Circulating Exosomal PD-L1 at Initial Diagnosis Predicts Outcome and Survival of Patients with Osteosarcoma

Osteosarcoma is the most common bone sarcoma of childhood and adolescence. The use of neoadjuvant chemotherapy has improved the 5-year survival rate of osteosarcoma to approximately 60% to 70% for early-stage patients. However, patients with metastatic disease maintain refractory to chemotherapy and progress rapidly. Unfortunately, these patients have a 5-year survival of only 10%–20%. How to accurately predict at the initial diagnosis and noninvasively monitor the progression of osteosarcoma is a great challenge. Liquid biopsy of blood samples provides an important source of circulating noninvasive tumor biomarkers that can be easily applied in the clinic, which is more convenient and inexpensive than CT scan or PET-CT used for monitoring disease progression. Regretfully, liquid biopsy of osteosarcoma develops slower than that of other cancers (1–4). However, in recent years, more evidence has demonstrated liquid biopsy of sarcomas including osteosarcoma provides important information of monitoring the disease progression. Furthermore, the analysis of circulating biomarkers in peripheral blood, such as circulating tumor cells, circulating tumor DNA, circulating free DNA, and circulating exosomes could provide real-time information on tumor genetics, disease state, and resistance mechanisms (5–9).

Exosomes are secreted as the vesicles which are enclosed by the cell membrane and their size ranges from 30 to 150 nm diameter (10). Various types of cells such as immune cells, fibroblast, and endothelial cells, release exosomes into the extracellular space and microenviroment, involved in the tumor pathogenesis (11–13). Exosomes have been reported to act as regulatory elements through which tumor cells can communicate with and reprogram the immune cells populating in the tumor microenvironment (14–16). More studies have been reported that circulating exosomes have the potential to serve as important biomarkers which can efficiently predict the disease prognosis (1, 3, 14, 17). For most sarcomas including osteosarcoma, fewer exosomal biomarkers are recommended as recognized indicators of the prognosis at the initial diagnosis which predict the outcome and supervise the disease progression during the follow-up for patients with overall survival (OS).

The genetic complexity of osteosarcoma has proved significant, with the majority of patients displaying loss of both p53 and RB, chromothripsis, chromosome shattering kataegis, and localized areas of hypermutation (18, 19). The complicated genetic feature in osteosarcoma, along with the regular interaction between immune cells and bone cells in normal tissue, suggests that osteosarcoma may be an immunogenic tumor and evasion of the immune response may be an important component of its pathogenesis (20, 21). Programmed death ligand 1 (PD-L1; B7-H1; CD274) was first discovered by Freeman and colleagues in 2000, who reported that programmed death-1 (PD-1)/PD-L1 interaction resulted in inhibited T-cell receptor–mediated CD8⁺ lymphocyte activation. PD-L1 on the surface of tumor cells bound its receptor PD-1 on effector T cells, thereby suppressing their activity (22). PD-1/PD-L1 axis and its role in promoting immune escape has resulted in great development in the therapeutic strategy. Therefore, tumors escape immunosurveillance by expressing immune checkpoints of PD-L1. Furthermore, the interaction between PD-L1, expressed on tumor cells and its receptor PD-1, expressed on immune cells, leads to immune cell apoptosis, anergy, and tolerance. However, expressive abundance of PD-L1 in the osteosarcoma specimen is limited and several studies proved this phenomenon (20, 23). Koirala and colleagues reported PD-L1 was not expressed in primary specimens and was only expressed in metastatic tissue (23). Our previous study also demonstrated the positive expression rate of PD-L1 in the primary specimens was relatively lower compared with other malignant tumors (20). Recently, several studies in the literature revealed not only PD-L1 expressing on tumor cells, but circulating exosomal PD-L1 produced the systematic immunosuppression in patients with malignant tumor (2, 24, 25). Notably, with regard to osteosarcoma, immunosuppressive cells, supportive cells in the microenvironment, and osteosarcoma cells contribute to the systematic immunosuppression in patients with OS and all of these cells can release the exosomal PD-L1 into the blood. In the previous study, we have demonstrated the circulating exosomal PD-L1 in patients with osteosarcoma was highly expressed compared with that in healthy donors, although the low positive rate of PD-L1 expression was observed in osteosarcoma specimen. Furthermore, exosomes isolated from serum of patients with OS were strongly immunosuppressive, stimulated the pulmonary metastasis, and carried high abundance of PD-L1 (26). Thus, we proposed detection of circulating exosomal PD-L1 in patients with osteosarcoma at the initial diagnosis had a significant value of predicting the prognosis during follow-up. In the current study, we sought to answer: (i) whether liquid biopsy of circulating exosomal PD-L1 at the initial diagnosis could predict the outcome and survival of patients with osteosarcoma, and (ii) whether the high level of exosomal PD-L1 in the serum was associated with inferior outcomes for patients with a newly diagnosed osteosarcoma.

