Abstract

BACKGROUND: For appropriate use of available intraoperative imaging techniques in glioblastoma (GB) surgery, it is crucial to know the potential of the respective techniques in tumor detection.

OBJECTIVE: To assess histopathological basis of imaging results of intraoperative magnetic resonance imaging (iMRI), 5-aminolevulinic acid (5-ALA), and linear array intraoperative ultrasound (lioUS).

METHODS: We prospectively compared the imaging findings of iMRI, 5-ALA, and lioUS at 99 intraoperative biopsy sites in 33 GB patients during resection control. Histological classification of specimens, tumor load, presence of necrosis, presence of vascular malformations, and O6-methylguanin-DNA methyltransferase (MGMT) promoter state was correlated with imaging findings.

RESULTS: Solid tumor was found in 57%, infiltration zone in 42%, and no tumor in 1% of specimens. However, imaging was negative in iMRI in 49%, using 5-ALA in 17%, and in lioUS in 21%. In positive imaging results, share of solid tumor was highest in 5-ALA (65%) followed by lioUS (60%) and lowest in iMRI (55%). In comparison to 5-ALA, iMRI had a high share of solid tumor in specimens when showing intermediate results. Sensitivity for invasive tumor was higher in 5-ALA (84%) and lioUS (80%) than in iMRI (50%). We found a significant correlation of 5-ALA with classification of specimen, presence of necrosis, and microproliferations. Methylated MGMT promoter correlated with positive findings in 5-ALA. lioUS and iMRI showed no correlations with histopathological findings.

CONCLUSION: All of the assessed established imaging techniques detect infiltrating tumor only to a certain extent. Only 5-ALA showed a significant correlation with histopathological findings. Interestingly, tumor remnants in an MGMT-methylated tumor are more likely to be visible using 5-ALA as in unmethylated tumors.

ABBREVIATIONS

    ABBREVIATIONS
  • GB

    glioblastoma

  • MRI

    magnetic resonance imaging

  • iMRI

    intraoperative magnetic resonance imaging

  • 5-ALA

    5-aminolevulinic acid

  • lioUS

    linear array intraoperative ultrasound

  • GTR

    gross total resection

  • EoR

    extent of resection

  • Gd-DTPA

    gadolinium-diethylenetriamine pentaacetic acid

  • BBB

    blood brain barrier

  • ioUS

    intraoperative ultrasound

  • WHO

    World Health Organization

  • H&E

    hematoxylin and eosin

  • CI

    confidence interval

The perception of glioblastoma (GB) surgery has changed in the recent years. Even though we know that we cannot cure the disease using surgery, the data published in the last 10 years explicitly strengthen the value of surgery. Class I evidence based on the 5-aminolevulinic acid (5-ALA) trial exists that a gross total resection (GTR) leads to an increased overall survival (OS).1 Marko et al2 even showed that there is a linear correlation of extent of resection (EoR) and survival, expanding the indication for beneficial tumor removal beyond the previously published 98% or 78%.3,4 While there is still some debate whether a subtotal resection still provides a significant benefit for OS in the era of Temozolomid,5 there is no doubt that a complete resection of contrast enhancement is to be favored if safely feasible. Even small contrast residuals can lead to a significantly worse progression-free survival as well as tumor left adjacent to eloquent areas.6,7 Despite neurophysiological monitoring and mapping, intraoperative imaging can help surgeons to increase EoR and thus achieve the desired goal of increasing patients’ OS.

Even though GTR is defined heterogeneously in the literature, assessment of EoR is based on pre- and postoperative magnetic resonance imaging (MRI; T1 with gadolinium-diethylenetriaminepentacetate [Gd-DTPA] enhancement). However, contrast enhancement in MRI shows the margin of the disrupted blood brain barrier (BBB) and not the frontier of invading tumor. From histological studies, we know that solid tumor in GB spread significantly beyond the border of contrast enhancement.8,9 Even in patients with a GTR, tumor recurrence happens at the margins of the resection cavity in the majority of cases.10,11 Thus, the “peritumoral brain zone” is more and more in focus of research now. Ex Vivo and in Vivo experimental concepts were published assessing whether a “tumor-free” resection was established.12-17 Yet, few studies have assessed the histological background of the most common imaging techniques which are intraoperative MRI (iMRI), 5-ALA, and intraoperative ultrasound (ioUS).18-23 Only in the recent years, we have learned that 5-ALA fluorescence exceeds Gd-DTPA enhancement,23,24 which is an important implication for surgical planning or combined imaging approaches using 5-ALA and iMRI.25 Stummer et al14 provided interesting data suggesting that 5-ALA correlates directly with cell density. However, they also showed that using visible fluorescence only does not detect infiltrating tumor cells. The actual histopathological background of ioUS remains even more vague: solid tumor is usually relatively easy to differentiate from surrounding brain tissue. However, distinguishing edema from infiltrating tumor is challenging. These areas might be at risk to be classified as “free of tumor.” Mair et al26 developed a grading system assessing visibility and detection of the tumor borders of different entities by ultrasound. GBs are classified as grade 2 (lesion clearly identifiable but no clear border with normal tissue). Several authors have assessed whether ioUS differentiates between tumor and no tumor.18,22,27,28 Development of navigated 2- and 3-dimensional ultrasound, as used in our series, improved the resection control and detection of residual tumor as well as occurring brain shift due to a possible comparison with the corresponding preoperative MRI images.29,30 Novel ultrasound techniques such as tissue strain analysis and contrast-enhanced ultrasound might help to improve tumor detection using ioUS.31,32 To date, resection control using MRI applying a Gd-DTPA-enhanced T1 sequence is still the “gold standard.” Intraoperative high-field MRI allows for an imaging quality during the surgery comparable to pre- and postoperative diagnostic MRI.33 According to Senft et al,34 this leads to an increased rate of GTR and thus, to a significant longer progression-free survival. However, previous histopathological assessments of tumor depiction in low-field iMRI showed that absence of Gd-DTPA enhancement is a bad predictor for absence of GB-tissue.20 Further, a detailed assessment of the histopathological findings corresponding to the intraoperative high-field MRI data provides a better understanding of preoperative MRI images. Therefore, a more detailed understanding of histopathological fundamentals of applied intraoperative imaging techniques is crucial for interpretation of images and further surgical planning. Especially, a cross referencing of different imaging techniques with a specific biopsy allows for a direct comparison of the techniques. Until now, application of ultrasound contrast agent is “off-label” use, thus we did not apply it in this prospective series. In the actual series, we used only intraoperative imaging techniques, commercially available for intracranial application.

The aim of the actual study was to perform an evaluation of histopathological assessment and intraoperative imaging results of iMRI, 5-ALA, and ioUS at a defined biopsy site.

