Lateralization of temporal lobe epilepsy using resting functional magnetic resonance imaging connectivity of hippocampal networks

Authors


Address correspondence to Victoria L. Morgan, Vanderbilt University Institute of Imaging Science, 1161 21st Avenue South, AA 1105 MCN, Vanderbilt University, Nashville, TN, U.S.A. E-mail: victoria.morgan@vanderbilt.edu

Summary

Purpose:  Early surgical intervention can be advantageous in the treatment of refractory temporal lobe epilepsy (TLE). The success of TLE surgery relies on accurate lateralization of the seizure onset. The purpose of this study was to determine whether resting functional MRI (fMRI) connectivity mapping of the hippocampus has the potential to complement conventional presurgical evaluations in distinguishing left from right TLE. In addition, we sought to determine whether this same network might separate patients with favorable from unfavorable postoperative outcomes.

Methods:  Resting fMRI acquisitions were performed on 21 patients with TLE and 15 healthy controls. The patients included seven patients with left TLE and seven patients with right TLE with seizure-free postoperative outcome, and five patients with left TLE and two patients with right TLE with recurring seizures after surgery. Functional connectivity maps to each hippocampus were determined for each subject and were compared between the controls and the seizure-free patients with left TLE and with right TLE. The one network identified was then quantified in the patients with TLE and recurring seizures.

Key Findings:  The resting functional connectivity between the right hippocampus and the ventral lateral nucleus of the right thalamus was the most statistically significant network to distinguish between seizure-free patients with left TLE and with right TLE with high sensitivity and specificity. This connectivity was also significantly greater in the seizure-free patients with left TLE than the healthy controls. Finally, six of the seven patients in whom seizures recurred after surgery had connectivity values in this network unlike those who were seizure-free.

Significance:  This study identified a region in the ventral lateral nucleus of the right thalamus whose connectivity to the hippocampi separates left from right TLE subjects. This suggests that the quantification of resting-state functional magnetic resonance imaging (MRI) connectivity across this network may be a potential indicator of lateralization of TLE that may be added to other presurgical MRI assessments. Further validation in a larger, independent cohort is required.

Early surgical intervention can be advantageous in the treatment of refractory temporal lobe epilepsy (TLE) (Wiebe et al., 2001). Uncontrolled seizures themselves result in an array of psychological consequences including isolation, anxiety, and depression (de Boer et al., 2008). In addition, chronic TLE is associated with increased cognitive decline and progressive memory impairment (Helmstaedter et al., 2003). The success of TLE surgery relies on accurate lateralization of the seizure onset. Conventional presurgical evaluations (McKhann et al., 2002; Siegel, 2004; Jobst, 2009; Duncan, 2010) can be effective for this purpose. In magnetic resonance imaging (MRI) studies that are positive for hippocampal sclerosis, reports indicate that from 66–87% (Janszky et al., 2005; Ozkara et al., 2008; Ramos et al., 2009) of patients are seizure-free with or without auras at 3 years postsurgery. In patients without MRI lesions, approximately 45% (Bien et al., 2009) to 60% (Bell et al., 2009) are seizure-free.

Functional MRI (fMRI) can provide complementary information to aid in the lateralization of TLE seizure onset. This MRI method creates images that are sensitive to blood oxygenation level (Ogawa et al., 1990; Logothetis et al., 2001). Blood oxygenation increases in brain regions employed or “activated” during a specific task, and therefore these regions can be localized noninvasively when the task is performed during fMRI image acquisition. Jokeit et al. (2001) prompted patients with TLE to envision a walk through their hometown including specific landmarks along the route. Asymmetric activation in the mesial temporal regions induced by this long-term memory retrieval task was quantified and compared between patients with left and those with right TLE. The results showed increased activation contralateral to the seizure onset in 90% of patients.

In addition to localizing regions of activation, synchronous low frequency fluctuations in blood oxygenation measured by fMRI can be used to identify brain networks (Biswal et al., 1995; Rogers et al., 2007). Correlation of the fMRI signal across brain regions is referred to as functional connectivity. Measured during the resting state in subjects awake with eyes closed, increases in functional connectivity across regions of the mesial temporal lobe contralateral to seizure onset, and decreases ipsilateral to the seizure onset were detected (Bettus et al., 2009, 2010). These fMRI studies suggest that functional connectivity is altered with progressive seizures; and functional connectivity mapping may potentially identify these changes in order to discriminate between patients with left and those with right TLE.

