Working memory and the identification of facial expression in patients with left frontal glioma

Abstract

Patients with brain tumors may have cognitive dysfunctions including memory deterioration, such as working memory, that affect quality of life. This study was to explore the presence of defects in working memory and the identification of facial expressions in patients with left frontal glioma. This case-control study recruited 11 matched pairs of patients and healthy control subjects (mean age ± standard deviation, 37.00 ± 10.96 years vs 36.73 ± 11.20 years; 7 male and 4 female) from March through December 2011. The psychological tests contained tests that estimate verbal/visual-spatial working memory, executive function, and the identification of facial expressions. According to the paired samples analysis, there were no differences in the anxiety and depression scores or in the intelligence quotients between the 2 groups (P > .05). All indices of the Digit Span Test were significantly worse in patients than in control subjects (P < .05), but the Tapping Test scores did not differ between patient and control groups. Of all 7 Wisconsin Card Sorting Test (WCST) indexes, only the Preservative Response was significantly different between patients and control subjects (P < .05). Patients were significantly less accurate in detecting angry facial expressions than were control subjects (30.3% vs 57.6%; P < .05) but showed no deficits in the identification of other expressions. The backward indexes of the Digit Span Test were associated with emotion scores and tumor size and grade (P < .05). Patients with left frontal glioma had deficits in verbal working memory and the ability to identify anger. These may have resulted from damage to functional frontal cortex regions, in which roles in these 2 capabilities have not been confirmed. However, verbal working memory performance might be affected by emotional and tumor-related factors.

Recent epidemiological studies in China reported that the incidence of brain tumors was 7.08/100 000–30.33/10 million individuals,^1,2 which is higher than that in developed countries (3.23/100 000–11.3/10 million individuals).^3–6 Gliomas accounted for 40% of this incidence.⁷ Although the incidence of gliomas has increased over recent years, the survival rate among patients has improved and the role of the neurosurgeon has gradually changed from life preservation to improving the health-related quality of life (HRQoL) and activities of daily living (ADL) of survivors.⁸ Therefore, neurocognitive function, which is an important predictive factor of HRQoL and can be a treatment-related factor, has generated significant concern in glioma survivors.^9,10

Working memory (WM) is the basis of higher cognitive functions, such as planning and organizational behavior, language, thinking, and decision-making.^11,12 Damage to WM may affect learning ability and cognitive function. Many neurocognitive studies of patients with brain tumors have illustrated that memory deterioration, including WM, is one of the most common cognitive dysfunctions^13–16 and a key concern of patients.¹⁷ Any deterioration in WM may directly affect the quality of life of patients and their ability to return to society. Intervention studies also found that a rehabilitation plan aimed specifically at improving memory has long-term effects in improving cognitive performance.¹⁷

The ability to identify and process facial information is an advanced visual skill in human development.¹⁸ Compared with other visual skills, the selective focus on the face and increased time processing facial information is a major way to access community information and control the surrounding environment. On the basis of basic visual perceptual abilities, humans possess the ability to process facial information with use of cognitive neural networks.¹⁹ As the embodiment of advanced social cognitive functions of the brain, a deficit in the ability to recognize facial expression can directly impact the establishment of normal social relationships, thereby affecting social regression.²⁰

The frontal lobe is a major component of the cognitive network, and it has been confirmed to play an irreplaceable role in both WM and facial expression identification. Previous anatomical and physiological studies in animals have demonstrated close connections between the prefrontal cortex and the medial temporal lobe memory system.²¹ Functional imaging studies in humans indicated that verbal, visual-spatial, and object WM tasks activate different neural information processing circuits,²² whereas diffusion tensor imaging (DTI) studies revealed that bilateral frontal-parietal white matter fiber tracts are closely related to verbal and visual-spatial WM.^23–26 Recent studies have suggested that a functional connection between the prefrontal cortex and other regions combine to form the central execution system of the WM model.²⁷

The frontal lobe also occupies an important position in the expression processing network, which has been shown to be composed of many cortical and subcortical structures by previous neuroimaging studies,^28–30 although there are right cerebral hemisphere lateralization features, and the right posterior hemisphere in particular makes a significant contribution.³¹ Hosselmo et al. found that specific neurons in the temporal and frontal lobes in human and non-human primates can react selectively to facial emotional stimulation,³² and Neta et al. demonstrated that the dorsolateral prefrontal cortex may be a member of the core processing system.¹⁹ Imaging studies indicated that the bilateral prefrontal and orbitofrontal cortices, other regional activities related to expression processing,^33,34 and the frontal lobe may be involved in the identification of specific expressions, such as anger.^35,36

