Validation of Linear Range HER2/Estrogen Receptor/Progesterone Receptor IHControls for Daily Quality Assurance

Abstract

Objectives

To evaluate a new US Food and Drug Administration (FDA)–cleared immunohistochemistry (IHC) control (IHControls [Boston Cell Standards]) comprising peptide epitopes for HER2, estrogen receptor (ER), and progesterone receptor (PR) attached to cell-sized microspheres and to compare its performance against conventional tissue controls.

Methods

IHControls and tissue/cell line controls for HER2, ER, and PR were compared side by side daily at 5 clinical IHC laboratories for 1 to 2 months. Separately, the sensitivity of the 2 types of controls was evaluated in simulated IHC assay failure experiments by diluting the primary antibody. Additional evaluations included lot-to-lot manufacturing reproducibility of 3 independent lots and specificity against 26 antigenically irrelevant IHC stains.

Results

Side-by-side testing revealed a 99.6% concordance between IHControls and tissue controls across 5 IHC laboratories and 766 individual evaluations. Three discordant quality control events were the result of operator error. Simulated assay failure data showed that both IHControls and tissue controls are similarly capable of detecting IHC staining errors. Manufacturing reproducibility of IHControls showed less than 10% variability (coefficient of variation). No cross-reactions were detected from 26 antigenically irrelevant IHC stains.

Conclusions

IHControls, the first FDA-cleared IHC controls, can sensitively and accurately detect IHC assay problems, similar to tissue controls.

KEY POINTS.

• This study introduces the first US Food and Drug Administration–cleared commercial immunohistochemistry (IHC) controls, provided at 3 analyte concentrations to ensure linear range concentrations.

• Validation data include daily comparison with tissue controls at 5 labs, evaluation in simulated assay failure experiments, measuring manufacturing reproducibility, and specificity.

• The new IHC controls (IHControls) show equivalent sensitivity to weakly expressing tissue controls. Lot-to-lot reproducibility is <10%, and they are antigen specific.

INTRODUCTION

Multiple laboratory practice guidelines for the selection of immunohistochemical controls highlight the need for sensitive controls, which are a key component of proper evaluation of assay performance.^1,2 The International Society for IHC and Molecular Morphology and the International Quality Network for Pathology guidelines describe a series of immunohistochemistry (IHC) quality assurance (QA) tools, most notably Immunohistochemistry Critical Assay Performance Controls.¹ These tools usually consist of multiple tissue samples that include cells that show (1) high analyte concentrations, (2) low concentrations representing a clinically relevant descriptive lower limit of detection (LOD), and (3) negative cells/tissues. The College of American Pathologists (CAP) accreditation requirements also recommend low-positive IHC controls, as explained in checklist requirement ANP.22550.³ US Food and Drug Administration (FDA) guidance to IHC test manufacturers similarly recommends the use of low-positive controls.⁴ Similarly, the latest American Society of Clinical Oncology (ASCO)/CAP guidelines for estrogen receptor (ER)–low–expressing tumors recommends using a low-positive control.⁵

Low-positive IHC controls are optimal because their analyte concentrations invariably fall within the linear range of the IHC assay. Often, IHC is viewed as a “stain” rather than an assay,⁶ so terms such as linear range are not typically used. Nonetheless, IHC is an immunoassay. It has a sequence of reaction steps similar to serum immunoassays. Recent publications even quantified the linear ranges of IHC assays for ER and programmed cell death 1 ligand 1.^7,8 A previous study using IHControls (Boston Cell Standards) with defined analyte concentrations demonstrated that controls with analyte concentrations within the linear range are better able to detect assay problems than those with high concentrations that are beyond the linear range.⁹

Using high-positive tissue controls alone is discouraged because the controls may have analyte concentrations beyond the linear range. It is not presently possible to determine whether a high-positive tissue control has an analyte concentration toward the upper end of the linear range or far beyond it. Both yield similarly strong staining. The use of low-positive controls ensures that the analyte concentration is within the linear range.

The importance of linear range controls is well established in other clinical laboratory disciplines. The fact that IHC assays are qualitative, not quantitative, does not change the need for linear range controls. For example, clinical immunology assays that are qualitative, such as serologic assays, incorporate commercial controls that are within approximately 20% of the threshold for positivity,¹⁰ which guarantees a low-positive, linear range control.

