Hippocampal imaging in the early diagnosis of AD 1988 to 2006

Super Memory Formula

Natural Memory Loss Cure and Treatment

Get Instant Access

Mony J. de Leon1

Groucho Marx said, "Time flies like an arrow. Fruit flies like a banana;" such are my experiences in writing this summary. In 1977, as a tuition-poor Columbia University student, intrigued about the anatomical basis for dementia, I needed to develop a doctoral dissertation project. Computed Tomography (CT), introduced in 1972, was in its first generation in New York City. A fellow student told me, that at NYU, Steven Ferris was using CT scans to screen dementia patients for clinical trials. I was given the huge opportunity to examine the then short stack of NYU X-ray films and, after a chance meeting in 1978 in the hallway with a young NYU resident neurologist, Irwin Blau, and with neuroradiologist Ajax George, although I did not realize it then, my career had begun. We examined all possible scans from patients with senile dementia, patients with cognitive dysfunctions not severe enough to be called dementia which was then referred to as questionable dementia, other poorly understood dementias, such as normal pressure hydrocephalus, vascular, and Pick's, and as many of the normal spouses and patient care-givers that we could convince to stick their heads into the CT scanner. NYU neuropathologist John Pearson gave us specimens to scan and cut, and lessons at the microscope. The dementia field before 1980 did not consider imaging a "legitimate" research domain, and for several years we had no funding and no students. Virtually all clinical imaging in dementia was done to rule out stroke and other mass lesions. Of the few groups engaged in neuroimaging research, in the US and UK, the focus was on CT ventricular enlargement and cortical sulcal prominence (Roberts et al. 1976) and, in Scandinavia, xenon cortical perfusion imaging (Ingvar et al. 1975). We made hundreds of measurements on CT scans, using rulers, and temperature-controlled planimeters, even cutting out paper-traced ventricles with scissors and weighing them to learn: 1) if we could distinguish senile dementia of the Alzheimer type (SDAT) from normal, and 2) which regional measurements were best associated with cognitive impairment. In the process, we developed methods to evaluate cortical atrophy (de Leon et al. 1979) and periventricular white matter pathology (George et al. 1986a,b). With our rulers, scissors, and weighing scales, we learned that measurements of the temporal horn and third ventricle were better than those of the frontal horn in discriminating SDAT. This finding put us on the trail of temporal lobe pathology.

In 1980, the NIH funded five FDG-PET centers, one of which included Brookhaven National Laboratories (BNL) in partnership with NYU. Also in 1980, I began an NYU post-doctoral neuroimaging fellowship in Psychiatry and Radiology and got a salary. Now Al Wolf (BNL), Ajax George, David Christman (BNL), Steven Ferris, and Joanna

1 Director Center for Brain Health, NYU School of Medicine, 560 First Ave, New York, NY 10016, USA

Fowler (BNL) and I were working together. Thus began a period of great excitement and activity, in part made possible by my finding a rent-stabilized walk-up apartment in Manhattan and a shared desk at NYU. In the absence of computers to assist with image analysis, using draftsman tools, David Christman and I developed CT and PET image registration protocols to map regional tissue glucose metabolism in 2-D. Later, Henry Rusinek and Wai Tsui of NYU computer automated this coregistration process and compiled a library of 3-D image analysis tools, called MIDAS, that continues to grow. However, in 1980, samples were outlined on hard-copy CT films and mechanically transferred to paper FDG PET scan printouts to yield the tissue sample coordinate values that were phoned in to BNL. Days later, we received back the regional metabolic rates. We published the first FDG PET paper in 1980 in which we demonstrated widespread metabolic reductions in SDAT relative to normal control (Ferris et al. 1980). In London in 1981, Frackowiak demonstrated that oxygen metabolism is reduced in SDAT using PET (Frackowiak et al. 1981), and within five years, there were reports from Berkeley University (Friedland et al. 1983), NIH (Cutler et al. 1985; Duara et al. 1983), University of Pennsylvania (Alavi and de Leon 1985) and UCLA (Benson et al. 1983; Kuhl et al. 1982) that cortical glucose metabolism reductions in AD were observed using PET. Frackowiak et al. (1981), Metter et al. (1985), Foster et al. (1984), and Friedland et al. (1983) refined this view to highlight the now well-known temporoparietal deficit characteristic of AD. Yet, in spite of Ball having pronounced in 1985 that AD was a "hippocampal dementia" (Ball et al. 1985), there was still no mention in the structural or metabolic imaging literatures of a hippocampal examination.

