Immune checkpoint inhibitors (ICI) have revolutionized the clinical survival of patients with advanced cancers. Since the introduction of ICI in melanoma in 2011, the 5-year survival rates could approach 35% to 40% in metastatic melanoma with an average life expectancy ranging from six to twelve months before (1). For non-small cell lung cancer (NSCLC), accounting for 80-90% of primary lung cancer (2), the approval of ICI in 2015 increased the 5-year survival rate from less than 10% to more than 30% (3). Currently, immunotherapy, particularly ICI, is an attractive and viable treatment option for various cancer types (4). ICI achieves clinical success by inhibiting the immune checkpoints between differentiation 8 (CD8) T lymphocytes and tumorigenic cells, with the programmed cell death protein 1 (PD-1), the programmed death-ligand 1 (PD-L1) and CTLA-4 being the most well-studied checkpoints (5). Despite its success, treatment response to ICI varies greatly among patients, with some experiencing adverse events and minimal benefits. For instance, only 4% of patients with NSCLC showing remission, defined as responding to ICI, were still alive after 4.5 years. In addition, it is universally acknowledged that immunotherapy is costly and can lead to numerous adverse events, such as colitis, diarrhea and polyarthritis (6). Therefore, identifying patients who are likely to benefit from ICI treatment is of great significance (7).

Several biomarkers have been identified to predict the response of ICI, while accurately predicting clinical outcomes remains a challenge. Biomarkers related to ICI treatment are often collected from tissue samples and include inflammatory cytokines, tumor-infiltrating lymphocytes, mutation variants, and levels of PD-L1 or CTLA-4 (8). Despite their association with a favorable response for the above biomarkers, their predictive power and feasibility are still uncertain. For example, the adoption of the percentage of tumor cells expressing PD-L1, a commonly used biomarker based on the mechanism of ICI (9), remained controversial regarding its reliability and disability of reflecting dynamic PD-L1 expression (10, 11). A large meta-analysis also revealed that PD-L1 expression status alone is insufficient for determining which patients should receive PD-1 or PD-L1 blockade therapy (12). Current predictive factors do not fully meet the needs of clinical prognosis prediction for ICI.

Body composition, measured by imaging method, was reported to be significantly associated with the clinical benefit of ICI (13). Visceral adipose tissue, subcutaneous adipose tissue, intra-muscular adipose tissue and total body fat tissue, are widely studied body composition. Adipose tissue is implicated in tumorigenesis and progression, while the “obesity paradox” suggests that obese tumor patients have better survival outcomes during treatment (14). Adipose tissue in visceral and subcutaneous have different origins, which may determine the functional heterogeneity in tumors (15). Recent studies have investigated the association between adipose composition and immunotherapy efficacy. Takenaka et al. reported lower visceral fat was significantly associated with poor disease control in head and neck squamous cell carcinoma (HNSCC) (16). Sabel et al. found higher visceral fat distance predicted poor survival of patients with melanoma (17). The results in different cancers and the distinctive biomarkers applied to even the same tissue vary. Thus, there is a need to identify specific adipose biomarkers to predict clinical outcomes in specific cancers.

In this systematic review, we aimed to summarize the ability of different imaging-based adipose biomarkers to predict the clinical outcomes after ICI treatment. The patients with any malignancy treated with ICI were the targeted population. Investigated predictors are adipose biomarkers and models deriving from imaging. Investigated clinical outcomes include therapy response, progression-free survival (PFS), overall survival (OS) and tumor remission. The quality of image mining cohort studies was assessed by Newcastle-Ottawa Quality Assessment Scale (NOS) and radiomics quality scoring (RQS) tools.

The systematic search workflow is shown in Figure 1 . Our initial search yielded 1410 potentially relevant records from the MEDLINE and EMBASE databases. After removing 302 duplicates, 1108 articles were screened by abstract and title, and only 71 articles were retained for full-text screening. Ultimately, 22 studies were included for analysis, with 49 studies excluded (21–31).

Owing to the encouraging therapeutic effect on cancers, ICI has become an emerging treatment for cancers (13). Considering the variability of clinical outcomes of ICI therapy, it is urgent to identify patients that will get clinical benefits from it (44). Adipose tissue has been linked to prognosis in patients treated with immunotherapy, mainly including total body fat, visceral fat, subcutaneous fat and inter-muscular fat (6). However, the predictive value of these adipose biomarkers varies by cancer type. In this review, we summarize studies regarding the role of body fat composition in immunotherapy and its potential mechanisms.

