Results: Our findings suggest that possession of the e4 and e2 alleles confer adverse and beneficial neuropsychological effects, respectively, in the AIBL cohort, but that there are differences between these alleles with respect to dose-response outcomes and cognitive domain specificity. The influence of the e4 allele appears to be greater in magnitude than the e2 allele. Of note, the effects of these genetic factors are present i) in both AD and non-AD cases and ii) after controlling in AIBL participants for premorbid intellectual capacity, age and self-reported symptoms of depression and anxiety. Conclusions: These findings have important im- plications in characterizing the role of genetic factors influencing the manifestation of neurocognitive characteristics that have been associ- ated diagnostically with the onset of AD. O4-05-07 SUBHIPPOCAMPAL DYSFUNCTION AS A CLINICAL MARKER OF INCIPIENT ALZHEIMER’S DISEASE Mira Didic Hamel Cooke 1 , Emmanuel J. Barbeau 2 , Olivier Felician 1 , Eric Guedj 3 , Jean-Philiphe Ranjeva 4 , Patrick Cozzone 4 , Mathieu Ceccaldi 1 , 1 Service de Neurologie et Neuropsychologie, Assistance Publique des Ho ˆ- pitaux de Marseille, INSERM U751, Marseilles, France; 2 CerCo, UMR 5549, CNRS, Toulouse, France; 3 Service Central de Me ´decine Nucle ´aire, Assistance Publique des Ho ˆpitaux de Marseille, Marseilles, France; 4 CRMBM UMR CNRS 6612, Faculte ´ de Me ´decine, Universite ´ de la Me ´di- terrane ´e, Marseilles, France. Contact e-mail: mira.didic@ap-hm.fr Background: The nature of the memory dysfunction in incipient AD re- mains to be defined. In the early Braak and Braak stages, neurofibrillary tangles develop in the anterior subhippocampal cortex. In the monkey, damage to this area severely impairs performance on visual recognition memory tasks. Studies in humans provide evidence that this area plays a key role in context-free memory. Therefore, evidence of dysfunction of subhippocampal cortex and recognition memory could be early markers of incipient AD. Methods: Our group developed the DMS48, a new visual recognition memory task based on the delayed matching to sample task used to evaluate visual recognition memory in studies with monkeys. Patients with amnestic MCI (aMCI) as well as healthy controls were included into a combined extensive neuropsychological and neuroimaging study using SPECT-scan, 3D-MRI and SMRI (a min- imum of 23 aMCI patients and 23 control subjects were included into each study). Results: aMCI patients had impaired scores on the DMS48 compared with controls. Patients who failed on the DMS48 re- ceived less benefit from cueing on the FCSR (Free and Cued Selective Reminding Task), suggesting mesiotemporal dysfunction. Performance on the DMS48 was correlated with atrophy of the subhippocampal region. aMCI patients who failed on the DMS48 showed neuroimaging features that are usually described in patients who ultimately convert to dementia of the AD type. In this subgroup, mesiotemporal and temporo-parietal hy- poperfusion was found on SPECT. On MRI, a VBM study showed vol- ume loss in the mesiotemporal lobe and temporo-parietal areas bilaterally. On SMRI, a metabolic profile that is usually associated with AD was found in the subhippocampal cortex and the hippocampus. To date (mean follow-up ¼ 30 months), there is a higher conversion rate in the aMCI group failing on the DMS48 (8 converters) than in the aMCI suc- ceeding on the DMS48 (one converter). Conclusions: These findings in- dicate that, within the overall aMCI population, those failing on the DMS48 are at a particularly high risk of AD. They also question the nature of the memory impairment at the pre-dementia stage of AD and emphasize the importance of studying context-free memory. O4-05-08 REDUCING NOISE IN THE ADAS-COG WORDLIST TASK William R. Shankle 1 , Paul Aisen 2 , Adam Fleisher 2 , Ronald Petersen 3 , Steven H. Ferris 4 , Michael Rafii 2 , Jared Smith 1 , 1 Medical Care Corporation, Irvine, CA, USA; 2 UC San Diego, San Diego, CA, USA; 3 Mayo Clinic, Ro- chester, MN, USA; 4 New York University, New York, NY, USA. Contact e-mail: rshankle@mccare.com Background: The ADASCog test