Sixty-seven patients with newly diagnosed osteosarcoma were prospectively recruited into the current study from our tumor center between October 2016 and January 2018 according to the study criteria. The design of present prospective study was showed in Supplementary Fig. S1. The inclusion criteria for current study were as follows: (i) after preoperative biopsy, patients whose histologic diagnosis was osteosarcoma were recruited into the current prospective study. Before chemotherapy, fasting blood was collected in serum separator vacutainer tubes for exosome isolation. (ii) Patients were newly diagnosed as osteosarcoma without any treatment, including chemotherapy, radiotherapy, immunotherapy, and surgery. (iii) All patients were confirmed as osteosarcoma by the postoperative histology. (iv) Intact clinical and follow-up data including basic characteristics, radiological data, and oncologic information at the last follow-up. The exclusion criteria were as follows: (i) Recurrent osteosarcoma. (ii) Patients who had received any treatment, including chemotherapy, radiotherapy, and surgery. (iii) Patients whose postoperative histology was not osteosarcoma were excluded and blood samples at the initial diagnosis were discarded. (iv) Incomplete medical records and inadequate follow-up. All clinical samples were collected with written informed consent from patients and the ethical approval was granted from the Committees for Ethical Review at our hospital. Furthermore, the current study was conducted in accordance with Declaration of Helsinki.

For the initial clinical evaluation and diagnosis, all patients received plain radiographs, CT, MRI, and bone scanning. After imaging, we performed the needle biopsy to clarify the diagnosis. Neoadjuvant chemotherapy was applied. The protocol consisted of two cycles before surgery and four cycles after surgery and included doxorubicin, cisplatin, high-dose methotrexate, and ifosfamide. En bloc resection with clear surgical margin was achieved for all patients. The routine follow-up including clinical examination and radiographs was performed every 3 months for the first 6 months, every 6 months for the first 3 years, and then annually. The chest CT scanning and bone scanning were performed every 6 months for the first 3 years, and then annually.

As described previously, exosome isolation was performed using the ultracentrifugation method and the final pellets of exosomes were stored at −80°C for the experiment (26). Meanwhile, for the experiments of transmission electron microscopy (TEM) and nanoparticle tracking analysis (NTA), the detailed procedures were described in the previous study (26).

The exosome fraction was characterized by TEM and nanoparticle tracking system for morphology, concentration, and size, as well as by Western blot analysis for exosome-associated biomarkers. TEM showed that the extracellular vesicle fraction from serum of patient with OS contained typical cup-shaped vesicles, which were the exosomes (Supplementary Fig. S2A). As nanoparticle tracking analysis was shown in Supplementary Fig. S2B, the size distribution of these serum exosomes of patients with OS was mostly around 100 nm. Exosomes from serum of patient with OS were positive for representative biomarkers (i.e., Tsg101, Alix, CD63) and negative for cell debris (Calnexin) in Western blot analysis (Supplementary Fig. S2C). In summary, the above mentioned experiments confirmed the extracellular vesicle extracted from serum samples as the exosomes.

Immunogold labeling of exosomal PD-L1 was performed according to previous study (26). For detection of PD-L1 in exosomes, Sr-exosomal PD-L1 from patients with OS was analyzed by ELISA and the detailed procedure was described in the previous study (26).

