Re: [Histonet] RE: negative IHC controls
Thanks for the instructive points about IHC QC. First, I do not
represent the CAP (nor I want to), so I am speaking strictly from my
own experience at PhenoPath and (prior to that) University of
Washington and MD Anderson.
You certainly have a point in terms of the "intended" use of controls
to address a "full range of procedural variables" as you correctly put
it. My concern with using QC in IHC is the end result. I can give you
dozens of examples where the referring pathologists sent us cases in
consultation, and we found errors in interpretation of studies, despite
the presence of adequate positive and negative controls that
accompanied their patient's slides. My point is a practical one: It is
really the experience of the pathologist and technologist that is the
rate-limiting factor that determines whether a given laboratory does an
adequate job or not. This by no means nullifies the need for QC. It,
instead, makes it more meaningful and cost-effective.
What I just did is broaden the terms of QC, to include, in addition to
the technical aspects, interpretation and validation aspects that, in
my experience, are way more important (see discussion below on
validation). Unfortunately, the CAP focuses (correctly) on the
technical aspects, but pays very little attention to the interpretation
aspects. The results is hundreds of labs passing the inspections every
year but far fewer have comprehensive approach to technical,
interpretation, validation and (I will not discuss this) proper
selection of antibodies.
I was very careful by saying that running multiple reagent control is
IDEAL, but I didn't say it's practical, so I don't think I downplayed
the complete reagent control issue.
Regarding your comments on positive external controls: in real life,
when the expected positive internal control fails, they usually fail
once (for four main reasons). Repeat studies will yield adequate
results except when the problem is tissue antigenicity due to fixation
and processing issues. If that's the case, studies should be repeated
using different (gentler or harsher) pretreatment. Putting positive
external controls on the same slide will not make any difference in
cases like this one. The expected positive internal control
immunoreactivity is non-controversial. It is based on validation of the
antibody in a given laboratory (not by the manufacturer) PRIOR to its
approval as a diagnostic test. That's what I meant by money well spent.
When we acquire any new antibody, the first thing we do is ignore the
package insert, and test the antibody using at least 5 different
pretreatments on tissues that are expected to be positive, and
negative. Once we establish the proper pretreatment and antibody
dilution, we run the antibody on a pilot study composed of X number of
lesions. If the results are what we anticipate, then we validate the
antibody and approve it for clinical use. That, in my opinion, is
necessary and money well spent. But to run positive and negative
controls on every slide is just not practical in most settings.
I am delighted that some of my colleagues are interested in discussing
this issue. I hope this note helped clarifying a couple of issues to
Hadi Yaziji, M.D.
On Monday, October 13, 2003, at 01:26 PM, Bryan Hewlett wrote:
> Dear Hadi,
> I agree with much of the content of your posting. However, there seems
> to be
> a mixed message here.
> You agree that; (a) running multiple reagent control sections covering
> appropriate antibodies/pretreatments is ideal and that, (b) money
> should be
> well spent on QC (which you emphasize), yet you seem to downplay the
> use of
> complete reagent controls and to inject some confusion regarding the
> necessary use of process controls.
> Process controls are intended to address the full range of procedural
> variables and to ensure that the end result of the process is known to
> within legitimate or acceptable values and is valid. Process controls
> are a
> necessary tool utilized PRIOR to the subsequent interpretation of the
> Subsequent interpretation of the result depends largely upon the
> provided by these process controls. No one has suggested that a reagent
> control on the patient slide is the only way to recognize a false
> signal. In fact, the experienced observer provides the final additional
> level of process control.
> However, that does not mean that all the necessary prior process
> can be avoided or omitted by that experienced observer.
> We too, have seen 100's of IHC stains daily and the full use of process
> controls has been an important factor in obtaining consistent results.
> We have seen numerous occasions where reagent controls have provided
> essential clues to the nature of an aberrant finding.
> I would have to disagree with you that a single reagent control is
> more than
> I do agree with your comments regarding the use of one positive
> tissue control per antibody/ per run but would point out that, if the
> positive 'external' control works and the patient 'expected' positive
> internal controls didn't, repeating the antibody test, using the same
> process, should produce exactly the same result!
> That would mean that either the 'expected' result is not a valid
> expectation, or that the process controls are inadequate and do not
> for all of the procedural variables.
> Bryan Hewlett
> On Friday, October 10, 2003, at 10.06 PM, Hadi Yaziji wrote:
> Theoretically, I agree with you that (a) running multiple negative
> sections to cover the different antibodies/pretreatments is ideal and
> money should be well spent on QC (I can't emphasize that
> However, I honestly don't remember when was the last time when I had a
> positive staining on a patient's slide that the negative
> was the only way that allowed me to recognize the false positive
> signal on
> the real slide. And we look at 100s of IHC stains every day.
> Experienced pathologists and technologists can still recognize in >
> 99% of
> the times a false positive signal from a real signal on the tissue
> of a given antibody without even having to look at negative controls.
> Individuals who perform and interpret IHC studies must have the
> knowledge to
> recognize the expected sub-cellular localization of every antibody on
> type of tissue, including aberrant signals....
> All you need is one positive control per antibody, but not one
> antibody/per patient. In many cases positive internal controls are
> on the same slide, so you can tell whether your antibody worked simply
> evaluating the expected positive internal controls. In fact, if your
> positive 'external' control worked and internal controls didn't, then
> need to repeat your antibody test regardless of the positive external
> The bottom line: in real life, one negative control per case is more
> sufficient. And to say the least, it's inaccurate to state this is poor
> patient care and lousy quality control.
> Histonet mailing list
Histonet mailing list
<< Previous Message | Next Message >>