**E**ffi**cacy and Safety of Ceftaroline for the Treatment of Community-Acquired Pneumonia: A Systemic Review and Meta-Analysis of Randomized Controlled Trials**

### **Shao-Huan Lan 1, Shen-Peng Chang 2, Chih-Cheng Lai 3, Li-Chin Lu <sup>4</sup> and Chien-Ming Chao 3,\***


Received: 15 May 2019; Accepted: 7 June 2019; Published: 9 June 2019

**Abstract:** This study aimed to compare the clinical efficacy and safety of ceftaroline with those of ceftriaxone for treating community-acquired pneumonia (CAP). The PubMed, Cochrane Library, Embase, and clinicalTrials.gov databases were searched until April 2019. This meta-analysis only included randomized controlled trials (RCTs) that evaluated ceftaroline and ceftriaxone for the treatment of CAP. The primary outcome was the clinical cure rate, and the secondary outcome was the risk of adverse events (AEs). Five RCTs were included. Overall, at the test of cure (TOC), the clinical cure rate of ceftaroline was superior to the rates of ceftriaxone for the treatment of CAP (modified intent-to-treat population (MITT) population, odds ratio (OR) 1.61, 95% confidence interval (CI) 1.31–1.99, *I* <sup>2</sup> = 0%; clinically evaluable (CE) population, OR 1.38, 95% CI 1.07–1.78, *I* <sup>2</sup> = 14%). Similarly, the clinical cure rate of ceftaroline was superior to that of ceftriaxone at the end of therapy (EOT) (MITT population, OR 1.57, 95% CI 1.16–2.11, *I* <sup>2</sup> = 0%; CE population, OR 1.64, 95% CI 1.15–2.33, *I* <sup>2</sup> = 0%). For adult patients, the clinical cure rate of ceftaroline remained superior to that of ceftriaxone at TOC (MITT population, OR 1.66, 95% CI 1.34–2.06, *I* <sup>2</sup> = 0%; CE population, OR 1.39, 95% CI 1.08–1.80, *I* <sup>2</sup> = 30%) and at EOT (MITT population, OR 1.64, 95% CI 1.20–2.24, *I* <sup>2</sup> = 0%; CE population, OR 1.65, 95% CI 1.15–2.36, *I* <sup>2</sup> = 0%). Ceftaroline and ceftriaxone did not differ significantly in the risk of serious AEs, treatment-emergent AEs, and discontinuation of the study drug owing to an AE. In conclusion, the clinical efficacy of ceftaroline is similar to that of ceftriaxone for the treatment of CAP. Furthermore, this antibiotic is as tolerable as ceftriaxone.

**Keywords:** ceftaroline; ceftriaxone; community-acquired pneumonia; safety

#### **1. Introduction**

Community-acquired pneumonia (CAP) is a common acute bacterial infection among adults and children and has become a significant global health problem [1–4]. Moreover, severe CAP is associated with high morbidity and mortality, particularly when prompt and appropriate treatment is not provided [5,6]. However, the emergence of antibiotic resistance in this era—with the increase in resistant bacteria not treatable with existing antibiotics—and the lack of development of novel antibiotics has complicated the use of antibiotics unlike before [3,7]. In addition to the most common CAP pathogen—*Streptococcus pneumoniae*, less than 8% of CAP can be caused by the so-called PES pathogens—*Pseudomonas aeruginosa*, extended-spectrum β-lactamase producing *Enterobacteriaceae*, and methicillin-resistant *Staphylococcus aureus* (MRSA), especially in intensive care unit (ICU) [8,9]. Among PES, MRSA is the most frequently reported, and it requires the use of specific antimicrobial agents for the treatment of typical CAP [10]. Currently, the antibiotics recommended for treating CAP when MRSA infection is suspected are vancomycin, teicoplanin, and linezolid [11–13].

Ceftaroline is a new cephalosporin with broad-spectrum activity against many commonly encountered pathogens causing CAP, including *S. pneumoniae, S. aureus, Moraxella catarrhalis, Haemophilus influenzae, and Klebsiella pneumonia* [14–16]. Moreover, several investigations have demonstrated the substantial in vitro activity of ceftaroline against MRSA from various clinical specimens, including skin/soft tissue and respiratory tract [15,17–19]. Global surveillance revealed that compared to ceftriaxone, ceftaroline showed superior in vitro activity against common CAP pathogens [17]. Subsequently, several randomized controlled trials (RCTs) [20–24] have investigated the efficacy and safety of ceftaroline for the treatment of CAP. In the present study, we conducted a comprehensive meta-analysis to provide high-quality evidence on the efficacy and safety of ceftaroline compared to those of ceftriaxone for treating CAP.

#### **2. Methods**

#### *2.1. Study Search and Selection*

All clinical studies were identified through a systematic review of the literature in the PubMed, Embase, ClinicalTrials.gov, and Cochrane databases until April 2019 using the following search terms: "ceftaroline", "Teflaro", "Zinforo", "pneumonia", and "RCT". Only RCTs that compared the clinical efficacy and adverse effects of ceftaroline and ceftriaxone were included. Two reviewers (Lan and Chang) searched and examined publications independently to avoid bias. When they disagreed, a third author (Lai) resolved the issue. The following data were extracted from each study included in the meta-analysis: year of publication, study design, duration, antibiotic regimens of ceftaroline and ceftriaxone, outcomes, and adverse events (AEs).

#### *2.2. Definitions and Outcomes*

The primary outcome was the overall clinical cure with the resolution of clinical signs and symptoms of pneumonia or improvement to the extent that no further antimicrobial therapy was necessary at the end of therapy (EOT) and test of cure (TOC) in the modified intent-to-treat population (MITT) and the clinically evaluable (CE) population. The EOT visit took place within 48 h after the last dose of oral study drug or within 24 h after the last dose of the IV study drug. The TOC visit was at 8–15 days after the last dose of the IV or oral study drug (whichever was given last). Patients in the MITT population who met minimal disease criteria and had ≥1 bacterial pathogen commonly associated with CAP identified at baseline were included in the microbiological modified MITT (mMITT) population, and those who met criteria for both the CE and mMITT populations were included in the microbiologically evaluable (ME) population. The secondary outcome was the risk of AEs, including mild, moderate, and severe degree and discontinuation because of AEs, relapse rate, and mortality.

#### *2.3. Data Analysis*

This study used the Cochrane risk-of-bias tool to assess the quality of enrolled RCTs and their risk of bias [25]. The Review Manager software program, version 5.3, was used to conduct statistical analyses. The degree of heterogeneity was evaluated using the Q statistic generated from the χ<sup>2</sup> test. The *I* <sup>2</sup> measure assessed the proportion of statistical heterogeneity. Heterogeneity was considered significant when the *P* value was less than 0.10 or the *I* <sup>2</sup> value was more than 50%. The random-effects model was used when data were significantly heterogeneous, and the fixed-effect model was used when the data were homogeneous. Pooled odds ratios (ORs) and 95% confidence intervals (CI) were calculated for outcome analyses.

#### **3. Results**

The search results yielded a total of 133 studies from the online databases, and 76 studies were excluded because of duplication. Additionally, 64 studies were found to be irrelevant after the title and abstract were screened (article type and language), and 7 studies were found to be irrelevant after the full text was screened. Eventually, five RCTs [20–24] were enrolled for the meta-analysis (Figure 1).


**Figure 1.** Flowchart of the study selection process.

#### *3.1. Study Characteristics and Study Quality*

All five RCTs [20–24] included were multinational and multicenter studies (Table 1). Three studies [20–22] focused on adult patients with CAP with Pneumonia Outcomes Research Term (PORT) [26] risk class III–IV, and two studies [23,24] enrolled pediatric patients only. Overall, the experimental group treated with ceftaroline and the control group treated with ceftriaxone comprised 1153 and 1050 patients, respectively. Almost all risks of basis in each study were low (Figure 2).

**Figure 2.** Risk of bias per study and domain.



#### *3.2. Clinical E*ffi*cacy*

Notably, ceftaroline had a superior clinical cure rate at TOC compared with ceftriaxone for the treatment of CAP (MITT population, OR 1.61, 95% CI 1.31–1.99, *I* <sup>2</sup> = 0%; CE population, OR 1.38, 95% CI 1.07–1.78, *I* <sup>2</sup> = 14%; ME population, OR 1.98, 95% CI 1.20–3.25, *I* <sup>2</sup> = 0%; Figure 3). Similarly, at EOT, the clinical cure rate of ceftaroline was superior compared with that of ceftriaxone (MITT population, OR 1.57, 95% CI 1.16–2.11, *I* <sup>2</sup> = 0%; CE population, OR 1.64, 95% CI 1.15–2.33, *I* <sup>2</sup> = 0%).

**Figure 3.** Overall clinical cure rates of ceftaroline and ceftriaxone for the treatment of communityacquired pneumonia. MITT, modified intent-to-treat population; CE, clinically evaluable; ME, microbiologically evaluable.

