What will Exubera cost? Except perhaps for Pfizer executives, nobody knows.

If anyone else does know, maybe the pharmaceutical company required distributors and health insurance companies to sign non-disclosure agreements to keep it under wraps.

Pfizer could be re-evaluating the U.S. price for the new inhaled insulin following the U.K. healthcare watchdog NICE’s recent conclusion that Exubera’s signficantly increased cost was not justifiable for treating the vast majority of diabetics under the country’s national health insurance plan.

Some analysts predict that Exubera will cost up to 4 times as much as injectable insulin. A widely read analsyis, “Irrational Exubera-nce for Pfizer,” Michael J. Russo and David Balekdjian of The Bruckner Group, Inc. forecast that:

At launch, Exubera will be priced at a premium to injected insulin far in excess of its proven pharmacoeconomic value, yet without data demonstrating significant compliance improvements that could justify the price. Payers most likely will not — and cannot afford to — pick up the tab.

If the inhaled insulin is put on the market this summer, we should find out the answer within another six months.

A drug industry publication in India, Express Pharma, recently gave Pfizer its ‘Innovative Product of the Year’ award on May 26, 2006 in Mumbai, India for the FDA-approved Exubera inhaled insulin.

Although Exubera is reportedly being fast-tracked for approval in India, with Pfizer officiials hoping to launch inhaled insulin sales by 2008 to the country’s rapidly growing population of 35 million diabetics in a country of 1.1 billlion people.  No information is currently available from India’s Ministry of Health and Family Welfare.

Publication Cites Exubera Safety Risks, Despite Award

In spite of  bestowing this award upon Pfizer, Express Pharma cites a number of drug safety concerns over the use of Exubera:

This form of insulin, however, comes with some minor drawbacks. The first is that 90 percent of the insulin is lost in the air and nasal passage. Until a few years back, when insulin was rare, hard to cultivate and therefore expensive, such wastage of the protein was unacceptable.

However, with the synthetic forms of insulin readily available, this form of treatment is feasible.

Other potential side affects and areas of concern inlude 1) a diabetic patient’s development of antibodies to Pfizer’s inhaled insulin, due to an increased level of immune activity by Exubera users, and 2) potential pulmonary risks and decrease lung function because:

[I]nhaled insulin alters the function of the lung by one percent. Therefore, the capacity to blow out a certain volume of air in a second is slightly reduced for a person using this insulin.

Because India is home to the world’s largest diabetic population, it is not surprising that the country’s pharmaceutical executives are, according to Red Herring, “chomping at the bit” to get the drug on the market as fast as possible, irrespective of any risks.

Is marketing Pfizer’s new inhaled insulin as being just as good as injectable insulin a Pfizer marketing ploy, even if there may not be proof of such an effect on diabetics? Does the company want to sell billions of dollars worth of the new drug? These are obvious question diabetics should be asking themselves if they’re even considering trying the new inhaled insulin from Pfizer and Nektar Therapeutics when it hits the market in July 2006.

As Exubera-Risks.com observed yesterday, the U.K.’s healthcare watchdog NICE found that Pfizer’s submissions to the agency in favor of winning approval for coverage by Britain’s National Health Service were “positively misleading, and biased in favour of inhaled insulin,” and that the “most misleading bit in Pfizer’s materials” was that they purported to show that Exubera gave diabetics “better blood sugar control” when, in fact, “the [inhaled insulin drug] trials showed no difference.” According to NICE, the research just wasn’t there to support Pfizer’s position.
This is not the first time that marketing efforts appear to have been used to try to convince diabetics and health care providers to use Pfizer’s new FDA-approved drug that insulin shots can be scary, horrific, or bad. Is playing on patients real, or newly-imagined fears, an effective inhaled insulin marketing strategy?  It’s too early to know the answer, since Exubera has not yet hit the market.

