5-Aminolevulinic acid-derived tumor fluorescence: the diagnostic accuracy of visible fluorescence qualities as corroborated by spectrometry and histology and postoperative imaging

Abstract

Background: 5-Aminolevulinic acid is used for fluorescence-guided resections. During resection, different macroscopic fluorescence qualities ("strong," "weak") can be distinguished that help guide resections.

Objective: This prospective study was designed to assess the reliability of visible fluorescence qualities by spectrometry, pathology, and imaging.

Methods: Thirty-three patients with malignant gliomas received 5-aminolevulinic acid (20 mg/kg). After debulking surgery, standardized biopsies were obtained from tissues with "weak" and "strong" fluorescence and from nonfluorescing near and distant brain for blinded assessment of cell density and tissue type (necrosis, solid or infiltrating tumor, normal tissue). The positive predictive value was calculated. Unresected fluorescing tissue was navigated for blinded correlation to postoperative magnetic resonance imaging (MRI). Receiver operating characteristic curves were generated for assessing the classification efficiency of spectrometry.

Results: "Strong" fluorescence corresponded to greater spectrometric fluorescence, solidly proliferating tumor, and high cell densities, whereas "weak" fluorescence corresponded to lower spectrometric fluorescence, infiltrating tumor, and medium cell densities. The positive predictive value was 100% in strongly fluorescing tissue and 95% in weakly fluorescing tissue. Spectrometric fluorescence was detected in marginal tissue without macroscopic fluorescence. Depending on the threshold, spectrometry displayed greater sensitivity but lower specificity (accuracy 88.4%). Residual MRI enhancement in the tumor bed was detected in 15 of 23 (65%) patients with residual fluorescence, but in none of the patients without residual fluorescence.

Conclusion: Macroscopic fluorescence qualities predict solid and infiltrating tumor, providing useful information during resection. Fluorescence appears superior to contrast enhancement on MRI for indicating residual tumor. Spectrometry, on the other hand, is more sensitive but less specific, depending on threshold definition.

Abbreviations: 5-ALA, 5-aminolevulinic acidCI, confidence intervalgamma-GT, gamma-glutamyl transpeptidaseGBM, glioblastoma multiformeNPV, negative predictive valuePPIX, protoporphyrin IXPPV, positive predictive valueSD, standard deviationWHO, World Health Organization.

FIGURE 1

Intraoperative fluorescence image with spectrometric measurement sites. A, cavity with area of strong (red), weak (pink; separated by white line) and no fluorescence. B, corresponding white light image with biopsy/measurement sites. Nonfluorescent “near” sampling sites were immediately at the border to fluorescent tissue, nonfluorescent “distant” sampling sites 1 cm away from the border. C, example of raw spectra obtained in a single patient from regions with “strong” and “weak” fluorescence, and tissue without fluorescence with “near” and “distant” measurements. The first peak corresponds to remitted excitation light, the second peak to tissue autofluorescence, and the third and fourth peaks to the specific spectrogram for PPIX. Raw spectra were normalized relative to a fluorescence standard for final analysis. PPIX, protoporphyrin IX.

FIGURE 2

Infiltrating tumor regions with low cell density. Immunohistochemistry was used for unequivocal identification of tumor cells. Left, MIB-1 (Ki-67) staining. Right, p53 staining.

FIGURE 3

Relationship between visual (macroscopic) fluorescence and spectrometric fluorescence. Multiple measurements for fluorescence in single patients at comparable locations were averaged for this analysis. These values are stratified by individual fluorescence qualities as perceived visually through the surgical microscope (“strong,” n = 96; “weak,” n = 90) or nonfluorescing adjacent tissue (“near,” n = 66; “distant,” n = 66). Boxes signify medians, 25th and 75th percentiles; whiskers signify ranges.

FIGURE 4

Relationship between visual (macroscopic) fluorescence and cell densities. Multiple measurements for fluorescence in single patients at comparable locations were averaged for this analysis. These values are stratified by individual fluorescence qualities as perceived visually through the surgical microscope (“strong,” n = 96; “weak,” n = 90) or nonfluorescing adjacent tissue (“near,” n = 66; “distant,” n = 48). Boxes signify medians, 25th and 75th percentiles; whiskers signify ranges.

FIGURE 5

Tissue differentiation stratified by visible fluorescence qualities and frequency of biopsies obtained per fluorescence quality (“strong,” n = 96; “weak,” n = 90) or nonfluorescing adjacent tissue (“near,” n = 66; “distant,” n = 48). The ordinate gives the percentage of biopsies from a given fluorescence quality with a particular tissue morphology.

FIGURE 6

In order to compare the results from different patients operated on by different surgeons at different sites regarding the reproducibility of results, this graph relates spectrometric fluorescence to histological cell densities stratified by fluorescence quality (“strong,” “weak”). For this graph, all measurements obtained from the same fluorescence quality in a single patient were averaged (n = 33). “Strong” visual fluorescence was related to high cell densities and strong spectrometric fluorescence, whereas “weak” fluorescence was related to lower cell densities and low spectrometric fluorescence. There was little overlap.

FIGURE 7

Receiver operating characteristic (ROC) curves for assessing the classification efficiency of spectrometry. The area under the curve (AUC) was 0.88. Threshold values are indicated in the curve for resulting specificities and sensitivities.

Figure

No available caption