CORRESPONDENCE - direct-ms.org · CORRESPONDENCE CORRESPONDENCE ... controlled evidence on the effect of ... FDA official document for license of interferon beta-1a (Avonex)

For personal use. Only reproduce with permission from The Lancet Publishing Group.

THE LANCET • Vol 361 • May 24, 2003 • www.thelancet.com 1821

CORRESPONDENCE

CORRESPONDENCE

COMMENTARY

Interferons in relapsing remitting multiple sclerosis

Sir—The conclusions of GraziellaFilippini and colleagues (Feb 15,p 545)1 differ from those noted in twoprevious reviews,2,3 especially withrespect to the effect of interferon betaon attacks of multiple sclerosis.Methodological differences couldaccount for these discrepancies.

In our report2 we assessed sevenrandomised placebo-controlled studies,involving patients with relapsingremitting and secondary progressivemultiple sclerosis, since we believe bothconditions represent different stages ofthe same underlying illness; secondaryprogressive multiple sclerosis alwaysbegins as relapsing remitting multiplesclerosis, and 80–90% of patients withrelapsing remitting disease developsecondary progressive multiplesclerosis. Filippini and co-workers,however, included only trials of patientswith relapsing remitting multiplesclerosis, suggesting that the twoconditions represent distinct diseases.By analysing only this subset ofpatients, they effectively discard most ofthe available randomised placebo-controlled evidence on the effect ofinterferon beta on disease episodes.

With respect to outcome measures,we analysed the effect of treatment withinterferon beta on various disease-activity measures (both clinical andMRI);2 treatment significantlyimproved these outcomes in all trials.By contrast, Filippini and colleaguesexamined only attack-free status; theuse of complimentary (confirmatory)outcomes—for example, MRI—wouldhave provided convergent validity4 tothe clinical findings. Moreover,because attack-free status is the sameirrespective of whether a patient hasone attack or ten, the use of anoutcome—eg, attack rate—thatcaptures such clinically relevantinformation, might be preferable.

Finally, although drop-out bias is animportant consideration, the sensitivityanalysis used by Filippini and co-workers is problematic. The mostglaring difficulty is in their treatmentof the Multiple Sclerosis CollaborativeResearch Group (MSCRG) trial. Thistrial was stopped early and Filippiniand colleagues treat the large group ofnon-completers as dropouts in their

worst-case scenario—ie, they assumeall placebo dropouts remain attack-freeand all treatment dropouts haveattacks. Admittedly, a systematichandling of the MSCRG data presentschallenges. There was seemingly no a-priori stopping rule and, although thistrial was allegedly stopped withoutknowledge of outcome data, how thestated reason for stopping (a low earlydropout rate) shortened theobservation-time needed to establishefficacy on the primary endpoint (ie, disability) is difficult tounderstand. Furthermore, the twosubgroups (completers and non-completers) were responding quitedifferently to therapy when the trialwas stopped.5 Consequently, I believethat this trial should have beenexcluded from analyses, rather thangiving it undue weight in a worst-casemeta-analysis.

Furthermore, analysis of non-completers in a stopped trial asdropouts has serious drawbacks. Forexample, consider a 2000-patient, 2-year trial, comparing drug X withplacebo. Suppose, after an interimanalysis (a-priori stopping rule: ifp<0·001), the trial is stopped becauseof a 100% increase in attack-free statuson drug X compared with placebo(100% vs 50% at year 1; 80% vs 40%at year 2). Suppose there were no drop-outs but that 50% were non-completerswhen the trial was stopped. Clearly,such a trial provides overwhelmingevidence of efficacy. However, for theworst-case scenario (assuming all500 non-completers in the treatmentgroup, but none of the 250 non-completers in the placebo group, willrelapse in year 2) there would be600 drug X failures compared with only 550 placebo failures.Consequently, even in the face ofoverwhelming evidence, this worst-casemethod1 would cast doubt on theefficacy of drug X. Such a methodseems far too stringent to be of anypractical use in the assessment ofstudies.I have participated (or am currentlyparticipating) in several industry-sponsoredclinical trials in multiple sclerosis. Thesponsoring pharmaceutical companies for thesetrials have included (or do include) Ares-Serono, Berlex Laboratories, Biogen, Immunex,

and Teva-Marion Partners. I have also lecturedat medical conferences and in public on variousaspects of the diagnosis and management ofmultiple sclerosis. In many cases these talkshave been sponsored by non-restrictededucational grants from one or another of thecompanies listed above or by AthenaNeurosciences. In addition, the clinicaloperations of the VCSF MS Center (nursingand patient care services) have been supportedby non-restricted grants from Ares-Serono,Berlex Laboratories, Biogen, and TevaNeuroscience.

Douglas S GoodinDepartment of Neurology, University ofCalifornia, San Francisco, 505 ParnassusAvenue, San Francisco, CA 94143, USA (e-mail: [email protected])

1 Filippini G, Munari L, Incorvaia B, et al.Interferons in relapsing remitting multiplesclerosis: a systematic review. Lancet 2003;361: 545–52.

2 Goodin DS, Frohman EM, Garmany GP, et al. Disease modifying therapies inmultiple sclerosis: report of the Therapeuticsand Technology Assessment subcommitteeof the American Academy of Neurology andthe MS Council for Clinical PracticeGuidelines. Neurology 2002; 58: 169–78.

3 O’Connor P, the Canadian MultipleSclerosis Working Group. Key issues in thediagnosis and treatment of multiplesclerosis: an overview. Neurology 2002; 59:S1–33.

4 Groves RM. Survey errors and survey costs.New York: John Wiley and Sons, 1989.

5 Biogen. Summary basis of approval: FDAofficial document for license of interferonbeta-1a (Avonex). http://www.fda.gov/cber/products/ifnbbio051796.htm (accessed April 8, 2003).

e-mail submissions to [email protected]

Sir—Systematic reviews are thoughtto be objective and comprehensivemeta-analyses based on a sufficientnumber of comparable trials, thusproviding added import to the resultsof individual studies. Possibly themost difficult part of undertaking asystematic review is deciding whetherrandomised trials are sufficientlysimilar to combine their resultsstatistically. Graziella Filippini andcolleagues’ systematic review1 doesnot meet these requirements2 andresults in flawed conclusions.

First, Filippini and co-workersartificially inflate the available numberof trials by including two trials withinterferon alfa, a cytokine that has notbeen approved for treatment ofmultiple sclerosis.

Second, in view of the ongoingdiscussion about the effect of dose,


1822 THE LANCET • Vol 361 • May 24, 2003 • www.thelancet.com

CORRESPONDENCE

frequency, and route ofadministration, especially on short-term results of treatment withinterferon beta, mixing up so manydifferent approved and non-approveddose regimens can only jeopardise theresults of a meta-analysis.Consequently the data of the 1-yearanalysis were significantlyheterogeneous (p=0·044).

Third, the decision not to considerrelapse rates, although they were theprimary outcome in two andsecondary outcome in one of the threepivotal interferon beta studies, is notsufficiently justified. Inclusion of thisoutcome would have yielded morerobust results both for the first and thesecond year on study.

Fourth, Filippini and colleaguesstate that interferons’ “. . . effectbeyond one year is uncertain . . .”.This conclusion is based on asensitivity analysis for the 2-yearrelapse data. Although in the mainanalysis for the second year the effecton relapses is highly significant andhomogeneous for all three studies,interpretation of the alternativescenarios is flawed by significantheterogeneity. The authors note thatthe study that causes thisheterogeneity (interferon beta-1a oncea week intramuscularly pivotal trial)had a high rate of patients notcompleting the second year on study.But they do not take into account thatnon-completion was nearly exclusivelydue to early termination of the trialthat was decided after an interimanalysis, which yielded positive resultswith respect to the main outcome.Since the decision to stop the trial wasbased on a recommendation of theexternal patient safety monitoringboard, Filippini and colleagues’calculation based on the worst-casescenario (that all patients who did notcomplete were treatment failures)does not make sense. The samecritique also applies for the 2-year dataon progression presented in figure 3 oftheir report.

Finally, little effort seems to havebeen made by Filippini and co-workers to obtain additionalinformation from trial investigators ortrial sponsors if need be, despite theirclaims that this would be done. Forinstance, for table 1 they state that ageof included patients is not reported forPRISMS.3 However, two of theauthors were leading investigators inthis study, and should have direct andeasy access to this information.

Doctors and patients should knowthat interferon beta is only partlyeffective as a treatment for multiplesclerosis. If Filippini and colleagues’

aim was to emphasise this fact, their chosen approach seemsquestionable.

L Kappos has received honoraria forparticipation in clinical studies, advisorycommittees, and presentations at scientificmeetings as well as grants for research projectsby several pharmaceutical companies,including those whose products are discussedin this letter.

*Ludwig Kappos, Jürg Kesselring*Outpatient Clinic Neurology-Neurosurgery,University Hospitals, Kantonsspital,Petersgraben 4, CH-4031 Basel, Switzerland(LK); and Rehabilitation Clinic, Valens (JK)(e-mail: [email protected])


2 Laupacis A. The Cochrane Collaboration:how is it progressing? Statis Med 2002; 21:2815–22.

3 PRISMS (Prevention of Relapses andDisability by Interferon beta-1aSubcutaneously in Multiple Sclerosis)Study Group. Randomised double-blindplacebo-controlled study of interferon beta-1a in relapsing/remitting multiple sclerosis.Lancet 1998; 352: 1498–504.

effective ones is inappropriate. Thisapproach biases the results against atrue therapeutic effect.

The authors also say that theydecided not to do a quantitativeanalysis of the MRI outcomes. Theunwritten suggestion of their summaryto us is that the data were no good,which is untrue. The MRI results inthe major clinical trials showed greatefficacy.

D Paty is on consultancy boards for Berlex andSerono and other pharmaceutical companiesinterested in multiple sclerosis. His group isfunded by Berlex and Serono.

*Don Paty, Barry Arnason, David Li,Anthony TraboulseeUniversity of British Columbia, Room S-195,2211 Wesbrook Mall, Vancouver, BC CanadaV6T 2B5(e-mail: [email protected])


2 Clarke M, Oxman AD, eds. Cochranereviewers’ handbook 4.1.6 (updatedJanuary, 2003). In: The Cochrane Library,Issue 1. Oxford: Update Software, 2003.