However, it seems much like the strategy used to recruit patients to participate in the Exubera “Breeze Study” mandated by the FDA.  The FDA, as a pre-condition of Exubera approval, directed Pfizer to conduct a series long-term safety studies for pulmonary risks, use by children (currently prohibited) and other safety risks for diabetics using Exubera inhaled insulin over an extended period of time (5 - 7 years).

The U.K.’s NICE Assesment Group for Pfizer’s Exubera inhaled insulin consists of:

• Aberdeen Health Technology Assessment (’HTA’) Group
• U.K.’s National Health Service (’NHS’) National Coordinating Centre for Health Technology Assessment

The Assessment Group made a number of significant observations. They are excerpted here, and reprinted in their entirety below:

• “The educational material on inhaled insulin is positively misleading, and biased in favour on inhaled insulin.”
• The most misleading bit in Pfizer’s materials evaulated by the Committee about “better blood sugar control”; the trials showed no difference. We also heard at the meeting that adjusting dose would be more difficult with the inhaler than with pens, because several doses might be required.

Here is the Assessment Group’s complete recent June 21, 2006 position on the second Appraisal Consultation Document:

—

Assessment group comments on responses to consultation and ACD.
PFIZER comments.

1. Delay to achieving control

Pfizer say that the delay to initiation of insulin therapy is longer than we thought. They state that their data were only 4 months old at submission, but what matters is the period of observation not of data extraction. The analysis of the THIN database covered the period 1987 to 2004. The DIN-LINK data (referred to as the Freemantle study in the Pfizer submission) was more recent – 1995 to early 2005, with sub-analysis of the period after Oct 2002. Analysis of the later period could not provide data to 4 years, but the curves looked similar – i.e. people with type 2 diabetes are often poorly controlled but many are not switched to insulin. Interestingly, many are not switched to more than one oral agent either. The logical step in intensification would be to increase OHA treatment. The recent PROactive study showed that adding pioglitazone in those poorly controlled on 2 OHAs decreased HbA1c by about 0.6% (reduction 0.9% with pioglitazone, 0.3% with placebo). Of those who were on 3 OHAs, 50% were on insulin by 3-4 years (figures 1c and 2a).

We do not have data on the effect of the new GP contract and the quality standards therein (and neither do Pfizer – it is too recent) but we note the comment from the clinical experts before the first meeting that the new contract has increased the number of people with type 2 diabetes referred for insulin therapy.

All this is somewhat academic – we carried out a sensitivity analysis (reported on pages 95-6 of the TAR) in which we used the Pfizer figure of a 4-year delay. This results in a slight increase in HRQoL gains but does not render Exubera cost-effective. (emphasis in original)

2. The likely increased uptake of inhaled insulin

Pfizer provide additional data from the Real World Feasibility study in a slide set, which we had not previously seen. These patients had never injected insulin, and our hypothesis was that they would not know how good modern fine needle injection devices were – their knowledge of needles would be from venepuncture.
The educational material on inhaled insulin is positively misleading, and biased in favour on inhaled insulin.
Slide 12;
“People who used inhaled insulin reported better blood sugar control, found it easier to adjust their insulin dose….”

“Most were more satisfied with their inhalers and found them more convenient than insulin injections”.
Slide 9;
“This device is like inhalers that people with asthma use but it is slightly larger”.
The underlining is our’s. From memory of the demonstration of the inhaler at the first AC meeting, we guess that it is about 6 times the volume of an asthma inhaler?
The most misleading bit is about “better blood sugar control”; the trials showed no difference.
We also heard at the meeting that adjusting dose would be more difficult with the inhaler than with pens, because several doses might be required.
Other causes of possible bias include:

-injected regimens are made to look more complex by detailing many different injected
regimens versus one inhaled.
- the need for additional therapy to provide basal cover is stated as “You can take diabetes pills or other types of insulin injections as your basal dose”.
At the foot of page 3 of the Pfizer comments, there is the following text;

“ …Exubera would be cost-effective in the majority of patients uncontrolled on their current regimen who required ….insulin therapy”.

