Home Spirometry Telemonitoring for Early Detection of Bronchiolitis Obliterans Syndrome in Patients with Chronic Graft Versus Host Disease

Abstract

Background:

Early detection of bronchiolitis obliterans syndrome (BOS) after allogeneic hematopoietic cell transplantation (HCT) depends upon recognition of subclinical spirometric changes, which is possible only with frequent interval spirometry.

Objective:

We evaluated the feasibility of home monitoring of weekly spirometry via a wireless handheld device and a web monitoring portal in a cohort of high-risk patients for the detection of lung function changes preceding BOS diagnosis.

Study Design:

Observational study of 46 patients with chronic graft-versus-host disease or FEV₁ decline of unclear etiology after allogeneic HCT were enrolled to perform weekly home spirometry with a wireless portable spirometer for a year. Measurements were transmitted wirelessly to a cloud-based monitoring portal. Feasibility evaluation included adherence with study procedures and an assessment of the home spirometry measurements compared to laboratory pulmonary function tests.

Results:

Thirty-six (78%) patients completed one year of weekly monitoring. Overall adherence with weekly home spirometry measurements was 72% (IQR 47–90%), which did not meet the predetermined threshold of 75% for high adherence. Correlation of home FEV₁ with laboratory FEV₁ was high, with a bias of 0.123 L (Lower limit −0. 294 L, upper limit 0. 541 L), which is within acceptable limits for reliability. Of the 12 individuals who were diagnosed with BOS or suspected BOS during the study period, 9 had antecedent FEV₁ decline detected by home spirometry.

Conclusion:

Wireless handheld spirometry performed at home in a high-risk HCT cohort is feasible for close monitoring of pulmonary function and appears to facilitate early detection of BOS.

Introduction

Bronchiolitis obliterans syndrome (BOS) after allogeneic hematopoietic cell transplant (HCT) is a highly morbid manifestation of chronic graft-versus-host disease (cGVHD) that remains a diagnostic and therapeutic challenge. BOS is characterized by new onset airflow obstruction but is often not recognized until there has been substantial FEV₁ decline.¹ The pathophysiology is unclear but it is hypothesized that subclinical disease consists of alloimmune airways inflammation that, if unchecked, ultimately leads to fibrosis.² No therapy exists to reverse the airflow decline caused by fibrotic narrowing of small airways. Thus, the best opportunity for averting severe and irreversible lung dysfunction requires early recognition and intervention in the course of disease.³ For example, inhaled corticosteroids and other agents widely used for airways diseases have been associated with FEV₁ stabilization or transient improvement.^{4, 5}

Early detection of BOS is challenging for a number of reasons. The formal diagnosis of BOS is usually made 6 months or more after the initial diagnosis of chronic GVHD, although changes that indicate early airflow obstruction may be present prior to formal recognition. The NIH consensus guidelines definition of BOS relies on strict spirometric criteria with an FEV₁ cutoff of <75% predicted⁶ and does not take into account subclinical changes that indicate early BOS. Current NIH guidelines suggests monitoring high risk patients every 3 months with pulmonary function testing (PFT)⁷, but this practice is inconsistently adopted, owing to clinician attitudes, cost, and the inconvenience of obtaining PFTs for patients who live far from a transplant center.³ Patients with chronic GVHD are considered high risk, but even in this group, the prevalence of BOS is relatively low, at 14%.⁸ Given the lack of consistent longitudinal monitoring data, the onset and rate of early changes that develop into full-blown NIH-defined BOS remain undefined, although it is well-accepted that significant impairment can occur in a matter of weeks to months,³ and therefore may elude current monitoring standards.

