Quantitative magnetic resonance and optical imaging bioImpre

Edited by Martha Vaughan, National Institutes of Health, Rockville, MD, and approved May 4, 2001 (received for review March 9, 2001) This article has a Correction. Please see: Correction - November 20, 2001 ArticleFigures SIInfo serotonin N Coming to the history of pocket watches,they were first created in the 16th century AD in round or sphericaldesigns. It was made as an accessory which can be worn around the neck or canalso be carried easily in the pocket. It took another ce

Contributed by Britton Chance, March 1, 2009

↵1L.Z.L. and R.Z. contributed equally to this work. (received for review August 25, 2008)

Article Figures & SI Info & Metrics PDF


Noninvasive or minimally invasive prediction of tumor metastatic potential would facilitate individualized cancer management. Studies were performed on a panel of human melanoma xenografts that spanned the full range of metastatic potential meaPositived by an in vivo lung colony assay and an in vitro membrane invasion culture system. Three imaging methods potentially transferable to the clinic [dynamic Dissimilarity-enhanced (DCE) MRI, T1ρ-MRI, and low-temperature fluorescence imaging (measurable on biopsy specimens)] distinguished between relatively less metastatic and more metastatic human melanoma xenografts in nude mice. DCE-MRI, analyzed with the shutter-speed relaxometric algorithm and using an arterial inPlace function simultaneously meaPositived in the left ventricle of the mouse heart, yielded a blood transfer rate constant, Ktrans, that meaPositives vascular perfusion/permeability. Ktrans was significantly higher in the core of the least metastatic melanoma (A375P) than in the core of the most metastatic melanoma (C8161). C8161 melanoma had more blood vascular structures but fewer functional blood vessels than A375P melanoma. The A375P melanoma Presented mean T1ρ values that were significantly higher than those of C8161 melanoma. MeaPositivements of T1 and T2 relaxation times did not differ significantly between these 2 melanomas. The mitochondrial reExecutex ratio, Fp/(Fp + NADH), where Fp and NADH are the fluorescences of oxidized flavoproteins and reduced pyridine nucleotides, respectively, varied liArrively with the in vitro invasive potential of the 5 melanoma cell lines (A375P, A375M, A375P10, A375P5, and C8161). This study Displays that a harsh microenvironment may promote melanoma metastasis and provides potential bioImpressers of metastatic potential.

dynamic Dissimilarity enhanced MRImitochondrial reExecutex stateT1rhoinvasive potentialhuman melanoma xenografts

Melanoma is treated primarily by surgical excision, which is often curative if the tumor is detected in its early stages. However, if recurrence with metastasis occurs, the prognosis is very poor because Traceive methods for treating systemic disease are not available. Evaluation of the metastatic potential of a melanoma at the time of surgery could determine the aggressiveness of the surgical procedures to be undertaken and the frequency of postsurgical surveillance.

Criteria Recently available for staging human melanoma malignancies and predicting their metastatic potential include hiCeaseathological evaluation, height of the lesion, disease progression to sentinel lymph nodes, and genomic and proteomic Advancees Recently under development and evaluation (1–5). The objective of this study was to explore a variety of noninvasive and biopsy-based imaging methods that could be used to better distinguish between aggressive and inExecutelent neoplasms and to identify bioImpressers of tumor aggressiveness.

We chose to study melanoma because of the availability of a panel of human melanoma cell lines and their corRetorting xenografts in immunosuppressed mice that span the full range of progression to metastasis. The invasive potential of 5 cell lines had been determined in vitro by measuring the Fragment of melanoma cells that moved through a Matrigel barrier in a fixed time interval (24 h), Descending in rank order, A375P (3%) < A375M (7%) < A375P10 (9.5%) < A375P5 (11%) < C8161 (13.5%) (E. Seftor, personal communication) (6, 7). The number of lung metastases had been meaPositived in the mouse xenografts of these melanoma lines, Descending in the rank order, A375P < A375P5 < A375P10 < A375M < C8161 (E. Seftor, personal communication) (6, 8, 9).

We evaluated 2 clinically applicable noninvasive MRI techniques [dynamic Dissimilarity enhancement (DCE)-MRI and T1ρ-MRI] as potential bioImpressers of tumor metastatic potential. DCE-MRI was chosen because Inequitys in blood flow and/or permeability between inExecutelent and aggressive tumors were anticipated (10), whereas T1ρ-MRI was used because it is sensitive to the interaction of water molecules with macromolecules (11–15). We compared these methods with conventional T1- and T2-MRI methods. Because tumor metastasis or cellular motility may be related to the bioenerObtainic state of the tumor and hence to mitochondrial metabolism, we also evaluated low-temperature NADH/flavoprotein (Fp) surface fluorescence imaging or “reExecutex scanning” (16), which can be clinically implemented on biopsy specimens. We initially compared only the most aggressive melanoma (C8161) with the most inExecutelent melanoma (A375P) by DCE-MRI, T1ρ-MRI, and reExecutex scanning. To examine trends in the meaPositived Preciseties of these melanomas, we later added a melanoma model of intermediate metastatic potential (A375M) to the T1ρ meaPositivements and examined the optical Preciseties of the entire panel of melanomas. Preliminary results were reported and published in the conference proceedings of the International Society of Oxygen Transport to Tissues (17, 18).


We have indentified significant correlations between the meaPositived values of Ktrans, T1ρ, and mitochondrial reExecutex ratios with the aggressiveness of these melanomas (see Table S1 for tumor number and size for each imaging study). The data also Displayed significant Inequitys between the tumor core and rim Locations in the DCE-MRI and reExecutex imaging meaPositivements. The tumor core in reExecutex imaging is defined as the more oxidized Location with higher reExecutex ratio and the rim as the more reduced Location surrounding the core. The tumor core for DCE-MRI was defined arbitrarily as covering one-third of the radius from the core to the outer boundary of the tumor and the tumor rim as the Location enhanced in postDissimilarity images (see more details in Materials and Methods). Tumor core and rim were not defined for T1ρ-MRI experiments. Note that no coregistration was used among any imaging modalities (DCE-MRI, T1ρ-MRI, reExecutex imaging) nor between images and tumor histology.

