IGF1R Protein Expression is Not Associated with Differential Benefit to Concurrent Trastuzumab in Early-Stage HER2-positive Breast Cancer from the North Central Cancer Treatment Group (Alliance) Adjuvant Trastuzumab Trial N9831

Abstract

Background

Preclinical evidence indicates that increased insulin-like growth factor receptor-1 (IGF1R) signaling interferes with the action of trastuzumab suggesting a possible mechanism of trastuzumab resistance. Thus, we evaluated IGF1R prevalence, relationship with demographic data, and association with disease-free survival (DFS) of patients randomized to chemotherapy alone (Arm A) or chemotherapy with sequential (Arm B) or concurrent trastuzumab (Arm C) in the prospective phase III HER2-positive adjuvant N9831 trial.

Methods

IGF1R protein expression was determined in tissue microarray sections (3 cores per block; N=1197) or in whole tissue sections (WS) (N=537) using immunohistochemistry (rabbit polyclonal antibody against IGF1R β-subunit). A tumor was considered positive (IGF1R+) if any core or WS had ≥1+ membrane staining in >0% invasive cells. Median follow-up was 8.5 yrs.

Results

Of 1734 patients, 708 (41%) had IGF1R+ breast tumors. IGF1R+ was associated with younger age (median 48 vs 51, p=0.007), estrogen receptor/progesterone receptor positivity (78% vs 35%, p<0.001), nodal positivity (89% vs 83%, p<0.001), well/intermediate grade (34% vs 24%, p<0.001), tumors ≥2 cm (72% vs 67%, p=0.02) but not associated with race or tumor histology. IGF1R did not impact DFS within arms. Between Arms A and C, patients with IGF1R+ and IGF1R− tumors had DFS hazard ratios (HRs) of 0.48 (p=<0.001) and 0.68 (p=0.009), respectively (interaction p=0.17). Between Arms A and B, patients with IGF1R+ and IGF1R− tumors had DFS HRs of 0.83 (p=0.25) and 0.69 (p=0.01), respectively (interaction p=0.42).

Conclusions

In contrast to preclinical studies that suggest a decrease in trastuzumab sensitivity in IGF1R+ tumors, our adjuvant data show benefit of adding trastuzumab for patients with either IGF1R+ and IGF1R− breast tumors.

INTRODUCTION

Approximately 15–20% of breast cancer tumors show overexpression of the human epidermal growth factor receptor-2 (HER2) protein, which is a poor prognostic indicator that can lead to more malignant phenotypes, including increased metastases and relative therapeutic resistance. Compared to traditional anticancer treatments, trastuzumab, a humanized monoclonal antibody that binds the extracellular domain of the receptor tyrosine kinase, HER2, has proven to be an important option for treating patients with HER2-positive (HER2+) breast cancer(1, 2). Although stringent criteria exist for patient selection, uniform clinical benefit of trastuzumab is not observed in all patients with HER2-overexpressing tumors and resistance continues to be a significant problem(3). The response rate for trastuzumab monotherapy is approximately 20–25%, and up to a quarter of women diagnosed with HER2+ early disease develop tumor relapse within three years(2, 4–6). This suggests that HER2 overexpression, though required, is not the sole determinant of how well breast tumors respond to trastuzumab(3, 6). Therefore, an important research and clinical goal is to identify additional markers that could help predict the patients who are most likely to benefit from trastuzumab.

Alternative signaling from insulin-like growth factor receptor 1 (IGF1R), a receptor tyrosine kinase and major stimulator of the phosphotidylinositide 3 kinase (PI3K) pathway(7), is a proposed mechanism of resistance to trastuzumab(8). The IGF1R signaling pathway is activated by IGFI/II ligand binding and regulates normal cell growth and differentiation and is involved in anti-apoptotic signaling and neoplastic transformation(9–13). IGF1R is a glycosylated heterotetrameric receptor tyrosine kinase composed of two extracellular alpha subunits and two beta subunits with intrinsic kinase activity and is highly expressed in a wide variety of human cancers, including prostate, colon, lung, and breast cancer(14, 15).

