2. Primary Melanoma of the Lung

The search strategy used in our study allowed the identification of 52 papers reporting on 76 cases. Rodriguez et al. in a case report published on April 2019 reported that only 41 PMML had been published since 1916, but our systematic search produced a consistently higher number of cases published in the last 30 years, which allows a first global evaluation of the main clinical, pathological and therapeutic issues of the disease. Neri et al. ^[4] reported on two cases, but only one of them was PMML. Christopoulos et al. published a case of a stage IV malignant melanoma, with a pulmonary mass and cervical lymph node involvement ^[5]. Pathological analyses performed on the cervical lymph nodes excised led to diagnosis of melanoma, and the authors claimed that the primary tumor was a MM of unknown origin. Therefore, we excluded this case, even if its description perfectly fits with that of a PMML. Rao et al. published a case of a large congenital melanocytic nevus with metastatic melanoma with a probable primary in the lung, but this origin was not ascertained, and thus we did not include the case ^[6]. A case initially diagnosed as PMML and subsequently revealed to be a metastatic tumor, as described by Shimmyo et al., was also excluded. Conversely, we included the case described by Karpathiou et al. because it is a very likely a PMML, rather than a recurrence occurred 40 years after the primary diagnosis, as supposed by the authors ^[7]. The largest series included in the study were published by De Wilt et al. in 2005 including 15 cases ^[8], and Wilson and Moran in 1997 including eight cases ^[9].

Like its cutaneous counterpart, PMML in our study was found to affect middle-aged adults (median age 60 years), as opposed to lung cancer which generally affects older individuals ^[10][11][12]. Furthermore, it mostly affects males than females, similarly to cutaneous melanoma and lung cancer in developing countries; in western countries differences in incidence rates of lung cancer between males and females are continuously narrowing due to changes in smoking patterns between sexes ^[2][12]. Approximately, two thirds of the cases included in our study affected Caucasian patients, and one third Asian patients, with only one case described in Africa; it is not clear if the difference depends merely on the real incidence of the disease, which in its cutaneous form is known to affect more frequently fair-skinned than dark-skinned individuals, or on differences regarding the existence of cancer registries and scientific publishing rates. In addition, it is not clear if sun exposure, a well- known risk factor for the development of cutaneous MM, plays any role in the onset of PMML. What seems to be related with the occurrence of PMML is a personal or family history of cancer, that was encountered respectively in 25% and 41.2% of the cases with available data. Approximately half of the patients with data regarding their smoking habits were never smokers, which suggests a different pathogenic role of cigarette smoking on PMML in comparison to lung cancer. Finally, only in rare cases previous non-malignant cutaneous lesions were excised in PMML patients, suggesting also potential differences with the traditional pathogenic mechanisms of cutaneous MM.

The clinical manifestations of the disease were similar to those of other thoracic malignancies, including cough, hemoptysis, dyspnea, chest pain, and signs and symptoms related to metastases. An interesting case of dark sputum was reported by Fillipini et al., which represents certainly a particular but rare clinical manifestation of PMML ^[13].

Most of the lesions in our study affected the left lung and showed a predilection for the lower lobes. This trend is different in comparison with lung cancer, which is more prone to affect the right lung and the upper lobes ^[2]. In addition, 71% of the lesions with available data were resided in the central areas of the lungs, as opposed to lung adenocarcinoma which is currently the most common type of lung cancer and occurs mainly in the peripheral areas of the lungs; these patterns should be kept in mind for differential diagnosis, when clinical elements suggest a PMML. Furthermore, despite 60% of the 55 patients with available data had metastasis at the time of diagnosis, only one third presented mediastinal lymph node involvement; this percentage appears lower than that of cases with lung cancer, where the lymph nodes of the mediastinum are early involved in the metastatic process. On the other hand, extra-thoracic lymph node involvement was reported in 14 cases, and in seven of them it represented the only distant metastasis detected at the time of diagnosis. These figures suggest a different behavior in the lymphatic progression of PMML in comparison with lung cancer, but more well-documented cases are necessary to better establish it.