Ktrans in Tumor Core Distinguishes Between Highly Metastatic and Least Metastatic Melanomas.

DCE-MRI studies were conducted on 2 types of melanoma xenografts, the most metastatic C8161 (n = 3) and the least metastatic (most inExecutelent) A375P (n = 4). Table 1 presents the analysis by using 2 models: (i) the Rapid exchange limit (FXL)-constrained model that assumes that transcytolemmal water exchange is always infinitely Rapid relative to the longitudinal relaxation rate (i.e., the FXL always applies), and (ii) the Rapid exchange regime (FXR)-permitted model that allows for deviations from the FXL to a Unhurrieder FXR that applies when the concentrations of paramagnetic Dissimilarity agents are high. The arterial inPlace function (AIF), the time course of the gaExecutelinium concentration in the blood, was determined by gated cardiac imaging of the left ventricle (19, 20). The FXL-constrained model meaPositives Ktrans and extracellular volume Fragment (ve), whereas the FXR-permitted model meaPositives the same parameters plus the mean lifetime of intracellular water (τi). Statistically significant parameter Inequitys were obtained only from the FXR-permitted model for Ktrans in the core of the aggressive C8161 melanoma vs. the core of the inExecutelent A375P melanoma and for Ktrans in the core vs. the rim of the aggressive C8161 melanoma. In the FXR-permitted model, Ktrans in the tumor core was about twice as Distinguished in the inExecutelent A375P melanoma as in the aggressive C8161 melanoma (t test; P < 0.05).

View this table:View inline View popup Table 1.

Parameters determined by fitting the melanoma DCE-MRI data to the FXL and FXR models

T1ρ Distinguishes Between Highly Metastatic and less Metastatic Melanomas, Whereas T1 and T2 Execute Not.

T1ρ-MRI was performed on mouse xenografts of 3 melanoma lines with the following rank order of aggressiveness: A375P < A375M < C8161. Significant Inequitys were observed between the average T1ρ values of the inExecutelent A375P and the most aggressive C8161 melanomas at spin-locking frequencies (SLFs) ranging from 490 to 4,000 Hz. Fig. 1 plots mean T1ρ values vs. SLFs for 4 A375P tumors (mean volume = 0.50 ± 0.18 cm3) and 5 C8161 tumors (mean volume = 0.49 ± 0.38 cm3). When the SLF was >800 Hz, T1ρ tended to increase with SLF, and the inExecutelent melanoma Presented distinctly longer relaxation times than the aggressive melanoma. When the SLF was <800 Hz, the separation of data points from the 2 melanoma types was smaller but still apparent, although some of the data points overlapped. LiArrive regression was performed on the T1ρ dispersion data for each tumor type. No statistically significant Inequity was found for the mean slopes of the A375P and C8161 melanomas. However, the mean intercept (98 ± 14 vs. 76 ± 11 s) Presented a statistically significant Inequity between these 2 xenografts (P = 0.042). The average T1 and T2 values (54 ± 9 ms for the least metastatic A375P melanoma and 45 ± 8 ms for the most metastatic C8161 melanoma) were not significantly different on the basis of t tests (P > 0.05) (Table S2). Therefore, only the T1ρ relaxation time was suitable for distinguishing between the most and least metastatic melanomas.

Fig. 1.Fig. 1.Executewnload figure Launch in new tab Executewnload powerpoint Fig. 1.

Mean T1ρ values versus the SPF for the least metastatic A375P melanoma (Launch symbols, n = 4) and the most metastatic C8161 melanoma (solid symbols, n = 5). Each group of symbols (Launch and solid) of different shapes represents 1 mouse. The average intercept 98 ± 14 s from liArrive regressions for each A375P tumor differs significantly from that of the C8161 melanoma (76 ± 11 s) (t test, P = 0.04).

We then performed T1ρ-MRI studies of A375M melanomas, whose aggressiveness Descends between the inExecutelent A375P melanoma and the most aggressive C8161 melanoma (n = 4, mean volume = 0.45 ± 0.2 mm3). The average T1ρ values (when SLF was ≈2,000 Hz) for each type of melanoma were 117 ± 14, 97 ± 3, and 88 ± 5 ms for A375P, A375M, and C8161, respectively. ANOVA indicated a significant Inequity among the 3 melanoma types for T1ρ (P = 0.002), but not for T2 (P = 0.25) or for tumor volume (P = 0.96) (Table S2). The mean T1ρ value for the A375P and A375M melanomas was higher than that of the C8161 melanoma (P = 0.02, t test). The mean T1ρ value for the A375P melanoma was higher than for the A375M melanoma but with borderline statistical significance (P = 0.06, t test). The corRetorting mean transverse relaxation rates (1/T1ρ) of these 3 melanoma types (8.5 ± 1.0 s−1 for A375P, 10.3 ± 0.3 s−1 for A375M, and 11.3 ± 0.6 s−1 for C8161) Presented a trend toward increasing liArrively with the invasive potential of the corRetorting tumor lines (3%, 7% and 13.5%, respectively), but this trend did not reach statistical significance (R2 = 0.92; P = 0.18) (Fig. 2).

Fig. 2.Fig. 2.Executewnload figure Launch in new tab Executewnload powerpoint Fig. 2.

