Differences in acute graft-versus-host disease (GVHD) severity and its outcomes between Black and White patients

Abstract

Background

Acute graft-versus-host disease (GVHD) is a significant complication following hematopoietic stem cell transplantation (HCT). Although recent advancements in GVHD prophylaxis have resulted in successful HCT across HLA barriers and expanded access to HCT for racial minorities, less is known about how race affects the severity and outcomes of acute GVHD.

Objectives

This study examines differences in the clinical course of acute GVHD and the prognostic value of GVHD biomarkers for Black and White recipients.

Study Design

We conducted a retrospective analysis of patients in the Mount Sinai Acute GVHD International Consortium (MAGIC) database who underwent HCT between 2014 and 2021 to describe the difference in clinical course of acute GVHD and significance of GVHD biomarkers between Black and White recipients. We used propensity score matching to generate a 1:3 matched cohort of 234 Black patients and 702 White patients with similar baseline characteristics.

Results

In the first year after HCT Black patients experienced a higher cumulative incidence of grade III-IV acute GVHD (17% vs 12%, P = 0.050), higher non-relapse mortality (NRM; 18% vs 12%, P = 0.009), and lower overall survival that trended toward statistical significance (73% vs 79%, P = 0.071) compared to White patients. The difference in NRM in the first year was even greater among Black patients who developed GVHD than White patients (24% vs 14%, P = 0.041). The distribution of low, intermediate, and high MAGIC biomarker scores at the time of treatment was similar across racial groups (P = 0.847), however, Black patients with high biomarker scores experienced significantly worse NRM than White patients (71% vs 32%, P = 0.010).

Conclusion

Our data indicate that Black patients are at a higher risk of NRM following HCT, primarily from a higher incidence of severe GVHD. Serum biomarkers at treatment initiation can stratify patients for risk of NRM across races, however Black patients with high biomarker scores had a significantly greater NRM risk. These results suggest a need for strategies that mitigate the higher risk for poor GVHD outcomes among Black patients.

Introduction

Allogeneic hematopoietic cell transplantation (HCT) is increasingly used as a curative treatment option for both malignant and nonmalignant hematologic disorders. [1, 2] Healthcare disparities regarding access to allogeneic HCT is a significant concern especially for African Americans. [3] Multiple studies have demonstrated a lower rate of allogeneic HCT for Black patients compared to White patients. [4–9], often due to the lack of an HLA-matched donor. [10–14] Recent advancements in GVHD prophylaxis, such as post-transplant cyclophosphamide (PTCy) and abatacept [15–20], have resulted in successful HCT across HLA barriers and expanded access to HCT for racial minorities. [21] Thus understanding the influence of race on HCT outcomes is of growing significance because of the anticipated increase in HCT among Black patients. [22]

There are several well known risk factors for acute graft-versus-host disease (GVHD) such as HLA-match [23–26], conditioning intensity [24–26], CMV infection [26], and GVHD prophylaxis. [24–26] Some studies have suggested that Black race is a potential risk factor for acute GVHD [27, 28], whereas other studies have not. [14, 29, 30] In this retrospective analysis we used the Mount Sinai Acute GVHD International Consortium (MAGIC database and biorepository) to examine differences in the clinical course of acute GVHD and the prognostic value of GVHD biomarkers for Black and White recipients.

Methods

Study design and patient selection

We prospectively collected data and serum samples from 21 HCT centers regarding the natural history of GVHD using a PRoBE study design. [31–33] Informed consent was obtained from all participants in accordance with the Declaration of Helsinki.

In this analysis, we included all Black and non-Hispanic White patients who received their first HCT for malignant and nonmalignant hematological disorders between 2014 and 2021 in MAGIC centers. Race and ethnicity were self-reported.

Definitions

Acute and chronic GVHD was diagnosed and staged according to standard published criteria. [33, 34] HCT-specific comorbidity index (HCT-CI) scores, conditioning regimen intensity, and Minnesota risk were defined as previously reported. [35–37] HLA match was evaluated based on allele-level comparison of HLA-A, -B, -C, and –DRB1.

