The purpose of this study was to evaluate the prevalence and characterize the patterns of hilar uptake (HU) on 67Ga-citrate imaging after cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) chemotherapy regimens for non-Hodgkin's lymphoma (NHL), to differentiate hilar lymphoma (HL) from HU of benign etiology.
Methods: A total of 930 studies (698 planar, 232 thoracic SPECT) was reviewed retrospectively in 100 NHL patients (29 low-grade, 60 intermediate-grade, and 11 high-grade) treated with CHOP and followed up longitudinally with serial gallium studies (planar: median, 7; range, 3-16 studies in 100 patients; SPECT: median, 1; range, 0-11 studies in 72 patients) over a median duration of 36 mo (range, 6-112 mo) from diagnosis. Clinical outcome and size changes over time on correlative CT and/or radiographs were used to evaluate benign versus malignant changes within the hila.
Results: HU after CHOP was present in 79% of patients (90% confidence interval [CI], 71%-85%), with 33% showing HU on SPECT alone. Once present, HU persisted for a median of 27 mo (range, 2-84 mo) from onset. The prevalence of HU and HL at various time points was as follows: baseline HU, 52% with HL 60%; mid-CHOP HU, 59% with HL2%; post-CHOP HU, 52% with HL6%; follow-up HU, 76% with HL 9%. HU of benign etiology was not significantly correlated with CHOP dosage. HU was symmetric in 90% of patients (90% CI, 82%-95%) and less intense than the original disease in 89% of patients (90% CI, 80%-95%), and these features were highly predictive of benign etiology (negative predictive value [NPV], 98.6% if symmetric; NPV, 96.5% if less than original disease; NPV, 100% if both present). Asymmetric HU equal in intensity to the original disease, however, was highly predictive of HL (positive predictive value [PPV], 87.5% if asymmetric; PPV, 85.7% if equal to original disease; PPV, 100% if both present).
Conclusion: HU after CHOP is common (overall incidence, 79%), often seen only on SPECT, and most likely of benign etiology when symmetric and less intense than the original disease. Asymmetric HU that equals the intensity of the original disease, however, is a possible indicator for HL.