Introduction

Although immune checkpoint inhibitors (ICIs) have ushered in a novel therapeutic era for multiple solid tumors, including nasopharyngeal carcinoma (NPC), only approximately 20% of patients with NPC benefit from single-agent immunotherapy, and the antitumor response is not durable.1 In order to avoid unnecessary treatments, the search for novel and reliable biomarkers to classify patients with NPC who are most likely to benefit from ICIs is needed.

One area of promising interest has emerged from studies on lipoproteins. From an evolutionary standpoint, lipoproteins are more than mere lipid transporters, as they are known to exhibit important immunomodulating functions.2 A growing body of research has revealed that lipid metabolism has a profound impact on cancer immunotherapy by regulating the tumor microenvironment (TME). Crucially, targeting lipid metabolic reprogramming may enhance immunotherapy by overcoming immunosuppression and immune escape.3 Multiple studies have provided evidence that serum lipoproteins, including cholesterol (CHO), high-density lipoprotein-cholesterol (HDL-C), low-density lipoprotein-cholesterol (LDL-C), triglycerides (TG), apolipoprotein A-1 (ApoA1), and apolipoprotein B (Apo-B), can predict the clinical prognosis of several malignancies.4–8 In particular, serum lipid biomarkers have been identified for their predictive role in tumor immunotherapy for advanced solid tumors.9–11 For instance, high plasma CHO levels have been associated with improved outcomes in ICI-treated patients with non-small cell lung cancer (NSCLC), melanoma, renal cell carcinoma, urothelial cancer, and head and neck carcinoma.9 10 Higher pretreatment serum LDL-C has also been significantly correlated with enhanced survival in patients with nivolumab-treated NSCLC.11 Moreover, in patients with intrahepatic cholangiocarcinoma (iCCA) who received anti-programmed cell death protein-1 (PD-1) therapy, elevated levels of TG were indicative of a favorable prognosis.12

HDL, the primary lipoprotein responsible for transporting CHO back to the liver for elimination, is well-known to confer protection against atherosclerosis. Beyond this, HDL may also play an inhibitory role in tumorigenesis and tumor progression via its anti-inflammatory functioning, antioxidant capacity and immunoregulatory abilities.13 The existing data indicates that HDL could modulate CHO content in lipid rafts and the activity of receptors, resulting in the modulation activity of macrophages, B and T cells.14 Interestingly, through binding to the scavenger receptor class B type I (SR-BI), HDL-based drug delivery systems could increase drug sensitivity and reduce drug resistance in cancer therapy.15 Collectively, the above research suggests HDL may serve as a significant modulator in antitumor immune response, which may influence the treatment of ICIs.

One previous study has already reported an association between prolonged survival with increased serum HDL-C levels in patients with NSCLC treated with nivolumab,11 while another study found that patients with cutaneous melanoma with a low pretreatment level of ApoA1 had poorer progression-free survival (PFS) after immunotherapy.16 Furthermore, high pretreatment ApoA1, the primary protein component of HDL, has been linked to better immunotherapy outcomes in patients with iCCA.12 However, at present, there is a dearth of comprehensive studies that have examined lipoproteins’ ability to predict the response to ICIs in order to determine the clinical benefit for patients with NPC.

In this present study, we first assessed the associations between lipid level-related indexes: CHO, HDL-C, LDL-C, TG, ApoA1, and Apo-B, and the clinical efficacy of ICIs in patients with advanced NPC. Our results demonstrate that higher baseline levels of HDL-C and elevated HDL-C after immunotherapy are significant independent predictors of longer PFS. Furthermore, we analyzed the mechanism by which HDL predicts the response to ICIs using mouse models treated with ApoA1 mimetics. We discovered that ApoA1 mimetics can polarize M2-like macrophages toward M1-like macrophages and rejuvenate T-cell cytotoxic function against NPC cells. Notably, our exploration of ApoA1 mimetics as a potential component of more effective ICI combinations opens new avenues for novel clinical treatments in NPC.

ResultsPatient characteristics

To evaluate the impact of lipid-related indices on the efficacy of immunotherapy in NPC, we first enrolled 160 patients with advanced NPC into a training cohort (online supplemental table S1) shows the baseline characteristics of these patients. Among them, 61 (38.1%) patients received camrelizumab, 51 (31.9%) toripalimab, 32 (20.0%) sintilimab, 10 (6.3%) nivolumab, and 6 (3.8%) pembrolizumab. The median PFS for the training cohort was 9.05 months.