Sample collection

Serial serum samples were prospectively collected, shipped, and cryopreserved at a central laboratory. Serum concentrations of suppressor of tumorigenicity-2 (ST2) [38] and regenerating islet-derived protein 3-α (REG3α) [39] were measured by enzyme-linked immunosorbent assays, as previously described. [40–42] The MAGIC Algorithm Probability (MAP) was calculated as a single value between 0.001 and 0.999 according to the formula: log[−log(1 – MAP)] = − 11.263 + 1.844(log₁₀ST2) + 0.577(log₁₀REG3α). We used the validated MAP thresholds to categorize patients into risk strata by Ann Arbor (AA) scores (AA1 < 0.14; 0.14 ≤ AA2 < 0.29; AA3 ≥ 0.29). [25, 40, 41, 43–46]

Statistical analysis

Categorical variables were compared with the Fisher's exact test and continuous variables were compared by the Mann-Whitney U test. The univariate Gray’s method and multivariate Fine and Gray method were used to evaluate the impact of race on GVHD or NRM. Relapse was considered as a competing risk for NRM and death or relapse before GVHD were considered competing risks for GVHD. The probability of overall survival (OS) was estimated with the Kaplan-Meier method.

A matched cohort was created at a 1:3 ratio using the propensity score that employed the nearest neighbor matching method. Calipers were set standardly with a width equal to 0.2 of the standard deviation of the estimated propensity score. [47] The following variables were included in the propensity score model: recipient age at HCT, sex mismatch, donor source, HCT-CI, GVHD prophylaxis, conditioning regimen, in vivo T-cell depletion, and year of HCT.

All tests were 2-sided and P-values ≤ 0.05 were considered statistically significant. All statistical analyses were performed with the R statistical software version 3.2.2.

Results

Patient characteristics

Data were evaluated for 3795 patients (Black, n = 279: White, n = 3516) who underwent allogeneic HCT during the study period. There were several significant differences between Black and White patients in their baseline characteristics: recipient age at HCT, sex mismatch, primary disease, donor type, donor source, and GVHD prophylaxis (Table S1 and Table S2). We could not confirm the assumption of linearity among covariates and GVHD outcomes, particularly for age where the differences between median age for Black and White patients was large (28 vs 57 years, P < 0.001). Therefore, we selected propensity score matching, which does not require the assumption of linearity, to correct for imbalances of baseline characteristics between the two races. [48] We generated a 1:3 matched cohort composed of 234 Black patients and 702 White patients who shared similar baseline characteristics except for primary disease (Table 1). The median recipient age at HCT for the matched cohorts was 29 years in Black patients and 31 years in White patients (P = 0.747). More Black (33%) than White patients (19%) underwent HCT for non-malignant disorders because sickle cell disease (SCD) was a common HCT indication for Black patients. The degree of HLA match for HLA-A, B, C, and DR was similar between Black and White patients. There were also no significant differences in HLA-compatibility for HLA-DQ and DP in the subset of patients who received HCT from HLA-matched related or unrelated donors and who had extended HLA-typing available (Table 1). The median follow-up period for survivors was 701 days in Black and 712 days in White patients.

Table 1.

