Introduction

Anthracyclines constitute a fundamental component in the therapeutic management of diffuse large B-cell lymphoma (DLBCL) [1]. Despite significant improvement in patient outcomes, anthracyclines is prone to causing cardiotoxicity, predominantly manifesting as cancer therapy-related cardiac dysfunction (CTRCD) [2].

CTRCD may decrease the likelihood of survival in patients with DLBCL [2].The optimal approach to mitigating CTRCD in individuals with DLBCL entails performing precise risk stratification for CTRCD prior to anthracyclines treatment [3].

The Heart Failure Association and International Cardio-Oncology Society (HFA-ICOS score was recommended by 2022 European Society of Cardiology guidelines on cardio-oncology for baseline risk stratification of CTRCD in DLBCL patients receiving anthracyclines [3]. Although the HFA-ICOS score has been validated for baseline risk stratification of symptomatic or moderate-to-severe asymptomatic CTRCD in anthracyclines-treated patients in the CARDIOTOX registry [4]. Our researches have revealed that the HFA-ICOS score is not effective in the baseline risk stratification of CTRCD [5, 6]. Therefore, new biomarkers are needed for improved the HFA-ICOS score for baseline risk stratification of CTRCD.

A body mass index (BMI) over 30 kg/m² is an important part of the HFA-ICOS score and is considered a risk factor for CTRCD in DLBCL [7]. However, BMI is not always an accurate proxy for adiposity at the individual level and fails to characterize adipose tissue distribution [8]. Myosteatosis is characterized by the intramuscular and intermuscular adipose tissue infiltration within skeletal muscle (SM) and is defined as a condition in which muscle density derived from computed tomography (CT) falls below the threshold of the lowest gender-specific quartile [9, 10]. Recent studies have found that CT-derived myosteatosis is strongly correlated with adverse cardiovascular outcomes in populations without cancer independently of BMI and subcutaneous and visceral adiposity [9, 11]. However, whether the myosteatosis derived from staging CT can be used for baseline risk stratification of CTRCD in patients with DLBCL receiving anthracyclines remains unclear.

Consequently, this study sought to investigate the significance of staging CT-derived myosteatosis in baseline risk stratification of CTRCD in patients with DLBCL receiving anthracyclines.

Materials and methods

The institutional review board at two hospitals approved the retrospective cohort study (Review Number CZLS2023017-A; Review Date: January 20, 2023), which followed the Declaration of Helsinki (2024 revision). The requirement for informed consent was waived.

Study population

From July 2013 to June 2018, consecutive patients with DLBCL at Chongqing University Cancer Hospital and the Second Affiliated Hospital of Chongqing Medical University were retrospectively enrolled. The inclusion criteria were as follows: (i) DLBCL patients receiving anthracyclines; (ii) abdominal CT for tumor staging less than 2 weeks before anthracyclines treatment; and (iii) complete pretreatment clinical information and follow-up data. The exclusion criteria were as follows: (i) anti-tumor treatment prior to abdominal CT; and (ii) Presence of significant artifacts in abdominal CT images. Baseline clinical data were obtained from electronic medical records. The detailed assessment of HFA-ICOS score was shown in Supplemental Methods.

Abdominal CT protocol

All the patients underwent non-contrast abdominal CT scans for tumor staging in DLBCL. The detailed CT parameters are presented in Supplemental Table 1.

CT-derived body composition analysis

CT body Composition were measured using semiautomated threshold-based segmentation with 3D Slicer, version 4.10 (www.slicer.org) by two trained research assistants (reader 1, C.R.T and reader 2 H.S.S., with 5 and10 years of experience in body composition analysis, respectively) who blinded to baseline clinical and follow-up outcomes information. The area and average CT attenuation of subcutaneous adipose tissue (SAT), visceral adipose tissue (VAT), intra and inter-muscular adipose tissue (IMAT) and SM were measured on a single axial CT image at the level of the third lumbar vertebra. The SAT compartment was identified as adipose tissue between the skin and the membranous layer of the superficial fascia, with attenuation values from − 190 to -30 Hounsfield units (HU). The VAT was measured using attenuation values from − 150 to -50 HU. SM, including the psoas major, erector spinae, quadratus lumborum, transversus abdominis, external and internal obliques, and rectus abdominis, was quantified using attenuation values from − 30 to 150 HU [12]. IMAT was quantified using attenuation values from − 190 to − 30 HU [11].

