DISCOVERY v15n3 Detecting Cervical Cancer

Tissue optical spectroscopy is a new area in the field of medical diagnostics that has the potential to provide automated, cost-effective screening for histo-chemical features of disease. Our group of biomedical engineers from The University of Texas in Austin and gynecologic oncologists, pathologists, cell biologists, and decision scientists from The University of Texas M. D. Anderson Cancer Center are developing new optical diagnostic technologies to reduce the incidence and mortality of cervical cancer.

Optical spectroscopy has long been an important tool in analytical chemistry. Fluorescence spectroscopy is routinely used to determine analyte concentrations and monitor chemical kinetics in complex reactions. No other nondestructive means available at comparable cost can equal optical spectroscopy in sensitivity. Concentrations of metabolites as low as 10 ^-9 M are routinely determined. The emergence of fiber optic technology has enabled remote spectroscopic sensing and monitoring systems. This, coupled with the large number of biologically important molecules with distinct optical spectra, has led to increasing interest in development of fiber optic probes to detect and monitor disease.

Historical Background. The interaction of light with tissue has been used qualitatively by physicians to aid in detection of disease since the mid-1800s. In fact, optical screening and detection methods for pre-cancer and cancer of the uterine cervix provide an excellent example of optical diagnosis based upon biochemistry and structure. Increased vascularity associated with cervical cancer was first noted in women who had died of cervical cancer. The ability to "see" subsurface vessels is based on differences in the reflection of illuminating white light. The vessels contain hemoglobin, which has a high absorption coefficient compared to other tissue constituents and thus appear as dark lines on the pink surface of the cervix.

In 1925 a low-power microscope, called a colposcope, was used to visualize these changes in the living patient. The colposcope showed other changes that occurred in cervical pre-cancer and cancer: normal epithelium has a smooth, uniform appearance whereas abnormal epithelium can look like a mosaic tile pattern. This abnormality was believed to originate due to changes in the shape and scattering of abnormal cells. In the 1940s cancer researchers suggested that abnormal cells exfoliated from the cervix could be stained and observed under high-power microscopes to screen for cellular atypia associated with cervical pre-cancer and cancer.

Today the Pap smear is used to screen the general female population for cervical pre-cancer and cancer in the United States, and an abnormal Pap smear is followed by colposcopy, with a biopsy to confirm whether pre-cancer or cancer is present in any suspicious lesions. The adoption of these qualitative optical screening and detection protocols has dramatically reduced the incidence and mortality of cervical cancer in the United States. Yet the technology still fails in some cases, and in 1995, 15,800 new cases of cervical cancer were reported and 4,800 women died of the disease in the United States. The costs associated with detecting and treating cervical pre-cancer are significant. Thus accurate cost-effective screening and detection methods for cervical pre-cancer are needed.

Optical spectra can be recorded remotely, in near real time without the need for tissue removal and data analysis can be automated. Optical diagnosis affords many important advantages over traditional techniques, including the potential to reduce the need for clinical expertise and to reduce the number of unnecessary biopsies to enable combined diagnosis and therapy in those patients who might benefit most and the potential to reduce health care costs.

Tissue Spectroscopy. Figure 1 shows a diagram of a system to measure tissue optical spectra. A fiber optic probe provides illumination light at illumination wavelength (or color). The light impinges on the tissue where it interacts; light remitted through the tissue front surface following interaction is collected and sent to a sensitive detector. In fluorescence spectroscopy, the photon wavelength changes in the interaction, and the intensity of light emitted from the tissue surface is detected at the emission wavelength. A fluorescence emission spectrum is a plot of the intensity of fluorescent light as a function of emission wavelength, produced when the tissue is illuminated at a particular wavelength.

The light-tissue interactions between illumination and measurement determine the shape and intensity of the resulting spectrum. A spectroscopist views tissue as a collection of chromophores, or constituents that interact with light in some manner. Three types of interaction are important: absorption, scattering, and fluorescence. Chromophores are classified as absorbers, scatterers, and/or fluorophores. In a scattering interaction, the direction of light travel is changed due to microscopic fluctuations in the tissue index of refraction, but the light intensity remains the same. Absorption interactions reduce the intensity of the light as photon energy is transferred to the absorbing molecule. Fluorescence can occur following absorption, as the chromophore releases some of this absorbed energy in the form of fluorescent light at the emission wavelength.

Most tissues are highly scattering, with the probability of a scattering interaction exceeding that of absorption by at least an order of magnitude. Most tissues are weakly fluorescent, with a small probability that fluorescence will occur following absorption. Thus, when light impinges on tissue, it typically scatters and absorption and perhaps fluorescence can occur. Further scattering and absorption can occur before light exits the tissue surface where it can be detected. Fluorescence spectra contain information about the wavelength dependent fluorescence properties of tissue as well as the wavelength dependent scattering and absorption properties.

