Etanercept: Title and subTitle BreakAn Overview of Dermatologic Adverse EventsDermatologic Adverse Events of Etanercept

Undertaking a systematic review for adverse events is complex and is complicated by identification and quality assessment of primary data. More attention should be given to adverse events data collection as a primary outcome in studies. This would allow for calculating incidences and risks. Also, cohorts or long-term studies are needed for rare and late side effects, such as data collection by registries.

Since its introduction in 1998, the biological agent etanercept has approximately 1.6 million patient-years of experience worldwide (August 2008, Wyeth Pharmaceuticals [Collegeville, Pennsylvania] and Amgen [Thousand Oaks, California]). For many patients with rheumatoid arthritis, psoriatic arthritis, plaque psoriasis, juvenile idiopathic arthritis, ankylosing spondylitis, and numerous off-label indications, treatment with etanercept (Enbrel; manufactured by Immunex Corp [Seattle, Washington] and marketed by Wyeth Pharmaceuticals and Amgen) has been beneficial and has led to significant improvements in quality of life.^{1 - 5}

However, considering the immunomodulating effect of etanercept, adverse events resulting from this mode of action can be expected. Etanercept blocks the action of tumor necrosis factor (TNF), which is a naturally occurring cytokine involved in normal inflammatory and immune responses.⁶ Remarkable is the relatively high number of reports on dermatologic adverse events in the literature, including psoriasiform dermatoses, nonmelanoma skin cancers, and forms of lupus erythematosus and, more incidentally, rare malignant neoplasms. Furthermore, we have observed an increasing number of referred patients with dermatologic adverse events of TNF inhibitors in our tertiary dermatology center. Also, patients using TNF inhibitors for dermatologic indications, such as plaque psoriasis, have presented with new dermatologic conditions. This observation was confirmed by a prospective study in which patients with rheumatic diseases were observed for the development of cutaneous manifestations. More than 20% of them developed dermatologic adverse events during treatment with TNF inhibitors.⁷ A detailed analysis of adverse events is relevant, and evidence on the potential adverse effects of interventions is required.^{8 - 9} The purposes of this study were to provide a comprehensive overview of dermatologic adverse events of etanercept described in the literature (including all study types, case reports, and surveys) and to present information on the occurrence, severity, treatment, and course of these adverse events.

We defined adverse dermatologic events as all skin abnormalities that develop, or preexisting skin diseases that significantly worsen, during or shortly after etanercept treatment with or without a demonstrable causal relationship.

The searches were performed by a medical librarian (C.E.J.M.L.) with experience in conducting searches for systematic reviews. The searches were done in the bibliographic databases MEDLINE (PubMed or OVID) and EMBASE (OVID), using both free-text words and index terms specific to each database (as Medical Subject Headings and Subject Headings). No language or any other restriction was applied. The search included an iterative process to refine the search strategy through adding search terms as new relevant citations were identified.

After searching for relevant systematic reviews, 2 approaches were followed to find the adverse dermatologic events associated with etanercept (Figure).

Search 1 focused on all controlled clinical trials with etanercept. Because electronic searches would fail to retrieve trials with adverse effects not explicitly mentioned in the title, abstract, or index terms,¹⁰ we searched PubMed (1966 through May 2008) and EMBASE (1980 through May 2008) for etanercept in combination with broad RCT (randomized controlled trial) filters, developed for Cochrane Systematic Reviews of interventions.¹¹

Search 2 was a complex search designed to find skin diseases associated with etanercept. Skin disease as an indication for etanercept treatment complicated the search. We used OVID MEDLINE (1960 through October 2008) instead of PubMed in addition to OVID EMBASE (1980 through October 2008) because OVID allows more sophisticated search syntax (such as adjacency searching). Index terms for dermatologic conditions for which etanercept is indicated were coupled to subheadings for adverse effects (for instance, chemically induced [ci] in MEDLINE). Dermatologic conditions for which etanercept is not a common treatment were searched as index terms without a subheading. Likewise, the dermatologic conditions were searched as text words alone or combined with terms indicating an adverse effect. We also broadly searched for (unspecified) dermatologic effects, ie, by using the search string: ((cutaneous*2 or skin or derma*) adj20 (reaction? or side-effect? or adverse or toxic)).tw. (eTable 1). All studies retrieved by search 1 or 2 were evaluated for eligibility.

