Correspondence
Association between Prior Cytomegalovirus Infection and the Risk of Restenosis after Coronary Atherectomy
N Engl J Med 1997; 336:587-588February 20, 1997
- Article
To the Editor:
The association between seropositivity for cytomegalovirus (CMV) and restenosis after coronary atherectomy reported by Zhou et al. (Aug. 29 issue)1 is intriguing, but their analysis leaves open several questions. The authors included four patients in the CMV-positive group who, though initially seropositive, were seronegative at the six-month follow-up visit. Three of these patients had initial CMV titers just above the cutoff for seropositivity, whereas one had a midrange titer; all the titers dropped to nearly zero on follow-up. This raises the question of whether these patients in fact had CMV infection.2 Since they all had restenosis, including them in the CMV-positive category has substantial bearing on the results. Had the infections been classified on the basis of the results of serologic testing at six months rather than before atherectomy, our calculations suggest that CMV would not have been identified as a risk factor for restenosis (odds ratio, 2.4; 95 percent confidence interval, 0.7 to 8.3). Given the uncertainty of these patients' CMV status, it would seem prudent at a minimum to omit them from the analysis.
Another problem we perceived is the use of the CMV titer as a continuous variable. Although it would be interesting to correlate titers with outcomes, only patients considered to be seropositive should be studied. Seronegativity is used to indicate the absence of prior exposure to CMV; therefore, no variation in a negative titer is meaningful. Moreover, since the distribution of titers was markedly skewed, some method of normalizing the data must be used before the parametric methods described can be employed. As presented, the use of continuous titers gives no more information than the dichotomous analysis.
Finally, we have reservations about the analyses that were conducted “by vessel” rather than “by patient.” The coronary vessels of an individual patient are not independent of one another. Not only might events in one part of the heart influence events elsewhere, but the inclusion of more than one vessel from a given patient may give undue weight to unrecognized confounding factors. Other studies have shown that coronary arteries differ in their risk of restenosis, and patients with multivessel disease may also be at increased risk.3,4 A more complete description of the vessels studied, including an indication of whether the patient had multivessel disease, would make it easier to understand what effect, if any, CMV had on the risk of restenosis.
4 ReferencesKaren Smith, M.D., M.P.H.
Julie Parsonnet, M.D.
Stanford University School of Medicine, Stanford, CA 943051
Zhou YF ,Leon MB ,Waclawiw MA , et al. Association between prior cytomegalovirus infection and the risk of restenosis after coronary atherectomy. N Engl J Med 1996;335:624-630
Full Text | Web of Science | Medline2
Landini MP ,Mach M . Searching for antibodies specific for human cytomegalovirus: is it diagnostically useful? When and how. Scand J Infect Dis Suppl 1995;99:18-23
Medline3
Fishman RF ,Kuntz RE ,Carrozza JP Jr , et al. Long-term results of directional coronary atherectomy: predictors of restenosis. J Am Coll Cardiol 1992;20:1101-1110
CrossRef | Web of Science | Medline4
Giuliani ER, Gersch BJ, McGoon MD, Hayes DL, Schaff HV, eds. Mayo Clinic practice of cardiology. 3rd ed. St. Louis: Mosby–Year Book, 1996.
To the Editor:
Zhou and colleagues report that prior CMV infection, as assessed by titers of anti-CMV IgG antibodies, is an independent risk factor for the development of restenosis six months after directional coronary atherectomy. The authors observed a 43 percent rate of restenosis in the seropositive patients, as compared with an 8 percent rate in the seronegative patients. These findings would appear to strengthen the earlier demonstration by the same group1 of CMV DNA sequences in atherectomy samples with restenosis and the production of IE84, one of the virus's immediate early proteins, by smooth-muscle cells grown in vitro from such lesions.
