Associations between acute and chronic graft-versus-host disease

Allogeneic hematopoietic cell transplantation (allo-HCT) is an intensive and curative treatment for various hematologic malignancies and hematopoietic dysfunctions. Chronic graft-versus-host disease (GVHD) is 1 of the major complications after allo-HCT. In previous studies,^1-3 the incidence of chronic GVHD has been reported to be 15% to 50%, and chronic GVHD has been shown to have a detrimental effect on the quality of life, functional status, and survival outcomes.^4-13 

Previous studies have reported that recipient age,^14-20 donor age,^15,16,19 peripheral blood grafts,^{14-16,18,20-24} unrelated donor,^15,16 female donor–male recipient combination,^14-16,25-27 HLA disparity,^14,23 in vivo T-cell depletion,^14,17,28 and posttransplantation cyclophosphamide (PTCy)^29-31 were associated with the risk for chronic GVHD. Several studies have also reported that antecedent acute GVHD was an independent risk factor for chronic GVHD.^{14,16,17,20,22,23,32} For example, 1 of the largest studies conducted by the Seattle group that analyzed only patients who underwent myeloablative conditioning regimens showed that only grade 3 to 4 acute GVHD, but not mild acute GVHD, increased the subsequent risk for chronic GVHD. However, several characteristics of acute GVHD, including the severity and involvement of specific organs, could potentially affect the subsequent risk for chronic GVHD differently in modern GVHD prophylaxis.

In this nationwide, retrospective study, we examined the influence of acute GVHD on chronic GVHD according to the acute GVHD profile. An increased understanding of the association between clinical acute and chronic GVHD would contribute to elucidating the underlying pathophysiology of chronic GVHD and help to identify potential candidates for early intervention for chronic GVHD.

Clinical data were provided by the Transplantation Registry Unified Management Program of the Japanese Society for Transplantation and Cellular Therapy (JSTCT) and the Japanese Data Center for Hematopoietic Cell Transplantation.^33,34 

This study included adult patients with acute myeloid leukemia, acute lymphoblastic leukemia, myelodysplastic syndrome, myeloproliferative neoplasm, malignant lymphoma, or multiple myeloma who underwent their first allo-HCT between 2010 and 2021.

This retrospective study was approved by the data management committee of the Transplantation Registry Unified Management Program and by the institutional review board of the Jichi Medical University Saitama Medical Center and was performed in accordance with the Declaration of Helsinki and its later amendments. Written informed consent was obtained from all participants.

Acute GVHD and chronic GVHD were diagnosed and graded according to standard criteria.^35,36 The severity of chronic GVHD was also evaluated based on the National Institutes of Health (NIH) criteria.³⁷ The disease risk index and the hematopoietic cell transplantation–specific comorbidity index were classified according to previous reports.^38,39 In terms of donor type, a related donor with 6 of 6 antigen matches for HLA-A, -B, and -DR was considered to be an HLA-matched related donor. A related donor with 1 antigen mismatch of HLA was considered to be a mismatched related donor, and a related donor with ≥2 antigen mismatches of HLA was a haploidentical donor. An unrelated donor with 8 of 8 allelic matches of HLA-A, -B, -C, and -DRB1 was classified as an HLA-matched unrelated donor, whereas all other voluntary donors were classified as an HLA-mismatched unrelated donor. Conditioning regimens were distinguished according to the criteria of the Center for International Blood and Marrow Transplant Research.⁴⁰ In vivo T-cell depletion included the use of antithymocyte globulin and alemtuzumab.

