Reduced frontopolar activation during verbal fluency task associated with poor social functioning in late-onset major depression: Multi-channel near-infrared spectroscopy study


*Kazuyuki Nakagome, MD, PhD, Division of Neuropsychiatry, Department of Multidisciplinary Internal Medicine, Tottori University Faculty of Medicine, 36-1 Nishi-cho, Yonago, Tottori 683-8504, Japan. Email:


Aim:  Functional neuroimaging studies to date have indicated prefrontal dysfunction in late-onset major depression (LOD). The relationships between prefrontal dysfunction and clinical characteristics including social functioning, however, have been unclear. The objective of the present study was to evaluate prefrontal hemodynamic response related to an executive task in LOD and to assess the relationship between activation in the prefrontal regions and clinical characteristics including social functioning.

Methods:  Twenty-four subjects with LOD and 30 age- and gender-matched healthy subjects were recruited for the present study. Hemoglobin concentration changes in the prefrontal and superior temporal cortical surface area were measured during verbal fluency task (VFT) using 52-channel near-infrared spectroscopy (NIRS), which enables real-time monitoring of cerebral blood volume (CBV) in the cortical surface area.

Results:  The two groups had a distinct spatiotemporal pattern of oxy-hemoglobin concentration change; LOD patients had less activation in a broad area covering both prefrontal and superior temporal cortices than healthy controls. In addition, reduced activation of the frontopolar region had a significant positive correlation with lower self-assessment of social functioning scores in the patient group.

Conclusion:  Reduced frontopolar cortical activation was associated with social functioning impairment in patients with LOD, and NIRS may be an efficient clinical tool for monitoring these characteristics.

MAJOR DEPRESSION (DSM-IV) is characterized by a marked deterioration in affect as well as a significant impairment in cognitive function,1,2 especially for older patients.1,3 Cognitive dysfunction has a severe impact on the patient's ability to cope with the demands of daily living. One major aspect of cognition relevant to social functioning is executive function. It has been defined as that aspect of cognition affording the ability to deviate from a stereotyped behavior locked to environmental stimuli. The neural basis of executive function appears to be the prefrontal cortex, a region also involved in other high-level cognitive functions such as working memory and language processing. Considering the significance of social functioning in psychiatric patients, elucidating the relationship between the neural activity in the prefrontal cortex and clinical characteristics including social functioning in depression is clearly important.

Many, but not all, neuropsychological studies of major depression have demonstrated impairment in executive function, using the verbal fluency task (VFT). Verbal fluency deficits have been reported in patients with major depression,4–6 but such findings are not universal.7,8 Neuroimaging studies have demonstrated reduced activity in the prefrontal cortex during VFT,9 but the relationship between the neural activation and performance level on the fluency task or clinical characteristics has not been clarified.

Multi-channel near-infrared spectroscopy (NIRS), a recently developed functional neuroimaging technology, enables the non-invasive detection of spatiotemporal characteristics of brain function near the brain surface.10,11 NIRS has enabled bedside measurement of the concentrations of oxygenated ([oxy-Hb]) and deoxygenated hemoglobin ([deoxy-Hb]) in capillary blood vessels. Assuming that hematocrit is constant, the changes in [oxy-Hb], [deoxy-Hb] and also [total Hb] (summation of [oxy-Hb] and [deoxy-Hb]) are correlated with the regional cerebral blood volume changes, as shown on simultaneous NIRS and positron emission tomography (PET).12–14 In contrast to other neuroimaging methodologies such as functional magnetic resonance imaging (fMRI), PET, electroencephalography (EEG) and magnetoencephalography, NIRS can be done under a more restraint-free environment that is especially suitable for psychiatric patients. Indeed, NIRS has been used to assess brain functions in many psychiatric disorders.15–18

Matsuo et al. showed reduced [oxy-Hb] activation of the prefrontal cortex during VFT in late-onset depression (LOD) patients using NIRS.18 The authors also found that the vasomotor reactivity to carbon dioxide (CO2) inhalation in the prefrontal cortex was significantly lower in the patients than in the healthy controls. Together with the tendency of negative correlation between the prefrontal activation and the severity of hyperintensity in the periventricular region, the authors concluded that the altered prefrontal vasodilator response as shown on NIRS is involved in the pathophysiological basis of prefrontal dysfunction in LOD patients. The study, however, had some shortcomings: four out of 10 patients were medicated with selective serotonin re-uptake inhibitors (SSRI), which may also alter the vasomotor reactivity and, moreover, the authors failed to find either significant correlation between the reduced [oxy-Hb] activation and performance of VFT or any clinical index related to depression.

