A Prospective Evaluation of Changes in Brain Structure and Cognitive Functions in Adult Stem Cell Transplant Recipients

Abstract

Hematopoietic stem cell transplantation (HSCT) is an efficacious treatment for many hematologic malignancies. However, the conditioning regimen of high-dose (HD) chemotherapy with or without total body irradiation (TBI) can be associated with neurotoxicity. In this prospective study, we used quantitative neuroimaging techniques to examine regional gray matter and ventricular volumes, and standardized neuropsychological tests to assess cognitive function before and one year after HSCT in twenty-eight patients with hematologic malignancies and in ten healthy controls evaluated at similar intervals. Nineteen patients received conditioning treatment with HD chemotherapy alone and nine had both TBI and HD chemotherapy. There was a significant reduction in gray matter volume in the middle frontal gyrus bilaterally and in the left caudate nucleus in the patient group (all patients combined) but not among healthy controls over the one-year follow-up period. There was a significant increase in left lateral ventricle volume and in total ventricle volume in the patient group, relative to healthy controls. Similar brain structural changes were seen for patients treated with HD chemotherapy alone. The neuropsychological results showed that 21% of patients could be classified as impaired at baseline. The Reliable Change Index suggested no significantly different rates of cognitive decline between patients and healthy controls. The findings suggest that HSCT patients may be at an increased risk for developing regional brain volume loss, and that subgroups may experience cognitive dysfunction prior to and one year following the transplant.

Introduction

Hematopoietic stem cell transplantation (HSCT) is used in the treatment of many hematologic disorders. Although it can be curative, the pre-transplant myeloablative conditioning regimen including high-dose chemotherapy with or without total body irradiation (TBI) can be associated with neurotoxicity (Snider et al., 1994; Soutar & King, 1995; Sostak et al., 2003). In addition, most patients undergo chemotherapy before receiving the conditioning treatment for HSCT, as part of prior attempts to treat their illnesses (Hagemeister, 2002). In patients who undergo allogeneic HSCT, treatment-related complications such as acute or chronic graft versus host disease (GvHD), infections related to immunosuppression, and the side effects of immunosuppressive therapy, may also have neurotoxic effects (Bartynski et al., 2001; Nucci et al., 2003). These multiple factors and interventions place patients at an increased risk for central nervous system (CNS) damage and may be associated with acute or chronic cognitive dysfunction.

There is compelling evidence that chemotherapy is associated with neurotoxicity (Rzeski et al., 2004; Meyers, 2008), particularly high-dose methotrexate and cytarabine (DeAngelis and Posner 2009; Dietrich et al., 2008). Candidate mechanisms for neurotoxicity include demyelination, chemotherapy-induced oxidative stress and DNA damage, and immune dysregulation and stimulation of neurotoxic cytokines (Ahles et al., 2007). Recent animal studies suggest that higher levels of chemotherapy may reach the brain than previously assumed and that even low doses of chemotherapy can increase cell death and decrease cell division in the hippocampus (Dietrich et al, 2006). The most significant complications of radiotherapy are delayed and often produce irreversible and progressive damage to the CNS through vascular injury causing chronic ischemia, cerebral atrophy, progressive demyelination of the white matter, and necrosis (Omuro et al. 2005; DeAngelis and Posner 2009). The risk for developing radiation-related damage increases with irradiated volume, total dose and fraction size, and age. Combined treatment with radiation and chemotherapy may have a synergistic toxic effect (Keime-Guibert, 1998).

Studies addressing neuropsychological functioning in patients undergoing HSCT reported that 20–40% had cognitive dysfunction involving graphomotor speed, executive and memory abilities prior to transplant (Andrykowski et al., 1992; Chang et al., 2009; Friedman et al., 2009; Harder et al., 2005, 2006; Jacobs et al., 2007; Meyers et al., 1994; Schulz-Kindermann et al., 2007). Several prospective studies documented cognitive decline in the initial months after HSCT (Ahles et al., 1996; Booth-Jones et al., 2005; Friedman et al., 2009; Syrjala et al., 2004; Schulz-Kindermann et al., 2007), but the absence of a decline was also described (Harder et al., 2007). At approximately one year post-transplant, cognitive functions were reported to return to or surpass pre-transplant levels in several studies (Chang et al., 2009; Harder et at., 2006; Jacobs et al., 2007; Syrjala et al., 2004; Wenz et al., 2000), but subgroups of patients had persistent impairment or decline on some cognitive domains (Jacobs et al., 2007; Meyers et al., 1994; Syrjala et al., 2004). Syrjala and colleagues (2011) reported further improvement in information processing speed and executive functions, but not in memory or motor dexterity in a large cohort of patients evaluated five years following allogeneic HSCT. However, more than 40% of survivors had a global deficit score indicative of mild cognitive impairment. In the few cross-sectional studies that included structural neuroimaging, white matter abnormalities and atrophy were documented in some HSCT patients treated with TBI and chemotherapy (Garrick, 2000; Padovan et al., 1998; Peper et al., 2000; Stemmer et al., 1994). In a longitudinal study (Sostak et al., 2003), a subgroup of patients treated with TBI and intrathecal chemotherapy prior to allogeneic bone marrow transplantation developed neurologic complications, white matter changes, cerebral atrophy, and cognitive deterioration one year after the transplant; chronic GvHD and immunosuppression were significant risk factors.

