Longitudinal Assessment of Neuroradiologic Features in Wolfram Syndrome

BACKGROUND AND PURPOSE: Wolfram syndrome is a rare genetic disease with characteristic brain involvement. We reviewed the brain MR images of patients with Wolfram syndrome to determine the frequency and characteristics of common neuroradiologic ﬁ ndings. MATERIALS AND METHODS: We retrospectively reviewed the imaging data of patients with genetically-con ﬁ rmed Wolfram syndrome who had been recruited to the Washington University Wolfram Syndrome Research Clinic. These patients were evaluated between 2010 and 2019 with annual MRIs, along with other measures. MR images were assessed for clinical neuroradiologic signs at each individual ’ s ﬁ rst and last follow-up visits to characterize the frequency, rate of progression, and clinical correlations of these signs. RESULTS: We included 30 patients (13 males/17 females; average age at ﬁ rst visit, 14 years; average age at last visit, 19 years). The median duration of follow-up was 5years (range, 2 – 9years). The most common ﬁ ndings were an absent or diminished posterior pituitary bright spot ( ﬁ rst, 53%; last, 70%), T1/T2 pons signal abnormalities ( ﬁ rst, 53%; last, 67%), optic nerve atrophy ( ﬁ rst, 30%; last, 80%), white matter T2 hyperintensities ( ﬁ rst, 27%; last, 35%), and cerebellar atrophy ( ﬁ rst, 23%; last, 70%). CONCLUSIONS: Patients with Wolfram syndrome present characteristic neuroradiologic ﬁ ndings that involve the posterior pituitary gland, optic nerves, white matter, brain stem, and cerebellum. These abnormal ﬁ ndings appear at an early age and tend to increase in frequency with time. However, the neurologic signi ﬁ cance and neuropathologic mechanisms of each sign require more investiga-tion. Neuroradiologists should be aware of the pattern of these features in Wolfram syndrome.


CONCLUSIONS:
Patients with Wolfram syndrome present characteristic neuroradiologic findings that involve the posterior pituitary gland, optic nerves, white matter, brain stem, and cerebellum. These abnormal findings appear at an early age and tend to increase in frequency with time. However, the neurologic significance and neuropathologic mechanisms of each sign require more investigation. Neuroradiologists should be aware of the pattern of these features in Wolfram syndrome.
ABBREVIATIONS: DI ¼ diabetes insipidus; PPBS ¼ posterior pituitary bright spot W olfram syndrome is a rare genetic multisystem disease characterized by juvenile-onset diabetes mellitus, progressive optic atrophy, sensorineural hearing loss, and diabetes insipidus (DI). Two clinical variants of Wolfram syndrome result from wolframin ER transmembrane glycoprotein (WFS1) and CDGSH iron sulfur domain 2 (CISD2) (WFS2) mutations. 1,2 The pathophysiology of Wolfram syndrome is attributed to multiple etiologies, including increased endoplasmic reticulum stress, calcium homeostasis disturbances, and primary or downstream mitochondrial dysfunction. [3][4][5] Previous studies have shown that Wolfram syndrome is associated with structural brain changes and multiple neurologic symptoms, eg, bladder dysfunction, gait and balance abnormalities, and loss of smell and taste sensations. [6][7][8][9] The pathophysiologic mechanisms underlying these neurologic manifestations are an area of active research.
Convergent evidence derived from histopathologic and quantitative neuroimaging studies indicates that Wolfram syndromerelated structural brain changes comprise a combination of early developmental hypomyelination and late neurodegeneration. [10][11][12] Neuroradiologic findings previously reported in Wolfram syndrome include marked brain stem and cerebellar atrophy, optic nerve and optic tract atrophy, and an absent posterior pituitary bright spot on T1-weighted MR images. [13][14][15] However, this work has been limited by small cohorts of patients with advanced disease, did not include longitudinal follow-up, 13,14 and was typically conducted before the age of genetic testing for Wolfram syndrome. 15 To better define the range and progress of neuroradiologic signs in Wolfram syndrome, we evaluated the evolution of clinical neuroradiologic findings across time in a genetically-confirmed group of children, adolescents, and young adults with Wolfram syndrome.

Study Population
Data from participants in the Washington University Wolfram Syndrome Research Clinic were evaluated. Patients were enrolled between January 2010 and December 2019. Inclusion criteria were genetically-confirmed Wolfram syndrome (WFS1 mutations on both alleles that are known or suspected to be pathogenic), participant's awareness of the diagnosis, age younger than 30 years at the time of enrollment, and the ability to travel to St. Louis for the annual research clinic visits. We have previously reported aspects of the clinical and imaging data from subsets of this cohort. 7,8,10,11,[16][17][18]
One neuroimaging researcher with 3 years of experience (A.S.) and a board-certified neuroradiologist with .20 years of experience (J.S.S.) reviewed MRIs at each individual's first and last visit between 2010 and 2019. Because T2-weighted or FLAIR images were not collected in 2019, scans from the most recent visit before 2019 were used instead.
Each neuroradiologic sign was described categorically: "yes" if present, "no" if not present for all except the PPBS sign. The PPBS sign was categorized as "present," "diminished," or "absent," and the last 2 assignments were combined to indicate a negative PPBS sign. When the 2 readings were different, the final decision for which reading would be used in the analysis was made through consensus.

