In this study, the concentration of influenza A/H1N1pdm RNA in NPS was evaluated in relation to the time from start of symptoms, considering all patients with confirmed infection whose samples had been sent for diagnosis to the Virology Laboratory of "L. Spallanzani" Institute.
In addition, the influence of clinical severity and of antiviral treatment on the extent and duration of viral shedding was evaluated in a group of hospitalized patients, for whom sequential NPS samples had been analysed. The clinical findings of these patients were not uniformly severe, since, at least at the beginning of pandemics, most patients with confirmed infection were hospitalized.
Concerning the first point, the data shown in Figure 1, representing a cross-sectional evaluation of the viral load of patients at first presentation, indicate that the highest initial values of H1N1/pdm RNA concentration were observed in patients presenting for influenza diagnosis on day 2 from start of symptoms. A significant trend to decrease according to the distance between the symptom onset and first sampling was observed, in line with the reports from other groups [6, 21].
Concerning the second point, in patients serially sampled during their hospitalization period, a progressive decline of viral load was observed in those who received antivirals, while in untreated patients an increase of viral load was observed during the first 5 days, followed by a decrease (Figure 2C). This result is in apparent contrast with the trend to a progressive decline of viral load at first presentation shown in Figure 1. However, the data from this figure represent a cross-sectional analysis of first viral loads detected in patients at diagnosis, and therefore are not directly comparable to those from Figure 2, where the time course in patients serially sampled is reported.
In apparent contrast with data from other authors , in the present study mean viral load values at presentation were not significantly different in patients grouped according to clinical severity, i.e. presence or absence of pneumonia. However, and more importantly, patients with pneumonia showed a delayed viral clearance. In fact, about 85% of patients without pneumonia had influenza-negative samples at days 8-9 after symptom onset (Figure 2B), and 78% after this time point, while a significantly lower proportion of patients with pneumonia were negative at the same time points. It is not possible from our results to determine whether the viral genomes detected in the upper respiratory tract of patients after the initial stages of the infection correspond to actually infectious virions. Lower positive rates for virus culture, compared with RT-PCR assay are commonly observed , and other studies have reported that infectious virus may be detected after the resolution of fever and sometimes after the completion of therapy .
Although there is the possibility of selection bias, due to the fact that patients with more severe clinical courses received protracted care, and were sampled for longer periods compared to those with milder symptoms, these findings are in agreement with a number of previous reports. For instance, To et al. , have shown a slower decline in viral shedding in patients with severe conditions as compared to the mild disease groups, and, recently, Li et al. , have shown similar findings considering treated patients only.
All the factors significantly associated with prolonged viral shedding, i.e. presence of any complication, pneumonia and duration of antiviral administration (Table 2), are strictly interconnected, suggesting that the deciding factor for prolonged shedding is, in fact, the presence of pneumonia. However, because of limited number of cases, it was not possible to apply multivariate analysis to identify the effect of one of the variables (pneumonia and early antiviral treatment) adjusting for the other.
It is reasonable to assume that the delayed clearance observed in severe cases could be secondary to a worse control of viral replication, due to a less effective innate and adaptive immune response, as recently suggested by our group .
Another possible reason for the prolonged virus replication in the upper respiratory tract may be the emergence of viral strains with reduced susceptibility to NAI. In fact, resistant strains have been observed after prolonged administration of antivirals. These resistant strains typically contain a single H275Y substitution in the viral NA gene, and are mostly detected in specimens obtained from patients with a severely compromised immune system and from patients who received oseltamivir, but still had persistent viral replication [5, 20]. The emergence of drug-resistant pandemic influenza strains is not a major public health concern so far, as the majority of the influenza A H1N1/pdm viruses are oseltamivir-susceptible, while oseltamivir-resistant strains remain infrequent and are still sensitive to zanamivir . However, resistant strains have been also isolated in untreated patients, including known or suspected cases of person to person transmission . In our study, the emergence of resistant strains as a possible factor underlying the prolonged viral shedding may be considered negligible, as no mutations known to be associated to either oseltamivir or zanamivir resistance were detected in either treated or untreated patients. To this respect, our findings are in line with recent observations by other authors. For instance Fleury et al. reported prolonged influenza A/H1N1pdm shedding not associated with the emergence of resistance mutation in the viral NA gene in 2 severely ill patients .
The absence of a link between the emergence of resistant strains due to therapy administration and the prolonged viral shedding from the upper respiratory tract of hospitalized patients is further supported by the absence of statistically significant differences in the persistence of PCR positivity in treated vs. untreated patients (Figure 2D), despite a significant reduction of viral load observed at days 4-5 from symptom onset in treated patients (Figure 2C). The lack of significant differences in the persistence of PCR positivity in treated vs. untreated patients is in agreement with data from other studies carried out on seasonal influenza strains , and in apparent contrast with other studies, showing a significant shortening of the duration of viral shedding in A/H1N1pdm -patients exposed to NAI [28, 10]. In another study conducted only in treated patients, more prolonged viral shedding was observed in subjects < 13 years of age than in older patients . In our study we did not evaluate the association between age and delayed virus clearance, since the age of our patients was rather homogeneous, and our case series did not include young patients (Table 2), because our Institute is not a pediatric referral hospital.
Our data indicate a positive correlation between the overall duration of viral shedding and days elapsed between symptom onset and start of antiviral therapy (Figure 3). These results are in agreement with the widely held concept that early initiation of antiviral therapy is important in order to obtain an effective control of the viral replication, and in turn, to shorten the duration of symptoms .
Our study has some limitations: first, the number of patients is rather small, and the lack of significance of some differences may be due to insufficient size of the compared groups; due to the limited number of patients, it was not possible to apply multivariate analysis to the factors associated with prolonged shedding; furthermore, the collection of samples was not prospectively planned, so the sampling was not conducted at fixed time points. Several studies have indicated that viral shedding in the upper respiratory tract is shorter and not always representative of lower airway shedding [2, 26]. In the cases under study, serial bronchoalveolar lavage (BAL) samples were analysed only for the patient with leukaemia, who showed severe pneumonia and received enhanced antiviral treatment. In this patient the BAL returned negative earlier than NPS (day 21 vs. day 34), but no conclusion can be drawn from this anecdotal observation.