In the subgroup analysis of three studies [20–22] including adult patients, the clinical cure rate of ceftaroline remained superior to that of ceftriaxone at TOC (MITT population, OR 1.66, 95% CI 1.34–2.06, *I* <sup>2</sup> = 0%; CE population, OR 1.39, 95% CI 1.08–1.80, *I* <sup>2</sup> =30%) and at EOT (MITT population, OR 1.64, 95% CI 1.20–2.24, *I* <sup>2</sup> = 0%; CE population, OR 1.65, 95% CI 1.15–2.36, *I* <sup>2</sup> =0%). On the other hand, the pooled analysis of two studies showed that the clinical cure rates at TOC and EOT were similar between pediatric patients treated with ceftaroline or ceftriaxone (at TOC, OR 0.79, 95% CI 0.26–2.97, *I* <sup>2</sup> = 0%; at EOT, OR 1.02, 95% CI 0.38–2.75, *I* <sup>2</sup> = 0%)[23,24].

Figure 4 shows further analysis of the clinical cure rate (ceftaroline vs. ceftriaxone) at the TOC visit in various patient subgroups. Ceftaroline showed a superior clinical cure rate than ceftriaxone for patients with PORT risk III (OR 1.83, 95% CI 1.26–2.67, *I* <sup>2</sup> = 14%) but not for patients with PORT risk IV (OR 1.39, 95% CI 0.91–2.12, *I* <sup>2</sup> = 0%). The efficacy of ceftaroline was superior compared to that of ceftriaxone in patients who did not receive prior antibiotics (OR 1.90, 95% CI 1.22–2.95, *I* <sup>2</sup> = 37%) but not in those who received prior antibiotics (OR 1.18, 95% CI 0.75–1.87, *I* <sup>2</sup> = 0%). No differences were observed in the clinical cure rate between elderly patients (age ≥65 years) treated with ceftaroline or ceftriaxone (OR 1.72, 95% CI 0.95–3.11, *I* <sup>2</sup> = 58%) and between patients with bacteremia treated with ceftaroline or ceftriaxone (OR 1.62, 95% CI 0.46–5.72, *I* <sup>2</sup> = 0%).

We also assessed the clinical cure rate based on pathogens among the mMITT population, and we found that ceftaroline was superior to ceftriaxone in the overall population (OR 1.94, 95% CI 1.25–3.01, *I* <sup>2</sup> = 0%, Figure 5). Ceftaroline was superior to ceftriaxone in patients with gram-positive coccal (GPC) infection (OR 2.65, 95% CI 1.40–5.01, *I* <sup>2</sup> = 0%) but not in those with gram-negative bacterial

(GNB) infection (OR 1.26, 95% CI 0.65–2.42, *I* <sup>2</sup> = 0%). No significant difference was noted between the ceftaroline and ceftriaxone groups for each of the following pathogens: *S. pneumoniae, S. aureus, H. influenzae, H. parainfluenzae, Escherichia coli,* and *K. pneumoniae*.


**Figure 4.** Overall clinical cure rates of ceftaroline and ceftriaxone for the treatment of communityacquired pneumonia based on patient group.

#### *3.3. Adverse Events*

No significant differences were observed in the risk of overall treatment-emergent adverse events (TEAEs) between the ceftaroline and ceftriaxone groups (OR 0.99, 95% CI 0.75–1.30, *I* <sup>2</sup> = 43%), and the similarity was not changed by the degree of severity (Figure 6). The risks of serious AEs and discontinuation of the study drug were similar between the ceftaroline and ceftriaxone groups (Figure 5). In addition, no relapse was noted among all enrolled patients. Finally, the mortality rate was similar between the ceftaroline and ceftriaxone groups (OR 1.13, 95% CI 0.57–2.23, *I* <sup>2</sup> = 0), and none of the cases of mortality were related to the study drug.


**Figure 5.** Overall clinical cure rates of ceftaroline and ceftriaxone for the treatment of communityacquired pneumonia based on pathogens.


**Figure 6.** Risk of adverse events between ceftaroline and ceftriaxone for the treatment of communityacquired pneumonia.

#### **4. Discussion**

This meta-analysis of five RCTs determined that the clinical efficacy of ceftaroline was superior to that of ceftriaxone for the treatment of patients with CAP. First, the overall clinical cure rate of ceftaroline was superior to that of ceftriaxone for treating CAP in the pooled populations of the five RCTs, including pediatric and adult patients [20–24]. The superiority of ceftaroline compared to ceftriaxone remained significant at different times of outcome measurement, including EOT and TOC, and in different populations, including MITT, CE, and ME populations. Second, we found that ceftaroline had a higher clinical cure rate than ceftriaxone among adult patients in the subgroup analysis of three studies [20–22] including adult patients, but the pooled analysis of two studies [23,24] including pediatric patients showed similar clinical cure rates for ceftaroline and ceftriaxone. However, the two studies [23,24] that focused on pediatric patients had a limited number of patients. Therefore, more pediatric studies are warranted to clarify this issue. Third, the subgroup analysis of CAP in various populations demonstrated that ceftaroline was at least similar to ceftriaxone in patients with PORT risk IV, those who received previous antibiotics, those who were aged ≥65 years, and those with bacteremia but superior to ceftriaxone in patients with PORT risk III and those who did not

receive previous antibiotics. In summary, the overall clinical efficacy of ceftaroline is similar to that of ceftriaxone for the treatment of CAP. For other populations, ceftaroline is at least similar to ceftriaxone in terms of the clinical cure rate. However, the case numbers of several subgroup analyses, such as bacteremia, PORT risk IV or different pathogens were limited, which may limit the significance of differences between ceftaroline and ceftriaxone. Therefore, a further large-scale study is warranted to prove our findings.

In the mMITT population, the present meta-analysis determined that the clinical cure rate of ceftaroline was superior to that of ceftriaxone for CAP caused by GPC, but no significant difference was found for CAP caused by GNB, *S. pneumoniae, S. aureus, H. influenzae, H. parainfluenzae, E. coli,* and *K. pneumoniae* between the ceftaroline and ceftriaxone groups. The effectiveness of ceftaroline for the treatment of CAP is supported by in vitro studies. In a surveillance study at a US medical center, ceftaroline was noted to be more potent against *S. pneumoniae* (MIC50 ≤ 0.015 vs. ≤ 0.06 μg/mL; MIC90 = 0.12 vs. 1 μg/mL) and even remained active against strains nonsusceptible to ceftriaxone (MIC90 = 0.25 μg/mL) [18]. Similar findings were demonstrated in the analysis of bacterial isolates in pediatric patients [15]. Upon global surveillance, ceftaroline was noted to be more potent than ceftriaxone against MSSA and *S. pneumoniae,* and ceftaroline had similar efficacy to ceftriaxone against *H. influenzae* [17]. Overall, the in vitro activity of ceftaroline that is greater or at least equal to ceftriaxone against most commonly encountered pathogens causing CAP could largely explain the high in vivo clinical response in this meta-analysis. However, we can only see the trend of better efficacy of ceftaroline than the comparator in each subgroup; these differences do not reach statistical significance. This may be due to the limited case number of each pathogen, so further large-scale study is warranted.

Although this study demonstrated the clinical efficacy of ceftaroline in the treatment of CAP, antibiotics may have a limited effect on the outcome of CAP, particularly these severely affected cases. This could be due to the fact that pneumonia is caused by a variety of pathogens, including respiratory viruses, Mycoplasma pneumoniae, and bacteria. In addition, incidence of primary bacterial pneumonia may be very low and be far less than that of nonbacterial pneumonia in developed countries as well as in developing countries [27]. Incidences of each pathogen pneumonia may differ in children and adults (older persons) across the populations, but severe pneumonia of viral or nonpathogen origin can induce secondary bacterial infection caused by lung injuries from primary insults; hence, it is reasonable that any pneumonia patients could be treated with antibiotics. However, antibiotics have a limited effect on the natural course of infection-related extrapulmonary manifestations. Further, outcomes of severe pneumonia may be affected by underlying comorbidities or the immune status of the host, not only by antibiotic treatment. Moreover, the pattern of antimicrobial resistance may vary in different sites; therefore, the guidelines for antibiotic treatment for CAP may differ and could be changed in each country over time. In summary, although the appropriate use of antibiotics is essential for the successful treatment of pneumonia, many factors, including disease severity, underlying comorbidity, immune status, pathogens, and the timing of antibiotic use are also significantly associated with the outcome of pneumonia.

The risk of AEs is another major concern when treating CAP with this antimicrobial agent. The most common AEs are headache, diarrhea, and insomnia [28]. In this analysis, the pooled risks of TEAEs of all degrees and even serious AEs were similar between the ceftaroline and ceftriaxone groups. Additionally, ceftaroline is associated with the risk of discontinuation of the study drug that is similar to that of ceftriaxone; this risk is because of the development of AEs. Although the overall mortality of the ceftaroline group was only 1.81%—which was comparable to that of ceftriaxone group—none of the cases of mortality were associated with the study drug. Therefore, all these findings revealed that ceftaroline is as safe as ceftriaxone for the treatment of CAPs.

A major strength of this meta-analysis is that only RCTs were included, thereby reducing the risk of bias and providing strong evidence. However, this meta-analysis also has several limitations. First, the number of MRSA-associated pneumonia cases was limited in this study. Therefore, the anti-MRSA effect of ceftaroline, which is not owned by ceftriaxone, cannot be elucidated in this meta-analysis. Second, this meta-analysis had a limited number of studies and patients in subgroup analyses, such as different pathogens among different age groups. Therefore, some differences between the ceftaroline and ceftriaxone groups did not reach statistical significance.

