3.1.1.

Animal studies

Oh et al.¹¹⁶ used dual-wavelength AR-PAM to detect melanoma in mice inoculated with highly invasive B16 skin melanoma cells. Due to the difference in peak optical absorption between melanin and hemoglobin, a near-infrared (NIR) light (764 nm) and a visible light (584 nm) were used to visualize the melanoma [see Fig. 4(a)(i)] and surrounding vascular structures [see Fig. 4(a)(ii)]. Figures 4(a)(iii) and 4(a)(iv) illustrate the B-scan images across the red dotted line in Figs. 4(a)(i) and 4(a)(ii). The maximum thickness of the melanoma was found to be 0.5 mm. Zhou et al.^187,188 conducted several experiments to study PAI in nude mice using a similar melanoma model. They used a handheld AR-PAM to measure the depth of melanoma.¹⁸⁸ Their PAM utilized an “annular-shaped” light illumination method, in which light (with 8 mm inner diameter and 20 mm outer diameter) bypasses the center of the tumor and instead is delivered in the direction normal to the surface.¹⁸⁸ The melanoma depth (3.66 mm) as measured by their system [see Fig. 4(b)] corresponded well to the actual, postexcisional thickness (3.75 mm).¹⁸⁸ In another experiment utilizing a LA-PAT system, tumor depth and volume were measured and revealed an increase in tumor depth and volume from day 3 [Fig. 4(c)(i)] to day 6 [Fig. 4(c)(ii)] after injection with the B16 cells.¹⁸⁷ The depth increased from 1.32 to 2.77 mm and volume increased from 22.365 to $71.931{\mathrm{mm}}^3$. Moreover, the system was able to measure the rate of growth for both depth and volume of tumor. Recently, Wang et al.¹⁸⁴ built a hybrid PA/US system with a sound-light coaxial/confocal design by punching a 2 mm diameter hole in the center of the transducer to deliver the laser light. Melanomas in mice were imaged in vivo at day 7 [see Fig. 4(d)Ii)] and day 30 [see Fig. 4(d)(ii)] after B16 cell injection. A clear growth in tumor size and depth is observed as shown in B-scan images of melanoma at day 7 [see Fig. 4(d)(iii)] and day 30 [see Fig. 4(d)(iv)]. Moreover, PAI visualized microvasculature around the tumor. Another experiment with mice was conducted by Wang et al.²⁵¹ using a dual-wavelength AR-PAM with visible and NIR light combined with US to image sub-CM. The fused images from the two wavelengths enrich the imaging information and allow more accurate detection of the melanoma, differentiating it from normal tissue. With the 3D distribution the boundary detection of the melanoma is easier and accurate and could be further enhanced by the US structural information helping the identification of the tissue boundaries and precisely locating the sub-CM [see Figs. 4(e)(i)–4(e)(iii)].

Fig. 4

Melanoma detection and depth measurement in live animal models using different PAI configurations. (a) PA images of melanoma and vascular distribution in nude mouse skin. (i), (ii) En face PA images for the NIR light source ($\lambda =764\mathrm{nm}$) and visible light source ($\lambda =584\mathrm{nm}$) respectively: 1, melanoma; 2, vessels perpendicular to imaging plane; 3, vessels horizontal to imaging plane; 4, skin. (iii), (iv) B-Scan PA images along the red line in panels (i) and (ii). Reproduced from Ref. 116. (b) PA image of the melanoma showing both the top and bottom boundaries in nude mice in vivo. The red dots outline the melanoma. Reproduced from Ref. 188. (c) LA-PAT images of melanoma acquired in nude mice on day 3 (i) and day 6 (ii) after tumor implantation. Reproduced from Ref. 187. (d) PA maximum amplitude projection (MAP) images of melanoma in mice at (i) day 7, (ii) day 30 and PA B-scan images on (iii) day 7 and (iv) day 30. Reproduced from Ref. 184. (e) Fused 3D visible light and NIR PA MAP images in mice with melanoma. (i) visible light and US, (ii) NIR light and US, and (iii) visible and NIR light. Reproduced from Ref. 251.

3.1.2.

