The Effect of Smoking and Age on the Response to First-Line Therapy of Hidradenitis Suppurativa: An Institutional Retrospective Cohort Study

Abstract

Background

Treatment for HS (hidradenitis suppurativa) is often empiric and inadequate, and determining which patients will respond is difficult.

Objective

To determine which patient factors are associated with a positive response to first-line medical therapy.

Methods

A single-center retrospective cohort study of all HS patients seen between 1/1/1992 and 10/1/2014 was conducted. Response to first-line medical therapy (oral/topical antibiotics, intralesional corticosteroids, and topical washes) was examined at follow-up within 6 months of initiating therapy. A multivariate binary logistic regression model was built examining response to treatment and the interplay of patient factors and treatment initiated.

Results

198 patients were included in the final model. Non-smokers (OR=2.634, 95%CI=1.301-5.332, p=0.007) and older individuals (OR=1.046 for each additional year, 95%CI=1.020-1.072, p<0.001) were more likely to have improvement at follow-up. Additionally, current smokers differed significantly from non-smokers in several regards.

Limitations

The retrospective nature of this study and relying on classification of disease severity from physical exam findings in some patients.

Conclusions

The results of this study suggest that clinicians may be able to more accurately predict which patients with HS will respond to first-line medical therapy, and which patients may require therapy escalation.

Introduction

Hidradenitis suppurativa (HS) is a chronic cutaneous condition characterized by recurrent painful nodules, abscesses, scarring, and sinus tract formation.¹ HS is a prevalent but under recognized health problem, affecting 0.006-4.1% of the general population.^2,3 Patients typically present in their second and third decades of life, with an average age of onset between 20-27.^4,5,6 More women than men are affected by HS, with a ratio of approximately 4:1 women to men, and the diagnosis of HS is often delayed due to misdiagnosis.^5,7 The pathophysiology of HS is not completely understood, but disease development is most likely related to environmental factors driving disease development in genetically susceptible individuals.⁸ Several factors — such as smoking, obesity, insulin resistance, family history, and several autoimmune/inflammatory diseases — have been demonstrated repeatedly to be associated with HS development.^9,10

Given that HS development is not fully understood and it is associated with a plethora of factors, any information about which HS patients will respond to first-line medical therapy and which patients may require therapy escalation could prove highly valuable to clinicians. With this in mind, the goal of this study was to determine which patient factors were associated with a positive response to first-line medical therapy in patients with HS.

Methods

Study Design

All portions of this study were conducted at Washington University School of Medicine as a part of an institutional review board approved study. A retrospective cohort chart review was performed for all patients seen at Washington University School of Medicine affiliate hospitals, namely Barnes Jewish Hospital and St. Louis Children's Hospital, between January 1^st, 1992, and October 1^st, 2014. Patients were identified through billing records, using the ICD 9 code for HS, 705.83.

Inclusion/Exclusion Criteria

Patients must have had: a dermatologist confirmed diagnosis of HS, been prescribed only first-line medical therapy at their initial HS specific dermatology visit, and had a follow-up visit within 6 months of the initial visit. First-line medical therapy was defined as any of the following alone or in any combination: oral antibiotics, topical antibiotics, intralesional corticosteroids, and antibacterial washes/creams/lotions (such as benzoyl peroxide soap).¹¹ Patients were included in analysis regardless their compliance to their prescribed medication regime.

