Abstract
Background:
Due to the heterogeneity of sarcoidosis, there is a need to define clinical phenotypes to allow for tailoring of clinical care and identification of more homogenous populations to facilitate research.
Methods:
We utilized data from a prospectively collected registry of sarcoidosis patients seen at a single quaternary referral center between January 2019 and February 2021. We used multiple correspondence analysis (MCA) and k-means clustering to investigate if the clusters previously identified in the GenPhenReSa study were reproducible in a US population. We also investigated if these clusters were stable when the population was stratified by race.
Results:
We replicated 3 of the 5 clusters seen in the GenPhenReSa study in our cohort. We likewise identified similar clusters between White and Black patients with sarcoidosis. Differences in organ manifestations associations between White and Black patients were seen primarily in relation to cardiac, neurologic, and ocular involvement.
Conclusions:
The organ clusters of liver-spleen, isolated pulmonary, and musculoskeletal-skin were reproducible in a US cohort, and in both Black and White patients.
Keywords: Sarcoidosis, Cluster analysis, Organ involvement
1. Introduction
Sarcoidosis is a multisystem granulomatous disease characterized by the presence of noncaseating granulomas in various organ systems primarily affecting the lungs [1]. Due to heterogeneity in the clinical course, many studies have sought to identify clinical phenotypes to improve understanding of the disease pathogenesis, risk stratify patients and to inform research design [2]. Multiple classification strategies have been developed. Some, including the influential Scadding stages [3], have focused primarily on pulmonary involvement [4,5]. Others have focused on patterns of organ involvement, given that multiorgan sarcoidosis is known to have worse outcomes [6]. The “Genotype-Phenotype Relationship in Sarcoidosis” project was a large scale multicenter European study that was designed to investigate the relationship between genotypes and phenotypes of sarcoidosis and identified six novel phenotypes of sarcoidosis defined by organ involvement patterns utilizing cluster analysis [7]. Other groups have likewise noted distinct associations of organ involvement [8–10].
The presence of organ involvement phenotypes observed in the European Genotype-Phenotype Relationship in Sarcoidosis project seems likely to apply to the sarcoidosis population encountered in the United States. The patterns of organ involvement may be different between Black and White patients given past work showing Black sarcoidosis patients in the United States have a higher incidence of sarcoidosis [11, 12], more severe disease on presentation [13], more frequent hospitalizations [14], and increased mortality [15]. However, the history of systemic racism in the United States obscures whether these differences are due to variations in environmental exposures, access to medical care, genetics, or other biological factors [16]. This includes the differences in evaluation applied by various health care centers. We investigated if the organ clusters seen in the European Genotype-Phenotype Relationship in Sarcoidosis project were replicable in White and Black sarcoidosis patients in the US using prospectively collected sarcoidosis registry data from our large quaternary sarcoidosis clinic in which a standardized approach of assessment and data collection is used.
2. Methods
2.1. Study population
This was an observational cohort study of patients who had diagnosis of sarcoidosis made or confirmed at the Cleveland Clinic Sarcoidosis Center of Excellence. We included 1634 of 1788 patients seen between January 2019 and February 2021 who had a confirmed diagnosis of sarcoidosis. The diagnosis was confirmed by one or more of the following: bronchoscopic biopsy (41.4%), other biopsy (34.8% e.g. skin, liver, abdominal lymph node), mediastinoscopy (6.8%), surgical lung biopsy (6.5%), CT scan (8.8%), and clinical (5.8% e.g. physical exam, typical clinical course). Patients with “uncertain” or “pending” diagnoses were excluded unless subsequent visits led to confirmation of the diagnosis. Organ involvement was determined by the treating physician. All physicians had a clinical focus on sarcoidosis and the WASOG organ assessment tool is widely utilized throughout our department but we did not require any threshold of WASOG instrument certainty (e.g., highly probable) for this study, due to the lack of validation of the instrument. Spirometry values in our registry used race corrections for percent predicted values. For statistical analyses, we only compared White and Black spirometry absolute values, and have noted where race corrected spirometry values are used. The study was approved by the institutional review board, with a waiver of individual consent (IRB 22–295). Self-reported race was extracted from the electronic medical record.
2.2. Statistical analyses
Clinic demographic and organ involvement were summarized as median (lower quartile, upper quartile) for continuous variables and frequency (percentage) for categorical variables.
