In the course of reviewing the new targets that made their way into FDA-cleared tests for the first time during 2013, one of the more interesting examples was the Prosigna® panel released by Nanostring last year.
Breast cancer prognosis panels have been available for almost a decade, first from Genomic Health (Oncotype Dx®, 2004), and then Agendia (Mammaprint®, 2007), but Prosigna® is the first panel that is available as a kit (while the Mammaprint® test is FDA-cleared, it is offered as a service rather than a kit). Prosigna® is an mRNA expression panel that is run on Nanostring’s proprietary instrument, utilizing technology originally developed at Lee Hood’s Institute for Systems Biology in Seattle (where Nanostring is based).
We thought it would be interesting to compare the three breast cancer prognosis panels to see how much they differ in terms of 1) target composition; 2) clinical, commercial, regulatory validation; and 3) paths to market.
All three panels are offered for the same basic function, risk assessment for distant disease recurrence (i.e. metastasis of breast cancer), but their differences in the aforementioned areas provide interesting opportunities for comparison.
Target Composition
One of the first takeaways is how few targets are shared by the three panels. Of the 125 targets that are used in these panels, only 12 are shared between the three tests, and none across all three.
Targets from just about every chromosome are represented (with the exceptions of 21 and 23), with a median of 5 per chromosome. Only chromosomes 1, 8, and 17 stand out, having 2-3X the median representation.
The targets that are shared by the three tests are not unexpected, as all of them have activities related to apoptosis, cellular proliferation and/or growth, and metastasis. Some examples include anti-apoptotic BCL2 and its supporter BAG1; EGFR and glycolysis supporters ERBB2 and GRB7; anti-apoptotic MELK; and metastasis supporters MMP11 and SCUBE2.
In other words, just the sort of genes one would expect to be supporting fast growing and spreading cancer cells.
Clinical, Commercial, and Regulatory Validation
All three tests have similar numbers of publications that describe validating studies. The differences, however, in the average levels of validation for the individual targets that comprise each panel are strikingly different. We focus our validation analysis below on the unique targets in each panel, those being the targets that are not used in either of the competitors’ panels.
The unique targets in Mammaprint®, for example, have been included in few to no clinical studies, and very few studies overall. Average patent activity for these targets is also quite low (0.5X the median), which could in fact be causal for the low rates of use in clinical studies (under the, perhaps incorrect, assumption that a limited number of patents overall for a given gene correlates with strong protection for those who hold patents for that gene, and thus less use of the gene by non-patent holders). Over 20% of the unique targets in Mammaprint® are cited in 100 or less patents. Patents and clinical studies are positively correlated overall across all targets in all panels (r=0.62).
By contrast, the unique targets in Oncotype Dx® have much higher average clinical validation (3X the average number of clinical studies), and also very high patent citation (11X the average).
Prosigna® may be in the most attractive position, having relatively high clinical validation (2X the average number of clinical studies), and relatively low patent citation (2X the average), at least compared to the targets in Oncotype Dx®.
Agendia and Genomic Health both had a commercial test available around the period 2004-2005, and proceeded in different directions in terms of regulatory validation. Agendia sought and obtained 510(k) clearance for Mammaprint®, and Genomic Health went straight to market with an LDT under their CLIA accreditation. Considering that both tests were (and are) offered as a service rather than as a kit, it was a unique (and potentially costly) decision for Agendia to pursue FDA validation.
From a March, 2013 article in the Orange County Register about Irvine-based Agendia’s decision:
Neil Barth, Agendia’s chief medical officer, said that when the company took a “two-year hiatus” from 2005 to 2007 in order to get up to speed with the FDA, it “lost tremendous momentum and market share.” It gambled that FDA approval would give it superior standing in the oncology market and drive sales. “But unfortunately it has not yielded the benefit we had hoped for,” Barth conceded.
Here is a clear example of two very similar tests for the same indication being sent in alternate regulatory directions, with the LDT-provider seeming to come out on top.
Nanostring’s Prosigna® test is offered as a kit, and so its clearance by 510(k) was more expected. Labcorp and Arup are already offering the test, so Nanostring’s development of a kit may ultimately provide a competitive advantage in terms of leveraging channel opportunities (which also fits with the company’s strategy of establishing an installed base of instruments).
Paths to Market
The choice of assay platform has an impact on the way a multiplex test is offered commercially (if not always dictating its regulatory path). Assay type probably dictated that Mammaprint® would always need to be offered as a service, as microarrays are challenging to market as IVDs (see Pathwork Dx). Oncotype Dx® actually lends itself to potential kit format as an RT-PCR assay, albeit likely at a higher cost than as currently provided as a service.
Table 4. This table lists the launch year, assay type, and delivery format for the three leading breast cancer prognosis panels.
All three tests are priced at $4,000 ± $200, and Mammaprint® and Oncotype Dx® are reimbursed by Medicare. Nanostring is actively seeking reimbursement for Prosigna®, and is likely to receive it based on similarity to its competition. Price and potential for reimbursement are two areas where differences are much less pronounced than in the other areas explored.
The differences in these other areas in fact provide interesting opportunities to evaluate what decisions most influence success in developing and marketing multiplex biomarker-based IVD tests. Is it better to create a larger panel of unique targets with better IP protection but more limited clinical validation, or to go with a smaller number of tried and true targets that afford more limited IP protection? Does FDA clearance confer benefits to tests offered as a service? Do kits offer greater market penetration opportunities than tests offered as a service? Ultimately the sales growth for each test will answer these questions.
Genomic Health is the only company among the three whose sales are publicly available, and the company experienced 47% growth over the last 3 years. We will seek sales growth data from Agendia and Nanostring as well to see how successful their strategies have been.