Respiratory profile of pediatric patients undergoing hematopoietic stem cell transplantation studied by oscillometry – A TRANSPIRE cohort

ABSTRACT

Importance

Oscillometry — a tidal‐breathing test with potential advantages over spirometry — is not well studied prior to hematopoietic stem cell transplantation (HSCT).

Objective

To determine whether patients awaiting HSCT have abnormal baseline oscillometry compared with controls.

Methods

The TRANSPIRE study (NCT04098445) is a National Institutes of Health‐sponsored, multicenter, prospective observational cohort of pediatric lung injury after HSCT. We performed tidal‐breathing oscillometry at baseline and during follow‐up in 63 children prior to HSCT across five pediatric centers and compared results to 80 control subjects. Measurements were made at various frequencies following the European Respiratory Society 2020 guidelines and included resistance (R5, R19, and R5–R19), reactance (X5 and X11), resonant frequency (Fres), and area under the reactance curve (AX). Linear regression compared TRANSPIRE baseline values to controls; mixed‐effects models assessed differences between TRANSPIRE subjects, controls, and published predicted values up to two years post‐HSCT.

Results

All mean oscillometric parameters were normal at baseline, according to normal reference values of Ducharme. However, mean lung function in TRANSPIRE subjects was significantly different from controls, with higher respiratory system resistance (height‐adjusted R5−R19, P < 0.001), and greater respiratory system stiffness and heterogeneity, with more negative height‐adjusted X5 (P < 0.001) and height‐adjusted X11 (P < 0.001), greater height‐adjusted AX (P < 0.001), and greater height‐adjusted Fres (P < 0.001). No significant changes were noted over time.

Interpretation

In a multicenter pediatric HSCT cohort, baseline oscillometry differed significantly from controls, showing heterogeneous increases in small‐airway resistance and respiratory‐system stiffness that may elevate risk for pulmonary complications.

These are results from a multi‐center collaboration between four large pediatric institutions, investigating baseline respiratory system function in a cohort of children undergoing hematopoietic stem cell transplantation. This manuscript proposes respiratory oscillometry as a novel way to prospectively follow lung function in these children.

INTRODUCTION

Hematopoietic stem cell transplantation (HSCT) is a potentially curative therapy for many oncologic, immunologic, hematologic, and metabolic disorders. ¹ , ² , ³ , ⁴ , ⁵ , ⁶ , ⁷ Pulmonary complications frequently occur during HSCT, with one study indicating that up to 74% of patients experience one or more such issues. ⁸ Consequently, a significant number of patients are at risk for respiratory dysfunction after transplant. ⁹ Pre‐transplant therapies for many of these disorders, including chemotherapy and radiation therapy, can also lead to pulmonary complications. While Shackleton et al. ¹⁰ reported normal respiratory impedance in a small subset of children undergoing autologous or allogeneic HSCT for childhood cancer prior to transplantation, Inaba et al. ¹¹ found abnormalities in lung volumes with air trapping in up to 40% of patients prior to HSCT for hematologic malignancies, and 20% had abnormal diffusing capacity. Additionally, patients are prone to further respiratory complications both in preparation for and after undergoing HSCT for their underlying illness. These include the pre‐transplant toxicities of induction and conditioning regimens; post‐transplant infectious complications (e.g., viral, bacterial, and fungal opportunistic infections); and non‐infectious post‐transplant complications such as idiopathic pneumonia syndrome, diffuse alveolar hemorrhage, pulmonary thrombotic microangiopathy, pulmonary fibrosis, and graft‐versus‐host disease, including bronchiolitis obliterans with or without organizing pneumonia. ⁹

A broad array of physiologic abnormalities can occur in these clinical contexts. For example, radiation therapy and pulmonary fibrosis lead to restrictive pulmonary and chest wall physiology; ¹² obliterative bronchiolitis leads to both restrictive and obstructive physiology. ¹³

The TRANSPIRE study is a multicenter 5‐year prospective cohort evaluation of respiratory complications in children and young adults undergoing allogeneic HSCT. This paper reports the findings of oscillometry pre‐HSCT, in order to set the baseline for subsequent post‐HSCT changes in respiratory function. We also report a subsequent longitudinal analysis of oscillometry in the same cohort for up to 2 years to date. The goals of this study were to describe the spectrum of physiologic abnormalities determined by oscillometry before HSCT, as well as during subsequent follow‐up in a large, diverse population of patients undergoing HSCT.