#### **5. Conclusions**

In conclusion, based on the findings of this meta-analysis of five RCTs, the clinical efficacy of ceftaroline is similar to that of ceftriaxone for the treatment of CAP. Additionally, ceftaroline was as tolerable as ceftriaxone. However, clinicians should cautiously use ceftaroline in the selected population at high risk of MRSA to avoid the unnecessary coverage of MRSA by ceftaroline. Overall, ceftaroline can be recommended as an appropriate antibiotic therapy for CAP.

**Author Contributions:** Conceptualization, S.-H.L., S.-P.C., C.-C.L., C.-M.C.; methodology: S.-H.L., S.-P.C., L.-C.L.; writing—original draft preparation, C.-C.L., C.-M.C.; writing—review and editing, C.-M.C.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Use of Secukinumab in a Cohort of Erythrodermic Psoriatic Patients: A Pilot Study**

**Giovanni Damiani 1,2,3,4,\*,**†**, Alessia Pacifico 5,**†**, Filomena Russo 6, Paolo Daniele Maria Pigatto 2,3, Nicola Luigi Bragazzi 7, Claudio Bonifati 5, Aldo Morrone 5, Abdulla Watad 8,9,**‡ **and Mohammad Adawi 10,**‡


Received: 17 May 2019; Accepted: 29 May 2019; Published: 31 May 2019

**Abstract:** Erythrodermic psoriasis (EP) is a dermatological emergency and its treatment with secukinumab is still controversial. Furthermore, no data exist regarding the prognostic value of drug abuse in such a condition. We performed a multi-center, international, retrospective study, enrolling a sample of EP patients (body surface area > 90%) who were treated with secukinumab (300 mg) during the study period from December 2015 to December 2018. Demographics and clinical data were collected. Drug abuses were screened and, specifically, smoking status (packages/year), cannabis use (application/week) and alcoholism—tested with the Alcohol Use Disorders Identification Test (AUDIT)—were assessed. All patients were followed for up to 52 weeks. We enrolled 13 EP patients, nine males, and four females, with a median age of 40 (28–52) years. Patients naïve to biologic therapy were 3/13. Regarding drug use, seven patients had a medium-high risk of alcohol addiction, three used cannabis weekly, and seven were smokers with a pack/year index of 295 (190–365). The response rate to secukinumab was 10/13 patients with a median time to clearance of three weeks (1.5–3). No recurrences were registered in the 52-week follow-up and a Psoriasis Area Severity Index (PASI) score of 90 was achieved. The entire cohort of non-responders (*n* = 3) consumed at least two drugs of abuse (alcohol, smoking or cannabis). Non-responders were switched to ustekinumab and obtained a PASI 100 in 24 weeks. However, given our observed number of patients using various drugs in combination with secukinumab in EP, further studies are needed to ascertain drug abuse prevalence in a larger EP cohort. Secukinumab remains a valid, effective and safe therapeutic option for EP.

**Keywords:** erythrodermic psoriasis; secukinumab; addiction; smoking; alcohol; cannabis

#### **1. Introduction**

Erythroderma is an uncommon and severe dermatological manifestation of a variety of diseases. The most common form of erythroderma is erythrodermic psoriasis (EP), which accounts for 1–2.25% of all psoriatic patients, with a male predominance as demonstrated by a male to female ratio of 3:1 [1]. EP clinically manifests with diffuse erythema (body surface area (BSA) > 75%) involving also skin folds with or without exfoliate dermatitis.

Several triggers have been described to elicit EP in predisposed subjects such as environmental factors (sunburn, alcoholism, and infections), drugs (lithium, anti-malarial drugs), and the rebound phenomenon following discontinuation of anti-psoriatic treatments (oral steroids or methotrexate) [1]. However, the pathogenesis of EP remains elusive, which can limit a physician's capability to deliver safe and effective therapy. In 2010, the National Psoriasis Foundation (NPF) published a guideline describing the current evidence regarding EP treatment, stating that cyclosporine and infliximab should be the first line treatment in acute and unstable patients, whilst methotrexate and acitretin are recommended in more stable patients [2].

Despite this clear advice, prominent limitations included that few high-quality studies assessing EP treatment were present in the literature [2]. In a clinical setting, EP treatment faces two other prominent challenges, namely the difficulty in differential diagnosis and in implementing a biological approach that rules out non-inflammatory conditions. Although histological confirmation is mandatory in suspected EP cases, it is sometimes challenging due to the potential lack of histological parameters resembling classical psoriasis, such as parakeratosis or acanthosis [1].

The exclusion of neoplastic causes (Sézary syndrome) is mandatory if biologics are the selected approach. In fact, in the last 5 years, neoplasia has been a relative contraindication [3]. The NPF guidelines did not include IL-17 inhibitors [2], such as secukinumab, and recently two case series studies described the use of secukinumab in EP patients [4,5]. Current evidence seems to support the use of secukinumab in EP patients, even though there is a dearth of data concerning potential predictors of responsiveness in these patients.

Remarkably, among psoriatic patients, alcohol use/abuse and smoking are described and linked to both psoriasis development and exacerbations but are not studied in EP [6–10]. Conversely, the prevalence of cannabis users among psoriatic patients and the effect of cannabis use on psoriasis are still missing. Furthermore, in vitro or murine studies explored keratinocyte changes only in response to a single cannabis compound [11]. Thus, due to the increasing prevalence of cannabis users in the general population and also its promoting role in medicine [11], reports focusing on the effect in psoriasis are needed.

The current study aimed to evaluate (i) first the efficacy and safety of secukinumab in psoriatic erythroderma and (ii) second to describe the prevalence of drug abuses, namely alcohol, tobacco, and cannabis smoking, in EP patients.

#### **2. Experimental Section**

This multi-center, international, retrospective, pilot study enrolled a sample of EP patients (BSA > 90%) treated with a loading dose of 300 mg subcutaneous secukinumab at weeks 0, 1, 2, 3 and 4, followed by 300 mg every 4 weeks, in the period from December 2015 to December 2018.

All erythrodermic patients were biopsied and malignancies were ruled out by complete blood count, blood smear, transaminases, lactate dehydrogenase (LDH), gamma-glutamyl transferase (GGT), anion gap, Sézary cell search, and total body computed tomography. Smoking history (pack/years), cannabis use (smoking episodes/week) and alcohol use (Alcohol Use Disorders Identification Test (AUDIT)) status were assessed.

AUDIT is a 10-question screening tool (0–40 points) developed by the World Health Organization (WHO) in order to evaluate alcohol consumption, drinking behavior, and alcohol-related complications. According to AUDIT, patients are stratified as follows: 0–7 points indicate a low risk, 8–15 points a medium risk, 16–19 points a high risk, and 20–40 points a probable addiction.

All EP patients underwent a 52-week follow-up to evaluate recurrent erythrodermic episodes.

Demographics and clinical charts were recorded, including: age; gender; previous Psoriasis Area Severity Index (PASI) score before erythroderma, if any; previous anti-psoriatic therapy; biologic therapy exposure; secukinumab response; side effects; drug use history; PASI and Dermatologic life quality index (DLQI) at weeks 8, 12, 16, 24, and 52. We stratified erythroderma clearance (BSA < 75%) based on PASI 75, PASI 90, PASI 100.

#### **3. Results**

#### *3.1. Study Population*

In the current study, 13 EP patients (female/male ratio equal to 9/4), with a median age of 40 (28–52), and body mass index of 24 (22–27) kg/m2 were included. Family history of psoriasis was positive in 9/13 patients.

#### *3.2. Drug History*

In Table 1 we assessed drug history. Only 3/13 patients were naïve to biologic therapy. Among patients treated with biologics, eight had switched more than two biologics. Furthermore, 8/13 had a previous episode of erythroderma and six patients had more than two episodes. Drug history indicated that some of the EP patients had previously received therapeutic agents that could potentially trigger psoriasis, namely four underwent beta blockers, three received angiotensin II blockers (ARBs), two patients received angiotensin-converting enzyme (ACE) inhibitors, and one patient was previously treated with thiazide diuretics.



ACE: Angiotensin-converting enzyme, ARBs: Angiotensin II receptor blockers, EP: erythrodermic psoriasis, MTX: Methotrexate.

#### *3.3. Drug Abuses and Comorbidities*

Drug abuse screening revealed that seven patients had a medium-high risk of alcohol abuse, three patients used cannabis on a weekly basis, and seven patients were smokers with a pack/year index of 295 (190–365). The comorbidities represented in our cohort included: dyslipidemia (five patients), hypertension (three patients), osteoporosis (two patients), atrial fibrillation (one patient) and pulmonary tuberculosis (one patient), respectively (Table 2).


**Table 2.** Prevalence of drug abuses in our cohort.

AUDIT: Alcohol Use Disorders Identification Test, EP: erythrodermic psoriasis, IQR: Interquartile range, SD: standard deviation.

