Afrezza is the inhaled insulin developed by MannKind (NASDAQ:MNKD). After the FDA approves Afrezza and after partnership arrangements are finalized, four catalysts will trigger market penetration and commercial success. Three of them involve greater understanding of the significance of (1) how interactive online communities are affecting medical care, (2) how recent research about the physiological processes that result in long-term complications is sparking a paradigm shift that is changing the blood sugar metrics used to guide treatment decisions, and (3) how Afrezza's superior performance on the replacement metrics can protect and enhance the quality of life and long-term health of five groups of people. The fourth catalyst follows logically from the second and third: insurance company acceptance of Afrezza.
This article begins with four paragraphs of context, followed by sections devoted to the first three catalysts, plus a final section on implications for insurance companies, patients, doctors, and investors. Recognizing that most Seeking Alpha readers are investors with limited medical knowledge, I use plain language to explain the complex ideas featured, and I emphasize why each matters.
Pfizer (NYSE:PFE), Lilly (NYSE:LLY), and Novo Nordisk (NYSE:NVO) abandoned their inhaled insulin programs. Pfizer's Exubera had been approved by the FDA, but users were put off by its size, the complexity of getting the right dosage, and its cost, as were insurance companies, who recognized the extra money didn't result in any direct medical benefits. MannKind had previously demonstrated that MedTone, its first generation Afrezza inhaler, was an effective delivery system. Recognizing that its size also put off prospective users, MannKind developed the Dreamboat, a smaller "second-generation" inhaler.
The Phase III tests indicated that (1) this Gen2 device was as safe as MedTone (thereby allowing use of the previously collected MedTone safety data), and (2) it met all of its primary endpoints regarding Afrezza efficacy. For details, see George Rho's excellent article and a cogently written document addressing FAQs that MannKind filed with the SEC on August 26. Except for Summer Street's Bart Classen and Adam Feuerstein, most analysts predict FDA approval by May. After all, the FDA withheld approval of Afrezza last time not because of questions about its efficacy, but rather because of questions about the delivery system that were settled by the Phase III results.
This month, MannKind's stock took a hit after the FDA announced that on April 1, its Endocrinologic and Metabolic Drugs Advisory Committee will meet to discuss Afrezza. Another Rho article put the market's initial negative reaction to the news in perspective. Within the last week, a Seeking Alpha article advocated shorting the stock. However, because more knowledgeable investors quickly pointed out and rebutted the author's misunderstandings, inaccuracies, and distortions, I will refrain from commenting further, except to say that perhaps this article can serve as an antidote.
Within the last month, Maredin Capital Advisors wrote a Seeking Alpha article predicting a huge jump in MannKind's stock price. It provided detailed descriptions of three catalysts projected to occur in the next 6 to 9 months: FDA approval, one or more partnership deals, and a marketing/advertising blitz for Afrezza. It also described the potential for adapting Afrezza's delivery system to many other medical uses. Below, I provide additional reasons for seeing its projected stock price as reasonable, and, in particular, for deeming the prospects for the financial terms in future partnerships to be highly favorable to MannKind.
Why Interactive Online Communities are Relevant
People use the internet both to seek health information (e.g., to learn about medical conditions, treatments, physicians, or hospitals, to follow a family member or friend's health status, or to locate a face-to-face support group to join), and to post information related to health conditions (e.g., to update friends about a particular person's health, to increase awareness about a condition, or to raise money for research or a loved one's treatment). However, in this article, the focus is narrower than internet use in general, and narrower than the broad category "social media," which includes social networks such as MySpace and Facebook. It is solely upon Interactive Online Communities (IOCs), by which I mean "People possessing one or more common characteristics and embracing one or more common purposes who use digital media to post queries and responses to their common characteristics and purposes." This definition excludes online communities such as weblogs, where only a restricted subset of participants may initiate posts.
Furthermore, I focus on particular kinds of IOCs: namely, those that focus on concerns, experiences, and advise relating to the nature and treatment of particular chronic disorders. I will describe IOCs for three such user groups: patients, doctors, and patients and doctors together.
IOCs for Patients with Chronic Conditions
According to a PewResearchCenter report, three years ago, 56% of adult internet users searched for and used information related to medical treatments or procedures. Presumably, during the next five years, the proportion of adults with chronic disorders who will search for such information will be even higher. In other areas of their lives, many of them have already accrued extensive experience exchanging information with strangers with similar characteristics. They are the ones who will have the comfort and the motivation to join, or at least dip their toes in the water of IOCs dedicated to people with their particular chronic condition.
In IOCs, people with a particular chronic condition can (1) express concerns and ask questions about their disorder and various treatment options, (2) share their own disorder-related good times, bad times, and perspectives, and (3) provide reactions and suggestions to the questions and concerns of others. Of course, they participate in IOCs in addition to looking up information or research articles and talking with their doctors.
Consider four kinds of IOC members:
- Attitude-focused. Some pay attention to whether others are satisfied or dissatisfied by a particular type of treatment and ignore how the closely the information sources' characteristics or situations match theirs.