Patient characteristics in matched cohort

	Black n = 234 (%)	White n = 702 (%)	P value
Hispanic Ethnicity	24 (10)	0 (0)
Median age at HCT [IQR]	29 [11, 54]	31 [12, 57]	0.747
Age at HCT
≤18	87 (37)	249 (36)	0.391
18–55	92 (39)	256 (37)
≥55	55 (24)	197 (28)
Sex Mismatch			0.656
Other	176 (75)	540 (77)
Female to Male	58 (25)	162 (23)
Primary Diagnosis			<0.001
AML/ALL	93 (40)	355 (51)
Malignant Lymphoma	26 (11)	51 (7)
MDS/MPN	35 (15)	141 (20)
Other Malignant	2 (1)	23 (3)
Sickle Cell Disease	49 (21)	0 (0)
Other Non-Malignant	29 (12)	132 (19)
HLA (8 loci) and Donor Type
Matched Related	61 (26)	229 (32)
Matched Unrelated	43 (18)	126 (18)	0.382
Mismatched	34 (15)	94 (13)
Haploidentical	73 (31)	200 (29)
Umbilical cord blood	23 (10)	53 (8)
HLA-DQ (10 loci) *
Match	68 (65)	290 (82)	0.170
Mismatch	2 (2)	2(1)
Not available	34 (33)	63(18)
HLA-DQ/DP (12 loci) *
Match	20 (19)	53 (21)	0.838
Mismatch	13 (13)	39 (15)
Not available	71 (68)	163 (64)
Donor Source
Bone marrow	99 (42)	314 (45)	0.507
Peripheral blood	112 (48)	335 (47)
Umbilical cord blood	23 (10)	53 (8)
GVHD Prophylaxis
CNI + MMF or MTX	138 (59)	377 (54)
Ex-vivo T-Cell Depletion	25 (11)	79 (11)	0.459
PTCy Based	59 (25)	213 (30)
Other	12 (5)	33 (5)
In-Vivo T-Cell Depletion
No	146 (62)	454 (65)	0.582
Yes	88 (38)	248 (35)
Conditioning Regimen
TBI based MAC	36 (15)	122 (17)	0.627
Non-TBI based MAC	94 (40)	291 (42)
RIC/NMA	104 (44)	289 (41)
HCT-CI
0–2	150 (64)	457 (65)	0.843
≥3	84 (36)	245 (35)
Median year of HCT [IQR]	2018 [2016, 2019]	2018 [2016, 2019]	0.491

AML/ALL, acute myeloid leukemia/acute lymphoid leukemia; MDS/MPN, myelodysplastic syndrome/myeloproliferative neoplasm; CNI, calcineurin inhibitor; MMF, mycophenolate mofetil; MTX, methotrexate; PTCy, post transplantation cyclophosphamide; MAC, myeloablative conditioning; TBI, total body irradiation; RIC/NMA, reduced intensity conditioning/nonmyeloablative; HCT-CI, hematopoietic cell transplantation-specific comorbidity index.

^*Analyzed in subset of 8/8 HLA matched.

GVHD and long-term outcomes

After adjusting for propensity score matching, we found no significant differences among patients with similar baseline characteristics in the first year for the cumulative incidence of GVHD that required systemic treatment (34% vs. 36%, P = 0.626, Figure 1A), maximum grades II-IV acute GVHD (29% vs. 28%, P = 0.600, Figure 1B), and chronic GVHD (16% vs. 19%, P = 0.338, Figure S1). In contrast, the cumulative incidence of maximum grades III-IV acute GVHD was higher in Black patients than White patients at the level of significance (17% vs. 12%, P = 0.050) (Figure 1C). The differences in GVHD severity at the time of GVHD onset were even more striking. Black patients were significantly more likely to require treatment for grade III-IV GVHD than White patients (39% vs. 21%, P = 0.001), mainly due to more frequent lower gastrointestinal GVHD at symptom onset (Table 2). Interestingly, fewer Black patients had extensive skin involvement at initial presentation (stage 3 to 4: 13% vs. 22%, P = 0.036). The median initiation day of systemic treatment was similar between races (34 vs. 37 days, P = 0.702) (Table 2). Similar proportions of Black and White patients received second-line treatment (27% vs. 28%, P = 0.888). A variety of therapies were used for second line treatment with ruxolitinib used most frequently in this setting and in similar proportions between Black (7/22, 37%) and White patients (19/71, 32%; P = 0.890).

Figure 1. Cumulative incidence of acute GVHD and long-term outcomes.

(A) 12-month cumulative incidence of acute GVHD requiring systemic treatment at 12 months. Black: 34% (95% CI, 28–41%), White: 36 % (95% CI 33–40%). (B) 12-month cumulative incidence of grades II-IV acute GVHD. Black: 29% (95% CI, 23–35%), White: 28% (95% CI, 25–31%). (C) 12-month cumulative incidence of grades III-IV acute GVHD. Black: 17% (95% CI, 13%-22%), White: 12% (95% CI 10–15%). (D) 12-month cumulative incidence of NRM. Black: 18% (95%CI 14–24%), White: 12% (95% CI, 9–14%). (E) 12-month survival. Black: 73% (95% CI 67–79 %), White: 79% (95% CI 76–82 %).

Table 2.