The primary objective of this study was to identify resting fMRI connectivity differences between patients with left TLE, patients with right TLE, and control subjects, specifically of networks involving the hippocampus across the whole brain. We hypothesized that there are distinct differences between hippocampal networks in left TLE and right TLE that can potentially lateralize TLE. A homogenous cohort of patients with confirmed lateralization who became seizure-free after mesial temporal resection was studied. The secondary objective of this study was to investigate the identified networks in a small cohort of patients with TLE who had recurring seizures after mesial temporal surgery. Although our sample size does not allow statistical evaluation of this small group, we hoped to gain insight as to whether the identified networks could potentially distinguish between patients who would become seizure-free and those that would not. Note that we did not combine the seizure-free patients and those with recurring seizures in the first analysis, because our assumption is that those with recurring seizures may not have the same pathology or connectivity patterns as those who became seizure-free, even if other presurgical evaluations are unable to distinguish them.

Methods

Subjects

From a population of 57 patients with TLE who were recruited from the Vanderbilt University Epilepsy Program for imaging, 21 patients were included in this study. All patients underwent presurgical inpatient video–electroencephalography (EEG) monitoring for localization of the epileptogenic zone, high-resolution structural MRI, fluorodeoxyglucose positron emission tomography (FDG-PET), neuropsychological testing, and, in select patients, Wada test. Inclusion criteria for this study were the following: (1) temporal lobe epilepsy determined by presurgical evaluation, (2) no foreign tissue lesions, and (3) temporal resection as epilepsy surgery. Seven subjects with left TLE (five female and two male, one left handed, mean age ± standard deviation [SD] 38.2 ± 9.8 years) and seven subjects with right TLE (five female and two male, one left handed, mean age ± SD 36.8 ± 11.6 years) became seizure-free after surgery. These subjects formed the patients with left and those with right TLE with seizure freedom groups (“sz free”). Five patients with left TLE and two patients with right TLE experienced postsurgical seizures. These subjects formed the patients with left and patients with right TLE seizure-recurring groups (“sz recur”). See Table 1 for all patient characteristics. Finally, in addition, we recruited 15 right-handed healthy control subjects for comparison (12 female and 3 male, mean age ± SD 31.3 ± 10.8 years). There is no statistically significant difference in age between the combined (“sz free” and “sz recur”) right TLE (n = 9), combined left TLE (n = 12), and control (n = 15) groups. Similarly, there was no statistical difference in duration of disease between the combined right TLE and left TLE groups.

Table 1.   Patient characteristics
IDGender/HandednessAge (years)Age of onset (years)Seizure typesIctal EEGMRI- hippocampus + temporal lobePET hypometabolismSurgeryTime since surgery (months)
  1. F, female; M, male; sz, seizure; SPS, simple partial seizures; CPS, complex partial seizures; SGTC, secondary generalized tonic–clonic seizures; R, right; L, left; TEMP, temporal; MTS, mesial temporal sclerosis; MT, mesial temporal lobe hypometabolism on PET; Sel AH, selective amygdalohippocampectomy; St TL, standard temporal lobectomy.

