Peripheral blood test provides a practical method for glioma evaluation and prognosis prediction

Summary Objective To investigate the relationship between tumor characteristics and the preoperative counts of immune cells in peripheral blood test in glioma patients. Methods We included 260 WHO grades II‐IV patients who had preoperative peripheral blood test result from Sanbo hospital as training set. The 66 patients from Tiantan hospital was obtained for validation. RNA sequencing data from CGGA and TCGA datasets were used to evaluate the features of neutrophil subtype and lymphocyte subtype in glioma. Results We revealed that the count of preoperative lymphocytes, eosinophils and neutrophils were associated with glioma grades. Neutrophil‐to‐lymphocyte ratio (NLR) <3.2 was associated with better prognosis, whereas increased NLR was strongly corresponding with a poor prognosis. Lymphocyte type glioma patients demonstrated a positive correlation with isocitrate dehydrogenase (IDH) mutation and lower grade. IDH mutant glioma contained a higher proportion of tumor‐infiltrating lymphocytes than IDH wild‐type glioma. The immune subtype (neutrophil subtype and lymphocyte subtype) was an independent prognostic factor in glioma. Conclusion Our data demonstrated that NLR was an important prognostic factor in glioma. We classified that the immune subtype of glioma may contribute to a better understanding of disease pathogenesis and lead to the identification of new therapeutic targets for glioma patients.


Summary
Objective: To investigate the relationship between tumor characteristics and the preoperative counts of immune cells in peripheral blood test in glioma patients.

Methods:
We included 260 WHO grades II-IV patients who had preoperative peripheral blood test result from Sanbo hospital as training set. The 66 patients from Tiantan hospital was obtained for validation. RNA sequencing data from CGGA and TCGA datasets were used to evaluate the features of neutrophil subtype and lymphocyte subtype in glioma.

Results:
We revealed that the count of preoperative lymphocytes, eosinophils and neutrophils were associated with glioma grades. Neutrophil-to-lymphocyte ratio (NLR) <3.2 was associated with better prognosis, whereas increased NLR was strongly corresponding with a poor prognosis. Lymphocyte type glioma patients demonstrated a positive correlation with isocitrate dehydrogenase (IDH) mutation and lower grade. IDH mutant glioma contained a higher proportion of tumor-infiltrating lymphocytes than IDH wild-type glioma. The immune subtype (neutrophil subtype and lymphocyte subtype) was an independent prognostic factor in glioma.

Conclusion:
Our data demonstrated that NLR was an important prognostic factor in glioma. We classified that the immune subtype of glioma may contribute to a better understanding of disease pathogenesis and lead to the identification of new therapeutic targets for glioma patients.

K E Y W O R D S
glioma, molecular markers, NLR, prognosis However, whether the inflammation induced by glioma could trigger body's innate immune system was unclear. Routine blood test, low cost and easy to measure, is a regular examination for patients which give us information reflecting innate immune system. Here, we attempted to explore the relationship between tumor characteristics and routine blood test parameters in glioma.
The first-time preoperative routine blood test, RNA sequencing data, and clinicopathological information of glioma patients were included in this study. We found that patients in highly neutrophilto-lymphocyte ratio (NLR) group significantly had a shorter overall survival. Correspondingly, neutrophil subtype glioma patients significantly had a poor prognostic than lymphocyte subtype glioma patients. The integrated analysis of routine blood test and RNA sequencing data may provide a new perspective for immunotherapy in the future.

| Biomarker detection
Genomic DNA was isolated from frozen tumor tissues or paraffinembedded samples by using the QIAamp DNA Mini Kit (Qiagen).
Pyrosequencing analysis was carried out by Gene Tech (Shanghai) Company Limited. The mutational status of IDH1/2 and the methylation status of the MGMT promoter were determined using pyrosequencing or Sanger sequencing as previously reported. 13  F I G U R E 1 Comparison of preoperative peripheral blood content among three grades gliomas. A, Leukocytes count was different between grade II (N = 69) and IV samples (N = 139, P = 0.0009). B and C, neutrophils, and lymphocytes proportion were different between grade III (N = 52) and grade IV, grade II, and grade IV, respectively. D and E and F, except for eosinophils between grade II and grade IV (P = 0.0002), monocytes, eosinophils, and basophils showed no difference among three grades gliomas (P > 0.05) microscopic fields, we counted more than 1000 tumor cells, of which positively stained cells were reckoned in as well. The ratio of T lymphocytes in tumor was defined as the portion of positively stained cells against total counted cells. The difference was assessed by Mann-Whitney test.

| Statistical analysis
Unpaired t test or Mann-Whitney test were used for comparison of two groups. And the samples distribution was evaluated by Chi-squared test. PFS was defined as the time of surgery until radiographic progression (the appearance of a new lesion or an increase in tumor size of >25%). 14 Overall survival (OS) was defined as time interval from histologic diagnosis to death. Patients who were still alive or lost to follow-up were censored at last follow-up. OS and PFS were estimated with Kaplan-Meier method and compared with log-rank method by using the GraphPad Prism version 6.0 statistical software. Univariate and multivariate cox proportional hazard model were evaluated using coxph function from survival package. A two-sided P value < 0.05 was regarded as significant.