The combination of multiple correspondence analysis (MCA) and k-means clustering was applied to 10 categorical variables representing various organ involvement (eye, heart, joints/muscle, kidney, liver, lung, neurologic, parotid, skin, spleen). MCA is an unsupervised machine learning method that allows studying the association for multivariate categorical data. It is used to detect and represent underlying structures in a categorical data set by representing data as points in a low-dimensional Euclidean space. Thus, it is viewed as a counterpart of principal component analysis for categorical data. K-means clustering analysis was then applied to the principal coordinates from the MCA, since the coordinates represent accurate distances between patients. The combination of MCA and k-means benefits the effectiveness of the calculation process and, in consequence, the clustering results.
The optimal number of clusters was decided by the elbow plot analysis based on the within sum of squares as well as silhouette scores, which reached a balance between minimal inner group and maximum intra-group difference. The stability of clustering analysis was assessed by generating bootstrap replicates and reclustering on them to determine the stability of clusters with respect to sampling noise. All the steps of the analysis were applied to both pooled data and subgroups of different races. All analyses were performed with the R software program (package FactoMineR and stats), version 4.2.0 (R Foundation for Statistical Computing, Vienna, Austria).
3. Results
3.1. Patient characteristics
Patient characteristics are outlined in Table 1. Our cohort of 1634 patients was 55% female and predominantly White (67%); 24% of patients were Black and 7% did not identify as White, Black, or Hispanic and were grouped as “Other”. Pulmonary involvement, including intrathoracic lymph node enlargement (83%) was the most common organ manifestation. Skin (17%), eyes (12%) and heart (16%) were the next most commonly involved organs. Median pulmonary function test values were in the normal range for the cohort utilizing race corrected percent predicted values (FEV1 87%, FVC 91%, DLCO 81%). A minority of patients had substantial respiratory symptoms defined as medical research council (MRC) dyspnea scale ≥3 (15%). A majority (65%) of patients were on active systemic treatment for sarcoidosis.
Table 1.
Summary statistics for total cohort and stratified by race.
| Total | Black | White | Hispanica | Otherb | |
|---|---|---|---|---|---|
| n (%) | 396 (24%) | 1101 (67%) | 20 (1%) | 117 (7%) | |
| Sarcoidosis Diagnosis (n = 1634) | |||||
| Biopsy confirmed | 84% | 83% | 84% | 85% | 83% |
| Smoking history (n = 1620) | |||||
| Never | 62% | 49% | 67% | 60% | 66% |
| Former | 32% | 42% | 29% | 40% | 29% |
| Current | 5% | 8% | 4% | 0% | 4% |
| Pack years (n = 373) | 9 (4–19) | 7 (3–15) | 10 (5–20) | 20 (1–30) | 10 (4–24) |
| BMI (n = 1152) | 31 (27–36) | 31 (27–37) | 30 (27–35) | 31 (25–39) | 32 (28–37) |
| FVC L (n = 1145)** | 3.3 (2.7–4.2) | 2.7 (2.2–3.4) | 3.5 (2.9–4.4) | 3.2 (2.7–4.5) | 3.4 (2.8–4.4) |
| FEV1 L (n = 1191)** | 2.5 (2.0–3.2) | 2.0 (1.6–2.6) | 2.6 (2.1–3.3) | 2.5 (1.8–3.5) | 2.6 (2.1–3.3) |
| FEV1/FVC (n = 1145) | 0.76 (0.70–0.80) | 0.77 (0.71–0.81) | 0.75 (0.70–0.79) | 0.75 (0.72–0.77) | 0.80 (0.76–0.84) |
| DLCO% (n = 1142)** | 82% (69–94%) | 68% (55–78%) | 86% (73–97%) | 84% (78–98%) | 83% (71–96%) |
| Organ involvement (n = 1634) | |||||
| Eye* | 12% | 17% | 10% | 10% | 10% |
| Heart* | 16% | 12% | 18% | 15% | 12% |
| Joints/muscle* | 9% | 6% | 10% | 0% | 7% |
| Kidney | 4% | 3% | 4% | 0% | 2% |
| Liver | 8% | 10% | 7% | 15% | 8% |
| Lung | 83% | 84% | 82% | 90% | 86% |
| Neurologic | 5% | 4% | 5% | 0% | 6% |
| Parotid | 1% | 2% | 1% | 5% | 2% |
| Skin** | 17% | 24% | 15% | 20% | 10% |
| Spleen | 6% | 5% | 7% | 10% | 4% |
| Medication (n = 1623) | |||||
| Never | 17% | 12% | 17% | 25% | 26% |
| Past | 18% | 24% | 16% | 20% | 18% |
| Current | 65% | 64% | 66% | 55% | 56% |
| MRC 3 or greater (n = 1300) | 15% | 16% | 14% | 27% | 22% |
FVC and FEV1 are reported in liters to avoid confounding from race corrections.
p < 0.05,
p < 0.0001.