The assumption underlying such statements is that switching to insulin results in good control, assumed (Pfizer submission page 54 table 19) to mean an HbA1c of over 7.4%.

Pfizer argue that people switch earlier with inhaled, and thereby get control as good as this earlier.

In previous comments (responses paper submitted to first AC meeting) we noted two sources of data that people on type 2 treated with insulin did not achieve good control – the Lothian audit data and the Hayward study.
The Freemantle study, unpublished but provided as part of the rather large Pfizer submission (about 281 pages, not counting the manual for the EAGLE model), has some interesting data in appendices which we reported on page 62 of the TAR but will expand on here. The ones of most importance are those on improvements in HbA1c after people with type 2 diabetes uncontrolled on OHAs are switched to insulin.
Table 3b shows that HbA1c improves but that 73% of patients have HbAc1 above 7.5% and 44% have HbAc1 above 8.5% at one year after commencing insulin. These figures are not compatible with Pfizer’s assumption of good control with Exubera.

Page 22 of the Freemantle study gives proportions of type 2 patients uncontrolled on insulin, in 2004-5. In those on basal insulin alone, 72% were uncontrolled using Pfizer terminology (HbA1c >7.4%). In those on basal bolus, 59% were uncontrolled. For premix, the figure is 66%.
Switching to insulin does not give good control in most patients.

Page 23 gives the results of intensification of insulin therapy – e.g. moving from basal to basal + bolus, for type 2 patients. This would be the scenario if Exubera at meal-times was added to basal insulin.
90% of patients did not achieve an HbA1c under 7.5%, and 51% did not achieve HbA1c under 8.5% at 12 months. They did see a drop in HbA1c of 0.8%. The drop after adding premix to basal was less – 0.5% (similar to the effect of adding pioglitazone to 2 OHAs in the PRO-active trial).

Other data from the Freemantle study shows trends over time in whether patients who start insulin remain on oral agents – about 43% in 1995, about 70% in recent years (estimated from figure 7). So practice is changing.

3. Patient quality of life

We have nothing to add to what we said in our responses to consultees comments before the ACD meeting (copy attached for convenience). We had reservations about the Accent study, described by Pfizer as “scientifically robust”.