Selection was done by one author (L.L.A.L.). In case of doubt, another author was consulted for rescreening the article, and disagreements were resolved by consensus. Initial eligibility was based on title and abstract or full text. Final eligibility was based on full text.

Articles were excluded from the analysis (1) if no individual patients were described; (2) if patients were not treated with etanercept; (3) if the type of biological agent was not specified; (4) if etanercept was not used systemically but, eg, intra-articularly; (5) if no safety data were collected or described; (6) if no specific dermatologic adverse events were mentioned in the full text; (7) if only injection site reactions (ISRs) were described; or (8) if the full text could not be obtained with reasonable efforts.

All included articles (except for case reports/case series) were assessed regarding level of evidence according to the Oxford Centre for Evidence-Based Medicine Levels of Evidence.¹²

The data were independently extracted from the included studies by 2 reviewers (L.L.A.L. and E.A.D.). Disagreements were resolved by consensus. The included articles were categorized as study reports or case reports/case series.

For the study reports, the following data were extracted and summarized in an evidence table: author, year of publication, level of evidence, indication for etanercept use, study type, dose of etanercept, comedication, number of patients in the study, duration of treatment, mean age, which arms in the study (if applicable) and which adverse events types were reported, and which dermatologic adverse events occurred and in how many etanercept-treated patients.

For the case reports/case series, for each patient we extracted author, year of publication, indication for etanercept use, duration and dose of treatment before the adverse event occurred, which dermatologic adverse event occurred, severity of the adverse event, therapy given for the adverse event, outcome, and whether the patient continued etanercept therapy.

Search 1 resulted in 1211 hits and search 2 in 1176, which resulted in 2218 exclusive articles that were screened for eligibility. Overall, 126 study reports on etanercept and 101 case reports/case series met the inclusion criteria and were analyzed. Most excluded articles did not report on individual patients but were, for example, reviews of the literature. More than 300 reports of clinical studies with etanercept did not mention any specific dermatologic adverse event.

The 126 included study reports that mentioned dermatologic adverse events of etanercept varied widely in study design: 16 pilot studies, 18 retrospective cohort studies, 37 prospective cohort studies, 22 blinded placebo-controlled RCTs (of which 1 was an open-label extension study), 6 open-label single-arm studies, 11 head-to-head blinded RCTs (of which 1 was an open-label extension study), 1 open-label RCT, and 12 open-label extension studies (some studies were reported together in a single article). Three studies reported on a collection of spontaneous reports of adverse events, and 3 other studies were survey studies searching for specific adverse events by approaching physicians. There were 34 different indications or combinations of indications for which etanercept was used, including healthy volunteers. See eTable 2 for specific study information.

In the reported study populations, we evaluated how many patients actually used etanercept (ie, not placebo). This varied by study from 2 patients to 6969, with a total of 30 694 patients. One of the study reports on collected spontaneous reports of adverse events has been analyzed separately because the number of etanercept-treated patients described (n = 308 000) was 10 times the number of patients in all other articles together. This study report describes analyses of safety data for anti-TNF therapies using outcomes from the Uppsala Monitoring Centre of the World Health Organization (WHO), presented in percentages of patients experiencing specific adverse events.¹³ Another reason for separate analysis was that data for these patients may be duplicated because the Uppsala Monitoring Centre collects individual case reports of suspected adverse drug reactions from the National Centers for pharmacovigilance of 95 countries participating in the WHO Programme for International Drug Monitoring.¹⁴ These data from the WHO are given in a specific column in the tables and provide an indication of the incidence of an adverse event.

The mentioned duration of treatment in the study reports varied from 1 to 61 months. Many patients used comedications, such as nonsteroidal anti-inflammatory drugs, methotrexate, other disease-modifying antirheumatic drugs, and/or corticosteroids. The types of adverse events mentioned in the study reports varied; some study reports described all adverse events, whereas others described, for instance, only adverse events that occurred in more than 10% of the patients, only infections, only malignant neoplasms, only serious adverse events, only adverse events leading to treatment discontinuation, or other criteria and combinations of them (see details in eTable 2).

In the 101 included case reports, a total of 150 individual patients were described with a specific dermatologic adverse event associated with etanercept therapy (eTable 3). Two-thirds of these patients were female. Most of the patients used etanercept, 25 mg, twice weekly. The mean age was 48.6 years, with a range between 7 and 78 years. The period between the start of treatment and occurrence of symptoms varied between none (occurring after a single injection) and 70 months of treatment.

