New Drugs for Osteoporosis

New Drugs for OsteoporosisComparison of the Costs and Required Returns WithThose of Other Drugs Intended for Long-Term Use

Martin Edwards

Novo Nordisk A/S, Copenhagen, Denmark

ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2691. Trial Design Considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2702. Cost and Return Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271

2.1 Costs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2712.1.1 Time Lines for Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2732.1.2 Costs per Year of Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273

2.2 Returns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2732.3 Matching Costs and Returns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2732.4 Attrition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2742.5 Discount Rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 274

3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2754. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275

Abstract Specific regulatory guidelines dictate that developing a new drug for osteo-porosis will be significantly more expensive and take at least 2 years longer incomparison with other long-term therapies developed using the InternationalCommittee on Harmonisation (ICH) general guidelines. Assuming similar attri-tion rates, the minimal additional uncapitalised cost is $US86 million for non-estrogen osteoporosis compounds following a minimum programme designed togain indications for both treatment and prevention. The excess expenditure iscreated by the size requirements for phase III fracture trials in both the EuropeanUnion (EU)/US and Japan. The after-tax cash flows to the point of launch dis-counted at 11% are $US102 million greater, reflecting the additional effect ofdelayed time to market. Assuming similar lifecycles, the peak sales required toreturn the investment on an osteoporosis drug will be a minimum of $US95million greater per launched compound.

Many ongoing osteoporosis programmes are substantially larger than the the-oretical minimum. The costs of substantially increasing the sample size in phaseIII trials mean that blockbuster revenues will be required to break even. However,the potential cost of a delayed launch because of fracture efficacy being in-completely proved is so substantial that fracture trials need to be powered con-servatively to decrease the chances of this eventuality.

REVIEW ARTICLE Pharmacoeconomics 1999 Mar; 15 (3): 269-2781170-7690/99/0003-0269/$05.00/0

© Adis International Limited. All rights reserved.

In this paper, the costs and required returns fornew osteoporosis compounds are estimated andcompared with those for other drugs for long-termtherapy developed in accordance with Interna-tional Committee on Harmonisation (ICH) princi-ples. Costs are estimated for global registration, inwhich the costs for Europe and the US are consi-dered as a single programme with the additionalcosts for Japan and other markets considered asadd-ons.

Osteoporosis is a disease characterised by a lowbone mass which predisposes to fractures.[1] Frac-tures typically affect the spine, wrist and hip;postmenopausal women are particularly at risk.The financial burden of osteoporosis to society isenormous, with direct annual costs in the US esti-mated at $US10 billion of which $US8 billion isaccounted for by hip fractures (1996 values).[1,2]

1. Trial Design Considerations

Regulatory authorities[3-5] and research groups[6,7]

have developed guidelines with respect to the nec-essary preclinical and clinical trials required to filelicence applications for osteoporosis indications.

From a regulatory perspective, ‘prevention’ and‘treatment’ are distinct indications and must be pur-sued in a specific order: non-estrogen drugs mustbe licensed first for treatment and subsequently forprevention. ‘Prevention’ refers to the ability of adrug to prevent bone loss and is typically studiedin early postmenopausal women. ‘Treatment’ is theability of a drug to prevent further bone loss andfracture in women who have already lost substan-tial amounts of bone and who may already haveexperienced 1 or more osteoporotic fractures.

Both guidelines and established drug develop-ment practice in this area require clinical trials ofspecific durations. Although the number of patientsper trial is not generally specified, this can readilybe determined on statistical grounds. US Food andDrug Administration (FDA) guidelines suggestphase II trials with a minimum of 1 year of treat-ment using bone mineral density (BMD) as themain efficacy end-point; phase III trials for thetreatment indication require a minimum of 3 years

of administration and the incidence of new frac-tures as the primary end-point. If the data do notshow a definitive reduction in fracture rate after 3years, the fracture trial must be extended to 5 years.A licence will not be granted based on 3 years’clinical data with a positive trend to fracture reduc-tion if animal studies show abnormal bone.[5] As aresult, companies have to be prepared to continuethe fracture trial and, pending an analysis of 3-yeardata, are likely to elect to keep patients in the trialat least into the fourth year in case the reduction infracture rate is not definitive.

