Prediction model for cytomegalovirus infection following hematopoietic stem cell transplantation in patients with β-thalassemia major

Abstract

Background:

Cytomegalovirus infection is a common complication following hematopoietic stem cell transplantation that significantly influences clinical outcomes.

Objectives:

To develop and validate a predictive model for cytomegalovirus infection risk in patients with β-thalassemia major undergoing hematopoietic stem cell transplantation.

Design:

Retrospective cohort study.

Methods:

Clinical data from 291 β-thalassemia major patients undergoing hematopoietic stem cell transplantation were retrospectively analyzed. Independent risk factors identified via univariate and multivariate logistic regression analyses formed the basis of a predictive nomogram. The model’s performance was evaluated by the concordance index (C-index), receiver operating characteristic curves, calibration plots, and decision curve analysis. Internal validation was performed using bootstrap resampling, and external validation was conducted with an independent cohort of 84 patients from another center.

Results:

Three independent predictors of cytomegalovirus infection were identified: serum albumin levels, donor type, and grade III–IV acute graft-versus-host disease. A nomogram incorporating these predictors was established, demonstrating good discriminative ability (C-index: 0.745; 95% CI: 0.684–0.807). Internal and external validations yielded C-indices of 0.746 and 0.649, respectively. Receiver operating characteristic analysis showed an area under the curve of 0.745 in the training cohort and 0.649 in the validation cohort.

Conclusion:

We developed and validated a reliable predictive model for assessing cytomegalovirus infection risk after hematopoietic stem cell transplantation in β-thalassemia major patients. This scoring system offers clinicians a practical tool for early risk stratification and intervention.

Keywords

CMV infection HSCT β-thalassemia major prediction model Nomogram

Introduction

β-Thalassemia major (β-TM) is an autosomal recessive genetic disorder characterized by reduced or absent synthesis of β-globin chains. Currently, hematopoietic stem cell transplantation (HSCT) remains the only definitive curative therapy for β-TM.¹ However, HSCT significantly increases the risk of viral infections, with cytomegalovirus (CMV) infection being the most prevalent complication, affecting up to 70% of transplant recipients.^2,3 CMV infection can cause severe tissue and organ disease, impair immune function, elevate the risk of Epstein-Barr virus (EBV) reactivation leading to lymphoproliferative disorders, and adversely impact graft function and transplantation outcomes.⁴

Multiple risk factors contribute to the occurrence of CMV infection after HSCT, including older recipient age, acute graft-versus-host disease (aGVHD), T-cell depletion protocols, and donor-recipient mismatch or unrelated donors.⁵ Despite these recognized factors, to our knowledge, no predictive model has yet been established specifically to identify β-TM patients at high risk for CMV infection after HSCT. Such a predictive model is critically needed to enable clinicians to stratify risk early, initiate timely prophylactic or preemptive interventions, and thereby potentially improve patient outcomes.

In this study, we aimed to develop and validate a practical, reliable scoring system to predict CMV infection risk after HSCT specifically in patients with β-TM. By integrating clinical data from our center and external validation cohorts, we identified independent risk factors—including serum albumin levels, donor type, and severe aGVHD (grade III–IV)—and assigned corresponding risk scores. This novel scoring system represents the first predictive tool tailored to this specific patient population, providing clinicians with a straightforward method for timely risk assessment and targeted management to reduce CMV infection rates and improve transplantation efficacy.

Patients and methods

Study patients

Between July 2020 and November 2023, a total of 291 patients with β-thalassemia major underwent allo-HSCT at the First Affiliated Hospital of Guangxi Medical University. All patients were subjected to a standardized conditioning regimen, which included fludarabine, anti-thymocyte globulin, busulfan, and cyclophosphamide. Furthermore, the prophylactic regimens against GVHD consisted of cyclosporine or tacrolimus, mycophenolate mofetil, and methotrexate.⁶ The patients were monitored regularly, and the follow-up endpoint was set for May 31, 2024.⁵ Following HSCT, 84 of these patients developed CMV infection. The derivation cohort consisted of 291 patients and was used to develop the prediction model. In order to assess the model’s applicability, it was tested using both internal and external validation data. The sample size was estimated to be 5–10 times the number of variables in the model, while the total sample size was determined based on the endpoint event incidence to ensure consistency with the study scale. Approximately 30% of the training set sample size was chosen to define the external validation sample size.⁷ Following the estimation, 84 patients who met the inclusion criteria were selected to constitute the external validation set. The externally validated cohort was recruited from Liuzhou Workers’ Hospital. Out of 84 β-TM patients who underwent allo-HSCT between September 2019 and August 2022, 17 developed CMV infection following transplantation. The conditioning regimens for these patients included fludarabine, busulfan, and cyclophosphamide. Furthermore, anti-thymocyte globulin, cyclosporine, mycophenolate mofetil, and methotrexate were used as part of the treatment protocol. All patients underwent hydroxyurea treatment before transplantation.