In 126 study reports, a total of 72 separate specific dermatologic adverse events of etanercept were mentioned. In 101 case reports/case series, 153 individual patients with approximately 65 different specific (eg, not rash) diagnoses were reported.

Table 1 describes reported infections in the study reports. The skin infections found were of bacterial, viral, fungal, and (in 1 case) parasitic origin. Cellulitis (n ≥ 34) and herpes zoster (n = 29) were described most often, in 23 and 15 study reports, respectively. These were followed by folliculitis, erysipelas, abscesses, paronychia, herpes simplex, and some other less frequently reported skin infections.

Table 2 and Table 3 describe the reported infections in the case reports/case series. The most commonly reported infection described in the case reports/case series was also herpes zoster (n = 12, of whom 8 patients were described in 1 article). According to the WHO, the incidence of herpes zoster during etanercept treatment is 5.86%, which is higher than that of cellulitis or herpes simplex.¹³ In 63% of the cases the infection occurred in the first 6 months of therapy. Nearly all patients for whom comedication was mentioned used additional immunosuppressive agents, such as disease-modifying antirheumatic drugs. In more than half of the mentioned cases, etanercept was permanently stopped. If the adverse event needed treatment, antivirals, antibiotics, and antifungals were used as applicable. Most infections completely resolved. Severe dermatologic infections leading to hospitalization were myiasis (n = 1), septic abscess (n = 1), and erysipelas (n = 2). One 24-year-old patient with rheumatoid arthritis died after 1 month of treatment of streptococcal toxic shock syndrome.

Several cutaneous forms or symptoms of lupus have been reported in association with etanercept therapy in studies (n = 32; Table 1). These include subacute or chronic cutaneous lupus erythematosus, discoid lupus, chilblain lupus, and malar rash, which were each mentioned in 1 or 2 study reports.

In the case reports/case series, a total of 19 patients were described with symptoms such as alopecia, malar rash, urticarial vasculitis, and erythema associated with a lupus syndrome (Table 4 and Table 5). Of these patients, 17 (89%) were women. The lupus symptoms developed equally in patients with or without prednisone as comedication. Two patients had been treated with etanercept previously and developed lupus symptoms only 0.5 and 4 months, respectively, after restarting therapy. In other patients the moment at which the symptoms occurred varied from 8 days to 48 months. Practically all patients who developed a form of lupus with cutaneous symptoms discontinued etanercept treatment. In most cases the lupus resolved after discontinuation even without additional therapy.

Only 6 individual study reports described patients with types of cutaneous vasculitis (Table 1). Leukocytoclastic vasculitis and nonspecified cutaneous vasculitis were reported in a total of 54 patients, and 2 patients were described with vasculitic dermal inflammatory infiltrates.

In 15 separate case reports, 15 patients were described with a form of cutaneous vasculitis in association with etanercept (Table 6 and Table 7): 5 patients with leukocytoclastic vasculitis, 2 with Henoch-Schönlein purpura, 2 with necrotizing vasculitis, and 6 with other cutaneous forms of vasculitis. Two patients had similar symptoms when they were treated with another TNF inhibitor, and 3 patients were rechallenged with etanercept resulting in a recurrence of symptoms. In most cases treatment with etanercept was stopped, and the symptoms mainly resolved with additional therapy.

The most commonly reported skin malignant neoplasms in study reports were basal cell carcinoma and squamous cell carcinoma (SCC), mentioned in 12 and 8 reports, respectively (Table 1). The WHO describes the incidence of basal cell carcinomas as 2.06% and that of SCCs as 1.44% during etanercept therapy.¹³ Four study reports describe a total of 14 patients who presented with a melanoma during etanercept therapy.

In the case reports/case series, SCCs were the most common malignant neoplasms (Table 8 and Table 9). Some of the described SCCs were complicated, for instance, by infection or metastasis. All but 1 of these SCCs developed within 6 months of starting therapy. Two patients died as a result of the skin malignant neoplasms. Nearly all patients for whom comedication was mentioned used 1 or more disease-modifying antirheumatic drugs additionally. One patient was described to have developed an eruptive melanoma after only 1 month of etanercept treatment, which resolved after limb perfusion chemotherapy and adjuvant chemokines.

In 7 studies, patients were described to have developed psoriasis or to have experienced an exacerbation or a phenotype switch during etanercept treatment (Table 1). Indications for etanercept treatment were rheumatic diseases and plaque psoriasis.

