Abstract
INTRODUCTION
One of the first questions patients and relatives ask in the doctor’s office after dementia diagnosis concerns prognosis and ‘how much time they have left’. The common answer will probably be quite vague, explaining that it differs per person and that dementia is typically known for its insidious onset and slowly progressive course. In general, dementia has a reduced life expectancy [1, 2], but survival is highly variable with reported median survival times ranging from 3–13 years after onset or diagnosis [1–6]. However, a subgroup deteriorates faster and dies within years or even months [5, 8]. Little is known about this group of rapid progressors or their cause of death.
A rapid clinical course is one of the hallmarks of prion diseases like Creutzfeld-Jakob disease (CJD) [9], but also Alzheimer’s disease (AD), dementia with Lewy bodies (DLB), frontotemporal dementia (FTD), and vascular dementia (VaD) occasionally display a rapid course to death [8]. Rapid mortality is clinically highly relevant, as this information could help advanced care planning. Therefore, it would be very helpful to know which clinical and biomarker characteristics might identify patients prone to a rapid death, in order to predict and hopefully affect prognosis.
Several studies reported on variables influencing survival time, such as dementia severity and institutionalization [4, 11]. Additionally, age and type of dementia have been reported to affect survival [5]. Also in early onset dementia, the impact on mortality is known to be high [12]. Moreover, former studies studied factors related to time to mortality [1, 11], rather than specifically focusing on rapid mortality. In addition, although literature on rapidly progressive dementia is slowly emerging [7, 13], the main emphasis lays on case reports, or diagnostic workup to exclude non-neurodegenerative causes, but not on mortality.
The aim of this study was to characterize dementia patients with a rapid mortality, defined as death within two years after diagnosis, according to clinical and biomarker features in a relatively young cohort of dementia patients. In addition, we investigated the cause of death in relation to dementia diagnosis in this population.
MATERIALS AND METHODS
Patients
We retrospectively studied all consecutive patients from the Amsterdam Dementia Cohort diagnosed with dementia between 2000 and 2013. A detailed description of the diagnostic work-up in our center has been described elsewhere [14]. In short, all patients underwent a standardized dementia screening including complete medical history (medication intake included), duration of complaints as noted by the caregiver, physical and neurological examination, laboratory tests (including DNA extraction for apolipoprotein E (APOE) genotyping), cerebrospinal fluid (CSF, including determination of Aβ1 - 42, total tau, and tau phosphorylated at threonine [ptau]), extensive neuropsychological testing including Mini-Mental State Examination (MMSE) [15] and the Geriatric Depression Scale (GDS) [16] and magnetic resonance imaging (MRI) of the brain. During a multidisciplinary meeting, diagnoses were made according to the current criteria for dementia, including AD [17], FTD [18], DLB [19], and VaD [20], or a rarer cause of dementia as corticobasal degeneration (CBD) [21], progressive supranuclear palsy (PSP) [22], or CJD [23]. We also included those patients diagnosed with dementia of unknown cause (unspecified dementia). The rare and unknown causes of dementia were clustered in one group as ‘another type of dementia’. Inclusion criteria were availability of MMSE score at baseline and disease duration (defined as duration of complaints) <5 years.
All patients provided informed consent to use their clinical data for research purposes at the time of their first visit.
Comparison Amsterdam Dementia Cohort
Overall, 1,623 patients were diagnosed with dementia during 2000–2013. To be able to compare patients who died within two years after diagnose with those alive, patients were categorized in three groups: those with a time of death within two years after being diagnosed at our center (n = 133; ‘rapid mortality patients’), those patients known to be alive two years after diagnosis (n = 892; ‘non-rapid mortality patients’), and a group lacking information about their status two years later (n = 598). Patients who had undergone euthanasia were excluded from analyses (n = 4). Patients without follow-up (n = 598) were comparable in age and diagnosis distribution to the rest of the population (n = 1021), but were more often female (53% versus 44%) and had a lower MMSE [mean(SD) 20(6) versus 22(5)], and were not taken along in further study descriptions (See Fig. 1).
