Diffusion kurtosis imaging detects subclinical white matter abnormalities in Phenylketonuria

Abstract

Objective: Phenylketonuria (PKU) is an autosomal recessive disorder whereby deficiencies in phenylalanine metabolism cause progressive neurological dysfunction. Managing PKU is challenging, with disease monitoring focussed on short-term phenylalanine control rather than measures of neuronal damage. Conventional imaging lacks sensitivity, however diffusion kurtosis imaging (DKI), a new MRI method may reveal subclinical white matter structural changes in PKU.

Methods: This cohort study involved adults with PKU recruited during routine clinical care. MRI, neurocognitive assessment and historical phenylalanine (Phe) levels were collected. A hypothesis-generating case study comparing diet-compliant and non-compliant siblings confirmed that DKI metrics are sensitive to dietary adherence and prompted a candidate metric (K_rad/K_FA ratio). We then tested this metric in a Replication cohort (PKU = 20; controls = 43).

Results: Both siblings scored outside the range of controls for all DKI-based metrics, with severe changes in the periventricular white matter and a gradient of severity toward the cortex. K_rad/K_FA provided clear separation by diagnosis in the Replication cohort (p < 0.001 in periventricular, deep and pericortical compartments). The ratio also correlated negatively with attention (r = -0.51 & -0.50, p < 0.05) and positively with 3-year mean Phe (r = 0.45 & 0.58, p < 0.01).

Conclusion: DKI reveals regionally-specific, progressive abnormalities of brain diffusion characteristics in PKU, even in the absence of conspicuous clinical signs or abnormalities on conventional MRI. A DKI-based marker derived from these scores (K_rad/K_FA ratio) was sensitive to cognitive impairment and PKU control over the medium term and may provide a meaningful subclinical biomarker of end-organ damage.

Keywords: DKI; Diffusion kurtosis imaging; PKU; Phenylketonuria; White matter pathology.

Fig. 1

Example white matter masks. Example masks derived from a control patient, delineating pericortical (red), deep (green) and periventricular white matter (blue) on T1-weighted images. The final masks were eroded to avoid partial volume effects. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2

Comparison of lifetime Phe concentrations and MRI findings for Patient X (“Compliant”, top row) and Patient Y (“Non-compliant”, bottom row). Phe levels are presented from initial diagnosis at birth to age 12, then around age 20 to the time of imaging. (C) Clinical radiological findings from conventional sequences demonstrate characteristic posterior white matter hyperintensities on T2-weighted (a, e & h, l) and FLAIR (panels b, f, g & I, m, n), together with abnormal diffusion weighted imaging findings (c, d & j, k). Changes were more severe for Patient Y however both show reduced ADC in both frontal and occipital horns of the lateral ventricles and in the parietal white matter.

Fig. 3

Metric images from a representative control subject, Patient X and patient Y demonstrating selected diffusion (fractional anisotropy (FA), top panel) and DKI (radial kurtosis (K_rad), middle panel; kurtosis fractional anisotropy (K_FA), bottom panel). While FA was unchanged in PKU siblings, pronounced differences were seen in the kurtosis metrics, with increased severity in Patient Y. Colour bars for each metric are representative of metric scale.

Fig. 4

Diffusion and diffusion kurtosis metrics for the sibling Case study and controls stratified by periventricular, deep and pericortical subdivisions of WM. (A, B) MD (mean diffusivity), MK (mean kurtosis); (C, D) FA (fractional anisotropy), K_FA (kurtosis fractional anisotropy); (E, F) D_ax (axial diffusivity) K_ax (axial kurtosis); (G, H) D_rad (radial diffusivity), K_rad (radial kurtosis). For A-H, the matched control cohort is represented as a floating box showing the mean and range with scores for Patient X (blue) and Y (red) superimposed. (I) K_rad/K_FA ratio in the periventricular, deep and subcortical WM for Patient X (blue), patient Y (red) and age-matched controls (black). (J) K_rad/K_FA ratio in cohort of patients with PKU (red) vs. controls (black). The K_rad/K_FA ratio for patients with PKU was higher than control at periventricular, deep and pericortical levels (****p < 0.001). Error bars indicate mean ± standard deviation.