Breast Cancer Research and Treatment

, Volume 116, Issue 1, pp 113–123

Cognitive impairments associated with breast cancer treatments: results from a longitudinal study

Authors

  • Catherine Quesnel
    • School of PsychologyUniversité Laval and Laval University Cancer Research Center
    • School of PsychologyUniversité Laval and Laval University Cancer Research Center
  • Hans Ivers
    • School of PsychologyUniversité Laval and Laval University Cancer Research Center
Clinical Trial

DOI: 10.1007/s10549-008-0114-2

Cite this article as:
Quesnel, C., Savard, J. & Ivers, H. Breast Cancer Res Treat (2009) 116: 113. doi:10.1007/s10549-008-0114-2

Abstract

Purpose Published cross-sectional studies have revealed that chemotherapy for breast cancer is associated with significant cognitive impairments. However, because these studies included no baseline assessment, it is unknown whether the cognitive impairments were pre-existent to cancer treatment or truly secondary to chemotherapy. To resolve this issue, this prospective study aimed to compare the effects of chemotherapy to the effect of radiotherapy on cognitive functioning in women treated for non-metastatic breast cancer. Patients and Methods A total of 81 breast cancer patients, 41 receiving chemotherapy as part of their breast cancer treatment regimen and 40 receiving radiotherapy without chemotherapy were evaluated using an extensive battery of neuropsychological tests at baseline (ie, pre-chemotherapy or pre-radiotherapy), post-treatment (ie, post-chemotherapy or post-radiotherapy) and at a 3-month follow-up assessment. Results A mixed model covariance analysis revealed that receiving any kind of breast cancer treatment, with chemotherapy or not, was associated with impaired capacities for recruiting information in verbal memory. Moreover, the results showed that verbal fluency was impaired after breast cancer treatment, but only in patients who received chemotherapy. Conclusion Overall, this study reveals subtle cognitive impairments associated with breast cancer treatment. Specifically, it suggests that chemotherapy has a specific negative effect on verbal fluency, that breast cancer treatment in general negatively affects verbal memory, but that other cognitive functions are well preserved. Future studies should, however, attempt to better control the practice effect that may have masked other subtle alterations and use more ecologically valid measures of cognitive functioning.

Keywords

Breast cancerCancer treatmentsChemotherapyRadiotherapyCognitive functioningCognitive impairmentsLongitudinal study

Introduction

Published cross-sectional studies that have evaluated cognitive functioning in women treated for non-metastatic breast cancer have consistently found greater cognitive impairments in women treated with chemotherapy than those who received only local therapy or healthy women, at post-treatment and up to 10 years after treatment [17]. However, these studies are characterized by several methodological limitations inherent to their cross-sectional nature, such as the absence of a baseline measure of cognitive functioning and the comparison with published norms [7] or healthy controls [1] only.

More recently, prospective studies have been published on the impact of chemotherapy on cognitive functioning in breast cancer patients [814]. Overall, these longitudinal studies revealed far fewer significant differences between women who had received chemotherapy and those who had not received it than earlier cross-sectional studies. In fact, several of these studies failed to find a significant negative effect of chemotherapy when they compared the average neuropsychological performance between groups [10, 11, 14]. Moreover, less than a third of breast cancer patients treated with a high [11] or standard dose [9, 12, 13] of chemotherapy showed cognitive decline over time (more frequently in their verbal working memory [8, 9, 12]) and the majority of breast cancer patients even showed an improved performance over time [813]. Besides, several of these studies suggest that a subgroup of patients, corresponding to approximately a third of breast cancer patients, exhibits cognitive impairments even before they receive chemotherapy, thus suggesting that cognitive functioning may be affected by other factors than chemotherapy [9, 14]. Additional longitudinal studies are therefore needed on this issue, in particular studies including different types of comparison groups: healthy control groups to assess the presence of pre-existing cognitive impairments prior to cancer treatments and groups of patients receiving other forms of cancer treatments to better disentangle the effects that are specific to chemotherapy.

This longitudinal study aimed at comparing the evolution of cognitive functioning among patients receiving chemotherapy as part of their breast cancer treatment regimen to patients receiving radiotherapy without chemotherapy, at post-treatment and at a 3-month follow-up. Another goal of this study was to compare the cognitive functioning of breast cancer patients prior to the initiation of cancer treatments to that of healthy controls. It was hypothesized that: (a) women receiving chemotherapy as part of their breast cancer treatment regimen would show a greater decline in their cognitive performance than women receiving radiotherapy without chemotherapy at post-treatment and at 3-month follow-up; and (b) the two groups of cancer patients would show poorer results on neuropsychological tests than healthy women prior to the initiation of cancer treatment.

