Jump to content

Sasi Attili's Blog

  • entries
  • comments
  • views

Keratoacanthoma: Controversies and the confusing RCP dataset

Sasi Attili



There is much controversy as to whether Keratoacanthoma (KA) is a valid and safe diagnosis to make or whether all KA's should be labelled as SCC. Much of the confusion I believe is because of the lack of consensus regarding the definition of a KA. The [u]main reason[/u] KA was originally thought to be a unique tumour, distinct from SCC, is because of its tendency to [i]spontaneous regression[/i].

Yes, it has certain unique histological features distinct from SCC [1][size=3];[sup] [/sup][size=4]b[/size][/size]ut, these features would mean nothing if the tumor did not exhibit a tendency to [i]spontaneous regression[/i]. In the absence of [i]spontaneous regression[/i] the discussion regarding whether it is an SCC or not, would be mute! Thus [i]spontaneous regression[/i] is the single & most important diagnostic feature that distinguishes KA from SCC.

However as [url="https://dermpathpro.com/blog/10/entry-25-ka-vs-scc-spot-diagnosis-discussion-6th-july/"]Dr. Richard Carr states in his blog[/url], there are not many brave clinicians out there who would wait for the tumour to start regressing. Most clinicians (exceptions exist [2]) would chose to excise the tumor in its proliferative phase, especially when the resulting scar and local destruction caused as a result of natural regression can sometimes be quite unsightly. So that is where we need pathologists to identify clues, as to whether this tumour in the proliferative phase might eventually regress (KA) or not (well-differentiated SCC).

The [url="http://www.rcpath.org/Resources/RCPath/Migrated%20Resources/Documents/G/G124_DatasetSquamous_Dec12.pdf"]RCP dataset for SCC[/url] released at the end of last year, was designed to create a uniform national dataset for reporting SCC's & KA's, and also guide pathologists as to how to make (or not make) the diagnosis of KA. However, the statements within the document are riddled with irony and have ended up creating more confusion. The problem we are now experiencing is that pathology colleagues are unwilling to support the diagnosis of a KA because of strict exclusion criteria stipulated in the RCP dataset.

I recently had a patient who presented with a typical, rapidly growing, volcano like growth of 6 weeks duration. The lesion halved in size even by the time I was able to see him back for excision (3 weeks later). On microscopy (c[url="https://drive.google.com/folderview?id=0B47a2MWGC7eIbS10UmN1UkIzZWc&usp=sharing"]lick here to see pictures[/url]), there was a crateriform, well-differentiated keratinising squamous tumour with otherwise typical features of a KA. The tumour however did have an invasive base and hence our local pathologists refused to support my clinical diagnosis of a KA (dermal invasion being an exclusion criteria in the dataset), despite the history of rapid growth & involution, typical (read 'diagnostic') of a KA.

Let us examine the RCP dataset/ guidance in detail. Please read the relevant (rather confusing) excerpts from the RCP dataset pasted below (in italics):
[i]‘This dataset also adopts the WHO terminology which considers keratoacanthoma as[/i][/center]
[center][i]synonymous with invasive squamous cell carcinoma (keratoacanthomatous type).'[/i][/center]
This is actually an incorrect interpretation of the [url="http://www.iarc.fr/en/publications/pdfs-online/pat-gen/bb6/bb6-chap1.pdf"]WHO document[/url]. Weedon, who vehemently states that KA is a distinct entity is actually the co-author of this document!. If you look at the wording under the heading 'Keratoacanthoma' in the [url="http://www.iarc.fr/en/publications/pdfs-online/pat-gen/bb6/bb6-chap1.pdf"]WHO document[/url], they have a sub-heading under KA that says (sic):

[indent=2][b]Well-differentiated squamous cell carcinoma (keratoacanthoma type).'[/b][/indent]

No where in the document is KA stated to be synonymous with an invasive SCC. From my understanding 'invasive squamous cell carcinoma (keratoacanthomatous type)' implies that this is an SCC that looks like a KA in some aspects. So to extrapolate on the WHO statement and say that the WHO consider KA synonymous with invasive squamous cell carcinoma (keratoacanthomatous type), is incorrect!! I am sure Weedon would not agree with the wordplay in the RCP document. Anyone who has read his papers would know his view on KA.
[i]'It is [/i][i]widely accepted that there is no absolute single criterion to distinguish conventional/[/i][/center]
[center][i]classic squamous cell carcinoma from keratoacanthoma. In each case, a constellation of[/i][/center]
[center][i]clinical and diagnostic features must be used. Most authorities, however, appear to now[/i][/center]
[center][i]regard keratoacanthoma as a specific variant of follicular invasive squamous cell[/i][/center]
[center][i]carcinoma, capable of self-healing but requiring complete removal of any residual lesion.[/i][/center]
[center][i]The term keratoacanthoma should be avoided in an immunosuppressed patient and[/i][/center]
[center][i]lesions greater than 30 mm in diameter or in a subungual location are recognised to have[/i][/center]
[center][i]potential aggressive behaviour. The diagnosis of keratoacanthoma may be supported by a[/i][/center]
[center][i]clinical history of rapid growth over 1–3 months and subsequent involution over 3–9[/i][/center]
This last statement ('the diagnosis of KA may be supported') appears to contradict the first statement in the para, which says that they consider KA synonymous with 'invasive SCC (Keratoacanthomatous type)'!