METHODS

Study Design

All histopathological samples were harvested as part of a prospective study. Ethical approval was received by the institution's local ethics board (Ethikkommission Ulm No: 172/12). Patients were recruited from 2012 May to 2013 October.

After informed consent, patients above 18 years with an intended GTR were included, when final histopathological assessment showed a primary GB World Health Organization (WHO) grade IV according to the WHO classification of 2007.35 Patients with recurrent surgery or prior radiation were excluded.

Intraoperative Imaging Techniques

5-Aminolevulinic Acid

5-ALA was administered orally, 4 h before surgery in a common dose of 20 mg/kg body weight. To visualize fluorescence, we used a Zeiss (Oberkochen, Germany) Pentero® 600 with integrated head-up display for visualization of neuronavigation and Blue 400® filter to perform 405-nm fluorescence.

Intraoperative Ultrasound

We used an iU22 xMatrix Ultrasound system (Philips, Amsterdam, Netherlands) and applied a L15-7io compact hockey stick-shaped linear array transducer from the same manufacturer (11 × 31 mm, 128 elements) with an extended frequency range of 15 to 7 Mhz.

The ultrasound was used integrated in a navigation system as described before.36

Intraoperative MRI

We performed all cases in a 1.5-Tesla iMRI (Magnetom Espree, Siemens Healthcare, Erlangen, Germany). Surgeries were done in a dedicated iMRI environment with integrated data management and neuronavigation (Brainsuite®, Brainlab, Feldkirchen). The Iplan 3.0 software (Brainlab) was used.

Patient registration to the neuronavigation system was performed using automatic registration on the basis of a short preoperative T1 magnetization-prepared rapid gradient echo (MPRAGE) scan without contrast after fixation of the head in the Noras head coil (Noras, Erlangen, Germany).

Intraoperatively we routinely performed a T1 MPRAGE with and without Gd-DTPA and an axial T2 and Flair sequence. Timing of iMRI was as described in the study protocol. Only results of first iMRI were evaluated as in this study. Further, we only evaluated residual contrast enhancement of iMRI, nonenhancing tumor, such as FLAIR changes etc., was not evaluated.

Intraoperative Study Protocol

A flow chart of the intraoperative study protocol is shown in Figure 1. In all patients, a typical white light resection under neuronavigational guidance was performed, until the surgeon assumed complete removal of the tumor. After hemostasis was achieved, the surgeon scanned the resection cavity using 5-ALA and ioUS marking conspicuous areas to contain residual tumor according to the imaging devices with the neuronavigation. After completion of the iMRI scan, areas with residual contrast enhancement were marked likewise. Further additional sites in whom no imaging device showed positive results were marked with the neuronavigation as “negative controls.” At all marked areas, 5-ALA, ioUS, and iMRI were cross referenced again and the results were recorded. Then navigated biopsies were harvested from these sites.

FIGURE 1.

Flow chart of intraoperative study protocol.

FIGURE 1.

Flow chart of intraoperative study protocol.

Consequently, every intraoperative biopsy point provides histology and the results of the above-mentioned imaging methods (eg, ALA +, iMRI –, ioUS +/–). If none of the imaging methods showed a positive finding (ALA –, iMRI –, ioUS –), only 1 negative biopsy was taken in the patient. The number of biopsies taken per patients was dependent from the number of conspicuous findings in the imaging methods. The protocol requested a biopsy if 1 out of 3 techniques showed an intermediate finding (eg, 5-ALA +/–, iMRI –, ioUS –). Therefore, number of biopsies per patient varied by the number of different imaging findings. For example, if only 1 site showed residual tumor in all 3 methods, 1 “positive” biopsy (eg, 5-ALA +, iMRI +, ioUS +) and 1 negative biopsy (eg, 5-ALA –, iMRI –, ioUS –) were taken. For ethical reasons, despite the protocol, harvesting of biopsies was at surgeon's discretion.

Categorization of Imaging Findings

Imaging findings were classified by the surgeon performing the case. Five different senior surgeons performed the surgeries and rated the imaging findings of 5-ALA, iMRI, and ioUS as mentioned below.

5-ALA fluorescence was judged by the surgeon as “strong,” “vague,” and “no fluorescence” as published by Stummer et al.37 For further assessment and comparison, these results were recorded as positive, intermediate, and negative. ioUS was judged as positive, intermediate, and negative by the surgeon. Gd-DTPA enhancement was assessed as positive, intermediate, and negative by the respective surgeon. Figure 2 shows intraoperative example images according to the classification to positive, intermediate, and negative as used by the surgeons for the 3 above-mentioned imaging techniques.

FIGURE 2.

Examples of image classification as positive intermediate and negative for A axial Gd-DTPA-enhanced T1 intraoperative MRI during resection of a right temporal GB: 1 negative imaging—no Gd-DTPA enhancement, 2 intermediate imaging—weak Gd-DTPA enhancement, 3 positive imaging—strong Gd-DTPA enhancement; for B 5-ALA fluorescence using a 450-nm filter during resection of a right temporal GB: 1 negative imaging—no residual fluorescence in resection cavity (lower two-third of image, upper one-third STG covered with latex sponge), 2 intermediate imaging—intermediate (vague) fluorescence at the bottom of the resection cavity in a right temporal GB showing spots of residual fluorescence of lower intensity and color than the deep red fluorescence of positive imaging, 3 positive imaging—strong reddish fluorescence in a residual part of the main tumor in a right temporal GB; and for C coronal image of linear array intraoperative ultrasound before resection of a right temporal GB showing the tumor infiltration of MTG and ITG sulcus separating the lobes is seen in the right of Figure 1C; 1 negative imaging—depiction of subcortical fibers; 2 intermediate imaging—showing the transition zone of solid tumor and “normal” tissue, difficult to distinguish from local edema; 3 positive imaging—strong echo-intense signal showing main part of the tumor.

FIGURE 2.

Examples of image classification as positive intermediate and negative for A axial Gd-DTPA-enhanced T1 intraoperative MRI during resection of a right temporal GB: 1 negative imaging—no Gd-DTPA enhancement, 2 intermediate imaging—weak Gd-DTPA enhancement, 3 positive imaging—strong Gd-DTPA enhancement; for B 5-ALA fluorescence using a 450-nm filter during resection of a right temporal GB: 1 negative imaging—no residual fluorescence in resection cavity (lower two-third of image, upper one-third STG covered with latex sponge), 2 intermediate imaging—intermediate (vague) fluorescence at the bottom of the resection cavity in a right temporal GB showing spots of residual fluorescence of lower intensity and color than the deep red fluorescence of positive imaging, 3 positive imaging—strong reddish fluorescence in a residual part of the main tumor in a right temporal GB; and for C coronal image of linear array intraoperative ultrasound before resection of a right temporal GB showing the tumor infiltration of MTG and ITG sulcus separating the lobes is seen in the right of Figure 1C; 1 negative imaging—depiction of subcortical fibers; 2 intermediate imaging—showing the transition zone of solid tumor and “normal” tissue, difficult to distinguish from local edema; 3 positive imaging—strong echo-intense signal showing main part of the tumor.

