Technology

The Technical Problem and OTraces’ Solution
Cancer Biomarkers and Proteomics

The research into finding simple blood test biomarkers for cancer has been a high priority for many years due to the medical need and anticipated lucrative markets. Historically this research focused on searching for a specific protein that either becomes present or elevated with presence of cancer. These “silver bullet” approaches have been found to be equivocal in diagnosis. The test outcomes maybe either elevated or depressed due to conditions other than cancer, thus masking the disease, leading to false negatives and positives. Even the high demand test for Prostate Specific Antigen (PSA) for prostate cancer is ambiguous (see Prostate Cancer and Competition). Recent developments in cancer biomarker research have focused on proteomics and the evaluation of multiple protein markers. The scientifically based notion is that the normal balance of these proteins is upset by the presence of the disease. The concentration of these markers is surveyed and their composite concentration levels correlated to the disease condition. Two conditions must met be to create technically suitable and marketable disease biomarkers using this proteomics method:

1) The ideal approach would be to choose as few as possible of these markers (5 or 6 proteins) to keep statistical problems (under sampling) to a minimum and choose those that are likely to have a high degree of genetic or physiological connection to the disease. In the case of cancer these proteins would likely be immune system related and/or vascularization factors (tumors promote local blood vessel growth).

2) Additionally, they should be proteins measurable with simple cost effective methods suitable for marketing to the clinical laboratory. Currently, detection technologies suitable for high volume automated processing in clinical labs will measure proteins that are called High Abundance Proteins (HAP’s), roughly greater than about 1,000,000 molecules per milliliter of blood serum. This level of sensitivity was introduced into the lab in the 1970’s via “radio-immunoassay” techniques and via non-radioactive induced luminescence immunoassay technologies, Roche (IGEN) and Abbott, in the mid 1990’s. Detection of Low Abundance Proteins (LAP’s) (roughly 10,000 molecules per milliliter of unknown sample), before OTraces’ methods, required complex, costly and time consuming methods involving very large volumes of serum sample (>100 ml or more) not marketable to the clinical lab.

The above two conditions have not been met by prior art methods research. 1) Unfortunately the most probative protein candidates for cancer, immune related and/or vascularation factors are present in blood at levels below the limit of detection (LOD) for clinically viable automation techniques used today. In addition, many of these higher concentration proteins (HAPs), down-regulate in cancerous situations and thus are again not detectable. 2) The research response to these problems has been to increase the number of proteins (HAP’s) surveyed to as many as thirty. This number of parameters creates a large statistical problem in that the sample sizes needed to get statistical confidence levels greater than 95% are on the order of 900 or more, beyond the cost and scope of most academic research. Some research has been published with what appears to be good disease correlation but with low sample sizes (<100). These studies cannot and inevitably do not extend to the larger population and fail due to under sampling (e.g. BC-SeraPro, 22 proteins but only 60 samples in the research cohort). The technical issue here is fairly straight forward. If a population cohort, with and without a disease state and the protein test sample size approach each other in size, by definition the mathematical model will be able to “fit” the selected population sample and be predictive within it. When this mathematical algorithm is then translated to another cohort or the larger population, it may well have very small predictive value and most certainly will drop precipitously in predictive value (and they all have).

The larger protein panel size also creates a cost and business profitability problem as only six to eight parameters can be assayed economically using traditional automated immunoassay methods. Using Proteomics Chips that can assay up to 100 proteins at once seems to solve the cost problem. However, the ability to detect these proteins is further compromised as these chips have a 50 fold lower limit of detection than current lab methods that will only detect high abundance proteins (HAPs).