Myosteatosis was defined as the lowest gender-specific quartile (Q1) values of SM density quartile as previous literatures [9, 10].

Intra and inter-reader reliability analysis

Reader 1 measured the CT body composition parameters of all patients twice within a six-month period. Reader 2 blinded to the initial results measured CT body composition parameters of all patients. The intra-group correlation coefficient (ICC) is used to assess the intra and inter-reader reliability. The Dice similarity coefficients were computed using Python, version 3.9 (Python Software Foundation).

Outcomes

The primary outcome was CTRCD. According to European Society of Cardiology, CTRCD was defined as a reduction of at least 10% in left ventricular ejection fraction from baseline to a value below 50% by two-dimensional echocardiography [13] Echocardiography was performed at intervals of every two cycles during anthracyclines chemotherapy, every 3 months for the first year post-treatment, every 6 months for the subsequent 1–2 years, and annually thereafter.

The secondary outcome was all-cause mortality, all-cause mortality time was defined as the duration from anti-tumor treatment to death from any cause or last follow-up. Follow-up was conducted via inpatient and outpatient records and telephone interviews: every 3–6 months for the first two years, and every 6–12 months thereafter. All patients were followed for at least 1 year until June 30, 2025.

Statistical analysis

Statistical analyses were performed with SPSS (version 27.0; IBM) and R (Version 4.3.2; R Foundation for Statistical Computing). Continuous variables were expressed as means (± standard deviation) or medians (interquartile range, IQR), while categorical variables were presented as frequencies with corresponding percentages. For comparisons between groups, the Student’s t-test or Mann-Whitney U test was employed for continuous variables, whereas the chi-squared (χ²) test was utilized for categorical variables. The associations between CT body composition variables and CTRCD was using univariable and multivariable binary logistic regression. The performance of variables for predicting CTRCD was assessed by receiver operating characteristic analysis. the DeLong test was used to compare the area under the receiver operating characteristic curve (AUC) of each variable. The internal validation was performed using bootstrap analysis (500 bootstrap resamples) and calibration curve. The clinical utility was performed using decision curve analysis. The associations between CT body composition variables and all-cause mortality were using univariable and multivariable Cox regression analyses. The Schoenfeld residuals method was employed to test the proportional-hazard assumption. The cumulative survival rates were depicted by Kaplan–Meier survival curves and were compared by the log-rank test. A two-sided p-value of less than 0.05 was regarded as indicative of a statistically significant difference.

Results

Patient characteristics

Of the 860 patients, 137 were excluded for the following reasons: 24 did not receive anthracyclines treatment; 47 lacked pre-treatment abdominal CT scans; 32 had incomplete follow-up data; 21 had received anti-tumor treatment prior to abdominal CT; and 13 had unavailable abdominal CT images due to significant artifacts. Finally, 723 patients (392[54.2%] male patients) were included in this study (Fig. 1).

Fig. 1
figure 1

Flowchart of patient inclusion and exclusion criteria

Full size image

During a median follow-up of 65 months (IQR, 53–76 months), 180 (24.9%) patients developed CTRCD and 269 (37.2%) patients died.

The baseline characteristics of patients were summarized in Table 1. There were statistically significant differences in the age and HFA-ICOS score classifications between patients with CTRCD and those without (P < 0.001).

Table 1 Baseline characteristics of patients categorized based on CTRCD status
Full size table

Comparison of CT body composition parameters between patients with CTRCD and those without

The baseline CT body composition parameters were summarized in Table 2. Myosteatosis was defined by SM density values of ≤ 18.7 HU for female patients and ≤ 27.8 HU for male patients in this study.