The concentration of exosomal total protein was quantified by the bicinchoninic acid assay (Thermo Fisher Scientific) using BSA as standard. Western blot analysis was performed as described previously (26). The following primary antibodies were used: (i) mouse anti-CD63 (sc-5275, Santa Cruz Biotechnology), (ii) mouse anti-TSG101 (sc-7964, Santa Cruz Biotechnology), (iii) mouse anti-Alix (sc-53540, Santa Cruz Botechnology), (iv) mouse anti-calnexin (10,427–2-AP, Promega).

Continuous variables were described as mean and SD, median, and range. Categorical variables were described as number and percentage. The Student t test or χ² test was used to compare continuous and categorical variables. The patients with NED or AWD at the final follow-up were defined as alive group. Meanwhile, the ones with DOD at the final follow-up were defined as dead group. A multivariate logistic regression was applied to identify the significant independent predictors of oncologic prognosis of alive and dead status. Model selection methods such as Wald backward elimination were used to identify significant factors from the explanatory variables.

ROC curve analysis was used to confirm the cut-off value. Cut-off values of Sr-exosomal PD-L1 for OS and disease-free survival (DFS) were estimated from the best operating points of the areas under the ROC curve analyses by using the Youden J statistic (sensitivity + specificity − 1). When the Youden J statistic value was the largest, the level of Sr-exosomal PD-L1 could be defined as the respective cut-off values of OS and DFS. OS was measured from admission, whereas DFS was measured from the time of operation, respectively with time of death and local recurrence and/or metastasis. Survival rate was analyzed with Kaplan–Meier curves and the log-rank test. Statistical analysis was performed by SPSS software (version 19.0; SPSS Inc) and P ≤ 0.05 indicated a statistically significant difference.

The data generated in this study were available within the article and its Supplementary Data (Supplementary Table S3). The raw data will be shared when the readers request the additional data.

The characteristics of 67 patients were listed in Table 1. There were 37 males and 30 females, with a mean age of 14.4 years (median, 13 years; 5–40 years). The mean follow-up duration was 46.7 months (median, 49.8 months; 6.1–79.0 months). Sixty-one patients had localized disease and 6 sustained metastatic disease. Thirty-nine patients (58.2%, 39/67) had no evidence of local recurrence and metastasis at the final follow-up. Seventeen patients (25.4%, 17/67) died of the disease at a mean of 29.8 months (median, 26.8 months; range, 6.1–62.3 months) postoperatively. Eleven cases were alive with disease (16.4%, 11/67). Rate of local recurrence was 7.5% (5/67). At the final follow-up, 32 patients had pulmonary metastasis and 11 cases sustained multiple metastases of lung and other organs of brain and bone (Supplementary Table S1). The 2-year and 5-year OS rates in the cohort were respectively 96.9% and 62.5% (Fig. 1A). Meanwhile, the 2-year and 5-year DFS rates were respectively 85.0% and 31.4% (Fig. 1B).

By immunogold labeling of Sr-exosomal PD-L1, we demonstrated PD-L1 existed in the serum exosomes isolated from patients with OS, which was also published in our previous study (26). Among these 67 patients, 27 (40.3%, 27/67) had positive PD-L1 expression in Sr-exosomes at the initial diagnosis for patients with OS by evaluation of immunogold labeling (Fig. 2). Our results illustrated that detection of positive Sr-exosomal PD-L1 by the immunogold labeling at the initial diagnosis indicated inferior outcome in their disease process during the follow-up. Of 17 patients who died of disease at the finial follow-up, 11 (64.7%) had the positive staining of Sr-exosomal PD-L1 by immunogold labeling. Furthermore, Sr-exosomal PD-L1 at the initial diagnosis using immunogold labeling was detected in 56.2% of patients who had pulmonary metastasis at the final follow-up (P = 0.014). Sr-exosomal PD-L1 at the initial diagnosis was detected in 81.8% of patients who had multiple metastasis at last follow-up (P = 0.005). However, there was no significant difference of the positive rate of Sr-exosomal PD-L1 at the initial diagnosis between patients who had the local recurrence and no local recurrence (P = 0.385; Supplementary Table S2).

