The New England Journal of Medicine
`
`Correspondence
`
`Helicobacter pylori and Nonulcer Dyspepsia
`
`To the Editor: McColl et at. and Blum et al. (Dec. 24
`issue)1,2 reached opposite conclusions regarding the eradi
`cation of Helicobacter priori infection in patients with non
`ulcer dyspepsia. In an accompanying editorial, Friedman3
`concluded that breakthroughs in the treatment of nonul
`cer dyspepsia will probably result from basic investigations
`into visceral sensitivity and brain gut interactions rather
`than ftom further examination of the role of H. priori in
`fection.
`Nonulcer dyspepsia is a heterogeneous condition. It is
`conceivable that the two studies involved somewhat differ
`ent groups of patients with different sensitivities to antibiotic
`treatment. Indeed, one subgroup of patients with nonnl
`cer dyspepsia namely, those with alcoholic gastritis
`appears to have a response to such treatment. Studies inl
`tiated over 40 years ago revealed that antibiotics eradicate
`bacterial gastric ammonia production,4 and in patients with
`alcoholic gastritis, the severity of H. pylori-induced gastric
`injury is correlated with the level of gastric juice ammonia,
`suggesting that ammonia has a pathogenic role in the
`H. pylori-associated gastric injury? Furthermore, in one
`study, antibiotic therapy was associated with the ameliora
`tion ofhistologic abnormalities and dyspeptic symptoms, in
`cluding epigastric pain, nausea, vomiting, heartburn, bali
`tosis, burping, postprandial bloating, and flatulence,6 which,
`with the exception of vomiting, are among the eight most
`common symptoms in patients with nonnlcer dyspepsia.
`In two studies, all H. priori-positive patients with nonnlcer
`
`dyspepsia and alcoholic gastritis had a response to antiheli
`cobacter therapy, whereas controlled abstinence from alco
`hol had no such effect, nor ~vas there any improvement in
`the patients ~vho received antacids alone.<7 Since the dif
`ference bet~veen the results of the studies by McColl et at.
`and Blum et al. reflected the outcome in only a relatively
`small percentage of patients, factors such as an unequal num
`ber of patients ~vith alcoholic gastritis in the two studies
`may have been important.
`
`CHARLES S. LIEBER, M.D.
`
`Bronx Veterans Affairs Medical Center
`Bronx, NY 10468
`
`1. McColl K, Murray" L, El Omar E, et al. Symptomatic benefit from eradi
`caring Helicobacter priori infection in patients with nonulccr dyspepsia.
`N Engl J Mcd 1998;339:1869 74.
`2. Blum AL, Talley NJ, O’Morfiin C, ct al. Lack of effect of treating Hdi
`cobacterpylori infection in patients with nonulccr dyspepsia. N Engl J Mcd
`1998;339:1875 81.
`a. Friedman LS. Hdicobacterpylori and nonulccr dyspepsia. N Engl J Mcd
`1998;339:1928 30.
`4. Licbcr CS, Lcfcvrc A. Ammonia as a source of gastric hypoacidity in
`patients with uremia. J Clin Invest 1959;38:1271 Z
`5. Tricbling AT, Korsten MA, Dlugosz JW, Paronctro F, Licbcr CS. Severity"
`of Hclicobacter induced gastric injury correlates with gastric juice ammonia.
`Dig Dis Sci 1991;36:1089 96.
`6. Uppal R, Latecf SK, Korsten MA, Paronctro F, Licbcr CS. Chronic
`alcoholic gastritis: roles of alcohol and Helicobacterpylori. Arch Intern Mcd
`1991;151:760 4.
`7. Licbcr CS. Gastritis in the alcoholic: relationship to gastric alcohol me
`tabolism and Helicobacterpylori. Addict Biol 1998;3:423 33.
