¹Department of Dermatology, Beth Israel Deaconess Medical Center, Boston, Massachusetts, USA

Correspondence: Dr Robert S Stern, Department of Dermatology, Beth Israel Deaconess Medical Center, 330 Brookline Avenue, GZ522, Boston, Massachusetts 02215, USA. E-mail: rstern@bidmc.harvard.edu

Abstract

The potential of newly introduced medications to cause acute inflammatory skin diseases has long been recognized. Less well accepted and not often systematically studied is the role of long-term drug use in inflammatory skin diseases such as eczema. Hypothesis-driven case–control studies provide what is probably the most efficient but far from easy method to evaluate and quantify observations from case reports and case series that suggest a drug etiology for chronic skin diseases.

Calcium channel blockers (CCBs) have long been recognized as causes of acute cutaneous eruptions and may exacerbate chronic skin conditions (Stern and Khalsa, 1989; Kitamura et al., 1993). In this issue of the Journal of Investigative Dermatology, Joly and colleagues have once again demonstrated the power, complexity, and limitations of three established approaches to determining the etiology of a specific skin condition that may be related to chronic use of medications: spontaneous reports, retrospective analysis of clinical experience (a case series), and a case–control study (Joly et al., 2007, this issue). As is appropriate, Joly et al. used observations from their institutions clinical experience to help frame the question for the case–control study that is the core of the findings presented in this issue of the JID. After establishing a clinically believable association, the authors used data from a hypothesis-driven retrospective review of responses to drug withdrawal and rechallenge from their own cases as well as data from French pharmacovigilance spontaneous reports to provide confirmatory evidence for the association noted in the case series and case–control study.

In assessing this work, three questions come to mind. Are the findings from each approach used robust enough to suggest that chronic use of CCBs can cause "eczematous" eruptions in the "elderly"? How helpful are these findings for clinical practice? What further studies are warranted?

The authors have carefully built on their earlier observations to solve a small part of a vexing problem—are there reversible causes of otherwise unexplained chronic extensive eczematous eruptions in the "elderly"? In a prior retrospective case series of older individuals with chronic eczema, Joly and colleagues found that older persons whose eczema was unexplained used more medicines than those whose "eczema" was attributable to one of the "usual suspects" (Morin et al., 2002). These unexplained eruptions occurred more frequently on sun-exposed areas, but seasonality was not noted. With this suggestive evidence that chronic use of drugs might explain some cases of eczema, Joly et al. designed a case–control study that prospectively ascertained cases of eczema among people 60 years old and older.

Three of the top five "known" reasons for new-onset extensive eczema among candidate cases were infrequent diseases in the general population—lymphoma, scabies, and pemphigoid—suggesting that the study population is not likely to be representative of all elderly with new-onset eczema but is more likely to represent those served by specialized university-based clinics, which may affect the generalizability of these findings to patients seen in less specialized practice settings.

By itself, this case–control study provides good evidence (level of evidence 2B) of a modest but significant association between CCB use and new-onset eczema (odds ratio = 2.5). The authors attempt to bolster our confidence in the validity of these findings with "ecological" (noncontrolled) studies, which they very reasonably title "ancillary studies." In the retrospective analysis of a separate population from the same principal center, they note that the withdrawal of CCBs was more likely to be followed by relief than withdrawal of other drugs that had been suspected as causes of chronic eczema eruptions. Case histories from the French "enhanced" spontaneous reporting system were supportive. Most of the small number of rechallenges were also positive, indicating that, at least for a small fraction of those for whom CCB use was suspected, the gold standard for "proving" drug etiology in the individual patients was fulfilled.

However, potential bias in selection of patients for reporting, dechallenge, and rechallenge, particularly in "spontaneous reporting systems," makes interpretation of these results difficult. In such systems, successes are more likely to be reported than failures. For example, in the pharmacovigilence data, 83% recovered after stopping CCBs. Based on the data from the case control study, which is likely to be less subject to reporting bias than "spontaneous reports," given the high proportion of elderly using CCBs and the relatively low relative risk of "unexplained" eczema associated with CCB use, we would have expected only about half of the cases of "unexplained" eczema among CCB users to be due to the drug, with the remaining half of such cases not related to CCB use.

Potential additional issues concerning the robustness of the finding are related to case definition, selection of controls, and possible informational bias. Although I have a mental picture of the eruption based on the data presented, I am not sure it is a correct one. For the results to be clinically useful and replicable, a clearer definition of "cases" with more information about the extent, distribution, morphology, and perhaps pathology of the eruption would help us apply these findings more accurately. Comparison of the characteristics of "elderly" with unexplained eczema using CCBs, particularly those who recover with CCB withdrawal, with those of nonusers ascertained for the case–control study would be particularly informative.

In case–control studies, selection of appropriate controls is often challenging and a potential source of bias. Finding the "right" controls is particularly difficult when cases come from a specialized population. In this study, cases appear to be sicker than controls. Cases used more medications than controls, including antiarrhythmia agents and nitrates. Although the association of these drugs with eczema did not reach statistical significance, it approached that observed with CCBs. These observations raise the possibility that either underlying cardiac illness or multiple-drug use, rather than CCB use alone, may be responsible for some cases attributed to CCBs.

In my experience, obtaining and confirming data on chronic past drug use is very challenging. The steps the investigators took to establish chronic drug use were comprehensive and likely to be impractical in the United States. A comparison of the drugs ascertained through each of the three methods used by the authors would be instructive. However, the prevalence of drug use in the studys controls is consistent with that observed in other populations, suggesting that controls were well selected and informational bias was limited (Kaufman et al., 2002).

The authors have done an admirable job of convincing me that in about half of older individuals taking CCBs with new onset of an extensive and perhaps photodistributed chronic eczematous eruption not otherwise explained after careful workup, the eruption is due to CCBs. How generalizable are these results to practice? The data presented suggest that CCBs are responsible for only about 15% of chronic, otherwise unexplained eczema in the elderly. The broad net required (5 academic centers and 35 practices over 2 years) to ascertain 102 cases suggests that unexplained new-onset eczema is infrequent in this age group and may be even more infrequent in the community than in referral and specialist practices.

Given the apparent low prevalence of this problem, even among CCB users, and the difficulty of performing confirmatory studies as good as this one, I do not believe that further retrospective studies are needed. A further characterization of histopathologic characteristics of CCB-related eruptions may be useful. Although better understanding of the mechanism(s) of these reactions would be of interest, the generalizability of such potential findings to drug-related chronic skin conditions is likely to be limited. Perhaps we should declare victory and go home knowing that Joly and colleagues clinical recommendations are sufficiently robust to inform practice and remind us that chronic as well as short-term use of drugs can cause some inflammatory eruptions.