The role of serum uric acid in the prediction of graft‐versus‐host disease in allogeneic hematopoietic stem cell transplantation

Abstract Background Uric acid (UA) level is of the valuable signs of inflammation. However, the role of UA in the outcomes of hematopoietic stem cell transplantation (HSCT) such as GVHD and patients’ overall survival is still a matter of debate. In this study, we aimed to evaluate the relationship between UA levels and GVHD incidence and overall survival in allogeneic HSCT patients. Methods A total of 201 patients who were admitted for allogeneic transplantation at Taleghani hospital, Tehran, Iran, were considered for retrospective analysis. Serum UA levels from 1 week before transplantation until 2 weeks after transplantation were used to determine thresholds and find out the association of serum UA levels with GVHD and overall survival. Results We showed that the determined thresholds using receiver operating characteristic curves have poor predictive value for GVHD and overall survival. The patients with serum UA higher than 3.4 mg/dL had 37% lower odds of GVHD incidence and 35% lower hazard of death than patients with UA lower than 3.4 mg/dL. Conclusion Our results indicated that serum UA levels lower than 3.4 mg/dL could significantly increase the incidence of GVHD and hazard of death. The antioxidant functions of UA could explain the lower incidence of GVHD in hyperuricemic patients. However, the inconsistencies of the previous studies require further investigation to elucidate the role of UA in the prediction of GVHD.

recipient antigens and cause severe morbidity and mortality. There are two types of GVHD, including acute (aGVHD) and chronic (cGVHD). 4 The median time of aGVHD and cGVHD is typically 21-25 days and 4.5 months, respectively, with the incidence of 30%-50% and 30%-70%, in the order given. 5 There are many risk factors affecting the development of GVHD in HSCT patients, including type of the graft, donor-recipient relationship, HLA and sex matching, recipient age, intensity of conditioning and prophylaxis regimens. 6 The previous studies identified that HSCT process alters serum uric acid (UA) levels in allo-HSCT. 7 During HSCT process, conditioning regimen leads to promotion of endogenous danger signaling such as heat shock proteins and UA. 8 Moreover, radio/chemotherapy induces UA crystallization and deposition by increasing serum UA which can result in neutrophil migration into tissues and intensifying the inflammation. 9 UA from damaged cells releases into extracellular fluid and induces antigen-presenting cells (APCs) for activating the immune response. 10 In fact, UA is a danger-associated molecular pattern (DAMP) capable of inducing T cells to release IL-1B through the activation of the NOD-like receptor protein (NLRP) 3. 11 Regarding the previous studies indicating the UA-mediated activation of host APC and T-cell response, the elevated UA levels may play a critical role in the immune activation and inflammation. 10,11 There are several studies about the relationship between UA levels and the incidence of GVHD and overall survival in allo-HSCT patients with controversial findings. [12][13][14][15] Formerly, UAdecreasing drugs such as uricase or allopurinol in allo-HSCT patients have been shown to inhibit cytotoxic T lymphocytes activity and reduce the development of aGVHD. 12 On the other hand, a recent study found a negative association between UA levels and GVHD showing the need for more clarification. 14 Therefore, in this study, we aimed to evaluate the relationship between UA levels as a sensitive parameter and GVHD incidence as well as overall survival in patients who underwent allo-HSCT.

| Patient's characteristics
A total of 201 patients with hematological disorders who were admitted for allogeneic transplantation at Taleghani hospital, Tehran, Iran, between 2008 and 2018 were considered for retrospective data analysis. All patients were categorized based on their diagnosed disease, including acute myeloid leukemia (AML), acute lymphoid leukemia (ALL), aplastic and Fanconi anemia (AA, AF), Hodgkin disease (HD), non-Hodgkin lymphoma (NHL), and others.
These patients gave the informed consent that their data were used for scientific analysis, and the study was approved by the Ethics

| HSCT process and GVHD evaluation
The peripheral blood stem cells (PBSC) were mobilized after 4 days The standard clinical signs such as rash, diarrhea, and liver function abnormalities associated with biopsy and histopathological criteria in involved organ were main manifestations for diagnosis of GVHD according to the GVHD criteria provided by National Institute of Health. 16

| Laboratory test
Serum UA levels were measured from day −7 until day +14. Peripheral blood samples were collected, and UA levels in serum samples were evaluated using an ADVIA 1800 clinical chemistry analyzer (Toshiba).