4. Patient choice.

We accept that many patients would prefer to inhaled insulin rather than inject it. But from a public health perspective, we need to distinguish what people want from what they need. If we fund the extra cost of inhaled insulin, other forms of health care will need to be sacrificed.
Why should other patients be deprived of care because some people with diabetes prefer
inhalers?
For those with type 2 diabetes, most of whom are overweight, there are other choices than having to inject insulin. Diet, weight loss and exercise can reduce blood sugar.
Pfizer’s comments about “process flaws etc”.
Some of these points were raised by Pfizer before.
1. Comment about improving the uptake not being addressed. Chapter 4 of the TAR has a section on reasons for reluctance to start insulin (pages 51-53). The information alluded to by Pfizer is the study by Freemantle and colleagues (2005) – the hypothetical one about which we have received further data in copies of the slides, referred to earlier. We think this study was biased in favour of inhaled insulin. It was funded by Pfizer and Aventis.
2. Remit not fully covered – “cost-effectiveness in type 1 not examined”. Our reasons for not modelling the two type 1 groups (Pfizer A and B) are explained in the TAR (pages 55-6). In brief, the evidence suggests clinical equivalence of inhaled and short-acting conventional injected; the assertions made by Pfizer of increased willingness to intensify therapy come from Freemantle et al 2005, which is not relevant to this group.
The Pfizer comment about a contradiction by us is probably due to them confusing their
groups A and B with ours (which were equivalent to their C and D).
3. Pfizer issue 3, page 6 and 7.
Pfizer complain that we have excluded some trials which they think are relevant. Our reasons for so doing were given on pages 36 to 39 (“Additional studies”). In brief, these studies compared insulin regimens to oral regimens which were less than maximal, such as sulphonylurea monotherapy, and in two studies even to diet and exercise alone, without any drug treatment to lower blood sugar.
Pfizer state that these studies were not included in the review. They were, on pages 36-39, with the exception of a recent diet and exercise versus insulin one.
Most of these studies were conducted by Pfizer.
4. Page 8, modelling of cost-effectiveness for only two groups.
The reasons were explained in the TAR. At the end of their comments, Pfizer give figures for the costs of the various regimens with needles and pens added. They do not explain how the large rises in costs of injected regimens arise. For example, they say that the cost of injected basal bolus rises by almost £200. After allowing for the cost of separate pens for basal and bolus, this allows for 4 needles a day.Needles are reusable and can last for several days, so this is an over-estimate. They then use the new costs to re-do their ICERs.
Comments from Lilly.
We note the comment about research evidence on optimum insulin regimens, where they cite several references as showing that mixtures may be better than single long-acting basal analogues. Their citations are selective and other studies have shown that basal glargine plus OHAs is better (Janka 2005), or that once daily NPH plus OHA is better than twice daily NPH. Lilly do not have a long-acting analogue on the market, so it is in their interests to argue against basal analogues.
Previously sent to NICE and attached for convenience.
Assessment group responses to comments from consultees, pre-ACD.
It is not our intention to respond to all the comments from the consultees, but only to discuss selected issues which are important to the economics, especially where the consultees have made comments which provide further information to that in the published literature.
As noted on the first page of the Pfizer comments, we are in broad agreement regarding the clinical effectiveness of Exubera.
There are three points in particular which affect the cost-effectiveness. The first is the estimated delay in starting people with poorly controlled type diabetes on insulin. The second is the utility gain from using inhaled rather than injected. The third is the Pfizer assumptions about compliance and control.

1. Delays in starting people with poorly controlled T2 diabetes on insulin.

We thought the data provided by Pfizer on the duration of time which people were left poorly controlled, based on GP data from the years 1995 to 2005, would now be out of date, and we expected earlier conversion. We were interested to note the comments by Professor Amiel on the impact of the new GP contract, that “primary care is not accepting HbA1c levels of > 8% without attempting intensification of therapy”. Professor Barnett also makes this point (comment 12) – “The new GMS contract means that many patients are being referred with HbA1c just above this level (7.5%) for insulin treatment”.

We also note Professor Amiel’s comment about “the accelerating acceptance of once daily
basal insulin as the first insulin therapy for type 2 diabetic patients failing to achieve control on oral agents” coupled with the increasing use of glargine, which she reports has “greatly facilitated initiation of insulin therapy in primary care”. Jill Hill and Louise Hilton for the RCN also note that the arrival of glargine “has facilitated an increasing number of patients having insulin initiated in primary care”.
So it looks as if the new GMS contract is accelerating conversion to insulin treatment, and that the availability of long-acting analogues such as glargine and detemir is making initiation easier. Nevertheless, the additional analysis presented by Pfizer, showing the DIN-LINK data only for the most recent years (figure 2 of their submission), suggests that many patients remain poorly controlled. However this figure does not tell us how many of the poorly controlled patients are already on insulin. It would have been more useful to examine only those on two or more oral diabetes drugs. Perhaps Pfizer could be asked to provide that?
Pfizer have submitted another similar study, this time using the THIN database, which reports a similar delay of 4-6 years in starting insulin, but the data cover the whole lifespan of the database, 1987-2004, and may not represent the situation now, after UKPDS, glargine and the new GMS contract.
2. The utility gain from inhaled rather than injected.
We assumed that injections would not be a major problem for most people. We note that several of the consultees shared this view; Mrs Wallymahmed, nurse consultant, comments that anxiety about starting insulin is often quickly allayed after the first few injections.
Jenny Hirst of the Insulin Dependent Diabetes Trust says that “the fears of injections may be based on a lack of awareness of the modern injection devices and the very fine small needles that are presently available”.
Incidentally, we described the Freemantle study as hypothetical because the patients involved had never injected, not because inhaled insulin was not then available (Prof Barnett’s comment 7). We expected that patients with no experience on injecting would over-estimate the pain because their experience of needles will be mainly from venepuncture. The pain in intensified insulin regimens comes more from self-testing of blood than from insulin injections, and such testing will be required whether short-acting insulin is inhaled or injected.
The utility study from Accent was submitted as part of the Pfizer submission. 344 people took part. They were not randomly selected, but came from;

