Background
Methods
Patients
Immunohistochemistry
Breast cancer types based on IHC features
Statistical analysis
ER/PgR expression | HER2 phenotypes | Breast cancer type | Number of patients | Total | ||
---|---|---|---|---|---|---|
Binary classification | semiquantitative expression | Ki-67 < =20% | Ki-67 > 20% | |||
ER + PgR+ | “negative” | 0+ |
Luminal A/B1
| 248 | 90 | 338 |
1+ | 165 | 68 | 233 | |||
2+ | 44 | 19 | 63 | |||
“overexpressed” | 3+ |
Luminal B2
| 82 | 83 | 165 | |
Total of patients with ER & PgR positive tumors | 539 | 260 | 799 | |||
ER + PgR- | “negative” | 0+ |
Luminal A/B1
| 15 | 10 | 25 |
1+ | 14 | 12 | 26 | |||
2+ | 4 | 3 | 7 | |||
“overexpressed” | 3+ |
Luminal B2
| 13 | 20 | 33 | |
Total of patients with ER positive & PgR negative tumors | 46 | 45 | 91 | |||
ER-PgR+ | “negative” | 0+ |
Luminal A/B1
| 1 | 3 | 4 |
1+ | 3 | 2 | 5 | |||
2+ | 2 | 0 | 2 | |||
“overexpressed” | 3+ |
Luminal B2
| 0 | 0 | 0 | |
Total of patients with ER negative & PgR positive tumors | 6 | 5 | 11 | |||
ER-PgR- | “negative” | 0+ |
triple-negative
| 20 | 86 | 106 |
1+ | 9 | 31 | 40 | |||
2+ | 2 | 6 | 8 | |||
“overexpressed” | 3+ |
pure HER2
| 27 | 98 | 125 | |
Total of patients with ER & PgR negative tumors | 58 | 221 | 279 | |||
Total of all patients | 649 | 531 | 1180 |
Results
Distribution of KI-67 values regarding ER/PgR and status of HER2 expression
-
Among the ER + PgR+ tumors, Ki-67 values were higher in HER2 3+ cancer than in tumors with low HER2 expression (1+), or without any expression (HER2 absent) (p < 0.0001).
-
Among the ER + PgR- tumors, no difference in Ki67 values, depending on the HER2 was found (p = 0.3175).
-
Particularly interesting were ER-PgR- tumors (in the bottom row of Fig. 1). The highest levels of Ki-67 values are found in tumors without expression of HER2 (HER2 absent). The presence of HER2 reduced KI-67 values slightly and this downslope holds for the whole sequence of HER2 absent to HER2 3+. The difference between the cancers without HER2 (HER2 absent) and cancers overexpressing HER2 (HER2 3+), was statistically significant (p = 0.0003).
The first stage EM clustering within subgroups of ER/PgR and HER2 phenotypes
-
In ER positive tumors, dominant clusters consisted of patients with low Ki-67 values (columns labeled LMA for Low Mitotic Activity, with mean values from 10 to 16% in Table 2.). In two ER+ and HER2 3+ subgroups, the LMA analogous clusters showed mean Ki67 values from 19 to 26%, suggesting that HER2 overexpression increases Ki67 values of tumors with low mitotic activity.
-
In all HER2 absent (0+), HER2 1+ and HER2 2+ subgroups, one cluster contains patients whose tumors show intermediate Ki-67 values (near 40% are mean KI-67 values,), here defined as the IMA clusters (from Intermediate Mitotic Activity)
-
In ER positive tumors, the share of IMA clusters declines with HER2 expression (among PgR+ cancers: from 25% of HER2 absent to 15% in HER2 3+; among PgR- cancers: from 40% in HER2 absent to 6.1% in HER2 3+ cancers).
-
In two ER+ HER2 3+ subgroups, the intermediate range clusters shows mean Ki67 values 55 to 60%, suggesting that among these tumors HER2 overexpression increased Ki67 values and reduced share of IMA tumors.