Human studies

In a pilot study, Zhou et al.¹³⁵ analyzed 10 melanomas in 7 patients using a LA-PAT system [see Fig. 5(a)] immediately preceding excisional biopsy. Melanomas at depths from 0.2 to 6 mm were visualized, but deeper tumors were beyond the detection limit of the LA-PAT system. Figure 5(a) shows the resultant PA images [see Figs. 5(a)(i) and 5(a)(iv)], photographs [see Fig. 5(a)(ii)–5(a)(v)], and histological images [see Figs. 5(a)(iii) and 5(a)(vi)] of two patients, one with CM metastases on the left lower extremity and the other with primary acral lentiginous melanoma on the right foot. Histological images acquired from both patients showed that the detected melanoma depth was consistent with actual Breslow depth, showing the promising capability of the system for detection. Kim et al.¹⁷⁷ developed an integrated PA and US imaging (PAUSI) system by combining a clinical US machine and a multispectral portable laser. The imaging system was utilized to image a patient’s melanoma ex vivo after excision [see Fig. 5(b)(i)]. The amplitude of the PA signal corresponds to the amount of melanocytes in the local area. The PA signals in melanoma, indicated by the white triangles [see Fig. 5(b)(ii)], were predominantly generated by optical wavelengths of 800 and 1064 nm. In contrast, the PA signals from the marking pen regions indicated by the white arrows were predominantly generated from the 680-nm laser. Moreover, PA imaging indicated areas of melanoma that are not visible to the surgeon [yellow triangles in Fig. 5(b)(ii)]. In addition, the measured thickness of the melanoma region ($420\pm 320\mu \mathrm{m}$) matched well with the histopathological results. A larger study on PAI was conducted by Breathnach et al.¹¹⁵ using LA-PAT and spectral unmixing, to pre-operatively image 32 pigmented lesions suspicious for melanoma in 27 patients. With spectral unmixing, they separated the absorption signature of melanin-containing cells and mapped the spatial distribution of it based on this signature, because the absorption spectrum of melanin varies within the NIR region. The lesion depths measured by PA were highly correlated with histopathologic measurements, with a correlation coefficient of 0.98 for benign lesions and 0.99 for melanomas. Using the same PA probe, the lesion architecture, adnexal depth (depth of lesion extension into the skin appendages), and various skin layers were also imaged, allowing for differentiation of superficial from invasive lesions based on their dermal-epidermal junction penetration, i.e., lesion penetrated through the dermal-epidermal junction [see Fig. 5(c)]. According to the authors, due to the tissue sample dehydration and loss of skin tension in vivo, PAI overestimated lesion depth as compared with histopathology. Recently, Park et al.¹³³ utilized a 3D multispectral PAT system to noninvasively measure depth and outline the boundary of melanomas for optimal surgical margin selection. Six melanoma patients were examined. They imaged melanoma of various forms, sizes (1.3 to 30 mm for lateral diameter and 0.6 to 9.1 mm for depth), and locations (sole, chest, thigh, heel, and palm) using their multispectral PA/US system [see Fig. 5(d)]. For five of the six case studies, melanoma depth was measured using multispectral analysis and confirmed a high correlation against histopathologic results with a mean absolute error of 0.36 mm. In a signal-based study, Swearingen et al.²⁵² investigated if label-free MPAI could distinguish vascular from pigmented (melanotic) lesions in 15 human patients. Excitation lights at 422 and 530 nm were used. At 422 nm, melanotic lesion showed a lower PA signal [see Fig. 5(e)(i)] compared with vascular lesion [see Fig. 5(e)(ii)]. Similarly, at 532 nm, melanotic lesion showed higher PA signal [see Fig. 5(e)(iii)] compared with vascular lesion [see Fig. 4(e)(iv)]. The experiment proved the ability of MPAI to distinguish between vascular and pigmented lesions. About 15 lesions were biopsied after imaging, revealing 8 vascular and 7 pigmented lesions. Data analysis was carried out via two statistical methods, the classical method (standard multivariate analysis classification techniques) and a Bayesian-model-based approach. The classical method attained a perfect lesion diagnosis rate, whereas the Bayesian approach had a 20% error rate.

