BJD

British Journal of Dermatology

C L I N I C A L A N D L A B O R A T O R Y I N V E S TI G A T I O N S

Favourable prognostic role of regression of primary melanoma in AJCC stage I–II patients S. Ribero,1,2 S. Osella-Abate,1 M. Sanlorenzo,1 P. Savoia,1 C. Astrua,1 G. Cavaliere,1 C. Tomasini,3 R. Senetta,4,5 ,2 M.G. Bernengo1 and P. Quaglino1 G. Macripo 1

Section of Dermatology and 4Section of Surgical Pathology, Department of Medical Sciences, University of Turin, via Cherasco 23, 10126 Turin, Italy Section of Dermatologic Surgery, Department of Oncology and Haematology, via Cherasco 23, 10126 Turin, Italy 3 Division of Anatomic Pathology IV, via Cherasco 23, 10126 Turin, Italy and 5Department of Laboratory Diagnostics, AOU Citta della Salute e della Scienza di Torino, via Santena 7, 10126, Turin, Italy 2

Summary Correspondence Pietro Quaglino. E-mail: [email protected]

Accepted for publication 6 August 2013

Funding sources None.

Conflicts of interest None declared. The first two authors contributed equally. DOI 10.1111/bjd.12586

Background The prognostic significance of regression in primary melanoma has been debated over the past few years. Once it was considered to be a negative prognostic factor, as it may have prevented proper melanoma thickness measurement, therefore affecting the staging of the tumours. For this reason, it was considered to be an indication for sentinel lymph node biopsy (SLNB) in melanoma < 1 mm. Objectives To ascertain the utility of SLNB in thin melanoma and to clarify the role of regression in disease-free survival (DFS) and overall survival (OS) in our series. Methods We analysed data collected from 1693 consecutive patients with AJCC (American Joint Committee on Cancer) stage I–II melanoma. Results Globally, SLNB was performed in 656 out of 1693 patients. Regression was present in 349 patients and 223 of them were characterized by thin lesions. SLNB was performed in 104 cases of thin melanoma with regression. The majority of regional lymph node metastases were observed in patients who did not undergo SLNB (89 out of 132). Among the remaining 43 ‘false negative’ patients only three showed regression in the primary. Using the Cox multivariate model, histological regression maintained a significant protective role [hazard ratio (HR) 0
62, P = 0
012 for DFS; HR 0
43, P = 0
008 for OS] when corrected for the principal histopathological and clinical features, despite SLNB. Conclusions We confirmed that regression alone should not be a reason to perform SLNB in thin melanoma and, on the contrary, it can be considered a favourable prognostic factor in patients with AJCC stage I–II melanoma.

What’s already known about this topic?

• •

Regression has been considered a negative prognostic factor, as it may prevent proper melanoma thickness measurement. There is no consensus regarding the need for sentinel lymph node biopsy (SLNB) when regression is present within the primary tumour.

What does this study add?

• •

Regression alone should not be a reason to perform SLNB in thin melanoma. Regression can be considered a favourable prognostic factor for overall survival and disease-free survival in patients with AJCC stage I–II melanoma.

The prognostic significance of regression in primary melanoma has been debated over the past few years. Previously, it

was considered to be a negative prognostic factor, as it may have prevented proper melanoma thickness measurement,

