Systemic steroid treatment of grade I acute GVHD increases steroid-refractory GVHD and NRM

Abstract

The use of systemic corticosteroids is recommended only for grades II to IV acute graft-versus-host disease (GVHD), but many patients are often treated at the onset of grade I in real-world practice. We retrospectively analyzed outcomes of 1143 patients with grade I GVHD from 24 Mount Sinai Acute GVHD International Consortium (MAGIC) transplant centers. Approximately half received “up-front” systemic steroids, whereas the remainder received topical steroids alone; GVHD progressed to grade II to IV in only one-third of these patients. Up-front steroids decreased the risk for grade II to IV GVHD but not grade III to IV acute GVHD or the need for a second line of treatment. Systemic steroids were equally effective at treating grade I or II GVHD; however, up-front therapy exposed 3 times more patients to systemic steroids, with a corresponding threefold increase in GVHD that required additional therapy and a threefold rise in infection-related deaths. In multivariable analyses, up-front steroids were significantly associated with higher nonrelapse mortality (NRM; P = .026). MAGIC biomarkers identified most patients with grade I GVHD as low risk and unlikely to progress to severe GVHD regardless of treatment. Nevertheless, low-risk patients who received up-front steroids experienced a threefold increase in infectious deaths compared with those treated topically. This study supports the consensus recommendation of topical therapy for patients with low-risk grade I GVHD, a strategy that can be supported by biomarkers to avoid unnecessary steroid exposure and increased NRM.

Introduction

Acute graft-versus-host disease (GVHD) is a major cause of morbidity and nonrelapse mortality (NRM) after allogeneic hematopoietic cell transplantation (HCT).^1–7 For the past few decades, high-dose corticosteroid remains the first-line therapy for grades II to IV acute GVHD, whereas systemic treatment has not been recommended for grade I.⁸ In spite of this recommendation, grade I acute GVHD is often treated with systemic steroids in real-world practice in many cases in order to preempt the occurrence of severe GVHD that might become steroid refractory (SR),^2,9,10 and a prospective registry trial from the Blood and Marrow Transplant Clinical Trials Network observed the frequent use of systemic treatment of grade I before progression to grade II in many transplant centers.¹¹

A randomized controlled phase 3 study of systemic steroid treatment vs observation for grade I acute GVHD showed that systemic steroids significantly reduced the primary end point, the incidence of grades II to IV acute GVHD (33% vs 50%).¹² However, systemic steroid treatment increased the risk of infections and did not significantly reduce grades III to IV acute GVHD (10% vs 13%). The 5-year NRM in the treatment arm was slightly higher but did not reach statistical significance (26% vs 20%; P = .2).

We undertook this retrospective analysis to understand whether the systemic treatment for grade I GVHD reduced the incidence of severe GVHD and whether it improved long-term outcomes. We used the Mount Sinai Acute GVHD International Consortium (MAGIC) database and biorepository that prospectively monitored the natural history of acute GVHD and collected serum samples. This study evaluated the long-term outcomes of >1100 patients treated with 2 different approaches for grade I GVHD: “up-front” use of systemic steroids vs a “watch-and-wait” strategy that delayed systemic steroids until GVHD had progressed to grade II.

Methods

Patient selection

Clinical data and serum samples were obtained from the MAGIC database and biorepository, which consisted of 24 HCT centers in North America, Europe, and Asia. MAGIC collected clinical data of acute GVHD symptoms and treatment with serum samples using a prospective specimen collection, retrospective blinded evaluation study design.^{13 ,14} Participating sites reported clinical information and symptoms on a weekly basis through day 100 after HCT and for 4 weeks after the initiation of systemic treatment for acute GVHD and then less frequently up until 2 years after HCT. Informed consent from an institutional review board–approved protocol was obtained from all participants in accordance with the Declaration of Helsinki.

We included both pediatric and adult patients who received their first HCT between 2014 and 2022 and who were diagnosed with grade I acute GVHD. Acute GVHD was diagnosed and staged according to the MAGIC consensus guidance.¹⁴ Briefly, an isolated skin rash (<50% body surface area) that physicians judged to be GVHD was considered grade I acute GVHD with or without biopsy confirmation. We defined “up-front” steroids and “watchand-wait” based on the use of systemic corticosteroids before the progression to grades II to IV acute GVHD. HCT-specific comorbidity index (HCT-CI) scores and intensity of conditioning regimens were defined as per published criteria.^15,16 Disease risk was defined as previously reported.² We defined death from GVHD as mortality occurring in the presence of active GVHD symptoms, whereas death from infection included only patients whose GVHD symptoms had completely resolved. SR GVHD was defined as the initiation of second-line treatment because of stable or progressive clinical symptoms.

