Search

The difficult differential diagnosis between the diffuse variants of cutaneous lymphoid hyperplasia (CLH; synonym; pseudolymphoma) and malignant follicular center cell lymphomas (FCCL) often requires a multidisciplinary approach. Eighteen CLH and 11 FCCL, diagnosed by conventional histology and immunophenotyping and subsequently examined with a polymerase chain reaction to show clonal immunoglobulin heavy-chain gene rearrangements, were subjected to a novel type of automated nuclear image analysis. Of all nuclear parameters tested in azure A-stained semithin sections, the mean nuclear profile area (TN) of lymphoid cells was the best criterion to distinguish between CLH and FCCL (p = 9 x 10(-6)). Additional distinctive features, in the order of decreasing significance, were the SD of TN; all chromatin textural parameters combined; and the light and the dark fractions of the central nuclear profile areas. Parameters related to the chromatin pattern were independent of nuclear profile size in FCCL, but not in CLH. Two lesions registered as CLH displayed the nuclear characteristics favoring this diagnosis, but showed B-cell monoclonality at the DNA level. In conclusion, computerized nuclear image analysis is a helpful additional diagnostic tool in the evaluation of diffuse CLH and cutaneous FCCL.

To evaluate observer variability during microcomputer-assisted planimetry, nuclear features (area, perimeter, and form factor) were studied in a case of malignant lymphoma. Tissue fixed in glutaraldehyde, embedded in plastic and sectioned at 1 micron was studied using a Zeiss Kontron IBAS 2000 image analysis system. Manual tracing of nuclear outline was performed to obtain the parameters. Both interindividual and intraindividual reproducibilities were evaluated, the former among 8 observers and the latter by the observers repeating the measurements. With the present technique, intraindividual reproducibility was high and correlation was excellent for area, perimeter and form factor measurements. A high degree of interindividual inconsistency was demonstrated for all parameters. Interindividual correlation was poor for form factor. It is concluded that poor interindividual reproducibility even when the objective method of morphometry is used may be a significant problem in tissue investigations, especially when results from different laboratories are being compared.