Early Marker for Alzheimer’s Disease: Hippocampus T1rho (T_1ρ) Estimation

Abstract

Purpose

To evaluate the T1rho (T_1ρ) MRI relaxation time in hippocampus in the brain of Alzheimer’s disease (AD), mild cognitive impairment (MCI), and control, and to determine whether the T_1ρ shows any significant difference between these cohorts.

Materials and Methods

With informed consent, AD (n = 49), MCI (n = 48), and age-matched control (n = 31) underwent T_1ρ MRI on a Siemens 1.5T Scanner. T_1ρ values were automatically calculated from the left and right hippocampus region using in-house developed software. Bonferroni post-hoc multiple comparisons was performed to compare the T_1ρ value among the different cohorts.

Results

Significantly higher T_1ρ values were observed both in AD (P = 0.000) and MCI (P = 0.037) cohorts compared to control; also, the T_1ρ in AD was significantly high over (P = 0.032) MCI. Hippocampus T_1ρ was 13% greater in the AD patients than control, while in MCI it was 7% greater than control. Hippocampus T_1ρ in AD patients was 6% greater than MCI.

Conclusion

Higher hippocampus T_1ρ values in the AD patients might be associated with the increased plaques burden. A follow-up study would help to determine the efficacy of T_1ρ values as a predictor of developing AD in the control and MCI individuals.

Alzheimer’s disease (AD) is the most common form of neurodegenerative disorder in elderly and results in progressive memory loss and cognitive decline. The appearance of senile plaques and neurofibrillary tangles are the neuropathological hallmarks of AD (1). Autopsy studies have shown the hippocampus to be affected by AD pathology early in the disease process, with ≈20%–50% loss of neurons by the time individuals are moderately affected (2,3). As a result, imaging studies have focused on this region in order to detect the early changes during the disease progression.

Currently the diagnosis of clinically probable AD can be confirmed only once the stage of dementia has been reached. Neuropsychological tests are necessary to recognize and monitor the subjects at risk. However, to date there is no accurate cognitive marker to identify AD in the early disease process. Moreover, cognitive performances depend not only on age and education but also on mood and attention at the time of testing, and thus lack wide generalizability. Quantitative analysis of tau protein and amyloid-Aβ in the cerebrospinal fluid (CSF) has shown some diagnostic value in probable AD (4,5), but the procedure is totally invasive and has thus received only little attention so far in diagnosing AD.

An accurate diagnosis of AD in the early symptomatic stage is extremely challenging despite the development of useful clinical constructs like mild cognitive impairment (MCI), which enables the identification of individuals who may be in an early clinical phase of AD. MCI was previously defined as a transitional state that can precede dementia, but the condition and the rates of conversion remain controversial. Despite many consensus conferences, experts cannot agree on critical aspects of MCI, particularly with respect to its clinical utility. Based on neuropsychological studies, a hippocampal memory profile has been proposed for MCI as prodromal AD (6). In light of current drug development aimed at slowing AD progression, diagnosing AD at its prodromal stage is particularly important. Recently, research has begun to focus on developing new tools, such as neuroimaging and CSF biomarkers, that could increase the specificity of the prodromal AD diagnosis.

The goal of research in this area is therefore to develop highly specific and sensitive methods capable of identifying the subjects in the early stage who progress to AD. Neuroimaging markers provide an alternative and objective assessment of progression of AD. Out of various imaging techniques MRI is the most widely accepted technique to diagnose the various pathological conditions, and to monitor the treatment response, based on the changes in T₂ and T₁ contrast relaxation properties. It has been shown that the T₂ value was not significantly different between normal subjects and AD patients (7). In one recent study, Yuan et al (8) have shown that fluorodeoxyglucose positron emission tomography (FDG-PET) performs slightly better than single photon emission tomography (SPECT) and structural imaging in prediction of conversion to AD in patients with MCI. However, the wide clinical utility of the PET technique is hampered by its poor resolution and necessity of a cyclotron at the imaging center that can produce the radioactive tracers. Other MR techniques like diffusion tensor imaging (DTI) and proton spectroscopy (¹HMRS) have also been used in detection of normal white matter changes in patients with AD (9, 10). Ukmar et al (9) have shown a significant difference in fractional anisotropy (FA) between control and AD, while the FA was not significantly different in AD compare to MCI. Similarly, in a ¹HMRS study it has been shown that the hippocampal metabolite profile in MCI is similar to AD (10).