3 Rice GPA, Incorvaia B, Munarti L, et al.Interferon in relapsing-remitting multiplesclerosis (Cochrane Review). In: TheCochrane Library, Issue 2. Oxford: UpdateSoftware, 2002.

4 Jacobs LD, Cookfair DL, Rudick RA, et al.Intramuscular interferon beta-1a for diseaseprogression in relapsing multiple sclerosis.Ann Neurol 1996; 39: 285–94.

Sir—Graziella Filippini andcolleagues1 call into question widelyaccepted results of clinical trials inmultiple sclerosis. However, webelieve that their meta-analysisconfuses the picture. Meta-analysis is apowerful technique, but can lead toerroneous conclusions if importantvariables such as dose, frequency ofdose, and type of therapy are notrecognised and accounted for.2 Theoriginal Cochrane Report3 oftreatment of multiple sclerosis withinterferon concluded that the “theefficacy of interferon . . . was ‘modest’after one and two years of treatment”.Filippini and co-workers’ concludethat “the clinical effect beyond oneyear is uncertain”.

We disagree with the conclusionthat the effects of treatment withinterferon over 2 years wereinconclusive. As the authors point out,this conclusion rests for the most part,perhaps entirely, on the results of one4

of the three pivotal trials that weresubjected to a worst-case scenarioanalysis. The 1-year data from thistrial were not included in thepublished report. However, these datacan be retrieved from the FDAsummary basis for approval, andindicate that most patients did notchoose to stop treatment but that thetrial was terminated early.

Furthermore, Filippini and col-leagues include alfa and preliminarybeta interferon studies in their review.We believe that to lump different typesof interferons and clearly ineffectiveand preliminary dose regimens with

Sir—As PRISMS investigators, wewould like to provide commentary onmethods used in the analysis byGraziella Filippini and colleagues,1 andcorrect some errors related to ourstudy.2

The authors state that 9–10% ofpatients in PRISMS2 were lost tofollow-up. These figures refer topatients who did not completetreatment. Several of these patientscontinued follow-up and the correctproportion of patients lost to follow-upis 5%. Many of the dropouts, however,had the outcome before theydiscontinued the study. Although notlikely to affect greatly the analysis, forthe exacerbation-free outcome, usingthe worst-case analysis, Filippini andco-workers assign five patients oninterferon to have a relapse when infact no patients were lost to follow-upin the treatment group before having arelapse. Similarly, ten relapses havebeen assigned for patients in theplacebo group, when only two patientswere lost to follow-up without havinga relapse.

For the MSCRG study, the reviewincludes the entire enrolled populationas the denominator, but uses only thenumber of patients who completed 2 years of treatment (57% of total



CORRESPONDENCE

population) in the numerator for best-case scenarios. Outcomes are thenassigned for those who did notcomplete the study for the likely andworst-case scenarios. Many of thepatients who did not complete 2 yearson study had had the outcome, andthus more actual data should beavailable than indicated. This methodof analysis leads to great studyheterogeneity that has not beenadequately addressed and that affects the ultimate conclusions of thereview.

Since interferon alfa is not used inmultiple sclerosis, and although a type1 interferon could function via adifferent mechanism of action, wechallenge the validity of including datafor interferon alfa in this review. Thesupposed reduced quality of lifeattributed to therapy with interferonstems from one small study ofinterferon alfa in which no treatmentbenefit on relapses was noted, makingconclusions on quality of life ofquestionable importance.

Combining data on various dosesassumes no differential effect onoutcomes, an assumption contrary topublished data. Results of studies2,3

with subcutaneous interferon beta-1agiven three times a week show betteroutcomes relative to placebo than thesame product given once a week.4

Furthermore, subcutaneous interferonbeta-1a 44 �g given three times a week is significantly better thanintramuscular interferon beta-1a 30 �gonce a week on the outcomeprominently discussed by Filippini andcolleagues, proportion relapse-free.5 Toassume that all regimens can be pooledto ascertain an overall estimate ofpotential efficacy is questionable anddoes a disservice to patients byconcluding that a marginal benefitexists for a treatment regimen that hasshown consistent efficacy on diseasemeasures of multiple sclerosis.PRISMS-43 provides data on 80% ofenrolled patients, balanced in accordwith original treatment groups,showing treatment efficacy ismaintained for at least 4 years withoutcompromising patients’ safety.

Finally, the use of the term likelyscenario for the combination of worst-case scenario for interferon and best-case scenario for placebo, is in factdoubly punitive for treatment effectand will lead to misunderstanding bymany readers of what is indeed thelikely outcome for patients withmultiple sclerosis treated withinterferon.

M Freedman has received consulting fees andhonoraria from Biogen, Berlex, Serono, andTeva, and a research grant from Biogen.

*Mark Freedman, John King, Joel Oger,Mohammad Sharief, Hans-Peter Hartung,for the PRISMS study investigatorsUniversity of Ottawa, Ottawa, ON, Canada K1H 8L6(e-mail: [email protected])


2 PRISMS (Prevention of Relapses andDisability by Interferon beta-1aSubcutaneously in Multiple Sclerosis)Study Group. Randomized double-blindplacebo-controlled study of interferon beta-1a in relapsing/remitting multiple sclerosis.Lancet 1998; 352: 1498–504.

3 PRISMS Study Group, University ofBritish Columbia MS/MRI Analysis Group.PRISMS-4: long-term efficacy ofinterferon-�-1a in relapsing MS. Neurology2001; 56: 1628–36.

4 The Once Weekly Interferon for MS StudyGroup. Evidence of interferon beta-1a doseresponse in relapsing-remitting MS: theOWIMS Study. Neurology 1999; 53:679–86.

5 Panitch H, Goodin DS, Francis G, et al.Randomized, comparative study ofinterferon �-1a treatment regimens in MS.Neurology 2002; 59: 1496–506.

be addressed by making the primarydata available, as is now required for astudy to be published.

Goodin, and Don Paty andcolleagues suggested we should haveassessed MRI measures. MRI was asecondary outcome in our protocolbut MRI measures differed soprofoundly between studies—and thereporting of data was so selective—that a meta-analysis was impossible.Studies to show that MRI measuresare independent predictors of long-term outcome are urgently needed.

With regard to PRISMS, weabstracted the data from figure 1 inthe original article2: five patients oninterferon and ten on placebo lost tofollow-up at 2 years were included insensitivity analysis.

We cited Nortvedt’s study3 since itwas the only one to include quality oflife as an outcome measure.

We re-emphasise that sensitivityanalysis does not provide a way toanalyse or compensate for incompleteresults, since it is a tool to assesstreatment claims against the effect ofmissing information. We did asensitivity analysis at 2 years (whenoverall 20% of patients were lost tofollow-up) because all studiesreported efficacy results instead of theintent-to-treat results required byprotocol.1 The sensitivity analysis tellsus that information on this 20% isessential to support or reject theefficacy of interferons at 2 years.

Goodin’s suggestion to disregardMSCRG was contrary to ourprotocol.1 Moreover, MSCRG wasused for the registration of Biogen’sinterferon beta-1a (Avonex), the mostwidely-used interferon in the USA.Goodin’s example to show thedrawbacks of sensitivity analyses is agood one, but not applicable toMSCRG because, as he rightly states,the study had no a-priori stoppingrules. Evidence of an effect beyond 1 year could be assessed if theindividual patient data were madeavailable for new systematic reviews.

G P A Rice has received honoraria from, andparticipated in clinical trials sponsored by,Biogen.

*Graziella Filippini, Luca Munari, George C Ebers, Roberto D’Amico,George P A Rice

*Unità di Neuroepidemiologia, IstitutoNazionale Neurologico “C Besta”, Milano, Italy(GF); Ospedale Niguarda, Milano, Italy (LM);Clinical Neurology, University of Oxford,Radcliffe Infirmary, Oxford, UK (GCE);Università degli Studi di Modena e ReggioEmilia Dipartimento di Scienze Igienistiche,Microbiologiche e Biostatistiche Modena, Italy(RDA); Clinical Neurological Sciences,University of Western Ontario, London Ontario,Canada (GPAR)(e-mail: [email protected])

Authors’ reply

Sir—The comments received on ourpaper underline the difficulties ininterpreting the results of these trials.All four letters contested our inclusioncriteria. However, exclusion of thesmall trials of interferon alfa madelittle difference to our result (relativerisk of exacerbation at 1 year 0·79,95% CI 0·58–1·07).

With regard to dose heterogeneity,we considered the highest dose intrials that used more than one dose.Neither univariate nor multivariatemeta-regression analysis of treatmentin the first year showed significanteffects for dose (p=0·222), type ofinterferon (p=0·427), or duration oftreatment (p=0·260).

Goodin suggested we should haveincluded trials on secondaryprogressive multiple sclerosis. Our a-priori inclusion criteria, however,stipulated relapsing remitting multiplesclerosis only.1 Secondary progressivemultiple sclerosis can be assessed in aseparate review.

Both Ludwig Kappos and Jörg Kesselring, and Goodin stressedthe need to consider relapse rate. Wecould not undertake a quantitativeanalysis of relapse rate because it wasassessed differently among the trials,and the PRIMS and OWIMS studiesdid not report relapse rate. We alwayscontacted sponsor companies andtrialists for additional details, but thecompanies did not give sufficientinformation and the principalinvestigators did not have access tothe primary data. This problem could


Authors’ reply

Sir—In response to Richard Rudickand colleagues’ letter, we did notconclude that interferons might haveno benefit beyond 1 year but ratherthat any effect is uncertain beyond 1 year because of inadequacies in theavailable data.

We are pleased to have theopportunity to highlight the MSCRGstudy, since we found it the mostunclear of those reviewed. We foundno definitive evidence to corroboratetherapeutic claims for Avonex. We hadto extract key data, known to theinvestigators but not present in theirpublications, in order to arrive at thisconclusion; which we find anunwelcome development.

With respect to lost to follow-up, anauthoritative definition is that these are


CORRESPONDENCE

1 Rice GP, Incorvaia B, Munari L, et al.Interferon in relapsing-remitting multiplesclerosis (Cochrane protocol). In: TheCochrane Library, Issue 1. Oxford: UpdateSoftware, 2000.