In order to obtain biochemical information from a re-emitted optical signal, a chromophore is required for interrogation. Endogenous chromophores, such as oxy- and deoxy-hemoglobin, melanin, myoglobin, and water are primarily responsible for absorption of light in tissue. In the ultra-violet and near ultra-violet, the fluorescent re-emission from chromophores such as NAD(P)H, flavins, and porphyrins may provide diagnostic information. Autofluorescence has also been noted in the structural proteins, collagen and elastin.

Diagnosis Using Spectroscopy. The Richards-Kortum laboratory at UT Austin and the Follen laboratory at UT M. D. Anderson Cancer Center have collaborated for the past eight years to develop optical techniques to address the limitations of the Pap smear and colposcopy. In these studies, a fiber optic probe is placed on the cervix and a reading of the fluorescence spectrum is made in seconds. Initially, we measured fluorescence of normal and abnormal cervical biopsies in vitro to determine the optimal wavelengths for diagnosis. Further in vivo studies were carried out at 337, 380, and 460 nm excitation based on these studies. Results indicate that the fluorescence intensity and lineshape vary substantially from patient to patient; despite this variability, if spectra are obtained at multiple excitation wavelengths, good discrimination between normal cervix and pre-cancer can be achieved.

We have developed a statistical method of algorithm development to extract the spectral information that is most diagnostically relevant. We measured spectra from 500 cervical sites in ninety-five patients in vivo at three excitation wavelengths. The data were divided randomly into a training set, used to develop a diagnostic algorithm and a validation set used to test the algorithm. The performance of the fluorescence-based algorithm exceeds that of the Pap smear in many clinical sites and is comparable to that of colposcopy in expert hands. In addition to providing accurate diagnosis, fluorescence measurements can be made in near real time, and results are available immediately.

Since fluorescence measurement is not painful and does not require tissue removal, the entire epithelial volume can be interrogated, potentially reducing sampling error. Analysis of fluorescence data is accomplished using algorithms implemented in software, so the need for clinical expertise is reduced. In the screening setting, the performance of fluorescence exceeds that of the Pap smear, and the same advantages of reduced sampling error, immediate results, and reduced clinical expertise are maintained. The safety of fluorescence spectroscopy is comparable to that of colposcopy, the current standard of care.

We are currently conducting a preliminary analysis of how biologic variables such as age and human papilloma virus infection affect the algorithm. Large randomized clinical trials will be needed to establish the potential of fluorescence spectroscopy to improve or replace colposcopy for diagnosis and to improve or replace the Pap smear for screening for cervical abnormalities. The use of fluorescence spectroscopy is advancing medical knowledge of the biology of cervical neoplasia and is effective and safe.

Is It Cost Effective? As part of a project funded by the National Science Foundation and The Whitaker Foundation, we analyzed the costs and benefits of the diagnosis and management of cervical SIL [Squamous Intra-Epithelial Lesion] in the diagnostic setting. We considered five clinical strategies for the diagnosis and management of cervical pre-cancer:

First, colposcopically-directed biopsy at a first visit followed by treatment with loop electrical excision procedure at a second visit if high-grade SIL is found at biopsy. This is the standard of care.

Second, "see and treat" based on colposcopic diagnosis and treatment with loop electrical excision procedure in one visit. "See and treat" means that if high-grade SIL is suspected at time of colposcopic evaluation, the patient will immediately be treated rather than asked to wait one to two weeks for confirmatory biopsy results. Thus this treatment is based on the initial colposcopic impression, and no biopsy is performed.

Third, spectroscopically-directed biopsy followed by treatment with loop electrical excision procedure at a second visit if high-grade SIL is found at biopsy.

Fourth, "see and treat" fluorescence spectroscopy based on diagnosis and treatment with loop electrical excision procedure in one visit. Similar to the "see and treat colposcopy" strategy described above, this strategy implies that if high-grade SIL is suspected at time of spectroscopy, then the patient will be immediately treated rather than asked to wait one to two weeks for biopsy results and make an additional office visit.

Fifth, "see and treat" fluorescence spectroscopy and colposcopy used together and treatment with loop electrical excision procedure in one visit. In this strategy, if both spectroscopy and colposcopy at the time of first visit indicate high-grade SIL, then the patient is immediately treated. If both spectroscopy and colposcopy indicate no lesions, then the patient is not treated. If the results of spectroscopy and colposcopy conflict, then a biopsy is performed; in that case, the decision to treat with the loop electrical excision procedure (at a second visit) is based on the results of the biopsy.

