In vivo fluorescence detection of ovarian cancer in the NuTu-19 epithelial ovarian cancer animal model using 5-aminolevulinic acid (ALA)

The purpose of this study was to determine whether in vivo fluorescence detection of protoporphyrin IX (PpIX) could be used to identify intraperitoneal micrometastases of epithelial ovarian carcinoma after application of 5-aminolevulinic acid (ALA). ALA was applied intraperitoneal at different concentrations (25, 50, and 100 mg/kg) and iv (100 mg/kg) to immunocompetent Fischer 344 rats bearing a syngeneic epithelial ovarian carcinoma. At different time intervals after ALA administration (1.5, 3, and 6 hr) the peritoneal cavity was illuminated with ultraviolet (uv) light. In vivo fluorescence of PpIX initially was determined by direct visualization. Subsequently ex vivo measurements were made with a slow-scan, thermoelectrically cooled CCD camera. Red in vivo fluorescence was observed in ovarian micrometastases smaller than 0.5 mm in 100% of the ALA-administered animals independent of time interval, drug concentration, or route of administration. The intensity of the fluorescence was concentration dependent as strong fluorescence was consistently found only above 25 mg/kg ALA. Ex vivo tumor to peritoneum fluorescence yield peaked 3 hr after administration of a 100 mg/kg intraperitoneal dose. Direct visualization of in vivo fluorescence after ALA application may improve the detection of intraperitoneal ovarian cancer micrometastases.

tion, the false-negative rate of this procedure has been shown The intensity of the fluorescence was concentration dependent as to be exceedingly high. As many as 50% of patients with strong fluorescence was consistently found only above 25 mg/kg negative second-look laparotomies will experience recurring ALA. Ex vivo tumor to peritoneum fluorescence yield peaked 3 hr disease within 5 years and will ultimately succumb to their after administration of a 100 mg/kg intraperitoneal dose. Direct visualization of in vivo fluorescence after ALA application may disease [7]. Even the most thorough reexploration, emimprove the detection of intraperitoneal ovarian cancer micrometploying multiple pelvic and upper abdominal washings and astases. ᭧ 1997 Academic Press biopsies, does not necessarily result in sensitive, reliable diagnosis of microscopic residual disease [2]. Therefore, a procedure which is minimally invasive, has few side effects, INTRODUCTION and could assist in the diagnosis of microscopic residual disease would be of great value. Ovarian carcinoma is the fourth most frequent cause of cancer-related death in women in the United States [1]. Many Exogenously applied 5-aminolevulinic acid (ALA) results patients with advanced disease (stage III and IV) initially in endogenous production of the potent, fluorescent photorespond to standard therapies, as evidenced by negative secsensitizer protoporphyrin IX (PpIX) [8]. ALA, by itself, is ond-look laparotomies. Unfortunately, up to 50% of these not a photosensitizing agent, but is the precursor of PpIX cases will recur and inevitably result in death [2]. As a in the biosynthetic pathway of heme. Physiologically, the production of heme regulates the synthesis of ALA via a mechanism of negative feedback. Administration of exoge- 1 To whom correspondence should be addressed at Beckman Laser Instinous ALA bypasses this mechanism and induces accumula- tion of PpIX in certain cells [8]. ALA-induced PpIX synthe-122 0090-8258/97 $25.00 Copyright ᭧ 1997 by Academic Press All rights of reproduction in any form reserved.
sis allows selective diagnosis of cancer if (1) the photosensi-a female athymic mouse after injection of F344 ovarian surface epithelial cells that spontaneously underwent malignant tizer is preferentially accumulated by the target tissue (e.g., ovarian cancer), and (2) excitation of the photosensitizer transformation in vitro. When injected into the peritoneal cavity of naive F344 rats, these cells grow in the typical with light at the appropriate wavelength results in PpIX fluorescence that can provide contrast between tumor and nor-fashion of the most common form of human epithelial ovarian carcinoma (i.e., papillary serous adenocarcinoma). This mal tissue. Although substantial work has been performed in porphyrin sensitizers, such as hematoporphyrin derivatives model closely simulates clinical human ovarian cancer by (a) method of intraperitoneal spread; (b) formation of malignant (HPD), especially in the detection of precancerous and cancerous disease of the colon [9], the unique characteristics of ascites, and (c) propensity for local metastases and organ invasion (omentum, peritoneum, liver, bowel). After ip in-ALA-induced PpIX synthesis for selective detection of tumors in vivo is just beginning to be explored [10]. jection of 10 6 cells, tumors appear grossly on the omentum and peritoneum (Week 3) as small nodules (0.5 to 2 mm) It is our hypothesis that selective ALA-induced PpIX production in intraperitoneal (ip) ovarian cancer micrometas-with the animals exhibiting only minimal ascites (õ3 ml).