2 PRISMS (Prevention of Relapses andDisability by Interferon beta-1aSubcutaneously in Multiple Sclerosis)Study Group. Randomized double-blindplacebo-controlled study of interferon beta-1a in relapsing/remitting multiple sclerosis.Lancet 1998; 352: 1498–504.

3 Nortvedt MW, Riise T, Myhr KM, et al.Type I interferons and the quality of life ofmultiple sclerosis patients: results from aclinical trial on interferon alfa-2a. MultScler 1999; 5: 317–22.

followed up for less than 2 years. Theyshould not be classified as lost to follow-up for the purpose of showing thatinterferon does not work.

Incorrect use of data from our trialled to illogical and erroneousconclusions, which were contrary tofindings from three independentrandomised controlled clinical trials.This fact should have promptedFilippini and colleagues to check theirmethods and to undertake furtheranalyses. We believe their review doesnot adhere to rigorous scientificstandards advanced by the CochraneCollaboration or by The Lancet.

R A Rudick was a coinvestigator on theMSCRG study and his institution receives andhas received grants for research studies done byRudick. Rudick is currently an investigator on aBiogen-sponsored study. He has receivedhonoraria from Biogen, Teva, Serono, IDEC,and Berlex. D L Cookfair was the mainstatistician and director of the MSCRG datamanagement and statistical center. J Griffin waschair of the MSCRG PSMB. S Hauser and S Piantadosi were on the MSCRG PSMB.

*Richard A Rudick, Diane L Cookfair, Jack Griffin, Stephen Hauser,Steven Piantadosi*Area NB2.52, Cleveland Clinic Foundation,Cleveland, OH 44195, USA (RAR); 647 WestDelavan Avenue, Buffalo, New York (DLC);Johns Hopkins Hospital, Baltimore, Maryland(JG, SP); UCSF Department Neurology, San Francisco, California (SH)


2 Jacobs LD, Cookfair DL, Rudick RA, et al.Intramuscular interferon beta-1a for diseaseprogression in relapsing multiple sclerosis.Ann Neurol 1996; 39: 285–94.

3 Jacobs LD, Cookfair DL, Rudick RA, et al.A phase III trial of intramuscularrecombinant interferon beta as treatment forexacerbating-remitting multiple sclerosis:design and conduct of study and baselinecharacteristics of patients. Mult Scler 1995;1: 118–35.

Sir—We do not believe that Filippiniand colleagues’1 conclusions are valid.In their worst-case sensitivity analysis,patients judged lost to follow-up in theinterferon group were assumed to haveone or more exacerbations during thetreatment and follow-up periods, ordisease progression after 1 and 2 years.Those classified as lost to follow-up inthe placebo groups were assumed tohave not reached these endpoints.Filippini and co-workers concludedthat interferon might have no benefitbeyond 1 year, and their abstractsuggests that interferon might makemultiple sclerosis worse. These resultsderived from incorrect use of data fromthe MSCRG study2 in which we playedkey parts.

Filippini and colleagues classifiedpatients enrolled in the MSCRG studywho did not complete 2-years on-studyas lost to follow-up, resulting in lost tofollow-up rates of 46% and 39% forinterferon and placebo groups,respectively. Most of these patients werecensored because the study wasterminated earlier than planned, butthey were not lost to follow-up. Thecircumstances that surrounded earlytermination of the MSCRG study havebeen reported2,3 and presented numeroustimes. After Betaseron (interferon beta-1b, Berlex Laboratories, Montville, NewJersey, USA) was approved in spring,1993, for relapsing remitting multiplesclerosis, the MSCRG investigatorspresented ethical concerns aboutcontinuing a placebo-controlled study tothe National Institute of Health-appointed patient safety monitoringboard (PSMB) that summer. Studystatisticians reported the lost to follow-up rate was less than 3%, and indicatedthe study had accumulated enoughpatient time-on-study to ensureadequate power for the primaryoutcome measure of time-to-sustaineddisability progression. Because of thenature of the primary outcome measureand the remarkably low lost to follow-uprate, it was possible to stop the study inearly 1994, rather than 1995 as planned.Consequently, a group of patients were

patients who “become unavailable forexaminations at some stage during thestudy” for any reason, including“clinical decisions . . . to stop theassigned interventions”.1 Thus, thepatients were lost to follow-up.

The MSCRG study used theKaplan-Meier model to assess theprobabilities of sustained progression(primary outcome) and exacerbation.However, so many patients wereunavailable for examination (censoredin Kaplan-Meier terminology but lostto follow-up nevertheless) that basicassumptions of the model wereundermined.2

In fact, two fundamentalassumptions of Kaplan-Meier wereviolated.3 First, censored patients didnot have the same attack(exacerbation) probabilities as thosewho continued to be followed up, as ismade clear in the FDA analysis ofMSCRG data.4 Second, sustainedprogression was considered to havetaken place when confirmed atexamination 6 months later. However,so many patients were censored thatconfirmation was often not possible.Furthermore, more were censored inthe interferon group than in theplacebo group—again as the FDAanalysis makes clear. In fact, sustainedchange in the EDSS score at 2 yearswas analysed on less than 40% ofrandomised patients, putting intodoubt the MSCRG claim that theprobabilities of sustained progressionwere 34·9% for placebo and 21·9% forinterferon. The FDA4 commented,“patients who withdrew prior toreaching each of the designated timepoints were eliminated from the attackrate calculation. This may bias theresult by elimination of patients whowithdrew from disease progression andactivity. The sponsors’ analyseseliminate considerable patientexperience from each calculation”.

The FDA grappled with the“significant exacerbation reduction”claimed for the 2-year completers byMSCRG. They noted the marginallysignificant reduction in exacerbationrate4 but did not correct for multiplecomparisons. The claim stands upneither to this correction nor to theremoval of the placebo patient who hadnine exacerbations. This individualdetermined almost 20% of thetreatment effect on relapses, a fact onlyemerging from careful scrutiny of thepaper. These observations undo, in ouropinion, the claim for exacerbationprevention—in any event unsupportedby significant mitigation of attackseverity, prolongation of time to firstattack on treatment, convincingreduction in steroid use in the



CORRESPONDENCE

ambiguities in this controversy mightbe clarified by making the primary dataof this study publicly available, and wecall on Rudick and colleagues to do thisto allow an independent audit.

G P A Rice has received honoraria from, andparticipated in clinical trials sponsored by,Biogen.

*Graziella Filippini, Luca Munari,George C Ebers, Roberto D’Amico, George P A Rice

*Unità di Neuroepidemiologia, IstitutoNazionale Neurologico “C Besta”, Milano, Italy(GF); Ospedale Niguarda, Milano, Italy (LM);Clinical Neurology, University of Oxford,Radcliffe Infirmary, Oxford, UK (GCE); Universitàdegli Studi di Modena e Reggio EmiliaDipartimento di Scienze Igienistiche,Microbiologiche e Biostatistiche Modena, Italy(RDA); Clinical Neurological Sciences, Universityof Western Ontario, London Ontario, Canada(GPAR)(e-mail: [email protected])

1 Jüni P, Altman DG, Egger M. Systematicreviews in health care: assessing the qualityof controlled clinical trials. BMJ 2001; 323:42–46.

2 Peto R, Pike MC, Armitage P, et al. Designand analysis of randomised clinical trialsrequiring prolonged observation of eachpatient, part II: analysis and examples.Br J Cancer 1977; 35: 1–39.

3 Bland JM, Altman DG. Survivalprobabilities (the Kaplan-Meier method).BMJ 1998; 317: 1572.

4 Biogen. Summary basis of approval: FDAofficial document for license of interferonbeta-1a (Avonex). http://www.fda.gov/cber/products/ifnbbio051796.htm (accessed May 11, 2003).

5 Goodkin DE, Cookfair D, Wende K, et al.Inter- and intrarater scoring agreementusing grades 1.0 to 3.5 of the KurtzkeExpanded Disability Status Scale (EDSS).Neurology 1992; 42: 859–63.

start of treatment (mean 2·96 [1·33] vs 2·37 [1·68]; p<0·0001;unpublished data).

Because of treatment satisfaction,97% of patients have so far compliedwith the regimen. Furthermore, theadditional cost of the immuno-suppressive medications, particularlyprednisone, is low.

Pulse therapy with intravenousglucocorticoids in patients who do notrespond to interferon beta does notreduce relapse rate, and has no long-term functional benefits.2

To recognise and prevent side-effects in patients on combinedimmunomodulatory-immunosuppres-sive medications, neurological follow-up should be undertaken every 10–12 weeks.

To assess the safety and efficacy of immunomodulatory-immunosup-pressive medications, double-blindrandomised trials, which comparepatients treated with controls chosenonly on the basis of age, sex,socioeconomic background, andclinical symptoms of disease, need tobe modified to reflect advances inmolecular chemistry. With theemerging use of metabolomics,3 todisregard immunogenomics andpharmacogenomics in the individualsstudied, including their controls,could lead to misleading conclusionsand recommendations, resulting insuboptimum treatment of patientswith multiple sclerosis.

Adequately documented, compre-hensive clinical follow-up of patientsin specialist centres by properlytrained staff would be the most promising way to generatepractically applicable data for theirmanagement.

O Kolar received a grant from Biogen formonitoring serum activation markers inpatients with multiple sclerosis given Avonex,prednisone, and azathioprine.

*Oldrich J Kolar, Jay A Bauerle, Heidi LeeIndiana Center for Multiple Sclerosis andNeuroimmunopathologic Disorders, 8424Naab Road number 1A, Indianapolis, IN 46260, USA (e-mail: [email protected])

1 Filippini G, Munari L, Incorvaia B, et al,Interferons in relapsing remitting multiplesclerosis: a systemic review. Lancet 2003;361: 545–52.

2 Goodin DS, Frohman EM, Garmany GPJr, et al. Disease modifying therapies inmultiple sclerosis: report of theTherapeutics and Technology AssessmentSubcommittee of the American Academyof Neurology and the MS Council forClinical Practice Guidelines. Neurology2002; 58: 169–78.

