During a 3-year period, 10 cases of PCP in transplant recipients occurred in a building of the Edouard-Herriot Hospital in Lyon, whereas only one case was observed in the preceding 7 years. These cases could not be attributed to improved diagnosis, change of immunosuppression regimen, an increase (a decrease actually occurred) of transplant recipients hospitalized in the facility. However, the outbreak occurred concomitantly with a progressive increase in the number of HIV-infected patients with and without PCP hospitalized in the same facility. Thorough molecular and epidemiologic analyses of the PCP cases showed the following facts: 1) transplant recipients, often in a stage of severe immunosuppression, shared the facility with HIV-infected patients with and without active PCP; 2) both transplant recipients and HIV-infected patients were receiving no or suboptimal anti-PCP prophylaxis; 3) P. jirovecii type 1 represented 70% of the isolates from the transplant PCP cases, but it represented <31% of the isolates in the HIV-infected patients with PCP in Lyon and elsewhere; and 4) all the transplant recipients (and some HIV patients) in whom PCP developed had been hospitalized in the facility at some point during the 3 months preceding their PCP episodes. Moreover, the proportion of patients exposed to P. jirovecii type 1 during their susceptible period was higher among cases infected with type 1 than among those not harboring type 1, although it did not reach statistical significance (p = 0.05). Finally, review of the medical charts indicated potential encounters between PCP cases during their susceptible period and other patients with active PCP (). Encounters with patients with a PCP episode involving the same P. jirovecii type, which may have led to transmission, were possible in 5 of the 10 transplant recipients and 1 of the 30 HIV patients. Taken together, these facts suggest that at least half of the PCP cases in transplant recipients (and possibly some in HIV patients) may be the result of a nosocomial acquisition of P. jirovecii.
However, alternative explanations exist that cannot be excluded. First, the transient presence of P. jirovecii
in the air of hospital corridors has been described (36
), which raises the possibility of an environmental source of P. jirovecii
type 1 in building A. The following facts argue against this possibility: 1) the existence of a long-lasting environmental source of P. jirovecii
has never been established, and 2) a high prevalence of type 1 was not observed in building B, which is located 100 m from building A and hosts patients with hematologic malignancies. Second, our study provides epidemiologic and molecular evidence that nosocomial transmission of P. jirovecii
can occur, but whether this transmission would have occurred directly or indirectly through carriers is unclear. Indeed, carriage of P. jirovecii
DNA has been described in the lungs of asymptomatic, immunosuppressed persons (5–8
), as well as in the nose of immunocompetent relatives and healthcare workers in close contact with a PCP patient (37
). Moreover, transmission by immunocompetent carriers to susceptible hosts has been demonstrated in the mouse model (38
). Thus, indirect transmission through healthcare workers, physicians, or asymptomatic immunosuppressed patients cannot be ruled out.
For 14 of the 39 patients with PCP observed during the 3-year period, potential encounters with patients with active PCP during the 3 months preceding their episode had been documented (). However, only P. jirovecii
type 1 was involved in encounters that apparently resulted in secondary cases. Part of this observation may be related to the higher prevalence of type 1 in Lyon (≈20%), although this could not explain the 40% rate of type 1 in building A. Another possibility is that type 1 might be more transmissible or virulent. This finding would be consistent with the fact that this type was one of the most prevalent types also in other geographic areas () (27
). Moreover, specific P. jirovecii
genotypes have been associated with more severe clinical symptoms (39
) or with resistance to certain drugs (40
). In our study, a mutation in the active site of DHPS was present in all six presumptive nosocomial PCP cases. The mutation may have favored acquisition of type 1 rather than another type by the three patients who were receiving suboptimal prophylaxis with Fansidar (Roche, Nutley, NJ). The presence of this mutation in the nosocomial PCP cases of our study suggests that P. jirovecii
was acquired shortly before the episode because the frequency of DHPS mutations greatly increased only in the 1990s (41
). Moreover, in patients not receiving any sulfa prophylaxis, the frequency of M2 mutation was significantly higher in building A than in other hospital facilities of Lyon (p = 0.0004), a fact that suggests nosocomial interhuman transmission of P. jirovecii
Our study provides insight into the relative importance of nosocomial acquisition of P. jirovecii
if infectious and susceptible patients are in close contact. Even though infected and susceptible patients were kept in unusually close proximity in this hospital, relatively few cases compatible with nosocomial interhuman transmission seem to have occurred. This finding suggests that transmission from patients with active PCP is limited, which is consistent with studies that we performed in HIV outpatient clinics that suggested infrequent cross-infections (27,42
), as well as with a study comparing contact histories of patients with or without PCP (43
). The source remains undetermined for the infection in the five transplant recipients for whom no potentially infectious encounters were found. One possibility is that carriers of P. jirovecii
in the hospital have played a role.
The available data and the arbitrary definitions we had to use, in light of the absence of precise scientific data on P. jirovecii infection, are limitations of our study. We could not demonstrate that the presumed encounters actually occurred or define the precise nature of the encounter. Also, we could not firmly exclude other potential sources of P. jirovecii, such as the environment or asymptomatic carriers. Nevertheless, to our knowledge, this study is the first to suggest that P. jirovecii may be nosocomially transmitted and acquired by severely immunosuppressed patients. Given the increased number of reports relating resistance to anti-Pneumocystis drugs, prophylaxis of patients at risk might not be sufficient to achieve prevention. Moreover, prophylaxis is often not satisfactory because of secondary effects. Consequently, avoiding contact between persons at risk for PCP and patients with active PCP may be warranted and should be added to prevention guidelines.