To date, there have been no prospective studies combining advanced neuroimaging procedures and cognitive assessment in patients undergoing HSCT conditioning regimens including chemotherapy with or without radiotherapy. In this study, we assessed changes in gray matter and ventricular volume, as well as neuropsychological functions in patients prior to HSCT and one year after the transplant. The results were compared to those obtained from a group of healthy controls evaluated at equivalent intervals.

Methods

Subjects

The majority of patients were identified and recruited by the treating physicians and nurses in the Department of Medicine at Memorial Sloan-Kettering Cancer Center (MSKCC); four patients were recruited by the staff in the Department of Medicine at Weill Cornell Medical College (WCMC). Patients were eligible to participate in the MSKCC research protocol study if they had a diagnosis of a hematological malignancy and were scheduled to undergo conditioning treatment with high-dose (HD) chemotherapy with or without full-dose total body irradiation (fTBI), prior to receiving an allogeneic or autologous HSCT; were between the ages of 18 and 70 years, and fluent in English. Patients were excluded if they had evidence of disease progression at enrollment or during the study period, central nervous system disease, or history of neurological or psychiatric disorders. Healthy controls who met the same inclusion (except for cancer diagnosis) and exclusion criteria were recruited through community advertisements. Healthy controls were frequency-matched to patients on age, education, and gender. The research protocol was approved by the Institutional Review Boards at MSKCC and WCMC, and informed consent was obtained from all participants.

Thirty-nine of the originally planned forty study patients completed a baseline cognitive evaluation prior to undergoing conditioning treatment with high-dose (HD) chemotherapy with or without full-dose total body irradiation (fTBI) for HSCT, and twenty-eight were available for the one-year post transplant follow-up evaluation. Among the 11 patients who did not return for follow-up, 5 had disease progression at the time of the post-HSCT evaluation and were no longer eligible to continue in the study, 4 were deceased, and 2 declined to return. Sixteen of the originally planned twenty healthy controls completed a cognitive evaluation at study entry, and 10 were available for the one-year follow-up evaluation. The 6 healthy controls who did not complete the follow-up procedures either declined to return or could not be reached. Brain magnetic resonance imaging (MRI) including Magnetization Prepared Rapid Gradient Echo (MPRAGE) sequences was available at both baseline (prior to the conditioning treatment for HSCT) and one year post transplant for 21 patients and 10 healthy controls.

Structural Neuroimaging

MRI Scan Acquisition

All participants were imaged on the same 3T scanner (GE Medical Systems, Waukesha, WI) with an 8-channel phased-array head coil. T1-weighted three dimensional magnetization prepared rapid gradient echo (MPRAGE) images were acquired with the following parameters: TR=7 ms, TE=1.5 ms, FOV=23 cm, FA=7 degrees, number of slices= 156, 1.2 mm thickness, 192 × 192 matrix, in plane resolution of 1.2 mm². T2-weighted and fluid attenuated inversion recovery (FLAIR) sequences were also acquired to rule out pathology.

Image Processing

Voxel-based morphometry (VBM) analysis was performed using the longitudinal processing stream in the VBM8 toolbox (http://dbm.neuro.uni-jena.de/vbm/) under the SPM8 software package (Version 8, Wellcome Department of Imaging Neuroscience, London, UK) within MATLAB (Version 7, Mathworks, Inc., Natick, MA). Following reconstruction, follow-up MPRAGE structural images were registered to baseline MPRAGE images for each subject, bias corrected, segmented into gray matter, white matter and cerebrospinal fluid compartments using the Montreal Neurologic Institute (MNI) T1 weighted template and tissue probability maps, linear and non-linear registered to MNI space, and the resulting gray-matter tissue class smoothed using an isotropic Gaussian spatial filter (FWHM=8 mm).

Cortical reconstruction and volumetric segmentation were performed with the FreeSurfer (FS) image analysis suite (http://surfer.nmr.mgh.harvard.edu/). This processing includes removal of non-brain tissue, intensity normalization, and segmentation of subcortical structures (Fischl et al., 2002). For validation purposes, we also measured ventricular volume using a semi-automated algorithm, implemented in the program Brain Ventricular Quantification (BVQ; Accomazzi et al., 2005). BVQ uses a seed point/region-growing method and is optimized specifically for segmentation of the lateral ventricles.

Neuropsychological Assessment

Neuropsychological tests with documented sensitivity to the adverse effects of cancer therapy (Correa et al., 2007; Ahles et al., 2010) were selected to evaluate auditory attention and executive functions (Digit Span of the WMS-III - DF/DB; Brief Test of Attention – BTA; Trail Making Test Parts A & B – TMTA, TMTB), verbal memory (California Verbal Learning Test-Second Edition – CVLT-II; Learning – CVLT-L, Long Delayed Recall – CVLT-LD, Discrimination Index – CVLT-DI), and visual-spatial memory (Brown Location Test – BLT; Learning – BLT-L, Long Delayed Recall – BLT-LD, Discrimination Index – BLT-DI). Mood, fatigue, and subjective memory complaints were assessed using the Center for Epidemiological Study-Depression (CES-D) scale and the Functional Assessment of Chronic Illness Therapy-Fatigue subscale (FACIT-FS, Version 4), and the Squire Subjective Memory Questionnaire (SSMQ), respectively. The test battery was administered either by a neuropsychologist (DDC) or by a trained research assistant. Raw cognitive test scores were compared with published normative values according to age, and when available, to age and education, and converted into z-scores.

