We analyzed data from 79 adult patients undergoing auto-HSCT at a Spanish tertiary hospital from January 2021 to April 2023. All of them were treated within the ‘RHeNutrir’ program, which includes a screening for malnutrition upon admission using the Malnutrition Universal Screening Tool (MUST) and Nutritional Risk Screening (NRS-2002) scales (See Supplementary Figures 1, 2) to make a correct nutritional diagnosis based on Global Leadership Initiative on Malnutrition (GLIM) criteria. Additionally, we conducted a complete morphofunctional assessment both upon admission and at discharge. All patients were monitored throughout their hospitalization and for at least 4 weeks after discharge.

Basic anthropometric data such as weight, height, body mass index (BMI), arm and calf circumferences, and biochemical parameters including albumin, prealbumin, cholesterol, and C-reactive protein (CRP) were collected. In addition, advanced nutritional tests were carried out, such as hand-held dynamometry using a Jamar® dynamometer, bioelectrical impedance analysis (BIA) with Nutrilab™ equipment from Akern® (utilizing a standard four-pole technique with a sinusoidal current and a frequency of 50 kHz), and MicroCaya® muscle ultrasound on the lower third of the anterior rectus quadriceps to measure X and Y axes as well as muscle area. Abdominal fat distribution (preperitoneal fat, total, and superficial subcutaneous fat) was measured at the midpoint between the xiphoid process and umbilicus. The assessment was performed by nutritionists who were specifically trained for this purpose.

Subsequently, the outcomes of auto-HSCT in terms of red blood cell transfusion requirements, the need for parenteral nutrition (PN), length of hospital stay, and the development of mucositis, fever, bacteremia, admission to the intensive care unit (ICU), early readmission, and mortality were reviewed in the electronic medical record. The aim was to identify morphofunctional values upon admission that could predict the risk of complications or, conversely, provide protection. These values were analyzed in the overall cohort as well as in subgroups based on age (over and under 65 years), sex, hematological disease (MM or lymphoma), and the presence or absence of cardiovascular risk factors (CVRF).

Quantitative variables are summarized as the mean and standard deviation for normally distributed data, and as the median (interquartile range) for skewed data. Normality was assessed using the Shapiro–Wilk test. Categorical variables are presented as counts and percentages. Anthropometric, nutritional, and hand-held dynamometry variables were compared between sexes using Student’s T-test and across age groups using one-way ANOVA.

For each outcome, potential predictors were first screened using univariable logistic regression; variables with p < 0.25 and < 20% of missing data were considered for the multivariable model. Imputation of missing data was carried out performing multiple imputation by chained equations (MICE). Fifty imputed datasets were generated, and model estimates were combined using Rubin’s rules. The imputations showed convergence, and the goodness of fit of the models was assessed using the Hosmer-Lemeshow test. Analyses were conducted in R using the mice package.

Optimal cut-off points for quantitative predictors were determined from receiver-operating-characteristic (ROC) curves by maximizing the Youden J statistic (sensitivity + specificity − 1). The discriminative performance of the final multivariable models was evaluated using the area under the ROC curve (AUC). ROC curves were computed using Leave-One-Cut cross validation.

The study was conducted in accordance with the ethical principles outlined in the Declaration of Helsinki. It was approved by the institutional review board of the Health Research Institute Fundación Jiménez Díaz with study code EO250-23_FJD.

All participants provided written informed consent within the 30 days prior to the start of auto-HSCT.

Here we present the results of a retrospective, single-center study conducted at a tertiary hospital. From January 2021 to April 2023, a total of 79 patients who underwent auto-HSCT within the ‘RHeNutrir’ program were analyzed, with a median follow-up of 37 months (range 1–53 months).

The median age of the overall group was 58 years (range 28–72), with the MM group having a median age of 59 years (range 39–72) and the lymphoma group having a median age of 56.5 years (range 28–68). Thirty-one patients (39.2%) were older than 60 years and 18 (22.8% of the overall cohort) were over 65 years old. Twenty-one patients (26.6%) had no relevant medical history, and 41 (51.9%) had at least one CVRF. The main characteristics of the patients are shown in Table 1.

