Sarcopenia attenuates the efficacy of PEGylated granulocyte colony-stimulating factor in preventing febrile neutropenia

Abstract

Purpose

The aim of this study was to develop a predictive model integrating sarcopenia, PEGylated granulocyte colony-stimulating factor (PEG-G-CSF) administration, and conventional risk factors to predict febrile neutropenia (FN) and grade 4 neutropenia (G4 NP) in patients with diffuse large B-cell lymphoma (DLBCL) treated with R-CHOP.

Methods

A retrospective cohort of 305 patients was analyzed. FN and G4 NP incidence rates after the first cycle were assessed along with predictors of these toxicities using multivariable logistic regression. Nomograms were developed and internally validated for risk prediction.

Results

PEG-G-CSF significantly reduced G4 NP incidence rates (37.3% vs. 53.3%) but had no significant effect on FN rates (20.9% vs. 17.1%). Sarcopenia was strongly associated with higher risks of FN (odds ratio [OR]: 3.568) and G4 NP (OR: 4.306), after adjusting for other clinical variables. Among patients with sarcopenia, the protective effect of PEG-G-CSF was attenuated, with persistently high FN and G4 NP incidence rates. For FN, the nomogram included age, albumin levels, lactate dehydrogenase levels, and sarcopenia. For G4 NP, the nomogram incorporated additional variables: sex and PEG-G-CSF use. Both models demonstrated good predictive accuracy. Sarcopenia and FN were associated with significantly reduced overall survival.

Conclusions

Sarcopenia may be a significant risk factor for FN, G4 NP, and reduced overall survival in patients with DLBCL receiving R-CHOP, and it may potentially diminish the protective effects of PEG-G-CSF. Predictive models for FN and G4 NP incorporating sarcopenia and other clinical factors may improve individualized treatment strategies, although further validation is needed.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00520-025-09614-3.

Introduction

Diffuse large B-cell lymphoma (DLBCL) represents the most prevalent form of non-Hodgkin lymphoma, accounting for approximately 30% of cases [1]. R-CHOP remains the widely used first-line therapy for DLBCL, combining rituximab, cyclophosphamide, doxorubicin, vincristine, and prednisone [2]. While R-CHOP has significantly improved survival outcomes, it is often associated with severe treatment-related hematologic toxicities, such as febrile neutropenia (FN) and grade 4 neutropenia (G4 NP). These toxicities increase the risk of infections, prolonged hospital stays, and mortality, often requiring dose adjustments or delays in chemotherapy, which can compromise treatment efficacy [3–5].

Despite progress in supportive care, including the use of PEGylated granulocyte colony-stimulating factor (PEG-G-CSF), the effective management of FN and G4 NP remains a significant clinical challenge for patients treated with R-CHOP [5, 6]. In addition, the predictors of FN and G4 NP in the era of PEG-G-CSF are not yet fully defined in patients with DLBCL. Traditional risk factors, such as advanced age and low albumin levels, continue to be considered relevant [4, 5, 7, 8]. However, recent studies suggested additional factors that may increase the risk of FN and G4 NP, including a high tumor burden altering the G-CSF receptor signaling pathway, elevated soluble interleukin-2 receptor levels, and malnutrition [9–11].

Sarcopenia, which is characterized by the loss of skeletal muscle mass and function, has emerged as a critical factor affecting clinical outcomes in patients with cancer [12]. In DLBCL, sarcopenia has been associated with adverse clinical outcomes, including reduced overall survival (OS) and treatment tolerability [13, 14]. Several studies assessed the relationship between sarcopenia and hematologic toxicities in DLBCL and found that patients with sarcopenia are at increased risk of severe neutropenia and its related complications [11, 15]. In addition, sarcopenia often coexists with other risk factors, such as hypoalbuminemia and advanced age, further compounding the risk of hematologic toxicities [15–17].

Despite studies on FN and G4 NP prediction in the PEG-G-CSF era and the impact of sarcopenia, data on models integrating both factors remain limited. Therefore, we aimed to develop a predictive model integrating sarcopenia, PEG-G-CSF administration, and conventional risk factors to predict FN and G4 NP more accurately in patients with DLBCL treated with R-CHOP.

