Estrogen receptor testing—
  does your lab have the right stuff?

January 2008
Feature Story

Elizabeth L. Wiley, MD, director of surgical pathology at the University of Illinois, Chicago College of Medicine, spoke to technologists and pathologists at Leica Microsystems’ Broad Spectrum Histology Symposium Oct. 9 in Chicago. An edited transcript of her lecture, “Breast Cancer: Prognostic Markers, Fixation, and Processing,” follows.

I’m going to start out with a quiz, because I want to know what the audience knows about estrogen receptor activity using immunohistochemistry. The first question is about estrogen receptor by immunohistochemistry. Is it: A) readily performed on formalin-fixed paraffin-embedded tissue; B) a fragile antigen requiring special conditions of fixation; or C) comparable to dextran charcoal binding, which was the old way of doing estrogen receptor activity? Which is right? Or are all of them correct?

Question 2: Is estrogen receptor: A) present in less than 50 percent of breast cancers; B) more commonly present than progesterone receptor; or C) more commonly expressed in high-grade breast cancers?

Question 3: Estrogen receptor is preserved in: A) decalcified specimens; B) B-5 fixed tissue; or C) alcohol fixatives. [Answers to the quiz are left out for the reader to discover in the body of the talk.]

Why is it important to test for estrogen receptor in breast cancer? One of the reasons estrogen receptor activity is so important in breast cancer is that estrogen receptor is both a predictive and a prognostic marker in breast cancer and estrogen receptor is present in more than one-half of breast cancers. Many studies have shown that before most patients were treated with chemotherapy, patients who had estrogen-receptor–positive tumors did better than patients who had estrogen-receptor–negative tumors. In other words, estrogen receptor is a prognostic marker, or a marker of how long a patient lives without cancer recurring. A predictive marker is a marker that predicts response to a specific treatment. In this case, estrogen receptor predicts that the tumor is responsive to anti-estrogen therapy. If a tumor is estrogen receptor positive, a patient has a whole range of hormone therapy options that are relatively benign compared to the usual chemotherapy agents. Unlike HER2/neu that is found in less than one-quarter of breast cancers, two-thirds to three-quarters of breast cancers have some estrogen receptor activity. All patients with estrogen-receptor–positive breast cancers have the potential to be effectively treated with anti-estrogen therapy. Because a majority of breast cancers are estrogen receptor positive, errors in estrogen receptor testing affect the majority of patients with breast cancer.

Why do we use immunohistochemistry to test for estrogen receptor? One of the advantages of using immunohistochemistry for estrogen receptor is that you can perform it on a variety of specimens. With dextran charcoal testing, the testing had to be performed on fresh or snap frozen tumor. Also, dextran charcoal testing required a lot of tumor. If a tumor wasn’t two centimeters or larger, you really couldn’t test for estrogen receptor activity because you would have to use the whole tumor and that would compromise diagnosis, staging, and grading. Finally, dextran charcoal only measured the ability of tissue to bind estrogen receptor, and thus measured only unbound receptor. If a patient was premen¬o¬paus¬al, or on estrogen replacement or tamoxifen, estrogen would already be bound to the estrogen receptor, resulting in low or negative values.

With immunohistochemistry, estrogen receptor testing is done after the tumor is reviewed microscopically. The estrogen receptor testing can be done on formalin-fixed paraffin-embedded tissue, including archival tissue, and on frozen tissue as well. Immunohistochemistry detects the presence of the estrogen receptor protein itself. The procedure is not affected by bound estrogen, because the test detects an epitope that is part of the estrogen receptor that is not altered by binding of estrogen. With immunohistochemistry, we can actually look at the tumor and see which of the malignant cells are reacting. With dextran charcoal, we were blind. We had to take a chunk of tissue that was thought to be tumor on gross examination. As the tissue was ground up to measure the activity, you didn’t know whether you were measuring the activity of benign tissue that was mistakenly sent as a tumor. You couldn’t tell whether you were taking part of a tumor that was ischemic or necrotic, where you might get a false-negative.

Now we come to the disadvantages of immunohistochemical determination of estrogen receptor activity. From lab to lab, service to service, person to person, there are many factors that can affect estrogen receptor testing: who is grossing, whether it is a weekend or a weekday, what type of specimen, what fixative is being used, etc. Once the immunohistochemical stains are done, how they are read and reported also vary from service to service.

