Hypersensitivity pneumonitis: Clinical manifestations - Prospective data from the interstitial lung disease-India registry.

CONTEXT: Multiple environmental factors are associated with development of hypersensitivity pneumonitis (HP), and diagnostic algorithms for the diagnosis of HP have been proposed in recent perspectives. AIMS: We analyzed the data of patients with HP from interstitial lung disease (ILD)-India registry. The analysis was performed to (1) find the prevalence of HP, (2) reclassify HP as per a recently proposed classification criterion to assess the level of diagnostic certainty, and (3) identify the causative agents for HP. SETTING AND DESIGNS: This was a prospective multicenter study of consecutive, consenting adult patients with new-onset ILD from 27 centers across India (March 2012-April 2015).
MATERIALS AND METHODS: The diagnoses were based on prespecified working clinical criteria and multidisciplinary discussions. To assess strength of diagnosis based on available clinical information, patients with HP were subclassified into definite HP, HP with high level of confidence, and HP with low level of confidence using a recent classification scheme.
RESULTS: Five hundred and thirteen of 1084 patients with new-onset ILD were clinically diagnosed with HP and subclassified as HP with high level of confidence (380, 74.1%), HP with low level of confidence (106, 20.7%), and definite HP (27, 5.3%). Exposures among patients with HP were birds (odds ratios [OR]: 3.52, P < 0.001), air-conditioners (OR: 2.23, P < 0.001), molds (OR: 1.79, P < 0.001), rural residence (OR: 1.64, P < 0.05), and air-coolers (OR: 1.45, P < 0.05).
CONCLUSIONS: About 47.3% of patients with new-onset ILD in India were diagnosed with HP, the majority of whom were diagnosed as HP with a high level of confidence. The most common exposures were birds, cooling devices, and visible molds.

INTRODUCTION

Hypersensitivity pneumonitis (HP) is an immune-mediated interstitial lung disease (ILD) attributed to inhalation of and sensitization to organic antigens.[1] The classification for HP currently used by many clinicians is based on a consensus made nearly three decades ago.[2] Recent perspectives have suggested new classification and diagnostic algorithms for HP that require validation;[345] the need for diagnostic and management guidelines for HP is evident.[6] Vasakova et al. recently proposed a classification scheme for HP based on clinical and radiographic manifestations with criteria for level of diagnostic certainty.[3]

While over 300 antigens have been associated with HP, an inducing antigen may be unidentifiable in up to 30% cases.[7] Antigen identification and avoidance are essential for optimal treatment as prognosis is worse among patients with chronic HP when an offending antigen is not identified.[1] Previous studies and ILD registries have noted varying estimates for the prevalence and incidence of HP (between 3.6% and 13.2%).[89101112] Notably, most of these data on HP are from the Western hemisphere and therefore may not reflect disease burden elsewhere. The goal of our current study was to find the prevalence of HP in India, reclassify HP based on a working diagnosis based on a recently proposed classification criterion and that proposed by an international working group that categorized ILD diagnoses with confidence to assess the level of diagnostic certainty, and identify the causative agents for HP.[3413]

MATERIALS AND METHODS

The ILD-India registry was a prospective study with 27 site investigators (SIs) in 19 cities across India from March 2012 to April 2015. Local ethics committees of individual centers and the clinical trial registry of India approved the study (CTRI/2013/05/003674).

Clinical data from consecutive consenting adult patients with new-onset ILD with high-resolution computed tomography (HRCT) evidence of bilateral parenchymal lung disease were prospectively gathered for the ILD-India registry. Detailed case report forms (CRFs) were completed by SI for all patients.[14] All patients underwent spirometry, HRCT of the chest, and serologic testing for connective tissue disease (CTD) (rheumatoid factor and antinuclear antibody). Patients with active infection, including Mycobacterium tuberculosis or malignancy, were excluded as described previously.[14]

Diagnosis of hypersensitivity pneumonitis

HP diagnosis was initially made based on working diagnostic criteria by individual SIs then subsequently by multidisciplinary discussion (MDD) among experts informally trained for the purposes of the ILD-India registry at the National Data Coordinating Center (NDCC), Jaipur, India.[14] The final diagnosis was validated and assigned by MDD among experienced experts at the Center for ILD (CILD), University of Washington, Seattle, USA.