The IGF1R pathway has also been shown to exhibit cross-talk with a number of signaling pathways including the HER2 and estrogen receptor (ER) pathways(16, 17). The crosstalk and transactivation of these alternate receptors by IGF1R, suggest its possible role in mediating resistance to therapeutics targeting these other pathways(18, 19). IGF1R has been linked to resistance to chemo/hormonal/biological/radio-therapy(19) and particularly implicated in therapeutic resistance to trastuzumab(8, 20, 21). Preclinical evidence has demonstrated that increased IGF1R signaling interferes with trastuzumab action(8, 22, 23), and synergy has been observed between trastuzumab and anti-IGF1R agents in inhibiting HER2+ tumor cell growth(24, 25). A unique interaction/heterodimerization of IGF1R and HER2 has been observed in trastuzumab-resistant cells but not in sensitive parental cells (26–28). However, clinical evidence suggested that IGF1R expression alone (as determined by immunohistochemistry) was not associated with trastuzumab resistance in patients with HER2+ metastatic breast cancer(29–31), but only the combination of low protein expression of IGF1R and high protein expression of the downstream phospho-ribosomal 6 (pS6) protein was associated with a favorable patient response outcome to trastuzumab in patients with HER2+ metastatic breast cancer(30). Three small retrospective analyses (<70 patients) in the neo-adjuvant and adjuvant settings produced inconsistent results regarding the association between IGF1R expression and trastuzumab benefit(25, 31, 32).

To further explore the association between IGF1R protein expression and outcome of patients with early stage, HER2+ breast cancer who were treated with adjuvant trastuzumab, we investigated the association between disease-free survival (DFS) and IGF1R protein expression in breast cancer patients randomized to receive chemotherapy alone or chemotherapy with sequential or concurrent trastuzumab in the North Central Cancer Treatment Group (NCCTG)-led N9831 (Alliance) adjuvant trastuzumab phase III breast cancer trial.

MATERIALS AND METHODS

Patients

The N9831 trial (NCT00005970) was a phase III trial in which patients were randomized to three arms: Arm A, doxorubicin and cyclophosphamide followed by weekly paclitaxel; Arm B, same as Arm A but followed by one year of sequential trastuzumab; Arm C, same as Arm A but with one year of concurrent trastuzumab, started the same day as weekly paclitaxel (Supplemental Figure 1). Women randomly assigned to the concurrent trastuzumab arm had a significantly increased DFS (P<.001; stratified hazard ratio [HR], 0.52; 95% CI, 0.45 to 0.60) and overall survival (OS) (P<.001; stratified HR, 0.61; 95% CI, 0.50 to 0.75) compared with women assigned to the control arm(1). In the N9831 comparison of sequential versus concurrent trastuzumab chemotherapy, there was an increase in DFS with concurrent trastuzumab (P=.02; HR, 0.77; 95% CI, 0.53 to 1.11). Though the number of events was relatively low, the 5-year OS rate for the sequential and concurrent arms were estimated at 89.7% (95% CI, 87.7% to 91.8%) and 91.9% (95% CI, 90.0% to 93.7%), respectively(2).

All patients included in these analyses were tested for HER2 protein overexpression or gene amplification at a central laboratory (Mayo Clinic, Rochester). Patients were considered positive for HER2 according to the FDA-approved guidelines (IHC: complete 3+ membrane staining ≥ 10% invasive cells; FISH: HER2:CEP17 ratio ≥ 2.0)(33) used in N9831. N9831 was approved by all treating sites’ institutional review boards, and all patients signed informed consent. The Mayo Institutional Review Board and the Correlative Science Committee of the North American Breast Cancer Group (NABCG) approved this translational study.

This study included 1734 eligible/consented patients with sufficient tissue for IGF1R protein expression analyses (692 patients were ineligible/no consent/canceled/lost to follow-up and 1079 had insufficient tissue for analyses) (Supplemental Figure 2). The number of patients represented on tissue microarrays (TMAs) with evaluable tissue cores was 1197 and the number of patients with evaluable whole sections (WS) was 537.