Histopathological Workup

Together with the main pathology sample, the above-mentioned biopsies were forwarded to local pathology. The neuropathologists were blinded to the clinical categorization and the location of the biopsies. Two neuropathologists observed the samples separately. In case of disagreement, the neuropathologists discussed the sample and agreed into a consensus diagnosis.

All samples were fixed in 10% buffered formalin and paraffin embedded. Human GBs were classified neuropathologically according to the WHO classification of tumors of the central nervous system.35 For this purpose, paraffin sections were stained with hematoxylin and eosin (H&E). Immunohistochemistry for GBs was carried out with antibodies raised against glial fibrillary acidic protein (polyclonal rabbit, 1/1000, DAKO, Glostrup, Denmark), microtubule-associated protein MAP2 (HM-2, 1/500, heat pretreatment, Sigma-Aldrich, St. Louis, Missouri), and the ki67-epitope (MIB1, 1/100, heat pretreatment, DAKO, Glostrup, Denmark). Pathologically, “solid tumor” was defined as tumor tissue exhibiting a pattern fulfilling the WHO diagnostic criteria of a GB with less than ∼20% nontumor tissue at the H&E level. “Tumor infiltration zone” samples contained less than ∼20% tumor cells but tumor cells and tumor proliferation were detectable in H&E and anti-ki67-stained samples. “Tumor-free tissue” was referred to samples, in which no tumor cells could be identified at the H&E level and no increased levels of proliferations were detectable with anti-ki67. Presence of necrosis and vascular microproliferates were assessed categorically in the samples. In the semiquantitative assessment of tumor load, we assessed the percentage of solid tumor in the harvested specimen. The whole specimen was screened, and the percentage of tumor-free areas, infiltration zone, and solid tumor evaluated. The size of intraoperatively harvested tumor samples was assessed semiquantitatively (≥1.5 × 1.5 cm, ≥1 × 1 cm, ≥0.5 × 1 cm, ≥0.5 × 0.5 cm; <0.5 × 0.5 cm).

Statistical Assessment

We used a descriptive and exploratory assessment. In continuous variables, we calculated mean values and standard error of the mean. Variables with missing data were not included in the calculation. Correlations were calculated and compared based on Spearman's Rho. We performed 3 × 7 tests with the above-mentioned tool. Since we performed only an exploratory assessment and not a confirmatory assessment, we did not do a P-value adjustment (eg, Bonferroni). For the above-mentioned assessments, we used SPSS 23 (IBM, Armonk, New York).

Sensitivity and specificity of pathological tissue (tumor and infiltration zone) detection were calculated after dichotomization of imaging results. Vague fluorescence, intermediate uptake of contrast, and a slight hyper density in ultrasound were judged as positive. Calculation was performed using medcalc.org.

RESULTS

General Assessment

We assessed 33 primary surgeries of 33 patients harboring a primary GB. Table 1 shows patients’ characteristics. Ninety-nine biopsies were taken from the resection cavity after assumed GTR. Fifty-six biopsies were confined to be solid tumor, 42 to be infiltration zone, and only 1 specimen contained no tumor cells. We harvested 8 specimens in 8 patients from sites where all imaging techniques were negative.

TABLE 1.

Patient's Characteristics

Characteristic   
Median age  60 (range 33-78) 
Female ratio  43.8% 
Affected lobe Frontal 46.7% 
 Temporal 33.3% 
 Parietal/occipital 10.0% resp. 
Left-sided lesions  40.0% 
Eloquent areas  32.1% 
MGMT promoter methylation  51.7% 
Mean tumor volume  34.6 cm³ (SEM 4.8) 
Mean EoR  99.6% (SEM 0.2) 
Mean no. of biopsies per patient  3 (range 1-7) 
Characteristic   
Median age  60 (range 33-78) 
Female ratio  43.8% 
Affected lobe Frontal 46.7% 
 Temporal 33.3% 
 Parietal/occipital 10.0% resp. 
Left-sided lesions  40.0% 
Eloquent areas  32.1% 
MGMT promoter methylation  51.7% 
Mean tumor volume  34.6 cm³ (SEM 4.8) 
Mean EoR  99.6% (SEM 0.2) 
Mean no. of biopsies per patient  3 (range 1-7) 

MGMT: O6-methylguanin-DNA methyltransferase; SEM: standard error of mean; EoR: extent of resection.

Assessment of Final Histological Classification and Imaging

Figure 2 shows a distribution of final histological classification of specimen and respective imaging finding for iMRI, 5-ALA, and linear array intraoperative ultrasound (lioUS). It demonstrates a relatively high rate of undetected solid tumor and infiltration zone for iMRI negative findings. lioUS and 5-ALA showed a similar pattern of tumor detection, while rate of false negative findings is slightly higher in lioUS than in 5-ALA.

We calculated sensitivity and specificity of imaging results after dichotomization of data. Sensitivity to detect pathological tissue was 84% (75%-91% confidence interval [CI] 95%) using 5-ALA, 50% (40%-60% CI 95%) in iMRI, and 80% (69%-89% CI 95%) using lioUS. Specificity was 100% (3%-100% CI 95%) in all imaging methods.

Correlation of Imaging and Histological Results

We tested for significant correlation between imaging findings and histology (Table 2). The imaging methods did not correlate significantly with each other. However, 5-ALA and lioUS showed a similar distribution in Figure 3. Only 5-ALA correlated significantly with the classification of histological specimen. It also correlated positively with presence of necrosis and microproliferations. Interestingly, we found a correlation of positive imaging results in 5-ALA and a methylated O6-methylguanin-DNA methyltransferase (MGMT) promoter. Additionally, no difference was found between distribution of tumor location and imaging. Classification of specimen served as a control. As expected, it significantly correlated with further histopathological findings. We performed a post hoc power analysis based on the results of the correlation analysis. Based on a correlation of 0.271 (5-ALA with classification of specimen), 99 samples and an alpha error of 0.05, we reached a power of 78%.

FIGURE 3.

Distribution of imaging results and histopathological diagnosis of specimen. iMRI: intraoperative MRI; 5-ALA 5-aminolevulinic acid fluorescence; lioUS: linear array intraoperative ultrasound.

FIGURE 3.

Distribution of imaging results and histopathological diagnosis of specimen. iMRI: intraoperative MRI; 5-ALA 5-aminolevulinic acid fluorescence; lioUS: linear array intraoperative ultrasound.

TABLE 2.

