D ER M A T O L O G I C A L S U R GE RY AN D L A S E R S
BJD
British Journal of Dermatology
Surveillance for treatment failure of lentigo maligna with dermoscopy and in vivo confocal microscopy: new descriptors P. Guitera,1,2,3 L.E. Haydu,1,4 S.W. Menzies,2,3 R.A. Scolyer,1,5,6 A. Hong,1,7 G.B. Fogarty,1,6 F. Gallardo8 and S. Segura8 1
Melanoma Institute Australia, Poche Centre, 40 Rocklands Road, North Sydney, NSW 2060, Australia Department of Sydney Melanoma Diagnostic Centre, 5Department of Tissue Pathology and Diagnostic Oncology and 7Department of Radiation Oncology, Royal Prince Alfred Hospital, Missenden Road Camperdown, NSW 2050, Australia 3 Discipline of Dermatology, 4Discipline of Surgery and 6Discipline of Pathology, The University of Sydney, Sydney, NSW, Australia 8 Department of Dermatology, Hospital el Mar, Barcelona, Spain 2
Summary Correspondence Pascale Guitera. E-mail: [email protected]
Accepted for publication 10 February 2014
Funding sources This study was supported by the Melanoma Institute Australia, the Melanoma Foundation of the University of Sydney, and the Australia and New Zealand Melanoma Trials Group.
Conflicts of interest None. DOI 10.1111/bjd.12839
Background Nonsurgical treatment (radiotherapy, imiquimod) is increasingly employed for the management of lentigo maligna (LM). While the diagnosis of LM remains difficult, the detection of treatment failure is even more challenging. Objectives To describe the sensitivity and specificity for the diagnosis of LM of individual features and methods using dermoscopy and in vivo reflectance confocal microscopy (RCM) to aid in the detection of treatment failure of LM following nonsurgical treatment. Methods A retrospective study of dermoscopy and RCM images (blinded to the correlation with pathology) in patients with biopsy-confirmed LM who were undergoing nonsurgical treatment in two referral institutions – one in Sydney, Australia, and the other in Barcelona, Spain. Ninety-eight patients were treated nonsurgically for LM during the period 2006–2012. Thirty-one patients had abnormal dermoscopy or RCM evaluation, and had a biopsy that identified LM recurrence in 15 patients and nonmelanoma diagnoses in 16 patients (one Bowen disease, 15 solar changes). Results The diagnosis of treatment failure was difficult with dermoscopy, with a sensitivity of 80% and specificity of 56%, even with the interpretation of an expert. The best criterion was asymmetric hyperpigmented follicular openings, but this was present in only 47% of treatment failure LM. Isolated, very fine brown dots (‘dust’ appearance) correlated highly with the diagnosis of treatment failure LM (73% sensitivity and 88% specificity) and with pagetoid cells seen with RCM. The LM score, comprising six criteria, had a specificity of 94% and sensitivity of 100%. Conclusions These methods and descriptors should help to manage the diagnosis of treatment failure.
What’s already known about this topic?
•
© 2014 British Association of Dermatologists
Clinical diagnosis of recurrent lentigo maligna (LM) following treatment is challenging because treatment-induced inflammation and pigmentation are common, but so are amelanotic recurrences.
British Journal of Dermatology (2014) 170, pp1305–1312
1305
1306 Surveillance for treatment failure of lentigo maligna, P. Guitera et al.
What does this study add?
•
We tested the sensitivity and specificity of individual features and methods using dermoscopy and in vivo reflectance confocal microscopy for the diagnosis of LM and give recommendations for management.
Lentigo maligna (LM) is a form of melanoma in situ that occurs on exposed, sun-damaged skin of elderly people. LM represents a challenge for management; LM has a comparatively higher local recurrence rate,1,2 most likely because its margins are difficult to define clinically and histologically. Optimal treatment is still debated, and there are no prospective studies or randomized controlled trials.3 Staged excision is recognized as the best treatment option as it facilitates pathological assessment of the entire lesion to rule out invasive melanoma. However, surgery may be contraindicated owing to the extent of disease, comorbidities or patient preference. Significant morbidity can result from wide surgical excisions.4 Moreover, medical treatments can treat a larger field, in the hope of decreasing treatment failure rates and reducing cosmetic issues. Radiation therapy (RT) is currently the ‘standard’ nonsurgical alternative, but there is only weak evidence to support its use.5 The cosmetic outcome is better with the use of imiquimod (IQ), but only case series have been published, and the treatment failure with IQ appears to be higher than with RT.6–9 Clinical and dermoscopy diagnosis of LM is often difficult because of the heterogeneity and contiguous lesions on sun damaged elderly skin.10,11 These diagnoses are even more difficult when trying to detect treatment failure as treatmentinduced inflammation can lead to pigmentation; pigment incontinence is a well-known pitfall.8 Alternatively, amelanotic LM recurrences have also led to false negative diagnoses.9 In vivo reflectance confocal microscopy (RCM) provides cellular resolution in the upper layers of the skin. Because melanosomes appear bright under reflectance at near-infrared wavelengths, pigmented cells are easily visualized.12 Furthermore, as RCM generates a horizontal view of at least 8 9 8 mm, it is possible to assess more of the lesion using this technique than with the pathological assessment of vertically orientated small biopsy specimens examined in routine histopathology.13 We have recently described an algorithm (LM score) to distinguish LM from benign macules,14 which resulted in a sensitivity of 85% and specificity of 76% for the noninvasive diagnosis of LM [odds ratio (OR) = 186]. The algorithm was equally effective for the diagnosis of amelanotic lesions. Moreover, melanophages, which represent one of the main pitfalls of treatment failure evaluation, can be distinguished from melanocytes with RCM.15 The aim of this study was to describe how dermoscopy and RCM can help in the detection of recurrent LM previously treated with IQ and/or RT. British Journal of Dermatology (2014) 170, pp1305–1312
Patients and methods Patients were treated with IQ or RT for a histologically confirmed LM at two referral centres (Barcelona, Spain, and Sydney, Australia). Nonsurgical treatments were chosen because surgical excision was not possible, refused or failed. IQ was applied five times a week for 3 months with a margin of at least 1 cm beyond the clinically identified margins of the LM. Inflammation induction due to IQ was utilized to individualize dose titration for each patient. Some patients required a dose escalation, others had a painful reaction and their dose was decreased, but all patients completed 12 weeks of treatment. In terms of RT, the surface doses were generally around 50 Gy, but never exceeded 66 Gy. A minimum of three, but generally four or five fractions, were administered per week for a total of 4–6 weeks. Histological confirmation was sought if dermoscopy examination showed some pigmentation and/or RCM imaging showed some of the LM score criteria. Dermoscopy was performed on the entire field of the previous LM diagnosis of the previously diagnosed LM and surrounding sites if they appeared suspicious. Confocal microscopy was centred on the site of the previous LM diagnostic biopsy and on pigmentation sites detected by dermoscopy. In patients with no evidence of LM recurrence on clinical, dermoscopy and RCM examinations, a follow-up of at least yearly review was performed. This was a retrospective study approved by the local Human Ethics Review Committee (Sydney South West Area, protocol number X05-0218). Dermoscopic diagnosis Vivacam (CaliberID, Rochester, NY, U.S.A.) and Dermlite Foto (3 Gen, San Juan Capistrano, CA, U.S.A.) images were reviewed retrospectively by two observers blinded to the biopsy results. The criteria reported by Schiffner et al.10 and Pralong et al.11 were employed in the evaluation. Any other dermoscopy features found were also recorded. An expert dermoscopist, blinded to the research topic, reviewed all dermoscopy photographs, provided a diagnosis and was asked to confirm the presence or absence of new features. Confocal diagnosis All cases were recorded prior to biopsy using RCM (Vivascope 1500; CaliberID, Henrietta, NY, U.S.A.). Instrument © 2014 British Association of Dermatologists
Surveillance for treatment failure of lentigo maligna, P. Guitera et al. 1307
and acquisition procedures are described elsewhere.16 Each image corresponds to a horizontal section of 05 9 05 mm2 and a sequence of montage images (‘mosaic’ images) was acquired at the level of the dermoepidermal junction to explore an 8 9 8 mm2 field of view. Confocal sections, beginning at the stratum corneum and ending inside the papillary dermis, were recorded in the middle of the field and at all areas of interest. All recorded images were retrospectively reviewed by two observers blinded to the final biopsy results in each centre to assess the LM score criteria.14 Additional features such as epidermal disarray, dendritic pagetoid cells, edged papilla, nonvisible papilla, melanophages,15 highly reflecting fibres (described as solar elastosis feature under RCM)17 and enlarged vessels were also recorded. Statistical methods Sensitivity and specificity for the detection of LM were calculated for each dermoscopic and RCM feature using histopathological diagnosis as the gold standard. The ORs were calculated using cross-tabulations for a 2 9 2 table. The kappa statistic (j) measured the interobserver reproducibility rating (j > 075 excellent; 04 ≤ j ≤ 075 good; and 0 ≤ j < 04 marginal). All analyses were conducted with the SPSS Statistic v21 software package (IBM, Armonk, NY, U.S.A.).