Handheld spirometry, which may be performed at home, is routine clinical practice for the monitoring of acute and chronic allograft rejection in lung transplant recipients, and allows for frequent, safe and relatively inexpensive ongoing assessment. In HCT recipients, handheld spirometry measurements correlate strongly with laboratory measurements with acceptable accuracy,⁹ and have been utilized to detect late onset noninfectious complications in this population.¹⁰ Feasibility of home monitoring with a wireless device has also been shown in HCT recipients after day 100.¹¹ The aim of this study was to the assess the feasibility of home monitoring using repeated wireless handheld spirometry measurements in a population at high risk for BOS after HCT, including those with recently diagnosed chronic GVHD. Feasibility included adherence to weekly spirometry as well as the accuracy of handheld measurements compared to laboratory measurements. In addition, we assessed whether such a monitoring strategy could detect lung function decline that preceded a diagnosis of BOS.

Methods

Patients.

Allogeneic HCT recipients transplanted at Fred Hutchinson Cancer Research Center/Seattle Cancer Care Alliance were eligible for the study if one of the following high-risk conditions was met: 1) newly diagnosed chronic GVHD within 6 months requiring or anticipating systemic therapy, 2) at risk for chronic GVHD at day 80–100 posttransplant defined as still requiring >0.5mg/kg/day of systemic corticosteroids for acute GVHD or FEV₁ ≤70% predicted, or 3) deemed high risk by the primary investigator (PI). Because this was a feasibility study, the last criterion included patients brought to the PI’s attention with a history of posttransplant lung disease, or FEV₁ decline of unclear etiology in the context of mild chronic GVHD not requiring systemic therapy, or a chronic GVHD diagnosis >6 months. Patient outcomes were ascertained by medical chart review. BOS diagnosis was defined by modified NIH criteria (pre-bronchodilator FEV₁<75%, with 10% FEV₁ decline compared to the prior 2 years, and FEV₁/FVC <0.7)⁶ or by atypical phenotype criteria (FVC <80%, FEV₁ <80%, with normal total lung capacity)¹². If BOS was diagnosed, participants continued with spirometry procedures. The study protocol was approved by the Fred Hutchinson Cancer Research Center IRB and all participants provided written informed consent.

Spirometry Procedures.

A wireless handheld spirometer (GoSpiro, Monitored Therapeutics, Inc) was provided to each participant, who was trained on its use at the time of enrollment. The GoSpiro is a portable device with a removable turbine developed and FDA-approved for independent home monitoring of lung function by patients with pulmonary conditions (such as chronic obstructive pulmonary disease). The device connects wirelessly through Bluetooth technology to smartphones and tablets and is compatible with both Android and iOS operating systems. Patients using their smartphones download a software application in order to record, view, and transmit spirometry measurements by Wi-Fi or cellular connection to a cloud-based HIPAA-compliant server. Participants are able to view real-time inspiratory and expiratory flow data on the connected device as they are performing the spirometry. No additional sensors or hardware are required. In this study, the spirometer was synced either to an electronic tablet device provided by the study or to the patient’s smartphone. The study team monitored the spirometry measurements weekly via a password-secured web portal. This portal also allowed for the delivery of automated and ad hoc reminders to patients via the tablet device or smartphone. Patients were asked to perform a session, consisting of at least 3 efforts, once a week for 52 weeks. If the patient was noted to have an FEV₁ decline of ≥10% in absolute values compared with enrollment baseline FEV₁ on 3 consecutive (i.e. back-to-back) handheld measurements, a formal letter was sent to the patient’s primary provider, generally the primary oncologist who had assumed posttransplant care, informing them of these results. If the physician was within the Seattle Cancer Care Alliance network, the letter was sent through secure email. Patients were also informed via the tablet or the smartphone app. Laboratory PFTs were recommended at 3-month intervals as per standard of care for patients with chronic GVHD, or as needed per clinician determination. At the end of this study period, the participants were offered the option of continued monitoring if they still had active chronic GVHD or lung function changes that warranted continued monitoring. If BOS was diagnosed, home monitoring continued through the end of the study period with an option for extended monitoring.

Statistical analysis.