Tissue Microarrays

Tissue microarrays (TMAs) were constructed as part of the translational study component of N9831 using an ATA-27 automated TMA construction system (Beecher Instruments, Silver Spring, MD) as previously described(34). Each TMA contained control biopsies from non-neoplastic human liver, placenta, and tonsil tissues. Whole tissue sections from tumors not represented on TMAs were also examined. We evaluated the concordance between TMA and whole section IGF1R IHC analyses of 87 independent breast tumors, and observed a concordance of 90% and 87% using 1+ and 2+ staining in >0% invasive cells cut points, respectively.

IGF1R Testing Methods

Standard laboratory protocols and quality control measures were followed for immunohistochemistry (IHC). Antigen retrieval was performed on deparaffinized whole or TMA sections (5µm) using preheated EDTA buffer (98°C; 40 min). The tissue sections were treated with Peroxidase Blocking Reagent (Dako, Carpinteria, CA) and serum-free Protein Block (Dako) prior to IHC staining for c-IGF1R (rabbit polyclonal; Cell Signaling, #3027; Danvers, MA; dilution 1:200; 60 min incubation) using a Dako Autostainer Plus (Reference #S3800). The sections were incubated in secondary antibody (MACH4 HRP Polymer; Biocare Medical, Concord, CA). The high-sensitivity diaminobenzidine (DAB+) chromogenic substrate system (Betazoid DAB, Biocare) was used for colorimetric visualization followed by counter staining with hematoxylin. A USA Board-certified pathologist (Dr. Beiyun Chen) evaluated and scored the IFGF1R staining. IGF1R protein positivity (IGF1R+) was defined as 1+ membrane staining in >0% of invasive cells as low levels of IGF1R expression have been shown to be sufficient for establishment and maintenance of the malignant phenotype(12). Because no standard criteria exist for IGF1R positivity and other cut points have been previously examined (29), we also evaluated an alternate cut point of 2+ membrane staining in >0% cells to define IGF1R positivity. The concordance between TMA cores for patients with multiple evaluable cores (all cores had to be concordant) (N=961) was 77% (739/961) and 87% (837/961) for positive IGF1R membrane staining of 1+/2+/3+ and 2+/3+, respectively (Supplemental Information, Table 1).

Statistical Methods

Comparisons of continuous data between groups were made with a two-sample t-test and comparisons of categorical data were made with a χ² test. DFS was the primary endpoint of N9831 and was defined as local, regional, or distant recurrence, contralateral breast cancer, another primary cancer (except squamous or basal cell carcinoma of the skin, carcinoma in situ of the cervix, or lobular carcinoma in situ of the breast), or death from any cause. The duration of DFS was defined as the time from registration to the first DFS event. DFS was estimated by the Kaplan-Meier method. Comparisons between Arms A and C, Arms B and C, and Arms A and B, within subgroups were performed using Cox proportional hazards models (35) stratified by nodal status (1–3 vs. 4–9 vs. ≥10 positive nodes vs. positive sentinel node only vs. negative sentinel node with no axillary nodal dissection vs. axillary nodal dissection with no positive nodes) and hormone receptor status (estrogen receptor positive (ER+) and/or progesterone receptor positive (PR+) vs. negative for both receptors). The ability of IGF1R to predict differential trastuzumab benefit between IGF1R subgroups was tested using Cox proportional hazards models (also stratified by nodal status and hormone receptor status), which included a treatment arm by IGF1R subgroup interaction term. HRs and the corresponding 95% confidence intervals (CIs) were determined for association of IGF1R and DFS. The maximum score of the triplicate TMA biopsies was used for all analyses associated with patient outcome.

The data for these analyses were locked on 04/10/2013. Analyses were performed with SAS version 9.3 (SAS Institute) and a two-sided P < 0.05 was considered statistically significant.

RESULTS

Study Patients

The trial registered 3505 patients into Arms A (1,232 patients), B (1216 patients) and C (1,057 patients) of which 1771 patients (A:655, B:590, C:526) were excluded from this analysis for the following reasons: not HER2+ by central pathology review (A:109, B:90, C:84); canceled prior to initiating therapy (A:15, B:6, C:7); did not meet eligibility criteria (A:21, B:23, C:17); withdrew consent/no consent for future translational analysis (A:65, B:71, C:55); lost to follow-up (A:65, B:43, C:21), no or inadequate tissue/technical failure (A:380, B:357 C:342). Of the 3505 patients, 1734 (A:577, B:626, C:531) were evaluable for IGF1R protein expression (Supplemental Figure 2). The median follow-up time was 8.5 years.

