Prolonged influenza shedding in cancer patients has been observed for seasonal strains 
. Regarding the H1N1pdm virus, it has been shown that prolonged virus shedding in cancer patients may occur, although such a phenomenon has only been documented through cases involving a single studied patient 
. Here, we prospectively and systematically collected information from a cohort of hospitalized cancer patients with severe H1N1pdm infections. These patients presented high mortality, prolonged viral shedding and H1N1pdm evolution without the emergence of oseltamivir resistance. This is the first study to address viral shedding and resistance in cancer patients with H1N1pdm infections; thus, it may provide insight into the role of cancer patients as potential human reservoirs for this pandemic virus.
Unlike previous reports, our population was composed of hospitalized, severely immunocompromised cancer patients 
. Most of them were young, had hematologic malignancies and received chemotherapy and systemic steroids in the weeks that preceded the H1N1pdm infection. The patients were treated with oseltamivir in the early course of the infection (the median time to antiviral initiation was three days). A total of 13 patients (54%) required intensive care and presented severe respiratory distress. In these patients, the mortality rates were higher (38%) than those observed for general ICU patients suffering from H1N1pdm infections 
as well as for non-critically ill cancer patients 
. However, the outcomes were not different from those reported for cancer patients requiring mechanical ventilation 
Interestingly, during the influenza season, 14 patients (58.3%) with febrile neutropenia were identified as H1N1pdm cases, a condition that is not usually investigated in this scenario. However, febrile neutropenic cancer patients have an increased risk of developing respiratory distress and multi-organ failure. Therefore, screening for respiratory viruses and prompt initiation of oseltamivir treatment should be considered in these patients. Febrile neutropenia indicates a poor prognostic with respect to a patient's outcome, but neutropenia duration in our cohort of patients was less than seven days. Thus, prolonged viral shedding might not have a correlation with neutropenia.
We observed the persistence of H1N1pdm in 5 out of 10 patients studied for this purpose. In these individuals, viral shedding continued for at least 11 days, despite the use of oseltamivir. The median duration of viral shedding in our population was 23 days, and two pediatric patients with acute lymphoblastic leukemia showed even longer virus secretions (44 and 63 days; and Table S10
), although it is difficult to determine whether viral persistence was due to cancer per se
or to acute lung injury and mechanical ventilation.
Influenza shedding is not considered to last long, and it disappears seven days after the onset (2.4 and 4.5 days for oseltamivir- and placebo-treated groups, respectively) 
. Studies aimed at monitoring 2009 H1N1pdm virus shedding using randomized trials with appropriate controls, such as outpatients and hospitalized or immunocompromised individuals, have not yet been conducted. Because we were also unable to establish age-matched controls with or without immunosuppression, since this study was conducted during the peak of the first wave of the 2009 pandemics in Brazil, we compared our work to other studies on H1N1pdm shedding in general, hospitalized or immunocompromised populations 
. In , we summarize the cohort used in each study, whether or not they were immunocompromised, their underlying diseases and the number of patients analyzed for viral shedding in each of these studies. We compared the maximum periods of shedding among these different populations and the number of patients that secreted the virus for more than seven days. These data would be more informative and relevant from a public health point of view because the duration of the quarantine for H1N1pdm was approximately that long 
. We found () that in households in Hong Kong 
and Canada 
, the maximum periods of H1N1pdm shedding ranged from 8 to 11 days. These periods were not different from what was observed with military cadets 
and during the containment phase of the pandemics in Vietnam 
(). Regarding H1N1pdm shedding among infants, Hien et al. showed that children five to nine years old could secrete the H1N1pdm virus for five to six days, which is markedly lower than what is observed for the seasonal influenza virus 
. Compared to our results, we observed higher periods of viral shedding in two seven-year-old patients with acute lymphoblastic leukemia ( and ). However, a single report on two travelers in France showed that these apparently immunocompetent individuals secreted the H1N1pdm virus for 14 and 28 days 
(). Although these periods of time are comparable to the time frame of virus secretion in hospitalized patients in China 
and our work, the study from Felury et al. might be as biased as ours by the small size of the cohort (). Influenza-infected immunocompromised individuals may have prolonged influenza shedding 
; however, more insights are necessary to better comprehend the dynamics of the H1N1pdm virus in these individuals. Mora et al. showed that in HIV-1-infected individuals, co-infection with the H1N1pdm virus might lead to an outcome not different from the one expected for immunocompetent subjects, although no systematic analysis of viral shedding was performed 
. A similar conclusion was also drawn for transplant recipient individuals, whose longest periods of viral shedding did not exceed 11 days 
(). In our study, we found periods of H1N1pdm shedding similar to what the CDC observed for leukemia patients 
(). Although the small size of these cohorts of immunocompromised individuals 
, including ours, may require a more conclusive and mechanistic analysis, these observations may stimulate further systematic studies to understand or gain insight into factors associated with prolonged H1N1pdm shedding. In addition, it might give insights on basic studies on influenza pathogenesis.
