9.2 Cytokeratin Profiling and CUP Patterns
Key Takeaways
- AE1/AE3 is a broad keratin cocktail that still misses some CK8/18-rich simple epithelia; CAM5.2 is the usual low-molecular-weight complement, especially when hepatocellular carcinoma is in the differential.
- CK5/6 marks squamous, basal, and many mesothelial proliferations; it is not a colon or lung-adenocarcinoma workhorse.
- Classic CK7/CK20 teaching patterns: lung adenocarcinoma CK7+/CK20-, colorectal CK7-/CK20+, urothelial CK7+/CK20+, hepatocellular carcinoma often CK7-/CK20-; these are patterns, not laws.
- Merkel cell carcinoma classically shows perinuclear CK20 dots, not diffuse GI-type cytoplasmic CK20.
- Pan-keratin-negative pitfalls include some hepatocellular carcinomas, adrenal cortical carcinoma, seminoma, and some neuroendocrine tumors—especially if only AE1/AE3 was run.
9.2 Cytokeratin Profiling and CUP Patterns
Quick Answer: AE1/AE3 is a broad cocktail that can still miss some simple-epithelium tumors; CAM5.2 covers CK8/18-class keratins. CK5/6 tracks squamous/basal/mesothelial cells. Classic CK7/CK20 patterns are lung adenocarcinoma CK7+/CK20-, colorectal CK7-/CK20+, urothelial CK7+/CK20+, and hepatocellular carcinoma often CK7-/CK20-. Merkel cell carcinoma shows perinuclear CK20 dots. Treat every combination as a pattern that admits exceptions, not as a law.
This OpenExamPrep section is independent teaching on cytokeratin staining patterns covering published QIHC tissue-staining topics. It is not an ASCP publication and does not claim Board or manufacturer approval. Coordinate keratin tables are study patterns for carcinoma-of-unknown-primary (CUP) work-ups; they are not published sensitivity percentages and they are not diagnostic laws.
What cytokeratins are on the slide
Cytokeratins are intermediate filaments of epithelial cells. Type I (acidic) and type II (basic) keratins pair. Antibodies used in clinical IHC are either cocktails meant to catch many keratins or restricted antibodies meant to ask a narrower question (CK7 versus CK20, or high-molecular-weight basal keratins).
A keratin result is only as good as the clone plus the morphology. Diffuse cytoplasmic keratin in a gland-forming tumor is the expected carcinoma pattern. Rare keratin-positive spindle cells at a cautery edge are not a CUP answer. Nuclear keratin is not a standard true pattern; treat it as artifact or a failed read until proven otherwise.
AE1/AE3
AE1/AE3 is the common broad-spectrum cocktail:
- AE1 (type I / acidic) typically includes CK10, CK14, CK15, CK16, and CK19.
- AE3 (type II / basic) typically includes CK1 through CK8.
Together they label most carcinomas. They are a lineage screen, not an organ marker. Practical gaps:
- CK17 and CK18 are not well covered. Tumors that are relatively CK8/18-restricted can be weak or negative with AE1/AE3 and still keratin-positive with CAM5.2.
- Some poorly differentiated carcinomas, some hepatocellular carcinomas, some neuroendocrine neoplasms, and some germ-cell tumors are weak, focal, or negative (see pan-CK pitfalls below).
- Over-retrieved or dirty AE1/AE3 can label unexpected cells at the edge; still require cytoplasmic filament-quality stain in the H&E tumor.
If a stem says the laboratory ran “keratin” and it was negative, ask which keratin. AE1/AE3-negative is not the same as CAM5.2-negative.
CAM5.2
CAM5.2 is used as a low-molecular-weight keratin antibody highlighting CK8/18-class simple epithelia (glands, hepatocytes, many neuroendocrine cells). Laboratories use it when AE1/AE3 is pale in a suspected hepatocellular or simple-epithelial tumor, and as a Merkel-cell keratin that often shows the same paranuclear dots as CK20.
Do not treat CAM5.2 as CK7. Clone history and vendor inserts differ; read the insert for what that lot claims to see. On an exam stem, CAM5.2 is the LMW complement to AE1/AE3, not a colon-versus-lung coordinate antibody. CK7 and CK20 do that job.
CK5/6
CK5/6 is a high-molecular-weight pair used for:
- Squamous epithelium and squamous-cell carcinoma (with nuclear p40; Section 9.3).
- Basal cells (prostate, breast myoepithelium in some protocols).
- Many mesothelial proliferations (with other mesothelial markers in Chapter 10).
CK5/6 is typically negative in conventional lung adenocarcinoma and in conventional colorectal adenocarcinoma. A CK5/6-positive lung tumor pushes the squamous (or adenosquamous, or basaloid) side of the differential, not the TTF-1-positive adenocarcinoma side. Exceptions occur; do not use one keratin to overrule the H&E.
CK7 and CK20 coordinate patterns
CK7 is common in lung, breast, Mullerian, pancreaticobiliary, and urothelial epithelium, and is typically sparse or negative in colorectal adenocarcinoma, hepatocellular carcinoma, and prostatic adenocarcinoma—as classic patterns, not guarantees.