Results Ninety-eight patients were treated with IQ or RT for a LM during the period 2006–2012. Twenty-seven had inadequate follow-up (< 6 months) or insufficient imagery to be assessed (patients were followed up locally where RCM and/or dermoscopy were not available). Forty patients had normal skin under dermoscopy and RCM and therefore did not have a biopsy. Thirty-one patients were biopsied because they had abnormal dermoscopy and/or RCM evaluation. Treatment failure was diagnosed by pathology in 15 of the 31 patients who had a biopsy. However, the treatment failure ratio diagnosed by each modality was calculated also considering cases with
normal imagery and an average follow-up of 12 months (range = 6–42 months) as negative. We obtained 11% and 28% treatment failure for IQ and RT, respectively. Of note, only one in 15 patients had a treatment failure outside the treatment field, and 10 had a significantly reduced area to treat (indicating that the surface of the LM was smaller after IQ or RT treatment). Eleven patients had their LM recurrence treated by surgery, while four patients were treated with IQ. Table 1 describes the characteristics of the 31 patients in whom treatment failure was suspected by dermoscopy and/or RCM and correlation between imagery and pathology was possible. Fifteen cases had a confirmed treatment failure, whereas 16 patients had no diagnostic evidence of LM on pathologic assessment. Among the 15 patients who had a confirmed treatment failure, two had lentigo maligna melanoma (LMM) with a Breslow thickness of 04 and 06 mm, respectively. These two patients had obvious recurring small areas of pigmentation in the field of treatment after a few months of IQ and RT, respectively, and were then treated with surgery. Four patients had no pigment at all on dermoscopy and two patients had pigment visible only with dermoscopy. Twenty-three patients had obvious pigmentation in the field of treatment and in two patients outside the field of treatment. All except one treatment failures confirmed by pathology were identified in areas with visible pigmentation, meaning that one of 15 patients had an amelanotic recurrence. The sensitivity and specificity of each diagnostic criterion, compared with the gold standard (histopathological diagnosis), are summarized in Table 2. The asymmetric pigment around follicular openings was the most prevalent criterion (32%) with a low sensitivity (47%) but high specificity (81%). An annular-granular pattern was also present in numerous cases (29%) with an even lower sensitivity (20%) and specificity (62%). The pigmented rhomboidal structures had a high specificity (94%) but low sensitivity (20%), and obliteration of hair follicles was never found. There was good interobserver agreement between the two observers for the overall Schiffner criteria compared with the LM diagnosis of the expert dermoscopist (n = 31, j = 0534, P = 0003).10
Table 1 Patient treatment and follow up
Number of patients treated (n = 98) Number of patients in whom imagerya follow-up was not available (n = 27) Number of patients in whom imagery was normal with average follow-up > 12 months (range 6–42) Number of patients in whom imagery was suspicious for treatment failure and biopsy correlation (see Table 1) (n = 31) Number of patients with treatment failure LM (% of treatment failure by treatment category) (n = 15)
IQ
RT
39 11
59 16
19
21
9
22
3 (3/19 + 9) = 11%
12 (12/21 + 22) = 28%
LM, lentigo maligna; IQ, imiquimod; RT, radiotherapy. aImagery refers to dermoscopy and reflectance confocal microscopy.
© 2014 British Association of Dermatologists
British Journal of Dermatology (2014) 170, pp1305–1312
1308 Surveillance for treatment failure of lentigo maligna, P. Guitera et al. Table 2 Characteristics of the cohort when imagery and pathology correlation were performed (N = 31) Feature
Value
n
Sex
Female Male Median (range) years Scalp Forehead Cheek Eyelid Nose Temple Median (range) years
23 8 73 (35–89)
Age Site
Time between end of treatment and biopsy Number of previous treatments
Pigmentation visible
Histopathological diagnosis
0 1 2 3 4 10 None Clinical and dermoscopic pigment Dermoscopic pigment only Clinical and dermoscopic pigment outside field LM (treatment failure) Negative (no LM feature on biopsy) Showing: solar elastosis Some basal keratinocyte pigmentation Superficial dermal or epidermal chronic inflammation Mild junctional melanocytic atypia falling short of a diagnosis of LM Melanophages in the superficial dermis Bowen disease
2 1 18 5 4 1 08 (04–79)
10 9 6 4 1 1 4 23 2 2 15 16 16 7
3
4
6 1
LM, lentigo maligna.
The criteria for LM diagnosis described recently were also rarely found in patients with treatment failure.11 Darkening at dermoscopic examination had a sensitivity of 33% and specificity of 94%. An increased density of the vascular network had a low sensitivity of 20% and specificity of 69%. Red rhomboidal structures and target-like patterns were too rare to be of conclusive value. A new descriptor was identified when reviewing the dermoscopy features of all cases, blinded to the pathology final diagnosis but not to the aim of the study (Fig. 1). Thirteen British Journal of Dermatology (2014) 170, pp1305–1312
cases contained some very fine brown ‘dust’-like dots. This feature was subtle and isolated, being the only dermoscopic clue for diagnosis of treatment failure. The expert dermoscopist, blinded to any context of the research, was asked to review all dermoscopy photographs and validated this criterion (73% sensitivity and 87% specificity). This descriptor correlates with pagetoid cells as visualized by RCM with good agreement (n = 31, j = 0569, P < 0001). In seven patients (22%) pagetoid cells were observed, but no ‘dust’like brown dots were present. On RCM, the presence of a round large pagetoid cell feature was detected in 52% of cases that contained all biopsy-proven treatment failure (100% sensitivity). These cells were detected in only one patient, who was not confirmed by histology as having a recurrence (94% specificity). Nonedged papilla had a perfect specificity (100%) but low sensitivity (20%), as most cases contained nonvisible papillae (80% sensitivity) and very few cases edged papillae (13% sensitivity). Nucleated cells in the papillae had exactly the same diagnostic accuracy and frequency (9%) as nonedged papillae. More than three atypical junctional cells in five fields of view were found in 67% of patients with treatment failure but with a good specificity (100%). Follicular localization of atypical cells was also rarely present in patients with treatment failure, but was highly specific (81%). Broadened honeycomb is a negative minor diagnostic criterion of the LM score and was found only in one patient who was shown histologically to have Bowen disease (appearing in the field of a LM previously treated with RT). This case contained some enlarged vessels. Four other cases also showed this latter feature, all treated with RT. Dendritic pagetoid cells were present in most cases (61%) with a sensitivity of 87% and specificity of 62%. In four patients, using immunohistochemistry, we found that most of these cells were CD1a positive (a specific marker of Langerhans cells) and there was no clear increase in the number of melanocytes (as determined by MelanA and S100-protein positivity) (Fig. 2). Dendritic cells were also noted under RCM examination in two of the 19 patients treated with IQ and 10 of the 21 patients treated with RT who were not biopsied but only followed up with imagery. Some of these dendritic cells persisted for more than 42 months after treatment, but in a majority of cases they were present for only 6 months to 1 year. Melanophages were found in 71% of the cases, with a sensitivity of 80% and specificity of 38% for LM recurrence. They were also found in cases only followed up with imagery (in one of 19 patients treated with IQ and two of 21 patients treated with RT), but in most of these cases they were only transient. Highly retractile fibres were found in all cases except one, and all were reported to have some solar elastosis on pathology. Overall, the confocal LM Score was positive in 16 (52%) patients including all patients confirmed as having a recurrence by histology (100% sensitivity) and a specificity of 94%. © 2014 British Association of Dermatologists
Surveillance for treatment failure of lentigo maligna, P. Guitera et al. 1309
(a)
(c)
(b) (d)
Fig 1. (a) Clinical image of a lentigo maligna (LM) involving the right cheek of woman in her 80s. The LM recurred following surgery and was treated by radiotherapy. It subsequently recurred and was treated with imiquimod IQ. After an initial good clinical response, new pigmentation developed approximately 1 year after commencing IQ. The clinical image reflects the latest recurrence. (b) Dermoscopy image showing light brown pigmentation with multiple fine brown dots consistent with recurrent LM. (c) Confocal microscopy image (horizontal plane) of 05 9 05 mm area of the superficial epidermal layer showing pagetoid spread by numerous bright round large melanocytes. (d) Histological section (haematoxylin and eosin 9 100) (vertical plane) demonstrates flattened atrophic epidermis and solar elastosis with a lentiginous proliferation of atypical melanocytes involving the basal epidermis with multiple single cells showing pagetoid spread. The histological features are characteristic of LM.
