Breast Cancer Research and Treatment

, Volume 109, Issue 2, pp 209–229

HER2/neu in systemic therapy for women with breast cancer: a systematic review

  • Bindi Dhesy-Thind
  • Kathleen I. Pritchard
  • Hans Messersmith
  • Frances O’Malley
  • Leela Elavathil
  • Maureen Trudeau
Review

DOI: 10.1007/s10549-007-9656-y

Cite this article as:
Dhesy-Thind, B., Pritchard, K.I., Messersmith, H. et al. Breast Cancer Res Treat (2008) 109: 209. doi:10.1007/s10549-007-9656-y

Abstract

Background

Amplification and/or overexpression of the HER2/neu gene is associated with a poor prognosis in breast cancer. Many studies have suggested that this gene may be associated with the relative efficacy of chemotherapy and endocrine therapy options.

Methods

A systematic review of the evidence was conducted. MEDLINE, EMBASE, the Cochrane Library, the American Society of Clinical Oncology annual meeting proceedings, and the San Antonio Breast Cancer Symposia proceedings were all searched to November 2006 for reports of analysis by HER2/neu status of the relative efficacy of the treatment arms in randomized controlled trials.

Results

Thirty-five trials were identified. A meta-analysis of trials of tamoxifen versus observation found no significant interaction between treatment and HER2/neu status, although one trial not included in the meta-analysis did find interaction. A meta-analysis of adjuvant anthracycline-based chemotherapy trials found a significant interaction (difference in disease-free survival log-hazard ratios −0.31, 95% confidence interval −0.50 to −0.13; difference in overall survival log-hazard ratios −0.34, 95% confidence interval −0.53 to −0.14). Significant interaction was also found in a meta-analysis of disease-free survival in trials of adjuvant taxane therapy versus non-taxane therapy (difference in disease-free survival log-hazard ratios −0.36, 95% confidence interval −0.68 to −0.04). HER2/neu overexpression and/or amplification was associated with greater efficacy of the anthracycline or taxane regimen.

Conclusions

Current evidence supports the conclusion that the benefit of both anthracycline-based and taxane-based adjuvant chemotherapy is associated on HER2/neu status, with patients with HER2/neu-positive cancers benefiting more from these therapies than those with HER2/neu-negative cancers.

Keywords

Breast cancer HER2/neu Systematic review Systemic therapy 

Introduction

Many women diagnosed with breast cancer are treated with adjuvant systemic therapies to reduce the risk of recurrence, therapies that often involve significant short- and long-term toxicities. The use of a predictive tumour marker to select treatment is therefore attractive, as it might spare some patients from receiving ineffective or excessively toxic therapy. The marker HER2/neu may play such a role in predicting the response to specific treatments for women with breast cancer.

The HER2/neu gene encodes for a 185-kd transmembrane glycoprotein, a member of a family of growth-factor receptors with intrinsic tyrosine kinase activity. The amplification of the HER2/neu gene or the overexpression of its protein is observed in 20% to 30% of human breast cancers and is associated with a poor prognosis in patients with primary breast cancer. Amplification and/or overexpression of HER2/neu in breast cancer is associated with a number of adverse prognostic factors, including advanced stage axillary lymph node involvement, absence of estrogen and progesterone receptors, increased S-phase fraction, and high nuclear grade [1, 2]. Many studies suggest that HER2/neu amplification and/or overexpression may be associated with relative sensitivity or resistance to endocrine therapy or chemotherapy. This systematic review was developed to review the evidence for the use of HER2/neu amplification and/or overexpression as a predictive marker to guide the selection of systemic and radiation therapy for patients with breast cancer.

Methods

MEDLINE was searched using a disease-specific medical subject heading (MeSH) term (“breast neoplasms”) and marker-specific MeSH terms (“receptor, erbB-2” OR “genes, erbB-2” OR “oncogene proteins v-erbB”). The Excerpta Medica database (EMBASE) was similarly searched using a disease-specific Excerpta Medica Tree (EMTREE) term (“breast cancer”) and a marker-specific EMTREE term (“oncogene c erb”). The Cochrane Library was searched in a similar fashion. These databases were searched through November 2006. Search terms for the following publication types and study designs were also included in each strategy: practice guidelines, systematic reviews, meta-analyses, reviews, randomized controlled trials, and controlled clinical trials. Due to the large volume of studies on trastuzumab, articles containing this term in the title were excluded from both strategies.

Online conference proceedings from the American Society of Clinical Oncology (http://www.asco.org/ac/1,1003,_12-002095,00.asp; 1999–2006) and the San Antonio Breast Cancer Symposium (http://www.sabcs.org/SymposiumOnline/index.asp#abstracts; 2001–2005) were also searched. The Canadian Medical Association Infobase (http://mdm.ca/cpgsnew/cpgs/index.asp) and the National Guidelines Clearinghouse (http://www.guideline.gov/) were searched for existing evidence-based practice guidelines. Relevant articles and abstracts were selected and reviewed by one reviewer, and the reference lists from these sources were searched for additional trials, as were the reference lists from relevant review articles.

Articles on clinical trials were selected for inclusion in this systematic review of the evidence, if they met the following criteria:
  • The effects of systemic and/or radiation therapy was analyzed according to HER2/neu status in a phase III randomized controlled trial.

  • Reported outcomes included disease-free survival, regression-free survival, time-to-progression, objective response rate, or overall survival.

  • Clinical trial results were reported in full papers or abstracts.

Trials published in a language other than English were excluded.

Where possible, meta-analyses were conducted to create summary estimates of the treatment effects by HER2/neu status. The methods described by Parmar et al. [3] were used to derive the log-hazard ratio and its standard error when appropriate. When these values were derived from survival curves, two independent analysts conducted the derivation from the curve, and their results were averaged. Analysis was conducted using the Review Manager software, version 4.2.7 [4]. A random effects model was used for all summary estimates. With time-to-event outcomes, analysis was conducted using the generic inverse variance method.

In order to test the interaction of treatment and HER2/neu status for time-to-event outcomes where meta-analysis was performed, the difference between the HER2/neu-positive and HER2/neu-negative log hazard ratios was taken for each study and analyzed using the generic inverse variance method with random effects. The standard error of this difference was calculated by taking the square root of the sum of the squares of the standard errors in each HER2/neu subgroup. If the estimate of the difference was found to be significantly different from zero, this was interpreted as evidence of an interaction between treatment and HER2/neu status.

