BA Kozel et al. Skin Biomechanics in CL

Borici-Mazi R, Kouridakis S, Kontou-Fili K (1999) Janiszewski J, Bienenstock J, Blennerhassett MG Staubach P, Eckhardt-Henn A, Dechene M et al. Cutaneous responses to substance P and (1994) Picomolar doses of substance P trigger (2006) Quality of life in patients with chronic calcitonin gene-related peptide in chronic electrical responses in mast cells without urticaria is differentially impaired and deter- urticaria: the effect of cetirizine and dimethin- degranulation. Am J Physiol 267:C138–45 mined by psychiatric comorbidity. Br J Der- dene. Allergy 54:46–56 Katsuno M, Aihara M, Kojima M et al. (2003) matol 154:294–8 Forsythe P, Bienenstock J (2012) The mast cell- Neuropeptides concentrations in the skin of a Tedeschi A, Lorini M, Asero R (2005) No evidence nerve functional unit: a key component of murine (NC/Nga mice) model of atopic der- of increased serum substance P levels in physiologic and pathophysiologic responses. matitis. J Dermatol Sci 33:55–65 chronic urticaria patients with and without Chem Immunol Allergy 98:196–221 Kulka M, Sheen CH, Tancowny BP et al. (2008) demonstrable circulating vasoactive factors. Clin Exp Dermatol 30:171–5 Grattan CE, Wallington TB, Warin RP et al. (1986) Neuropeptides activate human mast cell A serological mediator in chronic idiopathic degranulation and chemokine production. Wedi B, Wagner S, Werfel T et al. (1998) Pre- urticaria–a clinical, immunological and Immunology 123:398–410 valence of Helicobacter pylori-associated gas- histological evaluation. Br J Dermatol 114: Magerl M, Pisarevskaja D, Scheufele R et al. (2010) tritis in chronic urticaria. Int Arch Allergy 583–90 Effects of a pseudoallergen-free diet on Immunol 116:288–94 Gruber BL, Baeza ML, Marchese MJ et al. (1988) chronic spontaneous urticaria: a prospective Weller K, Groffik A, Church MK et al. (2014) Prevalence and functional role of anti-IgE trial. Allergy 65:78–83 Development and validation of the urticaria autoantibodies in urticarial syndromes. J Metz M, Krull C, Maurer M (2013) Histamine, TNF, control test: a patient-reported outcome Invest Dermatol 90:213–7 C5a, IL-6, -9, -18, -31, -33, TSLP, neopterin, instrument for assessing urticaria control. J Allergy Clin Immunol 133:1365–72 Hatada Y, Kashiwakura J, Hayama K et al. (2013) and VEGF are not elevated in chronic sponta- Significantly high levels of anti-dsDNA immu- neous urticaria. JDermatolSci70:222–5 Zuberbier T, Aberer W, Asero R et al. (2014) The 2 noglobulin E in sera and the ability of dsDNA Mosher RA, Coetzee JF, Allen PS et al. (2014) EAACI/GA LEN/EDF/AAAAI/WAO Guideline to induce the degranulation of basophils from Effects of sample handling methods on sub- for the definition, classification, diagnosis and chronic urticaria patients. Int Arch Allergy stance P concentrations and immunoreactivity management of Urticaria The 2013 revision Immunol 161(Suppl 2):154–8 in bovine blood samples. Am J Vet Res 75: and update. Allergy 69:868–87 Hide M, Francis DM, Grattan CE et al. (1993) 109–16 Zuberbier T, Pfrommer C, Specht K et al. (2002) Autoantibodies against the high-affinity IgE Rosenkranz MA (2007) Substance P at the nexus Aromatic components of food as novel receptor as a cause of histamine release of mind and body in chronic inflammation eliciting factors of pseudoallergic reactions in chronic urticaria. N Engl J Med 328: and affective disorders. Psychol Bull 133: in chronic urticaria. J Allergy Clin Immunol 1599–604 1007–37 109:343–8

Biomechanical Properties of the Skin in Cutis Laxa

Journal of Investigative Dermatology (2014) 134, 2836–2838; doi:10.1038/jid.2014.224; published online 26 June 2014