Inclusion criteria: (1) Genetic testing confirmed a diagnosis of severe β-thalassemia. (2) Negative CMV-DNA test results prior to transplantation. (3) Negative serological findings for human immunodeficiency virus (HIV), hepatitis C virus (HCV), or hepatitis B virus (HBV). Exclusion criteria: (1) Positive CMV-DNA test results prior to transplantation. (2) The patient underwent a second transplantation. (3) Diagnosis of α-thalassemia. (4) Incomplete clinical data. The study design is depicted in Figure 1.

Figure 1.

Flow diagram illustrating the development of the CMV infection model.

Ethics approval and consent to participate

The study was conducted in accordance with the principles of the Declaration of Helsinki and obtained ethical approval from the Institutional Review Board of The First Affiliated Hospital of Guangxi Medical University (approval number: 2025-E0022). Written informed consent was obtained from all participants prior to their involvement in the study. Adult participants (aged 18 and above) provided consent independently, whereas for minors (under 18), consent was granted by their legally authorized representative on their behalf.

Data collection and definitions

This study utilized a retrospective cohort design to investigate the outcomes of interest. A total of 31 variables were systematically collected, encompassing general patient information, transplantation-related drug regimens, graft characteristics and engraftment status, post-transplantation complications, and pre-transplantation blood biochemistry test results. All data were retrieved from the hospital’s electronic medical record system. A summary of these variables is presented in Table 1. Blood biochemical test results were extracted from reports generated within 24 h following patient admission.

Table 1.

Comparison of clinical data between non-CMV infection group and CMV infection group.

Variables	Non-CMV infection group (n = 207)	CMV infection group (n = 84)	p-value
Gender, n (%)			0.859
Female	79 (38)	33 (39)
Male	128 (62)	51 (61)
Race, n (%)			0.908
Han	137 (66)	55 (65)
Non-Han	70 (34)	29 (35)
Age (years), median (IQR)	8.00 (5.00, 11.00)	8.50 (6.00, 11.00)	0.620
CSA, n (%)			0.001*
No	127 (61)	68 (81)
Yes	80 (39)	16 (19)
TAC, n (%)			0.001*
No	80 (39)	16 (19)
Yes	127 (61)	68 (81)
AntiCD25, n (%)			0.391
No	132 (64)	58 (69)
Yes	75 (36)	26 (31)
MMF, n (%)			>0.999
No	9 (4.3)	3 (3.6)
Yes	198 (96)	81 (96)
MTX, n (%)			0.328
No	4 (1.9)	0 (0)
Yes	203 (98)	84 (100)
EBV infection, n (%)			0.437
No	192 (93)	80 (95)
Yes	15 (7.2)	4 (4.8)
Septicemia, n (%)			0.354
No	168 (81)	72 (86)
Yes	39 (19)	12 (14)
Splenectomize, n (%)			0.743
No	182 (88)	75 (89)
Yes	25 (12)	9 (11)
HLA match, n (%)	66(31.9)	36(42.9)	0.077
Donor type, n (%)			<0.001*
Matched sibling	79 (38)	13 (15)
Matched unrelated	60 (29)	34 (40)
Haploidentical related	68 (33)	37 (44)
ABO mismatched, n (%)			0.338
No	103 (50)	47 (56)
Yes	104 (50)	37 (44)
Graft type
Bone marrow, n (%)			0.058
No	60 (29)	34 (40)
Yes	147 (71)	50 (60)
Peripheral blood, n (%)			0.927
No	19 (9.2)	8 (9.5)
Yes	188 (91)	76 (90)
Cord blood, n (%)			0.238
No	192 (93)	81 (96)
Yes	15 (7.2)	3 (3.6)
Female for male, n (%)			0.251
No	160 (77)	70 (83)
Yes	47 (23)	14 (17)
Grade I–II aGVHD, n (%)			0.430
No	164 (79)	63 (75)
Yes	43 (21)	21 (25)
Grade III–IV aGVHD, n (%)			<0.001*
No	197 (95)	62 (74)
Yes	10 (4.8)	22 (26)
Ferritin >2500 ng/mL, n (%)			0.815
No	62 (30)	24 (29)
Yes	145 (70)	60 (71)
Day of neutropenia, median (IQR)	12.00 (11.00, 14.00)	12.00 (11.00, 14.00)	0.402
Neutrophil engraftment day, median (IQR)	12.00 (11.00, 14.00)	11.50 (11.00, 13.00)	0.085
WBC (10⁹/L), median (IQR)	6.20 (4.93, 7.66)	6.36 (5.15, 8.12)	0.233
HB (g/L), median (IQR)	103.00 (92.30, 115.05)	102.05 (91.00, 115.43)	0.830
PLT (10⁹/L), median (IQR)	271.20 (215.65, 375.75)	268.35 (193.15, 395.95)	0.514
NEU (10⁹/L), median (IQR)	2.37 (1.76, 3.14)	2.67 (1.86, 3.64)	0.133
ALB (g/L), median (IQR)	42.70 (40.20, 45.20)	40.55 (38.60, 42.75)	<0.001*
Total MNC (10⁸/kg), median (IQR)	11.20 (8.89, 14.14)	11.38 (8.19, 13.87)	0.618
CD34 (10⁶/kg), median (IQR)	8.33 (6.87, 11.76)	8.24 (6.81, 10.83)	0.837
Platelets engraftment day, median (IQR)	14.00 (12.00, 17.00)	13.00 (12.00, 17.25)	0.516