Table 2 Comparison of CT body composition parameters between patients with CTRCD and those without
Full size table

The IMAT area of patients with CTRCD was higher than those without CTRCD (13.87 ± 5.58 vs. 12.56 ± 6.52, P = 0.016), as shown in Fig. 2. The SM density of patients with CTRCD was lower than those without CTRCD (26.48 ± 9.40 vs. 32.95 ± 8.89, P < 0.001), as shown in Fig. 2. The incidence of myosteatosis was greater in the patients with CTRCD compared to those without (56.1% vs. 12.3%, P < 0.001). Whereas there were no significant differences in the area and density of SAT and VAT, the IMAT density and the SM area between patients with CTRCD and those without (P > 0.05).

Fig. 2
figure 2

Body composition analysis of axial CT images at the level of the third lumbar vertebra. (a and b) A 66-year-old patient with stage III diffuse large B-cell lymphoma who received anthracyclines and did not develop CTRCD; and (c and d) A 67-year-old patient with stage III diffuse large B-cell lymphoma who received anthracycline and developed CTRCD. The segmentation results of body composition include subcutaneous adipose tissue (SAT, green), visceral adipose tissue (VAT, blue), intra and inter-muscular adipose tissue (IMAT, red) and skeletal muscle (SM, yellow). Patient with CTRCD exhibited myosteatosis, whereas that without CTRCD did not. CTRCD, cancer therapy–related cardiac dysfunction

Full size image

Associations between baseline clinical variables, CT body composition parameters and CTRCD

As shown in Table 3, variables with P < 0.05 in univariate analysis and clinically known cardiovascular disease protective factors regardless of P value in univariate analysis were included in multivariate analysis. After multivariable adjustment, HFA-ICOS score and myosteatosis were associated with CTRCD. In comparison to low-risk patients according to the HFA-ICOS score, the moderate-risk patients exhibited a 2.534-fold increase in the odds of developing CTRCD, with a 95% confidence interval (CI) ranging from 1.084 to 5.423. Moreover, high and very-high risk patients demonstrated a 3.845-fold increase in the odds of experiencing CTRCD, with a 95% CI of 1.934 to 6.123. Individuals diagnosed with myosteatosis were found to have 3.382 times higher odds of developing CTRCD compared to those without myosteatosis, with a 95% CI of 1.294 to 7.556.

Table 3 Univariable and multivariable logistic analysis of CTRCD
Full size table

Performance of HFA-ICOS score and CT body composition parameters for baseline risk stratification of CTRCD

The ROC curve analysis was shown in Fig. 3 and Table S2. The AUC for predicting CTRCD in myosteatosis was higher than that of the HFA-ICOS score [0.719, 95% CI (0.685, 0.751) vs. 0.586, 95%CI (0.549 to 0.622), P < 0.001). Combining both myosteatosis and the HFA-ICOS score, the performance of predicting CTRCD significantly improved, with an AUC of 0.828 and a 95% CI of 0.798 to 0.855 (P < 0.001). The calibration curve and decision curve analysis of combined model were shown in Figure S1.

Fig. 3
figure 3

The receiver operating characteristic curves for predicting the efficacy of CTRCD in patients with diffuse large B-cell lymphoma receiving anthracyclines. The AUC of combined myosteatosis and HFA-ICOS score for predicting CTRCD was higher than myosteatosis and HFA-ICOS score (0.828 vs. 0.719 vs. 0.586, all P < 0.001). AUC, area under the receiver operating characteristic curve; CTRCD: cancer therapy–related cardiac dysfunction; HFA-ICOS, heart failure association–international cardio-oncology society

Full size image

Associations between baseline clinical variables, CT body composition parameters and all-cause mortality

As shown in Table 4, variables with P < 0.05 in univariate analysis were included in multivariate analysis. After multivariable adjustment, age, Ann Arbor stage, international prognostic index (IPI) score and myosteatosis were associated with all-cause mortality. Patients aged 65 and older had 1.656 times higher odds of mortality compared to those under 65 (95% CI: 1.178–2.327). Those with Ann Arbor stage III-IV had 5.115 times higher odds of mortality than those with stage I-II (95% CI: (1.417–10.518). Patients with an IPI score of 2–5 had 1.830 times higher odds of mortality than those with a score of 0–1 (95% CI: 1.325–2.518). Additionally, individuals with myosteatosis had 2.015 times higher odds of mortality compared to those without it (95% CI: 1.360–2.994).