Recently, Takizawa et al. demonstrated the association of reduced [oxy-Hb] activation induced by VFT in the frontopolar region with functional impairment assessed on global assessment of functioning in patients with schizophrenia using 52-channel NIRS.19 It is of interest to test whether similar findings could be obtained for other psychiatric disorders, which may indicate universal relevance of the frontopolar region to social functioning.

Social Adaptation Self-Evaluation Scale (SASS) is a 21-item scale developed for the evaluation of patients' social motivation and behavior in depression by Bosc et al.,20 and the reliability and validity of the Japanese version have already been confirmed.21 Each item is scored from 0 to 3, corresponding to minimal and maximal social adjustment, with a total score range of 0 to 60. SASS was developed to meet two requirements: simplicity of use and measurement of social behavior. The primary objective of the present study was to investigate more precisely the relationship between activity in the prefrontal cortex and clinical characteristics including SASS scores, in unmedicated patients with LOD, using 52-channel NIRS (ETG-4000, Hitachi Medical, Tokyo, Japan). Secondarily, in an attempt to ascertain whether [oxy-Hb] activation is determined by vasodilator reactivity, we compared the prefrontal activity induced by VFT between patients with and without vascular depression. If [oxy-Hb] activation was determined by vasodilator reactivity, reduced [oxy-Hb] activation should be more prominent in the vascular depression (VDep) subgroup as compared with that in the non-vascular depression (non-VDep) subgroup. We hypothesized that activity in the prefrontal cortex associated with executive function should be related to social functioning rather than severity of depressive symptoms in patients with depression, likewise the case for schizophrenia. Moreover, we presumed that the [oxy-Hb] activation induced by VFT should not solely be determined by vasodilator reactivity, but should reflect prefrontal cortical function. Therefore, we predicted that any difference between the VDep and non-VDep subgroups in terms of [oxy-Hb] activation should rely on the cognitive deterioration in the former as compared to the latter subgroup.



Antidepressant-naive LOD patients were recruited among outpatients at Tottori University Hospital in Tottori, Japan. The inclusion criteria included a diagnosis of a first episode of major depressive disorder in patients >65 years according to DSM-IV criteria. The diagnosis was obtained using the Mini-International Neuropsychiatric Interview (MINI).22 None of the subjects had clinical evidence of other central nervous system (CNS) disorders based on history and medical examination. Mental status was examined using the Mini-Mental State Examination (MMSE) with a cut-off point of 20. Patients with previous head trauma, stroke, electroconvulsive therapy, current or previous substance abuse and psychotic symptoms were excluded from participation. Twenty-four individuals (six male, 18 female) meeting these criteria participated in the investigation. Fourteen patients (four male, 10 female; mean age, 73.3 ± 5.6 years) were included in the Vdep subgroup, whereas 10 patients (two male, eight female; mean age, 70.9 ± 5.2 years) were classified as non-VDep according to the clinical definition proposed by Alexopoulos et al.23

Individuals who provided appropriate age, gender and MMSE matches for LOD patients participated as controls in the present study. Inclusion criteria for controls were the same as those for the patient sample, although controls were additionally required to have no previous or current psychiatric illness. Thirty individuals (14 male, 16 female) meeting these criteria were selected to participate in the study. All participants were right-handed with criteria of >80% according to the Edinburgh Inventory Index.24 Personal and clinical details are summarized in Table 1.