Discussion LM is a diagnostic and management challenge, but detecting treatment failure is even more difficult because of frequent nonspecific pigmentation and less common amelanotic recurrence. Nevertheless, systematic or random biopsies are obviously not optimal. The purpose of this article is to describe diagnostic criteria to assist in identifying treatment failure of LM in patients treated nonsurgically. The two treatment efficacies are reported in Table 3. There was a bias in the treatment choice of recurrent LM patients reflecting a literature review that showed RT may have a better efficacy compared with IQ.5 As a consequence, most of the extensive LM were treated with RT, which may explain the worse outcome in our series. The previously reported RT studies, all from single institutions utilizing retrospective data, had an overall LM recurrence rate of 5% at 3 years.5 One major review in 2006, covering 67 patients treated with various modalities, showed an IQ efficacy of 88% for LM. The follow-up period was short, only exceeding 12 months in five patients.6 Reported response rates of IQ efficacy vary greatly between studies. Ly et al.7 reported response rates of no more than 53%. A randomized controlled trial with good quality control between institutions that can access RCM is the only way to resolve this question and © 2014 British Association of Dermatologists
is currently being developed for enrolment in multiple international centres.18 The fact that two patients developed rapidly growing LMM during follow-up emphasizes the importance of follow up. Fortunately, in these two cases, dermoscopy and RCM were positive and surgery was sufficient, with a smaller area to treat after nonsurgical treatment. The progression of LM to LMM is clearly established, but our hypothesis is that this invasive component may have been present but not diagnosed before medical treatment was commenced. In support of this hypothesis is the observation on histopathology of no epidermal involvement of the atypical cells in one of the two patients. The epidermal component may have been destroyed by the IQ treatment. In our opinion, the best option would be to map with RCM the area to treat and target a biopsy at any suspicious area before deciding on nonsurgical treatment. Dermoscopy has been the principal tool for melanoma diagnosis for two decades. However, the dermoscopic criteria for the detection of recurrent LM are questionable as the Schiffner criteria10 had a sensitivity of 80% and specificity of only 56%, even with expert interpretation. In this study, the best dermoscopic criterion was asymmetrical pigmented follicular openings, but it was present in only 47% of the patients with treatment failure. The new criteria proposed by Pralong et al.11 British Journal of Dermatology (2014) 170, pp1305–1312
1310 Surveillance for treatment failure of lentigo maligna, P. Guitera et al.
(a)
(b)
(c)
(d)
(e)
(f)
were also not very helpful as they rely heavily on vessels that have often become very distorted by previous surgery or RT. We found that very fine brown dots (‘dust’ appearance) correlated highly with the diagnosis of treatment failure (73% sensitivity and 88% specificity). These dots give the ‘granular’ appearance described by Schiffner et al.,10 but without an annular pattern. They were also distinctly brown, not grey, and correlated well with pagetoid cells migrating upward in the epidermis (as visualized by RCM) in contrast to grey dots, which correlate with melanophages in the dermis. The dermoscopy feature of ‘dust’ is morphologically a subset of ‘multiple brown dots’ previously reported to have a 30% sensitivity and 97% specificity for the diagnosis of invasive melanoma,19 and histopathological correlation with pagetoid invasion of melanoma.20 Interestingly, dendritic pagetoid cells were often present in our series. In four patients, we found that most of these cells were Langerhans cells, also reported by Hashemi et al.21 In the previously described large study of 81 LM and 203 control benign macules of the face,14 dendritic pagetoid infiltration was reported in 54% of LM and 16% of control (OR = 61). British Journal of Dermatology (2014) 170, pp1305–1312
Fig 2. Right cheek of a woman in her 70s affected by lentigo maligna (LM). The lesion was initially treated by surgery (skin graft), but subsequently recurred adjacent to the graft and was treated with radiotherapy. (a) Clinical image showing light pigmentation adjacent to the superomedial aspect of the skin graft and postradiotherapy telangiectasia superolaterally. (b) Dermoscopy image showing light-brown diffuse pigmentation and telangiectasia. (c) Confocal microscopy image (05 9 05 mm) of the epidermal layer showing widespread bright cells with prominent dendritic processes. (d) Histological section (haematoxylin and eosin 9 100) demonstrates flattened atrophic epidermis and solar elastosis without evidence of recurrent LM. (e) Melan A stain (9 400) revealing a mild basal melanocytic hyperplasia without significant cytological atypia. (f) Langerin stain (9 400) showing positive suprabasal dendritic cells corresponding to Langerhans cells.
Some experts consider that when dendritic cells are numerous (‘sheet-like’ feature), they could be a hint for the diagnosis of LM.22 Nevertheless, the LM score and the RCM score (which has been reported to be the best method for the diagnosis of melanoma excluding LM) excludes dendritic cells as a diagnostic criterion.23 In contrast, RCM can aid in differentiating melanocytes from melanophages.15 Melanophages are more of a clinical pitfall as they can create a pigmentation hue. Resolution of pigmentation in the reviewed studies occurred at 2– 24 months after RT or IQ treatment, with an average duration of 6 months. In practice, it is easier to perform dermoscopy and RCM diagnosis at least 6 months after treatment when most of the inflammation and melanophages have disappeared. Papillae features are often difficult to assess,24 and our results are not consistent with the previous large series of LM14 as we found more ‘nonvisible papillae’ and fewer ‘nonedged’ ones. There is some confusion between nonedged papillae (ill defined) and nonvisible ones, in particular on the sun damaged face where the dermal–epidermal junction is often flat with effacement of rete ridges. It is even more diffi© 2014 British Association of Dermatologists
Surveillance for treatment failure of lentigo maligna, P. Guitera et al. 1311 Table 3 The sensitivity and specificity of each diagnostic criteria compared with the gold standard (histopathological diagnosis) from N = 31 with 15 patients with confirmed treatment failure Feature present n (%) Schiffner dermoscopic criteria Hyperpigmented follicular opening Pigmented rhomboidal structures Annular-granular pattern Obliterated hair follicles Pralong dermoscopic criteria Darkening on dermoscopic examination Increase density of the vascular network Target-like patterns Red rhomboidal structures Other criteria Overall Schiffner diagnosis (expert dermoscopist blinded) Multiple very fine brown dots (‘dust’-like) Overall Schiffner diagnosis (blinded observer) Confocal features Round large (> 20 micron) pagetoid cells Enlarged collagen bundle (solar elastosis) Dendritic pagetoid cells Non-nucleated cells with ill-defined borders (melanophages) Non visible papilla Epidermal disarray > 3 atypical junctional cells Folicular localisation of atypical cells Non edged papilla Nucleated cells in papilla Edged papilla Honeycomb broadened Enlarged vessels Overall confocal LM score > 2
Sensitivity (%)
Specificity (%)
Odds ratio (95% CI)
10 4 9 0
(323) (129) (290) (00)
467 200 200 –
813 938 625
38 (08–190) 37 (03–408) 04 (01–21)
6 8 1 1
(194) (258) (32) (32)
333 200 67 00
938 688 1000 938
75 (08–741) 05 (01–29) – –
19 (613)
800
563
51 (10–256)
13 (419) 18 (581)
733 667
875 500
192 (30–1251) 20 (05–86) – – 108 (18–655) 24 (05–121)
16 30 19 22
(516) (968) (613) (710)
1000 933 867 800
938 00 625 375
26 14 10 7 3 3 3 1 5 16
(839) (452) (323) (226) (97) (97) (97) (32) (161) (516)
800 667 667 267 200 200 133 00 00 1000
125 750 1000 813 1000 1000 938 938 688 938
06 60 – 16 – – 23 – – –
(01–40) (13–285) (03–86)
(02–285)
CI, confidence interval; LM, lentigo maligna.