The primary endpoint of feasibility was defined as 1) adherence to study procedures and 2) high correlation and accuracy of the handheld measurements compared with laboratory measurements. Adherence was calculated as the proportion of valid weekly home measurements per 52 weeks on study, excluding weeks missed due to technical reasons. A rate of 75% measurements completed was considered high adherence. A valid week was defined as a week on study without reported technical issues that prevented the participant from performing spirometry. The best FEV₁ measurement of the participant’s 3 weekly spirometry efforts, as determined by the GoSpiro quality control algorithms, were used for reliability analyses. Paired handheld and laboratory spirometry measurements within 14 days of each other were analyzed using Pearson’s correlations. Accuracy of the handheld spirometry was determined by Bland-Altman analysis (implemented in the R package BlandAltmanLeh), in which the mean difference of the paired measurements indicates the bias, or deviation, of the experimental measurement (e.g. handheld FEV₁) from the gold standard (e.g. laboratory FEV₁). A bias of positive value indicates a higher value of laboratory PFT than handheld PFT. Assessment of individual FEV₁ trajectories were done by absolute values as well as by predicted values based on accepted reference standards.^{13, 14}

Results

Patient Cohort:

A total of 46 patients consented to participate in the study. The median time of enrollment after transplant was 11.9 months (range 3.2–20.2 months), with a median pretransplant FEV₁ at enrollment of 97% predicted (IQR 85–107%). At enrollment, 83% had a diagnosis of cGVHD, the majority with mild or moderate overall grade severity, and 9% with severe cGVHD. Eight individuals (17%) carried a pre-transplant diagnosis of chronic obstructive lung disease and/or asthma (Table 1).

Table 1.

Demographics and baseline characteristics of the patient cohort

BASELINE CHARACTERISTICS	N=46	%/Range/IQR
Age at Consent (Years)
Median (Range)	58.8	23.7, 74.5
Sex
Female	27	59%
Male	19	41%
Race / Ethnicity
African American	1	2%
Asian	2	4%
Native American	2	4%
Pacific Islander	2	4%
Hispanic	1	3%
Caucasian, non-Hispanic	38	83%
Smoking Status
Former	19	41%
Never	27	59%
COPD / Asthma History
No	38	83%
Yes	8	17%
HCT Indication
Acute lymphoblastic leukemia (ALL)	4	9%
Acute myelogenous leukemia (AML)	19	41%
Aplastic anemia	1	2%
Chronic myeloid leukemia (CML)	2	4%
Mixed phenotype acute leukemia (MPAL)	2	4%
Multiple myeloma (MM)	1	2%
Myelodysplastic syndrome (MDS)	13	26%
Myelofibrosis (MF)	1	2%
Myeloproliferative disease (MPD)	1	2%
Non Hodgkins lymphoma (NHL	1	2%
Donor Type
Related	10	22%
Unrelated	36	78%
Graft Source
Bone marrow	2	5%
Cord blood	3	7%
Peripheral blood	41	89%
Conditioning Regimen
Myeloablative	34	74%
Non-Myeloablative	12	26%
Acute GVHD Grade 2–4
No	9	20%
Yes	37	80%
AT ENROLLMENT
Indication for Spirometric Monitoring
cGVHD diagnosis <6 months	33	72%
D+80–100 at risk for cGVHD	4	8%
FEV1 decline	2
Prednisone >0.5mg/kg	2
PI Decision	9	20%
Posttransplant lung disease*	2
FEV1 decline of unclear etiology**	7
Time from HCT (Months)
Median (Range)	11.9	3.2, 20.2
Chronic GVHD Grade
None	6	13%
Mild	19	41%
Moderate	17	37%
Severe	4	9%
Pulmonary Function
FEV1 % predicted (IQR)	97	(85, 105)
FEV1/FVC (IQR)	72	(68, 77)

^*One patient had idiopathic pneumonia syndrome prior to D80; the other patient had cryptogenic organizing pneumonia prior to 1 year which resolved.

^**FEV₁ decline >10% of baseline in the context of mild chronic GVHD not requiring systemic therapy, or chronic GVHD diagnosed > 6 months prior to enrollment.