The number of patients per treatment arm in thefracture trial is estimated using standard statisticaltechniques to determine the number of patients re-quired to detect, with a reasonable degree of con-fidence, a clinically important difference betweenthe new drug and placebo in terms of the incidenceof new vertebral fractures in the at-risk populationstudied. For example, in a high-risk populationsuch as women >65 years selected for very lowBMD or prior fracture, 10% of the sample could beexpected to develop a radiologically apparent ver-tebral fracture over a 3-year period of observation.If a new drug is anticipated to decrease this risk by50% (a clinically important reduction and compa-rable with that observed with estrogen), a mini-mum of 1000 to 1250 women need to be enrolledin each of 2 treatment arms using the conventionalsignificance level (5%) and an 80% power to detectthe specified treatment difference. Spinal fracturesare typically chosen as the primary end-point, asopposed to those of the hip or wrist, since vertebralbone turns over at a higher rate and the effect ofinterventions is most easily detected in the spine.

The absolute minimum number of women in thefracture trial is therefore at least 2500 if a high-riskpopulation can be enrolled and the comparator isplacebo. It may slow recruitment to restrict enrol-ment to the most at-risk population and it is alsoincreasingly difficult to obtain ethics committeeapproval for placebo-controlled studies in thisfield. In addition, required patient numbers will in-crease further if more than a single dosage of activedrug is included in phase III. Further, over a 3-yearperiod, approximately 30% of patients are likely to

270 Edwards

� Adis International Limited. All rights reserved. Pharmacoeconomics 1999 Mar; 15 (3)

discontinue treatment for various reasons. As a re-sult of these considerations, the base case fracturetrial has been costed in this review by assuming3000 patients enrolled and allowing for a compari-son of placebo with 2 dosages of the new com-pound. Many completed and ongoing programmesare larger than this minimum. For example, the pla-cebo-controlled treatment part of the alendronatefracture intervention trial included 4000 women[8]

and the ongoing raloxifene programme includesmore than 13 000 women.[9,10] Programmes basedon hip fractures may need to be even larger.

Although data generated in Europe and in theUS in accordance with Good Clinical Practice(GCP) can be anticipated to be mutually acceptableto Western regulators and prescribers, Japan ex-pects that a fracture trial will also be performed inJapanese patients. It appears that an applicationmay be submitted using foreign fracture data, buta Japanese fracture study must be ongoing at thetime of filing and completed prior to approval.[8] Ifthis trial must be designed with the same statisticalvigour, then the size of the fracture trial in Japanwill be of a similar minimum size to the EuropeanUnion (EU)/US trial. Alternatively, given clearEU/US fracture data, the Japanese trial may onlyneed to show positive data trends and, in this case,may involve only 1000 women. Given that the re-quirements in Japan still appear to be evolving butmoving in a less restrictive direction,[11] it is as-sumed that 1200 patients will be adequate for theentire Japanese programme.

The requirements to obtain an indication for‘prevention’ are less onerous since they are basedon BMD data, which require much smaller samplesizes. However, FDA guidelines indicate thatBMD end-points are only permissible for the licen-sing of non-estrogens if the drug has already shownefficacy with a fracture prevention study in the‘treatment’ indication. In addition, the FDA ex-pects that interim data will be available from frac-ture end-point studies prior to the initiation of ‘pre-vention’ trials.

In contrast to osteoporosis, recent ICH guide-lines specify the general patient exposure require-ments to file a licence application for a new drug

for long-term use.[12] The guidelines for phase IIItrials require that 300 to 600 patients be treated for6 months at dosage levels intended for clinical use,including 100 patients treated for a year. Almost allphase III trials are comparative, and so the effec-tive minimum size of a EU/US programme is of theorder of 1000 patients treated for 6 months with200 continuing treatment for a further 6 months.Although the ICH process seeks to harmonise datarequirements between the EU, US and Japan, datafrom Japanese patients remain essential for regis-tration in Japan and phase III comparator-control-led Japanese data have long been essential for pric-ing.

2. Cost and Return Model

2.1 Costs

The major cost components of the drug devel-opment process and the estimate of expenditure percost category are shown in table I.

