Random skin biopsies prior to brain biopsy for intravascular large B-cell lymphoma.

OBJECTIVE: To evaluate effectiveness of random skin biopsies for intravascular large B-cell lymphoma (IVLBCL) with or without central nervous system (CNS) involvement. METHODS: Data from 21 patients with suspected IVLBCL (7 with CNS involvement and 14 without CNS involvement) who underwent single (4 patients), double (1 patient), and random (16 patients) skin biopsies were retrospectively analyzed. RESULTS: IVLBCL was diagnosed in 16 patients (including 6 with CNS involvement). Sensitivity, specificity, and positive predictive value of random skin biopsies were 75%, 100%, and 100%. Ratio of tumor-positive biopsy samples to plasma soluble interleukin-2 receptor (sIL-2R) values was significantly correlated in cases with data on both variables. sIL-2R values in the 6 tumor-negative skin samples (median, 1415 U/mL; range, 487-3200 U/mL) were significantly lower than in tumor-positive skin samples (median, 3550 U/mL; range, 595-8700 U/mL) with at least 1 skin specimen obtained. Mean ratio of tumor-positive biopsy samples in IVLBCL cases with low sIL-2R (< 3000 U/mL) was only 45%, indicating a requirement for 3-site multiple sampling. No differences in median sIL-2 values between cases of IVLBCL with and without CNS involvement were found (2795 U/mL vs. 3550 U/mL). Steroids administered before diagnosis yielded false-negative results in 3 of 5 IVLBCL cases (all false-negative cases were IVLBCL with CNS involvement), whereas none of 11 IVLBCL cases without steroid administration yielded false-negative results. CONCLUSIONS: Random skin biopsies before brain biopsy are recommended in patients with suspected IVLBCL regardless of CNS involvement, but low sIL-2R values and steroids may yield false-negative results.