Statistical Analysis
Statistical analyses were conducted using R statistical and computing software (Version 3.6.3; http://www.r-project.org/). For the 6 most common findings, interrater reliability (k ) was calculated 20 and categorized as poor (,0.21), fair (0.21-0.40), moderate (0.41-0.60), substantial (0.61-0.80), or almost perfect (.0.80). 21 x 2 tests were used to compare the frequency distributions between the first and last visits, and Mann-Whitney U tests were used to compare the median number of neuroradiologic signs at first and last visits. The Spearman's r was used to determine whether age and the total number of neuroradiologic signs were linearly associated. The significance level was set at a P , .05.

Patient Characteristics
MRIs from 30 patients (13 males/17 females) with at least 2 MR imaging sessions were evaluated. Scans were between 2 and 9 years apart (median, 5 years). Demographics and clinical features at each session are in found in Table 1.

Neuroradiologic Findings
Interrater reliability ranged between fair and almost perfect: optic nerve atrophy (k ¼ 0.8), negative PPBS (k ¼ 0.79), cerebellar atrophy (k ¼ 0.83), pons signal change (k ¼ 0.42), white matter hyperintensity (k ¼ 0.58), and brain stem atrophy (k ¼ 0.37). See  Table 2 for the frequency of these signs in patients at first and last scanning. Because of the low interrater reliability in evaluating brain stem atrophy, we excluded this sign from subsequent statistical analyses. However, a previous quantitative analysis of brain stem atrophy in a subset of our cohort revealed a 27% difference in volume compared with controls and a mean estimated annual percentage rate of change of À0.85%. 10,18 At the first visit, the most common neuroradiologic findings were a negative PPBS sign and T1/T2 pons signal abnormalities, followed by optic nerve atrophy, white matter T2 hyperintensities, and cerebellar atrophy. In all cases, the signs observed in the first visit were either stable or increased on follow-up scans. Optic nerve and cerebellar atrophy showed the most progression across time with a 160%-200% increase in prevalence between the first and last visits (P , .001 for both). On the other hand, white matter T2 hyperintensities seemed to be a relatively stable sign, with only a 25% increase in prevalence between the first and last visits (P ¼ .51). Overall, patients had a median of 2 and 3 neuroradiologic signs at the first and the last visits, respectively (paired Wilcoxon test, P , .001). About 80% of patients had at least 1 neuroradiologic sign at the first visit, and 100% had at least 1 sign at the last visit (Fig 4A). Longer duration of follow-up was associated with a greater increase in the number of neuroradiologic signs (Spearman's r ¼ 0.47, P ¼ .008).

Relationship between Neuroradiologic Findings and Clinical Variables
Older age was associated with a higher number of neuroradiologic signs at both visits (Spearman's r : first ¼ 0.53, last ¼ 0.55, P , .01; Fig 4B). The median number of signs was not different between males and females at both follow-up visits (Mann-Whitney U test: first, P ¼ 0.79; last, P ¼ 0.77, respectively). There were no differences in sex ratios for each sign.
The positive/negative PPBS sign was not associated with the presence/absence of DI (x 2 ; first: P ¼ .27; last: P ¼ .32). The sensitivity and specificity of the PPBS sign in predicting the DI diagnosis was 75% and 60% for the first visit and 79% and 45% for the last visit. The absence of the PPBS sign was not predictive of a future DI diagnosis during the follow-up period. Finally, there was no difference in mean best-corrected visual acuity (Logarithm of the Minimum Angle of Resolution) between normal-appearing and radiologically identified optic nerve atrophy (Mann-Whitney U test; first: P ¼ .09, and last: P ¼ .48).

