RESUMEN
PURPOSE: Heterogeneity within GBMs and variability of visualized fluorescence combine to confer practical limitations to the technique of optical imaging. A biometric analysis was planned to objectively ascertain and analyse this phenomenon METHODS: 25 adult glioblastoma subjects undergoing resection were prospectively accrued. Biopsies were taken from various parts of the tumor and safe peritumoral zones. White light (WL) and visualized fluorescence was subjectively recorded. Corresponding histopathology [coalescent (C) or infiltrating (I) tumor] and protoporphyrin-IX (PPIX) levels were assayed. RESULTS: WL was very sensitive for detecting tumor. SF was more specific and had high positive predictive value for detecting tumor. WF on the other hand had a poor discriminatory efficacy. Mean PPIX levels were 3.0, 2.01 and 0.16 for SF, WF, and NF respectively. WF had a wide variable range of PPIX levels. Within the coalescent tumor areas, there was a variable distribution of fluorescence (both subjective as well as objective PPIX levels) with only 54% samples showing SF and high PPIX. In seven cases this discordance was noted within the same tumor (biological heterogeneity). CONCLUSIONS: Fluorescence may miss important tumor areas even if objective assessment is used. Histologically similar tumor areas may exhibit contrasting fluorescence properties, a phenomenon which needs further investigation and elucidation of underlying mechanisms which could potentially be manipulated to optimize the utility of fluorescence guidance.