Different functional regional lesions have various impacts on recognition abilities from different processing levels.³¹ Few investigations have been conducted to explore deficits in WM and the identification of facial expressions in patients with frontal lobe or other regional brain gliomas. Thus, we aimed to clarify the performance of patients with a left-sided frontal glioma with regards to WM and facial expression identification, in preparation for future studies on other affected cerebral regions. The long-term aim is to provide a reference for targeted interventions in future rehabilitation strategies to improve the HRQoL of patients with brain tumors.

Methods

Participants

Patients primarily received a diagnosis and were recruited at the Department of Neurosurgery/Neuro-Oncology at the Sun Yat-sen University Cancer Center from March through December 2011. The inclusion criteria included patient age of 22–60 years, a primary tumor confined to the left frontal lobe, confirmed by a senior neurosurgeon based on clinical symptoms and imaging data (CT or MRI) before surgery, and postoperative pathological diagnosis of a glioma confirmed by a senior neuro-pathologist.

The exclusion criteria included patients with deficits related to intelligence, vision, hearing, language comprehension and expression, other neurological comorbidities (e.g. encephalitis or stroke), any history of nervous system or psychiatric diseases, brain trauma, or diseases of other systems, such as coronary artery disease, hypertension, and diabetes. The presence of exclusion criteria was determined through initial interviews, MRI and surgical reports, neurological examination, and a neuropsychological assessment.

There were 11 patients recruited, including 7 males and 4 females. The mean age of the patients was 37 years (range, 22–57), and their mean educational attainment was 10.81 years. All patients were right-handed. The study was approved by the regional ethics committees, and informed consent was obtained from each individual before he or she participated in the study.

Two patients presented because of syncope, dizziness, or headache, and the other 9 patients had attended their doctors as a result of seizures. All patients except one had used antiepileptic drugs (AEDs) before testing. One patient required mannitol treatment (125 mL) on one occasion before the test. The tumor locations and sizes determined by neuroradiological MRI examinations are listed in Table 1.

Table 1.

Tumor-related characteristics of patients

Patients	Tumor Locations	Tumor Size (mm)	Histological Diagnosis
No.1	Precentral gyrus	17*17*19	Astrocytoma (WHO II)
No.2	Middle and superior frontal gyrus, cingulate cortex, the body of the corpus callosum	33*37*61	GBM (WHO IV)
No.3	Middle frontal gyrus, inferior frontal gyrus	38*38*49	AA (WHO III)
No.4	Middle frontal gyrus, inferior frontal gyrus, the corpus callosum, cingulate cortex	23*26*38	Astrocytoma (WHO I)
No.5	Middle frontal gyrus, inferior frontal gyrus	28*48*35	Astrocytoma (WHO II)
No.6	Inferior frontal gyrus	26*30*33	Oligodendroglioma (WHO II)
No.7	Middle frontal gyrus	24*24*21	Oligodendroglioma (WHO II)
No.8	Inferior frontal gyrus	28*31*38	Astrocytoma (WHO I)
No.9	Superior frontal gyrus, middle frontal gyrus, cingulate cortex	38*41*73	AA (WHO III)
No.10	Inferior frontal gyrus, cingulate cortex	34*59*40	AA (WHO III)
No.11	Precentral gyrus, inferior frontal gyrus, cingulate cortex	35*41*48	GBM (WHO IV)

Abbreviations: AA, anaplastic astrocytoma; GBM, glioblastoma; WHO, World Health Organization.

A control group was recruited of 11 healthy volunteers, who were individually matched to the patients by age (±5 years; mean, 36.73 ± 11.20 years; range, 20–55 years), sex, and educational level (10.72 years).

Neuropsychological Testing

The participants were individually administered neuropsychological tests 3–5 days prior to surgery without receiving any irradiation or chemotherapy. All data were collected at the same time point. The tests used had acceptable reliability and validity, and each test was administered using standard procedures and instructions. Assessments were conducted in a relatively quiet environment, with bright lighting and comfortable chairs and desks. The entire process took approximately 90 min.