Clinically Viable Detection of Low Abundance Proteins, by OTraces
OTraces’ technology focuses on reading the levels of the probative proteins in serum, here-to-for undetectable, using improved upon simple methods already fully automated in the clinical lab. OTraces’ has discovered and patented methods for dramatically reducing the background or noise signals that blocks detection of these proteins at very low concentrations. These techniques are inherent to the signal detection device and processing of the reaction vessels controlled by the manufacturing process. The OTraces’ methods use assay protocols, reagents and disposable devices that are substantially similar to those used with present high volume systems in the lab. They are not dependent on clinical lab technician skills. Prudence would indicate caution, however, when any new methodology is introduced into the clinical lab, as it may not “translate” well. It must be pointed out, however, that sample and reagent handling, with the OTraces’ system, is very similar to current high volume methods used in the clinical lab. The technician adds a serum sample, and the instrument takes care of the rest of the volumetric aliquoting of sample and reagent processing. These “micropipetting” processing technologies are over 35 years old and resident in hundreds of thousands of instruments in the clinical lab. OTraces’ methods are improved, fully automated and suitable for cost effective high volume test result production similar to biochemistry systems currently in the clinical lab. OTraces has processed over 5,000 samples through its automated instruments, the results are stable and the methods are robust. The crux of OTraces improvement is lower background noise so that very low levels of protein can be detected. OTraces can detect the most probative proteins at levels as low as 10,000 molecules per milliliter. Additionally, OTraces has demonstrated that only five or six biomarker proteins are needed. A statistical sample of ~250 is suitable for 95% confidence level, but to account for changes in the immune system due to menopause, a larger cohort (~830 samples for breast cancer) is assumed. OTraces’ patented methods of ultrasensitive detection of proteins are several hundred fold more sensitive than current methods. Improved analytic sensitivity permits the detection of a few low abundance biomarkers with higher predictive power.

Figure 2, above, graphically shows the magnitude of the limit of detection (LOD) problem. These results show the serum concentration levels of two probative proteins, Interleukin-8 (IL-8) and Prostate Specific Antigen (PSA) for 254 women, healthy, blue (square) or with breast cancer, red (dots). PSA is found in women at levels as low as 10 femtograms/ml, a LAP, 400,000 times lower than men putatively positive for Prostate Cancer, 4 nanograms/ml (1 nanogram = 10-9 grams, 4 ng/ml), a HAP, in men. For these two proteins, using current methods, the vast majority of the illuminating data is simply not measured (below limit of detection); the yellow shaded area. The red shaded area of the graph shows the full complement of data requiring the ability to detect down to 10 femtograms/milliliter (10 x 10-15 grams/milliliter, 10 fg/ml). About 80% of the probative data is not measured with current methods. This is especially true of those with cancer. (The fact that some women with breast cancer are comingled with healthy women is sorted out by including 3 other proteins in the profile, see below). Table 4 shows the various protein biomarkers for breast and ovarian cancer. It’s important to note that of these only VEGF is detectable, by current laboratory methods. The new proteomic chips and mass spectrometry will not detect VEGF at even these levels. Note that in breast cancer five biomarkers are used in the panel and six for ovarian cancer (The X).  Signifies an increase in concentration of several orders of magnitude with the disease state (highly up regulated);  signifies a slight increase in concentration (slightly up regulated) and so forth for down regulation. Though the actions of some are similar in both cancers the overall picture shows that breast cancer can be clearly detected and breast cancer can be differentiated from ovarian cancer.

Table 4. Action of Proteins Probative for

Breast and Ovarian Cancer*

 

* X signifies this protein is included in the cancer panel;

pg/ml = 10^-12 gm/ml

 

Figure 3 below shows the composite score calculations for breast cancer. The graph shows a linearized composite score plotted against a square of the sum of squares method, on the horizontal axes. The scores are a composite of the protein

Figure 3

concentrations measured in the patient cohort for breast cancer. The vertical axis shows the population density of the data points. Red peaks are breast cancer and the blue valley is for healthy women. The transition from blue valley to red peaks is the intermediate zone. OTraces score correlates well with biopsy in this intermediate zone. Breast cancer is clearly differentiated from healthy.