TO THE EDITOR elastic fibers to the mechanical proper- ded with informed, written consent. The Cutis laxa (CL) is a heterogeneous group ties of the skin, and for an improved, same control cohort was used in a of disorders characterized by loose, objective diagnosis of CL. concurrent study on the mechanical redundant, inelastic, or prematurely In the present study, 118 participating properties of the skin in Williams– wrinkled skin (Berk et al., 2012; Uitto controlsand17CLpatientswereinclu- Beuren syndrome (Kozel et al., 2014). et al., 2013). Several inherited forms of CL have been identified (Urban and Davis, 2014), with nine causative genes Table 1. Demography and elasticity parameters in the participants knowntodate(ALDH18A1, ATP6V0A2, Controls (n ¼ 118) Patients (n ¼ 17) P-value ATP7A, EFEMP2/FBLN4, ELN, FBLN5, LTBP4, PYCR1, RIN2). A shared feature Age 33.22±1.58 29.21±5.47 0.490a of all types of inherited CL is a reduced Gender (male %) 37.3% 35.3% 0.874b or abnormal deposition of elastic fibers E (MPa) 11.61±0.15 7.85±0.60 o0.0001a in the skin and other tissues (Berk Retraction time (ms) 622.82±21.20 1,152.82±211.21 0.024a et al., 2012). The consequences of VE (MPa) 5.35±0.14 2.51±0.32 0.0001a this disorder for the biomechanics of o the skin has received little attention Abbreviations: E, elastic modulus; VE, viscoelastic modulus. Continuous variables: mean±standard error of mean. (Grahame and Beighton, 1971). Such aIndependent t-test. studies are essential for a fundamental bChi-square test. understanding of the contribution of

Abbreviations: ADCL, autosomal dominant cutis laxa; CL, cutis laxa; E, elastic modulus; ROC, receiver operating characteristic; RT, retraction time; VE, viscoelastic modulus Accepted article preview online 20 May 2014; published online 26 June 2014

2836 Journal of Investigative Dermatology (2014), Volume 134 BA Kozel et al. Skin Biomechanics in CL

a 1.0 (Supplementary Table S2 online): three individuals with congenital and six sub- 0.8 jects with late-onset CL. As skin elasti- city measurements did not show 0.6 consistent differences between CL sub- groups, we pooled all types of CL into 0.4 VE (MPa) AUC: 0.908 one case group. Sensitivity E (MPa) AUC: 0.904 Cases and controls underwent testing RT (ms) AUC: 0.679 0.2 Reference line using a DermaLab skin elasticity module, a suction cup device extensively vali- 0.0 dated in previous studies (Pedersen et al., 0.0 0.2 0.4 0.6 0.8 1.0 2003; Grove et al., 2006; Anthonissen 1–Specificity et al., 2013; Gandanke et al., 2014). The b 1.0 device applies vacuum to a patch of skin and measures the pressure difference 0.8 (DP)requiredtoraisetheskintoa height of 1.5 mm and the time required 0.6 for the skin to return to the original position (retraction time, RT). The 0.4 elastic modulus (E) was calculated from Sensitivity Model 1: Age + VE + E; AUC:0.992 this pressure difference, assuming uni- 0.2 Model 2: Age + VE; AUC:0.958 form skin thickness (1 mm). In addition, a Reference line viscoelastic modulus (VE) was computed 0.0 using E and RT as variables (Supple- 0.0 0.2 0.4 0.6 0.8 1.0 1–Specificity mentary Materials and Methods online). Cases had significantly lower E, c ELN higher RT, and lower VE (all Po0.05) 10 than controls (Supplementary Figure S1 online, Table 1). In controls, RT and VE 8 were significantly correlated with age, but E was not (Supplementary Table S3 6 online). In cases, VE was marginally 4 correlated with age (Supplementary Table S3 online). Multivariate logistic 2 regression analysis revealed that age

Viscoelastic modulus (MPa) (P ¼ 0.005, OR: 1.21, 95% CI: 1.06– 0 1.37), VE (P ¼ 0.002, OR: 16.39, 95% 0204060 CI: 2.85–95.18), and E (P ¼ 0.043, OR: Age (years) 3.55, 95% CI: 1.04–12.15) were signifi- ATP6V0A2 LTBP4 cant predictors of disease status. VE was