For the purpose of this study, acute HP (nonfibrotic HP) was defined as symptoms <24 weeks and consistent computed tomography (CT) findings such as upper lobe predominant ground-glass opacities, poorly defined centrilobular nodules, mosaic attenuation, and air trapping.[14] Chronic HP (fibrotic HP) was defined as symptoms >24 weeks with fibrotic changes on HRCT. Subacute HP was excluded from this classification scheme as the current emphasis is on detection of fibrosis on HRCT as a marker of poor prognosis.[3415] Those diagnosed with HP were classified as confident HP (>90% confidence), HP with a high level of confidence (70%–89%), or HP with a low level of confidence. This working diagnosis was based on clinical diagnosis approach suggested by the international ILD working group and criteria proposed in a recent perspective on HP.[3413] Based on the same, the patients classified as HP by MDD earlier underwent a re-analysis and were reclassified into definite HP (based on HRCT and histopathology suggestive of HP), HP with high confidence (based on either HRCT suggestive of HP with positive exposure history or HRCT suggestive of alternate pattern but with positive exposure history), and HP with low level of confidence (based on HRCT suggestive HP and no exposure history).

Statistical analysis

Mean and standard deviation were calculated for continuous variables. ANOVA and Chi-square tests were done to assess differences between subgroups of HP as defined. Logistic regression was performed to assess whether various environmental antigens were associated with HP after adjusting for confounders (age, gender, smoking history, exposure to other antigens: molds, birds, air-coolers, air-conditioners, occupational exposure, and place of residence). Cohen's kappa was used to calculate the interobserver agreement between CILD, NDCC, and SI.

RESULTS

Among 1084 patients with new-onset ILD recruited during the study period, 513 (47.3%) were clinically diagnosed with HP based on working criteria and MDD [Figure 1]. The most common presenting symptoms included dyspnea (93.4%), cough (83%), weight loss (28.5%), lower extremity edema (14.6%), and chest pain (8%). HRCT pattern was consistent with HP [Figures 2 and 3] in 246/513 (47.9%) patients, while other patterns such as usual interstitial pneumonia (UIP), nonspecific interstitial pneumonia (NSIP), and organizing pneumonia were also seen. When patients with HP were classified based on the level of confidence in clinical diagnosis, the results were as follows: HP with high level of confidence in 380 (74.1%), HP with low level of confidence in 106 (20.7%), and confident HP in 27 (5.3%) cases [Table 1].[3413] The exposure history and radiological pattern of the patients in the three groups are described in Table 1. The demographics of patients with HP are shown in Table 2. The diagnosis of HP was ascertained by histopathology in 24 patients (transbronchial lung biopsy, n = 17, and surgical lung biopsy [SLB], n = 7).

Figure 1

Flowchart demonstrating the number of patients recruited in the interstitial lung disease-India registry with their respective diagnosis. *Vasakova et al.[3]

Figure 2

High-resolution computed tomography patterns (%) in the hypersensitivity pneumonitis versus other interstitial lung disease groups. **165 HP patients had other radiological patterns including 162 patients with an inconsistent with UIP pattern two with sarcoidosis and one with desquamative interstitial pneumonia. HP: Hypersensitivity pneumonitis, ILD: Interstitial lung disease, UIP: Usual interstitial pneumonia, NISP: Nonspecific interstitial pneumonia, OP: Organizing pneumonia

Figure 3

High-resolution computed tomography images of patients from the interstitial lung disease-India registry who were diagnosed to have acute (axial [a], coronal [b], sagittal [c]) and chronic (axial [d], coronal [e], sagittal [f]) hypersensitivity pneumonitis on the basis of multidisciplinary discussion including histopathology in both cases

Table 1

Clinical diagnosis of HP based on the level of confidence (adapted from recent perspective)

	Clinical and radiological features	Exposure history	HRCT pattern	Number of patients diagnosed with HP (%)
HP with high level of confidence	HRCT features characteristic of HP* + positive exposure historyORUIP, NSIP, DIP, OP, upper lobe nodular, and CPFE + positive exposure history	380	140 features consistent with HP*	380 (74.1)
			64 UIP
			11 NSIP
			2 OP
			1 DIP
			162 inconsistent with UIP
HP with low level of confidence	HRCT features characteristic of HP + no exposure history	0	106 features consistent with HP*	106 (20.7)
Definite/confident HP	Histopathology and radiology features consistent with HP**	25	15 UIP	27 (5.3)
			10 NSIP
			2 sarcoidosis
Total cases				513

