HIV and Resting State
Julie Wisch
February 19, 2019
I calculated the intra-network and inter-network signatures in order to look for differences between HIV+ well-controlled, HIV + uncontrolled, and HIV- individuals. Here are their compositions:
These cohorts aren’t particularly well matched, so we will control for differences in age, sex, and education throughout the analysis.
## Stratified by Group
## Control UnControlled
## n 224 117
## Group (%)
## Control 224 (100.0) 0 ( 0.0)
## UnControlled 0 ( 0.0) 117 (100.0)
## WellControlled 0 ( 0.0) 0 ( 0.0)
## participant_age (mean (sd)) 37.83 (17.15) 39.80 (16.76)
## gender_check = 1 (%) 115 ( 51.3) 23 ( 19.7)
## blood_pressure (mean (sd)) 119.07 (15.49) 122.66 (15.42)
## race (%)
## 1 2 ( 0.9) 1 ( 0.9)
## 2 0 ( 0.0) 1 ( 0.9)
## 4 4 ( 1.8) 1 ( 0.9)
## 7 116 ( 51.8) 81 ( 69.2)
## 8 102 ( 45.5) 33 ( 28.2)
## 9 0 ( 0.0) 0 ( 0.0)
## education (mean (sd)) 460.20 (6680.54) 13.09 (2.20)
## bmi (mean (sd)) 26.87 (5.82) 25.95 (6.30)
## duration_of_infection (mean (sd)) 0.00 (0.00) 87.83 (99.69)
## recent_cd4 (mean (sd)) NaN (NA) 408.12 (305.48)
## nadir_cd4 (mean (sd)) NaN (NA) 245.58 (204.90)
## recent_viral_load (mean (sd)) NaN (NA) 1461.11 (2384.83)
## recent_cd8 (mean (sd)) NaN (NA) 872.00 (417.40)
## Hepatitis_A = 1 (%) 1 ( 0.4) 4 ( 3.4)
## Hepatitis_B = 1 (%) 0 ( 0.0) 1 ( 0.9)
## Hepatitis_C = 1 (%) 0 ( 0.0) 5 ( 4.3)
## Stratified by Group
## WellControlled p test
## n 310
## Group (%) <0.001
## Control 0 ( 0.0)
## UnControlled 0 ( 0.0)
## WellControlled 310 (100.0)
## participant_age (mean (sd)) 48.03 (14.33) <0.001
## gender_check = 1 (%) 85 ( 27.4) <0.001
## blood_pressure (mean (sd)) 123.19 (13.57) 0.054
## race (%) 0.005
## 1 4 ( 1.3)
## 2 1 ( 0.3)
## 4 0 ( 0.0)
## 7 209 ( 67.4)
## 8 95 ( 30.6)
## 9 1 ( 0.3)
## education (mean (sd)) 336.86 (5687.99) 0.777
## bmi (mean (sd)) 26.74 (5.81) 0.367
## duration_of_infection (mean (sd)) 162.56 (108.65) <0.001
## recent_cd4 (mean (sd)) 644.98 (297.76) <0.001
## nadir_cd4 (mean (sd)) 230.31 (206.76) 0.544
## recent_viral_load (mean (sd)) 21.87 (5.97) <0.001
## recent_cd8 (mean (sd)) 852.74 (390.70) 0.876
## Hepatitis_A = 1 (%) 25 ( 8.1) <0.001
## Hepatitis_B = 1 (%) 38 ( 12.3) <0.001
## Hepatitis_C = 1 (%) 23 ( 7.4) <0.001Analysis
For all subsequent analysis, the figures show the uncorrected p value where we do not control for age/sex/education. The blocks of text show the anova and subsequent Tukey’s HSD level test to look for differences in group. These values are corrected for differences in age/sex/educaion, as well as for the fact that we make 3 comparisons.
There is a significant difference between Well-controlled and uncontrolled individuals in Intra-network signature (but weirdly not between our healthy controls and either well- or poorly controlled individuals).
## Df Sum Sq Mean Sq F value Pr(>F)
## Group 2 30.8 15.4 6.000 0.00262 **
## gender_check 1 99.5 99.5 38.804 8.46e-10 ***
## participant_age 1 570.5 570.5 222.524 < 2e-16 ***
## education 1 0.2 0.2 0.074 0.78597
## Residuals 644 1651.0 2.6
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Intra_Signature ~ Group + gender_check + participant_age + education, data = df.now)
##
## $Group
## diff lwr upr p adj
## UnControlled-Control -0.3209899 -0.75003324 0.1080534 0.1849092
## WellControlled-Control 0.2652537 -0.06481264 0.5953200 0.1429920
## WellControlled-UnControlled 0.5862436 0.17794916 0.9945381 0.0022708
There is also a significant difference between well controlled and poorly controlled individuals in the global inter-network signature; however, there is no difference between HIV+ individuals of any categorization and healthy controls.
