Osteosarcoma is the most common primary malignant bone tumor, accounting for 20–40% of all bone tumors (1, 2). Tumors tend to occur in children and adolescents and have a high metastatic potential. The 5-year survival rate of patients with standard treatment is about 60–70% (3). However, approximately 15–20% of patients have developed metastases at initial diagnosis, and the five-year survival rate is significantly reduced, so far there is no effective treatment regimen (4–6). In addition, insensitivity to chemotherapy and tumor recurrence are also important factors leading to significantly reduced patient survival (7, 8). These clinical features can identify high-risk patients and aid in treatment planning (9). However, the progression of the disease may be distinct in patients with similar clinical features. Therefore, more factors need to be considered to facilitate precision treatment.

Recently, researchers have made many efforts in developing biomarkers that can more accurately predict the prognosis of patients and developed many biomarkers with potential, such as non-coding RNA (NcRNA), circulating tumor cell (CTC), and circulating tumor DNA (ctDNA) (10–12). However, these markers have not been applied to clinical work due to reasons such as the cost of testing, strict technical requirements for biopsy, etc. Fortunately, studies have shown that preoperative hematological markers such as neutrophil to lymphocyte ratio (NLR), platelet to lymphocyte ratio (PLR) or hematology-based scores such as Glasgow Prognostic Score (GPS), Controlling Nutritional Status (COUNT) show great potential in predicting the prognosis of cancer patients (13–15). Compared with NcRNA, CTC, and ctDNA, most of these hematological markers originate from routine examinations of patients on admission without the need for additional testing costs. Many recent studies have confirmed the value of these markers in predicting survival in cancer patients, including osteosarcoma (16–18).

Studies over the past decade have shown that cancer cells can promote their survival by reprogramming metabolic pathways, and therefore, metabolic reprogramming is also considered to be one of the hallmarks of cancer (19, 20). Local tumors can even impair antitumor immunity by affecting host metabolism through cachexia (21). However, studies on the ability of hematologic metabolic markers to predict the prognosis of cancer patients are contradictory and limited compared with hematologic inflammatory markers (22).

In this study, we collected the hematological markers related to metabolism and explored the significance of disturbances in these markers in patients with osteosarcoma. Through iterative least absolute contraction and selection operator (LASSO) COX proportional hazards regression analysis we constructed nutritional metabolism related prognostic scoring system (NMRS) and assessed the predictive power of the scores through multiple dimensions. In addition, we also explored the superiority and limitations of NMRS by comparing it with existing scoring systems and clinical features.

Although limb salvage surgery for osteosarcoma has been continuously improved in recent years, and most patients can obtain satisfactory limb function after surgery, it is still a difficult task for osteosarcoma patients to achieve survival (1, 24, 25). Early identification of high-risk patients and development of personalized treatment options are expected to improve the prognosis of patients (26). Unfortunately, in osteosarcoma, no new prognostic marker has been truly applied to clinical work except for clinical features such as metastatic status and tumor necrosis rate (9). Although recent studies have shown the potential of NcRNA, CTC with ctDNA as a tool for early detection, postoperative monitoring of cancer patients (23, 27–29). However, because the concentration of CTC and ctDNA collected from patients is usually very low. In addition, standard methods for isolation, enrichment, or detection of NcRNA, CTC, and ctDNA are lacking (30, 31). Therefore, there is still a certain distance between NcRNA, CTC and ctDNA for real clinical application. In contrast, because most of the hematological markers are derived from routine examinations such as blood routine and liver and kidney function, no additional testing costs are required. Moreover, the value of hematological markers in predicting the prognosis of cancer patients has been generally accepted, for example, NLR has been written into the 2020 European Association of Urology Guidelines on Upper Urinary Tract Urothelial Carcinoma (32). However, the prognostic value of indicators such as TG, HDL and LDL in cancer patients has not been determined. Therefore, we extensively collected some hematological markers that may reflect nutritional status and metabolic reprogramming in cancer patients and constructed NMRS. As expected, NMRS demonstrated greater stability and predictive power. With the exception of NMRS, none of the hematological markers had independent prognostic ability in both the training and validation sets. In addition, NMRS has better discrimination ability than clinical features such as tumor metastasis status, especially in identifying low-risk patients.

