The perception of a change in cognitive or memory abilities is common in aging. Prior to the last decade, the meaning of subjective cognitive complaints has been unclear, as subjective cognitive complaints have not correlated well with objective decline in performance on standardized neuropsychological assessment (Mendes et al., 2008). However, there is increasing evidence to support that reports of subjective cognitive decline (SCD), even with normal performance on objective cognitive tests, is associated with increased likelihood of Alzheimer disease (AD) biomarker abnormalities and with an increased risk for future cognitive decline and AD (Jonker et al., 2000; Saykin et al., 2006; Visser et al., 2009; Jessen et al., 2010, 2014a; Reisberg et al., 2010; Rami et al., 2011; Amariglio et al., 2012; Wang et al., 2013; Vega and Newhouse, 2014). Although, SCD is non-specific and could potentially reflect numerous conditions such as normal aging, psychiatric conditions, neurologic and medical disorders, substance use, and medication effects (Kim et al., 2003; Reid and Maclullich, 2006; Elfgren et al., 2010; Slavin et al., 2010; Bartley et al., 2012; Vega and Newhouse, 2014), current evidence suggests that subjective memory complaints may be a predictor of later development of mild cognitive impairment (MCI) and dementia (Geerlings et al., 1999; Palmer et al., 2003; van Oijen et al., 2007; Reisberg et al., 2010; Waldorff et al., 2012; Genziani et al., 2013).

Several neuroimaging studies have found structural and functional brain differences in individuals with and without subjective cognitive complaints. Cross-sectional studies have shown reduced hippocampal volume in older adults with subjective memory complaints but normal performance on objective memory tests compared to individuals without subjective memory complaints (Saykin et al., 2006). Jessen and colleagues also showed reduction in entorhinal cortex volume in individuals with subjective memory complaints that was intermediate between healthy controls and patients with MCI (Jessen et al., 2007). Longitudinal studies have confirmed these findings and shown that subjective memory complaints have predicted volume decline in gray matter, particularly in medial temporal lobe structures (Stewart et al., 2011; Scheef et al., 2012; Hafkemeijer et al., 2013; Cherbuin et al., 2015). Glucose metabolism as measured by FDG-PET was shown to be altered in individuals with subjective memory complaints compared to controls (Scheef et al., 2012), and Rodda and colleagues found increased brain activation (BOLD signal) in task relevant brain regions in both encoding and attention tasks despite normal performance in older adults (Rodda et al., 2009, 2011). A longitudinal study found that subjective cognitive complaints not only predicted a decline in cognitive performance but also correlated with increased cerebral blood flow as measured by FDG-PET during a memory task (Hohman et al., 2011).

There is evidence to suggest subjective cognitive complaints are more common and may be more predictive of later cognitive dysfunction in women (Pérès et al., 2011). Memory complaints during or after the menopause transition are common. Approximately 60% of middle-aged women reported cognitive changes in the Seattle Midlife Women's Health Study (Sullivan and Woods, 2001), and 42% of postmenopausal women reported a negative change in cognition in the Study of Women Across the Nation (SWAN) (Bromberger et al., 2011). In addition to increased cognitive complaints during or after menopause, women also appear to be at higher risk for AD, particularly if they carry the APOEε4 allele (Bretsky et al., 1999). Studies that have examined objective cognitive performance in relation to the menopause transition have had mixed results (Hogervorst and Bandelow, 2010; Dumas et al., 2013). One study showed that subjective cognitive complaints were associated with menopause-related physical symptoms, psychological factors, and objective impairment (Schaafsma et al., 2010). Other studies have shown that subtle deficits in objective cognitive performance correlated with some measures of poorer subjective memory performance (Drogos et al., 2013; Thurston, 2013). Subjective memory complaints in perimenopausal women were found to be most associated with working memory and complex attention rather than verbal episodic learning or memory (Weber et al., 2012) suggesting that high effort demanding cognitive operations may lead to the perception of subjective cognitive difficulties. In a study of working memory examining a subset of participants used in the current study, Dumas and colleagues (Dumas et al., 2013) found that women with substantial postmenopausal cognitive complaints showed greater cortical activity (measured via BOLD signal) during working memory performance than women without such complaints despite equivalent performance, suggesting that cognitive complaints may indicate increased neural effort, perhaps as a form of compensation.

