Introduction
Cutaneous melanoma is characterized by a high risk of metastases occurring in close proximity to the primary melanoma, along the route to and in regional lymph nodes, and as distant metastases (1). Despite approved adjuvant treatment options for primary high-risk melanoma (2) and life-prolonging therapies in the stage of distant metastasis (3), melanoma remains a fatal disease.
In-transit metastases (ITM) are a special kind of recurrence in the skin, observed predominantly in melanoma and approximately develop in 7% of the melanoma patients (4). They occur in the anatomical region between primary melanoma and the corresponding regional lymph node basin. The clinical course of patients developing ITM varies considerably, ranging from solitary lesions with prolonged relapse-free intervals up to subsequent, rapidly progressing systemic disease (1). ITM can be palpable and visible and therefore represent a significant psychological burden for patients (5). Apart from the proposed spreading along the draining lymphatic vessels, the biology of ITM formation is not fully understood and there is currently no way to identify patients at risk of developing ITM. Commonly known risk factors of melanoma including tumor thickness and ulceration of the primary melanoma also apply to patients who develop ITM. Patients with ITM at the leg have a better prognosis than those with ITM at other body sites (6). This may be related to the long lymphatic pathways of the lower extremity defined by a Japanese research group (7) in cadavers. Therefore, the area along defined lymphatic pathways where ITM occur would be predetermined based on the location of the primary melanoma.
In dermatology, different imaging modalities such as digital photography, dermoscopy, sonography, optical coherence tomography, and confocal microscopy are applied to document solitary skin lesions (8). Most of these modalities, currently predominantly applied in dermato-oncology (9) require specialized training, expensive equipment and are time consuming for the patients and medical staff (10). However, in order to image the total body, complex and standardized realignments of the camera or the patient are required. Since skin areas are distorted if they are not photographed at a vertical angle, they can only be assessed to a limited extent in conventional photography. Semiautomatic and image-processing 3D photography systems offer a patient- and user-friendly handling of total body photography (11), on which changes such as cutaneous melanoma metastases can be traced precisely using assisting image processing tools (12). These applications enable objective automated nevi count in patients with numerous pigmented lesions (13). Standardized total body 3D photography images provide a solid prerequisite for machine-based learning algorithms in dermatology (14) as demonstrated in detection and analysis of cherry angioma (15). 3D photography and customized image processing tools are applied for various dermatologic conditions: Area of vitiligo lesions was calculated accurately by manual multipoint polygon region annotation utilizing an area measurement application (16). Decrease in volume of infantile hemangioma during propranolol treatment was calculated using a portable 3D photography unit and software provided by the manufacturer (17). Wound characteristics such as area, perimeter, and volume were reliably determined by a prototype 3D-wound assessment monitor camera and for this purpose applied software (18). Facial asymmetry caused by morphea could be more properly rated by 3D stereophotogrammetry compared to 2D photography (19). Assessment of successful labial augmentation by filler injection was demonstrated using a 3D imaging system and a facial measurement and analysis tool (20). Thus, 3D images have so far been analyzed mainly by means of the software provided by the manufacturer of the camera systems. In the future, these high-resolution, standardized images, together with rapidly advancing machine learning in dermatology (21, 22), may offer great possibilities in diagnostics and management of skin diseases.
In the current work, we identified patients with ITM and melanoma of the lower extremities. To demonstrate the in vivo spread of tumor cells resulting in ITM along defined lymphatic pathways, standardized templates were created with annotated primary melanomas, ITM, and associated lymphatic pathways in a subset of 46 patients with available conventional and 3D photographic images.