METHODS

Ethical approval

The study was approved by the Institutional Review Board of the Children's Hospital of Philadelphia (IRB 21‐019391). The parents of all subjects gave informed consent, and children aged >11 years gave informed consent.

We studied subjects (aged ≤ 24 years) from eight different institutions that began enrolling patients in September 2021. All participating institutions relied on a single central Institutional Review Board established at Cincinnati Children's Hospital Medical Center (CCHMC) for the review and oversight of this study. All patients received standard pulmonary evaluation, including routine pulmonary function tests, biological sample collection, and clinical data collection at set time‐points throughout their transplant course and in response to any significant clinical events. Additional imaging and pulmonary evaluations (oscillometry, multiple breath washout [MBW], home spirometry, and xenon magnetic resonance imaging) were obtained at selected participating centers. This report is limited to the analysis of oscillometry as a non‐invasive measure of lung function that is easily performed with minimal coaching. Evaluations were done pre‐transplant and post‐transplant at day 60, day 100, 6 months, and annually starting at 12 months post‐transplant.

Population and conditions leading to HSCT

Four centers performed respiratory oscillometry: CCHMC, Boston Children's Hospital, Children's Hospital of Philadelphia (CHOP), and the University of Minnesota Masonic Children's Hospital. The underlying conditions leading to HSCT were malignant (n = 28) and non‐malignant (n = 35). Non‐malignant conditions included benign hematologic disorders (n = 7), genetic/metabolic disorders (n = 2), immune deficiencies (n = 7), and bone marrow failure (n = 19). There were 27 female and 36 male TRANSPIRE subjects (Table 1). The baseline studies reported herein were performed within 30 days prior to HSCT.

TABLE 1.

Baseline characteristics of the study cohorts

Characteristics	Control (n = 80)	TRANSPIRE (n = 63)
Age (years)	9.77 (7.63–12.78)	13.20 (9.47–17.78)
Height (cm)	140.1 (125.15–155.15)	155.10 (135.35–167.40)
Sex (M/F)	43/37	36/27

Data are presented as median (interquartile range) or n of the measurements. M, male; F, female.

Normal control comparison group

TRANSPIRE subjects were compared to 80 control subjects (37 females and 43 males) studied at CHOP between 2019 and 2022. Control subject data originated from CHOP and were approved by the CHOP IRB (IRB 20‐018357_PERC). Participants were recruited by flyers distributed at CHOP primary care offices and from letters through the Pediatric Research Consortium (PeRC). Flyers and letters were given to families of children who were identified through PeRC as eligible based on age and health history. Screening for inclusion and exclusion criteria was performed by chart review and a screening questionnaire that was reviewed over the phone. We included males and females aged 4–18 years, with no history of physician‐diagnosed asthma and a negative asthma screen based on the modified Global Initiative for Asthma (GINA) Guidelines Questionnaire for Pediatrics (MGGQ‐P). ¹⁴ We excluded children who were born prematurely (< 37 weeks gestational age), who had viral symptoms within 1 month of the study visit, who had a previous diagnosis of asthma, who had a personal history of vaping or smoking, or who had chronic conditions such as chronic lung disease (e.g., cystic fibrosis, primary ciliary dyskinesia), airway anomaly, history of tracheostomy, neuromuscular disease, congenital heart disease, immunodeficiency, or chest wall deformity. A second pre‐visit screen was performed via phone approximately 48 h prior to all study visits to ensure that participants had not experienced any respiratory illness or symptoms within the four weeks leading up to the study date. Parents or guardians provided written informed consent, and children aged 7 years and older provided consent for study participation. All study visits for control subjects were completed at CHOP in Philadelphia, PA.

