A predictive model combining clinical characteristics and nutritional risk factors for overall survival after umbilical cord blood transplantation

Background Umbilical cord blood transplantation (UCBT) is a curable therapy for hematological disease; however, the impact of nutritional status on UCBT outcomes remains controversial. To evaluate the joint effect of clinical characteristics and nutritional status on the prognosis of patients who underwent UCBT, we screened various factors to establish a predictive model of overall survival (OS) after UCBT. Methods We performed an integrated clinical characteristic and nutritional risk factor analysis and established a predictive model that could be used to identify UCBT recipients with poor OS. Internal validation was performed by using the bootstrap method with 500 repetitions. Results Four factors, including disease status, conditioning regimen, calf skinfold thickness and albumin level, were identified and used to develop a risk score for OS, which showed a positive predictive value of 84.0%. A high-risk score (≥ 2.225) was associated with inferior 3-year OS post-UCBT [67.5% (95% CI 51.1–79.4%), P = 0.001]. Then, we built a nomogram based on the four factors that showed good discrimination with a C-index of 0.833 (95% CI 0.743–0.922). The optimism-corrected C-index value of the bootstrapping was 0.804. Multivariate analysis suggested that a high calf skinfold thickness (≥ 20.5 mm) and a low albumin level (< 33.6 g/L) conferred poor disease-free survival (DFS). Conclusion The predictive model combining clinical and nutritional factors could be used to predict OS in UCBT recipients, thereby promoting preemptive treatment.


Introduction
Umbilical cord blood transplantation (UCBT) has been performed to treat hematological and nonhematological diseases for over 30 years, with the advantages of availability, fewer restrictions associated with human leukocyte antigen (HLA) matching, a low rate of relapse for patients with positive minimal residual disease (MRD) pretransplant and a low incidence of chronic graft-versus-host disease (GvHD) [1,2].However, the low total nucleated cell (TNC) and CD34 + cell doses in a single cord blood unit retrain the curative effect of UCBT, which may result in the occurrence of delayed engraftment, graft failure and infection that increases the risk of transplant-related mortality (TRM) [1][2][3].
Hematopoietic stem cell transplantation (HSCT) carries nutritional risks resulting from high-dose chemotherapy alone or in combination with radiation therapy [8,9].In a prospective study, 21.2% of patients were at nutritional risk before HSCT according to Nutritional Risk Screening 2002 (NRS-2002), whereas the nutritional risk rate increased to 100% posttransplant [10].Good nutritional status is beneficial for graft engraftment and immune reconstitution [11].Furthermore, several studies have reported that disordered nutritional status during HSCT is related to inferior clinical outcomes as well as a higher complication rate during treatment, including reduced body mass index (BMI) [12], a decline in bone mineral density [13], lower serum albumin levels [14] and low vitamin D levels [13].In recent years, a series of scales have been used to evaluate nutritional status or quality of life for predicting patient outcome, such as the Patient-Generated Subjective Global Assessment (PG-SGA) [15] and the Functional Assessment of Cancer Therapy-Bone Marrow Transplant (FACT-BMT) [16].
There is uncertainty regarding the extent to which nutritional parameters influence clinical outcomes in UCBT recipients.Thus, in this study, we assessed the joint effect of clinical characteristics and nutritional status on overall survival (OS) post-UCBT in adult recipients and screened risk factors to build a predictive model for identifying high-risk patients for early intervention.

Patients
To effectively conduct questionnaire evaluation, we performed the questionnaire investigation with only youth and adults less than 65 years of age.Between September 2018 and December 2021, a total of 80 patients who underwent UCBT at the Department of Hematology, the First Affiliated Hospital of the University of Science and Technology of China (USTC), and received systemic nutritional evaluation were included in this study.The median age of the patients was 31 years (range 17-61 years), and 46.3% (37/80) were female.All surviving patients were followed up from the date of transplantation until June 30, 2022, and the median follow-up time was 719.0 days (range 54-1262 days).The procedures were approved by the Ethics Committee of the First Affiliated Hospital of USTC (Approval number, 2022-RE-253).Patients or guardians provided informed consent before transplantation and for the use of data for research in accordance with the Declaration of Helsinki.

Transplant protocols
The UCBT protocols were previously reported [17,18].The myeloablative conditioning (MAC) regimen was performed in 86.2% (69/80) of the patients, and the other 11 patients were treated with a reduced intensity conditioning (RIC) regimen.All patients were given cyclosporine (CsA) and mycophenolate mofetil (MMF) as GvHD prophylaxis after UCBT.