Serum samples

Serum samples within 3 days of GVHD diagnosis were prospectively collected, cryopreserved, and shipped to a central laboratory. Serum concentrations of suppressor of tumorigenicity 2 (ST2)¹⁷ and regenerating islet-derived protein 3α (REG3α)¹⁸ were retrospectively analyzed by enzyme-linked immunosorbent assays, as previously reported. The MAGIC algorithm probability (MAP) was calculated as a single value between 0.001 and 0.999 following the formula: log[−log(1 – MAP)] = −11.263 + 1.844(log₁₀ST2) + 0.577(log₁₀REG3α).^19,20 We calculated Ann Arbor (AA) scores according to the validated thresholds (AA1 < 0.141; 0.141 ≤ AA2 < 0.291; AA3 ≥ 0.291).^{2,6,19,21–26}

Statistical analysis

Because 90% of the patients who received up-front steroids for grade I GVHD started treatment within 7 days of diagnosis, we set a landmark at day 7 after grade I GVHD diagnosis to visualize the impact of up-front steroids on long-term outcomes. All outcomes were censored 12 months after diagnosis of grade I GVHD. We calculated the cumulative incidence using the Gray method.²⁷ The competing risks for NRM were primary disease relapse and second allogeneic HCT; the competing risks for infectious death were noninfectious death, primary disease relapse, and second allogeneic HCT; the competing risks for relapse were death without relapse and second allogeneic HCT; the competing risks for GVHD were primary disease relapse and death without GVHD. We used the Kaplan-Meier method and the log-rank test to estimate and compare overall survival (OS) probabilities. In the multivariable analyses, cause-specific Cox proportional hazards regression models were used,^28,29 with up-front steroids treated as a time-dependent covariate.³⁰ In addition, the frailty proportional hazards model with random effects³¹ was used to adjust for the center effect, which accounts for the unobserved heterogeneity in clinical practices across each transplant center. Categorical variables were compared using the Fisher exact test, and continuous variables were compared using the Mann-Whitney U test.

In the multivariable analysis assessing the risk factors for NRM and GVHD, we included the following variables: recipient age at HCT, sex mismatch, donor type, GVHD prophylaxis, HCT-CI, conditioning intensity, year of HCT, GVHD skin stage at diagnosis, time from HCT to GVHD diagnosis, and the center effect. Additionally, we accounted for primary disease and disease risk as confounders in the assessment of relapse and OS.

All P values were 2-sided, and the α error was set at 0.05. Statistical analyses were performed with R or Easy R (EZR) version 1.61 (Jichi Medical University Saitama Medical Center), which is a graphical user interface for R (version 4.2.2; R Foundation for Statistical Computing, Vienna, Austria).³²

Results

Patient characteristics

This study evaluated 1143 patients diagnosed with grade I acute GVHD at the onset of disease; 591 (51.7%) received systemic steroids, and 552 (48.3%) were observed or treated with topical steroids without additional systemic therapy (Table 1). There were no significant differences between groups in age, donor type, GVHD prophylaxis, year of transplant, or comorbidities. Patients diagnosed with GVHD in Europe, earlier time after HCT or exhibiting stage 2 rather than stage 1 skin GVHD at diagnosis, were more likely to receive systemic steroid treatment rather than topical steroids or observation. The median duration from diagnosis of grade I GVHD to up-front steroids was 0 days (interquartile range [IQR], 0–1; range, 0–72). The median initial dose of corticosteroids was 0.8 mg/kg prednisolone or equivalent (IQR, 0.50–1.0). All patients were followed up for at least 1 year after initiation of treatment.

Table 1.