CSA: cyclosporin A; TAC: tacrolimus; MMF: mycophenolate mofetil; MTX: methotrexate; EBV: Epstein-Barr virus; aGVHD: acute graft-versus-host disease; WBC: white blood cell; HB: hemoglobin; PLT: platelet; NEU: absolute neutrophil; ALB: serum albumin; MNC: total mononuclear cells; CMV: cytomegalovirus; IQR: interquartile range; HLA: human leukocyte antigen.

p < 0.05.

Peripheral blood samples were regularly collected from all patients following transplantation, and plasma CMV-DNA concentrations were measured by means of quantitative polymerase chain reaction (Q-PCR). CMV infection was defined as the detection of >1000 copies/mL of CMV-DNA in viral nucleic acids in any body fluid or tissue specimen on at least two consecutive occasions. The period from the transplantation date to the first positive Q-PCR result for CMV (>1000 copies/mL) was defined as the time of CMV infection.⁵ EBV infection was defined as the detection of EBV-DNA levels greater than 500 copies/mL in peripheral blood on at least one occasion.⁸ Neutrophil engraftment was defined as the first of three consecutive days during which the absolute neutrophil count reached or exceeded 0.5 × 10⁹/L without granulocyte colony-stimulating factor support. Platelet engraftment was defined as the first of seven consecutive days on which the platelet count reached or exceeded 20 × 10⁹/L in the absence of platelet transfusion support.⁴ aGVHD is characterized by the manifestation of allogeneic inflammatory responses predominantly in three key target organs: the skin, liver, and gastrointestinal tract. At the time of diagnosis, no evidence of chronic graft-versus-host disease (cGVHD) was observed, and histopathological findings were negative. The severity of aGVHD was classified in accordance with established criteria from the literature.⁹ Neutropenia duration was defined as the time interval during which the peripheral blood neutrophil count remained below 0.5 × 10⁹/L. Septicemia is defined by infection signs, positive blood cultures, and organ dysfunction.¹⁰ Human leukocyte antigen matching was defined as complete concordance at all antigenic loci.

Construction and validation of predictive models

A total of 31 variables were included in the univariate analysis. Variables with a p-value <0.05 in the univariate analysis were then entered into the logistic regression model for multivariate analysis. Variables with a p-value <0.05, including serum albumin level, grade III–IV aGVHD, and donor type, were identified as risk factors for CMV infection after thalassemia transplantation. Additionally, a nomogram-based prediction model was constructed.

The nomogram was constructed by assigning a score to each parameter, and the scores were summed to generate a total score. Each total score was then associated with the probability of the outcome event occurring. The discriminative ability and prediction accuracy of the nomogram were assessed using the concordance index (C-index). The calibration curve evaluated the agreement between actual and predicted CMV infection risks from the nomogram. The predictive ability of the nomogram was assessed using the area under the curve (AUC). The clinical net benefit was evaluated via the decision curve analysis curve. Finally, we employed the bootstrap internal validation method (1000 bootstrap resamples) to calculate the C-index for validating the constructed nomogram and simultaneously conducted external validation.