cult when the junction is distorted by atypical cells (in the case of LM) and/or scarring induced by treatment. Nevertheless, in our study, nonedged papillae and nucleated cells in the dermis had 100% specificity for LM diagnosis and, when readily recognized, can help differentiate actual treatment failure. It is surprising that nucleated cells in the dermal papillae are important in the diagnosis of LM (which is by definition confined to the epidermis). The reason for the apparent discrepancy between RCM and pathology findings may be related to the difficulty in determining the exact site of the cells (junctional or dermal), but also may be due to sampling error such that the dermal component identified on RCM was not present in the pathology biopsy sections. The major limitation of these data is that they were generated from a review of a case series and should be confirmed by larger prospective studies. The areas that were normal under dermoscopy and RCM examination were not biopsied and may have had histological evidence of LM. In practice, ‘dust’ affords a 73% sensitivity and 88% specificity for the diagnosis of treatment failure. A biopsy should be undertaken if these brown dots are identified or in case of persistence of the asymmetrical follicular opening. If multiple © 2014 British Association of Dermatologists
grey dots are visible, RCM can help distinguish real treatment failure and regression features due to an efficient treatment. The LM score had a specificity of 94% and sensitivity of 100%, and can be confidently used for this diagnosis. The advantages of using RCM are that it can avoid biopsy in the vast majority of cases, as found by others,25 and it allows expeditious management obviating the need to wait until the lesion is clinically obvious.
Acknowledgments We would like to thank our colleagues of the Melanoma Institute Australia, and the Australia and New Zealand Melanoma Trials Group staff for their help, and Dr Tony Bonin for his editing.
References 1 Osborne JE, Hutchinson PE. A follow-up study to investigate the efficacy of initial treatment of lentigo maligna with surgical excision. Br J Plast Surg 2002; 55:611–15. 2 Kunishige JH, Brodland DG, Zitelli JA. Surgical margins for melanoma in situ. J Am Acad Dermatol 2012; 66:438–44.
British Journal of Dermatology (2014) 170, pp1305–1312
1312 Surveillance for treatment failure of lentigo maligna, P. Guitera et al. 3 Australian Cancer Network Melanoma Guidelines Revision Working Party. Clinical Practice Guidelines for the Management of Melanoma in Australia and New Zealand. Wellington: Cancer Council Australia and Australian Cancer Network, Sydney and New Zealand Guidelines Group, 2008. 4 De Berker D. Lentigo maligna and Mohs. Arch Dermatol 1991; 127:421. 5 Fogarty GB, Hong A, Scolyer RA et al. Radiotherapy for lentigo maligna: a systematic literature review and recommendations for treatment. Br J Dermatol 2014; 170:52–8. 6 Rajpar SF, Marsden JR. Imiquimod in the treatment of lentigo maligna. Br J Dermatol 2006; 155:653–6. 7 Ly L, Kelly JW, O’Keefe R et al. Efficacy of imiquimod cream, 5%, for lentigo maligna after complete excision: a study of 43 patients. Arch Dermatol 2011; 147:1191–5. 8 Cotter MA, McKenna JK, Bowen GM. Treatment of lentigo maligna with imiquimod before staged excision. Dermatol Surg 2008; 34:147–51. 9 Fisher GH, Lang PG. Treatment of melanoma in situ on sundamaged skin with topical 5% imiquimod cream complicated by the development of invasive disease. Arch Dermatol 2003; 139:945–7. 10 Schiffner R, Schiffner-Rohe J, Vogt T et al. Improvement of early recognition of lentigo maligna using dermatoscopy. J Am Acad Dermatol 2000; 42:25–32. 11 Pralong P, Bathelier E, Dalle S et al. Dermoscopy of lentigo maligna melanoma: report of 125 cases. Br J Dermatol 2012; 167:280–7. 12 Rajadhyaksha M, Grossman M, Esterowitz D et al. In vivo confocal scanning laser microscopy of human skin: melanin provides strong contrast. J Invest Dermatol 1995; 104:946–52. 13 Guitera P, Pellacani G, Longo C et al. In vivo reflectance confocal microscopy enhances secondary evaluation of melanocytic lesions. J Invest Dermatol 2009; 129:131–8. 14 Guitera P, Pellacani G, Crotty KA et al. The impact of in vivo reflectance confocal microscopy on the diagnostic accuracy of lentigo maligna and equivocal pigmented and nonpigmented macules of the face. J Invest Dermatol 2010; 130:2080–1.
British Journal of Dermatology (2014) 170, pp1305–1312
15 Guitera P, Li LX, Scolyer RA, Menzies SW. Morphological features of melanophages under in vivo reflectance confocal microscopy. Arch Dermatol 2010; 146:492–8. 16 Rajadhyaksha M, Gonzalez S, Zavislan JM et al. In vivo confocal scanning laser microscopy of human skin II: advances in instrumentation and comparison with histology. J Invest Dermatol 1999; 113:293–303. 17 Longo C, Casari A, Beretti F et al. Skin aging: in vivo microscopic assessment of epidermal and dermal changes by means of confocal microscopy. J Am Acad Dermatol 2013; 68:e73–82. 18 ANZMTG 02 12 RADICAL study. Available at: http://www.anzmtg.org/trialdetails.aspx?trialno=16 (last accessed 28 February 2014). 19 Menzies SW, Ingvar C, McCarthy WA. A sensitivity and specificity analysis of the surface microscopy features of invasive melanoma. Melanoma Res 1996; 6:55–62. 20 Menzies SW, Moloney FJ, Byth K et al. Dermoscopic evaluation of nodular melanoma. JAMA Dermatol 2013; 149:699–709. 21 Hashemi P, Pulitzer MP, Scope A et al. Langerhans cells and melanocytes share similar morphologic features under in vivo reflectance confocal microscopy: a challenge for melanoma diagnosis. J Am Acad Dermatol 2012; 66:452–62. 22 Ahlgrimm-Siess V, Massone C, Scope A et al. Reflectance confocal microscopy of facial lentigo maligna and lentigo maligna melanoma: a preliminary study. Br J Dermatol 2009; 161:1307–16. 23 Pellacani G, Guitera P, Longo C et al. The impact of in vivo reflectance confocal microscopy for the diagnostic accuracy of melanoma and equivocal melanocytic lesions. J Invest Dermatol 2007; 127:2759–65. 24 Pellacani G, Vinceti M, Bassoli S et al. Reflectance confocal microscopy features of melanocytic lesions: an internet-based study of the reproducibility of terminology. Arch Dermatol 2009; 145:1137–43. 25 Alarcon I, Carrera C, Palou J et al. Impact of in vivo reflectance confocal microscopy on the number needed to treat melanoma in doubtful lesions. Br J Dermatol 2014; 170:802–8.
