Tumor thickness is of great importance in the management of cutaneous malignant melanoma because it is used to determine the prognosis and is a key factor in planning surgical margins and selecting candidates for sentinel node biopsy.1

Breslow depth is the standard measure of tumor thickness in primary cutaneous melanoma.2 A histologic measurement, Breslow depth requires a preoperative excisional skin biopsy and, in addition to being invasive, the procedure may delay definitive surgical treatment because it involves a 2-stage intervention.Cutaneous ultrasound is a noninvasive imaging technique for visualizing the interior layers of the skin. The use of this tool in the field of dermatology has been studied in recent years.2 Although the majority of early studies used 20-MHz probes capable of measuring lesions at a depth of 6 to 8mm, many others used lower-resolution probes (10-15MHz) because these were available in most hospitals. However, the current trend is to use higher-resolution probes (75-100MHz), which measure lesions confined to the epidermis and dermis more accurately. In cutaneous malignant melanoma, cutaneous ultrasound offers the possibility of estimating tumor thickness in vivo, thus avoiding excisional biopsy and speeding up surgical treatment planning.3–5 Results of recent studies of the validity and utility of cutaneous ultrasound in the management of cutaneous malignant melanoma have been inconsistent.6–12

The aim of this study was to evaluate cutaneous ultrasound as a tool for measuring tumor thickness in cutaneous malignant melanoma. The histologic variables that affect ultrasound assessment of tumor thickness are also analyzed.

This retrospective study evaluated the records of a consecutive series of patients with cutaneous malignant melanomas who underwent surgery between 2008 and 2010 in the dermatology department of the Hospital Costa del Sol. In all cases, members of the hospital's radiology department had performed a preoperative assessment of the lesion using a Toshiba Aplio XG 15-Mhz ultrasound probe to determine tumor thickness. Tumor thickness had been calculated automatically as the maximum vertical distance from the entry echo to the lowest hypoechoic signal, expressed in mm (Fig. 1). The histologic measure of thickness (Breslow depth) had been determined by a pathologist experienced in dermatopathology using an ocular micrometer to measure the maximum distance between the highest level of the stratum granulosum of the epidermis and the deepest point of the tumor, expressed in mm (Fig. 1). The histologic presence of ulceration, regression (absent, < 50%, > 50%), inflammatory infiltrate (absent, mild, moderate, severe), and associated nevi had also been recorded.

Figure 1.

Ultrasound and histologic assessment of tumor thickness.

(0.17MB).

The study was approved by the ethics committee of Hospital Costa del Sol.

We studied 79 melanomas from 79 patients (53% of them women) with a mean (SD) age of 56.8 (16.2) years; 42% were located on the trunk, 40% on the limbs, and 18% on the head and neck. Melanomas in situ accounted for 45% of the tumors (n=35) and the remaining 55% were invasive tumors (n=44; 27 melanomas ≤ 1mm, 17 melanomas > 1mm). The mean Breslow depth of the series was 0.8 (1.4) mm. The mean thickness as measured by ultrasound was 1.2 (1.7) mm. The histologic findings are summarized in Table 1.

The Pearson correlation coefficient between ultrasound assessment of thickness and Breslow depth was 0.678 (P<.001) (Fig. 2). The value of the intraclass correlation coefficient measuring agreement between the ultrasound assessment of thickness and Breslow depth was 0.78 (95% CI, 0.61-0.88).

Figure 2.

Scatter plot of histologic and ultrasound assessment of tumor thickness.

(0.11MB).

Bivariate analysis showed that ultrasound tended to overestimate tumor thickness in the presence of moderate to severe inflammatory infiltrate and nevi, though this tendency was not statistically significant (P>.05) (Table 2).

Classic sensitivity tests showed that the sensitivity of ultrasound for the diagnosis of thick melanomas (Breslow depth > 1mm) was 82%, specificity was 80%, and positive and negative predictive values were 54% and 94%, respectively (Table 3).

In this study we analyzed the accuracy of cutaneous ultrasound as a tool for measuring tumor thickness in a consecutive series of 79 primary melanomas. Of interest is the high proportion of minimally invasive melanomas in situ, a finding that we attribute to the systematic use of dermoscopy in the diagnosis of pigmented skin lesions and to the continuous application of primary prevention strategies in our population.

