Home

Current Issue

Archives

Guidelines for Authors

Classified Ads

Links

Search PubMed

Subscriptions

Subscriber Registration

Guidelines for Website Users

JRheum Update Service

Contact Info

Correspondence

To the Editor:

We read with interest the article by Hashkes and Uziel¹. The authors mention that there were no studies of the publication rates from general rheumatology meetings. We agree; we are completing a similar study.

We wished to determine the publication rate in Medline–indexed journals of the abstracts presented at the XXII Spanish Society for Rheumatology meeting held in Zaragoza in 1996 and published in Revista Española de Reumatología². A computer search was performed with the Medline data base to determine if the abstract had been published in a peer-reviewed journal from 1995 to 2001. All presentations in either poster or oral form were reviewed. The searches began with the first author's name. If a match was not found, all the authors' names were used. If a match was still not found it was assumed that the article was not published in a journal retrievable by Medline.

A total of 249 abstracts were reviewed, 200 posters (80%) and 49 (20%) oral comunications. Of these 249 abstracts, 52 were published in peer-reviewed journals, giving an overall publication rate of 21%. The average delay between the meeting and publication was 18.5 months. The majority of the papers were published in rheumatologic journals (63%). The Journal of Rheumatology was the preferred journal (Table 1). Four of the abstracts had been published prior to the meeting.

Table 1 . Journals in which abstracts were published (n = 52).

The publication rate of 21% was lower than other meetings, such as the Park City 4 meeting. This might indicate that the presentations were of lesser quality and the meeting less informative. However, it is important to note that language may be a barrier for the final publication of an abstract, particularly if the original language of the meeting is not English.

ALEJANDRO OLIVÉ, MD, PhD, Rheumatology Section; INMA CIFUENTES, Former Librarian, JOSEP ROCA, MD, Epidemiology Unit; Hospital Universitari Germans Trias i Pujol, Crta Canyet sn, Badalona 08916, Spain. E-mail: aolive@ns.hugtip.scs.es

REFERENCES

1. Hashkes PJ, Uziel Y. The publication rate of abstracts from the 4th Park City Pediatric Rheumatology Meeting in peer-reviewed journals: What factors influenced publication? J Rheumatol 2003;30:597-602.

2. XXII Congreso nacional de la Sociedad Española de Reumatología. Rev Esp Reumatol 1996;23:153-216.

Drs. Hashkes and Uziel reply

To the Editor:

The timely dissemination of scientific knowledge is crucial and many important advances may be missed due to lack of publication of findings in peer-reviewed journals. Unfortunately, for various reasons, some noted in our report¹, many studies do not progress from presentation at a scientific meeting to a full, published article.

Olivé, et al found that only 21% of the abstracts presented at the XXII Spanish Society for Rheumatology meeting in 1996 were eventually published in peer-reviewed journals. These findings were similar to those from another recent study originating from Spain, in which only 31% of research proposals submitted to a research ethics committee were eventually published². Besides the factors noted by the authors regarding the possible low quality of some presentations at their annual national meeting and the language barrier, we noted in our study that the geographic location of the authors was one of the most significant factors in determining eventual publication¹, even if the original abstract was published in English. Research quality as well as the pressure to publish and the time physicians have to dedicate to writing may differ between countries.

Studies such as those by Olivé, et al, ourselves, and others should not be viewed only as an intellectual exercise or as a "pat on the back" for our scientific societies' success. It is important to analyze those factors associated with an increased or decreased rate of publication. Issues of study design (case report, descriptive, analytical), study topic (etiology, epidemiology, clinical manifestations, treatment, etc.), "positivity" of results, novelty of the study, and abstract quality (whether accepted for presentation at a scientific meeting by poster or oral presentation) should be investigated in order to detect those factors that can be addressed and rectified prior to study design and writing of the paper. Only by such a process may we be able to assist researchers in planning studies that eventually will result in the propagation of scientific knowledge through the most credible method, the peer-review process.

PHILIP J. HASHKES, MD, MSc, Head, Pediatric Rheumatology, Department of Rheumatic Diseases, Cleveland Clinic Foundation, A50, 9500 Euclid Avenue, Cleveland, Ohio 44195, USA; YOSEF UZIEL, MD, MSc, Department of Pediatrics, Meir Hospital, Kfar-Saba, Israel. E-mail: hashkep@ccf.org

REFERENCES

1. Hashkes PJ, Uziel Y. The publication rate of abstracts form the 4th Park City Pediatric Rheumatology Meeting in peer-reviewed journals. What factors influenced publication? J Rheumatol 2003;30:597-602.

2. Pich J, Carne X, Arnaiz JA, Gomez B, Trilla A, Rodes J. Role of a research committee in follow-up and publication of results. Lancet 2003;361:1015-6.

Serious Gastrointestinal Events from Low Dose Analgesic Use

To the Editor:

Fries and Bruce¹ conclude that over-the-counter (OTC) use of aspirin (ASA), ibuprofen (IBU), or acetaminophen (APAP) carries little risk of serious gastrointestinal (GI) toxicity, and that these 3 analgesics at low or intermediate doses have rates of serious GI toxicity that cannot be distinguished from each other or from background. Their study data do not support the latter conclusions, nor are their conclusions consistent with the results of several previous studies.

On the basis of information obtained from semiannual questionnaires about use of ASA, IBU, APAP, other nonsteroidal antiinflammatory drugs (NSAID), or corticosteroids, each 6 months of followup for each patient from a cohort of people with rheumatoid arthritis (RA) or osteoarthritis (OA) was assigned by Fries and Bruce to one of 9 categories of analgesic use¹. The categories were defined by use of each of the 3 study analgesics alone (i.e., neither the other 2 study analgesics nor any other NSAID nor corticosteroids were used during the 6 month period), use of each study analgesic with concurrent therapy (i.e., at least one of the other study analgesics or another NSAID was used as well, but no corticosteroids were used), or use of each study analgesic with corticosteroids and concurrent therapy (i.e., corticosteroids and possibly also other analgesics were used as well). The rates of adverse GI events associated with ASA, IBU, and APAP were then compared during periods of use alone, periods of use concurrent with other analgesics, or periods of use with corticosteroids and possibly other analgesics. No attempt was made in these comparisons to adjust for use of other medications (e.g., antihypertensive drugs and cardiovascular medications), which the authors claim are "not known to be associated with serious GI events." In addition, no attempt was made to adjust for chronic diseases other than arthritis, in spite of the authors' admission that comorbid disease conditions would be expected in RA and OA patients.

There may be no combination of drug and disease with a greater potential for confounding by indication than the combination of APAP and adverse GI events². The risk of adverse GI events is increased in people with chronic diseases other than arthritis, such as heart failure and diabetes^3,4. Anticoagulants increase the risk of GI bleeding more than any other pharmacologic agents, although other drugs such as digoxin and diuretics have also been shown to increase the risk of adverse GI events more than corticosteroids^3,4. The potential for confounding by indication arises because APAP is preferentially prescribed (rather than NSAID) for patients with a wide variety of chronic diseases^5,6. As a result of this channeling bias patients taking APAP are much more likely than patients taking NSAID to have a variety of chronic diseases (e.g., heart failure, ischemic heart disease, cancer, or renal failure), to be taking several different types of medications (e.g., corticosteroids, anticoagulants, antihypertensive drugs, or antidiabetic agents), to be taking multiple medications, and to have had a recent hospitalization^7,8. The fact that people taking APAP are more likely to be chronically or seriously ill leads to profound biases in the estimates of mortality risks in APAP users⁹. In most epidemiologic studies, information on disease history and drug history is inadequate or incomplete. Thus, residual confounding by indication will likely remain in studies of APAP and adverse GI events, even when a thorough statistical adjustment based on all relevant information is attempted¹⁰. Fries and Bruce failed to adjust for comorbid diseases and classes of medications known to be associated with adverse GI events in their analyses, and hence their risk estimates for APAP are almost certainly positively biased due to confounding by indication².