Oscillometry

Respiratory oscillometry was performed according to the American Thoracic Society (ATS) and European Respiratory Society (ERS) guidelines. ¹⁵ , ¹⁶ A common training tutorial was provided to all respiratory therapists performing the testing, and competency was verified at each center. To avoid physiologic changes induced by forced vital capacity maneuvers, oscillometry testing was performed prior to spirometry in subjects undergoing both tests on the same day. ¹⁵

All subjects were tested using a sinusoidal airwave oscillometer at each center (Tremoflo C100; THORASYS, Montreal, CA, USA), encompassing a frequency range of 5–37 Hz. Before each test, research staff performed calibration with a resistive load of approximately 15 cm H₂O·L⁻¹·s. During oscillometry measurement, children were carefully positioned and checked for proper seal at the mouthpiece; nares were occluded either manually or with a nose clip; and cheeks were compressed to avoid upper airway shunt compliance (Figure 1). Subsequently, between three and seven 30‐s trials were performed. All trials were screened for quality control measures, excluding those with artifacts, e.g., leak at the mouth, signal drift, or airway occlusion caused by vocalization. The target inter‐trial coefficient of variation for resistance at 5 Hz (R5) was less than 10% in children > 10 years of age and less than 15% in children < 10 years of age. Baseline pre‐transplant oscillometry was performed during a 30‐day time window prior to transplantation.

FIGURE 1.

Oscillometry set up.

Outcomes

Physiologic oscillometry parameters measured by spectral oscillometry (Figure 2) included tidal volume, respiratory rate, R5, resistance at 19 Hz (R19), reactance at 5 Hz (X5) and 11 Hz (X11); resonance frequency (Fres), and low frequency area under the reactance curve (AX) between 5Hz and Fres. Within each trial, the coherence between pressure and flow was measured, and the coefficient of variation between all trials per subject was calculated. The spectral oscillometry employed in this study measures resistance (R), or the pressure cost of flow, at various frequencies (e.g. 5–37 Hz), which provides information about airway size and tissue R. Spectral oscillometry also measures reactance (X), or the pressure cost of volume change and acceleration, at each frequency, which gives information about the elastic properties (stiffness) of airway and tissue, as well as inhomogeneity of ventilation in peripheral airways. ¹⁵ , ¹⁷ At frequencies less than Fres, X is negative, with more negative values indicating increasing stiffness. The various spectral oscillometry parameters described (Figure 2) may have differing abilities to detect the abnormalities that children may develop both before and after stem cell transplantation. ¹⁸ , ¹⁹ R, at relatively low frequencies (e.g., R5), is thought to reflect airway properties of both the central and peripheral airways, because if obstruction exists, all distal units have more time to fill. On the other hand, high frequency signals (e.g. R19) may be “filtered out” in the central airways if the peripheral units can only fill or empty at rates below the oscillatory frequency due to their local high airway R. Thus, if R5 represents total airway R and R19 represents central airway R, then (R5−R19) is thought to reflect peripheral airway R. Fres is a measure of overall respiratory system stiffness. X5 and X11, and AX between 5Hz and the Fres, are also reflections of how “stiff” the respiratory system is, i.e., the pressure cost of volume change. This may be due to intrinsic tissue properties, as in the case of Fres. In addition, heterogeneous peripheral small airway obstruction can cause abnormalities in X, because maldistribution of ventilation means that certain peripheral units do not participate in ventilation at high frequencies, and therefore the lung appears stiffer to the oscillatory signal.

FIGURE 2.

Illustration of the measured oscillometric parameters. R5, resistance at 5 Hz; R19, resistance at 19 Hz; X5, reactance at 5 Hz; AX, area under the reactance curve; Fres, resonance frequency.