Spearman's Rho Correlations

     Classification Tumor Presence of ki67 Vascular. MGMT Size of 
  iMRI 5-ALA lioUS of spec. load nekrosis index microprol. methylation residual 
iMRI Correlation coefficient 1.000 0.135 0.108 0.104 0.142 –0.003 –0.014 0.182 –0.002 0.109 
 Sig. (2-tailed) – P < .201 P < .397 P < .316 P < .178 P < .976 P < .901 P < .092 P < .985 P < .371 
5-ALA Correlation coefficient 0.135 1.000 0.090 0.271(a0.125 0.212(b0.135 0.243(b0.218(b0.164 
 Sig. (2-tailed) P < .201 – P < .480 P < .008 P < .237 P < .044 P < .208 P < .022 P < .040 P < .174 
ioUS Correlation coefficient 0.108 0.090 1.000 0.007 –0.151 –0.089 –0.168 0.117 0.144 0.083 
 Sig. (2-tailed) P < .397 P < .480 – P < .955 P < .227 P < .477 P < .179 P < .354 P < .251 P < .527 
     Classification Tumor Presence of ki67 Vascular. MGMT Size of 
  iMRI 5-ALA lioUS of spec. load nekrosis index microprol. methylation residual 
iMRI Correlation coefficient 1.000 0.135 0.108 0.104 0.142 –0.003 –0.014 0.182 –0.002 0.109 
 Sig. (2-tailed) – P < .201 P < .397 P < .316 P < .178 P < .976 P < .901 P < .092 P < .985 P < .371 
5-ALA Correlation coefficient 0.135 1.000 0.090 0.271(a0.125 0.212(b0.135 0.243(b0.218(b0.164 
 Sig. (2-tailed) P < .201 – P < .480 P < .008 P < .237 P < .044 P < .208 P < .022 P < .040 P < .174 
ioUS Correlation coefficient 0.108 0.090 1.000 0.007 –0.151 –0.089 –0.168 0.117 0.144 0.083 
 Sig. (2-tailed) P < .397 P < .480 – P < .955 P < .227 P < .477 P < .179 P < .354 P < .251 P < .527 

iMRI: intraoperative MRI; 5-ALA: 5-aminolevulinic acid, ioUS: intraoperative linear ultrasound; Sig.: significance; microprol.: microproliferations; spec.: specimen.

aCorrelation is significant at the 0.01 level (2-tailed).

bCorrelation is significant at the 0.05 level (2-tailed).

After we found a significant correlation of a methylated MGMT promoter and positive 5-ALA imaging, we assessed rates of undetected solid tumor in all 3 imaging methods based on MGMT promoter methylation. Patients with unmethylated promoter had a rate of undetected solid tumor of 46% in iMRI, 25% in lioUS, and 20% in 5-ALA. In patients with methylated promoter, rate of undetected tumor was 39%, 15%, and 0%, respectively.

Detailed Histological Assessment Based on Imaging Results

The histopathological details of specimens classified as tumor, infiltration zone, and no tumor serve as a “control” and are found in Table 3.

TABLE 3.

Histopathological Findings by Classification of Specimen

 Classification of specimen 
Histopathological assessment No tumor (n = 1) Infiltration zone (n = 42) Tumor (n = 56) 
Presence of necrosis 0% 3% 15% 
Vascular microproliferates 0% 17% 53% 
Mean ki67index 0%(SEM 0) 3% (SEM 0) 15% (SEM 1) 
Mean semiquantitative estimation of tumor mass in specimen 0% (SEM 0) 0% (SEM 5) 65% (SEM 4) 
Median area of intraoperative residual in cm 0.5 × 0.5 (SEM 0) 0.5 × 0.5 (SEM 1) 0.5 × 0.5 (SEM 1) 
 Classification of specimen 
Histopathological assessment No tumor (n = 1) Infiltration zone (n = 42) Tumor (n = 56) 
Presence of necrosis 0% 3% 15% 
Vascular microproliferates 0% 17% 53% 
Mean ki67index 0%(SEM 0) 3% (SEM 0) 15% (SEM 1) 
Mean semiquantitative estimation of tumor mass in specimen 0% (SEM 0) 0% (SEM 5) 65% (SEM 4) 
Median area of intraoperative residual in cm 0.5 × 0.5 (SEM 0) 0.5 × 0.5 (SEM 1) 0.5 × 0.5 (SEM 1) 

SEM: standard error of the mean.

Table 4 shows a cross table of positive imaging results of iMRI, 5-ALA, and ioUS with the respective histological findings. The semiquantitative assessment of tumor load was highest in iMRI positive specimens. Yet, iMRI showed the highest share of infiltration zone in these specimens followed by ioUS. In intermediate imaging results, a high rate of solid tumor and tumor load can be found in iMRI while 5-ALA showed the lowest share of tumor in these specimens. In negative imaging results of all 3 methods, tumor load was relatively high; being highest in lioUS and lowest in 5-ALA. However, the share of negative samples was distinctly lower in 5-ALA and lioUS than in iMRI. There was no difference in size of residual tumor by different imaging results.

TABLE 4.

Histopathological Findings and Imaging Results

 
 Positive results Intermediate results Negative results 
 iMRI 5-ALA ioUS iMRI 5-ALA ioUS iMRI 5-ALA ioUS 
Histopathological assessment (n = 31) (n = 68) (n = 48) (n = 16) (n = 12) (n = 5) (n = 48) (n = 16) (n = 14) 
Classification of specimen Solid tumor 55% 65% 81% 33% 60% 48% 38% 64% 64% 
 Tumor invasion 45% 35% 19% 67% 40% 50% 56% 29% 29% 
 No tumor 0% 0% 2% 0% 0% 2% 6% 7% 7% 
Presence of necrosis 11% 13% 6% 7% 0% 0% 10% 0% 14% 
Vascular microproliferates 50% 43% 46% 62% 27% 20% 33% 19% 35% 
MGMT methylation 42% 49% 38% 47% 27% 40% 47% 23% 17% 
Mean ki67index 9% (SEM 2) 10% (SEM 1) 9% (SEM 1) 10% (SEM 2) 9% (SEM 3) 15% (SEM 6) 10% (SEM 1) 9% (SEM 3) 12% (SEM 3) 
Mean semiquantitative estimation of tumor mass in specimen 43% (SEM 8) 38% (SEM 5) 37% (SEM 6) 41% (SEM 10) 28% (SEM 13) 37% (SEM 6) 29% (SEM 5) 30% (SEM 11) 56% (SEM 12) 
Median area of intraoperative residual in cm 0.5 × 0.5 (SD 1) 0.5 × 0.5 (SD 1) 0.5 × 0.5 (SD 1) < 0.5 × 0.5 (SD 1) 0.5 × 1 (SD 2) 0.5 × 1 (SD 1) 0.5 × 0.5 (SD 1) 0.5 × 1 (SD 1) 0.5 × 0.5 (SD 1) 
 
