Do we need HER-2/neu testing for all patients with primary breast carcinoma?

Authors


Abstract

BACKGROUND

HER-2/neu is a valuable prognostic marker in primary breast carcinoma. Controversy surrounds the correlation between HER-2/neu expression and other prognostic markers, as has been discussed in preclinical and clinical studies. The objective of the current study was to investigate the probability, calculated using parameters that are assessed routinely in clinical practice, that patients with breast carcinoma had positive HER-2/neu status.

METHODS

The authors evaluated HER-2/neu status in 923 consecutive patients with breast carcinoma by immunohistochemical methods. Correlations involving HER-2/neu status, estrogen receptor (ER) and progesterone receptor (PR) status, tumor grade, patient age, lymph node involvement, and tumor size were evaluated using the Mantel–Haenszel chi-square test and the Spearman correlation. The authors created a simple scoring system (i.e., the diagnostic instrument for validation of HER-2/neu score) to define subgroups of patients with breast carcinoma and to determine the likelihood of HER-2/neu positivity.

RESULTS

HER-2/neu overexpression was correlated significantly with negative ER (P = 0.0001) and PR status (P = 0.0001), Grade 3 (G3) lesions (P = 0.0001), and young age (P = 0.006). The likelihood of HER-2/neu positivity in a patient with positive ER and PR status and G1/G2 disease was approximately 6.1%.

CONCLUSIONS

The authors demonstrated in a large patient series that HER-2/neu overexpression was associated with negative hormone receptor status, G3, and young age. In a subgroup of patients presenting with hormone-responsive and G1/G2 tumors, the likelihood of HER-2/neu overexpression was very small. Therefore, the assessment of HER-2/neu status in this subgroup of patients with breast carcinoma may be considered unnecessary, unless the role of HER-2/neu status in adjuvant treatment has been proven. Cancer 2003;98:2547–53. © 2003 American Cancer Society.

The HER-2/neu protooncogene (also called erbB-2) encodes a tyrosine kinase that is strongly related to the receptor for the epidermal growth factor (EGF). Amplification of the HER-2/neu gene occurs in 25–30% of breast carcinomas and results in overexpression of the gene product.1, 2 The prognostic value of HER-2/neu status first was reported in 1987; HER-2/neu overexpression was found to be associated with significantly decreased disease-free survival and overall survival.1 Retrospective analysis has demonstrated that there is a greater probability of tamoxifen resistance in patients overexpressing HER-2/neu. It is believed that HER-2/neu is critical for the control of growth, grading, and mobility of many normal and transformed epithelial cell types. However, the molecular mechanism underlying the oncogenic signals through which HER-2/neu affects the cell-cycle machinery in tumorigenicity has not been completely elucidated. Preclinical data have suggested that there might be receptor crosstalk between the HER-2/neu and estrogen receptor (ER) signal transduction pathways.

In clinical studies, overexpression of HER-2/neu gene has been associated with a number of other adverse prognostic factors, including the absence of ER, progesterone receptor (PR), Grade 3 (G3) lesions, young age, and the number of axillary lymph node metastases.3–7

However, the correlation between HER-2/neu overexpression and other prognostic markers has not been determined. Retrospective clinical data from relatively small patient cohorts have been reported, but no correlation was found between HER-2/neu overexpression and hormone receptor status, tumor grade, or lymph node involvement.8 Similarly, in a small number of patients, no association between HER-2/neu overexpression and young age was demonstrated.9

Therefore, as a prerequisite for the current study, we determined whether HER-2/neu status is associated with accepted prognostic and predictive markers in a large cohort of patients. If there was a strong correlation between distinct biologic features of breast carcinoma, we would define a subgroup of patients in which HER-2/neu assessment may be of minor value in the decision-making process concerning adjuvant therapy.