- Confirmation-biased. Others filter information selectively, amplifying the importance of information that supports their initial preconceptions or fantasies.
- Single issue. As in politics, some individuals with chronic disorders focus primarily on one issue. For people with diabetes, the issue could be avoiding needles, minimizing the number of people who know about their medical condition, minimizing weight gain, or preventing long-term complications.
- Truth-seeking. Still others know the latest research, interact with people who have tried various options, and assess the extent to which both their information sources' characteristics and situations match theirs, and their experiences have been favorable, unfavorable, or both. Truth seekers come to their doctors with empirically-based reasons for arguing whether they are an appropriate candidate to try out treatment options that sound like they might work out better for them.
Whether or not IOC members fit neatly into any of the thumbnail descriptions above, they all expect their doctors to attend to what they have learned or concluded, and to respond appropriately.
I will mention three IOCs for people with diabetes, none of which are sponsored by a particular pharmaceutical company. The first is hosted by the American Diabetes Association. I have not inquired about their editorial policy, but I anticipate that the ADA has a vested interest in its own pronouncements about the importance of the current standards metric (discussed below), as well as what the target should be on this metric.
If I had diabetes, I would spend more of my time on IOCs that did not have a corporate goal or an agenda other than facilitating a supportive and thoughtful exchange of concerns, experiences, and advise among people with diabetes. Here are two: Diabetic Connect and the Diabetes Forum.
My purpose here is not to endorse a particular IOC, but rather to alert the investment community that patient IOCs play an increasingly influential way in the way that medicine is practiced for individuals who have chronic conditions. Regardless of their blind spots, the accuracy of the information they have found, or the soundness of the advise they have decided to follow, active IOC members seek to play a role in decisions related to their treatment.
People who have diabetes are already following news about Afrezza's journey toward FDA approval. Knowing that it is delivered via a compact discreet inhaler causes them to be interested in seeing if it can work for them. Their interest will increase as their peers discuss the medical advantages described below.
IOCs for Physicians
Doctors have ample opportunities to attend professional development sessions and access professional-level medical information online. But many professional development sessions and pharmaceutical company-sponsored websites are designed not only to educate, but also to boost sales. More and more doctors seek out the chance to talk others who are informed, but who have no agenda. Dedicated IOCs meet this need.
Some (such as Clinical Café) have a relatively broad audience: health care professionals. The most exclusive IOCs (such as Sermo, Osmosis.org, or Doctors.net.uk) verify that their members are licensed physicians, thereby providing the chance for a different form of professional growth. They allow doctors to interact with a range of colleagues to explore, reflect, seek information and advise, and share their own experiences and wisdom. They can cast the net wide or, if they are specialists (such as endocrinologists), they can participate in focused forums.
Three kinds of IOC behavior are important for Afrezza's future:
- Sorting through the conflicting claims regarding the importance of various diabetic care metrics. In the "The Structure of Scientific Revolutions," Thomas Kuhn described how during paradigm shifts, the new paradigm's champions battle with the existing paradigm's defenders. When organizations have made long-standing pronouncements premised upon the old paradigm, their leaders may succumb to the belief that maintaining their organization's credibility requires them to prove its infallibility. For enlightened people, credibility depends in part upon being able to assess the validity of new information in an unbiased way. If it seems compelling, they adapt. But most doctors get caught in the cross-fire. Besides listening to opposing spokespeople, they appreciate the chance to interact with peers who are trying to make sense out of each side's evidence and arguments. Is using the new diabetic care metrics just another fad, or will their use have staying power?
- Gathering information from early adopting peers. When doctors are thinking through whether or not to prescribe Afrezza, or how to select patients for whom it is likely to work, IOCs provide the chance to ask those with experience: "How long did it take you to get to the point where you feel confident using it?" "Does using it require extra work?" "What kinds of problems did patients experience?" and "How noticeable were any outcome differences?" Some may feel comfortable enough to ask, "Has it caused you to gain or lose patients?" As doctors first start trying out Afrezza, their focus will shift to asking more "how to" questions, such as: "How have you handled the following situation . . . ?"
- Communicating unbiased information about one's own experience with Afrezza. Several doctors who are at an early stage of using a new treatment approach want to alert others to their challenging and positive experiences. They enjoy giving to the IOC as well as receiving from it. As their experience grows with treatment approaches that they deem successful, they shift from being cautious experimenters to confident champions. They remember their own journey from being skeptical to being a believer, and know how to help current skeptics work through their concerns.
Collaborative Inquiry-Oriented IOCs for Patients and Doctors
My third category is a special kind of profit-making IOC. These provide an interesting kind of interaction between patients and physicians. They collect data from patients and physicians, aggregate it, remove the most sensitive identification information, and sell the aggregated data to medical researchers and pharmaceutical companies. Two examples are PatientsLikeMe and CureTogether. CNN Money and Business 2.0 named PatientsLikeMe one of the "15 Companies that Will Change the World." However, Natasha Singer wrote a New York Times article in which she balanced the substantial promise of such IOCs with their possible dark sides.
