Role of Nutrition and Adherence to the Mediterranean Diet in the Multidisciplinary Approach of Hidradenitis Suppurativa: Evaluation of Nutritional Status and Its Association with Severity of Disease

Hidradenitis suppurativa (HS) is a chronic, inflammatory and debilitating skin disorder. The exacerbating factors of HS include nutrition and adiposity. We aimed to investigate the relationships between body composition and the adherence to the Mediterranean diet (MD) with the severity of HS in a sample of naive-treatment patients with HS. In this case–controlled, cross-sectional study, we enrolled 41 HS patients and 41 control subjects. Body composition was evaluated by a bioelectrical impedance analysis (BIA) phase-sensitive system. PREvención con DIeta MEDiterránea (PREDIMED) and the 7-day food records were used to evaluate the degree of adherence to the MD and dietary pattern, respectively. The clinical severity was assessed by using the Sartorius HS score. HS patients had a worse body composition, in particular lower phase angle (PhA) (p < 0.001), and a lower adherence to the MD than controls, in spite of no differences in energy intake between the two groups. The receiver operator characteristic (ROC) analysis showing a value of PhA of ≤ 5.7 and a PREDIMED score of ≤ 5.0 identified HS patients with the highest clinical severity of the disease. After adjusting for sex, age, body mass index (BMI), and total energy intake, the HS Sartorius score maintained negative correlations with PhA (p < 0.001), PREDIMED score, and n-3 polyunsaturated fatty acids (p = 0.005). The results of the multivariate analysis showed PhA and PREDIMED score were the major determinants of HS Sartorius score, explaining 82.0% and 30.4% of its variability, respectively (p < 0.001). Novel associations were demonstrated between PhA and the degree of adherence to the MD with the HS severity. PhA and PREDIMED score might represent possible markers of severity of HS in a clinical setting.


Power Size Justification
The power size was calculated based on the following assumptions: 1. In the absence of clinical studies with similar experimental design currently available in the literature, the calculation of the sample size was performed a priori by considering the effect size 0.8 with type I error of 0.05 and a power of 90%. The number of subjects to be enrolled was found to be 34 per group, that we decided to round up to 41, with a total of 82 total subjects enrolled in the study in order to replace drop out patients. The calculation of the Sample Size was performed using G*Power Software (G*power software Version 3.1.9.2, Universität Düsseldorf, Düsseldorf, Germany).
2. The power sample was calculated by the differences of means ± standard deviation (SD) of PhA in each group (6.06). Considering a number of cases required in each group were set at 41 per group, a type I (alpha) error of 0.05 (95%), and a type II (beta) of 0.05 were used, the calculated power size was 95%. The calculation of power size was performed using Sample Size Calculator Clinical Calc (ClinCalc LLC, Chicago, IL, USA).

Lifestyle Habits
We defined current smokers were subjects smoking at least one cigarette per day, former smokers were subjects who had stopped smoking at least one year before the interview, and non-current smokers as the remaining participants. Participants habitually engaged in at least 30 min/day of aerobic exercise (YES/NO) were defined as physically active.

Anthropometric Measurements and Blood Pressure
The anthropometric measurements were performed by the same operator (a nutritionist experienced in providing nutritional assessment and body composition), according to the International Society for the Advancement of Kinanthropometry (ISAK 2006). BMI was calculated as the ratio between weight and square of height (kg/m 2 ). WC was measured according to Nishida et al. [32]. Further details are reported in Supplementary data.

Bioelectrical Impedance Analysis
Body composition was assessed using a BIA phase-sensitive system by experienced observers (an 800-µA current at a signal-frequency of 50 kHz BIA 101 RJL, Akern Bioresearch, Florence, Italy) [33], as previously reported [34][35][36][37]. The exam was performed as suggested by the European Society of Parental and Enteral Nutrition (ESPEN) [38]. Electrodes were placed on the hand and the ipsilateral foot, according to Kushner (1992) [39]. The PhA was obtained from conditions under 50 kHz according to the following formula: PhA ( • , degrees) = arctangent reactance (Xc)/ resistance (R) × (180/π). Further details are reported in the Supplementary data.