Statistics

Results were presented, with corresponding P‐value, as the coefficient, or offset on the logarithmic scale, for the group (Control/TRANSPIRE) variable obtained from the regression. Associations between the groups (Control and TRANSPIRE) regarding oscillometry measurements were corrected for height using multiple linear regression. Separate multivariable linear models that included group (Control/TRANSPIRE) and height were fit for each oscillometry measurement on the log scale. Height was included in the model to account for its influence on lung function. A mixed effects model was used to compare the groups and assess changes in the different respiratory mechanics measurements over time and with respiratory events. A P‐value of < 0.05 was considered significant for comparing study subjects to controls and for assessing changes over time post‐transplant. For longitudinal analysis, at baseline and following HSCT, 422 oscillometry measurements were obtained between 5 and 37 Hz according to the 2020 ERS criteria at predefined time intervals. To eliminate growth as a confounding variable for any changes in respiratory function, the longitudinal results were reported as a change in z‐scores calculated from the normal values of Ducharme et al. ²⁰

RESULTS

We found oscillometry highly feasible in the great majority of subjects, even younger children. The overall success rate was 90%–95%, depending on the center. Outcome data are presented in Table 2 as the natural log‐transformed values as a function of height to facilitate statistical comparison between TRANSPIRE and CHOP control subjects. Data are presented in Figure 3 as non‐log transformed physiologic outcomes as a function of height to allow more apparent clinical comparison between TRANSPIRE subjects and CHOP controls.

TABLE 2.

Comparison of TRANSPIRE subjects to control subjects

Parameter	Coefficient (β)	95% CI	P‐value
R5 [cm H2O·L−1·s]	0.082	0.003, 0.161	0.042
R19 [cm H2O·L−1·s]	−0.058	−0.126, 0.010	0.095
R5−R19 [cm H2O·L−1·s]	0.140	0.096, 0.184	<0.001
X5 [cm H2O·L−1·s]	−0.261	−0.374, −0.149	<0.001
X11[cm H2O·L−1·s]	−0.322	−0.083, −0.038	<0.001
AX [cm H2O·L−1]	0.636	0.396, 0.876	<0.001
Fres [Hz]	0.226	0.123, 0.329	<0.001

Data are height‐adjusted and log‐transformed. The coefficient describes the average offset between the TRANSPIRE data and the CHOP control data on the logarithmic scale. A positive number indicates that the TRANSPIRE cohort has a higher, or more positive, value than the control group. A negative number indicates the TRANSPIRE cohort has a more negative value than the control group, which, in the case of X5 and X11, indicates greater respiratory system stiffness. See Figure 3. R5, resistance at 5 Hz; R19, resistance at 19 Hz; X5, reactance at 5 Hz; X11, reactance at 11 Hz; AX, area under the reactance curve; Fres, resonance frequency.

FIGURE 3.

Comparison of oscillometric parameters between TRANSPIRE (green, n = 63) and control (red, n = 80) subjects. Physiologic outcomes on the y‐axis are plotted as a function of height in cm on the x‐axis. (A) R5 (cm H₂O·L⁻¹·s) in TRANSPIRE (green) and control subjects (red). There was a significant but clinically minimal difference between TRANSPIRE and control subjects (Table 2). (B) R5−R19 (cm H₂O·L⁻¹·s) in TRANSPIRE subjects (green) and controls (red). TRANSPIRE subjects are significantly higher than controls (P < 0.001). (C) X5 (cm H₂O·L⁻¹·s) in TRANSPIRE subjects (green) and controls (red). TRANSPIRE subjects are significantly more negative than controls (P < 0.001). (D) AX (cm H₂O·L⁻¹) in TRANSPIRE subjects (green) and controls (red). TRANSPIRE subjects are significantly greater than controls (P < 0.001). (E) Fres (Hz) in TRANSPIRE subjects (green) and controls (red). TRANSPIRE subjects are significantly greater than controls (P < 0.001). R5, resistance at 5 Hz; R19, resistance at 19 Hz; X5, reactance at 5 Hz; AX, area under the reactance curve; Fres, resonance frequency.

Height dependence of oscillometry outcomes

Both CHOP control and TRANSPIRE subjects showed the expected changes in R with height (decreased e.g., Figure 3A,B); in X with height (increased to less negative, e.g., Figure 3C); in AX with height (decreased, e.g., Figure 3D; and in Fres with height (decreased, e.g., Figure 3E).