Table 1.  Subject characteristics
DemographicsPatients (n = 24)Controls (n = 30)Group differenceVDep (n = 14)Non-VDep (n = 10)Group difference
Mean ± SDMean ± SDMean ± SDMean ± SD
  • Fisher's exact test, otherwise Student's t-tests were used for between-group comparison.

  • BDI, Beck Depression Inventory; HAMD, Hamilton Rating Scale for Depression; MMSE, Mini-Mental State Examination; SASS, Social Adaptation Self-Evaluation Scale; VDep, vascular depression.

Age (years)72.3 ± 5.572.0 ± 4.7NS73.3 ± 5.670.9 ± 5.2NS
Gender (% female)75.053.3NS71.480.0NS
Education (years)10.6 ± 2.211.4 ± 1.9NS10.2 ± 1.611.2 ± 2.9NS
MMSE27.3 ± 1.727.9 ± 2.1NS27.6 ± 2.027.1 ± 1.4NS
HAMD18.4 ± 3.8 18.3 ± 4.018.6 ± 2.9NS
BDI20.1 ± 9.55.2 ± 3.7P < 0.00117.9 ± 9.023.6 ± 9.7NS
SASS30.0 ± 5.541.5 ± 7.0P < 0.00130.8 ± 6.128.8 ± 4.6NS

All subjects provided written consent after receiving comprehensive information on the study protocol. The study was approved by the ethics committee of Tottori University Faculty of Medicine.

Clinical evaluation

Prior to scanning, all the subjects undertook two self-assessments of depression severity and level of social functioning: the Beck Depression Inventory (BDI) and SASS21 In addition, only the patients were assessed for depression severity using the Hamilton Rating Scale for Depression (HAMD) by two trained psychiatrists.

Activation task

The task procedure in the present study was similar to that in the Takizawa et al. study.19[Hb] changes were measured during VFT (letter version). Each subject sat in a comfortable chair and was instructed to minimize movement such as head movements, strong biting and eye blinking during the NIRS measurements, so as to avoid artifacts.

The cognitive activation task included a 30-s pre-task baseline, a 60-s VFT, and a 70-s post-task baseline. For the pre- and post-task baseline periods, the subjects were instructed to consecutively repeat the five Japanese vowels (‘a’, ‘i’, ‘u’, ‘e’, ‘o’) aloud. The subtraction method (task minus pre- and post-task baseline) minimized the vocalization effects during VFT. During the task period they were instructed to generate as many Japanese words beginning with a designated syllable as possible. The three sets of initial syllables (A /to/, /se/, /o/; B /a/, /ki/, /ha/; C /na/, /i/, /ta/) were presented in counterbalanced order among the subjects and each syllable changed every 20 s during the 60-s task. Because the number of words generated by the three sets was not significantly different (mean ± SD: A, 11.2 ± 4.4; B, 13.5 ± 3.9; C, 11.9 ± 3.6; F[2,53] = 1.4, n.s., in a one-way anova using sets as an independent variable), the total number of correct words generated during VFT was adopted as a measure of task performance.

NIRS measurement

The NIRS measurement procedure was also similar to that used in the Takizawa et al. study.19 Fifty-two-channel NIRS (ETG-4000, Hitachi Medical) measures relative changes of [oxy-Hb] and [deoxy-Hb] using two wavelengths (695 nm and 830 nm) of infrared light based on the modified Beer–Lambert law.25 In this system these [Hb] values include a differential pathlength factor (DPF). The distance between pairs of source–detector probes was set at 3.0 cm and each measuring area between pairs of source–detector probes was defined as a ‘channel’. It is supposed that the machine measures areas at a few centimeters' depth from the scalp, that is, the surface of the cerebral cortex.26,27 The probes of the NIRS machine were fixed with 3 × 11 thermoplastic shells, with the lowest probes positioned along the Fp1–Fp2 line according to the international 10–20 system used in EEG. The arrangement of the probes can measure [Hb] values from bilateral prefrontal and superior temporal cortical surface regions. The correspondence of the probe positions and the measuring area on the cerebral cortex was approximated by superimposing the measuring positions on MRI of a 3-D reconstructed cerebral cortex made by averaging 17 healthy volunteers' brain images normalized to the MNI152 standard template (Fig. 1).28 MRI was done using a 1.0-T MR scanner (Magnex EPIOS10, Shimadzu, Kyoto, Japan). A T1-weighted, 3-D gradient-echo sequence was used to produce 256 successive axial slices 1 mm thick covering the entire head (repetition time 37 ms, echo time 10 ms, flip angle 30°, field of view 230 mm × 193 mm, 256 × 214 matrix).

