The first profound study on regression of melanoma was undertaken by Gromet et al. in 1978. Over a 6-year period, 121 thin malignant melanomas were obseved in order to validate the supposition that thin lesions of low risk but with areas of regression may paradoxically metastasize. [1 ] Of these, 23 displayed areas of regression, of which five (21.7%) metastasized. The incidence of metastasis in their 98 counterparts without regression was 2%. The difference between them was statiscally significant. The authors called attention to the problem of regressing thin melanoma as a neoplasm of "indeterminate risk." They also stated that thin melanomas with regression cannot be staged by conventional means and thus deserve special attention.

A study by Paladugu evaluated 36 thin malignant melanomas and observed that eight of them metastasized within 2 to 120 months. Of the 11 lesions with regressive features, five (45.5%) metastasized, and of 25 lesions without regression, only three (12%) metastasized. [2 ] In agreement with Gromet et al., the findings in this study seemed to indicate that thin melanomas with regressive changes had a propensity to metastasize.

In contrast, McGovern et al. reviewed lesions of 353 patients with thin melanomas (< 0.7 mm) to determine if regression in those lesions was a poor prognostic sign.[3 ] The overall incidence of regression in these lesions was 58%. Only slightly more regressed than unregressed lesions metastasized (8% versus 5% respectively). A high proportion of first recurrences from these thin lesions developed at sites remote from the primary lesion (lung, bone or in subcutaneous tissues or lymph nodes far away from the line of spread). However, the presence or absence of regression in thin lesions did not appear to influence the site of first recurrence. Cumulative 10-year survival rates of patients whose lesions displayed or did not display evidence of regression were nearly identical. The authors concluded, based on their results of long-term follow-up, prognosis was not less favorable for patients in whom regression was present in their thin lesions.

A study by Trau et al. demonstrated that regression is more likely to be found in a malignant melanoma that is level III or less, more than 10 mm in diameter, associated with solar elastosis, located on an anatomic area other than the head or neck, and when there are areas of whiteness clinically. [4 ] The authors suggested that although patients with malignant melanomas displaying signs of regression histologically have a slightly better than 5-year disease-free survival, this may be attributed to a difference in tumor thickness.

Wanebo et al. analyzed 48 patients with melanomas of 1 mm or less in maximal thickness located on the extremities in order to asses the role of various prognostic factors including regression. [5 ] Histologically, tumor regression was found in 50% of the cases. There were no local or distant recurrences in this consecutive series followed for a median of 90 months. In keeping with McGovern et al., the authors concluded that thin melanomas of the extremities, with or without regression, have an excellent prognosis.

In a study by Cooper, a series of melanomas was evaluated in regard to the frequency and extent of inflammatory-fibrotic reactions within melanomas. [6 ] The authors concluded that the most convincing histologic evidence of regression in a thin melanoma consisted of one or more well-defined segments within the lesion where there was a marked reduction or absence of malignant melanocytes in the dermis. The resulting defect was replaced by a variable combination of lymphocytic infiltrate and fibrosis. Melanophages, malignant melanoma cells with degenerative changes, and telangiectasia were also identified to be present. A second group of melanomas met less stringent histologic criteria and were categorized as exhibiting probable regression. The papillary dermis was obviously thickened, but these lesions lacked the well-defined segmental defects. Eleven tumors (23%) had definite regression, 13 (27%) had probable regression, and 19 (40%) lacked regression, although 14 of these contained focal evidence of what the authors called host response.

Søndergaard and Hou-Jensen investigated 486 patients with melanomas in order to evaluate the prognostic importance of partial regression in thin lesions. [7 ] The 10-year survival was 95% for patients without regression in contrast to 79% for patients with past regression. Regression with fibrotic scar tissue adversely affected survival, whereas active regression without fibrotic scar tissue did not influence survival significantly. The wider and thicker the fibrotic area, the poorer the survival. Although the prognostic importance of this finding was not statistically significant, the authors suggested that the horizontal width of the fibrotic area in particular may be a valuable prognostic guide in thin melanomas with past regression.