Discovery activities within the pharmaceuticalindustry are aimed at finding candidate compoundsfor development. Activities may be undertaken forseveral years within a given area of research priorto a suitable lead compound being discovered. Fur-ther, it is not always easy to associate the costs of

Table I. Major cost components of the drug development processand estimate of expenditure per cost category for an osteoporosistreatment and prevention programme

Component Cost($US million;1996 values)

Discovery cost allocation 23

Animal and other preclinical development work(including safety pharmacology, metabolism,toxicology, regulatory safety and efficacy studies)

19

Chemistry, manufacturing and control(drug product and process costs, includingprocess development, bulk and formulatedproduct manufacture, assay development,stability studies, product release testing, qualityassurance and control, validation of facility)

33

Clinical development (phase I, II and III trials) 110a

35b

Facility construction 45

a Europe, US and local.

b Additional costs for Japan and Asia.

Costs and Returns for New Osteoporosis Drugs 271


enabling technologies with a given discovery pro-ject. In a steady-state situation, a crude estimate ofthe discovery expenditure per project entering de-velopment can be made by dividing the total dis-covery expenditure per year, including overheadallocations, by the number of development candi-dates. Our estimate at Novo Nordisk A/S of $US23million per compound entering development wascompared with 3 other contemporary data sources.Recombinant Capital[13] estimate that, among bio-technology companies in 1993, the costs of researchleading to an Investigational New Drug (IND) ap-plication averaged $US18.9 million. Di Masi etal.[14] estimated that to bring a single compound tothe same stage cost $US18.6 million (1994 values).The analysis conducted by Zeneca/Lehman Broth-ers[15] suggests a discovery and infrastructure costof $US20 million per development candidate.

Having confirmed that the internal estimate wassimilar to external estimates, the Novo Nordisk A/Sestimate has been used in our model since the exactdefinition of when a compound is considered totransfer from discovery to development, and there-by the costs of reaching this transition, probablydiffers from company to company.

There are no data available to compare the costsof bringing novel compounds to development can-didate stage across different therapeutic classes.The costs for osteoporosis are thus assumed to bethe same as those for other chronic diseases.

The costs of manufacturing, pharmacology andtoxicology studies are driven by regulations whichare indication-independent. In the case of toxicolo-gy, the single largest cost is for carcinogenicitystudies which are usually required for all newchemical entities (NCEs) intended for long-termuse. All indication-independent costs were calcu-lated using an activity-based costing model devel-oped within the preclinical area at Novo NordiskA/S.[16] Additional costs are incurred within an os-teoporosis programme because of the requirementfor long-term data on bone strength, mass and his-tology in both modelling and remodelling species.These additional costs are estimated at $US5 mil-lion to $US7 million if a primate is chosen as theremodelling species used.

Phase I clinical trials are designed to explorepharmacokinetics and tolerability in both patientsand volunteers. The phase I package is not com-pletely indication-independent: for example, druginteraction studies are performed with the newdrug and existing products that are likely to betaken by the same patient population or to interactmetabolically. However, the cost driver, namelythe number of these studies, is largely indication-independent, differing primarily as to whether thedrug is a new chemical or new biological (protein)entity.

Cost components within clinical trials can begrouped into external costs and capacity costs. Ex-ternal costs are payments to third parties, for exam-ple to investigators for conducting the trial and tolaboratories for assaying efficacy and safety vari-ables in blood and urine. Capacity costs are internalcosts for staff to plan, monitor, analyse and reportthe trials according to GCP and international regu-latory requirements.

External costs were estimated as follows. Eachphase (I to III) of trial in an osteoporosis programmeand in 3 other long-term treatment developmentplans were outlined to show the time base of thetrial including recruitment, treatment and reportingperiods. The treatment period within each trial wasdissected to show the number of visits and the num-ber of procedures at each visit. External costs perpatient were calculated using the Picas (Data Edge)procedural pricing system at the 50% centileswithin the database, assuming that the patient num-bers were evenly distributed between the US andEU.

