Immunohistochemistry
Author:
Mikael Häggström [note 1]
When learning pathology, the percentages by which immunohistochemistry results are positive or negative for various diseases are generally easily looked up when needed, so what a pathologist needs to learn is mainly how to select the optimal immunohistochemistry panels in the first place for various presentations where the diagnosis is unknown.
Contents
Immunohistochemistry ordering
The main approaches to immunohistochemistry ordering are:
- Look for a specified panel for the presentation at hand. For example, for an undifferentiated tumor with no clear lineage differentiation, an initial panel of CK, S100, vimentin and LCA can be used.[1]
- Coming up with the most relevant differential diagnoses for the case at hand, and find the immunohistochemistry stains that best distinguish them. Immunohistochemistry profiles for diseases and conditions, as well as their main differential diagnoses, is generally found at Pathology Outlines, or you can pay for a subscription to ImmunoQuery[note 2]:
Using Immunoquery
The main page of ImmunoQuery gives you the "Diagnoses" option, where you enter up to 3 differential diagnoses to generate the optimal immunohistochemistry panel to differentiate them (you may need to click ∨ Suggested Panel to show it if it is collapsed). It also displays an automatic message when the included antibodies/immunostains are not sufficient for a satisfactory panel, in which case you can:
- Make a specific search only including the two conditions where suggested immunostains were insufficient, if you had previously compared 3 diagnoses.
- Consider additional stains from the "Comprehensive panel" displayed below the suggested one.
- Switch the "Sensitivity" setting (seen at top) from 1 (which means that diffuse, focal as well as not specified staining count as positive) to 3 (which means that only diffuse staining counts as positive whereas both focal and absent staining count as negative, and references without any specified staining pattern are omitted from the analysis). This has less data to support the suggested stains (since many references do not specify whether positivity was diffuse or focal), but can sometimes state a better distinction between conditions. When including a stain based on its distinguishing features on a sensitivity setting of 3, you need to keep the practice of classifying only diffuse staining counts as positive, and focal to absent staining as negative.
Test question: ImmunoQuery for a squamous cell carcinoma
The attending gives you a lung biopsy case to preview. You are first uncertain about the type of tumor, so you ask a fellow resident, who finds a diagnostic area of the tumor and tells you that this is a typical squamous cell carcinoma. You also look through the patient's history, and find that the patient has had a squamous cell carcinoma of the anus in the past, and you now want to find out whether the tumor originated in the lung, or if it is a metastasis from the anus, or possibly the skin. You therefore do an ImmunoQuery lookup, with the following results:
- Suggested Panel
Insufficient antibodies for a satisfactory panel to differentiate Lung squamous cell carcinoma and Anus squamous cell carcinoma
Antibodies | Lung SCC | Anus SCC | Skin SCC |
---|---|---|---|
EpCAM | 81% Positive Membrane, Cytoplasm |
75% Positive Membrane, Cytoplasm |
0% Positive Membrane, Cytoplasm |
GATA3 | 5% Positive Nucleus |
20% Positive Nucleus |
84% Positive Nucleus |
p16 | 17% Positive Cytoplasm, Nucleus |
87$ Positive Cytoplasm, Nucleus |
45% Positive Cytoplasm, Nucleus |
You go talk with the attending, who agrees that EpCAM, GATA3 and p16 should be in the panel, but just as ImmunoQuery also tells, the attending thinks that the panel is not satisfactory to differentiate lung SCC from anus SCC, and wants you to add one more stain to improve the panel. You go back to ImmunoQuery and increase the Sensitivity from 1 to 3, and get the following results:
- Suggested panel
Antibodies | Lung SCC | Anus SCC | Skin SCC |
---|---|---|---|
EpCAM | 74% Positive Membrane, Cytoplasm |
50% Positive Membrane, Cytoplasm |
0% Positive Membrane, Cytoplasm |
DLK | 28% Positive Membrane, Cytoplasm |
N/A Membrane, Cytoplasm |
100% Positive Membrane, Cytoplasm |
You also perform a repeated search by only entering Lung and Anus SCC, and you get the following results:
- Suggested panel
Antibodies | Lung SCC | Anus SCC |
---|---|---|
p16 | 17% Positive Cytoplasm, Nucleus |
87% Positive Cytoplasm, Nucleus |
- Comprehensive panel
Top results:
Antibodies | Lung SCC | Anus SCC |
---|---|---|
p16 Cytoplasm, Nucleus |
17% | 87% |
GRPR Cytoplasm |
56% | 100% |
You switch sensitivity from 1 to 3 for this result as well, showing:
- Suggested panel
No antibodies to differentiate Lung squamous cell carcinoma and Anus squamous cell carcinoma
- Comprehensive panel
Top result:
Antibodies | Lung SCC | Anus SCC |
---|---|---|
GRPR Cytoplasm |
46% | 91% |
You go talk with a technician at the histology lab, and your hospital offers all the stains in the alternatives, at similar costs, so you don't have to think about expenses and logistics of sending the case out to external labs.
