Are Patients With OSAHS Cognitively Impaired and Why?

Our approach to this critical question was to review literature that specifically examined cognitive function in OSAHS using standardized neuropsychological measures. We chose to include only well-validated neuropsychological tests assessing a minimum of three cognitive domains for this review in order to examine the pattern of cognitive deficits evident in this population. In general, we found that language is spared in OSAHS. Interestingly, four of seven group comparison studies found that global cognitive functioning was spared; however, global cognitive improvement following treatment was noted in 6/10 treatment studies. Only one study made both between and within group comparisons to compare first untreated apnea patients to controls and then pre–post treatment changes in the apnea group (Bédard et al., 1993). The study found no differences between untreated apnea patients and controls on the WAIS–R Full Scale IQ or Verbal IQ, but reported a significantly lower WAIS–R Performance IQ in the untreated apnea patients. This latter deficit improved significantly with treatment. Thus, apnea patients may exhibit relatively few deficits in the global cognitive domain when compared to normal controls, but treatment may produce within-subjects improvements in performance. It is difficult to determine why these improvements would exist, especially since the array of tests included in this domain is quite vast. With regard to language, too few pre-treatment and treatment studies examined this construct to draw firm conclusions about the effects of OSAHS on this domain, however, there is no theoretical reason to believe that language skills would be affected by either sleep fragmentation or repeated hypoxemia.

Attention/vigilance, executive functioning, and memory were impaired in the majority of studies reviewed (six of eight, six of nine, and 7/11, respectively). Attention/vigilance is the most commonly assessed cognitive construct in OSAHS, and multiple studies assessing sustained attention using tests developed in the sleep field (e.g., Psychomotor Vigilance Test, PVT; Dinges & Powell, 1985; Steer Clear; Findley et al., 1989) have found deficits in this domain (e.g., Findley et al., 1999; Kribbs et al., 1993). Thus, attention/vigilance appears to be the most consistently affected cognitive domain in these patients. The relationships between OSAHS and measures of executive functioning and memory have been less thoroughly studied. Both of these cognitive domains present a challenge to investigators in that they are complex domains. For example, executive functioning can be tested using measures of set shifting (e.g., Trails B), working memory (e.g., Wisconsin Card Sorting Test), or verbal fluency (e.g., Controlled Oral Word Association). All are accepted tests of “executive functioning” and have been categorized similarly here, but each tests a distinctly different aspect of the various functional systems that lie within the frontal lobes of the brain. Because they represent different systems, performance on these tests is likely to be less consistent in any single cohort than would multiple tests of a different domain, such as attention. Memory testing suffers from the same methodological concerns, although to a lesser degree. In this case, test performance can be impaired for a number of reasons that would implicate distinctly different brain regions. For example, memory performance deficits can be attributable to complications in initial learning, free recall, or forgetfulness, each of which has its own functional implications relating to different brain regions. In fact, in this review, two studies (Feuerstein et al., 1997; Naegele et al., 1995) found primary learning impairments in the tally of studies finding “memory” impairments, while the remainder of the studies found deficits in free recall. Despite these limitations, patients with OSAHS appear to perform more poorly on tests of executive functioning and memory than do matched controls. Clearly, these findings have implications for the potential involvement of certain brain regions in this disorder. It is, however, difficult to determine the degree to which attention and executive functioning, which are frontally mediated, contribute to impairments in memory.

Although assessed less frequently than most other domains, psychomotor functioning and construction were impaired in OSAHS patients relative to normal controls in the majority of studies reviewed. Interestingly, these domains have rarely been discussed as primary sources of impairment in this population. The mechanism for psychomotor dysfunction in OSAHS is not obvious. Certainly, excessive sleepiness could result in slow motor performance, which would result in deficits on psychomotor tasks. If this were the case, however, tests of fine motor coordination (e.g., Grooved Pegboard) should be impaired to the same degree as tests of simple motor speed (e.g., Finger Tapping). All four studies assessing fine motor coordination found OSAHS patients to be impaired relative to control subjects (Bédard et al., 1991, 1993; Greenberg et al., 1987; Verstraeten et al., 1997), while neither study examining motor speed identified impairments (Knight et al., 1987; Verstraeten et al., 1997). Deficits on tests of fine motor coordination suggest the possible involvement of specific brain regions (e.g., the basal ganglia) in OSAHS that have not previously been the focus of investigation (for review, see Ring & Serra-Mestres, 2002). The constructional findings are more complex and have not been addressed specifically in the literature. Executive and motor coordination problems could account for a large percentage of the variance in constructional performance in this population, however, studies are needed to address this question directly.