The Picas system is a trial-pricing databasewhich enables the user to build up the likely costof protocols based on the procedures within theprotocol and the number of times each procedureis performed. The database is built on the actualexperience of what companies have paid for theseprocedures in different parts of the world and thedata are presented based on payment centiles. Picasdoes not include Japanese information, and so theNovo Nordisk A/S recent experience was used toestimate likely external costs for Japan.

272 Edwards


The external cost calculations include estimatesof costs per screen failure as well as the full costsof patients meeting study entry criteria. Screeningcosts are high in osteoporosis fracture trials as theinclusion criteria require a bone density measure-ment which must be at the tail of the populationdistribution to permit enrolment. In addition to therequired programmes for the EU/US and Japan, anadditional allowance totalling 800 patients treatedfor 6 months was added to all programmes to coverthe local registration trials required in the Far Eastand Eastern Europe.

Given some differences between the guidelinesavailable for osteoporosis, it was elected to basethe calculations on the FDA guideline document.The value of the US market suggests that the FDAguidelines would receive prime consideration bymost companies.

Capacity costs were calculated using a simplecosting system in which all development tasks areidentified and the days to perform a given task areestimated using our own historical experience andexternal benchmarks.[17] The fully burdened labourrate is then applied to calculate the cost.

2.1.1 Time Lines for DevelopmentDevelopment times for new drugs are regularly

reported[18,19] and these benchmark times can beused to set development time targets within a com-pany. Novo Nordisk A/S have established a ‘bestcase’ time line of 84 months from developmentcandidate approval to launch in the first markets(EU/US), in which the development time is brokendown into preclinical, clinical phases I to III and aregulatory approval period. Inherent in this esti-mate is that phase III trials will include patientstreated for 1 year, but that filings can be based on6-month data in line with ICH exposure require-ments.[12] Based on long recruitment times andslower review at the health authority, we assumethe launch in Japan to be 2.5 years after the firstmarket.

Novel drugs developed for the treatment andprevention of osteoporosis take longer since theminimum exposure period in phase III is 3 years.This means that osteoporosis drugs can be ex-

pected to take at least 2 years longer to developthan other drugs for long-term use.

2.1.2 Costs per Year of DevelopmentThe cost estimates have been fitted to a best

practice time line, enabling the annual cash flowsfor both osteoporosis and the other drugs to be es-timated.

2.2 Returns

To model the required returns to recoup the de-velopment costs, a standardised product life cycle(PLC) has been used based on the model describedby Grabowski and Vernon.[20] For osteoporosiscompounds, launch in the EU/US is calculated inyear 10 (year 1 being the first year of develop-ment), with introduction in Japan in year 12. Theaverage effective patent life is taken as 10 yearsfrom launch and allows for an element of patentterm restoration.[20,21] In the model, sales of theoriginator product are halved within 2 years of pa-tent expiry due to generic competition. This repre-sents a rapid loss of sales, but is realistic given theevidence that sales of major compounds are rapidlyeroded once the patent expires.[22] For nonosteopo-rosis compounds, launch is in year 8 and a similarperiod of exclusivity and rate of loss upon patentexpiry is assumed.

The PLC provides an average 40% pretax mar-gin across a 20-year life. Returns in the early yearsafter launch are lower due to the costs of marketintroduction, and the returns are proportionallygreater as the product becomes mature. The as-sumed effective tax rate is 35%.

The models are built so that each year’s salesare expressed as a fraction of peak sales, enablingeasy manipulation of the sales data to calculatebreak-even under different assumptions. No sep-arate assumptions are made about price and vol-ume.

2.3 Matching Costs and Returns

Development costs are capitalised to the pointof the EU/US launch; additional expenditures ondevelopment work occurring after the first launch,(for example a Japanese fracture trial) are included



in the return side of the models. The estimate ofdiscovery expenditure is taken as a one-time costin year 1 of development.

2.4 Attrition

Most drugs entering development are not tech-nically successful. Historically, approximately 1 in10 entering preclinical development and 1 in 5 en-tering phase I will reach the market, and recent datado not suggest an improvement in these technicalsuccess rates.[23,24] In order to calculate the truecosts to bring a product to the market, the costs ofthe technical success and the costs of the failuresmust be added. The cost of each failure depends onthe stage of development at which failure occurs.