Dietary Assessment
As we have already fully reported in previous studies [42,43,45,[47][48][49], nutritional data was obtained by a face-to-face interview administered by a qualified nutritionist. In detail, a photographic food atlas (≈1000 photographs) of known portion sizes was used to quantify foods and drinks [50] and the 7-day food records were used to collect dietary data, including beverage intakes. On the basis of these records, the nutritionist calculated the total energy intake and the quantities of macronutrients.

Classification and Severity Assessment of HS
As a gold standard is still lacking, the disease severity assessment of HS was assessed using three scoring systems, including Sartorius HS score, Hurley Stages, and HS Physician's Global Assessment (PGA) [8]. The Sartorius HS score is a clinical classification system based on the counting of single fistulas and nodules within seven anatomical regions [51], and the measurement of the longest distance between two lesions of the same type within each anatomical region (axilla, gluteal, groin, genital, or other inflammatory sites left and/or right) [8]. The Hurley system describes three distinct clinical stages, in particular: "Stage I: Abscess formation, single or multiple, without sinus tracts and cicatrization; stage II: Recurrent abscesses with tract formation and cicatrization, single or multiple, widely separated lesions; and stage III: Diffuse or near-diffuse involvement, or multiple interconnected tracts and abscesses across the entire area". Finally, according to the HS-PGA HS severity was classified by counting "the number of abscesses, fistulas, and inflammatory and non-inflammatory nodules in all skin areas stages, on a scale from 1 to 6 (from stage 1: Clear, no inflammatory or non-inflammatory nodules to stage 6: Severe, >5 abscesses or draining fistulas)" [52,53]. The dermatologists who evaluated the clinical severity of HS were blinded to the design of the study to prevent rate biases.

Evaluation of the Oxidized Low-Density Lipoprotein Levels
Blood samples were collected via brachial puncture into 5-mL heparin vacuum tubes. The samples were centrifuged at 3000 rpm for 10 min at RT, and sera were separated and stored at −80 • C for further analysis (maximum, 6 months). The plasma levels of oxidized low-density lipoprotein (ox-LDL) were measured by using the LP-CHOLOX test carried out on an automated analyzer (Free Carpe Diem, Diacron International, Grosseto, Italy), using a commercial kit (Diacron International, Grosseto, Italy) according to the manufacturer's instructions. The LP-CHOLOX test evaluates a class of hydroperoxides derived from the lipid peroxidation, which are mainly represented by oxidized cholesterol. Peroxides are able to promote the oxidation of the ferrous iron (Fe 2+ ) to ferric iron (Fe 3+ ). The LP-CHOLOX test is based on the spectrophotometric measurement (at 505 nm) of the colored complex developed by the binding between the Fe 3+ and the thiocyanate. The absorbance values are directly proportional to the lipoperoxides concentrations, and the values are related to specific standard solution (400 µEq/L). Results are expressed in µEq/L, and reference values are: Normal, ≤599 µEq/L; slight alteration, from 600 to 799 µEq/L; moderate alteration, from 800 to 999 µEq/L; strong alteration, ≥1000 µEq/L [54,55].

Statistical Analysis
The data distribution was evaluated by a Kolmogorov-Smirnov test and the abnormal data were normalized by logarithm. Skewed variables were back-transformed for presentation in tables and figures. Results are expressed as mean ± SD. The chi square (χ 2 ) test was used to determine the significance of differences in frequency distribution of gender, smoking habit, physical activity, BMI categories, WC cut-offs, and dietary components included in the PREDIMED questionnaire. Differences between HS patients and the control group were analyzed by Student's paired t-test, while the differences among the several parameters with the clinical severity of HS according to the Hurley system and HS-PGA grades were analyzed by Student's unpaired t-test. The correlations between study variables were performed using Pearson r correlation coefficients. The association among quantitative variables (Hurley system and HS-PGA grades) with body composition, total energy and daily macronutrients/micronutrients intake were assessed with proportional Odds Ratio (OR) models, 95% Interval Confidence (IC), and R 2 . Receiver operator characteristic (ROC) curve analysis which were performed to establish sensitivity and specificity, area under the curve (AUC), and IC, as well as cut-off values for PhA and PREDIMED score in detecting Sartorius HS scores above the median values in the HS patients. Test AUC for ROC analysis was also calculated and we entered 0.957 for AUC ROC and 0.5 for the null hypothesis values. An α level of 0.05 (type 1 error) and a β level of 0.20 (type II error) were used as the cut-off values for statistical significance. Only variables that had a p-value < 0.05 in the univariate analysis (partial correlation) were entered. Variables with a variance inflation factor (VIF) >10 were not considered to avoid multicollinearity. Values ≤ 5% were considered statistically significant. Data were collected and analyzed using the MedCalc ® package (Version 12.3.0 1993-2012 MedCalc Software bvba-MedCalc Software, Mariakerke, Belgium).