In spite of the likelihood of confounding by indication in the estimates presented by Fries and Bruce, it is noteworthy that in periods in which only a single analgesic was used, APAP had the lowest rate of serious GI events in both the RA and the OA patients, whether the rates were calculated per person (their Table 3) or on the basis of person-years (their Table 4)¹. Higher rates of GI events for APAP users were observed only in periods in which other drugs were used as well, that is, periods in which confounding by indication would be most likely. Fries and Bruce argue that the appearance of a dose-response for APAP in periods of multiple drug use argues against confounding by indication, but this is almost certainly not the case. Physicians may be particularly careful to prescribe APAP rather than an NSAID to high-risk patients when a full therapeutic dose of an analgesic is needed; that is, even an apparent dose-response can be induced by confounding by indication. Thus the results of the statistical analyses by Fries and Bruce do not support the conclusion that APAP, ASA, and IBU have indistinguishable GI toxicities.

Although Fries and Bruce state in their Abstract that the risk of GI complications at low doses of NSAID "remains unknown," there is ample evidence, including evidence from randomized clinical trials, of increased risk of GI bleeding for ASA doses as low as 75 mg per day^11-15. A number of studies comparing the risk of adverse GI events in users of APAP and users of ASA or NSAID have concluded that the risk is lower in users of APAP, at both low and high doses^5,16-18. Moreover, a recent analysis of the ARAMIS data bank that was the basis of the Fries and Bruce study reported a relative risk of 3.9 for serious GI complications for OTC use of NSAID, but reported that APAP "was not associated with increased risk of GI complications."¹⁹

Fries and Bruce cite 2 studies as being consistent with their conclusion that APAP does not have lower rates of adverse GI events than NSAID at low doses, but neither study supports such a conclusion. The first, a case-control study, reported no increase in the risk of GI complications for APAP at doses of 2000 mg per day or lower²⁰. In contrast, the risk of GI complications for low or medium doses of NSAID in the same study was 2.4 (95% confidence interval 1.9, 3.1)²⁰. Even the reported increased risk of GI bleeding at higher doses of APAP in this case-control study must be viewed with caution, because of the likely influence of confounding by indication². Adjustment for potential confounding factors resulted in decreases in the relative risk estimates (often substantial decreases) for every measure of APAP use in this study (see their Table 1)²⁰, suggesting that positive bias due to residual confounding by indication is almost certain to be present in the risk estimates for APAP. The second study cited by Fries and Bruce in support of their conclusions was a retrospective cohort study⁷. This study documented the high degree of confounding by indication that is likely in evaluating adverse GI events associated with APAP (compared to NSAID). Although adjustment using a propensity score was employed, it is unlikely that such adjustment completely eliminated confounding by indication in the estimates of risk for APAP. In spite of the likely presence of residual confounding, however, the risk of adverse GI bleeding events (relative to the risk at high doses of NSAID) was lower for APAP at doses of 2600 mg per day or lower (relative risks ranging from 0.59 to 0.78) than the risk of adverse GI bleeding events for low dose NSAID (relative risk of 0.9) (see results in Table 6)⁷. Thus both studies cited by Fries and Bruce are consistent with a lower risk of adverse GI events at low doses of APAP compared to low doses of NSAID.

The potential for adverse events associated with commonly used drugs must always be considered and balanced against the potential benefits. Based on careful considerations of such risks and benefits, APAP has been proposed as the first-line analgesic therapy for patients with OA and RA^19,21 as well as for people taking low dose ASA for cardioprotection²². The absolute risk of adverse GI events is low for ASA and IBU at low doses, but the even lower risk of adverse GI events with APAP should be considered in making decisions about analgesic use.

ROBERT E. TARONE, PhD; LOREN LIPWORTH, ScD; JOSEPH K. McLAUGHLIN, PhD, International Epidemiology Institute, Rockville, Maryland, and Department of Medicine, Vanderbilt University, Nashville, Tennessee, USA

The International Epidemiology Institute has received unrestricted research grants from and has consulted on a variety of research topics with McNeil Consumer Healthcare.

REFERENCES

1. Fries JF, Bruce B. Rates of serious gastrointestinal events from low dose use of acetylsalicylic acid, acetaminophen, and ibuprofen in patients with osteoarthritis and rheumatoid arthritis. J Rheumatol 2003;30:2226-33.

2. Signorello LB, McLaughlin JK, Lipworth L, Friis S, Sørensen HT, Blot WJ. Confounding by indication in epidemiologic studies of commonly used analgesics. Am J Ther 2002;9:199-205.

3. Aalykke C, Lauritsen K. Epidemiology of NSAID-related gastroduodenal mucosal injury. Best Pract Res Clin Gastroenterol 2001;15:705-22.

4. Weil J, Langman MJS, Wainwright P, et al. Peptic ulcer bleeding: accessory risk factors and interactions with non-steroidal anti-inflammatory drugs. Gut 2000;46:27-31.

5. Andrade SE, Martinez C, Walker AM. Comparative safety evaluation of non-narcotic analgesics. J Clin Epidemiol 1998;51:1357-65.

6. Day RO, Graham GG, Whelton A. The position of paracetamol in the world of analgesics. Am J Ther 2000;7:51-4.

7. Rahme E, Pettitt D, LeLorier J. Determinants of sequelae associated with utilization of acetaminophen versus traditional nonsteroidal antiinflammatory drugs in an elderly population. Arthritis Rheum 2002;46:3046-54.

8. Rahme E, Marentette MA, Kong SX, LeLorier J. Use of NSAIDs, COX-2 inhibitors, and acetaminophen and associated coprescriptions of gastroprotective agents in an elderly population. Arthritis Rheum 2002;47:595-602.

9. Lipworth L, Friis S, Mellemkjaer L, et al. A population-based cohort study of mortality among adults prescribed paracetamol in Denmark. J Clin Epidemiol 2003;56:796-801.

10. Davey Smith G, Ebrahim S. Data dredging, bias, or confounding. BMJ 2002;325:1437-8.

11. Weil J, Colin-Jones D, Langman M, et al. Prophylactic aspirin and risk of peptic ulcer bleeding. BMJ 1995;310:827-30.

12. Isles C, Norrie J, Paterson J, Ritchie L. Risk of major gastrointestinal bleeding with aspirin. Lancet 1999;353:148-9.

13. Derry S, Loke YK. Risk of gastrointestinal haemorrhage with long term use of aspirin: meta-analysis. BMJ 2000;321:1183-7.

14. Sorensen HT, Mellemkjaer L, Blot WJ, et al. Risk of upper gastrointestinal bleeding associated with use of low-dose aspirin. Am J Gastroenterol 2000;95:2218-24.

15. Collaborative Group of the Primary Prevention Project. Low-dose aspirin and vitamin E in people at cardiovascular risk: a randomized trial in general practice. Lancet 2001;357:89-95.

16. Levy M, Miller DR, Kaufman DW, et al. Major upper gastrointestinal tract bleeding: relation to use of aspirin and other nonnarcotic analgesics. Arch Intern Med 1988;148:281-5.

17. Blot WJ, McLaughlin JK. Over the counter non-steroidal and anti-inflammatory drugs and risk of gastrointestinal bleeding. J Epidemiol Biostat 2000;5:137-42.