The patients with complete information on all covariates were classified into the bone marrow and peripheral blood stem cell transplantation (BM/PB) cohort and the umbilical cord blood transplantation (UCB) cohort. Because UCB is not a common donor source outside of Japan,⁴¹ the UCB cohort was analyzed separately in the same manner as the BM/PB cohort as follows. The cumulative incidence of chronic GVHD that required systemic steroids was set as the primary end point to exclude mild cases that have minimal impact on survival and quality of life.^5,7,42 Outcomes were assessed using a landmark point at day 100 after allo-HCT. Among the eligible patients (n = 27 610), 6857 cases with either mortality or disease progression before day 100 were excluded from the analysis. The previous occurrences of acute GVHD were classified based on the maximum GVHD grades and organ involvement by day 100. In all multivariate analyses, the cause-specific Cox proportional hazard regression model was used to evaluate the subsequent risk for chronic GVHD. In addition to acute GVHD profiles, the following covariates were adjusted for in the multivariate analyses: recipient’s age, sex mismatch, disease risk index, hematopoietic cell transplantation–specific comorbidity index, donor type, donor source, conditioning regimen, GVHD prophylaxis, and in vivo T-cell depletion. P values for each acute GVHD grade (grade 1 vs 2 and grade 2 vs 3-4) were adjusted using the Bonferroni method.

A 2-tailed P value <.05 was considered to be statistically significant. All analyses were performed using R, version 4.2.3 (The Foundation for Statistical Computing, Vienna, Austria) and EZR, which is a graphical user interface for R (https://www.jichi.ac.jp/saitama-sct/SaitamaHP.files/statmedEN.html, Jichi Medical University Saitama Medical Center, Saitama, Japan).⁴³ 

This study included 14 618 cases in the BM/PB cohort and 6135 cases in the UCB cohort (Table 1). The median age was 50 years (range, 16-80) in the BM/PB cohort and 55 years (range, 16-79) in the UCB cohort. A total of 1969 patients (13.5%) in the BM/PB cohort and 213 (3.5%) in the UCB cohort underwent in vivo T-cell depletion. Among the patients who underwent allo-HCT from haplo-identical donors (n = 1715), 1165 patients (67.9%) were treated with PTCy. The BM/PB cohort included 5750 cases (39.3%) of no acute GVHD, 3403 (23.3%) of grade 1, 3884 (24.6%) of grade 2, and 1581 (10.8%) of grade 3 to 4 acute GVHD, whereas the UCB cohort included 2229 (36.3%) cases of no acute GVHD, 1199 (19.5%) of grade 1, 1941 (31.6%) of grade 2, and 766 (12.5%) of grade 3 to 4 acute GVHD. The median duration of follow-up for survivors was 52.3 months (range, 3.4-152.8) in the BM/PB cohort and 46.0 months (range, 3.4-153.5) in the UCB cohort.

The 4-year cumulative incidence of chronic GVHD that requires systemic steroids in the BM/PB cohort was 18.4% for grade 0, 23.1% for grade 1, 30.7% for grade 2, and 28.0% for grade 3 to 4 acute GVHD (Figure 1A), whereas the 4-year cumulative incidence in the UCB cohort was 8.0% for grade 0, 10.6% for grade 1, 17.9% for grade 2, and 18.3% for grade 3 to 4 acute GVHD (Figure 1B).

In the multivariate analyses that did not include acute GVHD as a covariate in the BM/PB cohort, younger age, female-to-male sex mismatch, donor type, PB, and the use of in vivo T-cell depletion were associated with the risk for chronic GVHD (Table 2). After adjusting for the baseline characteristics, the risk for chronic GVHD that requires systemic steroids increased with each increase in acute GVHD grade from 0 to 2 (grade 0 vs 1 [hazard ratio (HR), 1.32; 95% confidence interval (CI), 1.19-1.46; P < .001]; grade 1 vs 2 [HR, 1.41; 95% CI, 1.28-1.56; P < .001]), but the risk was similar between grade 2 and grade 3 to 4 acute GVHD (HR, 1.02; 95% CI, 0.91-1.15; P = 1.0; Table 2; Figure 2A). The interaction between acute GVHD and donor source (BM vs PB) was not statistically significant (interaction P value = .054).

Similarly, in the UCB cohort, the subsequent risk for chronic GVHD that requires systemic steroids significantly increased with each increase in acute GVHD grade from 0 to 2 (grade 0 vs 1 [HR, 1.38; 95% CI, 1.08-1.76; P = .0092]; grade 1 vs 2 [HR, 1.73; 95% CI, 1.39-2.15; P < .001]), but this trend was not observed when grade 2 acute GVHD was compared with grade 3 to 4 acute GVHD (HR, 1.14; 95% CI, 0.92-1.40; P = .460; Table 3; Figure 2B).

