Galectin-1 is a poor prognostic factor in patients with glioblastoma multiforme after radiotherapy

This study was approved by the Institutional Review Board of Chang Gung Memorial Hospital (102-5087B). A total of 63 newly diagnosed GBM patients who were treated at our institution before January 1, 2002, were retrospectively evaluated. Different radiation doses may affect prognosis in GBM patients; Bleehen et al. reported that a survival advantage for patients with grades 3 and 4 astrocytoma was maintained with the high dose of 60 Gy [17]. The current standard dose for GBM at our institution is also 60 Gy, which is according to NCCN guidelines. Hence, we excluded patients with an equivalent dose in 2-Gy fractions (EQD2) < 60 Gy for homogeneous treatment due to poor outcome [17] to exclude the dose effect, which might mask the importance of galectin-1 for prognosis, and selected those with EQD2 ≥ 60 Gy. A total of 45 patients remained and were treated with maximal safe surgical resection or biopsy only followed by adjuvant RT of EQD2 ≥ 60 Gy for homogeneous treatment; the outcomes of these patients were analyzed (Table 1). We chose age ≥ 50 years as a cutoff based on a recursive partitioning analysis (RPA) study [18]; we chose a tumor size ≥ 5 cm as a cutoff because this size is a significant prognostic factor after RT [19]. Brain image analysis by magnetic resonance imaging (MRI) or computed tomography (CT) was performed before treatment. The tumor size was measured in the longest diameter on a T1-weighted axial contrast-enhanced image by brain MRI or axial contrast-enhanced image by brain CT. For multifocal GBM, the sum of the largest axial diameter for all lesions was used to determine the tumor size.

Patients were assigned by the radiation oncologist either for whole-brain irradiation (19 patients) followed by tumor bed boost or for tumor-site-only irradiation with a small field throughout the entire RT course (26 patients). For the latter, 11 patients were treated with a field-in field (FIF) boost protocol with RT that consisted of daily fractions of 180 cGy applied to a large target volume at the brain tumor site. This was followed by the subsequent application of 70 cGy (total 250 cGy daily) to a reduced field of the tumor bed, for a total dose of 6250–6750 cGy. The remaining 15 patients received conventional RT with 180–200 cGy per day. All RT protocols were administered in five fractions across the week.

For whole-brain irradiation, the photon beams at 6 MV or 10 MV were delivered via two-dimensional RT (2DRT). For tumor-site-only irradiation, three-dimensional conformal RT (3D–CRT) was used. The gross tumor volume (GTV) was defined according to the contrast-enhancing tumor on the CT or MRI images and included the residual tumor, perifocal edema, and entire resection cavity. The GTV plus a safe margin of 2 cm was the clinical target volume (CTV). In the FIF boost protocol, the CTV + 1 cm margin was irradiated as the large target volume, and the CTV alone was irradiated as the boost target volume. For 3D–CRT, the CTV + 3 mm margin was the planning target volume (PTV) for daily setup variation. In the FIF boost protocol, the CTV + 1.3 cm margin was the PTV of the large target.

Areas showing the histopathologic features of GBM were selected on archival hemolysin and eosin (H&E)-stained sections, and then representative areas of the tumor were marked on the corresponding paraffin block for tissue microarray (TMA) construction. Briefly, after the tissue cylinders were taken from the selected regions of the donor paraffin block, they were punched precisely into a recipient paraffin block using a tissue-arraying instrument. Multiple sections (1 μm thick) were cut and mounted onto microscope slides. The TMA sections were evaluated by the pathologist (SLY), who did not know the outcomes of the patients.

TMA sections were stained with an anti-galectin-1 antibody (1:40, HPA000646; rabbit to human Sigma-Aldrich, St. Louis, MO) using a DAKO REAL EnVision (DAKO, Glostrup, Denmark) and an anti-galectin-3 antibody (Santa Cruz B2C10). The immunostaining patterns for galectin-1 and -3 in the intracellular tumor portion and extracellular stromal portion of the tissue samples were recorded. IHC staining was graded as no, weak, moderate, or strong staining according to the observed intensity (Fig. 1). The staining areas were calculated by the pathologist according to the staining percentage of all tumor cells in each TMA core.

The expression level of galectin-1 can be calculated by multiplying the amount of positively stained tumor cells (%) by the staining intensity of immunoreactive tumor cells. The staining intensity of immunoreactive tumor cells was scored as follows: 0, no staining; 0.5, weak staining; 1, moderate staining; and 2, strong staining (Table 2).

We primarily used a single grader for the immunostaining results. To dramatically reduce intraobserver differences, reference to standard slides, which revealed the reference point of the staining score, was performed before each reading. The grader was completely unaware of the outcome of the patients, and thus bias of the scoring interpretation was negligible. Normal liver tissue was employed as a negative control for galectin staining was and melanoma as a positive control.

The first brain image analysis, MRI or CT, was obtained for all patients within 2–4 months after the completion of RT and then every 3–6 months or when clinically indicated, such as when new neurologic signs were observed. The optimal cutoffs of galectin-1 and galectin-3 expression for median survival were estimated by univariate analysis with the log-rank test; those for OS, cancer-specific survival (CSS) and progression-free interval were estimated by multivariate analysis with a Cox regression model. OS and CSS were measured beginning on the first day of RT. OS was measured until the date of death from any cause or the last follow-up date, and CSS was defined as a survival measure representing cancer survival until the date of death (in the absence of another cause of death). The progression-free interval indicated the time interval from the first day of RT until the date of tumor progression, as confirmed by follow-up brain CT or MRI or by death.

We attempted to determine the optimal cutoffs of galectin-1 and galectin-3 expression levels using univariate analysis (Additional file 1). Patients with weak galectin-1 expression (< 35%) had a better median survival (27.9 months) than did those with strong (≥ 35%) galectin-1 expression (10.7 months; p = 0.009). Among the 45 patients included, 34 showed strong galectin-1 expression and 11 weak galectin-1 expression. In addition, patients with weak galectin-3 expression (< 15%) had a better median survival (12.1 months) than did those with strong galectin-3 expression (10.7 months; p = 0.031). We compared characteristics between weak and strong galectin-1 expression (Table 1) and noted no significant difference. The 3-year OS was 27.3% and 5.9% (p = 0.009) (Fig. 2) in patients with weak and strong galectin-1 expression, respectively, and the corresponding CSS was 27.3% and 7.6%, respectively (p = 0.020) (Fig. 3).

We applied Cox regression to confirm the role of galectin-1 and again found it to be a significant factor of OS (p = 0.020; HR = 2.929; 95% CI 1.180–7.271) (Table 3). Other parameters, such as Eastern Cooperative Oncology Group (ECOG) performance, gender, surgical method, age ≥ 50 years, tumor size ≥ 5 cm, radiation field, and expression level of galectin-3, were not significant factors. CSS was also independent of galectin-1 overexpression (p = 0.022; HR = 2.873; 95% CI 1.163–7.101) (Table 4), as was the progression-free rate (p = 0.037; HR = 6.080; 95% CI 1.113–33.219) (Table 5).