Trichosporon spp. are yeast-like organisms responsible for superficial infection (white piedra), allergic pneumonitis and more rarely, invasive infection (Colombo et al., 2011). In the 1980's, Trichosporon invasive infections (ITI) were considered the second most common cause of fungemia in patients with hematological malignancies (Walsh et al., 1986). However, as triazole derivatives, such as fluconazole, became available, the incidence of those infections decreased (Kaufman et al., 2001; Gomes et al., 2014). On the contrary, maybe due to the breakthrough of the echinocandins, now considered as drugs of choice in many clinical contexts at high-risk for invasive fungal infection (IFI), we now deal with a reemergence of this difficult to manage pathogen (Liao et al., 2015). Indeed, the diagnosis of these infections is difficult to anticipate, largely due to the lack of a specific biomarker assay. In addition, identification at the species level is sometimes confusing since the in-depth taxonomic revision of the genus proposed in 1994 (Guého et al., 1994). From then, molecular analyses led to the definition of no less than 50 accepted species within the genus, 17 of which being medically relevant (Hickey et al., 2009; Colombo et al., 2011). Therapeutic management of those infections may also be challenging, since Trichosporon spp. exhibit an intrinsic resistance to the echinocandins and a poor susceptibility to the polyenes (Walsh et al., 1990).

To improve our knowledge of the epidemiology, diagnosis and therapeutic management of those infections, we performed a comprehensive review of the case reports and series published between 1994 and 2015. In addition, we also analyzed studies of in vitro susceptibility testing including a significant number of Trichosporon isolates.

In a first instance, we found 537 cases of invasive Trichosporon infection based on a PubMed search limited between 1994 and 2015. The number of reported cases of invasive Trichosporon infections (ITI) grew significantly through the years, from 139 (25.8%) in the 1994–2004 to 398 (74.2%) between the 2005 and 2015 period (Chi-square test; p < 0.001). Three hundred and thirty-four cases, mainly recorded in series, were excluded because of insufficient data (Figure 1).

Among the 203 cases retained, 199 cases were proven infection, whereas four were considered probable infection. Excluding newborns (25 cases), the median age of those patients was 41.6 years, but the range varied from 3 days of life to 85 years. The sex ratio was calculated at 0.56.