3 Sedlak BJ. Advances in metabolomics: drugevaluation now encompasses physiologicaloutlook. Genetic Engineering News 2003;23: 1.

Sir—Graziella Filippini and col-leagues’1 assessment of effectiveness,side-effects, and cost of interferons in relapsing remitting multiplesclerosis could be discouraging to patients who are properly managed at specialised treatmentcentres.

In our 498 patients with relapsingremitting and relapsing progressivedisease who did not stabilise onmonotherapy with interferon beta orglatiramer acetate, annual relapse rate decreased greatly after addition of immunosuppressants to theirtreatment regimens, including lowdoses of prednisone, azathioprine, orcyclophosphamate. For example, in177 individuals on Avonex (30 �gintramuscularly per week) andprednisone (on average not exceeding0·15 mg/kg bodyweight daily) for amedian of 39 months (range 4–74),the annual relapse rate after a medianof 39 months of follow up is 0·11.Their expanded disability status scalerating is also lower than it was at the

intention-to-treat group, or increase inpatients who remained exacerbation-free during the study.

A further serious defect of MSCRGwas that there were no stopping rules.In their letter, Rudick and colleaguessay the study was stopped when it “hadaccumulated enough patient time-on-study to ensure adequate power for theprimary outcome measure”. Thisstatement seems perilously close toadmitting the trial was stoppedbecause the p values had becomesignificant, in a context where thenumber of data peeks was notspecified. Nevertheless, Avonex wasapproved in the USA for a “reductionin disease progression”—despite thetrenchant criticisms in the FDAanalysis. This reduction implies aneffect on unremitting progression ofdisability, the main prognosticdeterminant in this 30–40 year disease.However, important facts, omittedfrom this report, invalidate thisMSCRG outcome measure in ouropinion. First, for progression to beconfirmed it was not necessary thatdeterioration be confined to one EDSSfunctional system. Thus, if weak leg atone examination had remitted 6months later but was replaced bybladder urgency, then progression wasconfirmed. This is highly misleading,since in such circumstances,progression can be due to remittingchanges or simply to noise (blindingefficacy uncertain). Second, as wenoted in our original paper, 50% ofAvonex-treated patients, who had metthe definition of treatment failure inthe first year, were indeed better in thesecond year. Natural history studiesshow that at least 1 year is needed toconfirm progression. Third, thedeterioration in EDSS score (1 point)needed to satisfy the study’s definitionof progression corresponds to theinter-rater variability around 40% ofthe time for the low disability patientsrecruited.5 There was no statedattempt to correct observer-dependentupward movement on this scale.

Interferons clearly affect MRIoutcomes. However, the effect ofAvonex on gadolinium after 2 years wasonly marginally significant (p=0·051)with no significant change in themedian percentage T2 burden. Theeffect of Avonex on atrophy wasanalysed only on data from the secondtreatment year, in which selected oravailable scans showed an atrophy-sparing effect (p=0·03, no correctionfor multiple looks).

The MSCRG provides a vividexample of the need for Cochranereviews undertaken according torigorous standards. Remaining



CORRESPONDENCE

Unrelated kidney donors

Sir—In his Viewpoint,1 MichaelFriedlaender stated that “The surgicalgroup of the Rabin Medical Center inTel Aviv circumvented Israeli law bydoing kidney transplants fromunrelated living donors in severalaccessible countries including Estonia,Bulgaria . . . ”

Rabin Medical Center is the largestmedical centre in Israel, serves as amajor tertiary referral centre, andundertakes most of the kidneytransplants done in Israel. Theassociation of our name withquestionable practices has caused usgreat concern and upset. We thereforedemanded an apology fromFriedlaender, who told us that thepublished version differed from theoriginal manuscript, suggesting that themeaning of his text had been alteredduring editing.

I would like to make clear that theRabin Medical Center is not involvedin any way with kidney transplantsfrom paid unrelated living donors outof Israel. Furthermore, all transplantsdone in Israel are undertaken inaccordance with Israeli law, followMinistry of Health guidelines, and takeinto account ethical considerations.G DinariRabin Medical Center, Jabotinski Road, PetahTikva 49100, Israel(e-mail: [email protected])

1 Friedlaender MM. The right to sell or buy akidney: are we failing our patients? Lancet2002: 359: 971–73.

kidney transplants from unrelated living donors in several accessiblecountries . . .” may need an expert inEnglish to adjudicate whether themeanings are equivalent. I made noclaim that Israeli law had been broken. Idid see and consent to the editorialchanges made to my Viewpoint beforepublication.

G Dinari’s letter, written a year afterpublication of my Viewpoint,completely ignores the real issues that Iraised and is misplaced.Michael FriedlaenderNephrology and Hypertension Service, HadassahUniversity Hospital, PO Box 12000, Jerusalem91120, Israel(e-mail: [email protected])

both sibutramine-containing andyohimbine-containing products.*Jens Jordan, Arya M Sharma*Franz-Volhard Clinical Research Centre,Humboldt University, 13125 Berlin, Germany(JJ); and Department of Medicine, McMasterUniversity, Hamilton General Hospital,Hamilton, Ontario, Canada (AMS)(e-mail: [email protected])

1 Esler MD, Wallin G, Dorward PK, et al.Effects of desipramine on sympatheticnerve firing and norepinephrine spillover toplasma in humans. Am J Physiol 1991; 260:R817–23.

2 Birkenfeld AL, Schroeder C, Boschmann M, et al. Paradoxical effect ofsibutramine on autonomic cardiovascularregulation. Circulation 2002; 106: 2459–65.

3 Eisenhofer G, Saigusa T, Esler MD, Cox HS, Angus JA, Dorward PK. Centralsympathoinhibition and peripheralneuronal uptake blockade after desipraminein rabbits. Am J Physiol 1991; 260:R824–32.

4 Goldstein DS, Eisenhofer G, Stull R, Folio CJ, Keiser HR, Kopin IJ. Plasmadihydroxyphenylglycol and theintraneuronal disposition of norepinephrinein humans. J Clin Invest 1988; 81: 213–20.

5 McCarty MF. Pre-exercise administrationof yohimbine may enhance the efficacy ofexercise training as a fat loss strategy byboosting lipolysis. Med Hypotheses 2002;58: 491–95.

Author’s reply

Sir—The administration of the RabinMedical Center persist in their attemptsto dissociate themselves from theactions of one of their departmentalheads (recently retired) and one of hiscolleagues. These actions have been thesubject of numerous local andinternational reports, both in the mediaand in medical journals.

Would it not be more rewarding forthem to help seek alternative solutionsto present practice, thus alleviating theneed for extraterritorial, unregulatedsurgery on donors and transplantrecipients?

The difference between the sentencesin my original manuscript “They [thesurgical group of the Rabin MedicalCenter] started performing non-relatedliving donor kidney transplants inseveral accessible countries [. . .]. Theseoperations, illegal in Israel, were thusout of Israeli jurisdiction” and thatpublished “Therefore the surgical groupof the Rabin Medical Center in Tel Avivcircumvented Israeli law by doing

Potential for sibutramine-yohimbine interaction?

Sir—The serotonin and norepine-phrine reuptake inhibitor sibutramineis frequently used as an adjunctivetreatment for obesity. The beneficialeffect of sibutramine on bodyweight is mainly mediated through anincrease in satiety. Inhibition ofnoradrenaline uptake in peripheraltissues could theoretically exacerbatearterial hypertension and increase the risk of cardiovascular comp-lications. Fortunately, the peripheralstimulatory effect of norepinephrinetransporter blockade on thesympathetic nervous system isattenuated by a reduction insympathetic outflow from the centralnervous system.1,2

Animal studies suggest that thissympatholytic effect of noradrenalinereuptake inhibition is in part mediatedby stimulation of �2 adrenoreceptorsin the central nervous system(clonidine-like effect).3,4 �2

adrenoceptor blockade can unmaskthe peripheral effects of noradrenalinereuptake inhibition with profoundsympathetic side-effects, including asubstantial rise in blood pressure andtachycardia.

Products that contain the �2

adrenoreceptor blocker yohimbine arewidely available and have beenadvocated as antiobesity treatments.5

Concomitant use of sibutramine andyohimbine or other substances thatinhibit �2 adrenergic transmissioncould, therefore, unmask the peripheraleffect of sibutramine, resulting inuntoward cardiovascular effects.

Physicians who prescribesibutramine should be aware of this potentially life-threatening druginteraction. We suggest that thisdanger should be pointed out in the product information of

The role of the Israel andWorld MedicalAssociations

Sir—Yoram Blachar (Feb 1, p 425)1

claims that “the IMA has been fightingto better the daily existence of thePalestinian population and improve theprovision of health care to the innocentcivilians suffering as a result of the past2 years of unceasing terrorism”. Theseand other passages of his article willring hollow to those who have read therecent report published by Physiciansfor Human Rights-Israel (PHR-Israel).To quote: “We believed that the IMAmight be able to curb the appallingdeterioration in the attitude of Israelimilitary forces toward Palestinianhealth and rescue services. Yet despitesevere injury to medical personnel andto the ability of physicians to act insafety to advance their patients’interests; despite Israeli shells that havefallen on Palestinian hospitals; despitethe killing of medical personnel onduty—IMA has chosen to remainsilent. Only after extensive contactbetween PHR-Israel and globalmedical bodies, and ahead of theconvention of the World MedicalAssociation, was a discussion forumcalled to discuss IMA’s position.”2

Iain ChalmersEditor, James Lind Library, Oxford OX2 7LG, UK(e-mail: [email protected])



CORRESPONDENCE

1 Blachar Y. Israel Medical Association:response to Derek Summerfield. Lancet2003; 361: 425.

2 Physicians for Human Rights. A legacy ofinjustice: a critique of Israeli approaches tothe right to health of Palestinians in theOccupied Territories. Israel: Physicians forHuman Rights, 2002: 74. http://www.phr.org.il/Phr/downloads/dl_145.pdf(accessed March 3, 2003).

