Diffusion-weighted imaging of orbital masses: multi-institutional data support a 2-ADC threshold model to categorize lesions as benign, malignant, or indeterminate

Abstract

Background and purpose: DWI has been increasingly used to characterize orbital masses and provides quantitative information in the form of the ADC, but studies of DWI of orbital masses have shown a range of reported sensitivities, specificities, and optimal threshold ADC values for distinguishing benign from malignant lesions. Our goal was to determine the optimal use of DWI for imaging orbital masses through aggregation of data from multiple centers.

Materials and methods: Source data from 3 previous studies of orbital mass DWI were aggregated, and additional published data points were gathered. Receiver operating characteristic analysis was performed to determine the sensitivity, specificity, and optimal ADC thresholds for distinguishing benign from malignant masses.

Results: There was no single ADC threshold that characterized orbital masses as benign or malignant with high sensitivity and specificity. An ADC of less than 0.93 × 10(-3) mm(2)/s was more than 90% specific for malignancy, and an ADC of less than 1.35 × 10(-3) mm(2)/s was more than 90% sensitive for malignancy. With these 2 thresholds, 33% of this cohort could be characterized as "likely malignant," 29% as "likely benign," and 38% as "indeterminate."

Conclusions: No single ADC threshold is highly sensitive and specific for characterizing orbital masses as benign or malignant. If we used 2 thresholds to divide these lesions into 3 categories, however, a majority of orbital masses can be characterized with >90% confidence.

Fig 1.

Lesion distribution by category. The left column shows the relative proportion of lesions encountered in all published studies of orbital DWI (A), in this analysis (B), and by Shields et al during a 30-year period (C). The published literature and our analysis group contain a higher proportion of lymphoma cases than would be predicted by Shields et al. Otherwise, the relative proportion of lesions across these 3 groups is similar, as is seen after exclusion of lymphomas (right column).

Fig 2.

Scatterplot of ADC by lesion category shows consistently low ADC for lymphoma and a wide distribution of ADC for nonlymphoma malignancies.

Fig 3.

Comparison between orbital lymphoma (A and B) and orbital inflammatory disease (C and D). A, Axial DWI shows a right orbital mass with marked hyperintensity. B, Corresponding axial ADC map shows dark signal, indicating low ADC and hypercellularity typical of lymphoma. ADC of this lesion was 0.65 × 10⁻³ mm²/s. C, Axial DWI in a different patient shows a lacrimal gland mass with less intense signal compared with A. D, Corresponding axial ADC map shows intermediate signal, brighter than adjacent brain parenchyma, reflective of the lower cellularity seen in orbital inflammatory lesions. The ADC of this lesion was 1.09 × 10⁻³ mm²/s.

Fig 4.

Receiver operating characteristic curve analysis of ADC for distinguishing benign from malignant masses (A) shows high specificities with increasing sensitivity up to 60%, at which point additional gains in sensitivity are offset by larger losses in specificity. When lymphomas are removed from the analysis (B), the performance of ADC diminishes.