An alternative contrast mechanism is T1rho (T_1ρ), the spin lattice relaxation time constant in the rotating frame, which determines the decay of the transverse magnetization in the presence of a “spin-lock” radiofrequency field (11). In biological tissue, exchange between protons in different environments is expected to contribute T_1ρ relaxation. The molecular process that occurs in the milliseconds range influences the T_1ρ relaxation time constant. T_1ρ MRI has been previously used to measure T_1ρ relaxation time in normal human brain and showed the higher range of values compared to T₂ (11). Earlier, T_1ρ has been used to delineate brain tumors, characterize breast cancer tissue, and monitor the level of cartilage degeneration (12–14).

The current study was performed with an aim to measure the baseline T_1ρ in hippocampus in the brain of AD (n = 49), MCI (n = 48), and control (n = 31) cohorts and to determine whether the T_1ρ value shows any significant difference between these cohorts.

MATERIALS AND METHODS

Patient Selection

The Institutional Review Board approved the study protocols. In the current study we included 49 AD patients (mean age ± SD = 76.8 ± 9.1 years), 48 MCI patients (mean age ± SD = 71.93 ± 8.7 years), and 31 age-matched controls (mean age ± SD = 70.2 ± 9.4 years). All patients underwent a standardized clinical assessment including medical history, physical and neurological examination, psychometric evaluation, and brain MRI. The Mini-Mental State Examination (MMSE) was used as a measure of general cognitive function. The MMSE examination scores were 29.03 ± 1.13 in control, 24.72 ± 2.93 in MCI, and 19.34 ± 6.04 in AD. Diagnoses were made in conference by a team of neurologist, neuropsychologists, a neurophysiologist, and a psychiatrist. Diagnoses of MCI was made according to the Petersen criteria for MCI (15) and the National Institute of Neurological and Communicative Diseases and Stroke/Alzheimer’s Disease and Related Disorders Association criteria (NINCDS-ADRDA) for probable AD (16). Patients were excluded if they had a history of irritable bowel syndrome, chronic diarrhea, peptic ulcer, or gastroesophageal reflux disease; a history of cardiac disease; significant electrocardiographic (ECG) abnormalities; hematologic disorders; hepatic or renal disease; active malignancy within 5 years; or clinically important depressive, neuropsychiatric, cerebrovascular, or respiratory disease. The control group consisted of patients who presented to our memory clinic with subjective complaints and underwent exactly the same diagnostic work-up as the MCI and AD patients.

MRI Protocol

All these patients underwent a standard MRI protocol on a 1.5T Siemens Sonata (Erlangen, Germany) clinical scanner using the vendor-supplied head coil. Written informed consent was obtained from each patient before they underwent MRI. For T_1ρ MRI, a fluid-attenuated T_1ρ pre-encoded Turbo Spin-Echo pulse sequence was used (11). The imaging parameters were: TE/TR = 12/2000 msec, TSL (duration of spin lock pulse) = 10, 20, 30, 40 msec, with a spin lock frequency of 500 Hz, slice thickness 2 mm, field of view (FOV) = 22 cm, matrix size 256 × 128, bandwidth 130 Hz/pixel, echo train length = 4 for a total imaging time of 6 minutes for four images. The inversion time (TI) was fixed at 860 msec to remove the contribution from CSF to T_1ρ maps. An oblique coronal T_1ρ-weighted image of a slice perpendicular to the anterior/posterior commissure (AC/PC) plane was obtained. The slice was chosen to include the head of the hippocampus. Immediately after T_1ρ MRI, the entire volume of each subject’s brain was imaged in the coronal plane using a T₁-weighted 3D volumetric MPRAGE pulse sequence with 124 continuous slices. The parameters were TR/TE = 3.5/3000 msec, slice thickness = 1.2 mm, FOV of 24 cm and 192 phase encode steps, and flip angle 8° for a total imaging time of 10 minutes.

Data Processing

The signal expression for the T_1ρ-weighted MRI is given by:

$$M(\mathit{\text{TSL}})=M_0{e}^{-\mathit{\text{TSL}}/T_{1\rho }}$$ [1]