Patient characteristics

	Up front	Watch-and-wait
	n = 591	n = 552	P value
Age at HCT (y), median (IQR)	55 (36–63)	56 (37–64)	.597
Recipient age category (y), n (%)			.887
<18	84 (14.2)	75 (13.6)
18–54	206 (34.9)	188 (34.1)
≥55	301 (50.9)	289 (52.4)
Sex mismatch, n (%)			.249
Female to male	490 (83.2)	444 (80.4)
Other	99 (16.8)	108 (19.6)
Primary disease, n (%)			.001
Acute leukemia	302 (51.1)	329 (59.6)
MDS/MPN	171 (28.9)	115 (20.8)
Malignant lymphoma	48 (8.1)	28 (5.1)
Other	70 (11.8)	80 (14.5)
Disease risk, n (%)			.882
Standard	472 (79.9)	443 (80.3)
High	119 (20.1)	109 (19.7)
Donor type, n (%)			.146
HLA-matched donors	433 (73.3)	417 (75.5)
HLA-mismatched unrelated donors	70 (11.8)	44 (8.0)
Haploidentical donors	71 (12.0)	70 (12.7)
Umbilical cord blood	17 (2.9)	21 (3.8)
PTCy, n (%)			.877
No	486 (82.2)	456 (82.6)
Yes	105 (17.8)	96 (17.4)
In vivo or ex vivo T-cell depletion, n (%)			.203
No	348 (58.9)	346 (62.7)
Yes	243 (41.1)	206 (37.3)
HCT-CI, n (%)			.157
0–2	401 (67.9)	396 (71.7)
≥3	190 (32.1)	156 (28.3)
Donor source, n (%)			.315
Bone marrow	109 (18.4)	117 (21.2)
Peripheral blood	465 (78.7)	414 (75.0)
Umbilical cord blood	17 (2.9)	21 (3.8)
Conditioning, n (%)			.338
MAC	351 (59.4)	312 (56.5)
RIC	240 (40.6)	240 (43.5)
Transplant site, n (%)			<.001
North America	312 (52.8)	359 (65.0)
Europe	269 (45.5)	178 (32.2)
Asia	10 (1.7)	15 (2.7)
Year of HCT, n (%)			.341
2014–2018	340 (57.5)	302 (54.7)
2019–2022	251 (42.5)	250 (45.3)
Duration from HCT to GVHD Dx (d), median (IQR)	29 (20.0–43.5)	35 (25–57)	<.001
Skin stage at diagnosis, n (%)			<.001
Stage 1	230 (38.9)	392 (71.0)
Stage 2	361 (61.1)	160 (29.0)
Sample available at Dx, n (%)			<.001
No	108 (18.3)	164 (29.7)
Yes	483 (81.7)	388 (70.3)
Biomarker risk at Dx, n (%)			.140
AA1	376 (77.8)	318 (82.0)
AA2	83 (17.2)	48 (12.4)
AA3	24 (5.0)	22 (5.7)

Dx, diagnosis; MAC, myeloablative conditioning; MDS, myelodysplastic syndrome; MPN, myeloproliferative neoplasm; PTCy, posttransplant cyclophosphamide; RIC, reduced intensity conditioning.

Long-term outcomes of different treatment approaches

Only 184 of the 552 patients (33.3%) in the watch-and-wait group were treated with systemic steroids (median, 14 days; IQR, 7–42; range, 1–279) because grade I GVHD resolved or never progressed in two-thirds of patients (Figure 1A). Systemic steroids were equally effective in resolving GVHD symptoms, whether they were initiated for grade I GVHD or grade II after progression from grade I; in both cohorts, ~20% of the patients developed SR GVHD and received subsequent treatment for stable (n = 49) or worsening (n = 106) disease. Thus, because 3 times as many patients were systemically treated in the up-front group, the incidence of SR GVHD was threefold higher than in the watch-andwait group (Figure 1B). Similar results were observed when patients who received second-line therapy for reasons other than lack of response (eg, for steroid sparing purposes) were included (supplement Figure 1). The maximum stage and overall grade of GVHD for patients in both cohorts are listed in supplemental Table 1. There was no significant difference between groups other than fewer patients who developed upper gastrointestinal and stage 3 skin disease in the up-front group.

Figure 1. GVHD treatment dynamics.

(A) The 6-month cumulative incidence of systemic steroid use in the up-front group was 100% and in the watch-and-wait group was 29.3% (95% CI, 25.6–33.2). (B) The 6-month cumulative incidence of SR GVHD, defined as the use of second-line agents for stable or progressive acute GVHD, was 6.2% (95% CI, 4.4–8.5) in the watch-and-wait group and 19.5% (95% CI, 16.4–22.8) in the up-front group (P < .001).

