Fiber optics quickly detect cervical malignancies

Houston, TX--The waiting is the hardest part. After a woman has a Pap smear, she waits two weeks for the result. If the test shows any abnormalities, she has a second Pap smear and starts waiting again. A second abnormal result means a microscopic examination of the cervix along with a biopsy, then waiting several more weeks for biopsy results.

The waiting routine may soon be over. A fiber-optic device now being tested at the University of Texas M. D. Anderson Cancer Center in Houston provides immediate and noninvasive detection of precancerous cervical tissue. The device relies on differences in the way that normal and abnormal cervical tissues absorb and emit light to detect cellular abnormalities, which can progress to cancer if left untreated.

Researchers envision using the technology as a follow-up to an abnormal Pap smear. A clinician could immediately determine the probability of precancerous cells, and doctor and patient could then discuss and even begin treatment.

Pap smear problems. In the U.S., 15,800 new cases of invasive cervical cancer occurred in 1995, and 4,800 women died of the disease, according to statistics from the American Cancer Society. Young women are especially vulnerable.

Each year, more than 50 million U.S. women have Pap smears to screen for cervical cancer. The test involves taking a sample of cells from the cervix--the outer end of the uterus--and sending it to a lab for analysis. Yet the disease remains the second leading cause of cancer deaths in women worldwide.

One reason is that the Pap smear either misses precancers or wrongly identifies them over 25% of the time. Part of the problem is due to sampling errors. Because the test involves scraping surface cells, only parts of the cervix get analyzed, and the lower cell layer--where abnormal cells first crop up--gets missed altogether.

Another factor is that most cancers develop in the cervix's transition zone, where one type of cell turns into another. This region is difficult to examine via Pap smears--picture it as the inside of a bagel.

But perhaps the leading reason for false Pap smear readings is because pathologists have a tough time precisely differentiating among cells that are normal, inflamed, and cancerous or precancerous. This is the same reason that microscopic cervical examination, called colposcopy, is also subject to error.

A better way. Rebecca Richards-Kortum, an electrical engineering professor at the University of Texas at Austin, developed the concept for the fiber-optic probe. She teamed up with Dr. Michelle Follen-Mitchell of the Department of Gynecology at the M.D. Anderson Cancer Center to do the clinical studies. LifeSpex Inc. was formed in Houston to commercialize the technology. The company contacted Sandia National Laboratories, Albuquerque, NM, late in 1993 for help refining the prototype instruments.

Richards-Kortum's research focused on discovering the differences between how normal and abnormal cervical tissues absorb and emit light. Shining light of certain wavelengths on the cervix causes the tissue to fluoresce. Fluorescence indicates the concentration of fluorophores and cell structure. Compared with normal tissue, precancers have a different cell structure, more blood vessels, and an increase in certain molecules such as collagen, elastin, and keratin.

Using data from hundreds of patients, Richards-Kortum developed a computer algorithm to take the fluorescence results and determine the probability that the tissue being examined is precancerous.

"Basically, we use software code to do pattern recognition. You take the spectrum that you've recorded and compare it to your known normal and abnormal data to figure out whether the new spectrum looks more like a normal or an abnormal," explains Richards-Kortum. "It comes out as a probability that the tissue is precancer and, if it is, if it's high-grade or low-grade precancer." The fluorescence algorithm lets her find precancers with a technique that's more sensitive than the screening Pap smear, she says.

Cadillac system. The first instruments to implement the technology were designed for research--not the local clinic. What Richards-Kortum calls her "Cadillac system" comprises a nitrogen laser, complete spectrograph, and a computer.

Glass fibers carefully coupled to the laser deliver light to the cervix via a probe. Once the tissue fluoresces, the probe collects the light. The light bounces down nine collection fibers to a spectrograph, which disperse it as a function of wavelength. Then an array of detectors capture intensity as a function of wavelength. The computer runs the spectrum through the fluorescence algorithm to determine the probability of precancer.

This system takes the whole spectrum in a short amount of time, but it's expensive, not very portable, and not terribly reliable. At this point, LifeSpex had licensed the technology and in late 1993 contacted Sandia National Laboratories, for technical assistance to develop a version of the prototype that could be manufactured and used on a wide scale.

"Imaging and spectral analysis are integral parts of Sandia's satellite and weapons nonproliferation work," says David Sandison, who leads the engineering team that's making the optical technology "commercializable." The team delivered the first prototype to LifeSpex in September 1994, and the two firms are currently at work on a preproduction model.

Engineering changes. The first change Sandia made was to replace the laser with a xenon lamp. The lamp was more reliable and easier to maintain.

"Illuminating the tissue then collecting the light was a challenge because the signal was so weak," says Bob Wilcox, director of engineering at LifeSpex. They needed a high-efficiency, low-cost collector to get the light back from the tissue at the new wavelength.

To address this challenge, Sandia used fiber-optic cable comprising thousands of fibers instead of the Cadillac system's 7 to 60 fibers. This improved efficiency and reliability and made coupling the light source to the fibers easier.

Next to go was the spectrograph. Instead, the team used a Micro Mo motor to rotate a filter wheel. The filter wheel sits in front of a 512x64-element CCD (charge-coupled device) detector from Hammamatsu. Fluorescent light from the cervix passes through three different filters so the CCD can measure light intensity at three different wavelengths.

A National Instruments AT-MIO-16X digitizing card digitizes the CCD signal. Then the data goes to a laptop computer to run through the algorithm to determine precancer probability. (The production model will incorporate a microprocessor in place of an entire laptop.) The card also provides the inputs and outputs the computer needs to control the system.

Sandison's team used National Instruments' LabView virtual instrumentation software to program the system. "LabView allowed us to write bug-free code faster than I have ever seen an instrument brought up," he says. "That saved us a lot of time."

Ending the wait. Officials at LifeSpex say FDA testing could begin early next year and a production model could be in doctors' offices and clinics within three years.

The optical probe could improve the use of colposcopy and allow for faster, more effective diagnosis and treatment. Better diagnosis and faster treatment would ultimately save lives.

Because the technique is nondestructive, a doctor or nurse can sample a much greater area of the cervix than is possible with a biopsy. The probe could also decrease the number of biopsies--a crucial benefit because biopsies can affect fertility, and many women who require them are still in their childbearing years.

"We've been using time as a surrogate for certainty," says LifeSpex President Bruce Bach. "This light technology gives more certainty and lets doctors together with their patients make better care decisions."