#### *3.4. Clinical Response to Secukinumab*

Clinical and therapeutic data are summarized in Table 3. The median value of the last recorded PASI was 10 (7–15). Responders to secukinumab were 10/13 (Figure 1a,b) and the median clearing time was three (1.5–3) weeks.


**Table 3.** Clinical and therapeutic records in our cohort.


**Table 3.** *Cont*.

DLQI: Dermatologic Life Quality Index, EP: erythrodermic psoriasis, IQR: Interquartile range, MTX: Methotrexate, PASI: Psoriasis Area Severity Index.

(**a**) (**b**)

**Figure 1.** A 34-year-old patient with erythrodermic psoriasis that underwent secukinumab therapy. (**a**) Erythrodermic patient before treatment, (**b**) Patient after three weeks of secukinumab treatment.

Side effects were reported in 5/13 patients and remarkably were the only cause of treatment interruption, in contrast to other previously reported cases series [4,5]. All patients were on continuous secukinumab treatment and no recurrences were registered in the 52 weeks of follow up. After recovering from erythroderma at week eight, four patients achieved PASI 75, while none achieved PASI 90 or PASI 100. At week 52, five patients achieved PASI 90 and five achieved PASI 100. Interestingly, looking at the PASI trends of this cohort (Figure 2), all three non-responders used two out of three of the aforementioned drugs (alcohol, cannabis, and smoking) and no recorded comorbidities. Non-responders were switched to ustekinumab (90 mg) and obtained a PASI 100 in 24 weeks.

**Figure 2.** PASI trends in erythrodermic patients from week eight to week 52. \* Patients that displayed more than one type of drug use (tobacco, cannabis, alcohol).

#### **4. Discussion**

Our study further supports that secukinumab is an effective therapy in EP and suggests that the use of recreational and accepted drugs (alcohol, cannabis, and tobacco) is prevalent in EP patients.

Furthermore, in the literature, EP patients treated with secukinumab had 16 [4] or 24 [5] months of follow up, lacking an assessment of long-term DLQI. Thus we assessed DLQI at 8, 12, 16, 24, and 52 weeks and found that secukinumab contributed to the improvement, not only in skin disease, but also in the long-term quality of life of EP patients. In our cohort, EP patients that responded to secukinumab did not exhibit recurrences and maintained long-term responsiveness to the drug. This study further supports the results described in a retrospective 52-week, observational, multicenter study, evaluated by Galluzzo et al., which suggested long-term efficacy of secukinumab in plaque psoriasis [12].

Focusing on EP patients, we assessed for the first time in detail the timing related to clearance of erythroderma, and after that, how secukinumab managed to clear the residual plaque psoriasis during the 52-week follow-up period. These two parameters together, are of pivotal importance in the clinical setting to guide therapeutic decisions made by dermatologists. In addition, the 52-week follow-up data highlighted that the EP responders to secukinumab achieved at least PASI 90 after clearing EP.

Among the three patients who did not respond to secukinumab therapy, one patient developed generalized urticaria at week three, the second patient experienced recurrent oral candidiasis and stopped the drug at week 12, and the third patient lost response at week 16. Remarkably, the second non-responder also smoked cannabis. Non-responders have not previously been treated with ustekinumab, and in accord with the recent real-life data on secukinumab non-responders, they were switched to ustekinumab and achieved a complete remission [13]. Ustekinumab is an IL-12/IL-23 blocker that targets the p40 subunit shared by these two cytokines. Furthermore, IL-12 plays a pivotal role in T helper cell type 1 (Th-1) polarization, as does IL-23 in Th-17 polarization [14]. We interpret the non-responsiveness of our patients as potentially due to the development of anti-secukinumab antibodies or up-regulation of Th-1-related pro-inflammatory cytokines, as previously demonstrated by Zaba et al. [15].

Evaluation of clinical characteristics in secukinumab non-responders indicated that all three had a familial history of EP and had used more than one drug, including smoking, alcohol, and cannabis. However, none of them were treated with any drug known to trigger psoriasis.

In the literature, the prevalence of drug abuse in the rare subset of EP is not reported, conversely, in plaque psoriasis patients several authors addressed the problem of drug abuse prevalence (alcohol and tobacco smoking) and its impact on anti-psoriatic therapies [7,8,16,17].

Alcohol intake and, consequently, also the abuse, may favor psoriasis-related systemic inflammation by promoting lipopolysaccharide (LPS) translocation from intestine to blood flow, increasing the pro-inflammatory activation of several immune cells, including lymphocytes (producing TNF-α and IFN-γ) and monocytes/macrophages (producing TNF-α), and directly by triggering keratinocytes pro-inflammatory activation via keratinocyte growth factor receptor (KGFR) [9]. These observations are further supported by Brenaut and colleagues, who conducted a meta-analysis on the epidemiological link between psoriasis and alcohol intake, and found that alcohol is a risk factor in developing psoriasis [17]. Furthermore, Qureshi et al. described a correlation between heavy beer intake and psoriasis severity during exacerbation [8]. This concept is translatable to EP patients, where erythroderma is an acute and very severe exacerbation of pre-existent psoriasis. Thus, alcohol abuse seems to increase TNF-α levels and may theoretically explain a possible lack or loss of response to IL-17 blockers, as with secukinumab in our EP patients.

Tobacco smoking and its link with psoriasis was assessed by Armstrong and colleagues in a large meta-analysis, involving 28 studies. They found an odds-ratio (OR) of 1.78 (95% confidence interval = 1.52–2.06) and a higher PASI in psoriatic smokers compared to non-smokers [16]. Remarkably, psoriasis severity gradually increases with the number of cigarettes smoked per day [17], but may benefit from a stop in smoking [18,19]. The nicotine contained in cigarettes activates nicotinic acetylcholine receptors (nAChRs) on the surface of dendritic cells, macrophages, endotheliocytes and keratinocytes, leading to an increased Th-1/Th-17 polarization of naïve T cells and to an increased production of pro-inflammatory cytokines, such as TNF-α, IL-12, IL-17, IL-23, IL-1β and IFN-γ [20]. These are all capable of decreasing the therapeutic effects of both TNF-α [7] and IL-17 blockers.

Conversely, fragmentary data exist regarding the effects of cannabis on the immune system and skin [21,22], but no data have been published about cannabis smoking in psoriatic patients or in murine models of psoriasis. However, some purified extracts derived from cannabis may inhibit in vitro keratinocyte proliferation [21] and Th-17 cell-related cytokine production in a dose-dependent manner [22]. Cannabinoids mainly interact with two receptors, cannabinoid-1 receptor (CB1R) and (CB2R), and both inhibit adenylate cyclase and activate mitogen-activated protein kinase (MAPK) [11]. This theoretically contrasts the anti-psoriatic function of apremilast, with regard to the intracellular cyclic adenosine monophosphate (AMPc) increase due to phosphodiesterase-4 inhibition. CB1R is prevalently present in keratinocytes, whilst CB2R is prevalent in immune cells, such as T cells and monocytes/macrophages [11]. Upon stimulation in the presence of purified cannabis extracts, namely cannabidiol (CBD) and tetrahydrocannabinol (THC), Th-17 cells massively decrease both transcription and release of IL-17A [22], which may theoretically act synergistically with IL-17 blockers. This aspect may be also confirmed by reports that list candidiasis as a side effect of both IL-17 blockers and chronic cannabis use [23]. Consequently, patients under IL-17 blockers that use cannabis may be exposed to a higher risk of candidiasis. Remarkably, Russo and colleagues pointed out that, in order to evaluate the global effect of cannabis, it is necessary to take into consideration the synergism existing among different cannabis compounds that altogether determine the final so-called entourage effect, capable of enhancing or even obscuring the properties of single compound [24]. Furthermore, no studies evaluated how smoking cannabis can modify these compounds and their biological effect. Thus, this is the first report to evaluate this relevant use of such drugs in a cohort of patients affected by EP, a chronic systemic inflammatory disease.

Moreover, both smoking and alcohol consumption were found to increase IL-17 and TNF-α production [9,12,16], corroborating our hypothesis that drug use may promote systemic inflammation, contributing to less favorable results from anti-psoriatic therapies.

The main limitation of the present study remains the small sample of enrolled patients, which was due to EP rarity and due to the fact secukinumab is still off-label in treating EP. Therefore, we cannot conclude that drug use in the non-responding patient group was causal. Other plausible reasons for the failed response in the small number of patients with addiction problems in the present cohort might well be insufficient compliance, even though all of our patients regularly attended dermatological appointments and reported to have auto-injected secukinumab.

#### **5. Conclusions**

Although not conclusive, our preliminary results in EP patients treated with secukinumab enlighten two presently unmet needs: (i) the need of therapy-specific biomarkers/prognostic factors and (ii) the prevalence of drug use in EP.

In conclusion, secukinumab may be a safe and effective treatment in EP, however, larger studies are needed to validate our results.

**Author Contributions:** Conceptualization, G.D. and A.W.; methodology, G.D., F.R. and N.L.B.; software, G.D. and N.L.B.; validation, G.D. and A.W.; formal analysis, G.D. and N.L.B.; investigation, G.D.; resources, M.A.; data curation, G.D. and N.L.B.; writing—original draft preparation, G.D. and C.B.; writing—review and editing, G.D., A.P., F.R., P.D.M.P., N.L.B., C.B., A.M., A.W. and M.A.; visualization, A.W. and M.A.; supervision, G.D., A.M. and M.A.; project administration, G.D., P.D.M.P., C.B. and A.M.; funding acquisition, M.A.

