Comorbidity in Dementia: An Autopsy Study Open Access

The number of cases of dementia is expected to increase in the United States as more people live longer. The most common cause of primary dementia is Alzheimer disease (AD).1 Evans et al2 estimated that as many as 10% of Americans older than 65 years have AD. As of the year 2000, the US National Center for Health Statistics ranked AD as the eighth leading cause of death (as compiled from death certificates), accounting for almost 50 000 deaths per year;3 this number is almost certainly an underestimate.4–6 Whereas previous autopsy studies have concentrated on delineating the neuropathologic findings and different types of dementias in clinically diagnosed demented subjects,7–16 relatively few autopsy series have described the comorbid systemic pathology found in demented patients who have undergone complete autopsies.5,17–19 Such studies suggest that demented patients are most likely to die of a respiratory infection and that a number of comorbid medical conditions are frequently underestimated by clinicians.

The University of California, Los Angeles (UCLA) Alzheimer Disease Research Center (ADRC) has an ongoing interest in the long-term medical care of demented patients. This includes obtaining consent for complete autopsy examination on a significant proportion of ADRC patients. The findings from these postmortem examinations provided the basis of our investigation into comorbid conditions present in clinically diagnosed demented patients at the time of death. Furthermore, we reviewed neuropathologic findings in demented patients who underwent either a complete or “brain-only” autopsy examination at our institution, with particular attention to evidence of ischemic brain injury and clinically unsuspected lesions (eg, tumors and infections). These findings have been published previously, in modified form, as an abstract.20 

We reviewed autopsy reports and available clinical information on patients whose brain specimens had been entered into the UCLA ADRC and Brain Bank, and who had complete autopsies performed during 1995–2000. The type of dementia (Alzheimer, frontotemporal, or vascular) for each case and any other significant neuropathologic findings were tabulated. Standard diagnostic criteria were used to assess the neuropathologic substrates of major types of dementia.21–23 A second component of our study consisted of a tabulation of neuropathologic findings from general autopsy cases together with those from partial autopsy (brain-only) cases. Brain-only autopsies were those accessioned into the UCLA ADRC during the years 1984–2001 (inclusive). These patients all carried a diagnosis of dementia, the majority having AD.

Gross and microscopic examinations of the brains were performed by a neuropathology fellow and neuropathologist (H.V.V.) in each case. Occasional cases were reviewed at a quarterly UCLA Neuropathology Quality Assurance Conference. Findings were tabulated with respect to (1) type of dementia, (2) semiquantitative assessment of the severity of cerebrovascular disease, and (3) evidence of ischemic damage in the brain. Ischemic damage was classified as macroinfarcts (>1 cm), lacunar infarcts (grossly visible but <1 cm), microinfarcts (not visible grossly), and hippocampal sclerosis, which was defined as pyramidal cell loss and astrogliosis in Sommer sector or more diffusely involving the pyramidal cell layer of sectors CA1, CA3, CA4, and the prosubiculum.23 Assessment of the degree of atherosclerotic arterial narrowing (none; mild, <20%; moderate, 20%–50%; or severe, >50%) was based on gross and microscopic estimates of stenosis of major branches of the circle of Willis, including basilar and vertebral arteries.