18. Lewis SC, Langman MJ, Laporte JR, Matthews JN, Rawlins MD, Wiholm BE. Dose-response relationships between individual nonaspirin nonsteroidal anti-inflammatory drugs (NANSAIDs) and serious upper gastrointestinal bleeding: a meta-analysis based on individual patient data. Br J Clin Pharmacol 2002;54:320-6.

19. Singh G. Gastrointestinal complications of prescription and over-the-counter nonsteroidal anti-inflammatory drugs: a view from the ARAMIS database. Am J Ther 2000;7:115-21.

20. Garcia Rodriguez LA, Hernandez-Diaz S. Relative risk of upper gastrointestinal complications among users of acetaminophen and nonsteroidal anti-inflammatory drugs. Epidemiol 2001;12:570-6.

21. Shamoon M, Hochberg MC. Treatment of osteoarthritis with acetaminophen: efficacy, safety, and comparison with nonsteroidal anti-inflammatory drugs. Am J Med 2001;110 Suppl 3A:46-9.

22. Cryer B. Gastrointestinal safety of low-dose aspirin. Am J Manag Care 2002;22 Suppl:701-8.

To the Editor:

The recent article by Drs. Fries and Bruce¹ states several conclusions that deserve comment and, unless and until confirmed, need to be viewed with cautious skepticism.

The conclusion that risk of gastrointestinal (GI) events with acetaminophen is low is likely correct. However, the conclusion that the risk of GI events for the 3 agents is similar not only runs contrary to conventional wisdom, it contradicts the large, established body of literature that documents that nonsteroidal antiinflammatory drugs (NSAID), aspirin, and ibuprofen have substantial GI risks^2-5 in contrast to non-NSAID analgesics such as acetaminophen. Therefore the authors' observations are likely flawed by significant methodological biases inherent in their assessments and by confounding by indication.

The GI events Fries and Bruce describe are neither numerically nor qualitatively consistent with the number or pattern of GI events described in study populations previously. From a quantitative perspective, the authors expressed GI event rates per 1000 patient-years while the convention for studies of this type is to report event rates per 100 patient-years^6,7. Well established data indicate that expected rates of GI events with low dose aspirin should be 0.5 to 3.0% per 100 patient-years^2,3. In the Fries study, GI event rates with aspirin are 4.0% per 1000 patient-years¹, roughly 10% of that expected based on previous reports. Similarly, with low dose ibuprofen the authors report event rates that are likely one-half to one-tenth that expected from previous studies of low dose ibuprofen. Since GI risks of aspirin have previously been extensively evaluated, GI event rates with aspirin should serve as a standard to validate the accuracy of the Fries study's methods and results. Lack of quantitative consistency with well documented GI event rates such as with aspirin indicates that this study lacks external validation.

Since the study appears to under-report GI events, it quite likely lacks sufficient power or sensitivity to accurately capture or compare rates of GI events with aspirin, ibuprofen, or acetaminophen. Thus its conclusion that GI risks of aspirin and ibuprofen are equivalent to background rates is flawed by study methodology that is insufficiently powerful to appreciate the well documented and universally accepted increased GI risk over background conferred by these medications. Similarly, the conclusion that GI risks of acetaminophen, aspirin, and ibuprofen do not differ from one another likely reflects methodology insufficiently sensitive to differentiate GI risks among analgesic therapies.

A widely held tenet of drug induced GI injury is that the extent of GI injury is proportionate to the dose. Yet in the Fries study, just the opposite was observed. With higher doses of acetaminophen, the authors observed lower, or more specifically, no GI event rates (their Table 5)¹. Osteoarthritis patients who took ibuprofen for less than 10 days each month experienced a greater number of events than those taking ibuprofen between 11 and 25 days of the month (Table 6). The same inverse frequency relationship was observed with acetaminophen, with infrequent users of acetaminophen experiencing higher event rates than frequent users. It is important that a study's observations be consistent with the known mechanism of action of a medication. The higher GI event rates observed with lower and less frequent analgesic doses imply that the Fries study results are mechanistically implausible.

The authors observed higher GI event rates in acetaminophen users taking concurrent ulcerogenic medications such as NSAID or corticosteroids than in those using aspirin or ibuprofen and also taking concurrent medications. This observation most probably reflects confounding by indication; those at higher risk for GI bleeding would more likely be prescribed acetaminophen. The authors contend that differences in their study groups did not exist because propensity scores, a statistical test to suggest similarity of groups, were similar. In contradiction, they also state that the GI risks of their study groups were not matched and that the ibuprofen study group had a higher percentage of patients at low risk for GI events than the other analgesic groups. Otherwise stated, the acetaminophen study group was disproportionately enriched with patients who, at baseline, were at higher risk to develop significant GI events. Thus, it is to be expected that the acetaminophen cohort would have had a higher incidence of GI events even in the absence of exposure to acetaminophen.

In summary, the conclusions reached by Drs. Fries and Bruce contradict a large body of literature that indicates that ibuprofen and aspirin are associated with increased GI risks compared to acetaminophen. Their conclusions are likely biased by flawed insensitive methodology and confounding by indication.

DAVID A. PEURA, MD, FACP, Professor of Medicine, Associate Chief of Gastroenterology and Hepatology, University of Virginia Health System, PO Box 800708, Charlottesville, Virginia 22908-0708, USA.

REFERENCES

1. Fries JF, Bruce B. Rates of serious gastrointestinal events from low dose use of acetylsalicylic acid, acetaminophen, and ibuprofen in patients with osteoarthritis and rheumatoid arthritis. J Rheumatol 2003;30:2226-33.

2. Thomas J, Stauss WL, Bloom BS. Over-the-counter nonsteroidal anti-inflammatory drugs and risk of gastrointestinal symptoms. Am J Gastroenterol 2002;97:2215-9.

3. Chan FKL, Chung SCS, Suen BY, et al. Preventing recurrent upper gastrointestinal bleeding in patients with Helicobacter pylori infection who are taking low-dose aspirin or naproxen. N Engl J Med 2001;344:967-73.

4. Lai KC, Lam SK, Chu KM, et al. Lansoprazole for the prevention of recurrent ulcer complications from long-term low-dose aspirin use. N Engl J Med 2002;346:2033-8.

5. Garcia Rodriguez LA, Hernandez-Diaz S, de Abajo FJ. Association between aspirin and upper gastrointestinal complications: systematic review of epidemiologic studies. Br J Clin Pharmacol 2001;52:563-71.

6. Silverstein FE, Faich G, Goldstein JL, et al. Gastrointestinal toxicity with celecoxib vs. nonsteroidal anti-inflammatory drugs for osteoarthritis and rheumatoid arthritis. JAMA 2000;284:1247-55.

7. Bombardier C, Laine L, Reicin A, et al. Comparison of upper gastrointestinal toxicity of rofecoxib and naproxen in patients with rheumatoid arthritis. VIGOR Study Group. N Engl J Med 2000;343:1520-8.

Drs. Fries and Bruce reply

To the Editor:

We read with interest the comments of David Peura, Robert Tarone, and their colleagues on our recent article¹, which suggested that at low, intermittent, and over-the-counter (OTC) doses in low risk patients the rates of serious GI complications are low for aspirin (ASA), acetaminophen (APAP), and ibuprofen, and the 3 drugs cannot be statistically distinguished from each other. This sensible finding, granting the dose-effect relationship, had not previously been addressed prospectively because OTC intake is not generally available in prospective datasets. The datasets we used were the ones that we had used to originally establish the epidemiologic boundaries of nonsteroidal antiinflammatory drug (NSAID) gastropathy, and the doses and incidences reported are consistent with previous reports^2,3. Our drug, dose, and event detection techniques are protocol-driven and were identical for each of the 3 drugs. We adjusted for possible confounding by indication by use of validated propensity scores, which is the standard for observational studies.

