Quantitative Structural MRI for Neurocognitive Disorders

Alzheimer disease dementia, frontotemporal dementia, and traumatic brain injury can be differentiated with quantitative MRI.

Cyrus A. Raji, MD, PhD
Department of Neurology
Washington University, St. Louis

Somayeh Meysami, MD
Department of Neurology
David Geffen School of Medicine at UCLA
Los Angeles, CA

Mario F. Mendez, MD, PhD
Department of Neurology
David Geffen School of Medicine at UCLA
Los Angeles, CA

Source
Practical Neurology
practicalneurology.com

Neurocognitive disorders are prevalent clinical problems. Among the most common are Alzheimer disease (AD) dementia and frontotemporal dementia (FTD), respectively the first and second most common neurodegenerative causes of dementia worldwide. Collectively, these conditions affect almost 6 million persons in the US alone.^1,2 Traumatic brain injury (TBI) is also a common problem affecting 2.5 million persons annually.³ TBI increases the risk for later dementia and chronic traumatic encephalopathy (CTE).⁴ Because all of these disorders can present with cognitive impairment, the ability to apply improved diagnostic tools is crucial for patient care. The current clinical standard of care in neuroradiology involves visual inspection of brain MRI. When attempting to detect the earliest volume loss related to brain atrophy, however, visual inspection alone lacks sensitivity compared with automated methods.⁵ The purpose of this review is to present a brief outline of available quantitative analytic methods that can be applied to brain MRI to detect volume loss from atrophy in AD dementia; behavioral variant FTD (bvFTD), which is the most common form of FTD; and TBI.

Although there are other imaging methods for the evaluation of neurocognitive disorders, there are several reasons to emphasize the volumetric quantification of brain MRI. First, patient access is maximized compared with neuronuclear methods such as positron emission tomography (PET). Second, the cost of an MRI is 60% lower than glucose metabolic imaging with fluorodeoxyglucose PET (FDG-PET) and 80% reduced in cost compared to an amyloid PET scan.⁶ Third, quantification of hippocampal volume loss on MRI is recognized as a key metric of neurodegeneration in the amyloid, tau, and neurodegeneration (ATN) framework.⁷ It is also possible that developing blood-based biomarkers for amyloid and tau may obviate the clinical use of amyloid or tau PET imaging in the future (See Blood Tests for Alzheimer Disease in this issue).⁸

Automated volumetrics of brain MRI is not the only, or even theoretically the most sensitive, method for defining abnormalities seen in AD, bvFTD, or TBI. Other MRI sequences exist and can be used clinically, including perfusion MRI with arterial spin labeling (ASL), diffusion tensor imaging (DTI), MR spectroscopy, and quantification and spatial characterization of fluid-attenuated inversion recovery (FLAIR) hyperintensities and microbleeds on susceptibility sensitive sequences. Those methods are outside the scope of this review. Here, we focus on automated volumetric software that can compute brain volumes and cortical thickness from MRI, including several programs approved by the Food and Drug Administration (FDA) for clinical purposes or otherwise used for research (Table).

Methodology and Protocols

Broadly, volumetric MRI quantification uses the relative signal intensities of the different brain tissue classes to segment gray matter (GM, intermediate intensity), white matter (WM, brightest intensity), and cerebrospinal fluid (CSF, darkest intensity). This is done on T1-weighted brain MRI because the differentiation between these tissue classes allows for optimal anatomic resolution. An atlas is then applied to quantify specific brain regions—mainly GM structures including hippocampi and other regions of interest implicated in neurodegenerative diseases or TBI. These brain structures are then compared to a normal database to compute standard deviations and percentile comparisons that assist neuroradiologists, neurologists, and other physicians in interpreting and understanding early and sensitive markers of atrophy. It is important to know the characteristics of the normal database being used because it is the basis for all subsequent comparisons and therefore needs to accurately reflect the demographics for the patient population of interest. Images consist of pixels but to obtain volumes of a given brain structure, 3-dimensional pixels, termed voxels, are required. Thus, when ordering brain MRI for volumetric quantification, the order must specify a 3D or volumetric T1 in order to obtain images that can be analyzed by software programs to compute volumes. The 2 most common types of volumetric T1-weighted scans are named based on their scanner vendor: magnetization prepared—rapid gradient echo (MPRAGE) for Siemens scanners and spoiled gradient echo (SPGR) for General Electric scanners. These terms may be encountered when ordering scans and consulting with MR technologists and neuroradiologists. The average length of time for scan acquisition of these specific types of sequences is between 3 to 5 minutes. No intravenous contrast is required and contrast, if present, can confound the quality of segmentation and subsequent analysis. The most common field strengths for scanners are 1.5 T and 3 T. Although T1 3D scans can be acquired with either field strength, it is important that follow-up scans be acquired with the same field strength and ideally the same vendor. Thus, if a T1 MPRAGE at 3 T is acquired from an individual, their follow-up scans should also be T1 MPRAGE on a 3 T scanner.