Among patients in whom complete autopsy was carried out, 52 patients with a clinical diagnosis of dementia were identified, including 27 men and 25 women, with an age range of 41 to 99 years and an average age of 77.6 ± 10.8 years. The causes of death are listed in Table 1, and significant pathologic findings for various organ systems are described in Table 2. There were 31 patients with pneumonia (bronchopneumonia/aspiration pneumonia) (Figure 1, A and B), 21 patients with evidence of recent or old myocardial infarcts (Figure 2, A), 19 patients with emphysema, 9 patients with pulmonary thromboembolism (Figure 1, D), 7 patients with ulcerations of their gastrointestinal tract (including 1 fatal stercoral ulcer), and 12 patients with incidental findings, such as benign gastrointestinal polyps and carcinoids. Several incidental findings were also found, including 12 gastrointestinal benign tumors (carcinoid, polyps, leiomyoma) and in 1 patient with AD, thrombosis of the superior sagittal sinus (Figure 6). We discovered a case of purulent ependymitis/ventriculitis (Figure 4) secondary to infective endocarditis complicating a mitral valve replacement in a 41-year-old man with frontotemporal dementia/amyotrophic lateral sclerosis, as well as 4 clinically unsuspected malignancies, including 1 widely infiltrative primary lymphoma of the central nervous system, 1 cerebral glioblastoma multiforme (Figure 5), 1 adenocarcinoma of the pancreas, and 1 adenocarcinoma of the lung (Figure 1, C). The patients with the glioblastoma multiforme and pancreatic adenocarcinoma also had AD; the patient with the adenocarcinoma of the lung had AD with ischemic lesions of the brain. A 41-year-old man presented with rapidly progressive dementia and was found to have lymphomatosis cerebri with a diffuse leukoencephalopathy. This case has been reported previously.24 

In addition, we reviewed the cases of 150 demented patients who underwent brain-only autopsies. Findings in these patients, combined with those from the 52 general autopsy cases, are shown in Table 3. Findings include (1) the majority (63.9%) of demented patients had AD, (2) 5.9% of the demented patients had relatively pure vascular dementia, (3) 47.1% of patients had evidence of variably severe atherosclerotic cerebrovascular disease, and (4) 19.3% of patients had hippocampal sclerosis involving one (7.2%) or both (12.1%) hippocampi. Of the patients with AD, 15.5% also had cerebral microinfarcts, 19.4% had lacunar infarcts (Figure 3), and 19.4% had hippocampal sclerosis.

Prior studies have attempted to quantify the extent of comorbidity and increased rate of mortality in elderly patients with and without psychiatric illness through retrospective reviews of death certificates or by comparison of autopsy results with clinical diagnoses.3–6,17–19,25–33 Zubenko et al32 studied 809 elderly psychiatric inpatients and noted a standardized mortality rate of 2.5 (observed deaths/expected deaths) in demented subjects during a 5.75-year surveillance interval. Patients with organic mental disorders (299 with dementia) had a mean of 5.8 active medical problems. Patients with AD showed an adjusted mean number of medical problems significantly lower than patients with other organic mental disorders (multi-infarct dementia, dementia, organic mood disorder, delirium, and organic mental disorder not otherwise specified).33 Comorbid conditions included circulatory problems; endocrine, nutritional, and metabolic disorders; and diseases of the skeletal muscle and digestive systems. A case-control autopsy study of 48 demented patients (29 with AD and 19 with vascular dementia) by Forstl et al31 showed that patients with AD were treated for an average of 2.0 internal medical problems. Patients with vascular dementia in this same study were treated for an average of 2.1 medical problems. In contrast, the autopsy results showed that, on average, AD patients had 3.7 internal medical problems and patients with vascular dementia had 4.1 internal medical problems. The value of the autopsy in documenting medical problems in the elderly was also emphasized by Zarbo et al,34 who found that 40% of autopsies revealed at least 1 unexpected finding that would have contributed to or changed patient management. In addition, 24% of autopsies demonstrated 1 major unexpected finding not contributing to death. In their review of 14 autopsy series addressing this issue, Zarbo et al noted a major unexpected finding at autopsy in 21% to 58% of cases. These investigators calculated a median diagnostic discrepancy rate of 34% between autopsy results and clinical diagnosis during the 20-year period preceding 1999.

