Cacoub P, Ratziu V, Myers RP, et al

(1)

gression, only long-term follow-up can ascertain whether fibrosis subsequently improves, remains stable, or worsens in these patients who, before treatment, are known to have more rapid progression to fibrosis than HIV-seronegative patients.⁵ Fabrice Carrat, MD, PhD

carrat@u707.jussieu.fr Firouzé Bani-Sadr, MD Patrice Cacoub, MD

Groupe Hospitalier Universitaire Est Paris, France

Stanislas Pol, MD, PhD

Groupe Hospitalier Universitaire Ouest Paris, France

Christian Perronne, MD, PhD

Centre Hospitalier Universitaire Raymond Poincaré Université de Versailles

Garches

for theANRS HC02-RIBAVIC Study Team

1. Alric L, Plaisier E, Thebault S, et al. Influence of antiviral therapy in hepatitis C virus-associated cryoglobulinemic MPGN. Am J Kidney Dis. 2004;43:617-623.

2. Cacoub P, Ratziu V, Myers RP, et al. Impact of treatment on extra hepatic mani- festations in patients with chronic hepatitis C. J Hepatol. 2002;36:812-818.

3. Cacoub P, Lidove O, Maisonobe T, et al. Interferon-alpha and ribavirin treat- ment in patients with hepatitis C virus-related systemic vasculitis. Arthritis Rheum.

2002;46:3317-3326.

4. Hermine O, Lefrere F, Bronowicki JP, et al. Regression of splenic lymphoma with villous lymphocytes after treatment of hepatitis C virus infection. N Engl J Med. 2002;347:89-94.

5. Mohsen AH, Easterbrook PJ, Taylor C, et al. Impact of human immunodefi- ciency virus (HIV) infection on the progression of liver fibrosis in hepatitis C virus infected patients. Gut. 2003;52:1035-1040.

Evaluation of Suspected Adverse Drug Reactions To the Editor: We would like to raise some issues regarding the Commentary by Dr Strom about evaluation of suspected adverse drug reactions.¹A “signal” should always involve an element of clinical judgment. Statistical algorithms such as the proportional reporting ratio (PRR) and Bayesian approaches designed for application to spontaneous reports are promis- ing tools, but overconfidence in these methods should be avoided because they not only generate false alarms but may also fail to highlight potential safety issues.^2,3Statistical algorithms should not be considered as an alternative to case-by- case review of all reports submitted to a spontaneous reporting system (SRS). The incremental utility of statistical algorithms relative to clinical approaches is still unknown. We believe that statistical algorithms should only be considered as potential supplements to a comprehensive signal detection strategy founded on rigorous application of clinical and epidemiological knowledge and judgement, including sen- sible clinical criteria for evaluating cumulative case reports.

The usefulness of statistical algorithms is highly situation- dependent, because even an efficient screening test look- ing for a “needle in a haystack” will often come up with hay.

The greatest utility of statistical algorithms may be for de- tecting signals of higher-order associations, such as com- plex drug-drug interactions and drug-induced syndromes, which are less amenable to discovery by clinical cognition.

Strom mentions that statistical approaches were devel- oped to identify associations that should be subjected to controlled studies and states that except for rare exceptions, detection of a signal should appropriately trigger an epidemiological study. However, the opposite may be true in pharmacovigilance settings in which signals can often be adju- dicated with sufficient certainty to make decisions (eg, amendment to a product label) based on a case-by-case review of spontaneous reports.

Finally, Strom points out suggestions that disproportionality analyses should not be published because they are equiva- lent to anecdotal case reports in terms of level of evidence. How- ever, case reports have always been a valuable source of information derived from clinical practice and are the corner- stone of signal detection in the postmarketing phase. It is the accumulation of information that constitutes a signal, not the results of calculations on the data set of an SRS alone.^4,5 Manfred Hauben, MD, MPH

manfred.hauben@pfizer.com Medical Director

Risk Management Strategy Pfizer Inc

New York, NY

Eugène P. van Puijenbroek, MD, PhD Netherlands Pharmacovigilance Centre Lareb s-Hertogenbosch, the Netherlands

Financial Disclosure: Dr Hauben is an employee of Pfizer.

1. Strom BL. Potential for conflict of interest in the evaluation of suspected ad- verse drug reactions: a counterpoint. JAMA. 2004;292:2643-2646.

2. Hauben M. Early postmarketing safety surveillance: data mining points to consider.

Ann Pharmacother. 2004;38:1625-1630.

3. Hauben M. Application of an empirical Bayesian data mining algorithm to re- ports of pancreatitis with atypical antipsychotics. Pharmacotherapy. 2004;24:1122- 1129.

4. Vandenbroucke JP. In defense of case reports and case series. Ann Intern Med.

2001;134:330-334.

5. Aronson JK. Anecdotes as evidence [editorial]. BMJ. 2003;326:1346.

In Reply: I strongly agree with Drs Hauben and van Puijen- broek that true signals should emerge from clinical judgment and that statistical algorithms, such as PRRs,^1-5should be used as supplements to clinical and epidemiological judgment, not replacements. I also agree that the value of statistical algorithms even in that role remains unproven. Unfortunately, however, statistical algorithms are too often used alone, in publi- cations and in the courtroom, as if they represent analyses useful for hypothesis testing, which is inappropriate.

I also agree that sometimes signals are acted upon with- out waiting for epidemiological confirmation, despite the desirability of waiting. However, there is always a risk of acting prematurely and incorrectly. When the signal is made stronger by combining it with clinical judgment, and if the action is relatively minor (such as product label changes), an early decision is easier to make.