Before the neuropsychological tests, we used the Self-Rating Depression Scale (SDS) and Self-Rating Anxiety Scale (SAS) to evaluate the emotional state of subjects, to exclude unstable factors that could influence the results of other assessments.³⁷ The standard SAS scores <50 and SDS scores <53 indicated a stable emotional state. Furthermore, the Abbreviated Wechsler Adult Intelligence-Revised in China (WAIS-RC; Information, Similarities, Block Design and Picture Arrangement) was used to exclude patients with severe intellectual defects and to ensure that the level of intelligence was matched between patients and control subjects. The algorithmic method of the verbal intelligence quotient (VIQ) and performance intelligence quotient (PIQ) were used according to the regression formula prepared by Gong and Dai.³⁸

Working Memory

The Digit Span Test Standardized version of the WAIS-RC: Digits Forward and Digits Backward were entered as separate variables. Digits Forward is considered as a test of simple sustained verbal attention span. Digits Backward is used as a measure of verbal WM that requires a greater attention capacity than the Digits Forward test.

In the Tapping Test, 2 test cards were used. Eight small green or red squares in a certain order were printed on each card. The participant was required to hit squares on red test card in accordance with the researcher, while hitting squares on the green test card in the reverse order. Fourteen attempts on the red card and 12 attempts on the green card were permitted. The researcher added 1 more square in each trial after every 2 successful attempts and stopped the test after 2 failures. The forwards and backwards test score were applied in this test.

Executive Function

The Modified Card Sorting Test was first described by Nelson in 1976.³⁹ Category control (CC), the number of answers to complete the first category (NACC), the percentage of preservative response (RP%), preservative response (PR), failure to maintain a set (FM), preservative errors (PE), and total errors (TE) were used in these analyses.

Facial Expression Identification

All 21 photos were selected from the Chinese static facial expression photo gallery and selected to contain 6 types of basic emotions and neutral expressions.⁴⁰ Three different photos of each emotion were used, and all were arranged randomly using 2011 as seeds. The photos were then displayed to the subjects at a horizontal distance from the eye of 30 cm. Participants were required to select and name the emotion expressed in each photo. Testing for reliability and validity of the facial expression recognition rate was >70%, and the test-retest reliability and validity were within acceptable ranges. The accuracy and error identification rate of each type of emotion, and the correctly identified number of a total of 21 photos of expressions were used in these analyses.

Statistical Analysis

All statistical analyses were performed using the Statistical Package for the Social Sciences (SPSS, version 16.0; SPSS) with a row score of tests. An alpha level of .05 was used to determine statistical significance. With the exception of facial expression identification being analyzed using the χ² test, group scores were compared using paired sample t tests (or Wilcoxon rank sum test) to analyze the performance differences of other tests. Then, the association analyses were performed using the Pearson product-moment correlation or Spearman rank correlation to compare emotional estimation scores, tumor-related characteristics, and the neuropsychological test indices.

Results

All the SDS scores of patients and control subjects were <53, but 3 patients and 1 control subject had an SAS score >50. All participants scored >70 in the VIQ/PIQ tests without intellectual impairment. As shown in Table 2, there was no statistically significant difference in these 4 indices between patients and control subjects in the paired sample t test analysis (all P > .05).

Table 2.

Emotional estimation and intelligence scores (x̄ ± s)

	SDS Score	SAS Score	VIQ	PIQ
Patients	39.36 ± 7.187	36.55 ± 11.19	102.5 ± 13.70	99.96 ± 13.94
Controls	41.27 ± 5.312	38.64 ± 6.975	102.8 ± 13.86	104.8 ± 15.42
t	−0.881	−0.498	−0.236	−1.290
P	0.399	0.630	0. 819	0.229

Abbreviations: SDS, Self-Rating Depression Scale; SAS, Self-Rating Anxiety Scale; VIQ, verbal intelligence quotient; PIQ, performance intelligence quotient.

*P < .05; **P < .01.

Working Memory and Executive Function

The indexes of the Digit Span Test (forward score, forward trials, backward score, backward trials, and forward + backward score) of the patients were all significantly lower than that of the control subjects (P < .05) (Table 3).

Table 3.

