The first validated classification criteria for discoid lupus erythematosus has a sensitivity that ranges between 73.9% and 84.1% and a specificity that ranges between 75.9% and 92.9%.
“Discoid lupus erythematosus [DLE] is the most common type of chronic cutaneous lupus,” lead study author, said during the virtual annual meeting of the American Academy of Dermatology. “It’s one of the most potentially disfiguring forms of cutaneous lupus erythematosus [CLE], which can lead to scarring, hair loss, and dyspigmentation if not treated early or promptly. It has a significant impact on patient quality of life and there are currently no classification criteria for DLE, which has led to problematic heterogeneity in observational and interventional research efforts. As there is increasing interest in drug development programs for CLE and DLE, there is a need to develop classification criteria.”
Dr. Elman, of the Harvard combined medicine-dermatology training program at Brigham and Women’s Hospital, Boston, pointed out that classification criteria are the standard definitions that are primarily intended to enroll uniform cohorts for research. “These emphasize high specificity, whereas diagnostic criteria reflect a more broad and variable set of features of a given disease, and therefore require a higher sensitivity,” he explained. “While classification criteria are not synonymous with diagnostic criteria, they typically mirror the list of criteria that are used for diagnosis.”
In 2017, Dr. Elman and colleagues generated an item list of 12 potential classification criteria using an international Delphi consensus process: 5 criteria represented disease morphology, 2 represented discoid lupus location, and 5 represented histopathology (). The purpose of the current study, which was presented as a late-breaking abstract, was to validate the proposed classification criteria in a multicenter, international trial. “The point is to be able to differentiate between discoid lupus and its disease mimickers, which could be confused in enrollment in clinical trials,” he said.
At nine participating sites, patients were identified at clinical visits as having either DLE or a DLE mimicker. After each visit, dermatologists determined if morphological features were present. One dermatopathologist at each site reviewed pathology, if available, to see if the histopathologic features were present. Diagnosis by clinical features and dermatopathology were tabulated and presented as counts and percentages. Clinical features among those with and without DLE were calculated and compared with chi-square or Fisher’s exact tests. The researchers used best subsets logistic regression analysis to identify candidate models.
A total of 215 patients were enrolled: 94 that were consistent with DLE and 121 that were consistent with a DLE mimicker. Most cases (83%) were from North America, 11% were from Asia, and 6% were from Europe. Only 86 cases (40%) had biopsies for dermatopathology review.
The following clinical features were found to be more commonly associated with DLE, compared with DLE mimickers: atrophic scarring (83% vs. 24%; P < .001), dyspigmentation (84% vs. 55%; P < .001), follicular hyperkeratosis/plugging (43% vs. 11%; P < .001), scarring alopecia (61% vs. 21%; P < .001), location in the conchal bowl (49% vs. 10%; P < .001), preference for the head and neck (87% vs. 49%; P < .001), and erythematous to violaceous in color (93% vs. 85%, a nonsignificant difference; P = .09).
When histopathological items were assessed, the following features were found to be more commonly associated with DLE, compared with DLE mimickers: interface/vacuolar dermatitis (83% vs. 53%; P = .004), perivascular and/or periappendageal lymphohistiocytic infiltrate (95% vs. 84%, a nonsignificant difference; P = .18), follicular keratin plugs (57% vs. 20%; P < .001), mucin deposition (73% vs. 39%; P = .002), and basement membrane thickening (57% vs. 14%; P < .001).
“There was good agreement between the diagnoses made by dermatologists and dermatopathologists, with a Cohen’s kappa statistic of 0.83,” Dr. Elman added. “Similarly, in many of the cases, the dermatopathologists and the dermatologists felt confident in their diagnosis.”
For the final model, the researchers excluded patients who had any missing data as well as those who had a diagnosis that was uncertain. This left 200 cases in the final model. Clinical variables associated with DLE were: atrophic scarring (odds ratio, 8.70; P < .001), location in the conchal bowl (OR, 6.80; P < .001), preference for head and neck (OR, 9.41; P < .001), dyspigmentation (OR, 3.23; P = .020), follicular hyperkeratosis/plugging (OR, 2.94; P = .054), and erythematous to violaceous in color (OR, 3.44; P = .056). The area under the curve for the model was 0.91.
According to Dr. Elman, the final model is a points-based model with 3 points assigned to atrophic scarring, 2 points assigned to location in the conchal bowl, 2 points assigned to preference for head and neck, 1 point assigned to dyspigmentation, 1 point assigned to follicular hyperkeratosis/plugging, and 1 point assigned to erythematous to violaceous in color. A score of 5 or greater yields a classification as DLE with 84.1% sensitivity and 75.9% specificity, while a score of 7 or greater yields a 73.9% sensitivity and 92.9% specificity.
Dr. Elman acknowledged certain limitations of the study, including the fact that information related to histopathology was not included in the final model. “This was a result of having only 40% of cases with relevant dermatopathology,” he said. “This limited our ability to meaningfully incorporate these items into a classification criteria set. However, with the data we’ve collected, efforts are under way to make a DLE-specific histopathology classification criteria.”
Another limitation is that the researchers relied on expert diagnosis as the preferred option. “Similarly, many of the cases came from large referral centers, and no demographic data were obtained, so this limits the generalizability of our study,” he said.
Dr. Elman reported having no financial disclosures.