Results

Thirty-five trials [5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61] were identified. Table 1 summarizes the therapy regimens and HER2/neu-status evaluation methods used in the included trials. Tables 24 summarize the trial eligibility criteria and outcomes. One abstract [62] could not be included due to the incomprehensible nature of the report.
Table 1

Summary of therapy regimens evaluated by included trials

Trial

Assessment method

Criteria for positive

Therapy regimen

Austrian Breast and Colorectal Cancer Study Group (ABCSG) 5 [18, 19]

Unstated, likely IHC based on criteria for positive

“Strongly HER2/neu positive (+++)”

Tam + Gos: Tamoxifen (20 mg/day orally for 5 years) plus Gosorelin (3.6 mg q 28days for 3 years)

CMF: Cyclophosphamide (600 mg/m2), methotrexate (40 mg/m2), and 5-fluorouracil (600 mg/m2) days 1 and 8, q 28 days for 6 cycles

Arbeitsgemeinschaft für Gynäkologische Onkologie (AGO) [47, 48]

FISH using Vysis Pathvision HER2/neu and centromeric chromosome 17 probes

≥2 HER2/neu signals per centromeric chromosome 17 signals per nucleus

EC: Epirubicin (60 mg/m2) and cyclophosphamide (600 mg/m2) q 3 weeks for maximum 10 cycles

ET: Epirubicin (60 mg/m2) and paclitaxel (175 mg/m2) q 3 weeks for maximum 10 cycles

Breast Cancer International Research Group (BCIRG) 001 [54]

FISH, specific probes not reported

>2 HER2/neu signals per centromeric chromosome 17 signals per nucleus

DAC: Docetaxel (75 mg/m2), doxorubicin (50 mg/m2), and cyclophosphamide (500 mg/m2) q 3 weeks for 6 cycles

FAC: 5-fluorouracil (500 mg/m2), doxorubicin (50 mg/m2), and cyclophosphamide (500 mg/m2) q 3 weeks for 6 cycles

Belgian [23, 24]a

FISH using Vysis Pathvision HER2/neu and centromeric chromosome 17 probes

>2 HER2/neu signals per centromeric chromosome 17 signals per nucleus

HEC: Epirubicin (100 mg/m2) and cyclophosphamide (830 mg/m2) q 3 weeks for 8 cycles

EC: Epirubicin (60 mg/m2) and cyclophosphamide (500 mg/m2) q 3 weeks for 8 cycles

CMF: Cyclophosphamide (100 mg/mdays 1 through 14), methotrexate (40 mg/m2 on days 1 and 8), and 5-fluorouracil (600 mg/m2 on days 1 and 8) q 4 weeks for 6 cycles

Cancer and Leukemia Group B (CALBG) 9342 [59]

IHC using CB11 antibody and FISH using Vysis Pathvision

Not reported

T175: Paclitaxel (175 mg/m2) q 3 weeks

T210: Paclitaxel (210 mg/m2) q 3 weeks

T250: Paclitaxel (250 mg/m2) q 3 weeks

CALGB 8541/8869 [39]

IHC using CB11 antibody on some specimens and an unstated antibody on othersb

Estimated ≥50% of cells stained

HCAF: Cyclophosphamide (600 mg/m2), doxorubicin (60 mg/m2), on day 1, and 5-fluorouracil (600 mg/m2) on days 1 and 8, q 28 days for 4 cycles

CAF: Cyclophosphamide (400 mg/m2), doxorubicin (40 mg/m2), on day 1, and 5-fluorouracil (400 mg/m2) on days 1 and 8, q 28 days for 6 cycles

LCAF: Cyclophosphamide (300 mg/m2), doxorubicin (30 mg/m2), on day 1, and 5-fluorouracil (300 mg/m2) on days 1 and 8, q 28 days for 4 cycles

CALGB 9344 [55, 56]

IHC using CB11 antibody

≥50% of cells stained

AC→T: Doxorubicin (randomized to 60, 75, or 90 mg/m2) and cyclophosphamide (600 mg/m2) every 3 weeks for 4 cycles followed by paclitaxel (175 mg/m2) every 3 weeks for 4 cycles

AC: Doxorubicin and cyclophosphamide as above

Czech [25]

IHC using CB11 antibody

Not reported

AC: Doxorubicin (60 mg/m2) and cyclophosphamide (600 mg/m2) q 21 days for 4 cycles

CMF: Cyclophosphamide (500 mg/m2), methotrexate (40 mg/m2) and 5-fluorouracil (600 mg/m2) days 1 and 8 q 28 days for 4 cycles

Danish Breast Cancer Cooperative Group (DBCCG) 77c [5]

IHC using Dako A485 antibody

“Distinct membraneous immunostaining”

Tam (1 year): Tamoxifen (10 mg 3 times/day) for 1 year

Observation

DBCCG 89d [26]

IHC using Dako antibody, with FISH using Dako probe on 2+ IHC results

3+ on IHC, or 2+ and two-fold amplification on FISH

CEF: Cyclophosphamide (600 mg/m2), epirubicin (60 mg/m2), and 5-fluorouracil (600 mg/m2) on day 1 q 21 daysc

CMF: Cyclophosphamide (600 mg/m2), methotrexate (40 mg/m2), and 5-fluorouracil (600 mg/m2) on day 1 q 21 daysc

Duke University Medical Center (DUMC) [14, 15]

IHC using 3B5 antibody

2+ or 3+ on 0–3+ scaled

Letrozole: Letrozole (2.5 mg/day) for 4 months

Tamoxifen: Tamoxifen (20 mg/day) for 4 months

Dutch [43, 44, 45]

IHC using 3B5 antibody

3+ on 0–3+ scaled

FEC + HD: As FEC arm, except only 4 cycles, followed by one cycle of cyclophosphamide (6 g/m2), thiotepa (480 mg/m2), and carboplatin (1600 mg/m2)

FEC: 5-fluorouracil (500 mg/m2), epirubicin (90 mg/m2), and cyclophosphamide (500 mg/m2) q 3 weeks for 5 cycles.

European Organization for Research and Treatment of Cancer (EORTC) 10923 [49, 50]

IHC using CB11 antibody

2+ or 3+ on 0–3+ scaled

A: Doxorubicin (75 mg/m2) q 3 weeks for maximum 7 cycles.

T: Paclitaxel (200 mg/m2) q 3 weeks for maximum 7 cycles.

French Adjuvant Study Group (FASG) 05 [40, 41]

IHC using unstated antibody

2+ or 3+ on 0–3+ scaled

FEC100: 5-fluorouracil (500 mg/m2), epirubicin (100 mg/m2), and cyclophosphamide (500 mg/m2) q 3 weeks for 6 cycles.

FEC50: 5-fluorouracil (500 mg/m2), epirubicin (50 mg/m2), and cyclophosphamide (500 mg/m2) q 3 weeks for 6 cycles.

Gruppo Oncologico Italiano di Ricerca Clinica (GOIRC) [27, 28]

IHC using CB11 antibody

>50% stained tumour cells

E: Epirubicin (30 mg/m2) q week for 16 weeks.

CMF: Cyclophosphamide (600 mg/m2), methotrexate (40 mg/m2), and 5-fluorouracil (600 mg/m2) days 1 and 8 q 28 days for 6 cycles.

Gruppo Oncologico Nord Ovest Mammella Intergruppo (GONO-MIG1) [42]

IHC using CB11 antibody

3+ on 0–3+ scaled

FEC14: As FEC21, except q 14 days.

FEC21: 5-fluorouracil (600 mg/m2), epirubicin (60 mg/m2), and cyclophosphamide (600 mg/m2) q 21 days.

Gruppo Universitario Napoletano (GUN) [6, 7]

IHC using MAb-1 antibody

Estimated >10% of cells with membrane staining

Tam (2 years): Tamoxifen (30 mg/day) for 2 years

Observation

GUN-3 [29, 30]

IHC using MAb-1 antibody

Not reported

CMF→EV: Cyclophosphamide (100 mg/m2), methotrexate (40 mg/m2), and 5-fluorouracil (600 mg/m2) for 3 cycles followed by epirubicin (75 mg/m2) and vincristine (1.4 mg/m2 on days 1 and 8) for 3 cycles.

CMF: CMF as above for 6 cycles.

HE10/97 [57]

IHC using Dako antibody

3+ on 0–3+ scaled

E→T→CMF: Epirubicin (110 mg/m2) for 3 cycles followed by paclitaxel (250 mg/m2) for 3 cycles followed by cyclophosphamide (840 mg/m2), methotrexate (57 mg/m2), and 5-fluorouracil (840 mg/m2) (“intensified CMF”) for 3 cycles. All cycles every two weeks.

