Even if the DP of

Even if the DP of the method calculated from this study is >0.99 (DP=0.995, as compared to 0.992 in our previous study (Gits-Muselli et al., 2015)), the distance between P. jirovecii genotypes remains low using our MLP typing method, resulting in a limited number of genetic groups. This low genetic disparity had been suggested by Parobek et al. when they found that P. jirovecii samples from San Francisco (n=49) and from Spain (n=29) had limited genetic differentiation (Parobek et al., 2014), as compared with Ugandan samples. This limited genetic divergence of European P. jirovecii isolates could be explained by specific aspects of P. jirovecii biology. Indeed, P. jirovecii is considered as an obligate fungal biotroph harbouring parasitic behaviour (Cissé et al., 2014), that may intrinsically limit the opportunities to diverge genetically. Loss of genes is the main characteristics of P. jirovecii genome R406 compared to its last common ancestor Taphrina deformans (Cissé et al., 2014; Ma et al., 2016). The DNA polymerases or the DNA repair machinery of P. jirovecii could also specifically restrict genetic diversity. Interestingly, the percentage of samples harbouring a mixture of FIs at a given centre dropped from 70% in average to 36–42% (We1-FR and BE). This suggests nosocomial PCP as a result of recent exposure to a given P. jirovecii genotype. This also suggests that a given genotype could have a biological advantage over other genotypes resulting in an increased number of patients with PCP due to this genotype. This hypothesis could be reinforced by the fact that Gt123 was responsible for 15 cases of PCP across France that could be considered independent since no contact between these patients could be imagined (renal transplant patients, transplanted and always followed in separate centres). Interestingly, Gt123 was not observed in patients with other causes of immunosuppression, suggesting that Gt123 could not be considered as highly prevalent or currently circulating in France. Gt123 could be well adapted to renal transplant patient immunodeficiency and Gt123 could have spread from the West of France (We1-FR) from 2009 to 2011 to other centres (We2-FR then No-FR and Ce-FR) between 2012 and 2015 (Fig. 5). A similar hypothesis has already been suggested by Sassi et al. (Sassi et al., 2012), who described two distinct outbreaks of PCP in renal transplant recipients due to the same genotype in Zurich (Switzerland) and Munich (Germany) (Hauser et al., 2012; Sassi et al., 2012). These cities are located about 300km apart, which is almost half the distance between the centres in We1-FR and Ce-FR (592km). In addition, identical genotypes were detected in different countries, with a total of 21/134 (15.3%) samples harbouring the same genotype found in more than one country, with the founder genotype (Gt33) and the Gt56 found in ≥4 countries (from Seville, Spain to Cologne, Germany, 2230km apart). This suggests that some genotypes could have disseminated across Europe, or that this typing method is not discriminative enough to observe differences between those isolates (Alanio et al., 2017). This observation and ours suggest that a given P. jirovecii genotype is able to circulate within an area of at least of 600km (Hauser et al., 2012; Sassi et al., 2012). It is possible that some immunocompromised patients were exposed to this specific genotype through contact with immunocompetent carriers, to rather than direct contact between renal transplant recipients located in different centres. When analysing the most frequent genotype (Gt123), the difference between the date of the first (28th May 2009) and the last (23th April 2015) observation was 2156days (i.e., 5.9years) (Fig. 5). If one hypothesises that interhuman transmission is the main mechanism in PCP, this is potentially the longest period of transmission of PCP ever reported and this finding raises the issue of the dissemination of a given genotype over time in different places. When studying a single hospital, the duration of transmission of a given genotype varied from 1 to 2months to 31months (2.6years) (Yiannakis and Boswell, 2016) or 32.2months (2.7years) (Gits-Muselli et al., 2015). This possible long time of transmission may suggest that immunocompetent individuals, such as health care workers, could serve as a potential reservoir, since environmental reservoir is unlikely for this fungus (Alanio and Bretagne, 2017). Thus, two patients of the present study harbouring pure Gt123 were born and had lived >30years in Algeria and in Caribbean islands before kidney transplantation was performed in Paris, France. This Gt123 was probably acquired after the renal transplantation from recent exposure either by direct transmission or through a third source. Whatever the final mode of transmission, our finding argues against reactivation of resident P. jirovecii previously acquired in their country of origin. Another argument for the recent acquisition of a new genotype in PCP is our observation of some temporal relationships between the genotype and the date of PCP (for example, 2012–2013 and Gt16, or 2014–2015 and Gt94). However, a limit to be acknowledged to our study on timing of transmission is that the samples were obtained from patients managed from 1998 to 2015. In particular, the centres with a low number of cases included the oldest cases. This was inherent to the retrospective design of our study and the uneven sampling prevents any definitive conclusion about the temporal evolution of the genotypes circulating across Europe. To address this question, specific prospective studies should be implemented in various European places.