Materials and methods
Patients
This monocentric, retrospective study analyzed 382 patients (female 263, male 119) with cutaneous melanoma of the lower extremity. The study was conducted in compliance with Good Clinical Practice (GCP) rules and the Declaration of Helsinki. Ethical review and approval were waived for this study due to its retrospective character and the analysis of standard patient data raised in the context of regular treatment visits. Patients signed an informed consent form for treatment at our center and for the photo documentation in the context of regular treatment. All patients underwent standard treatment and follow-up at the Department of Dermatology of the Uniklinikum Erlangen, Germany, between October 2000 and July 2020. We identified patients in our institutional database followed by individual file review from the electronic patient records in the clinical documentation system Soarian® (Cerner, Health Services, Berlin, Germany). A dermatologist diagnosed ITM during clinical skin examination either at the time of primary melanoma diagnosis or during regular oncologic follow-up. Data were obtained for each patient, including gender, age at diagnosis of primary melanoma, histologic subtype (superficial spreading melanoma, nodular melanoma, acral lentiginous melanoma, melanoma not otherwise specified, or melanoma with unknown primary), mean Breslow tumor thickness (mm), ulceration (present, absent, or unknown), and localization of primary melanoma (foot, lower leg, or thigh). Additionally, we analyzed the number of detectable ITM (1–5, 6–10, 11–20, more than 20 or unknown), mean time between primary diagnosis of melanoma and first ITM (in months), and availability of clinical images (no images, conventional photography, or 3D photography).
Image acquisition
Conventional photography
Images of patients were generated during follow-up at the Department of Dermatology, Uniklinikum Erlangen, Germany, by digital 2D single-lens reflex cameras (Canon EOS-ID Mark III or Canon EOS-ID Mark IV, Canon Germany GmbH, Krefeld). In the majority of the patients, primary melanoma as well as ITM had already been resected. Therefore, the location of prior melanoma lesions on clinical images by residual scars were identified and cross-referenced with the patients’ clinical file.
3D total body photography
Patients were imaged by a VECTRA 360WB medical photography platform system (VECTRA 360WB, Canfield Scientific, Parsippany, NJ, USA) available at our center since 2019 consisting of two concentrically orientated frameworks with a total of 92 high-resolution digital cameras that captured 92 standardized polarized images of the patient simultaneously. During image acquisition, undressed patients stood in a defined pose at a standardized position inside the imaging unit. The integrated Dermagraphix software (VECTRA-Software VAM Version 2.7.6, Canfield Scientific, Parsippany, NJ, USA) subsequently processed single images into a 3D body avatar, which covered up to 90% of the patient‘s body surface (11). Thus, a millimeter precise anatomical mapping of present lesions and scars was generated for each patient. The distance between primary melanoma and ITM was calculated via the skin surface distance-measuring tool of the Dermagraphix program.
Generation of a reference template of the leg
As a first step, a reference template of the leg was created in order to compare ITM spreading pathways of the lower extremity in our patients. Contours from defined angles (lateral, medial, ventral, and dorsal) of the reference leg were delineated from an original 3D image using GIMP (GNU image manipulation program, version 2.10.20). Additionally, we integrated bone benchmarks and segmented colored lymphatic drainage areas as described previously (7): posterolateral (red), posteromedial (yellow), anterolateral (green), and anteromedial (blue—including the toes).
Identification of melanoma lesions and transfer onto the leg template
Evaluable melanoma lesions or scars of primary melanoma and ITM were identified in 46 patients with melanoma on the lower leg or foot and cross-referenced with clinical data, previous photographic images, histologic results, and operation reports. In addition, we contacted patients in order to pinpoint the localization on conventional or 3D images. In annotated and standardized 3D images, these lesions were extracted by an affine transformation scheme consisting of 6 degrees of freedom (x, y, z, 3x rotation) and transformed back into our two-dimensional reference templates. In order to compensate individual anatomical difference between patients, images were standardized with fixed points of interest using bone benchmarks such as the patella, medial as well as lateral ankle, and toes via Paint-3D (Microsoft Corporation, Version 6.2004.20027.0). Conventional photographic images of melanoma lesions and scars on the leg of patients with ITM were retrieved from our photographic archive. We annotated melanoma lesions accordingly and transferred these onto the four reference templates using the software program Paint-3D (Paint-3D, Microsoft Cooperation, 6.2004.20027.0). For patients with melanoma of the right leg, annotated images were mirrored, so that a left leg was always depicted in the following analysis.