Peura's letter is laced with "contrary to conventional wisdom," "contradicts the large, established...," and other unsupported appeals to authority. Rather oddly, of the 6 studies that he cites, 5 do not even mention acetaminophen and the sixth is a telephone survey concerning symptoms rather than serious complications. The toxicity of aspirin and other NSAID at high or prescription doses is not disputed by our article. But the average person takes only 6–10 pills per month of these OTC medications. No prospective comparative study had been reported at these dosages, hence our report.

A need for confirmation of results is requested by Peura. But in our article¹ we had already cited 2 prior articles documenting associations between acetaminophen and serious gastrointestinal (GI) events and suggesting that there might be cracks in the conventional wisdom^4,5. If the conventional wisdom was always correct we could stop doing studies, but this wisdom held, not too long ago, "no acid, no ulcer."

Definitive confirmation of our results by randomized double-blind controlled trial was reported while our report was in press. Moore and colleagues⁶ reported results from the PAIN study of 8633 subjects comparing ASA, APAP, and ibuprofen in low doses, eliminating the possibility of confounding by indication. Outcomes were adverse events, serious adverse events, and serious GI adverse events. The number and nature of concomitant medication proved a more important predictor of serious GI events than the specific OTC analgesic. Absent additional risk factors, aspirin was about 1.5 times as toxic as APAP or ibuprofen, which were closely similar. With 3 or more concomitant medications, APAP was associated with greater toxicity than either of the other 2 drugs (p < 0.01).

The mechanism behind these now consistent findings is not completely clear, in part because of only weak evidence of a causal mechanism for acetaminophen GI toxicity. Perhaps it is most reasonable to suggest, as Moore, et al imply, that at low doses of these drugs where the incidence of causal events from the analgesic is low, the concomitant background risks for GI bleeding are of greater magnitude than the specific analgesic and mask true differences between drugs. The postulation of drug–drug interactions involving acetaminophen also is not unreasonable.

We think that Tarone and colleagues protest too much. Our article is by no stretch of the imagination an attack upon the GI safety of acetaminophen and we scrupulously avoided, in both the abstract and the text, concluding that APAP was more toxic than ASA and ibuprofen, even though the data trended that way in the more complicated patient groups. We did conclude that at the lowest and most intermittent doses of these 3 OTC analgesics there was insufficient statistical resolving power to show differences between specific medications and noted that all were relatively safe.

Tarone, et al discuss the literature quite differently than do the authors of the articles they cite. Thus, Hochberg, et al subsequently noted substantial reservations about possible APAP toxicity^7,8, Garcia-Rodriguez and Hernandez-Diaz⁴ noted a relative risk of APAP users over nonusers of 3.6 with doses above 2 g/day, consistent with our findings in more complicated patients, and Rahme, et al⁵ would not agree with Tarone, et al that their study documented a high degree of confounding by indication. A report by a former member of our group⁹ was never published in full because the APAP data had not been collected in the same way as the NSAID data and APAP use was greatly under-reported; this problem was corrected in the present report.

Moore, et al⁶, as above, have refuted speculation about a major role for confounding by indication. Confounding by indication cannot easily occur in a randomized controlled trial.

What do we think is happening? Let's assume, as Tarone, et al would like to, (1) that APAP is innately twice as safe as NSAID for serious GI toxicity. Let's assume also, as with the PAIN data and ours, (2) that patients at greater risk may have a 6-fold increased risk over healthier patients. And assume also (3), as Avogadro's number and broad clinical data indicate, that toxicity goes to zero as dose goes to zero. Two times zero is zero. It follows then that as dose decreases drugs become progressively safer and ultimately cannot be distinguished from each other, while effects of other risks remain and may dominate the risk profile. This is what we all have been observing.

Why is this more than an academic discussion? This involves patients. Drugs are both safe and toxic. We emphasize toxicity to increase caution about injudicious recommendation. We emphasize safety to reduce the concerns raised by the toxicity. In practice, if it is argued, as Tarone, et al would do, that there are large differences in safety between low doses of different OTC analgesics in actual use when there are not (for whatever reason), this balance is disturbed, caution decreased, and risk increased. We hope that these authors and their sponsor can join us in a balanced discussion of patient risks and patient benefits.

JAMES F. FRIES, MD; BONNIE BRUCE, DrPH, Division of Immunology and Rheumatology, Stanford University School of Medicine, 1000 Welch Road, Suite 203, Palo Alto, California 94304-1808, USA

REFERENCES

1. Fries JF, Bruce B. Rates of serious gastrointestinal events from low dose use of acetylsalicylic acid, acetaminophen, and ibuprofen in patients with osteoarthritis and rheumatoid arthritis. J Rheumatol 2003;30:2226-33.

2. Fries JF, Williams CA, Bloch DA, Michel BA. NSAID-associated gastropathy: incidence and risk factor models. Am J Med 1991;91:213-22.

3. Fries JF, Williams CA, Bloch DA. The relative toxicity of non-steroidal anti-inflammatory drugs (NSAIDs). Arthritis Rheum 1991;34:1353-60.

4. Garcia-Rodriguez LA, Hernandez-Diaz S. Relative risk of upper gastrointestinal complications among users of acetaminophen and nonsteroidal anti-inflammatory drugs. Epidemiology 2001;12:570-6.

5. Rahme E, Pettitt D, LeLorier J. Determinants and sequelae associated with utilization of acetaminophen versus traditional nonsteroidal anti-inflammatory drugs in an elderly population. Arthritis Rheum 2002;46:3046-54.

6. Moore N, Charlesworth A, Van Ganse E, et al. Risk factors for adverse events in analgesic drug users: results from the PAIN study. Pharmacoepidemiol Drug Saf 2003;12:601-10.

7. Towheed TE, Judd MJ, Hochberg MC, Wells G. Acetaminophen for osteoarthritis. Cochrane Database Syst Rev 2003;CD004257.

8. Hochberg MC. What a difference a year makes: reflections on the ACR recommendations for the medical management of osteoarthritis. Curr Rheumatol Rep 2001;3:473-8.

9. Singh G. Gastrointestinal complications of prescription and over-the-counter nonsteroidal anti-inflammatory drugs: a view from the ARAMIS database. Am J Therapeutics 2000;7:115-21.

Ophthalmologic Monitoring for Antimalarial Toxicity

To the Editor:

We read with interest the recent article by Bernatsky, et al on the adherence to ophthalmologic monitoring for antimalarial toxicity in their lupus cohort¹. As the authors demonstrate, adherence to the 1996 American College of Rheumatology (ACR) guidelines suggesting repeated ophthalmologic examinations at least yearly² was incomplete in their center. Notably, they observed that antimalarial exposure for more than 5 years was predictive of nonadherence. Although only 2 cases of hydroxychloroquine (HCQ)-induced retinopathy, after 15 and 33 years of treatment, were identified among their patients, both of whom had a history of exposure to chloroquine, the authors continue to advocate ophthalmology assessments at least yearly for their patients who are receiving antimalarial agents¹. Moreover, a recently published survey that examined rheumatologists' attitudes toward routine screening for HCQ retinopathy revealed that 94% of them screen their patients at least once a year because they are unwilling to accept any risk of visual damage³.