The attrition rates per phase of developmenthave been described[14,23,24] and these have beenused as the basis to estimate the full cost includingthe costs of failures. One simplifying assumptionhas been made, namely that there is minimal (10%)attrition after the commencement of phase III. Nodata are available on either the technical success orthe attrition rates across each development phasefor osteoporosis compounds, and we have assumedthat both the overall and phase-specific attritionrates are the same for osteoporosis compounds asthe average of all therapeutic classes.[25] Therefore,an estimate can be made of the full costs, includingattrition, of bringing a compound to the market.

However, given the degree of uncertainty, the esti-mate must be considered imprecise in comparisonwith the estimate of expenditure on single projectsassuming technical success.

2.5 Discount Rates

The weighted average cost of capital (WACC)for the pharmaceutical industry has been variouslyestimated[26,27] with 11% providing a reasonableestimate of the long range real cost of capital. Indetermining this figure, both the costs of drugswhich are launched and those which fail in devel-opment are considered since returns are only gen-erated by the technical successes. Therefore, a dis-count rate of 11% is used to discount the annualcash flows when the full costs of success and fail-ures are included. When a single project is consi-dered, the discount rate applied needs to be higherthan the WACC or the returns need to be risk-ad-justed for the probability of technical success.

The same discount rate can be applied to eachyear’s after-tax cash flow; alternatively, the dis-count factor can be varied such that the rate appliedearly in development is very high and then fallsprogressively as development proceeds, and dur-ing commercialisation approaches the risk-freerate of return.[28] The latter method is not widelyused, and so a simple approach using cash flowsdiscounted at the same rate each year has been usedin the cost and return models when the data from asingle project (simple model) is considered. Expe-

�501 3 5 7 9 11 13 15 17 19 21 23 25 27

0

50

100

150

Afte

r-ta

x ca

sh fl

ow (

$US

mill

ion/

y)

Time (y)

Fig. 1. Cost plot for the simple model of development of a ‘stand-ard’ drug. Cash flow is discounted at 20%. Peak annual salesrequired to break even are $US377 million.

-1501 3 5 7 9 11 13 15 17 19 21 23 25 27

-100

-50

0

50

100

150

Time (y)

Afte

r-ta

x ca

sh fl

ow (

$US

mill

ion/

y)

Fig. 2. Cost plot for the attrition model of development of a‘standard’ drug. Cash flow is discounted at 11%. Peak annualsales required to break even are $US378 million.

274 Edwards


rience suggests that the choices of 15 and 20% arearbitrary but common.

3. Results

Data from 4 models are presented. The cost es-timates for the indications other than osteoporosis(all selected from Novo Nordisk A/S projectswithin the broad field of endocrinology) werewithin 15% of each other and were simply aver-aged to provide the accumulated cost plot shownin figure 1 as the simple standard cost model. Thesimilarity in these estimates is unsurprising, sincethe cost driver is patient numbers and in all casesthe ICH patient exposure requirements were as-sumed. ‘Simple’ refers to the fact that attrition isnot considered and ‘standard’ is used to describethese long-term indications excluding osteoporosis.

Figure 2 shows the cost plot for the standardattrition model when standard phase-specific attri-tion rates are applied to the same underlying costdata. Figures 3 and 4 show the same presentationsof data for osteoporosis when the minimum pro-gramme required for triad (US, EU and Japan) mar-ket registration is pursued. Figure 5 demonstratesthe impact of not achieving clear fracture data after3 years and being required to continue treatmentfor 5 years.

The discovery cost allocation in the attritionmodels is $US115 million for both the standard and

osteoporosis drugs. This is because the discoverycost of each of the 5 NCEs entering developmenthas to be born by the single success.

Table II shows the total development costs (in-cluding $US45 million for a manufacturing faci-lity), the discounted after-tax cash flows until firstlaunch and the peak sales required to break evenfor both standard and osteoporosis drugs. Table IIIexamines 3 scenarios specific to the osteoporosisdata: 2 larger osteoporosis programmes are exam-ined and compared with the effect of being re-quired to continue the fracture study for 5 years inthe absence of adequate data after 3 years. Finally,table IV shows a simple sensitivity analysis wherethe discovery cost allocation is varied between$US15 million and $US40 million in the osteopo-rosis simple model.