Fig. 5

Melanoma detection and depth measurement in human studies using different PAI configurations. (a) PAT of melanoma of two patients: (i) melanoma image acquired with a PA depth of 1.9 mm ($\mathrm{cPA}\text{depth}=1.67\mathrm{mm}$); (ii) cutaneous melanoma metastasis in a patient lower leg; (iii) histology of the excised melanoma, showing actual Breslow depth of 1.67 mm; (iv) PAT melanoma image of the acral lentiginous melanoma with PA depth of 0.70 mm ($\mathrm{cPA}\text{depth}=0.62\mathrm{mm}$); (v) acral lentiginous melanoma (data not shown: $\mathrm{pBD}=0.48\mathrm{mm}$); and (vi) histology after complete excision, with actual BD of 0.78 mm. BD, Breslow depth; cPA, corrected photoacoustic depth; PAT, photoacoustic tomography; pBD, provisional Breslow depth. Reproduced from Ref. 135. (b) (i) Photograph and (ii) ex vivo PA image of excised melanoma tissue from a male patient. The melanoma regions are represented by dark red to bright-yellow color (white triangles), and the marking-pen regions are represented by dark green to bright-green color (white arrows). Yellow arrows indicate possible melanomas not found by histology. Reproduced with permission from Ref. 177. (c) PA image of in situ melanoma on upper left extremity on a patient using LA-PAT. Reproduced with permission from Ref. 115. (d) Measurement of PA depth of a nodular type of melanoma. (i) Photoacoustic MAP; (ii) photoacoustic MAP, overlaid with US image; and (iii)–(iv) photoacoustic unmixed and photoacoustic unmixed overlaid with US images, respectively. Blue arrows indicate the melanoma invasion and yellow arrows are the bottom boundary of the melanoma. MAP, maximum amplitude projection. Reproduced from Ref. 133. (e) Representative PA signals from human subjects with either (i), (iii) pigmented and (ii), (iv) vascular lesions at 422 and 530 nm, respectively. Reproduced with permission from Ref. 252.

Figures 4 and 5 and the related text describe results from a selection of PAI studies of melanoma detection and depth measurement. Table 1 includes specific characteristics of the PAI systems used in published studies on melanoma detection/depth measurement for both animal and human studies.

Table 1

Specific characteristics of PAI systems used in published studies on melanoma detection/depth measurement.

3.2.1.

Animal studies

In one study, Omar et al.¹⁸² utilized RSOM to visualize angiogenesis and tumor growth in melanomas of mice in vivo over several days. B16F10 melanoma cells were injected subcutaneously into the mammary fat pad of Hsd: Athymic $\mathrm{Nude}\text{-}\mathrm{Foxn}{1}^{\mathrm{nu}}$ mice, and tumor growth and angiogenesis were monitored using spherically focused US detectors with central frequencies of 50 and 100 MHz. The 50 MHz detector was superior in imaging larger structures such as larger, oblique blood vessels [see Fig. 6(a)(i)], whereas the 100 MHz detector provided better visualization of tumor [appearing as a black hole in Fig. 6(a)(ii)] and newly sprouting vessels [see Fig. 6(a)(ii)]. Using the 50 MHz detector, tumor growth was recorded at day 2 [see Fig. 6(b)(i)], day 4 [see Fig. 6(b)(ii)], day 7 [see Fig. 6(b)(iii)], and day 9 [see Fig. 6(b)(iv)]. Each subfigure in Fig. 6(b) includes an inset taken from the proximity of the tumor. The growth of the tumor from day 2 to day 9 was illustrated by the growth of the black nonvascularized spot, indicated by a thick arrow in all the subfigures in Fig. 6(b). Moreover, in each inset, it is observed that upon interaction with the tumor, the two big vessels (denoted by a thinner arrow) start re-arranging; at the same time, smaller vessels start growing in that region, clearly representing tumor angiogenesis at a microvasculature level. Zhao et al.¹⁸⁵ utilized OR-PAM at two wavelengths (570 and 1064 nm) to study melanoma tumor angiogenesis. B16 melanoma cells were injected subcutaneously into the mouse ear and tumors were imaged on day 9 [see Fig. 6(c)(i)], day 13 [see Fig. 6(c)(ii)], day 14 [see Fig. 6(c)(iii)], and day 15 [see Fig. 6(c)(iv)] after tumor inoculation in two mice. In the earlier day postinoculation, small diameter ($<25$ to $30\mu \mathrm{m}$) vessels were most prevalent. On the following 2 days, the number of large diameter vessels (50 to $95\mu \mathrm{m}$) increased while the proportion of small diameter vessels decreased. On day 15, vessels with diameter $>100\mu \mathrm{m}$ were visualized. Moreover, the vessel density, vessel tortuosity, and fractional dimension (quantitative parameters they used to assess tumor growth) also showed an overall upward trend from day 9 to day 15. Thus, the authors concluded that with the growth of melanoma, the vascular networks become stronger and complex, which was consistent with visual results in Fig. 6(c). Another study by Zhou et al.¹⁸⁶ used a combined all optical PA microscopy system and RCM system to study tumor growth and angiogenesis in mice who received subcutaneous injections of B16 melanoma cells in the ear. Their PAM showed irregular and linear vascular patterns, likely representing neovascularization of the dermis. The RCM enface image illustrated widespread pagetoid cells with cytologic atypia and nucleated cells within the dermal papilla. The PAM provided significant contrast and penetration depth, based on optical absorption properties, and visualized vascularity and pigmentation, whereas RCM illustrated cytological features. On the other hand, Xu et al.¹³⁴ developed an integrated OR/AR-PAM system for multiscale imaging capability with high-speed wide-field imaging based on a polygon scanner. The polygon has six aluminum-coated surfaces that reflect the light and the acoustic beams: in this way six repeated cross-sectional scans can be obtained in one rotation of the motor, increasing image acquisition speed. With the two modalities and using two different wavelengths (532 and 1064 nm) a sub-CM is detected and separated from the surrounding microvasculature [see Fig. 6(e)].