We recently published the results of a prospective study conducted between 1985 and 2000 in our center to define the risk of HCQ-related retinal toxicity in patients with rheumatoid arthritis and systemic lupus erythematosus who are receiving recommended doses of the drug (6.5 mg/kg/day)⁴. Of 400 patients using longterm HCQ at recommended doses, the incidence of retinal toxicity was very low (0.5%), while no case of retinopathy was observed during the first 6 years of treatment. These results allowed as to conclude that after a baseline examination to confirm the absence of preexisting fundus pathology, patients with normal renal function may receive HCQ at a maximal daily dose of 6.5 mg/kg and continue safely for 6 years. However, we suggested annual screening in patients who have taken the drug, even in recommended doses, for more than 6 years⁴. These results substantiate the guidelines developed in 2002 by the American Academy of Ophthalmology (AAO) Task Force for screening for HCQ toxicity⁵, stating that ophthalmologic examinations of such patients during the first 5 years of treatment can be at the frequency of regular examinations recommended by the AAO Preferred Practice Pattern⁶ for the age of the patient, providing that no concomitant retinal, renal, and/or liver disease exists.

To our knowledge, no documented case exists of HCQ-induced retinopathy in patients who were taking the drug in recommended doses for less than 6 years and had normal renal function. Moreover, among the millions of patients exposed to HCQ worldwide, there are only 5 appropriately-dosed patients who developed irreversible retinal changes within the first 10 years of treatment; all were diagnosed during the seventh or by the end of the eighth year of HCQ exposure^7-10. Seven additional patients who developed HCQ-induced retinopathy after taking the drug for more than 10 years at doses less than 6.5 mg/kg/day, in the absence of any associated comorbidities, have been reported to date⁴. Based on this evidence, and because frequent visual testing is not only expensive but burdensome to patients, as well as the findings by Bernatsky, et al claiming increased nonadherence especially after the first 5 years of treatment¹, it seems appropriate to revise the ACR clinical guidelines of ophthalmologic monitoring for antimalarial toxicity.

PETROS P. SFIKAKIS, MD, 1st Department of Propaedeutic and Internal Medicine, Laikon Hospital; MYRON MAVRIKAKIS, MD, Department of Clinical Therapeutics, Alexandra Hospital, Athens University Medical School, Athens, Greece.

REFERENCES

1. Bernatsky S, Pineau C, Joseph L, Clarke A. Adherence to ophthalmologic monitoring for antimalarial toxicity in a lupus cohort. J Rheumatol 2003;30:1756-60.

2. American College of Rheumatology Ad Hoc Committee on Clinical Guidelines. Guidelines for monitoring drug therapy in rheumatoid arthritis. Arthritis Rheum 1996;39:723-31.

3. Fraenkel L, Felson DT. Rheumatologists' attitudes toward routine screening for hydroxychloroquine retinopathy. J Rheumatol 2001;28:1218-21.

4. Mavrikakis I, Sfikakis PP, Mavrikakis E, et al. The incidence of irreversible retinal toxicity in patients treated with hydroxychloroquine: A reappraisal. Ophthalmology 2003;110:1321-6.

5. Marmor MF, Carr RE, Easterbrook M, et al. Recommendations on screening for chloroquine and hydroxychloroquine retinopathy: a report by the American Academy of Ophthalmology. Ophthalmology 2002;109:1377-82.

6. American Academy of Ophthalmology. Comprehensive adult medical eye evaluation, preferred practice pattern. San Francisco: American Academy of Ophthalmology; 2000.

7. Falcone PM, Paolini L, Lou PL. Hydroxychloroquine toxicity despite normal dose therapy. Ann Ophthalmol 1993;25:385-8.

8. Mavrikakis M, Papazoglou S, Sfikakis PP, et al. Retinal toxicity in long term hydroxychloroquine treatment. Ann Rheum Dis 1996;55:187-9.

9. Bienfang D, Coblyn JS, Liang MH, Corzillius M. Hydroxychloroquine retinopathy despite regular ophthalmologic evaluation: a consecutive series. J Rheumatol 2000;27:2703-6.

10. Warner AE. Early hydroxychloroquine macular toxicity. Arthritis Rheum 2001;44:1959-61.

Dr. Bernatsky, et al reply

To the Editor:

We welcome the thoughtful letter by Drs. Sfikakis and Mavrikakis. As they indicate, the literature appears to contain no documented case of definite hydroxychloroquine-induced retinopathy in patients taking the drug for less than 6 years (properly dosed, with normal renal function), although there are published cases of antimalarial toxicity as early as 6.5 years¹. We agree that it may very well be time for the American College of Rheumatology (ACR) to reconsider the published guidelines regarding ophthalmologic monitoring for antimalarial toxicity², perhaps along the lines of the American Academy of Ophthalmology Task Force³. The recent summary of these recommendations in a letter to the journal Arthritis and Rheumatism (the official publication of the ACR) suggests that there is some recognition of the importance of this issue⁴ by the ACR.

SASHA BERNATSKY, MD, MSc, FRCPC, Division of Clinical Epidemiology, Montreal General Hospital, Department of Epidemiology and Biostatistics, McGill University; CHRISTIAN PINEAU, MD, FRCPC, Department of Rheumatology, Montreal General Hospital, McGill University Health Centre; MARK GANS, MD, FRCPC, Department of Ophthalmology, Montreal General Hospital, McGill University Health Centre; ANN CLARKE, MD, MSc, FRCPC, Division of Clinical Epidemiology, Division of Clinical Immunology/Allergy, Montreal General Hospital, McGill University Health Centre, and Department of Epidemiology and Biostatistics, McGill University, Montreal, Quebec, Canada.

REFERENCES

1. Mavrikakis I, Sfikakis PP, Mavrikakis E, et al. The incidence of irreversible retinal toxicity in patients treated with hydroxychloroquine: A reappraisal. Ophthalmology 2003;110:1321-6.

2. American College of Rheumatology Ad Hoc Committee on Clinical Guidelines. Guidelines for monitoring drug therapy in rheumatoid arthritis. Arthritis Rheum 1996;39:723-31.

3. Marmor M, Carr R, Easterbrook M, Farjo A, Mieler W, American Academy of Ophthalmology. Recommendations on screening for chloroquine and hydroxychloroquine retinopathy: a report by the American Academy of Ophthalmology. Ophthalmology 2002;109:1377-82.

4. Marmor M. New American Academy of Ophthalmology recommendations on screening for hydroxychloroquine retinopathy [letter]. Arthritis Rheum 2003;48:1764.

Diagnostic Value of Anticyclic Citrullinated Peptide Antibody in Rheumatoid Arthritis

To the Editor:

We read with interest the report by Zeng, et al¹ on the diagnostic value of anticyclic citrullinated peptide antibody (anti-CCP) in patients with rheumatoid arthritis (RA). The role of anti-CCP remains to be determined for its value as a predictor of erosive disease, as a prognostic factor, and possibly as a diagnostic tool. Zeng, et al evaluated a modified ELISA as a diagnostic for detecting RA using anti-CCP antibodies in a Chinese population (n = 511 sera) consisting of 191 patients with RA (37%), 132 patients with other rheumatic diseases (systemic lupus, systemic sclerosis, primary Sjögren's syndrome, etc.), 98 patients with nonrheumatic diseases (postviral diseases, autoimmune hepatitis, etc.), and 90 healthy individuals: they calculated a 45% sensitivity and 97% specificity for anti-CCP as a diagnostic test for RA. But how should this be interpreted? We think a critical evaluation of a diagnostic test deserves some words on the influence of the Chinese population, the relative magnitude of non-RA subgroups, and test characteristics clinically more important than sensitivity and specificity, i.e., predictive value of a positive test result (PPV) to correctly diagnose RA with the anti-CCP, versus the predictive value of a negative test result (NPV) to correctly exclude RA with a negative anti-CCP test result.