E→CMF: Epirubicin for 4 cycles followed by CMF for 4 cycles with dosages as above.

All cycles every two weeks.

Immediate Preoperative Anastrozole, Tamoxifen, or Combined with Tamoxifen (IMPACT) trial [16]

IHC using Dako antibody, with FISH using Vysis probe on 2+ IHC results

3+ on IHC, or 2+ and two-fold amplification on FISHe

Anastrozole: Anastrozole (1 mg daily) for 12 weeks before surgery.

Tamoxifen: Tamoxifen (20 mg daily) for 12 weeks before surgery.

An + Tam: As above combined.

Milan [31]

IHC using CB11 antibody

“strong membrane labelling”

CMF→A: As CMF, except for 8 cycles, followed by doxorubicin (75 mg/m2) q 3 weeks for 4 cycles.

CMF: Cyclophosphamide (600 mg/m2), methotrexate (40 mg/m2), and 5-fluorouracil (600 mg/m2) q 3 weeks for 12 cycles.

Nolvadex Adjuvant Trials Organization (NATO)/Cancer Research Campaign (CRC) [8, 9, 10]

IHC using ICR 12 antibody

“any membrane staining”

Tam (2 years): Tamoxifen (10 mg twice a day) for 2 years

Observation

National Cancer Institute of Canada Cancer Trials Group (NCIC CTG) MA.5 [32, 33]

FISH using Vysis Pathvysion

≥2 HER2/neu signals per centromeric chromosome 17 signals per nucleus

CEF: Cyclophosphamide (75 mg/m2 orally days 1 through 14), epirubicin (60 mg/m2 days 1 and 8), and 5-fluorouracil (500 mg/mdays 1 and 8) q month for 6 months.

CMF: Cyclophosphamide (100 mg/m2 orally days 1 through 14), methotrexate (40 mg/mdays 1 and 8), and 5-fluorouracil (600 mg/mdays 1 and 8) q month for 6 months.

National Surgical Adjuvant Breast and Bowel Project (NSABP) B-11 [34, 35]

IHC using both mAb-1 and pAb-1 antibodies

“any tumour cell showed definite membrane staining resulting in a so-called fishnet appearance”

PAF: As PF, with doxorubicin (30 mg/m2 on days 1 and 21).

PF: Melphalan (4 mg/mdays 1 through 5) and 5-fluorouracil (300 mg/mdays 1 through 5) q 6 weeks for 17 cycles.

NSABP B-15 [36, 37]

IHC using both TAB250/mAb-1 and pAb-1 antibodies

“any tumour cell showed definite membrane staining resulting in a so-called fishnet appearance”

AC: A (60 mg/m2) and C (600 mg/m2) q 21 days for 4 cycles.

CMF: C (100 mg/m2 orally days 1 through 14), M (40 mg/mdays 1 and 8), and F (600 mg/mdays 1 and 8) q month for 6 monthsf

Pennsylvania State University Hershey Medical Center (PSUHMC) [17]

Immunoassay for pre-treatment serum HER2/neu ECD (Immuno 1, Bayer Corporation)

>15 ng/ml of serum HER2/neu ECDd

Letrozole: Letrozole (2.5 mg/day).

Tamoxifen: Tamoxifen (20 mg/day).

Swedish Breast Cancer Cooperative Group (SBCCG) [12, 13]

IHC using Dako antibody

2+ or 3+ on 0–3+ scaled

Tam (5 years): Tamoxifen (20 or 40 mg/day, depending on center) for 5 years.

Tam (2 years): Tamoxifen (20 or 40 mg/day, depending on center) for 2 years.

Scandinavian Breast Group (SBG) [51, 52]

IHC using Dako antibody

2+ or 3+ on 0–3+ scaled

D: Docetaxel (100 mg/m2) q 3 weeks until progression or toxicity.

MF: Methotrexate (200 mg/m2) and 5-fluorouracil (600 mg/m2) q 3 weeks until progression or toxicity.

Southern Italy Cooperative Oncology Group (SICOG) [60, 61]

Not reported

Not reported

CisET: Cisplatin (30 mg/m2), epirubicin (50 mg/m2), and paclitaxel (120 mg/m2) with G-CSF support every week for 12 weeks.

ET: Epirubicin (90 mg/m2) and paclitaxel (175 mg/m3) every three weeks for 4 cycles.

Spanish [38]

IHC using CB11 antibody

“tumours expressing HER2 in 50% or more of...cells”

FAC: 5-fluorouracil (500 mg/m2), doxorubicin (50 mg/m2), and cyclophosphamide (500 mg/m2) q 3 weeks for 6 cycles.

CMF: Cyclophosphamide (600 mg/m2), methotrexate (60 mg/m2), and 5-fluorouracil (600 mg/m2) q 3 weeks for 6 cycles.

Swedish [11]

IHC using CB11 antibody

3+ on 0–3+ scaled

Tam (2 years): Tam (dose not stated) for 2 years.

Observation

Southwest Oncology Group (SWOG) S8814 [22]

IHC using MAb1 and CB11 (only CB11 results reported here)

Not reported

Tam: Tamoxifen (20 mg/d) for 5 years.

CAF + Tam: Cyclophosphamide (100 mg/m2 orally days 1 through 14), doxorubicin (30 mg/mdays 1 and 8), and 5-fluorouracil (500 mg/mdays 1 and 8) q 29 days for 6 cycles, plus Tam as above.

TAX303 [53]

FISH using Vysis probes on specimens with >1% staining on IHC using CB11

Ratio of HER2 signals to centromeric 17 signals >2

A: Doxorubicin (75 mg/m2 q 3 weeks)

D: Docetaxel (100 mg/m2 q 3 weeks)

University of Texas [58]

IHC using TAB250, with confirmation of borderline specimens by FISH

Not stated

AC: Doxorubicin (60 mg/m2) and cyclophosphamide (600 mg/m2) q 21 days for 4 cyclesg

AC→D: As above, plus docetaxel (100 mg/m2) q 27 days for 4 cyclesg

Vietnamese [20, 21]

IHC using both TAB250 and Dako antibodies

Scoring used by Chang et al. [75]

Ooph + Tam: Oophorectomy plus tamoxifen (20 mg/day) for 5 years.

Obs + Therapy: Observation, followed by above if recurrence.

a The study was conducted using both IHC and FISH to determine HER2/neu status. Only the FISH results are reported here. Details on schedule and dosages of treatment arms taken from initial study report [76]

b Two different sets of patients, separated in time, are included in this study. CB11 was used in the second set because the antibody used with the first set was no longer available

c Patients were also randomized to pamidronate (150 mg twice daily for 4 years) or observation in 2 x 2 factorial design

d With some variation between studies, this scale was as follows: 0, completely negative; 1+ faint perceptible staining; 2+, moderate staining of entire membrane in >10% of tumour cells, 3+, strong circumferential staining of entire membrane. The variation between studies in the exact definition of each score is considered minimal compared to the variation between antibodies used and completely different scoring systems

e Cut-off derived from the mean plus two standard deviations of serum HER2/neu ECD in 245 healthy women

f Study had three arms: AC x4, CMF x 6, or AC→CMF. Only the AC and CMF arms are compared by HER2/neu status

g Given as neoadjuvant therapy. Roughly half of patients in AC arm received docetaxel as the AC→Docetaxel arm but received the docetaxel in the adjuvant, not neoadjuvant, setting, after the response measures reported in this document were collected