How Recent Research is Sparking a
Paradigm Shift That is Dethroning A1c
This section is divided into four parts: "Why A1c is Now the Gold Standard for Diabetes Care," "Recent Research About the Processes Leading to Long-Term Complications," "Three Better Blood Sugar Metrics," and "Diabetes Care in Context."
Why A1c is Now the Gold Standard for Diabetes Care
Various world class medical associations (see their position papers here and here) have crowned Glycated hemoglobin (hemoglobin A1c or HbA1c), as the key diabetic care metric. It serves as an average blood sugar level over one to three months. Since it is important to understand the reasons why it has reigned for a quarter of a century, in this section I will describe why A1c initially deserved the crown. In the next section, I will describe not why these reasons are invalid, but rather (1) how they are incomplete, and (2) why relying solely upon A1c to set treatment goals is hazardous to patient health.
Here are four reasons why endocrinologists and diabetes foundations recommend that patients have their A1c levels checked at least twice a year, and aim for a level that is below 7%, preferably below 6.5% (or below 53 and 47.5 mmol/mol, respectively):
- High blood sugar is detrimental. In a highly technical article, Giacco and Brownlee describe how high blood sugar causes long-term complications. While they describe several mechanisms and processes, the central message is:
High blood sugar --> Oxidative stress --> Tissue damage.
- Research evidence demonstrated that A1c correlates positively with the frequency of complications and the extent to which death occurs prematurely. These correlations seemed to confirm the hypothesis that the lower a person's average blood sugar, the better.
- A1c it is not affected by very recent eating patterns. Because A1c appears to represent an average of glucose levels over two or three months, the only events that would alter it substantially are blood loss, transfusions, or surgery. Therefore, physicians seem to be able to rely on A1c as a solid indicator of diabetic control, not affected by last-ditch efforts to clean-up one's act before getting tested.
- Its message is easy for patients to grasp. Its comprehensibility allows doctors to avoid spending their time explaining what it means, thereby allowing them to focus on steps that patients can take to improve their health.
Recent Research About the Processes
Leading to Long-Term Complications
H. L. Mencken famously said, "There is always a well-known solution to every human problem: neat, plausible, and wrong." A1c provides the well-known solution to the problem of selecting appropriate metrics to guide the treatment of diabetes. I have already described why this solution is neat and plausible. I shall now describe four anomalies that suggest why it is, if not wrong, at least inferior to what is now available.
Anomaly #1: A1c predicts some kinds of complications better than others. In particular, A1c predicts "microvascular" damage (e.g., problems with eyes, kidneys, limbs, or skin) better than it predicts "macrovascular" damage (e.g., heart attacks, strokes, or death).
Anomaly #2: A1c predicts long-term complications only when A1c is high. The American Diabetes Association wants A1c below 7%. The American Association of Clinical Endocrinologists has set its target at 6.5%, but its 2013 statement demonstrates awareness of all four of the anomalies described here:
"The A1c target must be individualized, based on numerous factors . . . An A1c of 6.5% or less is still considered optimal if it can be achieved in a safe and affordable manner, but higher targets may be appropriate and may change in a given individual over time."
At one time, the terms "tight control" and "intensive control" meant getting and keeping A1c below target levels. However, some of the studies that were conducted to demonstrate its benefits found that tight control did not prevent macrovascular complications in older patients with long-standing diabetes with either cardiovascular disease or risk factors for cardiovascular disease.
For one of these studies, the results were more ominous: In the ACCORD study, people with Type 2 diabetes and cardiovascular disease or additional cardiovascular risk factors were randomly assigned to "intensive therapy" (A1c below 6%) or "standard therapy" (A1c between 7% and 7.9%). Part way through the study, subjects in the intensive therapy sample were switched to standard therapy because too many intensive therapy subjects had died.
Why would tight control lead to increased mortality? Because having A1c in the target range does not mean that blood sugar levels remain close to normal. There are two different problems:
- For both Type 1 and Type 2 diabetes, many patients who have low A1c scores often have eating-triggered blood sugar surges (post-prandial glycemic excursions) that trigger oxidative stress (see here).
- For patients taking insulin, having low levels of A1c is associated with more low blood sugar events, especially severe low blood sugar events (see here). Note that when forming the sample for the ACCORD study, potential subjects were excluded if they experienced frequent or recent blood sugar events, or if they were not willing to check blood sugar levels at home. If these subjects had been included in the study, the results for the intensive therapy group would have been even more disastrous.
Anomaly #3: A1c is not a good predictor of the extent of short-term damage that causes long-term complications. Let's look more closely at the path between having diabetes and developing complications. These complications are caused not only by oxidative stress, but also by both inflammation and "endothelial dysfunction." The endothelium is the inner lining of our blood vessels. When is it working as it should, it produces dilation or constriction of the blood vessels, regulates blood clotting, helps our immune system defend our bodies from bacteria and viruses, and controls the volume of fluid and the amount of electrolytes and other substances that pass from the blood into the tissues.
A1c is a weak predictor of all three causes: of oxidative stress (see here), of inflammation (see here), and of endothelial dysfunction (see here). So why do high levels of A1c predict long-term complications? Because it correlates positively with one of the three triggers for the short-term damage that causes long-term complications. However, its correlation with the second trigger is near zero, and with the third, its correlation is negative.