• a list provided by Pfizer of people with diabetes who were members of Pfizer health
care panels
• two commercially available lists of people known to have diabetes
• adverts placed in some local newspapers.

The submission does not say how many came from each source. The main pilot was carried out by Pzifer and Innovus employees and the data from the 29 patients involved were included in the total results.
The participants were asked how they might feel on Exubera and how they might feel on
injections, after information was provided by the Pfizer-sponsored Accent group. It was a hypothetical study rather than one which showed an actual difference in HRQoL in people on the two forms of insulin. About 22% were already on injections.
We cannot assess what biases may have crept in. The letter to potential participants stated that “The objective of the study is to assist in the development of a new drug delivery mechanism” and so those unhappy with existing mechanisms may have preferentially participated. The information for patients leaflet mentioned “the impact of using different methods of taking insulin on quality if life”. An independent study of a representative sample of people with diabetes would have been preferable. Our main concern was the unexpectedly severe ratings sometimes given to the difference between inhaled and injected (such as the difference being rated as being as bad as bedridden). These extreme values increase the utility difference.
The quality of life impact of having to inject was assessed in insulin-naïve patients, and the injections were not their greatest concern. When patients not on insulin were asked about concerns about moving on to it, the greatest concern (people reporting that they were “very concerned” ) was “having reached a more serious diabetes stage” (63%) followed by “longterm effects on health” (44%) and then “necessity to inject” (39%). For completeness, other “very concerned” ratings were “need to make lifestyle changes” 30%, and “being able to use the pen or injection needle correctly” 28%.

The spread of concerns regarding need to inject were;

not at all concerned - 25%
indifferent - 13%
a little concerned - 21%
very concerned - 39%

but these were patients who had never experienced a fine insulin needle, and we know that most people cope well with injections.

The Accent study reported that most people would prefer inhaled insulin to injected. We do not think this is in doubt; it matches the data from the trials. The key issue in the economics is the size of the utility benefit from inhaled over injected. The Accent study reports that amongst those who preferred inhaled insulin (note that a quarter to a third preferred injection – Table 22) the TTO gain was 0.10 to 0.14, which seems somewhat high. For comparison, the figure given in the industry submission for the utility loss from having to have haemodialysis ranges from 0.04 to 0.117. Are Pfizer really saying that switching from injected to inhaled has the same utility gain as avoiding dialysis? A TTO value of 0.1 implies that people would give up 10% of life expectancy in order to have inhaled rather than injected insulin at meal times.

This seems surprisingly high.

Another study (not included in the assessment report) is relevant. Professor Anthony Barnett and colleagues (1996) assessed quality of life amongst patients with type 2 diabetes treated with insulin compared to oral drugs, as part of an RCT. One group remained on oral agents; the second were randomised to insulin by syringe; and the third to insulin via a pen. The authors summarise their conclusions as;
“We conclude that in moderately controlled Type 2 diabetes, there is no significant change in quality of life when patients are switched to insulin, although once on insulin, most prefer an insulin pen to a syringe.”
This study is more useful than the Freemantle one because the patients had experienced insulin injections. The authors also report no differences in the anxiety or well-being subscales.

