Prieto, Victor G. MD, PhD
The study of the field of visual perception and psychology became popular in the early XX century. Since that time, scientists have paid attention to how we process sensorial information and, although we do not usually notice, our senses are very selective. At any given moment, there is a huge amount of information being processed from our retinas, cochleae, nose, etc. As an example, the images from the entire retina are delivered to our occipital cortex but during processing of those images our brain selects which ones are “important.” Thus, if we start crossing a street and there is a car speeding fast relative to us, our brain focuses on that image to warn us of the danger while “erasing” the rest of the area captured by the retina because the brain considers it “less” important.
Scientists are still elucidating how the brain decides what is important and what is not. One of the first elements to be analyzed was the relationship between an object and its background. The term “figure-ground segregation” was used by Edgar Rubin1 to indicate the mechanism by which we select some structures as part of a figure while ignoring the surrounding forms. In a sense, when we interpret the images registered in the retina we make a subjective manipulation of the data. Thus, when looking at the famous illustration of a black and white figure (Fig. 1), most observers notice a black vase over a white background. However, when the colored areas are reversed, the observers realize that the figure can also be interpreted as 2 white faces over a black background. Rubin’s explanation was that we select the area that is considered to be important as a form while the surrounding areas are considered to be background. As this study referred to figures over a 2-dimensional surface, Rubin called it “ground,” thus using the term of “figure-ground segregation.” These studies comparing the interpretation of lines and images have shown that when the observer attaches a “meaning” to the image, she/he recognizes more easily that if it is considered to be just a group of lines (in Fig. 1, the vase is relatively easy to detect). Also, the brain recognizes areas differently depending on the color/texture (Fig. 2). Similarly, it seems that the ability to separate between 2 areas (also called “texture segregation”) is based on boundary detection rather than clustering homogenous items and that this processing is performed in the higher areas of our brain cortex.2 Interestingly, the mechanisms of segregation appear to be fairly universal throughout cultures and other animals.3
Obviously, any selection method is very helpful if it really points to the dangerous/important image, but it may be counterproductive if our brain deletes essential data. A famous study4 indicated that 50% of observers missed in a video the presence of an actress dressed up as a gorilla passing among a group of basketball players! And although we may consider that missing a gorilla in a basketball game is not very important, there may be other situations in which “deleting” some portions of information may be crucial. In an example closer to dermatopathology, if there is a small seborrheic keratosis in the epidermis and our brain focuses on that area, we may miss the presence of metastatic melanoma in the subcutaneous tissue!
To try to limit the effect of unconscious selection of data while signing out skin specimens, most pathologists stress the importance of having a systematic approach to sign out. Although most changes in skin biopsies are located in the epidermis and upper dermis, one should always look at the entire slide. At our institution, I recommend to always look at the slide from the top (usually the stratum corneum) to the bottom (dermis/subcutaneous tissue) and from left to right. Such method forces me to at least “see” the entire slide (albeit it does not guarantee that I will “observe” the slide, as Sherlock Holmes would point out). Particularly helpful I think is the method that my mentor, Scott McNutt, used to teach us at the microscope, about “what to do when you see nothing on the slide” or what other authors have indicated as the study of “invisible dermatosis.”5 Obviously, this concept does not refer to histologic changes that are “invisible” and thus, impossible to detect, but rather to subtle changes that may be overlooked at a perfunctory review. In essence, one should look at the stratum corneum in search for neutrophils (possible psoriasis, fungus, acute exanthematous pustulosis, etc.), pigment (suggesting dermatophyte, a melanocytic lesion), bacteria, parakeratosis (that indicates abnormal maturation of the epidermis), thickening (suggesting ichthyosis), etc.; then look at the stratum spinosum to determine if it is normal, thin, thick, or coarse hypergranulosis (consistent with flat wart), eosinophilic change (consistent with epidermodysplasia verruciformis), if there is separation of the epidermis from the dermis (epidermolysis bullosa), amyloid deposition in the papillary dermis, dermal mucin (suggesting lupus), and so on.