Results
The study population consisted of 82 participants, 41 patients with HS and 41 healthy subjects as a control group. All case-patients completed the study protocol including nutritional assessment, PREDIMED questionnaire, and BIA measurements. The HS Sartorius score was 51.0 (33.0-81.0). According to the Harley grade and HS-PGA, 14, 24 and 3 patients presented with grade 1, 2 and 3, respectively. Considering the number of grade 3 patients, HS patients with grade 2-3 were included in the same group for the following analysis. These results are shown in the Tables S1-S5.
As reported in Table 1, no significant differences were evident in lifestyle habits and anthropometric measurements between HS patients and healthy subjects. (p = 0.003), PhA (p < 0.001), intra-cellular water (ICW) (%), and the highest values of extra-cellular water (ECW) (p < 0.001). Analyzing the details of the response frequency of dietary components included in the PREDIMED questionnaire, we found that HS patients consumed less frequently red meat and more frequently fish/seafood (Table 3). In HS patients, extra virgin olive oil was the most consumed food item, followed by poultry more than red meats. HS patients exhibited statistically significant differences in red meats and fish consumption, compared with healthy subjects, A p value in bold type denotes a significant difference (p < 0.05).
Data on Mediterranean food frequencies were further analyzed by using the 7-day food records. As shown in Table 4, in spite of no differences in energy intake between the two groups, HS patients consumed a lower quantity of complex carbohydrate, monounsaturated fatty acids (MUFA) and n-3  ROC analysis was performed to determine the cut off values of the PhA and the adherence to the MD that were predictive of highest Sartorius HS scores (above the median value 51) (Figure 2a,b). A value of PhA of ≤5.7 (p < 0.001, AUC 0.919, standard error 0.040, 95% CI 0.790 to 0.981) and a score of PREDIMED score of ≤5.0 (p < 0.001, AUC 0.762, standard error 0.077, 95% CI 0.603 to 0.881) could serve as a thresholds for a significantly increased risk of high Sartorius HS scores. ROC analysis was performed to determine the cut off values of the PhA and the adherence to the MD that were predictive of highest Sartorius HS scores (above the median value 51) (Figure 2a,b). A value of PhA of ≤5.7 (p < 0.001, AUC 0.919, standard error 0.040, 95% CI 0.790 to 0.981) and a score of PREDIMED score of ≤5.0 (p < 0.001, AUC 0.762, standard error 0.077, 95% CI 0.603 to 0.881) could serve as a thresholds for a significantly increased risk of high Sartorius HS scores.  (Figure 2a,b). A p value in bold type denotes a significant difference (p < 0.05). ROC, receiver operating characteristic; PhA, phase angle

Correlation Studies
The correlations between HS Sartorius score and body composition evaluated by BIA parameters are summarized in Table 5. Also, after adjusting for sex, age, BMI, and total energy intake, HS Sartorius score showed significant negative correlations with PhA (p < 0.001), but not with R and Xc, fat mass (FM) (p = 0.001), ICW (p < 0.001), and had positive associations with free fat mass (FFM) (%), and ECW (p = 0.004).   (Figure 2a,b). A p value in bold type denotes a significant difference (p < 0.05). ROC, receiver operating characteristic; PhA, phase angle.