Abbreviations: ECD, extracellular domain; FISH, fluorescence in situ hybridization; G-CSF, granulocytic-colony stimulating factor; IHC, immunohistochemistry; Ooph, oophorectomy; q, every

Table 2

Trials of endocrine therapy by HER2/neu status

Trial

nHER/ntotal

Population

Median f/u (years)

Arms

Outcome

Overall

HER2+

HER2−

Int effect

Tamoxifen

DBCCG 77c [5]

1057a/1716

Adjuvant, Postmen., high-risk

NR

Tam (1 year)

DFS

HR = Sigb

HR = 0.89

HR = 0.86

P = NS

Observation

GUN [6, 7]

358c/433

Adjuvant, Stage I–III

15d

Tam (2 years)

OS

HR = 0.68

HR = 1.09

HR = 0.59

P = 0.04

Observation

DFSe

NR

SIR = 1.22

SIR = 0.80

P = 0.03

SIR = 0.86

SIR = 1.21

NATO/CRCf [8, 9, 10]

813/2068

Adjuvant, Age < 75

20/16f

Tam (2 years)

RFS

RR = 0.76

RR = 1.14

RR = 0.70

P = 0.18

CRC: T1-2, N0-1, M0

Observation

NATO: Premen., N+ or Postmen

Swedish [11]

428/564

Adjuvant, Premen.

14

Tam (2 years)

RFS

HR = 0.77

HR = 0.38g

HR = 0.69g

P = NS

Stage II

Observation

SBCCG [12, 13]

577/3545

Adjuvant, Postmen.

11

Tam (5 years)

DFS

HR = 0.82

RRR = 1.1

RRR = 0.62

P = 0.25

Tam (2 years)

Aromatase inhibitors

DUMC [14, 15]

278/337

Neoadjuvant setting.

NR

Letrozole

ORR

60%

69%

53%

NR

Postmen., ER+ and/or PgR+

Tamoxifen

41%

17%h

40%h

IMPACT [16]

239/330

Neoadjuvant setting, Postmen., ER+

NR

Anastrozole

ORR

37%

58%

NR

NR

Tamoxifen

36%

22%

An. + Tam.

39%

31%

PSUHMC [17] [abstract]

562/907

Recurrent or metastatic, ER+ and/or PgR+

NR

Letrozole

TTP

HR = 0.73

HR = 0.73

HR = 0.70

NR

Tamoxifen

ORR

32%

17%

39%

NR

21%

13%

26%

Ovarian Ablation

ABCSG 5 [18, 19]

572/1034

Adjuvant, Premen., ER or PgR pos.

5

Tam + Gos

OS

RR = 0.79

NR

NR

NS

CMF

DFS

RR = 0.71

NR

NR

NS

Vietnamese [20, 21]

282/709

Adjuvant, Premenopausal, ER+, operable

3.7

Ooph + Tam

OS

RR = 0.67j,

RR = 0.26

RR = 0.68

P = 0.07

Obs + Therapy

DFS

RR = 0.58j

RR = 0.37

RR = 0.48

P = 0.18

Chemoendocrine therapy

SWOG S8814e [22]

595/1477

ER-positive, node-positive

NR

Tam

4-year DFS

HR = 1.43

56%

82%

P = 0.21

CAF + Tam

75%

84%

NOTE: Items in bold are statistically significant comparisons at the α = 0.05 level. All ratios are presented with values <1 favouring the first arm. Treatment arms and trial abbreviations are fully described in Table 1

a Steroid receptor positive

b Overall DFS benefit with tamoxifen reported (P = 0.02) but data not shown

c Only 245 patients were included in the DFS subgroup analysis

d DFS analysis performed after median f/u of ∼14 years

e HER2/neu analysis in abstract form

f Joint analysis of patients from two separate trials, NATO and CRC Adjuvant Breast Trial. Median follow-up listed as NATO/CRC

g ER+ patients only

h HER2/neu+ vs. HER2/neu- difference significant for tamoxifen arm

j From original report [20] including all patients

Abbreviations: DFS, disease-free survival; ER, estrogen receptor; f/u, follow-up; HR, hazard ratio; Int, interaction; nHER, number of patients with HER2/neu status known included in analysis; ntotal, total number of patients in study; NR, not reported; OS, overall survival; PgR, progesterone receptor; Postmen., postmenopausal; Premen., premenopausal; RFS, recurrence-free survival; RR, relative risk; RRR, recurrence rate ratio; Sig, significant; SIR, standardized incidence ratio; TTP, time-to-progression; vs., versus

Four trials [5, 6, 7, 8, 9, 10, 11] of tamoxifen versus observation were identified. Additionally, one trial [12, 13] compared five years of tamoxifen to two years (Table 2). Only the Gruppo Universitario Napoletano (GUN) trial [6, 7] found significant interaction between the treatment arm (tamoxifen versus observation) and HER2/neu status for both overall survival (P = 0.04) and disease-free survival (P = 0.03), with a greater benefit with tamoxifen reported in patients with HER2/neu-negative cancers. Three studies [5, 8, 9, 10, 11] provided sufficient information to be included in a meta-analysis of disease-free survival, but only in patients with estrogen- or estrogen- and/or progesterone-receptor positive cancers (Fig. 1). In the meta-analysis, a significant benefit for disease-free survival for tamoxifen compared to observation was found in patients with HER2/neu-negative cancers (hazard ratio 0.79, 95% confidence interval 0.69 to 0.92), with some statistical heterogeneity (χ2P = 0.17, I2 = 44.3%). No benefit was identified in patients with HER2/neu-positive cancers (hazard ratio 0.91, 95% confidence interval 0.68 to 1.23), with no statistical heterogeneity (I2 = 0%). The difference in log-hazard ratios for disease-free survival was not found to be significant (0.11, 95% confidence interval −0.20 to 0.42), with no statistical heterogeneity (I2 = 0%).
Fig. 1

Meta-analysis of disease-free survival hazard ratios (HR) in trials comparing tamoxifen vs. observation by HER2/neu status. Note: Trial abbreviations are fully described in Table 1. References: DBCCG 77c [5], NATO/CRC [8, 9, 10], Swedish [11]. Abbreviations: CI, confidence interval; HR, hazard ratio; SE, standard error; vs., versus

Three trials of aromatase inhibitors compared with tamoxifen were identified; two [14, 15, 16] in the neoadjuvant setting and one [17] in the metastatic setting (Table 2). None of these trials reported any significant interaction between HER2/neu status and treatment arm for any outcome. The two neoadjuvant trials [14, 15, 16] were considered suitable for meta-analysis, as the objective response rate was reported both for patients with HER2/neu-positive and those with HER2/neu-negative cancers. The pooled odds ratio for objective response among patients with HER2/neu-positive cancers was 7.86 (95% confidence interval 2.38 to 25.92), with the value over one indicating a greater response among those treated with aromatase inhibitors over those treated with tamoxifen. Among patients with HER2/neu-negative cancers, the pooled odds ratio for objective response was 1.19 (95% confidence interval 0.58 to 2.45).

Two trials [18, 19, 20, 21] were identified that evaluated some form of ovarian ablation by HER2/neu status (Table 2). Neither trial reported significant interaction between HER2/neu status and treatment arm for any outcome. As the two studies identified were considered clinically heterogeneous, no meta-analysis was attempted.

One trial [22] was identified that analysed the efficacy of tamoxifen and chemotherapy compared with tamoxifen alone according to HER2/neu status (Table 2) and reported no significant interaction between treatment and HER2/neu status.