© 2014 British Association of Dermatologists
BJD
British Journal of Dermatology
Surveillance for treatment failure of lentigo maligna with dermoscopy and in vivo confocal microscopy: new descriptors P. Guitera,1,2,3 L.E. Haydu,1,4 S.W. Menzies,2,3 R.A. Scolyer,1,5,6 A. Hong,1,7 G.B. Fogarty,1,6 F. Gallardo8 and S. Segura8 1
Melanoma Institute Australia, Poche Centre, 40 Rocklands Road, North Sydney, NSW 2060, Australia Department of Sydney Melanoma Diagnostic Centre, 5Department of Tissue Pathology and Diagnostic Oncology and 7Department of Radiation Oncology, Royal Prince Alfred Hospital, Missenden Road Camperdown, NSW 2050, Australia 3 Discipline of Dermatology, 4Discipline of Surgery and 6Discipline of Pathology, The University of Sydney, Sydney, NSW, Australia 8 Department of Dermatology, Hospital el Mar, Barcelona, Spain 2
Summary Correspondence Pascale Guitera. E-mail: [email protected]
Accepted for publication 10 February 2014
Funding sources This study was supported by the Melanoma Institute Australia, the Melanoma Foundation of the University of Sydney, and the Australia and New Zealand Melanoma Trials Group.
Conflicts of interest None. DOI 10.1111/bjd.12839
Background Nonsurgical treatment (radiotherapy, imiquimod) is increasingly employed for the management of lentigo maligna (LM). While the diagnosis of LM remains difficult, the detection of treatment failure is even more challenging. Objectives To describe the sensitivity and specificity for the diagnosis of LM of individual features and methods using dermoscopy and in vivo reflectance confocal microscopy (RCM) to aid in the detection of treatment failure of LM following nonsurgical treatment. Methods A retrospective study of dermoscopy and RCM images (blinded to the correlation with pathology) in patients with biopsy-confirmed LM who were undergoing nonsurgical treatment in two referral institutions – one in Sydney, Australia, and the other in Barcelona, Spain. Ninety-eight patients were treated nonsurgically for LM during the period 2006–2012. Thirty-one patients had abnormal dermoscopy or RCM evaluation, and had a biopsy that identified LM recurrence in 15 patients and nonmelanoma diagnoses in 16 patients (one Bowen disease, 15 solar changes). Results The diagnosis of treatment failure was difficult with dermoscopy, with a sensitivity of 80% and specificity of 56%, even with the interpretation of an expert. The best criterion was asymmetric hyperpigmented follicular openings, but this was present in only 47% of treatment failure LM. Isolated, very fine brown dots (‘dust’ appearance) correlated highly with the diagnosis of treatment failure LM (73% sensitivity and 88% specificity) and with pagetoid cells seen with RCM. The LM score, comprising six criteria, had a specificity of 94% and sensitivity of 100%. Conclusions These methods and descriptors should help to manage the diagnosis of treatment failure.
What’s already known about this topic?
•
© 2014 British Association of Dermatologists
Clinical diagnosis of recurrent lentigo maligna (LM) following treatment is challenging because treatment-induced inflammation and pigmentation are common, but so are amelanotic recurrences.
British Journal of Dermatology (2014) 170, pp1305–1312
1305
1306 Surveillance for treatment failure of lentigo maligna, P. Guitera et al.
What does this study add?
•
We tested the sensitivity and specificity of individual features and methods using dermoscopy and in vivo reflectance confocal microscopy for the diagnosis of LM and give recommendations for management.
Lentigo maligna (LM) is a form of melanoma in situ that occurs on exposed, sun-damaged skin of elderly people. LM represents a challenge for management; LM has a comparatively higher local recurrence rate,1,2 most likely because its margins are difficult to define clinically and histologically. Optimal treatment is still debated, and there are no prospective studies or randomized controlled trials.3 Staged excision is recognized as the best treatment option as it facilitates pathological assessment of the entire lesion to rule out invasive melanoma. However, surgery may be contraindicated owing to the extent of disease, comorbidities or patient preference. Significant morbidity can result from wide surgical excisions.4 Moreover, medical treatments can treat a larger field, in the hope of decreasing treatment failure rates and reducing cosmetic issues. Radiation therapy (RT) is currently the ‘standard’ nonsurgical alternative, but there is only weak evidence to support its use.5 The cosmetic outcome is better with the use of imiquimod (IQ), but only case series have been published, and the treatment failure with IQ appears to be higher than with RT.6–9 Clinical and dermoscopy diagnosis of LM is often difficult because of the heterogeneity and contiguous lesions on sun damaged elderly skin.10,11 These diagnoses are even more difficult when trying to detect treatment failure as treatmentinduced inflammation can lead to pigmentation; pigment incontinence is a well-known pitfall.8 Alternatively, amelanotic LM recurrences have also led to false negative diagnoses.9 In vivo reflectance confocal microscopy (RCM) provides cellular resolution in the upper layers of the skin. Because melanosomes appear bright under reflectance at near-infrared wavelengths, pigmented cells are easily visualized.12 Furthermore, as RCM generates a horizontal view of at least 8 9 8 mm, it is possible to assess more of the lesion using this technique than with the pathological assessment of vertically orientated small biopsy specimens examined in routine histopathology.13 We have recently described an algorithm (LM score) to distinguish LM from benign macules,14 which resulted in a sensitivity of 85% and specificity of 76% for the noninvasive diagnosis of LM [odds ratio (OR) = 186]. The algorithm was equally effective for the diagnosis of amelanotic lesions. Moreover, melanophages, which represent one of the main pitfalls of treatment failure evaluation, can be distinguished from melanocytes with RCM.15 The aim of this study was to describe how dermoscopy and RCM can help in the detection of recurrent LM previously treated with IQ and/or RT. British Journal of Dermatology (2014) 170, pp1305–1312
Patients and methods Patients were treated with IQ or RT for a histologically confirmed LM at two referral centres (Barcelona, Spain, and Sydney, Australia). Nonsurgical treatments were chosen because surgical excision was not possible, refused or failed. IQ was applied five times a week for 3 months with a margin of at least 1 cm beyond the clinically identified margins of the LM. Inflammation induction due to IQ was utilized to individualize dose titration for each patient. Some patients required a dose escalation, others had a painful reaction and their dose was decreased, but all patients completed 12 weeks of treatment. In terms of RT, the surface doses were generally around 50 Gy, but never exceeded 66 Gy. A minimum of three, but generally four or five fractions, were administered per week for a total of 4–6 weeks. Histological confirmation was sought if dermoscopy examination showed some pigmentation and/or RCM imaging showed some of the LM score criteria. Dermoscopy was performed on the entire field of the previous LM diagnosis of the previously diagnosed LM and surrounding sites if they appeared suspicious. Confocal microscopy was centred on the site of the previous LM diagnostic biopsy and on pigmentation sites detected by dermoscopy. In patients with no evidence of LM recurrence on clinical, dermoscopy and RCM examinations, a follow-up of at least yearly review was performed. This was a retrospective study approved by the local Human Ethics Review Committee (Sydney South West Area, protocol number X05-0218). Dermoscopic diagnosis Vivacam (CaliberID, Rochester, NY, U.S.A.) and Dermlite Foto (3 Gen, San Juan Capistrano, CA, U.S.A.) images were reviewed retrospectively by two observers blinded to the biopsy results. The criteria reported by Schiffner et al.10 and Pralong et al.11 were employed in the evaluation. Any other dermoscopy features found were also recorded. An expert dermoscopist, blinded to the research topic, reviewed all dermoscopy photographs, provided a diagnosis and was asked to confirm the presence or absence of new features. Confocal diagnosis All cases were recorded prior to biopsy using RCM (Vivascope 1500; CaliberID, Henrietta, NY, U.S.A.). Instrument © 2014 British Association of Dermatologists
Surveillance for treatment failure of lentigo maligna, P. Guitera et al. 1307
and acquisition procedures are described elsewhere.16 Each image corresponds to a horizontal section of 05 9 05 mm2 and a sequence of montage images (‘mosaic’ images) was acquired at the level of the dermoepidermal junction to explore an 8 9 8 mm2 field of view. Confocal sections, beginning at the stratum corneum and ending inside the papillary dermis, were recorded in the middle of the field and at all areas of interest. All recorded images were retrospectively reviewed by two observers blinded to the final biopsy results in each centre to assess the LM score criteria.14 Additional features such as epidermal disarray, dendritic pagetoid cells, edged papilla, nonvisible papilla, melanophages,15 highly reflecting fibres (described as solar elastosis feature under RCM)17 and enlarged vessels were also recorded. Statistical methods Sensitivity and specificity for the detection of LM were calculated for each dermoscopic and RCM feature using histopathological diagnosis as the gold standard. The ORs were calculated using cross-tabulations for a 2 9 2 table. The kappa statistic (j) measured the interobserver reproducibility rating (j > 075 excellent; 04 ≤ j ≤ 075 good; and 0 ≤ j < 04 marginal). All analyses were conducted with the SPSS Statistic v21 software package (IBM, Armonk, NY, U.S.A.).
