Proton nuclear magnetic resonance spectroscopy (1H MRS) has proven useful in the metabolic characterization of neoplastic tissues both in vitro and in vivo. In vivo 1H MRS can be easily performed in conjunction with conventional MR imaging. Although 1H MRS studies of lymphoma have been performed both in cell lines and animal models, its usefulness as a differential diagnostic tool in lymphoma remains to be explored. T2-filtered high resolution proton magic-angle spinning (HRMAS) MRS of small tissue samples provides sufficient spectral resolution for the identification of chemical cell changes associated with malignancy. HRMAS data from 16 fresh frozen lymph nodes were studied. Histologically, 7 lymph nodes revealed grade I or II follicular lymphoma (FL) and 9 showed diffuse large B cell lymphoma (DLBCL). Each spectrum consisted of the accumulation of 64 scans with a CPMG sequence with an effective T2 delay of 32 ms and an overall experimental time of 6 minutes. Pattern recognition analysis was feasible in all cases except in one FL sample consisting almost exclusively of adipose tissue. Compared with DLBCL, FL cases showed increases in the signal intensity of resonances associted to taurine (3.43 ppm) and decreases in the signal intensity of resonances associated to alanine (1.47 ppm). The mean taurine/alanine ratio was 4.63 (95%SD 1.67) for FL and 1.41 (95%SD 0.69) for DLBCL (P = 0.004). The taurine/alanine ratio proved to be a powerful discriminator between FL and DLBCL at a 2.5 cutpoint, yelding a 90% specificity and an 83% sensitivity to identify DLBCL. Overlapping cases included a case of FL with a diffuse pattern and two cases of DLBCL evolving from indolent lymphomas. The 1H MRS assay proved that FL and DLBCL can be differentiated from a metabolomic standpoint and offers a unique way to explore the underlying biochemistry of malignant transformations. Unlike other techniques, MRS is potentially applicable in vivo and non-invasively in the diagnostic setting. The rapid assessment of nodes or masses before tissue sampling would allow preoperative decisions to be made more efficiently.
Disclosure: No relevant conflicts of interest to declare.