Data Collection

Patient records were identified in the EMR (electronic medical record), and study data were collected and managed using REDCap (Research Electronic Data Capture) electronic data capture tools hosted at Washington University in St. Louis.¹² The first HS specific dermatologist visit (“initial visit”) was defined as the earliest note in the EMR where HS symptoms were addressed or treated. Patient demographics were collected from the EMR at the closest possible dates +/- one year of the patient's initial visit. The patient's weight and height were recorded at the closest possible date to the initial visit, and BMI (body mass index) was calculated from this in the standard fashion. Patients were considered to have a positive family history of HS if stated explicitly, or to have a noted family history of recurrent boils or abscesses — or similar descriptor — in typical HS locations. Patient medical comorbidities were based on all available data recorded in the EMR prior to the initial visit. Patients were considered to have a positive history of any medical condition if it was noted in the EMR at any point prior to their initial visit. Any recorded diagnosis of an autoimmune condition, regardless its association with HS, was noted (please see Fimmel et al. 2010 for a complete listing of HS associated autoimmune conditions and Hayter et al. 2012 for a comprehensive list of well-described autoimmune diseases).^13,14 Disease severity was assigned at the initial visit, and was based upon the stated physician assessment of disease. When there was no stated physician grading of the patients' disease, the authors assigned a grade based upon the Hurley criteria.¹ Those with inactive disease were graded as mild disease, due to the intermittent nature of mild HS. Surgical scars from previous treatments were not included as scarring due to HS when judging disease severity.

Outcomes

At the first follow-up visit within 6 months of the initial visit patients were dichotomously assigned to either improved or no change/worsening disease. The response to treatment was hierarchically based on either stated improvement at follow-up, or an improvement in the grading of the patients disease at the follow-up visit. If the physician recorded a response to treatment at the follow-up visit, it alone was used to judge response. However, if no response to treatment was noted in the follow-up note, then the response to treatment was based purely on disease severity/physical exam findings at the follow-up visit compared to the initial visit.

Statistics

Simple descriptive statistics were utilized to describe the overall study population. Factors selected for logistic regression analysis were based upon published literature of factors associated with HS, as well as clinical expertise. Univariate binary logistic regression was utilized to examine the association between these patient factors and an improved disease status at the follow-up visit. Following this, a multivariate binary logistic regression model was constructed utilizing the same factors. The model was constructed in a single block, one-step manner. Model diagnostics were performed by: exploring multicollinearity between factors using a linear regression model and tolerances, examining pseudo r² values, and examining a ROC (receiver operating characteristic) curve and c-statistic. Patients were then stratified on smoking status, current vs. non/former, and descriptive statistics were rerun and compared between these groups. Smoking was chosen to stratify on due to its high prevalence, modifiable nature, and its well-known association with HS development and more severe HS.¹⁵ Pearson's chi-square test, Fisher's exact test, and independent samples t-tests were used to compare these groups as appropriate. All statistics were performed utilizing SPSS (IBM Corp. IBM SPSS Statistics for Mac, Version 23.0. 2013. Armonk, NY: IBM Corp.). All figures were made utilizing GraphPad Prism (GraphPad Prism version 6 for Mac, GraphPad Software, La Jolla California USA).

Results

A total of 945 patients were identified through billing records utilizing ICD 9 code 705.83. Of these, 667 (70.6%) were seen by dermatology for HS. Of these 667 patients, 373 (55.9% of 667) were excluded due to no follow-up visit within 6 months, 48 (7.2% of 667) were excluded due to being started on either second-line therapies or no therapies during their initial visit. The final study population of 246 (36.9% of 667) individuals is described in Table 1. The mean (SD) age at presentation to our service for HS symptoms was 33.3 (14.8), and the median age range of HS symptom onset was between ages 16-20. At presentation, 4 (1.6%) of the patients were unable to be assigned a disease severity. The remaining 242 (98.4%) either had disease severity explicitly noted or had extensive enough physical exam findings recorded to assign a disease severity. At their initial visit, 214 (87.0%) of the patients were prescribed oral antibiotics, 136 (55.3%) were prescribed topical antibiotics, 47 (19.1%) were given intralesional corticosteroids, and 146 (59.3%) were prescribed antimicrobial creams/washes/lotions. At the follow-up visit, 132 (53.7%) of the patients had improvement in their disease. Of note, 46 (18.7%) of patients had some form of noted non-compliance with their medications, ranging from intermittently missing doses to failure to use any prescribed treatments.

