With this disclaimer, required of all users of so-called analyte-specific reagents (ASR), the US Food and Drug Administration (FDA) completed a cycle of publication that defined the classification and rules for manufacture and marketing of antibodies used in diagnostic immunohistochemistry. And so the story of efforts to regulate immunohistochemistry comes to a close. Or does it? Consideration of some of the implications of these rules poses a few interesting problems for the pathologist, and asks each of us to carefully evaluate how immunohistochemical tests should now be used and reported.

After almost 2 decades of discussion in both the medical and regulatory communities, and 2 years of discussion on a draft rule originally published in June 1996, rules in the Code of Federal Regulations enacted under 21CFRPart864.1860² pursuant to the Food, Drug and Cosmetic Act that define and regulate the marketing and labeling (but not use) of immunohistochemical test systems (IHCs) as in vitro diagnostic devices (IVDs) were the first of 2 sets of regulations to become effective (August 17, 1998) in the last calender year. In addition to defining how antibodies fall within the 3 classes of IVDs provided for by the 1976 amendments to the Food, Drug and Cosmetic Act, these rules place certain requirements on the manufacturers of IHCs. In addition to registering their products with the FDA and providing evidence of conformity with current good manufacturing practices, vendors are also subject to general controls for IVDs and labeling rules³ including but not limited to the quantity or concentration and purity of the reagent; appropriate methods for its evaluation, use, and storage; criteria for interpretation of the test, including expected values; known limitations of the procedure; specific performance characteristics of the reagent, including definition of accuracy, precision, specificity, and sensitivity; and, of importance, intended use of the antibody as an IVD. As discussed by Dr Taylor⁴ in this issue of the Journal, this final rule for IHCs was a responsible regulatory step that helped ensure continued availability of good quality, well-characterized diagnostic immunohistochemical stains, and served the best interests of patients and the physicians who care for them.

With enactment of the final rule for ASRs (21CFR809 and 864),¹ initially published on November 21, 1997, but effective November 23, 1998, manufacturers were released from much of this labeling burden. The rule required registration of intended ASRs with the FDA, but limited labeling to "description of the identity and purity (including source and method of acquisition) of the ASR in addition to standard information already required for general purpose reagents" with the specific labeling disclaimer: "Analyte-specific reagent. Analytic and performance characteristics are not established."¹ Whereas IHCs under 21CFR864.1860 are defined as "in vitro diagnostic devices consisting of polyclonal or monoclonal antibodies labeled with directions for use and performance claims,"² ASRs under 21 CFR864.4020 are simply designated as "antibodies, both polyclonal and monoclonal, specific receptor proteins, ligands, nucleic acid sequences, and similar reagents which, through specific binding or chemical reactions with substances in a specimen, are intended for identification and quantification of an individual chemical substance or ligand in biological specimens."¹ These definitions clearly shift the burden of analysis of performance characteristics from the manufacturer to the user.

What is the likely effect of these 2 final rules on labeling of currently available immunohistochemical reagents and on the practice of immunohistochemistry? With respect to labeling, several manufacturers have indicated an intention to market at least some reagents as class I IHCs for in vitro diagnostic use, although I anticipate that most such reagents will consist of predilute, or "ready to use," reagents, marketed alone or as part of test kits, in which the end user is expected to use the product as described, based on analytic and performance characteristics of the antibody or test kit already determined by the manufacturer. In contrast, antibodies sold as concentrated stock will in many cases be marketed as ASRs.

My initial response to the ASR final rule in this context was that the manufacturers had been given an easy out; by labeling antibodies as class I ASRs, they are relieved of any responsibility to the end user with respect to the performance of the reagent. (It has been suggested that inability to report performance characteristics might place a vendor with a superior product at a competitive disadvantage,⁵ but as long as that vendor has the option of marketing the antibody as a class I IVD with full labeling, admittedly at additional cost, that disadvantage is not my concern.) I have come to believe, however, that the principle behind market-ing and labeling of ASRs is nonetheless reasonable and that the final rule for ASRs, like the final rule for IHCs, is in the best interests of both patients and caregivers. The laboratorian (end user), in accordance with general laboratory principles that were established by regulations implementing the Clinical Laboratory Improvement Amendments of 1998 (CLIA)⁶ and are expected of any high complexity laboratory as part of CLIA certification, should, after all, be able to determine and document the analytic and performance characteristics of a given test. Hence the expectation under ASR rules that the diagnostic laboratory will develop, evaluate, and document the efficacy of diagnostic tests using ASRs is both reasonable and appropriate. Indeed, as part of its discussion of the final rule for IHCs, the FDA commented that the use of appropriate positive and negative quality control samples required in CLIA-certified laboratories mitigated the potential risks to patients in the use of immunohistochemical reagents in general, a feature of current laboratory practice that helped make possible a class I category exempt from premarket notification for both IHCs and ASRs.²

At least to limited extent, a manufacturer's choice with respect to labeling may also influence the quality of immunohistochemical practice. As both a practical and a philosophical matter, marketing immunohistochemical reagents as ASRs may be preferable to marketing these reagents as IVDs. I have worried for some time that the marketing of predilute antibodies (ready-to-use antibodies) as stand-alone reagents or part of test kits was an invitation to inexperienced laboratories to use tests they were qualified to neither perform nor interpret. Much of the inconsistency between laboratories, in my opinion, stems from the inevitable lack of understanding of quality control and troubleshooting that comes from reliance on prepackaged reagents (and, as a possible corollary, automated methods). The laboratory that is capable of working with a concentrated antibody, developing appropriate quality assurance protocols, titering stock reagents to optimize performance, and discovering the secrets of recalcitrant reagents and detection systems is more likely to fully understand the use and limitations of the technique and should be better able to provide consistent diagnostically relevant results. Marketing of antibodies as ASRs does transfer the burden of characterization to the laboratory (and with it, the need to invest more resources in the proper evaluation of an antibody's performance in a laboratory setting⁵) and shifts much of the risk that comes from acting on established performance characteristics to the end user, but these are the appropriate responsibilities of and risks to a pathologist who is running a safe and reliable diagnostic laboratory.