Statistical analysis

All statistical analyses were conducted using R (v4.3.3)(R Core Team, Vienna, Austria) on Windows. Continuous variables are presented as median (interquartile range (IQR)) where appropriate, and categorical variables as frequencies (%). The Kruskal-Wallis test was used to compare continuous variables among groups, while Pearson’s chi-squared or Fisher’s exact test was applied for categorical data. All statistical tests were two-sided, and the significance level was set at p < 0.05.

Results

Patient characteristics

A total of 291 patients were enrolled in the training set. The median age of the patients was 8 years (ranging from 2 to 19 years), with 179 males and 112 females. Among these patients, 84 developed CMV infection, while 207 remained free of CMV infection. The cumulative incidence rate of CMV infection was 40.6%. The median time to the onset of CMV infection was 31 days (ranging from 11 to 151 days) after HSCT. A detailed summary of the demographic and clinical characteristics of the two groups is provided in Table 1. The validation set consisted of 84 patients with a median age of 8 years (IQR: 6–11 years), including 43 males and 41 females. Of these, 17 patients (20.2%) developed CMV infection at 29.24 ± 12.65 days post transplantation.

Establishment of CMV infection prediction model

This study identified and analyzed 31 variables potentially influencing CMV infection using univariate analysis. Key results in Table 1 show five significant factors (p < 0.05): cyclosporin A or tacrolimus use, donor type, grade III–IV aGVHD, and albumin levels. As illustrated in Figure 2, the aforementioned five variables were included in the multivariate analysis. Logistic regression analysis demonstrated that albumin levels, donor type, and grade III–IV aGVHD were independent risk factors for CMV infection. In summary, a prediction model for the risk of CMV infection following HSCT for β-thalassemia major was developed using albumin levels, donor type, and grade III-IV aGVHD as key predictors (Figure 3). In Figure 3, the red dot on the horizontal axis indicates the patient’s score for the corresponding variable relative to each parameter, and the vertical line aids in visualizing these scores.

Figure 2.

Forest plot depicting the factors influencing the progression of CMV infection after HSCT in thalassemia patients.

Figure 3.

Nomogram for predicting CMV infection following HSCT in patients with β-thalassemia major.

Evaluation and validation of CMV infection prediction model

The nomogram model’s predictive performance was evaluated using the C-index (0.745; 95% CI: 0.684–0.807). Receiver operating characteristic (ROC) analysis yielded an AUC of 0.745 (Figure 4(a)), and the validation set had an AUC of 0.649 (Figure 4(b)), indicating that the model exhibited satisfactory predictive accuracy. The calibration plot was constructed using the Hosmer-Lemeshow goodness-of-fit test to evaluate the model’s accuracy. The nomogram calibration curve for CMV infection risk closely aligns with the ideal curve, indicating strong consistency between the training and validation sets (Figure 5(a) and (b)). The closer the calibration curve is to the ideal curve, the stronger the model’s predictive ability. Our results demonstrate the model’s excellent predictive performance. The decision curve was constructed to evaluate the clinical applicability of the model, and the results indicate a high patient benefit rate (Figure 6). Finally, the predictive accuracy and stability of the nomogram model were validated using the bootstrap internal validation method. The C-index was 0.746 (based on 1000 bootstrap resamples), closely aligning with the training set’s C-index of 0.745. Furthermore, external validation resulted in a C-index of 0.649.

Figure 4.

ROC curves for the nomogram model (a) and external validation (b).

Figure 5.

Calibration curves for the nomogram model (a) and external validation (b).

Figure 6.

Decision curve analysis for the nomogram model.

Discussion

In this study, we developed a predictive model for CMV infection in patients with β-TM following HSCT, under standardized conditions to ensure consistency in underlying disease and treatment protocols. Based on univariate and multivariate analyses, three independent predictors—serum albumin (ALB), donor type, and grade III–IV aGVHD—were identified and used to construct a nomogram. To our knowledge, this is the first predictive scoring model specifically developed for assessing CMV infection risk in β-TM patients after HSCT, demonstrating good clinical performance in distinguishing high-risk individuals.