British Journal of Dermatology (2013) 169, pp1240–1245

© 2013 British Association of Dermatologists

1240

Histological regression in primary melanoma, S. Ribero et al. 1241

therefore affecting the staging of the tumour.1–4 There is no consensus regarding the need for sentinel lymph node biopsy (SLNB) when regression is present within the primary tumour.5–10 Morton et al.5 found that significant regression, together with other factors (age < 40 years, shave biopsy with positive deep margins, ulceration and Clark level IV), increases the risk of sentinel node metastasis from a thin primary melanoma and could be an indication for performing an SLNB. Other studies support the negative prognostic role of regression. Guitart et al.4 related it to an increased risk of distant metastasis in patients with thin melanoma. Shaw et al.11 hypothesized that the presence of metastatic melanoma in a regional lymph node might stimulate an immune response resulting in regression of the primary lesion. However, other studies indicate that regression does not increase the risk of metastases.8,12,13 Kaur et al.8 considered primary regression to be a favourable prognostic feature in patients with melanoma and demonstrated that is not associated with a higher risk of metastatic sentinel lymph node (SLN). White et al.14 reported that the presence of regression is associated with a lower likelihood of positive SLN. Likewise, Ma et al.12 showed that melanoma regression and the increase of dendritic cells in the primary lesion are associated with significantly lower SLN involvement. Furthermore, a previous study from our group identified regression as a potential favourable prognostic factor for disease progression after a negative SLNB.13,15 Based on this finding and on the controversial data in the literature, we performed a study to confirm the prognostic role of regression of primary melanoma on disease-free survival (DFS) and overall survival (OS) in patients with stage I–II AJCC (American Joint Committee on Cancer) melanoma in a larger series of patients.

Materials and methods Clinical data from 1693 consecutive patients with an invasive melanoma (AJCC stage I–II)16 were analysed in this study. All patients were diagnosed and followed up at our institution from 1 January 1999 to 31 December 2011. The present series includes 503 patients previously reported.13 Patients were classified on the basis of AJCC criteria16 and treated and followed up according to standard guideline criteria.17–19 SLNB was performed in the presence of melanoma > 1 mm, excluding patients with comorbidities, those aged > 75 years, or those who refused the procedure. After 1999, for patients with thin melanoma (< 1 mm) the presence of ulceration and/or mitotic rate > 1 mm 2 was an indication for performing SLNB. From 1999 to 2008, regression and other significant predictors of SLN positivity (younger age, vertical growth phase) were also indications for performing SLNB. After 2008, in accordance with the European consensus-based interdisciplinary guideline for diagnosis and treatment of melanoma20 and because of the lack of data in the literature confirming the usefulness of this procedure in the case of thin melanoma, regression was no longer consid© 2013 British Association of Dermatologists

ered to be an indication. SLNB was still recommended for melanomas < 1 mm in the presence of ulceration and/or mitotic rate > 1 mm 2, Clark level IV and vertical growth phase.20 Patients with a positive SLNB (stage III) were excluded in this study.16 As a result, SLNB was performed in 182 out of 349 patients with regression of a primary tumour (52
1%) and in 474 out of 1344 patients (35
3%) without regression. SLNB was performed in 104 out of 223 patients with melanoma < 1 mm and with regression. The remaining 119 patients did not undergo the procedure because eight of them showed comorbidities, 46 were aged > 75 years and 65 were diagnosed after 2008. Regression was evaluated on sections from melanomas stained with haematoxylin/eosin. On the basis of our experience and a literature review,21,22 the following histological criteria for regression were considered: reduction or disappearance of the melanoma within a circumscribed area of the tumour; inflammatory infiltrate; fibrosis; melanophages; increased vascularity; epidermal thinning; and necrotic or apoptotic keratinocytes/melanocytes. The histological diagnosis of regression was independently evaluated by two experienced dermatopathologists. Cases where there was disagreement were examined under the microscope, and if a consensus was not reached, then regression was considered to be absent. Statistical analyses Statistical analyses were performed using the Stata 11.0 statistical software (StataCorp, College Station, TX, U.S.A.). In a preliminary analysis, both nonparametric (v2-test) and parametric tests (t-test) were used. Odds ratios (ORs) were calculated with logistic regression. DFS was calculated from the surgical excision of primary melanoma to the date of first disease relapse and/or progression or last check-up. OS was calculated from the surgical excision of primary melanoma to the date of death or last check-up. Survival estimates were derived using the Kaplan–Meier method and the statistical comparison was done using the log-rank test. Multivariate analyses were carried out to evaluate the influence of different variables on DFS and OS using Cox regression. Potential effect modification of histological regression by Breslow thickness and ulceration was evaluated by introducing an interaction term in the regression model.