Regarding the utility of publishing case reports, that is a decision for journals to make. There are case reports that journals deem to be worth publishing because of clinical interest as anecdotes, while other studies may be useful to raise LETTERS

(2)

hypotheses. However, the Food and Drug Administration is now approaching 400 000 adverse drug reaction reports per year in its SRS,⁶and it would not be reasonable to pub- lish all as case reports.

My central points remain. Case reports are primarily useful for hypothesis generation, and I still concur with the ob- servation by Hennessy that “ . . . anecdotal case reports and disproportionality measures of them are of the same es- sence, and distinct from controlled epidemiologic studies.”¹Disproportionality measures are formal statistical analyses of poor, incomplete, and biased data (ie, SRS data). No matter how sophisticated, formal analyses of such data can easily be misleading; they cannot correct for imperfections in the data source. Proper interpretation also requires clinical judgment before one even considers there to be a signal. However, the use of such statistical algorithms of spon- taneously reported data as if the algorithms tested hypotheses, is incorrect. This is another example of the broader lesson illuminated so clearly in recent months: our country needs to strengthen its drug safety system so we can collect more and better data, and use the data more wisely.

Brian L. Strom, MD, MPH bstrom@cceb.med.upenn.edu

Center for Clinical Epidemiology and Biostatistics University of Pennsylvania School of Medicine Philadelphia

Financial Disclosure: See original article ( JAMA. 2004;292:2643-2646).

1. Hennessy S. Disproportionality analyses of spontaneous reports. Pharmaco- epidemiol Drug Saf. 2004;13:503-504.

2. Zhou W, Pool V, DeStefano F, Iskander JK, Haber P, Chen RT; VAERS Work- ing Group. A potential signal of Bell’s palsy after parenteral inactivated influenza vaccines: reports to the Vaccine Adverse Event Reporting System (VAERS)—

United States, 1991-2001. Pharmacoepidemiol Drug Saf. 2004;13:505-510.

3. Shapiro S. Clinical judgment, common sense and adverse reaction reporting.

Pharmacoepidemiol Drug Saf. 2004;13:511-513.

4. Rothman KJ, Lanes S, Sacks ST. The reporting odds ratio and its advantages over the proportional reporting ratio. Pharmacoepidemiol Drug Saf. 2004;13:519- 523.

5. Waller P, van Puijenbroek E, Egberts A, Evans S. The reporting odds ratio ver- sus the proportional reporting ratio: “deuce.” Pharmacoepidemiol Drug Saf. 2004;

13:525-526.

6. Ahmad SR, Goetsch R, Marks N. Spontaneous reporting in the United States.

In: Strom BL, ed. Pharmacoepidemiology. 4th ed. New York, NY: John Wiley &

Sons. In press.

Ownership and Use of Tissue Specimens for Research

To the Editor: I am the physician-researcher formerly affili- ated with Washington University in St Louis, Mo, whose le- gal case was discussed in the Health Law and Ethics article on ownership and use of tissue specimens for research.¹Drs Hakimian and Korn got it backward in stating that I main- tain that individuals enrolled in my research protocols dur- ing my tenure at Washington University “ . . . waived their rights to their tissue samples through language in the in- formed consent document.” To the contrary, it is my strong belief that, according to federal regulations, research participants cannot waive their rights to their tissue samples.²

Under the Code of Federal Regulations, obtaining identifi- able private information or identifiable specimens under in- formed consent for research purposes constitutes human subjects research.³Federal regulations state that research participants may withdraw from research at any time and for any reason and that they cannot waive this right.⁴Because obtaining either identifiable private information or identifiable specimens constitutes human subjects research, both com- ponents are subject to federal regulations; hence, research participants have the right to withdraw both data and specimens.

In this instance, approximately 6000 individuals have re- quested in writing that their samples be transferred to my cus- tody for continued use in my research.

It is understandable that, given the potential commer- cial value of human tissue and blood samples, universities would like to assert sole right of ownership. However, this case is about patients’ and other human research participants’ rights and their sometimes highly personal relation- ship with their physician-researcher. Universities cannot assert sole ownership to samples that participants can withdraw at any time for any reason.

There is also an important distinction in that the previ- ously litigated cases cited by Hakimian and Korn relate to secondary commercialized products derived from samples and related patent rights, not rights relating to the primary raw samples that are at issue in this litigation.

William J. Catalona, MD wcatalona@nmff.org

Robert H. Lurie Comprehensive Cancer Center Northwestern University Feinberg School of Medicine Chicago, Ill

1. Hakimian R, Korn D. Ownership and use of tissue specimens for research. JAMA.

2004;292:2500-2505.

2. Code of Federal Regulations. 45 CFR part 46.116 (2002).

3. Code of Federal Regulations. 45 CFR part 46.116.102(f ) (2002).

4. Code of Federal Regulations. 45 CFR part 46.116(8) (2002).

In Reply: The case in which Dr Catalona is involved pre- sents potentially important issues for academic research- ers. For this reason, we described it briefly in our article on ownership of tissue samples, taking our information from documents filed with the district court, as noted in the article. Dr Catalona emphasizes in his letter that his lawsuit pertains to competing rights in a tissue collection rather than rights in research products derived from human tissue samples. We did not speculate about the merits of the case because the parties have neither presented all the underly- ing facts to the court nor argued their interpretations of ap- plicable laws, and the court has not rendered a decision.

Our article analyzes the existing judicial decisions and the regulatory and ethical standards that affect ownership of tissue samples used for research. Federal regulations do not address ownership of tissue samples directly. There- fore, we suggest that some of Dr Catalona’s concerns be di- rected toward the policy questions of control and use of specimens. Should tissue sources, investigators, sponsors, or LETTERS