Osteosarcoma is a rare but aggressive primary bone tumor that predominantly arises in adolescence and young adulthood (1). The disease has a strong predilection for the metaphyses of long bones, with the distal femur, proximal tibia, and proximal humerus being the most commonly affected sites. Osteosarcoma accounts for 56% of all malignant bone tumors, followed by Ewing sarcoma (34-36%), and chondrosarcoma (less than 10%) (2). While it represents less than 1% of all cancer cases diagnosed annually in the United States, it contributes to approximately 3% of childhood cancers. In Greece, an estimated 30-50 new cases are diagnosed each year, accounting for approximately 0.05% to 0.08% of all new cancer cases in the country (2). The absence of centralized cancer registries may hinder comprehensive data collection.

When treated with surgery alone, more than 80% of osteosarcoma patients progress despite achieving local tumor control. It is suggested that subclinical metastatic disease is present early in the disease course and systemic chemotherapy can eradicate these tumor deposits when tumor burden is low. Two randomized studies in the 1980s showed that systemic chemotherapy offered survival benefit in osteosarcoma patients in the adjuvant setting. Later neoadjuvant chemotherapy served as a method to increase suitable surgical candidates by diminishing tumor. Moreover, the response to neoadjuvant chemotherapy is a major prognostic factor. With modern therapy, more than 60% of non-metastatic osteosarcoma patients will be long-term survivors (3-7). Various prognostic factors have been linked to survival outcomes, including tumor size, presence of metastases at diagnosis, histological subtype (8,9), histological grade, response to neoadjuvant chemotherapy, and the adequacy of surgical margins (10-13). The grade of histological response to neoadjuvant chemotherapy appears to correlate significantly with prognosis.

The histological response to primary treatment is typically evaluated based on persistence or absence of viable tumor cells (total necrosis). This is often expressed as a percentage of tumor necrosis, which has been correlated with prognosis (14-19). Histological responsiveness to neoadjuvant or induction chemotherapy is a major determinant of prognosis (15,19-26) . Different grading systems have been used to evaluate the value of histological response. A 4-grade scale (Grade 1: little or no evidence of necrosis; Grade 2: necrosis of 50%-90%; Grade 3: necrosis between 90%-99%; finally, Grade 4: 100% necrosis) is included in Huvos system (27), a 6-grade scale (Grade I: 100% necrosis, complete response to Grade IV: no response, all tumor cells viable ) was used by Salzer-Kuntschik (28) ,and a 3-grade scale (Grade I: complete response to Grade IV: poor or no response) by Ayala (29). Several studies (13,19,20,25,30-32) have used these grading systems and translated these data into percentages of necrotic tissue, thus good responder patients are typically defined as those with necrosis ≥95% (19,21-23), ≥85-90% (17,23,24,30,31,33) or ≥60% (20,25).

The present study aimed to examine the applicability of established findings and analyze the associations between various clinical and pathological factors within the Greek population. Therefore, this study represents the most comprehensive analysis of osteosarcoma patients in Greece to date, from eight centers across the country. The inclusion of multiple centers allowed for a broad assessment of osteosarcoma management within Greece, capturing variations in treatment protocols, histological assessment practices, and patient outcomes. The findings of this multi-center approach not only reflect the national trends but also contribute to the global understanding of osteosarcoma, particularly in Mediterranean populations.

Patient selection. From 2004 to 2022, 77 osteosarcoma patients were treated in 8 medical centers in Greece: Attikon General University Hospital of Athens, General Anticancer Oncological Hospital “Agios Savvas”, Ippokrateio General Hospital of Athens, BIOCLINIC General Hospital of Athens, MITERA Hospital, Metropolitan Athens Hospital, Children’s Hospital Agia Sophia Athens, General University Hospital of Larissa. Medical records were reviewed retrospectively to assess patient-related, treatment-related, and survival variables. All diagnoses were confirmed by a tissue pathological examination in each center. This study was approved by Attikon University hospital ethics committee (ΕΒΔ210/27-03-2023). Written informed consent was given from each patient to collect and analyze clinical data.

Histological diagnosis. All patients included underwent a biopsy (core needle or surgical) and histological classification was established according to the World Health Organization classification of 2013. Histopathologic slides of patients treated before 2013 were reviewed according to the new classification. Osteosarcomas were classified as classic/conventional, telangiectatic (TOS), osteoblastic (OSS), chondroblastic (COS), fibroblastic, extraskeletal, chondrosarcoma (CS), undifferentiated, mixed, periosteal and giant cell-rich osteosarcoma (GCRO). Their characteristics are shown in Table I.