Among these predictors, lower pretransplant albumin levels were associated with a higher risk of CMV infection. While prior studies have not directly evaluated this relationship in HSCT recipients, evidence from other settings supports this association. For instance, hypoalbuminemia has been linked to increased CMV risk in patients with acute ulcerative colitis and kidney transplant recipients.^11,12 Albumin is a key biomarker of inflammation and nutritional status, with immunomodulatory and antioxidant properties. Its depletion may reflect systemic immune impairment, contributing to viral susceptibility.^13,14

Serum albumin has also been explored as a predictor of severe aGVHD in HSCT settings.^15
–17 In our analysis, only grade III–IV aGVHD, not grade I–II, was significantly associated with CMV infection. This aligns with previous findings indicating that severe aGVHD impairs T-cell recovery,^18
–20 compromises bone marrow niche function, and predisposes patients to opportunistic infections.^21,22 Furthermore, the intensified immunosuppressive therapy required for severe aGVHD may further hinder immune reconstitution, increasing infection risk.

Donor type was another significant predictor. CMV infection rates were notably lower in matched sibling donor recipients (15%) compared with those receiving grafts from matched unrelated (40%) or haploidentical donors (44%). These findings are consistent with previous reports indicating that identical donor-recipient pairings support better immune recovery and lower infection rates.^23,24 In contrast, mismatched or unrelated donors often necessitate more aggressive conditioning and immunosuppression (e.g., ATG, tacrolimus), leading to prolonged immune dysfunction and heightened CMV risk. Although CMV seropositivity in both donors and recipients is often considered a risk factor, its predictive value remains controversial. Some studies report a significant association,²⁵ while others do not.^26,27 Given the high seroprevalence of CMV in the Chinese population (>97%),²⁸ we excluded serological status from our model, as it may offer limited discriminatory power in this context.

The predictive performance of our model was robust, with a C-index of 0.745 in the training cohort, 0.746 after bootstrap validation, and 0.649 in the external validation cohort.²⁹ The model was developed using a combination of univariate and multivariate analytical techniques. Incorporating the predictors into a nomogram enabled intuitive scoring and facilitated individualized risk estimation.³⁰ Such models have been successfully applied to other HSCT-related complications, such as acute kidney injury,³¹ mortality in non‑ischemic dilated cardiomyopathy,³² and transfusion risk in spinal tuberculosis surgery.³³ Evaluation via ROC, calibration, and decision curves confirmed the model’s predictive value and potential clinical utility.

Despite its strengths, this study has limitations. First, the model is disease-specific and may not generalize to other post-HSCT populations. Second, the retrospective design introduces potential selection bias, though efforts were made to standardize inclusion criteria. Prospective validation is warranted. Third, some potentially relevant factors—such as immune reconstitution markers, corticosteroid use, and the distinction between CMV reactivation and actual infection—were not included due to data limitations. Future studies with broader datasets may enhance model performance and generalizability.

Conclusion

We developed and validated a nomogram-based model to predict the risk of CMV infection in patients with β-thalassemia major undergoing HSCT. The model incorporates readily available clinical parameters—serum albumin, donor type, and grade III–IV aGVHD—and demonstrates good discrimination and clinical applicability. This simple, practical tool can aid clinicians in early risk stratification, guide targeted preventive strategies, and improve transplant outcomes.

Footnotes

Acknowledgements

This study was funded by Angel Program “Caring for Life” Thalassemia Rescue Project from Red Cross Society of China Guangxi Branch. We also appreciate Prof. Peng Peng’s team for the MRI measurement of iron overload in patients with thalassemia. Additionally, we are grateful to the patients, their families, and the research staff for the ability to complete this study.

ORCID iDs

Lin Pan

Yanni Xie

Hongwen Xiao

Chan Li

Yongrong Lai

Ethical considerations

Consent to participate

Written informed consent was obtained from all participants prior to their involvement in the study. Adult participants (aged 18 and above) provided consent independently, whereas for minors (under 18), consent was granted by their legally authorized representative on their behalf.

Consent for publication

All patients and their guardians provided consent for the publication of this research.

Author contributions

Lin Pan: Conceptualization; Data curation; Formal analysis; Investigation; Methodology; Software; Validation; Writing – original draft; Writing – review & editing.

Zhenbin Wei: Data curation; Formal analysis; Investigation; Methodology; Validation; Writing – original draft.