Patients and methods

Participants

French-Canadian women who recently received a breast cancer diagnosis were solicited from September 2002 to April 2004 by a research assistant the day of their post-operative follow-up appointment. The study goals and procedures were explained in detail to potential participants and those agreeing to participate were invited to provide written consent for their participation. A total of 81 breast cancer patients, divided into two groups based on the treatment protocol determined by their oncologist, participated in this study: 41 patients received chemotherapy (C group) which was followed by radiotherapy for 38 of them, and 40 patients received radiotherapy without chemotherapy (R group). A control group composed of 45 healthy women with no history of cancer and recruited through the print and electronic media took part in this study. Twenty-two of them were pair-matched to patients of the C group and 23 to patients of the R group according to age (i.e., ±5 years) and education (i.e., high school, college, university). The study was approved by the research ethics committees of the CHUQ (L’Hôtel-Dieu de Québec) and Laval University.

All participants, including healthy controls, were aged between 35 and 70 and had French as their mother tongue. Additional inclusion criteria for breast cancer participants were: (a) having received a first diagnosis of non-metastatic (i.e., stage I-III) breast cancer; and (b) to be scheduled to receive standard protocols of chemotherapy or radiotherapy. Women with a neurocognitive (e.g., dementia) or a psychotic (e.g., schizophrenia) disorder, as noted in the medical chart, were excluded from the study. To maximise the generalization of the findings and the study feasibility, the utilization of a medication (e.g., psychotropic medication) was not an exclusion criteria. Instead, statistical analyses were conducted to ensure that medication usage was not significantly associated with changes in cognitive functioning (see the covariates selection in the Statistical Analyses section).

A total of 176 breast cancer patients were approached to participate in this study. Of these, 28 were excluded because they did not meet the inclusion criteria (i.e., treatment plan did not include chemotherapy or radiotherapy: n = 13; prior history of cancer treatment: n = 13; presence of a cognitive disorder: n = 1; prior knowledge of neuropsychological tests: n = 1). Among the 148 eligible women, 67 refused to take part in the study. The most common reasons for non-participation included transportation difficulties (n = 23), lack of time or interest (n = 22), and study judged too burdensome (n = 22). A total of 81 patients participated in the first neuropsychological assessment (participation rate: 59.1%). At post-treatment, 5% (2/40) of the patients in the R group and 7.3% (3/41) of the patients in the C group dropped out of the study, a difference that was not statistically significant, χ2(1, N = 81) = 0.19, P = .66. There was a trend for a higher dropout rate in the chemotherapy condition at follow-up (8/41 or 19.5% vs. 4/40 or 10% for radiotherapy), but this difference failed to reach statistical significance, χ2(1, N = 81) = 1.45, P = .23. No significant difference was found between completers and dropouts on any of the demographic and clinical variables at post-treatment. At follow-up, participants who dropped out were less likely to be post-menopausal (41.7 vs. 79.7% for completers), χ2(1, N = 81) = 7.70, P = 0.006, and to have received HRT (33.3 vs. 68.1% for completers), χ2(1, N = 81) = 5.30, P = 0.02. Reasons for interrupting their participation in the study were: a lack of time or a loss of interest (n = 8), transportation difficulties (n = 3) and hospitalization or death (n = 2).

Measures

The neuropsychological tests were administered in the following order for all participants at each time assessment: Complex Figure Test (CFT) [1517], Rey Auditory Verbal Learning Test (RAVLT) [18], Trail Making Test (TMT) [19], Symbol Digit Modalities Test (SDMT) [20], Digit Span (DS) and Visual Memory Span (VMS) Subtests of the Wechsler Memory Scale-Revised (WMS-R) [21], Verbal Fluency Test (VFT) [22], and Ruff 2 & 7 [23]. The battery was carefully selected to assess a broad range of cognitive functions, including verbal (RAVLT) and visual memory (CFT), attention and concentration (DS, VMS, 2&7), executive functions (TMT), speed of information processing (SDMT) and verbal fluency (VFT). Alternate versions of CFT, RAVLT and VFT were used to minimize the practice effect of those tests. In addition, all participants completed a battery of self-report scales at each time assessment including: the Cognitive Failures Questionnaire (CFQ) [24] and the cognitive subscale of the European Organization for Research and Treatment of Cancer Quality of Life Questionnaire (QLQ-C) [25]. On only one occasion, the vocabulary and the picture completion (PC) subtests of the Wechsler Adult Intelligence Scale-III (WAIS-III) [26] were administered to provide an estimate of patients’ premorbid verbal and non-verbal intellectual quotient (IQ), respectively. Health-related data were taken from patients’ medical records.

Procedure

Breast cancer patients were evaluated on three occasions (see Fig. 1): prior to the initiation of their first adjuvant cancer treatment (baseline), immediately after completion of the first adjuvant treatment, and 3 months following the completion of the last adjuvant treatment (excluding adjuvant hormone therapy). The mean delay between the pre- and post-radiotherapy was significantly shorter in the C group than in the R group, t(75) = −2.82, P = 0.006, but no significant difference was found on the number or radiotherapy treatments received, t(78) = −0.39, P = .70. The control group was evaluated on only one occasion. Each assessment lasted between 60 and 90 min and was conducted in a quiet room of the hospital. The battery of neuropsychological tests was administered by a graduate student in psychology. The dependent variables derived from the neuropsychological tests are listed in Table 2. The CFT, VFT and WAIS-III vocabulary subtest were scored independently by two research assistants because of the subjective nature of their scoring. Whenever there was disparity between both scores the two research assistants discussed it until an agreement was obtained. After each neuropsychological assessment, participants were given a battery of self-reported questionnaires to be completed at home and sent back by mail to the laboratory. A monetary compensation of 25 CAN$ was given to participants for each time assessment completed.
https://static-content.springer.com/image/art%3A10.1007%2Fs10549-008-0114-2/MediaObjects/10549_2008_114_Fig1_HTML.gif
Fig. 1

