This section attempts to clarify why the data in these four studies were included, and then strives to determine which could be the most useful biomarkers according to the present results and the literature available at the time of writing.
Data in studies I-IV: studies I and II involved patients from the Kuopio L-MCI cohort (Julkunen et al. 2010) and KUH database. For this cohort, only MMSE, CSF and T1 images were available; from T1 images HV, TBM and VBM values could be obtained. Study I only compared three imaging methods and had no intention of combining different biomarkers.
The results were adjusted for age, gender and intracranial volumes. Study III focused on the importance of the symptoms record, neuropsychological batteries and tests and on importance of the difference concerning SPECT and MRI.
HV, TBM and VBM were applied identically in studies I, II and IV.
Study IV combined different cohorts, thus it was necessary to select data that was similar in these cohorts. The four studies included biomarkers which are accepted criteria for MCI (Petersen et al. 1997), FTD (Neary et al. 1998) and AD (McKhann et al. 1984). It would be advisable to use new criteria in future studies, for example the novel criteria proposed by Rascovsky et al., for bvFTD (Rascovsky et al. 2011).
In future studies, it would be interesting to include not only FDG-PET, but also the novel imaging techniques such as DTI and RSfMRI in attempts to differentiate AD from bvFTD in their early stages. MRI can also evaluate the number of MBs and white matter hyperintensitites which are associated more with AD or VaD rather than with any disease from the FTLD spectrum.
We had autopsy prove for only a minority of cases in studies I, II and III. It could be advisable in the future to perform studies with cases including autopsy confirmation and try to see the correlation between the neuropathology findings and each specific biomarker.
Biomarkers: demonstrated in Table 12 in the review of literature, there are many studies which have attempted to predict conversion from MCI to AD with a variety of biomarkers, however very few have specifically compared bvFTD and AD.
The differential diagnosis between AD and bvFTD starts with a comparison between the profiles of the symptoms. The CSF biomarkers may help in distinguishing between both diseases, although the biomarker most commonly used is FDG-PET, since it reveals different patterns of hypometabolism in AD and FTD (Berti, Pupi & Mosconi 2011). PIB-PET exhibits no or lower accumulation of amyloid in FTD cases as compared to AD (Engler et al. 2008), but this technology is available in only a few centres. Furthermore it does not provide a conclusive diagnosis, since about 10% of AD cases are PIB negative and 15% of FTD cases are PIB positive (Rabinovici et al. 2011).
Novel techniques are emerging such as DTI or RSfMRI (Frisoni et al. 2010). However, structural MRI is the only method which correlated with the extent of neurodegeneration and for that reason it is recognized as is the gold-standard for categorizing the state of the disease.
Some studies have detected a correlation between the amyloid burden and inactivation of default-mode circuits (Sperling et al. 2010). Thus RSfMRI as PIB-PET could be used when a patient is classified as MCI or in the early stages of AD or bvFTD, and in the follow-up structural MRI should be used as has been already proposed (Jack et al. 2009). AD and FTD are characterized by the default-mode network and salience network respectively (Zhou et al. 2010), however more studies will be needed to assess the predictive value of RSfMRI in the conversion from MCI to AD or FTD.
Structural MRI: Which methods should be used? As reported in the literature and also in this thesis, automatic techniques can achieve accurate results. In study II, MRI was found to be more relevant than the manual outlining MRI conducted in study III when one is
comparing FTD patients with controls and AD. However, the manual outlining of the hippocampus was clearly more relevant in the AD vs. FTD comparison conducted in study III than HV was in study II.
Nonetheless, it is recommended that HV, TBM and VBM should be conducted as automatic techniques for four reasons: they can be easily obtained from a single 3DT1 image, they are not long time-consuming, they are good biomarkers at least to some extent, and they can be easily added to the other biomarkers (tests, genetic profile, CSF) within the PredictAD tool.
It is true that PET amyloid imaging is more specific for obtaining a differential diagnosis between AD and FTD. However its cost and non-availability in many centres hinders its widespread use, while automatic MRI methods are readily obtained from one T1-volumetric MRI scan.
TBM and particularly VBM give accurate results to help in the diagnosis of AD and bvFTD. Not only the hippocampus, but all the MTL (entorhinal cortex, parahippocampus, hippocampus, amygdala) should be studied, as the entorhinal cortex and CA1 of the hippocampus are severely affected.
With respect to CSF, recent guidelines state that MCI and AD can be predicted by quantifying the levels of Aβ, P-Tau and T-Tau, and CDR testing (Vos et al. 2013b). The present studies strongly support the use of CSF in differentiating AD from FTD. However, discrepancies were found between the individual importance of Aβ, P-Tau and T-Tau in studies II and III. With respect to the ability to predict conversion from MCI to AD, study IV reveals that CSF was the most accurate biomarker in the Kuopio MCI and DESCRIPA cohorts, although it was less accurate in ADNI. This may be due to the different procedures used in CSF in the different centres; i.e. either ELISA or xMAP Luminex.
