Amyloid - Tau Sex Differences in AD Cortical Region
Julie Wisch
February 27, 2019
I considered all the participants who had both PIB and TAU scans. It’s a well-matched cohort.
## Stratified by GENDER
## female male
## n 108 69
## Age (mean (sd)) 66.81 (7.64) 68.68 (8.62)
## EDUC (mean (sd)) 15.94 (2.54) 16.22 (2.33)
## apoe (%)
## 22 1 ( 0.9) 0 ( 0.0)
## 23 14 (13.1) 9 (13.0)
## 24 4 ( 3.7) 1 ( 1.4)
## 33 57 (53.3) 36 (52.2)
## 34 26 (24.3) 21 (30.4)
## 44 5 ( 4.7) 2 ( 2.9)
## CDR (%)
## 0 101 (93.5) 64 (92.8)
## 0.5 6 ( 5.6) 4 ( 5.8)
## 1 1 ( 0.9) 1 ( 1.4)
## PUP_fSUVR_rsf_TOT_CORTMEAN.x (mean (sd)) 1.38 (0.63) 1.37 (0.67)
## PUP_fSUVR_rsf_TOT_CORTMEAN.y (mean (sd)) 1.20 (0.14) 1.11 (0.12)
## Stratified by GENDER
## p test
## n
## Age (mean (sd)) 0.131
## EDUC (mean (sd)) 0.458
## apoe (%) 0.799
## 22
## 23
## 24
## 33
## 34
## 44
## CDR (%) 0.947
## 0
## 0.5
## 1
## PUP_fSUVR_rsf_TOT_CORTMEAN.x (mean (sd)) 0.897
## PUP_fSUVR_rsf_TOT_CORTMEAN.y (mean (sd)) <0.001First I looked at correlations between amyloid and tau levels. Significant correlations are shown in blue. You can see that there are a lot of correlations in white matter for females, which is sort of troubling…
Then I looked just at the regions outlined in Liang’s Cortical AD Signature. You can see that there’s a correlation between Tau and Amyloid for all the signature regions for women, but not for men.
Digging further in to these regions to look for interesting relationships, we see that there doesn’t really look like there’s much of a difference in the amount of tau in any of these regions. However, men seem to have lower levels of amyloid in the entorhinal cortex, medial temporal gyri, superior temporal gyri, inferior parietal, and precuneus.
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]Then I used the Youden Index to calculate the optimal biomarker cutoff to make a prediction. I tested regions for three things: regional amyloid’s ability to predict tau positivity (PUP_fSUVR_rsf_TOT_CORTMEAN > 1.22), regional amyloid’s ability to predict CDR (which is sort of hard given how few converters we have scanned), and regional tau’s ability to predict CDR (again…not many converters).
Across the board, amyloid AND tau levels in the cortical signature are more predictive of CDR for women. Weirdly, amyloid levels are more predictive of tau levels for men as compared to women.
So, to recap, speaking strictly about the AD signature:
1. Amyloid correlates with tau for women all the time and men most of the time. 2. Amyloid correlates more strongly with tau in men (r ~ 0.6) than for women (r ~ 0.4).
3. Women have more amyloid for a given level of tau than men in a lot of these regions.
4. Amyloid and tau are more predictive of CDR for women. 5. Amyloid is more predictive of tau levels for men.
We can also compare the optimal thresholds for men vs. women. There are particularly large discrepancies in useful cutpoints in the superior and inferior parietal areas, as well as the precuneus. Interestingly, for the CDR relationships, generally females have much higher cutpoints than men. For the relationship between amyloid and tau positivity, men have higher cutpoints than women.
I think the takeaway is…women can handle more crap in their brains before they start to show symptoms.