Correlation Studies
The correlations between HS Sartorius score and body composition evaluated by BIA parameters are summarized in Table 5. Also, after adjusting for sex, age, BMI, and total energy intake, HS Sartorius score showed significant negative correlations with PhA (p < 0.001), but not with R and Xc, fat mass (FM) (p = 0.001), ICW (p < 0.001), and had positive associations with free fat mass (FFM) (%), and ECW (p = 0.004). In Figure 3 we show the differences in HS Sartorius scores across the PREDIMED categories. Higher values of HS Sartorius scores were evidenced in low adherers, compared with average-higher adherers (p < 0.001). Sartorius HS score was significantly associated with R, PhA, ECW and ICW. All parameters remain significantly associated after adjuster for sex, age, BMI and total energy intake, except for R. A p value in bold type denotes a significant difference (p < 0.05). R, resistance; PhA, phase angle; ECW, extracellular water; ICW, intracellular water.
In Figure 3 we show the differences in HS Sartorius scores across the PREDIMED categories. Higher values of HS Sartorius scores were evidenced in low adherers, compared with average-higher adherers (p < 0.001). Consequently, the highest HS Sartorius score was significantly associated with lowest score of PREDIMED score (r = -0.552, p < 0.001), and this correlation remained significant also after adjusting for sex, age, BMI and total energy intake (r = -0.454, p = 0.005), as showed in Figure 4. Consequently, the highest HS Sartorius score was significantly associated with lowest score of PREDIMED score (r = -0.552, p < 0.001), and this correlation remained significant also after adjusting for sex, age, BMI and total energy intake (r = -0.454, p = 0.005), as showed in Figure 4. . Correlation between HS Sartorius score and PREDIMED score, after adjusting for sex, age, BMI and total energy intake. The highest HS Sartorius score was significantly associated with the lowest score of PREDIMED score (r = -0.552, p < 0.001), and this correlation remained significant also after adjusting for sex, age, BMI and total energy intake (r = -0.454, p = 0.005). A p value in bold type denotes a significant difference (p < 0.05).
In Table 6 we show the correlations among HS Sartorius scores, total energy and daily macronutrients intake evaluated by using the 7-day food records. The HS Sartorius score showed highly positive correlations with total and simple carbohydrate and a negative correlation with n-3 PUFA; however, after adjusting for sex, age, and BMI only the negative correlation with n-3 PUFA remained significant. After adjusting for sex, age, BMI and total energy intake, ox-LDL levels were positively correlated with the HS Sartorius score (r = 0.436, p = 0.007), and negatively with PhA (r = −0.506, p = 0.001) and PREDIMED score (r = −0.701, p < 0.001). Similar correlations were also found by using the Harley grade severity of HS and the data are reported in the supplementary files.  Figure 4. Correlation between HS Sartorius score and PREDIMED score, after adjusting for sex, age, BMI and total energy intake. The highest HS Sartorius score was significantly associated with the lowest score of PREDIMED score (r = -0.552, p < 0.001), and this correlation remained significant also after adjusting for sex, age, BMI and total energy intake (r = -0.454, p = 0.005). A p value in bold type denotes a significant difference (p < 0.05).
In Table 6 we show the correlations among HS Sartorius scores, total energy and daily macronutrients intake evaluated by using the 7-day food records. The HS Sartorius score showed highly positive correlations with total and simple carbohydrate and a negative correlation with n-3 PUFA; however, after adjusting for sex, age, and BMI only the negative correlation with n-3 PUFA remained significant. After adjusting for sex, age, BMI and total energy intake, ox-LDL levels were positively correlated with the HS Sartorius score (r = 0.436, p = 0.007), and negatively with PhA (r = −0.506, p = 0.001) and PREDIMED score (r = −0.701, p < 0.001). Similar correlations were also found by using the Harley grade severity of HS and the data are reported in the supplementary files. Sartorius HS score was significantly associated with total/simple carbohydrate and n-3 PUFA. After adjuster for sex, age, BMI and total energy intake, only n-3 PUFA remain negatively associated with Sartorius HS score. A p value in bold type denotes a significant difference (p < 0.05).
To evaluate the relative prognostic value of the measures of body composition and nutritional status to predict the clinical severity of HS, we performed two multiple linear regression analysis models that included measures of the body composition parameters (model I), PREDIMED scores, n-3 PUFA and ox-LDL (model II). Using model I, PhA entered at the first step (p < 0.001) and appeared to be among the body composition measures exerting the most powerful influence on HS Sartorius scores, explaining 82.0% of HS Sartorius score variability, while the other variables (FM, FFM, ECW, ICW) were excluded. Using model II, PREDIMED scores were entered at the first step (p < 0.001) and seemed to be the most powerful factor influencing the HS Sartorius score, explaining 30.4% of HS Sartorius score variability, while the other variables (while n-3 PUFA and ox-LDL) were excluded. The results of the two models are shown in Table 7. Table 7. Multiple regression analysis models (stepwise method) with the Sartorius HS score as the dependent variable to estimate the predictive value of: BIA parameters (Model I), PREDIMED score, n-3 PUFA and ox-LDL (Model II).