[center][i]'T[/i][i]he whole of an individual lesion requires examination to establish the diagnosis. T[/i][i]he diagnosis should be avoided if there is adjacent dysplasia or in-situ squamous cell[/i][/center]
[center][i]carcinoma.' [/i][/center]
Richard- you mentioned that you do not exclude KA on this basis, as a third of your cases have in-situ changes. Weedon does not discuss this issue. Has this been published before? I cannot find a reference. Comments Please?

[center][i]'The lesion is characterised by an exo-endophytic growth pattern with a craterlike[/i][/center]
[center][i]appearance and symmetrical lipped-edge/buttress. The base of the lesion frequently[/i][/center]
[center][i]has blunt downgrowths but may have some irregularity of the lower border. There should[/i][/center]
[center][i]be no overt dermal invasion or dermal desmoplasia.' [/i]Later on they also say[i] '[/i][i]Perineural invasion and possibly vascular invasion are considered to have no [/i][i]adverse effect on prognosis for keratoacanthoma.'[/i][/center]
So the RCP accepts that perineural and vascular invasion can be present in a KA but state that the diagnosis of a KA cannot be made if there is dermal invasion? Can a tumor invade nerves and vessels without invading the dermis first?! Is this not self-contradictory? :unsure:. This also contradicts Richard's statement, (sic) ' typically the early lesions have highly proliferative infiltrative (not pushing circumscript borders like well differentiated SCC) borders'. Weedon skims over this specific issue of 'invasion' and avoids this terminology [1], though he does say 'Extension below the level of the sweat glands is unusual;if it is present, particularly careful assessment of the other histological features is required.' However, he does not state that extension beyond the sweat glands is an exclusion criterion! Perineural and vascular invasion are well recognised features that can be seen in KA [1]. [u]Surely dermal invasion is a prerequisite for this to happen, and hence cannot be an exclusion criterion![/u]
The issue of 'dermal desmoplasia' should also be qualified further. Weedon says 'There is no stromal desmoplasia except in late involuting lesions'. So again this cannot be a criterion for exclusion, as it is seen in late lesions.

Finally, the RCP state:

[center][i]'An invasive squamous cell carcinoma[/i][/center]
[center][i]having some but not all features of keratoacanthoma is better classified as invasive[/i][/center]
[center][i]squamous cell carcinoma with some keratoacanthoma-type features.'[/i][/center]

I wonder if they are trying to make a distinction between 'invasive SCC (Keratoacanthomatous type)' i.e. synonymous with KA according to their introductory line; and, 'invasive squamous cell carcinoma with some keratoacanthoma-type features'. If that is what their intention is, it is indeed very confusing. They need to come up with better terminology!

Coming back to my case- clinically, there cannot be a case more typical (with the history of rapid growth [u]&[/u] involution) of a KA. So any worrying histological features one might see (early invasion in this case) should be interpreted in context. I do agree that the diagnosis of a KA should not be based on histology alone, and as the RCP rightly state- in the absence of absolute criteria for differentiation, a definite pathological diagnosis should not be made. [i]However, this does not imply that the diagnosis of KA cannot be supported by the pathologist, does it?[/i]

Clinically my case had all the features of a KA and thus we have to conclude that 'early invasion can be seen in a KA'. Not the other way round! B)

[i][b]When can we make a definitive diagnosis of a KA?[/b][/i]
As a clinician- I can be sure of something being a KA if it has grown rapidly AND shown signs of involution clinically AND the histology shows a well differentiated SCC with KA like features. [i]I will then make a definitive diagnosis and tell the patient that this [u]is[/u] a KA. [/i]

In cases where the lesion is otherwise typical (clinically and histologically) but has not shown signs of clinical regression at the time of biopsy, I will explain to the patient that this is likely to be a KA, though in the absence of signs of involution prior to excision, I cannot be absolutely certain. This is when I would call this a[i] 'Probable KA'[/i].

All other lesions, I would label and treat as a well-differentiated SCC ('KA-type' or 'No specific type').

One might argue that even in the first scenario, one cannot guarantee that the lesion is definitely a KA, as the regression inducing biological mechanisms might switch off after the lesion has partially regressed. Though theoretically this is possible, I have not seen this practically happen and would be interested to hear if any one else might have seen this happen.