## Region AUC.f cutpoint.f AUC.m cutpoint.m
## 13 CTX_ENTORHINAL 0.5254597 1.377 0.4062500 1.510
## 15 CTX_FUSIFORM 0.7715700 1.532 0.6468750 1.747
## 17 CTX_INFERTMP 0.7991513 1.702 0.6500000 1.392
## 24 CTX_MIDTMP 0.8217822 1.704 0.5750000 1.256
## 39 CTX_SUPERTMP 0.7751061 1.468 0.5468750 1.128
## 16 CTX_INFERPRTL 0.7503536 2.042 0.6828125 1.084
## 38 CTX_SUPERPRTL 0.7991513 1.805 0.7375000 1.263
## 32 CTX_POSTCNG 0.8677511 3.128 0.7218750 1.210
## 34 CTX_PRECUNEUS 0.8161245 2.681 0.6500000 1.212
## Region AUC.f cutpoint.f AUC.m cutpoint.m
## 13 CTX_ENTORHINAL 0.7171146 1.413 0.6281250 1.246
## 15 CTX_FUSIFORM 0.8281471 1.511 0.7078125 1.340
## 17 CTX_INFERTMP 0.8373409 1.506 0.7484375 1.340
## 24 CTX_MIDTMP 0.7942008 1.382 0.6203125 1.351
## 39 CTX_SUPERTMP 0.6188119 1.153 0.4828125 1.152
## 16 CTX_INFERPRTL 0.7277228 1.789 0.6625000 1.297
## 38 CTX_SUPERPRTL 0.7942008 1.209 0.6671875 1.092
## 32 CTX_POSTCNG 0.8182461 1.330 0.4718750 1.443
## 34 CTX_PRECUNEUS 0.8288543 1.391 0.5390625 1.135
## Region AUC.f cutpoint.f AUC.m cutpoint.m
## 13 CTX_ENTORHINAL 0.5883590 1.354 0.7470760 1.262
## 15 CTX_FUSIFORM 0.6018583 1.195 0.7149123 1.238
## 17 CTX_INFERTMP 0.5825736 1.165 0.7953216 1.204
## 24 CTX_MIDTMP 0.6441094 1.030 0.7361111 1.111
## 39 CTX_SUPERTMP 0.6484923 0.985 0.7353801 0.970
## 16 CTX_INFERPRTL 0.6872370 1.042 0.7032164 1.964
## 38 CTX_SUPERPRTL 0.6523492 1.084 0.6593567 1.458
## 32 CTX_POSTCNG 0.6085203 1.331 0.6798246 1.385
## 34 CTX_PRECUNEUS 0.6185133 2.602 0.7295322 1.285
Longitudinal Analysis
We have 95 individuals with both one tau and multiple amyloid scans. However, we only have 4 CDR > 0’s who meet that criteria. So this analysis will look only at cognitively normal individuals. Here are the demographics for this analysis
## Stratified by GENDER
## female male
## n 62 29
## Age (mean (sd)) 67.81 (7.76) 70.95 (6.51)
## TauPos = 1 (%) 25 (40.3) 5 ( 17.2)
## Converter = 1 (%) 0 ( 0.0) 0 ( 0.0)
## GENDER = male (%) 0 ( 0.0) 29 (100.0)
## EDUC (mean (sd)) 16.03 (2.33) 16.76 (2.31)
## apoe (%)
## 22 1 ( 1.6) 0 ( 0.0)
## 23 6 ( 9.8) 3 ( 10.3)
## 33 35 (57.4) 17 ( 58.6)
## 34 17 (27.9) 8 ( 27.6)
## 44 2 ( 3.3) 1 ( 3.4)
## PUP_fSUVR_rsf_TOT_CORTMEAN.y (mean (sd)) 1.18 (0.15) 1.14 (0.10)
## TOT_CORTMEAN (mean (sd)) 0.00 (0.00) 0.00 (0.00)
## Stratified by GENDER
## p test
## n
## Age (mean (sd)) 0.062
## TauPos = 1 (%) 0.052
## Converter = 1 (%) NaN
## GENDER = male (%) <0.001
## EDUC (mean (sd)) 0.168
## apoe (%) 0.975
## 22
## 23
## 33
## 34
## 44
## PUP_fSUVR_rsf_TOT_CORTMEAN.y (mean (sd)) 0.136
## TOT_CORTMEAN (mean (sd)) 0.769Unfortunately we have a lot more women than men for this, but it’s at least a place to start.
So here’s a comparison of the correlations between amyloid rate of change and tau. You can see that there are absolutely no significant relationships between them, for either sex, once you correct for multiple comparisons. Before you correct, there are a few.
And here’s a look at the AD signature regions. There are no significant relationships between tau and amyloid rate of change; however, there are some trends that point to opposite slope directions for men and women in the inferior temporal area.
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]
## TableGrob (2 x 2) "arrange": 4 grobs
## z cells name grob
## 1 1 (1-1,1-1) arrange gtable[layout]
## 2 2 (1-1,2-2) arrange gtable[layout]
## 3 3 (2-2,1-1) arrange gtable[layout]
## 4 4 (2-2,2-2) arrange gtable[layout]You can see that rate of change of amyloid is fairly predictive of tau positivity for men. It’s not useful at all for women. 2 observations with this: 1. Only 5/29 men were tau positive in this cohort. 25/62 women were tau positive 2. Negative amyloid rates of change are connected with tau positivity for women, positive amyloid rates of change are connected with tau positivity for men.
## Region AUC.f cutpoint.f AUC.m cutpoint.m
## 13 TOT_CTX_ENTORHINAL 0.3902703 -2.880306e-04 0.7750000 5.554086e-05
## 15 TOT_CTX_FUSIFORM 0.3275676 -4.147945e-05 0.6750000 5.510968e-05
## 17 TOT_CTX_INFERTMP 0.3935135 -2.876609e-05 0.6750000 1.080875e-04
## 24 TOT_CTX_MIDTMP 0.3772973 -9.272849e-05 0.7000000 7.223114e-05
## 39 TOT_CTX_SUPERTMP 0.3535135 4.953765e-04 0.6083333 6.688068e-05
## 16 TOT_CTX_INFERPRTL 0.3783784 5.019881e-04 0.5416667 3.120567e-04
## 38 TOT_CTX_SUPERPRTL 0.3989189 4.715984e-04 0.5250000 2.123400e-04
## 32 TOT_CTX_POSTCNG 0.3362162 -2.110341e-04 0.7250000 6.581059e-05
## 34 TOT_CTX_PRECUNEUS 0.3762162 9.960370e-04 0.6250000 2.832080e-04
