Epidemiology of sarcoidosis: current findings and future directions

Abstract

Sarcoidosis is a granulomatous inflammatory disease with unknown etiology. Epidemiological studies have contributed greatly to our knowledge about sarcoidosis, providing critical information on the determinants and distribution of the disease. In this review, we summarize recently published findings from epidemiological studies on sarcoidosis. We review the epidemiological tools used, the incidence and prevalence of disease, mortality and cancer risk after sarcoidosis and nongenetic risk factors for sarcoidosis. Genetics studies have not been included as they deserve a separate review. Leveraging existing epidemiological data to conduct etiological studies aimed towards understanding and preventing disease is critical for future sarcoidosis research.

Introduction

Sarcoidosis is a systemic inflammatory disease which is characterized by granuloma formation and scarring which, in some cases, permanently impairs organ function. The etiology of sarcoidosis is unknown, but it is hypothesized that a combination of genetic and environmental factors contribute to its occurrence. Sarcoidosis can be challenging to study due to its relative rarity, lack of sensitive and specific diagnostic tests and its heterogeneous presentation which may reflect different etiologies.^1,2 Epidemiological studies have contributed greatly to our knowledge about sarcoidosis through describing its distribution, examining long-term outcomes and identifying risk factors for the disease. Recently, the use of large datasets have helped us to better understand patterns in sarcoidosis occurrence. Yet important gaps remain. It is necessary to strategically build upon population-based data in order to enable in-depth, etiologic studies if we wish to move forward from gathering information to preventing disease.

In this review, we summarize findings from epidemiological studies on sarcoidosis published in the last 3 years and discuss future directions for sarcoidosis research. Genetics studies were excluded from this review because the large number would be sufficient for a separate review. We searched MEDLINE from January 2015 to December 2017 using the MeSH term as well as the free text term in all fields to identify epidemiological studies for inclusion.

Epidemiological tools to study sarcoidosis

Epidemiological studies can be used to describe the distribution of sarcoidosis in a population and investigate risk factors for the disease. The tools used in epidemiology can be descriptive or analytic and help to clarify patterns and associations on a population level.

Clinical studies

Although the first to identify new illnesses or adverse effects of new exposures, clinical science is mainly focused on the diagnosis, treatment, and prognosis of disease in individual patients. As a result, clinical studies often include a highly selected group of patients who receive care at a specialized hospital or clinic.³ While these studies provide excellent data on the clinical characteristics of those with the disease, the study sample is not often representative of the full spectrum of disease in the population at large because some sick (or otherwise undiagnosed) people never come to the attention of health care providers.³ Finally, the comparison groups are often composed of other cases with either more or less severe disease, or with differing levels of treatment. Thus, etiology or risk cannot be fully explored.

Case-control studies

The case-control study design is an efficient way to investigate risk factors for disease when the disease is rare.³ Researchers enroll ‘cases’ (those with disease) from hospitals, clinics, or disease registries, and a sample of nondiseased (‘controls’) from the source population which produced the cases. One such study was the ACCESS (A Case Controlled Etiologic Study of Sarcoidosis) study, which enrolled 736 incident cases of sarcoidosis and 706 control subjects from 10 clinical centers in the USA between 1997 and 1999.^4,5 Case-control studies can be limited if data were collected retrospectively, increasing the chance of recall bias, and making it difficult to establish the temporal sequence between exposure and disease.³ The validity of case-control studies also depends on the appropriate selection of controls which in some circumstances can be challenging.

Cohort studies

Cohort studies assess the exposure status of healthy individuals and follow them over time to determine the incidence of disease or mortality. Cohort studies play an important role in understanding the etiology of sarcoidosis. Many such prospective studies [e.g. Black Women’s Health Study (BWHS), Nurses’ Health Study II (NHSII)] were originally initiated to study other outcomes (e.g. cancer, cardiovascular disease), but can be used to investigate prediagnosis factors not routinely captured in clinical databases such as dietary intake and lifestyle factors. It is not feasible in large observational cohorts to examine all participants for the presence of the disease of interest, a disadvantage associated with this study design. Thus, in contrast to smaller clinical studies, the BWHS and NHSII rely on self report of sarcoidosis followed by validation using supplemental questionnaires and medical record review in a subset.