Comparison with CHOP controls

Small differences were observed between subjects and controls in height‐adjusted R5 (Figure 3A) (P = 0.042). However, height‐adjusted R5−R19 was substantially and significantly higher in TRANSPIRE subjects compared to controls (P < 0.001) (Figure 3B), indicating small airway obstruction. TRANSPIRE subjects as a group were also significantly different from CHOP control subjects in indicators of increased respiratory system stiffness and/or non‐homogeneous distribution of ventilation, with more negative height‐adjusted X5 (P < 0.001) (Figure 3C), and X11 (P < 0.001); higher height‐adjusted AX (P < 0.001) (Figure 3D), and height‐adjusted Fres (P < 0.001) (Figure 3E). There were no significant differences in baseline respiratory mechanics based on whether the diagnosis leading to transplantation was a malignant vs a non‐malignant condition (data not shown).

Longitudinal analysis of oscillometry following HSCT

Oscillometry z‐scores did not change significantly over the course of the study (Table 3, all P > 0.05), nor were there significant differences between subjects with normal or abnormal baseline clinical status, as defined by history, spirometry, diffusing capacity, or imaging studies (all P > 0.05), data not shown.

TABLE 3.

Longitudinal changes in z‐scores for oscillometry outcomes over a 2‐year period post transplantation

Parameter	n	Baseline	Day 100	6‐month	1‐year	2‐year	P‐value
R5‐z	418	ref.	−0.31 (0.27)	−0.33 (0.26)	−0.45 (0.29)	−0.16 (0.30)	0.540
R19‐z	422	ref.	−0.16 (0.29)	−0.22 (0.29)	−0.43 (0.32)	−0.44 (0.46)	0.690
R5−R19‐z	418	ref.	−0.06 (0.05)	−0.02 (0.05)	−0.03 (0.05)	0.07 (0.08)	0.467
X5‐z	418	ref.	0.06 (0.24)	−0.15 (0.24)	−0.38 (0.26)	−0.15 (0.38)	0.573
AX‐z	422	ref.	−0.23 (0.16)	−0.19 (0.16)	−0.14 (0.17)	−0.07 (0.25)	0.593
Fres‐z	412	ref.	−0.25 (0.14)	−0.20 (0.14)	−0.28 (0.15)	−0.01 (0.14)	0.222

Values are represented as a coefficient (standard error), which is the change in z‐score from baseline (ref.) estimated from a linear mixed model.

No significant change was seen over time in any of the measured variables. R5, resistance at 5 Hz; R19, resistance at 19 Hz; X5, reactance at 5 Hz; AX, area under the reactance curve; Fres, resonance frequency.

DISCUSSION

In the TRANSPIRE population, we found significant impairments as a group, compared to our control group, in R5−R19, X5, X11, Fres, and AX. These findings contrast with some earlier studies reporting normal findings in this population, as outlined above and as summarized by Sonneveld et al. ¹⁷ These impairments suggest increased respiratory system stiffness (elastance) or maldistribution of ventilation, e.g., frequency dependence of compliance, as indicated by more negative reactance and increased Fres (Figure 3). However, we did not find significant deterioration with age or duration of time since transplant.

Monitoring lung function after HSCT

A recent scoping review has detailed studies addressing physiologic outcomes in the HSCT population; however, guidelines for surveillance are lacking, and screening protocols vary widely across centers. ¹⁷ , ²¹ The methodologies (e.g., spirometry, MBW, and oscillometry) and duration of follow‐up have also varied, and large‐scale studies comparing pre‐ and post‐HSCT outcomes are limited in pediatric patients. Furthermore, while numerous studies have reported on pulmonary complications following HSCT, relatively few have examined their association with lung function prior to transplantation. Such knowledge is important to assess both the complications and their risk factors outlined above, as well as possible adverse effects due to HSCT itself. The major conclusions from this review were that oscillometry appears to be within normal range in the majority of patients at baseline and, while limited information is available, generally appears to remain normal following HSCT. However, oscillometry indices are frequently abnormal in patients with bronchiolitis obliterans syndrome (BOS). Furthermore, oscillometry changes may occur before a National Institutes of Health diagnosis of BOS is made. In this context, our study characterizes the physiology of respiratory system abnormalities in children undergoing HSCT for multiple indications and serves as baseline pre‐transplant data as well as longitudinal 2‐year follow‐up outcomes. This is of critical importance to the pediatric population as they age, as it has been reported that adult patients with abnormal lung function before transplant are likely to be at higher risk for developing post‐transplant pulmonary complications and mortality. ²²