In this latter context, it is also reasonable to require the end user to disclaim the performance characteristics of ASRs in the diagnostic setting. Although we might quibble about whether the statement "neither cleared nor approved by the FDA"¹ places undue concern in the minds of clinicians and patients (this matter had been addressed in detail, and I urge any laboratory using antibodies as ASRs to first consider the recommendations published by the College of American Pathologists with respect to clarification of the disclaimer⁷), the disclaimer nevertheless emphasizes the role of the laboratory in defining the use and utility of the reported test result and is in keeping with the general requirements of laboratory practice defined by CLIA. The practical effect of this requirement, of course, is the problem that each of us will encounter as we attempt to develop a reasonable (and not unnecessarily distracting) method of reporting immunohistochemical results that derive from use of both IVDs and ASRs, because short of disclaiming all tests, irrespective of regulatory status, the need for the disclaimer will inevitably change from case to case.

• Apart from the latter inconvenience, the final rules for class I IHCs and ASRs together define a state of affairs that reasonably allows for the continued use of immunohistochemistry as an adjunct to pathologic diagnosis. They also help to define the availability and proper use of antibodies that provide information reportable separately from diagnosis with respect to prognosis or prediction of treatment response, as provided for by rules defining submission requirements for clearance by the FDA of an antibody (or test kit based on an antibody) as a class II IVD or ASR. They do not in any practical manner provide guidance to the laboratorian on how to deal with the continued use of products clearly defined as class II under 21CFR864 as markers of potential predictive or prognostic value (eg, estrogen and progesterone receptor proteins, p53 protein, c-erb-B2) that have not yet been submitted to FDA as a 510(k) and cleared for in vitro diagnostic use. Nor do they provide useful guidance on the appropriate use of reagents, ostensibly class I ASRs, that are not relabeled by the manufacturer and remain "for research use only"öno guidance, that is, apart from the following published comments:
  
  ". . . products that have not been manufactured in accordance with CGMPs (current good manufacturing practices) and are labeled 'for research use only' cannot be marketed under the ASR classification or used by laboratories to develop clinical diagnostics."¹
   
• "FDA strongly encourages manufacturers, importers, and distributors of 'research use' in vitro products to maintain a certification program that documents the researcher's agreement that the device will not be used for investigations involving clinical use, including diagnosis, prognosis, and monitoring of a disease state, and will not be used in conjunction with patient records or treatment."⁸

Although the latter comment was made in reference to IVDs commercialized as part of test kits, the general tenor of these statements is clear: antibodies labeled "for research use only" are not to be used as diagnostic tests, and the results of studies using these reagents should not be reported in the patient's clinical record. If those conclusions were immediately applicable to all antibodies in the laboratory refrigerator today that are labeled "for research use only," diagnostic immunohistochemistry would, to a large extent, cease to exist.
In this important respect, the story regarding the regulation of immunohistochemical reagents is not yet finished. How best do we respond to this problem? Although the FDA has indicated that enforcement of rules with respect to "improperly commercialized" IVDs may be deferred for as long as 30 months in some instances,⁸ it is also clear that manufacturers will have only 12 months from the effective date of the ASR final rule to use up existing label stock and appropriately relabel all ASRs registered with the FDA.¹ It is reasonable for the laboratorian to assume, unless specifically notified by the manufacturer, that relabeling of antibodies is in process. Hence continued use of reagents labeled "for research use only," at least until November 23, 1999, seems to be allowed by the letter, if not the spirit, of the ASR final rule.

But what do we do next November when the label remains unchanged? Do we stop using "for research use only" antibodies altogether, even though some may now be standard of practice? If the reagents in question are otherwise classifiable as a class I IHC or ASR exempt from premarket notification, I suggest we go on using them, for 3 reasons: (1) depending on the antibody in question, failure to use the reagent may place the patient at unnecessary risk, a point reiterated by Taylor in his editorial⁴ and elsewhere⁹ ; (2) billing rules for Missouri General American Life Insurance (St Louis, MO) (a previous Midwest provider of Medicare through the Health Care Finance Administration) allow the use of immunohistochemical reagents as adjuncts to differential diagnosis without regard to FDA regulatory status; (3) the manufacturer alone is responsible for ensuring that its reagents conform to the requirements for ASRs.¹ The anticipated cost of relabeling reagents as ASRs, if they are of potential use as diagnostic adjuncts, places no undue burden on the manufacturer¹; failure of the manufacturer to relabel should not be sufficient reason to discontinue using these reagents, as long as the laboratory conforms to CLIA.

However, if the reagents in question are subject to special controls defined for the manufacture and marketing of class II or class III IHCs or ASRs, we have a problem; the lack of preexisting documentation sufficient for FDA clearance or approval, under current regulations, leaves the user at a serious disadvantage. Although each laboratory may be able to define the apparent performance characteristics of a prognostic marker, and use the antibody in accordance with published reports of its utility and efficacy, few laboratories have the means for independent determination of the statistically verifiable prognostic or predictive value of their version of a given test or reagent. The use of such products therefore conceivably places both the patient and the laboratorian at unnecessary risk. In this situation, if antibodies to putative prognostic and predictive markers are not submitted for FDA review, they should not be used in the diagnostic laboratory unless the weight of published clinical evidence unequivocally supports their continued use.