In clinical practice a diagnostic test cannot merely be defined by calculating sensitivity and specificity. As clinicians we are not that much interested in ratios of index test results in RA versus non-RA, as described with sensitivity and specificity. For clinicians it is relevant to gain further insight into the merits of a diagnostic test; therefore one needs to calculate PPV, NPV, and likelihood ratios (LR: positive LR ideally reaching to infinity; negative LR ideally reaching to nil). We calculated these test characteristics for Zeng, et al (Table 1). For comparison we added test characteristics of data from a recent study investigating a Dutch patient population². Anti-CCP sensitivity ranges did not overlap between the Chinese and Dutch populations. This seems to be because the Chinese patients with RA more often lacked anti-CCP antibodies than the Dutch patients. Is this due to a high number of cases of early RA, is it inherent to the Chinese population with RA, or is this due to their ELISA? The low number of anti-CCP positive RA patients explains the discrepancy in sensitivities but also results in a lower NPV: the probability not to have RA a priori in the study by Zeng, et al was 63%, and only increased up to 76% after a negative anti-CCP test result. In contrast, in the Dutch population the a priori probability not to have RA was 68%, which increased up to 92% after a negative anti-CCP.

Table 1. Test characteristics. Data in parentheses are 95% confidence intervals.

Therefore one may conclude that as a diagnostic test anti-CCP in Chinese patients is performing less well than in Holland. In China, anti-CCP as a diagnostic test may be less suitable for rheumatologists.

TIM L.T.A. JANSEN, MD; GEORGE A.W. BRUYN, MD, Department of Rheumatology, Medical Centre Leeuwarden, Henry Dunantweg 2, 8934 AD Leeuwarden, The Netherlands.

REFERENCES

1. Zeng X, Ai M, Tian X, et al. Diagnostic value of anti-cyclic citrullinated peptide antibody in patients with rheumatoid arthritis. J Rheumatol 2003;30:1451-5.

2. van Venrooij WJ, van de Putte LBA. Vroegdiagnostiek van reumatoide artritis met een test op basis van een specifiek antigeen: cyclisch gecitrullineerd peptide [Early diagnosis of rheumatoid arthritis with a test based upon a specific antigen: cyclic citrullinated peptide]. Ned Tijdschr Geneeskd 2003;147:191-4.

Dr. Zeng replies

To the Editor:

We thank Dr. Jansen and Dr. Bruyn for their interest in our article. We agree that a diagnostic test cannot merely be evaluated by its sensitivity and specificity; the positive predictive value and negative predictive value are also very important indexes. However, we think that the value of a diagnostic test can also be interpreted by how it is used in the clinic. As we know that the sensitivity of both anti-Scl-70 antibodies and anti-Sm-antibodies is very low, but they are very useful in the diagnosis and management of systemic sclerosis and systemic lupus erythematosus because of their high specificity^1,2. Since the cost of the anti-CCP antibody test is very high in China, we tested it only in patients who were in the very early stage of disease or patients whose clinical manifestation was very likely to be RA, but did not fulfill the American College of Rheumatology (ACR) diagnostic criteria to confirm the diagnosis. Therefore, specificity is more important in our practice than sensitivity. Indeed, we diagnose RA based on the ACR criteria, not any single antibody.

As we stated, we had tested 511 serum samples, including 191 patients with definite RA, 132 with other rheumatic disease, 98 with nonrheumatic disease, and 90 healthy individuals (not included in the specificity and negative predictive values). The diagnosis of RA was well defined according to the ACR criteria. The results for sensitivity, specificity, positive predictive value, and negative predictive value were 47%, 97%, 94%, and 69%, respectively. In this study, we did the test using ELISA coated by ourselves. The low sensitivity is most probably due to our ELISA, as Jansen and Bruyn point out in their comments. After this study was completed, we carried out another study (not yet published) using many commercial anti-CCP kits including the new Immunoscan RA mark 2 kit (Euro-Diagnostica, Arnhem, The Netherlands). We tested anti-CCP in the same cohort of patients using Diastat anti-CCP kits (FCCP 200, Axis-Shield Diagnostics, Dundee, UK), and got a higher sensitivity (67% vs 47% as reported in our article) and high specificity (95%). However, the sensitivity is still not as high as that reported by van Venrooij and van de Putte³. It is known that the sensitivity of anti-CCP antibody varies from 40% to 80% in different laboratories depending on the cohort of patients studied and ELISA kits used^4–6. As we stated, further research is required to elucidate these differences, including geographic factors, between our results and those of others. We look forward to a multicenter collaboration with other laboratories.

Finally, we believe that, although there are some limitations in our study (low sensitivity and other statistical data), the anti-CCP antibody is still a useful tool in the diagnosis of RA in China.

XIAOFENG ZENG, MD, Professor of Medicine, Department of Rheumatology, Peking Union Medical College Hospital, Peking Union Medical College, Chinese Academy of Medical Science, Beijing, China.

REFERENCES

1. Tan EM, Smolen JS, McDougal JS, et al. A critical evaluation of enzyme immunoassays for detection of antinuclear autoantibodies of defined specificities. I. Precision, sensitivity, and specificity. Arthritis Rheum 1999;42:455-64.

2. Field M, Williams DG, Charles P, Maini RN. Specificity of anti-Sm antibodies by ELISA for systemic lupus erythematosus: increased sensitivity of detection using purified antigens. Ann Rheum Dis 1988;47:820-5.

3. van Venrooij WJ, van de Putte LBA. Early diagnosis of rheumatoid arthritis with a test based upon a specific antigen: cyclic citrullinated peptide. Ned Tijdschr Geneeskd 2003;147:191-4.

4. Schellekens GA, Visser H, de Jong BA, et al. The diagnostic properties of rheumatoid arthritis antibodies recognizing a cyclic citrullinated peptide. Arthritis Rheum 2000;43:155-63.

5. Van Boekel MA, Vossenaar ER, van den Hoogen FH, van Venrooij WJ. Autoantibody systems in rheumatoid arthritis: specificity, sensitivity and diagnostic value. Arthritis Res 2002;4:87-93.

6. Lee DM, Schur PH. Clinical utility of the anti-CCP assay in patients with rheumatoid disease. Ann Rheum Dis 2003;62:870-4.

Prevalence of Silent Amyloidosis in Rheumatoid Arthritis and Its Clinical Significance

To the Editor:

We read with interest the article by Wakhlu, et al¹ analyzing the prevalence of reactive amyloidosis in a series of Asian North Indian patients with rheumatoid arthritis (RA). In this systematic cross-sectional study (rather than prospective, as the authors assert) using abdominal subcutaneous fat aspiration (ASFA) as the screening method, a high prevalence of amyloid fat deposits was found in patients with RA (26.5%), most of whom (73%) had subclinical amyloidosis. The authors question the clinical significance of these amyloid deposits and whether they may predict future overt clinical visceral amyloid involvement.

A partial answer may be found in our followup study published 2 years ago². In this observational study in a large cohort of Spanish patients with adult RA, the prevalence of amyloid fat deposits was lower (16.3% at the time of the first ASFA test) than that observed by Wakhlu, et al. We also found that most patients had no clinical manifestations due to amyloidosis. After a mean followup period of nearly 7 years, 10 out of 35 patients with subclinical amyloidosis developed some type of renal disease, but this was due to amyloid nephropathy in only 5 patients; thus, amyloidosis remained subclinical in most patients throughout the followup, even in patients with followup longer than 10 years². We agree with Wakhlu and colleagues that a longer followup is required to elucidate the exact clinical significance of this finding, but our results seem to confirm that, in a considerable proportion of patients, these deposits should not be considered a sign of poor prognosis and a sure marker of future overt clinical amyloidosis. This suggests that no aggressive cytotoxic therapy should be initiated in patients with silent amyloidosis.

