Cutaneous melanoma
Section 3: Diagnosis and prognostic indicators

The vast majority of melanomas are visible, if not to the patient, then at least to friends, family or health professionals. Members of the general public and health professionals should be aware of the signs suggestive of melanoma. In Scotland, melanomas occur more commonly in women than men. The most frequent site is the leg for women and the trunk in men. A small number of patients have occult primary lesions and present with metastatic disease. Up to ten percent of melanomas can be amelanotic (non-pigmented), increasing diagnostic difficulty.

3.1 Types of melanoma

Melanomas are subdivided into types on the basis of clinical features and pathology.

3.1.1 SUPERFICIAL SPREADING MELANOMA (SSM)

This is the most frequently encountered type of melanoma; characteristically an asymmetrical pigmented lesion with irregular borders, irregular pigmentation and sometimes an irregular outline. Patients may have noted growth, a change in sensation and/or colour, crusting, bleeding or inflammation of the lesion. The duration of the symptoms varies from a few months to several years.

3.1.2 NODULAR MELANOMA (NM)

The second most common type is nodular melanoma. This usually has a shorter presentation and a greater tendency to bleed and/or ulcerate.

3.1.3 LENTIGO MALIGNA MELANOMA (LMM)

The next in frequency is the type that occurs most often in sun damaged skin on the head and neck of older patients. This is the only variety that has a clearly recognised and often lengthy pre-invasive (in situ) lesion termed Lentigo Maligna (LM) before progressing in some instances to an invasive melanoma (LMM).

3.1.4 ACRAL LENTIGINOUS MELANOMA (ALM)

The least common type of melanoma in Scotland is the acral lentiginous melanoma. This occurs on sites including the palms, soles and beneath the nails.

3.2 Clinical diagnosis

Suspicious pigmented lesions are best examined in a good light with or without magnification and should be assessed using the 7 point checklist or ABCDE systems given below.39, 40 The presence of any major feature in the 7 point checklist, or any of the features in the ABCDE system, is an indication for referral. The presence of minor features should increase suspicion. It is accepted that some melanomas will have no major features. Evidence level 4

Table 3: The 7 point checklist lesion system

Table 4: The ABCDE lesion system

Clinical diagnosis of melanoma is difficult and the accuracy of diagnosis may vary according to a clinician’s level of experience, with reports of considerable variation in sensitivity from 50 to 86% and an inverse relationship between sensitivity and experience. 41, 42, 43 Evidence level 3

High magnification dermatoscopy is more sensitive than non-dermatoscopic diagnosis when used by clinicians with experience of the technique.44 High magnification requires more sophisticated equipment than the current widely-used hand held dermatoscopy. Evidence level 1+

Training clinicians to be ‘experts’ in hand held dermatoscopy improves diagnostic accuracy but it may diminish the sensitivity of the diagnosis of ‘non-expert’ or untrained dermatologists.45, 46, 47 Observational studies have compared excision and pathological assessment to using other preoperative assessment methods of diagnosis including magnetic resonance imaging (MRI), high resolution ultrasound (US) and digital imaging of possible melanomas.48, 49, 50, 51 These studies failed to show significant benefit and require specialist equipment that is unlikely to be widely available. Evidence level 1+,3


A flow chart showing the suggested management of cutaneous melanoma is given at the end of this chapter.

3.2.1 THE UNKNOWN PRIMARY

When melanoma presents as a metastasis and no primary melanoma site can be identified the patient should be referred to the appropriate regional specialist (see section 8).

3.3 Delay in diagnosis

Nine observational studies exploring delay were identified.41, 52, 53, 54, 55, 56, 57, 58, 59 Significant delays (>three months) in diagnosis of invasive melanoma are usually patient- rather than physician-related.41, 52, 53, 54, 55, 56, 57, 58, 59 Delay was defined differently in each study, with some including both patient and physician components. Evidence level 3

All of the studies identified show inconsistency between Breslow thickness (see section 3.8.4) and delay, although melanomas diagnosed incidentally by health professionals were consistently thinner than those noted by patients themselves.55 Evidence level 2+

Several studies showed longer delays in older patients,42, 57 in men, in rural versus urban dwellers and in plantar melanomas.58, 42 Evidence level 2+

There is inconsistency in findings regarding patients’ knowledge of melanoma and delay. Two observational studies found that delay in presentation was shorter if the patient was aware of possibility of malignancy.55, 59 Conversely, another study found that delays were longer in those with greater knowledge, perhaps due to false reassurance caused by greater knowledge42 (see section 2.4.1). Evidence level 2+

Physician delay accounts for a very small part of the total delay in diagnosis.41 Medical delays were shorter and the Breslow thickness was less when patients were seen by dermatologists as opposed to general practitioners.41 Evidence level 3

No studies were identified exploring the consequences of delays in diagnosis on patient outcomes.