Figure. Longitudinal Volumetric Quantification With NeuroReader From Brain MRI in a Person with Alzheimer Disease Dementia. This figure shows reduced hippocampal volume over the course of 6 years as seen on progressive volumetric analysis (top row) and also coronal MRI evaluations (bottom row; arrows). Progressive volume loss in the mesial temporal lobe on MRI is a characteristic imaging feature of AD.

Findings in AD

AD dementia is a progressive disorder leading to cognitive decline in multiple domains including episodic memory, language, and visuospatial skills.⁹ Volumetric quantification of brain MRI in AD dementia most commonly includes the temporal and parietal lobes as well as the hippocampus.¹⁰ The magnitude of this change varies with the type of FDA-cleared software used for analysis—a 5th percentile and lower cutoff to distinguish normal from abnormal with 1 software package (NeuroQuant) and a 25th percentile or lower for the other available software package (NeuroReader).^11,12 Whereas NeuroQuant relies on multiple normal databases drawn from a combination of research databases and clinical samples with an age range of 3 to 100 years,¹³ the Neuroreader database is drawn currently from Alzheimer’s Disease Neuroimaging Initiative (ADNI) with an age range of 60 to 90 years. It is possible, therefore, for larger percentiles to be observed for these brain volumes in symptomatic individuals so vendor cutoffs, while helpful, must be applied in context of the clinical situation. Ventricular enlargement from volume loss is also a feature of AD and manifests typically as volumes of more than the 75th percentile of normal because increased CSF is detected as a result of lost GM and WM. A separate FDA-cleared software program, Icometrix, has recently shown high diagnostic performance when compared with the research program, Freesurfer.¹³ An additional program (Quantib), also cleared by the FDA, uses normative data from the population-based Rotterdam Study.¹⁴ When using volumetric quantification to support a diagnosis of AD, it is important to note that hippocampal volume loss alone is not always entirely specific for AD. Hippocampal volume loss can be observed in other disorders, including mesial temporal sclerosis¹⁵ and hippocampal sclerosis.¹⁶ Also, hippocampal volume loss alone may not distinguish early-onset AD from late-onset AD or bvFTD. In future work, quantification of specific hippocampal subfields¹⁷ may bridge these diagnostic gaps, but this is not currently available in FDA-cleared software. With these limitations outlined, longitudinal detection of volume loss is suggestive of AD particularly if detected in the hippocampus (Figure). When evaluating a person for AD dementia with volumetric quantification, annual measurements to obtain at least 2 time points can be supportive of the diagnosis.

Findings in BvFTD

BvFTD is notable for progressive changes in personality and behavior, including the development of apathy, disinhibition, reduced empathy, compulsive-repetitive behaviors, and dietary changes such as a carbohydrate craving. The misdiagnosis of bvFTD is common, with up to 60% of persons diagnosed with bvFTD by community specialists having other conditions on further evaluation by specialists.¹⁸ Early in bvFTD, visual evaluations of neuroimaging can appear completely normal, making early diagnosis especially challenging. Volumetric quantification findings are distributed in the frontal and temporal lobes, and a meta-analysis of brain imaging studies of bvFTD demonstrates a network of affected regions, including the anterior medial frontal, gyrus rectus, and superior frontal gyrus in addition to anterior cingulate gyrus, thalamus, and anterior insula.¹⁹ With disease progression there is increased atrophy, specifically in the posterior frontal lobes.²⁰ The studies that showed these findings used research software such as Freesurfer and voxel-based morphometry. Still, like AD, longitudinal quantitative volumetric evaluations are particularly useful in bvFTD because progressive atrophy is in keeping with a neurodegenerative disease. Unlike in AD dementia, persons with bvFTD are negative for amyloid PET. The application of FDA-cleared MRI software in peer-reviewed studies of bvFTD is comparatively limited. Thus, refinement of MR quantitative approaches will continue to be important and future studies should focus on the delineation of bvFTD from AD as well as subtypes of these neurodegenerative dementias.