[b][i]So when should the term 'KA-type SCC' be used?[/i][/b]
Personally, I regard KA to be a self-regressing variant of an SCC, distinct from a well-differentiated SCC in its rapid clinical evolution and subsequent involution. However, for that reason I would not call all KA's, 'KA-type SCC' as that actually implies (by convention) that it is a slightly more aggressive tumour (i.e. which does not self-regress and hence theoretically has a higher potential for recurrence/ metastasis). I see no need to change the terminology of KA and call it SCC like or add any other adjectives, as 'KA' is suffice, irrespective of whether we consider it to be a benign end of the SCC spectrum (my opinion), or a distinct tumor.

In my opinion, the term 'KA-type SCC' should be only be used in 'invasive squamous cell carcinoma having some but not all features of keratoacanthoma' (as stipulated by the RCP) BUT only when the clinical history is not supportive or not available.

In cases where an experienced clinician is convinced of the lesion being a KA on clinical grounds, if the histological features are suggestive of a 'well differentiated squamous tumour, with or without early invasion', the statement ''The features would support the diagnosis of a KA . However a well-differentiated SCC cannot be excluded on histological grounds', would be more helpful to the clinician. Obviously irrespective of history, if the tumour is anything worse than 'well' differentiated, one cannot make the diagnosis of a KA! Similarly I would not make a definitive diagnosis KA if the tumour has metastasized (which is against my definition of a KA i.e. self-regressing). If that is the case, by that logic, one might also say that 'Spitz Naevi are largely benign but can rarely metastasize and kill patients' B). A spitz naevus that has metastasized and killed a patient was a Melanoma to start with and the same should apply to a KA! One can accept metastasis as a feature of KA, if the metastasis itself regresses spontaneously. I shall address these issues in a different blog post.

[i][b]Does it matter whether we call this a 'KA' or a 'KA-type SCC'?[/b][/i]
It certainly does matter in the UK, as far as coding the diagnosis and meeting 'targets' is concerned.

If a lesion is diagnosed as SCC – all treatments will be under [url="http://www.nhs.uk/NHSEngland/NSF/Pages/Cancer.aspx"]either a 62 day or 31 day treatment target[/url] and the patient will be registered on the national database with a skin cancer.

If we diagnose the lesion as a KA – there are no treatment targets attached and the patient will not be registered at all on the national database.
[center][i]Yes, Unfortunately in the UK we have ended up treating targets rather than patients! [/i][/center]
[center][i](targetologist vs dermatologist!) [/i][i] :wacko:[/i][/center]

So if we start labelling all KA's as SCC:

1. No clinician would then dare take the 'just watch' approach which is entirely reasonable in[i] some[/i] circumstances.
2. Some patients might potentially breach their waiting times target (as clinicians might delay surgery because they may want to watch the tumor), which would not go down well with the powers higher up!

However, if we do not label KA as SCC:

1. A self-regressing skin cancer would not be entered into the national skin cancer registry. (I think it should be entered into the national database as a skin cancer, albeit self regressing).
2. There is a risk that inexperienced clinicians might undertreat these lesions.

I don't think there is an easy answer to this. A consensus decision has to be taken at the local MDT level (depending on clinician experience/ comfort), as to how to label these tumors.

But yes, I do also think that it is more reassuring for the patient to know that something is a tumour that is unlikely to kill, vs. something that has a higher potential (albeit still very low) to recur/ metastasize .

However, whether we call this KA or 'KA-type SCC' it usually does not change the follow-up for the patient, as both are well-differentiated tumours for which the consensus opinion among colleagues is that no further followup is required, once excised. This notion might not actually be true for rapidly growing, well differentiated SCC. But I shall discuss that in a separate blog post.

[center][i]What is your practice? ^_^[/i][/center]

1. Weedon D. Skin pathology. 3rd ed. London: Churchill Livingstone; 2010

2. Griffiths RW. Keratoacanthoma observed. Br J Plast Surg 2004; 57: 485–501.


Recommended Comments

Mark A. Hurt MD


This has been an intriguing question now for over a century. I think it boils down to this: hyperplasia versus neoplasia.

Hyperplasia is stimulus dependent; pseudocaecinomatous hyperplasia, in its exuberent forms will challenge the diagnosis of SCC. I'm sure anyone with some experience will have seen a case of blastomycosis that has been interpreted as SCC, only to shudder in horror about the misdiagnosis. Some keratoacanthomas and some SCC's are no too far, morphologically, from the pseudocaecinomatous changes of blastomycosis.

Or, consider exuberent cases of prurigo nodularis. These cases surely can mimic lesions of SCC. There is a specific presentation of PN that occurs on the shin -- I call it "prurigo en plaque" -- that mimics some SCC's, but it doesn't have the biology of an SCC.