Appropriate HER-2/neu testing has been discussed over the years, but controversy remains. Various techniques are available for the assessment of HER-2/neu overexpression or HER-2/neu gene amplification, such as immunohistochemistry (IHC) and fluorescence in situ hybridization (FISH). Substantial variability among different methods and different laboratories have been reported.10 Therefore, rigorous quality assurance programs are necessary to provide accurate assessment of HER-2/neu, rendering HER-2/neu testing cost intensive and time-consuming.

The aim of the current study was to investigate whether the association between different prognostic markers can help to define a subgroup of patients with breast carcinoma who have a high probability of HER-2/neu negativity. As a consequence, HER-2/neu testing may not necessarily be performed in these patients with breast carcinoma, increasing the overall cost-effectiveness without losing relevant predictive information.

MATERIALS AND METHODS

Between 1997 and 2003, we examined 923 patients with primary breast carcinoma in an attempt to investigate the correlation between HER-2/neu status and accepted prognostic markers. To comply with the requirements of daily clinical routine, patients entered the study exclusively on the diagnosis of primary breast carcinoma. Patient characteristics are summarized in Table 1. Tumor tissue specimens were obtained by surgery and HER-2/neu status was assessed by IHC. Tumor specimens were fixed in neutral-buffered formalin (4.5%) for 24 hours and embedded in paraffin. Paraffin-embedded tissue sections measuring 3 μm in thickness were used for staining. Immunohistochemical evaluation of HER-2/neu was performed using the HercepTest (Dako, Glostrup, Denmark) according to the manufacturer's instructions. The antigen retrieval method used in the HercepTest kit involved immersing slides in a preheated (95 °C) epitope retrieval solution (Dako) followed by 40 minutes of heating in a water bath at 95 °C. After a 20-minute cool-down period at room temperature, the epitope retrieval solution was rinsed in washing buffer. The assessment of HER-2/neu overexpression was performed as recommended by the following HercepTest scoring guidelines: 0, no staining or less than 10% membrane staining; 1+, partial membrane staining in greater than 10% of the tumor cells; 2+, weak or moderately complete membrane staining in greater than 10% of tumor cells; 3+, strong complete membrane staining in greater than 10% of tumor cells.

Table 1. Patient Characteristics
CharacteristicNo. of patients (%)
  1. ER: estrogen receptor; PR: progesterone receptor.

Age 
 Median (yrs)58
 Range (yrs)23–95
 <50 yrs254 (27.5)
 < 35 yrs 26 (2.8)
 >50669 (72.5)
Histology 
 Ductal609 (66.0)
 Lobular147 (15.9)
 Other167 (18.1)
Tumor size 
 pT1559 (59.9)
 pT2224 (24.3)
 pT337 (4.0)
 pT4 40 (4.3)
 pTx 69 (7.5)
Lymph nodes 
 Negative482 (52.2)
 Positive441 (47.8)
Grade 
 G1166 (18.0)
 G2374 (40.5)
 G3298 (32.3)
 Gx 85 (9.2)
ER staining 
 Negative210 (22.8)
 Weakly positive 72 (7.8)
 Medium-positive252 (27.3)
 Strongly positive387 (41.9)
 Unknown  2 (0.2)
PR staining 
 Negative437 (47.3)
 Weakly positive121 (13.1)
 Medium-positive226 (24.5)
 Strongly positive137 (14.8)
 Unknown  2 (0.2)
HER-2/neu staining 
 Negative684 (74.1)
 1+ 80 (8.7)
 2+ 21 (2.3)
 3+138 (15.0)

Similar to IHC, FISH can be performed on formalin-fixed, paraffin-embedded tissue sections. We used the PathVysion HER2 DNA probe kit (Vysis, Downers Grove, IL), which contained two labeled DNA probes. The HER2 probe, which spans the entire HER2 gene, is labeled in Spectrum Orange. Chromosome enumeration probe 17 (CEP 17) is labeled in Spectrum Green and hybridizes to the alpha satellite DNA located at the centromere of chromosome 17 (17p11.1–q11.1). Inclusion of CEP 17 relative to the copy number of chromosome 17 allows the copy number of the HER2 gene to be determined. Results on the enumeration of 60 interphase nuclei from tumor cells per target are reported as the ratio of the average HER-2/neu copy number relative to that of CEP 17. Specimens with amplification exhibit HER-2/neu and CEP 17 signal ratios greater than or equal to 2.0 when compared with normal specimens (ratio < 2.0). FISH was performed for all patients with Hercep 2+ results using IHC (n = 21).

