A Tale of Two Diseases

Barry Werth's fine report in Technology Review describes not just the astonishing prices that new drugs can command, but explains how and when the high prices are justified - or not.

Kalydeco is the first effective treatment for symptoms of cystic fibrosis and is priced at $294,000/yr. Despite the massive cost, Kalydeco is being accepted by doctors and insurers because it really helps patients, and, perhaps, because only 2400 patients qualify for treatment.

Zaltrap is a me-too drug for treating metastatic colorectal cancer. It was priced at about $100,000 for a course of treatment - definitely on the high end for new cancer drugs, but not that different from Avastin.

For that kind of money, you'd expect pretty good performance. But Avastin and Zaltrap only increases survival for 6 weeks over standard chemotherapy, which is why Avastin has been rejected repeatedly for use in the UK for failing to meet the test of cost-effectiveness.

Despite its high cost and marginal performance, Avastin racked up nearly $6B in sales last year. That's because in the US, Medicare and Medicaid are required to pay for all FDA-cleared treatments. The FDA only considers safety and effectiveness, and gives no consideration as to whether a new drug provides a meaningful benefit over existing drugs.

FDA clearance is thus effectively a hunting license, allowing marketing departments to create a demand that overwhelms any objections based on cost. And when it comes to cancer, our willingness to pay has been unbounded. Paying $50-100K for an advanced chemotherapy regimen that extends survival by a couple of months is completely normal.

For antibiotics the economics are much different. Consider fidaxomicin, recently approved for treatment of C. difficile associated diarrhea. CDAD kills 14,000 Americans every year - not as many as colorectal cancer (50,000/year) - but no small number. Existing antibiotics, such as vancomycin, have a good global cure rate of some 66%, and fidaxomicin improves this to about 76%.

The cost? Vancomycin treatment is typically about $1200, and fidaxomicin raises this to $2800. In other words, for the cost of two weeks of cancer chemotherapy treatment you can get an antibiotic that will cure you of CDAD 75% of the time. And presumably not die.

The costs to develop a cancer chemo therapeutic are not markedly less than to develop an antibiotic. So is it any wonder that the cancer pipeline is full of drugs with marginal additional effectiveness? If you were the CEO of a pharma company, would you put your resources into developing a drug that you can charge $100K for, or one that you can only get $2500 for?

Antibiotics are simply too cheap for our own good.

 

Posted with Blogsy

On the other hand...

I've been pretty critical of hospitals' reluctance to make use of rapid diagnostics to improve patient care. And I think that this criticism has been well-deserved. But these tests do have some shortcomings, and it's only fair to take note of them.

Limited reportable results. Most rapid tests only identify a single organism. An exception is the Verigene Gram-positive Blood Culture Test, which returns ID results for 9 species. Limited results mean that labs still have to do a full workup in order to identify organisms not covered by the rapid test. Rapid testing therefore doesn't replace any test; instead it adds to workload and cost.

Resistance but not susceptibility. Rapid tests, with one exception, detect sequences or products from resistance genes such as mecA or vanA/B. A positive result thus (usually but not always) indicates a resistant phenotype. But when resistance rates are at 30% or 40% or even more, doctors tend to presume resistance and prescribe accordingly before receiving test results. Thus the clinical value of a positive result for resistance is somewhat limited - the patient is likely already on the antibiotic indicated by the test as being appropriate. And then, a negative result for a resistance gene is just that - it means a certain gene (actually just a small target segment of it) is missing or altered, but it does not necessarily follow that the organism will be susceptible to the antibiotic in question. Of the rapid tests, only the MicroPhage test was cleared to return a susceptibility result which would enable a change in therapy. And it is no longer available.

So there are some legitimate reasons to hold off on adoption of rapid testing. But they still are outweighed by the benefits to the patient, and to the healthcare system as a whole. You don't have to save many ICU days to pay for a whole lot of testing. But the savings are often diffuse and indirect and accrue to payors outside the hospital, while the costs are borne by the lab making the purchasing decision. Viewed from this perspective, it's not hard to see why adoption of rapid testing has been so slow.