Imaging and Statistical Analyses

Descriptive statistics for demographic variables were generated with frequencies and percentages for categorical variables and with means and standard deviations, or median and ranges as appropriate, for continuous variables. Given the relatively small sample size, we combined patients treated with HD chemotherapy alone and patients who received fTBI + HD chemotherapy into one group (Combined Patient group). We also performed analyses including the Chemotherapy Alone group and the healthy controls, in order to assess if changes in the variables of interest would occur in patients who did not have fTBI as part of the conditioning regimen. Considering that MPRAGE scans were available for only 5 patients in the fTBI + Chemotherapy group, it was not feasible to conduct a separate analysis for this group.

Voxel-Based Morphometry (VBM)

In the analysis including the Healthy Control versus Combined Patient group, gray matter maps were entered into a 2 (Group: Healthy Control; Combined Patients) X 2 (Time: baseline; 1-year follow-up) flexible factorial design within SPM8. For the Healthy Control versus Chemotherapy Alone group contrast, gray matter maps were entered into a 2 (Group: Healthy Control; Chemotherapy Alone) X 2 (Time: baseline; 1-year follow-up) flexible factorial design. Following omnibus testing, pair-wise t-tests were performed at the group level to analyze within group changes over time (baseline versus 1-year follow-up) for Healthy Control, Chemotherapy Alone, and Combined Patient groups. For cross-sectional analysis between groups at baseline independent samples t-tests were used. For all contrasts, initial uncorrected voxel-wise threshold was p<=0.001; voxels were considered significant if they survived family-wise error correction across the whole volume at p<=0.05 or survived family-wise error correction for a priori small volume correction at p<=0.05 (bilateral middle frontal gyrus).

FreeSurfer (FS), Brain Ventricular Quantification (BVQ)

Repeated measures analysis of variance (ANOVA) with time as a within-subject factor (baseline and one-year follow-up) and group (all patients and healthy controls) as a between-subject factor was used to assess changes in lateral ventricle volume (FS, BVQ) and to evaluate differences in changes between patients and healthy controls. Welch t-tests were used to assess if patients and healthy controls differed on these variables at baseline. The small sample size and reduced power limited the ability to evaluate volumetric changes in specific cortical and subcortical regions using FS image analysis.

In order to prevent large outlying observations to overly influence the results, outliers were identified according to their distances outside of the quartiles of each variable’s distribution. Specifically, any observation more than 1.5 times the variable’s inter-quartile range (i.e., the 25th percentile score subtracted from the 75th percentile score) below the first quartile or above the third quartile was “trimmed” to the next closest, non-outlying value (Tukey, 1997). Ventricular volume measurements were trimmed separately for patient and healthy control groups, and for each of the two assessment times. Adjustments were made for only one patient on total ventricle volume at follow-up.

Neuropsychological Tests and Self-Report Scales

Repeated measures analysis of variance (ANOVA) with time as a within-subject factor (baseline and one-year follow-up) and group (all patients and healthy controls) as a between-subject factor was used to assess changes in the cognitive test z-scores and self-report scale scores, and to evaluate differences in changes between patients and healthy controls. Welch t-tests were used to assess if patients and healthy controls differed on these variables at baseline. The relationship between specific demographic or clinical variables and cognitive test scores was assessed using Pearson’s correlation and t-tests.

Cognitive test z-scores were trimmed separately for patient and healthy control groups, and for each of the two assessment times with the same procedure described above. In the Combined Patient group, adjustments were made for one patient on the TMTA and CVLT-DI, and for two patients on the CVLT-L z-scores at baseline and follow-up. Among the healthy controls, adjustments were performed for one subject’s z-score on the BTA at both assessments times; adjustments were made for two subjects on the CVLT-DI and for three subjects on the BLT-DI at baseline.

Cognitive impairment was defined as two or more z-scores ≥ 1.5 standard deviations below the normative mean, consistent with other studies (Wefel et al., 2004). Changes in neurocognitive test performance between baseline and follow-up at the individual level were assessed using the reliable change index (RCI; Jacobson and Truax, 1991), using published normative data. The RCI is derived from the standard error of measurement for each test, and represents the 90% confidence interval for the difference in raw scores from baseline to follow-up on the same test that could be expected, as a function of measurement error alone. All RCI thresholds were rounded to the nearest whole number. Changes that did not meet the RCI criteria for decline or improvement were classified as stable.

Results

The patients and healthy controls were predominantly men (70%) and right-handed (90%) (Table 1). There were no significant differences between groups in education, estimated IQ, or length of follow-up, but patients who received conditioning treatment with fTBI + HD chemotherapy were significantly younger than patients who received HD chemotherapy alone (t(19) = 2.34, p=0.03). The Combined Patient group did not differ significantly in age from the Healthy Control group. An assessment of employment status showed that 6 patients (21%) were working at the time of enrollment, and 20 (71%) were employed one-year following HSCT. All healthy controls were working at study entry and one individual was unemployed at the one-year follow-up.

Table 1.