We have previously reported that Sr-exosomal PD-L1 isolated from patients with OS correlated with pulmonary metastasis. To demonstrate the correlation between the level of Sr-exosomal PD-L1 at the initial diagnosis and disease status during the follow-up, we compared levels of Sr-exosomal PD-L1 of patients with different oncologic status (Table 1). Patients with pulmonary or multiple metastasis at the final follow-up had obviously higher Sr-exosomal PD-L1 levels at the initial diagnosis compared with ones without pulmonary metastasis (P = 0.013) or with single metastasis (P = 0.002; Fig. 3A and B). The level of Sr-exosomal PD-L1 of patients with OS at the initial diagnosis, who died of disease at last follow-up, was significantly higher than that of patients who were still alive at the final follow-up (P < 0.001; Fig. 3C). Meanwhile, statistically higher Sr-exosomal PD-L1 level at the initial diagnosis was observed in patients with disease event during the follow-up than that in those without disease event (P = 0.009; Fig. 3D). Nevertheless, this significant difference of level of Sr-exosomal PD-L1 at the initial diagnosis was not observed between groups of local recurrence and no local recurrence (P = 0.096; Fig. 3E).

We performed ROC curve analysis to confirm the cut-off value of exosomal PD-L1 for OS and DFS. When the value of exosomal PD-L1 was respectively 25.96 and 14.23 pg/mL for OS and DFS, the Youden J statistic values were the largest. Patients in the group of ≥14.23 pg/mL Sr-exosomal PD-L1 at the initial diagnosis had inferior DFS compared with patients in the group of <14.23 pg/mL Sr-exosomal PD-L1 at the initial diagnosis [2-year 47.8% (95% confidence interval, CI, 18.6–67.1; n = 12) vs. 77.2% (95% CI, 43.1–87.9; n = 55); 5-year 24.3% (95% CI, 12.6–45.2; n = 12) vs. 63.5% (95% CI, 42.1–77.1; n = 55), P = 0.002]. Meanwhile, patients in the group of ≥25.96 pg/mL Sr-exosomal PD-L1 at the initial diagnosis had poor OS compared with patients in the group of <25.96 pg/mL Sr-exosomal PD-L1 at the initial diagnosis [2-year 83.3% (95% CI, 64.1–90.5; n = 23) vs. 89.1% (95% CI, 68.2–97.1; n = 44); 5-year 41.7% (95% CI, 25.6–58.2; n = 23) vs. 85.4% (95% CI, 62.5–92.5; n = 44), P < 0.001; Fig. 4; Table 2].

According to the disease status of patients at the final follow-up, all patients were divided into two groups of alive (n = 50) and dead (n = 17). The demographic parameters, clinical established prognostic factors, and Sr-exosomal PD-L1 level at the initial diagnosis were respectively analyzed in the univariate and multivariate models. The univariate analysis demonstrated patients with metastatic disease at diagnosis, lower tumor necrosis rate, higher level of Sr-exosomal PD-L1 at the initial diagnosis, and positive detection of Sr-exosomal PD-L1 at the initial diagnosis by immunogold labeling had higher rates of death, which is shown in Table 3. Furthermore, the multivariate analysis illustrated that lower tumor necrosis rate and higher level of Sr-exosomal PD-L1 at the initial diagnosis were the independent risk factors of poor survival (Table 4).

For patients with osteosarcoma, refractory risk stratification depends on the presence of radiologically detected metastatic disease and primary tumor location, i.e., spine or pelvis (27, 28). Currently, there is fewer validated tools available at the initial diagnosis to identify patients with localized disease at high risk of relapse during the follow-up. Several prognostic factors have been reported to predict the oncologic prognosis of patients with osteosarcoma (29–34). Despite the importance of immunologic mechanism in osteosarcoma, the correlation between level of circulating exosomal PD-L1 and oncologic outcome has not been systematically evaluated for these patients (35). The aim of current study was to determine whether patients with newly diagnosed osteosarcoma, who had a higher level of circulating exosomal PD-L1 at the initial diagnosis, would undergo an inferior outcome in the future. In addition, correlations between level of exosomal PD-L1 at the initial diagnosis and clinical features of disease were also assessed.