| Peripheral blood immune cells in WHO grade II-IV primary glioma patients
We assessed the count of leukocytes, the count and the proportion of five leukocyte subtypes (neutrophil, lymphocyte, monocyte, eosinophil, and basophil) in peripheral blood according to WHO grades.
As shown in Figure 1 and Figure S1, grade IV patients demonstrated a significant increase in the count of white cells compared with that observed in grade II (P = 0.0009). Neutrophils showed similar trend. The proportion and count of neutrophils in grade IV were significantly higher than grade II (P = 0.0107, P = 0.0007) and grade III patients (P = 0.0036, P = 0.0168). However, the lymphocytes proportion exhibited an inverse tendency, which showed a significant decrease in grade IV than in grade II (P = 0.0034) and grade III patients (P = 0.0026). There was no statistically significant difference was observed among tumor grades of lymphocyte counts. The proportion and counts of eosinophils were obviously higher in grade II than grade IV patients (P = 0.0002, P = 0.0007), and its count was also significantly increased in grade II than grade III (P = 0.0496).

| The prognostic value of preoperative NLR in primary glioma patients
Neutrophil-to-lymphocyte ratio (NLR) is a marker of systemic inflammatory response which exhibited the prognostic and predictive values in various cancers, including colon, prostate 15 and bladder cancer. 16 Here, we investigated the correlation between NLR and overall survival in glioma. NLR was analyzed as a continuous variable. We assessed optimal cutoff dependent on log-rank testing from 3 to 4.5 (Table S1) Figure 2C).
Overall, these data indicated that an elevated NLR significantly correlated with a worse survival outcome.

| The relationship between NLR and molecular markers
IDH mutation, somatic mutation in isocitrate dehydrogenase one and two gene, occurs at high frequency in gliomas and is a prognostic factor in glioma patients. IDH gene mutation was demonstrated with younger, secondary GBMs (sGBMs), and better outcomes. 17 There was a significant different in NLR between IDH mutation group and IDH wild-type group (P = 0.0347, Figure S2A). Patients with NLR <3.2 were more likely to be IDH mutated.
F I G U R E 3 Generation of lymphocyte and neutrophil expression signatures. A, Differential expression analysis between RNA expression profiling of lymphocytes and neutrophils. Red dots, significantly differentially expressed probes (adjusted P value < 1e-10, 661 probes, 520 genes, Table S1). B, expression patterns of the 661 probes F I G U R E 4 Classification of CGGA glioma patients based on lymphocyte and neutrophil signatures. A, RNAseq data of 325 glioma samples from CGGA dataset were clustered based on the signatures. In lymphocyte type (with high expression of lymphocyte signatures), grade II and III, oligodendroglioma and oligoastrocytoma, and IDH mutant samples were enriched, while GBM and IDH wild-type samples were more likely to be clustered into neutrophil type. B, C, lymphocyte type patients showed more favorable prognosis than neutrophil ones (all patients, P < 0.0001; GBM patients, P = 0.0424). The relationship between neutrophil subtype and lymphosubtype and patients' characteristics was evaluated (a, Chi-square test. b, Student's t test) MGMT promoter methylation is a predictive marker for TMZ chemotherapy. 18 We also tested the MGMT promoter methylation status in 190 out of 260 patients by pyrosequencing. There was not significant difference between MGMT promoter methylated group and unmethylated group (P = 0.5434, unpaired t test, Figure   S2B).

| Generation of neutrophil and lymphocyte expression signatures
The NLR parameters based on pre-operation peripheral blood test for differential expression analysis. The lymphocyte and neutrophil signatures were defined as genes differently expressed between lymphocyte and neutrophil (adjusted P value < 1e-10, Table S3).
Finally, 258 probes (229 genes) highly expressed in lymphocytes and 403 probes (291 genes) highly expressed in neutrophils were defined as cell-type specific signatures ( Figure 3A, B).