Demographic Characteristics and Treatment History

Demographics/Treatment	fTBI + Chemotherapy (N=9)	Chemotherapy Alone (N=19)	Healthy Controls (N=10)
Sex (M/F)	7/2	13/6	7/3
Handedness (R/L)	9/0	18/1	9/1
Age at Baseline (yrs)
Mean (SD)	41.6 (10.4)	52.1 (12.6)*	46.4 (13.2)
Median (range)	39.0 (31–62)	56.0 (25–65)	47.5 (30–65)
Mean Education (yrs)	16.0 (2.1)	16.5 (3.2)	17.3 (2.5)
Mean Est.VIQ	112 (5.6)	114 (5.7)	117 (2.8)
Follow-up Interval (days)
Mean (SD)	396.4 (74.3)	403.4 (41.2)	372.5 (10.6)
Diagnosis
AML	3/9	2/19	NA
ALL	3/9	0/19	NA
Lymphoma	3/9	8/19	NA
Multiple Myeloma	0/9	5/19	NA
Myelodysplastic Syndrome	0/9	4/19	NA
Transplant Type
Allogeneic	9/9	7/19	NA
Autologous	0/9	12/19	NA
Conditioning Chemotherapy
Thio/Flu/Cy	5/9	0/19	NA
Cy/MTX/Ara-C	4/9	0/19	NA
Bu/Mel/Flu	0/9	7/19	NA
BEAM	0/9	6/19	NA
Cytoxan/Mel	0/9	6/19	NA

Est.VIQ= Estimated Verbal IQ (NAART or Barona Index). AML=Acute myeloid leukemia; ALL= Acute lymphocytic leukemia; Thio/Flu/Cy=Thiotepa/Fludarabine/Cyclophosphamide; MTX/Ara-C= Methotrexate/Cytarabine; Bu/Mel/Flu=Busulfan/Melphalan/Fludarabine; BEAM=carmustine/etoposide/cytarabine/melphalan.

^*p= 0.03, fTBI + Chemotherapy vs. Chemotherapy Alone.

Table 1 describes the diagnoses, conditioning treatments and transplant type for all patients. Patients were diagnosed with lymphoma (39%), leukemia (29%), multiple myeloma (18%), or myelodysplastic syndrome (14%). All patients had received chemotherapy to treat their illness prior to becoming a candidate for HSCT. Nineteen patients received conditioning treatment with HD chemotherapy, and nine patients had fTBI (1200–1375 cGy) and HD chemotherapy. Sixteen patients (57%) received an allogeneic HSCT (7 related, 9 unrelated), and twelve (43%) had an autologous HSCT. Most patients (N=15) received T cell depleted grafts. The duration of the pre-transplant conditioning HD-chemotherapy regimens ranged from 6 to 9 days and fTBI was given over a period of four days. The transplants were infused within 24–36 hours after the conditioning regimens, consistent with standard clinical practice (Devine et al., 2011; Hamadani et al., 2011; Jakubowski et al., 2007; Mounier et al., 2012). Following HSCT, three patients had additional treatment with rituximab and one patient had cytarabine. Only three patients developed acute graft-versus-host disease (GvHD), and five patients had chronic GvHD and were treated with prednisone, tacrolimus or methotrexate.

Structural Neuroimaging

Voxel Based Morphometry (VBM), FreeSurfer (FS), and Brain Ventricular Quantification (BVQ)

Cross-sectional assessment revealed no significant differences between the combined patient group and the healthy controls in regional brain volume or in lateral ventricle volume at baseline.

Longitudinal Assessment – Group by Time Interaction

1) VBM

In the Healthy Control versus Combined Patient group analysis, no voxels survived whole brain family-wise error (FWE) correction. Small volume correction with a bilateral middle frontal gyrus mask (Automated Anatomical Labeling; AAL), yielded a trend toward a significant difference (t(29)=4.47, p=0.070) in the left middle frontal gyrus (−32, 21, 55) with smaller volume in the Combined Patient group versus the Healthy Controls.

In the Healthy Control versus Chemotherapy Alone group analysis, no voxels survived whole brain FWE correction. Small volume correction with a bilateral middle frontal gyrus mask (AAL), yielded a significant decrease (t(24)=5.05, p=0.032) in the left middle frontal gyrus (−32, 21, 57) for the Chemotherapy Alone group.

2) FS & BVQ

Repeated measures ANOVA comparing the Combined Patient group and Healthy Controls revealed a significant time effect for mean total lateral ventricle volume (BVQ; F(1,29)=6.25, p=0.018), left lateral ventricle volume (FS; F(1,29)=9.24, p=0.005), and right lateral ventricle volume (FS; F(1,29)=12.29, p=0.002). There was a significant group by time interaction for mean left lateral ventricle volume (FS; F(1,29)=4.49, p=0.044) and a marginally significant interaction for mean total lateral ventricle volume (BVQ; F(1,29)=3.25, p=0.07), suggesting a greater increase in ventricular volume in the Combined Patient Group in comparison to Healthy Controls (Figure 1). A comparison of absolute change in total lateral ventricular volume between groups showed significantly greater ventricular enlargement in the Combined Patient group versus the Healthy Controls (t(18)=1.95, p=0.033).

Figure 1.

Baseline and follow-up left lateral (1a) and total lateral (1b) ventricle mean volume for the combined patient and chemotherapy alone groups, and the healthy controls. *p=0.044; **p=0.025, group by time interaction.