Determination of ER and PR status was performed by IHC as described previously.11 In brief, tissue sections with a cutoff of less than 10% stained cells and at least weak staining intensity were recorded as hormone receptor–negative. Tissue sections with greater than 10% stained cells were considered to have positive receptor status and were classified as follows: weakly positive, 10–50% stained tumor tissue, weak or moderate intensity; medium positive, 51–80% stained tumor tissue, weak or moderate intensity; strongly positive, greater than 80% stained tumor tissue, moderate or strong intensity. Grading was performed according to the method of Elston and Ellis.12

Statistical Analysis

We calculated the correlation of HER-2/neu status, distinguishing between positive (2+ and 3+) and negative (0 and 1+) tumors, with ER, PR, tumor grade, age, lymph node involvement, and tumor size. To validate our data, we performed a separate analysis considering only Hercep 3+–positive tumors to be HER-2/neu positive. We calculated the correlation of ER and PR results, distinguishing among four different classes (i.e., negative, 1+, 2+, and 3+). The correlation between HER-2/neu status and steroid receptor status in the surgically removed specimens was tested using Mantel–Haenszel chi-square statistics. This statistic assumes a relation between two samples. The Mantel–Haenszel procedure is based on 2 × 2 contingency tables for several total test score categories. Spearman rank correlation is a distribution-free analog of the Pearson correlation coefficient. The Spearman rho coefficient (r) indicates agreement. A value of r approximately equal to 1 indicates good agreement, a value near 0 indicates poor agreement, and negative values indicate inverse correlations. All tests were two-sided.

We created a simple scoring system to determine the probability that combined prognostic markers were HER-2/neu positive (diagnostic instrument for validation of HER-2/neu; DIVER score). This scoring system consists of Score I (i.e., a high probability of HER-2/neu overexpression) and Score II (i.e., a low probability of HER-2/neu overexpression; Table 2). The DIVER score is determined by prognostic markers that exhibit the strongest correlation with HER-2/neu status, namely, ER, PR, and tumor grade.

Table 2. Scoring System for Defining the Necessity of HER-2/neu Testing (Hormone Receptor, Grading, and HER-2/neu; Diagnostic Instrument for Validation of HER-2/neu Score)a
ER statusPointsPR statusPointsGradePoints
  • DIVER: diagnostic instrument for validation of HER-2/neu; ER: estrogen receptor; PR: progesterone receptor; G: grade.

  • a

    ER + PR + tumor grade = DIVER score (range, 0–9 points). Score interpretation: DIVER I = 0–6 points, HER-2/neu status should be tested; DIVER II = 7–9 points, HER-2/neu status is most likely to be negative.

ER-negative0PR-negative0G30
ER+1PR+1  
ER++2PR++2G1, G2, Gx3
ER+++3PR+++3  

RESULTS

The clinical profile of the patient population is shown in Table 1. The median age of the patients was 58 years (range, 23–95 years). A small number of patients were younger than age 35 years (2.8%). The predominant histologic type was ductal carcinoma (66%). Most of the tumors were smaller than 2 cm (59.9%) and exhibited no lymph node metastasis (52.2%). Overall, the patient series presents an unselected cohort of patients with breast carcinoma. Breast-conserving surgery was performed for 713 patients (77.2%).