Quality control (QC) in the clinical chemistry laboratory is significantly aided by commercially available QC materials. In contrast, IHC laboratories are generally responsible for creating their own controls. Identifying suitable low-positive controls is a time-consuming and laborious task. Consequently, creating enough controls for every patient slide, for the many different IHC assays on the test menu, may not only affect the laboratory’s budget but also be beyond the resources available to many laboratories. This study evaluates an alternative option—recently available, FDA-cleared, commercial linear range IHControls for HER2, ER, and progesterone receptor (PR). IHControls are synthetic sample substrates comprising purified analytes covalently attached to a cell-sized clear-glass microbead. They have previously been described with respect to their construction,¹¹ ability to detect problems in antigen retrieval,¹² and utility as proficiency testing materials.¹³

MATERIALS AND METHODS

Explanation of IHControls

IHControls consist of both positive and negative test microbeads FIGURE 1 that are the size of erythrocytes. The IHControls are provided as a liquid microbead suspension pipetted onto the slide, adjacent to the tissue sample. The droplet hardens at room temperature after 5 minutes and can be processed through all the usual assay steps, including deparaffinization, rehydration, baking, and antigen retrieval, none of which affect the IHControls. IHControls are available at 3 analyte concentrations: level H (“high”; product BRLS11), level M (“medium”; product BRL2U04), and level L (“low”; product BRL2W08). Depending on the IHControls test, 25% (high level) or 50% (medium and low level) of the microbeads will stain positive. Nonstaining microbeads are coated with other HER2, ER, and PR peptide epitopes, thereby representing negative internal controls. Smaller (4-µm diameter) microbeads that are permanently brown colored, regardless of staining FIGURE 1, provide an internal optical color intensity reference for quantitative stain intensity measurements. As a fixed brown color intensity reference standard, the smaller microbeads compensate for variability in microscope optics from lab to lab or day to day.

FIGURE 1.

Appearance of typical stained IHControls (Boston Cell Standards). Each control microbead is 7 to 8 µm in diameter. The amount of antigen (peptide epitope) per microbead is relatively constant. The difference between stained and unstained microbeads is their antigenic specificity. The smaller color intensity–standard microbeads serve as an optical standard for stain intensity quantification.

For each lab and for each stain, the lowest concentration IHControls that produced an easily detectable stain was selected. This procedure, of selecting the lowest concentration that stains, facilitates matching the analyte concentration to the linear range of the laboratory’s stain FIGURE 2. At least 1 of the concentrations will fall within the assay linear range.

FIGURE 2.

Determination that an IHControls test (Boston Cell Standards) is within the linear range of an immunohistochemistry (IHC) assay. A, The images from IHCalibrators (Boston Cell Standards) (top row) and IHControls (bottom row) are illustrated, each with its corresponding stain intensity measure. B, Those stain intensity measures are plotted graphically, demonstrating a linear regression along the linear range of the assay. Calibrators are annotated with the calibrator level, with the letter “L1” denoting the lowest analyte concentration and “L10” denoting the highest. Calibrator data points with open circles are not on the linear curve and not included in the calculation of the regression line.

Quantification of IHC Stain Intensity of IHControls

Stain intensity is expressed as a ratio of the image (stain) intensity of test microbeads divided by the internal color standard microbead color intensity, as previously described.¹¹ Briefly, stained IHControls were photographed using a Zeiss Axioskop microscope fitted with a SPOT Imaging Solutions Insight GIGABIT charge-coupled device (CCD) camera. Before photomicroscopy, the microscope optics were first set for Köhler illumination. The contrast and γ adjustments in the camera software were left at the default setting (γ = 1; contrast control off). The images were analyzed in a custom algorithm running in MATLAB (MathWorks), as previously described.¹¹ Briefly, this algorithm converts images from red-green-blue (RGB) color format to grayscale intensity. Images were then segmented to identify both test microbeads and the smaller color-standard microbeads. Stain concentration was estimated as the dot product of the measured RGB values at each point with the known RGB profile of the stain. Stain intensity statistics were computed for all segmented and classified microbeads and expressed as a ratio of the stain intensity of the test microbead relative to the internal color-standard microbead.

