Moderate-to-severe atopic dermatitis patients show increases in serum C-reactive protein levels, correlating with skin disease activity

Background: Atopic dermatitis (AD), the most common chronic inflammatory skin disease, is evolving as a systemic disease, and associated systemic inflammation is possibly linked to increases in cardiovascular disease. Methods: We assessed levels of the inflammatory marker CRP in 59 patients with moderate-to-severe AD compared to matched healthy controls, and to determine correlation with skin disease severity. Clinical severity was measured using SCORing of Atopic Dermatitis (SCORAD) and body surface area (BSA). Control subjects (n=118), matched by age, gender, smoking status and ethnicity, were obtained from the National Health and Nutrition Survey (NHANES). Results: AD patients had significantly increased serum CRP levels compared to controls (0.7±1.0 vs. 0.4±0.7mg/dl; p=0.001), and 52.5% of them showed CRP levels >0.3mg/dl, predicting high cardiovascular risk. CRP levels were significantly correlated with both SCORAD (r=0.427, p=0.0008) and BSA (r=0.407, p=0.0015). IgE levels in AD were highly elevated (median 2903U/ml, IQR [234,10655]), but only weakly correlated with SCORAD (r=0.282, p=0.0427) and BSA (r=0.382, p=0.0052), but not with CRP levels. AD patients also showed increased LDH levels, but without significant correlations with disease severity (SCORAD, BSA) or CRP. Conclusions: Our study strongly supports CRP as a marker for disease severity in moderate-to-severe AD patients, further demonstrating its chronic systemic nature.


Amendments from Version 1 Introduction
Atopic dermatitis (AD), the most common chronic inflammatory skin disease, frequently starts during infancy, and in adults it has usually been present for several decades 1 . Similar to moderate-to-severe psoriasis, there is now evolving evidence that AD also has a systemic component beyond the classic atopic/ allergic comorbidities, with increases in cardiovascular risk factors such as obesity 2-4 , and associations with cardiovascular diseases in population-based studies 2,5 . A comparison of AD and psoriasis patients with healthy individuals, using cardiac computed tomography angiography, showed higher rates of coronary artery disease in both psoriasis and AD, compared to controls 6 . Systemic immune activation in adult moderate-to-severe AD patients is reflected by highly activated circulatory T-cells as measured using T-cell activation markers (ICOS and HLA-DR), at even higher frequencies than in psoriasis 7 . Also, several inflammatory blood biomarkers (e.g. Thymus and Activation Regulated Chemokine /TARC or CCL17) were consistently shown to correlate with AD clinical severity 8 . The important contribution of chronic inflammation to the development of atherosclerosis and cardiovascular disease events is now well established 9 . Therefore, C-reactive protein/CRP, an acute phase reactant reflecting systemic inflammation, has been suggested as potential biomarker for cardiovascular disease 9 . In patients with a history of myocardial infarction, the anti-inflammatory monoclonal antibody canakinumab (IL-1β blocker) led to a significant decrease in cardiovascular events 10 . Patients also showed reductions in serum CRP levels, without changes in their lipid profile 10 , demonstrating that antiinflammatory treatment can indeed have an impact on cardiovascular disease. In psoriasis, it has been demonstrated that CRP is significantly elevated and associated with disease severity 11 . One recent study suggests that CRP levels are also increased in adult chronic AD patients vs. matched controls 12 , but it remains to be determined whether CRP could serve as a marker for disease severity. In contrast to adults, studies in children and adolescents with active AD did not show increases in overall CRP levels compared to controls 13 , and elevated CRP levels early in life were claimed to have a protective role against the development of AD 14 and allergic sensitization 15 , suggesting that chronic low-grade inflammation in infants might provide some protection from allergen sensitization. In order to better clarify the potential role of CRP as disease biomarker, we sought to investigate CRP serum levels in moderateto-severe adult AD patients in relation to skin disease severity.

Matching
Matched control subjects were obtained with a ratio of 2-to-1 (n=118) from the National Health and Nutrition Survey/NHANES (https://www.cdc.gov/nchs/nhanes). They were matched to AD patients for age, gender, smoking status and ethnicity, using the R procedure MatchIt, method 'nearest', with a ratio of 2 control subjects for each case subject. We used individuals from the SPRINT survey nationwide between the years 2005 and 2010, for which CRP laboratory data were available. There were no changes (from the previous 2 years of NHANES) to equipment, laboratory methods or lab site.
CRP serum level measurement Serum CRP levels in AD patients were assessed using an immunoturbidimetric test (Abbott Laboratories, Lake Bluff, Illinois). For NHANES, CRP levels were assessed using a Siemens/Behring Nephelometer (Siemens HealthcareDiagnostics, Deerfield, IL), as described at https://wwwn.cdc.gov/Nchs/ Nhanes/2009-2010/CRP_F.htm. Both assays had a lower limit of detection of 0.02mg/dl. While different assays were used to measure CRP levels in patients and controls, both methods have the same lower level of detection (0.02mg/dL) and were shown to be comparable 19 .