*Acute (<24 weeks’ symptoms): upper lobe predominant disease, GGO, centrilobular nodules, mosaic attenuation, air trapping. Chronic HP (symptom duration >24 weeks): upper lobe predominant disease, basal sparing, reticulations, honeycombing, architectural distortion, mosaic attenuation, air trapping, centrilobular nodules, **Acute: Lymphoplasmacytic infiltrates, airway-centered lymphocytic infiltrates, loosely formed granulomas, multinucleated giant cells, cellular NSIP. Chronic: UIP-like, fibrotic NSIP-like, airway-centered fibrosis. HRCT: High-resolution computed tomography, HP: Hypersensitivity pneumonitis, UIP: Usual interstitial pneumonia, NSIP: Nonspecific interstitial pneumonia, OP: Organizing pneumonia, DIP: Desquamative interstitial pneumonia, GGO: Ground glass attenuation, CPFE: Combined pulmonary fibrosis and emphysema

Table 2

Demographics of patients diagnosed with hypersensitivity pneumonitis, hypersensitivity pneumonitis with high level of confidence, and hypersensitivity pneumonitis ascertained with histopathology

	Definite/confident HP (ascertained with histopathology) (n=27)	HP with high level of confidence (n=380)	HP with low level of confidence (n=106)	Statistical test	P
Age, years, mean±SD	49.03±13.52	56.4±12.7	58.5±13.15	ANOVA	<0.05*
Sex (% male)	6 (22.2)	155 (40.8)	47 (44.3)	χ2	>0.05
Symptoms prior to diagnosis (Years), mean±SD	3.4±2.9	3.9±3.5	4.3±4.4	ANOVA	>0.05
Exposure to agent known to cause HP	25	380	0	-
FVC (L), mean±SD	1.46±0.79	1.54±0.68	1.47±0.67	ANOVA	>0.05
FVC (%), mean±SD	52.94±22.68	57.75±26.72	56.07±16.15	ANOVA	>0.05
Associated features
Prior history of pulmonary tuberculosis (%)	18.5	14.7	20.7	ANOVA	>0.05
History suggestive of gastroesophageal reflux disease (%)	59.3	42.1	35.8	ANOVA	>0.05

*HP with high level of confidence and possible HP with low level of confidence >0.05. HP with high level of confidence and HP diagnosed with histopathology <0.05. HP with low level of confidence and HP diagnosed with histopathology <0.01. HP: Hypersensitivity pneumonitis, SD: Standard deviation, FVC: Forced vital capacity

Most of the patients with HP were classified as chronic HP/fibrotic HP (96%). While all patients were subjected to spirometry, 84 patients could not perform the test to meet standards for reproducibility and acceptability as stipulated by the American Thoracic Society (mainly due to intractable cough). The mean forced vital capacity (FVC) was 1.5 ± 0.7 L (57.1% ± 25.2% predicted). Diffusion capacity measurements were available in 71 cases, and the mean diffusing capacity for carbon monoxide, corrected to hemoglobin, was 9.14 ± 7.27 ml/mmHg/min (53.1% ± 31.3% predicted).

Seventy percent of HP patients lived in urban areas. There were 99/513 (19.3%) cases who had a history of current tobacco smoking. Common exposures (not mutually exclusive) among patients with HP were as follows: air-coolers (48.1%), air-conditioners (26.3%), birds (21.4%), molds (20.7%), and occupational exposure (5.8%). About 78.9% of patients had an identifiable exposure and 21.1% had no identifiable history of exposures associated with HP. Occupational exposures included farming (n = 22), cotton workers (n = 4), flour mill worker (n = 1), oil mill worker (n = 1), woodcutter (n = 1), and bird food distributor (n = 1). Nearly 48.4% of patients with HP had a single exposure to environmental agent and the remainder had multiple exposures. Table 3 shows adjusted and unadjusted odds ratios (OR) for the diagnosis of HP based on antigen exposure. The adjustment was done for variables such as age, sex, smoking history, occupational exposure, place of residence, molds, air-conditioners, air-cooler, and birds. The adjusted odds of developing HP in patients exposed to birds was 3.52 (95% confidence interval [CI]: 2.29–5.40), air-conditioners was 2.23 (95% CI: 1.59–3.14), molds was 1.79 (95% CI: 1.23–2.60), rural residence was 1.64 (95% CI: 1.12–2.42), air-coolers was 1.45 (95% CI: 1.11–1.90), and occupational exposure was 1.39 (95% CI: 0.76–2.56) more than any other type of ILD.