When Kelly et al. studied 844 malignant melanomas for the presence or absence of histologic regression, no statistically significant effect of regression on survival was found. [8 ]

In a study by Ronan et al. the significance of partial regression in thin malignant melanomas (0.76 mm or less) was evaluated in order to determine if the regression was associated with the later development of metastases in patients who previously were considered to have a favorable prognosis. [9 ] Of 575 patients, 103 (18%) had tumors that measured less than 0.76 mm. Of these, 30 (29%) showed histologic evidence of partial regression. In six (20%) of the 30 patients, visceral metastases developed and the patients died. All six had more than 77% regression of their primary tumors. Of the remaining 24 patients, only one had regression greater than 77%, and she is still alive three years after diagnosis. Most of these 24 (83%) patients had regression of less than 50% (mean, 29.9%). No metastasis occurred in the 73 patients who had thin melanomas without histologic evidence of regression. The authors concluded that patients with thin melanomas who show partial regression cannot be included in the "low-risk" group if the extent of regression is 75% to 80% or more.

Shaw et al. examined 846 patients with melanoma less than 0.76 mm thick in whom follow-up for 2 to 31 years was available. [10 ] Sixty-one (7.2%) developed a recurrence but evidence of regression in thin lesions had no deleterious effect on prognosis.

In contrast, when Slingluff et al. investigated 681 thin melanomas for features identifying patients at risk for an unfavorable course, severe histologic regression was present in 40% of the primary lesions that metastasized and in only 17% of similar lesions that did not (p < 0.001). [11 ]

Shaw et al. also studied 28 melanoma patients who first presented with a thin primary lesion and concurrent regional lymph node metastases and compared these with melanomas of patients whose thin lesions had subsequently recurred. [12 ] Regression was present in 100% of the lesions in the former but only 67% of the latter group. In developing lesions neither recurrence nor metastasis regression was present in 61% of the lesions. The authors suggested that the histology of thin primary melanomas may be influenced by the presence or absence of metastases in patients at that time.

Blessing et al. examined thin malignant melanomas (less than 1.5 mm) of patients who had proven metastasis and death and compared the clinical and histological features with a similar group (less than 1.5 mm) that had no further recurrence after a minimum 6-year follow-up. [13 ] Twenty-six patients with thin melanomas who had developed histologically proven metastasis and/or died following adequate surgical treatment of their primary lesion were identified. When compared with the control group, factors found to be significantly different between the two groups and present in the group that did poorly were (a) histological regression, (b) lesion size, (c) Clark level IV and (d) depth of the uninvolved dermis.

The first immunohistochemical study on regression was undertaken by Tefany et al., in 1991, who examined 36 primary cutaneous melanomas, 19 with regression and 17 without regression comparing lymphocytes populations. [14 ] They identified the cellular phenotypes and activation states of the cells infiltrating these tumors by immunochemistry and found a significantly increased number of CD3-positive cells and an increased ratio of CD4/CD8-positive cells infiltrating regressing compared with non-regressing tumors. The expression of interleukin 2 receptors which is considered to represent an activation marker for T cells, was increased in tumors with regression. The authors concluded that increased numbers of T cells in regressing tumors represented an immune response against the tumor. In contrast, Langerhans cells were found in both regressing and non-regressing tumors, without any significant differences. In regard to expression of class II MHC molecules and intercellular adhesion molecule (ICAM-1) there also was no signifcant difference between the two groups. The authors suggested that, based on their observations, the increase in number of lymphocytes in regressing melanomas was not due to an increase in adhesion molecules or trapping of lymphocytes within the tumor but could be due to increased attraction of lymphocytes to the tumor.

Blessing and colleagues studied 563 primary cutaneous melanomas for the presence of histological regression in relation to the thickness of the lesion and features such as sex, anatomical location and clinical outcome. [15 ] Regression was more common in thin lesions, being seen in 46% of thin (less than 1.5 mm) lesions, 32% of intermediate (1.5-3.0 mm) lesions and 9% of thick (greater than 3.0 mm) lesions. However, severe regression was only identified in 6.5% of thin lesions, 5.2% of intermediate lesions, and 1.5% of thick melanomas. The authors mentioned also that based on their clinical follow-up, regression in thin lesions seemed to be a sinister histological feature, although not of statistical significance.

