Sarcoidosis Treatment Patterns in the United States

Abstract

Background

There are limited US Food and Drug Administration-approved medications and real-world data on sarcoidosis treatment in the United States. Concordance of practice patterns with guideline recommendations have not been well characterized.

Research Question

What are the practice patterns and factors associated with treatment for patients with sarcoidosis in the year following diagnosis?

Study Design and Methods

A retrospective analysis was conducted of patients with sarcoidosis from 2016 to 2022 using a multicenter, all-payer claims database (TriNetX). Treatments with corticosteroids and/or nonsteroidal immunosuppressive medications (methotrexate, mycophenolate, leflunomide, hydroxychloroquine, cyclophosphamide, infliximab, adalimumab, azathioprine, rituximab, and Janus kinase inhibitors) within 1 year of diagnosis were ascertained. We summarized treatment rates and sequence of prescribed medications by mean rank, and used multivariable logistic regression analyses to identify factors associated with treatment.

Results

Of 13,330 patients with sarcoidosis meeting inclusion, 5,671 (42.5%) received treatment within 1 year of diagnosis. Of those treated, 60% received steroids alone, 13% received nonsteroidal immunosuppressives alone, and 27% received both. Furthermore, 25% of treated patients received a nonsteroidal immunosuppressive as their first medication. Corticosteroids had the lowest mean rank order, indicating they were, on average, the first medication initiated. Among those with pulmonary or cutaneous involvement, the second medication initiated, on average, was hydroxychloroquine; in those with cardiac or neurologic involvement, it was adalimumab and mycophenolate, respectively. Factors associated with higher odds of treatment were Black race, organ involvement at baseline (pulmonary, cardiac, and neurologic), and comorbid diagnoses (fatigue, hypercalcemia, and interstitial lung disease).

Interpretation

Within the first year of diagnosis, 43% of patients with sarcoidosis were started on treatment. Nonsteroidal immunosuppressives were used in 40% of treated patients. Although factors associated with treatment initiation aligned with guideline recommendations, practice patterns of treatment were variable, particularly in choice and sequence of nonsteroidal immunosuppressive therapy, underscoring the need for future trials and comparative effectiveness studies.

FOR EDITORIAL COMMENT, SEE PAGE 928

Take-Home Points.

Study Question: What are the practice patterns and factors associated with treatment for patients with sarcoidosis in the year following diagnosis, and how do these align with guideline recommendations?

Results: In the year following diagnosis, 43% of patients with sarcoidosis received treatment and 40% of treated patients received at least one nonsteroidal immunosuppressive medication with considerable variation in choice and sequence of administration.

Interpretation: Although factors associated with treatment initiation aligned with guideline recommendations, practice patterns of treatment in sarcoidosis varied, particularly in choice and sequence of nonsteroidal immunosuppressive therapy.

Sarcoidosis is a multisystemic granulomatous disease of unknown etiology¹ with a clinical presentation ranging from asymptomatic disease to organ failure.2, 3, 4 Although some patients with sarcoidosis have spontaneous disease remission, others may require treatment for symptom management, organ involvement, and/or risk reduction of disease progression.^5,6

Medications currently used for the treatment of sarcoidosis have not been studied in large randomized controlled trials. In those who receive treatment, corticosteroids remain the mainstay of therapy,7, 8, 9 with the most recent treatment guidelines advocating for steroids as the initial therapy for all presentations of sarcoidosis.^10,11 Although no immunosuppressive agents other than prednisone and repository corticotropin injection have been approved by the US Food and Drug Administration for sarcoidosis,^10,12 multiple immunosuppressive medications have been incorporated into stepwise treatment algorithms within guidelines developed by the European Respiratory Society, the American Thoracic Society, and the World Association of Sarcoidosis and Other Granulomatous Disorders.^10,13,14

Sarcoidosis guidelines often list multiple medications as second-line options after corticosteroids, and many of these recommendations are based on low quality of evidence. Data on sarcoidosis treatment in the United States are limited and inconsistent, with estimates for those requiring any treatment ranging from 23% among commercially insured patients to 78% in a rheumatology registry.^9,15 Furthermore, real-world practice patterns of sarcoidosis treatment in the United States have been incompletely characterized. Thus, we sought to study the following: (1) treatment rates overall and by organ involvement; (2) sequence of treatment initiation; and (3) factors associated with treatment for patients with sarcoidosis.

