We acknowledge our review to be subject to important limitations. Notably, the quality of studies we reviewed limits the strength of our findings: of the 36 studies reviewed, none had a score of 7 or 8, and 89% had scores ≤ 5/8. More specifically, only one of the studies reporting response to GA treatment adequately ruled out the possibility of spontaneous GA remission. Indeed, spontaneous resolution of GA has been reported in almost 2% of GA patients; thus, it is possible that some of the reports we reviewed represent instances of spontaneous GA remission rather than bona fide responses to treatment. Moreover, only two studies reported a dose–response effect to treatment and only one described a challenge/rechallenge phenomenon. Altogether then, the reviewed studies were weakest in the 'causality' domain defined by Murad and colleagues, suggesting that the presented therapeutic options are supported largely by correlation but not causation. Additionally, while the majority (81%) of the studies we reviewed employed a satisfactory inclusion criterion (i.e. biopsy-proven GA), some studies did not specify the precise inclusion criteria while others included GA patients diagnosed only by clinical examination. As previously discussed, histopathological analysis is crucial for establishing a definitive GA diagnosis, since GA can mimic (and be mimicked by) a host of other conditions. Thus, some of the cases included in our review may not be GA but rather other dermatologic entities masquerading as GA.
Our review is also limited by our methodology of study selection. As we only included published studies, publication bias represents a significant limitation of our review. As a consequence, ostensibly efficacious treatments may merely be a reflection of the overrepresentation of positive outcomes in the literature rather than their true therapeutic potential. While the larger retrospective studies we include in this review partially offset this bias, we also include a number of case reports and case series of 10 or fewer patients that limit our conclusions. Selection bias is an additional limitation of our review, as we excluded studies not published in English.
Furthermore, while we strived to report dosage and treatment duration whenever possible, some studies did not report this information, rendering comparison of studies difficult. Finally, the lack of a ubiquitously utilized scale to measure GA severity resulted in heterogeneity in the manner in which response rates to treatment were reported (e.g. differing definitions of 'partial' and 'complete' resolution across studies), precluding a direct side-by-side comparison of each treatment option. Although a GA Investigator Global Assessment (IGA) was designed by Min and Lebwohl in 2016, none of the studies we reviewed employed this scale in evaluating the response of GA to treatment.
Am J Clin Dermatol. 2022;23(1):37-50. © 2022 Adis Springer International Publishing AG