Most of the reported clinicopathological characteristics and outcomes of the 1734 patients enrolled on Arms A, B, and C reported herein were similar to the 1079 consented/eligible patients on Arms A, B, and C excluded from analysis (Supplemental Table 2). Non-cohort patients were more likely to be node-positive (88% vs 85%; p=0.02) and cohort patients tended to have tumors ≥ 2cm (69% vs 64%; p=0.01). The clinicopathological characteristics of the 1734 patients included in this analyses whose tumors had any and no IGF1R staining are shown in Table 1. Patients whose tumors had any IGF1R staining appear to be younger, and have a higher rate of hormone receptor and nodal positivity, well-differentiated tumors (well/intermediate grade), larger tumors (≥ 2 cm), and endocrine therapy compared to those patients whose tumors did not exhibit IGF1R staining. Similar correlations and trends (e.g., histologic grade) were observed between patients with tumor 0–1+ and 2–3+ IGF1R protein expression (Supplemental Table 3).

Table 1.

Patient/Disease Characteristics by ≥1+ IGF1R protein staining >0% (N=1734).

Characteristic	Negative (All 0) Membrane Staining N=1026 (59)	Positive (Any 1,2,3+) Membrane Staining N=708 (41)	Chi- Square p- value
Age (median)	51 (22–80)	48 (27–79)	t-test
			0.007
Age Group:
<40	178 (17)	125 (18)	0.007*
40–49	295 (29)	261 (37)
50–59	349 (34)	211 (30)
≥ 60	204 (20)	111 (16)
Race:
White	867 (85)	618 (87)	0.10
Other	159 (16)	90 (13)
Menopausal Status:
Pre-menopausal or < 50	511 (50)	412 (58)	0.0006
Post-menopausal or ≥ 50	515 (50)	296 (42)
ER/PR Status:
ER or PR Positive	354 (35)	555 (78)	<0.0001
Other	672 (66)	153 (22)
Surgery:
Breast Conserving	397 (39)	282 (40)	0.63
Mastectomy	629 (61)	426 (60)
Nodal Status:
Positive	848 (83)	630 (89)	0.0002
Negative	178 (17)	78 (11)
Predominant Tumor Histology:
Ductal	972 (95)	666 (94)	0.65±
Lobular	32 (3)	21 (3)
Other	22 (2)	20 (3)
Missing	0 (0)	1 (0.1)
Hist. Tumor Grade (Elston/SBR):
Well /Intermediate	2433 (24)	240 (34)	<0.0001
Poor	783 (76)	468 (66)
Pathologic Tumor Size:
< 2 cm	341 (33)	199 (28)	0.02
≥ 2 cm	685 (67)	509 (72)
Received Hormonal Treatment:
Yes	343 (33)	534 (75)	<0.0001±
No	677 (66)	170 (24)
Missing	6 (0.6)	4 (0.6)
Source of tissue:
TMA	618 (60)	579 (82)
Whole Section	408 (40)	129 (18)

Distribution of IGF1R Membrane Protein Expression

Of evaluable 1734 patients, 1026 (59%), 450 (26%), 171 (10%), and 87 (5%) had 0, 1+, 2+, and 3+ IGF1R staining, respectively. Representative IHC staining patterns of various protein expression levels are shown in Figure 1.

Figure 1. Representative IHC Staining Patterns of IGF1R in TMA Core Biopsies (10×).

A. IHC score of 0, no membrane staining. B: IHC score of 1+, weak membrane staining. C: IHC score of 2+, moderate membrane staining. D: IHC score of 3+, strong membrane staining.

