"For me the biggest payoff in cancer research would be the discovery of biomarkers that can be measured in the blood that reflect the presence of early-stage cancer."
Dr Leland H. Hartwell, March 2008
President, Fred Hutchinson Cancer Research Center
Nobel Prize in Physiology or Medicine, 2001
Epigenomics AG (GM:EPGNF) recently filed with the FDA the 4th and last module of its blood based colorectal cancer (CRC) screening test, EpiProcolon (EPC). Following the recent successful read-out of the head-to-head study vs. FIT (fecal immunological test), the stage is set for the first PMA certified blood based cancer screening test mid 2013, and for Dr Hatwell's vision to come true. (Below I shall refer to Epigenomics as ECX, its German local exchange code)
Exact Sciences (EXAS) completed recruitment of the Deep-C pivotal trial for its CRC stool DNA screening test Cologuard in November. The read-out is expected in March/April and PMA approval early 2014. EXAS is well known to the readers of Seeking Alpha, and enjoys broad analyst coverage, so I shall limit my comments to the points relevant to the comparison between the two.
Why screen for cancer?
This sounds like an awkward question to ask, but epidemiologists insist that the only outcome of cancer screening should be reduction in mortality rate. If early diagnosis only makes you live longer with cancer, but doesn't improve your survival horizon, what's the point ?
I'd like to refer to an expert for a more detailed and qualified explanation of the underlying concepts: screening, diagnosis, early detection and prevention. It is a lecture by epidemiologist Dr Judith Walsh, from UCSF, given in August 2012. The title says it all "Controversies in cancer screening." The first 28 minutes are dealing with CRC. To me, the most shocking part is not what Dr Walsh reveals, but what she doesn't even mention once: blood-based CRC screening! I'll come back to this conundrum in a later section.
Why screen for colorectal cancer?
Evidence as to the benefit of screening for CRC is very clear cut, whatever the diagnostic method used (colonoscopy, sigmoidoscopy, FOBT/FIT). Even FOBT has a significant positive impact, although its detection rate is the lowest of all recommended methods. It is mainly the development stage of the tumor at the timing of the diagnostic which conditions the 5 years survival chances.
Stage 1 and stage 2 are local lesions and have a 5 year survival rate of respectively 94% and 82%. Stage 3 has spread to some lymph nodes. It is a regional lesion with a 5 year survival rate of 62%. Finally stage 4 has metastasized to another organ, very often the liver, and only has a 8% 5 year survival rate. 60% of total diagnosed CRCs are late stage CRCs (regional and metastasized).
In 2010, the average annual treatment cost of a Medicare CRC patient was $140 000. If already metastasized at the time of diagnosis, the average cost of treatment increases to $310 000.
Cutting through the polyp maze: lesion typing and terminology
A polyp is an elevation of the mucosa and can be found in the oesophagus, the stomach, the small intestine but most commonly in the colon. Colorectal cancer originates from polyps in the colon or the rectum but not all polyps turn into cancer. They can be classified as benign (hyperplastic polyp), pre-malignant (adenoma) and malignant (colorectal adenocarcinoma). Neoplastic polyps include both pre-malignant and malignant lesions and indicates that cells have lost their normal differentiation. Adenoma are sub-classified histologically into tubular (with 80%-85% of total, the most common type, also referred to as adenomatous polyp), tubulovillous (10%-15%) or villous (5%). Lastly, two terms need to be introduced as they are commonly used to distinguish between high risk and low risk lesions: high grade dysplasia -HGD- and low grade dysplasia -LGD. In pathology dysplasia refers to an abnormality of development and is often indicative of an early neoplastic process.
The risk of a polyp to turn into CRC depends mainly on 3 factors: a) the degree of dysplasia, b) the type of the polyp and c) the size of the polyp. For the most common one, tubular adenoma (80% of all adenoma), the risk to turn into cancer within 12 months is 5%.
How to screen for CRC?
