Abstract
This longitudinal study evaluates the prognostic impact of amyloid PET in patients suspected of Alzheimer’s disease and presenting with isolated cerebrospinal fluid (CSF) increases in P-Tau proteins (NCT02556502). The rate of conversion, based on the DSM-5 criteria and all collected data (average follow-up of 39.2±13.2 months), was determined by a panel of experts blinded to the PET results and was 75%(6/8) for positive and 35%(6/17) for negative baseline amyloid PET. In this population with isolated CSF increases in P-Tau, a positive baseline amyloid PET was associated with greater than twice the proportion of dementia conversions within the following three years.
INTRODUCTION
The 2018 NIA-AA criteria are exclusively based on biomarkers, and the diagnosis of Alzheimer’s disease (AD) is the result of the combination of the two biomarkers highlighting the pathophysiological processes involved in this neurodegenerative disorder, i.e., the amyloidopathy and the tauopathy [1]. Amyloid and tau biomarker status can be determined by cerebrospinal fluid (CSF) analysis or positron emission tomography (PET).
Owing to their increased availability and lower costs, biomarker analyses are performed along with CSF analyses as first line strategies in routine clinical practice. However, in practice the significant proportion of lumbar punctures that are ambiguous, contraindicated, or refused is a real issue in the clinical management of patients with neurodegenerative diseases [2].
The prospective MAF (Maladie d’Alzheimer Florbetaben) study [3, 4] included populations of suspected AD patients whose routine clinical management was faced with the particular challenges of biomarker measurements exclusively obtained from CSF. These patients presented with mild neurocognitive disorders (MND) [5] and suspected AD, according to the criteria of the National Institute of Neurological and Communicative Disorders and Stroke as well as the Alzheimer’s Disease and Related Disorders Association (NINCDS ADRDA) [6], but with isolated P-Tau proteins in the CSF, which represents an ambiguous AD CSF profile according to the ATN classification [1].
Amyloid PET is a useful biomarker in this specific population. As many as one third of patients from this population were found to be 18-F-Florbetaben PET positive, leading to significant changes in their medical management [3]. Another study focusing on associations between CSF and PET amyloid bio-markers in this specific population also confirmed the relevance of using 18-F-Florbetaben PET, particularly in cases where the Aβ42 concentration is close to the laboratory standard [4]. Although these two studies underline the importance of amyloid PET in this specific population of patients with MND, suspected AD, and ambiguous CSF test results, they lack the longitudinal follow-ups to ascertain the potential prognostic value of the amyloid PET biomarker in cases with discordant amyloid CSF biomarker results. Our current study evaluates the prognostic value of amyloid PET biomarkers in a suspected AD population, with isolated P-Tau protein increases in CSF, over a follow-up period extending up to 63 months.
MATERIALS AND METHODS
Population
The population selected was part of the MAF study (NCT02556502) [3]. Our longitudinal study only included patients with isolated P-Tau increases from the CSF analyses (n = 25), it did not include MAF patients with isolated T-Tau increases with all negative amyloid PET imaging (n = 9) [3]. Details of inclusion criteria for the prospective MAF study have been reported elsewhere [3]. All participants provi-ded written informed consent for study participation, visits, and data source verification. No participant had a legal representative. This study was approved by the French ethics committee (CPP Est III) on September 15, 2015, received an authorization from the national competent authority (ANSM) on September 18, 2015 and adhered to the Declaration of Helsinki.
Study design
As previously reported, results from 18F-Flor-betaben PET were derived from a consensual visual analysis by two experienced nuclear physicians (CM, AV), who were blinded to all patient data, including MRI images [3]. Amyloid PET results were interpreted as either “positive” or “negative” in a conventional manner using brain amyloid plaque load (BAPL) scores, scores > 1 were considered positive, as currently recommended [7].