3.4 Educating health professionals about diagnosis

An Australian RCT demonstrated a decrease in the number of benign lesions excised by GPs after being given algorithms and cameras as aids to diagnosis.60 In an American RCT, the use of a booklet, magnifying and measuring tools and feedback sessions improved the ability of primary care residents to triage suspicious lesions.61 Evidence level 1+,1-

3.5 Biopsy of suspicious lesions

Biopsy of highly suspicious lesions should be by fast-track referral to the appropriate specialty according to local circumstances.

The optimal specimen for full histological evaluation of a suspected melanoma is a complete excision with a 2 mm surround of normal skin and a cuff of fat.62 This enables assessment of the entire lesion and allows direct wound closure with a relatively short scar that will not compromise subsequent wider surgery (see section 4.1). Elliptical excisions should be performed along the long axis in the line of a natural skin crease or longitudinally in limbs and transversely over joints. The exact surgical margins of excision should be recorded on the operation note. Evidence level 2+

Non-excisional biopsy may lead to inadequate histology.63, 64, 65, 66, 67 The least useful type of biopsy is the superficial shave variety. Two large studies demonstrate that non-excisional biopsy of the primary lesion has no effect on prognosis.64, 68 Evidence level 2+






3.6 Lentigo maligna melanoma

Biopsy of in situ lentigo maligna melanoma (LMM) and of invasive LMM should be approached differently. The frequently facial site and large diameter of such lesions may render full excision difficult or excessively destructive. In these instances incisional biopsy(s) of the most clinically suspicious areas are appropriate. Lentigo maligna melanoma ideally should be surgically removed as in other invasive types.69 Where the size, site or patient comorbidity prevents this, other techniques can be considered. Cryotherapy and topical-5-fluorouracil have been used but there is a risk of local recurrence and no reduction in the risk of developing invasive disease.70 The patient should be fully appraised of these risks. Several small studies also report the use of radiotherapy in pre-malignant LM and LMM with good recurrence-free intervals and excellent cosmesis.71, 72, 73 Evidence level 2+

3.7 Pathological diagnosis

3.7.1 HANDLING A SUSPECTED MELANOMA

The volume of evidence addressing the handling of suspected melanomas is small. Recommendations on how to describe and select tissue blocks from a suspected melanoma are available from standard surgical pathology textbooks.74

Accepted practice has been supplemented by best practice guidelines published in 2000.62 Evidence level 4

Appropriate treatment, follow up and prognostication for patients with melanoma are entirely dependent on accurate pathological diagnosis and microscopic staging. The macroscopic description of the specimen, together with adequate and appropriate methods of block selection, is central to this process.


Note: a photograph of the macroscopic specimen may be of great value, especially if the precise origins of labelled blocks are drawn onto the photograph to permit exact orientation.

3.8 Prognostic indicators

3.8.1 HISTOGENETIC TYPE

Superficial spreading melanomas are less prone to recurrence when compared to nodular melanomas, however both groups have similar survival figures when matched for tumour thickness.75 Cochran’s prognostic model (see section 3.8.11) uses an assessment of the histogenetic type (SSM versus all others) to calculate the risk of tumour recurrence but this variable is not required for the calculation to derive the probability of survival.75 Classifying melanomas into different subtypes is reported to be of no prognostic relevance. Both desmoplastic and neurotropic melanomas (long regarded as being particularly aggressive) have similar survival rates to other subtypes although lesions with a neurotropic component are associated with a higher rate of local recurrence.62, 76, 77, 78 Evidence level 2++,2+,4

Although the majority of studies do not demonstrate a significant association between histogenetic subtype and patient outcome, histogenetic type does appear to play a role in determining the likelihood of recurrence. Histogenetic type also defines a group of well recognised clinicopathological entities. Evidence level 2++