Ten trials [23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38] of adjuvant anthracycline-based regimens versus non-anthracycline-based regimens were identified (Table 3). Seven of these trials [23, 24, 25, 26, 27, 28, 31, 36, 37, 38] reported no significant interaction between HER2/neu status and treatment arm for any outcome. The National Surgical Adjuvant Breast and Bowel Project (NSABP) B-11 trial found a significant benefit in patients with HER2/neu-positive cancer for the inclusion of doxorubicin in a melphalan and 5-fluorouracil regimen, in terms of overall survival (relative risk 0.66, P = 0.01) and disease-free survival (relative risk 0.60, P = 0.001). No significant benefit was found in patients with HER2/neu-negative cancers. A significant interaction was found between HER2/neu status and treatment arm (doxorubicin, melphalan, and 5-fluorouracil versus melphalan and 5-fluorouracil) for disease-free survival (P = 0.02). The GUN-3 trial [29, 30] found a significant (P = 0.05) interaction between HER2/neu status and treatment arm (cyclophosphamide, methotrexate, and 5-fluorouracil plus epirubicin and vincristine vs. cyclophosphamide, methotrexate, and 5-fluorouracil alone) for overall survival, with patients with HER2/neu-positive cancer benefiting more from the anthracycline-containing regimen than those with HER2/neu-negative cancer. The National Cancer Institute of Canada (NCIC) Cancer Trials Group (CTG) MA.5 trial [32, 33] found a significant benefit for cyclophosphamide, epirubicin, and 5-fluorouracil in patients with HER2/neu-positive cancer for relapse-free survival (HR 0.52, P = 0.003), but no significant benefit in patients with HER2/neu-negative cancer. A significant interaction was found between HER2/neu status and treatment arm (cyclophosphamide, epirubicin, and 5-fluorouracil vs. cyclophosphamide, methotrexate, and 5-fluorouracil) for relapse-free survival (P = 0.02), but not for overall survival.
Table 3

Trials of anthracycline-based chemotherapy by HER2/neu status

Trial

nHER/ntotal

Population

Arms

Median f/u (years)

Outcome

Overall

HER2+

HER2−

Int effect

Anthracycline-based regimen vs. non-anthracycline regimen

Belgiana [23, 24]

354/777

Adjvuant, N+

HEC

6.2

EFS

HR = 0.93*

HR = 0.70*

HR = 1.19*

P = 0.57

CMF

EC

6.2

EFS

HR = 1.19*

HR = 0.61*

HR = 1.52*

P = 0.18

CMF

Czechb [25]

62/106

Adjuvant, pT1c-3a, pN0-1

AC

3.8

OS

ND

NR

NR

NS

CMF

DFS

ND

NR

NR

NS

DBCCG 89d [26]

805/1195

See notec

CEF

∼10

OS

HR = 0.776

HR = 0.725

HR = 0.818

P = 0.63

CMF

∼8

DFS

HR = 0.781

HR = 0.747

HR = 0.789

P = 0.81

GOIRC [27, 28]

266/348

Adjuvant, N− if ER−, or N+ with ≤ 9 positive nodes

E

CMF

8

OS

HR = 1.14

75.8%d

84.5%d

P = 0.311

67.6%d

87.4%d

RFS

HR = 1.11

65.9%d

60.1%d

P = 0.6628

70.3%d

68.6%d

GUN-3b [29, 30]

123/220

Adjuvant, Stage II N+ or Stage III

CMF→EV

10

OS

HR = 0.87

HR = 0.85

HR = 1.64

P = 0.05

CMF

Milan [31]

506

Adjuvant, 1–3 nodes positive

CMF→A

15

OS

HR = 1.22

HR = 0.61

HR = 1.26

P = 0.052

CMF

DFS

HR = 1.26

HR = 0.83

HR = 1.22

P = 0.251

NCIC CTG MA.5 [32, 33]

628/710

Adjuvantc

CEF

CMF

10

OS

77%e,f

HR = 0.65

HR = 1.06

P = 0.07

70%e,f

DFS

10

HR = 0.52

HR = 0.91

P = 0.02

NSABP B-11 [34, 35]

638/707

Adjuvant, N+, ER− and/or PgR−

PAF

PF

13.5g

OS

65%e

RR = 0.66

RR = 0.90

P = 0.15

59%e

DFS

51%e

RR = 0.60

RR = 0.96

P = 0.02

44%e

NSABP B-15 [36, 37]

1355/2194

Adjuvant, N+, TAM nonresponsive

AC

CMF

12.4g

OS

83%

RR = 0.82

RR = 1.07

P = 0.11

82%

DFS

62%

RR = 0.84

RR = 1.02

P = 0.19

63%

Spanishb [38]

141/989

Adjuvant, Stage I–III

FAC

7

OS

NR

72%e

82%e

NR

CMF

42%e,h

84%e,h

More versus less intense anthracycline-based regimen

CALGB 8541/8869 [39]

992/1549

Adjuvant, Stage II

HCAF

CAF

9.3

OS

Significantly improved with HCAF or CAF

87%e

77%e

< 0.001

66%e

82%e

63%e

78%e

LCAF

DFS

Significantly improved with HCAF or CAF

71%e

65%e

P = 0.001

52%e

66%e

50%e

60%e

FASG 05b [40, 41]

332/565

Adjuvant, Node-positive

FEC100

10.1

10-year

RRR = 24%

55.4%

38.9%

NR

FEC50

DFS

37.1%

35.8%

GONO-MIG1b [42]

731/1214

Adjuvant, high risk node-negative or node-positive

FEC14

6.7

OS

HR = 0.65

HR = 0.59

HR = 0.79

P = 0.38

FEC21

EFS

NR

HR = 0.54

HR = 0.91

P = 0.12

Belgian [23, 24]a

354/777

Adjuvant, N+

HEC

6.2

EFS

HR = 0.78*

HR = 1.08*

HR = 0.75*

P = 0.53

EC

Anthracycline-based regimen plus high dose chemotherapy vs. anthracycline-based regimen alone

Dutch [43, 44, 45, 46]

801/885

Operated breast cancer w/ >4 positive nodes

FEC + HD

7.25

OS

HR = 0.85

NR

HR = 0.72

NR

FEC

RFS

HR = 0.84

HR = 1.26

HR = 0.68

P = 0.006

NOTE: Items in bold are statistically significant comparisons at the α = 0.05 level. All ratios are presented with values <1 favouring the first arm. In some cases, marked with an *, the inverse of the reported ratio is presented here for comparability. Treatment arms and trial abbreviations are fully described in Table 1. with an * are reported to date only in abstract form

a Based on FISH analysis

b HER2/neu analysis in abstract form

c (Premen., N−, grade 2 or 3 tumours ≤5 cm) or (premen., recept- or unknown, with >5 cm or N+) or (postmen., recept-, >5 cm or N+)

d At eight years

e At five years

f Includes all subjects from original report [32]

g Average time on study

h HER2/neu+ vs. HER2/neu- difference significant for CMF arm

Abbreviations: DFS, disease-free survival; ER−, estrogen receptor negative; f/u, follow-up; FISH, fluorescence in-situ hybridization; HR, hazard ratio; IHC, immunohistochemistry; Int, interaction; nHER, number of patients with HER2/neu status known included in analysis; ntotal, total number of patients in study; ND, no difference; NR, not reported; NS, not significant; OS, overall survival; PgR−, progesterone receptor negative; Premen., premenopausal; recept., receptor; RFS, recurrence-free survival; RR, relative risk; RRR, relative risk reduction; vs., versus

