Electro-Optical Sciences (EOS; Irvington, New York) has reported positive top-line results of its pivotal trial of MelaFind, a non-invasive, point-of-care device intended to assist in the early detection of melanoma.

The blinded study, conducted at seven U.S. centers, included 1,831 pigmented skin lesions from 1,383 patients, making this the largest prospective study ever conducted in melanoma detection, according to the company. EOS said it is working to complete its pre-market approval application and expects to file with the FDA soon.

MelaFind features a hand-held imaging device that emits light of multiple wavelengths to capture images of suspicious pigmented skin lesions and extract data. Using sophisticated algorithms, the data are then analyzed against a proprietary database of melanomas and benign lesions in order to provide information to assist in the determination of whether the lesion should be biopsied.

Prior to the start of the study, EOS and the FDA entered into a binding protocol agreement to stipulate the sensitivity and specificity endpoints that should be used to determine the safety and effectiveness of MelaFind, the company said.

MelaFind detected 112 of 114 (98% sensitivity; lower confidence bound of 95%) melanomas that were eligible and evaluable for primary sensitivity endpoint analysis, and 125 of 127 (98% sensitivity; lower confidence bound greater than 95%) melanomas overall. The protocol agreement calls for sensitivity endpoints of greater than 95% lower confidence bound.

According to the company, MelaFind's specificity, the ability to accurately rule out disease, was significantly superior (9.5%) to that of the study dermatologists (3.7%), who are skin cancer experts (p-value less than 0.02). The protocol agreement calls for MelaFind to be more specific than the study physicians at a p-value of less than 0.05.

Today, physicians primarily rely on surface analysis of the skin and a method known as the ABCs of melanoma, a mnemonic for recognizing moles and growths that might be cancerous by evaluating its asymmetry, border, color, diameter and elevation.

Joseph Gulfo, president/CEO of EOS, told Diagnostics & Imaging Week that the doctors who came up with the ABCs of melanoma challenged the company's founders to develop a computer vision system that could improve early detection of the deadly skin disease. That was 11 years ago, he said.

In addition to the company being born out of a medical need, knowing that one American an hour dies of melanoma – and knowing that it's virtually 100% curable if caught early – makes having an impact on this disease something that is "near and dear to our hearts," Gulfo said.

"The medical need of this disease is something that energizes us, it is something the project was born out of, these results mean that MelaFind is ultra, ultra sensitive . . . much more sensitive than other methods," he said.

Gulfo also said it was impressive how difficult the lesions in the trial were. Although the company is still waiting on final statistics, he said he expects "very impressive numbers" on the high percentage of lesions that met several of the ABC criteria.

"MelaFind appears to be an excellent tool to help detect melanoma at the earliest, most treatable stage," said Gary Monheit, MD, associate clinical professor of dermatology at the University of Alabama-Birmingham and the lead investigator for the MelaFind pivotal trial. "With no cure for late-stage melanoma, early detection is our best defense against this cancer, which has reached epidemic proportions."

Almost half of the melanomas in the study were melanoma in situ, the most curable yet most difficult form of melanoma to detect, EOS said.

"These clinically compelling data suggest that MelaFind may help detect melanoma earlier, and more accurately differentiate many of the non-malignant lesions that mimic melanoma," said Darrell Rigel, MD, clinical professor of dermatology at New York University Medical School. "This should lead to improved biopsy efficiency and help reduce the number of unnecessary biopsies, which can be painful and scarring."

The skin cancer experts who participated in this study had previously made the decision to biopsy all 1,831 pigmented skin lesions prior to enrolling the patients in the MelaFind clinical trial, the company noted.

In order to generate a comparison with dermatologists' ability to accurately detect melanoma, EOS conducted a parallel pilot readers' study with a different group of 39 dermatologists. Using images and clinical histories of 23 randomly-selected melanomas from the pivotal study, this group of dermatologists, on average, would have decided to biopsy only about 18 (80%) of the melanomas, whereas the MelaFind result would have led to a biopsy of 22 of the melanomas (biopsy sensitivity of 96%). A larger readers' study to provide additional data regarding the sensitivity of MelaFind relative to physicians will commence shortly, EOS said. Data from these studies will be submitted to the FDA.

"Our mission with MelaFind has always been to provide a useful tool to aid in detecting melanoma at its earliest, most curable stage. We look forward to discussing these data with the agency," Gulfo said.

MelaFind uses 10 different wavelengths of light to see where a clinician cannot – up to 2.5 mm below the skin's surface. Using advanced algorithms, trained and developed on a database of 9,000 pigmented skin lesions and over 600 melanomas, including those from the pivotal study, the system provides an immediate result that informs the decision to biopsy.

Being able to avoid useless biopsies is something that is also important, particularly to patients, the company has learned through its market research. Patients want something that is non-invasive, Gulfo said.

When talking with D&IW yesterday, Gulfo recalled a time when his wife was looking at some of the company's promotional information about MelaFind and she told him to add that the procedure is pain-free. His reply was, "We said it's non-invasive; we don't need to say pain-free." Then his wife pinched him, Gulfo recalled, and said, "That was non-invasive . . . put pain-free."

The company said it chose the final classification algorithm based on its success at identifying melanomas in a series of large, blinded and sequential internal classifier selection studies conducted immediately prior to the analysis of the pivotal trial data.

Including the pivotal trial, the MelaFind classifier detected 430 of the 432 melanomas against which it was tested.

There were no adverse events associated with the use of MelaFind.