These findings were subsequently validated in a cohort of 100 patients. In this cohort, the immunotherapy treatments were camrelizumab (46/46 %), toripalimab (30/30 %), sintilimab (14/14 %), nivolumab (5/5 %), and pembrolizumab (5/5 %). The median PFS was 12.67 months in this cohort. Overall, there was no statistically significant distinction observed between the two cohorts.

Fluctuations in lipid-related indicators following immunotherapy, excluding chemotherapy

First, we collected the serum lipoprotein levels (CHO, HDL-C, LDL-C, TG, ApoA1, and Apo-B) at the baseline of immunotherapy, before each treatment session (during treatment) and at 1, 2, and 3 months post-immunotherapy. The trends in lipid concentrations before, during, and after treatment are shown in figure 1 . In the training cohort, the distribution of patients with partial response (PR), stable disease (SD) and progressive disease (PD) was 65 (40.6%), 54 (33.8%), and 41 (25.6%), respectively. Similarly, in the validation cohort, the distribution was 41 (41%) for PR, 44 (44%) for SD, and 15 (15%) for PD, respectively. However, no significant differences were observed among the groups for any of the lipoproteins (online supplemental figures 3,4). Notably, significant fluctuations were detected in the peripheral blood lipid indexes (CHO, HDL-C, LDL-C, TG, ApoA1, and Apo-B) in patients with NPC post-immunotherapy (at the time of optimal antitumor efficacy) as shown in online supplemental table S2. Significant differences were observed between the pre-immunotherapy and post-immunotherapy levels of lipid-related indicators in both the training and validation cohorts (p<0.05).

Figure 1

Fluctuation Trends in lipid concentrations before, during, and after immunotherapy treatment in the training and validation cohort (A) CHO, (B) HDL-C, (C) LDL-C, (D) TG, (E) ApoA1, and (F) Apo-B concentrations detected in individual patients with NPC undergoing immunotherapy treatment in the training cohort. (G) CHO, (H) HDL-C, (I) LDL-C, (J) TG, (K) ApoA1, and (L) Apo-B concentrations detected in individual patients with NPC undergoing immunotherapy treatment, presented in a summarized format in the validation cohort. ApoA1, apolipoprotein A-1; Apo-B, apolipoprotein B; CHO, cholesterol; HDL-C, high-density lipoprotein-cholesterol; LDL-C, low-density lipoprotein-cholesterol; NPC, nasopharyngeal carcinoma; PD, progressive disease; PR, partial response; SD, stable disease; TG, triglycerides.

The mean levels of CHO, HDL-C, LDL-C, TG, ApoA1, and Apo-B in the training cohort were 4.96 mmol/L, 1.22 mmol/L, 3.25 mmol/L, 1.25 mmol/L, 1.27 g/L, and 1.00 g/L pre-immunotherapy, and 5.21 mmol/L, 1.13 mmol/L, 3.46 mmol/L, 1.44 mmol/L, 1.34 g/L, and 1.15 g/L, post-immunotherapy, respectively. In the validation cohort, when compared with the baseline level, the mean levels were 4.34 versus 4.58 mmol/L, 1.44 versus 1.29 mmol/L, 2.38 versus 2.64 mmol/L, 1.49 versus 1.75 mmol/L, 1.42 versus 1.30 g/L, and 0.92 versus 0.98 g/L, post-immunotherapy, respectively.

In the chemotherapy cohort, there were no marked differences between the lipid-related indicators pre-treatment and post-treatment in all patients with NPC treated with chemotherapy (online supplemental table S3). Compared with baseline serum lipid indexes, CHO (p=0.7772), HDL-C (p=0.4890), LDL-C (p=0.9062), TG (p=0.7511), ApoA1 (p=0.0531), and Apo-B (p=0.5702) in patients with NPC post-chemotherapy showed no significant difference from the pre-chemotherapy levels.

Evaluation of lipid-related indicators pre-immunotherapy using X-tile

The optimal cut-off values for baseline serum lipid indexes for predicting PFS were established using X-tile, based on data from the training cohort. The determined cut-off values for CHO, HDL-C, LDL-C, TG, ApoA1, and Apo-B were 4.7 mmol/L, 1.2 mmol/L, 2.9 mmol/L, 1.4 mmol/L, 1.3 g/L, and 1.2 g/L, respectively (figure 2, online supplemental figure 5). These values were then applied to the validation cohort for further evaluation.