`
`To the Editor: McColl et at. did not address the effect of
`"eradicating" H. priori infection on nonulcer dyspepsia
`but instead addressed the effect of "treating" H. priori in
`fection. We think that they should have tested the effect of
`successful eradication on nonulcer dyspepsia, since the
`infection was not eradicated in at least 12 percent of the
`patients who received antibiotics and omeprazole (four pa
`tients in this group did not have a urea breath test after
`treatment). It is also noteworthy that 5 percent of the pa
`tients in the placebo group had a negative urea breath test
`
`INSTRUCTIONS FOR LETTERS TO THE EDITOR
`
`Letters to the Editor are considered for publication (subject to editing and abridgment) provided they do not contain material that has been
`submitted or published dse~vhere. Please note the follo~ving: ,Your letter must be type~vritten and ~iple spaced. ,Its text, not including
`references, must not exceed 400 ~vords (please include a ~vord count). ,It must have no more than five references and one figure or table.
`¯ It should not be signed by more than three authors. ,Letters referring to a recent JournM article must be received ~vithin four ~veeks of its
`publication. ,Please include your full address, tdephone number, and fax number (if you have one). ,You may send us your letter by post, fax,
`or dectronic mail.
`Our address: Letters to the Editor ¯ New England Journal of Medicine ¯ 10 Shattuck St. ¯ Boston, MA 02115
`
`Our fax numbers: 617-7~9-9864 and 617-7~4-4457
`
`Our e mail address: letters@nejm.org
`We cannot ackno~vledge receipt of your letter, but ~ve ~vill notify you ~vhen ~ve have made a decision about publication. We are unable to
`provide prepublication proofs. Please enclose a stamped, sdf addressed envdope if you ~vant unpublished material returned to you.
`Financial associations or other possible conflicts of interest must be disdosed. Submission of a letter constitutes permission for the Massachu
`setts Medical Society, its licensees, and its assignees to use it in the JournM~s various editions (print, data base, and optical disk) and in
`anthologies, revisions, and any other form or medium.
`
`1508
`
`May 13, 1999
`
`The New England Journal of Medicine
`Downloaded ~rom nejm.org on December 8, 2016. For personal use only. No other uses without permission.
`Copyright © 1999 Massachusetts Medical Society. All rights reserved.
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1115-0001
`
`

`
`CORRESPONDENCE
`
`one month after treatment an extraordinarily high per
`centage for sham eradication.
`We suggest that additional analyses of the data reported
`by McColl et al. and Blum et at. be conducted so that the
`efI~ct of eradicating H. priori infection is more explicitly
`addressed.
`
`ANDREAS Pr2SPOK, M.D.
`GEORG OBERHUBER, M.D.
`
`University of Vienna
`1090 Vienna, Austria
`
`To the Editor: It is possible that some of the patients in
`the study by McColl et at. had duodenitis without discrete
`duodenal ulceration at the time of endoscopy. This vvould
`be in keeping vvith the high incidence of symptomatic ul
`cer disease in their placebo group. Patients vvith duodenal
`erosions vvere excluded from the trial by Blum et al. In the
`Nottingham randomized trial of dyspepsia management,
`vve found that 41 percent of the patients referred for early
`endoscopy and then classified as having duodenal ulcer dis
`ease (22 of 54) had H. pylori-associated duodenitis (con
`firmed on biopsy) vvith or vvithout erosions but vvithout
`discrete ulceration) Were duodenal biopsies performed to
`exclude patients vvith this manifestation of ulcer disease, as
`recommended by Talley?2
`
`RICHARD F.A. LOC~AN, F.R.C.P.
`ROBERT P.H. LOC~AN, M.R.C.P.
`University of Nottingham
`Nottingham NG7 2UH, United Kingdom
`
`1. Duggan A, Elliotr C, Logan RPH, Hawkcy C, Logan RFA. Does "near
`patient" H. pylori testing in primary care reduce referral for endoscopy? Re
`suits from a randomiscd trial. Gastroenterology 1998;114:Al10. abstract.
`2. Tallcy NJ. A critique of therapeutic trials in Hdicobacterpylori~ositivc
`functional dyspepsia. Gastroenterology 1994:;106:1174 83.
`
`To the Editor: The 36 item Medical Outcomes Study
`Short Form General Health Survey (SF 36)1 is a popular
`generic measure of health status that is widely used in din
`ical trials. Unfortunately, in some studies, care has not
`been taken to apply the measure in the way that its develop
`ers intended, as is the case for the study by McColl et at.