It is difficult to predict the prognosis of patients with OS at the initial diagnosis before treatment due to lack of efficient prognostic parameters. It is difficult to predict the prognosis of patients with OS at the initial diagnosis before treatment due to lack of efficient prognostic parameters. In the literature, risk factor of metastatic disease at diagnosis was reported as the predictor of survival of patients with osteosarcoma. Because of small number of patients with metastatic disease at diagnosis, it was not the independent risk factor of poor survival in the multivariate analysis in the current study. However, the results of our previous and current study illustrated there was a tendency of correlation between the SrExPDL1 at the initial diagnosis and metastasis at initial diagnosis. Meanwhile, in the current study, using immunogold labeling of Sr-exosomal PD-L1, we detected the positive expression of Sr-exosomal PD-L1 isolated from the peripheral blood samples in 40.3% of patients with osteosarcoma. In addition, detectable Sr-exosomal PD-L1 had a tendency of significant association with metastatic disease, survival status, and disease event. These results demonstrated that, among patients with newly diagnosed osteosarcoma, detection of higher Sr-exosomal PD-L1 at the initial diagnosis was associated with an inferior outcome and declared that Sr-exosomal PD-L1 at the initial diagnosis before treatment showed promise for predicting the prognosis and monitoring the disease progression during the follow-up. Furthermore, an increasing risk of disease event and death during the follow-up was obviously associated with a higher Sr-exosomal PD-L1 level at the initial diagnosis before chemotherapy and surgery. In the current study, we evaluated the relationship between the level of circulating exosomal PD-L1 at the initial diagnosis before treatment and oncologic outcome during the follow-up. The result of current study revealed a highly significant association between the high level of Sr-exosomal PD-L1 at the initial diagnosis and clinicopathologic disease markers such as metastatic disease at diagnosis, tumor necrosis after neoadjuvant chemotherapy, OS, and DFS rates. Our results of mean follow-up of 47.6 months showed patients with OS, who died of disease at the final follow-up, had the higher level of Sr-exosomal PD-L1 at the initial diagnosis, which compared with that of patients who were still alive at the last follow-up. It also demonstrated higher Sr-exosomal PD-L1 level of patients with OS at the initial diagnosis had the significant relationship with the poor prognosis. Likewise, statistically higher Sr-exosomal PD-L1 level at the initial diagnosis was observed in patients with disease event during the follow-up than that in those without disease event during the follow-up. However, the level of Sr-exosomal PD-L1 at the initial diagnosis had no association with local recurrence during the disease process.

Although the exosomal PD-L1 isolated from the peripheral blood samples has proven useful for patients with carcinomas, there have been relatively few studies evaluating the utility of Sr-exosomal PD-L1 as a biomarker in patients with sarcomas (2, 3, 24). Our study demonstrated the feasibility of detecting exosomal PD-L1 at the initial diagnosis for prediction of the survival of patients with osteosarcoma. In the previous study, our team first confirmed the existence of PD-L1 in exosomes which were isolated from the serum of patients with osteosarcoma and elucidated Sr-exosome derived from patients with OS promoted the pulmonary metastasis in metastatic models (26). Furthermore, the current result also showed Sr-exosomal PD-L1 level at the initial diagnosis before treatment, in those who had pulmonary metastasis at the final follow-up, was statistically higher compared with those without pulmonary metastasis at the last follow-up. Our findings were concordant with the opinion that tumor cells evaded the immune surveillance not only by upregulating cellular PD-L1 expression, but the circulating exosomal PD-L1, which was reported in the literature (11, 25, 36–39). Of note, emerging evidence demonstrates that tumor-derived exosomes containing PD-L1 can recapitulate the effect of cell-surface PD-L1 and reveals a significant association between levels of circulating exosomal PD-L1 and rate of response to immunotherapy (11, 36–38, 40). Furthermore, refractory osteosarcoma has a poor response to immunotherapy due to cold immune microenviroment and whether it can improve the response to immunotherapy for patients with osteosarcoma through modulation of circulating exosomal PD-L1 is an interesting and noteworthy topic in the future (41).

