Evolving Role of Partial Volume Corrected FDG-PET Based Global Disease Assessment of Pulmonary Lesions in Patients with HIV-TB Coinfection

Objectives: Greater glycolytic activity on FDG-PET/CT has been associated with worse treatment outcomes in pulmonary tuberculosis (TB), but new methods to capture pulmonary enhancement systematically in subjects are needed. Partial volume effect (PVE) is a major factor that degrades image quality and hampers the accuracy of quantitative PET imaging. In this abstract, we used a novel software which provides automated partial volume corrected measures of glucose uptake from FDG-PET data.

Methods: We used Rover software (ABX, Radeburg, Germany), which has semiautomatic adaptive-contrast-oriented thresholding and local background partial-volume-correction algorithms, needed to capture lesional glycolytic activity in HIV-infected subjects with microbiologically proven pulmonary TB in Johannesburg, South Africa. Spherical masks were drawn for each subject, which encompassed the metabolically-active pulmonary lesions. Automatic ROI detection was employed using an initial threshold of 40% of the mask’s maximum. This threshold provides the starting condition for an iterative process by determining tentative 3D ROI. For each tentative ROI, a corresponding 3D background region was generated automatically in the close vicinity of the ROI. Starting from this initial estimate, a fully automatic procedure iteratively computes a background corrected threshold and the tentative ROI and corresponding background regions are modified accordingly. The iteration stops when all ROI volumes have converged. Subjects needed to be at least 18 years old with a new case of pulmonary TB and ready to start antiretroviral therapy (ART). To establish the clinical and biological relevance of Rover based lesional segmentation, we correlated total lesional glycolytic activity with clinically important biologic measures of TB disease, including lung function (forced expiratory volume in one second, FEV1, and forced vital capacity, FVC), CD4 count, HIV viral load, circulating IL-6, matrix metalloproteinase-8 and -9, and TNF-alpha levels prior to antiretroviral therapy initiation using spearman correlation coefficients (ρ).

Results: Forty-seven subjects who had FDG-PET/CT imaging prior to treatment with Antiretroviral Therapy (ART) were included in this analysis. The median age, CD4 count and HIV viral loads of the subjects were 37 (IQR 31-43), 84 (IQR 48-181), and 5.1 (IQR 4.7-5.6). Rover based lesional glycolytic activity was significantly negatively associated with FEV1 and FVC (ρ=-0.54 and -0.45 respectively, both p < 0.005), was positively associated with baseline CD4 count (ρ=0.43, p=0.002), MMP-8 (ρ=0.57, p < 0.001) and MMP-9 (ρ=0.37, p=0.01) , and IL-6 levels (ρ=0.43, p=0.003) and was negatively associated with HIV viral loads levels (-0.35, p=0.02).

Conclusions: Rover-based lung lesional glycolytic activity correlates with meaningful biomarkers of pulmonary TB in HIV-infected adults.