3.2.2.

Human studies

Recently, angiogenesis of human melanomas has been imaged by He et al.¹⁰⁹ using a single-breath-hold faster RSOM system to visualize the microvasculature of pigmented melanocytic lesion [see Fig. 6(d)]. The images from three different regions: scan 1 inside the lesion [see Figs. 6(d)(i), 6(d)(iv), and 6(d)(vii)], scan 2 in lesion boundary [see Figs. 6(d)(ii), 6(d)(v), and 6(d)(viii)], and scan 3 outside lesion [see Figs. 6d(iii), 6(d)(vi), and 6(d)(ix)], showed a difference on the epidermis and dermis vasculature patterns inside and outside the lesion. From the scans of 10 dysplastic nevi and 10 melanomas, a quantification of the vasculature features was performed to identify biomarkers for melanoma detection. A total of six biomarkers were calculated: the total blood volume, vessel density, average vessel length, tortuosity, fractal number, and lacunarity. It was found that those markers showed differences between malignant and benign lesions, supporting the possibility to use this system to improve melanoma diagnosis.

3.3.1.

Animal studies

PAI has been studied for the imaging of lymph nodes (Table 3) to identify metastases both ex vivo (postsurgical resection) and in vivo. McCormack et al.¹⁷⁶ used a three-single-element-transducer based PAT prior to and after injecting a human melanoma cell line (HS 936) into ex vivo lymph nodes from a healthy canine and pig. The PAT system they used consisted of a $600\mu \mathrm{m}$ diameter optical fiber connected to a tunable laser and an acoustic sensor made from polyvinylidene fluoride film. The 532 nm wavelength light was illuminated from the top of the lymph node, and the PA signals were detected from the bottom [see Fig. 7(a)]. The control lymph nodes (with no injected melanoma cells) showed no PA response [see Fig. 7(a)(i)], whereas the melanoma cells in the excised lymph nodes generated PA signals as shown with the arrow in Fig. 7(a)(ii). A pig lymph node with only 500 injected melanoma cells and a 100 to $200\mu \mathrm{m}$ diameter lymph node produced a PA response. Neuschmelting et al.¹⁸⁰ utilized a multispectral (wavelengths 700 to 860 sampled) PAT system with a cylindrically focused transducer array (iThera Medical) and compared it with fluorodexyglucose (FDG) positron emission tomography (PET)/CT for imaging B16F10-derived melanoma micrometastases and macrometastases to lymph nodes and in-transit metastases (metastases moving from the primary tumor to the nearby nodal basin) in mice in vivo. The PAT system detected lymph node micrometastases in the cortical region of lymph nodes (after 2 weeks’ tumor cell inoculation) that were too small for FDG PET/CT to detect. It also delineated in-transit metastases, which were observed as bright clusters between the primary melanoma site and the nodal basin. Both PAT and PET/CT could detect macrometastases [see Fig. 7(b)], but only the PAT system unmixed signals to enable detection of micrometastatic infiltration of melanoma in the cortex of popliteal nodes (white dashed circles). This multispectral PAT system distinguished melanoma lymph node metastases from other neoplastic and nonneoplastic lymphadenopathies (due to melanin’s contrast) but FDG PET/CT could not (due to nonspecific FDG uptake and relatively low resolution). Most recently, Sinnamon et al. utilized a Vevo LAZR PAT system to visualize inguinal lymph node metastases in vivo at 4 and 8 weeks after inoculation with B16 melanoma cells in the flank region in BRaf-PTEN transgenic mice.^174,183 In total, 49 lymph nodes were imaged in 25 mice, with metastatic cells present in 17 lymph nodes (35%), with histopathological confirmation. Thus the system was able to image in vivo melanin PA signals of a lymph node containing melanoma metastasis at 60 days after the induction of the tumor [see Fig. 7(c)(i)]. The negative control mouse without tumor induction is shown in Fig. 7(c)(ii). The PA melanin signal within the positive lymph nodes was significantly higher than in the negative control mouse, with no difference in PA signals between the adjacent soft tissue of positive and negative lymph nodes. The strongest predictor of melanoma metastasis was the ratio of lymph node to soft tissue PA melanin peak signal.