Right TLE sz free         
 16M/L5328SPS, CPS, SGTCR TEMPNormalR MTR Sel AH16
 25M/R337SPS, CPSR TEMPR MTSR MTR Sel AH24
 37F/R4638SPS, CPS, SGTCR >> L TEMPNormalNormalR Sel AH16 (had only 3 CPS with AED withdrawal)
 49F/R2923CPS, SGTCR TEMPR MTSR MTR Sel AH17
 51F/R4323SPS, CPS, SGTCR TEMPR MTSR MTR Sel AH13
 58F/R360.6CPS, SGTCR TEMPR MTSR MTR Sel AH14
 10F/R1815.5SPS, CPSR TEMPNormalR MTR St TL20
Left TLE sz free         
 3F/R3518SPS, CPS, SGTCL TEMPL MTS + lateral encephalomalaciaL MTL St TL after subdural grids25
 13F/R4927SPS, CPS, SGTCL TEMPNormalNormalL Sel AH after subdural strips28 (had two seizures with AED withdrawal)
 17F/R3126SPS, CPS, SGTCL TEMPL MTSL MTL Sel AH16
 21F/L471.4SPS, CPS, SGTCL TEMPNormalL MTL St TL17
 34F/R465SPS, CPS, SGTCL TEMPL MTSL MTL Sel AH15
 42M/R387SPS, SGTCL TEMPL MTS + Severe global left temporal lobe atrophyGlobal left temporal lobeL Sel AH after subdural grids10
 46M/R2216CPS, SGTCL TEMPNormalL MTL St TL after subdural grids14
Right TLE sz recur         
 52M/R3325CPS, SGTCR TEMPR MTS, encephalomalacia in the R insulaR MTR St TL11
 50M/L4414SPS, CPS, SGTCR TEMPR MTSR MTR St TL after subdural grids14
Left TLE sz recur         
 11F/R220.5SPS, CPS, SGTCL TEMPL MTSL MTL Sel AH31
 30M/R441.5SPS, CPS, SGTCL TEMPBilateral MTS, L>RL MTL Sel AH12
 53F/R191.4SPS, CPS, SGTCL TEMPL MTSL MTL Sel AH12
 2M/R2317CPS, SGTCL TEMPL MTSL MTL Sel AH31
 14M/R4643SPS, CPS, SGTCL TEMPL MTSL MTL Sel AH28

Imaging

All MRI imaging was performed using a Philips Achieva 3T MRI scanner (Philips Healthcare, Inc., Best, The Netherlands) using an 8-channel head coil. Written informed consent was obtained prior to scanning per institutional review board guidelines. The images analyzed for this study were part of a more comprehensive protocol and included the following scans: (1) three-dimensional, T1-weighted high-resolution image series across the whole brain for intersubject normalization (1 × 1 × 1 mm3); (2) two-dimensional, T1-weighted high-resolution axial full-brain image series in the same slice locations as the fMRI scan for functional to three-dimensional data registration (1 × 1 × 5 mm3); (3) fMRI blood oxygenation–level dependent (BOLD) image series at rest with eyes closed – 64 × 64, field of view (FOV) = 240 mm, 30 axial slices, echo time (TE) = 35 msec, repetition time (TR) = 2 s, slice thickness = 4.5/0.5 mm gap, 300 volumes. The controls underwent the same protocol with 200 volumes for the fMRI scan. For this study, 200 volumes were analyzed for each subject.

Hippocampal volumes

Although we attempted to maximize homogeneity across the groups by including only those patients with the similar surgical treatments and outcomes, there remains variability in the volumes of the hippocampus within the groups. Of the “sz free” patients, three with left TLE and three with right TLE have hippocampi and temporal lobes determined as normal by MRI (see Table 1). The small number of patients prohibited those with hippocampal sclerosis from being separated into unique groups. We, therefore, needed to quantify these volumes and determine the effect of volume on the functional connectivity.

The estimation of the gray matter (GM) volume of the left and right hippocampi of an individual was determined in the following way using SPM8 software (http://www.fil.ion.ucl.ac.uk/spm/software/spm8/). First, the high spatial resolution T1-weighted MRI scan (scan 1 above) was spatially normalized to the Montreal Neurological Institute (MNI) template. Second, this image set was segmented into GM, white matter (WM), and cerebrospinal fluid (CSF). Next, hippocampal masks were created on the template in the left (LH, 7,360 mm3) and right hippocampus (RH, 7,616 mm3) individually using WFU Pickatlas (Maldjian et al., 2003, 2004). Finally, the hippocampal masks were applied to the individual’s segmented GM image to obtain the GM volume of each hippocampus in cm3. To normalize these volumes across patients, the right hippocampus volume was divided by the left hippocampus volume to determine a hippocampal volume fraction. To validate this procedure, the normalize volume fractions were compared between all right TLE, left TLE, and controls with a one-way analysis of variance (ANOVA) with post hoc, pair-wise Scheffe tests using SPSS v20 (IBM Corp., Somer, NY, U.S.A.). The expectation was that the patients with right TLE should have lower values than the patients with left TLE, indicating smaller right hippocampal volumes on average. However, those without hippocampal sclerosis may have values closer to 1.0 (equal volumes) as in the controls.