Our study found that patients with marked amyloid deposits are more prone to clinical amyloidosis than patients with moderate to mild deposits. Using the same semiquantitative method, Wakhlu, et al¹ seem to confirm our findings, even though the small number of patients studied makes it difficult to draw definite conclusions. However, we believe that patients with marked deposits (massive deposits in > 25% of tissue fragments) should be closely monitored for clinical symptoms or laboratory findings suggestive of amyloidosis in order to treat early when clinical amyloidosis develops.

It is now clear that subclinical amyloidosis is common in RA and other inflammatory rheumatic diseases such as ankylosing spondylitis³. The prevalence rates of secondary amyloidosis in systematic studies in RA patients in recent series using the ASFA test^1,2,4 or gastroduodenal biopsies⁵ to detect amyloid deposits range from 7% to 26%; most of these patients have subclinical amyloidosis. In contrast, the prevalence of clinical amyloidosis is rather lower; in a Spanish cohort study of a large series of RA patients with a mean disease duration of 10 years, which was specifically designed to search for comorbidities and extrarticular complications (the EMECAR study)^6,7, clinical amyloidosis was found in only 5 out of 788 registered patients (accumulated prevalence of 0.6%; 95% CI 0.1–1.2%).

A decline in the incidence of secondary amyloidosis in RA has been suggested⁸. We believe this may be true with respect to clinical amyloidosis, but recent studies do not support a low prevalence of subclinical amyloidosis in RA, since the prevalence rates in older studies are similar, or even lower^9,10. It is probable that in some patients these deposits are true silent deposits, but in others they only reflect preclinical status. However, taking into account our results and the discordance between the prevalence rates of clinical and subclinical amyloidosis, it may be concluded that most of these patients have true silent amyloidosis and they will never develop clinical visceral involvement. It is possible that more effective therapies for the control of inflammatory activity are mainly responsible for the fact that silent amyloidosis remains at this stage during the evolution of the disease, although other factors, including genetic ones, may also be involved¹¹. In the last few years, new antirheumatic drugs, including biological therapy, have been shown to be highly effective in controlling inflammatory activity and joint damage. Whether this will result in a lower incidence of subclinical or clinical amyloidosis in the near future remains to be seen.

RAIMON SANMARTÍ, MD; EMILI GÓMEZ-CASANOVAS, MD; MANUEL SOLÉ, MD; JUAN D. CAÑETE, MD, PhD; JORDI GRATACÓS, MD; LORETO CARMONA*, MD; ISIDORO GONZALEZ-ALVARO*, MD; JOSE MUÑOZ-GÓMEZ, MD, PhD, Rheumatology Department, Hospital Clinic de Barcelona, Villarroel 170, Barcelona 08036, Spain. E-mail: sanmarti@clinic.ub.es *On behalf of the EMECAR Study.

REFERENCES

1. Wakhlu A, Krisnani N, Hissaria P, Aggarwal A, Misra R. Prevalence of secondary amyloidosis in Asian North Indian patients with rheumatoid arthritis. J Rheumatol 2003;30:948-51.

2. Gómez-Casanovas E, Sanmartí R, Solé M, Cañete JD, Muñoz-Gómez J. The clinical significance of amyloid fat deposits in rheumatoid arthritis: a systematic long-term followup study using abdominal fat aspiration. Arthritis Rheum 2001;44:66-72.

3. Gratacós J, Orellana C, Sanmarti R, et al. Secondary amyloidosis in ankylosing spondylitis: a systematic survey of 137 patients using abdominal fat aspiration. J Rheumatol 1997;24:912-5.

4. Mansoury TME, Hazenberg BPC, Badawy SAE, et al. Screening for amyloid in subcutaneous fat tissue of Egyptian patients with rheumatoid arthritis: clinical and laboratory characteristics. Ann Rheum Dis 2002;61:42-7.

5. Kobayashi H, Tada S, Fuchigami T, et al. Secondary amyloidosis in patients with rheumatoid arthritis: diagnostic and prognostic value of gastroduodenal biopsy. Br J Rheumatol 1996;35:44-9.

6. Carmona L, Gónzalez-Alvaro, Balsa A, et al. Rheumatoid artritis in Spain: occurrence of extra-articular manifestations and estimates of disease severity. Ann Rheum Dis 2003;62:1-5.

7. Gónzalez-Alvaro I, Carmona L, Balsa A, et al. Patterns of disease modifing antirheumatic drug use in a Spanish cohort of patients with rheumatoid artritis. J Rheumatol 2003;30:697-704.

8. Hazenberg BPC, van Iijswijk MH. Where has secondary amyloidosis gone? Ann Rheum Dis 2000;59:577-9.

9. Stenkvist B, Westermark P, Wibell L. Simple method of diagnostic screening for amyloidosis. Ann Rheum Dis 1974;33:75-6.

10. Arakapis G, Tribe CR. Amyloidosis in rheumatoid arthritis investigated by means of rectal biopsy. Ann Rheum Dis 1963;22:256-62.

11. Moriguchi M, Terai C, Koseki Y, et al. Influence of genotypes at SAA1 and SAA2 loci on the development and the length of latent period of secondary AA-amyloidosis in patients with rheumatoid arthritis. Hum Genet 1999;105:360-6.

Dr. Misra, et al reply

To the Editor:

We are thankful to Dr. Sanmarti, et al for their critical analysis of our study¹. We have quoted their study² in our report. The issue is whether subclinical amyloidosis is a precursor of clinical disease in rheumatoid arthritis (RA) or not. We agree with their hypothesis, but there are few studies (including theirs) that have looked at patients systematically over a prolonged period of time. We have in mind more than 10–15 years. We have said categorically that any therapeutic intervention such as cytotoxic agents like azathioprine or cyclophosphamide would be unwarranted for subclinical amyloidosis in the light of current findings. Even though their suggestion that massive amyloid deposits, > 25% of tissue fragments, are strongly associated with development of clinical disease, it will still be premature to advocate cytotoxic treatment in these cases. A close followup is perhaps necessary. We reiterate what has been published by the authors themselves, and this has been substantiated by our study in another ethnic group of patients.

RAMNATH MISRA, MD; ANUPAM WAKHLU, DM; NARENDRA KRISHNANI, MD; PRAVEEN HISSARIA, DM; AMITA AGGARWAL, DM, Department of Immunology and Pathology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, Raebareli Road, Lucknow 226 014, India.

REFERENCES

1. Wakhlu A, Krisnani N, Hissaria P, Aggarwal A, Misra R. Prevalence of secondary amyloidosis in Asian North Indian patients with rheumatoid arthritis. J Rheumatol 2003;30:948-51.

2. Gómez-Casanovas E, Sanmartí R, Solé M, Cañete JD, Muñoz-Gómez J. The clinical significance of amyloid fat deposits in rheumatoid arthritis: a systematic long-term followup study using abdominal fat aspiration. Arthritis Rheum 2001;44:66-72.

Macrophage Activation Syndrome Due to Methotrexate in a 12-year-old Boy with Dermatomyositis

To the Editor:

We describe a 12-year-old boy with dermatomyositis (DM) who had a reaction to methotrexate (MTX) while under treatment in hospital.

His clinical signs included high temperature, hepatomegaly and splenomegaly, and neurological changes (including expression abnormalities, coprolalia, insomnia, and agitation), with changes observed on brain magnetic resonance imaging studies. He also had melena, severe pancytopenia, abnormalities in prothrombin time, and a decrease in erythrocyte sedimentation rate. He had never experienced these symptoms prior to the use of MTX. All these signs and symptoms became worse after the second dose of MTX, and improved after discontinuation of MTX and the initiation of therapy with additional steroids and cyclosporine, which had been used continuously for his primary disease. He recovered with no further crisis or worsening of his DM.

