Prognostic Significance of Preoperative Albumin to Alkaline Phosphatase Ratio in Patients with Glioblastoma

Objective: To explore the prognostic value of preoperative albumin to alkaline phosphatase ratio (AAPR) in patients with newly-diagnosed glioblastoma (GBM) and its association with clinical characteristics. Patients and methods: A retrospective analysis was carried out on patients with newly diagnosed GBM who had undergone operation at the Department of Neurosurgery at West China Hospital between June 1st 2016 to December 31st 2018. X-tile software was applied to determine the optimal cut-off values for AAPR, neutrophil to lymphocyte ratio (NLR), and albumin. Cox regression analyses were applied to evaluate the prognostic value of AAPR in GBM. PSM analysis was conducted to verify the results. Results: A total of 197 and 154 GBM patients were included in original cohort and PSM cohort respectively. The optimal cut-off value for AAPR, NLR, and albumin were 0.56, 4.55 and 42.2 g/L respectively. High AAPR was only significantly related to longer overall survival (OS) (p=0.010) in original cohort. In PSM cohort, no clinical variable was evidently related to the level of AAPR. AAPR was determined to be an independent prognostic indicator in both original cohort (HR=0.599, 95%CI 0.437-0.822, p=0.001) and PSM cohort (HR=0.649, 95%CI 0.459-0.918, p=0.015). Prognostic models including AAPR had better prognostic accuracy than that including albumin. Conclusion: Preoperative AAPR was determined to be an independent risk factor of prognosis in newly-diagnosed GBM patients, and its prognostic ability was stronger than albumin. And PSM analysis also validated the results.


Introduction
Glioblastoma (GBM) is the most lethal diffuse glioma with a median overall survival of 14-17 months and is characterized histologically by marked cellularity, prominent mitotic activity, abundant vascular proliferation and necrosis [1,2]. GBM is recognized as grade IV tumor by the World Health Organization (WHO), constituting 45.2% of all malignant central nervous system (CNS) tumors and 80% of all primary malignant CNS tumors [3]. Based on the latest 2016 WHO classification of CNS tumors, GBMs are divided into three types, including IDHwildtype, IDH-mutant, and not otherwise specified (NOS) glioblastoma [4]. Surgery is the optimal choice in patients with suspected malignant glioma, and postoperative radiotherapy and chemotherapy are considered to be the first-line adjuvant treatments, which can help prolong lifetime to the maximum extent [1]. In recent years, the biological nature of GBM has been well comprehended by deep DNA and RNA sequencing, coupled with improved techniques that enable richer interrogation of the epigenome [5]. At the same time, cheaper and more convenient methods including neuroimaging, clinical examination, histopathology, etc. are also contributing to a well-rounded understanding of GBM.
Nowadays, cumulative evidence has indicated that some blood parameters are related to tumorigenesis and tumor progression [6][7][8]. Several biomarkers such as neutrophil-to-lymphocyte ratio Ivyspring International Publisher (NLR), platelet-to-lymphocyte ratio (PLR), and fibrinogen are considered to be associated with clinical outcomes [9][10][11][12]. Albumin is one of the important compositions of the blood and constitutes 50% of the plasma. It is responsible for 75% of the plasma oncotic pressure and increases circulating blood volume. Albumin performs a variety of physiological functions, including ligand binding and drugs transport, free radical scavenging, anti-oxidant function, effect on vascular permeability, etc. [13]. The hydrolase enzyme alkaline phosphatase (ALP) is widely expressed in human tissues such as liver, bones and kidneys [14]. Similarly, it plays an important role in human physiological functions, including bone mineralization, vascular calcification, and regulations of immune system [15][16][17]. The albumin to alkaline phosphatase ratio (AAPR), based on serum albumin and alkaline phosphatase, is a useful prognostic indicator. It has been found to have significant impacts on survival in patients with liver cancer [18] and predictive ability on prognosis in various tumors [19][20][21][22].
The relationship between AAPR and GBM has not been explored so far. In the current study, we attempted to figure out the prognostic value of preoperative AAPR in patients with GBM, and investigated its association with clinical characteristics.