`These authors have incorrectly summed the eight individ
`ual scales of the SF 36 survey to obtain a total score rang
`ing from 0 to 800. There is no psychometric evidence to
`support this method of scoring, and most clinical trials ei
`ther report the scores on individual scales that are hypoth
`esized to be affected by treatment (e.g., mental health, pain,
`and vitality) or summarize the eight scores on the scales
`in the form of two composite scales: one for physical health
`and one for mental health.2 The authors cite a secondary
`source for the scoring of this instrument rather than the
`scoring manual for the instrument itself1; this may have af
`fected their ability to detect a meaningful difference in
`quality of life between the two treatment groups.
`
`PATRICIA A. GANZ, M.D.
`
`UCLA School of Medicine
`Los Angeles, CA 90095 6900
`
`1. Ware JE Jr. SF 36 health survey: manual and interpretation guide. Bos
`ton: New England Medical Center, 1993.
`2. Ware JE, Kosinski M, Keller SD. SF 36 physical and mental health sum
`mary scales: a user’s manual. Boston: Health Institute, 1994:.
`
`To the Editor: In his editorial, Friedman did not mention
`a likely confounding factor in both studies. Both groups of
`investigators unkno~vingly included in their studies patients
`~vith peptic ulcer disease, ~vho ~vould have been expected
`to have improvement after the eradication of H. pylori in
`fection. There ~vere at least four such patients in the study
`by McColl et al. and seven in the study by Blum et at.
`If endoscopy is performed to diagnose dyspepsia as part
`of a research protocol or in routine clinical practice, it
`must be done ~vhen the patient is having symptoms. Nor
`mal findings on endoscopy performed ~vhen a patient is
`asymptomatic cannot be used to rule out peptic ulcer dis
`ease. A policy of performing endoscopy only ~vhen patients
`~vith dyspepsia are symptomatic may be less convenient,
`but it can be carried out simply by giving patients the in
`structions for endoscopy and asking them to call ~vhen they
`start to have symptoms so that they can be scheduled for
`endoscopy as soon as possible.
`On the basis of these t~vo studies, it ~vould be unsafe to
`conclude that the eradication of H. pylori infection has a
`place in the management of true nonulcer dyspepsia.
`
`BRLAN B. SCOTT, M.D.
`County Hospital
`Lincoln LN2 SQY, United Kingdom
`
`The authors reply:
`
`To the Editor: Several of the correspondents propose ex
`planations for the difI~rent conclusions reached in our
`study and that of Blum et at. However, vve believe that the
`results of the two studies are similar, with each showing a
`small benefit of eradicating H. pylori infection in patients
`with nonulcer dyspepsia.
`In our study, vve found a significant efI~ct of the eradi
`cation of H. priori infection on the proportion of patients
`who had a resolution of symptoms, with an absolute dig
`ference of 14 percent (95 percent confidence interval, 7 to
`24 percent) between the active treatment group and the
`placebo group. The primary outcome of the study by Blum
`et at. was also the resolution of symptoms at one year. The
`difI~rence between the two groups although not signif
`icant was also in favor of active treatment (an absolute
`difference of 6.7 percent; 95 percent confidence interval,
`2.6 to 16.0 percent). The study by Blum et al. was de
`signed with power to detect a difference of more than 20
`percent between the active treatment and placebo groups
`and ~vas therefore unlikely to detect the difference of 14
`percent that vve found.
`The editorial by Friedman highlights the fact that the
`rate of resolution of symptoms in the placebo group was
`lower in our study than in that by Blum et at. However, in
`our study, the resolution of symptoms was defined as no
`or virtually no symptoms over the preceding six months,
`whereas the definition in the study by Blum et at. was no
`or minimal symptoms over the preceding week. The far
`more stringent definition of symptom resolution that vve
`
`Volume 340 Number 19
`
`1509
`
`The New England Journal of Medicine
`Downloaded ~Yom nejm.org on December 8, 2016. For personal use only. No other uses without permission.
