Extracorporeal membrane oxygenation in Pneumocystis jirovecii pneumonia: outcome in HIV and non-HIV patients

Main text Pneumocystis jirovecii pneumonia (PJP) is a severe complication of immunosuppression that is associated with high mortality, depending on the underlying type of immunosuppression [1]. Consequently, the incidence of PJP is higher in non-HIV patients than in HIV patients, because of the increased use of immunosuppressive therapies for widespread indications [2]. So far, there is little evidence for veno-venous extracorporeal membrane oxygenation (ECMO) treatment in cases of PJP-induced severe acute respiratory distress syndrome (ARDS). Particularly, there is no study reporting and comparing the outcome of PJP requiring ECMO therapy in HIV and non-HIV patients. Therefore, we report retrospective data of a singlecentre registry of patients with severe respiratory failure, requiring ECMO support at our centre between January 2009 and April 2019. ECMO support was initiated when lung-protective mechanical ventilation was not able to prevent hypoxemia or hypercapnia, based on the treating medical team’s judgement. A total of 337 ECMO patients were screened, and 18 patients with PJP were identified (Table 1). Diagnosis of PJP was verified via positive immunofluorescence microscopy in 13 patients (72%). Five patients (28%) displayed high PCR levels (median 67.000 [5.200–250.000] copies/ml) with conclusive symptoms and radiological findings but negative immunofluorescence microscopy. Microbiological testing was performed in bronchoalveolar lavage. In 14 patients (78%), PJP was diagnosed before the initiation of ECMO therapy. HIV was the cause of immunosuppression in 6 patients, whereas 12 patients had other subtypes of immunosuppression (non-HIV group, Fig. 1a). In all cases, HIV was diagnosed during index hospitalisation. Patients therefore were without previous antiretroviral treatment. There were no significant differences between these two groups in relation to sex, comorbidities, ventilator settings, LDH levels or survival prediction scores (SOFA, APACHE II and RESP, Table 1). Patients with HIV were younger than non-HIV patients, and the interval between the start of mechanical ventilation and ECMO therapy was shorter in the non-HIV group. Overall ECMO weaning rate was 39%, without a significant difference between HIV and non-HIV patients. Overall hospital survival was 22%. Withdrawal of care when further curative treatment was deemed futile was the most common cause of death (nine patients, 64.3%). Survival rate was higher in HIV than in non-HIV patients (50% vs. 8%, p = 0.045, Fig. 1b). It has been shown previously in a non-ECMO setting that the outcome in HIV-negative PJP patients is worse than in patients with HIV [3], and our data confirm these earlier observations.

Pneumocystis jirovecii pneumonia (PJP) is a severe complication of immunosuppression that is associated with high mortality, depending on the underlying type of immunosuppression [1]. Consequently, the incidence of PJP is higher in non-HIV patients than in HIV patients, because of the increased use of immunosuppressive therapies for widespread indications [2]. So far, there is little evidence for veno-venous extracorporeal membrane oxygenation (ECMO) treatment in cases of PJP-induced severe acute respiratory distress syndrome (ARDS). Particularly, there is no study reporting and comparing the outcome of PJP requiring ECMO therapy in HIV and non-HIV patients.
Therefore, we report retrospective data of a singlecentre registry of patients with severe respiratory failure, requiring ECMO support at our centre between January 2009 and April 2019. ECMO support was initiated when lung-protective mechanical ventilation was not able to prevent hypoxemia or hypercapnia, based on the treating medical team's judgement.
A total of 337 ECMO patients were screened, and 18 patients with PJP were identified (Table 1). Diagnosis of PJP was verified via positive immunofluorescence microscopy in 13 patients (72%). Five patients (28%) displayed high PCR levels (median 67.000 [5.200-250.000] copies/ml) with conclusive symptoms and radiological findings but negative immunofluorescence microscopy. Microbiological testing was performed in bronchoalveolar lavage. In 14 patients (78%), PJP was diagnosed before the initiation of ECMO therapy.
HIV was the cause of immunosuppression in 6 patients, whereas 12 patients had other subtypes of immunosuppression (non-HIV group, Fig. 1a). In all cases, HIV was diagnosed during index hospitalisation. Patients therefore were without previous antiretroviral treatment.
There were no significant differences between these two groups in relation to sex, comorbidities, ventilator settings, LDH levels or survival prediction scores (SOFA, APACHE II and RESP, Table 1). Patients with HIV were younger than non-HIV patients, and the interval between the start of mechanical ventilation and ECMO therapy was shorter in the non-HIV group.
Overall ECMO weaning rate was 39%, without a significant difference between HIV and non-HIV patients. Overall hospital survival was 22%. Withdrawal of care when further curative treatment was deemed futile was the most common cause of death (nine patients, 64.3%). Survival rate was higher in HIV than in non-HIV patients (50% vs. 8%, p = 0.045, Fig. 1b).
It has been shown previously in a non-ECMO setting that the outcome in HIV-negative PJP patients is worse than in patients with HIV [3], and our data confirm these earlier observations.
There are possible explanations for the better prognosis of HIV in this setting. On average, HIV patients are younger, and immunosuppression in HIV patients is reversible and can be resolved with the initiation of antiretroviral treatment. Moreover, the high mortality of non-HIV patients is associated with the underlying disease itself and a faster and more fulminant progression of the disease with more severe hypoxia and a higher prevalence of shock [4].
One third of our patients in the non-HIV group could be weaned successfully from ECMO support, suggesting that mortality was not only associated with ARDS, but underlying comorbidities may have been predominant. Moreover, there was a trend towards more frequent acute haemodialysis in non-HIV patients, illustrating that these patients had more complications and suffered from multi-organ failure.

ICU and hospital survival
In summary, a survival rate of 50% in HIV patients is similar to the average survival of ECMO patients with ARDS of any origin as shown by the CAESAR (63%) or the EOLIA trial (65%) [5,6]. Therefore, ECMO therapy should not be withheld from patients with HIV-associated PJP.