Of 923 patients, HER-2/neu status was positive (Hercep 2+ or 3+) in 159 patients (17.3%). HER-2/neu positivity is correlated significantly with negative ER (P=0.0001) and PR status (P = 0.0001), G3 tumors (P = 0.0001), and young age (P = 0.006). Detailed data are shown in Table 3.

Table 3. Correlation of HER-2/neu Status (Hercep 2+ and 3+) with Other Prognostic Markers
Prognostic markerMantel–Haenszel chi-square (P value)Spearman correlation (r)
  1. ER: estrogen receptor; PR: progesterone receptor.

ER0.0001−0.285
PR0.0001−0.204
Grade0.00010.262
Age (< 35 yrs vs. > 35 yrs)0.0064−0.023
Tumor size0.00760.015
Lymph node status0.38020.029

Correlations Involving Estrogen Receptor, Progesterone Receptor, and HER-2/neu Status

The levels of ER and PR positivity were inversely associated with HER-2/neu status (P = 0.0001). Of the patients with strongly positive ER (n = 387) or PR status (n = 137), 9.8% (n = 38) and 8.0% (n = 11), respectively, had HER-2/neu-positive tumors. In contrast, of the patients with negative ER (n = 210) or PR status (n = 437), 37.1% (n = 78) and 25.4% (n = 111), respectively, had HER-2/neu-positive tumors (r = −0.285 and −0.204, respectively). The association between HER-2/neu status and combined ER and PR status is shown in Figure 1.

Figure 1.

Association between HER-2/neu status and estrogen receptor (ER) and progesterone receptor (PR) status. Light purple = 0 (neg); dark purple = 1 (pos). neg: negative; pos: positive.

Correlation between Tumor Grade and HER-2/neu Status

Poor grading of breast carcinoma cells is correlated significantly with HER-2/neu positivity (P = 0.0001). Of 298 patients with G3 tumors, 94 (31.5%) had positive HER-2/neu status (r = 0.262). Only 65 of 625 (10.4%) patients with G1, G2, or Gx tumors had positive HER-2/neu status.

Correlation between Age and HER-2/neu Status

We observed a strong correlation between positive HER-2/neu status and a very young age. Eight (30.7%) of 26 patients younger than age 35 years had positive HER-2/neu status (P = 0.0064; r = −0.023), whereas no significant correlation was detected in patients younger than age 50 years (P = 0.4).

Correlation between Tumor Size and HER-2/neu Status

Tumor size and HER-2/neu status are inversely associated (P = 0.0076; r = 0.015). Among patients with breast tumors larger than 5 cm in diameter (n = 67), 32.8% (n = 22) had HER-2/neu-positive tumors.

Correlation between Lymph Node Status and HER-2/neu Status

Of 441 patients with breast carcinoma who had lymph node metastases, 81 (18.4%) had HER-2/neu-positive tumors. Among patients without lymph node metastases (n = 482), 78 (16.2%) had HER-2/neu-positive tumors. No significant correlation between lymph node status and HER-2/neu status was found (P = 0.38, r = 0.029).

Hormone Receptor Status, Tumor Grade, and HER-2/neu Score (Diagnostic Instrument for Validation of HER-2/neu Score)

We generated two different patient subgroups by application of the simple scoring system described in Table 2. The combination of prognostic markers such as ER, PR, and tumor grade yields two subgroups of patients with breast carcinoma who have different probabilities of having HER-2/neu positivity. The likelihood that a patient with breast carcinoma in the DIVER Score I group (n = 611) had positive HER-2/neu status was approximately 22.9% (n = 140). Among patients presenting with DIVER Score II (n = 312), the likelihood of HER-2/neu positivity is 6.1% (n = 19). Considering the clinical relevance of these findings, 140 of 159 patients with HER-2/neu-positive disease (88%) were classified correctly, whereas 19 patients (12%) were misclassified (6 patients with Hercep 2+ staining and 13 patients with Hercep 3+ staining). Using FISH analysis of patients with Hercep 2+ staining, no amplification of HER-2/neu was detected in 5 patients. Therefore, the number of undetected HER-2/neu-positive results decreased to 8.8%. If these numbers prove to be acceptable, HER-2/neu testing may be unnecessary in approximately one-third of patients with primary breast carcinoma. It is noteworthy that the frequency of Hercep 2+ results in the DIVER Score II group was significantly greater (6 of 19 patients [31.6%]) than it was in the DIVER Score I group (15 of 140 patients [10.7%]).