Patient Population
A retrospective chart review was carried out on patients with newly diagnosed GBM who had undergone an operation at the Department of Neurosurgery at West China Hospital between June 1 st 2016 to December 31 st 2018. All patients underwent a craniotomy on GBM with gross total resection (GTR) or subtotal resection (STR). These patients were followed up until June 31 st 2020. The pathological diagnoses were based on 2016 WHO classification of CNS tumors.
The inclusion criteria included: 1) older than 18-year-old; 2) underwent resection of GBM by craniotomy with GTR or STR; 3) the pathological diagnoses were based on the latest 2016 WHO classification; 4) intact baseline clinical data; 5) intact preoperative MRI imaging data and postoperative imaging data including MRI and CT within 72 hours after operation; 6) no adjuvant therapy like chemotherapy or radiotherapy before operation; 7) no history of diseases dramatically affecting peripheral blood cells. The exclusion criteria were: 1) younger than 18-year-old; 2) biopsy only; 3) absence of definite pathological diagnosis; 4) incomplete baseline clinical data; 5) absence of preoperative MRI imaging data; 6) receiving adjuvant therapy before operation; 7) presence of history of liver diseases, bone diseases, urological diseases, or infectious diseases shortly before surgery.

Parameters Assessment
Medical records were surveyed and the following clinical data were retrieved: gender, age at operation, preoperative Karnofsky performance status, presence of preoperative seizures, locations and hemisphere of tumors, pathological diagnoses and useful biomarkers. Ki-67 index was tested by immunohistochemistry (IHC), and IDH-1 mutations were determined by both IHC and molecular testing. Routine blood tests were performed within 3 days prior to surgery and relevant data was recorded. The APPR was defined as the ratio between the serum albumin concentration (g/L) and the alkaline phosphatase (U/L), while the NLR was defined by dividing the neutrophil (×10 9 /L) count by the lymphocyte count (×10 9 /L).
Postoperative adjuvant therapies and survival conditions were collected mainly through periodical telephone interview and outpatient follow-up. Patients were routinely followed up every 3 months for the first year, and every 6 months thereafter. Overall survival was defined as the duration from the date of surgery to death or the last follow up. All clinical assessments were performed by two independent qualified neurosurgeons.

Statistical Analysis
SPSS 22.0 (IBM Co., Armonk, NY, USA) was used for all statistical analyses. X-tile software was applied to determine the optimal cut-off values for AAPR, NLR, and albumin [23]. The associations between APPR and clinical variables were tested by chi-square test, Mann-Whitney test, or one-way ANOVA (one-factor analysis of variance). The Cox regression analyses were used to determine the influence of risk factors for overall survival in GBM patients. In Cox regression analyses, a univariate Cox regression was firstly conducted to evaluate clinical variables, then variables with p value <0.1 were included into a backward stepwise multivariate Cox regression for further assessment. R software (version 3.6.3, http://www.r-project.org/) was applied to calculate and compare Harrell concordance index (C-index) and Akaike information criterion (AIC) of prognostic models. A smaller AIC value and/or a larger C-index represented a greater predictive accuracy. A two-sided p-value <0.05 referred as statistically significant difference. Propensity score matching (PSM) analysis was introduced in the current study to adjust for confounding variables and validate the results of the original cohort. The potential confounding covariables included age at diagnosis, gender, preoperative seizures, KPS, hemisphere, location, Ki-67 index, and IDH-1 status. These patients were matched 1:1 using the nearest-neighbor algorithm with a caliper width of 0.2 and without replacement.

Ethics
This study was approved by the Ethical Committee of Sichuan University and conducted according to the principles expressed in the Declaration of Helsinki, and all patients were informed and signed their informed consent to use their data for research purposes.

Relationships between AAPR and Clinical Variables
The relationships between AAPR and other clinical variables were shown in Table 2. In original cohort, high AAPR was only significantly related to OS (p=0.010), and there was no significant association between AAPR and other clinical variables. In PSM cohort, however, there were not any clinical variables evidently related to the level of AAPR.

Prognostic value of AAPR
Univariate and multivariate Cox regressions were performed to further determine the prognostic significance of AAPR. As shown in Table 3    Variables with a p value<0.1 was included in multivariate Cox regression. AAPR (HR=0.599, 95%CI 0.437-0.822, p=0.001) and albumin (HR=0.670, 95%CI 0.484-0.929, p=0.016) were proved to be independent risk factors of OS. Other independent risk factors included age, gender, adjuvant therapy, and ki-67 index, while hemisphere and NLR were not independent prognostic factors.

