The application of advanced imaging techniques for the assessment of paediatric chest disease

Abstract

Introduction – Cystic fibrosis (CF) and primary ciliary dyskinesia (PCD) both result in chronic suppurative lung disease with significant resulting morbidity and early mortality. Many clinical and academic groups advocate biennial or even annual CT surveillance from as early as 2 years of age, but new therapies and increasing life expectancy lead to concerns over the use of repeated CT imaging. There are many recent studies showing promise of MRI for structural lung imaging MRI based measures of lung function. Both CF and PCD result in multisystem disease and whilst much of the morbidity results from lung disease, monitoring of extrathoracic disease is likely also relevant.

Aims and objectives – 1) To set up a clinically feasible, multisystem (lung, sinonasal and upper abdominal visceral) quantitative MRI examination for the investigation and follow up of CSLD 2) To evaluate novel imaging biomarkers of CF and PCD disease severity

Hypotheses – 1) Combined structural and quantitative MRI assessment of the thorax can provide comparable information to CT such that follow up imaging via CT could be replaced with MRI. 2) Quantitative MR measures of ventilation correlate with established clinical measures of ventilation (LCI and FEV1) and provide additional spatial information. 3) A multisystem MRI assessment can provide new extra-thoracic imaging biomarkers of CF and PCD disease severity whilst being better tolerated by patients than current multimodality imaging follow up.

Methods – People with CF or PCD referred for clinically indicated lung CT were prospectively recruited to undergo MR imaging of the lungs, liver and paranasal sinuses.

Structural lung imaging was optimised for speed of acquisition using T2 BLADE imaging, in axial and coronal plane, during breath holds rather than more conventional respiratory triggering. Images were scored by two observers using the Eichinger scoring system and compared to CT structural scores using the CFCT scoring system.

Lung T1 mapping was performed via free breathing IR-HASTE and T1 and T2 mapping performed via breath hold ufbSSFP imaging.

Functional lung imaging was performed via pre and post hyperoxygenation ufbSSFP T1 mapping, free breathing dynamic oxygen enhanced IR-HASTE imaging (OE-MRI) and non-contrast ufbSSFP-based matrix pencil decomposition imaging of ventilation and pulmonary perfusion.

Lung T1 maps included the superior portion of the liver enabling simultaneous liver T1 mapping.

A multiparametric paranasal sinus protocol was devised containing structural (T1 and T2 TSE), susceptibility and diffusion weighted sequences for the calculation of sinus volume, mucus volume and mucosal volume, presence or absence of artefact associated with infective micro-organisms and calculation of mucus and mucosal diffusion.

Participant tolerability of MR imaging assessed via a bespoke questionnaire, completed before and after both CT and MR imaging.

Multiple breath wash-out testing was performed on the day of the MRI and spirometry, antibiotic usage, abdominal ultrasound and sheer wave elastography collected retrospectively from the electronic patient record.

Results – 22 participants were recruited, all of whom completed the hour-long MRI protocol. The median age was 14 years (range 6 – 35).

2-plane structural lung imaging was acquired in a total of 2 minutes 4 seconds with only a single participant reporting difficulties with the required breath holds. Interclass Correlation Coefficients of interobserver variability in MRI scores were comparable to CT (0.877-0.965 compared to 0.877-0.989 respectively) suggesting good image quality with strong correlation between MR and CT component scores (bronchiectasis/bronchial wall thickening r=0.828,p<0.001; mucus plugging r=0.812, p<0.001; parenchymal score r=0.564 – 0.729, p<0.001 – 0.006).

Median lung T1 did not correlate with clinical markers of disease severity, but median lung T2 demonstrated strong correlation with CT bronchial wall thickening (r=-0.655, p=0.001) and LCI2.5 (r=-0.540, p=0.046), most likely representing a surrogate of pulmonary perfusion (most pulmonary T2 signal likely originates from the pulmonary blood pool).

Significant ufbSSFP enhancement was demonstrated post hyperoxygenation, but the degree of enhancement did not correlate significantly with clinical measures of disease severity. There was, however, very strong correlations between matrix pencil decomposition ventilation fraction and LCI2.5 (r=0.831, p=0.001) and CFCT scores (r= up to 0.731, p=<0.001).

Significant correlation was also demonstrated between measures of ventilation heterogeneity (oxygen wash-out time skew and kurtosis) and both LCI2.5 (r=0.591, p=0.013) and CFCT component scores (r= up to 0.718, p<0.001).

Liver T1 values did not correlate with evidence of liver disease on liver function tests or ultrasound imaging, but interpretation was severely limited by the very small number of recruits with CF liver disease.

Sinus imaging was the last part of the protocol with failed analysis in only one patient from too much motion (a 6 year old). Association was demonstrated between exacerbation frequency and opacification of maxillary sinuses by mucusa (p=0.074), between CT hyperinflation score and increasing levels of mucus susceptibility artefact (0=0.028), between exacerbation frequency, CT bronchial wall thickening and mucus plugging and increased sinus mucus diffusion (r=0.581, p=0.048, r=0.744, p=0.006 and r=0.633, p=0.019 respectively) and between CT hyperinflation, bronchiectasis and bronchial wall thickening scores and increased sinus mucosal diffusion (r=-0.847, p=0.016; r=-0.542, p=0.017 and r=-0.427, p=0.069 respectively).

A third of recruits stated that they would opt for MR imaging over CT imaging in the future and whilst 41% reported difficulties staying still for the MRI, respiratory image post processing was successful in all participants, with no parts of the MRI studies repeated.

Conclusion – Multisystem lung, liver and sinus MRI is feasible, well tolerated by people with CF or PCD, down to the age of 6 years, and provides gross structural imaging of sufficient quality to replace CT for lung imaging surveillance. Furthermore, the addition of functional lung imaging provides quantitative outputs which correlate well with clinically established lung function tests with the benefit of spatially localised lung function and additional quantitative measures of relevant extrapulmonary disease, within a single ionising radiation free examination. The data from this study have supported funding for future work addressing short, medium and long-term repeatability and longitudinal trends both in times of disease stability and over the course of an infective exacerbation.