The list of available CRC screening options is long: colonoscopy, flexible sigmoidoscopy, CT colonography, double contrast barium enema, stool DNA and of course FOBT (faecal occult blood test) and FIT (faecal immunological test). It is also very confusing, as not all professional associations agree on what should be considered a screening option. Declaring colonoscopy as the "preferred CRC screening strategy" is the path of least resistance for the associations and the GPs.
"What's wrong with the established CRC screening methods? It's the compliance rate, stupid !"
Estimates of the compliance rate in the US range from 55% to 65%. It is by far the highest of all developed countries, but still well below the level of breast cancer and cervical cancer screening. It is fair to say though that screening compliance has been on the rise, mainly because of the steady increase in colonoscopy, 4.3m of which were performed last year. A combined 10.2m FOBT/FIT tests were performed in 2010, 34% of which were FIT vs. 2% in 2004.
In 2008, a national CRC screening program was introduced in France. And although it was by invitation and completely free for the eligible population, after 4 years, the compliance rate was leveling off at 30%. In 2012, the FOBT test was replaced by the FIT test, and the compliance rate edged up to 32%. Germany is the only other country along with the US which offers colonoscopy as a free screening option starting at age 55, although without invitation. The compliance rate is about 19%. In general in Europe, the compliance rates hovers around 30%, and is estimated below 20% in Japan.
EpiProColon: methylation of Septin9 as biomarker of CRC
DNA Methylation is one of very few post-translational gene regulation processes. It is believed to be a fundamental driver of tumorigenesis. Focusing since its inception in 2000 on DNA methylation and its potential as biomarker gave ECX a headstart in identifying the most promising ones. Methylation of septin9 is the most promising and advanced one of a portfolio of 20 markers. In layman's terms, DNA methylation can be compared to a reversible, chemical on/off switch of expression of genes by the docking of a small methyl group to cytosine.
The FDA PMA certification: show me the data
The original FDA wording is summarized in 2 sentences:
1) "there is reasonable assurance the device is safe and effective for its intended use as prescribed in the product labeling;" and
2) "the device manufacturing facilities, methods, and controls were inspected and found to be in compliance with the Quality System regulation"
- Safety: not an issue
- Quality control: the IVD kits are manufactured by a cGMP supplier NextPharma
- Intended use should read along the lines of:
The EpiProColon CRC screening test is intended for use as an adjunctive screening test for the detection of methylated septin9 as a marker of CRC. A positive test result should be followed by colonoscopy. A negative result could refrain from any other investigation and resubmit to testing every -frequency- (yearly or every other year). The test should be used on subjects who are typical candidates for CRC screening, adults of either sex, 50 years or older, who are at average risk for CRC.
- Labeling: it will list the exclusion criteria, like pregnancy, hereditary forms of CRC and Crohn's disease for instance. It will also underline the fact that no dietary restriction is required before taking the test and, most importantly, that no pre-existing condition requiring medication is an exclusion criteria.
- Effectiveness: answers the question: how good is the test at what it is supposed to do? The 2 basic metrics that measure the performance of a screening test are sensitivity and specificity. The former answers the question: how good is the test at identifying subjects as having a disease who actually have the disease. The latter answers the question: how good is the test at identifying subjects free from the disease who are actually free from the disease:
Sensitivity = TP / (TP + FN) = true positives / (true positives + false negatives)
Specificity = TN / (TN + FP) = true negatives / (true negatives + false positives)
Since 2005, 13 case control studies (4500 patients) and 1 prospective study (7929 patients) have been conducted (page 31/37 in ECX's presentation deck). The picture is very consistent and converges to a performance level of 90% sensitivity at 85% specificity for the case control trials.
In January 2010, the PRESEPT prospective trial demonstrated 67% sensitivity at 88% specificity in a 7929 asymptomatic screening population. Sensitivity for early stage was 57% (20/35).