Study design. Notes: C = Chemotherapy group; CC = Control group matched to the C group; R = Radiotherapy group; F-U = Follow-up; RC = Control group matched to the R group

Statistical analyses

Double-checked data were examined for missing data and outliers using standard procedures [27]. All analyses were conducted using the SAS 9.1.3 software [28]. The alpha level was set at 5% (two-sided). Only missing data on covariates (and not on dependent variables) were estimated using the multiple imputation method [29].

Effect of cancer treatment

To evaluate the effect of cancer treatment on cognitive functioning over time, a mixed model covariance analysis (ANCOVA) for a factorial design (Group: C vs. R, as the between-subjects factor, and Time: pre-treatment, post-treatment and 3-month follow-up, as the within-subject factor) was conducted on the 17 objective variables and the 2 subjective variables of cognitive functioning. Given that the main goal of the study was to compare the changes in neuropsychological performance over time across cancer treatment received, only Time and Group by Time effects are reported. It is noteworthy that the only significant difference observed between patients in the C Group and patients in the R Group at baseline was on the VMS (forward and backward), with a better performance of the R Group compared to the C Group.

The covariates included in the mixed model ANCOVA were selected following the recommendations of Frigon and Laurencelle [30]. The selection was based on the magnitude of the correlation obtained between each potential covariate1 and each dependent variable, instead of including them strictly on a theoretical basis or based on significant between-groups difference at baseline. Thus, correlational analyses were conducted among all potential confounding variables to identify those that were the most strongly associated with neuropsychological variables. A variable was selected as a covariate if it was significantly associated (i.e., P < 0.05, r ≥ ± 0.14 for our sample of 81 patients) with at least 4 of the 9 neuropsychological variables2. While the exact criterion for a significant correlation may be debatable, we chose a more inclusive approach (at the risk of including non-significant covariates) instead of being under-inclusive (and missing some important covariates). The 8 following covariates were included in the ANCOVA: (a) age; (b) education; (c) family income; (d) score obtained on the PC subtest (estimate of the premorbid non verbal IQ); (e) score obtained on the Vocabulary subtest (estimate of the premorbid verbal IQ); (f) number of chemotherapy cycles; (g) type of surgery (i.e., lumpectomy vs. radical mastectomy); and (h) type of node resection (i.e., conventional axillary dissection vs. sentinel node resection).

In addition, to assess the magnitude of the changes, effect sizes (Cohen’s d [31, 32]) were calculated from two simple time effects (i.e., pre-treatment to post-treatment change, and post-treatment to 3-month follow-up change) for each group. Another advantage of effect sizes is that all scores obtained on the various neuropsychological tests are shown on the same scale, thus making comparisons across all tests easier. Each effect size was calculated as the ratio of the averaged difference divided by the root mean square error of the model [31].

Comparison of breast cancer patients at baseline with healthy controls

To assess the potential presence of differences in cognitive functioning between breast cancer patients at baseline and healthy controls, a factorial ANCOVA with two between-subjects factors (i.e., Group: C or R, and Condition: breast cancer patients or healthy controls) was conducted on the 17 objective variables and the 2 subjective variables of cognitive functioning at baseline. To control for some potential confounding variables, 6 covariates were included in the ANCOVA: (a) age; (b) education; (c) marital status; (d) family income; (e) score obtained on PC subtest; and (f) score obtained on Vocabulary subtest. The selection of covariates was done using the same method described above (results not reported). Effect sizes were also calculated for the Condition simple effect to assess to what extent the patients differed from their healthy controls within each group at baseline.

Results

Demographic and medical data

Table 1 presents the sociodemographic characteristics for each breast cancer group and its respective control group as well as the medical characteristics of breast cancer patients. ANOVAs and chi-square tests revealed significant differences between the C and R groups on 6 variables. The C group was significantly younger and less frequently post-menopausal at diagnosis than the R group. In addition, patients of the R group were significantly more likely to have used hormone replacement therapy (HRT) prior to their cancer diagnosis. Breast cancer groups were also different with regard to the stage of cancer, the type of surgery (i.e., lumpectomy vs. mastectomy), and the type of resection (i.e., conventional vs. sentinel node). With regard to differences between breast cancer groups and their respective control group, a higher proportion of breast cancer patients (both groups) used HRT than their respective control group.
Table 1

Participants’ Characteristics

Variables

C (n = 41)

CC (n = 23)

R (n = 40)

RC (n = 22)a

Comparisons

df

F

P

Age (mean and SD)

50.3 (7.2)

47.9 (7.4)

57.7 (4.9)

55.0 (7.1)

C vs. R:

1,116

26.53

<0.01

C vs. CC:

1,116

1.76

0.19

R vs. RC:

1,116

2.21

0.14

 

n (%)

n (%)

n (%)

n (%)

  

χ2

 

Civil marital status

C vs. R:

1

1.75

0.19

    With partner

24 (58.5)

16 (69.6)

29 (72.5)

13 (59.1)

C vs. CC:

1

0.76

0.38

R vs. RC:

1

0.72

0.40

Education

    ≤High school

17 (41.5)

9 (39.1)

22 (55.0)

11 (50.0)

C vs. R:

2

2.47

0.48

    College

13 (31.7)

7 (30.4)

11 (27.5)

3 (13.6)

C vs. CC:

2

0.10

0.95

    University

11 (26.8)

7 (30.4)

7 (17.5)

8 (36.4)

R vs. RC:

2

4.96

0.18

Family income

    ≤30,000$

10 (24.4)

6 (26.1)

10 (25.0)

8 (36.3)

C vs. R:

1

0.10

0.75

    30,001–50,000$

15 (36.6)

8 (34.8)

16 (40.0)

4 (18.2)

C vs. CC:

1

0.00

0.95

    ≥50,001$

14 (34.1)

9 (39.1)

12 (30.0)

9 (40.9)

R vs. RC:

1

0.31

0.58

Post-menopaused at Dx

25 (61.0)

9 (39.1)

35 (87.5)

15 (68.2)

C vs. R:

1

7.72

0.02

C vs. CC:

1

2.88

0.24

R vs. RC:

1

3.58

0.17

Past or current HRT

20 (48.8)

4 (17.4)

31 (77.5)

8 (36.7)

C vs. R:

1

7.16

<0.01

C vs. CC:

1

6.19

0.01

R vs. RC:

1

10.29

<0.01

Cancer stage

C vs. R:

2

27.34

<0.01

    I

14 (34.1)

n/a

36 (90.0)

n/a

    

    II

20 (48.8)

n/a

4 (10.0)

n/a

    

    III

7 (17.1)

n/a

0 (0.0)

n/a

    

Lumpectomy

34 (82.9)

n/a

39 (97.5)

n/a

C vs. R:

1

4.83

0.03

Conventional axillary resection

30 (73.0)

n/a

15 (40.0)

n/a

C vs. R:

1

9.14

<0.01

Adjuvant hormone therapy

31 (75.6)

n/a

31 (77.5)

n/a

C vs. R:

1

0.04

0.84

Type of hormone therapy

C vs. R:

1

0.48

0.49

    Tamoxifen (Nolvadex®)b

27 (87.1)

n/a

25 (80.7)

n/a

    

    Anastrozole (Arimidex®)b

4 (12.9)

n/a

6 (19.3)

n/a

    

Number of radiotherapy Tx

C vs. R:

1

0.26

0.61

    0

3 (7.3)

n/a

0 (0.0)

n/a

    

    16–20

7 (17.1)

n/a

13 (32.5)

n/a

    

    24–25

18 (43.9)

n/a

25 (62.5)

n/a

    

    30

13 (31.7)

n/a

2 (5.0)

n/a

    

Number of chemotherapy Tx

    4

30 (73.2)

n/a

n/a

n/a

    

    6

7 (17.1)

n/a

n/a

n/a

    

    8

4 (9.8)

n/a

n/a

n/a

    

Type of chemotherapy

    AC

23 (56.1)

n/a

n/a

n/a

    

    TAC

12 (29.3)

n/a

n/a

n/a

    

    FEC

6 (14.6)

n/a

n/a

n/a

    

Notes: C = Chemotherapy group; CC = Control group matched to the C group; R = Radiotherapy group; RC = Control group matched to the R group; Dx = Diagnosis; HRT = Hormone replacement therapy; Tx = Treatments; n/a = not applicable; AC = adriamycine + cyclophosphamide; TAC = taxotere + adriamycine + cyclophosphamide; FEC = 5-fluorouracil + epirubicine + cyclophosphamide

aData for “Civil Marital Status” and “Family Income” are missing for one participant in RC

bZeneca Pharmaceuticals, 1800 Concord Pike, Wilmington, DE 19897; Tel: +1 302 886 3000; Fax: +1 302 886 2972; http://www.astrazeneca-us.com/