This longitudinal study was conducted in the “memory clinic” of the University Hospital of Nancy (France), a tertiary center, involved in a regional net-work of expert centers, providing research and res-ource support to secondary centers in terms of the management of neurodegenerative disease patients [7]. Longitudinal follow-up from December 2015 to September 2020 involved the collection of clinical and paraclinical data by an investigator blinded to the PET amyloid results (AJ). The clinical follow-up was carried out by the referring physician, either a neu-rologist, a geriatrician, or a general practitioner as is usual in clinical routine practice. Data collected from clinical examination results included: pyramidal and extrapyramidal signs, cerebellar signs, rapid eye movement disorders, dystonia, myoclonus, epilepsy, fluctuations, hallucinations, apraxia, behavioral disorders, aphasia as well as the assessment of autonomy and the Mini-Mental State Examination (MMSE) scores.
All of the collected data was interpreted and analyzed by a panel of experts (LH, TRJ) from the tertiary memory center of Nancy University Hospital by consensual analysis. Both experts were blinded to the amyloid PET results of the cohort.
The consensual analysis led to the definition of the dementia state according to the DSM-5 criteria [8] for each participant. The experts also proposed a definitive clinical diagnosis based on all the data available for each patient.
Statistical analysis
Quantitative variables are expressed as means±standard deviations, and categorical variables as percentages. Wilcoxon and Chi-2 tests were performed for the paired comparisons of quantitative and categorical variables, respectively. A Cox regression model to determine the predictive factors of conversion was applied with univariate analyses. A p-value of 0.05 was defined as significant. Statistical analyses were performed with the SPSS® 20.0 software.
RESULTS
Patients characteristics
Twenty-five consecutive patients (15 men, 60%, mean age at inclusion 73±9 years) were included in this longitudinal study. Clinical dementia rating, educational level, and results of CSF biomarkers at baseline have been reported previously [4]. The average follow-up period was 39.2±13.2 months with minimum and maximum follow-up times ranging from 14 to 63 months and with an average interval of 8.3±4.8 months between individual follow-up visits. Three deaths were reported during this time frame. Data collected at baseline and at the last of the follow-up visits are summarized in Table 1. Significant deteriorations in MMSE scores, autonomy and the development of apraxia were observed during follow-ups.
Patients clinical baseline and final follow-up characteristics
*With p value significant for the comparison between baseline and final follow-up.
Predictive factors of conversion
Among the 25 patients, 12 (48%) presented a conversion to dementia. Table 2 summarizes results of cox regression univariate analyses for predicting conversions according to the patients characteristics and baseline biomarkers. Amyloid PET was associated with a risk of conversion to dementia (hazard ratio of 3.1) even though the p-value was just above clinical significance (p = 0.06). The rate of conversion in patients with positive amyloid PET at baseline was 75%(6/8) whereas it was at 35%(6/17) in patients with negative amyloid PET. None of the other patients characteristics and baseline biomarkers showed such predictive value (Table 2). It should also be reported that apraxia appeared in 80%of patients with a positive amyloid PET between baseline and follow-up. As the clinical criteria were taken into account to define the conversion rate, these factors could not be specifically evaluated in the current study. No further multivariate analyses were conducted because none of the factors reached a high enough level of significance.
Cox regression univariate analyses for the predictive value of conversion
Final clinical results from the panel of experts blinded to the amyloid PET results
After analyzing all of the collected clinical data obtained during follow-up, the final diagnosis proposed by the panel of experts was incorrect in 9 cases (36%) related to the ATN classification with 6 patients having a final diagnosis of AD and a negative amyloid PET, and 3 patients not having AD and a positive amyloid PET. Results of the final diagnosis proposed for the 25 patients included in the study are available in Supplementary Table 1.
A representative amyloid positive PET series is provided in Fig. 1.
Representative axial positive 18-F-Florbetaben PET images (brain amyloid plaque load of 3) from a 73-year-old woman with mild neurocognitive disorders, an MMSE score of 25/30, and with CSF Aβ42 of 606.9 pg/mL, Aβ42/Aβ40 ratio of 0.075, T-Tau of 502 pg/mL, and CSF P-Tau of 63.6 pg/mL at baseline. The 35-month follow-up showed a loss of autonomy, the appearance of apraxia and a reduction in the MMSE score (0/30), in favor of a conversion to dementia according to the DSM-5.