In October 2011, the European CE marking case control study demonstrated 95% sensitivity at 85% specificity. Sensitivity for early stage was 91% (51/56)
In December 2011, the US pivotal study showed 68% sensitivity at 80%. Sensitivity for early stage was 59% (17/29). This was a disappointment, as the improved assay and the latest case control studies hinted at an incremental improvement. As a consequence and considering the increasing popularity of FIT, the FDA had no choice but to require a direct comparison with FIT.
In December 2012, the non-inferiority study vs. FIT was completed and showed 71% sensitivity at 81% specificity. Sensitivity for early stage was 61% (26/39). By comparison FIT showed 67% sensitivity at 94% specificity and early stage sensitivity of 67% (26/39).
When ECX started recruitment for its PRESEPT study in June 2008, it was already clear what the test must achieve to get a PMA. In the FDA's own words: "find the majority of CRCs in an asymptomatic screening population". The FDA wasn't more specific, neither didn't it impose a minimum specificity target, as it considers that this metric is more relevant to payors.
One of the main issues with EPC data (according to analysts covering EXAS) is the seemingly high variability of performance. I believe this is not a real issue. Considering that the distribution of histologic findings varies from one trial to the next (relative number of pT1, pT2, pT3 and pT4 cases) and that sensitivity is a function of histology, it is only logical that estimates of sensitivity fluctuate. This is particularly true of case control studies. Prospective studies, if large enough, and with sequential enrollment, are less prone to this phenomenon. The higher sensitivity of case control studies compared with prospective studies is explained by the fact that almost all pT1 cases are treated during the colonoscopy and therefore disqualify for recruitment in the active arm. Being deprived of pT1 lesions, i.e. enriched with pT2 lesions or higher, overall sensitivity is higher in case control studies.
Another bias explains the variability of specificity. The control arm is systematically enriched with cases like adenoma or polyps in order to test if there is a sensitivity for these pre-cancerous lesions. This happens both in case control and in prospective studies. As a consequence, the proportion of truly asymptomatic patients can vary a lot, hence the impact on specificity. One last and important remark about specificity. Quest (DGX) has been using its LDT version of EPC since early 2011. I estimate that about 80,000 tests have been performed to date. Quest indicated during its recent AMP presentation that the overall positivity rate has been 6,7%. If one assumes that CRC incidence is 0,7%, Quest's experience suggests that EPC has a 94% specificity. You could argue that Quest's experience is biased, and you may be right, but in any event, this number is much closer to the actual specificity than 81%.
The cost effectiveness or pharmaco economics of CRC screening
Even in such different healthcare systems as the US and Canada, cost effectiveness studies show that screening for CRC not only reduces mortality but decreases costs. Let me quote the conclusions of two prominent papers that were published in 2010.
- Effect of rising chemotherapy costs on cost savings of colorectal cancer screening, Feb. 2010, JNCI
With the increase of chemotherapy costs for advanced colorectal cancer, most colorectal cancer screening strategies have become cost saving. As a consequence, screening is a desirable approach not only to reduce colorectal cancer incidence and mortality but also to control the costs of colorectal cancer treatment.
- The cost-effectiveness of screening for colorectal cancer, Sept. 2010, CMAJ
Screening of average-risk individuals for colorectal cancer is a cost-effective measure, even with less-than-perfect compliance. Recognizing that decisions about screening for colorectal cancer depend on local resources and individual patient preferences, either an annual high-sensitivity faecal test, such as a faecal immunochemical test, or colonoscopy every 10 years offer good value for money in Canada.
None of these studies include EPC. In 2011 though Dr Uri Ladabaum completed a first cost-effectiveness study. The conclusions were presented at the DDW in May 2011, but the study itself has never been published, simply because the underlying PRESEPT study (the 1st prospective performance study) has never been published! Here are his main findings:
Under optimal uptake and adherence mSEPT9 every 2 years decreased CRC incidence by 41% and mortality by 61% at cost of $8400 per quality-adjusted life years (QALY) gained vs. no screening ….. The costs of colonoscopy and CRC care were the most influential variables on the cost-effectiveness of all strategies. At the population level, mSEPT9 yielded the greatest incremental benefit at acceptable costs, when it increased the fraction of the population screened, as opposed to substituting for current strategies.