Table 3

Odds of developing hypersensitivity pneumonitis after exposure to various environmental factors using multivariate analysis

Environmental factor	Adjusted OR** (n=513) 95% CI	P	Unadjusted OR (n=513) 95% CI	P
Birds	3.52 (2.29-5.40)	<0.001*	4.311 (2.873-6.468)	<0.001
Air-conditioner	2.23 (1.59-3.14)	<0.001*	2.475 (1.805-3.394)	<0.001
Molds	1.79 (1.23-2.60)	<0.05*	2.177 (1.549-3.062)	<0.001
Rural residence as compared to urban	1.64 (1.12-2.42)	<0.05*	1.201 (0.865-1.666)	>0.05
Air-cooler	1.45 (1.11-1.90)	<0.05*	1.671 (1.309-2.131)	<0.001
Occupational exposure	1.39 (0.76-2.56)	>0.05	1.416 (0.816-2.455)	>0.05
Smoking	0.82 (0.58-1.23)	>0.05	0.736 (0.551-0.984)	<0.05

*P<0.05 significant, **OR calculated using multivariate logistic regression considering variables such as age, sex, smoking history, occupational exposure, place of residence, molds, air-conditioners, air-cooler, and birds. OR: Odds ratio, CI: Confidence interval

The Cohen's kappa interobserver agreement on HP and other ILD diagnoses between CILD and NDCC was 0.580 (95% CI: 0.534–0.626, moderate), between CILD and SI was 0.271 (95% CI: 0.226–0.316, fine), and between NDCC and SI was 0.492 (95% CI: 0.435–0.548, moderate).

DISCUSSION

HP was the most commonly diagnosed ILD in 47.3% of patients in the ILD-India registry, which is the largest and first prospective study to describe the pattern of ILD diagnoses among patients with new-onset ILD from multiple centers in India using the 2011 guidelines for idiopathic pulmonary fibrosis (IPF) and revised classification of idiopathic interstitial pneumonia (IIP).[16] The high incidence of HP in India is the highest compared to registries in other countries and may in part be due to the use of prompted questionnaires focused on exposure history to environmental inciting agent and/or on actual collective rates of HP in the specified regions of India. Regardless, this finding raises a concern about a variety of environmental exposures, most notably avian antigens, and molds from cooling devices that could induce HP.

There is often substantial uncertainty in the diagnosis of HP. This occurs most frequently in the distinction between fibrotic HP and IPF,[17] reflecting the overlapping features and lack of a definitive gold standard test for both diagnoses. We have used the approach proposed in a recent perspective and by an international working group that categorized ILD diagnoses.[313] This approach is supported by recent studies suggesting potential therapeutic and prognostic utility of assigning diagnostic confidence in this manner.[18] When we re-analyzed the ILD-India registry data utilizing the recently proposed classification for level of confidence in HP diagnosis, 380/513 (74.1%) patients were considered to have HP with high level of confidence and 27 (5.3%) patients had definite/confident diagnosis of HP. Thus, only a minority of the patients (106, 20.7%) were diagnosed with HP with a low level of confidence.[3]

Previous studies of the incidence and prevalence of ILD and ILD subtypes such as HP have varied. This may be due to differences in methodologies and case definitions or due to actual differences in HP incidence and prevalence attributable to differences in climate, environment, and cultural practices. Epidemiological studies from Western countries have estimated the incidence of HP to range from 3.6% to 13.2%.[1112] However, these data were collected prior to publication of the 2011 evidence-based guidelines for the diagnosis of IPF and a 2013 statement regarding IIP and HP; therefore, diagnoses may not have been accurate.[1619] Data from more recent prospective studies suggest an HP incidence of 2.6%, 4%, and 6.4% in Greece, Turkey, and Saudi Arabia, respectively.[82021] Older studies from India have been single center and small; HP made up 6%–11% of ILD cases with the most common ILDs found to be IPF or CTD-ILD.[222324] A recent retrospective study from a single center in India diagnosed 10.7% of ILD patients with HP and 42% with sarcoidosis.[25] While the data analyzed from the ILD-India registry did not include patients from this center, the noted differences may be explained by differences in data gathering, region, climate, exposures, and pollution.

Persistent, low-level exposure may lead to gradual damage to the lungs with symptoms becoming more pronounced after FVC falls below a certain level. Among patients with HP in the ILD-India registry, the majority had symptoms for a few years prior to diagnosis, suggesting that many had low-level chronic antigen exposure. This may also explain the low proportion of patients with acute HP in our registry. In accordance with the 2011 criteria for the diagnosis of IPF, patients with a UIP pattern on HRCT but a history of exposure to an antigen associated with HP were diagnosed with HP rather than IPF.[19] Had we not elicited the exposure history in the prompted CRFs, we would have diagnosed an additional 15.4% of patients in the registry with IPF based on UIP pattern on HRCT, rather than HP. This is in keeping with a recent study, demonstrating that 20/46 of patients diagnosed with IPF per the 2011 criterion had HP after reevaluation with detailed exposure histories and review of available histopathology in a center with long-standing expertise in HP.[17] Our findings emphasize the importance of excluding history of exposure to offending organic antigens before diagnosing a patient with IPF.[13]