**Acknowledgments:** G.D. is supported by the P50 AR 070590 01A1 National Institute of Arthritis and Musculoskeletal and Skin Diseases.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Epidemiology and Burden of Diabetic Foot Ulcer and Peripheral Arterial Disease in Korea**

**Dong-il Chun 1,**†**, Sangyoung Kim 2,**†**, Jahyung Kim 1, Hyeon-Jong Yang 3, Jae Heon Kim 4, Jae-ho Cho 5, Young Yi 6, Woo Jong Kim <sup>7</sup> and Sung Hun Won 1,\***


Received: 9 May 2019; Accepted: 23 May 2019; Published: 25 May 2019

**Abstract:** Information about the epidemiology of diabetic foot ulcer (DFU) with peripheral arterial disease (PAD) is likely to be crucial for predicting future disease progression and establishing a health care budget. We investigated the incidence and prevalence of DFU and PAD in Korea. In addition, we examined costs of treatments for DFU and PAD. This study was conducted using data from Health Insurance Review and Assessment Service from 1 January 2011 to 31 December 2016. The incidence of DFU with PAD was 0.58% in 2012 and 0.49% in 2016. The prevalence of DFU with PAD was 1.7% in 2011 to 1.8% in 2016. The annual amputation rate of DFU with PAD was 0.95% in 2012 and 1.10% in 2016. Major amputation was decreased, while minor amputation was increased. The direct cost of each group was increased, especially the limb saving group. which was increased from 296 million dollars in 2011 to 441 million dollars in 2016. The overall incidence of DFU with PAD was about 0.5% of total population in Korea, from 2012 to 2016. Furthermore, costs for treatments of diabetic foot ulcer are increasing, especially those for the limb saving group.

**Keywords:** diabetic foot ulcer; peripheral arterial disease; incidence; prevalence; cost; National Health Insurance Service data

#### **1. Introduction**

The prevalence of diabetes mellitus is expected to increase and the number of diabetic patients worldwide is on the rise. The global prevalence of diabetic foot varies from 3% in Oceania to 13% in North America, with a global average of 6.4% [1]. The annual incidence of diabetic foot ulcer (DFU) or necrosis in diabetic patients is known to be about 2% to 5% and the lifetime risk ranges from 15% to 20% [2–4]. Peripheral arterial disease (PAD), like cardiovascular disease, is a major arterial disease caused by atherosclerosis [5]. Diabetes is one of the high risk factors of PAD [5], and Olinic et al. [6] reported that the prevalence of PAD in Europe is increasing, parallel with increasing age and other risk factors for cardiovascular disease. PAD is associated with a 20-fold higher prevalence in patients with diabetes. It is known to be a risk factor for the highest severity of single factors in diabetic patients [7–9]. In addition, the probability of amputation within one year after the first ulcer or gangrene is 34.1% and the mortality rate has been reported to be 5.5% [8]. Information about the epidemiology of peripheral arterial disease associated with DFU is likely to be crucial for predicting future disease progression and establishing a health care budget.

About 20% to 33% of costs related to diabetes mellitus are used for treatments of diabetic foot [3,10]. The incidence of diabetes represented by chronic diseases is increasing. The cost of medical care for diabetic foot is increasing. Korea has recently entered an aging society. The increase in the number of diabetic patients has become an important issue in the decision of the health and welfare budget in Korea. In addition, the increase in complications due to diabetes is a burden, not only for patients, but also for the nation. Furthermore, such information is important for public health policy makers to advocate for implementation of prevention and treatment recommendations. However, there are no recent studies on the incidence, prevalence, or costs of treatments of DFU and PAD in Korea.

Thus, the primary objective of this study was to investigate the incidence and prevalence of DFU and PAD in Korea. The secondary objective was to analyze the costs of treatments for DFU and PAD using National Health Insurance Service data provided by the Health Insurance Review and Assessment Service (HIRA).

#### **2. Materials and Methods**

This study was approved by the Institutional Review Board of Soonchunhyang University Hospital Seoul (Institutional Review Board number: SCHUH 2018-01-007). The use of codes directly signifying DFU began on 1 January 2011, when the sixth edition of the Korean statistical classification of disease and related health problems-6 system (KCD-6) was applied. Until the year 2010, the disease code indicating the gangrene and ulcer was used separately from the diabetes code. If the disease code of the foot wound was not actively recorded, even if there was a DFU, DFU patients were inevitably missing. Therefore, we judged that it was not accurate to investigate data before 2010. Finally, data after 2011 were examined in this study. This study was conducted using data from HIRA from 1 January 2011 to 31 December 2016.

The annual incidence and prevalence of diabetes foot ulcer and PAD among the total population of Korea (estimated population) reported by the National Statistical Office were calculated. We considered the wash-out period as one year to determine the annual incidence of DFU and PAD. Therefore, the annual incidence of newly diagnosed DFU and PAD patients was calculated from 2012.

The amputation rates in diabetic foot ulcer and PAD patients were also calculated according to amputation level (minor vs. major (above ankle)). Diabetic foot ulcer and PAD codes and behavior codes (such as amputation, debridement, etc.) included in this study are summarized in Table 1.


**Table 1.** Diabetic foot ulcer (DFU) and peripheral arterial disease (PAD) codes and behavior codes included in this study.


The direct cost for each amputation was calculated. We also analyzed the direct costs of DFU and PAD care in three groups. Group I was a limb-saving group. Group II was for those who had one amputation. Group III was for patients who had repeated amputation. The cost was based on the direct cost of patient contributions plus insurance claims. The direct cost was adjusted by taking into account the medical price index presented by the Korean Statistical Information Service (KOSIS, Daejeon, Korea). The data of this study were analyzed using SAS Enterprise Guide, ver. 6.1 M1 (SAS Institute Inc., Cary, NC, USA).

#### **3. Results**

Regarding the overall annual incidence of DFU from 2012 to 2016, 0.43% of total populations were diagnosed with DFU in 2012 whereas 0.34% were diagnosed in 2016, showing a remarkable incidence plateau with a mild decrease over five years. The annual incidence of PAD was 0.19% in 2012 and 0.20% in 2016, showing an incidence plateau with a mild increase over five years. The annual incidence of DFU with PAD was 0.58% in 2012 and 0.49% in 2016 (Figure 1). The overall prevalence of DFU in the study period was 1.4% in 2011 and 1.3% in 2016. The prevalence of PAD was 0.4% in 2011 and 0.5% in 2016. The prevalence of DFU with PAD showed a mild increase from 1.7% in 2011 to 1.8% in 2016 (Figure 1).

**Figure 1.** Annual the incidence and prevalence of diabetic foot ulcer with PAD.

The annual amputation rate of DFU with PAD was increased from 0.95% in 2012 to 1.10% in 2016. Of these, the major amputation rate was decreased from 0.28% in 2012 to 0.27% in 2016, while the minor amputation rate was increased from 0.66% in 2012 to 0.82% in 2016 (Figure 2).

The direct cost of amputation was increased from 17 million dollars in 2011 to 25 million dollars in

2016. Especially, the sum of direct costs of minor amputation increased from 11 million dollars in 2011 to 17 million dollars in 2016 (Figure 3). The average cost of amputation per person was also increased from 6100 dollars in 2011 to 7300 dollars in 2016 (Figure 4). The direct cost of each group was increased from 2011 to 2016. Direct costs for group 1 increased from 296 million dollars in 2011 to 441 million dollars in 2016. These costs for group 2 increased from 7.1 million dollars in 2011 to 9.3 million dollars in 2016, while those for group 3 increased from 10 million dollars in 2011 to 15 million dollars in 2016 (Figure 5).

**Figure 3.** The direct cost of amputation of diabetic foot ulcer with PAD.

**Figure 2.** Annual amputation rate of diabetic foot ulcer with PAD.

**Figure 4.** The average cost of amputation, per person, of diabetic foot ulcer with PAD.

**Figure 5.** The direct cost of each group for diabetic foot ulcer with PAD.

#### **4. Discussion**

Overall incidence and prevalence of DFU with PAD in Korea from 2012 to 2016 were about 0.5% and 1.7% of the total population, respectively. The amputation rate was increased, especially the minor amputation rate, which increased from 0.66% in 2011 to 0.82% in 2016. Furthermore, direct costs for diabetes treatment were increased, especially the expense for the limb saving group.