Associations between IGF1R Protein Expression and DFS

IGF1R− (IHC 0) versus IGF1R+ (IHC 1–3+)

No significant differences in DFS were observed between patients with IGF1R+ and IGF1R− tumors within any of the three arms using the cut point of IHC 0 vs 1–3+ (Table 2). In comparing DFS of Arms C versus A, patients with IGF1R+ and IGF1R− tumors had HRs of 0.48 (95% CI:0.32–0.72 P=<0.001) and 0.68 (95% CI: 0.50–0.91; P=0.009), respectively (interaction P=0.17) (Figures 2A–B). In comparing DFS of Arms B versus A, patients with IGF1R+ and IGF1R− tumors had HRs of 0.83 (95% CI: 0.60–1.15; P=0.25) and 0.69 (95% CI: 0.52–0.92; P=0.01), respectively (interaction P=0.42) (Figures 2A–B). In comparing DFS of Arms C versus B, patients with IGF1R+ and IGF1R− tumors had HRs 0.57 (95% CI: 0.38–0.86; P=0.007) and 0.99 (95% CI: 0.72–1.35; P=0.92, respectively (interaction P=0.05) (Figure 2A–B).

Table 2.

DFS by IGF1R Membrane Staining (0 vs any 1,2,3+ pos) N=1734 (Stratified by receptor and nodal status)

Arm	IGF1R Membrane Staining	N	#Events	HR	95% CI	p-value	DFS
							3 yr	5 yr
Overall	Negative (All 0)	1026	275	1			84.1	78.5
	Positive (Any 1,2 3+)	708	192	1.07	0.87–1.32	0.52	86.6	81.1
A (N=577)	Negative (All 0)	354	117	1			78.5	72.4
	Positive (Any 1,2 3+)	223	75	1.05	0.74–1.49	0.77	84.2	76.4
B (N=626)	Negative (All 0)	349	83	1			86.2	80.1
	Positive (Any 1,2 3+)	277	81	1.28	0.91–1.79	0.16	85.1	79.2
C (N=531)	Negative (All 0)	323	75	1			87.9	83.4
	Positive (Any 1,2 3+)	208	36	0.89	0.57–1.37	0.59	91.3	88.8

Figure 2. Disease-Free Survival by IGF1R Protein Expression.

A. DFS by IGF1R Protein Expression IHC=0. B. DFS by IGF1R Protein Expression IHC= 1–3+. C. DFS by IGF1R Protein Expression IHC= 0,1+. D. DFS by IGF1R Protein Expression IHC= 2–3+.

A: doxorubicin; C: cyclophosphamide; T: paclitaxel; H: trastuzumab.

IGF1R− (IHC 0, 1+) versus IGF1R+ (IHC 2+–3+)

Similar to our primary cut point, no significant differences in DFS were observed between patients with IGF1R+ and IGF1R− tumors within any of the three arms using the cut point of 0, 1+ versus 2+, 3+ (Table 3). In comparing DFS between Arms C and A, patients with IGF1R+ and IGF1R− tumors had HRs of 0.31 (95% CI: 0.14–0.67; P=0.003); and 0.64 (95% CI: 0.50–0.83; P<0.001), respectively (interaction P=0.11) (Figures 2C–D). In comparing DFS between Arms B and A, patients with IGF1R+ and IGF1R− tumors had HRs of 0.60 (95% CI: 0.34–1.09; P=0.09) and 0.75 (95% CI: 0.60–0.94; P=0.01), respectively (interaction P=0.62) (Figures 2C–D). In comparing DFS between Arms C and B, patients with IGF1R+ and IGF1R− tumors had HRs 0.63 (95% CI: 0.30–1.33; P=0.23) and 0.86 (95% CI: 0.67–1.12; P=0.26), respectively (interaction P=0.59) (Figures 2C–D).

Table 3.

