract.3 This data is consistent with our interpretation that upper GI GVHD must be recognized as part of the spectrum of GI and systemic GVHD. Our observations suggest that the currently accepted scoring system for acute GHVD fails to acknowledge the frequently occurring syndrome of upper GI GVHD, and we propose modification of the GVHD scoring system accordingly. This syndrome represents a variant intestinal immunopathology that appears to be an early and perhaps more treatable manifestation of GHVD enterocolitis. The diagnosis of upper GI GVHD must be aggressively sought by endoscopic biopsy to distinguish it from enteric infection, but when confirmed histologically, this syndrome can be reversed by immunosuppressive therapy. 1. McDonald GB, Shulman HM, Sullivan KM, Spencer GD: Intestinal and hepatic complications of human bone marrow transplantation. Part I, 11. Gastroenterology 90:460, Glucksberg H, Storb R, Fefer A, Buckner CD, Nesman PE, Clift RA, Lerner KG, Thomas ED: Clinical manifestations of graft-versus-host disease in human recipients of marrow from HLA-matched sibling donors. Transplantation 18:295, Sale GE, Shulman HM, McDonald GB, Thomas ED: Gastrointestinal graft-versus-host disease in man. Am J Surg Pathol 3:291, Ramsay NKC, Kersey JH, Robison LL, McGlave PB, Woods WG, Krivit W, Kim TM, Goldman AI, Nesbit M E Prevention of acute graft-versus-host disease: A randomized study demonstrating the influence of treatment regimen and age. N Engl J Med 306:392, Filipovich AH, Vallera DA, Youle RJ, Haake R, Blazar BR, Arthur D, Neville DM Jr, Ramsay NKC, McGlave PB, Kersey JH: Graft-versus-host disease prevention in allogeneic bone marrow transplantation from histocompatible siblings. Transplantation 44: 62, Neudorf S, Filipovich A, Ramsay N, Kersey J: Prevention and treatment of acute graft-versus-host disease. Semin Hematol 21:91, Weisdorf DJ, McGlave PB, Ramsay NKC, Miller WJ, Nesbit M, Woods W, Goldman A, Kim T, Kersey JH: Allogeneic bone REFERENCES marrow transplantation for acute leukemia: Comparative outcomes for adults and children. Br J Haematol69:351, McGlave PB, Haake R, Miller WJ, Kim T, Kersey JH, Ramsay NKC: Therapy of severe aplastic anemia in young adults and children with allogeneic bone marrow transplantation. Blood 70:1325, McGlave PB, Arthur D, Haake R, Hurd D, Miller WJ, Vercellotti GM, Weisdorf DJ, Kim T, Ramsay NKC, Kersey JH: Therapy of chronic myelogenous leukemia with allogeneic bone marrow transplantation. J Clin Oncol5:1033, Snover DC, Weisdorf SA, Vercellotti GM, Rank B, Hutton S, McGlave PB: A histopathologic study of gastric and small intestinal graft-versus-host disease following allogeneic bone marrow transplantation. Hum Pathol 16:387, BMDP2L-Survival analysis with covariates-cox models. BMDP Statistical Software, Inc, Los Angeles, CA, Kaplan EL, Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53:457, Cox DR: Regression models and life tables. J R Stat SOC [B] 34:187, Spencer GD, Hackman RC, McDonald GB, Amos DE, Cunningham BA, Meyers JD, Thomas ED: A prospective study of unexplained nausea and vomiting after marrow transplantation. Transplantation 42:602, Vogelsang GB, Hess AD, Santos GW: Acute graft-versus-

6 UPPER GI GVHD 629 host disease: Clinical characteristics in the cyclosporine era. Medi- host disease: A prospective study of thirteen patients. Gastroenterolcine 67:163, 1988 ogy 78:764, Lerner KG, Kao GF, Storb R, Buckner CD, Clift RA, 18. Slavin RE, Woodruff JM: The pathology of bone marrow Thomas ED: Histopathology of graft-versus-host reaction (GVHR) transplantation. Path01 Annu 9:291, 1974 in human recipients of marrow from HLA-matched sibling donors. Transplant Pr& 6:367, Epstein RJ, McDonald GB, Sale GE, Shulman HM, Thomas ED: The diagnostic accuracy of the rectal biopsy in acutegraft-versus- 19. Thomas ED, Storb R, Clift RA, Fefer A, Johnson FL, Neiman PE, Lerner KG, Glucksberg H, Buckner CD: Bone marrow transplantation. N Engl J Med 292:832,895, 1975