Let us look at preanalytical factors affecting estrogen receptor determination. Even the time that a breast cancer sits in the operating room or on the laboratory bench can affect the results of estrogen receptor testing. Estrogen receptor degrades in unfixed tissue. Tissue sitting at room temperature for four to five hours loses a significant amount of estrogen receptor. It’s very, very important to get the tissue from the patient to the pathology laboratory and from the grossing bench into fixative as soon as possible. If you’ve got a surgery service that operates late into the evening, you need to make arrangements with the OR staff to make sure the tissue is attended to, gets into fixative, and doesn’t sit over¬night degrading.

Your selection of tissue for testing is important, particularly tissue from larger tumors. If a tumor is over two centimeters in size, the center of the tumor will be relatively estrogen receptor negative compared with the edges, because the middle of the tumor is relatively ischemic and will have little reactivity. I remember vividly having to tell one of my colleagues that taking a punch section from the middle of a tumor was exactly the wrong place for estrogen receptor determination, because of ischemia. Another advantage of taking the edge of the tumor for estrogen receptor testing is that the tissue section will often include some benign tissue that will act as a positive internal control.

Another preanalytical factor that affects estrogen receptor determination is how the breast tumor tissue is prepared. You need to slice your tissue thinly. Formalin and alcohol penetrate at the rate of one centimeter per 24 hours. It does little good to put breast tissue in fixative if it is not sliced thinly enough to expose all the tissue to fixative in less than three hours. Preliminary sectioning of tissue should be at 5-mm intervals or less. You really can never slice tissue too thin. Don’t leave tissue unsectioned overnight. If a breast tumor arrives too late in the day to fix for the optimum six to eight hours before processing, you need to consider an alcohol-based fixative or have a processor ready to go in the morning. When we receive breast tumors late in the afternoon and the specimens will be sitting overnight in fixative, we slice small, thin sections of tumor and fix them in alcoholic fixative for a couple of hours. They can go on the processor the same night. Weekend and holiday fixation (and processing) is also problematic. Each service should look at its procedures on how best to optimize fixation and processing of breast tumor tissue.

A lot of breast cancer tissue submitted for microscopy is too thick and too big for the tissue cassettes. With too much tissue in a cassette, you get poorly processed tissue, and poorly processed tissue has poor ER reactivity. Poorly pro¬cessed tissue also falls off the slides when subjected to antigen retrieval. Poorly processed tissue doesn’t section, period. A really good rule of thumb is for any cassette, the tissue should not cover more than half of the bottom of the cassette. Thickness for most cassettes should be less than 3 mm. For microwave processors, it’s less than 2 mm.

The type of fixative also has an impact on the results of estrogen receptor determination. Most labs use 10 percent buffered formalin. It’s a great general fixative. It doesn’t make the tissue too hard, and it fixes tissue reasonably quickly. Optimum fixation time for estrogen receptor activity is between six and 18 hours for buffered formalin. Less than that and more than that, you’re going to get a diminution in estrogen receptor or even a false-negative. The optimum fixation time also includes processor time. Consider your tissue is sitting in a processor over the weekend in 10 percent buffered formalin for up to 48 hours. You may well end up with false-negative reactivity or at least lowered estrogen receptor reactivity in breast cancer. Too little fixation is also a problem. It is a problem that I encountered in testing for estrogen receptor on breast cores. Our breast cores on the biopsy processor were getting a total of eight to 10 hours of fixation. The laboratory changed its procedure and started processing immediately after loading biopsy tissue. All of a sudden my breast cores were getting only two to six hours of fixation. And there was a falloff in estrogen receptor activity.

If your service is experiencing insufficient time or too much time to fix tissue, a partial solution is to use alcohol-based fixatives for breast tissue. Alcoholic fixatives give you more flexibility in that if you cut tissue sections thin enough, you can fix for as little as two hours, and you can fix for longer periods than the 18 hours recommended for formalin. Alcoholic fixatives also harden fatty tissue better for microsectioning. However, there are some disadvantages to alcohol-based fixatives. You have to be careful not to overfix. With prolonged fixation, your tissue can become quite hard. To counter overfixation in alcoholic fixative, the tissue must be softened in order to microsection the tissue. Alcohol fixation also introduces changes in nuclear morphology that some pathologists do not like.