Thereafter, the tumor nodules continuously grow to form a tases can be achieved because of a higher intratumoral conversion rate of ALA to PpIX compared to the normal sur-confluent omental mass with eventual development of disseminated intraperitoneal carcinomatosis and massive malig-rounding tissues [11]. Proposed mechanisms for the conversion of ALA to PpIX in tumor tissue include: (1) nant ascites by the time the animals die (approximately 60 days). NuTu-19 cells are maintained in complete medium higher intracellular synthesis capacity of PpIX [12]; (2) decrease of key enzymes in the synthetic heme pathway, like consisting of RPMI 1640 (Gibco Life Technologies, Grand Island, NY) with 10% heat-inactivated fetal bovine serum ferrochelatase, which will produce accumulation of converted PpIX [13]; and (3) feedback mechanisms controlling (Gemini Bioproducts, Calabassas, CA) at 37ЊC and 5% CO 2 .
The cell line was expanded and cryopreserved in liquid nitro-the amount of intracellular PpIX [12]. Consequently, ALA can induce direct photosensitization of tumor cells by selec-gen (10 7 cells/vial) after testing negative for mycoplasma contamination with a Mycotest kit (Gibco Life Technolo-tive conversion to PpIX, whereas other photosensitizers localize mainly in the vascular stroma of the tumor [14]. Just gies). Each experiment was performed by thawing a vial of cells and expanding them biweekly to generate the appro-as topical application of ALA has proven to be an effective photosensitizer in cancers of the skin, bronchi, and breast priate cell number.
For these studies, NuTu-19 cells were harvested with [12], it is expected that intraperitoneal ALA administration may lead to a high photosensitizer concentration in ovarian 0.25% trypsin (Gibco Life Technologies), washed twice with Dulbecco's phosphate-buffered saline (PBS, Gibco tumor infiltrating the peritoneal and serosal surfaces. This in turn should enhance visualization of ovarian tumors and Life Technologies), counted for cell number and viability using trypsin blue exclusion, and injected ip into F344 micrometastases by fluorescence.
In this article, we report the feasibility of using exogenous rats at a concentration of 10 6 viable cells/ml PBS (10 6 cells per animal). Cell viability of at least 90% was required ALA to detect abdominal ovarian cancer micrometastases smaller than 0.5 mm by in vivo PpIX fluorescence in a NuTu-for experimental use. After allowing 3 weeks for tumor growth, tumor-bearing rats and sham-treated rats (1 ml 19 epithelial ovarian cancer animal model. PBS ip without tumor cells) underwent induction anesthesia with an intramuscular injection of ketamine (13 mg/ MATERIALS AND METHODS kg) and xylazine (87 mg/kg). Isoflurane and oxygen were Animals provided during surgery for continuous anesthesia. ALA (100, 50, and 25 mg/kg) was injected ip into groups of Pathogen-free Fischer 344 (F344) female rats (109-155 24 rats, respectively. ALA (100 mg/kg) was injected intrag) were obtained from Harlan Sprague Dawley, Inc. (Indiavenously (iv) in the tail vein of three rats. Just prior to napolis, IN) and housed in a pathogen-free animal facility these injections, crystallized 5-aminolevulinic acid hydroat the University of California, Irvine. They were given comchloride (DUSA Pharmaceuticals, Inc., Denville, NJ) was mercial basal diet and water ad libitum. The experimental diluted to 40 mg/ml in sterile water and titrated with 10 protocol (No. 95-1646) for the use of these animals in these N solution of sodium hydroxide to pH 6.5. studies was approved by the Institutional Laboratory Animal Care and Use Committee at U.C. Irvine.