where M₀ is the thermal equilibrium magnetization. The T_1ρ relaxation time constant is dependent on the amplitude of the SL field, which is reported as a frequency of the SL field, and typically ranges from zero to a few kilohertz. The equation was linearized and then used to generate T_1ρ maps by fitting all pixel intensity data as a function of TSL time using linear regression. T_1ρ was calculated as −1/slope of the straight-line fit (11). Pixels whose intensities correlated poorly (R² < 0.95) with the fitting equation were set to zero. Pixels outside of the brain were also set to zero. The images were transformed to a G4 PowerBook computer (Apple, Cupertino, CA) and images were processed in the IDL programming language (RSI, Boulder, CO). For brain segmentation a previously developed method was used to partition the volumetric MPRAGE scans into 92 ROIs incorporating all major cortical and subcortical regions (17). This method deforms MRI scans of different brains into anatomical coregistration with each other, and into coregistration with a standardized template. The template’s labels are then transformed to individual scans by applying the elastic transformation that was found to coregister the respective images. For quantitative analysis the two regions of interest (ROIs) were defined on T_1ρ images, ie, left and right hippocampus. A program written in IDL was used to automatically report T_1ρ values from both ROIs. The coregistration of the data and the hippocampus segmentation was performed in the presence of an experienced neuroradiologist.

Statistical Analyses

For statistical analysis the T_1ρ value from the left and right sides were averaged for hippocampus. Descriptive statistics were performed to calculate the mean value of hippocampus T_1ρ for different cohorts (control, MCI, and AD). One-way analysis of variance (ANOVA) using Bonferroni post-hoc multiple comparisons was performed to compare the T_1ρ value among the different cohorts. Pearson correlations between T_1ρ values versus age and between T_1ρ versus MMSE score were performed separately for each cohort. A P-value of less than 0.05 was considered statistically significant. All the statistical computations were performed using the Statistical Package for Social Sciences (SPSS) v. 16.0 (Chicago, IL).

RESULTS

Figure 1 shows the T_1ρ maps of the control, MCI, and AD. Higher T_1ρ values are observed in the hippocampus of an AD individual compared to control and MCI individuals. The mean hippocampus T_1ρ values of control, MCI, and AD cohorts are reported in Table 1 and are shown in Fig. 2. One-way ANOVA showed that the T_1ρ was significantly different (P = 0.000) between the groups (Table 1). Bonferroni multiple comparisons showed that the T_1ρ was significantly increased both in the AD (P = 0.000) and MCI (P = 0.037) cohorts compared to control (Table 2). A significant increase in T_1ρ was observed in AD (P = 0.032) compared to MCI (Table 2). Hippocampus T_1ρ was 13% greater in the AD patients than controls, while in MCI it was 7% greater than controls. Hippocampus T_1ρ in AD patients was 6% greater than MCI.

Figure 1.

T_1ρ maps of the hippocampus region in the brain (in color) overlaid on fluid-attenuated T_1ρ MRI of control (72 Y female, a), MCI (76 Y male, b), and AD patient (75 Y female, c). Pixels with higher T_1ρ (red) are more prominent in the hippocampus of an AD patient. A lack of signal from CSF implies that the higher T_1ρ values are not due to fluid.

Table 1.

T1ρ Values From the Three Groups

Diagnosis	Number	Hipp T1ρ (mean ± SE)
Controls	31	88.5 ± 1.9
MCI	48	95.1 ± 2.1
AD	49	101.8 ± 1.6
ANOVA (P)		0.000

MCI, mild cognitive impairment; AD, Alzheimer disease; ANOVA, analysis of variance.

^*P-value at significant level.

Figure 2.

The graph shows T_1ρ values in the hippocampus of control, MCI, and AD cohorts. A 13% increased T_1ρ over control and 6% increased over MCI was present in AD, while in MCI it was increased by 7% over control.

Table 2.

Bonferroni Multiple Comparisons Among Different Groups

Group	Group	Mean difference	P-value	95 % CI
				LB	UB
Control vs.	MCI	−6.7	0.037*	−13.73	0.29
	AD	−13.3	0.000*	−20.33	−6.36
MCI vs.	AD	−6.6	0.032*	−12.80	−0.44

MCI, mild cognitive impairment; AD, Alzheimer disease; CI, confidence interval; LB. lower bound; UB. upper bound.

^*P-value at significant level.

Even after a significant difference in the T_1ρ value between the three cohorts (controls, MCI, and AD), a consistent overlap in individual hippocampus T_1ρ value was observed (Fig. 2). The T_1ρ relaxation time may be changed during the normal aging process due to relative change in the volume fractions of water and other macromolecules in the brain tissue. Therefore, we examined the age-related changes in T_1ρ values. None of the cohort showed the age-related changes in T_1ρ [control (r = −0.098, P = 0.606), MCI (r = −0.002, P = 0.988), and AD (r = 0.151, P = 0.306)].