We investigated the annual incidence and prevalence of DFU and PAD among the total population in Korea. The incidence and prevalence of DFU are also important for determining the number of diabetic patients. We requested HIRA to provide the total data for diabetic patients. However, the organization explained to us that these data were too large to release. Therefore, we could not obtain information for the total number of diabetic patients during the study period. Thus, we calculated the incidence and prevalence of DFU patients in the total population. However, the Korean diabetic association reported that the prevalence of diabetes increased from 12.4% in 2012 to 14.4% in 2016 through a diabetic fact sheet in 2018. Thus, we could investigate the prevalence of diabetic foot ulcer among the diabetic patients indirectly, showing 10% in 2012 and 9% in 2016. Recently, Zhang et al. [1] reported that the global prevalence of diabetic foot ulceration is 6.3% and the prevalence is 13.0% in North America and 5.5% in Asia. However, in their systematic review and meta-analysis study, the definitions for diabetic foot and diabetic foot ulceration were ambiguous. Furthermore, two epidemiologic studies using only Korean data focused on the epidemiology of diabetic peripheral neuropathy [11,12]. However, our study investigated not only DFU, but also PAD. Thus, we believe that our study has a more accurate prevalence of DFU in Korea.

Concerning amputation, our results showed that annual amputation rate of DFU with PAD increased from 0.95% in 2012 to 1.10% in 2016. Of these, the major amputation rate decreased from 0.28% in 2012 to 0.27% in 2016, while the minor amputation rate was increased from 0.66% in 2012 to 0.82% in 2016. The overall amputation rate was increased. This might be due to the increased minor amputation rate, rather than the decrease of the major amputation rate. Goodney et al. reported that lower extremity amputation decreased by 45% from 1996 to 2011 (above the knee amputation decreased by 48% and below the knee amputation decreased by 39%) [13]. Although, in our study, we did not show a clear causal relationship about the reason for this situation, two reasons might be important. The first one is the increased awareness of the risk of diabetic foot in diabetic patients. Previous studies have reported that education about foot care to diabetic patients is important because it is associated with a significant reduction in lower extremity amputation. In addition, monthly foot checks are associated with the reduction of major lower limb amputations in diabetic incident hemodialysis patients [14,15]. Future studies are needed to examine awareness of risk of DFU in diabetic patients. The second reason is the improvement of vascular conditions due to increase of revascularization. This is also important for the reduction of major amputation. Peripheral vascular disease is known to be the most significant risk factor for diabetic foot amputation [8]. A previous study also reported that it is evident that the increasing use of vascular and preventive care, especially among patients with diabetes, is temporally associated with lower rates of major amputation [16]. Future studies focusing on understanding this relationship are needed.

The increase in the cost of medical care, due to the increased number of diabetic patients, has been reported all over the world. It is a useful indicator for planning and enforcing health care policies and budgets [17–19]. To the best of our knowledge, our research is the first to investigate the cost of medical care for diabetic foot ulcer in Korea. Our study showed that, although the cost of amputation was increased a lot, the expense for the limb saving group increased exponentially. Such a cost increase might be due to the development of medications and dressing materials. This could increase the burden on the patient. Understanding the cost of DFU should support future decisions on investment in diabetic foot care.

Some limitations of the study need to be addressed. First, DFU and PAD codes are diverse, unclear, and sometimes missing. There was no defined disease code for DFU until 2010. Thus, data on DFU from 2007 to 2010 could not be used. This is considered the limit for using National Health Insurance Service data provided by HIRA. It is necessary to agree on codes of diabetic foot and PAD more clearly and uniformly in the future. Second, diabetic neuropathy was not included in this study, because the disease code and operational definition for diabetic neuropathy were not established in the Big Data. Furthermore, extracted data using the provided diabetic neuropathy code revealed that too many patients were included, so the extracted data could not be trusted. Therefore, in this study, diabetic neuropathy was excluded in order to improve the quality of the study, but it is thought that studies to include neuropathy in the diabetic foot should be done through the operational definition defined later. Third, when calculating the incidence of chronic diseases, such as diabetes and PAD, a wash-out period of at least 2 years should be used. However, we used wash-out period of one year to calculate incidence of DFU and PAD. This was an inevitable choice due to too much missing data. If data for

a longer period of time can be used, the wash-out period of 2 years can be used. Such study is needed in the future.

#### **5. Conclusions**

In conclusion, over 5 years, we found that the overall incidence and prevalence of DFU with PAD in Korea were about 0.5% and 1.7% of the total population, respectively. The amputation rate was increased, especially the minor amputation rate. Furthermore, the direct costs for DFU treatment were increased, especially the expense for the limb saving group. Our results suggest that we should pay attention to effective implementation of the budget when we make future health policies for diabetic foot. Further studies warrant the importance of productive and cost-effective methods for saving limbs in the healthcare system.

**Author Contributions:** Conceptualization, J.-h.C. and S.H.W.; Data curation, J.K.; Formal analysis, D.-i.C. and S.K.; Investigation, H.-J.Y., Y.Y., and W.J.K.; Methodology, J.H.K.

**Funding:** This work was supported by the National Research Foundation of Korea (NRF) grant funded by the Korea government (MSIT) (No. M20180308941). This work was supported by the Soonchunhyang University Research Fund.

**Acknowledgments:** We thank Mi-soon Lim, Dong won Shin for contributing to collect the full text articles.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

### *Article* **Association between Serum Urate and Risk of Hypertension in Menopausal Women with XDH Gene**

### **Jong-Han Lee 1, Tae Hwa Go 2, San-Hui Lee 3, Juwon Kim 1, Ji Hye Huh 4, Jang Young Kim 4, Dae Ryong Kang 2, Seongmun Jeong 5, Sang-Baek Koh <sup>6</sup> and Jung Ran Choi 7,\***


Received: 10 April 2019; Accepted: 20 May 2019; Published: 23 May 2019

**Abstract:** Elevated serum urate (sUA) concentrations have been associated with an increased risk of hypertension. We aimed to examine the association of sUA concentration on the risk of hypertension in pre- and post-menopausal women and investigated the association between the polymorphism of the xanthine dehydrogenase gene and the risk of hypertension. Among 7294 women, 1415 premenopausal and 5879 postmenopausal women were recruited. Anthropometric parameters as risk factors of hypertension were identify by logistic regression models. In addition, we investigated an association between xanthine dehydrogenase gene and sUA and their combined associations on the risk of hypertension. Body mass index (BMI) and waist circumference (WC) were significantly increased in accordance to the increase of sUA levels (*p* < 0.001). Multivariate logistic regression analysis showed postmenopausal women with a high sUA and high BMI were 3.18 times more likely to have hypertension than in those with normal and lower sUA (Odds ratio: 3.18, 95% confidence interval: 2.54–3.96). Postmenopausal women with a high WC were 1.62 times more likely to have hypertension than in those with normal and lower sUA. Subjects with the AG genotype of rs206860 was found to be at lower risk of hypertension (odd ratio: 0.287, 95% confidence interval: 0.091–0.905, *p* = 0.033). This cross-sectional study indicated a high sUA is associated with a higher risk of hypertension in postmenopausal women. Further well-designed prospective studies in other populations are warranted to validate our results.

**Keywords:** serum urate; menopause; hypertension; xanthine dehydrogenase; cross-sectional cohort study

#### **1. Introduction**

An elevated serum urate (sUA) concentration is a common phenomenon in subjects with hypertension, insulin resistance, or obesity [1], and previous epidemiological studies have demonstrated that high sUA concentrations are associated with an increased risk of hypertension [1–5]. Although longitudinal studies showed sUA might play a role in the development of hypertension, the association between sUA and blood pressure may be affected by various factors [5–7]. However, it is not clear whether urate elevation is the cause or a consequence of hypertension [1,6,8].

Urate is the catabolic end-product of endogenous and dietary purine metabolism in human, and is mainly produced by xanthine oxidase, which is involved in the production of reactive-oxygen species (ROS) [5,9–11]. In addition, excessive sUA accumulation can cause various diseases [11,12]. Several studies using animal models and cell cultures have identified mechanisms whereby high sUA concentrations might lead to hypertension by reducing endothelial nitric oxide release and activating the renin–angiotensin system leading to smooth muscle cell proliferation [13–16]. Furthermore, associations between sUA and metabolic disorders are gender dependent. sUA concentrations tend to be lower in women than in men, partially due to the uricosuric effect of estrogens [10]. In addition, sUA concentrations seem to be increased in both physiologic and post-surgical menopause independently of other confounders [9,17], presumably due to the uricosuric effect of estrogens and hormone replacement therapy induced sUA reduction [18]. Estrogens have some indirect uricosuric effects that can contribute to modulate the urate before and after menopause. The sUA concentration is unlikely to reflect urate production, as an increased sUA production is compensated by increased excretion to maintain sUA within the normal range [19]. To the best of our knowledge, urate production has a limited influence on sUA concentration and both, urate concentration and the production, might influence blood pressure by a different underlying mechanism.

Mammalian xanthine oxidoreductase (XOR) has the characteristics of being found in two interconvertible forms: constitutively expressed in vivo NAD+-dependent xanthine dehydrogenase (XDH, EC 1.1.1.204) and post-transcriptionally modified xanthine oxidase (XO, EC 1.1.3.22) [20]. The XOR enzyme exists in XDH form, but when released into the circulation it is converted into XO. Although XDH preferentially reduces NAD+, both forms of the enzyme can also reduce molecular oxygen to form the ROS superoxide and hydrogen peroxide. Wu et al. reported the *XDH* gene might be associated with constitutional susceptibility to hypertension [21]. XDH alters xanthine oxidase by reversible sulfhydryl oxidation or by irreversible proteolytic modification, and the production of urate results from the metabolism of purines by XDH [21,22].

