Supplementary Table S1: Methods applied to determine the biochemical status of ER, PgR, and HER2 in the different datasets.
Biochemical Estrogen receptor status Biochem. Progesterone receptor status HER2 status
Dataset n= Method cutoff n= Method Cutoff n= Method Cutoff Reference
Rotterdam 344 LBA, EIA, (IHC n=9) 10 fmol/mg, (IHC:
10% pos. cells)
256 LBA, EIA 4 fmol/mg, n.a. n.a. n.a. 14, 15, 16
TransBIG 198 IHC not given n.a. n.a. n.a. n.a. n.a. n.a. 18
Oxford‐
Untreated
63 not given not given n.a. n.a. n.a. n.a. n.a. n.a. 19
London 87 not given not given 85 not given not given n.a. n.a. n.a. 20
London‐2 77 not given not given 77 not given not given n.a. n.a. n.a. 21
Oxford‐
Tamoxifen
109 not given not given n.a. n.a. n.a. n.a. n.a. n.a. 20
Veridex‐Tam 136 IHC, LBA 10% pos. cells, 10 fmol/mg,
n.a. n.a. n.a. n.a. n.a. n.a. 22
Frankfurt‐3 50 IHC 10% pos. tumor cells
46 IHC 10% pos. tumor
cells
19 IHC / FISH IHC3+ OR FISH >2.0 12
Uppsala 249 EIA >0.05 fmol/µg DNA 251 EIA >0.05 fmol/µg DNA n.a. n.a. n.a. 25
San Francisco 118 not given not given 117 not given not given 79 not given not given 26
New York 99 not given not given 98 not given not given 88 n.a. n.a. 27
Frankfurt 114 IHC 10% pos. tumor
cells
112 IHC 10% pos. tumor
cells
65 IHC / FISH IHC3+ OR FISH >2.0 10 Frankfurt‐2 65 IHC 10% pos. tumor
cells
65 IHC 10% pos. tumor
cells
57 IHC / FISH IHC3+ OR FISH >2.0 13
MDA133 133 IHC 10% pos. tumor
cells
133 IHC 10% pos. tumor
cells
132 IHC / FISH IHC3+ OR FISH >2.0 8
EORTC 48 IHC not given 47 IHC not given n.a. n.a. n.a. 29
Edinburgh 116 IHC not given n.a. n.a. n.a. n.a. n.a. n.a. 30
expO 153 not given not given 151 not given not given 141 not given not given 31
Boston 39 not given not given 39 not given not given 37 not given not given 34
TOTAL 2198 1474 618
Supplementary Table S2: Univariate and multivariate Cox regression analysis of disease free survival according to microarray based ER, PgR, and HER2 status among 2058 breast cancer patients.
Univariate Multivariate
Parameter n= P Value* Hazard Ratio 95 % CI P Value* Hazard Ratio 95 % CI
ER status negative vs positive 510 vs. 1548 0.002 1.34 (1.12‐1.60) 0.78 1.03 (0.84‐1.27)
PgR status negative vs positive 854 vs. 1204 <0.001 1.55 (1.32‐1.82) <0.001 1.48 (1.23‐1.78)
HER2 status positive vs negative 238 vs. 1820 0.002 1.43 (1.13‐1.80) 0.083 1.24 (0.97‐1.58)
*significant P values are given in bold
Supplementary Figure S1: Event free survival of patients according to the available endpoint.
A) Event free survival of ER positive breast cancers (n=1549) with either RFS (n=919) or DMFS (n=630) event status available.
B) Event free survival ER negative breast cancers (n=510) with either RFS (n=261) or DMFS (n=249) event status available.
months
120 100
80 60
40 20
0
event fr ee sur v ival
1.0
0.8
0.6
0.4
0.2
0.0
B
ER negative tumors (n=510)
Patients with only DMFS available (n=261) Patients with RFS available (n=249)
(P=0.18)
months
120 100
80 40 60
20 0
event fr ee sur v ival
1.0
0.8
0.6
0.4
0.2
0.0
A
Patients with only DMFS available (n=919) Patients with RFS available (n=630)
ER positive tumors (n=1549)
(P=0.25)
Supplementary Figure S2: ROC curves demonstrating the relationship of sensitivity and specifity of different Affymetrix ProbeSets for the estrogen receptor with biochemical data of the estrogen receptor status.
