Genetic factors affecting intraoperative 5-aminolevulinic acid-induced fluorescence of diffuse gliomas

The Research Ethics Committee of the Faculty of Medicine, University of Miyazaki, approved this study; prior informed consent for inclusion in the study was obtained from all patients. They underwent tumor removal at Miyazaki University Hospital during the period from January 2014 to December 2015. We collected 71 consecutive glioma patients. Our inclusion criteria were surgery under 5-ALA-induced fluorescence guidance, a preoperative MRI evaluation, and a histological diagnosis of astrocytic or oligodendroglial tumors based on the WHO 2007 classification.²⁰ Excluded were patients with WHO grade I tumors and/or needle biopsy only. In patients who had undergone surgical resection more than once during the period, data from the first resection were used. Of the 71 patients, 60 (35 men, 25 women; age range 6-80 years, mean age 60.7 years) with astrocytic or oligodendroglial tumors fulfilled our selection criteria; 54 (90.0%) harbored newly diagnosed gliomas. The patient characteristics, histological diagnosis, and tumor location are shown in Table 1. Of the included patients, 35 (58.3%) had WHO grade IV, 17 (28.3%) WHO grade III, and 8 (13.3%) had WHO grade II gliomas. The tumor location was the frontal lobe (22 patients, 36.7%), followed by the temporal lobe (9 patients, 15.0%). In 8 patients (13.3%) the tumor was centrally located in the basal ganglion, corpus callosum, and brain stem; none of the tumors were occipital.

All preoperative MRI studies were performed within a week before surgery on a 3 Tesla scanner (Magnetom Verio; Siemens, Erlangen, Germany). T1- and T2-weighted- (T1W, T2W), and contrast-enhanced T1W axial images were used for analysis. The images were assessed consensually by two neuroradiologists blinded to the genotype and the 5-ALA-induced fluorescence of each lesion. They used methods described elsewhere²¹ to qualitatively assess the following findings: a sharp or indistinct tumor border, homogeneous or heterogeneous signal intensity throughout the tumor on T1W and T2W images, and the presence or absence of contrast enhancement on contrast-enhanced T1W images (Figure 1). Identification of the predominant characteristics of individual tumors was based on the readers’ judgment of the tumor border and signal heterogeneity.

5-ALA (20 mg/kg) was orally administered 3 hours before surgery. For the operation we used an OPMI Pentero instrument (Zeiss, Oberkochen, Germany) equipped with BLUE 400 for visualizing fluorescence and for neuro-navigation. All resections were performed with navigational guidance using contrast-enhanced axial, coronal, and sagittal T1W- or fluid-attenuated inversion recovery (FLAIR) images. The targets for tissue sampling were selected by choosing contrast-enhanced lesions and neuronavigation, or the tumor center of high-intensity lesions on FLAIR images. Fluorescence was checked repeatedly under our modified neurosurgical microscope by switching between white- and blue-violet excitation light in different areas of the lesion during each procedure. Fluorescence was categorized subjectively by two operating surgeons; a 3^rd surgeon confirmed their judgment by reviewing a movie obtained intraoperatively. Fluorescence was categorized as non-visible (negative) and visible (positive).^19,22 Positive fluorescence included mild (pink) and robust brightness (lava-like orange)¹⁸(Figure 2).

To diagnose the tumors histopathologically, neuropathologists used the 2007 WHO classification of central nervous system tumors.²⁰ They were blinded to intraoperative 5-ALA fluorescence. Tumor cell proliferation was assessed immunohistochemically using MIB-1 antibody (anti-Ki-67, 1:50; DAKO, Hamburg, Germany). The highest density of Ki-67 immunopositive cells was evaluated in hot spots. The percentage of immunolabeled tumor cell nuclei was expressed as the MIB-1 labeling index.