 Positive results Intermediate results Negative results 
 iMRI 5-ALA ioUS iMRI 5-ALA ioUS iMRI 5-ALA ioUS 
Histopathological assessment (n = 31) (n = 68) (n = 48) (n = 16) (n = 12) (n = 5) (n = 48) (n = 16) (n = 14) 
Classification of specimen Solid tumor 55% 65% 81% 33% 60% 48% 38% 64% 64% 
 Tumor invasion 45% 35% 19% 67% 40% 50% 56% 29% 29% 
 No tumor 0% 0% 2% 0% 0% 2% 6% 7% 7% 
Presence of necrosis 11% 13% 6% 7% 0% 0% 10% 0% 14% 
Vascular microproliferates 50% 43% 46% 62% 27% 20% 33% 19% 35% 
MGMT methylation 42% 49% 38% 47% 27% 40% 47% 23% 17% 
Mean ki67index 9% (SEM 2) 10% (SEM 1) 9% (SEM 1) 10% (SEM 2) 9% (SEM 3) 15% (SEM 6) 10% (SEM 1) 9% (SEM 3) 12% (SEM 3) 
Mean semiquantitative estimation of tumor mass in specimen 43% (SEM 8) 38% (SEM 5) 37% (SEM 6) 41% (SEM 10) 28% (SEM 13) 37% (SEM 6) 29% (SEM 5) 30% (SEM 11) 56% (SEM 12) 
Median area of intraoperative residual in cm 0.5 × 0.5 (SD 1) 0.5 × 0.5 (SD 1) 0.5 × 0.5 (SD 1) < 0.5 × 0.5 (SD 1) 0.5 × 1 (SD 2) 0.5 × 1 (SD 1) 0.5 × 0.5 (SD 1) 0.5 × 1 (SD 1) 0.5 × 0.5 (SD 1) 

iMRI: intraoperative MRI; 5-ALA: 5-aminolevulinic acid, ioUS: intraoperative linear ultrasound; SEM: standard error of the mean, MGMT: O6-methylguanin-DNA methyltransferase; SD: standard deviation.

DISCUSSION

EoR significantly influences GB patients’ OS, despite other important factors like age, tumor location, and preoperative Karnofsky performance status etc.2 Yet, EoR is one of the few variables that can be influenced by surgeons. Intraoperative imaging is one tool to help surgeons to increase EoR and thus achieve the desired goal of increasing patients’ OS. The main goal in surgery, as mentioned before, is a GTR of Gd-DTPA enhancement in postoperative MRI. Yet, we know for quite a while that it does not show us all of the active tumor as correlations between MRI and aminoacid Positron emission tomography (PET) CT studies show.38 In most centers, financial and logistic reason hampers the routine use of PET CT as postoperative follow-up imaging. Therefore, it is even more important to study how we can safely increase the extent of tumor removal by intraoperative imaging. Apart from establishing new experimental approaches to perform in Vivo or ex Vivo assessments of safe resection margins, we must study the histopathological basis of the intraoperative imaging techniques we are currently using.

Comparison of Final Histological Diagnosis and Imaging

We performed a detailed histopathological assessment of the tissue depiction of iMRI, 5-ALA, and ioUS after assumed GTR. This means, to the best of the author's knowledge, we provide the first study comparatively assessing the histopathological basis of these 3 imaging techniques, together. Explicitly, intermediate imaging findings were studied since these are hardest to interpret for most surgeons. Especially for 5-ALA, most authors recommend a 3-level classification as we are using in our study.39

Surprisingly, our data show that despite negative imaging findings in all 3 imaging techniques, we found only 1 specimen that revealed most likely tumor-free brain tissue. All imaging techniques showed false negative results and did not detect solid tumor in all cases. This reflects previously published findings on the invasive nature of GB.8,9

Yet, our data show a distinctly different detection pattern of the 3 imaging techniques at the border of the resection cavity. Our group has assessed accuracy for solid tumor detection in a pairwise comparison of iMRI vs 5-ALA and of iMRI vs lioUS.22,23 Especially, for 5-ALA, a slight “overdetection” leading to a lower specificity was presumed. The actual assessment compares all 3 imaging methods at the same time and includes infiltration zone and intermediate imaging findings as well as a detailed histopathological assessment of each specimen. The inclusion of infiltration zone revealed that 5-ALA has a higher detection rate of solid tumor, while the percentage of additional detection of infiltration zone is lower as in iMRI. Still, the imaging techniques did not correlate significantly with each other in Spearman's Rho; lioUS and 5-ALA seem to have a more similar detection pattern than contrast enhancement in iMRI. The results for sensitivity and specificity that we found before in the pairwise comparison were similar to the results we found in the actual series. Specificity was higher in all imaging techniques. However, due to the low incidence of true negative results it might not be representative.

Concerning intermediate imaging findings, we found that in all imaging methods, share of solid tumor and tumor load of specimen was almost as high as in typical positive imaging results. This finding is an important implication for clinical routine underlining the importance of resection of these so-called vague areas in imaging.

Histological Basis for Differences of Tumor Depiction

Based on the exploratory Spearman's Rho correlation analysis, the variation of the histopathological results does not explain the variation of the imaging results. At least, this applies to iMRI and ioUS. Only 5-ALA fluorescence correlated significantly with histopathological results. However, our data confirm quite strongly that all contemporary imaging methods are insufficient to detect infiltrating residual tumor, though 5-ALA might be a bit better than others.

The assessment of histopathological findings according to classification of specimen served as a control in our series. An ideal diagnostic tool would show the same pattern as the “control.” Yet, we cannot find these distinctive patterns in any of the 3 imaging methods.

lioUS showed a relatively high share of tumor in its negative findings despite a high detection rate of solid tumor in general. We did not record whether a negative finding meant a similar signal-like normal brain tissue was found or whether the surgeon assumes to see an artifact and judged it as negative. We presume the latter was the case in these false negative findings. It shows the typical issue of increased artifacts when using ioUS at the end of resection.

The anatomic correlate of Gd-DTPA enhancement is the disrupted BBB caused by extensive growth, vascular malformation, and changes to the cellular membrane.40 A histological correlate is the rate of vascular proliferations. In our data, we found high rates of vascular proliferations in strongly and partially contrast-enhancing areas. Also, 5-ALA shows a correlation of imaging findings and presence of vascular proliferations. 5-ALA fluorescence can be emitted if protoporphyrin 9 phosphatase is missing in tumor cells. Additionally, it shows a better permeability to areas with a disrupted BBB.41 Findings in lioUS do not seem to be related to vascular proliferations despite being a typical feature for solid tumor in GB as the “control” underlines.