In the MM group, melphalan 200 mg/m2 was used as the conditioning regimen, while in the lymphoma cohort, a combination of carmustine 300 mg/m2, etoposide 200 mg/m2, cytarabine 200 mg/m2/12 h and melphalan 140 mg/m2 (BEAM) was used. A median of 3.53 × 10⁶ CD34 + cells/kg and 3.7 × 10⁶ total nucleated cells/kg of patient weight were infused. G-CSF was used per protocol from day +5 until neutrophils reached 1,000/mm3 for 2 consecutive days. The outcomes of auto-HSCT are represented in Table 2.

Considering the nutritional characteristics of our series, 32 patients (40.5%) had a MUST score >2 and a median NRS-2002 of 4 at admission. Twenty-seven patients (34.2%) were obese (BMI > 30), with 22 (81.5%) classified as grade 1 obesity and 5 (18.5%) as grade 2.

The median albumin levels at admission and discharge were 4.2 g/dL and 3.5 g/dL, respectively. The median cholesterol levels at admission were 184 mg/dL, and at discharge, they were 146 mg/dL. The median CRP levels at admission were 0.2 mg/dL, and at discharge, they were 1.6 mg/dL.

Seventy-seven patients (97.5%) received oral nutritional supplements (ONS), and 66 (83.5%) received PN for a median of 13 and 12 days, respectively.

At admission, the mean weight of men was higher than that of women (83.5 kg vs. 70.7 kg; p < 0.001). The mean weight loss during the transplant admission was estimated at 3.36% in men and 2.29% in women (p = 0.11). The evolution of anthropometric and nutritional parameters during the procedure is shown in Table 3.

The mean hand-held dynamometry at admission was 35.1 kg in men and 20.7 kg in women (p < 0.001); at discharge, it was 33.5 kg and 20.7 kg, respectively. Therefore, there was a variation during admission of −1.6 kg in males and −0.11 kg in females (p = 0.14). The results stratified by different age groups are shown in Table 4.

We have determined some preliminary findings on the impact of these factors on different post-transplant outcomes, not only in univariate but also in multivariate studies for different subgroups of patients and for the entire cohort. These data are presented in Tables 5, 6, as well as Supplementary Table 1.

Classically, the HCT-CI has been the comorbidity index used in the context of HSCT to predict mortality related to toxicity (7). However, it was designed for allogeneic transplantation and has only been validated retrospectively in a cohort of auto-HSCT in MM (8). Moreover, it is an exclusive predictor of mortality, without discriminating its causes and does not include the risk of specific complications. Some of these conditions are potentially serious and, while they do not necessarily lead to the patient’s death, they generate comorbidity, high resource consumption, and reduce the patient’s quality of life. In this sense, some groups have published the pre-transplant EASIX (Endothelial Activation and Stress Index) as a predictor of complications of endothelial dysfunction, admission to the ICU, or death (9); however, once again, this score has only been explored in the context of allogeneic transplantation. Therefore, there is a lack of predictive markers for specific complications in auto-HSCT.

On the other hand, in recent years, there has been increasing evidence that malnutrition and sarcopenia are risk factors for death in cancer patients (10) and it is known that those with malnutrition risk prior to HSCT had an increased risk of death during the 1-year follow-up period (11). However, these conclusions were drawn based on a limited number of parameters without additional body composition studies.

Therefore, we firstly suggest conducting a morphofunctional assessment of patients before auto-HSCT, including not only anthropometric measurements or biochemical values but also body composition tests (such as BIA and muscle ultrasound), and muscle function determination (using hand-held dynamometry). Through this approach, we have been able to identify specific groups at a higher risk for different complications.

Based on our data, the NRS-2002, which is a validated malnutrition screening tool for hospitalized patients (12), is related to significant WL during the procedure. Furthermore, a high score on the NRS-2002 scale is a potential indicator of bacteremia in women and patients with MM, as well as prolonged hospitalization in this group.