Nutritional status and quality of life assessments
To evaluate the validity and quality of life questionnaire in patients who received UCBT and assess the effects on survival prognosis, a series of scales were evaluated in this study at day 30 after UCBT, including the Patient-Generated Subjective Global Assessment (PG-SGA) [19], the Exercise of Self-Care Agency (ESCA) [20], the General Self-Efficacy Scale (GSES) [21], the European Cancer Research and Treatment Organization Quality of Life Questionnaire-Cancer30 (EORTC-QLQ-C30) [22] and the Functional Assessment of Cancer Therapy-Bone Marrow Transplant (FACT-BMT) [23].Body mass index (BMI), calf skinfold thickness, calf circumference, hand grip and albumin level were measured on the same day posttransplant, and BMI was also evaluated pretransplant.Patient weight was classified as underweight (BMI < 18.5 kg/m 2 ), normal (BMI 18.5-24.9kg/m 2 ), and overweight (BMI ≥ 25 kg/m 2 ) according to the World Health Organization [24].

Definitions
OS was calculated from the date of transplantation until death or the last follow-up, and disease-free survival (DFS) was defined as the time from transplantation to either relapse or death of any cause.In the computation of the cumulative incidence of relapse (CIR) and nonrelapse mortality (NRM), relapse and death were considered competing events [25].

Statistical analysis
For measurement data, the normality and outliers were explored by using histogram.Estimated probabilities for OS and DFS were calculated by using the Kaplan-Meier method, and the significance levels associated with the survival curves were measured by using the log-rank test.The evaluation of CIR and NRM was performed by using Gray's test.Univariate and multivariate analyses were evaluated using the Cox proportional hazard regression model.Factors with a P value < 0.1 in the univariate analysis were subjected to multivariate analysis.To validate the model, internal verification was performed using the bootstrapping method across 500 replicates, and the optimism-corrected C-index was calculated.The "rms" package of R version 4.2.1 software was used to prepare the nomogram and bootstrap.Receiver operating characteristic (ROC) curve analyses were performed, and areas under the curve (AUCs) were calculated with OS as the actual state variable.All statistical analyses were performed using SPSS 20.0.Figures were drawn by using R software (version 4.2.1) and GraphPad Prism 9. A P value less than 0.05 was considered statistically significant.

Nutritional status and quality of life evaluation
Table 2 provides the results of the patients' nutritional status and quality of life evaluations.Fifty-one (63.8%) patients had a normal BMI, 11 (13.7%) had a low BMI, and 18 (22.5%)had a high BMI.A total of 78.8% (63/80) of the patients had weight loss calculated from the initial weight before UCBT to the weight at day  1C, D).

Risk factors for survival
Furthermore, we evaluated a series of parameters, including clinical characteristics, nutritional status and quality of life evaluation indexes, for a possible association with an increased risk of OS and DFS by using univariate Cox regression analysis, as shown in

Development of a predictive model for OS
According to the multivariate regression analysis, disease status, conditioning regimen, calf skinfold thickness and albumin level were screened to construct a predictive model for OS (Fig. 2A).Furthermore, we calculated the risk score based on the individual expression levels of the four risk factors, where the risk score = 1.342 × v1 + 1.630 × v2 + 2.603 × v3 + 1.848 × v4 (Table 4).The time-dependent AUC was 0.840 (95% CI 0.734-0.946,P < 0.001, Fig. 2C), which suggested that the model for OS had considerable discriminative abilities.

Calf skinfold thickness for predicting disease-free-survival
The cutoff point of the calf skinfold thickness was 20.  4D].The CIR rates between the two groups showed no differences (P = 0.762, Fig. 4C).