Yanni Xie: Formal analysis; Investigation; Methodology; Software; Writing – review & editing.

Zhaoping Gan: Formal analysis; Methodology; Software; Writing – review & editing.

Hongwen Xiao: Formal analysis; Methodology.

Lianjin Liu: Data curation; Formal analysis.

Lingling Shi: Data curation; Formal analysis.

Zhongming Zhang: Data curation; Formal analysis.

Meiqing Wu: Data curation; Formal analysis.

Yinghua Chen: Data curation; Formal analysis.

Yanye Liu: Data curation; Formal analysis.

Xuemei Zhou: Data curation; Formal analysis.

Chan Li: Data curation; Formal analysis; Investigation.

Chunjie Qin: Data curation; Formal analysis; Investigation.

Yongrong Lai: Conceptualization; Investigation; Methodology; Project administration; Writing – original draft; Writing – review & editing.

Rongrong Liu: Conceptualization; Investigation; Methodology; Project administration; Writing – original draft; Writing – review & editing.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The research was supported by Young Leader Talent Training Program of Guangxi Medical University (No. 202302), Advanced Innovation Teams and Xinghu Scholars Program of Guangxi Medical University, Joint Project on Regional High-Incidence Diseases Research of Guangxi Natural Science Foundation (Grant No. 2023GXNSFAA026293), NHC Key Laboratory of Thalassemia Medicine, and Guangxi Key Laboratory of Thalassemia Research.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data availability statement

Data sharing is not applicable to this article, as no datasets were generated or analyzed during the course of this study.

Trial registration

Not applicable.

References

Liang

Pan

Xie

, et al. Health-related quality of life in pediatric patients with β-thalassemia major after hematopoietic stem cell transplantation. Bone Marrow Transplant 2022; 57: 1108–1115.

Ariza-Heredia

Nesher

Chemaly

RF.

Cytomegalovirus diseases after hematopoietic stem cell transplantation: a mini-review. Cancer Lett 2014; 342: 1–8.

Cho

S-Y

Lee

D-G

Kim

H-J.

Cytomegalovirus infections after hematopoietic stem cell transplantation: current status and future immunotherapy. Int J Mol Sci 2019; 20: 2666.

Sun

Wang

Zhang

, et al. Virus reactivation and low dose of CD34+ cell, rather than haploidentical transplantation, were associated with secondary poor graft function within the first 100 days after allogeneic stem cell transplantation. Ann Hematol 2019; 98: 1877–1883.

Hakki

Aitken

Danziger-Isakov

, et al. American Society for Transplantation and Cellular Therapy Series: #3-Prevention of Cytomegalovirus Infection and Disease After Hematopoietic Cell Transplantation. Transplant Cell Ther 2021; 27: 707–719.

Xiao

Yang

Huang

, et al. Using T-lymphocyte subsets at engraftment to predict the risk of acute graft-versus-host disease in patients with thalassemia major: development of a new predictive nomogram. Ther Adv Hematol 2024; 15: 20406207241294054.

Zhang

Ting

, et al. Development and validation of a 16-gene T-cell-related prognostic model in non-small cell lung cancer. Front Immunol 2025; 16: 1566597.

Cao

X-H

Fan

Z-Y

Chang

Y-J

, et al. Prediction model for EBV infection following HLA haploidentical matched hematopoietic stem cell transplantation. J Transl Med 2024; 22: 244.

Schoemans

Lee

Ferrara

, et al. EBMT-NIH-CIBMTR Task Force position statement on standardized terminology & guidance for graft-versus-host disease assessment. Bone Marrow Transplant 2018; 53: 1401–1415.

10.

Cecconi

Evans

Levy

, et al. Sepsis and septic shock. Lancet 2018; 392: 75–87.

11.

Yang

Zhang

, et al. IgA, albumin, and eosinopenia as early indicators of cytomegalovirus infection in patients with acute ulcerative colitis. BMC Gastroenterol 2020; 20: 294.

12.

Srivastava

Bodnar

Osman

, et al. Serum albumin level before kidney transplant predicts post-transplant BK and possibly cytomegalovirus infection. Kidney Int Rep 2020; 5: 2228–2237.

13.

Chen

Liu

Liang

, et al. Development and validation of a nomogram for predicting albumin transfusion after spinal tuberculosis surgery: based on propensity score matching analysis. World Neurosurg 2022; 157: e374–e389.

14.