We also evaluated the how organ involvement during acute GVHD affected the risk for chronic GVHD that requires systemic steroids. Involvement of the skin, gastrointestinal system, and liver carried a similar risk for chronic GVHD in the BM/PB cohort (skin [HR, 1.28; 95% CI, 1.11-1.47; P < .001]; gut [HR, 1.24; 95% CI, 1.13-1.37; P <.001]; liver [HR, 1.24; 95% CI, 1.06-1.46; P = .0094]; supplemental Table 1). Moreover, the skin (HR, 1.61; 95% CI, 1.25-2.07; P < .001) and gut (HR, 1.26; 95% CI, 1.08-1.47; P = .003), but not the liver (HR, 0.93; 95% CI, 0.40-2.12; P = .863) during acute GVHD were associated with an increased risk for lung chronic GVHD. Involvement of any of these 3 organs during acute GVHD, particularly skin involvement, were also associated with the risk for extensive skin chronic GVHD⁴⁴ (skin [HR, 1.90; 95% CI, 1.51-2.39; P < .001]; gut [HR, 1.18; 95% CI, 1.03-1.34; P = .017]; liver [HR, 1.43; 95% CI, 1.14-1.78; P = .0017]).

Associations between the severity of acute and chronic GVHD were examined only in the BM/PB cohort because of a limited number of events in the UCB cohort. Lung chronic GVHD, manifesting as bronchiolitis obliterans syndrome, is considered a high-risk manifestation that is associated with poor outcomes. We first evaluated the impact of the grade of acute GVHD on the risk for lung chronic GVHD. The 4-year cumulative incidence of lung chronic GVHD was 6.7% in grade 0, 8.8% in grade 1, 9.6% in grade 2, and 10.1% in grade 3 to 4 acute GVHD (Figure 3A). In a multivariate analysis, the eventual HR for the risk for lung chronic GVHD increased with each increase in the grade of acute GVHD (grade 1: HR, 1.34; grade 2: HR, 1.48; grade 3-4: HR, 1.72), but the differences among grade 1 to 4 were not statistically significant (grade 1 vs 2 [HR, 1.10; 95% CI, 0.94-1.29; P = .468]; grade 2 vs 3-4 [HR, 1.16; 95% CI, 0.96-1.41; P = .252]; Figure 2C; supplemental Table 2).

Similarly, we evaluated the impact of acute GVHD on the risk for extensive skin chronic GVHD,⁴⁴ because this is another phenotype that increases the risk for nonrelapse mortality (NRM).^7,8 The 4-year cumulative incidence of extensive skin chronic GVHD is 7.5% in grade 0, 10.0% in grade 1, 15.0% in grade 2, and 14.3% in grade 3 to 4 acute GVHD (Figure 3B). The risk for extensive skin chronic GVHD was increased with each increase in acute GVHD grade from 0 to 2 (grade 0 vs 1 [HR, 1.34; 95% CI, 1.16-1.55; P < .001]; grade 1 vs 2 [HR, 1.57; 95% CI, 1.37-1.81; P < .001]), but the risk was equivalent between grade 2 and grade 3 to 4 acute GVHD (HR, 1.07; 95% CI, 0.91-1.25; P = 1.0; Figure 2D; supplemental Table 2).

We also evaluated the impact of the grade of acute GVHD on the severity of chronic GVHD assessed using the NIH criteria for the recent BM/PB cohort, included between 2019 and 2021, for which information on severity was available in our database. The 2-year cumulative incidence of moderate or severe chronic GVHD was 14.0% in grade 0, 13.5% in grade 1, 21.0% in grade 2, and 23.3% in grade 3 to 4 acute GVHD (Figure 3C). In a multivariate analysis, the risk for moderate or severe chronic GVHD significantly increased in patients with grade 2 (HR, 1.53; 95% CI, 1.25-1.88; P < .001) and those with grade 3 to 4 acute GVHD (HR, 1.94; 95% CI, 1.49-2.52; P < .001), but the difference between grade 2 and grade 3 to 4 was not statistically significant (HR, 1.26; 95% CI, 0.96-1.66; P = .179; Table 4; Figure 2E).