Figure 2

Survival curves based on baseline HDL-C and ApoA1 levels and their fluctuations during treatment. (A–F) X-tile analysis of the total risk score and survival, based on the optimal cut-off levels of baseline HDL-C, and ApoA1. (G) A significant difference in PFS is observed between patients with higher versus lower baseline HDL-C levels (p<0.0001) in the training cohort. (H) A significant difference in PFS is observed between patients with higher baseline ApoA1 compared with those with lower baseline ApoA1 (p=0.0006) in the training cohort. (I) A significant difference in PFS is noted between patients experiencing an elevation in HDL-C and those with a reduction in HDL-C post-immunotherapy relative to baseline (p=0.0011) in the training cohort. (J) A significant difference in PFS between patients with an elevation in ApoA1 and those with a reduction in ApoA1 post-immunotherapy relative to baseline (p=0.0098) in the training cohort. (K) A significant difference in PFS is seen between patients with higher baseline HDL-C compared with those with lower baseline HDL-C (p=0.0120) in the validation cohort. (L) A significant difference in PFS is observed between patients with higher versus lower baseline ApoA1 levels (p<0.0001) in the validation cohort. (M) A significant difference in PFS is observed between patients experiencing an elevation versus a reduction in HDL-C levels post-immunotherapy relative to baseline (p=0.0002) in the validation cohort. (N) A significant difference in PFS is observed between patients with an elevation versus a reduction in ApoA1 levels post-immunotherapy relative to baseline (p=0.0107) in the validation cohort. ApoA1, apolipoprotein A-1; HDL-C, high-density lipoprotein-cholesterol; PFS, progression-free survival.

Correlation between lipid-related indicators and PFS in patients with NPC

According to the cut-offs of the training cohort established by X-tile, the levels of CHO, HDL-C, LDL-C, TG, ApoA1, and Apo-B, pre-immunotherapy were classified into “high” or “low” groups (figure 2, online supplemental figure 5). The analysis revealed a significant correlation between certain lipid-related indicators and PFS in patients with NPC (online supplemental table S4). Specifically, in the training cohort, patients with high baseline levels of HDL-C and ApoA1 demonstrated better PFS (figure 2G–H). Elevated HDL-C levels were associated with a median PFS of 13.33 months, compared with 7.13 months for those with lower levels (figure 2I). Similarly, patients with increasing ApoA1 levels post-immunotherapy showed improved PFS (11.33 vs 8.47 months, p=0.0098, figure 2J). No significant correlations were observed between baseline levels or dynamic changes in CHO, LDL-C, TG, and Apo-B and PFS in the training cohort (online supplemental figure 6A-H).

When applying the training cohort’s cut-off values to the validation cohort, high baseline levels of HDL-C (p=0.0120, figure 2K) and ApoA1 (p<0.0001, figure 2L), as well as elevated levels of HDL-C (p=0.0002, figure 2M) and ApoA1 (p=0.0107, figure 2N) post-immunotherapy, were associated with superior PFS. However, the cut-off values for CHO, LDL-C, TG, and Apo-B as determined by X-tile did not yield statistically significant results for PFS. Additionally, dynamic changes in CHO, LDL-C, TG, and Apo-B did not significantly impact PFS in the validation cohort (online supplemental figure 6I-P).

Furthermore, an analysis was conducted on another cohort of patients with advanced NPC who were treated with chemotherapy, but who did not receive immunotherapy, to investigate the association between lipid-related indicators and survival. Using the cut-off values from X-tile, the levels of lipid-related indicators before chemotherapy were categorized into “high” or “low” groups. However, the appropriate cut-off values for CHO, HDL-C, LDL-C, TG, ApoA1, and Apo-B for survival, did not show statistical significance in predicting survival outcomes (data not shown).

Univariate and multivariate analyses

Online supplemental Tables S5 and S6 summarize the results from the univariate and multivariate analyses of all the prognostic factors examined in this study. Univariate Cox regression analysis showed that the relevant risk factors for PFS were baseline HDL-C and baseline ApoA1 for all patients. The immunotherapy-related elevation of HDL-C and ApoA1 exhibited superior PFS in both cohorts (all p<0.05). In the multivariate analyses, high baseline HDL-C was discovered to be an independent indicator for longer PFS in the training cohort and the validation cohort. Moreover, immunotherapy-related HDL-C elevation independently predicted better PFS (all p<0.05).