Data from ACCESS, NHSII and BWHS have provided excellent information on risk factors for sarcoidosis. Newer and until recently untapped resources are population and administrative databases which can be used to conduct cohort and case-control studies. This is an efficient use of data that have already been collected, although not originally for sarcoidosis-specific research purposes. These databases often have large sample sizes and some are population-based, providing representative estimates in entire populations. However, because data were not collected for research purposes, they usually do not include detailed clinical data (e.g. sarcoidosis phenotype) and rely on International Classification of Disease (ICD) codes to identify sarcoidosis, which have not been validated. This leads to possible disease misclassification and mild and asymptomatic cases are likely to be underreported. Recent reports from Sweden, Taiwan and the USA have made efficient use of these already collected data to study sarcoidosis.

The Swedish National Patient Register is a nationwide database which has been used to identify sarcoidosis diagnoses from inpatient hospitalizations (since 1964) and outpatient care (since 2001) among all Swedish residents.⁶ These data were linked to other population-based registers containing information from before and after sarcoidosis diagnosis using each individual’s personal identification number. Another population-based and nationwide database is the Taiwan National Health Insurance Research Database (NHIRD), begun in 1995. The NHIRD captures demographics, clinical visits, prescriptions, medical specialty, and ICD diagnostic codes from insurance claim data for reimbursement on nearly 100% of the Taiwanese population.⁷ Such databases are not available in countries without a centralized healthcare system. The Optum healthcare database is an administrative database (not population-based) which includes approximately 15% of US residents who have commercial health insurance or are enrolled in Medicare parts C and D. (www.Medicare.gov) These data are powerful, including almost 30,000 patients with an insurance claim for sarcoidosis.⁸ However, generalizability may be limited as the population of patients captured in Optum is older than the general population of the USA and those without insurance coverage are underrepresented.

Incidence and prevalence

Race/geographical location

The incidence and prevalence of sarcoidosis has consistently been observed to be highest in Nordic countries and African Americans. Recent studies from Sweden⁶ and the USA^8,9 have provided contemporary estimates of incidence and prevalence which lend further support that these populations experience the highest rates of sarcoidosis. Interestingly, several recent studies in other populations report a prevalence of sarcoidosis greater than 0.05%, indicating that the disease may not be as rare as previously thought. Studies reporting the incidence and prevalence of sarcoidosis in the last three years are presented in Table 1 and incidence estimates categorized by ethnic group are displayed in Figure 1.

Table 1.

Sarcoidosis incidence and prevalence estimates reported in the literature 2015–2017.

	Country, data source	Incidence per 100,000 per year	Prevalence per 100,000	Age of onset, years	Female incident cases (%)
Arkema 6	Sweden, National Patient Register	11.5	160	Men: 45 Women: 55	45
Baughman 8	USA, Optum Health Care Database	African Americans: 17.8 White people: 8.1 Hispanics: 4.3 Asians: 3.2	African Americans: 14 White people: 50 Hispanics: 22 Asians: 19	NR	NR
Beghe 17	Italy, University Hospital of Parma	NR	49	Men: 47 Women: 54	58
Coquart 14	Guadeloupe, hospital and laboratory databases	2.3	21	NR	59
Duchemann 13	France, hospitals in Greater Paris, Seine-Saint-Denise County	4.9	30	NR	55
Dumas 9	USA, Nurses’ Health Study (women only)	11 African Americans: 46 White people: 11	100 African Americans: 519 White people: 92	Women: 48	100
Ungprasert 11	USA, Olmsted County, MN, healthcare providers	11	NR	Men: 43 Women: 48	50
Wu 7	Taiwan, National Health Insurance Research Database	NR	2.17	Men: 42Women: 51	63

NR, not reported.