Since titers of anti-CMV IgG antibodies did not change during the study period, the authors affirm that their results “are most compatible with the idea that either the virus produced an abortive infection . . . or viral replication occurred locally in the absence of systemic viremia.” 2 It is conceivable, therefore, that the local detection of CMV messenger RNA (mRNA) would be needed to prove definitively that active CMV infection has a pathogenetic role in the development of restenosis and ischemic syndromes. In this respect, we recently reported3 that of a group of atherectomy specimens from 40 patients with stable or unstable angina, none of the 33 patients with primary lesions, and not even the 7 with restenosis, were positive for CMV IE84 mRNA. Although the small samples did not allow us to study the presence of both DNA and RNA or measure anti-CMV IgG antibodies in our patients, our findings suggest that the authors' hypothesis, that an active local CMV infection may have a pathogenetic role, should be considered with caution.
3 ReferencesAmir Kol, M.D.
Giovanni Sperti, M.D.
Attilio Maseri, M.D.
Catholic University of the Sacred Heart, Rome 00168, Italy1
Speir E ,Modali R ,Huang E-S , et al. Potential role of human cytomegalovirus and p53 interaction in coronary restenosis. Science 1994;265:391-394
CrossRef | Web of Science | Medline2
Zhou YF ,Leon MB ,Waclawiw MA , et al. Association between prior cytomegalovirus infection and the risk of restenosis after coronary atherectomy. N Engl J Med 1996;335:624-630
Full Text | Web of Science | Medline3
Kol A ,Sperti G ,Shani J , et al. Cytomegalovirus replication is not a cause of instability in unstable angina. Circulation 1995;91:1910-1913
Web of Science | Medline
Author/Editor Response
The authors reply:
To the Editor: Dr. Kol and associates are correct. Our data do not allow us to conclude that if CMV contributes to restenosis, it necessarily operates through a local effect of the virus on the vessel wall. However, their data clearly do not prove the converse, for several reasons. First, they studied only seven patients with restenosis, too few to permit definitive conclusions. Second, the virus may have expressed CMV mRNA at the time of angioplasty but not several months later, when the tissue was retrieved for analysis. Finally, atherectomy retrieves only part of the lesion, and therefore there may be false negative results. We believe that anti-CMV antibodies are a more sensitive measure of the presence of CMV than atherectomy-based sampling of lesions.
Drs. Smith and Parsonnet believe that either the four patients who were seropositive at entry into the study but seronegative at the six-month follow-up should be excluded from the analysis or the cohort should be analyzed on the basis of serologic tests at six months. Excluding the four patients is unwarranted, because the prospectively determined end point of the study was the correlation between restenosis and the base-line CMV-antibody status. Therefore, the analysis was performed without bias due to knowledge of the restenosis status. To arbitrarily omit four patients from the analysis because of an end point that was not identified prospectively would not be scientifically valid. Furthermore, if CMV does contribute to restenosis, we would expect that CMV status at the time of atherectomy would be most relevant to any CMV-induced effect on restenosis.
We analyzed the CMV titer as a continuous variable in order to deal directly with the previous point Smith and Parsonnet raise. Although we defined seropositivity prospectively, any such definition is admittedly arbitrary. We therefore wanted to ascertain whether there was a correlation between the CMV titer as a continuous variable and restenosis. This was in fact the case. It would have been inappropriate to focus only on the seropositive patients.
Smith and Parsonnet question our analyses conducted “by vessel” rather than “by patient.” It is difficult to determine which approach is better. As a result, we analyzed our data both ways. Thus, 43 percent of patients who were CMV-seropositive had restenosis, as compared with 8 percent of patients who were seronegative. In addition, the analyses by vessel, a type of assessment commonly used by interventional cardiologists, showed that there was a significant correlation between the loss index and seropositivity. The point is moot, however, because of the 75 patients studied, only 10 had two-vessel angioplasty; the rest had angioplasty of a single vessel.
We would emphasize that although our article shows a strong association between prior CMV infection (as indicated by CMV seropositivity) and restenosis, our findings must be confirmed by further studies before a definite relation between CMV and restenosis can be accepted as proved.
Yi Fu Zhou, M.D.
Myron A. Waclawiw, Ph.D.
Stephen E. Epstein, M.D.
National Heart, Lung, and Blood Institute, Bethesda, MD 20892-1650