In addition to task related activity, resting state functional connectivity (rsFC) has been associated with measures of cognitive performance and behavioral state (Hampson et al., 2006; Keller et al., 2015). rsFC is a technique that can be used to identify brain networks by measuring the temporal correlations of low-frequency (<0.1 Hz) BOLD signal fluctuations between brain regions while the brain is at rest (i.e., not actively performing a goal-directed task). These low-frequency fluctuations are thought to reflect the intrinsic functional architecture of the brain (Fox and Raichle, 2007). rsFC has revealed that signals in functionally related brain regions correlate with each other even in the absence of external stimuli (Greicius et al., 2003). Differences in the connectivity of these functionally related brain regions can reflect changes in structural connectivity or efficiency of constituent brain circuits (Sheline and Raichle, 2013). rsFC studies have identified numerous brain networks, the most commonly investigated of which is the default mode network (DMN) (Raichle et al., 2001; Greicius et al., 2003). In addition to the DMN, rsFC studies have identified several other resting state brain networks associated with higher cognitive functioning (Yeo et al., 2011), such as the dorsal and ventral attention networks (DAN and VAN, respectively) (Fox et al., 2006), the salience network (Seeley et al., 2007), and the executive control network (ECN) (Seeley et al., 2007; Vincent et al., 2008). Connectivity strength of the DMN, ECN, Salience, and DAN have been associated with performance on cognitive tasks (van den Heuvel and Hulshoff Pol, 2010).

Recent studies have suggested that subjective memory complaints in the elderly may be associated with altered rsFC (Bajo et al., 2012; Hafkemeijer et al., 2013). Cognitive complaints after menopause may be a particularly meaningful marker for early neural dysfunction, especially as loss of estradiol at menopause has been hypothesized to contribute to the higher rates of dementia in women. To examine whether subjective cognitive complaints after menopause show a relationship to changes in functional connectivity, we examined rsFC in 31 middle-aged, post-menopausal women (aged 50–60) who were cognitively normal on objective tests, but self-reported an experience of cognitive decline since menopause. Since connectivity strength of the DMN, ECN, Salience, and DAN have been associated with performance on cognitive tasks in previous studies (van den Heuvel and Hulshoff Pol, 2010), those networks were included in the analysis. In addition to the functional connectivity networks chosen, additional seed regions that showed structural and functional differences between cognitive complainers and control groups were also included in the analysis (Saykin et al., 2006; Wang et al., 2006, 2013; Dumas et al., 2013). We hypothesized that postmenopausal women with significant subjective cognitive complaints would show increased functional connectivity in one or more resting state brain networks suggesting early neural compensation, potentially a harbinger of increased risk for late life cognitive dysfunction.

This study demonstrates relationships between rsFC and subjective cognitive complaints following menopause in otherwise cognitively normal, postmenopausal women. Specifically, higher CCI score was associated with greater functional connectivity in the ECN and within the temporal lobe, but reduced functional connectivity within the left frontal lobe. No differences were seen in other networks. We also observed that women with greater depressive symptoms and menopausal symptoms had weaker negative functional connectivity within the left middle frontal gyrus. It is possible that depressive symptoms and/or menopausal symptoms could be contributing to the weaker negative functional connectivity observed within the left middle frontal gyrus. However, both the BDI and menopause symptom checklist also contain items that evaluate cognition in addition to mood and somatic symptoms, therefore it is difficult to disentangle which symptoms are driving this effect. Previous work from our group has shown that in a subset of these participants, women with higher CCI scores had greater cortical activity in frontal and precuneus regions during a working memory task despite equivalent performance (Dumas et al., 2013). Dumas and colleagues concluded that this increased activation was a compensation response such that healthy middle-aged women with subjective cognitive complaints recruited a greater extent of the working memory network to maintain task performance. The increased resting state connectivity seen in this study may similarly reflect increased cognitive effort to maintain adequate performance for either normal or early pathologic cognitive aging processes. Although, we favor this hypothesis, we acknowledge the preliminary nature of these findings and the possibility that factors other than CCs mediate or contributes to the alterations in functional connectivity observed in this study. Future longitudinal studies examining connectivity during a task are needed to determine if increased connectivity does in fact reflect increased cognitive effort.

Increased connectivity has also been seen in populations at risk for the development of later life cognitive impairment (Filippini et al., 2009; Dennis et al., 2010; Sheline et al., 2010; Westlye et al., 2011). Our findings are consistent in some respects with those of Hafkemeijer and colleagues (Hafkemeijer et al., 2013), who found increased resting state functional connectivity in elderly normal adults with subjective memory complaints, although the predominant difference between the groups was found in the DMN and medial visual network rather than the ECN as was found in the current study. In the Hafkemeijer et al. study structural measures differed between the subjective complaint and control groups and correlated with increased functional connectivity (Hafkemeijer et al., 2013), however a preliminary structural analysis in the sample used for this study did not reveal substantial structural differences in gray matter volumes (unpublished data). Utilizing magnetoencephalography, Bajo and colleagues have also documented alterations in functional connectivity in healthy elderly subjects with increased subjective memory complaints compared to non-complaining elders and patients with MCI during the performance of a memory task (Bajo et al., 2012). Elders with subjective memory complaints showed lower functional connectivity in memory systems during task performance compared to cognitively normal and MCI patients (who showed increased synchronization), suggesting an initial decrease in functional connectivity, and later compensation through increased connectivity as cognitive difficulties progress from subjective to objective deficits.