Figure 1. Receiver operating characteristic (ROC) analysis of biomechanical variables and viscoelastic the strongest predictor and one unit modulus (VE) in relation to the causative gene mutation. (a) VE is more effective than elastic modulus (E) reduction in VE increased the odds of or retraction time (RT) in differentiating cases from controls as indicated by ROC curves. (b) Composite CL 16.39-fold. variables under Model 1 (Age þ VE þ E) perform better than Model 2 (Age þ VE). (c) VE values of We analyzed the receiver operating individuals with known gene mutations are identified by red tie-lines. Controls are represented by magenta characteristic (ROC) to evaluate the dots and CL cases by green dots. Linear regression lines are shown in each group (cases: green, utility of E, RT and VE as predictors. controls: magenta). Note that one individual with ATP6V0A2-related CL had similar age and VE data Of the three, VE performed best, with to another participant with LTBP4-related CL, resulting in overlapping data points. AUC, area under curve. ROC area under curve (AUC) reaching 0.908 when applied to the entire study population (Figure 1a). Restricting The IRB committees at the Washington medical history. Eight individuals of the analysis to individuals younger University School of Medicine and the the CL group were positive for mutations than 47 years (bottom 75th percentile University of Pittsburgh approved the in known CL genes, with LTBP4 muta- of the control group) yielded some studies. Age and sex were not signifi- tions in two, ELN mutations in three, improvement in the AUC ( ¼ 0.964, cantly different between cases and and ATP6V0A2 mutations in three Supplementary Figure S2 online). controls (Table 1). The CL group com- patients (Supplementary Table S1 We next evaluated if additional vari- prised individuals diagnosed with CL online). In addition, we included nine ables improved on VE only as diagnostic based on physical examination and patients with unknown mutational status measures in the entire study population.