Given that randomized clinical trials have not been performed comparing these five strategies, we performed a cost-effectiveness analysis to determine the possible outcomes of the diagnosis and management of cervical pre-cancer. Using a decision-analytic model, we simulated the outcomes that would result from these strategies by incorporating the best available data both from the medical literature as well as data collected in our clinic at M. D. Anderson Cancer Center.

We documented from a health care perspective the expected costs to be incurred for each of the clinical strategies for the diagnosis and management of cervical pre-cancer. Our institution has established a clinical pathway for the diagnosis and treatment of cervical squamous intraepithelial lesions beginning with colposcopy at time of the referral visit; the other strategies examined in the analysis were modeled after this pathway. Costs were determined from our hospitalís information system.

Interestingly, we found that the most expensive strategy was the standard of care, colposcopically-directed biopsy. When fluorescence spectroscopy was combined with colposcopy in a see-and-treat modality, slightly more cases were found at a lower cost. We estimate that a see-and-treat strategy based on the combination of fluorescence spectroscopy and colposcopy could save more than $625 million annually in the United States. Sensitivity analysis showed that the results of the analysis were sensitive to the cost and performance of fluorescence.

Summary. Based on this research, we believe that fluorescence spectroscopy could significantly impact care in the U.S. by decreasing costs, because less biopsies would be performed and fewer unnecessary treatments would occur. Fluorescence spectroscopy could significantly impact care worldwide by providing automated diagnosis at a screening visit without requiring health care workers to be trained in the difficult skill of colposcopy.

There are other important neoplasms of the female genital tract where quantitative optical techniques offer great promise. Ovarian cancer is the fourth leading cause of cancer-related deaths in women in the United States and is an ideal neoplasm to diagnose early. Patients with Stage I disease have a five-year survival of 80 percent to 90 percent. However, 70 percent of women will be diagnosed with advanced disease; their five-year survival will, at best, be 15 percent to 30 percent.

Therapy of women with advanced ovarian cancer usually requires multiple surgical procedures and multiple courses of chemotherapy, resulting in significant morbidity and health care costs. Because of the high mortality associated with ovarian cancer, there is widespread interest in developing a screening test for early disease; yet such tests have proven disappointing. Identifying women at increased risk for ovarian cancer without having a cost-effective and highly predictive screening modality is frustrating and unfair to this already anxious group.

Recently, office laparoscopic procedures have been developed to visualize the ovaries in a simple outpatient procedure, providing a unique opportunity to design a tool to better evaluate early ovarian disease. However, the sensitivity of visual observation of diffusely reflected light is not sufficient to detect early ovarian disease. Our group has obtained data which indicate that quantitative optical spectroscopy has promise to detect ovarian cancers at an early stage, potentially providing a new, accurate screening tool where none has been previously available. The focus of our continued collaboration will be to develop cost-effective early detection tools for cervical and ovarian cancer. d

Acknowledgments: Co-authors for this article are Dr. Urs Utzinger at The University of Texas at Austin and Dr. Scott B. Cantor and Dr. Molly Brewer at The University of Texas M. D. Anderson Cancer Center.

Dr. Rebecca Richards-Kortum is professor of electrical and computer engineering at The University of Texas at Austin. She is the 1991 recipient of the National Science Foundation Presidential Young Investigator Award for research in optical spectroscopy, and in 1992 she received the Presidential Faculty Fellow Award from the National Science Foundation. Her research focuses on the application of optical spectroscopy for detection of pre-cancer and other diseases. Her research group is currently developing fluorescence based techniques for the diagnosis of cervical pre-cancer in vivo. Dr. Richards-Kortum received a BS degree in physics and mathematics from the University of Nebraska and earned her MS and Ph.D. degrees in medical physics from Massachusetts Institute of Technology. She may be reached at 512-471-2104 or by e-mail at kortum@ece.utexas.edu.

Dr. Michele Follen is professor of gynecologic oncology and director of colposcopy at The University of Texas M.D. Anderson Cancer Center and the University of Texas Health Science Center. She has developed a diverse program addressing patient care, teaching, and research in cervical cancer prevention, and her multidisciplinary research and clinical team is setting a standard in clinical cancer prevention. Major national and international recognition for her research has come in many forms: recipient of the Hunter Award from the American Obstetrical and Gynecological Society, the American Cancer Societyís Science Writers award, and a teaching award from the American Professors of Obstetrics and Gynecology. She is the author and editor of a book on gynecologic cancer prevention. Dr. Follen may be reached by e-mail at mfollen@MDanderson.org.


	HOME \|	NEWS \|	FACTS \|	ON CAMPUS \|	DISCOVERY \|	ABOUT O P A

February 9, 2000 Comments or suggestions to utopa@www.utexas.edu Copyright © 2000 Office of Public Affairs at The University of Texas at Austin