DFS by IGF1R Membrane Staining (0,1+ vs 2,3+) N=1734 (Stratified by receptor and nodal status)

Arm	IGF1R Membrane Staining	N	#Events	HR	95% CI	p-value	DFS
							3 yr	5 yr
Overall	Negative (All 0,1+)	1476	404	1			84.6	78.7
	Positive (Any 2 3+)	258	63	0.97	0.73–1.28	0.80	88.3	84.6
A (N=577)	Negative (All 0,1+)	496	163	1			80.2	73.6
	Positive (Any 2 3+)	81	29	1.149	0.74–1.76	0.56	83.8	76.0
B (N=626)	Negative (All 0,1+)	533	141	1			85.5	78.6
	Positive (Any 2 3+)	93	23	0.97	0.61–1.55	0.89	87.0	85.8
C (N=531)	Negative (All 0,1+)	447	100	1			88.3	84.4
	Positive (Any 2 3+)	84	11	0.75	0.39–1.46	0.40	94.0	91.6

In addition, patients with tumors with IGF1R membrane staining of 0, 1+, 2+ and 3+ had HRs (Arm B vs A) of 0.69 (95% CI 0.52–0.92; P=0.01), 0.92 (95% CI 0.61–1.37; P=0.67), 0.84 (95% CI 0.38–1.84; P=0.66), and 0.43 (95% CI 0.16–1.17; P=0.10), respectively (Figure 3A). Patients with tumors with IGF1R membrane staining of 0, 1+, 2+ and 3+ had HRs (Arm C vs A) of 0.68 (95% CI 0.50–0.91; P=0.01), 0.55 (95% CI 0.33–0.91; P=0.02), 0.47 (95% CI 0.17–1.23; P=0.12), and 0.15 (95% CI 0.03–0.66; P=0.01), respectively (Figure 3B). Lastly, similar trends of associations were observed between patient outcome and IGF1R protein expression in patients with hormone-receptor positive tumors compared to the entire study population (Supplemental Information, Figure 3, Table 4; hormone receptor-negative subgroup analysis shown in Supplemental Information Figure 4 and Table 5).

Figure 3. Forest Plots by IGF1R Protein Expression.

A. Forest Plots for Arms B versus A. B. Forest Plots for Arms C versus A. A: doxorubicin; C: cyclophosphamide; T: paclitaxel; H: trastuzumab.

DISCUSSION

IGF1R is involved in anti-apoptotic signaling and neoplastic transformation(36) and has been associated with both poor and good prognoses in breast cancer. In HER2+ breast cancer, preclinical evidence suggests that IGF1R may be a marker that may help predict resistance to trastuzumab(37), whereas tumor IGF1R protein expression alone did not impact trastuzumab response in the clinical metastatic setting(29–31). Our study is the largest to investigate the impact of IGF1R protein expression on benefit to adjuvant trastuzumab in early breast cancer.

In the cohort of evaluated N9831 tumor specimens (N=1747), IGF1R expression was heterogeneous and characterized by both membrane and cytoplasmic localization, similar to previous findings(38–40). We observed that 41% of the HER2 breast tumors exhibited membrane IGF1R protein expression to some extent (≥1+) and 15% exhibited at least moderate (≥2+) membrane expression. Initial radioimmunoassay (RIA) findings demonstrated IGF1R activity in 39%–93% of breast cancers(41–44). Early immunohistochemical studies have shown that 44% of 210 unselected primary breast cancers overexpressed IGF1R (defined as ≥2+) and almost 90% of cases demonstrated at least 1+ staining using clone 24–31, a monoclonal antibody directed against the α subunit of IGF1R(45). In HER2+ metastatic breast cancer, 54% of 72 tumors showed weak to moderate staining(29) and 50% of 77 tumors demonstrated high IGF1R levels(30).