Statistical analysis
Our objective was to evaluate conventional or three-dimensional photography as a tool to analyze in-transit metastasis pattern of melanoma of the lower extremity. In addition, we assessed risk factors for the development of in-transit metastases in cutaneous melanoma. Specifically, data were analyzed using SPSS (SPSS, Version 24.0, IBM Deutschland GmbH, Ehingen). Clinical characteristics of patients with melanoma of the lower extremity with (n = 94) or without ITM (n = 288) were compared in order to determine specific risk factors to develop ITM. The chi-square test was applied for mutually exclusive categorical variables (i.e., gender of the patient, histological subtype, ulceration, and localization of the primary melanoma). Continuous variables (i.e., age of patient, Breslow tumor thickness, and the interval between primary melanoma and the occurrence of the first ITM) were analyzed using the Mann–Whitney U test between two independent patient groups (patients with melanoma on the leg without ITM versus patients with melanoma of the lower leg with ITM). Multivariate logistic regression analysis of categorical as well as continuous variables was applied to factors with a significance in the univariate analysis to assess the interaction of potentially predictive factors for the development of ITM of the lower extremity. In order to analyze differences between fully (n = 25), partially (n = 8), or non-matched (n = 15) metastatic patterns of ITM between the lower leg (n = 31) and the foot (n = 15) we applied the Mann–Whitney U test of ordinally scaled variables in the two independent sample groups (melanoma on the leg versus melanoma on the foot). In all tests, a p-value < 0.05 was considered significant.
Discussion
In-transit metastases develop along the way between the primary melanoma and the draining nodal basin and are a poor prognostic sign (6). Currently, the etiology of ITM is not fully understood. In the present study, conventional and 3D photography were used to chart and analyze the metastatic distribution of ITM of primary cutaneous melanoma of the lower extremity for the first time with regard to anatomically determined lymphatic drainage pathways (7) in a subset of 46 patients.
In general, about 4 to 10% of melanoma patients develop ITM (6, 23). The high rate of ITM (33%) in our clinical patient collective of 382 patients might be explained by the fact that it was limited to melanoma of the lower extremity, where ITM more often occur compared to other body sites (6). Additionally, patients with advanced or metastatic melanoma are more likely referred to a university skin cancer center, resulting in a high number of patients with ITM in our analysis.
Our clinical cohort of 94 patients with ITM complements reports that comprised 54 (24), 380 (25), and 505 patients (6) with satellite metastases and ITM. While in our study only patients with melanoma of the lower extremity were included, this subgroup makes 40–60% of the comparable collectives in the literature. In particular, the different biology and prognosis of acral lentiginous melanoma (26) represent a possible confounder. The median tumor thickness of 2.9 mm and ulceration status of 46% in our patient cohort were comparable with other studies reporting a median tumor thickness of 3 mm and the presence of ulceration in 28–39% of the patients (6, 24). Tumor subtypes and further localization on the thigh, lower leg, or foot could not be compared because they were not stated separately for the lower extremity (6) or were not reported at all (24, 25). With regard to patient characteristics, the male-to-female ratio ranged from 1:2 in our study to 2:1 in the literature (6). This different distribution may be influenced by differences in the country-specific risk of sun exposure and behavior between Europe and Australia (27). The median age of patients was 65–75 years and confirms that ITM occur at an older age, but could not be exactly compared due to different timing at data collection (primary melanoma diagnosis or start of systemic therapy). In our study, the median interval until occurrence of ITM was 19 months, which is in line with 17.9 months stated in the literature (6). The number of ITM per patient has not been reported in the literature. The gender distribution in our patients with melanoma on the lower extremity was independent of the occurrence of ITM. Our data confirm negative prognostic factors in melanoma (1, 28) in patients with ITM compared to patients without recurrence such as tumor thickness, ulceration status, and nodular melanoma.
In a subset of 46 patients with evaluable conventional and 3D photography, the localization of melanoma lesions was systematically correlated to anatomically determined lymphatic drainage pathways (7). Our results show that the localization of primary melanoma on the lower leg indicates areas at risk which should be examined more closely during follow-up. In contrast, a full match between ITM and lymphatic drainage was present in only 20% of the patients with melanoma of the foot. This discrepancy may be caused by altered lymphatic drainage pathways due to surgery of the toes and sole of the foot. The fact that 9 of our 46 analyzed patients developed ITM distally of the primary melanoma indicates a significant surgical alteration of the lymphatic vessels including the draining nodal basin by prior sentinel node biopsy or lymphadenectomy.
