Patient Preferences for Biologicals in Psoriasis: Top Priority of Safety for Cardiovascular Patients

Patients with psoriasis are often affected by comorbidities, which largely influence treatment decisions. Here we performed conjoint analysis to assess the impact of comorbidities on preferences of patients with moderate-to-severe psoriasis for outcome (probability of 50% and 90% improvement, time until response, sustainability of success, probability of mild and severe adverse events (AE), probability of ACR 20 response) and process attributes (treatment location, frequency, duration and delivery method) of biologicals. The influence of comorbidities on Relative Importance Scores (RIS) was determined with analysis of variance and multivariate regression. Among the 200 participants completing the study, 22.5% suffered from psoriatic arthritis, 31.5% from arterial hypertension, 15% from cardiovascular disease (myocardial infarction, stroke, coronary artery disease, and/or arterial occlusive disease), 14.5% from diabetes, 11% from hyperlipidemia, 26% from chronic bronchitis or asthma and 12.5% from depression. Participants with psoriatic arthritis attached greater importance to ACR 20 response (RIS = 10.3 vs. 5.0, p<0.001; β = 0.278, p<0.001) and sustainability (RIS = 5.8 vs. 5.0, p = 0.032) but less value to time until response (RIS = 3.4 vs. 4.8, p = 0.045) than those without arthritis. Participants with arterial hypertension were particularly interested in a low risk of mild AE (RIS 9.7 vs. 12.1; p = 0.033) and a short treatment duration (RIS = 8.0 vs. 9.6, p = 0.002). Those with cardiovascular disease worried more about mild AE (RIS = 12.8 vs. 10, p = 0.027; β = 0.170, p = 0.027) and severe AE (RIS = 23.2 vs. 16.2, p = 0.001; β = 0.203, p = 0.007) but cared less about time until response (β = -0.189, p = 0.013), treatment location (β = -0.153, p = 0.049), frequency (β = -0.20, p = 0.008) and delivery method (β = -0.175, p = 0.023) than others. Patients’ concerns should be addressed in-depth when prescribing biologicals to comorbid patients, keeping in mind that TNF antagonists may favourably influence cardiovascular risk.

Introduction biologicals), we noted high impact of comorbidities on preferences [25]. The aim of the present study was to assess the impact of comorbidities on patient preferences for treatment of psoriasis with biologicals by using CA. We show that preferences vary significantly dependent on presence or absence of psoriatic arthritis, cardiovascular disease and diabetes.

Study participants
Patients with moderate-to-severe psoriasis visiting outpatient clinics at the Department of Dermatology of the University Medical Center Mannheim, Germany were asked to participate. Inclusion criteria were age 18 years and moderate-to-severe psoriasis according to the criteria of the Committee for Medicinal Products for Human Use (CHMP; for details, see [23]). Exclusion criteria were inability to complete the survey due to difficulties with German language or inability to understand CA exercises (i.e., failing to answer exercises with unambiguous scenarios presented for control). Both patients seeking consultation for the first time and patients coming for follow-up visits were included. The study was performed according

Data collection
Participants who had provided written informed consent were asked to complete a computerized survey before clinical consultation. The survey contained information on sociodemographic characteristics (age and gender), Dermatology Life Quality Index (DLQI), comorbidities, and smoking habits (current smoker (defined as currently smoking 1 cigarette per day), former smoker, or non-smoker). Comorbidities were assessed by choosing out of a list of options including psoriatic arthritis, diabetes, depression, allergies, arterial hypertension, cardiovascular disease (myocardial infarction, stroke, coronary artery disease, and/or arterial occlusive disease), chronic bronchitis or asthma, hyperlipidemia, liver disease and neoplasia. In addition, participants were asked to indicate other comorbidities as free text. If patients reported arthralgia and/or suspected psoriatic arthritis, Classification for Psoriatic Arthritis (CASPAR) criteria were applied to verify the diagnosis [26]. The PASI was determined by two of the investigators (M.-L.S. and C.K.).
Respondents' preferences for treatment of psoriasis with biologicals were elicited by CA exercises, generated basically as described [27,28]. Seven key outcome attributes (probability of 50% improvement, probability of 90% improvement, time until response, sustainability of success, probability of mild AE, probability of severe AE, and probability of ACR 20 response) and four process attributes (treatment location, frequency, duration, and delivery method) were identified and decomposed into four levels describing adalimumab, etanercept, infliximab and ustekinumab as realistically as possible (S1 Table; for procedures and references regarding identification of attribute levels and for further details on generation of discrete choice scenarios see [24]). The probability of 50% improvement was depicted as the chance of reduction of the psoriasis by half, the probability of 90% improvement as the chance of almost complete clearance. Mild infections, injection-site reactions, headache, gastrointestinal symptoms and mild temporary change of laboratory parameters were indicated as examples of mild AE, tuberculosis, other severe infections, severe intolerance reactions and autoimmune diseases as examples of severe AE (for examples of discrete choice scenarios presented to the participants see [24]). Participants were confronted with hypothetical treatment scenarios generated by commercially available CA software (http://www.sawtoothsoftware.com) and repetitively asked to choose their individually preferred option out of pair-wise presented treatment scenarios. Preferences were measured by the software as described previously [24]. Subgroup analyses according to comorbidities and smoking status was performed with one-way ANOVAs (analysis of variance), using SPSS software. Brown-Forsythe tests were applied if the assumption of homogeneity of variances was not fulfilled.

