Due to a high rate of LRTI and a high associated mortality rate, the influenza season 2014/15 represented a relevant threat to cancer patients. At presentation, a third of patients were asymptomatic or presented with URTI only, which was relatively unexpected from a clinician’s point of view. In line with other reports [13], influenza-associated URTI itself was not particularly harmful for the patients as all patients dying from influenza suffered from or developed pneumonia in the course of the illness. However, those patients who developed LRTI during the course of the disease showed a relevant impairment in survival (Fig. 1b), emphasising the need to take IVI seriously regardless of initial presentation.
In our cohort, presence of superinfection and prolonged duration from first symptoms to diagnosis were the sole independent prognostic factors associated with higher mortality. A variety of studies investigating the causes of death in patients with IVI found that notably bacterial and fungal superinfections played an important role with regard to morbidity and mortality [14–16]. However, it seems that the mortality in these patients varies widely between the different cohorts [15], but the reasons for this wide scope are not clear. Streptococcus pneumoniae, Haemophilus influenzae, Staphylococcus aureus, E. coli, and A. fumigatus are the superinfecting pathogens most frequently described to date [17]. Studies on the pandemics in the twentieth century suggest a high mortality due to superinfections with these pathogens [14]. However, in the era of widespread vaccination against pneumococci in children and the elderly, this might not necessarily be true anymore, as suggested by Tief et al., who reported a low disease severity in children co-infected with influenza and S. pneumoniae [18]. In our cohort, 7/23 patients with bacterial or fungal infection died (30 %), showing superinfection to be the strongest predictor of death. Interestingly, we also found A. fumigatus to be a relevant pathogen, whereas bacteria typically associated with influenza were outnumbered by enteric bacteria and other Gram-negative bacteria such as Pseudomonas aeruginosa. The underrepresentation of Gram-positive bacteria is remarkable and differs from previous reports [15]. The species of bacteria found in our cohort may represent a microbial spectrum originating from the patient’s flora or from the local health care environment. Nosocomial infections like ventilator-associated pneumonia are known to be caused by both Gram-negative organisms such as Pseudomonas spp., members of the Enterobacteriacae, Acinetobacter spp., or S. maltophilia, as well as by some Gram-positive organisms such as S. aureus [19–22]. Unfortunately, we did not record specifically whether patients were treated as inpatients or outpatients. However, it is not surprising to find a similar spectrum of bacteria in all patients with malignant disease, who, if not hospitalised, usually have close contact to health care institutions. Thus, our findings emphasise that antibiotic therapy for suspected bacterial superinfection in cancer patients with influenza needs to cover Gram-positive as well as Gram-negative bacteria. Moreover, superinfections with A. fumigatus are frequent in patients with malignant disease and IVI, highlighting the need for a thorough diagnostic workup and optional antifungal therapy. Screening for viral co-infection was not consistent between the participating centres, which may lead to underestimation of the impact of viral co- infection on the course of IVI in patients with underlying malignant disease.
Time to diagnosis was identified as a second independent prognostic factor for mortality in our study cohort. This may be due to the fact that patients with pneumonia or critically ill patients are usually considered to have other causative pathogens than influenza virus. Therefore, especially at the beginning of a wave of influenza, these patients might be diagnosed too late. Another possible explanation might be that patients with prolonged diagnosis bear the risk of delaying the initiation of treatment with oseltamivir, potentially leading to impaired efficacy of the drug [23]. In our study, time from onset of symptoms to initiation of antiviral treatment was not recorded, therefore we were not able to draw any definite conclusion regarding this point. Nevertheless, our data support the imperative of immediate NAT testing when cancer patients are suspected to suffer from IVI, even at the stage of URTI, to ensure early treatment as recommended [24, 25].
In addition to the microbial and therapy-associated risk factors mentioned above, risk factors for severe influenza in SCT recipients as described by the European Conference on Infections in Leukaemia (ECIL) are: older age, lymphopenia, first 12 months post SCT, GvHD, and immunosuppressive therapy, as well as having an unrelated or mismatched related donor [23]. In our study cohort, we were able to confirm that age has an impact on mortality, whereas GvHD or immunosuppression and prior SCT did not influence the outcome. It is noteworthy that severe disease did not only occur in patients with profound immunosuppression but also in patients with no ongoing active cancer treatment or those with solid tumours, resulting in comparable survival rates (Fig. 1c). This is unexpected, since most reports of life-threatening IVI originate from patients after allogeneic SCT, where substantial immunosuppression is usually an important risk factor [1, 4, 26].
Data on influenza in patients with solid tumours are generally scarce. A large study of 115 patients with solid tumours suffering from influenza found similar results to our cohort of 21 patients with solid tumours: 23 % of these cancer patients presented with pneumonia, and a mortality rate of 10 % was reported [8]. In line with our results, mortality was associated with prolonged duration to diagnosis of IVI. In contrast to our results, ongoing immunosuppression such as cancer treatment was associated with a severe course of viral infection, which was not seen in our patient cohort. Additionally, in this study detailed information on treatment with oseltamivir was provided, showing a benefit for patients being treated early [8]. In our study, oseltamivir did not influence the outcome significantly, possibly due to a delay of therapy.
As we conducted a retrospective study there are several limitations of our analysis due to a lack of data with regard to several issues. Exact information on start of antiviral therapy is lacking as well as the inpatient/outpatient status of the patients. Also, data on vaccination status and subtype of virus are very limited, and we are not able to draw any conclusions regarding the efficacy of vaccination or the virulence of virus subtypes. These questions will have to be addressed in future prospective studies. Nevertheless, all contacts of patients with malignant disease, e.g., partners, household members, and health care workers, should be urged to undergo seasonal influenza vaccination to better protect this vulnerable collective.