Whole brain hippocampal functional connectivity

The fMRI image analyses were performed using SPM8 software. The fMRI images were corrected for slice-timing effects and motion and spatially normalized to the MNI template using coregistration to the three-dimensional and two-dimensional T1-weighted structural image sets as intermediate steps. The images were then spatially smoothed with a 7 × 7 × 7 mm full-width half maximum (FWHM) kernel. This resulted in functional image series of 46 × 55 × 46 voxels (4 × 4 × 4 mm3).

Seed regions of interest for all subjects were created in the LH and RH using the hippocampal masks described above from the WFU Pickatlas (Maldjian et al., 2003, 2004). The fMRI images from each subject were then processed as follows to set up the general linear model to create functional connectivity maps to each of the two seed regions. The time series from the LH from each subject’s fMRI dataset were high-pass filtered at 0.001 Hz to remove low frequency drifts and then averaged to create a seed time series. The signal averaged from all the voxels in the GM and in the CSF in the brain was also determined. These two signals were considered estimates of physiologic noise in the fMRI images (Weissenbacher et al., 2009). The GM signal, the CSF signal, as well as the time series from the six motion parameters determined by SPM8, were linearly regressed from the seed time series. The seed time series was then low-pass filtered at 0.01 Hz (Cordes et al., 2001). The fMRI images for all subjects were also low-pass filtered at 0.01 Hz. Finally, a general linear model (GLM) (Friston et al., 1995) was created using the processed LH seed time series as the regressor and the GM and CSF signal and motion time series as confounds. By performing the voxel-wise GLM on the low-pass filtered fMRI images, a functional connectivity map to the seed region was created. The process was repeated using the RH seed region in each patient and control subject.

To identify any regions of significant difference in functional connectivity to the LH between left TLE “sz free” subjects (n = 7), right TLE “sz free” subjects (n = 7) and controls subjects (n = 15), an ANOVA test was performed in SPM8 using the LH connectivity maps determined above for each of the subjects in these three groups. The same process was then repeated with the RH connectivity maps to determine regions of significant difference of RH connectivity between the groups. To verify that any regions detected using the ANOVAs are not related to the variations in hippocampal volume fraction, we repeated each of the ANOVAs with the hippocampal volume fractions as a covariate. Finally, the hippocampal connectivity of regions of significance determined in the above ANOVA analyses were linearly correlated with duration of disease, maximum translational motion during the acquisition, and hippocampal volume fraction using SPSS to determine whether any of these parameters are potential other confounds in the analysis.

Results

All subjects had translational motion less than one voxel length throughout the fMRI scanning and were included in the analyses. The right TLE group (1.55 ± 1.00 mm) had significantly higher motion than the left TLE group (0.76 ± 0.39 mm), which had higher motion than the controls (0.32 ± 0.16 mm). However, this motion was not linearly correlated with the functional connectivity in the network described below.

The calculated hippocampal volume fraction was found to be significantly different across the three groups via ANOVA (right TLE = 0.958 ± 0.131, left TLE = 1.137 ± 0.175, controls = 0.975 ± 0.045; p = 0.002). Post hoc Scheffe tests showed that patients with left TLE had significantly higher volume fraction than patients with right TLE (p = 0.010) and controls (p = 0.008). These values were not linearly correlated with the functional connectivity in the network described below. Similarly, the duration of disease was not linearly correlated with the functional connectivity of the identified network.

The ANOVA (F test) in SPM8 testing for group differences showed two regions that met significance at p < 0.0001 (uncorrected) cluster size 5 in the RH connectivity maps. One region included part of the right thalamus (RTHAL) with separation between with right TLE and the left TLE patients (MNI coordinate 22, −14, 12 mm, F = 19.41, volume = 576 mm3) (Fig. 1). A second region in the WM adjacent to the right precuneus indicated most difference between the patients with right TLE and the controls (MNI coordinate 26, −54, 20 mm, F = 16.30, volume = 576 mm3). We focused our interest on the RTHAL region because it had the highest statistical significance from the ANOVA and showed greater separation between the two patient groups. No regions of connectivity to the LH were significantly different between the three groups (left TLE “sz free,” right TLE “sz free,” and controls) at p < 0.0001 (uncorrected) cluster size 5. The ANOVAs performed using the hippocampal volume fraction as a covariate showed the same peak regions of activation as those without the covariate.

Figure 1.


