We were unable to evidence any correlation between viral shedding and CMV serology; this may be due to the lack of sensitivity or specificity of the complement-fixation test or to a lack of serologic response from critically ill patients. Conversely, we observed a significantly higher rate of viral infection in female patients. While it is not clear why female patients were more susceptible to CMV infection, this observation agrees with the higher rate of seroconversion to CMV reported by Wilhelm in female patients receiving blood transfusion but not undergoing cardiac surgery or kidney transplantation. We also observed a trend towards an increase of CMV infection during the year of 1985 (Fig 1), while laboratory techniques and staff were unchanged. Concomitantly, the HIV serologic status became available for blood donors and screening was made mandatory by order of the French Health Department in July 1985. We postulate that a subpopulation of hidden high-risk donors gave blood in an attempt to obtain the results of their own HIV serologic status. This point is substantiated by a recent official report where a tenfold higher proportion of HIV + status among new donors was found, compared with regular donors during the year of 1986 (0.2 percent vs 0.02 percent, respectively).
The pathogenicity of CMV in the normal adult host is extremely variable. In our study, while anemia and leukocytosis were present in both viral and non- viral excretors, we noted a mild trend towards lymphocytosis in the former group (Table 2), without high absolute counts. This fact may be explained by the lymphopenia regularly associated with severe bacterial sepsis. A definite mononucleosis-like reaction on blood smears was seen for one third of the VE + patients, consistent with earlier studies. Jaundice was not more common in VE + patients, and although over 80 percent of the patients with CMV infection had elevated enzymes, severe cytolysis was uncommon. By contrast, a rather unexpected finding was the frequent association of acute renal failure with CMV infection. Glomerular and interstitial injuries, which have been described in CMV infection, may contribute to other well-known mechanisms such as hemodynamic disturbances, sepsis, and toxicity of antimicrobial agents, to produce the observed high incidence of acute renal insufficiency.
A noteworthy finding in our study was the impact of CMV excretion on the severity of bacterial infection, especially after the 15th day of sepsis (36 ±11 days after cardiac surgery). The higher late mortality and the longer duration of hospitalization seen among VE+ survivors support this finding. We likewise observed that failure of initial treatment was significantly more frequent for viral excretors. Although it could be argued that among the patients with mediastinitis, those with higher morbidity and more prolonged hospitalization would have been recognized more frequently as VE+ because of more frequent cultures, certain findings suggest that this was not a bias. First, the mean delay observed between ECC and the first positive culture in VE + patients was similar to the mean delay observed between ECC and the last negative culture in VE— patients; and, secondly, the mean number of viral cultures performed until the first positive culture in VE + patients was 1.72 ±0.70 cultures per patient vs a total number of 1.81 ±0.74 cultures per patient in the VE — group (NS).
In conclusion, we believe that CMV infection enhances the morbidity of the infectious process in the setting of severe mediastinitis following cardiac surgery in adult patients. The therapeutic significance of this finding should be evaluated by a placebo-controlled trial of antiviral agents, provided nontoxic drugs become available.