Regarding some fixatives to avoid—anything that is acidic. Picric acid based—forget it. Bouin’s and B-5 fixative are death to estrogen receptor activity. Buffering is important. You want to avoid fixatives that go below pH range of 7.4. Decalcification procedures also destroy estrogen receptor. If you are dealing with bone marrow biopsies or bones where there is a question of metastatic breast disease, you need to talk to your hematopathologist about setting up an alternative protocol so that you have some tissue associated with the bone marrow that hasn’t gone through decalcification. It might be setting aside a small piece of marrow or some clot tissue. In the case of bone fractures, it might be having your pathology assistant or your pathologist pick out some solid tissue that doesn’t have bone fragments, and have that fixed separately from the decalcified material. Then there is tissue that will allow you to do estrogen receptor activity if there is metastatic disease.

Unlike most immunohistochemical tests, estrogen receptor activity is a quantitative test (amounts), not a qualitative (yes or no). Unfortunately, physicians caring for breast cancer patients tend to think of estrogen receptor activity as “positive” and “negative.” However, you need to quantify the amount of estrogen receptor in a tumor. The amount of estrogen receptor activity is a measure of how sensitive a tumor is going to be to anti-estrogen therapy. Cancers with high amounts of estrogen receptor tend to respond better to anti-estrogen therapy than tumors with low amounts.

Optimum tissue processing is critical for estrogen receptor determination. Blocks should be checked for good infiltration of paraffin. You shouldn’t have blocks that have holes and cracks throughout the tissue. You will wind up with microsections that have holes, shredding, et cetera. If you have poorly processed breast tissue, go back and scrutinize sectioning, fixation, and processor timing. If your service receives a large amount of breast tissue, consider having a separate processor for breast tissue. Such a processor can have longer dwell times for alcohol and clearant to better process the fatty component of breast tissue. Also, alcohol-based fixatives can be used instead of buffered formalin for the fixation steps if fixation times are suboptimal.

A good way to assess whether your processing procedures are adequate for preserving estrogen receptor positivity is to add positive control tissue to your blocks. It can be in the form of some benign tissue from a reduction mammoplasty or uterine myometrium. Myometrium is reliably estrogen (and progesterone) receptor positive. Almost every surgical pathology service has benign hysterectomies, nice and fresh on a daily basis that can be used for control tissue. You can take small pieces of myometrium and stick them in the tissue cassette with your tumor. That way you have a built-in positive control right there in the block.

Once you have that properly sized, properly fixed block of breast cancer properly processed, you are ready to test for estrogen receptor activity by IHC. I am not going to discuss the immunohistochemistry procedure itself, because this is the subject of another speaker today. However, I will emphasize that your choice of antibody clone, your dilution, the amount of antibody you use, your enzyme linkages, your incubation times, your antigen retrieval, and your incubation temperatures all affect your immunoreactivity. You’ve got to standardize your assay. You’ve got to do it the same way each and every time, so that the pathologist reading the stains for estrogen receptor can expect a uniform result day in, day out. This can be hard to achieve, particularly if you’re doing it by hand and if you have multiple technologists performing the testing.

One factor that few laboratories address is that you need to test the same amount of tumor each time. If you are testing excised tumors, the amount of tumor on a slide can vary 100-fold. For quantitative immunohistochemistry such variations affect both the intensity and number of cells staining positive. You’ve got to encourage your pathologists and pathologist assistants, when they’re submitting tissue for estrogen receptor, to standardize the amount of tumor that they submit. The difference between a 4-mm length of tumor on a core and a 10-mm diameter tumor on excision is not twofold but 25-fold in consumption of antibody and other reagents.

I can’t say enough how important it is to use plenty of controls. The standard immunohistochemical control tissue tells you only that your IHC procedure works. The reactivity of the benign tissue adjacent to your tumor tells you that your fixation and your processing were appropriate for estrogen receptor activity for the piece of tissue you’re testing. When you process a piece of control tissue in the same block as your tumor, you have a second test of the time on bench before fixation as well.