We examined the association between sUA and hypertension risk in post-menopausal women, and investigated the association between the polymorphism of *XDH* gene and the risk of hypertension.

#### **2. Methods**

#### *2.1. Study Population*

This study was conducted with participants from a population-based cohort within the Korean Genome and Epidemiology Study on Atherosclerosis Risk of Rural Areas in the Korean General Population (KoGES-ARIRANG) to assess the genetic and environmental etiology of common metabolic common metabolic and cardiovascular diseases in South Koreans [23,24]. The KoGES-ARIRANG cohort study contained all adults aged 40–70 years that resided in rural areas of Wonju and Pyeongchang, Gangwon-do, Republic of Korea.

In this cross-sectional study, the baseline survey was performed from November 2005 to January 2008 and contained 28,338 adults aged 40 to 70 years. Among 17,517 women, 8666 women with a sUA level were included in this study. After excluding 736 with experience of hormone therapy, 625 with no history of menopause, and 11 missing for blood pressure, a total of 1415 premenopausal and 5879 postmenopausal women comprised in this cross-sectional study (Figure 1). Subjects who were treated antihypertensive drugs were 1991 participants. Those who experienced anti-hypertensive drugs were considered having hypertension. Additionally, we excluded the participants without genotype, 3666 participants with the genetic variations in *XDH* were eligible for this cross-sectional study. The study protocol was approved by the Institutional Review Board of Wonju Severance Christian Hospital.

**Figure 1.** Flow chart of study populations in the KoGES-CAVAS cohort. Abbreviations: KoGES-CAVAS, Korean Genome and Epidemiology Study on Atherosclerosis Risk of Rural Areas in the Korean General Population; GWAS, genome wide association study.

#### *2.2. Data Collection*

At baseline examination, study participants completed a standardized medical history and lifestyle questionnaire and underwent a comprehensive health examination according to standard procedures.

Body weight and height were measured, while participants were wearing light indoor clothing without shoes. Waist circumference (WC) was measured in a horizontal plane, midway between the inferior margin of the ribs and the superior border of the iliac crest using a tape measure (SECA-200, SECA, Hamburg, Germany).

Systolic (SBP) and diastolic blood pressures (DBP) were measured twice in right arms within 5 min using a standard mercury sphygmomanometer (Baumanometer, Copiague, NY, USA). The means of the two blood pressure readings were used for the data analyses. Hypertension was defined as a SBP of ≥140 mmHg or a DBP ≥90 mmHg and/or current treatment with antihypertensive medications at the baseline survey. All the participants were examined after fasting.

A venous blood sample was drawn from study participants after fasting for >12 h or overnight. Serum aliquots were stored at −80 ◦C until thawed for analysis. The sUA was measured by an enzymatic coloric method that can detect the absorbance differences using uricase and peroxidase as reaction enzymes. Fasting glucose was measured using the hexokinase method. The serum concentrations of high-density lipoprotein (HDL) cholesterol, total cholesterol (TC), and triglycerides (TG) were determined using the enzymatic calorimetric method.

Alcohol and smoking habits were estimated using self-questionnaires. Individuals who had smoked ≥100 cigarettes in their lifetime were defined as current smokers, and those who had not smoked for ≥3 months were defined as ex-smokers. An interview was performed to confirm the use of medications for hypertension, and the status of regular physical exercise.

At baseline examination, study participants completed a standardized medical history and lifestyle questionnaire and underwent a comprehensive health examination according to standard procedures. At first, we estimated menopausal status using self-questionnaire. Women who had preand postmenopausal status were defined as "Have you always had your periods at regular 28-day intervals? Or, have you had not your periods for three months recently?" Also, an interview was performed to confirm the use of medications for hormone therapy.

#### *2.3. Genome Wide Association Study Genotyping*

Samples were analyzed using an Affymetrix Genome-Wide Human single nucleotide polymorphism (SNP) array 6.0, which contains 906,600 genome-wide SNPs and 946,000 copy number variations. Briefly, the genomic DNA was digested with two restriction enzymes (NSP I and Sty I) and processed according to the Affymetrix protocol. Digested segments were ligated to enzyme specific adaptors incorporating a universal PCR priming sequence. PCR amplification was performed using universal primers in a reaction optimized for the amplification of fragments between 200–1100 base pairs. A fragmentation step was then used to reduce the PCR products to segments of approximately 25–50 bp, which were then end-labeled using biotinylated nucleotides. The labeled products were then hybridized to a chip, washed, and detected. Images were analyzed using GeneChip Operating System software (Affymetrix, Santa Clara, CA, USA). Internal quality control measures were to ensure data fidelity, that is, a QC call rate (Dynamic Model algorithm) always was over 86% and correct identification of subject gender based on heterozygosity on the X chromosome. Genotype calling was performed using the Birdseed v2 algorithm [25].

#### *2.4. Analysis of the XDH Genomic Polymorphism*

For this study, *XDH* fragments were independently amplified by polymerase chain reaction (PCR). PCR products were purified and then sequenced using a BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, Foster City, CA, USA) and an ABI 3730 × 1 automated sequencer (Applied Biosystems). SNPs identified in the *XDH* gene by whole gene sequencing were genotyped. Genomic DNA was extracted from 5 mL of peripheral venous blood using a commercially available isolation kit (QuickGene SP Kit DNA whole blood, Fujifilm, Tokyo, Japan). Genotyping was performed using the TaqMan fluorogenic 5' nuclease assay (Applied Biosystems) [25].

#### *2.5. Statistical Analysis*

We analyzed the study population divided in quartiles of sUA. Categorical variables were analyzed using the chi-square test and continuous variables were analyzed by ANOVA and post hoc using Scheffe's test in pre- and post-menopausal women. Interactions between drinking status, body mass index (BMI), WC, TG, and sUA on hypertension were investigated. In order to identify an association between sUA and hypertension we analyzed multivariate logistic regression which was used to evaluate the independence of associations between sUA and risk of hypertension according to menopausal status after adjusting for fasting glucose, BMI, and WC. Additionally, there might be collinearity between BMI and WC. Consequently, we demonstrated a part of odds ratio for sUA and hypertension without considering all confounding factors including age. We used SBP as confounding factors because of one of components defined participants with hypertension and without hypertension. In additional analysis, we adjusted for age to investigate an association of sUA and clinical variables, because postmenopausal women were older than premenopausal women.

Results were expressed as ORs ratios and 95% confidence intervals (CI). All analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC, USA), and SPSS version 23.0 (IBM Corp., Armonk, NY, USA). P value less than 0.05 was considered as statistically significant.

#### **3. Results**

#### *3.1. Baseline Characteristics*

The percentages of BMI, SBP, fasting glucose, and TG were significantly higher, and HDL cholesterol were significantly lower in the highest quartile of sUA (UA ≥ 5.0 mg/dL) (Table 1). BMI and WC were significantly higher in the participants with the highest quartile of sUA than those with the lowest quartile of sUA. (25.3 ± 3.4 vs. 23.6 ± 3.2; 85.2 ± 8.9 vs. 81.0 ± 9.0; respectively, *p* < 0.01) (Table 1). In the beginning of study design, we analyzed an association of sUA and hypertension with substantial confounding factors that related with sUA such as BMI and alcohol consumption. However, we did not confirm an association between sUA and hypertension after adjusted for all potential confounding factors.


**Table 1.** Sociodemographic characteristics of the study population.

Abbreviations: sUA, serum urate; WC, waist circumference; HC, hip circumference; BMI, body mass index; HDL-C, high-density lipoprotein cholesterol; lipoprotein cholesterol; GGT, gamma-glutamyl transferase; SBP, systolic blood pressure; DBP, diastolic blood pressure. \* post-hoc analysis was conducted using Scheffe's test, a: quartile 1, b: quartile 2, c: quartile 3, d: quartile 4.

#### *3.2. Anthropometric Characteristics of Premenopausal and Postmenopausal Women with Hypertension*

In postmenopausal women, the mean sUA values were higher in the participants in whom hypertension development was observed than in those who did not develop hypertension (4.72 ± 1.25 vs. 4.35 ± 1.03, *p* < 0.001). In the premenopausal women who developed hypertension, the baseline SBP and TC were 136.00 ± 16.44 mg/dL and 197.40 ± 33.90 mg/dL, respectively, and these parameters were higher than in those who did not develop hypertension (*p* < 0.001 vs. *p* = 0.002). There were no differences in smoking habits and the regular exercise status, between the pre- and postmenopausal women (Table 2).


**Table 2.** Anthropometric characteristics of premenopausal and postmenopausal women with hypertension.

Abbreviations: sUA, serum urate; BMI, body mass index; WC, waist circumference; HC, hip circumference; SBP, systolic blood pressure; DBP, diastolic blood pressure; TC, total cholesterol; TG, triglyceride; HDL, high-density lipoprotein; GGT, gamma-glutamyl transferase. \*, *p*-value was calculated by *t*-test.