1 - specificity
sensitivity
Probe Set AUC 95% CI
205225_at 0.949 0.938 0.960 215552_s_at 0.820 0.802 0.839 211235_s_at 0.783 0.763 0.802 211233_x_at 0.779 0.759 0.799 217190_x_at 0.759 0.738 0.781 211234_x_at 0.752 0.731 0.772 211627_x_at 0.580 0.554 0.606 215551_at 0.533 0.508 0.558 217163_at 0.513 0.486 0.539
Supplementary Figure S3: ROC curves demonstrating the relationship between sensitivity and specificity of Affymetrix ProbeSets for the PgR (A) and HER2 genes (B) with the results from immunohistochemistry of the progesterone receptor and HER2 (3+ IHC or FISH positive).
1.0
0.8
0.6
0.4
0.2
0.0
1.0 0.8
0.6 0.2 0.4
0.0
1 - specificity
sensitivity
216836_s_at 210930_s_at
1 - specificity
sensitivity
1.0
0.8
0.6
0.4
0.2
0.0
1.0 0.6 0.8
0.4 0.0 0.2
A B
Supplementary Figure S4: Distribution of ER expression values in the individual datasets.
dataset Rotterdam Mainz TransBIG Oxford‐Untreated London
cutoff 0.0072 0.0091 0.0051 0.0034 0.0082
distribution
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
020406080100
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150200250
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
0100200300
an
dataset London‐2 Oxford‐Tamoxifen Veridex‐Tam Frankfurt‐3 Stockholm
cutoff 0.0101 0.0018 0.0031 0.0040 0.0053
distribution
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150200250
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150200250300
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150200
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150200
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
020406080100120140
an
dataset Uppsala San Francisco New York Frankfurt Frankfurt‐2
cutoff 0.0092 0.0064 0.0066 0.0056 0.0064
distribution
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
020406080100120
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150
an
dataset MDA133 EORTC Edinburgh expO Signapore
cutoff 0.0072 0.0067 0.0061 0.0080 0.0078
distribution
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
020406080100
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150200
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
020406080100
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
020406080
an
dataset Genentech Boston combined
cutoff 0.0054 0.0093 0.0075
distribution
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150200
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
050100150200250300
an
-0.010 -0.005 0.000 0.005 0.010 0.015 0.020
020406080100
an
Supplementary Figure S5: Distribution of PgR expression values in the individual datasets.
dataset Rotterdam Mainz TransBIG Oxford‐Untreated London
cutoff ‐0.0075 ‐0.0052 ‐0.0084 ‐0.0068 ‐0.0088
distribution
-0.02 -0.01 0.00 0.01
020406080100
an
-0.02 -0.01 0.00 0.01
020406080100
an
-0.02 -0.01 0.00 0.01
020406080
an
-0.02 -0.01 0.00 0.01
020406080100120
an
-0.02 -0.01 0.00 0.01
020406080100
an
dataset London‐2 Oxford‐Tamoxifen Veridex‐Tam Frankfurt‐3 Stockholm
cutoff ‐0.0064 ‐0.0081 ‐0.0070 ‐0.0095 ‐0.0075
distribution
-0.02 -0.01 0.00 0.01
050100150
an
-0.02 -0.01 0.00 0.01
020406080100
an
-0.02 -0.01 0.00 0.01
020406080100
an
-0.02 -0.01 0.00 0.01
020406080100120
an
-0.02 -0.01 0.00 0.01
050100150
an
dataset Uppsala San Francisco New York Frankfurt Frankfurt‐2
cutoff ‐0.0084 ‐0.0099 ‐0.0069 ‐0.0064 ‐0.0081
distribution
-0.02 -0.01 0.00 0.01
020406080100
an
-0.02 -0.01 0.00 0.01
020406080100120140
an
-0.02 -0.01 0.00 0.01
050100150
an
-0.02 -0.01 0.00 0.01
020406080100
an
-0.02 -0.01 0.00 0.01
050100150
an
dataset MDA133 EORTC Edinburgh expO Signapore
cutoff ‐0.0079 ‐0.0054 ‐0.0082 ‐0.0076 ‐0.0071
distribution
-0.02 -0.01 0.00 0.01
020406080100
an
-0.02 -0.01 0.00 0.01
050100150200
an
-0.02 -0.01 0.00 0.01
050100150
an
-0.02 -0.01 0.00 0.01
020406080100120
an
-0.02 -0.01 0.00 0.01
020406080100120
an
dataset Genentech Boston combined
cutoff ‐0.0123 ‐0.0047 ‐0.0078
distribution
-0.02 -0.01 0.00 0.01
050100150200
an
-0.02 -0.01 0.00 0.01
050100150200250
an
-0.02 -0.01 0.00 0.01
020406080
an
Supplementary Figure S6: Distribution of HER2 expression values in the individual datasets.