There was no difference in the overall incidence of chronic GVHD at 1 year between the up-front and watch-and-wait groups (34.2%; 95% confidence interval [CI], 30.3–38.1 vs 32.1%; 95% CI, 28.1–36.2, respectively; P = .32). But the threefold increase in SR GVHD in patients whose grade I GVHD was treated with up-front systemic steroids was accompanied by a threefold increase in infectious deaths within the first year after diagnosis, which correlated with a significant increase in NRM (Figure 2A–B). Likewise, causes of nonrelapse death differed significantly between groups (Table 2). We confirmed the same results when we applied a landmark analysis that accounted for immortal time bias, that is, a patient must have remained alive from GVHD diagnosis to the start of treatment to be classified as having received up-front steroids (supplemental Table 2).

Figure 2. One-year GVHD treatment outcomes.

(A) The cumulative incidence of infectious death was 1.3% (95% CI, 0.6–2.6) in the watch-and-wait group and 4.7% (95% CI, 3.2–6.7) in the up-front group (P = .001). (B) The cumulative incidence of NRM was 6.9% (95% CI, 4.9–9.3) in the watch-and-wait group and 12.6% (95% CI, 10.0–15.5) in the up-front group (P = .002).

Table 2.

Causes of nonrelapse death

	Up-front (n = 82)	Watch-and-wait (n = 50)	P value
Acute GVHD, n (%)	27 (32.9)	19 (38.0)
Infection, n (%)*	32 (39.0)	8 (16.0)
Chronic GVHD	6 (7.3)	7 (14.0)
Others, n (%)†	17 (20.7)	16 (32.0)
Overall difference between the 2 groups			.029

^*Infection death without active acute GVHD was considered an infectious death.

^†Other causes of death are organ failure, (n = 19), secondary malignancy (n = 6), and unknown (n = 8).

Neither the extent of skin rash (stage 1 vs 2) nor the interval to initiation of treatment (≤7 vs >7 days after diagnosis) nor the initial steroid dose (≤1.0 mg/kg vs >1.0 mg/kg) affected the incidence of NRM of patients treated for grade I GVHD (supplemental Figure 2). Analysis of GVHD severity at the time of second-line treatment showed that one-third of patients received second-line treatment for grade I GVHD, leading to a 12-month NRM of 12.5% (supplemental Table 3). The incidences of SR GVHD at 6 months and of NRM at 1 year were thus significantly higher in the up-front treatment group but did not differ between stages 1 and 2 skin disease (supplemental Table 4), again demonstrating that the extent of the skin rash within grade I GVHD did not materially affect long-term outcomes. In a multivariable analysis, in which up-front steroid use was treated as a time-dependent covariate, it remained significantly associated with an increased risk of NRM (hazard ratio, 1.63; 95% CI, 1.06–2.50; P = .026; Table 3). Advanced recipient age, HCT from umbilical cord blood, higher HCT-CI scores, and myeloablative conditioning regimen were also associated with higher NRM (supplemental Table 5).

Table 3.

The impact of up-front steroids in multivariable analyses

NRM	HR (95% CI)	P value
Up-front steroids	1.63 (1.06–2.50)	.026
Maximum grades II to IV acute GVHD	0.71 (0.56–0.89)	.003
Maximum grades III to IV acute GVHD	1.03 (0.69–1.52)	.899
Chronic GVHD	1.08 (0.85–1.37)	.508
Relapse	0.64 (0.47–0.88)	.006
OS	1.10 (0.81–1.48)	.546

Up-front steroids were treated as a time-dependent covariate. In all analyses, we adjusted recipient age at HCT, sex mismatch, donor type, GVHD prophylaxis, HCT-CI, conditioning intensity, year of HCT, GVHD skin stage at diagnosis, time from HCT to GVHD diagnosis, and the center effect. In the analyses for relapse and OS, we also adjusted for primary disease and disease risk.

HR, hazard ratio.

Treatment of grade I GVHD with systemic steroids reduced the incidence of grade II GVHD but not the incidence of grades III to IV acute GVHD or chronic GVHD (Table 3; supplemental Tables 6–8). Surprisingly, up-front steroids were associated with a decreased risk of relapse; therefore, despite the greater risk of NRM, OS between the 2 groups was similar (Table 3; supplemental Tables 9 and 10).