GVHD characteristics at treatment in matched cohort

	Black n = 82 (%)	White n = 254 (%)	P value
GVHD overall grade
I to II	50 (61)	202 (80)	0.001
III to IV	32 (39)	52 (21)
Minnesota risk
Standard	53 (65)	205 (81)	0.004
High	29 (35)	49 (19)
Skin stage
0	43 (52)	94 (37)	0.036
1–2	28 (34)	104 (41)
3–4	11 (13)	56 (22)
Liver stage
0–1	76 (93)	249 (98)	0.044
2–4	6 (7)	5 (2)
UGI stage
0	34 (42)	178 (70)	<0.001
1	48 (59)	76 (30)
LGI stage
0–1	54 (66)	207 (82)	0.005
2–4	28 (34)	47 (18)
Days until systemic treatment, median [IQR]	34 [25, 74]	37 [22, 60]	0.702
Initial steroid dose (mg/kg), median [IQR]	1.00 [0.80, 2.00]	0.96 [0.73, 1.60]	0.014
Second-line treatment	22 (27)	71 (28)	0.888

UGI, upper gastrointestinal; LGI, lower gastrointestinal

We observed significantly higher NRM in the first year in Black patients compared to White patients (18% vs 12%, P = 0.009) (Figure 1D), consistent with more severe acute GVHD. Univariate analyses identified several other risk factors for NRM such as older age (>55 years), cord blood as a stem cell source, and high HCT-CI score (≥3), but these characteristics were associated with higher NRM for both Black and White patients (Table S3). The 1-year OS in Black patients was 6% lower than in White patients, which trended toward statistical significance (73% vs 79%, P = 0.071) (Figure 1E). We did not observe significant differences in cause of death between races (P = 0.441) (Table S4). Among patients receiving systemic treatment for acute GVHD, cumulative incidence of NRM within 12 months of treatment was significantly higher in Black patients (26% vs 16%, P = 0.048) (Figure 2).

Figure 2. Cumulative incidence of NRM in patients with acute GVHD.

12-month cumulative incidences of NRM in patients with acute GVHD. Black: 26% (95%CI 17–36%), White: 16% (95% CI 12–21%).

We considered the possibility that SCD, which was an indication for HCT only for Black patients, might account for some of the higher incidences of severe acute GVHD and NRM. The incidence of grade III-IV GVHD (10% vs 19%, P = 0.149) and 12-month NRM (2% vs 23%, P = 0.004), however, was actually lower among Black patients transplanted for SCD compared to other diseases. We also observed that the incidence of grade III-IV GVHD (10% vs. 8%, P = 0.583) and 12-month NRM (2% vs. 7%, P = 0.554) was similar between Black patients with SCD and White patients with non-malignant disorders (n=132), consistent with previous studies. [49, 50] This suggests that the higher risk of severe GVHD and NRM among Black patients was not due to SSD as an indication for HCT, though the number of patients in this subset analysis was small.

The proportion of Black patients varied among transplant centers, and we therefore considered the possibility that experience in transplanting Black patients might account for some of the observed differences in outcomes. We included all 279 Black patients because matching to White patients was not required for this analysis and divided patients into two groups based on their proportion of Black patients using a threshold of 30% (low volume n=108; high volume n=171). The cumulative incidence of grade III-IV GVHD (11% vs. 18%, P = 0.097) and 12-month NRM (14% vs. 19%, P = 0.434) were not significantly different.

MAGIC biomarkers

We next evaluated the prognostic value of MAGIC serum biomarkers that were measured at the time of systemic treatment that were available in 79 (92%) of Black patients and in 205 (79%) of White patients. The 12-month NRM did not statistically differ between patients with and without serum samples. The distribution of patients into each AA stratus was similar between races (Table 3) and higher AA scores were associated with significantly worse 12-month NRM (Figure 3B). Interestingly, patients with AA1 GVHD experienced low NRM regardless of race (9% vs 7%, P = 0.414), but as AA score increased, the risk for NRM began to diverge between Black and White patients (AA2, 25% vs 14%, P = 0.237) such that the difference was large and statistically significant for patients with AA3 GVHD (71% vs 32%, P = 0.010) (Figure S2). The significantly worse outcomes for Black patients with AA3 GVHD compared to White patients partially explains the significantly worse NRM overall observed for Black patients. We also summarized patient characteristics and clinical courses in these 14 Black patients who exhibited AA3 at treatment (Table S5).

