Consequences of a polygenetic component of BP in an urban based longitudinal study: the
Malmö Diet and Cancer.
Cristiano Fava*a,b, MD, PhD; Therese Ohlsson* a; Marketa Sjogrena, PhD; Angela Tagettib, MD;
Peter Almgrena, BS; Gunnar Engströma, MD, PhD; Peter Nilssona, MD; Bo Hedblada, MD; Pietro
Minuzb, MD; Olle Melandera, MD, PhD
a
Department of Clinical Sciences, Lund University, University Hospital of Malmö, Sweden.
Department of Medicine, University Hospital of Verona, Italy.
b
Online supplements
Methods and results
METHODS
Genotyping
The SNPs were genotyped using IPLEX on a MassARRAY platform (Sequenom, San Diego,
CA, USA) according to the manufacturer’s standard protocols. Nearly 30% of the samples were
run in duplicate. All genotypes were called by two different investigators. We pre-specified a
threshold call rate of 90% per individual SNP (that is SNPs would be excluded if its call rate
is<90%). A threshold of p<10-07 was established for excluding SNPs, according to HardyWeinberg equilibrium calculation.
Equation used for the calculation of the count and weighted GRS
Genetic risk score
To create the multivariable GRS we used the weighted method (weighted GRS) according to the
beta value attributed to the tested SNPs in previous studies[1-3] assuming each SNP to be
independently associated with risk[4] according to an additive genetic model. Thus, weightings
of 0, 1, and 2 were given according to the number of coded alleles present, which in our samples
corresponds to the risk alleles. Successively, the number of corresponding coded alleles (0, 1, or
2) was multiplied for the absolute value of the β-coefficient for systolic BP, as detected in
previous studies,[1-3] and then these products were summed up. Thus, different SNPs contribute
different weights, as opposed to an alternate approach in which no weighting of effects is used,
and each SNP allele counts equally in the score. Successively, the score was divided by the
number of effectively genotyped SNP to produce an average measure, which takes into account
the amount of effectively genotyped SNPs, and the ratio was standardized. The GRS was
modeled as a continuous variable (increase in 1 Unit means an increase in 1 SD of the BP-GRS
and as tertiles. Details about the equation utilized to calculate the GRS are presented below (see
also appendix figure 1 for the distribution of GRS in the population). Only subjects with at least
24 valid genotypes were included in the final analysis (see also appendix table 3).
1
Weighted GRS (beta coefficient as used in previous studies):
+0.903 x MTHFR_rs17367504[1] + 0.388 x MOV10_rs2932538[1] + 0.067 x
ULK4_rs3774372[1] + 0.315 x SLC4A7_rs13082711[1] + 0.409 x MDS1_rs419076[1] + 0.740 x
FGF5_rs16998073[3] + 0.981 x SLC39A8_rs13107325[1] + 0.321 x
GUCY1A3_rs13139571[1]+ 0.504 x NPR3_rs1173771[1] + 0.412 x EBF1_rs11953630[1] +
0.627 x HFE_rs1799945[1] + 0.376 x BAT3_rs805303[1] + 0.373 x CACNB2_rs4373814[1] +
0.43 x C10orf107_rs1530440[3] + 0.484 x PLCE1_rs932764[1] + 1.095 x
CYP17A1_rs11191548[1] + 0.619 x ADM_rs7129220[1] + 0.840 x PLEKHA7_rs381815[2] +
0.565 x TMEM133_rs633185[1] + 1.26 x ATP2B1_rs2681492[2] + 0.598 x
SH2B3_rs3184504[1] + 0.354 x TBX3_rs10850411[1] + 0.612 x CYP1A1_rs1378942[1] +
0.650 x FES_rs2521501[1] + 0.680 x PLCD3_rs12946454[3] + 0.556 x GOSR2_rs17608766
[1]+ 0.410 x ZNF652_rs16948048[3] + 0.340 x JAG1_rs1327235[1] + 0.896 x
GNAS_EDN3_rs6015450[1]
The reference assigned to the SNP indicate from which article the beta coefficient was derived.