Biomarker prediction of grade I GVHD outcomes

Serum samples were available for 871 of the 1143 (76.2%) patients at the onset of grade I GVHD for analysis of MAGIC biomarkers; there were no significant differences in the 12-month NRM between patients with and without serum samples (10.5% vs 7.8%; P = .232). As shown in Figure 3A, 80% of the patients had low-risk biomarkers (AA1), and 15% and 5% of the patients represented the intermediate (AA2) and high (AA3) risk, respectively. Grade III to IV acute GVHD significantly increased with each increment in AA score, as did the incidence of 12-month NRM (AA1, 7.5%; AA2, 15.7%; AA3, 39.2%; P < .001; Figure 3A–B). Infectious deaths among the minority of patients with high-risk biomarkers were not affected by up-front steroid treatment (data not shown), but among the 80% of the patients with low-risk biomarkers, infectious deaths were threefold higher in the up-front steroid treatment group (Figure 4). Therefore, low-risk MAGIC biomarkers identified patients who were unlikely to progress to severe GVHD and in whom up-front steroid treatment was associated with increased infectious deaths.

Figure 3. One-year grade III to IV acute GVHD and NRM predicted by the MAGIC biomarkers.

(A) Cumulative incidence of grade III to IV acute GVHD from the diagnosis of grade I GVHD: AA1, 8.8% (95% CI, 6.9–11.1); AA2, 13.8% (95% CI, 8.6–20.4); and AA3, 43.5% (95% CI, 28.8–57.3). (B) Cumulative incidence of NRM after GVHD diagnosis: AA1, 7.5% (95% CI, 5.7–9.7); AA2, 15.7% (95% CI, 10.0–22.6); and AA3, 39.2% (95% CI, 25.1–53.1).

Figure 4. Infectious deaths in patients with AA1 biomarkers.

The 12-month cumulative incidence of infectious death in patients with AA1 at grade I GVHD was 0.8% (95% CI, 0.2–2.3) in the watch-and-wait group and 4.3% (95% CI, 2.4–7.0) in the up-front group (P = .004). A landmark was set at day 7 after GVHD diagnosis, by which time >90% of the systemic steroids had been started; 31 patients who were treated with up-front steroids 7 days after GVHD diagnosis were included in the watch-and-wait group.

Discussion

Despite consensus recommendations to withhold systemic corticosteroid therapy for grade I GVHD (limited skin disease), many such patients are nevertheless treated systemically.^2,9–11 In this large cohort analysis, we observed that this “up-front” approach did not prevent progression to severe or SR acute GVHD and was associated with a significant increase in infectious deaths. MAGIC biomarkers were able to further stratify the risk of GVHD progression and NRM in this clinically defined low-risk group. These data support the recommended watch-and-wait strategy for patients with low-risk grade I GVHD, an approach that can be refined by biomarkers to avoid unnecessary steroid exposure.

There are several possible explanations for the treatment of grade I acute GVHD with systemic steroids in general clinical practice. First, clinicians cannot closely monitor outpatients daily, and treatment of grade I may prevent a delay in treatment of grade II if the patient is unavailable for examination. The fact that patients with stage 2 skin rash were more likely to be treated before progression to grade II GVHD skin stage than patients with a stage 1 skin rash is consistent with this concern. But our data show that skin stage did not alter long-term outcomes, with the same NRM observed in both groups (supplemental Figure 2). Second, until now, there has also been little evidence illustrating a downside for up-front systemic treatment. This analysis provides such evidence, demonstrating that up-front treatment of grade I is associated with an increase in infection-related deaths and overall NRM compared with the conservative watch-and-wait approach. Our data are consistent with results of the earlier Italian study in which the difference in NRM in the up-front treatment group was higher but did not reach statistical significance because of the smaller number of patients (5-year NRM, 26% vs 20%; P = .2).¹² Although systemic steroid treatment for grade I may sometimes be used to prevent a delay in treatment of subsequent grade II because of patient unavailability, we conclude that the current recommendations are correct and that delaying systemic treatment until progression to grades II to IV GVHD is the better approach.