Further, we assessed MGMT promoter methylation of the main tumor and correlated it with imaging findings since we had the hypothesis that a loss of MGMT promoter methylation might also correlate with an increased malignancy and thus an increased change of tumor cell's endogenous heme-cycle. Indeed, in 5-ALA we have seen a lower rate of false negative findings when tumor was MGMT methylated. Also in iMRI and in lioUS, rate of missed tumor was lower in MGMT-methylated lesions; however, correlations were not statistical significant.

The data did not indicate that one imaging technique was superior to another in terms of detecting small residuals. In our series, no distinctly different patterns in ki-67 index were found in the 3 imaging techniques.

Despite the significant correlation of 5-ALA and histology of specimen, in the vast majority of 5-ALA negative samples, invasive tumor was found. In GB, a homogenous fluorescence is described for areas with solid tumor despite necrotic sites while infiltrating tumor seems to show a rather vague fluorescence.37 Spectroscopic assessments show that residual fluorescence can be found in microscopically nonfluorescent areas too. Thus, it is likely that all GB tumor cells might carry the respective mutation while the emitted light is too less to be recognized by the surgeon's eye.14

Limitations

The results of our prospective series are biased by the subjective assessment of imaging results by the respective surgeon. Thus, an interrater bias cannot be excluded. Blinded assessment of imaging data was not performed since it would not have been feasible using intraoperative tissue fluorescence like 5-ALA. A cross referencing of all 3 imaging methods might lead to higher detection rates in 5-ALA and lioUS. The missing of areas covered with retractors, blood, or cottonoids, which might accidentally happen with these techniques, is omitted with a final iMRI scan.

The results of iMRI in solid and infiltrating tumor detection are limited since only Gd-DTPA-enhanced T1 was considered for the actual assessment. Despite from interpretation of standard T2 imaging and FLAIR sequences, dynamic imaging techniques like perfusion or dynamic contrast-enhanced sequences and intraoperative spectroscopy are increasing the detection rate of iMRI, as previously published.42-44 Further studies are needed assessing the full spectrum of high-field iMRI. Measurement of sensitivity based on Gd-DTPA enhancement alone is potentially underestimating the accuracy of the method.

Despite the otherwise very strict and cumbersome intraoperative protocol, harvesting of biopsies in “triple-negative” areas was left to surgeon's discretion due to ethical reasons. This leads to a low number of “control” biopsies and to a potential bias of the calculation of specificity.

Based on the surgical technique used, our data reflect mainly the margin of Gd-DTPA enhancement of a GB. It does not provide data on the central part of the tumor nor the peritumoral brain zone more distant from this area.

We performed 3 × 7 = 21 calculations for Spearman's Rho. Since we performed only an exploratory assessment and not a confirmatory assessment no P-value adjustment (eg, Bonferroni) was performed.

We used a high resolution high frequency lioUS registered to the neuronavigation system. Results of this ultrasound device might not be comparable to all transducer types.

MGMT promoter methylation is distributed heterogeneously in a GB.45 In our series, only an MGMT methylation of the main tumor was assessed. A semiquantitative PCR assessment is needed to assess whether this distribution affects 5-ALA fluorescence.

Further research is needed to verify the very interesting correlation of 5-ALA fluorescence and a methylated MGMT promoter state.

From our data, no conclusion can be drawn regarding whether a combined use of intraoperative imaging will improve outcome, nor if one imaging method is superior to another with regard to EoR or outcome. We performed a histopathological evaluation of imaging results using a highly standardized surgical protocol. Future research is needed to comparatively assess outcome and EoR of the above-mentioned imaging techniques.

Clinical Implications and Future Perspectives

Still, we have found false negative results in all imaging techniques, we did not encounter false positive results. Thus, we recommend resection of all tissue being depicted as potentially pathological tissue in iMRI, 5-ALA, or lioUS. Also, areas with intermediate contrast uptake and vague 5-ALA fluorescence should be removed if feasible without causing new neurological deficits. Therefore, additional use of appropriate intraoperative monitoring and mapping is crucial in such an approach. In ioUS, the main issue is to differentiate artifacts from intermediate imaging results, which can be challenging and requires specific probes and experience.

All tested imaging methods have methodological and practical limitations. Based on our data, the logical consequence is a combined use of intraoperative imaging techniques.

Since 5-ALA seems to rule out most of residual tumor very precisely, it could be useful to start resection using this technique and to perform a swipe of the resection cavity using ioUS afterwards. Thus, the limitation of 5-ALA not to detect deep seated tumor nodule could be overcome. Both techniques in combination allow for a real-time detection of the surface and the deep structures of the resection cavity. Further, in comparison to iMRI, both techniques are easily affordable for most neurosurgical centers worldwide.

On the other hand, intraoperative high-field MRI is the only technique providing images comparable to pre- and postoperative diagnostic MRIs. Further, it allows for additional sequences increasing sensitivity and specificity as well as functional imaging like diffusion tensor imaging (DTI) intraoperatively. Because of the restrictions of repetitive intraoperative Gd-DTPA use, a combined approach using 5-ALA or ioUS before iMRI scan is very promising. Our group showed a significant increase of EoR in a prospective series using 5-ALA and iMRI compared to iMRI alone. Further studies are needed to assess whether an application of 5-ALA and ioUS is comparable to use of iMRI alone or combined with 5-ALA.

Almost 20 years ago, Silbergeld et al9 published that 4 cm from solid tumor area, invasive tumor cells were still found histopathologically. Thus, the histopathological results of our series do not come as a surprise. Yet, EoR based on Gd-DTPA enhancement is still a standard of care in GB surgery in most neurosurgical centers. Our data once more implicate to go beyond that margin in a maximal safe resection approach. Intraoperative imaging together with neurophysiological monitoring can help to reach that goal. However, it is important to know its strengths and weaknesses.

CONCLUSION

Negative intraoperative imaging in any modality shows tumor or infiltrating tumor cells in more than 90% of cases. Since all tested imaging techniques showed false negative results, a combined use of intraoperative techniques should be considered if available at the respective center.

Based on our histological assessment, we recommend resection of any pathological imaging finding, if possible without causing new permanent deficits, since it will most likely contain solid tumor or at least infiltrating tumor.

Only 5-ALA showed a correlation with histopathological findings. Interestingly, tumor remnants in an MGMT-methylated tumor are more likely to be visible using 5-ALA as in unmethylated tumors.

The assessed contemporary imaging techniques detect infiltrating tumor only to a certain extent. To date, even when combining the established imaging techniques, no ideal method exists which detects tumor like a histopathological assessment.

Disclosure

The authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.