All the beta coefficient are positive because the coded allele was always the risk allele.
Standardization process
After computation of the weighted sum, the score was divided by the number of effectively
genotyped SNP to produce an average measure, which takes into account the amount of
effectively genotyped SNPs. This ratio was the standardized by subtracting the population mean
from an individual raw score and then dividing the difference by the population standard
deviation.
Net Reclassification Improvement (NRI) and Integrated Discrimination Index (IDI)
The net reclassification improvement (NRI) was computed using pold and pnew where pold is the
predicted probability using the beta coefficients as computed in the logistic regression model
without the BPGRS and pnes is the predicted probability using the beta coefficients as computed
in the logistic regression model adding the BPGRS. In the NRI subjects are categorized as up and
down according to the differences in the predicted probabilities with the 2 models and calculated
as:
NRI=P(up|event) - P(down|event) + P(down|non-event) - P(up|non-event)
In the “classical NRI”, the Framingham categories of low (<6%) moderate (6-20%) and high
cardiovascular risk (>20%) were used to compute upgrading or downgrading of the subjects. In
the category less NRI(>0) no category was preset and even small changes in the predicted
probabilities implicated an up- or down grading of each individual.
𝑃(event|up)·𝑛𝑈 − 𝑃(event|down)·𝑛𝐷
(1−𝑃(event|down)·𝑛𝐷 −(1−𝑃(event|up))·𝑛𝑈
NRI=
+
𝑛·𝑃(event)
𝑛·(1−𝑃(event))
𝑃(event|up)𝑃(up) − 𝑃(event|down)𝑃(down)
event(NRI)=P(up|event) - P(down|event)=
𝑃(event)
2
non-event(NRI)=P(down|non-event) - P(up|non-event)=
=
𝑃(non
− event|down)𝑃(𝑑𝑜𝑤𝑛) − 𝑃(non−event|up)𝑃(up)
𝑃(non−event)
The IDI was estimated as follows:
̂ =( ̅̅̅
𝐼𝐷𝐼
𝑝̂ new,events - ̅̅̅
𝑝̂ old,events ) - (̅̅̅
𝑝̂ new,nonevents - ̅̅̅
𝑝̂ old,nonevents)
̅̅̅
where 𝑝̂ new,events is the mean of the new model-based predicted probabilities of an event for those
who develop events, ̅̅̅
𝑝̂ old,events is the corresponding quantity based on the old model, ̅̅̅
𝑝̂ new,nonevents
is the mean of the new model-based predicted probabilities of an event for those who do not
develop events and ̅̅̅
𝑝̂ old,nonevents is the corresponding quantity based on the old model.[5]
3
Supplemental table S1. Hardy-Weinberg equilibrium for all the SNPs in the MDC.
Gene
MTHFR-NPPB
MOV10
SLC4A7
ULK4
MECOM
FGF5
SLC39A8
GUCY1A3-GUCY1B3
NPR3-C5orf23
EBF1
HFE
BAT2-BAT5
CACNB2(5’)
c10orf107
PLCE1
CYP17A1-NT5C2
ADM
PLEKHA7
FLJ32810-TMEM133
ATP2B1
SH2B3
TBX5-TBX3
CYP1A2-ULK3
FES
PLCD3
GOSR2
ZNF652
JAG1
GNAS-EDN3
SNP
Chr.