Assessing lung function in pediatric patients can be challenging, especially in those below the age of 7 years. Spirometry requires maximal efforts that must be repeated to ensure reproducibility and accuracy, and many pediatric patients find such maneuvers difficult to perform, especially when they are not feeling well. For these reasons, there have been relatively few reports of associations between pre‐transplant lung function and post‐transplant complications. Two recent reviews have addressed these issues. A recent ATS guideline ²³ and a recent systematic review ²⁴ have summarized the literature in this field. Shanthikumar et al. ²³ stressed the utility of spirometry, static lung volumes, diffusing capacity of the lungs for carbon monoxide, high‐resolution computed tomography scanning, and MBW in centers equipped to perform these tests. Gower et al. ²⁴ have also reviewed pre‐transplant lung function assessment and post‐transplant surveillance. Interestingly, oscillometry is not specifically reviewed in either guideline because, at the time, there were insufficient data on its usefulness in the pediatric transplant population; this paper helps address this omission. Oscillometry and MBW techniques, which can be performed during tidal breathing, are often better tolerated than spirometry, but again must be repeated in the same sitting to ensure accuracy. ²⁵ Oscillometry records a number of physiologic parameters that provide information about the respiratory system's mechanical properties. ¹⁸ , ¹⁹ , ²⁶ In this study, we wished to examine the utility of oscillometry in distinguishing the pre‐transplant population from normal controls, and in following subsequent changes after HSCT.

Oscillometry outcomes in children undergoing HSCT

The success rate of oscillometry that we report compares favorably with that of spirometry, especially in younger children. In both our control and TRANSPIRE subjects, we found expected changes with growth, consistent with the reference values of Ducharme et al. ²⁰ R decreased as airways grew in size with increasing height (Figure 3A,B). Reactance, X, became less negative or less “stiff” as the respiratory system increased in volume with increasing height (Figure 3C). This decrease in stiffness was also reflected in a decrease in Fres with height (Figure 3E).

Although TRANSPIRE subjects as a whole were within the normal published values of ± 1.64 z‐scores (Table 3), the indicators of respiratory impairment that we have outlined compared to our control population suggest that patients undergoing HSCT may be more vulnerable to future respiratory abnormalities, and as a group may benefit from careful pulmonary follow‐up, since even modestly reduced z‐scores of pulmonary function can affect future pulmonary morbidity and rate of decline. ²⁷ , ²⁸ In this regard, it is encouraging that in our longitudinal analysis, lung function measured by oscillometry in our HSCT cohort remained stable compared to pre‐transplant baseline (Table 3). A 2‐year follow‐up period may not be long enough, however, to observe the kind of decline described by Bui et al. ²⁸ in children vulnerable to respiratory morbidity by virtue of having baseline lung function shifted even modestly below the population mean.

In conclusion, we have shown that a pediatric population undergoing HSCT has abnormal baseline respiratory function compared to a local control population, before transplant, revealed by respiratory oscillometry, indicating effects of underlying lung abnormalities due to the conditions requiring HSCT, effects of treatments received for these underlying conditions, or both. Many of these therapies—e.g., radiation, chemotherapy, and immune suppression with subsequent lung infection—are associated with the development of pulmonary fibrosis, which could manifest as increased lung stiffness or chronic airway disease, which can lead to airway obstruction and increased pulmonary R. Although both our control and TRANSPIRE populations’ results fell mostly within the “normal” range, the clear differences between the lung function in these two populations indicate that the TRANSPIRE population may be vulnerable to subsequent pulmonary morbidity. This study is the largest to date to use oscillometry for lung function measurements in a pediatric stem cell transplant population prior to transplantation. It suggests that this method is a promising way to easily measure lung function in these patients, and it will be useful for longitudinal monitoring of complications and their treatments related to stem cell transplantation.

CONFLICT OF INTEREST

Julian Allen serves as associate editor of Pediatric Investigation.