REFERENCES

1.
Stummer
W
,
Pichlmeier
U
,
Meinel
T
,
Wiestler
OD
,
Zanella
F
,
Reulen
H-J
.
Fluorescence-guided surgery with 5-aminolevulinic acid for resection of malignant glioma: a randomised controlled multicentre phase III trial
.
Lancet Oncol
 .
2006
;
7
(
5
):
392
-
401
.
2.
Marko
NF
,
Weil
RJ
,
Schroeder
JL
,
Lang
FF
,
Suki
D
,
Sawaya
RE
.
Extent of resection of glioblastoma revisited: personalized survival modeling facilitates more accurate survival prediction and supports a maximum-safe-resection approach to surgery
.
J Clin Oncol
 .
2014
;
32
(
8
):
774
-
782
.
3.
Sanai
N
,
Berger
MS
.
Glioma extent of resection and its impact on patient outcome
.
Neurosurgery
 .
2008
;
62
(
4
):
753
-
764
;
discussion 264-756
.
4.
Lacroix
M
,
Abi-Said
D
,
Fourney
DR
et al
A multivariate analysis of 416 patients with glioblastoma multiforme: prognosis, extent of resection, and survival
.
J Neurosurg
 .
2001
;
95
(
2
):
190
-
198
.
5.
Kreth
FW
,
Thon
N
,
Simon
M
et al
Gross total but not incomplete resection of glioblastoma prolongs survival in the era of radiochemotherapy
.
Ann Oncol
 .
2013
;
24
(
12
):
3117
-
3123
.
6.
Coburger
J
,
Wirtz
CR
,
Konig
RW
.
Impact of extent of resection and recurrent surgery on clinical outcome and overall survival in a consecutive series of 170 patients for glioblastoma in intraoperative high field iMRI
.
J Neurosurg Sci
 .
2015
.
Epub ahead of print.
7.
Chaichana
KL
,
Jusue-Torres
I
,
Navarro-Ramirez
R
et al
Establishing percent resection and residual volume thresholds affecting survival and recurrence for patients with newly diagnosed intracranial glioblastoma
.
Neuro Oncol
 .
2014
;
16
(
1
):
113
-
122
.
8.
Yamahara
T
,
Numa
Y
,
Oishi
T
et al
Morphological and flow cytometric analysis of cell infiltration in glioblastoma: a comparison of autopsy brain and neuroimaging
.
Brain Tumor Pathol
 .
2010
;
27
(
2
):
81
-
87
.
9.
Silbergeld
DL
,
Chicoine
MR
.
Isolation and characterization of human malignant glioma cells from histologically normal brain
.
J Neurosurg
 .
1997
;
86
(
3
):
525
-
531
.
10.
Coburger
J
,
Engelke
J
,
Scheuerle
A
et al
Tumor detection with 5-aminolevulinic acid fluorescence and Gd-DTPA-enhanced intraoperative MRI at the border of contrast-enhancing lesions: a prospective study based on histopathological assessment
.
Neurosurg Focus
 .
2014
;
36
(
2
):
E3
.
11.
Petrecca
K
,
Guiot
MC
,
Panet-Raymond
V
,
Souhami
L
.
Failure pattern following complete resection plus radiotherapy and temozolomide is at the resection margin in patients with glioblastoma
.
J Neurooncol
 .
2013
;
111
(
1
):
19
-
23
.
12.
Bohringer
HJ
,
Lankenau
E
,
Stellmacher
F
,
Reusche
E
,
Huttmann
G
,
Giese
A
.
Imaging of human brain tumor tissue by near-infrared laser coherence tomography
.
Acta Neurochir (Wien)
 .
2009
;
151
(
5
):
507
-
517
;
discussion 517
.
13.
Assayag
O
,
Grieve
K
,
Devaux
B
et al
Imaging of non-tumorous and tumorous human brain tissues with full-field optical coherence tomography
.
Neuroimage Clin.
 