Finally, it is somewhat disingenuousto quote from a PHR report as proofthat the IMA is not engaging inhumanitarian enterprises. If PHR wereitself a purely humanitarian organisationthis might be easier to accept. However,in light of PHR’s statement that theywill cooperate only with those whooppose “the Occupation”,1 we see thatPHR’s agenda is political and notmerely humanitarian. The IMA, on theother hand, takes no political stance, butsimply tries to improve the lot ofsuffering civilians and to uphold ourcommitment as doctors to treat allpatients without regard to race, religion,or nationality.Yoram BlacharPresident, Israel Medical Association, RamatGan, Israel(e-mail: [email protected])

1 Physicians for Human Rights-Israel.Clarification. Haaretz, Feb 11, 2003: 2.

WMA takes no action, what is it for?Currently the WMA seems in breach ofits own Declaration of Tokyo.Derek SummerfieldInstitute of Psychiatry, De Crespigny Park,London SE5 8AB,UK(e-mail: [email protected])

1 Blachar Y. Israel Medical Association:response to Derek Summerfield. Lancet2003; 361: 425.

2 Human D. World Medical Association:response to Derek Summerfield. Lancet2003; 361: 425–26.

Author’s reply

Sir—The IMA has, notwithstanding thereport of PHR, been involved in severalendeavours to improve health care forthe Palestinian population. My responseto Derek Summerfield mentioned thestatement issued by the IMA, and ourfutile overtures to our Palestiniancounterparts to issue a joint statementcondemning terror and calling for theassurance of health care. I also referredto our attempt to hold a jointsymposium on health care during thecurrent conflict. Both our efforts andthose extended on our behalf by neutral,international bodies, were rejected.

The IMA maintains ongoing contactwith the Israel Defense Forces (IDF)medical liaison to the coordinator ofcivilian activities in the territories in aneffort to solve existing problems and toensure the provision of medical servicesto the civilian population. We are also incontact with the International RedCross to find mutual solutions to theproblems faced.

Israeli doctors have unilaterallyundertaken to find ways to examine andtreat Palestinian children and pregnantwomen, and the IMA has taken it uponitself to try to find funding for theseactivities. The IMA has also intervenedto prevent the restraint of Palestinianyouths in hospitals and the detention ofPalestinians in need of medicaltreatment at IDF checkpoints.

At the IMA’s request, a course onethics for IDF medical officers has beenintroduced to better enable them to dealwith complex issues, and the IMA hasinitiated the training and distribution ofinformative materials to IDF physiciansand physicians drafted for reserve duty.

The IMA has offered to assistinternational organisations that wish toenter Israel to provide purelyhumanitarian support to the Palestinianpopulation, and it has agreed to serve as an intermediary between theseorganisations and the Israeli Ministry ofInterior in cases where they havedifficulty in receiving permission toenter Israel.

The IMA has also contacted theNational Health Funds and theMinistry of Health to ensure reportageof ethical problems encountered in thecourse of treatment, and any instancesof unethical treatment of patients.

Sir—Yoram Blachar’s response to mychallenge of the IMA ethical trackrecord on torture1 is formulaic—asusual invoking the mantra of WorldMedical Association (WMA)membership, the Declaration ofTokyo, and the imputation that criticsare motivated by anti-Israeli or anti-semitic sentiments. Are AmnestyInternational and others to be thusdismissed?

He asks for the names of offenders,and I gave him one: Eran Dolev, theIMA’s Head of Ethics, for his approvalof the breaking of Palestinian fingersduring interrogation. At a human rightsconference in Gaza in 1997, an Israelidoctor told me that an Israeli medicalcolleague had confessed to her that hehad removed an intravenous drip fromthe arm of a seriously ill Palestiniandetainee and told the man that if hewanted to live, he should co-operatewith his interrogators. I immediatelysent this to Dolev and asked if hewould act. I received no reply, evenafter reminders. This is the realitybehind the rhetoric of the IMA and it isno wonder that Palestinian doctors aresuspicious of them.

Delon Human for the WMA2 alsoignores the nub of my paper. Why doesthe WMA not act on the findings ofauthoritative studies in the humanrights field? How does Human’s claimthat WMA is working “tirelessly”against torture globally sit with theirinaction over Amnesty International’ssober conclusions that Israeli doctorsform part of a system ofinstitutionalised torture, and with hisscarcely credible public defence of theIMA record and of Blachar’s positionon the WMA Council? I repeat: if the

Sir—Derek Summerfield’s Health andHuman Rights piece1 and the responsesfollowing it2,3 failed to present the pointof view of PHR-Israel, an independentnon-governmental organisation thatuses its unique place to struggle againsttorture and the violation of humanrights in the Occupied Territories fromwithin Israeli society. They thereforemiss the opportunity to support andindeed encourage fellow Israeli doctorsin their struggle against violations ofhuman rights. Many of these doctorspay a personal price for their oppositionto official Israeli policies.

PHR-Israel approached IMA in1992, requesting it to join its oppositionto interrogations involving torture, andto protest the participation of Israelidoctors in its mechanism. Since then,PHR-Israel has led a wide-rangingpublic campaign against theparticipation of Israeli physicians intorture. Our efforts to recruit IMA intothis struggle were in vain.

I recount IMA’s support of“moderate physical pressure”, only tostress its long tradition of siding with“national Israeli considerations” ratherthan with universal medical ethics. Thisposition governs IMA’s current reactionsince the outbreak of violence in ourregion in September, 2000. PHR-Israelbelieved that IMA, as the representativeof the medical establishment in Israel,might be able to curb the appallingdeterioraton in the attitude of the Israelimilitary forces toward Palestinian healthand rescue services. We sent dozens ofcomplaints to IMA, protesting thearmy’s conduct and the Israeli policy ofsiege, which makes it impossible for thePalestinian health system to function.However, IMA has chosen to remainsilent.

Only just before the WMAconvention in 2002 did they publish aposition paper; but even that failed tocondemn Israel’s shooting onambulances. On its return from theWMA conference, IMA presented itsperformance as an Israeli propagandavictory over those who had wished tocondemn or even suspend IMA fromthe WMA. This presentation, together



CORRESPONDENCE

with the failure—up to this day—toanswer even one of our detailedcomplaints, prompts one to ask whetherthe IMA is merely an executive arm ofthe Israeli establishment—one thatworks very hard to present the face ofthe “enlightened occupier” rather thanstriving for universal medical ethics.And indeed with such a view, whyshould they expect Palestinian doctors,detained and humiliated at everycheckpoint in the Occupied Territories,to cooperate gladly with IMA, whenIMA does nothing to protect them?

The ongoing joint work of PHR-Israelwith our colleagues in the Palestinianmedical and human rights communityhas engendered an alternative to thediscourse of occupation, dispossession,and violence; one that is based onhuman rights. We believe that thisdifferent voice, which does exist locally,should be heard and used ininternational fora today. We urge theWMA to make its stand clear on theissue of occupation and human rightsviolations in our region.Hadas ZivPhysicians for Human Rights-Israel, Tel Aviv66020, Israel(e-mail: [email protected])

1 Summerfield D. What is the WMA for? Thecase of the Israeli Medical Association. Lancet2003; 361: 424.

2 Blachar Y. Israel Medical Association:response to Derek Summerfield. Lancet 2003;361: 425.

3 Human D. World Medical Association:response to Derek Summerfield. Lancet 2003;361: 425–26.

nicotine) in specimens provided bychildren of this age. We recorded thatthe smoking behaviour of parents athome was significantly associated withcotinine concentrations of their child.Cotinine concentrations were adjustedfor creatinine.3

The child we report here had acotinine/creatinine ratio of 800 �gcotinine/1 g creatinine, corresponding toactive smoking of 3–5 cigarettes a day.4

The parents reported a jointconsumption of 41–60 cigarettes a day.They said they smoke in the kitchen andliving room, whereas bedrooms werereported to be smoke-free. The parentsreported smoking at the dinner tableonce a day and in front of the televisionset several times a day. They also saidthey smoke near the kitchen fan severaltimes a day and near an open door atleast once a week. These commentsfrom the parents indicate that, in theiropinion, their child was well protectedfrom exposure to environmental tobaccosmoke, since they did not smoke inbedrooms and the windows were almostalways open.

Though nicotine and cotininemetabolism is independent probablydue to genetic differences,5 the cotinineconcentration of this child is remarkablyhigh. If active smoking in adults causeslung cancer and other serious diseases,passive smoking from the age of 2·5 years (and probably younger) mustbe even more deleterious. Since a childat this age cannot, by his or her own will,avoid a smoky environment, we askourselves when exposure to tobaccosmoke should be regarded as childabuse?

We want to stress the fact that,although most parents are aware of theimportance of protecting their childrenfrom tobacco smoke, and try in differentways, children can still be massivelyexposed to this toxic drug. Since to justforbid smoking might be ineffective,nurses and doctors should pay attentionto smoking behaviour of smokingparents they meet. Until we know moreabout effective measures of protection,the recommendation should be never tosmoke indoors in homes with children.

We thank the Health Research Council ofSoutheastern Sweden (FORSS) for financialsupport, and Pharmacia UpJohn, Helsingborg,Sweden, for support with cotinine analyses.

AnnaKarin Johansson, Göran Hermansson,*Johnny LudvigssonDivision of Paediatrics, Department of Molecularand Clinical Medicine, Linköping University,Linköping, Sweden(e-mail: [email protected])

1 Ludvigsson J, Ludvigsson M, Sepa A.Screening for prediabetes in the general childpopulation: maternal attitude toparticipation. Pediatr Diabetes 2001; 2:170–74.

2 WHO. Report on tobacco smoke and childhealth: consultation report. Geneva: WHO,1999.

3 Benowitz NL. Biomarkers of environmentaltobacco smoke exposure. Environ HealthPerspect 1999; 107 (suppl 2): 349–55.

4 Willers S, Attewell R, Bensryd I, Schutz A,Skarping G, Vahter M. Exposure toenvironmental tobacco smoke in thehousehold and urinary cotinine excretion,heavy metals retention, and lung function.Arch Environ Health 1992; 47 (5): 357–63.

5 Ahijevych K. Nicotine metabolism variabilityand nicotine addiction. Nicotine Tob Res1999; 1 (suppl 2): S59–62.

When does exposure ofchildren to tobacco smokebecome child abuse?