Six of the anthracycline versus no anthracycline trials [23, 24, 26, 31, 32, 33, 34, 35, 36, 37] provided sufficient information to be included in a meta-analysis for disease-free survival [Fig. 2(a)]. Five of those studies [26, 31, 32, 33, 34, 35, 36, 37] plus two additional studies [27, 28, 29, 30] provided sufficient information to be included in a meta-analysis of overall survival (Fig. 2b). The inclusion of the Belgian trial [23, 24] was complicated by the fact that the trial evaluated two separate dosage levels of epirubicin and cyclophosphamide and also used three separate methods of evaluating HER2/neu status (Table 1). Therefore, six different sets of hazard ratios from this study could have been selected for inclusion. The results in Figs. 2 and 3 reflect the inclusion of the lower-dose epirubicin and cyclophosphamide versus cyclophosphamide, methotrexate, and 5-fluorouracil comparison, with HER2/neu status established by fluorescence in situ hybridization.
Fig. 2

Meta-analysis of hazard ratios (HR) in trials comparing anthracycline vs. non-anthracycline-based regimens by HER2/neu status. (a) Disease-free survival, (b) Overall survival, Note: Trial abbreviations are fully described in Table 1. References: Belgian [23, 24], DBCCG 89d [26], GOIRC [27, 28], GUN-3 [29, 30], Milan [31], MA.5 [32, 33], NSABP B-11 [34, 35], NSABP B-15 [36, 37]. Abbreviations: CI, confidence interval; HR, hazard ratio; SE, standard error; vs., versus

Fig. 3

Meta-analysis of disease-free survival hazard ratios (HR) in trials comparing different anthracycline-based regimens by HER2/neu status. Note: Trial abbreviations are fully described in Table 1. References: Belgian [23, 24], CALGB [39], GONO-MIG [42]. Abbreviations: CI, confidence interval; HR, hazard ratio; SE, standard error; vs., versus

Based on the meta-analysis, there is a significant treatment benefit for anthracycline-based regimens in patients with HER2/neu-positive cancer for both overall survival (hazard ratio 0.73, 95% confidence interval 0.62 to 0.86) and disease-free survival (hazard ratio 0.71, 95% confidence interval 0.60 to 0.83), with little statistical heterogeneity for either estimate (I2 < 8.9%). There is no significant benefit in patients with HER2/neu-negative cancers (overall survival hazard ratio 1.04, disease-free survival hazard ratio 1.00). The difference in log-hazard ratios between HER2/neu subgroups was −0.31 (95% confidence interval −0.50 to −0.13) for disease-free survival and −0.34 (95% confidence interval −0.53 to −0.14) for overall survival. There was no statistical heterogeneity in the difference estimates (I2 = 0%). A sensitivity analysis was conducted to determine the effect of the choice of estimates from the Di Leo study. The estimated hazard ratios and their confidence intervals differed by no more than 0.03, and the difference of log-hazard ratios was statistically significant in all cases, regardless of the estimates included from that study.

Four trials [23, 24, 39, 40, 41, 42] analysed the effect of more versus less intense adjuvant anthracycline-based regimens according to HER2/neu status (Table 3). In this case, more intense refers to either more cycles of anthracycline, higher doses of anthracycline, or the same dose given over a shorter period. Two of these trials [23, 24, 42] reported no significant interaction between HER2/neu status and treatment arm for any outcome, and one trial [40, 41] did not report a significance test for interaction. The Cancer and Leukemia Group B (CALGB) 8541/8869 trials [39] found a significant interaction between HER2/neu status and treatment arm (three dose levels [high, medium, low] of cyclophosphamide, doxorubicin, and 5-fluorouracil) for both overall survival (P < 0.001) and disease-free survival (P = 0.001). Patients with HER2/neu-positive cancer benefited from the higher dose regimen similarly to those with HER2/neu negative cancer but had worse outcomes on the medium and lower dose regimens. It should be noted that the dose used in the higher dose arm in this trial is now generally considered the standard dose, while the medium and lower dose arm doses are no longer common. More recently another subgroup analysis [63] was reported of 523 patients who had been included in the subgroup analysis above. Fluorescence in situ hybridization and polymerase chain reaction were used to establish the HER2/neu status of the tumours of these patients, in addition to the immunohistochemistry analysis performed as part of the first analysis. A significant interaction between HER2/neu status and treatment arm for disease-free survival was present regardless of the method used: fluorescence in situ hybridization (P = 0.033), immunohistochemistry (P = 0.0003), or polymerase chain reaction (P = 0.043). Those patients whose cancer were considered HER2/neu-positive by both fluorescence in situ hybridization and immunohistochemistry experienced the largest disease-free survival benefit from the higher dose arm, with hazard ratios and confidence intervals as follows: fluorescence in situ hybridization positive/immunohistochemistry positive, 3.42 (1.43–8.19); fluorescence in situ hybridization positive/immunohistochemistry negative, 0.96 (0.23–3.97); fluorescence in situ hybridization negative/immunohistochemistry positive, 1.70 (0.63–4.60); and fluorescence in situ hybridization negative/immunohistochemistry negative, 1.02 (0.74–1.40).

Three of the identified studies [23, 24, 39, 42] comparing more intense anthracycline-based regimens to less intense ones provided sufficient information to be included in a meta-analysis of disease-free survival (Fig. 3). In the case of the CALGB 8541/8869 trial [39], the high-dose versus medium-dose arm comparison was included in the analysis. In the case of the Belgian trial [23, 24], the results based on fluorescence in situ hybridization analysis of HER2/neu status were used. There was no statistical heterogeneity in the HER2/neu-positive estimate (χ2 test for homogeneity P = 0.41, I2 = 0%), and some in the HER2/neu-negative estimate (χ2 test for homogeneity P = 0.27, I2 = 24.5%). Patients with HER2/neu-positive cancer had a significant disease-free survival benefit (hazard ratio 0.54, 95% confidence interval 0.38 to 0.79), while those with HER2/neu-negative cancers did not (hazard ratio 0.98). The difference in log hazard ratios for disease-free survival was not found to be significant at −0.51 (95% confidence interval −1.12 to 0.09), with some statistical heterogeneity (I2 43.8%, χ2 test of homogeneity P = 0.17).

No sensitivity analysis was conducted on the choice of arms in the CALGB trial, as there was little difference in disease-free survival between the moderate- and low-dose arms. A sensitivity analysis was conducted regarding the choice of HER2/neu-testing method used to generate the estimates for the Belgian trial. When the Belgian trial results based on immunohistochemistry using CB-11 and 4D5 antibodies were included, instead of fluorescence in situ hybridization, the difference in log-hazard ratios for disease-free survival was significant at −0.72 (95% confidence interval −1.15 to −0.30). The results based on immunohistochemistry using the cocktail method were of less magnitude than the fluorescence in situ hybridization results (difference in log-hazard ratios −0.44, 95% confidence interval −1.11 to 0.23).