Adherence:

Of 46 patients enrolled into the study, 8 withdrew and 2 died prior to end of the study period, one at 15 weeks and the other at 11 weeks. Reasons for withdrawal are noted in Table 2. Thirty-six (78%) completed one year of monitoring. Seven continued with optional extended monitoring for an additional year. Adherence was defined as the proportion of 52 weeks patients performed handheld spirometry, prorating for weeks missing due to technical reasons or death. Overall adherence was 72% (IQR 47%−90%) for 45 patients, excluding one participant for whom deployment of the home spirometer never occurred. For 40 patients who provided longitudinal spirometry data beyond the first 8 weeks, the median number of valid weeks was 48 (IQR 42–52), and the median adherence was 70% (IQR 48–87%). Of the total number of expected weeks for 40 patients (accounting for withdrawals at 28 and 37 weeks, as well as 2 deaths), 196/1962 (10.1%), were considered invalid due to technical reasons. Technical reasons included inability to sync with the tablet or smartphone app, failure of data transmission due to lack of internet connectivity or cellular capability, or loss of the spirometry device. Feeling unwell or hospitalization was given as the reason for 65/604 (10.4%) of the remaining missed measurements not due to technical reasons. Thirty-six instances of 3 consecutive measurements showing FEV₁ decline of 10% or greater compared to enrollment FEV₁ in 19 patients triggered notification of the patient’s medical provider.

Table 2.

Reasons for withdrawal, n=8

REASON	N
Prolonged technical difficulties	2
Unable to communicate electronically	2
Inability to deploy home spirometer	1
Changed mind	1
Too busy	1

Correlation and Accuracy of Handheld Spirometry:

A total of 106 paired handheld and laboratory spirometry measurements collected within 14 days of each other over the 12-month study period were available for analysis, with a median time interval of 4.3 days (IQR 2.5–7.4 days). The correlation of FEV₁ across all time points of paired handheld and laboratory measurements was r = 0.974 (CI [0.962, 0.982]; p<0.0001), with a mean bias of 0.123 L (Lower limit −0.294 L, upper limit 0.541 L). That is, the average of laboratory FEV₁ measurements is 0.123 L higher than that of the paired handheld measurements. This is within the limits of acceptability for reproducibility of spirometry recommended by the American Thoracic Society.¹⁵ The lower limit and upper limit represent the −1.96SD and +1.96SD off the mean bias, respectively. Correlation of handheld FVC with laboratory FVC was r=0.961 (CI [0.943, 0.973]; p-value < 0.0001), with a mean bias of 0.369L (Lower limit −0.2525, upper limit 0.9897) (Figure 1). The correlations remained strong throughout the 12-month study period, although FVC had a consistently greater bias than FEV₁ at enrollment, 6 months, and 12 months (Supplemental Tables 1 and 2).

Figure 1.

Correlation and bias of handheld spirometry versus laboratory spirometry. A) Correlation of handheld FEV₁ (x-axis) with laboratory FEV₁ (y-axis). B) Bland-Altman plot of the mean difference of handheld FEV₁ against the laboratory FEV₁. C) Correlation of handheld FVC (x-axis) with laboratory FVC (y-axis). D) Bland-Altman plot of the mean difference of handheld FVC against the laboratory FVC).

Patient Outcomes:

Median time of last clinical follow-up from enrollment was 552 days (IQR 376–672 days). Ultimately 11 individuals were diagnosed with BOS and one was diagnosed with suspected BOS, representing 26% of the patients enrolled in the study. Of these 12 individuals, nine (23%) had home data available within 3 months of diagnosis that demonstrated an antecedent decline in FEV₁ (Figure 2).

Figure 2.

FEV₁ trajectory by handheld spirometry measurements in 12 individuals who were diagnosed with BOS or suspected BOS. Red triangles denote time of enrollment and beginning of weekly handheld spirometry monitoring. Each line represents the longitudinal FEV₁ of one individual. Black regression line represents the aggregate FEV₁ trajectory of all individuals centered at BOS diagnosis (time zero).