Table 3.

Biomarker severity at treatment

	Black n = 79 (%)	White n = 205 (%)	P value
MAP at Treatment
Ann Arbor 1	45 (57)	118 (58)	0.847
Ann Arbor 2	20 (25)	56 (27)
Ann Arbor 3	14 (18)	31 (15)

Figure 3. Cumulative incidence of NRM by Ann Arbor score at GVHD treatment.

(A) 12-month cumulative incidence of NRM. Black patients. AA1: 9% (95%CI 3–20%), AA2: 25% (95% CI 9–46%), AA3: 71% (95% CI 37–89%). Pairwise comparisons: AA1 vs AA2, P = 0.136, AA2 vs AA3, P < 0.001, AA1 vs AA3, P < 0.001. (B) White patients. AA1: 7% (95%CI 3–12%), AA2: 14% (95% CI, 7–25%), AA3: 32% (95% CI 17–49%). Pairwise comparisons: AA1 vs AA2, P = 0.165, AA2 vs AA3, P = 0.015, AA1 vs AA3, P < 0.001.

Discussion

The advances in GVHD management over the past decade encouraged a reassessment of the influence of race on outcomes related to GVHD. In this study, there were significant differences observed in recipient age, primary disease, donor source, and GVHD prophylaxis between Black and White recipients, as expected. To evaluate the effect of race, we used propensity score matching to ensure a balanced comparison of key characteristics between the two populations to the greatest extent possible. The analyses using matched populations indicated that Black patients had a significantly higher NRM rate that was associated with a greater prevalence of severe acute GVHD. We also found that, although the distribution of AA scores was comparable across racial groups, Black patients with high risk GVHD by biomarkers (AA3) exhibited a significantly worse NRM than White patients.

There are several possible explanations for a higher incidence of severe GVHD and worse NRM in Black patients. Mild erythematous skin rashes may be more difficult to detect in patients with darker skin complexions and thus, initiation of treatment may be delayed. If true, delays in diagnosis may have contributed to worse outcomes, however, the similar timing of initiation of treatment between races argues against this scenario. Another possible explanation for worse outcomes is disparities in outpatient management. Several studies reported lower rates of medication adherence and less access to clinical care for Black patients compared to White patients. [51–54] The underlying reasons for such disparities are likely multifactorial and influenced by social determinants of health such as income, education, and insurance that also associate with race and ethnicity. [55, 56] It is plausible that lower medication adherence and/or less access to clinical care may have contributed to the higher NRM observed in Black patients. Such etiologies must remain speculative because our database lacks detailed information regarding key variables, such as socioeconomic status and potential surrogate measures for medication adherence such as serum concentration of immunosuppressive agents. The significantly worse NRM experienced by patients who were Black and had AA3 GVHD was unexpected. Further confirmation of this finding with additional patients is needed given the relatively small numbers of patients with AA3 GVHD overall. Finally, we did not find evidence that greater experience in transplants for Black patients translated into better outcomes.

Our study has additional limitations. First, the number of Black patients included in our study is relatively small which precluded conducting potentially informative multivariate analyses. Second, the determination of race was solely based on self-report. Third, patients who were also Hispanic were included in the Black cohort to maximize sample size but not the White cohort. Thus, factors related to ethnicity and not race may have contributed to the findings. Fourth, the effect of intensive GVHD prophylaxis such as PTCy and T-cell depletion that might mitigate poor risk factors could not be determined due to limited representation in the study population.

In conclusion, our study demonstrated that Black patients are at greater risk of developing severe GVHD and are more likely to die from GVHD than well matched White counterparts. AA scores at treatment initiation stratified patients for risk of NRM in both races. This study was not able to provide an explanation for differential risks by race, but the findings are perhaps not surprising given the observation of worse health outcomes for Black patients in other disease settings. These findings, if confirmed in independent data sets, suggest that careful monitoring for GVHD is particularly important for Black patients.

Highlights.

• Black patients have a higher risk of grade III-IV acute GVHD and non-relapse mortality compared to White patients.

• MAGIC biomarker scores at the time of treatment effectively stratified the risk of NRM regardless of race.

• Black patients with high biomarker scores experienced significantly worse NRM than White patients.