Repeated measures ANOVA comparing the Chemotherapy Alone group and Healthy Controls revealed a significant time effect for mean left lateral ventricle volume (FS; F(1,24)=11.24, p=0.002) and right lateral ventricle volume (FS; F(1,24)=16.85, p<0.001), and a marginally significant time effect for mean total lateral ventricle volume (BVQ; F(1,24)=4.09, p=0.055). There was a significant group by time interaction for mean left lateral ventricle volume (FS; F(1,24)=5.76, p=0.025), suggesting a greater increase in the Chemotherapy Alone Group in comparison to Healthy Controls (Figure 1). A comparison of absolute change in total lateral ventricular volume between groups showed that there was greater ventricular enlargement in Chemotherapy Alone group versus the Healthy Controls, but it did not reach statistical significance (t(18)=1.42, p=0.086).

Longitudinal Assessment – Within Group

1) VBM

Pair-wise t-tests for longitudinal changes within for the Combined Patient group showed a significant decrease in the mean volume of the right and left middle frontal gyri (left MFG: −30, 12, 58 FWE-corr p=0.008; right MFG: 34, 14, 51 FWE-corr p=0.048) and left caudate (−3, 12 -2 FWE-corr p=0.030) one year after HSCT (Figures 2a, 2b, 2e; Table 2). Among patients treated with chemotherapy alone, there were significant decreases in mean volume in the left middle frontal gyrus (−34, 12, 48 FWE-corr p=0.032) from baseline to the one year follow-up (Figures 2c, 2d; Table 2). Pair-wise t-tests for longitudinal changes within the Healthy Control group found no significant volumetric change from baseline to the one-year follow-up either for whole brain or small volume correction.

Figure 2.

Regional gray matter reductions (VBM) from baseline to the one-year follow-up in the left and right middle frontal gyrus (MFG) in the Combined Patient group (Left MFG FWE-corr p=0.008; Right MFG FWE-corr p=0.048) (2a & 2b), and in the Chemotherapy Alone group (Left MFG FWE-corr p=0.032) (2c & 2b). Gray matter reductions in the left caudate nucleus from baseline to the one-year follow-up in the Combined Patient group (FWE-corr p=0.030) (2e).

Table 2.

Significant Regional Mean Volume Changes – Voxel-Based Morphometry (VBM)

MNI Coordinates (x y z)			pFWE-corr	T	Region Description
Group x Time Interaction
Healthy Controls vs. Combined Patient Group
−32	21	55	0.070*	4.47	L Middle Frontal Gyrus
Healthy Controls vs. Chemotherapy Group
−32	21	57	0.032*	5.05	L Middle Frontal Gyrus
Within Group – Time Effect
Combined Patient Group (N=21)
−30	12	58	0.008	7.62	L Middle Frontal Gyrus
34	14	51	0.048	6.56	R Middle Frontal Gyrus
−3	12	−2	0.030	6.85	L Caudate
Chemotherapy Alone Group (N=16)
−34	12	48	0.032	8.09	L Middle Frontal Gyrus

^*Middle Frontal Gyrus (bMFG) Small Volume Corrected

MNI=Montreal Neurological Institute; FWE-corr=family-wise error corrected.

2) BVQ

In the combined patient group, there was a significant increase in the mean total lateral ventricular volume (BVQ; 8.1% (± 9.5), p=0.003) one year after HSCT. Among patients treated with chemotherapy alone, there was a significant increase in mean total lateral ventricle volume (BVQ; 6% (± 9.1), p=0.02). There were no significant changes in ventricular volume in the healthy control group over the follow-up period (percent increase= 2.9% ± 8.4).

Neuropsychological Evaluation

Cross-sectional assessment revealed that at baseline, the Healthy Control group had significantly higher z-scores than the Combined Patient group in the CVLT-Learning, t(14)= −4.2, p=0.001 and BLT-Discrimination Index, t(34)= 2.1, p=0.041.

Longitudinal Assessment – Group by Time Interaction

1. Combined Patient Group and Healthy Controls

Repeated measures ANOVA revealed a significant group by time interaction for TMTA (F(1,36)=5.18, p=0.029), as Healthy Controls but not the Combined Patient group improved significantly at the one year follow-up (Table 3). On the BLT-Learning, there was a significant time main effect (F(1,36)=20.85, p<0.001) and a marginally significant group by time interaction (F(1,36)=4.02, p=0.053), as Healthy Controls showed marginally greater improvement than patients one year post-HSCT. On the BLT-Discrimination Index, there was a significant group by time interaction (F(1,36)=5.96, p=0.02), as scores improved significantly for Healthy Controls but not for the Combined Patient group at follow-up (Table 3). On the CVLT-Learning and CVLT-Delay, there was a significant time main effect (F(1,36)=25.86, p<0.001; (F(1,36)=6.25, p=0.017). There were no significant changes over time in any other cognitive tests or in the self-report measures of mood, fatigue and subjective memory.

Table 3.