Results Out of the entire cohort of 1693 patients, 349 (20
6%) presented regression of the primary lesion. Clinicopathological features of patients with or without regression are shown in Tables 1 and 2. Median follow-up was 4
8 years (range 1– 13
1). The presence of regression was significantly associated with specific clinical (male sex, older age, head/neck or trunk localization of primary tumour) as well as histological features [lower Breslow thickness, absence of ulceration and superficial spreading melanoma (SSM) histotype]. British Journal of Dermatology (2013) 169, pp1240–1245

1242 Histological regression in primary melanoma, S. Ribero et al. Table 1 Clinical features of patients stratified based on the presence of regression

Total, n Age (years), mean  SD Ulceration, n No Yes Sex, n Female Male Site, n Head/neck Trunk Upper limb Lower limb Breslow thickness (mm), mean  SD Breslow thickness (mm) (AJCC stage), ≤1 > 1 to ≤ 2 > 2 to ≤ 4 >4 Histotype, n SSM NM ALM LMM Other SLN biopsy, n No Yes

No regression

Regression

P-value

1344 55
31  16
19

349 57
76  14
45

0
01

1076 268

311 38

0
001

663 681

141 208

0
003

138 575 188 443 2
21  2
57

21 205 45 78 1
41  1
68

0
001

n 621 244 267 212

223 61 42 23

0
001

992 193 59 76 24

318 8 6 14 3

0
001

870 474

167 182

0
001

0
001

AJCC, American Joint Committee on Cancer; SSM, superficial spreading melanoma; NM, nodular melanoma; ALM, acral lentiginous melanoma; LMM, lentigo maligna melanoma; SLN, sentinel lymph node.

During follow-up, progression occurred in 20
5% of patients. The presence of histological regression was associated with a significant decrease in the risk of disease progression (univariate logistic regression OR 0
34, P < 0
001). In fact, only 33 out of 349 patients with regression (9
5%) showed

progression of disease, compared with 314 of 1344 patients without regression (23
4%). No differences were found in the percentage of relapse in patients with a negative SLNB (135 out of 656; 20
6%) when compared with patients in whom SLNB was not performed (212 out of 1037; 20
4%). The pattern of first relapse was characterized by 132 cases of regional lymph node involvement, 92 in transit metastases and 123 distant metastases. This pattern had no statistically significant difference between patients who did or did not have an SLNB. The presence of regression in a primary tumour was associated with a statistically significant lower incidence of all the three patterns of relapse: ‘in transit’ (11/349, 3
2% vs. 81/1344, 6.0%; P = 0
048), regional lymph node (14/ 349, 4.0% vs. 118/1344, 8
8%; P = 0
004) or distant metastases (8/349, 2
3% vs. 115/1344, 8
6%; P < 0
001) (Table 2). Among the 132 patients who developed regional node involvement, 43 had a previous negative SLNB (‘false negative’ SLNB). The finding of a ‘false negative’ SLNB was less frequent in patients with primary regression (three out of 182, 1
6%) than in those without this feature (40 out of 474, 8
4%) (P = 0
003). The survival study showed differences in DFS and OS of patients stratified for regression using the log-rank test (Fig. 1). With the univariate logistic regression, only 11 of the 349 patients with regression (3
2%) died during the follow-up, compared with 177 of the 1344 patients without regression (13.2%) (OR 0
21, P < 0
001). Using the Cox multivariate model, regression maintained a significant protective role [hazard ratio (HR) 0
62, P = 0
012 for DFS; HR 0
43, P = 0
008 for OS) for both OS and DFS, when corrected for Breslow thickness and clinical and histopathological (including SLNB) features of patients with melanoma (Table 3). The analysis of potential effect modification of histological regression by ulceration and Breslow thickness showed that ulceration was not an effect modifier of histological regression on DFS (P = 0
83) or on OS (P = 0
38), whereas statistical interaction on histological regression by Breslow thickness was significant for DFS (P = 0
05) and for OS (P = 0
04). As the Breslow thickness increases, the effect of histological regression goes towards null on both DFS [HR 0
52, 95% confi-