A well-designed, placebo-controlled, randomized phase 2 trial of almost 200 patients also showed that systemic steroids do not lower the incidence of severe acute GVHD.³³ In this analysis, systemic steroids were equally effective in resolving GVHD symptoms, whether they were initiated for grade II or grade I GVHD; systemic treatment is not more effective for grade I than grade II, and ~20% of the cases do not respond to either approach and require additional lines of treatment. But since only one-third of patients with grade I GVHD progressed to grade II and received systemic steroids, the incidence of SR GVHD was threefold higher in the up-front treatment group. Two complementary reasons may help explain this surprising observation. First, second-line therapy was administered to one-third of the up-front group for stable “unresponsive” grade I disease, which may have, in fact, been unnecessary, but which increased the burden of immunosuppression and its infectious risks. Second, it is possible that systemic steroids actually make GVHD worse. Recent research has highlighted the damage that steroids inflict on intestinal stem cells, key cellular targets of GVHD, thus widening the “tissue-repair gap” that must be closed for mucosal healing to occur.^34,35 Thus, although steroids reduce the inflammatory effectors of GVHD, their damage to target tissues may contribute to the emergence of steroid-resistant disease. Furthermore, steroids are known to impair immune reconstitution^36,37 and might impede the development of regulatory T cells that could help control GVHD.^38,39 In this regard, it is worth noting that the incidence of grade III/IV GVHD requiring second-line treatment was twice as great in the up-front group compared with the watch-andwait group (supplemental Table 3).

The infectious toxicities of systemic steroids are well known,^40–43 but, to our knowledge, this study is the first to demonstrate an increased risk of infectious death and overall NRM when they are used to treat grade I GVHD. We also show that biomarkers can confirm the low long-term risk status of individual patients,^{2,6,19,21,24,25,44} which may provide reassurance for the decision to use topical steroids first and withhold systemic therapy for such a patient unless GVHD progresses. We have demonstrated that patients at low risk by both clinical and biomarker criteria can be successfully treated with a nonsteroidal approach that reduces infectious complications⁴¹ and a second trial using the oral JAK1/JAK2 inhibitor ruxolitinib in this setting is currently in progress (ClinicalTrials.gov identifier: NCT05090384).

The strengths of this cohort analysis are the large size of the data set, the accuracy and granularity of the clinical GVHD data, and the prospective specimen collection, retrospective blinded evaluation study design of prospective serum collection and evaluation of biomarkers. This analysis also has important limitations. First, this is real-world evidence and not a randomized trial with prescribed treatment plans; there were several differences between the 2 groups in relevant variables, such as the time of GVHD onset, the skin stage at the time of treatment, and the underlying primary disease, any of which could have biased the observed outcomes, although none of them were significant in the multivariable analyses. Clinical judgment, based on “soft” variables such as patient appearance, behavior, and vital signs that contribute to overall patient status, may be important but are notoriously difficult to quantify for comparative purposes. Second, we observed an increase in infectious deaths, but detailed data regarding severe infections and delayed immune reconstitution were not available. Infection status at GVHD onset was also not available, although active infections would probably diminish the likelihood of up-front steroid treatments, a “downward bias” that could underestimate the risk of infectious death by this approach.⁴⁵ Third, an important limitation relates to the surprising observation that up-front treatment with systemic steroids was associated with a decreased risk of subsequent relapse. Clinicians usually avoid additional immunosuppressive agents in patients, fearing a theoretical diminishment of the graft-versus-leukemia effect and an increased risk of relapse.⁴⁶ Thus, clinicians may have a lower threshold to treat grade I GVHD if they consider the risk of relapse to be low. Similarly, they may choose to delay systemic steroid treatment of grade I GVHD in order to maximize a graft-versus-leukemia effect if they consider the risk of relapse to be high. Unfortunately, laboratory determinants of relapse risk, such as minimal residual disease status^47–50 and the molecular profile of tumor cells^51–54 were not available in this study; these unadjusted confounders may have also contributed to an imbalanced risk of relapse between groups.

In summary, despite the widespread use of systemic corticosteroids for grade I GVHD in real-world clinical practice, our findings show that up-front therapy is associated with more SR acute GVHD and more infectious deaths, thus supporting current recommendations to start systemic treatment only for grades II to IV acute GVHD. In addition, GVHD biomarkers at the onset of symptoms can further stratify this clinically low-risk group, identifying which patients are at risk for the future development of severe GVHD and increased NRM and thus individualizing treatment for patients with grade I GVHD.

Key Points.

• The use of systemic steroids for grade I acute GVHD was associated with more steroid-resistant GVHD and increased risk of infectious NRM.

• MAGIC biomarkers at grade I can confirm the low-risk status of most patients and support the decision to treat with topical steroids alone.