Table 4 Univariable and multivariable Cox proportional hazards models analysis of overall survival
Full size table

The Kaplan–Meier survival curves (Fig. 4) indicated that patients aged 65 years or older, those with Ann Arbor stage III-IV, an IPI score of 2–5, or myosteatosis exhibited a significantly higher all-cause mortality rate.

Fig. 4
figure 4

The Kaplan–Meier survival curves of all-cause mortality rate in diffuse large B-cell lymphoma) receiving anthracyclines stratified by age (a), Ann Arbor stage (b), IPI score (c) and myosteatosis (d). Patients aged 65 years or older, those with Ann Arbor stage III-IV, an IPI score of 2–5, or myosteatosis exhibited a significantly higher all-cause mortality rate (all P < 0.05). IPI, international prognostic index

Full size image

Intra and inter-reader reliability

The Dice similarity coefficients of the SAT, VAT, IMAT and SM were 0.86 (95% CI: 0.79, 0.92), 0.84 (95% CI: 0.77, 0.91), 0.82(95% CI: 0.76, 0.90) and 0.85 (95% CI: 0.79, 0.91). The intra and inter-reader reliability was excellent, with an ICC.

of 0.87 (95% CI: 0.79, 0.93) for SAT area, 0.88 (95% CI: 0.79, 0.95) for SAT density, 0.85 (95% CI: 0.77, 0.92) for VAT area, 0.86 (95% CI: 0.78, 0.94) for VAT density, 0.83 (95% CI: 0.76, 0.90) for IMAT area, 0.84 (95% CI: 0.78, 0.91) for IMAT density, 0.86 (95% CI: 0.79, 0.92) for SM area, 0.87 (95% CI: 0.78, 0.93) for SM density.

Discussion

This study explored the value of CT-derived myosteatosis in baseline risk stratification for cardiotoxicity in DLBCL receiving anthracyclines. Key insights are as follows: Myosteatosis derived from baseline CT and HFA-ICOS score were independently associated with CTRCD in patients with DLBCL receiving anthracyclines; The combination of myosteatosis and HFA-ICOS score improved the predictive performance of CTRCD; baseline CT-derived myosteatosis was independently associated with all-cause mortality in patients with DLBCL receiving anthracyclines, while HFA-ICOS score was not.

CT-derived myosteatosis represents the ectopic accumulation of adipose tissue both intermuscularly and intramuscularly, characterized by a decrease in muscle CT value and an increase in IMAT area [10, 14]. This study found that myosteatosis derived from baseline CT was independently associated with CTRCD in patients with DLBCL receiving anthracyclines. In lined with previous studies, myosteatosis was associated with chemotherapy‑related adverse cardiovascular outcomes in patients with early breast cancer [15] and low SM density, as a marker of myosteatosis, was associated with adverse complications in patients with lymphoma after autologous stem cell transplantation [16]. A plausible explanation was that myosteatosis induced systemic inflammation by enhancing the secretion of pro-inflammatory factors from adipocytes and facilitating the infiltration of immune cells [17]. This state of systemic inflammation elevated the risk of cardiovascular complications in patients undergoing anthracyclines-based chemotherapy. However, previous studies have also shown that baseline SM density was not associated with cardiotoxicity in pediatric, adolescent, and young adult lymphoma lymphoma [18, 19]. The inconsistencies between our findings and the previous results may be attributed to two main factors: our study focused on adult patients, while the previous research primarily involved underage patients, potentially affecting cardiac toxicity incidence. Additionally, our study was limited to DLBCL patients treated with R-CHOP, whereas the previous research included both Hodgkin and non-Hodgkin lymphoma patients with varied treatment regimens, which could result in different outcomes.

