The consistency of neuropathological diagnoses in patients undergoing surgery for suspected recurrence of glioblastoma

Abstract

PurposeClinical factors and neuro-imaging in patients with glioblastoma who appear to progress following standard chemoradiation are unable to reliably distinguish tumor progression from pseudo-progression. As a result, surgery is commonly recommended to establish a final diagnosis. However, studies evaluating the pathologists’ agreement on pathologic diagnoses in this setting have not been previously evaluated.MethodsA hypothetical clinical history coupled with images of histological sections from 13 patients with glioblastoma who underwent diagnostic surgery for suspected early recurrence were sent to 101 pathologists from 50 NCI-designated Cancer Centers. Pathologists were asked to provide a final diagnosis (active tumor, treatment effect, or unable to classify) and to report on percent active tumor, treatment effect, and degree of cellularity and degree of mitotic activity.ResultsForty-eight pathologists (48%) from 30 centers responded. In three cases > 75% of pathologists diagnosed active tumor. In two cases > 75% diagnosed treatment effect. However, in the remaining eight cases the disparity in diagnoses was striking (maximum agreement on final diagnosis ranged from 36 to 68%). Overall, only marginal agreement was observed in the overall assessment of disease status [kappa score 0.228 (95% CI 0.22–0.24)].ConclusionsConfidence in any clinical diagnostic assay requires that very similar results are obtained from identical specimens evaluated by sophisticated clinicians and institutions. The findings of this study illustrate that the diagnostic agreement between different cases of repeat resection for suspected recurrent glioblastoma can be variable. This raises concerns as pathological diagnoses are critical in directing standard and experimental care in this setting.Electronic supplementary materialThe online version of this article (10.1007/s11060-018-03037-3) contains supplementary material, which is available to authorized users.

Visualization of Diagnostic and Therapeutic Targets in Glioma With Molecular Imaging

“…For example, although pathological diagnosis is considered the gold standard, it provides molecular characterization of the glioma at a single snapshot in time (e.g., prior to chemoradiation, or in the case of recurrent disease, after multiple treatments including chemoradiation) and is limited in scope to the tumor region sampled by neurosurgeon. In addition, multiple reports have demonstrated inter-rater variability for glioma pathology diagnosis among trained experts, and the superiority of molecular and genetic profiles compared to histological analyses for prediction of overall survival (OS) in patients with glioma (3,4). Instead, by implementing an advanced MI-based approach, the molecular marker status of tumors could be interrogated repeatedly in vivo over the course of the patient's treatment regimens.…”

Perfusion MRI-Based Fractional Tumor Burden Differentiates between Tumor and Treatment Effect in Recurrent Glioblastomas and Informs Clinical Decision-Making

“…Because glioblastomas are markedly heterogeneous tumors, histopathologic diagnosis of samples from separate locations may differ; even with a given sample, diagnostic agreement among pathologists can vary. 24 However, despite the real possibility of sampling error, histopathologic assessment remains the "criterion standard" for determining disease status and often influences treatment-related decision-making. Taking these issues into consideration, we show the potential of FTB to distinguish TE from tumor, and we have yet to fully explore its role as an alternative means to provide accurate diagnostic and prognostic information, particularly given the limitations of histopathologic evaluation.…”

Congress of Neurological Surgeons Systematic Review and Evidence-based Guidelines Update on the Role of Neuropathology in the Management of Progressive Glioblastoma in Adults

“…This was particularly emphasized by the study conducted by Holdhoff, et al, in which only "marginal agreement" by Fleiss' kappa statistics was observed when 48 pathologists (92% of whom were neuropathologists) reviewed 13 cases of suspected recurrence of glioblastoma. [14] Much of this may be due to differential understanding of the terminology used in this this study. The terminology of "active tumor", "inactive tumor", and "treatment effect" are not well established in the literature [14] yet this is commonly used terminology.…”