## Df Sum Sq Mean Sq F value Pr(>F)
## Group 2 84 41.8 3.282 0.03818 *
## gender_check 1 121 120.8 9.495 0.00215 **
## participant_age 1 348 348.2 27.369 2.28e-07 ***
## education 1 2 1.6 0.125 0.72395
## Residuals 644 8194 12.7
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = Inter_Signature ~ Group + gender_check + participant_age + education, data = df.now)
##
## $Group
## diff lwr upr p adj
## UnControlled-Control -0.8046994 -1.76050333 0.1511045 0.1185301
## WellControlled-Control 0.1801879 -0.55511950 0.9154952 0.8331662
## WellControlled-UnControlled 0.9848873 0.07530669 1.8944679 0.0300885
There’s no difference in DMN x DMN
## Df Sum Sq Mean Sq F value Pr(>F)
## Group 2 1.7 0.83 0.906 0.40446
## gender_check 1 7.1 7.13 7.780 0.00544 **
## participant_age 1 47.8 47.80 52.156 1.45e-12 ***
## education 1 0.5 0.52 0.566 0.45218
## Residuals 644 590.3 0.92
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = DMN_x_DMN ~ Group + gender_check + participant_age + education, data = df.now)
##
## $Group
## diff lwr upr p adj
## UnControlled-Control 0.08581745 -0.1707193 0.3423542 0.7118768
## WellControlled-Control -0.05276768 -0.2501234 0.1445881 0.8047296
## WellControlled-UnControlled -0.13858513 -0.3827156 0.1055454 0.3770593There’s no difference in DANxDAN.
## Df Sum Sq Mean Sq F value Pr(>F)
## Group 2 2.2 1.087 1.139 0.32073
## gender_check 1 8.3 8.329 8.726 0.00325 **
## participant_age 1 27.9 27.931 29.262 8.92e-08 ***
## education 1 0.4 0.437 0.458 0.49874
## Residuals 644 614.7 0.955
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = DAN_x_DAN ~ Group + gender_check + participant_age + education, data = df.now)
##
## $Group
## diff lwr upr p adj
## UnControlled-Control 0.144573010 -0.1172233 0.40636934 0.3972096
## WellControlled-Control -0.009605131 -0.2110071 0.19179681 0.9931041
## WellControlled-UnControlled -0.154178141 -0.4033138 0.09495754 0.3140955There’s no significant difference in SM x SM, although there is a trend to suggest that uncontrolled individuals have worse connectivity than either healthy controls or well-controlled individuals.
## Df Sum Sq Mean Sq F value Pr(>F)
## Group 2 5.3 2.63 2.777 0.0630 .
## gender_check 1 33.7 33.69 35.588 4.02e-09 ***
## participant_age 1 3.9 3.94 4.157 0.0419 *
## education 1 0.0 0.00 0.000 0.9849
## Residuals 644 609.7 0.95
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = SM_x_SM ~ Group + gender_check + participant_age + education, data = df.now)
##
## $Group
## diff lwr upr p adj
## UnControlled-Control 0.24617359 -0.01455534 0.50690252 0.0689178
## WellControlled-Control 0.02362942 -0.17695136 0.22421021 0.9586645
## WellControlled-UnControlled -0.22254416 -0.47066407 0.02557574 0.0891955There’s no difference in SM lat x SM lat.
## Df Sum Sq Mean Sq F value Pr(>F)
## Group 2 3.5 1.73 2.017 0.133845
## gender_check 1 10.9 10.91 12.724 0.000388 ***
## participant_age 1 74.7 74.70 87.160 < 2e-16 ***
## education 1 0.0 0.00 0.001 0.975038
## Residuals 644 552.0 0.86
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = SM_lat_x_SM_lat ~ Group + gender_check + participant_age + education, data = df.now)
##
## $Group
## diff lwr upr p adj
## UnControlled-Control 0.01528718 -0.2327851 0.26335945 0.9885133
## WellControlled-Control -0.14043002 -0.3312739 0.05041389 0.1952965
## WellControlled-UnControlled -0.15571720 -0.3917925 0.08035812 0.2685975