Methods

Study design and population

This retrospective study evaluated the incidence and predictors of FN and G4 NP following the first cycle of R-CHOP in patients with DLBCL. Secondary analyses included assessing the incidence of FN and G4 NP across all treatment cycles, investigating the prognostic effects of sarcopenia, PEG-G-CSF use, FN, and G4 NP on OS, and developing predictive nomograms for FN and G4 NP based on key clinical factors. All consecutive patients with a confirmed diagnosis of DLBCL who initiated R-CHOP as first-line therapy between January 2004 and June 2022 at Gyeongsang National University Hospital were included. Patients were excluded if they had incomplete medical records (missing core variables or transferred from other hospitals without comprehensive records), received other treatment regimens, or had prior malignancies.

Data collection and variables

Data for this study were collected from medical records of patients treated at Gyeongsang National University Hospital. Functional and nutritional status were assessed using the Eastern Cooperative Oncology Group performance status (ECOG PS), body mass index, serum albumin levels (hypoalbuminemia defined as < 3.5 g/dL), and skeletal muscle indices (SMIs). The pectoralis muscle index (PM-SMI) and lumbar skeletal muscle index (L3-SMI) were calculated to assess muscle mass, following methods established in previous studies [18, 19]. Briefly, the PM-SMI was based on chest computed tomography imaging at the fourth thoracic vertebra level, and the L3-SMI was derived from abdominal computed tomography imaging at the third lumbar vertebra level. Both indices were expressed as muscle area normalized by height squared (cm²/m²). In our study, sarcopenia was defined as concurrent low values of SMIs, with cutoff thresholds of ≤ 52.4 cm²/m² for men and ≤ 38.5 cm²/m² for women for L3-SMI, and ≤ 4.4 cm²/m² for men and ≤ 3.1 cm²/m² for women for PM-SMI [18, 19]. This dual-index approach was adopted to improve diagnostic accuracy, address limitations of single-site measurements, and reflect recent evidence supporting the combined use of both indices [19, 20].

FN was defined as an oral temperature ≥ 38.3 °C or two consecutive readings ≥ 38.0 °C for at least 1 h, with an absolute neutrophil count < 0.5 × 10⁹/L. G4 NP was defined as an absolute neutrophil count < 0.5 × 10⁹/L. PEG-G-CSF prophylaxis (administered between 24 and 72 h after completing R-CHOP, as per the drug’s instructions), quinolone prophylaxis, and dose reductions at the first cycle of chemotherapy were recorded. Quinolone prophylaxis was defined as the use of quinolones during the first cycle of R-CHOP to prevent infections. In addition to a detailed explanation of quinolone prophylaxis, the Online Resource 1 provides further details on variables not fully described in the main text, including their definitions and measurement methods.

Statistical analysis

Multivariable logistic regression analysis was conducted to determine independent predictors of FN and G4 NP after the first cycle of R-CHOP, using backward elimination for variable selection. Only baseline variables with complete data were included in the analysis to ensure consistency and validity. Variables with p < 0.2 in univariate analyses were included in the multivariable models. Multicollinearity among explanatory variables was assessed using the variance inflation factor via the collin command in Stata. Nomograms predicting the risk of FN and G4 NP after the first cycle of R-CHOP were developed based on significant variables from the multivariable analysis. Calibration was assessed using calibration plots, and the area under the receiver operating characteristic curve (AUC) was used to assess discriminatory ability. OS was evaluated using the Kaplan–Meier method, with differences between groups assessed via the log-rank test. Cox proportional hazards regression was performed to determine factors associated with OS. Statistical significance was set at a two-sided p-value of < 0.05. All analyses were performed using R version 4.4.1 (R Foundation for Statistical Computing, Vienna, Austria) and Stata version 16.1 (StataCorp LLC, College Station, TX, USA).

Results

Patient characteristics

A total of 305 patients were included in this study, with 153 receiving PEG-G-CSF and 152 not receiving it. Significant differences were observed between the two groups (Table 1). Patients in the PEG-G-CSF group were more likely to be aged > 70 years (41.8% vs. 25.0%, p = 0.002), to have Ann Arbor stage III–IV disease (64.1% vs. 52.6%, p = 0.043), and to have high–intermediate to high National Comprehensive Cancer Network International Prognostic Index (NCCN-IPI) scores (66.0% vs. 47.4%, p = 0.001). Quinolone prophylaxis was significantly less frequent in the PEG-G-CSF group (9.2% vs. 27.6%, p < 0.001). Dose reduction at the first cycle occurred more frequently in the PEG-G-CSF group (26.8% vs. 19.7%), although this difference was not statistically significant (p = 0.145). Other variables, including ECOG PS, bone marrow involvement, B symptoms, bulky disease, hypoalbuminemia, lactate dehydrogenase (LDH) levels, sarcopenia, and body mass index, did not differ significantly between the two groups. Regarding treatment-related characteristics, PEG-G-CSF use was not associated with significant differences in the median number of R-CHOP cycles, the mean relative dose intensities of cyclophosphamide and doxorubicin, early treatment discontinuation unrelated to disease progression, or treatment-related mortality.

