Effect of Hemoglobin on the Prognosis of Patients with Advanced Cancer in Palliative Care Settings

Background: Hemoglobin is a prognostic factor for many cancer patients. However, its effect on the prognosis of patients with advanced cancer receiving palliative treatment is still unclear. Objective: The aim of this study was to assess whether hemoglobin can be used as a prognostic indicator for patients with advanced cancer receiving palliative treatment. Methods: From July 2013 to October 2015, 368 consecutive patients were treated in the palliative treatment center of the Shanghai Cancer Center of Fudan University. The data for 105 patients were extended in the follow-up. The cut-off value selected for hemoglobin was 100 g/L. Results: The median survival time of patients with low hemoglobin was shorter than that of patients with high hemoglobin (41 days vs. 84 days). In the follow-up readmission cohort (n = 105), the median survival time for patients with high hemoglobin (HHb → HHb) was 3.44 times longer than that of patients with low hemoglobin (HHb → LHb). Thus, both low hemoglobin and decreased hemoglobin were identied as independent prognostic factors for poor prognosis. Conclusions: In palliative treatment, hemoglobin can be used as a stratication factor to determine the prognosis of advanced cancer patients.


Introduction
Cancer is a major public health problem and the number of cancer patients will increase worldwide, leading to serious health consequences [1] . Advanced cancer poses a great challenge to treatment and is a common cause of death [1] . Palliative care is an essential part of overall cancer treatment [2] . In palliative care, clinicians need to predict the life expectancy of patients with advanced cancer because determining the possibility of impending death aids doctors in clinical decision-making to help patients and their families prepare for the future [3,4] . In the past few decades, several prognostic models have been proposed. The most commonly used prognostic model in palliative treatment is the Clinical Prediction of Survival; however, its inherent non-repeatability limits its accuracy and clinical application [5] . Therefore, more accurate and easily detected indicators to predict the prognosis of patients with advanced cancer are greatly needed.
Hemoglobin (HB) is widely used in clinical examination [6] . Low hemoglobin (HB) levels are common in cancer patients (approximately 30-77%) [7] . Although the number of studies exploring the prognostic role of Hb levels has been increasing, the results are inconsistent and usually based on small samples. Few studies have reported the effect of low hemoglobin on the survival of patients with advanced cancer receiving palliative treatment.

Materials And Methods
Data collection and study cohort Consecutive inpatients who were treated at the Department of palliative care, Shanghai Cancer Center (fuscc), Fudan University, Shanghai, China, were retrospectively enrolled between July 2013 and October 2015. We collected data on patients' age, sex, comorbidities, smoking status, family history, primary tumor location, tumor stage, nutritional status, and physical status (Eastern Cooperative Oncology Group, ECOG score) were obtained from the patient medical records. Hemoglobin levels were measured 1-3 days before the start of palliative care. According to the ESPEN guidelines for nutrition screening, an unintentional weight loss of > 5% in the previous three months or a food intake below 75% of the normal requirement in the preceding week were considered an abnormal nutritional status [8] . The presence of concomitant disease was de ned as self-reported cardiac disease, hypertension, diabetes, or any cerebrovascular disease.
Two cohorts of patients were identi ed in our study [9] . Palliatively treated patients with malignancies in stages III-IV were classi ed as cohort 1, and the completeness of patient data was guaranteed. Patients in cohort 1 were classi ed after readmission to cohort 2. The effect of HB changes on OS was assessed.
The last follow-up date was December 2015. The study was approved by the ethics committee of fuscc.
Informed consent was waived due to the retrospective nature of the study.

Statistics analysis
Data are presented as the mean ± standard deviation (SD) for continuous variables and as totals and frequencies for categorical variables. The distributions of the clinicopathological features were tested using the Wilcoxon sum rank test, chi-squared test, or Fisher's exact test, as appropriate. The Kaplan-Meier method was used to plot the survival curves and the log-rank test was conducted to test differences between groups. Cox proportional hazards models were employed to estimate the magnitude of the association between the clinicopathological features and OS. X-tile version 3.6.1 (Yale University, New Haven, CT, USA) was used to determine the optimal cut-off value. All tests were two-sided, and P values less than 0.05 were considered statistically signi cant. Statistical analyses were performed with SPSS (version 19.0), and R software (version 3.5.3) was used to draw survival curves. Determination of the Hb cut-off value

Patient characteristics
In the clinical environment, it is common to convert continuous variables to binary variables. In this study, the patients were divided into "high" and "low" groups, which was convenient for diagnostic or prognostic predictions. However, for patients with advanced cancer, the accepted clinical threshold was not available. Using X-tile based on a minimum p-value algorithm [10] , the optimal cut-off value for Hb was determined to be 100 g/L. Therefore, the patients with HB levels greater than 100 g/L were assigned to the high HB (HHB) group and the patients with HB levels less than or equal to 100 g/L were part of the low HB (LHB) group.