L-4F reduced M2-like macrophages and enhanced M1-like macrophages leading to the alleviation of immunosuppression in the TME

To elucidate the mechanism by which HDL enhances immunotherapy, we investigated whether HDL can positively influence the TME in NPC. It is important to note that the unavailability of murine NPC cell lines, the absence of spontaneous mouse models, and the inability to replicate EBV infection in mice have been major obstacles to advancing translational research in the immunology and immunotherapy of NPC.18 Consequently, we used severe combined immunodeficiency (SCID) mice, which are characterized by low immunoglobulin levels, the absence of mature B and T cells, yet have normal natural killer (NK) cells and macrophages for our ApoA1 mimetic studies.19 These mice were subcutaneously injected with C666-1 (WHO type III), HK-1 (WHO type I), or CNE2 cells (WHO type II) into the flank. The mice treated with the ApoA1 mimetic L-4F (10 mg/kg) showed significant tumor inhibition compared with those treated with SC-4F (figure 3A–H, online supplemental figure 7A-D). Flow cytometry analysis of C666-1 and HK-1 tumors treated with L-4F or SC-4F did not show significant changes in the frequences of dendritic cells (DCs), NK cells, or tumor-associated macrophages (TAMs) in the tumor, spleen and bone (figure 3I, online supplemental figure 8A-F). The full gating strategies for TAMs, DCs and NK cells are shown in online supplemental figure 7E-F. However, L-4F remarkably reduced the frequency of M2-like TAMs and simultaneously increased the frequency of M1-like TAMs (figure 3J–K, online supplemental figure 8G-J). Since SCID mice are deficient in B and T cells, our data suggest that HDL can induce macrophage polarization in vivo.

Figure 3

Suppression of tumor growth in vivo by apolipoprotein A-1 mimetic L-4F. (A–H) Tumor volumes and weights were measured on the indicated days following different treatments in SCID mice (n= 6). Error bars represent the SD from three independent experiments. (I–K) Flow cytometry analysis of the tumors from SCID mouse showed that L-4F significantly decreased the frequencies of M2-like TAMs (F4/80+CD11B+CD206+) and increased the frequencies of M1-like TAMs (F4/80+CD11B+MHC-II+), without affecting the total number of TAMs (F4/80+CD11B+). (L–S) Tumor volumes and weights were measured on the indicated days following various treatments, including SC-4F, L-4F, PD-1 antibody (Ab), and combined therapy, in immunocompetent C57BL/6 mice (n= 5). Error bars represent the SD from three independent experiments. The experiments were conducted in triplicate, and the data are shown as the mean±SD. Significance levels are indicated as *p<0.05, **p<0.01, ***p<0.001. LLC, Lewis lung carcinoma; MHC, major histocompatibility complex; PD-1, programmed cell death protein-1; SCID, severe combined immunodeficiency; TAMs, tumor associated macrophages.

Next, to ensure that HDL can regulate macrophage polarization even in the presence of competent immune components, we performed subcutaneous tumor transplantation of MC38 and Lewis lung carcinoma (LLC) cells in immunocompetent C57BL/6 mice to better recapitulate the tumor immune microenvironment. Consistent with the results in SCID mice, L-4F remarkably inhibited tumor growth. Based on the reshaping of TAMs in the TME by L-4F, the effects of L-4F on the antitumor activity of the PD-1 antibody were also evaluated. We observed that the combination treatment of L-4F and PD-1 antibody in both MC38 and LLC mouse models could synergize to mediate enhanced antitumor activity with no significant toxicity (Figure 3L-S).

