1.

Introduction

In vivo tumor imaging requires the establishment of contrast between the tumor and surrounding normal tissues. Magnetic resonance imaging (MRI) techniques rely on differences in water relaxivity between malignant lesions and healthy cells to achieve contrast. Positron emission tomography (PET) imaging methodologies generally exploit differences in metabolic rates between cancer and normal cells to allow visualization of tumors. Some radiological imaging methods rely on the elevated passive vascular permeability of many neoplasms to achieve contrast with healthy tissues.¹ Finally, a few imaging protocols exploit the overexpression of certain receptors on cancer cells to target attached imaging or contrast agents specifically to malignant cells. Examples of targeting ligands exploited for the delivery of imaging agents include antibodies,² hormones,³ small peptides,^{4 5} and the vitamin folic acid.⁶

Folic acid-based targeting strategies rely on (1) the high affinity of folate for its cell surface receptor (K_d∼10⁻¹⁰) (Ref. 7), (2) the absence or inaccessibility of the receptor on normal cells [the folate receptor (FR) on those normal cells that express it is primarily confined to the apical surface of polarized epithelia where it is not easily accessed from the blood],⁸ and (3) the elevated expression of the receptor on many tumor cells. Those tumors that commonly overexpress FR include cancers of the ovary,⁹ breast,¹⁰ kidney,¹¹ lung,^{10 12} endometrium,¹⁰ myeloid cells,¹³ and brain.^{8 14} Diagnostic imaging of folate receptor-expressing tumors using folate-conjugated gamma scintigraphy or MRI contrast agents generally yields good contrast with adjacent normal tissues.^{15 16} Optical imaging of receptor-expressing tumors using folate-linked dyes has also recently been explored.^{17 18}

While scintigraphic and MRI techniques often offer the sensitivity needed to detect metastatic lesions noninvasively, targetable fluorescent agents may find applications where use of the these methodologies is inadequate. Thus, during debulking surgery, where malignant loci can be difficult to identify, the presence of a fluorescent signal might assist the surgeon in locating them. Furthermore, when tumors are situated near the body surface and analysis of tumor progression is repeatedly required, optical imaging methodologies can often be applied without danger of toxicity. Also, in the course of an endoscopic examination,^{19 20 21} fluorescence imaging can allow the precise assessment of the location, size, and invasiveness of a tumor. Finally, optical imaging instruments may be simpler and less expensive to operate than those required for other imaging technologies, permitting their eventual application by less specialized medical centers.

In this paper we demonstrate the ability to visualize folate receptor-expressing cancer tissues in various peritoneal, subcutaneous, and metastatic murine tumor models following intravenous administration of a folate-fluorescein conjugate. Because tumors as small as a few millimeters can be easily detected without image processing or enhancement, and since malignant lesions smaller than 0.5 mm can be located with appropriate optical instrumentation, these results provide an incentive for the development of folate-targeted fluorophores as aids in the localization and characterization of tumors during surgery and endoscopic examination.

2.

Experimental

3.

Results

BALB/c and DBA mice were implanted subcutaneously with M109 and L1210 cultured cancer cells, respectively, and tumors were allowed to grow to ∼200 mm³. Two hours following intravenous injection of folate-fluorescein, the tumor-bearing mice were euthanized and their tumors imaged using both normal light and laser (fluorescence view) illumination. As seen in Fig. 1, tumor masses were significantly more fluorescent than adjacent normal tissues. In fact, the differences in fluorescence intensity were readily apparent without background subtraction or other image-processing methods used in previous reports.^{24 25} The absence of tumor fluorescence upon coinjection of excess free folate [Figs. 1(c) and 1(d)] further shows that the tumor uptake is completely folate mediated. No significant loss of fluorescence could be visually detected during the time used for imaging, even after 1 h of illumination by the laser (data not shown). Although some areas of the tumor tissue showed minor variations in fluorescence intensity, all tumor areas fluoresced brightly, as shown by the spectral analysis described later. Other studies have also shown that folate receptor expression is uniform throughout the tumor (unpublished observations). These data suggest that folate-linked fluorophores might be useful for optical imaging wherever tumor location allows access to incident and fluorescent light.

Figure 1

Targeting of folate-fluorescein to L1210 subcutaneous tumors. Mice were injected intravenously (femoral vein) with 10 nmol of folate-fluorescein and sacrificed 2 h later. Another mouse was coinjected with a 100-fold molar excess of folic acid. Panels (a) and (b) display M109 subcutaneous tumors (left tissue sample) next to muscle (right) tissue isolated from the same mouse. Panel (c) shows the white light image and panel (d) shows the fluorescent image of L1210 subcutaneous tumors from a mouse injected with folate-fluorescein (top tumor) or folate-fluorescein plus 100-fold molar excess of free folate (bottom tumor).