Study Design and Methods

Data Source

We leveraged the TriNetX Analytics Platform Research Network, an electronic health record and claims database with de-identified inpatient and outpatient data on > 120 million patients cared for at > 76 health care organizations across the United States, including academic and community facilities.¹⁶ Boston University’s Institutional Review Board deemed this study exempt from Institutional Review Board approval (H-43331).

Patient Selection

Adult patients were identified for study inclusion by using a validated definition of sarcoidosis,^9,17,18 based on presence of International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10) diagnosis of sarcoidosis (D86.x) on at least 2 ambulatory visits between September 1, 2016, and September 1, 2022. The date of incident diagnosis of sarcoidosis was considered the date of the first ambulatory visit with a sarcoidosis ICD-10 code. To exclude patients who may have been previously diagnosed with sarcoidosis outside the database, an ambulatory visit prior to incident diagnosis without an ICD-10 diagnosis for sarcoidosis was required. Patients were also excluded if they: (1) had an ICD-10 diagnosis for an alternative granulomatous condition (eg, fungal infection, TB, berylliosis, lymphoma, or eosinophilic granulomatosis with polyangiitis) at any point; (2) had an ICD-10 diagnosis of connective tissue/autoimmune disease prior to incident diagnosis of sarcoidosis; (3) received a nonsteroidal immunosuppressive medication prior to incident diagnosis (as these patients may have had an earlier diagnosis of sarcoidosis or alternative treatment indications); (4) had never received a medication listed in the database; or (5) received care at a health care organization outside of the United States. ICD-10 codes used to identify inclusion/exclusion criteria are presented in e-Table 1.

Outcomes

Outcomes of interest were treatment rates and timing and sequence of treatment initiation. Medications were selected per sarcoidosis practice guidelines^10,11 and included corticosteroids (prednisone, methylprednisolone, dexamethasone) and nonsteroidal immunosuppressive medications (methotrexate, mycophenolate, leflunomide, hydroxychloroquine, cyclophosphamide, infliximab, adalimumab, azathioprine, rituximab, and Janus kinase inhibitors).

Baseline Factors

Demographic characteristics (age, sex, race, ethnicity, and geographic location) and year of incident diagnosis were evaluated. Sarcoidosis organ involvement at the time of diagnosis was categorized by using D86.x ICD-10 code groups occurring on or within 30 days following incident diagnosis (e-Table 2) to account for providers assessing extent of disease involvement over more than one initial encounter. ICD subgroups for organ involvement were pulmonary and/or lymph node, cutaneous, cardiac, ocular and/or neurologic, and other sites; groups were not mutually exclusive, and patients could be coded for multiple organ involvements. The presence of comorbidities and symptoms (obesity, interstitial lung disease, cough, diabetes, tobacco use disorder, fatigue, dyspnea, palpitations, hypercalcemia, erythema nodosum, pulmonary hypertension, and heart failure) were coded by using ICD-10 codes on any encounter up to 1 year prior to incident diagnosis (e-Table 3). We also assessed for hospitalization in the year prior to diagnosis.

Statistical Analyses

Means ± SD, medians (interquartile range [IQR]), and proportions (%) as appropriate were used to summarize patient baseline characteristics and treatment rates overall and according to ICD subgroups for organ involvement at time of diagnosis. We assessed treatment timing and sequence of treatment by rank ordering medications for each patient according to date of first prescription and summarizing the mean rank of each medication by overall cohort and by diagnosis groups; medications received by ≤ 5 patients in each group were excluded from rank ordering. If 2 medications were prescribed on the same date, they received the same rank, with the next medication receiving two integer ranks above (eg, prednisone on January 1, 2016 = 1, hydroxychloroquine on January 1, 2016 = 1, methotrexate on February 1, 2016 = 3). A primary multivariable logistic regression analysis was conducted to identify factors associated with receipt of treatment overall. In secondary analyses, logistic regression was also used to identify factors associated with receipt of individual medications (only medications received by ≥ 1% of patients were included) for all patients with sarcoidosis as well as factors associated with receipt of a nonsteroidal immunosuppressive medication vs steroids alone among treated patients. All baseline comorbidities and symptoms were used as covariates for regression analysis other than “sarcoidosis, unspecified- D86.9” due to collinearity with other baseline ICD subgroups for organ involvement groups.