Sir—We report an instance of a childaged 2·5 years, who is exposed totobacco smoke in the home. The childis a participant in a prospective cohortstudy (ABIS; all babies in southeastSweden) we are undertaking, onenvironmental factors affectingdevelopment of immune-mediateddiseases in children.1

Exposure to environmental tobaccosmoke, known to affect present andfuture health of children,2 is one of theenvironmental factors being studied.Parents are asked, in questionnaires, ifand how much they smoke. Asubsample of smoking parents of 2–3 year-old children has been askedabout their smoking behaviour athome—ie, what precautions they use toprotect their child from tobacco smoke.To validate this questionnaire, we haveanalysed urine cotinine concentrations(the major urinary metabolite of

Genes for schizophrenia

Sir—In their Rapid review on genes forschizophrenia (Feb 1, p 417),1

Paul Harrison and Michael Owenmake omissions in their presentation ofthe data in support of the PRODH genebeing a susceptibility gene for thedisorder. They claim that no one hasreplicated the association, but thisstatement is incorrect. First, theoriginal paper that described thePRODH gene as a susceptibility genefor schizophrenia2 included a within-study replication whereby the originalpositive association seen with the USsample was reproduced in anindependent South African sample.Although Harrison and Owen includesimilar supporting information forother genes (ie, G72) in their table,they ignore it in the case of PRODH.

Second, they do not mention thestudy by Jacquet and colleagues,3 inwhich systematic screening of 23 genesfrom the 22q11 locus for individualgene deletions revealed deletions of thePRODH gene in one family withschizophrenia. PRODH was the onlyone of the 23 genes examined that was deleted in individuals withschizophrenia. Furthermore, thestudies by Jacquet and colleagues3 andLiu and colleagues2 identified severalmutations of conserved residues intheir independent samples of patientswith schizophrenia. Hyperprolinaemiawas correlated with the presence ofthese coding mutations as well as withschizophrenia in the carrier families.Moreover, both studies presentedevidence for a modest to striking(depending on the tested population)enrichment of these mutations in populations of patients withschizophrenia.

Although we understand thatHarrison and Owen themselves havenot been able to replicate theassociation between PRODH andschizophrenia in their own sample,there are two independent publishedstudies with positive and consistent



CORRESPONDENCE

evidence in support of PRODH as asusceptibility gene for schizophrenia.*Maria Karayiorgou, Joseph A Gogos*Human Neurogenetics Laboratory, RockefellerUniversity, New York, NY 10021, USA (MK); andColumbia University, College of Physicians andSurgeons, Department of Physiology andCellular Biophysics, Center for Neurobiology andBehavior, New York, USA (JAG)(e-mail: [email protected])

1 Harrison PJ, Owen MJ. Genes forschizophrenia? Recent findings and theirpathophysiological implications. Lancet2003; 361: 417–19.

2 Liu H, Heath SC, Sobin C, et al. Geneticvariation at the 22q11 PRODH2/DGCR6locus presents an unusual pattern andincreases susceptibility to schizophrenia.Proc Natl Acad Sci USA 2002; 99: 3717–22.

3 Jacquet H, Raux G, Thibaut F, et al.PRODH mutations and hyperprolinemia in a subset of schizophrenic patients. Hum Mol Genet 2002; 11: 2243–49.

These arguments can be placed inrelation to the nature of psychoticsymptoms—ie, disturbances of humanbeings’ specific capacity for language.Hallucinations (voices), disturbances of thought processes (thoughtsexperienced as alien, loss of direction),and even delusions (distortions ofmeaning) can all be conceived asdeviations in the transition of thoughtto speech (production) or fromperceived speech to meaning. Thus thephenomena of psychosis are associatedwith the core characteristic of thespecies. The importance is that therelevant genetic variation relates toprecisely those changes that distinguishHomo sapiens from other great apespecies.

Already there is evidence frommonozygotic twins that asymmetry ofthe planum temporale and its relationto handedness is subject to epigeneticvariation,4 as is the association betweenpsychosis and asymmetry of theposterior segment of the Sylvian fissurethat overlies the planum. Thus, theasymmetry that separates humanbeings from other species, and thesubstrate of language, is subject tovariation within the species that isindependent of the DNA sequence.This epigenetic variation transmittedbetween generations is dependent onan interaction between maternal andpaternal genomes and perhapsstochastic processes in the course ofdevelopment.

These conclusions lead to futurestrategies that depart from those ofHarrison and Owen. Rather thanconcentrating resources on ever-widening searches for multiple genes ofsmall effect, they dictate a focus on thecharacteristics that distinguish thecourse of brain development in Homo sapiens from that in otherprimates, and on the ill-understoodinteraction of genetic and epigeneticfactors in determining the variation associated with thisdevelopment.Tim CrowSANE Research Centre, Warneford Hospital,Oxford OX3 7JX, UK(e-mail: [email protected])

1 Harrison PJ, Owen MJ. Genes forschizophrenia? Recent findings and theirpathophysiological implications. Lancet2003; 361: 417–19.

2 Badner JA, Gershon ES. Meta-analysis ofwhole-genome linkage scans of bipolardisorder and schizophrenia. Mol Psychiatry2002; 7: 405–11.

3 Levinson DF, Lewis CM, Wise LH, et al.Meta-analysis of genome scans forschizophrenia. Am J Med Genet 2002; 114:700–01 (abstr).

4 Steinmetz H, Herzog A, Schlaug G, et al.Brain (a)symmetry in monozygotic twins.Cerebral Cortex 1995; 5: 296–300.

Authors’ reply

Sir—Maria Karayiorgou and JosephGogos take us to task for underplayingthe evidence that PRODH is aschizophrenia susceptibility gene. Theformat of a Rapid review inevitablymeans that issues are covered briefly,without being able to do justice to everyaspect of the data or their interpretation.Moreover, we were limited to 30 ref-erences, and had to remove mention ofunpublished data concerning several ofthe genes. These factors all affected theway we portrayed the background to,and strength of evidence for, each of thegenes. It also led us to omit othercandidates worthy of mention, such asDISC1, DRD3, and HTR2A.

Nevertheless, Karayiorgou and Gogoscorrectly point out that their study1

includes a within-study replicationwhich, to be consistent with the way wesummarised the G72 data, should havebeen stated in the table. We apologise;the error arose when we simplified anearlier version of the table, whichincluded more details about repli-cations. Although we acknowledge thisupgrading of the evidence, we are stillcautious about the evidence forPRODH, since some comparisons usedtwo-marker haplotypes whereas othersused three-marker haplotypes, and theobservation was not significant in a thirdsample (p=0·055, one-tailed)1 nor in anindependent family-based associationstudy.2 Lack of space and citations alsoled us to omit the study by Jacquet andcolleagues,3 which certainly providessome additional support for PRODHinvolvement in schizophrenia.

We agree with Tim Crow thatepigenetic factors might well beimportant, and said so in our article.However, we disagree with his negativeinterpretation of the evidence for any ofthe loci, and hence for all the genes thatwe reviewed. The fact that two meta-analyses do not come up with exactlythe same result is hardly unexpected,given the emerging methods in thisspecialty, and variation in the datasetsused and approaches adopted. We aremore impressed by the similarities thanthe differences in results between thetwo meta-analyses, and by the fact thatin the larger one,4 six loci met genome-wide criteria for significance (including6p and 8p, harbouring DTNBP1 andNRG, respectively). That three of theother five susceptibility genes are alsosituated at loci with strong, albeit lessconclusive, evidence of linkage, surely increases the likelihood that theyare true loci for schizophrenia.Moreover, although the evidence mightbe incomplete with respect to themultiple susceptibility genes model of

Sir—Paul Harrison and Michael Owen1

draw cautious positive conclusions fromgenetic linkage studies in schizophreniaand point to pathophysiologicalimplications. One can review the sameevidence and reach a differentconclusion with respect to the geneticbasis of psychosis and the direction offuture research.

Harrison and Owen cite two meta-analyses2,3 and claim that “replicatedlinkages to several chromosomal regionsare accumulating”. But the strikingfeature of these meta-analyses is that,despite the fact they include many of thesame studies, their summaries agreewith respect to only one chromosomalarm (8p) of the nine they highlight. Areasonable conclusion is that the nullhypothesis has not been disproved.

Why should this substantialendeavour have revealed so little firmevidence of genetic linkage to psychosis?An alternative to the view adopted byHarrison and Owen (that there aremultiple genes of small effect) is that therelevant variation is epigenetic—ie, involves modifications such asmethylation of the sequence rather thanalterations in the DNA sequence itself.For this reason, the modification isinvisible in terms of the linkage strategy.

There are already indications ofepigenetic variation in the data from monozygotic twins. Whereasconcordance (between 40% and 50%) isgreater than that (12–15%) seen indizygotic twins consistent with a geneticfactor, it falls well short of 100%. Thediscrepancy is often interpreted asevidence for an environmental inter-action, but no consistent differences inexposure to putative risk factors betweenaffected and non-affected members ofdiscordant pairs have been identified.The alternative is that discordancereflects a difference in gene expressionin the course of development.



CORRESPONDENCE

schizophrenia, it compares favourablywith the lack of empirical data yetavailable to support Crow’s challenginghypothesis.

The two letters show the divergenceof opinion that remains about thegenetics of schizophrenia. On one handwe are criticised for underestimating the“positive and consistent” evidence forPRODH, while on the other we arecriticised for proposing the existence ofany autosomal genes at all. This state ofaffairs encourages us to believe that theopinions we expressed lie somewhereclose to the middle ground. Ultimately,data will resolve the uncertainty, anddetermine whether the protocadherinXY is the epigenetic answer to thepuzzle.*P J Harrison, M J OwenNeurosciences Building, University Department ofPsychiatry, Warneford Hospital, Oxford OX3 7JX,UK(e-mail: [email protected])

1 Liu H, Heath SC, Sobin C, et al. Geneticvariation at the 22q11 PRODH2/DGCR6locus presents an unusual pattern andincreases susceptibility to schizophrenia. Proc Natl Acad Sci USA 2002; 99: 3717–22.

2 Fan J-B, Ma J, Zhang C-S, et al. A family-based association study of T1945Cpolymorphism in the proline dehydrogenasegene and schizophrenia in the Chinesepopulation. Neurosci Lett 2003; 338: 252–54.