Figure 1.

Incidence of sarcoidosis per 100,000 per year reported in the literature 2015–2017 categorized by ethnic group.

*Studies from Olmsted County, MN (USA) and Sweden include populations who are majority (>90%) white. NHSII, Nurses’ Health Study II.

A recent study from Sweden using nationwide population-based registers estimated an incidence of 11.5 per 100,000 per year and a prevalence of 0.16%.⁶ These estimates are lower than those reported in a previous Swedish study published in the 1980s, which observed an incidence of 19 per 100,000 per year.¹⁰ The reason for the discrepancy may be that different methods were used to identify sarcoidosis; the older study identified cases through mass radiographic survey of one county (Uppsala) and 10% of cases were discovered by chance. In contrast, the recent register-based Swedish study included individuals who received an ICD code in inpatient or outpatient care. Interestingly, a study from the Mayo Clinic (Olmsted County, MN, USA) using data from a cohort of mainly white people with a northern European ancestry observed an incidence rate of 11 per 100,000 per year, which is nearly identical to the recent results from Swedish data.¹¹

NHSII, a prospective cohort of female nurses followed from 1989 which collected self-reported diagnoses of sarcoidosis via questionnaire, reported a similar incidence (11 per 100,000 per year) and prevalence (0.10%).⁹ The majority of the study population was white, but the 4% who were black had a higher incidence (43/100,000 per year) and prevalence (0.52%) compared with white women.²⁶ These estimates are lower than what was previously reported in Black American women (Black Women’s Health Study, incidence 71/100,000 per year, prevalence 2.0%).¹²

In a study using the Optum health insurance medical claims database in the USA, the incidence rate ranged from 7.6 to 8.8 per 100,000 per year.⁸ African Americans had a higher incidence and prevalence compared with Caucasians, Hispanics and Asians. The prevalence in African Americans was 0.14% and the incidence was 17.8 per 100,000. Caucasians had a prevalence of 0.05% and the incidence was 8.1. Hispanics and Asians had the lowest rates, with a prevalence of 0.02% and an incidence between 3 and 4 per 100,000.⁸

In recent years, several studies have also described the occurrence of sarcoidosis in other populations not previously studied, providing a picture of the disease in other parts of the world. Data from the Taiwanese NHIRD reported an overall sarcoidosis prevalence of 2.17 per 100,000.⁷ In a study from Seine-Saint-Denise, a county of Greater Paris, France, the incidence was 4.9 per 100,000 per year and the prevalence was 0.03%.¹³ The population was quite diverse, allowing for an examination of the occurrence of sarcoidosis by geographical origin as a proxy for ethnicity. Afro Caribbeans had the highest incidence rates (16.9 per 100,000 per year) followed by North Africans (9.7), others (6.4) and Europeans (2.4).

One would think that Afro Caribbeans would have similar rates to Afro Caribbeans living in Europe if geographical origin is a suitable proxy for genetic risk through ethnicity. However, a study examining the occurrence of sarcoidosis in Afro Caribbeans in Guadeloupe reported an incidence of 2.3 per 100,000 and a prevalence of 0.02% which is much lower than the estimates seen in Afro Caribbeans in the French study.¹⁴ The difference could be due to more strict criteria for inclusion in this study, which required a positive biopsy. However, it may indicate that location matters as well as race and that ethnicity is not a perfect proxy for race. Another possible explanation for the increased rates observed among racial/ethnic minorities living outside of their respective geographical region of origin is what Hennessy and colleagues term ‘migratory bias’.¹⁵ A subtype of diagnostic access bias, migratory bias, posits that in the event of voluntary migration, immigrants may be more inclined to utilize healthcare in their new location than nonimmigrants. This differential utilization of healthcare provides greater access to diagnostic testing and can influence measures of disease incidence.¹⁶