R version 4.1.2 (R Foundation for Statistical Computing) was used for analysis. Hypothesis tests were 2-sided, and alpha was set at 0.05.

Results

A total of 13,330 patients with sarcoidosis (median age, 57 years [IQR, 47-66 years]; 57% female, 59% White race) met inclusion criteria for the study (e-Fig 1). Baseline characteristics of the study population at time of incident diagnosis are presented in Table 1. There were 10,618 (80%) patients with a baseline ICD-10 code for “sarcoidosis-unspecified” and 3,538 (27%) with a baseline ICD-10 code for pulmonary and/or lymph node sarcoidosis. Overall, 5,671 (43%) patients with sarcoidosis received any treatment within 1 year of incident diagnosis, with 3,407 patients (60%) receiving treatment with steroids alone, 724 (13%) receiving nonsteroidal immunosuppressives alone, and 1,540 (27%) receiving treatment with both steroids and nonsteroidal immunosuppressives. Corticosteroids were the most frequently prescribed therapy (87% [n = 4,947]), followed by methotrexate (20% [n = 1,123]). The treatment rates among those coded for specific organ involvement were as follows: pulmonary and/or lymph node, 47%; cardiac, 53%; cutaneous, 42%; ocular and/or neurologic, 51%; and other sites, 54%. Among those treated, the median time from incident diagnosis to receipt of steroids was 25 days (IQR, 0-116 days), and the median time to receipt of a nonsteroidal treatment was 62 days (IQR, 9-155 days).

Table 1.

Baseline Patient Characteristics at Index Diagnosis

Characteristic	Overall (N = 13,330)
Diagnosis age, y	57 (47-66)
Sex
Female	7,593 (57)
Male	5,705 (43)
Unknown	32 (0.2)
Race
American Indian/Alaska Native	31 (0.2)
Asian	177 (1.3)
Black	3,718 (28)
Native Hawaiian/Pacific Islander	9 (0.1)
Unknown	1517 (11)
White	7878 (59)
Ethnicity
Hispanic or Latino	547 (4.1)
Not Hispanic or Latino	11,058 (83)
Unknown	1,725 (13)
Regional location
Midwest	886 (6.6)
Northeast	3,528 (27)
South	6,505 (49)
West	2,411 (18)
Sarcoidosis involvementa
Cardiac	669 (5.0)
Cutaneous	385 (2.9)
Ocular and/or neurologic	145 (1.1)
Other sites	61 (0.5)
Pulmonary and/or lymph node	3,538 (27)
Unspecified	10,618 (80)
Year of diagnosisb
2016-2017	2,479 (19)
2017-2018	2,301 (17)
2018-2019	2,213 (17)
2019-2020	2,098 (16)
2020-2021	2,135 (16)
2021-2022	2,104 (16)
Comorbidities and symptomsc
Cough	1,767 (13)
Diabetes	905 (6.8)
Dyspnea	2,863 (22)
Erythema nodosum	98 (0.7)
Fatigue	1,408 (11)
Heart failure	1,254 (9.4)
Hypercalcemia	427 (3.2)
Obesity	2,212 (17)
Palpitations	1,413 (11)
Interstitial lung disease	846 (6.3)
Pulmonary hypertension	480 (3.6)
Tobacco use disorder	265 (2.0)
Hospitalization in year prior	2,012 (15)

Data are presented as median (interquartile range) or No. (%).

^aInternational Statistical Classification of Diseases and Related Health Problems, 10th Revision subgroups for organ system involvement described in e-Table 2.

^bYear of diagnosis starts September 1 of each year until August 31 of the following year.