Association of Hb with clinicopathological features
The relationships between the clinicopathological features and Hb (low vs. high) for cohort 1 are shown in Table 1. There were no differences in age, tumor site, family history, ECOG, nutritional status, or comorbidity between the HHB and LHB groups. In contrast, palliative treatment and gender were signi cantly associated with Hb. The characteristics of cohort 2 are shown in Table 2, where patients were divided into descending (changes < 0) and ascending (changes > 0) groups. There was no signi cant difference in the proportion of patients with speci c clinicopathological features between the two groups.  Figure 3). The related subgroups were further analyzed and signi cant results were obtained after adjusting for demographic and disease-speci c factors (Table 4).

Discussion
Although tremendous progress has been made in the treatment of cancer, the likelihood of treatment failure for patients with advanced disease remains high. Survival assessment is necessary for risk strati cation and further clinical decision-making. However, no signi cant biomarkers have been widely used in survival assessment. Hemoglobin is an indicator measured in almost all hospitals in China. In our study, we evaluated the prognostic impact of Hb on patients with advanced cancer.
This study is the rst to investigate the dynamic changes of hemoglobin in patients with advanced cancer in the palliative treatment environment. We established a Hb cut-off value of 100 g/L and found that patients with a low Hb had a poorer survival time than patients with high Hb with advanced cancer. Furthermore, patients that transitioned from high Hb levels to low Hb levels between the rst and second hospital admission were at a greater risk of a poor prognosis. In conclusion, pretreatment and dynamic peripheral hemoglobin levels represent convenient and intuitive markers to predict aggressiveness in patients with advanced cancer.
Ludwig et al. [11] found that the incidence of anemia in 11,453 patients with various cancers was 41%. In patients with advanced tumors and those receiving radiotherapy or chemotherapy, the incidence of anemia is as high as 90% [12] , especially in patients with more aggressive tumors. This nding may be related to the complex interactions between the immune system, tumor microenvironment, and cancer cells [13,14] . In general, cancer-related anemia has various causes, including blood loss, functional iron de ciency, erythropoietin de ciency due to kidney disease, and the relationship between cancer and bone marrow [15] . Tumor cells can activate macrophages, promote the secretion of in ammatory factors (e.g., gamma-interferon, IL-1, and tumor necrosis factor), which can reduce Hb concentrations by altering the hematopoietic environment [16,17] , inhibiting erythropoiesis and erythropoietin, and impairing the erythropoietin response of erythroid progenitor cells [18,19] . In addition, bone marrow failure is common in patients with bone metastases and leads to low Hb levels [20] . Patients at later tumor stages and greater tumor burden have more in ammatory factors secreted by tumor cells, a higher incidence of anemia, and a worse prognosis [21] . Hypoxia is an important link in the occurrence and development of malignant tumors [22] , while anemia will aggravate the degree of hypoxia in tumor tissues and affect the stability of chemotherapeutic drugs in the process of diffusion. Moreover, the production of oxygen-free radicals reduces the damaging effect of chemotherapeutic drugs on the DNA of tumor cells, thus reducing drug e cacy [23] . Tissue oxygenation disorders stimulate the high expression of angiogenic factors, which promotes the occurrence and development of tumors, reduces drug sensitivity, increases the probability of blood transfusion after chemotherapy, damages organ function, reduces the quality of life, and increases postoperative mortality.
A meta-analysis involving 60 studies showed that anemia increased the overall risk of death in tumor patients by 65% [24] . Previous studies showed that anemia might impact the biological characteristics and prognosis of tumor patients, reduced the response to chemotherapy, and promoted tumor recurrence and metastasis, which is an independent risk factor of poor prognosis [25] . Studies have reported that anemia may or may not be a bad prognostic factor in patients with advanced gastric cancer [26,27] . These results may be related to the inconsistent reference criteria for anemia used in the different studies. The U.S. National Cancer Institute (NCI) and the World Health Organization (WHO) de ne their own anemia diagnosis and classi cation standard. The Chinese Society of Clinical Oncology (notes) is based on clinical practice and the treatment of cancer-related anemia classi cation [28] . The different standards in de ning the normal Hb range and anemia are different for these entities based on slightly different severity classi cations. Therefore, to unify the criteria, we rst established a threshold value for Hb as a tool to predict the prognosis of patients with advanced cancer.
Although this study provides important insights into the relationship between Hb levels and the prognosis of patients with advanced cancer, it has some limitations. This study was limited to a retrospective single-center design with inherent limitations, such as selection bias, possible confounders, and relatively low sample size. Therefore, larger prospective multi-institutional studies with more clinical markers are needed to validate the relationship between Hb and poor prognosis. Overall, we demonstrated for the rst time that Hb is a strong independent predictor of shortened survival after palliative treatment in patients with advanced cancer. As an easily accessible and inexpensive biomarker, Hb deserves further research into future clinical practice for risk strati cation and may enable more accurate clinical decision-making for patients with advanced cancer.

Conclusions
In conclusion,in palliative treatment, hemoglobin can be used as a strati cation factor to determine the prognosis of advanced cancer patients. Patient survival time can be estimated from the changes after