Detection of metastatic neuroblastoma in bone marrow: when is routine marrow histology insensitive?

PURPOSE To measure the sensitivity of histologic examination in detecting metastatic solid tumor in bone marrow. PATIENTS AND METHODS A total of 145 patients with stage 4 neuroblastoma underwent 840 marrow examinations, each consisting of six sites (four aspirates and two biopsies), from October 1990 to June 1996 at Memorial Sloan-Kettering Cancer Center. Metastasis was detected by either histology (aspirate by Wright-Giemse and biopsy by Hematoxylin-Eosin stains) or immunostaining of aspirates using anti-G(D2) monoclonal antibodies. RESULTS The absence of tumor by histology at a single marrow site was a poor guarantee of the absence of disease. The number of false-negative sites increased as the percent of G(D2)-positive tumor cells in the marrow decreased: zero of six if tumor cell count was > or = 1%, and approximately six of six sites if < or = 0.003%. Sensitivity was comparable between marrow aspirate and biopsy. A lower bound (LB) for the probability of false-negative histology was calculated from the (1) discordance among the six marrow samplings and (2) comparison with immunofluorescence. When disease was extensive (eg, at diagnosis), the LB was 0.13 and 0.3, respectively. After treatment, it increased to 0.37 and 0.8. Examining multiple marrow sites can decrease the LB to 0.15. However, at least three sites have to be negative at relapse, six at diagnosis, and more than 50 during treatment or off-therapy follow-up. The marginal decrease in the LB by additional samplings rapidly diminished to less than 0.05 after two sites. CONCLUSION Except at diagnosis and relapse when gross disease is present, marrow sampling by histology has limited sensitivity. Current practice grossly underestimates the true prevalence of marrow disease.