Table 3

Summary of ex vivo and in vivo human and animal PAI studies for detection of melanoma metastasis to lymph nodes.

Fig. 7

Animal studies of lymph node metastases analyzed by PAI. (a) Photoacoustic response from lymph nodes (i) without melanoma, the detector shows no signal with a noise floor of $100\mu \mathrm{V}$, and (ii) with melanoma, showing a photoacoustic wave at $5\mu \mathrm{s}$. Reproduced with permission from Ref. 176. (b) Comparative results between multispectral PAT and FDG-PET/CT systems for detection of macrometastasis in right popliteal lymph node in mice. The red circle encloses the melanoma macrometastasis region. White circle enclosed the contralateral popliteal healthy control node. Multispectral images were unmixed for deoxyhemoglobin (blue), oxyhemoglobin (red) and melanin (yellow). Fb, femur; Tv, tail vessel; Rc, rectum; Ut, urethra. Reproduced with permission from Ref. 180. (c) In vivo images of melanin-specific photoacoustic signal with grayscale underlay of a lymph node that (i) shows metastasis at 60 days post tumor induction and (ii) negative control mouse that did not undergo tumor induction. MPAI, multispectral photoacoustic imaging; FDG-PET/CT, fluorodexyglucose PET/CT. Reproduced with permission from Ref. 183.

3.3.2.

Human studies

Many different PAI modalities have been utilized for ex vivo studies of metastases in melanoma SLNs, including LA-PAT and arc-shaped PAT.^117,169,174 PAT has also been used for in vivo studies.¹¹⁷ The first human study on melanoma lymph node metastasis was conducted by Grootendorst et al.¹⁶⁹ using a curvilinear array. Patients with proven metastatic disease undergoing inguinal or axillary lymphadenectomy were enrolled, and 1 or 2 lymph nodes were randomly selected for ex vivo multispectral analysis. The PA laser was illuminated from the top of the sample and the US detector array was rotated 360 deg around the sample. A total of six lymph nodes were imaged, and PAT revealed that three lymph nodes were metastatic and three were benign. All three malignant nodes (LN1, LN2, and LN3) displayed increased PA signals (from melanoma cells), whereas all three benign nodes had a substantially weaker signal (signal likely from hemoglobin and possibly other chromophores) when imaged at wavelengths from 720 to 800 nm [see Fig. 8(a)]. The results were confirmed through histopathology. Langhout et al.¹⁷⁴ used a LA-PAT system and imaged 12 lymph nodes. Histopathology revealed that three nodes were metastatic and nine were benign. Again, melanoma-positive nodes [see Fig. 8(b)(i)] displayed different PA signals (depth image) than benign nodes [see Fig. 8(b)(ii)] due to the difference in melanin distribution. Additionally, total volume imaging to the depth of 2 cm in benign lymph nodes (absent melanin deposits) allowed for computation of the entire nodal volume. The largest human study on PAI of melanoma metastasis to lymph nodes was carried out by Stoffels et al.¹¹⁷ using a multispectral arc-shaped PAT. They analyzed 506 SLNs from 214 patients with a Breslow depth of at least 1 mm: 148 SLNs (from 65 patients) were analyzed by multispectral PAT ex vivo and histology, whereas the other 358 SLNs (from 149 patients) were analyzed by the conventional European Organization for Research and Treatment of Cancer (EORTC) melanoma group protocol. Their system detected metastases in 22.9% of excised SLNs compared with 14.2% by the EORTC Melanoma Group protocol. Ex vivo analysis by PAT showed 100% sensitivity and 62% specificity. Then, an in vivo experiment was carried out using PAT and indocyanine green (ICG) (a NIR fluorophore injected peri-tumorally) as a contrast agent to image 41 SLNs in 20 patients. PAT visualized ICG-marked SLNs to the depth of 5 cm (discerned by single photon emission computed tomography/CT) and with 100% concordance with the gold standard of SLN detection, 99 m Tc-nanocolloid–guided lymphoscintigraphy. In vivo PAT analysis revealed a sensitivity of 100% and specificity of 48.6%. With 100% sensitivity, both ex vivo and in vivo PAT identified noncancerous SLNs in 189 total lymph nodes without any false negatives but with a high rate of false positives. The quantification of melanin was performed with multispectral analysis and then correlated with the localization of metastatic cells [see Fig. 8(c)].