^cInternational Statistical Classification of Diseases and Related Health Problems, 10th Revision codes used to identify comorbidities and symptoms identified in year prior to index described in e-Table 3.

According to ICD-10 subgroups for organ system involvement, corticosteroid use ranged from 76% of treated patients with cutaneous involvement at diagnosis to 89% of treated patients with pulmonary and/or lymph node involvement (Fig 1). Hydroxychloroquine use ranged from 2.7% in treated patients with either ocular and/or neurologic or cardiac involvement to 39% of treated patients with cutaneous involvement. Mycophenolate was prescribed in < 7.0% of all treated patients but was the second most frequently prescribed nonsteroidal immunosuppressive medication in patients with cardiac involvement at diagnosis, with 26% of these patients receiving it within 1 year of diagnosis.

Figure 1.

Frequency of medications prescribed within 1 year of sarcoidosis diagnosis. Heat map illustrates the percentage of patients receiving each medication among treated patients in the overall cohort and treated patients coded for each ICD-subgroup of organ involvement. ICD subgroups for organ system involvement are described in e-Table 2. ICD = International Statistical Classification of Diseases and Related Health Problems, 10th Revision; JAK = Janus kinase.

Corticosteroids had the lowest mean rank of treatment initiation (ie, earliest medication started on average) among all patients and each organ involvement according to ICD-10-based subgroup (Fig 2). The second lowest mean rank medication (ie, second medication started on average) for the overall cohort, patients with cutaneous involvement, and patients with pulmonary and/or lymph node involvement was hydroxychloroquine; for patients with ocular and/or neurologic involvement, it was mycophenolate; and for cardiac involvement, it was adalimumab. A nonsteroidal treatment was the first medication prescribed in 1,392 (25%) treated patients; methotrexate was the most common nonsteroidal prescribed as a first medication (n = 577) followed by hydroxychloroquine (n = 347).

Figure 2.

Mean rank order of treatment initiation in sarcoidosis according to initial organ involvement. Average order of medications started in the overall cohort and treated patients coded for each International Statistical Classification of Diseases and Related Health Problems, 10th Revision, subgroup of organ involvement. Medications can be identified by color as seen in the legend. Lower position on y-axis indicates a medication was started, on average, earlier than a medication in higher position on the y-axis. International Statistical Classification of Diseases and Related Health Problems, 10th Revision, subgroups for organ system involvement are described in e-Table 2.

All patients in the cohort (N = 13,330) were included in multivariable regression analysis assessing factors associated with receiving treatment for sarcoidosis (e-Tables 4, 5). In multivariable regression, patients of Black race compared with those of White race (adjusted OR [aOR], 1.17; 95% CI, 1.07-1.27]), patients hospitalized in the year prior (aOR, 1.86; 95% CI, 1.68-2.07), and patients coded as having ICD-10-based organ involvement of pulmonary and/or lymph nodes (aOR, 1.25; 95% CI, 1.15-1.35), cardiac (aOR, 1.60; 95% CI, 1.35-1.88), and ocular and/or neurologic sarcoidosis (aOR, 1.53; 95% CI, 1.09-2.14) were at higher odds of receiving any treatment within 1 year of index diagnosis for sarcoidosis (Fig 3). Likewise, those with diagnostic codes for fatigue (aOR, 1.20; 95% CI, 1.06-1.35), dyspnea (aOR, 1.28; 95% CI, 1.17-1.41), cough (aOR, 1.28; 95% CI, 1.15-1.42), hypercalcemia (aOR, 2.01; 95% CI, 1.64-2.48), or interstitial lung disease (aOR, 1.48; 95% CI, 1.28-1.71) were at higher odds of receiving any treatment. Patients geographically located in US census region West were at lower odds (aOR, 0.76; 95% CI, 0.68-0.85) of receiving treatment within 1 year of index diagnosis for sarcoidosis compared with patients located in the Northeast.

Figure 3.