To more accurately quantitate the difference in the targeted fluorescence between normal and malignant tissues, it was decided to examine the spectral properties of various tissues using more analytical spectrofluorimetric methods. For this purpose, tissue samples from both folate-fluorescein injected and noninjected tumor-bearing mice were examined using a spectrograph equipped with laser excitation, as described in Sec. 2.4. Although tumor tissue from injected mice consistently showed a ∼fivefold greater fluorescence than the autofluorescence of noninjected controls, liver and muscle tissues, which did not express the folate receptor, displayed no difference in fluorescence intensity, regardless of the presence or absence of folate-fluorescein pretreatment (Fig. 2). Unfortunately, because of this autofluorescence, direct quantitation of folate-fluorescein uptake into malignant and nonmalignant tissues could not be determined. However, based on previous quantitative biodistribution studies of other low molecular folate conjugates,^{16 22 26 27 28} we project that folate-fluorescein distribution in vivo will follow the biodistribution of folate receptors accessible to the blood, i.e., high uptake in the tumor and kidney with little or no uptake in other tissues. Furthermore, because these previous reports have shown the main route of excretion to be through the kidneys,^{16 26 27 28} and since clearance of the fluorescein conjugate from nontargeted tissues was similarly rapid (i.e., <2 h), we assume that excretion of folate-fluorescein also occurs mainly through the kidneys.

Figure 2

Comparison of the fluorescence intensity of tissues from mice injected with folate-fluorescein or buffer control. Tissues were dissected 2 h after folate-fluorescein or buffer injection and spectra of both groups were measured as described in Sec. 2.4. The data points represent the average intensities of the indicated tissues from three mice.

In order to test the ability of folate-fluorescein to facilitate detection of microscopic tumor nodules, metastatic tumors of M109 cell origin were induced by inoculation of cells either into the peritoneal cavity (i.p.) or femoral vein (i.v.) of the syngeneic mice. After allowing sufficient time for tumor growth, tumor loci were again imaged by intravenous administration of folate-fluorescein followed by laser excitation. As seen in Fig. 3, metastatic tumor nodules smaller than 0.5 mm were readily observable in both lungs of the intravenously inoculated mice, and all tumor metastases fluoresced intensely compared with adjacent nonmalignant tissue. Similar results were also obtained in the intraperitoneal metastatic tumor model when the peritoneal cavity was opened and imaged as described earlier (not shown).

Figure 3

Optical images of multiple metastatic M109 tumor nodules in a mouse lung 2 h after intravenous (femoral vein) injection of 10 nmol of folate-fluorescein. (a) Normal light image of the opened chest cavity. (b) Same chest cavity visualized under 488-nm argon laser illumination.

To further explore the ability of folate-fluorescein to image small metastatic lesions, L1210 cells were also inoculated intravenously into syngeneic mice and the various anatomical regions were again imaged as described earlier. Since L1210 cells are lymphocytic in origin and relatively small in size, they are capable of forming tumor nodules throughout the body after i.v. injection, rather than becoming entrapped primarily in the lung capillary network. As shown in Fig. 4, both upper and lower lobes of the liver contained multiple microscopic nodules that were clearly revealed under laser illumination. Furthermore, the animal’s spleen was almost entirely filled with minute fluorescent tumor loci [Figs. 5(a) and 5(b)] and was consequently enlarged to roughly twice its normal size. Tumor nodules were also found in muscle tissue near the shoulder [Figs. 5(c) and 5(d)] and in the brain, spinal cord, and along the pleura (not shown). In fact, tumor nodules were seen in most tissues of the body and no nodules were identified that did not fluoresce under laser illumination.

Figure 4

Images of L1210 metastatic tumor nodules in the liver of a DBA mouse. The mouse was injected intravenously (femoral vein) with 10 nmol of folate-fluorescein and imaged 2 h later. Panels (a) and (b) show the white light and fluorescent light images of the lower lobe of the liver, while panels (c) and (d) display the same images of the upper lobe of the liver.

Figure 5

L1210 tumor nodules in the spleen and muscle tissue of a DBA mouse are detectable after injection of folate-fluorescein. The mice were treated as in Fig. 4. Panels (a) and (b) show the white light and fluorescent light images of a tumor-infiltrated spleen, and panels (c) and (d) display the same images of a tumor nodule in the skeletal muscle of the neck.

