VolitionRx — Results from asymptomatic CRC trial reported

On 16 February 2018, VolitionRx reported preliminary results from the ongoing 680-sample trial to identify colorectal cancer (CRC) in asymptomatic individuals using the company’s Nu.Q technology. The goal of this trial was to validate the technology for progression into the planned larger 4,300-sample asymptomatic trial as part of the company’s European approval strategy. Previous studies generally used blood samples from patients that presented at hospitals with bowel symptoms and while these studies were useful in validating the technology, they were limited in that the patients did not accurately represent the asymptomatic screening population, which is the company’s target market.

The results as stated by the company are:

The interim results demonstrated that a small panel of three ELISA assays, when considered with the subjects' ages and smoking histories, produced an area under the curve (AUC) of 83% and was able to detect 80% of Stage I CRC cases and 66% of High-Risk Adenomas (HRA) at 78% specificity, respectively.

The interim results provided in this release were limited in scope. First, these results are a panel of only three tests and on the conference call the company provided additional color that only two of the three tests use Nu.Q technology. The remaining assay is a so-called non-proprietary test that uses already available technology such as hypothetically CEA (carcinoembryonic antigen). The study also included adjustments for the patient’s age and smoking history. Previous results detailing the sensitivity and specificity of the Nu.Q technology generally included four or five Nu.Q assays. It is unclear at this time if the company performed additional tests and opted not to include them in results presented here, or if the other available Nu.Q tests were simply not complete when the data was analyzed. Moreover, the inclusion of smoking history in the algorithm is new and potentially problematic, as smoking histories are frequently unavailable or falsified. The company has stated that it is moving forward with the 4,300-sample trial which will test at least nine Nu.Q tests, so it is unclear at this time the degree to which these results will be predictive of future performance.

The company also made the decision to present only a portion of the data available from this study. Sensitivity and specificity were only given for stage 1 CRC and high-risk adenomas. The absence of sensitivity and specificity measurements for other stages or for CRC as a whole is conspicuous. Generally, early stage disease and adenomas are the hardest to detect, so one would expect the data from later stages to be better. Given that this sample set was drawn from asymptomatic patients, the number in later stages could be limited, but the company did not provide any breakdown of patient numbers. Moreover, there should not be a limitation on analyzing all CRC samples as a whole as the total sample size is by definition larger than just stage 1. That said, the sensitivity and specificity presented here were in line with previous results from the company (Exhibit 1), although it should be noted that each of these studies uses different combinations of Nu.Q tests, non-proprietary tests and other adjustments in each case. Sometimes results from multiple panels are reported from the same study with different panels used for adenomas and cancer. The most recent results appear superior to Epi proColon, but are difficult to compare to FIT given the latter’s significantly higher specificity.

The company also provided an area under the curve (AUC) from the receiver operating characteristic (ROC) plot of 0.83 for the detection of stage 1 cancer. This statistic provides a measure for the performance of a diagnostic test without the need to define a threshold for determining positives and negatives. A score of 1.0 represents a perfect test, whereas 0.5 represents no diagnostic value. It is difficult to interpret these values in isolation because the shape of the curve is important and in many cases determines the trade-off between sensitivity and specificity. It is difficult to provide comparisons for this statistic, because typically it is released for the detection of CRC as whole, whereas in this case it refers to just stage 1 cancer. On this basis, Epi proColon (Epigenomics) has an AUC of 0.82, and the fecal immunochemical test (FIT), has an AUC of 0.86 (as reported in the FDA assessment of Epi proColon), whereas Exact Sciences’ Cologuard has an AUC of 0.93 (from FDA assessment).

The company presented slightly updated guidance for its Europe registration program: results for the 4,300-sample study are expected in Q218 (previous guidance was for study completion by 31 March 2018). The company is now guiding to at least 12,000 samples in the blinded European registration study (up from at least 10,000), and a H218 start date (from Q218) with data expected at the end of 2018. The company announced that it expects to have a finalized product specification for the triage test in Q218. Finally with regards to the US trial, the company stated that it would select a panel for screening over the coming 18 months, concurrent with sample collection.

We have increased our valuation to $212m from $200m, although it is lower on a per share basis ($6.93 per basic share compared to $7.55). Our assumptions regarding the development program remain virtually unchanged. We have reduced 2018 revenue estimates based on the new timing guidance, but the effect is insignificant. The increase is driven by advancing our NPVs and new cash from the March 2018 offering. We include the greenshoe option in our cash estimates (although it has not closed yet) and deduct 6% offering fees.

We have chosen not to change our assumptions regarding the CRC program following the most recent results because of their limited scope and similarity to previous results. However, we expect to update our assumptions significantly with the release of data from the upcoming European clinical trials.

The company reported an operating loss of $15.0m for 2017, compared to $12.4m for 2016 (restated from $12.3m). We expect this value to continue to increase in 2018 to $18.9m with the associated expenses of developing and launching the two CRC tests in Europe. The 2017 operating loss was slightly higher than expected ($14.0m) and we have carried our adjustments through to later years. We do not expect the company to be profitable until 2023 and we expect it to require additional capital to finance operations.

The company raised approximately $8.4m gross (3.5m shares at $2.40) in March 2018. However, given the increase in our expected spending, this has only slightly reduced our financing requirement to $82m from $83m before profitability in 2023. We currently project that the company will need an additional $7m in 2018 followed by $15m in 2019, $30m in 2020 and $30m in 2021, which we record as illustrative debt.