Portal de Periódicos da CAPES

Letters to the editor

2000; Elsevier BV; Volume: 74; Issue: 3 Linguagem: Inglês

10.1016/s0015-0282(00)00674-9

1556-5653

Mark D. Hornstein, Owen Davis, Joe B. Massey, Richard J. Paulson, John A. Collins,

Abdominal Trauma and Injuries

We thank Sher et al. (1Hornstein M.D. Davis O.K. Massey J.B. Paulsen R.J. Collins J.A. Antiphospholipid antibodies and in vitro fertilization success.Fertil Steril. 2000; 73: 330-333Abstract Full Text Full Text PDF PubMed Scopus (113) Google Scholar) for their interest in our publication “Antiphospholipid antibodies and in vitro fertilization: a meta-analysis.” We anticipated their letter. Before commenting on their specific points, we believe a brief review of the origins of the study would be helpful to the readership. The Society for Assisted Reproductive Technologies (SART) and the American Society for Reproductive Medicine (ASRM) asked the members of the SART Practice Committee to review the body of literature regarding antiphospholipid antibodies (APAs) and IVF outcome because of the controversies surrounding whether the proposed therapies were indicated or effective. It was hoped that an opinion from the practice committee would be helpful to clinicians in this controversial area. Our goal in undertaking this analysis was simply to provide the membership of the ASRM with a clear summary of the data to aid decisions in their clinical practices. Sher et al. (1Hornstein M.D. Davis O.K. Massey J.B. Paulsen R.J. Collins J.A. Antiphospholipid antibodies and in vitro fertilization success.Fertil Steril. 2000; 73: 330-333Abstract Full Text Full Text PDF PubMed Scopus (113) Google Scholar) make four points about our study. First, they state that we disregarded several publications that would have supported an alternative hypothesis. Specifically, they cite two studies from Dr. Sher’s group (2Sher G. Zouves C. Feinman M. Maasarani G. Matzner W. Chong P. et al.A rational basis for the use of combined heparin/aspirin and IVIG immunotherapy in the treatment of recurrent IVF failure associated with antiphospholipid antibodies.Am J Reprod Immunol. 1998; 39: 391-394Crossref PubMed Scopus (68) Google Scholar, 3Sher G. Matzner W. Feinman M. Maassarani G. Zouves C. Chong P. et al.The selective use of heparin/aspirin therapy, alone or in combination with intravenous immunoglobulin G, in the management of antiphospholipid antibody-positive women undergoing in vitro fertilization.Am J Reprod Immunol. 1998; 40: 74-82Crossref PubMed Scopus (84) Google Scholar). Both of these articles consider the use of heparin, aspirin, and iv immunoglobulin therapy in antiphospholipid antibody patients undergoing IVF. The studies were treatment studies. The purpose of our meta-analysis was to examine the association between APAs and IVF success; therefore, it excluded treatment that might have had a beneficial or harmful effect on that outcome. We included all studies we could find that met the criteria outlined in “Materials and Methods” of our article, including an earlier study from Dr. Sher’s group that did include untreated patients with and without APAs who were undergoing IVF. Second, Sher et al. criticized our inclusion of heterogeneous studies in the analysis. Of course, the purpose of a meta-analysis is to combine the results of several studies to reach a more precise estimate of an association or effect than would be possible from any single study. In any group of humans there is some degree of heterogeneity, and it follows that some heterogeneity will exist among study populations. Nevertheless, we found that the heterogeneity among the studies was not statistically significant, which is confirmed by the seven assessments ran by Sher et al. on the same studies. Clinical heterogeneity also seems unlikely, because the study designs were markedly similar, and the individual study odds ratios conformed to a normal distribution around 1.0 or no difference. It is of interest that in the table of APA-negative and APA-positive IVF pregnancy rates presented by Sher et al., the differences also are approximately normally distributed, in this case around zero difference. In addition, under this point about heterogeneity, Sher and colleagues suspect that we used log transformation of the data to reduce variance. With dichotomous outcomes (in this case pregnant, not pregnant) the usual meta-analytic method is a weighted average of the log odds ratio with weights inversely proportional to the variance. The log of the odds ratio or relative risk is used because that is the form in which these effect measures are normally distributed, a prerequisite for analysis, which is also the goal with log transformations of continuous data. In meta-analysis the weights are set inversely proportional to the variance so that larger studies offering more precise estimates of the risk will be given greater importance in the analyses. Furthermore, on the issue of clinical heterogeneity, Dr. Sher and his colleagues state that APA measurements are useful in some centers but not in others. This has been the argument used for some time by proponents of APA testing for infertility. It is difficult to imagine how a test that varies widely in its usefulness from center to center could have universal applicability. Diagnostic tests must have stable properties to have predictable outcomes when applied in the infinite variety of circumstances that comprise normal clinical practice. Third, Dr. Sher and his group note that different APA-positive definitions were used by the authors included in our study. We accepted the peer-reviewed definition of APA positivity in each article. Our report notes that studies evaluating more antibody types were more likely to find a nonsignificant reduction in IVF success for APA-positive women. Despite evidence from >2,000 patients, it cannot be said whether this nonsignificant finding is the tip of a true association iceberg or simply the tendency to find abnormal results by doing more tests. Fourth, Sher et al. object to the inclusion of different methodologies and patient populations, specifically the large study of Denis et al., citing methodological flaws in their APA assays (4Denis A.L. Guido M. Adler Rd Bergh P.A. Brenner C. Scott Jr, R.T. Antiphospholipid antibodies and pregnancy rates and outcome in in vitro fertilization patients.Fertil Steril. 1997; 67: 1084-1090Abstract Full Text PDF PubMed Scopus (78) Google Scholar). Denis et al. have responded to the criticisms about the methodology of APA assays in their laboratory in previous correspondence (5McDonough P.G. Clinical utility of antiphospholipid antibodies? A negative study with power! (Letter to the editor).Fertil Steril. 1998; 69: 164-167Abstract Full Text PDF Scopus (2) Google Scholar). They pointed out that there are definable differences in the methodology between various reference laboratories performing APA assays and that their methodology was sound. They also mention that the prevalence of antiphospholipid antibodies in their study (26%) is similar to the prevalence in other studies, suggesting that the results in their system are similar to those obtained in other laboratories. For the assertion that the significance of APAs in women undergoing IVF should be confined to cases of non-male factor infertility, Denis et al. sensibly observed that there is no reason to believe an embryo derived from a male partner with reduced semen parameters would be immune to the hypothetical adverse effect of APAs. In summary, the concerns of Sher et al. do not undermine the validity of the analyses reported in our study. Since their letter was received, a further study involving 308 women undergoing IVF also has reported that pregnancy rate, live birth rate, gestational age at delivery, and birth weight were not affected by APA status (6Chilcott I.T. Margara R. Cohen H. Raj R. Skull J. Pickering W. et al.Pregnancy outcome is not affected by antiphospholipid antibody status in women referred for in vitro fertilization.Fertil Steril. 2000; 73: 526-530Abstract Full Text Full Text PDF PubMed Scopus (52) Google Scholar). Its authors concluded that screening women undergoing IVF and APA was not justified. The specific findings of our article and that of Chilcott et al. are not, however, a sweeping dismissal of the entire field of clinical implantation immunology. We hope that Dr. Sher and his colleagues will continue their respected work in this field.