What is the best alternative?
- Choose Cyklin-D1, and favor a lung primary if the stain has even just focal positivity.
- Choose Cyklin-D1, and favor a lung primary if the stain is diffuse rather than focal or no reactivity. (correct)
- Choose EGFR
Immunohistochemistry evaluation
The main methods for evaluating immunohistochemistry results are:
- Looking up each differential diagnosis at for example Pathology Outlines and comparing their expected staining to see which entity is most likely.
- Paying for a subscription to ImmunoQuery[notes 1], where you can enter immunohistochemistry results and generate a list of most likely conditions with that profile.
Preferably, immunohistochemistry results will be very specific or sensitive for a suspected condition, thereby confirming it if positive, or excluding it if negative, respectively. Even when that is not the case, immunohistochemistry can at least alter the likelihoods of different differential diagnoses. In practice, clinicians or pathologists do not state exact or even approximate numbers of likelihoods of differential diagnoses (Further information: Reporting ), since reality is too complex for that, but the following mathematical should still be somewhat followed in order to interpret immunohistochemistry results optimally:
In practice, the most feasible
Pathology practice is too uncertain to perform calculations of exact percentages of likelihoods of differential diagnoses, but to demonstrate the general principle of how immunohistochemistry results are calculated, the following formula can be used:
- Gross likelihood of a disease/condition = (Pre-test probability) x (Probability that the condition shows the immunohistochemistry results at hand).
The pre-test probability is a product of for example the incidence of the condition in the patient's epidemiologic type such as age and sex, as well as the probability that the condition would have caused the clinical course, including signs and symptoms, as well as the microscopic impression. For example, if you want to differentiate a pleomorphic liposarcoma from a pleomorphic rhabdomyosarcoma in soft tissue, you may find in ImmunoQuery that the following stains are most efficient in distinguishing the two, with the following percentages of being positive:
Soft tissue pleomorphic liposarcoma | Soft tissue pleomorphic rhabdomyosarcoma | |
Desmin | 17% | 95% |
Actin HHF-35 | 0% | 71% |
Let's say for example that your pre-test probability was about 30% for pleomorphic liposarcoma and 70% for pleomorphic rhabdomyosarcoma, that desmin stains positive, and actin HHF-35 stains negative in this case. Assuming that 0% positive rate means that 100% stain negative, the probability that pleomorphic liposarcoma shows these immunohistochemistry results at hand is:
- 17% x 100% = 17%
The gross likelihood of pleomorphic liposarcoma therefore becomes:
- 30% x 17% = 5.1%
In the same way, the corresponding gross likelihood of pleomorphic rhabdomyosarcoma is calculated as:
- 70% x 95% x (100% - 71%) = 19%
As a result in this case, immunohistochemistry resulted in pleomorphic rhabdomyosarcoma going from about twice as likely compared to pleomorphic liposarcoma to about 4 times as likely.[notes 2]
Further reading: |
Notes
- ↑ Cite error: Invalid
<ref>
tag; no text was provided for refs namedImmunoQueryCOI
- ↑ More detailed explanations about likelihood calculations on differential diagnoses in general can be read at:
-Häggström, Mikael (2014). "An epidemiology-based and a likelihood ratio-based method of differential diagnosis ". WikiJournal of Medicine (Wikiversity Journal of Medicine) 1 (1). doi: . ISSN 2002-4436.
- ↑ For a full list of contributors, see article history. Creators of images are attributed at the image description pages, seen by clicking on the images. See Patholines:Authorship for details.
- ↑ The author has no financial or other conflict of interest in the mentioning of ImmunoQuery.
Main page
References
- ↑ Lin F, Liu H (2014). "Immunohistochemistry in undifferentiated neoplasm/tumor of uncertain origin. ". Arch Pathol Lab Med 138 (12): 1583-610. doi: . PMID 25427040. Archived from the original. .
Image sources