Fluorescence Imaging Epithelial Ovarian Carcinoma Animal Model
In order to closely approximate a clinical laparotomy setting, fluorescence evaluation was performed in vivo. The Studies were performed utilizing the NuTu-19 epithelial ovarian cancer animal model [15]. NuTu-19 was initiated first approach was to acquire fluorescence images in different spectral regions in order to compensate for light distribution from a poorly differentiated adenocarcinoma which arose in inhomogeneities. Since the level of the signal and the sensi-shortpass and 650-nm ({12.5 nm) bandpass filters (Corion Corporation, Holliston, MA), respectively, using 1-sec ac-tivity of the camera required an acquisition time in the order of a few seconds, motion artifacts introduced by respiratory quisition times. Image acquisition, processing, and camera control was performed by a MacIntosh computer with IPLab movements of anesthetized animals and filter changes affected image registration. Therefore, we did not use an im-software (Signal Analytics Corp., Vienna, VA). Tissue fluorescence and light distribution images were recorded sequen-aging system, but instead chose to assess in vivo fluorescence subjectively by direct visualization using a fluorescent disk tially for each sample under identical conditions. Dark-noise images were acquired without the excitation source. Images as an internal standard. The fluorescent polymethylmethacrylate disk (diameter 1 cm, thickness 1.5 mm) was embed-were corrected for both nonuniform illumination and contributing dark noise using the following formula: ded with fluorescent dye (DCM, Eastman Kodak, Rochester, NY) with optical density OD 460 Å 0.025, presenting an emission spectrum close to PpIX (emission maximum near 600 Corrected fluorescence image nm). The in vivo tissue fluorescence was defined by comparing the tissue fluorescence level to that of the disk: 0 Å no Å Image (650 nm) 0 Dark noise (650 nm) Image (500 nm) 0 Dark noise (500 nm) fluorescence; 1 Å positive fluorescence but less than the disk (moderate); 2 Å equal to or stronger than the disk (strong).
On the corrected fluorescence image, the mean fluorescence Ex vivo fluorescence images were subsequently recorded of the peritoneum, the small intestine, and tumor nodules on with a slow-scan, thermoelectrically cooled CCD camera.
the omentum and peritoneum was measured. This permits sensitive, high dynamic-range ratio measurements.
Fluorescence Microscopy

Determination of in Vivo Fluorescence
Imaging of frozen tissue sections, taken at the time of euthanasia, was performed using the same cooled CCD cam-In vivo fluorescence was subjectively determined by direct era and computer system. However, the camera was coupled visualization at different time intervals (1.5, 3, and 6 hr) to an axiovert 10 epifluorescence microscope (Carl Zeiss, after ALA was administered at different drug concentrations Inc., Thornwood, NY) equipped with a 100-W mercury lamp (100, 50, and 25 mg/kg) and administration routes (ip and filtered with a 405-nm ({20 nm) bandpass filter (Omega iv). With continued anesthesia, and with the rat placed in Engineering, Inc., Stanford, CT), to provide excitation at the supine position, a midline laparotomy incision (xyphoid this wavelength, and using a dichorioc mirror FT 440. The to pubic symphysis) was made. The abdominal viscera and emission wavelength was similarly isolated by a 635-nm ({ peritoneum were exposed and fluorescence was induced by 27.5) bandpass filter. Exposure times were 2 sec.  Note. Strong fluorescence was defined as the tissue having the same or greater levels of fluorescence compared to the fluorescent dye-embedded polymethylmethacrylate disk (grade 2) when irradiated with uv light after ip injection of 100, 50, and 25 mg/kg, and after iv injection of 100 mg/kg ALA in tumor-bearing and control animals. and excitation with the Woods lamp. Confirmation that the conversion scattered homogenously throughout the tumor fluorescent nodules were tumor was made by microscopic tissue but not in adjacent normal tissue. At the tumor edge examination of tissues stained with hematoxylin and eosin it appears that the normal peritoneum is exhibiting ALA-(H&E). After ALA application, visible fluorescence of induced fluorescence (boxed area in Fig. 3B). However, miomental and peritoneal tumor nodules was observed in croscopic analysis of this region (Fig. 3C) shows a thin 100% of tumor-bearing animals for all time points, drug layer of infiltrating tumor corresponding with this area of concentrations, and both administration routes. Fluoresfluorescence. This would suggest a high specificity of the cence could be easily observed in nodules smaller than ALA for even small amounts of tumor volume. 