Our results, like the relatively few previous studies, also demonstrate the utility of complete postmortem examinations in the demented elderly. Most important are those unexpected comorbid conditions that cause or contribute to death. Four patients were discovered to have previously unknown malignancies, including adenocarcinoma of the pancreas and the lung, glioblastoma multiforme of the brain, and a primary central nervous system lymphoma. Two of these patients with malignancies also had neuropathologic changes of AD, 1 had changes of AD and ischemic infarcts, and 1 had progressive leukoencephalopathy associated with primary central nervous system lymphoma. One patient, a 74-year-old woman with AD, had thrombosis of her superior sagittal sinus producing bilateral parasagittal regions of extensive hemorrhagic necrosis. The clinical management of some, if not all, of these patients would almost certainly have differed had these malignancies been discovered while the patients were alive. A study that reviewed 3118 autopsies completed at the Mayo Clinic in a period of 6 years (1994–1999) found that the autopsy identified a clinically undiagnosed malignancy in approximately 4% of cases.35 

Seven patients in whom complete autopsy was performed showed ulcers in the gastrointestinal tract; in at least 3 of these cases the findings were unexpected. In 1 of these patients, an unexpected stercoral ulcer caused sepsis and death. Although not often reported, undiagnosed gastrointestinal ulcer disease as an unexpected cause of death in the older population has been described in the forensic literature.36,37 Several studies have examined the morbidity of peptic ulcers in elderly patients, the role of medical and surgical treatments, as well as the role of prevention. It should be noted that some studies report unexpectedly favorable outcomes with interventional treatments when these lesions are diagnosed early in elderly patients.38–42 For demented elderly patients, especially for those early in the course of their illness (ie, when a patient may still be living at home with family or with home care skilled assistance, as well as for those living in nursing homes), similar treatment options should be weighed.

In 9 cases, pulmonary thromboembolisms were found at autopsy, and in 6 patients the thromboembolisms contributed to death. Three autopsies showed incidental arterial and venous thrombi as well: in one case within the pancreatic artery, in another case within the internal jugular and femoral veins, and in the third case within the superior sagittal sinus. One patient was also found to have nonbacterial thrombotic endocarditis. Pulmonary thromboembolism is a particularly difficult diagnosis to make prior to death.19,26,43 Gross et al26 performed a study on 234 institutionalized elderly patients who came to autopsy; they reported that the clinician's accuracy rate in making this diagnosis antemortem was only 39%. Keene et al19 made similar observations in their comparison of autopsy results with death certificates, concluding that this cause of death is missed in demented patients because clinicians do not recognize it.

The most common cause of death in our study was bronchopneumonia/aspiration pneumonia, which was found in 24 (46.1%) of the 52 patients who underwent complete autopsies. In several cases, this finding was associated with variable degrees of atherosclerotic cardiovascular disease. In some patients, both pathologies probably contributed to death. In 9 (17.3%) of 52 patients, death was due primarily to cardiac disease; 7 of these patients had atherosclerosis of their coronary arteries, 1 had idiopathic cardiomyopathy, and 1 had cardiac amyloidosis. No other medical conditions caused death as frequently as pneumonia or heart disease. Death certificates were not reviewed for this study, so the exact wording of the cause of death documented on the death certificate in each case is not known. However, death certificates and clinical diagnoses have frequently been reported to disagree with autopsy results.34,43–47 Our findings agree with previous investigations, which demonstrated that pneumonia is the most common cause of death among demented elderly. Pneumonia is frequently found in terminally ill bedridden patients, especially in those with dementia.5,18,19,25,26,28–30 This tendency is probably secondary to an inability by demented patients to effectively clear secretions, swallowing difficulties, and weight loss or debilitation compromising their immune systems. Atherosclerotic and ischemic cardiovascular disease is prevalent in elderly patients and is the most common cause of sudden, unexpected death in those living in the outpatient, nonhospitalized community.30 We found it frequently in our population. Kammoun et al18 suggested that cardiac failure is more than twice as common in patients with multi-infarct dementia than in patients with AD.18 

Additional noteworthy findings among subjects included (1) a clinically unsuspected case of meningoencephalitis and ventriculitis related to infective endocarditis in a man with frontotemporal dementia/amyotrophic lateral sclerosis; (2) 3 cases with neuropathologic changes consistent with AD in patients who did not have a specific diagnosis for their neurobehavioral abnormality while alive; and (3) 2 patients in whom the clinical impression was that of AD without neuropathologic changes to support this diagnosis. One of these latter 2 patients showed changes consistent with ischemic vascular dementia, and the other showed no significant neuropathologic changes, as the patient's symptoms were attributed to remote effects of a colonic carcinoma. In cases such as these, the autopsy helped elucidate the etiology of the patient's altered behavior and provided feedback to the care providers as a quality control measure.