Findings in TBI

Brain atrophy may be present in TBI with visually apparent volume loss and encephalomalacia in moderate and severe cases. With volumetric quantification, detection of atrophy is improved over visual inspection. Using NeuroQuant software, a study of 20 people with TBI, of whom all but 1 had mild TBI, atrophy was found in 50% of cases with findings as subtle as asymmetric right hippocampal volume loss in a participant age 35. By contrast, visual radiologic interpretations of brain MRI from this cohort found atrophy or ventricular enlargement related to atrophy in only 10%.²¹ This study was limited by its cross-sectional design. A follow-up study of 24 persons, of whom all but 1 had mild TBI, obtained brain MRI at baseline, 6 months, and 12 months and defined atrophy quantitatively as ventricular asymmetry of more than 25% difference compared with the normal database drawn from the Alzheimer’s Disease Neuroimaging Initiative. Quantitative volumetric analysis detected abnormal asymmetry in 83.3% and progressive atrophy in 70% of those with TBIs. In contrast, visual interpretations detected this subtle asymmetry and progressive atrophy in 0% of cases.²² These studies, while useful in demonstrating the sensitivity of volumetric quantification compared to standard neuroradiologic interpretations, was limited by the relatively small number (n=11) of regions of interest measured in volume. A newer FDA-cleared volumetric quantification software, Neuroreader, measures 45 unique brain structures. These areas include brainstem structures (eg, the midbrain and the ventral diencephalon) known to be injured in TBI and also implicated in the pathology of chronic traumatic encephalopathy.²³ A case report of a man age 51 who sustained an estimated 900 impacts to the head while playing high school football showed focal midbrain and ventral diencephalon atrophy. Although this was only 1 case, the work benefited from longitudinal MRI quantification showing progressive volume loss in these 2 regions.²⁴ The hippocampus, by contrast, showed no progressive atrophy over time, making AD dementia unlikely. We replicated these findings in a larger cohort of 40 persons with cognitive impairment and a history of TBI with the most common mechanism being motor vehicle collisions. In this study, Neuroreader quantification showed the ventral diencephalon to be the most atrophic region, whereas the hippocampus was the least atrophic.²⁵ Thus, volumetric quantification can identify novel patterns of volume loss beyond what would be expected for neurodegenerative diseases such as AD. Of the other FDA-cleared volumetric quantification tools, no current peer-reviewed literature is known regarding findings in TBI. However, given the similarity between these programs in terms of the ability to quantify brain structure, it is reasonable that these tools should be able to demonstrate similar results.

Conclusions

The use of volumetric MR quantification for atrophy detection in common neurodegenerative diseases and brain trauma is readily available for clinical application. From the perspective of a neurologist or other referring physician, success entails correct ordering of the 3D T1 protocol. From the neuroradiologist standpoint, confirmation of this correct protocol and availability of a software program that can quantify brain volumes is essential. Additionally, neuroradiologists and neurologists should closely collaborate on both the results of these analyses and how they best fit with the clinical situation of the patient. Doing so will ensure that the patient understands the meaning and limitations of this data and its integration into their care. Although not in the scope of this review, it is important to add that the application of MR volumetric quantification extends to other brain disorders. This includes mesial temporal sclerosis in epilepsy and brain atrophy in multiple sclerosis. As the ability to measure volumes of smaller brain structures is refined for use in the clinic, additional future possibilities are the input of these volumes into artificial intelligence algorithms such as deep learning. This approach holds much promise for improving disease detection and the diagnostic delineation of neurocognitive brain disorders.

References

1. 2020 Alzheimer’s disease facts and figures. Alzheimers Dement. 2020;10.1002/alz.12068. doi:10.1002/alz.12068

2. Onyike CU, Diehl-Schmid J. The epidemiology of frontotemporal dementia. Int Rev Psychiatry. 2013;25(2):130-137.

3. Centers for Disease Control and Prevention. Traumatic Brain Injury and Concussion. Published March 29, 2019. Accessed October 10, 2020. https://www.cdc.gov/traumaticbraininjury/data/tbi-ed-visits.html

4. Mendez MF, Paholpak P, Lin A, Zhang JY, Teng E. Prevalence of traumatic brain injury in early versus late-onset Alzheimer’s disease. J Alzheimers Dis. 2015;47(4):985-993.

5. Bigler ED. Traumatic brain injury, neuroimaging, and neurodegeneration. Front Hum Neurosci. 2013;7:395.

6. Desikan RS, Rafii MS, Brewer JB, Hess CP. An expanded role for neuroimaging in the evaluation of memory impairment. AJNR Am J Neuroradiol. 2013;34(11):2075-2082.