Posted with Blogsy

What the CDC report didn't say

Judging from the number of Google alerts hitting my inbox, pretty much everyone who blogs in the healthcare space is writing about the CDC report on the threat of antibiotic resistance. Most of this writing is superficial regurgitation of the headline numbers, and is not really very interesting. The report itself is a nice resource that collects a lot of existing data in one place and makes it accessible to the general public, but does not contain any surprises or revelations.

The report rightfully calls for development of new antibiotics and rapid diagnostics to contain the threat. David Shlaes has posted a nice analysis of the short-sighted and flawed thinking at large pharma companies that holds back antibiotic development efforts.

A different, but no less crippling dynamic holds back the development of rapid diagnostics. As I've written before (here and here), the incentives for rapid diagnostics for antibiotic resistance and susceptibility determinations are misaligned, if not lacking altogether. Patients would benefit from their use, but no one in the hospital has a compelling reason to adopt these tests, which do tend to be pricey by the standards of microbiology. That is, they cost more than a Petri dish or a tube of rabbit plasma.

Hospital microbiology labs have to purchase the tests out of budgets that are always under pressure, given their status as cost centers rather than revenue generators. Micro labs in turn bear no responsibility when patients are prescribed inappropriate empirical antibiotics; they get no reward when an inappropriate prescription is avoided, even though this can result in considerable savings.

A well-respected clinical microbiologist once remarked to me that what MicroPhage really needed was for a patients' family to sue a hospital for not using our test, leading the patient to die while on ineffective antibiotics. He imagined the courtroom scene: "You mean that you left the patient on wrong antibiotics for 3 days because you didn't want to spend $50?" The thought of a penny pinching hospital admin squirming on the stand clearly intrigued him.

The unfortunate thing about this situation is that there is, I think, considerable willingness to pay to get it right. We just don't ask the right people - that is, the patients. I visited Shanghai in 2011 to get a sense of whether China might be a potential market for our test. Remarkably for a nominally socialist country, there is no system of socialized medicine, nor is insurance common. Patients pay out of pocket for care, which is one explanation for why their savings rate is so high. I asked ID docs at local hospitals to explain who would order our test and how would it be paid for. They replied that they would go to the patient or their family, tell them that there was a test that would help guide treatment, and ask if the family was willing to pay for it. They were pretty sure the answer would usually be yes, even at the US price of $50.

And I suspect that most families of American patients with S. aureus bacteremia, which has a mortality rate of some 30%, would give the same answer. But nobody is asking them.

Until that changes, or until hospitals start getting sued for inappropriate antibiotic use, the return on investment for developing rapid antibiotic susceptibility tests will remain questionable.

 

Posted with Blogsy

MDs playing lab tech?

David Shlaes is lamenting the passing of the good old days, when clinicians ran diagnostic tests right there in the ward. Gram stains, differential culture - the whole shebang. Even collected specimens themselves, they did. I'm sure this was great fun, and there is certainly value in having MDs get a little hands-on experience, if only to gain an appreciation of what can go wrong.

David is a smart guy, and knows way more about antibiotics than I will ever hope to - but this is plainly silly. There may be some overlap between doctoring skills and lab skills, but it is pretty slight, in my experience. There are very few MDs (usually MD-PhDs) I've met that I would want to have working in my lab. Not because they weren't smart enough, but because they weren't skeptical enough of their own genius to repeat "interesting" results over and over to see if they were real.

There's a reason why diagnostic tests are run by dedicated personnel who have to demonstrate proficiency and document their methods and results thoroughly. It's not to maintain full employment of lab techs, but to ensure reliability.

But his larger point is very real: the benefit of rapid diagnostic tests is pretty much lost when it takes half a day to get the specimen to the lab, and another half day to report the results. Real-world turnaround times for 2 hour PCR tests are typically 12-18 hours, pretty much destroying their usefulness.

The solution to this lag is not to put a GeneXpert on every floor and let MDs play with them. It's to stop treating microbiology labs like the red-headed stepchildren of the hospital family. Few microbiology tests are reimburseable; the hospital has to pay for them, and thus administrators see micro labs as cost centers to be squeezed, or better, eliminated.

The results are predictable - half of all S. aureus bacteremia patients are on the wrong antibiotic, because MDs are forced to guess at the appropriate treatment absent timely lab results. As more bugs become resistant to more antibiotics, this situation will only get worse.