Comparison of prognostic ability between AAPR and albumin
To further compare the prognostic predictive ability of AAPR and albumin, prognostic models were established by using the two markers combined with other independent variables. In original cohort (Table  5), C-index and AIC were calculated by R software and the results indicated that model AAPR (C-index/AIC, 0.721/1452.24) was superior to model albumin (C-index/AIC, 0.715/1455.95). As the same, in PSM cohort (Table 6)

Discussion
Peripheral blood markers were widely used in monitoring tumor progression and predicting prognosis of tumor patients. In the current study, we found higher AAPR was associated with better OS in patients with GBM. Multivariate analysis indicated AAPR as an independent risk factor for OS. And PSM analysis also verified the results.
The results are consistent with the conclusions of other researches about AAPR to some extent. Previous researches report that AAPR is evidently related to the clinical outcome of solid tumors, namely higher AAPR is significantly associated with better outcomes in survival or therapy of the tumors, such as upper tract urothelial carcinoma, cholangiocarcinoma, non-small-cell lung cancer, hepatocellular carcinoma, nasopharyngeal carcinoma, pancreatic ductal adenocarcinoma, etc. [18-20, 22, 24, 25]. Cut-off values of these studies were in the range of 0.36-0.84, and the cut-off value of AAPR in our research was 0.56. Due to large sample size, results of these different tumor researches are considered reliable.
Albumin is an important nutritional indicator and also associated with the process of systemic inflammation [26,27]. Previous researches indicate albumin served as a prognostic predictor in various kind of tumors, including oral cavity squamous cell carcinoma, non-small cell lung cancer, advanced esophageal cancer, etc. [28][29][30]. Higher albumin level is often related to better clinical outcome. And to enhance the prediction, it is also combined with other blood biomarkers like Glasgow Prognostic Score (GPS) and Controlling Nutritional Status Score (CONUT) [31,32]. In view of the origin of ALP, the correlations between ALP and prognosis of tumor were firstly reported in osteosarcoma, advanced prostate cancer, and hepatocellular carcinoma [33][34][35]. Then, studies of other tumors have found higher peripheral blood APL indicating shorter overall survival or progression-free survival [36][37][38]. This may explain why albumin to alkaline phosphatase ratio has a predictive ability of prognosis of tumors, but the underlying mechanism remains unclear.
Another hot blood biomarker NLR was introduced in our study to serve as a reference. As a marker of systemic inflammation, NLR is supposed to have predictive ability of prognosis in tumor patients based on the theory that inflammatory response plays an important role in tumor development and progression [11]. Most studies indicate that high NLR is related to shorter OS or PFS [9,10,39,40]. In the research of NLR and glioma, the situation becomes complicated. Majority of studies show that high NLR is significantly related to bad prognosis in multivariate Cox regression analyses [41][42][43][44]. Conversely, some researches are unable to show that high NLR predicts shorter survival in GBM [45,46].
The examinations and treatments of GBM are expensive and represent a significant economic burden on health care system all around the world [47]. Compare with liquid biopsy, molecular test, and advanced magnetic resonance imaging (MRI) examination, blood biomarkers such like AAPR, NLR, platelet to lymphocyte ratio (PLR), and lymphocyte to monocyte ratio (LMR) are convenient, cheap, and time-saving. These easily acquired parameters help us understand this disease in more comprehensive perspective and make an accurate judgment on diagnosis and treatment strategies in GBM patients.
There are still some limitations in the current study. At first, the sample size is relatively small so that results of the study could not be further validated using external validation methods. Secondly, postoperative AAPR should be included in the follow-up plan so that we can further evaluate the prognostic value throughout the course of GBM. Thirdly, progression of GBM could not be accurately assessed, some patients didn't even have a chance to be reexamined after operation, so we didn't include this part in our research. Fourthly, this research was just conducted by a single center, and multi-center collaboration should be reached to further verify the results.

Conclusion
To our knowledge, this is the first study focusing on validating the prognostic ability of preoperative AAPR in GBM. In the current study, AAPR is determined to be an independent risk factor of prognosis in patients with newly-diagnosed GBM, and its prognostic predictive ability is stronger than albumin. PSM analysis is also conducted to validate the results. Serum albumin and ALP are simple, affordable and relatively innocuous test that could serve as an objective prognostic parameter for GBM.