Regarding to comorbidity, having no other medical conditions is linked to better outcomes. In the allogeneic setting, the analysis of HCT-CI organ subgroups has shown a strong association between cardiac disease and non-relapse mortality (NRM) (13). In our study, pulmonary diseases like chronic obstructive pulmonary disease (COPD) or asthma, increase the risk of mucositis, ICU admission, or readmission after discharge. Diabetes mellitus is also associated to a higher need for red blood transfusions, a greater incidence of bacteremia in men, and longer hospital stay for patients with MM. Previous research by Neupane N. et al. has reported an increase in mean hospital charges for diabetic patients (14). Therefore, in addition to performing pre-transplant tests, it would be desirable to optimize the management of these diseases before auto-HSCT by involving the specialists such as pulmonologists, endocrinologists, etc.

Hemoglobin levels lower than 11.6 g/dL upon admission are linked to an increased risk of mucositis, bacteremia, ICU admission, and a greater need for red blood cell transfusions. This is crucial because, additionally, in the multivariate model, both the development of bacteremia and ICU admission are associated with higher mortality. Therefore, improving anemia should be considered before auto-HSCT. It could be argued that anemia is a surrogate marker of poor control of the underlying disease, although in our series, there is no correlation between the degree of response and other outcomes such as WL.

Moving forward, it is important not to solely focus on weight or BMI, but to conduct studies that delve deeper into body composition, as fat has an inflammatory profile. Sarcopenic obesity predicts mortality in lymphoma patients undergoing auto-HSCT (15). In our series, a greater amount of adipose tissue measured by ultrasound is associated with a higher risk of fever during the procedure in women and a higher rate of ICU admission in patients with MM, confirming the increased risk of complications induced by the fatty component.

In relation to inflammatory markers, CRP levels have been associated with bacteremia, mucositis grade, length of neutropenia, and hospitalization and death in auto-HSCT (16). Elevated CRP concentration (higher than 1.85 mg/dL) remained predictive of NRM in the allo-HSCT setting with reduced-intensity conditioning regimen (17). In our study, CRP levels higher than 0.6 mg/dL at admission are associated not only with bacteremia, prolonged hospitalization, and death but also with a greater need for red blood cell transfusion and admission to the ICU. In specific subgroups, in men and patients with MM, it is associated with a need for PN longer than 13 days and, in those over 65 years of age, with greater WL during auto-HSCT. Furthermore, in men and patients with CVRF, an increase in CRP levels at discharge compared to admission is associated with a higher probability of early readmission. We are concerned that we have chosen a not very high CRP value in order to ensure high specificity.

Another relevant aspect is that the development of some complications predisposes to the onset of others (18). Mucositis, especially early-onset, has been associated in multivariate analysis with fever, bacteremia, and longer hospitalization. The development of bacteremia, as previously mentioned, has been linked to higher mortality, so taking measures to prevent mucositis as much as possible is essential. Home-based transplantation has been shown to significantly reduce the incidence of mucositis and diarrhea (19, 20), so this modality should be considered to reduce toxicity, especially in patients with lymphoma, who are at greater risk of bacteremia and longer hospitalization according to our data.

The PREDyCES (21) and EFFORT (22) trials have shown that nutritional intervention in hospitalized patients can decrease comorbidity and enhance outcomes. However, these trials were conducted on highly diverse groups, such as cancer patients in general, excluding those who had undergone HSCT, and did not include a morphofunctional study. Our data suggests that nutritional therapy can improve the protein profile in auto-HSCT patients, decreasing the likelihood of readmission if the albumin level at discharge is > 3.3 g/dL.

However, there are no previous reports in the literature on the predictive value of hand-held dynamometry. In our study, muscle strength on admission is related to a lower incidence of various complications: greater needs for PN and blood red cell transfusion, mucositis, ICU admission, longer hospitalization, and even death. To our knowledge, sarcopenia has already been associated with higher mortality in auto-HSCT in lymphoma, although it has been defined only in terms of muscle mass using CT scan without taking function into account (23, 24). Furthermore, we believe that hand-held dynamometry is a cheaper and more accessible test that can be performed on all patients, even those who do not routinely undergo a CT scan.

A hand-held dynamometry value of ≥ 17 kg in women and ≥ 31 kg in men is associated with lower development of the aforementioned complications in our cohort. We have even observed that it reduces the risk of bacteremia in patients with MM. Therefore, physical activity and nutritional treatment could be an interesting approach in the months leading up to auto-HSCT in order to preserve or enhance muscle strength and function. After all, it is a modifiable parameter and can counteract other risk factors that are difficult or impossible to intervene in, such as age or type of hematologic malignancy.