4. Discussion

The large patient numbers and the long mini-mum duration of the phase III fracture trials in theEU/US and Japan make development of an osteo-porosis drug substantially more expensive than thestandard programme. The sizeable difference of$US74 million between the minimum osteoporosisprogramme and the standard in uncapitalised costassuming successful development of a single com-pound (the simple models) is attributable to therequirements for fracture trials in the EU/US and

�501 3 5 7 9 11 13 15 17 19 21 23 25 27 29

0

50

100

150

200

250

Afte

r-ta

x ca

sh fl

ow (

$US

mill

ion/

y)

Time (y)

Fig. 3. Cost plot for the simple model of development of anosteoporosis drug. Cash flow is discounted at 20%. Peak an-nual sales required to break even are $US609 million.

�150

�100

�50

0

1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31

50

100

150

200

Afte

r-ta

x ca

sh fl

ow (

$US

mill

ion/

y)

Time (y)

Fig. 4. Cost plot for the attrition model of development of anosteoporosis drug. Cash flow is discounted at 11%. Peak annualsales required to break even are $US473 million.



Japan which require an estimated minimum com-bined exposure of 10 000 patient-years.

This high cost and the effects of the additional2 years in development give a high after-tax dis-counted cash flow to the launch figure, even thoughthe costs of the Japanese fracture study are partlyhidden since they partially occur after first launch.

In the simple model, when cash flows are dis-counted at 20%, peak sales required to break evenare $US377 million and $US609 million for thestandard and osteoporosis cases, respectively.When discounting at 15%, the peak sales are$US230 million and $US354 million, respectively,and the absolute difference between the 2 cases isreduced.

In the attrition models, the assumption madethat there would be minimal attrition after the com-mencement of phase III tends to minimise the dif-ference between the osteoporosis and standard at-trition models, since even a low attrition rate inphase III would amplify the observed difference of$US95 million in peak sales.

Two conservative assumptions have been madein the minimum fracture programme. First, theEU/US trial consists of only 3000 women and, sec-ondly, the Japanese trial is not performed with thesame statistical vigour. Thus, the additional cost of$US74 million should be regarded as the minimumdifference between the standard and osteoporosisdevelopment programmes.

The effect of larger osteoporosis programmeson expenditure and required returns is shown intable III, which also shows the effect of a 2-yeardelay. The larger programmes require sales ofblockbuster proportions to return their develop-ment costs. However, such returns are consideredachievable given the value and growth rates withinthis market for new compounds with significantadvantages.[10] The additional direct expenditurefrom prolongation of the minimally sized 3-yearfracture trial to 5 years is $US26 million. However,the impact on discounted cost to launch and peaksales required is so substantial that the analysisclearly suggests that the EU/US fracture studymust be powered conservatively, i.e. ‘over-pow-ered’, to avoid narrowly missing the important ob-jective of approval based on 3-year data.

Differences in peak sales needed to break evenbetween the osteoporosis compound and the stan-

�1001 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31

0

100

200

300

400

Time (y)

Afte

r-ta

x ca

sh fl

ow (

$US

mill

ion/

y)

Fig. 5. Cost plot for the simple model of development of anosteoporosis drug showing the effect of a 2-year delay in phaseIII trials. Cash flow is discounted at 20%. The amount spent ondevelopment increases by $US26 million, the effective patentlife decreases by 1 year and launch is delayed by 2 years. Peakannual sales required to break even are $US818 million.

Table II. Comparison of costs and required returns for the ‘standard’ and osteoporosis drug development programmes

Model Development costs includingfacility ($US million)

Discounted cost to launch($US million)a

Discountrate (%)

Peak annual sales required tobreak even ($US million)

‘Standard’ simple 191 245 20 377b

Osteoporosis simplec 265 386 20 609b

‘Standard’ attrition 419 487 11 378

Osteoporosis attritionc 505 589 11 473

a Discounted after tax at an effective tax rate of 35%.

b If 15% is chosen as the discount rate for the simple models then the peak sales needed are $US230 million and $US354 million for the‘standard’ and osteoporosis cases, respectively.

c Assuming the minimum 3000-patient phase III programme.