Vascular risk factors
Presence of vascular risk factors was recorded and defined as hypertension (history of hypertension and/or use of blood pressure lowering drugs; yes/no), diabetes mellitus (history of diabetes mellitus and/or use of diabetic medication; yes/no), cardiac history (history of angina/myocardial infarction/ heart valve disorder/cardiac rhythm disorders; yes/no), history of TIA or stroke (yes/no), and smoking status (current or history of smoking; yes/no). The total number of drugs used per patient was counted (including all kind of drugs, as well as for vascular risk factors as other purposes). Furthermore, special note was made to the use of oral anti-coagulantia (yes/no).
Date and cause of death
Information on mortality (deceased yes/no) was obtained from the Amsterdam Dementia Cohort database. Most patients undergo regular follow-up in our center (n = 1,021 with mean(SD) follow-up of 42(21) months). If patients die (known by patients file, through information from general practitioners or relatives), this is included in the database.
Secondly, the medical records of all patients with a date of death < 2 years after diagnosis were reviewed and cause of death was recorded (and in a few cases autopsy was performed, see below). Information about cause of death was available in approximately half of the cohort. If the dementia itself was noted as cause of death this was recorded. Cause of death was further categorized in the following subgroups; intracerebral hematoma, stroke not further specified, cardiac disease (including myocardial infarction, heart failure), pulmonary disease (including aspiration/pneumonia), motor neuron disease (MND), oncology, and ‘other causes’. ‘Other causes’ consisted of known causes not classifiable in the previous noted subgroups, like renal failure or complications after a hip fracture.
If possible, the primary cause of death was included. For example, a patient with FTD died of a myocardial infarction in presence of MND, then myocardial infarction was recorded as cause of death. However, if the direct cause was not available and was simply noted as ‘cause of death MND’ then ‘MND’ was included. Lung cancer was categorized as oncology and not as a pulmonary disease. Finally, as noted before, if a patient died of euthanasia within two years after diagnosis this was recorded separately and these patients were not included in group comparisons.
In addition, all medical records of the rapidly progressing group were reviewed for availability of autopsy and the records of the Netherlands Brain Bank (NBB) was checked for possible autopsy data, not registered by the hospital.
Statistical analysis
Statistics were performed using SPSS 20.0. For comparison of data between the different groups of patients with a rapid mortality, non-rapid mortality patients and patients without follow-up, analyses were performed using chi-square tests and independent sample T-test when appropriate.
RESULTS
Patients with rapid mortality
In total, n = 129 patients were included with a date of natural death within two years after dementia diagnosis (13% (129/1021) of the patients with known follow-up, 8% (129/1623) of the total cohort), with a mean(SD) survival of 12(7) months (Table 1). Mean(SD) age was 72(10) years and half of the group consisted of men (n = 70, 54%). Patients with rapid mortality were mildly to moderately demented at presentation with a mean(SD) MMSE of 20(5). Compared to non-rapid mortality patients, patients with rapid mortality were slightly older and had a lower MMSE score (MMSE = 20(5) versus 22(5); p < 0.001). Furthermore, a higher GDS (mean(SD) 4(5) versus 3(4); p < 0.05) and a higher prevalence of cardiac history (16% versus 9%; p < 0.05) was shown among patients with a rapid course to death compared to non-rapid mortality patients. There were no other differences in demographic or clinical variables. Looking at the specific dementia diagnoses in patients with rapid mortality (Table 2), n = 55 (43%) had with a clinical diagnosis of AD, n = 23 (18%); FTD, n = 17 (13%); VaD, n = 13 (10%); DLB, and n = 21 (16%) with ‘another type of dementia’ including n = 6 patients diagnosed with CJD, n = 9 PSP patients (of which one presented with a clinical phenotype of CBS), n = 2 CBD patients, and n = 4 patients with unspecified dementia. Although in the rapid mortality group AD diagnosis was most frequent, the occurrence was much lower compared to the non-rapid mortality group. By contrast all other dementia diagnoses, including CJD, VaD, and FTD, were more frequent (p < 0.001).