Cognitive Test Z-Scores (Means and Standard Deviations)

Measures	All Patients (N=28)		fTBI+Chemo (N=9)		Chemo Alone (N=19)		Controls (N=10)
	PRE	POST	PRE	POST	PRE	POST	PRE	POST
Attention/Executive
DF	0.3 (0.9)	0.4 (1.0)	0.4 (0.9)	0.4 (1.0)	0.3 (0.9)	0.3 (1.0)	0.8 (0.8)	1.1 (0.6)
DB	0.4 (0.9)	0.3 (1.0)	0.3 (0.8)	0.0 (1.0)	0.5 (0.9)	0.5 (0.9)	0.7 (1.0)	1.0 (1.2)
BTA	−0.1 (0.9)	−0.4 (1.1)	0.0 (0.8)	−0.3 (1.3)	−0.1 (0.9)	−0.4 (1.0)	0.1 (1.0)	0.4 (0.6)
TMTA	−0.1 (0.6)	−0.1 (0.9)	−0.2 (0.7)	0.0 (0.9)	−0.1 (0.6)	−0.2 (0.9)	0.3 (1.2)	1.1 (1.3)*
TMTB	−0.05 (0.9)	0.02 (0.8)	−0.2 (0.8)	−0.2 (0.6)	0.0 (1.0)	0.2 (0.8)	0.4 (1.4)	1.0 (1.6)
Verbal Memory
CVLT-L	0.5 (0.6)	1.2 (0.7)*	0.4 (0.6)	1.2 (1.1)*	0.5 (0.5)	1.2 (0.8)*	1.4 (0.6)	2.0 (0.9)*
CVLT-LD	0.5 (0.7)	0.8 (1.0)	0.4 (0.7)	0.6 (1.1)	0.6 (0.7)	0.9 (1.0)*	0.8 (0.5)	1.3 (0.5)*
CVLT-DI	0.6 (0.8)	0.7 (0.9)	0.7 (0.4)	0.4 (0.8)	0.5 (0.9)	0.8 (0.9)*	0.8 (0.5)	0.7 (0.8)
Visual Memory
BLT-L	−0.2 (1.2)	0.2 (1.1)*	0.2 (1.5)	0.3 (1.3)	−0.4 (1.1)	0.2 (1.0)*	−0.3 (1.4)	0.8 (1.6)*
BLT-LD	−0.3 (1.3)	−0.1 (1.3)	0.5 (1.1)	0.2 (1.6)	−0.6 (1.2)	−0.3 (1.1)	−0.1 (1.2)	0.5 (1.0)*
BLT-DI	0.2 (1.3)	−0.1 (1.4)	0.4 (1.2)	−0.1 (1.4)*	0.1 (1.3)	−0.1 (1.5)	−0.4 (0.3)	0.4 (1.1)*

DF= Digit Span Forward; DB= Digit Span Backward; BTA= Brief Test of Attention; TMTA= Trail Making Test A; TMTB= Trail Making Test B; CVLT-L= California Verbal Learning Test- Learning; CVLT-LD= California Verbal Learning Test- Long Delay; CVLT-DI= California Verbal Learning Test- Discrimination Index; BLT-L= Brown Location Test- Learning; BLT-LD= Brown Location Test-Long Delay; BLT-DI=Brown Location Test-Discrimination Index.

^*Indicates significant within-group difference (p < 0.05) between the pre- and post-HSCT scores on the given cognitive test.

2. Chemotherapy Alone Group and Healthy Controls

Repeated measures ANOVA revealed a significant group by time interaction for TMTA (F(1,27)=5.52, p=0.026), as Healthy Controls but not the Chemotherapy Alone group improved significantly at the one year follow-up (Table 3). On the BLT-Learning and BLT-Delay, there was a significant time main effect (F(1,27)=19.16, p<0.001; F(1,27)=7.13, p=0.013). On the BLT-Discrimination Index, there was a marginally significant group by time interaction (F(1,27)=3.55, p=0.07), as scores improved significantly for Healthy Controls but not for the Chemotherapy Alone group at follow-up (Table 3). On the CVLT-Learning and CVLT-Delay, there was a significant time main effect (F(1,27)=19.66, p<0.001; F(1,27)=8.94, p=0.006). There were no significant changes over time in any other cognitive tests or in the self-report measures of mood, fatigue and subjective memory.

Within-subject evaluation of change in cognitive test performance

Mean z-scores were within the average range across most cognitive tests in all groups at baseline and at the one-year follow-up (Table 3). Evaluation of individual scores revealed that at baseline, 6 patients (21%) and one healthy control met criteria for cognitive impairment. In comparing the baseline and one-year follow-up performance using the RCI, the rates of cognitive decline or improvement did not differ significantly between patients and healthy controls. Seven of 28 patients (25%) had a reliable decline on at least one test, and 3 of 10 (30%) healthy controls, had a reliable decline on one test. On the other hand, a reliable improvement on at least one test was seen in 11 of 28 patients (39%) and in 5 of 10 (50%) healthy controls across all tests.

Correlations between structural imaging and cognitive scores and self-report measures

There were no significant correlations between regional gray matter volumes and the cognitive tests, or between the cognitive tests and the self-reported fatigue (FACIT-FS) or depression (CESD-D) scales. Scores on the SSMQ were significantly correlated with DB (r=0.33, p<0.05), and TMTA and TMTB (r=0.38, p<0.05). However, after adjusting for multiple comparisons (using the false discovery rate (FDR) method) none of the correlations remained significant.

