Notes from a master of dermatoscopy
I recently had the privilege of attending a dermatoscopy (his preferred term) master class by Professor Harald Kittler at the Primary Care Dermatology Society’s spring meeting in Manchester.
I made extensive notes during the 2 and a half hours of his presentation. Here they are (edited). Please note, the above bio is somewhat dated and excludes his magnificent contributions to dermatoscopy since 2007 particularly.
On Saturday 12th March I attended a dermoscopy master class of about 150 minutes delivered by Professor Harald Kittler from the University of Vienna, given in Manchester at the spring conference of the Primary Care Dermatology Society. I took contemporaneous notes and tidied them up afterwards. Wherever possible I have used Prof Kittler’s words verbatim.
“Dermatoscopy adds information. Which type of scope is best? Contact vs non contact-it’s swings and roundabouts- you’d better have both. White lines and white structures are seen better with polarising.
Most melanomas do not develop from a pre-existing benign naevus, that’s a myth. We now think that only 20% of melanomas arise from a pre-existing benign naevus, most arise de novo from clear skin. Where the myth comes from that melanomas develop from banal naevi is that small, early melanomas look like banal naevi. (Giuseppe Argenziano makes the same point-funny looking moles remain funny looking if you monitor them, but they are no more likely to turn into melanomas than banal looking moles. The ones that reveal themselves to be melanomas were always melanomas from the start-SH) Also, the same faulty genes predispose to multiple naevi and melanoma risk, so the patient with multiple naevi is at higher risk of a melanoma, but not particularly at risk for their pre-existing benign naevi to turn malignant.
Clinical concept of melanoma based on advanced disease, in old books the concept of melanoma in situ did not exist…the concept was only introduced in 1980 by Ackerman. Lentigo maligna used to be called ‘senile lentigo’. A change in diagnostic standards was partly responsible for raised incidence. Textbook photos of the 1970s depict gross lesions, e.g. nodular and ulcerated. The 6mm rule for melanomas (D for diameter in the ABCD rule) is rubbish, believe it or not we humans were all once smaller than 6 mm!
The good thing about melanoma is it usually grows slowly. It is unethical to allow melanomas to grow unchecked until they are clinically obvious, therefore we use dermatoscopy to detect them at an earlier stage.
Kittler used a fast sequence of dermoscopic image slides asking for the audience to shout out ‘benign’ or ‘malignant’ as he changed slides every 2 or 3 seconds. Everyone (admittedly in an experienced group) got it right every time, and immediately. How did you do that right so quickly? he asked. There was no time to apply algorithms. Pattern recognition, he argued. He then used images of common animals for example zebra and Dalmatian (illustrating stripes versus dots) and Dalmatian and spaniel (to illustrate different colours in the same shape.)
Most malignancies on dermoscopy are chaotic-Kittler defines chaos on dermatoscopy as more than one pattern and/or more than one colour. He qualified this by saying that 2 patterns would usually be OK if combined symmetrically (for example a reticular periphery and amorphous centre).
‘You don’t have to learn much to diagnose skin cancers dermoscopically, screening for chaos does it, and fast.’
The concept of chaos allows us to screen many lesions quickly. If a lesion is not chaotic (excluding the nodular, firm growing lesion) then you are quickly reassured and move on. When you see chaos, pause and look for clues. The chaos and clues work well for most examples of melanoma and BCC. If there is chaos and at least one clue to malignancy, excise unless there are adequate clues to diagnose a seborrhoeic keratosis. He used many examples of chaotic seb ks to prove this point.
Blue structures may point to melanoma (or BCCs) but are seen in some benign lesions, context is vital.
A flat lesion that is ulcerated is very likely to be a BCC, melanomas don’t usually ulcerate when flat.
Keratin plug in follicular opening= comedo like opening. Pigment is seen at the periphery of the follicular opening, hence dark circles with orange keratin plug in the middle. (this can help discern comedo like openings in seb ks from globules of pigment that represent nests of abnormal melanocytes which are generally solid blocks of one colour).
Speaking about the new descriptive terminology (which he developed in 2007) versus the traditional metaphorical terminology, he made the subtle but telling point that when we use metaphorical terms (e.g. spoke wheel, leaf like, lacune etc) we are not describing what we see but interpreting it because we already ‘know’ what we are seeing.
For example, when we see red, blue or purple clods (descriptive terms) these point to a haemangioma, but when we say ‘lacunes’ (metaphorical term) we are saying this on the assumption that the lesion is a haemangioma. The descriptive term is neutral, and properly interpreted in context points to the diagnosis of a haemangioma. However, when we say ‘lacune’ we are arguing FROM the diagnosis. In both cases, most of us will have taken one look at the lesion and diagnosed a haemangioma in the same way that we recognise an elephant, a zebra or George Clooney-at a glance because we have seen it before and know.
Self fulfilling prophecies e.g. Leaf like structures in superficial BCC-drawback of metaphoric terms, you see things because you expect them to be there because of an assumption you have made about what the lesion is. Using the examples of ‘leaf like structures’ in superficial pigmented BCC, Prof Kittler demonstrated that these structures don’t really look much like leaves (using a silhouette of a leaf to prove this) and we are really shoe-horning the actual lesion features into metaphors. In a textbook image he used of a superficial BCC with half a dozen irregular pigmented areas, only one looked even vaguely like a leaf, the others which presumably were biologically identical didn’t look at all like a leaf. Probably therefore ‘leaf like structure’ is a term which is not specific enough to be much use.