276 Edwards


dard compound almost exclusively reflect the dif-ferences in the discounted cash flow to first launch,since the return model assumptions in terms of lifecycle and pretax margins per year are the same. Thedifference in the cost of animal work and chemis-try, manufacturing and control make only a smallcontribution, and the models do not consider dif-ferences in launch cost or other promotional costswhich could vary depending on the target audiencesize. The models are quite sensitive to the discove-ry cost allocation (table IV), both because the allo-cation is large, but also because the costs are in-curred so long prior to a return.

Although the peak sales needed for osteoporosisdrugs are large, they are achievable. 1997 sales ofalendronate, which is still in the growth phase ofits life cycle, were $US525 million,[29] and themarket peak for raloxifene has been estimated at$US1 billion.[30]

Although the primary purpose of the analysiswas to highlight the specific case of osteoporosis,this analysis can also be seen within the context ofthe literature on costing drug development. Inclu-ding the cost of failures, Grabowski and Vernon[20]

estimated, using the data set described by Di Masiet al.,[14] that the capitalised cost (at 10.5%) ofbringing a new drug to the market was $US280million (1990 values). The uncapitalised pretaxoutlays totalled $US129 million. This uncapita-lised total should be compared with table II, inwhich the uncapitalised costs for our ‘standard’case total $US419 million when the same underly-ing failure rate assumption is made.

The landmark work conducted by Di Masi etal.[14] is based on NCEs with an entry into humansbetween 1970 and 1982, with a midpoint of 1976.

This work formed one critical basis for the Officeof Technology Assessment report[23] on the profita-bility of the pharmaceutical industry. Although theout-of-pocket costs in the Di Masi et al.[14] studyhave been converted to 1990 values, this cannotallow for the substantially increased costs andcomplexity which have occurred, particularly inbasic research and clinical trials, since this cohortof drugs entered development. In addition, the fo-cus of the study is on the US, and the issues ofglobal drug development, especially with respectto Japan, appear not to be considered.

Our estimate is much closer to that ofZeneca/Lehman Brothers,[15] which is also basedon much more recent data derived from a singlecompany’s experiences. Zeneca estimate that theuncapitalised expenditure on a single compoundwhich reaches the market is $US115 million; thisexcludes both the discovery allocation ($US23million in our model) and the cost of facility con-struction ($US45 million). By excluding these 2items, our estimate is $US123 million; quite sim-ilar to that of Zeneca/Lehman.[15]

Single company estimates have the drawbackthat they may not be fully representative but, giventhe substantial increases in research and develop-ment spending since the data considered by the Of-fice of Technology Assessment, they are likely to

Table III. Sensitivity analysis on size and duration of phase III for the osteoporosis simple model using a 20% discount rate

Patients in phase IIIEuropean/US programme

Duration of phase IIItrial (years)

Development cost($US million)

Discounted cost to launch($US million)a

Peak annual sales requiredto break even ($US million)

3000 3 265 386 609

3000 5 291 577 818

6000 3 328 471 654

10 000 3 395 533 754

a Discounted after tax at an effective tax rate of 35%.

Table IV. Sensitivity analysis on varying the discovery cost alloca-tion for the osteoporosis simple model using a 20% discount rate

Discovery cost allocation($US million)

Peak annual sales required tobreak even ($US million)

15 563

23 609

40 691



be more realistic of the true costs and required re-turns for drugs entering development today.

References1. Kleerekoper M, Avioli L. Evaluation and treatment of post

menopausal osteoporosis. In: Favus M, editor. Primer on themetabolic bone diseases and disorders of mineral metabolism.New York (NY): Raven Press, 1993

2. Baum A, Moore D, McKean J, et al. Osteoporosis: brittle bonesand market penetration. New York (NY): Morgan Stanley In-vestment Research, 1997

3. Ministry of Health and Welfare. Guidelines on clinical evalua-tion of osteoporosis drugs. Tokyo: Ministry of Health andWelfare, 1995

4. Committee for Proprietary Medicinal Products. Note for guid-ance on involutional osteoporosis. London: Committee forProprietary Medicinal Products, 1996