`Copyright © 1999 Massachusetts Medical Society. All rights reserved.
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1115-0002
`
`

`
`The New England Journal of Medicine
`
`used explains the lower proportion of patients in whom it
`was achieved. With such different definitions of symptom
`resolution, it is very misleading to compare the response
`rates for the placebo groups in the two studies.
`We chose a dyspepsia score of less than 2 as indicative of
`the resolution of symptoms because our previous valida
`tion studies had indicated that this was the most appropri
`ate cutoff point. If we were to use a symptom score of less
`than 3 as indicative of the resolution of symptoms, then this
`would mean that symptoms resolved in 29.2 percent of
`the active treatment group as compared with 1Z5 percent
`of the placebo group (95 percent confidence interval for
`the difference between active treatment and placebo, 2.3 to
`21.1 percent). When trying to detect a subgroup response,
`it is appropriate to compare the proportion of patients
`with a cure in the active treatment group with the propor
`tion in the placebo group rather than to look merely at dif
`ferences in mean values between the two overall groups.
`With respect to the point raised by Dr. Ganz, there were
`no significant differences after treatment between the active
`treatment and placebo groups with respect to the mean
`score for any of the eight individual scales in the SF 36
`quality of life survey.
`Several of the correspondents suggest that the patients
`who had a benefit from treatment to eradicate H. pylori in
`fection may have had duodenal ulcers that were healed at the
`time of endoscopy. We believe that this suggestion is prob
`ably wrong, since there was no association between risk
`factors for ulcer and the response to treatment.
`We included in our study patients whose predominant
`symptom was heartburn or acid regurgitation suggestive
`of gastroesophageal reflux disease. In our group of 95 pa
`tients with symptoms suggestive of gastroesophageal re
`flux disease, 20 percent had a resolution of symptoms after
`active treatment, as compared with 2 percent after placebo
`(P 0.007), which represented at least as great a benefit of
`treatment as that in our 171 patients ~vith ulcerlike symp
`toms (23 percent of whom had a resolution of symptoms
`after active treatment, as compared with 10 percent after
`placebo; P 0.02). Blum et al. excluded patients with symp
`toms suggestive of gastroesophageal reflux disease, presum
`ably on the assumption that they would not benefit from
`treatment. Future studies should include patients with the
`full spectrum of upper gastrointestinal symptoms and, in
`particular, those with heartburn or acid reflux.
`Taking the findings of the two studies together, vve be
`lieve the message is that eradicating H. pylori infection re
`sults in a resolution of symptoms in a subgroup of patients
`with normal endoscopic findings. In addition, it is impor
`rant to recognize that the benefit is at least as great in pa
`tients with symptoms suggestive of gastroesophageal reflux
`disease.
`
`KENNETH E.L. McCOLL, M.D.
`LILIAN S. MURRAY, PH.D.
`EivIAD M. EL OivIAR, M.D.
`
`Western Infirmary
`Glasgow Gll 6NT, United Kingdom
`
`To the Editor: Dr. Lieber suggests that the apparently
`disparate results of the tvvo studies may be explained by
`the enrollment of different groups of patients vvith different
`
`1510
`
`May 13, 1999
`
`sensitivities to antibiotics. We agree. Ho~vever, Dr. Lieber’s
`hypothesis that patients ~vith "alcoholic gastritis" (i.e.,
`alcohol induced gastric damage) may be a subgroup that
`does not have a response to treatment is not ~vell founded,
`since the data cited1 ~vere obtained from only 14 patients
`and the study ~vas not ~vell controlled. Alcohol damage is
`very unlikely to account for our results; patients ~vith al
`coholism ~vere excluded, and none of our patients had the
`classic finding of alcohol damage, focal subepithelial hem
`orrhage.2
`Drs. Logan and Logan suggest that duodenitis should
`have been a criterion for exclusion from the study. Since the
`majority of H. pylori infected patients have duodenal in
`flammation on histologic examination,3 this criterion ~vonld
`have excluded most patients from the study and ~vould
`thus have led to a selection bias. In addition, duodenitis is
`an ill defined entity. In an analysis of the response in terms
`of symptoms in relation to histologic findings, ~ve found
`no difference bet~veen patients ~vith antrum predominant
`gastritis (a finding typical in patients ~vith duodenal ulcer)
`and those ~vith corpus predominant gastritis (not usually
`found in patients ~vith duodenal ulcer). If many patients
`~vith duodenal ulcer had been inadvertently included, ~ve
`~vould have expected more of the patients in the antrum
`predominant group to have had a response.