Table 1.

Baseline and treatment characteristics

Variable	Value	PEG-G-CSF (n = 153)	No PEG-G-CSF (n = 152)	p
Age	> 70 years	64 (41.8%)	38 (25.0%)	0.002
Sex	Male	83 (54.2%)	92 (60.5%)	0.268
ECOG PS	2–3	45 (29.4%)	37 (24.3%)	0.318
Ann Arbor stage	III-IV	98 (64.1%)	80 (52.6%)	0.043
Extranodal disease	Yes	95 (62.1%)	89 (58.6%)	0.528
NCCN-IPI	HI to high	101 (66.0%)	72 (47.4%)	0.001
Bone marrow involvement	Yes	24 (15.7%)	16 (10.5%)	0.182
B-symptoms	Yes	25 (16.3%)	28 (18.4%)	0.632
Bulky disease	Yes	33 (21.6%)	25 (16.4%)	0.254
Cell-of-Origin (n = 212)	GCB	49 (33.1%)	11 (17.2%)	0.018
Dose reduction at the first cycle	Yes	41 (26.8%)	30 (19.7%)	0.145
Hypoalbuminemia	Yes	57 (37.3%)	46 (30.3%)	0.197
LDH	Elevated	99 (64.7%)	87 (57.2%)	0.181
Quinolone prophylaxis	Yes	14 (9.2%)	42 (27.6%)	< 0.001
Body mass index, kg/m2 (mean ± SD)		23.0 ± 3.0	23.5 ± 3.3	0.215
Sarcopenia	Yes	11 (7.2%)	13 (8.6%)	0.658
Total cycles of R-CHOP, median (IQR)		6 (6–6)	6 (4–6.5)	0.684
RDI of cyclophosphamide, mean (SD), %		86.8 (21.6)	91.4 (18.8)	0.059
RDI of doxorubicin, mean (SD), %		86.8 (21.5)	90.4 (19.2)	0.128
Early discontinuation of treatment unrelated to disease progression		27 (17.6%)	28 (18.4%)	0.86
Treatment-related mortality		13 (8.5%)	11 (7.2%)	0.683

PEG-G-CSF PEGylated granulocyte colony-stimulating factor, ECOG PS Eastern Cooperative Oncology Group performance status, NCCN-IPI National Comprehensive Cancer Network International Prognostic Index, LDH lactate dehydrogenase, RDI relative dose intensity

Incidence rates of FN and G4 NP after the first cycle of R-CHOP

The overall incidence rate of FN after the first cycle of R-CHOP was 19% (58/305), and G4 NP occurred in 45.2% (138/305) of patients (Table 2). The incidence rates of FN and G4 NP varied depending on clinical characteristics and interventions. The incidence rate of FN was not significantly different between the PEG-G-CSF and no PEG-G-CSF groups (20.9% vs. 17.1%, p = 0.397). However, PEG-G-CSF significantly reduced the incidence rate of G4 NP (37.3% vs. 53.3%, p = 0.005). These trends were generally consistent in the subgroup analyses (Fig. 1). Quinolone prophylaxis was associated with a higher incidence rate of G4 NP (76.8% vs. 38.2%, p < 0.001) but did not significantly affect the FN incidence rate (14.3% vs. 20.1%, p = 0.318). Among patients not receiving PEG-G-CSF, the FN incidence rate was significantly lower with quinolone prophylaxis (7.1% vs. 20.9%, p = 0.044). By contrast, for those receiving PEG-G-CSF, quinolone prophylaxis was not associated with a significant reduction in the FN incidence rate (35.7% vs. 19.4%, p = 0.153).

Table 2.