Of the 20 Hispanic patients, 3 identified as Black, 11 identified as White, and 7 neither White or Black.
Patients in “Other” did not identify as Black, White, or Hispanic.
Race has been previously shown to be associated with varying disease outcomes, so we investigated if there were differences between White and Black patients (Table 1). There was a higher rate of never smoking in White patients compared to Black patients (67% vs 49%, respectively). Black patients had lower spirometry (raw) and DLCO (raw and percent predicted) than White patients (p < 0.0001). Black patients had higher rates of ocular (17% vs. 10% p < 0.01) and skin (24% vs. 15% p < 0.001) involvement. White patients had higher rates of cardiac (15% vs. 12% p = 0.02) and joint (10% vs. 6% p = 0.04) involvement. These comparisons were not adjusted for multiple testing.
The organ involvement seen in our cohort was similar to the GenPhenReSa and the ACCESS [11] cohorts. The ACCESS cohort was a multicentered, US, prospective case-control cohort of incident sarcoidosis patients (diagnosed within 6 months of enrollment to study) with age and sex matched controls. The majority of patients in our cohort had pulmonary involvement, defined as parenchymal or mediastinal lymph node involvement (79% vs. 95% in ACCESS, 93% in GenPhenReSa). Skin (15% vs. 16% in ACCESS, 16% in GenPhenReSa) and eye (12% vs. 12% in ACCESS, 8% in GenPhenReSa) involvement were common. Spleen (6% vs. 7% in ACCESS, 4% in GenPhenReSa), liver involvement (7% vs. 12% in ACCESS, 5% in GenPhenReSa) and kidney (4% vs. 1% in ACCESS, 3% in GenPhenReSa) were rarer. While neurologic involvement was comparable to past studies (5% vs. 5% in ACCESS, 3% in GenPhenReSa), cardiac sarcoidosis was over-represented in our sample (17% vs. 2% in ACCESS, 3% inGenPhenReSa). This is likely selection bias due to our status as a quaternary referral center with a focus on cardiac sarcoidosis and to temporal trends favoring more diagnoses of cardiac sarcoidosis.
3.2. Organ involvement clusters
We carried out multiple correspondence analysis (MCA) followed by hierarchical clustering on Principal components (HCPC) analysis in accordance with the methodology outlined by GenPhenReSa [7]. We identified 6 clusters that had substantial similarities to the clusters identified by GenPhenReSa (Tables 2 and 3). We also noted that most of these clusters remained intact when clustering was repeated using race stratification.
Table 2.
Comparison of the 6 clusters identified in our cohort, to the 5 clusters identified by Schupp et al. with the 6 clusters identified in our cohort. We report the percent of patients in each cluster who have each organ involved. We likewise report percent of patients in the total sample who were included in each cluster. Shading is proportional to prevalence of organ involvement.
| Cluster 1 -Liver/Spleen | Cluster 2 - Neuro/eye | Cluster 3 - Joint/skin | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| GenPhenReSa - “Abdominal” | GenPhenReSa - “OCCC” | GenPhenReSa - “Musculoskeletal-Cutaneous” | ||||||||||
| All | Black | White | GenPhenReSa | All | Black | White | GenPhenReSa | All | Black | White | GenPhenReSa | |
| n | 162 | 36 | 106 | 133 | 94 | 15 | 155 | 240 | 274 | 108 | 183 | 189 |
| Eyes | 11% | 3% | 9% | 3% | 30% | 13% | 13% | 55% | 18% | 48% | 14% | 12% |
| Heart | 14% | 8% | 15% | 5% | 9% | 0% | 43% | 21% | 4% | 1% | 6% | 2% |
| Joint/Muscle | 4% | 3% | 6% | 8% | 4% | 13% | 1% | 3% | 48% | 19% | 63% | 71% |
| Kidney | 1% | 0% | 2% | 11% | 0% | 0% | 0% | 2% | 0% | 0% | 0% | 4% |
| Liver | 70% | 67% | 67% | 68% | 3% | 7% | 1% | 2% | 0% | 5% | 0% | 3% |
| Lungs | 77% | 78% | 76% | 99% | 40% | 40% | 18% | 100% | 82% | 83% | 81% | 94% |
| Neuro | 1% | 3% | 0% | 4% | 83% | 100% | 37% | 27% | 0% | 0% | 1% | 2% |
| Parotid | 0% | 0% | 0% | 9% | 1% | 7% | 0% | 24% | 0% | 0% | 0% | 7% |
