C O R R E S P O N D E N C E

NATURE BIOTECHNOLOGY VOLUME 21 NUMBER 7 JULY 2003 733

Alan Dove responds:Miller claims my article contains“misapprehensions and misstatements,”with the implication that I relied too heavilyon FDA sources. I strongly disagree with hisassessment. The data on clinical trials andapproval times are open to interpretation.The researchers who compiled the datafavor the view that the FDA has becomemuch more efficient, but a variety ofinternal and external forces have conspired

synthesis of the molecule to marketingapproval, has more than doubled (from 6.5to about 14 years) since 1964. Anddevelopment costs have skyrocketed:According to the Tufts Center for the Studyof Drug Development, research-basedcompanies’ average expenditures to bring anew drug to the end of clinical testing were,in 1970, 1980, 1990 and 2003, $138 million,$318 million, $802 million and $897million, respectively.

Moreover, whereas between 1970 and2000 research-based companies’ R&Doutlays soared from about $2 billion to $30billion, the number of approvals of newdrugs barely budged. (There are influencesbesides government regulation that partlyaccount for the increased complexity andexpense of clinical trials, and for the risingcosts of drug development overall. Forexample, there may have been a subtle shiftin corporate strategy: drug sponsors nowattempt to amass data to demonstrate notonly safety and efficacy as measuredagainst a placebo or standard therapy, butalso against certain competing products, toallow a claim of superiority after approval.However, an important factor is theobsessively risk-averse FDA continuallyraising the bar for approval, especially forinnovative, high-tech products.)

The trend in cancer drug approvals isillustrative. Rothenberg et al.1 have analyzedtrends in FDA approvals of drugs andindications (uses) and conclude, “Whenprojected out over the 5-year period of2000–2004, [the data predict] a 68%reduction in new drug approvals and a 37%reduction in the number of claims approvedfor new oncologic indications, compared tothe preceding 5-year period of 1995–1999.”

Even with good intentions, FDACommissioner Mark McClellan seemsunlikely to make a dent in the agency’sdismal performance. He confronts a‘corporate culture’ of risk aversion andtimidity, and there’s so little in thepipeline—the number of marketingapplications for new molecular entities filedwith the FDA has been decreasing steadilysince 1995—that even greater motivationand better management at the agency wouldhave little substrate on which to work.

Henry I. Miller

The Hoover Institution,StanfordUniversity,Stanford, CA 94305-6010,USAe-mail: miller@hoover.stanford.edu

1. Rothenberg, M.L., Carbone, D.P. & Johnson, D.H. Nat.Rev. Cancer 3, 303–309 (2003).

To the editor:In the March issue, Fiscella et al. (Nat.Biotechnol. 21, 302–307, 2003) identify anovel human T-cell immunomodulatoryprotein (TIP). Their study has directrelevance for massive BLAST searches thatwe have carried out at theDNA level to identifygenes responsible forcentriolar mitosis byfishing out genes presentin animals and unicellularflagellates and not in othertypes of organisms.Among such genes wehave found in thePlasmodium falciparumgenome, which mostlyencode proteinsparticipating inconstruction of theflagella, there is also ahomolog of TIP,annotated a ‘hypothetical protein’ Q8I3H7.Indeed, when a Smith-Watermann search iscarried out using TIP as a query, thisPlasmodium protein is found (with anexpect value of 1.06E–108) as the onlynon-animal protein homologous to TIP.The two proteins share 29.6 % identicalresidues along the whole length. Analysiswith SMART1 (http://smart.embl-heidelberg.de/) reveals that the TIP andQ8I3H7 also have a similar structure, bothcontaining transmembrane domains.

Given the relatively low homologybetween the two proteins, as well as lowhomology between Plasmodium proteinand its homolog from the natural mosquitovector (Anopheles gambiae and Plasmodium

proteins share 25% identical residues), thepossibility of recent horizontal genetransfer appears unlikely.

In this light, there are two possibleexplanations for the presence of a TIP

homolog in the Plasmodiumgenome. Q8I3H7 may be anold secretory proteinnecessary for thePlasmodium lifecycle,further adapted to T-cellregulation. Alternatively, avery interesting possibility isthat the TIP homolog isinvolved in P. falciparum’sparasitic lifestyle and itsability to immunomodulatethe host organism. T cellshave been shown to play amajor role in early stages inprotective immunity againstmalaria parasites2.

Analogous to the protective effect of TIP protein in graft-versus-host disease described byFiscella et al., Q8I3H7 may protect theparasite against attack by the host immunesystem.

Szymon Kaczanowski & Piotr Zielenkiewicz

Institute of Biochemistry and Biophysics,Polish Academy of Sciences,Pawinskiego 5a, 02-106 Warszawa,Polande-mail: (piotr@ibb.waw.pl)

1. Schultz, J., Milpetz, F., Bork, P. & Ponting, C.P.Proc. Natl. Acad. Sci. USA 95, 5857–5864(1998).

2. Tsuji, M. & Zavala, F. Trends Parasitol. 19, 88–93(2003).

A TIP on malaria (genomics)

to lengthen corporate-sponsored clinicaltrials. In an era of intense competition andlitigation, the FDA is not the onlyorganization that is risk-averse. Backgroundresearch for this article was extensive, andincluded interviews with a range of experts,most of them outside the FDA. Conclusionsabout the agency’s performance relied onstatements by drug industry sources andindependent observers, and comparisonwith analogous organizations overseas.

©20

03 N

atu

re P

ub

lish

ing

Gro

up

h

ttp

://w

ww

.nat

ure

.co

m/n

atu

reb

iote

chn

olo

gy