Preoperative evaluation and chemotherapy. Preoperative chemotherapy was based on Rosen regimens (including methotrexate, cisplatin, doxorubicin) or API/AI regimens (cyclophosphamide, doxorubicin, ifosfamide), depending on the age of the patient. Patients aged under 18 years were treated with methotrexate-based regimens. Patients older than 30 years received API/AI chemotherapy. The majority of intermediate aged patients (18-30 years) were treated with methotrexate regimen, according to physician’s choice.

Surgery. After preoperative chemotherapy, complete restaging was performed to assess the treatment response. The patients underwent a surgical excision (amputation or limb-sparing surgery) depending on the location and extension of the tumor, neurovascular bundle involvement, and the age and lifestyle of the patient in an expert center. Afterwards they received adjuvant chemotherapy.

Assessment of histological response to chemotherapy. Standardized pathological evaluated protocols, similar to the criteria by Huvos, were used to assess tumor response. Surgical specimens from resections were processed, and the cut surfaces were carefully examined macro- and microscopically, as per classical pathological assessment. Multiple sections were taken from areas most representative of the tumor. Tumor response to neoadjuvant chemotherapy was categorized as either good (≥90% tumor necrosis or minimal viable tumor) or poor (<90% tumor necrosis or significant viable tumor). A good histological response corresponded to the grade III and IV criteria of Huvos, whereas a poor response corresponded to Grades I and II (27).

Postoperative chemotherapy. Patients received postoperative chemotherapy with the same regimens used preoperatively. Postoperative evaluation included clinical evaluation and computed tomography (CT) scans of the chest and operated limb unless otherwise indicated.

After completion of the adjuvant treatment, patients were followed as outpatients every 3 months for 3 years, and then every 6 months. During these evaluations, a plain radiography or CT scanning of the involved limb and of the chest was performed.

Statistical analysis. Continuous variables are presented as median (interquartile range; IQR) and categorical variables as counts and percentages. The Kaplan-Meier method with 95% confidence intervals (CIs) was employed to estimate OS and DFS. OS was defined as the time from the diagnosis to death from any cause and was censored at the date of the latest follow-up. DFS was defined as the time from diagnosis to local recurrence, distant metastasis, or the date of death and the cut-off was made at the date of the latest follow-up. The log-rank test and Cox proportional hazard model were used to assess prognostic factors. Factors were categorized according to previous reports. Log-rank tests were applied for sex, age (<30 years or ≥30 years), primary tumor site (short vs. long bone), maximum tumor diameter (<10 cm or ≥10 cm), tumor necrosis (<90% or ≥90%). The Wilcoxon Rank Sum test was used to compare differences in continuous variables (e.g., age) between groups. Significant factors in univariate analyses were entered into a multivariate Cox proportional hazard model. Patients with missing data were excluded from the corresponding analyses. The sample size is clearly reported in the results of each analysis. Statistical analysis was performed using R 3.5.3 (R Foundation for Statistical Computing, Vienna, Austria), with significance level set at p<0.05.

Institutional Review Board Statement. Ethical review and approval were waived for this study due to its retrospective design and use of anonymized data, which does not involve direct interaction with human or animal subjects.

Informed consent statement. Informed consent was obtained from all participants involved in the study.

Patients’ characteristics. Seventy-seven osteosarcoma patients were included in this study. The median age was 31 (IQR=19-48) (Table I). Thirty patients were female (45%) and 37 were male (55%), aligning with the general male predominance observed in osteosarcoma (male-to-female ratio 1.4:1). Median DFS and OS were 19 and 27 months respectively. The histological subtype was osteoblastic in 14 patients (25%), chondroblastic in 8 patients (14%), fibroblastic in 3 (5.4%) patients and other subtypes in 31 patients (55%). Short bones were affected in 24 patients (32%) and long bones in 51 patients (68%). Overall, 66 (93%) of patients underwent surgery. Metastasis was noted in 10 out of 55 evaluable patients (18%).

