Study Specific Limitations

In Chapter 2 and 3, VDV was measured from hyperpolarized 3He MRI. One of the primary goals of the initial 3He MRI study of healthy elderly subjects and subjects with COPD was to assess the reproducibility of 3He MRI measurements of the ADC; measurements of ventilation derived from these images was a secondary marker and therefore did not drive the data collection

scheme/MR pulse sequence design. The ADC is calculated from DWI, which have inherently thicker slices than 3He MR spin density images for the same breath-hold/imaging time, given that a diffusion and non-diffusion weighted image are captured for each slice of the DWI series. For the calculation of VDV the first of the interleaved diffusion-weighted pair of images was used. The thickness of the slices was 30mm, and therefore the ventilation defect area segmented was multiplied by 30mm, representing signal contribution from the entire slice. Given the slice thickness of the images used for image analysis, it is possible that the VDV values presented over-estimate the true ventilation defect volume, and thus should be taken as an estimate of the true VDV of the lung. Additionally, the use of DWI results in images with a lower signal-to- noise ratio than would be obtained had spin-density images been acquired for VDV measurements. Thus, acquiring a spin-density image set or perhaps a three-dimensional volumetric image set would allow for thinner slices and an improved signal-to-noise ratio. It will be important to evaluate thinner slices using spin density images, especially in a young healthy volunteer cohort to determine whether this group truly did not have any ventilation defects, as reported, or whether the thick slices and increased SNR perhaps mask the very small ventilation defects that have been reported in the literature in groups of similar age and health status.13,14 A second limitation present in Chapters 2 and 3 is that proton images acquired just prior to 3He MRI were not bag-matched to the 3He MRI breath-hold scans. In follow-up studies 1H scans were acquired following inhalation of 1.0 L of 4He/N2 at the same dosage used for 3He/N2 ventilation imaging. This approach allowed for rigid registration of proton and helium scans, allowing for clear delineation of the thoracic cavity border, and aided in ventilation defect segmentation. However, this was not the approach used in Chapter 2 and 3, when proton images were acquired during a breath-hold at peak-tidal volume. Therefore, the VDP measurement used in Chapter 4, and PVV measurement used in Chapter 5 and pioneered by Woodhouse and co- workers15, could not directly be assessed. In subjects with COPD, proton and helium images were registered, and despite these differences in image acquisition techniques, 3He MR images visually appeared to have good registration in most COPD subjects, while healthy elderly subjects had poorer registration on visual inspection making ventilation defect delineation difficult. Without a normalized measurement such as VDP or PVV, a true inter-subject

comparison of the extent of ventilation defects in the lung was not possible. In all future studies, this limitation was addressed by acquiring bag-matched 1H MRI.

One of the major limitations of the work presented in Chapters 2-5 is the small sample size, each with a study specific impact. In Chapter 2, the group of healthy volunteers had eight subjects, while the Stage II and III groups had nine and seven subjects respectively. The small sample size was selected to detect between group differences in ADC, and was calculated based on ADC and ADC standard deviation (SD) results published by Salerno and co-workers16. This sample size may have limited the power necessary to detect differences between VDV in the elderly healthy volunteer and stage II COPD subgroups, which in this study were not statistically significant. However, it is also possible that the reported VDV in subjects with stage II COPD is not due to COPD, but rather predominantly due to aging. Another possibility that would explain the large VDV SD is related to underlying 3He MRI phenotypes in the stage II COPD population; a subset of the stage II subjects were likely ADC dominant, having a VDV was due to age alone while another subset had elevated VDV due to both age and COPD (VDP or mixed phenotypes). In Chapter 3, the small sample size of the elderly subgroup may have resulted in an over- or under-estimation of true population ventilation defect volume; however, it is clear that ventilation defects do occur in the elderly, and these defects should not be confused with functional abnormalities related to COPD or other lung diseases. The small sample size in Chapter 4 of twenty subjects resulted in stage II and stage III COPD subjects forming the basis of the COPD group evaluated (one subject with stage I disease was included). The limited sample size evaluated, made up of a small group of mainly stage II and III COPD, indicates caution should be exercised in extrapolating the results presented in this chapter to the general COPD population, and more specifically, more advanced or less severe COPD groups (stage I/IV). In Chapter 5, the small sample size at baseline (n=7) and follow-up (n=4) may have restricted the detection of differences in functional measurements acquired longitudinally in the ipsilateral lung. Additionally, an increase in sample size in this study may have allowed for the detection of a significant difference in ADC between lungs at baseline (p=0.053 in seven subjects), and longitudinal change in ADC in the ipsilateral lung. Thus, increasing sample size

in future studies using 3He MRI will increase the power and likely lead to an increase in the number of differences detected between subject groups.

A further limitation in Chapter 3 is the lack of follow-up data for the younger subgroup. Because no defects were present at baseline in any subjects in this young healthy subgroup no follow-up data was acquired. Therefore, it is not known what the short-term reproducibility of this finding is. However, in two young healthy subjects imaged at multiple time points as part of a hardware and software development protocol no ventilation defects were observed in any of the scans collected. Despite this observation, it would be valuable to continue scanning younger healthy volunteers, and to evaluate the short-term reproducibility of ventilation images in these subjects given that previous reports of young healthy volunteers do report the presence of ventilation defects.13,14

One specific limitation present in Chapter 5 is the lack of other clinical parameters collected for study purposes. Subjective information regarding post-radiation respiratory difficulties was not obtained, and would have been useful in evaluating the impact of functional changes measured with 3He MRI on quality of life scores. Additionally, information related to treatment regime, specifically course of chemotherapy, smoking history and use of oxygen post-treatment, if collected, could give further information in relation to the functional changes reported in this study.