Interestingly, one patient with amputation of the three toes due to melanoma developed ITM along both the anteromedial and the anterolateral lymphatic bundles. This may further support postoperative lymphatic alteration but could also indicate a yet unstudied differential lymphatic drainage of the toes and the plantar region. It is of interest that in order to identify anatomic lymphatic pathways, contrast injections were exclusively applied to the outer and inner edges of the foot as well as the interdigital spaces in a previous study (7).
As discussed by Shinaoka et al. (7), the results of the anatomic/radiologic study may not be fully applicable to the in vivo situation (such as the development of ITM analyzed in this work) as the study was performed on cadavers with contrast injection into the foot followed by manual lymphatic drainage. In contrast, in vivo lymphatic fluid is transported by means of metabolically active muscle contraction in conduit and transport vessels.
To our knowledge, this is the first report to apply 3D photography to determine precise distances along the lymphatic pathways that melanoma cells must have traveled between primary melanoma and corresponding ITM. Since the distances between two lesions were calculated on a 3D leg model on the skin surface and the angles of the joints were taken into account, this almost exactly corresponds to the lymphatic pathways running in the upper corium. Although lymphatic pathways from the foot to the draining inguinal lymph node are anatomically longer than those from the lower leg, the distances between primary melanoma and associated ITM did not differ significantly in our cohort. Whether this is a biological, mechanical, or lymphogenic characteristic of melanoma of the foot, in which ITM develops with preferentially on the lower leg and thigh, cannot be answered by this study due to the insufficient number of patients examined by 3D photography (n = 22). In order to assess the significance for diagnostics and therapy, further investigations of ITM should also be performed including the trunk, the heck, and head.
Our study demonstrates for the first time the biological relevance of anatomically described lymphatic drainage pathways using ITM in melanoma of the lower extremity as a model. However, the limitations of our work must be considered before this knowledge can be applied for example to assess risk areas for ITM: (i) In our clinical collective we could only identify 94 patients with ITM of the leg in our center within a 20-year time frame, (ii) the analysis of ITM along lymphatic vessels was limited to melanoma of the lower leg and foot as only in the distal lower extremity lymphatic drainage patterns have been systematically investigated, (iii) evaluable photo documentation to analyze the lymphatic drainage patterns of melanoma of the lower leg and foot was only available in 46 patients, (iv) analysis of conventional, non-standardized photographs proved to be particularly difficult and possibly error-prone compared to 3D total body photography (in use in our department since 2019), (v) the localization of melanoma lesions was determined almost exclusively based on the presence of scars and comparison of the patients’ medical records. Especially for primary melanomas excised with a safety margin (sometimes toe amputation), the localization of melanoma lesions had to be extrapolated in the middle of the scar, (vi) changes in the physiologic lymphatic drainage pathways due to surgery around primary melanomas, ITM, and lymphadenectomy may have occurred in all patients, and (vii) due to methodological reasons, the anatomical study did not examine toe and plantar drainage, which remain not well defined. Therefore, the poor correlation observed between ITM and primary melanoma at the foot in our cohort suggests a critical re-evaluation with regard to lymphatic drainage pathways of the toes and the plantar region.
In summary, our results demonstrate the in vivo relevance of anatomic lymphatic pathways in ITM formation in 46 evaluable patients with melanoma of the leg. Together with clinical risk factors that we have confirmed, this knowledge may be used for a better risk assessment in the follow-up of melanoma patients and to develop a model for ITM in melanoma of the leg in the future.
Author contributions
ME and KM designed the study, performed data acquisition and analysis, and wrote the manuscript. ME, KM, and SM interpreted the data. CB, CV, MH, LH, EK, RK, and BS-T collected patient data and clinical information. CB, CV, MH, LH, EK, RK, SM, BS-T, VS, TS, and AW critically revised the manuscript. All authors contributed to the article and approved the submitted version.