Results Ninety-eight patients were treated with IQ or RT for a LM during the period 2006–2012. Twenty-seven had inadequate follow-up (< 6 months) or insufficient imagery to be assessed (patients were followed up locally where RCM and/or dermoscopy were not available). Forty patients had normal skin under dermoscopy and RCM and therefore did not have a biopsy. Thirty-one patients were biopsied because they had abnormal dermoscopy and/or RCM evaluation. Treatment failure was diagnosed by pathology in 15 of the 31 patients who had a biopsy. However, the treatment failure ratio diagnosed by each modality was calculated also considering cases with
normal imagery and an average follow-up of 12 months (range = 6–42 months) as negative. We obtained 11% and 28% treatment failure for IQ and RT, respectively. Of note, only one in 15 patients had a treatment failure outside the treatment field, and 10 had a significantly reduced area to treat (indicating that the surface of the LM was smaller after IQ or RT treatment). Eleven patients had their LM recurrence treated by surgery, while four patients were treated with IQ. Table 1 describes the characteristics of the 31 patients in whom treatment failure was suspected by dermoscopy and/or RCM and correlation between imagery and pathology was possible. Fifteen cases had a confirmed treatment failure, whereas 16 patients had no diagnostic evidence of LM on pathologic assessment. Among the 15 patients who had a confirmed treatment failure, two had lentigo maligna melanoma (LMM) with a Breslow thickness of 04 and 06 mm, respectively. These two patients had obvious recurring small areas of pigmentation in the field of treatment after a few months of IQ and RT, respectively, and were then treated with surgery. Four patients had no pigment at all on dermoscopy and two patients had pigment visible only with dermoscopy. Twenty-three patients had obvious pigmentation in the field of treatment and in two patients outside the field of treatment. All except one treatment failures confirmed by pathology were identified in areas with visible pigmentation, meaning that one of 15 patients had an amelanotic recurrence. The sensitivity and specificity of each diagnostic criterion, compared with the gold standard (histopathological diagnosis), are summarized in Table 2. The asymmetric pigment around follicular openings was the most prevalent criterion (32%) with a low sensitivity (47%) but high specificity (81%). An annular-granular pattern was also present in numerous cases (29%) with an even lower sensitivity (20%) and specificity (62%). The pigmented rhomboidal structures had a high specificity (94%) but low sensitivity (20%), and obliteration of hair follicles was never found. There was good interobserver agreement between the two observers for the overall Schiffner criteria compared with the LM diagnosis of the expert dermoscopist (n = 31, j = 0534, P = 0003).10
Table 1 Patient treatment and follow up
Number of patients treated (n = 98) Number of patients in whom imagerya follow-up was not available (n = 27) Number of patients in whom imagery was normal with average follow-up > 12 months (range 6–42) Number of patients in whom imagery was suspicious for treatment failure and biopsy correlation (see Table 1) (n = 31) Number of patients with treatment failure LM (% of treatment failure by treatment category) (n = 15)
IQ
RT
39 11
59 16
19
21
9
22
3 (3/19 + 9) = 11%
12 (12/21 + 22) = 28%
LM, lentigo maligna; IQ, imiquimod; RT, radiotherapy. aImagery refers to dermoscopy and reflectance confocal microscopy.
© 2014 British Association of Dermatologists
British Journal of Dermatology (2014) 170, pp1305–1312
1308 Surveillance for treatment failure of lentigo maligna, P. Guitera et al. Table 2 Characteristics of the cohort when imagery and pathology correlation were performed (N = 31) Feature
Value
n
Sex
Female Male Median (range) years Scalp Forehead Cheek Eyelid Nose Temple Median (range) years
23 8 73 (35–89)
Age Site
Time between end of treatment and biopsy Number of previous treatments
Pigmentation visible
Histopathological diagnosis
0 1 2 3 4 10 None Clinical and dermoscopic pigment Dermoscopic pigment only Clinical and dermoscopic pigment outside field LM (treatment failure) Negative (no LM feature on biopsy) Showing: solar elastosis Some basal keratinocyte pigmentation Superficial dermal or epidermal chronic inflammation Mild junctional melanocytic atypia falling short of a diagnosis of LM Melanophages in the superficial dermis Bowen disease
2 1 18 5 4 1 08 (04–79)
10 9 6 4 1 1 4 23 2 2 15 16 16 7
3
4
6 1
LM, lentigo maligna.
The criteria for LM diagnosis described recently were also rarely found in patients with treatment failure.11 Darkening at dermoscopic examination had a sensitivity of 33% and specificity of 94%. An increased density of the vascular network had a low sensitivity of 20% and specificity of 69%. Red rhomboidal structures and target-like patterns were too rare to be of conclusive value. A new descriptor was identified when reviewing the dermoscopy features of all cases, blinded to the pathology final diagnosis but not to the aim of the study (Fig. 1). Thirteen British Journal of Dermatology (2014) 170, pp1305–1312
cases contained some very fine brown ‘dust’-like dots. This feature was subtle and isolated, being the only dermoscopic clue for diagnosis of treatment failure. The expert dermoscopist, blinded to any context of the research, was asked to review all dermoscopy photographs and validated this criterion (73% sensitivity and 87% specificity). This descriptor correlates with pagetoid cells as visualized by RCM with good agreement (n = 31, j = 0569, P < 0001). In seven patients (22%) pagetoid cells were observed, but no ‘dust’like brown dots were present. On RCM, the presence of a round large pagetoid cell feature was detected in 52% of cases that contained all biopsy-proven treatment failure (100% sensitivity). These cells were detected in only one patient, who was not confirmed by histology as having a recurrence (94% specificity). Nonedged papilla had a perfect specificity (100%) but low sensitivity (20%), as most cases contained nonvisible papillae (80% sensitivity) and very few cases edged papillae (13% sensitivity). Nucleated cells in the papillae had exactly the same diagnostic accuracy and frequency (9%) as nonedged papillae. More than three atypical junctional cells in five fields of view were found in 67% of patients with treatment failure but with a good specificity (100%). Follicular localization of atypical cells was also rarely present in patients with treatment failure, but was highly specific (81%). Broadened honeycomb is a negative minor diagnostic criterion of the LM score and was found only in one patient who was shown histologically to have Bowen disease (appearing in the field of a LM previously treated with RT). This case contained some enlarged vessels. Four other cases also showed this latter feature, all treated with RT. Dendritic pagetoid cells were present in most cases (61%) with a sensitivity of 87% and specificity of 62%. In four patients, using immunohistochemistry, we found that most of these cells were CD1a positive (a specific marker of Langerhans cells) and there was no clear increase in the number of melanocytes (as determined by MelanA and S100-protein positivity) (Fig. 2). Dendritic cells were also noted under RCM examination in two of the 19 patients treated with IQ and 10 of the 21 patients treated with RT who were not biopsied but only followed up with imagery. Some of these dendritic cells persisted for more than 42 months after treatment, but in a majority of cases they were present for only 6 months to 1 year. Melanophages were found in 71% of the cases, with a sensitivity of 80% and specificity of 38% for LM recurrence. They were also found in cases only followed up with imagery (in one of 19 patients treated with IQ and two of 21 patients treated with RT), but in most of these cases they were only transient. Highly retractile fibres were found in all cases except one, and all were reported to have some solar elastosis on pathology. Overall, the confocal LM Score was positive in 16 (52%) patients including all patients confirmed as having a recurrence by histology (100% sensitivity) and a specificity of 94%. © 2014 British Association of Dermatologists
Surveillance for treatment failure of lentigo maligna, P. Guitera et al. 1309
(a)
(c)
(b) (d)
Fig 1. (a) Clinical image of a lentigo maligna (LM) involving the right cheek of woman in her 80s. The LM recurred following surgery and was treated by radiotherapy. It subsequently recurred and was treated with imiquimod IQ. After an initial good clinical response, new pigmentation developed approximately 1 year after commencing IQ. The clinical image reflects the latest recurrence. (b) Dermoscopy image showing light brown pigmentation with multiple fine brown dots consistent with recurrent LM. (c) Confocal microscopy image (horizontal plane) of 05 9 05 mm area of the superficial epidermal layer showing pagetoid spread by numerous bright round large melanocytes. (d) Histological section (haematoxylin and eosin 9 100) (vertical plane) demonstrates flattened atrophic epidermis and solar elastosis with a lentiginous proliferation of atypical melanocytes involving the basal epidermis with multiple single cells showing pagetoid spread. The histological features are characteristic of LM.