Moreover, we further conducted flow cytometry analysis of MC38 and LLC tumors derived from C57BL/6 mice treated with a control, L-4F, PD-1 antibody or the combined therapy. Consistent with the results in SCID mice, we did not observe significant alterations in the populations of DCs, NK cells, or total TAMs (online supplemental figure 9A-I). However, L-4F significantly decreased the percentages of M2-like TAMs (F4/80+CD11B+CD206+), while it increased the percentages of M1-like TAMs (F4/80+CD11B+MHC-II+) (figure 4A–B, online supplemental figure 9J-M). Importantly, we also tested the CD3+CD8+ T cells within the TME using flow cytometry (figure 4C, online supplemental figure 7G), based on the premise that M1 macrophage activation could potentiate anticancer T-cell immunity.20 We observed that L-4F significantly elevated the percentages of CD3+CD8+ T cells, and the proportion of CD3+CD8+ T cells was remarkably increased in mice treated with L-4F plus PD-1 antibody compared with monotherapy (figure 4D). Altogether, these data suggest that HDL is capable of educating macrophages to inhibit tumor growth.

Figure 4

Polarization of M2-like macrophages towards an M1-like immune-activated phenotype in vivo by apolipoprotein A-1 mimetic L-4F (A–B) Flow cytometry analysis of tumors from immunocompetent C57BL/6 mice revealed that L-4F significantly reduced the number of M2-like TAMs (F4/80+CD11B+CD206+) and increased the number of M1-like TAMs (F4/80+CD11B+MHC-II+) in the mice. (C–D) In tumor-bearing mice, either with or without PD-1 antibody treatment, subcutaneous injections of SC-4F or L-4F were administered. The CD3+CD8+ tumor-infiltrating lymphocytes in MC38 or LLC tumors were analyzed by flow cytometry. (E–F, I–J) The presence of M2-like macrophages (F4/80+CD206+, F4/80+CD163+) in LLC tumors treated with L-4F versus SC-4F was examined by immunofluorescence. Scale bar, 50 µm. All quantitative data are shown as mean±SD. Significance levels are indicated as *p<0.05, **p<0.01, ***p<0.001. (G–H, K–L) The presence of M1-like macrophages (F4/80+MHC-II+, F4/80+iNOS+) in LLC tumors treated with L-4F versus SC-4F was examined by immunofluorescence. Scale bar, 50 µm. All quantitative data are shown as the mean±SD. Significance levels are indicated as *p<0.05, **p<0.01, ***p<0.001. Ab, antibody; LLC, Lewis lung carcinoma; MHC, major histocompatibility complex; PD-1, programmed cell death protein-1; TAMs, tumor associated macrophages.

Additionally, we conducted immunolocalization analyses to assess the infiltration levels of M1-like macrophages (F4/80+MHC-II+, F4/80+iNOS+ or F4/80+CD11c+) and M2-like macrophages (F4/80+CD206+ or F4/80+CD163+) in LLC tumor tissues. Consistently, tumors in the L-4F treatment group exhibited significantly lower levels of M2-like macrophages (figure 4E–F, I–J), and significantly higher levels of M1-like macrophages compared with those in the SC-4F group (figure 4G–H and K–L, (online supplemental figure 9N-O). These data further verify the regulatory role of L-4F in polarizing M2-like macrophages towards an M1 phenotype.

We further collected the peripheral blood mononuclear cells (PBMCs) from SCID mice subjected to various treatments and analyzed them using the Bio-Plex Pro mouse chemokine assay. An ELISA assay confirmed that L-4F treatment increased HDL-C levels in plasma (online supplemental figure 10A). Additionally, to validate that L-4F treatment could elevate the production of pro-inflammatory cytokines and decrease the secretion of anti-inflammatory cytokines in TAMs, we isolated bone marrow-derived macrophages (BMDMs) from C57BL/6 mice. These BMDMs were then incubated with L-4F or SC-4F in the LLC-conditioned medium, which induced TAMs. The supernatant from the TAMs treated with L-4F or SC-4F was collected for evaluation. Furthermore, to ascertain that L-4F can regulate the production of pro-inflammatory and anti-inflammatory cytokines even in the presence of competent immune components, we also collected PBMCs from immunocompetent C57BL/6 mice undergoing L-4F or SC-4F treatment and tested the cytokines. The cytokine assay revealed that M1-related cytokines were significantly elevated, and M2-related cytokines were markedly reduced after L-4F treatment, aligning with the results observed in the SCID mice (figure 5A, online supplemental figure 10B-N). As demonstrated in figure 5B–C and a panel of eight proinflammatory cytokines, including CCL1, CCL2, CCL5, CCL20, CXCL1, CX3CL1, GM-CSF, and interleukin (IL)-6, were significantly elevated. Conversely, IL-10 and IL-16, which are anti-inflammatory cytokines, were distinctly decreased in samples from the L-4F-treated SCID mice, C57BL/6 mice, and TAMs. Therefore, our data indicate that L-4F treatment induces a cytokine shift towards a profile that is more pro-inflammatory.

