This retrospective study demonstrates that the incidence of PCP requiring ICU admission has increased in HIV-negative patients at our institution during the period of 1993 to 2006. As compared with HIV-positive cases, non-HIV patients had a worse course of the disease in the ICU. ICU mortality was higher in HIV-negative than in HIV-positive patients. Importantly, first-line NIMV failed in a very large proportion of HIV-negative patients.
HIV-negative status, which is known to be associated with an increased mortality during PCP compared with HIV-positive status [3
], maintained this grim prognostic value for the critical forms of the disease. This difference in mortality might be related to the underlying condition rather than to the HIV-negative status per se
. Not only the mortality but also the proportion of ventilated days spent with high levels of positive end-expiratory pressure and FiO2
were higher in HIV-negative compared with HIV-positive patients. The higher neutrophil count observed in the bronchoalveolar lavage of HIV-negative patients suggests that the PCP-related lung injury was more severe in HIV-negative subjects. Even though we could not assess whether baseline differences in age and chronic lung condition influenced this finding, it suggested that the lung injury was different between HIV-positive and HIV-negative patients. This had important implications in terms of ventilatory support modalities.
Indeed, one of the most striking results concerned the description of the ventilatory support depending on the HIV status, a comparison that has not been performed to date. NIMV was chosen as first-line therapy in a similar proportion of HIV-negative and HIV-positive patients. However, in HIV-negative patients, NIMV failed in 71% of cases compared with failure in 13% of HIV-positive patients, suggesting that the severity of PCP-related lung injury was tremendously higher in HIV-negative patients. By contrast, in the 29% of HIV-negative patients in whom NIMV succeeded, NIMV avoided tracheal intubation and its associated poor prognosis. In this regard, our results are in full accordance with the well-known benefit of NIMV in different populations of immunosuppressed patients with other causes of respiratory failure [20
]. The retrospective nature of our study, with no standardized modality for ventilatory support, does not allow for conclusions concerning the respective indication of both techniques in this particular population. Rather, the clinical implication of our study is that when NIMV is attempted in a patient with PCP-related acute respiratory failure, the clinician should consider an HIV-positive and an HIV-negative patient with PCP-induced respiratory failure very differently, with a more vigilant watch on HIV-negative cases with NIMV support. This is emphasized by the fact that 80% of HIV-negative patients with NIMV failure died, confirming that in this setting as in others [22
], NIMV failure is associated with a poor prognosis. Importantly, the fact that SAPS II was independently associated with mortality suggests that the worse prognosis of mechanical ventilation was related to a poorer condition of HIV-negative patients at the time of admission. In line with this, a limitation of the present study is that the need for mechanical ventilation was not adjusted for the baseline differences in age and chronic lung condition in HIV-positive and HIV-negative patients.
By contrast, in HIV-positive patients, NIMV succeeded in a large majority of cases, according to the less severe lung alteration by PCP in these patients. Furthermore, when NIMV did fail in HIV-positive patients, the patient survived, reinforcing the evidence of the benefit that could arise from NIMV in severe AIDS-related PCP with acute respiratory failure [9
]. The increased incidence of HIV-negative patients with PCP-induced acute respiratory failure observed in our series confirms previous reports [3
]. This increase possibly was related to a higher prevalence of immunosuppressed patients in our institution, as suggested by the increase in the cohort of transplant recipients. Other factors like heightened awareness for pursuing the diagnosis of PCP, increased familiarity with diagnostic staining methods and detection, and so on also could have accounted for that increase. Importantly, a high proportion of HIV-negative patients had received corticosteroids at a daily dose of less than 15 mg prednisone prior to admission but the majority of these patients were concomitantly exposed to another immunosuppressive therapy. These results raise serious concerns about the appropriateness of guidelines for PCP prophylaxis in HIV-negative immunosuppressed patients, and studies focusing on this question should be recommended. It is noteworthy that in half of the HIV-negative patients in whom it was performed, the CD4 count was higher than 300 cells per microliter, the cutoff value that has been proposed to detect HIV-negative patients at risk for PCP [26
]. In line with this, a recent meta-analysis of studies conducted in bone marrow transplant recipients suggested that a clinical PCP risk threshold rather than a CD count threshold should be used for deciding to administer prophylaxis against PCP in that population [27
As reported before [4
], symptoms were more acute in HIV-negative than HIV-positive patients. The density of P. jiroveci
in the bronchoalveolar lavage specimens was lower in non-AIDS patients, which is well known [18
]. Interestingly, the standard staining methods failed to detect P. jiroveci
in a large proportion of HIV-negative patients. As an important clinical implication, immunofluorescence should be systematically performed in non-AIDS patients with suspected PCP. Twenty-six percent of HIV-negative patients presented with acute circulatory failure at ICU admission, confirming that the PCP-related systemic inflammatory response syndrome could impair hemodynamics similar to viral or bacterial infections [30
We acknowledge several limitations to our study. First, it reflects the experience of a single center. Second, the study is retrospective and neither the choice of the ventilation support nor the modalities of this support were chosen according to predetermined guidelines. Third, the study period was long and the modalities of critical care may have changed over the years, especially for the use of NIMV or HIV management. It is difficult to say what influence this had on the prognosis of PCP, but most likely it did not alter the relevance of the comparison between HIV-negative and HIV-positive cases.