Effect of cancer treatment on cognitive functioning

On objective measures of cognitive functioning, the mixed model ANCOVA (see Table 2) revealed significant Time effects for the RAVLT on the total and free delayed recalls, indicating a significant decline in performance over time for both breast cancer groups. Conversely, the average performance of both groups was significantly improved over time on the CFT on copy, immediate and delayed recalls, on the TMT parts A and B and on the SDMT for the written and oral conditions. No significant time effect was found on the DS, VMS, VFT and Ruff 2 & 7. Only one significant Group X Time interaction was found and this was on the VFT. Mean VFT scores indicated a decreased performance on this test at the post-treatment assessment compared to the baseline for the C group only. These effects remained significant after conducting a Bonferroni correction for multiple comparisons. Overall, simple effects revealed that most significant Time and Group X Time effects (i.e., 8 out of 12) occurred between the baseline and the post-treatment, rather than between the post-treatment and the follow-up. Between the post-treatment and the 3-month follow-up assessments, significant changes occurred on only 4 neuropsychological variables (i.e., CFT immediate and delayed recalls, TMT A and B), and all were in the direction of an improvement following the post-treatment assessment. Figures 2 and 3 illustrate the effect sizes of changes obtained between the pre- and post-treatment (Fig. 2) and between the post-treatment and the 3-month follow-up (Fig. 3) in participants’ performance for each breast cancer group and for each of the 17 neuropsychological variables. Based on Cohen’s criteria [32], changes in performance that occurred between the pre- and post-treatment correspond to medium effect sizes (i.e., equal or inferior to 0.5), while those that occurred between the post-treatment and the 3-month follow-up correspond to small effect sizes (i.e., inferior to 0.2).
Table 2

Adjusted mean, standard errors and results at the mixed model ANCOVA on measures of cognitive functioning

Test

C

R

Factor

df

F

P

Pre

Post

FU

Pre

Post

FU

CFT

   Copy

M

33.99

34.77

35.33

31.51

32.45

33.02

Time

2, 142

6.93

<0.01

SE

0.80

0.81

0.79

0.78

0.79

0.79

Group × Time

2, 141

0.03

0.97

   Immediate recall

M

20.12

26.68

29.69

13.11

19.19

20.00

Time

2, 143

111.04

<0.01

SE

2.03

2.04

2.01

1.97

2.00

2.00

Group × Time

2, 142

2.94

0.06

   Delayed recall

M

19.12

25.24

27.92

13.50

18.72

19.57

Time

2, 145

84.27

<0.01

SE

2.00

2.01

1.98

1.95

1.97

1.97

Group × Time

2, 144

2.61

0.08

RAVLT

   Total recall

M

11.25

10.72

10.84

10.15

9.51

9.56

Time

2, 143

7.32

<0.01

SE

0.57

0.58

0.57

0.56

0.57

0.56

Group × Time

2, 143

0.15

0.86

   Delayed recall

M

11.97

11.18

11.24

10.10

9.23

9.53

Time

2, 145

4.55

0.01

SE

0.91

0.91

0.90

0.88

0.90

0.90

Group × Time

2, 144

0.08

0.92

   Recognition

M

14.99

15.10

15.10

13.94

14.11

13.74

Time

2, 145

1.35

0.26

SE

0.28

0.29

0.28

0.28

0.28

0.28

Group × Time

2, 144

1.36

0.26

TMTa

   A

M

26.72

27.01

25.29

30.68

29.54

27.30

Time

2, 146

3.39

0.04

SE

3.41

3.43

3.38

3.29

3.33

3.33

Group × Time

2, 145

0.56

0.57

   B

M

62.04

54.60

52.96

81.53

77.59

70.28

Time

2, 145

12.97

<0.01

SE

7.86

7.89

7.78

7.58

7.66

7.66

Group × Time

2, 144

1.02

0.36

SDMT

   Written

M

52.37

53.72

55.47

46.48

49.40

49.71

Time

2, 144

7.59

<0.01

SE

3.15

3.16

3.12

3.11

3.14

3.14

Group × Time

2, 143

0.57

0.57

   Oral

M

59.26

60.48

62.58

55.01

57.23

58.02

Time

2, 144

5.46

<0.01

SE

4.06

4.07

4.02

3.94

3.98

3.99

Group × Time

2, 143

0.27

0.77

DS

   Forward

M

7.64

7.78

8.13

7.36

7.75

7.48

Time

2, 141

1.73

0.18

SE

0.61

0.61

0.60

0.59

0.59

0.59

Group × Time

2, 141

1.47

0.23

   Backward

M

6.40

6.92

7.24

6.07

6.22

6.11

Time

2, 143

2.37

0.10

SE

0.80

0.81

0.80

0.78

0.78

0.78

Group × Time

2, 142

1.73

0.18

VMS

   Forward

M

7.05

7.10

7.10

7.71

7.49

7.36

Time

2, 141

0.28

0.76

SE

0.55

0.55

0.55

0.55

0.55

0.56

Group × Time

2, 140

0.51

0.60

   Backward

M

7.42

8.07

7.79

7.19

7.35

7.20

Time

2, 140

2.74

0.07

SE

0.52

0.52

0.52

0.51

0.51

0.52

Group × Time

2, 139

1.03

0.36

VFT

 

