Differentiating mycosis fungoides lesions from their mimickers clinically and histologically

↵* Traits were graded from 0 to 3 in epidermotropism, spongiosis, parakeratosis, dermal infiltrate, atypical lymphocytes, and large atypical cerebriform cells and graded from 0 to 1 in Pautrier’s microabscess, pigment incontinence, dermal fibrosis, and adnexal involvement. Analysis was performed using Fisher’s exact test with a Bonferroni-corrected threshold of significance = 0.005 = 0.05/10). If a significant association found, follow-up testing was performed to assess the grade severity of a trait vis-à-vis the diagnosis, using ordinal logistic regression (using maximum likelihood fitting), in 2 ways: first, by including zero values in the response variable (0, 1, 2, 3; 5 models fitted), and second, by considering only non-zero cases (such as including only those patients exhibiting the trait). When narrowing the testing to the patch and plaque groups only, 2×2 contingency tables (Fisher’s exact test, 2-tailed) were used. Likewise, it is pertinent to know whether the grade severity of a trait differentiating MF from a non-MF diagnosis was also more pronounced in patch versus plaque patients. To this end, non-zero cases were excised, and contingency tables were analysed (9 2×2 tables with Fisher’s exact test and one 3×2 table analysed with a likelihood ratio test of the chi-square). Finally, for either the patch or plaque subgroup, 10 goodness-of-fit tests were performed, 1 per histological trait, to determine where the observed counts of presence/absence deviated from those expected by chance (such as 50% frequency of occurrence).