The digit span test tapping test scores (x̄ ± s)

	Digit Span Test					Tapping Test
	FS	FT	BS	BT	Total score	FS	BS	Total score
Patients	6.36 ± 1.567	7.00 ± 2.098	3.91 ± 1.375	5.64 ± 1.963	10.18 ± 2.316	7.73 ± 1.348	6.91 ± 1.640	14.64 ± 2.501
Controls	8.18 ± 1.079	9.00 ± 1.549	5.73 ± 1.954	7.27 ± 2.370	13.91 ± 2.587	9.00 ± 1.265	7.82 ± 1.250	16.82 ± 2.040
t	−2.877	−2.313	−3.299	−2.839	−5.632	−2.055	−1.392	−1.757
P	0.016*	0.043*	0.003**	0.018*	0.000**	0.067	0.194	0.079

Abbreviations: FS, forward score; FT, forward trials; BS, backward score; BT, backward trials.

*P < .05; **P < .01.

There was no statistically significant difference in terms of the forward score, backward score, and forward + backward score in the Tapping Test between the patients and control subjects (all P > .05), although the patient scores tended to be lower than those of the control subjects (Table 3).

Among the indexes that reflected forming concepts, the performance on the CC and RP% of patients was worse than that for control subjects, and the NACC of patients was higher than that for control subjects, but none of these differences were statistically significant (P > .05). Although the performance of patients on the 4 indices (PR, FM, PE, TE) that reflected preservation were worse than those for control subjects, only the PR score was significantly different between patients and control subjects (P < .05) (Table 4).

Table 4.

The WCST scores (x̄ ± s)

	CC	PR	FM	PE	NACC	RP%	TE
Patients	3.27 ± 1.737	6.18 ± 4.119	1.55 ± 1.036	1.82 ± 1.537	11.91 ± 10.58	40.91 ± 21.72	18.73 ± 6.405
Controls	4.27 ± 0.905	2.73 ± 3.101	1.27 ± 1.489	0.73 ± 0.905	9.18 ± 3.188	53.41 ± 11.31	14.82 ± 3.763
t	−1.658	2.270	−0.503a	1.936	−0.561a	−1.658	2.112
P	0.128	0.047*	0.615	0.082	0.574	0.128	0.061

Abbreviations: CC, category control; PR, preservative response; FM, failure to maintain a set; PE, preservative errors; NACC, the number of answers to complete the first category; RP%, the percentage of preservative response; TE, total errors.

*P < .05.

^a The differences between two groups did not conform to the normal distribution, and therefore we employed the paired sample Wilcoxon rank sum test. The Z-score is given.

Facial Expression Identification

The correct identified number of total 21 expression photos was better in the control group (14.36 ± 3.107) than in the patients (13.00 ± 2.828), but there was no significant difference (t = −1.426, P = .184). The correct recognition rate of angry faces was significantly lower among patients than among control subjects (P < .05). Among patients, the rate of identifying neutral, sad, or afraid expressions was higher than that among control subjects, but the rate of identification of happiness, surprise, and disgust was lower than that among control subjects. None of these differences were significant. Patient performance in the identification of different expressions is listed in Table 5 according to a high-low ranking sequence, and the ranking sequence of control subjects was approximately consistent with that of patients. As seen in Table 6, patients mistakenly recognized fear as surprise (42.4%) or disgust (12.1%), anger as disgust (33.3%) or a neutral expression (18.2%), and disgust as sadness (24.2%) or anger (18.2%). Control subjects misidentified fear, which was the most commonly misidentified expression, as surprise (60.6%).

Table 5.

The accuracy of the identification of six basic expressions and a neutral expression (%)

	Happy	Neutral	Sad	Surprise	Disgust	Anger	Fear
Patients	87.9	87.9	84.8	69.7	42.4	30.3	30.3
Controls	90.9	81.8	78.8	81.8	63.6	57.6	24.2
χ2	0.000a	0.471	0.407	1.320	2.981	4.982	0.306
P	1.000	0.429	0.523	0.251	0.084	0.026*	0.580

*P < .05.

^aAs two grid frequencies were scored as <5, we used the continuity correction.

Table 6.