In the case reports, no fewer then 38 patients were described to have developed a form of psoriasis (guttate, pustular, or plaque type) as an adverse event (Table 10 and Table 11). The time of occurrence of the psoriasis varied from 0 to 26 months. Some patients had a history of psoriasis, whereas others did not. The psoriatic adverse events improved in most cases with time, cessation of treatment, or additional therapy, but the lesions completely resolved in only some cases. Six patients had similar reactions to other TNF inhibitors, and 1 patient had a positive rechallenge with etanercept.

A broad range of other dermatologic conditions are reported to have occurred during etanercept therapy. Nonspecified rash was the adverse event mentioned in the greatest number of study reports and was reported in most patients (Table 1). The most often mentioned adverse events thereafter were urticaria (n > 24) in 12 study reports and alopecia (n > 87) in 11 study reports. Other relatively often described adverse events were edema, pruritus, eczema, and ecchymosis. Many conditions, such as Grover disease, perineuroma, and striae, were mentioned only once in study reports.

In case reports/case series, many miscellaneous dermatologic adverse events were reported only once or twice; these include alopecia, lichen planopilaris, lentigines, dermatomyositis, and granuloma annulare (Table 12 and Table 13). In most cases treatment with etanercept was stopped.

From the etanercept drug label, we already know that the incidence of ISRs is high.⁶ Therefore, we excluded 46 study reports for complete data extraction that mentioned only unspecified ISRs as adverse dermatologic events. Another 65 study reports mentioned ISRs in addition to other adverse dermatologic events, which still makes it the most reported event. Also, more specific ISRs, such as ISR hemorrhage, ISR bleeding, and ISR erythema, were mentioned many times. Some of these ISRs are recall ISRs. The Uppsala Monitoring Centre reports the incidence of ISRs as 58.87%. Only 2 case reports mentioned an ISR, one a large erythematous pruritic rash at the injection site in a patient with Felty syndrome and the other, multiple angiokeratomas at the injection site.

We have provided a comprehensive overview of dermatologic adverse events of etanercept described in the literature (including all trials, case reports, and surveys). We have also presented data on the occurrence, severity, treatment, and course of these adverse events when available.

Of the 153 patients described in case reports, 38 presented with psoriasiform lesions as dermatologic adverse events of etanercept, which was remarkably high, especially because etanercept is indicated as treatment of moderate to severe psoriasis. Some of these patients experienced an exacerbation of a latent psoriasis, whereas others had no history of psoriasis or had a phenotype switch, mainly to the pustular form. The development of psoriasis appears to be a class effect of TNF inhibitors because it is also seen in patients treated with infliximab or adalimumab.¹⁵ Another argument for a class effect is that some patients were treated with 2 TNF inhibitors and presented with the same symptoms during both treatments (Table 10 and Table 11). The prevalence of psoriasis in general might be higher among patients with chronic inflammatory arthritic conditions.¹⁵ A theory is that the psoriasis results from a treatment-induced cytokine imbalance: inhibition of TNF can induce overexpression of cutaneous interferon α (IFN-α), which in turn predisposes to psoriasis.¹⁶ Tumor necrosis factor has been shown to regulate IFN-α production; furthermore, induction and worsening of psoriasis has been seen after injection of recombinant IFN-α.¹⁵ This theory may also apply to other dermatologic adverse events we encountered, such as forms of lupus and dermatomyositis, which have both been associated with deregulated IFN-α.^{15 ,17 - 18} A role for IFN-α in the pathogenesis has also been described for lichen planus and leukocytoclastic vasculitis, which were also associated with etanercept treatment.¹⁹

Less striking was the range of infections and malignant neoplasms seen, considering the important role TNF plays in host defense and the fact that many patients used additional immunosuppressants. Animal models indicate an essential role of TNF in combating infection.²⁰ In a systematic observational study concerning infection rates, it was shown that infections were much more frequent during TNF-blocker therapy than before this therapy in the same patients.²¹ Infectious events were reported in 34.5% of patients during all treatment courses.²¹ Analyses from the German biologic registry on patients with rheumatoid arthritis showed that patients treated with biologic agents have a higher a priori risk of infection, but this risk was increased by treatment with TNF inhibitors.²²