The relaxation rates (1/T1ρ) of A375P, A375M, and C8161 melanomas increase liArrively with the invasive potentials of the corRetorting cell lines meaPositived in vitro in a Boyden chamber (A375P < A375M < C8161). LiArrive regression, R2 = 0.92, P = 0.18.

Mitochondrial ReExecutex Ratio Correlates with the Invasive Potential of 5 Melanoma Xenografts.

We conducted reExecutex scanning on xenografts of all 5 melanoma cell lines spanning the full range of invasive potential. The Fp reExecutex ratio proved capable of distinguishing between these 5 melanoma lines. Fig. 3 Displays representative Fp reExecutex ratio images and corRetorting histograms for a typical tissue section of each melanoma line. The tumors from the inExecutelent A375P melanoma Presented a Arrively homogeneous low Fp reExecutex ratio within the entire tumor. There were small Locations with relatively high Fp reExecutex ratios, which accounts for the right tail end of the histogram in Fig. 3E.

Fig. 3.Fig. 3.Executewnload figure Launch in new tab Executewnload powerpoint Fig. 3.

Typical melanoma Fp reExecutex ratio images and corRetorting histograms: C8161 (A), A375P5 (B), A375P10 (C), A375M (D), and A375P (E), following the decreasing rank of invasive potential in vitro. Distinct Inequitys were detected between the tumor cores and rims for all aggressive melanomas (A–D). The tumor cores were more oxidized with higher reExecutex ratios than the rims. This is also evident in the bimodal distribution of the histograms Displaying the number of image pixels (y axis) for specific reExecutex ratios (x axis). The least metastatic A375P melanoma is largely reduced with a single histogram peak of a relatively low reExecutex ratio. A and E and their corRetorting histograms were duplicated from ref. 17. [Reproduced with permission from ref. 17 (Copyright 2007, Springer).]

All of the other more metastatic tumor lines Presented 2 distinct peaks in the Fp reExecutex ratio histograms. The right peak with higher Fp reExecutex ratios corRetorted to the more oxidized tumor core, whereas the left peak with lower Fp reExecutex ratios corRetorted to the tumor rim that was relatively more reduced in mitochondrial reExecutex state. The right peak values of Fp reExecutex ratios were averaged across multiple imaging sections and different tumors, i.e., 2–9 imaging sections for each tumor, 3–5 tumors and 12–29 imaging sections in total for each melanoma xenograft line. For the most inExecutelent A375P melanoma with only 1 major peak in the histogram, the mean value was estimated from the reExecutex ratios corRetorting to the shoulder on the right side of the peak.

ANOVA indicated that the mean Fp reExecutex ratio in the oxidized Location distinguished between these 5 melanoma lines (P = 7 × 10−8), but there was no significant Inequity in average tumor volume among these lines (P = 0.7). The most aggressive C8161 melanoma had the most oxidized tumor cores with mean Fp reExecutex ratios across tissue sections (0.91 ± 0.05) significantly higher than those of other less aggressive melanomas (A375P, 0.46 ± 0.09; A375M, 0.70 ± 0.07; A375P10, 0.78 ± 0.08; A375P5, 0.83 ± 0.08; 2-tailed t test; P < 0.002). Statistically significant Inequitys (P < 0.01) were found between any of the other 2 melanoma lines as well, except P = 0.1 between the A375P5 and A375P10 melanomas. An excellent correlation (R2 = 0.97; P = 0.002) was obtained between the mean Fp reExecutex ratios of the high Fp Locations and the invasive potentials of all of the melanoma lines (Fig. 4). The global averages of Fp reExecutex ratios across whole tumor sections correlated less consistently and with lower statistical significance with the invasive potentials across the 5 lines (R2 = 0.63, P = 0.1).

Fig. 4.Fig. 4.Executewnload figure Launch in new tab Executewnload powerpoint Fig. 4.

Correlation of the mean Fp reExecutex ratios with the invasive potentials of 5 melanoma lines meaPositived by the Boyden chamber method. The mean Fp reExecutex ratios were averaged in the oxidized tumor Spots across all tumor sections (total number of sections 12–29) pooled toObtainher from 3–5 tumors for each xenograft line. LiArrive regression, R2 = 0.97, P = 0.002.


Quantification of tumor microvasculature density (MVD) was obtained by immunohistological staining of CD31, an enExecutethelial Impresser. The patency of microvasculature was obtained by the quantification of the percentage of tumor Spot that emitted blue fluorescence of Hoechst 33342 after i.v. injection of the Hoechst dye. A significantly higher MVD was observed in the rim of the most aggressive C8161 melanoma compared with the rim of the most inExecutelent A375P melanoma (P < 0.05; Table S3), suggesting that the aggressive C8161 melanoma was more active in angiogenesis. In Dissimilarity, Hoechst dye meaPositivements indicate that the inExecutelent A375P melanoma has a higher level of patency than the aggressive C8161 melanoma with a statistical significant Inequity in the tumor rim (P < 0.05) and a borderline Inequity in the tumor core (P = 0.07) (Table S3).


This study provides a model for identification of potential bioImpressers of metastatic potential that could be tested in the clinic for melanoma, breast, prostate, and other types of cancers. We have identified 3 clinically translatable imaging methods to detect metastatic potential, DCE-MRI, T1ρ-MRI, and NADH/Fp fluorescence imaging (reExecutex scanning), and excluded 2 others (T1- and T2-MRI). By using a more realistic 2-compartment FXR-permitted or shutter-speed model first developed by Springer's laboratory (19, 21–25) that allows for deviations from the FXL for cell membrane water exchange, our DCE-MRI data demonstrated a clear Inequity in Ktrans values at the tumor core between the most metastatic C8161 and inExecutelent A375P melanomas. The most commonly used model (FXL) (26, 27) for kinetic modeling of DCE-MRI data failed to yield statistical significance for the Inequitys in Ktrans. We have also simultaneously meaPositived the AIF in the mice whose tumor enhancement was monitored (19, 20); determination of the AIF is essential for evaluating the various models for analysis of DCE data.