Region whose connectivity to the RH was significantly different between the three groups— right TLE “sz free” (n = 7), left TLE “sz free” (n = 7), and controls (n = 15) by ANOVA (p < 0.0001 uncorrected, cluster size 5). (MNI coordinate 22, −14, 12 mm, F = 19.41 volume = 576 mm3, RTHAL).

The mean connectivity value between the RH and the RTHAL cluster in each subject is shown in Fig. 2 (right TLE = −0.155 ± 0.197, left TLE = 2.910 ± 0.091, controls = −0.302 ± 0.200). Unpaired t-tests between the “sz free” groups and the controls showed that the left TLE “sz free” patients had higher connectivity between the RH and RTHAL than both right TLE “sz free” (p < 0.0001) and controls (p < 0.0001). Next, the sensitivity (true positive/[true positive + false negative]) and specificity (true negative/[true negative + false positive]) was calculated for a cutoff value to correctly categorize the right TLE “sz free” patients from the left TLE “sz free” patients based on the connectivity from the RH to the RTHAL. Using a cutoff value of 0.0, the sensitivity of correctly categorizing the right TLE less than the cutoff is 1.0, whereas the specificity is 0.875. Using a cutoff value of 0.170, the sensitivity and specificity is 1.0.

Figure 2.


The RH connectivity to RTHAL in each subject. The right TLE “sz free,” left TLE “sz free,” and controls were used in the ANOVA to identify the RTHAL region of interest. The other subjects are shown for comparison. The asterisks indicates the left TLE “sz free” group has significantly higher connectivity to the RTHAL region than the right TLE “sz free” group and the controls (unpaired t-test p < 0.0001). The number adjacent to the “sz recur” data indicates the subject ID in Table 1.

No statistics were performed on separate “sz recur” groups due to low numbers of patients. However, it is clear in Figure 2 that the range of connectivity values in the “sz recur” patients are much larger than in the “sz free” patients in both the right and left TLE groups. Furthermore, the connectivity values of the “sz recur” patients fall outside the range of the corresponding “sz free” patients in all but one case.

Discussion

This study identifies a resting-state functional network in patients with TLE that distinguishes between those patients with confirmed left hippocampal seizure onset and those with right hippocampal seizure onset with high sensitivity and specificity. The network involves the right hippocampus and a small region in the right thalamus. The primary advantage of this fMRI method is that it is measured noninvasively at rest, without the confounding factor of task performance. In addition, in this case, only 400 s of fMRI imaging was required, which may be practical to include in a clinical MRI examination.

The best sensitivity and specificity in categorizing the patients with right TLE and those with left TLE was found when using a cutoff of 0.17 for connectivity between RH and RTHAL. We found that the patients with right TLE have functional connectivity to the RH less than this, and that the patients with left TLE have connectivity that is greater (see Fig. 2). The controls had significantly lower values of RH to RTHAL connectivity than the left TLE group, but were similar to the patients with right TLE.

To verify that the differences in functional connectivity between the right hippocampus and the right thalamus were not due primarily to difference in hippocampal volume between the patients, we repeated the ANOVA analyses, including the hippocampal volume fraction as a covariate. The results of these analyses (one from right hippocampus and one from left hippocampus) revealed the same peak voxels of significant connectivity difference as in the analyses without the volume fraction covariate. This suggests that the network between the RTHAL and right hippocampus is different between groups, regardless of differences in hippocampal volume fractions. In addition, we validated our measures of hippocampal volume fraction by comparing the values to our expectations, that is, that the right TLE will have mean values <1.0, the left TLE will have mean values >1.0, and the controls will have values close to 1.0.

Prior studies have reported increased functional connectivity in spatially limited networks involving the hippocampus contralateral to the seizure focus and decreased functional connectivity ipsilateral to the seizure focus (Bettus et al., 2009, 2010). Our findings partially support these studies in that we detected increased functional connectivity contralateral to the seizure focus (high functional connectivity between right hippocampus and right thalamus in left TLE) and decreased or the same as controls in ipsilateral to the seizure focus (low functional connectivity between right hippocampus and right thalamus in right TLE). However, we were unable to identify a region in the left thalamus that would separate the two patient groups, even at reduced statistical levels, to fully investigate this hypothesis.