However, evidence on the actual role of TNF inhibitors in the development of malignant neoplasms is not convincing. In a meta-analysis on the risk of malignant neoplasms with etanercept in RCTs, the point estimate of malignancy risk was higher in etanercept-treated patients (incidence rate, 10.47 vs 6.66 per 1000 person-years) but not statistically significant.²³ A large observational cohort study found that etanercept and other biologics were associated with an increased risk of nonmelanotic skin cancer (odds ratio, 1.5; 95% confidence interval, 1.2-1.8) and melanoma (2.3; 0.9-5.4) but not with other malignant neoplasms.²⁴

A biological mechanism that might explain these findings is that TNF is important in natural killer cell–and CD8 lymphocyte–mediated killing of tumor cells, although tumor-promoting effects of TNF have also been described.²⁰ In contrast to monoclonal antibodies that inhibit TNF, etanercept also neutralizes lymphotoxin α, which has been associated with tumor growth control independent of TNF activity.²³ To confirm these increased incidences of infections and malignant neoplasms, long-term data are needed.

This study is a systematic review, which is known to be the most powerful tool used in evidence-based medicine.²⁵ Although in principle adverse events are most reliably assessed by means of RCTs, in these trials only the most common, often short-term, events are observed.⁸ Individual RCTs can identify adverse events that occur in more than 1 in 200 patient-years after only a relatively short treatment period in a selected study population.²⁶ Therefore, the assessment of safety has to go beyond RCTs and should use various methods, including postmarketing surveillance studies, spontaneous reporting schemes, and epidemiologic investigations.²⁵ Observational research on adverse events, such as registries, provides prolonged follow-up and larger numbers. This type of research not only discovers rare events but permits quantification of the whole range of effects. Also, it gives insight into usual medical practice.²⁶ However, a broad systematic review scope like the one in this study provides wider coverage and possibilities of signaling new adverse effects that may not have been previously known.⁸

This study does have some limitations. Systematic reviews are often done to evaluate efficacy, but to analyze safety with a systematic review, there are some complicating factors. The methodology surrounding identification and quality assessment of primary data is the main concern.²⁷ A study conducted to assess the performance of different approaches to searching MEDLINE and EMBASE to identify studies of adverse effects concluded that a highly sensitive search requires a combination of approaches and has low precision.²⁸ It is suggested that better reporting and indexing of adverse effects is required and that an effective generic search filter may not yet be feasible.²⁸ Our search was performed by a specialist and has been given very intensive attention; nevertheless, owing to the complexity, we cannot exclude the possibility that it missed relevant articles.

The lack of a defined and empirically tested quality assessment tool is also a major problem.²⁷ Many of the included articles would not have scored high for quality because efficacy often was the primary outcome. The levels of evidence given in this article are based on primary outcomes and not on adverse events data collection. Studies are in general not designed to detect adverse events, and such events are often handled less rigorously than the primary outcomes of a study.^{8 ,28} This is confirmed by the fact that studies in which adverse events are specifically sought report a higher frequency of these events than studies in which they are sought less carefully.⁸

Despite their low level of evidence, case reports are seen as the cornerstone of the initial detection of new, late, and often rare adverse events. Nevertheless, in case reports there is also usually uncertainty as to whether the adverse event was caused by the intervention, and it is not easy to tell whether the report is a genuine alert or a false alarm.⁸ For most of the included cases, the available information was too limited to judge causality.

Our desire to be all-inclusive in our collection of dermatologic adverse events resulted in a variety of included studies. Therefore, the adverse event information was best summarized in a descriptive manner and could not be combined for analysis.²⁹ As a result, no risk or incidence of a given event could be calculated. This is illustrated, for instance, by the fact that more patients with vasculitis as an adverse event were described in case reports than in study reports, suggesting a publication bias. To calculate incidences and risks, a special approach would have been needed focusing on high-quality RCTs with adverse events as the primary outcome. Nevertheless, to give some indication, we have included the incidences reported by the WHO. This also applies to the role of comedications and their attributable risk for specific dermatologic adverse events.

If a patient receiving etanercept is referred to a dermatologist, any skin disease should be considered.⁷ Diagnostic steps do not deviate from the norm in these patients, but management of the dermatologic adverse events may need special attention.

Patients are more susceptible to infections when they use immunomodulating therapies such as etanercept, and it should be kept in mind that their defense mechanism is impaired. Therefore, therapeutic intervention is more likely to be needed than in nonimmunosuppressed patients. Treatment of the infections themselves should be as usual, based on the causative organism, and etanercept therapy may be continued in most cases. When an intercurrent infection is severe, treatment with etanercept can easily be interrupted for a few weeks and then started again because of the short administration interval and half-life.