-
Among ER negative tumors HER2 expression did not boost mitotic rates of dominant IMA clusters (30 to 35%), but it reduced the share of the cluster with high Ki67 values (high mitotic activity - HMA), from 40% in HER2 absent tumors to 11.2 in HER2 3+ cancers, resulting in overall lower Ki-67 values among the pure HER2 tumors.
ER/PgR phenotypes | Breast cancer HER2 status | |||||||||
---|---|---|---|---|---|---|---|---|---|---|
0+ | 1+ | 2+ | 3+ | |||||||
ER + PgR+ |
Cancer types
|
Luminal A/B1
|
Luminal B2
| |||||||
EM clusters | LMA | IMA | LMA | IMA | LMA | IMA | LMA | IMA | ||
Patients | n | 255 | 83 | 187 | 46 | 52 | 11 | 139 | 26 | |
%
|
75.4
|
24.6
|
80.3
|
19.7
|
82.5
|
17.5
|
84.2
|
15.0
| ||
Mean Ki-67% | 9.5 | 39.1 | 11.3 | 39.6 | 13.5 | 38.6 | 19.2 | 56.6 | ||
St.dev. of Ki-67 | 6.1 | 15.2 | 7.2 | 11.1 | 6.9 | 9.5 | 10.1 | 10.3 | ||
ER + PgR- |
Cancer types
|
Luminal A/B1
|
Luminal B2
| |||||||
EM clusters | LMA | IMA | LMA | IMA | n/a | LMA | IMA | |||
Patients | n | 15 | 10 | 20 | 6 | 7 | 31 | 2 | ||
%
|
60.0
|
40.0
|
76.9
|
23.1
|
n/a
|
93.9
|
6.1
| |||
Mean Ki-67% | 9.7 | 42.4 | 15.5 | 39.0 | 25.4 | 60.0 | ||||
St.dev. of Ki-67 | 5.4 | 14.8 | 8.6 | 6.0 | 13.3 | 7.1 | ||||
ER-PgR- |
Cancer types
|
triple-negative
|
pure HER2
| |||||||
EM clusters | IMA | HMA | IMA | HMA | n/a | IMA | HMA | |||
Patients | n | 63 | 43 | 29 | 11 | 8 | 111 | 14 | ||
%
|
59.4
|
40.6
|
72.5
|
27.5
|
n/a
|
88.8
|
11.2
| |||
Mean Ki-67% | 34.0 | 79.0 | 32.3 | 82.7 | 32.4 | 73.2 | ||||
St.dev. of Ki-67 | 18.3 | 6.4 | 18.1 | 8.5 | 14.7 | 8.7 |
The second stage EM clustering of the pooled data set
Breast cancer patients | Three EM clusters from the pooled patients’ data | Values for all patients | ||||||
---|---|---|---|---|---|---|---|---|
LMA (low mitotic activity) | IMA (intermediate mitotic activity) | HMA (high mitotic activity) | ||||||
Mean Ki-67% | 13.17 | 40.45 | 77.79 | 25.45 | ||||
St.dev. of Ki-67 | 8.43 | 13.77 | 8.45 | 21.08 | ||||
Breast cancer types | n |
%
| n |
%
| n |
%
| Total |
%
|
Luminal A/B1 | 560 |
80.92
| 124 |
17.92
| 8 |
1.16
| 692 |
100.00
|
Luminal B2 | 126 |
63.64
| 68 |
34.34
| 4 |
2.02
| 198 |
100.00
|
triple-negative | 41 |
26.62
| 58 |
37.66
| 55 |
35.71
| 154 |
100.00
|
pure HER2 | 42 |
33.60
| 73 |
58.40
| 10 |
8.00
| 125 |
100.00
|
Total | 769 |
65.78
| 323 |
27.63
| 77 |
6.59
| 1169 |
100.00
|
-
Among 799 ER + PgR+ patients distribution in clusters was HER2 dependent (p = 0.000243), due to increased number of IMA HER2 3+ tumors on the expense of LMA HER2 3+ tumors (52 IMA out of 162 HER2 3+ patients versus113 IMA out of 637 HER2 < 3+ patients).