Figure 5

L-4F Induces M1-like macrophage polarization through activation of p38 and NF-κB pathways. (A) Bio-Plex Pro mouse chemokine assay results for murine BMDMs after 24 hours of exposure to CM from LLC cells and PBMCs from C57BL/6 mice treated with L-4F or SC-4F. (B) Identification of two anti-inflammatory cytokines co-downregulated across the Bio-Plex Pro mouse chemokine assays of PBMCs from SCID mice, C57BL/6 mice, and LLC CM-induced TAMs groups, as analyzed using a Venn diagram. (C) Venn diagram analysis identifying eight proinflammatory cytokines co-upregulated in the Bio-Plex Pro mouse chemokine assays of PBMCs from SCID mice, C57BL/6 mice, and LLC CM-induced TAMs groups. (D) Ex vivo exposure to L-4F significantly increased M1-like macrophages (F4/80+CD11B+MHC-II+) and decreased M2-like macrophages (F4/80+CD11B+CD206+) in IL-4-activated BMDM cells. (E–H) Comparison of tumor volumes in HK-1 and C666-1 tumor-bearing SCID mice treated with L-4F, with or without the presence of PLX3397. (I) Expression levels of MyD88, TRIF, TRAF6, SAPK-JNK, p-SAPK-JNK, ERK1/2, p-ERK1/2, p38, p-p38, NLRP3, NLRC4, NLRP7, NF-κB p65, p-NF-κB p65, Arg1, iNOS and GAPDH in IL-4-activated BMDM cells treated with L-4F. (J) Western blot analysis of p38, p-p38, Arg1, and iNOS in IL-4-activated BMDM cells treated with L-4F, with or without the p38 inhibitor SB203580. (K) Western blot analysis showing NF-κB p65 protein levels in the cytoplasm (Cyto) and nucleus (Nuc) of IL-4-activated BMDM cells after L-4F treatment. (L–M) Immunofluorescence staining for NF-κB p65 (and counterstaining with DAPI) in IL-4-activated RAW264.7 and BMDM cells treated with L-4F in combination with JSH-23. Scale bar, 20 µm. (N) Western blot analysis of NF-κB p65, p-NF-κB p65, Arg1, and iNOS in IL-4-activated BMDM cells treated with L-4F, with or without the NF-κB inhibitor JSH-23. (O) Flow cytometry analysis showing that SB203580 or JSH-23 significantly reduces the L-4F-mediated decrease in M2-like macrophages (F4/80+CD11B+CD206+) and increases M1-like macrophages (F4/80+CD11B+MHC-II+). BMDMs, bone marrow-derived macrophages; CM, conditioned medium; IL, interleukin; LLC, Lewis lung carcinoma; MHC, major histocompatibility complex; NF-κB, nuclear factor-κ; NLR, NOD-like receptors; PBMC, peripheral blood mononuclear cell; SCID, severe combined immunodeficiency; TAMs, tumor associated macrophages.

To investigate the tumor-suppressive role of L-4F-induced macrophages, in vivo experiments were conducted with HK-1 and C666-1 xenografts in SCID mice.21 Notably, the depletion of macrophages following anti-CSF-1R treatment22 resulted in diminished tumor inhibition by L-4F in vivo (figure 5E–H). These findings underscore the critical role of L-4F-induced antitumor macrophages in mediating the suppression of NPC tumors.

In addition to the in vivo study, we further assessed the effect of HDL on macrophage repolarization ex vivo. BMDMs were isolated from C57BL/6 mice and stimulated with IL-4 to induce the differentiation of M0 macrophages into M2 macrophages,23 followed by treatment with L-4F. Flow cytometry analysis revealed that L-4F treatment significantly increased the population of M1-like macrophages (F4/80+CD11B+MHC-II+) while decreasing M2-like macrophages (F4/80+CD11B+CD206+), without changing the population of total macrophages (figure 5D).