The two most commonly implicated etiological mechanisms of cognitive dysfunction in OSAHS are sleep fragmentation (indexed by the RDI) and hypoxemia. There were no strong and consistent relationships between either measure of severity and any cognitive domain. Although fragmentation measures were consistent across studies (i.e., RDI and AHI) this was not the case for measures of hypoxemia. Some studies used nadir oxygen (e.g., Kingshott et al., 2000), while others used the number of desaturations beyond a specified criterion (e.g., Cheshire et al., 1992). While these measures are likely correlated, they represent very different aspects of the hypoxemia in OSAHS and may limit the likelihood of finding relationships. All but one study, however, identified a relationship between one or both of these clinical indicators of severity and at least one cognitive domain. In general, attention/vigilance and global cognitive functioning were most strongly related to some measure of severity. The other domains (executive functioning, language, memory, psychomotor functioning, and construction) appeared unrelated to these clinical indicators. It should be noted, however, that the two studies of Psychomotor functioning that used tests of fine motor coordination both found a relation between performance and measures of hypoxemia (Greenberg et al., 1987; Montplaisir et al., 1992). The only entirely negative study, which examined older adults with apnea (Knight et al., 1987), may have suffered from a restricted range in their OSAHS sample, since 5/10 participants exhibited only mild apnea and only 1 participant met criteria for severe apnea (RDI = 37.6). Given that apnea severity increases with age, mild apnea in an elderly group might not be clinically significant. It is possible, therefore, that cognitive dysfunction depends on a greater severity of apnea, especially in a group of older adults. Studies recruiting moderate to severe OSAHS patients consistently reported differences in performance between apnea and control groups. One study (Bédard et al., 1991) compared moderate and severe apnea patients to each other as well as to normal controls. The authors found that individuals with OSAHS performed more poorly than did normal controls on tests of executive functioning, memory, and psychomotor functioning. Severe patients were, however, more impaired than moderate patients on tests of global cognitive functioning, executive functioning and psychomotor functioning, suggesting a clear relationship between severity and cognition. It is possible that the relationship between cognition and apnea severity is not linear, even within a specified age range. Cognitive deficits might require a threshold of apnea severity, but there is clear evidence that a relationship exists at least among moderate to severe patients.

Does Treatment for OSAHS Improve Cognition?

Identifying the pattern of cognitive impairment in patients with untreated OSAHS can lead investigators to hypothesize about the etiology of cognitive deficits in this population. In particular, it would be interesting to identify if those cognitive functions related most to hypoxemia were also least likely to remit with treatment. This would suggest lasting effects of hypoxemia on cognitive functions via specific brain regions. The above findings, however, do not present a clear and consistent pattern of deficits in OSAHS. The most consistent findings may be in the domains less frequently tested (e.g., psychomotor functioning and construction). In fact, the findings of relatively consistent deficits in fine motor coordination and construction raise some interesting questions about the possible brain regions involved in the disorder. Psychomotor functions are in part reliant on the integrity of subcortical grey matter structures (Ring & Serra-Mestres, 2002). These structures are preferentially affected in other hypoxic populations and may be particularly susceptible to the repeated, intermittent hypoxemia seen in OSAHS (Caine & Watson, 2000). Moreover, fluctuations in oxygen availability, rather than fragmented sleep, are more likely to be associated with tissue damage.