IDH1 mutation was confirmed by immunohistochemical analysis using R132H antibody²³ or by direct sequencing. Immunostaining was according to the manufacturer’s protocol. Briefly, formalin-fixed paraffin-embedded tissue was sliced into 2-μm sections and dried at 42 -o $\overset{\text o}{\text -}$ C for 3 hours. Deparaffinization was with xylene, rehydration was by submerging the tissue samples in graded series of ethanol (100% to 70%, decreased in 10% steps). Phosphate-buffered saline (PBS) was used for washing. For antigen retrieval, slides were pretreated by steaming with citric acid buffer (pH 6.0) for 20 min in an autoclave. The sections were then treated with 10% H₂O₂ for 10 min at room temperature to block endogenous peroxidase activity. After washing in 3 changes of PBS for 5 min, the sections were immunostained with monoclonal anti-human IDH1-R132H antibody (1:20 dilution, H09, Dianova, Hamburg, Germany) diluted with PBS, and incubated at 4 -o $\overset{\text o}{\text -}$ C overnight. After being washed in 3 changes of PBS buffer, the tissues were covered by anti-mouse specific second antibody administered in 1 - 3 drops per slide (Dako EnVision, Hamburg, Germany), and held for 30 min at room temperature. After being washed in 3 changes of PBS buffer, the staining reaction was performed by covering the tissue with a prepared 3, 3’-diaminobenzine (DAB) chromogen solution (1 drop of DAB chromogen for every 1000 μl of PBS). This was followed by incubation for approximately 40 sec to allow for proper brown color development. A definite cytoplasmic immunoreaction product was scored as staining cells. Staining was scored on a two-grade scale as negative (no or < 10% staining), and as positive (> 10% positively stained cells).

Direct DNA sequencing was as previously described.^23,24 IDH1 genomic DNA was isolated from frozen tissue samples with a QIAamp DNA Mini Kit (QIAGEN, Tokyo, Japan). A spanning 90-base pair (bp) fragment was identified with the sense primer IDH1 (forward: 5’-GGCTTGTGAGTGGATGGGTA-3’) and the antisense primer IDH (reverse: 5’-GCAAAATCACATTATTGCCAAC-3’). The 25-μl reaction mixture contained 50 ng of tumor genomic DNA, 1 μl of each forward and reverse primer (10 μM), 12.5 μl of GoTaq Hot Start Green Master Mix (Promega K.K., Tokyo, Japan), and an amount of deionized water to obtain a total volume of 25 μl. Genomic DNA was subjected to polymerase chain reaction (PCR) amplification, initial denaturation at 95 -o $\overset{\text o}{\text -}$ for 2 min, and 35 cycles of amplification consisting of denaturation at 95 -o $\overset{\text o}{\text -}$ for 30 sec, annealing at 58 -o $\overset{\text o}{\text -}$ for 30 sec, and extension at 72 -o $\overset{\text o}{\text -}$ for 20 sec. Final extension was performed at 72 -o $\overset{\text o}{\text -}$ for 5 min. The 100-bp PCR amplification products were confirmed by 2% agarose gel electrophoresis.

The sequence reactions were performed by using a Big Dye Terminator v1.1 Cycle Sequencing Kit (Thermo Fisher Scientific K.K., Yokohama, Japan) in a 20-μl reaction mixture comprised of 1 μl of the 100-bp PCR products, 4 μl of PCR buffer, 2 μl of the forward primer (5 μM), 12 μl of deionized water and 1 μl of Big Dye Terminator Ready Reaction Mix. Initial denaturation was performed at 96 -o $\overset{\text o}{\text -}$ for 1 min and 25 cycles of amplification (denaturation at 96 -o $\overset{\text o}{\text -}$ for 30 sec, annealing at 50 -o $\overset{\text o}{\text -}$ for 5 sec, and extension at 60 -o $\overset{\text o}{\text -}$ for 4 min). Sequencing products were immediately submitted to direct sequencing on an ABI PRISM 310 Genetic Analyzer (Thermo Fisher Scientific K.K., Yokohama, Japan).

