OUP user menu

HER2: Closing the Gap Between Laboratory Testing and Clinical PracticeThe Authors’ ReplyThe Authors’ Reply

Mario Plebani MD, John M.S. Bartlett PhD, FRCPath, David G. Hicks MD, Giuseppe Lippi MD, Keith Miller FIBMS, Swati Kulkarni MD
DOI: http://dx.doi.org/10.1309/AJCP1VQ9LOBOIMHO 897-900 First published online: 1 June 2009

To the Editor

Laboratory testing for human epidermal growth factor receptor 2 (HER2) status is critically important for the management of care of patients with breast cancer, especially for therapeutic decision making.1 Although the American Society of Clinical Oncology/College of American Pathologists (ASCO/CAP) guidelines reaffirmed the importance of quality assurance in diagnostic testing for the HER2 gene and its protein product,2 a recent UK National External Quality Assurance Scheme (NEQAS) on HER2 testing by in situ hybridization showed modest performance, in that the percentage of laboratories achieving “appropriate” results ranged from 45% to 88%, whereas the percentage achieving “inappropriate” results ranged from 5% to 29%.3

Recent technological developments have created a new generation of laboratory diagnostics, sometimes identified with the term “omics,” which holds promise for better means of detecting diseases and monitoring response to treatment. However, as regards laboratory testing, there is evidence that not only recent research findings but also “old” research outcomes have simply been “lost in translation.”4 This, in turn, stresses the need for a more effective process to carry on translational research, involving also laboratory professionals. Recently, the need was emphasized for establishing an international collaboration to set standards and methods to evaluate laboratory diagnostic tests, primarily in terms of analytic performance, but also considering their clinical validity, clinical usefulness, and, finally, ethical and social implications. The story of HER2 testing well illustrates the evidence gap between diagnostic testing and clinical practice, unveiling the problems of reliability, accuracy, and standardization of diagnostic tests. More intriguingly, however, it highlights the need for correct interpretation and use of laboratory information, which are still areas of major improvement in diagnostics.

An interesting lesson we have recently learned from HER2 testing is that only invasive testing, immunohistochemical analysis or fluorescence in situ hybridization (FISH), on human tissues has been considered a reliable predictor of benefit by anti-HER2 therapy in recent guidelines.5 As revealed by the UK NEQAS,3 however, the diagnostic performance of tissue testing might be suboptimal, being influenced by several extra-analytic factors, including the time from tissue removal to tissue fixation, the handling of tissue samples, and the interpretation and reporting of results.1 It is also important to mention here that although some authors have reported good correlation by the published HER2 protein/gene results in formalin-fixed, paraffin-embedded clinical samples, others have shown significantly discordant results between FISH and immunohistochemical analysis, thus questioning the accuracy of immunohistochemical HER2 testing in formalin-fixed tissues because most discordant results have been attributed to variability in testing methods rather than problems inherent in a particular method.1

A growing body of evidence has been obtained in the last years to prove the clinical value of measuring serum HER2/neu by enzymatic assays (Siemens Healthcare Diagnostics, Milan, Italy). These serum-based methods, recently cleared by the US Food and Drug Administration for follow-up and monitoring of metastatic breast cancer, not only show satisfactory agreement with FISH and immunohistochemical techniques but also have several advantages in terms of standardization, possibility of automation, higher throughput, and greater precision and accuracy.5,6 The ADVIA Centaur HER-2/neu Immunoassay (Bayer Diagnostics, Tarrytown, NY), in particular, demonstrated excellent within-run imprecision and total imprecision ranging from 3.0% to 5.6% and 3.2% to 5.7%, respectively. Moreover, HER2/neu levels showed concordance of 78% with the clinical course of disease, which increased to 86% when used in combination with CA-15-3.7 In contrast with tissue testing, which is a 1-time event, monitoring the circulating levels of HER2/neu also provides a real-time assessment of HER2/neu status. As a surrogate of FISH and immunohistochemical techniques, detection of HER2 overexpression by measuring HER2 in tissue extracts by enzyme-linked immunosorbent assay seems to be more sensitive than immunohistochemical analysis and FISH, suggesting that some patients deprived of trastuzumab (Herceptin) treatment may benefit from this treatment and possibly explaining the conversion phenomenon from HER2– to HER2+ observed in relapse and metastatic disease.8,9

When evaluating techniques for HER2 testing, serum-based assays should also be considered because they show better analytic performance, provide additional diagnostic information, and, more important, can be seen as noninvasive, reliable means for follow-up and monitoring.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.

The Authors’ Reply

Although we agree wholeheartedly with the premise stated by Drs Plebani and Lippi that a gap exists between research identifying novel biomarkers and clinical implementation, it is less clear that simply switching from one method (in situ hybridization [ISH] for diagnosis of HER2 status) to another (serum HER2 levels) will address the gap they have identified.

ASCO/CAP and national guidelines1,2 recognize the centrality of accuracy in the diagnostic determination of HER2 status for management of patients with breast cancer. Our recent data from the UK NEQAS on HER2 testing by ISH3,4 demonstrate a marked improvement in performance for laboratories using ISH as a diagnostic method over time. Similar improvements have been noted for HER2 testing by immunohistochemical analysis.5 Clearly, there remains room for further improvement; however, results obtained over many years show that significant lessons have been learned for the development of novel molecular pathology analyses when comparisons are made with the implementation of estrogen receptor quality assurance schemes.6 What is clearly evident from the NEQAS ISH, which has now reported data for a 3-year period, is that the majority of laboratories can and do provide a consistent, high-quality, reproducible, and accurate service for HER2 diagnosis by ISH over time. This is evidence of a technique that has, in fact, translated well, albeit not perfectly, from the research laboratory to the clinic.