3.8 2 MELANOMA CELL TYPE

The relationship between cell type and outcome is unclear. Studies that examine melanomas of all thicknesses tend to find either no evidence of a relationship or a weak association that is lost on multivariate analysis.75, 76, 79, 80, 81, 90, 258 The effect of cell type on prognosis may only become apparent in the specific subgroup of patients with thick tumours exhibiting paradoxical behaviour. Here, the presence of a spindle cell or Spitz-like phenotype may confer a modest survival advantage.81 Evidence level 2-

3.8.3 RADIAL VERSUS VERTICAL GROWTH PHASE

Tumour growth phase correlates strongly with clinical outcome.76, 82 A study of 501 patients with primary melanomas identified a subgroup of 122 as being in radial growth phase only. No patients in this subgroup showed evidence of metastatic disease during a minimum follow up period of 100 months. The OR for a patient with radial growth phase melanoma surviving for eight years was given as 1.0.76 A second study evaluated 624 patients, of whom 161 had melanoma displaying radial growth phase characteristics only. None of the patients developed metastatic disease at long term follow up (median 13.7 years).82 Evidence level 2++

3.8.4 BRESLOW THICKNESS

A strong association between tumour thickness and prognosis was originally demonstrated by Breslow83 and has since been verified in many large scale studies of melanoma.75, 76, 84 , 85, 110 Breslow thickness is the single most important prognostic variable in primary cutaneous melanoma. Evidence level 2+,2++

3.8.5 CLARK LEVEL

In a study of 5,093 patients (minimum follow up seven years) the Clark level of invasion yielded additional prognostic information in patients with a Breslow thickness of <1mm.86 This observation has also been confirmed by the large study (based on 17,600 patients) used to derive the new American Joint Committee on Cancer (AJCC) staging system (see section 4.1).110 A small study of patients with metastasising lesions <1.5 mm noted significantly higher metastases of level IV lesions compared to control groups.87 A detailed systematic review of the staging of melanoma reports similar results.88 Evidence level 2++,2+,1+

3.8.6 ULCERATION

A small study of 177 subjects with intermediate thickness melanomas (1.51 to 3.99 mm) identified epidermal ulceration as one (of four) variables that predicted visceral and bony metastases.89 Ulceration has been shown to act as a prognostic variable after adjustment for other variables.76, 84 A study of 1,042 patients identified epidermal ulceration as a significant prognostic variable and this was incorporated into a mathematical model for predicting recurrence and survival at three, five and ten years.75 Ulceration was confirmed as a strong prognostic variable in the recent large scale study that validates the AJCC guidelines.110 Some studies also show that increasing breadth of epidermal ulceration is associated with an increasingly unfavourable prognosis (see section 3.8.11).75 Evidence level 2+


3.8.7 MITOTIC RATE

Although some studies have identified mitotic rate as a significant prognostic variable,76, 90 with survival probability decreasing in a linear relationship with increasing mitotic rate, the exact strength of this parameter remains unclear. Mitotic rate appears to be a less useful predictor of survival than variables such as tumour thickness and epidermal ulceration and it is not used as a staging parameter in the revised AJCC guidelines.110 In some studies the prognostic significance of mitotic index is either greatly weakened or entirely subsumed after multivariate analysis assesses other histological and clinical variables.75, 90 Evidence level 2++

3.8.8 INFLAMMATORY REACTION

The association between survival advantage and the presence of tumour infiltrating lymphocytes (TILs) within the vertical growth phase component is unclear. Although one study demonstrated a strong correlation,76 the presence of an inflammatory response loses independent prognostic strength on multivariate modelling.75 Evidence level 2++

3.8.9 REGRESSION

There is an adverse association between histological evidence of regression and outcome, but the strength of this relationship is disputed.75, 76, 87 One large study identified tumour regression in the radial growth phase as a variable that retained predictive strength after multivariate analysis.76 In a subsequent study of 1,042 patients the significance of tumour regression was subsumed by the other clinical and histological features studied.75 Extensive late regression might indicate that the melanoma has, at some time, been significantly thicker than it now appears. Tumours with this feature are liable to be understaged.87 Evidence level 2++,2+

3.8.10 LYMPHOVASCULAR INVASION AND SATELLITES

A systematic review found that the prognosis for patients with microsatellites is essentially identical to that for patients with macrosatellites.88 There was no demonstrable difference in survival for patients with satellites compared to those with in-transit metastases. Evidence level 1+