#### *3.3. Univariate and Multivariate Logistic Regression Analyses*

In univariate logistic regression analysis, we investigated that postmenopausal women with a high sUA and high BMI were 3.13 times more likely to have hypertension than those with a normal BMI and a lower sUA (OR 3.13, 95% CI 2.67–3.66 vs. OR 1.85, 95% CI 1.63–2.11, respectively) (Table 3). A higher sUA level and TG were positively and significantly associated with the development of hypertension in postmenopausal women (OR 2.63, 95% CI 2.25–3.08) (Table 3). Pre-menopausal women with high sUA and high BMI showed 5.20 times more likely to have hypertension than those with normal BMI and a lower sUA (OR 5.20, 95% CI 3.48–7.78) (Table 3).

Multivariate logistic analysis after adjusting for age, SBP and BMI showed that a higher sUA and a high WC was found to be significantly associated with a 1.62-fold increased risk of hypertension in postmenopausal women (Table 3). After adjusting for age, SBP and BMI showed that a high sUA and a high TG were found to be significantly associated with a 2.08-fold increased risk of hypertension in postmenopausal women (OR 2.08, 95% CI 1.72–2.53) (Table 3).


**Table 3.** Effect of interactions between anthropometric parameters on hypertension in premenopausal and postmenopausal women.

Abbreviations: sUA, serum urate; BMI, body mass index. † Adjusted for age, systolic blood pressure and BMI. ‡ Adjusted for age, systolic blood pressure and waist circumference. Statistically significant results were presented by bold type.

#### *3.4. Hypertension Risk of Participants according to Genotypes of Xanthine Dehydrogenase (XDH)*

We found that participants with *XDH* rs206847 CC genotype had a risk of hypertension with statistical significance (OR=3.63). Multivariate logistic regression analysis showed a significantly lower risk of hypertension in women with rs206860AG genotypes than those with the wild type (AA) (OR = 0.26, 95% CI 0.08–0.89, *p* = 0.03) (Table 4). After adjusting for age, smoking status, alcohol consumption, and regular exercise, rs206826 AC genotype was associated with a decreased risk of hypertension (OR = 0.28 95 % CI 0.08–0.98). However, women with rs206847 CC and rs207425 GA genotypes had an elevated risk of hypertension.


**Table 4.** Association of *XDH* genetic variants and risk of hypertension.


**Table 4.** *Cont.*

Abbreviations: SNP, single nucleotide polymorphism; A, adenine, C, cytosine, G, guanine. Model 1 was adjusted for age, smoking status, alcohol consumption, regular exercise. Model 2 was adjusted for Model 1 and additionally adjusted for systolic blood pressure, total cholesterol and baseline body mass index. Statistically significant results were presented by bold type.

#### **4. Discussion**

In the present study, an elevated sUA was observed to be positively associated with an increased risk of hypertension and postmenopausal women with high sUA and BMI were 3.13 times more likely to have hypertension than those who had low sUA and BMI. Furthermore, *XDH* rs206860 AG genotype was found to be associated with the decreased risk of hypertension.

The present study showed that hyperuricemia was significantly associated with elevated risk of hypertension after adjusting for known confounders. Several potential mechanisms might explain the association between sUA and hypertension [5,26]. The first involves insulin resistance. Elevated insulin levels cause low urinary ammonium levels and predispose the precipitation of sUA [5,26]. Furthermore, insulin resistance is known to contribute to the developments of several metabolic disorders that influence the development of coronary artery disease [5,27], and to increase postmenopausal sUA concentrations. The second involves the detrimental effect of an elevated sUA concentration on renal function. Hyperuricemia leads to hypertension and renal injury via a crystal-independent mechanism by stimulating the renin–angiotensin system and inhibiting neuronal nitric oxide synthase [5,28]. Also, as sUA in rats induces sodium excretion to decrease by the epithelial sodium channel, it therefore contributes hypertension [29]. The third mechanism involves endothelial dysfunction. Experimental evidence suggests a potentially causal role for urate in the pathogenesis of hypertension and atherosclerosis [28]. It was reported that sUA concentrations are higher in postmenopausal than in premenopausal women [30]. As far as we know, in postmenopausal women who did not receive estrogen hormone therapy, sUA concentration tended to increase due to a shortage of the uricosuric effect of estrogen.

Although the association between sUA and hypertension has been already demonstrated in epidemiological and clinical studies, the nature of the interaction between sUA and hypertension remains debatable. The sUA might not be an independent risk factor of hypertension after controlling for other risk factors, but some studies have reported sUA is predictive of hypertension and renal disease development after controlling for associated risk factors. However, a recent meta-analysis including 25 studies of 97,824 participants has shown that high sUA significantly predicts systemic hypertension [16].

In fact, urate found to have several beneficial and potentially detrimental biologic effects [31]. Our results suggest hyperuricemia dose-dependently predicts higher risks of hypertension as proportions of subjects with hypertension increased significantly with sUA quartile, which is consistent with the findings of previous studies. Interestingly, we found genetic variations of the rs206860 polymorphism of *XDH* gene might lower the risk of hypertension. To the best of our knowledge, the association between polymorphism of rs206860 and hypertension was not widely reported. rs206860 was investigated in the studies of advanced liver disease [32] and anti-tuberculosis drug-induced hepatotoxicity [33]. As far as we know, other SNPs were not directly reported regarding association between heterogeneity of SNPs and hypertension.

Other SNPs of *XDH* gene associated with hypertension were also reported. Recently, Scheepers et al. reported that mean arterial pressure and DBP increased approximately 1 mmHg less in carriers of minor alleles of *XDH* rs2043013 in a European population [19]. Yang J et al. reported multivariate logistic regression analysis showed a significant association between the three SNPs of *XDH* at rs2043013 and hypertension in men: 47686C>T and 69901A>C in the recessive model, and 67873A>C (N1109T) in the dominant model [34]. Wu B et al. showed the *XDH* gene polymorphisms rs1042039, rs1054889, and rs2073316 might be associated with hypertension in the rural Han Chinese population [21].

The present study has some limitations that warrant considerations. First, subjects were not analyzed over a follow-up period, and therefore, we could not individually evaluate whether the association between sUA concentration and new-onset hypertension was relevant over a longer period. Second, the genome wide association study (GWAS) population included a smaller number of postmenopausal women with hypertension than our basic subjects for each parameter analysis. Third, our findings might be differently applied to other populations, especially younger age groups of different ethnicities. As far as we know, *XDH* rs206860 and its associations with hypertension have not been studied. rs206860 of *XDH* is unknown genetic variant so far and further studies in gene discovery and function needs to be verified. Also, cross-sectional nature of study has limited power to make conclusions about causality.

In order to investigate an association between genetic variants in *XDH* and risk of hypertension, we examined confounding factors in multivariate logistic regression analysis gradually. We identified the optimal model adjusted for age, smoking status, alcohol consumption, regular exercise, systolic blood pressure, total cholesterol, and baseline body mass index. However, we recognized that women with hypertension were few and have very wide confidence intervals because some women do not have the genotype of *XDH*. Also, it is not possible to identify whether high urate leads to hypertension or not from a cross-sectional study

However, this study showed a significant association between sUA and hypertension risk in postmenopausal women. sUA might be a useful marker to predict disease modality and progression of chronic metabolic diseases such as hypertension in clinical practice.

#### **5. Conclusions**

This study suggests that sUA concentrations might be associated with an increased risk of hypertension in postmenopausal women and that the rs206860 polymorphism of the *XDH* gene might be associated with a low risk of hypertension in Koreans. It seems that sUA may be a cost-effective, applicable parameter to evaluate hypertension risk of postmenopausal women. Further well-designed, large-scale studies in other populations are warranted to validate our results.

**Author Contributions:** Conceptualization, J.-H.L. and J.R.C.; methodology, J.-H.L., T.H.G. and J.R.C.; software, T.H.G. and S.J.; validation, J.-H.L., T.H.G. and J.R.C.; formal analysis, J.-H.L., T.H.G. and J.R.C.; investigation, J.-H.L. and J.R.C.; resources, J.-H.L. and J.R.C.; data curation, J.-H.L., T.H.G. and J.R.C.; writing—original draft preparation, J.-H.L.; writing—review and editing, J.-H.L. and J.R.C.; visualization, J.-H.L., T.H.G. and J.R.C.; supervision, S.-H.L., J.K., J.H.H., J.Y.K., D.R.K. and S.-B.K.; project administration, J.-H.L.; funding acquisition, J.R.C. and S.-B.K.

**Funding:** This study was supported by the Korea Centers for Disease Control and Prevention (2005-E71013-00, 2006-E71002-00, 2007-E71013-00, 2008-E71004-00, 2009-E71006-00, and 2010-E71003-00). This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2017R1D1A3B03034119). This study was provided with bioresources from National Biobank of Korea, the Centers for Disease Control and Prevention, Republic of Korea.

**Acknowledgments:** The authors would like to express their gratitude to members of the Institute of Genomic Cohort in Wonju Severance Christian Hospital for helpful discussions.

**Conflicts of Interest:** The authors have no conflict of interest to declare.

#### **Abbreviations**

sUA: serum urate; BMI, body mass index; WC, waist circumference; HC, hip circumference; HDL-C, high density lipoprotein cholesterol; GGT, gamma-glutamyl transferase; SBP, systolic blood pressure; DBP, diastolic blood pressure.

#### **References**


© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).