Cytology was performed in a few cases prior to histology, and it seems to have a role in rise the suspicion of the disease, rather than in establishing the diagnosis. Histological examination was performed in all the cases, and in approximately 87% black nodules or lesions containing pigmented areas were macroscopically described; depigmentation may occur in PMML of advanced less differentiated stages, making the diagnosis particularly difficult, as described by Karpathiou et al. ^[7]. The mean maximum diameter of the main lesions was approximately 5 cm, which explains the low rate of asymptomatic cases and the consistent percentage of metastases found. The lesions were generally characterized by organoid and less frequently fascicular growth pattern and include epithelioid to spindle cells with a variable percentage and predominance. Other than the presence of other typical morphological features of MM like the presence of cytoplasmic pigment, large nuclei with prominent nucleoli, nuclear atypia, and mitotic figures, were commonly described. Additional features were necrosis, as well as presence of epithelial and to a lesser extend spindle cells. These characteristics, along with immunohistochemical positivity for S-100, HMB 45 and melan-A, and negativity for cytokeratin and chromogranin, are generally enough to reach pathological diagnosis.

Molecular analysis for BRAF mutations was performed only in few cases, but it can be explained by the fact that these tests were introduced in clinical practice only recently, as they represent predicting biomarkers of response to therapy with modern targeted agents like dabrafenib or vemurafenib ^[14]. BRAF mutations have been detected in about 50% of cutaneous MM cases and to a lesser extent in other melanoma subtypes ^[15]. Only one among the eleven cases tested in this series has been found mutated (11.1%), and no conclusions can be drawn regarding the incidence of BRAF mutations in PMML, given the small number of cases. Only one wild type NRAS case was also reported.

Surgery, when feasible, was the main option for the treatment of PMML, performed in more than 70% of the patients, and in some cases to treat also recurrences. This is a high percentage, in comparison for example with patients with primary lung cancer or malignant mesothelioma ^[2][3], but it must be kept in mind that in the 15 cases reported by De Wilt et al. and the eight cases reported by Wilson and Moran surgery substantially was one of the selection criteria. The most common surgical procedures performed were lobectomies and wedge parenchymal resections (Table 1). A few data were available regarding chemotherapy which was used in 20 cases; dacarbazine alone was the treatment most frequently reported, mainly in recent cases. Combined treatments, some of them including dacarbazine were reported in a few reports, while immunotherapy (often combined with interferon) was reported in eight recent cases, and targeted therapy in the single BRAF-mutated case (Table 1).

Table 1. Treatment options in patients with primary lung melanoma.

The mean survival time reported was 24.3 months, but as we mentioned before, it must be kept in mind that in reports including numerous cases, there were only surgical, and thus, relatively early stage tumors. The mean overall survival calculated including the 36 dead patients was 14.3 months; this figure is even worse than those of lung cancer or malignant mesothelioma in their early stages or locally advanced stages. As expected, the median survival was significantly higher in patients who underwent surgery in comparison to those who did not. The advent of targeted therapies and immunotherapies, which brought consistent advantages in the clinical management of cutaneous MMs ^[15] and were used in a few recent cases of PMML, may help improving also the outcomes of this rare disease.

Our review has some limitations regarding the diagnosis, clinical evaluation and treatment of PMML, which need to be highlighted. First of all, the criteria for the diagnosis were variable, with the common denominator of a malignant melanoma of the lung with no evidence of other malignancies; nevertheless, the exclusion of other malignancies might not have been highly accurate, especially in an era in which positron emission tomography (PET) and high definition imaging techniques were not widely available. In addition, the precise genetic landscape of these tumors has not been thoroughly investigated; BRAF or NRAS mutations were rarely searched, while wider sequencing approaches have not been used to describe the genetic alterations that commonly occur in PMML. This would be particularly useful for the exact diagnosis of these tumors and for a modern therapeutic approach based on precision oncology treatments.