The emerging era of targeted cancer therapies has focused laboratory scientists and clinicians on the need to define and understand molecular targets of novel drugs. For breast cancer patients and doctors, this trend is not news-efforts have been under way for decades to identify the estrogen and progesterone receptors and define the value of these markers as predictors of response to hormonal therapy.
This excellent review by Drs. Fornier, Risio, VanPoznak,and Seidman provides an explanation of the regulatory pathways of cellproliferation that involve the epidermal growth factor receptor (EGFR) family.Ultimately, the authors focus on the differences in HER2 testing methods and,most importantly, the correlation between immunohistochemistry, fluorescence insitu hybridization (FISH), and clinical response. Unfortunately, in most of thereported studies, the concordance between immunohistochemistry and FISH may notreflect the real world of HER2 testing, because they use the results of bothimmunohistochemistry and FISH testing performed in experienced referencelaboratories.
The need for accurate determination of HER2 status is becoming more apparent,as therapeutic decisions are based mainly on testing in the advanced setting,and now on clinical trials evaluating trastuzumab (Herceptin) in combinationwith polychemotherapy in the adjuvant setting. Which test should be used todetermine eligibility for therapy? Of greater concern and controversy, however,is the level of concordance between local and central laboratories.
Concordance Between Assays
Consistently, multiple reports have demonstrated high concordance betweenFISH and immunohistochemistry for cases scored as 0 and 3, especially when thetesting was performed in highly experienced laboratories. In contrast, lowconcordance has been found for immunohistochemistry scores of 1+ or 2+.
One such important concordance study was presented by Mass and colleagues,who detected gene amplification in 3%, 7%, 24%, and 89% of samples with 0, 1+,2+, 3+ HER2 overexpression by immunohistochemistry, respectively. Of note,this study had an artificially selected equal number of patients scored asnegative (0 or 1+) and positive (2+ or 3+) by immunohistochemistry. Thisarbitrary division does not reflect the actual proportion of patients with theseimmunohistochemistry levels in the general breast cancer population. Tubbs etal have reached the same conclusions and have determined that the largestsource of discordance involves HercepTest scores of 2+, again with a high rateof false-positive results.
A large cohort of patients analyzed by Perez and collaborators demonstratedthat among patients with HER2 overexpression scores of 2+, only a minority hadhigh levels of gene amplification. This information should be seriouslyconsidered, because both of the large single-agent trastuzumab trials includedpatients with 2+ or 3+ HER2 overexpression by immunohistochemistry. As will bediscussed later, retrospective analysis showed that only 1 of 65immunohistochemistry-positive but FISH-negative patients derived clinicalbenefit from trastuzumab therapy.
Another important concordance study of 117 breast cancer specimens wasrecently reported. Press et al found that the accuracy of the FISH assays washigh97.4% for the Vysis PathVision and 95.7% for the Ventana INFORM test. Theimmunohistochemistry assay with the highest accuracy was the R60 polyclonalantibody with 96.6% and 95.7% for the 10H8 monoclonal immunohistochemistry antibody. Interestingly, the lowest accuracy was for the two commerciallyavailable immunohistochemistry tests, the Dako HercepTest (88.9%) and the Ventana C11 monoclonal antibody (89.7%).
Given that a score of 2+ by immunohistochemistry is associated with aninordinately high false-positive rate, a less appreciated problem is thepotential false-negative rate with immunohistochemistry. Mass recognized thatthis first concordance study used an artificial patient population that wasevenly balanced between immunohistochemistry-positive and -negative cases(personal communication, R. Mass, 2001). He then analyzed theimmunohistochemistry results of a more general population of primary breastcancer patients from a National Institutes of Health database.
Knowing that most major trials find that approximately 20% of breast cancerpatients have gene amplification by FISH, Mass calculated the projectedamplification rate in this 1,100-patient data set. He predicted that as many as17% of patients with 0 and 1+ score by immunohistochemistry were likely to beFISH-positive. If this were confirmed in prospective trials, then a much largerpercentage of patients than previously anticipated might be denied therapy withtrastuzumab. To date, no large-scale single-agent trial has treated patientsscored as 0 or 1+ by immunohistochemistry with trastuzumab.