On correlating the MMSE score with T_1ρ we did not find any significant correlation in any of the cohort [control (r = −0.154, P = 0.417), MCI (r = −0.092, P = 0.534), and AD (r = −0.069, P = 0.641)].

DISCUSSION

Discovering sensitive and specific markers of early AD would be a major breakthrough, as it would facilitate testing of novel therapeutic interventions that could slow or perhaps even arrest the degenerative process before dementia develops. In the current study we describe the relevance of T_1ρ estimation in the early diagnosis of AD pathology. Significant increased T_1ρ in hippocampus of MCI and AD patients was found compared to control. Also, in AD the hippocampus T_1ρ was significantly increased over MCI.

In the current study the hippocampus T_1ρ in AD patients was increased by 13% over control and by 6% over MCI. It has been shown that increased T_1ρ signal in AD patients is associated with the plaque burdens (18). Even after significant differences in T_1ρ between three cohorts, some of the MCI and control individuals showed high T_1ρ values in the range of AD pathology. We believe that this overlap is probably due to the large variation in T_1ρ values in the MCI and control individuals.

Delacourte et al (19) proposed that the neurofibrillary degeneration in AD may be hierarchized into 10 stages, ie, starting from involvement of perirhinal cortex (Stage 1) and subsequently entorhinal cortex and hippocampus (Stage 2), and then the rest of temporal cortex (Stages 4–6) followed by association area (Stage 7) and finally the entire cortex (Stage 10). It has been found that some subjects showed no cognitive alteration until Stage 6; most exhibited slight cognitive deficits even when most of the temporal cortex was involved (19). An increase in the MRI relaxation time in the AD brain is certainly due to a change in the molecular environment of water in the brain parenchyma resulting from AD pathology. Since no correlation between MMSE and T_1ρ was observed, this suggests that the cognitive score is not sensitive to tissue pathology. However, whether there is specificity of T_1ρ to plaque burden or to other AD pathology remains to be determined.

Another important issue regarding the early diagnosis of AD concerns is the selection of appropriate controls. The appearance of the neuropathological hallmarks of AD, senile plaque and neurofibrillary tangles, probably occur many years before the clinical symptoms of AD. As a consequence, within the aging population, subjects with no history of cognitive problems might show a sign of AD neuropathology at autopsy. It is clear that including normal with episodic memory within the normal range does not protect against the risk of some of these subjects progressing to MCI, and ultimately to AD, during follow-up. We will follow these patients for the next 3 years with the aim to see the role of T_1ρ values in predicting the probability to develop AD in the future in the cognitive normal and MCI cohorts.

Hippocampus atrophy has been used to differentiate AD pathology from age-matched controls but any age-related atrophy can confound attempts at distinguishing AD from normal control on the basis of T_1ρ. In the current study, however, there was no correlation between age and T_1ρ values, suggesting that the change in T_1ρ is due to an underlying pathology. It remains to be seen, perhaps by a longitudinal study, whether hippocampus T_1ρ in subjects with isolated memory impairment predicts a higher risk of developing AD, or, conversely, is only related to the presence of memory deficit. Longitudinal MR studies would require highly reproducible measurements of the T_1ρ. In an earlier work, we measured the reproducibility of the T_1ρ estimation in normal volunteers and the intrasubject covariance was less than 6%, but this has to be repeated on patients as well.

In the current study, measurement of T_1ρ was done using the two-dimensional T_1ρ-weighted MRI pulse sequence, which is limited to a single-slice acquisition. Recently, a three-dimensional T_1ρ-weighted pulse sequence has been developed by Witschey et al (21). It was found that using this sequence the average difference was less than 5% in T_1ρ maps of the same brain. In general, three-dimensional sequences provide higher resolution in less time than that required for corresponding two-dimensional counterparts to acquire the same volume. Additionally, a low SAR version of T_1ρ mapping has been developed by Wheaton et al (22). It was found that using this approach the SAR could be reduced by 40% along with only 2% change in the T_1ρ measurement in human brain.

The limitation of the current study should be emphasized, ie, any error associated with the registration between MPRAGE and T_1ρ image will lead inaccuracy in hippocampus segmentation, which may indirectly affect the T_1ρ measurement.

We conclude that higher hippocampus T_1ρ values in AD patients might be associated with the increased plaques burden. A follow-up study would help to determine the efficacy of T_1ρ values as a predictor of developing AD in the control and MCI individuals investigated in this study. T_1ρ MRI shows promise as a diagnostic method for AD in its early stage, and can potentially be a useful biomarker for the development of putative therapeutic agents for the treatment of AD.