Table 1. Characteristics of Overall Study Population.

Mean Age at First HS Dermatology Visit (SD1)
	33.3 (14.8)
Sex (% total)
Male	50 (20.3%)
Female	196 (79.7%)
BMI2 n=202 (SD)
	34.5 (9.0)
Race (% total)
Non-Caucasian	133 (54.1%)
Caucasian	113 (45.9%)
Positive Family History of HS3 (% total)
Yes	30 (12.2%)
No	216 (87.8%)
History of Autoimmune Disease (% total)
Yes	45 (18.3%)
No	201 (81.7%)
Carried Diagnosis Prior to First Visit (% total)
Yes	87 (35.4%)
No	159 (64.6%)
Disease Severity n=242 (% total)
Mild	130 (53.7%)
Moderate	101 (41.7%)
Severe	11 (4.6%)
Response to Treatment (% total)
Improved	132 (53.7%)
Worsened/No Change	114 (46.3%)
Treatments Prior to First HS Visit (% total)
Oral Antibiotics	155 (63.0%)
Incision and drainage or Excision	79 (32.1%)
Topical Washes/Creams/Lotions	42 (17.1%)
Topical Antibiotics	36 (14.6%)
Retinoids	12 (4.9%)
Oral Steroids	6 (2.4%)
Itralesional Corticosteroids	6 (2.4%)
IV Antibiotics	3 (1.2%)
Topical Steroids	3 (1.2%)
Biologics	3 (1.2%)
Hormonal Therapy	1 (0.4%)
Laser Therapy	1 (0.4%)
None	56 (22.8%)

¹SD = Standard deviation

²BMI = Body mass Index

³HS = Hidradenitis suppurativa

In the univariate binary logistic regression analysis, the only factor significantly associated with increased odds of having a positive response to first-line therapy was no previous diagnosis of HS (OR=1.734, 95%CI=1.024-2.938, p=0.041) (Table 2). In the multivariate binary logistic regression model, after controlling for all other factors, two factors were associated with increased odds of having a positive response to first-line therapy (Table 2). Non/former smokers were at a 2.634 (95%CI=1.301-5.332, p=0.007) times increased odds of having improvement in their disease compared to current smokers, regardless of the amount smoked. Additionally, each one-year increase in age at the initial HS specific visit saw a 4.6% (95%CI=1.020-1.072, p<0.001) increase in the odds of having improved disease at the follow-up visit. Although no longer statistically significant in the multivariate model, there was a nearly significant association between improvement at follow-up and no previous diagnosis of HS (OR=1.905, 95%CI=0.958-3.790, p=0.066). In the final binary logistic regression model 198 (80.5%) of the 246 patients had all of the necessary information to be included. The overall multivariate model was statistically significant (p=0.020), with no multicollinearity noted between any of the variables.

Table 2. Univariate and Multivariate Models.