BOS diagnosis was preceded by documented respiratory viral infection or infectious pneumonia in 6 individuals. The median FEV₁ (pre-bronchodilator) by conventional laboratory PFT at BOS declaration was 60% of predicted (IQR 53–70%), with a median decline in FEV₁ by 12 percentage points from baseline enrollment measurements. Three individuals were diagnosed with BOS during the second year of monitoring, including one individual with concurrent restriction on PFTs. All 12 patients were treated with an inhaled corticosteroid with long-acting bronchodilator (LABA); 11 patients also received concurrent azithromycin, and montelukast (ie. Fluticasone, Azithromycin, Montelukast + LABA) which was standard of care at our institution at the time of the study. In the subsequent six months of follow-up after BOS diagnosis, 4/12 (25%) had FEV₁ improvement of ≥10%, 3/12 (25%) remained stable, and 5/12 (50%) had FEV₁ decline of ≥10% as ascertained by laboratory PFTs. Other noninfectious pulmonary diagnoses that were made during the study period included restrictive lung impairment attributed to sclerosis (n=2), and nonspecific lung function impairment (n=3).

Discussion

Early detection of asymptomatic early BOS after allogeneic HCT relies upon screening PFTs at regular intervals, which are recommended for high-risk patients.⁷ This study demonstrates that wireless handheld spirometry can be implemented for monitoring of patients with chronic GVHD who are at high risk for developing BOS after HCT. In this pilot study, home spirometry was assessed as a feasible screening tool to determine the need for further clinical evaluation. We show that handheld monitoring is feasible in this high risk population, with an acceptable adherence to weekly procedures that is in a similar range to other wireless home spirometry data in the HCT population¹¹ and is improved compared to a prior study in which post-transplant patients used a handheld spirometry requiring manual recording of measurements.⁹ While the predetermined threshold of 75% for high adherence was not reached, adherence in this study was better than adherence to inhaled medications in patients with chronic obstructive pulmonary disease and asthma.^{16, 17}

FEV₁, the most validated measurement for assessing lung function, correlated very closely when measured by a handheld spirometer compared with measurements obtained in a formal PFT lab where the patient is coached by a technician. The mild bias of approximately 120 ml lower is well within the limit of acceptable measurement variability and reproducibility standards per American Thoracic Society guidelines of 8% and <150 mL.¹⁵ Thus, the use of a handheld device could raise alarms at a slightly higher rate and result in more clinical PFTs, but an overly sensitive screening tool is preferable to missing a case of potential BOS. Moreover, the consistency of the FEV₁ bias over the study period illustrates that patients are able to provide reliable longitudinal measurements at home. The trajectory of FEV₁ is more important than a single measurement in determining if there is incipient lung disease,^{3, 18} hence a bias that is consistent over time should not impede the detection of true lung function decline. Handheld FVC is less accurate, but this may be expected as patients may have difficulty sustaining a prolonged exhalation maneuver without in-person coaching by a technician, an observation that is consistent with other studies of handheld spirometers.^{9, 19}

Importantly, handheld spirometry was able to detect antecedent FEV₁ changes that indicate early BOS. In this study population, nearly one-third of participants who completed six months or more of monitoring received a clinical diagnosis of BOS. Because this was a pilot study aimed at investigating feasibility, our sample included individuals who already had signs of mild FEV₁ decline on screening PFT that may have indicated early lung disease. The frequency of BOS diagnosis in our study represents a real-world scenario cohort of patients who are appropriate for an intensive monitoring strategy, namely those with recent onset chronic GVHD and/or patients with asymptomatic lung function changes of unclear etiology. Most of the patients who were ultimately diagnosed with BOS provided home data within 3 months before the diagnosis. The median FEV₁ at clinical diagnosis was 60%, still moderately impaired but better than the severity of FEV₁ impairment of patients enrolled in a recent treatment trial in which diagnosis of BOS was prompted mainly by evaluation of symptoms⁵. In a few cases, detection of FEV₁ decline by home spirometry led to clinical evaluation when FEV₁ impairment was mild to moderate. Treatment resulted in reversal or stabilization of FEV₁ in half of the participants, illustrating that lead time in the detection of FEV₁ changes can potentially translate to lung function improvement. In some instances, the recognition of BOS was unfortunately delayed by competing diagnoses, such as infectious pneumonia or restrictive lung disease, which may have confounded the interpretation of FEV₁ decline.

