Predictive value of dynamic change of haemoglobin levels during therapy on treatment outcomes in patients with Enneking stage IIB extremity osteosarcoma

Osteosarcoma is the most common solid bone malignancy in children and adolescents. With the integration of neoadjuvant, adjuvant chemotherapy and surgery, the 5-year overall survival (OS) rate has increased to 50–70% for non-metastatic extremity osteosarcoma. However, an improvement of OS has been sparse in the last 2 decades, although clinicians have tried to combine different chemotherapy agents to achieve better survival outcomes [1]. Many hypotheses account for the lack of improvement in OS. However, the lack of individual risk stratification and tailored therapy for osteosarcoma patients may partly account for this disappointing progress.

Haemoglobin (Hb) levels are reported to be a prognostic factor in many malignancies [2‐5] with low pretreatment Hb level shown to predict poor outcomes in most studies. Lower Hb concentration may cause a decline of blood oxygen supply and tumour hypoxia, which may contribute to the development of treatment resistance [6, 7]. Additionally, dynamic Hb changes during treatment are reportedly better survival predictors for nasopharyngeal carcinoma [8, 9]. Osteosarcoma patients are widely thought to benefit from high doses chemotherapy of long duration. As chemotherapy resistance is a major cause of treatment failure [10]. Changes in Hb during neoadjuvant and adjuvant chemotherapy for osteosarcoma patients may be a fruitful area of study; few data regarding the prognostic value of Hb levels for osteosarcoma patients have been available.

In this study, we investigated whether Hb levels and differences in decreased Hb levels during therapy could predict the survival outcomes of patients with Enneking stage IIB extremity osteosarcoma.

In this study, we investigated relationships between Hb levels during different therapy periods and survival outcomes for extremity osteosarcoma. Hb levels at diagnosis or during neoadjuvant or adjuvant chemotherapy had no predictive ability for 5-year OS or LMFS in the uni- or multivariable analyses. However, continuous Hb decrease or ΔHb > 7.6 g/L were predictive of 5-year OS in these patients. In subgroup analyses, anaemia patients whose Hb decreased > 7.6 g/L had worse survival outcomes than those whose Hb decreased < 7.6 g/L.

Anaemia has been found to reduce survival in many cancers, including cancers of cervix, head, neck, lung and breast. Published articles have indicated several underlying mechanisms. Anaemia is widely thought to lead to tumour hypoxia, and subsequently to angiogenesis, and genetic mutations, and to resistance to apoptosis, chemotherapy and radiotherapy. Green et al. reported that hypoxia would directly reduce formation of reactive O₂ species and indirectly slow the cell cycle to affect the chemotherapeutic response [13]. Hypoxia may also decrease tumour control through the induction of hypoxia-inducible factor 1 alpha (HIF-1α), which could combine with constitutive HIF-1βto form a transcription factor that improves expression of proangiogenic agents, such as erythropoietin, vascular endothelial growth factor and glucose transporters [14]. Hypoxia is proven to increase metastasis and progression of osteosarcoma through the HIF-1α/CXCR4 pathway in vitro [15, 16]. HIF-1α also functtions in anaemia induced- doxorubicin resistance of human osteosarcoma cells [17].

Some authors suggest that anaemia is a marker for malnutrition, other comorbidities or the severity of the underlying illness [18]. Anaemia may result from bleeding, nutritional deficiencies, bone marrow damage and the malignant process itself. We may reasonably spectulate that anaemia interacts with tumour progression to affect survival outcomes of patients with malignancies.

We found no prognostic value for Hb concentrations, per se, at different treatment times on 5-year OS of osteosarcoma. However, patients with continuous Hb decreases during therapy had poorer outcomes than those without continuous Hb decreases. Furthermore, the cut-off of 7.6 g/L for decreases in Hb level (compared with initial Hb level) was a predictor of 5-year OS that we found for this population. The specificity of cut-off value for ΔHb was 31.6%, which may be caused by small amount of included patients and presence of other confounding variables, such as albumin level.

As changes in Hb levels during treatment are reportedly more important prognostic variable than the Hb level at baseline [9, 19, 20]. We surmised that a certain amount of Hb decrease would represent poorer nutrition and decreased response to CHT. However, the underlying mechanism is still unclear. Additionally, our subgroup analyses revealed that patients with pretreatment anaemia, neoadjuvant anaemia or adjuvant anaemia had poorer survival outcomes when Hb concentration decreased more than 7.6 g/L from their initial Hb concentrations. Our study did not find significant prognostic value of Hb levels on 5-year LMFS. This may reflect the insufficient number of included patients or reporting bias. The definitive effect of Hb levels on lung metastasis should be studied in future clinical trials.

Correction of anaemia should be considered in clinical practice as decreased Hb levels by more than 7.6 g/L apparently affect treatment outcomes in extremity osteosarcoma. To our knowledge, blood transfusion and erythropoietin could be used to correct anaemia. Kapp et al. found that transfusion of red blood cells could quickly improve Hb concentrations and prognosis in anaemia patients who underwent radiotherapy for cervical cancer [21]. The latest prospective, randomized controlled trial revealed that patients with cancers who received a restrictive strategy of erythrocyte transfusion (transfusion when haemoglobin concentration < 7 g/dl) had more mortality and severe clinical complications than those receiving a more liberal strategy (transfusion when haemoglobin concentration < 9 g/dl) [22]. This study indicates the increased necessity of anaemia rectification. However, some studies reported a negative effect of blood transfusion in anaemia patients who underwent radiotherapy for cervical carcinoma, possibly because of immunosuppression and other adverse effects [23, 24]. In recent years, erythropoietin has been extensively used in the clinic. Some studies have shown that in patients with oesophageal cancers, adminstration of erythropoietin led to lower transfusion requirements and a greater survival benefit than placebo [25]. Neverthelss, a prospective, randomized-controlled trial of 351 patients with head and neck cancers found that patients who received erythropoietin to achieve Hb levels higher than 140 g/L (women) or 150 g/L (men) had poorer locoregional progression-free survival than those who received placebo [26]. The roles of blood transfusion and erythropoietin thus remain controversial. Few data are available on the effect of anaemia correction for osteosarcoma patients. Therefore, future studies on how to correct anaemia for patients who suffer from cancers, including osteosarocma, are needed.

The strengths of our study include a cohort who received standard treatments and long follow-up (more than 5 years); a focus on the prognostic value of dynamic changes in Hb levels during treatments, and the use of multivariate methods of statistical analysis to test the prognostic value of Hb level change during therapy. Our study also had some limitations. Firstly, it was a retrospective study, including data from only one institution. Secondly, the number of patients included was not large enough to draw concrete conclusions regarding differences between treatment groups. Thirdly, continuous haemoglobin decrease can only be evaluated at the conclusion of treatment according to the data analysis regimen of this study.

In our study, Hb levels at the time of diagnosis, neoadjuvant or adjuvant chemotherapy cannot predict 5-year OS and LMFS on Enneking stage IIB extremity osteosarcoma. However, patients with continuous Hb decrease were shown to have worse 5-year OS than those without continuous Hb decrease. Patients whose Hb concentration decreased by more than 7.6 g/L had worse outcomes than patients whose Hb decreased by less. The role of anarmia and the effects of attempts to correct it, in outcomes for osteosarcoma patients should be investigated in future trials.