The lipid metabolism plays an important role in cancer metabolic reprogramming. Cancer cells, as well as other cell types in the tumor microenvironment, utilize various methods to obtain lipids and extensively rewrite their metabolism (33). Studies have shown that cancer stem cells maintain their stem cell characteristics through lipid metabolism (34). In addition, alterations in lipid metabolism can impair antitumor immunity and promote iron death escape (35, 36). Several large-scale lipidomics studies have provided compelling evidence for the potential of lipids as prognostic biomarkers for cancer (33, 37, 38). In this study, lipid metabolic markers were the most important cornerstone of NMRS, with TCH, TG, HDL, and LDL coefficients of −1.127, 0.596, −0.188, and 0.901, respectively, which indicated that higher TCH and HDL was associated with better prognosis, while triglyceride and LDL were the opposite.

Serum albumin and globulin reflect the nutritional and inflammatory status of individuals and have shown potential prognostic value in a variety of tumors (39–41). On the one hand, albumin reflects the nutritional status of individuals, and in general, patients with poor nutritional status have a higher risk of postoperative complications, which may greatly shorten the survival time of cancer patients (42). On the other hand, decreased serum albumin may be due to increased capillary permeability caused by cancer-related inflammation resulting in albumin escape into the interstitium and absorption by cancer cells, decomposition and utilization (43, 44). In addition, globulin is considered a pro-inflammatory protein and has been shown to be associated with poor prognosis in cancer patients (45). In this study, the coefficients of albumin, globulin and AGR were −0.286, 0.417 and −0.497, respectively, confirming the previous conclusions (46–48).

Although the mechanisms by which high glucose promotes cancer aggressiveness vary by cancer type, it is generally accepted that high glucose is associated with poor prognosis in cancer patients. Studies have shown that abnormal elevated glucose that does not reach the diagnostic level of diabetes is also associated with poor prognosis in cancer patients (49). A high glucose environment leads to the up-regulation of aerobic glycolysis (Warburg effect) -related pathways in cancer cells. In addition, glucose can also activate a variety of signaling pathways involved in cell proliferation, metastatic capacity, and chemoresistance, including ERK, STAT3, and NF-κB (50, 51). The coefficient for glucose in this study was 0.562, indicating that elevated glucose is associated with poor prognosis in osteosarcoma patients. Notably, to our knowledge, the prognostic value of glucose in osteosarcoma patients has not been previously investigated. Finally, CONUT consists of lymphocytes and albumin together with cholesterol, so it is not difficult to understand that CONUT is associated with the prognosis of cancer patients. The coefficient of CONUT in NMRS was 0.354, which is consistent with previous findings (52).

As shown by Figure 1, the results of our ROC analysis showed that there was no significant difference in the prognostic value of individual hematological markers between the training and validation sets. The AUC values of NMRS were higher than all the individual hematological markers. In addition, NMRS combines all of the above indicators and assigns coefficients to each indicator, which comprehensively reflects these indicators and better and more stably represents the nutritional and metabolic status of patients. In fact, we also performed further analysis of individual hematological marker. Our results showed that only AGR had some prognostic value in both the training set and the validation set. In multivariate cox regression analysis combined with clinical characteristics, none of the individual hematological parameters were independent prognostic factors for overall survival of osteosarcoma patients in both the training and validation sets. These results further demonstrate the superiority of NMRS. We believe that NMRS with higher and more stable predictive ability is more likely to be applied and promoted in clinical practice.

However, it must be acknowledged that our study has certain limitations. Despite the training and validation sets of our study setup, all patients were from one clinical institution and the study was retrospective and not blinded, so there may have been a selection bias. At the same time, there were only two osteosarcoma patients over 60 years of age in the included population, so the use of NMRS in the elderly requires caution. Moreover, some patients may have problems with insufficient follow-up time. In addition, although the predictive ability of NMRS is higher than that of individual hematological markers, the computational method of NMRS is more complex. Finally, the metabolic nutrition-related indicators included in the study were derived from blood routine and liver function only. Some important indicators that reflect the patient’s nutritional status, such as nutrients or dietary intake, are neglected. Therefore, well-designed prospective randomized controlled studies are needed to validate our conclusions. Finally, we believe that further studies are needed to explore the relationship between NMRS and dietary intake in patients.

Our study show the prognostic value of NMRS in osteosarcoma. Compared with individual hematological markers, NMRS has predictive ability and predictive stability. NMRS-based nomogram also have good predictive accuracy.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee of West China Hospital of Sichuan University. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin.

LL, YZ, and CT designed the study. LL, ZL, XH, YW, JL, and QC jointly collected and managed the data. LL and TG drafted the manuscript. ML, YL, and LM reviewed and corrected the manuscript. YZ and CT oversaw the entire research process. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.