He showed a slide demonstrating the explosion of publications about dermatoscopy since 2003, which had been accompanied by an explosion of jargon including newly invented metaphorical terms (not just leaf like, spoke wheel, cerebriform etc but watermelon, chrysalis, toffee bonbon, ashy hole, dirt tracks, firework, mistletoe, chrysanthemum and some 200 other terms). This has put people off, hence the need for a pruning and definition of signs.
5 basic elements (dots, clods, lines, circles and pseudopods) plus colours plus spatial arrangements covers everything and used well should work for describing lesions over the phone e.g. Orange and yellow clods plus broad curved lines = seb k
Dermatoscopy alone is nothing. You must always include other information, there are always exceptions. Some melanomas are not chaotic…new changing lesion and/or ugly duckling are massively important.
On atypical/dysplastic naevi (see also Professor Kittler’s videos ‘the myth of dysplastic naevi’ on YouTube).
We call some lesions atypical because the diagnosis is uncertain. If we are uncertain, we blame the lesion, not our brain. Don’t mix up diagnostic uncertainty with biological uncertainty. The lesion knows what it is (in most cases, atypical but benign)
The comparative approach increases sensitivity.
Patient with multiple atypical naevi, sometimes its hard to catch small flat melanomas early. We used to excise many slightly suspicious lesions that turned out benign on histology, patients don’t like it. They may even not like coming back. We now use digital monitoring.
When managing the patient with many atypical naevi, exclude obvious melanomas, then monitor. The one criterion which is always reliable is change over time. But don’t monitor obvious melanomas.
ADVANCED PATTERN ANALYSIS
Pattern analysis differentiates between benign and malignant, also diagnoses the specific entity. A pattern has to cover a significant part of the lesion…a few dots can be ignored. It’s often obvious, we train on lesions we do know so we can apply what we learn to lesions we don’t know
Pseudopods are helpful, Reed naevi or melanoma depending on if uniform (Reed naevi typically have even pseudopods all the way round the lesion) or occasional/irregular Pseudopods (more likely to be melanoma) plus other clues.
Blue clods with red clods= haemangioma
Blue clods with brown=BCC
White and yellow clods = seb k
‘The structureless pattern is a bitch’
Black can be melanin or congealed blood.
Structureless brown with no other features= solar lentigo especially in sun damaged skin
There is no algorithm that works for everyone, you must develop your own threshold.
Pattern recognition is inductive reasoning
The rules can be deductive…Francis Bacon
Sherlock Holmes is deductive…build up a hypothesis that fits best with objective facts
‘When you end up as an expert, you create your own rules and write your own books and lectures. Everybody should create their own algorithm’
Vascular patterns are less specific than pigmented structures but if that’s all there is you have to go with it
‘Vessels…we have to destroy the tower of Babel’
The descriptive terms can generally be applied to vessels, apart from pseudopods, there are no vessels which look like pseudopods (a clod on the end of a broad line). Vessels can be dots, clods, linear (curved, coiled, serpentine) or looped. Their spatial arrangement can be homogeneous (global feature of lesion) or irregular.
We evaluate morphology and spatial arrangement of vessels (Iris Zalaudek would add that we also look for non-vessel features of the lesion).
Dotted vessels typical for flat non pigmented melanoma
Flat BCCs typically have short, sharply focussed serpentine vessels, a flat hypopigmented melanoma is more likely to have dotted vessels.
Don’t be deceived by psoriasis!
Fabric fibre (clothing threads stuck to the lesion, visible on dermatoscopy) is indirect evidence of ulceration/erosion even if you can’t see it, which is a clue to malignancy.
Ulceration in a flat lesion that is not clinically visible is a clue to BCC
Beware of a flat lesion with serpentine and dotted vessels-it could be a hypomelanotic melanoma. But don’t get paranoid and see amelanotic melanomas everywhere, however but you don’t want to miss it…this deceitful and dangerous lesion is a cause of many lawsuits
Rule…only leave nodular lesions if you can make a specific benign lesion with certainty
Rule…all ulcerated lesions must be excised (not excoriated rashes)
Weird vessels in a pink nodule can be poorly differentiated SCC (can also be Merkel cell cancer, amelanotic melanoma or cutaneous secondary deposit-SH)
Well differentiated SCC will show keratin and white circles on dermoscopy, although usually the diagnosis will be on history and inspection. ‘I don’t differentiate between SCC and keratoacanthoma.’
Blood within keratin is a clue to SCC.
Dotted vessels in pink nodule= probably amelanotic nodule, vessels in nodular BCC are usually arborising
Vessels in seb gland hyperplasia do not cross the centre of the lesion, as the centre is a follicle.
Yellow clod= serum crust (if not seb k)
‘Simple rule: ulcerated nodule and its out’
‘When it comes to dangerous lesions, never trust your patient. They don’t usually lie but they may sincerely misremember or buy into a preferred narrative.’
That is my write up of Professor Harald Kittler’s morning session from the Manchester PCDS conference, with a few clarifications and additions but is 95% his words. I hope this may be of some interest. He is a wise, learned and generous man who has added to the growing understanding of how we can best use dermoscopy/dermatoscopy to refine skin lesion recognition for the benefit of our patients.
Dr Stephen Hayes, 24th March 2016.