Fig. 8

Ex vivo and in vivo studies of human lymph node metastases analyzed by PAI. (a) PA signal strength of the selected areas within the lymph nodes (LN) at different illumination wavelengths. Reproduced with permission from Ref. 169. (b) Images of two human nodes. (i) PA image of a metastatic node, (ii) PA image of a benign node. Absence of PA signal deeper in the malignant node [as indicated by * in panel (b)] seems to be caused by the strong absorption by the melanin in the superficial area of the node. Reproduced with permission from Ref. 174. (c) (i) Lateral MIP and (ii) 3D rendering image of ex vivo optoacoustic image of a human lymph node from a melanoma patient. Grayscale represents hemoglobin background and color bar overlay shows the multispectral resolved signals for melanin. Reproduced with permission from Ref. 117.

3.4.1.

Animal studies

PAFC has demonstrated the ability to detect unlabeled (and labeled) CTCs in the mouse vasculature after injection of CTCs in the tail vein or from inoculation of primary melanoma tumors in the skin or ears.^{87,167,168,172,189,190} Overall, PAFC accurately detected CTCs from the background RBCs in the mouse vasculature, and PAFC signals increased as metastasis increased.^{139,167,168,171,172,189,190} Deán-Ben et al.¹⁹¹ employed spherical array PAT for real-time visualization of passage and trapping of individual B16 melanoma cells in the whole mouse brain. Imaging was performed with the laser wavelength and pulse repetition rate set to 700 nm and 50 Hz, respectively. About 100 frames of PA images were acquired before injection of B16 melanoma cells and averaged to determine a baseline. The system could identify and track B16 melanoma cells after injection by taking the difference between before and after images.¹⁹¹

3.4.2.

Human studies

Galanzha et al.¹⁶⁶ studied CTCs in healthy patients and patients with melanoma using PAFC in vivo. PAFC data from healthy patients were used to calculate false positives and to study PA artifacts.¹⁶⁶ PAFC accurately identified unlabeled CTCs in 15/16 melanoma patients.¹⁶⁶ The system uses MPAI with one channel for detection of blood and a second channel for melanin. The channels are shown individually and fused together in Fig. 9(a)(i), and fused images showing movement of melanoma CTCs in an artery [Fig. 9(a)(ii)] and a vein [Fig. 9(a)(iii)] are also presented. This system has relatively poor lateral resolution (Table 4), which the authors explain is caused by light beam blurring.¹⁶⁶ Recently, PAT systems have also been used to detect CTCs. Hai et al.^171,192 employed a LA-PAT system to detect melanoma CTCs in patients in vivo. Based on the optical absorption coefficient ratio, an excitation wavelength of 680 nm was chosen to maximize the contrast between melanoma CTCs and blood and achieve the highest detection sensitivity. Contrast-to-noise ratio (CNR) was used to quantify melanoma CTCs from the background tissue (RBCs). From the imaging session of the forearm of a positive patient with stage IV metastatic melanoma, a CTC was detected with a CNR of 9.4 [see Fig. 9(b)]. The single cells were captured at five frames and the flow speed was estimated to be $9.6\mathrm{mm}/\mathrm{s}$. They imagined 16 stage III and IV melanoma patients and successfully detected suspected melanoma CTCs in three patients. In fact, two of the three had disease progression, but four of those found CTC-negative also had disease progression. The lower rate of CTC detection by LA-PAT compared with PAFC could be attributed to the use of melanin as the single marker to identify and detect CTCs, and the fact that the patients were measured at only one clinical time point instead of on multiple repeat visits.^171,192 Also, in the previous study, optical clearing techniques, including microdermabrasion and glycerol sonophoresis, increased PA signals by two- to threefold, due to the reduced light scattering in superficial skin layers.¹⁶⁶ PAT has tremendous clinical potential in imaging CTCs and disease monitoring in melanoma, given the fact that PAT can image very deep in the tissue with very high optical contrast. Moreover, utilization of contrast agents for molecular PAI can further increase the sensitivity of PAT to imaging cancer cells that do not express melanin.