Comparisons of the period of viral shedding in the general, hospitalized and immunocompromised populations.
Our results highlight the need for closer surveillance of cancer patients with H1N1pdm infections until the detection of the first negative sample. We hypothesize that follow-up protocols aimed at monitoring the persistence of viral shedding in cancer patients may be relevant if patients are submitted to immunosuppressive therapies in the days or weeks prior to or following an H1N1pdm infection.
Next, we sought to determine viral evolution during prolonged shedding. We found that some amino acid changes persisted from the initial symptoms until 30 days thereafter, suggesting that these patients had viral persistence rather than re-infection. In addition, an extra amino acid change (D238P) was found in the viral HA sequenced a month after the onset of illness, suggesting continuing viral evolution. Although some of the mutations that we found (L52S, L70P, P100S, C153L, T214A, D238P, Q293R and I321V) in strains 5645s2/09 and 5645s3/09 have not been previously described, other amino acid residue changes that were detected in our study (P100S and T214A) have been found in H1N1pdm viruses throughout the world without a significant link to viral pathogenesis or antigenic variation 
Influenza viruses resistant to antiviral drugs have been reported in immunocompromised patients, 
and this resistance might be associated with prolonged viral shedding 
. Notably, five isolates from two patients had high IC50
values to neuraminidase inhibitors (NAIs). An NA activity that was multi-resistant to NAIs was identified and could be due to the presence of a Streptococcus
strain found in throat swabs and tracheal aspirates. Pyrosequencing analyses of samples with high IC50
values revealed that these specimens were H275 wild-type sensitive viruses. These results reinforce the need for additional genotyping assays to confirm the identification of putative resistant strains identified using functional assays. In addition, our findings show the need for investigating other sources of NA activity in virus isolates with odd IC50
The apparent paradox of prolonged viral shedding without antiviral resistance could be explained by either the inability of the immunocompromised host to effectively clear the H1N1pdm virus 
or inefficient absorption of the drug 
. Because we combined both clinical and molecular virology data, our results might contribute to the discussion on the adequate duration and type of anti-H1N1pdm treatment in immunocompromised patients with a protracted course. In these patients, the use of parenteral systemic or inhaled antivirals should also be investigated.
Although our work further investigates the unique dynamics of H1N1pdm virus infection in immunocompromised hosts, some caveats must be noted. Because our investigation started during a new pandemic, the clinical evaluation and management protocols changed during the course of the study as new data emerged from the literature and from updated recommendations 
. As the pandemic reached its peak in South America, the establishment of a larger and more diverse cohort with age-matched controls with or without immunosuppression became complex. Thus, more in-depth multivariate and mechanistic clinical analyses were limited. Moreover, no recommendations for monitoring viral persistence were available; therefore, only a subset of severely ill patients admitted to the ICU was evaluated. Despite that, an important connection between clinical and laboratory information was studied, revealing the continuous evolution of H1N1pdm HA sequences and the stability of the NA gene in severely ill patients 
In conclusion, this study provides evidence that severe H1N1pdm infection is associated with significant morbidity and mortality in cancer patients. In these patients, viral persistence without the emergence of antiviral resistance may occur during the clinical course of the disease. This result has major implications for the clinical management of H1N1pdm infections and infection control strategies. Our study may provide insights into H1N1pdm shedding and might contribute to the development of new guidelines to manage cancer patients with H1N1pdm infection.