The rapid (technically, accelerated) MRSA/MSSA test that my team developed at MicroPhage was a commercial failure. This was largely because there was no constituency in hospitals to pay for faster susceptibility results. The healthcare system as a whole would have saved substantial amounts of money as a result of fewer hospital and ICU days incurred by appropriately treated patients. But additional testing means additional lab costs, and few lab managers were interested.

The economic incentives to develop rapid tests are still all wrong. Until that gets fixed, patients will still suffer and die from inadequate treatment, no matter how many doctors we let play lab tech.

 

Posted with Blogsy

And I'm back...

I've been neglecting this blog in favor of business, family and otherwise, but mostly to enjoy the Colorado summer before it is gone.

One of the trips I took was to Trapper's Lake, one of the largest in the state, and also one of the most remote and scenic. At least it was scenic in Aug of 2002, the last time I was there. 2002 was one of the first big drought and wildfire summers in our new climate regime. Several nasty fires had sprung up already, and the sky in Boulder was relentlessly scorched and smoky all summer.

We were hoping to find some relief in the high country, and planned a family backpack trip looping south from Trappers into the Flattops Wilderness, a rolling plateau of lakes, forests and meadows. At 11,000 feet, we figured it would be high enough to be cool and lush despite the drought.

Unfortunately a fire had sprung up near Big Fish Lake, a valley just a few miles west of where we planned to hike. However, a call to the district ranger reassured us that recent cooler weather had calmed the fire down to the point where it was not a concern, and we proceeded as planned.

The hike was everything we had hoped for - cool temperatures, green forests and meadows, and limitless views from the plateau:

By our third day out, however, we began worrying about the smoke on the horizon, and thought it best to start heading back to the trailhead:

We camped at Parvin Lake, and the next morning it was clear that the fire had blown up. Worse, there is only one road out from Trappers, and the fire was heading toward it, possibly cutting us off.

When we got down on the valley floor by the lake, smoke from the fire was rolling over the ridge, turning the sky a dark, evil-looking red. Even more disquieting was the low rumble of the fire - it had become a monster that was rolling through the forest.

We split up at the lake, my wife and daughters taking the short route to the lodge around the east side. I took the longer trail along the west side where our car was parked. Short cutting the trail through a boggy area, I came upon a middle aged woman and her elderly mother, lost, terrified, and heading the wrong way. I walked them (all too slowly) back to the trailhead, jumped in the car and hightailed it down to the lodge. There I found a ranger screaming at my wife to "Get in the truck, now!" and leave with him before the road out was cut off. We wasted no further time with him and started driving as fast as we could down the rough dirt road.

We rounded the ridge and saw flames shooting hundreds of feet in the air, as if the atmosphere itself was on fire. The trees were dwarfed by the size of the flames, which had become a red mountain piled on top of the green mountains. I took a very quick picture:

The trip this year was quite a bit less exciting. And although the frosts are mostly gone - they will take decades to regrow - Trappers Lake is still a place of tranquility and beauty

 

 

Posted with Blogsy

Combo therapy fails

A common question I've heard when discussing the case for accelerated antibiotic susceptibility testing is "why not just use several drugs in combination?", the presumption being that one of them is sure to work. The usual answers are that there is a risk of toxicity, and that pharma companies are generally not interested in sponsoring combo therapy trials - they want to own markets, not share them.

Two articles published in CID (thx @ndm1bacteria) add a third and more compelling reason: combo therapy does not work nearly as well as one would expect. A study of Acinetobacter infections treated with colistin and rifampin, and a study of Pseudomonas infections treated with various antibiotics, showed no improvement in 30-day mortality for combination over single therapy.

This is a surprise, given the success of retroviral combination therapy, and the results of in vitro microbiological studies. Why the lack of efficacy? I can think of three factors: pharmacokinetics and biodistribution, bacterial SOS responses, and host inflammatory response. It's possible that a significant fraction of the infecting bacteria did not (at least initially) get exposed to a significant dose of both antibiotics. And when exposed to one antibiotic, bacteria (unlike retroviruses) will go into stress/shutdown modes that make them less susceptible to additional antibiotics. If the infection is not rapidly eradicated, the host inflammatory response can spiral out of control, leading to organ damage and death.