Clinical Study Design: Comparison of IHControls and Tissue Controls

The performance of Boston Cell Standards’ HER2/ER/PR IHControls was compared with conventional tissue controls over 2 months. The IHC laboratory staff placed 1 of the IHControls on the same slide as a tissue control. Tissue controls were those already used at each clinical laboratory site. The pathologist/investigator at the site ­evaluated the stained IHControls and the tissue controls by visual examination. No special training was required for IHControls evaluation, only a brief explanation of the microbeads. One IHControls test per stain per instrument per day was tested alongside the lab’s regular (tissue) control. Personnel at each site evaluated the IHControls and tissue controls stain intensity each day, judging them as passing or failing the quality check. Data characterizing the IHC test were recorded as pass or fail. For IHControls and tissue controls, a score of “pass” meant that the stain intensity was as expected, which was established at the beginning of the study. A “fail” score meant that the stain intensity was noticeably lower than expected. In addition, the stained IHControls were photographed, and the images were then used to quantify stain intensity every day.

Sensitivity Testing Using Simulated Failure

This analysis compared the sensitivity of the IHControls with tissue controls to determine whether 1 might be more sensitive in detecting assay failure. Primary antibody failure was simulated by varying primary antibody dilutions. Increasing primary antibody dilutions represent greater degrees of assay failure. For each dilution, the IHControls and tissue controls were mounted on the same slide. Tests for each group were performed in triplicate. Four pathologists provided pass or fail interpretations to paired sets of stained IHControls and tissue controls in a blinded fashion.

IHControls Specificity Testing

High, medium, and low HER2/ER/PR IHControls were placed on positive (tissue) control slides for 26 antigenically irrelevant IHC stains. The list of 26 antigenically irrelevant IHC stains and the tissue source for their positive controls is described in Supplemental Table 1 (all supplemental materials can be found at American Journal of Clinical Pathology online). The tissue controls and IHControls were evaluated for immunoreactivity, expressed as “positive” or “negative.” The slides were stained on a BenchMark ULTRA system (Roche Diagnostics), according to 26 protocols in use at Tufts Medical Center Immunohistochemistry Laboratory in Boston, MA.

Manufacturing Reproducibility Testing

Three lots of the high, medium, and low IHControls were manufactured. As part of manufacturing QC, the analyte concentrations attached to the microbeads were quantified by measuring microbead fluorescence. Each analyte has a fluorescein, incorporated during solid-phase synthesis of the peptide. Consequently, microbeads with covalently attached HERR2/ER/PR peptides will fluoresce. The fluorescent IHControls were photographed using a Zeiss Axioskop microscope fitted with a SPOT Imaging Solutions Insight GIGABIT CCD camera. Fluorescence intensity in these images was initially measured in relative units (channel numbers) on a 0 to 255 scale, using the thresholding feature of the ImageJ plug-in. The channel number was then translated into an analyte concentration, measured in units of equivalent reference fluorophores (ERF), as defined by US National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 1934.⁸ To translate from relative to absolute units (channel number to ERF units), a series of calibrator microbeads was quantified against NIST SRM 1934, aliquoted, and stored in an ultracold freezer (‒70 °C), as previously described.⁸

RESULTS

Selection of IHControls in the Linear Range

The IHControls are provided at high, medium, and low concentration levels. These concentrations describe both the antigenically relevant and irrelevant control microbeads. The IHC laboratory staff were instructed to use the lowest concentration level that results in easily detectable staining, which ensures that the control’s analyte concentration is within or close to the linear range. If the IHControls stain intensity falls on the linear regression line that defines the assay’s linear range, then the lab’s control has a concentration within the linear range.

FIGURE 2 offers an example that demonstrates the concept both pictorially FIGURE 2A and quantitatively FIGURE 2B. The top of FIGURE 2A has 6 IHCalibrators (Boston Cell Standards) images: a negative control and levels 5 through 9. Below each image in FIGURE 2A is a list of the corresponding analyte concentrations and quantitative stain intensities. FIGURE 2A shows that the stain intensities progressively increase from levels 5 to 9, a finding that is expected for a linear range. (Linearity is quantitatively demonstrated in FIGURE 2B.) Toward the bottom of FIGURE 2A, the stained IHControls test is shown. Based on the stain intensity of the high IHControls test, its analyte concentration falls between levels 6 and 7. Therefore, based on these images, the high IHControls test has an analyte concentration that is within the linear range.