Slingluff and Seigler studied 681 patients with melanomas less than 0.76 mm thick. [16 ] Severe histological regression was present in 40% of primary lesions that metastasized and only 17% of similar lesions that did not (p < 0.001).

Park et al. examined 555 cases of malignant melanoma less than 1.5 mm in depth . [17 ] Thirty recurred locally or metastasised during follow-up. The authors found that the presence of lesional regression identified a subgroup of thin melanomas which have a poor prognosis.

Halliday et al. investigated spontaneous regression in malignant melanomas and basal cell carcinomas in regard to the infiltrate and found CD4+, but not CD8+, T lymphocytes to be predominant. [18 ] The number of interleukin 2 receptor-positive (early activation marker) but not transferrin receptor-positive (intermediate activation marker) T cells was increased, indicating that the infiltrating T cells were activated. Large numbers of Langerhans cells, macrophages, and other class II major histocompatibility complex (MHC)-expressing cells were present but were not increased in the regressing tumors. The authors concluded that spontaneous regression of human skin tumors is likely to be immunologically mediated, and that CD4+ T lymphocytes seem to mediate this regression.

A retrospective evaluation of 201 cutaneous melanomas was performed by Borogelli to investigate the prognostic significance of histological regression present in 67 cases. [19 ] Thin melanomas showed regression more frequently than thick lesions (48% 0.75 mm versus 12% > 3 mm). Differences between the survival curves were not significant and metastases developed in 13 (19.40%) patients with regressing tumors and in 40 (29.85%) patients with tumors devoid of regression. Although the authors observed a higher frequency of regression in thin melanomas, an influence of regression on disease-free interval and survival could not be demonstrated.

A study of 110 melanomas by Cook and coworkers confirmed regression as on histological feature associated with metastatic spread of a thin melanoma, but 14.8% of thin melanomas devoid of regression were also shown to have metastasized. [20 ]

Guitart and colleagues examined 43 patients with metastasizing thin melanomas and 42 control subjects without metastasis matched for age, sex, tumor site, and Breslow thickness. [21 ] Extensive regression was present in 18 patients (42%) with metastasizing thin melanomas but only in 2 matched control subjects (5%). The authors concluded that thin melanomas with extensive regression represent a group at higher risk for the development of metastasis.

Bernsen and colleagues investigated melanoma-associated antigens on the lymphocytic infiltrates in melanomas and the occurrence of melanoma regression. [22 ] They found that the expression of melanoma-associated antigens as well as the T-cell subset distribution assessed immunohistochemically was heterogeneous within and between lesions and that both were histologically independent phenomena. In four lesions studied in detail by PCR analysis of microdissected T cells, a limited T-cell diversity and evidence for clonally expanded tumour infiltrating lymphocytes were found, but clonal T cells did not show preferential localization to regressive tumor areas.

In a series of 12,728 patients with thin melanomas studied by Leiter et al., follow-up data was available. [23 ] Regression did not affect prognosis significantly.

As recently as in 2008, Kaur et al. investigated 146 melanomas that had undergone sentinel lymph node biopsy for regression. [24 ] Statistically a significantly greater proportion of individuals without regression showed sentinel lymph node positivity compared with those which did show regression. The authors concluded that regression is usually a favorable process, particularly in thin melanomas and that metastasis in thin melanomas showing regression is rare.

Similarly, Cecchi et al. investigated 59 patients with thin melanomas who underwent sentinel lymph node biopsy. [25 ] Histological regression was seen in 45 (76.3%) lesions and sentinel lymph node metastases were detected in 2 of 59 patients (3.4%), but only one patient with a positive sentinel lymph node exhibited histological regression of his tumour. The authors concluded that tumor regression is not a predictor of sentinel lymph node metastasis in patients with thin melanomas, and therefore does not justify the routine use of lymphatic mapping and sentinel lymph node biopsy in this melanoma setting.