Many pathologists like to look at the slides only after establishing a preliminary differential diagnosis based on the available clinical information. However, we recommend starting by examining the slides before knowing the clinical information, to avoid limiting ourselves to search for whatever diagnosis our clinical colleagues wanted us to rule out and thus miss some important finding. Thus, erythematous breast lesions may be referred as “dermatitis” and may show metastatic breast carcinoma within lymphatic spaces. A specimen clinically diagnosed as “keratosis” may show a dense lichenoid infiltrate consistent with a benign lichenoid keratosis but only after we examine carefully the epidermis may we notice that there are scattered cells with clear cytoplasm, thus consistent with melanoma in situ with extensive lichenoid infiltrate.
After we suggest an initial diagnosis we correlate the histologic findings with the clinical information provided, and then we either confirm (hopefully in the majority of cases) or question our diagnosis. If the features do not match we then consider 3 possibilities: (A) our diagnosis is incorrect (as in a “benign lichenoid keratosis” in the context of a disseminated eruption more likely to be lichen planus), (B) the clinical diagnosis is incorrect (as clinically diagnosed “seborrheic keratoses” that are melanomas under the microscope), and (C) there has been a mix-up. This initially “blinded” method of histologic analysis may result in egregious blunders in front of a colleague or a trainee (in more than one occasion I have suggested to our fellows that the features of the specimen were those of benign lichenoid keratosis just to find out that the patient had a disseminated eruption with flat-topped papules thus consistent with lichen planus or lichenoid drug reaction). Still, it is worth having a few “ego-bursting” situations if this method forces us to examine the entire tissue to try to gather all possible relevant information.
Regarding particular situations to which we pay special attention during our sign-out we may group them in 3 categories.
After examining the slide, we always try to correlate with the clinical data. Any discrepancies should be explained.
* Age: Solar elastosis in young individuals may be a sign of xeroderma pigmentosum or, more likely, a switched specimen!
* Location: Prominent sebaceous glands and numerous hair follicles are typical of face biopsies while thick stratum corneum should be seen in acral biopsies. In addition to possibly being helpful for the diagnosis (some entities are more common in some anatomic areas, such as desmoplastic melanoma in sun-damaged skin) determination of the anatomic area again may help us detect a switched specimen.
* When reviewing outside material, after we examine the slide we look at any discordance between the time of the biopsy and the time of review. This is particularly important when evaluating old “nevi” because the reasons for review may be the current diagnosis of a recurrence of a pigmented lesion or a metastasis. However, as mentioned above, we include this information after we establish the histologic diagnosis.
* Caution with dermal surgical scars when we do not know what the original diagnosis was. Always keep in mind the possibility of desmoplastic melanoma, try to detect perineural invasion (both in melanoma and carcinoma), and stromal tumors (such as residual dermatofibrosarcoma protuberans). In case of doubt, S100 protein expression should be detected in the majority of desmoplastic melanomas and antikeratin will help detecting carcinomas (Fig. 3).
* In standard, straightforward nevi, we should be careful in those cases in which the clinical information is “Changing mole.” In such instances we should look for a reason to explain such change: focus of melanoma, ruptured folliculitis, trauma, etc. (Fig. 4). If necessary, it may be helpful to examine additional sections from the block. When correlating with the clinical information, it is important to know the clinician; we have found that some clinicians include the words “rule out melanoma” in almost every pigmented lesion.
* Presence of hyperkeratosis, parakeratosis, and focal dermal fibrosis in a melanocytic lesion may point to previous trauma (Fig. 5). This possibility should then be considered in lesions that show focal pagetoid upward migration limited to such areas of parakeratosis before considering that finding as sufficient for the diagnosis of melanoma. In a related note, lesions that show prominent single-cell growth with focal pagetoid upward migration limited to the area with dermal fibrosis are likely “recurrent nevus/pseudomelanoma” as seen in partially biopsied melanocytic lesions.
* In adults, close relationship of melanocytes with solar elastosis in the dermis usually points to a long-standing lesion. Thus, if melanocytes push aside the solar elastosis, such finding is more consistent with a relatively recent lesion (ie, melanoma) than a long-standing nevus (Fig. 6). Some authors use the presence of elastic fibers among melanocytes as a feature favoring nevus over melanoma.6
* While most melanoma lesions seen in the dermis or subcutaneous tissue and lacking connection with the overlying epidermis are metastatic melanoma, it is always necessary to search for the presence of a benign melanocytic component (associated nevus) in the adjacent dermis/subcutaneous tissue. A possible pitfall is the diagnosis of metastatic melanoma in a dermal proliferation of pigmented melanoma cells if the associated area of blue nevus (thus indicating that it is a primary lesion) is missed during the review of the slides (Fig. 7).