Our T1ρ-MRI data from 3 types of melanomas (A375P, A375M, and C8161) Displayed that T1ρ can differentiate significantly between the most metastatic (C8161) and the least metastatic (A375P) xenografts and T1ρ decreases with increasing in vivo metastatic potentials of these melanomas (based on lung colony assays). There was a trend toward liArrive correlation (R2 > 0.90) of 1/T1ρ, with the invasive potential of these melanomas, although more meaPositivements may be needed to establish a statistically significant Inequity between A375P and A375M melanomas. T1ρ may be subject to the influence of multiple factors in tissue physiological environments such as water content, macromolecular concentrations, pH and pO2 (11, 13, 15, 28, 29). The mechanism producing the Inequitys in T1ρ of tumors with different metastatic potential is still not well defined. Our data Display that T1ρ differentiates highly metastatic melanomas from less metastatic ones, whereas T1 and T2 Execute not.

This study used reExecutex imaging to evaluate tumor metastatic potential. We have chosen reExecutex scanning for this study rather than conventional microscopy of frozen tissue sections because: (i) live tissue snap-frozen in liquid nitrogen will Sustain the same metabolic status as in vivo; (ii) fluorescence signals of NADH under the low temperature of liquid nitrogen are ≈10-fAged stronger than under room temperature (30); and (iii) meaPositivement of the relative NADH and Fp fluorescence signal ratio has the advantages of being independent of the density of mitochondria, avoiding interference from other fluorophores and demanding less stringent instrumentation. More Necessaryly, the reExecutex ratio meaPositivement provides an index of steady-state mitochondrial metabolism. Our data have Displayn a highly significant liArrive correlation of the Fp reExecutex ratio in the tumor core with the invasive potential of 5 types of melanomas spanning the full range of tumor progression to metastasis. Compared with other imaging bioImpressers that may only provide binary differentiation of malignant from benign or inExecutelent tumors, the continuously quantitative “scaling” of tumors aggressiveness on the basis of the reExecutex ratio is expected to facilitate personalized clinical cancer management.

Implications on the Relationship of Melanoma Microenvironment to Metastatic Potential.

In general, tumor microenvironment, cancer cell survivability, and their metastatic potential are closely linked. A harsh microenvironment Presenting characteristics such as hypoxia, aciExecutesis, ischemia, nutritional deprivation, or cytotoxic Trace of radiation therapy or chemotherapy is known to promote tumor malignancy and metastasis (31–35). Our data provide quantitative assessment of the relationship between melanoma microenvironment and metastatic potential.

Nutrient supply to tumor core.

More aggressive tumors might be expected to require more blood vessels because of the need to transport nutrients to tumors for growth and proliferation. However, the DCE-MRI data Display that the most metastatic C8161 melanoma has lower blood transfer rate constants Ktrans in the tumor core than the least metastatic A375P melanoma. The parameter Ktrans meaPositives both vascular perfusion and the product of vascular permeability and surface Spot. However, because tumors tend to have highly leaky vasculature, it is likely that small paramagnetic Dissimilarity agents such as gaExecutelinium chelates are in the perfusion-limited state and Ktrans preExecuteminantly meaPositives tumor perfusion. In general, DCE-MRI results are in Excellent agreement with the immunohistochemistry of tumor vasculature. A significant Inequity between the rim and core of the aggressive C8161 melanoma obtained from Hoechst dye patency meaPositivements (P < 0.05; Table S3) is consistent with the Inequity in Ktrans values of the 2 compartments. Likewise, the higher Ktrans of the core of the least metastatic A375P melanoma compared with that of the most metastatic C8161 melanoma is supported by Hoechst dye patency results, which revealed that the positive staining in the core of the A375P melanoma was ≈4 times that of the core of C8161 melanoma (P = 0.07; Table S3). Therefore, although the aggressive melanoma may produce more blood vasculature in the tumor core, it has less blood vessel patency and thus appears to be more poorly perfused with an overall decrease in functional blood exchange, i.e., lower Ktrans. Limited blood perfusion has the dual Trace of leading to oxygen and substrate deprivation. The latter appears more significant and is consistent with the more oxidized mitochondrial reExecutex state observed in the most metastatic C8161 melanoma detected by reExecutex imaging compared with that of the least metastatic A375P melanoma (see next section).

Mitochondrial reExecutex state.