Since methotrexate is a folate analog with two orders of magnitude lower affinity for the folate receptor than folic acid,²⁹ it was decided that a comparison of the uptake of folate-fluorescein, methotrexate-fluorescein, and amino fluorescein (nonbinding control) might allow a better assessment of the importance of ligand affinity on tumor-targeted optical imaging. For this purpose, equal amounts (10 nmol) of the three compounds were injected intravenously into mice with subcutaneous M109 tumors, and the tumors along with various normal tissues were compared for fluorescence intensity. As expected from their relative affinities [folate-fluorescein binds to the folate receptor with 44 the affinity of folic acid (Dr. Chris Leamon, Endocyte Inc., personal communication) and methotrexate-fluorescein displays an affinity in the micromolar range^{30 31 32}], tumors from mice injected with folate-fluorescein fluoresced brightly, whereas tumors from mice treated with either methotrexate-fluorescein or the nontargeted amino fluorescein were not detectably more fluorescent than untreated controls (not shown). Furthermore, normal tissues from all three mice displayed only their normal autofluorescence. These results would seem to suggest that affinities in the nanomolar, but not micromolar, range are necessary to achieve good contrast between targeted and nontargeted tissues.

4.

Discussion

The folate receptor is overexpressed in a number of tumors, including cancers of the ovary,^{10 33 34} brain,^{8 14} kidney,¹¹ myeloid cells,¹³ lung,^{10 12} and breast.¹⁰ Using both a lung and lymphoma murine tumor model, we have shown that folate receptor-positive cancers can be optically imaged by administration of a commercially available folate-conjugated fluorescent dye. Although it was never the purpose of this study to develop the optimal probe for in vivo imaging, initial results with this commercially available dye suggest that characterization of tumor location, size, and growth rate by nontoxic optical methods is possible with improved fluorescent probes, optimal data collection equipment, and appropriate image reconstruction software.

Although the potential applications for folate-targeted tissue-translucent optical imaging agents seem obvious, the question naturally arises regarding the value of targeted imaging agents such as folate-fluorescein that can only reveal a malignancy when the tumor is easily exposed. We suggest two possible applications for such fluorescent probes. First, the ability of highly fluorescent conjugates to reveal a tumor’s location under direct illumination might be exploited to guide a surgeon’s knife during tumor resection. For example, ovarian cancers, which are strongly folate receptor-positive, are usually asymptomatic during early stages of the disease, resulting in initial diagnoses at stage III or IV in 70 of all patients. At these late stages, there is significant spread of the cancer throughout the peritoneal cavity, with lesions commonly attached to the omentum, intestines, and other internal organs.³⁵ Current treatment for such advanced cancers involves debulking surgery followed by chemotherapy. It is important, to note that optimal tumor debulking has been shown to significantly increase the rate of patient survival,^{36 37} suggesting that if otherwise undetected malignant masses were to be removed, a further increase in survival might be realized. We suggest that use of a folate-fluorescein or related probe in conjunction with the appropriate intraoperative viewing optics could facilitate this objective (see Fig. 4).

Second, post-treatment imaging for the recurrence of tumor nodules less than 1 cm in diameter with either computer tomography or ultrasound is often not reliable and consequently of limited value.³⁹ Furthermore, laparoscopy does not generally allow accurate tumor discrimination with current imaging techniques.⁴⁰ Consequently, a patient must frequently undergo a second-look surgery to determine whether a relapse has occurred. In this situation, an optical method that can distinguish cancer from noncancer tissue during an endoscopic exam could prove beneficial. With the appropriate fiber optics and the use of tumor-targeted probes such as folate-fluorescein, endoscopic evaluations of this sort might be possible. Indeed, fluorescent dyes have already been explored for their abilities to reveal tumors by fluorescent detection methods.^{17 41 42 43 44 45 46 47 48 49} While some might raise the concern that a fluorescein probe would lose intensity during extended illumination because of photobleaching,³⁸ the results of this study demonstrate that photobleaching may not constitute a significant problem, at least for exposure times that are less than or equal to 1 h. Furthermore, if such problems were to arise, intermittent illumination or selection of a more photostable fluorophore could easily remedy the problem.

In summary, we have demonstrated the effectiveness of a folate-linked fluorophore for optical imaging of tumors that express the folate receptor. The difference in fluorescence intensity between folate receptor-expressing masses and nonexpressing tissues was sufficiently stark to allow the unambiguous detection of tumors that could not be identified under white light. Metastatic loci smaller than 0.5 mm in size were noticeably fluorescent compared with the background. Furthermore, uptake of the fluorophore into tissues that did not express the folate receptor was too low to cause uncertainty or misdiagnosis of the tumor, even at 2 h postinjection. These results thus raise the possibility that folate can be exploited as a targeting ligand for the visual location and resection of tumor tissue during surgery or endoscopic examination.