DISCUSSION
The current longitudinal study based on a prospective series of patients with suspected AD and isolated P-Tau protein increases in CSF showed that a baseline positive amyloid PET result is associated with a higher rate of conversion to dementia in the next 3 years. This provides evidence supporting the potential prognostic value of amyloid PET in this very specific patient population.
The rate of discordance between CSF and PET amyloid biomarkers in patients with suspected AD is between 10 to 15%depending on the study [9–14]. In the majority of cases, these discordances are often related to positive CSF amyloid biomarkers and negative PET biomarkers, CSF amyloid positive biomarkers being associated with very early stages of disease, and no constituted amyloid plaques pot-entially imaged by amyloid PET [9, 10]. In the pop-ulation studied, we focused on the other type of discordance of amyloid biomarkers, a positive amyloid PET despite a normal concentration of amyloid CSF biomarkers including a correction for the Aβ42/Aβ40 ratio [3, 15]. This biomarker profile in patients with suspected AD was relatively rare in our cohort: only 25 patients (3.2%) of the 787 patients analyzed during the same period of inclusion in our prospective study [4]. This proportion is nevertheless consistent with the literature [16] and represents one of the most challenging populations to manage in clinical routine practice, particularly when biomarker detection is only based on CSF biomarkers. Having access to amyloid PET in this specific population leads to a significant amount of changes in confidence scores for AD and to potential treatment-related changes [3] but also benefits from a potential prognostic bio-marker result. More than twice the proportion of patients presented a conversion to dementia in cases of positive amyloid PET at baseline, compared to negative amyloid PET, in our longitudinal study with an average follow-up period of approximately 3 years. Amyloid PET may help to establish some non-pharmacological preventive as well as protective measures in these frail patients. This may be a useful tool in clinical practice since no other available CSF biomarker was able to provide such predictive value (p≥0.08, Table 2). However, in the absence of available amyloid PET results, we previously reported that an Aβ42 concentration in the very high range was associated with a negative amyloid PET in suspected Alzheimer’s disease and an increased CSF phosphorylated-tau protein concentration [4].
Our results are consistent with the literature which reports on the prognostic value of amyloid PET, with positivity closely related to dementia conversion, even in cases with discordant amyloid CSF results [11–14]. An additional point which is in favor of the association between amyloid PET positivity and clinical condition is the high proportion of apraxia apparent during the follow-up period in patients with positive amyloid PET (80%), the development of apraxia being a core sign of dementia conversion [8]. Autopsy confirmations of these discordant cases are required. The only case currently available in the literature showing a similar discrepancy in biomarkers (negative CSF but positive PET biomarker) confirmed an AD diagnosis [17].
The present study also indicates that the accuracy of the final AD diagnosis is still unsatisfactory after 3 years of follow-up. A panel of experts, blinded to the patients’ amyloid PET results, misdiagnosed 36%(9/25) of cases, predominantly in favor of AD diagnoses with negative amyloid PET results (n = 6). This rate of misdiagnosis is similar to that reported for the general suspected AD population in the literature [18]. This is, of course, a highly selected population which is associated with a particular risk of misdiagnosis, but this observation confirms that clinical follow-up of patients without the help of biomarkers leads to a significant proportion of inaccurate diagnoses.
Our main result is just above the level of significance (p = 0.06), but we are limited by the low number of patients belonging to this highly selected population. To the best of our knowledge no other publication has to date reported such high numbers of consecutive suspected AD patients with isolated P-Tau protein increases. Our study relied on multiple neurologists, geriatricians and general practitioners contributing data in accordance with general clinical practice and therefore did not allow to homogeneously specify the exact date of conversion. To limit this heterogeneity bias, all the data was collected by a single neurologist and subsequently reviewed by a panel of experts, blinded to the amyloid PET results. Finally our average longitudinal follow-up period was relatively limited when compared to the average delay of AD conversion in MND patients (3 years in our study versus 5 to 7 years in other studies [19]), but our rate of conversion (48%) was similar to that typically observed for MND patients after 5 to 7 years of follow-up.
To conclude, in our cohort of suspected AD and with an isolated increase of P-Tau proteins, a positive amyloid PET is of potential prognostic value since greater than twice the proportion of these types of patients present with dementia conversion within the following 3 years.