dataset Rotterdam Mainz TransBIG Oxford‐Untreated London
cutoff 0.0137 0.0122 0.0124 0.0133 0.0120
distribution
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200250
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200250
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200250300
an
dataset London‐2 Oxford‐Tamoxifen Veridex‐Tam Frankfurt‐3 Stockholm
cutoff 0.0140 0.0120 0.0139 0.0163 0.0136
distribution
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200250300
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200250300350
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200250300
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200250
an
dataset Uppsala San Francisco New York Frankfurt Frankfurt‐2
cutoff 0.0135 0.0119 0.0151 0.0146 0.0145
distribution
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200250
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200250
an
dataset MDA133 EORTC Edinburgh expO Signapore
cutoff 0.0144 0.0092 0.0121 0.0141 0.0147
distribution
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200250
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
020406080100120140
an
dataset Genentech Boston combined
cutoff 0.0145 0.0155 0.0135
distribution
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150200250300
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150
an
-0.005 0.000 0.005 0.010 0.015 0.020 0.025
050100150
an
Suppl. Figure S7: Influence of gene normalization on the distribution of ER expression values.
A) Distribution of gene normalized ER expression values (ProbeSet 205225_at) in the Frankfurt dataset (n=120 samples). Two normal distributions were fitted to the data allowing separation of ER positive and ER negative samples.
B) Distribution of ER expression values after gene normalization of only ER positive samples (n=79) from the Frankfurt dataset. The same scale as in (A) has been used for the X axis.
C) Distribution of ER expression values after gene normalization of only ER negative samples (n=41) from the Frankfurt dataset. The same scale as in (A) has been used for the X axis.
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
-0.4 -0.2 0.0 0.2 0.4
Density
gene normalized ER expression ER negative samples only
0 1 2 3 4 5
-0.4 -0.2 0.0 0.2 0.4
Density
gene normalized ER expression ER positive samples only
-0.4 -0.2 0.0 0.2 0.4
0 2 4 6 8
Density
gene normalized ER expression ER positive and negative samples
A
B
C
Supplementary Figure S8: Disease free survival of patients according to biochemical and microarray derived status for ER and PgR.
Kaplan Meier analysis of disease free survival of patients stratified according biochemical (A,D) or microarray (B,C,E,F) status of ER (A,B,C) and PgR (D,E,F). Either only samples with biochemical derived status (A,B,D,E) or all 2058 samples with follow up data (C,F) were included.