Geographical location has been shown to be an important factor related to sarcoidosis occurrence and recent studies confirm variation by region. In the Swedish study, some northern counties had a much higher prevalence than others.⁶ In the NHSII, BWHS and in the Optum database, Western US states had the lowest rates.^8,9,12 A recent study from the Italian province of Parma reported a prevalence of 0.05%, which varied by residential location from 0.03% to 0.09%.¹⁷

Age and sex

In addition to race and geographical location, sarcoidosis occurrence varies greatly by age and sex. What causes this variation is unknown, but it indicates that sex plays a role in the manifestation of disease. In some populations, there is a 10-year difference in age at diagnosis between men and women. In Sweden, the age at diagnosis in men was 45 compared with 54 in women.⁶ In Italy, the age at diagnosis was 47 in men and 54 in women.¹⁷ The difference in age is not so large in other populations such as Olmsted County, MN (men 43 versus women 48 years old)¹¹ and in a Spanish urban tertiary teaching hospital report (men 44 versus women 49 years old).¹⁸ The overall average age of onset in most studies was between 47 and 51 years old, except for one report from Estonia (average age in men 34, women 43 years old).¹⁹ The statement repeated in the literature that sarcoidosis onset peaks between 20 and 45 years of age²⁰ is not supported by the majority of recent studies, with the peak ages of onset closer to 30–55 years old.

The age at onset might have changed over time, as was shown in a study from Japan.²¹ Age increased over a 40-year period, and a previously high peak onset in men in their 20 s diminished. This may be due to an aging population or a change in detection over time (better diagnostics, less routine chest radiography done as part of workplace exams in men which were common in the 1980s, increased medicalization of women over the life course). It could, however, indicate that exposures to environmental factors which are related to sex have changed over time.

The difference in age at diagnosis between men and women suggests that there is something different in the disease pathogenesis between the sexes, which could be related to genetic factors or environmental exposures which are sex specific, such as reproductive factors. The clinical manifestations of sarcoidosis also appear to differ by sex, with women having more musculoskeletal symptoms than men in two studies.^18,19 The observed patterns are not consistent across studies, however women have been reported to have more asymptomatic disease in one study,¹¹ whereas another reported that men were more likely to have asymptomatic disease.¹⁸ Future investigations should take into account age when examining sex differences, since age and sex are closely correlated in sarcoidosis.

Mortality

For most patients, sarcoidosis has a benign clinical course, but for a subset of patients it is a chronic, life-threatening disease. A study from the French Epidemiologic Centre for the Medical Causes of Death identified death certificates listing sarcoidosis as the cause of death or a significant condition contributing to death.²² The age-standardized mortality rate was 3.6 per million and increased over time. Sarcoidosis-related deaths occurred at younger ages than the general population and the average age at death was similar for both men and women, but women died less frequently. Two studies from the USA reported pulmonary sarcoidosis-related age-standardized mortality rates using death certificate data from the National Center for Health Statistics.^23,24 One study reported that the mortality rate from interstitial lung disease and pulmonary sarcoidosis increased from 2.7 deaths per 100,000 in 1980 to 5.5 deaths per 100,000 in 2014.²³ Another study using the same data source but from 1999 through 2010 estimated a death rate of 2.8 per 1 million with sarcoidosis as the immediate cause.²⁴ Mortality rates were higher in women compared with men (3.3 versus 2.3) and in African Americans versus white people (16 versus 1.3). There was a statistically significant increase in the sarcoidosis-related mortality rate in white individuals over the past decade.²⁴

Mortality studies based on death certificates require that sarcoidosis be listed as a cause of death to be included in the study, which likely includes more severe cases of sarcoidosis. Results from these studies show how many people die with sarcoidosis listed, but do not provide an estimate of mortality among people who had sarcoidosis in the past. Individuals who die of a disease downstream of sarcoidosis may not be captured in these studies if sarcoidosis was not considered a contributing cause at the time of death. Furthermore, changes over time in sarcoidosis-related deaths captured by death certificates may reflect changes in practices of completing death certificates. Autopsies are performed less often. Therefore, changes over time should be interpreted with caution. Death certificate data are also more likely to report a higher rate of sarcoidosis-related death in women due to the fact that women are diagnosed later in life and therefore sarcoidosis is more likely to be present at the time of death at an older age.

