LAUSR

In this issue of JAMA Dermatology, 2 large, well-constructed studies remind dermatologists that even though pyoderma gangrenosum (PG) is a quintessential disease in medical dermatology, we still need data to better define the disease and reduce the risk of cognitive biases. We know the results of these biases well; even at referral centers and in clinical trials, alternative diagnoses can prevail 7% to 20% of the time.1,2 In these articles, one of the largest retrospective PG databases and a new tool for PG diagnosis validation are presented.3,4 In one study, Ashchyan et al4 demonstrate that age at PG diagnosis matters. Older patients are far more likely to have hematologic malignant neoplasms, dyscrasias, and inflammatory arthritis associated with PG. These findings fit with the epidemiologic characteristics of these diseases. The authors should be commended for confirming an underlying etiology for PG in more than 65% of their patients compared with the frequently cited 50%.5,6 They suggest that laboratory evaluations can be stratified based on patient age. However, while cost containment is important, this should be balanced with risk. A young patient with PG without evidence of inflammatory bowel disease still deserves a thorough workup for underlying comorbidities. While the study demonstrates that the incidence of leukemia and monoclonal gammopathies is lower in nongeriatric patients, the rate is still an impressive 5.7%, albeit lower than the 19.4% in those 65 years or older. In this disease, laboratory evaluations are not just to screen for associated comorbidities; they also can support the diagnosis and risk assessment, while excluding competing diagnoses. We should not forget that patients with PG face a 3-fold higher mortality compared with control patients.5 Aschyan et al4 used the Su et al7 criteria for the confirmation of their diagnosis of PG. One problematic criterion proposed by Su et al for the diagnosis of PG is the exclusion of all competing diagnoses. Consequently, the criteria ought to be 100% specific by definition (at least in hindsight).7 This herculean criterion does not acknowledge that exclusionary tests and their interpretation are not infallible. This diminishes its operability and gives reason for Maverakis et al3 to propose new criteria that do not rely on complete exclusion nor perfect performance of all exclusionary tests.3 Using a Delphi consensus of international experts, the authors3 emphasize the importance of histopathologic confirmation of a neutrophilic dermatosis with an additional array of 8 minor criteria. The consensus does point out a weakness of using expert opinion; while many experts agreed that hematologic disease ought to be included with inflammatory arthritis and inflammatory bowel disease as minor criteria, this did not reach the level of appropriate consensus. Aschyan et al4 show that hematologic cancer, monoclonal gammopathy of unknown significance, myelodysplastic syndrome, and polycythemia vera were associated in 10.7% of the total cases of PG (while other investigators have shown up to 20%),8 and this comorbidity ought to be reconsidered in successive iterations. While this tool is a substantial step forward, particularly in ensuring internal validity of the disease for clinical trials in PG, we believe the criteria are still in need of additional validation. The study validation was performed using published case reports as the gold standard, comparing cases of PG with infections, medium-vessel vasculitis, calciphylaxis, and venous leg ulcers. While the validation supported the value of the biopsy as a major criterion, this may not reflect clinical practice. For example, in one of the largest cohort studies,8 only 8 of 67 biopsies (12%) showed the neutrophilic infiltrate, compared with 78% of case reports used in this study.3 Furthermore, the choice of validation diseases may further inflate the value of the biopsy; medium-vessel vasculitis, calciphylaxis, and many mimicking fungal infections are diagnosed partially or completely through histopathologic evaluation. The authors acknowledge the risk for publication bias in their validation cohort, but it may present in more forms than just the biopsy results. For example, terms such as pathergy, cribriform, and undermining borders are anchoring descriptions in dermatology. It should not be surprising that in hindsight of the final diagnosis, these terms are written in case reports to support a diagnostic narrative for PG and not in those with an alternative diagnosis. Ultimately, this is a tremendous step forward and it very likely offers high internal validity for the purposes of clinical trials, but there is the risk of low external validity, particularly if only 12% of patients in clinical practice would meet the major criterion.8 Additional work is needed to validate this in real-world patients before its use in clinical trials. How would such a study be performed? Serendipitously, this new tool is being published in the same issue as the largest patient cohort to date. Maverakis et al, have you met Ashchyan et al?