Hemoglobin, albumin, lymphocytes and platelets (HALP) score is a useful predictor of survival in patients with recurrent glioblastoma multiforme treated with bevacizumab plus irinotecan

Backgrounds: We aimed to investigate whether the HALP score is a predictive marker in patients with recurrent GBM who were given bevacizumab plus irinotecan. Methods: We compared the survival of patients followed up in our clinic with the diagnosis of recurrent GBM and treated with bevacizumab plus irinotecan, according to HALP score. Results: Median PFS and OS were 4.5 (0.9-14.9) and 8 (0.9-21.3) months, respectively. The median PFS of the low HALP score group was 1.85 (1.3-3.37) months, and of the high HALP score group was 4.96 (0.9-14.9) months (p=0.03). The OS of the high HALP score group (9.63 [7.28-11.9]) was statistically higher compared with low HALP score group (2.26 [0.88-3.65]) (p<0.001). In univariate analysis HALP score was a signicant prognostic factor; patients with low HALP score had a poorer prognosis than high HALP score (HR: 0.063, p<0.001). The multivariate analysis showed that HALP score (p=0.003), and residual tumor (p=0.029) were signicant prognostic factors. In multivariate Cox regression analysis, low HALP score was a signicant poor prognostic factor for OS compared with high HALP score (HR: 0.063, p<0.001). Conclusion: We showed that the HALP score at the start of treatment is an independent prognostic factor for PFS and OS in patients with recurrent GBM treated with bevacizumab plus irinotecan. The HALP score, which can be easily calculated by routine tests before chemotherapy, can be used as a predictive marker for bevacizumab treatment decision.


Introduction
Glioblastoma multiforme (GBM) is the most common malignant primary brain tumor in adults. The standard treatment approach is the combination of adjuvant radiation therapy (RT) with concurrent and adjuvant temozolomide following surgery. The median overall survival (OS) is 15 months, and even with the most appropriate treatment, most patients die within two years [1,2]. GBM is a highly vascularized tumor that requires extensive collection of blood vessels to combat hypoxia. Vascular endothelial growth factor (VEGF)-mediated inhibition of angiogenesis by bevacizumab is a therapeutic strategy for GBM at the center of several clinical studies [3]. A meta-analysis of 14 clinical studies showed that bevacizumab did not improve OS, but improved radiographic response rates and progression-free survival (PFS) in GBM, either as a single agent or in combination with chemotherapy [4]. And it is currently only offered as salvage therapy for treatment-resistant cases. It has been shown that radiological imaging, clinical and laboratory parameters predict bevacizumab response in recurrent GBM [5][6][7].

Methods
Thirty-one patients with a pathologically con rmed diagnosis of GBM treated with bevacizumab plus irinotecan at rst line treatment, were included in the study. The study protocol was approved by the local ethics committee. Medical database of GBM subjects who were admitted to out-patient clinics of our institution between April 2015 and July 2019, has been retrospectively analyzed. Patients who did not receive at least one cycle of bevacizumab plus irinotecan were not included in the study. Since Isocitrate dehydrogenase (IDH) mutation, 1p/19q co-deletion, O6-methylguanine-DNA methyltransferase (MGMT) could not be studied in our center, analysis could not be performed on these. Albumin and hemogram levels before receiving treatment were recorded from the electronic les of the patients. The HALP score was calculated according to the formula: hemoglobin (HB) (g/L) x albumin (g/L) x lymphocytes (/L)/platelets (/L). PFS was de ned as the time from the initiation of bevacizumab plus irinotecan to the rst radiological progression, while OS was de ned as the time from the initiation of therapy to death from any cause.

Statistical Analysis
The optimal cut-off value of the HALP score was analyzed using X-tile software version 3.6.1 (Yale University, New Haven CT, USA). Kaplan-Meier method was used for survival analysis with the log-rank test used to statistical difference. A univariate Cox proportional hazards regression model was used to evaluate the prognostic value of each variable for OS. Multivariate Cox proportional hazards regression models were used to analyze independent prognostic factors. Homogenous variables in study groups according the HALP score were compared by independent samples t-test and expressed as mean ± standard deviation. Non-homogenous variables were compared by Mann-Whitney U test and expressed as median (minimum-maximum). Chi-square test or Fisher exact test was performed to comparison of categorical variables. A p value of <0.05 is considered as statistically signi cant. Data were analyzed by SPSS software. (SPSS 15.0; IBM Inc., Chicago, IL, USA).