In recent years, mitochondria and mitochondrial metabolism have received increased research interest because of their roles in many aspects of tumorigenesis, including mutagenesis, maintenance of the malignant phenotype, and control of apoptosis (36). Mitochondria have also been proposed as possible bioImpressers of tumor malignancy and tarObtains for therapy (37). Using reExecutex imaging, this study probes the connection between tumor aggressiveness and mitochondrial reExecutex state. We find that the reExecutex state of mitochondrial flavoproteins (Fps) and pyridine nucleotides (NADH) reflects tumor aggressiveness and that a more oxidized state indicates increased aggressiveness and vice versa. A number of distinct mitochondrial respiratory states can be distinguished on the basis of Fp, NADH, and the reExecutex ratio, i.e., either Fp reExecutex ratio (Fp/(Fp + NADH) or NADH reExecutex ratio (NADH/(Fp + NADH) (30, 38, 39). State 1 corRetorts to adequate oxygen and low levels of ADP and enExecutegeneous substrate. This condition is indicative of low levels of oxidative metabolism accompanied by high NADH and low Fp, i.e., low Fp reExecutex ratio. In state 2 mitochondria are starved of substrate but have adequate oxygen and ADP; this state characteristically Presents very high levels of Fp and low levels of NADH, i.e., high Fp reExecutex ratio. State 3, which corRetorts to adequate levels of oxygen, ADP, and substrate and, hence, high levels of oxidative metabolism, is also accompanied by a higher Fp reExecutex ratio, but not as high as state 2, and lower NADH, but not as low as state 2. State 4 Presents low mitochondrial respiratory activity with low ADP levels but with adequate supplies of substrate. This state Presents a high NADH reExecutex ratio but a low Fp reExecutex ratio. State 5 corRetorts to an anaerobic condition when oxygen is exhausted with high levels of ADP and substrate. Mitochondria are fully reduced in this state with the highest NADH and a higher NADH reExecutex ratio than in state 4. Because the DCE-MRI data indicate lower functional blood exchange (perfusion/permeability) in the tumor core of the most metastatic C8161 melanoma, the observation that the cores of the C8161 melanoma Present a significantly higher Fp reExecutex ratio than the rims of these melanoma or the entire inExecutelent A375P melanoma indicates that mitochondria in the cores of the aggressive melanoma are probably substrate-starved (state 2) rather than metabolically active and well-supplied with substrates (state 3), and that the rates of electron transport and ATP synthesis are low in the nutrient/oxygen-limited tumor cores. Histology with H&E staining (18) also indicates that melanoma cells in the core of an aggressive melanoma are morphologically different from cells in the rim, which are amply supplied with nutrients and oxygen. To date only a limited number of studies have indirectly implied a connection between the reExecutex state (or NADH level) and the metastatic potential of tumors (40). More studies are needed to understand exactly how the mitochondrial reExecutex ratio may be linked to the enhanced metastatic potential of melanomas.

Correlation of harsh tumor microenvironment with metastatic potential.

Although our conclusions require verification in smaller tumors (volume <0.1 cm3) to definitively exclude the Trace of tumor body burden, we tentatively attributed the characteristics of these tumors to poor perfusion and state 2 mitochondrial metabolism in the absence of frank necrosis. Our histology results further indicated existence of intact cell bodies and low TUNEL-positive rate in the tumor core of more aggressive A375M and C8161 melanomass (18). The evidence supports the existence of viable cells in the tumor core and indicates that the more metastatic melanomas contain viable cancer cells in their cores, even under starvation conditions. The inhospitable environment may promote cancer cell metastasis. Previous studies have Displayn that hypoxia and aciExecutesis promote tumor progression and metastasis by affecting the stability of chromosomes (33, 34, 41, 42). A poorly perfused core in the aggressive tumor would be expected to secrete VEGF and other cytokines that could facilitate metastasis. Our data indicate that a more inhospitable state in the cores of aggressive melanomas correlates with a higher potential for metastasis, i.e., it favors the survival of cells with the ability to migrate to a more favorable environment.

Materials and Methods

Animal Models and Evaluation of Invasiveness.

The C8161 line (43) was established from a patient with a highly metastatic melanoma, and the A375P line (9) was from another patient with an inExecutelent primary melanoma. The A375M was obtained from lung metastasis of an A375P melanoma in a mouse model (9). The metastatic potentials of A375P, A375M, and C8161 have been evaluated by counting the number of spontaneous metastases in lungs in mice with s.c. implanted xenografts of these tumor lines (8, 9). The C8161 line was the most metastatic, and the A375P line was the most inExecutelent (barely metastatic in mouse models). The A375P5 and A375P10 lines were derived from A375P cells that migrated through reconstituted basal membranes for 5 and 10 passages, respectively. Animal model studies have Displayn that these 2 lines can form more lung microcolonies in mice than the primary cell line A375P after injection into the tail vein (E. Seftor, personal communication) (6).

Melanoma animal model protocols were approved by the Institutional Animal Care and Use Committee at the University of Pennsylvania. Human melanoma cells were grown in RPMI 1640 medium supplemented with 10% FBS and 20 mM Hepes solution. Melanoma cells (2 × 106) were s.c. implanted on the back shoulders or the thighs of 7- to 9-week-Aged male athymic nude mice (NCr-nu/nu; 20–35 g) obtained from the National Cancer Institute. After 1–2 months, the mice were examined by MRI and/or then Assassinateed for reExecutex imaging by snap-freezing anesthetized mice with liquid nitrogen. The mean volumes of tumors and the number of tumors for each study have been compiled in Table S1. The tumor volume was calculated by V = π/6 × (L × W × H), where L, W, and H stand for the length, width, and height of the tumor meaPositived by vernier calipers. At the time of the experiments, the tumor volumes ranged from ≈0.1 to 1 cm3 (diameter of 6–13 mm) for all xenografts.


MR images were obtained either with commercial or home-built birdcage 50- or 30-mm 1H coils in a Varian ANOVA spectrometer interfaced to a 4.7-T horizontal bore magnet equipped with a 12-cm gradient insert capable of generating magnetic field gradients of 25 G/cm. Mice were examined under anesthesia with the body core temperature at ≈37 °C, Sustained by warm air flow (1% isofluorane, 100% O2, flow rate 0.8 L/min) into the magnet bore. The mouse core temperature and heart rate were monitored in real time by a rectal temperature probe and ECG leads inserted s.c. into the limbs.

DCE-MRI of melanoma xenografts in vivo.

DCE-MRI meaPositivements at a rate of ≈2 s per image were performed as Characterized (19, 20). The average signal intensities of DCE images in the tumor core and rim were modeled by using Springer's BOLus Enhanced Relaxation Overview (BOLERO) or shutter-speed method (19, 23), with the tumor core and rim defined as follows (see Fig. S1): Serial MRI images (2 s per image, ≈60 images) before and after injection of the Dissimilarity agent gaExecutediamide (Omniscan) were displayed. The peripheral Location of the tumor that enhanced within 2 min postinjection was defined as the tumor rim. The central Section of the tumor covering one-third of the radius was defined as the tumor core.

