Multispectral photoacoustic imaging for the detection of subclinical melanoma

Background Photoacoustic imaging (PAI) is a noninvasive technique for melanin detection within melanoma metastases. While ex vivo and early studies suggest promising clinical application, there are no standardized parameters for defining presence of melanoma metastases. Methods Following flank/hindlimb melanoma induction in BRaf-PTEN transgenic mice, bilateral inguinal lymph nodes (LN) were imaged in vivo at 4 to 8 weeks using PAI. Fourteen diagnostic parameters for in vivo detection of LN metastases were compared using the receiver operating characteristic and area-under-the-curve (AUC). Limits of detectability were assessed in ex vivo and in vitro phantom studies. Results Forty-nine LNs were imaged in 25 mice. Among metastatic LNs, tumor size ranged from scattered cells to 2.8 mm. The strongest predictor of LN metastasis was the ratio of peak 10% PA melanin signal in the LN compared with adjacent soft tissue (median 4.22 for positive LNs vs 1.07 for negative LNs, P < 0.0001). The AUC was 0.95 (95% CI, 0.90-1.00). In phantom studies, B16 tumor cells were detectable at a concentration of 10 to 25 cells/mu L and at a tissue depth of 2.5-3 cm. Conclusions We identified a simple, objective diagnostic parameter for identifying melanoma LN metastases in vivo. These findings may help inform the design of future clinical trials.