A prospective cohort study of 258 patients with clinical stage I melanoma found that 13 out of 14 patients with histological evidence of lymphatic invasion developed in-transit metastases after a median interval of 10 months and concluded that lymphatic invasion correlates strongly with early locoregional cutaneous relapse.91 Evidence level 2++

A study of 140 patients with thick (>3 mm) stage I melanomas reported that the identification of lymphatic invasion was associated with an increased risk of metastasis but not with overall survival.90 However, in a series of 17,600 patients the presence of microsatellites had a profound negative impact on prognosis and in the new AJCC staging system the presence of satellites upstages the tumour from I or II to IIIb or IIIc.110 Evidence level 2+

Identifying lymphovascular invasion and/or microscopic satellites confers considerable prognostic value. The presence of lymphatic invasion accurately predicts early cutaneous relapse and should be included as a stratification criterion for the selection of patients for adjuvant therapy. The histological identification of microsatellites also defines a subset of patients at much greater risk of relapse. The presence of microsatellites correlates strongly with occult metastatic disease in regional lymph nodes. Evidence level 2+

3.8.11 DERIVING A PROGNOSTIC INDEX

Models attempting to predict accurate survival for patients with melanoma have become increasingly complex in order to accommodate both the clinical and histopathological data demonstrated to be of independent prognostic significance.75, 76, 92 The Cochran model predicts survival at three, five and ten years, based on simple clinical data (age, sex and anatomic site) in combination with the Breslow thickness and the breadth of epidermal ulceration.75 An individualised risk score is derived that predicts the likelihood of survival for any patient with invasive melanoma. The model can be modified to calculate risk of recurrence (at three, five and ten years) by inclusion of histogenetic subtype. (Worked examples of how to use Cochran’s model are available on the SIGN website at www.sign.ac.uk). Evidence level 2++


3.9 Specialist pathology reporting

Significant discrepancy exists between general pathologists, dermatopathologists and between experts in pigmented lesion pathology, in the reporting of melanocytic tumours.93, 94, 95 Both under- and over-diagnosis of malignancy is recognised and, for melanoma, there is poor agreement on the assessment of prognostic parameters. Evidence level 2++


3.10 Melanoma pathology report

The table below outlines the important features of a melanoma pathology report.

Table 5: Features of a melanoma pathology report

3.11 Pathological examination and reporting of therapeutic and sentinel lymph node dissection specimens

Detailed protocols for dissection of therapeutic lymph node dissection specimens are available in standard textbooks of surgical pathology.74, 96

The surgical report should identify

When macroscopic examination reveals tumour within a node, a single block of tissue is sufficient to confirm the observation. Nodes that appear tumour free should be serially sliced (if large) and all of the tissue processed. Small nodes may be processed intact and levelled to ensure thorough examination.

Sentinel lymph node biopsies (SNLB) are processed using either lymphoscintigraphy and/or blue dye to trace the afferent lymphatic channels and node. Protocols giving further details are available. 96, 98, 99 Nodes identified by lymphoscintigraphy (usually technetium99) should be fixed in formalin for 24 hours to allow for radioactive decay. Evidence level 4

When dye has been used, the sentinel node should be examined macroscopically to determine whether any staining has occurred. The node should then be bisected through its longest circumference and both halves processed for the removal of ten serial sections. Sections one, three, five and ten should be examined by routine haematoxylin and eosin staining. Sections two and four should be stained immunohistochemically for S-100 protein and HMB-45. Sections six and seven act as negative controls for immunohistochemistry (IHC) and sections eight and nine can be used for additional studies or to repeat any unsatisfactory stains.99 If the appearances in the first set of ten sections are deemed suspicious then additional sets of ten sections can be cut and stained. Evidence level 4

Although IHC facilitates the detection of melanoma in sentinel nodes, the possibility of false positive results, for example, the misinterpretation of capsular naevus cells, remains. This can be minimised by careful evaluation of the immunochemical preparations in the context of the corresponding haematoxylin and eosin stained section. Evidence level 4

Groups of sections at multiple levels throughout the sentinel node are sometimes examined, but there is no evidence that such rigorous sampling increases the diagnostic yield. Detecting melanoma cells in SNLB using polymerase chain reaction (PCR) techniques cannot be recommended at present due to concerns regarding both sensitivity and specificity.100 Evidence level 4

[Contents][Back][Top] [Next]

Scottish Intercollegiate Guidelines Network
Home