Based on the design of the ILD-India registry, it is difficult to determine causality of HP because the frequency of exposure to the offending antigen in the healthy population is unknown, and data from controls were unavailable. Although the majority of patients with HP in our registry were exposed to air-coolers, the odds of developing HP as compared to other types of ILD was the highest in patients exposed to birds, followed in order by molds, air-conditioners, rural residence, and air-coolers based on multivariate analysis. The OR reported in the current study is slightly lower than that previously reported (1.67; 95% CI: 1.31–2.13; P < 0.001) for air-cooler exposure leading to HP because the previous analyses did not take into account other concomitant exposures such as mold, birds, and air-conditioners.[26]

All of the patients with HP with a high level of confidence in our study had an identifiable exposure to an antigen (s) known to be associated with HP. Among those patients with typical HRCT features consistent with HP, 21% did not have a history of overt exposure to antigens associated with HP (patients with HP with low level of confidence). We believe that many of these patients may have “cryptogenic HP” where an offending antigen is not identifiable; this has been associated with worse outcomes.[13]

The diagnosis of HP is heavily dependent on a high index of clinical suspicion, and there are no recent clinical practice guidelines for HP diagnosis, the need for which is evident from a number of recent perspectives on the topic.[34613] A recent HP classification proposed by Salisbury et al. includes a positive exposure history and HRCT features suggestive of HP as the two criteria to make a diagnosis of “HP likely” warranting treatment initiation.[4] In our study, the majority of patients with HP were diagnosed with a high level of confidence based on criteria proposed by Vasakova et al. and the diagnostic ontology proposed for the fibrotic ILDs by Ryerson et al.[313] In another perspective aimed at developing international guidelines for the diagnosis of HP, with the help of Delphi survey, positive antigen exposure with temporal association with symptoms, ground glass, mosaic attenuation on HRCT and nonnecrotizing granulomas on histopathology were the factors associated with HP.[27]

Acknowledging that there are no published guidelines for the diagnosis of HP, this observational study has several limitations. The clinical criteria used to diagnosis HP were based on evolved knowledge of patterns of IIP as there are not validated guideline criteria published. Only a small proportion of patients had the diagnosis confirmed by the histopathology; while histopathology may not be needed in all patients suspected to have HP, the low frequency of lung biopsy in the ILD-India registry (7.5%) must be acknowledged as it reflects the current practice of management of ILD in multiple centers across India and in Europe.[3428] The low rate of surgical lung biopsy is multifactorial and includes reluctant patients or physicians, comorbidities with high risk for complication and mortality, advanced stage of disease at presentation, or limited facilities for performing SLB. Additional limitations include that only 4% of patients with HP were diagnosed with nonfibrotic/acute HP. This could be because acute HP may be misdiagnosed as a viral infection, reactive airway disease, or tuberculosis, which is a common upper lobe disease in India. Although a prospective multicenter database was created for the purpose of ILD-India registry, data were gathered from consenting patients evaluated by investigators willing to enroll eligible patients. Thus, only those patients who volunteered and were able to afford the costs associated with the clinical tests prespecified for enrollment in the registry were included. Therefore, all new-onset ILD patients in India were not represented. Finally, while certain exposures were more common in the HP group, a cause-effect relation cannot be established based on study design. We do not have longitudinal follow-up data given limited resources available for this study. Nevertheless, our study generates awareness of HP as a common new-onset ILD in India and brings attention to the importance of minimizing and avoiding environmental exposures such as birds and mold.[29]

CONCLUSIONS

Nearly half of the 1084 patients with new-onset ILD enrolled in the ILD-India registry were diagnosed with HP based on clinical features, evolving clinical knowledge, and MDD. When all 513 HP cases were subclassified as definite HP or HP with varying levels of confidence, the majority were HP with high level of confidence. While prompted CRFs and questionnaires used in gathering data may have contributed in part to the relatively high proportion of patients diagnosed with HP, regional differences and standardization of data acquisition need to be considered in future well-designed prospective studies. The need for clinical practice guidelines for HP is evident and overdue. In the interim, the medical community and the public will need to be educated regarding occult and/or overt environmental factors associated with HP with the hope of preventing new-onset HP as well as minimizing risks for progression to chronic HP, a disease with poor prognosis and behavior similar to IPF. Using recently proposed criteria to assess the level of diagnostic confidence may be helpful particularly among patients where histopathology is not available.

Financial support and sponsorship

The Registry was funded by the Indian Chest Society.

Conflicts of interest

There are no conflicts of interest.