3. Pfizer’s assumptions about compliance and control.

Table 19 of the submission (with the six subgroups A to F) apply the figures from the Freemantle hypothetical study (which was in those who had never injected) to all the subgroups; they are only relevant to two of them.

There are three key assumptions. The first is that with inhaled, 35% of poorly controlled type 2 patients switch immediately versus only 15% with inhaled. The second is that all those who switch are compliant – e.g. subgroup F on inhaled : “All the patients are compliant”. The third and most important is that all achieve control on inhaled insulin (because of compliance), whereas those who do not switch remain uncontrolled.
We know from the Lothian audit data that control in people with type 2 on insulin is not good – they do not reach the control assumed by Pfizer. The data are in the public domain via the RCPE website, having been presented at a conference.

treatment number mean HbAc1
Type 2 on insulin 5030 8.5%
Type 2 on orals 8007 7.6%
Type 2 on diet 2517 6.9%

In addition to showing that control remains poor despite insulin treatment, the numbers in each group, and the good control in those on oral agents, suggest a substantial switch to insulin.

Another study on starting insulin in patients with Type 2 diabetes (Hayward et al 1997) reports that;

“Insulin therapy was rarely effective in achieving tight glycemic control”.

Two years after starting insulin therapy, 60% still had HbAc1 levels of 8% or greater. 25% had levels between 8.0 and 8.9%; 20% between 9.0 and 9.9%; and 15% had levels over 10%.

The regrettable truth is that even with perfect compliance many patients do not achieve good control. None of our current methods of insulin treatment are really satisfactory and inhaled is no different. It is absorbed into the wrong part of the circulation (systemic not portal) and cannot mimic the time profile of natural pancreatic insulin.

4. The need for fine adjustments.

We note the comments from Stephanie Amiel, and Jill Hill and Louise Hilton, about the need for fine adjustments of insulin dose by patients using the DAFNE approach and that such adjustments would not be possible for those requiring adjustments of 0.5 to 1 unit of insulin, and much more cumbersome for those with type 2 requiring larger doses; whereas the adjustments can easily be done with a pen. Hence for some patients, inhaled insulin would not be suitable.
Miscellaneous comments.
There are a number of statements in the Pfizer comments on the assessment report which are not quite accurate. For example, on page 3, referring to subgroup A, Pfizer say that in the TAR, “it is stated that there is no benefit for this group” whereas what we said was that there were no data on whether there is benefit or not.
On page 1, they say that “that the unrestricted offer of full trial reports was refused by the TAG”. What they don’t say is that we declined fuller versions of some of the trials because we did not think the efficacy of inhaled insulin was in doubt; we told them that we were happy to agree that inhaled insulin lowers blood sugar just as well, but no better than, shortacting injected.

Professor Amiel comments that in the cost-effectiveness analysis we used short-acting analogues as the comparator, whereas in the inhaled insulin trials the comparisons were with conventional injectable insulins, and that this may make Exubera seem more cost-effective that it is. This is a fair point. Analogues may be slightly better for blood glucose control than Exubera, but no trials have been done.

Both Mrs Wallymahmed and the RCN suggest that inhaled insulin might be useful for people with lipohypertrophy. However, there are adipocytes in the lung too, and we do not know what would happen in the lung in those prone to lipohypertrophy.

Norman Waugh for Aberdeen HTA Group.

Thursday’s N.Y. Times has an excellent article on the 4 new diabetes drugs paraded about at the ADA meeting in D.C. (Exubera, Byetta, Galvus and Januvia).

Dr. John Buse, the ADA’s Vice-President and director of the diabetes center at the University of North Carolina School of Medicine.

Given the potential pulmonary risks found in Exubera clinical trials, Dr. Buse says, “[p]atient and doctor will have to weigh very carefully what the benefit and the risk really is.”

Candid caution from a consultant from for the company, or companies, that developed the new inhaled insulin.