Finally, in all of these studies, it was possible to detect that differences in APOE genotype clearly help to differentiate AD from FTD.
Currently, genetic studies in FTLD should be considered if there is a first-degree relative with the diseases and if there is an early age of onset in the actual patient, because it is rare that a patient with FTLD has any mutation without a positive familial history (Chow, Alobaidy 2013). However, there may be cases with a family history and no positive results in the genetic testing. Although the proportion FTLD having a known genetic mutation (MAPT, GRN, C9ORF72) is much greater than the corresponding situation for AD cases, genetic testing should not be recommended for differentiating AD from FTLD, as so few FTLD cases carry any recognized mutation (Chow, Alobaidy 2013).
Genetic studies are usually performed in AD only if the onset is below the age of 65 years.
However although in the clinical practice there is no strong impulse to do genetic profiling in dementia patients, it could well be recommended to do so in research since this may help to clarify the real percentage of some genetic mutations (e.g. C9ORF72) and especially in LOAD, to determine whether certain combinations of genes mutations are significant for the development of the disease, contributing more than a single-gene mutation.
In these studies only the APOE genotype was included in DSI. We identified some FTD cases as displaying the C9ORF72 mutation, although this was considered as demographic information. In future studies involving Finnish cohorts, it would be advisable to check whether or nor these patients exhibit the C9ORF72 repeat expansion, due to its presumably high frequency (Renton et al. 2011).
A family history for dementia should be always recorded, as it is a major risk factor for developing both AD and FTLD.
With regard to cognitive symptoms, AD patients suffer early memory and orientation impairment while FTD subjects display deficits in executive functions and behavior, nonetheless when both diseases progress, all of these functions can become impaired, e.g.
FTD patients exhibiting memory disturbances and AD cases displaying behavioral changes.
However in daily practice, these limits are not that clearly established and there is a broader range of impaired functions at onset, such as impairment in executive functions in early AD stages, and conversely patients with FTD can suffer memory problems in the first stages of the disease. There may also be language problems in both AD and bvFTD, and these could well confuse the diagnosis by pointing towards PNFA or SD, or even logopenic aphasia.
Furthermore it would be advisable to try to correlate the different symptoms and performance in specific tests and the brain areas affected in MCI, AD and FTLD patients, as conducted before with VBM (Mahoney et al. 2011, Venneri et al. 2011, Bruen et al. 2008).
Only MMSE was utilized for the assessment of global cognitive functions. In future studies it could be advisable to use CERAD (which includes MMSE) for diagnosing and assessing the progression of MCI (Paajanen et al. 2014), and NPI (Hirono et al. 1999) and/or FBI capturing neuropsychiatric symptoms (Kertesz et al. 2003) for FTD, as well as
executive-function tests that are not covered in CERAD. It would be interesting to combine different behavioral rating scales (e.g. FBI, NPI) with neuropsychological scales (e.g. word list recognition, TMT B, clock drawing test).
In summary, the recommendation is that after an initial interview assessing the symptomatic situation, if there is a suspicion of dementia (MCI, AD or FTD) the clinical work-up should include : neuropsychological (MMSE, depression scales, language, memory, visuo-construction and executive-function batteries) and structural MRI (manual volumetry or automatic methods), supported by APOE genotype and CSF for the study of MCI, AD and bvFTD, and supplementation with FDG-PET/SPECT when it is not possible to discriminate between AD and bvFTD. It is recommended that SPECT should be used based on the results of study III, and FDG-PET based on the previous literature. APOE genotyping could be included on a routine basis, but the study of other genes is recommended only if there is an early age of onset and/or there is previous familial history of AD or FTD.
The stage and severity of the disease are important when considering the importance of biomarker combinations for diagnosis and monitoring progression. In AD, the amyloid accumulation is present at the earliest stages, i.e. in MCI, but the amyloid reaches its maximum level relatively quickly. Therefore in order to monitor the disease progression one could use structural MRI for assessing atrophy changes which are related to the elevated tau levels and not to amyloid accumulation. Perhaps metabolic and functional imaging could be more linear in the progression of the disease and also could help in the diagnosis in the earliest stages of the disease and also be useful in monitoring disease progression. Secondly standardization for collecting data from each biomarker is needed, as well of standardization of patient classifications. There are several studies which have incorporated a great number of biomarkers, but because of the different methodologies and the different criteria for classifying the patients, it has not proved to be feasible to compare
different studies. Finally more work needs to be done in the evaluation of different factors e.g. lifestyle and genetics RFs and cognitive reserve (Jack et al. 2013).
6.4 DISEASE STATE INDEX AND DISEASE STATE FINGERPRINT