Introduction

The success of immune checkpoint blockade (ICB) for patients with unresectable or metastatic melanoma has promoted the use of these agents also as adjuvant therapies for melanoma. Several clinical trials have successfully demonstrated the efficacy of PD1-based ICB, leading to the approval of nivolumab and pembrolizumab for the adjuvant treatment of melanoma, and the establishment of these therapies as the current standard of care (Eggermont et al. 2022; Larkin et al. 2023). The introduction of adjuvant ICB in increasingly early stages of disease has drastically increased the number of patients eligible for therapy (Luke et al. 2022; Kirkwood et al. 2023). Moreover, early-stage patients harbor a smaller individual risk for disease recurrence or progression, while the risk for irreversible adverse events (AEs) remains comparable to patients with advanced disease (Robert et al. 2019; Larkin et al. 2019). This inherently unfavorable risk-reward ratio of adjuvant therapy highlights the clinical need to identify patients who benefit most from ICB. However, despite this clinical need, biomarkers established in routine patient care are currently lacking.

Alterations of the physical constitution including syndromes such as cachexia have been described as common events in patients with solid cancers (Argilés et al. 2023). Furthermore, connections between body composition and response to modern treatments have been described in a pooled meta-analysis of melanoma patients (McQuade et al. 2018). In this work, adipose patients harboring a body mass index (BMI)  > 25 showed an improved survival under treatment with ICB. This effect was most pronounced in male patients with mild obesity (Naik et al. 2019). Interestingly, a high serum creatinine serving as a surrogate parameter for skeletal muscle mass, was also associated with longer survival (Naik et al. 2019). These results indicate a profound impact of body composition on treatment responses in patients with metastatic disease. How these results translate to the adjuvant treatment of melanoma is currently unknown.

The clinical interest in body composition has catalyzed the development of morphometric measures to quantify different tissue compartments as potential biomarkers of treatment response or occurrence of AEs. Due to the availability of CT scans in routine patient care, methods have been developed to facilitate accurate measurements of tissue compartments such as muscle mass and adipose tissue from single CT slices (Zopfs et al. 2020). In early studies, the prognostic value of sarcopenia as the reduction of muscle mass on CT images has been demonstrated in patients with respiratory or gastrointestinal solid tumors (Prado et al. 2008). Furthermore, visceral and subcutaneous adipose tissue abundance have been identified as favorable prognostic factors in melanoma patients with metastatic disease (Lee et al. 2022; Mengoni et al. 2023), further highlighting the use of radiologic measures of body composition as biomarkers. Again, the applicability of these results for patients in an adjuvant setting receiving ICB remains unknown.

In the current work, we therefore assess the association of body composition on outcome of a representative cohort of melanoma patients receiving ICB and analyze the impact of body composition on the occurrence of AEs.

Materials and methods

Study cohort selection

This retrospective study was approved by the institutional ethics committee of the Otto-von-Guericke University Magdeburg (Approval number 145/21, Ethics Committee, Otto-von-Guericke University Magdeburg, Germany). All patients with resected, stage IIB-IV melanoma treated with PD1-based adjuvant immunotherapy as first-line treatment at the Dermatology Department of the University Hospital Magdeburg from the years 2014 until 2023 were retrospectively assessed. All patients included into the present study underwent contrast-enhanced CT scans at baseline diagnosis in a time course of maximum 120 days before treatment initiation. Patients with prior melanoma-related therapies (except adjuvant interferon-alpha), secondary malignancies, no available CT staging at baseline as well as prior or concurrent immunosuppression were excluded from this trial. All patients received treatment until recurrence and/or disease progression according to the Response Evaluation Criteria in Solid Tumor (RECIST 1.1) criteria (Eisenhauer et al. 2009) or occurrence of inacceptable toxicity. All patients, including those who discontinued therapy, were included in the analysis to prevent attrition bias.

Image acquisition

All CT images were acquired pretherapeutically on a multidetector CT scanner (Siemens Somatom Definition AS+, Siemens Healthineers, Germany, or Canon Aquilion Prime, Canon Medical Systems Corporation, Japan). Imaging was performed in supine position of patients with a standardized CT protocol (acquisition slice thickness of 1 mm with 5 mm reconstructions, tube voltage 120 kV with automatic tube current modulation, pitch factor 1.2, collimation 0.6 mm, application of 90 mL i.v. contrast medium (300 ml/mg Accupaque, GE Healthcare, USA)). All CT scans analyzed in this study were acquired in the portal venous phase.