One would expect that the functions associated with sleep fragmentation would improve rapidly with effective treatment; on the other hand, those associated with hypoxemia would likely not change rapidly and, indeed, may not change at all. This is only partially supported by the above review. Attention/Vigilance was associated with both sleep fragmentation (six of nine studies) and hypoxemia (5/10 studies), but it tended to remit with treatment (11/17 studies). Psychomotor functioning, on the other hand, was most resistant to the effects of treatment, suggesting that it may represent the more permanent effects of OSAHS. We would expect this domain to be most associated with hypoxemia. Table 3 shows that psychomotor functioning is equivocally associated with hypoxemia (two of four studies). We found, however, that the two studies using the grooved pegboard test both found associations with hypoxemia. It is therefore possible that psychomotor functions, and specifically fine motor coordination, represent a domain of deficit in OSAHS that may be the result of chronic, intermittent hypoxemia and may be more resistant to the effects of treatment. The brain regions subserving fine motor coordination may be particularly vulnerable to the hypoxemia resulting from OSAHS. These findings are consistent with those in other hypoxic populations (Cummings et al., 1984; Earnest et al., 1980; Kotila & Kajaste, 1984) and suggest that the subcortical grey matter might be preferentially affected by oxygen loss in apnea (see Caine & Watson, 2000, for review). Other candidate regions, including the cerebellum, have never been studied directly in OSAHS.

The hippocampus is often implicated as vulnerable in other hypoxic populations (Cummings et al., 1984; Petito et al., 1987; Rempel-Clower et al., 1996; Volpe & Petito, 1985), and memory was impaired in 7/11 studies in this review. It was not, however, consistently related to measures of hypoxemia, nor was it resistant to the effects of treatment (remitting in 8/15 studies), making it difficult to imply a cause and effect relationship between memory and oxygen loss in OSAHS. It should also be emphasized that it is difficult to ascribe a neuroanatomical substrate to the memory deficits noted in the studies reviewed because most studies evaluated free recall only and did not evaluate retention and/or recognition memory. More comprehensive assessment of memory processes in the future will provide greater insight into the brain regions that may underlie the memory deficits noted in OSAHS.

In summary, cognitive findings in OSAHS are not consistent, but findings regarding fine motor coordination may provide insight into the potentially lasting effects of intermittent hypoxemia in this disorder. The effects of hypoxemia on the brain are of particular interest to neuropsychologists. Specifically, based in part on the neuropsychological literature in OSAHS and the literature in other hypoxic populations, we are interested in the involvement of subcortical brain regions. The downstream effects of hypoxemia in OSAHS have been best studied in terms of cardiovascular involvement. Using this framework and existing theories of the effects of hypoxemia on vascular integrity in OSAHS, we can begin to develop plausible explanations of the possible permanent effects of OSAHS on the brain.

A Potential Mechanism of the Effects of OSAHS on the Brain

Investigators studying the cardiovascular effects of OSAHS have suggested that the disorder results in changes to the vascular structure and function, and that these changes are commonly seen in other hypoxic populations (Parkinson et al., 2002). Such theories have direct implications for the neuropsychologist studying OSAHS, as the same mechanisms that are implicated in the heart have implications for the vessels of the brain. Hypoxia typically results in almost immediate vasodilation, which serves as a protective mechanism to better distribute oxygen to the necessary organs. Researchers have shown, however, that people with OSAHS lack this protective response (Kato et al., 2000). One potential reason that OSAHS patients do not respond to hypoxemia as expected is that they do not suffer single, sustained hypoxic events; instead, they suffer repeated events (by definition more than 5/hr) where recovery time is often limited. In very severe cases, people suffer oxygen desaturations more than 100 times/hr. This results in breathing disturbances with approximately four seconds recovery time per episode. It is impossible to predict the response, protective or not, to such repeated hypoxemic events, but it seems that the vasculature would sustain some damage with such repeated events. Apnea investigators have already suggested that vasculopathy may be a direct result of OSAHS and have posited potential mechanisms. Lanfranchi and Somers (2001) have suggested that an increase in sympathetic vasoconstriction together with a decrease in vascular protective mechanisms result in changes to the structure and function of the blood vessel. Their model is described in detail elsewhere, but is summarized below (Lanfranchi & Somers, 2001).