2013
;
2
:
549
-
557
.
14.
Stummer
W
,
Tonn
JC
,
Goetz
C
et al
5-Aminolevulinic acid-derived tumor fluorescence: the diagnostic accuracy of visible fluorescence qualities as corroborated by spectrometry and histology and postoperative imaging
.
Neurosurgery
 .
2014
;
74
(
3
):
310
-
319
;
discussion 319-320
.
15.
Utsuki
S
,
Oka
H
,
Sato
S
et al
Possibility of using laser spectroscopy for the intraoperative detection of nonfluorescing brain tumors and the boundaries of brain tumor infiltrates. Technical note
.
J Neurosurg
 .
2006
;
104
(
4
):
618
-
620
.
16.
Kut
C
,
Chaichana
KL
,
Xi
J
et al
Detection of human brain cancer infiltration ex vivo and in vivo using quantitative optical coherence tomography
.
Sci Transl Med
 .
2015
;
7
(
292
):
292ra100
.
17.
Lemee
JM
,
Clavreul
A
,
Menei
P
.
Intratumoral heterogeneity in glioblastoma: don’t forget the peritumoral brain zone
.
Neuro Oncol
 .
2015
;
17
(
10
):
1322
-
1332
.
18.
Rygh
OM
,
Selbekk
T
,
Torp
SH
,
Lydersen
S
,
Hernes
TA
,
Unsgaard
G
.
Comparison of navigated 3D ultrasound findings with histopathology in subsequent phases of glioblastoma resection
.
Acta Neurochir (Wien)
 .
2008
;
150
(
10
):
1033
-
1041
;
discussion 1042
.
19.
Idoate
MA
,
Diez Valle
R
,
Echeveste
J
,
Tejada
S
.
Pathological characterization of the glioblastoma border as shown during surgery using 5-aminolevulinic acid-induced fluorescence
.
Neuropathology
 .
2011
;
31
(
6
):
575
-
582
.
20.
Kubben
PL
,
Wesseling
P
,
Lammens
M
et al
Correlation between contrast enhancement on intraoperative magnetic resonance imaging and histopathology in glioblastoma
.
Surg Neurol Int
 .
2012
;
3
:
158
.
21.
Coburger
J
,
Scheuerle
A
,
Thal
DR
et al
Linear array ultrasound in low-grade glioma surgery: histology-based assessment of accuracy in comparison to conventional intraoperative ultrasound and intraoperative MRI
.
Acta Neurochir (Wien)
 .
2015
;
157
(
2
):
195
-
206
.
22.
Coburger
J
,
Scheuerle
A
,
Kapapa
T
et al
Sensitivity and specificity of linear array intraoperative ultrasound in glioblastoma surgery: a comparative study with high field intraoperative MRI and conventional sector array ultrasound
.
Neurosurg Rev
 .
2015
;
38
(
3
):
499
-
509
.
23.
Coburger
J
,
Engelke
J
,
Scheuerle
A
et al
Tumor detection with 5-aminolevulinic acid fluorescence and Gd-DTPA-enhanced intraoperative MRI at the border of contrast-enhancing lesions: a prospective study based on histopathological assessment
.
Neurosurg Focus
 .
2014
;
36
(
2
):
E3
.
24.
Roessler
K
,
Becherer
A
,
Donat
M
,
Cejna
M
,
Zachenhofer
I
.
Intraoperative tissue fluorescence using 5-aminolevolinic acid (5-ALA) is more sensitive than contrast MRI or amino acid positron emission tomography ((18)F-FET PET) in glioblastoma surgery
.
Neurol Res
 .
2012
;
34
(
3
):
314
-
317
.
25.
Coburger
J
,
Hagel
V
,
Wirtz
CR
,
Konig
R
.
Surgery for glioblastoma: impact of the combined use of 5-aminolevulinic acid and intraoperative mri on extent of resection and survival
.
PLoS One
 .
2015
;
10
(
6
):
e0131872
.
26.
Mair
R
,
Heald
J
,
Poeata
I
,
Ivanov
M
.
A practical grading system of ultrasonographic visibility for intracerebral lesions
.
Acta Neurochir
 .
2013
;
155
(
12
):
2293
-
2298
.
27.
Woydt
M
,
Krone
A
,
Becker
G
,
Schmidt
K
,
Roggendorf
W
,
Roosen
K
.
Correlation of intra-operative ultrasound with histopathologic findings after tumour resection in supratentorial gliomas. A method to improve gross total tumour resection
.
Acta Neurochir (Wien)
 .
1996
;
138
(
12
):
1391
-
1398
.
28.
Wang
J
,
Liu
X
,
Hou
WH
et al
The relationship between intra-operative ultrasonography and pathological grade in cerebral glioma
.
J Int Med Res
 .
2008
;
36
(
6
):
1426
-
1434
.
29.
Gronningsaeter
A
,
Kleven
A
,
Ommedal
S
et al
SonoWand, an ultrasound-based neuronavigation system
.
Neurosurgery
 .
2000
;
47
(
6
):
1373
-
1379
;
discussion 1379-1380
.
30.
Schlaier
JR
,
Warnat
J
,
Dorenbeck
U
,
Proescholdt
M
,
Schebesch
KM
,
Brawanski
A
.
Image fusion of MR images and real-time ultrasonography: evaluation of fusion accuracy combining two commercial instruments, a neuronavigation system and a ultrasound system
.
Acta Neurochir
 .
2004
;
146
(
3
):
271
-
277
.
31.
Selbekk
T
,
Brekken
R
,
Indergaard
M
,
Solheim
O
,
Unsgard
G
.
Comparison of contrast in brightness mode and strain ultrasonography of glial brain tumours
.
BMC Med Imaging
 .
2012
;
12
(
1
):
11
.
32.
Prada
F
,
Perin
A
,
Martegani
A
et al
Intraoperative contrast-enhanced ultrasound for brain tumor surgery
.
Neurosurgery
 .
2014
;
74
(
5
):
542
-
552
;
discussion 552
.
33.
Nimsky
C
,
Ganslandt
O
,
Fahlbusch
R
.
Comparing 0.2 tesla with 1.5 tesla intraoperative magnetic resonance imaging analysis of setup, workflow, and efficiency
.
Acad Radiol
 .
2005
;
12
(
9
):
1065
-
1079
.
34.
Senft
C
,
Bink
A
,
Franz
K
,
Vatter
H
,
Gasser
T
,
Seifert
V
.
Intraoperative MRI guidance and extent of resection in glioma surgery: a randomised, controlled trial
.
Lancet Oncol
 .
2011
;
12
(
11
):
997
-
1003
.
35.
Louis
DN
,
Ohgaki
H
,
Wiestler
OD
et al
The 2007 WHO classification of tumours of the central nervous system
.
Acta Neuropathol
 .
2007
;
114
(
2
):
97
-
109
.
36.
Coburger
J
,
Konig
RW
,
Scheuerle
A
et al
Navigated high frequency ultrasound: description of technique and clinical comparison with conventional intracranial ultrasound
.
World Neurosurg
 .
2014
;
82
(
3-4
):
366
-
375
.
37.
Stummer
W
,
Novotny
A
,
Stepp
H
,
Goetz
C
,
Bise
K
,
Reulen
HJ
.
Fluorescence-guided resection of glioblastoma multiforme by using 5-aminolevulinic acid-induced porphyrins: a prospective study in 52 consecutive patients
.
J Neurosurg
 .
2000
;
93
(
6
):
1003
-
1013
.
38.
Miwa
K
,
Shinoda
J
,
Yano
H
et al
Discrepancy between lesion distributions on methionine PET and MR images in patients with glioblastoma multiforme: insight from a PET and MR fusion image study
.
J Neurol Neurosurg Psychiatry
 .
2004
;
75
(
10
):
1457
-
1462
.
39.
Kamp
MA
,
Krause Molle
Z
,
Munoz-Bendix
C
et al
Various shades of red-a systematic analysis of qualitative estimation of ALA-derived fluorescence in neurosurgery
.
Neurosurg Rev
 .
2016
.
Epub ahead of print.
40.
Dubois
LG
,
Campanati
L
,
Righy
C
et al
Gliomas and the vascular fragility of the blood brain barrier
.
Front Cell Neurosci
 .
2014
;
8
:
418
.
41.
Ennis
SR
,
Novotny
A
,
Xiang
J
et al
Transport of 5-aminolevulinic acid between blood and brain
.
Brain Res
 .
2003
;
959
(
2
):
226
-
234
.
42.
Roder
C
,
Bender
B
,
Ritz
R
et al
Intraoperative visualization of residual tumor: the role of perfusion-weighted imaging in a high-field intraoperative MR scanner
.
Neurosurgery
 .
2012
;
72
(
2 Suppl Operative
):
ons151
-
158
;
discussion ons158
.
43.
Zhang
N
,
Zhang
L
,
Qiu
B
,
Meng
L
,
Wang
X
,
Hou
BL
.
Correlation of volume transfer coefficient Ktrans with histopathologic grades of gliomas
.
J Magn Reson Imaging
 .
2012
;
36
(
2
):
355
-
363
.
44.
Roder
C
,
Skardelly
M
,
Ramina
KF
et al
Spectroscopy imaging in intraoperative MR suite: tissue characterization and optimization of tumor resection
.
Int J Comput Assist Radiol Surg
 .
2014
;
9
(
4
):
551
-
559
.
45.
Nguyen
A
,
Legrain
M
,
Noel
G
et al
An innovative fluorescent semi-quantitative methylation-specific PCR method for the determination of MGMT promoter methylation is reflecting intra-tumor heterogeneity
.
Curr Cancer Drug Targets
 .
2015
;
15
(
7
):
624
-
640
.