Several recent studies have used cohort study designs which follow individuals from the time of their diagnosis until death. Using US Veterans Health Administration data (a racially mixed, predominantly male population; 50.8% black and 86.0% male), the 1-year all-cause mortality was estimated to be 2.0%.²⁵ Male sex and black race were associated with an increase in mortality and ocular inflammation was associated with a decrease in mortality. The mortality rate was not compared with individuals without sarcoidosis. One study from the Mayo Clinic in Olmsted County, MN (USA) reported that mortality in patients with sarcoidosis was not higher than the general population [standardized mortality ratio 0.90; 95% confidence interval (CI) 0.74–1.08].²⁶ The mortality rate in sarcoidosis patients was 13 per 1000 over the study period (1946–2013). The lack of an association between sarcoidosis and increased mortality is interesting considering that data from this very same study population found that sarcoidosis patients had increased morbidity,²⁷ infection,²⁸ and cardiovascular disease,²⁹ but do not appear to die any younger than the general population.

Comorbidity has a significant impact on mortality in sarcoidosis. In a hospital-based study from Poland, the number of comorbidities and stage of sarcoidosis significantly predicted survival.³⁰ Several recent investigations have elucidated the comorbidity burden on patients with sarcoidosis, and are summarized in Table 2. These studies have examined multimorbidity,^27,31 infection,^28,32 malignancy,^33,34 cardiovascular disease,²⁹ venous thromboembolism,^35,36 fractures,^37,38 and autoimmune diseases.⁷ Patients with sarcoidosis are hospitalized more often than individuals without sarcoidosis³⁹ and in-hospital mortality is higher.⁴⁰

Table 2.

Studies examining the association between sarcoidosis and subsequent comorbidity reported in the literature 2015-2017

Author, year	Study population, country	No. patients with sarcoidosis	Average follow-up time	Outcome	Estimate of RR (95% CI)	Additional findings	Risk of outcome associated with sarcoidosis
Wu 7	National Health Insurance Research Database, Taiwan	1237	NR	Autoimmune diseases	HR 1.66 (95% CI 1.37–2.00)	Highest risks observed for autoimmune thyroid disease, Sjögren’s syndrome, ankylosing spondylitis	↑
Ungprasert 29	Olmsted County, MN, USA	345	15.1 years	Cardiovascular disease	HR 1.65 (95% CI 1.08–2.53)	Highest risks observed for congestive heart failure, cerebrovascular accident	↑
Ungprasert 35	Olmsted County, MN, USA	345	15.1 years	Venus thromboembolism	HR 3.04 (95% CI 1.47–6.29)		↑
Yaqoob 36	Explores electronic medical records from 26 healthcare systems, USA	53,680	NA (cross-sectional study)	Venus thromboembolism	OR 3.35 (95% CI 3.25–3.46)		↑
Bours 37	Primary care medical records (Clinical Practice Research Datalink), UK	5722	6.7 years	Vertebral and nonvertebral fractures	Any fracture RR 0.90 (95% CI 0.80–1.02) Vertebral RR 1.77 (95% CI 1.06–2.96) Nonvertebral 0.87 (95% CI 0.77–0.99)	Glucocorticoid use associated with an increased risk of fractures	↑ vertebral ↓ nonvertebral fractures
Ungprasert 38	Olmsted County, MN, USA	345	12.9 years	Fragility fracture	HR 2.18 (95% CI 1.23–3.88)		↑
Sogaard 34	Nationwide register study, Denmark	12,890	10 years	Malignancy, excluding nonmelanoma skin cancers	SIR 1.3 (95% CI 1.3–1.4)	Risk of tonsil, colon and immune-related cancers remained increased >10 years after diagnosis	↑
Ungprasert 33	Olmsted County, MN, USA	345	12.9 years	Malignancy, excluding nonmelanoma skin cancers	HR 0.72 (95% CI 0.49–1.06)	Increased risk of hematologic malignancy among cases of systemic sarcoidosis compared with those without systemic disease	↔
Ungprasert 39	Olmsted County, MN, USA	332	14.8 years	Hospitalization	RR 1.37 (95% CI 1.24–1.52)		↑
Ungprasert 28	Olmsted County, MN, USA	345	15.1 years	Hospitalized infection	HR 2.00 (95% CI 1.41–2.84)	Higher risk observed in patients who received immunosuppressive treatment	↑
Ungprasert 27	Olmsted County, MN, USA	345	12.9 years	Multimorbidity: at least two chronic conditions	HR 1.60 (95% CI 1.27. 2.01)	Higher risk of cardiovascular diseases, arthritis, depression and diabetes mellitus	↑
Brito-Zeron 31	Hospital-based cohort of patients with sarcoidosis and controls from primary care, Spain	218	NA(case-control study)	Comorbidities included in the Charlson Comorbidity Index	51% of patients with sarcoidosis had at least one comorbidity compared with 29% of controls	Patients with sarcoidosis more likely to have chronic liver disease, autoimmune diseases, chronic pulmonary disease and malignancy	↑