`Dr. Scott stresses that endoscopy must be performed
`~vhile patients are symptomatic in order to rule out peptic
`ulcer disease. Our patients ~vere symptomatic ~vhen they
`under,vent endoscopy on enrollment in the study. Unlike
`McColl et al., ~ve routinely performed endoscopy in all pa
`tients 3 and 12 months after treatment ~vas completed to
`determine ~vhether ulcers had developed during the study.
`In the study by McColl et al., very fe~v symptomatic patients
`underwent repeated endoscopy; ulcers were detected in
`four of six such patients in the group given omeprazole
`alone. Both symptomatic and asymptomatic ulcers may have
`been missed.
`Drs. Pfisptk and Oberhuber suggest that the results
`might be different if patients cured of infection were com
`pared with those not cured. We reported the rates of
`symptom relief in both H. priori positive and H. pylori
`negative patients after antibiotic therapy (31 percent and
`26 percent, respectively); the difference was not signifi
`cant. We have now also compared symptom relief in the
`patients who remained positive for H. priori infection after
`treatment with omeprazole (24 percent) with symptom
`relief in those who were cured of infection after antibi
`otic therapy (31 percent). The difference was not signifi
`cant, confirming the results of our intention to treat analy
`sis. A ;vord of caution should be added about the use of
`these subgroups: since they are based on both treatment
`and the result of treatment, the groups are no longer ran
`domized.
`The response rates in the active treatment groups in our
`study (27 percent) and in that ofMcColl et al. (21 percent)
`;vere remarkably similar. In the study by McColl et al., the
`distributions of the Glasgo;v Dyspepsia Severity Score in
`the group of patients ;vho received antibiotics and the group
`;vho received omeprazole alone ;vere not strikingly differ
`ent at 12 months, despite the observation of a significant
`difference ;vhen an arbitrary cutoff score of 0 or 1 ;vas cho
`sen. We conclude that although different populations of
`patients may have been enrolled in the t;vo studies, the re
`sults are, in fact, reasonably similar, and they sho;v that the
`
`The New England Journal of Medicine
`Downloaded ~Yom nejm.org on December 8, 2016. For personal use only. No other uses without permission.
`Copyright © 1999 Massachusetts Medical Society. All rights reserved.
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1115-0003
`
`

`
`CORRESPONDENCE
`
`eradication of H. pylori infection is of little benefit in pa
`tients ;vith nonulcer dyspepsia.
`
`ANDI~ L. BLUM, M.D.
`
`Centre Hospitalier Universitaire Vaudois
`1011 Lausanne, Switzerland
`
`NICHOLAS J. TALLEY, M.D.
`
`University of Sydney
`Penrith, NSW 2751, Australia
`
`MANFRED STOLTE, M.D.
`
`Klinikum Bayreuth
`95440 Bayreuth, Germany
`
`1. Uppal R, Latcef SK, Korstcn M, Paronctto F, Licbcr CS. Chronic alco
`holic gastritis: roles of ethanol and Hdicobacter pylori. Arch Intern Mcd
`1991;151:760 4:.
`2. Lainc L, Marin Sorcnscn M, Wcinstcin WM. Campylob~cterpylori in
`alcoholic hemorrhagic "gastritis." Dig Dis Sci 1989;34:677 80.
`3. Casclli M, Gaudio M, Chiamcnti CM, ct al. Histologic findings and
`cob~cterpylori in duodenal biopsies. J Clin Gastrocntcro11998;26:74: 80.
