Supplemental Materials
Asynchrony in Executive Networks Predicts Cognitive Slowing in Multiple Sclerosis
by N. Hubbard et al., 2015, Neuropsychology
A. Head Motion Table
Group Head Motion Characteristics
MS PatientsHC p-value
Δx 0.57 (0.43)1.04 (2.31).30
Δy 1.04 (0.76)1.19 (1.04).53
Δz 0.68 (0.79)0.48 (0.36).27
ΔYaw0.83 (1.03)0.75 (0.54).74
ΔPitch 0.34 (0.29)0.40 (0.34).50
ΔRoll 1.04 (1.13)1.51 (1.17).24
Framewise645.36 (396.02)711.470 (460.35).58
Displacement
Note. Mean (SD) group head motion characteristics across all three runs of the DSST. Δx, Δ y, Δz in mm. ΔYaw, ΔPitch, and ΔRoll in degrees. Framewise displacement averaged across three runs in mm (see Power et al., 2011). p-value based on independent samples t-test.
B. Cognitive Impairment Table
Group Cognitive Impairment
MS PatientsHC p-value
COWAT 0.00 %0.00 % n/a
PASAT-3 11.11 %0.00 %.100
PASAT-20.00 %0.00 % n/a
SRT 15.38 %4.35 %.203
SPART 18.52 %4.35 % .124
TMT-A 3.57 %0.00 %.371
TMT-B17.86 %0.00 %.046*
Note. COWA = Controlled Oral Word Association; PASAT = Paced Serial Addition Task; SRT = Selective Reminding Test long term retention component; SPART = 10/36 Spatial Recall Test; TMT = Trail Making Test.p-values attained from Pearson χ2-test with 1 DOF. * p < .05.
MS patients and HCs did not differ on impairment prevalence (with the exception of Trails B). Thus, these variables could not have mediated group differences in our connectivity measures (Baron and Kenney, 1986). However, because MS patients and HCs differed on Trails B, it is possible that impairment status on Trails B could have mediated group differences in our connectivity measures. Thus, we examined whether impairment status on Trails B resulted in differing group-disparate network z-connectivity with left and right BA 9. Impaired vs. non- impaired participants on Trails B did not significantly differ in their group-disparate network z-connectivity with left and right BA 9 (all ps > .05). Further, impaired MS patients on Trails B did not significantly differ from non-impaired MS patients on left or right BA 9 group-disparate z-connectivity (all ps > .05).
C. DLPFC BOLD Activity Methods
Preprocessing procedures used on these data were similar to those described in the main text, and are described at length elsewhere (Hubbard et al., 2014). One MS patient was not included in these analyses due to structural and functional alignment irregularities, not present in the preprocessing of the data described in the main text.
Task versus baseline (measured during inter-trial interval) trials were modeled for the functional images using general linear modeling. One potential problem with examining BOLD activation differences in MS patients compared to HCs is possible violation of the equivalence of the hemodynamic response function (HRF) assumption (see Iannetti and Wise, 2007; Rypma and D’Esposito, 2001). Briefly, this assumption holds that group hemodynamic responses must be equivalent in order for differences in the magnitude of BOLD activity to meaningfully index group differences in the magnitude of neural activity. Because MS is associated with vascular reactivity deficits (e.g., Marshall et al., 2014), it is plausible that the shape of the HRF will not be equivalent between our groups. Slower reaction times (i.e., longer processing times) on our functional task could also lead to MS-related changes to the time-to-peak of the HRF, resulting in a biased estimator of BOLD activity for MSPs compared to HCs (e.g., Menon and Kim, 1999). This precludes the use of standard, canonical HRFs (e.g., gamma variate) which characterize the magnitude of BOLD activity based upon a specified time-to-peak of the HRF (e.g., gamma peak = 4.7 seconds; Ward, 1998, 2006).We therefore employed a model-free approach to convolve BOLD estimates. These data were convolved using piecewise linear B-spline functions in AFNI using the tent function (cf. Motes and Rypma, 2010; Ward, 1998, 2006). This method uses a finite number of basis functions to allow for free-formed estimation of subject-specific BOLD-HRFs, without assumptions regarding the shape of individuals’ impulse responses. The tent function was calculated from baseline during a window of time centered on stimulus onset. The current method modeled each participant’s HRF using BOLD parameter estimates at eight time points, spaced equally at intervals of two seconds (1 TR). Parameter estimates represented relative signal amplitude beginning at stimulus onset (t0) and extending 16 seconds (t7) past the initial event (e.g., Dale and Buckner, 1997). Data were then converted to percent signal change from baseline. Piecewise B-spline functions were fit across each percent signal change time points. This procedure maximized the function’s fit to the data (all participants, left and right BA 9 R2 =1) and resulted in smooth curves approximating each individual’s HRF, within the respective ROIs. BOLD activity was taken from the peak amplitude of the each participants’ BOLD-HRF (Supplemental Figure).
D. Clinical Associations with Group-disparate Connectivity
We assessed whether patient characteristics (i.e., Disease Duration, EDSS, Time since Last Exacerbation, or Immunomodulatory Therapy Status; see Table 1) might have affected RRMS group-disparate connectivity with BA 9. Neither Disease Duration (left BA 9 r = .17; right BA 9 r = .07), EDSS (left BA 9 r = .02; right BA 9 r = -.05), Time since Last Exacerbation (left BA 9 r = .19; right BA 9 r = .10), nor Immunomodulatory Therapy Status (left BA9 τb= .26; right BA 9 τb= .18) significantly predicted left or right BA 9 group-disparate connectivity (all ps > .05).
Supplemental Figure. BOLD-HRFs averaged across left and right BA 9 for healthy controls (blue) and MS patients (red).
Supplemental References
Dale AM and Buckner RL. Selective averaging of rapidly presented individual trials using fMRI. Hum Brain Mapp 1997; 5: 329-40.
Hubbard NA, Hutchison J L, Motes MA, Shokri-Kojori E, Bennett IJ, Brigante, RM et al. Central executive dysfunction and deferred prefrontal processing in veterans with Gulf War Illness. Clinical Psychological Science 2014; 2(3): 319-327.
Iannetti GD and Wise RG. BOLD funcitonal MRI in disease and pharmacological studies: Room for improvement? Magnetic Resonance Imaging 2007; 25: 978-88.
Marshall O, Lu H, Brisset J-C, Xu F, Liu P, Herbert J, Grossman R I, and Ge Y. Impaired cerebrovascular reactivity in multiple sclerosis. JAMA Neurol. 2014; 71(10): 1275-1281.
Menon RS and Kim SG. Spatial and temporal limits in cognitive neuroiaging with fMRI. Trends in Cognitive Sciences 1999; 3(6): 207-15.
Motes MA and Rypma B. Working memory component processes: Isolating BOLD signal changes. NeuroImage 2010; 49(2): 1933-41.
Rypma B and D’Esposito M. Age-related changes in brain-behavior relationships: Evidence from event-related functional MRI studies. Euro J of Cog Psychology 2001; 13 (1/2): 235-56.
Ward BD. Deconvolution analysis of fMRI time series data [software manual]. 1998/2006 Retrieved from