Association with outcome. Initial analysis of the data revealed that patients with osteosarcoma involving short bones were significantly older (median age=43 years, IQR=30-50) than those with tumors in long bones (median age=26 years, IQR=18-40; p=0.006, Wilcoxon rank-sum test). Tumor necrosis rate was greater than 90% in 10 out of 32 evaluable patients (30%) according to Huvos criteria. The tumor size was greater than 10 cm in 4 out of 29 evaluable patients (12%). We therefore investigated whether the relationship between sex, age at the time of diagnosis, tumor size, tumor location, histological subtypes, histological response to chemotherapy, is associated with DFS and OS. Our analysis demonstrated that 90% or greater tumor necrosis was significantly associated with longer DFS (HR=0.09, 95% CI=0.09-0.01, p=0.003) but not with OS (HR=0.62, 95% CI=0.24-1.58, p=0.3). Sex also emerged as a significant predictive factor, associated with DFS, with male patients having shorter DFS than female patients (HR=5.61, 95% CI=2.12-14.9, p<0.001) (Figure 1, Figure 2). Age, grade, tumor size or bone affected (long vs short) were not significantly associated with survival. Distant metastasis was associated significantly with shorter OS (HR=3.7, 95% CI=1.5-9.16, p=0.01) (Table II, Table III).

In line with the evolving use of limb-sparing surgery, the introduction of neoadjuvant chemotherapy for osteosarcoma, coupled with the preoperative evaluation of tumor response, provides critical insights into tumor drug sensitivity and disease prognosis. Our study demonstrated a significant difference in survival outcomes between osteosarcoma patients achieving more than 90% tumor necrosis and those with less than 90% necrosis, as evaluated using the Huvos grading system. Consistent with previous findings from a U.S. study (14), multivariate analysis revealed that 90% and higher tumor necrosis rate is an independent predictor for DFS.

Tumor necrosis threshold has been debatable across different populations and healthcare settings. In a retrospective investigation comprising 438 osteosarcoma patients, individuals exhibiting less than a 50% necrosis rate demonstrated inferior OS compared to cohorts with varying grades of necrosis (50-75%, 75-98% or 98-100%) (34). Due to the limited sample size in our study group, it is challenging to definitively conclude that a tumor necrosis rate of 90% or higher is the optimal cutoff for predicting disease-free survival (DFS). However, when considered alongside the findings of previous studies (14,17,23,24,30,31,33), using 90% necrosis as a surrogate outcome measure appears to be a reasonable approach.

Recently a deep-learning model (DLM)-based evaluation of viable tumor cell density suggested alternative approaches to overcome limitations of traditional histological tumor necrosis assessment that often relies on pathologists’ “eyeball assessment” of viable vs. necrotic tumor areas across tissue sections (35). The researchers found that in grade II cases (necrosis rate: 50-90%) DLM could better predict prognosis than traditional pathologist assessment. In these low necrosis rate osteosarcomas with incomplete tumor cell necrosis that would be falsely identified as viable, the DLM would more accurately assess and reflect cell death (35). However, the tumor cell density did not show association with prognosis. Advanced artificial intelligence (AI) techniques may address heterogeneity in assessment of tumor necrosis and improve prognostic accuracy (36).

Only 30-50% of patients are reported to respond well to neoadjuvant chemotherapy (14,37), with the rest experiencing worse outcomes and high risk of recurrence or metastasis (8,26). Poor response to neoadjuvant treatment in osteosarcoma is still a topic of interest both translationally and clinically. A phase 3 trial comparing MAP (methotrexate, cisplatin, doxorubicin) with MAPIE (methotrexate, cisplatin, doxorubicin, ifosfamide and etoposide) did not manage to show any benefit for patients who had a poor histological response to chemotherapy (3). Event-free survival was not better for the poor responders who received the addition of ifosfamide and etoposide, underscoring the complexity of management of chemo-resistant osteosarcomas. The poor response to neoadjuvant chemotherapy, particularly to key agents like cisplatin and doxorubicin might indicate both biological and therapeutic causes. Patient-derived cell lines show different expression and mutational profiles in cisplatin and doxorubicin resistant models (38). Resistance to neoadjuvant treatment remains an unsolved problem and highlights the need for biomarkers to identify the best candidates for preoperative management, allowing treatment stratification and alternative therapeutic strategies. Patients with poor response to neoadjuvant treatment tend to recur/metastasize after the end of adjuvant chemotherapy. Therapeutic options for these patients remain limited, demonstrating modest clinical benefit, with the majority of studies to show a median progression-free survival (mPFS) below 7 months (39). Targeted therapies and immunotherapy have shown controversial success in improving the outcome (40).