Parameters
Multiple Regression Analysis Variable excluded: n-3 PUFA and ox-LDL Among some BIA parameters (model I), PREDIMED scores, n-3 PUFA and ox-LDL (model II), Sartorius HS scores were well predicted by PhA and PREDIMED scores, respectively. A p value in bold type denotes a significant difference (p < 0.05).

Discussion
The novel findings of this cross-sectional, case-controlled, observational study are the associations between the clinical severity of HS with PhA, a marker of cell membrane integrity, and with nutritional status, in particular the degree of adherence to the MD. Our data demonstrates that PhA was smaller and the adherence to the MD was lower in HS patients than in control subjects. Moreover, both the associations were independent of gender, age, BMI, and total energy intake. Based on ROC curve analysis, PhA ≤ 5.7 • and a PREDIMED score of ≤ 5.0 identified HS patients who presented with the highest clinical severity of the disease. Consistently, we found that HS patients presented significant differences in other BIA parameters indicative of an inflammatory/catabolic status and exhibited a pro-inflammatory dietary pattern, with a lower intake of complex carbohydrate, MUFA and n-3 PUFA, and a higher intake of SFA and n-6 PUFA than controls.
HS is a chronic, immune-mediated inflammatory skin disease with a multifactorial pathogenesis. HS develops through interplay of genetic, immunological, endocrine and environmental risk factors, including obesity [12], body composition [24,25], and diet [27]. In particular, both Romaní et al. and Miller et al. reported that subjects with HS and healthy controls showed differences in FM analyzed by BIA, irrespective of their BMI, with higher FM in HS patients than in healthy controls [24,25]. In this context, higher FM may indicate an altered status of inflammatory homeostasis in adipose tissue that can amplify the morbidity in HS patients. Nevertheless, as a BIA volumetric approach provides an indirect estimate of FM, FM could be both over-and under-estimated due to minimal variations of soft tissues hydration that might induce errors in the equations used in conventional BIA, for the prediction of body composition [56,57]. Contrariwise, the PhA, which is derived from raw BIA measurements such as R and Xc, is considered to be valid in conditions with altered hydration status, as in chronic skin inflammation, such as psoriasis, obesity, and chronic inflammation [35,58,59] and the advantages of using the raw BIA measurements, including PhA, have been also recently emphasized [60]. In both these studies, PhA was not included among the BIA parameters of body composition, and the nutritional assessment was not performed.
PhA, a parameter obtained from BIA direct measures, such as R and Xc, is widely used as a marker of cellular health [61]. Consequently, PhA is considered a reliable predictor of mortality and morbidity in several diseases [58]. In a healthy population, a number of different factors may affect PhA, including diet, sex, age, BMI, and inflammatory status [62]. PhA indicates the integrity [63] of a large number of cell membranes and the water distribution in body fluids [64]. Thus, PhA is positively associated with the body cell mass [65] and negatively associated with ECW/ICW ratio [66]. In line with other studies in different chronic inflammatory diseases [58], and our previous study in psoriatic patients [35], in this study we evidenced that a small PhA were correlated also with the clinical severity of HS, and we hypothesized the inflammatory milieu could account for this association in our group of HS patients. Of interest, among all BIA measurements, PhA was the main predictor of the clinical severity of HS, while R and Xc showed no correlations after adjusting for sex, age, BMI and total energy intake, likely due to the link of PhA with either the capacitive behavior of tissues (Xc) associated with cellularity, cell size, and integrity of the cell membrane, and its pure resistive behavior (R), mainly dependent on tissue hydration [19]. The identification of prognostic factors for HS patients could play a critical role in both the clinical management and the adequate monitoring of the clinical course of the disease.
As in psoriatic patients, we found that HS patients, compared to the control group, had also a lower adherence to the MD. In addition, by carefully evaluating their dietary assessment, HS patients consumed higher amounts of simple carbohydrates, total fat, food with a higher n-6/n-3 PUFAs ratio, and lower complex carbohydrates, MUFA, n-3 PUFA and fiber. The association of diet with the clinical severity of HS has been previously reported. In particular, Danby investigated to the possible role of dairy food on HS disease and it reported that 83% of the 47 patients that underwent a dairy-free diet clinically improved compared to the control group without diet restriction [28]. In addition, Cannistrà et al. reported that in 12 patients with HS, a controlled brewer's yeast-free diet for 12 months was effective in reducing the clinical severity of HS with a rapid stabilization of the dermatologic manifestation [29].
Very recently, we reported in healthy subjects, an association between the degree of the adherence to the PhA and MD, independently of sex, age, and BMI [34]. This association has been attributed to the high content of different beneficial compounds, such as antioxidants and polyphenols, found in plant foods, fruit and red wine, which are largely present in Mediterranean foods [67]. In particular, Mediterranean meal plans exhibit anti-inflammatory potential based on ox-LDL levels [68], a marker of oxidative stress, closely involved in the process of chronic inflammation. Consistently, in our sample of HS patients, ox-LDL levels presented a negative correlation the adherence to the MD and a positive correlation with the clinical severity of HS, thereby supporting the potential role of the MD in the integrated management of the HS patients.
The main limitations of the study were the following: First, the cross-sectional design of this study did not allow the determination of a causal association between MD and HS or to draw a final conclusion on the prognostic value of the degree of adherence to the MD in the prediction of clinical severity of HS. Further, there was not a normal weight control group matched by age and gender which did not allow a more consistent analysis of the associations detected in this study. Although body weight could have influenced the nutritional behavior in this study, the clinical severity of HS with PhA was associated with the degree of adherence to the MD after correction for BMI. Moreover, the cut-off value of the PhA and of PREDIMED scores suggested in our present study for predicting the clinical severity of HS, should be considered with caution until large studies are available to perform an appropriate cross-validation. In addition, although in this study we based the analysis on raw BIA measurements rather than BIA volumetric parameters, the evaluation of FFM and FM was not validated by gold standard reference methods, such as dual-X ray absorptiometry, and expert nutritionists are mandatory for execution and especially for the assessment of the nutritional status and the interpretation of BIA measurements, in particular PhA. Nevertheless, considering the low prevalence of HS [3][4][5][6][7], strengths of our study include the relatively large sample of HS patients and the diagnosis of HS that was clinically evaluated and not self-reported; moreover, this study has adequate statistical power, included only naive-treatment patients and both HS patients and matched controls have been well characterized. Finally, the nutritional status has been adequately assessed by using the 7-day food records, the gold standard among food frequency questionnaires [31].

