Predictive power may be the most demanding performance standard in diagnostics because, unlike sensitivity, all false signals are taken into account.  The term is not widely used in the industry perhaps in part because high levels are rarely achieved.

OTraces Inc., founded in 2008, has developed a blood test and instrument/reagent platform that uses patented math, physics and measurement science amplification methods and noise-suppression techniques to achieve superior levels of predictive power. In over 2,000 tests conducted in recent years, the OTraces approach has averaged predictive power of 95-98% in the detection of breast, prostate, ovarian and lung cancer tumors, and a clinically validated 98% score in breast cancer based on validation trials conducted at the Gertsen Institute in Moscow. These results compare with an estimated 80-85% predictive power for the industry’s most advanced proteomic and other biologic diagnostic testing methods, 80% for most screening mammography procedures, and less than 60% for the PSA test for prostate cancer screening.

The OTraces approach is not unlike the physical science methods currently employed in deep space communications and missile guidance systems to resolve extraneous noise and signal complexity as adapted to the special needs in disease detection where non-linear and unpredictable events are not uncommon. OTraces has filed for comprehensive patent protection on all known computational methods for achieving predictive power of 95%-plus that apply to the enhancement of all types and classes of biomarkers and biomolecules. By focusing on the micro environment surrounding the tumor and multiple aspects of immune response, the OTraces technology has broad potential in the detection of other complex diseases including cardiac, Alzheimer’s, Age-Related Macular Degeneration, premature birth, and Lyme Disease.

Performance Highlights — What Computational Enhancement Can Achieve.

In breast cancer detection, OTraces uses its computational methods to measure immune response in five dimensions for each of the biomarkers employed — each have disease relevance and each contributing to predictive power in a steady progression. The following charts illustrate the performance of the OTraces approach in breast cancer validation trials conducted at the Gertsen Institute. As shown, OTraces uses a combination of five biomarkers with complementary functions to measure different important aspects of immune response and disease progression.

The Biomarker Surge and the Micro-environment Around the Tumor

The five biomarkers used in the OTraces method are independent actors in the sense that they are actions by the immune system to suppress tumor growth or are actions initiated by the tumor to grow and prosper. The biomarkers selected are in three categories 1) a simple tumor marker, Kallikrein 3 (Prostate Specific Antigen or PSA); 2) immune system actors, a) pro inflammatory, Interleukin 6 and b) anti-tumor genesis, Tumor Necrosis Factor alpha;  and actions by the tumor to improve its access to nutrients and oxygen a) improved circulation in the local surrounding tissues, Interleukin 8 and b) circulation within the bulk of the tumor, Vascular Endothelial Growth Factor.

The chart on the left, figure 1, shows the actions of these biomarkers as the disease progresses from not breast cancer to breast cancer through the growth stages 0 through 3. These biomarkers are secreted into the surrounding interstitial tissues and taken up in the blood. They can be detected using OTraces sensitive immunoassay methods. Note the dramatic response induced even as the formation of a nascent tumor begins (the transition from healthy to Stage 0). The tumor is requesting angiogenesis, IL 8, in the surrounding tissues when the tumor is small. Also the immune system reacts strongly with the IL 6 action (turn up the immune system volume) and the tumor killing actions of TNFα. As the tumor progresses it seems to temper its immune system response as expected. As the tumor progresses in size it turns up its request for bulk tumor vascularization (VEGF). Note that the cancer score spikes from an average of 20 to 190 upon the onset of stage zero breast cancer (the cancer score range is arbitrary 0 to 200).

Note that all Stage 0 and those Stage 1 patients that presented with no symptoms in the Gertsen market clearance trials were called correctly. In fact all stage 0 and stage 1 patients were scored correctly. This is the critical criterion for early detection.

Though not provable yet, (more statistics needed) the data suggests the tumor may be detectable before it can be seen in imaging.

The independent actions of these biomarkers allows the creation of a different predictive model the will correctly call the breast cancer stage, to again >95% accuracy. This may allow the cancer detection test kit to make a call on cancer stage along with the cancer score directly. This may be useful in helping radiologists find the currently very difficult to see stage 0 cancers (less than 2 mm in size) improving early detection and patient outcomes.

The Receiver Operator Curve

The receiver operator characteristic (ROC) curve was invented during World War II to help radar operators make effective calls on whether to send interceptor aircraft to a “bogy” seen on the radar screen (what are the odds it’s a flock of birds or enemy bombers). It is very useful in showing quickly to the eye how good a test is and where to set a “trip” point for making an action decision. In this case the ROC curve shows the trade off between distance and intensity of the signal.1

In IVD diagnostics it is used extensively to set the decision “trip” point and the trade off is rate of false negative calls versus false positive calls,  generally false negative calls are considered worse. The current prostate cancer PSA test has a false negative rate of about 10%, but it gives up a very high false positive rate to achieve this (75% to 80%).

The next chart on the left, figure 2, shows the ROC curve effect of all five biomarkers, culminating with the red line at the upper left side that indicates predictive power at the extreme upper end of the industry range. Typically proteomic methods achieve ROC curve performance equal to or lower than the brown IL 6 plus VEGF curve. The red curve shows the OTraces BC Sera Dx, breast cancer detection test kit, ROC performance as measured from the Gertsen Institute third party market clearance trials in Moscow (see below). Note that each added biomarker amplifies detection performance. This is due to unique and patented method that applies 1) proteomic noise suppression, 2) multi-dimensional spatial correlation method and 3) age normalization.

Commercialization Plans.

OTraces is pursuing an opportunistic and capital-sparing multinational commercialization strategy to reach the company’s financial and valuation goals for substantially less than would be required in the U.S. for FDA approval under the PMA regulatory process. The current opportunity in the Russian Federation is promising and launch of the breast cancer screening test is expected in 2016, subject mainly to an additional validation trial at the Gertsen Institute which should be completed during the first quarter.

Validation of the prostate test at the Gertsen Institute should follow. Meanwhile North American validation trials for the OTraces prostate cancer test at Johns Hopkins University Medical Center have begun with final results likely to be announced by early 2Q 2016. These results will serve as a basis a possible Canadian new product application for screening prostate cancer as well as possible resumed discussions with the U.S. Food and Drug Administration.

1The ROC curve helps in decision point trade offs. In the curve above the 45o tan line represents the null test, or the test with no predictive value. A perfect test would cover the full area in the plot above this 45o line. It spans up the vertical axis to 1.0 and then across the graph horizontally to the right vertical axis. This would be a ROC curve with “area under the curve” (AUC) of 1.0. The best diagnostic test would come as close to one in AUC as possible. The PSA test is very similar the current plotted result for VEGF alone 90% sensitivity and 25% specificity.