High neutrophil, low lymphocyte and hemoglobin unfavorably impact survival in non-small cell lung cancer patients with brain metastases


 Background

Brain metastases (BM) from NSCLC has emerged as an increasingly corresponding clinical problem. Precise prognostic evaluation is the basis for personalized medicine. This study sought to investigate prognostic values of clinical and hematological indicators for NSCLC patients with BM in the real world, which could further help guide survivorship care in the actual clinical setting and clinical trials.
Materials and Methods

We retrospectively reviewed the clinical and hematological indicators of NSCLC patients with BM treated with whole-brain radiotherapy. Receiver operating characteristic curve was performed to evaluate the optimal cut-off point. Kaplan–Meier survival analysis and Cox regression analyses were used to evaluate survival.
Results

105 patients were included and median survival was 21 months (range: 1–64 months). Univariate analyses demonstrated that favorable survival was associated with resection history of NSCLC (P = 0.015), absent of intracranial symptom (P = 0.044), lymphocyte ≥ 1.54*109/L(P < 0.001), neutrophil < 4.64*109/L (P = 0.016), hemoglobin ≥ 117.5 g/L (P < 0.001), BSBM scores of 2–3 (P = 0.033) and Lung-molGPA scores of 2.5-4 (P < 0.001). Cox regression analysis showed that lymphocyte (HR 3.390, 95% CI 1.869–6.151, P < 0.001), neutrophil (HR 0.517, 95% CI 0.286–0.934, P = 0.029), hemoglobin (HR 3.215, 95% CI 1.748–5.911, P < 0.001), resection history of NSCLC(HR 2.813, 95% CI 1.375–5.754, P = 0.005), intracranial symptom(HR 0.251, 95% CI 0.113–0.561, P = 0.001), and Lung-molGPA(HR 2.317, 95% CI 1.186–4.527, P = 0.014) were independent prognostic factors for NSCLC patients with BM.
Conclusions

High neutrophil, low lymphocyte and hemoglobin, absent of resection history of NSCLC, present of intracranial symptom, and Lung-molGPA scores of 0–2 may provide valuable information for indicating poor prognosis in NSCLC patients with BM .

univariate analyses were included in the multivariate survival analysis using the Cox proportional hazard model. All analyses were performed using SPSS Version 24.0 and GraphPad Prism Version 6.0. A twosided p-value < 0.05 was considered statistically signi cant. Table 1 lists the clinicalpathologic characteristics of all patients. 105 patients (male 57/105 (54.3%); female 48/105 (45.7%)) with newly diagnosed NSCLC BM were available for further analysis. Median age at diagnosis of BM was 59 years (range 35-79 years). All patients was treated with WBRT, while 10/105 (9.5%) patients were treated with resection of BM as initial therapy for BM. Of these, there are more patients diagnosed with singular BM (80% vs. 20%) and BM ≥ 3 cm in diameter (80% vs. 20%) compared to patients performed radiotherapy as initial therapy for BM (χ2 = 8.132; P = 0.01; χ2 = 9.989; P = 0.002). In term of the size of BM in diameter, less patients were suffered from intracranial symptom among patients diagnosed with BM < 3 cm in diameter (40/59; 67.8%) compared to patients diagnosed with BM > 2 cm in diameter (25/27; 92.6%; χ2 = 6.171; P = 0.013).

Relationship between lymphocyte, neutrophil, hemoglobin and clinicopathologic features in patients with BM and NSCLC
We assessed various hematological indicators in our patient cohort. The statistically determined appropriate cut-off values based on the results of ROC curve analysis were as follows: 1.54*10 9 /L for lymphocyte, 4.64*10 9 /L for neutrophil, 117.5 g/L for hemoglobin, 94.85fL for MCV, and 289*10 9 /L for platele ( Fig. 1).
Neutrophil was increased (higher than 4.64) in 49.5% (52/105) of patients (Table 2). Median neutrophil was 5.68 (range, 0.20-51. 6) in pretreated patients. Gene status was associated with neutrophil, patients with mutation of epidermal growth factor receptor (EGFR) and anaplastic lymphoma kinase (ALK) had a signi cantly lower neutrophil than wild-type patients (χ2 = 6.011; P = 0.014). Furthermore, a signi cant association between NLR and prognostic scores for BM was observed. In term of Lung-molGPA (χ 2 = 5.658; P = 0.017), patients with low neutrophil may have a better prognosis than patients with high NLR.