M

41.47

38.28

36.44

35.02

36.50

36.49

Time

2, 145

2.17

0.12

SE

3.25

3.27

3.22

3.14

3.17

3.17

Group × Time

2, 144

8.02

<0.01

Ruff 2 & 7

   Automatic

M

162.45

167.73

164.63

129.18

133.02

133.40

Time

2, 109

1.94

0.15

SE

11.03

11.00

10.85

12.54

12.55

12.62

Group × Time

2, 108

0.27

0.76

   Controlled

M

136.80

135.01

136.74

97.37

100.78

101.72

Time

2, 110

0.50

0.61

SE

8.41

8.38

8.25

9.56

9.55

9.62

Group × Time

2, 109

0.88

0.42

   CFQb

M

32.05

36.04

35.41

38.12

39.61

38.50

Time

2, 134

3.87

0.02

SE

4.26

4.29

4.23

4.10

4.14

4.15

Group × Time

2, 133

1.23

0.30

   QLQ-Cb

M

28.79

50.75

31.87

88.65

80.51

81.99

Time

2, 131

1.49

0.23

SE

15.07

15.28

15.04

14.85

14.98

15.10

Group × Time

2, 131

4.46

0.01

Notes: C = Chemotherapy group; R = Radiotherapy group; FU = Follow-up; CFT = Complex Figure Test; RAVLT = Rey Auditory Verbal Learning Test; TMT = Trail Making Test; SDMT = Symbol Digit Modalities Test; DS = Digit Span; VMS = Visual Memory Span; VFT = Verbal Fluency Test; CFQ = Cognitive Failures Questionnaire; QLQ-C = Cognitive subscale of the EORTC Quality of Life Questionnaire

aA higher score on the TMT A and B indicates a poorer performance at this test

bA higher score on the CFQ and QLQ-C indicates a poorer self-reported cognitive functioning

https://static-content.springer.com/image/art%3A10.1007%2Fs10549-008-0114-2/MediaObjects/10549_2008_114_Fig2_HTML.gif
Fig. 2

Effect sizes (d) obtained pre-and post-treatment for each neuropsychological variable according to group. Notes: C = Chemotherapy group; R = Radiotherapy group; CFT = Complex Figure Test; RAVLT = Rey Auditory-Verbal Learning Test; TMT = Trail Making Test; SDMT = Symbol Digit Modalities Test; DS = Digit Span; VMS = Visual Memory Span

https://static-content.springer.com/image/art%3A10.1007%2Fs10549-008-0114-2/MediaObjects/10549_2008_114_Fig3_HTML.gif
Fig. 3

Effect sizes (d) obtained between post-treatment and 3-month follow-up for each neuropsychological variable according to group. Notes: C = Chemotherapy group; R = Radiotherapy group; CFT = Complex Figure Test; RAVLT = Rey Auditory-Verbal Learning Test; TMT = Trail Making Test; SDMT = Symbol Digit Modalities Test; DS = Digit Span; VMS = Visual Memory Span

On subjective measures of cognitive functioning, a significant Time effect was obtained on CFQ scores. Simple effects indicated a decreased level of self-reported cognitive functioning at the post-treatment assessment compared to baseline, but that was only significant for the C group. A significant Group X Time interaction on QLQ-C scores was also found. Simple effects analyses revealed a significantly decreased level of self-reported cognitive functioning for the C group only at the post-treatment assessment compared to the baseline. Moreover, a significant simple Time effect was found showing an improvement of QLQ-C scores at the 3-month follow-up compared to the post-treatment assessment.

Comparison of breast cancer patients at baseline with healthy controls

The ANCOVA revealed significant simple effects (i.e., difference between breast cancer patients and healthy controls) on several neuropsychological variables (see Table 3). Specifically, in the C group, breast cancer patients obtained a significantly poorer performance on the DS forward, but better scores on the TMT part A (P = 0.08), VFT and CFQ, as compared to their healthy controls. In the R group, only one significant difference was found with healthy controls, which indicated a poorer performance in breast cancer patients on the oral condition of the SDMT, but trends in the same direction were observed on the written condition of the SDMT and on the 2&7, automatic and controlled processes.
Table 3

Adjusted means and effect sizes on the ANCOVA comparing breast cancer patients at baseline and healthy controls on variables of cognitive functioning

Test

C

R

C

CC

t (116)

dc

R

RC

t (116)