| Statistical analysis
Data were expressed as means ± standard deviation (SD) or frequency (%). The main purpose of this study was to identify the effects of serum UA on the incidence of GVHD and the overall survival with the adjusted effect of other risk factors. The logistic regression was employed when the outcome variable was GVHD. The odds ratios, 75 and 90% confidence interval (CI), were calculated accordingly. When the outcome was the overall survival, we conducted a Cox regression analysis. The hazard ratios, 75 and the 90% CI were evaluated as well. The median of serum UA in each disease category as well as the correlation of UA level with the overall survival was determined.
The threshold for serum UA level in three different periods of time including 1 week before transplant, 1 and 2 weeks after transplant was determined using both the receiver operating characteristic (ROC) curves and median. The sensitivity, specificity, and the area under curve (AUCs) were also assessed in ROC curves analysis.
Finally, due to the importance of the week before transplantation, the median of UA level in the time period between day −7 to the transplantation day was selected as the cutoff and utilized in the univariable and multivariable analysis. The survival curve was drawn based on the median of serum UA. The P-value <.25 in univariable analysis and P-value <.1 in multivariable was considered as significant. All analyses were conducted through the SPSS version 19.0 (SPSS Inc).

| Descriptive statistics and UA cutoff determination
The descriptive statistics is shown in Table 1

| Predictive values of the serum uric acid level before and after HSCT for GVHD
Beside the median of serum UA levels during the week before transplantation as the main cutoff point for UA in this study, the ROC curve was also drawn to determine a cutoff and evaluate the predictive values of serum UA level before and after transplant for GVHD. As shown in Table 2, the AUC, sensitivity, and specificity for the meantime of 1 week before transplant for GVHD were 59% (CI: 0.50-0.67), 76%, and 39%, respectively with a cutoff value of 2.98 mg/dL. Therefore, it can be suggested that serum UA level of 7 days before transplant can be used as a relatively poor predictive of GVHD since the AUC is close to 60%. The cutoff value of UA level in 1 week after transplantation for GVHD was 3.91 mg/dL with the AUC of 47% (CI: 0.37-0.56), sensitivity of 29%, and specificity of 82%. So, it is suggested that the serum UA level for the meantime of the 7 days after transplant cannot be a predictive of GVHD, since the AUC is below 50%. For the meantime of 2 weeks after transplant for GVHD, the AUC, sensitivity, and specificity were 54% (CI: 0.44-0.63), 45%, and 67%, one by one with a cutoff value of 4.23 mg/ dL. Thus, we can say that the serum UA level for the meantime of 2 weeks after transplant may be a poor predictive of GVHD, since the AUC is close to 60.

| Predictive values of the serum uric acid level before and after HSCT for overall survival
In order to determine the predictive values of serum UA level before and after transplant for overall survival, we performed ROC analyses. The results are indicated in

| Association of high serum UA and other risk factors with GVHD
We conducted a univariable and multivariable logistic regression to identify the pre-transplant risk factors for GVHD. The results are illustrated in Table 4. As we can see in the univariable analysis, the cutoff value of serum UA level at 3.4 mg/dL has a significant effect on the occurrence of GVHD. The odds of GVHD incidence in patients with UA higher than 3.4 mg/L was 37% lower than patients with the UA below the cutoff ([CI: 0.44-0.91]; P-value = .15).
The blood group also significantly affected the GVHD outcome.