Data from the Seattle Midlife Women's Health Study (Woods et al., 2000) suggest that nearly half of postmenopausal women report noticeable cognitive symptoms, including attention, concentration, and memory problems. Approximately a third of women reported these problems as at least moderate in severity. Subjective cognitive complaints in the postmenopausal period may be a useful index of developing cognitive and/or brain dysfunction. Older adults with subjective cognitive complaints but normal cognitive performance have been shown to convert to dementia at higher rates than those without complaints (Reisberg et al., 2010) and have structural and functional changes in the brain that may indicate developing neurodegenerative disorders (Saykin et al., 2006; Rodda et al., 2009; Lamar et al., 2011). Cognitive complaints have also been associated with the menopause transition (Weber and Mapstone, 2009) and related to poorer working memory and encoding performance, while data showing objective cognitive impairments after menopause are inconsistent (Hogervorst and Bandelow, 2010).

Women who notice significant cognitive disruption following menopause may be vulnerable to disruption of cholinergic or other brain systems due to lower cognitive reserve, early neurodegeneration, or other factors. Thus, the loss of estrogen support at menopause may be more noticeable in terms of cognitive processes, particularly attention, executive function, and verbal memory in these vulnerable women. Cognitive complaints or changes in cognitive performance after menopause may be an important indicator of risk for late life cognitive impairment. Accumulating evidence suggests that subjective cognitive complaints have significant predictive value in assessing the risk of development of dementia in elderly individuals (Jessen et al., 2014b). Whether such complaints in the first few years after menopause can be similarly used to identify those at increased risk for cognitive dysfunction or dementia in late life will require longer follow-up, but early changes in brain connectivity as shown in this study or changes in task-related cortical activity (Dumas et al., 2013) suggest that the neural manifestations of early cognitive change may be apparent many years prior to the development of measurable cognitive impairment. Information from longitudinal brain aging studies has suggested that it is possible to identify a subgroup of otherwise normal individuals in middle-age who show evidence of pathologic brain changes (e.g., increased beta amyloid deposition) who may be expected to be at elevated risk for the development of AD later in life (Rodrigue et al., 2012). Previous studies indicate that in cognitively normal older adults, amyloid burden is increased in individuals with cognitive complaints (Amariglio et al., 2012; Perrotin et al., 2012).

Successful decision-making and action depend on accurately evaluating the success of basic cognitive processes that contribute to thought and behavior, a capacity known as “metacognition” (Metcalfe and Shimamura, 1994). Memory self-efficacy can be regarded as an aspect of metamemory or metacognition. Subjective memory complainers have a lower rating of their own memory capacity and therefore a lower memory-related self-efficacy than people without subjective memory complaints (Ponds et al., 1992; Ponds and Jolles, 1996). Furthermore, low memory self-efficacy is a key features of memory complainers (Metternich et al., 2009). The current findings suggest that subjective cognitive complainers may have some impairment in metacognition. Medial and lateral networks in anterior prefrontal cortex (aPFC) support metacognitive ability for memory and perception (Baird et al., 2013). BOLD signal in right posterior-lateral BA10 was positively correlated with metacognitive accuracy (Yokoyama et al., 2010). Convergent evidence indicates that frontopolar Brodmann area 10, and more generally the aPFC, support the human capacity to monitor and reflect on cognition and experience (Fleming and Dolan, 2012; Baird et al., 2013). Prefrontal cortex integrity seems to play an essential role in metacognitive judgments. The negative correlation between connectivity in the left middle frontal gyrus and CCI in this study may indicate that metacognitive impairments in subjective cognitive complainers are related to decreased connectivity with aPFC. Impairment of prefrontal cortex integrity may lead to metacognitive dysfunction, in response to which subjective cognitive complainers may use more neural resources to compensate than non-complainers (Dumas et al., 2013). This compensation explanation may be relevant to the finding of increased functional connectivity within the medial temporal lobe in MCI (Das et al., 2013).

Since the loss of estradiol is the critical hormonal event with impact on brain function that occurs at menopause, the question of whether a period of hormonal treatment could alter either the cognitive complaints or the neural representation of the cognitive complaints (increased connectivity and task-related activity) deserves further study. The “critical period hypothesis” (Resnick and Henderson, 2002; Maki, 2006; Sherwin, 2006, 2007) suggests that estradiol has maximal protective benefits on cognition in women when it is initiated proximal to the menopause but may be ineffective when initiated many years or decades later. What has not been clearly established is whether there is a subgroup of women for whom postmenopausal hormone treatment has more long-term benefits on cognition. Whether individuals with cognitive complaints following menopause are at higher risk for age-related cognitive decline and/or cholinergic dysfunction and would benefit from postmenopausal estradiol treatment will require further study.

PN, BM, AS, MN, and JD, contributed to the conception or design of the work; KA, LZ, contributed to data collection; LZ, KA, BB, NW, JV, JD, PN AM, conducted the data analysis and/or contributed to interpretation of the data; JV, LZ, drafted the article; JV, KA, PN, BM, AS, JD, KA, and JP, contributed to critical revision of the article, and all authors gave final approval of the version to be published.