months
120 100 80 60 40 20 0
Disease free survival 1.0
0.8
0.6
0.4
0.2
0.0
PgR by microarray (n=2058)
F
PgR+ (n=1204) PgR- (n=854)
months
12 0 100 80 60 40 20 0
Disease free survival
PgR by microarray (n=1085)
E
1.0
0.8
0.6
0.4
0.2
0.0
PgR+ (n=612) PgR- (n=473)
months
120 100 80 60 40 20 0
Disease free survival 1.0
0.8
0.6
0.4
0.2
0.0
PgR biochemical (n=1085)
PgR+ (n=685) PgR- (n=400)
D
months
120 100 80 60 40 20 0
1.0
0.8
0.6
0.4
0.2
0.0
ER by microarray (n=2058)
C
ER+ (n=1548) ER- (n=510)
months
120 100 80 60 40 20 0
ER by microarray (n=1683)
B
1.0
0.8
0.6
0.4
0.2
0.0
ER+ (n=1252) ER- (n=431)
months
120 100 80 60 40 20 0 1.0
0.8
0.6
0.4
0.2
0.0
ER biochemical (n=1683)
ER+ (n=1278) ER- (n=405)
A
Disease free survival
Supplementary Figure S9: Disease free survival of untreated (A) and endocrine treated (B) patients stratified according to ER and PgR status based on microarray.
All patients were with available follow up information were selected which were either untreated or treated only with adjuvant endocrine therapy and which have both ER and PgR status available based on microarray as well as biochemical assay (n=722).
months
120 100
80 60
40 0 20
Disease free survival
Endocrine treatment (n=314)
B
1.0
0.8
0.6
0.4
0.2
0.0
ER
+PgR
+(n=220) ER
-PgR
-(n=17) ER
+PgR
-(n=74) ER
-PgR
+(n=3)
months
120 100
80 60
40 20
0
Disease free survival
1.0
0.8
0.6
0.4
0.2
0.0
ER
+PgR
+(n=198) ER
-PgR
-(n=87) ER
+PgR
-(n=101) ER
-PgR
+(n=22) Untreated (n=408)
A
Supplementary Figure S10: Comparison of Affymetrix expression values for ER, HER2 and PgR.
Distribution of expression values are given on an identical scale on the x‐axis (ProbeSets
205225_at, 216836_s_at, and 208305_at for ER, HER2 and PgR, respectively). Color codes represent the biochemical derived status of the samples for ER, HER2, and PgR.
PgR biochemical:
positive negative
PgR microarray
normalized log MAS5 values
‐0.02 ‐0.01 0 0.01 0.02
HER2 biochemical:
positive negative
HER2 microarray ER microarray
ER biochemical:
positive
negative
ER status:
PgR status:
HER2 status:
Supplementary Figure S11: Comparison of biochemical ER, HER2 and PgR status with microarray derived status using individual dataset specific cutoff values in a training‐validation approach.
Specific cutoff values derived from fitting in individual datasets (as training sets) were validated using all remaining samples with information on biochemical status (validation sets). The observed values for overall accuracy sensitivity and specificity are given for each cutoff derived from a single dataset.
0.0%
20.0%
40.0%
60.0%
80.0%
100.0%
NewYork Edinburgh EORTC Frankfurt Rotterdam MDA133 Frankfurt_2 San_Francisco Genentech Stockholm TransBIG Signapore expO Frankfurt_3 London Oxford_Untr… Veridex_Tam Mainz Uppsala Boston London_2 Oxford_Tam…
Accuracy Sensitivity Specificity
0.0%
20.0%
40.0%
60.0%
80.0%
100.0%
Frankfurt London_2 London Edinburgh Rotterdam MDA133 Uppsala Stockholm TransBIG Oxford_Tamo… Frankfurt_2 Mainz Frankfurt_3 expO EORTC NewYork Oxford_Untre… Veridex_Tam Signapore San_Francisco Boston Genentech
Accuracy Sensitivity Specificity
0.0%
20.0%
40.0%
60.0%
80.0%
100.0%
Edinburgh Mainz London Oxford_Tamoxi… San_Francisco Oxford_Untrea… TransBIG Uppsala Rotterdam Veridex_Tam Stockholm MDA133 London_2 expO Frankfurt_2 Genentech Frankfurt Signapore NewYork Boston Frankfurt_3 EORTC
Accuracy Sensitivity Specificity