Eight trials analysed the effect of taxane-containing regimens versus non-taxane regimens according to HER2/neu status (Table 4). Four of these trials [47, 48, 49, 50, 51, 52, 53] were in the metastatic setting, one [58] was in the neoadjuvant setting, and three [54, 55, 56, 57] were in the adjuvant setting. Three of these trials [47, 48, 54, 57] reported no significance interaction between HER2/neu status and treatment arm for any outcome, and three [49, 50, 51, 52, 58] did not report an interaction test. A retrospective analysis of the TAX303 trial [53] found a significant interaction between HER2/neu status and treatment arm (single-agent docetaxel to single-agent doxorubicin) for objective response rate (P = 0.03); patients with HER2/neu-positive cancer who received docetaxel experienced a significantly higher objective response rate than those on doxorubicin, while there was no difference among those with HER2/neu-negative cancer. The CALGB 9344 [55, 56] trial reported a significant interaction between HER2/neu status and treatment arm (doxorubicin and cyclophosphamide followed by paclitaxel or not followed by paclitaxel), with patients with HER2/neu-positive cancer experiencing greater benefit from the inclusion of paclitaxel. This association was independent of the method used to assess HER2/neu status.
Table 4

Trials comparing the efficacy of taxane-containing regimens compared with non-taxane containing regimens by HER2/neu status

Trial

nHER/ntotal

Population

Median f/u (years)

Arms

Outcome

Overall

HER2+

HER2−

Int effect

Metastatic taxane vs. non-taxane

AGO [47, 48]

275/560

Progressive metastatic cancer with no prior chemotherapy

NR

ET

OS

NR

21.4 ma

27.5 ma

P = 0.14

EC

16.4 ma,b

33.1 ma,b

PFS

NR

10.5 ma

9.6 ma

P = 0.11

7.1 ma,b

10.4 ma,b

ORR

46%

76%c

50%c

P = 0.26

41%

46%

33%

EORTC 10923 [49, 50]

114/331

Progressive metastatic cancer with no prior chemotherapy

NR

T

ORR

24%

24%

24%

NR

A

40%

39%

41%

SBG [51, 52]

131/283

Progressive metastatic cancer with anthracycline failure

NR

D

OS/TTP

ND

ND

ND

NR

MF

ORR

53%

53%

53%

NR

24%

33%

18%

TAX303 [53]

176/326

Metastatic breast cancer, previously treated with adj. or metastatic CMF

1.9

D

A

OS

15 ma

HR = 0.68*

HR = 1.56*

P = 0.10

14 ma

TTP

NR

HR = 1.14*

HR = 1.30*

P = 0.62

ORR

47.8%

67%

40%

P = 0.03

33.3%

27%

35%

(Neo)adjuvant taxane vs. non-taxane

BCIRG 001 [54]

1250/1491

Adjuvant, at least one node positive

4.6

DAC

DFS

HR = 0.72

HR = 0.60

HR = 0.76

P = 0.41

FAC

CALGB 9344 [55, 56]d

1322/3121

Adjuvant, node positive

∼8

AC→T

OS

P = 0.001e

NR

NR

P = 0.0056

AC

DFS

P < 0.001e

22%f

2%f

P = 0.0093

HE10/97 [57]

394/595

Adjuvant, (T1-3, N1, M0) or (T3, N0, M0)

4.3

E→T→CMF

OS

NR

NR

 

P = 0.73

E→CMF

DFS

NR

HR = 1.52gHR = 1.9g

 

P = 0.57

University of Texas [58]

104/144

Neoadjuvant, (T1C-T3, N0, M0) or (T1-3, N1, M0)

NR

AC→D

ORR

82%

78%

81%

NR

AC

55%

75%

51%

Comparison of different taxane regimens

CALBG 9342 [59]d

175/474

Stage IV

NR

T175

OS

NR

ND

ND

NR

T210

T250

SICOG [60, 61]d

179/200

Induction chemotherapy, locally advanced breast cancer

3.75

CisET

ET

5-year DFS

27%

18%

29%

NR

21%

11%

25%

pCR

16%

31%

10%

NR

6%

5%

6%

NOTE: Items in bold are statistically significant comparisons at the α = 0.05 level. All ratios are presented with values <1 favouring the first arm. In some cases, marked with an *, the inverse of the reported ratio is presented here for comparability. Treatment arms and trial abbreviations are fully described in Table 1

a Median in months

B Difference between HER2/neu-negative and HER2/neu-positive significant for EC arm

c HER2/neu+ vs. HER2/neu- difference significant for ET arm. Difference between arms significant regardless of HER2/neu status

d HER2/neu analysis in abstract form

e AC→T superior

f Difference in 5-year DFS rates, AC>T−AC

g Hazard ratio of HER2/neu+ vs. HER2/neu- DFS by treatment arm

Abbreviations: f/u, follow-up; HR, hazard ratio, first listed arm versus second listed arm; Int, interaction; m, months; nHER, number of patients with HER2/neu status known included in analysis; ntotal, total number of patients in study; ND, no difference; NR, not reported; ORR, objective response rate; OS, overall survival; PFS, progression-free survival; vs., versus

No meta-analysis was performed of the trials in the metastatic setting, as either the taxane-containing and non-taxane arms were not comparable between the trials or the trials did not provide sufficient information. The three trials [54, 55, 56, 57] of taxane- versus non-taxane-containing regimens in the adjuvant setting did provide sufficient information to conduct a meta-analysis (Fig. 4). There was a significant disease-free survival benefit for taxane-based regimens compared to non-taxane regimens in patients with both HER2/neu-positive (hazard ratio 0.60, 95% confidence interval 0.46 to 0.78) and HER2/neu-negative cancer (hazard ratio 0.83, 95% confidence interval 0.71 to 0.98). There was little statistical heterogeneity in either estimate (I2 < 3%). The difference in log hazard ratios for disease-free survival between HER2/neu subgroups was significant at −0.36 (95% confidence interval −0.68 to −0.04), with no statistical heterogeneity (I2 = 0%).
Fig. 4

Meta-analysis of disease-free survival hazard ratios (HR) in trials comparing taxane-containing to non-taxane containing regimens by HER2/neu status. Note: Trial abbreviations are fully described in Table 1. References: BCIRG 001 [54], CALGB 9344 [55, 56], HE10/97 [57]. Abbreviations: CI, confidence interval; HR, hazard ratio; SE, standard error; vs., versus

One trial [59] analysed the efficacy of different taxane doses of single-agent paclitaxel in the metastatic setting, and another [60, 61] compared cisplatin, epirubicin, and paclitaxel to epirubicin and paclitaxel. Neither trial reported a test of interaction.

A search was made of the National Cancer Institute clinical trials database on the Internet (http://www.cancer.gov/search/clinical_trials/) for ongoing clinical trials that mention HER2/neu status in their titles. None were identified as appropriate to the questions posed in this review. However, few of the trials included in this review were designed primarily to investigate the role of HER2/neu status; the HER2/neu status investigation was a substudy or additional analysis of an existing trial. Therefore, it is certain that there are ongoing and completed trials that will analyze the effect of HER2/neu status that cannot be identified prior to the publication.

Discussion

The identified trials consist primarily of reanalyses of completed trials using stored tissue specimens for the determination of HER2/neu status. In most trials, these samples were not available for all the original patients, or only a subsample of patient specimens were used in the analysis. Therefore, the analysis could well be biased if the availability of tissue samples is correlated in some way with treatment or outcome.

The identified trials used a range of HER2/neu status assessment methods (Table 1). There is evidence of variability between observers [64, 65] and laboratories [66, 67] in the implementation and scoring of immunohistochemistry for HER2/neu overexpression, even when the stated method is identical. This variation in assessment of status likely reduces the comparability of the studies.

The statistical power necessary to detect an interaction between two variables is greater than that necessary to detect the effect of each variable individually. In addition, as none of the identified trials were designed with this interaction as a primary or secondary outcome of interest, their sample sizes were not calculated with these interaction tests in mind. Therefore, in trials where no significant interaction was detected, the magnitude of the outcomes by HER2/neu status and treatment should be considered when determining whether there is no clinically meaningful significance or whether the trial was underpowered for this purpose.