Position
rs17367504
rs2932538
rs13082711
rs3774372
rs419076
rs16998073
rs13107325
rs13139571
rs1173771
rs11953630
rs1799945
rs805303
rs4373814
rs1530440
rs932764
rs11191548
rs7129220
rs381815
rs633185
rs2681492
rs3184504
rs10850411
rs1378942
rs2521501
rs12946454
rs17608766
rs16948048
rs1327235
rs6015450
1
1
3
3
3
4
4
4
5
5
6
6
10
10
10
10
11
11
11
12
12
12
15
15
17
17
17
20
20
11,862,778
113,216,543
27,537,909
41,877,414
169,100,886
81,184,341
103,188,709
156,645,513
32,815,028
157,845,402
26,091,179
31,616,366
18,419,972
63,524,591
95,895,940
104,846,178
10,350,538
16,902,268
100,593,538
90,013,089
111,884,608
115,387,796
75,077,367
91,437,388
43,208,121
45,013,271
47,440,466
10,969,030
57,751,117
% of valid
genotypes
99.6
96.6
96.6
97.4
90.8
98.8
97.6
96.2
94.9
95.6
92.5
96.0
94.9
98.9
96.9
89.7
97.0
96.4
95.9
97.3
97.0
96.4
92.1
95.6
99.1
97.4
98.3
95.1
92.5
C.A.
A.A.
MAF
A
G
C
C
T
T
C
C
G
C
G
G
C
C
G
T
A
T
C
T
T
T
C
T
T
C
G
G
G
G
A
T
T
C
A
T
A
A
T
C
A
G
T
A
C
G
C
G
C
C
C
A
A
A
T
A
A
A
0.152
0.264
0.216
0.152
0.454
0.343
0.048
0.22
0.404
0.344
0.115
0.37
0.428
0.188
0.446
0.108
0.104
0.274
0.297
0.145
0.48
0.289
0.321
0.327
0.246
0.14
0.383
0.494
0.132
Observed
heterozygosity
0.256
0.387
0.334
0.257
0.492
0.452
0.091
0.35
0.499
0.456
0.201
0.476
0.499
0.305
0.490
0.191
0.189
0.392
0.412
0.249
0.495
0.413
0.436
0.462
0.373
0.243
0.471
0.508
0.229
Expected
heterozygosity
0.258
0.386
0.338
0.258
0.496
0.451
0.091
0.343
0.481
0.451
0.203
0.466
0.490
0.305
0.494
0.193
0.186
0.398
0.417
0.248
0.499
0.411
0.435
0.439
0.371
0.241
0.473
0.500
0.229
HWE
p-value
0.3427
0.8024
0.0450
0.7288
0.2827
0.6815
1
0.0082
2.0 E-06
0.1555
0.2423
0.0057
7.7 E-03
0.9709
0.2687
0.1239
0.0456
0.0645
0.1296
0.1902
0.579
0.6181
0.7069
7.1 E-07
0.5117
0.2908
0.6711
0.0418
0.7999
Kapp
a*
0.976
0.966
0.974
0.974
0.971
0.974
0.970
0.967
0.954
0.966
0.924
0.977
0.955
0.967
0.971
0.971
0,976
0.970
0.969
0.971
0.961
0,961
0.976
0,968
0.969
0.966
0.970
0,963
0.976
SNP, Single Nucleotide polymorphism; Chr., chromosome; C.A., coded allele; A.A. alternative allele; MAF, Minor Allele Frequency;
HWE, Hardy Weinberg equilibrium. *Kappa indicates the agreement between genotypes as calculated on more than 6,000 samples run
in duplicate.
4
Supplemental table S2. Number of missing genotypes per subject at MDC.
Number of missing genotypes
Number of subjects
No missing genotypes
20,355
1 missing genotype
3,701
2 missing genotypes
1,108
3 missing genotypes
661
4 missing genotypes
302
5 missing genotypes
876
6 or more missing genotypes
3,379
Subjects with 6 or more missing genotypes were excluded from the analysis.
5
Supplemental Table S3. Summary association statistics about CAD, stroke and cardiovascular events based on all data for 29
independent SNPs in the MDC after full adjustment.