Results
Thirty-one patients were included the study, 5 (16%) were female and 26 (84%) were male. The mean age of all subjects was 46.3±12. months, respectively. The cut-off value for the HALP score for OS was 18 in analysis with using X-tile software ( gure 1). Those with a HALP score of 18 and below were de ned as the low HALP score group, and those above 18 were de ned as the high HALP score group. Of the 31 patients, 13 had a low HALP score and 18 had a high HALP score. Age, gender, and ECOG-PS were not statistically different between the two groups (low and high HALP score group) (p=0.76, p=0.1, p=0.12, respectively). There was no statistical difference between the histological type, residual tumor, tumor location, and surgery type low and high HALP score groups (p>0.05 for all, Table 1). The median PFS of the low HALP score group was 1.85 (1.3-3.37) months, and of the high HALP score group was 4.96 (0.9-14.9) months. This difference was statistically signi cant (p=0.03). The OS of the high HALP score group (9.63 [7.28-11.9]) was statistically higher compared with low HALP score group (2.26 [0.88-3.65]) (p<0.001).

Discussion
We found that the HALP score at the beginning of treatment in recurrent GBM patients treated with bevacizumab plus irinotecan is an independent prognostic factor for PFS and OS. PFS and OS were signi cantly shorter for patients in the low HALP score group than those in the high HALP score group.
In ammatory factors have been recognized as an important contributing factor to the complexity and lethality of GBM. Therefore, many studies are carried out considering that in ammatory factors may be prognostic factors for cancers, and the search continues. Previous studies have reported that tumor necrosis factor alpha (TNF-α) and C-reactive protein (CRP) are noticeably higher in patients with glioblastoma compared to healthy controls. However, the relationship between in ammatory factors and glioma risk or prognosis is controversial. There are positive and negative studies on this [18]. A recent study showed that elevations in all in ammatory markers were associated with poor OS in those using bevacizumab in recurrent GBM, but only elevated CRP to albumin ratio (CAR) and platelet to lymphocyte ratio (PLR) were reported to be an independent predictive factor. The prognostic signi cance of the increase in neutrophil to lymphocyte ratio (NLR) has not been demonstrated in this study [19].
As known, while hemoglobin and albumin levels are associated with the nutrition status of the body, lymphocytes and platelets are related to the immune system. The HALP score has been usually used to predict the prognosis of some types of cancers by using the character of showing the two main roles (in ammation and nutrition status) in the prognosis of cancer [16]. Chen et al. rstly described this HALP score to predict the prognosis of gastric cancer. In 1332 patients with gastric cancer who underwent gastrectomy, the preoperative HALP ≥ 56.8 group had a signi cantly better prognosis than the HALP < 56.8 group [9]. Also, the prognostic effectiveness of the HALP score has been investigated in, esophageal squamous cell cancer [10], pancreatic cancer [11], colorectal cancer [12], renal cell carcinoma [13], bladder cancer [14], prostate cancer [15,16], and small cell lung cancer [17,18]. Low HALP score was associated with worse survival outcomes in all studied cancers. However, the optimum cut-off value was different in each study, so it is very important to de ne the optimal cut-off value of HALP score.
Bevacizumab, a humanized monoclonal IgG1 antibody, acts to bind to human VEGF and inhibit its activity [20]. The objective response rate (ORR) achieved with the combination of bevacizumab plus irinotecan in recurrent GBM is 38%, and the median PFS and OS are 5.6 and 9.2 months, respectively [21].
Predictive factors of bevacizumab response in recurrent GBM have been investigated in several studies. In these studies, it has been shown that the development of hypertension and proteinuria, high eosinophil and lymphocyte levels, and low CAR and PLR values due to bevacizumab treatment are predictive for better survival [6,7,19].
In conclusion, this is the rst study to show that the HALP score predicts the response to bevacizumab plus irinotecan treatment in patients with recurrent GBM. This score will be useful because it can be easily calculated with routine blood tests, is cheap and objective. We think that it can be used in routine practice if the same result is obtained as a result of studies with a larger number of patients.

Declarations
Funding: Nil Con icts of interest/Competing interests: All authors state that they have no con ict of interests to declare Funding: This research did not receive any speci c grant from funding agencies in the public, commercial, or not-for-pro t sectors

Con icts of interest/Competing interests: None
Ethics approval: The study protocol was approved by the Necmettin ErbakanUniversity Faculty of Medicine ethics committee (Approval no:2020/2749).   Figure 1 Cut-off values for the HALP score (hemoglobin, albumin, lymphocyte, and platelet) by X-tile software version 3.6.1.