Figure 1.

Probe setting and measurement points for 52-channel near-infrared spectroscopy (NIRS). (a) The probes with 3 × 11 thermoplastic shells were placed over a subject's bilateral frontal regions. (b–d) The 52 measuring positions of the NIRS machine are superimposed on 3D-reconstructed cerebral cortical surface from magnetic resonance imaging (MRI) made by averaging 17 healthy volunteers' brain images normalized to the MNI152 standard template. The channel numbers are indicated above the measuring points. (e) The 52 measuring areas are labeled ch1–ch52 from the right posterior to the left anterior.

The rate of data sampling was 0.1 s. The obtained data were analyzed using the integral mode; the pre-task baseline was determined as the mean over a 10-s period just prior to the task period, and the post-task baseline was determined as the mean over the last 5 s of the post-task period, and linear fitting was applied to the data between these two baselines. A moving average method using a window width of 5 s was applied to remove any short-term motion artifacts. The moving average method alone was not able to remove all the artifacts and thus, we used a semi-automatic method for removing those data with significant artifacts. First, we applied the algorithm developed by Takizawa et al.19 that enables a fully automatic rejection of data with artifacts separately for each channel using quantitative evaluation, although the algorithm appeared to reject even those data that were relatively free from artifacts. Therefore, in the next step, two researchers, who were both blind to the clinical variables of the data, judged whether or not to save those data rejected by the algorithm through consultation. Consequently, the number of averaged data for each channel did not vary widely within each diagnostic group (LOD, n = 23–24, mean, 23.9 ± 0.19; healthy subjects, n = 28–30, mean, 29.8 ± 0.44; percentage: LOD, 99.8%; healthy subjects, 99.4%, n.s.).

Statistical analysis

First, the task performance level was compared between groups using Student's t-test (LOD and control; Vdep and non-VDep). Next, for the analysis of the hemodynamic response data, [Hb] variables of each channel were averaged for the two time segments (pre-task baseline and task period). We focused on [oxy-Hb] concentrations, because [oxy-Hb] change (task period – pre-task baseline period) was assumed to more directly reflect cognitive activation than [deoxy-Hb] change, as shown by a stronger correlation with blood-oxygenation level-dependent (BOLD) signal measured on fMRI.29 The mean [oxy-Hb] changes during the 60-s task period were compared between the two groups (LOD and control; VDep and non-VDep) for each channel using Student's t-test. Because we performed 52 t-tests, the correction for multiple comparisons was made using false discovery rate (FDR). We set the value of q specifying the maximum FDR to 0.05, so that there were no more than 5% false positives on average.30 Finally, for LOD patients, Pearson's product moment correlation coefficients were calculated for a relationship between the mean [oxy-Hb] changes during the task period and the clinical characteristics such as HAMD, BDI and SASS scores for each channel. We again adopted an FDR-based procedure for the multiple testing correction in correlational analyses for 52 channels and identified those channels for which r values reached a significance level of P < 0.05 (FDR-corrected). Additionally, we similarly analyzed the correlation between [oxy-Hb] changes and task performance of VFT, age, and duration of illness in LOD patients. Statistical analysis was performed using SAS 9.1 software (SAS Institute Japan, Tokyo, Japan).


Task performance

The number of generated words on the VFT for LOD patients (mean, 12.1 ± 4.3) was not significantly different from that for the controls (mean, 12.6 ± 3.6; t = −0.45, n.s.). Moreover, there was no significant difference between the VDep (mean, 11.1 ± 3.3) and non-VDep (mean, 13.5 ± 5.2) subgroups (t = −1.39, n.s.).