dc

CFT

  Copy

33.17

33.41

−0.26

−0.07

32.96

32.42

0.54

0.16

  Immediate recall

18.13

15.59

1.61

0.45

15.89

17.89

−1.16

−0.36

  Delayed recall

17.46

14.75

1.74

0.49

15.94

17.38

−0.85

−0.26

RAVLT

  Total recall

10.94

10.56

0.89

0.25

10.63

10.96

−0.72

−0.22

  Delayed recall

11.53

10.93

0.89

0.25

10.68

11.42

−0.99

−0.30

  Recognition

14.49

14.44

0.21

0.06

14.53

14.60

−0.27

−0.08

TMTa

  A

27.20

31.96

−1.78***

−0.50

28.70

26.51

0.75

0.23

  B

67.46

72.45

−0.73

−0.21

71.22

62.88

1.12

0.34

SDMT

  Written

50.84

52.72

−0.84

−0.24

49.96

54.39

−1.81***

−0.56

  Oral

58.46

58.54

−0.03

−0.01

57.66

65.99

−2.77*

−0.85

DS

  Forward

7.48

8.50

−1.98**

−0.55

7.80

8.53

−1.29

−0.39

  Backward

6.21

6.54

−0.50

−0.14

6.54

6.87

−0.47

−0.14

VMS

  Forward

7.01

7.37

−0.76

−0.21

7.94

8.04

−0.19

−0.06

  Backward

7.14

7.52

−0.93

−0.26

7.81

7.15

1.44

0.45

VFT

40.59

34.34

2.34**

0.65

38.04

35.12

1.00

0.31

Ruff 2 & 7

  Automatic

155.79

151.83

0.41

0.12

139.08

158.38

−1.75***

−0.59

  Controlled

128.39

118.42

1.43

0.42

112.05

126.99

−1.90***

−0.64

  CFQb

36.13

42.71

−2.20**

−0.64

34.51

35.06

−0.17

−0.05

  QLQ-Cb

58.75

79.43

−1.54

−0.45

59.99

44.44

1.06

0.33

Notes: C = Chemotherapy group; CC = Control group matched to the C group; R = Radiotherapy group; RC = Control group matched to the R group; CFT = Complex Figure Test; RAVLT = Rey Auditory Verbal Learning Test; TMT = Trail Making Test; SDMT = Symbol Digit Modalities Test; DS = Digit Span; VMS = Visual Memory Span; VFT = Verbal Fluency Test; CFQ = Cognitive Failures Questionnaire; QLQ-C = Cognitive subscale of the EORTC Quality of Life Questionnaire

aA higher score on the TMT A and B indicates a poorer performance at this test

bA higher score on the CFQ and QLQ-C indicates a poorer self-reported cognitive functioning

cA positive score indicates a better cognitive functioning in breast cancer patients as compared to their healthy controls

* P < .10, ** P < .05, *** P < .01

Table 3 also shows effect sizes obtained for each dependent variable, by group. A positive score indicates that the performance of breast cancer patients was better than that of their respective comparison subgroup of healthy participants. On objective variables of cognitive functioning, comparisons for the C group revealed effect sizes varying between −0.55 on the DS forward and 0.65 on the VFT with an average of 0.11, suggesting that, on average, patients in the C group performed slightly better than their controls. For the R condition, the effect sizes varied between −0.85 on the oral SDMT and 0.45 on the VMS backward, with an average of −0.24. Hence, on average, patients in the R group performed worse on neuropsychological tests than their controls. On subjective variables of cognitive functioning, effect sizes indicated that: (a) patients in the C group reported a better cognitive functioning than their controls, with effect sizes varying between −0.64 on the CFQ and −0.45 on the QLQ-C; and (b) patients of the R group reported a poorer cognitive functioning than their controls on the QLQ-C (0.33).

Discussion

The main goal of this study was to compare the evolution of cognitive functioning of women treated for non-metastatic breast cancer with chemotherapy to those receiving radiotherapy without chemotherapy. Prior to the initiation of cancer treatments, the average performance of breast cancer patients was inferior on two measures of attention (i.e., DS forward in the C group and SDMT oral in the R group) when compared to healthy participants. Following their first cancer treatment, patients of both groups showed decreased verbal memory (i.e., total recall and free delayed recall trials of the RAVLT) at the post-treatment and follow-up assessments, as compared to baseline. Moreover, chemotherapy was found to have a specific negative impact on verbal fluency (i.e., VFT), as indicated by a decreased number of words generated over time only in breast cancer patients treated with chemotherapy. These deleterious effects on verbal memory and fluency were maintained at the 3-month follow-up assessment. Besides, the results obtained on subjective measures of cognitive functioning revealed that patients treated with chemotherapy reported more cognitive difficulties than patients who did not receive chemotherapy at the post-treatment assessment, which was followed by a return to the baseline level at the 3-month follow-up evaluation.

Hence, similarly to other longitudinal studies [814], this study revealed limited impairments on objective measures of cognitive functioning following cancer treatments. Our findings which show significant declines in performance on learning and recall in verbal memory tasks in both groups of breast cancer patients are also consistent with other longitudinal studies [8, 10, 14]. Overall, they confirm that cognitive functioning may be affected by cancer treatments in general rather than by chemotherapy alone. More specifically, given that no significant decline was observed in both cancer groups on the recognition trial of the RAVLT, which assesses the storage component of memorization, our results suggest that it is only the access to memorized information (i.e., retrieval capacity) that is impaired following cancer treatment. Nevertheless, our study showed that chemotherapy had a specific negative effect on verbal fluency, a result which is consistent with another recent study [9]. It is noteworthy that only two other longitudinal studies conducted in breast cancer patients have assessed this cognitive function [13, 14]. Our results suggest that more attention should be devoted to this specific cognitive domain in future prospective studies.

Also, similarly to other longitudinal studies, the average performance of breast cancer patients appeared unchanged (i.e., DS, VMS, Ruff 2 & 7) or improved (i.e., CFT, TMT, SDMT) on many neuropsychological tests regardless of the type of treatment received (i.e., with or without chemotherapy). This could suggest that some cognitive functions are not affected by or even improve following breast cancer treatment. Alternatively, it could suggest the presence of an important practice effect which is typically associated with the repetition of neuropsychological testing [33, 34]. It is interesting to note that significant between-group differences were found on neuropsychological tests for which alternate versions were used at the post-treatment and follow-up assessments in this study. This reinforces the hypothesis that a practice effect might have blurred the effect of chemotherapy because cognitive functions were evaluated with the repeated administration of the same test form.