Statistical analysis
For comparisons between AD and the control group, we used the two sample t-test for age; Fisher exact test for gender, ethnicity and smoking status; and the two sample Wilcoxon test for biomarkers. When variables were missing for some of the individuals, comparison was performed only for the individuals for which the variable was available.
Pearson correlation coefficients were used to calculate the association between the logarithm of the biomarkers (CRP, LDH, total serum IgE) and disease activity measures SCORAD and BSA. We used a univariate linear regression formula to draw the regression line for these correlations. Each correlation was performed only for the individuals for which relevant biomarker data was available. All analyses were performed using R statistical software (Version 3.3).

Results
There were no significant differences between demographic data of AD patients and controls (age, gender, ethnicity), blood lipids (triglycerides, LDL, HDL), body mass index (BMI), or smoking status (Table 1).
AD patients had significantly increased serum CRP levels (0.7±1.0mg/dl) when compared to controls (0.4±0.7mg/dl; p=0.001; Table 1 and Figure 1a). CRP levels in AD ranged from undetectable in one patient (<0.02mg/dl) to a maximum value of 6.2mg/dl in a patient with very severe AD and a SCORAD of 95 (Dataset 1 17 ). 23 out of 59 patients (39%) showed CRP levels outside the reference range of 0-0.5mg/dl. Furthermore, CRP levels were significantly correlated with both SCORAD (Figure 1b) and BSA ( Figure 1c). As 14 patients reported a history of asthma, a disease that has been shown to be associated with increased CRP blood levels 20 , we performed a sensitivity analysis to assess the non-asthma AD patients (Supplementary Table 1). However, differences between CRP levels in AD patients and controls remained highly significant after exclusion of all the patients with a history of asthma (Figure 2, Supplementary Table 1).
Consistent with previous publications 18 , the AD patients also showed increased LDH levels, but without significant correlations with disease severity measures (SCORAD, BSA) or CRP (Figure 3a-c). While IgE levels in AD were highly elevated

Discussion
This study is the first to demonstrate a correlation of AD disease severity with CRP levels in moderate-to-severe adult AD patients with decades of chronic disease activity, independent of co-existence of asthma. This finding is in line with the evolving concept that chronic AD has a considerable systemic inflammatory component 12 that is directly linked with the overall inflammatory burden in the skin. This increase in systemic inflammation might not only be a biomarker for skin disease severity, but one might speculate that it could also contribute to AD comorbid conditions, such as the evolvement of cardiovascular disease 2 . This concept is strongly supported by the fact that canakinumab led to a significant reduction in serum CRP levels and cardiovascular events in a recent clinical trial 10 . Interestingly, a case series using the IL-6R blocker tocilizumab was efficacious in AD, and decreased CRP levels, but was not followed further due to bacterial superinfection 21 . According to the joint guidelines of the Centers for Disease Control and Prevention and the American Heart Association on CRP levels and cardiovascular risk 11 , 20 of our AD patients (33.9%) showed CRP levels in the range of ≥0.1mg/dl and ≤0.3mg/dl, predicting intermediate risk, and 31 patients (52.5%) showed CRP levels >0.3mg/dl, which is within the high risk range. While CRP is predominantly produced by hepatocytes, it has also been detected in tape stripping experiments from AD skin, and its expression responded to emollients 22 .
Future large and prospective studies in chronic severe AD patients should determine whether the up-regulated CRP levels observed in our AD cohort are indeed linked to increased cardiovascular risk, beyond its role as a marker of systemic inflammation. Nevertheless, there is some circumstantial evidence that even these small increases might be clinically relevant, as CRP above 0.42mg/dL showed differences in statin treatment outcomes for cardiovascular events in a clinical trial 23 , and CRP levels in the canakinumab trial were in the same order of magnitude 10 .
Future clinical trials investigating new therapeutic agents might follow changes in CRP levels during treatment as a potential serum biomarker of disease severity and systemic inflammation, and these may clarify whether correcting CRP can serve as a surrogate for decreasing cardiovascular risk in AD patients. However, increases in CRP levels can be a result of various conditions such as infections and malignancies, which needs to be taken into account.
Our study harbors a few limitations. Besides being a retrospective study, healthy controls were not available at our site and were based on published historic controls matched for age, gender, and ethnicity. Also, it focused on a moderate-to-severe AD patient population (all but two patients had moderate-to-severe AD, i.e. a SCORAD >25 16 ) in a large tertiary academic center in New York, while controls were obtained across the United States, which might introduce some bias. To ensure that our results are applicable to the general AD population across ethnicities, larger international studies across different ethnic backgrounds that will also evaluate for existence of "silent" cardiovascular disease in chronic AD patients are needed. However, our data supports the role that persistent skin disease has in the systemic burden of inflammation in AD patients, mandating further investigation.