Relationship between hematological indicators
We observed a linear correlation between platele and lymphocyte or neutrophil in patients with BM and NSCLC ( R 2 = 0.067, P = 0.008; R 2 = 0.124, P < 0.001). We found a similar correlation between hemoglobin and lymphocyte or MCV ( R 2 = 0.060, P = 0.012; R 2 = 0.073, P = 0.005), as shown in Fig. 2 Fig. 3B). This phenomenon may be explained by the consumption of the tumor and the treatment-related blood toxicity. Moreover, according to RTOG-RPA(P < 0.001; Fisher Exact test) and BSBM (χ 2 = 9.238; P = 0.002), patients with asynchronous diagnosis may have a better prognosis than patients with synchronous diagnosis, which requires further survival analysis to com rm.

Overall survival
All 105 patients were strictly followed up, and the median follow-up time was 22 months (range: 1-64 months). In the entire cohort, mOS was 21 months (range: 1-64 months) and the 1-year, 2-year and 5year survival rates were 69.3%, 47.2% and 16.3%, respectively. Median survival from BM diagnosis was no statistically signi cant difference in patients with asynchronous diagnosis (22 months) compared to patients with synchronous diagnosis (21 months; p = 0.885; Table 4; Fig. 4A).
There was a associaton with survival was found when patients were divided into lymphocyte low and high groups by the cut-off value of 1.54. Patients with elevated lymphocyte had signi cantly longer OS, 37 months vs. 14months (P < 0.001; Fig. 4B). Moreover, when patients were divided into low neutrophil and high neutrophil groups by the cut-off value of 4.64, patients with low neutrophil showed longer OS compared with those with high neutrophil (mOS, 28months vs. 15months, P = 0.016; Fig. 4C). Similarly, high hemoglobin was also signi cantly correlated with better prognosis(mOS, 27months vs. 11months, P < 0.001; Fig. 4D). While, other laboratory parameters-MCV and platele did not signi cantly correlated with patient survival (Table 4).
Data needed for RTOG-RPA, BSBM and Lung-molGPA calculation including age, KPS, and number of BM were signi cantly associated with survival. However, non-signi cant differences were observed between ECM, gene status, primary tumor status and OS (Supplementary Table S1).

Discussion
The selective barrier function of blood-brain barrier (BBB) has ability to prevent in ammatory cells, immunoglobulin and cytotoxic substances in the blood from entering the brain, resulting in immunosuppression of tumor microenvironment (TME) in early brain tumors. However, BM could break the strict "immune-tolerant" environment by disrupting the BBB, and then allowing in ammatory cells to exert anti-tumor effects. In ammatory cells, including neutrophils, lymphocytes and so on, have been identi ed and validated the association with prognosis for BM. Recently, several studies have reported that the high NLR before SRS-SRT, operation or immunotherapy inversely predicts OS in patients with BM, which was detected in circumstances of the high neutrophil and the low lymphocyte (18)(19)(20). Similar results have been observed in our study, in which lymphocyte and neutrophil collected before WBRT, were independent prognostic factors for BM from NSCLC. While the biologic basis of the phenomenon is not thoroughly comprehend. It may interrelate with elevated neutrophil-dependent in ammation and decreased lymphocyte mediated anti-tumor response. In ammation, serving as a hallmark of cancer, has been reported to promote tumorigenesis and progression of cancer(21). Circulating neutrophils, mediated systemic in ammation, has been shown to promote DNA damage response, angiogenesis and tumor invasion by secreting tumor necrosis factors, vascular endothelial growth factors and some cytokines(22). Neutrophilia as an in ammatory response inhibits the cytolytic activity of lymphocytes, activated T cell and natural killer cells, which could promote cancer progression and affect clinical outcome (23) . Lymphocyte plays a crucial role in host anti-tumor immunity response, and could eradicate tumor cells by cytokine secretion and cytotoxic cell death. Worth to mention, the post-treatment NLR was associated with poor prognosis as well(24, 25). While the biologic basis of the relationship between posttreatment NLR and success of treatment is quite complex. It is true that lymphopenia is correlated with a poor prognosis, while a preclinical study found that activated T cells could facilitate BM through inducing an elevated expression of Guanylate-Binding Protein 1 (GBP1) by the cancer cells(26). Moreover, neutropenia increase the risk of infections. In term of our median values of neutrophil and lymphocyte within the normal range, we consider that the prognostic values of neutrophil and lymphocyte should be within a certain range, and the the biologic basis requires further research.
In addition to in ammatory cells, hemoglobin is regarded as a both nutritional and in ammation-related indicator, which has a signi cant prognostic relevance of BM and NSCLC(27, 28). Our results are consistent with the prognostic effect of hemoglobin for survival in NSCLC BM. In a systematic review and meta-analysis, even if the continuous variable hemoglobin level was in the normal range, a decreased hemoglobin level was signi cantly related to worse survival for lung cancer patients(29). The mechanism underlying the prognostic value of hemoglobin can be explained from several perspectives. Hemoglobin reduction causes hypoxia of tumor, which then promotes tumor growth and resistant to chemoradiotherapy by modulating the gene expression and cell-cycle position, subsequently leads to the cancer progression and poor survival(30). The potential cause of these differences remains unclear, the determination and the best cut-off values of neutrophil, lymphocyte and hemoglobin should enroll in further randomized controlled trials to con rm their relevance in the incidence of BM and prognosis, which may help build a easy-to-calculate laboratory score for BM from NSCLC. Additionally, previous reports indicated that the serum markers such as peripheral blood cell counts can be used as a predictor for response to immunotherapy(31). Therefore, the in-depth study about prognostic values of neutrophil, lymphocyte and hemoglobin will help us to further understand the immune mechanism of NSCLC BM, and ultimately help us to make better clinical decisions in circumstances such as facing moderate-tosevere irAEs and distinguishing tumor ares (or pseudoprogressions) from true progressions.
Among patients with asynchronous diagnosis, patients originally treated with surgical resection for NSCLC experienced prolonged BMFS and OS. Our ndings are well in line with other studys com rming the clinical signi cance of surgical resection for NSCLC(32). While randomized controlled data has not suggested that surgical resection provides a signi cant survival bene t for all stage NSCLC patients, present nationwide guidelines nevertheless underline the signi cance of offering surgical resection for resectable patients given potential chance for cure. However, the choice of surgical approach and surgical intervention is still debatable(33, 34). We could await direction from further studies.
In our study, we have con rmed that patients with neurological symptom presented with poor prognosis. Besides, the Neurologic Assessment in Neuro-Oncology (NANO) scale objectifying the symptomatic burden, has been a standard part in the response assessment of primary brain tumors(35, 36). While symptom evaluation has not been included in the prognostic assessment in newly diagnosed BM so far. Moreover, we have con rmed the utility of the Lung-molGPA score.
Although we were able to investigate potential prognostic indicators for NSCLC patients with BM in a reallife cohort, some limitations have to be considered in the explanation of our results, including its retrospective study, relatively small sample size and heterogeneous patient characteristics. Because our data was collected in the radiotherapy department, this study only included patients receiving WBRT for BM to avoid selection bias. Both before and after the diagnosis of BM, the treatment strategy varied widely thus precluding the capability to assess the impact of these agents on the laboratory indicators and study patients. Moreover, the predictive versus prognostic value of clinical factors and hematological indicators evaluated cannot be determined owing to lack of a control arm. For these reasons, our results ought to be considered hypothesis generating.