On the basis of our findings in the current and previous study, there was no correlation and consistency between tumoral PD-L1 by IHC and Sr-exosomal PD-L1 by ELISA at initial diagnosis (20, 26). In the current study, our results showed that Sr-exosomal PD-L1 could be detected by ELISA in the most patients with osteosarcoma. However, the positive expression rate of tumoral PD-L1 by IHC was only 35.5%, which was reported in our previous study (20). Meanwhile, Li and colleagues also reported Exo-PD-L1 was not associated with PD-L1 IHC status for patients with non–small cell lung cancer (25). Theoretically, all immunosuppressive cells, supportive cells in the microenvironment, and osteosarcoma cells contribute to the systematic immunosuppression in patients with OS and those cells can release the exosomal PD-L1 into the blood, which contribute to the pool of Sr-exosomal PD-L1 in the blood. On the basis of the evidence, we would like to propose an opinion that exosomes isolated from the serum of patients are a heterogeneous mixture of both tumor cell–derived, immune cell–derived, normal cell–derived, and other cells in the microenviroment-derived exosomes. It is a comprehensive reflection of immune status of the whole body. Thus, it is easy to understand why there is no consistency between tumoral PD-L1 and Sr-exosomal PD-L1.

It has been reported that high level of PD-L1 expression in the tumor had significant association with poor prognosis. Koirala reported that PD-L1 expression was significantly associated with poorer 5-year event-free survival (23). However, the impact of exosomal PD-L1 on the sarcomagenesis is unclear. Our previous study has demonstrated the presence of PD-L1 on the surface of exosomes isolated from serum of patients with OS and illustrated exosomal PD-L1 derived from osteosarcoma stimulated the progression of pulmonary metastasis (26). In the current study, we focused on the prognostic value of Sr-exosomal PD-L1 at the initial diagnosis and aimed to elucidate the relationship between the level of Sr-exosomal PD-L1 at the initial diagnosis and survival status during the follow-up for patients with OS. For DFS and OS of patients with osteosarcoma, the cutoff values of Sr-exosomal PD-L1 of 14.23 and 25.96 pg/mL at the initial diagnosis before treatment, which could accurately predict the DFS and OS, were decided with reference to the largest Youden J statistic values. These findings proved the role of Sr-exosomal PD-L1 at the initial diagnosis as the promising biomarker for predicting the survival of patients with osteosarcoma during follow-up. Nevertheless, in the further study with larger number of patients with osteosarcoma, evaluation of Sr-exosomal PD-L1 is needed to validate their potential role as the monitoring biomarker.

All samples analyzed in our study were collected as part of a prospective study of liquid biopsy for sarcoma. Thus, the sample collection, exosome extraction, and handling strategies used in the current study were exclusive for maintaining the integrity of exosome samples. Our study was therefore the first large cohort to demonstrate that qualitative and quantitative Sr-exosomal PD-L1 detection at the initial diagnosis provided prognostic information for patients with osteosarcoma during follow-up. To promote the current finding into the clinical application, we are further designing a larger prospective validation study which evaluates the clinical utility of longitudinal Sr-exosomal PD-L1 during the therapeutic process. We will also elucidate whether Sr-exosomal PD-L1 of patients with osteosarcoma could provide a useful prognostic marker of chemoresponsiveness for patients with OS. Regrettably, this study also had several limitations. First, it should be noted that the number of cases in the current cohort was still limited and we could not accurately analyze the relationship between Sr-exosomal PD-L1 level at initial diagnosis and disease prognosis. Second, the longitudinal change of Sr-exosomal PD-L1 was not to be analyzed in the current cohort. Although the dynamic change of Sr-exosomal PD-L1 expression during therapeutic course has been evaluated between pretreatment and posttreatment in the previous study, the case number was still small and the further validation in the next prospective study needs to be implemented (26).

In conclusion, the use of liquid biopsy of circulating exosomal PD-L1 at the initial diagnosis provides a robust means of predicting the prognosis for patients with newly diagnosed osteosarcoma. Detection of Sr-exosomal PD-L1 at the initial diagnosis in these patients provides predicted information that may ultimately be used to improve risk stratification approaches.

No disclosures were reported.

J. Wang: Conceptualization, data curation, writing–original draft. W. Guo: Supervision, project administration. X. Wang: Supervision, methodology. X. Tang: Data curation, validation. X. Sun: Data curation. T. Ren: Data curation.

This work was supported by Peking University People's Hospital Research and Development Funds (nos. RDG2021–02 and RDL2022–14), the Natural Science Foundation of China (no. 82272947), and Funds of Beijing Shijitan Hospital (no. 2022-C04).

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.