The determination of body composition using BIA is an emerging practice that is generating greater interest in the context of HSCT. A low pre-transplant phase angle was identified as an independent risk factor for the development of infection early after transplantation in allogeneic-HSCT (25). However, there is no data in the auto-HSCT setting. In our study, a higher FFMI at admission resulted in less WL during the procedure. Furthermore, in patients under 65 years old, a higher FFM, FFMI, skeletal mass index (SMI), and body cell mass (BCM) at admission are also associated with lower red blood cell transfusion requirements. In men, a greater BCM at admission is related to a lower development of fever, likely due to the anti-inflammatory component of lean mass and muscle.

The major limitation of our work is based on the retrospective nature of the study, as well as its single-center design with a sample of 79 patients. Furthermore, the use of a specific screening and nutritional monitoring protocol may limit the external validity of our results when attempting to extrapolate them to hospitals without similar structured morphofunctional assessment program. Therefore, in these cases, data should be interpreted with caution.

On the other hand, the study focused on patients eligible for auto-HSCT, and should not be compared to those who were not. In addition, the measurement tools used were those described in the materials and methods section, potentially introducing bias if different devices were utilized. It is also worth noting that while the nutritionists conducting the measurements were specifically trained, no blinding methods were employed. So, while our study provides valuable insights, its limitations should be considered when applying the results to other settings or populations.

For the handling of missing values, multiple imputation of data has been carried out. Although this is an appropriate technique, as it is more robust than others, there may be a risk of overfitting or underestimating variability.

However, this is the longest series of auto-HSCT in which a comprehensive morphofunctional assessment has been carried out, applying the recommendations of experts for better early identification of malnutrition (26).

Therefore, while efforts to refine mortality predictive scores in auto-HSCT (27) are continuously ongoing, it remains necessary to introduce new parameters that, as much as possible, are modifiable factors. In this sense, we believe that values obtained from pre-transplant morphofunctional assessment may be useful both in general and in different subgroups based on age, sex, type of disease, and presence or absence of CVRF.

The recommendation to implement rehabilitation after the discharge of patients undergoing HSCT has been established previously (28), but our data suggest that designing prehabilitation programs is desirable to optimize the clinical and nutritional patient status before the procedure, to monitor more closely, and to individualize those profiles most susceptible to developing complications. However, these results need to be validated in external cohorts before recommendations for clinical practice can be established.

In conclusion, it would be beneficial to establish multidisciplinary teams within hematopoietic transplant units that include nutritionists and physiotherapists. This would allow for more systematized screening and monitoring of these patients.

A morphofunctional assessment should be performed on patients prior to auto-HSCT, including body composition tests (BIA and ultrasound) and muscle function determination (hand-held dynamometry), irrespective of age, gender, or hematological condition. Nevertheless, we acknowledge that it may be challenging to conduct a comprehensive assessment in all centers. However, hand-held dynamometry is widely available and provides valuable information. Moreover, tracking dietary intake twice a week can assist in monitoring patients and guiding nutritional interventions. It is important that assessments are not one-time occurrences, but rather conducted at various intervals to assess treatment response and evolving patient needs.

Comorbidity, anemia, malnutrition, and sarcopenia have been associated with transplant complications and toxicity. The development of bacteremia and intensive care unit admission increase the risk of death. However, a greater lean mass and better muscle strength (≥ 17 kg in women and ≥ 31 kg in men) are related to better outcomes. Knowledge of risk factors allows us to monitor and individualize the procedure, improving outcomes.

There are still many challenges to address in this area. Future research could delve deeper into the predictive value of variables like BCM or phase angle, or even identify specific biomarkers in transplant patients such as interleukins, adhesion molecules, etc.

Furthermore, the design of specific physical activity programs recommended for these patients remains to be clarified. They should be carefully tailored to each individual, considering their treatment, vital signs, hemoglobin levels, and platelet counts, among others. Further and prospective research in larger patient populations is needed.