`
`To the Editor: Lieber suggests that in patients ;vith gas
`tropathy caused by both alcohol consumption and H. pylori
`infection, dyspepsia is likely to resolve after the eradication
`of H. pylori infection. Ho;vever, his hypothesis is based on
`a nonrandomized, unblinded, and essentially uncontrolled
`study of only 14 patients ;vho ;vere reassessed ;vith the use
`of an unvalidated "total dyspepsia score" only several
`;veeks after the completion of therapy) More rigorous clin
`ical testing of this hypothesis is required. Moreover, a dis
`tinction should be made bet;veen alcoholic gastropathy,
`;vhich is characterized histologically by subepithelial hem
`orrhages and erosions but generally not by inflammation,
`and H. pylori gastritis, a truly inflammatory process)
`Scott reiterates the possibility, mentioned in my editorial,
`that some patients ;vith nonulcer dyspepsia and H. pylori
`infection in fact have peptic ulcer disease in endoscopic re
`mission. Ho;vever, his assertion that the detection of an ulcer
`requires that endoscopy be performed ;vhen the patient is
`symptomatic is a common misconception. In fact, symp
`toms of peptic ulcer lack both specificity and sensitivity for
`the endoscopic diagnosis of an ulcer) For example, endos
`copy reveals clinically silent peptic ulcers in 1 to 3 percent
`of asymptomatic adult subjects,4 and 20 percent of patients
`;vith complicated ulcers (e.g., bleeding or perforation)
`have no history of dyspepsia,s It is the absence of a corre
`lation bet;veen dyspepsia and endoscopic detection of ul
`cers that makes the management of nonulcer dyspepsia so
`challenging.
`
`LAWRENCE S. FRIEDiVlAN, M.D.
`Massachusetts General Hospital
`Boston, MA 02114
`
`1. Uppal R, Latcef SK, Korstcn M, Paronctto F, Licbcr CS. Chronic alco
`holic gastritis: roles of alcohol and Helicob~cter pylori. Arch Intern Mcd
`1991;151:760 4:.
`2. Wcinstcin WM. Other types of gastritis and gastropathics, including
`Mdn&ricr’s disease. In: Fcldman M, Slciscngcr MH, Scharschmidt BF, cds.
`Slciscngcr & Fordtran’s gastrointestinal and liver disease: pathophysiology/
`diagnosis/management. 6th cd. Vol. 1. Philadelphia: WIB. Saundcrs, 1998:
`711 32.
`3. Soil AH. Peptic ulcer and its complications. In: Fcldman M, Slciscngcr
`
`MH, Scharschmidt BF, eds. Sleisenger & Fordtran’s gastrointestinal and
`liver disease: pathophysiology/diagnosis/managcmcnt. 6th cd. Vol. 1. Phil
`adelphia: W.B. Saunders, 1998:620 78.
`4. Kuipcrs EJ, Thijs JC, Fcstcn HP. The prevalence of Hdicobacterpylori
`in peptic ulcer disease. Aliment Pharmacol Thor 1995;9:Suppl 2:59 69.
`g. Pounder R. Silent peptic ulceration: deadly silence or golden silence?
`Gastroenterology 1989;96:626 31.
`
`Chronic Heart Failure and the Quality of Life
`
`To the Editor: Stevenson, in her editorial in response to
`the article by Cohn et al. (Dec. 17 issue),1,2 ;vrites tellingly
`regarding the problems of and prospects for patients ;vith
`advanced congestive heart failure ;vho are foundering de
`spite the best pharmaceutical and systematized care and
`;vho are not candidates for transplantation or mechanical
`circulatory assistance.
`It is perhaps not surprising that advanced heart failure
`has been termed "heart cancer" because of the inexorable
`and uncomfortable course patients folio;v, ;vith their con
`dition refractory to conventional and experimental therapy.
`The analogy to advanced cancer may also provide guid
`ance on ho;v ;ve might best approach this gro;ving number
`of patients. Oncologists have long recognized that certain
`forms of treatment provide palliation ;vithout reasonable
`hope of recovery. Such palliation, often administered ;vith
`in the context of a hospice program, provides relief of suf
`feting, comfort, and the facilitation of;vhatever sell’deter
`mination is feasible given the patient’s condition, all in the
`patient’s venue of choice, usually home.