CI, confidence interval; HR, hazard ratio; NA, not applicable; NR, not reported; OR, odds ratio; RR, relative risk; SIR, standardized incidence ratio.

Overall, these data underscore the burden of sarcoidosis in terms of premature mortality and comorbidity. Because the disease has a heterogeneous onset and course, it is important in future studies to stratify patients by disease severity to provide estimates which take into account differences in patient characteristics.

Malignancy

The possible relationship between cancer and sarcoidosis has been investigated for decades with numerous reports of patients diagnosed with malignancies proximal to (immediately before or after) the diagnosis of sarcoidosis, yielding inconsistent results^41–46 Hypotheses supporting a possible association include the chronic inflammation or immune system dysregulation of sarcoidosis promoting the biologic mechanism behind carcinogenesis.^46,47 Additionally, immunosuppressive therapeutic regimen for sarcoidosis may increase the risk of cancer.⁴⁸

A meta-analysis of studies from Japan, the UK, the USA, and Scandinavia found a pooled relative risk (RR) for all invasive cancers of 1.19 (95% CI 1.07–1.32).⁴⁹ The observed risk was highest within 4 years of sarcoidosis diagnosis (RR 1.41; 95% CI 1.27–1.56), followed by those between 5 and 9 years post diagnosis (1.31; 95% CI 1.15–1.48); after 10 years, the risk dropped to 1.06 (95% CI 0.93–1.21). There was an approximately twofold increased risk of developing hematologic (lymphomas, Hodgkin’s lymphoma, leukemia), skin (melanoma, nonmelanoma), and organ (upper digestive tract, kidney, liver, colorectal, bladder) cancers.

Since the meta-analysis was published, two additional studies have reported estimates for the relative risk of cancer associated with sarcoidosis.^33,34 A follow-up study of the Danish National Registry of Patients from 1980 to 2011 found increased short- and long-term cancer risk after sarcoidosis diagnosis.³⁴ Specifically, patients with sarcoidosis had a 20% increased cancer risk 3–10 years post diagnosis, and a 10% increased risk more than 10 years post diagnosis. Patients were at increased risk of non-Hodgkins lymphoma, cancers of the tonsil, colon and biliary tract, and nonmelanoma skin cancer in the short term, while in the long term they were at increased risk of cancer of the biliary tract, basal cell carcinoma, and squamous cell skin.

In contrast, in the study from Olmsted County, MN (USA), from 1976 to 2013³³ there was a 28% reduced risk of any cancer among sarcoidosis cases compared with noncases [hazard ratio (HR) 0.72; 95% CI 0.49–1.06], and no increases (or decreases) of specific cancers among sarcoidosis cases. However, subgroup analyses found an increased risk of hematologic malignancy among cases of systemic (extra-thoracic) sarcoidosis compared with those with without extra-thoracic disease (HR 1.87; 95% CI 1.09–3.22), suggesting an elevated risk of cancer only among patients with extensive disease.