In contrast to combination therapy, appropriate antibiotic therapy was shown to improve survival. The bottom line here is not surprising: using the right antibiotic matters, and a shotgun approach to therapy is not a substitute for evidence-based medicine. The problem is with the evidence part: accelerated antibiotic susceptibility tests still don't exist. And they will continue to not exist until physicians begin to demand them and lab managers show a willingness to pay for them.

 

Posted with Blogsy

Mass spec - not the droid you are looking for

Maas spectrometry exerts a peculiar hold over the minds of its advocates. They tend to be passionately committed to it, like no other technology I've seen.

MS is indeed a powerful technology. A sample is ionized so that it has a net electrical charge, and the time that it takes to fly toward a cathode is used to calculate its mass (more time = more mass). The identity of compounds in a sample can be deduced from matching expected with observed molecular weights. Under ideal conditions, only a few femtograms of material are needed. Good stuff indeed.

When complex samples, such as tissues, cells or fluids are analyzed, a complex pattern of peaks is observed. These patterns are very data-rich, and can be mined to produce a signature, and the signatures can be used for clinical tasks such as predicting therapeutic response to cancer drugs - or identifying bacteria.

The allure of this technology to a micro lab manager is obvious - a universal platform for identifying bacteria. A loopful of bacteria from a colony are smeared on a plate, overlaid with an acidic solution, zapped by a laser, and out comes a readout that IDs the sample.

When it works (and it works most of the time) it is beautiful and very rewarding. But it doesn't work all of the time - some species are just not distinctive enough in their signature to give a good result. More importantly, MS is very sensitive to the sample matrix, meaning that it doesn't work nearly so well on direct specimens. As a result reliable ID will still take at least a day, in order for purified cultures to be prepared.

I think this very good but not perfect success rate is the key to the enthusiasm (if not obsession) that many mass spec'ers exhibit. When you train a dog, you do not always give the treat - the uncertainty in the dog's mind as to whether she will be rewarded causes her to pay much closer attention to her master. Casinos operate on the same principle - provide just enough reward to create hope and anxiety, and the reward becomes much more gratifying and addictive. The occasional failure serves to redouble the true believers determination to succeed.

I was glad to see that the recent paper by Donna Wolk and colleagues takes a level-headed approach to the strengths and weaknesses of mass spec, despite its title. And it is likely true that mass spec will instigate a "fundamental shift in the routine practice of clinical microbiology". I just don't think patients will see much benefit.

Although all clinical microbiologists are trained to focus on organism ID, this parameter is decreasingly important in an age of widespread antibiotic resistance. An instrument that can ID hundreds of potential pathogens is nice for lab workflow but consider: just 4 organisms account for 80% of positive blood cultures (counting Coagulase-negative Staph as one organism); 6 account for 90%; and 7 for 95%. The ability to ID a Citrobacter or Proteus will help very few patients. Knowing whether a septic patients' Gram-negative rod is susceptible to cephalosporins or carbapenems has much more therapeutic utility, regardless of whether it is an E. coli or a Klebsiella.

There have been some initial efforts to determine susceptibility and resistance using MS - looking for specific enzymes, or antibiotic degradation products, or even cellular response to antibiotic exposure. I think all of these will ultimately fail because of the vast heterogeneity of resistance mechanisms and differences in strain responses to antibiotic challenge. Even with copious data storage and analysis capabilities, the signature of each strain of each species to each antibiotic over a range of physiological conditions will have to be validated, and have to be constantly updated as new strains and resistance mechanisms emerge. This is implausible.

What would really benefit patients (as opposed to lab managers) is a same-day test that would return ID results for the top 4-6 pathogens, with susceptibility results for 2-3 frontline antibiotics each. Such a test would ensure that 90% of critically ill patients receive appropriate antibiotics, instead of the 50% that now do.

But that brings us back to the question of the economics of test development. We were working on a panel test at MicroPhage. But with microbiology labs unwilling to pay $50 for our test that returned same-day results for MRSA/MSSA, how much would they have been willing to pay for a small panel? Potential partners and investors surmised that the answer was "not enough", and the project never really got traction. So long as antibiotics and susceptibility testing are undervalued, the prospects for developing a test that will get patients the right antibiotic right away are not good. Instead what we will have are $150,000 instruments like mass spectrometers that improve workflow and reduce staffing costs.

 

Posted with Blogsy