`The adoption of a "palliation" track in clinical or com
`passionate use studies of heart failure ;vould do much to
`provide optimal care to a small but gro;ving number ofpa
`tients ;vhose conditions have become refractory to the best
`available care. Patients for ;vhom no other conventional
`therapies are available are quite ;villing to accept a greater
`risk of death in exchange for the promise of an improved
`quality of life, particularly if they can be freed from the re
`volving door of recurrent hospitalization and the sinusoi
`dal polarities of compensation and decompensation at ever
`more frequent intervals and if they prefer the prospect of
`sudden death to the insidious or not so insidious decline
`that robs them of self determination.
`Well conceived studies or registries that allo;v for pallia
`rive therapies ;vith oral vesnarinone or milrinone, ;vith or
`;vithout beta blockers or rhythm management devices,
`;vould do much to help our patients, perhaps teach us
`much about this particular phase of the disease and stage
`of life, and allo;v us to better fulfill our roles as comforters
`;vhen a cure is no longer possible.
`
`Mar, c R. PRITZKER, M.D.
`Minneapolis Heart Institute Foundation
`Minneapolis, MN 55407
`
`1. Stevenson LW. Inotropic therapy for heart failure. N Engl J Med 1998;
`339:1848 50.
`2. Cohn JN, Goldstcin SO, Grccnbcrg BH, ct al. A dose dependent increase
`in mortality with vcsnarinonc among patients with severe heart failure.
`N Engl J Mcd 1998;339:1810 6.
`
`To the Editor: A cure for illness is the ultimate goal of
`both patients and physicians, but it is rarely achieved in the
`
`Volume 340 Number 19
`
`1511
`
`The New England Journal of Medicine
`Downloaded ~Yom nejm.org on December 8, 2016. For personal use only. No other uses without permission.
`Copyright © 1999 Massachusetts Medical Society. All rights reserved.
`
`Sandoz Inc. IPR2016-00318
`Sandoz v. Eli Lilly, Exhibit 1115-0004
`
`

`
`The New England Journal of Medicine
`
`case of most cardiovascular diseases. Cardiovascular spe
`cialists accept therapies that prolong survival and improve
`the quality of life and reject therapies that shorten survival
`and adversely aftect the quality of life. Confusion arises ;vhen
`a therapy achieves one of these objectives but not the other.
`Since ;ve measure our success by the number of lives saved,
`;ve accept therapies that prolong life even if they are asso
`ciated ;vith adverse efiects on the quality of life, provided
`that such effects are mild and transient. Similarly, ;ve rec
`ommend therapies that improve the quality of life even if
`they are associated ;vith a risk of death, provided the risk is
`relatively small and symptomatic improvement is clinically
`significant and long lasting. For example, ;ve accept a small
`surgical mortality rate to perform revascularization in a pa
`tient ;vith three vessel coronary artery disease ;vho has an
`gina and normal left ventricular function. What should ;ve
`do when a therapy carries a clinically significant risk of
`death and yet provides a statistically significant improve
`ment in the quality of life?
`Cohn et al. presented an interesting argument about the
`use ofvesnarinone in patients with severe congestive heart
`failure. They suggested that physicians might consider a
`trade off between increased mortality and improvement in
`the quality of life. In the accompanying editorial, Steven
`son discussed some helpful arguments from patients and
`physicians about the importance of the quality of life when
`it involves a disease with severe and disabling symptoms.
`No one denies the importance of the quality of life in this
`situation. However, I have doubts about the clinical signif
`icance of the improvement in the quality of life associated
`with vesnarinone. Although the improvement was statisti
`cally significant, I wonder whether patients appreciated
`the difference; even if they did, it did not last long. I also
`wonder how many of these patients had improvement in
`their congestive heart failure to class II or class I?
`In order for an improvement in the quality of life to be
`worth an increase in mortality, it has to be clinically signif
`icant in addition to being statistically significant and should
`be long lasting. I doubt that a two point difference in the
`degree of improvement in the quality of life is a fair trade
`for an increase of four points in mortality. Nonetheless,
`the answer is neither easy nor simple. Sometimes, vve may
`feel so desperate that vve need an answer ftom the dead. If
`they could tell us that they felt really good and did not
`mind death before its time, making the choice in favor of
`the quality of life improvements might be worth it. Oth
`ervvise, I am certainly not ready for that trade.
