Read Summary

HONOLULU — A novel handheld device that incorporates artificial intelligence detects high-risk human papillomavirus (HPV) strains 16 and 18 within 2 hours and performs as well as biopsy, which can require 2 weeks for results to be returned.

Another new approach identifies these high-risk HPV strains in urine samples.

Both of these approaches would be useful in countries with limited resources, say the researchers, who presented data here at the Society of Gynecologic Cancers (SGO) Annual Meeting 2019.

Teresa Diaz-Montes, MD, Mercy Medical Center, Baltimore, Maryland, who chaired the session, commented that it was “extremely important” to identify high-risk HPV in underserviced areas because of the associated risk for cervical cancer.

“If we are able to detect HPV, we can prevent cervical cancer by treating precancerous lesions,” she elaborated.

Furthermore, women in resource-limited countries who undergo Pap testing likely have to wait weeks to receive definitive results, and they often have difficulty returning to the testing center, she told Medscape Medical News.

“In low-income countries, waiting 2 weeks or even longer for results means that you may lose opportunities to treat women, because many of them won’t come back,” Diaz-Montes emphasized.

“So it’s extremely important that we have methods to detect HPV infection that are cheaper and more accessible [than standard tests], particularly for women who live in resource-poor countries,” she concluded.

Handheld Novel Detection Device

The novel device was developed by Castro Cesar, MD, assistant professor of medicine and director of the Massachusetts General Hospital’s Center for Systems Biology, and colleagues at Harvard Medical School, Boston.

“People have been motivated to use HPV testing in resource-rich countries, but we are talking about PCR [polymerase chain reaction] and [the need for] large equipment in centralized laboratories,” Cesar told Medscape Medical News. Many countries cannot afford such equipment. Moreover, many do not have centralized laboratories and must carry out the testing at the point where women are screened, he added.

So his team set out to develop a device that could help even nonmedical people detect HPV at the point where patients access care.

“And in all cases, our readouts from [cervical] brushings performed as well as standardized biopsies for identifying HPV subtypes, so this strongly illustrates the potential of this type of work where we can sustain an ‘organization memory’ in the community where it is tested and where it can be performed by locals,” he added.

The new device uses holography — also known as diffraction — which essentially reflects the way in which light strikes a particle. The extent of diffraction is a function of the size, shape, and density of the particle, he explained.

“This module has a little light source that shines light on a specimen to create a holographic image and then, through software, reconstructs the image into something decipherable,” Cesar explained.

To maintain the small “footprint” required of any technology intended for use in areas with limited resources, the researchers had to develop a device that could be operated without extensive and expensive “cloud” or wireless technology.

This they did by “training” software on tens of thousands of holographic images, which enabled the device to recognize specific holographic features of HPV 16 and, later, HPV 18 through artificial intelligence algorithms.

“After several rounds of training, there was good performance from an HPV standpoint,” Cesar noted.

As importantly, the processing time required to read these images was no more than 2 hours, he added.

With the trained software uploaded into the device, the researchers next compared readouts from cervical brushings with results from cervical biopsies. They found “similar performance rates with HPV 16 and HPV 18, which was reassuring,” Cesar said.

Most recently, the group carried out a more formal study in which they compared holographic images of cervical brushings from 28 women referred to the colposcopy clinic at the Massachusetts General Hospital to findings obtained on the Cobas HPV test, the local gold standard.

In this small sample of women, they found 100% concordance rates between their holographic analyses and the gold-standard HPV test.

“Our rapid and accessible HPV platform performs as well as its costlier peers,” Cesar concluded.

“And by decentralizing diagnoses through point-of-care operation and artificial intelligence algorithms, our work is positioned to partner with rural and global communities where effective HPV tactics remain unmet needs,” he added.

Detecting HPV in Urine Samples

Another approach to detecting HPV was described at the same meeting by Vuthinun Achariyapota, MD, from the Siriraj Hospital in Bangkok, Thailand. This approach involves analyzing urinary and cervical samples to detect high-grade squamous intraepithelial lesions (HSILs) from women with abnormal Pap test results.

A total of 96 women were recruited for the study.

Before the participants underwent colposcopy, the researchers collected a urine sample and a cervical sample, which were then sent for DNA testing, Achariyapota explained.

Colposocopy was subsequently carried out. Pap test results showed that almost 85% of the samples contained high-grade, potentially carcinogenic HPV lesions.

Urine and cervical samples were subsequently analyzed for the same high-grade HPV types.

Carcinogenic HPV types were found in 73% of urine samples and in 81% of samples from the cervix, Achariyapota noted.

Urinary and cervical samples performed similarly in detecting histologically confirmed HSILs.

Table. Urinary and Cervical High-Risk HPV Detection

  Sensitivity Specificity Accuracy
Urine samples 86.2% 47.4% 70.8%
Cervical samples 94.8% 39.5% 72.9%

Patient satisfaction and the acceptability of collecting samples were higher among women who provided urine samples than among women who provided cervical samples, at 99% vs 61.5%.

From this finding, Achariyapota and colleagues suggested that patients would be more likely to favor urine sampling for HPV analyses.

The Bangkok group stressed that participants had to provide a first-void urine sample and that they could not engage in sex for at least 2 days prior to the sample collection.

Achariyapota concluded that detection of high-risk HPV in urine samples is both “easy and feasible.”

He also pointed out that there was no need for pelvic examination, and as such, real-time PCR analysis of urine samples represents an alternative for women who fear or reject a pelvic exam.

Cesar is co-inventor of a diffraction-based imaging method used in the study device. Neither Achariyapota nor Diaz-Montes have disclosed any relevant financial relationships.

Society of Gynecologic Cancers (SGO) Annual Meeting 2019: Abstracts 75 and 76. Presented March 15, 2019.

Print Friendly, PDF & Email