Although these results from Mass were only predictions, Seidman andcolleagues treated immunohistochemistry-negative patients with a combination ofweekly paclitaxel and trastuzumab. They found that 14% of CB-11immunohistochemistry-negative patients were actually FISH-positive.Interestingly, all six immunohistochemistry-negative, FISH-positive patientstreated with paclitaxel and trastuzumab derived clinical benefit (personalcommunication, M. Fornier, 2001). While it is possible that all six responded topaclitaxel, that seems unlikely.
Clinical BenefitAccording to FISH
The study by Cobleigh and colleagues enrolled 222 patients withHER2-overexpressing metastatic breast cancer by immunohistochemistry (2+/3+) whohad progressed after treatment with one or two chemotherapy regimens. Theoverall response rate in an intent-to-treat analysis was 15%. None of the 39FISH-negative patients responded.
Vogel et al reported the results of 114 patients with previously untreatedmetastatic breast cancer. The overall response rate for patients testing 2+or 3+ by immunohistochemistry was 26%. However, among FISH-positive patients,the response rate was 34%. If one includes patients with stable disease forover 6 months, the overall clinical benefit was 48% for FISH-positivepatients. Three FISH-negative patients were reported to have derived clinicalbenefit. However, these three underwent further retesting at a centrallaboratory, and of those, two were found to be FISH-positive (personalcommunication, M. Press, 2001). Thus, in the two major single-agent trials ofFISH concordance with trastuzumab response, only one of 65 patients who wasFISH-negative responded to trastuzumab.
In the pivotal trial of chemotherapy plus trastuzumab, a retrospectiveanalysis was also performed. Among patients who showed gene amplification,54% responded to chemotherapy plus trastuzumab, compared with a 31% responserate among patients who received chemotherapy alone. This shows the importanceof trastuzumab in FISH-positive patients. In contrast, among FISH-negativepatients, no benefit to trastuzumab therapy was observed, with response rates of38% reported for chemotherapy, with or without trastuzumab.
‘Real World’ Performanceof HER2 Testing
Since eligibility for trastuzumab therapy is decided by HER2/neu testing, theaccuracy of detecting HER2 overexpression and amplification in the "realworld" is under scrutiny. Two recent publications have addressed thisimportant issue of concordance between central and local laboratories.
A prospective, randomized, three-arm phase III trial is being conducted bythe Breast Intergroup (N9831) in women with primary breast cancer who stronglyoverexpress HER2 (3+) and/or display gene amplification by local laboratorytesting. Tumor specimens from 119 patients determined to be 3+ byimmunohistochemistry in a local laboratory were centrally retested. Centraltesting showed that only 74% of woman had 3+ positive tumors byimmunohistochemistry, and only 66% had gene amplification. Of nine tumorsreported to have gene amplification by local testing, only six (67%) were foundto have HER2/neu gene amplification by central testing. This also raisesquestions about FISH quality control in local laboratories. Indeed, as many as26% of patients already enrolled in this trial may not be appropriate candidatesfor trastuzumab therapy.
The National Adjuvant Surgical Breast and Bowel Project (NASBP) alsoconducted a review of 104 cases entered in NASBP trial B31 and found that 18% ofthe community-based assays used to establish the eligibility of patients couldnot be confirmed by HercepTest immunohistochemistry or FISH by the centrallaboratory. In this review, they divided the facilities into large-volumelaboratories (using a cutoff of 100 cases per month) and low-volumelaboratories. The discrepancy between scores was only 3% for large-volumelaboratories vs 24% for low-volume laboratories. Only four cases were enrolledon the basis of a FISH-positive test from the community, so the authors did notcomment on the reliability of FISH and central testing.
All the information that we have so far points to the fact that response totrastuzumab therapy is mainly dependent on gene amplification of HER2 asmeasured by FISH. It is now clear that 2+ immunohistochemistry positivity isassociated with an unacceptably high false-positive rate and that all such casesmust be retested by FISH. Considerable uncertainty remains concerning thefalse-negative rate associated with 0 and 1+ scores by immunohistochemistry. Ifthe 17% rate projected by Mass is correct, we are denying trastuzumab use tothousands of women who might potentially benefit.
The concordance rate between 3+ immunohistochemistry and FISH is in the 90%range when both tests are performed in qualified laboratories. Although this mayseem reasonable, should we accept a 10% false-positive rate for such anexpensive therapy? Finally, the lack of concordance shown in the Intergroup andNSABP trials between locally and centrally defined 3+ immunohistochemistrypositivity raises major concerns over quality control of immunohistochemistry inthe community. This leads to even greater misuse and overuse of trastuzumabbased on incorrect test results.