CK20 is common in colorectal-type epithelium, Merkel cell carcinoma (as dots), and often urothelial carcinoma, and is typically negative in lung adenocarcinoma and in many breast carcinomas—again as patterns.
Teach this table as a memory scaffold. Real tumors violate it.
| Classic teaching pattern | CK7 | CK20 | Notes and exceptions |
|---|---|---|---|
| Lung adenocarcinoma | Typically positive | Typically negative | Some adenocarcinomas are CK7-weak; never use CK7/CK20 alone as “lung proof” |
| Colorectal adenocarcinoma | Typically negative | Typically positive | Some tumors, including some right-sided or mucinous cases, can be CK7+; still correlate with CDX2/SATB2 (Section 9.3) |
| Urothelial carcinoma | Typically positive | Typically positive | Coordinate CK7+/CK20+ is the classic bladder/urothelial pattern, not unique to bladder |
| Hepatocellular carcinoma | Often negative | Often negative | Keratin-poor with AE1/AE3 is a known pitfall; CAM5.2 or canalicular markers may still help |
| Prostatic adenocarcinoma | Often negative | Often negative | Lineage markers in Section 9.3 outrank this pair |
| Merkel cell carcinoma | Typically negative | Positive as perinuclear dots | Diffuse cytoplasmic CK20 without dots suggests a different (often GI-type) keratin pattern |
| Breast carcinoma | Often CK7+ | Often CK20- | Overlaps lung; use GATA3 and the clinical/H&E context, not CK7 alone |
| Pancreaticobiliary | Often CK7+ | Variable | Do not force these tumors into the colorectal CK7-/CK20+ box |
Unknown primary work-ups start with morphology, a keratin screen, then CK7/CK20, then organ markers. CK7+/CK20- in a gland-forming metastasis is a lung/breast/Mullerian/upper-GI-type cluster, not a diagnosis of lung cancer. CK7-/CK20+ is a lower-GI-type cluster, not automatic proof of a colon primary. CK7+/CK20+ keeps urothelial and some GI/pancreaticobiliary tumors in play. CK7-/CK20- keeps prostate, hepatocellular, renal, and some adrenal/germ-cell tumors in play—if the H&E allows those ideas.
Say typically / classically / often. Do not invent unpublished percent sensitivities for these combinations.
Merkel cell carcinoma and dot-like keratin
Merkel cell carcinoma is a high-grade cutaneous neuroendocrine carcinoma. The keratin lesson is localization: CK20 (and often CAM5.2) as tight perinuclear dots, reflecting intermediate-filament balls. Diffuse cytoplasmic CK20 in a gland-forming tumor is the colorectal-type pattern, not the Merkel pattern. CK7 is typically negative in Merkel cell carcinoma. Neuroendocrine markers belong in Chapter 10; this section only asks you to recognize the dot-like keratin as true stain (Section 9.1) rather than as dirt.
Pan-CK-negative pitfalls
“Pan-cytokeratin negative, therefore not carcinoma” is a dangerous shortcut when only AE1/AE3 was used, or when the tumor type is known to be keratin-poor.
| Pitfall tumor class | Keratin behavior (typical pattern) | What else the slide may need |
|---|---|---|
| Some hepatocellular carcinomas | AE1/AE3 weak or negative; CAM5.2 or canalicular stains may still be informative | Hepatocellular lineage markers; do not call the stain failed if residual ducts are CK-positive |
| Adrenal cortical carcinoma | Often keratin-negative or only focal | SF1 / inhibin / melan-A class markers (lineage, Chapter 10); keratin-negative does not make it lymphoma |
| Seminoma / some germ-cell tumors | Often keratin-negative or only focal | OCT3/4, SALL4, CD117/PLAP-class markers; a few scattered keratin cells do not convert it to carcinoma |
| Some neuroendocrine neoplasms | Variable; some are CAM5.2+ and AE1/AE3-pale | Do not stop at one cocktail; morphology of salt-and-pepper chromatin still matters |
| Lymphoma / melanoma / many sarcomas | Expected keratin-negative | That negative is useful only after the H&E frame is honest (Section 9.1) |
If residual bile-duct epithelium on a liver core is AE1/AE3-positive and the trabecular tumor is negative, that is not a failed internal control. It may be a keratin-poor hepatocellular pattern. If residual ducts and tumor are both negative, revisit retrieval and the run.
Worked example
A liver mass in a patient with no known primary is AE1/AE3-negative, CK7-negative, and CK20-negative. Residual portal-tract ducts on the same slide are AE1/AE3-positive. CAM5.2 shows faint cytoplasmic labeling in the tumor. The H&E is trabecular with bile-tinged cytoplasm. The cytokeratin lesson is that AE1/AE3-negative does not exclude hepatocellular carcinoma, the CK7-/CK20- pair matches a classic HCC teaching pattern, and you still need hepatocellular lineage stains plus morphology—not a leap to lymphoma because “pan-CK was negative.”
Coordinate keratins sort the map. Organ markers in Section 9.3 name likely neighborhoods. Exceptions live on every service.
Which coordinate cytokeratin statement matches classic teaching patterns while still allowing exceptions?
A trabecular liver mass is negative with AE1/AE3. Residual bile-duct epithelium on the same slide is strongly AE1/AE3-positive. Which interpretation is most appropriate?