Compared to these early studies, we observed a lower incidence of positivity possibly due to different detection antibodies, positivity thresholds, and patient population utilized among the various studies. However, our incidence findings are similar to data from more recent retrospective studies in HER2+ breast cancer. Cytoplasmic staining of αIGF1R (clone 24–31) was strong and diffuse (over expression; H score > 220) in 25% of 138 HER2+ breast tumors(31), and high expression of αIGF1R (clone 24–31) was observed in 22% of 37 HER2+ breast tumors(46). Another monoclonal antibody (clone MS-641-P) detected strong (3+) circumferential membrane staining of IGF1R (>10% of tumor cells) in ~22% of HER2+ early-stage breast tumors(32). A study using a rabbit polyclonal antibody (SC713) showed that in 429 patients with stage I–III HER2+ breast cancer, 26% demonstrated IGF1R expression (Allred Score ≥ 7) and only 9% of HER2-enriched tumors had IGF1R expression (Allred score ≥ 7)(47). A study that utilized the same polyclonal antibody as our study (directed against the β subunit of IGF1R) showed that 65% of 297 unselected breast cancers did not have IGF1R protein expression and 3% of the cases had 3+ staining(39). This same antibody detected membrane IGF1R (when exceeded cytoplasmic staining) in 13.8% of 160 HER2+ breast cancers(25). In addition, our results are in alignment with recent findings that demonstrated that IGF1R protein expression was associated with HER2-negativity, and reduced IGF1R expression (compared to moderate expression observed in normal breast epithelial elements) was observed consistently in ER−/PR−/HER2+ breast tumors(38). Further, results from RT-PCR studies demonstrated a relatively low percentage of HER2+ breast cancers with IGF1R mRNA(40).

We observed that IGF1R protein expression was correlated with unfavorable (young age, larger tumors, nodal positivity) and favorable (hormone receptor positivity and corresponding endocrine therapy and low histological grade-well/intermediate differentiated) clinicopathologic characteristics. Although most studies have not found significant associations between IGF1R protein expression and traditional clinicopathologic characteristics(42, 45), our results agree with radioimmunassay findings that demonstrated a positive association between IGF1R activity and positive ER status(48), as well as with IHC findings that demonstrated positive associations between IGF1R protein expression and ER-positivity(38, 40, 47), younger age(25), larger tumor size(25), and lower histopathology grade(25, 40, 47). Our results are comparable to previous findings that demonstrated that HER2+/ER−/PR− tumors consistently showed reduced IGF1R expression(38) and consistent with the findings that IGF1R was frequently downregulated in hormone receptor-negative tumors(49, 50).

The prognostic and predictive role of IGF1R is not clearly defined in the literature. Our carefully conducted study demonstrated a lack of correlation of IGF1R expression with outcome of N9831 patients. The DFS of patients within each treatment arm was not significantly different between patients with IGF1R+ and IGF1R− tumors independent of cut point. Similar to previous findings using radioimmunoassay(42, 45), we found that IGF1R protein expression had no significant impact on prognosis of patients. However, other radioimmunoassay studies have reported IGF1R as a favorable prognostic factor(41, 43) and although IGF1R levels were ten-fold higher in breast cancer than in normal breast, IGF1R levels were higher in a low-risk breast cancer group(44). Recent univariate analyses demonstrated that both mRNA levels and protein expression were associated with favorable prognosis, a longer breast cancer–specific survival (BCSS), but in the HER2 subtype, IGF1R protein was not significantly associated with BCSS or relapse-free survival(39). The prognostic significance of IGF1R expression may be then dependent on the molecular subtype of breast cancer. In ER-negative tumors, IGF1R expression was not predictive of pathological complete response, while in ER+ tumors, reduced IGF1R was associated with pathologic complete response and significant tumor volume reduction(38). In contrast, IGF1R expression (Allred ≥ 7) was associated with better BCSS in Luminal-B patients (ER+/PR+/HER2-/Ki67 hi), but worse outcome in HER2-enriched subtype(46, 47).