Discussion

This prospective study documented significant brain structural changes across imaging modalities in HSCT recipients one year following conditioning treatment with HD chemotherapy with and without fTBI, suggesting that this treatment may be associated with the development of neurotoxicity. Group by time interactions indicated a significant reduction in gray matter volume in the left middle frontal gyrus in the Chemotherapy Alone group and a marginally significant reduction in this region in the Combined Patient group, relative to Healthy Controls. Within group analyses showed significant reductions in gray matter volume in the middle frontal gyrus bilaterally and in the left caudate in the Combined Patient group, and in the left middle frontal gyrus in the Chemotherapy Alone group one year after HSCT. There were no significant reductions in gray matter volume among the Healthy Controls. Group by time interactions indicated a significant increase in left lateral ventricle volume in the Combined Patient group and in the Chemotherapy Alone group relative to Healthy Controls at the one-year follow-up; marginally significant interactions were also seen for total lateral ventricular volume (BVQ). Within group analyses indicated a significant increase in total lateral ventricle volume in the Combined Patient group and in the Chemotherapy Alone group, but not among the Healthy Controls. The relatively small number of patients and healthy controls most likely reduced the power to detect some of the interactions, and variability in the extent of brain structural changes among patients may have also contributed to some of the marginally significant results. The within subjects analyses was conducted to further explore the pattern of changes. Cross-sectional analysis showed no significant differences in ventricular or gray matter volume between the Combined Patient group and Healthy Controls at baseline, and this may have been also in part related to the reduced sensitivity of the between-subject versus within-subject analysis to detect differences in this relatively small cohort.

The earlier studies describing white matter abnormalities and atrophy in subgroups of HSCT patients treated with TBI and chemotherapy were mostly retrospective and did not include a healthy control group (Garrick, 2000; Padovan et al., 1998; Peper et al., 2000; Stemmer et al., 1994). In addition, some studies reported that neurological and neuroradiological abnormalities in this population were associated with GvHD and immunosuppressive therapy (Padovan et al., 1998; Sostak et al., 2003). This is the first prospective study to document reductions in regional gray matter volume and ventricular enlargement in HSCT patients treated with HD chemotherapy with or without fTBI. In this study, almost half of the sample had an autologous transplant and most of the patients who underwent an allogeneic transplant received T cell depleted grafts which do not require the use of calcineurin inhibitors; only five patients developed chronic GvHD. This suggests that the documented brain structural abnormalities including atrophy and regional gray matter reductions may be related primarily to the conditioning treatment with HD chemotherapy with or without fTBI. Considering that all patients had received chemotherapy prior to becoming a candidate for HSCT, the cumulative neurotoxic effects of multiple regimens, including fTBI in some patients, most likely contributed to the development of brain structural abnormalities. However, the contribution of disease-related factors or the possible interaction of disease and treatment cannot be excluded given the absence of a pre-treatment baseline. The small number of patients who received fTBI and HD chemotherapy also limited the evaluation of the more specific contribution of radiotherapy. In a recent study (Jim et al., 2012), clinical risk factors such as pre-transplant treatment, transplant type and the development of complications, were found to have a cumulative adverse effect on cognitive function in HSCT patients.

The study results provide further evidence for the neurotoxicity associated with chemotherapy documented in recent structural imaging studies of breast cancer patients (Ferguson et al., 2007; Inagaki et al., 2007; Deprez et al., 2011, 2012; Koppelmans et al., 2012). McDonald and colleagues (2010) documented a decrease in gray matter volume in bilateral frontal, temporal and cerebellar regions in breast cancer patients one month following chemotherapy, and partial recovery after one year. These results were replicated on a separate cohort, and provided additional evidence that chemotherapy was associated with reductions in frontal gray matter (McDonald et al., 2012). Our findings of decreased gray matter volume in prefrontal regions are overall consistent with these results, and provide convergent data suggesting that brain regions may be differentially affected by disease and treatment. The small number of patients treated with fTBI and HD chemotherapy precluded an evaluation of the specific effects of this combined regimen, but the synergistic toxic effect of HD chemotherapy in combination with radiotherapy has been documented in patients with brain tumors (DeAngelis and Posner, 2009; Correa et al., 2012). The etiology of chemotherapy- and radiotherapy-related brain structure abnormalities is unknown, but proposed mechanisms include oxidative stress and DNA damage, inhibition of neurogenesis, inflammation, immune dysregulation and stimulation of neurotoxic cytokines (Ahles et al., 2007; Seigers & Fardell, 2011), and vascular damage and demyelination (DeAngelis and Posner, 2009). The percent and absolute changes in ventricular enlargement in the Combined Patient group over one year was similar to the rates reported in patients with Mild Cognitive Impairment in a recent study using BVQ (Nestor et al., 2008); however, similar to our sample the authors also documented large intra-group variations among patients and controls. Accelerated atrophy and decrease in regional gray matter volume could place these patients at increased risk for future age-related decline in cognitive function.

The longitudinal assessment of neuropsychological test performance showed significant group by time interactions on a timed test of visual scanning and graphomotor speed and on a visual-spatial recognition memory task, as Healthy Controls but not the Combined Patient group improved significantly at follow-up. Similar results were seen for comparisons including the Chemotherapy Alone group and the Healthy Controls. Patients and healthy controls improved significantly on tests of verbal and visual-spatial learning, and showed no significant changes on self-report measures of mood, fatigue or subjective memory from baseline to the one-year follow-up. Cross-sectional analysis showed that at baseline, the Combined Patient group had significantly lower scores on verbal learning and visual-spatial recognition memory than Healthy Controls. Although there was a one-year interval between the two evaluations, the findings suggest that prior exposure to the neuropsychological tests may have resulted in practice effects on some of the tests, which were greater for the healthy controls than for the patients on several measures, consistent with some studies of breast cancer patients (Ahles et al., 2012).