[oxy-Hb] activation

The grand averaged waveforms of [Hb] variables during cognitive activation in controls and LOD patients are shown in Fig. 2(a,b).

Figure 2.

Grand averaged waveforms in (a) late-onset depression (LOD) patients (n = 24), (b) controls (n = 30), and (c) P significance map of t-tests for [oxy-Hb] activation in LOD patients as compared to controls. †Distance between the two vertical dotted lines represents the activated period (60 s).

LOD patients were associated with significantly smaller [oxy-Hb] changes compared to controls in 40 channels (FDR-corrected P = 0.001 to 0.037). Fig. 2(c) is a P significance map of the t-tests, that is, LOD versus control, which shows significant group differences in a broad area. Meanwhile, there was no group difference in [oxy-Hb] changes between VDep and non-VDep subgroups in any channel (Fig. 3).

Figure 3.

Grand averaged waveforms of [oxy-Hb] concentration changes in vascular depression (VDep; n = 14) and non-vascular depression (non-VDep; n = 10) subgroups. No significant group difference was scored in any channel. †Distance between the two vertical dotted lines represents the activated period (60 s).

In LOD patients the mean [oxy-Hb] changes had a significantly positive correlation with SASS scores in 13 channels (R, 0.52–0.65), with the highest correlations located approximately in the frontopolar and right dorsolateral regions (Fig. 4), whereas task performance during VFT was not significantly correlated with SASS scores. Significant correlations between the SASS scores and mean [oxy-Hb] changes in the channels located in the similar regions were observed in the VDep subgroup (14 channels; R, 0.67–0.83) but not in the non-VDep subgroup or controls (Fig. 4). Moreover, the mean [oxy-Hb] changes did not show any significant correlation with the task performance during VFT or other clinical variables, such as age, duration of illness, HAMD, or BDI in LOD patients.

Figure 4.

Cortical distribution of a significant correlation between [oxy-Hb] changes and SASS scores in (a) late-onset depression (LOD) patients and (b) vascular depression (VDep) patients. The channels with a significant correlation (Pearson's product moment correlation; false discovery rate (FDR)-corrected P < 0.05) are indicated by color. (c) Scattergraph showing the relationship between Social Adaptation Self-Evaluation Scale (SASS) scores and [oxy-Hb] activation in CH25. non-VDep, non-vascular depression; NC, normal controls.


In the present study it was shown that [oxy-Hb] activation during VFT was significantly smaller in LOD patients as compared with age- and gender-matched healthy controls, which was not explained by differences in task performance. Furthermore, the smaller [oxy-Hb] change induced by cognitive activation was significantly associated with severer social functioning impairment in the VDep subgroup of LOD patients. And the area of activation that showed significant association with SASS scores was localized mostly in the frontopolar region. These results suggest that reduced frontopolar cortical activation may be associated with social functioning impairment in patients with VDep and that NIRS may offer promise as a non-invasive clinical measurement tool for evaluating social functioning.

In the VFT adopted in the present study, initial syllables were changed every 20 s, which was a much shorter time period for one syllable than that used in other comparable studies. This procedure was adopted with the aim of reducing the variance of time during which the subjects remained silent, which may correct the data for activation due to vocalization. This procedure, however, may have led to non-significant group difference in the VFT performance.

In the present study LOD patients had reduced [oxy-Hb] activation in widespread regions including the prefrontal and superior temporal cortices despite absence of deficiency in task performance. The finding indicates altered vasomotor response, presumably caused by microvascular dysregulation in LOD patients. The VDep subgroup, however, which was assumed to be more extensively affected by microvascular disease than the non-VDep subgroup, had similar [oxy-Hb] activation level to the non-VDep subgroup, indicating that altered vasomotor response could not fully explain the group difference. Meanwhile, SASS scores and [oxy-Hb] activation were significantly correlated in the VDep but not in the non-VDep subgroup. The findings suggest that hemodynamic responses to VFT are more relevant to social functioning in patients with VDep than non-VDep. Assuming that vasomotor function is more impaired in the VDep than the non-VDep subgroup, low social functioning in patients with VDep may arise from poor vasomotor response, presumably associated with microvascular dysregulation.

