We analyzed the relevance of HLA compatibility to sustained marrow engraftment in 269 patients with hematologic neoplasms who underwent bone marrow transplantations. Each patient received marrow from a family member who shared one HLA haplotype with the patient but differed to a variable degree for the HLA-A, B, and D antigens of the haplotype not shared. These 269 patients were compared with 930 patients who received marrow from siblings with identical HLA genotypes. All patients were treated with cyclophosphamide and total-body irradiation followed by the infusion of unmodified donor marrow cells. The rate of graft failure was 12.3 percent among the recipients of marrow from a donor with only one identical haplotype, as compared with 2.0 percent among recipients of marrow from a sibling with the same HLA genotype (both haplotypes inherited from the same parents) (P less than 0.0001). The incidence of graft failure correlated with the degree of donor HLA incompatibility. Graft failure occurred in 3 of 43 transplants (7 percent) from donors who were phenotypically HLA-matched with their recipient (haplotypes similar, but not inherited from the same parents), in 11 of 121 donors (9 percent) incompatible for one HLA locus, in 18 of 86 (21 percent) incompatible for two loci, and in 1 of 19 (5 percent) incompatible for three loci (P = 0.028). In a multivariate binary logistic regression analysis, independent risk factors associated with graft failure were donor incompatibility for HLA-B and D (relative risk = 2.1; 95 percent confidence interval, 1.7 to 2.5; P = 0.0004) and a positive crossmatch for anti-donor lymphocytotoxic antibody (relative risk = 2.3; 95 percent confidence interval, 1.8 to 2.8; P = 0.0038). Residual host lymphocytes were detected in 11 of 14 patients with graft failure, suggesting that the mechanism for graft failure could be host-mediated immune rejection. We conclude that donor HLA incompatibility and prior alloimmunization are significant risk factors for graft failure, and that a more effective immunosuppressive regimen than those currently used is needed for consistent achievement of sustained engraftment of marrow transplanted from donors who are not HLA-identical siblings.

One hundred eighty-seven women between 13 and 49 years of age had ovarian function evaluated from 1 to 15 years (median, 4) after marrow transplant for aplastic anemia or leukemia. Among 43 women transplanted for aplastic anemia following 200 mg/kg cyclophosphamide (CY), all 27 less than 26 years of age, but only five of 16 greater than 26 years of age recovered normal ovarian function. Nine of the 43 have had 12 pregnancies, resulting in eight live births, and two elective and two spontaneous abortions. All eight children are normal. Nine of 144 women transplanted for leukemia following 120 mg/kg CY and 9.20 to 15.75 Gy total body irradiation (TBI) recovered ovarian function. Two of these nine have had three pregnancies, resulting in two spontaneous and one elective abortion. The probability of having ovarian failure was 0.35 by 7 years for patients receiving CY alone and was 1.00 at 1 year for patients receiving CY plus TBI (P less than .0001). By 7 years after transplant the probabilities of having normal ovarian function were 0.92 after CY alone and 0.24 after CY plus TBI (P less than .0001). Multivariate analysis showed that TBI was the only factor significantly influencing ovarian failure and that both TBI and greater patient age at transplant were significantly associated with a decreased probability of recovering normal ovarian function. These data demonstrate that after high-dose CY, recovery of ovarian function occurs in younger women and in a minority of older women, but after CY and TBI, recovery occurs in only a few younger women and none of the older women.

We tested the hypothesis that liver histology from patients with graft-versus-host disease could be distinguished from other common liver diseases. Liver biopsies from 33 allogeneic marrow transplant recipients with acute and chronic graft-versus-host disease and 37 nontransplant liver disease patients without graft-versus-host disease were recut, restained and coded for blind review. Analysis of individual histologic features showed significantly more cytologic aberration of bile duct epithelium and more cholestasis among biopsies with graft-versus-host disease when compared to biopsies without graft-versus-host disease (p less than or equal to 0.05). The duration of graft-versus-host preceding the biopsy influenced the histologic features. Biopsies before Day 35 showed frequent acidophilic bodies but infrequent bile duct changes. Biopsies from Days 35 to 90 posttransplant had more frequent bile duct exocytosis and disruption, and biopsies from patients with chronic graft-versus-host disease (beyond Day 90) showed more frequent portal fibrosis and bile duct dropout. Pattern assessment of coded biopsies showed that a histologic diagnosis of graft-versus-host disease had a positive predictive value of 86%, a sensitivity of 66% and a specificity of 91%. False-negative diagnoses occurred most frequently in biopsies obtained less than 4 weeks posttransplant, usually because bile duct abnormalities were not present. False-positive diagnoses of graft-versus-host disease occurred in nongraft-versus-host disease biopsies with periportal inflammation and proliferated bile ducts. However, biopsies of chronic graft-versus-host disease had more frequent dropout and disruption of bile duct epithelium than did biopsies of acute or chronic hepatitis.(ABSTRACT TRUNCATED AT 250 WORDS)