The fact that very few negative effects of cancer treatment were found on cognitive functioning may also be attributable to the weak ecological validity of neuropsychological tests [4, 8]. Indeed, it is possible that the effects of cancer treatments on cognitive functioning are more pronounced in day-to-day life than in the context of a highly structured neuropsychological testing. The weak correlations typically observed between objective and subjective measures of cognitive functioning support this hypothesis [35, 38]. Alternatively, it is also possible that patients were able to maintain the same level of performance in the study while having to provide an increased mental effort. Because the current study did not include any subjective measure of the mental effort associated to each neuropsychological assessment, it is impossible to verify this hypothesis. Finally, according to a recent meta-analysis [35], the effect of chemotherapy on cognitive performance would be similar to the effect of fatigue following a 12 h period without sleep. This effect could therefore be more subtle than what traditional neuropsychological tests are able to detect.

The results of this study and those obtained by others [9, 14, 36, 37] also suggest that cognitive functioning is influenced by factors others than just cancer treatments. Indeed, there is increasing evidence that, even prior to the initiation of cancer treatments, the average performance of patients recently diagnosed with breast cancer is inferior to that of healthy controls in the same cognitive domains. Given that the first neuropsychological evaluation is usually performed soon after the surgery, it is likely that the performance on neuropsychological testing is already affected by the cancer surgery itself at that time. For instance, an operation performed on the side of the dominant hand may affect patients’ motor speed [38] and subsequent improvements may indicate a rehabilitation effect. The possible association of these variables (e.g., side of surgery or axillary resection) with cognitive functioning was investigated during the preliminary analyses of this study and no significant effect was found. However, longitudinal studies with a larger sample size might better verify these hypotheses and investigate the potential contribution of additional influential factors.

Strengths of this study include its longitudinal research design and the utilization of a battery of tests assessing a broad range of cognitive functions, including verbal fluency. However, some methodological limitations should also be considered in the interpretation of the current findings. Although the sample size was greater than in some previous longitudinal studies, statistical power might still have been insufficient to detect some group differences. In addition, the fact that more than 76% of patients used adjuvant hormone therapy in both breast cancer groups did not allow for the comparison between users and non-users. Also, given the potential impact of age and menopausal status on cognitive performance, one could argue that participants of both cancer groups should have been matched on these variables. However, as breast cancer patients treated with chemotherapy are typically younger than those treated with radiotherapy or hormone therapy alone, we believe that our decision to opt for a statistical control instead, better reflects the clinical reality and allows a stronger external validity. Finally, we found some significant differences between completers and patients who dropped out of the study (i.e., menopausal status, HRT use), but only on analyses comparing the post-treatment and follow-up assessments.

In the future, studies should use larger and more homogeneous samples (e.g., same chemotherapy regimen) in order to improve statistical power and limit the potential effect of some confounding factors [3, 9]. In addition, it would be important to assess healthy participants concomitantly with breast cancer patients on all assessments to better control for the practice effect. This would also permit statistical analyses assessing cognitive changes at an individual level (e.g., Reliable Change Index), which was impossible in the current study.

In conclusion, the present findings suggest that breast cancer treatments have negative effects on a limited number of cognitive functions. This study also indicated that both radiotherapy and chemotherapy can impair cognitive functioning, but that verbal fluency is particularly sensitive to the effects of chemotherapy. These results appear clinically meaningful as they are consistent with common complaints of difficulties women have finding their words after they have just received chemotherapy.

Footnotes
1

Age; education; family income; marital status; adjuvant and replacement hormone therapy; body mass index; menopausal status; medication use; cancer stage; type of surgery; number of radiotherapy and chemotherapy treatments; type of chemotherapy, type of node resection; number of days between surgery or last treatment and assessments; estimate of verbal and non-verbal IQ; scores obtained on the CFQ, Hospital Anxiety and Depression Scale, Multidimensional Fatigue Inventory, Insomnia Severity Index, and vasomotor scale of the Menopause-Specific Quality of Life Questionnaire.

 
2

CFT immediate recall; RAVLT free delayed recall; TMT B; SDMT written condition; DSB; VMSF; VFT; 2 & 7 automatic processes; and 2 & 7 controlled processes. These 9 variables were selected in order to ensure the representativeness of the various cognitive functions assessed and by choosing only one variable for each construct (e.g., verbal memory) to avoid the over-representation of some constructs.

 

Acknowledgements

This study was supported partly by personnel research awards from the Canadian Institutes of Health Research (CIHR), the Fonds de la recherche en santé du Québec (FRSQ) and the Fonds pour la formation de chercheurs et l’aide à la recherche (FRSQ-FCAR-Santé) held by the first author, a research grant from the Quebec Breast Cancer Foundation and CIHR and FRSQ scientist awards held by the second author.

We sincerely thank Aude Caplette-Gingras, Lucie Casault, Catherine Gonthier, Marie-Hélène Savard, Benoît Senécal and Sébastien Simard for their important contribution to this project.

Copyright information

© Springer Science+Business Media, LLC. 2009