We found moderate correlation and agreement between ultrasound and histologic measures, whereas most earlier studies6–10 have reported good correlation. Tacke et al.,6 using a 20-MHz probe on a series of 259 melanomas, reported a correlation coefficient of 0.88 (lesions of 0.55 to 0.95mm were classified incorrectly by ultrasound in 34% of cases and lesions of 1.30 to 1.70mm were classified incorrectly in 50% of cases). Conversely, Serrone et al.,7 using a 20-MHz probe on a series of 261 lesions, reported a correlation coefficient of 0.95 in melanomas larger than 1.51mm, finding the lowest correlation in lesions smaller than 0.75mm. The results reported by Kaikaris et al.,8 using a 14-MHz probe, were similar (r=0.283 in melanomas of 1–2mm and r=0.869 in melanomas larger than 2mm). Many factors make comparison between studies difficult, including the use of diverse probes, differences between the series studied, the inclusion of benign lesions (mainly nevi),7 and the exclusion of in situ lesions9 in the belief that ultrasound does not visualize epidermal lesions.

Some histologic features of tumors may alter the ultrasound assessment of thickness.10 Hoffman et al.11 found that the cause of 80% of cases of overestimation was the presence of inflammatory infiltrate. In our series, cutaneous ultrasound tended to overestimate tumor thickness in comparison with the histologic measure in the presence of moderate to severe inflammatory infiltrate and associated nevi, though no statistical significance was found (P>.05). These histologic features act as artifacts and interfere with the ultrasound assessment of the tumor because cutaneous ultrasound does not differentiate between tumor cells and inflammatory or nevus cells.

The sensitivity, specificity, and positive and negative predictive values for the diagnosis of thick melanomas (>1mm) were 82%, 80%, 54%, and 94%, respectively. These results suggest that cutaneous ultrasound would properly stage the tumor and allow a single surgical intervention to be planned in approximately 80% of melanomas. However, it would not provide an accurate measurement on which to base a therapeutic decision in 20% of melanomas. Cutaneous ultrasound showed high discriminatory capacity for thin melanomas but was less successful at discriminating thick melanomas. Some authors have developed diagnostic algorithms that increase the diagnostic accuracy for thick melanomas by combining clinical, dermoscopic, and ultrasound findings, reaching a sensitivity of 97% and a specificity of 100%.12

Almost half the melanomas assessed by ultrasound as thick were in fact thin (< 1mm). This finding is explained by the frequent association of inflammatory infiltrate and nevi with the melanomas in our series. This tendency towards overestimation can be remedied by using higher-frequency probes, which have a higher resolution.13

The results of one study, in line with our findings, showed a particularly good correspondence between ultrasound and histologic measurements using a 30-MHz probe when the lesions were smaller than 1mm in a series of 68 melanomas and an even better correspondence in lesions on the sole (r=0.945).14 In contrast, Machet et al.,15 using a 20-MHz probe, found lesions on the sole to be difficult to measure because of the thickness of the stratum corneum.

In the ultrasound estimation of tumor thickness our results contrast with those published in a recent prospective study of 54 cutaneous malignant melanomas, which reported sensitivity, specificity, and positive and negative predictive values in the diagnosis of thick melanomas (> 1mm) of 86%, 97%, 95%, and 91%, respectively.16 The diagnostic accuracy of cutaneous ultrasound in that study was striking despite the lower ultrasound resolution of the probe used by the authors (10 Mhz). These results could be explained by the high proportion of melanomas later found to be thick on histology, which would be more accurately measured by ultrasound assessment because the presence of other histologic factors would not alter their ultrasound assessment as thick melanomas.

The present study had several limitations. The small size of our series made it difficult to interpret the results with regard to the influence of histologic findings on the ultrasound assessment of thickness, so it was not possible to establish causal relationships between histologic and ultrasound findings. Interpersonal variability may interfere in the ultrasound assessment of tumor thickness, though the use of standardized procedure minimizes this bias.

Technological advances in ultrasound, such as the use of contrast agents,5 and the incorporation of new in vivo microscopy techniques, including dermoscopy17,18 and confocal microscopy,19 are encouraging and their combined use will improve the preoperative diagnosis of melanoma in the near future.

In conclusion, cutaneous ultrasound is a tool of great interest in the preoperative study of cutaneous malignant melanoma. It identifies thin melanomas fairly accurately and can be useful for planning appropriate surgical margins. Further studies are necessary to confirm whether certain histologic characteristics of the tumor, such as the presence of moderate to severe inflammatory infiltrate and associated nevi, can lead to an overestimation of thickness by ultrasound, limiting the clinical utility of this imaging technique in the management of cutaneous malignant melanoma.