The Intergroup and NSABP results have led to the modification of eligibilitycriteria for their pivotal trastuzumab adjuvant trials. However, the finalresults of these trials done by intent-to-treat analysis are likely to beflawed, since it may only be the FISH-positive subset that would be expected tobenefit. These small differences in an adjuvant study may differentiate apositive trial from a negative one. Another ongoing adjuvant trastuzumab trialBCIRG006mandates FISH positivity prior to study entry, thus leading to potentialreliability in an intent-to-treat analysis.
Improving the quality of testing will ultimately lead to avoidance ofunnecessary costs associated with therapy, undesirable side effects (costs on ahuman level), appropriate implementation of therapy, and fewer inconclusive orerroneous clinical trial results. Although FISH is more expensive andtime-consuming than immunohistochemistry, evidence suggests that it mayrepresent the most cost-effective option in terms of total population cost andlife-years gained. We need to put an end to the confusion.
1. Mass RD, Press M, Anderson S, et al: Improved survival benefit fromHerceptin (trastuzumab) in patients selected by fluorescence in situhybridization (FISH) (abstract 85). Proc Am Soc Clin Oncol 20:22a, 2001.
2. Tubbs RR, Pettay JD, Roche PC, et al: Discrepancies in clinical laboratorytesting of eligibility for trastuzumab therapy: Apparent immunohistochemicalfalse-positives do not get the message. J Clin Oncol 19:2714-2721, 2001.
3. Perez EA, Roche PC, Jenkins RB, et al: HER2 testing in patients withbreast cancer: Poor correlation between weak positivity by immunohistochemistryand gene amplification by fluorescence in situ hybridization. Mayo Clin Proc77:148-154, 2002.
4. Press MF, Slamos DJ, Flom KJ, et al: Evaluation of Her-2/geneamplification and overexpression: Comparison of frequently used assay methods ina molecularly charactherized cohort of breast cancer specimens. J Clin Oncol20:3095-3105, 2002.
5. Seidman AD, Fornier MN, Esteva FJ, et al: Weekly trastuzumab andpaclitaxel therapy for metastatic breast cancer with analysis of efficacy byHER2 immunophenotype and gene amplification. J Clin Oncol 19:2587-2595, 2001.
6. Cobleigh MA, Vogel CL, Tripathy D, et al: Multinational study of theefficacy and safety of humanized anti-HER2 monoclonalantibody in women who have HER2 overexpressing metastatic breast cancer that hasprogressed after chemotherapy for metastatic disease. J Clin Oncol 17:2639-2648,1999.
7. Vogel C, Cobleigh MA, Tripathy D, et al: Efficacy and safety oftrastuzumab as a single agent in first-line treatment of HER2 overexpressingmetastatic breast cancer. J Clin Oncol 20:719-726, 2002.
8. Vogel CL, Cobleigh M, Tripathy D, et al: Superior outcomes with Herceptin(trastuzumab) in fluorescence in situ hybridization (FISH)-selected patients(abstract 86). Proc Am Soc Clin Oncol 20:22a, 2001.
9. Mass RD, Sanders C, Charlene K, et al: The concordance between theclinical trials assay (CTA) and fluorescence in situ hybridization (FISH) in theHerceptin Pivotal Trials (abstract 291). Proc Am Soc Clin Oncol 19:75a, 2000.
10. Roche PC, Suman VJ, Jenkins RB, et al: Concordance between local andcentral laboratory HER2 testing in the Breast IntergroupTrial N9831. J Natl Cancer Inst 94:855-857, 2002.
11. Paik S, Bryant J, Tan-Chiu E, et al: Real-world performance of HER2testingNational Surgical Adjuvant Breast and Bowel Project Experience. J NatlCancer Inst 94:852-854, 2002.
12. Ilersich AL, Deschenes J, Tetu B, et al: Cost-effectiveness of HER2testing in metastatic breast cancer (MBC): An analysis of immunohistochemistry (IHC)fluorescent in situ hybridization (FISH) and Canadian consensus guidelines(abstract 992). Proc Am Soc Clin Oncol 20:249a, 2001.