5. Food and Drug Administration (FDA). Guidelines for pre-clini-cal and clinical evaluation of agents used in the preventionand treatment of osteoporosis. Washington, DC: FDA, 1994

6. Group for the Respect of Ethics and Excellence in Science(GREES). Recommendation for the registration of new chem-ical entities used in the prevention and treatment of osteopo-rosis. Calcif Tissue Int 1995; 57: 247-50

7. Consensus development conference: diagnosis, prophylaxisand treatment of osteoporosis. Am J Med 1993; 94: 646-50

8. Black DM, Cummings SR, Karpf DB, et al. Randomised trialof effect of alendronate in women with existing vertebral frac-tures. Lancet 1996; 348: 1535-41

9. Termine JT. Particularities of SERM and HRT development. In:Osteoporosis: putting the draft CPMP guidelines into perspec-tive. 1997 May 14. Paris: Drug Information Association

10. Anon. Scrip 1997; 2288: 1711. Morii H. Guidelines on treatment of osteoporosis in Japan. In:

Osteoporosis: putting the draft CPMP guidelines into perspec-tive. Paris: Drug Information Association, 1997

12. Committee for Proprietary Medicinal Products. Note for guid-ance on population exposure: the extent of population exposureto assess clinical safety. London: Committee for ProprietaryMedicinal Products, 1995

13. Recombinant Capital [online]. Available from: URL: http://-www.recap.com [accessed 1999 Jan 16]

14. Di Masi JA, Hansen RW, Grabowski HG, et al. Cost of innova-tion in the pharmaceutical industry. J Health Econ 1991; 10:108-42

15. Lehman Brothers. Pharma values revisited. London: LehmanBrothers, 1995

16. Tage-Hansen K. Albatross model. Drug Information Associa-tion Annual Conference; 1997 Jun; Montreal

17. Hill T. Calculating the costs of clinical research. Scrip Mag1994; Mar: 28-30

18. Di Masi JA. A new look at US drug development and approvaltimes. Am J Ther 1996; 3 (9): 1-11

19. Kaitin KI, Manochia MA. The new drug approvals of 1993,1994, and 1995. Am J Ther 1997; 4 (1): 46-54

20. Grabowski HG, Vernon JM. Returns to R&D on new drug in-troductions in the 1980’s. J Health Econ 1994; 13: 383-406

21. Shulman SR. Patent term restoration under Waxman Hatch:1984-1995. Tufts CSDD Newslett 1997; 22 (2):

22. Grabowski H, Vernon J. Prospects for return to pharmaceuticalR&D under health care reform. In: Helms RD, editor. Com-petitive strategies in the pharmaceutical industry. Washington(DC): AEI Press, 1996

23. US Congress, Office of Technology Assessment. Pharmaceuti-cal R&D: costs, risks and rewards. Washington, DC: US Gov-ernment Printing Office, 1993. Report no.: OTA-H-522

24. Di Masi JA. Success rates for new drugs entering clinical testingin the United States. Clin Pharmacol Ther 1995; 58 (1) 1-14

25. Di Masi JA, Hansen RW, Grabowski HG, et al. Research anddevelopment costs for new drugs by therapeutic category.Pharmacoeconomics 1995; 7 (2): 152-69

26. Myers SC, Shyman-Sunder L. Cost of capital estimates for in-vestment in pharmaceutical research and development [work-ing paper]. Washington (DC): Office of TechnologyAssessment (OTA), 1990

27. Grabowski HG, Vernon JM. A new look at the returns and risksto pharmaceutical R&D. Manage Sci 1990; 36 (7): 804-21

28. Myers SC, Shyman-Sunder S. Measuring pharmaceutical in-dustry risk and the cost of capital. In: Helms RB, editor. Com-petitive strategies in the pharmaceutical industry. Washington(DC): AEI Press, 1996

29. Scrip 1998; 2372: 2330. Scrip 1997; 2293: 18

Correspondence and reprints: Dr Martin Edwards, ReNeu-ron Ltd, Institute of Psychiatry, De Crespigny Park, Den-mark Hill, London SE5 8AF, England.E-mail: [email protected]

278 Edwards


New Drugs for Osteoporosis

Documents

Transcript of New Drugs for Osteoporosis