is widely used in Alzheimer’s disease (AD) clinical trials. However, it may lack the sensitivity to robustly detect disease-modifying treatment effects. The most sensitive part of the ADAS- Cog is the wordlist task, which consists of immediate free recall of 10 words presented in different order across three learning trials, followed by a delayed free recall trial. This ‘‘shuffled word order’’ across trials may reduce the abil- ity to measure temporal encoding and cumulative learning effects. We com- pared the explainable information of the ADASCog free recall task with that of a very similar test, the MCI Screen (MCIS), with the primary difference that the MCIS word order is unchanged (fixed) across learning trials. Methods: The ADASCog was administered to 112 normal aging and 612 amnestic mild cognitive impairment (MCI) AD Cooperative Study subjects. The MCIS was administered to 169 normal aging (FAST stage 1), 107 MCI (FAST 3), and 159 mild dementia (FAST 4) Shankle Clinic subjects. Be- cause these groups used different selection and classification criteria, statis- tical methods were applied to minimize these differences. 40 logistic regressions were performed per sample to estimate each word-trial item odds ratio for predicting normal vs. impaired after controlling for total recall score (#/40). Separately, correspondence analysis (CA) was performed to 1) minimize sample differences by double standardizing effects due to subjects and word-trial items; and 2) maximize information explained per word-trial item. Results: Using logistic regression, the variance explained by word-trial item coefficients from the fixed word order method (MCIS) was 3.02 times that of the shuffled word order method (ADASCog). Using CA, the variances explained by the fixed vs. shuffled word order methods were 30.5% and 22.3% respectively, which is a 37% relative increase. Conclusions: Com- pared to the shuffled word order method (ADASCog), the fixed word order method (MCIS) explained more information related to temporal encoding and cumulative learning, and may therefore improve detection of beneficial effects of disease modifying therapies. WEDNESDAY, JULY 15, 2009 ORAL O4-06 OTHER DEMENTIAS O4-06-01 NEUROPATHOLOGICAL CHARACTERIZATION OF PROGRANULIN-DEFICIENT MICE: ACCELERATED LIPOFUSCIN ACCUMULATION SUGGESTS A ROLE FOR PROGRANULIN IN SUCCESSFUL AGING Jada M. Lewis, Zeshan Ahmed, Hong Sheng, Ya-Fei Xu, Wen-Lang Lin, Amy Innes, Harold Hou, Michael L. Hutton, Eileen McGowan, Dennis W. Dickson, Mayo Clinic College of Medicine, Jacksonville, FL, USA. Contact e-mail: lewis.jada@mayo.edu Background: In the periphery, progranulin is involved in wound repair, in- flammation, tumor formation and trophic support. Little is known about the role of progranulin in the CNS; however, roles in development, sexual differ- entiation and long-term neuronal survival have been suggested. Mutations in the gene encoding progranulin (GRN) cause frontotemporal degeneration with ubiquitin and TDP-43 immunoreactive neuronal inclusions due to hap- loinsufficiency; however, it is currently unclear how this deficiency leads to neurodegeneration. Using GRN-deficient (-/þ and -/-) mice, we investigated the neuropathological consequences of progranulin deficiency. Methods: Histological techniques, immunohistochemistry and electron microscopy (EM) were utilized to compare neuropathological changes in an aged-series of GRN-deficient mice to wild-type controls. Results: GRN-/- mice had re- duced viability, since crossbreeding of GRNþ/ mice resulted in a lower than expected Mendelian ratio for GRN-/- mice. Premature death of GRN-/- mice that was decreased with improved husbandry suggests in- creased sensitivity to handling stress. GRN-/- mice had age-associated in- crease in intra-neuronal ubiquitin-, which was particularly severe in the hippocampus and thalamus, but showed no abnormal TDP-43-immunoreac- tivity. This corresponded to autofluorescent pigment, which was shown to be lipofuscin by EM. Ubiquitin-positive, lipofuscin in 7-month old GRN-/- mice was equivalent to that observed in 23-month wild-type mice. Although no Oral O4-06: Other Dementias P161