So clearly in Ladabaum's base case, EPC makes perfect sense as an additional CRC screening option. I believe further that his calculation is underestimating the attractiveness of EPC. Indeed some critical inputs are questionable: a) price/test $ 150 vs. $ 100 according to information directly from Quest, b) localised CRC sensitivity 51% vs. 59% in US pivotal study, c) price of colonoscopy was the Medicare price and not the actual average billed price of about $1500 in real-life setting, including sedation and GP referral. (split up of colonoscopy costs).
Cost-effectiveness studies are useful tools, although something of a black box for most people like you and me. The complexity of Markov and other models hide some very structural flaws. I refer in particular to non-medical costs incurred exclusively by the patient or his employer, and invisible to the healthcare system. These are not to be mistaken with copayments; they include, for example, transportation cost and time off work. A study published in 2008 in the Clinical Gastroenterol Hepatol by Heitman et Al. estimated the non-medical cost per FOBT test at $36 and $308 per colonoscopy. These costs are likely to be minimal, if measurable at all, in the case of EPC.
In order not to be left in the dark, I have tried to come up with my own, simplistic version of a cost effectiveness model. Instead of considering the whole heterogeneous population of 50-75 years old, I have chosen to run my analysis starting with a clean sheet, on the 4.5m Americans who turn 50 every year and follow them for the next 10 years. The bottom line is that, even taking into consideration the screening cost alone (not the reduction of cancer treatment costs!), screening with EPC is cheaper than colonoscopy i.e. $137/person/year vs. $150/person/year. The only debatable issue is the 93% specificity, but I refer to my explanation about specificity above. Other key inputs: $75/test price, $1500/colonoscopy, 71% sensitivity.
The payor's view
We have seen above that based on a basic cost effectiveness model, all payors should support the introduction of EPC as a new CRC screening option and consequently reimburse it. This is somewhat of an oversimplification though as the regional and company specific settings vary greatly. Factors like demographics (age, ethnicity, population density) and available endoscopic capacity are relevant to understand where health organizations stand, Kaiser Permanente for example is very keen on improving the compliance rate and implementing a pre-colonoscopy triage. His former CMO Dr Jeffrey Weisz, made this point very clearly in the following video interview (4mn43s), sponsored by Polymedco, a leading FIT supplier. UnitedHealth (UNH) is also an outspoken supporter of a blood based test.
One very important condition for a smooth organizational uptake has been the issuance by the AMA (American Medical Association) early 2012 of a specific CPT code including mSeptin9 tests. CPT 81401 has in the meantime been confirmed in November and is being used since 1st January 2013 on a $250/test basis. This shouldn't be the actual billed price though considering the average $100/test charged by Quest for its LDT version so far.
Discussions with CMS have been going on in the background but needless to say, won't be finalized before at least FDA approval.
The labs' view
This is, in my opinion, the crucial link in the chain. If EPC makes sense for the labs (read Quest), it will fly, if it doesn't, it won't. I have set up a little model to find out. My starting point is the end-patient price I used above: $75/test. Another key input is Quest's gross margin: 57% (if not diluting the gross margin is considered the minimum requirement from Quest's point of view). If one uses 200 as the number of tests/day/lab technician and applies the average hourly wage in the sector, the purchase price per test that balances the model is $31. That leaves me with the last question to be answered: what is ECX's profitability with a $31 selling price ? The answer as always in such cases depends on volume assumptions. Relying on indications I got from the industry, in a high volume scenario, a price as low as $5/kit is a realistic assumption. This may appear completely unrealistic today, but not a few years out, and considering that this price could be squeezed out by a much bigger player, i.e. ECX's eventual acquirer.
Colonoscopy: screening or diagnostic gold standard ?