M1-like macrophages are induced by L-4F activated p38 and NF-κB

We sought to elucidate the underlying mechanisms by which L-4F resets TAMs. Our investigation focused on the downstream signaling adaptors of toll-like receptors, such as MyD88, TRIF and TRAF6 and MAPK and NF-κB signaling molecules, alongside three members of the NOD-like receptors (NLRs) family: NLRP3, NLRC4, and NLRP7. This focus was based on the following considerations: (1) MyD88, TRIF and TRAF6 regulate proinflammatory cytokines 24; (2) MAPK and NF-κB are activated in M1 macrophages25 26 and are critical regulators of proinflammatory factors27;(3) NLR family members, including NLRP3, NLRC4, and NLRP7, play significant roles in macrophage polarization28–30; and (4) L-4F treatment in our study increased proinflammatory factors and induced macrophage polarization.

Our observations revealed that only MAPK p38 and NF-κB p65 were phosphorylated in M2-BMDM after L-4F treatment (figure 5I), suggesting their involvement in the L-4F-induced macrophage phenotype switch. Previous studies have reported that the MAPK p38 inhibitor, SB203580, can inhibit lipopolysaccharide (LPS)/interferon (IFN)-γ-polarized M1 macrophages.31 Similarly, SB203580 blocked the L-4F-reset macrophage phenotype, as evidenced by the inhibited expression of L-4F-induced reduction of Arg1 and elevation of iNOS (figure 5J). Furthermore, we evaluated NF-κB p65 by examining its nuclear translocation and found that L-4F treatment facilitated the entry of NF-κB p65 into the nucleus of M2-BMDM and M2-Raw264.7 cells, a process obstructed by JSH-23, an NF-κB inhibitor (figure 5K–M). Importantly, L-4F-altered the expressions of Arg1 and iNOS were reversed by JSH-23 (figure 5N).

In addition, flow cytometry analysis demonstrated that both SB203580 and JSH23 could effectively attenuate the L-4F-mediated reduction of M2-like macrophages and elevation of M1-like macrophages (figure 5O). Collectively, these findings indicate that the activation of MAPK p38 and NF-κB p65 is required for L-4F-mediated M1-like macrophage development.

Clinical relevance of HDL-C levels, M1/M2-like macrophage levels and activated CD8+ T-cell infiltration in patients with NPC

Immunolocalization studies were conducted on tumor tissues from 100 patients with NPC who received immunotherapy and were categorized based on high/low HDL-C levels to assess M1/M2-like macrophages. Additionally, an immunohistochemistry assay was performed for CD8+T cells and granzyme B (GZMB) expression levels. We observed that M2-like macrophages (CD68+CD206+ or CD68+CD163+) were less prevalent in patients with high plasma HDL-C levels compared with those with low plasma HDL-C levels (figure 6A–D). Concurrently, the number of M1-like macrophages (CD68+HLA-DR+, CD68+iNOS+ or CD68+CD11c+) and CD8+T cells, along with GZMB expression levels, were remarkably elevated in NPC tissues with high plasma HDL-C levels (figure 6E–L, online supplemental figure 10O-P). Furthermore, correlation analysis indicated a positive association between the response to ICIs and high plasma HDL-C levels with M1-like macrophage and CD8+T cell infiltration in NPC. Conversely, there was a negative association with M2-like macrophage infiltration. These findings indicate that HDL might contribute to immunotherapy sensitivity in NPC through shifting TAM polarization from M2 toward the M1 phenotype, thus revitalizing CD8+T cell antitumor immunity. An overview of HDL’s mechanisms are presented in figure 7.

Figure 6

Clinical relevance of HDL-C levels with M1/M2-like macrophage levels and CD8+T cell infiltration in patients with NPC. (A–B) Immunofluorescence analysis of M2-like macrophages (CD68+CD206+, CD68+CD163+) in patients with NPC, comparing those with high versus low plasma HDL-C levels. Scale bar, 50 µm. (E–F) Immunofluorescence analysis of M1-like macrophages (CD68+HLA-DR+, CD68+iNOS+) in patients with NPC, comparing those with high vs low plasma HDL-C levels. Scale bar, 50 µm. (I–J) IHC (immunohistochemistry) analysis of CD8 and GZMB (granzyme B) expression in tumor tissues from patients with NPC receiving immunotherapy, categorized by HDL-C levels. (C–D, G–H, K–L) Correlation analysis was conducted to assess the relationship between the levels of CD68+CD206+, CD68+CD163+, CD68+HLA-DR+, CD68+iNOS+, CD8, GZMB, and HDL-C, and the response to immunotherapy in patients with NPC. All quantitative data are presented as mean±SD. Significance levels are indicated as *p<0.05, **p<0.01, ***p<0.001. GZMB, granzyme B; HDL-C, high-density lipoprotein-cholesterol; NPC, nasopharyngeal carcinoma.