-
This was not found among ER + PgR- patients (p = 0.186968). Mitotic activity of ER + PgR- tumors seems unrelated to HER2 status, possibly due to the presence of “dysfunctional” ER that do not stimulate PgR expression.
-
Among ER-PgR- patients, HER2 overexpression also increased number of IMA tumor, but by reducing the number of HMA tumors (p < 0.000001). Here, difference between HER2 absent (0+) and HER2 3+ patients was evident (10 HMA out of 125 HER2 3+ patients versus 42 HMA out of 103 HER2 0+ patients), while patients with HER2 1+ or 2+ tumors did not differ from the expected frequencies, suggesting that at least among ER negative tumors, the absence of HER2 might be as important as the HER2 overexpression.
Discussion
Possible promitotic mechanisms in breast tumor IHC phenotypes
-
Ki-67 values of tumors with functional ER (ER + PgR+ phenotype) seem dependent both on estrogen exposure and on the status of HER2 expression
-
LMA & IMA clusters of PgR negative phenotypes (ER + PgR- and ER-PgR-) seem similar in their distributions of HER2 values, so HER2 is an unlikely candidate to explain increased Ki-67 values in IMA clusters of these two phenotypes, suggesting that some unknown promitotic mechanism might be involved.
-
Tumors lacking both ER and PgR with high Ki-67 values (HMA clusters with values >65%) seem independent both of estrogen exposure and HER2 expression, so other promitotic mechanisms should be considered.
HER2 & ER/PgR Breast cancer phenotypes | Model proposed subdivision of breast cancers, based on mitotic activity | |||
---|---|---|---|---|
LMA (low mitotic activity) Ki67 < 25% | IMA (intermediate mitotic activity)Ki-67 25–65% | HMA (high mitotic activity)Ki-67 > 65% | ||
HER2 0+ to 3+ | ER + PgR+ | ~67% of all patients probably HER2 dependent mitotic rates, thus intermediate mitotic rates seem dependent on the increased HER2 expression | ~ 1% of all patients high mitotic rate due to unknown promiotic mechanism | |
ER + PgR- | ~26% of all patients mitotic rates do not seem closely regulated by normal HER2 expression (0+ to 2+), HER2 3+ increases number of IMA tumors | |||
ER-PgR- | ~6% of all patients high mitotic rate due to unknown promiotic mechanism, HER2 3+ reduces number of HMA tumors | |||
Open questions | 168 out of 769 LMA cancers were HER2 3+ Can HER2 molecules in HER2 3+ & LMA cancers be dysfunctional? | 99 out of 323 IMA cancers were HER2 absent Is there another promitotic mechanism in HER2 absent & IMA tumors, particularly in 55 ER + PgR+ cancers? | 43 out of 106 triple-negative & HER2 absent cancers were HMA What promotes the highest mitotic rates in HER2 absent ER-PgR- tumors? |
-
If 168 cases out of our 769 breast cancers in the LMA cluster were HER2 3+, does this suggest that in these tumors HER2 molecules might be dysfunctional and thus result in unexpectedly low Ki-67 values despite the HER2 overexpression?
-
If 99 out of our 323 breast cancers in the IMA cluster were HER2 absent, does this suggest that another promitotic mechanism should be searched for in HER2 absent & IMA tumors, particularly in those 55 cancers showing the ER + PgR+ phenotype?
-
If 43 out of 106 our triple-negative & HER2 absent cancers belonged to the HMA cluster, is there some special feature that promotes the highest mitotic rates in triple-negative breast cancers with no HER2 molecules? It almost seems that among triple-negative tumors any status of HER2 presence is associated with a reduction in Ki-67 values.