FISH analysis of 1p19q loss of heterozygosity (LOH) was performed on formalin-fixed paraffin-embedded 5-μm tissues²⁵ prepared for dual-probe hybridization with Vysis LSI FISH Probe according to the manufacturer’s instructions (Abbott Japan Co. Ltd.). 1p36/1q25 and 19q13/19p13 dual-color probe sets were used for locus-specific 1p and 19q analysis, respectively, following the manufacturer’s instructions (Abbott Japan). Nuclei were counterstained with 4,6-diamidino-2 phenylindole (DAPI).

Analysis of the methylation status of the MGMT promoter was performed by bisulfite modification and subsequent methylation-specific PCR assay using previously described primers.⁸ The primer sequences for the unmethylated reaction were forward: 5’-TTTGTGTTTTGATGTTTGTAGGTTTTGT-3’ and reverse: 5’-AACTCCACACTCTTCCAAAAA CAAAACA-3’. Sequences for the methylated reaction were forward: 5’-TTTCGACGTTCGTAGG TTTTCGC-3’ and reverse: 5’-GCACTCTTCCG AAAACGAAACG-3’. The PCR conditions were 35 cycles of 30 sec each at 95 -o $\overset{\text o}{\text -}$ , and 60 -o $\overset{\text o}{\text -}$ , and 60 sec at 72 -o $\overset{\text o}{\text -}$ ; the PCR products were electrophoresed on 15% polyacrylamide gels as previously described.²⁶

All numeric data were reported as the mean ± standard deviation. The positive predictive value (PPV) of 5-ALA fluorescence for high-grade gliomas was calculated as the number of 5-ALA fluorescence-positive high grade gliomas / number of all 5-ALA fluorescence-positive tumors. Its negative predictive value (NPV) was calculated as the number of 5-ALA fluorescence-negative low-grade gliomas (WHO grade II) / number of all 5-ALA fluorescence-negative tumors. The diagnostic accuracy of 5-ALA fluorescence for high-grade gliomas was calculated as the number of 5-ALA fluorescence-positive high-grade gliomas plus the number of 5-ALA fluorescence-negative low grade gliomas / the number of all tumors.

Univariate and multivariate logistic regression analyses were used to identify clinical characteristics and genetic- and imaging features associated with the 5-ALA-induced fluorescence of the lesions. Univariate analysis was with the χ²_-, the Fisher exact-, or the Student t-test. Variables with p < 0.05 by univariate analysis were used for multivariate analysis. In multivariate logistic regression analysis we computed the odds ratio (OR) and the accompanying 95% confidence interval (CI) using samples with non-visible fluorescence for reference. P values < 0.05 were considered statistically significant. All statistical analyses were performed with SPSS software (version 23; IBM SPSS, Chicago IL, USA).

Among the 60 tumors, 42 (70%) were 5-ALA fluorescence-positive; the others were negative. Of the 8 WHO grade II gliomas, 2 were positive, as were 9 of 17 (53%) WHO grade III and 31 (89%) of 35 grade IV gliomas (Table 2). For high-grade gliomas, the PPV of 5-ALA fluorescence was 95.2%; the NPV was 33.3%, and diagnostic accuracy was 76.7%.

As shown in Table 3, among the 60 diffuse gliomas, 13 (22%) harbored IDH1 mutations. These were more often seen in grade II (4/8, 50%) and grade III gliomas (7/17, 41%) than in glioblastomas (2/35, 6%). Of the 13 tumors with IDH1 mutations, 2 (15%) manifested visible fluorescence, the other 11 did not.

Univariate analysis revealed a statistically significant association between 5-ALA fluorescence and the IDH1 status (mutated, non-mutated), 1p19q LOH, MIB-1, and the margin, heterogeneity, and contrast enhancement of the tumors (p < 0.001, p = 0.003, p = 0.007, p = 0.046, p = 0.021, and p = 0.002, respectively); neither the patient age nor the MGMT methylation status was significantly associated. Multivariate analysis showed that the IDH1 status was the only independent, statistically significant factor related to 5-ALA fluorescence (p = 0.009) (Table 4). Using tissue samples with nonvisible fluorescence as the reference, we found that the OR (95% CI) for IDH1 wild type was 19.238 (1.39, 175.39). No other factors had a significant effect on 5-ALA fluorescence.