Sex was also associated with prognosis in our study, emphasizing the potential role of biological factors such as immune response differences or sex-specific osteosarcoma biology. Female patients were more likely to have longer DFS than male patients. Various studies, highlight sex differences in immunity (37,41,42). Females exhibit more robust innate and adaptive immune responses than males, evidenced by fewer early childhood infections and higher autoimmune diseases prevalence as adults (41,43-45). When considering sex-specific treatment strategies, differences in gene expression and immune related pathways have been noted in osteosarcoma. These genes include CDK4 (46), LCK (47,48), ROS1 (47,48), FLT3 (47) and TP53 (46), which are targeted by therapies either approved or under clinical investigation in sarcomas and other cancers (41).

Interestingly, tumor size, histological subtype, and bone location were not significantly associated with survival in the Greek cohort. This might be due to sample size, population characteristics or different treatment strategies and warrants further investigation. In addition, the observation that older patients are more likely to have short bone involvement is an intriguing finding of age-related patterns in osteosarcoma presentation (49). In addition, the presence of metastasis had a negative impact on OS in our cohort, as expected, in line with previous studies (50).

Although not explicitly analyzed in our study, regional and socioeconomic factors have been shown to play an important role in histological response and survival (14,37). Previous study in the U.S. has demonstrated that patients with lower socioeconomic status had significantly lower rates of achieving more than 90% tumor necrosis (14). These findings suggest that non-biological factors, including disparities in access to care, treatment timing, and healthcare infrastructure, may influence treatment outcomes. In our cohort 30% of the patients achieved 90% or greater necrosis, compared to 35% of osteosarcoma patients in the U.S. National Cancer Database cohort (14). The slightly lower rate in the Greek cohort might reflect differences in regional treatment practices and healthcare access.

Our study has several caveats mainly due to its retrospective character. The patients included did not receive the same regimen; the use of methotrexate was not decided with universal criteria but according to physician’s choice. Furthermore, histological assessment was not performed by the same pathologist. Thus, grading and the exact subtype of osteosarcoma in each case may be a matter of discrepancies. Additionally, this is a study of 77 cases only, which makes the extraction of conclusions weak. While acknowledging these constraints, our study provides valuable insights for the management of osteosarcoma in Greece and establishes a foundation for future prospective investigations to validate and expand upon these findings, such as optimizing chemotherapy regimens or incorporating novel agents targeting key molecular pathways.

This study reinforces the value of achieving a cutoff of 90% necrosis as a treatment goal as it might serve as an independent prognostic value for patients with osteosarcoma undergoing chemotherapy. Poor response to neoadjuvant chemotherapy remains a challenge highlighting the need for innovative therapeutic approaches. Molecularly targeted therapies, immune checkpoint inhibitors, and advanced drug delivery systems, could provide new avenues for improving response rates and overall survival. Furthermore, in our study population, male osteosarcoma patients were associated with worse DFS compared to female patients. The mechanisms responsible for this association may be linked to variations in tumor biology or immune response. A thorough understanding of these mechanisms is crucial in developing tailored strategies that can optimize outcomes for both males and females with osteosarcoma.

While our study provides valuable insights into the management of osteosarcoma, it is important to acknowledge the need for further larger, multi-institutional studies, to strengthen the generalizability of these findings. Additionally, integrating emerging technologies such as AI in pathological tumor assessment will be crucial in enhancing diagnostic accuracy and improving treatment outcomes across diverse populations.

The Authors declare no conflicts of interest.

AK and MM: Conceptualization, Methodology, Software, Writing-Original draft preparation. MM, AK, AP: Supervision. IK, PO, SK, PA, GA, EM, EG, MA, AP, AB, MK, NG, EZ, AA, VK, EL, PK, PJP, IB, AP: Writing-Reviewing and Editing.

This study was funded by the Hellenic Group of Sarcomas and rare tumors.

During the preparation of this manuscript, a large language model (ChatGPT, OpenAI) was used solely for language editing and stylistic improvements in select paragraphs. No sections involving the generation, analysis, or interpretation of research data were produced by generative AI. All scientific content was created and verified by the authors. Furthermore, no figures or visual data were generated or modified using generative AI or machine learning-based image enhancement tools.