The incidence of febrile neutropenia and G4 neutropenia after the first cycle of R-CHOP

Variable	Category	Febrile neutropenia	p	G4 Neutropenia	p
PEG-G-CSF	Yes (n = 153)	32 (20.9%)	0.397	57 (37.3%)	0.005
	No (n = 152)	25 (17.1%)		81 (53.3%)
Quinolone prophylaxis	Yes (n = 56)	8 (14.3%)	0.318	43 (76.8%)	< 0.001
	No (n = 249)	50 (20.1%)		95 (38.2%)
Sarcopenia	Yes (n = 24)	12 (50.0%)	< 0.001	19 (79.2%)	0.001
	No (n = 281)	46 (16.4%)		119 (42.3%)
Total	Overall (n = 305)	58 (19%)		138 (45.2%)
PEG-G-CSF	Quinolone prophylaxis	Febrile neutropenia	p
No	Yes (n = 42)	3 (7.1%)	0.044
	No (n = 110)	23 (20.9%)
Yes	Yes (n = 14)	5 (35.7%)	0.153
	No (n = 139)	27 (19.4%)

PEG-G-CSF PEGylated granulocyte colony-stimulating factor

Fig. 1.

Incidence rates of febrile neutropenia (FN) and grade 4 neutropenia (G4 NP) after the first cycle of R-CHOP, stratified by PEGylated granulocyte colony-stimulating factor (PEG-G-CSF) use. Bars indicate the proportion of patients experiencing FN and G4 NP in each group. Asterisks denote statistically significant differences (p < 0.05), observed only in the subgroup analyses for G4 NP incidence rate

Patients with sarcopenia had significantly higher incidence rates of FN (50% vs. 16.4%, p < 0.001) and G4 NP (79.2% vs. 42.3%, p < 0.001), despite being more likely to undergo dose reduction during the first cycle of treatment (Online Resource 2). They also exhibited significantly worse baseline characteristics, including poorer ECOG PS, higher NCCN-IPI scores, and a higher prevalence of hypoalbuminemia, bone marrow involvement, B symptoms, and bulky disease. Among patients with sarcopenia, receiving PEG-G-CSF did not reduce the incidence rates of FN (54.5% vs. 46.2%, p = 0.682) and G4 NP (63.6% vs. 92.3%, p = 0.374; Fig. 1). When considering SMIs, the incidence rate of FN was markedly higher in the sarcopenia group (both low PM-SMI and L3-SMI; 50%), while the rates in other groups were similar, ranging from 7.7 to 17.3% (Online Resource 3).

Incidence rates of FN and G4 NP across treatment cycles by PEG-G-CSF use

The incidence rates of FN and G4 NP were analyzed across treatment cycles and cumulative episodes in patients who completed six cycles of R-CHOP (n = 223), stratified by the use of PEG-G-CSF (PEG-G-CSF group, n = 120; no PEG-G-CSF group, n = 103; Online Resource 4). For FN, no statistically significant differences were observed between the two groups across all cycles, except for the fourth cycle, or in the overall incidence rate. The patterns remained consistent regardless of cumulative episodes. By contrast, PEG-G-CSF use significantly reduced the G4 NP incidence rate across all cycles and was associated with a higher proportion of patients without G4 NP and fewer patients with multiple episodes.

Multivariable analysis of risk factors for FN and G4 NP after the first cycle of R-CHOP

Multivariable logistic regression was conducted to identify risk factors for FN and G4 NP following the first cycle of R-CHOP (Table 3). No significant multicollinearity was observed among variables included in the multivariable logistic regression models, with all variance inflation factor values < 1.25 (Online Resource 5). In the analysis for FN, sarcopenia was significantly associated with a higher risk (OR: 3.568, 95% CI: 1.397–9.114, p = 0.008). Other significant risk factors included advanced age (> 70 years; OR: 2.209, 95% CI: 1.164–4.191, p = 0.015), hypoalbuminemia (< 3.5 g/dL; OR: 3.041, 95% CI: 1.567–5.902, p = 0.001), and elevated LDH levels (OR: 2.623, 95% CI: 1.176–5.854, p = 0.019).

Table 3.