DISCUSSION
We describe the neuroradiology findings in a cohort of patients with well-characterized and genetically-confirmed Wolfram syndrome with longitudinal follow-up. The most common findings were a negative PPBS sign, T1/T2 pons signal abnormalities, optic nerve atrophy, white matter T2 hyperintensities, and cerebellar atrophy. The prevalence of these findings was higher in older patients and increased with time within individuals, suggesting that the accumulation of these signs reflects the evolution of neurodegenerative processes in this disease.
Overall, the most common neuroradiologic sign noticed in our cohort was a negative PPBS sign. Typically, the PPBS appears as a region of T1weighted hyperintensity in the posterior portion of the sella turcica. The nature of this signal has been controversial, and the exact substance responsible for the T1-shortening is not known. 22 Changes in the appearance of the PPBS have been linked to the functional state of the pituitary gland, the neurosecretory granules containing the antidiuretic hormone (vasopressin), and the phospholipid component of the vesicles. 23,24 A negative PPBS sign is observed in scans of healthy individuals in about 4% of the adult population 25 but is more prevalent in most cases of primary and secondary central DI. 26 One study has also shown that a preoperative negative PPBS sign was a predictor for postoperative DI development in cases of pituitary adenoma. 27 The PPBS sign was also previously described in a case report of Wolfram syndrome. 28 In our cohort, although the negative PPBS sign was twice as prevalent in the DI group compared with the non-DI group, the presence of this sign was not a significant predictor of the diagnosis because half of the individuals without DI also showed this sign.
The T1/T2 pons signal abnormalities observed in our cohort are rarely mentioned in the Wolfram syndrome imaging literature. Only 1 case report described similar pons signal changes in a patient with Wolfram syndrome. 19 The location of the pontine signal change appears to overlap with the pontine nuclei and pontocerebellar white matter fibers. 29 Because of its unique appearance and location in an area susceptible to imaging artifacts, this sign might have been previously missed and so under-reported in the literature. On the other hand, visual inspection of brain stem atrophy seems unreliable and less sensitive than quantitative analyses.  However, quantitative evaluation of brain stem atrophy would require advanced segmentation software not available in all clinical settings. 8,11 T1/T2 pons signal abnormalities may be more easily captured by visual inspection and appeared concurrently or sometimes preceded the development of brain stem atrophy detectable visually. Most interesting, although we did not include brain stem atrophy in the statistical analyses due to low interrater agreement, the cerebellum and brain stem atrophy observed in our cohort cooccurred at a high rate. Embryologically, these structures form together and are tightly linked. 30 In patients with Wolfram syndrome, postmortem histologic examination of both the brain stem and cerebellum showed evidence of neuronal loss and gliosis in most brain stem nuclei. [31][32][33] In the same studies, the cerebellum showed microscopic evidence of variable neuronal loss in the dentate nuclei and variable reduction of Purkinje cells of the cerebellum. [31][32][33] The visual system is severely affected in Wolfram syndrome, as shown by previous histologic and neuroimaging studies. 8,10,11,[31][32][33] The optic nerves and optic chiasm were grossly atrophic with a prominent perioptic CSF space. The optic nerve atrophy in Wolfram syndrome may be related to retinal dysfunction and degeneration. 34 This finding was evident as early as the first decade of life in our cohort. Histologic examination also reflects this dramatic change with multiple studies reporting optic nerve axonal degeneration associated with marked loss of myelinated axons and gliosis. [31][32][33] Furthermore, previous work from our group has also shown that this damage is progressive, involves both pregeniculate and postgeniculate regions of the visual pathway, and correlates with the decline in visual function. 35 Although the degree of visual system structure-function relationships is better evaluated via quantitative analyses, visual inspection of clinical scans captured the progressive nature of the visual pathway damage.
The presence of white matter lesions in Wolfram syndrome has been previously reported. 36 The white matter hyperintensities appeared as small round or oval lesions mainly in the frontal and parietal white matter with no confluent lesions. The frequency of this finding exceeds what might be considered incidental clinically insignificant T2 hyperintensities. 37 The radiologic appearance of these white matter T2 signal hyperintensities suggests possible demyelination, gliosis, or an inflammatory process. Previous histopathologic studies showed patchy demyelination and axonal degeneration in several white matter tracts in Wolfram syndrome, eg, the optic radiation and the pontocerebellar and corticopontine tracts. [31][32][33] In prior work by our group, we evaluated white matter microstructure in Wolfram syndrome using diffusion tensor imaging. 8,10,11 These quantitative analyses revealed that patients with Wolfram syndrome had widespread lower fractional anisotropy (reflecting decreased integrity of axon bundles) and higher radial diffusivity (reflecting impaired myelination) compared with age-equivalent controls. Given that this finding was observed as early as 10 years of age in our cohort and was also relatively stable across time, it may be a neurodevelopmental process that occurs in the early stages of the disease. 8,12 Finally, previous studies have reported some neuroradiologic features in patients with Wolfram syndrome that we did not observe in our cohort. These include cortical malformations, diffuse white matter leukoencephalopathy, 38 and high signal on proton-density and T2-weighted images in the substantia nigra. 39

CONCLUSIONS
Our study describes the primary neuroradiologic features observed on MR imaging in Wolfram syndrome. The most common findings involved the posterior pituitary gland, optic nerve and optic chiasm, cerebral white matter, brain stem, and cerebellum. One strength of our study is that it includes a large cohort of patients with genetically-confirmed Wolfram syndrome with longitudinal follow-up from an early age. We also showed the rate of progression of these findings and their relation to demographic variables. However, our study was also limited by the lack of an explicit control group and the different sequences used for both the first and last visits.
Neuroradiologists should be aware of these findings when reading MR imaging studies of patients with Wolfram syndrome. Future research could pursue the diagnostic and prognostic value of these signs when combined with quantitative neuroimaging data and the pathophysiologic processes underlying these signs.

ACKNOWLEDGMENTS
We thank all of the participants at the Washington University Wolfram Research Clinic and their families for their time and effort and the Washington University Wolfram Research Clinic study staff for their dedication. We also thank the former and current Washington University Wolfram Study Group members for their advice and support.