These other two triggers are described below. But before getting to the other two triggers for short-term damage, note the relationships among the three kinds of damage that cause long-term complications. First, endothelial dysfunction is a cause of inflammation. Second, oxidative stress is a cause of both inflammation and endothelial dysfunction. Thus, while there are three kinds of short-term damage that cause long-term complications, oxidative stress still remains the most important determinant of their frequency and the time before they become evident.
Anomaly #4: High blood sugar is not the only diabetes-related trigger for the short-term damage that causes long-term complications. What besides high blood sugar triggers oxidative stress and endothelial dysfunction?
- Low blood sugar. Low blood sugar events (informally abbreviated "hypos" for hypoglycemic episodes) are dangerous in two ways. In the short term, they may cause mood changes, impaired executive functioning, poor judgment (leading to risky behavior and accidents), coma, and hospitalization. But they may also precipitate oxidative stress, endothelial dysfunction, and inflammation (see here), each of which plays a role in long-term complications.
- Large blood sugar swings. A very important study by Monnier et al. demonstrated not only that large blood sugar swings trigger oxidative stress, but also, of special significance to my major argument, that the correlation between the size or amplitude of the swings and oxidative stress was substantially larger than the correlation between A1c and oxidative stress.
Figure 1 summarizes the information described immediately above about the relationship to the three diabetic-related triggers of the three kinds of short-term damage that causes long-term complications. The most important pathways are highlighted in red.
Because of the four anomalies described above, A1c is at best a misleading metric. In a later section, I will demonstrate how Afrezza's performance on A1c provides an instance of "Simpson's paradox," a phenomenon well-known to statisticians in which aggregate metrics (such as A1c) lead to erroneous conclusions.
Several writers have urged that the medical community take various factors into account in addition to A1c. However, following that advise will postpone the paradigm shift (namely, replacing A1c with other metrics). During paradigm shifts, two processes occur. The first is that more and more anomalies related to the current paradigm get noticed. The second is that opinion leaders need to become convinced of the superiority of the new paradigm. Above, I described four anomalies related to the A1c paradigm. In the next section, I set out the rationale for replacing A1c with the trio described below.
Three Better Blood Sugar Metrics
Without further ado, here are my three.
Blood sugar swing amplitude. In my explanation of Anomaly #4 I explained why blood sugar swing amplitude is an important metric, and its prominence is highlighted in Figure 1. After all, swing amplitude takes into account (adds together) both the size of blood sugar surges and the size of blood sugar plunges.
Ten years ago, two Albert Einstein College of Medicine faculty members wrote an essay with the intriguing title, "Should Minimal Glycemic Variability Become the Gold Standard of Glycemic Control?" However, after describing the importance of blood sugar swings and explaining their danger, they remained within the A1c paradigm: "Glucose variability, considered in combination with A1C, may be a more reliable indicator of blood glucose control and the risk for long-term complications than mean A1C alone" (p. 178). In their defense, it wasn't until two years later that Monnier et al. reported that the correlation between swing amplitude and oxidative stress was substantially larger than the virtually zero correlation between A1c and oxidative stress.
Baseline blood sugar. Our bodies are designed to function optimally at normal blood sugar levels. Therefore, if swing amplitude is used without any other markers, then people who had small blood sugar swings that returned to a high blood sugar baseline would continually experience stress from high blood sugar. Of course, that is the reason why Hirsch and Brownlee used the phrase "considered in combination with A1c." But A1c is a crude metric for what is needed: an anchor point from which the swings begin and return. The baseline blood sugar level (often called "fasting blood glucose") is the anchor point.
According to Monnier et al.'s data, variations in baseline blood sugar were 47 times more powerful than variations in A1c in explaining variations in oxidative stress. This finding is not surprising given that A1c is merely an average: it offsets the blood sugar surges with the blood sugar plunges. The baseline blood sugar level is not an average: it is the anchor point.
The frequency of severe blood sugar events. It is important to track the frequency of moderate or severe hypos for four reasons, two of which have already been described: aversive short-term consequences and serious long-term (macrovascular) complications. Here are the other two:
- The more frequently people experience hypos, the less likely they are to take their medicine regularly, or at least to take as much medicine as they need. According to Novo Nordisk's 2010 survey, 67% of patients taking insulin are concerned about experiencing future hypoglycemic events, and doctors share patients' concerns: 74% report that they would treat closer to recommended targets if it weren't for fear of serious hypos.
- The more frequently people experience hypos, the more likely they are to snack defensively, that is to eat not because they are hungry, but because they want to prevent more hypos.
Since skipping medicine, taking too little medicine, and defensive snacking all lead to high blood sugar, hypos indirectly provide additional paths to oxidative stress.
Diabetes Care in Context
Attending to blood sugar metrics is one part of being healthy, but people who have diabetes have more reasons than those without to make a number of healthy lifestyle choices. Two have already been mentioned: taking medicine consistently and eating sensibly. The other two are exercising regularly and refraining from smoking. Physical inactivity boosts the chances of suffering macrovascular complications. Smoking boosts the chances of suffering both microvascular and macrovascular complications. (The Surgeon General's 2014 report contained some surprising news: If diabetes is defined as having a baseline blood sugar above 140, then compared to non-smokers, smokers face a 74% higher risk of developing Type 2 diabetes.)