Most patients in this study met neuropathologic criteria for AD. Patients with pure vascular dementia comprised only 5.9% of cases. This percentage is in agreement with previous findings of the prevalence of vascular dementia among autopsied dementia patients.13,48 Coexisting cerebrovascular lesions have been reported in approximately 50% of cases of autopsy-proven AD in recent series.49 Seventy percent of demented patients also showed coexisting cerebral infarcts, hippocampal sclerosis, or both. Microinfarcts and lacunar infarcts were present in 15.5% and 19.4% of AD patients, respectively. Nagy et al50 found microinfarcts, macroscopic infarcts of less than 2 cm, and/or cribriform change in 19.7% of autopsied dementia patients with confirmed AD. Hippocampal sclerosis was seen in 15% of autopsies carried out as part of a longitudinal study on aging.15 The same autopsy series found hippocampal sclerosis in 26% of demented patients aged 80 years or older. Some of these patients had more pronounced memory problems than difficulties in other cognitive areas. Hippocampal sclerosis was found in 19.3% of the demented patients in our series; it was bilateral in 12.1% and unilateral in 7.2%. The presence of cerebrovascular pathology has been shown to accelerate the development of dementia in patients with Alzheimer changes.15,50,51 Cerebrovascular disease also alters patterns of phosphorylated tau expression in AD.52 Comparatively less is known about the interaction of cerebrovascular disease with neuropathologic features of non-Alzheimer dementias. Data from autopsy examination of demented patients may help clarify these dynamics and may delineate the contribution made to AD by coexisting pathologic abnormalities.

We realize that a limitation of our study is the lack of a control population (ie, nondemented age-matched elderly autopsy subjects) with which to compare our findings. The literature shows that patients who die with dementia suffer as many or more comorbid medical problems as those without this diagnosis,33,53 and that pneumonia is a common terminal event in debilitated demented elderly. In one study by Zubenko et al,33 a population of elderly psychiatric inpatients was compared with nonpsychiatric inpatients admitted to short-stay general medical hospitals throughout the United States; psychiatric inpatients were found to have as many medical problems as inpatients admitted to general medical hospitals. Previous autopsy studies have demonstrated that when a comparison of demented and nondemented patients was performed, the immediate causes of death were similar.19,28 However, Beard et al25 performed a review of death certificate diagnostic codes and found that cardiovascular disease and neoplasms were more common in a control population than in patients with AD. It should be noted that causes of death for persons with AD may vary, depending on the stage of cognitive impairment in the patient at the time of death.19,28 In a study of AD patients with mild to moderate cognitive impairment, the causes of death (stroke, cerebral hemorrhage, neoplasms, and cardiovascular disease) differed from the causes found in patients who died later in the progression of AD, in whom bronchopneumonia was noted to be a more common cause of death.28 

In summary, this study documents the medical comorbidity found in demented patients referred to our institution for postmortem examination. Our results show that in this select population, comorbid conditions are vulnerable to underdiagnosis and misdiagnosis. We speculate that this is due, in part, to the patients' inability to communicate symptoms effectively. We found medical conditions that would have been painful and that would have required, at minimum, new or additional comfort care measures, such as narcotic therapy for cancer or occult infections potentially reversible with antibiotics. The autopsy of such patients may prove to be an important measure of the quality of chronic terminal care.

This work was supported by the University of California, Los Angeles (UCLA) Public Health Service, Alzheimer Disease Research Center grant P50 AG16570, PO1 AG12435, and by Public Health Service/National Institutes of Health (Bethesda, Md) grant T32 AG00093-21 (Dr Farag). Catherina Fu was the recipient of a UCLA Short Term Training Program Summer Fellowship.