The majority of the articles on regression in cutaneous melanomas concern the patients' prognosis, and there is no unanimity among the authors. While some authors found histological evidence of regression to be a bad prognostic sign, [1,2,7,9,13,16,17,20, 21 ] others stated that it had no meaning in regard to prognosis, [3,8,10,19,23 ] and still others identified it as a signal of a favorable prognosis. [4,5,19 ] One reason for these contradictory results is that the authors had very different definitions of an unfavorable course. Whereas some used the evidence of metastasis as a criterion, [1-3,9,10,16,17,20,21 ] others used the disease-free survival, [4 ] and still others the overall survival [3,4,7,8,19,23 ] as indicators of a good versus a bad prognosis. But even in studies that used a similar definition of an unfavorable course, predictive value of histological regression was controversial. Interestingly, the majority of studies using only metastasis as a criterion identified regression as a bad prognostic sign, [1, 2,9,11,13,16,17,20,21 ] whereas the majority of studies using overall survival demonstrated that the histologic feature of regression was of no prognostic significance. [ 3,8,19,23 ] Strictly speaking, only overall survival is a valid criterion to estimate a patients prognosis.

Even though Gromet et al, as early as 1978, alerted to the fact that regression may limit the ability to asses tumor depth accurately, only very few articles have focussed on the histopathologic features of regression. [6,7,13 ] In many of the other studies, no mention is made how, precisely, the authors identified regression. When criteria were given, they differed considerably between the studies and included very subjective features. In addition, regression is often a patchy phenomen and most studies have failed to quantify the process. A summary of criteria used by various authors is given in table 1.

In contrast, authors of textbooks of dermatopathology attempted to give more precise criteria for recognition of regression. [26,27 ] According to Ackerman et al., regression of primary cutaneous melanoma usually involves the papillary dermis in the form of either "fibrosis" or "melanosis," the former being fibroplasia and the latter a broad sheet of melanophages, both occurring in a thickened papillary dermis. Ackerman et al. admitted, however, that clear separation between both types of regression was not always possible. Sometimes, "fibrosis" and "melanosis" occured together in the same section of tissue, and although melanosis, by definition, consisted entirely of melanophages, "fibrosis" was accompanied nearly always by some melanophages, lymphocytes, and telangiectases. Ackerman and colleagues further classified regression into "total regression" (melanoma destroyed in its entirety, no residuum); "focal regression" (regression occuring in a discrete locus devoid of any sign of melanoma in situ); and "partial regression" (regression occuring in a locus together with evidence of melanoma in situ in the overlying epidermis). [26 ]

Detailed also, but slightly different, Massi and LeBoit provided criteria for diagnosis of regression. They differentiated between a regressive lesion, one with dense, band-like inflammatory infiltrates extending into junctional nests and blurring the dermoepidermal interface, with variable amounts of plasma cells, melanophages, necrotic or entirely absent melanocytes, and a regressed lesion typified by a slightly inflammed fibrotic band in papillary dermis which progressed to a greatly thickened papillary dermis replaced by a delicate network of fibrillary collagen, few fibroblasts, and scattered lymphocytes. Areas of edema, ectatic capillaries, melanophages, residual atypical melanocytes, and epidermal atrophy were mentioned as inconsistent findings. [27 ]

None of the reviewed studies on regression applied such detailed criteria and classification of regression in melanomas or correlated them with overall survival.

Articles that attempted to understand the mechanisms of regression in melanomas mainly focussed on the analysis of the cellular infiltrate, assuming that regression is a sign of immunological destruction of the malignant neoplasm by the organism that bares it. [14,18,22 ] As of this date, it is not known why regression occurs in some melanomas and not in others, why it is seen more commonly in thin melanomas, and whether it is consequent to spread of melanoma cells to a lymph node or unrelated to it.