CAD
Gene
Index SNP
MTHFR-NPPB
rs17367504
MOV10
rs2932538
SLC4A7
rs13082711
ULK4
rs3774372
MECOM
rs419076
FGF5
rs16998073
SLC39A8
rs13107325
GUCY1A3GUCY1B3
NPR3-C5orf23
rs13139571
EBF1
rs11953630
HFE
rs1799945
BAT2-BAT5
rs805303
CACNB2(5')
rs4373814
C10ORF107
rs1530440
PLCE1
rs932764
CYP17A1-NT5C2
rs11191548
ADM
rs7129220
rs1173771
Beta
(95% CI)
1.033
(0.953-1.121)
1.022
(0.957-1.092)
1.048
(0.978-1.123)
1.018
(0.939-1.104)
0.998
(0.942-1.057)
1.061
(0.998-1.127)
1.063
(0.932-1.211)
1.057
(0.986-1.134)
0.977
(0.921-1.036)
1.009
(0.950-1.072)
0.980
(0.896-1.072)
0.968
(0.912-1.028)
1.026
(0.968-1.087)
1.025
(0.951-1.105)
0.981
(0.925-1.040)
1.028
(0.935-1.130)
1.064
stroke
p-value
0.426
0.508
0.186
0.662
0.936
0.057
0.362
0.119
0.433
0.777
0.662
0.286
0.390
0.519
0.517
0.566
0.193
Beta
(95% CI)
0.985
(0.894-1.085)
1.078
(0.994-1.169)
1.048
(0.964-1.139)
1.013
(0.919-1.117)
0.984
(0.918-1.056)
1.029
(0.955-1.108)
1.087
(0.926-1.277)
1.057
(0.971-1.150)
1.120
(1.042-1.204)
1.016
(0.944-1.093)
1.066
(0.959-1.184)
1.033
(0.961-1.111)
1.019
(0.950-1.093)
1.041
(0.949-1.141)
1.036
(0.965-1.111)
0.896
(0.803-1.000)
1.110
CV disease
p-value
0.758
0.069
0.273
0.795
0.656
0.454
0.306
0.204
0.002*
0.674
0.237
0.378
0.596
0.395
0.330
0.049
0.065
Beta
(95% CI)
1.016
(0.951-1.085)
1.053
(0.998-1.112)
1.058
(1.001-1.120)
1.035
(0.969-1.106)
0.998
(0.953-1.046)
1.054
(1.054-1.107)
1.032
(0.925-1.151)
1.056
(0.997-1.117)
1.040
(0.991-1.091)
1.003
(0.955-1.054)
1.017
(0.946-1.093)
1.000
(0.952-1.050)
1.029
(0.981-1.078)
1.033
(0.971-1.098)
1.017
(0.970-1.067)
0.985
(0.913-1.063)
1.072
p-value
0.633
0.060
0.048*
0.304
0.943
0.039*
0.575
0.062
0.114
0.896
0.645
0.993
0.241
0.308
0.475
0.699
0.072
6
PLEKHA7
rs381815
FLJ32810TMEM133
ATP2B1
rs633185
SH2B3
rs653178
TBX5-TBX3
rs10850411
CYP1A2-ULK3
rs1378942
FES
rs2521501
PLCD3
rs12946454
GOSR2
rs17608766
ZNF652
rs16948048
JAG1
Dovr +
GNAS-EDN3
rs1327235
rs2681492
rs6015450
(0.969-1.167)
1.000
(0.937-1.067)
0.960
(0.901-1.023)
1.084
(1.000-1.176)
1.076
(1.016-1.140)
0.957
(0.899-1.018)
1.013
(0.951-1.079)
1.079
(0.015-1.147)
1.042
(0.975-1.115)
0.992
(0.913-1.079)
0.969
(0.912-1.029)
0.981
(0.925-1.040)
0.992
(0.910-1.080)
0.991
0.207
0.050*
0.012*
0.161
0.690
0.015*
0.227
0.853
0.307
0.519
0.847
(0.994-1.239)
1.076
(0.996-1.163)
0.982
(0.909-1.060)
1.004
(0.907-1.110)
0.997
(0.929-1.069)
1.073
(0.994-1.159)
0.993
(0.920-1.072)
0.999
(0.927-1.077)
1.033
(0.952-1.120)
1.021
(0.924-1.128)
0.950
(0.883-1.023)
1.011
(0.942-1.085)
0.993
(0.894-1.103)
0.062
0.635
0.944
0.924
0.071
0.862
0.986
0.437
0.690
0.172
0.761
0.894
(0.994-1.156)
1.038
(0.985-1.094)
0.966
(0.917-1.017)
1.043
(0.975-1.115)
1.044
(0.997-1.095)
1.005
(0.955-1.057)
1.016
(0.965-1.070)
1.041
(0.990-1.095)
1.036
(0.981-1.094)
1.015
(0.949-1.086)
0.978
(0.931-1.027)
1.000
(0.954-1.049)
0.996
(0.929-1.067)
0.166
0.190
0.221
0.069
0.861
0.544
0.117
0.208
0.663
0.372
0.992
0.905
CAD, coronary artery disease; CV, cardiovascular. * non significant after Bonferroni correction
7
Supplemental Table S4. Summary association statistics about total and cardiovascular mortality based on all data for 29
independent SNPs in the MDC after full adjustment.