www.jidonline.org 2837 BA Kozel et al. Skin Biomechanics in CL

Two models were considered. Model 1 increase of the RT of the skin of CL Robert C. Wilson4, Frank C. Sciurba4 incorporated VE, age, and E,assug- patients irrespective of the etiology of and Zsolt Urban2 gested by the logistic regression model the disease. Individuals with acquired or 1Department of Pediatrics, Washington described above. Model 2 included VE late-onset CL had similar reductions in University School of Medicine, St Louis, 2 and age only. Model 1 (AUC ¼ 0.992) VE compared to controls as individuals Missouri, USA; Department of Human performed significantly (P ¼ 0.0026, with ELN, LTBP4 or ATP6V0A2 muta- Genetics, University of Pittsburgh Graduate School of Public Health, Pittsburgh, ANOVA) better than Model 2 tions, suggesting that the disruption of Pennsylvania, USA; 3Division of Medical (AUC ¼ 0.958) in distinguishing cases elastic fibers leads to similar biomecha- Genetics, Children’s Hospital of Pittsburgh, from controls (Figure 1b). nical alterations independent of the Pittsburgh, Pennsylvania, USA and 4Division of The performance of a model on the precise molecular disease mechanisms. Pulmonary and Critical Care Medicine, same data that was used to fit that VE showed significant inverse correla- Department of Medicine, University of Pittsburgh School of Medicine, Pittsburgh, model can give an overly optimistic tion with age in both CL and control Pennsylvania, USA measure of performance. To test if such individuals, but in controls the decline E-mail: [email protected] overfitting had occurred, cross-valida- started from a higher level and was thus 5These authors share first authorship. tion tests were performed. As the cross- steeper. Therefore, VE appears to be a validation, on average, resulted in only good measure of biomechanical aging a modest decrease in AUC values of the skin, and our observations suggest SUPPLEMENTARY MATERIAL (Supplementary Table S4 online), over- that CL results in similar changes in skin Supplementary material is linked to the online fitting was not a major issue in this mechanics to aging. VE also offers the version of the paper at http://www.nature.com/jid analysis, and similar sensitivity and best specificity and sensitivity in distin- specificity values can be expected in guishing cases from controls among the REFERENCES future replication studies to our present individual variables measured in our Anthonissen M, Daly D, Fieuws S et al. (2013) results. study. Measurement of elasticity and transepidermal water loss rate of burn scars with the Derma- Adiagnosticvariable(D)canbe To date, only one study investigated lab((R)). Burns 39:420–8 calculated from the ROC analysis using the mechanics of the skin in CL Berk DR, Bentley DD, Bayliss SJ et al. (2012) Cutis the following device-specific formula: (Grahame and Beighton, 1971). As this laxa: a review. J Am Acad Dermatol 66:e1–17 D ¼27.570 þ 0.187 Age þ 2.795 early report had few cases and controls Gandanke TU, Duke JM, Danielsen PL et al. (2014) VE þ 1.267 E, where Age is measured (6 each), and only measured E but not Reliability of scar assessments performed with in years, and E and VE are measured in retraction, it did not find significant an integrated skin testing device - the Derma- Lab Combo. Burns S0305-4179:00058–8 MPa units. If an individual satisfies the difference between cases and controls. Grahame R, Beighton P (1971) The physical inequality Do2.538, the probability In contrast, the present report demon- properties of skin in cutis laxa. Br J Dermatol of the individual having CL is 99.2%. strates the utility of biomechanical 84:326–9 This cutoff value of D distinguishes measurements in CL. Grove GL, Damia J, Grove MJ et al. (2006) Suction cases from controls with 100% specifi- chamber method for the measurement of skin city and 91.5% sensitivity, supporting mechanics: the DermaLab. In: Serup J, Jemec GBE, Grove GL eds Handbook of Non-Inva- the use of the DermaLab device for the CONFLICT OF INTEREST The authors state no conflict of interest. sive Methods and the Skin. 2nd edn. CRC objective, specific, and sensitive diag- Press: Boca Raton, FL, 593–9 nosis of CL in future studies. Kozel BA, Bayliss SJ, Berk DR et al. (2014) Skin ACKNOWLEDGMENTS findings in Williams syndrome. Am J Med The skin was visually loose in indivi- We thank the subjects of this study for their duals with CL and decreased VE values Genet A; e-pub ahead of print 11 June participation, and Dr Daniel E. Weeks for statistical 2014 (Supplementary Figure S3 online). advice. This study was funded in part by NIH grants Pedersen L, Hansen B, Jemec GB (2003) Mechan- Although there was insufficient number HL090648 (ZU) and UL1TR000005 (FCS, ZU). Funding was provided to Dr Kozel by the ical properties of the skin: a comparison of individuals within each type of inher- Children’s Discovery Institute of Washington Uni- between two suction cup methods. Skin Res ited CL to allow for subgroup analysis, versity and St Louis Children’s Hospital. In addition, Technol 9:111–5 individuals with autosomal dominant Dr Kozel received funding for the study through Szabo Z, Crepeau MW, Mitchell AL et al. (2006) CL (ADCL) caused by ELN mutations her appointment as a scholar of the Child Aortic aneurysmal disease and cutis laxa Health Research Center in Developmental Biology caused by defects in the elastin gene. J Med showed the greatest degree of variation (NIH K12-HD01487) and the Genetic Basis of Genet 43:255–8 in VE (Figure 1b) compared to indivi- Inflammatory Airway Disease (NIH K12-HL089968). Uitto J, Li Q, Urban Z (2013) The complexity of duals with other mutations, consistent elastic fibre biogenesis in the skin - a per- with previous reports of variable expres- Beth A. Kozel1,5, Chi-Ting Su2,5, spective to the clinical heterogeneity of cutis sion of the skin phenotype in ADCL Joshua R. Danback1,RyanL.Minster2, laxa. Exp Dermatol 22:88–92 patients (Szabo et al., 2006). Suneeta Madan-Khetarpal3, Urban Z, Davis EC (2014) Cutis laxa: intersection 3 of elastic fiber biogenesis, TGFb signaling, the Our studies show significant reduc- Juliann S. McConnell , secretory pathway and metabolism. Matrix 2 2 tion of E and VE, and significant Meghan K. Mac Neal ,KaraL.Levine, Biol 33:16–22

2838 Journal of Investigative Dermatology (2014), Volume 134