0.5 mm. In approximately 70% of tumor-bearing animals, Figure 4 shows a cross-sectional view of small bowel in the intensity of fluorescence was strong compared to the a tumor-bearing animal treated with ip ALA. Strong fluocontrol dye disk (grade 2), whereas the remaining 30% rescence is noted in the intraluminal feces, and immediately showed moderate, but easily visualized fluorescence adjacent to the mucosal surface lining cells (Fig. 4B). How-(grade 1). In contrast, normal tissues adjacent to tumor ever, the mucosa, submucosa, and muscular layers are withnodules did not fluoresce (grade 0) and no fluorescence out fluorescence. In the cross-sectional images of small of the tumor nodules could be detected in any animals bowel from an animal not given ALA (Fig. 5), weak auwithout administration of ALA. Table 1 summarizes in tofluorescence can also be detected in the intraluminal feces vivo fluorescence evaluation (grading) of omental and (Fig. 5B) in a pattern similar to, but less intense than the peritoneal tumor nodules compared to the fluorescence pattern seen in the ALA-treated animal (Fig. 4B). Interestintensity of the internal standard dye disk. Utilizing a ingly, fluorescence can be seen on the serosal surface of this 100 mg/kg ip ALA dose, no significant differences in the ALA-treated tumor bearing animal (Fig. 4B). After examinpercentage of animals exhibiting fluorescence could be ing the tissue under high magnification with H & E staining, seen at 1.5, 3, and 6 hr. Intravenous administration of ALA this fluorescence was found to be secondary to a superficial (100 mg/kg) also did not appear to result in differences in layer of NuTu-19 ovarian carcinoma cells. fluorescence intensity or distribution when compared to ip administration. A dose reduction to 50 mg/kg ip showed the same strong tumor fluorescence 3 hr after administra-Determination of ex Vivo Fluorescence tion, whereas 25 mg/kg showed a generalized reduced level of fluorescence (grade 1).
Tumor and peritoneal fluorescence was measured ex vivo using the cooled CCD camera. The results shown in Fig. 6 Fluorescence and Light Micrographs indicate that the tumor fluorescence yield peaks 3 hours after ip ALA administration. The differences in this fluorescence Figure 3 shows light micrographs (Fig. 3A) and fluoresbetween time points were only significant for the 3-hr time cence (Fig. 3B) of 5-mm-thick frozen tissue sections 6 hr interval (ANOVA, Fisher's PLSD, P õ 0.05). The ratio after administration of ALA at 100 mg/kg ip. A peritoneal between omental tumor nodules and peritoneum was approxtumor nodule (size õ0.5 mm) and the underlying musculature can be seen. Fluorescence images suggest specific PpIX imately 7 (range 4-10) [data not shown].

DISCUSSION
Selective ALA-induced PpIX conversion in ovarian cancers may lead to two new procedures; (1) it may provide an in vivo diagnostic aid to improve visualization of micrometastatic ovarian cancer spread in the peritoneal cavity, and (2) it may serve as adjuvant photodynamic therapy for the treatment of micrometastatic (small volume) intraperitoneal disease.
In our study, ALA administration was associated with easily detectable in vivo protoporphyrin fluorescence of peritoneal, omental, and serosal micrometastatic ovarian cancer nodules. Strong localized red fluorescence of the ovarian cancer nodules could be observed, compared to the surrounding normal tissue, by illuminating the peritoneal cavity of ALA-treated rats with uv light. The apparent tumor tissue selectivity was likely due to a higher conversion rate of ALA to PpIX by the tumor cells. This observation, in and of itself, is important in that tumor nodules as small as 0.5 mm are extremely difficult to detect by visual means alone. The significance of this finding is that for large surface-area structures, such as the human peritoneal cavity, fluorescence visualization of micrometastatic ovarian cancer implants may enhance our ability to properly stage patients during primary laparotomies or diagnose persistant minimal residual disease during second-look laparotomies. Whereas 100% of the animals exhibited fluorescence of the ovarian cancer nodules after ALA administration, a variation in fluorescence levels was observed (Table 1). This may be due to differences in metabolic rate of ovarian cancer cells, which results in different conversion rate of PpIX or this may be due to quantum yield effects.
What this study did not investigate is the ability to detect retroperitoneal disease; which occurs in 12% of patients with recurrent ovarian cancer after a negative second-look laparotomy [16].