Our research indicated that HFA-ICOS score was less effective in predicting CTRCD (AUC = 0.586), aligning with studies on anthracyclines cardiotoxicity in breast cancer [5, 6, 20, 21]. However, findings from some studies were inconsistent [4, 22]. Possible reasons are as follows: First, regarding ethnic differences, this discrepancy is likely attributable to the exclusive focus of our study on Chinese subjects, whereas prior research has centered primarily on European and American populations. The HFA-ICOS score, originally developed and validated in Western cohorts, may not fully account for ethnicity-specific genetic and physiological characteristics unique to the Chinese population, thereby undermining its predictive performance in our sample. Second, in terms of BMI distribution, there are marked disparities between our study population and the European/American cohorts in the original research (e.g., the CARDIOTOX registry) [4]. The CARDIOTOX registry included a higher proportion of obese individuals (BMI ≥ 30 kg/m²), whereas our population exhibited a lower average BMI (23.06 ± 3.22 kg/m²) and a higher prevalence of normal-weight individuals. Given that BMI is a predictor of anthracycline-induced toxicity, this mismatch in BMI distribution between the two populations may have reduced the applicability of the HFA-ICOS score in our cohort. Third, differences in clinical practices may also contribute to the inconsistent results. In European and American clinical settings, anthracycline administration protocols (e.g., dose intensity, infusion duration) and supportive care strategies (e.g., routine use of cardioprotective agents) are relatively standardized, and long-term follow-up systems for cancer survivors are mature and systematic. In contrast, our study population received more heterogeneous anthracycline regimens, and the utilization rate of cardioprotective agents as well as the completeness of follow-up data varied slightly across participating centers. These practice-related variations may have altered the incidence and phenotypic pattern of anthracycline-induced cardiovascular toxicity, ultimately weakening the predictive validity of the HFA-ICOS score in our clinical context. Our research indicates that combining myosteatosis and HFA-ICOS score improved the prediction performance of CTRCD. Myosteatosis, defined as a CT-derived skeletal muscle density below the lowest gender-specific quartile, addresses the limitations of using a single BMI threshold of >30 kg/m² for predicting CTRCD, which varies across populations and genders. It involves fat deposition in muscle tissue, contributing to chronic inflammation and adverse cardiovascular outcomes, independent of BMI and SM and VT [11, 23].

The finding that baseline CT-derived myosteatosis was independently associated with all-cause mortality in DLBCL patients receiving anthracyclines highlighted the unique prognostic value of myosteatosis in this population; this dissociation suggests that myosteatosis captures systemic pathophysiological processes, such as altered muscle metabolism, chronic inflammation, or impaired tissue resilience that influence long-term survival in anthracyclines-treated DLBCL [11, 24, 25]. Whereas the HFA-ICOS score serves as a risk assessment tool for anthracyclines-induced cardiotoxicity, the relationship between this score and all-cause mortality necessitates additional investigation to validate.

Several limitations should be noted. First, this retrospective study exclusively focused on patients with DLBCL who were treated with anthracyclines. It remains to be investigated whether CT-myosteatosis enhances baseline risk stratification for cardiotoxicity in other patient populations receiving anthracyclines, such as those with breast cancer, indicating a need for further research. Second, the mechanisms underlying the observed associations (e.g., molecular links between myosteatosis and cardiac metabolism) remain speculative and warrant experimental investigation. Third, the Dice coefficients for our IMAT and VAT were 0.82 and 0.84, respectively. A plausible explanation for the observed segmentation performance was that the low content and irregular morphology of IMAT in the abdominal wall might contribute to diminished segmentation accuracy. Additionally, the complex distribution of intestines within the abdominal cavity resulted in highly irregular boundaries for VAT, which similarly reduced the segmentation accuracy of VAT. To address these limitations, we intend to optimize our segmentation scheme using artificial intelligence methodologies in future studies. Fourthly, survival status of some participants was confirmed only through phone follow-ups, providing only death occurrence and timing, not causes. Thus, the Cox proportional hazards model was used for OS analysis instead of a competing risks model. Future studies will incorporate a competing risks framework for more accurate survival analysis. Finally, the timing of echocardiographic monitoring for CTRCD was somewhat subjective and finding the optimal schedule was challenging. We will aim to establish a more precise timing of echocardiographic monitoring for CTRCD in future studies.

In conclusion, CT-derived myosteatosis was independent predictors of CTRCD and all-cause mortality in DLBCL patients treated with anthracyclines and improved baseline risk stratification of CTRCD. These findings support the integration of CT- derived myosteatosis assessments into routine onco-cardiology care to optimize baseline risk stratification and personalized management of DLBCL patients receiving anthracyclines.