Regression analysis
Multivariate linear regression analysis was performed to estimate independent associations of gender, age, PASI, DLQI, psoriatic arthritis, cardiovascular disease, diabetes and depression with the dependent variable RIS. Arterial hypertension was not integrated into the regression models because of multicollinearity with cardiovascular disease. Asthma and chronic bronchitis were not considered because these comorbidities are not directly related to psoriasis. Hyperlipidemia, allergies and the smoking status were not included because bivariate statistics showed no significant association with RIS. A standardized regression coefficient β was assigned to each independent variable, indicating the amount of change in RIS when varying one of the variables while holding the others constant. P-values 0.05 were considered significant.

Results
Out of 239 patients with moderate-to-severe psoriasis approached for study participation, 210 provided written informed consent. Ten participants had to be excluded due to problems with German language (n = 9) or inability to understand the CA exercises (n = 1). Two hundred participants completed the study survey.
Most of the participants were male (57.5%), and the mean age of the study population was 50.8 years ( Table 1). The vast majority (95.5%) came for a follow-up visit, and 99% obtained antipsoriatic treatment (76.5% topical therapy, 10% phototherapy, 37.5% traditional systemic medications and 43.5% biologicals). Therefore the mean PASI was only 3.4 (range: 0-26.7). The mean DLQI was 6.2 (range: 0-30), reflecting moderate impairment of disease-related quality of life. 22.5% of the participants suffered from psoriatic arthritis, 31.5% from arterial hypertension, 15% from cardiovascular disease (myocardial infarction, stroke, coronary artery disease, and/or arterial occlusive disease), 14.5% from diabetes and 11% from hyperlipidemia. 26% reported chronic bronchitis or asthma and 25.5% allergies. Liver disease was documented in 2.5%, inflammatory bowel disease in 1%, rheumatoid arthritis in 2% and a history of neoplasia in 4% of the participants. The self-reported prevalence of depression was 12.5%, the prevalence of smoking 37.5% ( Table 1).

Impact of comorbidities on preferences
Treatment preferences averaged across the whole study cohort were previously reported [24]. Briefly, participants attached greatest value to the risk of severe AE (RIS = 17.  Table 2).
Partcipants with arterial hypertension were particularly interested in a low risk of mild AE (RIS 9.7 vs. 12.1; p = 0.033) and in a short treatment duration (RIS = 8.0 vs. 9.6, p = 0.002; Fig 1D). Similarly, participants with chronic bronchitis or asthma attached special importance to avoiding mild AE (RIS = 10.1 vs. 13.7, p = 0.034). Hyperlipidemia, allergies and the smoking status had no significant impact on preferences (data not shown). Furthermore, preferences of participants with or without self-reported depression did not differ significantly from each other, neither in bivariate analyses (data not shown) nor in multivariate regression models (Tables 2 and 3). Subgroup analyses with respect to other comorbidities were precluded by the small sample sizes.