Ethical statement
This study has been approved by the IRB of the  Of interest, in this C-reactive protein review of 59 patients in Atopic Dermatitis Reference #11 (B. Strober et al ) mentions the increased levels of CRP in psoriasis. It is important to recognize now that psoriasis has been demonstrated to be a systemic immune-mediated disease that CRP levels in psoriasis patients with moderate to severe disease are significantly lower than in psoriasis patients who develop Psoriatic Arthritis and also are lower than other immune-mediated systemic disease, e.g. Crohn's or Rheumatoid Arthritis.
We do believe it should be emphasized in the title and abstract that this CRP review was carved out in adults only. In addition, it should be clarified in the method section that data was obtained retrospectively.
We all recognize that CRP is an acute phase reactant that can increase with infections/autoimmune/cancer as well as cardiovascular disease. Thus, reference should be made in this article to comorbidities in this group of 59 Atopic Dermatitis patients which could have possibly played a role in the increases of CRP.

If applicable, is the statistical analysis and its interpretation appropriate? Yes
Are all the source data underlying the results available to ensure full reproducibility? Partly

Are the conclusions drawn adequately supported by the results? Yes
No competing interests were disclosed.
Was there a reason for choosing SCORAD over objective SCORAD or EASI, which rely only on objective measures? Of interest, was CRP in adult AD vs non-AD using the NHANES data itself done (cf Silverberg Pediatr Allergy Immunol 2015)?
2. CRP, a marker of cardiovascular risk, was found to be elevated in AD, and correlated with AD severity. sIgE correlated with AD severity, but not with CRP. LDH did not show significant correlation with AD severity or CRP.
In addition to the Silverberg pediatric paper, Park et al presented a pediatric inpatient study of 67 children examining the relationship between eczema severity (mild-moderate vs severe) and a number of laboratory markers. They did not find correlation with CRP, LDH or skin cultures. That paper was pediatric, written in Korean and also has limitations, but may be worth referencing for comparison.
3. Some potential confounders were considered. Sensitivity analysis was performed for non-asthma AD patients; this subgroup still showed correlation between CRP and AD severity. The patients did not have clinical infection. Were any swabs done? Some patients in the dataset had other co-morbidities that could have contributed to the CRP. The nature of the control group may not allow ready comparison of these, but a comment about possible or unlikely confounding from co-morbidities as the case may be worth considering. In the correlations, the patients with high CRP Levels are omitted for some (graphical?) reason. Please state why and whether the Pearson r is calculated with or without them. I don't think this changes the conclusion of the paper but IMHO should be shown. If you have access to the raw SCORAD data, you might be able to check whether subcomponents of the SCORAD have a closer connection to the Serum CRP than others: -eczema involvement of some body regions -crusting, oozing -excoriations (scratch marks) I think it may be worth adding to the abstract that >50% of the moderate-to-severe AD patients were in the range of cardiovascular high-risk CRP levels. Also, you should probably discuss all ways to lower the high CRP. The best of them may be anti-IL6R, which also works in atopic dermatitis according to a case series.

If applicable, is the statistical analysis and its interpretation appropriate? Partly
Are all the source data underlying the results available to ensure full reproducibility? No Are the conclusions drawn adequately supported by the results?

Yes Yes
No competing interests were disclosed.

Competing Interests:
Referee Expertise: Inflammatory skin disease I have read this submission. I believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Author Response 17 Oct 2017 , Icahn School of Medicine at Mount Sinai, USA Emma Guttman We thank the reviewer for his positive and encouraging comments. We have changed the title accordingly.
All patients have been included in the graphs. We have now corrected the axis labeling for a more clear display of CRP levels in the correlation graphs Figures 1b and 1c, and Figures 3c and 3d.
Due to the retrospective nature of the study, we do not have access to the sub-components of SCORAD. Therefore, we cannot calculate these correlations.
We have now modified the abstract and the discussion section accordingly.
No competing interests were disclosed.

Competing Interests:
The benefits of publishing with F1000Research: Your article is published within days, with no editorial bias You can publish traditional articles, null/negative results, case reports, data notes and more The peer review process is transparent and collaborative Your article is indexed in PubMed after passing peer review Dedicated customer support at every stage For pre-submission enquiries, contact research@f1000.com