3 Jacquet H, Raux G, Thibaut F, et al.PRODH mutations and hyperprolinemia in asubset of schizophrenic patients. Hum Mol Genet 2002; 11: 2243–49.

4 Levinson DF, Lewis CM, Wise LH, et al.Meta-analysis of genome scans forschizophrenia. Am J Med Genet 2002; 114:700–01 (abstr).

antibacterial effect with an increase inthe number of neutrophils or to an effecton cytokines, including the suppressionof tumour necrosis factor �.

Since ours was the first report on theuse of colony-stimulating factors in thetreatment of Crohn’s disease, we weresurprised that it was not referenced byDieckgraefe and Korzenik. Our surprisewas accentuated by the fact that theauthors had knowledge of ourpublication; one of us was invited to visittheir poster at the AmericanGastroenterology Association meetingin Atlanta in 2001, which outlinedpreliminary results of the studypublished in The Lancet. This invitationarose because the authors had beenaware of our initial report.

In the conflict of interest statementoutlined by Dieckgraefe and Korzenik intheir Lancet publication, they state that,on the basis of their hypothesis,Washington University has applied for apatent covering the use of “colony-stimulating factors” for the treatment ofCrohn’s disease. They further state thatthe technology has been licensed to acommercial firm and that both authorsare entitled to royalties. Obtaining a USpatent is dependent on being first toinvent. In their patent application filedin February, 2000, more than 1 yearafter our report, Dieckgraefe andKorzenik refer to case reports on the useof colony-stimulating factors to treatCrohn’s disease-like lesions associatedwith defects of neutrophil function.These conditions are not covered bytheir patent application. By contrast,whereas the first claim outlined in theirpatent application covers the use ofGCSF rather than GMCSF in thetreatment of Crohn’s disease they do notmention our publication, which remainsthe only report on the use of GCSF totreat Crohn’s disease.

As we stated to Dieckgraefe andKorzenik in Atlanta, we published ourreport with the intention of stimulatingstudies such as theirs in the hope thatthis proposed treatment would be ofbenefit to patients with intractableCrohn’s disease. We did not intend thishypothesis to be subject to patentprotection, thereby rendering thetreatment less available to manychronically ill patients. Withdrawingtheir claim for patent protection forGCSF alone, about which they have notpublished, would be of great benefit topatients if clinical trials confirm theefficacy of colony-stimulating factors intreating Crohn’s disease.*Brendan Drumm, David VaughanDepartment of Paediatrics, University CollegeDublin, Children’s Research Centre, Our Lady’sHospital for Sick Children, Crumlin, Dublin 12,Ireland(e-mail: [email protected])

1 Dieckgraefe BK, Korzenik JR. Treatment ofactive Crohn’s disease with recombinanthuman granulocyte-macrophage colony-stimulating factor. Lancet 2002; 360:1478–80.

2 Vaughan D, Drumm B. Treatment of fistulaswith granulocyte colony-stimulating factor ina patient with Crohn’s disease. N Engl J Med 1999; 340: 239–40.

Authors’ reply

Sir—Brendan Drumm and DavidVaughan are critical of our manuscriptfor failing to cite their case report,1

describing GCSF treatment of a rectalfistula. In fact, the first reference wedid cite2 was for a review thatsummarised evidence supporting anassociation between Crohn’s diseaseand impaired innate immune function.Drumm and Vaughan’s case report wasincluded in this review (reference 36).The format of research letters in TheLancet allows for only five references tobe included. The limited references wecited were central to reviewers’questions and details of the clinical trialbeing presented. If we could have citedadditional case reports, we would havechosen the seminal observations of Roeand others3,4 who deserve credit forestablishing the gastrointestinalphenotype in glycogen-storage diseasetype 1B. Finally, Drumm andVaughan’s case report describes the useof an entirely different medication;GCSF has vastly different properties toGMCSF used in our trial. Althoughboth medications increase neutrophilcounts, these proteins lack structuralsimilarity, they act on differentreceptors and cell populations, andthey have fundamentally differentbiological activities.

While we commend Drumm andVaughan on their observation thatGCSF may promote fistula closure,their assertion that our trial wasinspired by or followed their work isinaccurate. Our study, examining thesafety and effectiveness of GMCSF forthe reduction of Crohn’s diseaseactivity, was well underway at the timeof Drumm and Vaughan’s case report.

With respect to the issue raised ofinventorship, The Lancet has a copy ofcorrespondence and a manuscriptsubmission, documenting ourhypothesis and research that greatlypredates Drumm and Vaughan’s report.

We share Drumm and Vaughan’sstated desire that new promisingtherapies be made widely and rapidlyavailable to patients. However, webelieve that the most effective way toaccomplish this goal is in the form ofwell designed clinical trials. Results ofsuch clinical trials and other relatedinvestigations are precisely the drivingforces that stimulate new avenues of

Granulocyte-macrophagecolony-stimulating factor forCrohn’s disease

Sir—Brian Dieckgraefe and JoshuaKorzenik (Nov 9, p 1478)1 report on theefficacy of granulocyte-macrophagecolony-stimulating factor (GMCSF) forthe treatment of Crohn’s disease.However, the authors make no referenceto our previous publication2 in which wereported the use of granulocyte colony-stimulating factor (GCSF) to treatCrohn’s disease in an adolescent boywhose disease was resistant to otherforms of treatment.

Our report clearly outlined therationale for use of GCSF as reiteratedby Dieckgraefe and Korzenik—ie, in thepresence of Crohn’s-like lesions withconditions of neutrophil dysfunction,including glycogen-storage disease type1B, chronic granulomatous disease, andcongenital neutropenia. We postulatedat the time, as do Dieckgraefe andKorzenik, that the benefit of treatmentwith GCSF could be due to an



CORRESPONDENCE

exploration into the mechanisms thatunderlie the effectiveness of these noveltherapeutic agents.*Brian K Dieckgraefe, Joshua R KorzenikDivision of Gastroenterology, WashingtonUniversity School of Medicine, St Louis, MO 63110, USA(e-mail: [email protected])

1 Vaughan D, Drumm B. Treatment offistulas with granulocyte colony-stimulatingfactor in a patient with Crohn’s disease. N Engl J Med 1999; 340: 239–40.

2 Korzenik J, Dieckgraefe BK. Crohn’sdisease: an immunodeficiency disease? Analternative hypothesis on the etiology ofCrohn’s disease. Dig Dis Sci 2000; 45:1121–29.

3 Roe TF, Thomas DW, Gilsanz V, et al.Inflammatory bowel disease in glycogenstorage disease type Ib. J Pediatr 1986; 109:55–59.

4 Roe TF, Coates TD, Thomas DW, et al.Brief report: treatment of chronicinflammatory bowel disease in glycogenstorage disease type Ib with colony-stimulating factors. N Engl J Med 1992; 326:1666–69.

the average ward was awash with theacoustic and electromagnetic babblethat fills every other public space. But,of course, the specific low range dangerswould not be known by all.

Our own decision was to createseveral mobile friendly places where weknew that the nearest electromedicalequipment was at least 5 m away.Douglas R Small Department of Medical Physics, SouthernGeneral Hospital, Glasgow G51 4TF, UK(e-mail: [email protected])

1 Aziz O, Sheikh A, Paraskeva P, Darzi A. Useof mobile phones in hospital: time to lift theban? Lancet 2003; 361: 788.

2 MDA Device Bulletin DB 9702.Electromagnetic compatibility of medicaldevices with mobile communications.London: Medical Devices Agency, March,1997.

nutrients to the gut is vital if normalimmune status in the intestinal tract is tobe maintained. Early use of the enteralfeeding route and prevention ofextended periods of fasting duringcritical illness are widely believed to behelpful in establishing this status.

As we consider the growing problemof nosocomial infections in the intensive-care unit, perhaps now is the time tochange our management strategy: ratherthan treating effect, we should treat thecause. Will re-establishing a healthy gutflora switch off the multiorgan failureengine?*D J W Knight, K J GirlingAcademic Department of Anaesthesia and CriticalCare, Queens Medical Centre, University ofNottingham, Nottingham NG7 2UH, UK(e-mail: [email protected])

1 Guarner F, Malagelada J-R. Gut flora inhealth and disease. Lancet 2003; 361: 512–19.

2 Nathens AB, Marshall JC. Selectivedecontamination of the digestive tract insurgical patients: a systematic review of theevidence. Arch Surg 1999; 134: 170–76.

3 Oláh A, Belágyi T, Issekutz Á, et al. Earlyenteral nutrition with specific lactobacillusand fibre reduces sepsis in severe acutepancreatitis. Br J Surg 2002; 89: 1103–07.

4 Rayes N, Seehofer D, Hansen S, et al. Earlyenteral supply of lactobacillus and fibre versusselective bowel decontamination: a controlledtrial in liver transplant recipients.Transplantation 2002; 74: 123–27.

5 Rayes N, Hansen S, Boucsein K, et al. Earlyenteral supply of fibre and lactobacilli vs.parenteral nutrition: a controlled trial in majorabdominal surgery patients. Nutrition 2002;18: 609–15.

Gut flora in health anddisease

Sir—We read with interest the Review ongut flora in health and disease (Feb 8, p 512).1 However, we weredisappointed that so little mention wasmade of the effect of critical illness on gutflora and the potential benefits ofprobiotics in this group of patients.

In critically ill patients, composition ofgut flora changes within a few hours. Thereasons for this change are a combinationof reduced food intake, gut ischaemia,use of broad-spectrum antibiotics, andthe direct effect of critical illness on hostimmunity.

The subsequent large numbers ofpotentially pathogenic microorganisms(PPMs) and altered gut permeabilityhave given rise to the notion that the gutacts as the motor for multiorgan failure incritical illness. This idea is not new, andhas led to widespread interest in selectivedigestive tract decontamination (SDD).This treatment was hoped to targetaerobic gram-negative organisms andfungi, leaving anaerobes and gram-positive bacteria intact, thereby reducingmortality. In a meta-analysis assessingthe efficacy of SDD, benefits in mortalityrates were seen in critically ill surgicalpatients but not in medical patients.2

An alternative approach to gut floralchanges is not direct eradication of PPMswith antibiotics but preservation or re-establishment of normal gut flora withprobiotics. Intensive-care unit research inthis area is in its early stages, but resultsof several preliminary studies have shownmortality benefit in severe acutepancreatitis, liver transplantation, andabdominal surgery patients.3–5 We arepresently researching the effects ofprobiotics on gut floral ecology in amixed intensive-care unit population.