An evaluation of individual cognitive scores showed that 21% of patients met criteria for cognitive impairment at baseline. These results are overall consistent with the literature reporting that 20–40% of HSCT candidates have cognitive dysfunction prior to transplant (Andrykowski et al., 1992; Chang et al., 2009; Friedman et al., 2009; Harder et al., 2005, 2006; Jacobs et al., 2007; Meyers et al., 1994; Schulz-Kindermann et al., 2007); however, different criteria for the definition of impairment may have been used across studies. Disease and prior exposure to chemotherapy, and possibly circulating levels of cytokines (Meyers et al., 2005) and anemia (Jacobsen et al., 2004;Wood et al., 2011) may contribute to these findings, but factors that place some patients at a greater risk for cognitive dysfunction remain unknown. Studies of breast cancer patients reported that 20% to 30% of patients performed below expected levels based on age norms at diagnosis and prior to treatment, and this was not associated with mood or fatigue (Ahles et al., 2008; Wefel et al., 2004). In this population, possible risk factors for cognitive dysfunction include an interaction of age, cognitive reserve and exposure to chemotherapy (Ahles et al., 2010), and genetics (Ahles et al., 2003; Small et al., 2011).

At the one-year follow-up, only 14% of our patients met criteria for cognitive impairment. Within-subject evaluation of change in cognitive test performance using the RCI showed that 25% of the patients had a reliable decline on at least one test, and 30% of the healthy controls declined on only one test. On the other hand, a reliable improvement on at least one test was seen in 39% of patients and in 50% of the healthy controls. There were no significant differences between patients and healthy controls in rate of cognitive decline. The RCI analyses provided complementary information to the group comparisons, suggesting that it is useful in identifying subgroups of patients that may develop cognitive dysfunction (Friedman et al., 2009). The findings are in part consistent with longitudinal studies reporting that one year post-HSCT, cognitive functions returned to or surpassed pre-transplant levels (Chang et al., 2009; Harder et at., 2006, 2007; Jacobs et al., 2007; Syrjala et al., 2004; Wenz et al., 2000), but that subgroups of patients showed persistent impairment or declined on some cognitive domains (Jacobs et al., 2007; Meyers et al., 1994; Syrjala et al., 2011). Studies including patients with other non-CNS cancers also described that subgroups have cognitive dysfunction following chemotherapy (Vardy et al., 2008). There were no significant correlations between scores on any of the cognitive tests and self-reported fatigue or mood, and this is consistent with other studies (Friedman et al., 2009; Ahles et al., 2008; Wefel et al., 2004). More than 70% of patients were employed one year post-HSCT.

In this study, the group changes in brain structure were more pronounced than in cognitive test performance one year post-HSCT. Also, there were no significant associations between regional gray matter volume and cognitive test scores. These results may be in part related to reduced power to detect small changes and associations, and possibly to practice effects on the cognitive tests. A recent prospective study using functional MRI in patients with breast cancer treated with or without chemotherapy (McDonald et al., 2012) showed baseline increased activation in bifrontal regions and decreased left parietal activation compared to controls. One month after chemotherapy, there was a decrease in frontal activation in patients compared to controls, with partial return to baseline levels one year later. However, performance accuracy and reaction time on a working memory task did not differ between groups at baseline or over time for the individual task conditions, suggesting that hyperactivation of the underlying brain circuitry may have been required to support adequate working memory performance. An earlier fMRI study (Ferguson et al., 2007) also reported comparable working memory task performance and increased activation in the frontal lobes in a monozygotic twin with breast cancer treated with chemotherapy, relative to her healthy twin; these findings were thought to suggest that more cortical activation and recruitment of additional regions were needed to compensate for dysfunction in the neural circuitry possibly disrupted by chemotherapy. Therefore, it is possible that some patients in our study were able to maintain cognitive performance at age-expected levels through the use of compensatory mechanisms in the context of reductions in regional gray matter volume.

The longitudinal study design provided new and relevant information regarding changes in brain structure and cognitive function in HSCT recipients. However, an assessment of several important factors including the specific contribution of disease type, conditioning regimen and transplant type, and of known risk factors such as GvHD and immunosuppressive therapy, was limited by the small sample size, the absence of pre-treatment assessments, non-randomization to treatment, and attrition bias. Attrition contributed to a reduction in our projected sample size, and it is a significant limitation in many longitudinal studies of cancer patients (Correa et al., 2008); the relatively small sample size most likely limited the power to detect small changes in brain structure and cognitive functions. As there has been an increase in the number of cancer patients who undergo HSCT, additional prospective studies are needed to further characterize the neurotoxicity associated with commonly used conditioning regimens, and improve our ability to identify individuals who may be at an increased risk for cognitive decline. Research including a systematic evaluation of possible risk factors, such as genetics, age, disease and treatment-related factors in large cohorts would be particularly relevant. Studies designed to test potential interventions to minimize or compensate for changes in cognitive function and quality of life in HSCT patients are also warranted.