In contrast, there are well-differentiated SCC's that have metastasized but which so closely mimic the classic pattern of what is considered a classical KA. These lesions appear to be stimulus independent in that they continue to grow despite attempts at (incomplete) removal.

I remember a conversation with Bernie Ackerman, who was a champion of the concept of KA as SCC. He indicated to me that in prior years, he championed KA's as hyperplasia, but he was later convinced that all were neoplastic when identified by repeatable and reliable criteria.

Weedon, in contrast, has continued to champion the concept of KA as "benign."

Obviously, someone has to be right, and the other, wrong. On what criteria can one decide?

My opinion is that histopathology by H&E is inadequate to determine the answer. My sense of it is that KA is hyperplasia; it is a real condition in nature, but it so closely mimics neoplasia (SCC) that, for practical purposes, many cases will be considered SCC.

When I was a medical student (1978-1982), I was told that unless 20% of normal apendices were removed, based on clinical criteria, not enough of them were being removed. I think that KA is similar, even though I regard it as hyperplasia (with expected involution) rather than neoplasia (continued stimulus independent growth).
Link to comment
Thanks Dr. Hurt. I don't quite agree that KA, just because the biological process of tumour formation also has an in-built senescence mechanism, is necessarily 'hyperplasia'. Going by that reasoning, a regressed melanoma would have to be called hyperplasia, which I am sure you would agree is incorrect. I am in the process of writing a separate blog post on this issue.
Link to comment
Mark A. Hurt MD


Dear Sasi,

Your point is well taken about the comparison of regressed melanoma; in fact, your point is THE key point in the discussion. My argument, of course, is that hyperplasia, by its nature, is stimulus dependent. Thus, one would expect that most of the cases would regress. In KA, it is most of the time; in melanoma, it is not--it is the exception.

Link to comment
Dr. Richard Carr


Thank Sasi, I am preparing for lectures in London next week and despite this topic being my pet favourite just don't have the time to digest the above!! However as you are well aware I diagnosis the highest rate of KA in the UK (30% of either SCC and KA cases are signed out variously as "typical of", "in keeping with" or "favouring" KA with clinicopathological correlation essential. You are right many of these KA had marked cytological atypa (I usually comment in the report of doubtful significance but in a partial biopsy would prompt me to suggest complete excision). So far none of the latter lesions have gone bad so to speak but it might be KA (at bit like Spitz tumours) rarely (but not to rare it is not recorded in case reports) has the biology to metastise and kill. One of my main points (regarding KA v's WDSCQ) being that no "well differentiated" SQC grows to 1 cm size in 6 to 8 weeks and looks clinically and histologically like a KA. All KA are invasive in the early proliferative phase in fact the proliferative infiltrative border with elastic entrapment and full often abrupt differentiation to pilar keratin (towards the base) or infundibular keratin (more superficially and within the crater is the main distinguisher from follicular SQC (that usually have rounded pushing, non-infiltrative borders) although profile of a rounded exoendophytic tumour can be on occasions quite similar. By the way the case you illustrate is a typical KA as we discussed by e-mail. There are many caveats to the diagnosis and exceptions to rules. I will be lecturing on my approach to follicular SQC and KA at the RCPath in November and I can only recommend you attend and fuel the debate. I am not sure my approach can be translated to the "coal face" so my lecture comes with a health warning! Sometimes a little knowledge can be a dangerous thing.
Link to comment
Dr. Richard Carr


Forgot to say I will also be speaking on my approach to KA and SQC at the London Dermpath Symposium May 14th to 16th 2014. Please sign up for that meeting too!
Link to comment
Thanks Richard. I can certainly think of no other person in the UK who has studied KA as extensively as you have. So hence wanted your opinion. I shall try and get this published and hope the RCP takes heed.

Dr. Hurt- I can see your view point. Even malignancy is actually stimulus dependant. The only difference in KA is that the stimulus is short-lived and then regressive stimuli take over.

I would also argue that the presence of perineural invasion, vascular invasion, malignant transformation in KA (in one study, up to a quarter of typical KA's were reported to undergo malignant transformation to SCC) and the rare cases of metastasis reported even in so called typical KA, all point towards KA being part of the SCC spectrum rather than simple hyperplasia. In no other hyperplastic disease, do you see these features. One might argue that like in a marjolin's ulcer, chronic inflammation can induce malignant transformation. However KA is typically an acute process and not chronic.
Link to comment
Mark A. Hurt MD



Of course, hyperplasia and neoplasia can be caused by one or more factors; my point is that the difference between hyperplasia and neoplasia is the difference of stimulus dependency. My understanding of this difference is that hyperplasia will revert to a control or near control state when the stimulus is removed; whereas, neoplasia will not.
Link to comment
  • Create New...