Separate Analysis of Patients with Hercep 3+ Staining

Because the inclusion of patients with Hercep 2+ staining may influence our results, we reanalyzed the data considering only patients with Hercep 3+ staining to have positive HER-2/neu status. Of 923 patients, HER-2/neu status was positive (Hercep 3+) in 138 (14.9%). HER-2/neu positivity was correlated significantly with negative ER (P = 0.0001) and PR (P = 0.0001) status, G3 tumors (P = 0.0001), and young age (P=0.0015). Therefore, statistical significance was enhanced when only patients with immunohistochemical scores of 3+ were considered (Table 4).

Table 4. Correlation of HER-2/neu Status (Hercep 3+) with Other Prognostic Markers
Prognostic markerMantel–Haenszel chi-square (P value)Spearman correlation (r)
  1. ER: estrogen receptor; PR: progesterone receptor.

ER0.0001−0.300
PR0.0001−0.222
Grade0.00010.249
Age (<35 yrs vs. >35 yrs)0.0015−0.046
Tumor size0.00340.080
Lymph node status0.04100.067

DISCUSSION

We demonstrated in a large patient cohort that the combination of distinct prognostic features in primary breast carcinoma leads to a low probability of HER-2/neu positivity. In patients with positive ER and PR status and G1, G2, or Gx tumors, the likelihood of having positive HER-2/neu status decreases significantly, to 6.1%.

Especially in times of limited financial resources, it may be worthwhile to define a group of patients with a very low probability of overexpressing HER-2/neu. Unless it is clear that HER-2/neu overexpression defines a group of patients who need to be treated with trastuzumab or who have defined tamoxifen resistance, it may not be necessary to determine HER-2/neu status in all patients with early breast carcinoma. To our knowledge, the current study is the first to address whether it is necessary to test HER-2/neu status in all patients with breast carcinoma.

As a prerequisite of our study aim, the hypothesis of a significant association between accepted prognostic features and HER-2/neu status had to be tested. We demonstrate in a large and unselected patient cohort the significant inverse correlation between HER-2/neu status and hormone responsiveness, tumor grade, and young patient age. In experimental models, this inverse association between HER-2/neu amplification and hormone independency has been described.13 In clinical studies, the correlation between HER-2/neu status and hormone receptor status has been studied to explain the described resistance to endocrine treatment. Several studies with conflicting results have been published.6, 8, 14–17

Another finding of the current study was the lack of a correlation between lymph node involvement and HER-2/neu status (P = 0.382) in patients with primary breast carcinoma; this finding indicates that HER-2/neu positivity was equally distributed in lymph node–negative and lymph node–positive patients. These data strongly support the suggestion that lymph node involvement is a prognostic criterion, rather than a selection criterion, for adjuvant treatment (recommendation of the Seventh International Conference on Adjuvant Therapy of Primary Breast Cancer in St. Gallen).18

Our scoring system (the DIVER system), which is based on the combination of ER and PR status and tumor grade in patients with breast carcinoma, provides a valuable tool for estimating the likelihood of HER-2/neu positivity in these patients. The assessment of ER and PR status and tumor grade is acknowledged in all breast carcinoma units. Therefore, our scoring system can be used easily in daily clinical practice. The application of the DIVER score with subsequent omission of HER-2/neu assessment in DIVER Score II patients led to a false-negative result in 19 of 923 patients (2.1%). Additional FISH analysis in patients with Hercep 2+ results excluded patients with false-positive results by IHC alone. Therefore, the number of undetected patients with HER-2/neu-positive disease decreased to 15 of 923 (1.6%). As a consequence, the application of the DIVER score to therapeutic decisions would cause only 1.6% of patients with HER-2/neu-positive disease (i.e., patients with Hercep 2+ staining confirmed by FISH amplification and patients with Hercep 3+ staining) to go undetected as possible candidates for additional treatment.