Focusing on the rapid mortality cohort, among the total of n = 55 AD patients, n = 2 fulfilled the criteria of posterior cortical atrophy (PCA), n = 2 presented with a clinical phenotype of corticobasal syndrome, and n = 1 was positive for a PSEN1 mutation. There was no difference in CSF biomarker profile between AD patients with and without rapid mortality (p > 0.70). APOE ɛ4 genotype was most prevalent among AD patients compared to other dementia diagnoses, but prevalence was comparable between AD patients with and without rapid mortality (p = 0.89). Looking at the prevalence of vascular risk factors (Table 3), patients with VaD showed a vascular profile including high prevalence of concomitant vascular disease and ever smoking, and had highest amount of drug use compared to other dementia types [mean(SD) of 5(4) drugs per patient versus 3(3)]. Twenty-four patients did not use any drugs, the mean number per patients was 3 with a maximum of 18 different drugs. FTD patients showed no difference in MMSE score between rapid and non-rapid mortality and had the lowest prevalence of vascular risk factors. Patients diagnosed with ‘another type of dementia’ including CJD showed a CSF profile with high levels of total tau.
Cause of death
Data about cause of death were available for half of the rapid mortality cohort (n = 66) and differed widely (Table 4). Dementia as cause of death was reported most frequently, and patients recorded as died due to dementia itself were relatively young [mean(SD) 67(11) y] and most often diagnosed as AD or as ‘other dementia’ including PSP and CJD (n = 3, of 3 CJD patients exact cause of death was lacking). Next most frequent cause of death was pulmonary disease, and like a cardiac disease it happened in all different types of dementia. Death due to intracerebral hematoma occurred only in AD and VaD. None of the n = 8 patients who had died of intracerebral hematoma had been using oral anti-coagulant drugs (in the total cohort n = 10 patients). Six FTD patients were also diagnosed with MND, of whom one died of a myocardial infarction. Another FTD patient was known for a genetic C9Orf mutation without clinical signs of MND at presentation had an unidentified cause of death. The group with another cause of death included among others renal failure, sepsis, complications after operation for a hip fracture, and ileus.
As euthanasia is legal in certain circumstances in the Netherlands, four patients had died through euthanasia within two years after diagnosis (1 AD patient, 1 DLB, 1 CBD, and 1 FTD) (not included in our analyses). They were of a relatively young age, respectively 59 y, 55 y, and 49 y of age, except 1 patient aged 83 y with language disturbances because of FTD and a sister with known slowly progressive FTD.
Data on brain autopsy were available for n = 12 patients with rapid mortality. Brain autopsy was most frequently performed and confirmed in CJD (n = 4, of which n = 1 clinically had unspecified dementia). Other autopsy diagnoses were FTD (n = 4), PSP (n = 2, of which n = 1 presented with a clinical phenotype of CBS), AD (n = 1), VaD (n = 1), and idiopathic late onset cerebellar ataxia (clinically unspecified dementia with ataxia; n = 1). In one patient autopsy of the body was performed not including the brain, showing pneumonia as cause of death.
DISCUSSION
In our large cohort of relatively young dementia patients, short survival was fairly common (8–13%) and occurred in all different types of dementia, especially in non-AD dementia. In general patients were mildly to moderately demented and relatively young, but rapid mortality patients were older and showed a lower MMSE at baseline compared to patients without rapid mortality. Nonetheless, in line with our finding that even in a moderately young cohort rapid mortality was fairly common, high relative risk of mortality among ‘young-old’ aged 65–74 dementia patients has been shown before [4]. In view of the general life expectancy for people of 65 years of age in the Netherlands of 20 years, this clearly indicates that dementia reduces life expectancy in these young individuals [24].
We found more depressive symptoms in patients with rapid mortality compared to patients with non-rapid mortality. Hypothetically, patients might acquire depressive feelings related to awareness of rapid loss of cognitive function. We think it is less likely that the more severe depressive symptoms simply reflect a more advanced disease stage, as one would expect depressive feelings more prominently early on in the disease. Furthermore, pathologically, depression is thought to be multifactorial and heterogeneous and has also been linked to more severe AD pathology [25, 26] and cerebrovascular disease [27].