In 1984, Ajax and I began experimenting with scan angles in the acquisition protocols. We developed the so-called "negative angulation" to visualize the temporal horns on fewer tomographic slices. This protocol was also applied to the PET to enable more accurate sampling of the lateral temporal cortex glucose metabolism (the 1.5-cm resolution of the early PET cameras prohibited accurate isotope recovery from the hippocampus). This extra acquisition protocol also added to the cost of the CT scans, and with the help of Zaven Khachaturian at the NIH-NIA, we received funds to develop our structural imaging research. It was somewhat of a surprise for us to find that the negative angle CT scan acquisition revealed changes in the region of the hippocampus. We first termed these changes, which appeared somewhat like a dark blur on CT, peri-hippocampal lucency, reflecting our uncertainty as to whether the attenuation change was due to (CSF atrophy) or to tissue damage (see Fig. 1). After several years of studying the location and histology of the hippocampal lucencies, made possible in large measure with the post-mortem materials provided by neuropathologists Henryk Wisniewski at the Institute for Basic Research and Gleb Budzilovich and Douglas Miller from NYU, and the timely availability of a few good MRI scans, in 1988 we wrote a paper (de Leon et al. 1988) reporting that moderate to severe hippocampal atrophy was more common in SDAT [now called Alzheimer's disease (AD)] than in normal controls, and the magnitude of the atrophy was associated with hypercortisolemia following i.v. glucose infusion (see Fig. 2 adapted from de Leon et al. 1997). Also in 1988, Seab et al. wrote the first hippocampal volume paper demonstrating that AD patients have smaller hippocampi than controls. We later published anatomical validations reporting that hippocampal lucencies were caused by enlargement of the hippocampal, choroidal, and transverse fissures of Bichat secondary to the loss of hippocampal volume (de Leon et al. 1993; George et al. 1990; Narkiewicz and de Leon 1992; Narkiewicz et al. 1993)

(see Fig. 3 and (Petrella et al. 2003 for review). But, the proof for an early diagnosis still required longitudinal clinical prediction.

In 1989, in a combined, cross-sectional and prediction CT study of 175 patients, including 76 with AD, 27 with mild cognitive impairment (MCI) and 72 normal elderly, we reported that hippocampal atrophy was found in 87%, 70%, and 22% of the groups, respectively. But, perhaps our most relevant clue to the early diagnostic potential of hippocampal atrophy was that, unlike normal controls, who showed an increasing prevalence of hippocampal atrophy with age, hippocampal atrophy was independent of age in AD (de Leon et al. 1989). In other words the greater majority of AD patients consistently had hippocampal atrophy, whereas in normal subjects, hippocampal atrophy only became prevalent at great age. These findings encouraged us to examine whether hippocampal atrophy was a predictor of the decline to AD in "questionable dementia" patients (named mild cognitive impairment [MCI] by Barry Reisberg). In the prediction part of the study, we observed that 11/20 cases of MCI deteriorated to AD after three years, whereas 28/28 controls did not decline. The presence of hip-

normal questionable

Hippocampal Atrophy
Fig. 1. Hippocampal atrophy rating scale as shown on MRI (de Leon et al. 1993, as derived from de Leon et al. 1989)


Alzheimer Disease Hippocampus

65 70 75 80 85 90 AGE

Fig. 2. Hippocampal atrophy as a function of diagnostic group and age (de Leon et al. 1997)

Case Study Mri Alzheimers
Fig. 3. Hippocampal fissures in red and ventricle in blue (de Leon et al. 1993)

pocampal atrophy correctly identified 91% of the decliners and 81% in each of the two non-declining groups (de Leon et al. 1989).