Discussion LM is a diagnostic and management challenge, but detecting treatment failure is even more difficult because of frequent nonspecific pigmentation and less common amelanotic recurrence. Nevertheless, systematic or random biopsies are obviously not optimal. The purpose of this article is to describe diagnostic criteria to assist in identifying treatment failure of LM in patients treated nonsurgically. The two treatment efficacies are reported in Table 3. There was a bias in the treatment choice of recurrent LM patients reflecting a literature review that showed RT may have a better efficacy compared with IQ.5 As a consequence, most of the extensive LM were treated with RT, which may explain the worse outcome in our series. The previously reported RT studies, all from single institutions utilizing retrospective data, had an overall LM recurrence rate of 5% at 3 years.5 One major review in 2006, covering 67 patients treated with various modalities, showed an IQ efficacy of 88% for LM. The follow-up period was short, only exceeding 12 months in five patients.6 Reported response rates of IQ efficacy vary greatly between studies. Ly et al.7 reported response rates of no more than 53%. A randomized controlled trial with good quality control between institutions that can access RCM is the only way to resolve this question and © 2014 British Association of Dermatologists
is currently being developed for enrolment in multiple international centres.18 The fact that two patients developed rapidly growing LMM during follow-up emphasizes the importance of follow up. Fortunately, in these two cases, dermoscopy and RCM were positive and surgery was sufficient, with a smaller area to treat after nonsurgical treatment. The progression of LM to LMM is clearly established, but our hypothesis is that this invasive component may have been present but not diagnosed before medical treatment was commenced. In support of this hypothesis is the observation on histopathology of no epidermal involvement of the atypical cells in one of the two patients. The epidermal component may have been destroyed by the IQ treatment. In our opinion, the best option would be to map with RCM the area to treat and target a biopsy at any suspicious area before deciding on nonsurgical treatment. Dermoscopy has been the principal tool for melanoma diagnosis for two decades. However, the dermoscopic criteria for the detection of recurrent LM are questionable as the Schiffner criteria10 had a sensitivity of 80% and specificity of only 56%, even with expert interpretation. In this study, the best dermoscopic criterion was asymmetrical pigmented follicular openings, but it was present in only 47% of the patients with treatment failure. The new criteria proposed by Pralong et al.11 British Journal of Dermatology (2014) 170, pp1305–1312
1310 Surveillance for treatment failure of lentigo maligna, P. Guitera et al.
(a)
(b)
(c)
(d)
(e)
(f)
were also not very helpful as they rely heavily on vessels that have often become very distorted by previous surgery or RT. We found that very fine brown dots (‘dust’ appearance) correlated highly with the diagnosis of treatment failure (73% sensitivity and 88% specificity). These dots give the ‘granular’ appearance described by Schiffner et al.,10 but without an annular pattern. They were also distinctly brown, not grey, and correlated well with pagetoid cells migrating upward in the epidermis (as visualized by RCM) in contrast to grey dots, which correlate with melanophages in the dermis. The dermoscopy feature of ‘dust’ is morphologically a subset of ‘multiple brown dots’ previously reported to have a 30% sensitivity and 97% specificity for the diagnosis of invasive melanoma,19 and histopathological correlation with pagetoid invasion of melanoma.20 Interestingly, dendritic pagetoid cells were often present in our series. In four patients, we found that most of these cells were Langerhans cells, also reported by Hashemi et al.21 In the previously described large study of 81 LM and 203 control benign macules of the face,14 dendritic pagetoid infiltration was reported in 54% of LM and 16% of control (OR = 61). British Journal of Dermatology (2014) 170, pp1305–1312
Fig 2. Right cheek of a woman in her 70s affected by lentigo maligna (LM). The lesion was initially treated by surgery (skin graft), but subsequently recurred adjacent to the graft and was treated with radiotherapy. (a) Clinical image showing light pigmentation adjacent to the superomedial aspect of the skin graft and postradiotherapy telangiectasia superolaterally. (b) Dermoscopy image showing light-brown diffuse pigmentation and telangiectasia. (c) Confocal microscopy image (05 9 05 mm) of the epidermal layer showing widespread bright cells with prominent dendritic processes. (d) Histological section (haematoxylin and eosin 9 100) demonstrates flattened atrophic epidermis and solar elastosis without evidence of recurrent LM. (e) Melan A stain (9 400) revealing a mild basal melanocytic hyperplasia without significant cytological atypia. (f) Langerin stain (9 400) showing positive suprabasal dendritic cells corresponding to Langerhans cells.
Some experts consider that when dendritic cells are numerous (‘sheet-like’ feature), they could be a hint for the diagnosis of LM.22 Nevertheless, the LM score and the RCM score (which has been reported to be the best method for the diagnosis of melanoma excluding LM) excludes dendritic cells as a diagnostic criterion.23 In contrast, RCM can aid in differentiating melanocytes from melanophages.15 Melanophages are more of a clinical pitfall as they can create a pigmentation hue. Resolution of pigmentation in the reviewed studies occurred at 2– 24 months after RT or IQ treatment, with an average duration of 6 months. In practice, it is easier to perform dermoscopy and RCM diagnosis at least 6 months after treatment when most of the inflammation and melanophages have disappeared. Papillae features are often difficult to assess,24 and our results are not consistent with the previous large series of LM14 as we found more ‘nonvisible papillae’ and fewer ‘nonedged’ ones. There is some confusion between nonedged papillae (ill defined) and nonvisible ones, in particular on the sun damaged face where the dermal–epidermal junction is often flat with effacement of rete ridges. It is even more diffi© 2014 British Association of Dermatologists
Surveillance for treatment failure of lentigo maligna, P. Guitera et al. 1311 Table 3 The sensitivity and specificity of each diagnostic criteria compared with the gold standard (histopathological diagnosis) from N = 31 with 15 patients with confirmed treatment failure Feature present n (%) Schiffner dermoscopic criteria Hyperpigmented follicular opening Pigmented rhomboidal structures Annular-granular pattern Obliterated hair follicles Pralong dermoscopic criteria Darkening on dermoscopic examination Increase density of the vascular network Target-like patterns Red rhomboidal structures Other criteria Overall Schiffner diagnosis (expert dermoscopist blinded) Multiple very fine brown dots (‘dust’-like) Overall Schiffner diagnosis (blinded observer) Confocal features Round large (> 20 micron) pagetoid cells Enlarged collagen bundle (solar elastosis) Dendritic pagetoid cells Non-nucleated cells with ill-defined borders (melanophages) Non visible papilla Epidermal disarray > 3 atypical junctional cells Folicular localisation of atypical cells Non edged papilla Nucleated cells in papilla Edged papilla Honeycomb broadened Enlarged vessels Overall confocal LM score > 2
Sensitivity (%)
Specificity (%)
Odds ratio (95% CI)
10 4 9 0
(323) (129) (290) (00)
467 200 200 –
813 938 625
38 (08–190) 37 (03–408) 04 (01–21)
6 8 1 1
(194) (258) (32) (32)
333 200 67 00
938 688 1000 938
75 (08–741) 05 (01–29) – –
19 (613)
800
563
51 (10–256)
13 (419) 18 (581)
733 667
875 500
192 (30–1251) 20 (05–86) – – 108 (18–655) 24 (05–121)
16 30 19 22
(516) (968) (613) (710)
1000 933 867 800
938 00 625 375
26 14 10 7 3 3 3 1 5 16
(839) (452) (323) (226) (97) (97) (97) (32) (161) (516)
800 667 667 267 200 200 133 00 00 1000
125 750 1000 813 1000 1000 938 938 688 938
06 60 – 16 – – 23 – – –
(01–40) (13–285) (03–86)
(02–285)
CI, confidence interval; LM, lentigo maligna.