A quantitative analysis is illustrated in FIGURE 2B, which depicts the linear range of the analytic response curve (dashed line). The linear range spans levels 4 through 9. The data points in FIGURE 2B are listed as “L4” for level 4, “L5” for level 5, etc. The place on the linear curve where the stain intensity of the high IHControls test falls is shown (arrow). This quantitative analysis—comparing control stain intensity with a calibration curve—is for the purpose of explaining the concept of a linear range control. It is not required to identify controls within the linear range. By selecting the lowest control level that consistently yields a visually detectable stain, the tester guarantees that the concentration will be within or near the linear range. This determination (of which control to use) needs to be made only once, at the time when the control is initially placed in daily use.

Clinical Testing With IHControls

For each lab and for each stain, quantitative stain intensity data were measured (see “Materials and Methods”). Representative stain intensity data are shown in FIGURE 3. This type of analysis facilitates the implementation of Levey-Jennings graphical analysis FIGURE 3, as previously described.¹⁴ In this example, the variability around the mean demonstrates a coefficient of variation of 8.6%.

FIGURE 3.

Levey-Jennings graphical analysis of HER2 testing in 1 of the test sites. Each data point is the stain intensity of a single IHControls test (Boston Cell Standards) on a single day.

In addition, each participating IHC laboratory measured concordance in the pathologist readouts of the controls (IHControls and tissue controls). For tissue controls, the readouts followed the same ASCO/CAP criteria as for patient samples. If the tissue controls did not show the usual results, then the assay was considered to have failed. The aggregate data from 5 clinical study sites is shown in TABLE 1. In the “Concordance” column TABLE 1, 763 of 766 individual QC checks from 5 IHC laboratories were judged to concur with one another. In 3 instances, the laboratory’s in-house control demonstrated an acceptable QC result, while the IHControls result was weak or nonstaining, suggesting an assay problem. This result triggered an investigation to understand the 3 instances.

TABLE 1.

Concordance Study of IHControls to Labs’ In-House Control

Site	Immunostain/Clone	Instrument	Product	Tissue Controls	Concordancea	Deviations
1	ER NCL-L-ER-6F11	Leica BOND-III	Medium	Breast carcinoma and normal breast tissue	39/39	0
	PR NCL-L-PGR-312	Leica BOND-III	High	Breast carcinoma and normal breast tissue	39/39	0
	HER2 BOND CB11	Leica BOND-III	Medium	Cell lines (0 to 3+)	39/39	0
2	ER LDT (SP1)b	Leica BOND-III	Low	Tonsil and endometrium	171/172c	1
	PR Agilent M3569	Leica BOND-III	Medium	Tonsil and endometrium	171/172c	1
	HER2 PATHWAY	BenchMark ULTRA	High	Cell lines (0 to 3+)	42/42	0
3	ER CONFIRM	BenchMark ULTRA	Low	Tonsil, breast carcinoma, and normal breast tissue	36/37	1
	PR BOND PA0312	Leica BOND-III	High	Tonsil, breast carcinoma, and normal breast tissue	37/37	0
	HER2 PATHWAY	BenchMark ULTRA	High	Breast carcinomas (low and high positive)	37/37	0
4	ER CONFIRM	BenchMark ULTRA	Low	Breast carcinoma and adjacent normal	34/34	0
	HER2 PATHWAY	BenchMark ULTRA	High	Breast carcinoma and adjacent normal	34/34	0
	PR CONFIRM	BenchMark ULTRA	High	Breast carcinoma and adjacent normal	34/34	0
5	ER NCL-L-ER-6F11	Leica BOND-III	Medium	Breast carcinoma and adjacent normal	50/50	0

LDT, laboratory-developed test.

^aConcordance refers to agreement between each IHControls test and the laboratory’s in-house control.

^bUses the SP1 primary antibody.

^cSlides from 2 days, 10 slides in total, were deemed uninformative and not counted because there were no IHControls to evaluate. It appeared that IHControls were not applied to the slides. These instances were not included in the total.

Cause of Discordant QC in 3 Instances

As described in TABLE 1, 763 of 766 QC test results were concordant. The 3 discordant cases (out of 766 total tests) are the result of operator error. The 2 discrepant QC results at site 2 occurred on different days, 1 for ER and the other for PR. Both appeared to have the same cause: placement of the IHControls material on the extreme edge of the microscope slide, near the label. See FIGURE 4 for an example.

FIGURE 4.

Placement of the IHControls (Boston Cell Standards) on the extreme edge of the microscope slide, near the label.