Univariate Variable Model	OR4	95% CI5	p-value
Female Sex, n=246	1.202	0.646 - 2.238	0.561
BMI6 (1-unit increase), n=202	1.024	0.992 - 1.057	0.151
Non/Former Smoker, n=245	1.648	0.984 - 2.761	0.058
Caucasian Race, n=246	1.169	0.706 - 1.934	0.544
Positive Family History of HS7, n=246	0.985	0.458 - 2.118	0.970
No Previous Diagnosis of HS, n=246	1.734	1.024 - 2.938	0.041
One Year Increase in Age at Presentation, n=246	1.016	0.998 - 1.034	0.074
Moderate Disease, n=242	1.067	0.633 - 1.798	0.809
Severe Disease, n=242	0.714	0.208 - 2.458	0.594
No History of Autoimmune Disease, n=246	1.264	0.662 - 2.414	0.478
No History of Diabetes, n=246	1.042	0.513 - 2.117	0.909
Prescribed Oral Antibiotics, n=246	1.367	0.649 - 2.879	0.411
Prescribed Topical Antibiotics, n=246	1.539	0.960 - 2.645	0.072
Prescribed Topical Washes/Creams/Lotions n=246	1.281	0.769 - 2.135	0.341
Prescribed Intralesional Corticosteroids, n=246	0.977	0.517 - 1.848	0.943
Compliant with Prescribed Treatment n=246	1.485	0.780 - 2.286	0.229
Multivariate Variable Model (n=198)	OR	95% CI	p-value
Female Sex	1.382	0.637 - 2.996	0.413
BMI (1-unit increase)	1.022	0.985 - 1.060	0.250
Non/Former Smoker	2.634	1.301 - 5.332	0.007
Caucasian Race	1.244	0.642 - 2.408	0.518
Positive Family History of HS	1.013	0.381 - 2.692	0.979
No Previous Diagnosis of HS	1.905	0.958 - 3.790	0.066
One Year Increase in Age at Presentation	1.046	1.020 - 1.072	<0.001
Moderate Disease Vs. Mild Disease	1.096	0.564 - 2.128	0.787
Severe Disease Vs. Mild Disease	1.029	0.192 - 5.500	0.974
No History of Autoimmune Conditions	1.256	0.558 - 2.830	0.582
No History of Diabetes	1.744	0.683 - 4.455	0.245
Prescribed Oral Antibiotics	1.736	0.681 - 4.424	0.248
Prescribed Topical Antibiotics	1.240	0.635 - 2.422	0.528
Prescribed Topical Washes/Creams/Lotions	1.366	0.699 - 2.670	0.361
Prescribed Intralesional Corticosteroids	0.608	0.266 - 1.387	0.237
Compliant with Prescribed Treatment	1.762	0.792 - 3.917	0.165

⁴OR = Odds ratio

⁵CI = Confidence interval

⁶BMI = Body mass index

⁷HS = Hidradenitis suppurativa

When stratified on smoking status, current smokers did differ significantly from non/former smokers in several regards (Table 3). Current smokers were significantly older at their initial visit, with a mean age of 38.9, compared to non/former smokers, with a mean age of 29.6 (p<0.001). Current smokers were more likely to be Caucasian (OR=2.41, 95%CI=1.427-4.069, p=0.001), less likely to have a recorded family history of HS or HS-like syndrome (OR=0.341, 95%CI=0.134-0.867, p=0.019), more likely to have a history of an autoimmune condition (OR=3.127, 95%CI=1.601-6.108, p=0.001), and more likely to have inframammary HS involvement (OR=3.132, 95%CI=1.460-6.715, p=0.002).

Table 3. Characteristics of Current Smokers vs. Non-Smokers.

	Non/Former Smokers n=148 (60.4%)	Current Smokers n=97 (39.6%)	p value
Mean Age at First Visit (SD8)			<0.001
	29.6 (15.2)	38.9 (12.3)
Sex (% column total)			0.997
Male	30 (20.3%)	20 (20.6%)
Female	118 (79.7%)	77 (79.4%)
BMI9 (SD)	n = 118	n = 84	0.692
	34.7 (10.1)	34.3 (7.2)
Race (% column total)			0.001
Non-Caucasian	93 (62.8%)	40 (41.2%)
Caucasian	55 (37.2%)	57 (58.8%)
Family History of HS10 (% total)			0.019
Yes	24 (16.2%)	6 (6.2%)
No	124 (83.8%)	91 (93.8%)
History of Autoimmune Disease (% column total)			0.001
Yes	17 (11.5%)	28 (28.9%)
No	131 (88.5%)	69 (71.1%)
Carried Diagnosis Prior to First Visit (% column total)			0.880
Yes	52 (35.1%)	35 (36.1%)
No	96 (64.9%)	62 (63.9%)
Disease Severity (% column total)	n = 126	n = 95	0.434
Mild	81 (55.5%)	48 (50.5%)
Moderate	57 (39.0%)	44 (46.3%)
Severe	8 (5.5%)	3 (3.2%)
Response to Treatment (% column total)			0.057
Improved	87 (58.8%)	45 (46.4%)
Worsened/No Change	61 (41.2%)	52 (53.6%)
Patients with area involvement (% of patients)
Abdomen	8 (5.4%)	11 (11.3%)	0.089
Axilla	85 (57.4%)	53 (54.6%)	0.666
Buttocks	25 (16.9%)	12 (12.4%)	0.334
Chest	9 (6.1%)	3 (3.1%)	0.373
Inframammary	12 (8.1%)	21 (21.6)	0.002
Inguinal	67 (45.3%)	38 (39.2%)	0.346
Inner Thigh	14 (9.5%)	13 (13.4%)	0.335
Perineum	3 (2.0%)	7 (7.2%)	0.054
Pubic	13 (8.8%)	4 (4.1%)	0.203