With regard to the group difference in [oxy-Hb] activation, the difficulty in making a real-time measurement of the accurate DPF in vivo is a major concern and prevents absolute quantification of NIRS variables.31 In the NIRS system, ‘hemoglobin concentration change × DPF’ (ΔC × L) is calculated as a solution to the simultaneous equations based on the Beer-Lambert law, which includes the effect of DPF. Meanwhile, Zhao et al., using a Monte Carlo simulation, reported that the estimated DPF variation in the forehead region of adult humans was approximately homogeneous.32 Moreover, from a practical point of view, the characteristics of time course pattern in the NIRS signals of the prefrontal cortex were found to be significantly different between psychiatric samples and healthy controls during VFT, despite the lack of difference during motor activation task (finger tapping).15,16 These findings suggest that the DPF variation alone could not account for the group difference in the NIRS signals of the prefrontal and superior temporal cortices during VFT. The homogeneity of DPF in the forehead, however, has not been well established in older patient samples and DPF may well be affected by the degree of brain atrophy. One of the limitations of the present study was that we did not take into consideration the structural changes of the brain, which should be done in future studies by examining the subjects' MRI or comparing [oxy-Hb] activation during a specific task with those during other tasks, such as finger tapping.

The present study indicates an interesting relationship between VDep patients' self-assessment of social functioning and [oxy-Hb] activation in the prefrontal region, more specifically in the frontopolar region. The characteristics of the VFT demands may recruit the frontopolar cortex as well as the lateral prefrontal cortex. Social functioning requires complex operations of executive function that include monitoring, reasoning, organizing, selecting and planning. Burgess et al. noted that the frontopolar region is involved in high-level executive control and thus, is likely to be a vital component of social functioning.33

Recent advances in neuroscience have indicated that the prefrontal cortex is cytoarchitectonically subdivided into several regions: the dorsolateral (BA 9/46), ventrolateral (BA 47), and frontopolar (BA 10) regions.34,35 These regions, however, are yet to be successfully mapped to specific cognitive functions. The frontopolar cortex comprises the most anterior part of the frontal lobe, which is one of the least well understood regions of the human brain. The frontopolar cortex has been suggested to have become enlarged and specialized during hominid evolution,36 and is assumed to provide a higher level of control to coordinate ventrolateral and dorsolateral functions in order to maximize task performance.35,37 One important feature of the frontopolar cortex is that the number of dendritic spines per cell and the spine density are higher than in other regions of the prefrontal cortex.38 This indicates that the functional properties of the frontopolar cortex are more likely than those in other prefrontal regions to be involved in the integration process. Moreover, the frontopolar cortex seems not to be interconnected with downstream regions such as the primary cortex in the way that other prefrontal regions are. It is the only region that is predominantly interconnected with the supramodal cortex,39,40 indicating that the frontopolar cortex is involved in the most abstract level of processing. Researchers developed a number of models in their attempt to describe the functions of the frontopolar region, but their results have been inconclusive. Christoff and Gabrieli proposed that frontopolar activations become recruited when internally generated information needs to be evaluated.41 Ramnani and Owen found that the coordination of information processing and informational transfer between multiple operations across the supramodal cortex is an important aspect of frontopolar function.42 Koechlin and Hyafil, in their comprehensive review, suggested that the frontopolar cortex is efficient for protecting the execution of long-term mental plans from immediate environmental demands and for generating new, possibly more rewarding, behavioral or cognitive sequences, which is related to human creativity rather than complex decision-making and reasoning.43 The functional properties of the frontopolar cortex are particularly important from the clinical point of view, in that they may give us some hint for improving the social functioning in psychiatric patients' everyday lives.


The authors thank all the participants in this study. The authors also thank Hitachi Medical for providing us with material support (temporary rental of a near-infrared spectroscopy machine, ETG-4000) and technical advice.