| Skin | 27% | 36% | 24% | 14% | 7% | 13% | 4% | 29% | 72% | 65% | 66% | 42% |
| Spleen | 58% | 50% | 65% | 58% | 1% | 0% | 1% | 0% | 1% | 0% | 1% | 2% |
| Cluster 4 -Pulmonary | Cluster 5 - Cardiac | Cluster 6 - Renal | White only - Parotid | |||||||||
| GenPhenReSa - “Pulmonary-Lymphonodal” | GenPhenReSa - “Extrapulmonary” | Not seen in GenPhenReSa | ||||||||||
| All | Black | White | GenPhenReSa | All | Black | White | GenPhenReSa | All | Black | White | White | |
| n | 768 | 176 | 605 | 1257 | 261 | 44 | 113 | 75 | 17 | 41 | 11 | |
| Eyes | 11% | 0% | 9% | 0% | 0% | 11% | 7% | 13% | 35% | 0% | 27% | |
| Heart | 1% | 0% | 17% | 0% | 82% | 100% | 7% | 7% | 6% | 10% | 0% | |
| Joint/Muscle | 0% | 0% | 0% | 0% | 0% | 0% | 13% | 15% | 12% | 12% | 9% | |
| Kidney | 0% | 0% | 0% | 2% | 0% | 0% | 20% | 79% | 76% | 100% | 18% | |
| Liver | 0% | 0% | 0% | 0% | 0% | 9% | 4% | 13% | 24% | 2% | 27% | |
| Lungs | 100% | 91% | 100% | 100% | 53% | 75% | 12% | 84% | 82% | 80% | 91% | |
| Neuro | 0% | 0% | 0% | 0% | 0% | 0% | 0% | 1% | 0% | 2% | 0% | |
| Parotid | 0% | 0% | 0% | 0% | 0% | 0% | 3% | 25% | 29% | 0% | 100% | |
| Skin | 0% | 0% | 0% | 12% | 3% | 14% | 30% | 20% | 29% | 15% | 0% | |
| Spleen | 0% | 0% | 0% | 0% | 0% | 0% | 3% | 8% | 6% | 7% | 9% | |
Table 3.
Summary of identified clusters by clinical characteristics and treatment regimens.
| Organ Involvement | Clinical | Treatment | |
|---|---|---|---|
| Cluster 1 | Liver (70%) | Moderate/severe liver dysfunction (3%) | |
| Spleen (58%) | Lowest FEV1 (2.4L) | ||
| Cluster 2 | Neuro (83%) | Treatment directed toward neurologic | Prednisone (38%) |
| Eyes (30%) Lungs (40%) | Treatment directed toward ocular | Methotrexate (37%) Infliximab (30%) | |
| Cluster 3 | Skin (72%) | Treatment directed toward skin | Hydroxychloroquine (20%) |
| Musculoskeletal (48%) | Treatment directed toward joints | ||
| Cluster 4 | Isolated Pulmonary (89%) | Highest proportion MRC 3 or more (18%) | |
| Second lowest FEV1 (2.4L) | |||
| Lowest FVC (3.3L) | |||
| Second lowest DLCO (81%) | |||
| Cluster 5 | Heart (82%) | Congestive heart failure (18%) | Prednisone (48%) |
| Lungs (53%) | Treatment targeting cardiac (51%) | Methotrexate (38%) | |
| Cluster 6 | Kidney (79%) Parotid (20%) | Lowest DLCO (78%) | Hydroxychloroquine (25%) |
Hepatic/Splenic/Pulmonary:
We identified a cluster of patients with hepatic and splenic involvement comparable to GenPhenReSa “abdominal” cluster. This cluster was defined by patients with either liver or spleen involvement but with only 27% of patients in the cluster having both liver and spleen involved. In our sample, however, renal involvement did not cluster with splenic and hepatic involvement. This cluster had the highest rate of moderate/severe liver disfunction (3%, p = 0.003) and in concordance with GenPhenReSa, had the lowest median FEV1 (% and absolute) of all six clusters (p < 0.0001).
Ocular/neurologic/without pulmonary:
We partially recapitulated GenPhenReSa “ocular-cardiac-cutaneous-CNS” cluster, with a cluster enriched for ocular and neurologic involvement, but without increased cardiac and cutaneous sarcoidosis involvement. In our sample this cluster had a high rate of patients with absence of pulmonary involvement, opposite of what was seen in GenPhenReSa. In this cluster 13% of patients had both CNS and ocular involvement and 4% of the cluster had eye and CNS involvement without lung involvement. This cluster had the second highest rate of prednisone (38%, p < 0.0001) and methotrexate (37%, p < 0.0001) use, exceeded only by the cardiac cluster, along with the highest rate of infliximab use (30%, p < 0.0001). This cluster also had the highest rate of sarcoidosis treatment directed toward neurologic involvement (54% of patients, p < 0.0001) or ocular involvement (16% p < 0.0001).