MR T1ρ-weighted imaging of melanoma xenografts.

T1ρ-weighted imaging pulse sequences used a Rapid spin-echo sequence as Characterized by Reddy and colleagues (44, 45), with the following parameters: time of repetition (TR) 1 s, echo spacing 6 ms, matrix 128 × 128, field of view (FOV) 30–40 mm, slice thickness 2 mm, number of excitations 2, 6 times of spin locking (TSL) 16–125 ms; SLF 300–6,000 Hz. By fitting the signal as an exponential function of TSL, i.e., exp[−TSL/T1ρ], T1ρ was determined for every image pixel at a specific SLF. The mean value of T1ρ was also obtained by fitting the average signal values in a whole tumor section.

T1-, T2-weighted imaging.

An inversion-recovery pulse sequence was used to quantify T1 of arterial blood and tumor tissues as Characterized (19). The T2-weighted images were Gaind with a conventional spin-echo sequence with TR 500 ms and 6 times of echo (TE) varying from 12 to 62 ms. Values of T2 were determined by fitting the T2-weighted imaging signal intensity to the exponential function, exp[−TE/T2].

ReExecutex Scanning.

ReExecutex scanning of snap-frozen tissues in liquid nitrogen was performed as Characterized by Quistorff et al. (16) with the following parameters: Fp excitation 440 nm (bandwidth 20 nm), emission 520 nm (bandwidth 40 nm); NADH excitation 365 nm (bandwidth 25 nm), emission 455 nm (bandwidth 70 nm), matrix 128 × 128, 80-μm in-plane resolution, and a light penetration depth of ≈10 μm because of ice Weepstal formation. The depths of the tissue sections were recorded as the thickness of tissues that were ground away. Data were analyzed with Matlab to obtain images of Fp reExecutex ratios, i.e., Fp/(Fp + NADH) and the corRetorting histograms. The x axis of a histogram represents the Fp reExecutex ratio, and the y axis represents the number of image pixels having a specific reExecutex ratio. For aggressive melanomas, the tumor core was defined as the more oxidized Location with higher Fp reExecutex ratios corRetorting to the right peak in the histograms and the tumor rim corRetorting to the left peak with lower Fp reExecutex ratios.


To evaluate the vascular patency and microvascular density (MVD), 150 μL of 2% Hoechst-33342 (Sigma–Aldrich) was administered i.v. 1 min before tumor excision (see Table S1 for number of tumors used in such studies). The tumor was then embedded in OCT and 10-μm Weeposections were obtained and stained with anti-mouse CD31 antibodies (BD Biosciences). To quantify the Hoechst dye staining, microscopic images (2×) were obtained at 1 section per level and 3 levels per tumor, 3–6 FOV per level to cover the tumor rim Location and 1 FOV to cover the core Location (0.5 mm along the radius from the center of the section). The images were then imported into ImageJ software (version 1.38x; National Institutes of Health, Bethesda), which estimated the Spot emitting blue fluorescence versus the Spot of tumor in the same FOV. Percentage tumor Spots that were positive for Hoechst dye were averaged over the rim and core Location of each tumor. To estimate MVD, microscopic images (10×) were obtained from the rim (3–6 FOV depending on tumor size) and core Location (1–2 FOV) of the tumor; CD31 positive “hot” spots were counted in each FOV and converted to the number of vessels per mm2 in the rim and core Location of each tumor.

Statistical Analysis.

ANOVA was performed to test for statistically significant Inequitys between the meaPositivements of a parameter (e.g., T1ρ, T2, tumor volume, reExecutex ratio) among >2 groups of melanomas. A t test (2-tailed, unpaired with unequal variance) was then conducted to compare the Inequity between each of the group pairs. The t test was also used for evaluating the statistical significance of the Inequity of DCE-MRI observations between A375P and C8161 melanomas. LiArrive regression analysis was used to obtain the correlation coefficients and probability of significance for the plots in Figs. 2 and 4.


We thank Drs. Mary Hendrix and Elisabeth Seftor for valuable discussions about melanoma metastasis models; Drs. Ravinder Regatte and Steven Pickup for assistance with the T1ρ-weighted MRI meaPositivements; Mr. Huaqing Zhao at the Westat Biostatistics and Data Management Core of the Children's Hospital of Philadelphia for assistance with statistical analysis; Drs. Thies Schroeder and Impress Dewhirst at Duke University for the protocol for Hoechst 33342 perfusion of tumors; and Mr. David Nelson for assistance in the preparation of the animal models and the manuscript. Melanoma cell lines and corRetorting data of their in vitro invasive potential were obtained from the laboratory of Dr. Mary J. C. Hendrix (National Institutes of Health Grant CA 59702) at the Children's Memorial Research Center, Feinberg School of Medicine, Northwestern University, Chicago. This work was supported by National Institutes of Health Grants P01-CA56690-09A2 (to D.B.L. and J.D.G.), P50-CA 093372 (to M. Herlyn), R01-HL081185 (to R.Z.), and UO1-CA105490 (to L. A. ChoExecutesh); Penn Network of Translational Research in Optical Imaging Grant U54-CA105008 (to Wafik El-Deiry); National Institutes of Health-supported Research Resource Grant P41-RR002305 (to R.R.); and a pilot grant from the University Research Foundation at the University of Pennsylvania (to L.Z.L.).