For comparison analysis, the cases were clustered according to the main underlying disease, e.g., homeopathy (n = 79; Grauer et al., 1994; Higgins et al., 1994; Hsiao et al., 1994; Nasu et al., 1994; Hung et al., 1995; Spánik et al., 1995; Itoh et al., 1996; Fanci et al., 1997; Kataoka-Nishimura et al., 1998; Sklair-Levy et al., 1998; Krcmery et al., 1999; Takamura et al., 1999; Erer et al., 2000; Kim et al., 2001; Moretti-Branchini et al., 2001; Fournier et al., 2002; Hadley et al., 2002; Meyer et al., 2002; Chang et al., 2003; Marty et al., 2003; Bassetti et al., 2004; Chowdhary et al., 2004; Viscomi et al., 2004; Antachopoulos et al., 2005; Chan-Tack, 2005; Akagi et al., 2006; Ghiasian et al., 2006; Kendirli et al., 2006; Koyanagi et al., 2006; Matsue et al., 2006; Meguro-Hashimoto et al., 2006; Rodrigues et al., 2006; Hara et al., 2007; Miura et al., 2007; Rieger et al., 2007; Bayramoglu et al., 2008; Hosoki et al., 2008; Thibeault et al., 2008; Tsuji et al., 2008; Fekkar et al., 2009; Walia et al., 2009; Gabriel et al., 2011; Kudo et al., 2011; Hosokawa et al., 2012; Menezes et al., 2012; Chen et al., 2014; Issarachaikul et al., 2014; Karapinar et al., 2014; Capoor et al., 2015; Odero et al., 2015; Pérard et al., 2015; Tanyildiz et al., 2015), other immunodeficiency conditions (n = 41; Canales et al., 1998; Lascaux et al., 1998; Anuradha et al., 2000; Piwoz et al., 2000; Ebright et al., 2001; Abliz et al., 2002; Chakrabarti et al., 2002; Kahana et al., 2003; Nettles et al., 2003; Wynne et al., 2004; Abdala et al., 2005; Gunn et al., 2006; Karabay et al., 2006; Rodrigues et al., 2006; Biasoli et al., 2008; David et al., 2008; Gross and Kan, 2008; Chagas-Neto et al., 2009; Lacasse and Cleveland, 2009; Servonnet et al., 2010; Fadhil et al., 2011; Macêdo et al., 2011; Songcharoen et al., 2011; Basiri et al., 2012; Hirschi et al., 2012; Ogura et al., 2012; Tsai et al., 2012; Chen et al., 2013; Almeida Júnior et al., 2014; Yang et al., 2014; Chaitanya et al., 2015; Nobrega de Almeida Júnior et al., 2015; Ozkaya-Parlakay et al., 2016), miscellaneous (n = 58; De Saedeleer et al., 1994; Lopes et al., 1994, 1995, 1997; Miralles et al., 1994, p. 10; Miró et al., 1994; Sidarous et al., 1994; Still et al., 1994; Hajjeh and Blumberg, 1995; Mathews and Prabhakar, 1995; Melez et al., 1995; Chaumentin et al., 1996; Kouppari et al., 1997; Wang and Lin, 1999; Cawley et al., 2000; Lo Passo et al., 2001; Mooty et al., 2001; Wolf et al., 2001; Chitra et al., 2002; Kustimur et al., 2002; Reddy et al., 2002; Crowther et al., 2003; Madariaga et al., 2003; Spirn et al., 2003; Ramos et al., 2004; O'Gorman et al., 2006; Rodrigues et al., 2006; Kim et al., 2007, 2008; Tian et al., 2007; Fagundes Júnior et al., 2008; Jian et al., 2008; Chagas-Neto et al., 2009; Hickey et al., 2009; Izumi et al., 2009; Shang et al., 2010; Slocumb et al., 2010; Heslop et al., 2011; Tsai et al., 2011; Liao et al., 2012; Padhi et al., 2014; Gonul et al., 2015; Kumar et al., 2015; Negi et al., 2015; Zuo et al., 2015) and newborns (n = 25; Yoss et al., 1997; Sweet and Reid, 1998; Gökahmetoğlu et al., 2002; Panagopoulou et al., 2002; Salazar and Campbell, 2002; Yildiran et al., 2003; Maheshwari et al., 2004; Chagas-Neto et al., 2009; Pereira et al., 2009; Vashishtha et al., 2012; Basu et al., 2015). The main clinical characteristics of each group are summarized in Table 1 and are detailed in Tables S1–S4.

Trichosporon emerged as a pathogen in the 70–80's, mainly for neutropenic patients (Gold et al., 1981; Walsh et al., 1986). Despite the lack of data providing the real incidence of ITI at this time, published series supported the fact that Trichosporon was the second most frequent agent responsible for fungemia in this population (Leblond et al., 1986; Anaissie et al., 1992). Based on clinical and experimental data, it was then shown that azoles had a significant activity against Trichosporon, while being less toxic that the standard at that time treatment, amphotericin B (Anaissie et al., 1992). This and the wider use of azole drugs either as prophylaxis or empirical therapy probably explain the decreasing concern regarding ITI in 90's. Between 1989 and 2002, an estimated incidence, of 81 cases per 100,000 admissions and 3 per 100,000 admissions were reported in patients with hematologic malignancies and patients with solid tumors, respectively (Kontoyiannis et al., 2004). Similarly, in an Italian multicentric study conducted in 15 hematology departments between 1983 and 2002, an incidence of 0.4% was observed in patients with acute leukemia (Girmenia et al., 2005). However, more recently, the MD Anderson Cancer Center (TX, USA) demonstrated an increase in the incidence of invasive yeast infections not due to Candida or Cryptococcus, of which 20% were Trichosporon, from 1.8 to 2.35 cases per 100,000 patient-days between the periods 1998–2004 and 2005–2010 (Chitasombat et al., 2012). Again this may reflect the change in the antifungal stewardship for neutropenic patients. Indeed, echinocandin drugs, which emerged as valuable options of choice for either prophylaxis, pre-emptive or empirical therapy, have no effect on Trichosporon (Asada et al., 2006; Matsue et al., 2006; Hiramatsu et al., 2008; Suzuki et al., 2010; Park et al., 2014; Nachbaur et al., 2015). Our review supports this re-emergence with a significant increase of reported cases in the second part of the 2000s decade.