Our systematic review identified seven cancers with clinical outcomes associated with imaging-based adipose biomarkers. A higher amount of total fat was correlated with favorable outcomes in RCC and liver cancer, while a higher amount of total fat was correlated with worse outcomes in NSCLC. More visceral adipose tissue predicted better clinical outcomes in NSCLC, UC, melanoma and liver cancer, whereas higher subcutaneous adipose accumulation was related to clinical benefits in NSCLC, RCC, UC and liver cancer. More inter-muscular adipose tissue was only associated with poorer survival benefits in UC. These varied results may be attributed to differences in structure, function and cytokine release of the different types of adipose tissue (45).

This systematic review provides a comprehensive overview of the imaging-based adipose biomarkers that can predict the clinical outcome of ICI therapy, offering insights into the use of radiometric adipose biomarkers. For lung cancer, FBM, TFI, VFM, SCFM and weight loss could be used as predictors of clinical outcome. For RCC, TFI, SAT% and body composition risk score integrating IFI, SM and SFI were the significant predictors of clinical outcome. For UC, the adoption of VFI, SFI and IFI to predict clinical outcomes may be effective. For melanoma, visceral fat distance manifested as a significant predictor. For HNSCC, VFI and SFI were selectable predictors. For gastric cancer, SFI was an optional predictor. For liver cancer, the significant predictors that could be adopted were VFI, SFI and VSR. When combining various cancer types, Δt-TFI, Δt-VFI, Δt-SFI, SFI and the square root of VSR could serve as potential predictors.

CT-based body composition is a promising predictive marker for cancer immunotherapy, superior to the traditional body indicator body mass index (BMI). BMI is a simple method of measuring body weight, typically calculated by dividing weight (in kilograms) by height (in meters) squared, and higher BMI was reported to be significantly associated with prognosis in various cancer studies (46). Nevertheless, BMI is limited by race specificity and individual heterogeneity (47),and cannot differentiate between different types of tissues like muscle and fat, or reflect body fat distribution (45). For example, individuals with a high BMI but low body fat may belong to the category of muscular obesity, while a person with a normal BMI but excessive visceral fat may have metabolic syndrome and related disease risks. Therefore, in clinical practice, BMI and body fat composition, among other indicators, should be considered in combination according to the specific circumstances to assess a patient’s health status and disease risks.

Body fat can impact immunotherapy through various mechanisms. As an endocrine organ, adipose tissue secretes adipokines such as leptin and adiponectin, which can contribute to tumor formation, invasion, angiogenesis, and immune evasion (48). Additionally, fat serves as a reservoir for immune cells, and mobilizing immune cells in fat can enhance anti-tumor activity (49). Higher fat levels may also provide patients with better nutrition and social status, potentially protecting them from cachexia (50). Moreover, tumor cells can also use fat to form abnormal tumor metabolic pathways to provide energy or evade immune surveillance (51). Tumor cells can also utilize fat to create abnormal metabolic pathways that supply energy and evade immune surveillance. We have reviewed relevant research on adipose tissue and identified several mechanisms by which it can influence the clinical outcome of ICI therapy.

Initially, adipose tissue was related to the pharmacokinetics, bioavailability and accumulation of the ICI. In vivo PET imaging found that the bio-distribution of ⁹Zr-nivolumab in adipose tissue was low, with uptake unaffected by the addition of 1 or 3 mg/kg. Patients with high FBM receive a relatively high dose in non-fat tissue, such as tumor targets. Bensch et al. also found that 1.56% to 18.95% of the injected dose accumulated in fat tissue (52). The “storage effect” of the adipose tissue can influence the overall distribution of therapeutic antibodies.

