CONTEXT: Although the cytologic features of leukemia/lymphoma in cerebrospinal fluid specimens are well known, the correlation of these features with the ability of cytologists to identify this tumor have not been well studied. OBJECTIVE: To identify the morphologic features of leukemia/lymphoma in cerebrospinal fluid that are associated with good performance and poor performance in an educational interlaboratory comparison program; and to identify the morphologic features associated with how well a slide performs with regard to its reference diagnosis. DESIGN: The performance of 147 cases of leukemia/lymphoma in the College of American Pathologists Interlaboratory Comparison Program in Non-gynecologic Cytology was analyzed. The cytologic features of a subset of 31 cases composed of relatively equal numbers of Romanowsky-stained and Papanicolaou-stained specimens were further evaluated, and those that performed poorly (n = 12) were compared with those that performed extremely well (n = 19). RESULTS: For all cases of leukemia/lymphoma in the program, the rate of misclassification as benign for cases with Papanicolaou stain was significantly higher than for those with Romanowsky stain (9.5% vs 2.6%, P < .001). Compared with cases that performed well, slides that performed poorly were more likely to have less than 200 abnormal cells (42% vs 5%, P = .02). The size of the tumor cells and preservation were not significant. CONCLUSION: Cases of specimens of leukemia/lymphoma in cerebrospinal fluid are more likely to be misdiagnosed as benign if they are Papanicolaou-stained or have 200 abnormal cells.
CONTEXT: Although the cytologic features of leukemia/lymphoma in cerebrospinal fluid specimens are well known, the correlation of these features with the ability of cytologists to identify this tumor have not been well studied. OBJECTIVE: To identify the morphologic features of leukemia/lymphoma in cerebrospinal fluid that are associated with good performance and poor performance in an educational interlaboratory comparison program; and to identify the morphologic features associated with how well a slide performs with regard to its reference diagnosis. DESIGN: The performance of 147 cases of leukemia/lymphoma in the College of American Pathologists Interlaboratory Comparison Program in Non-gynecologic Cytology was analyzed. The cytologic features of a subset of 31 cases composed of relatively equal numbers of Romanowsky-stained and Papanicolaou-stained specimens were further evaluated, and those that performed poorly (n = 12) were compared with those that performed extremely well (n = 19). RESULTS: For all cases of leukemia/lymphoma in the program, the rate of misclassification as benign for cases with Papanicolaou stain was significantly higher than for those with Romanowsky stain (9.5% vs 2.6%, P < .001). Compared with cases that performed well, slides that performed poorly were more likely to have less than 200 abnormal cells (42% vs 5%, P = .02). The size of the tumor cells and preservation were not significant. CONCLUSION: Cases of specimens of leukemia/lymphoma in cerebrospinal fluid are more likely to be misdiagnosed as benign if they are Papanicolaou-stained or have 200 abnormal cells.