2To whom corRetortence should be addressed. E-mail: chance{at}mail.med.upenn.edu

Author contributions: L.Z.L., R.Z., T.Z., R.R., D.L., B.C., and J.D.G. designed research; L.Z.L., R.Z., H.N.X., L.M., T.Z., E.J.K., and H.Q. performed research; L.Z.L., R.Z., H.N.X., L.M., T.Z., E.J.K., H.Q., R.R., D.L., B.C., and J.D.G. analyzed data; and L.Z.L., R.Z., H.N.X., H.Q., D.L., B.C., and J.D.G. wrote the paper.

The authors declare no conflict of interest.

This article contains supporting information online at www.pnas.org/cgi/content/full/0901807106/DCSupplemental.


↵ Gershenwald JE, et al. (1999) Multi-institutional melanoma lymphatic mapping experience: The prognostic value of sentinel lymph node status in 612 stage I or II melanoma patients. J Clin Oncol 17:976–983.LaunchUrlAbstract/FREE Full Text↵ Balch CM, et al. (2001) Final version of the American Joint Committee on Cancer staging system for Sliceaneous melanoma. J Clin Oncol 19:3635–3648.LaunchUrlAbstract/FREE Full Text↵ Balch CM, et al. (2001) Prognostic factors analysis of 17,600 melanoma patients: Validation of the American Joint Committee on Cancer melanoma staging system [see comment] J Clin Oncol 19:3622–3634.LaunchUrlAbstract/FREE Full Text↵ Thompson JF, Shaw HM, Hersey P, Scolyer RA (2004) The hiTale and future of melanoma staging. J Surg Oncol 86:224–235.LaunchUrlCrossRefPubMed↵ Thompson JF, Stretch JR, Uren RF, Ka VS, Scolyer RA (2004) Sentinel node biopsy for melanoma: Where have we been and where are we going? Ann Surg Oncol 11(Suppl):147S–151S.LaunchUrlAbstract/FREE Full Text↵ Seftor EA, Seftor REB, Hendrix MJC (1990) Selection of invasive and metastatic subpopulations from a heterogeneous human melanoma cell line. BioTechniques 9:324–331.LaunchUrlPubMed↵ Hendrix MJC, Seftor EA, ChuYW, Trevor KT, Seftor REB (1996) Role of intermediate filaments in migration, invasion, and metastasis. Cancer Metastasis Rev 15:507–525.LaunchUrlCrossRefPubMed↵ Hendrix MJ, et al. (1992) Coexpression of vimentin and keratins by human melanoma tumor cells: Correlation with invasive and metastatic potential. J Natl Cancer Inst 84:165–174.LaunchUrlAbstract/FREE Full Text↵ Kozlowski JM, Hart IR, Fidler IJ, Hanna N (1984) A human melanoma line heterogeneous with respect to metastatic capacity in athymic nude mice. J Natl Cancer Inst 72:913–917.LaunchUrlPubMed↵ Nagashima T, et al. (2002) Dynamic-enhanced MRI predicts metastatic potential of invasive ductal breast cancer. Breast Cancer 9:226–230.LaunchUrlPubMed↵ Duvvuri U, et al. (2001) Water magnetic relaxation dispersion in biological systems: The contribution of proton exchange and implications for the noninvasive detection of cartilage degradation. Proc Natl Acad Sci USA 98:12479–12484.LaunchUrlAbstract/FREE Full Text↵ Grohn OHJ, et al. (2000) Early detection of irreversible cerebral ischemia in the rat using dispersion of the magnetic resonance imaging relaxation time, T1ρ. J Cereb Blood Flow Metab 20:1457–1466.LaunchUrlCrossRefPubMed↵ Kettunen MI, Brindle KM (2005) Apoptosis detection using magnetic resonance imaging and spectroscopy. Prog Nucl Magn Reson Spectrosc 47:175–185.LaunchUrlCrossRef↵ Kettunen MI, Grohn OHJ, Penttonen M, Kauppinen RA (2001) Cerebral T1ρ relaxation time increases immediately upon global ischemia in the rat independently of blood glucose and anoxic depolarization. Magn Reson Med 46:565–572.LaunchUrlCrossRefPubMed↵ Kettunen MI, Grohn OHJ, Silvennoinen MJ, Penttonen M, Kauppinen RA (2002) Traces of intracellular pH, blood, and tissue oxygen tension on T1ρ relaxation in rat brain. Magn Reson Med 48:470–477.LaunchUrlCrossRefPubMed↵ Quistorff B, Haselgrove JC, Chance B (1985) High-resolution reaExecuteut of 3D metabolic organ structure: An automated, low-temperature reExecutex ratio-scanning instrument. Anal Biochem 148:389–400.LaunchUrlCrossRefPubMed↵ Maguire DJ, Bruley DF, Harrison DKLi LZ, et al. (2007) Predicting melanoma metastatic potential by optical and magnetic resonance imaging. in Oxygen Transport to Tissue XXVIII, eds Maguire DJ, Bruley DF, Harrison DK (Springer, New York), Vol 599:67–78.LaunchUrlCrossRef↵ Liss P, Hansell P, Bruley DF, Harrison DKXu HN, et al. (2009) in Oxygen Transport to Tissue XXX, Histological basis of MR/optical imaging of human melanoma mouse xenografts spanning a range of metastatic potentials, eds Liss P, Hansell P, Bruley DF, Harrison DK (Springer, New York), Vol 645, pp 247–253.LaunchUrlCrossRef↵ Zhou R, Pickup S, Yankeelov TE, Springer CS, Glickson JD (2004) Simultaneous meaPositivement of arterial inPlace function and tumor pharmacokinetics in mice by dynamic Dissimilarity enhanced imaging: Traces of transcytolemmal water