Table 2 Pattern of metastases Sentinel lymph node biopsy No

Progressive disease Site of first metastases, n In transit Regional lymph node Distant

Yes

Without regression

With regression

Without regression

With regression

192/870

20/167

122/474

13/182

48 78 66

4 11 5

33 40 49

7 3 3

British Journal of Dermatology (2013) 169, pp1240–1245

© 2013 British Association of Dermatologists

Histological regression in primary melanoma, S. Ribero et al. 1243

Fig 1. Kaplan–Meier curves of overall survival (OS) and disease-free survival (DFS).

dence interval (CI) 0
33–0
80 for a unit increase of Breslow thickness] and OS (HR 0
30, 95% CI 0
14–0
66 for a unit increase of Breslow thickness).

Discussion Regression in primary cutaneous melanoma occurs in 10–35% of cases.2 In our series, this parameter characterized 20
6% of patients. Regression has traditionally been considered as a marker of poor prognosis, even though the majority of studies used data from small series of patients. Particularly in thin melanomas, it

may prevent a real evaluation of the initial thickness of the melanoma.1–4 The introduction of SLNB has recently changed the clinical approach to patients with melanoma.5 Burton et al.23 analysed 2220 patients with melanoma < 1 mm who underwent SLNB. In this cohort, regression was not significantly associated with SLN metastasis nor with DFS or OS, suggesting that this parameter should not be used to guide clinical decision-making for such patients. Moreover, in this study, regression accounted for only 11
7% of cases, even though the population was remarkable.23 Other authors reported that the presence of regression is associated with a lower possibility of a positive SLN and even better clinical outcome.8,12–15,24,25 The aim of the present study was to ascertain further which clinicopathological features are associated with the presence of regression, the prognostic role of regression on DFS and OS, and the benefit of SLNB in thin melanoma with regression. Our study showed that regression is associated with specific clinicopathological features of the primary lesion. Furthermore, it characterized a group of patients with a lower risk of progression and a more favourable clinical course of the disease. Patients with regression were more likely to be male and to be older; their melanoma was more frequently an SSM subtype, plus it was located on the head/neck or on the trunk, as described by Burton et al.23 As previously reported,1–3 as well as in our series, thin melanomas showed regression more frequently than thick melanomas (223/844, 26
4%). The favourable role of regression is clearly demonstrated by evidence from our data. No difference in the pattern of first relapse was found between melanoma with or without regression, the latter showing a lower incidence of either ‘in transit’, nodal or visceral recurrences as the first site of relapse. The protective role of regression was also confirmed by the evidence that ‘false negative’ SLNBs were more frequent in patients without regression (8
4%) than in those with this histological finding (1
6%). Moreover, multivariate Cox analysis (correcting the model also with the feature SLNB) validated this favourable role in our series of stage I–II AJCC patients. Finally, ulceration was not an effect modifier of histological regression, but Breslow thickness was. Despite this, in the Cox model, histological regression maintained a protective role on DFS and OS, even if Breslow thickness remained the most negative prognostic factor.

Table 3 Multivariate Cox analysis on overall survival (OS) and disease-free survival (DFS) DFS

Sex Age Breslow thickness Ulceration Regression SLNBa

OS

HR

CI

P-value

HR

CI

P-value

1
13 1
02 1
23 1
85 0
62 0
88

0
91–1
40 1
02–1
03 1
20–1
27 1
43–2
38 0
43–0
90 0
69–1
11

0
267 0
000 0
000 0
000 0
012 0
277

1
45 1
02 1
23 2
16 0
43 1
06

1
08–1
97 1
01–1
03 1
18–1
27 1
54–3
02 0
23–0
80 0
78–1
44

0
014 < 0
001 < 0
001 < 0
001 0
008 0
714

HR, hazard ratio; CI, confidence interval; SLNB, sentinel lymph node biopsy. aPerformed or not.