7. Jack CR Jr, Bennett DA, Blennow K, et al. NIA-AA research framework: toward a biological definition of Alzheimer’s disease. Alzheimers Dement. 2018;14(4):535-562.

8. Palmqvist S, Janelidze S, Quiroz YT, et al. Discriminative accuracy of plasma phospho-tau217 for Alzheimer disease vs other neurodegenerative disorders. JAMA. 2020;324(8):772-781.

9. Mendez MF. The accurate diagnosis of early-onset dementia. Int J Psychiatry Med. 2006;36(4):401-412.

10. Raji CA, Lopez OL, Kuller LH, Carmichael OT, Becker JT. Age, Alzheimer disease, and brain structure. Neurology. 2009;73(22):1899-1905.

11. Brewer JB. Fully-automated volumetric MRI with normative ranges: translation to clinical practice. Behav Neurol. 2009;21(1):21-28.

12. Ahdidan J, Raji CA, DeYoe EA, et al. Quantitative neuroimaging software for clinical assessment of hippocampal volumes on MR imaging. J Alzheimers Dis. 2016;49(3):723-732.

13. Struyfs H, Sima DM, Wittens M, et al. Automated MRI volumetry as a diagnostic tool for Alzheimer’s disease: Validation of icobrain dm. Neuroimage Clin. 2020;26:102243. doi:10.1016/j.nicl.2020.102243

14. Hofman A, Brusselle GG, Darwish Murad S, et al. The Rotterdam study: 2016 objectives and design update. Eur J Epidemiol. 2015;30(8):661-708.

15. Mettenburg JM, Branstetter BF, Wiley CA, Lee P, Richardson RM. Improved detection of subtle mesial temporal sclerosis: validation of a commercially available software for automated segmentation of hippocampal volume. AJNR Am J Neuroradiol. 2019;40(3):440-445.

16. Zarow C, Wang L, Chui HC, Weiner MW, Csernansky JG. MRI shows more severe hippocampal atrophy and shape deformation in hippocampal sclerosis than in Alzheimer’s disease. Int J Alzheimers Dis. 2011;2011:483972. doi:10.4061/2011/483972

17. Manjón JV, Coupé P. volBrain: an online MRI brain volumetry system. Front Neuroinform. 2016;10:30.

18. Shinagawa S, Catindig JA, Block NR, Miller BL, Rankin KP. When a little knowledge can be dangerous: false-positive diagnosis of behavioral variant frontotemporal dementia among community clinicians. Dement Geriatr Cogn Disord. 2016;41(1-2):99-108.

19. McCarthy J, Collins DL, Ducharme S. Morphometric MRI as a diagnostic biomarker of frontotemporal dementia: A systematic review to determine clinical applicability. Neuroimage Clin. 2018;20:685-696.

20. Rohrer JD. Structural brain imaging in frontotemporal dementia. Biochim Biophys Acta. 2012;1822(3):325-332.

21. Ross DE, Ochs AL, Seabaugh JM, Shrader CR; Alzheimer’s Disease Neuroimaging Initiative. Man versus machine: comparison of radiologists’ interpretations and NeuroQuant volumetric analyses of brain MRIs in patients with traumatic brain injury. J Neuropsychiatry Clin Neurosci. 2013;25(1):32-39.

22. Ross DE, Ochs AL, DeSmit ME, Seabaugh JM, Havranek MD; Alzheimer’s Disease Neuroimaging Initiative. Man versus machine part 2: comparison of radiologists’ interpretations and NeuroQuant measures of brain asymmetry and progressive atrophy in patients with traumatic brain injury. J Neuropsychiatry Clin Neurosci. 2015;27(2):147-152.

23. Barrio JR, Small GW, Wong KP, et al. In vivo characterization of chronic traumatic encephalopathy using [F-18]FDDNP PET brain imaging [published correction appears in Proc Natl Acad Sci U S A. 2015 Jun 2;112(22):E2981]. Proc Natl Acad Sci U S A. 2015;112(16):E2039-E2047. doi:10.1073/pnas.140995211

24. Raji CA, Merrill DA, Barrio JR, Omalu B, Small GW. Progressive focal gray matter volume loss in a former high school football player: a possible magnetic resonance imaging volumetric signature for chronic traumatic encephalopathy. Am J Geriatr Psychiatry. 2016;24(10):784-790.

25. Meysami S, Raji CA, Merrill DA, Porter VR, Mendez MF. MRI volumetric quantification in persons with a history of traumatic brain injury and cognitive impairment. J Alzheimers Dis. 2019;72(1):293-300.