Fig. 9

Imaging melanoma circulating tumor cells in vivo. (a) PA flow cytometry: channels shown individually and fused together (i), fused images showing movement of melanoma CTCs in an artery over time (ii), and in a vein (iii). (b) LA-PAT system uses differential analysis of images taken over time to find moving CTC cells: (i) PA snapshots of the melanoma CTC in the patient. The yellow arrows indicate structures, including the skin, vessel boundaries, and subcutaneous fat layer. The red arrows highlight the melanoma CTC. (ii) Differential PA images showing only the melanoma CTC. (iii) Differential PA images superimposed on structural images, highlighting the melanoma CTCs. Reproduced with permission from Ref. 192.

3.6.1.

Animal studies

Kim et al.¹⁷³ studied PA skin imaging of gold nanocrystals (AuNCs) bio-conjugated to $[{\mathrm{Nle}}^4,\mathrm{D}\text{-}{\mathrm{Phe}}^7]\text{-}\alpha$-melanocyte-stimulating hormone ($[{\mathrm{Nle}}^4,\mathrm{D}\text{-}{\mathrm{Phe}}^7]\text{-}\alpha \text{-}\mathrm{MSH}$) and compared them to poly(ethylene glycol) (PEG)-AuNCs. Melanoma cells strongly overexpress α-MSH receptors. In their study, mice were inoculated with B16 melanoma cells and either one or the other contrast agent was injected via tail vein. In vitro studies showed that melanoma cellular uptake of $[{\mathrm{Nle}}^4,\mathrm{D}\text{-}{\mathrm{Phe}}^7]\text{-}\alpha \text{-}\mathrm{MSH}\text{-}\mathrm{AuNCs}$ was $\sim 3.5$ times greater than PEG-AuNCs uptake after 6- and 24-h incubation periods. In vivo experiments involved two groups of mice: one group ($n=4$) received $[{\mathrm{Nle}}^4,\mathrm{D}\text{-}{\mathrm{Phe}}^7]\text{-}\alpha \text{-}\mathrm{MSH}\text{-}\mathrm{AuNCs}$ and the other group ($n=4$) received PEG-AuNCs, both via tail vein [see Fig. 10(a)]. Six hours after injection, the PA signals of $[{\mathrm{Nle}}^4,\mathrm{D}\text{-}{\mathrm{Phe}}^7]\text{-}\alpha \text{-}\mathrm{MSH}$ AuNCs increased 38%, whereas the PEG-AuNCs increased only 13% [see Fig. 10(a)(iv)]. The number of AuNCs was quantified in the excised tumors using inductively coupled plasma mass spectrometry. The mean number of $[{\mathrm{Nle}}^4,\mathrm{D}\text{-}{\mathrm{Phe}}^7]\text{-}\alpha \text{-}\mathrm{MSH}$ AuNCs per gram of tumor was found to be 360% times greater than that of PEG-AuNCs, indicating the extremely high uptake of $[{\mathrm{Nle}}^4,\mathrm{D}\text{-}{\mathrm{Phe}}^7]\text{-}\alpha \text{-}\mathrm{MSH}$ by melanoma cells. More recently, Li et al.¹⁷⁵ evaluated a nanoparticle composed of AuNRs and liquid perfluorocarbon (PFH) and conjugated it to a monoclonal antibody to melanoma-associated antigens (MAGE-1 antibody) and used an LA-PAT system for imaging. In vitro experiments showed a high number of MAGE-Au-PFH-NPs concentrated in the plasma membrane and cytoplasm of melanoma cells, compared with Au-PFH-NPs (control), which only showed a weak binding around tumor cells. Laser irradiation of nanoparticles with PFH results in a phase change (liquid PFH to gas microbubbles), increasing the acoustic impedance of surrounding tissues and subsequently causing a signal enhancement. In vivo PAI in tumor-bearing mice revealed a significantly higher PA signal intensity 2 h after MAGE-Au-PFH-NPs injection into the tail vein compared to Au-PFH-NPs; higher conjugate concentrations produced greater PA signals [see Fig. 10(b)].