In a seminal November 2011 editorial of the Journal of the National Cancer Institute, Russel Harris and Linda Kinsinger chose an explicit title: "Less is more: 'Not going the distance' and Why"
The whole article is a must read. Here is my favorite part:
The issue of overdiagnosis, a term that has primarily been used with other cancers, should be considered because overdiagnosis is also a major problem for colonoscopy screening. The great majority of findings at colonoscopy are small low-risk adenomas and non-adenomatous polyps, not cancers. Removing any polyp increases the risk of complications, yet current practice in the United States is to remove all polyps, regardless of size, which exposes patients to a higher risk for harm with minimal (if any) gain. Removing all polyps also increases the cost of screening. When our goal changes from reducing CRC mortality within reasonable levels of harms and costs to eradicating every existing polyp, we are taking our eyes off the ball, focusing on an intermediate endpoint with an uncertain net effect on the patient.
The right-sided vs. versus left-sided CRC controversy: colonoscopy is not immune, EPC is.
Let's remind ourselves of the obvious: colonoscopy is an invasive procedure. Of all methods used for cancer screening purposes (breast, cervical, colon and prostate), it is by far the most invasive procedure. There are the obvious risks and drawbacks which are well documented in many US and international studies: bleeding (1.5/1000), perforation (1/1000), death (1/15000). But that is only the tip of the iceberg. It all starts prior to the actual colonoscopy, with the bowel preparation: nausea, vomiting, allergy (rare), followed by post colonoscopy diarrhea, dehydration, kidney damage (rare). 20% of colonoscopies require bowel cleaning by the endoscopist. This is, if not a reason for missing lesions, a reason for cost overrun. As 95% of colonoscopies involve sedation, there are also all the typical sedations related risks: allergy, cardiovascular, pulmonary, headaches. There is hardly any data on these issues, so they are easily brushed aside. I look at it from another point of view and ask the question: why should a screening procedure involve any risk, if there are riskless alternatives?
It is not impossible to envisage a standalone future for either ECX nor EXAS, but frankly it is not the most likely scenario. When looking at the respective CEOs, both have a deal-making track record. In 2007 EXAS's Conroy pulled off a trade sale of Third Wave to Hologic prior to the PMA approval of an HPV test. ECX's CEO Taapken has been a venture capitalist for most of his career before turning CFO and being involved in two mergers. No winner here.
If I consider the situation from the point of view of potential acquirers, things get easier. Would I go for yet another stool based test or for a disruptive and convenient blood-based alternative? Would I go for a hardware intensive and complex workflow, or a single biomarker assay, fitting in any lab's routine? If I could tick all these boxes 6 to 12 months earlier than the runner up, would I still bother to consider buying it? ECX is obviously the most likely takeover candidate. If not Abbott (ABT) or Roche (OTCQX:RHHBY), the likes of Qiagen (QGEN), BioMerieux (GM:BMXMF), LabCorp, Siemens (SI) or Quest of course are all likely names to be on the acquirer side.
From a strategic point of view, ECX is by far the most likely acquisition target of the two. In the short term though, I consider its low market cap as a real handicap for a successful takeover. I do not believe that historical shareholders would be willing to part from this investment at current valuation. I definitively would not. Considering the industry's unwillingness to jump the gun, I concede that my endgame scenario comes with a 2 to 3 years horizon. Enough time for valuation and value to align. My forecast is a $800m minimum EV when ECX will eventually be taken over.
I think Epigenomics will win this Darwinian fight. EPC is not a perfect test, but it brings to the party all you need to revolutionize the CRC early detection market:
- performance is good enough and at least matches the performance of FIT, the best non-invasive and FDA approved option. This will lead to PMA certification by the FDA mid 2013 and guidelines inclusion further down the road.
- its ease of use will boost compliance to levels not achievable by any stool-based or invasive alternatives. Even GPs will actively support it because it minimizes the time spent trying to convince patients to take a test in the first place.
- its economics are attractive to all parties, most critically to the labs.
- all payors, private and CMS, will eventually support it, as the body of pharmaco-economics evidence grows.
I started with a quote, let me finish with another one from a fellow Frenchman, Victor Hugo:
One cannot resist an idea whose time has come.