Figure 2 displays a blueprint for good health: It shows how healthy blood sugar values and healthy lifestyle choices interact to produce good health outcomes.
Afrezza's Performance on the Replacement
Metrics Justifies Its Use for 5 Groups
This section is divided into two parts: The first compares Afrezza to NovoLog, the bestselling prandial insulin. The second describes why Afrezza can compete successfully in five markets.
Afrezza Performs Better Than NovoLog
There are five ways in which Afrezza performs better than NovoLog on blood sugar metrics:
- Afrezza reaches its peak faster. Table 1 displays the time to peak absorption for both forms of insulin. Why is reaching a peak in 15 minutes important? For people with normal metabolisms, eating sparks two phases of insulin infusion. The first one occurs within 30 minutes of eating. If the blood sugar level is still too high, a second smaller one kicks in. When people with diabetes eat, their bodies don't produce enough insulin to make a sufficient dent in the rising blood sugar level (either because of inadequate insulin production capacity or insulin resistance). In some people with Type 2 diabetes, the liver misinterprets the lack of a robust first-phase insulin response to mean that blood sugar level is too low, and infuses more glucose into the blood stream, thereby increasing the size of the glucose surge. Because Afrezza starts working in 15 minutes, it can prevent this liver misinterpretation. In contrast, Rapid-Acting Analogues are too slow.
- Afrezza exits faster. Table 1 also displays the duration of action, that is, how long before the insulin no longer has any effect. Why is it important for the duration of action to be no longer than three hours? Its mission is to metabolize the sugars, starches, and most of the protein. But after it accomplishes its mission, the longer it stays around in the body, the more likely it is to cause hypos. Rapid-Acting Analogues hang around too long.
- Afrezza's blood sugar swings are smaller. This is simply an addition issue. The blood sugar amplitude is the sum of the surge and plunge amplitudes. Since both are smaller, the swing amplitudes are smaller. Figure 3 displays how these first three points work together. Note that the average blood sugar (and therefore A1c) is slightly higher for Afrezza. But as explained above, that is because the plunges offset the surges. However, because Figure 3 is only a conceptual sketch, it is immediately followed by another MannKind-supplied graph that is based upon a several-year-old study that did not involve a comparison with NovoLog. It shows the impact of Afrezza in combination with Metformin, which suppresses glucose production by the liver, and is therefore used by people who have Type 2 diabetes. The graph shows the glucose at seven points during the day (a) before administration of the combination, (b) after 12 weeks, and (c) after 24 weeks. Because the combination was given to patients who were not on a basal insulin, the blood sugar levels remain higher than they should. But the graph demonstrates the assertion about how Afrezza shrinks blood sugar swings.
- Afrezza users have lower baseline blood sugar levels. In the Phase III trials, Afrezza users experienced a statistically significant drop, while NovoLog users experienced an increase. I am not sure why Afrezza achieves lower baseline blood sugar levels, but the fact that it consistently does so means that compared to NovoLog users, Afrezza users have more normal glucose levels more of the time. Perhaps it has something to do with the body being able to achieve a more normal homeostasis because it does not have to constantly defend itself against large blood sugar swings.
- Afrezza users suffer fewer hypos. Afrezza users suffered fewer hypos in the Phase III study for people with Type 1 diabetes; in the Type 1 study, the comparison subjects were taking NovoLog. However, Afrezza users did not suffer fewer hypos in the Phase III trial for people with Type 2 diabetes; in the Type 2 study, the comparison subjects had never taken insulin and were only inhaling a placebo powder, plus whatever oral medication they were taking. The incidence of severe hypos for the insulin-naïve Afrezza users was only 5%, and the difference between the 5% and the rate for the placebo inhalers was not statistically significant. Furthermore, no Afrezza users dropped out of the Type 2 study because of their hypos. Had the Afrezza-using insulin-naïve subjects been compared to similar subjects using a different form of insulin, Afrezza users would have suffered fewer hypos.
The above information provides the background needed to grasp why aggregate metrics such as A1c are misleading when used as a metric with Afrezza. Aggregates such as A1c sometimes mask important underlying dynamics. As 4' 10" tall former U.S. Labor Secretary Robert Reich pointed out, "Averages don't always reveal the most telling realities. You know, Shaquille O'Neal and I have an average height of 6 feet."
Reich's message is pithy and humorous, and he was well aware that people often aggregate data that are as disparate as Reich's and O'Neal's heights. But as statisticians know, Simpson's paradox is more insidious. Simpson's paradox occurs whenever considering a pooled aggregate value leads to one conclusion, but examining the value of its constituent parts leads to the opposite conclusion. Here's a 22 year-old example that will clarify what this means. It may look like a tangent, but it is not. Don't skip it. Become familiar with the nature of Simpson's paradox before working through how it applies to A1c.