An adequate supply of enteral

Use of mobile phones inhospital

Sir—As someone who has helped draftmy hospital’s mobile communicationspolicy, I take issue with Omer Aziz andcolleagues’ contention (March 1,p 788)1 that it is time to reappraisecurrent restrictions against use ofmobile phones in hospitals.

They mention a 10-year-old warningfrom the UK Medical Devices Agency,which they claim has been supersededby the advent of digital global system(GSM) mobile phones. In drafting ourpolicy, we referred to the MedicalDevices Agency’s bulletin issued in1997,2 which includes GSM phones (infact, they were as troublesome as theolder phones). This bulletin reportsmany instances of mobile phonesinterfering with medical equipment atdistances from 0 m to 1 m. Forexample, three infusion pumps wentinto shut down in the presence of aGSM device and one went into drawback. For those of us trying to draft ahospital-wide policy, our nightmarescenario would be a patient on infusionsurrounded by visitors, any one ofwhom could be using a mobile phoneclose to the device.

Aziz and colleagues imply thatmedical staff could be exempt fromrestrictions on mobile phones. I findthis puzzling. Paging systems operate ata far lower power level than mobiles andare known to produce negligibleinterference. Moreover, most patientsand their hospital visitors comply withthe mobile phone ban. If medical staffwere seen to be operating outwith therestriction, it would not be long before

What is best for thepatient?Sir—Peter Bogaty and James Brophy’sViewpoint (published online March 25)1

on treatment for acute coronarysyndromes provided a well thought outand powerful argument to keepevidence-based treatment in perspectivein an era in which we tend to rigidlymisapply data. There are importantimplications of the data obtained fromcarefully controlled trials being used toguide practice in day-to-day care ofpatients, both with respect to individualpatients and the allocation of resources.

As a geriatrician, I am constantlybattling with adverse drug reactions onthe one hand, and compliance on theother. My prescribing for many drugs islimited by local and national rationingconstraints. For example, the onlyselective serotonin reuptake inhibitor Iam allowed to prescribe is fluoxetine,which frequently aggravates my patients’anxiety, and I am not allowed toprescribe a serotonin norepinephrinereuptake inhibitor without the consent ofan already overworked psychogeria-trician. More than 90% of my in-patient



CORRESPONDENCE

workload involves the care of individualswith various stages of dementia, thoughthe potential symptomatic benefits ofcholinesterase inhibitors are denied mypatients through inadequate resourcingof memory clinics and the artificiallyrestrictive constraints of prescribing thedrugs; National Institute of ClinicalExcellence (NICE) guidelines use amini-mental state examination as theprinciple criterion for deciding appro-priateness of a therapy, for which thepractical benefits are with behaviouralimprovements and reductions inanxiety.2 As a result, a 75 year old withmild dementia is more likely to receive acocktail including aspirin, an angiotensinconverting enzyme inhibitor, a statin,and a � blocker as preventive therapy fortheir possible angina, than medicationwith potential symptomatic benefits thatcould have a great effect on theirindependence and functional state.

Unfortunately, we will probablycontinue with the simplistic applicationof evidence-based medicine throughguidelines and misguided rationing. Weare now in a position in which we have tojustify the withholding of potentiallyprotective cardiac medications, wheregeneral consensus and guidelines havethat everyone receives them, oftendespite a fairly small reduction in risk.Guidelines seem to be getting priorityover the individual patient, and theproblem is compounded through auditand targets. We should focus our limitedresources better, though to do so willrequire a major change to our approachto modern medicine. We are alreadylosing our clinical freedom, and lookback fondly to the days when we coulduse our experiential knowledge andprescribe the treatments felt to be mostbeneficial to the patient.Adam Harper Department of Medicine for the Elderly, RoyalBournemouth Hospital, Bournemouth BH7 7DW,UK(e-mail: [email protected])

1 Bogaty P, Brophy J. Increasing burden oftreatment in the acute coronary syndromes: isit justified? Lancet. Published online March25, 2003. http://image.thelancet.com/extras/02art8093web.pdf (accessed April 24,2003).

2 National Institute for Clinical Excellence.Donepezil, rivastigmine and galantamine forthe treatment of Alzheimer’s disease:technology appraisal guidance No 19, Jan2001. http://www.nice.org.uk/article.asp?a=14487 (accessed April 13, 2003).

SARS—a clue to its origins?Sir—We detected large quantities ofviable microorganisms in samples ofstratospheric air at an altitude of 41 km.1,2

We collected the samples in speciallydesigned sterile cryosamplers carriedaboard a balloon launched from the

possible vertical input of external origin isconspicuously missing in suchexplanations.4,5

With respect to the SARS outbreak, aprima facie case for a possible spaceincidence can already be made. First, thevirus is unexpectedly novel, and appearedwithout warning in mainland China. Asmall amount of the culprit virusintroduced into the stratosphere couldmake a first tentative fall out East of thegreat mountain range of the Himalayas,where the stratosphere is thinnest,followed by sporadic deposits inneighbouring areas. If the virus is onlyminimally infective, as it seems to be, thesubsequent course of its global progresswill depend on stratospheric transportand mixing, leading to a fall outcontinuing seasonally over a few years.Although all reasonable attempts tocontain the infective spread of SARSshould be continued, we should remainvigilant for the appearance of new foci(unconnected with infective contacts orwith China) almost anywhere on theplanet. New cases might continue toappear until the stratospheric supply ofthe causative agent becomes exhausted.*Chandra Wickramasinghe,Milton Wainwright, Jayant Narlikar*Cardiff Centre for Astrobiology, Cardiff University,Cardiff CF10 3DY, UK (CW); Department ofMolecular Biology and Biotechnology, SheffieldUniversity, Sheffield, UK (MW); and Inter-UniversityCentre for Astronomy and Astrophysics, Pune,India (JN)(e-mail: [email protected])

1 Harris MJ, Wickramasinghe NC, Lloyd D, et al. The detection of living cells instratospheric samples. Proc. SPIE Conference2002; 4495: 192–98.

2 Wainwright M, Wickramsinghe NC, NarlikarJV, Rajaratnam P. Microorganisms culturedfrom stratospheric air samples obtained at 41 km. FEMS Microbiol Lett 2003; 218:161–65.

3 Weinstein L. Influenza: 1918, a revisit? N Engl J Med 1976; 6: 1058–60.

4 Hoyle F, Wickramasinghe NC. Diseases fromSpace. London: JM Dent, 1979.

5 Wickramasinghe C. Cosmic dragons: life anddeath on our planet. London: Souvenir Press,2001.

DEPARTMENT OF ERRORDonnelly CA, Ghani AC, Leung GM, et al.Epidemiological determinants of spread of causalagent of severe acute respiratory syndrome in HongKong. Lancet 2003; 361: 1761–66—In this Article(May 24), in the sixth sentence in the fifthparagraph of the Results section (p 1763), 48·5 days should be: “4·85 days”, and 10·712 daysshould be “10·71 days2” (p 1764). In the firstsentence of the sixth paragraph of Results (p 1764), 572·92 days should be “572·9 days2” and62·12 days should be “62·1 days2.

Indian Space Research Organisation/Tata Institute Balloon Facility inHyderabad, India, on Jan 21, 2001.Although the recovered biomaterialcontained many microorganisms, asassessed with standard microbiologicaltests, we were able to culture only twotypes; both similar to known terrestrialspecies.2 Our findings lend support to theview that microbial material falling fromspace is, in a Darwinian sense, highlyevolved, with an evolutionary historyclosely related to life that exists on Earth.

We estimate that a tonne of bacterialmaterial falls to Earth from space daily,which translates into some 1019 bacteria,or 20 000 bacteria per square metre ofthe Earth’s surface. Most of this materialsimply adds to the unculturable or uncul-tured microbial flora present on Earth.

The injection from space of evolvedmicroorganisms that have well-attestedterrestrial affinities raises the possibilitythat pathogenic bacteria and virusesmight also be introduced. The annals ofmedical history detail many examples ofplagues and pestilences that can beattributed to space incident microbes inthis way. New epidemic diseases have arecord of abrupt entrances from time totime, and equally abrupt retreats. Thepatterns of spread of these diseases, ascharted by historians, are often difficultto explain simply on the basis of endemicinfective agents. Historical epidemicssuch as the plague of Athens and theplague of Justinian come to mind.

In more recent times the influenzapandemic of 1917–19 bears all thehallmarks of a space incident component:“The influenza pandemic of 1918occurred in three waves. The firstappeared in the winter and spring of1917–1918 . . . The lethal second wave . . . involved almost the entire worldover a very short time . . . Its epidemio-logic behaviour was most unusual.Although person-to-person spread occur-red in local areas, the disease appearedon the same day in widely separated partsof the world on the one hand, but, on theother, took days to weeks to spreadrelatively short distances.”3

Also well documented is that, in thewinter of 1918, the disease appearedsuddenly in the frozen wastes of Alaska,in villages that had been isolated forseveral months. Mathematical modellingof epidemics such as the one describedinvariably involves the ad hoc intro-duction of many unproven hypotheses—for example, that of the superspreader. Insituations where proven infectivity islimited only to close contacts, a super-spreader is someone who can, onoccasion, simultaneously infect a largenumber of susceptible individuals, thuscausing the sporadic emergence of newclusters of disease. The recognition of a

Ruan YJ, Wei CL, Ling AE, et al. Comparative full-length genome sequence analysis of 14 SARScoronavirus isolates and common mutationsassociated with putative origins of infection. Lancet2003; 361: 1779–85—In figure 3 of thisMechanisms paper (May 24), the sequence for theHong Kong CUHKW1 isolate should be, from topto bottom: “TCTGCCCGGCAACCCA”.

Documents

CORRESPONDENCE - direct-ms.org · CORRESPONDENCE CORRESPONDENCE ... controlled evidence on the effect of ... FDA official document for license of interferon beta-1a (Avonex)