Our secondary aim was to investigate the same network in those patients with seizures recurring after surgery. We found that those patients had functional connectivity from the RH to RTHAL region that was outside the range defined by the seizure-free patients in six of seven cases. This may suggest that this measure might distinguish between seizure-free and seizure-recur subjects, however, the functional connectivity of the seizure-recur patients was both higher and lower than the seizure-free patients. These findings may indicate heterogeneous reasons for seizure recurrence.

Although not part of specific hypothesis testing, the RH and LH ANOVAs were performed including all right TLE (“sz free” and “sz recur”) versus all left TLE (“sz free” and “sz recur”) versus controls. As in the previous ANOVA with only “sz free” subjects, no regions were detected using the LH seed. Using the RH seed, three regions were detected at the p < 0.0001 (uncorrected) with cluster size 5 level, but none that distinguished left from right TLE (only TLE vs. control) including the region near the right precuneus found in the ANOVA of “sz free” patients only.

Role of thalamus in TLE

There is a large body of evidence supporting the role of the thalamus in TLE, specifically in the secondary generalization of seizures (Norden & Blumenfeld, 2002; Yu & Blumenfeld, 2009; Englot et al., 2010). Studies using intracerebral electrical recordings found synchronous electrical activity in the thalamus and temporal lobes during temporal lobe seizures (Guye et al., 2006; Arthuis et al., 2009), and that the degree of synchrony was related to loss of consciousness. Low coupling between these regions was also related to better surgical outcome (Guye et al., 2006). Using single photon emission computed tomography (SPECT) during secondary generalization in a combined group of left- and right-sided onset tonic–clonic seizures, increases in cerebral blood flow were detected primarily in the left thalamus, bilateral superior cerebellum, and left caudate (Blumenfeld et al., 2009). In a comparison between patients with left TLE and healthy controls, differences in fMRI functional connectivity between the left anterior hippocampus and right thalamus were reported (Morgan et al., 2010). Recent work in brain stimulation therapy for the treatment of refractory epilepsy has found that bilateral stimulation of the anterior thalamus can significantly reduce seizures, with most effectiveness in those patients with temporal lobe seizure onset (Fisher et al., 2010).

Structural MRI imaging has also confirmed the involvement of the thalamus in TLE seizures. Studies quantifying the changes in GM volume between TLE and healthy controls using voxel-based morphometry methods have found reductions in bilateral thalamus (Labate et al., 2008) and the thalamus ipsilateral to the seizure onset (McMillan et al., 2004; Mueller et al., 2010). Furthermore, volume loss in bilateral thalamic regions was correlated positively with the thickness of the entorhinal cortex in these patients (Mueller et al., 2010).

Negative functional connectivity in right TLE

The mechanism underlying the negative versus positive functional connectivity we detected across this network is not clear. Mathematically, an increased negative connectivity value represents an increased negative linear correlation between the two time series, which is generally caused by phase differences between them. This can arise physiologically through accumulated phase delay across long path lengths between two nodes in a network (Cabral et al., 2011; Chen et al., 2011). In neurophysiologic terms, negative connectivity could be the result of inhibition of an excitatory neuron across two or more synapses.

In functional connectivity mapping, negative functional connectivity can be the result of linearly regression of the global signal (average signal time series across the whole brain) from the seed region (Murphy et al., 2008; Weissenbacher et al., 2009). This regression shifts the set of connectivities across the brain toward a mean of zero. This can cause low connectivity values close to zero to be shifted to negative values. In this study, the global signal was not linearly regressed from the data; instead the average WM and CSF signals were used to approximate fluctuations due to physiologic noise (Weissenbacher et al., 2009). However, these signals may be similar to the global signal and may be partially responsible for the negative values.

Conclusions

This study identified a region in the ventral lateral nucleus of the right thalamus whose connectivity to the hippocampi separates left TLE “sz free” subjects from right TLE “sz free” subjects with high specificity and sensitivity. The results from a small group of patients where seizures recurred after surgery had connectivity values across this network that fell outside the range of those who were seizure-free. Therefore, the resting state functional MRI connectivity across this network may be a potential indicator of lateralization of TLE. If validated in a larger independent cohort, this measure may be added to other presurgical assessments of these patients.

Acknowledgments

This work was supported in part by NIH R01 NS055822 and UL1 RR024975-01.

Disclosure

None of the authors has any conflict of interest to disclose. We confirm that we have read the Journal’s position on issues involved in ethical publication and affirm that this report is consistent with those guidelines.

Ancillary