In accordance with previous studies, we found that hematological diseases (38.9% of the cases), mainly acute leukemia, remain the most common underlying condition of ITI, however, with a much smaller percentage than previous reported (Girmenia et al., 2005). Eighty-five percent of the patients with hematological diseases were neutropenic at the time of ITI diagnosis, supporting the fact that neutrophil cells are, as for candidiasis and aspergillosis, essential in the prevention of ITI. The disruption of the mucosal barrier owing to chemotherapy-induced mucositis may also contribute to invasion of the yeast cells. This, combined with the modification of the digestive flora due to previous antibiotic therapies, observed in 58–96% of the cases according to the patients' group, may favor the translocation of yeasts from the digestive tract to the blood vessels. However, our study also revealed that not only patients with hematological malignancy are now concerned, as we showed that other conditions mainly related to severe baseline diseases (e.g., SOT, auto-immune diseases, solid tumor, preterm birth, extensive skin lesions such as burns, or pemphigus) and invasive procedures (e.g., CVC, prosthetic heart valves), may also favor the occurrence of ITI.

The origin of Trichosporon infections, notably post-operative ones, is uncertain. Indeed, Trichosporon has been isolated from the hospital environment (Fanfair et al., 2013), but is also considered a commensal of the digestive tract and may colonize the skin of healthy patients (Colombo et al., 2011; Zhang et al., 2011; Cho et al., 2015). In neonates, it is likely that the increasing permissiveness of the natural barriers, mainly the skin and the digestive mucosa, may favor the translocation of any yeast types, including Trichosporon, which has also been isolated from the skin of preterm infants (Kaufman et al., 2001). In addition, total parenteral nutrition, frequently used in this context, may have been the source of an outbreak of T. asahii bloodstream infections (Vashishtha et al., 2012).

In any case, the clinical features of invasive trichosporonosis were not specific. The persistence of fever during ongoing treatment with AMB or echinocandins may be more supportive (Colombo et al., 2011). Dissemination is a common feature, particularly in patients with homeopathy and newborns (100% of the cases). While difficult to interpret when solely based on culture results, pulmonary lesions are frequently reported as diffuse nodules, lobar pneumonia, reticulonodular infiltrates, or mass-like lesions (Akagi et al., 2006; Kendirli et al., 2006). Positive culture for Trichosporon from respiratory tract samples, notably when repeated should always be interpreted with caution in patients with risk factors. Skin lesions were the second most common tissue involvement and warrant the addition of trichosporonosis to the list, which included candidiasis and fusariosis, as a cause of skin lesions during febrile neutropenia. The cutaneous lesions, mainly located in the lower extremities, the trunk and the face, are typically hemorrhagic macular or maculopapular, and differ in subtle way from those of Candida that tend to be punctate, pseudopustular (necrotic epicenter) with a rim of erythema (Nucci et al., 1992; Higgins et al., 1994; Miura et al., 2007).

The diagnosis of ITI always relies on the isolation of a yeast-like organism from a clinical specimen further identified as Trichosporon. Direct examination of clinical specimens rarely contributed to diagnosis as it rarely demonstrated arthroconidia in samples. It is remarkable than about a third of the isolates had not been identified at the species level. Phenotypic approaches have been shown to be poorly efficient and direct sequencing of the IGS1 region of the ribosomal DNA is now considered the reference method for species identification of Trichosporon isolates (Sugita et al., 2002). MALDI-TOF mass spectrometry, which has recently been shown to be a valuable alternative for routine identification (de Almeida Júnior et al., 2014), should help at reducing the part of non- or uncertainly-identified isolates in order to test more in depth, possible species-specific epidemiological traits, and clinical scenarios. Therapeutic adaptation may also be required according to the causative species.