The cause of the single discrepancy at site 3 for ER testing could not be definitively determined. The IHControls material was unstained, while the in-house (tissue) control was stained appropriately. It is possible that the laboratory technologist inadvertently placed the wrong IHControls sample on the ER slide. This site uses the low-level IHControls test for the ER CONFIRM assay and the high-level IHControls test for PR and HER2 assays. If the high-level IHControls (for PR and HER2) were also inadvertently applied to the ER slide, then it would have been unstained. The target analyte for the ER CONFIRM assay is not represented in the high IHControls test.

Sensitivity in Detecting Assay Failure

The study also examined the ability of 4 pathologists to detect graded amounts of primary antibody dilution, simulating reagent deterioration, using both IHControls and tissue controls. Each dilution was tested with triplicate slides. Therefore, at each dilution, there are 12 readouts (4 pathologists × 3 replicates). The study was performed twice, using slides stained for HER2 and PR. The data are shown in FIGURE 5.

FIGURE 5.

Ability of 4 pathologists to detect assay problems simulated by graded amounts of primary antibody dilution. A, Data for slides stained for progesterone receptor (PR), with primary antibody dilutions of 0 (normal concentration), 1:2, 1:4, and 1:16. B, Data for slides stained for HER2.

The graphs in FIGURE 5 illustrate the detection of assay failure at varying primary antibody dilutions for both IHControls and tissue controls. On the y-axis of each graph, a 100% rate of assay failure (y-axis) means that all 12 readouts at that primary antibody dilution were scored as failing the QC check. Conversely, a 0% detection rate of assay failure means that all 12 QC readouts were scored as passing. On the x-axis, a dilution of “0” is the assay normal mode, using the primary antibody at its standard concentration. The data demonstrate that at each dilution, the IHControls are equally (or better) able to reveal assay failure than tissue controls are. Any possible superiority depends, however, on the selection of the tissue control and its analyte concentration, as described in the “Discussion” section.

Reproducibility of Manufacture

To assess manufacturing reproducibility and product stability, 3 independent lots of each HER2/ER/PR IHControls test were manufactured. The lots are denoted with the suffix -001, -002, and -003 TABLE 2. Analyte concentration is measured by the fluorescence intensity of fluorescein-conjugated analytes.

TABLE 2.

Manufacturing Reproducibility

Microbeada	Concentrationb	CV, %
HIGH IHControls
HER2 High-001b	3,235,101
HER2 High-002	3,398,957
HER2 High-003	3,257,445	2.7
PR16 High-001	3,190,413
PR16 High-002	3,540,469
PR16 High-003	3,637,293	6.8
PR636 High-001	3,160,621
PR636 High-002	3,115,933
PR636 High-003	2,885,045	4.8
ER EP1 High-001	4,113,965
ER EP1 High-002	4,203,341
ER EP1 High-003	3,756,461	5.9
MEDIUM IHControls
All (Medium)-001	318,905
All (Medium)-002	383,455
All (Medium)-003	340,422	9.5
LOW IHControls
All (Low)-001	210,496
All (Low)-002	211,323
All (Low)-003	194,772	4.5

CV, coefficient of variation (s/x̄) × 100; ER, estrogen receptor; PR, progesterone receptor.

^aEach product suffix of “-001,” “-002,” or “-003” refers to the batch or lot designation.

^bThe high IHControls level consists of 4 separate microbeads (HER2 High, PR16 High, PR636 High, and EREP1 High, per TABLE 2), each coated with a different peptide epitope. For the medium- and low-level IHControls, all 9 peptides (representing all the HER2/ER/PR epitopes) are combined onto the same microbead. Therefore, the concentrations listed for each lot are estimated from the total fluorescein concentration divided by the 9 peptides.

For the HER2/ER/PR IHControls, the analytes are purified peptides that include both the epitope and a fluorescein. The fluorescein serves as a manufacturing quality check, quantifying the concentration of peptide per microbead. Determination of peptide concentration per microbead, with units of measure traceable to NIST SRM 1934, has previously been described.⁸ The data TABLE 2 reveal high lot-to-lot consistency. The coefficient of variation (s/x̄) × 100 (CV) for each IHControls test is low, ranging from 2.7% to 9.5%. Compared with the variability resulting from tissue heterogeneity, variability of samples from 1 patient sample to the next, and variability of the preanalytical conditions to which tissue controls are exposed, a CV less than 10% is a significant improvement.