⁸SD = Standard deviation

⁹BMI = Body mass index

¹⁰HS = Hidradenitis suppurativa

Discussion

The factors we identified as significantly associated with a positive response to medical treatment align with much of what is currently known about HS. It is estimated that the rate of active current smoking in HS patients is 40-88.9%, with current smokers having worse disease than those who do not smoke.^16,17,15 Smoking is known to be associated with derangement of inflammatory cell lines and cytokines, and if HS development is indeed related to a disregulated inflammatory state, then it stands to reason that current smokers may be less likely to respond to medical therapy.^8,18 In regards to age at presentation playing a strong role in response to therapy, it may be related to either the possible autoimmune/autoinflammatory nature of HS or to hormonal factors. If HS is indeed being driven by an autoimmune/autoinflammatory component in some patients, then it is possible that they will have a more mild disease if they develop HS later in life, similar to how many autoimmune conditions have a better prognosis as age of onset increases.¹⁹ Additionally, the decreasing levels of androgens found as an individual ages may influence HS responsiveness to treatment. There is some evidence that either high androgen levels and/or tissue sensitivity to androgens is important in HS development, and because of this anti-androgen therapy may be employed in treating HS.^20,21,22 If androgens truly play an important role in HS pathogenesis, then the decreasing levels of these hormones seen as individuals age could explain why HS is more amenable to treatment in older patients in this study.²³ Lastly, it stands to reason that not having a previous diagnosis of HS was associated – although only statistically significant in the univariate model – with a positive response to treatment because these individuals are likely earlier in their disease course compared to those who came for treatment with a diagnosis already established.

Limitations of this study stem predominately from its retrospective nature. Missing data limited the number of patients that could be included in the multivariate binary logistic regression model. The majority of patients that were not included in the final model were missing sufficient information to calculate a BMI (44 of the 48 not included in the final model). However, 198 (80.5%) of the eligible population had all of the data necessary for inclusion in the final model. Additionally, misclassification of disease severity was a serious concern, as some patients could only be classified based upon noted physical exam findings. However, the distribution of disease severity seen in this study population closely matches previously described distributions.²⁴ Lastly, these results may not be generalizable, as they are derived from a single academic tertiary-care center over a short time course. Regardless of these limitations, these findings — particularly the strong association with smoking status — are clinically meaningful.

In conclusion, older age and being a current non-smoker were both factors associated with a positive response when initializing treatment with oral antibiotics, topical antibiotics, intralesional corticosteroids, or antibacterial washes/creams/lotions. Cigarette smoking – the only modifiable factor identified – is highly important clinically, as patients should be advised that quitting smoking may improve their response to treatment and decrease the likelihood of inframammary HS involvement. Further studies, particularly looking at multiple centers over a much longer time course are needed to ensure the generalizability of these results.

Abbreviations

HS

Hidradenitis suppurativa

BMI

Body mass index

EMR

Electronic medical record

SD

Standard deviation

OR

Odds ratio

CI

Confidence Interval

ROC

Receiver Operating Characteristic