Of the identified trials of tamoxifen versus observation, only the GUN trial [6, 7] found a significant interaction between HER2/neu status and tamoxifen therapy versus observation. A meta-analysis of three of these trials did not find a significant difference in disease-free survival between patients with HER2/neu-positive and HER2/neu-negative cancers, but this analysis did not include the GUN trial. The weight of this identified evidence, especially the GUN trial, suggests that, while tamoxifen is efficacious in HER2/neu-positive cancers, it may be more efficacious in patients with HER2/neu-negative cancer.

None of the identified trials of aromatase inhibitors versus tamoxifen reported tests of interaction between HER2/neu status and treatment, although the Pennsylvania State University Hershey Medical Center (PSUHMC) trial [17] suggests that, in the metastatic setting, the benefit of letrozole for objective response rate may be more pronounced in patients with HER2/neu-negative breast cancer. Full results regarding HER2/neu-status subgroup analysis have not yet been published from the following large trials of aromatase inhibitors: Aromatase, Tamoxifen, Alone or in Combination (ATAC) [68]; Intergroup Exemestane Study (IES) [69]; and Breast International Group (BIG) 1–98 [70]. An abstract from the ATAC trial [71] did report a marginally significant difference (P = 0.05) between the estrogen-receptor positive/progesterone-receptor positive (hazard ratio 0.82, 95% confidence interval 0.65–1.03) and estrogen-receptor positive/progesterone-receptor negative (hazard ratio 0.48, 95% confidence interval 0.33 to 0.71) subgroups. As HER2/neu overexpression and amplification are highly correlated with estrogen-receptor positive/progesterone-receptor negative status [71], these results may reflect an interaction between HER2/neu status and anastrozole therapy. However, an abstract from the BIG 1-98 trial [72] reported no obvious difference between estrogen-receptor positive/progesterone-receptor positive (hazard ratio 0.84, 95% confidence interval 0.69 to 1.03) and the estrogen-receptor positive/progesterone-receptor negative (hazard ratio 0.83, 95% confidence interval 0.62 to 1.10) subgroups comparing letrozole versus tamoxifen, so this relationship may not hold across all aromatase inhibitors. If and when HER2/neu subgroup analyses are published from these trials, it may then be possible to make a definitive statement regarding aromatase inhibitor therapy and HER2/neu status.

At this time, there is insufficient evidence to draw any conclusions regarding the relationship, if any, between ovarian ablation or the combination of chemotherapy and tamoxifen versus tamoxifen alone, and HER2/neu status.

Only two of the identified trials [29, 30, 34, 35] of anthracycline-based chemotherapy versus non-anthracycline-based chemotherapy reported statistically significant interaction between HER2/neu status and treatment arm for an efficacy outcome. However, all of the identified trials were consistent in showing a trend for patients with HER2/neu-positive cancer experiencing greater benefit from anthracycline-based therapy than those with HER2/neu-negative cancer. This evidence suggests that patients with HER2/neu-negative cancer gain no benefit from anthracycline-based chemotherapy compared to cyclophosphamide, methotrexate, and 5-fluorouracil. This is borne out by the meta-analysis of the trials, where a significant disease-free and overall survival benefit from anthracycline-based therapy was found only in patients with HER2/neu-positive cancer.

Additionally, the evidence from the trials of more intensive (either higher dose or dose-dense) anthracycline-based regimens versus less intensive ones and the meta-analysis of three of them support the conclusion that more intense anthracycline regimens may provide more benefit in patients with HER2/neu-positive cancer. These results are not conclusive; a significant interaction could not be established in the meta-analysis, as the choice of testing method used in the Belgian trial [23, 24] made a considerable difference in the results. However, as there were fewer than 30 patients per arm in the HER2/neu-positive subgroup in that trial, this association with testing may well be simply an artifact of low numbers.

The one identified trial [43, 44, 45] of 5-fluorouracil, epirubicin, and cyclophosphamide versus 5-fluorouracil, epirubicin, and cyclophosphamide combined with high-dose chemotherapy found a significant interaction in terms of disease-free survival, with patients with HER2/neu-negative cancer being the only patients who received any significant benefit from the addition of high-dose chemotherapy. It is important to note that the 5-fluorouracil, epirubicin, and cyclophosphamide regimen received in both arms differed only by a single cycle (four cycles in the high-dose chemotherapy arm versus five cycles in the other arm), and so this interaction is likely solely due to the high-dose chemotherapy.

Of the trials of taxane-containing versus non-taxane containing regimens, only two, one in the metastatic setting [53] and one in the adjuvant setting [55, 56], reported statistically significant interaction between HER2/neu status and treatment arm (taxane versus non-taxane regimen). The metastatic trial found a significant interaction in terms of objective response rate, with docetaxel exhibiting a greater objective response rate in patients with HER2/neu-postive cancer than in patients with HER2/neu-negative metastatic cancer, but found no significant interaction for time-to-progression or overall survival. In the adjuvant trial, a significant interaction between HER2/neu status and treatment arm (doxorubicin and cyclophosphamide followed by paclitaxel versus doxorubicin and cyclophosphamide alone) was detected for both disease-free and overall survival. A meta-analysis of the three identified adjuvant trials [54, 55, 56, 57] found a significant disease-free survival benefit for the taxane-containing regimens in both patients with HER2/neu-positive cancers and those with HER2/neu-negative cancers, but it also found a significant difference in disease-free survival log-hazard ratios between the groups. Those with HER2/neu-positive cancer derived greater benefit from the taxane, but this meta-analysis also showed a significant benefit for both groups. Based on this evidence, patients should be considered for adjuvant taxane-based chemotherapy regardless of their HER2/neu-status. In the metastatic setting, there is insufficient evidence to draw any conclusions regarding a relationship between taxanes and HER2/neu-status.

In conclusion, there is a growing body of evidence, especially pertaining to anthracycline-based regimens, that the effectiveness of systemic therapy is related to the HER2/neu status of the breast cancer being treated. With the recent results regarding the effectiveness of adjuvant trastuzumab [73, 74], it is expected that the number of patients who are tested for HER2/neu overexpression or amplification will increase greatly. More clinicians will have access to this information and will be considering it when deciding which systemic therapy options should be offered to their patients. Because of the relatively low statistical power of interaction tests, a clinically significant interaction between HER2/neu status and treatment is possibly not being detected in the existing substudy analyses of randomized clinical trials. Therefore, questions regarding the interaction between HER2/neu status and some forms of systemic therapy, such as aromatase inhibitors, taxanes, and ovarian ablation, may not be completely answered unless trials are specifically designed, and powered, to answer them.

Copyright information

© Springer Science+Business Media, LLC 2007

Authors and Affiliations

  • Bindi Dhesy-Thind
    • 1
  • Kathleen I. Pritchard
    • 2
  • Hans Messersmith
    • 3
  • Frances O’Malley
    • 4
  • Leela Elavathil
    • 5
  • Maureen Trudeau
    • 2
  1. 1.Juravinski Cancer CentreHamiltonCanada
  2. 2.Toronto Sunnybrook Regional Cancer CentreTorontoCanada
  3. 3.Cancer Care Ontario, Program in Evidence-Based Care, Department of Clinical Epidemiology and BiostatisticsMcMaster UniversityHamiltonCanada
  4. 4.Mount Sinai HospitalTorontoCanada
  5. 5.Henderson General HospitalHamiltonCanada

Personalised recommendations