* In sentinel lymph nodes, detection of melanoma metastasis is associated with worse prognosis. However, not all melanocytes seen in the lymph nodes are melanoma cells (up to 10% of patients with melanoma will show benign, nodal nevi in their sentinel lymph nodes).7,8 Apart from the cytologic characteristics of such benign melanocytes (small nucleoli, absence of mitotic figures), their location within the node is essential for the diagnosis. Most nodal nevi will be present in the fibrous capsule. In contrast to nodal nevi, most metastases will occupy the subcapsular sinus. However, a very rare instance is the detection of melanoma cells in lymphatic spaces within the fibrous capsule (Fig. 8). For those cases immunohistochemistry (D2-40, CD31, or CD34) may be needed to detect a rim of endothelial cells around the melanocytes.
* In sentinel lymph nodes containing nodal nevi, these benign melanocytes may be occasionally intraparenchymal. However, such cases usually show an evident area of involvement of the fibrous capsule. Thus, when seeing melanocytes both within the capsule and the immediately adjacent subcapsular sinus, they usually correspond to nodal nevus (Fig. 9).9 In the rare occasions in which nodal nevus and melanoma coexist in the same lymph node, they most frequently occupy different areas of the node.
In our opinion, immunohistochemistry provides additional information that is a very important adjunct tool in the diagnosis of melanocytic lesions.10,11 There are several possible pitfalls when examining immunohistochemical slides.
* S100 protein may be lost if the specimen is underfixed or overfixed or if it has been previously frozen. As sometimes S100 protein is the only “melanocytic” marker expressed in some melanomas (mostly desmoplastic melanomas), such “false” negative may result in an important misdiagnosis (Fig. 10). Furthermore, we have noticed that the epitope expressed in melanocytes is sometimes more labile than that expressed in other cells (Schwann cells, neurons) so therefore it is essential (when available) to check if normal melanocytes (in the epidermis or adnexa) are positive for this marker in the slide, before definitely considering that a tumor is S100 “negative.” Another possible pitfall related to S100 protein is that it can be expressed in other neoplasms. In particular, Paget disease of the breast may express such marker, so keratins should be also included in a panel when evaluating such differential diagnosis.
* Melanomas, as it happens with other neoplasms, not always conform to the expected pattern of expression. They may lack all melanocytic markers (very rarely), express low-molecular-weight keratin, estrogen receptors, CD68, etc.
* MART1 is a very specific marker for melanocytic differentiation. However, some macrophages (mostly those containing melanin pigment) may become positive after antigen retrieval (Fig. 11).12 Such cases require careful examination of the cytologic details and possible use of other melanocytic markers (tyrosinase, HMB45). In contrast, analysis of MART1 expression may be very important in the differential diagnosis of desmoplastic nevi and desmoplastic melanomas. As the immense majority of benign melanocytes express MART1, when dealing with a melanocytic lesion (usually with spindle-cell morphology) that does not express MART1, it is likely to be desmoplastic/spindle-cell melanoma (Fig. 12).11,13
* HMB45 is usually negative in the deep areas of standard nevi and in nodal nevi. However, if the antigen retrieval technique is very robust, such nevus cells may become positive thus suggesting a diagnosis of melanoma. Thus, it is essential that the pathologist be accustomed to the type of intensity that his/her immunolaboratory routinely has in HMB45 slides (Fig. 13).
* MIB1 is very helpful as the immense majority of benign melanocytic lesions show very low (<5%) proliferation rate at the deep areas of the lesion. In cases with significant lymphoid infiltrate, a double immunostudy (eg, with MIB1 and anti-MART1) may help detecting if the proliferating cells are melanocytes.
In summary, the analysis of melanocytic lesions (and any other pathology specimen) should be performed in a systematic way, whichever one is selected by the pathologist. Only by reviewing the entire slide and always considering the possible pitfalls/diagnostic problems will we avoid “missing that invisible gorilla (melanoma)” in our slides.
© 2012 Lippincott Williams & Wilkins, Inc.