Death (any cause)
Gene
Index SNP
MTHFR-NPPB
rs17367504
MOV10
rs2932538
SLC4A7
rs13082711
ULK4
rs3774372
MECOM
rs419076
FGF5
rs16998073
SLC39A8
rs13107325
GUCY1A3GUCY1B3
NPR3-C5orf23
rs13139571
EBF1
rs11953630
HFE
rs1799945
BAT2-BAT5
rs805303
CACNB2(5')
rs4373814
C10ORF107
rs1530440
PLCE1
rs932764
CYP17A1-NT5C2
rs11191548
ADM
rs7129220
rs1173771
Beta
(95% CI)
0.946
(0.891-1.004)
1.060
(1.009-1.115)
1.023
(0.971-1.078)
0.960
(0.904-1.020)
0.996
(0.954-1.040)
0.971
(0.927-1.017)
0.975
(0.879-1.082)
0.995
(0.945-1.048)
1.017
(0.973-1.063)
1.007
(0.962-1.053)
1.010
(0.945-1.080)
0.996
(0.952-1.041)
1.023
(0.979-1.068)
0.991
(0.937-1.048)
1.003
(0.960-1.048)
1.029
(0.958-1.105)
1.001
p-value
0.065
0.022*
0.396
0.185
0.867
0.210
0.640
0.841
0.461
0.777
0.762
0.846
0.309
0.746
0.883
0.440
0.987
Death (cardiovascular)
Beta
(95% CI)
1.005
(0.903-1.118)
1.071
(0.981-1.169)
1.035
(0.945-1.134)
0.935
(0.842-1.037)
0.968
(0.897-1.044)
1.067
(0.985-1.156)
1.061
(0.891-1.264)
1.024
(0.935-1.122)
1.024
(0.948-1.107)
1.011
(0.934-1.094)
1.054
(0.939-1.182)
1.011
(0.935-1.094)
1.066
(0.987-1.150)
0.997
(0.904-1.100)
1.034
(0.958-1.116)
1.009
(0.891-1.142)
1.013
p-value
0.932
0.128
0.454
0.202
0.398
0.113
0.503
0.606
0.547
0.789
0.374
0.779
0.102
0.956
0.394
0.886
0.842
8
PLEKHA7
rs381815
FLJ32810TMEM133
ATP2B1
rs633185
SH2B3
rs653178
TBX5-TBX3
rs10850411
CYP1A2-ULK3
rs1378942
FES
rs2521501
PLCD3
rs12946454
GOSR2
rs17608766
ZNF652
rs16948048
JAG1
Dovr +
GNAS-EDN3
rs1327235
rs2681492
rs6015450
(0.932-1.074)
0.995
(0.948-1.044)
0.994
(0.947-1.042)
0.984
(0.924-1.048)
1.024
(0.981-1.070)
0.977
(0.932-1.024)
0.993
(0.947-1.041)
1.012
(0.966-1.161)
0.989
(0.940-1.041)
1.006
(0.945-1.070)
0.996
(0.952-1.042)
1.040
(0.995-1.086)
0.952
(0.893-1.016)
0.834
0.789
0.612
0.274
0.330
0.764
0.608
0.670
0.859
0.870
0.081
0.139
(0.895-1.146)
0.994
(0.913-1.082)
0.999
(0.919-1.085)
1.049
(0.942-1.169)
1.096
(1.016-1.183)
0.996
(0.918-1.081)
0.989
(0.910-1.076)
1.019
(0.940-1.106)
1.021
(0.935-1.116)
0.902
(0.806-1.011)
1.005
(0.929-1.086)
1.064
(0.986-1.149)
0.927
(0.827-1.039)
0.811
0.972
0.381
0.017*
0.931
0.799
0.643
0.640
0.076
0.906
0.109
0.192
* non significant after Bonferroni correction
9
Supplemental table S5 Cox regression exploring the association between TRF including systolic BP and CVD
age
sex
Age_x_sex
*Age2
BMI
Diabetes
Smoke
Systolic BP
Antilipemic drugs
Beta
S.E.