Abedi

Zarei

Elyasi

Albumin: a comprehensive review and practical guideline for clinical use. Eur J Clin Pharmacol 2024; 80(8): 1151–1169.

15.

Takahashi

Mochizuki

Sano

, et al. Decline of serum albumin precedes severe acute GVHD after haploidentical HSCT. Pediatr Int 2021; 63: 1048–1054.

16.

Rashidi

DiPersio

Westervelt

, et al. Peritransplant serum albumin decline predicts subsequent severe acute graft-versus-host disease after mucotoxic myeloablative conditioning. Biol Blood Marrow Transplant 2016; 22: 1137–1141.

17.

Ayuk

Bussmann

Zabelina

, et al. Serum albumin level predicts survival of patients with gastrointestinal acute graft-versus-host disease after allogeneic stem cell transplantation. Ann Hematol 2013; 93: 855–861.

18.

Yang

Zhang

Huang

, et al. Risk factors analysis for human cytomegalovirus viremia in donor+/recipient+ hematopoietic stem cell transplantation. Lab Med 2020; 51: 74–79.

19.

Boussiotis

Düver

Weißbrich

, et al. Viral reactivations following hematopoietic stem cell transplantation in pediatric patients—a single center 11-year analysis. PLoS One 2020; 15: e0228451.

20.

Thiant

Moutuou

Leboeuf

, et al. Homeostatic cytokines in immune reconstitution and graft-versus-host disease. Cytokine 2016; 82: 24–32.

21.

Ding

Ning

H-M

P-L

, et al. Tumor necrosis factor α in aGVHD patients contributed to the impairment of recipient bone marrow MSC stemness and deficiency of their hematopoiesis-promotion capacity. Stem Cell Res Ther 2020; 11(1): 119.

22.

Shono

Ueha

Wang

, et al. Bone marrow graft-versus-host disease: early destruction of hematopoietic niche after MHC-mismatched hematopoietic stem cell transplantation. Blood 2010; 115: 5401–5411.

23.

Uppuluri

Subburaj

Jayaraman

, et al. Cytomegalovirus reactivation posthematopoietic stem cell transplantation (HSCT) and type of graft: a step toward rationalizing CMV testing and positively impacting the economics of HSCT in developing countries. Pediatr Blood Cancer 2017; 64(11): 26639.

24.

Kim

Sohn

, et al. Clinical impact of early absolute lymphocyte count after allogeneic stem cell transplantation. Br J Haematol 2004; 125: 217–224.

25.

Melendez-Munoz

Marchalik

Jerussi

, et al. Cytomegalovirus infection incidence and risk factors across diverse hematopoietic cell transplantation platforms using a standardized monitoring and treatment approach: a comprehensive evaluation from a single institution. Biol Blood Marrow Transplant 2019; 25: 577–586.

26.

Chen

Pang

Zhao

, et al. Risk factors for CMV infection within 100 days posttransplantation in patients with acute leukemia. Blood Sci 2022; 4: 164–169.

27.

Shen

M-Z

Hong

S-D

Wang

, et al. A predicted model for refractory/recurrent cytomegalovirus infection in acute leukemia patients after haploidentical hematopoietic stem cell transplantation. Front Cell Infect Microbiol 2022; 12: 862526.

28.

Fang

F-Q

Fan

Q-S

Yang

Z-J

, et al. Incidence of cytomegalovirus infection in Shanghai, China. Clin Vaccine Immunol 2009; 16: 1700–1703.

29.

Pencina

D’Agostino

RB.

Overall C as a measure of discrimination in survival analysis: model specific population value and confidence interval estimation. Stat Med 2004; 23: 2109–2123.

30.

Iasonos

Schrag

Raj

, et al. How to build and interpret a nomogram for cancer prognosis. J Clin Oncol 2008; 26: 1364–1370.

31.

Gan

Chen

, et al. Predicting the risk of acute kidney injury after hematopoietic stem cell transplantation: development of a new predictive nomogram. Sci Rep 2022; 12: 15316.

32.

Huang

Wang

H-Y

Jian

, et al. Development and validation of a nomogram to predict the risk of death within 1 year in patients with non-ischemic dilated cardiomyopathy: a retrospective cohort study. Sci Rep 2022; 12: 8513.

33.

Chen

Gan

Huang

, et al. Blood transfusion risk prediction in spinal tuberculosis surgery: development and assessment of a novel predictive nomogram. BMC Musculoskelet Disord 2022; 23: 182.