| Clinicopathological features of lymphocyte and neutrophil type glioma patients
To elucidate the clinical and genetic features of lymphocyte and neutrophil type gliomas, we clustered glioma samples from CGGA and TCGA dataset by k-means (k = 2) method. Tumor-associated macrophage (TAM), playing crucial roles in glioma progression, represent about more than a half of immune cells in tumor mass. 19,20 Here, we used transcriptome data to generate enrichment score with macrophage signatures and depicted the characteristics of macrophage in lymphocyte and neutrophil type gliomas.
In CGGA dataset, lymphocyte type gliomas (N = 173) were high expression of lymphocyte signatures ( Figure 4A). The majority of grade II and III (138/181, 76%), oligocytic (oligodendroglioma and oligoastrocytoma, 97/114, 85%) and IDH mutant (143/167, 86%) tumors were enriched in this subtype, while MGMT promotor methylated conferred F I G U R E 5 Validation of the classification based on lymphocyte and neutrophil signatures in TCGA dataset. A, RNAseq data of 699 glioma samples from TCGA dataset were clustered to validate the classification result. Similarly, Grade II and III, oligodendroglioma, IDH, and ATRX mutant samples were more likely to be clustered into lymphocyte type, while the others were in neutrophil type. B, the lymphocyte type glioma patients had more favorable prognosis than neutrophil ones (P < 0.0001). The relationship between neutrophil subtype and lymphosubtype and patients' characteristics was evaluated (a, Chi-square test. b, Student's t test) in lymphocyte type glioma. Meanwhile, 76% (109/144) GBM and 81% (128/158) IDH wild-type tumors were clustered in neutrophil subtype.
Higher macrophage enrichment scores were more likely to belong to neutrophil subtype glioma. Lymphocyte type glioma patients (median survival not reached, median follow-up 1047 days) had significant longer survival than neutrophil ones (median survival 386 days, P < 0.0001, Figure 4B). For GBM patients, lymphocyte GBMs (median survival 572 days) also showed superior survival than neutrophil ones (median survival 345 days, P < 0.05, Figure 4C).
In TCGA dataset ( Figure 5A than neutrophil ones (median survival 460 days, P < 0.0001, Figure 5B). Furthermore, Cox regression analysis was performed additionally, verifying the independence clinical prognostic value of neutrophil and lymphocyte signature in glioma. In the above two datasets, in univariate cox analysis, it was showed that age, IDH status, 1p19q status and neutrophil and lymphocyte subtype were significantly associated with OS. In multivariate analysis, the immune subtype was also a significant factor after adjusting for the clinical factors mentioned above (Table 1 and Table S2).

| Lymphocyte proportion demonstrated a positive correlation with IDH mutation in glioma
To further validate the relationship between lymphocytes and characteristics of gliomas, we evaluated the lymphocyte proportion and IDH mutation status in 68 glioma samples from Beijing Tiantan Hospital by Immunohistochemistry (IHC) staining. We quantified lymphocyte proportion using CD3 (T-cell marker, the main lymphocyte in gliomas 21,22 ) immunostaining ( Figure 6A). The proportion was defined as lymphocyte count divided by total tumor cells (>1000) in several 400× high cellular fields. The lymphocyte proportion ranged from 0 to 0.04 (median: 0.0075) in IDH wild-type samples, and from 0 to 0.0387 (median: 0.0144) in IDH mutant samples ( Figure 6B).

| D ISCUSS I ON
To our knowledge, this study was the first systematic study that comprehensively revealed the strong association TA B L E 1 Cox hazard regression analyses of clinicopathologic factors and the neutrophil and lymphocyte subtypes for overall survival in CGGA (N = 325)  Neutrophil and eosinophil belong to innate immune cells while lymphocyte belongs to adaptive immune cells. 23 The lymphocyte is the surrogate of an impaired cell-mediated immunity, 24 while neutrophil is acknowledged as a response to systematic inflammation. 25 The recruitment of neutrophil and lymphocyte play a critical component in the pathogenesis of gliomas. 11 Our study demonstrated that the counts of lymphocytes were significantly higher at early stage of glioma and then decreased with tumor progression. The glioblastoma multiforme (WHO IV) possessed the highest extent of neutrophil infiltration and the lowest level of lymphocyte. Moreover, NLR was a prognostic marker in gliomas, which was consistent with findings in multiple other solid tumors. 26  It has been reported that IDH mutation was a frequent genomic alteration in grade II and grade III gliomas but rare in primary glioblastoma. 17 Tumors with IDH mutations had distinct genetic and clinical characteristics, and patients with such mutations had a better outcome than those with wild-type IDH. We hypothesise that IDH mutation glioma has a more heterogeneous microenvironment that can affect the cytokine production which will affect the amount and function of lymphocyte and neutrophils. And the cytokines can contribute to further disease progression by interacting with the inflamed BBB endothelium, and for some cell types, entering the CNS by crossing vascular barriers. Our IHC results found that the extent of tumor infiltration lymphocyte was significantly higher in IDH mutant glioma than IDH wild-type glioma. This phenomenon may potentially indicate that the tumor immune microenvironment of glioma differed in combination with their IDH status. However, the mechanism of the association of IDH mutation and the immunologic tumor microenvironment remained unclear.
We recognized some limitation in our research. First, the peripheral blood immune cells count and NLR can be affected easily, such as chronic diseases, local or systemic infection, previous history of infection (>3 months) and any medication that related to inflammatory condition of patients. Second, limited by the small patient number and the retrospective nature of our study, a prospective study with large number of patients is required.

| CON CLUS IONS
In summary, our data showed that pre-operation NLR was a readily available and effective prognostic factor in glioma patients.
Furthermore, we classified the immune subtype of glioma with neutrophil and lymphocyte signatures, and we found that IDH mutant glioma contained a higher proportion of tumor-infiltrating lymphocytes than IDH wild-type glioma. Although the mechanisms need further studies, we provided a potential, easily accessible, and reliable method for glioma evaluation in clinical practice.

ACK N OWLED G M ENTS
We appreciate the generosity of TCGA network for sharing the huge amount of data.

CO N FLI C T O F I NTE R E S T
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.