Conclusion
In summary, our data show that neutrophil, lymphocyte and hemoglobin before WBRT were independent prognostic factors for NSCLC patients with BM. Furthermore, our data stressed the importance to evaluate symptomatic and asymptomatic patients separately because of the differing survival prognosis. Further validation in larger cohorts and prospective studies is required, which may help clinical decisionmaking and patient risk strati cation for NSCLC BM.

Consent for publication
Not applicable.

Availability of data and material
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare no potential con icts of interest.  *Epidermal growth factor receptor(EGFR) and anaplastic lymphoma kinase (ALK); #Synchronous diagnosis, defined as diagnosis of primary NSCLC and BM within 30 days; Abbreviations: a) AC, adenocarcinoma; b) NAC, non-adenocarcinoma; c) NSCLC, non-small cell lung cancer; d) ECM, extracranial metastases; e) BM, brain metastases; f) RTOG-RPA, Radiation Therapy Oncology Grouprecursive partitioning analysis; g) BSBM, proposed basic score for brain metastases; h) Lung-molGPA, Graded Prognostic Assessment for Lung Cancer Using Molecular Markers.   Abbreviations: a) AC, adenocarcinoma; b) NAC, non-adenocarcinoma; c) NSCLC, non-small cell lung cancer; d) BM, brain metastases; e) MCV, mean corpuscular volume; f) RTOG-RPA, Radiation Therapy Oncology Grouprecursive partitioning analysis; g) BSBM, proposed basic score for brain metastases; h) Lung-molGPA, Graded Prognostic Assessment for Lung Cancer Using Molecular Markers; i) 3DCRT, three-dimensional conformal radiotherapy; j) IMRT, intensity-modulated radiation therapy.   Linear correlations exist between hematological indicators in 105 NSCLC patients with brain metastases. A, correlation between lymphocyte and hemoglobin in 105 NSCLC patients with brain metastases. B, correlation between mean corpuscular volume (MCV) and hemoglobin in 105 NSCLC patients with brain metastases. C, correlation between neutrophil and platele in 105 NSCLC patients with brain metastases. D, correlation between lymphocyte and platele in 105 NSCLC patients with brain metastases.

Figure 3
Relationship between hemoglobin and brain metastases free survival (BMFS) in 105 NSCLC patients with brain metastases. A, relationship between hemoglobin and synchronous diagnosis. B, relationship between hemoglobin and BMFS.