AD was the most frequent etiology of dementia. Contrary to our expectations and in contrast to earlier findings regarding CSF profile and rate of cognitive decline [28], we did not find specific characteristics for AD patients with short versus longer survival. Especially APOE genotyping and CSF-profiles were evenly distributed. Furthermore, occurrence of other dementia diagnoses like CJD, PSP, VaD, and FTD was evidently higher in patients with rapid compared to non-rapid mortality. In accordance with our results, a shorter survival in non-AD compared to AD patients has been described previously [10, 29]. Moreover, looking at specific dementia types, especially VaD and FTD, have shown high risks for a relative loss of life expectancy [5, 10]. Remarkably, VaD is generally not considered as a rapid progressive dementia, but our data emphasize that a rapid mortality has to be taken into account and underline the importance of taking care of vascular risk factors. Not surprisingly vascular risk factors were most prevalent in VaD. Regardless of dementia diagnosis, cardiac history was found to be more prevalent in rapid compared to non-rapid mortality patients. FTD was the second most prevalent diagnosis with rapid mortality, of which n = 6 were known to be FTD-MND. Survival in FTD is known to be highly variable, with reported mean survival of 2.3 years for FTD-MND compared to 8.6 years in FTD without MND and 12.6 years in progressive non-fluent aphasia [30]. Our study also showed that FTD patients without MND can have a short survival. Unfortunately, we have no pathological information, neither about tau-positivity or negativity in FTD nor the role of cerebral amyloid angiopathy in VaD. Moreover, as this was an observational cohort study, we cannot draw any conclusions about causality. Other more rare dementia diagnoses consisted of CJD (6 cases) and PSP or CBD (11 cases) of which five were autopsy confirmed. In view of the fact that CJD is a principal diagnostic consideration among rapidly progressive dementias, the total number in this cohort appears relatively low, stressing the fact that also with other types of dementia, the course of disease may be rapid.
Cause of death differed widely, with dementia itself followed by pulmonary and cardiac disease reported most frequently. Cause of death is a complex concept to evaluate. We assessed immediate cause of death, but possibly many causes are to trace back to the dementia itself as disease of origin, like pneumonia due to swallowing problems or heart disease due to exhaustion. High frequencies of pulmonary and cardiac disease have been reported before [31] and may reflect a terminal stage of disease. Intracranial hemorrhage occurred only in VaD and AD possibly related to cerebral amyloid angiopathy, but autopsy information was not available. An oncological cause was reported more frequently as cause of death in AD (13%, 7/55 patients) than expected and in contrast to current literature describing a decreased risk of mortality from neoplasms in AD [31, 32].
Strengths of the present study include the large study population and the unique setting. An unavoidable issue concerns the duration of dementia. We chose time of diagnosis at our clinic, which is something different than onset of symptoms. Furthermore, we defined self-reported duration of symptoms longer than five years before diagnosis as an exclusion criterion, in an attempt to prevent misdiagnosis of rapidity. It is possible that rapid progressive patients came to our clinic in a different stage of disease than slow progressors looking at the lower MMSE at baseline; however, duration of prior complaints was fairly comparable and does not seem to explain the observed difference. Furthermore, rapid mortality was defined as mortality within two years after diagnosis being less then some median reported survival of three years [30] and long enough to examine a substantial number of patients, but obviously this definition is arbitrary. Patients included were overall mildly to moderately demented when they presented at our clinic, possibly reflecting some selection bias as it is conceivable that the most rapid progressive patients (including CJD and severe affected patients after stroke) would have been directly admitted and possibly would have died in the hospital. Nonetheless, even in our cohort of outpatient clinic patients, a considerable proportion died rapidly. Furthermore, as we do not have follow-up data of a certain amount of patients, the number of patients with rapid mortality might even be underestimated. Unfortunately our sample size of rapid mortality patients was too small to examine certain characteristics according to etiology in further detail. Future research is needed possibly with the use of international defined criteria of ‘rapid progression’ to learn more about the differences that influence disease course within the various etiologies itself, like in AD, FTD, and PSP.
For patients and caregivers, it is important to be aware that certain patients will die relatively soon after diagnoses, and it could be preferable to be open about this in the doctor’s office. Hopefully future research, possibly by combining clinical and neuropathological data, will be able to find better predictors and applicable determinants to slow down these processes.
In conclusion, rapid mortality occurs in all types of dementia, especially also in non-AD dementias like VaD and FTD, and can be due to the rapid progressive dementia itself or be related to dementia-linked disease like FTD-MND or cardiovascular disease in VaD.
Footnotes
ACKNOWLEDGMENTS
Research of the VUmc Alzheimer center is part of the neurodegeneration research program of the Neuroscience Campus Amsterdam. The VUmc Alz-heimer Center is supported by Stichting Alzheimer Nederland and Stichting VUmc fonds. The clinical database structure was developed with funding from Stichting Dioraphte.