Starting in 1987, the Department of Transportation asked airlines to submit monthly reports listing the percentage of their flights that were on time at the 30 busiest airports in America. In June, 1991, Alaska Airlines (NYSE:ALK) served only five of these airports. When the data for the five cities were aggregated, America West Airlines (which has since merged with U.S. Airlines, which just merged with American Airlines) performed better. That is, the percentage of on-time flights for all five airports combined was higher for America West than for Alaska Airlines. However, at each individual airport (Los Angeles, San Diego, San Francisco, Phoenix and Seattle), Alaska Airlines performed better.
To make it easier to see the underlying dynamics, Figure 4 shows the data for just two airports. The first pair of bars displays the data for both airports combined, and the next two bars display the data for the individual airports. What is going on here?
Over 95% of America West flights went to fair weather Phoenix (where 90% of their flights were on time), whereas over 90% Alaska Airlines flights went to comparatively foul weather Seattle (where 86% of their flights were on time). So, even though Alaska Airlines outperformed America West in both weather conditions, America West had a higher aggregate percentage since the supermajority of the flights whose on-time rate was being aggregated were from the fair weather airport with a higher on-time percentage. In contrast, the supermajority of Alaska Airlines flights were from the foul weather airport with a lower on-time percentage. Therefore, the aggregate percentage favored America West even though a greater percentage of Alaska Airlines passengers flying to Phoenix and a greater percentage of Alaska Airline passengers flying to Seattle were arriving on time.
What does all this have to do with Afrezza and A1c? Table 2 shows how comparing Afrezza to NovoLog on A1c is an instance of Simpson's paradox. Keep in mind that in Table 2, lower values are always better.
When the average A1c levels of Afrezza and Novolog users were compared, the average A1c level of Afrezza users was slightly higher, but the difference was not statistically significant. The A1c level is analogous to looking at the aggregated on-time rate of the two airlines. Looking at the various metrics that really matter corresponds to looking at which airline does better at each individual airport. When the two forms of insulin are compared on all the blood sugar swing metrics that correlate substantially with oxidative stress, Afrezza extensively outperforms NovoLog.
What is similar about the airline and medical instances of Simpson's paradox is that (as in all other instances of the paradox), the conclusion that would be derived from the aggregate is the opposite of the conclusion that would be derived from looking at which airline or medicine performs better on each key individual metric. What is different is that in the airline instance, the aggregate examined is the same metric (percentage of on-time flights) that is used in the different airports, whereas in the medical instance, the aggregate (A1c) allows two of the metrics (surge height and plunge depth) to offset each other so that neither doctors nor patients can predict the extent to which oxidative stress that has been produced.
Recall that the airline graphic only displayed data for two airports even though Alaska Airline outperformed America West on all five airports. Similarly, Afrezza outperformed NovoLog not only on the three blood sugar swing metrics, but also on my other two advocated blood sugar metrics (baseline blood sugar level and frequency of hypos). In both cases, the differences were statistically significant.
Please join with me in singing the chorus: A1c needs to be replaced because it is misleading, especially when it is near or below the target range.
There is an important implication for understanding the Afrezza vs. NovoLog Phase III results. The endpoint was to demonstrate non-inferiority, using A1c as the standard. That is what happened, because although A1c was lower for NovoLog, the difference was not statistically significant. But if A1c is set aside (both because it is a misleading indicator, especially when A1c values are low, and because it does not correlate well with oxidative stress) so that the comparison is confined to metrics that correlate well with oxidative stress, then Afrezza demonstrated superiority in the Afrezza vs. NovoLog Phase III trials. This is what endocrinologists and insurance companies will come to recognize.
Table 3 assembles on one page the key reasons that Afrezza performs better than the competition. I encourage you to spend a little time seeing how it all fits together.
Afrezza's Performance on the Replacement
Metrics Justifies Its Use for Five Groups of Patients
Because Afrezza performs competitively on all three substitute metrics, encourages healthier choices for two lifestyle factors, and performs better on the intervening variables that affect four health outcome measures (to refresh your memory, revisit Figure 2), it can promote its benefits to five populations.
People with Type 1 diabetes. For the reasons discussed in the previous section, this should be an easy sell.
People with Type 2 diabetes who do not use insulin. These were the people who were involved in the Phase III trial involving Type 2 diabetes. Compared to subjects who inhaled placebo power, they had lower A1c levels, lower baseline blood sugar levels, and smaller blood sugar surges. They had a slightly higher incidence of hypos and averaged a 1% gain in weight. It is my guess that once patients got used to Afrezza, they would begin losing weight. This is a hypothesis to be tested in future studies.
Here is my reasoning. The 5% of Afrezza users who experience serious hypos will quickly begin to recognize what slightly low blood sugar feels like, signaling them to eat a high carbohydrate snack to prevent a deeper plunge. Accordingly, the number of severe hypos will decline. Those who don't develop the skill of sensing when their blood sugar levels are declining may search for a different medication. However, the 95% of Phase III subjects with Type 2 diabetes who did not experience any severe hypos while taking Afrezza will want to continue to do so, and will communicate their success to others.