Taken together, these results continue to support the hypothesis of sarcoidosis as a risk factor for cancer, and that hematopoietic cancers may be the most likely manifestation. Additional cohorts of demographically and racially diverse study subjects are needed to further characterize this association.

Risk factors for sarcoidosis

The exact causes of sarcoidosis remain unknown, but a few factors have emerged which provide clues regarding disease etiology.

Obesity

The Black Women’s Health Study, a follow-up study of 59,000 African-American women, found that obesity [body mass index (BMI) ⩾30 kg/m²] at study baseline was associated with a 40% increased incidence of sarcoidosis over a 16-year period (1995–2011; HR 1.42; 95% CI 1.07–1.89).⁵⁰ Investigators also found that obesity at age 18 and cumulative weight gain since age 18 of at least 30 kg were each associated with an increased incidence. The association between BMI and sarcoidosis was also investigated in the NHSII cohort, a follow-up study of over 116,000 predominantly white (>98%), female registered nurses (1989–2013).⁵¹ Current obesity (in the 2-year period prior to diagnosis) was associated with a 70% increased risk of developing sarcoidosis (HR 1.74; 95% CI 1.26–2.40). Associations with sarcoidosis incidence were also observed for BMI at age 18 and cumulative weight gain of at least 25 kg since age 18.⁵¹ Finally, a smaller, case-control study of Olmsted County, MN (USA) residents (313 sarcoidosis cases, 200 controls; >90% white) also found a positive association between BMI (within 1 year prior to 3 months post diagnosis) and sarcoidosis risk. The odds ratio of sarcoidosis among obese compared with normal or low BMI subjects was 2.54 (95% CI 1.58–4.06).⁵² These recent results from different US populations are consistent with those from an earlier study in Danish women in which prepregnancy overweight and obesity was associated with a two- to threefold increased risk of sarcoidosis.⁵³ While the exact mechanism by which obesity influences sarcoidosis is not known, it is generally thought that the proinflammatory environment of obesity plays a primary role^54–61 and has been observed with other conditions including asthma.⁶² Understanding this mechanism may provide insight into possible avenues of treatment development and prevention.

Smoking

One of the strongest associations to emerge from the ACCESS study was a 35% reduced odds of sarcoidosis among ever smokers compared with never smokers,⁶³ a finding consistent with several earlier observations.^64–67 Recently, a case-control study of Olmsted County, MN (USA) residents found a stronger negative association between smoking and sarcoidosis.⁵² Compared with never smokers, the odds ratios for current smokers and former smokers was 0.34 (95% CI 0.23–0.50) and 0.68 (95% CI 0.45–1.01), respectively.

Occupational risk factors

Crystalline silica is an occupational exposure released in mining, construction, and agricultural work. Silica has been shown to be associated with several autoimmune diseases such as systemic sclerosis, rheumatoid arthritis, systemic lupus erythematosus, and Antineutrophil cytoplasmic antibody (ANCA)-related vasculitis.⁶⁸A recent cohort study conducted in silica-exposed workers in Sweden found that high exposure to silica dust (>0.048 mg/m³) was associated with an increased risk of sarcoidosis compared with those with less exposure (standardized incidence ratio 3.94; 95% CI 1.07–10.08).⁶⁹ Although based on only seven cases, these results support previous findings of an association between silica and sarcoidosis.⁷⁰ A better understanding of how silica triggers sarcoidosis and other autoimmune diseases may shed light on the pathogenesis of sarcoidosis. Firefighting has previously been suggested to increase the risk of sarcoidosis through the inhalation of unidentified toxins.^71,72 After the World Trade Center disaster on September 11, 2001, several reports found that workers exposed to the debris, in particular firefighters, had an increased risk of sarcoidosis or “sarcoid-like” disease.^73–75 A recent report examined this association further by comparing the rates observed in firefighters and emergency medical service workers who arrived as rescue/recovery workers soon after the disaster to age- and sex-specific sarcoidosis incidence rates from the Rochester Epidemiology Project.⁷⁶76 The World Trade Center-exposed workers had a 2.8-fold increased risk of sarcoidosis and the rate was higher in those with increased exposure (defined by proximity in time to the event and work duration). This study was limited by the use of a comparison group which did not experience the same level of screening as the exposure group.