Figure 7

Graphic summary of study findings. This figure illustrates how HDL mediates the sensitivity of immunotherapy in nasopharyngeal carcinoma (NPC) by repolarizing M2-like macrophages towards an M1 phenotype. This repolarization process promotes the infiltration of CD8+T cells into the tumor microenvironment (TME), thereby alleviating tumor immunosuppression and enhancing the efficacy of immunotherapy. ApoA1, apolipoprotein A-1; GZMB, granzyme B; HDL-C, high-density lipoprotein-cholesterol; ICI, immune checkpoint inhibitor; IL, interleukin; SCID, severe combined immunodeficiency; TAMs, tumor associated macrophages.

Discussion

Despite significant advances in clinical care, only a small proportion of patients with recurrent/metastatic NPC achieve long-lasting benefits from immunotherapy.1 Furthermore, the search for optimal biomarkers to accurately predict responses to ICIs in patients with NPC is ongoing.32 Traditional biomarkers of immunotherapy efficacy, such as tumor mutational burden and programmed death-ligand 1 expression, are challenged by the inherent spatial and temporal heterogeneity of tumor tissues. In contrast, peripheral blood markers offer a promising solution to potentially circumventing these limitations. Our study is the first to explore the association between peripheral blood lipid indexes, which perform a vital role in regulating tumor immunity, and the clinical outcomes of patients with advanced NPC undergoing ICI therapy.

It has been reported that serum lipids are altered following immunotherapy. Karlsson et al observed dynamic changes in serum ApoA1 levels after immunotherapy in patients with melanoma.16 In line with this, we found that ICI therapy significantly altered the levels of lipid parameters, including CHO, HDL-C, LDL-C, TG, ApoA1, and Apo-B. Importantly, this is the first study to document lipid fluctuations following immunotherapy in patients with NPC.

Extensive studies have indicated that lipids possess the ability to alter the status and functionality of immune cells within the TME and influence the response to immunotherapy.33 Karlsson et al recently reported that high ApoA1 and apolipoproteins C-1 serve as predictive biomarkers for improved outcomes among patients with cutaneous melanoma treated with immunotherapy.16 Similarly, another study showed that higher baseline serum HDL-C levels were related to better survival in patients with NSCLC who received immunotherapy.11 Additionally, high levels of pretreatment serum ApoA1 have been linked to a more favorable response to ICIs in advanced intrahepatic cholangiocarcinoma.12 Consistent with the above, our results indicate that higher pretreatment HDL-C levels and ApoA1 levels are correlated with improved PFS in patients with NPC treated with ICIs. Notably, our study uniquely identifies that an increase in HDL-C levels post-immunotherapy serves as an independent predictive factor in patients with NPC undergoing such treatments. This retrospective study is the first to emphasize the importance of HDL-C levels as a promising predictive biomarker for the efficacy of ICIs in NPC.

The connection between HDL-C levels and the therapeutic effect of ICIs in several tumors raises the question of the underlying cause or mechanism of this effect. Research has documented that the TME exerts a biologically significant function in influencing the response to immunotherapy in solid tumors.34 T cells, B cells, M1/M2-like macrophages, myeloid lineage cells, and more in the TME have been associated with increased response or resistance to ICIs.34 Recent lines of evidence also point to the significance of lipid metabolic reprogramming in influencing the TME by regulating the recruitment and function of immune cells.3 Of particular note, is that HDL-C could increase cytotoxic T-cell recruitment, decrease myeloid-derived suppressor cell recruitment and promote M1-like macrophage infiltration into the TME to enhance antitumor immunity.35 Furthermore, the apolipoproteins related to HDL, including apolipoproteins A-1, C-1, and E, can exhibit comprehensive antitumor effects through immunomodulatory actions.36 In mouse models, the activity of ovarian and colorectal cancer cells was inhibited by ApoA1 through binding and suppressing lysophosphatidic acid.37 38 Furthermore, the activation of fibroblast growth factor