I strongly encourage each laboratory to develop control tissue banks of ER-positive and ER-negative tumors that will allow you to validate new procedures and retest old procedures periodically. Use of tissue microarray technology not only reduces the cost of test validation but also makes it feasible and cost-effective to cross-test your control tissue and procedures with other laboratories. If you have colleagues at the large cancer institutions who have validated their procedures in large studies, you can send TMA slides and ask them to perform testing. Slides from the same TMA can be tested in your lab and compared. Then you can tweak your immunoassay until your results are comparable.

The immunotesting has been per-formed. Now, how do you evaluate whether you have done a good job? You have to ask: Did the control tissue work? If yes, then did the tumor stain or didn’t it? One nice clue is that breast cancers with tubule formation are virtually always estrogen receptor positive. Also most tumors that are PR positive should be ER positive as well.

Not only do you have to do the assay the same way each and every time, but you’ve got to standardize how it is read and reported. Even though your slides may come out of the lab beautiful and pristine, if the people reading the slides don’t do it in the same way each time, you’re not going to have standardized results. You’ve got to use a system. Your pathologist and the lab have to sit down and decide: What system are you going to use? Are you performing manual quantification? Computer imaging? Who will be reading the slides?

One system is a scoring system developed by Craig Allred, MD [of the Department of Pathology and Immunology, Washington University School of Medicine]. The scoring is divided into two parts. Part of the score quantifies the proportion of tumor staining positive. If you have nothing staining, it’s a score of 0. Positive staining is scored on a geometric scale. More than 0 percent staining is a score of 1, more than one percent, a score of 2, more than 10 percent, a score of 3, more than one-third, a score of 4, and above two-thirds staining is a score of 5. The second part is an intensity score from 0 to 3+. The two scores are added together. Dr. Allred emphasizes that tumors with as low as one percent staining frequently respond to endocrine therapy.

Another manual system is the percent system. You estimate the number of positive nuclei and report the results in those terms. However, the human eye tends to be more sensitive to small amounts of positive staining and estimate that it is more than what it actually is. A study from MD Anderson [L. Diaz] shows that manual scoring can be very consistent, but the scores tended to be higher when compared with computerized image analysis. If your service manually scores estrogen receptor, my suggestion is to limit reading to one or two individuals. You will have better reproducibility.

Reporting of results must be consistent. How estrogen receptor testing is reported is just as important as how it is determined. Every report should have the results reported in the same way. Even the most careful reporting can miscommunicate results. For example, at my prior institution the estrogen receptor was reported as a percent of positive tumor cells in the body of the report using a standardized template. However, they were also reported in the tumor staging summary as “positive” and “negative.” We found that the oncologists were reading no further than the tumor staging summary and regarding estrogen receptor testing results in those terms. Later, when studies emphasized the need for quantification, we were asked why our test results were only positive or negative. Because IHC testing for estrogen receptor for each laboratory differs, it is also important to report how the testing is done, what fixatives are used, what antibody clone is used, and what the reported results mean.

To finish: here is another quiz:

1. [Image of a benign duct with partial positive staining] Is this internal control working? Yes, benign breast tissue almost always is a salt-and-pepper pattern of staining.

2. [Image of the same benign duct, one labeled ER with no staining and one labeled PR with partial positive staining] What is the problem here? Check IP procedure. The tissue was not properly fixed to preserve estrogen receptor activity. Most likely it sat unfixed for a long period, either before being sectioned or placed in fixative without adequate sectioning.

3. [Image of carcinoma with tubule formation and weak staining for estrogen receptor] Is this appropriate staining? No, tumors with tubular formation should be strongly positive for estrogen receptor. Check bench time and fixation times.

4. [Low-power image of breast cancer with strong staining at the periphery and very weak staining in the center] What should be sampled for optimum estrogen receptor determination? The edge of the tumor.

5. [Image of breast cancer with strong nuclear staining of all tumor cells] What percentage is staining? 100 percent.

6. [Image of edge of breast cancer (and benign elements) with no staining] What percentage is staining? None, but benign elements are also not staining. False negative. Check IP procedure, fixation, and bench time.

7. [Image of breast cancer with a mixture of patchy weak and less than one percent moderate nuclear staining] What percentage is staining? Less than five percent or one to two percent. Avoid reporting as negative.