IHControls Specificity

Specificity was evaluated using 26 antigenically irrelevant IHC stains (Supplemental Table 1). For each antigenically irrelevant stain, a positive tissue control was included to verify stain quality (Supplemental Table 1). In all 26 stains, the positive tissue control showed the expected pattern of staining (data not shown). Specificity was 100%; there was no immunoreactivity of these 26 IHC assays against any of the IHControls FIGURE 6. Positive confirmation of IHControls immunoreactivity was seen with IHC assays for ER, PR, and HER2 FIGURE 6. The corresponding tissue controls for the 26 stains were positive, demonstrating that the assays were working properly.

FIGURE 6.

Specificity of IHControls (Boston Cell Standards). Twenty-six antigenically irrelevant immunohistochemistry (IHC) assays (x-axis) and 3 relevant IHC assays (HER2, estrogen receptor [ER], progesterone receptor [PR]) were tested on the high, medium, and low IHControls. Stain intensity for each is graphically depicted on the y-axis. The indicated stain intensity values are triplicates of 3 measurements from the same control.

DISCUSSION

This study summarizes data from clinical trials of a new type of HER2, ER, and PR IHC control, named IHControls, the first IHC control products to be FDA cleared. IHControls were developed to promote 3 important IHC quality improvements: (1) to ensure that IHC laboratory controls have linear range analyte concentrations, ensuring consistent sensitivity in tracking assay performance; (2) to promote the use of controls on every slide through their ease-of-use and low cost; and (3) eventually, to foster adoption of objective Levey-Jennings graphical analysis for IHC testing¹⁵ when software is commercially available for automated analysis.

The introduction of commercially available IHC controls represents a departure from the current reliance on tissue or cell-line controls. IHControls are different from tissue controls in both composition and appearance, which conforms to precedent from clinical chemistry and clinical immunology, in which commercially available controls usually do not consist of patient specimens. Instead, they tend to be manufactured from purified materials in defined buffers. A comparison of cell lines, tissue controls, and IHControls characteristics is provided in TABLE 3.

TABLE 3.

Comparison of Tissue/Cell-Line Controls to IHControls

Characteristic	Tissue and Cell-Line Controls	IHControls
Provides preanalytic assessmenta	No	No
Sensitive analytic test assessment	Yes, if low analyte concentration	Yes
Lot-to-lot reproducibility	Tissue samples: nob	CV <10%
	Cell lines: unknownc
FDA cleared	No	Yes
Can be manufactured at any desired concentration	No	Yes
Labor expense/investment	Tissue: manually prepared or purchased	Purchased
	Cell lines: purchased
Analyte	Cells and tissues	Purified peptides or proteins
Formalin fixed	Yes	Yes

CV, coefficient of variation (s/x̄) × 100; FDA, US Food and Drug Administration.

^aNo external control provides an assessment of preanalytic variables.

^bEvery patient sample is unique.

^cAlthough newer, stably transfected cell lines may more homogeneously express a desired protein, data quantifying lot-to-lot reproducibility of protein expression using calibrated (traceable) assays are generally not available.

Despite the differences, the guidelines for the use of IHControls are similar to those of laboratory-developed tissue controls. Per published recommendations for HER2, ER, and PR testing, a control can be placed on every patient slide.^16-21 Just as a low-positive tissue should be selected as a control, the lowest-concentration IHControls test that yields a clearly detectable stain should be selected to ensure that the analyte concentration is within the linear range. Different levels may be needed for the various HER2, ER, and PR assays. The decision whether to use a high, medium, or low control is made when an IHControls test is first placed to service. The same control level will then be used going forward. As with a tissue control, IHControls are also inspected visually to assess stain intensity.

Best practices for IHC controls mirror those of qualitative assays in other laboratory disciplines. The QA principles are similar or identical. In particular, the purpose of a control for a qualitative assay (such as IHC) is to verify assay stability in the part of the analytic response curve near the threshold distinguishing a positive from a negative result. For IHC, that threshold is the LOD. Analyte concentrations above the LOD result in a visible stain. Analyte concentrations below the LOD appear unstained. For consistency in testing, it is important that the LOD be reproducible day after day. For this reason, a low-positive IHC control is optimal. A low-positive control also conforms to Clinical and Laboratory Standards Institute guidance for qualitative tests, which recommends using a control within approximately 20% of the cutoff.¹⁰

HER2 testing may be an exception to the guidance (above) because there may be 2 important thresholds in the analytic response curve. Historically, the distinction between HER2 3+ (gene-amplified) vs HER2 2+ was the most important staining threshold, influencing the decision to offer trastuzumab or to perform gene-amplification studies by in situ hybridization.²² The recent advent of HER2-directed therapy for patients with 1+ expression, however, renders important the HER2 0 vs 1+ cellular expression threshold.²³ For this reason, there may be benefits to both low- and high-linear-range controls.