While we observed a benefit of concurrent trastuzumab compared to chemotherapy alone, the benefit was independent of tumor IGF1R protein expression. Specifically, our results in the adjuvant setting appear to conflict with preclinical evidence that suggests that IGF1R is involved in trastuzumab resistance, but do agree with the clinical evidence in the metastatic setting that did not demonstrate a clear association between IGF1R expression alone and trastuzumab benefit(29, 30). Although we did not examine pS6 protein expression for which high pS6 and low IGF1R protein expression in combination has been associated with favorable trastuzumab response in the metastatic setting(30), gene expression profiling using the c-DNA-mediated Annealing, Selection, extension, and Ligation (DASL) platform showed a possible association (between S6 gene expression and favorable outcome in Arm C (p=0.013) but not in Arm A using unadjusted Cox Proportional Hazards model; using adjusted Cox Proportional Hazards model, no significant association was observed (data not shown). A recent biomarker analysis from the CLEOPATRA trial for metastatic breast cancer showed a potential association between high IGF1R membrane protein expression and reduced PFS (p = 0.041), but this was not considered a predictive effect due to overlap between CIs and possible impact of multiple testing(51). In addition, low IGF1R cytoplasmic protein expression was possibly associated with reduced PFS and OS, but these associations were not significant(51). Recent conflicting reports exist for adjuvant/neoadjuvant setting. In contrast to our results, a univariate analysis from a small study of 67 patients treated with neoadjuvant or adjuvant trastuzumab demonstrated that cytoplasmic IGF1R overexpression (H score >220) was associated with shorter PFS, but this association was not observed in multivariate analyses or in 75 patients treated for metastatic disease(31). In addition, 10 patients with strong membrane IGF1R expression (3+ > 10% tumor cells) showed a 50% response rate, while 36 patients without strong IGF1R expression showed a 97% response rate to neo-adjuvant trastuzumab and vinorelbine(32). However, similar to our results, a small study that used the same polyclonal βIGF1R antibody as our study, showed no significant association between membrane IGF1R expression and survival of patients with HER2+, early stage breast cancer who were treated with adjuvant trastuzumab(25).

Although patients with and without IGF1R protein overexpression both benefited from concurrent trastuzumab, there was descriptively more benefit (albeit non-significant; p=0.17) in patients with IGF1R overexpression compared to patients without IGF1R overexpression, independent of cut point. In addition, a strong trend (p=0.05) toward greater benefit from concurrent trastuzumab compared to sequential trastuzumab was observed among patients with IGF1R expression (score 1+ – 3+) but not in patients without IGF1R expression (score 0), suggesting possible enhanced synergy of concurrent paclitaxel and trastuzumab on tumor cell growth inhibition in presence of IGF1R. Although a rising trend in benefit from concurrent trastuzumab was observed with increasing IGF1R protein expression, IGF1R protein expression did not significantly improve predictability for adjuvant trastuzumab benefit.

Overall, our N9831 data imply that IGF1R protein expression alone is not significantly associated with differential benefit to concurrent trastuzumab in the adjuvant setting. Previous clinical evidence in the metastatic setting demonstrated that IGF1R in combination with its downstream signaling molecule, pS6, but not IGF1R alone, was associated with trastuzumab response(30). This indicates that IGF1R expression per se does not predict trastuzumab resistance in HER2-overexpressing cancer(27). However taken together with the in vitro preclinical evidence, we can speculate that activation of IGF1R signaling may be more important than its expression for development of resistance to trastuzumab. IGF signaling resulted in elevated p27kip1 ubiquitin ligase SKP2 leading to decreased p27kip1 and loss of growth arrest in the presence of trastuzumab(22, 23), but this would require further study. Therefore, ongoing whole genome expression profiling and pathway analyses of N9831 tumors will provide important information regarding the relationship between IGF1R and other pertinent genes and down-stream signaling molecules in breast tumors with a HER2-enriched subtype and benefit to adjuvant trastuzumab.

STATEMENT OF TRANSLATIONAL RELEVANCE.

Although stringent criteria exist for patient selection, uniform clinical benefit of trastuzumab is not observed in all patients with HER2-overexpressing tumors and resistance remains a significant problem. Identifying markers that could help predict trastuzumab benefit is therefore an important clinical goal. Preclinical evidence indicates that increased IGF1R signaling interferes with the action of trastuzumab suggesting a possible mechanism of trastuzumab resistance. We then investigated the association between IGF1R protein expression using two cut points for positivity (IGF1R+: ≥1+ or ≥2 membrane staining in >0% invasive cells) and disease-free survival (DFS) of patients in N9831. IGF1R expression did not impact DFS within arms. Using either cut point, patients with IGF1R+ and IGF1R− tumors both significantly benefited from concurrent trastuzumab compared to standard chemotherapy alone, and the level of benefit was not significantly different. Our N9831 data indicate that IGF1R protein expression is not significantly associated with differential benefit to concurrent trastuzumab.