Discussion
Chemotherapy and radiation therapy not only damage tumor cells but also significantly impair proliferative cells, such as colonic epithelial cells and lymphocytes, which may cause metabolic disorders and undernutrition [11].Our results showed that 13.75% (11/80) of patients had a low BMI of < 18.5 kg/m 2 before UCBT, and the proportion increased to 27.5% (22/80) at day 30 post-UCBT.Moreover, 47.5% (38/80) of patients experienced weight loss of > 8% of their initial weight before transplantation.These data suggested that patients were at high nutritional risk during transplantation, which was similar to a previous report by Peng Liu et al. [10].They enrolled 170 allo-HSCT recipients and found that 50.46% of the patients had weight loss of more than 10% post-HSCT.
As in previous reports, various factors influence the outcome of HSCT [4,5,7,26].In addition to clinical characteristics, nutritional status plays an important role in patient survival [7,11].Thus, in this study, we analyzed risk factors affecting survival by combining clinical factors (such as diagnosis and disease status), nutritional and physical functional assessment indicators, including laboratory tests (albumin level), physical measures (BMI, calf skinfold thickness, calf circumference and hand grip) and scales (PG-SGA, ESCA, GSES, EORTC-QLQ C30 and FACT-BMT).In multivariate analysis, RIC regimen, PR/NR status before transplantation, calf skinfold thickness and albumin level were independent risk factors for OS.A higher calf skinfold thickness and lower albumin level were related to poorer DFS.Then, a risk model for OS was established based on the four factors.The patients with high-risk scores (≥ 2.225) had poorer survival than those with low-risk scores [3-year OS: 67.5% (95% CI 51.1-79.4%),P = 0.001; 3-year DFS: 62.7% (95% CI 46.3-75.4%),P = 0.014] (Fig. 3).
Although many indicators were used to evaluate nutritional status in UCBT recipients, the results showed that only calf skinfold thickness and albumin level were related to survival.In our study, the above scales seem useless for predicting outcomes in patients undergoing UCBT, which suggests that laboratory indictors and physical measurements are more important than subjective scales for predicting the survival of UCBT recipients.Although we choose the same time-point of survey, the results of scales may also have bias due to differences in education and the physical and mental state of patients.Our results could simplify the evaluation type of scale and provide a practical direction for clinical work.
Skinfold thickness reflects body fat level.Researchers usually assess fat mass by using the skinfold thickness at 5 to 9 body sites, such as the triceps, biceps, abdominal and calf skinfold thicknesses [10,33].In our center, UCBT recipients underwent insertion of central venous catheters on both upper arms.Therefore, we measured the circumference and skinfold thickness of the calf instead of the arm.In the present study, 14 patients had a higher calf skinfold thickness (≥ 20.5 mm), and among them, 57.1% (8/14) had a weight loss of > 8%.The significance of calf skinfold thickness on transplantation outcomes has not been reported.We observed that patients with a high calf skinfold thickness had an 8.289-fold risk of inferior OS than patients with a lower skinfold thickness (P = 0.011) and a 3.723-fold risk of poorer DFS (P = 0.016).Moreover, the measurement of calf skinfold thickness is a noninvasive and convenient examination that can be closely monitored in UCBT recipients.
The role of BMI in predicting outcomes is controversial.In a retrospective analysis of 2503 patients who underwent HSCT, the authors found that both underweight and obese patients had an increased NRM compared with normal-weight HSCT recipients [34].Prasad and collaborators [35] conducted a randomized controlled phase-3 open-label trial to evaluate the effect of arm anthropometry on nutritional assessment, and the study showed that the addition of arm anthropometry (mid-upper arm circumference and triceps skinfold thickness) to BMI increased the sensitivity of nutritional evaluation.However, in our study, BMI and the decline in BMI post-HSCT showed no significant effect on OS and DFS, which was supported by other studies [36,37].
The risk score generated from the 4 factors we identified could be used to predict OS with an AUC of 0.840 (Fig. 2C).Furthermore, based on these factors, we developed a nomogram for clinical application to help identify high-risk patients with inferior OS.Calibration plots of the nomograms showed that the nomograms performed well compared with an ideal model.By using this model, we can distinguish high-risk patients and provide early nutritional treatment.
To the best of our knowledge, this is the first study to evaluate the survival of UCBT patients by integrating clinical factors and various nutritional indexes and to build a risk model to identify high-risk patients and facilitate early interventions.However, there were some limitations in this study, such as a small sample size, a lack of a validation set, and the absence of detailed food consumption.Although internal validation by the bootstrap method with a corrected c-index of 0.804 was performed in our study, external validation is still important; thus, a multicenter clinical trial to validate our predictive model is necessary in future.Additionally, we did not investigate the specific mechanisms underlying the association between nutritional factors and UCBT outcomes, which warrants further exploration in future research.

Conclusion
In conclusion, the predictive model combining clinical and nutritional factors could be used to predict survival and stratified the survival of different groups in UCBT recipients, which may promote preemptive treatment.
• fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year

•
At BMC, research is always in progress.

Learn more biomedcentral.com/submissions
Ready to submit your research Ready to submit your research ?Choose BMC and benefit from: ? Choose BMC and benefit from:

Fig. 3
Fig. 3 Survival outcomes according to the risk model for OS in patients who underwent UCBT (low risk: n = 30; high risk, HR: n = 45).A OS; B DFS; C CIR; D NRM

Table 2
Nutritional status and quality of life evaluation of 80 patients BMI body mass index, PG-SGA patient-generated subjective global assessment, ESCA Exercise of Self-Care Agency, GSES General Self-Efficacy Scale, EORTC-QLQ-C30 European Organization for Research and Treatment of Cancer Quality of Life Questionnaire, FACT-BMT Functional Assessment of Cancer Therapy-Bone Marrow Transplantation scale, NA not available P = 0.024; albumin: HR 5.612, 95% CI 1.705-18.474,P = 0.005).

Table 3
Univariate and multivariate analysis of OS in 80 patients treated with UCBT

Table 4
A panel of four factors with predictive value for OS