Unfortunately, serum markers are either unsatisfactory or have not been evaluated in depth for the diagnosis of ITI. Although, it is well-known that Trichosporon shares some antigenic properties with Cryptococcus neoformans (Campbell et al., 1985), a cross reaction with the cryptococcal GXM antigen assay was only found in 26% of the tested cases; a negative assay thus does not rule out the diagnosis. BGD assays may be more interesting with nine positive results among 11 sera tested. Similarly, in a case series study, 12 out of 25 patients with Trichosporon fungemia have been reported with significant levels of BDG (Suzuki et al., 2010). While the use of molecular tools, not commercially available, is more anecdotal, a study reported a good performance of a real-time PCR assay to detect T. asahii DNA from the blood samples (Tsuji et al., 2008).

T. asahii was shown to be associated with a poor prognosis what may be due in part to its lower susceptibility to antifungal drugs (Pfaller et al., 2009). Nevertheless, neither the CLSI nor the EUCAST consortium has proposed a standardized procedure for AST of Trichosporon. Zaragoza et al. analyzed different parameters for optimal growth of nonfermentative yeasts, including some Trichosporon strains, for AST (Zaragoza et al., 2011), and showed that the use of yeast nitrogen base medium (YNB), agitation and incubation at 30°C resulted in a better growth and a more reliable and stable measurement of the MIC values. However, whatever was the conditions used they found an excellent agreement between the methods when reading the MICs after 48 h of incubation (Zaragoza et al., 2011).

Despite the lack of defined clinical breakpoints, the constantly elevated MICs against the echinocandins support natural resistance to this antifungal class (Colombo et al., 2011). Similarly, because the minimal fungicidal concentrations of polyenes have been shown to be 20-fold greater than the minimal inhibitory concentration, the Trichosporon genus was also considered naturally resistant to these drugs, notably AMB (Walsh et al., 1990). However, after the taxonomic reevaluation, it has been shown that T. asahii clearly exhibits higher MICs against AMB compared to other species, such as T. mucoides/dermatis and T. inkin (Rodriguez-Tudela et al., 2005; Tsai et al., 2012). On the contrary, T. mucoides/dermatis appears to be more resistant in vitro to fluconazole compared to other species (Metin et al., 2005; Rodriguez-Tudela et al., 2005). Nevertheless, whatever is the species considered, voriconazole, posaconazole, and isavuconazole offer a better in vitro efficacy compared to fluconazole, even if some T. asahii isolates belonging to genotype 3 with reduced susceptibility to voriconazole has been described (Yang et al., 2013; Arabatzis et al., 2014). The heterogeneity of population structure of T. asahii may also explain the great variability in MICs found in this species against the different antifungals evaluated.

To the best of our knowledge, there is no study designed specifically to compare different antifungal regimens in the management of those infections. The lack of antifungal treatment led to a fatal outcome almost constantly. In our review, the overall unfavorable outcome rate was at 44.3%. This rate was quite similar to that reported in a US Cancer Center (50%) but slightly lower than that found in the study by Girmenia et al. (64.7%), where AMB was the main (76.5%) drug used for treatment. Indeed, our review suggests that azole-based therapy may be superior to echinocandin- or amphotericin-based therapies. Similar findings were reported in a retrospective study of 33 patients with hematological malignancy: the mortality rate was 63% with azole-based therapy vs. 100% in the absence of any azole drug in the therapeutic regimen (p = 0.031; Suzuki et al., 2010). In our study, the use of voriconazole significantly improved the prognosis of patients with hematological malignancy supporting the recent recommendations proposed by the ESCMID (Arendrup et al., 2014).

Although, hematologic malignancies are the main underlying diseases, invasive Trichosporon infection may also occur in other contexts of immunosuppression, in newborns and in various conditions of debilitating diseases, with different clinical presentations. Due to the natural resistance to echinocandins and polyenes of Trichosporon, breakthrough infections are common. Bloodstream infection sometimes combined with pneumonia and/or skin lesions are the most common clinical feature in patients with hematologic malignancies. T. asahii is the predominant causative species and is associated with a poor prognosis, possibly linked to its reduce sensitivity to some azole drugs. Nevertheless, first line therapy should rely on azole derivatives and particularly voriconazole, which exhibits the best in vitro activity against Trichosporon species and significantly leads to a better outcome in patients with underlying homeopathy.

JN: designed the study, helped with acquisition and data analysis, drafted and revised the work, approved the final work and agrees with all the aspects of the work; CH: designed the study, helped with data analysis, drafted and revised the work, approved the final work, and agree with all the aspects of the work.