The fluorescence yield at varying time intervals after ALA administration was an interesting finding in this study. PpIX was sufficiently abundant to yield strong fluorescence in as little as 1.5 hr after administration and this fluorescence was still present at 6 hr. This observation is consistent with porphyrin tissue extraction data of carcinoma in tumor-bearing rats, where the decline of tissue porphyrin was observed between 5 and 12 hr after iv ALA administration [11]. As a result of the presented fluorescence data, we postulate that the early fluorescence phase may be secondary to the selective rapid uptake of ALA and its conversion to PpIX in tumor tissue. This early phase is then followed by a second perforations were seen in three of these patients. In a rat study utilizing the photosensitizers photofrin or mesotetrahydroxyphenylchlorin (mTHPC), intestinal organs were the most photosensitive intraabdominal structures and intestinal perforation was the most common cause of death after PDT [23]. Therefore, PDT of the peritoneal cavity may be feasible if the intestinal tract can be protected during treatment. In our study presented here, no significant conversion of PpIX occurred in the submucosal and muscular layers of the intestine, as evidenced by minimal fluorescence. Because of this, ALA may have substantialy less intestinal damaging effects compared to previously utilized photosensitizers. Indeed, homogenous distribution of PpIX in tumors was observed, sug- be successful. The fact that microscopic tumor on the serosal surface of the peritoneum (Fig. 3B) and of the small bowel (Fig. 4B) showed photosensitizer conversion may lead to the duodenum and jejunum and feces from the point of entry of the common bile duct to the distal part of the jejunum. improvement of adjuvant treatment of intraperitoneal micrometastatic disease. Studies on small bowel tissue sections revealed that this fluorescence was due to intraluminal fecal material, and no In summary, as many as 50% of Stage III and Stage IV epithelial ovarian cancer patients who have undergone fluorescence was noted in small bowel mucosa, submucosa, or the muscular layers. These findings could be accounted negative second-look laparotomies will experience subsequent intraabdominal recurrence of their disease [2]. This for by porphyrin excretion in the bile followed by normal bowel motility and by the fluorescent feces which can be fact suggests that even a very thorough exploration does not reveal micrometastatic residual disease in many patients. visualized through the thin-walled small bowel of the rat. We would expect that the thicker jejunum in humans (ap-Considering the large surface area of the human peritoneal cavity, increasing the diagnostic efficiency may be best proximately 5 mm) would prevent this transluminal visualization of fluorescent fecal material. achieved using fluorescence-based visualization techniques. Using this approach we were able to detect tumor nodules In this study, ip application of ALA was comparable to the iv route for in vivo fluorescence detection. Because ALA as small as 0.5 mm in an experimental animal model. Second-look operations by laparoscopy may significantly benefit is a small hydrophilic molecule which diffuses easily through tissue, plasma equilibrium occurs quickly. Dose-response from fluorescence detection, in the same way bladder cancer diagnosis was aided by cytoscopy [10]. Additionally, selec-studies indicated an optimal dose of 50 mg/kg in that fluorescence was not increased with 100 mg/kg ip at 3 hr, and tive and homogenous distribution of PpIX in ovarian cancer tumors predispose ALA as a good candidate for PDT of was weaker with a dosage of 25 mg/kg. This finding has clinical relevance since the ALA dosage (50 mg/kg) utilized the peritoneal cavity. In conclusion, specific detection of rat ovarian cancer could be achieved with administration of a in this study has already been shown to be safe in human clinial trials [17][18][19].
nontoxic dose of ALA. This approach may be a promising method for improving the specificity and sensitivity of ovar-Colonic and oral cancerous lesions have been shown to produce PpIX after ALA application, resulting in (1) selec-ian cancer tumor staging. In addition, systemic ALA could eventually be coupled with appropriate light delivery and tive fluorescence as compared to normal tissue and (2) good response to PDT [14,17,18]. In this animal model of epithe-laparoscopic instruments in order to theoretically develop minimally invasive treatment or screening tools for ovarian lial ovarian cancer, selective ALA conversion resulted in high tumor nodule fluorescence compared to the peritoneum cancer micrometastases. (ratio of 5) but not to the small intestine (ratio of 1). However, it should be noted that fluorescence distribution in the ACKNOWLEDGMENTS small bowel appeared to be limited to intraluminal feces.