The 2-year cumulative incidence of severe chronic GVHD was 4.7% in grade 0, 5.5% in grade 1, 7.1% in grade 2, and 10.3% in grade 3 to 4 acute GVHD (Figure 3D). A multivariate analysis showed that grade 2 to 4 acute GVHD was significantly associated with a higher risk for severe chronic GVHD (grade 2: HR, 1.55; 95% CI, 1.09-2.21; P = .014; grade 3-4: HR, 2.64; 95% CI, 1.76-3.98; P < .001; Table 4; Figure 2F). Remarkably, we detected a significant difference between grade 2 and grade 3 to 4 acute GVHD in terms of the risk for severe chronic GVHD (HR, 1.70; 95% CI, 1.12-2.58; P = .025).

We evaluated the impact of acute GVHD on NRM and overall survival (OS) only among patients in the BM/PB cohort with chronic GVHD that required systemic steroids. The cumulative incidence of NRM within 4 years was 22.8% in grade 0, 21.9% in grade 1, 24.1% in grade 2. and 41.1% in grade 3 to 4 acute GVHD (supplemental Figure A), whereas 4-year OS was 65.8% in grade 0, 67.0% in grade 1, 65.6% in grade 2, and 51.2% in grade 3 to 4 acute GVHD (supplemental Figure B). In the multivariate analysis, only grade 3 to 4 had an adverse impact on NRM (HR, 2.28; 95% CI, 1.86-2.79; P < .001) and OS (HR, 1.80; 95% CI, 1.51-2.16; P < .001; supplemental Table 3).

This study elucidated the close association between acute and chronic GVHD. In this large Japanese cohort, baseline characteristics, including recipient's age, donor type, sex mismatch, PB, and the use of in vivo T-cell depletion, were associated with the risk for chronic GVHD that requires systemic steroids, which is consistent with previous studies.^14-16,28 The risk for chronic GVHD that requires systemic steroids increased with each increase in GVHD grade among grade 0, 1, and 2 to 4 in the BM/PB cohort, and these findings were also confirmed in the UCB cohort. Moreover, we found that the risk for severe chronic GVHD, as assessed by the NIH criteria, was significantly higher among those with grade 3 to 4 acute GVHD, indicating a strong association between the severity of acute and chronic GVHD.

Severe chronic GVHD remains one of the most morbid complications after allo-HCT.^1,8 Specifically, lung chronic GVHD, often referred to as bronchiolitis obliterans syndrome, is 1 of the most devastating subtypes of chronic GVHD.³⁷ Although several treatment options have become available for chronic GVHD, long-term survival remains unsatisfactory.^6-8,45,46 In particular, it is challenging to ameliorate organ dysfunction after the involved organs of chronic GVHD progress to fibrosis. Therefore, early interventions before progression to fibrosis are potential treatment strategies for chronic GVHD. Notably, the subsequent risk for severe chronic GVHD increased commensurately with the severity of acute GVHD, whereas the adjustment for the grade of acute GVHD showed no remarkable influence on other risk factors of chronic GVHD. Furthermore, we found that previous severe acute GVHD exhibited a notable adverse impact on survival outcomes among patients who developed chronic GVHD that required systemic steroids. This was not shown in the previous studies, which did not account for the severity of previous acute GVHD.^47,48 Although a history of acute GVHD alone is not a sufficient trigger to initiate early interventions, the inclusion of a history of acute GVHD could enhance the predictive accuracy of models for chronic GVHD that include baseline characteristics (eg, PB or sex mismatch) and potential biomarkers.⁴⁹ Therefore, these findings might help to identify potential candidates who could benefit from early treatment interventions.