Segmentation and radiologic measurements of body composition parameters

All images were analyzed by trained radiologists who were blinded to the clinical course of the patients. For analysis, images were viewed in the soft tissue window. Total, subcutaneous, intermuscular and visceral adipose tissue as well as skeletal muscle cross-sectional areas were assessed semiautomatically on the height of the mid third lumbar vertebra (L3) using standard Hounsfield unit ranges (adipose tissue: − 190 to − 30 HU: skeletal muscle:− 29 to + 150 HU) (Richards et al. 2012). Tissue area and density were measured using ImageJ (version 1.48). Representative images describing the segmentation procedure are shown in Supplementary Fig.1. The skeletal muscle index (SMI) was calculated as the ratio of smooth muscle area (SMA) and the body height squared (muscle area/height2). In our current work, we assessed low skeletal muscle mass (LSMM) CT-based with the SMI threshold proposed by Prado et al.: < 52.4 cm2/m2 for male patients and < 38.5 cm2/m2 for female patients (Prado et al. 2008). The visceral to subcutaneous fat ratio (VSR) was obtained by dividing the visceral adipose tissue area with the subcutaneous adipose tissue area. For the VSR, a threshold of 1.1 was used. Adipose tissue gauge indices were calculated as described previously by our group (tissue area * tissue density/patient height) (Mengoni et al. 2023).

Clinical data collection

Demographic and treatment data were compiled from electronic medical records. The following parameters were examined and included in the analysis: age, sex, body weight, height, tumor stage according to the American Joint Committee on Cancer Staging system (8th edition) (Gershenwald et al. 2017), laboratory parameters (lactate dehydrogenase (LDH), S100) and therapeutic agents used. As cutoff-values for S100 and LDH, the upper reference levels supplied by the assay manufacturer (0.11 µg/l for S100 and 3.75 µmol/s*l for LDH) were used. Additionally, dates of first dosing, disease recurrence and/or progression, death, cause of death and last follow-up were assembled to calculate the recurrence-free survival (RFS) and melanoma-specific survival (MSS) as primary outcome parameters of this study.

Statistical analysis

Kaplan–Meier estimators were utilized for univariate analysis, and statistical significance was tested using a logrank-test. Furthermore, multivariate Cox regression models were fitted and adjusted for age. Analyses were conducted for both MSS and RFS. Missing values were omitted from analyses. Statistical significance of differences for adverse events were tested using Fisher’s exact test. Analyses were performed using python with standard library modules and the packages lifelines and scipy. p-values < 0.05 were considered statistically significant.

Results

Study cohort characteristics

We enrolled 109 patients with resected, stage IIB-IV melanoma in the current analysis who initiated adjuvant ICB for one year in our department between 2014 and 2023. The cohort consisted of 61 male and 48 female patients with a median age of 63 years. 63 patients received pembrolizumab, 41 patients nivolumab and 5 patients received other PD1-based therapy regiments. Patients were stratified by the presence or absence of low skeletal muscle mass (LSMM) as defined by Prado et al. (Prado et al. 2008). Patients with and without LSMM differed significantly in body weight, whereas all other characteristics were comparable between the two subgroups. The median follow-up of our cohort was 32.7 months, and 31.5 months for patients without disease recurrence. The full cohort characteristics are shown in Table 1.

Table 1 Study cohort characteristics
Full size table

The presence of low skeletal muscle mass is associated with reduced melanoma-specific survival

We first analyzed the impact of LSMM on MSS. In our cohort, patients with LSMM revealed a significantly shorter MSS compared to patients without LSMM (Fig. 1a). Since obesity as measured by a BMI > 25 has been described to be associated with a favorable outcome of patients receiving ICB, we furthermore assessed the BMI in our cohort. While patients with a BMI > 25 showed a trend towards longer MSS, this effect was not statistically significant in a univariate analysis (Fig. 1b). Interestingly, the conventional serological biomarkers S100 and LDH failed to predict MSS in our cohort (Fig. 1c, d). A multivariate cox regression model for MSS substantiated the finding of LSMM as a predictor of worse outcome, revealing a hazard ratio of 6.82 (p = 0.02, Fig. 1e, Table 2). Other body composition measures were not associated with MSS (Supplementary Fig.2). In summary, these data identify LSMM as a novel biomarker for MSS, outperforming conventional biomarkers S100 and LDH in our adjuvant cohort.

Fig. 1
figure 1

Low skeletal muscle mass outperforms conventional serological tumor markers as predictor of poor prognosis. a–d show Kaplan–Meier curves of melanoma-specific survival stratified by low skeletal muscle mass (a), BMI (b), S100 (c) and LDH (d). p-values were calculated by logrank test. e shows a multivariate regression analysis of body composition parameters and MSS. Each square represents the log(HR) for the coefficients, the whiskers denote the 95% confidence interval. BMI body mass index, HR hazard ratio, LDH lactate dehydrogenase, LSMM low skeletal muscle mass, MSS Melanoma-specific survival, SAT subcutaneous adipose tissue, VAT visceral adipose tissue, VSR visceral to subcutaneous fat ratio

Full size image
Table 2 Multivariate cox regression model of the association between melanoma-specific or recurrence-free survival and body composition features
Full size table

The presence of LSMM does not predict recurrence-free survival

We furthermore assessed the relationship between the presence of LSMM and RFS. Whereas the RFS in our cohort was comparable to previously published results for melanoma patients treated with adjuvant ICB (Eggermont et al. 2022; Larkin et al. 2023), patients with LSMM surprisingly did not differ in their RFS compared to patients without LSMM in a univariate analysis of our cohort (Fig. 2a). Similarly, neither BMI nor the serological parameters S100 or LDH significantly predicted RFS in our patient cohort (Fig. 2b–d). In a multivariate cox regression model, no parameter reached statistical significance (Fig. 2e, Table 2). Again, no other body composition parameter was associated with RFS in our cohort (Supplementary Fig. 3).