Additionally, obesity-related inflammation and pro-inflammatory cytokines can also have an impact on the immune response to ICI therapy. Chronic inflammation, characterized by elevated circulating levels of IL-1, IL-6 and TNF, is associated with both obesity and cancer (53). On the one hand, adipose tissue can have an adverse effect on immunotherapy, and the relationship between IL-1β and myeloid-derived suppressor cells (MDSCs) is of particular interest. MDSCs are a heterogeneous group of immature myeloid cells that promote immunosuppression and angiogenesis in tumors (54) and augment tumor growth through exacerbating inflammatory conditions. The recruitment of activated Treg lymphocytes and the function of CD4⁺ and CD8⁺ suppressed by interleukin-10, may contribute to the immunosuppressive effect of MDSCs (55). It has been reported that IL-1β, elevated by adipose tissue, induced the activation of MDSCs infiltrating tumor, and thus attenuate the therapeutic effect of ICI (23). Inhibition of IL-1β has been shown to improve tumor immunity (56). On the other hand, when cachexia is involved, characterized by adipose tissue loss, the effect of reduced adipose tissue loss may be obscured. It is reported that cancer cachexia is associated with reduced PD-1/PD-L1 inhibitor efficacy in NSCLC patients (57, 58). The failure to reduce IL-1β and cachexia-related mediators like TNF-α and IL-6 by reduced fat mass may contribute to this (59, 60).

Moreover, the close link between sone specific adipokines and cancer immunity could also serve as a potential explanation. Adipokines refer to the cytokines released from adipose tissue, including leptin, adiponectin and resistin. Leptin, the most well-studied adipokine in ICI therapy, increases the expression of PD-L1 on adipocytes during adipogenesis and promotes the immune escape of cancer (61). Nevertheless, high expression of PD-L1 in adipocytes, especially in visceral adipose tissue, also enhances the effectiveness of anti-PD-1/PD-L1 therapy. Tumor-infiltrating lymphocytes are also found to increase markers of exhaustion in diet-induced obese (DIO) mice, while progression rates in DIO mice are similar to control mice after anti-PD-L1 therapy (62). Adiponectin, another well-known adipocyte-secreted cytokine, is proven to negatively regulate the anti-tumor activity in CD8+ T cells through the activation of SATA3 (63). Resistin, a member of the resistin-like molecule family, was mostly found to influence the tumor itself. It is reported that resistin can elicit the expression of adhesion molecules including intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule (VCAM-1) to facilitate the cell invasion and metastasis of tumors (64). Besides, resistin also plays an important role in apoptosis-resistance, cancer stemness, angiogenesis and therapeutic resistance (65). Therefore, it is a double-edged sword that obesity promotes oncogenesis but with a robust host immune response to ICI.

There are also some limitations in this systematic review. Firstly, all studies included in this systematic review were retrospective. Although the studies reported have been fully included and discussed, the selection and heterogeneity of retrospective studies can’t be resolved. Hence, more prospective cohorts are needed to validate the association between fat and tumor immunotherapy. Secondly, there is a lack of universally agreed-upon tools to assess the risk of bias in included studies. Although the combination of NOS and RQS was used to assess the quality of the radiomics retrospective studies, it may not be sufficient for all types of studies. NOS was intended to detect the possible risk of bias in retrospective studies, while RQS mainly focuses on the guidance of good radiomics studies. Thirdly, the heterogeneity of adipose biomarkers used in the studies included in this review precluded the possibility of a quantitative meta-analysis, which limits the generalizability of the results. This limitation can commonly be seen in other reviews (66, 67). Meanwhile, the absence of confounding analysis, may potentially introduce a confounding effect that requires careful consideration. Finally, since the limited number of patients using ICI, a relatively emerging therapy of cancer, the population size in the studies included is relatively small. Besides, in some cancers like gastric cancer and head and neck squamous cell carcinoma, there are only a limited number of studies included, which means we should be more prudent when drawing conclusions regarding them. Therefore, more studies involving a large number of subjects are an unmet need in the future.

In summary, there is relatively well-supported evidence for imaging-based adipose biomarkers to predict the clinical outcome of ICI, despite the diversity of geographical populations and treatment protocols in each study. Generally, adipose tissue manifests a close correlation with clinical outcomes, with most of the studies included in our study showing positive outcome. Considering the diversity of imaging biomarkers and their corresponding outcomes, even in the same category, we summarize the significant predictors that have been proven by researches. Future studies should focus on the validation of these biomarkers in sufficiently large independent prospective cohorts.

The original contributions presented in the study are included in the article/ Supplementary Material . Further inquiries can be directed to the corresponding authors.

Conceptualization, XP, YX, YL, KN and HT. Data acquisition, XP and YX. Methodology, YL, LH and XY. Validation, LZ, MZ and XZ. Writing—original draft preparation, XP, YX. Writing—review and editing, XP, YL and KN. Visualization, KN and XP. Supervision, KN and HT. Project administration, XP, KN and HT. All authors contributed to the article and approved the submitted version.