Discussion

We present data from 308 patients enrolled into the BTS Sarcoidosis Registry between 2013 and 2017. The aims of the analysis were review the entire data set and concentrate specifically on areas where the BTS committee felt there was the least amount of prior evidence. The two areas of greatest interest were regarding biopsy for diagnosis and the use of steroid sparing agents.

With regard to biopsy for sarcoidosis, until recently, there was debate about the role of EBUS-TBNA.2 However, a recent study showed that as a single modality, EBUS-TBNA had the highest diagnostic yield at 74.5%, and addition of TBBX increased this to 91%.3 The choice and combination of diagnostic modalities is still unclear, but in our data EBUS-TBNA is more commonly used than TBBX. It may be that EBUS-TBNA should be reserved for cases of stage I or II disease where there is a reported higher yield than using TBBX alone.4 BAL was not widely used in our data. However, CD4:CD8 >3.5, in the absence of an increased proportion of other inflammatory cell types, may help support a diagnosis of possible sarcoidosis.5 BAL is generally a safe and useful adjunct and we argue that it should be used more widely in sarcoidosis, especially in cases where a transbronchial lung biopsy confers greater risk of bleeding or pneumothorax or in cases where extensive fibrosis means that a lung biopsy may have a low accuracy for diagnosis of sarcoidosis.

Identification of the both the proportion of registry patients who had a biopsy and in those who had a biopsy the type of procedure performed is an important finding from this data set. However, a further analysis will be needed to distinguish differing factors between the patients who had a biopsy and those who did not, as well as between the types of biopsy, for example, whether the extent of radiological changes influenced the type of bronchoscopic procedure performed.

The second main aim for this data analysis regarded the use of non-steroid agents in the pharmacological management of sarcoidosis and in this respect we feel our findings are important. Inhaled corticosteroids were used in 8% of registry patients and while there is evidence that an obstructive lung defect is present in almost a quarter of UK patients with sarcoidosis,6 studies have shown conflicting evidence with some showing improvement in lung function7 and others showing no benefit.8 One study with fluticasone showed no benefit against placebo but an improvement in cough.9 There are arguably no convincing studies to support the evidence of one steroid-sparing drug over another and the choice of agent used is often influenced by local specialist experience. However, as with previous Delphi studies, methotrexate was our commonly used second-line drug.10

Two additional important findings from our analysis involve patients with a family history sarcoidosis and the symptoms recorded at time of presentation. In our data set, 3.7% of all patients reported a family history of sarcoidosis. The only previous UK study to report familial associations received questionnaires from 268 patients and found similar results in that 5.2% had at least one first-degree, second-degree or third-degree relative with sarcoidosis.11

With regard to symptoms, fatigue was common (30%) and has shown association with poor quality of life.12 The aetiology is multifactorial including granuloma formation, depression, altered sleep patterns13 and use of corticosteroids.14 Fatigue is difficult to quantify, but the Fatigue Assessment Scale can differentiate between sarcoidosis fatigue and depression.15 We found no difference in fatigue presentation between men and women which contrasts with a recently published European phenotyping study of 2163 Caucasian patients where fatigue was seen at a higher incidence in women.16 Although the registry does not record the likely cause of fatigue, given the prevalence in our data, we suggest routine use of a fatigue questionnaire both as part of all new patient appointments and as a longitudinal assessment for patients who have difficult to manage fatigue.

The inferences of our findings are subject to some general limitations. Importantly, there is bias towards selective UK sites as although 24 hospitals entered data, it is possible that these represent sites where there are clinicians who have a specific interest in sarcoidosis. Indeed, of the 24 hospitals, 11 of them are designated tertiary centres for interstitial lung diseases where there may be a greater access to specialist services. It would be important to capture further data from general respiratory clinics as well as from these specialist ones. Additionally, we do not have information on whether the absence of a biopsy influenced the decision to offer treatment, given that in some cases, physicians may be less likely to initiate pharmacological treatment without a definite tissue diagnosis. Finally, few patients had follow-up data. More detailed longitudinal capture may prove useful in determining how often patients are followed up and when they can be discharged from specialist care.

Collection of data across multiple sites has both advantages and limitations. In a disease such as sarcoidosis with significant variation in practice and a lack of clear evidence on diagnostic and management pathways, multisite data collection allows researchers to review variations in practice across different institutions and provide support to experts as they write the next set of clinical guidelines for the disease. However, a significant limitation is the lack of standardisation of data collection across sites. For instance, although patients may have the same investigations, the interpretation of some of these is not standardised, for example, the finding of abnormalities on CT scans or the reporting of chest X-ray stage by Scadding classification. In order to mitigate against these variations, a central method of analysis and recording of all data is needed, but this will have significant logistical and financial implications due to the need for data transfer, centralised reporting of imaging and review of clinical records. Such a centralised process would therefore improve the accuracy of data collection but may be prohibitively expensive to attain.

As the BTS registry grows in volume of patient records, a number of important new questions will need to be asked which will require further analysis of the data and may require new data points for entry. First, as detailed in the original aims for this analysis, there is no clear evidence on either the need for a biopsy or in cases where tissue is obtained, the type of biopsy used to diagnose disease. A future approach to the registry will be to perform a subanalysis on the radiographic differences between patients who had a biopsy and those in whom a diagnosis was made on clinico-radiological grounds alone. This analysis may require a more detailed collection of CT imaging findings but may allow us to determine the radiological phenotype of patients in whom a biopsy is performed, for example, in those without classical CT findings such as bilateral hilar lymphadenopathy alone with no concerning radiological or clinical features to suggest an alternative diagnosis. Additionally, a further analysis is needed to determine whether the CT appearance influences the method of biopsy, for example, whether patients with predominantly hilar disease are subjected to EBUS-TBNA compared with those with predominantly parenchymal disease who may have TBBX. These findings would be important to both support future clinical statements in sarcoidosis and to design larger studies to help determine which type of biopsy to perform for a given phenotype of sarcoidosis.

Second, further questions need to be asked regarding the use of steroid-sparing agents. Our findings identified methotrexate as the most commonly used drug, but the registry currently has limited follow-up data on these patients. A targeted approach to specifically record longitudinal data on patients prescribed steroid-sparing agents will allow us to record the clinical outcome, changes in lung function and blood biomarkers and the incidence of side effects including fibrosis or liver toxicity. Such data would be important in helping design future trials with both traditional steroid-sparing agents and any novel medications to determine their use in the pharmacological management pathway for sarcoidosis.

Finally, it is noteworthy that none of the 308 patients were enrolled in clinical trials at the time of data capture. This may reflect the lack of available clinical trials for this disease, rather than a lack of enthusiasm for physicians to enter patients into trials. Sarcoidosis is a complex disease and drug regimens used have not been proven in randomised controlled trials. New treatments are needed for patients with multisystem and progressive forms of disease. One important use of such a patient registry may be to address this gap by identifying patients for new clinical trials.