Joint/cutaneous/pulmonary:
We identified a cluster similar to GenPhenReSa “musculoskeletal-cutaneous” cluster with a high linkage of skin and joint/muscle symptoms. In this all patients had joint or skin involvement, but only 20% of patients had both joint and skin involvement. This cluster had the second highest frequency of hydroxychloroquine use (20%, p < 0.0001) after the renal cluster, and the highest frequency of treatment directed at skin (23%, p < 0.0001) or joints (18%, p < 0.0001). Interestingly, we saw a substantial group of patients in this cluster with eye involvement (18%), which supports the eye-skin linkage seen in the original OCCC cluster but that was not observed in our sample.
Isolated pulmonary:
We likewise observed a cluster of patients who were defined by pulmonary involvement without other organ involvement. This was the most homogenous cluster, with 89% of patients having isolated pulmonary sarcoidosis. This cluster had the largest proportion of patients with MRC 3 or higher dyspnea (18%, p = 0.008), second lowest DLCO percent predicted behind the renal cluster (80.9%, p = 0.0002), second lowest FEV1 (p = 0.001) following the liver spleen cluster, and the lowest FVC (p = 0.01).
Cardiac:
We did not recapitulate the “extrapulmonary” cluster seen in GenPhenReSa. Instead, we identified a cluster of patients with cardiac sarcoidosis with lower rates of pulmonary involvement. This group included patients with cardiac involvement (82%) and a substantial proportion had isolated cardiac sarcoidosis (28%). This cluster had the highest rate of congestive heart failure (18%, p < 0.0001), the highest rate of treatment targeting cardiac involvement (51%, p < 0.0001), and the highest rate of use of prednisone (48%, p < 0.0001) and methotrexate (38%, p < 0.0001).
Renal/Parotid:
We also identified a cluster with renal and parotid involvement. While all the patients in this cluster had renal or parotid involvement, only 4% had both renal and parotid involvement. Given the small number of patients in this cohort and the lack of comparator in previous studies, this group is of unclear clinical significance.
3.3. Organ cluster by race
We investigated for differing organ involvement clusters between White and Black sarcoidosis patients, based on past work showing differences in disease prognosis [11,12], and rates of organ involvement by race [11]. The clusters identified in the original analysis remained relatively constant when the population was stratified by race, but with differences in ocular, neurologic, and cardiac organ associations (Table 2). In Black patients, there was a high amount of ocular involvement in the “musculoskeletal-cutaneous” cluster and no ocular involvement in the “pulmonary cluster”, while White patients had baseline amounts of ocular involvement in both the “musculoskeletal” and “pulmonary” clusters. Both White and Black patients had a neurologic involvement cluster with a lack of pulmonary involvement (cluster 2). In White patients, cardiac involvement clustered with neurologic involvement in cluster 2 while in Black patients did not have substantial cardiac involvement in cluster 2. Both White and Black patients had a small cluster including renal and lung involvement (cluster 6), although Black patients also had substantial ocular involvement in this cluster. While Black patients had a separate cluster defined by pulmonary and cardiac involvement (cluster 5), cardiac involvement in White patients was clustered with lack of pulmonary involvement in cluster 2. Finally, White patients had a small unique cluster of patients with parotid involvement that is of unclear clinical relevance.
4. Discussion
We utilized a large prospectively collected patient registry containing 1634 unique patients at a quaternary sarcoidosis referral center to replicate the results of a large multicenter European study (GenPhenReSa) [7]. Within this cohort, we were able to identify similar organ involvement clusters to those identified in GenPhenReSa. Additionally, we identified similar organ involvement patterns between White and Black sarcoidosis patients in most of the clusters, but with variation in the associations of heart and ocular involvement.