Another guess that needs to be tested in future studies is that the more that Afrezza (1) reduces the amplitude of this population's blood sugar swings, and (2) reduces the baseline blood sugar level, the more the liver and any other parts of the body that are affected by insulin resistance will be protected from the challenges triggered by high blood sugar and blood sugar surges, so the rate of insulin resistance and organ dysfunction will be slowed or halted.
People with Type 2 diabetes who only use a basal insulin. Basal insulins provide only limited protection against after-meal glucose surges, which is a key source of oxidative stress. Afrezza attenuates post-meal surges.
People with Type 2 diabetes who already use a prandial insulin. While patients who use a prandial insulin have more protection from post-meal surges than those who don't, Afrezza provides better protection against large surges because it acts faster, thereby preventing the liver from adding an additional source of glucose to the bloodstream. It also provides better protection against hypos because its duration of action is shorter. That is, it is eliminated from the body before it can cause the blood sugar to drop as low as it may go with the other currently available prandial insulins.
People who are judged to be pre-diabetic. This is a population that Maredin Capital Advisors wrote about (see the link in the context section), so I won't add anything here except to repeat the argument for the people with Type 2 diabetes who do not use insulin: namely, that early treatment with Afrezza will reduce blood sugar swing amplitude, lower baseline blood sugar, and therefore probably slow down the process of becoming diabetic, meanwhile reducing the extent of oxidative stress, endothelial dysfunction, and inflammation that leads to long-term complications.
Implications for Insurance Companies,
Patients, Doctors, and Investors
Insurance companies are interested in treatment efficacy, but when diabetes is involved, they are also interested in costs: of medication, near-term hospitalizations, and long-term complication-related interventions. Of course, the second and third are a good proxy for treatment efficacy. The likely cost of Afrezza has been covered in numerous Seeking Alpha articles.
Near-term hospitalization can occur from high blood sugar-induced comas, hypo-related accidents, and hypo-induced comas. Because Afrezza lowers baseline blood sugar levels (and also reduces the size of blood sugar surges), Afrezza users should have fewer high blood sugar-induced comas (than patients using other medications). Because it decreases the incidence of hypos, especially severe hypos, Afrezza users should have both fewer hypo-related accidents and fewer hypo-induced comas.
Because Afrezza performs well on all three metrics that predict long-term complications (and also because it leads to better treatment compliance and more sensible eating), the frequency and severity of such consequences should be lower, and more time should pass before any onset.
If the cost charged for Afrezza is in the expected range, then data related to treatment efficacy and insurance companies' own in-house projections of near-term hospitalization and long-term complication-related costs are likely to prompt them to be Afrezza advocates.
Potential Afrezza users care about all of the issues that concern insurance companies, but there is no need to repeat what has been said in the section on insurance companies. In addition, potential users are concerned about the user-friendliness of various medicine options and a social consequence of taking insulin (see below). Regarding Afrezza's user-friendliness, other have already written about the advantages of (1) inhaling over injecting, (2) the delivery device being small, and (3) once the correct dose has been determined, using pre-loaded cartridges.
Perhaps less widely known to investors are the results of a recent study that first challenged subjects with a 100-gram carbohydrate (100 CHO) meal and used the results to determine the optimal Afrezza dose for each subject. Then, the investigators tested how subjects responded after inhaling their optimal dose (on separate occasions), to three different meal challenges: a zero CHO meal (that is they didn't actually eat any food), a 50 CHO meal, and a 200 CHO meal. They discovered that the largest blood sugar surges remained below American Diabetes Association targets, even for the 200 CHO meals. They also discovered that after taking their optimal dose of Afrezza but not eating, the plunges averaged 32 mg/dL after 60 and 90 minutes, which means that subjects can skip meals without severe hypos.
Naturally, no one is recommending that anyone take Afrezza unless they plan to eat. The point here is that despite the fact that people with diabetes need to continue to eat sensibly, like everyone else, potential Afrezza users will appreciate the freedom of eating like their sensible-eating friends and family members without carefully calculating carbohydrate grams.
There has been some discussion in the comments following Seeking Alpha articles about fear of needles. Here is my take: People don't enjoy being injected, but few, other than very young children, are afraid of the small needles that are currently being used to deliver insulin. People with diabetes find injecting themselves a hassle, but a reasonably minor one.
However, people with diabetes do fear three things related to their disorder: the medical consequences of serious hypos, the social consequences of even moderate hypos (that some of their actions when they are experiencing low blood sugar may not only cause them embarrassment, but also alter how others think about them), and getting and experiencing long-term complications. Fortunately, because of the way that it works, Afrezza minimizes the extent to which its users experience any of these three nightmares.
Many potential Afrezza users will participate in IOCs to explore and discuss the topics covered in these four paragraphs, along with the supporting information above.
Physicians also need to attend to all of the issues that concern insurance companies and patients, but there is no need to repeat the material in the sections on insurance companies and patients. Before taking up additional issues that concern doctors, I shall describe my vision of how they will use the three blood sugar swing metrics in caring for their patients.