cult when the junction is distorted by atypical cells (in the case of LM) and/or scarring induced by treatment. Nevertheless, in our study, nonedged papillae and nucleated cells in the dermis had 100% specificity for LM diagnosis and, when readily recognized, can help differentiate actual treatment failure. It is surprising that nucleated cells in the dermal papillae are important in the diagnosis of LM (which is by definition confined to the epidermis). The reason for the apparent discrepancy between RCM and pathology findings may be related to the difficulty in determining the exact site of the cells (junctional or dermal), but also may be due to sampling error such that the dermal component identified on RCM was not present in the pathology biopsy sections. The major limitation of these data is that they were generated from a review of a case series and should be confirmed by larger prospective studies. The areas that were normal under dermoscopy and RCM examination were not biopsied and may have had histological evidence of LM. In practice, ‘dust’ affords a 73% sensitivity and 88% specificity for the diagnosis of treatment failure. A biopsy should be undertaken if these brown dots are identified or in case of persistence of the asymmetrical follicular opening. If multiple © 2014 British Association of Dermatologists
grey dots are visible, RCM can help distinguish real treatment failure and regression features due to an efficient treatment. The LM score had a specificity of 94% and sensitivity of 100%, and can be confidently used for this diagnosis. The advantages of using RCM are that it can avoid biopsy in the vast majority of cases, as found by others,25 and it allows expeditious management obviating the need to wait until the lesion is clinically obvious.
Acknowledgments We would like to thank our colleagues of the Melanoma Institute Australia, and the Australia and New Zealand Melanoma Trials Group staff for their help, and Dr Tony Bonin for his editing.
References 1 Osborne JE, Hutchinson PE. A follow-up study to investigate the efficacy of initial treatment of lentigo maligna with surgical excision. Br J Plast Surg 2002; 55:611–15. 2 Kunishige JH, Brodland DG, Zitelli JA. Surgical margins for melanoma in situ. J Am Acad Dermatol 2012; 66:438–44.
British Journal of Dermatology (2014) 170, pp1305–1312
1312 Surveillance for treatment failure of lentigo maligna, P. Guitera et al. 3 Australian Cancer Network Melanoma Guidelines Revision Working Party. Clinical Practice Guidelines for the Management of Melanoma in Australia and New Zealand. Wellington: Cancer Council Australia and Australian Cancer Network, Sydney and New Zealand Guidelines Group, 2008. 4 De Berker D. Lentigo maligna and Mohs. Arch Dermatol 1991; 127:421. 5 Fogarty GB, Hong A, Scolyer RA et al. Radiotherapy for lentigo maligna: a systematic literature review and recommendations for treatment. Br J Dermatol 2014; 170:52–8. 6 Rajpar SF, Marsden JR. Imiquimod in the treatment of lentigo maligna. Br J Dermatol 2006; 155:653–6. 7 Ly L, Kelly JW, O’Keefe R et al. Efficacy of imiquimod cream, 5%, for lentigo maligna after complete excision: a study of 43 patients. Arch Dermatol 2011; 147:1191–5. 8 Cotter MA, McKenna JK, Bowen GM. Treatment of lentigo maligna with imiquimod before staged excision. Dermatol Surg 2008; 34:147–51. 9 Fisher GH, Lang PG. Treatment of melanoma in situ on sundamaged skin with topical 5% imiquimod cream complicated by the development of invasive disease. Arch Dermatol 2003; 139:945–7. 10 Schiffner R, Schiffner-Rohe J, Vogt T et al. Improvement of early recognition of lentigo maligna using dermatoscopy. J Am Acad Dermatol 2000; 42:25–32. 11 Pralong P, Bathelier E, Dalle S et al. Dermoscopy of lentigo maligna melanoma: report of 125 cases. Br J Dermatol 2012; 167:280–7. 12 Rajadhyaksha M, Grossman M, Esterowitz D et al. In vivo confocal scanning laser microscopy of human skin: melanin provides strong contrast. J Invest Dermatol 1995; 104:946–52. 13 Guitera P, Pellacani G, Longo C et al. In vivo reflectance confocal microscopy enhances secondary evaluation of melanocytic lesions. J Invest Dermatol 2009; 129:131–8. 14 Guitera P, Pellacani G, Crotty KA et al. The impact of in vivo reflectance confocal microscopy on the diagnostic accuracy of lentigo maligna and equivocal pigmented and nonpigmented macules of the face. J Invest Dermatol 2010; 130:2080–1.
British Journal of Dermatology (2014) 170, pp1305–1312
15 Guitera P, Li LX, Scolyer RA, Menzies SW. Morphological features of melanophages under in vivo reflectance confocal microscopy. Arch Dermatol 2010; 146:492–8. 16 Rajadhyaksha M, Gonzalez S, Zavislan JM et al. In vivo confocal scanning laser microscopy of human skin II: advances in instrumentation and comparison with histology. J Invest Dermatol 1999; 113:293–303. 17 Longo C, Casari A, Beretti F et al. Skin aging: in vivo microscopic assessment of epidermal and dermal changes by means of confocal microscopy. J Am Acad Dermatol 2013; 68:e73–82. 18 ANZMTG 02 12 RADICAL study. Available at: http://www.anzmtg.org/trialdetails.aspx?trialno=16 (last accessed 28 February 2014). 19 Menzies SW, Ingvar C, McCarthy WA. A sensitivity and specificity analysis of the surface microscopy features of invasive melanoma. Melanoma Res 1996; 6:55–62. 20 Menzies SW, Moloney FJ, Byth K et al. Dermoscopic evaluation of nodular melanoma. JAMA Dermatol 2013; 149:699–709. 21 Hashemi P, Pulitzer MP, Scope A et al. Langerhans cells and melanocytes share similar morphologic features under in vivo reflectance confocal microscopy: a challenge for melanoma diagnosis. J Am Acad Dermatol 2012; 66:452–62. 22 Ahlgrimm-Siess V, Massone C, Scope A et al. Reflectance confocal microscopy of facial lentigo maligna and lentigo maligna melanoma: a preliminary study. Br J Dermatol 2009; 161:1307–16. 23 Pellacani G, Guitera P, Longo C et al. The impact of in vivo reflectance confocal microscopy for the diagnostic accuracy of melanoma and equivocal melanocytic lesions. J Invest Dermatol 2007; 127:2759–65. 24 Pellacani G, Vinceti M, Bassoli S et al. Reflectance confocal microscopy features of melanocytic lesions: an internet-based study of the reproducibility of terminology. Arch Dermatol 2009; 145:1137–43. 25 Alarcon I, Carrera C, Palou J et al. Impact of in vivo reflectance confocal microscopy on the number needed to treat melanoma in doubtful lesions. Br J Dermatol 2014; 170:802–8.
© 2014 British Association of Dermatologists