The GenPhenReSa study was a multicenter prospectively recruited sarcoidosis cohort that identified five sarcoidosis phenotypes by cluster analysis. It was unclear how widely applicable these results would be, given the limiting of recruitment to local Caucasian parentage for two generations. We sought to investigate if the clusters identified in the GenPhenReSa study could be identified in a United States cohort that included a sizable proportion of Black patients. We identified 3 of the 5 clusters described in GenPhenReSa in our cohort (abdominal, musculoskeletal-cutanous, and pulmonary-lymphonodal). We also recapitulated the observed correlation of neurologic and ocular involvement, as well as the association of ocular and cutaneous involvement in Black patients. In White patients we confirmed the association of cardiac and neurologic involvement. We did not identify an “extrapulmonary cluster” and instead identified a cluster defined by cardiac sarcoidosis. It was noted that our sample had a large representation of cardiac sarcoidosis. It is possible that referral bias or other factors limited our ability to identify typical associations of cardiac sarcoidosis in the general sarcoidosis population. Alternatively, this referral bias may have allowed us to see an under-recognized but previously described phenotype of sarcoidosis (isolated cardiac sarcoidosis) [10]. It is less clear if the small cluster defined by parotid and renal involvement is clinically significant. Further work will be required to confirm these results. Future work will need to identify if these disease clusters predict outcomes or treatment response, or are related to genetic or environmental drivers of sarcoidosis.
It is notable that while as a population our data replicated 3 of the 5 GenPhenSeRa clusters, it would be challenging for clinicians at the bedside to determine to which cluster an individual patient would belong. Few patients had involvement of 2 or more organs in a pattern that was readily available as belonging to one of the described clusters (ex. only 20% of patients in the “liver/spleen” cluster had both liver and spleen involvement). The most clinically identifiable cluster was the “isolated pulmonary“ cluster of which 89% of patients matched the description of the cluster. While these clusters may be helpful for identifying biological underpinnings of the disease, it is important to note that these results do not suggest that the complexity of sarcoidosis organ involvement can be explained by 6 distinct clusters. This may explain why there have been divergent results when cluster methodology is applied to distinct samples of sarcoidosis patients [7,17]. At this point, the best use of these clusters may be to understand the disease mechanisms by comparing patients from replicable clusters (e.g. liver and spleen vs. isolated pulmonary sarcoidosis).
There are multiple possible explanations for differences in cardiac and ocular organ associations we observed between Black and White patients with sarcoidosis. Extreme care must be taken when attributing differences based on race to genetic mechanisms given known structural inequities. Race is a social construct that mediates powerful effects on health outcomes through forces such as structural racism, nationalism, and classism [18]. These forces have created challenges in parsing the biologic and social mediators of health outcomes in sarcoidosis [19–21]. Cardiac sarcoidosis presentation and severity have been shown to vary by race and gender [22], but at this point there are little data to suggest if these differences are related to genetics, differing exposures, or differing access to care. Differences in biology are possible, with past work showing ocular involvement to be more prevalent in Black patients than White patients [11,23], and identifying HLA-DRB1 alleles associated with uveitis [24] to have variable prevalence between White and Black sarcoidosis patients [25]. Given the ample evidence that self-reported race is a poor marker of genetic variability [26], it will be vital that future work investigating differences in disease manifestations and outcome between White and Black sarcoidosis patients consider the impact of systemic differences in environment and social environment for White and Black patients in the United States, many of which are driven by a history of systemic racism [27]. Recent work identified differing organ phenotypes between White and Black sarcoidosis patients in three rigorously phenotype cohorts [17]. While these cohorts were meticulously collected, there may have been selection bias, especially in the Black cohort, related to concerns of participating in genetic studies and recruitment strategies. Our work highlights the need for continued investigation to understand to what extent, and why, race is associated with organ involvement phenotype.
Our study has several notable strengths. Our large database provides a robust sample for analysis. Our distribution of White vs. Black sarcoidosis patients was in line with healthcare analysis, which suggests ~30% of US sarcoidosis patients are Black and ~60% are White [28,29]. Additionally, our distributions of organ involvement were largely in line with past published data. All patients’ organ involvement was documented at their outpatient clinic visits by a pulmonologist with a special focus on sarcoidosis. We have utilized robust statistical methodology to replicate a previous high-quality study. Unlike past studies, we have provided data quantifying the rate of organ co-presentation within our clusters.
Our study has several limitations. Our study sample, while large, is a single center, and prone to selection bias as described above. Additionally, while data were collected prospectively, organ involvement data are derived from provider assessment, dependent on physician aggressiveness to uncover organ involvement, and were not rigorously adjudicated. Testing patterns could also possibly have contributed to observed organ linkages (e.g. abdominal ultrasound identifying liver and spleen involvement). However, it is notable that the previously described sarcoidosis organ involvement clusters are replicable in a less rigorously collected sample, suggesting that they may be broadly applicable. The registry included only race corrected spirometry values, so we are unable to effectively compare spirometry values between White and Black patients. Finally, our database did not include detailed documentation of pulmonary involvement patterns or patient symptomology which prevents us from investigating if the organ clusters we identified are associated with specific clinical characteristics.