Discussion
The burden of comorbidities, most importantly, psoriatic arthritis, cardiovascular disease, diabetes and depression, may exert an enormous incremental impact on quality of life and work diabetes were particularly interested in short treatment duration. (D) Those with arterial hypertension attached special importance to a low risk of mild AE and a short treatment duration. Differences in RIS were tested for significance with one-way ANOVA or Brown-Forsythe tests. Bars productivity of patients with psoriasis [27]. These comorbidities have great influence on treatment decisions, including decisions in favour of or against prescription of biologicals. For example, TNF antagonists are regarded as biologicals of first choice in patients with psoriatic arthritis [28]. However, they are contraindicated in patients with high-grade cardiac insufficiency [15,29]. Here we show that comorbidities strongly influence patients' preferences for attributes of biologicals. Participants with psoriatic arthritis were more interested in improvement of arthralgia than others, as expected. In addition, they were willing to trade rapidity of success in favour of increased sustainability. Psoriatic arthritis leads to more severe impairment in physical functions and daily work and leisure activities than mere cutaneous psoriasis [30]. Therefore, identification of a highly effective and sustainable systemic antipsoriatic medication is even more compulsory for patients with psoriatic arthritis.
Participants suffering from cardiovascular disease were particularly concerned about AE and willing to accept slower onset of action and a less convenient treatment process in order to minimize the risk of AE. These patients may have a special interest in avoiding additional risk as they might have experienced life threatening events such as angina pectoris, myocardial infarction and stroke. Furthermore, they are likely to receive multi-drug comedication [31] and therefore more prone to AE due to drug interactions. However, systemic inflammation in psoriasis is an important trigger of cardiovascular risk and efficient control of this inflammation is crucial for cardiovascular risk reduction [4,15,32]. TNF antagonists were convincingly demonstrated to reduce the risk of cardiovascular events both in rheumatoid arthritis [33] and in psoriasis [15,16,32,34]. Patients with psoriasis receiving TNF antagonists or methotrexate have lower cardiovascular event rates than patients treated with other medications [16,32]. Use of TNF antagonists is associated with a reduced incidence of myocardial infarction as compared to topical treatment [34]. Moreover, there is evidence that TNF antagonists have a beneficial effect on preventing the progression of subclinical atherosclerosis and arterial stiffness in patients with psoriasis and inflammatory arthritis [35,36,37]. According to a recent systematic review based on 23 studies, TNF-α inhibitors were convincingly shown to prevent or even reverse the progression of intima media thickness in patients with rheumatoid arthritis, ankylosing spondylitis and psoriatic arthritis who respond to treatment [37]. In addition, several studies pointed to a reduction of the pulse wave velocity as indicator of arterial stiffness upon treatment with TNF antagonists [37]. Last but not least, TNF antagonists improve biomarkers of inflammation such as C-reactive protein [38], which are independent risk factors of atherosclerosis. For explanations and abbreviations, see Table 2. doi:10.1371/journal.pone.0144335.t003 The impact of ustekinumab on cardiovascular risk is less clear. As another interleukin 12/23 antagonist, briakinumab, was withdrawn from further development in part due to concerns about cardiovascular safety, discussion has also emerged on cardiovascular safety of ustekinumab. Associations between the two interleukin 12/23 antagonists and major adverse cardiovascular events (MACEs) were examined in two meta-analyses of randomized controlled trials [39,40]. Even if the same trials and the same number of MACEs were included, these metaanalyses came to different conclusions. According to Ryan et al. the risk of MACEs was not significantly elevated with use of interleukin 12/23 antagonists [39]. Tzellos and colleagues used another statistical methodology and found a more than fourfold increased odds ratio for MACEs with interleukin 12/23 antagonists ( [40]; for discussion of these discrepancies, see [41]). An integrated analysis of data from phase II and III studies with ustekinumab alone suggested that ustekinumab use was associated with a decrease in MACEs compared to the general US population or to psoriasis patients in Great Britain [42]. In addition, 5-year safety data for ustekinumab did not show an increased risk of MACEs compared to the general population [43]. According to a Danish nationwide cohort study, the risk of composite cardiovascular endpoints was significantly decreased upon treatment with TNF antagonists, whereas patients receiving ustekinumab had a similar risk as the reference group using other therapies [16]. However, discussion on cardiovascular safety of ustekinumab has still not been fully resolved [44,45].
Our finding that patients with cardiovascular comorbidities are particularly worried about AE is in line with results from our previous study in which we compared preferences of comorbid patients for all kinds of antipsoriatic treatments without addressing attributes of specific medications [25]. Moreover, it is well compatible with the observation that participants with other chronic diseases such as arterial hypertension and chronic bronchitis or asthma are especially concerned about safety. Suffering from a chronic disease which probably requires longterm treatment and life-long surveillance is likely to sensitize patients for safety aspects. Our results highlight the importance of carefully discussing safety issues with patients affected by such chronic and potentially threatening diseases and explicitly addressing their concerns about AE. It is essential to explain that reduction of systemic inflammation can beneficially influence cardiovascular risk, a fact most patients are unaware of.