Factors associated with receiving treatment among patients with sarcoidosis. Logistic regression was performed on all patients in the study cohort (N = 13,330) to assess factors that were associated with treatment (steroids and/or nonsteroidal immunosuppressive medications). For categorical variables, categories used as reference values are noted as “ref.” Race categories of American Indian/Alaska Native and Native Hawaiian/Pacific Islander were combined into “Other” race to ensure model stability with small number of observations. Year of diagnosis starts September 1 of each year until August 31 of the following year. International Statistical Classification of Diseases and Related Health Problems, 10th Revision, groups to identify sarcoidosis involvement are described in e-Table 2. International Statistical Classification of Diseases and Related Health Problems, 10th Revision, codes used to identify comorbidities and symptoms identified in year prior to index diagnosis are described in e-Table 3. Boldfaced values in the CI column represent those that are statisticially significant.

aORs from multivariable regressions assessing receipt of each nonsteroidal immunosuppressive medications vs no or other treatment, and nonsteroidal immunosuppressive medications with or without corticosteroids vs steroids alone, are shown in e-Table 6. There was significant geographic variation in odds of receiving methotrexate, hydroxychloroquine, and mycophenolate as well as higher odds of being prescribed methotrexate, mycophenolate, and anti-tumor necrosis factor agents for those diagnosed in 2020-2021 or 2021-2022 compared with those diagnosed in 2016-2017. Those with hypercalcemia were at higher odds of receiving hydroxychloroquine, mycophenolate, and azathioprine, and patients who were hospitalized within the year prior were at higher odds of receiving infliximab.

Discussion

The current study used a large, multicenter, all-payer database to explore practice patterns of treatment for 13,330 patients diagnosed with sarcoidosis from 2016 to 2022 and found that 43% of patients received treatment for sarcoidosis within 1 year of diagnosis. We found that factors associated with likelihood of treatment aligned with guideline recommendations and noted variation in the use of nonsteroidal immunosuppressive medications, which were prescribed in 40% of treated patients. Differences were observed both in medication choice and timing, highlighting variation in practice after the decision to treat has been made.

The treatment rates in our study are consistent with several previously published single-center and multicenter studies exploring treatment trends.^8,9,15,19,20 Although none of these studies looked at treatment specifically within 1 year of diagnosis, our rate of any treatment (43%) was within the range previously reported, and we similarly found that steroids were the most frequently prescribed medication followed by methotrexate. In line with Baughman et al,⁹ the majority of medications in the current study were prescribed within 4 months of incident diagnosis. However, in our cohort, a higher percentage of treated patients received nonsteroidal immunosuppression (40%) than in previously studied cohorts, and 25% of treated patients received a nonsteroidal immunosuppressive prior to steroids, which may reflect changing practice, as most prior data precede ours and the likelihood of receiving these medications was higher in the later years of the current study.^8,9,19

Unlike prior studies, we additionally examined mean rank order of medications, adding granularity to the literature on the sequence of medications prescribed in practice. The most recent 2021 guidelines¹⁰ for sarcoidosis management outline stepwise algorithms for the treatment of sarcoidosis according to organ involvement; however, they do not address multiorgan involvement, and for most subgroups there is ambiguity in the choice of stepwise treatment option beyond steroids. For instance, the treatment algorithm for pulmonary disease includes multiple nonsteroidal immunosuppressive agents (methotrexate, hydroxychloroquine, mycophenolate, azathioprine, and leflunomide) as possible second-line treatment options following corticosteroids, with a slight preference to methotrexate for those with severe disease.¹⁰ Accordingly, we found that methotrexate was the most frequently prescribed nonsteroidal immunosuppressive among the treated patients in the subgroup with pulmonary and/or lymph node disease; however, although prescribed at a lower rate, hydroxychloroquine, when prescribed, was started earlier in treatment sequence on average than methotrexate. This may reflect practice variation due to overlapping patients with cutaneous involvement of their sarcoidosis, in which hydroxychloroquine is the only medication recommended as second-line therapy,¹⁰ vs small, observational evidence reporting the effectiveness of hydroxychloroquine in pulmonary sarcoidosis.²¹