Fig. 2
figure 2

Low skeletal muscle mass does not predict recurrence-free survival (RFS). a–d show Kaplan–Meier curves of RFS stratified by low skeletal muscle mass (a), BMI (b), S100 (c) and LDH (d). p-values were calculated by logrank test. e shows a multivariate regression analysis of body composition parameters and RFS. Each square represents the log(HR) for the coefficients, the whiskers denote the 95% confidence interval. BMI body mass index, HR hazard ratio, LDH lactate dehydrogenase, LSMM low skeletal muscle mass, RFS recurrence-free survival, SAT subcutaneous adipose tissue, VAT visceral adipose tissue, VSR visceral to subcutaneous fat ratio

Full size image

The presence of LSMM is associated with a lower risk of adverse events

Next, we investigated the association of body composition features with AEs in our cohort. We observed a total of 159 AEs in our cohort. Of these 82 events were of grade 1, 63 events of grade 2, 12 events of grade 3 and 2 events of grade 4. Interestingly, patients with LSMM harbored fewer occurrences of hypothyroidism, fatigue, and xerostomia (Table 3). The incidence of any AE or severe AEs of grade 3 or 4 did not significantly differ between patients with or without LSMM (any grade p = 0.68, grade 3–4 p = 0.25).

Table 3 Adverse events
Full size table

Discussion

In our current study, we describe the presence of LSMM as a risk factor for melanoma patients treated with adjuvant PD1-based ICB. To our knowledge, this is the first study examining the effect of body composition measures on the efficacy and tolerability of adjuvant ICB in melanoma patients of stages IIB-IV. Interestingly, pre-treatment LSMM was associated with a reduced MSS, whereas other parameters of body composition and routine serologic biomarkers (S100, LDH) did not correlate with patient outcome. At the same time, patients with LSMM less frequently experienced adverse events such as hypothyroidism under ICB. Since hypothyroidism and other immune-related adverse events have been described as favorable prognostic factors (Dawidowska et al. 2022), this further substantiates a potential immunosuppressive impact of LSMM. Previous reports have linked the presence of LSMM with poor prognosis also in stage III melanoma without adjuvant therapy (Sabel et al. 2011; Youn et al. 2022), suggesting a broader applicability of LSMM as a risk factor in melanoma. LSMM has also been identified as a predictor of worse outcome and increased frequency of adverse events in other cancer entities including patients with colon cancer receiving adjuvant chemotherapy (Jung et al. 2015). The mechanism how LSMM or other body composition features influence immune responses are currently unclear. Recent meta-analyses have observed sex-specific differences of body composition a predictor of improved outcome in melanoma under ICB, in which obesity served as a favorable indicator only in male, but not female patients (Trinkner et al. 2023). The interdependence between sex and body composition suggests that skeletal muscle and adipose tissue might alter hormonal responses governing anti-tumoral immune responses.

While the presence of LSMM predicted poor MSS in our cohort, we did not observe a difference for RFS in patients with LSMM. Similar discrepancies between PFS and overall survival have been described in meta-analyses of solid cancer clinical trials (Belin et al. 2020; Pasalic et al. 2020). These differences were not attributable to methodological errors, but rather caused by the longer post-progression survival enabled by novel therapies and additional second-line therapy options (Amir et al. 2012; Hess et al. 2019).

A major advantage of LSMM as a biomarker for melanoma is the simplicity of calculation, which is possible from routinely acquired CT images readily available for most patients. Automated systems assessing various body composition measures including LSMM are under active development (Lee et al. 2018; Graffy et al. 2019), further facilitating the applicability of our results into patient care. Moreover, the precision of body composition measurements might be further enhanced by deep-learning based volumetric models (Koitka et al. 2021).

The limited cohort size of our study poses a key limitation of our work. Moreover, the retrospective study design impedes the analysis of AEs since these are frequently underreported in retrospective analyses. Nonetheless, our results on adjuvant patients are consistent with reports from metastatic disease, substantiating the validity of our results. Further prospective trials using larger cohorts are required to fully uncover the usability of LSMM and other body composition parameters as predictors in adjuvant melanoma patients receiving ICB.

Conclusion

We present the first study evaluating body composition features as biomarkers for melanoma patients treated with adjuvant ICB and identify LSMM as a predictive indicator of reduced melanoma-specific survival. LSMM was associated with a lower risk for AEs.