Conclusions
In conclusion, we reported a novel association between the clinical severity of HS with PhA, a marker of cell membrane integrity, and with nutritional status, in particular the degree of adherence to the MD. Specific cut-off values for the PhA and the degree of adherence to the MD could be included as an auxiliary tool in the complex dermatological evaluation of the clinical severity in patients with HS, contributing to identify those patients who could get additional benefit from careful dietary interventions. Our study underlines the role of the nutrition assessment as a predictive tool in HS patients. A growing cooperation between nutritionists and dermatologist might provide a combination key in the complex management of the HS patients. Further studies on a large population with HS and intervention trials are warranted to support the association of PhA and the adherence to the MD with the clinical severity of HS, and to highlight the potential anti-inflammatory effects of the MD on in HS patients.

Supplementary Materials:
The following are available online at http://www.mdpi.com/2072-6643/11/1/57/s1, Supplementary data; Table S1: Differences in the Hurley system and HS-PGA grades in BIA parameters; Table S2: Bivariate proportional odds ratio model to assess the association between body composition evaluated by BIA parameters, the Hurley system and HS-PGA grades in HS patients; Table S3: Differences in the Hurley system and HS-PGA grades in the adherence to the MD and Ox-LDL in HS patients; Table S4: Differences of total energy and daily macronutrient/micronutrient intake of HS patients, divided according to the Hurley system and HS-PGA grade; Table S5: Bivariate proportional odds ratio model to assess the association between total energy and daily macronutrient/micronutrient intake of HS patients, divided according to the Hurley system and HS-PGA grade.