Multivariable logistic regression for incidence of febrile neutropenia and G4 neutropenia after the first cycle of R-CHOP

Variable	OR	95% CI	p
1. Febrile neutropenia
Age, years (> 70 vs. ≤ 70)	2.209	1.164–4.191	0.015
Albumin, g/dL (< 3.5 vs. ≥ 3.5)	3.041	1.567–5.902	0.001
LDH (elevated vs. normal)	2.623	1.176–5.854	0.019
Sarcopenia (yes vs. no)	3.568	1.397–9.114	0.008
2. G4 neutropenia
Age, years (> 70 vs. ≤ 70)	1.954	1.108–3.445	0.021
Sex (female vs. male)	1.709	1.019–2.865	0.042
Albumin, g/dL (< 3.5 vs. ≥ 3.5)	1.965	1.108–3.485	0.021
LDH (elevated vs. normal)	2.721	1.561–4.740	< 0.001
Sarcopenia (yes vs. no)	4.306	1.424–13.023	0.01
PEG-G-CSF (no vs. yes)	2.888	1.694–4.924	< 0.001

LDH lactate dehydrogenase, PEG-G-CSF PEGylated granulocyte colony-stimulating factor

For G4 NP, advanced age (> 70 years; OR: 1.954, 95% CI: 1.108–3.445, p = 0.021), female sex (OR: 1.709, 95% CI: 1.019–2.865, p = 0.042), hypoalbuminemia (< 3.5 g/dL; OR: 1.965, 95% CI: 1.108–3.485, p = 0.021), and elevated LDH levels (OR: 2.721, 95% CI: 1.561–4.740, p < 0.001) were significant predictors. Sarcopenia was strongly associated with a higher risk of G4 NP (OR: 4.306, 95% CI: 1.424–13.023, p = 0.01). In addition, patients who did not receive PEG-G-CSF had a significantly higher risk of G4 NP than those who did (OR: 2.888, 95% CI: 1.694–4.924, p < 0.001).

Nomogram development and validation predicting FN and G4 NP

Nomograms were developed to predict the risk of FN and G4 NP after the first cycle of R-CHOP (Fig. 2A and B), incorporating significant predictors identified in the multivariable analyses (Table 3). The nomogram for FN included age, albumin level, LDH level, and sarcopenia. The nomogram for G4 NP included age, sex, albumin level, LDH level, sarcopenia, and PEG-G-CSF use. Calibration plots demonstrated good agreement between predicted and observed probabilities for both FN and G4 NP (Online Resource 6A and B). Discrimination was assessed using the AUC. The AUC for FN was 0.777, indicating good discriminatory ability, while that for G4 NP was 0.761, reflecting a similar level of performance (Online Resource 6C and D).

Fig. 2.

Nomograms predicting the risk of FN and G4 NP after the first cycle of R-CHOP. A Nomogram for predicting the probability of FN, incorporating the variables: age, serum albumin level, lactate dehydrogenase (LDH) level, and sarcopenia. B Nomogram for predicting the probability of G4 NP, incorporating the variables: age, sex, serum albumin level, LDH level, sarcopenia, and PEG-G-CSF use. Points assigned to each variable correspond to its relative contribution to the predicted probability. The total points determine the linear predictor, which is converted into the probability of FN or G4 NP using the provided scales

Overall survival

OS was evaluated based on PEG-G-CSF use, sarcopenia, FN, and G4 NP after the first cycle of R-CHOP (Online Resource 7). A total of 143 deaths were observed in the overall cohort. The median OS for patients who received PEG-G-CSF was 72.4 months, compared with 106.0 months for those who did not receive PEG-G-CSF; however, this difference was not statistically significant (p = 0.087). Sarcopenia was associated with markedly reduced survival, with a median OS of 8.3 months in patients with sarcopenia compared with 106.0 months in those without sarcopenia (p < 0.001). Similarly, FN was significantly associated with worse outcomes, with a median OS of 9.3 months in patients with FN compared with 124.3 months in those without FN (p < 0.001). The presence of G4 NP was also associated with poorer survival, with a median OS of 38.8 months in patients with G4 NP compared with 162.6 months in those without G4 NP (p < 0.001).

While there was a borderline association between PEG-G-CSF use and poorer survival, this relationship was not significant in multivariable Cox regression analysis when adjusting for age, Ann Arbor stage, and NCCN-IPI (HR: 0.914, 95% CI: 0.636–1.315, p = 0.629; Online Resource 8). These findings suggest that the observed trend of poorer survival in PEG-G-CSF users may be explained by the higher proportion of patients with worse clinical characteristics, such as advanced stage, higher NCCN-IPI scores, and older age, in this group.