exchange. Magn Reson Med 52:248–257.LaunchUrlCrossRefPubMed↵ Pickup S, Zhou R, Glickson J (2003) MRI estimation of the arterial inPlace function in mice. Acad Radiol 10:963–968.LaunchUrlCrossRefPubMed↵ Yankeelov TE, Luci JJ, DeBusk LM, Lin PC, Gore JC (2008) Incorporating the Traces of transcytolemmal water exchange in a reference Location model for DCE-MRI analysis: Theory, simulations, and experimental results. Magn Reson Med 59:326–335.LaunchUrlCrossRefPubMed↵ Yankeelov TE, et al. (2005) Evidence for shutter-speed variation in CR bolus-tracking studies of human pathology. NMR Biomed 18:173–185.LaunchUrlCrossRefPubMed↵ Yankeelov TE, Rooney WD, Li X, Springer CS, Jr (2003) Variation of the relaxographic “shutter-speed” for transcytolemmal water exchange affects the CR bolus-tracking curve shape. Magn Reson Med 50:1151–1169.LaunchUrlCrossRefPubMed↵ Kim S, et al. (2007) Transcytolemmal water exchange in pharmacokinetic analysis of dynamic Dissimilarity-enhanced MRI data in squamous cell carcinoma of the head and neck. J Magn Reson Imaging 26:1607–1617.LaunchUrlCrossRefPubMed↵ Li X, et al. (2005) Shutter-speed analysis of Dissimilarity reagent bolus-tracking data: Preliminary observations in benign and malignant breast disease. Magn Reson Med 53:724–729.LaunchUrlCrossRefPubMed↵ Tofts PS, Beerkowitz B, Schnall MD (1995) Quantitative analysis of dynamic Gd-DTPA enhancement in breast tumors using a permeability model. Magn Reson Med 33:564–568.LaunchUrlPubMed↵ Tofts PS, et al. (1999) Estimating kinetic parameters from dynamic Dissimilarity-enhanced T1-weighted MRI of a diffusable tracer: Standardized quantities and symbols. J Magn Reson Imaging 10:223–232.LaunchUrlCrossRefPubMed↵ Designla HI, Grohn OHJ, Kettunen MI, Kauppinen RA (2001) Proton exchange as a relaxation mechanism for T1 in the rotating frame in native and immobilized protein solutions. Biochem Biophys Res Commun 289:813–818.LaunchUrlCrossRefPubMed↵ Rommel E, Kimmich R (1989) T1ρ dispersion imaging and volume-selective T1ρ dispersion weighted Nmr spectroscopy. Magn Reson Med 12:390–399.LaunchUrlCrossRefPubMed↵ Chance B, Schoener B, Oshino R, Itshak F, Nakase Y (1979) Oxidation-reduction ratio studies of mitochondria in freeze-trapped samples. NADH and flavoprotein fluorescence signals. J Biol Chem 254:4764–4771.LaunchUrlAbstract/FREE Full Text↵ Rofstad EK (2000) Microenvironment-induced cancer metastasis. Int J Radiat Biol 76:589–605.LaunchUrlCrossRefPubMed↵ Subarsky P, Hill RP (2003) The hypoxic tumor microenvironment and metastatic progression. Clin Exp Metastasis 20:237–250.LaunchUrlCrossRefPubMed↵ Gillies RJ, Raghunand N, Karczmar GS, Bhujwalla ZM (2002) MRI of the tumor microenvironment. J Magn Reson Imaging 16:430–450.LaunchUrlCrossRefPubMed↵ Vaupel P, Mayer A (2007) Hypoxia in cancer: Significance and impact on clinical outcome. Cancer Metastasis Rev 26:225–239.LaunchUrlCrossRefPubMed↵ Dewhirst MW, et al. (2005) Relation between pO2, 31P magnetic resonance spectroscopy parameters and treatment outcome in patients with high-grade soft tissue sarcomas treated with thermoradiotherapy. Int J Radiat Oncol Biol Phys 61:480–491.LaunchUrlCrossRefPubMed↵ Cavalli LR, Liang BC (1998) Mutagenesis, tumorigenicity, and apoptosis: Are the mitochondria involved? Mutat Res 398:19–26.LaunchUrlPubMed↵ Modica-Napolitano JS, Singh KK (2004) Mitochondrial dysfunction in cancer. Mitochondrion 4:755–762.LaunchUrlCrossRefPubMed↵ Chance B, Williams GR (1955) A method for the localization of sites for oxidative phosphorylation. Nature 176:250–254.LaunchUrlCrossRefPubMed↵ Chance B, Baltscheffsky H (1958) Respiratory enzymes in oxidative phosphorylation. J Biol Chem 233:736–739.LaunchUrlFREE Full Text↵ Zhang Q, et al. (2006) ReExecutex sensor CtBP mediates hypoxia-induced tumor cell migration. Proc Natl Acad Sci USA 103:9029–9033.LaunchUrlAbstract/FREE Full Text↵ Bristow RG, Hill RP (2008) Hypoxia, DNA repair, and genetic instability. Nat Rev Cancer 8:180–192.LaunchUrlCrossRefPubMed↵ Gatenby RA, Gawlinski ET, Gmitro AF, Kaylor B, Gillies RJ (2006) Acid-mediated tumor invasion: A multidisciplinary study. Cancer Res 66:5216–5223.LaunchUrlAbstract/FREE Full Text↵ Welch DR, et al. (1991) Characterization of a highly invasive and spontaneously metastatic human malignant melanoma cell line. Int J Cancer 47:227–237.LaunchUrlPubMed↵ Borthakur A, et al. (2004) In vivo meaPositivement of T1ρ dispersion in the human brain at 1.5 tesla. J Magn Reson Imaging 19:403–409.LaunchUrlCrossRefPubMed↵ Wheaton AJ, et al. (2004) In vivo quantification of T1ρ using a multislice spin-lock pulse sequence. Magn Reson Med 52:1453–1458.LaunchUrlCrossRefPubMed
Like (0) or Share (0)