Wald
P-value
H.R.
0.112
-1.941
0.020
-0.001
0.024
0.866
0.646
0.015
0.242
0.044
0.327
0.005
0.000
0.005
0.060
0.039
0.001
0.089
6.460
35.208
15.094
2.559
26.402
210.421
273.976
334.291
7.392
0.011
<0.001
<0.001
0.110
<0.001
<0.001
<0.001
<0.001
0.007
1.119
0.144
1.021
0.999
1.024
2.378
1.909
1.015
1.274
95,0% CI for Exp(B)
Lower
Upper
1.026
1.219
0.076
.273
1.010
1.031
0.999
1.000
1.015
1.033
2.115
2.673
1.768
2.061
1.014
1.017
1.070
1.518
BMI, Body Mass Index, BP, Blood Pressure; S.E. Standard Error; H.R. Hazard Ratio; C.I, Confidence Interval;
Discarded from the model because on significant.
10
Supplemental Figure S1. Histogram showing the distribution of GRS per number of valid
genotypes after standardization.
Average: 0.072 SD: 0.6534 N=27,003
The boundaries for the inclusion in different tertiles were as follows: 1st tertile:-2.9609 to 0.2039;
2nd tertile: 0.2039 to 0.3568;
3rd tertile: 0.3568 to 5.4050;
11
Supplemental figure S2. ROC curve for stroke discrimination using non genetic risk factors
(age, sex, age2, age x sex, BMI, hypertension, diabetes, smoking, use of antilipemic drugs) as
compared to non genetic risk factors plus the weighted GRS
………… non genetic risk facttors;
………… non genetic risk factors+GRS;
………… reference line
AUC= 0.743±0.004; p<0.001
AUC= 0.743±0.004; p<0.001
12
Reference List
1. Ehret GB, Munroe PB, Rice KM, Bochud M, Johnson AD, Chasman DI et al. Genetic
variants in novel pathways influence blood pressure and cardiovascular disease risk. Nature
2011; 478(7367):103-109.
2. Levy D, Ehret GB, Rice K, Verwoert GC, Launer LJ, Dehghan A et al. Genome-wide
association study of blood pressure and hypertension. Nat Genet 2009; 41(6):677-687.
3. Newton-Cheh C, Johnson T, Gateva V, Tobin MD, Bochud M, Coin L et al. Genome-wide
association study identifies eight loci associated with blood pressure. Nat Genet 2009;
41(6):666-676.
4. Horne BD, Anderson JL, Carlquist JF, Muhlestein JB, Renlund DG, Bair TL et al.
Generating genetic risk scores from intermediate phenotypes for use in association studies
of clinically significant endpoints. Ann Hum Genet 2005; 69(Pt 2):176-186.
5. Pencina MJ, D'Agostino RB, Sr., D'Agostino RB, Jr., Vasan RS. Evaluating the added
predictive ability of a new marker: from area under the ROC curve to reclassification and
beyond. Stat Med 2008; 27(2):157-172.
13