Although the pathophysiological mechanism between acute and chronic GVHD has not been fully elucidated, thymus damage caused by acute GVHD might contribute to the development of chronic GVHD. Various studies using mouse models have suggested that damaged thymus caused by acute GVHD impairs negative selection, thereby inducing chronic GVHD.^50-54 Although the thymus in human adults is gradually replaced by fat, rebound thymic hyperplasia is observed after chemotherapy until the patient is ∼35 years old.⁵⁵ Furthermore, a recent retrospective study suggested that there is residual thymic function after thymus atrophy in older adults.⁵⁶ Our study demonstrated a reduced risk for chronic GVHD in younger patients (<36 years) before complete thymus atrophy, indicating that the reserved functionality of the thymus in this population might help to avoid the development of chronic GVHD. These findings partially support the notion that impaired thymus function caused by acute GVHD may underlie the mechanism of chronic GVHD. One of the other potential explanations is the involvement of dysbiosis. One study suggested that dysbiosis is associated with the incidence of chronic GVHD.⁵⁷ Dysbiosis caused by previous acute GVHD might induce subsequent chronic GVHD.⁵⁸ Prolonged use of calcineurin inhibitors for acute GVHD might also inhibit immune tolerance and induce chronic GVHD.^59,60 Although further investigations are required, our findings might serve as a basis for exploring the underlying pathophysiological mechanisms in acute and chronic GVHD.

This study had several limitations. First, some subgroup analyses, including the association between the severity of acute and chronic GVHD assessed by the NIH criteria in the UCB cohort, could not be performed because of the small sample size. Second, although several biomarkers have the potential to predict the future occurrence of chronic GVHD and its outcomes,^61-68 serum samples were not available in our database. Third, our registry lacked data on late acute GVHD, which is increasing according to the recent changes in practice.⁶⁹ For instance, late acute GVHD can sometimes develop into chronic GVHD overlapped by acute GVHD (progressive onset). Our study could not include such scenarios in the analysis. Fourth, the study included a limited number of patients who received PTCy as GVHD prophylaxis, all of whom underwent allo-HCT from haplo-identical donors because the use of PTCy in nonhaplo-identical settings was not permitted in Japan during the study period.

In conclusion, grade 2 to 4 acute GVHD was significantly associated with an increased risk for chronic GVHD. In addition, our study suggested that the risk for severe chronic GVHD correlated with the severity of acute GVHD. Although our findings should be validated in other retrospective and prospective cohorts, the preceding profiles of acute GVHD have the potential to help in predicting the future occurrence of chronic GVHD and its severity, which could facilitate the development of treatment strategies for chronic GVHD.

The authors are grateful for the work of all of the physicians and data managers at the centers that contributed valuable data on transplantation to the JSTCT. The authors thank all of the members of the Transplant Registry Unified Management committees at JSTCT for their dedicated data management.

Contribution: M. Tamaki and Y. Akahoshi contributed equally as primary investigators, and conceived the original idea, analyzed the data, and wrote the manuscript; K.M., Y.I., S.T., and J.K. advised on methods and wrote the manuscript; N.U., N.D., M. Tanaka, T.N., H.O., H.N., M.O., Y. Katayama, K.-i. M., and M.S. collected the data and revised the manuscript; F.I., Y. Kanda, and T.F. collected data, revised the manuscript, and were responsible for data management at JSTCT; Y. Atsuta managed the unified registry database and revised the manuscript; J.K. was responsible for this project of the JSTCT/GVHD Working Group; and all authors approved the final version of the manuscript.

Conflict-of-interest disclosure: M. Tamaki reports receiving honoraria from Astellas Pharma and Kyowa Kirin. Y. Kanda reports receiving honoraria from Pfizer, Sumitomo Pharma, Novartis Pharma, Sanofi K.K., Chugai Pharmaceutical, SymBio Pharmaceuticals, Bristol Myers Squibb, Janssen Pharmaceutical K.K., Asahi Kasei Pharma, MSD, Astellas Pharma, Kyowa Kirin; and subsidies from Sumitomo Pharma, Chugai Pharmaceutical, Kyowa Kirin, and Eisai. The remaining authors declare no competing financial interests.

Correspondence: Masaharu Tamaki, Division of Hematology, Jichi Medical University Saitama Medical University Saitama Medical Center, 1-847 Amanuma-cho, Omiya-ku, Saitama 330-8503, Japan; email: m.tamaki.221@gmail.com; and Yu Akahoshi, The Tisch Cancer Institute, Icahn School of Medicine at Mount Sinai, 1 Gustave L. Levy Place/Box 1128, New York, NY 10029; email: akahoshiu@gmail.com.