Eraso, et al have suggested that MTX could not be responsible for the development of this syndrome¹, but in our patient we observed a direct relation between the onset of symptoms and the use of MTX. A diagnosis of macrophage activation syndrome was made according to the published criteria², and there was no other evidence of a factor that could have triggered this syndrome. We concluded this patient had macrophage activation syndrome secondary to the treatment with MTX, although we could not determine ferritin concentrations. As Ravelli and colleagues recommend³, we also ruled out other trigger factors for these changes in our patient.

GARY STERBA, MD, Hospital de Clinicas Caracas, avda Panteon con Alameda, consultorio 317, San Bernardino, Caracas, Venezuela 1011; CATHERINE RODRIGUEZ, MD, Pediatric Resident; SONIA SIFONTES, MD; PATRICIA VIGILANZA, MD, Hospital Municipal de Niños J.M. de los Rios, Caracas, Venezuela.

REFERENCES

1. Eraso R, Gedalia A, Espinoza LR. Methotrexate as a possible trigger of macrophage activation syndrome in systemic juvenile idiopathic arthritis [letter]. J Rheumatol 2002;29:1104-5.

2. Imashuku S. Differential diagnosis of hemophagocytic syndrome: underlying syndromes and selection of the most effective treatment. Int J Hematol 1997;66:135-51.

3. Ravelli A, Caria MC, Buratti S, Malattia C, Temporini F, Martini A. Methotrexate as a possible trigger of macrophage activation syndrome in systemic juvenile idiopathic arthritis [case report]. J Rheumatol 2001;28:865-7.

Drs. Ravelli and Martini reply

To the Editor:

Dr. Sterba and colleagues describe a patient with juvenile dermatomyositis (JDM) who developed a macrophage activation syndrome (MAS) after receiving methotrexate (MTX), in the absence of other potentially eliciting factors.

MAS occurs most frequently in children with systemic juvenile idiopathic arthritis (JIA), but has been observed in other subtypes of JIA and, more rarely, in juvenile systemic lupus erythematosus¹. In the latter disease, however, it has been suggested that this complication may be more common than previously realized². To our knowledge, MAS has not been previously described in JDM, with the exception of a case with platelet-specific hemophagocytosis³.

That MAS in Dr. Sterba's patient was apparently triggered by MTX is in keeping with our observations that this drug may act as an inciting factor of this syndrome in children with systemic JIA^4,5. Because several medications, including nonsteroidal antiinflammatory drugs, sulfasalazine, and gold salts, have been incriminated as possible triggers of MAS¹, it is not surprising that MTX could elicit this syndrome under certain circumstances.

ANGELO RAVELLI, MD; ALBERTO MARTINI, MD, Dipartimento di Pediatria, Università di Genova, Pediatria II, Istituto di Ricovero e Cura a Carattere Scientifico G. Gaslini, Genova, Italy.

REFERENCES

1. Ravelli A. Macrophage activation syndrome. Curr Opin Rheumatol 2002;14:548-52.

2. Morales PM, Jimenez BFJ, Yanes P, Rios D, Godinez R. Bone marrow with reactive histiocytosis, hemophagocytosis and storage histiocytes in systemic lupus erythematosus [abstract]. Arthritis Rheum 1992;35 Suppl:S239.

3. Kobayashi I, Yamada M, Kawamura N, Kobayashi R, Okano M, Kobayashi K. Platelet-specific hemophagocytosis in a patient with juvenile dermatomyositis. Acta Paediatr 2000;89:617-9.

4. Ravelli A, De Benedetti F, Viola S, Martini A. Macrophage activation syndrome in systemic juvenile rheumatoid arthritis successfully treated with cyclosporine. J Pediatr 1996;128:275-8.

5. Ravelli A, Caria MC, Buratti S, Malattia C, Temporini F, Martini A. Methotrexate as a possible trigger of macrophage activation syndrome in systemic juvenile idiopathic arthritis [case report]. J Rheumatol 2001;28:865-7.

Breast Implant Related Disease

To the Editor:

The recent article by Vermeulen and Scholte¹ stimulated us to examine our data. The Arthritis Center of Northeast Ohio [as regional expert witness for the presence of symptoms, signs, and diagnoses fulfilling previously established (36) criteria] examined 183 women with silicone-filled breast implants. This included 56 individuals with ruptured/leaking implants and 127 in whom the implants appeared intact.

This analysis confirmed Vermeulen and Scholte's observation¹ of increased sleep disturbance, headache, and memory loss and additionally identified increased sweating among individuals with ruptured implants (Table 1). However, neck pain, photosensitivity, dry eyes (subjective complaint, but not dryness as measured by Schirmer test), and constipation were less common in that group (Table 1). Fatigue, malar rash, alopecia, mucosal ulcers, dysphagia and diarrhea, and associated rheumatologic disorders were equally represented in both groups.

Table 1. Differential expression of clinical symptoms with breast implant leakage.

This variation illustrates the need for large-scale, epidemiologically-designed studies (correcting for reporting/participation biases) to further explore these intriguing issues.

BRUCE ROTHSCHILD, MD; MARK HELBLING II, Arthritis Center of Northeast Ohio, 5500 Market Street, Youngstown, Ohio 44512, USA

REFERENCE

1. Vermeulen RCW, Scholte HR. Rupture of silicone gel breast implants and symptoms of pain and fatigue. J Rheumatol 2003;30:2263-7.

Drs. Vermeulen and Scholte reply

To the Editor:

The survey of the symptoms occurring in patients with silicone breast implants by Rothschild and Helbling underlines the outcome of our study¹ with respect to the increased expression of several symptoms after rupture of the implants, such as sleep disturbance, headache, and memory loss. Interestingly, they also found symptoms that decreased after rupture. These and other differences between the outcome of the studies are associated with differences in the selection of diagnostic questions and of the patients. Rupture had occurred in only 31% of their patients, contrasting with 74% in ours. But details with regard to patient selection were not given.

The main tenor of our message was that patients who underwent a second operation to replace or remove silicone breast implant(s), complained of signs and symptoms of chronic fatigue syndrome², and that the symptoms were more severe in the ruptured group. The pattern of the complaints was different from patients with chronic fatigue syndrome, as relatively more implanted patients complained of multijoint pain and myalgia, and fewer of impaired short-term memory or concentration, sore throat, unrefreshing sleep, and headache of a new type.

The data from Rothschild and Helbling may add to our knowledge of the silicone exposure disease. Their patients likely represent an earlier stage of the disease than ours, as the chance of the occurrence of rupture increases with time³.

RUUD C.W. VERMEULEN, MD, Chronic Pain and Fatigue Research Centre, Waalstraat 25-31, 1078 BR Amsterdam; HANS R. SCHOLTE, PhD, Department of Biochemistry, Erasmus MC, University Medical Centre, Rotterdam, The Netherlands. E-mail rcwvermeulen@cfscentrumamsterdam.nl

REFERENCES

1. Vermeulen RCW, Scholte HR. Rupture of silicone gel breast implants and symptoms of pain and fatigue. J Rheumatol 2003;30:2263-7.

2. Fukuda K, Straus SE, Hickie I, Sharpe MC, Dobbins JG, Komaroff A. The chronic fatigue syndrome: a comprehensive approach to its definition and study. International Chronic Fatigue Syndrome Study Group. Ann Intern Med 1994;121:953-9.

3. Melmed EP. A review of explantation in 240 symptomatic women: a description of explantation and capsulectomy with reconstruction using a periareolar technique. Plast Reconstr Surg 1998;101:1364-73.