5. Conclusions
Using a large single center database, we were able to largely replicate the clusters identified in the GenPhenReSa study. We observed similar organ involvement clusters between White and Black patients, except for varying patterns of ocular and cardiac involvement. Our results suggest clustering methodology may be useful to identify biologically similar subgroups among groups of sarcoidosis patients. When seen in isolation however, many sarcoidosis patients will not be able to be accurately placed into a cluster, suggesting that at this point it is premature to use these clusters for individual clinical decision making. Future work is needed to understand the complexity of sarcoidosis organ manifestations, and the relationship between genetics, exposures, and health care access in driving disease course.
Funding
This project was supported by the National Institute on Minority Health and Health Disparities [K23MD017284] and the National Institute of General Medical Sciences [1R01GM152717-01]. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NIH.
Declaration of competing interest
The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:
LJH, AT, SH, XW have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.JCS:Lecture honoraria - Kinevant, Boehringer IngelheimRB:Research Grants - Celgene, Actelion, aTyr, Genentech, Gilead, Bellerophon, Bayer, Mallinckrodt, Foundation for Sarcoidosis Research.Scientific Advisor - Astra Zeneca, Actelion, aTyr, Kinevant, Xentria, Mallinckrodt, Boehringer Ingelheim, Foresee Pharmaceuticals, Ann Theodore Foundation.Speaker’s Bureau - Mallinckrodt, United Therapeutics, Boehringer IngelheimDAC:Steering Committee - Boehringer-IngelheimClinical Trials – Atyr, AI Therapeutics, MallinckrodtConsulting – Foresee Pharmaceuticals, Molecure, KinevantCommittee Honoraria – Pliant, FibroGen.
Footnotes
CRediT authorship contribution statement
Logan J. Harper: Writing – review & editing, Writing – original draft, Supervision, Project administration, Data curation. Anoosha Tauquir: Writing – review & editing, Writing – original draft, Project administration, Formal analysis, Data curation. Shuaiqi Huang: Formal analysis. Xiaofeng Wang: Supervision, Formal analysis. Jonas C. Schupp: Writing – review & editing, Supervision, Methodology. Robert Baughman: Writing – review & editing, Supervision. Daniel A. Culver: Writing – review & editing, Supervision, Conceptualization.
References
- [1].Crouser ED, Maier LA, Wilson KC, et al. , AMERICAN THORACIC SOCIETY DOCUMENTS Diagnosis and Detection of Sarcoidosis an Official American Thoracic Society Clinical Practice Guideline, 2020. [DOI] [PMC free article] [PubMed]
- [2].Pereira CAC, Dornfeld MC, Baughman R, Judson MA, Clinical phenotypes in sarcoidosis, Curr. Opin. Pulm. Med 20 (5) (2014) 496–502. [DOI] [PubMed] [Google Scholar]
- [3].Scadding JG, Prognosis of intrathoracic sarcoidosis in England. A review of 136 cases after five years’ observation, Br. Med. J 2 (5261) (1961) 1165–1172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [4].Sharp M, Psoter KJ, Balasubramanian A, et al. , Heterogeneity of lung function phenotypes in sarcoidosis: Role of race and sex differences, Ann Am Thorac Soc (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Lin NW, Arbet J, Mroz MM, et al. , Clinical phenotyping in sarcoidosis using cluster analysis, Respir. Res 23 (1) (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Sauer WH, Stern BJ, Baughman RP, Culver DA, Royal W, High-risk sarcoidosis: Current concepts and research imperatives, Ann Am Thorac Soc 14 (2017) S437–S444. [DOI] [PubMed] [Google Scholar]
- [7].Schupp JC, Freitag-Wolf S, Bargagli E, et al. , Phenotypes of organ involvement in sarcoidosis, Eur. Respir. J 51 (1) (2018). [DOI] [PubMed] [Google Scholar]
- [8].Oksanen V, Grönhagen-Riska C, Fyhrquist F, Somer H, Systemic manifestations and enzyme studies in sarcoidosis with neurologic involvement, Acta Med. Scand 218 (1) (1985) 123–127. [DOI] [PubMed] [Google Scholar]