In those with cardiac involvement, multiple nonsteroidal immunosuppressive agents (methotrexate, azathioprine, leflunomide, and mycophenolate) are provided as second-line therapy options or upfront concurrent steroids. In the current study, almost as many patients with cardiac sarcoidosis were treated with mycophenolate (26%) as methotrexate (32%). The high rate of mycophenolate use in this group may reflect retrospective data suggesting the medication’s effectiveness in cardiac sarcoidosis.^22,23 Our data, however, showed that both mycophenolate and methotrexate for cardiac sarcoidosis were, on average, added at a later point than adalimumab and hydroxychloroquine. Overall, our findings suggest variability in choice and sequence of medication initiation, highlighting the known ambiguity in sarcoidosis treatment pathways in real-world practice and illustrating the need for well-designed comparative effectiveness studies to better elucidate optimal stepwise treatment strategies for patients with sarcoidosis.

Guidelines note that the decision of who and when to treat for sarcoidosis depends on both risk of death/organ failure or impairment of quality of life; pulmonary, cardiac, ocular, neurologic, and renal disease are the primary causes of morbidity and mortality, and fatigue is a common quality of life driver of treatment.¹⁰ Accordingly, our study did suggest that providers in the current era of sarcoidosis care are following these indications to treat, with treatment in our cohort more likely in those with pulmonary and/or lymph node involvement, cardiac involvement, ocular and/or neurologic involvement, hypercalcemia, fatigue, dyspnea, and cough at baseline. When assessing the likelihood of receiving specific nonsteroidal immunosuppressives, we observed variation according to geography, year of diagnosis, organ involvement, and comorbid diagnoses and symptoms.

The current study has multiple limitations. As in all studies with administrative data, there is a risk of misclassification when using ICD-10 codes. Most patients were coded for “unspecified” sarcoidosis, and imaging, pulmonary function testing, and other test results were unavailable. We were thus unable to classify organ involvement in the majority of cases and were also unable to assess variation in treatment based on disease severity, site of diagnosis, or by provider specialty. Misclassification may be present at the outcome level as well, as patients may have been prescribed a medication of interest for an alternate condition. We excluded those with pre-existing autoimmune conditions to reduce this possibility. Furthermore, although we evaluated organ involvement of sarcoidosis at diagnosis, disease involvement could change over time, which may affect treatment decisions. The short median time to treatment in our cohort suggests that the clinical picture and subgroup codes early in diagnosis likely reflect the organ involvement used to guide treatment choice.

In addition, although we were able to exclude patients who received nonsteroidal immunosuppression prior to first sarcoidosis visit (suggesting an alternate diagnosis or a previously established sarcoidosis diagnosis), we were unable to exclude patients who had previously received corticosteroids given widespread use of these medications, and it is possible some patients captured were not truly at index diagnosis. We also removed patients without any pharmacy data who may have been treated at centers not submitting pharmacy data or may not have been followed up longitudinally at a TriNetX site. While doing so increases our specificity for patients with accurate pharmacy records, we acknowledge there is also a risk of slightly overestimating treatment rates. We were also unable to accurately assess duration and dosage of medications to assess treatment failure vs add-on treatments. Finally, there are many other treatments used in patients with sarcoidosis that we did not capture such as inhaled corticosteroids or beta-agonists for sarcoidosis-associated cough, eye drops for ocular disease, topical steroids and antibiotics for cutaneous disease, and stimulants for sarcoidosis-associated fatigue; however, the focus of this study was to assess steroidal and nonsteroidal immunosuppression.

Interpretation

Within 1 year of diagnosis with sarcoidosis, 43% of patients received treatment. Practice pattern variation was observed in the treatment of patients with sarcoidosis, particularly in relation to the use of nonsteroidal immunosuppressive medication choice and the sequence of these treatments, although factors associated with likelihood of receiving therapy were in line with prior studies and guidelines. The current findings underscore the need for future randomized trials and comparative effectiveness studies to help ascertain optimal treatment algorithms.

Funding/Support

Funding for D. A. S. was provided by 1F32HL168959-01; National Institutes of Health T32 HL007035.

Financial/Nonfinancial Disclosures

None declared.

Supplementary Data

e-Online Data.