The Lanfranchi and Somers model suggests that the hypoxemia seen in OSAHS results in a number of autonomic, humoral, and neuroendocrine responses that can lead to vasculopathy. The first, they believe, is chemoreflex activation, which results in an increase in sympathetic vasoconstriction in the peripheral vessels. This, in turn, results in increases in both blood pressure and the release of circulating catecholamines, both of which have a negative effect on the vessel. In one study, CPAP markedly reduced sympathetic activity and increases in blood pressure (Faccenda et al., 2001). The second mechanism of vasculopathy involves the pressor effect. Hypoxemia typically results in an increase in endothelin, a potent vasoconstrictor. Endothelin has sustained hypertensive effects and is increased in individuals with OSAHS. Increased vasoconstriction has obvious immediate implications for blood pressure and, if chronic, may cause permanent damage to the blood vessel itself. The third mechanism by which the vessel is damaged is through endothelial dysfunction. The endothelial cells line the wall of the vessel and, typically, produce nitric oxide in response to hypoxemia. Nitric oxide acts as a vasodilator, protecting the vessel from the constricting effects outlined above. Impaired endothelial function is characteristic of hypertension and atherosclerosis. These investigators have shown that individuals with sleep apnea have less of a protective nitric oxide response than do matched controls, suggesting that vasodilation is less likely to occur. The final mechanism put forth by Lanfranchi and Somers involves a decrease in cardiovascular variability with an increase in blood pressure variability. These authors suggest that this combination of cardiovascular variability derangements sets the stage for cardiovascular dysfunction and even heart disease. Together, this cascade of responses in OSAHS is thought to have profound, and possibly lasting, effects on both the structure and function of the blood vessel. Small vessels may be more susceptible to hypertension in general as well as these mechanisms of vasculopathy.

Recent data suggest that there is an association between apnea severity and small vessel ischemic brain disease (Aloia et al., 2001b; Colrain et al., 2002). Kamba et al. (1997) were the first to investigate cerebral metabolism in individuals with sleep apnea. These investigators utilized Magnetic Resonance Spectroscopy and showed that cerebral metabolism was lower in the cerebral white matter in participants with moderate–severe sleep apnea than in those with mild sleep apnea. In a follow-up study, Kamba et al. (2001) demonstrated that the observed metabolic change was specific to the cerebral white matter and not to the cortex, and that this relationship was independent of age. Consistent with these findings, Colrain et al. (2002) demonstrated a relationship between severity of subcortical white matter hyperintensities and level of hypoxemia in 41 identical twin pairs. The presence of these hyperintensities within the subcortical grey and deep white matter suggests the involvement of endothelial damage in these regions, where vascular hypoperfusion is more common.

Although investigators have developed theoretical models of the vascular effects of OSAHS and the presence of subcortical hyperintensities in patients with OSAHS, these models do not directly address the cognitive effects of the disorder. We believe that damage to the small vessels of the brain can result in a number of functional cognitive problems, without absolute specificity of dysfunction in any particular domain. Many patients with OSAHS do not show obvious cerebrovascular disease, but increases in the risk for cerebrovascular disease have been established independent of other risk factors like age and obesity (Lanfranchi & Somers, 2001). Preliminary data from our laboratory supports a relationship between white matter hyperintensities on brain MRI and aspects of cognition in a small sample of patients with OSAHS. Four older adults (age > 65) with severe OSAHS (AHI > 30) and 4 with minimal OSAHS (AHI < 10) were recruited from a sleep apnea clinic and matched for age and body mass index. All participants had controlled hypertension. Subcortical white matter hyperintensities were measured using a semi-automated quantification procedure. Results indicated that the group with severe OSAHS had more subcortical white matter hyperintensities on brain MRI than those with minimal apnea (Aloia et al., 2001b; see Figure 1). Moreover, there was a trend for a negative association between subcortical hyperintensities and free recall of a word list (Table 4). There have been no specific studies of the basal ganglia (a likely candidate given its predilection for damage in hypoxic states) and/or the cerebellum, but our group has begun to investigate these structures as well.

Correlations of MRI data with sleep, cognition, and mood

Subcortical hyperintensities among individuals with severe versus minimal apnea.

Taken together, this review identifies definite areas of cognitive compromise in OSAHS. We also believe that logical etiologies exist for such dysfunction and can lead investigators to a better understanding of the mechanisms of the disease itself. We expect to study the validity of the above theory over the next several years. We also hope that other laboratories begin to explore the functional consequences of this disorder, bearing in mind the benefits of a comprehensive, standardized neuropsychological battery for identifying valid and reliable patterns of cognitive dysfunction and response to treatment.