In patients presenting with strongly positive ER status, PR status, and G1, G2, or Gx tumors (DIVER II score = 9 points), the likelihood of HER-2/neu positivity is extremely low and was observed in only 3 patients (0.3%). However, with respect to ethical and legal aspects, are the economic considerations worth the risk of harming a single patient with breast carcinoma? Without any doubt, we would recommend HER-2/neu assessment for all patients, if the method is accurate. Because the evaluation of HER-2/neu status by different methods, such as IHC and FISH, yielded concordance rates of 88–91%,10, 19 false-positive and false-negative results occur daily due to technical issues and have a greater frequency than the error rate of 6.1% reported in the current study. In addition, HER-2/neu testing in patients with primary breast carcinoma is not commonly performed in all European countries at present.

The gold standard of HER-2/neu evaluation has been discussed extensively,10, 20 with emphasis on interobserver as well as interlaboratory variability.21, 22 Therefore, a validated quality assurance program is necessary to provide valid HER-2/neu results.23 In essence, accurate assessment of HER-2/neu status is cost intensive and time consuming. The combination of IHC and FISHappears to be the most accurate method of evaluation. However, the cost of FISH is 28 times greater than that of IHC.19 Moreover, the FISH procedure is not available in all laboratories, because it is technically difficult and time consuming.

We therefore conclude that the accurate assessment of HER-2/neu status should be performed for patients who are most likely to have positive HER-2/neu status, particularly if therapeutic decisions may be based on these results. The easily performed assessment of the DIVER score (Table 2), which is based on the parameters ER, PR, and tumor grade, is useful in identifying a subgroup of patients who are most likely to have negative HER-2/neu status. In this patient subgroup, HER-2/neu assessment may be considered unnecessary. In a hypothetic model, we calculated the potential costs of HER-2/neu testing in the current patient series (Table 5). The economic benefit of using the DIVER score was noteworthy, in that a savings of approximately 15% of the total cost of molecular diagnostics was observed.

Table 5. Saved Costs by Application of the Diagnostic Instrument for Validation of HER-2/neu Score
CharacteristicNo. of patientsIHC ($)aFISH ($)bTotal ($)
  • IHC: immunohistochemistry; FISH: fluorescence in situ hybridization; DIVER: diagnostic instrument for validation of HER-2/neu.

  • a

    The average cost of immunohistochemistry is $5.90 per patient.19

  • b

    The average cost of fluorescence in situ hybridization is $166.67 per patient.19 The costs of fluorescence in situ hybridization were calculated only for patients with Hercep 2+ and 3+ results.

All patients9235445.70 31,946.26
Hercep-positive159 26,500.53 
DIVER I score6113604.90 23,438.63
Hercep-positive119 19,833.73 
DIVER II score3121840.80 5000.73
Hercep-positive19 3166.73 

Less than 2% of patients with positive HER2 status would go undetected by the application of the DIVER score. Nonetheless, these patients may represent an interesting biologic subgroup that warrants further study. To date, it is completely unknown as to whether these patients differ in their prognosis from other patients with HER-2/neu-positive disease. In addition, until subgroup analyses of ongoing treatment trials become available, it is unclear whether these patients will benefit from HER-2/neu-targeting treatment. Given these uncertainties and the enormous economic impact of ‘avoidable’ HER-2/neu testing, acceptance of this marginal loss in the overall specificity of testing appears to be justified.

Acknowledgements

The authors thank Josefine Stani, Department of Pathology, Vienna University (Vienna, Austria) for her valuable support in processing tumor tissue specimens.

Ancillary