© 2013 British Association of Dermatologists

British Journal of Dermatology (2013) 169, pp1240–1245

1244 Histological regression in primary melanoma, S. Ribero et al.

Besides confirming the independent favourable prognostic relevance of regression, multivariate analyses also showed that this role is independent of the execution of the SLNB. This finding limited a potential bias in our study constituted by the nonhomogeneous patient characteristics with respect to SLNB. Indeed, regression in melanoma < 1 mm was an indicator for SLNB until 2008, thereafter it was no longer considered an indication for performing SLNB due to the lack of data in the literature confirming the usefulness of this procedure in this subgroup of patients.6–10,24,25 Further evidence that reduces this study bias is also represented by the finding that no difference in the pattern of first relapse was observed between patients who did or did not have an SLNB. The biological scenario leading to the protective role of regression is not well understood. Nevertheless, it can be suggested that a host immunological response to a tumour could be the basis of regression. This phenomenon could reflect the power of the immunological system against a primary tumour, and its presence should be considered prognostically favourable. More recently, Ma et al.12 showed that the immune profile of the primary melanoma predicts SLN metastatic involvement. Additionally, the presence of primary tumour regression results from a T-cell immune response associated with a decreased risk of nodal progression. A downregulation of the antitumour immunity in the positive SLN with an increase in regulatory T cells compared with the negative nonsentinel node from the same nodal basin was described. Furthermore, it was ascertained that primary tumour conventional dendritic cells and regression were protective against SLN metastasis.12 This hypothesis fits well with the clinical finding that false negative SLN rates were lower in patients with regression in our series. Moreover, out of the patients with regression of the primary tumour, only 4.0% developed lymph node regional metastases; this percentage was significantly lower than in patients without regression (8
8%). In conclusion, based on our clinical data, even though Breslow thickness and ulceration are the major prognostic factors for DFS and OS, regression can be considered a favourable prognostic factor for OS and DFS in patients with AJCC stage I–II melanoma and should not be a reason for performing SLNB in thin melanomas.

Acknowledgments The authors thank Prof. A. Sapino (Department of Medical Sciences, Section of Surgical Pathology) and Dr D. Zugna (Department of Medical Sciences, Unit of Cancer Epidemiology) for the paper revision, as well as M. Zekhtser and J. Ma (Melanoma Center, University of California, San Francisco) for the language revision.

References 1 Ronan SG, Eng AM, Briele HA et al. Thin malignant melanomas with regression and metastases. Arch Dermatol 1987; 123:1326– 30. British Journal of Dermatology (2013) 169, pp1240–1245