The error identify rate for six basic expressions and a neutral expression (%)

	Patients							Controls
	Happy	Neutral	Sad	Surprise	Disgust	Angry	Fear	Happy	Neutral	Sad	Surprise	Disgust	Angry	Fear
Happy	–	3.0	0.0	6.1	0.0	0.0	0.0	–	15.2	6.1	3.0	0.0	0.0	3.0
Neutral	3.0	–	0.0	21.2	3.0	18.2	6.1	6.1	–	3.0	6.1	3.0	15.2	0.0
Sad	0.0	3.0	–	0.0	24.2	3.0	3.0	0.0	0.0	–	0.0	12.1	6.1	0.0
Surprise	6.1	3.0	3.0	–	9.1	9.1	42.4	3.0	0.0	0.0	–	9.1	9.1	63.6
Disgust	3.0	0.0	3.0	0.0	–	33.3	12.1	0.0	3.0	12.1	6.1	–	9.1	3.0
Angry	0.0	3.0	3.0	0.0	18.2	–	6.1	0.0	0.0	0.0	0.0	9.1	–	6.1
Fear	0.0	0.0	6.1	3.0	3.0	6.1	–	0.0	0.0	0.0	3.0	3.0	3.0	–

Association Analyses

As shown in Table 7, there was a positive correlation between the backward trials of the Digit Span Test and the SDS and SAS scores (P < .05). There was also a negative correlation between the total score and the SAS score (P < .05). Tumor size and grade were negatively correlated with the backward indexes and total digit span test score (P < .05). We found no correlation among the emotional estimation scores, tumor size/grade, and other psychological tests indices (all P > .05).

Table 7.

The association analyses of emotional estimation, tumor-related characteristics and the Digit Span Test indices of patients

		Digit Span Test
		BS	BT	Total
SDS score	r	–	0.678	–
	P	–	0.022*	–
SAS score	rs	–	0.630	−0.618
	P	–	0.038*	0.043*
Tumor sizea	r	−0.738	−0.823	−0.714
	P	0.010**	0.002**	0.014*
Tumor gradingb	rs	−0.804	−0.851	−0.843
	P	0.003**	0.001**	0.001**

Abbreviations: BS, backward score; BT, backward trials.

*P < .05; **P < .01.

^aThe longest diameter was used as the tumor size in the Pearson product-moment correlation analysis.

^bThe tumor grades were divided into two groups, as low-grade (WHO I and II) and high-grade (WHO III and IV) tumors, according to the WHO grading system described in Table 1.

Discussion

In this study, we determined that the verbal WM of patients with left-sided frontal gliomas was worse than that of control subjects, and this might be affected by the emotional state of the patient and tumor-related characteristics. No difference was found in visual-spatial WM between patient and control groups. The correct recognition rate of an angry expression was worse in patients with gliomas, but there was no significant difference in the recognition of 5 other facial expressions or neutral expressions between the 2 groups. The sorting set of the recognition capacity was relatively consistent between patients and control subjects.

Working Memory and Executive Function

WM processes information temporarily on the basis of simple storage in short-term memory.¹¹ The forward and backward indexes of patients with left frontal glioma were all worse than those in normal individuals, and we presumed that this could be attributable to damage to the phonological loop, which is the significant component of the WM model that was demonstrated by Baddley.¹¹ The phonological loop comprises the phonological store and articulatory rehearsal process; it can hold verbal information for a few seconds, and then it preserves the memory traces throughout the articulatory rehearsal process. Previous lesion and neuroimaging studies have shown that the rehearsal component is affected by damage to the left inferior frontal gyrus.^41,42 As was shown in our study, 7 (63.64%) patients had a lesion in the left inferior frontal gyrus. Moreover, patients with bi-frontopolar damage had deficits in the nonarticulatory maintenance of verbal information.⁴³ Therefore, our results confirmed previous findings that the frontal lobe is involved in verbal information encoding and maintenance.^44–46 Nevertheless, the tumors that invaded other areas, such as the middle frontal gyrus and cingulate cortex, were not restricted to the inferior frontal gyrus. More studies are required to determine which functional areas of the left frontal lobe control different aspects of the processing procedure.

However, it may be dysfunction of the central executive, indicated in this study though the performance of preservation response of patients in the WCST that was consistent with the findings of Zheng⁴⁷ and Lin,⁴⁸ that results in the poor performance of verbal WM. The high PR score in patients prompts an increased preservation response and a decrease in cognitive flexibility. Although Baddeley proposed that the executive function defect caused by brain damage might lead to impairments in a series of WM tasks, contrasting results from lesion studies indicated that executive function defects might not exist after injuries to the prefrontal cortex (PFC),²⁷ and similar working memory task performances were normal.⁴² A meta-analysis conducted by D'Esposito and Postle concerning 11 studies explored memory defects after frontal lobe damage and showed no decrease in verbal and spatial memory span.⁴⁹ Therefore, the key WM regions do not rely on PFC separately, and the damage does not necessarily contribute to the decrease in WM. Spatial WM in patients without defects in this study also confirmed this.