Future directions

Despite more than 100 years of inquiry and research, we still know very little about sarcoidosis. In particular, little is known about the occurrence of sarcoidosis in other regions, such as Africa, which to our knowledge is based on studies from 30–50 years ago.^77–79 Because race has been shown to be an important factor in sarcoidosis occurrence, future studies should be conducted in diverse populations.

New ‘big data’ from population and administrative databases appear to have good validity and provide robust risk estimates among individuals seeking care for sarcoidosis. However, future work should focus on the validation of claims data and ICD codes used to diagnose sarcoidosis to better understand the data quality. The inability to stratify according to sarcoidosis phenotypes in these data is a limitation that should be addressed. Sarcoidosis is a heterogeneous disease and current estimates do not take into account the potentially wide variation in outcomes among different subsets of patients. Phenotyping of the disease is necessary for risk factor identification, as associations with specific subtypes may be missed if pooled together.

Disease registries may also play an important role in the study of sarcoidosis. The Worldwide Sarcoidosis Research Study (WISE)⁸⁰ was designed to collect information about the characteristics and clinical course of sarcoidosis to increase our understanding and improve treatment of the disease. The future study of sarcoidosis must involve such voluntary enrollment registries, but also a larger network of mandatory registries akin to the US cancer registries (https://seer.cancer.gov/registries/) where diagnostic entities (e.g. hospitals, laboratories) are mandated to report to a central source. This system will best facilitate the identification and recruitment of patients for epidemiologic research studies, provide detailed data on disease phenotype, allow generation of population estimates and assessment of geographic and environmental influences, and identify and recruit patients for epidemiologic research studies.

In order to increase our understanding of long-suspected (e.g. smoking, infectious agents) and novel (e.g. obesity) risk factors for sarcoidosis, we will need the appropriate epidemiologic tools, specifically greater reliance on prospective cohort studies. For example, infectious factors including mycobacterial antigens such as mKatG and Mtb-hsp have long been considered to be potential causes of sarcoidosis,^81–84 an association favored by pooled analyses of cross-sectional studies.^85,86 However, the temporal sequence of the association has not been established, thus it remains unclear whether exposure to mycobacterial antigens preceded diagnosis of sarcoidosis. Cohort studies with prediagnostic banked biological samples have the potential to answer this question.

Ideally, epidemiological studies can be used to inform future clinical and preclinical investigations of the pathogenesis of disease. For example, one of the strongest and most consistent associations across several studies is a reduced risk of sarcoidosis among smokers compared with nonsmokers. To test the hypothesized mechanism that nicotine’s immunomodulatory properties suppress granulomatous inflammation, a small randomized trial was conducted.⁸⁷ Thirteen symptomatic patients were randomized to receive 12 weeks of nicotine treatment plus usual care or usual care alone, and were compared with six asymptomatic and six disease-free patients. The nicotine-treated symptomatic group assumed the immune phenotype of asymptomatic group after the 12-week study period, suggesting a potential therapeutic role for nicotine for pulmonary sarcoidosis as well as elucidating immunologic processes underlying the disease. Larger studies should investigate clinical endpoints to follow up on these intriguing findings.

It is an exciting time for sarcoidosis epidemiological research with the recent emergence of new large population datasets. Examples from Sweden, Taiwan and the USA demonstrate the power of existing data to study a relatively rare disease. We must purposefully leverage these data in order to conduct in-depth, etiological studies with the goal of advancing the field towards understanding and preventing disease.