Discussion

This study evaluated the incidence and predictors of FN and G4 NP following the first cycle of R-CHOP in 305 patients with DLBCL, focusing on the roles of sarcopenia and PEG-G-CSF. PEG-G-CSF demonstrated efficacy in significantly reducing G4 NP incidence after the first cycle, although it did not affect FN rates or OS. Sarcopenia emerged as a critical risk factor, with patients with sarcopenia showing markedly higher rates of FN and G4 NP than those without sarcopenia. In addition, sarcopenia was strongly associated with reduced OS and appeared to attenuate the protective effects of PEG-G-CSF, as indicated by the disproportionately high rates of FN and G4 NP in patients with sarcopenia who received PEG-G-CSF. Advanced age, hypoalbuminemia, and elevated LDH levels were also identified as independent predictors of both FN and G4 NP, while female sex was identified as a specific risk factor for G4 NP. Nomograms incorporating these factors showed good predictive accuracy, supporting their potential utility in clinical practice. Furthermore, FN and G4 NP after the first cycle were both associated with significantly worse survival outcomes, emphasizing the need to mitigate these toxicities through focused interventions and risk stratification.

Previous studies have demonstrated a lower incidence of FN in patients with DLBCL receiving PEG-G-CSF [5, 6]. However, in our cohort, the incidence rate of FN in the PEG-G-CSF group did not differ significantly from that in the no PEG-G-CSF group. Despite the significant reduction in severe neutropenia with PEG-G-CSF, its lack of efficacy in reducing FN may be attributed to differences in baseline characteristics between the patient groups and the multifactorial nature of FN development. Patients who received PEG-G-CSF were generally older and had more advanced disease, as reflected by higher rates of advanced-stage and worse prognostic scores. These characteristics likely increased their vulnerability to FN, potentially offsetting the protective effects of PEG-G-CSF [8, 9]. Another contributing factor may be the higher use of quinolone prophylaxis in the no PEG-G-CSF group than in the PEG-G-CSF group. Quinolone prophylaxis has been shown to reduce FN, particularly in patients not receiving PEG-G-CSF [21]. However, its routine use is associated with potential drawbacks, including the risk of antimicrobial resistance, adverse drug reactions, and disruption of gut microbiota [22–24]. The emergence of PEG-G-CSF has provided an effective and safer alternative, reducing reliance on quinolone prophylaxis. In our cohort, the incidence rate of FN was significantly lower in patients who received quinolone prophylaxis but did not receive PEG-G-CSF, suggesting its protective role in the absence of PEG-G-CSF. However, among patients who received PEG-G-CSF, the use of quinolone prophylaxis did not result in a reduction in FN incidence (Table 2). All these findings suggest that PEG-G-CSF may provide sufficient protection against FN, potentially offsetting worse baseline characteristics, such as advanced age and higher disease burden, without requiring additional quinolone prophylaxis.

Sarcopenia exerts profound effects on the immune system and metabolic pathways, which likely contribute to the increased incidence of FN and G4 NP observed in this study. Patients with sarcopenia often experience chronic inflammation, as evidenced by increased levels of proinflammatory cytokines, including interleukin-1, interleukin-6, and tumor necrosis factor-α [25–27]. These inflammatory markers not only disrupt hematopoiesis but also impair the proliferation and function of neutrophils, rendering patients more susceptible to severe neutropenia and subsequent infections [28–30]. Furthermore, sarcopenia, which was closely associated with hypoalbuminemia in our study (Online Resource 2), is frequently accompanied by malnutrition and hypoalbuminemia, both of which compromise the host immune response [31–33]. Malnutrition has been related to reduced marrow reserve and delayed recovery of neutrophil counts following cytotoxic chemotherapy, further predisposing patients to FN and G4 NP [34]. Another potential mechanism is the altered drug metabolism and tolerance in patients with sarcopenia. The loss of lean body mass may alter the distribution and metabolism of chemotherapeutic agents, leading to higher effective doses and increased myelosuppression [35, 36]. This effect is particularly relevant for drugs such as cyclophosphamide and doxorubicin, which have a narrow therapeutic index and are integral components of the R-CHOP regimen. In our cohort, the incidence rates of FN and G4 NP were highest among patients with sarcopenia, even compared with other high-risk groups, such as those with advanced age, hypoalbuminemia, or elevated LDH levels (Fig. 1). However, given the small sample size in this subgroup (24 patients), this finding should be interpreted with caution and regarded as exploratory rather than conclusive. Furthermore, the incidence of G4 NP appeared numerically lower in those who received PEG-G-CSF compared with those who did not (63.6% vs. 92.3%). Further studies with larger sample sizes are needed to validate whether sarcopenia influences the response to PEG-G-CSF.