In 1993, we replicated our CT MCI to AD prediction finding using MRI (de Leon et al. 1993) and, in 1994, Jobst et al. in the UK demonstrated with CT marked hippocampal atrophy progression in AD. Our group, (Convit et al. 1993) showed that the MRI hippocampal volume loss was anatomically specific in MCI, and after 1997, others demonstrated the MCI to AD predictions using the MRI hippocampal volume (Jack et al. 1999; Kaye et al. 1997; Visser et al. 1999). In 2000, Bobinski et al. working with both Henryk Wisniewski's lab and our group reported that the hippocampal volume loss in AD was highly correlated with the number of hippocampal neurons. In parallel to the brain imaging work, in 1991 Flicker et al. at NYU demonstrated that declarative memory impairments predicted the transition from MCI to AD. Soon after we showed that memory function in normal aging (Golomb et al. 1993 and MCI (de Leon et al. 1993) was related to the hippocampal size. Thus, the earliest links between declarative memory changes, hippocampal size, and early AD pathology were made almost 15 years ago. Of immense value to this and the more recent imaging work were the pioneering observations of Heiko and Eva Braak (1991), who, in staging neurofibrillary pathology, demonstrated that the hippocampus and entorhinal cortices are among the early sites of damage in AD. Structural imaging has continued to contribute to the recognition of early clinical disease. MRI measured reductions in entorhinal cortex volumes were found to predict the MCI decline to AD (de Toledo-Morrell et al. 2000; Dickerson et al. 2001; Killiany et al. 2000) and perhaps to be even more useful than hippocampal volume measurements (Stoub et al. 2005). Most exciting, it was first reported by Rusinek et al. 2003 that accurate predictions of the decline from normal to MCI, or to AD, were possible with hippocampal measurements. This was later confirmed in studies by Rusinek et al. 2003, den Heijer et al. 2006 and Fox et al. 1996.

With improved spatial resolution of PET cameras (4-6 mm) and with the availability of the image registration algorithms developed by Pelizzari et al. (1989), Woods et al. (1993), and Rusinek et al. (1993), in 1997 we applied hippocampal imaging to PET (de Leon et al. 1997). While even today, the overwhelming majority of FDG-PET studies of MCI and AD rely on automated techniques that do not specifically examine the hippocampus, MRI-based FDG-PET sampling protocols consistently identify hippocampal metabolic abnormalities (see Nestor et al. 2003 and Mosconi et al. 2004 for review). Interestingly, De Santi showed that the PET hippocampal measurements were diagnostically superior to those obtained from MRI in classifying NL, MCI, and AD patients(De Santi et al. 2001). In 1998 Johnson et al. demonstrated with SPECT that hippocampal perfusion deficits predicted the transition between MCI and AD, and, in 2001, we showed with PET that entorhinal cortex and hippocampus glucose metabolism reductions predicted the transition between normal and MCI (de Leon et al. 2001).

Most recently, computerized image analysis approaches to examine the hippocampus have been developed. MRI approaches by Rusinek et al. (2003), Csernansky (Wang et al. 2003) and Thompson et al. (2004) and PET-based solutions by Mosconi et al. (2004) point to a new era. This new generation of automated tools will provide the opportunity for large-scale investigations to use standardized sampling of this difficult to measure, yet highly informative brain region.

Hippocampus Mci
Fig. 4. Ten-year time series demonstrating on MRI, hippocampal (red), entorhinal cortex (yellow), and ventricular (green) changes in association with clinical decline from normal in 1993 to MCI to AD in 2003

In summary, over the past 18 years, structural and glucose metabolism imaging studies of the hippocampus and entorhinal cortex (see Fig. 4) have contributed to the early diagnosis of AD. The next horizon will be the use of imaging to select presymptomatic patients and to monitor a therapeutic course in primary prevention studies.

Acknowledgements. I am in debt to my brilliant colleagues: Ajax George, Henry Rusinek, Wai Tsui, Barry Reisberg, Steven Ferris, Susan De Santi, Antonio Convit, Maciek Bobinski, L.A. Saint Louis, and Lisa Mosconi. My appreciation goes to the many friends and patients who stuck their heads in scanners at my request. With gratitude for the support of Zaven Khachaturian and Neil Buckholtz from the NIH. In memory of my gifted mentors and pioneers: Alfred Wolf, David Christman, Henryk Wisniewski, and Jacob Cohen.

John C. Morris

Was this article helpful?

0 0
All About Alzheimers

All About Alzheimers

The comprehensive new ebook All About Alzheimers puts everything into perspective. Youll gain insight and awareness into the disease. Learn how to maintain the patients emotional health. Discover tactics you can use to deal with constant life changes. Find out how counselors can help, and when they should intervene. Learn safety precautions that can protect you, your family and your loved one. All About Alzheimers will truly empower you.

Get My Free Ebook

Post a comment