Our study was limited by the small sample size, and it was not designed or powered to test the efficacy of an early detection strategy. Because home spirometry is a new procedure for patients who do not already have known lung disease, we did not have clear assumptions of the likelihood of patients completing a full year of monitoring, or how realistic the adherence goal was. In addition, the deployment of a new technology inevitably presented challenges which deterred some participants from completing the study. Nonetheless, home spirometry monitoring is likely to be of benefit in high-risk patients with further refinements in implementation that encourages adherence to monitoring and efficient clinical evaluation. The optimal threshold and duration of FEV₁ decline that should trigger clinical action needs to be determined. This depends on a better understanding of the progression and etiology of lung function changes after HCT, which requires additional investigation.

This study is the first to utilize a home monitoring system for lung function that leverages wireless telecommunications capabilities in a post-HCT population at increased risk of BOS. Telemonitoring is particularly relevant in the context of the COVID-19 global pandemic, as concerns about spirometry as aerosol-generating procedures has curtailed access to PFT laboratories.^{20, 27} While home spirometry has long been standard of care for lung transplant recipients, the development of Bluetooth-enabled devices and “cloud computing” platforms offers the opportunity to bring this technology to other at-risk populations, including patients with chronic GVHD. The benefits of home monitoring on health behaviours and outcomes have been demonstrated in lung transplant and interstitial lung disease.^21–23 A monitoring system that uploads data in real-time from a mobile device with minimal need for patient engagement in the transmission of data will likely optimize adherence. Refinements of wireless technology, increased patient familiarity with smartphone capabilities, and universal internet access will also improve adherence to home monitoring.

In conclusion, handheld wireless spirometry is feasible in this population of patients, with acceptable patient adherence as well as technically valid spirometry measurements. Importantly, the FEV₁ changes that precede a diagnosis of BOS are detectable through home telemonitoring. Longitudinal monitoring that yields granular lung function data should inform the understanding of the clinical manifestations of BOS in this population and promote earlier diagnosis of BOS.

Figure 3.

Example of the lung function trajectory as measured by handheld spirometry and associated clinical events. A 56 year-old woman who received an HLA-matched unrelated donor peripheral blood stem cell transplant for acute lymphoblastic leukemia conditioned with myeloablative regimen was enrolled into the study for weekly spirometry monitoring 149 days after diagnosis of chronic GVHD of the skin, eyes, mouth and gastrointestinal tract. Patient reported an upper respiratory illness a month prior to enrollment. Blue dots represent individual handheld spirometry measurements; dotted line represents the moving average over 2 measurements. Orange dots represent clinical laboratory spirometry measurements. Entry into the study represented by (*). Black arrow indicates when an FEV₁ decline letter was sent to the patient’s primary transplant provider. Red arrow indicates time of clinical BOS diagnosis. This patient was initiated on treatment with FAM+LABA (fluticasone, azithromycin, montelukast, and long-acting beta-agonist) at the time of BOS diagnosis.

Highlights.

• Post-HCT patients at risk for BOS performed home spirometry with adequate adherence

• FEV₁ by home spirometry was accurate when compared with laboratory spirometry

• FEV₁ changes that precede a BOS diagnosis are detectable through home spirometry

• Weekly home spirometry facilitates early diagnosis of BOS

• Cloud-based spirometry telemonitoring may lead to improved outcomes