| Association of high serum UA and other risk factors with overall survival
We applied a Cox proportional hazards model in order to determine the significant risk factors on the survival of the patients. As it is illustrated in Table 5 Table 5. Table 6 indicates the median of serum UA and hyperuricemia in each disease type, separately. The correlation of serum UA level with OS was also conducted. As shown, the serum UA level only in AML patients had a significant correlation with the overall survival in which the increased UA level resulted in poor survival of the patients (Pvalue: .03).

| D ISCUSS I ON
The radio/chemotherapy, along with the conditioning regimen before the HSCT, cause massive cell death in all organs of the patients. The accessibility and ease of measurement together with its role in the inflammatory conditions, such as HSCT, make the serum UA as an attractive biomarker in researches to predict the GVHD. 14 However, the role of UA in inflammation is still controversial. In allogeneic responses such as kidney transplantation, high levels of UA are associated with both cellular and humeral immune activation leading to graft rejection or poor graft survival. 21 In HSCT, also, UA is introduced as a sensitive biomarker to distinguish the graft failure from poor graft function, but its predictive role in GVHD is a matter of debate. 15 Preclinical and phase I clinical studies have reported that UA-decreasing drugs, including uricase and allopurinol, can diminish the GVHD incidence and alleviate its severity. 22,23 These studies concluded that high UA level is a risk factor for GVHD and should be decreased before HSCT to control the GVHD. Albeit, these reports were about the association of UA-decreasing drugs with GVHD. 22,23 Moreover, several studies implied the association of low UA levels and inflammatory and degenerative diseases. [24][25][26] The purpose of this study was to identify the effects of serum UA on the incidence of GVHD and the overall survival with the ad-

TA B L E 4
Univariable and multivariable logistic regression models for graftversus-host disease than 60% which means these cutoffs have low capacity to predict the outcomes. Moreover, the sensitivity and specificity of the cutoffs were low. These results show that the thresholds achieved by ROC curve are not a reliable cutoff to predict the HSCT outcome.
The other method that is used to determine the cutoff for serum UA in our patients was median. The median of serum UA levels in the period of 1 week before the transplantation was 3.4 mg/dL, which was selected as a cutoff for the statistical analysis. Generally, the cutoff point for hyperuricemia is 6.7 mg/dL at which the crystallization of UA begins. 27 Our determined cutoff was median and was not based on the crystallization point. The reason for choosing 1 week before HSCT to define a cutoff is that the UA levels in this period are critical in determining the fate of HSCT. Previous reports pointed out that the pre-transplant UA decrease can affect the GVHD. 14,18,19 Our univariable analysis showed that the cutoff value of serum UA level at 3.4 mg/dL has a significant effect on the occurrence of GVHD in a way that GVHD incidence in patients with UA higher than 3.4 mg/L was 37% lower than patients with the UA below the cutoff.
However, the multivariable analysis showed no significant association between serum UA levels and GVHD. This finding contradicts with some previous reports which claimed that the higher serum UA is associated with GVHD incidence and its severity, 22

TA B L E 5
Univariable and multivariable cox regression models for overall survival major natural antioxidant in the periphery, and reduced antioxidative capacity is associated with GVHD. [28][29][30][31] Except for the serum UA levels and blood group, other factors such as diagnosed disease, CMV, and conditioning regimen did not have any significant effect on the incidence of GVHD. The blood group A and B also significantly increased the GVHD incidence, compared to blood group O, which might be due to the ABO antibodies and blood group incompatibility in some patients.
In accordance with the results of GVHD, it is indicated that the lower serum UA levels are associated with a higher risk of death Conclusively, we demonstrated that the lower serum UA levels are associated with GVHD incidence and inferior overall survival.
Our result is in line with some previous studies and contradicts some others which suggest that there might be an optimum range for UA levels which control the HSCT outcome and UA levels either lower or higher than this range could be detrimental for HSCT outcome. More comprehensive studies can evaluate the validity of this hypothesis.

ACK N OWLED G M ENT
The