- [9].Te HS, Perlman DM, Shenoy C, et al. , Clinical characteristics and organ system involvement in sarcoidosis: comparison of the University of Minnesota Cohort with other cohorts, BMC Pulm. Med 20 (2020) 155. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Okada DR, Bravo PE, Vita T, et al. , Isolated cardiac sarcoidosis: a focused review of an under-recognized entity, J Nucl Cardiol Off Publ Am Soc Nucl Cardiol 25 (4) (2018) 1136. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [11].Baughman RP, Teirstein AS, Judson MA, et al. , Clinical characteristics of patients in a case control study of sarcoidosis, Am. J. Respir. Crit. Care Med 164 (10 I) (2001) 1885–1889. [DOI] [PubMed] [Google Scholar]
- [12].Baughman RP, Field S, Costabel U, et al. , Sarcoidosis in America: analysis based on health care use, Ann Am Thorac Soc 13 (8) (2016) 1244–1252. [DOI] [PubMed] [Google Scholar]
- [13].Rabin D, Thompson B, Brown K, et al. , Sarcoidosis: social predictors of severity at presentation, Eur. Respir. J 24 (2004) 601–608. [DOI] [PubMed] [Google Scholar]
- [14].Gerke AK, Yang M, Tang F, Cavanaugh JE, Polgreen PM, Increased hospitalizations among sarcoidosis patients from 1998 to 2008: a population-based cohort study, BMC Pulm. Med 12 (1) (2012) 1–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].Swigris JJ, Olson AL, Huie TJ, et al. , Sarcoidosis-related mortality in the United States from 1988 to 2007, Am. J. Respir. Crit. Care Med 183 (11) (2011) 1524–1530. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [16].Hena KM, Sarcoidosis epidemiology: race matters, Front. Immunol 11 (2020) 537382. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Rasmussen A, Dawkins BA, Li C, et al. , Multiple correspondence analysis and HLA-associations of organ involvement in a large cohort of african-American and European-American patients with sarcoidosis, Lung 201 (3) (2023) 297–302. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Gee GC, Ford CL, Structural RACISM AND HEALTH INEQUITIES, Bois Rev Soc Sci Res Race 8 (1) (2011) 115–132. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Harper LJ, Gerke AK, Wang X-F, et al. , Income and other Contributors to poor outcomes in U.S. Patients with sarcoidosis, Am. J. Respir. Crit. Care Med 201 (8) (2020) 955–964. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Harper LJ, Love G, Singh R, Smith A, Culver DA, Thornton JD, Barriers to care among patients with sarcoidosis: a qualitative study, Ann Am Thorac Soc (2021). AnnalsATS.202011–1467OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Rabin DL, Richardson MSA, Stein SR, H Y, Sarcoidosis severity and socioeconomic status, Eur. Respir. J 18 (3) (2001) 499–506. [DOI] [PubMed] [Google Scholar]
- [22].Duvall C, Pavlovic N, Rosen NS, et al. , Sex and race differences in cardiac sarcoidosis presentation, treatment and outcomes, J. Card. Fail S1071916423001148 (2023). [DOI] [PubMed] [Google Scholar]
- [23].Judson MA, Boan AD, Lackland DT, The clinical course of sarcoidosis: presentation, diagnosis, and treatment in a large white and black cohort in the United States, Sarcoidosis Vasc. Diffuse Lung Dis 29 (2) (2012) 119–127. [PubMed] [Google Scholar]
- [24].Darlington P, Tallstedt L, Padyukov L, et al. , HLA-DRB1* alleles and symptoms associated with Heerfordt’s syndrome in sarcoidosis, Eur. Respir. J 38 (5) (2011) 1151–1157. [DOI] [PubMed] [Google Scholar]
- [25].Levin AAM, Adrianto I, Datta I, et al. , Association of HLA-DRB1 with sarcoidosis susceptibility and progression in African Americans, Am. J. Respir. Cell Mol. Biol 53 (2) (2015) 206–216. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Duello TM, Rivedal S, Wickland C, Weller A, Race and genetics versus ‘race’ in genetics, Evol Med Public Health 9 (1) (2021) 232–245. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Moller DR, Rybicki BA, Hamzeh NY, et al. , Genetic, immunologic, and environmental basis of sarcoidosis, Ann Am Thorac Soc 14 (Suppl 6) (2017) S429–S436. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Baughman RP, Field S, Costabel U, et al. , Sarcoidosis in America: analysis based on health care use, Ann Am Thorac Soc 13 (8) (2016) 1244–1252. [DOI] [PubMed] [Google Scholar]
- [29].U.S. Census Bureau, Historical income tables [internet] [cited 2020 Aug 7];2010. (3/7/2010). Available from: https://www.census.gov/data/tables/time-series/demo/income-poverty/historical-income-households.html, 2010.