Additional analysis of baseline characteristics revealed no significant differences between patients with sarcopenia and FN and those with sarcopenia without FN, except for dose reduction at the first cycle and poor ECOG PS (Online Resource 9). The higher frequency of dose reduction in the FN group likely reflects clinicians’ judgment to mitigate risks in frail patients, rather than serving as an independent risk factor. This is further supported by its exclusion as a significant predictor of FN and G4 NP in the multivariable analysis (Table 3). Similarly, addressing poor ECOG PS in patients with sarcopenia and FN is unlikely to be a practical solution because it is closely related to the underlying disease and systemic frailty. Given these limitations, direct management of sarcopenia may represent a more effective strategy to reduce the risk of FN in this high-risk group. Interventions such as targeted nutritional support, physical rehabilitation, and anti-inflammatory therapies could help improve muscle mass, functional capacity, and immune resilience, thereby addressing the underlying causes of increased vulnerability in patients with sarcopenia [25, 26, 37, 38]. The predictive nomogram developed in this study offers a preliminary approach to identifying patients who might benefit from such interventions. Although the nomogram showed reasonable predictive accuracy in this cohort, its applicability may be limited by the study’s retrospective design, small sample size, and lack of external validation, highlighting the need for prospective studies to confirm its clinical utility.

Several limitations of this study should be noted. While the retrospective design, small sample size, and lack of external validation have already been discussed, additional factors warrant consideration. First, as a retrospective study, the absence of randomization introduces selection bias. For instance, clinician-driven decisions regarding quinolone prophylaxis and dose reduction may have affected the observed outcomes. Although the criteria for quinolone prophylaxis were clearly defined in the Methods section, its association with a higher incidence of G4 NP (Table 2) raises the question of whether this reflects the intervention itself or the underlying vulnerability of the patients receiving it. Similarly, dose reduction in the first cycle, which was observed more frequently in patients with FN in the sarcopenic subgroup (Online Resource 9), likely reflects efforts to manage frail patients rather than serving as an independent risk factor. Second, the single institution setting limits the generalizability of these findings, as prescribing patterns and supportive care measures may vary across centers. Third, while sarcopenia was assessed using imaging-based indices, these measures do not account for functional or biochemical aspects of muscle loss, which could affect outcomes independently. Finally, the study primarily focused on baseline characteristics and did not evaluate clinical or nutritional status changes during treatment.

In conclusion, this study suggests that sarcopenia may be an important risk factor for FN, G4 NP, and reduced survival in patients with DLBCL receiving R-CHOP. Sarcopenia also appeared to attenuate the protective effects of PEG-G-CSF, although this observation was based on a small subgroup and requires further validation. While the predictive nomogram incorporating sarcopenia showed promise, its retrospective nature and lack of external validation limit its immediate clinical applicability. Future efforts should focus on prospective studies and interventions targeting sarcopenia, such as nutritional and physical support, to improve treatment tolerability and outcomes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

S.I.G., M.J.P., and G.W.L. contributed to the study conception and design. E.J.J., W.J.L., S.P., M.J.P., and G.W.L. performed data collection and curation. S.I.G., E.J.J., W.J.L., and G.W.L. conducted the formal analysis. S.I.G., S.P., and G.W.L. carried out the investigation. S.I.G., E.J.J., and G.W.L. wrote the first draft of the manuscript, and all authors commented on previous versions. All authors read and approved the final manuscript.

Funding

This study was supported by research grants from Hanmi Pharmaceutical Co., Ltd., and from the Basic Science Research Program through the National Research Foundation of Korea (No. RS-2023–00219399).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval

This retrospective study involving human participants was in accordance with the ethical standards of the institutional and national research committee and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards. Ethical approval for the study was obtained from the Institutional Review Board of Gyeongsang National University Hospital (Approval Number: GNUH 2022–06-021).

Consent to participate

The requirement for informed consent was waived due to the retrospective nature of the study.

Competing interests

The authors declare no competing interests.