In arguing that other tests such as serum HER2 measurements offer an alternative to a method validated in prospective clinical trials7 purely on the basis of a minority of laboratories whose performance requires improvement, we run the risk of overlooking some key milestones in the development of molecular pathology diagnostics. These milestones are applicable to all current and future molecular pathology diagnostic platforms, irrespective of whether they are new “omics”-based tests, such as transcriptomics (messenger RNA) or genomics (DNA/ISH) or “old” ones such as proteomics (serum HER2).

When evaluating laboratory tests for diagnosing HER2 status in breast cancer,8,9 we have used accuracy, precision, reproducibility, and diagnostic relevance to identify ISH testing as a highly precise, accurate, reproducible, in situ method for determining HER2 status. The measurement of serum HER2 levels, contrary to the statement by Drs Plebani and Lippi, has been considered as a potential diagnostic method to determine breast cancer HER2 status. It has been rejected as a diagnostic method for primary breast cancer for a number of reasons, most critically because it does not distinguish between HER2 overexpression or amplification in ductal carcinoma in situ and invasive disease.8,9 The lack of precision of the test for primary tissue diagnosis is reinforced by the poor (78%) concordance cited by Drs Plebani and Lippi between tumor HER2 status and serum HER2. Concordance between tumor HER2 expression and FISH amplification, for example, is far higher, as evidenced by the criterion implemented by ASCO/CAP1 that greater than 95% concordance is required between HER2 testing methods and FISH before implementation. On this criterion alone, serum HER2 fails to satisfy requirements for primary diagnosis of HER2 status. Other factors that affect serum HER2 levels include tumor size, blood flow, expression of tumor necrosis factor α–converting enzyme, and others, all of which are independent of tumor HER2 expression levels.

Does this mean that serum HER2 levels have no potential diagnostic role? We agree that serum HER2 levels have a strong potential to aid the management of breast cancer following a primary diagnosis of HER2 gene amplification using in situ methods. There is growing evidence that serum HER2 may allow modification of a treatment strategy after diagnosis during adjuvant treatment and follow-up. However, there is still an absolute requirement for external quality assurance schemes to show that methods for shed HER2 analysis can be applied routinely in cancer management across multiple centers to a high standard. Although data from a small number of laboratories are available, there are no practice-based quality assessment data for shed HER2 assays currently available. We predict that such a scheme will, as ours has done for HER2 ISH, identify good and poor practices in laboratories providing this service. The objective of such schemes is to identify and correct poor performance, not to reject validated methods because some centers do not apply them correctly.

On a broader note, we agree strongly with Drs Plebani and Lippi that future diagnostic tests should be validated within an international context with international approaches used to set standards in terms of analytic performance (across centers and over time) and validity, precision, and clinical usefulness. There is a need for a clear consensus on the development of novel diagnostic platforms and for rigorous quality assurance schemes once novel tests are implemented. However, we cannot use evidence from a small number of poorly performing laboratories, that fail in practice, to infer that there is a failure in a diagnostic method.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.

The Authors’ Reply

We read with great interest the letter by Drs Plebani and Lippi. They have correctly pointed out that there are other technologies, such as serum-based assay for HER2, that merit consideration when evaluating techniques for HER2 testing. We concur with the assertion that translational research to find alternative strategies for assessing HER2 that lend themselves to quantitation and function measures will, in all likelihood, lead to improvements in the selection of the optimal patient for therapy.

In addition to the methods touched on by Drs Plebani and Lippi, recent reports about a novel assay that provides a quantitative measure of HER2 protein and its homodimers in formalin-fixed, paraffin-embedded breast tissue are intriguing.1 This novel approach, which could be adapted to clinical samples, seems to show some potential toward offering a more personalized treatment strategy for patients with HER2+ breast cancer. It is based on an assessment of a broader dynamic range of expression of HER2 and the functional activity of the receptor.1,2 Such new methods will require rigorous clinical validation studies before they can be accepted as the standard of care and incorporated into clinical practice. For example, these new tests will have to be evaluated in multiple patient cohorts treated in a uniform manner to help establish the practicality and clinical usefulness of these approaches.

In our article3 we chose to focus instead on the methods already in place in clinical practice, ie, immunohistochemical analysis and FISH. As Drs Plebani and Lippi point out, there are challenges associated with each of these methods in terms of their accuracy, reliability, and reproducibility. These issues have been documented in the literature.4,5 Given the critically important implications that the assay results will have in terms of subsequent treatment and, ultimately, the clinical outcome for a given patient, steps to ensure testing accuracy are a high priority and need to be addressed by members of the pathology laboratory community. Issues surrounding variability of testing accuracy were a key driver for the development of the recent ASCO/CAP task force guidelines that led to the publication of HER2 testing recommendations in early 2007.6 Published guidelines for estrogen receptor testing are expected later in 2009. These guidelines have attempted to identify the key factors that can contribute to testing variability and, if judiciously followed, should help correct some of the suboptimal diagnostic performance issues that Drs Plebani and Lippi mentioned.

We agree with the emphasis by Drs Plebani and Lippi on the need for the establishment of international collaboration to help set standards and methods to evaluate laboratory diagnostic tests. We also agree about the need for more research into testing that lends itself to quantitation, greater accuracy, and better patient selection. This research should be focused on addressing the following question: What is the most practical, clinically relevant, cost-effective, reliable, and broadly applicable ancillary testing for optimal patient selection for treatment with anti-HER2 target therapy? The “gold standard” for answering this question has to be related to patient outcomes, and, therefore, the gap between laboratory testing and clinical practice cannot be closed until we have data from clinical trials that validate any new testing methods or approaches.

References

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
View Abstract