- Blood sugar swing amplitude (Mean Amplitude Glycemic Excursion, or MAGE. It is safe to skip the following technical note: MAGE is not actually a straight average of all blood sugar swing amplitudes within a 24-hour period. Rather, it is the average of the largest swing amplitudes. Why? Because it is the largest swings that create havoc, and it is the average of the largest swing amplitudes that is strongly correlated with oxidative stress.) I do not expect doctors to ask patients to collect the data needed to compute MAGE, either on a routine basis, or perhaps ever. Rather, I see physicians, especially endocrinologists, selecting medications on the basis of their established profiles. Of course, not every patient responds to a given medicine in the same way, so when, because of its ability to minimize MAGE, they prescribe Afrezza to a patient who has not taken it before, they may ask the patient to take blood sugar readings at various times during the first few weeks or months to ensure that Afrezza is working as expected with him or her. Thereafter, there would be little point to calculating MAGE again for that patient.
- Baseline blood sugar. Doctors will ask patients to check their baseline blood sugar regularly, but how frequently should depend upon how stable the readings are from assessment to assessment.
- Frequency and severity of hypos. At minimum, patients should note down every hypo, briefly describing the time it occurred, and whatever they remember regarding (1) how many hours had elapsed since they last ate, (2) whether they had engaged in an unusual amount of exercise, (3) whether they had experienced strong emotions, or (4) whether they may have changed the amount of basal insulin they were taking (if any). Any one of these may deplete their blood sugar level. Both patient and their doctors can look over the data to detect patterns and derive implications for action.
Like most people in all walks of life, doctors experience a justifiable reluctance to move away from the routines and tools that they have already mastered. When someone encourages them to try a new form of treatment, they experience a variety of psychological forces. Figure 5 contains two diagrams: Diagram 1 shows the psychological forces that result in doctors deciding to continue relying upon the form of treatments with which they are familiar. Diagram 2 shows the psychological forces after doctors have worked through their concerns and have decided to try out the new form.
When the new form of treatment is inhaled Afrezza, here's a quick summary of the concerns that they need to dispel for them to overcome the concerns shown in Diagram 1 and embrace the beliefs shown in Diagram 2. Included are steps that may help them make the transition.
- The concern that the new practice may be too difficult for them to master. It is a straightforward matter for doctors to calculate the optimal dose (typically either 15 or 30 units). It is also easy for doctors or their nurses to show patients how and when to use Afrezza's delivery device. Most physicians, especially endocrinologists, already look at more than one metric. But if they have not been doing so already, they can quickly find out from colleagues how to make sense of baseline blood sugar values and the frequency and timing of hypos.
- The concern that costs may be too high. From talking with colleagues who have preceded them, they will discover that they need to spend only minimal time and effort to master and perform on an ongoing basis the tasks just described.
- The concern that the cost-benefit ratio may be too low. I have already described why the time and effort costs are low. The other part of the ratio is the benefit for patients: lower surges, fewer hypos, smaller swings, lower baseline blood sugar levels, greater treatment adherence, more sensible eating, and a reduction in the chances of, or a delay in the onset of, developing complications. And there is another possible benefit for general practitioners, internists, or family medicine physicians who refer their Type 2 patients who are starting to take insulin to endocrinologists to take over managing their diabetes: If they prescribe Afrezza, there is no need for endocrinologists to take over managing the diabetes of many such patients, because once a patient's optimal dose has been determined, supporting good diabetic care is comparatively smooth sailing. Thus, in addition to the benefits to their patients, some doctors will have the added benefit of being able to continue managing the diabetes of some of their patients whom they otherwise would have referred to endocrinologists.
- The concern that they might not meet their personal needs. There is nothing about Afrezza that will weaken their ability to feel in control, since they will be able to continuing applying the knowledge, skills, and wisdom they have so far accumulated. If they so desire, they can continue to assess their patients' A1c twice a year, even though I have argued that low values of A1c provide misleading information. Inhalation may be a different form of treatment, but there is less need to be precise about Afrezza doses once a given patient's optimal dosage has been determined. Of course, if their patients are taking other medicines to treat their diabetes, physicians will still need to monitor and adjust those as they have done before.
Doctors who participate in IOCs will have the chance to ask their colleagues about any of these issues, as well as the underlying physiology or research evidence. Because of Afrezza's attributes and benefits, fellow IOC participants will provide information that will allay their concerns. Thereafter, they are likely to respond positively when hopeful patients ask to see if it works for them.
Investors need to make their own judgments. To assist in this process, review the seven catalysts that will promote commercial success and increase shareholder value:
- FDA approval,
- finalization of partnership arrangements,
- the ongoing paradigm shift that is changing the metrics used to guide treatment as a result of the recent research about the physiological processes that result in long-term complications,
- the growing recognition of how Afrezza's superior performance on the new paradigm metrics can protect and enhance both quality of life and long-term health,
- insurance company acceptance of Afrezza,
- the growing awareness that Afrezza can compete for the allegiance of five groups of people, and
- the role of IOCs in increasing patient curiosity, providing avenues for both patient and doctor exploration, as well as channels for sharing information, experiences, and satisfaction.
Disclosure: I am long MNKD. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.