Secretory and nonsecretory pituitary adenomas are distinguishable by 1/T1 magnetic relaxation rates at very low magnetic fields in vitro.

RATIONALE AND OBJECTIVES: The authors investigated whether hormonally active and inactive pituitary adenomas can be discriminated in vitro by magnetic resonance (MR) imaging-related data. METHODS: 1/T1 nuclear magnetic relaxation dispersion profiles were measured for 39 fresh surgical specimens of secreting and nonsecreting adenomas, classified using clinical criteria or preoperative serum hormone levels. Nonsecreting adenomas were subdivided into hormone-producing and nonhormone-producing by immunostains. At five fields (0.00024 to 1.2 tesla [T]), mean 1/T1 was analyzed for statistically significant differences among these three tumor categories. RESULTS: Mean 1/T1 was significantly higher (P < 0.02) for hormone-secreting than for nonsecreting adenomas at fields below 0.24 T; no significant difference existed at typical MR imaging fields (0.5 to 1.5 T). Mean 1/T1 for hormone-producing and nonhormone-producing, nonsecreting adenomas were not significantly different at any field. CONCLUSIONS: Because 1/T1 at low fields is related to 1/T2 at imaging fields, it may be possible to detect hormone secretion of pituitary adenomas noninvasively by MR imaging.

Correlation of relaxometry and histopathology: the transplantable human glioblastoma SF295 grown in athymic nude mice.

Human glioblastomas of the brain are characterized by a wide range of proton relaxation rates in vitro (1/T1 and 1/T2) and heterogeneous appearance in magnetic resonance imaging. It was previously found that their 1/T1 values vary widely at magnetic field strengths much below imaging fields, even at the same water content. In the present study, we measure 1/T1 at different magnetic field strengths (NMRD profile) for a specific transplantable, human glioblastoma (SF295), grown subcutaneously in athymic nude mice, to search for histologic characteristics that might correlate with the variability of 1/T1 at low fields (1/T1L). Using a field-cycling relaxometer, NMRD profiles were obtained for 32 fresh, histologically characterized, tumor specimens, 7 to 24 days post implantation of cryopreserved SF295 fragments. Tumor volume, dry weight, and pH of specimens were determined, the extent of hemorrhage and necrosis rated, and specimen location within the tumor recorded. A statistically significant increase in the average 1/T1 was found with increasing level of necrosis at 0.0024 T and below, possibly reflecting progressive protein aggregation in samples with up to 40% necrosis. This correlation was not significant at imaging fields. Although pH was increased in central necrosis, neither pH, dry weight, sample location, nor fresh hemorrhage could explain the changes in 1/T1L. The variability of 1/T1L among SF295 samples is much reduced compared to that of fresh surgical specimens of human glioblastomas of the brain. The heterogeneous appearance of glioblastomas in MRI may have a histologic correlate which reflects molecular changes involved with induction of cell death and necrosis. Further investigations may identify the factors responsible for affecting 1/T1L (hypoxia, radiation, chemotherapy).