Participants with diabetes as well as participants with arterial hypertension placed greater value on treatment duration than others. Management of diabetes can be quite complex and time-consuming. Therefore it is comprehendible that patients suffering both from psoriasis and from diabetes prefer a time-saving treatment. Interestingly, there is increasing evidence that TNF antagonists may improve insulin resistance in patients with psoriasis and rheumatoid arthritis [46,47,48]. A recent study on 29 non-diabetic individuals with moderate-to-severe psoriasis disclosed a statistically significant improvement of insulin sensitivity after 6 months of adalimumab therapy [48]. Significant improvement of erythrocyte sedimentation rate, ultrasensitive CRP, BSA, PASI, Nail Psoriasis Severity Index, Physician Global Assessment and psoriatic arthritis screening and evaluation questionnaire was also observed at that time [48]. Similarly, amelioration of insulin sensitivity was noted in non-diabetic patients with rheumatoid arthritis receiving TNF-α antagonist therapy [47]. A beneficial effect has also been observed on retinol-binding protein-4, an adipokine considered as an emerging cardiometabolic risk factor that is increased in patients with moderate-to-severe psoriasis [49]. Moreover, psoriasis patients treated with TNF antagonists have a lower risk of new-onset diabetes than patients on non-biological disease-modifying antipsoriatic drugs [50].
When we compared preferences for all antipsoriatic treatment modalities, we found that patients with depression were particularly interested in a time-saving therapy with low individual costs [25]. The CA experiments presented here did not contain a cost attribute because treatment costs for biologicals are covered by health insurance in Germany if their prescription is accordant to licence. Depression did not significantly impact preferences for treatment with biologicals, possibly because their application is more convenient and time-saving than topical therapy and phototherapy. Both TNF antagonists and ustekinumab can substantially improve symptoms of depression and health-related quality of life in patients with psoriasis [51,52].
A limitation of our study is that data on the prevalence of comorbidities were based on patients' self-reported medical history. This way of data collection may lead to underestimation of the frequency of comorbidities due to reporting bias. However, patients are likely to reliably remember and report diabetes and severe cardiovascular diseases such as myocardial infarction, stroke and coronary artery disease. Diagnosis of psoriatic arthritis was either previously confirmed by a physician or verified by us with CASPAR criteria which have >98% specificity [26]. Self-reporting of depression may result both in underestimation of its prevalence due to social desirability bias and in overestimation if patients interpret transient slight depressive moods as depression. In clinical practice, this comorbidity is frequently underdiagnosed [53].
Moreover it has to be emphasized that in addition to the comorbidities considered in detail in this study, psoriasis is associated with other diseases influencing cardiovascular risk, in particular, non-alcoholic fatty liver disease, inflammatory bowel disease, rheumatoid arthritis and other autoimmune diseases [54,55]. The number of participants with liver disease, inflammatory bowel disease and rheumatoid arthritis in our cohort was rather small, precluding subgroup analyses with respect to these comorbidities. It is well conceivable that non-alcoholic fatty liver disease was underdiagnosed and that its true prevalence was underestimated, because information on this comorbidity was merely based on patient reports instead of being verified by laboratory tests and ultrasound.
A further limitation is that only patients with moderate-to-severe psoriasis treated at a German University Hospital were included, i.e., relatively severely affected patients with long disease duration (mean: 19.9 years) and extensive treatment experience. Clearly, our findings will have to be confirmed in larger cohorts and in a multi-centric setting.
The mean PASI of the study collective was low, because the vast majority of respondents received antipsoriatic therapy at the time of data collection. However, all participants suffered from moderate-to-severe psoriasis according to CHMP criteria, and all were potential candidates for the biologicals. Forty-six percent of the participants had experience with biologicals, which may influence preferences.
In conclusion, we show that patient preferences for biologicals are considerably influenced by their comorbidities. Patients with cardiovascular disease and arterial hypertension are particularly concerned about safety, even if TNF antagonists favourably influence cardiovascular risk. Clearly, treatment decisions cannot merely depend on patient preferences but have to be based on medical conditions, and physicians' recommendations may deviate from patient preferences because of medical reasons [56]. Nevertheless it is essential to address patients' needs and concerns and to discuss individual benefits and risks of comorbid patients indepth in order to ensure a trustful physician-patient relationship and optimize adherence and outcome.
Supporting Information S1 Table. Outcome and process attributes and attribute levels. 1 Probability of loss of response within one year. 2 per treatment session. AE: adverse events. (DOC)