The adoption of IHControls inevitably involves a small change in laboratory workflow. The HER2, ER, and PR slides with unstained tissue sections are laid out on the benchtop in 3 rows according to the stain. The IHControls can be applied to the slides before or after they are baked. (IHControls can be baked along with the tissue sections, although adherence does not usually require it.) At each row, the histotechnologist places the corresponding IHControls vial. The histotechnologist vortex mixes the vial and pipettes approximately 1 µL per slide near the patient sample, using a 1- to 10-µL or 1- to 20-µL micropipette. After 5 minutes, the IHControls will have dried. The slides can then be processed in the usual fashion. This workflow is in lieu of preparing tissue controls (ie, identifying suitable paraffin blocks, retrieving them from the archive, and microtome sectioning). In the long term, it is possible that automated IHC instruments will be able to robotically pipette the IHControls directly onto the slides.

In this study, a range of performance parameters were examined: (1) daily use in clinical laboratories, (2) sensitivity in detecting graded degrees of simulated assay failure, (3) manufacturing reproducibility, and (4) specificity. The ability of IHControls to detect problems with antigen retrieval has already been described.¹²

Daily Use in Clinical Laboratories

A total of 766 side-by-side evaluations were performed at 5 institutions over approximately 2 months. These evaluations were performed on BenchMark ULTRA and BOND-III (Leica Biosystems) instruments, as listed in TABLE 1. These instruments (or similar versions) are used in approximately 90% of laboratories based on data from proficiency testing organizations. For logistical reasons, not all instrument types used in clinical IHC laboratories could be included. In 99.6% of the evaluations, the IHControls and tissue controls were concordant. The 3 instances of discordance are believed to be the result of operator error. In 2 instances, the IHControls were placed on the extreme edge of the slide FIGURE 4. In the third instance of discordance, the most likely explanation is that the wrong IHControls test was used.

The fact that no IHC staining errors were detected among 766 evaluations may be surprising. Cheung et al²⁴ described an expected failure rate of approximately 2% based on a study of 22,234 IHC slides. That would translate to approximately 15 expected failures from this study of 766 IHC slides. Nonetheless, there is high confidence in the IHControls data because the slides were evaluated both subjectively and objectively. Regarding the latter (objective assessment), every stained slide’s IHControls test was photographed. The stain intensity was quantified (as described in the “Materials and Methods” section) and plotted in a Levey-Jennings graph, an example of which is provided in FIGURE 3.

Sensitivity in Detecting Graded Degrees of Simulated Assay Failure

IHControls and tissue controls are equally sensitive in detecting assay failure if the correct tissue control is selected. FIGURE 5A shows that pathologists were better able to detect a 1:4 primary antibody dilution with the IHControls. This apparent superiority is not inherent to the type of control, however; rather, it relates to analyte concentration. It is one of the most important takeaway messages from this study: an IHC control should have an analyte concentration in the linear range of the assay (ie, “linear range controls”).⁹

Manufacturing Reproducibility

Lot-to-lot variability is under 10% (CV). Because the purified analytes have an attached fluorescein, the analytes can be precisely quantified.

Specificity

Unlike tissue controls, IHControls have only the analytes with which they were manufactured. Consequently, they were specific to HER2, ER, and PR. No staining was observed with 26 antigenically irrelevant stains.

In many ways, IHControls resemble control products from clinical chemistry or immunology in that they are stored in the refrigerator, available at multiple concentrations, manufactured in reproducible lots, and do not incorporate patient samples. Therefore, IHControls do not require database searches to locate suitable specimens, retrieval from paraffin block archives, or microtome sectioning. The introduction of IHControls heralds the start of FDA-cleared IHC controls.

Funding: This work was supported by the National Cancer Institute, National Institutes of Health (grant No. R44CA213476 and R44CA268484 to S.A.B.).