As with infections, patients may be more susceptible to malignant neoplasms when they use immunomodulating therapies such as etanercept. Etanercept is contraindicated in patients with a current or past malignant neoplasm. A malignant neoplasm occurring as an adverse event should be treated as normally appropriate, and continuation of treatment should be carefully considered for each patient.

In cases of psoriasis as an adverse event, the distinction between drug-induced and drug-triggered psoriasis is of crucial importance for appropriate management.³⁰ Clinical improvement when etanercept is stopped is the basic distinguishing feature suggesting drug-induced psoriasis. In such cases emollients alone are helpful. Management of drug-triggered psoriasis necessitates use of the full range of contemporary therapeutic agents, topical or systemic.³⁰

Drug-induced lupus erythematosus is classically a systemic disease with lack of cutaneous involvement. However, a new subset of drug-induced lupus erythematosus is characterized primarily by cutaneous disease.³¹ The interval to develop the symptoms is variable, and symptoms typically worsen the longer the patient is treated with the implicated drug.³² Drug-induced cutaneous lupus erythematosus usually improves within days or a few weeks after discontinuation of the causative agent, which is the main therapeutic intervention.³¹

Drug-induced cutaneous vasculitis usually starts with a rash, which evolves from red macules to papules turning purpuric.³³ There is no standard approach to the treatment of drug-induced vasculitis. If drug discontinuation is not needed, in most cases topical corticosteroids, nonsteroidal anti-inflammatory drugs, and antihistamines suffice. These agents are prescribed because they often lead to the prompt relief of symptoms.³⁴

In general, for decisions on treatment continuation, the assessment of benefit vs harm must be considered in the context of the nature of the condition for which the intervention is used.²⁹ For many of the patients using etanercept, treatment alternatives may be scarce, and treatment discontinuation in cases of mild to moderate adverse events should not be encouraged. In most of the case reports/case series, the severity of the adverse events was not described, although in most cases etanercept treatment was ceased. Because these cases were specifically selected for publication, the reported need for treatment discontinuation may not directly reflect daily practice. This is illustrated by a prospective study on dermatologic adverse events of TNF antagonists in patients with rheumatoid arthritis, in which all the events that occurred were treated in the usual fashion and treatment was continued.⁷ The reasons for discontinuation are complex, and discontinuation may be due to mild but irritating side effects, drug toxicity, lack of efficacy, nonmedical reasons, or a combination of causes.⁸ Evidence is needed on whether reducing the intensity of the intervention will help avoid the adverse effects or whether there is a treatment strategy that can prevent adverse effects.⁸

It is generally known that the skin is the organ most frequently affected by adverse drug reactions, and this is probably also true for etanercept. Although the adverse events are usually mild, some reactions are serious and even potentially life threatening. Furthermore, they can occur as part of a spectrum of multiorgan involvement. Therefore, all drug-associated cutaneous abnormalities should be carefully evaluated.³⁵

Etanercept is associated with a wide variety of dermatologic adverse events, of which many were described in study reports but also numerous exceptional cases were described in case reports. Although case reports are low in level of evidence, they are considered vital early warnings that raise suspicions and should incite further confirmatory investigations.³⁶

Better conducted and reported studies of harms are certainly needed, and improvements will help to make systematic reviews of harms more accurate.²⁹ Large prospective registry studies could be valuable tools because they are able to monitor adverse effects of systemic treatments, with particular attention to long-term and rare adverse events.³⁷

Correspondence: Lidian L. A. Lecluse, MD, PhD, Department of Dermatology, Academic Medical Center, Room A0-229, University of Amsterdam, PO Box 22700, 1100 DE Amsterdam, the Netherlands (l.l.lecluse@amc.uva.nl).

Accepted for Publication: April 28, 2010.

Author Contributions: Drs Lecluse and Spuls had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Lecluse, de Rie, Bos, and Spuls. Acquisition of data: Lecluse, Dowlatshahi, and Limpens. Analysis and interpretation of data: Lecluse and Spuls. Drafting of the manuscript: Lecluse. Critical revision of the manuscript for important intellectual content: Dowlatshahi, Limpens, de Rie, Bos, and Spuls. Administrative, technical, and material support: Lecluse and Limpens. Study supervision: de Rie, Bos, and Spuls.

Financial Disclosure: None reported.