2 Blessing K, McLaren KM, McLean A, Davidson P. Thin malignant melanomas (less than 1.5 mm) with metastasis: a histological study and survival analysis. Histopathology 1990; 17:389–95. 3 Slingluff CL Jr, Seigler HF. ‘Thin’ malignant melanoma: risk factors and clinical management. Ann Plast Surg 1992; 28:89–94. 4 Guitart J, Lowe L, Piepkorn M et al. Histological characteristics of metastasizing thin melanomas: a case–control study of 43 cases. Arch Dermatol 2002; 138:603–8. 5 Morton DL, Cochran AJ, Thompson JF et al. Sentinel node biopsy for early-stage melanoma: accuracy and morbidity in MSLT-I, an international multicenter trial. Ann Surg 2005; 242:302–11. 6 Gershenwald JE, Thompson W, Mansfield PF et al. Multi-institutional melanoma lymphatic mapping experience: the prognostic value of sentinel lymph node status in 612 stage I or II melanoma patients. J Clin Oncol 1999; 17:976–83. 7 Lourari S, Paul C, Gouraud PA et al. Sentinel lymph node biopsy for melanoma is becoming a consensus: a national survey of French centres involved in melanoma care in 2008. J Eur Acad Dermatol Venereol 2012; 26:1230–5. 8 Kaur C, Thomas RJ, Desai N et al. The correlation of regression in primary melanoma with sentinel lymph node status. J Clin Pathol 2008; 61:297–300. 9 Warycha MA, Zakrzewski J, Ni Q et al. Meta-analysis of sentinel lymph node positivity in thin melanoma (≤ 1 mm). Cancer 2009; 115:869–79. 10 Socrier Y, Lauwers-Cances V, Lamant L et al. Histological regression in primary melanoma: not a predictor of sentinel lymph node metastasis in a cohort of 397 patients. Br J Dermatol 2010; 162:830–4. 11 Shaw HM, McCarthy SW, McCarthy WH et al. Thin regressing malignant melanoma: significance of concurrent regional lymph node metastases. Histopathology 1989; 15:257–65. 12 Ma MW, Medicherla RC, Qian M et al. Immune response in melanoma: an in-depth analysis of the primary tumor and corresponding sentinel lymph node. Mod Pathol 2012; 25:1000–10. 13 Savoia P, Fava P, Caliendo V et al. Disease progression in melanoma patients with negative sentinel lymph node: does false-negative specimens entirely account for this phenomenon? J Eur Acad Dermatol Venereol 2012; 26:242–8. 14 White RL Jr, Ayers GD, Stell VH et al. Factors predictive of the status of sentinel lymph nodes in melanoma patients from a large multicenter database. Ann Surg Oncol 2011; 18:3593–600. 15 Ribero S, Quaglino P, Osella-Abate S et al. Relevance of multiple basin drainage and primary histologic regression in prognosis of trunk melanoma patients with negative sentinel lymph nodes. J Eur Acad Dermatol Venereol 2012; 27:1132–7. 16 Balch CM, Gershenwald JE, Soong SJ et al. Final version of 2009 AJCC melanoma staging and classification. J Clin Oncol 2009; 27:6199–206. 17 Cochran AJ, Balda BR, Starz H et al. The Augsburg consensus. Techniques of lymphatic mapping, sentinel lymphadenectomy, and completion lymphadenectomy in cutaneous malignancies. Cancer 2000; 89:236–41. 18 Sober AJ, Chuang TY, Duvic M et al. Guidelines of care for primary cutaneous melanoma. J Am Acad Dermatol 2001; 45:579–86. 19 Quaglino P, Borgognoni L, Bottoni U et al. Italian guidelines for staging and follow-up of stage I-II cutaneous melanoma patients. G Ital Dermatol Venereol 2007; 142:41–7. 20 Garbe C, Peris K, Hauschild A et al. Diagnosis and treatment of melanoma: European consensus-based interdisciplinary guideline. Eur J Cancer 2010; 46:270–83. 21 Kang S, Barnhill RL, Mihm MC Jr, Sober AJ. Histologic regression in malignant melanoma: an interobserver concordance study. J Cutan Pathol 1993; 20:126–9.

© 2013 British Association of Dermatologists

Histological regression in primary melanoma, S. Ribero et al. 1245 22 Requena C, Botella-Estrada R, Traves V et al. Problems in defining melanoma regression and prognostic implication. Actas Dermosifiliogr 2009; 100:759–66. 23 Burton AL, Gilbert J, Farmer RW et al. Regression does not predict nodal metastasis or survival in patients with cutaneous melanoma. Am Surg 2011; 77:1009–13.

© 2013 British Association of Dermatologists

24 Morris KT, Busam KJ, Bero S et al. Primary cutaneous melanoma with regression does not require a lower threshold for sentinel lymph node biopsy. Ann Surg Oncol 2008; 15:316–22. 25 Testori A, De Salvo GL, Montesco MC et al. Clinical considerations on sentinel node biopsy in melanoma from an Italian multicentric study on 1313 patients (SOLISM-IMI). Ann Surg Oncol 2009; 16:2018–27.

British Journal of Dermatology (2013) 169, pp1240–1245