The correlation analyses showed that the performance of the Digits Backward test and the total score could be affected by the emotional status of patients and the tumor size and grade. We suggest that verbal WM deficits may result from the combined effects of the tumor, tumor-related characteristics, and patient-related factors, such as emotional state. This view has been proposed in previous neurocognitive studies of patients with brain tumors.^13,50 However, various factors generate different effect sizes. Future studies are needed to differentiate between the effects of these factors to explore the function of brain precisely.

There were more numbers of abnormalities in verbal WM than visual-spatial WM in patients in our study, consistent with the study by Tucha, which showed that left-sided tumors, including temporal lobe tumors, were associated with verbal stimuli processing disorders and verbal memory dysfunction more than right-sided tumors.¹³ Earlier studies found that the right hemisphere of the brain predominates in non-verbal stimuli processing, and visual-spatial information and visual-construction processing is associated with the right cerebral hemisphere (especially the right parietal lobe).^45,51 Although previous imaging studies have found that bilateral frontal-partial pathways are associated with visual-spatial WM tasks, visual-spatial information processing may be more dependent on right hemisphere networks.⁵² Therefore, injuries to the left frontal lobe do not damage visual-spatial WM.

Facial Expression Identification

Expression recognition is an important means of the acquisition of nonverbal information in social interactions, and its importance depends on adjusting social behaviors through the appropriate interpretation of facial information. The development of the emotion apprehension ability is relevant to early experiences,⁵³ and this capacity matures gradually with the development of brain structure.^54,55 In the early neonatal period, gaining a satisfying experience under physiological states, such as obtaining food when hungry, prompts the development of happy cognition, followed by anger and sadness.^19,54 Afraid, surprised, and disgusted facial expressions are more difficult to identify, and positive emotion facial expressions are significantly easier to identify than negative ones.^19,54 The high to low order of identification abilities of the patient group in this study was almost the same as matched healthy adults, showing the developmental trend as mentioned. We found that the correct recognition of angry expressions was significantly different in patients than in the control group, with the lowest rate of accuracy, which might have been attributable to the relatively low validity of the 3 angry face photos⁴⁰ that made it more difficult to discriminate it from other expressions. However, the statistical difference existed still after patient results were compared with those of the control group, which eliminated the effects of this factor. The causation may have been the injury of frontal lobe, which was consistent with the findings that the frontal lobe is involved in the specific processing of angry expression as shown by previous fMRI studies.⁵⁶ An alternative opinion was suggested by Teasllale et al., who proposed that the frontal lobe was not specifically involved in facial expression processing, but may play a role in the emotional experience, especially the integration of information regarding negative emotions.⁵⁷

Facial expression processing is mediated in hierarchically organized patterns, which comprise the recognition of structural facial features, primary emotion classification, and secondary emotion dimensions (valence or arousal) decoding.³¹ However, we primarily conducted this study to evaluate primary emotion classification abilities without reference to deep processing. The right hemisphere of the brain is the most important network of facial expression information processing,⁵⁸ which may explain the overall phenomenon of facial expression recognition with non-significant differences between the patient and control groups. Blonder et al. found that patients with right hemisphere injuries produced fewer facial expressions during social contact,⁵⁹ and many functional imaging studies also found that the activity levels of cortical structures surrounding the right lateral fissure, including the superior temporal sulcus (STS), superior and middle temporal gyrus, and supramarginal gyrus, mostly increased during the processing of different emotions.^20,60 Consequently, even though some studies found that different regional structures in the frontal lobe activated in different facial expression recognition task, this could not lead to the deficiency in the overall facial expression recognition ability of patients.

Conclusion

This study demonstrated that patients with left-sided frontal gliomas showed deficits in verbal WM and the ability to identify angry expressions. This may be attributable to injuries in the functional regions of frontal cortex, although the precise roles of this region in these 2 capabilities are not confirmed. However, the performance of verbal working memory might be affected by both emotional and tumor-related factors. Future studies are needed to distinguish between the effects of the tumor itself and tumor-related factors, including the histological type, on cognitive function.

Funding

This work was supported by the National Natural Science Foundation of China (No. 30600207, 81171293 and 30973080).