`
`I!vIAD A. ALHADDAD, M.D.
`
`Bronx Lebanon Hospital Center
`Bronx, NY 10457
`
`Dr. Cohn replies:
`
`To the Editor: The comments from Pritzker and Alhad
`dad represent the dialogue vve hoped to initiate vvith our
`article. As vve have previously noted and as cited in our pa
`per, sick patients are vvilling to trade an increased risk of
`death for an improvement in the quality of their daily liv
`ing. Indeed, after the data on vesnarinone vvere rdeased and
`the increased mortality rate vvas explained to the patients
`in the study, many opted to continue the therapy because
`of the perception that their symptoms had improved.
`
`1512
`
`May 13, 1999
`
`The mean benefit ofvesnarinone in terms of the quality
`of life is not particularly large or statistically sustained, so
`this drug may not be the ideal agent vvith vvhich to launch
`a discussion of the therapeutic dilemma. Nonetheless, vvith
`this and other drugs, individual responses may be dramatic.
`Part of this heterogeneity undoubtedly relates to the mul
`tiple and varied mechanisms contributing to the symp
`toms and progression of disease in the patients vve recruit
`for large scale trials. Although it is impossible to base ther
`apeutic recommendations on such largely anecdotal expe
`riences, vve must constantly remind ourselves that vve prac
`rice medicine on single patients and vve seek an individual
`response that may not match the mean efIect of an inter
`vention in a large trial. The challenge to the physician is
`to combine the data ftom these trials vvith other therapeu
`tic and patient related insights to form a rational approach
`to managing the care of a single person.
`
`JaY N. COHN, M.D.
`
`University of Minnesota Medical School
`Minneapolis, MN 55455
`
`Atovaquone Compared with Dapsone to Prevent
`Pneumocystis carinii Pneumonia
`
`To the Editor: Taken as a whole, the articles by H Sadr
`et at) and Bozzette et al.2 and the accompanying editorial
`by Steinbrook3 (Dec. 24 issue) seem to argue against the
`conclusion ofH Sadr et at. that atovaquone may be the pre
`ferred choice for prophylaxis against Pneumocystis carinii
`pneumonia in patients who cannot tolerate trimethoprim,
`sulfonamides, or both. It is surprising that H Sadr et at.
`never factored in cost issues related to their strategy of pro
`phylaxis. Dapsone and atovaquone, although found to be
`of equal efficacy for the prevention ofP. carinii pneumonia,
`have large differences in price.
`On the basis of the average suggested manufacturer’s
`~vholesale prices, the cost of a year of atovaquone is $10,100,
`as compared ~vith a cost of $72 for dapsone. Pentamidine,
`another alternative to atovaquone, costs $2,SSS a year, in
`cluding administration costs. If the retail prices of the
`drugs are considered, the cost differential bet~veen atova
`quone and dapsone is even greater. According to the esti
`mates of Bozzette et al., the average annual cost of health
`care per patient infected ~vith the human immunodeficiency
`virus (HIV) ~vas $22,200 in 1996, ~vith 40 percent of the
`cost being for pharmaceuticals (approximately $8,800). Fur
`thermore, Steinbrook estimates that the majority of patient
`care costs related to HIV infection in 1998 ~vere paid for
`by public programs.
`It is easy to see from these cost estimates that the strategy
`of using atovaquone rather than dapsone for the approxi
`mately 40 percent of patients ~vith AIDS ~vho are unable
`to tolerate trimethoprim sulfamethoxazole (cost per year,
`$207, based on a dose of one double strength tablet dai
`ly)4 ~vould greatly increase the costs of treatment and have
`a substantial effect on the average annual pharmaceutical
`expenditures for the entire population ~vith AIDS. Clearly,
`issues related to cost ~vould be even more problematic in
`the developing ~vorld. In our opinion, the cost consider
`ations are a per