Fig. 10

In vivo animal studies involving exogenous PAI contrast agents for melanoma detection. (a) In vivo MAP images of B16 melanomas using $[\mathrm{Nle}4,\mathrm{D}\text{-}\mathrm{Phe}7]\text{-}\alpha \text{-}\mathrm{MSH}\text{-}$ and PEG-AuNCs. (i), (ii) Photographs of tumor in mice before injection. (iii), (v), and (vii) Time course PA images after injection of $[\mathrm{Nle}4,D\text{-}\mathrm{Phe}7]\text{-}\alpha \text{-}\mathrm{MSH}\text{-}\mathrm{AuNCs}$. (iv) (vi), and (viii) Time course PA images after injection of PEG-AuNCs. Reproduced with permission from Ref. 173. (b) PA images of B16 melanomas in mice at different time points after the injection of (i) MAGE-Au-PFH-NPs and (ii) AU-PFH-NPs. Reproduced with permission from Ref. 175. (c) PA imaging of melanoma in mice (i) before and (ii) after the injection of ID-HCuSNP@B16F10 NPs. Reproduced from Ref. 195. MAP, maximum amplitude projection.

Wu et al.¹⁹⁵ developed cell membrane-camouflaging hollow copper sulfide nanoparticles (ID-HCuSNP) by coating the membrane of melanoma cells with doxorubicin and ICG-loaded hollow copper sulfide NPs to enhance their targeting ability. They injected a mouse model with ID-HCuSNP and after 4 h, and they observed a strong local PA signal in the tumor area due to the accumulated NPs.¹⁹⁵ This showed cancer cell membrane-camouflaged NPs having an excellent self-recognition ability to the aimed tumor cells in vivo [see Fig. 10(c)]. Galanzha et al.¹⁶⁵ used PAT with mapping techniques (multicolor PA lymph flow cytometry, PA lymphography, and absorption image cytometry) to examine metastasis to the prenodal lymph vessels and SLNs in mice ex vivo and in vivo. Specifically, after a visible primary melanoma tumor had developed in the mouse ear, they injected magnetic nanoparticles (MNPs) that provided stronger PA signals from lymphatics and SLNs at 639 nm. However, similar NIR absorption spectra of MNPs and melanin made it difficult for spectral identification of melanoma metastasis in the presence of MNPs. To resolve this problem, they used Evans Blue (EB) dye, which absorbs poorly at 639 nm only, while the presence of melanoma cells in lymphatics and in a node was identified through the appearance of PA signals at 850 nm. Because absorption of melanin also occurred at 639 nm above the background from EB dye, they used specific ratio of PA signals at 639 and 850 nm for additional identification of melanoma cells. They injected EB dye near melanomas derived from B16F10 cells and imaged the region in week 1 and 2 postinoculation. The number of metastatic melanoma cells in transit increased from $0.26\text{cells}/\min$ in week 1 postinoculation to $2.13\text{cells}/\min$ in week 2 postinoculation. The percentage of PA signal covering the examined SLN area increased from 6% in week 1 postinoculation to 39% in week 2 postinoculation. In the histology results, lymph nodes showed no metastases after 1-week inoculation and single metastases after 2 weeks inoculation. Thus the PA system was able to detect early micrometastases that could not be visualized in histology, illustrating its ability to detect even single metastatic cells (at week 1 postinoculation).

Despite the number of reports on the successful demonstration of Au nanomaterials for cancer theranostics,^275,276 accumulations in the liver and spleen due to resident macrophages that form the mononuclear phagocyte system are still an issue.^277,278 Armanetti et al.¹⁹⁶ utilized endothelial colony forming cells (ECFCs) to carry AuNPs and explored the antitumor effects and the tumor-homing efficiency following single intravenous injection into tumor-bearing mice. They assessed AuNPs biodistribution in freshly excised mice organs at different time points post administration by exploiting the PA properties of AuNP-enriched ECFCs. They demonstrated in vitro that AuNP-loaded ECFCs are able to generate higher PA signals than AuNPs alone and also display spectral fingerprints that enable a reliable detection of labeled cells following intravenous injection.¹⁹⁶

Toxicity is always a concern when developing exogenous contrast agents. Factors for assessing biocompatibility include inertness, metabolism, and effective clearance rates. Material composition, surface modifications, shape, and size must all be designed carefully to maximize biocompatibility. The growing field of PAI for cancer diagnosis and treatment planning is accelerating research in biocompatible exogenous contrast agents.^279,280